aboutsummaryrefslogtreecommitdiff
path: root/arch/mips/math-emu/sp_simple.c
diff options
context:
space:
mode:
Diffstat (limited to 'arch/mips/math-emu/sp_simple.c')
-rw-r--r--arch/mips/math-emu/sp_simple.c39
1 files changed, 10 insertions, 29 deletions
diff --git a/arch/mips/math-emu/sp_simple.c b/arch/mips/math-emu/sp_simple.c
index ae4fcfafd85..f1ffaa9a17e 100644
--- a/arch/mips/math-emu/sp_simple.c
+++ b/arch/mips/math-emu/sp_simple.c
@@ -5,8 +5,6 @@
* MIPS floating point support
* Copyright (C) 1994-2000 Algorithmics Ltd.
*
- * ########################################################################
- *
* This program is free software; you can distribute it and/or modify it
* under the terms of the GNU General Public License (Version 2) as
* published by the Free Software Foundation.
@@ -18,33 +16,17 @@
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
- * 59 Temple Place - Suite 330, Boston MA 02111-1307, USA.
- *
- * ########################################################################
+ * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*/
-
#include "ieee754sp.h"
-int ieee754sp_finite(ieee754sp x)
-{
- return SPBEXP(x) != SP_EMAX + 1 + SP_EBIAS;
-}
-
-ieee754sp ieee754sp_copysign(ieee754sp x, ieee754sp y)
-{
- CLEARCX;
- SPSIGN(x) = SPSIGN(y);
- return x;
-}
-
-
-ieee754sp ieee754sp_neg(ieee754sp x)
+union ieee754sp ieee754sp_neg(union ieee754sp x)
{
COMPXSP;
EXPLODEXSP;
- CLEARCX;
+ ieee754_clearcx();
FLUSHXSP;
/*
@@ -55,30 +37,29 @@ ieee754sp ieee754sp_neg(ieee754sp x)
SPSIGN(x) ^= 1;
if (xc == IEEE754_CLASS_SNAN) {
- ieee754sp y = ieee754sp_indef();
- SETCX(IEEE754_INVALID_OPERATION);
+ union ieee754sp y = ieee754sp_indef();
+ ieee754_setcx(IEEE754_INVALID_OPERATION);
SPSIGN(y) = SPSIGN(x);
- return ieee754sp_nanxcpt(y, "neg");
+ return ieee754sp_nanxcpt(y);
}
return x;
}
-
-ieee754sp ieee754sp_abs(ieee754sp x)
+union ieee754sp ieee754sp_abs(union ieee754sp x)
{
COMPXSP;
EXPLODEXSP;
- CLEARCX;
+ ieee754_clearcx();
FLUSHXSP;
/* Clear sign ALWAYS, irrespective of NaN */
SPSIGN(x) = 0;
if (xc == IEEE754_CLASS_SNAN) {
- SETCX(IEEE754_INVALID_OPERATION);
- return ieee754sp_nanxcpt(ieee754sp_indef(), "abs");
+ ieee754_setcx(IEEE754_INVALID_OPERATION);
+ return ieee754sp_nanxcpt(ieee754sp_indef());
}
return x;
le='width: 4.4%;'/> -rw-r--r--arch/sparc/kernel/cpumap.c435
-rw-r--r--arch/sparc/kernel/cpumap.h16
-rw-r--r--arch/sparc/kernel/devices.c51
-rw-r--r--arch/sparc/kernel/dma.c12
-rw-r--r--arch/sparc/kernel/ds.c1273
-rw-r--r--arch/sparc/kernel/dtlb_miss.S39
-rw-r--r--arch/sparc/kernel/dtlb_prot.S54
-rw-r--r--arch/sparc/kernel/ebus.c552
-rw-r--r--arch/sparc/kernel/entry.S986
-rw-r--r--arch/sparc/kernel/entry.h254
-rw-r--r--arch/sparc/kernel/errtbls.c144
-rw-r--r--arch/sparc/kernel/etrap_32.S (renamed from arch/sparc/kernel/etrap.S)71
-rw-r--r--arch/sparc/kernel/etrap_64.S240
-rw-r--r--arch/sparc/kernel/fpu_traps.S384
-rw-r--r--arch/sparc/kernel/ftrace.c147
-rw-r--r--arch/sparc/kernel/getsetcc.S24
-rw-r--r--arch/sparc/kernel/head.S1325
-rw-r--r--arch/sparc/kernel/head_32.S815
-rw-r--r--arch/sparc/kernel/head_64.S984
-rw-r--r--arch/sparc/kernel/helpers.S63
-rw-r--r--arch/sparc/kernel/hvapi.c200
-rw-r--r--arch/sparc/kernel/hvcalls.S823
-rw-r--r--arch/sparc/kernel/hvtramp.S137
-rw-r--r--arch/sparc/kernel/idprom.c76
-rw-r--r--arch/sparc/kernel/init_task.c29
-rw-r--r--arch/sparc/kernel/iommu.c862
-rw-r--r--arch/sparc/kernel/iommu_common.h59
-rw-r--r--arch/sparc/kernel/ioport.c573
-rw-r--r--arch/sparc/kernel/irq.c674
-rw-r--r--arch/sparc/kernel/irq.h157
-rw-r--r--arch/sparc/kernel/irq_32.c362
-rw-r--r--arch/sparc/kernel/irq_64.c973
-rw-r--r--arch/sparc/kernel/itlb_miss.S39
-rw-r--r--arch/sparc/kernel/ivec.S51
-rw-r--r--arch/sparc/kernel/jump_label.c41
-rw-r--r--arch/sparc/kernel/kernel.h186
-rw-r--r--arch/sparc/kernel/kgdb_32.c172
-rw-r--r--arch/sparc/kernel/kgdb_64.c200
-rw-r--r--arch/sparc/kernel/kprobes.c603
-rw-r--r--arch/sparc/kernel/kstack.h83
-rw-r--r--arch/sparc/kernel/ktlb.S352
-rw-r--r--arch/sparc/kernel/ldc.c2374
-rw-r--r--arch/sparc/kernel/led.c39
-rw-r--r--arch/sparc/kernel/leon_kernel.c499
-rw-r--r--arch/sparc/kernel/leon_pci.c100
-rw-r--r--arch/sparc/kernel/leon_pci_grpci1.c722
-rw-r--r--arch/sparc/kernel/leon_pci_grpci2.c914
-rw-r--r--arch/sparc/kernel/leon_pmc.c93
-rw-r--r--arch/sparc/kernel/leon_smp.c469
-rw-r--r--arch/sparc/kernel/mdesc.c974
-rw-r--r--arch/sparc/kernel/misctrap.S97
-rw-r--r--arch/sparc/kernel/module.c176
-rw-r--r--arch/sparc/kernel/muldiv.c240
-rw-r--r--arch/sparc/kernel/nmi.c273
-rw-r--r--arch/sparc/kernel/of_device.c586
-rw-r--r--arch/sparc/kernel/of_device_32.c429
-rw-r--r--arch/sparc/kernel/of_device_64.c726
-rw-r--r--arch/sparc/kernel/of_device_common.c178
-rw-r--r--arch/sparc/kernel/of_device_common.h36
-rw-r--r--arch/sparc/kernel/pci.c1009
-rw-r--r--arch/sparc/kernel/pci_common.c553
-rw-r--r--arch/sparc/kernel/pci_fire.c522
-rw-r--r--arch/sparc/kernel/pci_impl.h186
-rw-r--r--arch/sparc/kernel/pci_msi.c446
-rw-r--r--arch/sparc/kernel/pci_psycho.c618
-rw-r--r--arch/sparc/kernel/pci_sabre.c614
-rw-r--r--arch/sparc/kernel/pci_schizo.c1509
-rw-r--r--arch/sparc/kernel/pci_sun4v.c1024
-rw-r--r--arch/sparc/kernel/pci_sun4v.h92
-rw-r--r--arch/sparc/kernel/pci_sun4v_asm.S362
-rw-r--r--arch/sparc/kernel/pcic.c370
-rw-r--r--arch/sparc/kernel/pcr.c300
-rw-r--r--arch/sparc/kernel/perf_event.c1804
-rw-r--r--arch/sparc/kernel/pmc.c79
-rw-r--r--arch/sparc/kernel/power.c76
-rw-r--r--arch/sparc/kernel/process_32.c (renamed from arch/sparc/kernel/process.c)407
-rw-r--r--arch/sparc/kernel/process_64.c772
-rw-r--r--arch/sparc/kernel/prom.h11
-rw-r--r--arch/sparc/kernel/prom_32.c (renamed from arch/sparc/kernel/prom.c)328
-rw-r--r--arch/sparc/kernel/prom_64.c635
-rw-r--r--arch/sparc/kernel/prom_common.c162
-rw-r--r--arch/sparc/kernel/prom_irqtrans.c842
-rw-r--r--arch/sparc/kernel/psycho_common.c470
-rw-r--r--arch/sparc/kernel/psycho_common.h48
-rw-r--r--arch/sparc/kernel/ptrace_32.c (renamed from arch/sparc/kernel/ptrace.c)104
-rw-r--r--arch/sparc/kernel/ptrace_64.c1106
-rw-r--r--arch/sparc/kernel/reboot.c53
-rw-r--r--arch/sparc/kernel/rtrap_32.S (renamed from arch/sparc/kernel/rtrap.S)120
-rw-r--r--arch/sparc/kernel/rtrap_64.S371
-rw-r--r--arch/sparc/kernel/sbus.c676
-rw-r--r--arch/sparc/kernel/sclow.S86
-rw-r--r--arch/sparc/kernel/semaphore.c155
-rw-r--r--arch/sparc/kernel/setup_32.c (renamed from arch/sparc/kernel/setup.c)263
-rw-r--r--arch/sparc/kernel/setup_64.c635
-rw-r--r--arch/sparc/kernel/signal.c1123
-rw-r--r--arch/sparc/kernel/signal32.c794
-rw-r--r--arch/sparc/kernel/signal_32.c556
-rw-r--r--arch/sparc/kernel/signal_64.c547
-rw-r--r--arch/sparc/kernel/sigutil.h9
-rw-r--r--arch/sparc/kernel/sigutil_32.c121
-rw-r--r--arch/sparc/kernel/sigutil_64.c95
-rw-r--r--arch/sparc/kernel/smp.c430
-rw-r--r--arch/sparc/kernel/smp_32.c398
-rw-r--r--arch/sparc/kernel/smp_64.c1523
-rw-r--r--arch/sparc/kernel/sparc-stub.c724
-rw-r--r--arch/sparc/kernel/sparc_ksyms.c297
-rw-r--r--arch/sparc/kernel/sparc_ksyms_32.c31
-rw-r--r--arch/sparc/kernel/sparc_ksyms_64.c52
-rw-r--r--arch/sparc/kernel/spiterrs.S245
-rw-r--r--arch/sparc/kernel/sstate.c122
-rw-r--r--arch/sparc/kernel/stacktrace.c85
-rw-r--r--arch/sparc/kernel/starfire.c116
-rw-r--r--arch/sparc/kernel/sun4c_irq.c261
-rw-r--r--arch/sparc/kernel/sun4d_irq.c859
-rw-r--r--arch/sparc/kernel/sun4d_smp.c461
-rw-r--r--arch/sparc/kernel/sun4m_irq.c691
-rw-r--r--arch/sparc/kernel/sun4m_smp.c318
-rw-r--r--arch/sparc/kernel/sun4setup.c75
-rw-r--r--arch/sparc/kernel/sun4v_ivec.S341
-rw-r--r--arch/sparc/kernel/sun4v_tlb_miss.S428
-rw-r--r--arch/sparc/kernel/sunos_asm.S67
-rw-r--r--arch/sparc/kernel/sunos_ioctl.c230
-rw-r--r--arch/sparc/kernel/sys32.S278
-rw-r--r--arch/sparc/kernel/sys_solaris.c35
-rw-r--r--arch/sparc/kernel/sys_sparc.c513
-rw-r--r--arch/sparc/kernel/sys_sparc32.c256
-rw-r--r--arch/sparc/kernel/sys_sparc_32.c223
-rw-r--r--arch/sparc/kernel/sys_sparc_64.c647
-rw-r--r--arch/sparc/kernel/sys_sunos.c1210
-rw-r--r--arch/sparc/kernel/syscalls.S260
-rw-r--r--arch/sparc/kernel/sysfs.c320
-rw-r--r--arch/sparc/kernel/systbls.h103
-rw-r--r--arch/sparc/kernel/systbls_32.S (renamed from arch/sparc/kernel/systbls.S)139
-rw-r--r--arch/sparc/kernel/systbls_64.S169
-rw-r--r--arch/sparc/kernel/tadpole.c126
-rw-r--r--arch/sparc/kernel/tick14.c86
-rw-r--r--arch/sparc/kernel/time.c601
-rw-r--r--arch/sparc/kernel/time_32.c382
-rw-r--r--arch/sparc/kernel/time_64.c850
-rw-r--r--arch/sparc/kernel/trampoline_32.S (renamed from arch/sparc/kernel/trampoline.S)90
-rw-r--r--arch/sparc/kernel/trampoline_64.S411
-rw-r--r--arch/sparc/kernel/traps_32.c (renamed from arch/sparc/kernel/traps.c)85
-rw-r--r--arch/sparc/kernel/traps_64.c2796
-rw-r--r--arch/sparc/kernel/tsb.S582
-rw-r--r--arch/sparc/kernel/ttable_32.S417
-rw-r--r--arch/sparc/kernel/ttable_64.S272
-rw-r--r--arch/sparc/kernel/una_asm_32.S153
-rw-r--r--arch/sparc/kernel/una_asm_64.S146
-rw-r--r--arch/sparc/kernel/unaligned_32.c (renamed from arch/sparc/kernel/unaligned.c)289
-rw-r--r--arch/sparc/kernel/unaligned_64.c700
-rw-r--r--arch/sparc/kernel/utrap.S29
-rw-r--r--arch/sparc/kernel/vio.c463
-rw-r--r--arch/sparc/kernel/viohs.c824
-rw-r--r--arch/sparc/kernel/visemul.c898
-rw-r--r--arch/sparc/kernel/vmlinux.lds.S182
-rw-r--r--arch/sparc/kernel/windows.c12
-rw-r--r--arch/sparc/kernel/winfixup.S158
-rw-r--r--arch/sparc/kernel/wof.S93
-rw-r--r--arch/sparc/kernel/wuf.S90
172 files changed, 55632 insertions, 14104 deletions
diff --git a/arch/sparc/kernel/.gitignore b/arch/sparc/kernel/.gitignore
new file mode 100644
index 00000000000..c5f676c3c22
--- /dev/null
+++ b/arch/sparc/kernel/.gitignore
@@ -0,0 +1 @@
+vmlinux.lds
diff --git a/arch/sparc/kernel/Makefile b/arch/sparc/kernel/Makefile
index e795f282dec..7cf9c6ea3f1 100644
--- a/arch/sparc/kernel/Makefile
+++ b/arch/sparc/kernel/Makefile
@@ -1,30 +1,118 @@
-# $Id: Makefile,v 1.62 2000/12/15 00:41:17 davem Exp $
+
+#
# Makefile for the linux kernel.
#
-extra-y := head.o init_task.o vmlinux.lds
+asflags-y := -ansi
+ccflags-y := -Werror
+
+extra-y := head_$(BITS).o
-EXTRA_AFLAGS := -ansi
+# Undefine sparc when processing vmlinux.lds - it is used
+# And teach CPP we are doing $(BITS) builds (for this case)
+CPPFLAGS_vmlinux.lds := -Usparc -m$(BITS)
+extra-y += vmlinux.lds
-IRQ_OBJS := irq.o sun4m_irq.o sun4c_irq.o sun4d_irq.o
-obj-y := entry.o wof.o wuf.o etrap.o rtrap.o traps.o $(IRQ_OBJS) \
- process.o signal.o ioport.o setup.o idprom.o \
- sys_sparc.o sunos_asm.o systbls.o \
- time.o windows.o cpu.o devices.o sclow.o \
- tadpole.o tick14.o ptrace.o sys_solaris.o \
- unaligned.o muldiv.o semaphore.o prom.o of_device.o devres.o
+ifdef CONFIG_FUNCTION_TRACER
+# Do not profile debug and lowlevel utilities
+CFLAGS_REMOVE_ftrace.o := -pg
+CFLAGS_REMOVE_time_$(BITS).o := -pg
+CFLAGS_REMOVE_perf_event.o := -pg
+CFLAGS_REMOVE_pcr.o := -pg
+endif
-devres-y = ../../../kernel/irq/devres.o
+obj-$(CONFIG_SPARC32) += entry.o wof.o wuf.o
+obj-$(CONFIG_SPARC32) += etrap_32.o
+obj-$(CONFIG_SPARC32) += rtrap_32.o
+obj-y += traps_$(BITS).o
-obj-$(CONFIG_PCI) += pcic.o
-obj-$(CONFIG_SUN4) += sun4setup.o
-obj-$(CONFIG_SMP) += trampoline.o smp.o sun4m_smp.o sun4d_smp.o
-obj-$(CONFIG_SUN_AUXIO) += auxio.o
-obj-$(CONFIG_PCI) += ebus.o
-obj-$(CONFIG_SUN_PM) += apc.o pmc.o
-obj-$(CONFIG_MODULES) += module.o sparc_ksyms.o
-obj-$(CONFIG_SPARC_LED) += led.o
+# IRQ
+obj-y += irq_$(BITS).o
+obj-$(CONFIG_SPARC32) += sun4m_irq.o sun4d_irq.o
-ifdef CONFIG_SUNOS_EMUL
-obj-y += sys_sunos.o sunos_ioctl.o
-endif
+obj-y += process_$(BITS).o
+obj-y += signal_$(BITS).o
+obj-y += sigutil_$(BITS).o
+obj-$(CONFIG_SPARC32) += ioport.o
+obj-y += setup_$(BITS).o
+obj-y += idprom.o
+obj-y += sys_sparc_$(BITS).o
+obj-$(CONFIG_SPARC32) += systbls_32.o
+obj-y += time_$(BITS).o
+obj-$(CONFIG_SPARC32) += windows.o
+obj-y += cpu.o
+obj-$(CONFIG_SPARC32) += devices.o
+obj-y += ptrace_$(BITS).o
+obj-y += unaligned_$(BITS).o
+obj-y += una_asm_$(BITS).o
+obj-y += prom_common.o
+obj-y += prom_$(BITS).o
+obj-y += of_device_common.o
+obj-y += of_device_$(BITS).o
+obj-$(CONFIG_SPARC64) += prom_irqtrans.o
+
+obj-$(CONFIG_SPARC32) += leon_kernel.o
+obj-$(CONFIG_SPARC32) += leon_pmc.o
+
+obj-$(CONFIG_SPARC64) += reboot.o
+obj-$(CONFIG_SPARC64) += sysfs.o
+obj-$(CONFIG_SPARC64) += iommu.o
+obj-$(CONFIG_SPARC64) += central.o
+obj-$(CONFIG_SPARC64) += starfire.o
+obj-$(CONFIG_SPARC64) += power.o
+obj-$(CONFIG_SPARC64) += sbus.o
+obj-$(CONFIG_SPARC64) += ebus.o
+obj-$(CONFIG_SPARC64) += visemul.o
+obj-$(CONFIG_SPARC64) += hvapi.o
+obj-$(CONFIG_SPARC64) += sstate.o
+obj-$(CONFIG_SPARC64) += mdesc.o
+obj-$(CONFIG_SPARC64) += pcr.o
+obj-$(CONFIG_SPARC64) += nmi.o
+obj-$(CONFIG_SPARC64_SMP) += cpumap.o
+
+obj-y += dma.o
+
+obj-$(CONFIG_PCIC_PCI) += pcic.o
+obj-$(CONFIG_LEON_PCI) += leon_pci.o
+obj-$(CONFIG_SPARC_GRPCI2)+= leon_pci_grpci2.o
+obj-$(CONFIG_SPARC_GRPCI1)+= leon_pci_grpci1.o
+
+obj-$(CONFIG_SMP) += trampoline_$(BITS).o smp_$(BITS).o
+obj-$(CONFIG_SPARC32_SMP) += sun4m_smp.o sun4d_smp.o leon_smp.o
+obj-$(CONFIG_SPARC64_SMP) += hvtramp.o
+
+obj-y += auxio_$(BITS).o
+obj-$(CONFIG_SUN_PM) += apc.o pmc.o
+
+obj-$(CONFIG_MODULES) += module.o
+obj-$(CONFIG_MODULES) += sparc_ksyms_$(BITS).o
+obj-$(CONFIG_SPARC_LED) += led.o
+obj-$(CONFIG_KGDB) += kgdb_$(BITS).o
+
+
+obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o
+obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
+
+obj-$(CONFIG_EARLYFB) += btext.o
+obj-$(CONFIG_STACKTRACE) += stacktrace.o
+# sparc64 PCI
+obj-$(CONFIG_SPARC64_PCI) += pci.o pci_common.o psycho_common.o
+obj-$(CONFIG_SPARC64_PCI) += pci_psycho.o pci_sabre.o pci_schizo.o
+obj-$(CONFIG_SPARC64_PCI) += pci_sun4v.o pci_sun4v_asm.o pci_fire.o
+obj-$(CONFIG_SPARC64_PCI_MSI) += pci_msi.o
+
+obj-$(CONFIG_COMPAT) += sys32.o sys_sparc32.o signal32.o
+
+obj-$(CONFIG_US3_MC) += chmc.o
+
+obj-$(CONFIG_KPROBES) += kprobes.o
+obj-$(CONFIG_SUN_LDOMS) += ldc.o vio.o viohs.o ds.o
+
+obj-$(CONFIG_AUDIT) += audit.o
+audit--$(CONFIG_AUDIT) := compat_audit.o
+obj-$(CONFIG_COMPAT) += $(audit--y)
+
+pc--$(CONFIG_PERF_EVENTS) := perf_event.o
+obj-$(CONFIG_SPARC64) += $(pc--y)
+
+obj-$(CONFIG_SPARC64) += jump_label.o
diff --git a/arch/sparc/kernel/apc.c b/arch/sparc/kernel/apc.c
index d06a405ca71..eefda32b595 100644
--- a/arch/sparc/kernel/apc.c
+++ b/arch/sparc/kernel/apc.c
@@ -11,13 +11,16 @@
#include <linux/init.h>
#include <linux/miscdevice.h>
#include <linux/pm.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/module.h>
#include <asm/io.h>
-#include <asm/sbus.h>
#include <asm/oplib.h>
#include <asm/uaccess.h>
#include <asm/auxio.h>
#include <asm/apc.h>
+#include <asm/processor.h>
/* Debugging
*
@@ -28,11 +31,10 @@
#define APC_OBPNAME "power-management"
#define APC_DEVNAME "apc"
-volatile static u8 __iomem *regs;
-static int apc_regsize;
-static int apc_no_idle __initdata = 0;
+static u8 __iomem *regs;
+static int apc_no_idle = 0;
-#define apc_readb(offs) (sbus_readb(regs+offs))
+#define apc_readb(offs) (sbus_readb(regs+offs))
#define apc_writeb(val, offs) (sbus_writeb(val, regs+offs))
/* Specify "apc=noidle" on the kernel command line to
@@ -55,7 +57,7 @@ __setup("apc=", apc_setup);
* CPU idle callback function
* See .../arch/sparc/kernel/process.c
*/
-void apc_swift_idle(void)
+static void apc_swift_idle(void)
{
#ifdef APC_DEBUG_LED
set_auxio(0x00, AUXIO_LED);
@@ -68,9 +70,9 @@ void apc_swift_idle(void)
#endif
}
-static inline void apc_free(void)
+static inline void apc_free(struct platform_device *op)
{
- sbus_iounmap(regs, apc_regsize);
+ of_iounmap(&op->resource[0], regs, resource_size(&op->resource[0]));
}
static int apc_open(struct inode *inode, struct file *f)
@@ -83,16 +85,14 @@ static int apc_release(struct inode *inode, struct file *f)
return 0;
}
-static int apc_ioctl(struct inode *inode, struct file *f,
- unsigned int cmd, unsigned long __arg)
+static long apc_ioctl(struct file *f, unsigned int cmd, unsigned long __arg)
{
- __u8 inarg, __user *arg;
+ __u8 inarg, __user *arg = (__u8 __user *) __arg;
- arg = (__u8 __user *) __arg;
switch (cmd) {
case APCIOCGFANCTL:
if (put_user(apc_readb(APC_FANCTL_REG) & APC_REGMASK, arg))
- return -EFAULT;
+ return -EFAULT;
break;
case APCIOCGCPWR:
@@ -110,77 +110,87 @@ static int apc_ioctl(struct inode *inode, struct file *f,
return -EFAULT;
apc_writeb(inarg & APC_REGMASK, APC_FANCTL_REG);
break;
+
case APCIOCSCPWR:
if (get_user(inarg, arg))
return -EFAULT;
apc_writeb(inarg & APC_REGMASK, APC_CPOWER_REG);
break;
+
case APCIOCSBPORT:
if (get_user(inarg, arg))
return -EFAULT;
apc_writeb(inarg & APC_BPMASK, APC_BPORT_REG);
break;
+
default:
return -EINVAL;
- };
+ }
return 0;
}
static const struct file_operations apc_fops = {
- .ioctl = apc_ioctl,
- .open = apc_open,
- .release = apc_release,
+ .unlocked_ioctl = apc_ioctl,
+ .open = apc_open,
+ .release = apc_release,
+ .llseek = noop_llseek,
};
static struct miscdevice apc_miscdev = { APC_MINOR, APC_DEVNAME, &apc_fops };
-static int __init apc_probe(void)
+static int apc_probe(struct platform_device *op)
{
- struct sbus_bus *sbus = NULL;
- struct sbus_dev *sdev = NULL;
- int iTmp = 0;
-
- for_each_sbus(sbus) {
- for_each_sbusdev(sdev, sbus) {
- if (!strcmp(sdev->prom_name, APC_OBPNAME)) {
- goto sbus_done;
- }
- }
- }
-
-sbus_done:
- if (!sdev) {
- return -ENODEV;
- }
+ int err;
- apc_regsize = sdev->reg_addrs[0].reg_size;
- regs = sbus_ioremap(&sdev->resource[0], 0,
- apc_regsize, APC_OBPNAME);
- if(!regs) {
+ regs = of_ioremap(&op->resource[0], 0,
+ resource_size(&op->resource[0]), APC_OBPNAME);
+ if (!regs) {
printk(KERN_ERR "%s: unable to map registers\n", APC_DEVNAME);
return -ENODEV;
}
- iTmp = misc_register(&apc_miscdev);
- if (iTmp != 0) {
+ err = misc_register(&apc_miscdev);
+ if (err) {
printk(KERN_ERR "%s: unable to register device\n", APC_DEVNAME);
- apc_free();
+ apc_free(op);
return -ENODEV;
}
/* Assign power management IDLE handler */
- if(!apc_no_idle)
- pm_idle = apc_swift_idle;
+ if (!apc_no_idle)
+ sparc_idle = apc_swift_idle;
printk(KERN_INFO "%s: power management initialized%s\n",
- APC_DEVNAME, apc_no_idle ? " (CPU idle disabled)" : "");
+ APC_DEVNAME, apc_no_idle ? " (CPU idle disabled)" : "");
+
return 0;
}
+static struct of_device_id apc_match[] = {
+ {
+ .name = APC_OBPNAME,
+ },
+ {},
+};
+MODULE_DEVICE_TABLE(of, apc_match);
+
+static struct platform_driver apc_driver = {
+ .driver = {
+ .name = "apc",
+ .owner = THIS_MODULE,
+ .of_match_table = apc_match,
+ },
+ .probe = apc_probe,
+};
+
+static int __init apc_init(void)
+{
+ return platform_driver_register(&apc_driver);
+}
+
/* This driver is not critical to the boot process
* and is easiest to ioremap when SBus is already
* initialized, so we install ourselves thusly:
*/
-__initcall(apc_probe);
-
+__initcall(apc_init);
diff --git a/arch/sparc/kernel/asm-offsets.c b/arch/sparc/kernel/asm-offsets.c
index 6773ed76e41..f76389a3234 100644
--- a/arch/sparc/kernel/asm-offsets.c
+++ b/arch/sparc/kernel/asm-offsets.c
@@ -12,33 +12,47 @@
#include <linux/sched.h>
// #include <linux/mm.h>
+#include <linux/kbuild.h>
-#define DEFINE(sym, val) \
- asm volatile("\n->" #sym " %0 " #val : : "i" (val))
+#include <asm/hibernate.h>
-#define BLANK() asm volatile("\n->" : : )
+#ifdef CONFIG_SPARC32
+int sparc32_foo(void)
+{
+ DEFINE(AOFF_thread_fork_kpsr,
+ offsetof(struct thread_struct, fork_kpsr));
+ return 0;
+}
+#else
+int sparc64_foo(void)
+{
+#ifdef CONFIG_HIBERNATION
+ BLANK();
+ OFFSET(SC_REG_FP, saved_context, fp);
+ OFFSET(SC_REG_CWP, saved_context, cwp);
+ OFFSET(SC_REG_WSTATE, saved_context, wstate);
+
+ OFFSET(SC_REG_TICK, saved_context, tick);
+ OFFSET(SC_REG_PSTATE, saved_context, pstate);
+
+ OFFSET(SC_REG_G4, saved_context, g4);
+ OFFSET(SC_REG_G5, saved_context, g5);
+ OFFSET(SC_REG_G6, saved_context, g6);
+#endif
+ return 0;
+}
+#endif
int foo(void)
{
- DEFINE(AOFF_task_thread, offsetof(struct task_struct, thread));
BLANK();
- /* XXX This is the stuff for sclow.S, kill it. */
- DEFINE(AOFF_task_pid, offsetof(struct task_struct, pid));
- DEFINE(AOFF_task_uid, offsetof(struct task_struct, uid));
- DEFINE(AOFF_task_gid, offsetof(struct task_struct, gid));
- DEFINE(AOFF_task_euid, offsetof(struct task_struct, euid));
- DEFINE(AOFF_task_egid, offsetof(struct task_struct, egid));
- /* DEFINE(THREAD_INFO, offsetof(struct task_struct, stack)); */
- DEFINE(ASIZ_task_uid, sizeof(current->uid));
- DEFINE(ASIZ_task_gid, sizeof(current->gid));
- DEFINE(ASIZ_task_euid, sizeof(current->euid));
- DEFINE(ASIZ_task_egid, sizeof(current->egid));
- BLANK();
- DEFINE(AOFF_thread_fork_kpsr,
- offsetof(struct thread_struct, fork_kpsr));
+ DEFINE(AOFF_task_thread, offsetof(struct task_struct, thread));
BLANK();
DEFINE(AOFF_mm_context, offsetof(struct mm_struct, context));
+ BLANK();
+ DEFINE(VMA_VM_MM, offsetof(struct vm_area_struct, vm_mm));
/* DEFINE(NUM_USER_SEGMENTS, TASK_SIZE>>28); */
return 0;
}
+
diff --git a/arch/sparc/kernel/audit.c b/arch/sparc/kernel/audit.c
new file mode 100644
index 00000000000..24361b494a9
--- /dev/null
+++ b/arch/sparc/kernel/audit.c
@@ -0,0 +1,79 @@
+#include <linux/init.h>
+#include <linux/types.h>
+#include <linux/audit.h>
+#include <asm/unistd.h>
+
+#include "kernel.h"
+
+static unsigned dir_class[] = {
+#include <asm-generic/audit_dir_write.h>
+~0U
+};
+
+static unsigned read_class[] = {
+#include <asm-generic/audit_read.h>
+~0U
+};
+
+static unsigned write_class[] = {
+#include <asm-generic/audit_write.h>
+~0U
+};
+
+static unsigned chattr_class[] = {
+#include <asm-generic/audit_change_attr.h>
+~0U
+};
+
+static unsigned signal_class[] = {
+#include <asm-generic/audit_signal.h>
+~0U
+};
+
+int audit_classify_arch(int arch)
+{
+#ifdef CONFIG_COMPAT
+ if (arch == AUDIT_ARCH_SPARC)
+ return 1;
+#endif
+ return 0;
+}
+
+int audit_classify_syscall(int abi, unsigned syscall)
+{
+#ifdef CONFIG_COMPAT
+ if (abi == AUDIT_ARCH_SPARC)
+ return sparc32_classify_syscall(syscall);
+#endif
+ switch(syscall) {
+ case __NR_open:
+ return 2;
+ case __NR_openat:
+ return 3;
+ case __NR_socketcall:
+ return 4;
+ case __NR_execve:
+ return 5;
+ default:
+ return 0;
+ }
+}
+
+static int __init audit_classes_init(void)
+{
+#ifdef CONFIG_COMPAT
+ audit_register_class(AUDIT_CLASS_WRITE_32, sparc32_write_class);
+ audit_register_class(AUDIT_CLASS_READ_32, sparc32_read_class);
+ audit_register_class(AUDIT_CLASS_DIR_WRITE_32, sparc32_dir_class);
+ audit_register_class(AUDIT_CLASS_CHATTR_32, sparc32_chattr_class);
+ audit_register_class(AUDIT_CLASS_SIGNAL_32, sparc32_signal_class);
+#endif
+ audit_register_class(AUDIT_CLASS_WRITE, write_class);
+ audit_register_class(AUDIT_CLASS_READ, read_class);
+ audit_register_class(AUDIT_CLASS_DIR_WRITE, dir_class);
+ audit_register_class(AUDIT_CLASS_CHATTR, chattr_class);
+ audit_register_class(AUDIT_CLASS_SIGNAL, signal_class);
+ return 0;
+}
+
+__initcall(audit_classes_init);
diff --git a/arch/sparc/kernel/auxio.c b/arch/sparc/kernel/auxio_32.c
index baf4ed3fb0f..ae88c223e4d 100644
--- a/arch/sparc/kernel/auxio.c
+++ b/arch/sparc/kernel/auxio_32.c
@@ -6,10 +6,17 @@
#include <linux/stddef.h>
#include <linux/init.h>
#include <linux/spinlock.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/export.h>
+
#include <asm/oplib.h>
#include <asm/io.h>
#include <asm/auxio.h>
#include <asm/string.h> /* memset(), Linux has no bzero() */
+#include <asm/cpu_type.h>
+
+#include "kernel.h"
/* Probe and map in the Auxiliary I/O register */
@@ -21,13 +28,13 @@ static DEFINE_SPINLOCK(auxio_lock);
void __init auxio_probe(void)
{
- int node, auxio_nd;
+ phandle node, auxio_nd;
struct linux_prom_registers auxregs[1];
struct resource r;
switch (sparc_cpu_model) {
+ case sparc_leon:
case sun4d:
- case sun4:
return;
default:
break;
@@ -59,10 +66,9 @@ void __init auxio_probe(void)
r.flags = auxregs[0].which_io & 0xF;
r.start = auxregs[0].phys_addr;
r.end = auxregs[0].phys_addr + auxregs[0].reg_size - 1;
- auxio_register = sbus_ioremap(&r, 0, auxregs[0].reg_size, "auxio");
- /* Fix the address on sun4m and sun4c. */
- if((((unsigned long) auxregs[0].phys_addr) & 3) == 3 ||
- sparc_cpu_model == sun4c)
+ auxio_register = of_ioremap(&r, 0, auxregs[0].reg_size, "auxio");
+ /* Fix the address on sun4m. */
+ if ((((unsigned long) auxregs[0].phys_addr) & 3) == 3)
auxio_register += (3 - ((unsigned long)auxio_register & 3));
set_auxio(AUXIO_LED, 0);
@@ -74,18 +80,14 @@ unsigned char get_auxio(void)
return sbus_readb(auxio_register);
return 0;
}
+EXPORT_SYMBOL(get_auxio);
void set_auxio(unsigned char bits_on, unsigned char bits_off)
{
unsigned char regval;
unsigned long flags;
spin_lock_irqsave(&auxio_lock, flags);
- switch(sparc_cpu_model) {
- case sun4c:
- regval = sbus_readb(auxio_register);
- sbus_writeb(((regval | bits_on) & ~bits_off) | AUXIO_ORMEIN,
- auxio_register);
- break;
+ switch (sparc_cpu_model) {
case sun4m:
if(!auxio_register)
break; /* VME chassis sun4m, no auxio. */
@@ -97,19 +99,19 @@ void set_auxio(unsigned char bits_on, unsigned char bits_off)
break;
default:
panic("Can't set AUXIO register on this machine.");
- };
+ }
spin_unlock_irqrestore(&auxio_lock, flags);
}
-
+EXPORT_SYMBOL(set_auxio);
/* sun4m power control register (AUXIO2) */
-volatile unsigned char * auxio_power_register = NULL;
+volatile u8 __iomem *auxio_power_register = NULL;
void __init auxio_power_probe(void)
{
struct linux_prom_registers regs;
- int node;
+ phandle node;
struct resource r;
/* Attempt to find the sun4m power control node. */
@@ -117,7 +119,7 @@ void __init auxio_power_probe(void)
node = prom_searchsiblings(node, "obio");
node = prom_getchild(node);
node = prom_searchsiblings(node, "power");
- if (node == 0 || node == -1)
+ if (node == 0 || (s32)node == -1)
return;
/* Map the power control register. */
@@ -128,8 +130,8 @@ void __init auxio_power_probe(void)
r.flags = regs.which_io & 0xF;
r.start = regs.phys_addr;
r.end = regs.phys_addr + regs.reg_size - 1;
- auxio_power_register = (unsigned char *) sbus_ioremap(&r, 0,
- regs.reg_size, "auxpower");
+ auxio_power_register =
+ (u8 __iomem *)of_ioremap(&r, 0, regs.reg_size, "auxpower");
/* Display a quick message on the console. */
if (auxio_power_register)
diff --git a/arch/sparc/kernel/auxio_64.c b/arch/sparc/kernel/auxio_64.c
new file mode 100644
index 00000000000..86e55778e4a
--- /dev/null
+++ b/arch/sparc/kernel/auxio_64.c
@@ -0,0 +1,152 @@
+/* auxio.c: Probing for the Sparc AUXIO register at boot time.
+ *
+ * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
+ *
+ * Refactoring for unified NCR/PCIO support 2002 Eric Brower (ebrower@usa.net)
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/ioport.h>
+#include <linux/of_device.h>
+
+#include <asm/prom.h>
+#include <asm/io.h>
+#include <asm/auxio.h>
+
+void __iomem *auxio_register = NULL;
+EXPORT_SYMBOL(auxio_register);
+
+enum auxio_type {
+ AUXIO_TYPE_NODEV,
+ AUXIO_TYPE_SBUS,
+ AUXIO_TYPE_EBUS
+};
+
+static enum auxio_type auxio_devtype = AUXIO_TYPE_NODEV;
+static DEFINE_SPINLOCK(auxio_lock);
+
+static void __auxio_rmw(u8 bits_on, u8 bits_off, int ebus)
+{
+ if (auxio_register) {
+ unsigned long flags;
+ u8 regval, newval;
+
+ spin_lock_irqsave(&auxio_lock, flags);
+
+ regval = (ebus ?
+ (u8) readl(auxio_register) :
+ sbus_readb(auxio_register));
+ newval = regval | bits_on;
+ newval &= ~bits_off;
+ if (!ebus)
+ newval &= ~AUXIO_AUX1_MASK;
+ if (ebus)
+ writel((u32) newval, auxio_register);
+ else
+ sbus_writeb(newval, auxio_register);
+
+ spin_unlock_irqrestore(&auxio_lock, flags);
+ }
+}
+
+static void __auxio_set_bit(u8 bit, int on, int ebus)
+{
+ u8 bits_on = (ebus ? AUXIO_PCIO_LED : AUXIO_AUX1_LED);
+ u8 bits_off = 0;
+
+ if (!on) {
+ u8 tmp = bits_off;
+ bits_off = bits_on;
+ bits_on = tmp;
+ }
+ __auxio_rmw(bits_on, bits_off, ebus);
+}
+
+void auxio_set_led(int on)
+{
+ int ebus = auxio_devtype == AUXIO_TYPE_EBUS;
+ u8 bit;
+
+ bit = (ebus ? AUXIO_PCIO_LED : AUXIO_AUX1_LED);
+ __auxio_set_bit(bit, on, ebus);
+}
+EXPORT_SYMBOL(auxio_set_led);
+
+static void __auxio_sbus_set_lte(int on)
+{
+ __auxio_set_bit(AUXIO_AUX1_LTE, on, 0);
+}
+
+void auxio_set_lte(int on)
+{
+ switch(auxio_devtype) {
+ case AUXIO_TYPE_SBUS:
+ __auxio_sbus_set_lte(on);
+ break;
+ case AUXIO_TYPE_EBUS:
+ /* FALL-THROUGH */
+ default:
+ break;
+ }
+}
+EXPORT_SYMBOL(auxio_set_lte);
+
+static const struct of_device_id auxio_match[] = {
+ {
+ .name = "auxio",
+ },
+ {},
+};
+
+MODULE_DEVICE_TABLE(of, auxio_match);
+
+static int auxio_probe(struct platform_device *dev)
+{
+ struct device_node *dp = dev->dev.of_node;
+ unsigned long size;
+
+ if (!strcmp(dp->parent->name, "ebus")) {
+ auxio_devtype = AUXIO_TYPE_EBUS;
+ size = sizeof(u32);
+ } else if (!strcmp(dp->parent->name, "sbus")) {
+ auxio_devtype = AUXIO_TYPE_SBUS;
+ size = 1;
+ } else {
+ printk("auxio: Unknown parent bus type [%s]\n",
+ dp->parent->name);
+ return -ENODEV;
+ }
+ auxio_register = of_ioremap(&dev->resource[0], 0, size, "auxio");
+ if (!auxio_register)
+ return -ENODEV;
+
+ printk(KERN_INFO "AUXIO: Found device at %s\n",
+ dp->full_name);
+
+ if (auxio_devtype == AUXIO_TYPE_EBUS)
+ auxio_set_led(AUXIO_LED_ON);
+
+ return 0;
+}
+
+static struct platform_driver auxio_driver = {
+ .probe = auxio_probe,
+ .driver = {
+ .name = "auxio",
+ .owner = THIS_MODULE,
+ .of_match_table = auxio_match,
+ },
+};
+
+static int __init auxio_init(void)
+{
+ return platform_driver_register(&auxio_driver);
+}
+
+/* Must be after subsys_initcall() so that busses are probed. Must
+ * be before device_initcall() because things like the floppy driver
+ * need to use the AUXIO register.
+ */
+fs_initcall(auxio_init);
diff --git a/arch/sparc/kernel/btext.c b/arch/sparc/kernel/btext.c
new file mode 100644
index 00000000000..987f7ec497c
--- /dev/null
+++ b/arch/sparc/kernel/btext.c
@@ -0,0 +1,672 @@
+/*
+ * Procedures for drawing on the screen early on in the boot process.
+ *
+ * Benjamin Herrenschmidt <benh@kernel.crashing.org>
+ */
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/console.h>
+
+#include <asm/btext.h>
+#include <asm/oplib.h>
+#include <asm/io.h>
+
+#define NO_SCROLL
+
+#ifndef NO_SCROLL
+static void scrollscreen(void);
+#endif
+
+static void draw_byte(unsigned char c, long locX, long locY);
+static void draw_byte_32(unsigned char *bits, unsigned int *base, int rb);
+static void draw_byte_16(unsigned char *bits, unsigned int *base, int rb);
+static void draw_byte_8(unsigned char *bits, unsigned int *base, int rb);
+
+#define __force_data __attribute__((__section__(".data")))
+
+static int g_loc_X __force_data;
+static int g_loc_Y __force_data;
+static int g_max_loc_X __force_data;
+static int g_max_loc_Y __force_data;
+
+static int dispDeviceRowBytes __force_data;
+static int dispDeviceDepth __force_data;
+static int dispDeviceRect[4] __force_data;
+static unsigned char *dispDeviceBase __force_data;
+
+#define cmapsz (16*256)
+
+static unsigned char vga_font[cmapsz];
+
+static int __init btext_initialize(phandle node)
+{
+ unsigned int width, height, depth, pitch;
+ unsigned long address = 0;
+ u32 prop;
+
+ if (prom_getproperty(node, "width", (char *)&width, 4) < 0)
+ return -EINVAL;
+ if (prom_getproperty(node, "height", (char *)&height, 4) < 0)
+ return -EINVAL;
+ if (prom_getproperty(node, "depth", (char *)&depth, 4) < 0)
+ return -EINVAL;
+ pitch = width * ((depth + 7) / 8);
+
+ if (prom_getproperty(node, "linebytes", (char *)&prop, 4) >= 0 &&
+ prop != 0xffffffffu)
+ pitch = prop;
+
+ if (pitch == 1)
+ pitch = 0x1000;
+
+ if (prom_getproperty(node, "address", (char *)&prop, 4) >= 0)
+ address = prop;
+
+ /* FIXME: Add support for PCI reg properties. Right now, only
+ * reliable on macs
+ */
+ if (address == 0)
+ return -EINVAL;
+
+ g_loc_X = 0;
+ g_loc_Y = 0;
+ g_max_loc_X = width / 8;
+ g_max_loc_Y = height / 16;
+ dispDeviceBase = (unsigned char *)address;
+ dispDeviceRowBytes = pitch;
+ dispDeviceDepth = depth == 15 ? 16 : depth;
+ dispDeviceRect[0] = dispDeviceRect[1] = 0;
+ dispDeviceRect[2] = width;
+ dispDeviceRect[3] = height;
+
+ return 0;
+}
+
+/* Calc the base address of a given point (x,y) */
+static unsigned char * calc_base(int x, int y)
+{
+ unsigned char *base = dispDeviceBase;
+
+ base += (x + dispDeviceRect[0]) * (dispDeviceDepth >> 3);
+ base += (y + dispDeviceRect[1]) * dispDeviceRowBytes;
+ return base;
+}
+
+static void btext_clearscreen(void)
+{
+ unsigned int *base = (unsigned int *)calc_base(0, 0);
+ unsigned long width = ((dispDeviceRect[2] - dispDeviceRect[0]) *
+ (dispDeviceDepth >> 3)) >> 2;
+ int i,j;
+
+ for (i=0; i<(dispDeviceRect[3] - dispDeviceRect[1]); i++)
+ {
+ unsigned int *ptr = base;
+ for(j=width; j; --j)
+ *(ptr++) = 0;
+ base += (dispDeviceRowBytes >> 2);
+ }
+}
+
+#ifndef NO_SCROLL
+static void scrollscreen(void)
+{
+ unsigned int *src = (unsigned int *)calc_base(0,16);
+ unsigned int *dst = (unsigned int *)calc_base(0,0);
+ unsigned long width = ((dispDeviceRect[2] - dispDeviceRect[0]) *
+ (dispDeviceDepth >> 3)) >> 2;
+ int i,j;
+
+ for (i=0; i<(dispDeviceRect[3] - dispDeviceRect[1] - 16); i++)
+ {
+ unsigned int *src_ptr = src;
+ unsigned int *dst_ptr = dst;
+ for(j=width; j; --j)
+ *(dst_ptr++) = *(src_ptr++);
+ src += (dispDeviceRowBytes >> 2);
+ dst += (dispDeviceRowBytes >> 2);
+ }
+ for (i=0; i<16; i++)
+ {
+ unsigned int *dst_ptr = dst;
+ for(j=width; j; --j)
+ *(dst_ptr++) = 0;
+ dst += (dispDeviceRowBytes >> 2);
+ }
+}
+#endif /* ndef NO_SCROLL */
+
+static void btext_drawchar(char c)
+{
+ int cline = 0;
+#ifdef NO_SCROLL
+ int x;
+#endif
+ switch (c) {
+ case '\b':
+ if (g_loc_X > 0)
+ --g_loc_X;
+ break;
+ case '\t':
+ g_loc_X = (g_loc_X & -8) + 8;
+ break;
+ case '\r':
+ g_loc_X = 0;
+ break;
+ case '\n':
+ g_loc_X = 0;
+ g_loc_Y++;
+ cline = 1;
+ break;
+ default:
+ draw_byte(c, g_loc_X++, g_loc_Y);
+ }
+ if (g_loc_X >= g_max_loc_X) {
+ g_loc_X = 0;
+ g_loc_Y++;
+ cline = 1;
+ }
+#ifndef NO_SCROLL
+ while (g_loc_Y >= g_max_loc_Y) {
+ scrollscreen();
+ g_loc_Y--;
+ }
+#else
+ /* wrap around from bottom to top of screen so we don't
+ waste time scrolling each line. -- paulus. */
+ if (g_loc_Y >= g_max_loc_Y)
+ g_loc_Y = 0;
+ if (cline) {
+ for (x = 0; x < g_max_loc_X; ++x)
+ draw_byte(' ', x, g_loc_Y);
+ }
+#endif
+}
+
+static void btext_drawtext(const char *c, unsigned int len)
+{
+ while (len--)
+ btext_drawchar(*c++);
+}
+
+static void draw_byte(unsigned char c, long locX, long locY)
+{
+ unsigned char *base = calc_base(locX << 3, locY << 4);
+ unsigned char *font = &vga_font[((unsigned int)c) * 16];
+ int rb = dispDeviceRowBytes;
+
+ switch(dispDeviceDepth) {
+ case 24:
+ case 32:
+ draw_byte_32(font, (unsigned int *)base, rb);
+ break;
+ case 15:
+ case 16:
+ draw_byte_16(font, (unsigned int *)base, rb);
+ break;
+ case 8:
+ draw_byte_8(font, (unsigned int *)base, rb);
+ break;
+ }
+}
+
+static unsigned int expand_bits_8[16] = {
+ 0x00000000,
+ 0x000000ff,
+ 0x0000ff00,
+ 0x0000ffff,
+ 0x00ff0000,
+ 0x00ff00ff,
+ 0x00ffff00,
+ 0x00ffffff,
+ 0xff000000,
+ 0xff0000ff,
+ 0xff00ff00,
+ 0xff00ffff,
+ 0xffff0000,
+ 0xffff00ff,
+ 0xffffff00,
+ 0xffffffff
+};
+
+static unsigned int expand_bits_16[4] = {
+ 0x00000000,
+ 0x0000ffff,
+ 0xffff0000,
+ 0xffffffff
+};
+
+
+static void draw_byte_32(unsigned char *font, unsigned int *base, int rb)
+{
+ int l, bits;
+ int fg = 0xFFFFFFFFUL;
+ int bg = 0x00000000UL;
+
+ for (l = 0; l < 16; ++l)
+ {
+ bits = *font++;
+ base[0] = (-(bits >> 7) & fg) ^ bg;
+ base[1] = (-((bits >> 6) & 1) & fg) ^ bg;
+ base[2] = (-((bits >> 5) & 1) & fg) ^ bg;
+ base[3] = (-((bits >> 4) & 1) & fg) ^ bg;
+ base[4] = (-((bits >> 3) & 1) & fg) ^ bg;
+ base[5] = (-((bits >> 2) & 1) & fg) ^ bg;
+ base[6] = (-((bits >> 1) & 1) & fg) ^ bg;
+ base[7] = (-(bits & 1) & fg) ^ bg;
+ base = (unsigned int *) ((char *)base + rb);
+ }
+}
+
+static void draw_byte_16(unsigned char *font, unsigned int *base, int rb)
+{
+ int l, bits;
+ int fg = 0xFFFFFFFFUL;
+ int bg = 0x00000000UL;
+ unsigned int *eb = (int *)expand_bits_16;
+
+ for (l = 0; l < 16; ++l)
+ {
+ bits = *font++;
+ base[0] = (eb[bits >> 6] & fg) ^ bg;
+ base[1] = (eb[(bits >> 4) & 3] & fg) ^ bg;
+ base[2] = (eb[(bits >> 2) & 3] & fg) ^ bg;
+ base[3] = (eb[bits & 3] & fg) ^ bg;
+ base = (unsigned int *) ((char *)base + rb);
+ }
+}
+
+static void draw_byte_8(unsigned char *font, unsigned int *base, int rb)
+{
+ int l, bits;
+ int fg = 0x0F0F0F0FUL;
+ int bg = 0x00000000UL;
+ unsigned int *eb = (int *)expand_bits_8;
+
+ for (l = 0; l < 16; ++l)
+ {
+ bits = *font++;
+ base[0] = (eb[bits >> 4] & fg) ^ bg;
+ base[1] = (eb[bits & 0xf] & fg) ^ bg;
+ base = (unsigned int *) ((char *)base + rb);
+ }
+}
+
+static void btext_console_write(struct console *con, const char *s,
+ unsigned int n)
+{
+ btext_drawtext(s, n);
+}
+
+static struct console btext_console = {
+ .name = "btext",
+ .write = btext_console_write,
+ .flags = CON_PRINTBUFFER | CON_ENABLED | CON_BOOT | CON_ANYTIME,
+ .index = 0,
+};
+
+int __init btext_find_display(void)
+{
+ phandle node;
+ char type[32];
+ int ret;
+
+ node = prom_inst2pkg(prom_stdout);
+ if (prom_getproperty(node, "device_type", type, 32) < 0)
+ return -ENODEV;
+ if (strcmp(type, "display"))
+ return -ENODEV;
+
+ ret = btext_initialize(node);
+ if (!ret) {
+ btext_clearscreen();
+ register_console(&btext_console);
+ }
+ return ret;
+}
+
+static unsigned char vga_font[cmapsz] = {
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7e, 0x81, 0xa5, 0x81, 0x81, 0xbd,
+0x99, 0x81, 0x81, 0x7e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7e, 0xff,
+0xdb, 0xff, 0xff, 0xc3, 0xe7, 0xff, 0xff, 0x7e, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x6c, 0xfe, 0xfe, 0xfe, 0xfe, 0x7c, 0x38, 0x10,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x38, 0x7c, 0xfe,
+0x7c, 0x38, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18,
+0x3c, 0x3c, 0xe7, 0xe7, 0xe7, 0x18, 0x18, 0x3c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x18, 0x3c, 0x7e, 0xff, 0xff, 0x7e, 0x18, 0x18, 0x3c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x3c,
+0x3c, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0xff, 0xff,
+0xff, 0xff, 0xe7, 0xc3, 0xc3, 0xe7, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x3c, 0x66, 0x42, 0x42, 0x66, 0x3c, 0x00,
+0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xc3, 0x99, 0xbd,
+0xbd, 0x99, 0xc3, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00, 0x00, 0x1e, 0x0e,
+0x1a, 0x32, 0x78, 0xcc, 0xcc, 0xcc, 0xcc, 0x78, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x3c, 0x66, 0x66, 0x66, 0x66, 0x3c, 0x18, 0x7e, 0x18, 0x18,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3f, 0x33, 0x3f, 0x30, 0x30, 0x30,
+0x30, 0x70, 0xf0, 0xe0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7f, 0x63,
+0x7f, 0x63, 0x63, 0x63, 0x63, 0x67, 0xe7, 0xe6, 0xc0, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x18, 0x18, 0xdb, 0x3c, 0xe7, 0x3c, 0xdb, 0x18, 0x18,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfe, 0xf8,
+0xf0, 0xe0, 0xc0, 0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x06, 0x0e,
+0x1e, 0x3e, 0xfe, 0x3e, 0x1e, 0x0e, 0x06, 0x02, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x18, 0x3c, 0x7e, 0x18, 0x18, 0x18, 0x7e, 0x3c, 0x18, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x66, 0x66, 0x66, 0x66, 0x66, 0x66,
+0x66, 0x00, 0x66, 0x66, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7f, 0xdb,
+0xdb, 0xdb, 0x7b, 0x1b, 0x1b, 0x1b, 0x1b, 0x1b, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x7c, 0xc6, 0x60, 0x38, 0x6c, 0xc6, 0xc6, 0x6c, 0x38, 0x0c, 0xc6,
+0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0xfe, 0xfe, 0xfe, 0xfe, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x3c,
+0x7e, 0x18, 0x18, 0x18, 0x7e, 0x3c, 0x18, 0x7e, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x18, 0x3c, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x7e, 0x3c, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x18, 0x0c, 0xfe, 0x0c, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x30, 0x60, 0xfe, 0x60, 0x30, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc0, 0xc0,
+0xc0, 0xfe, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x24, 0x66, 0xff, 0x66, 0x24, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x10, 0x38, 0x38, 0x7c, 0x7c, 0xfe, 0xfe, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0xfe, 0x7c, 0x7c,
+0x38, 0x38, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x18, 0x3c, 0x3c, 0x3c, 0x18, 0x18, 0x18, 0x00, 0x18, 0x18,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x66, 0x66, 0x66, 0x24, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x6c,
+0x6c, 0xfe, 0x6c, 0x6c, 0x6c, 0xfe, 0x6c, 0x6c, 0x00, 0x00, 0x00, 0x00,
+0x18, 0x18, 0x7c, 0xc6, 0xc2, 0xc0, 0x7c, 0x06, 0x06, 0x86, 0xc6, 0x7c,
+0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc2, 0xc6, 0x0c, 0x18,
+0x30, 0x60, 0xc6, 0x86, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x38, 0x6c,
+0x6c, 0x38, 0x76, 0xdc, 0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x30, 0x30, 0x30, 0x60, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0c, 0x18, 0x30, 0x30, 0x30, 0x30,
+0x30, 0x30, 0x18, 0x0c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x30, 0x18,
+0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x18, 0x30, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x66, 0x3c, 0xff, 0x3c, 0x66, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x7e,
+0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0x30, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7e, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x02, 0x06, 0x0c, 0x18, 0x30, 0x60, 0xc0, 0x80, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xce, 0xde, 0xf6, 0xe6, 0xc6, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x38, 0x78, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x7e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6,
+0x06, 0x0c, 0x18, 0x30, 0x60, 0xc0, 0xc6, 0xfe, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x7c, 0xc6, 0x06, 0x06, 0x3c, 0x06, 0x06, 0x06, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0c, 0x1c, 0x3c, 0x6c, 0xcc, 0xfe,
+0x0c, 0x0c, 0x0c, 0x1e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0xc0,
+0xc0, 0xc0, 0xfc, 0x06, 0x06, 0x06, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x38, 0x60, 0xc0, 0xc0, 0xfc, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0xc6, 0x06, 0x06, 0x0c, 0x18,
+0x30, 0x30, 0x30, 0x30, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6,
+0xc6, 0xc6, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0x7e, 0x06, 0x06, 0x06, 0x0c, 0x78,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x00, 0x00,
+0x00, 0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x18, 0x18, 0x00, 0x00, 0x00, 0x18, 0x18, 0x30, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x06, 0x0c, 0x18, 0x30, 0x60, 0x30, 0x18, 0x0c, 0x06,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7e, 0x00, 0x00,
+0x7e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x60,
+0x30, 0x18, 0x0c, 0x06, 0x0c, 0x18, 0x30, 0x60, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x7c, 0xc6, 0xc6, 0x0c, 0x18, 0x18, 0x18, 0x00, 0x18, 0x18,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xde, 0xde,
+0xde, 0xdc, 0xc0, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x38,
+0x6c, 0xc6, 0xc6, 0xfe, 0xc6, 0xc6, 0xc6, 0xc6, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xfc, 0x66, 0x66, 0x66, 0x7c, 0x66, 0x66, 0x66, 0x66, 0xfc,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3c, 0x66, 0xc2, 0xc0, 0xc0, 0xc0,
+0xc0, 0xc2, 0x66, 0x3c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xf8, 0x6c,
+0x66, 0x66, 0x66, 0x66, 0x66, 0x66, 0x6c, 0xf8, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xfe, 0x66, 0x62, 0x68, 0x78, 0x68, 0x60, 0x62, 0x66, 0xfe,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0x66, 0x62, 0x68, 0x78, 0x68,
+0x60, 0x60, 0x60, 0xf0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3c, 0x66,
+0xc2, 0xc0, 0xc0, 0xde, 0xc6, 0xc6, 0x66, 0x3a, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xc6, 0xc6, 0xc6, 0xc6, 0xfe, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3c, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x3c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1e, 0x0c,
+0x0c, 0x0c, 0x0c, 0x0c, 0xcc, 0xcc, 0xcc, 0x78, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xe6, 0x66, 0x66, 0x6c, 0x78, 0x78, 0x6c, 0x66, 0x66, 0xe6,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xf0, 0x60, 0x60, 0x60, 0x60, 0x60,
+0x60, 0x62, 0x66, 0xfe, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0xe7,
+0xff, 0xff, 0xdb, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xc6, 0xe6, 0xf6, 0xfe, 0xde, 0xce, 0xc6, 0xc6, 0xc6, 0xc6,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6,
+0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfc, 0x66,
+0x66, 0x66, 0x7c, 0x60, 0x60, 0x60, 0x60, 0xf0, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xd6, 0xde, 0x7c,
+0x0c, 0x0e, 0x00, 0x00, 0x00, 0x00, 0xfc, 0x66, 0x66, 0x66, 0x7c, 0x6c,
+0x66, 0x66, 0x66, 0xe6, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6,
+0xc6, 0x60, 0x38, 0x0c, 0x06, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xff, 0xdb, 0x99, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6,
+0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0xc3,
+0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0x66, 0x3c, 0x18, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xdb, 0xdb, 0xff, 0x66, 0x66,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0xc3, 0x66, 0x3c, 0x18, 0x18,
+0x3c, 0x66, 0xc3, 0xc3, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0xc3,
+0xc3, 0x66, 0x3c, 0x18, 0x18, 0x18, 0x18, 0x3c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xff, 0xc3, 0x86, 0x0c, 0x18, 0x30, 0x60, 0xc1, 0xc3, 0xff,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3c, 0x30, 0x30, 0x30, 0x30, 0x30,
+0x30, 0x30, 0x30, 0x3c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80,
+0xc0, 0xe0, 0x70, 0x38, 0x1c, 0x0e, 0x06, 0x02, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x3c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x3c,
+0x00, 0x00, 0x00, 0x00, 0x10, 0x38, 0x6c, 0xc6, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0x00, 0x00,
+0x30, 0x30, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x78, 0x0c, 0x7c,
+0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xe0, 0x60,
+0x60, 0x78, 0x6c, 0x66, 0x66, 0x66, 0x66, 0x7c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6, 0xc0, 0xc0, 0xc0, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1c, 0x0c, 0x0c, 0x3c, 0x6c, 0xcc,
+0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x7c, 0xc6, 0xfe, 0xc0, 0xc0, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x38, 0x6c, 0x64, 0x60, 0xf0, 0x60, 0x60, 0x60, 0x60, 0xf0,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x76, 0xcc, 0xcc,
+0xcc, 0xcc, 0xcc, 0x7c, 0x0c, 0xcc, 0x78, 0x00, 0x00, 0x00, 0xe0, 0x60,
+0x60, 0x6c, 0x76, 0x66, 0x66, 0x66, 0x66, 0xe6, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x18, 0x18, 0x00, 0x38, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x06, 0x00, 0x0e, 0x06, 0x06,
+0x06, 0x06, 0x06, 0x06, 0x66, 0x66, 0x3c, 0x00, 0x00, 0x00, 0xe0, 0x60,
+0x60, 0x66, 0x6c, 0x78, 0x78, 0x6c, 0x66, 0xe6, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x38, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xe6, 0xff, 0xdb,
+0xdb, 0xdb, 0xdb, 0xdb, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0xdc, 0x66, 0x66, 0x66, 0x66, 0x66, 0x66, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xdc, 0x66, 0x66,
+0x66, 0x66, 0x66, 0x7c, 0x60, 0x60, 0xf0, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x76, 0xcc, 0xcc, 0xcc, 0xcc, 0xcc, 0x7c, 0x0c, 0x0c, 0x1e, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0xdc, 0x76, 0x66, 0x60, 0x60, 0x60, 0xf0,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c, 0xc6, 0x60,
+0x38, 0x0c, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x30,
+0x30, 0xfc, 0x30, 0x30, 0x30, 0x30, 0x36, 0x1c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0xcc, 0xcc, 0xcc, 0xcc, 0xcc, 0xcc, 0x76,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0xc3, 0xc3,
+0xc3, 0x66, 0x3c, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0xc3, 0xc3, 0xc3, 0xdb, 0xdb, 0xff, 0x66, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0x66, 0x3c, 0x18, 0x3c, 0x66, 0xc3,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc6, 0xc6, 0xc6,
+0xc6, 0xc6, 0xc6, 0x7e, 0x06, 0x0c, 0xf8, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0xfe, 0xcc, 0x18, 0x30, 0x60, 0xc6, 0xfe, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x0e, 0x18, 0x18, 0x18, 0x70, 0x18, 0x18, 0x18, 0x18, 0x0e,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x00, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x70, 0x18,
+0x18, 0x18, 0x0e, 0x18, 0x18, 0x18, 0x18, 0x70, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x76, 0xdc, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x38, 0x6c, 0xc6,
+0xc6, 0xc6, 0xfe, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3c, 0x66,
+0xc2, 0xc0, 0xc0, 0xc0, 0xc2, 0x66, 0x3c, 0x0c, 0x06, 0x7c, 0x00, 0x00,
+0x00, 0x00, 0xcc, 0x00, 0x00, 0xcc, 0xcc, 0xcc, 0xcc, 0xcc, 0xcc, 0x76,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x0c, 0x18, 0x30, 0x00, 0x7c, 0xc6, 0xfe,
+0xc0, 0xc0, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x38, 0x6c,
+0x00, 0x78, 0x0c, 0x7c, 0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xcc, 0x00, 0x00, 0x78, 0x0c, 0x7c, 0xcc, 0xcc, 0xcc, 0x76,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x60, 0x30, 0x18, 0x00, 0x78, 0x0c, 0x7c,
+0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00, 0x00, 0x38, 0x6c, 0x38,
+0x00, 0x78, 0x0c, 0x7c, 0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x3c, 0x66, 0x60, 0x60, 0x66, 0x3c, 0x0c, 0x06,
+0x3c, 0x00, 0x00, 0x00, 0x00, 0x10, 0x38, 0x6c, 0x00, 0x7c, 0xc6, 0xfe,
+0xc0, 0xc0, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc6, 0x00,
+0x00, 0x7c, 0xc6, 0xfe, 0xc0, 0xc0, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x60, 0x30, 0x18, 0x00, 0x7c, 0xc6, 0xfe, 0xc0, 0xc0, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x66, 0x00, 0x00, 0x38, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x3c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x3c, 0x66,
+0x00, 0x38, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x60, 0x30, 0x18, 0x00, 0x38, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0xc6, 0x00, 0x10, 0x38, 0x6c, 0xc6, 0xc6,
+0xfe, 0xc6, 0xc6, 0xc6, 0x00, 0x00, 0x00, 0x00, 0x38, 0x6c, 0x38, 0x00,
+0x38, 0x6c, 0xc6, 0xc6, 0xfe, 0xc6, 0xc6, 0xc6, 0x00, 0x00, 0x00, 0x00,
+0x18, 0x30, 0x60, 0x00, 0xfe, 0x66, 0x60, 0x7c, 0x60, 0x60, 0x66, 0xfe,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x6e, 0x3b, 0x1b,
+0x7e, 0xd8, 0xdc, 0x77, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3e, 0x6c,
+0xcc, 0xcc, 0xfe, 0xcc, 0xcc, 0xcc, 0xcc, 0xce, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x10, 0x38, 0x6c, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc6, 0x00, 0x00, 0x7c, 0xc6, 0xc6,
+0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x60, 0x30, 0x18,
+0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x30, 0x78, 0xcc, 0x00, 0xcc, 0xcc, 0xcc, 0xcc, 0xcc, 0xcc, 0x76,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x60, 0x30, 0x18, 0x00, 0xcc, 0xcc, 0xcc,
+0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc6, 0x00,
+0x00, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7e, 0x06, 0x0c, 0x78, 0x00,
+0x00, 0xc6, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0xc6, 0x00, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6,
+0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x7e,
+0xc3, 0xc0, 0xc0, 0xc0, 0xc3, 0x7e, 0x18, 0x18, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x38, 0x6c, 0x64, 0x60, 0xf0, 0x60, 0x60, 0x60, 0x60, 0xe6, 0xfc,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0x66, 0x3c, 0x18, 0xff, 0x18,
+0xff, 0x18, 0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfc, 0x66, 0x66,
+0x7c, 0x62, 0x66, 0x6f, 0x66, 0x66, 0x66, 0xf3, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x0e, 0x1b, 0x18, 0x18, 0x18, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18,
+0xd8, 0x70, 0x00, 0x00, 0x00, 0x18, 0x30, 0x60, 0x00, 0x78, 0x0c, 0x7c,
+0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0c, 0x18, 0x30,
+0x00, 0x38, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x18, 0x30, 0x60, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x30, 0x60, 0x00, 0xcc, 0xcc, 0xcc,
+0xcc, 0xcc, 0xcc, 0x76, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x76, 0xdc,
+0x00, 0xdc, 0x66, 0x66, 0x66, 0x66, 0x66, 0x66, 0x00, 0x00, 0x00, 0x00,
+0x76, 0xdc, 0x00, 0xc6, 0xe6, 0xf6, 0xfe, 0xde, 0xce, 0xc6, 0xc6, 0xc6,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x3c, 0x6c, 0x6c, 0x3e, 0x00, 0x7e, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x38, 0x6c, 0x6c,
+0x38, 0x00, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x30, 0x30, 0x00, 0x30, 0x30, 0x60, 0xc0, 0xc6, 0xc6, 0x7c,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0xc0,
+0xc0, 0xc0, 0xc0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0xfe, 0x06, 0x06, 0x06, 0x06, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0xc0, 0xc0, 0xc2, 0xc6, 0xcc, 0x18, 0x30, 0x60, 0xce, 0x9b, 0x06,
+0x0c, 0x1f, 0x00, 0x00, 0x00, 0xc0, 0xc0, 0xc2, 0xc6, 0xcc, 0x18, 0x30,
+0x66, 0xce, 0x96, 0x3e, 0x06, 0x06, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18,
+0x00, 0x18, 0x18, 0x18, 0x3c, 0x3c, 0x3c, 0x18, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x6c, 0xd8, 0x6c, 0x36, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xd8, 0x6c, 0x36,
+0x6c, 0xd8, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x11, 0x44, 0x11, 0x44,
+0x11, 0x44, 0x11, 0x44, 0x11, 0x44, 0x11, 0x44, 0x11, 0x44, 0x11, 0x44,
+0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa,
+0x55, 0xaa, 0x55, 0xaa, 0xdd, 0x77, 0xdd, 0x77, 0xdd, 0x77, 0xdd, 0x77,
+0xdd, 0x77, 0xdd, 0x77, 0xdd, 0x77, 0xdd, 0x77, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0xf8, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0xf8, 0x18, 0xf8,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0xf6, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0x36, 0x00, 0x00, 0x00, 0x00, 0x00, 0xf8, 0x18, 0xf8,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x36, 0x36, 0x36, 0x36,
+0x36, 0xf6, 0x06, 0xf6, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0x36, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0x06, 0xf6,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x36, 0xf6, 0x06, 0xfe, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0xfe, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0xf8, 0x18, 0xf8,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0xf8, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x1f, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0xff,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0xff, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x1f, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0xff, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x1f, 0x18, 0x1f, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x37,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x37, 0x30, 0x3f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x3f, 0x30, 0x37, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0xf7, 0x00, 0xff,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0xff, 0x00, 0xf7, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x37, 0x30, 0x37, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0x36, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0x00, 0xff,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x36, 0x36, 0x36,
+0x36, 0xf7, 0x00, 0xf7, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x18, 0x18, 0x18, 0x18, 0x18, 0xff, 0x00, 0xff, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0xff,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0xff, 0x00, 0xff, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x3f,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x1f, 0x18, 0x1f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x1f, 0x18, 0x1f, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3f,
+0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x36, 0x36, 0x36, 0xff, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36, 0x36,
+0x18, 0x18, 0x18, 0x18, 0x18, 0xff, 0x18, 0xff, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0xf8,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x1f, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+0xff, 0xff, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff,
+0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xf0, 0xf0, 0xf0, 0xf0,
+0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
+0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
+0x0f, 0x0f, 0x0f, 0x0f, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x76, 0xdc, 0xd8, 0xd8, 0xd8, 0xdc, 0x76, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x78, 0xcc, 0xcc, 0xcc, 0xd8, 0xcc, 0xc6, 0xc6, 0xc6, 0xcc,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xfe, 0xc6, 0xc6, 0xc0, 0xc0, 0xc0,
+0xc0, 0xc0, 0xc0, 0xc0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0xfe, 0x6c, 0x6c, 0x6c, 0x6c, 0x6c, 0x6c, 0x6c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0xfe, 0xc6, 0x60, 0x30, 0x18, 0x30, 0x60, 0xc6, 0xfe,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7e, 0xd8, 0xd8,
+0xd8, 0xd8, 0xd8, 0x70, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x66, 0x66, 0x66, 0x66, 0x66, 0x7c, 0x60, 0x60, 0xc0, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x76, 0xdc, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7e, 0x18, 0x3c, 0x66, 0x66,
+0x66, 0x3c, 0x18, 0x7e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x38,
+0x6c, 0xc6, 0xc6, 0xfe, 0xc6, 0xc6, 0x6c, 0x38, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x38, 0x6c, 0xc6, 0xc6, 0xc6, 0x6c, 0x6c, 0x6c, 0x6c, 0xee,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1e, 0x30, 0x18, 0x0c, 0x3e, 0x66,
+0x66, 0x66, 0x66, 0x3c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x7e, 0xdb, 0xdb, 0xdb, 0x7e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x03, 0x06, 0x7e, 0xdb, 0xdb, 0xf3, 0x7e, 0x60, 0xc0,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1c, 0x30, 0x60, 0x60, 0x7c, 0x60,
+0x60, 0x60, 0x30, 0x1c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x7c,
+0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0xfe, 0x00, 0x00, 0xfe, 0x00, 0x00, 0xfe, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x7e, 0x18,
+0x18, 0x00, 0x00, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x30,
+0x18, 0x0c, 0x06, 0x0c, 0x18, 0x30, 0x00, 0x7e, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x0c, 0x18, 0x30, 0x60, 0x30, 0x18, 0x0c, 0x00, 0x7e,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0e, 0x1b, 0x1b, 0x1b, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18,
+0x18, 0x18, 0x18, 0x18, 0xd8, 0xd8, 0xd8, 0x70, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x00, 0x7e, 0x00, 0x18, 0x18, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x76, 0xdc, 0x00,
+0x76, 0xdc, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x38, 0x6c, 0x6c,
+0x38, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x18, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x18, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0f, 0x0c, 0x0c,
+0x0c, 0x0c, 0x0c, 0xec, 0x6c, 0x6c, 0x3c, 0x1c, 0x00, 0x00, 0x00, 0x00,
+0x00, 0xd8, 0x6c, 0x6c, 0x6c, 0x6c, 0x6c, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x70, 0xd8, 0x30, 0x60, 0xc8, 0xf8, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x7c, 0x7c, 0x7c, 0x7c, 0x7c, 0x7c, 0x7c, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+0x00, 0x00, 0x00, 0x00,
+};
diff --git a/arch/sparc/kernel/central.c b/arch/sparc/kernel/central.c
new file mode 100644
index 00000000000..052b5a44318
--- /dev/null
+++ b/arch/sparc/kernel/central.c
@@ -0,0 +1,272 @@
+/* central.c: Central FHC driver for Sunfire/Starfire/Wildfire.
+ *
+ * Copyright (C) 1997, 1999, 2008 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/export.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+
+#include <asm/fhc.h>
+#include <asm/upa.h>
+
+struct clock_board {
+ void __iomem *clock_freq_regs;
+ void __iomem *clock_regs;
+ void __iomem *clock_ver_reg;
+ int num_slots;
+ struct resource leds_resource;
+ struct platform_device leds_pdev;
+};
+
+struct fhc {
+ void __iomem *pregs;
+ bool central;
+ bool jtag_master;
+ int board_num;
+ struct resource leds_resource;
+ struct platform_device leds_pdev;
+};
+
+static int clock_board_calc_nslots(struct clock_board *p)
+{
+ u8 reg = upa_readb(p->clock_regs + CLOCK_STAT1) & 0xc0;
+
+ switch (reg) {
+ case 0x40:
+ return 16;
+
+ case 0xc0:
+ return 8;
+
+ case 0x80:
+ reg = 0;
+ if (p->clock_ver_reg)
+ reg = upa_readb(p->clock_ver_reg);
+ if (reg) {
+ if (reg & 0x80)
+ return 4;
+ else
+ return 5;
+ }
+ /* Fallthrough */
+ default:
+ return 4;
+ }
+}
+
+static int clock_board_probe(struct platform_device *op)
+{
+ struct clock_board *p = kzalloc(sizeof(*p), GFP_KERNEL);
+ int err = -ENOMEM;
+
+ if (!p) {
+ printk(KERN_ERR "clock_board: Cannot allocate struct clock_board\n");
+ goto out;
+ }
+
+ p->clock_freq_regs = of_ioremap(&op->resource[0], 0,
+ resource_size(&op->resource[0]),
+ "clock_board_freq");
+ if (!p->clock_freq_regs) {
+ printk(KERN_ERR "clock_board: Cannot map clock_freq_regs\n");
+ goto out_free;
+ }
+
+ p->clock_regs = of_ioremap(&op->resource[1], 0,
+ resource_size(&op->resource[1]),
+ "clock_board_regs");
+ if (!p->clock_regs) {
+ printk(KERN_ERR "clock_board: Cannot map clock_regs\n");
+ goto out_unmap_clock_freq_regs;
+ }
+
+ if (op->resource[2].flags) {
+ p->clock_ver_reg = of_ioremap(&op->resource[2], 0,
+ resource_size(&op->resource[2]),
+ "clock_ver_reg");
+ if (!p->clock_ver_reg) {
+ printk(KERN_ERR "clock_board: Cannot map clock_ver_reg\n");
+ goto out_unmap_clock_regs;
+ }
+ }
+
+ p->num_slots = clock_board_calc_nslots(p);
+
+ p->leds_resource.start = (unsigned long)
+ (p->clock_regs + CLOCK_CTRL);
+ p->leds_resource.end = p->leds_resource.start;
+ p->leds_resource.name = "leds";
+
+ p->leds_pdev.name = "sunfire-clockboard-leds";
+ p->leds_pdev.id = -1;
+ p->leds_pdev.resource = &p->leds_resource;
+ p->leds_pdev.num_resources = 1;
+ p->leds_pdev.dev.parent = &op->dev;
+
+ err = platform_device_register(&p->leds_pdev);
+ if (err) {
+ printk(KERN_ERR "clock_board: Could not register LEDS "
+ "platform device\n");
+ goto out_unmap_clock_ver_reg;
+ }
+
+ printk(KERN_INFO "clock_board: Detected %d slot Enterprise system.\n",
+ p->num_slots);
+
+ err = 0;
+out:
+ return err;
+
+out_unmap_clock_ver_reg:
+ if (p->clock_ver_reg)
+ of_iounmap(&op->resource[2], p->clock_ver_reg,
+ resource_size(&op->resource[2]));
+
+out_unmap_clock_regs:
+ of_iounmap(&op->resource[1], p->clock_regs,
+ resource_size(&op->resource[1]));
+
+out_unmap_clock_freq_regs:
+ of_iounmap(&op->resource[0], p->clock_freq_regs,
+ resource_size(&op->resource[0]));
+
+out_free:
+ kfree(p);
+ goto out;
+}
+
+static const struct of_device_id clock_board_match[] = {
+ {
+ .name = "clock-board",
+ },
+ {},
+};
+
+static struct platform_driver clock_board_driver = {
+ .probe = clock_board_probe,
+ .driver = {
+ .name = "clock_board",
+ .owner = THIS_MODULE,
+ .of_match_table = clock_board_match,
+ },
+};
+
+static int fhc_probe(struct platform_device *op)
+{
+ struct fhc *p = kzalloc(sizeof(*p), GFP_KERNEL);
+ int err = -ENOMEM;
+ u32 reg;
+
+ if (!p) {
+ printk(KERN_ERR "fhc: Cannot allocate struct fhc\n");
+ goto out;
+ }
+
+ if (!strcmp(op->dev.of_node->parent->name, "central"))
+ p->central = true;
+
+ p->pregs = of_ioremap(&op->resource[0], 0,
+ resource_size(&op->resource[0]),
+ "fhc_pregs");
+ if (!p->pregs) {
+ printk(KERN_ERR "fhc: Cannot map pregs\n");
+ goto out_free;
+ }
+
+ if (p->central) {
+ reg = upa_readl(p->pregs + FHC_PREGS_BSR);
+ p->board_num = ((reg >> 16) & 1) | ((reg >> 12) & 0x0e);
+ } else {
+ p->board_num = of_getintprop_default(op->dev.of_node, "board#", -1);
+ if (p->board_num == -1) {
+ printk(KERN_ERR "fhc: No board# property\n");
+ goto out_unmap_pregs;
+ }
+ if (upa_readl(p->pregs + FHC_PREGS_JCTRL) & FHC_JTAG_CTRL_MENAB)
+ p->jtag_master = true;
+ }
+
+ if (!p->central) {
+ p->leds_resource.start = (unsigned long)
+ (p->pregs + FHC_PREGS_CTRL);
+ p->leds_resource.end = p->leds_resource.start;
+ p->leds_resource.name = "leds";
+
+ p->leds_pdev.name = "sunfire-fhc-leds";
+ p->leds_pdev.id = p->board_num;
+ p->leds_pdev.resource = &p->leds_resource;
+ p->leds_pdev.num_resources = 1;
+ p->leds_pdev.dev.parent = &op->dev;
+
+ err = platform_device_register(&p->leds_pdev);
+ if (err) {
+ printk(KERN_ERR "fhc: Could not register LEDS "
+ "platform device\n");
+ goto out_unmap_pregs;
+ }
+ }
+ reg = upa_readl(p->pregs + FHC_PREGS_CTRL);
+
+ if (!p->central)
+ reg |= FHC_CONTROL_IXIST;
+
+ reg &= ~(FHC_CONTROL_AOFF |
+ FHC_CONTROL_BOFF |
+ FHC_CONTROL_SLINE);
+
+ upa_writel(reg, p->pregs + FHC_PREGS_CTRL);
+ upa_readl(p->pregs + FHC_PREGS_CTRL);
+
+ reg = upa_readl(p->pregs + FHC_PREGS_ID);
+ printk(KERN_INFO "fhc: Board #%d, Version[%x] PartID[%x] Manuf[%x] %s\n",
+ p->board_num,
+ (reg & FHC_ID_VERS) >> 28,
+ (reg & FHC_ID_PARTID) >> 12,
+ (reg & FHC_ID_MANUF) >> 1,
+ (p->jtag_master ?
+ "(JTAG Master)" :
+ (p->central ? "(Central)" : "")));
+
+ err = 0;
+
+out:
+ return err;
+
+out_unmap_pregs:
+ of_iounmap(&op->resource[0], p->pregs, resource_size(&op->resource[0]));
+
+out_free:
+ kfree(p);
+ goto out;
+}
+
+static const struct of_device_id fhc_match[] = {
+ {
+ .name = "fhc",
+ },
+ {},
+};
+
+static struct platform_driver fhc_driver = {
+ .probe = fhc_probe,
+ .driver = {
+ .name = "fhc",
+ .owner = THIS_MODULE,
+ .of_match_table = fhc_match,
+ },
+};
+
+static int __init sunfire_init(void)
+{
+ (void) platform_driver_register(&fhc_driver);
+ (void) platform_driver_register(&clock_board_driver);
+ return 0;
+}
+
+fs_initcall(sunfire_init);
diff --git a/arch/sparc/kernel/cherrs.S b/arch/sparc/kernel/cherrs.S
new file mode 100644
index 00000000000..4ee1ad42086
--- /dev/null
+++ b/arch/sparc/kernel/cherrs.S
@@ -0,0 +1,579 @@
+ /* These get patched into the trap table at boot time
+ * once we know we have a cheetah processor.
+ */
+ .globl cheetah_fecc_trap_vector
+ .type cheetah_fecc_trap_vector,#function
+cheetah_fecc_trap_vector:
+ membar #Sync
+ ldxa [%g0] ASI_DCU_CONTROL_REG, %g1
+ andn %g1, DCU_DC | DCU_IC, %g1
+ stxa %g1, [%g0] ASI_DCU_CONTROL_REG
+ membar #Sync
+ sethi %hi(cheetah_fast_ecc), %g2
+ jmpl %g2 + %lo(cheetah_fast_ecc), %g0
+ mov 0, %g1
+ .size cheetah_fecc_trap_vector,.-cheetah_fecc_trap_vector
+
+ .globl cheetah_fecc_trap_vector_tl1
+ .type cheetah_fecc_trap_vector_tl1,#function
+cheetah_fecc_trap_vector_tl1:
+ membar #Sync
+ ldxa [%g0] ASI_DCU_CONTROL_REG, %g1
+ andn %g1, DCU_DC | DCU_IC, %g1
+ stxa %g1, [%g0] ASI_DCU_CONTROL_REG
+ membar #Sync
+ sethi %hi(cheetah_fast_ecc), %g2
+ jmpl %g2 + %lo(cheetah_fast_ecc), %g0
+ mov 1, %g1
+ .size cheetah_fecc_trap_vector_tl1,.-cheetah_fecc_trap_vector_tl1
+
+ .globl cheetah_cee_trap_vector
+ .type cheetah_cee_trap_vector,#function
+cheetah_cee_trap_vector:
+ membar #Sync
+ ldxa [%g0] ASI_DCU_CONTROL_REG, %g1
+ andn %g1, DCU_IC, %g1
+ stxa %g1, [%g0] ASI_DCU_CONTROL_REG
+ membar #Sync
+ sethi %hi(cheetah_cee), %g2
+ jmpl %g2 + %lo(cheetah_cee), %g0
+ mov 0, %g1
+ .size cheetah_cee_trap_vector,.-cheetah_cee_trap_vector
+
+ .globl cheetah_cee_trap_vector_tl1
+ .type cheetah_cee_trap_vector_tl1,#function
+cheetah_cee_trap_vector_tl1:
+ membar #Sync
+ ldxa [%g0] ASI_DCU_CONTROL_REG, %g1
+ andn %g1, DCU_IC, %g1
+ stxa %g1, [%g0] ASI_DCU_CONTROL_REG
+ membar #Sync
+ sethi %hi(cheetah_cee), %g2
+ jmpl %g2 + %lo(cheetah_cee), %g0
+ mov 1, %g1
+ .size cheetah_cee_trap_vector_tl1,.-cheetah_cee_trap_vector_tl1
+
+ .globl cheetah_deferred_trap_vector
+ .type cheetah_deferred_trap_vector,#function
+cheetah_deferred_trap_vector:
+ membar #Sync
+ ldxa [%g0] ASI_DCU_CONTROL_REG, %g1;
+ andn %g1, DCU_DC | DCU_IC, %g1;
+ stxa %g1, [%g0] ASI_DCU_CONTROL_REG;
+ membar #Sync;
+ sethi %hi(cheetah_deferred_trap), %g2
+ jmpl %g2 + %lo(cheetah_deferred_trap), %g0
+ mov 0, %g1
+ .size cheetah_deferred_trap_vector,.-cheetah_deferred_trap_vector
+
+ .globl cheetah_deferred_trap_vector_tl1
+ .type cheetah_deferred_trap_vector_tl1,#function
+cheetah_deferred_trap_vector_tl1:
+ membar #Sync;
+ ldxa [%g0] ASI_DCU_CONTROL_REG, %g1;
+ andn %g1, DCU_DC | DCU_IC, %g1;
+ stxa %g1, [%g0] ASI_DCU_CONTROL_REG;
+ membar #Sync;
+ sethi %hi(cheetah_deferred_trap), %g2
+ jmpl %g2 + %lo(cheetah_deferred_trap), %g0
+ mov 1, %g1
+ .size cheetah_deferred_trap_vector_tl1,.-cheetah_deferred_trap_vector_tl1
+
+ /* Cheetah+ specific traps. These are for the new I/D cache parity
+ * error traps. The first argument to cheetah_plus_parity_handler
+ * is encoded as follows:
+ *
+ * Bit0: 0=dcache,1=icache
+ * Bit1: 0=recoverable,1=unrecoverable
+ */
+ .globl cheetah_plus_dcpe_trap_vector
+ .type cheetah_plus_dcpe_trap_vector,#function
+cheetah_plus_dcpe_trap_vector:
+ membar #Sync
+ sethi %hi(do_cheetah_plus_data_parity), %g7
+ jmpl %g7 + %lo(do_cheetah_plus_data_parity), %g0
+ nop
+ nop
+ nop
+ nop
+ nop
+ .size cheetah_plus_dcpe_trap_vector,.-cheetah_plus_dcpe_trap_vector
+
+ .type do_cheetah_plus_data_parity,#function
+do_cheetah_plus_data_parity:
+ rdpr %pil, %g2
+ wrpr %g0, PIL_NORMAL_MAX, %pil
+ ba,pt %xcc, etrap_irq
+ rd %pc, %g7
+#ifdef CONFIG_TRACE_IRQFLAGS
+ call trace_hardirqs_off
+ nop
+#endif
+ mov 0x0, %o0
+ call cheetah_plus_parity_error
+ add %sp, PTREGS_OFF, %o1
+ ba,a,pt %xcc, rtrap_irq
+ .size do_cheetah_plus_data_parity,.-do_cheetah_plus_data_parity
+
+ .globl cheetah_plus_dcpe_trap_vector_tl1
+ .type cheetah_plus_dcpe_trap_vector_tl1,#function
+cheetah_plus_dcpe_trap_vector_tl1:
+ membar #Sync
+ wrpr PSTATE_IG | PSTATE_PEF | PSTATE_PRIV, %pstate
+ sethi %hi(do_dcpe_tl1), %g3
+ jmpl %g3 + %lo(do_dcpe_tl1), %g0
+ nop
+ nop
+ nop
+ nop
+ .size cheetah_plus_dcpe_trap_vector_tl1,.-cheetah_plus_dcpe_trap_vector_tl1
+
+ .globl cheetah_plus_icpe_trap_vector
+ .type cheetah_plus_icpe_trap_vector,#function
+cheetah_plus_icpe_trap_vector:
+ membar #Sync
+ sethi %hi(do_cheetah_plus_insn_parity), %g7
+ jmpl %g7 + %lo(do_cheetah_plus_insn_parity), %g0
+ nop
+ nop
+ nop
+ nop
+ nop
+ .size cheetah_plus_icpe_trap_vector,.-cheetah_plus_icpe_trap_vector
+
+ .type do_cheetah_plus_insn_parity,#function
+do_cheetah_plus_insn_parity:
+ rdpr %pil, %g2
+ wrpr %g0, PIL_NORMAL_MAX, %pil
+ ba,pt %xcc, etrap_irq
+ rd %pc, %g7
+#ifdef CONFIG_TRACE_IRQFLAGS
+ call trace_hardirqs_off
+ nop
+#endif
+ mov 0x1, %o0
+ call cheetah_plus_parity_error
+ add %sp, PTREGS_OFF, %o1
+ ba,a,pt %xcc, rtrap_irq
+ .size do_cheetah_plus_insn_parity,.-do_cheetah_plus_insn_parity
+
+ .globl cheetah_plus_icpe_trap_vector_tl1
+ .type cheetah_plus_icpe_trap_vector_tl1,#function
+cheetah_plus_icpe_trap_vector_tl1:
+ membar #Sync
+ wrpr PSTATE_IG | PSTATE_PEF | PSTATE_PRIV, %pstate
+ sethi %hi(do_icpe_tl1), %g3
+ jmpl %g3 + %lo(do_icpe_tl1), %g0
+ nop
+ nop
+ nop
+ nop
+ .size cheetah_plus_icpe_trap_vector_tl1,.-cheetah_plus_icpe_trap_vector_tl1
+
+ /* If we take one of these traps when tl >= 1, then we
+ * jump to interrupt globals. If some trap level above us
+ * was also using interrupt globals, we cannot recover.
+ * We may use all interrupt global registers except %g6.
+ */
+ .globl do_dcpe_tl1
+ .type do_dcpe_tl1,#function
+do_dcpe_tl1:
+ rdpr %tl, %g1 ! Save original trap level
+ mov 1, %g2 ! Setup TSTATE checking loop
+ sethi %hi(TSTATE_IG), %g3 ! TSTATE mask bit
+1: wrpr %g2, %tl ! Set trap level to check
+ rdpr %tstate, %g4 ! Read TSTATE for this level
+ andcc %g4, %g3, %g0 ! Interrupt globals in use?
+ bne,a,pn %xcc, do_dcpe_tl1_fatal ! Yep, irrecoverable
+ wrpr %g1, %tl ! Restore original trap level
+ add %g2, 1, %g2 ! Next trap level
+ cmp %g2, %g1 ! Hit them all yet?
+ ble,pt %icc, 1b ! Not yet
+ nop
+ wrpr %g1, %tl ! Restore original trap level
+do_dcpe_tl1_nonfatal: /* Ok we may use interrupt globals safely. */
+ sethi %hi(dcache_parity_tl1_occurred), %g2
+ lduw [%g2 + %lo(dcache_parity_tl1_occurred)], %g1
+ add %g1, 1, %g1
+ stw %g1, [%g2 + %lo(dcache_parity_tl1_occurred)]
+ /* Reset D-cache parity */
+ sethi %hi(1 << 16), %g1 ! D-cache size
+ mov (1 << 5), %g2 ! D-cache line size
+ sub %g1, %g2, %g1 ! Move down 1 cacheline
+1: srl %g1, 14, %g3 ! Compute UTAG
+ membar #Sync
+ stxa %g3, [%g1] ASI_DCACHE_UTAG
+ membar #Sync
+ sub %g2, 8, %g3 ! 64-bit data word within line
+2: membar #Sync
+ stxa %g0, [%g1 + %g3] ASI_DCACHE_DATA
+ membar #Sync
+ subcc %g3, 8, %g3 ! Next 64-bit data word
+ bge,pt %icc, 2b
+ nop
+ subcc %g1, %g2, %g1 ! Next cacheline
+ bge,pt %icc, 1b
+ nop
+ ba,pt %xcc, dcpe_icpe_tl1_common
+ nop
+
+do_dcpe_tl1_fatal:
+ sethi %hi(1f), %g7
+ ba,pt %xcc, etraptl1
+1: or %g7, %lo(1b), %g7
+ mov 0x2, %o0
+ call cheetah_plus_parity_error
+ add %sp, PTREGS_OFF, %o1
+ ba,pt %xcc, rtrap
+ nop
+ .size do_dcpe_tl1,.-do_dcpe_tl1
+
+ .globl do_icpe_tl1
+ .type do_icpe_tl1,#function
+do_icpe_tl1:
+ rdpr %tl, %g1 ! Save original trap level
+ mov 1, %g2 ! Setup TSTATE checking loop
+ sethi %hi(TSTATE_IG), %g3 ! TSTATE mask bit
+1: wrpr %g2, %tl ! Set trap level to check
+ rdpr %tstate, %g4 ! Read TSTATE for this level
+ andcc %g4, %g3, %g0 ! Interrupt globals in use?
+ bne,a,pn %xcc, do_icpe_tl1_fatal ! Yep, irrecoverable
+ wrpr %g1, %tl ! Restore original trap level
+ add %g2, 1, %g2 ! Next trap level
+ cmp %g2, %g1 ! Hit them all yet?
+ ble,pt %icc, 1b ! Not yet
+ nop
+ wrpr %g1, %tl ! Restore original trap level
+do_icpe_tl1_nonfatal: /* Ok we may use interrupt globals safely. */
+ sethi %hi(icache_parity_tl1_occurred), %g2
+ lduw [%g2 + %lo(icache_parity_tl1_occurred)], %g1
+ add %g1, 1, %g1
+ stw %g1, [%g2 + %lo(icache_parity_tl1_occurred)]
+ /* Flush I-cache */
+ sethi %hi(1 << 15), %g1 ! I-cache size
+ mov (1 << 5), %g2 ! I-cache line size
+ sub %g1, %g2, %g1
+1: or %g1, (2 << 3), %g3
+ stxa %g0, [%g3] ASI_IC_TAG
+ membar #Sync
+ subcc %g1, %g2, %g1
+ bge,pt %icc, 1b
+ nop
+ ba,pt %xcc, dcpe_icpe_tl1_common
+ nop
+
+do_icpe_tl1_fatal:
+ sethi %hi(1f), %g7
+ ba,pt %xcc, etraptl1
+1: or %g7, %lo(1b), %g7
+ mov 0x3, %o0
+ call cheetah_plus_parity_error
+ add %sp, PTREGS_OFF, %o1
+ ba,pt %xcc, rtrap
+ nop
+ .size do_icpe_tl1,.-do_icpe_tl1
+
+ .type dcpe_icpe_tl1_common,#function
+dcpe_icpe_tl1_common:
+ /* Flush D-cache, re-enable D/I caches in DCU and finally
+ * retry the trapping instruction.
+ */
+ sethi %hi(1 << 16), %g1 ! D-cache size
+ mov (1 << 5), %g2 ! D-cache line size
+ sub %g1, %g2, %g1
+1: stxa %g0, [%g1] ASI_DCACHE_TAG
+ membar #Sync
+ subcc %g1, %g2, %g1
+ bge,pt %icc, 1b
+ nop
+ ldxa [%g0] ASI_DCU_CONTROL_REG, %g1
+ or %g1, (DCU_DC | DCU_IC), %g1
+ stxa %g1, [%g0] ASI_DCU_CONTROL_REG
+ membar #Sync
+ retry
+ .size dcpe_icpe_tl1_common,.-dcpe_icpe_tl1_common
+
+ /* Capture I/D/E-cache state into per-cpu error scoreboard.
+ *
+ * %g1: (TL>=0) ? 1 : 0
+ * %g2: scratch
+ * %g3: scratch
+ * %g4: AFSR
+ * %g5: AFAR
+ * %g6: unused, will have current thread ptr after etrap
+ * %g7: scratch
+ */
+ .type __cheetah_log_error,#function
+__cheetah_log_error:
+ /* Put "TL1" software bit into AFSR. */
+ and %g1, 0x1, %g1
+ sllx %g1, 63, %g2
+ or %g4, %g2, %g4
+
+ /* Get log entry pointer for this cpu at this trap level. */
+ BRANCH_IF_JALAPENO(g2,g3,50f)
+ ldxa [%g0] ASI_SAFARI_CONFIG, %g2
+ srlx %g2, 17, %g2
+ ba,pt %xcc, 60f
+ and %g2, 0x3ff, %g2
+
+50: ldxa [%g0] ASI_JBUS_CONFIG, %g2
+ srlx %g2, 17, %g2
+ and %g2, 0x1f, %g2
+
+60: sllx %g2, 9, %g2
+ sethi %hi(cheetah_error_log), %g3
+ ldx [%g3 + %lo(cheetah_error_log)], %g3
+ brz,pn %g3, 80f
+ nop
+
+ add %g3, %g2, %g3
+ sllx %g1, 8, %g1
+ add %g3, %g1, %g1
+
+ /* %g1 holds pointer to the top of the logging scoreboard */
+ ldx [%g1 + 0x0], %g7
+ cmp %g7, -1
+ bne,pn %xcc, 80f
+ nop
+
+ stx %g4, [%g1 + 0x0]
+ stx %g5, [%g1 + 0x8]
+ add %g1, 0x10, %g1
+
+ /* %g1 now points to D-cache logging area */
+ set 0x3ff8, %g2 /* DC_addr mask */
+ and %g5, %g2, %g2 /* DC_addr bits of AFAR */
+ srlx %g5, 12, %g3
+ or %g3, 1, %g3 /* PHYS tag + valid */
+
+10: ldxa [%g2] ASI_DCACHE_TAG, %g7
+ cmp %g3, %g7 /* TAG match? */
+ bne,pt %xcc, 13f
+ nop
+
+ /* Yep, what we want, capture state. */
+ stx %g2, [%g1 + 0x20]
+ stx %g7, [%g1 + 0x28]
+
+ /* A membar Sync is required before and after utag access. */
+ membar #Sync
+ ldxa [%g2] ASI_DCACHE_UTAG, %g7
+ membar #Sync
+ stx %g7, [%g1 + 0x30]
+ ldxa [%g2] ASI_DCACHE_SNOOP_TAG, %g7
+ stx %g7, [%g1 + 0x38]
+ clr %g3
+
+12: ldxa [%g2 + %g3] ASI_DCACHE_DATA, %g7
+ stx %g7, [%g1]
+ add %g3, (1 << 5), %g3
+ cmp %g3, (4 << 5)
+ bl,pt %xcc, 12b
+ add %g1, 0x8, %g1
+
+ ba,pt %xcc, 20f
+ add %g1, 0x20, %g1
+
+13: sethi %hi(1 << 14), %g7
+ add %g2, %g7, %g2
+ srlx %g2, 14, %g7
+ cmp %g7, 4
+ bl,pt %xcc, 10b
+ nop
+
+ add %g1, 0x40, %g1
+
+ /* %g1 now points to I-cache logging area */
+20: set 0x1fe0, %g2 /* IC_addr mask */
+ and %g5, %g2, %g2 /* IC_addr bits of AFAR */
+ sllx %g2, 1, %g2 /* IC_addr[13:6]==VA[12:5] */
+ srlx %g5, (13 - 8), %g3 /* Make PTAG */
+ andn %g3, 0xff, %g3 /* Mask off undefined bits */
+
+21: ldxa [%g2] ASI_IC_TAG, %g7
+ andn %g7, 0xff, %g7
+ cmp %g3, %g7
+ bne,pt %xcc, 23f
+ nop
+
+ /* Yep, what we want, capture state. */
+ stx %g2, [%g1 + 0x40]
+ stx %g7, [%g1 + 0x48]
+ add %g2, (1 << 3), %g2
+ ldxa [%g2] ASI_IC_TAG, %g7
+ add %g2, (1 << 3), %g2
+ stx %g7, [%g1 + 0x50]
+ ldxa [%g2] ASI_IC_TAG, %g7
+ add %g2, (1 << 3), %g2
+ stx %g7, [%g1 + 0x60]
+ ldxa [%g2] ASI_IC_TAG, %g7
+ stx %g7, [%g1 + 0x68]
+ sub %g2, (3 << 3), %g2
+ ldxa [%g2] ASI_IC_STAG, %g7
+ stx %g7, [%g1 + 0x58]
+ clr %g3
+ srlx %g2, 2, %g2
+
+22: ldxa [%g2 + %g3] ASI_IC_INSTR, %g7
+ stx %g7, [%g1]
+ add %g3, (1 << 3), %g3
+ cmp %g3, (8 << 3)
+ bl,pt %xcc, 22b
+ add %g1, 0x8, %g1
+
+ ba,pt %xcc, 30f
+ add %g1, 0x30, %g1
+
+23: sethi %hi(1 << 14), %g7
+ add %g2, %g7, %g2
+ srlx %g2, 14, %g7
+ cmp %g7, 4
+ bl,pt %xcc, 21b
+ nop
+
+ add %g1, 0x70, %g1
+
+ /* %g1 now points to E-cache logging area */
+30: andn %g5, (32 - 1), %g2
+ stx %g2, [%g1 + 0x20]
+ ldxa [%g2] ASI_EC_TAG_DATA, %g7
+ stx %g7, [%g1 + 0x28]
+ ldxa [%g2] ASI_EC_R, %g0
+ clr %g3
+
+31: ldxa [%g3] ASI_EC_DATA, %g7
+ stx %g7, [%g1 + %g3]
+ add %g3, 0x8, %g3
+ cmp %g3, 0x20
+
+ bl,pt %xcc, 31b
+ nop
+80:
+ rdpr %tt, %g2
+ cmp %g2, 0x70
+ be c_fast_ecc
+ cmp %g2, 0x63
+ be c_cee
+ nop
+ ba,pt %xcc, c_deferred
+ .size __cheetah_log_error,.-__cheetah_log_error
+
+ /* Cheetah FECC trap handling, we get here from tl{0,1}_fecc
+ * in the trap table. That code has done a memory barrier
+ * and has disabled both the I-cache and D-cache in the DCU
+ * control register. The I-cache is disabled so that we may
+ * capture the corrupted cache line, and the D-cache is disabled
+ * because corrupt data may have been placed there and we don't
+ * want to reference it.
+ *
+ * %g1 is one if this trap occurred at %tl >= 1.
+ *
+ * Next, we turn off error reporting so that we don't recurse.
+ */
+ .globl cheetah_fast_ecc
+ .type cheetah_fast_ecc,#function
+cheetah_fast_ecc:
+ ldxa [%g0] ASI_ESTATE_ERROR_EN, %g2
+ andn %g2, ESTATE_ERROR_NCEEN | ESTATE_ERROR_CEEN, %g2
+ stxa %g2, [%g0] ASI_ESTATE_ERROR_EN
+ membar #Sync
+
+ /* Fetch and clear AFSR/AFAR */
+ ldxa [%g0] ASI_AFSR, %g4
+ ldxa [%g0] ASI_AFAR, %g5
+ stxa %g4, [%g0] ASI_AFSR
+ membar #Sync
+
+ ba,pt %xcc, __cheetah_log_error
+ nop
+ .size cheetah_fast_ecc,.-cheetah_fast_ecc
+
+ .type c_fast_ecc,#function
+c_fast_ecc:
+ rdpr %pil, %g2
+ wrpr %g0, PIL_NORMAL_MAX, %pil
+ ba,pt %xcc, etrap_irq
+ rd %pc, %g7
+#ifdef CONFIG_TRACE_IRQFLAGS
+ call trace_hardirqs_off
+ nop
+#endif
+ mov %l4, %o1
+ mov %l5, %o2
+ call cheetah_fecc_handler
+ add %sp, PTREGS_OFF, %o0
+ ba,a,pt %xcc, rtrap_irq
+ .size c_fast_ecc,.-c_fast_ecc
+
+ /* Our caller has disabled I-cache and performed membar Sync. */
+ .globl cheetah_cee
+ .type cheetah_cee,#function
+cheetah_cee:
+ ldxa [%g0] ASI_ESTATE_ERROR_EN, %g2
+ andn %g2, ESTATE_ERROR_CEEN, %g2
+ stxa %g2, [%g0] ASI_ESTATE_ERROR_EN
+ membar #Sync
+
+ /* Fetch and clear AFSR/AFAR */
+ ldxa [%g0] ASI_AFSR, %g4
+ ldxa [%g0] ASI_AFAR, %g5
+ stxa %g4, [%g0] ASI_AFSR
+ membar #Sync
+
+ ba,pt %xcc, __cheetah_log_error
+ nop
+ .size cheetah_cee,.-cheetah_cee
+
+ .type c_cee,#function
+c_cee:
+ rdpr %pil, %g2
+ wrpr %g0, PIL_NORMAL_MAX, %pil
+ ba,pt %xcc, etrap_irq
+ rd %pc, %g7
+#ifdef CONFIG_TRACE_IRQFLAGS
+ call trace_hardirqs_off
+ nop
+#endif
+ mov %l4, %o1
+ mov %l5, %o2
+ call cheetah_cee_handler
+ add %sp, PTREGS_OFF, %o0
+ ba,a,pt %xcc, rtrap_irq
+ .size c_cee,.-c_cee
+
+ /* Our caller has disabled I-cache+D-cache and performed membar Sync. */
+ .globl cheetah_deferred_trap
+ .type cheetah_deferred_trap,#function
+cheetah_deferred_trap:
+ ldxa [%g0] ASI_ESTATE_ERROR_EN, %g2
+ andn %g2, ESTATE_ERROR_NCEEN | ESTATE_ERROR_CEEN, %g2
+ stxa %g2, [%g0] ASI_ESTATE_ERROR_EN
+ membar #Sync
+
+ /* Fetch and clear AFSR/AFAR */
+ ldxa [%g0] ASI_AFSR, %g4
+ ldxa [%g0] ASI_AFAR, %g5
+ stxa %g4, [%g0] ASI_AFSR
+ membar #Sync
+
+ ba,pt %xcc, __cheetah_log_error
+ nop
+ .size cheetah_deferred_trap,.-cheetah_deferred_trap
+
+ .type c_deferred,#function
+c_deferred:
+ rdpr %pil, %g2
+ wrpr %g0, PIL_NORMAL_MAX, %pil
+ ba,pt %xcc, etrap_irq
+ rd %pc, %g7
+#ifdef CONFIG_TRACE_IRQFLAGS
+ call trace_hardirqs_off
+ nop
+#endif
+ mov %l4, %o1
+ mov %l5, %o2
+ call cheetah_deferred_handler
+ add %sp, PTREGS_OFF, %o0
+ ba,a,pt %xcc, rtrap_irq
+ .size c_deferred,.-c_deferred
diff --git a/arch/sparc/kernel/chmc.c b/arch/sparc/kernel/chmc.c
new file mode 100644
index 00000000000..dbb210d74e2
--- /dev/null
+++ b/arch/sparc/kernel/chmc.c
@@ -0,0 +1,864 @@
+/* chmc.c: Driver for UltraSPARC-III memory controller.
+ *
+ * Copyright (C) 2001, 2007, 2008 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/list.h>
+#include <linux/string.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <asm/spitfire.h>
+#include <asm/chmctrl.h>
+#include <asm/cpudata.h>
+#include <asm/oplib.h>
+#include <asm/prom.h>
+#include <asm/head.h>
+#include <asm/io.h>
+#include <asm/memctrl.h>
+
+#define DRV_MODULE_NAME "chmc"
+#define PFX DRV_MODULE_NAME ": "
+#define DRV_MODULE_VERSION "0.2"
+
+MODULE_AUTHOR("David S. Miller (davem@davemloft.net)");
+MODULE_DESCRIPTION("UltraSPARC-III memory controller driver");
+MODULE_LICENSE("GPL");
+MODULE_VERSION(DRV_MODULE_VERSION);
+
+static int mc_type;
+#define MC_TYPE_SAFARI 1
+#define MC_TYPE_JBUS 2
+
+static dimm_printer_t us3mc_dimm_printer;
+
+#define CHMCTRL_NDGRPS 2
+#define CHMCTRL_NDIMMS 4
+
+#define CHMC_DIMMS_PER_MC (CHMCTRL_NDGRPS * CHMCTRL_NDIMMS)
+
+/* OBP memory-layout property format. */
+struct chmc_obp_map {
+ unsigned char dimm_map[144];
+ unsigned char pin_map[576];
+};
+
+#define DIMM_LABEL_SZ 8
+
+struct chmc_obp_mem_layout {
+ /* One max 8-byte string label per DIMM. Usually
+ * this matches the label on the motherboard where
+ * that DIMM resides.
+ */
+ char dimm_labels[CHMC_DIMMS_PER_MC][DIMM_LABEL_SZ];
+
+ /* If symmetric use map[0], else it is
+ * asymmetric and map[1] should be used.
+ */
+ char symmetric;
+
+ struct chmc_obp_map map[2];
+};
+
+#define CHMCTRL_NBANKS 4
+
+struct chmc_bank_info {
+ struct chmc *p;
+ int bank_id;
+
+ u64 raw_reg;
+ int valid;
+ int uk;
+ int um;
+ int lk;
+ int lm;
+ int interleave;
+ unsigned long base;
+ unsigned long size;
+};
+
+struct chmc {
+ struct list_head list;
+ int portid;
+
+ struct chmc_obp_mem_layout layout_prop;
+ int layout_size;
+
+ void __iomem *regs;
+
+ u64 timing_control1;
+ u64 timing_control2;
+ u64 timing_control3;
+ u64 timing_control4;
+ u64 memaddr_control;
+
+ struct chmc_bank_info logical_banks[CHMCTRL_NBANKS];
+};
+
+#define JBUSMC_REGS_SIZE 8
+
+#define JB_MC_REG1_DIMM2_BANK3 0x8000000000000000UL
+#define JB_MC_REG1_DIMM1_BANK1 0x4000000000000000UL
+#define JB_MC_REG1_DIMM2_BANK2 0x2000000000000000UL
+#define JB_MC_REG1_DIMM1_BANK0 0x1000000000000000UL
+#define JB_MC_REG1_XOR 0x0000010000000000UL
+#define JB_MC_REG1_ADDR_GEN_2 0x000000e000000000UL
+#define JB_MC_REG1_ADDR_GEN_2_SHIFT 37
+#define JB_MC_REG1_ADDR_GEN_1 0x0000001c00000000UL
+#define JB_MC_REG1_ADDR_GEN_1_SHIFT 34
+#define JB_MC_REG1_INTERLEAVE 0x0000000001800000UL
+#define JB_MC_REG1_INTERLEAVE_SHIFT 23
+#define JB_MC_REG1_DIMM2_PTYPE 0x0000000000200000UL
+#define JB_MC_REG1_DIMM2_PTYPE_SHIFT 21
+#define JB_MC_REG1_DIMM1_PTYPE 0x0000000000100000UL
+#define JB_MC_REG1_DIMM1_PTYPE_SHIFT 20
+
+#define PART_TYPE_X8 0
+#define PART_TYPE_X4 1
+
+#define INTERLEAVE_NONE 0
+#define INTERLEAVE_SAME 1
+#define INTERLEAVE_INTERNAL 2
+#define INTERLEAVE_BOTH 3
+
+#define ADDR_GEN_128MB 0
+#define ADDR_GEN_256MB 1
+#define ADDR_GEN_512MB 2
+#define ADDR_GEN_1GB 3
+
+#define JB_NUM_DIMM_GROUPS 2
+#define JB_NUM_DIMMS_PER_GROUP 2
+#define JB_NUM_DIMMS (JB_NUM_DIMM_GROUPS * JB_NUM_DIMMS_PER_GROUP)
+
+struct jbusmc_obp_map {
+ unsigned char dimm_map[18];
+ unsigned char pin_map[144];
+};
+
+struct jbusmc_obp_mem_layout {
+ /* One max 8-byte string label per DIMM. Usually
+ * this matches the label on the motherboard where
+ * that DIMM resides.
+ */
+ char dimm_labels[JB_NUM_DIMMS][DIMM_LABEL_SZ];
+
+ /* If symmetric use map[0], else it is
+ * asymmetric and map[1] should be used.
+ */
+ char symmetric;
+
+ struct jbusmc_obp_map map;
+
+ char _pad;
+};
+
+struct jbusmc_dimm_group {
+ struct jbusmc *controller;
+ int index;
+ u64 base_addr;
+ u64 size;
+};
+
+struct jbusmc {
+ void __iomem *regs;
+ u64 mc_reg_1;
+ u32 portid;
+ struct jbusmc_obp_mem_layout layout;
+ int layout_len;
+ int num_dimm_groups;
+ struct jbusmc_dimm_group dimm_groups[JB_NUM_DIMM_GROUPS];
+ struct list_head list;
+};
+
+static DEFINE_SPINLOCK(mctrl_list_lock);
+static LIST_HEAD(mctrl_list);
+
+static void mc_list_add(struct list_head *list)
+{
+ spin_lock(&mctrl_list_lock);
+ list_add(list, &mctrl_list);
+ spin_unlock(&mctrl_list_lock);
+}
+
+static void mc_list_del(struct list_head *list)
+{
+ spin_lock(&mctrl_list_lock);
+ list_del_init(list);
+ spin_unlock(&mctrl_list_lock);
+}
+
+#define SYNDROME_MIN -1
+#define SYNDROME_MAX 144
+
+/* Covert syndrome code into the way the bits are positioned
+ * on the bus.
+ */
+static int syndrome_to_qword_code(int syndrome_code)
+{
+ if (syndrome_code < 128)
+ syndrome_code += 16;
+ else if (syndrome_code < 128 + 9)
+ syndrome_code -= (128 - 7);
+ else if (syndrome_code < (128 + 9 + 3))
+ syndrome_code -= (128 + 9 - 4);
+ else
+ syndrome_code -= (128 + 9 + 3);
+ return syndrome_code;
+}
+
+/* All this magic has to do with how a cache line comes over the wire
+ * on Safari and JBUS. A 64-bit line comes over in 1 or more quadword
+ * cycles, each of which transmit ECC/MTAG info as well as the actual
+ * data.
+ */
+#define L2_LINE_SIZE 64
+#define L2_LINE_ADDR_MSK (L2_LINE_SIZE - 1)
+#define QW_PER_LINE 4
+#define QW_BYTES (L2_LINE_SIZE / QW_PER_LINE)
+#define QW_BITS 144
+#define SAFARI_LAST_BIT (576 - 1)
+#define JBUS_LAST_BIT (144 - 1)
+
+static void get_pin_and_dimm_str(int syndrome_code, unsigned long paddr,
+ int *pin_p, char **dimm_str_p, void *_prop,
+ int base_dimm_offset)
+{
+ int qword_code = syndrome_to_qword_code(syndrome_code);
+ int cache_line_offset;
+ int offset_inverse;
+ int dimm_map_index;
+ int map_val;
+
+ if (mc_type == MC_TYPE_JBUS) {
+ struct jbusmc_obp_mem_layout *p = _prop;
+
+ /* JBUS */
+ cache_line_offset = qword_code;
+ offset_inverse = (JBUS_LAST_BIT - cache_line_offset);
+ dimm_map_index = offset_inverse / 8;
+ map_val = p->map.dimm_map[dimm_map_index];
+ map_val = ((map_val >> ((7 - (offset_inverse & 7)))) & 1);
+ *dimm_str_p = p->dimm_labels[base_dimm_offset + map_val];
+ *pin_p = p->map.pin_map[cache_line_offset];
+ } else {
+ struct chmc_obp_mem_layout *p = _prop;
+ struct chmc_obp_map *mp;
+ int qword;
+
+ /* Safari */
+ if (p->symmetric)
+ mp = &p->map[0];
+ else
+ mp = &p->map[1];
+
+ qword = (paddr & L2_LINE_ADDR_MSK) / QW_BYTES;
+ cache_line_offset = ((3 - qword) * QW_BITS) + qword_code;
+ offset_inverse = (SAFARI_LAST_BIT - cache_line_offset);
+ dimm_map_index = offset_inverse >> 2;
+ map_val = mp->dimm_map[dimm_map_index];
+ map_val = ((map_val >> ((3 - (offset_inverse & 3)) << 1)) & 0x3);
+ *dimm_str_p = p->dimm_labels[base_dimm_offset + map_val];
+ *pin_p = mp->pin_map[cache_line_offset];
+ }
+}
+
+static struct jbusmc_dimm_group *jbusmc_find_dimm_group(unsigned long phys_addr)
+{
+ struct jbusmc *p;
+
+ list_for_each_entry(p, &mctrl_list, list) {
+ int i;
+
+ for (i = 0; i < p->num_dimm_groups; i++) {
+ struct jbusmc_dimm_group *dp = &p->dimm_groups[i];
+
+ if (phys_addr < dp->base_addr ||
+ (dp->base_addr + dp->size) <= phys_addr)
+ continue;
+
+ return dp;
+ }
+ }
+ return NULL;
+}
+
+static int jbusmc_print_dimm(int syndrome_code,
+ unsigned long phys_addr,
+ char *buf, int buflen)
+{
+ struct jbusmc_obp_mem_layout *prop;
+ struct jbusmc_dimm_group *dp;
+ struct jbusmc *p;
+ int first_dimm;
+
+ dp = jbusmc_find_dimm_group(phys_addr);
+ if (dp == NULL ||
+ syndrome_code < SYNDROME_MIN ||
+ syndrome_code > SYNDROME_MAX) {
+ buf[0] = '?';
+ buf[1] = '?';
+ buf[2] = '?';
+ buf[3] = '\0';
+ return 0;
+ }
+ p = dp->controller;
+ prop = &p->layout;
+
+ first_dimm = dp->index * JB_NUM_DIMMS_PER_GROUP;
+
+ if (syndrome_code != SYNDROME_MIN) {
+ char *dimm_str;
+ int pin;
+
+ get_pin_and_dimm_str(syndrome_code, phys_addr, &pin,
+ &dimm_str, prop, first_dimm);
+ sprintf(buf, "%s, pin %3d", dimm_str, pin);
+ } else {
+ int dimm;
+
+ /* Multi-bit error, we just dump out all the
+ * dimm labels associated with this dimm group.
+ */
+ for (dimm = 0; dimm < JB_NUM_DIMMS_PER_GROUP; dimm++) {
+ sprintf(buf, "%s ",
+ prop->dimm_labels[first_dimm + dimm]);
+ buf += strlen(buf);
+ }
+ }
+
+ return 0;
+}
+
+static u64 jbusmc_dimm_group_size(u64 base,
+ const struct linux_prom64_registers *mem_regs,
+ int num_mem_regs)
+{
+ u64 max = base + (8UL * 1024 * 1024 * 1024);
+ u64 max_seen = base;
+ int i;
+
+ for (i = 0; i < num_mem_regs; i++) {
+ const struct linux_prom64_registers *ent;
+ u64 this_base;
+ u64 this_end;
+
+ ent = &mem_regs[i];
+ this_base = ent->phys_addr;
+ this_end = this_base + ent->reg_size;
+ if (base < this_base || base >= this_end)
+ continue;
+ if (this_end > max)
+ this_end = max;
+ if (this_end > max_seen)
+ max_seen = this_end;
+ }
+
+ return max_seen - base;
+}
+
+static void jbusmc_construct_one_dimm_group(struct jbusmc *p,
+ unsigned long index,
+ const struct linux_prom64_registers *mem_regs,
+ int num_mem_regs)
+{
+ struct jbusmc_dimm_group *dp = &p->dimm_groups[index];
+
+ dp->controller = p;
+ dp->index = index;
+
+ dp->base_addr = (p->portid * (64UL * 1024 * 1024 * 1024));
+ dp->base_addr += (index * (8UL * 1024 * 1024 * 1024));
+ dp->size = jbusmc_dimm_group_size(dp->base_addr, mem_regs, num_mem_regs);
+}
+
+static void jbusmc_construct_dimm_groups(struct jbusmc *p,
+ const struct linux_prom64_registers *mem_regs,
+ int num_mem_regs)
+{
+ if (p->mc_reg_1 & JB_MC_REG1_DIMM1_BANK0) {
+ jbusmc_construct_one_dimm_group(p, 0, mem_regs, num_mem_regs);
+ p->num_dimm_groups++;
+ }
+ if (p->mc_reg_1 & JB_MC_REG1_DIMM2_BANK2) {
+ jbusmc_construct_one_dimm_group(p, 1, mem_regs, num_mem_regs);
+ p->num_dimm_groups++;
+ }
+}
+
+static int jbusmc_probe(struct platform_device *op)
+{
+ const struct linux_prom64_registers *mem_regs;
+ struct device_node *mem_node;
+ int err, len, num_mem_regs;
+ struct jbusmc *p;
+ const u32 *prop;
+ const void *ml;
+
+ err = -ENODEV;
+ mem_node = of_find_node_by_path("/memory");
+ if (!mem_node) {
+ printk(KERN_ERR PFX "Cannot find /memory node.\n");
+ goto out;
+ }
+ mem_regs = of_get_property(mem_node, "reg", &len);
+ if (!mem_regs) {
+ printk(KERN_ERR PFX "Cannot get reg property of /memory node.\n");
+ goto out;
+ }
+ num_mem_regs = len / sizeof(*mem_regs);
+
+ err = -ENOMEM;
+ p = kzalloc(sizeof(*p), GFP_KERNEL);
+ if (!p) {
+ printk(KERN_ERR PFX "Cannot allocate struct jbusmc.\n");
+ goto out;
+ }
+
+ INIT_LIST_HEAD(&p->list);
+
+ err = -ENODEV;
+ prop = of_get_property(op->dev.of_node, "portid", &len);
+ if (!prop || len != 4) {
+ printk(KERN_ERR PFX "Cannot find portid.\n");
+ goto out_free;
+ }
+
+ p->portid = *prop;
+
+ prop = of_get_property(op->dev.of_node, "memory-control-register-1", &len);
+ if (!prop || len != 8) {
+ printk(KERN_ERR PFX "Cannot get memory control register 1.\n");
+ goto out_free;
+ }
+
+ p->mc_reg_1 = ((u64)prop[0] << 32) | (u64) prop[1];
+
+ err = -ENOMEM;
+ p->regs = of_ioremap(&op->resource[0], 0, JBUSMC_REGS_SIZE, "jbusmc");
+ if (!p->regs) {
+ printk(KERN_ERR PFX "Cannot map jbusmc regs.\n");
+ goto out_free;
+ }
+
+ err = -ENODEV;
+ ml = of_get_property(op->dev.of_node, "memory-layout", &p->layout_len);
+ if (!ml) {
+ printk(KERN_ERR PFX "Cannot get memory layout property.\n");
+ goto out_iounmap;
+ }
+ if (p->layout_len > sizeof(p->layout)) {
+ printk(KERN_ERR PFX "Unexpected memory-layout size %d\n",
+ p->layout_len);
+ goto out_iounmap;
+ }
+ memcpy(&p->layout, ml, p->layout_len);
+
+ jbusmc_construct_dimm_groups(p, mem_regs, num_mem_regs);
+
+ mc_list_add(&p->list);
+
+ printk(KERN_INFO PFX "UltraSPARC-IIIi memory controller at %s\n",
+ op->dev.of_node->full_name);
+
+ dev_set_drvdata(&op->dev, p);
+
+ err = 0;
+
+out:
+ return err;
+
+out_iounmap:
+ of_iounmap(&op->resource[0], p->regs, JBUSMC_REGS_SIZE);
+
+out_free:
+ kfree(p);
+ goto out;
+}
+
+/* Does BANK decode PHYS_ADDR? */
+static int chmc_bank_match(struct chmc_bank_info *bp, unsigned long phys_addr)
+{
+ unsigned long upper_bits = (phys_addr & PA_UPPER_BITS) >> PA_UPPER_BITS_SHIFT;
+ unsigned long lower_bits = (phys_addr & PA_LOWER_BITS) >> PA_LOWER_BITS_SHIFT;
+
+ /* Bank must be enabled to match. */
+ if (bp->valid == 0)
+ return 0;
+
+ /* Would BANK match upper bits? */
+ upper_bits ^= bp->um; /* What bits are different? */
+ upper_bits = ~upper_bits; /* Invert. */
+ upper_bits |= bp->uk; /* What bits don't matter for matching? */
+ upper_bits = ~upper_bits; /* Invert. */
+
+ if (upper_bits)
+ return 0;
+
+ /* Would BANK match lower bits? */
+ lower_bits ^= bp->lm; /* What bits are different? */
+ lower_bits = ~lower_bits; /* Invert. */
+ lower_bits |= bp->lk; /* What bits don't matter for matching? */
+ lower_bits = ~lower_bits; /* Invert. */
+
+ if (lower_bits)
+ return 0;
+
+ /* I always knew you'd be the one. */
+ return 1;
+}
+
+/* Given PHYS_ADDR, search memory controller banks for a match. */
+static struct chmc_bank_info *chmc_find_bank(unsigned long phys_addr)
+{
+ struct chmc *p;
+
+ list_for_each_entry(p, &mctrl_list, list) {
+ int bank_no;
+
+ for (bank_no = 0; bank_no < CHMCTRL_NBANKS; bank_no++) {
+ struct chmc_bank_info *bp;
+
+ bp = &p->logical_banks[bank_no];
+ if (chmc_bank_match(bp, phys_addr))
+ return bp;
+ }
+ }
+
+ return NULL;
+}
+
+/* This is the main purpose of this driver. */
+static int chmc_print_dimm(int syndrome_code,
+ unsigned long phys_addr,
+ char *buf, int buflen)
+{
+ struct chmc_bank_info *bp;
+ struct chmc_obp_mem_layout *prop;
+ int bank_in_controller, first_dimm;
+
+ bp = chmc_find_bank(phys_addr);
+ if (bp == NULL ||
+ syndrome_code < SYNDROME_MIN ||
+ syndrome_code > SYNDROME_MAX) {
+ buf[0] = '?';
+ buf[1] = '?';
+ buf[2] = '?';
+ buf[3] = '\0';
+ return 0;
+ }
+
+ prop = &bp->p->layout_prop;
+ bank_in_controller = bp->bank_id & (CHMCTRL_NBANKS - 1);
+ first_dimm = (bank_in_controller & (CHMCTRL_NDGRPS - 1));
+ first_dimm *= CHMCTRL_NDIMMS;
+
+ if (syndrome_code != SYNDROME_MIN) {
+ char *dimm_str;
+ int pin;
+
+ get_pin_and_dimm_str(syndrome_code, phys_addr, &pin,
+ &dimm_str, prop, first_dimm);
+ sprintf(buf, "%s, pin %3d", dimm_str, pin);
+ } else {
+ int dimm;
+
+ /* Multi-bit error, we just dump out all the
+ * dimm labels associated with this bank.
+ */
+ for (dimm = 0; dimm < CHMCTRL_NDIMMS; dimm++) {
+ sprintf(buf, "%s ",
+ prop->dimm_labels[first_dimm + dimm]);
+ buf += strlen(buf);
+ }
+ }
+ return 0;
+}
+
+/* Accessing the registers is slightly complicated. If you want
+ * to get at the memory controller which is on the same processor
+ * the code is executing, you must use special ASI load/store else
+ * you go through the global mapping.
+ */
+static u64 chmc_read_mcreg(struct chmc *p, unsigned long offset)
+{
+ unsigned long ret, this_cpu;
+
+ preempt_disable();
+
+ this_cpu = real_hard_smp_processor_id();
+
+ if (p->portid == this_cpu) {
+ __asm__ __volatile__("ldxa [%1] %2, %0"
+ : "=r" (ret)
+ : "r" (offset), "i" (ASI_MCU_CTRL_REG));
+ } else {
+ __asm__ __volatile__("ldxa [%1] %2, %0"
+ : "=r" (ret)
+ : "r" (p->regs + offset),
+ "i" (ASI_PHYS_BYPASS_EC_E));
+ }
+
+ preempt_enable();
+
+ return ret;
+}
+
+#if 0 /* currently unused */
+static void chmc_write_mcreg(struct chmc *p, unsigned long offset, u64 val)
+{
+ if (p->portid == smp_processor_id()) {
+ __asm__ __volatile__("stxa %0, [%1] %2"
+ : : "r" (val),
+ "r" (offset), "i" (ASI_MCU_CTRL_REG));
+ } else {
+ __asm__ __volatile__("ldxa %0, [%1] %2"
+ : : "r" (val),
+ "r" (p->regs + offset),
+ "i" (ASI_PHYS_BYPASS_EC_E));
+ }
+}
+#endif
+
+static void chmc_interpret_one_decode_reg(struct chmc *p, int which_bank, u64 val)
+{
+ struct chmc_bank_info *bp = &p->logical_banks[which_bank];
+
+ bp->p = p;
+ bp->bank_id = (CHMCTRL_NBANKS * p->portid) + which_bank;
+ bp->raw_reg = val;
+ bp->valid = (val & MEM_DECODE_VALID) >> MEM_DECODE_VALID_SHIFT;
+ bp->uk = (val & MEM_DECODE_UK) >> MEM_DECODE_UK_SHIFT;
+ bp->um = (val & MEM_DECODE_UM) >> MEM_DECODE_UM_SHIFT;
+ bp->lk = (val & MEM_DECODE_LK) >> MEM_DECODE_LK_SHIFT;
+ bp->lm = (val & MEM_DECODE_LM) >> MEM_DECODE_LM_SHIFT;
+
+ bp->base = (bp->um);
+ bp->base &= ~(bp->uk);
+ bp->base <<= PA_UPPER_BITS_SHIFT;
+
+ switch(bp->lk) {
+ case 0xf:
+ default:
+ bp->interleave = 1;
+ break;
+
+ case 0xe:
+ bp->interleave = 2;
+ break;
+
+ case 0xc:
+ bp->interleave = 4;
+ break;
+
+ case 0x8:
+ bp->interleave = 8;
+ break;
+
+ case 0x0:
+ bp->interleave = 16;
+ break;
+ }
+
+ /* UK[10] is reserved, and UK[11] is not set for the SDRAM
+ * bank size definition.
+ */
+ bp->size = (((unsigned long)bp->uk &
+ ((1UL << 10UL) - 1UL)) + 1UL) << PA_UPPER_BITS_SHIFT;
+ bp->size /= bp->interleave;
+}
+
+static void chmc_fetch_decode_regs(struct chmc *p)
+{
+ if (p->layout_size == 0)
+ return;
+
+ chmc_interpret_one_decode_reg(p, 0,
+ chmc_read_mcreg(p, CHMCTRL_DECODE1));
+ chmc_interpret_one_decode_reg(p, 1,
+ chmc_read_mcreg(p, CHMCTRL_DECODE2));
+ chmc_interpret_one_decode_reg(p, 2,
+ chmc_read_mcreg(p, CHMCTRL_DECODE3));
+ chmc_interpret_one_decode_reg(p, 3,
+ chmc_read_mcreg(p, CHMCTRL_DECODE4));
+}
+
+static int chmc_probe(struct platform_device *op)
+{
+ struct device_node *dp = op->dev.of_node;
+ unsigned long ver;
+ const void *pval;
+ int len, portid;
+ struct chmc *p;
+ int err;
+
+ err = -ENODEV;
+ __asm__ ("rdpr %%ver, %0" : "=r" (ver));
+ if ((ver >> 32UL) == __JALAPENO_ID ||
+ (ver >> 32UL) == __SERRANO_ID)
+ goto out;
+
+ portid = of_getintprop_default(dp, "portid", -1);
+ if (portid == -1)
+ goto out;
+
+ pval = of_get_property(dp, "memory-layout", &len);
+ if (pval && len > sizeof(p->layout_prop)) {
+ printk(KERN_ERR PFX "Unexpected memory-layout property "
+ "size %d.\n", len);
+ goto out;
+ }
+
+ err = -ENOMEM;
+ p = kzalloc(sizeof(*p), GFP_KERNEL);
+ if (!p) {
+ printk(KERN_ERR PFX "Could not allocate struct chmc.\n");
+ goto out;
+ }
+
+ p->portid = portid;
+ p->layout_size = len;
+ if (!pval)
+ p->layout_size = 0;
+ else
+ memcpy(&p->layout_prop, pval, len);
+
+ p->regs = of_ioremap(&op->resource[0], 0, 0x48, "chmc");
+ if (!p->regs) {
+ printk(KERN_ERR PFX "Could not map registers.\n");
+ goto out_free;
+ }
+
+ if (p->layout_size != 0UL) {
+ p->timing_control1 = chmc_read_mcreg(p, CHMCTRL_TCTRL1);
+ p->timing_control2 = chmc_read_mcreg(p, CHMCTRL_TCTRL2);
+ p->timing_control3 = chmc_read_mcreg(p, CHMCTRL_TCTRL3);
+ p->timing_control4 = chmc_read_mcreg(p, CHMCTRL_TCTRL4);
+ p->memaddr_control = chmc_read_mcreg(p, CHMCTRL_MACTRL);
+ }
+
+ chmc_fetch_decode_regs(p);
+
+ mc_list_add(&p->list);
+
+ printk(KERN_INFO PFX "UltraSPARC-III memory controller at %s [%s]\n",
+ dp->full_name,
+ (p->layout_size ? "ACTIVE" : "INACTIVE"));
+
+ dev_set_drvdata(&op->dev, p);
+
+ err = 0;
+
+out:
+ return err;
+
+out_free:
+ kfree(p);
+ goto out;
+}
+
+static int us3mc_probe(struct platform_device *op)
+{
+ if (mc_type == MC_TYPE_SAFARI)
+ return chmc_probe(op);
+ else if (mc_type == MC_TYPE_JBUS)
+ return jbusmc_probe(op);
+ return -ENODEV;
+}
+
+static void chmc_destroy(struct platform_device *op, struct chmc *p)
+{
+ list_del(&p->list);
+ of_iounmap(&op->resource[0], p->regs, 0x48);
+ kfree(p);
+}
+
+static void jbusmc_destroy(struct platform_device *op, struct jbusmc *p)
+{
+ mc_list_del(&p->list);
+ of_iounmap(&op->resource[0], p->regs, JBUSMC_REGS_SIZE);
+ kfree(p);
+}
+
+static int us3mc_remove(struct platform_device *op)
+{
+ void *p = dev_get_drvdata(&op->dev);
+
+ if (p) {
+ if (mc_type == MC_TYPE_SAFARI)
+ chmc_destroy(op, p);
+ else if (mc_type == MC_TYPE_JBUS)
+ jbusmc_destroy(op, p);
+ }
+ return 0;
+}
+
+static const struct of_device_id us3mc_match[] = {
+ {
+ .name = "memory-controller",
+ },
+ {},
+};
+MODULE_DEVICE_TABLE(of, us3mc_match);
+
+static struct platform_driver us3mc_driver = {
+ .driver = {
+ .name = "us3mc",
+ .owner = THIS_MODULE,
+ .of_match_table = us3mc_match,
+ },
+ .probe = us3mc_probe,
+ .remove = us3mc_remove,
+};
+
+static inline bool us3mc_platform(void)
+{
+ if (tlb_type == cheetah || tlb_type == cheetah_plus)
+ return true;
+ return false;
+}
+
+static int __init us3mc_init(void)
+{
+ unsigned long ver;
+ int ret;
+
+ if (!us3mc_platform())
+ return -ENODEV;
+
+ __asm__ __volatile__("rdpr %%ver, %0" : "=r" (ver));
+ if ((ver >> 32UL) == __JALAPENO_ID ||
+ (ver >> 32UL) == __SERRANO_ID) {
+ mc_type = MC_TYPE_JBUS;
+ us3mc_dimm_printer = jbusmc_print_dimm;
+ } else {
+ mc_type = MC_TYPE_SAFARI;
+ us3mc_dimm_printer = chmc_print_dimm;
+ }
+
+ ret = register_dimm_printer(us3mc_dimm_printer);
+
+ if (!ret) {
+ ret = platform_driver_register(&us3mc_driver);
+ if (ret)
+ unregister_dimm_printer(us3mc_dimm_printer);
+ }
+ return ret;
+}
+
+static void __exit us3mc_cleanup(void)
+{
+ if (us3mc_platform()) {
+ unregister_dimm_printer(us3mc_dimm_printer);
+ platform_driver_unregister(&us3mc_driver);
+ }
+}
+
+module_init(us3mc_init);
+module_exit(us3mc_cleanup);
diff --git a/arch/sparc/kernel/compat_audit.c b/arch/sparc/kernel/compat_audit.c
new file mode 100644
index 00000000000..7062263d09c
--- /dev/null
+++ b/arch/sparc/kernel/compat_audit.c
@@ -0,0 +1,44 @@
+#define __32bit_syscall_numbers__
+#include <asm/unistd.h>
+#include "kernel.h"
+
+unsigned sparc32_dir_class[] = {
+#include <asm-generic/audit_dir_write.h>
+~0U
+};
+
+unsigned sparc32_chattr_class[] = {
+#include <asm-generic/audit_change_attr.h>
+~0U
+};
+
+unsigned sparc32_write_class[] = {
+#include <asm-generic/audit_write.h>
+~0U
+};
+
+unsigned sparc32_read_class[] = {
+#include <asm-generic/audit_read.h>
+~0U
+};
+
+unsigned sparc32_signal_class[] = {
+#include <asm-generic/audit_signal.h>
+~0U
+};
+
+int sparc32_classify_syscall(unsigned syscall)
+{
+ switch(syscall) {
+ case __NR_open:
+ return 2;
+ case __NR_openat:
+ return 3;
+ case __NR_socketcall:
+ return 4;
+ case __NR_execve:
+ return 5;
+ default:
+ return 1;
+ }
+}
diff --git a/arch/sparc/kernel/cpu.c b/arch/sparc/kernel/cpu.c
index 259a559d4ce..82a3a71c451 100644
--- a/arch/sparc/kernel/cpu.c
+++ b/arch/sparc/kernel/cpu.c
@@ -4,164 +4,528 @@
* Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
*/
+#include <linux/seq_file.h>
#include <linux/kernel.h>
+#include <linux/export.h>
#include <linux/init.h>
#include <linux/smp.h>
#include <linux/threads.h>
+
+#include <asm/spitfire.h>
+#include <asm/pgtable.h>
#include <asm/oplib.h>
+#include <asm/setup.h>
#include <asm/page.h>
#include <asm/head.h>
#include <asm/psr.h>
#include <asm/mbus.h>
#include <asm/cpudata.h>
+#include "kernel.h"
+#include "entry.h"
+
DEFINE_PER_CPU(cpuinfo_sparc, __cpu_data) = { 0 };
+EXPORT_PER_CPU_SYMBOL(__cpu_data);
+
+int ncpus_probed;
+unsigned int fsr_storage;
-struct cpu_iu_info {
- int psr_impl;
- int psr_vers;
- char* cpu_name; /* should be enough I hope... */
+struct cpu_info {
+ int psr_vers;
+ const char *name;
+ const char *pmu_name;
};
-struct cpu_fp_info {
- int psr_impl;
- int fp_vers;
- char* fp_name;
+struct fpu_info {
+ int fp_vers;
+ const char *name;
};
+#define NOCPU 8
+#define NOFPU 8
+
+struct manufacturer_info {
+ int psr_impl;
+ struct cpu_info cpu_info[NOCPU];
+ struct fpu_info fpu_info[NOFPU];
+};
+
+#define CPU(ver, _name) \
+{ .psr_vers = ver, .name = _name }
+
+#define CPU_PMU(ver, _name, _pmu_name) \
+{ .psr_vers = ver, .name = _name, .pmu_name = _pmu_name }
+
+#define FPU(ver, _name) \
+{ .fp_vers = ver, .name = _name }
+
+static const struct manufacturer_info __initconst manufacturer_info[] = {
+{
+ 0,
+ /* Sun4/100, 4/200, SLC */
+ .cpu_info = {
+ CPU(0, "Fujitsu MB86900/1A or LSI L64831 SparcKIT-40"),
+ /* borned STP1012PGA */
+ CPU(4, "Fujitsu MB86904"),
+ CPU(5, "Fujitsu TurboSparc MB86907"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(0, "Fujitsu MB86910 or Weitek WTL1164/5"),
+ FPU(1, "Fujitsu MB86911 or Weitek WTL1164/5 or LSI L64831"),
+ FPU(2, "LSI Logic L64802 or Texas Instruments ACT8847"),
+ /* SparcStation SLC, SparcStation1 */
+ FPU(3, "Weitek WTL3170/2"),
+ /* SPARCstation-5 */
+ FPU(4, "Lsi Logic/Meiko L64804 or compatible"),
+ FPU(-1, NULL)
+ }
+},{
+ 1,
+ .cpu_info = {
+ /* SparcStation2, SparcServer 490 & 690 */
+ CPU(0, "LSI Logic Corporation - L64811"),
+ /* SparcStation2 */
+ CPU(1, "Cypress/ROSS CY7C601"),
+ /* Embedded controller */
+ CPU(3, "Cypress/ROSS CY7C611"),
+ /* Ross Technologies HyperSparc */
+ CPU(0xf, "ROSS HyperSparc RT620"),
+ CPU(0xe, "ROSS HyperSparc RT625 or RT626"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(0, "ROSS HyperSparc combined IU/FPU"),
+ FPU(1, "Lsi Logic L64814"),
+ FPU(2, "Texas Instruments TMS390-C602A"),
+ FPU(3, "Cypress CY7C602 FPU"),
+ FPU(-1, NULL)
+ }
+},{
+ 2,
+ .cpu_info = {
+ /* ECL Implementation, CRAY S-MP Supercomputer... AIEEE! */
+ /* Someone please write the code to support this beast! ;) */
+ CPU(0, "Bipolar Integrated Technology - B5010"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(-1, NULL)
+ }
+},{
+ 3,
+ .cpu_info = {
+ CPU(0, "LSI Logic Corporation - unknown-type"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(-1, NULL)
+ }
+},{
+ PSR_IMPL_TI,
+ .cpu_info = {
+ CPU(0, "Texas Instruments, Inc. - SuperSparc-(II)"),
+ /* SparcClassic -- borned STP1010TAB-50*/
+ CPU(1, "Texas Instruments, Inc. - MicroSparc"),
+ CPU(2, "Texas Instruments, Inc. - MicroSparc II"),
+ CPU(3, "Texas Instruments, Inc. - SuperSparc 51"),
+ CPU(4, "Texas Instruments, Inc. - SuperSparc 61"),
+ CPU(5, "Texas Instruments, Inc. - unknown"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ /* SuperSparc 50 module */
+ FPU(0, "SuperSparc on-chip FPU"),
+ /* SparcClassic */
+ FPU(4, "TI MicroSparc on chip FPU"),
+ FPU(-1, NULL)
+ }
+},{
+ 5,
+ .cpu_info = {
+ CPU(0, "Matsushita - MN10501"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(0, "Matsushita MN10501"),
+ FPU(-1, NULL)
+ }
+},{
+ 6,
+ .cpu_info = {
+ CPU(0, "Philips Corporation - unknown"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(-1, NULL)
+ }
+},{
+ 7,
+ .cpu_info = {
+ CPU(0, "Harvest VLSI Design Center, Inc. - unknown"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(-1, NULL)
+ }
+},{
+ 8,
+ .cpu_info = {
+ CPU(0, "Systems and Processes Engineering Corporation (SPEC)"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(-1, NULL)
+ }
+},{
+ 9,
+ .cpu_info = {
+ /* Gallium arsenide 200MHz, BOOOOGOOOOMIPS!!! */
+ CPU(0, "Fujitsu or Weitek Power-UP"),
+ CPU(1, "Fujitsu or Weitek Power-UP"),
+ CPU(2, "Fujitsu or Weitek Power-UP"),
+ CPU(3, "Fujitsu or Weitek Power-UP"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(3, "Fujitsu or Weitek on-chip FPU"),
+ FPU(-1, NULL)
+ }
+},{
+ PSR_IMPL_LEON, /* Aeroflex Gaisler */
+ .cpu_info = {
+ CPU(3, "LEON"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(2, "GRFPU"),
+ FPU(3, "GRFPU-Lite"),
+ FPU(-1, NULL)
+ }
+},{
+ 0x17,
+ .cpu_info = {
+ CPU_PMU(0x10, "TI UltraSparc I (SpitFire)", "ultra12"),
+ CPU_PMU(0x11, "TI UltraSparc II (BlackBird)", "ultra12"),
+ CPU_PMU(0x12, "TI UltraSparc IIi (Sabre)", "ultra12"),
+ CPU_PMU(0x13, "TI UltraSparc IIe (Hummingbird)", "ultra12"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(0x10, "UltraSparc I integrated FPU"),
+ FPU(0x11, "UltraSparc II integrated FPU"),
+ FPU(0x12, "UltraSparc IIi integrated FPU"),
+ FPU(0x13, "UltraSparc IIe integrated FPU"),
+ FPU(-1, NULL)
+ }
+},{
+ 0x22,
+ .cpu_info = {
+ CPU_PMU(0x10, "TI UltraSparc I (SpitFire)", "ultra12"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(0x10, "UltraSparc I integrated FPU"),
+ FPU(-1, NULL)
+ }
+},{
+ 0x3e,
+ .cpu_info = {
+ CPU_PMU(0x14, "TI UltraSparc III (Cheetah)", "ultra3"),
+ CPU_PMU(0x15, "TI UltraSparc III+ (Cheetah+)", "ultra3+"),
+ CPU_PMU(0x16, "TI UltraSparc IIIi (Jalapeno)", "ultra3i"),
+ CPU_PMU(0x18, "TI UltraSparc IV (Jaguar)", "ultra3+"),
+ CPU_PMU(0x19, "TI UltraSparc IV+ (Panther)", "ultra4+"),
+ CPU_PMU(0x22, "TI UltraSparc IIIi+ (Serrano)", "ultra3i"),
+ CPU(-1, NULL)
+ },
+ .fpu_info = {
+ FPU(0x14, "UltraSparc III integrated FPU"),
+ FPU(0x15, "UltraSparc III+ integrated FPU"),
+ FPU(0x16, "UltraSparc IIIi integrated FPU"),
+ FPU(0x18, "UltraSparc IV integrated FPU"),
+ FPU(0x19, "UltraSparc IV+ integrated FPU"),
+ FPU(0x22, "UltraSparc IIIi+ integrated FPU"),
+ FPU(-1, NULL)
+ }
+}};
+
/* In order to get the fpu type correct, you need to take the IDPROM's
* machine type value into consideration too. I will fix this.
*/
-struct cpu_fp_info linux_sparc_fpu[] = {
- { 0, 0, "Fujitsu MB86910 or Weitek WTL1164/5"},
- { 0, 1, "Fujitsu MB86911 or Weitek WTL1164/5 or LSI L64831"},
- { 0, 2, "LSI Logic L64802 or Texas Instruments ACT8847"},
- /* SparcStation SLC, SparcStation1 */
- { 0, 3, "Weitek WTL3170/2"},
- /* SPARCstation-5 */
- { 0, 4, "Lsi Logic/Meiko L64804 or compatible"},
- { 0, 5, "reserved"},
- { 0, 6, "reserved"},
- { 0, 7, "No FPU"},
- { 1, 0, "ROSS HyperSparc combined IU/FPU"},
- { 1, 1, "Lsi Logic L64814"},
- { 1, 2, "Texas Instruments TMS390-C602A"},
- { 1, 3, "Cypress CY7C602 FPU"},
- { 1, 4, "reserved"},
- { 1, 5, "reserved"},
- { 1, 6, "reserved"},
- { 1, 7, "No FPU"},
- { 2, 0, "BIT B5010 or B5110/20 or B5210"},
- { 2, 1, "reserved"},
- { 2, 2, "reserved"},
- { 2, 3, "reserved"},
- { 2, 4, "reserved"},
- { 2, 5, "reserved"},
- { 2, 6, "reserved"},
- { 2, 7, "No FPU"},
- /* SuperSparc 50 module */
- { 4, 0, "SuperSparc on-chip FPU"},
- /* SparcClassic */
- { 4, 4, "TI MicroSparc on chip FPU"},
- { 5, 0, "Matsushita MN10501"},
- { 5, 1, "reserved"},
- { 5, 2, "reserved"},
- { 5, 3, "reserved"},
- { 5, 4, "reserved"},
- { 5, 5, "reserved"},
- { 5, 6, "reserved"},
- { 5, 7, "No FPU"},
- { 9, 3, "Fujitsu or Weitek on-chip FPU"},
-};
-#define NSPARCFPU ARRAY_SIZE(linux_sparc_fpu)
-
-struct cpu_iu_info linux_sparc_chips[] = {
- /* Sun4/100, 4/200, SLC */
- { 0, 0, "Fujitsu MB86900/1A or LSI L64831 SparcKIT-40"},
- /* borned STP1012PGA */
- { 0, 4, "Fujitsu MB86904"},
- { 0, 5, "Fujitsu TurboSparc MB86907"},
- /* SparcStation2, SparcServer 490 & 690 */
- { 1, 0, "LSI Logic Corporation - L64811"},
- /* SparcStation2 */
- { 1, 1, "Cypress/ROSS CY7C601"},
- /* Embedded controller */
- { 1, 3, "Cypress/ROSS CY7C611"},
- /* Ross Technologies HyperSparc */
- { 1, 0xf, "ROSS HyperSparc RT620"},
- { 1, 0xe, "ROSS HyperSparc RT625 or RT626"},
- /* ECL Implementation, CRAY S-MP Supercomputer... AIEEE! */
- /* Someone please write the code to support this beast! ;) */
- { 2, 0, "Bipolar Integrated Technology - B5010"},
- { 3, 0, "LSI Logic Corporation - unknown-type"},
- { 4, 0, "Texas Instruments, Inc. - SuperSparc-(II)"},
- /* SparcClassic -- borned STP1010TAB-50*/
- { 4, 1, "Texas Instruments, Inc. - MicroSparc"},
- { 4, 2, "Texas Instruments, Inc. - MicroSparc II"},
- { 4, 3, "Texas Instruments, Inc. - SuperSparc 51"},
- { 4, 4, "Texas Instruments, Inc. - SuperSparc 61"},
- { 4, 5, "Texas Instruments, Inc. - unknown"},
- { 5, 0, "Matsushita - MN10501"},
- { 6, 0, "Philips Corporation - unknown"},
- { 7, 0, "Harvest VLSI Design Center, Inc. - unknown"},
- /* Gallium arsenide 200MHz, BOOOOGOOOOMIPS!!! */
- { 8, 0, "Systems and Processes Engineering Corporation (SPEC)"},
- { 9, 0, "Fujitsu or Weitek Power-UP"},
- { 9, 1, "Fujitsu or Weitek Power-UP"},
- { 9, 2, "Fujitsu or Weitek Power-UP"},
- { 9, 3, "Fujitsu or Weitek Power-UP"},
- { 0xa, 0, "UNKNOWN CPU-VENDOR/TYPE"},
- { 0xb, 0, "UNKNOWN CPU-VENDOR/TYPE"},
- { 0xc, 0, "UNKNOWN CPU-VENDOR/TYPE"},
- { 0xd, 0, "UNKNOWN CPU-VENDOR/TYPE"},
- { 0xe, 0, "UNKNOWN CPU-VENDOR/TYPE"},
- { 0xf, 0, "UNKNOWN CPU-VENDOR/TYPE"},
-};
+static const char *sparc_cpu_type;
+static const char *sparc_fpu_type;
+const char *sparc_pmu_type;
-#define NSPARCCHIPS ARRAY_SIZE(linux_sparc_chips)
-char *sparc_cpu_type;
-char *sparc_fpu_type;
+static void __init set_cpu_and_fpu(int psr_impl, int psr_vers, int fpu_vers)
+{
+ const struct manufacturer_info *manuf;
+ int i;
-unsigned int fsr_storage;
+ sparc_cpu_type = NULL;
+ sparc_fpu_type = NULL;
+ sparc_pmu_type = NULL;
+ manuf = NULL;
+
+ for (i = 0; i < ARRAY_SIZE(manufacturer_info); i++)
+ {
+ if (psr_impl == manufacturer_info[i].psr_impl) {
+ manuf = &manufacturer_info[i];
+ break;
+ }
+ }
+ if (manuf != NULL)
+ {
+ const struct cpu_info *cpu;
+ const struct fpu_info *fpu;
+
+ cpu = &manuf->cpu_info[0];
+ while (cpu->psr_vers != -1)
+ {
+ if (cpu->psr_vers == psr_vers) {
+ sparc_cpu_type = cpu->name;
+ sparc_pmu_type = cpu->pmu_name;
+ sparc_fpu_type = "No FPU";
+ break;
+ }
+ cpu++;
+ }
+ fpu = &manuf->fpu_info[0];
+ while (fpu->fp_vers != -1)
+ {
+ if (fpu->fp_vers == fpu_vers) {
+ sparc_fpu_type = fpu->name;
+ break;
+ }
+ fpu++;
+ }
+ }
+ if (sparc_cpu_type == NULL)
+ {
+ printk(KERN_ERR "CPU: Unknown chip, impl[0x%x] vers[0x%x]\n",
+ psr_impl, psr_vers);
+ sparc_cpu_type = "Unknown CPU";
+ }
+ if (sparc_fpu_type == NULL)
+ {
+ printk(KERN_ERR "FPU: Unknown chip, impl[0x%x] vers[0x%x]\n",
+ psr_impl, fpu_vers);
+ sparc_fpu_type = "Unknown FPU";
+ }
+ if (sparc_pmu_type == NULL)
+ sparc_pmu_type = "Unknown PMU";
+}
-void __init cpu_probe(void)
+#ifdef CONFIG_SPARC32
+static int show_cpuinfo(struct seq_file *m, void *__unused)
+{
+ seq_printf(m,
+ "cpu\t\t: %s\n"
+ "fpu\t\t: %s\n"
+ "promlib\t\t: Version %d Revision %d\n"
+ "prom\t\t: %d.%d\n"
+ "type\t\t: %s\n"
+ "ncpus probed\t: %d\n"
+ "ncpus active\t: %d\n"
+#ifndef CONFIG_SMP
+ "CPU0Bogo\t: %lu.%02lu\n"
+ "CPU0ClkTck\t: %ld\n"
+#endif
+ ,
+ sparc_cpu_type,
+ sparc_fpu_type ,
+ romvec->pv_romvers,
+ prom_rev,
+ romvec->pv_printrev >> 16,
+ romvec->pv_printrev & 0xffff,
+ &cputypval[0],
+ ncpus_probed,
+ num_online_cpus()
+#ifndef CONFIG_SMP
+ , cpu_data(0).udelay_val/(500000/HZ),
+ (cpu_data(0).udelay_val/(5000/HZ)) % 100,
+ cpu_data(0).clock_tick
+#endif
+ );
+
+#ifdef CONFIG_SMP
+ smp_bogo(m);
+#endif
+ mmu_info(m);
+#ifdef CONFIG_SMP
+ smp_info(m);
+#endif
+ return 0;
+}
+#endif /* CONFIG_SPARC32 */
+
+#ifdef CONFIG_SPARC64
+unsigned int dcache_parity_tl1_occurred;
+unsigned int icache_parity_tl1_occurred;
+
+
+static int show_cpuinfo(struct seq_file *m, void *__unused)
+{
+ seq_printf(m,
+ "cpu\t\t: %s\n"
+ "fpu\t\t: %s\n"
+ "pmu\t\t: %s\n"
+ "prom\t\t: %s\n"
+ "type\t\t: %s\n"
+ "ncpus probed\t: %d\n"
+ "ncpus active\t: %d\n"
+ "D$ parity tl1\t: %u\n"
+ "I$ parity tl1\t: %u\n"
+#ifndef CONFIG_SMP
+ "Cpu0ClkTck\t: %016lx\n"
+#endif
+ ,
+ sparc_cpu_type,
+ sparc_fpu_type,
+ sparc_pmu_type,
+ prom_version,
+ ((tlb_type == hypervisor) ?
+ "sun4v" :
+ "sun4u"),
+ ncpus_probed,
+ num_online_cpus(),
+ dcache_parity_tl1_occurred,
+ icache_parity_tl1_occurred
+#ifndef CONFIG_SMP
+ , cpu_data(0).clock_tick
+#endif
+ );
+ cpucap_info(m);
+#ifdef CONFIG_SMP
+ smp_bogo(m);
+#endif
+ mmu_info(m);
+#ifdef CONFIG_SMP
+ smp_info(m);
+#endif
+ return 0;
+}
+#endif /* CONFIG_SPARC64 */
+
+static void *c_start(struct seq_file *m, loff_t *pos)
+{
+ /* The pointer we are returning is arbitrary,
+ * it just has to be non-NULL and not IS_ERR
+ * in the success case.
+ */
+ return *pos == 0 ? &c_start : NULL;
+}
+
+static void *c_next(struct seq_file *m, void *v, loff_t *pos)
+{
+ ++*pos;
+ return c_start(m, pos);
+}
+
+static void c_stop(struct seq_file *m, void *v)
+{
+}
+
+const struct seq_operations cpuinfo_op = {
+ .start =c_start,
+ .next = c_next,
+ .stop = c_stop,
+ .show = show_cpuinfo,
+};
+
+#ifdef CONFIG_SPARC32
+static int __init cpu_type_probe(void)
{
int psr_impl, psr_vers, fpu_vers;
- int i, psr;
+ int psr;
- psr_impl = ((get_psr()>>28)&0xf);
- psr_vers = ((get_psr()>>24)&0xf);
+ psr_impl = ((get_psr() >> PSR_IMPL_SHIFT) & PSR_IMPL_SHIFTED_MASK);
+ psr_vers = ((get_psr() >> PSR_VERS_SHIFT) & PSR_VERS_SHIFTED_MASK);
psr = get_psr();
put_psr(psr | PSR_EF);
- fpu_vers = ((get_fsr()>>17)&0x7);
+
+ if (psr_impl == PSR_IMPL_LEON)
+ fpu_vers = get_psr() & PSR_EF ? ((get_fsr() >> 17) & 0x7) : 7;
+ else
+ fpu_vers = ((get_fsr() >> 17) & 0x7);
+
put_psr(psr);
- for(i = 0; i<NSPARCCHIPS; i++) {
- if(linux_sparc_chips[i].psr_impl == psr_impl)
- if(linux_sparc_chips[i].psr_vers == psr_vers) {
- sparc_cpu_type = linux_sparc_chips[i].cpu_name;
- break;
- }
- }
+ set_cpu_and_fpu(psr_impl, psr_vers, fpu_vers);
- if(i==NSPARCCHIPS)
- printk("DEBUG: psr.impl = 0x%x psr.vers = 0x%x\n", psr_impl,
- psr_vers);
+ return 0;
+}
+#endif /* CONFIG_SPARC32 */
- for(i = 0; i<NSPARCFPU; i++) {
- if(linux_sparc_fpu[i].psr_impl == psr_impl)
- if(linux_sparc_fpu[i].fp_vers == fpu_vers) {
- sparc_fpu_type = linux_sparc_fpu[i].fp_name;
- break;
- }
+#ifdef CONFIG_SPARC64
+static void __init sun4v_cpu_probe(void)
+{
+ switch (sun4v_chip_type) {
+ case SUN4V_CHIP_NIAGARA1:
+ sparc_cpu_type = "UltraSparc T1 (Niagara)";
+ sparc_fpu_type = "UltraSparc T1 integrated FPU";
+ sparc_pmu_type = "niagara";
+ break;
+
+ case SUN4V_CHIP_NIAGARA2:
+ sparc_cpu_type = "UltraSparc T2 (Niagara2)";
+ sparc_fpu_type = "UltraSparc T2 integrated FPU";
+ sparc_pmu_type = "niagara2";
+ break;
+
+ case SUN4V_CHIP_NIAGARA3:
+ sparc_cpu_type = "UltraSparc T3 (Niagara3)";
+ sparc_fpu_type = "UltraSparc T3 integrated FPU";
+ sparc_pmu_type = "niagara3";
+ break;
+
+ case SUN4V_CHIP_NIAGARA4:
+ sparc_cpu_type = "UltraSparc T4 (Niagara4)";
+ sparc_fpu_type = "UltraSparc T4 integrated FPU";
+ sparc_pmu_type = "niagara4";
+ break;
+
+ case SUN4V_CHIP_NIAGARA5:
+ sparc_cpu_type = "UltraSparc T5 (Niagara5)";
+ sparc_fpu_type = "UltraSparc T5 integrated FPU";
+ sparc_pmu_type = "niagara5";
+ break;
+
+ case SUN4V_CHIP_SPARC64X:
+ sparc_cpu_type = "SPARC64-X";
+ sparc_fpu_type = "SPARC64-X integrated FPU";
+ sparc_pmu_type = "sparc64-x";
+ break;
+
+ default:
+ printk(KERN_WARNING "CPU: Unknown sun4v cpu type [%s]\n",
+ prom_cpu_compatible);
+ sparc_cpu_type = "Unknown SUN4V CPU";
+ sparc_fpu_type = "Unknown SUN4V FPU";
+ sparc_pmu_type = "Unknown SUN4V PMU";
+ break;
}
+}
+
+static int __init cpu_type_probe(void)
+{
+ if (tlb_type == hypervisor) {
+ sun4v_cpu_probe();
+ } else {
+ unsigned long ver;
+ int manuf, impl;
- if(i == NSPARCFPU) {
- printk("DEBUG: psr.impl = 0x%x fsr.vers = 0x%x\n", psr_impl,
- fpu_vers);
- sparc_fpu_type = linux_sparc_fpu[31].fp_name;
+ __asm__ __volatile__("rdpr %%ver, %0" : "=r" (ver));
+
+ manuf = ((ver >> 48) & 0xffff);
+ impl = ((ver >> 32) & 0xffff);
+ set_cpu_and_fpu(manuf, impl, impl);
}
+ return 0;
}
+#endif /* CONFIG_SPARC64 */
+
+early_initcall(cpu_type_probe);
diff --git a/arch/sparc/kernel/cpumap.c b/arch/sparc/kernel/cpumap.c
new file mode 100644
index 00000000000..de1c844dfab
--- /dev/null
+++ b/arch/sparc/kernel/cpumap.c
@@ -0,0 +1,435 @@
+/* cpumap.c: used for optimizing CPU assignment
+ *
+ * Copyright (C) 2009 Hong H. Pham <hong.pham@windriver.com>
+ */
+
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/kernel.h>
+#include <linux/cpumask.h>
+#include <linux/spinlock.h>
+#include <asm/cpudata.h>
+#include "cpumap.h"
+
+
+enum {
+ CPUINFO_LVL_ROOT = 0,
+ CPUINFO_LVL_NODE,
+ CPUINFO_LVL_CORE,
+ CPUINFO_LVL_PROC,
+ CPUINFO_LVL_MAX,
+};
+
+enum {
+ ROVER_NO_OP = 0,
+ /* Increment rover every time level is visited */
+ ROVER_INC_ON_VISIT = 1 << 0,
+ /* Increment parent's rover every time rover wraps around */
+ ROVER_INC_PARENT_ON_LOOP = 1 << 1,
+};
+
+struct cpuinfo_node {
+ int id;
+ int level;
+ int num_cpus; /* Number of CPUs in this hierarchy */
+ int parent_index;
+ int child_start; /* Array index of the first child node */
+ int child_end; /* Array index of the last child node */
+ int rover; /* Child node iterator */
+};
+
+struct cpuinfo_level {
+ int start_index; /* Index of first node of a level in a cpuinfo tree */
+ int end_index; /* Index of last node of a level in a cpuinfo tree */
+ int num_nodes; /* Number of nodes in a level in a cpuinfo tree */
+};
+
+struct cpuinfo_tree {
+ int total_nodes;
+
+ /* Offsets into nodes[] for each level of the tree */
+ struct cpuinfo_level level[CPUINFO_LVL_MAX];
+ struct cpuinfo_node nodes[0];
+};
+
+
+static struct cpuinfo_tree *cpuinfo_tree;
+
+static u16 cpu_distribution_map[NR_CPUS];
+static DEFINE_SPINLOCK(cpu_map_lock);
+
+
+/* Niagara optimized cpuinfo tree traversal. */
+static const int niagara_iterate_method[] = {
+ [CPUINFO_LVL_ROOT] = ROVER_NO_OP,
+
+ /* Strands (or virtual CPUs) within a core may not run concurrently
+ * on the Niagara, as instruction pipeline(s) are shared. Distribute
+ * work to strands in different cores first for better concurrency.
+ * Go to next NUMA node when all cores are used.
+ */
+ [CPUINFO_LVL_NODE] = ROVER_INC_ON_VISIT|ROVER_INC_PARENT_ON_LOOP,
+
+ /* Strands are grouped together by proc_id in cpuinfo_sparc, i.e.
+ * a proc_id represents an instruction pipeline. Distribute work to
+ * strands in different proc_id groups if the core has multiple
+ * instruction pipelines (e.g. the Niagara 2/2+ has two).
+ */
+ [CPUINFO_LVL_CORE] = ROVER_INC_ON_VISIT,
+
+ /* Pick the next strand in the proc_id group. */
+ [CPUINFO_LVL_PROC] = ROVER_INC_ON_VISIT,
+};
+
+/* Generic cpuinfo tree traversal. Distribute work round robin across NUMA
+ * nodes.
+ */
+static const int generic_iterate_method[] = {
+ [CPUINFO_LVL_ROOT] = ROVER_INC_ON_VISIT,
+ [CPUINFO_LVL_NODE] = ROVER_NO_OP,
+ [CPUINFO_LVL_CORE] = ROVER_INC_PARENT_ON_LOOP,
+ [CPUINFO_LVL_PROC] = ROVER_INC_ON_VISIT|ROVER_INC_PARENT_ON_LOOP,
+};
+
+
+static int cpuinfo_id(int cpu, int level)
+{
+ int id;
+
+ switch (level) {
+ case CPUINFO_LVL_ROOT:
+ id = 0;
+ break;
+ case CPUINFO_LVL_NODE:
+ id = cpu_to_node(cpu);
+ break;
+ case CPUINFO_LVL_CORE:
+ id = cpu_data(cpu).core_id;
+ break;
+ case CPUINFO_LVL_PROC:
+ id = cpu_data(cpu).proc_id;
+ break;
+ default:
+ id = -EINVAL;
+ }
+ return id;
+}
+
+/*
+ * Enumerate the CPU information in __cpu_data to determine the start index,
+ * end index, and number of nodes for each level in the cpuinfo tree. The
+ * total number of cpuinfo nodes required to build the tree is returned.
+ */
+static int enumerate_cpuinfo_nodes(struct cpuinfo_level *tree_level)
+{
+ int prev_id[CPUINFO_LVL_MAX];
+ int i, n, num_nodes;
+
+ for (i = CPUINFO_LVL_ROOT; i < CPUINFO_LVL_MAX; i++) {
+ struct cpuinfo_level *lv = &tree_level[i];
+
+ prev_id[i] = -1;
+ lv->start_index = lv->end_index = lv->num_nodes = 0;
+ }
+
+ num_nodes = 1; /* Include the root node */
+
+ for (i = 0; i < num_possible_cpus(); i++) {
+ if (!cpu_online(i))
+ continue;
+
+ n = cpuinfo_id(i, CPUINFO_LVL_NODE);
+ if (n > prev_id[CPUINFO_LVL_NODE]) {
+ tree_level[CPUINFO_LVL_NODE].num_nodes++;
+ prev_id[CPUINFO_LVL_NODE] = n;
+ num_nodes++;
+ }
+ n = cpuinfo_id(i, CPUINFO_LVL_CORE);
+ if (n > prev_id[CPUINFO_LVL_CORE]) {
+ tree_level[CPUINFO_LVL_CORE].num_nodes++;
+ prev_id[CPUINFO_LVL_CORE] = n;
+ num_nodes++;
+ }
+ n = cpuinfo_id(i, CPUINFO_LVL_PROC);
+ if (n > prev_id[CPUINFO_LVL_PROC]) {
+ tree_level[CPUINFO_LVL_PROC].num_nodes++;
+ prev_id[CPUINFO_LVL_PROC] = n;
+ num_nodes++;
+ }
+ }
+
+ tree_level[CPUINFO_LVL_ROOT].num_nodes = 1;
+
+ n = tree_level[CPUINFO_LVL_NODE].num_nodes;
+ tree_level[CPUINFO_LVL_NODE].start_index = 1;
+ tree_level[CPUINFO_LVL_NODE].end_index = n;
+
+ n++;
+ tree_level[CPUINFO_LVL_CORE].start_index = n;
+ n += tree_level[CPUINFO_LVL_CORE].num_nodes;
+ tree_level[CPUINFO_LVL_CORE].end_index = n - 1;
+
+ tree_level[CPUINFO_LVL_PROC].start_index = n;
+ n += tree_level[CPUINFO_LVL_PROC].num_nodes;
+ tree_level[CPUINFO_LVL_PROC].end_index = n - 1;
+
+ return num_nodes;
+}
+
+/* Build a tree representation of the CPU hierarchy using the per CPU
+ * information in __cpu_data. Entries in __cpu_data[0..NR_CPUS] are
+ * assumed to be sorted in ascending order based on node, core_id, and
+ * proc_id (in order of significance).
+ */
+static struct cpuinfo_tree *build_cpuinfo_tree(void)
+{
+ struct cpuinfo_tree *new_tree;
+ struct cpuinfo_node *node;
+ struct cpuinfo_level tmp_level[CPUINFO_LVL_MAX];
+ int num_cpus[CPUINFO_LVL_MAX];
+ int level_rover[CPUINFO_LVL_MAX];
+ int prev_id[CPUINFO_LVL_MAX];
+ int n, id, cpu, prev_cpu, last_cpu, level;
+
+ n = enumerate_cpuinfo_nodes(tmp_level);
+
+ new_tree = kzalloc(sizeof(struct cpuinfo_tree) +
+ (sizeof(struct cpuinfo_node) * n), GFP_ATOMIC);
+ if (!new_tree)
+ return NULL;
+
+ new_tree->total_nodes = n;
+ memcpy(&new_tree->level, tmp_level, sizeof(tmp_level));
+
+ prev_cpu = cpu = cpumask_first(cpu_online_mask);
+
+ /* Initialize all levels in the tree with the first CPU */
+ for (level = CPUINFO_LVL_PROC; level >= CPUINFO_LVL_ROOT; level--) {
+ n = new_tree->level[level].start_index;
+
+ level_rover[level] = n;
+ node = &new_tree->nodes[n];
+
+ id = cpuinfo_id(cpu, level);
+ if (unlikely(id < 0)) {
+ kfree(new_tree);
+ return NULL;
+ }
+ node->id = id;
+ node->level = level;
+ node->num_cpus = 1;
+
+ node->parent_index = (level > CPUINFO_LVL_ROOT)
+ ? new_tree->level[level - 1].start_index : -1;
+
+ node->child_start = node->child_end = node->rover =
+ (level == CPUINFO_LVL_PROC)
+ ? cpu : new_tree->level[level + 1].start_index;
+
+ prev_id[level] = node->id;
+ num_cpus[level] = 1;
+ }
+
+ for (last_cpu = (num_possible_cpus() - 1); last_cpu >= 0; last_cpu--) {
+ if (cpu_online(last_cpu))
+ break;
+ }
+
+ while (++cpu <= last_cpu) {
+ if (!cpu_online(cpu))
+ continue;
+
+ for (level = CPUINFO_LVL_PROC; level >= CPUINFO_LVL_ROOT;
+ level--) {
+ id = cpuinfo_id(cpu, level);
+ if (unlikely(id < 0)) {
+ kfree(new_tree);
+ return NULL;
+ }
+
+ if ((id != prev_id[level]) || (cpu == last_cpu)) {
+ prev_id[level] = id;
+ node = &new_tree->nodes[level_rover[level]];
+ node->num_cpus = num_cpus[level];
+ num_cpus[level] = 1;
+
+ if (cpu == last_cpu)
+ node->num_cpus++;
+
+ /* Connect tree node to parent */
+ if (level == CPUINFO_LVL_ROOT)
+ node->parent_index = -1;
+ else
+ node->parent_index =
+ level_rover[level - 1];
+
+ if (level == CPUINFO_LVL_PROC) {
+ node->child_end =
+ (cpu == last_cpu) ? cpu : prev_cpu;
+ } else {
+ node->child_end =
+ level_rover[level + 1] - 1;
+ }
+
+ /* Initialize the next node in the same level */
+ n = ++level_rover[level];
+ if (n <= new_tree->level[level].end_index) {
+ node = &new_tree->nodes[n];
+ node->id = id;
+ node->level = level;
+
+ /* Connect node to child */
+ node->child_start = node->child_end =
+ node->rover =
+ (level == CPUINFO_LVL_PROC)
+ ? cpu : level_rover[level + 1];
+ }
+ } else
+ num_cpus[level]++;
+ }
+ prev_cpu = cpu;
+ }
+
+ return new_tree;
+}
+
+static void increment_rover(struct cpuinfo_tree *t, int node_index,
+ int root_index, const int *rover_inc_table)
+{
+ struct cpuinfo_node *node = &t->nodes[node_index];
+ int top_level, level;
+
+ top_level = t->nodes[root_index].level;
+ for (level = node->level; level >= top_level; level--) {
+ node->rover++;
+ if (node->rover <= node->child_end)
+ return;
+
+ node->rover = node->child_start;
+ /* If parent's rover does not need to be adjusted, stop here. */
+ if ((level == top_level) ||
+ !(rover_inc_table[level] & ROVER_INC_PARENT_ON_LOOP))
+ return;
+
+ node = &t->nodes[node->parent_index];
+ }
+}
+
+static int iterate_cpu(struct cpuinfo_tree *t, unsigned int root_index)
+{
+ const int *rover_inc_table;
+ int level, new_index, index = root_index;
+
+ switch (sun4v_chip_type) {
+ case SUN4V_CHIP_NIAGARA1:
+ case SUN4V_CHIP_NIAGARA2:
+ case SUN4V_CHIP_NIAGARA3:
+ case SUN4V_CHIP_NIAGARA4:
+ case SUN4V_CHIP_NIAGARA5:
+ case SUN4V_CHIP_SPARC64X:
+ rover_inc_table = niagara_iterate_method;
+ break;
+ default:
+ rover_inc_table = generic_iterate_method;
+ }
+
+ for (level = t->nodes[root_index].level; level < CPUINFO_LVL_MAX;
+ level++) {
+ new_index = t->nodes[index].rover;
+ if (rover_inc_table[level] & ROVER_INC_ON_VISIT)
+ increment_rover(t, index, root_index, rover_inc_table);
+
+ index = new_index;
+ }
+ return index;
+}
+
+static void _cpu_map_rebuild(void)
+{
+ int i;
+
+ if (cpuinfo_tree) {
+ kfree(cpuinfo_tree);
+ cpuinfo_tree = NULL;
+ }
+
+ cpuinfo_tree = build_cpuinfo_tree();
+ if (!cpuinfo_tree)
+ return;
+
+ /* Build CPU distribution map that spans all online CPUs. No need
+ * to check if the CPU is online, as that is done when the cpuinfo
+ * tree is being built.
+ */
+ for (i = 0; i < cpuinfo_tree->nodes[0].num_cpus; i++)
+ cpu_distribution_map[i] = iterate_cpu(cpuinfo_tree, 0);
+}
+
+/* Fallback if the cpuinfo tree could not be built. CPU mapping is linear
+ * round robin.
+ */
+static int simple_map_to_cpu(unsigned int index)
+{
+ int i, end, cpu_rover;
+
+ cpu_rover = 0;
+ end = index % num_online_cpus();
+ for (i = 0; i < num_possible_cpus(); i++) {
+ if (cpu_online(cpu_rover)) {
+ if (cpu_rover >= end)
+ return cpu_rover;
+
+ cpu_rover++;
+ }
+ }
+
+ /* Impossible, since num_online_cpus() <= num_possible_cpus() */
+ return cpumask_first(cpu_online_mask);
+}
+
+static int _map_to_cpu(unsigned int index)
+{
+ struct cpuinfo_node *root_node;
+
+ if (unlikely(!cpuinfo_tree)) {
+ _cpu_map_rebuild();
+ if (!cpuinfo_tree)
+ return simple_map_to_cpu(index);
+ }
+
+ root_node = &cpuinfo_tree->nodes[0];
+#ifdef CONFIG_HOTPLUG_CPU
+ if (unlikely(root_node->num_cpus != num_online_cpus())) {
+ _cpu_map_rebuild();
+ if (!cpuinfo_tree)
+ return simple_map_to_cpu(index);
+ }
+#endif
+ return cpu_distribution_map[index % root_node->num_cpus];
+}
+
+int map_to_cpu(unsigned int index)
+{
+ int mapped_cpu;
+ unsigned long flag;
+
+ spin_lock_irqsave(&cpu_map_lock, flag);
+ mapped_cpu = _map_to_cpu(index);
+
+#ifdef CONFIG_HOTPLUG_CPU
+ while (unlikely(!cpu_online(mapped_cpu)))
+ mapped_cpu = _map_to_cpu(index);
+#endif
+ spin_unlock_irqrestore(&cpu_map_lock, flag);
+ return mapped_cpu;
+}
+EXPORT_SYMBOL(map_to_cpu);
+
+void cpu_map_rebuild(void)
+{
+ unsigned long flag;
+
+ spin_lock_irqsave(&cpu_map_lock, flag);
+ _cpu_map_rebuild();
+ spin_unlock_irqrestore(&cpu_map_lock, flag);
+}
diff --git a/arch/sparc/kernel/cpumap.h b/arch/sparc/kernel/cpumap.h
new file mode 100644
index 00000000000..9dac398c434
--- /dev/null
+++ b/arch/sparc/kernel/cpumap.h
@@ -0,0 +1,16 @@
+#ifndef _CPUMAP_H
+#define _CPUMAP_H
+
+#ifdef CONFIG_SMP
+void cpu_map_rebuild(void);
+int map_to_cpu(unsigned int index);
+#define cpu_map_init() cpu_map_rebuild()
+#else
+#define cpu_map_init() do {} while (0)
+static inline int map_to_cpu(unsigned int index)
+{
+ return raw_smp_processor_id();
+}
+#endif
+
+#endif
diff --git a/arch/sparc/kernel/devices.c b/arch/sparc/kernel/devices.c
index b240b8863fd..8d5d09f09ca 100644
--- a/arch/sparc/kernel/devices.c
+++ b/arch/sparc/kernel/devices.c
@@ -17,12 +17,11 @@
#include <asm/oplib.h>
#include <asm/prom.h>
#include <asm/smp.h>
-#include <asm/system.h>
#include <asm/cpudata.h>
+#include <asm/cpu_type.h>
+#include <asm/setup.h>
-extern void cpu_probe(void);
-extern void clock_stop_probe(void); /* tadpole.c */
-extern void sun4c_probe_memerr_reg(void);
+#include "kernel.h"
static char *cpu_mid_prop(void)
{
@@ -31,9 +30,9 @@ static char *cpu_mid_prop(void)
return "mid";
}
-static int check_cpu_node(int nd, int *cur_inst,
- int (*compare)(int, int, void *), void *compare_arg,
- int *prom_node, int *mid)
+static int check_cpu_node(phandle nd, int *cur_inst,
+ int (*compare)(phandle, int, void *), void *compare_arg,
+ phandle *prom_node, int *mid)
{
if (!compare(nd, *cur_inst, compare_arg)) {
if (prom_node)
@@ -51,15 +50,15 @@ static int check_cpu_node(int nd, int *cur_inst,
return -ENODEV;
}
-static int __cpu_find_by(int (*compare)(int, int, void *), void *compare_arg,
- int *prom_node, int *mid)
+static int __cpu_find_by(int (*compare)(phandle, int, void *),
+ void *compare_arg, phandle *prom_node, int *mid)
{
struct device_node *dp;
int cur_inst;
cur_inst = 0;
for_each_node_by_type(dp, "cpu") {
- int err = check_cpu_node(dp->node, &cur_inst,
+ int err = check_cpu_node(dp->phandle, &cur_inst,
compare, compare_arg,
prom_node, mid);
if (!err) {
@@ -71,7 +70,7 @@ static int __cpu_find_by(int (*compare)(int, int, void *), void *compare_arg,
return -ENODEV;
}
-static int cpu_instance_compare(int nd, int instance, void *_arg)
+static int cpu_instance_compare(phandle nd, int instance, void *_arg)
{
int desired_instance = (int) _arg;
@@ -80,13 +79,13 @@ static int cpu_instance_compare(int nd, int instance, void *_arg)
return -ENODEV;
}
-int cpu_find_by_instance(int instance, int *prom_node, int *mid)
+int cpu_find_by_instance(int instance, phandle *prom_node, int *mid)
{
return __cpu_find_by(cpu_instance_compare, (void *)instance,
prom_node, mid);
}
-static int cpu_mid_compare(int nd, int instance, void *_arg)
+static int cpu_mid_compare(phandle nd, int instance, void *_arg)
{
int desired_mid = (int) _arg;
int this_mid;
@@ -98,7 +97,7 @@ static int cpu_mid_compare(int nd, int instance, void *_arg)
return -ENODEV;
}
-int cpu_find_by_mid(int mid, int *prom_node)
+int cpu_find_by_mid(int mid, phandle *prom_node)
{
return __cpu_find_by(cpu_mid_compare, (void *)mid,
prom_node, NULL);
@@ -108,18 +107,19 @@ int cpu_find_by_mid(int mid, int *prom_node)
* address (0-3). This gives us the true hardware mid, which might have
* some other bits set. On 4d hardware and software mids are the same.
*/
-int cpu_get_hwmid(int prom_node)
+int cpu_get_hwmid(phandle prom_node)
{
return prom_getintdefault(prom_node, cpu_mid_prop(), -ENODEV);
}
void __init device_scan(void)
{
- prom_printf("Booting Linux...\n");
+ printk(KERN_NOTICE "Booting Linux...\n");
#ifndef CONFIG_SMP
{
- int err, cpu_node;
+ phandle cpu_node;
+ int err;
err = cpu_find_by_instance(0, &cpu_node, NULL);
if (err) {
/* Probably a sun4e, Sun is trying to trick us ;-) */
@@ -132,19 +132,6 @@ void __init device_scan(void)
}
#endif /* !CONFIG_SMP */
- cpu_probe();
-#ifdef CONFIG_SUN_AUXIO
- {
- extern void auxio_probe(void);
- extern void auxio_power_probe(void);
- auxio_probe();
- auxio_power_probe();
- }
-#endif
- clock_stop_probe();
-
- if (ARCH_SUN4C_SUN4)
- sun4c_probe_memerr_reg();
-
- return;
+ auxio_probe();
+ auxio_power_probe();
}
diff --git a/arch/sparc/kernel/dma.c b/arch/sparc/kernel/dma.c
new file mode 100644
index 00000000000..b667aa6f28f
--- /dev/null
+++ b/arch/sparc/kernel/dma.c
@@ -0,0 +1,12 @@
+#include <linux/kernel.h>
+#include <linux/dma-mapping.h>
+#include <linux/dma-debug.h>
+
+#define PREALLOC_DMA_DEBUG_ENTRIES (1 << 15)
+
+static int __init dma_init(void)
+{
+ dma_debug_init(PREALLOC_DMA_DEBUG_ENTRIES);
+ return 0;
+}
+fs_initcall(dma_init);
diff --git a/arch/sparc/kernel/ds.c b/arch/sparc/kernel/ds.c
new file mode 100644
index 00000000000..dff60abbea0
--- /dev/null
+++ b/arch/sparc/kernel/ds.c
@@ -0,0 +1,1273 @@
+/* ds.c: Domain Services driver for Logical Domains
+ *
+ * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+#include <linux/delay.h>
+#include <linux/mutex.h>
+#include <linux/kthread.h>
+#include <linux/reboot.h>
+#include <linux/cpu.h>
+
+#include <asm/hypervisor.h>
+#include <asm/ldc.h>
+#include <asm/vio.h>
+#include <asm/mdesc.h>
+#include <asm/head.h>
+#include <asm/irq.h>
+
+#include "kernel.h"
+
+#define DRV_MODULE_NAME "ds"
+#define PFX DRV_MODULE_NAME ": "
+#define DRV_MODULE_VERSION "1.0"
+#define DRV_MODULE_RELDATE "Jul 11, 2007"
+
+static char version[] =
+ DRV_MODULE_NAME ".c:v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
+MODULE_AUTHOR("David S. Miller (davem@davemloft.net)");
+MODULE_DESCRIPTION("Sun LDOM domain services driver");
+MODULE_LICENSE("GPL");
+MODULE_VERSION(DRV_MODULE_VERSION);
+
+struct ds_msg_tag {
+ __u32 type;
+#define DS_INIT_REQ 0x00
+#define DS_INIT_ACK 0x01
+#define DS_INIT_NACK 0x02
+#define DS_REG_REQ 0x03
+#define DS_REG_ACK 0x04
+#define DS_REG_NACK 0x05
+#define DS_UNREG_REQ 0x06
+#define DS_UNREG_ACK 0x07
+#define DS_UNREG_NACK 0x08
+#define DS_DATA 0x09
+#define DS_NACK 0x0a
+
+ __u32 len;
+};
+
+/* Result codes */
+#define DS_OK 0x00
+#define DS_REG_VER_NACK 0x01
+#define DS_REG_DUP 0x02
+#define DS_INV_HDL 0x03
+#define DS_TYPE_UNKNOWN 0x04
+
+struct ds_version {
+ __u16 major;
+ __u16 minor;
+};
+
+struct ds_ver_req {
+ struct ds_msg_tag tag;
+ struct ds_version ver;
+};
+
+struct ds_ver_ack {
+ struct ds_msg_tag tag;
+ __u16 minor;
+};
+
+struct ds_ver_nack {
+ struct ds_msg_tag tag;
+ __u16 major;
+};
+
+struct ds_reg_req {
+ struct ds_msg_tag tag;
+ __u64 handle;
+ __u16 major;
+ __u16 minor;
+ char svc_id[0];
+};
+
+struct ds_reg_ack {
+ struct ds_msg_tag tag;
+ __u64 handle;
+ __u16 minor;
+};
+
+struct ds_reg_nack {
+ struct ds_msg_tag tag;
+ __u64 handle;
+ __u16 major;
+};
+
+struct ds_unreg_req {
+ struct ds_msg_tag tag;
+ __u64 handle;
+};
+
+struct ds_unreg_ack {
+ struct ds_msg_tag tag;
+ __u64 handle;
+};
+
+struct ds_unreg_nack {
+ struct ds_msg_tag tag;
+ __u64 handle;
+};
+
+struct ds_data {
+ struct ds_msg_tag tag;
+ __u64 handle;
+};
+
+struct ds_data_nack {
+ struct ds_msg_tag tag;
+ __u64 handle;
+ __u64 result;
+};
+
+struct ds_info;
+struct ds_cap_state {
+ __u64 handle;
+
+ void (*data)(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len);
+
+ const char *service_id;
+
+ u8 state;
+#define CAP_STATE_UNKNOWN 0x00
+#define CAP_STATE_REG_SENT 0x01
+#define CAP_STATE_REGISTERED 0x02
+};
+
+static void md_update_data(struct ds_info *dp, struct ds_cap_state *cp,
+ void *buf, int len);
+static void domain_shutdown_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len);
+static void domain_panic_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len);
+#ifdef CONFIG_HOTPLUG_CPU
+static void dr_cpu_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len);
+#endif
+static void ds_pri_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len);
+static void ds_var_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len);
+
+static struct ds_cap_state ds_states_template[] = {
+ {
+ .service_id = "md-update",
+ .data = md_update_data,
+ },
+ {
+ .service_id = "domain-shutdown",
+ .data = domain_shutdown_data,
+ },
+ {
+ .service_id = "domain-panic",
+ .data = domain_panic_data,
+ },
+#ifdef CONFIG_HOTPLUG_CPU
+ {
+ .service_id = "dr-cpu",
+ .data = dr_cpu_data,
+ },
+#endif
+ {
+ .service_id = "pri",
+ .data = ds_pri_data,
+ },
+ {
+ .service_id = "var-config",
+ .data = ds_var_data,
+ },
+ {
+ .service_id = "var-config-backup",
+ .data = ds_var_data,
+ },
+};
+
+static DEFINE_SPINLOCK(ds_lock);
+
+struct ds_info {
+ struct ldc_channel *lp;
+ u8 hs_state;
+#define DS_HS_START 0x01
+#define DS_HS_DONE 0x02
+
+ u64 id;
+
+ void *rcv_buf;
+ int rcv_buf_len;
+
+ struct ds_cap_state *ds_states;
+ int num_ds_states;
+
+ struct ds_info *next;
+};
+
+static struct ds_info *ds_info_list;
+
+static struct ds_cap_state *find_cap(struct ds_info *dp, u64 handle)
+{
+ unsigned int index = handle >> 32;
+
+ if (index >= dp->num_ds_states)
+ return NULL;
+ return &dp->ds_states[index];
+}
+
+static struct ds_cap_state *find_cap_by_string(struct ds_info *dp,
+ const char *name)
+{
+ int i;
+
+ for (i = 0; i < dp->num_ds_states; i++) {
+ if (strcmp(dp->ds_states[i].service_id, name))
+ continue;
+
+ return &dp->ds_states[i];
+ }
+ return NULL;
+}
+
+static int __ds_send(struct ldc_channel *lp, void *data, int len)
+{
+ int err, limit = 1000;
+
+ err = -EINVAL;
+ while (limit-- > 0) {
+ err = ldc_write(lp, data, len);
+ if (!err || (err != -EAGAIN))
+ break;
+ udelay(1);
+ }
+
+ return err;
+}
+
+static int ds_send(struct ldc_channel *lp, void *data, int len)
+{
+ unsigned long flags;
+ int err;
+
+ spin_lock_irqsave(&ds_lock, flags);
+ err = __ds_send(lp, data, len);
+ spin_unlock_irqrestore(&ds_lock, flags);
+
+ return err;
+}
+
+struct ds_md_update_req {
+ __u64 req_num;
+};
+
+struct ds_md_update_res {
+ __u64 req_num;
+ __u32 result;
+};
+
+static void md_update_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len)
+{
+ struct ldc_channel *lp = dp->lp;
+ struct ds_data *dpkt = buf;
+ struct ds_md_update_req *rp;
+ struct {
+ struct ds_data data;
+ struct ds_md_update_res res;
+ } pkt;
+
+ rp = (struct ds_md_update_req *) (dpkt + 1);
+
+ printk(KERN_INFO "ds-%llu: Machine description update.\n", dp->id);
+
+ mdesc_update();
+
+ memset(&pkt, 0, sizeof(pkt));
+ pkt.data.tag.type = DS_DATA;
+ pkt.data.tag.len = sizeof(pkt) - sizeof(struct ds_msg_tag);
+ pkt.data.handle = cp->handle;
+ pkt.res.req_num = rp->req_num;
+ pkt.res.result = DS_OK;
+
+ ds_send(lp, &pkt, sizeof(pkt));
+}
+
+struct ds_shutdown_req {
+ __u64 req_num;
+ __u32 ms_delay;
+};
+
+struct ds_shutdown_res {
+ __u64 req_num;
+ __u32 result;
+ char reason[1];
+};
+
+static void domain_shutdown_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len)
+{
+ struct ldc_channel *lp = dp->lp;
+ struct ds_data *dpkt = buf;
+ struct ds_shutdown_req *rp;
+ struct {
+ struct ds_data data;
+ struct ds_shutdown_res res;
+ } pkt;
+
+ rp = (struct ds_shutdown_req *) (dpkt + 1);
+
+ printk(KERN_ALERT "ds-%llu: Shutdown request from "
+ "LDOM manager received.\n", dp->id);
+
+ memset(&pkt, 0, sizeof(pkt));
+ pkt.data.tag.type = DS_DATA;
+ pkt.data.tag.len = sizeof(pkt) - sizeof(struct ds_msg_tag);
+ pkt.data.handle = cp->handle;
+ pkt.res.req_num = rp->req_num;
+ pkt.res.result = DS_OK;
+ pkt.res.reason[0] = 0;
+
+ ds_send(lp, &pkt, sizeof(pkt));
+
+ orderly_poweroff(true);
+}
+
+struct ds_panic_req {
+ __u64 req_num;
+};
+
+struct ds_panic_res {
+ __u64 req_num;
+ __u32 result;
+ char reason[1];
+};
+
+static void domain_panic_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len)
+{
+ struct ldc_channel *lp = dp->lp;
+ struct ds_data *dpkt = buf;
+ struct ds_panic_req *rp;
+ struct {
+ struct ds_data data;
+ struct ds_panic_res res;
+ } pkt;
+
+ rp = (struct ds_panic_req *) (dpkt + 1);
+
+ printk(KERN_ALERT "ds-%llu: Panic request from "
+ "LDOM manager received.\n", dp->id);
+
+ memset(&pkt, 0, sizeof(pkt));
+ pkt.data.tag.type = DS_DATA;
+ pkt.data.tag.len = sizeof(pkt) - sizeof(struct ds_msg_tag);
+ pkt.data.handle = cp->handle;
+ pkt.res.req_num = rp->req_num;
+ pkt.res.result = DS_OK;
+ pkt.res.reason[0] = 0;
+
+ ds_send(lp, &pkt, sizeof(pkt));
+
+ panic("PANIC requested by LDOM manager.");
+}
+
+#ifdef CONFIG_HOTPLUG_CPU
+struct dr_cpu_tag {
+ __u64 req_num;
+ __u32 type;
+#define DR_CPU_CONFIGURE 0x43
+#define DR_CPU_UNCONFIGURE 0x55
+#define DR_CPU_FORCE_UNCONFIGURE 0x46
+#define DR_CPU_STATUS 0x53
+
+/* Responses */
+#define DR_CPU_OK 0x6f
+#define DR_CPU_ERROR 0x65
+
+ __u32 num_records;
+};
+
+struct dr_cpu_resp_entry {
+ __u32 cpu;
+ __u32 result;
+#define DR_CPU_RES_OK 0x00
+#define DR_CPU_RES_FAILURE 0x01
+#define DR_CPU_RES_BLOCKED 0x02
+#define DR_CPU_RES_CPU_NOT_RESPONDING 0x03
+#define DR_CPU_RES_NOT_IN_MD 0x04
+
+ __u32 stat;
+#define DR_CPU_STAT_NOT_PRESENT 0x00
+#define DR_CPU_STAT_UNCONFIGURED 0x01
+#define DR_CPU_STAT_CONFIGURED 0x02
+
+ __u32 str_off;
+};
+
+static void __dr_cpu_send_error(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ struct ds_data *data)
+{
+ struct dr_cpu_tag *tag = (struct dr_cpu_tag *) (data + 1);
+ struct {
+ struct ds_data data;
+ struct dr_cpu_tag tag;
+ } pkt;
+ int msg_len;
+
+ memset(&pkt, 0, sizeof(pkt));
+ pkt.data.tag.type = DS_DATA;
+ pkt.data.handle = cp->handle;
+ pkt.tag.req_num = tag->req_num;
+ pkt.tag.type = DR_CPU_ERROR;
+ pkt.tag.num_records = 0;
+
+ msg_len = (sizeof(struct ds_data) +
+ sizeof(struct dr_cpu_tag));
+
+ pkt.data.tag.len = msg_len - sizeof(struct ds_msg_tag);
+
+ __ds_send(dp->lp, &pkt, msg_len);
+}
+
+static void dr_cpu_send_error(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ struct ds_data *data)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&ds_lock, flags);
+ __dr_cpu_send_error(dp, cp, data);
+ spin_unlock_irqrestore(&ds_lock, flags);
+}
+
+#define CPU_SENTINEL 0xffffffff
+
+static void purge_dups(u32 *list, u32 num_ents)
+{
+ unsigned int i;
+
+ for (i = 0; i < num_ents; i++) {
+ u32 cpu = list[i];
+ unsigned int j;
+
+ if (cpu == CPU_SENTINEL)
+ continue;
+
+ for (j = i + 1; j < num_ents; j++) {
+ if (list[j] == cpu)
+ list[j] = CPU_SENTINEL;
+ }
+ }
+}
+
+static int dr_cpu_size_response(int ncpus)
+{
+ return (sizeof(struct ds_data) +
+ sizeof(struct dr_cpu_tag) +
+ (sizeof(struct dr_cpu_resp_entry) * ncpus));
+}
+
+static void dr_cpu_init_response(struct ds_data *resp, u64 req_num,
+ u64 handle, int resp_len, int ncpus,
+ cpumask_t *mask, u32 default_stat)
+{
+ struct dr_cpu_resp_entry *ent;
+ struct dr_cpu_tag *tag;
+ int i, cpu;
+
+ tag = (struct dr_cpu_tag *) (resp + 1);
+ ent = (struct dr_cpu_resp_entry *) (tag + 1);
+
+ resp->tag.type = DS_DATA;
+ resp->tag.len = resp_len - sizeof(struct ds_msg_tag);
+ resp->handle = handle;
+ tag->req_num = req_num;
+ tag->type = DR_CPU_OK;
+ tag->num_records = ncpus;
+
+ i = 0;
+ for_each_cpu(cpu, mask) {
+ ent[i].cpu = cpu;
+ ent[i].result = DR_CPU_RES_OK;
+ ent[i].stat = default_stat;
+ i++;
+ }
+ BUG_ON(i != ncpus);
+}
+
+static void dr_cpu_mark(struct ds_data *resp, int cpu, int ncpus,
+ u32 res, u32 stat)
+{
+ struct dr_cpu_resp_entry *ent;
+ struct dr_cpu_tag *tag;
+ int i;
+
+ tag = (struct dr_cpu_tag *) (resp + 1);
+ ent = (struct dr_cpu_resp_entry *) (tag + 1);
+
+ for (i = 0; i < ncpus; i++) {
+ if (ent[i].cpu != cpu)
+ continue;
+ ent[i].result = res;
+ ent[i].stat = stat;
+ break;
+ }
+}
+
+static int dr_cpu_configure(struct ds_info *dp, struct ds_cap_state *cp,
+ u64 req_num, cpumask_t *mask)
+{
+ struct ds_data *resp;
+ int resp_len, ncpus, cpu;
+ unsigned long flags;
+
+ ncpus = cpumask_weight(mask);
+ resp_len = dr_cpu_size_response(ncpus);
+ resp = kzalloc(resp_len, GFP_KERNEL);
+ if (!resp)
+ return -ENOMEM;
+
+ dr_cpu_init_response(resp, req_num, cp->handle,
+ resp_len, ncpus, mask,
+ DR_CPU_STAT_CONFIGURED);
+
+ mdesc_populate_present_mask(mask);
+ mdesc_fill_in_cpu_data(mask);
+
+ for_each_cpu(cpu, mask) {
+ int err;
+
+ printk(KERN_INFO "ds-%llu: Starting cpu %d...\n",
+ dp->id, cpu);
+ err = cpu_up(cpu);
+ if (err) {
+ __u32 res = DR_CPU_RES_FAILURE;
+ __u32 stat = DR_CPU_STAT_UNCONFIGURED;
+
+ if (!cpu_present(cpu)) {
+ /* CPU not present in MD */
+ res = DR_CPU_RES_NOT_IN_MD;
+ stat = DR_CPU_STAT_NOT_PRESENT;
+ } else if (err == -ENODEV) {
+ /* CPU did not call in successfully */
+ res = DR_CPU_RES_CPU_NOT_RESPONDING;
+ }
+
+ printk(KERN_INFO "ds-%llu: CPU startup failed err=%d\n",
+ dp->id, err);
+ dr_cpu_mark(resp, cpu, ncpus, res, stat);
+ }
+ }
+
+ spin_lock_irqsave(&ds_lock, flags);
+ __ds_send(dp->lp, resp, resp_len);
+ spin_unlock_irqrestore(&ds_lock, flags);
+
+ kfree(resp);
+
+ /* Redistribute IRQs, taking into account the new cpus. */
+ fixup_irqs();
+
+ return 0;
+}
+
+static int dr_cpu_unconfigure(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ u64 req_num,
+ cpumask_t *mask)
+{
+ struct ds_data *resp;
+ int resp_len, ncpus, cpu;
+ unsigned long flags;
+
+ ncpus = cpumask_weight(mask);
+ resp_len = dr_cpu_size_response(ncpus);
+ resp = kzalloc(resp_len, GFP_KERNEL);
+ if (!resp)
+ return -ENOMEM;
+
+ dr_cpu_init_response(resp, req_num, cp->handle,
+ resp_len, ncpus, mask,
+ DR_CPU_STAT_UNCONFIGURED);
+
+ for_each_cpu(cpu, mask) {
+ int err;
+
+ printk(KERN_INFO "ds-%llu: Shutting down cpu %d...\n",
+ dp->id, cpu);
+ err = cpu_down(cpu);
+ if (err)
+ dr_cpu_mark(resp, cpu, ncpus,
+ DR_CPU_RES_FAILURE,
+ DR_CPU_STAT_CONFIGURED);
+ }
+
+ spin_lock_irqsave(&ds_lock, flags);
+ __ds_send(dp->lp, resp, resp_len);
+ spin_unlock_irqrestore(&ds_lock, flags);
+
+ kfree(resp);
+
+ return 0;
+}
+
+static void dr_cpu_data(struct ds_info *dp, struct ds_cap_state *cp, void *buf,
+ int len)
+{
+ struct ds_data *data = buf;
+ struct dr_cpu_tag *tag = (struct dr_cpu_tag *) (data + 1);
+ u32 *cpu_list = (u32 *) (tag + 1);
+ u64 req_num = tag->req_num;
+ cpumask_t mask;
+ unsigned int i;
+ int err;
+
+ switch (tag->type) {
+ case DR_CPU_CONFIGURE:
+ case DR_CPU_UNCONFIGURE:
+ case DR_CPU_FORCE_UNCONFIGURE:
+ break;
+
+ default:
+ dr_cpu_send_error(dp, cp, data);
+ return;
+ }
+
+ purge_dups(cpu_list, tag->num_records);
+
+ cpumask_clear(&mask);
+ for (i = 0; i < tag->num_records; i++) {
+ if (cpu_list[i] == CPU_SENTINEL)
+ continue;
+
+ if (cpu_list[i] < nr_cpu_ids)
+ cpumask_set_cpu(cpu_list[i], &mask);
+ }
+
+ if (tag->type == DR_CPU_CONFIGURE)
+ err = dr_cpu_configure(dp, cp, req_num, &mask);
+ else
+ err = dr_cpu_unconfigure(dp, cp, req_num, &mask);
+
+ if (err)
+ dr_cpu_send_error(dp, cp, data);
+}
+#endif /* CONFIG_HOTPLUG_CPU */
+
+struct ds_pri_msg {
+ __u64 req_num;
+ __u64 type;
+#define DS_PRI_REQUEST 0x00
+#define DS_PRI_DATA 0x01
+#define DS_PRI_UPDATE 0x02
+};
+
+static void ds_pri_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len)
+{
+ struct ds_data *dpkt = buf;
+ struct ds_pri_msg *rp;
+
+ rp = (struct ds_pri_msg *) (dpkt + 1);
+
+ printk(KERN_INFO "ds-%llu: PRI REQ [%llx:%llx], len=%d\n",
+ dp->id, rp->req_num, rp->type, len);
+}
+
+struct ds_var_hdr {
+ __u32 type;
+#define DS_VAR_SET_REQ 0x00
+#define DS_VAR_DELETE_REQ 0x01
+#define DS_VAR_SET_RESP 0x02
+#define DS_VAR_DELETE_RESP 0x03
+};
+
+struct ds_var_set_msg {
+ struct ds_var_hdr hdr;
+ char name_and_value[0];
+};
+
+struct ds_var_delete_msg {
+ struct ds_var_hdr hdr;
+ char name[0];
+};
+
+struct ds_var_resp {
+ struct ds_var_hdr hdr;
+ __u32 result;
+#define DS_VAR_SUCCESS 0x00
+#define DS_VAR_NO_SPACE 0x01
+#define DS_VAR_INVALID_VAR 0x02
+#define DS_VAR_INVALID_VAL 0x03
+#define DS_VAR_NOT_PRESENT 0x04
+};
+
+static DEFINE_MUTEX(ds_var_mutex);
+static int ds_var_doorbell;
+static int ds_var_response;
+
+static void ds_var_data(struct ds_info *dp,
+ struct ds_cap_state *cp,
+ void *buf, int len)
+{
+ struct ds_data *dpkt = buf;
+ struct ds_var_resp *rp;
+
+ rp = (struct ds_var_resp *) (dpkt + 1);
+
+ if (rp->hdr.type != DS_VAR_SET_RESP &&
+ rp->hdr.type != DS_VAR_DELETE_RESP)
+ return;
+
+ ds_var_response = rp->result;
+ wmb();
+ ds_var_doorbell = 1;
+}
+
+void ldom_set_var(const char *var, const char *value)
+{
+ struct ds_cap_state *cp;
+ struct ds_info *dp;
+ unsigned long flags;
+
+ spin_lock_irqsave(&ds_lock, flags);
+ cp = NULL;
+ for (dp = ds_info_list; dp; dp = dp->next) {
+ struct ds_cap_state *tmp;
+
+ tmp = find_cap_by_string(dp, "var-config");
+ if (tmp && tmp->state == CAP_STATE_REGISTERED) {
+ cp = tmp;
+ break;
+ }
+ }
+ if (!cp) {
+ for (dp = ds_info_list; dp; dp = dp->next) {
+ struct ds_cap_state *tmp;
+
+ tmp = find_cap_by_string(dp, "var-config-backup");
+ if (tmp && tmp->state == CAP_STATE_REGISTERED) {
+ cp = tmp;
+ break;
+ }
+ }
+ }
+ spin_unlock_irqrestore(&ds_lock, flags);
+
+ if (cp) {
+ union {
+ struct {
+ struct ds_data data;
+ struct ds_var_set_msg msg;
+ } header;
+ char all[512];
+ } pkt;
+ char *base, *p;
+ int msg_len, loops;
+
+ if (strlen(var) + strlen(value) + 2 >
+ sizeof(pkt) - sizeof(pkt.header)) {
+ printk(KERN_ERR PFX
+ "contents length: %zu, which more than max: %lu,"
+ "so could not set (%s) variable to (%s).\n",
+ strlen(var) + strlen(value) + 2,
+ sizeof(pkt) - sizeof(pkt.header), var, value);
+ return;
+ }
+
+ memset(&pkt, 0, sizeof(pkt));
+ pkt.header.data.tag.type = DS_DATA;
+ pkt.header.data.handle = cp->handle;
+ pkt.header.msg.hdr.type = DS_VAR_SET_REQ;
+ base = p = &pkt.header.msg.name_and_value[0];
+ strcpy(p, var);
+ p += strlen(var) + 1;
+ strcpy(p, value);
+ p += strlen(value) + 1;
+
+ msg_len = (sizeof(struct ds_data) +
+ sizeof(struct ds_var_set_msg) +
+ (p - base));
+ msg_len = (msg_len + 3) & ~3;
+ pkt.header.data.tag.len = msg_len - sizeof(struct ds_msg_tag);
+
+ mutex_lock(&ds_var_mutex);
+
+ spin_lock_irqsave(&ds_lock, flags);
+ ds_var_doorbell = 0;
+ ds_var_response = -1;
+
+ __ds_send(dp->lp, &pkt, msg_len);
+ spin_unlock_irqrestore(&ds_lock, flags);
+
+ loops = 1000;
+ while (ds_var_doorbell == 0) {
+ if (loops-- < 0)
+ break;
+ barrier();
+ udelay(100);
+ }
+
+ mutex_unlock(&ds_var_mutex);
+
+ if (ds_var_doorbell == 0 ||
+ ds_var_response != DS_VAR_SUCCESS)
+ printk(KERN_ERR "ds-%llu: var-config [%s:%s] "
+ "failed, response(%d).\n",
+ dp->id, var, value,
+ ds_var_response);
+ } else {
+ printk(KERN_ERR PFX "var-config not registered so "
+ "could not set (%s) variable to (%s).\n",
+ var, value);
+ }
+}
+
+static char full_boot_str[256] __attribute__((aligned(32)));
+static int reboot_data_supported;
+
+void ldom_reboot(const char *boot_command)
+{
+ /* Don't bother with any of this if the boot_command
+ * is empty.
+ */
+ if (boot_command && strlen(boot_command)) {
+ unsigned long len;
+
+ snprintf(full_boot_str, sizeof(full_boot_str), "boot %s",
+ boot_command);
+ len = strlen(full_boot_str);
+
+ if (reboot_data_supported) {
+ unsigned long ra = kimage_addr_to_ra(full_boot_str);
+ unsigned long hv_ret;
+
+ hv_ret = sun4v_reboot_data_set(ra, len);
+ if (hv_ret != HV_EOK)
+ pr_err("SUN4V: Unable to set reboot data "
+ "hv_ret=%lu\n", hv_ret);
+ } else {
+ ldom_set_var("reboot-command", full_boot_str);
+ }
+ }
+ sun4v_mach_sir();
+}
+
+void ldom_power_off(void)
+{
+ sun4v_mach_exit(0);
+}
+
+static void ds_conn_reset(struct ds_info *dp)
+{
+ printk(KERN_ERR "ds-%llu: ds_conn_reset() from %pf\n",
+ dp->id, __builtin_return_address(0));
+}
+
+static int register_services(struct ds_info *dp)
+{
+ struct ldc_channel *lp = dp->lp;
+ int i;
+
+ for (i = 0; i < dp->num_ds_states; i++) {
+ struct {
+ struct ds_reg_req req;
+ u8 id_buf[256];
+ } pbuf;
+ struct ds_cap_state *cp = &dp->ds_states[i];
+ int err, msg_len;
+ u64 new_count;
+
+ if (cp->state == CAP_STATE_REGISTERED)
+ continue;
+
+ new_count = sched_clock() & 0xffffffff;
+ cp->handle = ((u64) i << 32) | new_count;
+
+ msg_len = (sizeof(struct ds_reg_req) +
+ strlen(cp->service_id));
+
+ memset(&pbuf, 0, sizeof(pbuf));
+ pbuf.req.tag.type = DS_REG_REQ;
+ pbuf.req.tag.len = (msg_len - sizeof(struct ds_msg_tag));
+ pbuf.req.handle = cp->handle;
+ pbuf.req.major = 1;
+ pbuf.req.minor = 0;
+ strcpy(pbuf.req.svc_id, cp->service_id);
+
+ err = __ds_send(lp, &pbuf, msg_len);
+ if (err > 0)
+ cp->state = CAP_STATE_REG_SENT;
+ }
+ return 0;
+}
+
+static int ds_handshake(struct ds_info *dp, struct ds_msg_tag *pkt)
+{
+
+ if (dp->hs_state == DS_HS_START) {
+ if (pkt->type != DS_INIT_ACK)
+ goto conn_reset;
+
+ dp->hs_state = DS_HS_DONE;
+
+ return register_services(dp);
+ }
+
+ if (dp->hs_state != DS_HS_DONE)
+ goto conn_reset;
+
+ if (pkt->type == DS_REG_ACK) {
+ struct ds_reg_ack *ap = (struct ds_reg_ack *) pkt;
+ struct ds_cap_state *cp = find_cap(dp, ap->handle);
+
+ if (!cp) {
+ printk(KERN_ERR "ds-%llu: REG ACK for unknown "
+ "handle %llx\n", dp->id, ap->handle);
+ return 0;
+ }
+ printk(KERN_INFO "ds-%llu: Registered %s service.\n",
+ dp->id, cp->service_id);
+ cp->state = CAP_STATE_REGISTERED;
+ } else if (pkt->type == DS_REG_NACK) {
+ struct ds_reg_nack *np = (struct ds_reg_nack *) pkt;
+ struct ds_cap_state *cp = find_cap(dp, np->handle);
+
+ if (!cp) {
+ printk(KERN_ERR "ds-%llu: REG NACK for "
+ "unknown handle %llx\n",
+ dp->id, np->handle);
+ return 0;
+ }
+ cp->state = CAP_STATE_UNKNOWN;
+ }
+
+ return 0;
+
+conn_reset:
+ ds_conn_reset(dp);
+ return -ECONNRESET;
+}
+
+static void __send_ds_nack(struct ds_info *dp, u64 handle)
+{
+ struct ds_data_nack nack = {
+ .tag = {
+ .type = DS_NACK,
+ .len = (sizeof(struct ds_data_nack) -
+ sizeof(struct ds_msg_tag)),
+ },
+ .handle = handle,
+ .result = DS_INV_HDL,
+ };
+
+ __ds_send(dp->lp, &nack, sizeof(nack));
+}
+
+static LIST_HEAD(ds_work_list);
+static DECLARE_WAIT_QUEUE_HEAD(ds_wait);
+
+struct ds_queue_entry {
+ struct list_head list;
+ struct ds_info *dp;
+ int req_len;
+ int __pad;
+ u64 req[0];
+};
+
+static void process_ds_work(void)
+{
+ struct ds_queue_entry *qp, *tmp;
+ unsigned long flags;
+ LIST_HEAD(todo);
+
+ spin_lock_irqsave(&ds_lock, flags);
+ list_splice_init(&ds_work_list, &todo);
+ spin_unlock_irqrestore(&ds_lock, flags);
+
+ list_for_each_entry_safe(qp, tmp, &todo, list) {
+ struct ds_data *dpkt = (struct ds_data *) qp->req;
+ struct ds_info *dp = qp->dp;
+ struct ds_cap_state *cp = find_cap(dp, dpkt->handle);
+ int req_len = qp->req_len;
+
+ if (!cp) {
+ printk(KERN_ERR "ds-%llu: Data for unknown "
+ "handle %llu\n",
+ dp->id, dpkt->handle);
+
+ spin_lock_irqsave(&ds_lock, flags);
+ __send_ds_nack(dp, dpkt->handle);
+ spin_unlock_irqrestore(&ds_lock, flags);
+ } else {
+ cp->data(dp, cp, dpkt, req_len);
+ }
+
+ list_del(&qp->list);
+ kfree(qp);
+ }
+}
+
+static int ds_thread(void *__unused)
+{
+ DEFINE_WAIT(wait);
+
+ while (1) {
+ prepare_to_wait(&ds_wait, &wait, TASK_INTERRUPTIBLE);
+ if (list_empty(&ds_work_list))
+ schedule();
+ finish_wait(&ds_wait, &wait);
+
+ if (kthread_should_stop())
+ break;
+
+ process_ds_work();
+ }
+
+ return 0;
+}
+
+static int ds_data(struct ds_info *dp, struct ds_msg_tag *pkt, int len)
+{
+ struct ds_data *dpkt = (struct ds_data *) pkt;
+ struct ds_queue_entry *qp;
+
+ qp = kmalloc(sizeof(struct ds_queue_entry) + len, GFP_ATOMIC);
+ if (!qp) {
+ __send_ds_nack(dp, dpkt->handle);
+ } else {
+ qp->dp = dp;
+ memcpy(&qp->req, pkt, len);
+ list_add_tail(&qp->list, &ds_work_list);
+ wake_up(&ds_wait);
+ }
+ return 0;
+}
+
+static void ds_up(struct ds_info *dp)
+{
+ struct ldc_channel *lp = dp->lp;
+ struct ds_ver_req req;
+ int err;
+
+ req.tag.type = DS_INIT_REQ;
+ req.tag.len = sizeof(req) - sizeof(struct ds_msg_tag);
+ req.ver.major = 1;
+ req.ver.minor = 0;
+
+ err = __ds_send(lp, &req, sizeof(req));
+ if (err > 0)
+ dp->hs_state = DS_HS_START;
+}
+
+static void ds_reset(struct ds_info *dp)
+{
+ int i;
+
+ dp->hs_state = 0;
+
+ for (i = 0; i < dp->num_ds_states; i++) {
+ struct ds_cap_state *cp = &dp->ds_states[i];
+
+ cp->state = CAP_STATE_UNKNOWN;
+ }
+}
+
+static void ds_event(void *arg, int event)
+{
+ struct ds_info *dp = arg;
+ struct ldc_channel *lp = dp->lp;
+ unsigned long flags;
+ int err;
+
+ spin_lock_irqsave(&ds_lock, flags);
+
+ if (event == LDC_EVENT_UP) {
+ ds_up(dp);
+ spin_unlock_irqrestore(&ds_lock, flags);
+ return;
+ }
+
+ if (event == LDC_EVENT_RESET) {
+ ds_reset(dp);
+ spin_unlock_irqrestore(&ds_lock, flags);
+ return;
+ }
+
+ if (event != LDC_EVENT_DATA_READY) {
+ printk(KERN_WARNING "ds-%llu: Unexpected LDC event %d\n",
+ dp->id, event);
+ spin_unlock_irqrestore(&ds_lock, flags);
+ return;
+ }
+
+ err = 0;
+ while (1) {
+ struct ds_msg_tag *tag;
+
+ err = ldc_read(lp, dp->rcv_buf, sizeof(*tag));
+
+ if (unlikely(err < 0)) {
+ if (err == -ECONNRESET)
+ ds_conn_reset(dp);
+ break;
+ }
+ if (err == 0)
+ break;
+
+ tag = dp->rcv_buf;
+ err = ldc_read(lp, tag + 1, tag->len);
+
+ if (unlikely(err < 0)) {
+ if (err == -ECONNRESET)
+ ds_conn_reset(dp);
+ break;
+ }
+ if (err < tag->len)
+ break;
+
+ if (tag->type < DS_DATA)
+ err = ds_handshake(dp, dp->rcv_buf);
+ else
+ err = ds_data(dp, dp->rcv_buf,
+ sizeof(*tag) + err);
+ if (err == -ECONNRESET)
+ break;
+ }
+
+ spin_unlock_irqrestore(&ds_lock, flags);
+}
+
+static int ds_probe(struct vio_dev *vdev, const struct vio_device_id *id)
+{
+ static int ds_version_printed;
+ struct ldc_channel_config ds_cfg = {
+ .event = ds_event,
+ .mtu = 4096,
+ .mode = LDC_MODE_STREAM,
+ };
+ struct mdesc_handle *hp;
+ struct ldc_channel *lp;
+ struct ds_info *dp;
+ const u64 *val;
+ int err, i;
+
+ if (ds_version_printed++ == 0)
+ printk(KERN_INFO "%s", version);
+
+ dp = kzalloc(sizeof(*dp), GFP_KERNEL);
+ err = -ENOMEM;
+ if (!dp)
+ goto out_err;
+
+ hp = mdesc_grab();
+ val = mdesc_get_property(hp, vdev->mp, "id", NULL);
+ if (val)
+ dp->id = *val;
+ mdesc_release(hp);
+
+ dp->rcv_buf = kzalloc(4096, GFP_KERNEL);
+ if (!dp->rcv_buf)
+ goto out_free_dp;
+
+ dp->rcv_buf_len = 4096;
+
+ dp->ds_states = kmemdup(ds_states_template,
+ sizeof(ds_states_template), GFP_KERNEL);
+ if (!dp->ds_states)
+ goto out_free_rcv_buf;
+
+ dp->num_ds_states = ARRAY_SIZE(ds_states_template);
+
+ for (i = 0; i < dp->num_ds_states; i++)
+ dp->ds_states[i].handle = ((u64)i << 32);
+
+ ds_cfg.tx_irq = vdev->tx_irq;
+ ds_cfg.rx_irq = vdev->rx_irq;
+
+ lp = ldc_alloc(vdev->channel_id, &ds_cfg, dp);
+ if (IS_ERR(lp)) {
+ err = PTR_ERR(lp);
+ goto out_free_ds_states;
+ }
+ dp->lp = lp;
+
+ err = ldc_bind(lp, "DS");
+ if (err)
+ goto out_free_ldc;
+
+ spin_lock_irq(&ds_lock);
+ dp->next = ds_info_list;
+ ds_info_list = dp;
+ spin_unlock_irq(&ds_lock);
+
+ return err;
+
+out_free_ldc:
+ ldc_free(dp->lp);
+
+out_free_ds_states:
+ kfree(dp->ds_states);
+
+out_free_rcv_buf:
+ kfree(dp->rcv_buf);
+
+out_free_dp:
+ kfree(dp);
+
+out_err:
+ return err;
+}
+
+static int ds_remove(struct vio_dev *vdev)
+{
+ return 0;
+}
+
+static const struct vio_device_id ds_match[] = {
+ {
+ .type = "domain-services-port",
+ },
+ {},
+};
+
+static struct vio_driver ds_driver = {
+ .id_table = ds_match,
+ .probe = ds_probe,
+ .remove = ds_remove,
+ .name = "ds",
+};
+
+static int __init ds_init(void)
+{
+ unsigned long hv_ret, major, minor;
+
+ if (tlb_type == hypervisor) {
+ hv_ret = sun4v_get_version(HV_GRP_REBOOT_DATA, &major, &minor);
+ if (hv_ret == HV_EOK) {
+ pr_info("SUN4V: Reboot data supported (maj=%lu,min=%lu).\n",
+ major, minor);
+ reboot_data_supported = 1;
+ }
+ }
+ kthread_run(ds_thread, NULL, "kldomd");
+
+ return vio_register_driver(&ds_driver);
+}
+
+fs_initcall(ds_init);
diff --git a/arch/sparc/kernel/dtlb_miss.S b/arch/sparc/kernel/dtlb_miss.S
new file mode 100644
index 00000000000..09a6a15a710
--- /dev/null
+++ b/arch/sparc/kernel/dtlb_miss.S
@@ -0,0 +1,39 @@
+/* DTLB ** ICACHE line 1: Context 0 check and TSB load */
+ ldxa [%g0] ASI_DMMU_TSB_8KB_PTR, %g1 ! Get TSB 8K pointer
+ ldxa [%g0] ASI_DMMU, %g6 ! Get TAG TARGET
+ srlx %g6, 48, %g5 ! Get context
+ sllx %g6, 22, %g6 ! Zero out context
+ brz,pn %g5, kvmap_dtlb ! Context 0 processing
+ srlx %g6, 22, %g6 ! Delay slot
+ TSB_LOAD_QUAD(%g1, %g4) ! Load TSB entry
+ cmp %g4, %g6 ! Compare TAG
+
+/* DTLB ** ICACHE line 2: TSB compare and TLB load */
+ bne,pn %xcc, tsb_miss_dtlb ! Miss
+ mov FAULT_CODE_DTLB, %g3
+ stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Load TLB
+ retry ! Trap done
+ nop
+ nop
+ nop
+ nop
+
+/* DTLB ** ICACHE line 3: */
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+
+/* DTLB ** ICACHE line 4: */
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
diff --git a/arch/sparc/kernel/dtlb_prot.S b/arch/sparc/kernel/dtlb_prot.S
new file mode 100644
index 00000000000..b2c2c5be281
--- /dev/null
+++ b/arch/sparc/kernel/dtlb_prot.S
@@ -0,0 +1,54 @@
+/*
+ * dtlb_prot.S: DTLB protection trap strategy.
+ * This is included directly into the trap table.
+ *
+ * Copyright (C) 1996,1998 David S. Miller (davem@redhat.com)
+ * Copyright (C) 1997,1998 Jakub Jelinek (jj@ultra.linux.cz)
+ */
+
+/* Ways we can get here:
+ *
+ * [TL == 0] 1) User stores to readonly pages.
+ * [TL == 0] 2) Nucleus stores to user readonly pages.
+ * [TL > 0] 3) Nucleus stores to user readonly stack frame.
+ */
+
+/* PROT ** ICACHE line 1: User DTLB protection trap */
+ mov TLB_SFSR, %g1
+ stxa %g0, [%g1] ASI_DMMU ! Clear FaultValid bit
+ membar #Sync ! Synchronize stores
+ rdpr %pstate, %g5 ! Move into alt-globals
+ wrpr %g5, PSTATE_AG|PSTATE_MG, %pstate
+ rdpr %tl, %g1 ! Need a winfixup?
+ cmp %g1, 1 ! Trap level >1?
+ mov TLB_TAG_ACCESS, %g4 ! For reload of vaddr
+
+/* PROT ** ICACHE line 2: More real fault processing */
+ bgu,pn %xcc, winfix_trampoline ! Yes, perform winfixup
+ ldxa [%g4] ASI_DMMU, %g5 ! Put tagaccess in %g5
+ ba,pt %xcc, sparc64_realfault_common ! Nope, normal fault
+ mov FAULT_CODE_DTLB | FAULT_CODE_WRITE, %g4
+ nop
+ nop
+ nop
+ nop
+
+/* PROT ** ICACHE line 3: Unused... */
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+
+/* PROT ** ICACHE line 4: Unused... */
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
diff --git a/arch/sparc/kernel/ebus.c b/arch/sparc/kernel/ebus.c
index d850785b208..acf8314cec4 100644
--- a/arch/sparc/kernel/ebus.c
+++ b/arch/sparc/kernel/ebus.c
@@ -1,392 +1,256 @@
-/* $Id: ebus.c,v 1.20 2002/01/05 01:13:43 davem Exp $
- * ebus.c: PCI to EBus bridge device.
+/* ebus.c: EBUS DMA library code.
*
* Copyright (C) 1997 Eddie C. Dost (ecd@skynet.be)
- *
- * Adopted for sparc by V. Roganov and G. Raiko.
- * Fixes for different platforms by Pete Zaitcev.
+ * Copyright (C) 1999 David S. Miller (davem@redhat.com)
*/
+#include <linux/export.h>
#include <linux/kernel.h>
#include <linux/types.h>
-#include <linux/init.h>
-#include <linux/slab.h>
-#include <linux/string.h>
-
-#include <asm/system.h>
-#include <asm/page.h>
-#include <asm/pbm.h>
-#include <asm/ebus.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+
+#include <asm/ebus_dma.h>
#include <asm/io.h>
-#include <asm/oplib.h>
-#include <asm/prom.h>
-#include <asm/bpp.h>
-struct linux_ebus *ebus_chain = NULL;
+#define EBDMA_CSR 0x00UL /* Control/Status */
+#define EBDMA_ADDR 0x04UL /* DMA Address */
+#define EBDMA_COUNT 0x08UL /* DMA Count */
+
+#define EBDMA_CSR_INT_PEND 0x00000001
+#define EBDMA_CSR_ERR_PEND 0x00000002
+#define EBDMA_CSR_DRAIN 0x00000004
+#define EBDMA_CSR_INT_EN 0x00000010
+#define EBDMA_CSR_RESET 0x00000080
+#define EBDMA_CSR_WRITE 0x00000100
+#define EBDMA_CSR_EN_DMA 0x00000200
+#define EBDMA_CSR_CYC_PEND 0x00000400
+#define EBDMA_CSR_DIAG_RD_DONE 0x00000800
+#define EBDMA_CSR_DIAG_WR_DONE 0x00001000
+#define EBDMA_CSR_EN_CNT 0x00002000
+#define EBDMA_CSR_TC 0x00004000
+#define EBDMA_CSR_DIS_CSR_DRN 0x00010000
+#define EBDMA_CSR_BURST_SZ_MASK 0x000c0000
+#define EBDMA_CSR_BURST_SZ_1 0x00080000
+#define EBDMA_CSR_BURST_SZ_4 0x00000000
+#define EBDMA_CSR_BURST_SZ_8 0x00040000
+#define EBDMA_CSR_BURST_SZ_16 0x000c0000
+#define EBDMA_CSR_DIAG_EN 0x00100000
+#define EBDMA_CSR_DIS_ERR_PEND 0x00400000
+#define EBDMA_CSR_TCI_DIS 0x00800000
+#define EBDMA_CSR_EN_NEXT 0x01000000
+#define EBDMA_CSR_DMA_ON 0x02000000
+#define EBDMA_CSR_A_LOADED 0x04000000
+#define EBDMA_CSR_NA_LOADED 0x08000000
+#define EBDMA_CSR_DEV_ID_MASK 0xf0000000
+
+#define EBUS_DMA_RESET_TIMEOUT 10000
+
+static void __ebus_dma_reset(struct ebus_dma_info *p, int no_drain)
+{
+ int i;
+ u32 val = 0;
-/* We are together with pcic.c under CONFIG_PCI. */
-extern unsigned int pcic_pin_to_irq(unsigned int, const char *name);
+ writel(EBDMA_CSR_RESET, p->regs + EBDMA_CSR);
+ udelay(1);
-/*
- * IRQ Blacklist
- * Here we list PROMs and systems that are known to supply crap as IRQ numbers.
- */
-struct ebus_device_irq {
- char *name;
- unsigned int pin;
-};
-
-struct ebus_system_entry {
- char *esname;
- struct ebus_device_irq *ipt;
-};
-
-static struct ebus_device_irq je1_1[] = {
- { "8042", 3 },
- { "SUNW,CS4231", 0 },
- { "parallel", 0 },
- { "se", 2 },
- { NULL, 0 }
-};
-
-/*
- * Gleb's JE1 supplied reasonable pin numbers, but mine did not (OBP 2.32).
- * Blacklist the sucker... Note that Gleb's system will work.
- */
-static struct ebus_system_entry ebus_blacklist[] = {
- { "SUNW,JavaEngine1", je1_1 },
- { NULL, NULL }
-};
+ if (no_drain)
+ return;
-static struct ebus_device_irq *ebus_blackp = NULL;
+ for (i = EBUS_DMA_RESET_TIMEOUT; i > 0; i--) {
+ val = readl(p->regs + EBDMA_CSR);
-/*
- */
-static inline unsigned long ebus_alloc(size_t size)
-{
- return (unsigned long)kmalloc(size, GFP_ATOMIC);
+ if (!(val & (EBDMA_CSR_DRAIN | EBDMA_CSR_CYC_PEND)))
+ break;
+ udelay(10);
+ }
}
-/*
- */
-int __init ebus_blacklist_irq(const char *name)
+static irqreturn_t ebus_dma_irq(int irq, void *dev_id)
{
- struct ebus_device_irq *dp;
-
- if ((dp = ebus_blackp) != NULL) {
- for (; dp->name != NULL; dp++) {
- if (strcmp(name, dp->name) == 0) {
- return pcic_pin_to_irq(dp->pin, name);
- }
- }
+ struct ebus_dma_info *p = dev_id;
+ unsigned long flags;
+ u32 csr = 0;
+
+ spin_lock_irqsave(&p->lock, flags);
+ csr = readl(p->regs + EBDMA_CSR);
+ writel(csr, p->regs + EBDMA_CSR);
+ spin_unlock_irqrestore(&p->lock, flags);
+
+ if (csr & EBDMA_CSR_ERR_PEND) {
+ printk(KERN_CRIT "ebus_dma(%s): DMA error!\n", p->name);
+ p->callback(p, EBUS_DMA_EVENT_ERROR, p->client_cookie);
+ return IRQ_HANDLED;
+ } else if (csr & EBDMA_CSR_INT_PEND) {
+ p->callback(p,
+ (csr & EBDMA_CSR_TC) ?
+ EBUS_DMA_EVENT_DMA : EBUS_DMA_EVENT_DEVICE,
+ p->client_cookie);
+ return IRQ_HANDLED;
}
- return 0;
+
+ return IRQ_NONE;
+
}
-void __init fill_ebus_child(struct device_node *dp,
- struct linux_ebus_child *dev)
+int ebus_dma_register(struct ebus_dma_info *p)
{
- const int *regs;
- const int *irqs;
- int i, len;
-
- dev->prom_node = dp;
- regs = of_get_property(dp, "reg", &len);
- if (!regs)
- len = 0;
- dev->num_addrs = len / sizeof(regs[0]);
-
- for (i = 0; i < dev->num_addrs; i++) {
- if (regs[i] >= dev->parent->num_addrs) {
- prom_printf("UGH: property for %s was %d, need < %d\n",
- dev->prom_node->name, len,
- dev->parent->num_addrs);
- panic(__FUNCTION__);
- }
+ u32 csr;
- /* XXX resource */
- dev->resource[i].start =
- dev->parent->resource[regs[i]].start;
- }
+ if (!p->regs)
+ return -EINVAL;
+ if (p->flags & ~(EBUS_DMA_FLAG_USE_EBDMA_HANDLER |
+ EBUS_DMA_FLAG_TCI_DISABLE))
+ return -EINVAL;
+ if ((p->flags & EBUS_DMA_FLAG_USE_EBDMA_HANDLER) && !p->callback)
+ return -EINVAL;
+ if (!strlen(p->name))
+ return -EINVAL;
- for (i = 0; i < PROMINTR_MAX; i++)
- dev->irqs[i] = PCI_IRQ_NONE;
+ __ebus_dma_reset(p, 1);
- if ((dev->irqs[0] = ebus_blacklist_irq(dev->prom_node->name)) != 0) {
- dev->num_irqs = 1;
- } else {
- irqs = of_get_property(dp, "interrupts", &len);
- if (!irqs) {
- dev->num_irqs = 0;
- dev->irqs[0] = 0;
- if (dev->parent->num_irqs != 0) {
- dev->num_irqs = 1;
- dev->irqs[0] = dev->parent->irqs[0];
- }
- } else {
- dev->num_irqs = len / sizeof(irqs[0]);
- if (irqs[0] == 0 || irqs[0] >= 8) {
- /*
- * XXX Zero is a valid pin number...
- * This works as long as Ebus is not wired
- * to INTA#.
- */
- printk("EBUS: %s got bad irq %d from PROM\n",
- dev->prom_node->name, irqs[0]);
- dev->num_irqs = 0;
- dev->irqs[0] = 0;
- } else {
- dev->irqs[0] =
- pcic_pin_to_irq(irqs[0],
- dev->prom_node->name);
- }
- }
- }
+ csr = EBDMA_CSR_BURST_SZ_16 | EBDMA_CSR_EN_CNT;
+
+ if (p->flags & EBUS_DMA_FLAG_TCI_DISABLE)
+ csr |= EBDMA_CSR_TCI_DIS;
+
+ writel(csr, p->regs + EBDMA_CSR);
+
+ return 0;
}
+EXPORT_SYMBOL(ebus_dma_register);
-void __init fill_ebus_device(struct device_node *dp, struct linux_ebus_device *dev)
+int ebus_dma_irq_enable(struct ebus_dma_info *p, int on)
{
- const struct linux_prom_registers *regs;
- struct linux_ebus_child *child;
- struct dev_archdata *sd;
- const int *irqs;
- int i, n, len;
- unsigned long baseaddr;
-
- dev->prom_node = dp;
-
- regs = of_get_property(dp, "reg", &len);
- if (!regs)
- len = 0;
- if (len % sizeof(struct linux_prom_registers)) {
- prom_printf("UGH: proplen for %s was %d, need multiple of %d\n",
- dev->prom_node->name, len,
- (int)sizeof(struct linux_prom_registers));
- panic(__FUNCTION__);
- }
- dev->num_addrs = len / sizeof(struct linux_prom_registers);
-
- for (i = 0; i < dev->num_addrs; i++) {
- /*
- * XXX Collect JE-1 PROM
- *
- * Example - JS-E with 3.11:
- * /ebus
- * regs
- * 0x00000000, 0x0, 0x00000000, 0x0, 0x00000000,
- * 0x82000010, 0x0, 0xf0000000, 0x0, 0x01000000,
- * 0x82000014, 0x0, 0x38800000, 0x0, 0x00800000,
- * ranges
- * 0x00, 0x00000000, 0x02000010, 0x0, 0x0, 0x01000000,
- * 0x01, 0x01000000, 0x02000014, 0x0, 0x0, 0x00800000,
- * /ebus/8042
- * regs
- * 0x00000001, 0x00300060, 0x00000008,
- * 0x00000001, 0x00300060, 0x00000008,
- */
- n = regs[i].which_io;
- if (n >= 4) {
- /* XXX This is copied from old JE-1 by Gleb. */
- n = (regs[i].which_io - 0x10) >> 2;
- } else {
- ;
- }
+ unsigned long flags;
+ u32 csr;
-/*
- * XXX Now as we have regions, why don't we make an on-demand allocation...
- */
- dev->resource[i].start = 0;
- if ((baseaddr = dev->bus->self->resource[n].start +
- regs[i].phys_addr) != 0) {
- /* dev->resource[i].name = dev->prom_name; */
- if ((baseaddr = (unsigned long) ioremap(baseaddr,
- regs[i].reg_size)) == 0) {
- panic("ebus: unable to remap dev %s",
- dev->prom_node->name);
- }
+ if (on) {
+ if (p->flags & EBUS_DMA_FLAG_USE_EBDMA_HANDLER) {
+ if (request_irq(p->irq, ebus_dma_irq, IRQF_SHARED, p->name, p))
+ return -EBUSY;
}
- dev->resource[i].start = baseaddr; /* XXX Unaligned */
- }
-
- for (i = 0; i < PROMINTR_MAX; i++)
- dev->irqs[i] = PCI_IRQ_NONE;
- if ((dev->irqs[0] = ebus_blacklist_irq(dev->prom_node->name)) != 0) {
- dev->num_irqs = 1;
+ spin_lock_irqsave(&p->lock, flags);
+ csr = readl(p->regs + EBDMA_CSR);
+ csr |= EBDMA_CSR_INT_EN;
+ writel(csr, p->regs + EBDMA_CSR);
+ spin_unlock_irqrestore(&p->lock, flags);
} else {
- irqs = of_get_property(dp, "interrupts", &len);
- if (!irqs) {
- dev->num_irqs = 0;
- if ((dev->irqs[0] = dev->bus->self->irq) != 0) {
- dev->num_irqs = 1;
-/* P3 */ /* printk("EBUS: child %s irq %d from parent\n", dev->prom_name, dev->irqs[0]); */
- }
- } else {
- dev->num_irqs = 1; /* dev->num_irqs = len / sizeof(irqs[0]); */
- if (irqs[0] == 0 || irqs[0] >= 8) {
- /* See above for the parent. XXX */
- printk("EBUS: %s got bad irq %d from PROM\n",
- dev->prom_node->name, irqs[0]);
- dev->num_irqs = 0;
- dev->irqs[0] = 0;
- } else {
- dev->irqs[0] =
- pcic_pin_to_irq(irqs[0],
- dev->prom_node->name);
- }
+ spin_lock_irqsave(&p->lock, flags);
+ csr = readl(p->regs + EBDMA_CSR);
+ csr &= ~EBDMA_CSR_INT_EN;
+ writel(csr, p->regs + EBDMA_CSR);
+ spin_unlock_irqrestore(&p->lock, flags);
+
+ if (p->flags & EBUS_DMA_FLAG_USE_EBDMA_HANDLER) {
+ free_irq(p->irq, p);
}
}
- sd = &dev->ofdev.dev.archdata;
- sd->prom_node = dp;
- sd->op = &dev->ofdev;
- sd->iommu = dev->bus->ofdev.dev.parent->archdata.iommu;
-
- dev->ofdev.node = dp;
- dev->ofdev.dev.parent = &dev->bus->ofdev.dev;
- dev->ofdev.dev.bus = &ebus_bus_type;
- sprintf(dev->ofdev.dev.bus_id, "ebus[%08x]", dp->node);
-
- /* Register with core */
- if (of_device_register(&dev->ofdev) != 0)
- printk(KERN_DEBUG "ebus: device registration error for %s!\n",
- dp->path_component_name);
-
- if ((dp = dp->child) != NULL) {
- dev->children = (struct linux_ebus_child *)
- ebus_alloc(sizeof(struct linux_ebus_child));
-
- child = dev->children;
- child->next = NULL;
- child->parent = dev;
- child->bus = dev->bus;
- fill_ebus_child(dp, child);
-
- while ((dp = dp->sibling) != NULL) {
- child->next = (struct linux_ebus_child *)
- ebus_alloc(sizeof(struct linux_ebus_child));
-
- child = child->next;
- child->next = NULL;
- child->parent = dev;
- child->bus = dev->bus;
- fill_ebus_child(dp, child);
- }
- }
+ return 0;
}
+EXPORT_SYMBOL(ebus_dma_irq_enable);
-void __init ebus_init(void)
+void ebus_dma_unregister(struct ebus_dma_info *p)
{
- const struct linux_prom_pci_registers *regs;
- struct linux_pbm_info *pbm;
- struct linux_ebus_device *dev;
- struct linux_ebus *ebus;
- struct ebus_system_entry *sp;
- struct pci_dev *pdev;
- struct pcidev_cookie *cookie;
- struct device_node *dp;
- struct resource *p;
- unsigned short pci_command;
- int len, reg, nreg;
- int num_ebus = 0;
-
- dp = of_find_node_by_path("/");
- for (sp = ebus_blacklist; sp->esname != NULL; sp++) {
- if (strcmp(dp->name, sp->esname) == 0) {
- ebus_blackp = sp->ipt;
- break;
- }
+ unsigned long flags;
+ u32 csr;
+ int irq_on = 0;
+
+ spin_lock_irqsave(&p->lock, flags);
+ csr = readl(p->regs + EBDMA_CSR);
+ if (csr & EBDMA_CSR_INT_EN) {
+ csr &= ~EBDMA_CSR_INT_EN;
+ writel(csr, p->regs + EBDMA_CSR);
+ irq_on = 1;
}
+ spin_unlock_irqrestore(&p->lock, flags);
- pdev = pci_get_device(PCI_VENDOR_ID_SUN, PCI_DEVICE_ID_SUN_EBUS, NULL);
- if (!pdev)
- return;
-
- cookie = pdev->sysdata;
- dp = cookie->prom_node;
-
- ebus_chain = ebus = (struct linux_ebus *)
- ebus_alloc(sizeof(struct linux_ebus));
- ebus->next = NULL;
-
- while (dp) {
- struct device_node *nd;
-
- ebus->prom_node = dp;
- ebus->self = pdev;
- ebus->parent = pbm = cookie->pbm;
-
- /* Enable BUS Master. */
- pci_read_config_word(pdev, PCI_COMMAND, &pci_command);
- pci_command |= PCI_COMMAND_MASTER;
- pci_write_config_word(pdev, PCI_COMMAND, pci_command);
-
- regs = of_get_property(dp, "reg", &len);
- if (!regs) {
- prom_printf("%s: can't find reg property\n",
- __FUNCTION__);
- prom_halt();
- }
- nreg = len / sizeof(struct linux_prom_pci_registers);
-
- p = &ebus->self->resource[0];
- for (reg = 0; reg < nreg; reg++) {
- if (!(regs[reg].which_io & 0x03000000))
- continue;
-
- (p++)->start = regs[reg].phys_lo;
- }
-
- ebus->ofdev.node = dp;
- ebus->ofdev.dev.parent = &pdev->dev;
- ebus->ofdev.dev.bus = &ebus_bus_type;
- sprintf(ebus->ofdev.dev.bus_id, "ebus%d", num_ebus);
+ if (irq_on)
+ free_irq(p->irq, p);
+}
+EXPORT_SYMBOL(ebus_dma_unregister);
- /* Register with core */
- if (of_device_register(&ebus->ofdev) != 0)
- printk(KERN_DEBUG "ebus: device registration error for %s!\n",
- dp->path_component_name);
+int ebus_dma_request(struct ebus_dma_info *p, dma_addr_t bus_addr, size_t len)
+{
+ unsigned long flags;
+ u32 csr;
+ int err;
+
+ if (len >= (1 << 24))
+ return -EINVAL;
+
+ spin_lock_irqsave(&p->lock, flags);
+ csr = readl(p->regs + EBDMA_CSR);
+ err = -EINVAL;
+ if (!(csr & EBDMA_CSR_EN_DMA))
+ goto out;
+ err = -EBUSY;
+ if (csr & EBDMA_CSR_NA_LOADED)
+ goto out;
+
+ writel(len, p->regs + EBDMA_COUNT);
+ writel(bus_addr, p->regs + EBDMA_ADDR);
+ err = 0;
+
+out:
+ spin_unlock_irqrestore(&p->lock, flags);
+
+ return err;
+}
+EXPORT_SYMBOL(ebus_dma_request);
+void ebus_dma_prepare(struct ebus_dma_info *p, int write)
+{
+ unsigned long flags;
+ u32 csr;
- nd = dp->child;
- if (!nd)
- goto next_ebus;
+ spin_lock_irqsave(&p->lock, flags);
+ __ebus_dma_reset(p, 0);
- ebus->devices = (struct linux_ebus_device *)
- ebus_alloc(sizeof(struct linux_ebus_device));
+ csr = (EBDMA_CSR_INT_EN |
+ EBDMA_CSR_EN_CNT |
+ EBDMA_CSR_BURST_SZ_16 |
+ EBDMA_CSR_EN_NEXT);
- dev = ebus->devices;
- dev->next = NULL;
- dev->children = NULL;
- dev->bus = ebus;
- fill_ebus_device(nd, dev);
+ if (write)
+ csr |= EBDMA_CSR_WRITE;
+ if (p->flags & EBUS_DMA_FLAG_TCI_DISABLE)
+ csr |= EBDMA_CSR_TCI_DIS;
- while ((nd = nd->sibling) != NULL) {
- dev->next = (struct linux_ebus_device *)
- ebus_alloc(sizeof(struct linux_ebus_device));
+ writel(csr, p->regs + EBDMA_CSR);
- dev = dev->next;
- dev->next = NULL;
- dev->children = NULL;
- dev->bus = ebus;
- fill_ebus_device(nd, dev);
- }
+ spin_unlock_irqrestore(&p->lock, flags);
+}
+EXPORT_SYMBOL(ebus_dma_prepare);
- next_ebus:
- pdev = pci_get_device(PCI_VENDOR_ID_SUN,
- PCI_DEVICE_ID_SUN_EBUS, pdev);
- if (!pdev)
- break;
+unsigned int ebus_dma_residue(struct ebus_dma_info *p)
+{
+ return readl(p->regs + EBDMA_COUNT);
+}
+EXPORT_SYMBOL(ebus_dma_residue);
- cookie = pdev->sysdata;
- dp = cookie->prom_node;
+unsigned int ebus_dma_addr(struct ebus_dma_info *p)
+{
+ return readl(p->regs + EBDMA_ADDR);
+}
+EXPORT_SYMBOL(ebus_dma_addr);
- ebus->next = (struct linux_ebus *)
- ebus_alloc(sizeof(struct linux_ebus));
- ebus = ebus->next;
- ebus->next = NULL;
- ++num_ebus;
- }
- if (pdev)
- pci_dev_put(pdev);
+void ebus_dma_enable(struct ebus_dma_info *p, int on)
+{
+ unsigned long flags;
+ u32 orig_csr, csr;
+
+ spin_lock_irqsave(&p->lock, flags);
+ orig_csr = csr = readl(p->regs + EBDMA_CSR);
+ if (on)
+ csr |= EBDMA_CSR_EN_DMA;
+ else
+ csr &= ~EBDMA_CSR_EN_DMA;
+ if ((orig_csr & EBDMA_CSR_EN_DMA) !=
+ (csr & EBDMA_CSR_EN_DMA))
+ writel(csr, p->regs + EBDMA_CSR);
+ spin_unlock_irqrestore(&p->lock, flags);
}
+EXPORT_SYMBOL(ebus_dma_enable);
diff --git a/arch/sparc/kernel/entry.S b/arch/sparc/kernel/entry.S
index c2eed8f7151..33c02b15f47 100644
--- a/arch/sparc/kernel/entry.S
+++ b/arch/sparc/kernel/entry.S
@@ -7,24 +7,19 @@
* Copyright (C) 1997 Anton Blanchard (anton@progsoc.uts.edu.au)
*/
+#include <linux/linkage.h>
#include <linux/errno.h>
#include <asm/head.h>
#include <asm/asi.h>
#include <asm/smp.h>
-#include <asm/kgdb.h>
#include <asm/contregs.h>
#include <asm/ptrace.h>
#include <asm/asm-offsets.h>
#include <asm/psr.h>
#include <asm/vaddrs.h>
-#include <asm/memreg.h>
#include <asm/page.h>
-#ifdef CONFIG_SUN4
-#include <asm/pgtsun4.h>
-#else
-#include <asm/pgtsun4c.h>
-#endif
+#include <asm/pgtable.h>
#include <asm/winmacro.h>
#include <asm/signal.h>
#include <asm/obio.h>
@@ -45,91 +40,20 @@
_SV; _SV; _SV; _SV; _SV; _SV; _SV; \
_RS; _RS; _RS; _RS; _RS; _RS; _RS;
-/* First, KGDB low level things. This is a rewrite
- * of the routines found in the sparc-stub.c asm() statement
- * from the gdb distribution. This is also dual-purpose
- * as a software trap for userlevel programs.
- */
- .data
- .align 4
-
-in_trap_handler:
- .word 0
-
.text
- .align 4
-
-#if 0 /* kgdb is dropped from 2.5.33 */
-! This function is called when any SPARC trap (except window overflow or
-! underflow) occurs. It makes sure that the invalid register window is still
-! available before jumping into C code. It will also restore the world if you
-! return from handle_exception.
-
- .globl trap_low
-trap_low:
- rd %wim, %l3
- SAVE_ALL
-
- sethi %hi(in_trap_handler), %l4
- ld [%lo(in_trap_handler) + %l4], %l5
- inc %l5
- st %l5, [%lo(in_trap_handler) + %l4]
-
- /* Make sure kgdb sees the same state we just saved. */
- LOAD_PT_GLOBALS(sp)
- LOAD_PT_INS(sp)
- ld [%sp + STACKFRAME_SZ + PT_Y], %l4
- ld [%sp + STACKFRAME_SZ + PT_WIM], %l3
- ld [%sp + STACKFRAME_SZ + PT_PSR], %l0
- ld [%sp + STACKFRAME_SZ + PT_PC], %l1
- ld [%sp + STACKFRAME_SZ + PT_NPC], %l2
- rd %tbr, %l5 /* Never changes... */
-
- /* Make kgdb exception frame. */
- sub %sp,(16+1+6+1+72)*4,%sp ! Make room for input & locals
- ! + hidden arg + arg spill
- ! + doubleword alignment
- ! + registers[72] local var
- SAVE_KGDB_GLOBALS(sp)
- SAVE_KGDB_INS(sp)
- SAVE_KGDB_SREGS(sp, l4, l0, l3, l5, l1, l2)
-
- /* We are increasing PIL, so two writes. */
- or %l0, PSR_PIL, %l0
- wr %l0, 0, %psr
- WRITE_PAUSE
- wr %l0, PSR_ET, %psr
- WRITE_PAUSE
-
- call handle_exception
- add %sp, STACKFRAME_SZ, %o0 ! Pass address of registers
-
- /* Load new kgdb register set. */
- LOAD_KGDB_GLOBALS(sp)
- LOAD_KGDB_INS(sp)
- LOAD_KGDB_SREGS(sp, l4, l0, l3, l5, l1, l2)
- wr %l4, 0x0, %y
-
- sethi %hi(in_trap_handler), %l4
- ld [%lo(in_trap_handler) + %l4], %l5
- dec %l5
- st %l5, [%lo(in_trap_handler) + %l4]
-
- add %sp,(16+1+6+1+72)*4,%sp ! Undo the kgdb trap frame.
-
- /* Now take what kgdb did and place it into the pt_regs
- * frame which SparcLinux RESTORE_ALL understands.,
- */
- STORE_PT_INS(sp)
- STORE_PT_GLOBALS(sp)
- STORE_PT_YREG(sp, g2)
- STORE_PT_PRIV(sp, l0, l1, l2)
- RESTORE_ALL
+#ifdef CONFIG_KGDB
+ .align 4
+ .globl arch_kgdb_breakpoint
+ .type arch_kgdb_breakpoint,#function
+arch_kgdb_breakpoint:
+ ta 0x7d
+ retl
+ nop
+ .size arch_kgdb_breakpoint,.-arch_kgdb_breakpoint
#endif
#if defined(CONFIG_BLK_DEV_FD) || defined(CONFIG_BLK_DEV_FD_MODULE)
- .text
.align 4
.globl floppy_hardint
floppy_hardint:
@@ -200,22 +124,11 @@ floppy_tdone:
set auxio_register, %l7
ld [%l7], %l7
- set sparc_cpu_model, %l5
- ld [%l5], %l5
- subcc %l5, 1, %g0 /* enum { sun4c = 1 }; */
- be 1f
- ldub [%l7], %l5
+ ldub [%l7], %l5
or %l5, 0xc2, %l5
stb %l5, [%l7]
andn %l5, 0x02, %l5
- b 2f
- nop
-
-1:
- or %l5, 0xf4, %l5
- stb %l5, [%l7]
- andn %l5, 0x04, %l5
2:
/* Kill some time so the bits set */
@@ -304,7 +217,7 @@ real_irq_entry:
#ifdef CONFIG_SMP
.globl patchme_maybe_smp_msg
- cmp %l7, 12
+ cmp %l7, 11
patchme_maybe_smp_msg:
bgu maybe_smp4m_msg
nop
@@ -341,32 +254,59 @@ smp4m_ticker:
WRITE_PAUSE
RESTORE_ALL
+#define GET_PROCESSOR4M_ID(reg) \
+ rd %tbr, %reg; \
+ srl %reg, 12, %reg; \
+ and %reg, 3, %reg;
+
/* Here is where we check for possible SMP IPI passed to us
* on some level other than 15 which is the NMI and only used
* for cross calls. That has a separate entry point below.
+ *
+ * IPIs are sent on Level 12, 13 and 14. See IRQ_IPI_*.
*/
maybe_smp4m_msg:
GET_PROCESSOR4M_ID(o3)
- set sun4m_interrupts, %l5
- ld [%l5], %o5
- sethi %hi(0x40000000), %o2
- sll %o3, 12, %o3
+ sethi %hi(sun4m_irq_percpu), %l5
+ sll %o3, 2, %o3
+ or %l5, %lo(sun4m_irq_percpu), %o5
+ sethi %hi(0x70000000), %o2 ! Check all soft-IRQs
ld [%o5 + %o3], %o1
- andcc %o1, %o2, %g0
+ ld [%o1 + 0x00], %o3 ! sun4m_irq_percpu[cpu]->pending
+ andcc %o3, %o2, %g0
be,a smp4m_ticker
cmp %l7, 14
- st %o2, [%o5 + 0x4]
+ /* Soft-IRQ IPI */
+ st %o2, [%o1 + 0x04] ! sun4m_irq_percpu[cpu]->clear=0x70000000
WRITE_PAUSE
- ld [%o5], %g0
+ ld [%o1 + 0x00], %g0 ! sun4m_irq_percpu[cpu]->pending
WRITE_PAUSE
or %l0, PSR_PIL, %l4
wr %l4, 0x0, %psr
WRITE_PAUSE
wr %l4, PSR_ET, %psr
WRITE_PAUSE
- call smp_reschedule_irq
+ srl %o3, 28, %o2 ! shift for simpler checks below
+maybe_smp4m_msg_check_single:
+ andcc %o2, 0x1, %g0
+ beq,a maybe_smp4m_msg_check_mask
+ andcc %o2, 0x2, %g0
+ call smp_call_function_single_interrupt
nop
-
+ andcc %o2, 0x2, %g0
+maybe_smp4m_msg_check_mask:
+ beq,a maybe_smp4m_msg_check_resched
+ andcc %o2, 0x4, %g0
+ call smp_call_function_interrupt
+ nop
+ andcc %o2, 0x4, %g0
+maybe_smp4m_msg_check_resched:
+ /* rescheduling is done in RESTORE_ALL regardless, but incr stats */
+ beq,a maybe_smp4m_msg_out
+ nop
+ call smp_resched_interrupt
+ nop
+maybe_smp4m_msg_out:
RESTORE_ALL
.align 4
@@ -375,16 +315,16 @@ linux_trap_ipi15_sun4m:
SAVE_ALL
sethi %hi(0x80000000), %o2
GET_PROCESSOR4M_ID(o0)
- set sun4m_interrupts, %l5
- ld [%l5], %o5
- sll %o0, 12, %o0
- add %o5, %o0, %o5
- ld [%o5], %o3
+ sethi %hi(sun4m_irq_percpu), %l5
+ or %l5, %lo(sun4m_irq_percpu), %o5
+ sll %o0, 2, %o0
+ ld [%o5 + %o0], %o5
+ ld [%o5 + 0x00], %o3 ! sun4m_irq_percpu[cpu]->pending
andcc %o3, %o2, %g0
- be 1f ! Must be an NMI async memory error
- st %o2, [%o5 + 4]
+ be sun4m_nmi_error ! Must be an NMI async memory error
+ st %o2, [%o5 + 0x04] ! sun4m_irq_percpu[cpu]->clear=0x80000000
WRITE_PAUSE
- ld [%o5], %g0
+ ld [%o5 + 0x00], %g0 ! sun4m_irq_percpu[cpu]->pending
WRITE_PAUSE
or %l0, PSR_PIL, %l4
wr %l4, 0x0, %psr
@@ -395,28 +335,6 @@ linux_trap_ipi15_sun4m:
nop
b ret_trap_lockless_ipi
clr %l6
-1:
- /* NMI async memory error handling. */
- sethi %hi(0x80000000), %l4
- sethi %hi(0x4000), %o3
- sub %o5, %o0, %o5
- add %o5, %o3, %l5
- st %l4, [%l5 + 0xc]
- WRITE_PAUSE
- ld [%l5], %g0
- WRITE_PAUSE
- or %l0, PSR_PIL, %l4
- wr %l4, 0x0, %psr
- WRITE_PAUSE
- wr %l4, PSR_ET, %psr
- WRITE_PAUSE
- call sun4m_nmi
- nop
- st %l4, [%l5 + 0x8]
- WRITE_PAUSE
- ld [%l5], %g0
- WRITE_PAUSE
- RESTORE_ALL
.globl smp4d_ticker
/* SMP per-cpu ticker interrupts are handled specially. */
@@ -475,6 +393,36 @@ linux_trap_ipi15_sun4d:
/* FIXME */
1: b,a 1b
+ .globl smpleon_ipi
+ .extern leon_ipi_interrupt
+ /* SMP per-cpu IPI interrupts are handled specially. */
+smpleon_ipi:
+ SAVE_ALL
+ or %l0, PSR_PIL, %g2
+ wr %g2, 0x0, %psr
+ WRITE_PAUSE
+ wr %g2, PSR_ET, %psr
+ WRITE_PAUSE
+ call leonsmp_ipi_interrupt
+ add %sp, STACKFRAME_SZ, %o1 ! pt_regs
+ wr %l0, PSR_ET, %psr
+ WRITE_PAUSE
+ RESTORE_ALL
+
+ .align 4
+ .globl linux_trap_ipi15_leon
+linux_trap_ipi15_leon:
+ SAVE_ALL
+ or %l0, PSR_PIL, %l4
+ wr %l4, 0x0, %psr
+ WRITE_PAUSE
+ wr %l4, PSR_ET, %psr
+ WRITE_PAUSE
+ call leon_cross_call_irq
+ nop
+ b ret_trap_lockless_ipi
+ clr %l6
+
#endif /* CONFIG_SMP */
/* This routine handles illegal instructions and privileged
@@ -781,376 +729,37 @@ setcc_trap_handler:
jmp %l2 ! advance over trap instruction
rett %l2 + 0x4 ! like this...
- .align 4
- .globl linux_trap_nmi_sun4c
-linux_trap_nmi_sun4c:
- SAVE_ALL
-
- /* Ugh, we need to clear the IRQ line. This is now
- * a very sun4c specific trap handler...
- */
- sethi %hi(interrupt_enable), %l5
- ld [%l5 + %lo(interrupt_enable)], %l5
- ldub [%l5], %l6
- andn %l6, INTS_ENAB, %l6
- stb %l6, [%l5]
-
- /* Now it is safe to re-enable traps without recursion. */
- or %l0, PSR_PIL, %l0
- wr %l0, PSR_ET, %psr
+sun4m_nmi_error:
+ /* NMI async memory error handling. */
+ sethi %hi(0x80000000), %l4
+ sethi %hi(sun4m_irq_global), %o5
+ ld [%o5 + %lo(sun4m_irq_global)], %l5
+ st %l4, [%l5 + 0x0c] ! sun4m_irq_global->mask_set=0x80000000
WRITE_PAUSE
-
- /* Now call the c-code with the pt_regs frame ptr and the
- * memory error registers as arguments. The ordering chosen
- * here is due to unlatching semantics.
- */
- sethi %hi(AC_SYNC_ERR), %o0
- add %o0, 0x4, %o0
- lda [%o0] ASI_CONTROL, %o2 ! sync vaddr
- sub %o0, 0x4, %o0
- lda [%o0] ASI_CONTROL, %o1 ! sync error
- add %o0, 0xc, %o0
- lda [%o0] ASI_CONTROL, %o4 ! async vaddr
- sub %o0, 0x4, %o0
- lda [%o0] ASI_CONTROL, %o3 ! async error
- call sparc_lvl15_nmi
- add %sp, STACKFRAME_SZ, %o0
-
- RESTORE_ALL
-
- .align 4
- .globl invalid_segment_patch1_ff
- .globl invalid_segment_patch2_ff
-invalid_segment_patch1_ff: cmp %l4, 0xff
-invalid_segment_patch2_ff: mov 0xff, %l3
-
- .align 4
- .globl invalid_segment_patch1_1ff
- .globl invalid_segment_patch2_1ff
-invalid_segment_patch1_1ff: cmp %l4, 0x1ff
-invalid_segment_patch2_1ff: mov 0x1ff, %l3
-
- .align 4
- .globl num_context_patch1_16, num_context_patch2_16
-num_context_patch1_16: mov 0x10, %l7
-num_context_patch2_16: mov 0x10, %l7
-
- .align 4
- .globl vac_linesize_patch_32
-vac_linesize_patch_32: subcc %l7, 32, %l7
-
- .align 4
- .globl vac_hwflush_patch1_on, vac_hwflush_patch2_on
-
-/*
- * Ugly, but we cant use hardware flushing on the sun4 and we'd require
- * two instructions (Anton)
- */
-#ifdef CONFIG_SUN4
-vac_hwflush_patch1_on: nop
-#else
-vac_hwflush_patch1_on: addcc %l7, -PAGE_SIZE, %l7
-#endif
-
-vac_hwflush_patch2_on: sta %g0, [%l3 + %l7] ASI_HWFLUSHSEG
-
- .globl invalid_segment_patch1, invalid_segment_patch2
- .globl num_context_patch1
- .globl vac_linesize_patch, vac_hwflush_patch1
- .globl vac_hwflush_patch2
-
- .align 4
- .globl sun4c_fault
-
-! %l0 = %psr
-! %l1 = %pc
-! %l2 = %npc
-! %l3 = %wim
-! %l7 = 1 for textfault
-! We want error in %l5, vaddr in %l6
-sun4c_fault:
-#ifdef CONFIG_SUN4
- sethi %hi(sun4c_memerr_reg), %l4
- ld [%l4+%lo(sun4c_memerr_reg)], %l4 ! memerr ctrl reg addr
- ld [%l4], %l6 ! memerr ctrl reg
- ld [%l4 + 4], %l5 ! memerr vaddr reg
- andcc %l6, 0x80, %g0 ! check for error type
- st %g0, [%l4 + 4] ! clear the error
- be 0f ! normal error
- sethi %hi(AC_BUS_ERROR), %l4 ! bus err reg addr
-
- call prom_halt ! something weird happened
- ! what exactly did happen?
- ! what should we do here?
-
-0: or %l4, %lo(AC_BUS_ERROR), %l4 ! bus err reg addr
- lduba [%l4] ASI_CONTROL, %l6 ! bus err reg
-
- cmp %l7, 1 ! text fault?
- be 1f ! yes
- nop
-
- ld [%l1], %l4 ! load instruction that caused fault
- srl %l4, 21, %l4
- andcc %l4, 1, %g0 ! store instruction?
-
- be 1f ! no
- sethi %hi(SUN4C_SYNC_BADWRITE), %l4 ! yep
- ! %lo(SUN4C_SYNC_BADWRITE) = 0
- or %l4, %l6, %l6 ! set write bit to emulate sun4c
-1:
-#else
- sethi %hi(AC_SYNC_ERR), %l4
- add %l4, 0x4, %l6 ! AC_SYNC_VA in %l6
- lda [%l6] ASI_CONTROL, %l5 ! Address
- lda [%l4] ASI_CONTROL, %l6 ! Error, retained for a bit
-#endif
-
- andn %l5, 0xfff, %l5 ! Encode all info into l7
- srl %l6, 14, %l4
-
- and %l4, 2, %l4
- or %l5, %l4, %l4
-
- or %l4, %l7, %l7 ! l7 = [addr,write,txtfault]
-
- andcc %l0, PSR_PS, %g0
- be sun4c_fault_fromuser
- andcc %l7, 1, %g0 ! Text fault?
-
- be 1f
- sethi %hi(KERNBASE), %l4
-
- mov %l1, %l5 ! PC
-
-1:
- cmp %l5, %l4
- blu sun4c_fault_fromuser
- sethi %hi(~((1 << SUN4C_REAL_PGDIR_SHIFT) - 1)), %l4
-
- /* If the kernel references a bum kernel pointer, or a pte which
- * points to a non existant page in ram, we will run this code
- * _forever_ and lock up the machine!!!!! So we must check for
- * this condition, the AC_SYNC_ERR bits are what we must examine.
- * Also a parity error would make this happen as well. So we just
- * check that we are in fact servicing a tlb miss and not some
- * other type of fault for the kernel.
- */
- andcc %l6, 0x80, %g0
- be sun4c_fault_fromuser
- and %l5, %l4, %l5
-
- /* Test for NULL pte_t * in vmalloc area. */
- sethi %hi(VMALLOC_START), %l4
- cmp %l5, %l4
- blu,a invalid_segment_patch1
- lduXa [%l5] ASI_SEGMAP, %l4
-
- sethi %hi(swapper_pg_dir), %l4
- srl %l5, SUN4C_PGDIR_SHIFT, %l6
- or %l4, %lo(swapper_pg_dir), %l4
- sll %l6, 2, %l6
- ld [%l4 + %l6], %l4
-#ifdef CONFIG_SUN4
- sethi %hi(PAGE_MASK), %l6
- andcc %l4, %l6, %g0
-#else
- andcc %l4, PAGE_MASK, %g0
-#endif
- be sun4c_fault_fromuser
- lduXa [%l5] ASI_SEGMAP, %l4
-
-invalid_segment_patch1:
- cmp %l4, 0x7f
- bne 1f
- sethi %hi(sun4c_kfree_ring), %l4
- or %l4, %lo(sun4c_kfree_ring), %l4
- ld [%l4 + 0x18], %l3
- deccc %l3 ! do we have a free entry?
- bcs,a 2f ! no, unmap one.
- sethi %hi(sun4c_kernel_ring), %l4
-
- st %l3, [%l4 + 0x18] ! sun4c_kfree_ring.num_entries--
-
- ld [%l4 + 0x00], %l6 ! entry = sun4c_kfree_ring.ringhd.next
- st %l5, [%l6 + 0x08] ! entry->vaddr = address
-
- ld [%l6 + 0x00], %l3 ! next = entry->next
- ld [%l6 + 0x04], %l7 ! entry->prev
-
- st %l7, [%l3 + 0x04] ! next->prev = entry->prev
- st %l3, [%l7 + 0x00] ! entry->prev->next = next
-
- sethi %hi(sun4c_kernel_ring), %l4
- or %l4, %lo(sun4c_kernel_ring), %l4
- ! head = &sun4c_kernel_ring.ringhd
-
- ld [%l4 + 0x00], %l7 ! head->next
-
- st %l4, [%l6 + 0x04] ! entry->prev = head
- st %l7, [%l6 + 0x00] ! entry->next = head->next
- st %l6, [%l7 + 0x04] ! head->next->prev = entry
-
- st %l6, [%l4 + 0x00] ! head->next = entry
-
- ld [%l4 + 0x18], %l3
- inc %l3 ! sun4c_kernel_ring.num_entries++
- st %l3, [%l4 + 0x18]
- b 4f
- ld [%l6 + 0x08], %l5
-
-2:
- or %l4, %lo(sun4c_kernel_ring), %l4
- ! head = &sun4c_kernel_ring.ringhd
-
- ld [%l4 + 0x04], %l6 ! entry = head->prev
-
- ld [%l6 + 0x08], %l3 ! tmp = entry->vaddr
-
- ! Flush segment from the cache.
-#ifdef CONFIG_SUN4
- sethi %hi((128 * 1024)), %l7
-#else
- sethi %hi((64 * 1024)), %l7
-#endif
-9:
-vac_hwflush_patch1:
-vac_linesize_patch:
- subcc %l7, 16, %l7
- bne 9b
-vac_hwflush_patch2:
- sta %g0, [%l3 + %l7] ASI_FLUSHSEG
-
- st %l5, [%l6 + 0x08] ! entry->vaddr = address
-
- ld [%l6 + 0x00], %l5 ! next = entry->next
- ld [%l6 + 0x04], %l7 ! entry->prev
-
- st %l7, [%l5 + 0x04] ! next->prev = entry->prev
- st %l5, [%l7 + 0x00] ! entry->prev->next = next
- st %l4, [%l6 + 0x04] ! entry->prev = head
-
- ld [%l4 + 0x00], %l7 ! head->next
-
- st %l7, [%l6 + 0x00] ! entry->next = head->next
- st %l6, [%l7 + 0x04] ! head->next->prev = entry
- st %l6, [%l4 + 0x00] ! head->next = entry
-
- mov %l3, %l5 ! address = tmp
-
-4:
-num_context_patch1:
- mov 0x08, %l7
-
- ld [%l6 + 0x08], %l4
- ldub [%l6 + 0x0c], %l3
- or %l4, %l3, %l4 ! encode new vaddr/pseg into l4
-
- sethi %hi(AC_CONTEXT), %l3
- lduba [%l3] ASI_CONTROL, %l6
-
- /* Invalidate old mapping, instantiate new mapping,
- * for each context. Registers l6/l7 are live across
- * this loop.
- */
-3: deccc %l7
- sethi %hi(AC_CONTEXT), %l3
- stba %l7, [%l3] ASI_CONTROL
-invalid_segment_patch2:
- mov 0x7f, %l3
- stXa %l3, [%l5] ASI_SEGMAP
- andn %l4, 0x1ff, %l3
- bne 3b
- stXa %l4, [%l3] ASI_SEGMAP
-
- sethi %hi(AC_CONTEXT), %l3
- stba %l6, [%l3] ASI_CONTROL
-
- andn %l4, 0x1ff, %l5
-
-1:
- sethi %hi(VMALLOC_START), %l4
- cmp %l5, %l4
-
- bgeu 1f
- mov 1 << (SUN4C_REAL_PGDIR_SHIFT - PAGE_SHIFT), %l7
-
- sethi %hi(KERNBASE), %l6
-
- sub %l5, %l6, %l4
- srl %l4, PAGE_SHIFT, %l4
- sethi %hi((SUN4C_PAGE_KERNEL & 0xf4000000)), %l3
- or %l3, %l4, %l3
-
- sethi %hi(PAGE_SIZE), %l4
-
-2:
- sta %l3, [%l5] ASI_PTE
- deccc %l7
- inc %l3
- bne 2b
- add %l5, %l4, %l5
-
- b 7f
- sethi %hi(sun4c_kernel_faults), %l4
-
-1:
- srl %l5, SUN4C_PGDIR_SHIFT, %l3
- sethi %hi(swapper_pg_dir), %l4
- or %l4, %lo(swapper_pg_dir), %l4
- sll %l3, 2, %l3
- ld [%l4 + %l3], %l4
-#ifndef CONFIG_SUN4
- and %l4, PAGE_MASK, %l4
-#else
- sethi %hi(PAGE_MASK), %l6
- and %l4, %l6, %l4
-#endif
-
- srl %l5, (PAGE_SHIFT - 2), %l6
- and %l6, ((SUN4C_PTRS_PER_PTE - 1) << 2), %l6
- add %l6, %l4, %l6
-
- sethi %hi(PAGE_SIZE), %l4
-
-2:
- ld [%l6], %l3
- deccc %l7
- sta %l3, [%l5] ASI_PTE
- add %l6, 0x4, %l6
- bne 2b
- add %l5, %l4, %l5
-
- sethi %hi(sun4c_kernel_faults), %l4
-7:
- ld [%l4 + %lo(sun4c_kernel_faults)], %l3
- inc %l3
- st %l3, [%l4 + %lo(sun4c_kernel_faults)]
-
- /* Restore condition codes */
- wr %l0, 0x0, %psr
+ ld [%l5 + 0x00], %g0 ! sun4m_irq_global->pending
WRITE_PAUSE
- jmp %l1
- rett %l2
-
-sun4c_fault_fromuser:
- SAVE_ALL
+ or %l0, PSR_PIL, %l4
+ wr %l4, 0x0, %psr
+ WRITE_PAUSE
+ wr %l4, PSR_ET, %psr
+ WRITE_PAUSE
+ call sun4m_nmi
nop
-
- mov %l7, %o1 ! Decode the info from %l7
- mov %l7, %o2
- and %o1, 1, %o1 ! arg2 = text_faultp
- mov %l7, %o3
- and %o2, 2, %o2 ! arg3 = writep
- andn %o3, 0xfff, %o3 ! arg4 = faulting address
-
- wr %l0, PSR_ET, %psr
+ st %l4, [%l5 + 0x08] ! sun4m_irq_global->mask_clear=0x80000000
WRITE_PAUSE
+ ld [%l5 + 0x00], %g0 ! sun4m_irq_global->pending
+ WRITE_PAUSE
+ RESTORE_ALL
- call do_sun4c_fault
- add %sp, STACKFRAME_SZ, %o0 ! arg1 = pt_regs ptr
+#ifndef CONFIG_SMP
+ .align 4
+ .globl linux_trap_ipi15_sun4m
+linux_trap_ipi15_sun4m:
+ SAVE_ALL
- RESTORE_ALL
+ ba sun4m_nmi_error
+ nop
+#endif /* CONFIG_SMP */
.align 4
.globl srmmu_fault
@@ -1158,8 +767,11 @@ srmmu_fault:
mov 0x400, %l5
mov 0x300, %l4
- lda [%l5] ASI_M_MMUREGS, %l6 ! read sfar first
- lda [%l4] ASI_M_MMUREGS, %l5 ! read sfsr last
+LEON_PI(lda [%l5] ASI_LEON_MMUREGS, %l6) ! read sfar first
+SUN_PI_(lda [%l5] ASI_M_MMUREGS, %l6) ! read sfar first
+
+LEON_PI(lda [%l4] ASI_LEON_MMUREGS, %l5) ! read sfsr last
+SUN_PI_(lda [%l4] ASI_M_MMUREGS, %l5) ! read sfsr last
andn %l6, 0xfff, %l6
srl %l5, 6, %l5 ! and encode all info into l7
@@ -1186,36 +798,6 @@ srmmu_fault:
RESTORE_ALL
-#ifdef CONFIG_SUNOS_EMUL
- /* SunOS uses syscall zero as the 'indirect syscall' it looks
- * like indir_syscall(scall_num, arg0, arg1, arg2...); etc.
- * This is complete brain damage.
- */
- .globl sunos_indir
-sunos_indir:
- mov %o7, %l4
- cmp %o0, NR_SYSCALLS
- blu,a 1f
- sll %o0, 0x2, %o0
-
- sethi %hi(sunos_nosys), %l6
- b 2f
- or %l6, %lo(sunos_nosys), %l6
-
-1:
- set sunos_sys_table, %l7
- ld [%l7 + %o0], %l6
-
-2:
- mov %o1, %o0
- mov %o2, %o1
- mov %o3, %o2
- mov %o4, %o3
- mov %o5, %o4
- call %l6
- mov %l4, %o7
-#endif
-
.align 4
.globl sys_nis_syscall
sys_nis_syscall:
@@ -1224,31 +806,20 @@ sys_nis_syscall:
call c_sys_nis_syscall
mov %l5, %o7
- .align 4
- .globl sys_execve
-sys_execve:
- mov %o7, %l5
- add %sp, STACKFRAME_SZ, %o0 ! pt_regs *regs arg
- call sparc_execve
- mov %l5, %o7
+sunos_execv:
+ .globl sunos_execv
+ b sys_execve
+ clr %i2
.align 4
- .globl sys_pipe
-sys_pipe:
+ .globl sys_sparc_pipe
+sys_sparc_pipe:
mov %o7, %l5
add %sp, STACKFRAME_SZ, %o0 ! pt_regs *regs arg
call sparc_pipe
mov %l5, %o7
.align 4
- .globl sys_sigaltstack
-sys_sigaltstack:
- mov %o7, %l5
- mov %fp, %o2
- call do_sigaltstack
- mov %l5, %o7
-
- .align 4
.globl sys_sigstack
sys_sigstack:
mov %o7, %l5
@@ -1268,7 +839,7 @@ sys_sigreturn:
nop
call syscall_trace
- nop
+ mov 1, %o1
1:
/* We don't want to muck with user registers like a
@@ -1287,8 +858,9 @@ sys_rt_sigreturn:
be 1f
nop
+ add %sp, STACKFRAME_SZ, %o0
call syscall_trace
- nop
+ mov 1, %o1
1:
/* We are returning to a signal handler. */
@@ -1366,20 +938,16 @@ flush_patch_four:
.align 4
linux_sparc_ni_syscall:
sethi %hi(sys_ni_syscall), %l7
- b syscall_is_too_hard
+ b do_syscall
or %l7, %lo(sys_ni_syscall), %l7
-linux_fast_syscall:
- andn %l7, 3, %l7
- mov %i0, %o0
- mov %i1, %o1
- mov %i2, %o2
- jmpl %l7 + %g0, %g0
- mov %i3, %o3
-
linux_syscall_trace:
+ add %sp, STACKFRAME_SZ, %o0
call syscall_trace
- nop
+ mov 0, %o1
+ cmp %o0, 0
+ bne 3f
+ mov -ENOSYS, %o0
mov %i0, %o0
mov %i1, %o1
mov %i2, %o2
@@ -1390,25 +958,40 @@ linux_syscall_trace:
.globl ret_from_fork
ret_from_fork:
call schedule_tail
- mov %g3, %o0
+ ld [%g3 + TI_TASK], %o0
b ret_sys_call
ld [%sp + STACKFRAME_SZ + PT_I0], %o0
- /* Linux native and SunOS system calls enter here... */
+ .globl ret_from_kernel_thread
+ret_from_kernel_thread:
+ call schedule_tail
+ ld [%g3 + TI_TASK], %o0
+ ld [%sp + STACKFRAME_SZ + PT_G1], %l0
+ call %l0
+ ld [%sp + STACKFRAME_SZ + PT_G2], %o0
+ rd %psr, %l1
+ ld [%sp + STACKFRAME_SZ + PT_PSR], %l0
+ andn %l0, PSR_CWP, %l0
+ nop
+ and %l1, PSR_CWP, %l1
+ or %l0, %l1, %l0
+ st %l0, [%sp + STACKFRAME_SZ + PT_PSR]
+ b ret_sys_call
+ mov 0, %o0
+
+ /* Linux native system calls enter here... */
.align 4
.globl linux_sparc_syscall
linux_sparc_syscall:
+ sethi %hi(PSR_SYSCALL), %l4
+ or %l0, %l4, %l0
/* Direct access to user regs, must faster. */
- cmp %g1, NR_SYSCALLS
+ cmp %g1, NR_syscalls
bgeu linux_sparc_ni_syscall
sll %g1, 2, %l4
ld [%l7 + %l4], %l7
- andcc %l7, 1, %g0
- bne linux_fast_syscall
- /* Just do first insn from SAVE_ALL in the delay slot */
- .globl syscall_is_too_hard
-syscall_is_too_hard:
+do_syscall:
SAVE_ALL_HEAD
rd %wim, %l3
@@ -1427,9 +1010,9 @@ syscall_is_too_hard:
call %l7
mov %i5, %o5
+3:
st %o0, [%sp + STACKFRAME_SZ + PT_I0]
- .globl ret_sys_call
ret_sys_call:
ld [%curptr + TI_FLAGS], %l6
cmp %o0, -ERESTART_RESTARTBLOCK
@@ -1465,6 +1048,8 @@ ret_sys_call:
st %l2, [%sp + STACKFRAME_SZ + PT_NPC]
linux_syscall_trace2:
+ add %sp, STACKFRAME_SZ, %o0
+ mov 1, %o1
call syscall_trace
add %l1, 0x4, %l2 /* npc = npc+4 */
st %l1, [%sp + STACKFRAME_SZ + PT_PC]
@@ -1472,170 +1057,6 @@ linux_syscall_trace2:
st %l2, [%sp + STACKFRAME_SZ + PT_NPC]
- /*
- * Solaris system calls and indirect system calls enter here.
- *
- * I have named the solaris indirect syscalls like that because
- * it seems like Solaris has some fast path syscalls that can
- * be handled as indirect system calls. - mig
- */
-
-linux_syscall_for_solaris:
- sethi %hi(sys_call_table), %l7
- b linux_sparc_syscall
- or %l7, %lo(sys_call_table), %l7
-
- .align 4
- .globl solaris_syscall
-solaris_syscall:
- cmp %g1,59
- be linux_syscall_for_solaris
- cmp %g1,2
- be linux_syscall_for_solaris
- cmp %g1,42
- be linux_syscall_for_solaris
- cmp %g1,119
- be,a linux_syscall_for_solaris
- mov 2, %g1
-1:
- SAVE_ALL_HEAD
- rd %wim, %l3
-
- wr %l0, PSR_ET, %psr
- nop
- nop
- mov %i0, %l5
-
- call do_solaris_syscall
- add %sp, STACKFRAME_SZ, %o0
-
- st %o0, [%sp + STACKFRAME_SZ + PT_I0]
- set PSR_C, %g2
- cmp %o0, -ERESTART_RESTARTBLOCK
- bgeu 1f
- ld [%sp + STACKFRAME_SZ + PT_PSR], %g3
-
- /* System call success, clear Carry condition code. */
- andn %g3, %g2, %g3
- clr %l6
- b 2f
- st %g3, [%sp + STACKFRAME_SZ + PT_PSR]
-
-1:
- /* System call failure, set Carry condition code.
- * Also, get abs(errno) to return to the process.
- */
- sub %g0, %o0, %o0
- mov 1, %l6
- st %o0, [%sp + STACKFRAME_SZ + PT_I0]
- or %g3, %g2, %g3
- st %g3, [%sp + STACKFRAME_SZ + PT_PSR]
-
- /* Advance the pc and npc over the trap instruction.
- * If the npc is unaligned (has a 1 in the lower byte), it means
- * the kernel does not want us to play magic (ie, skipping over
- * traps). Mainly when the Solaris code wants to set some PC and
- * nPC (setcontext).
- */
-2:
- ld [%sp + STACKFRAME_SZ + PT_NPC], %l1 /* pc = npc */
- andcc %l1, 1, %g0
- bne 1f
- add %l1, 0x4, %l2 /* npc = npc+4 */
- st %l1, [%sp + STACKFRAME_SZ + PT_PC]
- b ret_trap_entry
- st %l2, [%sp + STACKFRAME_SZ + PT_NPC]
-
- /* kernel knows what it is doing, fixup npc and continue */
-1:
- sub %l1, 1, %l1
- b ret_trap_entry
- st %l1, [%sp + STACKFRAME_SZ + PT_NPC]
-
-#ifndef CONFIG_SUNOS_EMUL
- .align 4
- .globl sunos_syscall
-sunos_syscall:
- SAVE_ALL_HEAD
- rd %wim, %l3
- wr %l0, PSR_ET, %psr
- nop
- nop
- mov %i0, %l5
- call do_sunos_syscall
- add %sp, STACKFRAME_SZ, %o0
-#endif
-
- /* {net, open}bsd system calls enter here... */
- .align 4
- .globl bsd_syscall
-bsd_syscall:
- /* Direct access to user regs, must faster. */
- cmp %g1, NR_SYSCALLS
- blu,a 1f
- sll %g1, 2, %l4
-
- set sys_ni_syscall, %l7
- b bsd_is_too_hard
- nop
-
-1:
- ld [%l7 + %l4], %l7
-
- .globl bsd_is_too_hard
-bsd_is_too_hard:
- rd %wim, %l3
- SAVE_ALL
-
- wr %l0, PSR_ET, %psr
- WRITE_PAUSE
-
-2:
- mov %i0, %o0
- mov %i1, %o1
- mov %i2, %o2
- mov %i0, %l5
- mov %i3, %o3
- mov %i4, %o4
- call %l7
- mov %i5, %o5
-
- st %o0, [%sp + STACKFRAME_SZ + PT_I0]
- set PSR_C, %g2
- cmp %o0, -ERESTART_RESTARTBLOCK
- bgeu 1f
- ld [%sp + STACKFRAME_SZ + PT_PSR], %g3
-
- /* System call success, clear Carry condition code. */
- andn %g3, %g2, %g3
- clr %l6
- b 2f
- st %g3, [%sp + STACKFRAME_SZ + PT_PSR]
-
-1:
- /* System call failure, set Carry condition code.
- * Also, get abs(errno) to return to the process.
- */
- sub %g0, %o0, %o0
-#if 0 /* XXX todo XXX */
- sethi %hi(bsd_xlatb_rorl), %o3
- or %o3, %lo(bsd_xlatb_rorl), %o3
- sll %o0, 2, %o0
- ld [%o3 + %o0], %o0
-#endif
- mov 1, %l6
- st %o0, [%sp + STACKFRAME_SZ + PT_I0]
- or %g3, %g2, %g3
- st %g3, [%sp + STACKFRAME_SZ + PT_PSR]
-
- /* Advance the pc and npc over the trap instruction. */
-2:
- ld [%sp + STACKFRAME_SZ + PT_NPC], %l1 /* pc = npc */
- add %l1, 0x4, %l2 /* npc = npc+4 */
- st %l1, [%sp + STACKFRAME_SZ + PT_PC]
- b ret_trap_entry
- st %l2, [%sp + STACKFRAME_SZ + PT_NPC]
-
/* Saving and restoring the FPU state is best done from lowlevel code.
*
* void fpsave(unsigned long *fpregs, unsigned long *fsr,
@@ -1730,11 +1151,13 @@ fpload:
.globl __ndelay
__ndelay:
save %sp, -STACKFRAME_SZ, %sp
- mov %i0, %o0
- call .umul ! round multiplier up so large ns ok
- mov 0x1ae, %o1 ! 2**32 / (1 000 000 000 / HZ)
- call .umul
- mov %i1, %o1 ! udelay_val
+ mov %i0, %o0 ! round multiplier up so large ns ok
+ mov 0x1ae, %o1 ! 2**32 / (1 000 000 000 / HZ)
+ umul %o0, %o1, %o0
+ rd %y, %o1
+ mov %i1, %o1 ! udelay_val
+ umul %o0, %o1, %o0
+ rd %y, %o1
ba delay_continue
mov %o1, %o0 ! >>32 later for better resolution
@@ -1743,18 +1166,21 @@ __udelay:
save %sp, -STACKFRAME_SZ, %sp
mov %i0, %o0
sethi %hi(0x10c7), %o1 ! round multiplier up so large us ok
- call .umul
- or %o1, %lo(0x10c7), %o1 ! 2**32 / 1 000 000
- call .umul
- mov %i1, %o1 ! udelay_val
+ or %o1, %lo(0x10c7), %o1 ! 2**32 / 1 000 000
+ umul %o0, %o1, %o0
+ rd %y, %o1
+ mov %i1, %o1 ! udelay_val
+ umul %o0, %o1, %o0
+ rd %y, %o1
sethi %hi(0x028f4b62), %l0 ! Add in rounding constant * 2**32,
or %g0, %lo(0x028f4b62), %l0
addcc %o0, %l0, %o0 ! 2**32 * 0.009 999
bcs,a 3f
add %o1, 0x01, %o1
3:
- call .umul
- mov HZ, %o0 ! >>32 earlier for wider range
+ mov HZ, %o0 ! >>32 earlier for wider range
+ umul %o0, %o1, %o0
+ rd %y, %o1
delay_continue:
cmp %o0, 0x0
@@ -1781,9 +1207,25 @@ breakpoint_trap:
RESTORE_ALL
+#ifdef CONFIG_KGDB
.align 4
- .globl __handle_exception, flush_patch_exception
-__handle_exception:
+ .globl kgdb_trap_low
+ .type kgdb_trap_low,#function
+kgdb_trap_low:
+ rd %wim,%l3
+ SAVE_ALL
+ wr %l0, PSR_ET, %psr
+ WRITE_PAUSE
+
+ call kgdb_trap
+ add %sp, STACKFRAME_SZ, %o0
+
+ RESTORE_ALL
+ .size kgdb_trap_low,.-kgdb_trap_low
+#endif
+
+ .align 4
+ .globl flush_patch_exception
flush_patch_exception:
FLUSH_ALL_KERNEL_WINDOWS;
ldd [%o0], %o6
@@ -1835,7 +1277,7 @@ restore_current:
retl
nop
-#ifdef CONFIG_PCI
+#ifdef CONFIG_PCIC_PCI
#include <asm/pcic.h>
.align 4
@@ -1881,6 +1323,46 @@ pcic_nmi_trap_patch:
rd %psr, %l0
.word 0
-#endif /* CONFIG_PCI */
+#endif /* CONFIG_PCIC_PCI */
+
+ .globl flushw_all
+flushw_all:
+ save %sp, -0x40, %sp
+ save %sp, -0x40, %sp
+ save %sp, -0x40, %sp
+ save %sp, -0x40, %sp
+ save %sp, -0x40, %sp
+ save %sp, -0x40, %sp
+ save %sp, -0x40, %sp
+ restore
+ restore
+ restore
+ restore
+ restore
+ restore
+ ret
+ restore
+
+#ifdef CONFIG_SMP
+ENTRY(hard_smp_processor_id)
+661: rd %tbr, %g1
+ srl %g1, 12, %o0
+ and %o0, 3, %o0
+ .section .cpuid_patch, "ax"
+ /* Instruction location. */
+ .word 661b
+ /* SUN4D implementation. */
+ lda [%g0] ASI_M_VIKING_TMP1, %o0
+ nop
+ nop
+ /* LEON implementation. */
+ rd %asr17, %o0
+ srl %o0, 0x1c, %o0
+ nop
+ .previous
+ retl
+ nop
+ENDPROC(hard_smp_processor_id)
+#endif
/* End of entry.S */
diff --git a/arch/sparc/kernel/entry.h b/arch/sparc/kernel/entry.h
new file mode 100644
index 00000000000..ebaba6167dd
--- /dev/null
+++ b/arch/sparc/kernel/entry.h
@@ -0,0 +1,254 @@
+#ifndef _ENTRY_H
+#define _ENTRY_H
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/init.h>
+
+/* irq */
+void handler_irq(int irq, struct pt_regs *regs);
+
+#ifdef CONFIG_SPARC32
+/* traps */
+void do_hw_interrupt(struct pt_regs *regs, unsigned long type);
+void do_illegal_instruction(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+
+void do_priv_instruction(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void do_fpd_trap(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void do_fpe_trap(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void handle_tag_overflow(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void handle_watchpoint(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void handle_reg_access(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void handle_cp_disabled(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+void handle_cp_exception(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+
+
+
+/* entry.S */
+void fpsave(unsigned long *fpregs, unsigned long *fsr,
+ void *fpqueue, unsigned long *fpqdepth);
+void fpload(unsigned long *fpregs, unsigned long *fsr);
+
+#else /* CONFIG_SPARC32 */
+
+#include <asm/trap_block.h>
+
+struct popc_3insn_patch_entry {
+ unsigned int addr;
+ unsigned int insns[3];
+};
+extern struct popc_3insn_patch_entry __popc_3insn_patch,
+ __popc_3insn_patch_end;
+
+struct popc_6insn_patch_entry {
+ unsigned int addr;
+ unsigned int insns[6];
+};
+extern struct popc_6insn_patch_entry __popc_6insn_patch,
+ __popc_6insn_patch_end;
+
+struct pause_patch_entry {
+ unsigned int addr;
+ unsigned int insns[3];
+};
+extern struct pause_patch_entry __pause_3insn_patch,
+ __pause_3insn_patch_end;
+
+void __init per_cpu_patch(void);
+void sun4v_patch_1insn_range(struct sun4v_1insn_patch_entry *,
+ struct sun4v_1insn_patch_entry *);
+void sun4v_patch_2insn_range(struct sun4v_2insn_patch_entry *,
+ struct sun4v_2insn_patch_entry *);
+void __init sun4v_patch(void);
+void __init boot_cpu_id_too_large(int cpu);
+extern unsigned int dcache_parity_tl1_occurred;
+extern unsigned int icache_parity_tl1_occurred;
+
+asmlinkage void sparc_breakpoint(struct pt_regs *regs);
+void timer_interrupt(int irq, struct pt_regs *regs);
+
+void do_notify_resume(struct pt_regs *regs,
+ unsigned long orig_i0,
+ unsigned long thread_info_flags);
+
+asmlinkage int syscall_trace_enter(struct pt_regs *regs);
+asmlinkage void syscall_trace_leave(struct pt_regs *regs);
+
+void bad_trap_tl1(struct pt_regs *regs, long lvl);
+
+void do_fpieee(struct pt_regs *regs);
+void do_fpother(struct pt_regs *regs);
+void do_tof(struct pt_regs *regs);
+void do_div0(struct pt_regs *regs);
+void do_illegal_instruction(struct pt_regs *regs);
+void mem_address_unaligned(struct pt_regs *regs,
+ unsigned long sfar,
+ unsigned long sfsr);
+void sun4v_do_mna(struct pt_regs *regs,
+ unsigned long addr,
+ unsigned long type_ctx);
+void do_privop(struct pt_regs *regs);
+void do_privact(struct pt_regs *regs);
+void do_cee(struct pt_regs *regs);
+void do_cee_tl1(struct pt_regs *regs);
+void do_dae_tl1(struct pt_regs *regs);
+void do_iae_tl1(struct pt_regs *regs);
+void do_div0_tl1(struct pt_regs *regs);
+void do_fpdis_tl1(struct pt_regs *regs);
+void do_fpieee_tl1(struct pt_regs *regs);
+void do_fpother_tl1(struct pt_regs *regs);
+void do_ill_tl1(struct pt_regs *regs);
+void do_irq_tl1(struct pt_regs *regs);
+void do_lddfmna_tl1(struct pt_regs *regs);
+void do_stdfmna_tl1(struct pt_regs *regs);
+void do_paw(struct pt_regs *regs);
+void do_paw_tl1(struct pt_regs *regs);
+void do_vaw(struct pt_regs *regs);
+void do_vaw_tl1(struct pt_regs *regs);
+void do_tof_tl1(struct pt_regs *regs);
+void do_getpsr(struct pt_regs *regs);
+
+void spitfire_insn_access_exception(struct pt_regs *regs,
+ unsigned long sfsr,
+ unsigned long sfar);
+void spitfire_insn_access_exception_tl1(struct pt_regs *regs,
+ unsigned long sfsr,
+ unsigned long sfar);
+void spitfire_data_access_exception(struct pt_regs *regs,
+ unsigned long sfsr,
+ unsigned long sfar);
+void spitfire_data_access_exception_tl1(struct pt_regs *regs,
+ unsigned long sfsr,
+ unsigned long sfar);
+void spitfire_access_error(struct pt_regs *regs,
+ unsigned long status_encoded,
+ unsigned long afar);
+
+void cheetah_fecc_handler(struct pt_regs *regs,
+ unsigned long afsr,
+ unsigned long afar);
+void cheetah_cee_handler(struct pt_regs *regs,
+ unsigned long afsr,
+ unsigned long afar);
+void cheetah_deferred_handler(struct pt_regs *regs,
+ unsigned long afsr,
+ unsigned long afar);
+void cheetah_plus_parity_error(int type, struct pt_regs *regs);
+
+void sun4v_insn_access_exception(struct pt_regs *regs,
+ unsigned long addr,
+ unsigned long type_ctx);
+void sun4v_insn_access_exception_tl1(struct pt_regs *regs,
+ unsigned long addr,
+ unsigned long type_ctx);
+void sun4v_data_access_exception(struct pt_regs *regs,
+ unsigned long addr,
+ unsigned long type_ctx);
+void sun4v_data_access_exception_tl1(struct pt_regs *regs,
+ unsigned long addr,
+ unsigned long type_ctx);
+void sun4v_resum_error(struct pt_regs *regs,
+ unsigned long offset);
+void sun4v_resum_overflow(struct pt_regs *regs);
+void sun4v_nonresum_error(struct pt_regs *regs,
+ unsigned long offset);
+void sun4v_nonresum_overflow(struct pt_regs *regs);
+
+extern unsigned long sun4v_err_itlb_vaddr;
+extern unsigned long sun4v_err_itlb_ctx;
+extern unsigned long sun4v_err_itlb_pte;
+extern unsigned long sun4v_err_itlb_error;
+
+void sun4v_itlb_error_report(struct pt_regs *regs, int tl);
+
+extern unsigned long sun4v_err_dtlb_vaddr;
+extern unsigned long sun4v_err_dtlb_ctx;
+extern unsigned long sun4v_err_dtlb_pte;
+extern unsigned long sun4v_err_dtlb_error;
+
+void sun4v_dtlb_error_report(struct pt_regs *regs, int tl);
+void hypervisor_tlbop_error(unsigned long err,
+ unsigned long op);
+void hypervisor_tlbop_error_xcall(unsigned long err,
+ unsigned long op);
+
+/* WARNING: The error trap handlers in assembly know the precise
+ * layout of the following structure.
+ *
+ * C-level handlers in traps.c use this information to log the
+ * error and then determine how to recover (if possible).
+ */
+struct cheetah_err_info {
+/*0x00*/u64 afsr;
+/*0x08*/u64 afar;
+
+ /* D-cache state */
+/*0x10*/u64 dcache_data[4]; /* The actual data */
+/*0x30*/u64 dcache_index; /* D-cache index */
+/*0x38*/u64 dcache_tag; /* D-cache tag/valid */
+/*0x40*/u64 dcache_utag; /* D-cache microtag */
+/*0x48*/u64 dcache_stag; /* D-cache snooptag */
+
+ /* I-cache state */
+/*0x50*/u64 icache_data[8]; /* The actual insns + predecode */
+/*0x90*/u64 icache_index; /* I-cache index */
+/*0x98*/u64 icache_tag; /* I-cache phys tag */
+/*0xa0*/u64 icache_utag; /* I-cache microtag */
+/*0xa8*/u64 icache_stag; /* I-cache snooptag */
+/*0xb0*/u64 icache_upper; /* I-cache upper-tag */
+/*0xb8*/u64 icache_lower; /* I-cache lower-tag */
+
+ /* E-cache state */
+/*0xc0*/u64 ecache_data[4]; /* 32 bytes from staging registers */
+/*0xe0*/u64 ecache_index; /* E-cache index */
+/*0xe8*/u64 ecache_tag; /* E-cache tag/state */
+
+/*0xf0*/u64 __pad[32 - 30];
+};
+#define CHAFSR_INVALID ((u64)-1L)
+
+/* This is allocated at boot time based upon the largest hardware
+ * cpu ID in the system. We allocate two entries per cpu, one for
+ * TL==0 logging and one for TL >= 1 logging.
+ */
+extern struct cheetah_err_info *cheetah_error_log;
+
+/* UPA nodes send interrupt packet to UltraSparc with first data reg
+ * value low 5 (7 on Starfire) bits holding the IRQ identifier being
+ * delivered. We must translate this into a non-vector IRQ so we can
+ * set the softint on this cpu.
+ *
+ * To make processing these packets efficient and race free we use
+ * an array of irq buckets below. The interrupt vector handler in
+ * entry.S feeds incoming packets into per-cpu pil-indexed lists.
+ *
+ * If you make changes to ino_bucket, please update hand coded assembler
+ * of the vectored interrupt trap handler(s) in entry.S and sun4v_ivec.S
+ */
+struct ino_bucket {
+/*0x00*/unsigned long __irq_chain_pa;
+
+ /* Interrupt number assigned to this INO. */
+/*0x08*/unsigned int __irq;
+/*0x0c*/unsigned int __pad;
+};
+
+extern struct ino_bucket *ivector_table;
+extern unsigned long ivector_table_pa;
+
+void init_irqwork_curcpu(void);
+void sun4v_register_mondo_queues(int this_cpu);
+
+#endif /* CONFIG_SPARC32 */
+#endif /* _ENTRY_H */
diff --git a/arch/sparc/kernel/errtbls.c b/arch/sparc/kernel/errtbls.c
deleted file mode 100644
index ed14df7116e..00000000000
--- a/arch/sparc/kernel/errtbls.c
+++ /dev/null
@@ -1,144 +0,0 @@
-/* errtbls.c: Error number conversion tables.
- *
- * Copyright (C) 1995, 2007 David S. Miller (davem@davemloft.net)
- *
- * Based upon preliminary work which is:
- *
- * Copyright (C) 1995 Adrian M. Rodriguez (adrian@remus.rutgers.edu)
- */
-
-#include <asm/solerrno.h> /* Solaris errnos */
-
-/* Here is the table which converts between Linux error number values
- * to the equivalent under Solaris. Note that since the Linux ones
- * have been set up to match exactly those of SunOS, no translation
- * table is needed for that OS.
- */
-
-int solaris_errno[] = {
- 0,
- SOL_EPERM,
- SOL_ENOENT,
- SOL_ESRCH,
- SOL_EINTR,
- SOL_EIO,
- SOL_ENXIO,
- SOL_E2BIG,
- SOL_ENOEXEC,
- SOL_EBADF,
- SOL_ECHILD,
- SOL_EAGAIN,
- SOL_ENOMEM,
- SOL_EACCES,
- SOL_EFAULT,
- SOL_NOTBLK,
- SOL_EBUSY,
- SOL_EEXIST,
- SOL_EXDEV,
- SOL_ENODEV,
- SOL_ENOTDIR,
- SOL_EISDIR,
- SOL_EINVAL,
- SOL_ENFILE,
- SOL_EMFILE,
- SOL_ENOTTY,
- SOL_ETXTBSY,
- SOL_EFBIG,
- SOL_ENOSPC,
- SOL_ESPIPE,
- SOL_EROFS,
- SOL_EMLINK,
- SOL_EPIPE,
- SOL_EDOM,
- SOL_ERANGE,
- SOL_EWOULDBLOCK,
- SOL_EINPROGRESS,
- SOL_EALREADY,
- SOL_ENOTSOCK,
- SOL_EDESTADDRREQ,
- SOL_EMSGSIZE,
- SOL_EPROTOTYPE,
- SOL_ENOPROTOOPT,
- SOL_EPROTONOSUPPORT,
- SOL_ESOCKTNOSUPPORT,
- SOL_EOPNOTSUPP,
- SOL_EPFNOSUPPORT,
- SOL_EAFNOSUPPORT,
- SOL_EADDRINUSE,
- SOL_EADDRNOTAVAIL,
- SOL_ENETDOWN,
- SOL_ENETUNREACH,
- SOL_ENETRESET,
- SOL_ECONNABORTED,
- SOL_ECONNRESET,
- SOL_ENOBUFS,
- SOL_EISCONN,
- SOL_ENOTONN,
- SOL_ESHUTDOWN,
- SOL_ETOOMANYREFS,
- SOL_ETIMEDOUT,
- SOL_ECONNREFUSED,
- SOL_ELOOP,
- SOL_ENAMETOOLONG,
- SOL_EHOSTDOWN,
- SOL_EHOSTUNREACH,
- SOL_ENOTEMPTY,
- SOL_EPROCLIM,
- SOL_EUSERS,
- SOL_EDQUOT,
- SOL_ESTALE,
- SOL_EREMOTE,
- SOL_ENOSTR,
- SOL_ETIME,
- SOL_ENOSR,
- SOL_ENOMSG,
- SOL_EBADMSG,
- SOL_IDRM,
- SOL_EDEADLK,
- SOL_ENOLCK,
- SOL_ENONET,
- SOL_ERREMOTE,
- SOL_ENOLINK,
- SOL_EADV,
- SOL_ESRMNT,
- SOL_ECOMM,
- SOL_EPROTO,
- SOL_EMULTIHOP,
- SOL_EINVAL, /* EDOTDOT XXX??? */
- SOL_REMCHG,
- SOL_NOSYS,
- SOL_STRPIPE,
- SOL_EOVERFLOW,
- SOL_EBADFD,
- SOL_ECHRNG,
- SOL_EL2NSYNC,
- SOL_EL3HLT,
- SOL_EL3RST,
- SOL_NRNG,
- SOL_EUNATCH,
- SOL_ENOCSI,
- SOL_EL2HLT,
- SOL_EBADE,
- SOL_EBADR,
- SOL_EXFULL,
- SOL_ENOANO,
- SOL_EBADRQC,
- SOL_EBADSLT,
- SOL_EDEADLOCK,
- SOL_EBFONT,
- SOL_ELIBEXEC,
- SOL_ENODATA,
- SOL_ELIBBAD,
- SOL_ENOPKG,
- SOL_ELIBACC,
- SOL_ENOTUNIQ,
- SOL_ERESTART,
- SOL_EUCLEAN,
- SOL_ENOTNAM,
- SOL_ENAVAIL,
- SOL_EISNAM,
- SOL_EREMOTEIO,
- SOL_EILSEQ,
- SOL_ELIBMAX,
- SOL_ELIBSCN,
-};
diff --git a/arch/sparc/kernel/etrap.S b/arch/sparc/kernel/etrap_32.S
index a8b35bed12a..e3e80d65e39 100644
--- a/arch/sparc/kernel/etrap.S
+++ b/arch/sparc/kernel/etrap_32.S
@@ -1,4 +1,4 @@
-/* $Id: etrap.S,v 1.31 2000/01/08 16:38:18 anton Exp $
+/*
* etrap.S: Sparc trap window preparation for entry into the
* Linux kernel.
*
@@ -216,9 +216,7 @@ tsetup_patch6:
/* Call MMU-architecture dependent stack checking
* routine.
*/
- .globl tsetup_mmu_patchme
-tsetup_mmu_patchme:
- b tsetup_sun4c_stackchk
+ b tsetup_srmmu_stackchk
andcc %sp, 0x7, %g0
/* Architecture specific stack checking routines. When either
@@ -228,53 +226,6 @@ tsetup_mmu_patchme:
*/
#define glob_tmp g1
- .globl tsetup_sun4c_stackchk
-tsetup_sun4c_stackchk:
- /* Done by caller: andcc %sp, 0x7, %g0 */
- bne trap_setup_user_stack_is_bolixed
- sra %sp, 29, %glob_tmp
-
- add %glob_tmp, 0x1, %glob_tmp
- andncc %glob_tmp, 0x1, %g0
- bne trap_setup_user_stack_is_bolixed
- and %sp, 0xfff, %glob_tmp ! delay slot
-
- /* See if our dump area will be on more than one
- * page.
- */
- add %glob_tmp, 0x38, %glob_tmp
- andncc %glob_tmp, 0xff8, %g0
- be tsetup_sun4c_onepage ! only one page to check
- lda [%sp] ASI_PTE, %glob_tmp ! have to check first page anyways
-
-tsetup_sun4c_twopages:
- /* Is first page ok permission wise? */
- srl %glob_tmp, 29, %glob_tmp
- cmp %glob_tmp, 0x6
- bne trap_setup_user_stack_is_bolixed
- add %sp, 0x38, %glob_tmp /* Is second page in vma hole? */
-
- sra %glob_tmp, 29, %glob_tmp
- add %glob_tmp, 0x1, %glob_tmp
- andncc %glob_tmp, 0x1, %g0
- bne trap_setup_user_stack_is_bolixed
- add %sp, 0x38, %glob_tmp
-
- lda [%glob_tmp] ASI_PTE, %glob_tmp
-
-tsetup_sun4c_onepage:
- srl %glob_tmp, 29, %glob_tmp
- cmp %glob_tmp, 0x6 ! can user write to it?
- bne trap_setup_user_stack_is_bolixed ! failure
- nop
-
- STORE_WINDOW(sp)
-
- restore %g0, %g0, %g0
-
- jmpl %t_retpc + 0x8, %g0
- mov %t_kstack, %sp
-
.globl tsetup_srmmu_stackchk
tsetup_srmmu_stackchk:
/* Check results of callers andcc %sp, 0x7, %g0 */
@@ -283,7 +234,8 @@ tsetup_srmmu_stackchk:
cmp %glob_tmp, %sp
bleu,a 1f
- lda [%g0] ASI_M_MMUREGS, %glob_tmp ! read MMU control
+LEON_PI( lda [%g0] ASI_LEON_MMUREGS, %glob_tmp) ! read MMU control
+SUN_PI_( lda [%g0] ASI_M_MMUREGS, %glob_tmp) ! read MMU control
trap_setup_user_stack_is_bolixed:
/* From user/kernel into invalid window w/bad user
@@ -298,18 +250,25 @@ trap_setup_user_stack_is_bolixed:
1:
/* Clear the fault status and turn on the no_fault bit. */
or %glob_tmp, 0x2, %glob_tmp ! or in no_fault bit
- sta %glob_tmp, [%g0] ASI_M_MMUREGS ! set it
+LEON_PI(sta %glob_tmp, [%g0] ASI_LEON_MMUREGS) ! set it
+SUN_PI_(sta %glob_tmp, [%g0] ASI_M_MMUREGS) ! set it
/* Dump the registers and cross fingers. */
STORE_WINDOW(sp)
/* Clear the no_fault bit and check the status. */
andn %glob_tmp, 0x2, %glob_tmp
- sta %glob_tmp, [%g0] ASI_M_MMUREGS
+LEON_PI(sta %glob_tmp, [%g0] ASI_LEON_MMUREGS)
+SUN_PI_(sta %glob_tmp, [%g0] ASI_M_MMUREGS)
+
mov AC_M_SFAR, %glob_tmp
- lda [%glob_tmp] ASI_M_MMUREGS, %g0
+LEON_PI(lda [%glob_tmp] ASI_LEON_MMUREGS, %g0)
+SUN_PI_(lda [%glob_tmp] ASI_M_MMUREGS, %g0)
+
mov AC_M_SFSR, %glob_tmp
- lda [%glob_tmp] ASI_M_MMUREGS, %glob_tmp ! save away status of winstore
+LEON_PI(lda [%glob_tmp] ASI_LEON_MMUREGS, %glob_tmp)! save away status of winstore
+SUN_PI_(lda [%glob_tmp] ASI_M_MMUREGS, %glob_tmp) ! save away status of winstore
+
andcc %glob_tmp, 0x2, %g0 ! did we fault?
bne trap_setup_user_stack_is_bolixed ! failure
nop
diff --git a/arch/sparc/kernel/etrap_64.S b/arch/sparc/kernel/etrap_64.S
new file mode 100644
index 00000000000..1276ca2567b
--- /dev/null
+++ b/arch/sparc/kernel/etrap_64.S
@@ -0,0 +1,240 @@
+/*
+ * etrap.S: Preparing for entry into the kernel on Sparc V9.
+ *
+ * Copyright (C) 1996, 1997 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1997, 1998, 1999 Jakub Jelinek (jj@ultra.linux.cz)
+ */
+
+
+#include <asm/asi.h>
+#include <asm/pstate.h>
+#include <asm/ptrace.h>
+#include <asm/page.h>
+#include <asm/spitfire.h>
+#include <asm/head.h>
+#include <asm/processor.h>
+#include <asm/mmu.h>
+
+#define TASK_REGOFF (THREAD_SIZE-TRACEREG_SZ-STACKFRAME_SZ)
+#define ETRAP_PSTATE1 (PSTATE_TSO | PSTATE_PRIV)
+#define ETRAP_PSTATE2 \
+ (PSTATE_TSO | PSTATE_PEF | PSTATE_PRIV | PSTATE_IE)
+
+/*
+ * On entry, %g7 is return address - 0x4.
+ * %g4 and %g5 will be preserved %l4 and %l5 respectively.
+ */
+
+ .text
+ .align 64
+ .globl etrap_syscall, etrap, etrap_irq, etraptl1
+etrap: rdpr %pil, %g2
+etrap_irq: clr %g3
+etrap_syscall: TRAP_LOAD_THREAD_REG(%g6, %g1)
+ rdpr %tstate, %g1
+ or %g1, %g3, %g1
+ sllx %g2, 20, %g3
+ andcc %g1, TSTATE_PRIV, %g0
+ or %g1, %g3, %g1
+ bne,pn %xcc, 1f
+ sub %sp, STACKFRAME_SZ+TRACEREG_SZ-STACK_BIAS, %g2
+ wrpr %g0, 7, %cleanwin
+
+ sethi %hi(TASK_REGOFF), %g2
+ sethi %hi(TSTATE_PEF), %g3
+ or %g2, %lo(TASK_REGOFF), %g2
+ and %g1, %g3, %g3
+ brnz,pn %g3, 1f
+ add %g6, %g2, %g2
+ wr %g0, 0, %fprs
+1: rdpr %tpc, %g3
+
+ stx %g1, [%g2 + STACKFRAME_SZ + PT_V9_TSTATE]
+ rdpr %tnpc, %g1
+ stx %g3, [%g2 + STACKFRAME_SZ + PT_V9_TPC]
+ rd %y, %g3
+ stx %g1, [%g2 + STACKFRAME_SZ + PT_V9_TNPC]
+ rdpr %tt, %g1
+ st %g3, [%g2 + STACKFRAME_SZ + PT_V9_Y]
+ sethi %hi(PT_REGS_MAGIC), %g3
+ or %g3, %g1, %g1
+ st %g1, [%g2 + STACKFRAME_SZ + PT_V9_MAGIC]
+
+ rdpr %cansave, %g1
+ brnz,pt %g1, etrap_save
+ nop
+
+ rdpr %cwp, %g1
+ add %g1, 2, %g1
+ wrpr %g1, %cwp
+ be,pt %xcc, etrap_user_spill
+ mov ASI_AIUP, %g3
+
+ rdpr %otherwin, %g3
+ brz %g3, etrap_kernel_spill
+ mov ASI_AIUS, %g3
+
+etrap_user_spill:
+
+ wr %g3, 0x0, %asi
+ ldx [%g6 + TI_FLAGS], %g3
+ and %g3, _TIF_32BIT, %g3
+ brnz,pt %g3, etrap_user_spill_32bit
+ nop
+ ba,a,pt %xcc, etrap_user_spill_64bit
+
+etrap_save: save %g2, -STACK_BIAS, %sp
+ mov %g6, %l6
+
+ bne,pn %xcc, 3f
+ mov PRIMARY_CONTEXT, %l4
+ rdpr %canrestore, %g3
+ rdpr %wstate, %g2
+ wrpr %g0, 0, %canrestore
+ sll %g2, 3, %g2
+
+ /* Set TI_SYS_FPDEPTH to 1 and clear TI_SYS_NOERROR. */
+ mov 1, %l5
+ sth %l5, [%l6 + TI_SYS_NOERROR]
+
+ wrpr %g3, 0, %otherwin
+ wrpr %g2, 0, %wstate
+ sethi %hi(sparc64_kern_pri_context), %g2
+ ldx [%g2 + %lo(sparc64_kern_pri_context)], %g3
+
+661: stxa %g3, [%l4] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g3, [%l4] ASI_MMU
+ .previous
+
+ sethi %hi(KERNBASE), %l4
+ flush %l4
+ mov ASI_AIUS, %l7
+2: mov %g4, %l4
+ mov %g5, %l5
+ add %g7, 4, %l2
+
+ /* Go to trap time globals so we can save them. */
+661: wrpr %g0, ETRAP_PSTATE1, %pstate
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ SET_GL(0)
+ .previous
+
+ stx %g1, [%sp + PTREGS_OFF + PT_V9_G1]
+ stx %g2, [%sp + PTREGS_OFF + PT_V9_G2]
+ sllx %l7, 24, %l7
+ stx %g3, [%sp + PTREGS_OFF + PT_V9_G3]
+ rdpr %cwp, %l0
+ stx %g4, [%sp + PTREGS_OFF + PT_V9_G4]
+ stx %g5, [%sp + PTREGS_OFF + PT_V9_G5]
+ stx %g6, [%sp + PTREGS_OFF + PT_V9_G6]
+ stx %g7, [%sp + PTREGS_OFF + PT_V9_G7]
+ or %l7, %l0, %l7
+ sethi %hi(TSTATE_TSO | TSTATE_PEF), %l0
+ or %l7, %l0, %l7
+ wrpr %l2, %tnpc
+ wrpr %l7, (TSTATE_PRIV | TSTATE_IE), %tstate
+ stx %i0, [%sp + PTREGS_OFF + PT_V9_I0]
+ stx %i1, [%sp + PTREGS_OFF + PT_V9_I1]
+ stx %i2, [%sp + PTREGS_OFF + PT_V9_I2]
+ stx %i3, [%sp + PTREGS_OFF + PT_V9_I3]
+ stx %i4, [%sp + PTREGS_OFF + PT_V9_I4]
+ stx %i5, [%sp + PTREGS_OFF + PT_V9_I5]
+ stx %i6, [%sp + PTREGS_OFF + PT_V9_I6]
+ mov %l6, %g6
+ stx %i7, [%sp + PTREGS_OFF + PT_V9_I7]
+ LOAD_PER_CPU_BASE(%g5, %g6, %g4, %g3, %l1)
+ ldx [%g6 + TI_TASK], %g4
+ done
+
+3: mov ASI_P, %l7
+ ldub [%l6 + TI_FPDEPTH], %l5
+ add %l6, TI_FPSAVED + 1, %l4
+ srl %l5, 1, %l3
+ add %l5, 2, %l5
+
+ /* Set TI_SYS_FPDEPTH to %l5 and clear TI_SYS_NOERROR. */
+ sth %l5, [%l6 + TI_SYS_NOERROR]
+ ba,pt %xcc, 2b
+ stb %g0, [%l4 + %l3]
+ nop
+
+etraptl1: /* Save tstate/tpc/tnpc of TL 1-->4 and the tl register itself.
+ * We place this right after pt_regs on the trap stack.
+ * The layout is:
+ * 0x00 TL1's TSTATE
+ * 0x08 TL1's TPC
+ * 0x10 TL1's TNPC
+ * 0x18 TL1's TT
+ * ...
+ * 0x58 TL4's TT
+ * 0x60 TL
+ */
+ TRAP_LOAD_THREAD_REG(%g6, %g1)
+ sub %sp, ((4 * 8) * 4) + 8, %g2
+ rdpr %tl, %g1
+
+ wrpr %g0, 1, %tl
+ rdpr %tstate, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x00]
+ rdpr %tpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x08]
+ rdpr %tnpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x10]
+ rdpr %tt, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x18]
+
+ wrpr %g0, 2, %tl
+ rdpr %tstate, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x20]
+ rdpr %tpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x28]
+ rdpr %tnpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x30]
+ rdpr %tt, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x38]
+
+ sethi %hi(is_sun4v), %g3
+ lduw [%g3 + %lo(is_sun4v)], %g3
+ brnz,pn %g3, finish_tl1_capture
+ nop
+
+ wrpr %g0, 3, %tl
+ rdpr %tstate, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x40]
+ rdpr %tpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x48]
+ rdpr %tnpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x50]
+ rdpr %tt, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x58]
+
+ wrpr %g0, 4, %tl
+ rdpr %tstate, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x60]
+ rdpr %tpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x68]
+ rdpr %tnpc, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x70]
+ rdpr %tt, %g3
+ stx %g3, [%g2 + STACK_BIAS + 0x78]
+
+ stx %g1, [%g2 + STACK_BIAS + 0x80]
+
+finish_tl1_capture:
+ wrpr %g0, 1, %tl
+661: nop
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ SET_GL(1)
+ .previous
+
+ rdpr %tstate, %g1
+ sub %g2, STACKFRAME_SZ + TRACEREG_SZ - STACK_BIAS, %g2
+ ba,pt %xcc, 1b
+ andcc %g1, TSTATE_PRIV, %g0
+
+#undef TASK_REGOFF
+#undef ETRAP_PSTATE1
diff --git a/arch/sparc/kernel/fpu_traps.S b/arch/sparc/kernel/fpu_traps.S
new file mode 100644
index 00000000000..a6864826a4b
--- /dev/null
+++ b/arch/sparc/kernel/fpu_traps.S
@@ -0,0 +1,384 @@
+ /* This is trivial with the new code... */
+ .globl do_fpdis
+ .type do_fpdis,#function
+do_fpdis:
+ sethi %hi(TSTATE_PEF), %g4
+ rdpr %tstate, %g5
+ andcc %g5, %g4, %g0
+ be,pt %xcc, 1f
+ nop
+ rd %fprs, %g5
+ andcc %g5, FPRS_FEF, %g0
+ be,pt %xcc, 1f
+ nop
+
+ /* Legal state when DCR_IFPOE is set in Cheetah %dcr. */
+ sethi %hi(109f), %g7
+ ba,pt %xcc, etrap
+109: or %g7, %lo(109b), %g7
+ add %g0, %g0, %g0
+ ba,a,pt %xcc, rtrap
+
+1: TRAP_LOAD_THREAD_REG(%g6, %g1)
+ ldub [%g6 + TI_FPSAVED], %g5
+ wr %g0, FPRS_FEF, %fprs
+ andcc %g5, FPRS_FEF, %g0
+ be,a,pt %icc, 1f
+ clr %g7
+ ldx [%g6 + TI_GSR], %g7
+1: andcc %g5, FPRS_DL, %g0
+ bne,pn %icc, 2f
+ fzero %f0
+ andcc %g5, FPRS_DU, %g0
+ bne,pn %icc, 1f
+ fzero %f2
+ faddd %f0, %f2, %f4
+ fmuld %f0, %f2, %f6
+ faddd %f0, %f2, %f8
+ fmuld %f0, %f2, %f10
+ faddd %f0, %f2, %f12
+ fmuld %f0, %f2, %f14
+ faddd %f0, %f2, %f16
+ fmuld %f0, %f2, %f18
+ faddd %f0, %f2, %f20
+ fmuld %f0, %f2, %f22
+ faddd %f0, %f2, %f24
+ fmuld %f0, %f2, %f26
+ faddd %f0, %f2, %f28
+ fmuld %f0, %f2, %f30
+ faddd %f0, %f2, %f32
+ fmuld %f0, %f2, %f34
+ faddd %f0, %f2, %f36
+ fmuld %f0, %f2, %f38
+ faddd %f0, %f2, %f40
+ fmuld %f0, %f2, %f42
+ faddd %f0, %f2, %f44
+ fmuld %f0, %f2, %f46
+ faddd %f0, %f2, %f48
+ fmuld %f0, %f2, %f50
+ faddd %f0, %f2, %f52
+ fmuld %f0, %f2, %f54
+ faddd %f0, %f2, %f56
+ fmuld %f0, %f2, %f58
+ b,pt %xcc, fpdis_exit2
+ faddd %f0, %f2, %f60
+1: mov SECONDARY_CONTEXT, %g3
+ add %g6, TI_FPREGS + 0x80, %g1
+ faddd %f0, %f2, %f4
+ fmuld %f0, %f2, %f6
+
+661: ldxa [%g3] ASI_DMMU, %g5
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ ldxa [%g3] ASI_MMU, %g5
+ .previous
+
+ sethi %hi(sparc64_kern_sec_context), %g2
+ ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2
+
+661: stxa %g2, [%g3] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g2, [%g3] ASI_MMU
+ .previous
+
+ membar #Sync
+ add %g6, TI_FPREGS + 0xc0, %g2
+ faddd %f0, %f2, %f8
+ fmuld %f0, %f2, %f10
+ membar #Sync
+ ldda [%g1] ASI_BLK_S, %f32
+ ldda [%g2] ASI_BLK_S, %f48
+ membar #Sync
+ faddd %f0, %f2, %f12
+ fmuld %f0, %f2, %f14
+ faddd %f0, %f2, %f16
+ fmuld %f0, %f2, %f18
+ faddd %f0, %f2, %f20
+ fmuld %f0, %f2, %f22
+ faddd %f0, %f2, %f24
+ fmuld %f0, %f2, %f26
+ faddd %f0, %f2, %f28
+ fmuld %f0, %f2, %f30
+ b,pt %xcc, fpdis_exit
+ nop
+2: andcc %g5, FPRS_DU, %g0
+ bne,pt %icc, 3f
+ fzero %f32
+ mov SECONDARY_CONTEXT, %g3
+ fzero %f34
+
+661: ldxa [%g3] ASI_DMMU, %g5
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ ldxa [%g3] ASI_MMU, %g5
+ .previous
+
+ add %g6, TI_FPREGS, %g1
+ sethi %hi(sparc64_kern_sec_context), %g2
+ ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2
+
+661: stxa %g2, [%g3] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g2, [%g3] ASI_MMU
+ .previous
+
+ membar #Sync
+ add %g6, TI_FPREGS + 0x40, %g2
+ faddd %f32, %f34, %f36
+ fmuld %f32, %f34, %f38
+ membar #Sync
+ ldda [%g1] ASI_BLK_S, %f0
+ ldda [%g2] ASI_BLK_S, %f16
+ membar #Sync
+ faddd %f32, %f34, %f40
+ fmuld %f32, %f34, %f42
+ faddd %f32, %f34, %f44
+ fmuld %f32, %f34, %f46
+ faddd %f32, %f34, %f48
+ fmuld %f32, %f34, %f50
+ faddd %f32, %f34, %f52
+ fmuld %f32, %f34, %f54
+ faddd %f32, %f34, %f56
+ fmuld %f32, %f34, %f58
+ faddd %f32, %f34, %f60
+ fmuld %f32, %f34, %f62
+ ba,pt %xcc, fpdis_exit
+ nop
+3: mov SECONDARY_CONTEXT, %g3
+ add %g6, TI_FPREGS, %g1
+
+661: ldxa [%g3] ASI_DMMU, %g5
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ ldxa [%g3] ASI_MMU, %g5
+ .previous
+
+ sethi %hi(sparc64_kern_sec_context), %g2
+ ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2
+
+661: stxa %g2, [%g3] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g2, [%g3] ASI_MMU
+ .previous
+
+ membar #Sync
+ mov 0x40, %g2
+ membar #Sync
+ ldda [%g1] ASI_BLK_S, %f0
+ ldda [%g1 + %g2] ASI_BLK_S, %f16
+ add %g1, 0x80, %g1
+ ldda [%g1] ASI_BLK_S, %f32
+ ldda [%g1 + %g2] ASI_BLK_S, %f48
+ membar #Sync
+fpdis_exit:
+
+661: stxa %g5, [%g3] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g5, [%g3] ASI_MMU
+ .previous
+
+ membar #Sync
+fpdis_exit2:
+ wr %g7, 0, %gsr
+ ldx [%g6 + TI_XFSR], %fsr
+ rdpr %tstate, %g3
+ or %g3, %g4, %g3 ! anal...
+ wrpr %g3, %tstate
+ wr %g0, FPRS_FEF, %fprs ! clean DU/DL bits
+ retry
+ .size do_fpdis,.-do_fpdis
+
+ .align 32
+ .type fp_other_bounce,#function
+fp_other_bounce:
+ call do_fpother
+ add %sp, PTREGS_OFF, %o0
+ ba,pt %xcc, rtrap
+ nop
+ .size fp_other_bounce,.-fp_other_bounce
+
+ .align 32
+ .globl do_fpother_check_fitos
+ .type do_fpother_check_fitos,#function
+do_fpother_check_fitos:
+ TRAP_LOAD_THREAD_REG(%g6, %g1)
+ sethi %hi(fp_other_bounce - 4), %g7
+ or %g7, %lo(fp_other_bounce - 4), %g7
+
+ /* NOTE: Need to preserve %g7 until we fully commit
+ * to the fitos fixup.
+ */
+ stx %fsr, [%g6 + TI_XFSR]
+ rdpr %tstate, %g3
+ andcc %g3, TSTATE_PRIV, %g0
+ bne,pn %xcc, do_fptrap_after_fsr
+ nop
+ ldx [%g6 + TI_XFSR], %g3
+ srlx %g3, 14, %g1
+ and %g1, 7, %g1
+ cmp %g1, 2 ! Unfinished FP-OP
+ bne,pn %xcc, do_fptrap_after_fsr
+ sethi %hi(1 << 23), %g1 ! Inexact
+ andcc %g3, %g1, %g0
+ bne,pn %xcc, do_fptrap_after_fsr
+ rdpr %tpc, %g1
+ lduwa [%g1] ASI_AIUP, %g3 ! This cannot ever fail
+#define FITOS_MASK 0xc1f83fe0
+#define FITOS_COMPARE 0x81a01880
+ sethi %hi(FITOS_MASK), %g1
+ or %g1, %lo(FITOS_MASK), %g1
+ and %g3, %g1, %g1
+ sethi %hi(FITOS_COMPARE), %g2
+ or %g2, %lo(FITOS_COMPARE), %g2
+ cmp %g1, %g2
+ bne,pn %xcc, do_fptrap_after_fsr
+ nop
+ std %f62, [%g6 + TI_FPREGS + (62 * 4)]
+ sethi %hi(fitos_table_1), %g1
+ and %g3, 0x1f, %g2
+ or %g1, %lo(fitos_table_1), %g1
+ sllx %g2, 2, %g2
+ jmpl %g1 + %g2, %g0
+ ba,pt %xcc, fitos_emul_continue
+
+fitos_table_1:
+ fitod %f0, %f62
+ fitod %f1, %f62
+ fitod %f2, %f62
+ fitod %f3, %f62
+ fitod %f4, %f62
+ fitod %f5, %f62
+ fitod %f6, %f62
+ fitod %f7, %f62
+ fitod %f8, %f62
+ fitod %f9, %f62
+ fitod %f10, %f62
+ fitod %f11, %f62
+ fitod %f12, %f62
+ fitod %f13, %f62
+ fitod %f14, %f62
+ fitod %f15, %f62
+ fitod %f16, %f62
+ fitod %f17, %f62
+ fitod %f18, %f62
+ fitod %f19, %f62
+ fitod %f20, %f62
+ fitod %f21, %f62
+ fitod %f22, %f62
+ fitod %f23, %f62
+ fitod %f24, %f62
+ fitod %f25, %f62
+ fitod %f26, %f62
+ fitod %f27, %f62
+ fitod %f28, %f62
+ fitod %f29, %f62
+ fitod %f30, %f62
+ fitod %f31, %f62
+
+fitos_emul_continue:
+ sethi %hi(fitos_table_2), %g1
+ srl %g3, 25, %g2
+ or %g1, %lo(fitos_table_2), %g1
+ and %g2, 0x1f, %g2
+ sllx %g2, 2, %g2
+ jmpl %g1 + %g2, %g0
+ ba,pt %xcc, fitos_emul_fini
+
+fitos_table_2:
+ fdtos %f62, %f0
+ fdtos %f62, %f1
+ fdtos %f62, %f2
+ fdtos %f62, %f3
+ fdtos %f62, %f4
+ fdtos %f62, %f5
+ fdtos %f62, %f6
+ fdtos %f62, %f7
+ fdtos %f62, %f8
+ fdtos %f62, %f9
+ fdtos %f62, %f10
+ fdtos %f62, %f11
+ fdtos %f62, %f12
+ fdtos %f62, %f13
+ fdtos %f62, %f14
+ fdtos %f62, %f15
+ fdtos %f62, %f16
+ fdtos %f62, %f17
+ fdtos %f62, %f18
+ fdtos %f62, %f19
+ fdtos %f62, %f20
+ fdtos %f62, %f21
+ fdtos %f62, %f22
+ fdtos %f62, %f23
+ fdtos %f62, %f24
+ fdtos %f62, %f25
+ fdtos %f62, %f26
+ fdtos %f62, %f27
+ fdtos %f62, %f28
+ fdtos %f62, %f29
+ fdtos %f62, %f30
+ fdtos %f62, %f31
+
+fitos_emul_fini:
+ ldd [%g6 + TI_FPREGS + (62 * 4)], %f62
+ done
+ .size do_fpother_check_fitos,.-do_fpother_check_fitos
+
+ .align 32
+ .globl do_fptrap
+ .type do_fptrap,#function
+do_fptrap:
+ TRAP_LOAD_THREAD_REG(%g6, %g1)
+ stx %fsr, [%g6 + TI_XFSR]
+do_fptrap_after_fsr:
+ ldub [%g6 + TI_FPSAVED], %g3
+ rd %fprs, %g1
+ or %g3, %g1, %g3
+ stb %g3, [%g6 + TI_FPSAVED]
+ rd %gsr, %g3
+ stx %g3, [%g6 + TI_GSR]
+ mov SECONDARY_CONTEXT, %g3
+
+661: ldxa [%g3] ASI_DMMU, %g5
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ ldxa [%g3] ASI_MMU, %g5
+ .previous
+
+ sethi %hi(sparc64_kern_sec_context), %g2
+ ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2
+
+661: stxa %g2, [%g3] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g2, [%g3] ASI_MMU
+ .previous
+
+ membar #Sync
+ add %g6, TI_FPREGS, %g2
+ andcc %g1, FPRS_DL, %g0
+ be,pn %icc, 4f
+ mov 0x40, %g3
+ stda %f0, [%g2] ASI_BLK_S
+ stda %f16, [%g2 + %g3] ASI_BLK_S
+ andcc %g1, FPRS_DU, %g0
+ be,pn %icc, 5f
+4: add %g2, 128, %g2
+ stda %f32, [%g2] ASI_BLK_S
+ stda %f48, [%g2 + %g3] ASI_BLK_S
+5: mov SECONDARY_CONTEXT, %g1
+ membar #Sync
+
+661: stxa %g5, [%g1] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g5, [%g1] ASI_MMU
+ .previous
+
+ membar #Sync
+ ba,pt %xcc, etrap
+ wr %g0, 0, %fprs
+ .size do_fptrap,.-do_fptrap
diff --git a/arch/sparc/kernel/ftrace.c b/arch/sparc/kernel/ftrace.c
new file mode 100644
index 00000000000..0a2d2ddff54
--- /dev/null
+++ b/arch/sparc/kernel/ftrace.c
@@ -0,0 +1,147 @@
+#include <linux/spinlock.h>
+#include <linux/hardirq.h>
+#include <linux/ftrace.h>
+#include <linux/percpu.h>
+#include <linux/init.h>
+#include <linux/list.h>
+#include <trace/syscall.h>
+
+#include <asm/ftrace.h>
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+static const u32 ftrace_nop = 0x01000000;
+
+static u32 ftrace_call_replace(unsigned long ip, unsigned long addr)
+{
+ u32 call;
+ s32 off;
+
+ off = ((s32)addr - (s32)ip);
+ call = 0x40000000 | ((u32)off >> 2);
+
+ return call;
+}
+
+static int ftrace_modify_code(unsigned long ip, u32 old, u32 new)
+{
+ u32 replaced;
+ int faulted;
+
+ __asm__ __volatile__(
+ "1: cas [%[ip]], %[old], %[new]\n"
+ " flush %[ip]\n"
+ " mov 0, %[faulted]\n"
+ "2:\n"
+ " .section .fixup,#alloc,#execinstr\n"
+ " .align 4\n"
+ "3: sethi %%hi(2b), %[faulted]\n"
+ " jmpl %[faulted] + %%lo(2b), %%g0\n"
+ " mov 1, %[faulted]\n"
+ " .previous\n"
+ " .section __ex_table,\"a\"\n"
+ " .align 4\n"
+ " .word 1b, 3b\n"
+ " .previous\n"
+ : "=r" (replaced), [faulted] "=r" (faulted)
+ : [new] "0" (new), [old] "r" (old), [ip] "r" (ip)
+ : "memory");
+
+ if (replaced != old && replaced != new)
+ faulted = 2;
+
+ return faulted;
+}
+
+int ftrace_make_nop(struct module *mod, struct dyn_ftrace *rec, unsigned long addr)
+{
+ unsigned long ip = rec->ip;
+ u32 old, new;
+
+ old = ftrace_call_replace(ip, addr);
+ new = ftrace_nop;
+ return ftrace_modify_code(ip, old, new);
+}
+
+int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
+{
+ unsigned long ip = rec->ip;
+ u32 old, new;
+
+ old = ftrace_nop;
+ new = ftrace_call_replace(ip, addr);
+ return ftrace_modify_code(ip, old, new);
+}
+
+int ftrace_update_ftrace_func(ftrace_func_t func)
+{
+ unsigned long ip = (unsigned long)(&ftrace_call);
+ u32 old, new;
+
+ old = *(u32 *) &ftrace_call;
+ new = ftrace_call_replace(ip, (unsigned long)func);
+ return ftrace_modify_code(ip, old, new);
+}
+
+int __init ftrace_dyn_arch_init(void)
+{
+ return 0;
+}
+#endif
+
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+extern void ftrace_graph_call(void);
+
+int ftrace_enable_ftrace_graph_caller(void)
+{
+ unsigned long ip = (unsigned long)(&ftrace_graph_call);
+ u32 old, new;
+
+ old = *(u32 *) &ftrace_graph_call;
+ new = ftrace_call_replace(ip, (unsigned long) &ftrace_graph_caller);
+ return ftrace_modify_code(ip, old, new);
+}
+
+int ftrace_disable_ftrace_graph_caller(void)
+{
+ unsigned long ip = (unsigned long)(&ftrace_graph_call);
+ u32 old, new;
+
+ old = *(u32 *) &ftrace_graph_call;
+ new = ftrace_call_replace(ip, (unsigned long) &ftrace_stub);
+
+ return ftrace_modify_code(ip, old, new);
+}
+
+#endif /* !CONFIG_DYNAMIC_FTRACE */
+
+/*
+ * Hook the return address and push it in the stack of return addrs
+ * in current thread info.
+ */
+unsigned long prepare_ftrace_return(unsigned long parent,
+ unsigned long self_addr,
+ unsigned long frame_pointer)
+{
+ unsigned long return_hooker = (unsigned long) &return_to_handler;
+ struct ftrace_graph_ent trace;
+
+ if (unlikely(atomic_read(&current->tracing_graph_pause)))
+ return parent + 8UL;
+
+ if (ftrace_push_return_trace(parent, self_addr, &trace.depth,
+ frame_pointer) == -EBUSY)
+ return parent + 8UL;
+
+ trace.func = self_addr;
+
+ /* Only trace if the calling function expects to */
+ if (!ftrace_graph_entry(&trace)) {
+ current->curr_ret_stack--;
+ return parent + 8UL;
+ }
+
+ return return_hooker;
+}
+#endif /* CONFIG_FUNCTION_GRAPH_TRACER */
diff --git a/arch/sparc/kernel/getsetcc.S b/arch/sparc/kernel/getsetcc.S
new file mode 100644
index 00000000000..a14d272d206
--- /dev/null
+++ b/arch/sparc/kernel/getsetcc.S
@@ -0,0 +1,24 @@
+ .globl getcc
+ .type getcc,#function
+getcc:
+ ldx [%o0 + PT_V9_TSTATE], %o1
+ srlx %o1, 32, %o1
+ and %o1, 0xf, %o1
+ retl
+ stx %o1, [%o0 + PT_V9_G1]
+ .size getcc,.-getcc
+
+ .globl setcc
+ .type setcc,#function
+setcc:
+ ldx [%o0 + PT_V9_TSTATE], %o1
+ ldx [%o0 + PT_V9_G1], %o2
+ or %g0, %ulo(TSTATE_ICC), %o3
+ sllx %o3, 32, %o3
+ andn %o1, %o3, %o1
+ sllx %o2, 32, %o2
+ and %o2, %o3, %o2
+ or %o1, %o2, %o1
+ retl
+ stx %o1, [%o0 + PT_V9_TSTATE]
+ .size setcc,.-setcc
diff --git a/arch/sparc/kernel/head.S b/arch/sparc/kernel/head.S
deleted file mode 100644
index 9a219e8b5dd..00000000000
--- a/arch/sparc/kernel/head.S
+++ /dev/null
@@ -1,1325 +0,0 @@
-/* $Id: head.S,v 1.105 2001/08/12 09:08:56 davem Exp $
- * head.S: The initial boot code for the Sparc port of Linux.
- *
- * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
- * Copyright (C) 1995,1999 Pete Zaitcev (zaitcev@yahoo.com)
- * Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx)
- * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
- * Copyright (C) 1997 Michael A. Griffith (grif@acm.org)
- *
- * CompactPCI platform by Eric Brower, 1999.
- */
-
-#include <linux/version.h>
-#include <linux/init.h>
-
-#include <asm/head.h>
-#include <asm/asi.h>
-#include <asm/contregs.h>
-#include <asm/ptrace.h>
-#include <asm/psr.h>
-#include <asm/page.h>
-#include <asm/kdebug.h>
-#include <asm/winmacro.h>
-#include <asm/thread_info.h> /* TI_UWINMASK */
-#include <asm/errno.h>
-#include <asm/pgtsrmmu.h> /* SRMMU_PGDIR_SHIFT */
-
- .data
-/*
- * The following are used with the prom_vector node-ops to figure out
- * the cpu-type
- */
-
- .align 4
- .globl cputyp
-cputyp:
- .word 1
-
- .align 4
- .globl cputypval
-cputypval:
- .asciz "sun4c"
- .ascii " "
-
-cputypvalend:
-cputypvallen = cputypvar - cputypval
-
- .align 4
-/*
- * Sun people can't spell worth damn. "compatability" indeed.
- * At least we *know* we can't spell, and use a spell-checker.
- */
-
-/* Uh, actually Linus it is I who cannot spell. Too much murky
- * Sparc assembly will do this to ya.
- */
-cputypvar:
- .asciz "compatability"
-
-/* Tested on SS-5, SS-10. Probably someone at Sun applied a spell-checker. */
- .align 4
-cputypvar_sun4m:
- .asciz "compatible"
-
- .align 4
-
-#ifndef CONFIG_SUN4
-sun4_notsup:
- .asciz "Sparc-Linux sun4 needs a specially compiled kernel, turn CONFIG_SUN4 on.\n\n"
- .align 4
-#else
-sun4cdm_notsup:
- .asciz "Kernel compiled with CONFIG_SUN4 cannot run on SUN4C/SUN4M/SUN4D\nTurn CONFIG_SUN4 off.\n\n"
- .align 4
-#endif
-
-sun4e_notsup:
- .asciz "Sparc-Linux sun4e support does not exist\n\n"
- .align 4
-
-#ifndef CONFIG_SUNOS_EMUL
-#undef SUNOS_SYSCALL_TRAP
-#define SUNOS_SYSCALL_TRAP SUNOS_NO_SYSCALL_TRAP
-#endif
-
- /* The Sparc trap table, bootloader gives us control at _start. */
- .text
- .globl start, _stext, _start, __stext
- .globl trapbase
-_start: /* danger danger */
-__stext:
-_stext:
-start:
-trapbase:
-#ifdef CONFIG_SMP
-trapbase_cpu0:
-#endif
-/* We get control passed to us here at t_zero. */
-t_zero: b gokernel; nop; nop; nop;
-t_tflt: SPARC_TFAULT /* Inst. Access Exception */
-t_bins: TRAP_ENTRY(0x2, bad_instruction) /* Illegal Instruction */
-t_pins: TRAP_ENTRY(0x3, priv_instruction) /* Privileged Instruction */
-t_fpd: TRAP_ENTRY(0x4, fpd_trap_handler) /* Floating Point Disabled */
-t_wovf: WINDOW_SPILL /* Window Overflow */
-t_wunf: WINDOW_FILL /* Window Underflow */
-t_mna: TRAP_ENTRY(0x7, mna_handler) /* Memory Address Not Aligned */
-t_fpe: TRAP_ENTRY(0x8, fpe_trap_handler) /* Floating Point Exception */
-t_dflt: SPARC_DFAULT /* Data Miss Exception */
-t_tio: TRAP_ENTRY(0xa, do_tag_overflow) /* Tagged Instruction Ovrflw */
-t_wpt: TRAP_ENTRY(0xb, do_watchpoint) /* Watchpoint Detected */
-t_badc: BAD_TRAP(0xc) BAD_TRAP(0xd) BAD_TRAP(0xe) BAD_TRAP(0xf) BAD_TRAP(0x10)
-t_irq1: TRAP_ENTRY_INTERRUPT(1) /* IRQ Software/SBUS Level 1 */
-t_irq2: TRAP_ENTRY_INTERRUPT(2) /* IRQ SBUS Level 2 */
-t_irq3: TRAP_ENTRY_INTERRUPT(3) /* IRQ SCSI/DMA/SBUS Level 3 */
-t_irq4: TRAP_ENTRY_INTERRUPT(4) /* IRQ Software Level 4 */
-t_irq5: TRAP_ENTRY_INTERRUPT(5) /* IRQ SBUS/Ethernet Level 5 */
-t_irq6: TRAP_ENTRY_INTERRUPT(6) /* IRQ Software Level 6 */
-t_irq7: TRAP_ENTRY_INTERRUPT(7) /* IRQ Video/SBUS Level 5 */
-t_irq8: TRAP_ENTRY_INTERRUPT(8) /* IRQ SBUS Level 6 */
-t_irq9: TRAP_ENTRY_INTERRUPT(9) /* IRQ SBUS Level 7 */
-t_irq10:TRAP_ENTRY_INTERRUPT(10) /* IRQ Timer #1 (one we use) */
-t_irq11:TRAP_ENTRY_INTERRUPT(11) /* IRQ Floppy Intr. */
-t_irq12:TRAP_ENTRY_INTERRUPT(12) /* IRQ Zilog serial chip */
-t_irq13:TRAP_ENTRY_INTERRUPT(13) /* IRQ Audio Intr. */
-t_irq14:TRAP_ENTRY_INTERRUPT(14) /* IRQ Timer #2 */
- .globl t_nmi
-#ifndef CONFIG_SMP
-t_nmi: NMI_TRAP /* Level 15 (NMI) */
-#else
-t_nmi: TRAP_ENTRY(0x1f, linux_trap_ipi15_sun4m)
-#endif
-t_racc: TRAP_ENTRY(0x20, do_reg_access) /* General Register Access Error */
-t_iacce:BAD_TRAP(0x21) /* Instr Access Error */
-t_bad22:BAD_TRAP(0x22) BAD_TRAP(0x23)
-t_cpdis:TRAP_ENTRY(0x24, do_cp_disabled) /* Co-Processor Disabled */
-t_uflsh:SKIP_TRAP(0x25, unimp_flush) /* Unimplemented FLUSH inst. */
-t_bad26:BAD_TRAP(0x26) BAD_TRAP(0x27)
-t_cpexc:TRAP_ENTRY(0x28, do_cp_exception) /* Co-Processor Exception */
-t_dacce:SPARC_DFAULT /* Data Access Error */
-t_hwdz: TRAP_ENTRY(0x2a, do_hw_divzero) /* Division by zero, you lose... */
-t_dserr:BAD_TRAP(0x2b) /* Data Store Error */
-t_daccm:BAD_TRAP(0x2c) /* Data Access MMU-Miss */
-t_bad2d:BAD_TRAP(0x2d) BAD_TRAP(0x2e) BAD_TRAP(0x2f) BAD_TRAP(0x30) BAD_TRAP(0x31)
-t_bad32:BAD_TRAP(0x32) BAD_TRAP(0x33) BAD_TRAP(0x34) BAD_TRAP(0x35) BAD_TRAP(0x36)
-t_bad37:BAD_TRAP(0x37) BAD_TRAP(0x38) BAD_TRAP(0x39) BAD_TRAP(0x3a) BAD_TRAP(0x3b)
-t_iaccm:BAD_TRAP(0x3c) /* Instr Access MMU-Miss */
-t_bad3d:BAD_TRAP(0x3d) BAD_TRAP(0x3e) BAD_TRAP(0x3f) BAD_TRAP(0x40) BAD_TRAP(0x41)
-t_bad42:BAD_TRAP(0x42) BAD_TRAP(0x43) BAD_TRAP(0x44) BAD_TRAP(0x45) BAD_TRAP(0x46)
-t_bad47:BAD_TRAP(0x47) BAD_TRAP(0x48) BAD_TRAP(0x49) BAD_TRAP(0x4a) BAD_TRAP(0x4b)
-t_bad4c:BAD_TRAP(0x4c) BAD_TRAP(0x4d) BAD_TRAP(0x4e) BAD_TRAP(0x4f) BAD_TRAP(0x50)
-t_bad51:BAD_TRAP(0x51) BAD_TRAP(0x52) BAD_TRAP(0x53) BAD_TRAP(0x54) BAD_TRAP(0x55)
-t_bad56:BAD_TRAP(0x56) BAD_TRAP(0x57) BAD_TRAP(0x58) BAD_TRAP(0x59) BAD_TRAP(0x5a)
-t_bad5b:BAD_TRAP(0x5b) BAD_TRAP(0x5c) BAD_TRAP(0x5d) BAD_TRAP(0x5e) BAD_TRAP(0x5f)
-t_bad60:BAD_TRAP(0x60) BAD_TRAP(0x61) BAD_TRAP(0x62) BAD_TRAP(0x63) BAD_TRAP(0x64)
-t_bad65:BAD_TRAP(0x65) BAD_TRAP(0x66) BAD_TRAP(0x67) BAD_TRAP(0x68) BAD_TRAP(0x69)
-t_bad6a:BAD_TRAP(0x6a) BAD_TRAP(0x6b) BAD_TRAP(0x6c) BAD_TRAP(0x6d) BAD_TRAP(0x6e)
-t_bad6f:BAD_TRAP(0x6f) BAD_TRAP(0x70) BAD_TRAP(0x71) BAD_TRAP(0x72) BAD_TRAP(0x73)
-t_bad74:BAD_TRAP(0x74) BAD_TRAP(0x75) BAD_TRAP(0x76) BAD_TRAP(0x77) BAD_TRAP(0x78)
-t_bad79:BAD_TRAP(0x79) BAD_TRAP(0x7a) BAD_TRAP(0x7b) BAD_TRAP(0x7c) BAD_TRAP(0x7d)
-t_bad7e:BAD_TRAP(0x7e) BAD_TRAP(0x7f)
-t_sunos:SUNOS_SYSCALL_TRAP /* SunOS System Call */
-t_sbkpt:BREAKPOINT_TRAP /* Software Breakpoint/KGDB */
-t_divz: TRAP_ENTRY(0x82, do_hw_divzero) /* Divide by zero trap */
-t_flwin:TRAP_ENTRY(0x83, do_flush_windows) /* Flush Windows Trap */
-t_clwin:BAD_TRAP(0x84) /* Clean Windows Trap */
-t_rchk: BAD_TRAP(0x85) /* Range Check */
-t_funal:BAD_TRAP(0x86) /* Fix Unaligned Access Trap */
-t_iovf: BAD_TRAP(0x87) /* Integer Overflow Trap */
-t_slowl:SOLARIS_SYSCALL_TRAP /* Slowaris System Call */
-t_netbs:NETBSD_SYSCALL_TRAP /* Net-B.S. System Call */
-t_bad8a:BAD_TRAP(0x8a) BAD_TRAP(0x8b) BAD_TRAP(0x8c) BAD_TRAP(0x8d) BAD_TRAP(0x8e)
-t_bad8f:BAD_TRAP(0x8f)
-t_linux:LINUX_SYSCALL_TRAP /* Linux System Call */
-t_bad91:BAD_TRAP(0x91) BAD_TRAP(0x92) BAD_TRAP(0x93) BAD_TRAP(0x94) BAD_TRAP(0x95)
-t_bad96:BAD_TRAP(0x96) BAD_TRAP(0x97) BAD_TRAP(0x98) BAD_TRAP(0x99) BAD_TRAP(0x9a)
-t_bad9b:BAD_TRAP(0x9b) BAD_TRAP(0x9c) BAD_TRAP(0x9d) BAD_TRAP(0x9e) BAD_TRAP(0x9f)
-t_getcc:GETCC_TRAP /* Get Condition Codes */
-t_setcc:SETCC_TRAP /* Set Condition Codes */
-t_getpsr:GETPSR_TRAP /* Get PSR Register */
-t_bada3:BAD_TRAP(0xa3) BAD_TRAP(0xa4) BAD_TRAP(0xa5) BAD_TRAP(0xa6)
-t_slowi:INDIRECT_SOLARIS_SYSCALL(156)
-t_bada8:BAD_TRAP(0xa8) BAD_TRAP(0xa9) BAD_TRAP(0xaa) BAD_TRAP(0xab)
-t_badac:BAD_TRAP(0xac) BAD_TRAP(0xad) BAD_TRAP(0xae) BAD_TRAP(0xaf) BAD_TRAP(0xb0)
-t_badb1:BAD_TRAP(0xb1) BAD_TRAP(0xb2) BAD_TRAP(0xb3) BAD_TRAP(0xb4) BAD_TRAP(0xb5)
-t_badb6:BAD_TRAP(0xb6) BAD_TRAP(0xb7) BAD_TRAP(0xb8) BAD_TRAP(0xb9) BAD_TRAP(0xba)
-t_badbb:BAD_TRAP(0xbb) BAD_TRAP(0xbc) BAD_TRAP(0xbd) BAD_TRAP(0xbe) BAD_TRAP(0xbf)
-t_badc0:BAD_TRAP(0xc0) BAD_TRAP(0xc1) BAD_TRAP(0xc2) BAD_TRAP(0xc3) BAD_TRAP(0xc4)
-t_badc5:BAD_TRAP(0xc5) BAD_TRAP(0xc6) BAD_TRAP(0xc7) BAD_TRAP(0xc8) BAD_TRAP(0xc9)
-t_badca:BAD_TRAP(0xca) BAD_TRAP(0xcb) BAD_TRAP(0xcc) BAD_TRAP(0xcd) BAD_TRAP(0xce)
-t_badcf:BAD_TRAP(0xcf) BAD_TRAP(0xd0) BAD_TRAP(0xd1) BAD_TRAP(0xd2) BAD_TRAP(0xd3)
-t_badd4:BAD_TRAP(0xd4) BAD_TRAP(0xd5) BAD_TRAP(0xd6) BAD_TRAP(0xd7) BAD_TRAP(0xd8)
-t_badd9:BAD_TRAP(0xd9) BAD_TRAP(0xda) BAD_TRAP(0xdb) BAD_TRAP(0xdc) BAD_TRAP(0xdd)
-t_badde:BAD_TRAP(0xde) BAD_TRAP(0xdf) BAD_TRAP(0xe0) BAD_TRAP(0xe1) BAD_TRAP(0xe2)
-t_bade3:BAD_TRAP(0xe3) BAD_TRAP(0xe4) BAD_TRAP(0xe5) BAD_TRAP(0xe6) BAD_TRAP(0xe7)
-t_bade8:BAD_TRAP(0xe8) BAD_TRAP(0xe9) BAD_TRAP(0xea) BAD_TRAP(0xeb) BAD_TRAP(0xec)
-t_baded:BAD_TRAP(0xed) BAD_TRAP(0xee) BAD_TRAP(0xef) BAD_TRAP(0xf0) BAD_TRAP(0xf1)
-t_badf2:BAD_TRAP(0xf2) BAD_TRAP(0xf3) BAD_TRAP(0xf4) BAD_TRAP(0xf5) BAD_TRAP(0xf6)
-t_badf7:BAD_TRAP(0xf7) BAD_TRAP(0xf8) BAD_TRAP(0xf9) BAD_TRAP(0xfa) BAD_TRAP(0xfb)
-t_badfc:BAD_TRAP(0xfc) BAD_TRAP(0xfd)
-dbtrap: BAD_TRAP(0xfe) /* Debugger/PROM breakpoint #1 */
-dbtrap2:BAD_TRAP(0xff) /* Debugger/PROM breakpoint #2 */
-
- .globl end_traptable
-end_traptable:
-
-#ifdef CONFIG_SMP
- /* Trap tables for the other cpus. */
- .globl trapbase_cpu1, trapbase_cpu2, trapbase_cpu3
-trapbase_cpu1:
- BAD_TRAP(0x0) SRMMU_TFAULT TRAP_ENTRY(0x2, bad_instruction)
- TRAP_ENTRY(0x3, priv_instruction) TRAP_ENTRY(0x4, fpd_trap_handler)
- WINDOW_SPILL WINDOW_FILL TRAP_ENTRY(0x7, mna_handler)
- TRAP_ENTRY(0x8, fpe_trap_handler) SRMMU_DFAULT
- TRAP_ENTRY(0xa, do_tag_overflow) TRAP_ENTRY(0xb, do_watchpoint)
- BAD_TRAP(0xc) BAD_TRAP(0xd) BAD_TRAP(0xe) BAD_TRAP(0xf) BAD_TRAP(0x10)
- TRAP_ENTRY_INTERRUPT(1) TRAP_ENTRY_INTERRUPT(2)
- TRAP_ENTRY_INTERRUPT(3) TRAP_ENTRY_INTERRUPT(4)
- TRAP_ENTRY_INTERRUPT(5) TRAP_ENTRY_INTERRUPT(6)
- TRAP_ENTRY_INTERRUPT(7) TRAP_ENTRY_INTERRUPT(8)
- TRAP_ENTRY_INTERRUPT(9) TRAP_ENTRY_INTERRUPT(10)
- TRAP_ENTRY_INTERRUPT(11) TRAP_ENTRY_INTERRUPT(12)
- TRAP_ENTRY_INTERRUPT(13) TRAP_ENTRY_INTERRUPT(14)
- TRAP_ENTRY(0x1f, linux_trap_ipi15_sun4m)
- TRAP_ENTRY(0x20, do_reg_access) BAD_TRAP(0x21) BAD_TRAP(0x22)
- BAD_TRAP(0x23) TRAP_ENTRY(0x24, do_cp_disabled) SKIP_TRAP(0x25, unimp_flush)
- BAD_TRAP(0x26) BAD_TRAP(0x27) TRAP_ENTRY(0x28, do_cp_exception)
- SRMMU_DFAULT TRAP_ENTRY(0x2a, do_hw_divzero) BAD_TRAP(0x2b) BAD_TRAP(0x2c)
- BAD_TRAP(0x2d) BAD_TRAP(0x2e) BAD_TRAP(0x2f) BAD_TRAP(0x30) BAD_TRAP(0x31)
- BAD_TRAP(0x32) BAD_TRAP(0x33) BAD_TRAP(0x34) BAD_TRAP(0x35) BAD_TRAP(0x36)
- BAD_TRAP(0x37) BAD_TRAP(0x38) BAD_TRAP(0x39) BAD_TRAP(0x3a) BAD_TRAP(0x3b)
- BAD_TRAP(0x3c) BAD_TRAP(0x3d) BAD_TRAP(0x3e) BAD_TRAP(0x3f) BAD_TRAP(0x40)
- BAD_TRAP(0x41) BAD_TRAP(0x42) BAD_TRAP(0x43) BAD_TRAP(0x44) BAD_TRAP(0x45)
- BAD_TRAP(0x46) BAD_TRAP(0x47) BAD_TRAP(0x48) BAD_TRAP(0x49) BAD_TRAP(0x4a)
- BAD_TRAP(0x4b) BAD_TRAP(0x4c) BAD_TRAP(0x4d) BAD_TRAP(0x4e) BAD_TRAP(0x4f)
- BAD_TRAP(0x50)
- BAD_TRAP(0x51) BAD_TRAP(0x52) BAD_TRAP(0x53) BAD_TRAP(0x54) BAD_TRAP(0x55)
- BAD_TRAP(0x56) BAD_TRAP(0x57) BAD_TRAP(0x58) BAD_TRAP(0x59) BAD_TRAP(0x5a)
- BAD_TRAP(0x5b) BAD_TRAP(0x5c) BAD_TRAP(0x5d) BAD_TRAP(0x5e) BAD_TRAP(0x5f)
- BAD_TRAP(0x60) BAD_TRAP(0x61) BAD_TRAP(0x62) BAD_TRAP(0x63) BAD_TRAP(0x64)
- BAD_TRAP(0x65) BAD_TRAP(0x66) BAD_TRAP(0x67) BAD_TRAP(0x68) BAD_TRAP(0x69)
- BAD_TRAP(0x6a) BAD_TRAP(0x6b) BAD_TRAP(0x6c) BAD_TRAP(0x6d) BAD_TRAP(0x6e)
- BAD_TRAP(0x6f) BAD_TRAP(0x70) BAD_TRAP(0x71) BAD_TRAP(0x72) BAD_TRAP(0x73)
- BAD_TRAP(0x74) BAD_TRAP(0x75) BAD_TRAP(0x76) BAD_TRAP(0x77) BAD_TRAP(0x78)
- BAD_TRAP(0x79) BAD_TRAP(0x7a) BAD_TRAP(0x7b) BAD_TRAP(0x7c) BAD_TRAP(0x7d)
- BAD_TRAP(0x7e) BAD_TRAP(0x7f)
- SUNOS_SYSCALL_TRAP
- BREAKPOINT_TRAP
- TRAP_ENTRY(0x82, do_hw_divzero)
- TRAP_ENTRY(0x83, do_flush_windows) BAD_TRAP(0x84) BAD_TRAP(0x85)
- BAD_TRAP(0x86) BAD_TRAP(0x87) SOLARIS_SYSCALL_TRAP
- NETBSD_SYSCALL_TRAP BAD_TRAP(0x8a) BAD_TRAP(0x8b) BAD_TRAP(0x8c)
- BAD_TRAP(0x8d) BAD_TRAP(0x8e) BAD_TRAP(0x8f)
- LINUX_SYSCALL_TRAP BAD_TRAP(0x91) BAD_TRAP(0x92) BAD_TRAP(0x93) BAD_TRAP(0x94)
- BAD_TRAP(0x95) BAD_TRAP(0x96) BAD_TRAP(0x97) BAD_TRAP(0x98) BAD_TRAP(0x99)
- BAD_TRAP(0x9a) BAD_TRAP(0x9b) BAD_TRAP(0x9c) BAD_TRAP(0x9d) BAD_TRAP(0x9e)
- BAD_TRAP(0x9f) GETCC_TRAP SETCC_TRAP GETPSR_TRAP
- BAD_TRAP(0xa3) BAD_TRAP(0xa4) BAD_TRAP(0xa5) BAD_TRAP(0xa6)
- INDIRECT_SOLARIS_SYSCALL(156) BAD_TRAP(0xa8) BAD_TRAP(0xa9) BAD_TRAP(0xaa) BAD_TRAP(0xab)
- BAD_TRAP(0xac) BAD_TRAP(0xad) BAD_TRAP(0xae) BAD_TRAP(0xaf) BAD_TRAP(0xb0)
- BAD_TRAP(0xb1) BAD_TRAP(0xb2) BAD_TRAP(0xb3) BAD_TRAP(0xb4) BAD_TRAP(0xb5)
- BAD_TRAP(0xb6) BAD_TRAP(0xb7) BAD_TRAP(0xb8) BAD_TRAP(0xb9) BAD_TRAP(0xba)
- BAD_TRAP(0xbb) BAD_TRAP(0xbc) BAD_TRAP(0xbd) BAD_TRAP(0xbe) BAD_TRAP(0xbf)
- BAD_TRAP(0xc0) BAD_TRAP(0xc1) BAD_TRAP(0xc2) BAD_TRAP(0xc3) BAD_TRAP(0xc4)
- BAD_TRAP(0xc5) BAD_TRAP(0xc6) BAD_TRAP(0xc7) BAD_TRAP(0xc8) BAD_TRAP(0xc9)
- BAD_TRAP(0xca) BAD_TRAP(0xcb) BAD_TRAP(0xcc) BAD_TRAP(0xcd) BAD_TRAP(0xce)
- BAD_TRAP(0xcf) BAD_TRAP(0xd0) BAD_TRAP(0xd1) BAD_TRAP(0xd2) BAD_TRAP(0xd3)
- BAD_TRAP(0xd4) BAD_TRAP(0xd5) BAD_TRAP(0xd6) BAD_TRAP(0xd7) BAD_TRAP(0xd8)
- BAD_TRAP(0xd9) BAD_TRAP(0xda) BAD_TRAP(0xdb) BAD_TRAP(0xdc) BAD_TRAP(0xdd)
- BAD_TRAP(0xde) BAD_TRAP(0xdf) BAD_TRAP(0xe0) BAD_TRAP(0xe1) BAD_TRAP(0xe2)
- BAD_TRAP(0xe3) BAD_TRAP(0xe4) BAD_TRAP(0xe5) BAD_TRAP(0xe6) BAD_TRAP(0xe7)
- BAD_TRAP(0xe8) BAD_TRAP(0xe9) BAD_TRAP(0xea) BAD_TRAP(0xeb) BAD_TRAP(0xec)
- BAD_TRAP(0xed) BAD_TRAP(0xee) BAD_TRAP(0xef) BAD_TRAP(0xf0) BAD_TRAP(0xf1)
- BAD_TRAP(0xf2) BAD_TRAP(0xf3) BAD_TRAP(0xf4) BAD_TRAP(0xf5) BAD_TRAP(0xf6)
- BAD_TRAP(0xf7) BAD_TRAP(0xf8) BAD_TRAP(0xf9) BAD_TRAP(0xfa) BAD_TRAP(0xfb)
- BAD_TRAP(0xfc) BAD_TRAP(0xfd) BAD_TRAP(0xfe) BAD_TRAP(0xff)
-
-trapbase_cpu2:
- BAD_TRAP(0x0) SRMMU_TFAULT TRAP_ENTRY(0x2, bad_instruction)
- TRAP_ENTRY(0x3, priv_instruction) TRAP_ENTRY(0x4, fpd_trap_handler)
- WINDOW_SPILL WINDOW_FILL TRAP_ENTRY(0x7, mna_handler)
- TRAP_ENTRY(0x8, fpe_trap_handler) SRMMU_DFAULT
- TRAP_ENTRY(0xa, do_tag_overflow) TRAP_ENTRY(0xb, do_watchpoint)
- BAD_TRAP(0xc) BAD_TRAP(0xd) BAD_TRAP(0xe) BAD_TRAP(0xf) BAD_TRAP(0x10)
- TRAP_ENTRY_INTERRUPT(1) TRAP_ENTRY_INTERRUPT(2)
- TRAP_ENTRY_INTERRUPT(3) TRAP_ENTRY_INTERRUPT(4)
- TRAP_ENTRY_INTERRUPT(5) TRAP_ENTRY_INTERRUPT(6)
- TRAP_ENTRY_INTERRUPT(7) TRAP_ENTRY_INTERRUPT(8)
- TRAP_ENTRY_INTERRUPT(9) TRAP_ENTRY_INTERRUPT(10)
- TRAP_ENTRY_INTERRUPT(11) TRAP_ENTRY_INTERRUPT(12)
- TRAP_ENTRY_INTERRUPT(13) TRAP_ENTRY_INTERRUPT(14)
- TRAP_ENTRY(0x1f, linux_trap_ipi15_sun4m)
- TRAP_ENTRY(0x20, do_reg_access) BAD_TRAP(0x21) BAD_TRAP(0x22)
- BAD_TRAP(0x23) TRAP_ENTRY(0x24, do_cp_disabled) SKIP_TRAP(0x25, unimp_flush)
- BAD_TRAP(0x26) BAD_TRAP(0x27) TRAP_ENTRY(0x28, do_cp_exception)
- SRMMU_DFAULT TRAP_ENTRY(0x2a, do_hw_divzero) BAD_TRAP(0x2b) BAD_TRAP(0x2c)
- BAD_TRAP(0x2d) BAD_TRAP(0x2e) BAD_TRAP(0x2f) BAD_TRAP(0x30) BAD_TRAP(0x31)
- BAD_TRAP(0x32) BAD_TRAP(0x33) BAD_TRAP(0x34) BAD_TRAP(0x35) BAD_TRAP(0x36)
- BAD_TRAP(0x37) BAD_TRAP(0x38) BAD_TRAP(0x39) BAD_TRAP(0x3a) BAD_TRAP(0x3b)
- BAD_TRAP(0x3c) BAD_TRAP(0x3d) BAD_TRAP(0x3e) BAD_TRAP(0x3f) BAD_TRAP(0x40)
- BAD_TRAP(0x41) BAD_TRAP(0x42) BAD_TRAP(0x43) BAD_TRAP(0x44) BAD_TRAP(0x45)
- BAD_TRAP(0x46) BAD_TRAP(0x47) BAD_TRAP(0x48) BAD_TRAP(0x49) BAD_TRAP(0x4a)
- BAD_TRAP(0x4b) BAD_TRAP(0x4c) BAD_TRAP(0x4d) BAD_TRAP(0x4e) BAD_TRAP(0x4f)
- BAD_TRAP(0x50)
- BAD_TRAP(0x51) BAD_TRAP(0x52) BAD_TRAP(0x53) BAD_TRAP(0x54) BAD_TRAP(0x55)
- BAD_TRAP(0x56) BAD_TRAP(0x57) BAD_TRAP(0x58) BAD_TRAP(0x59) BAD_TRAP(0x5a)
- BAD_TRAP(0x5b) BAD_TRAP(0x5c) BAD_TRAP(0x5d) BAD_TRAP(0x5e) BAD_TRAP(0x5f)
- BAD_TRAP(0x60) BAD_TRAP(0x61) BAD_TRAP(0x62) BAD_TRAP(0x63) BAD_TRAP(0x64)
- BAD_TRAP(0x65) BAD_TRAP(0x66) BAD_TRAP(0x67) BAD_TRAP(0x68) BAD_TRAP(0x69)
- BAD_TRAP(0x6a) BAD_TRAP(0x6b) BAD_TRAP(0x6c) BAD_TRAP(0x6d) BAD_TRAP(0x6e)
- BAD_TRAP(0x6f) BAD_TRAP(0x70) BAD_TRAP(0x71) BAD_TRAP(0x72) BAD_TRAP(0x73)
- BAD_TRAP(0x74) BAD_TRAP(0x75) BAD_TRAP(0x76) BAD_TRAP(0x77) BAD_TRAP(0x78)
- BAD_TRAP(0x79) BAD_TRAP(0x7a) BAD_TRAP(0x7b) BAD_TRAP(0x7c) BAD_TRAP(0x7d)
- BAD_TRAP(0x7e) BAD_TRAP(0x7f)
- SUNOS_SYSCALL_TRAP
- BREAKPOINT_TRAP
- TRAP_ENTRY(0x82, do_hw_divzero)
- TRAP_ENTRY(0x83, do_flush_windows) BAD_TRAP(0x84) BAD_TRAP(0x85)
- BAD_TRAP(0x86) BAD_TRAP(0x87) SOLARIS_SYSCALL_TRAP
- NETBSD_SYSCALL_TRAP BAD_TRAP(0x8a) BAD_TRAP(0x8b) BAD_TRAP(0x8c)
- BAD_TRAP(0x8d) BAD_TRAP(0x8e) BAD_TRAP(0x8f)
- LINUX_SYSCALL_TRAP BAD_TRAP(0x91) BAD_TRAP(0x92) BAD_TRAP(0x93) BAD_TRAP(0x94)
- BAD_TRAP(0x95) BAD_TRAP(0x96) BAD_TRAP(0x97) BAD_TRAP(0x98) BAD_TRAP(0x99)
- BAD_TRAP(0x9a) BAD_TRAP(0x9b) BAD_TRAP(0x9c) BAD_TRAP(0x9d) BAD_TRAP(0x9e)
- BAD_TRAP(0x9f) GETCC_TRAP SETCC_TRAP GETPSR_TRAP
- BAD_TRAP(0xa3) BAD_TRAP(0xa4) BAD_TRAP(0xa5) BAD_TRAP(0xa6)
- INDIRECT_SOLARIS_SYSCALL(156) BAD_TRAP(0xa8) BAD_TRAP(0xa9) BAD_TRAP(0xaa) BAD_TRAP(0xab)
- BAD_TRAP(0xac) BAD_TRAP(0xad) BAD_TRAP(0xae) BAD_TRAP(0xaf) BAD_TRAP(0xb0)
- BAD_TRAP(0xb1) BAD_TRAP(0xb2) BAD_TRAP(0xb3) BAD_TRAP(0xb4) BAD_TRAP(0xb5)
- BAD_TRAP(0xb6) BAD_TRAP(0xb7) BAD_TRAP(0xb8) BAD_TRAP(0xb9) BAD_TRAP(0xba)
- BAD_TRAP(0xbb) BAD_TRAP(0xbc) BAD_TRAP(0xbd) BAD_TRAP(0xbe) BAD_TRAP(0xbf)
- BAD_TRAP(0xc0) BAD_TRAP(0xc1) BAD_TRAP(0xc2) BAD_TRAP(0xc3) BAD_TRAP(0xc4)
- BAD_TRAP(0xc5) BAD_TRAP(0xc6) BAD_TRAP(0xc7) BAD_TRAP(0xc8) BAD_TRAP(0xc9)
- BAD_TRAP(0xca) BAD_TRAP(0xcb) BAD_TRAP(0xcc) BAD_TRAP(0xcd) BAD_TRAP(0xce)
- BAD_TRAP(0xcf) BAD_TRAP(0xd0) BAD_TRAP(0xd1) BAD_TRAP(0xd2) BAD_TRAP(0xd3)
- BAD_TRAP(0xd4) BAD_TRAP(0xd5) BAD_TRAP(0xd6) BAD_TRAP(0xd7) BAD_TRAP(0xd8)
- BAD_TRAP(0xd9) BAD_TRAP(0xda) BAD_TRAP(0xdb) BAD_TRAP(0xdc) BAD_TRAP(0xdd)
- BAD_TRAP(0xde) BAD_TRAP(0xdf) BAD_TRAP(0xe0) BAD_TRAP(0xe1) BAD_TRAP(0xe2)
- BAD_TRAP(0xe3) BAD_TRAP(0xe4) BAD_TRAP(0xe5) BAD_TRAP(0xe6) BAD_TRAP(0xe7)
- BAD_TRAP(0xe8) BAD_TRAP(0xe9) BAD_TRAP(0xea) BAD_TRAP(0xeb) BAD_TRAP(0xec)
- BAD_TRAP(0xed) BAD_TRAP(0xee) BAD_TRAP(0xef) BAD_TRAP(0xf0) BAD_TRAP(0xf1)
- BAD_TRAP(0xf2) BAD_TRAP(0xf3) BAD_TRAP(0xf4) BAD_TRAP(0xf5) BAD_TRAP(0xf6)
- BAD_TRAP(0xf7) BAD_TRAP(0xf8) BAD_TRAP(0xf9) BAD_TRAP(0xfa) BAD_TRAP(0xfb)
- BAD_TRAP(0xfc) BAD_TRAP(0xfd) BAD_TRAP(0xfe) BAD_TRAP(0xff)
-
-trapbase_cpu3:
- BAD_TRAP(0x0) SRMMU_TFAULT TRAP_ENTRY(0x2, bad_instruction)
- TRAP_ENTRY(0x3, priv_instruction) TRAP_ENTRY(0x4, fpd_trap_handler)
- WINDOW_SPILL WINDOW_FILL TRAP_ENTRY(0x7, mna_handler)
- TRAP_ENTRY(0x8, fpe_trap_handler) SRMMU_DFAULT
- TRAP_ENTRY(0xa, do_tag_overflow) TRAP_ENTRY(0xb, do_watchpoint)
- BAD_TRAP(0xc) BAD_TRAP(0xd) BAD_TRAP(0xe) BAD_TRAP(0xf) BAD_TRAP(0x10)
- TRAP_ENTRY_INTERRUPT(1) TRAP_ENTRY_INTERRUPT(2)
- TRAP_ENTRY_INTERRUPT(3) TRAP_ENTRY_INTERRUPT(4)
- TRAP_ENTRY_INTERRUPT(5) TRAP_ENTRY_INTERRUPT(6)
- TRAP_ENTRY_INTERRUPT(7) TRAP_ENTRY_INTERRUPT(8)
- TRAP_ENTRY_INTERRUPT(9) TRAP_ENTRY_INTERRUPT(10)
- TRAP_ENTRY_INTERRUPT(11) TRAP_ENTRY_INTERRUPT(12)
- TRAP_ENTRY_INTERRUPT(13) TRAP_ENTRY_INTERRUPT(14)
- TRAP_ENTRY(0x1f, linux_trap_ipi15_sun4m)
- TRAP_ENTRY(0x20, do_reg_access) BAD_TRAP(0x21) BAD_TRAP(0x22)
- BAD_TRAP(0x23) TRAP_ENTRY(0x24, do_cp_disabled) SKIP_TRAP(0x25, unimp_flush)
- BAD_TRAP(0x26) BAD_TRAP(0x27) TRAP_ENTRY(0x28, do_cp_exception)
- SRMMU_DFAULT TRAP_ENTRY(0x2a, do_hw_divzero) BAD_TRAP(0x2b) BAD_TRAP(0x2c)
- BAD_TRAP(0x2d) BAD_TRAP(0x2e) BAD_TRAP(0x2f) BAD_TRAP(0x30) BAD_TRAP(0x31)
- BAD_TRAP(0x32) BAD_TRAP(0x33) BAD_TRAP(0x34) BAD_TRAP(0x35) BAD_TRAP(0x36)
- BAD_TRAP(0x37) BAD_TRAP(0x38) BAD_TRAP(0x39) BAD_TRAP(0x3a) BAD_TRAP(0x3b)
- BAD_TRAP(0x3c) BAD_TRAP(0x3d) BAD_TRAP(0x3e) BAD_TRAP(0x3f) BAD_TRAP(0x40)
- BAD_TRAP(0x41) BAD_TRAP(0x42) BAD_TRAP(0x43) BAD_TRAP(0x44) BAD_TRAP(0x45)
- BAD_TRAP(0x46) BAD_TRAP(0x47) BAD_TRAP(0x48) BAD_TRAP(0x49) BAD_TRAP(0x4a)
- BAD_TRAP(0x4b) BAD_TRAP(0x4c) BAD_TRAP(0x4d) BAD_TRAP(0x4e) BAD_TRAP(0x4f)
- BAD_TRAP(0x50)
- BAD_TRAP(0x51) BAD_TRAP(0x52) BAD_TRAP(0x53) BAD_TRAP(0x54) BAD_TRAP(0x55)
- BAD_TRAP(0x56) BAD_TRAP(0x57) BAD_TRAP(0x58) BAD_TRAP(0x59) BAD_TRAP(0x5a)
- BAD_TRAP(0x5b) BAD_TRAP(0x5c) BAD_TRAP(0x5d) BAD_TRAP(0x5e) BAD_TRAP(0x5f)
- BAD_TRAP(0x60) BAD_TRAP(0x61) BAD_TRAP(0x62) BAD_TRAP(0x63) BAD_TRAP(0x64)
- BAD_TRAP(0x65) BAD_TRAP(0x66) BAD_TRAP(0x67) BAD_TRAP(0x68) BAD_TRAP(0x69)
- BAD_TRAP(0x6a) BAD_TRAP(0x6b) BAD_TRAP(0x6c) BAD_TRAP(0x6d) BAD_TRAP(0x6e)
- BAD_TRAP(0x6f) BAD_TRAP(0x70) BAD_TRAP(0x71) BAD_TRAP(0x72) BAD_TRAP(0x73)
- BAD_TRAP(0x74) BAD_TRAP(0x75) BAD_TRAP(0x76) BAD_TRAP(0x77) BAD_TRAP(0x78)
- BAD_TRAP(0x79) BAD_TRAP(0x7a) BAD_TRAP(0x7b) BAD_TRAP(0x7c) BAD_TRAP(0x7d)
- BAD_TRAP(0x7e) BAD_TRAP(0x7f)
- SUNOS_SYSCALL_TRAP
- BREAKPOINT_TRAP
- TRAP_ENTRY(0x82, do_hw_divzero)
- TRAP_ENTRY(0x83, do_flush_windows) BAD_TRAP(0x84) BAD_TRAP(0x85)
- BAD_TRAP(0x86) BAD_TRAP(0x87) SOLARIS_SYSCALL_TRAP
- NETBSD_SYSCALL_TRAP BAD_TRAP(0x8a) BAD_TRAP(0x8b) BAD_TRAP(0x8c)
- BAD_TRAP(0x8d) BAD_TRAP(0x8e) BAD_TRAP(0x8f)
- LINUX_SYSCALL_TRAP BAD_TRAP(0x91) BAD_TRAP(0x92) BAD_TRAP(0x93) BAD_TRAP(0x94)
- BAD_TRAP(0x95) BAD_TRAP(0x96) BAD_TRAP(0x97) BAD_TRAP(0x98) BAD_TRAP(0x99)
- BAD_TRAP(0x9a) BAD_TRAP(0x9b) BAD_TRAP(0x9c) BAD_TRAP(0x9d) BAD_TRAP(0x9e)
- BAD_TRAP(0x9f) GETCC_TRAP SETCC_TRAP GETPSR_TRAP
- BAD_TRAP(0xa3) BAD_TRAP(0xa4) BAD_TRAP(0xa5) BAD_TRAP(0xa6)
- INDIRECT_SOLARIS_SYSCALL(156) BAD_TRAP(0xa8) BAD_TRAP(0xa9) BAD_TRAP(0xaa) BAD_TRAP(0xab)
- BAD_TRAP(0xac) BAD_TRAP(0xad) BAD_TRAP(0xae) BAD_TRAP(0xaf) BAD_TRAP(0xb0)
- BAD_TRAP(0xb1) BAD_TRAP(0xb2) BAD_TRAP(0xb3) BAD_TRAP(0xb4) BAD_TRAP(0xb5)
- BAD_TRAP(0xb6) BAD_TRAP(0xb7) BAD_TRAP(0xb8) BAD_TRAP(0xb9) BAD_TRAP(0xba)
- BAD_TRAP(0xbb) BAD_TRAP(0xbc) BAD_TRAP(0xbd) BAD_TRAP(0xbe) BAD_TRAP(0xbf)
- BAD_TRAP(0xc0) BAD_TRAP(0xc1) BAD_TRAP(0xc2) BAD_TRAP(0xc3) BAD_TRAP(0xc4)
- BAD_TRAP(0xc5) BAD_TRAP(0xc6) BAD_TRAP(0xc7) BAD_TRAP(0xc8) BAD_TRAP(0xc9)
- BAD_TRAP(0xca) BAD_TRAP(0xcb) BAD_TRAP(0xcc) BAD_TRAP(0xcd) BAD_TRAP(0xce)
- BAD_TRAP(0xcf) BAD_TRAP(0xd0) BAD_TRAP(0xd1) BAD_TRAP(0xd2) BAD_TRAP(0xd3)
- BAD_TRAP(0xd4) BAD_TRAP(0xd5) BAD_TRAP(0xd6) BAD_TRAP(0xd7) BAD_TRAP(0xd8)
- BAD_TRAP(0xd9) BAD_TRAP(0xda) BAD_TRAP(0xdb) BAD_TRAP(0xdc) BAD_TRAP(0xdd)
- BAD_TRAP(0xde) BAD_TRAP(0xdf) BAD_TRAP(0xe0) BAD_TRAP(0xe1) BAD_TRAP(0xe2)
- BAD_TRAP(0xe3) BAD_TRAP(0xe4) BAD_TRAP(0xe5) BAD_TRAP(0xe6) BAD_TRAP(0xe7)
- BAD_TRAP(0xe8) BAD_TRAP(0xe9) BAD_TRAP(0xea) BAD_TRAP(0xeb) BAD_TRAP(0xec)
- BAD_TRAP(0xed) BAD_TRAP(0xee) BAD_TRAP(0xef) BAD_TRAP(0xf0) BAD_TRAP(0xf1)
- BAD_TRAP(0xf2) BAD_TRAP(0xf3) BAD_TRAP(0xf4) BAD_TRAP(0xf5) BAD_TRAP(0xf6)
- BAD_TRAP(0xf7) BAD_TRAP(0xf8) BAD_TRAP(0xf9) BAD_TRAP(0xfa) BAD_TRAP(0xfb)
- BAD_TRAP(0xfc) BAD_TRAP(0xfd) BAD_TRAP(0xfe) BAD_TRAP(0xff)
-
-#endif
- .align PAGE_SIZE
-
-/* This was the only reasonable way I could think of to properly align
- * these page-table data structures.
- */
- .globl pg0, pg1, pg2, pg3
- .globl empty_bad_page
- .globl empty_bad_page_table
- .globl empty_zero_page
- .globl swapper_pg_dir
-swapper_pg_dir: .skip PAGE_SIZE
-pg0: .skip PAGE_SIZE
-pg1: .skip PAGE_SIZE
-pg2: .skip PAGE_SIZE
-pg3: .skip PAGE_SIZE
-empty_bad_page: .skip PAGE_SIZE
-empty_bad_page_table: .skip PAGE_SIZE
-empty_zero_page: .skip PAGE_SIZE
-
- .global root_flags
- .global ram_flags
- .global root_dev
- .global sparc_ramdisk_image
- .global sparc_ramdisk_size
-
-/* This stuff has to be in sync with SILO and other potential boot loaders
- * Fields should be kept upward compatible and whenever any change is made,
- * HdrS version should be incremented.
- */
- .ascii "HdrS"
- .word LINUX_VERSION_CODE
- .half 0x0203 /* HdrS version */
-root_flags:
- .half 1
-root_dev:
- .half 0
-ram_flags:
- .half 0
-sparc_ramdisk_image:
- .word 0
-sparc_ramdisk_size:
- .word 0
- .word reboot_command
- .word 0, 0, 0
- .word _end
-
-/* Cool, here we go. Pick up the romvec pointer in %o0 and stash it in
- * %g7 and at prom_vector_p. And also quickly check whether we are on
- * a v0, v2, or v3 prom.
- */
-gokernel:
- /* Ok, it's nice to know, as early as possible, if we
- * are already mapped where we expect to be in virtual
- * memory. The Solaris /boot elf format bootloader
- * will peek into our elf header and load us where
- * we want to be, otherwise we have to re-map.
- *
- * Some boot loaders don't place the jmp'rs address
- * in %o7, so we do a pc-relative call to a local
- * label, then see what %o7 has.
- */
-
- mov %o7, %g4 ! Save %o7
-
- /* Jump to it, and pray... */
-current_pc:
- call 1f
- nop
-
-1:
- mov %o7, %g3
-
- tst %o0
- be no_sun4u_here
- mov %g4, %o7 /* Previous %o7. */
-
- mov %o0, %l0 ! stash away romvec
- mov %o0, %g7 ! put it here too
- mov %o1, %l1 ! stash away debug_vec too
-
- /* Ok, let's check out our run time program counter. */
- set current_pc, %g5
- cmp %g3, %g5
- be already_mapped
- nop
-
- /* %l6 will hold the offset we have to subtract
- * from absolute symbols in order to access areas
- * in our own image. If already mapped this is
- * just plain zero, else it is KERNBASE.
- */
- set KERNBASE, %l6
- b copy_prom_lvl14
- nop
-
-already_mapped:
- mov 0, %l6
-
- /* Copy over the Prom's level 14 clock handler. */
-copy_prom_lvl14:
-#if 1
- /* DJHR
- * preserve our linked/calculated instructions
- */
- set lvl14_save, %g1
- set t_irq14, %g3
- sub %g1, %l6, %g1 ! translate to physical
- sub %g3, %l6, %g3 ! translate to physical
- ldd [%g3], %g4
- std %g4, [%g1]
- ldd [%g3+8], %g4
- std %g4, [%g1+8]
-#endif
- rd %tbr, %g1
- andn %g1, 0xfff, %g1 ! proms trap table base
- or %g0, (0x1e<<4), %g2 ! offset to lvl14 intr
- or %g1, %g2, %g2
- set t_irq14, %g3
- sub %g3, %l6, %g3
- ldd [%g2], %g4
- std %g4, [%g3]
- ldd [%g2 + 0x8], %g4
- std %g4, [%g3 + 0x8] ! Copy proms handler
-
-/* Must determine whether we are on a sun4c MMU, SRMMU, or SUN4/400 MUTANT
- * MMU so we can remap ourselves properly. DON'T TOUCH %l0 thru %l5 in these
- * remapping routines, we need their values afterwards!
- */
- /* Now check whether we are already mapped, if we
- * are we can skip all this garbage coming up.
- */
-copy_prom_done:
- cmp %l6, 0
- be go_to_highmem ! this will be a nop then
- nop
-
- set LOAD_ADDR, %g6
- cmp %g7, %g6
- bne remap_not_a_sun4 ! This is not a Sun4
- nop
-
- or %g0, 0x1, %g1
- lduba [%g1] ASI_CONTROL, %g1 ! Only safe to try on Sun4.
- subcc %g1, 0x24, %g0 ! Is this a mutant Sun4/400???
- be sun4_mutant_remap ! Ugh, it is...
- nop
-
- b sun4_normal_remap ! regular sun4, 2 level mmu
- nop
-
-remap_not_a_sun4:
- lda [%g0] ASI_M_MMUREGS, %g1 ! same as ASI_PTE on sun4c
- and %g1, 0x1, %g1 ! Test SRMMU Enable bit ;-)
- cmp %g1, 0x0
- be sun4c_remap ! A sun4c MMU or normal Sun4
- nop
-srmmu_remap:
- /* First, check for a viking (TI) module. */
- set 0x40000000, %g2
- rd %psr, %g3
- and %g2, %g3, %g3
- subcc %g3, 0x0, %g0
- bz srmmu_nviking
- nop
-
- /* Figure out what kind of viking we are on.
- * We need to know if we have to play with the
- * AC bit and disable traps or not.
- */
-
- /* I've only seen MicroSparc's on SparcClassics with this
- * bit set.
- */
- set 0x800, %g2
- lda [%g0] ASI_M_MMUREGS, %g3 ! peek in the control reg
- and %g2, %g3, %g3
- subcc %g3, 0x0, %g0
- bnz srmmu_nviking ! is in mbus mode
- nop
-
- rd %psr, %g3 ! DO NOT TOUCH %g3
- andn %g3, PSR_ET, %g2
- wr %g2, 0x0, %psr
- WRITE_PAUSE
-
- /* Get context table pointer, then convert to
- * a physical address, which is 36 bits.
- */
- set AC_M_CTPR, %g4
- lda [%g4] ASI_M_MMUREGS, %g4
- sll %g4, 0x4, %g4 ! We use this below
- ! DO NOT TOUCH %g4
-
- /* Set the AC bit in the Viking's MMU control reg. */
- lda [%g0] ASI_M_MMUREGS, %g5 ! DO NOT TOUCH %g5
- set 0x8000, %g6 ! AC bit mask
- or %g5, %g6, %g6 ! Or it in...
- sta %g6, [%g0] ASI_M_MMUREGS ! Close your eyes...
-
- /* Grrr, why does it seem like every other load/store
- * on the sun4m is in some ASI space...
- * Fine with me, let's get the pointer to the level 1
- * page table directory and fetch its entry.
- */
- lda [%g4] ASI_M_BYPASS, %o1 ! This is a level 1 ptr
- srl %o1, 0x4, %o1 ! Clear low 4 bits
- sll %o1, 0x8, %o1 ! Make physical
-
- /* Ok, pull in the PTD. */
- lda [%o1] ASI_M_BYPASS, %o2 ! This is the 0x0 16MB pgd
-
- /* Calculate to KERNBASE entry. */
- add %o1, KERNBASE >> (SRMMU_PGDIR_SHIFT - 2), %o3
-
- /* Poke the entry into the calculated address. */
- sta %o2, [%o3] ASI_M_BYPASS
-
- /* I don't get it Sun, if you engineered all these
- * boot loaders and the PROM (thank you for the debugging
- * features btw) why did you not have them load kernel
- * images up in high address space, since this is necessary
- * for ABI compliance anyways? Does this low-mapping provide
- * enhanced interoperability?
- *
- * "The PROM is the computer."
- */
-
- /* Ok, restore the MMU control register we saved in %g5 */
- sta %g5, [%g0] ASI_M_MMUREGS ! POW... ouch
-
- /* Turn traps back on. We saved it in %g3 earlier. */
- wr %g3, 0x0, %psr ! tick tock, tick tock
-
- /* Now we burn precious CPU cycles due to bad engineering. */
- WRITE_PAUSE
-
- /* Wow, all that just to move a 32-bit value from one
- * place to another... Jump to high memory.
- */
- b go_to_highmem
- nop
-
- /* This works on viking's in Mbus mode and all
- * other MBUS modules. It is virtually the same as
- * the above madness sans turning traps off and flipping
- * the AC bit.
- */
-srmmu_nviking:
- set AC_M_CTPR, %g1
- lda [%g1] ASI_M_MMUREGS, %g1 ! get ctx table ptr
- sll %g1, 0x4, %g1 ! make physical addr
- lda [%g1] ASI_M_BYPASS, %g1 ! ptr to level 1 pg_table
- srl %g1, 0x4, %g1
- sll %g1, 0x8, %g1 ! make phys addr for l1 tbl
-
- lda [%g1] ASI_M_BYPASS, %g2 ! get level1 entry for 0x0
- add %g1, KERNBASE >> (SRMMU_PGDIR_SHIFT - 2), %g3
- sta %g2, [%g3] ASI_M_BYPASS ! place at KERNBASE entry
- b go_to_highmem
- nop ! wheee....
-
- /* This remaps the kernel on Sun4/4xx machines
- * that have the Sun Mutant Three Level MMU.
- * It's like a platypus, Sun didn't have the
- * SRMMU in conception so they kludged the three
- * level logic in the regular Sun4 MMU probably.
- *
- * Basically, you take each entry in the top level
- * directory that maps the low 3MB starting at
- * address zero and put the mapping in the KERNBASE
- * slots. These top level pgd's are called regmaps.
- */
-sun4_mutant_remap:
- or %g0, %g0, %g3 ! source base
- sethi %hi(KERNBASE), %g4 ! destination base
- or %g4, %lo(KERNBASE), %g4
- sethi %hi(0x300000), %g5
- or %g5, %lo(0x300000), %g5 ! upper bound 3MB
- or %g0, 0x1, %l6
- sll %l6, 24, %l6 ! Regmap mapping size
- add %g3, 0x2, %g3 ! Base magic
- add %g4, 0x2, %g4 ! Base magic
-
- /* Main remapping loop on Sun4-Mutant-MMU.
- * "I am not an animal..." -Famous Mutant Person
- */
-sun4_mutant_loop:
- lduha [%g3] ASI_REGMAP, %g2 ! Get lower entry
- stha %g2, [%g4] ASI_REGMAP ! Store in high entry
- add %g4, %l6, %g4 ! Move up high memory ptr
- subcc %g3, %g5, %g0 ! Reached our limit?
- blu sun4_mutant_loop ! Nope, loop again
- add %g3, %l6, %g3 ! delay, Move up low ptr
- b go_to_highmem ! Jump to high memory.
- nop
-
- /* The following is for non-4/4xx sun4 MMU's. */
-sun4_normal_remap:
- mov 0, %g3 ! source base
- set KERNBASE, %g4 ! destination base
- set 0x300000, %g5 ! upper bound 3MB
- mov 1, %l6
- sll %l6, 18, %l6 ! sun4 mmu segmap size
-sun4_normal_loop:
- lduha [%g3] ASI_SEGMAP, %g6 ! load phys_seg
- stha %g6, [%g4] ASI_SEGMAP ! stort new virt mapping
- add %g3, %l6, %g3 ! increment source pointer
- subcc %g3, %g5, %g0 ! reached limit?
- blu sun4_normal_loop ! nope, loop again
- add %g4, %l6, %g4 ! delay, increment dest ptr
- b go_to_highmem
- nop
-
- /* The following works for Sun4c MMU's */
-sun4c_remap:
- mov 0, %g3 ! source base
- set KERNBASE, %g4 ! destination base
- set 0x300000, %g5 ! upper bound 3MB
- mov 1, %l6
- sll %l6, 18, %l6 ! sun4c mmu segmap size
-sun4c_remap_loop:
- lda [%g3] ASI_SEGMAP, %g6 ! load phys_seg
- sta %g6, [%g4] ASI_SEGMAP ! store new virt mapping
- add %g3, %l6, %g3 ! Increment source ptr
- subcc %g3, %g5, %g0 ! Reached limit?
- bl sun4c_remap_loop ! Nope, loop again
- add %g4, %l6, %g4 ! delay, Increment dest ptr
-
-/* Now do a non-relative jump so that PC is in high-memory */
-go_to_highmem:
- set execute_in_high_mem, %g1
- jmpl %g1, %g0
- nop
-
-/* The code above should be at beginning and we have to take care about
- * short jumps, as branching to .text.init section from .text is usually
- * impossible */
- __INIT
-/* Acquire boot time privileged register values, this will help debugging.
- * I figure out and store nwindows and nwindowsm1 later on.
- */
-execute_in_high_mem:
- mov %l0, %o0 ! put back romvec
- mov %l1, %o1 ! and debug_vec
-
- sethi %hi(prom_vector_p), %g1
- st %o0, [%g1 + %lo(prom_vector_p)]
-
- sethi %hi(linux_dbvec), %g1
- st %o1, [%g1 + %lo(linux_dbvec)]
-
- ld [%o0 + 0x4], %o3
- and %o3, 0x3, %o5 ! get the version
-
- cmp %o3, 0x2 ! a v2 prom?
- be found_version
- nop
-
- /* paul@sfe.com.au */
- cmp %o3, 0x3 ! a v3 prom?
- be found_version
- nop
-
-/* Old sun4's pass our load address into %o0 instead of the prom
- * pointer. On sun4's you have to hard code the romvec pointer into
- * your code. Sun probably still does that because they don't even
- * trust their own "OpenBoot" specifications.
- */
- set LOAD_ADDR, %g6
- cmp %o0, %g6 ! an old sun4?
- be sun4_init
- nop
-
-found_version:
-#ifdef CONFIG_SUN4
-/* For people who try sun4 kernels, even if Configure.help advises them. */
- ld [%g7 + 0x68], %o1
- set sun4cdm_notsup, %o0
- call %o1
- nop
- b halt_me
- nop
-#endif
-/* Get the machine type via the mysterious romvec node operations. */
-
- add %g7, 0x1c, %l1
- ld [%l1], %l0
- ld [%l0], %l0
- call %l0
- or %g0, %g0, %o0 ! next_node(0) = first_node
- or %o0, %g0, %g6
-
- sethi %hi(cputypvar), %o1 ! First node has cpu-arch
- or %o1, %lo(cputypvar), %o1
- sethi %hi(cputypval), %o2 ! information, the string
- or %o2, %lo(cputypval), %o2
- ld [%l1], %l0 ! 'compatibility' tells
- ld [%l0 + 0xc], %l0 ! that we want 'sun4x' where
- call %l0 ! x is one of '', 'c', 'm',
- nop ! 'd' or 'e'. %o2 holds pointer
- ! to a buf where above string
- ! will get stored by the prom.
-
- subcc %o0, %g0, %g0
- bpos got_prop ! Got the property
- nop
-
- or %g6, %g0, %o0
- sethi %hi(cputypvar_sun4m), %o1
- or %o1, %lo(cputypvar_sun4m), %o1
- sethi %hi(cputypval), %o2
- or %o2, %lo(cputypval), %o2
- ld [%l1], %l0
- ld [%l0 + 0xc], %l0
- call %l0
- nop
-
-got_prop:
- set cputypval, %o2
- ldub [%o2 + 0x4], %l1
-
- cmp %l1, ' '
- be 1f
- cmp %l1, 'c'
- be 1f
- cmp %l1, 'm'
- be 1f
- cmp %l1, 's'
- be 1f
- cmp %l1, 'd'
- be 1f
- cmp %l1, 'e'
- be no_sun4e_here ! Could be a sun4e.
- nop
- b no_sun4u_here ! AIEEE, a V9 sun4u... Get our BIG BROTHER kernel :))
- nop
-
-1: set cputypval, %l1
- ldub [%l1 + 0x4], %l1
- cmp %l1, 'm' ! Test for sun4d, sun4e ?
- be sun4m_init
- cmp %l1, 's' ! Treat sun4s as sun4m
- be sun4m_init
- cmp %l1, 'd' ! Let us see how the beast will die
- be sun4d_init
- nop
-
- /* Jump into mmu context zero. */
- set AC_CONTEXT, %g1
- stba %g0, [%g1] ASI_CONTROL
-
- b sun4c_continue_boot
- nop
-
-/* CPUID in bootbus can be found at PA 0xff0140000 */
-#define SUN4D_BOOTBUS_CPUID 0xf0140000
-
-sun4d_init:
- /* Need to patch call to handler_irq */
- set patch_handler_irq, %g4
- set sun4d_handler_irq, %g5
- sethi %hi(0x40000000), %g3 ! call
- sub %g5, %g4, %g5
- srl %g5, 2, %g5
- or %g5, %g3, %g5
- st %g5, [%g4]
-
-#ifdef CONFIG_SMP
- /* Get our CPU id out of bootbus */
- set SUN4D_BOOTBUS_CPUID, %g3
- lduba [%g3] ASI_M_CTL, %g3
- and %g3, 0xf8, %g3
- srl %g3, 3, %g4
- sta %g4, [%g0] ASI_M_VIKING_TMP1
- sethi %hi(boot_cpu_id), %g5
- stb %g4, [%g5 + %lo(boot_cpu_id)]
- sll %g4, 2, %g4
- sethi %hi(boot_cpu_id4), %g5
- stb %g4, [%g5 + %lo(boot_cpu_id4)]
-#endif
-
- /* Fall through to sun4m_init */
-
-sun4m_init:
- /* XXX Fucking Cypress... */
- lda [%g0] ASI_M_MMUREGS, %g5
- srl %g5, 28, %g4
-
- cmp %g4, 1
- bne 1f
- srl %g5, 24, %g4
-
- and %g4, 0xf, %g4
- cmp %g4, 7 /* This would be a HyperSparc. */
-
- bne 2f
- nop
-
-1:
-
-#define PATCH_IT(dst, src) \
- set (dst), %g5; \
- set (src), %g4; \
- ld [%g4], %g3; \
- st %g3, [%g5]; \
- ld [%g4+0x4], %g3; \
- st %g3, [%g5+0x4];
-
- /* Signed multiply. */
- PATCH_IT(.mul, .mul_patch)
- PATCH_IT(.mul+0x08, .mul_patch+0x08)
-
- /* Signed remainder. */
- PATCH_IT(.rem, .rem_patch)
- PATCH_IT(.rem+0x08, .rem_patch+0x08)
- PATCH_IT(.rem+0x10, .rem_patch+0x10)
- PATCH_IT(.rem+0x18, .rem_patch+0x18)
- PATCH_IT(.rem+0x20, .rem_patch+0x20)
- PATCH_IT(.rem+0x28, .rem_patch+0x28)
-
- /* Signed division. */
- PATCH_IT(.div, .div_patch)
- PATCH_IT(.div+0x08, .div_patch+0x08)
- PATCH_IT(.div+0x10, .div_patch+0x10)
- PATCH_IT(.div+0x18, .div_patch+0x18)
- PATCH_IT(.div+0x20, .div_patch+0x20)
-
- /* Unsigned multiply. */
- PATCH_IT(.umul, .umul_patch)
- PATCH_IT(.umul+0x08, .umul_patch+0x08)
-
- /* Unsigned remainder. */
- PATCH_IT(.urem, .urem_patch)
- PATCH_IT(.urem+0x08, .urem_patch+0x08)
- PATCH_IT(.urem+0x10, .urem_patch+0x10)
- PATCH_IT(.urem+0x18, .urem_patch+0x18)
-
- /* Unsigned division. */
- PATCH_IT(.udiv, .udiv_patch)
- PATCH_IT(.udiv+0x08, .udiv_patch+0x08)
- PATCH_IT(.udiv+0x10, .udiv_patch+0x10)
-
-#undef PATCH_IT
-
-/* Ok, the PROM could have done funny things and apple cider could still
- * be sitting in the fault status/address registers. Read them all to
- * clear them so we don't get magic faults later on.
- */
-/* This sucks, apparently this makes Vikings call prom panic, will fix later */
-2:
- rd %psr, %o1
- srl %o1, 28, %o1 ! Get a type of the CPU
-
- subcc %o1, 4, %g0 ! TI: Viking or MicroSPARC
- be sun4c_continue_boot
- nop
-
- set AC_M_SFSR, %o0
- lda [%o0] ASI_M_MMUREGS, %g0
- set AC_M_SFAR, %o0
- lda [%o0] ASI_M_MMUREGS, %g0
-
- /* Fujitsu MicroSPARC-II has no asynchronous flavors of FARs */
- subcc %o1, 0, %g0
- be sun4c_continue_boot
- nop
-
- set AC_M_AFSR, %o0
- lda [%o0] ASI_M_MMUREGS, %g0
- set AC_M_AFAR, %o0
- lda [%o0] ASI_M_MMUREGS, %g0
- nop
-
-
-sun4c_continue_boot:
-
-
-/* Aieee, now set PC and nPC, enable traps, give ourselves a stack and it's
- * show-time!
- */
-
- sethi %hi(cputyp), %o0
- st %g4, [%o0 + %lo(cputyp)]
-
- /* Turn on Supervisor, EnableFloating, and all the PIL bits.
- * Also puts us in register window zero with traps off.
- */
- set (PSR_PS | PSR_S | PSR_PIL | PSR_EF), %g2
- wr %g2, 0x0, %psr
- WRITE_PAUSE
-
- /* I want a kernel stack NOW! */
- set init_thread_union, %g1
- set (THREAD_SIZE - STACKFRAME_SZ), %g2
- add %g1, %g2, %sp
- mov 0, %fp /* And for good luck */
-
- /* Zero out our BSS section. */
- set __bss_start , %o0 ! First address of BSS
- set end , %o1 ! Last address of BSS
- add %o0, 0x1, %o0
-1:
- stb %g0, [%o0]
- subcc %o0, %o1, %g0
- bl 1b
- add %o0, 0x1, %o0
-
- /* Initialize the uwinmask value for init task just in case.
- * But first make current_set[boot_cpu_id] point to something useful.
- */
- set init_thread_union, %g6
- set current_set, %g2
-#ifdef CONFIG_SMP
- sethi %hi(boot_cpu_id4), %g3
- ldub [%g3 + %lo(boot_cpu_id4)], %g3
- st %g6, [%g2]
- add %g2, %g3, %g2
-#endif
- st %g6, [%g2]
-
- st %g0, [%g6 + TI_UWINMASK]
-
-/* Compute NWINDOWS and stash it away. Now uses %wim trick explained
- * in the V8 manual. Ok, this method seems to work, Sparc is cool...
- * No, it doesn't work, have to play the save/readCWP/restore trick.
- */
-
- wr %g0, 0x0, %wim ! so we do not get a trap
- WRITE_PAUSE
-
- save
-
- rd %psr, %g3
-
- restore
-
- and %g3, 0x1f, %g3
- add %g3, 0x1, %g3
-
- mov 2, %g1
- wr %g1, 0x0, %wim ! make window 1 invalid
- WRITE_PAUSE
-
- cmp %g3, 0x7
- bne 2f
- nop
-
- /* Adjust our window handling routines to
- * do things correctly on 7 window Sparcs.
- */
-
-#define PATCH_INSN(src, dest) \
- set src, %g5; \
- set dest, %g2; \
- ld [%g5], %g4; \
- st %g4, [%g2];
-
- /* Patch for window spills... */
- PATCH_INSN(spnwin_patch1_7win, spnwin_patch1)
- PATCH_INSN(spnwin_patch2_7win, spnwin_patch2)
- PATCH_INSN(spnwin_patch3_7win, spnwin_patch3)
-
- /* Patch for window fills... */
- PATCH_INSN(fnwin_patch1_7win, fnwin_patch1)
- PATCH_INSN(fnwin_patch2_7win, fnwin_patch2)
-
- /* Patch for trap entry setup... */
- PATCH_INSN(tsetup_7win_patch1, tsetup_patch1)
- PATCH_INSN(tsetup_7win_patch2, tsetup_patch2)
- PATCH_INSN(tsetup_7win_patch3, tsetup_patch3)
- PATCH_INSN(tsetup_7win_patch4, tsetup_patch4)
- PATCH_INSN(tsetup_7win_patch5, tsetup_patch5)
- PATCH_INSN(tsetup_7win_patch6, tsetup_patch6)
-
- /* Patch for returning from traps... */
- PATCH_INSN(rtrap_7win_patch1, rtrap_patch1)
- PATCH_INSN(rtrap_7win_patch2, rtrap_patch2)
- PATCH_INSN(rtrap_7win_patch3, rtrap_patch3)
- PATCH_INSN(rtrap_7win_patch4, rtrap_patch4)
- PATCH_INSN(rtrap_7win_patch5, rtrap_patch5)
-
- /* Patch for killing user windows from the register file. */
- PATCH_INSN(kuw_patch1_7win, kuw_patch1)
-
- /* Now patch the kernel window flush sequences.
- * This saves 2 traps on every switch and fork.
- */
- set 0x01000000, %g4
- set flush_patch_one, %g5
- st %g4, [%g5 + 0x18]
- st %g4, [%g5 + 0x1c]
- set flush_patch_two, %g5
- st %g4, [%g5 + 0x18]
- st %g4, [%g5 + 0x1c]
- set flush_patch_three, %g5
- st %g4, [%g5 + 0x18]
- st %g4, [%g5 + 0x1c]
- set flush_patch_four, %g5
- st %g4, [%g5 + 0x18]
- st %g4, [%g5 + 0x1c]
- set flush_patch_exception, %g5
- st %g4, [%g5 + 0x18]
- st %g4, [%g5 + 0x1c]
- set flush_patch_switch, %g5
- st %g4, [%g5 + 0x18]
- st %g4, [%g5 + 0x1c]
-
-2:
- sethi %hi(nwindows), %g4
- st %g3, [%g4 + %lo(nwindows)] ! store final value
- sub %g3, 0x1, %g3
- sethi %hi(nwindowsm1), %g4
- st %g3, [%g4 + %lo(nwindowsm1)]
-
- /* Here we go, start using Linux's trap table... */
- set trapbase, %g3
- wr %g3, 0x0, %tbr
- WRITE_PAUSE
-
- /* Finally, turn on traps so that we can call c-code. */
- rd %psr, %g3
- wr %g3, 0x0, %psr
- WRITE_PAUSE
-
- wr %g3, PSR_ET, %psr
- WRITE_PAUSE
-
- /* First we call prom_init() to set up PROMLIB, then
- * off to start_kernel().
- */
-
- sethi %hi(prom_vector_p), %g5
- ld [%g5 + %lo(prom_vector_p)], %o0
- call prom_init
- nop
-
- call start_kernel
- nop
-
- /* We should not get here. */
- call halt_me
- nop
-
-sun4_init:
-#ifdef CONFIG_SUN4
-/* There, happy now Adrian? */
- set cputypval, %o2 ! Let everyone know we
- set ' ', %o0 ! are a "sun4 " architecture
- stb %o0, [%o2 + 0x4]
-
- b got_prop
- nop
-#else
- sethi %hi(SUN4_PROM_VECTOR+0x84), %o1
- ld [%o1 + %lo(SUN4_PROM_VECTOR+0x84)], %o1
- set sun4_notsup, %o0
- call %o1 /* printf */
- nop
- sethi %hi(SUN4_PROM_VECTOR+0xc4), %o1
- ld [%o1 + %lo(SUN4_PROM_VECTOR+0xc4)], %o1
- call %o1 /* exittomon */
- nop
-1: ba 1b ! Cannot exit into KMON
- nop
-#endif
-no_sun4e_here:
- ld [%g7 + 0x68], %o1
- set sun4e_notsup, %o0
- call %o1
- nop
- b halt_me
- nop
-
- __INITDATA
-
-sun4u_1:
- .asciz "finddevice"
- .align 4
-sun4u_2:
- .asciz "/chosen"
- .align 4
-sun4u_3:
- .asciz "getprop"
- .align 4
-sun4u_4:
- .asciz "stdout"
- .align 4
-sun4u_5:
- .asciz "write"
- .align 4
-sun4u_6:
- .asciz "\n\rOn sun4u you have to use UltraLinux (64bit) kernel\n\rand not a 32bit sun4[cdem] version\n\r\n\r"
-sun4u_6e:
- .align 4
-sun4u_7:
- .asciz "exit"
- .align 8
-sun4u_a1:
- .word 0, sun4u_1, 0, 1, 0, 1, 0, sun4u_2, 0
-sun4u_r1:
- .word 0
-sun4u_a2:
- .word 0, sun4u_3, 0, 4, 0, 1, 0
-sun4u_i2:
- .word 0, 0, sun4u_4, 0, sun4u_1, 0, 8, 0
-sun4u_r2:
- .word 0
-sun4u_a3:
- .word 0, sun4u_5, 0, 3, 0, 1, 0
-sun4u_i3:
- .word 0, 0, sun4u_6, 0, sun4u_6e - sun4u_6 - 1, 0
-sun4u_r3:
- .word 0
-sun4u_a4:
- .word 0, sun4u_7, 0, 0, 0, 0
-sun4u_r4:
-
- __INIT
-no_sun4u_here:
- set sun4u_a1, %o0
- set current_pc, %l2
- cmp %l2, %g3
- be 1f
- mov %o4, %l0
- sub %g3, %l2, %l6
- add %o0, %l6, %o0
- mov %o0, %l4
- mov sun4u_r4 - sun4u_a1, %l3
- ld [%l4], %l5
-2:
- add %l4, 4, %l4
- cmp %l5, %l2
- add %l5, %l6, %l5
- bgeu,a 3f
- st %l5, [%l4 - 4]
-3:
- subcc %l3, 4, %l3
- bne 2b
- ld [%l4], %l5
-1:
- call %l0
- mov %o0, %l1
-
- ld [%l1 + (sun4u_r1 - sun4u_a1)], %o1
- add %l1, (sun4u_a2 - sun4u_a1), %o0
- call %l0
- st %o1, [%o0 + (sun4u_i2 - sun4u_a2)]
-
- ld [%l1 + (sun4u_1 - sun4u_a1)], %o1
- add %l1, (sun4u_a3 - sun4u_a1), %o0
- call %l0
- st %o1, [%o0 + (sun4u_i3 - sun4u_a3)]
-
- call %l0
- add %l1, (sun4u_a4 - sun4u_a1), %o0
-
- /* Not reached */
-halt_me:
- ld [%g7 + 0x74], %o0
- call %o0 ! Get us out of here...
- nop ! Apparently Solaris is better.
-
-/* Ok, now we continue in the .data/.text sections */
-
- .data
- .align 4
-
-/*
- * Fill up the prom vector, note in particular the kind first element,
- * no joke. I don't need all of them in here as the entire prom vector
- * gets initialized in c-code so all routines can use it.
- */
-
- .globl prom_vector_p
-prom_vector_p:
- .word 0
-
-/* We calculate the following at boot time, window fills/spills and trap entry
- * code uses these to keep track of the register windows.
- */
-
- .align 4
- .globl nwindows
- .globl nwindowsm1
-nwindows:
- .word 8
-nwindowsm1:
- .word 7
-
-/* Boot time debugger vector value. We need this later on. */
-
- .align 4
- .globl linux_dbvec
-linux_dbvec:
- .word 0
- .word 0
-
- .align 8
-
- .globl lvl14_save
-lvl14_save:
- .word 0
- .word 0
- .word 0
- .word 0
- .word t_irq14
-
- .section ".fixup",#alloc,#execinstr
- .globl __ret_efault
-__ret_efault:
- ret
- restore %g0, -EFAULT, %o0
diff --git a/arch/sparc/kernel/head_32.S b/arch/sparc/kernel/head_32.S
new file mode 100644
index 00000000000..3d92c0a8f6c
--- /dev/null
+++ b/arch/sparc/kernel/head_32.S
@@ -0,0 +1,815 @@
+/*
+ * head.S: The initial boot code for the Sparc port of Linux.
+ *
+ * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1995,1999 Pete Zaitcev (zaitcev@yahoo.com)
+ * Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx)
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ * Copyright (C) 1997 Michael A. Griffith (grif@acm.org)
+ *
+ * CompactPCI platform by Eric Brower, 1999.
+ */
+
+#include <linux/version.h>
+#include <linux/init.h>
+
+#include <asm/head.h>
+#include <asm/asi.h>
+#include <asm/contregs.h>
+#include <asm/ptrace.h>
+#include <asm/psr.h>
+#include <asm/page.h>
+#include <asm/kdebug.h>
+#include <asm/winmacro.h>
+#include <asm/thread_info.h> /* TI_UWINMASK */
+#include <asm/errno.h>
+#include <asm/pgtsrmmu.h> /* SRMMU_PGDIR_SHIFT */
+
+ .data
+/* The following are used with the prom_vector node-ops to figure out
+ * the cpu-type
+ */
+ .align 4
+ .globl cputypval
+cputypval:
+ .asciz "sun4m"
+ .ascii " "
+
+/* Tested on SS-5, SS-10 */
+ .align 4
+cputypvar:
+ .asciz "compatible"
+
+ .align 4
+
+notsup:
+ .asciz "Sparc-Linux sun4/sun4c or MMU-less not supported\n\n"
+ .align 4
+
+sun4e_notsup:
+ .asciz "Sparc-Linux sun4e support does not exist\n\n"
+ .align 4
+
+/* The trap-table - located in the __HEAD section */
+#include "ttable_32.S"
+
+ .align PAGE_SIZE
+
+/* This was the only reasonable way I could think of to properly align
+ * these page-table data structures.
+ */
+ .globl empty_zero_page
+empty_zero_page: .skip PAGE_SIZE
+
+ .global root_flags
+ .global ram_flags
+ .global root_dev
+ .global sparc_ramdisk_image
+ .global sparc_ramdisk_size
+
+/* This stuff has to be in sync with SILO and other potential boot loaders
+ * Fields should be kept upward compatible and whenever any change is made,
+ * HdrS version should be incremented.
+ */
+ .ascii "HdrS"
+ .word LINUX_VERSION_CODE
+ .half 0x0203 /* HdrS version */
+root_flags:
+ .half 1
+root_dev:
+ .half 0
+ram_flags:
+ .half 0
+sparc_ramdisk_image:
+ .word 0
+sparc_ramdisk_size:
+ .word 0
+ .word reboot_command
+ .word 0, 0, 0
+ .word _end
+
+/* Cool, here we go. Pick up the romvec pointer in %o0 and stash it in
+ * %g7 and at prom_vector_p. And also quickly check whether we are on
+ * a v0, v2, or v3 prom.
+ */
+gokernel:
+ /* Ok, it's nice to know, as early as possible, if we
+ * are already mapped where we expect to be in virtual
+ * memory. The Solaris /boot elf format bootloader
+ * will peek into our elf header and load us where
+ * we want to be, otherwise we have to re-map.
+ *
+ * Some boot loaders don't place the jmp'rs address
+ * in %o7, so we do a pc-relative call to a local
+ * label, then see what %o7 has.
+ */
+
+ mov %o7, %g4 ! Save %o7
+
+ /* Jump to it, and pray... */
+current_pc:
+ call 1f
+ nop
+
+1:
+ mov %o7, %g3
+
+ tst %o0
+ be no_sun4u_here
+ mov %g4, %o7 /* Previous %o7. */
+
+ mov %o0, %l0 ! stash away romvec
+ mov %o0, %g7 ! put it here too
+ mov %o1, %l1 ! stash away debug_vec too
+
+ /* Ok, let's check out our run time program counter. */
+ set current_pc, %g5
+ cmp %g3, %g5
+ be already_mapped
+ nop
+
+ /* %l6 will hold the offset we have to subtract
+ * from absolute symbols in order to access areas
+ * in our own image. If already mapped this is
+ * just plain zero, else it is KERNBASE.
+ */
+ set KERNBASE, %l6
+ b copy_prom_lvl14
+ nop
+
+already_mapped:
+ mov 0, %l6
+
+ /* Copy over the Prom's level 14 clock handler. */
+copy_prom_lvl14:
+#if 1
+ /* DJHR
+ * preserve our linked/calculated instructions
+ */
+ set lvl14_save, %g1
+ set t_irq14, %g3
+ sub %g1, %l6, %g1 ! translate to physical
+ sub %g3, %l6, %g3 ! translate to physical
+ ldd [%g3], %g4
+ std %g4, [%g1]
+ ldd [%g3+8], %g4
+ std %g4, [%g1+8]
+#endif
+ rd %tbr, %g1
+ andn %g1, 0xfff, %g1 ! proms trap table base
+ or %g0, (0x1e<<4), %g2 ! offset to lvl14 intr
+ or %g1, %g2, %g2
+ set t_irq14, %g3
+ sub %g3, %l6, %g3
+ ldd [%g2], %g4
+ std %g4, [%g3]
+ ldd [%g2 + 0x8], %g4
+ std %g4, [%g3 + 0x8] ! Copy proms handler
+
+/* DON'T TOUCH %l0 thru %l5 in these remapping routines,
+ * we need their values afterwards!
+ */
+
+ /* Now check whether we are already mapped, if we
+ * are we can skip all this garbage coming up.
+ */
+copy_prom_done:
+ cmp %l6, 0
+ be go_to_highmem ! this will be a nop then
+ nop
+
+ /* Validate that we are in fact running on an
+ * SRMMU based cpu.
+ */
+ set 0x4000, %g6
+ cmp %g7, %g6
+ bne not_a_sun4
+ nop
+
+halt_notsup:
+ ld [%g7 + 0x68], %o1
+ set notsup, %o0
+ sub %o0, %l6, %o0
+ call %o1
+ nop
+ ba halt_me
+ nop
+
+not_a_sun4:
+ /* It looks like this is a machine we support.
+ * Now find out what MMU we are dealing with
+ * LEON - identified by the psr.impl field
+ * Viking - identified by the psr.impl field
+ * In all other cases a sun4m srmmu.
+ * We check that the MMU is enabled in all cases.
+ */
+
+ /* Check if this is a LEON CPU */
+ rd %psr, %g3
+ srl %g3, PSR_IMPL_SHIFT, %g3
+ and %g3, PSR_IMPL_SHIFTED_MASK, %g3
+ cmp %g3, PSR_IMPL_LEON
+ be leon_remap /* It is a LEON - jump */
+ nop
+
+ /* Sanity-check, is MMU enabled */
+ lda [%g0] ASI_M_MMUREGS, %g1
+ andcc %g1, 1, %g0
+ be halt_notsup
+ nop
+
+ /* Check for a viking (TI) module. */
+ cmp %g3, PSR_IMPL_TI
+ bne srmmu_not_viking
+ nop
+
+ /* Figure out what kind of viking we are on.
+ * We need to know if we have to play with the
+ * AC bit and disable traps or not.
+ */
+
+ /* I've only seen MicroSparc's on SparcClassics with this
+ * bit set.
+ */
+ set 0x800, %g2
+ lda [%g0] ASI_M_MMUREGS, %g3 ! peek in the control reg
+ and %g2, %g3, %g3
+ subcc %g3, 0x0, %g0
+ bnz srmmu_not_viking ! is in mbus mode
+ nop
+
+ rd %psr, %g3 ! DO NOT TOUCH %g3
+ andn %g3, PSR_ET, %g2
+ wr %g2, 0x0, %psr
+ WRITE_PAUSE
+
+ /* Get context table pointer, then convert to
+ * a physical address, which is 36 bits.
+ */
+ set AC_M_CTPR, %g4
+ lda [%g4] ASI_M_MMUREGS, %g4
+ sll %g4, 0x4, %g4 ! We use this below
+ ! DO NOT TOUCH %g4
+
+ /* Set the AC bit in the Viking's MMU control reg. */
+ lda [%g0] ASI_M_MMUREGS, %g5 ! DO NOT TOUCH %g5
+ set 0x8000, %g6 ! AC bit mask
+ or %g5, %g6, %g6 ! Or it in...
+ sta %g6, [%g0] ASI_M_MMUREGS ! Close your eyes...
+
+ /* Grrr, why does it seem like every other load/store
+ * on the sun4m is in some ASI space...
+ * Fine with me, let's get the pointer to the level 1
+ * page table directory and fetch its entry.
+ */
+ lda [%g4] ASI_M_BYPASS, %o1 ! This is a level 1 ptr
+ srl %o1, 0x4, %o1 ! Clear low 4 bits
+ sll %o1, 0x8, %o1 ! Make physical
+
+ /* Ok, pull in the PTD. */
+ lda [%o1] ASI_M_BYPASS, %o2 ! This is the 0x0 16MB pgd
+
+ /* Calculate to KERNBASE entry. */
+ add %o1, KERNBASE >> (SRMMU_PGDIR_SHIFT - 2), %o3
+
+ /* Poke the entry into the calculated address. */
+ sta %o2, [%o3] ASI_M_BYPASS
+
+ /* I don't get it Sun, if you engineered all these
+ * boot loaders and the PROM (thank you for the debugging
+ * features btw) why did you not have them load kernel
+ * images up in high address space, since this is necessary
+ * for ABI compliance anyways? Does this low-mapping provide
+ * enhanced interoperability?
+ *
+ * "The PROM is the computer."
+ */
+
+ /* Ok, restore the MMU control register we saved in %g5 */
+ sta %g5, [%g0] ASI_M_MMUREGS ! POW... ouch
+
+ /* Turn traps back on. We saved it in %g3 earlier. */
+ wr %g3, 0x0, %psr ! tick tock, tick tock
+
+ /* Now we burn precious CPU cycles due to bad engineering. */
+ WRITE_PAUSE
+
+ /* Wow, all that just to move a 32-bit value from one
+ * place to another... Jump to high memory.
+ */
+ b go_to_highmem
+ nop
+
+srmmu_not_viking:
+ /* This works on viking's in Mbus mode and all
+ * other MBUS modules. It is virtually the same as
+ * the above madness sans turning traps off and flipping
+ * the AC bit.
+ */
+ set AC_M_CTPR, %g1
+ lda [%g1] ASI_M_MMUREGS, %g1 ! get ctx table ptr
+ sll %g1, 0x4, %g1 ! make physical addr
+ lda [%g1] ASI_M_BYPASS, %g1 ! ptr to level 1 pg_table
+ srl %g1, 0x4, %g1
+ sll %g1, 0x8, %g1 ! make phys addr for l1 tbl
+
+ lda [%g1] ASI_M_BYPASS, %g2 ! get level1 entry for 0x0
+ add %g1, KERNBASE >> (SRMMU_PGDIR_SHIFT - 2), %g3
+ sta %g2, [%g3] ASI_M_BYPASS ! place at KERNBASE entry
+ b go_to_highmem
+ nop ! wheee....
+
+
+leon_remap:
+ /* Sanity-check, is MMU enabled */
+ lda [%g0] ASI_LEON_MMUREGS, %g1
+ andcc %g1, 1, %g0
+ be halt_notsup
+ nop
+
+ /* Same code as in the srmmu_not_viking case,
+ * with the LEON ASI for mmuregs
+ */
+ set AC_M_CTPR, %g1
+ lda [%g1] ASI_LEON_MMUREGS, %g1 ! get ctx table ptr
+ sll %g1, 0x4, %g1 ! make physical addr
+ lda [%g1] ASI_M_BYPASS, %g1 ! ptr to level 1 pg_table
+ srl %g1, 0x4, %g1
+ sll %g1, 0x8, %g1 ! make phys addr for l1 tbl
+
+ lda [%g1] ASI_M_BYPASS, %g2 ! get level1 entry for 0x0
+ add %g1, KERNBASE >> (SRMMU_PGDIR_SHIFT - 2), %g3
+ sta %g2, [%g3] ASI_M_BYPASS ! place at KERNBASE entry
+ b go_to_highmem
+ nop ! wheee....
+
+/* Now do a non-relative jump so that PC is in high-memory */
+go_to_highmem:
+ set execute_in_high_mem, %g1
+ jmpl %g1, %g0
+ nop
+
+/* The code above should be at beginning and we have to take care about
+ * short jumps, as branching to .init.text section from .text is usually
+ * impossible */
+ __INIT
+/* Acquire boot time privileged register values, this will help debugging.
+ * I figure out and store nwindows and nwindowsm1 later on.
+ */
+execute_in_high_mem:
+ mov %l0, %o0 ! put back romvec
+ mov %l1, %o1 ! and debug_vec
+
+ sethi %hi(prom_vector_p), %g1
+ st %o0, [%g1 + %lo(prom_vector_p)]
+
+ sethi %hi(linux_dbvec), %g1
+ st %o1, [%g1 + %lo(linux_dbvec)]
+
+ /* Get the machine type via the romvec
+ * getprops node operation
+ */
+ add %g7, 0x1c, %l1
+ ld [%l1], %l0
+ ld [%l0], %l0
+ call %l0
+ or %g0, %g0, %o0 ! next_node(0) = first_node
+ or %o0, %g0, %g6
+
+ sethi %hi(cputypvar), %o1 ! First node has cpu-arch
+ or %o1, %lo(cputypvar), %o1
+ sethi %hi(cputypval), %o2 ! information, the string
+ or %o2, %lo(cputypval), %o2
+ ld [%l1], %l0 ! 'compatible' tells
+ ld [%l0 + 0xc], %l0 ! that we want 'sun4x' where
+ call %l0 ! x is one of 'm', 'd' or 'e'.
+ nop ! %o2 holds pointer
+ ! to a buf where above string
+ ! will get stored by the prom.
+
+
+ /* Check value of "compatible" property.
+ * "value" => "model"
+ * leon => sparc_leon
+ * sun4m => sun4m
+ * sun4s => sun4m
+ * sun4d => sun4d
+ * sun4e => "no_sun4e_here"
+ * '*' => "no_sun4u_here"
+ * Check single letters only
+ */
+
+ set cputypval, %o2
+ /* If cputypval[0] == 'l' (lower case letter L) this is leon */
+ ldub [%o2], %l1
+ cmp %l1, 'l'
+ be leon_init
+ nop
+
+ /* Check cputypval[4] to find the sun model */
+ ldub [%o2 + 0x4], %l1
+
+ cmp %l1, 'm'
+ be sun4m_init
+ cmp %l1, 's'
+ be sun4m_init
+ cmp %l1, 'd'
+ be sun4d_init
+ cmp %l1, 'e'
+ be no_sun4e_here ! Could be a sun4e.
+ nop
+ b no_sun4u_here ! AIEEE, a V9 sun4u... Get our BIG BROTHER kernel :))
+ nop
+
+leon_init:
+ /* LEON CPU - set boot_cpu_id */
+ sethi %hi(boot_cpu_id), %g2 ! boot-cpu index
+
+#ifdef CONFIG_SMP
+ ldub [%g2 + %lo(boot_cpu_id)], %g1
+ cmp %g1, 0xff ! unset means first CPU
+ bne leon_smp_cpu_startup ! continue only with master
+ nop
+#endif
+ /* Get CPU-ID from most significant 4-bit of ASR17 */
+ rd %asr17, %g1
+ srl %g1, 28, %g1
+
+ /* Update boot_cpu_id only on boot cpu */
+ stub %g1, [%g2 + %lo(boot_cpu_id)]
+
+ ba continue_boot
+ nop
+
+/* CPUID in bootbus can be found at PA 0xff0140000 */
+#define SUN4D_BOOTBUS_CPUID 0xf0140000
+
+sun4d_init:
+ /* Need to patch call to handler_irq */
+ set patch_handler_irq, %g4
+ set sun4d_handler_irq, %g5
+ sethi %hi(0x40000000), %g3 ! call
+ sub %g5, %g4, %g5
+ srl %g5, 2, %g5
+ or %g5, %g3, %g5
+ st %g5, [%g4]
+
+#ifdef CONFIG_SMP
+ /* Get our CPU id out of bootbus */
+ set SUN4D_BOOTBUS_CPUID, %g3
+ lduba [%g3] ASI_M_CTL, %g3
+ and %g3, 0xf8, %g3
+ srl %g3, 3, %g4
+ sta %g4, [%g0] ASI_M_VIKING_TMP1
+ sethi %hi(boot_cpu_id), %g5
+ stb %g4, [%g5 + %lo(boot_cpu_id)]
+#endif
+
+ /* Fall through to sun4m_init */
+
+sun4m_init:
+/* Ok, the PROM could have done funny things and apple cider could still
+ * be sitting in the fault status/address registers. Read them all to
+ * clear them so we don't get magic faults later on.
+ */
+/* This sucks, apparently this makes Vikings call prom panic, will fix later */
+2:
+ rd %psr, %o1
+ srl %o1, PSR_IMPL_SHIFT, %o1 ! Get a type of the CPU
+
+ subcc %o1, PSR_IMPL_TI, %g0 ! TI: Viking or MicroSPARC
+ be continue_boot
+ nop
+
+ set AC_M_SFSR, %o0
+ lda [%o0] ASI_M_MMUREGS, %g0
+ set AC_M_SFAR, %o0
+ lda [%o0] ASI_M_MMUREGS, %g0
+
+ /* Fujitsu MicroSPARC-II has no asynchronous flavors of FARs */
+ subcc %o1, 0, %g0
+ be continue_boot
+ nop
+
+ set AC_M_AFSR, %o0
+ lda [%o0] ASI_M_MMUREGS, %g0
+ set AC_M_AFAR, %o0
+ lda [%o0] ASI_M_MMUREGS, %g0
+ nop
+
+
+continue_boot:
+
+/* Aieee, now set PC and nPC, enable traps, give ourselves a stack and it's
+ * show-time!
+ */
+ /* Turn on Supervisor, EnableFloating, and all the PIL bits.
+ * Also puts us in register window zero with traps off.
+ */
+ set (PSR_PS | PSR_S | PSR_PIL | PSR_EF), %g2
+ wr %g2, 0x0, %psr
+ WRITE_PAUSE
+
+ /* I want a kernel stack NOW! */
+ set init_thread_union, %g1
+ set (THREAD_SIZE - STACKFRAME_SZ), %g2
+ add %g1, %g2, %sp
+ mov 0, %fp /* And for good luck */
+
+ /* Zero out our BSS section. */
+ set __bss_start , %o0 ! First address of BSS
+ set _end , %o1 ! Last address of BSS
+ add %o0, 0x1, %o0
+1:
+ stb %g0, [%o0]
+ subcc %o0, %o1, %g0
+ bl 1b
+ add %o0, 0x1, %o0
+
+ /* If boot_cpu_id has not been setup by machine specific
+ * init-code above we default it to zero.
+ */
+ sethi %hi(boot_cpu_id), %g2
+ ldub [%g2 + %lo(boot_cpu_id)], %g3
+ cmp %g3, 0xff
+ bne 1f
+ nop
+ mov %g0, %g3
+ stub %g3, [%g2 + %lo(boot_cpu_id)]
+
+1: sll %g3, 2, %g3
+
+ /* Initialize the uwinmask value for init task just in case.
+ * But first make current_set[boot_cpu_id] point to something useful.
+ */
+ set init_thread_union, %g6
+ set current_set, %g2
+#ifdef CONFIG_SMP
+ st %g6, [%g2]
+ add %g2, %g3, %g2
+#endif
+ st %g6, [%g2]
+
+ st %g0, [%g6 + TI_UWINMASK]
+
+/* Compute NWINDOWS and stash it away. Now uses %wim trick explained
+ * in the V8 manual. Ok, this method seems to work, Sparc is cool...
+ * No, it doesn't work, have to play the save/readCWP/restore trick.
+ */
+
+ wr %g0, 0x0, %wim ! so we do not get a trap
+ WRITE_PAUSE
+
+ save
+
+ rd %psr, %g3
+
+ restore
+
+ and %g3, 0x1f, %g3
+ add %g3, 0x1, %g3
+
+ mov 2, %g1
+ wr %g1, 0x0, %wim ! make window 1 invalid
+ WRITE_PAUSE
+
+ cmp %g3, 0x7
+ bne 2f
+ nop
+
+ /* Adjust our window handling routines to
+ * do things correctly on 7 window Sparcs.
+ */
+
+#define PATCH_INSN(src, dest) \
+ set src, %g5; \
+ set dest, %g2; \
+ ld [%g5], %g4; \
+ st %g4, [%g2];
+
+ /* Patch for window spills... */
+ PATCH_INSN(spnwin_patch1_7win, spnwin_patch1)
+ PATCH_INSN(spnwin_patch2_7win, spnwin_patch2)
+ PATCH_INSN(spnwin_patch3_7win, spnwin_patch3)
+
+ /* Patch for window fills... */
+ PATCH_INSN(fnwin_patch1_7win, fnwin_patch1)
+ PATCH_INSN(fnwin_patch2_7win, fnwin_patch2)
+
+ /* Patch for trap entry setup... */
+ PATCH_INSN(tsetup_7win_patch1, tsetup_patch1)
+ PATCH_INSN(tsetup_7win_patch2, tsetup_patch2)
+ PATCH_INSN(tsetup_7win_patch3, tsetup_patch3)
+ PATCH_INSN(tsetup_7win_patch4, tsetup_patch4)
+ PATCH_INSN(tsetup_7win_patch5, tsetup_patch5)
+ PATCH_INSN(tsetup_7win_patch6, tsetup_patch6)
+
+ /* Patch for returning from traps... */
+ PATCH_INSN(rtrap_7win_patch1, rtrap_patch1)
+ PATCH_INSN(rtrap_7win_patch2, rtrap_patch2)
+ PATCH_INSN(rtrap_7win_patch3, rtrap_patch3)
+ PATCH_INSN(rtrap_7win_patch4, rtrap_patch4)
+ PATCH_INSN(rtrap_7win_patch5, rtrap_patch5)
+
+ /* Patch for killing user windows from the register file. */
+ PATCH_INSN(kuw_patch1_7win, kuw_patch1)
+
+ /* Now patch the kernel window flush sequences.
+ * This saves 2 traps on every switch and fork.
+ */
+ set 0x01000000, %g4
+ set flush_patch_one, %g5
+ st %g4, [%g5 + 0x18]
+ st %g4, [%g5 + 0x1c]
+ set flush_patch_two, %g5
+ st %g4, [%g5 + 0x18]
+ st %g4, [%g5 + 0x1c]
+ set flush_patch_three, %g5
+ st %g4, [%g5 + 0x18]
+ st %g4, [%g5 + 0x1c]
+ set flush_patch_four, %g5
+ st %g4, [%g5 + 0x18]
+ st %g4, [%g5 + 0x1c]
+ set flush_patch_exception, %g5
+ st %g4, [%g5 + 0x18]
+ st %g4, [%g5 + 0x1c]
+ set flush_patch_switch, %g5
+ st %g4, [%g5 + 0x18]
+ st %g4, [%g5 + 0x1c]
+
+2:
+ sethi %hi(nwindows), %g4
+ st %g3, [%g4 + %lo(nwindows)] ! store final value
+ sub %g3, 0x1, %g3
+ sethi %hi(nwindowsm1), %g4
+ st %g3, [%g4 + %lo(nwindowsm1)]
+
+ /* Here we go, start using Linux's trap table... */
+ set trapbase, %g3
+ wr %g3, 0x0, %tbr
+ WRITE_PAUSE
+
+ /* Finally, turn on traps so that we can call c-code. */
+ rd %psr, %g3
+ wr %g3, 0x0, %psr
+ WRITE_PAUSE
+
+ wr %g3, PSR_ET, %psr
+ WRITE_PAUSE
+
+ /* Call sparc32_start_kernel(struct linux_romvec *rp) */
+ sethi %hi(prom_vector_p), %g5
+ ld [%g5 + %lo(prom_vector_p)], %o0
+ call sparc32_start_kernel
+ nop
+
+ /* We should not get here. */
+ call halt_me
+ nop
+
+no_sun4e_here:
+ ld [%g7 + 0x68], %o1
+ set sun4e_notsup, %o0
+ call %o1
+ nop
+ b halt_me
+ nop
+
+ __INITDATA
+
+sun4u_1:
+ .asciz "finddevice"
+ .align 4
+sun4u_2:
+ .asciz "/chosen"
+ .align 4
+sun4u_3:
+ .asciz "getprop"
+ .align 4
+sun4u_4:
+ .asciz "stdout"
+ .align 4
+sun4u_5:
+ .asciz "write"
+ .align 4
+sun4u_6:
+ .asciz "\n\rOn sun4u you have to use sparc64 kernel\n\rand not a sparc32 version\n\r\n\r"
+sun4u_6e:
+ .align 4
+sun4u_7:
+ .asciz "exit"
+ .align 8
+sun4u_a1:
+ .word 0, sun4u_1, 0, 1, 0, 1, 0, sun4u_2, 0
+sun4u_r1:
+ .word 0
+sun4u_a2:
+ .word 0, sun4u_3, 0, 4, 0, 1, 0
+sun4u_i2:
+ .word 0, 0, sun4u_4, 0, sun4u_1, 0, 8, 0
+sun4u_r2:
+ .word 0
+sun4u_a3:
+ .word 0, sun4u_5, 0, 3, 0, 1, 0
+sun4u_i3:
+ .word 0, 0, sun4u_6, 0, sun4u_6e - sun4u_6 - 1, 0
+sun4u_r3:
+ .word 0
+sun4u_a4:
+ .word 0, sun4u_7, 0, 0, 0, 0
+sun4u_r4:
+
+ __INIT
+no_sun4u_here:
+ set sun4u_a1, %o0
+ set current_pc, %l2
+ cmp %l2, %g3
+ be 1f
+ mov %o4, %l0
+ sub %g3, %l2, %l6
+ add %o0, %l6, %o0
+ mov %o0, %l4
+ mov sun4u_r4 - sun4u_a1, %l3
+ ld [%l4], %l5
+2:
+ add %l4, 4, %l4
+ cmp %l5, %l2
+ add %l5, %l6, %l5
+ bgeu,a 3f
+ st %l5, [%l4 - 4]
+3:
+ subcc %l3, 4, %l3
+ bne 2b
+ ld [%l4], %l5
+1:
+ call %l0
+ mov %o0, %l1
+
+ ld [%l1 + (sun4u_r1 - sun4u_a1)], %o1
+ add %l1, (sun4u_a2 - sun4u_a1), %o0
+ call %l0
+ st %o1, [%o0 + (sun4u_i2 - sun4u_a2)]
+
+ ld [%l1 + (sun4u_1 - sun4u_a1)], %o1
+ add %l1, (sun4u_a3 - sun4u_a1), %o0
+ call %l0
+ st %o1, [%o0 + (sun4u_i3 - sun4u_a3)]
+
+ call %l0
+ add %l1, (sun4u_a4 - sun4u_a1), %o0
+
+ /* Not reached */
+halt_me:
+ ld [%g7 + 0x74], %o0
+ call %o0 ! Get us out of here...
+ nop ! Apparently Solaris is better.
+
+/* Ok, now we continue in the .data/.text sections */
+
+ .data
+ .align 4
+
+/*
+ * Fill up the prom vector, note in particular the kind first element,
+ * no joke. I don't need all of them in here as the entire prom vector
+ * gets initialized in c-code so all routines can use it.
+ */
+
+prom_vector_p:
+ .word 0
+
+/* We calculate the following at boot time, window fills/spills and trap entry
+ * code uses these to keep track of the register windows.
+ */
+
+ .align 4
+ .globl nwindows
+ .globl nwindowsm1
+nwindows:
+ .word 8
+nwindowsm1:
+ .word 7
+
+/* Boot time debugger vector value. We need this later on. */
+
+ .align 4
+ .globl linux_dbvec
+linux_dbvec:
+ .word 0
+ .word 0
+
+ .align 8
+
+ .globl lvl14_save
+lvl14_save:
+ .word 0
+ .word 0
+ .word 0
+ .word 0
+ .word t_irq14
+
+ .section ".fixup",#alloc,#execinstr
+ .globl __ret_efault
+__ret_efault:
+ ret
+ restore %g0, -EFAULT, %o0
diff --git a/arch/sparc/kernel/head_64.S b/arch/sparc/kernel/head_64.S
new file mode 100644
index 00000000000..452f04fe8da
--- /dev/null
+++ b/arch/sparc/kernel/head_64.S
@@ -0,0 +1,984 @@
+/* head.S: Initial boot code for the Sparc64 port of Linux.
+ *
+ * Copyright (C) 1996, 1997, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1996 David Sitsky (David.Sitsky@anu.edu.au)
+ * Copyright (C) 1997, 1998 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ * Copyright (C) 1997 Miguel de Icaza (miguel@nuclecu.unam.mx)
+ */
+
+#include <linux/version.h>
+#include <linux/errno.h>
+#include <linux/threads.h>
+#include <linux/init.h>
+#include <linux/linkage.h>
+#include <asm/thread_info.h>
+#include <asm/asi.h>
+#include <asm/pstate.h>
+#include <asm/ptrace.h>
+#include <asm/spitfire.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/errno.h>
+#include <asm/signal.h>
+#include <asm/processor.h>
+#include <asm/lsu.h>
+#include <asm/dcr.h>
+#include <asm/dcu.h>
+#include <asm/head.h>
+#include <asm/ttable.h>
+#include <asm/mmu.h>
+#include <asm/cpudata.h>
+#include <asm/pil.h>
+#include <asm/estate.h>
+#include <asm/sfafsr.h>
+#include <asm/unistd.h>
+
+/* This section from from _start to sparc64_boot_end should fit into
+ * 0x0000000000404000 to 0x0000000000408000.
+ */
+ .text
+ .globl start, _start, stext, _stext
+_start:
+start:
+_stext:
+stext:
+! 0x0000000000404000
+ b sparc64_boot
+ flushw /* Flush register file. */
+
+/* This stuff has to be in sync with SILO and other potential boot loaders
+ * Fields should be kept upward compatible and whenever any change is made,
+ * HdrS version should be incremented.
+ */
+ .global root_flags, ram_flags, root_dev
+ .global sparc_ramdisk_image, sparc_ramdisk_size
+ .global sparc_ramdisk_image64
+
+ .ascii "HdrS"
+ .word LINUX_VERSION_CODE
+
+ /* History:
+ *
+ * 0x0300 : Supports being located at other than 0x4000
+ * 0x0202 : Supports kernel params string
+ * 0x0201 : Supports reboot_command
+ */
+ .half 0x0301 /* HdrS version */
+
+root_flags:
+ .half 1
+root_dev:
+ .half 0
+ram_flags:
+ .half 0
+sparc_ramdisk_image:
+ .word 0
+sparc_ramdisk_size:
+ .word 0
+ .xword reboot_command
+ .xword bootstr_info
+sparc_ramdisk_image64:
+ .xword 0
+ .word _end
+
+ /* PROM cif handler code address is in %o4. */
+sparc64_boot:
+ mov %o4, %l7
+
+ /* We need to remap the kernel. Use position independent
+ * code to remap us to KERNBASE.
+ *
+ * SILO can invoke us with 32-bit address masking enabled,
+ * so make sure that's clear.
+ */
+ rdpr %pstate, %g1
+ andn %g1, PSTATE_AM, %g1
+ wrpr %g1, 0x0, %pstate
+ ba,a,pt %xcc, 1f
+
+ .globl prom_finddev_name, prom_chosen_path, prom_root_node
+ .globl prom_getprop_name, prom_mmu_name, prom_peer_name
+ .globl prom_callmethod_name, prom_translate_name, prom_root_compatible
+ .globl prom_map_name, prom_unmap_name, prom_mmu_ihandle_cache
+ .globl prom_boot_mapped_pc, prom_boot_mapping_mode
+ .globl prom_boot_mapping_phys_high, prom_boot_mapping_phys_low
+ .globl prom_compatible_name, prom_cpu_path, prom_cpu_compatible
+ .globl is_sun4v, sun4v_chip_type, prom_set_trap_table_name
+prom_peer_name:
+ .asciz "peer"
+prom_compatible_name:
+ .asciz "compatible"
+prom_finddev_name:
+ .asciz "finddevice"
+prom_chosen_path:
+ .asciz "/chosen"
+prom_cpu_path:
+ .asciz "/cpu"
+prom_getprop_name:
+ .asciz "getprop"
+prom_mmu_name:
+ .asciz "mmu"
+prom_callmethod_name:
+ .asciz "call-method"
+prom_translate_name:
+ .asciz "translate"
+prom_map_name:
+ .asciz "map"
+prom_unmap_name:
+ .asciz "unmap"
+prom_set_trap_table_name:
+ .asciz "SUNW,set-trap-table"
+prom_sun4v_name:
+ .asciz "sun4v"
+prom_niagara_prefix:
+ .asciz "SUNW,UltraSPARC-T"
+prom_sparc_prefix:
+ .asciz "SPARC-"
+prom_sparc64x_prefix:
+ .asciz "SPARC64-X"
+ .align 4
+prom_root_compatible:
+ .skip 64
+prom_cpu_compatible:
+ .skip 64
+prom_root_node:
+ .word 0
+prom_mmu_ihandle_cache:
+ .word 0
+prom_boot_mapped_pc:
+ .word 0
+prom_boot_mapping_mode:
+ .word 0
+ .align 8
+prom_boot_mapping_phys_high:
+ .xword 0
+prom_boot_mapping_phys_low:
+ .xword 0
+is_sun4v:
+ .word 0
+sun4v_chip_type:
+ .word SUN4V_CHIP_INVALID
+1:
+ rd %pc, %l0
+
+ mov (1b - prom_peer_name), %l1
+ sub %l0, %l1, %l1
+ mov 0, %l2
+
+ /* prom_root_node = prom_peer(0) */
+ stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "peer"
+ mov 1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 1
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1
+ stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1, 0
+ stx %g0, [%sp + 2047 + 128 + 0x20] ! ret1
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ ldx [%sp + 2047 + 128 + 0x20], %l4 ! prom root node
+ mov (1b - prom_root_node), %l1
+ sub %l0, %l1, %l1
+ stw %l4, [%l1]
+
+ mov (1b - prom_getprop_name), %l1
+ mov (1b - prom_compatible_name), %l2
+ mov (1b - prom_root_compatible), %l5
+ sub %l0, %l1, %l1
+ sub %l0, %l2, %l2
+ sub %l0, %l5, %l5
+
+ /* prom_getproperty(prom_root_node, "compatible",
+ * &prom_root_compatible, 64)
+ */
+ stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "getprop"
+ mov 4, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 4
+ mov 1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1
+ stx %l4, [%sp + 2047 + 128 + 0x18] ! arg1, prom_root_node
+ stx %l2, [%sp + 2047 + 128 + 0x20] ! arg2, "compatible"
+ stx %l5, [%sp + 2047 + 128 + 0x28] ! arg3, &prom_root_compatible
+ mov 64, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4, size
+ stx %g0, [%sp + 2047 + 128 + 0x38] ! ret1
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ mov (1b - prom_finddev_name), %l1
+ mov (1b - prom_chosen_path), %l2
+ mov (1b - prom_boot_mapped_pc), %l3
+ sub %l0, %l1, %l1
+ sub %l0, %l2, %l2
+ sub %l0, %l3, %l3
+ stw %l0, [%l3]
+ sub %sp, (192 + 128), %sp
+
+ /* chosen_node = prom_finddevice("/chosen") */
+ stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "finddevice"
+ mov 1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 1
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1
+ stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1, "/chosen"
+ stx %g0, [%sp + 2047 + 128 + 0x20] ! ret1
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ ldx [%sp + 2047 + 128 + 0x20], %l4 ! chosen device node
+
+ mov (1b - prom_getprop_name), %l1
+ mov (1b - prom_mmu_name), %l2
+ mov (1b - prom_mmu_ihandle_cache), %l5
+ sub %l0, %l1, %l1
+ sub %l0, %l2, %l2
+ sub %l0, %l5, %l5
+
+ /* prom_mmu_ihandle_cache = prom_getint(chosen_node, "mmu") */
+ stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "getprop"
+ mov 4, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 4
+ mov 1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1
+ stx %l4, [%sp + 2047 + 128 + 0x18] ! arg1, chosen_node
+ stx %l2, [%sp + 2047 + 128 + 0x20] ! arg2, "mmu"
+ stx %l5, [%sp + 2047 + 128 + 0x28] ! arg3, &prom_mmu_ihandle_cache
+ mov 4, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4, sizeof(arg3)
+ stx %g0, [%sp + 2047 + 128 + 0x38] ! ret1
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ mov (1b - prom_callmethod_name), %l1
+ mov (1b - prom_translate_name), %l2
+ sub %l0, %l1, %l1
+ sub %l0, %l2, %l2
+ lduw [%l5], %l5 ! prom_mmu_ihandle_cache
+
+ stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "call-method"
+ mov 3, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 3
+ mov 5, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 5
+ stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1: "translate"
+ stx %l5, [%sp + 2047 + 128 + 0x20] ! arg2: prom_mmu_ihandle_cache
+ /* PAGE align */
+ srlx %l0, 13, %l3
+ sllx %l3, 13, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x28] ! arg3: vaddr, our PC
+ stx %g0, [%sp + 2047 + 128 + 0x30] ! res1
+ stx %g0, [%sp + 2047 + 128 + 0x38] ! res2
+ stx %g0, [%sp + 2047 + 128 + 0x40] ! res3
+ stx %g0, [%sp + 2047 + 128 + 0x48] ! res4
+ stx %g0, [%sp + 2047 + 128 + 0x50] ! res5
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ ldx [%sp + 2047 + 128 + 0x40], %l1 ! translation mode
+ mov (1b - prom_boot_mapping_mode), %l4
+ sub %l0, %l4, %l4
+ stw %l1, [%l4]
+ mov (1b - prom_boot_mapping_phys_high), %l4
+ sub %l0, %l4, %l4
+ ldx [%sp + 2047 + 128 + 0x48], %l2 ! physaddr high
+ stx %l2, [%l4 + 0x0]
+ ldx [%sp + 2047 + 128 + 0x50], %l3 ! physaddr low
+ /* 4MB align */
+ srlx %l3, ILOG2_4MB, %l3
+ sllx %l3, ILOG2_4MB, %l3
+ stx %l3, [%l4 + 0x8]
+
+ /* Leave service as-is, "call-method" */
+ mov 7, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 7
+ mov 1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1
+ mov (1b - prom_map_name), %l3
+ sub %l0, %l3, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x18] ! arg1: "map"
+ /* Leave arg2 as-is, prom_mmu_ihandle_cache */
+ mov -1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x28] ! arg3: mode (-1 default)
+ /* 4MB align the kernel image size. */
+ set (_end - KERNBASE), %l3
+ set ((4 * 1024 * 1024) - 1), %l4
+ add %l3, %l4, %l3
+ andn %l3, %l4, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4: roundup(ksize, 4MB)
+ sethi %hi(KERNBASE), %l3
+ stx %l3, [%sp + 2047 + 128 + 0x38] ! arg5: vaddr (KERNBASE)
+ stx %g0, [%sp + 2047 + 128 + 0x40] ! arg6: empty
+ mov (1b - prom_boot_mapping_phys_low), %l3
+ sub %l0, %l3, %l3
+ ldx [%l3], %l3
+ stx %l3, [%sp + 2047 + 128 + 0x48] ! arg7: phys addr
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ add %sp, (192 + 128), %sp
+
+ sethi %hi(prom_root_compatible), %g1
+ or %g1, %lo(prom_root_compatible), %g1
+ sethi %hi(prom_sun4v_name), %g7
+ or %g7, %lo(prom_sun4v_name), %g7
+ mov 5, %g3
+90: ldub [%g7], %g2
+ ldub [%g1], %g4
+ cmp %g2, %g4
+ bne,pn %icc, 80f
+ add %g7, 1, %g7
+ subcc %g3, 1, %g3
+ bne,pt %xcc, 90b
+ add %g1, 1, %g1
+
+ sethi %hi(is_sun4v), %g1
+ or %g1, %lo(is_sun4v), %g1
+ mov 1, %g7
+ stw %g7, [%g1]
+
+ /* cpu_node = prom_finddevice("/cpu") */
+ mov (1b - prom_finddev_name), %l1
+ mov (1b - prom_cpu_path), %l2
+ sub %l0, %l1, %l1
+ sub %l0, %l2, %l2
+ sub %sp, (192 + 128), %sp
+
+ stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "finddevice"
+ mov 1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 1
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1
+ stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1, "/cpu"
+ stx %g0, [%sp + 2047 + 128 + 0x20] ! ret1
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ ldx [%sp + 2047 + 128 + 0x20], %l4 ! cpu device node
+
+ mov (1b - prom_getprop_name), %l1
+ mov (1b - prom_compatible_name), %l2
+ mov (1b - prom_cpu_compatible), %l5
+ sub %l0, %l1, %l1
+ sub %l0, %l2, %l2
+ sub %l0, %l5, %l5
+
+ /* prom_getproperty(cpu_node, "compatible",
+ * &prom_cpu_compatible, 64)
+ */
+ stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "getprop"
+ mov 4, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 4
+ mov 1, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1
+ stx %l4, [%sp + 2047 + 128 + 0x18] ! arg1, cpu_node
+ stx %l2, [%sp + 2047 + 128 + 0x20] ! arg2, "compatible"
+ stx %l5, [%sp + 2047 + 128 + 0x28] ! arg3, &prom_cpu_compatible
+ mov 64, %l3
+ stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4, size
+ stx %g0, [%sp + 2047 + 128 + 0x38] ! ret1
+ call %l7
+ add %sp, (2047 + 128), %o0 ! argument array
+
+ add %sp, (192 + 128), %sp
+
+ sethi %hi(prom_cpu_compatible), %g1
+ or %g1, %lo(prom_cpu_compatible), %g1
+ sethi %hi(prom_niagara_prefix), %g7
+ or %g7, %lo(prom_niagara_prefix), %g7
+ mov 17, %g3
+90: ldub [%g7], %g2
+ ldub [%g1], %g4
+ cmp %g2, %g4
+ bne,pn %icc, 89f
+ add %g7, 1, %g7
+ subcc %g3, 1, %g3
+ bne,pt %xcc, 90b
+ add %g1, 1, %g1
+ ba,pt %xcc, 91f
+ nop
+
+89: sethi %hi(prom_cpu_compatible), %g1
+ or %g1, %lo(prom_cpu_compatible), %g1
+ sethi %hi(prom_sparc_prefix), %g7
+ or %g7, %lo(prom_sparc_prefix), %g7
+ mov 6, %g3
+90: ldub [%g7], %g2
+ ldub [%g1], %g4
+ cmp %g2, %g4
+ bne,pn %icc, 4f
+ add %g7, 1, %g7
+ subcc %g3, 1, %g3
+ bne,pt %xcc, 90b
+ add %g1, 1, %g1
+
+ sethi %hi(prom_cpu_compatible), %g1
+ or %g1, %lo(prom_cpu_compatible), %g1
+ ldub [%g1 + 6], %g2
+ cmp %g2, 'T'
+ be,pt %xcc, 70f
+ cmp %g2, 'M'
+ bne,pn %xcc, 49f
+ nop
+
+70: ldub [%g1 + 7], %g2
+ cmp %g2, '3'
+ be,pt %xcc, 5f
+ mov SUN4V_CHIP_NIAGARA3, %g4
+ cmp %g2, '4'
+ be,pt %xcc, 5f
+ mov SUN4V_CHIP_NIAGARA4, %g4
+ cmp %g2, '5'
+ be,pt %xcc, 5f
+ mov SUN4V_CHIP_NIAGARA5, %g4
+ ba,pt %xcc, 49f
+ nop
+
+91: sethi %hi(prom_cpu_compatible), %g1
+ or %g1, %lo(prom_cpu_compatible), %g1
+ ldub [%g1 + 17], %g2
+ cmp %g2, '1'
+ be,pt %xcc, 5f
+ mov SUN4V_CHIP_NIAGARA1, %g4
+ cmp %g2, '2'
+ be,pt %xcc, 5f
+ mov SUN4V_CHIP_NIAGARA2, %g4
+
+4:
+ /* Athena */
+ sethi %hi(prom_cpu_compatible), %g1
+ or %g1, %lo(prom_cpu_compatible), %g1
+ sethi %hi(prom_sparc64x_prefix), %g7
+ or %g7, %lo(prom_sparc64x_prefix), %g7
+ mov 9, %g3
+41: ldub [%g7], %g2
+ ldub [%g1], %g4
+ cmp %g2, %g4
+ bne,pn %icc, 49f
+ add %g7, 1, %g7
+ subcc %g3, 1, %g3
+ bne,pt %xcc, 41b
+ add %g1, 1, %g1
+ mov SUN4V_CHIP_SPARC64X, %g4
+ ba,pt %xcc, 5f
+ nop
+
+49:
+ mov SUN4V_CHIP_UNKNOWN, %g4
+5: sethi %hi(sun4v_chip_type), %g2
+ or %g2, %lo(sun4v_chip_type), %g2
+ stw %g4, [%g2]
+
+80:
+ BRANCH_IF_SUN4V(g1, jump_to_sun4u_init)
+ BRANCH_IF_CHEETAH_BASE(g1,g7,cheetah_boot)
+ BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,cheetah_plus_boot)
+ ba,pt %xcc, spitfire_boot
+ nop
+
+cheetah_plus_boot:
+ /* Preserve OBP chosen DCU and DCR register settings. */
+ ba,pt %xcc, cheetah_generic_boot
+ nop
+
+cheetah_boot:
+ mov DCR_BPE | DCR_RPE | DCR_SI | DCR_IFPOE | DCR_MS, %g1
+ wr %g1, %asr18
+
+ sethi %uhi(DCU_ME|DCU_RE|DCU_HPE|DCU_SPE|DCU_SL|DCU_WE), %g7
+ or %g7, %ulo(DCU_ME|DCU_RE|DCU_HPE|DCU_SPE|DCU_SL|DCU_WE), %g7
+ sllx %g7, 32, %g7
+ or %g7, DCU_DM | DCU_IM | DCU_DC | DCU_IC, %g7
+ stxa %g7, [%g0] ASI_DCU_CONTROL_REG
+ membar #Sync
+
+cheetah_generic_boot:
+ mov TSB_EXTENSION_P, %g3
+ stxa %g0, [%g3] ASI_DMMU
+ stxa %g0, [%g3] ASI_IMMU
+ membar #Sync
+
+ mov TSB_EXTENSION_S, %g3
+ stxa %g0, [%g3] ASI_DMMU
+ membar #Sync
+
+ mov TSB_EXTENSION_N, %g3
+ stxa %g0, [%g3] ASI_DMMU
+ stxa %g0, [%g3] ASI_IMMU
+ membar #Sync
+
+ ba,a,pt %xcc, jump_to_sun4u_init
+
+spitfire_boot:
+ /* Typically PROM has already enabled both MMU's and both on-chip
+ * caches, but we do it here anyway just to be paranoid.
+ */
+ mov (LSU_CONTROL_IC|LSU_CONTROL_DC|LSU_CONTROL_IM|LSU_CONTROL_DM), %g1
+ stxa %g1, [%g0] ASI_LSU_CONTROL
+ membar #Sync
+
+jump_to_sun4u_init:
+ /*
+ * Make sure we are in privileged mode, have address masking,
+ * using the ordinary globals and have enabled floating
+ * point.
+ *
+ * Again, typically PROM has left %pil at 13 or similar, and
+ * (PSTATE_PRIV | PSTATE_PEF | PSTATE_IE) in %pstate.
+ */
+ wrpr %g0, (PSTATE_PRIV|PSTATE_PEF|PSTATE_IE), %pstate
+ wr %g0, 0, %fprs
+
+ set sun4u_init, %g2
+ jmpl %g2 + %g0, %g0
+ nop
+
+ __REF
+sun4u_init:
+ BRANCH_IF_SUN4V(g1, sun4v_init)
+
+ /* Set ctx 0 */
+ mov PRIMARY_CONTEXT, %g7
+ stxa %g0, [%g7] ASI_DMMU
+ membar #Sync
+
+ mov SECONDARY_CONTEXT, %g7
+ stxa %g0, [%g7] ASI_DMMU
+ membar #Sync
+
+ ba,pt %xcc, sun4u_continue
+ nop
+
+sun4v_init:
+ /* Set ctx 0 */
+ mov PRIMARY_CONTEXT, %g7
+ stxa %g0, [%g7] ASI_MMU
+ membar #Sync
+
+ mov SECONDARY_CONTEXT, %g7
+ stxa %g0, [%g7] ASI_MMU
+ membar #Sync
+ ba,pt %xcc, niagara_tlb_fixup
+ nop
+
+sun4u_continue:
+ BRANCH_IF_ANY_CHEETAH(g1, g7, cheetah_tlb_fixup)
+
+ ba,pt %xcc, spitfire_tlb_fixup
+ nop
+
+niagara_tlb_fixup:
+ mov 3, %g2 /* Set TLB type to hypervisor. */
+ sethi %hi(tlb_type), %g1
+ stw %g2, [%g1 + %lo(tlb_type)]
+
+ /* Patch copy/clear ops. */
+ sethi %hi(sun4v_chip_type), %g1
+ lduw [%g1 + %lo(sun4v_chip_type)], %g1
+ cmp %g1, SUN4V_CHIP_NIAGARA1
+ be,pt %xcc, niagara_patch
+ cmp %g1, SUN4V_CHIP_NIAGARA2
+ be,pt %xcc, niagara2_patch
+ nop
+ cmp %g1, SUN4V_CHIP_NIAGARA3
+ be,pt %xcc, niagara2_patch
+ nop
+ cmp %g1, SUN4V_CHIP_NIAGARA4
+ be,pt %xcc, niagara4_patch
+ nop
+ cmp %g1, SUN4V_CHIP_NIAGARA5
+ be,pt %xcc, niagara4_patch
+ nop
+
+ call generic_patch_copyops
+ nop
+ call generic_patch_bzero
+ nop
+ call generic_patch_pageops
+ nop
+
+ ba,a,pt %xcc, 80f
+niagara4_patch:
+ call niagara4_patch_copyops
+ nop
+ call niagara4_patch_bzero
+ nop
+ call niagara4_patch_pageops
+ nop
+
+ ba,a,pt %xcc, 80f
+
+niagara2_patch:
+ call niagara2_patch_copyops
+ nop
+ call niagara_patch_bzero
+ nop
+ call niagara_patch_pageops
+ nop
+
+ ba,a,pt %xcc, 80f
+
+niagara_patch:
+ call niagara_patch_copyops
+ nop
+ call niagara_patch_bzero
+ nop
+ call niagara_patch_pageops
+ nop
+
+80:
+ /* Patch TLB/cache ops. */
+ call hypervisor_patch_cachetlbops
+ nop
+
+ ba,pt %xcc, tlb_fixup_done
+ nop
+
+cheetah_tlb_fixup:
+ mov 2, %g2 /* Set TLB type to cheetah+. */
+ BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,1f)
+
+ mov 1, %g2 /* Set TLB type to cheetah. */
+
+1: sethi %hi(tlb_type), %g1
+ stw %g2, [%g1 + %lo(tlb_type)]
+
+ /* Patch copy/page operations to cheetah optimized versions. */
+ call cheetah_patch_copyops
+ nop
+ call cheetah_patch_copy_page
+ nop
+ call cheetah_patch_cachetlbops
+ nop
+
+ ba,pt %xcc, tlb_fixup_done
+ nop
+
+spitfire_tlb_fixup:
+ /* Set TLB type to spitfire. */
+ mov 0, %g2
+ sethi %hi(tlb_type), %g1
+ stw %g2, [%g1 + %lo(tlb_type)]
+
+tlb_fixup_done:
+ sethi %hi(init_thread_union), %g6
+ or %g6, %lo(init_thread_union), %g6
+ ldx [%g6 + TI_TASK], %g4
+ mov %sp, %l6
+
+ wr %g0, ASI_P, %asi
+ mov 1, %g1
+ sllx %g1, THREAD_SHIFT, %g1
+ sub %g1, (STACKFRAME_SZ + STACK_BIAS), %g1
+ add %g6, %g1, %sp
+ mov 0, %fp
+
+ /* Set per-cpu pointer initially to zero, this makes
+ * the boot-cpu use the in-kernel-image per-cpu areas
+ * before setup_per_cpu_area() is invoked.
+ */
+ clr %g5
+
+ wrpr %g0, 0, %wstate
+ wrpr %g0, 0x0, %tl
+
+ /* Clear the bss */
+ sethi %hi(__bss_start), %o0
+ or %o0, %lo(__bss_start), %o0
+ sethi %hi(_end), %o1
+ or %o1, %lo(_end), %o1
+ call __bzero
+ sub %o1, %o0, %o1
+
+#ifdef CONFIG_LOCKDEP
+ /* We have this call this super early, as even prom_init can grab
+ * spinlocks and thus call into the lockdep code.
+ */
+ call lockdep_init
+ nop
+#endif
+
+ mov %l6, %o1 ! OpenPROM stack
+ call prom_init
+ mov %l7, %o0 ! OpenPROM cif handler
+
+ /* Initialize current_thread_info()->cpu as early as possible.
+ * In order to do that accurately we have to patch up the get_cpuid()
+ * assembler sequences. And that, in turn, requires that we know
+ * if we are on a Starfire box or not. While we're here, patch up
+ * the sun4v sequences as well.
+ */
+ call check_if_starfire
+ nop
+ call per_cpu_patch
+ nop
+ call sun4v_patch
+ nop
+
+#ifdef CONFIG_SMP
+ call hard_smp_processor_id
+ nop
+ cmp %o0, NR_CPUS
+ blu,pt %xcc, 1f
+ nop
+ call boot_cpu_id_too_large
+ nop
+ /* Not reached... */
+
+1:
+#else
+ mov 0, %o0
+#endif
+ sth %o0, [%g6 + TI_CPU]
+
+ call prom_init_report
+ nop
+
+ /* Off we go.... */
+ call start_kernel
+ nop
+ /* Not reached... */
+
+ .previous
+
+ /* This is meant to allow the sharing of this code between
+ * boot processor invocation (via setup_tba() below) and
+ * secondary processor startup (via trampoline.S). The
+ * former does use this code, the latter does not yet due
+ * to some complexities. That should be fixed up at some
+ * point.
+ *
+ * There used to be enormous complexity wrt. transferring
+ * over from the firmware's trap table to the Linux kernel's.
+ * For example, there was a chicken & egg problem wrt. building
+ * the OBP page tables, yet needing to be on the Linux kernel
+ * trap table (to translate PAGE_OFFSET addresses) in order to
+ * do that.
+ *
+ * We now handle OBP tlb misses differently, via linear lookups
+ * into the prom_trans[] array. So that specific problem no
+ * longer exists. Yet, unfortunately there are still some issues
+ * preventing trampoline.S from using this code... ho hum.
+ */
+ .globl setup_trap_table
+setup_trap_table:
+ save %sp, -192, %sp
+
+ /* Force interrupts to be disabled. */
+ rdpr %pstate, %l0
+ andn %l0, PSTATE_IE, %o1
+ wrpr %o1, 0x0, %pstate
+ rdpr %pil, %l1
+ wrpr %g0, PIL_NORMAL_MAX, %pil
+
+ /* Make the firmware call to jump over to the Linux trap table. */
+ sethi %hi(is_sun4v), %o0
+ lduw [%o0 + %lo(is_sun4v)], %o0
+ brz,pt %o0, 1f
+ nop
+
+ TRAP_LOAD_TRAP_BLOCK(%g2, %g3)
+ add %g2, TRAP_PER_CPU_FAULT_INFO, %g2
+ stxa %g2, [%g0] ASI_SCRATCHPAD
+
+ /* Compute physical address:
+ *
+ * paddr = kern_base + (mmfsa_vaddr - KERNBASE)
+ */
+ sethi %hi(KERNBASE), %g3
+ sub %g2, %g3, %g2
+ sethi %hi(kern_base), %g3
+ ldx [%g3 + %lo(kern_base)], %g3
+ add %g2, %g3, %o1
+ sethi %hi(sparc64_ttable_tl0), %o0
+
+ set prom_set_trap_table_name, %g2
+ stx %g2, [%sp + 2047 + 128 + 0x00]
+ mov 2, %g2
+ stx %g2, [%sp + 2047 + 128 + 0x08]
+ mov 0, %g2
+ stx %g2, [%sp + 2047 + 128 + 0x10]
+ stx %o0, [%sp + 2047 + 128 + 0x18]
+ stx %o1, [%sp + 2047 + 128 + 0x20]
+ sethi %hi(p1275buf), %g2
+ or %g2, %lo(p1275buf), %g2
+ ldx [%g2 + 0x08], %o1
+ call %o1
+ add %sp, (2047 + 128), %o0
+
+ ba,pt %xcc, 2f
+ nop
+
+1: sethi %hi(sparc64_ttable_tl0), %o0
+ set prom_set_trap_table_name, %g2
+ stx %g2, [%sp + 2047 + 128 + 0x00]
+ mov 1, %g2
+ stx %g2, [%sp + 2047 + 128 + 0x08]
+ mov 0, %g2
+ stx %g2, [%sp + 2047 + 128 + 0x10]
+ stx %o0, [%sp + 2047 + 128 + 0x18]
+ sethi %hi(p1275buf), %g2
+ or %g2, %lo(p1275buf), %g2
+ ldx [%g2 + 0x08], %o1
+ call %o1
+ add %sp, (2047 + 128), %o0
+
+ /* Start using proper page size encodings in ctx register. */
+2: sethi %hi(sparc64_kern_pri_context), %g3
+ ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2
+
+ mov PRIMARY_CONTEXT, %g1
+
+661: stxa %g2, [%g1] ASI_DMMU
+ .section .sun4v_1insn_patch, "ax"
+ .word 661b
+ stxa %g2, [%g1] ASI_MMU
+ .previous
+
+ membar #Sync
+
+ BRANCH_IF_SUN4V(o2, 1f)
+
+ /* Kill PROM timer */
+ sethi %hi(0x80000000), %o2
+ sllx %o2, 32, %o2
+ wr %o2, 0, %tick_cmpr
+
+ BRANCH_IF_ANY_CHEETAH(o2, o3, 1f)
+
+ ba,pt %xcc, 2f
+ nop
+
+ /* Disable STICK_INT interrupts. */
+1:
+ sethi %hi(0x80000000), %o2
+ sllx %o2, 32, %o2
+ wr %o2, %asr25
+
+2:
+ wrpr %g0, %g0, %wstate
+
+ call init_irqwork_curcpu
+ nop
+
+ /* Now we can restore interrupt state. */
+ wrpr %l0, 0, %pstate
+ wrpr %l1, 0x0, %pil
+
+ ret
+ restore
+
+ .globl setup_tba
+setup_tba:
+ save %sp, -192, %sp
+
+ /* The boot processor is the only cpu which invokes this
+ * routine, the other cpus set things up via trampoline.S.
+ * So save the OBP trap table address here.
+ */
+ rdpr %tba, %g7
+ sethi %hi(prom_tba), %o1
+ or %o1, %lo(prom_tba), %o1
+ stx %g7, [%o1]
+
+ call setup_trap_table
+ nop
+
+ ret
+ restore
+sparc64_boot_end:
+
+#include "etrap_64.S"
+#include "rtrap_64.S"
+#include "winfixup.S"
+#include "fpu_traps.S"
+#include "ivec.S"
+#include "getsetcc.S"
+#include "utrap.S"
+#include "spiterrs.S"
+#include "cherrs.S"
+#include "misctrap.S"
+#include "syscalls.S"
+#include "helpers.S"
+#include "hvcalls.S"
+#include "sun4v_tlb_miss.S"
+#include "sun4v_ivec.S"
+#include "ktlb.S"
+#include "tsb.S"
+
+/*
+ * The following skip makes sure the trap table in ttable.S is aligned
+ * on a 32K boundary as required by the v9 specs for TBA register.
+ *
+ * We align to a 32K boundary, then we have the 32K kernel TSB,
+ * the 64K kernel 4MB TSB, and then the 32K aligned trap table.
+ */
+1:
+ .skip 0x4000 + _start - 1b
+
+! 0x0000000000408000
+
+ .globl swapper_tsb
+swapper_tsb:
+ .skip (32 * 1024)
+
+ .globl swapper_4m_tsb
+swapper_4m_tsb:
+ .skip (64 * 1024)
+
+! 0x0000000000420000
+
+ /* Some care needs to be exercised if you try to move the
+ * location of the trap table relative to other things. For
+ * one thing there are br* instructions in some of the
+ * trap table entires which branch back to code in ktlb.S
+ * Those instructions can only handle a signed 16-bit
+ * displacement.
+ *
+ * There is a binutils bug (bugzilla #4558) which causes
+ * the relocation overflow checks for such instructions to
+ * not be done correctly. So bintuils will not notice the
+ * error and will instead write junk into the relocation and
+ * you'll have an unbootable kernel.
+ */
+#include "ttable_64.S"
+
+! 0x0000000000428000
+
+#include "systbls_64.S"
+
+ .data
+ .align 8
+ .globl prom_tba, tlb_type
+prom_tba: .xword 0
+tlb_type: .word 0 /* Must NOT end up in BSS */
+ .section ".fixup",#alloc,#execinstr
+
+ .globl __ret_efault, __retl_efault, __ret_one, __retl_one
+ENTRY(__ret_efault)
+ ret
+ restore %g0, -EFAULT, %o0
+ENDPROC(__ret_efault)
+
+ENTRY(__retl_efault)
+ retl
+ mov -EFAULT, %o0
+ENDPROC(__retl_efault)
+
+ENTRY(__retl_one)
+ retl
+ mov 1, %o0
+ENDPROC(__retl_one)
+
+ENTRY(__ret_one_asi)
+ wr %g0, ASI_AIUS, %asi
+ ret
+ restore %g0, 1, %o0
+ENDPROC(__ret_one_asi)
+
+ENTRY(__retl_one_asi)
+ wr %g0, ASI_AIUS, %asi
+ retl
+ mov 1, %o0
+ENDPROC(__retl_one_asi)
+
+ENTRY(__retl_o1)
+ retl
+ mov %o1, %o0
+ENDPROC(__retl_o1)
diff --git a/arch/sparc/kernel/helpers.S b/arch/sparc/kernel/helpers.S
new file mode 100644
index 00000000000..314dd0c9fc5
--- /dev/null
+++ b/arch/sparc/kernel/helpers.S
@@ -0,0 +1,63 @@
+ .align 32
+ .globl __flushw_user
+ .type __flushw_user,#function
+__flushw_user:
+ rdpr %otherwin, %g1
+ brz,pn %g1, 2f
+ clr %g2
+1: save %sp, -128, %sp
+ rdpr %otherwin, %g1
+ brnz,pt %g1, 1b
+ add %g2, 1, %g2
+1: sub %g2, 1, %g2
+ brnz,pt %g2, 1b
+ restore %g0, %g0, %g0
+2: retl
+ nop
+ .size __flushw_user,.-__flushw_user
+
+ /* Flush %fp and %i7 to the stack for all register
+ * windows active inside of the cpu. This allows
+ * show_stack_trace() to avoid using an expensive
+ * 'flushw'.
+ */
+ .globl stack_trace_flush
+ .type stack_trace_flush,#function
+stack_trace_flush:
+ rdpr %pstate, %o0
+ wrpr %o0, PSTATE_IE, %pstate
+
+ rdpr %cwp, %g1
+ rdpr %canrestore, %g2
+ sub %g1, 1, %g3
+
+1: brz,pn %g2, 2f
+ sub %g2, 1, %g2
+ wrpr %g3, %cwp
+ stx %fp, [%sp + STACK_BIAS + RW_V9_I6]
+ stx %i7, [%sp + STACK_BIAS + RW_V9_I7]
+ ba,pt %xcc, 1b
+ sub %g3, 1, %g3
+
+2: wrpr %g1, %cwp
+ wrpr %o0, %pstate
+
+ retl
+ nop
+ .size stack_trace_flush,.-stack_trace_flush
+
+#ifdef CONFIG_SMP
+ .globl hard_smp_processor_id
+ .type hard_smp_processor_id,#function
+hard_smp_processor_id:
+#endif
+ .globl real_hard_smp_processor_id
+ .type real_hard_smp_processor_id,#function
+real_hard_smp_processor_id:
+ __GET_CPUID(%o0)
+ retl
+ nop
+#ifdef CONFIG_SMP
+ .size hard_smp_processor_id,.-hard_smp_processor_id
+#endif
+ .size real_hard_smp_processor_id,.-real_hard_smp_processor_id
diff --git a/arch/sparc/kernel/hvapi.c b/arch/sparc/kernel/hvapi.c
new file mode 100644
index 00000000000..c0a2de0fd62
--- /dev/null
+++ b/arch/sparc/kernel/hvapi.c
@@ -0,0 +1,200 @@
+/* hvapi.c: Hypervisor API management.
+ *
+ * Copyright (C) 2007 David S. Miller <davem@davemloft.net>
+ */
+#include <linux/kernel.h>
+#include <linux/export.h>
+#include <linux/init.h>
+
+#include <asm/hypervisor.h>
+#include <asm/oplib.h>
+
+/* If the hypervisor indicates that the API setting
+ * calls are unsupported, by returning HV_EBADTRAP or
+ * HV_ENOTSUPPORTED, we assume that API groups with the
+ * PRE_API flag set are major 1 minor 0.
+ */
+struct api_info {
+ unsigned long group;
+ unsigned long major;
+ unsigned long minor;
+ unsigned int refcnt;
+ unsigned int flags;
+#define FLAG_PRE_API 0x00000001
+};
+
+static struct api_info api_table[] = {
+ { .group = HV_GRP_SUN4V, .flags = FLAG_PRE_API },
+ { .group = HV_GRP_CORE, .flags = FLAG_PRE_API },
+ { .group = HV_GRP_INTR, },
+ { .group = HV_GRP_SOFT_STATE, },
+ { .group = HV_GRP_TM, },
+ { .group = HV_GRP_PCI, .flags = FLAG_PRE_API },
+ { .group = HV_GRP_LDOM, },
+ { .group = HV_GRP_SVC_CHAN, .flags = FLAG_PRE_API },
+ { .group = HV_GRP_NCS, .flags = FLAG_PRE_API },
+ { .group = HV_GRP_RNG, },
+ { .group = HV_GRP_PBOOT, },
+ { .group = HV_GRP_TPM, },
+ { .group = HV_GRP_SDIO, },
+ { .group = HV_GRP_SDIO_ERR, },
+ { .group = HV_GRP_REBOOT_DATA, },
+ { .group = HV_GRP_NIAG_PERF, .flags = FLAG_PRE_API },
+ { .group = HV_GRP_FIRE_PERF, },
+ { .group = HV_GRP_N2_CPU, },
+ { .group = HV_GRP_NIU, },
+ { .group = HV_GRP_VF_CPU, },
+ { .group = HV_GRP_KT_CPU, },
+ { .group = HV_GRP_VT_CPU, },
+ { .group = HV_GRP_DIAG, .flags = FLAG_PRE_API },
+};
+
+static DEFINE_SPINLOCK(hvapi_lock);
+
+static struct api_info *__get_info(unsigned long group)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(api_table); i++) {
+ if (api_table[i].group == group)
+ return &api_table[i];
+ }
+ return NULL;
+}
+
+static void __get_ref(struct api_info *p)
+{
+ p->refcnt++;
+}
+
+static void __put_ref(struct api_info *p)
+{
+ if (--p->refcnt == 0) {
+ unsigned long ignore;
+
+ sun4v_set_version(p->group, 0, 0, &ignore);
+ p->major = p->minor = 0;
+ }
+}
+
+/* Register a hypervisor API specification. It indicates the
+ * API group and desired major+minor.
+ *
+ * If an existing API registration exists '0' (success) will
+ * be returned if it is compatible with the one being registered.
+ * Otherwise a negative error code will be returned.
+ *
+ * Otherwise an attempt will be made to negotiate the requested
+ * API group/major/minor with the hypervisor, and errors returned
+ * if that does not succeed.
+ */
+int sun4v_hvapi_register(unsigned long group, unsigned long major,
+ unsigned long *minor)
+{
+ struct api_info *p;
+ unsigned long flags;
+ int ret;
+
+ spin_lock_irqsave(&hvapi_lock, flags);
+ p = __get_info(group);
+ ret = -EINVAL;
+ if (p) {
+ if (p->refcnt) {
+ ret = -EINVAL;
+ if (p->major == major) {
+ *minor = p->minor;
+ ret = 0;
+ }
+ } else {
+ unsigned long actual_minor;
+ unsigned long hv_ret;
+
+ hv_ret = sun4v_set_version(group, major, *minor,
+ &actual_minor);
+ ret = -EINVAL;
+ if (hv_ret == HV_EOK) {
+ *minor = actual_minor;
+ p->major = major;
+ p->minor = actual_minor;
+ ret = 0;
+ } else if (hv_ret == HV_EBADTRAP ||
+ hv_ret == HV_ENOTSUPPORTED) {
+ if (p->flags & FLAG_PRE_API) {
+ if (major == 1) {
+ p->major = 1;
+ p->minor = 0;
+ *minor = 0;
+ ret = 0;
+ }
+ }
+ }
+ }
+
+ if (ret == 0)
+ __get_ref(p);
+ }
+ spin_unlock_irqrestore(&hvapi_lock, flags);
+
+ return ret;
+}
+EXPORT_SYMBOL(sun4v_hvapi_register);
+
+void sun4v_hvapi_unregister(unsigned long group)
+{
+ struct api_info *p;
+ unsigned long flags;
+
+ spin_lock_irqsave(&hvapi_lock, flags);
+ p = __get_info(group);
+ if (p)
+ __put_ref(p);
+ spin_unlock_irqrestore(&hvapi_lock, flags);
+}
+EXPORT_SYMBOL(sun4v_hvapi_unregister);
+
+int sun4v_hvapi_get(unsigned long group,
+ unsigned long *major,
+ unsigned long *minor)
+{
+ struct api_info *p;
+ unsigned long flags;
+ int ret;
+
+ spin_lock_irqsave(&hvapi_lock, flags);
+ ret = -EINVAL;
+ p = __get_info(group);
+ if (p && p->refcnt) {
+ *major = p->major;
+ *minor = p->minor;
+ ret = 0;
+ }
+ spin_unlock_irqrestore(&hvapi_lock, flags);
+
+ return ret;
+}
+EXPORT_SYMBOL(sun4v_hvapi_get);
+
+void __init sun4v_hvapi_init(void)
+{
+ unsigned long group, major, minor;
+
+ group = HV_GRP_SUN4V;
+ major = 1;
+ minor = 0;
+ if (sun4v_hvapi_register(group, major, &minor))
+ goto bad;
+
+ group = HV_GRP_CORE;
+ major = 1;
+ minor = 1;
+ if (sun4v_hvapi_register(group, major, &minor))
+ goto bad;
+
+ return;
+
+bad:
+ prom_printf("HVAPI: Cannot register API group "
+ "%lx with major(%lu) minor(%lu)\n",
+ group, major, minor);
+ prom_halt();
+}
diff --git a/arch/sparc/kernel/hvcalls.S b/arch/sparc/kernel/hvcalls.S
new file mode 100644
index 00000000000..f3ab509b76a
--- /dev/null
+++ b/arch/sparc/kernel/hvcalls.S
@@ -0,0 +1,823 @@
+ /* %o0: devhandle
+ * %o1: devino
+ *
+ * returns %o0: sysino
+ */
+ENTRY(sun4v_devino_to_sysino)
+ mov HV_FAST_INTR_DEVINO2SYSINO, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o1, %o0
+ENDPROC(sun4v_devino_to_sysino)
+
+ /* %o0: sysino
+ *
+ * returns %o0: intr_enabled (HV_INTR_{DISABLED,ENABLED})
+ */
+ENTRY(sun4v_intr_getenabled)
+ mov HV_FAST_INTR_GETENABLED, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o1, %o0
+ENDPROC(sun4v_intr_getenabled)
+
+ /* %o0: sysino
+ * %o1: intr_enabled (HV_INTR_{DISABLED,ENABLED})
+ */
+ENTRY(sun4v_intr_setenabled)
+ mov HV_FAST_INTR_SETENABLED, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_intr_setenabled)
+
+ /* %o0: sysino
+ *
+ * returns %o0: intr_state (HV_INTR_STATE_*)
+ */
+ENTRY(sun4v_intr_getstate)
+ mov HV_FAST_INTR_GETSTATE, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o1, %o0
+ENDPROC(sun4v_intr_getstate)
+
+ /* %o0: sysino
+ * %o1: intr_state (HV_INTR_STATE_*)
+ */
+ENTRY(sun4v_intr_setstate)
+ mov HV_FAST_INTR_SETSTATE, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_intr_setstate)
+
+ /* %o0: sysino
+ *
+ * returns %o0: cpuid
+ */
+ENTRY(sun4v_intr_gettarget)
+ mov HV_FAST_INTR_GETTARGET, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o1, %o0
+ENDPROC(sun4v_intr_gettarget)
+
+ /* %o0: sysino
+ * %o1: cpuid
+ */
+ENTRY(sun4v_intr_settarget)
+ mov HV_FAST_INTR_SETTARGET, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_intr_settarget)
+
+ /* %o0: cpuid
+ * %o1: pc
+ * %o2: rtba
+ * %o3: arg0
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_cpu_start)
+ mov HV_FAST_CPU_START, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_cpu_start)
+
+ /* %o0: cpuid
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_cpu_stop)
+ mov HV_FAST_CPU_STOP, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_cpu_stop)
+
+ /* returns %o0: status */
+ENTRY(sun4v_cpu_yield)
+ mov HV_FAST_CPU_YIELD, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_cpu_yield)
+
+ /* %o0: type
+ * %o1: queue paddr
+ * %o2: num queue entries
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_cpu_qconf)
+ mov HV_FAST_CPU_QCONF, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_cpu_qconf)
+
+ /* %o0: num cpus in cpu list
+ * %o1: cpu list paddr
+ * %o2: mondo block paddr
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_cpu_mondo_send)
+ mov HV_FAST_CPU_MONDO_SEND, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_cpu_mondo_send)
+
+ /* %o0: CPU ID
+ *
+ * returns %o0: -status if status non-zero, else
+ * %o0: cpu state as HV_CPU_STATE_*
+ */
+ENTRY(sun4v_cpu_state)
+ mov HV_FAST_CPU_STATE, %o5
+ ta HV_FAST_TRAP
+ brnz,pn %o0, 1f
+ sub %g0, %o0, %o0
+ mov %o1, %o0
+1: retl
+ nop
+ENDPROC(sun4v_cpu_state)
+
+ /* %o0: virtual address
+ * %o1: must be zero
+ * %o2: TTE
+ * %o3: HV_MMU_* flags
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_mmu_map_perm_addr)
+ mov HV_FAST_MMU_MAP_PERM_ADDR, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_mmu_map_perm_addr)
+
+ /* %o0: number of TSB descriptions
+ * %o1: TSB descriptions real address
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_mmu_tsb_ctx0)
+ mov HV_FAST_MMU_TSB_CTX0, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_mmu_tsb_ctx0)
+
+ /* %o0: API group number
+ * %o1: pointer to unsigned long major number storage
+ * %o2: pointer to unsigned long minor number storage
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_get_version)
+ mov HV_CORE_GET_VER, %o5
+ mov %o1, %o3
+ mov %o2, %o4
+ ta HV_CORE_TRAP
+ stx %o1, [%o3]
+ retl
+ stx %o2, [%o4]
+ENDPROC(sun4v_get_version)
+
+ /* %o0: API group number
+ * %o1: desired major number
+ * %o2: desired minor number
+ * %o3: pointer to unsigned long actual minor number storage
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_set_version)
+ mov HV_CORE_SET_VER, %o5
+ mov %o3, %o4
+ ta HV_CORE_TRAP
+ retl
+ stx %o1, [%o4]
+ENDPROC(sun4v_set_version)
+
+ /* %o0: pointer to unsigned long time
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_tod_get)
+ mov %o0, %o4
+ mov HV_FAST_TOD_GET, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_tod_get)
+
+ /* %o0: time
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_tod_set)
+ mov HV_FAST_TOD_SET, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_tod_set)
+
+ /* %o0: pointer to unsigned long status
+ *
+ * returns %o0: signed character
+ */
+ENTRY(sun4v_con_getchar)
+ mov %o0, %o4
+ mov HV_FAST_CONS_GETCHAR, %o5
+ clr %o0
+ clr %o1
+ ta HV_FAST_TRAP
+ stx %o0, [%o4]
+ retl
+ sra %o1, 0, %o0
+ENDPROC(sun4v_con_getchar)
+
+ /* %o0: signed long character
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_con_putchar)
+ mov HV_FAST_CONS_PUTCHAR, %o5
+ ta HV_FAST_TRAP
+ retl
+ sra %o0, 0, %o0
+ENDPROC(sun4v_con_putchar)
+
+ /* %o0: buffer real address
+ * %o1: buffer size
+ * %o2: pointer to unsigned long bytes_read
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_con_read)
+ mov %o2, %o4
+ mov HV_FAST_CONS_READ, %o5
+ ta HV_FAST_TRAP
+ brnz %o0, 1f
+ cmp %o1, -1 /* break */
+ be,a,pn %icc, 1f
+ mov %o1, %o0
+ cmp %o1, -2 /* hup */
+ be,a,pn %icc, 1f
+ mov %o1, %o0
+ stx %o1, [%o4]
+1: retl
+ nop
+ENDPROC(sun4v_con_read)
+
+ /* %o0: buffer real address
+ * %o1: buffer size
+ * %o2: pointer to unsigned long bytes_written
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_con_write)
+ mov %o2, %o4
+ mov HV_FAST_CONS_WRITE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_con_write)
+
+ /* %o0: soft state
+ * %o1: address of description string
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_mach_set_soft_state)
+ mov HV_FAST_MACH_SET_SOFT_STATE, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_mach_set_soft_state)
+
+ /* %o0: exit code
+ *
+ * Does not return.
+ */
+ENTRY(sun4v_mach_exit)
+ mov HV_FAST_MACH_EXIT, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_mach_exit)
+
+ /* %o0: buffer real address
+ * %o1: buffer length
+ * %o2: pointer to unsigned long real_buf_len
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_mach_desc)
+ mov %o2, %o4
+ mov HV_FAST_MACH_DESC, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_mach_desc)
+
+ /* %o0: new timeout in milliseconds
+ * %o1: pointer to unsigned long orig_timeout
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_mach_set_watchdog)
+ mov %o1, %o4
+ mov HV_FAST_MACH_SET_WATCHDOG, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_mach_set_watchdog)
+
+ /* No inputs and does not return. */
+ENTRY(sun4v_mach_sir)
+ mov %o1, %o4
+ mov HV_FAST_MACH_SIR, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_mach_sir)
+
+ /* %o0: channel
+ * %o1: ra
+ * %o2: num_entries
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_tx_qconf)
+ mov HV_FAST_LDC_TX_QCONF, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ldc_tx_qconf)
+
+ /* %o0: channel
+ * %o1: pointer to unsigned long ra
+ * %o2: pointer to unsigned long num_entries
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_tx_qinfo)
+ mov %o1, %g1
+ mov %o2, %g2
+ mov HV_FAST_LDC_TX_QINFO, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ stx %o2, [%g2]
+ retl
+ nop
+ENDPROC(sun4v_ldc_tx_qinfo)
+
+ /* %o0: channel
+ * %o1: pointer to unsigned long head_off
+ * %o2: pointer to unsigned long tail_off
+ * %o2: pointer to unsigned long chan_state
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_tx_get_state)
+ mov %o1, %g1
+ mov %o2, %g2
+ mov %o3, %g3
+ mov HV_FAST_LDC_TX_GET_STATE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ stx %o2, [%g2]
+ stx %o3, [%g3]
+ retl
+ nop
+ENDPROC(sun4v_ldc_tx_get_state)
+
+ /* %o0: channel
+ * %o1: tail_off
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_tx_set_qtail)
+ mov HV_FAST_LDC_TX_SET_QTAIL, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ldc_tx_set_qtail)
+
+ /* %o0: channel
+ * %o1: ra
+ * %o2: num_entries
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_rx_qconf)
+ mov HV_FAST_LDC_RX_QCONF, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ldc_rx_qconf)
+
+ /* %o0: channel
+ * %o1: pointer to unsigned long ra
+ * %o2: pointer to unsigned long num_entries
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_rx_qinfo)
+ mov %o1, %g1
+ mov %o2, %g2
+ mov HV_FAST_LDC_RX_QINFO, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ stx %o2, [%g2]
+ retl
+ nop
+ENDPROC(sun4v_ldc_rx_qinfo)
+
+ /* %o0: channel
+ * %o1: pointer to unsigned long head_off
+ * %o2: pointer to unsigned long tail_off
+ * %o2: pointer to unsigned long chan_state
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_rx_get_state)
+ mov %o1, %g1
+ mov %o2, %g2
+ mov %o3, %g3
+ mov HV_FAST_LDC_RX_GET_STATE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ stx %o2, [%g2]
+ stx %o3, [%g3]
+ retl
+ nop
+ENDPROC(sun4v_ldc_rx_get_state)
+
+ /* %o0: channel
+ * %o1: head_off
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_rx_set_qhead)
+ mov HV_FAST_LDC_RX_SET_QHEAD, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ldc_rx_set_qhead)
+
+ /* %o0: channel
+ * %o1: ra
+ * %o2: num_entries
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_set_map_table)
+ mov HV_FAST_LDC_SET_MAP_TABLE, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ldc_set_map_table)
+
+ /* %o0: channel
+ * %o1: pointer to unsigned long ra
+ * %o2: pointer to unsigned long num_entries
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_get_map_table)
+ mov %o1, %g1
+ mov %o2, %g2
+ mov HV_FAST_LDC_GET_MAP_TABLE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ stx %o2, [%g2]
+ retl
+ nop
+ENDPROC(sun4v_ldc_get_map_table)
+
+ /* %o0: channel
+ * %o1: dir_code
+ * %o2: tgt_raddr
+ * %o3: lcl_raddr
+ * %o4: len
+ * %o5: pointer to unsigned long actual_len
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_copy)
+ mov %o5, %g1
+ mov HV_FAST_LDC_COPY, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ retl
+ nop
+ENDPROC(sun4v_ldc_copy)
+
+ /* %o0: channel
+ * %o1: cookie
+ * %o2: pointer to unsigned long ra
+ * %o3: pointer to unsigned long perm
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_mapin)
+ mov %o2, %g1
+ mov %o3, %g2
+ mov HV_FAST_LDC_MAPIN, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ stx %o2, [%g2]
+ retl
+ nop
+ENDPROC(sun4v_ldc_mapin)
+
+ /* %o0: ra
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_unmap)
+ mov HV_FAST_LDC_UNMAP, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ldc_unmap)
+
+ /* %o0: channel
+ * %o1: cookie
+ * %o2: mte_cookie
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ldc_revoke)
+ mov HV_FAST_LDC_REVOKE, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ldc_revoke)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: pointer to unsigned long cookie
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_get_cookie)
+ mov %o2, %g1
+ mov HV_FAST_VINTR_GET_COOKIE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ retl
+ nop
+ENDPROC(sun4v_vintr_get_cookie)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: cookie
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_set_cookie)
+ mov HV_FAST_VINTR_SET_COOKIE, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_vintr_set_cookie)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: pointer to unsigned long valid_state
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_get_valid)
+ mov %o2, %g1
+ mov HV_FAST_VINTR_GET_VALID, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ retl
+ nop
+ENDPROC(sun4v_vintr_get_valid)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: valid_state
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_set_valid)
+ mov HV_FAST_VINTR_SET_VALID, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_vintr_set_valid)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: pointer to unsigned long state
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_get_state)
+ mov %o2, %g1
+ mov HV_FAST_VINTR_GET_STATE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ retl
+ nop
+ENDPROC(sun4v_vintr_get_state)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: state
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_set_state)
+ mov HV_FAST_VINTR_SET_STATE, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_vintr_set_state)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: pointer to unsigned long cpuid
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_get_target)
+ mov %o2, %g1
+ mov HV_FAST_VINTR_GET_TARGET, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%g1]
+ retl
+ nop
+ENDPROC(sun4v_vintr_get_target)
+
+ /* %o0: device handle
+ * %o1: device INO
+ * %o2: cpuid
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_vintr_set_target)
+ mov HV_FAST_VINTR_SET_TARGET, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_vintr_set_target)
+
+ /* %o0: NCS sub-function
+ * %o1: sub-function arg real-address
+ * %o2: sub-function arg size
+ *
+ * returns %o0: status
+ */
+ENTRY(sun4v_ncs_request)
+ mov HV_FAST_NCS_REQUEST, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_ncs_request)
+
+ENTRY(sun4v_svc_send)
+ save %sp, -192, %sp
+ mov %i0, %o0
+ mov %i1, %o1
+ mov %i2, %o2
+ mov HV_FAST_SVC_SEND, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%i3]
+ ret
+ restore
+ENDPROC(sun4v_svc_send)
+
+ENTRY(sun4v_svc_recv)
+ save %sp, -192, %sp
+ mov %i0, %o0
+ mov %i1, %o1
+ mov %i2, %o2
+ mov HV_FAST_SVC_RECV, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%i3]
+ ret
+ restore
+ENDPROC(sun4v_svc_recv)
+
+ENTRY(sun4v_svc_getstatus)
+ mov HV_FAST_SVC_GETSTATUS, %o5
+ mov %o1, %o4
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_svc_getstatus)
+
+ENTRY(sun4v_svc_setstatus)
+ mov HV_FAST_SVC_SETSTATUS, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_svc_setstatus)
+
+ENTRY(sun4v_svc_clrstatus)
+ mov HV_FAST_SVC_CLRSTATUS, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_svc_clrstatus)
+
+ENTRY(sun4v_mmustat_conf)
+ mov %o1, %o4
+ mov HV_FAST_MMUSTAT_CONF, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_mmustat_conf)
+
+ENTRY(sun4v_mmustat_info)
+ mov %o0, %o4
+ mov HV_FAST_MMUSTAT_INFO, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_mmustat_info)
+
+ENTRY(sun4v_mmu_demap_all)
+ clr %o0
+ clr %o1
+ mov HV_MMU_ALL, %o2
+ mov HV_FAST_MMU_DEMAP_ALL, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_mmu_demap_all)
+
+ENTRY(sun4v_niagara_getperf)
+ mov %o0, %o4
+ mov HV_FAST_GET_PERFREG, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_niagara_getperf)
+
+ENTRY(sun4v_niagara_setperf)
+ mov HV_FAST_SET_PERFREG, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_niagara_setperf)
+
+ENTRY(sun4v_niagara2_getperf)
+ mov %o0, %o4
+ mov HV_FAST_N2_GET_PERFREG, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_niagara2_getperf)
+
+ENTRY(sun4v_niagara2_setperf)
+ mov HV_FAST_N2_SET_PERFREG, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_niagara2_setperf)
+
+ENTRY(sun4v_reboot_data_set)
+ mov HV_FAST_REBOOT_DATA_SET, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_reboot_data_set)
+
+ENTRY(sun4v_vt_get_perfreg)
+ mov %o1, %o4
+ mov HV_FAST_VT_GET_PERFREG, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ retl
+ nop
+ENDPROC(sun4v_vt_get_perfreg)
+
+ENTRY(sun4v_vt_set_perfreg)
+ mov HV_FAST_VT_SET_PERFREG, %o5
+ ta HV_FAST_TRAP
+ retl
+ nop
+ENDPROC(sun4v_vt_set_perfreg)
diff --git a/arch/sparc/kernel/hvtramp.S b/arch/sparc/kernel/hvtramp.S
new file mode 100644
index 00000000000..b7ddcdd1dea
--- /dev/null
+++ b/arch/sparc/kernel/hvtramp.S
@@ -0,0 +1,137 @@
+/* hvtramp.S: Hypervisor start-cpu trampoline code.
+ *
+ * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
+ */
+
+
+#include <asm/thread_info.h>
+#include <asm/hypervisor.h>
+#include <asm/scratchpad.h>
+#include <asm/spitfire.h>
+#include <asm/hvtramp.h>
+#include <asm/pstate.h>
+#include <asm/ptrace.h>
+#include <asm/head.h>
+#include <asm/asi.h>
+#include <asm/pil.h>
+
+ .align 8
+ .globl hv_cpu_startup, hv_cpu_startup_end
+
+ /* This code executes directly out of the hypervisor
+ * with physical addressing (va==pa). %o0 contains
+ * our client argument which for Linux points to
+ * a descriptor data structure which defines the
+ * MMU entries we need to load up.
+ *
+ * After we set things up we enable the MMU and call
+ * into the kernel.
+ *
+ * First setup basic privileged cpu state.
+ */
+hv_cpu_startup:
+ SET_GL(0)
+ wrpr %g0, PIL_NORMAL_MAX, %pil
+ wrpr %g0, 0, %canrestore
+ wrpr %g0, 0, %otherwin
+ wrpr %g0, 6, %cansave
+ wrpr %g0, 6, %cleanwin
+ wrpr %g0, 0, %cwp
+ wrpr %g0, 0, %wstate
+ wrpr %g0, 0, %tl
+
+ sethi %hi(sparc64_ttable_tl0), %g1
+ wrpr %g1, %tba
+
+ mov %o0, %l0
+
+ lduw [%l0 + HVTRAMP_DESCR_CPU], %g1
+ mov SCRATCHPAD_CPUID, %g2
+ stxa %g1, [%g2] ASI_SCRATCHPAD
+
+ ldx [%l0 + HVTRAMP_DESCR_FAULT_INFO_VA], %g2
+ stxa %g2, [%g0] ASI_SCRATCHPAD
+
+ mov 0, %l1
+ lduw [%l0 + HVTRAMP_DESCR_NUM_MAPPINGS], %l2
+ add %l0, HVTRAMP_DESCR_MAPS, %l3
+
+1: ldx [%l3 + HVTRAMP_MAPPING_VADDR], %o0
+ clr %o1
+ ldx [%l3 + HVTRAMP_MAPPING_TTE], %o2
+ mov HV_MMU_IMMU | HV_MMU_DMMU, %o3
+ mov HV_FAST_MMU_MAP_PERM_ADDR, %o5
+ ta HV_FAST_TRAP
+
+ brnz,pn %o0, 80f
+ nop
+
+ add %l1, 1, %l1
+ cmp %l1, %l2
+ blt,a,pt %xcc, 1b
+ add %l3, HVTRAMP_MAPPING_SIZE, %l3
+
+ ldx [%l0 + HVTRAMP_DESCR_FAULT_INFO_PA], %o0
+ mov HV_FAST_MMU_FAULT_AREA_CONF, %o5
+ ta HV_FAST_TRAP
+
+ brnz,pn %o0, 80f
+ nop
+
+ wrpr %g0, (PSTATE_PRIV | PSTATE_PEF), %pstate
+
+ ldx [%l0 + HVTRAMP_DESCR_THREAD_REG], %l6
+
+ mov 1, %o0
+ set 1f, %o1
+ mov HV_FAST_MMU_ENABLE, %o5
+ ta HV_FAST_TRAP
+
+ ba,pt %xcc, 80f
+ nop
+
+1:
+ wr %g0, 0, %fprs
+ wr %g0, ASI_P, %asi
+
+ mov PRIMARY_CONTEXT, %g7
+ stxa %g0, [%g7] ASI_MMU
+ membar #Sync
+
+ mov SECONDARY_CONTEXT, %g7
+ stxa %g0, [%g7] ASI_MMU
+ membar #Sync
+
+ mov %l6, %g6
+ ldx [%g6 + TI_TASK], %g4
+
+ mov 1, %g5
+ sllx %g5, THREAD_SHIFT, %g5
+ sub %g5, (STACKFRAME_SZ + STACK_BIAS), %g5
+ add %g6, %g5, %sp
+ mov 0, %fp
+
+ call init_irqwork_curcpu
+ nop
+ call hard_smp_processor_id
+ nop
+
+ call sun4v_register_mondo_queues
+ nop
+
+ call init_cur_cpu_trap
+ mov %g6, %o0
+
+ wrpr %g0, (PSTATE_PRIV | PSTATE_PEF | PSTATE_IE), %pstate
+
+ call smp_callin
+ nop
+
+ call cpu_panic
+ nop
+
+80: ba,pt %xcc, 80b
+ nop
+
+ .align 8
+hv_cpu_startup_end:
diff --git a/arch/sparc/kernel/idprom.c b/arch/sparc/kernel/idprom.c
index 3a5bad52539..6bd75012109 100644
--- a/arch/sparc/kernel/idprom.c
+++ b/arch/sparc/kernel/idprom.c
@@ -1,4 +1,4 @@
-/* $Id: idprom.c,v 1.24 1999/08/31 06:54:20 davem Exp $
+/*
* idprom.c: Routines to load the idprom into kernel addresses and
* interpret the data contained within.
*
@@ -8,64 +8,58 @@
#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/init.h>
+#include <linux/export.h>
#include <asm/oplib.h>
#include <asm/idprom.h>
-#include <asm/machines.h> /* Fun with Sun released architectures. */
-#ifdef CONFIG_SUN4
-#include <asm/sun4paddr.h>
-extern void sun4setup(void);
-#endif
struct idprom *idprom;
+EXPORT_SYMBOL(idprom);
+
static struct idprom idprom_buffer;
+#ifdef CONFIG_SPARC32
+#include <asm/machines.h> /* Fun with Sun released architectures. */
+
/* Here is the master table of Sun machines which use some implementation
* of the Sparc CPU and have a meaningful IDPROM machtype value that we
* know about. See asm-sparc/machines.h for empirical constants.
*/
-struct Sun_Machine_Models Sun_Machines[NUM_SUN_MACHINES] = {
-/* First, Sun4's */
-{ "Sun 4/100 Series", (SM_SUN4 | SM_4_110) },
-{ "Sun 4/200 Series", (SM_SUN4 | SM_4_260) },
-{ "Sun 4/300 Series", (SM_SUN4 | SM_4_330) },
-{ "Sun 4/400 Series", (SM_SUN4 | SM_4_470) },
-/* Now, Sun4c's */
-{ "Sun4c SparcStation 1", (SM_SUN4C | SM_4C_SS1) },
-{ "Sun4c SparcStation IPC", (SM_SUN4C | SM_4C_IPC) },
-{ "Sun4c SparcStation 1+", (SM_SUN4C | SM_4C_SS1PLUS) },
-{ "Sun4c SparcStation SLC", (SM_SUN4C | SM_4C_SLC) },
-{ "Sun4c SparcStation 2", (SM_SUN4C | SM_4C_SS2) },
-{ "Sun4c SparcStation ELC", (SM_SUN4C | SM_4C_ELC) },
-{ "Sun4c SparcStation IPX", (SM_SUN4C | SM_4C_IPX) },
+static struct Sun_Machine_Models Sun_Machines[] = {
+/* First, Leon */
+{ .name = "Leon3 System-on-a-Chip", .id_machtype = (M_LEON | M_LEON3_SOC) },
/* Finally, early Sun4m's */
-{ "Sun4m SparcSystem600", (SM_SUN4M | SM_4M_SS60) },
-{ "Sun4m SparcStation10/20", (SM_SUN4M | SM_4M_SS50) },
-{ "Sun4m SparcStation5", (SM_SUN4M | SM_4M_SS40) },
+{ .name = "Sun4m SparcSystem600", .id_machtype = (SM_SUN4M | SM_4M_SS60) },
+{ .name = "Sun4m SparcStation10/20", .id_machtype = (SM_SUN4M | SM_4M_SS50) },
+{ .name = "Sun4m SparcStation5", .id_machtype = (SM_SUN4M | SM_4M_SS40) },
/* One entry for the OBP arch's which are sun4d, sun4e, and newer sun4m's */
-{ "Sun4M OBP based system", (SM_SUN4M_OBP | 0x0) } };
+{ .name = "Sun4M OBP based system", .id_machtype = (SM_SUN4M_OBP | 0x0) } };
static void __init display_system_type(unsigned char machtype)
{
char sysname[128];
register int i;
- for (i = 0; i < NUM_SUN_MACHINES; i++) {
- if(Sun_Machines[i].id_machtype == machtype) {
+ for (i = 0; i < ARRAY_SIZE(Sun_Machines); i++) {
+ if (Sun_Machines[i].id_machtype == machtype) {
if (machtype != (SM_SUN4M_OBP | 0x00) ||
prom_getproperty(prom_root_node, "banner-name",
sysname, sizeof(sysname)) <= 0)
- printk("TYPE: %s\n", Sun_Machines[i].name);
+ printk(KERN_WARNING "TYPE: %s\n",
+ Sun_Machines[i].name);
else
- printk("TYPE: %s\n", sysname);
+ printk(KERN_WARNING "TYPE: %s\n", sysname);
return;
}
}
- prom_printf("IDPROM: Bogus id_machtype value, 0x%x\n", machtype);
- prom_halt();
+ prom_printf("IDPROM: Warning, bogus id_machtype value, 0x%x\n", machtype);
}
-
+#else
+static void __init display_system_type(unsigned char machtype)
+{
+}
+#endif
/* Calculate the IDPROM checksum (xor of the data bytes). */
static unsigned char __init calc_idprom_cksum(struct idprom *idprom)
{
@@ -84,24 +78,14 @@ void __init idprom_init(void)
idprom = &idprom_buffer;
- if (idprom->id_format != 0x01) {
- prom_printf("IDPROM: Unknown format type!\n");
- prom_halt();
- }
+ if (idprom->id_format != 0x01)
+ prom_printf("IDPROM: Warning, unknown format type!\n");
- if (idprom->id_cksum != calc_idprom_cksum(idprom)) {
- prom_printf("IDPROM: Checksum failure (nvram=%x, calc=%x)!\n",
+ if (idprom->id_cksum != calc_idprom_cksum(idprom))
+ prom_printf("IDPROM: Warning, checksum failure (nvram=%x, calc=%x)!\n",
idprom->id_cksum, calc_idprom_cksum(idprom));
- prom_halt();
- }
display_system_type(idprom->id_machtype);
- printk("Ethernet address: %x:%x:%x:%x:%x:%x\n",
- idprom->id_ethaddr[0], idprom->id_ethaddr[1],
- idprom->id_ethaddr[2], idprom->id_ethaddr[3],
- idprom->id_ethaddr[4], idprom->id_ethaddr[5]);
-#ifdef CONFIG_SUN4
- sun4setup();
-#endif
+ printk(KERN_WARNING "Ethernet address: %pM\n", idprom->id_ethaddr);
}
diff --git a/arch/sparc/kernel/init_task.c b/arch/sparc/kernel/init_task.c
deleted file mode 100644
index d9d4f96360c..00000000000
--- a/arch/sparc/kernel/init_task.c
+++ /dev/null
@@ -1,29 +0,0 @@
-#include <linux/mm.h>
-#include <linux/fs.h>
-#include <linux/module.h>
-#include <linux/sched.h>
-#include <linux/init_task.h>
-#include <linux/mqueue.h>
-
-#include <asm/pgtable.h>
-#include <asm/uaccess.h>
-
-static struct fs_struct init_fs = INIT_FS;
-static struct files_struct init_files = INIT_FILES;
-static struct signal_struct init_signals = INIT_SIGNALS(init_signals);
-static struct sighand_struct init_sighand = INIT_SIGHAND(init_sighand);
-struct mm_struct init_mm = INIT_MM(init_mm);
-struct task_struct init_task = INIT_TASK(init_task);
-
-EXPORT_SYMBOL(init_mm);
-EXPORT_SYMBOL(init_task);
-
-/* .text section in head.S is aligned at 8k boundary and this gets linked
- * right after that so that the init_thread_union is aligned properly as well.
- * If this is not aligned on a 8k boundry, then you should change code
- * in etrap.S which assumes it.
- */
-union thread_union init_thread_union
- __attribute__((section (".text\"\n\t#")))
- __attribute__((aligned (THREAD_SIZE)))
- = { INIT_THREAD_INFO(init_task) };
diff --git a/arch/sparc/kernel/iommu.c b/arch/sparc/kernel/iommu.c
new file mode 100644
index 00000000000..bfa4d0c2df4
--- /dev/null
+++ b/arch/sparc/kernel/iommu.c
@@ -0,0 +1,862 @@
+/* iommu.c: Generic sparc64 IOMMU support.
+ *
+ * Copyright (C) 1999, 2007, 2008 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1999, 2000 Jakub Jelinek (jakub@redhat.com)
+ */
+
+#include <linux/kernel.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/errno.h>
+#include <linux/iommu-helper.h>
+#include <linux/bitmap.h>
+
+#ifdef CONFIG_PCI
+#include <linux/pci.h>
+#endif
+
+#include <asm/iommu.h>
+
+#include "iommu_common.h"
+#include "kernel.h"
+
+#define STC_CTXMATCH_ADDR(STC, CTX) \
+ ((STC)->strbuf_ctxmatch_base + ((CTX) << 3))
+#define STC_FLUSHFLAG_INIT(STC) \
+ (*((STC)->strbuf_flushflag) = 0UL)
+#define STC_FLUSHFLAG_SET(STC) \
+ (*((STC)->strbuf_flushflag) != 0UL)
+
+#define iommu_read(__reg) \
+({ u64 __ret; \
+ __asm__ __volatile__("ldxa [%1] %2, %0" \
+ : "=r" (__ret) \
+ : "r" (__reg), "i" (ASI_PHYS_BYPASS_EC_E) \
+ : "memory"); \
+ __ret; \
+})
+#define iommu_write(__reg, __val) \
+ __asm__ __volatile__("stxa %0, [%1] %2" \
+ : /* no outputs */ \
+ : "r" (__val), "r" (__reg), \
+ "i" (ASI_PHYS_BYPASS_EC_E))
+
+/* Must be invoked under the IOMMU lock. */
+static void iommu_flushall(struct iommu *iommu)
+{
+ if (iommu->iommu_flushinv) {
+ iommu_write(iommu->iommu_flushinv, ~(u64)0);
+ } else {
+ unsigned long tag;
+ int entry;
+
+ tag = iommu->iommu_tags;
+ for (entry = 0; entry < 16; entry++) {
+ iommu_write(tag, 0);
+ tag += 8;
+ }
+
+ /* Ensure completion of previous PIO writes. */
+ (void) iommu_read(iommu->write_complete_reg);
+ }
+}
+
+#define IOPTE_CONSISTENT(CTX) \
+ (IOPTE_VALID | IOPTE_CACHE | \
+ (((CTX) << 47) & IOPTE_CONTEXT))
+
+#define IOPTE_STREAMING(CTX) \
+ (IOPTE_CONSISTENT(CTX) | IOPTE_STBUF)
+
+/* Existing mappings are never marked invalid, instead they
+ * are pointed to a dummy page.
+ */
+#define IOPTE_IS_DUMMY(iommu, iopte) \
+ ((iopte_val(*iopte) & IOPTE_PAGE) == (iommu)->dummy_page_pa)
+
+static inline void iopte_make_dummy(struct iommu *iommu, iopte_t *iopte)
+{
+ unsigned long val = iopte_val(*iopte);
+
+ val &= ~IOPTE_PAGE;
+ val |= iommu->dummy_page_pa;
+
+ iopte_val(*iopte) = val;
+}
+
+/* Based almost entirely upon the ppc64 iommu allocator. If you use the 'handle'
+ * facility it must all be done in one pass while under the iommu lock.
+ *
+ * On sun4u platforms, we only flush the IOMMU once every time we've passed
+ * over the entire page table doing allocations. Therefore we only ever advance
+ * the hint and cannot backtrack it.
+ */
+unsigned long iommu_range_alloc(struct device *dev,
+ struct iommu *iommu,
+ unsigned long npages,
+ unsigned long *handle)
+{
+ unsigned long n, end, start, limit, boundary_size;
+ struct iommu_arena *arena = &iommu->arena;
+ int pass = 0;
+
+ /* This allocator was derived from x86_64's bit string search */
+
+ /* Sanity check */
+ if (unlikely(npages == 0)) {
+ if (printk_ratelimit())
+ WARN_ON(1);
+ return DMA_ERROR_CODE;
+ }
+
+ if (handle && *handle)
+ start = *handle;
+ else
+ start = arena->hint;
+
+ limit = arena->limit;
+
+ /* The case below can happen if we have a small segment appended
+ * to a large, or when the previous alloc was at the very end of
+ * the available space. If so, go back to the beginning and flush.
+ */
+ if (start >= limit) {
+ start = 0;
+ if (iommu->flush_all)
+ iommu->flush_all(iommu);
+ }
+
+ again:
+
+ if (dev)
+ boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
+ 1 << IO_PAGE_SHIFT);
+ else
+ boundary_size = ALIGN(1UL << 32, 1 << IO_PAGE_SHIFT);
+
+ n = iommu_area_alloc(arena->map, limit, start, npages,
+ iommu->page_table_map_base >> IO_PAGE_SHIFT,
+ boundary_size >> IO_PAGE_SHIFT, 0);
+ if (n == -1) {
+ if (likely(pass < 1)) {
+ /* First failure, rescan from the beginning. */
+ start = 0;
+ if (iommu->flush_all)
+ iommu->flush_all(iommu);
+ pass++;
+ goto again;
+ } else {
+ /* Second failure, give up */
+ return DMA_ERROR_CODE;
+ }
+ }
+
+ end = n + npages;
+
+ arena->hint = end;
+
+ /* Update handle for SG allocations */
+ if (handle)
+ *handle = end;
+
+ return n;
+}
+
+void iommu_range_free(struct iommu *iommu, dma_addr_t dma_addr, unsigned long npages)
+{
+ struct iommu_arena *arena = &iommu->arena;
+ unsigned long entry;
+
+ entry = (dma_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT;
+
+ bitmap_clear(arena->map, entry, npages);
+}
+
+int iommu_table_init(struct iommu *iommu, int tsbsize,
+ u32 dma_offset, u32 dma_addr_mask,
+ int numa_node)
+{
+ unsigned long i, order, sz, num_tsb_entries;
+ struct page *page;
+
+ num_tsb_entries = tsbsize / sizeof(iopte_t);
+
+ /* Setup initial software IOMMU state. */
+ spin_lock_init(&iommu->lock);
+ iommu->ctx_lowest_free = 1;
+ iommu->page_table_map_base = dma_offset;
+ iommu->dma_addr_mask = dma_addr_mask;
+
+ /* Allocate and initialize the free area map. */
+ sz = num_tsb_entries / 8;
+ sz = (sz + 7UL) & ~7UL;
+ iommu->arena.map = kmalloc_node(sz, GFP_KERNEL, numa_node);
+ if (!iommu->arena.map) {
+ printk(KERN_ERR "IOMMU: Error, kmalloc(arena.map) failed.\n");
+ return -ENOMEM;
+ }
+ memset(iommu->arena.map, 0, sz);
+ iommu->arena.limit = num_tsb_entries;
+
+ if (tlb_type != hypervisor)
+ iommu->flush_all = iommu_flushall;
+
+ /* Allocate and initialize the dummy page which we
+ * set inactive IO PTEs to point to.
+ */
+ page = alloc_pages_node(numa_node, GFP_KERNEL, 0);
+ if (!page) {
+ printk(KERN_ERR "IOMMU: Error, gfp(dummy_page) failed.\n");
+ goto out_free_map;
+ }
+ iommu->dummy_page = (unsigned long) page_address(page);
+ memset((void *)iommu->dummy_page, 0, PAGE_SIZE);
+ iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page);
+
+ /* Now allocate and setup the IOMMU page table itself. */
+ order = get_order(tsbsize);
+ page = alloc_pages_node(numa_node, GFP_KERNEL, order);
+ if (!page) {
+ printk(KERN_ERR "IOMMU: Error, gfp(tsb) failed.\n");
+ goto out_free_dummy_page;
+ }
+ iommu->page_table = (iopte_t *)page_address(page);
+
+ for (i = 0; i < num_tsb_entries; i++)
+ iopte_make_dummy(iommu, &iommu->page_table[i]);
+
+ return 0;
+
+out_free_dummy_page:
+ free_page(iommu->dummy_page);
+ iommu->dummy_page = 0UL;
+
+out_free_map:
+ kfree(iommu->arena.map);
+ iommu->arena.map = NULL;
+
+ return -ENOMEM;
+}
+
+static inline iopte_t *alloc_npages(struct device *dev, struct iommu *iommu,
+ unsigned long npages)
+{
+ unsigned long entry;
+
+ entry = iommu_range_alloc(dev, iommu, npages, NULL);
+ if (unlikely(entry == DMA_ERROR_CODE))
+ return NULL;
+
+ return iommu->page_table + entry;
+}
+
+static int iommu_alloc_ctx(struct iommu *iommu)
+{
+ int lowest = iommu->ctx_lowest_free;
+ int n = find_next_zero_bit(iommu->ctx_bitmap, IOMMU_NUM_CTXS, lowest);
+
+ if (unlikely(n == IOMMU_NUM_CTXS)) {
+ n = find_next_zero_bit(iommu->ctx_bitmap, lowest, 1);
+ if (unlikely(n == lowest)) {
+ printk(KERN_WARNING "IOMMU: Ran out of contexts.\n");
+ n = 0;
+ }
+ }
+ if (n)
+ __set_bit(n, iommu->ctx_bitmap);
+
+ return n;
+}
+
+static inline void iommu_free_ctx(struct iommu *iommu, int ctx)
+{
+ if (likely(ctx)) {
+ __clear_bit(ctx, iommu->ctx_bitmap);
+ if (ctx < iommu->ctx_lowest_free)
+ iommu->ctx_lowest_free = ctx;
+ }
+}
+
+static void *dma_4u_alloc_coherent(struct device *dev, size_t size,
+ dma_addr_t *dma_addrp, gfp_t gfp,
+ struct dma_attrs *attrs)
+{
+ unsigned long flags, order, first_page;
+ struct iommu *iommu;
+ struct page *page;
+ int npages, nid;
+ iopte_t *iopte;
+ void *ret;
+
+ size = IO_PAGE_ALIGN(size);
+ order = get_order(size);
+ if (order >= 10)
+ return NULL;
+
+ nid = dev->archdata.numa_node;
+ page = alloc_pages_node(nid, gfp, order);
+ if (unlikely(!page))
+ return NULL;
+
+ first_page = (unsigned long) page_address(page);
+ memset((char *)first_page, 0, PAGE_SIZE << order);
+
+ iommu = dev->archdata.iommu;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ iopte = alloc_npages(dev, iommu, size >> IO_PAGE_SHIFT);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (unlikely(iopte == NULL)) {
+ free_pages(first_page, order);
+ return NULL;
+ }
+
+ *dma_addrp = (iommu->page_table_map_base +
+ ((iopte - iommu->page_table) << IO_PAGE_SHIFT));
+ ret = (void *) first_page;
+ npages = size >> IO_PAGE_SHIFT;
+ first_page = __pa(first_page);
+ while (npages--) {
+ iopte_val(*iopte) = (IOPTE_CONSISTENT(0UL) |
+ IOPTE_WRITE |
+ (first_page & IOPTE_PAGE));
+ iopte++;
+ first_page += IO_PAGE_SIZE;
+ }
+
+ return ret;
+}
+
+static void dma_4u_free_coherent(struct device *dev, size_t size,
+ void *cpu, dma_addr_t dvma,
+ struct dma_attrs *attrs)
+{
+ struct iommu *iommu;
+ unsigned long flags, order, npages;
+
+ npages = IO_PAGE_ALIGN(size) >> IO_PAGE_SHIFT;
+ iommu = dev->archdata.iommu;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ iommu_range_free(iommu, dvma, npages);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ order = get_order(size);
+ if (order < 10)
+ free_pages((unsigned long)cpu, order);
+}
+
+static dma_addr_t dma_4u_map_page(struct device *dev, struct page *page,
+ unsigned long offset, size_t sz,
+ enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ struct iommu *iommu;
+ struct strbuf *strbuf;
+ iopte_t *base;
+ unsigned long flags, npages, oaddr;
+ unsigned long i, base_paddr, ctx;
+ u32 bus_addr, ret;
+ unsigned long iopte_protection;
+
+ iommu = dev->archdata.iommu;
+ strbuf = dev->archdata.stc;
+
+ if (unlikely(direction == DMA_NONE))
+ goto bad_no_ctx;
+
+ oaddr = (unsigned long)(page_address(page) + offset);
+ npages = IO_PAGE_ALIGN(oaddr + sz) - (oaddr & IO_PAGE_MASK);
+ npages >>= IO_PAGE_SHIFT;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ base = alloc_npages(dev, iommu, npages);
+ ctx = 0;
+ if (iommu->iommu_ctxflush)
+ ctx = iommu_alloc_ctx(iommu);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (unlikely(!base))
+ goto bad;
+
+ bus_addr = (iommu->page_table_map_base +
+ ((base - iommu->page_table) << IO_PAGE_SHIFT));
+ ret = bus_addr | (oaddr & ~IO_PAGE_MASK);
+ base_paddr = __pa(oaddr & IO_PAGE_MASK);
+ if (strbuf->strbuf_enabled)
+ iopte_protection = IOPTE_STREAMING(ctx);
+ else
+ iopte_protection = IOPTE_CONSISTENT(ctx);
+ if (direction != DMA_TO_DEVICE)
+ iopte_protection |= IOPTE_WRITE;
+
+ for (i = 0; i < npages; i++, base++, base_paddr += IO_PAGE_SIZE)
+ iopte_val(*base) = iopte_protection | base_paddr;
+
+ return ret;
+
+bad:
+ iommu_free_ctx(iommu, ctx);
+bad_no_ctx:
+ if (printk_ratelimit())
+ WARN_ON(1);
+ return DMA_ERROR_CODE;
+}
+
+static void strbuf_flush(struct strbuf *strbuf, struct iommu *iommu,
+ u32 vaddr, unsigned long ctx, unsigned long npages,
+ enum dma_data_direction direction)
+{
+ int limit;
+
+ if (strbuf->strbuf_ctxflush &&
+ iommu->iommu_ctxflush) {
+ unsigned long matchreg, flushreg;
+ u64 val;
+
+ flushreg = strbuf->strbuf_ctxflush;
+ matchreg = STC_CTXMATCH_ADDR(strbuf, ctx);
+
+ iommu_write(flushreg, ctx);
+ val = iommu_read(matchreg);
+ val &= 0xffff;
+ if (!val)
+ goto do_flush_sync;
+
+ while (val) {
+ if (val & 0x1)
+ iommu_write(flushreg, ctx);
+ val >>= 1;
+ }
+ val = iommu_read(matchreg);
+ if (unlikely(val)) {
+ printk(KERN_WARNING "strbuf_flush: ctx flush "
+ "timeout matchreg[%llx] ctx[%lx]\n",
+ val, ctx);
+ goto do_page_flush;
+ }
+ } else {
+ unsigned long i;
+
+ do_page_flush:
+ for (i = 0; i < npages; i++, vaddr += IO_PAGE_SIZE)
+ iommu_write(strbuf->strbuf_pflush, vaddr);
+ }
+
+do_flush_sync:
+ /* If the device could not have possibly put dirty data into
+ * the streaming cache, no flush-flag synchronization needs
+ * to be performed.
+ */
+ if (direction == DMA_TO_DEVICE)
+ return;
+
+ STC_FLUSHFLAG_INIT(strbuf);
+ iommu_write(strbuf->strbuf_fsync, strbuf->strbuf_flushflag_pa);
+ (void) iommu_read(iommu->write_complete_reg);
+
+ limit = 100000;
+ while (!STC_FLUSHFLAG_SET(strbuf)) {
+ limit--;
+ if (!limit)
+ break;
+ udelay(1);
+ rmb();
+ }
+ if (!limit)
+ printk(KERN_WARNING "strbuf_flush: flushflag timeout "
+ "vaddr[%08x] ctx[%lx] npages[%ld]\n",
+ vaddr, ctx, npages);
+}
+
+static void dma_4u_unmap_page(struct device *dev, dma_addr_t bus_addr,
+ size_t sz, enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ struct iommu *iommu;
+ struct strbuf *strbuf;
+ iopte_t *base;
+ unsigned long flags, npages, ctx, i;
+
+ if (unlikely(direction == DMA_NONE)) {
+ if (printk_ratelimit())
+ WARN_ON(1);
+ return;
+ }
+
+ iommu = dev->archdata.iommu;
+ strbuf = dev->archdata.stc;
+
+ npages = IO_PAGE_ALIGN(bus_addr + sz) - (bus_addr & IO_PAGE_MASK);
+ npages >>= IO_PAGE_SHIFT;
+ base = iommu->page_table +
+ ((bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
+ bus_addr &= IO_PAGE_MASK;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ /* Record the context, if any. */
+ ctx = 0;
+ if (iommu->iommu_ctxflush)
+ ctx = (iopte_val(*base) & IOPTE_CONTEXT) >> 47UL;
+
+ /* Step 1: Kick data out of streaming buffers if necessary. */
+ if (strbuf->strbuf_enabled)
+ strbuf_flush(strbuf, iommu, bus_addr, ctx,
+ npages, direction);
+
+ /* Step 2: Clear out TSB entries. */
+ for (i = 0; i < npages; i++)
+ iopte_make_dummy(iommu, base + i);
+
+ iommu_range_free(iommu, bus_addr, npages);
+
+ iommu_free_ctx(iommu, ctx);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
+ int nelems, enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ struct scatterlist *s, *outs, *segstart;
+ unsigned long flags, handle, prot, ctx;
+ dma_addr_t dma_next = 0, dma_addr;
+ unsigned int max_seg_size;
+ unsigned long seg_boundary_size;
+ int outcount, incount, i;
+ struct strbuf *strbuf;
+ struct iommu *iommu;
+ unsigned long base_shift;
+
+ BUG_ON(direction == DMA_NONE);
+
+ iommu = dev->archdata.iommu;
+ strbuf = dev->archdata.stc;
+ if (nelems == 0 || !iommu)
+ return 0;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ ctx = 0;
+ if (iommu->iommu_ctxflush)
+ ctx = iommu_alloc_ctx(iommu);
+
+ if (strbuf->strbuf_enabled)
+ prot = IOPTE_STREAMING(ctx);
+ else
+ prot = IOPTE_CONSISTENT(ctx);
+ if (direction != DMA_TO_DEVICE)
+ prot |= IOPTE_WRITE;
+
+ outs = s = segstart = &sglist[0];
+ outcount = 1;
+ incount = nelems;
+ handle = 0;
+
+ /* Init first segment length for backout at failure */
+ outs->dma_length = 0;
+
+ max_seg_size = dma_get_max_seg_size(dev);
+ seg_boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
+ IO_PAGE_SIZE) >> IO_PAGE_SHIFT;
+ base_shift = iommu->page_table_map_base >> IO_PAGE_SHIFT;
+ for_each_sg(sglist, s, nelems, i) {
+ unsigned long paddr, npages, entry, out_entry = 0, slen;
+ iopte_t *base;
+
+ slen = s->length;
+ /* Sanity check */
+ if (slen == 0) {
+ dma_next = 0;
+ continue;
+ }
+ /* Allocate iommu entries for that segment */
+ paddr = (unsigned long) SG_ENT_PHYS_ADDRESS(s);
+ npages = iommu_num_pages(paddr, slen, IO_PAGE_SIZE);
+ entry = iommu_range_alloc(dev, iommu, npages, &handle);
+
+ /* Handle failure */
+ if (unlikely(entry == DMA_ERROR_CODE)) {
+ if (printk_ratelimit())
+ printk(KERN_INFO "iommu_alloc failed, iommu %p paddr %lx"
+ " npages %lx\n", iommu, paddr, npages);
+ goto iommu_map_failed;
+ }
+
+ base = iommu->page_table + entry;
+
+ /* Convert entry to a dma_addr_t */
+ dma_addr = iommu->page_table_map_base +
+ (entry << IO_PAGE_SHIFT);
+ dma_addr |= (s->offset & ~IO_PAGE_MASK);
+
+ /* Insert into HW table */
+ paddr &= IO_PAGE_MASK;
+ while (npages--) {
+ iopte_val(*base) = prot | paddr;
+ base++;
+ paddr += IO_PAGE_SIZE;
+ }
+
+ /* If we are in an open segment, try merging */
+ if (segstart != s) {
+ /* We cannot merge if:
+ * - allocated dma_addr isn't contiguous to previous allocation
+ */
+ if ((dma_addr != dma_next) ||
+ (outs->dma_length + s->length > max_seg_size) ||
+ (is_span_boundary(out_entry, base_shift,
+ seg_boundary_size, outs, s))) {
+ /* Can't merge: create a new segment */
+ segstart = s;
+ outcount++;
+ outs = sg_next(outs);
+ } else {
+ outs->dma_length += s->length;
+ }
+ }
+
+ if (segstart == s) {
+ /* This is a new segment, fill entries */
+ outs->dma_address = dma_addr;
+ outs->dma_length = slen;
+ out_entry = entry;
+ }
+
+ /* Calculate next page pointer for contiguous check */
+ dma_next = dma_addr + slen;
+ }
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (outcount < incount) {
+ outs = sg_next(outs);
+ outs->dma_address = DMA_ERROR_CODE;
+ outs->dma_length = 0;
+ }
+
+ return outcount;
+
+iommu_map_failed:
+ for_each_sg(sglist, s, nelems, i) {
+ if (s->dma_length != 0) {
+ unsigned long vaddr, npages, entry, j;
+ iopte_t *base;
+
+ vaddr = s->dma_address & IO_PAGE_MASK;
+ npages = iommu_num_pages(s->dma_address, s->dma_length,
+ IO_PAGE_SIZE);
+ iommu_range_free(iommu, vaddr, npages);
+
+ entry = (vaddr - iommu->page_table_map_base)
+ >> IO_PAGE_SHIFT;
+ base = iommu->page_table + entry;
+
+ for (j = 0; j < npages; j++)
+ iopte_make_dummy(iommu, base + j);
+
+ s->dma_address = DMA_ERROR_CODE;
+ s->dma_length = 0;
+ }
+ if (s == outs)
+ break;
+ }
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ return 0;
+}
+
+/* If contexts are being used, they are the same in all of the mappings
+ * we make for a particular SG.
+ */
+static unsigned long fetch_sg_ctx(struct iommu *iommu, struct scatterlist *sg)
+{
+ unsigned long ctx = 0;
+
+ if (iommu->iommu_ctxflush) {
+ iopte_t *base;
+ u32 bus_addr;
+
+ bus_addr = sg->dma_address & IO_PAGE_MASK;
+ base = iommu->page_table +
+ ((bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
+
+ ctx = (iopte_val(*base) & IOPTE_CONTEXT) >> 47UL;
+ }
+ return ctx;
+}
+
+static void dma_4u_unmap_sg(struct device *dev, struct scatterlist *sglist,
+ int nelems, enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ unsigned long flags, ctx;
+ struct scatterlist *sg;
+ struct strbuf *strbuf;
+ struct iommu *iommu;
+
+ BUG_ON(direction == DMA_NONE);
+
+ iommu = dev->archdata.iommu;
+ strbuf = dev->archdata.stc;
+
+ ctx = fetch_sg_ctx(iommu, sglist);
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ sg = sglist;
+ while (nelems--) {
+ dma_addr_t dma_handle = sg->dma_address;
+ unsigned int len = sg->dma_length;
+ unsigned long npages, entry;
+ iopte_t *base;
+ int i;
+
+ if (!len)
+ break;
+ npages = iommu_num_pages(dma_handle, len, IO_PAGE_SIZE);
+ iommu_range_free(iommu, dma_handle, npages);
+
+ entry = ((dma_handle - iommu->page_table_map_base)
+ >> IO_PAGE_SHIFT);
+ base = iommu->page_table + entry;
+
+ dma_handle &= IO_PAGE_MASK;
+ if (strbuf->strbuf_enabled)
+ strbuf_flush(strbuf, iommu, dma_handle, ctx,
+ npages, direction);
+
+ for (i = 0; i < npages; i++)
+ iopte_make_dummy(iommu, base + i);
+
+ sg = sg_next(sg);
+ }
+
+ iommu_free_ctx(iommu, ctx);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static void dma_4u_sync_single_for_cpu(struct device *dev,
+ dma_addr_t bus_addr, size_t sz,
+ enum dma_data_direction direction)
+{
+ struct iommu *iommu;
+ struct strbuf *strbuf;
+ unsigned long flags, ctx, npages;
+
+ iommu = dev->archdata.iommu;
+ strbuf = dev->archdata.stc;
+
+ if (!strbuf->strbuf_enabled)
+ return;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ npages = IO_PAGE_ALIGN(bus_addr + sz) - (bus_addr & IO_PAGE_MASK);
+ npages >>= IO_PAGE_SHIFT;
+ bus_addr &= IO_PAGE_MASK;
+
+ /* Step 1: Record the context, if any. */
+ ctx = 0;
+ if (iommu->iommu_ctxflush &&
+ strbuf->strbuf_ctxflush) {
+ iopte_t *iopte;
+
+ iopte = iommu->page_table +
+ ((bus_addr - iommu->page_table_map_base)>>IO_PAGE_SHIFT);
+ ctx = (iopte_val(*iopte) & IOPTE_CONTEXT) >> 47UL;
+ }
+
+ /* Step 2: Kick data out of streaming buffers. */
+ strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static void dma_4u_sync_sg_for_cpu(struct device *dev,
+ struct scatterlist *sglist, int nelems,
+ enum dma_data_direction direction)
+{
+ struct iommu *iommu;
+ struct strbuf *strbuf;
+ unsigned long flags, ctx, npages, i;
+ struct scatterlist *sg, *sgprv;
+ u32 bus_addr;
+
+ iommu = dev->archdata.iommu;
+ strbuf = dev->archdata.stc;
+
+ if (!strbuf->strbuf_enabled)
+ return;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ /* Step 1: Record the context, if any. */
+ ctx = 0;
+ if (iommu->iommu_ctxflush &&
+ strbuf->strbuf_ctxflush) {
+ iopte_t *iopte;
+
+ iopte = iommu->page_table +
+ ((sglist[0].dma_address - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
+ ctx = (iopte_val(*iopte) & IOPTE_CONTEXT) >> 47UL;
+ }
+
+ /* Step 2: Kick data out of streaming buffers. */
+ bus_addr = sglist[0].dma_address & IO_PAGE_MASK;
+ sgprv = NULL;
+ for_each_sg(sglist, sg, nelems, i) {
+ if (sg->dma_length == 0)
+ break;
+ sgprv = sg;
+ }
+
+ npages = (IO_PAGE_ALIGN(sgprv->dma_address + sgprv->dma_length)
+ - bus_addr) >> IO_PAGE_SHIFT;
+ strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static struct dma_map_ops sun4u_dma_ops = {
+ .alloc = dma_4u_alloc_coherent,
+ .free = dma_4u_free_coherent,
+ .map_page = dma_4u_map_page,
+ .unmap_page = dma_4u_unmap_page,
+ .map_sg = dma_4u_map_sg,
+ .unmap_sg = dma_4u_unmap_sg,
+ .sync_single_for_cpu = dma_4u_sync_single_for_cpu,
+ .sync_sg_for_cpu = dma_4u_sync_sg_for_cpu,
+};
+
+struct dma_map_ops *dma_ops = &sun4u_dma_ops;
+EXPORT_SYMBOL(dma_ops);
+
+int dma_supported(struct device *dev, u64 device_mask)
+{
+ struct iommu *iommu = dev->archdata.iommu;
+ u64 dma_addr_mask = iommu->dma_addr_mask;
+
+ if (device_mask >= (1UL << 32UL))
+ return 0;
+
+ if ((device_mask & dma_addr_mask) == dma_addr_mask)
+ return 1;
+
+#ifdef CONFIG_PCI
+ if (dev_is_pci(dev))
+ return pci64_dma_supported(to_pci_dev(dev), device_mask);
+#endif
+
+ return 0;
+}
+EXPORT_SYMBOL(dma_supported);
diff --git a/arch/sparc/kernel/iommu_common.h b/arch/sparc/kernel/iommu_common.h
new file mode 100644
index 00000000000..1ec0de4156e
--- /dev/null
+++ b/arch/sparc/kernel/iommu_common.h
@@ -0,0 +1,59 @@
+/* iommu_common.h: UltraSparc SBUS/PCI common iommu declarations.
+ *
+ * Copyright (C) 1999, 2008 David S. Miller (davem@davemloft.net)
+ */
+
+#ifndef _IOMMU_COMMON_H
+#define _IOMMU_COMMON_H
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/scatterlist.h>
+#include <linux/device.h>
+#include <linux/iommu-helper.h>
+
+#include <asm/iommu.h>
+#include <asm/scatterlist.h>
+
+/*
+ * These give mapping size of each iommu pte/tlb.
+ */
+#define IO_PAGE_SHIFT 13
+#define IO_PAGE_SIZE (1UL << IO_PAGE_SHIFT)
+#define IO_PAGE_MASK (~(IO_PAGE_SIZE-1))
+#define IO_PAGE_ALIGN(addr) ALIGN(addr, IO_PAGE_SIZE)
+
+#define IO_TSB_ENTRIES (128*1024)
+#define IO_TSB_SIZE (IO_TSB_ENTRIES * 8)
+
+/*
+ * This is the hardwired shift in the iotlb tag/data parts.
+ */
+#define IOMMU_PAGE_SHIFT 13
+
+#define SG_ENT_PHYS_ADDRESS(SG) (__pa(sg_virt((SG))))
+
+static inline int is_span_boundary(unsigned long entry,
+ unsigned long shift,
+ unsigned long boundary_size,
+ struct scatterlist *outs,
+ struct scatterlist *sg)
+{
+ unsigned long paddr = SG_ENT_PHYS_ADDRESS(outs);
+ int nr = iommu_num_pages(paddr, outs->dma_length + sg->length,
+ IO_PAGE_SIZE);
+
+ return iommu_is_span_boundary(entry, nr, shift, boundary_size);
+}
+
+unsigned long iommu_range_alloc(struct device *dev,
+ struct iommu *iommu,
+ unsigned long npages,
+ unsigned long *handle);
+void iommu_range_free(struct iommu *iommu,
+ dma_addr_t dma_addr,
+ unsigned long npages);
+
+#endif /* _IOMMU_COMMON_H */
diff --git a/arch/sparc/kernel/ioport.c b/arch/sparc/kernel/ioport.c
index ad0ede24ca1..7f08ec8a7c6 100644
--- a/arch/sparc/kernel/ioport.c
+++ b/arch/sparc/kernel/ioport.c
@@ -1,4 +1,4 @@
-/* $Id: ioport.c,v 1.45 2001/10/30 04:54:21 davem Exp $
+/*
* ioport.c: Simple io mapping allocator.
*
* Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
@@ -35,27 +35,41 @@
#include <linux/slab.h>
#include <linux/pci.h> /* struct pci_dev */
#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
#include <linux/scatterlist.h>
+#include <linux/of_device.h>
#include <asm/io.h>
#include <asm/vaddrs.h>
#include <asm/oplib.h>
#include <asm/prom.h>
-#include <asm/of_device.h>
-#include <asm/sbus.h>
#include <asm/page.h>
#include <asm/pgalloc.h>
#include <asm/dma.h>
+#include <asm/iommu.h>
+#include <asm/io-unit.h>
+#include <asm/leon.h>
-#define mmu_inval_dma_area(p, l) /* Anton pulled it out for 2.4.0-xx */
+const struct sparc32_dma_ops *sparc32_dma_ops;
-struct resource *_sparc_find_resource(struct resource *r, unsigned long);
+/* This function must make sure that caches and memory are coherent after DMA
+ * On LEON systems without cache snooping it flushes the entire D-CACHE.
+ */
+static inline void dma_make_coherent(unsigned long pa, unsigned long len)
+{
+ if (sparc_cpu_model == sparc_leon) {
+ if (!sparc_leon3_snooping_enabled())
+ leon_flush_dcache_all();
+ }
+}
static void __iomem *_sparc_ioremap(struct resource *res, u32 bus, u32 pa, int sz);
static void __iomem *_sparc_alloc_io(unsigned int busno, unsigned long phys,
unsigned long size, char *name);
static void _sparc_free_io(struct resource *res);
+static void register_proc_sparc_ioport(void);
+
/* This points to the next to use virtual memory for DVMA mappings */
static struct resource _sparc_dvma = {
.name = "sparc_dvma", .start = DVMA_VADDR, .end = DVMA_END - 1
@@ -114,6 +128,7 @@ void __iomem *ioremap(unsigned long offset, unsigned long size)
sprintf(name, "phys_%08x", (u32)offset);
return _sparc_alloc_io(0, offset, size, name);
}
+EXPORT_SYMBOL(ioremap);
/*
* Comlimentary to ioremap().
@@ -123,7 +138,11 @@ void iounmap(volatile void __iomem *virtual)
unsigned long vaddr = (unsigned long) virtual & PAGE_MASK;
struct resource *res;
- if ((res = _sparc_find_resource(&sparc_iomap, vaddr)) == NULL) {
+ /*
+ * XXX Too slow. Can have 8192 DVMA pages on sun4m in the worst case.
+ * This probably warrants some sort of hashing.
+ */
+ if ((res = lookup_resource(&sparc_iomap, vaddr)) == NULL) {
printk("free_io/iounmap: cannot free %lx\n", vaddr);
return;
}
@@ -135,15 +154,7 @@ void iounmap(volatile void __iomem *virtual)
kfree(res);
}
}
-
-/*
- */
-void __iomem *sbus_ioremap(struct resource *phyres, unsigned long offset,
- unsigned long size, char *name)
-{
- return _sparc_alloc_io(phyres->flags & 0xF,
- phyres->start + offset, size, name);
-}
+EXPORT_SYMBOL(iounmap);
void __iomem *of_ioremap(struct resource *res, unsigned long offset,
unsigned long size, char *name)
@@ -161,13 +172,6 @@ void of_iounmap(struct resource *res, void __iomem *base, unsigned long size)
EXPORT_SYMBOL(of_iounmap);
/*
- */
-void sbus_iounmap(volatile void __iomem *addr, unsigned long size)
-{
- iounmap(addr);
-}
-
-/*
* Meat of mapping
*/
static void __iomem *_sparc_alloc_io(unsigned int busno, unsigned long phys,
@@ -182,7 +186,7 @@ static void __iomem *_sparc_alloc_io(unsigned int busno, unsigned long phys,
if (name == NULL) name = "???";
- if ((xres = xres_alloc()) != 0) {
+ if ((xres = xres_alloc()) != NULL) {
tack = xres->xname;
res = &xres->xres;
} else {
@@ -223,7 +227,7 @@ _sparc_ioremap(struct resource *res, u32 bus, u32 pa, int sz)
}
pa &= PAGE_MASK;
- sparc_mapiorange(bus, pa, res->start, res->end - res->start + 1);
+ srmmu_mapiorange(bus, pa, res->start, resource_size(res));
return (void __iomem *)(unsigned long)(res->start + offset);
}
@@ -235,72 +239,31 @@ static void _sparc_free_io(struct resource *res)
{
unsigned long plen;
- plen = res->end - res->start + 1;
+ plen = resource_size(res);
BUG_ON((plen & (PAGE_SIZE-1)) != 0);
- sparc_unmapiorange(res->start, plen);
+ srmmu_unmapiorange(res->start, plen);
release_resource(res);
}
#ifdef CONFIG_SBUS
-void sbus_set_sbus64(struct sbus_dev *sdev, int x)
+void sbus_set_sbus64(struct device *dev, int x)
{
printk("sbus_set_sbus64: unsupported\n");
}
-
-extern unsigned int sun4d_build_irq(struct sbus_dev *sdev, int irq);
-void __init sbus_fill_device_irq(struct sbus_dev *sdev)
-{
- struct linux_prom_irqs irqs[PROMINTR_MAX];
- int len;
-
- len = prom_getproperty(sdev->prom_node, "intr",
- (char *)irqs, sizeof(irqs));
- if (len != -1) {
- sdev->num_irqs = len / 8;
- if (sdev->num_irqs == 0) {
- sdev->irqs[0] = 0;
- } else if (sparc_cpu_model == sun4d) {
- for (len = 0; len < sdev->num_irqs; len++)
- sdev->irqs[len] =
- sun4d_build_irq(sdev, irqs[len].pri);
- } else {
- for (len = 0; len < sdev->num_irqs; len++)
- sdev->irqs[len] = irqs[len].pri;
- }
- } else {
- int interrupts[PROMINTR_MAX];
-
- /* No "intr" node found-- check for "interrupts" node.
- * This node contains SBus interrupt levels, not IPLs
- * as in "intr", and no vector values. We convert
- * SBus interrupt levels to PILs (platform specific).
- */
- len = prom_getproperty(sdev->prom_node, "interrupts",
- (char *)interrupts, sizeof(interrupts));
- if (len == -1) {
- sdev->irqs[0] = 0;
- sdev->num_irqs = 0;
- } else {
- sdev->num_irqs = len / sizeof(int);
- for (len = 0; len < sdev->num_irqs; len++) {
- sdev->irqs[len] =
- sbint_to_irq(sdev, interrupts[len]);
- }
- }
- }
-}
+EXPORT_SYMBOL(sbus_set_sbus64);
/*
* Allocate a chunk of memory suitable for DMA.
* Typically devices use them for control blocks.
* CPU may access them without any explicit flushing.
- *
- * XXX Some clever people know that sdev is not used and supply NULL. Watch.
*/
-void *sbus_alloc_consistent(struct sbus_dev *sdev, long len, u32 *dma_addrp)
+static void *sbus_alloc_coherent(struct device *dev, size_t len,
+ dma_addr_t *dma_addrp, gfp_t gfp,
+ struct dma_attrs *attrs)
{
- unsigned long len_total = (len + PAGE_SIZE-1) & PAGE_MASK;
+ struct platform_device *op = to_platform_device(dev);
+ unsigned long len_total = PAGE_ALIGN(len);
unsigned long va;
struct resource *res;
int order;
@@ -326,39 +289,37 @@ void *sbus_alloc_consistent(struct sbus_dev *sdev, long len, u32 *dma_addrp)
printk("sbus_alloc_consistent: cannot occupy 0x%lx", len_total);
goto err_nova;
}
- mmu_inval_dma_area(va, len_total);
- // XXX The mmu_map_dma_area does this for us below, see comments.
- // sparc_mapiorange(0, virt_to_phys(va), res->start, len_total);
+
+ // XXX The sbus_map_dma_area does this for us below, see comments.
+ // srmmu_mapiorange(0, virt_to_phys(va), res->start, len_total);
/*
* XXX That's where sdev would be used. Currently we load
* all iommu tables with the same translations.
*/
- if (mmu_map_dma_area(dma_addrp, va, res->start, len_total) != 0)
+ if (sbus_map_dma_area(dev, dma_addrp, va, res->start, len_total) != 0)
goto err_noiommu;
- /* Set the resource name, if known. */
- if (sdev) {
- res->name = sdev->prom_name;
- }
+ res->name = op->dev.of_node->name;
return (void *)(unsigned long)res->start;
err_noiommu:
release_resource(res);
err_nova:
- free_pages(va, order);
-err_nomem:
kfree(res);
+err_nomem:
+ free_pages(va, order);
err_nopages:
return NULL;
}
-void sbus_free_consistent(struct sbus_dev *sdev, long n, void *p, u32 ba)
+static void sbus_free_coherent(struct device *dev, size_t n, void *p,
+ dma_addr_t ba, struct dma_attrs *attrs)
{
struct resource *res;
struct page *pgv;
- if ((res = _sparc_find_resource(&_sparc_dvma,
+ if ((res = lookup_resource(&_sparc_dvma,
(unsigned long)p)) == NULL) {
printk("sbus_free_consistent: cannot free %p\n", p);
return;
@@ -369,19 +330,18 @@ void sbus_free_consistent(struct sbus_dev *sdev, long n, void *p, u32 ba)
return;
}
- n = (n + PAGE_SIZE-1) & PAGE_MASK;
- if ((res->end-res->start)+1 != n) {
- printk("sbus_free_consistent: region 0x%lx asked 0x%lx\n",
- (long)((res->end-res->start)+1), n);
+ n = PAGE_ALIGN(n);
+ if (resource_size(res) != n) {
+ printk("sbus_free_consistent: region 0x%lx asked 0x%zx\n",
+ (long)resource_size(res), n);
return;
}
release_resource(res);
kfree(res);
- /* mmu_inval_dma_area(va, n); */ /* it's consistent, isn't it */
- pgv = mmu_translate_dvma(ba);
- mmu_unmap_dma_area(ba, n);
+ pgv = virt_to_page(p);
+ sbus_unmap_dma_area(dev, ba, n);
__free_pages(pgv, get_order(n));
}
@@ -391,8 +351,13 @@ void sbus_free_consistent(struct sbus_dev *sdev, long n, void *p, u32 ba)
* CPU view of this memory may be inconsistent with
* a device view and explicit flushing is necessary.
*/
-dma_addr_t sbus_map_single(struct sbus_dev *sdev, void *va, size_t len, int direction)
+static dma_addr_t sbus_map_page(struct device *dev, struct page *page,
+ unsigned long offset, size_t len,
+ enum dma_data_direction dir,
+ struct dma_attrs *attrs)
{
+ void *va = page_address(page) + offset;
+
/* XXX why are some lengths signed, others unsigned? */
if (len <= 0) {
return 0;
@@ -401,177 +366,72 @@ dma_addr_t sbus_map_single(struct sbus_dev *sdev, void *va, size_t len, int dire
if (len > 256*1024) { /* __get_free_pages() limit */
return 0;
}
- return mmu_get_scsi_one(va, len, sdev->bus);
+ return mmu_get_scsi_one(dev, va, len);
}
-void sbus_unmap_single(struct sbus_dev *sdev, dma_addr_t ba, size_t n, int direction)
+static void sbus_unmap_page(struct device *dev, dma_addr_t ba, size_t n,
+ enum dma_data_direction dir, struct dma_attrs *attrs)
{
- mmu_release_scsi_one(ba, n, sdev->bus);
+ mmu_release_scsi_one(dev, ba, n);
}
-int sbus_map_sg(struct sbus_dev *sdev, struct scatterlist *sg, int n, int direction)
+static int sbus_map_sg(struct device *dev, struct scatterlist *sg, int n,
+ enum dma_data_direction dir, struct dma_attrs *attrs)
{
- mmu_get_scsi_sgl(sg, n, sdev->bus);
-
- /*
- * XXX sparc64 can return a partial length here. sun4c should do this
- * but it currently panics if it can't fulfill the request - Anton
- */
+ mmu_get_scsi_sgl(dev, sg, n);
return n;
}
-void sbus_unmap_sg(struct sbus_dev *sdev, struct scatterlist *sg, int n, int direction)
+static void sbus_unmap_sg(struct device *dev, struct scatterlist *sg, int n,
+ enum dma_data_direction dir, struct dma_attrs *attrs)
{
- mmu_release_scsi_sgl(sg, n, sdev->bus);
+ mmu_release_scsi_sgl(dev, sg, n);
}
-/*
- */
-void sbus_dma_sync_single_for_cpu(struct sbus_dev *sdev, dma_addr_t ba, size_t size, int direction)
-{
-#if 0
- unsigned long va;
- struct resource *res;
-
- /* We do not need the resource, just print a message if invalid. */
- res = _sparc_find_resource(&_sparc_dvma, ba);
- if (res == NULL)
- panic("sbus_dma_sync_single: 0x%x\n", ba);
-
- va = page_address(mmu_translate_dvma(ba)); /* XXX higmem */
- /*
- * XXX This bogosity will be fixed with the iommu rewrite coming soon
- * to a kernel near you. - Anton
- */
- /* mmu_inval_dma_area(va, (size + PAGE_SIZE-1) & PAGE_MASK); */
-#endif
-}
-
-void sbus_dma_sync_single_for_device(struct sbus_dev *sdev, dma_addr_t ba, size_t size, int direction)
-{
-#if 0
- unsigned long va;
- struct resource *res;
-
- /* We do not need the resource, just print a message if invalid. */
- res = _sparc_find_resource(&_sparc_dvma, ba);
- if (res == NULL)
- panic("sbus_dma_sync_single: 0x%x\n", ba);
-
- va = page_address(mmu_translate_dvma(ba)); /* XXX higmem */
- /*
- * XXX This bogosity will be fixed with the iommu rewrite coming soon
- * to a kernel near you. - Anton
- */
- /* mmu_inval_dma_area(va, (size + PAGE_SIZE-1) & PAGE_MASK); */
-#endif
-}
-
-void sbus_dma_sync_sg_for_cpu(struct sbus_dev *sdev, struct scatterlist *sg, int n, int direction)
+static void sbus_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
+ int n, enum dma_data_direction dir)
{
- printk("sbus_dma_sync_sg_for_cpu: not implemented yet\n");
+ BUG();
}
-void sbus_dma_sync_sg_for_device(struct sbus_dev *sdev, struct scatterlist *sg, int n, int direction)
+static void sbus_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
+ int n, enum dma_data_direction dir)
{
- printk("sbus_dma_sync_sg_for_device: not implemented yet\n");
-}
-
-/* Support code for sbus_init(). */
-/*
- * XXX This functions appears to be a distorted version of
- * prom_sbus_ranges_init(), with all sun4d stuff cut away.
- * Ask DaveM what is going on here, how is sun4d supposed to work... XXX
- */
-/* added back sun4d patch from Thomas Bogendoerfer - should be OK (crn) */
-void __init sbus_arch_bus_ranges_init(struct device_node *pn, struct sbus_bus *sbus)
-{
- int parent_node = pn->node;
-
- if (sparc_cpu_model == sun4d) {
- struct linux_prom_ranges iounit_ranges[PROMREG_MAX];
- int num_iounit_ranges, len;
-
- len = prom_getproperty(parent_node, "ranges",
- (char *) iounit_ranges,
- sizeof (iounit_ranges));
- if (len != -1) {
- num_iounit_ranges =
- (len / sizeof(struct linux_prom_ranges));
- prom_adjust_ranges(sbus->sbus_ranges,
- sbus->num_sbus_ranges,
- iounit_ranges, num_iounit_ranges);
- }
- }
-}
-
-void __init sbus_setup_iommu(struct sbus_bus *sbus, struct device_node *dp)
-{
-#ifndef CONFIG_SUN4
- struct device_node *parent = dp->parent;
-
- if (sparc_cpu_model != sun4d &&
- parent != NULL &&
- !strcmp(parent->name, "iommu")) {
- extern void iommu_init(int iommu_node, struct sbus_bus *sbus);
-
- iommu_init(parent->node, sbus);
- }
-
- if (sparc_cpu_model == sun4d) {
- extern void iounit_init(int sbi_node, int iounit_node,
- struct sbus_bus *sbus);
-
- iounit_init(dp->node, parent->node, sbus);
- }
-#endif
+ BUG();
}
-void __init sbus_setup_arch_props(struct sbus_bus *sbus, struct device_node *dp)
-{
- if (sparc_cpu_model == sun4d) {
- struct device_node *parent = dp->parent;
-
- sbus->devid = of_getintprop_default(parent, "device-id", 0);
- sbus->board = of_getintprop_default(parent, "board#", 0);
- }
-}
+static struct dma_map_ops sbus_dma_ops = {
+ .alloc = sbus_alloc_coherent,
+ .free = sbus_free_coherent,
+ .map_page = sbus_map_page,
+ .unmap_page = sbus_unmap_page,
+ .map_sg = sbus_map_sg,
+ .unmap_sg = sbus_unmap_sg,
+ .sync_sg_for_cpu = sbus_sync_sg_for_cpu,
+ .sync_sg_for_device = sbus_sync_sg_for_device,
+};
-int __init sbus_arch_preinit(void)
+static int __init sparc_register_ioport(void)
{
- extern void register_proc_sparc_ioport(void);
-
register_proc_sparc_ioport();
-#ifdef CONFIG_SUN4
- {
- extern void sun4_dvma_init(void);
- sun4_dvma_init();
- }
- return 1;
-#else
return 0;
-#endif
}
-void __init sbus_arch_postinit(void)
-{
- if (sparc_cpu_model == sun4d) {
- extern void sun4d_init_sbi_irq(void);
- sun4d_init_sbi_irq();
- }
-}
+arch_initcall(sparc_register_ioport);
+
#endif /* CONFIG_SBUS */
-#ifdef CONFIG_PCI
/* Allocate and map kernel buffer using consistent mode DMA for a device.
* hwdev should be valid struct pci_dev pointer for PCI devices.
*/
-void *pci_alloc_consistent(struct pci_dev *pdev, size_t len, dma_addr_t *pba)
+static void *pci32_alloc_coherent(struct device *dev, size_t len,
+ dma_addr_t *pba, gfp_t gfp,
+ struct dma_attrs *attrs)
{
- unsigned long len_total = (len + PAGE_SIZE-1) & PAGE_MASK;
- unsigned long va;
+ unsigned long len_total = PAGE_ALIGN(len);
+ void *va;
struct resource *res;
int order;
@@ -583,34 +443,33 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t len, dma_addr_t *pba)
}
order = get_order(len_total);
- va = __get_free_pages(GFP_KERNEL, order);
- if (va == 0) {
+ va = (void *) __get_free_pages(GFP_KERNEL, order);
+ if (va == NULL) {
printk("pci_alloc_consistent: no %ld pages\n", len_total>>PAGE_SHIFT);
- return NULL;
+ goto err_nopages;
}
if ((res = kzalloc(sizeof(struct resource), GFP_KERNEL)) == NULL) {
- free_pages(va, order);
printk("pci_alloc_consistent: no core\n");
- return NULL;
+ goto err_nomem;
}
if (allocate_resource(&_sparc_dvma, res, len_total,
_sparc_dvma.start, _sparc_dvma.end, PAGE_SIZE, NULL, NULL) != 0) {
printk("pci_alloc_consistent: cannot occupy 0x%lx", len_total);
- free_pages(va, order);
- kfree(res);
- return NULL;
+ goto err_nova;
}
- mmu_inval_dma_area(va, len_total);
-#if 0
-/* P3 */ printk("pci_alloc_consistent: kva %lx uncva %lx phys %lx size %lx\n",
- (long)va, (long)res->start, (long)virt_to_phys(va), len_total);
-#endif
- sparc_mapiorange(0, virt_to_phys(va), res->start, len_total);
+ srmmu_mapiorange(0, virt_to_phys(va), res->start, len_total);
*pba = virt_to_phys(va); /* equals virt_to_bus (R.I.P.) for us. */
return (void *) res->start;
+
+err_nova:
+ kfree(res);
+err_nomem:
+ free_pages((unsigned long)va, order);
+err_nopages:
+ return NULL;
}
/* Free and unmap a consistent DMA buffer.
@@ -621,12 +480,12 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t len, dma_addr_t *pba)
* References to the memory and mappings associated with cpu_addr/dma_addr
* past this call are illegal.
*/
-void pci_free_consistent(struct pci_dev *pdev, size_t n, void *p, dma_addr_t ba)
+static void pci32_free_coherent(struct device *dev, size_t n, void *p,
+ dma_addr_t ba, struct dma_attrs *attrs)
{
struct resource *res;
- unsigned long pgp;
- if ((res = _sparc_find_resource(&_sparc_dvma,
+ if ((res = lookup_resource(&_sparc_dvma,
(unsigned long)p)) == NULL) {
printk("pci_free_consistent: cannot free %p\n", p);
return;
@@ -637,70 +496,38 @@ void pci_free_consistent(struct pci_dev *pdev, size_t n, void *p, dma_addr_t ba)
return;
}
- n = (n + PAGE_SIZE-1) & PAGE_MASK;
- if ((res->end-res->start)+1 != n) {
+ n = PAGE_ALIGN(n);
+ if (resource_size(res) != n) {
printk("pci_free_consistent: region 0x%lx asked 0x%lx\n",
- (long)((res->end-res->start)+1), (long)n);
+ (long)resource_size(res), (long)n);
return;
}
- pgp = (unsigned long) phys_to_virt(ba); /* bus_to_virt actually */
- mmu_inval_dma_area(pgp, n);
- sparc_unmapiorange((unsigned long)p, n);
+ dma_make_coherent(ba, n);
+ srmmu_unmapiorange((unsigned long)p, n);
release_resource(res);
kfree(res);
-
- free_pages(pgp, get_order(n));
-}
-
-/* Map a single buffer of the indicated size for DMA in streaming mode.
- * The 32-bit bus address to use is returned.
- *
- * Once the device is given the dma address, the device owns this memory
- * until either pci_unmap_single or pci_dma_sync_single_* is performed.
- */
-dma_addr_t pci_map_single(struct pci_dev *hwdev, void *ptr, size_t size,
- int direction)
-{
- BUG_ON(direction == PCI_DMA_NONE);
- /* IIep is write-through, not flushing. */
- return virt_to_phys(ptr);
-}
-
-/* Unmap a single streaming mode DMA translation. The dma_addr and size
- * must match what was provided for in a previous pci_map_single call. All
- * other usages are undefined.
- *
- * After this call, reads by the cpu to the buffer are guaranteed to see
- * whatever the device wrote there.
- */
-void pci_unmap_single(struct pci_dev *hwdev, dma_addr_t ba, size_t size,
- int direction)
-{
- BUG_ON(direction == PCI_DMA_NONE);
- if (direction != PCI_DMA_TODEVICE) {
- mmu_inval_dma_area((unsigned long)phys_to_virt(ba),
- (size + PAGE_SIZE-1) & PAGE_MASK);
- }
+ free_pages((unsigned long)phys_to_virt(ba), get_order(n));
}
/*
* Same as pci_map_single, but with pages.
*/
-dma_addr_t pci_map_page(struct pci_dev *hwdev, struct page *page,
- unsigned long offset, size_t size, int direction)
+static dma_addr_t pci32_map_page(struct device *dev, struct page *page,
+ unsigned long offset, size_t size,
+ enum dma_data_direction dir,
+ struct dma_attrs *attrs)
{
- BUG_ON(direction == PCI_DMA_NONE);
/* IIep is write-through, not flushing. */
return page_to_phys(page) + offset;
}
-void pci_unmap_page(struct pci_dev *hwdev,
- dma_addr_t dma_address, size_t size, int direction)
+static void pci32_unmap_page(struct device *dev, dma_addr_t ba, size_t size,
+ enum dma_data_direction dir, struct dma_attrs *attrs)
{
- BUG_ON(direction == PCI_DMA_NONE);
- /* mmu_inval_dma_area XXX */
+ if (dir != PCI_DMA_TODEVICE)
+ dma_make_coherent(ba, PAGE_ALIGN(size));
}
/* Map a set of buffers described by scatterlist in streaming
@@ -718,18 +545,17 @@ void pci_unmap_page(struct pci_dev *hwdev,
* Device ownership issues as mentioned above for pci_map_single are
* the same here.
*/
-int pci_map_sg(struct pci_dev *hwdev, struct scatterlist *sgl, int nents,
- int direction)
+static int pci32_map_sg(struct device *device, struct scatterlist *sgl,
+ int nents, enum dma_data_direction dir,
+ struct dma_attrs *attrs)
{
struct scatterlist *sg;
int n;
- BUG_ON(direction == PCI_DMA_NONE);
/* IIep is write-through, not flushing. */
for_each_sg(sgl, sg, nents, n) {
- BUG_ON(page_address(sg_page(sg)) == NULL);
- sg->dvma_address = virt_to_phys(sg_virt(sg));
- sg->dvma_length = sg->length;
+ sg->dma_address = sg_phys(sg);
+ sg->dma_length = sg->length;
}
return nents;
}
@@ -738,19 +564,16 @@ int pci_map_sg(struct pci_dev *hwdev, struct scatterlist *sgl, int nents,
* Again, cpu read rules concerning calls here are the same as for
* pci_unmap_single() above.
*/
-void pci_unmap_sg(struct pci_dev *hwdev, struct scatterlist *sgl, int nents,
- int direction)
+static void pci32_unmap_sg(struct device *dev, struct scatterlist *sgl,
+ int nents, enum dma_data_direction dir,
+ struct dma_attrs *attrs)
{
struct scatterlist *sg;
int n;
- BUG_ON(direction == PCI_DMA_NONE);
- if (direction != PCI_DMA_TODEVICE) {
+ if (dir != PCI_DMA_TODEVICE) {
for_each_sg(sgl, sg, nents, n) {
- BUG_ON(page_address(sg_page(sg)) == NULL);
- mmu_inval_dma_area(
- (unsigned long) page_address(sg_page(sg)),
- (sg->length + PAGE_SIZE-1) & PAGE_MASK);
+ dma_make_coherent(sg_phys(sg), PAGE_ALIGN(sg->length));
}
}
}
@@ -765,21 +588,19 @@ void pci_unmap_sg(struct pci_dev *hwdev, struct scatterlist *sgl, int nents,
* must first perform a pci_dma_sync_for_device, and then the
* device again owns the buffer.
*/
-void pci_dma_sync_single_for_cpu(struct pci_dev *hwdev, dma_addr_t ba, size_t size, int direction)
+static void pci32_sync_single_for_cpu(struct device *dev, dma_addr_t ba,
+ size_t size, enum dma_data_direction dir)
{
- BUG_ON(direction == PCI_DMA_NONE);
- if (direction != PCI_DMA_TODEVICE) {
- mmu_inval_dma_area((unsigned long)phys_to_virt(ba),
- (size + PAGE_SIZE-1) & PAGE_MASK);
+ if (dir != PCI_DMA_TODEVICE) {
+ dma_make_coherent(ba, PAGE_ALIGN(size));
}
}
-void pci_dma_sync_single_for_device(struct pci_dev *hwdev, dma_addr_t ba, size_t size, int direction)
+static void pci32_sync_single_for_device(struct device *dev, dma_addr_t ba,
+ size_t size, enum dma_data_direction dir)
{
- BUG_ON(direction == PCI_DMA_NONE);
- if (direction != PCI_DMA_TODEVICE) {
- mmu_inval_dma_area((unsigned long)phys_to_virt(ba),
- (size + PAGE_SIZE-1) & PAGE_MASK);
+ if (dir != PCI_DMA_TODEVICE) {
+ dma_make_coherent(ba, PAGE_ALIGN(size));
}
}
@@ -789,86 +610,104 @@ void pci_dma_sync_single_for_device(struct pci_dev *hwdev, dma_addr_t ba, size_t
* The same as pci_dma_sync_single_* but for a scatter-gather list,
* same rules and usage.
*/
-void pci_dma_sync_sg_for_cpu(struct pci_dev *hwdev, struct scatterlist *sgl, int nents, int direction)
+static void pci32_sync_sg_for_cpu(struct device *dev, struct scatterlist *sgl,
+ int nents, enum dma_data_direction dir)
{
struct scatterlist *sg;
int n;
- BUG_ON(direction == PCI_DMA_NONE);
- if (direction != PCI_DMA_TODEVICE) {
+ if (dir != PCI_DMA_TODEVICE) {
for_each_sg(sgl, sg, nents, n) {
- BUG_ON(page_address(sg_page(sg)) == NULL);
- mmu_inval_dma_area(
- (unsigned long) page_address(sg_page(sg)),
- (sg->length + PAGE_SIZE-1) & PAGE_MASK);
+ dma_make_coherent(sg_phys(sg), PAGE_ALIGN(sg->length));
}
}
}
-void pci_dma_sync_sg_for_device(struct pci_dev *hwdev, struct scatterlist *sgl, int nents, int direction)
+static void pci32_sync_sg_for_device(struct device *device, struct scatterlist *sgl,
+ int nents, enum dma_data_direction dir)
{
struct scatterlist *sg;
int n;
- BUG_ON(direction == PCI_DMA_NONE);
- if (direction != PCI_DMA_TODEVICE) {
+ if (dir != PCI_DMA_TODEVICE) {
for_each_sg(sgl, sg, nents, n) {
- BUG_ON(page_address(sg_page(sg)) == NULL);
- mmu_inval_dma_area(
- (unsigned long) page_address(sg_page(sg)),
- (sg->length + PAGE_SIZE-1) & PAGE_MASK);
+ dma_make_coherent(sg_phys(sg), PAGE_ALIGN(sg->length));
}
}
}
-#endif /* CONFIG_PCI */
+
+struct dma_map_ops pci32_dma_ops = {
+ .alloc = pci32_alloc_coherent,
+ .free = pci32_free_coherent,
+ .map_page = pci32_map_page,
+ .unmap_page = pci32_unmap_page,
+ .map_sg = pci32_map_sg,
+ .unmap_sg = pci32_unmap_sg,
+ .sync_single_for_cpu = pci32_sync_single_for_cpu,
+ .sync_single_for_device = pci32_sync_single_for_device,
+ .sync_sg_for_cpu = pci32_sync_sg_for_cpu,
+ .sync_sg_for_device = pci32_sync_sg_for_device,
+};
+EXPORT_SYMBOL(pci32_dma_ops);
+
+/* leon re-uses pci32_dma_ops */
+struct dma_map_ops *leon_dma_ops = &pci32_dma_ops;
+EXPORT_SYMBOL(leon_dma_ops);
+
+struct dma_map_ops *dma_ops = &sbus_dma_ops;
+EXPORT_SYMBOL(dma_ops);
+
+
+/*
+ * Return whether the given PCI device DMA address mask can be
+ * supported properly. For example, if your device can only drive the
+ * low 24-bits during PCI bus mastering, then you would pass
+ * 0x00ffffff as the mask to this function.
+ */
+int dma_supported(struct device *dev, u64 mask)
+{
+ if (dev_is_pci(dev))
+ return 1;
+
+ return 0;
+}
+EXPORT_SYMBOL(dma_supported);
#ifdef CONFIG_PROC_FS
-static int
-_sparc_io_get_info(char *buf, char **start, off_t fpos, int length, int *eof,
- void *data)
+static int sparc_io_proc_show(struct seq_file *m, void *v)
{
- char *p = buf, *e = buf + length;
- struct resource *r;
+ struct resource *root = m->private, *r;
const char *nm;
- for (r = ((struct resource *)data)->child; r != NULL; r = r->sibling) {
- if (p + 32 >= e) /* Better than nothing */
- break;
- if ((nm = r->name) == 0) nm = "???";
- p += sprintf(p, "%016llx-%016llx: %s\n",
+ for (r = root->child; r != NULL; r = r->sibling) {
+ if ((nm = r->name) == NULL) nm = "???";
+ seq_printf(m, "%016llx-%016llx: %s\n",
(unsigned long long)r->start,
(unsigned long long)r->end, nm);
}
- return p-buf;
+ return 0;
}
-#endif /* CONFIG_PROC_FS */
-
-/*
- * This is a version of find_resource and it belongs to kernel/resource.c.
- * Until we have agreement with Linus and Martin, it lingers here.
- *
- * XXX Too slow. Can have 8192 DVMA pages on sun4m in the worst case.
- * This probably warrants some sort of hashing.
- */
-struct resource *
-_sparc_find_resource(struct resource *root, unsigned long hit)
+static int sparc_io_proc_open(struct inode *inode, struct file *file)
{
- struct resource *tmp;
-
- for (tmp = root->child; tmp != 0; tmp = tmp->sibling) {
- if (tmp->start <= hit && tmp->end >= hit)
- return tmp;
- }
- return NULL;
+ return single_open(file, sparc_io_proc_show, PDE_DATA(inode));
}
-void register_proc_sparc_ioport(void)
+static const struct file_operations sparc_io_proc_fops = {
+ .owner = THIS_MODULE,
+ .open = sparc_io_proc_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = single_release,
+};
+#endif /* CONFIG_PROC_FS */
+
+static void register_proc_sparc_ioport(void)
{
#ifdef CONFIG_PROC_FS
- create_proc_read_entry("io_map",0,NULL,_sparc_io_get_info,&sparc_iomap);
- create_proc_read_entry("dvma_map",0,NULL,_sparc_io_get_info,&_sparc_dvma);
+ proc_create_data("io_map", 0, NULL, &sparc_io_proc_fops, &sparc_iomap);
+ proc_create_data("dvma_map", 0, NULL, &sparc_io_proc_fops, &_sparc_dvma);
#endif
}
diff --git a/arch/sparc/kernel/irq.c b/arch/sparc/kernel/irq.c
deleted file mode 100644
index 01a6756ba37..00000000000
--- a/arch/sparc/kernel/irq.c
+++ /dev/null
@@ -1,674 +0,0 @@
-/* $Id: irq.c,v 1.114 2001/12/11 04:55:51 davem Exp $
- * arch/sparc/kernel/irq.c: Interrupt request handling routines. On the
- * Sparc the IRQs are basically 'cast in stone'
- * and you are supposed to probe the prom's device
- * node trees to find out who's got which IRQ.
- *
- * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
- * Copyright (C) 1995 Miguel de Icaza (miguel@nuclecu.unam.mx)
- * Copyright (C) 1995,2002 Pete A. Zaitcev (zaitcev@yahoo.com)
- * Copyright (C) 1996 Dave Redman (djhr@tadpole.co.uk)
- * Copyright (C) 1998-2000 Anton Blanchard (anton@samba.org)
- */
-
-#include <linux/module.h>
-#include <linux/sched.h>
-#include <linux/ptrace.h>
-#include <linux/errno.h>
-#include <linux/linkage.h>
-#include <linux/kernel_stat.h>
-#include <linux/signal.h>
-#include <linux/interrupt.h>
-#include <linux/slab.h>
-#include <linux/random.h>
-#include <linux/init.h>
-#include <linux/smp.h>
-#include <linux/delay.h>
-#include <linux/threads.h>
-#include <linux/spinlock.h>
-#include <linux/seq_file.h>
-
-#include <asm/ptrace.h>
-#include <asm/processor.h>
-#include <asm/system.h>
-#include <asm/psr.h>
-#include <asm/smp.h>
-#include <asm/vaddrs.h>
-#include <asm/timer.h>
-#include <asm/openprom.h>
-#include <asm/oplib.h>
-#include <asm/traps.h>
-#include <asm/irq.h>
-#include <asm/io.h>
-#include <asm/pgalloc.h>
-#include <asm/pgtable.h>
-#include <asm/pcic.h>
-#include <asm/cacheflush.h>
-#include <asm/irq_regs.h>
-
-#include "irq.h"
-
-#ifdef CONFIG_SMP
-#define SMP_NOP2 "nop; nop;\n\t"
-#define SMP_NOP3 "nop; nop; nop;\n\t"
-#else
-#define SMP_NOP2
-#define SMP_NOP3
-#endif /* SMP */
-unsigned long __raw_local_irq_save(void)
-{
- unsigned long retval;
- unsigned long tmp;
-
- __asm__ __volatile__(
- "rd %%psr, %0\n\t"
- SMP_NOP3 /* Sun4m + Cypress + SMP bug */
- "or %0, %2, %1\n\t"
- "wr %1, 0, %%psr\n\t"
- "nop; nop; nop\n"
- : "=&r" (retval), "=r" (tmp)
- : "i" (PSR_PIL)
- : "memory");
-
- return retval;
-}
-
-void raw_local_irq_enable(void)
-{
- unsigned long tmp;
-
- __asm__ __volatile__(
- "rd %%psr, %0\n\t"
- SMP_NOP3 /* Sun4m + Cypress + SMP bug */
- "andn %0, %1, %0\n\t"
- "wr %0, 0, %%psr\n\t"
- "nop; nop; nop\n"
- : "=&r" (tmp)
- : "i" (PSR_PIL)
- : "memory");
-}
-
-void raw_local_irq_restore(unsigned long old_psr)
-{
- unsigned long tmp;
-
- __asm__ __volatile__(
- "rd %%psr, %0\n\t"
- "and %2, %1, %2\n\t"
- SMP_NOP2 /* Sun4m + Cypress + SMP bug */
- "andn %0, %1, %0\n\t"
- "wr %0, %2, %%psr\n\t"
- "nop; nop; nop\n"
- : "=&r" (tmp)
- : "i" (PSR_PIL), "r" (old_psr)
- : "memory");
-}
-
-EXPORT_SYMBOL(__raw_local_irq_save);
-EXPORT_SYMBOL(raw_local_irq_enable);
-EXPORT_SYMBOL(raw_local_irq_restore);
-
-/*
- * Dave Redman (djhr@tadpole.co.uk)
- *
- * IRQ numbers.. These are no longer restricted to 15..
- *
- * this is done to enable SBUS cards and onboard IO to be masked
- * correctly. using the interrupt level isn't good enough.
- *
- * For example:
- * A device interrupting at sbus level6 and the Floppy both come in
- * at IRQ11, but enabling and disabling them requires writing to
- * different bits in the SLAVIO/SEC.
- *
- * As a result of these changes sun4m machines could now support
- * directed CPU interrupts using the existing enable/disable irq code
- * with tweaks.
- *
- */
-
-static void irq_panic(void)
-{
- extern char *cputypval;
- prom_printf("machine: %s doesn't have irq handlers defined!\n",cputypval);
- prom_halt();
-}
-
-void (*sparc_init_timers)(irq_handler_t ) =
- (void (*)(irq_handler_t )) irq_panic;
-
-/*
- * Dave Redman (djhr@tadpole.co.uk)
- *
- * There used to be extern calls and hard coded values here.. very sucky!
- * instead, because some of the devices attach very early, I do something
- * equally sucky but at least we'll never try to free statically allocated
- * space or call kmalloc before kmalloc_init :(.
- *
- * In fact it's the timer10 that attaches first.. then timer14
- * then kmalloc_init is called.. then the tty interrupts attach.
- * hmmm....
- *
- */
-#define MAX_STATIC_ALLOC 4
-struct irqaction static_irqaction[MAX_STATIC_ALLOC];
-int static_irq_count;
-
-struct {
- struct irqaction *action;
- int flags;
-} sparc_irq[NR_IRQS];
-#define SPARC_IRQ_INPROGRESS 1
-
-/* Used to protect the IRQ action lists */
-DEFINE_SPINLOCK(irq_action_lock);
-
-int show_interrupts(struct seq_file *p, void *v)
-{
- int i = *(loff_t *) v;
- struct irqaction * action;
- unsigned long flags;
-#ifdef CONFIG_SMP
- int j;
-#endif
-
- if (sparc_cpu_model == sun4d) {
- extern int show_sun4d_interrupts(struct seq_file *, void *);
-
- return show_sun4d_interrupts(p, v);
- }
- spin_lock_irqsave(&irq_action_lock, flags);
- if (i < NR_IRQS) {
- action = sparc_irq[i].action;
- if (!action)
- goto out_unlock;
- seq_printf(p, "%3d: ", i);
-#ifndef CONFIG_SMP
- seq_printf(p, "%10u ", kstat_irqs(i));
-#else
- for_each_online_cpu(j) {
- seq_printf(p, "%10u ",
- kstat_cpu(j).irqs[i]);
- }
-#endif
- seq_printf(p, " %c %s",
- (action->flags & IRQF_DISABLED) ? '+' : ' ',
- action->name);
- for (action=action->next; action; action = action->next) {
- seq_printf(p, ",%s %s",
- (action->flags & IRQF_DISABLED) ? " +" : "",
- action->name);
- }
- seq_putc(p, '\n');
- }
-out_unlock:
- spin_unlock_irqrestore(&irq_action_lock, flags);
- return 0;
-}
-
-void free_irq(unsigned int irq, void *dev_id)
-{
- struct irqaction * action;
- struct irqaction **actionp;
- unsigned long flags;
- unsigned int cpu_irq;
-
- if (sparc_cpu_model == sun4d) {
- extern void sun4d_free_irq(unsigned int, void *);
-
- sun4d_free_irq(irq, dev_id);
- return;
- }
- cpu_irq = irq & (NR_IRQS - 1);
- if (cpu_irq > 14) { /* 14 irq levels on the sparc */
- printk("Trying to free bogus IRQ %d\n", irq);
- return;
- }
-
- spin_lock_irqsave(&irq_action_lock, flags);
-
- actionp = &sparc_irq[cpu_irq].action;
- action = *actionp;
-
- if (!action->handler) {
- printk("Trying to free free IRQ%d\n",irq);
- goto out_unlock;
- }
- if (dev_id) {
- for (; action; action = action->next) {
- if (action->dev_id == dev_id)
- break;
- actionp = &action->next;
- }
- if (!action) {
- printk("Trying to free free shared IRQ%d\n",irq);
- goto out_unlock;
- }
- } else if (action->flags & IRQF_SHARED) {
- printk("Trying to free shared IRQ%d with NULL device ID\n", irq);
- goto out_unlock;
- }
- if (action->flags & SA_STATIC_ALLOC)
- {
- /* This interrupt is marked as specially allocated
- * so it is a bad idea to free it.
- */
- printk("Attempt to free statically allocated IRQ%d (%s)\n",
- irq, action->name);
- goto out_unlock;
- }
-
- *actionp = action->next;
-
- spin_unlock_irqrestore(&irq_action_lock, flags);
-
- synchronize_irq(irq);
-
- spin_lock_irqsave(&irq_action_lock, flags);
-
- kfree(action);
-
- if (!sparc_irq[cpu_irq].action)
- __disable_irq(irq);
-
-out_unlock:
- spin_unlock_irqrestore(&irq_action_lock, flags);
-}
-
-EXPORT_SYMBOL(free_irq);
-
-/*
- * This is called when we want to synchronize with
- * interrupts. We may for example tell a device to
- * stop sending interrupts: but to make sure there
- * are no interrupts that are executing on another
- * CPU we need to call this function.
- */
-#ifdef CONFIG_SMP
-void synchronize_irq(unsigned int irq)
-{
- unsigned int cpu_irq;
-
- cpu_irq = irq & (NR_IRQS - 1);
- while (sparc_irq[cpu_irq].flags & SPARC_IRQ_INPROGRESS)
- cpu_relax();
-}
-#endif /* SMP */
-
-void unexpected_irq(int irq, void *dev_id, struct pt_regs * regs)
-{
- int i;
- struct irqaction * action;
- unsigned int cpu_irq;
-
- cpu_irq = irq & (NR_IRQS - 1);
- action = sparc_irq[cpu_irq].action;
-
- printk("IO device interrupt, irq = %d\n", irq);
- printk("PC = %08lx NPC = %08lx FP=%08lx\n", regs->pc,
- regs->npc, regs->u_regs[14]);
- if (action) {
- printk("Expecting: ");
- for (i = 0; i < 16; i++)
- if (action->handler)
- printk("[%s:%d:0x%x] ", action->name,
- (int) i, (unsigned int) action->handler);
- }
- printk("AIEEE\n");
- panic("bogus interrupt received");
-}
-
-void handler_irq(int irq, struct pt_regs * regs)
-{
- struct pt_regs *old_regs;
- struct irqaction * action;
- int cpu = smp_processor_id();
-#ifdef CONFIG_SMP
- extern void smp4m_irq_rotate(int cpu);
-#endif
-
- old_regs = set_irq_regs(regs);
- irq_enter();
- disable_pil_irq(irq);
-#ifdef CONFIG_SMP
- /* Only rotate on lower priority IRQs (scsi, ethernet, etc.). */
- if((sparc_cpu_model==sun4m) && (irq < 10))
- smp4m_irq_rotate(cpu);
-#endif
- action = sparc_irq[irq].action;
- sparc_irq[irq].flags |= SPARC_IRQ_INPROGRESS;
- kstat_cpu(cpu).irqs[irq]++;
- do {
- if (!action || !action->handler)
- unexpected_irq(irq, NULL, regs);
- action->handler(irq, action->dev_id);
- action = action->next;
- } while (action);
- sparc_irq[irq].flags &= ~SPARC_IRQ_INPROGRESS;
- enable_pil_irq(irq);
- irq_exit();
- set_irq_regs(old_regs);
-}
-
-#if defined(CONFIG_BLK_DEV_FD) || defined(CONFIG_BLK_DEV_FD_MODULE)
-
-/* Fast IRQs on the Sparc can only have one routine attached to them,
- * thus no sharing possible.
- */
-static int request_fast_irq(unsigned int irq,
- void (*handler)(void),
- unsigned long irqflags, const char *devname)
-{
- struct irqaction *action;
- unsigned long flags;
- unsigned int cpu_irq;
- int ret;
-#ifdef CONFIG_SMP
- struct tt_entry *trap_table;
- extern struct tt_entry trapbase_cpu1, trapbase_cpu2, trapbase_cpu3;
-#endif
-
- cpu_irq = irq & (NR_IRQS - 1);
- if(cpu_irq > 14) {
- ret = -EINVAL;
- goto out;
- }
- if(!handler) {
- ret = -EINVAL;
- goto out;
- }
-
- spin_lock_irqsave(&irq_action_lock, flags);
-
- action = sparc_irq[cpu_irq].action;
- if(action) {
- if(action->flags & IRQF_SHARED)
- panic("Trying to register fast irq when already shared.\n");
- if(irqflags & IRQF_SHARED)
- panic("Trying to register fast irq as shared.\n");
-
- /* Anyway, someone already owns it so cannot be made fast. */
- printk("request_fast_irq: Trying to register yet already owned.\n");
- ret = -EBUSY;
- goto out_unlock;
- }
-
- /* If this is flagged as statically allocated then we use our
- * private struct which is never freed.
- */
- if (irqflags & SA_STATIC_ALLOC) {
- if (static_irq_count < MAX_STATIC_ALLOC)
- action = &static_irqaction[static_irq_count++];
- else
- printk("Fast IRQ%d (%s) SA_STATIC_ALLOC failed using kmalloc\n",
- irq, devname);
- }
-
- if (action == NULL)
- action = kmalloc(sizeof(struct irqaction),
- GFP_ATOMIC);
-
- if (!action) {
- ret = -ENOMEM;
- goto out_unlock;
- }
-
- /* Dork with trap table if we get this far. */
-#define INSTANTIATE(table) \
- table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_one = SPARC_RD_PSR_L0; \
- table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_two = \
- SPARC_BRANCH((unsigned long) handler, \
- (unsigned long) &table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_two);\
- table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_three = SPARC_RD_WIM_L3; \
- table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_four = SPARC_NOP;
-
- INSTANTIATE(sparc_ttable)
-#ifdef CONFIG_SMP
- trap_table = &trapbase_cpu1; INSTANTIATE(trap_table)
- trap_table = &trapbase_cpu2; INSTANTIATE(trap_table)
- trap_table = &trapbase_cpu3; INSTANTIATE(trap_table)
-#endif
-#undef INSTANTIATE
- /*
- * XXX Correct thing whould be to flush only I- and D-cache lines
- * which contain the handler in question. But as of time of the
- * writing we have no CPU-neutral interface to fine-grained flushes.
- */
- flush_cache_all();
-
- action->flags = irqflags;
- cpus_clear(action->mask);
- action->name = devname;
- action->dev_id = NULL;
- action->next = NULL;
-
- sparc_irq[cpu_irq].action = action;
-
- __enable_irq(irq);
-
- ret = 0;
-out_unlock:
- spin_unlock_irqrestore(&irq_action_lock, flags);
-out:
- return ret;
-}
-
-/* These variables are used to access state from the assembler
- * interrupt handler, floppy_hardint, so we cannot put these in
- * the floppy driver image because that would not work in the
- * modular case.
- */
-volatile unsigned char *fdc_status;
-EXPORT_SYMBOL(fdc_status);
-
-char *pdma_vaddr;
-EXPORT_SYMBOL(pdma_vaddr);
-
-unsigned long pdma_size;
-EXPORT_SYMBOL(pdma_size);
-
-volatile int doing_pdma;
-EXPORT_SYMBOL(doing_pdma);
-
-char *pdma_base;
-EXPORT_SYMBOL(pdma_base);
-
-unsigned long pdma_areasize;
-EXPORT_SYMBOL(pdma_areasize);
-
-extern void floppy_hardint(void);
-
-static irq_handler_t floppy_irq_handler;
-
-void sparc_floppy_irq(int irq, void *dev_id, struct pt_regs *regs)
-{
- struct pt_regs *old_regs;
- int cpu = smp_processor_id();
-
- old_regs = set_irq_regs(regs);
- disable_pil_irq(irq);
- irq_enter();
- kstat_cpu(cpu).irqs[irq]++;
- floppy_irq_handler(irq, dev_id);
- irq_exit();
- enable_pil_irq(irq);
- set_irq_regs(old_regs);
- // XXX Eek, it's totally changed with preempt_count() and such
- // if (softirq_pending(cpu))
- // do_softirq();
-}
-
-int sparc_floppy_request_irq(int irq, unsigned long flags,
- irq_handler_t irq_handler)
-{
- floppy_irq_handler = irq_handler;
- return request_fast_irq(irq, floppy_hardint, flags, "floppy");
-}
-EXPORT_SYMBOL(sparc_floppy_request_irq);
-
-#endif
-
-int request_irq(unsigned int irq,
- irq_handler_t handler,
- unsigned long irqflags, const char * devname, void *dev_id)
-{
- struct irqaction * action, **actionp;
- unsigned long flags;
- unsigned int cpu_irq;
- int ret;
-
- if (sparc_cpu_model == sun4d) {
- extern int sun4d_request_irq(unsigned int,
- irq_handler_t ,
- unsigned long, const char *, void *);
- return sun4d_request_irq(irq, handler, irqflags, devname, dev_id);
- }
- cpu_irq = irq & (NR_IRQS - 1);
- if(cpu_irq > 14) {
- ret = -EINVAL;
- goto out;
- }
- if (!handler) {
- ret = -EINVAL;
- goto out;
- }
-
- spin_lock_irqsave(&irq_action_lock, flags);
-
- actionp = &sparc_irq[cpu_irq].action;
- action = *actionp;
- if (action) {
- if (!(action->flags & IRQF_SHARED) || !(irqflags & IRQF_SHARED)) {
- ret = -EBUSY;
- goto out_unlock;
- }
- if ((action->flags & IRQF_DISABLED) != (irqflags & IRQF_DISABLED)) {
- printk("Attempt to mix fast and slow interrupts on IRQ%d denied\n", irq);
- ret = -EBUSY;
- goto out_unlock;
- }
- for ( ; action; action = *actionp)
- actionp = &action->next;
- }
-
- /* If this is flagged as statically allocated then we use our
- * private struct which is never freed.
- */
- if (irqflags & SA_STATIC_ALLOC) {
- if (static_irq_count < MAX_STATIC_ALLOC)
- action = &static_irqaction[static_irq_count++];
- else
- printk("Request for IRQ%d (%s) SA_STATIC_ALLOC failed using kmalloc\n", irq, devname);
- }
-
- if (action == NULL)
- action = kmalloc(sizeof(struct irqaction),
- GFP_ATOMIC);
-
- if (!action) {
- ret = -ENOMEM;
- goto out_unlock;
- }
-
- action->handler = handler;
- action->flags = irqflags;
- cpus_clear(action->mask);
- action->name = devname;
- action->next = NULL;
- action->dev_id = dev_id;
-
- *actionp = action;
-
- __enable_irq(irq);
-
- ret = 0;
-out_unlock:
- spin_unlock_irqrestore(&irq_action_lock, flags);
-out:
- return ret;
-}
-
-EXPORT_SYMBOL(request_irq);
-
-void disable_irq_nosync(unsigned int irq)
-{
- return __disable_irq(irq);
-}
-EXPORT_SYMBOL(disable_irq_nosync);
-
-void disable_irq(unsigned int irq)
-{
- return __disable_irq(irq);
-}
-EXPORT_SYMBOL(disable_irq);
-
-void enable_irq(unsigned int irq)
-{
- return __enable_irq(irq);
-}
-
-EXPORT_SYMBOL(enable_irq);
-
-/* We really don't need these at all on the Sparc. We only have
- * stubs here because they are exported to modules.
- */
-unsigned long probe_irq_on(void)
-{
- return 0;
-}
-
-EXPORT_SYMBOL(probe_irq_on);
-
-int probe_irq_off(unsigned long mask)
-{
- return 0;
-}
-
-EXPORT_SYMBOL(probe_irq_off);
-
-/* djhr
- * This could probably be made indirect too and assigned in the CPU
- * bits of the code. That would be much nicer I think and would also
- * fit in with the idea of being able to tune your kernel for your machine
- * by removing unrequired machine and device support.
- *
- */
-
-void __init init_IRQ(void)
-{
- extern void sun4c_init_IRQ( void );
- extern void sun4m_init_IRQ( void );
- extern void sun4d_init_IRQ( void );
-
- switch(sparc_cpu_model) {
- case sun4c:
- case sun4:
- sun4c_init_IRQ();
- break;
-
- case sun4m:
-#ifdef CONFIG_PCI
- pcic_probe();
- if (pcic_present()) {
- sun4m_pci_init_IRQ();
- break;
- }
-#endif
- sun4m_init_IRQ();
- break;
-
- case sun4d:
- sun4d_init_IRQ();
- break;
-
- default:
- prom_printf("Cannot initialize IRQs on this Sun machine...");
- break;
- }
- btfixup();
-}
-
-void init_irq_proc(void)
-{
- /* For now, nothing... */
-}
diff --git a/arch/sparc/kernel/irq.h b/arch/sparc/kernel/irq.h
index 32ef3ebd0a8..70a0b8ddd0b 100644
--- a/arch/sparc/kernel/irq.h
+++ b/arch/sparc/kernel/irq.h
@@ -1,68 +1,101 @@
-#include <asm/btfixup.h>
-
-/* Dave Redman (djhr@tadpole.co.uk)
- * changed these to function pointers.. it saves cycles and will allow
- * the irq dependencies to be split into different files at a later date
- * sun4c_irq.c, sun4m_irq.c etc so we could reduce the kernel size.
- * Jakub Jelinek (jj@sunsite.mff.cuni.cz)
- * Changed these to btfixup entities... It saves cycles :)
+#include <linux/platform_device.h>
+
+#include <asm/cpu_type.h>
+
+struct irq_bucket {
+ struct irq_bucket *next;
+ unsigned int real_irq;
+ unsigned int irq;
+ unsigned int pil;
+};
+
+#define SUN4M_HARD_INT(x) (0x000000001 << (x))
+#define SUN4M_SOFT_INT(x) (0x000010000 << (x))
+
+#define SUN4D_MAX_BOARD 10
+#define SUN4D_MAX_IRQ ((SUN4D_MAX_BOARD + 2) << 5)
+
+/* Map between the irq identifier used in hw to the
+ * irq_bucket. The map is sufficient large to hold
+ * the sun4d hw identifiers.
+ */
+extern struct irq_bucket *irq_map[SUN4D_MAX_IRQ];
+
+
+/* sun4m specific type definitions */
+
+/* This maps direct to CPU specific interrupt registers */
+struct sun4m_irq_percpu {
+ u32 pending;
+ u32 clear;
+ u32 set;
+};
+
+/* This maps direct to global interrupt registers */
+struct sun4m_irq_global {
+ u32 pending;
+ u32 mask;
+ u32 mask_clear;
+ u32 mask_set;
+ u32 interrupt_target;
+};
+
+extern struct sun4m_irq_percpu __iomem *sun4m_irq_percpu[SUN4M_NCPUS];
+extern struct sun4m_irq_global __iomem *sun4m_irq_global;
+
+/* The following definitions describe the individual platform features: */
+#define FEAT_L10_CLOCKSOURCE (1 << 0) /* L10 timer is used as a clocksource */
+#define FEAT_L10_CLOCKEVENT (1 << 1) /* L10 timer is used as a clockevent */
+#define FEAT_L14_ONESHOT (1 << 2) /* L14 timer clockevent can oneshot */
+
+/*
+ * Platform specific configuration
+ * The individual platforms assign their platform
+ * specifics in their init functions.
*/
+struct sparc_config {
+ void (*init_timers)(void);
+ unsigned int (*build_device_irq)(struct platform_device *op,
+ unsigned int real_irq);
+
+ /* generic clockevent features - see FEAT_* above */
+ int features;
+
+ /* clock rate used for clock event timer */
+ int clock_rate;
+
+ /* one period for clock source timer */
+ unsigned int cs_period;
+
+ /* function to obtain offsett for cs period */
+ unsigned int (*get_cycles_offset)(void);
+
+ void (*clear_clock_irq)(void);
+ void (*load_profile_irq)(int cpu, unsigned int limit);
+};
+extern struct sparc_config sparc_config;
-BTFIXUPDEF_CALL(void, disable_irq, unsigned int)
-BTFIXUPDEF_CALL(void, enable_irq, unsigned int)
-BTFIXUPDEF_CALL(void, disable_pil_irq, unsigned int)
-BTFIXUPDEF_CALL(void, enable_pil_irq, unsigned int)
-BTFIXUPDEF_CALL(void, clear_clock_irq, void)
-BTFIXUPDEF_CALL(void, clear_profile_irq, int)
-BTFIXUPDEF_CALL(void, load_profile_irq, int, unsigned int)
-
-static inline void __disable_irq(unsigned int irq)
-{
- BTFIXUP_CALL(disable_irq)(irq);
-}
-
-static inline void __enable_irq(unsigned int irq)
-{
- BTFIXUP_CALL(enable_irq)(irq);
-}
-
-static inline void disable_pil_irq(unsigned int irq)
-{
- BTFIXUP_CALL(disable_pil_irq)(irq);
-}
-
-static inline void enable_pil_irq(unsigned int irq)
-{
- BTFIXUP_CALL(enable_pil_irq)(irq);
-}
-
-static inline void clear_clock_irq(void)
-{
- BTFIXUP_CALL(clear_clock_irq)();
-}
-
-static inline void clear_profile_irq(int irq)
-{
- BTFIXUP_CALL(clear_profile_irq)(irq);
-}
-
-static inline void load_profile_irq(int cpu, int limit)
-{
- BTFIXUP_CALL(load_profile_irq)(cpu, limit);
-}
-
-extern void (*sparc_init_timers)(irq_handler_t lvl10_irq);
-
-extern void claim_ticker14(irq_handler_t irq_handler,
- int irq,
- unsigned int timeout);
+unsigned int irq_alloc(unsigned int real_irq, unsigned int pil);
+void irq_link(unsigned int irq);
+void irq_unlink(unsigned int irq);
+void handler_irq(unsigned int pil, struct pt_regs *regs);
+
+unsigned long leon_get_irqmask(unsigned int irq);
+
+/* irq_32.c */
+void sparc_floppy_irq(int irq, void *dev_id, struct pt_regs *regs);
+
+/* sun4m_irq.c */
+void sun4m_nmi(struct pt_regs *regs);
+
+/* sun4d_irq.c */
+void sun4d_handler_irq(unsigned int pil, struct pt_regs *regs);
#ifdef CONFIG_SMP
-BTFIXUPDEF_CALL(void, set_cpu_int, int, int)
-BTFIXUPDEF_CALL(void, clear_cpu_int, int, int)
-BTFIXUPDEF_CALL(void, set_irq_udt, int)
-#define set_cpu_int(cpu,level) BTFIXUP_CALL(set_cpu_int)(cpu,level)
-#define clear_cpu_int(cpu,level) BTFIXUP_CALL(clear_cpu_int)(cpu,level)
-#define set_irq_udt(cpu) BTFIXUP_CALL(set_irq_udt)(cpu)
+/* All SUN4D IPIs are sent on this IRQ, may be shared with hard IRQs */
+#define SUN4D_IPI_IRQ 13
+
+void sun4d_ipi_interrupt(void);
+
#endif
diff --git a/arch/sparc/kernel/irq_32.c b/arch/sparc/kernel/irq_32.c
new file mode 100644
index 00000000000..a979e99f875
--- /dev/null
+++ b/arch/sparc/kernel/irq_32.c
@@ -0,0 +1,362 @@
+/*
+ * Interrupt request handling routines. On the
+ * Sparc the IRQs are basically 'cast in stone'
+ * and you are supposed to probe the prom's device
+ * node trees to find out who's got which IRQ.
+ *
+ * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1995 Miguel de Icaza (miguel@nuclecu.unam.mx)
+ * Copyright (C) 1995,2002 Pete A. Zaitcev (zaitcev@yahoo.com)
+ * Copyright (C) 1996 Dave Redman (djhr@tadpole.co.uk)
+ * Copyright (C) 1998-2000 Anton Blanchard (anton@samba.org)
+ */
+
+#include <linux/kernel_stat.h>
+#include <linux/seq_file.h>
+#include <linux/export.h>
+
+#include <asm/cacheflush.h>
+#include <asm/cpudata.h>
+#include <asm/setup.h>
+#include <asm/pcic.h>
+#include <asm/leon.h>
+
+#include "kernel.h"
+#include "irq.h"
+
+/* platform specific irq setup */
+struct sparc_config sparc_config;
+
+unsigned long arch_local_irq_save(void)
+{
+ unsigned long retval;
+ unsigned long tmp;
+
+ __asm__ __volatile__(
+ "rd %%psr, %0\n\t"
+ "or %0, %2, %1\n\t"
+ "wr %1, 0, %%psr\n\t"
+ "nop; nop; nop\n"
+ : "=&r" (retval), "=r" (tmp)
+ : "i" (PSR_PIL)
+ : "memory");
+
+ return retval;
+}
+EXPORT_SYMBOL(arch_local_irq_save);
+
+void arch_local_irq_enable(void)
+{
+ unsigned long tmp;
+
+ __asm__ __volatile__(
+ "rd %%psr, %0\n\t"
+ "andn %0, %1, %0\n\t"
+ "wr %0, 0, %%psr\n\t"
+ "nop; nop; nop\n"
+ : "=&r" (tmp)
+ : "i" (PSR_PIL)
+ : "memory");
+}
+EXPORT_SYMBOL(arch_local_irq_enable);
+
+void arch_local_irq_restore(unsigned long old_psr)
+{
+ unsigned long tmp;
+
+ __asm__ __volatile__(
+ "rd %%psr, %0\n\t"
+ "and %2, %1, %2\n\t"
+ "andn %0, %1, %0\n\t"
+ "wr %0, %2, %%psr\n\t"
+ "nop; nop; nop\n"
+ : "=&r" (tmp)
+ : "i" (PSR_PIL), "r" (old_psr)
+ : "memory");
+}
+EXPORT_SYMBOL(arch_local_irq_restore);
+
+/*
+ * Dave Redman (djhr@tadpole.co.uk)
+ *
+ * IRQ numbers.. These are no longer restricted to 15..
+ *
+ * this is done to enable SBUS cards and onboard IO to be masked
+ * correctly. using the interrupt level isn't good enough.
+ *
+ * For example:
+ * A device interrupting at sbus level6 and the Floppy both come in
+ * at IRQ11, but enabling and disabling them requires writing to
+ * different bits in the SLAVIO/SEC.
+ *
+ * As a result of these changes sun4m machines could now support
+ * directed CPU interrupts using the existing enable/disable irq code
+ * with tweaks.
+ *
+ * Sun4d complicates things even further. IRQ numbers are arbitrary
+ * 32-bit values in that case. Since this is similar to sparc64,
+ * we adopt a virtual IRQ numbering scheme as is done there.
+ * Virutal interrupt numbers are allocated by build_irq(). So NR_IRQS
+ * just becomes a limit of how many interrupt sources we can handle in
+ * a single system. Even fully loaded SS2000 machines top off at
+ * about 32 interrupt sources or so, therefore a NR_IRQS value of 64
+ * is more than enough.
+ *
+ * We keep a map of per-PIL enable interrupts. These get wired
+ * up via the irq_chip->startup() method which gets invoked by
+ * the generic IRQ layer during request_irq().
+ */
+
+
+/* Table of allocated irqs. Unused entries has irq == 0 */
+static struct irq_bucket irq_table[NR_IRQS];
+/* Protect access to irq_table */
+static DEFINE_SPINLOCK(irq_table_lock);
+
+/* Map between the irq identifier used in hw to the irq_bucket. */
+struct irq_bucket *irq_map[SUN4D_MAX_IRQ];
+/* Protect access to irq_map */
+static DEFINE_SPINLOCK(irq_map_lock);
+
+/* Allocate a new irq from the irq_table */
+unsigned int irq_alloc(unsigned int real_irq, unsigned int pil)
+{
+ unsigned long flags;
+ unsigned int i;
+
+ spin_lock_irqsave(&irq_table_lock, flags);
+ for (i = 1; i < NR_IRQS; i++) {
+ if (irq_table[i].real_irq == real_irq && irq_table[i].pil == pil)
+ goto found;
+ }
+
+ for (i = 1; i < NR_IRQS; i++) {
+ if (!irq_table[i].irq)
+ break;
+ }
+
+ if (i < NR_IRQS) {
+ irq_table[i].real_irq = real_irq;
+ irq_table[i].irq = i;
+ irq_table[i].pil = pil;
+ } else {
+ printk(KERN_ERR "IRQ: Out of virtual IRQs.\n");
+ i = 0;
+ }
+found:
+ spin_unlock_irqrestore(&irq_table_lock, flags);
+
+ return i;
+}
+
+/* Based on a single pil handler_irq may need to call several
+ * interrupt handlers. Use irq_map as entry to irq_table,
+ * and let each entry in irq_table point to the next entry.
+ */
+void irq_link(unsigned int irq)
+{
+ struct irq_bucket *p;
+ unsigned long flags;
+ unsigned int pil;
+
+ BUG_ON(irq >= NR_IRQS);
+
+ spin_lock_irqsave(&irq_map_lock, flags);
+
+ p = &irq_table[irq];
+ pil = p->pil;
+ BUG_ON(pil > SUN4D_MAX_IRQ);
+ p->next = irq_map[pil];
+ irq_map[pil] = p;
+
+ spin_unlock_irqrestore(&irq_map_lock, flags);
+}
+
+void irq_unlink(unsigned int irq)
+{
+ struct irq_bucket *p, **pnext;
+ unsigned long flags;
+
+ BUG_ON(irq >= NR_IRQS);
+
+ spin_lock_irqsave(&irq_map_lock, flags);
+
+ p = &irq_table[irq];
+ BUG_ON(p->pil > SUN4D_MAX_IRQ);
+ pnext = &irq_map[p->pil];
+ while (*pnext != p)
+ pnext = &(*pnext)->next;
+ *pnext = p->next;
+
+ spin_unlock_irqrestore(&irq_map_lock, flags);
+}
+
+
+/* /proc/interrupts printing */
+int arch_show_interrupts(struct seq_file *p, int prec)
+{
+ int j;
+
+#ifdef CONFIG_SMP
+ seq_printf(p, "RES: ");
+ for_each_online_cpu(j)
+ seq_printf(p, "%10u ", cpu_data(j).irq_resched_count);
+ seq_printf(p, " IPI rescheduling interrupts\n");
+ seq_printf(p, "CAL: ");
+ for_each_online_cpu(j)
+ seq_printf(p, "%10u ", cpu_data(j).irq_call_count);
+ seq_printf(p, " IPI function call interrupts\n");
+#endif
+ seq_printf(p, "NMI: ");
+ for_each_online_cpu(j)
+ seq_printf(p, "%10u ", cpu_data(j).counter);
+ seq_printf(p, " Non-maskable interrupts\n");
+ return 0;
+}
+
+void handler_irq(unsigned int pil, struct pt_regs *regs)
+{
+ struct pt_regs *old_regs;
+ struct irq_bucket *p;
+
+ BUG_ON(pil > 15);
+ old_regs = set_irq_regs(regs);
+ irq_enter();
+
+ p = irq_map[pil];
+ while (p) {
+ struct irq_bucket *next = p->next;
+
+ generic_handle_irq(p->irq);
+ p = next;
+ }
+ irq_exit();
+ set_irq_regs(old_regs);
+}
+
+#if defined(CONFIG_BLK_DEV_FD) || defined(CONFIG_BLK_DEV_FD_MODULE)
+static unsigned int floppy_irq;
+
+int sparc_floppy_request_irq(unsigned int irq, irq_handler_t irq_handler)
+{
+ unsigned int cpu_irq;
+ int err;
+
+
+ err = request_irq(irq, irq_handler, 0, "floppy", NULL);
+ if (err)
+ return -1;
+
+ /* Save for later use in floppy interrupt handler */
+ floppy_irq = irq;
+
+ cpu_irq = (irq & (NR_IRQS - 1));
+
+ /* Dork with trap table if we get this far. */
+#define INSTANTIATE(table) \
+ table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_one = SPARC_RD_PSR_L0; \
+ table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_two = \
+ SPARC_BRANCH((unsigned long) floppy_hardint, \
+ (unsigned long) &table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_two);\
+ table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_three = SPARC_RD_WIM_L3; \
+ table[SP_TRAP_IRQ1+(cpu_irq-1)].inst_four = SPARC_NOP;
+
+ INSTANTIATE(sparc_ttable)
+
+#if defined CONFIG_SMP
+ if (sparc_cpu_model != sparc_leon) {
+ struct tt_entry *trap_table;
+
+ trap_table = &trapbase_cpu1;
+ INSTANTIATE(trap_table)
+ trap_table = &trapbase_cpu2;
+ INSTANTIATE(trap_table)
+ trap_table = &trapbase_cpu3;
+ INSTANTIATE(trap_table)
+ }
+#endif
+#undef INSTANTIATE
+ /*
+ * XXX Correct thing whould be to flush only I- and D-cache lines
+ * which contain the handler in question. But as of time of the
+ * writing we have no CPU-neutral interface to fine-grained flushes.
+ */
+ flush_cache_all();
+ return 0;
+}
+EXPORT_SYMBOL(sparc_floppy_request_irq);
+
+/*
+ * These variables are used to access state from the assembler
+ * interrupt handler, floppy_hardint, so we cannot put these in
+ * the floppy driver image because that would not work in the
+ * modular case.
+ */
+volatile unsigned char *fdc_status;
+EXPORT_SYMBOL(fdc_status);
+
+char *pdma_vaddr;
+EXPORT_SYMBOL(pdma_vaddr);
+
+unsigned long pdma_size;
+EXPORT_SYMBOL(pdma_size);
+
+volatile int doing_pdma;
+EXPORT_SYMBOL(doing_pdma);
+
+char *pdma_base;
+EXPORT_SYMBOL(pdma_base);
+
+unsigned long pdma_areasize;
+EXPORT_SYMBOL(pdma_areasize);
+
+/* Use the generic irq support to call floppy_interrupt
+ * which was setup using request_irq() in sparc_floppy_request_irq().
+ * We only have one floppy interrupt so we do not need to check
+ * for additional handlers being wired up by irq_link()
+ */
+void sparc_floppy_irq(int irq, void *dev_id, struct pt_regs *regs)
+{
+ struct pt_regs *old_regs;
+
+ old_regs = set_irq_regs(regs);
+ irq_enter();
+ generic_handle_irq(floppy_irq);
+ irq_exit();
+ set_irq_regs(old_regs);
+}
+#endif
+
+/* djhr
+ * This could probably be made indirect too and assigned in the CPU
+ * bits of the code. That would be much nicer I think and would also
+ * fit in with the idea of being able to tune your kernel for your machine
+ * by removing unrequired machine and device support.
+ *
+ */
+
+void __init init_IRQ(void)
+{
+ switch (sparc_cpu_model) {
+ case sun4m:
+ pcic_probe();
+ if (pcic_present())
+ sun4m_pci_init_IRQ();
+ else
+ sun4m_init_IRQ();
+ break;
+
+ case sun4d:
+ sun4d_init_IRQ();
+ break;
+
+ case sparc_leon:
+ leon_init_IRQ();
+ break;
+
+ default:
+ prom_printf("Cannot initialize IRQs on this Sun machine...");
+ break;
+ }
+}
+
diff --git a/arch/sparc/kernel/irq_64.c b/arch/sparc/kernel/irq_64.c
new file mode 100644
index 00000000000..666193f4e8b
--- /dev/null
+++ b/arch/sparc/kernel/irq_64.c
@@ -0,0 +1,973 @@
+/* irq.c: UltraSparc IRQ handling/init/registry.
+ *
+ * Copyright (C) 1997, 2007, 2008 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1998 Eddie C. Dost (ecd@skynet.be)
+ * Copyright (C) 1998 Jakub Jelinek (jj@ultra.linux.cz)
+ */
+
+#include <linux/sched.h>
+#include <linux/linkage.h>
+#include <linux/ptrace.h>
+#include <linux/errno.h>
+#include <linux/kernel_stat.h>
+#include <linux/signal.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <linux/slab.h>
+#include <linux/random.h>
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
+#include <linux/ftrace.h>
+#include <linux/irq.h>
+#include <linux/kmemleak.h>
+
+#include <asm/ptrace.h>
+#include <asm/processor.h>
+#include <linux/atomic.h>
+#include <asm/irq.h>
+#include <asm/io.h>
+#include <asm/iommu.h>
+#include <asm/upa.h>
+#include <asm/oplib.h>
+#include <asm/prom.h>
+#include <asm/timer.h>
+#include <asm/smp.h>
+#include <asm/starfire.h>
+#include <asm/uaccess.h>
+#include <asm/cache.h>
+#include <asm/cpudata.h>
+#include <asm/auxio.h>
+#include <asm/head.h>
+#include <asm/hypervisor.h>
+#include <asm/cacheflush.h>
+
+#include "entry.h"
+#include "cpumap.h"
+#include "kstack.h"
+
+#define NUM_IVECS (IMAP_INR + 1)
+
+struct ino_bucket *ivector_table;
+unsigned long ivector_table_pa;
+
+/* On several sun4u processors, it is illegal to mix bypass and
+ * non-bypass accesses. Therefore we access all INO buckets
+ * using bypass accesses only.
+ */
+static unsigned long bucket_get_chain_pa(unsigned long bucket_pa)
+{
+ unsigned long ret;
+
+ __asm__ __volatile__("ldxa [%1] %2, %0"
+ : "=&r" (ret)
+ : "r" (bucket_pa +
+ offsetof(struct ino_bucket,
+ __irq_chain_pa)),
+ "i" (ASI_PHYS_USE_EC));
+
+ return ret;
+}
+
+static void bucket_clear_chain_pa(unsigned long bucket_pa)
+{
+ __asm__ __volatile__("stxa %%g0, [%0] %1"
+ : /* no outputs */
+ : "r" (bucket_pa +
+ offsetof(struct ino_bucket,
+ __irq_chain_pa)),
+ "i" (ASI_PHYS_USE_EC));
+}
+
+static unsigned int bucket_get_irq(unsigned long bucket_pa)
+{
+ unsigned int ret;
+
+ __asm__ __volatile__("lduwa [%1] %2, %0"
+ : "=&r" (ret)
+ : "r" (bucket_pa +
+ offsetof(struct ino_bucket,
+ __irq)),
+ "i" (ASI_PHYS_USE_EC));
+
+ return ret;
+}
+
+static void bucket_set_irq(unsigned long bucket_pa, unsigned int irq)
+{
+ __asm__ __volatile__("stwa %0, [%1] %2"
+ : /* no outputs */
+ : "r" (irq),
+ "r" (bucket_pa +
+ offsetof(struct ino_bucket,
+ __irq)),
+ "i" (ASI_PHYS_USE_EC));
+}
+
+#define irq_work_pa(__cpu) &(trap_block[(__cpu)].irq_worklist_pa)
+
+static struct {
+ unsigned int dev_handle;
+ unsigned int dev_ino;
+ unsigned int in_use;
+} irq_table[NR_IRQS];
+static DEFINE_SPINLOCK(irq_alloc_lock);
+
+unsigned char irq_alloc(unsigned int dev_handle, unsigned int dev_ino)
+{
+ unsigned long flags;
+ unsigned char ent;
+
+ BUILD_BUG_ON(NR_IRQS >= 256);
+
+ spin_lock_irqsave(&irq_alloc_lock, flags);
+
+ for (ent = 1; ent < NR_IRQS; ent++) {
+ if (!irq_table[ent].in_use)
+ break;
+ }
+ if (ent >= NR_IRQS) {
+ printk(KERN_ERR "IRQ: Out of virtual IRQs.\n");
+ ent = 0;
+ } else {
+ irq_table[ent].dev_handle = dev_handle;
+ irq_table[ent].dev_ino = dev_ino;
+ irq_table[ent].in_use = 1;
+ }
+
+ spin_unlock_irqrestore(&irq_alloc_lock, flags);
+
+ return ent;
+}
+
+#ifdef CONFIG_PCI_MSI
+void irq_free(unsigned int irq)
+{
+ unsigned long flags;
+
+ if (irq >= NR_IRQS)
+ return;
+
+ spin_lock_irqsave(&irq_alloc_lock, flags);
+
+ irq_table[irq].in_use = 0;
+
+ spin_unlock_irqrestore(&irq_alloc_lock, flags);
+}
+#endif
+
+/*
+ * /proc/interrupts printing:
+ */
+int arch_show_interrupts(struct seq_file *p, int prec)
+{
+ int j;
+
+ seq_printf(p, "NMI: ");
+ for_each_online_cpu(j)
+ seq_printf(p, "%10u ", cpu_data(j).__nmi_count);
+ seq_printf(p, " Non-maskable interrupts\n");
+ return 0;
+}
+
+static unsigned int sun4u_compute_tid(unsigned long imap, unsigned long cpuid)
+{
+ unsigned int tid;
+
+ if (this_is_starfire) {
+ tid = starfire_translate(imap, cpuid);
+ tid <<= IMAP_TID_SHIFT;
+ tid &= IMAP_TID_UPA;
+ } else {
+ if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+ unsigned long ver;
+
+ __asm__ ("rdpr %%ver, %0" : "=r" (ver));
+ if ((ver >> 32UL) == __JALAPENO_ID ||
+ (ver >> 32UL) == __SERRANO_ID) {
+ tid = cpuid << IMAP_TID_SHIFT;
+ tid &= IMAP_TID_JBUS;
+ } else {
+ unsigned int a = cpuid & 0x1f;
+ unsigned int n = (cpuid >> 5) & 0x1f;
+
+ tid = ((a << IMAP_AID_SHIFT) |
+ (n << IMAP_NID_SHIFT));
+ tid &= (IMAP_AID_SAFARI |
+ IMAP_NID_SAFARI);
+ }
+ } else {
+ tid = cpuid << IMAP_TID_SHIFT;
+ tid &= IMAP_TID_UPA;
+ }
+ }
+
+ return tid;
+}
+
+struct irq_handler_data {
+ unsigned long iclr;
+ unsigned long imap;
+
+ void (*pre_handler)(unsigned int, void *, void *);
+ void *arg1;
+ void *arg2;
+};
+
+#ifdef CONFIG_SMP
+static int irq_choose_cpu(unsigned int irq, const struct cpumask *affinity)
+{
+ cpumask_t mask;
+ int cpuid;
+
+ cpumask_copy(&mask, affinity);
+ if (cpumask_equal(&mask, cpu_online_mask)) {
+ cpuid = map_to_cpu(irq);
+ } else {
+ cpumask_t tmp;
+
+ cpumask_and(&tmp, cpu_online_mask, &mask);
+ cpuid = cpumask_empty(&tmp) ? map_to_cpu(irq) : cpumask_first(&tmp);
+ }
+
+ return cpuid;
+}
+#else
+#define irq_choose_cpu(irq, affinity) \
+ real_hard_smp_processor_id()
+#endif
+
+static void sun4u_irq_enable(struct irq_data *data)
+{
+ struct irq_handler_data *handler_data = data->handler_data;
+
+ if (likely(handler_data)) {
+ unsigned long cpuid, imap, val;
+ unsigned int tid;
+
+ cpuid = irq_choose_cpu(data->irq, data->affinity);
+ imap = handler_data->imap;
+
+ tid = sun4u_compute_tid(imap, cpuid);
+
+ val = upa_readq(imap);
+ val &= ~(IMAP_TID_UPA | IMAP_TID_JBUS |
+ IMAP_AID_SAFARI | IMAP_NID_SAFARI);
+ val |= tid | IMAP_VALID;
+ upa_writeq(val, imap);
+ upa_writeq(ICLR_IDLE, handler_data->iclr);
+ }
+}
+
+static int sun4u_set_affinity(struct irq_data *data,
+ const struct cpumask *mask, bool force)
+{
+ struct irq_handler_data *handler_data = data->handler_data;
+
+ if (likely(handler_data)) {
+ unsigned long cpuid, imap, val;
+ unsigned int tid;
+
+ cpuid = irq_choose_cpu(data->irq, mask);
+ imap = handler_data->imap;
+
+ tid = sun4u_compute_tid(imap, cpuid);
+
+ val = upa_readq(imap);
+ val &= ~(IMAP_TID_UPA | IMAP_TID_JBUS |
+ IMAP_AID_SAFARI | IMAP_NID_SAFARI);
+ val |= tid | IMAP_VALID;
+ upa_writeq(val, imap);
+ upa_writeq(ICLR_IDLE, handler_data->iclr);
+ }
+
+ return 0;
+}
+
+/* Don't do anything. The desc->status check for IRQ_DISABLED in
+ * handler_irq() will skip the handler call and that will leave the
+ * interrupt in the sent state. The next ->enable() call will hit the
+ * ICLR register to reset the state machine.
+ *
+ * This scheme is necessary, instead of clearing the Valid bit in the
+ * IMAP register, to handle the case of IMAP registers being shared by
+ * multiple INOs (and thus ICLR registers). Since we use a different
+ * virtual IRQ for each shared IMAP instance, the generic code thinks
+ * there is only one user so it prematurely calls ->disable() on
+ * free_irq().
+ *
+ * We have to provide an explicit ->disable() method instead of using
+ * NULL to get the default. The reason is that if the generic code
+ * sees that, it also hooks up a default ->shutdown method which
+ * invokes ->mask() which we do not want. See irq_chip_set_defaults().
+ */
+static void sun4u_irq_disable(struct irq_data *data)
+{
+}
+
+static void sun4u_irq_eoi(struct irq_data *data)
+{
+ struct irq_handler_data *handler_data = data->handler_data;
+
+ if (likely(handler_data))
+ upa_writeq(ICLR_IDLE, handler_data->iclr);
+}
+
+static void sun4v_irq_enable(struct irq_data *data)
+{
+ unsigned int ino = irq_table[data->irq].dev_ino;
+ unsigned long cpuid = irq_choose_cpu(data->irq, data->affinity);
+ int err;
+
+ err = sun4v_intr_settarget(ino, cpuid);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_intr_settarget(%x,%lu): "
+ "err(%d)\n", ino, cpuid, err);
+ err = sun4v_intr_setstate(ino, HV_INTR_STATE_IDLE);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_intr_setstate(%x): "
+ "err(%d)\n", ino, err);
+ err = sun4v_intr_setenabled(ino, HV_INTR_ENABLED);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_intr_setenabled(%x): err(%d)\n",
+ ino, err);
+}
+
+static int sun4v_set_affinity(struct irq_data *data,
+ const struct cpumask *mask, bool force)
+{
+ unsigned int ino = irq_table[data->irq].dev_ino;
+ unsigned long cpuid = irq_choose_cpu(data->irq, mask);
+ int err;
+
+ err = sun4v_intr_settarget(ino, cpuid);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_intr_settarget(%x,%lu): "
+ "err(%d)\n", ino, cpuid, err);
+
+ return 0;
+}
+
+static void sun4v_irq_disable(struct irq_data *data)
+{
+ unsigned int ino = irq_table[data->irq].dev_ino;
+ int err;
+
+ err = sun4v_intr_setenabled(ino, HV_INTR_DISABLED);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_intr_setenabled(%x): "
+ "err(%d)\n", ino, err);
+}
+
+static void sun4v_irq_eoi(struct irq_data *data)
+{
+ unsigned int ino = irq_table[data->irq].dev_ino;
+ int err;
+
+ err = sun4v_intr_setstate(ino, HV_INTR_STATE_IDLE);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_intr_setstate(%x): "
+ "err(%d)\n", ino, err);
+}
+
+static void sun4v_virq_enable(struct irq_data *data)
+{
+ unsigned long cpuid, dev_handle, dev_ino;
+ int err;
+
+ cpuid = irq_choose_cpu(data->irq, data->affinity);
+
+ dev_handle = irq_table[data->irq].dev_handle;
+ dev_ino = irq_table[data->irq].dev_ino;
+
+ err = sun4v_vintr_set_target(dev_handle, dev_ino, cpuid);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_vintr_set_target(%lx,%lx,%lu): "
+ "err(%d)\n",
+ dev_handle, dev_ino, cpuid, err);
+ err = sun4v_vintr_set_state(dev_handle, dev_ino,
+ HV_INTR_STATE_IDLE);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_vintr_set_state(%lx,%lx,"
+ "HV_INTR_STATE_IDLE): err(%d)\n",
+ dev_handle, dev_ino, err);
+ err = sun4v_vintr_set_valid(dev_handle, dev_ino,
+ HV_INTR_ENABLED);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_vintr_set_state(%lx,%lx,"
+ "HV_INTR_ENABLED): err(%d)\n",
+ dev_handle, dev_ino, err);
+}
+
+static int sun4v_virt_set_affinity(struct irq_data *data,
+ const struct cpumask *mask, bool force)
+{
+ unsigned long cpuid, dev_handle, dev_ino;
+ int err;
+
+ cpuid = irq_choose_cpu(data->irq, mask);
+
+ dev_handle = irq_table[data->irq].dev_handle;
+ dev_ino = irq_table[data->irq].dev_ino;
+
+ err = sun4v_vintr_set_target(dev_handle, dev_ino, cpuid);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_vintr_set_target(%lx,%lx,%lu): "
+ "err(%d)\n",
+ dev_handle, dev_ino, cpuid, err);
+
+ return 0;
+}
+
+static void sun4v_virq_disable(struct irq_data *data)
+{
+ unsigned long dev_handle, dev_ino;
+ int err;
+
+ dev_handle = irq_table[data->irq].dev_handle;
+ dev_ino = irq_table[data->irq].dev_ino;
+
+ err = sun4v_vintr_set_valid(dev_handle, dev_ino,
+ HV_INTR_DISABLED);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_vintr_set_state(%lx,%lx,"
+ "HV_INTR_DISABLED): err(%d)\n",
+ dev_handle, dev_ino, err);
+}
+
+static void sun4v_virq_eoi(struct irq_data *data)
+{
+ unsigned long dev_handle, dev_ino;
+ int err;
+
+ dev_handle = irq_table[data->irq].dev_handle;
+ dev_ino = irq_table[data->irq].dev_ino;
+
+ err = sun4v_vintr_set_state(dev_handle, dev_ino,
+ HV_INTR_STATE_IDLE);
+ if (err != HV_EOK)
+ printk(KERN_ERR "sun4v_vintr_set_state(%lx,%lx,"
+ "HV_INTR_STATE_IDLE): err(%d)\n",
+ dev_handle, dev_ino, err);
+}
+
+static struct irq_chip sun4u_irq = {
+ .name = "sun4u",
+ .irq_enable = sun4u_irq_enable,
+ .irq_disable = sun4u_irq_disable,
+ .irq_eoi = sun4u_irq_eoi,
+ .irq_set_affinity = sun4u_set_affinity,
+ .flags = IRQCHIP_EOI_IF_HANDLED,
+};
+
+static struct irq_chip sun4v_irq = {
+ .name = "sun4v",
+ .irq_enable = sun4v_irq_enable,
+ .irq_disable = sun4v_irq_disable,
+ .irq_eoi = sun4v_irq_eoi,
+ .irq_set_affinity = sun4v_set_affinity,
+ .flags = IRQCHIP_EOI_IF_HANDLED,
+};
+
+static struct irq_chip sun4v_virq = {
+ .name = "vsun4v",
+ .irq_enable = sun4v_virq_enable,
+ .irq_disable = sun4v_virq_disable,
+ .irq_eoi = sun4v_virq_eoi,
+ .irq_set_affinity = sun4v_virt_set_affinity,
+ .flags = IRQCHIP_EOI_IF_HANDLED,
+};
+
+static void pre_flow_handler(struct irq_data *d)
+{
+ struct irq_handler_data *handler_data = irq_data_get_irq_handler_data(d);
+ unsigned int ino = irq_table[d->irq].dev_ino;
+
+ handler_data->pre_handler(ino, handler_data->arg1, handler_data->arg2);
+}
+
+void irq_install_pre_handler(int irq,
+ void (*func)(unsigned int, void *, void *),
+ void *arg1, void *arg2)
+{
+ struct irq_handler_data *handler_data = irq_get_handler_data(irq);
+
+ handler_data->pre_handler = func;
+ handler_data->arg1 = arg1;
+ handler_data->arg2 = arg2;
+
+ __irq_set_preflow_handler(irq, pre_flow_handler);
+}
+
+unsigned int build_irq(int inofixup, unsigned long iclr, unsigned long imap)
+{
+ struct ino_bucket *bucket;
+ struct irq_handler_data *handler_data;
+ unsigned int irq;
+ int ino;
+
+ BUG_ON(tlb_type == hypervisor);
+
+ ino = (upa_readq(imap) & (IMAP_IGN | IMAP_INO)) + inofixup;
+ bucket = &ivector_table[ino];
+ irq = bucket_get_irq(__pa(bucket));
+ if (!irq) {
+ irq = irq_alloc(0, ino);
+ bucket_set_irq(__pa(bucket), irq);
+ irq_set_chip_and_handler_name(irq, &sun4u_irq,
+ handle_fasteoi_irq, "IVEC");
+ }
+
+ handler_data = irq_get_handler_data(irq);
+ if (unlikely(handler_data))
+ goto out;
+
+ handler_data = kzalloc(sizeof(struct irq_handler_data), GFP_ATOMIC);
+ if (unlikely(!handler_data)) {
+ prom_printf("IRQ: kzalloc(irq_handler_data) failed.\n");
+ prom_halt();
+ }
+ irq_set_handler_data(irq, handler_data);
+
+ handler_data->imap = imap;
+ handler_data->iclr = iclr;
+
+out:
+ return irq;
+}
+
+static unsigned int sun4v_build_common(unsigned long sysino,
+ struct irq_chip *chip)
+{
+ struct ino_bucket *bucket;
+ struct irq_handler_data *handler_data;
+ unsigned int irq;
+
+ BUG_ON(tlb_type != hypervisor);
+
+ bucket = &ivector_table[sysino];
+ irq = bucket_get_irq(__pa(bucket));
+ if (!irq) {
+ irq = irq_alloc(0, sysino);
+ bucket_set_irq(__pa(bucket), irq);
+ irq_set_chip_and_handler_name(irq, chip, handle_fasteoi_irq,
+ "IVEC");
+ }
+
+ handler_data = irq_get_handler_data(irq);
+ if (unlikely(handler_data))
+ goto out;
+
+ handler_data = kzalloc(sizeof(struct irq_handler_data), GFP_ATOMIC);
+ if (unlikely(!handler_data)) {
+ prom_printf("IRQ: kzalloc(irq_handler_data) failed.\n");
+ prom_halt();
+ }
+ irq_set_handler_data(irq, handler_data);
+
+ /* Catch accidental accesses to these things. IMAP/ICLR handling
+ * is done by hypervisor calls on sun4v platforms, not by direct
+ * register accesses.
+ */
+ handler_data->imap = ~0UL;
+ handler_data->iclr = ~0UL;
+
+out:
+ return irq;
+}
+
+unsigned int sun4v_build_irq(u32 devhandle, unsigned int devino)
+{
+ unsigned long sysino = sun4v_devino_to_sysino(devhandle, devino);
+
+ return sun4v_build_common(sysino, &sun4v_irq);
+}
+
+unsigned int sun4v_build_virq(u32 devhandle, unsigned int devino)
+{
+ struct irq_handler_data *handler_data;
+ unsigned long hv_err, cookie;
+ struct ino_bucket *bucket;
+ unsigned int irq;
+
+ bucket = kzalloc(sizeof(struct ino_bucket), GFP_ATOMIC);
+ if (unlikely(!bucket))
+ return 0;
+
+ /* The only reference we store to the IRQ bucket is
+ * by physical address which kmemleak can't see, tell
+ * it that this object explicitly is not a leak and
+ * should be scanned.
+ */
+ kmemleak_not_leak(bucket);
+
+ __flush_dcache_range((unsigned long) bucket,
+ ((unsigned long) bucket +
+ sizeof(struct ino_bucket)));
+
+ irq = irq_alloc(devhandle, devino);
+ bucket_set_irq(__pa(bucket), irq);
+
+ irq_set_chip_and_handler_name(irq, &sun4v_virq, handle_fasteoi_irq,
+ "IVEC");
+
+ handler_data = kzalloc(sizeof(struct irq_handler_data), GFP_ATOMIC);
+ if (unlikely(!handler_data))
+ return 0;
+
+ /* In order to make the LDC channel startup sequence easier,
+ * especially wrt. locking, we do not let request_irq() enable
+ * the interrupt.
+ */
+ irq_set_status_flags(irq, IRQ_NOAUTOEN);
+ irq_set_handler_data(irq, handler_data);
+
+ /* Catch accidental accesses to these things. IMAP/ICLR handling
+ * is done by hypervisor calls on sun4v platforms, not by direct
+ * register accesses.
+ */
+ handler_data->imap = ~0UL;
+ handler_data->iclr = ~0UL;
+
+ cookie = ~__pa(bucket);
+ hv_err = sun4v_vintr_set_cookie(devhandle, devino, cookie);
+ if (hv_err) {
+ prom_printf("IRQ: Fatal, cannot set cookie for [%x:%x] "
+ "err=%lu\n", devhandle, devino, hv_err);
+ prom_halt();
+ }
+
+ return irq;
+}
+
+void ack_bad_irq(unsigned int irq)
+{
+ unsigned int ino = irq_table[irq].dev_ino;
+
+ if (!ino)
+ ino = 0xdeadbeef;
+
+ printk(KERN_CRIT "Unexpected IRQ from ino[%x] irq[%u]\n",
+ ino, irq);
+}
+
+void *hardirq_stack[NR_CPUS];
+void *softirq_stack[NR_CPUS];
+
+void __irq_entry handler_irq(int pil, struct pt_regs *regs)
+{
+ unsigned long pstate, bucket_pa;
+ struct pt_regs *old_regs;
+ void *orig_sp;
+
+ clear_softint(1 << pil);
+
+ old_regs = set_irq_regs(regs);
+ irq_enter();
+
+ /* Grab an atomic snapshot of the pending IVECs. */
+ __asm__ __volatile__("rdpr %%pstate, %0\n\t"
+ "wrpr %0, %3, %%pstate\n\t"
+ "ldx [%2], %1\n\t"
+ "stx %%g0, [%2]\n\t"
+ "wrpr %0, 0x0, %%pstate\n\t"
+ : "=&r" (pstate), "=&r" (bucket_pa)
+ : "r" (irq_work_pa(smp_processor_id())),
+ "i" (PSTATE_IE)
+ : "memory");
+
+ orig_sp = set_hardirq_stack();
+
+ while (bucket_pa) {
+ unsigned long next_pa;
+ unsigned int irq;
+
+ next_pa = bucket_get_chain_pa(bucket_pa);
+ irq = bucket_get_irq(bucket_pa);
+ bucket_clear_chain_pa(bucket_pa);
+
+ generic_handle_irq(irq);
+
+ bucket_pa = next_pa;
+ }
+
+ restore_hardirq_stack(orig_sp);
+
+ irq_exit();
+ set_irq_regs(old_regs);
+}
+
+void do_softirq_own_stack(void)
+{
+ void *orig_sp, *sp = softirq_stack[smp_processor_id()];
+
+ sp += THREAD_SIZE - 192 - STACK_BIAS;
+
+ __asm__ __volatile__("mov %%sp, %0\n\t"
+ "mov %1, %%sp"
+ : "=&r" (orig_sp)
+ : "r" (sp));
+ __do_softirq();
+ __asm__ __volatile__("mov %0, %%sp"
+ : : "r" (orig_sp));
+}
+
+#ifdef CONFIG_HOTPLUG_CPU
+void fixup_irqs(void)
+{
+ unsigned int irq;
+
+ for (irq = 0; irq < NR_IRQS; irq++) {
+ struct irq_desc *desc = irq_to_desc(irq);
+ struct irq_data *data = irq_desc_get_irq_data(desc);
+ unsigned long flags;
+
+ raw_spin_lock_irqsave(&desc->lock, flags);
+ if (desc->action && !irqd_is_per_cpu(data)) {
+ if (data->chip->irq_set_affinity)
+ data->chip->irq_set_affinity(data,
+ data->affinity,
+ false);
+ }
+ raw_spin_unlock_irqrestore(&desc->lock, flags);
+ }
+
+ tick_ops->disable_irq();
+}
+#endif
+
+struct sun5_timer {
+ u64 count0;
+ u64 limit0;
+ u64 count1;
+ u64 limit1;
+};
+
+static struct sun5_timer *prom_timers;
+static u64 prom_limit0, prom_limit1;
+
+static void map_prom_timers(void)
+{
+ struct device_node *dp;
+ const unsigned int *addr;
+
+ /* PROM timer node hangs out in the top level of device siblings... */
+ dp = of_find_node_by_path("/");
+ dp = dp->child;
+ while (dp) {
+ if (!strcmp(dp->name, "counter-timer"))
+ break;
+ dp = dp->sibling;
+ }
+
+ /* Assume if node is not present, PROM uses different tick mechanism
+ * which we should not care about.
+ */
+ if (!dp) {
+ prom_timers = (struct sun5_timer *) 0;
+ return;
+ }
+
+ /* If PROM is really using this, it must be mapped by him. */
+ addr = of_get_property(dp, "address", NULL);
+ if (!addr) {
+ prom_printf("PROM does not have timer mapped, trying to continue.\n");
+ prom_timers = (struct sun5_timer *) 0;
+ return;
+ }
+ prom_timers = (struct sun5_timer *) ((unsigned long)addr[0]);
+}
+
+static void kill_prom_timer(void)
+{
+ if (!prom_timers)
+ return;
+
+ /* Save them away for later. */
+ prom_limit0 = prom_timers->limit0;
+ prom_limit1 = prom_timers->limit1;
+
+ /* Just as in sun4c PROM uses timer which ticks at IRQ 14.
+ * We turn both off here just to be paranoid.
+ */
+ prom_timers->limit0 = 0;
+ prom_timers->limit1 = 0;
+
+ /* Wheee, eat the interrupt packet too... */
+ __asm__ __volatile__(
+" mov 0x40, %%g2\n"
+" ldxa [%%g0] %0, %%g1\n"
+" ldxa [%%g2] %1, %%g1\n"
+" stxa %%g0, [%%g0] %0\n"
+" membar #Sync\n"
+ : /* no outputs */
+ : "i" (ASI_INTR_RECEIVE), "i" (ASI_INTR_R)
+ : "g1", "g2");
+}
+
+void notrace init_irqwork_curcpu(void)
+{
+ int cpu = hard_smp_processor_id();
+
+ trap_block[cpu].irq_worklist_pa = 0UL;
+}
+
+/* Please be very careful with register_one_mondo() and
+ * sun4v_register_mondo_queues().
+ *
+ * On SMP this gets invoked from the CPU trampoline before
+ * the cpu has fully taken over the trap table from OBP,
+ * and it's kernel stack + %g6 thread register state is
+ * not fully cooked yet.
+ *
+ * Therefore you cannot make any OBP calls, not even prom_printf,
+ * from these two routines.
+ */
+static void notrace register_one_mondo(unsigned long paddr, unsigned long type,
+ unsigned long qmask)
+{
+ unsigned long num_entries = (qmask + 1) / 64;
+ unsigned long status;
+
+ status = sun4v_cpu_qconf(type, paddr, num_entries);
+ if (status != HV_EOK) {
+ prom_printf("SUN4V: sun4v_cpu_qconf(%lu:%lx:%lu) failed, "
+ "err %lu\n", type, paddr, num_entries, status);
+ prom_halt();
+ }
+}
+
+void notrace sun4v_register_mondo_queues(int this_cpu)
+{
+ struct trap_per_cpu *tb = &trap_block[this_cpu];
+
+ register_one_mondo(tb->cpu_mondo_pa, HV_CPU_QUEUE_CPU_MONDO,
+ tb->cpu_mondo_qmask);
+ register_one_mondo(tb->dev_mondo_pa, HV_CPU_QUEUE_DEVICE_MONDO,
+ tb->dev_mondo_qmask);
+ register_one_mondo(tb->resum_mondo_pa, HV_CPU_QUEUE_RES_ERROR,
+ tb->resum_qmask);
+ register_one_mondo(tb->nonresum_mondo_pa, HV_CPU_QUEUE_NONRES_ERROR,
+ tb->nonresum_qmask);
+}
+
+/* Each queue region must be a power of 2 multiple of 64 bytes in
+ * size. The base real address must be aligned to the size of the
+ * region. Thus, an 8KB queue must be 8KB aligned, for example.
+ */
+static void __init alloc_one_queue(unsigned long *pa_ptr, unsigned long qmask)
+{
+ unsigned long size = PAGE_ALIGN(qmask + 1);
+ unsigned long order = get_order(size);
+ unsigned long p;
+
+ p = __get_free_pages(GFP_KERNEL, order);
+ if (!p) {
+ prom_printf("SUN4V: Error, cannot allocate queue.\n");
+ prom_halt();
+ }
+
+ *pa_ptr = __pa(p);
+}
+
+static void __init init_cpu_send_mondo_info(struct trap_per_cpu *tb)
+{
+#ifdef CONFIG_SMP
+ unsigned long page;
+
+ BUILD_BUG_ON((NR_CPUS * sizeof(u16)) > (PAGE_SIZE - 64));
+
+ page = get_zeroed_page(GFP_KERNEL);
+ if (!page) {
+ prom_printf("SUN4V: Error, cannot allocate cpu mondo page.\n");
+ prom_halt();
+ }
+
+ tb->cpu_mondo_block_pa = __pa(page);
+ tb->cpu_list_pa = __pa(page + 64);
+#endif
+}
+
+/* Allocate mondo and error queues for all possible cpus. */
+static void __init sun4v_init_mondo_queues(void)
+{
+ int cpu;
+
+ for_each_possible_cpu(cpu) {
+ struct trap_per_cpu *tb = &trap_block[cpu];
+
+ alloc_one_queue(&tb->cpu_mondo_pa, tb->cpu_mondo_qmask);
+ alloc_one_queue(&tb->dev_mondo_pa, tb->dev_mondo_qmask);
+ alloc_one_queue(&tb->resum_mondo_pa, tb->resum_qmask);
+ alloc_one_queue(&tb->resum_kernel_buf_pa, tb->resum_qmask);
+ alloc_one_queue(&tb->nonresum_mondo_pa, tb->nonresum_qmask);
+ alloc_one_queue(&tb->nonresum_kernel_buf_pa,
+ tb->nonresum_qmask);
+ }
+}
+
+static void __init init_send_mondo_info(void)
+{
+ int cpu;
+
+ for_each_possible_cpu(cpu) {
+ struct trap_per_cpu *tb = &trap_block[cpu];
+
+ init_cpu_send_mondo_info(tb);
+ }
+}
+
+static struct irqaction timer_irq_action = {
+ .name = "timer",
+};
+
+/* Only invoked on boot processor. */
+void __init init_IRQ(void)
+{
+ unsigned long size;
+
+ map_prom_timers();
+ kill_prom_timer();
+
+ size = sizeof(struct ino_bucket) * NUM_IVECS;
+ ivector_table = kzalloc(size, GFP_KERNEL);
+ if (!ivector_table) {
+ prom_printf("Fatal error, cannot allocate ivector_table\n");
+ prom_halt();
+ }
+ __flush_dcache_range((unsigned long) ivector_table,
+ ((unsigned long) ivector_table) + size);
+
+ ivector_table_pa = __pa(ivector_table);
+
+ if (tlb_type == hypervisor)
+ sun4v_init_mondo_queues();
+
+ init_send_mondo_info();
+
+ if (tlb_type == hypervisor) {
+ /* Load up the boot cpu's entries. */
+ sun4v_register_mondo_queues(hard_smp_processor_id());
+ }
+
+ /* We need to clear any IRQ's pending in the soft interrupt
+ * registers, a spurious one could be left around from the
+ * PROM timer which we just disabled.
+ */
+ clear_softint(get_softint());
+
+ /* Now that ivector table is initialized, it is safe
+ * to receive IRQ vector traps. We will normally take
+ * one or two right now, in case some device PROM used
+ * to boot us wants to speak to us. We just ignore them.
+ */
+ __asm__ __volatile__("rdpr %%pstate, %%g1\n\t"
+ "or %%g1, %0, %%g1\n\t"
+ "wrpr %%g1, 0x0, %%pstate"
+ : /* No outputs */
+ : "i" (PSTATE_IE)
+ : "g1");
+
+ irq_to_desc(0)->action = &timer_irq_action;
+}
diff --git a/arch/sparc/kernel/itlb_miss.S b/arch/sparc/kernel/itlb_miss.S
new file mode 100644
index 00000000000..5a8377b5495
--- /dev/null
+++ b/arch/sparc/kernel/itlb_miss.S
@@ -0,0 +1,39 @@
+/* ITLB ** ICACHE line 1: Context 0 check and TSB load */
+ ldxa [%g0] ASI_IMMU_TSB_8KB_PTR, %g1 ! Get TSB 8K pointer
+ ldxa [%g0] ASI_IMMU, %g6 ! Get TAG TARGET
+ srlx %g6, 48, %g5 ! Get context
+ sllx %g6, 22, %g6 ! Zero out context
+ brz,pn %g5, kvmap_itlb ! Context 0 processing
+ srlx %g6, 22, %g6 ! Delay slot
+ TSB_LOAD_QUAD(%g1, %g4) ! Load TSB entry
+ cmp %g4, %g6 ! Compare TAG
+
+/* ITLB ** ICACHE line 2: TSB compare and TLB load */
+ bne,pn %xcc, tsb_miss_itlb ! Miss
+ mov FAULT_CODE_ITLB, %g3
+ sethi %hi(_PAGE_EXEC_4U), %g4
+ andcc %g5, %g4, %g0 ! Executable?
+ be,pn %xcc, tsb_do_fault
+ nop ! Delay slot, fill me
+ stxa %g5, [%g0] ASI_ITLB_DATA_IN ! Load TLB
+ retry ! Trap done
+
+/* ITLB ** ICACHE line 3: */
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+
+/* ITLB ** ICACHE line 4: */
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
+ nop
diff --git a/arch/sparc/kernel/ivec.S b/arch/sparc/kernel/ivec.S
new file mode 100644
index 00000000000..d29f92ebca5
--- /dev/null
+++ b/arch/sparc/kernel/ivec.S
@@ -0,0 +1,51 @@
+ /* The registers for cross calls will be:
+ *
+ * DATA 0: [low 32-bits] Address of function to call, jmp to this
+ * [high 32-bits] MMU Context Argument 0, place in %g5
+ * DATA 1: Address Argument 1, place in %g1
+ * DATA 2: Address Argument 2, place in %g7
+ *
+ * With this method we can do most of the cross-call tlb/cache
+ * flushing very quickly.
+ */
+ .align 32
+ .globl do_ivec
+ .type do_ivec,#function
+do_ivec:
+ mov 0x40, %g3
+ ldxa [%g3 + %g0] ASI_INTR_R, %g3
+ sethi %hi(KERNBASE), %g4
+ cmp %g3, %g4
+ bgeu,pn %xcc, do_ivec_xcall
+ srlx %g3, 32, %g5
+ stxa %g0, [%g0] ASI_INTR_RECEIVE
+ membar #Sync
+
+ sethi %hi(ivector_table_pa), %g2
+ ldx [%g2 + %lo(ivector_table_pa)], %g2
+ sllx %g3, 4, %g3
+ add %g2, %g3, %g3
+
+ TRAP_LOAD_IRQ_WORK_PA(%g6, %g1)
+
+ ldx [%g6], %g5
+ stxa %g5, [%g3] ASI_PHYS_USE_EC
+ stx %g3, [%g6]
+ wr %g0, 1 << PIL_DEVICE_IRQ, %set_softint
+ retry
+do_ivec_xcall:
+ mov 0x50, %g1
+ ldxa [%g1 + %g0] ASI_INTR_R, %g1
+ srl %g3, 0, %g3
+
+ mov 0x60, %g7
+ ldxa [%g7 + %g0] ASI_INTR_R, %g7
+ stxa %g0, [%g0] ASI_INTR_RECEIVE
+ membar #Sync
+ ba,pt %xcc, 1f
+ nop
+
+ .align 32
+1: jmpl %g3, %g0
+ nop
+ .size do_ivec,.-do_ivec
diff --git a/arch/sparc/kernel/jump_label.c b/arch/sparc/kernel/jump_label.c
new file mode 100644
index 00000000000..48565c11e82
--- /dev/null
+++ b/arch/sparc/kernel/jump_label.c
@@ -0,0 +1,41 @@
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/mutex.h>
+#include <linux/cpu.h>
+
+#include <linux/jump_label.h>
+#include <linux/memory.h>
+
+#include <asm/cacheflush.h>
+
+#ifdef HAVE_JUMP_LABEL
+
+void arch_jump_label_transform(struct jump_entry *entry,
+ enum jump_label_type type)
+{
+ u32 val;
+ u32 *insn = (u32 *) (unsigned long) entry->code;
+
+ if (type == JUMP_LABEL_ENABLE) {
+ s32 off = (s32)entry->target - (s32)entry->code;
+
+#ifdef CONFIG_SPARC64
+ /* ba,pt %xcc, . + (off << 2) */
+ val = 0x10680000 | ((u32) off >> 2);
+#else
+ /* ba . + (off << 2) */
+ val = 0x10800000 | ((u32) off >> 2);
+#endif
+ } else {
+ val = 0x01000000;
+ }
+
+ get_online_cpus();
+ mutex_lock(&text_mutex);
+ *insn = val;
+ flushi(insn);
+ mutex_unlock(&text_mutex);
+ put_online_cpus();
+}
+
+#endif
diff --git a/arch/sparc/kernel/kernel.h b/arch/sparc/kernel/kernel.h
new file mode 100644
index 00000000000..e7f652be9e6
--- /dev/null
+++ b/arch/sparc/kernel/kernel.h
@@ -0,0 +1,186 @@
+#ifndef __SPARC_KERNEL_H
+#define __SPARC_KERNEL_H
+
+#include <linux/interrupt.h>
+#include <linux/ftrace.h>
+
+#include <asm/traps.h>
+#include <asm/head.h>
+#include <asm/io.h>
+
+/* cpu.c */
+extern const char *sparc_pmu_type;
+extern unsigned int fsr_storage;
+extern int ncpus_probed;
+
+#ifdef CONFIG_SPARC64
+/* setup_64.c */
+struct seq_file;
+void cpucap_info(struct seq_file *);
+
+static inline unsigned long kimage_addr_to_ra(const void *p)
+{
+ unsigned long val = (unsigned long) p;
+
+ return kern_base + (val - KERNBASE);
+}
+
+/* sys_sparc_64.c */
+asmlinkage long sys_kern_features(void);
+
+/* unaligned_64.c */
+asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn);
+int handle_popc(u32 insn, struct pt_regs *regs);
+void handle_lddfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr);
+void handle_stdfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr);
+
+/* smp_64.c */
+void __irq_entry smp_call_function_client(int irq, struct pt_regs *regs);
+void __irq_entry smp_call_function_single_client(int irq, struct pt_regs *regs);
+void __irq_entry smp_new_mmu_context_version_client(int irq, struct pt_regs *regs);
+void __irq_entry smp_penguin_jailcell(int irq, struct pt_regs *regs);
+void __irq_entry smp_receive_signal_client(int irq, struct pt_regs *regs);
+
+/* kgdb_64.c */
+void __irq_entry smp_kgdb_capture_client(int irq, struct pt_regs *regs);
+
+/* pci.c */
+int pci64_dma_supported(struct pci_dev *pdev, u64 device_mask);
+
+/* signal32.c */
+void do_sigreturn32(struct pt_regs *regs);
+asmlinkage void do_rt_sigreturn32(struct pt_regs *regs);
+void do_signal32(struct pt_regs * regs);
+asmlinkage int do_sys32_sigstack(u32 u_ssptr, u32 u_ossptr, unsigned long sp);
+
+/* compat_audit.c */
+extern unsigned sparc32_dir_class[];
+extern unsigned sparc32_chattr_class[];
+extern unsigned sparc32_write_class[];
+extern unsigned sparc32_read_class[];
+extern unsigned sparc32_signal_class[];
+int sparc32_classify_syscall(unsigned syscall);
+#endif
+
+#ifdef CONFIG_SPARC32
+/* setup_32.c */
+struct linux_romvec;
+void sparc32_start_kernel(struct linux_romvec *rp);
+
+/* cpu.c */
+void cpu_probe(void);
+
+/* traps_32.c */
+void handle_hw_divzero(struct pt_regs *regs, unsigned long pc,
+ unsigned long npc, unsigned long psr);
+/* irq_32.c */
+extern struct irqaction static_irqaction[];
+extern int static_irq_count;
+extern spinlock_t irq_action_lock;
+
+void unexpected_irq(int irq, void *dev_id, struct pt_regs * regs);
+void init_IRQ(void);
+
+/* sun4m_irq.c */
+void sun4m_init_IRQ(void);
+void sun4m_unmask_profile_irq(void);
+void sun4m_clear_profile_irq(int cpu);
+
+/* sun4m_smp.c */
+void sun4m_cpu_pre_starting(void *arg);
+void sun4m_cpu_pre_online(void *arg);
+void __init smp4m_boot_cpus(void);
+int smp4m_boot_one_cpu(int i, struct task_struct *idle);
+void __init smp4m_smp_done(void);
+void smp4m_cross_call_irq(void);
+void smp4m_percpu_timer_interrupt(struct pt_regs *regs);
+
+/* sun4d_irq.c */
+extern spinlock_t sun4d_imsk_lock;
+
+void sun4d_init_IRQ(void);
+int sun4d_request_irq(unsigned int irq,
+ irq_handler_t handler,
+ unsigned long irqflags,
+ const char *devname, void *dev_id);
+int show_sun4d_interrupts(struct seq_file *, void *);
+void sun4d_distribute_irqs(void);
+void sun4d_free_irq(unsigned int irq, void *dev_id);
+
+/* sun4d_smp.c */
+void sun4d_cpu_pre_starting(void *arg);
+void sun4d_cpu_pre_online(void *arg);
+void __init smp4d_boot_cpus(void);
+int smp4d_boot_one_cpu(int i, struct task_struct *idle);
+void __init smp4d_smp_done(void);
+void smp4d_cross_call_irq(void);
+void smp4d_percpu_timer_interrupt(struct pt_regs *regs);
+
+/* leon_smp.c */
+void leon_cpu_pre_starting(void *arg);
+void leon_cpu_pre_online(void *arg);
+void leonsmp_ipi_interrupt(void);
+void leon_cross_call_irq(void);
+
+/* head_32.S */
+extern unsigned int t_nmi[];
+extern unsigned int linux_trap_ipi15_sun4d[];
+extern unsigned int linux_trap_ipi15_sun4m[];
+
+extern struct tt_entry trapbase_cpu1;
+extern struct tt_entry trapbase_cpu2;
+extern struct tt_entry trapbase_cpu3;
+
+extern char cputypval[];
+
+/* entry.S */
+extern unsigned long lvl14_save[4];
+extern unsigned int real_irq_entry[];
+extern unsigned int smp4d_ticker[];
+extern unsigned int patchme_maybe_smp_msg[];
+
+void floppy_hardint(void);
+
+/* trampoline_32.S */
+extern unsigned long sun4m_cpu_startup;
+extern unsigned long sun4d_cpu_startup;
+
+/* process_32.c */
+asmlinkage int sparc_do_fork(unsigned long clone_flags,
+ unsigned long stack_start,
+ struct pt_regs *regs,
+ unsigned long stack_size);
+
+/* signal_32.c */
+asmlinkage void do_sigreturn(struct pt_regs *regs);
+asmlinkage void do_rt_sigreturn(struct pt_regs *regs);
+void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0,
+ unsigned long thread_info_flags);
+asmlinkage int do_sys_sigstack(struct sigstack __user *ssptr,
+ struct sigstack __user *ossptr,
+ unsigned long sp);
+
+/* ptrace_32.c */
+asmlinkage int syscall_trace(struct pt_regs *regs, int syscall_exit_p);
+
+/* unaligned_32.c */
+asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn);
+asmlinkage void user_unaligned_trap(struct pt_regs *regs, unsigned int insn);
+
+/* windows.c */
+void try_to_clear_window_buffer(struct pt_regs *regs, int who);
+
+/* auxio_32.c */
+void __init auxio_probe(void);
+void __init auxio_power_probe(void);
+
+/* pcic.c */
+extern void __iomem *pcic_regs;
+void pcic_nmi(unsigned int pend, struct pt_regs *regs);
+
+/* time_32.c */
+void __init time_init(void);
+
+#else /* CONFIG_SPARC32 */
+#endif /* CONFIG_SPARC32 */
+#endif /* !(__SPARC_KERNEL_H) */
diff --git a/arch/sparc/kernel/kgdb_32.c b/arch/sparc/kernel/kgdb_32.c
new file mode 100644
index 00000000000..dcf210811af
--- /dev/null
+++ b/arch/sparc/kernel/kgdb_32.c
@@ -0,0 +1,172 @@
+/* kgdb.c: KGDB support for 32-bit sparc.
+ *
+ * Copyright (C) 2008 David S. Miller <davem@davemloft.net>
+ */
+
+#include <linux/kgdb.h>
+#include <linux/kdebug.h>
+#include <linux/sched.h>
+
+#include <asm/kdebug.h>
+#include <asm/ptrace.h>
+#include <asm/irq.h>
+#include <asm/cacheflush.h>
+
+extern unsigned long trapbase;
+
+void pt_regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs)
+{
+ struct reg_window32 *win;
+ int i;
+
+ gdb_regs[GDB_G0] = 0;
+ for (i = 0; i < 15; i++)
+ gdb_regs[GDB_G1 + i] = regs->u_regs[UREG_G1 + i];
+
+ win = (struct reg_window32 *) regs->u_regs[UREG_FP];
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_L0 + i] = win->locals[i];
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_I0 + i] = win->ins[i];
+
+ for (i = GDB_F0; i <= GDB_F31; i++)
+ gdb_regs[i] = 0;
+
+ gdb_regs[GDB_Y] = regs->y;
+ gdb_regs[GDB_PSR] = regs->psr;
+ gdb_regs[GDB_WIM] = 0;
+ gdb_regs[GDB_TBR] = (unsigned long) &trapbase;
+ gdb_regs[GDB_PC] = regs->pc;
+ gdb_regs[GDB_NPC] = regs->npc;
+ gdb_regs[GDB_FSR] = 0;
+ gdb_regs[GDB_CSR] = 0;
+}
+
+void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *p)
+{
+ struct thread_info *t = task_thread_info(p);
+ struct reg_window32 *win;
+ int i;
+
+ for (i = GDB_G0; i < GDB_G6; i++)
+ gdb_regs[i] = 0;
+ gdb_regs[GDB_G6] = (unsigned long) t;
+ gdb_regs[GDB_G7] = 0;
+ for (i = GDB_O0; i < GDB_SP; i++)
+ gdb_regs[i] = 0;
+ gdb_regs[GDB_SP] = t->ksp;
+ gdb_regs[GDB_O7] = 0;
+
+ win = (struct reg_window32 *) t->ksp;
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_L0 + i] = win->locals[i];
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_I0 + i] = win->ins[i];
+
+ for (i = GDB_F0; i <= GDB_F31; i++)
+ gdb_regs[i] = 0;
+
+ gdb_regs[GDB_Y] = 0;
+
+ gdb_regs[GDB_PSR] = t->kpsr;
+ gdb_regs[GDB_WIM] = t->kwim;
+ gdb_regs[GDB_TBR] = (unsigned long) &trapbase;
+ gdb_regs[GDB_PC] = t->kpc;
+ gdb_regs[GDB_NPC] = t->kpc + 4;
+ gdb_regs[GDB_FSR] = 0;
+ gdb_regs[GDB_CSR] = 0;
+}
+
+void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs)
+{
+ struct reg_window32 *win;
+ int i;
+
+ for (i = 0; i < 15; i++)
+ regs->u_regs[UREG_G1 + i] = gdb_regs[GDB_G1 + i];
+
+ /* If the PSR register is changing, we have to preserve
+ * the CWP field, otherwise window save/restore explodes.
+ */
+ if (regs->psr != gdb_regs[GDB_PSR]) {
+ unsigned long cwp = regs->psr & PSR_CWP;
+
+ regs->psr = (gdb_regs[GDB_PSR] & ~PSR_CWP) | cwp;
+ }
+
+ regs->pc = gdb_regs[GDB_PC];
+ regs->npc = gdb_regs[GDB_NPC];
+ regs->y = gdb_regs[GDB_Y];
+
+ win = (struct reg_window32 *) regs->u_regs[UREG_FP];
+ for (i = 0; i < 8; i++)
+ win->locals[i] = gdb_regs[GDB_L0 + i];
+ for (i = 0; i < 8; i++)
+ win->ins[i] = gdb_regs[GDB_I0 + i];
+}
+
+int kgdb_arch_handle_exception(int e_vector, int signo, int err_code,
+ char *remcomInBuffer, char *remcomOutBuffer,
+ struct pt_regs *linux_regs)
+{
+ unsigned long addr;
+ char *ptr;
+
+ switch (remcomInBuffer[0]) {
+ case 'c':
+ /* try to read optional parameter, pc unchanged if no parm */
+ ptr = &remcomInBuffer[1];
+ if (kgdb_hex2long(&ptr, &addr)) {
+ linux_regs->pc = addr;
+ linux_regs->npc = addr + 4;
+ }
+ /* fallthru */
+
+ case 'D':
+ case 'k':
+ if (linux_regs->pc == (unsigned long) arch_kgdb_breakpoint) {
+ linux_regs->pc = linux_regs->npc;
+ linux_regs->npc += 4;
+ }
+ return 0;
+ }
+ return -1;
+}
+
+extern void do_hw_interrupt(struct pt_regs *regs, unsigned long type);
+
+asmlinkage void kgdb_trap(struct pt_regs *regs)
+{
+ unsigned long flags;
+
+ if (user_mode(regs)) {
+ do_hw_interrupt(regs, 0xfd);
+ return;
+ }
+
+ flushw_all();
+
+ local_irq_save(flags);
+ kgdb_handle_exception(0x172, SIGTRAP, 0, regs);
+ local_irq_restore(flags);
+}
+
+int kgdb_arch_init(void)
+{
+ return 0;
+}
+
+void kgdb_arch_exit(void)
+{
+}
+
+void kgdb_arch_set_pc(struct pt_regs *regs, unsigned long ip)
+{
+ regs->pc = ip;
+ regs->npc = regs->pc + 4;
+}
+
+struct kgdb_arch arch_kgdb_ops = {
+ /* Breakpoint instruction: ta 0x7d */
+ .gdb_bpt_instr = { 0x91, 0xd0, 0x20, 0x7d },
+};
diff --git a/arch/sparc/kernel/kgdb_64.c b/arch/sparc/kernel/kgdb_64.c
new file mode 100644
index 00000000000..cbf21d0870e
--- /dev/null
+++ b/arch/sparc/kernel/kgdb_64.c
@@ -0,0 +1,200 @@
+/* kgdb.c: KGDB support for 64-bit sparc.
+ *
+ * Copyright (C) 2008 David S. Miller <davem@davemloft.net>
+ */
+
+#include <linux/kgdb.h>
+#include <linux/kdebug.h>
+#include <linux/ftrace.h>
+#include <linux/context_tracking.h>
+
+#include <asm/cacheflush.h>
+#include <asm/kdebug.h>
+#include <asm/ptrace.h>
+#include <asm/irq.h>
+
+#include "kernel.h"
+
+void pt_regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs)
+{
+ struct reg_window *win;
+ int i;
+
+ gdb_regs[GDB_G0] = 0;
+ for (i = 0; i < 15; i++)
+ gdb_regs[GDB_G1 + i] = regs->u_regs[UREG_G1 + i];
+
+ win = (struct reg_window *) (regs->u_regs[UREG_FP] + STACK_BIAS);
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_L0 + i] = win->locals[i];
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_I0 + i] = win->ins[i];
+
+ for (i = GDB_F0; i <= GDB_F62; i++)
+ gdb_regs[i] = 0;
+
+ gdb_regs[GDB_PC] = regs->tpc;
+ gdb_regs[GDB_NPC] = regs->tnpc;
+ gdb_regs[GDB_STATE] = regs->tstate;
+ gdb_regs[GDB_FSR] = 0;
+ gdb_regs[GDB_FPRS] = 0;
+ gdb_regs[GDB_Y] = regs->y;
+}
+
+void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *p)
+{
+ struct thread_info *t = task_thread_info(p);
+ extern unsigned int switch_to_pc;
+ extern unsigned int ret_from_fork;
+ struct reg_window *win;
+ unsigned long pc, cwp;
+ int i;
+
+ for (i = GDB_G0; i < GDB_G6; i++)
+ gdb_regs[i] = 0;
+ gdb_regs[GDB_G6] = (unsigned long) t;
+ gdb_regs[GDB_G7] = (unsigned long) p;
+ for (i = GDB_O0; i < GDB_SP; i++)
+ gdb_regs[i] = 0;
+ gdb_regs[GDB_SP] = t->ksp;
+ gdb_regs[GDB_O7] = 0;
+
+ win = (struct reg_window *) (t->ksp + STACK_BIAS);
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_L0 + i] = win->locals[i];
+ for (i = 0; i < 8; i++)
+ gdb_regs[GDB_I0 + i] = win->ins[i];
+
+ for (i = GDB_F0; i <= GDB_F62; i++)
+ gdb_regs[i] = 0;
+
+ if (t->new_child)
+ pc = (unsigned long) &ret_from_fork;
+ else
+ pc = (unsigned long) &switch_to_pc;
+
+ gdb_regs[GDB_PC] = pc;
+ gdb_regs[GDB_NPC] = pc + 4;
+
+ cwp = __thread_flag_byte_ptr(t)[TI_FLAG_BYTE_CWP];
+
+ gdb_regs[GDB_STATE] = (TSTATE_PRIV | TSTATE_IE | cwp);
+ gdb_regs[GDB_FSR] = 0;
+ gdb_regs[GDB_FPRS] = 0;
+ gdb_regs[GDB_Y] = 0;
+}
+
+void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs)
+{
+ struct reg_window *win;
+ int i;
+
+ for (i = 0; i < 15; i++)
+ regs->u_regs[UREG_G1 + i] = gdb_regs[GDB_G1 + i];
+
+ /* If the TSTATE register is changing, we have to preserve
+ * the CWP field, otherwise window save/restore explodes.
+ */
+ if (regs->tstate != gdb_regs[GDB_STATE]) {
+ unsigned long cwp = regs->tstate & TSTATE_CWP;
+
+ regs->tstate = (gdb_regs[GDB_STATE] & ~TSTATE_CWP) | cwp;
+ }
+
+ regs->tpc = gdb_regs[GDB_PC];
+ regs->tnpc = gdb_regs[GDB_NPC];
+ regs->y = gdb_regs[GDB_Y];
+
+ win = (struct reg_window *) (regs->u_regs[UREG_FP] + STACK_BIAS);
+ for (i = 0; i < 8; i++)
+ win->locals[i] = gdb_regs[GDB_L0 + i];
+ for (i = 0; i < 8; i++)
+ win->ins[i] = gdb_regs[GDB_I0 + i];
+}
+
+#ifdef CONFIG_SMP
+void __irq_entry smp_kgdb_capture_client(int irq, struct pt_regs *regs)
+{
+ unsigned long flags;
+
+ __asm__ __volatile__("rdpr %%pstate, %0\n\t"
+ "wrpr %0, %1, %%pstate"
+ : "=r" (flags)
+ : "i" (PSTATE_IE));
+
+ flushw_all();
+
+ if (atomic_read(&kgdb_active) != -1)
+ kgdb_nmicallback(raw_smp_processor_id(), regs);
+
+ __asm__ __volatile__("wrpr %0, 0, %%pstate"
+ : : "r" (flags));
+}
+#endif
+
+int kgdb_arch_handle_exception(int e_vector, int signo, int err_code,
+ char *remcomInBuffer, char *remcomOutBuffer,
+ struct pt_regs *linux_regs)
+{
+ unsigned long addr;
+ char *ptr;
+
+ switch (remcomInBuffer[0]) {
+ case 'c':
+ /* try to read optional parameter, pc unchanged if no parm */
+ ptr = &remcomInBuffer[1];
+ if (kgdb_hex2long(&ptr, &addr)) {
+ linux_regs->tpc = addr;
+ linux_regs->tnpc = addr + 4;
+ }
+ /* fallthru */
+
+ case 'D':
+ case 'k':
+ if (linux_regs->tpc == (unsigned long) arch_kgdb_breakpoint) {
+ linux_regs->tpc = linux_regs->tnpc;
+ linux_regs->tnpc += 4;
+ }
+ return 0;
+ }
+ return -1;
+}
+
+asmlinkage void kgdb_trap(unsigned long trap_level, struct pt_regs *regs)
+{
+ enum ctx_state prev_state = exception_enter();
+ unsigned long flags;
+
+ if (user_mode(regs)) {
+ bad_trap(regs, trap_level);
+ goto out;
+ }
+
+ flushw_all();
+
+ local_irq_save(flags);
+ kgdb_handle_exception(0x172, SIGTRAP, 0, regs);
+ local_irq_restore(flags);
+out:
+ exception_exit(prev_state);
+}
+
+int kgdb_arch_init(void)
+{
+ return 0;
+}
+
+void kgdb_arch_exit(void)
+{
+}
+
+void kgdb_arch_set_pc(struct pt_regs *regs, unsigned long ip)
+{
+ regs->tpc = ip;
+ regs->tnpc = regs->tpc + 4;
+}
+
+struct kgdb_arch arch_kgdb_ops = {
+ /* Breakpoint instruction: ta 0x72 */
+ .gdb_bpt_instr = { 0x91, 0xd0, 0x20, 0x72 },
+};
diff --git a/arch/sparc/kernel/kprobes.c b/arch/sparc/kernel/kprobes.c
new file mode 100644
index 00000000000..98d71284341
--- /dev/null
+++ b/arch/sparc/kernel/kprobes.c
@@ -0,0 +1,603 @@
+/* arch/sparc64/kernel/kprobes.c
+ *
+ * Copyright (C) 2004 David S. Miller <davem@davemloft.net>
+ */
+
+#include <linux/kernel.h>
+#include <linux/kprobes.h>
+#include <linux/module.h>
+#include <linux/kdebug.h>
+#include <linux/slab.h>
+#include <linux/context_tracking.h>
+#include <asm/signal.h>
+#include <asm/cacheflush.h>
+#include <asm/uaccess.h>
+
+/* We do not have hardware single-stepping on sparc64.
+ * So we implement software single-stepping with breakpoint
+ * traps. The top-level scheme is similar to that used
+ * in the x86 kprobes implementation.
+ *
+ * In the kprobe->ainsn.insn[] array we store the original
+ * instruction at index zero and a break instruction at
+ * index one.
+ *
+ * When we hit a kprobe we:
+ * - Run the pre-handler
+ * - Remember "regs->tnpc" and interrupt level stored in
+ * "regs->tstate" so we can restore them later
+ * - Disable PIL interrupts
+ * - Set regs->tpc to point to kprobe->ainsn.insn[0]
+ * - Set regs->tnpc to point to kprobe->ainsn.insn[1]
+ * - Mark that we are actively in a kprobe
+ *
+ * At this point we wait for the second breakpoint at
+ * kprobe->ainsn.insn[1] to hit. When it does we:
+ * - Run the post-handler
+ * - Set regs->tpc to "remembered" regs->tnpc stored above,
+ * restore the PIL interrupt level in "regs->tstate" as well
+ * - Make any adjustments necessary to regs->tnpc in order
+ * to handle relative branches correctly. See below.
+ * - Mark that we are no longer actively in a kprobe.
+ */
+
+DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
+DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
+
+struct kretprobe_blackpoint kretprobe_blacklist[] = {{NULL, NULL}};
+
+int __kprobes arch_prepare_kprobe(struct kprobe *p)
+{
+ if ((unsigned long) p->addr & 0x3UL)
+ return -EILSEQ;
+
+ p->ainsn.insn[0] = *p->addr;
+ flushi(&p->ainsn.insn[0]);
+
+ p->ainsn.insn[1] = BREAKPOINT_INSTRUCTION_2;
+ flushi(&p->ainsn.insn[1]);
+
+ p->opcode = *p->addr;
+ return 0;
+}
+
+void __kprobes arch_arm_kprobe(struct kprobe *p)
+{
+ *p->addr = BREAKPOINT_INSTRUCTION;
+ flushi(p->addr);
+}
+
+void __kprobes arch_disarm_kprobe(struct kprobe *p)
+{
+ *p->addr = p->opcode;
+ flushi(p->addr);
+}
+
+static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
+{
+ kcb->prev_kprobe.kp = kprobe_running();
+ kcb->prev_kprobe.status = kcb->kprobe_status;
+ kcb->prev_kprobe.orig_tnpc = kcb->kprobe_orig_tnpc;
+ kcb->prev_kprobe.orig_tstate_pil = kcb->kprobe_orig_tstate_pil;
+}
+
+static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb)
+{
+ __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
+ kcb->kprobe_status = kcb->prev_kprobe.status;
+ kcb->kprobe_orig_tnpc = kcb->prev_kprobe.orig_tnpc;
+ kcb->kprobe_orig_tstate_pil = kcb->prev_kprobe.orig_tstate_pil;
+}
+
+static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
+ struct kprobe_ctlblk *kcb)
+{
+ __get_cpu_var(current_kprobe) = p;
+ kcb->kprobe_orig_tnpc = regs->tnpc;
+ kcb->kprobe_orig_tstate_pil = (regs->tstate & TSTATE_PIL);
+}
+
+static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs,
+ struct kprobe_ctlblk *kcb)
+{
+ regs->tstate |= TSTATE_PIL;
+
+ /*single step inline, if it a breakpoint instruction*/
+ if (p->opcode == BREAKPOINT_INSTRUCTION) {
+ regs->tpc = (unsigned long) p->addr;
+ regs->tnpc = kcb->kprobe_orig_tnpc;
+ } else {
+ regs->tpc = (unsigned long) &p->ainsn.insn[0];
+ regs->tnpc = (unsigned long) &p->ainsn.insn[1];
+ }
+}
+
+static int __kprobes kprobe_handler(struct pt_regs *regs)
+{
+ struct kprobe *p;
+ void *addr = (void *) regs->tpc;
+ int ret = 0;
+ struct kprobe_ctlblk *kcb;
+
+ /*
+ * We don't want to be preempted for the entire
+ * duration of kprobe processing
+ */
+ preempt_disable();
+ kcb = get_kprobe_ctlblk();
+
+ if (kprobe_running()) {
+ p = get_kprobe(addr);
+ if (p) {
+ if (kcb->kprobe_status == KPROBE_HIT_SS) {
+ regs->tstate = ((regs->tstate & ~TSTATE_PIL) |
+ kcb->kprobe_orig_tstate_pil);
+ goto no_kprobe;
+ }
+ /* We have reentered the kprobe_handler(), since
+ * another probe was hit while within the handler.
+ * We here save the original kprobes variables and
+ * just single step on the instruction of the new probe
+ * without calling any user handlers.
+ */
+ save_previous_kprobe(kcb);
+ set_current_kprobe(p, regs, kcb);
+ kprobes_inc_nmissed_count(p);
+ kcb->kprobe_status = KPROBE_REENTER;
+ prepare_singlestep(p, regs, kcb);
+ return 1;
+ } else {
+ if (*(u32 *)addr != BREAKPOINT_INSTRUCTION) {
+ /* The breakpoint instruction was removed by
+ * another cpu right after we hit, no further
+ * handling of this interrupt is appropriate
+ */
+ ret = 1;
+ goto no_kprobe;
+ }
+ p = __get_cpu_var(current_kprobe);
+ if (p->break_handler && p->break_handler(p, regs))
+ goto ss_probe;
+ }
+ goto no_kprobe;
+ }
+
+ p = get_kprobe(addr);
+ if (!p) {
+ if (*(u32 *)addr != BREAKPOINT_INSTRUCTION) {
+ /*
+ * The breakpoint instruction was removed right
+ * after we hit it. Another cpu has removed
+ * either a probepoint or a debugger breakpoint
+ * at this address. In either case, no further
+ * handling of this interrupt is appropriate.
+ */
+ ret = 1;
+ }
+ /* Not one of ours: let kernel handle it */
+ goto no_kprobe;
+ }
+
+ set_current_kprobe(p, regs, kcb);
+ kcb->kprobe_status = KPROBE_HIT_ACTIVE;
+ if (p->pre_handler && p->pre_handler(p, regs))
+ return 1;
+
+ss_probe:
+ prepare_singlestep(p, regs, kcb);
+ kcb->kprobe_status = KPROBE_HIT_SS;
+ return 1;
+
+no_kprobe:
+ preempt_enable_no_resched();
+ return ret;
+}
+
+/* If INSN is a relative control transfer instruction,
+ * return the corrected branch destination value.
+ *
+ * regs->tpc and regs->tnpc still hold the values of the
+ * program counters at the time of trap due to the execution
+ * of the BREAKPOINT_INSTRUCTION_2 at p->ainsn.insn[1]
+ *
+ */
+static unsigned long __kprobes relbranch_fixup(u32 insn, struct kprobe *p,
+ struct pt_regs *regs)
+{
+ unsigned long real_pc = (unsigned long) p->addr;
+
+ /* Branch not taken, no mods necessary. */
+ if (regs->tnpc == regs->tpc + 0x4UL)
+ return real_pc + 0x8UL;
+
+ /* The three cases are call, branch w/prediction,
+ * and traditional branch.
+ */
+ if ((insn & 0xc0000000) == 0x40000000 ||
+ (insn & 0xc1c00000) == 0x00400000 ||
+ (insn & 0xc1c00000) == 0x00800000) {
+ unsigned long ainsn_addr;
+
+ ainsn_addr = (unsigned long) &p->ainsn.insn[0];
+
+ /* The instruction did all the work for us
+ * already, just apply the offset to the correct
+ * instruction location.
+ */
+ return (real_pc + (regs->tnpc - ainsn_addr));
+ }
+
+ /* It is jmpl or some other absolute PC modification instruction,
+ * leave NPC as-is.
+ */
+ return regs->tnpc;
+}
+
+/* If INSN is an instruction which writes it's PC location
+ * into a destination register, fix that up.
+ */
+static void __kprobes retpc_fixup(struct pt_regs *regs, u32 insn,
+ unsigned long real_pc)
+{
+ unsigned long *slot = NULL;
+
+ /* Simplest case is 'call', which always uses %o7 */
+ if ((insn & 0xc0000000) == 0x40000000) {
+ slot = &regs->u_regs[UREG_I7];
+ }
+
+ /* 'jmpl' encodes the register inside of the opcode */
+ if ((insn & 0xc1f80000) == 0x81c00000) {
+ unsigned long rd = ((insn >> 25) & 0x1f);
+
+ if (rd <= 15) {
+ slot = &regs->u_regs[rd];
+ } else {
+ /* Hard case, it goes onto the stack. */
+ flushw_all();
+
+ rd -= 16;
+ slot = (unsigned long *)
+ (regs->u_regs[UREG_FP] + STACK_BIAS);
+ slot += rd;
+ }
+ }
+ if (slot != NULL)
+ *slot = real_pc;
+}
+
+/*
+ * Called after single-stepping. p->addr is the address of the
+ * instruction which has been replaced by the breakpoint
+ * instruction. To avoid the SMP problems that can occur when we
+ * temporarily put back the original opcode to single-step, we
+ * single-stepped a copy of the instruction. The address of this
+ * copy is &p->ainsn.insn[0].
+ *
+ * This function prepares to return from the post-single-step
+ * breakpoint trap.
+ */
+static void __kprobes resume_execution(struct kprobe *p,
+ struct pt_regs *regs, struct kprobe_ctlblk *kcb)
+{
+ u32 insn = p->ainsn.insn[0];
+
+ regs->tnpc = relbranch_fixup(insn, p, regs);
+
+ /* This assignment must occur after relbranch_fixup() */
+ regs->tpc = kcb->kprobe_orig_tnpc;
+
+ retpc_fixup(regs, insn, (unsigned long) p->addr);
+
+ regs->tstate = ((regs->tstate & ~TSTATE_PIL) |
+ kcb->kprobe_orig_tstate_pil);
+}
+
+static int __kprobes post_kprobe_handler(struct pt_regs *regs)
+{
+ struct kprobe *cur = kprobe_running();
+ struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+
+ if (!cur)
+ return 0;
+
+ if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) {
+ kcb->kprobe_status = KPROBE_HIT_SSDONE;
+ cur->post_handler(cur, regs, 0);
+ }
+
+ resume_execution(cur, regs, kcb);
+
+ /*Restore back the original saved kprobes variables and continue. */
+ if (kcb->kprobe_status == KPROBE_REENTER) {
+ restore_previous_kprobe(kcb);
+ goto out;
+ }
+ reset_current_kprobe();
+out:
+ preempt_enable_no_resched();
+
+ return 1;
+}
+
+int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
+{
+ struct kprobe *cur = kprobe_running();
+ struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+ const struct exception_table_entry *entry;
+
+ switch(kcb->kprobe_status) {
+ case KPROBE_HIT_SS:
+ case KPROBE_REENTER:
+ /*
+ * We are here because the instruction being single
+ * stepped caused a page fault. We reset the current
+ * kprobe and the tpc points back to the probe address
+ * and allow the page fault handler to continue as a
+ * normal page fault.
+ */
+ regs->tpc = (unsigned long)cur->addr;
+ regs->tnpc = kcb->kprobe_orig_tnpc;
+ regs->tstate = ((regs->tstate & ~TSTATE_PIL) |
+ kcb->kprobe_orig_tstate_pil);
+ if (kcb->kprobe_status == KPROBE_REENTER)
+ restore_previous_kprobe(kcb);
+ else
+ reset_current_kprobe();
+ preempt_enable_no_resched();
+ break;
+ case KPROBE_HIT_ACTIVE:
+ case KPROBE_HIT_SSDONE:
+ /*
+ * We increment the nmissed count for accounting,
+ * we can also use npre/npostfault count for accounting
+ * these specific fault cases.
+ */
+ kprobes_inc_nmissed_count(cur);
+
+ /*
+ * We come here because instructions in the pre/post
+ * handler caused the page_fault, this could happen
+ * if handler tries to access user space by
+ * copy_from_user(), get_user() etc. Let the
+ * user-specified handler try to fix it first.
+ */
+ if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr))
+ return 1;
+
+ /*
+ * In case the user-specified fault handler returned
+ * zero, try to fix up.
+ */
+
+ entry = search_exception_tables(regs->tpc);
+ if (entry) {
+ regs->tpc = entry->fixup;
+ regs->tnpc = regs->tpc + 4;
+ return 1;
+ }
+
+ /*
+ * fixup_exception() could not handle it,
+ * Let do_page_fault() fix it.
+ */
+ break;
+ default:
+ break;
+ }
+
+ return 0;
+}
+
+/*
+ * Wrapper routine to for handling exceptions.
+ */
+int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
+ unsigned long val, void *data)
+{
+ struct die_args *args = (struct die_args *)data;
+ int ret = NOTIFY_DONE;
+
+ if (args->regs && user_mode(args->regs))
+ return ret;
+
+ switch (val) {
+ case DIE_DEBUG:
+ if (kprobe_handler(args->regs))
+ ret = NOTIFY_STOP;
+ break;
+ case DIE_DEBUG_2:
+ if (post_kprobe_handler(args->regs))
+ ret = NOTIFY_STOP;
+ break;
+ default:
+ break;
+ }
+ return ret;
+}
+
+asmlinkage void __kprobes kprobe_trap(unsigned long trap_level,
+ struct pt_regs *regs)
+{
+ enum ctx_state prev_state = exception_enter();
+
+ BUG_ON(trap_level != 0x170 && trap_level != 0x171);
+
+ if (user_mode(regs)) {
+ local_irq_enable();
+ bad_trap(regs, trap_level);
+ goto out;
+ }
+
+ /* trap_level == 0x170 --> ta 0x70
+ * trap_level == 0x171 --> ta 0x71
+ */
+ if (notify_die((trap_level == 0x170) ? DIE_DEBUG : DIE_DEBUG_2,
+ (trap_level == 0x170) ? "debug" : "debug_2",
+ regs, 0, trap_level, SIGTRAP) != NOTIFY_STOP)
+ bad_trap(regs, trap_level);
+out:
+ exception_exit(prev_state);
+}
+
+/* Jprobes support. */
+int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
+{
+ struct jprobe *jp = container_of(p, struct jprobe, kp);
+ struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+
+ memcpy(&(kcb->jprobe_saved_regs), regs, sizeof(*regs));
+
+ regs->tpc = (unsigned long) jp->entry;
+ regs->tnpc = ((unsigned long) jp->entry) + 0x4UL;
+ regs->tstate |= TSTATE_PIL;
+
+ return 1;
+}
+
+void __kprobes jprobe_return(void)
+{
+ struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+ register unsigned long orig_fp asm("g1");
+
+ orig_fp = kcb->jprobe_saved_regs.u_regs[UREG_FP];
+ __asm__ __volatile__("\n"
+"1: cmp %%sp, %0\n\t"
+ "blu,a,pt %%xcc, 1b\n\t"
+ " restore\n\t"
+ ".globl jprobe_return_trap_instruction\n"
+"jprobe_return_trap_instruction:\n\t"
+ "ta 0x70"
+ : /* no outputs */
+ : "r" (orig_fp));
+}
+
+extern void jprobe_return_trap_instruction(void);
+
+int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
+{
+ u32 *addr = (u32 *) regs->tpc;
+ struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+
+ if (addr == (u32 *) jprobe_return_trap_instruction) {
+ memcpy(regs, &(kcb->jprobe_saved_regs), sizeof(*regs));
+ preempt_enable_no_resched();
+ return 1;
+ }
+ return 0;
+}
+
+/* The value stored in the return address register is actually 2
+ * instructions before where the callee will return to.
+ * Sequences usually look something like this
+ *
+ * call some_function <--- return register points here
+ * nop <--- call delay slot
+ * whatever <--- where callee returns to
+ *
+ * To keep trampoline_probe_handler logic simpler, we normalize the
+ * value kept in ri->ret_addr so we don't need to keep adjusting it
+ * back and forth.
+ */
+void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
+ struct pt_regs *regs)
+{
+ ri->ret_addr = (kprobe_opcode_t *)(regs->u_regs[UREG_RETPC] + 8);
+
+ /* Replace the return addr with trampoline addr */
+ regs->u_regs[UREG_RETPC] =
+ ((unsigned long)kretprobe_trampoline) - 8;
+}
+
+/*
+ * Called when the probe at kretprobe trampoline is hit
+ */
+static int __kprobes trampoline_probe_handler(struct kprobe *p,
+ struct pt_regs *regs)
+{
+ struct kretprobe_instance *ri = NULL;
+ struct hlist_head *head, empty_rp;
+ struct hlist_node *tmp;
+ unsigned long flags, orig_ret_address = 0;
+ unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline;
+
+ INIT_HLIST_HEAD(&empty_rp);
+ kretprobe_hash_lock(current, &head, &flags);
+
+ /*
+ * It is possible to have multiple instances associated with a given
+ * task either because an multiple functions in the call path
+ * have a return probe installed on them, and/or more than one return
+ * return probe was registered for a target function.
+ *
+ * We can handle this because:
+ * - instances are always inserted at the head of the list
+ * - when multiple return probes are registered for the same
+ * function, the first instance's ret_addr will point to the
+ * real return address, and all the rest will point to
+ * kretprobe_trampoline
+ */
+ hlist_for_each_entry_safe(ri, tmp, head, hlist) {
+ if (ri->task != current)
+ /* another task is sharing our hash bucket */
+ continue;
+
+ if (ri->rp && ri->rp->handler)
+ ri->rp->handler(ri, regs);
+
+ orig_ret_address = (unsigned long)ri->ret_addr;
+ recycle_rp_inst(ri, &empty_rp);
+
+ if (orig_ret_address != trampoline_address)
+ /*
+ * This is the real return address. Any other
+ * instances associated with this task are for
+ * other calls deeper on the call stack
+ */
+ break;
+ }
+
+ kretprobe_assert(ri, orig_ret_address, trampoline_address);
+ regs->tpc = orig_ret_address;
+ regs->tnpc = orig_ret_address + 4;
+
+ reset_current_kprobe();
+ kretprobe_hash_unlock(current, &flags);
+ preempt_enable_no_resched();
+
+ hlist_for_each_entry_safe(ri, tmp, &empty_rp, hlist) {
+ hlist_del(&ri->hlist);
+ kfree(ri);
+ }
+ /*
+ * By returning a non-zero value, we are telling
+ * kprobe_handler() that we don't want the post_handler
+ * to run (and have re-enabled preemption)
+ */
+ return 1;
+}
+
+static void __used kretprobe_trampoline_holder(void)
+{
+ asm volatile(".global kretprobe_trampoline\n"
+ "kretprobe_trampoline:\n"
+ "\tnop\n"
+ "\tnop\n");
+}
+static struct kprobe trampoline_p = {
+ .addr = (kprobe_opcode_t *) &kretprobe_trampoline,
+ .pre_handler = trampoline_probe_handler
+};
+
+int __init arch_init_kprobes(void)
+{
+ return register_kprobe(&trampoline_p);
+}
+
+int __kprobes arch_trampoline_kprobe(struct kprobe *p)
+{
+ if (p->addr == (kprobe_opcode_t *)&kretprobe_trampoline)
+ return 1;
+
+ return 0;
+}
diff --git a/arch/sparc/kernel/kstack.h b/arch/sparc/kernel/kstack.h
new file mode 100644
index 00000000000..53dfb92e09f
--- /dev/null
+++ b/arch/sparc/kernel/kstack.h
@@ -0,0 +1,83 @@
+#ifndef _KSTACK_H
+#define _KSTACK_H
+
+#include <linux/thread_info.h>
+#include <linux/sched.h>
+#include <asm/ptrace.h>
+#include <asm/irq.h>
+
+/* SP must be STACK_BIAS adjusted already. */
+static inline bool kstack_valid(struct thread_info *tp, unsigned long sp)
+{
+ unsigned long base = (unsigned long) tp;
+
+ /* Stack pointer must be 16-byte aligned. */
+ if (sp & (16UL - 1))
+ return false;
+
+ if (sp >= (base + sizeof(struct thread_info)) &&
+ sp <= (base + THREAD_SIZE - sizeof(struct sparc_stackf)))
+ return true;
+
+ if (hardirq_stack[tp->cpu]) {
+ base = (unsigned long) hardirq_stack[tp->cpu];
+ if (sp >= base &&
+ sp <= (base + THREAD_SIZE - sizeof(struct sparc_stackf)))
+ return true;
+ base = (unsigned long) softirq_stack[tp->cpu];
+ if (sp >= base &&
+ sp <= (base + THREAD_SIZE - sizeof(struct sparc_stackf)))
+ return true;
+ }
+ return false;
+}
+
+/* Does "regs" point to a valid pt_regs trap frame? */
+static inline bool kstack_is_trap_frame(struct thread_info *tp, struct pt_regs *regs)
+{
+ unsigned long base = (unsigned long) tp;
+ unsigned long addr = (unsigned long) regs;
+
+ if (addr >= base &&
+ addr <= (base + THREAD_SIZE - sizeof(*regs)))
+ goto check_magic;
+
+ if (hardirq_stack[tp->cpu]) {
+ base = (unsigned long) hardirq_stack[tp->cpu];
+ if (addr >= base &&
+ addr <= (base + THREAD_SIZE - sizeof(*regs)))
+ goto check_magic;
+ base = (unsigned long) softirq_stack[tp->cpu];
+ if (addr >= base &&
+ addr <= (base + THREAD_SIZE - sizeof(*regs)))
+ goto check_magic;
+ }
+ return false;
+
+check_magic:
+ if ((regs->magic & ~0x1ff) == PT_REGS_MAGIC)
+ return true;
+ return false;
+
+}
+
+static inline __attribute__((always_inline)) void *set_hardirq_stack(void)
+{
+ void *orig_sp, *sp = hardirq_stack[smp_processor_id()];
+
+ __asm__ __volatile__("mov %%sp, %0" : "=r" (orig_sp));
+ if (orig_sp < sp ||
+ orig_sp > (sp + THREAD_SIZE)) {
+ sp += THREAD_SIZE - 192 - STACK_BIAS;
+ __asm__ __volatile__("mov %0, %%sp" : : "r" (sp));
+ }
+
+ return orig_sp;
+}
+
+static inline __attribute__((always_inline)) void restore_hardirq_stack(void *orig_sp)
+{
+ __asm__ __volatile__("mov %0, %%sp" : : "r" (orig_sp));
+}
+
+#endif /* _KSTACK_H */
diff --git a/arch/sparc/kernel/ktlb.S b/arch/sparc/kernel/ktlb.S
new file mode 100644
index 00000000000..605d4920458
--- /dev/null
+++ b/arch/sparc/kernel/ktlb.S
@@ -0,0 +1,352 @@
+/* arch/sparc64/kernel/ktlb.S: Kernel mapping TLB miss handling.
+ *
+ * Copyright (C) 1995, 1997, 2005, 2008 David S. Miller <davem@davemloft.net>
+ * Copyright (C) 1996 Eddie C. Dost (ecd@brainaid.de)
+ * Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx)
+ * Copyright (C) 1996,98,99 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+
+#include <asm/head.h>
+#include <asm/asi.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/tsb.h>
+
+ .text
+ .align 32
+
+kvmap_itlb:
+ /* g6: TAG TARGET */
+ mov TLB_TAG_ACCESS, %g4
+ ldxa [%g4] ASI_IMMU, %g4
+
+ /* sun4v_itlb_miss branches here with the missing virtual
+ * address already loaded into %g4
+ */
+kvmap_itlb_4v:
+
+ /* Catch kernel NULL pointer calls. */
+ sethi %hi(PAGE_SIZE), %g5
+ cmp %g4, %g5
+ blu,pn %xcc, kvmap_itlb_longpath
+ nop
+
+ KERN_TSB_LOOKUP_TL1(%g4, %g6, %g5, %g1, %g2, %g3, kvmap_itlb_load)
+
+kvmap_itlb_tsb_miss:
+ sethi %hi(LOW_OBP_ADDRESS), %g5
+ cmp %g4, %g5
+ blu,pn %xcc, kvmap_itlb_vmalloc_addr
+ mov 0x1, %g5
+ sllx %g5, 32, %g5
+ cmp %g4, %g5
+ blu,pn %xcc, kvmap_itlb_obp
+ nop
+
+kvmap_itlb_vmalloc_addr:
+ KERN_PGTABLE_WALK(%g4, %g5, %g2, kvmap_itlb_longpath)
+
+ TSB_LOCK_TAG(%g1, %g2, %g7)
+
+ /* Load and check PTE. */
+ ldxa [%g5] ASI_PHYS_USE_EC, %g5
+ mov 1, %g7
+ sllx %g7, TSB_TAG_INVALID_BIT, %g7
+ brgez,a,pn %g5, kvmap_itlb_longpath
+ TSB_STORE(%g1, %g7)
+
+ TSB_WRITE(%g1, %g5, %g6)
+
+ /* fallthrough to TLB load */
+
+kvmap_itlb_load:
+
+661: stxa %g5, [%g0] ASI_ITLB_DATA_IN
+ retry
+ .section .sun4v_2insn_patch, "ax"
+ .word 661b
+ nop
+ nop
+ .previous
+
+ /* For sun4v the ASI_ITLB_DATA_IN store and the retry
+ * instruction get nop'd out and we get here to branch
+ * to the sun4v tlb load code. The registers are setup
+ * as follows:
+ *
+ * %g4: vaddr
+ * %g5: PTE
+ * %g6: TAG
+ *
+ * The sun4v TLB load wants the PTE in %g3 so we fix that
+ * up here.
+ */
+ ba,pt %xcc, sun4v_itlb_load
+ mov %g5, %g3
+
+kvmap_itlb_longpath:
+
+661: rdpr %pstate, %g5
+ wrpr %g5, PSTATE_AG | PSTATE_MG, %pstate
+ .section .sun4v_2insn_patch, "ax"
+ .word 661b
+ SET_GL(1)
+ nop
+ .previous
+
+ rdpr %tpc, %g5
+ ba,pt %xcc, sparc64_realfault_common
+ mov FAULT_CODE_ITLB, %g4
+
+kvmap_itlb_obp:
+ OBP_TRANS_LOOKUP(%g4, %g5, %g2, %g3, kvmap_itlb_longpath)
+
+ TSB_LOCK_TAG(%g1, %g2, %g7)
+
+ TSB_WRITE(%g1, %g5, %g6)
+
+ ba,pt %xcc, kvmap_itlb_load
+ nop
+
+kvmap_dtlb_obp:
+ OBP_TRANS_LOOKUP(%g4, %g5, %g2, %g3, kvmap_dtlb_longpath)
+
+ TSB_LOCK_TAG(%g1, %g2, %g7)
+
+ TSB_WRITE(%g1, %g5, %g6)
+
+ ba,pt %xcc, kvmap_dtlb_load
+ nop
+
+ .align 32
+kvmap_dtlb_tsb4m_load:
+ TSB_LOCK_TAG(%g1, %g2, %g7)
+ TSB_WRITE(%g1, %g5, %g6)
+ ba,pt %xcc, kvmap_dtlb_load
+ nop
+
+kvmap_dtlb:
+ /* %g6: TAG TARGET */
+ mov TLB_TAG_ACCESS, %g4
+ ldxa [%g4] ASI_DMMU, %g4
+
+ /* sun4v_dtlb_miss branches here with the missing virtual
+ * address already loaded into %g4
+ */
+kvmap_dtlb_4v:
+ brgez,pn %g4, kvmap_dtlb_nonlinear
+ nop
+
+#ifdef CONFIG_DEBUG_PAGEALLOC
+ /* Index through the base page size TSB even for linear
+ * mappings when using page allocation debugging.
+ */
+ KERN_TSB_LOOKUP_TL1(%g4, %g6, %g5, %g1, %g2, %g3, kvmap_dtlb_load)
+#else
+ /* Correct TAG_TARGET is already in %g6, check 4mb TSB. */
+ KERN_TSB4M_LOOKUP_TL1(%g6, %g5, %g1, %g2, %g3, kvmap_dtlb_load)
+#endif
+ /* TSB entry address left in %g1, lookup linear PTE.
+ * Must preserve %g1 and %g6 (TAG).
+ */
+kvmap_dtlb_tsb4m_miss:
+ /* Clear the PAGE_OFFSET top virtual bits, shift
+ * down to get PFN, and make sure PFN is in range.
+ */
+661: sllx %g4, 0, %g5
+ .section .page_offset_shift_patch, "ax"
+ .word 661b
+ .previous
+
+ /* Check to see if we know about valid memory at the 4MB
+ * chunk this physical address will reside within.
+ */
+661: srlx %g5, MAX_PHYS_ADDRESS_BITS, %g2
+ .section .page_offset_shift_patch, "ax"
+ .word 661b
+ .previous
+
+ brnz,pn %g2, kvmap_dtlb_longpath
+ nop
+
+ /* This unconditional branch and delay-slot nop gets patched
+ * by the sethi sequence once the bitmap is properly setup.
+ */
+ .globl valid_addr_bitmap_insn
+valid_addr_bitmap_insn:
+ ba,pt %xcc, 2f
+ nop
+ .subsection 2
+ .globl valid_addr_bitmap_patch
+valid_addr_bitmap_patch:
+ sethi %hi(sparc64_valid_addr_bitmap), %g7
+ or %g7, %lo(sparc64_valid_addr_bitmap), %g7
+ .previous
+
+661: srlx %g5, ILOG2_4MB, %g2
+ .section .page_offset_shift_patch, "ax"
+ .word 661b
+ .previous
+
+ srlx %g2, 6, %g5
+ and %g2, 63, %g2
+ sllx %g5, 3, %g5
+ ldx [%g7 + %g5], %g5
+ mov 1, %g7
+ sllx %g7, %g2, %g7
+ andcc %g5, %g7, %g0
+ be,pn %xcc, kvmap_dtlb_longpath
+
+2: sethi %hi(kpte_linear_bitmap), %g2
+
+ /* Get the 256MB physical address index. */
+661: sllx %g4, 0, %g5
+ .section .page_offset_shift_patch, "ax"
+ .word 661b
+ .previous
+
+ or %g2, %lo(kpte_linear_bitmap), %g2
+
+661: srlx %g5, ILOG2_256MB, %g5
+ .section .page_offset_shift_patch, "ax"
+ .word 661b
+ .previous
+
+ and %g5, (32 - 1), %g7
+
+ /* Divide by 32 to get the offset into the bitmask. */
+ srlx %g5, 5, %g5
+ add %g7, %g7, %g7
+ sllx %g5, 3, %g5
+
+ /* kern_linear_pte_xor[(mask >> shift) & 3)] */
+ ldx [%g2 + %g5], %g2
+ srlx %g2, %g7, %g7
+ sethi %hi(kern_linear_pte_xor), %g5
+ and %g7, 3, %g7
+ or %g5, %lo(kern_linear_pte_xor), %g5
+ sllx %g7, 3, %g7
+ ldx [%g5 + %g7], %g2
+
+ .globl kvmap_linear_patch
+kvmap_linear_patch:
+ ba,pt %xcc, kvmap_dtlb_tsb4m_load
+ xor %g2, %g4, %g5
+
+kvmap_dtlb_vmalloc_addr:
+ KERN_PGTABLE_WALK(%g4, %g5, %g2, kvmap_dtlb_longpath)
+
+ TSB_LOCK_TAG(%g1, %g2, %g7)
+
+ /* Load and check PTE. */
+ ldxa [%g5] ASI_PHYS_USE_EC, %g5
+ mov 1, %g7
+ sllx %g7, TSB_TAG_INVALID_BIT, %g7
+ brgez,a,pn %g5, kvmap_dtlb_longpath
+ TSB_STORE(%g1, %g7)
+
+ TSB_WRITE(%g1, %g5, %g6)
+
+ /* fallthrough to TLB load */
+
+kvmap_dtlb_load:
+
+661: stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB
+ retry
+ .section .sun4v_2insn_patch, "ax"
+ .word 661b
+ nop
+ nop
+ .previous
+
+ /* For sun4v the ASI_DTLB_DATA_IN store and the retry
+ * instruction get nop'd out and we get here to branch
+ * to the sun4v tlb load code. The registers are setup
+ * as follows:
+ *
+ * %g4: vaddr
+ * %g5: PTE
+ * %g6: TAG
+ *
+ * The sun4v TLB load wants the PTE in %g3 so we fix that
+ * up here.
+ */
+ ba,pt %xcc, sun4v_dtlb_load
+ mov %g5, %g3
+
+#ifdef CONFIG_SPARSEMEM_VMEMMAP
+kvmap_vmemmap:
+ sub %g4, %g5, %g5
+ srlx %g5, ILOG2_4MB, %g5
+ sethi %hi(vmemmap_table), %g1
+ sllx %g5, 3, %g5
+ or %g1, %lo(vmemmap_table), %g1
+ ba,pt %xcc, kvmap_dtlb_load
+ ldx [%g1 + %g5], %g5
+#endif
+
+kvmap_dtlb_nonlinear:
+ /* Catch kernel NULL pointer derefs. */
+ sethi %hi(PAGE_SIZE), %g5
+ cmp %g4, %g5
+ bleu,pn %xcc, kvmap_dtlb_longpath
+ nop
+
+#ifdef CONFIG_SPARSEMEM_VMEMMAP
+ /* Do not use the TSB for vmemmap. */
+ mov (VMEMMAP_BASE >> 40), %g5
+ sllx %g5, 40, %g5
+ cmp %g4,%g5
+ bgeu,pn %xcc, kvmap_vmemmap
+ nop
+#endif
+
+ KERN_TSB_LOOKUP_TL1(%g4, %g6, %g5, %g1, %g2, %g3, kvmap_dtlb_load)
+
+kvmap_dtlb_tsbmiss:
+ sethi %hi(MODULES_VADDR), %g5
+ cmp %g4, %g5
+ blu,pn %xcc, kvmap_dtlb_longpath
+ mov (VMALLOC_END >> 40), %g5
+ sllx %g5, 40, %g5
+ cmp %g4, %g5
+ bgeu,pn %xcc, kvmap_dtlb_longpath
+ nop
+
+kvmap_check_obp:
+ sethi %hi(LOW_OBP_ADDRESS), %g5
+ cmp %g4, %g5
+ blu,pn %xcc, kvmap_dtlb_vmalloc_addr
+ mov 0x1, %g5
+ sllx %g5, 32, %g5
+ cmp %g4, %g5
+ blu,pn %xcc, kvmap_dtlb_obp
+ nop
+ ba,pt %xcc, kvmap_dtlb_vmalloc_addr
+ nop
+
+kvmap_dtlb_longpath:
+
+661: rdpr %pstate, %g5
+ wrpr %g5, PSTATE_AG | PSTATE_MG, %pstate
+ .section .sun4v_2insn_patch, "ax"
+ .word 661b
+ SET_GL(1)
+ ldxa [%g0] ASI_SCRATCHPAD, %g5
+ .previous
+
+ rdpr %tl, %g3
+ cmp %g3, 1
+
+661: mov TLB_TAG_ACCESS, %g4
+ ldxa [%g4] ASI_DMMU, %g5
+ .section .sun4v_2insn_patch, "ax"
+ .word 661b
+ ldx [%g5 + HV_FAULT_D_ADDR_OFFSET], %g5
+ nop
+ .previous
+
+ be,pt %xcc, sparc64_realfault_common
+ mov FAULT_CODE_DTLB, %g4
+ ba,pt %xcc, winfix_trampoline
+ nop
diff --git a/arch/sparc/kernel/ldc.c b/arch/sparc/kernel/ldc.c
new file mode 100644
index 00000000000..e01d75d4032
--- /dev/null
+++ b/arch/sparc/kernel/ldc.c
@@ -0,0 +1,2374 @@
+/* ldc.c: Logical Domain Channel link-layer protocol driver.
+ *
+ * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
+ */
+
+#include <linux/kernel.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <linux/delay.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/scatterlist.h>
+#include <linux/interrupt.h>
+#include <linux/list.h>
+#include <linux/init.h>
+#include <linux/bitmap.h>
+
+#include <asm/hypervisor.h>
+#include <asm/iommu.h>
+#include <asm/page.h>
+#include <asm/ldc.h>
+#include <asm/mdesc.h>
+
+#define DRV_MODULE_NAME "ldc"
+#define PFX DRV_MODULE_NAME ": "
+#define DRV_MODULE_VERSION "1.1"
+#define DRV_MODULE_RELDATE "July 22, 2008"
+
+static char version[] =
+ DRV_MODULE_NAME ".c:v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
+#define LDC_PACKET_SIZE 64
+
+/* Packet header layout for unreliable and reliable mode frames.
+ * When in RAW mode, packets are simply straight 64-byte payloads
+ * with no headers.
+ */
+struct ldc_packet {
+ u8 type;
+#define LDC_CTRL 0x01
+#define LDC_DATA 0x02
+#define LDC_ERR 0x10
+
+ u8 stype;
+#define LDC_INFO 0x01
+#define LDC_ACK 0x02
+#define LDC_NACK 0x04
+
+ u8 ctrl;
+#define LDC_VERS 0x01 /* Link Version */
+#define LDC_RTS 0x02 /* Request To Send */
+#define LDC_RTR 0x03 /* Ready To Receive */
+#define LDC_RDX 0x04 /* Ready for Data eXchange */
+#define LDC_CTRL_MSK 0x0f
+
+ u8 env;
+#define LDC_LEN 0x3f
+#define LDC_FRAG_MASK 0xc0
+#define LDC_START 0x40
+#define LDC_STOP 0x80
+
+ u32 seqid;
+
+ union {
+ u8 u_data[LDC_PACKET_SIZE - 8];
+ struct {
+ u32 pad;
+ u32 ackid;
+ u8 r_data[LDC_PACKET_SIZE - 8 - 8];
+ } r;
+ } u;
+};
+
+struct ldc_version {
+ u16 major;
+ u16 minor;
+};
+
+/* Ordered from largest major to lowest. */
+static struct ldc_version ver_arr[] = {
+ { .major = 1, .minor = 0 },
+};
+
+#define LDC_DEFAULT_MTU (4 * LDC_PACKET_SIZE)
+#define LDC_DEFAULT_NUM_ENTRIES (PAGE_SIZE / LDC_PACKET_SIZE)
+
+struct ldc_channel;
+
+struct ldc_mode_ops {
+ int (*write)(struct ldc_channel *, const void *, unsigned int);
+ int (*read)(struct ldc_channel *, void *, unsigned int);
+};
+
+static const struct ldc_mode_ops raw_ops;
+static const struct ldc_mode_ops nonraw_ops;
+static const struct ldc_mode_ops stream_ops;
+
+int ldom_domaining_enabled;
+
+struct ldc_iommu {
+ /* Protects arena alloc/free. */
+ spinlock_t lock;
+ struct iommu_arena arena;
+ struct ldc_mtable_entry *page_table;
+};
+
+struct ldc_channel {
+ /* Protects all operations that depend upon channel state. */
+ spinlock_t lock;
+
+ unsigned long id;
+
+ u8 *mssbuf;
+ u32 mssbuf_len;
+ u32 mssbuf_off;
+
+ struct ldc_packet *tx_base;
+ unsigned long tx_head;
+ unsigned long tx_tail;
+ unsigned long tx_num_entries;
+ unsigned long tx_ra;
+
+ unsigned long tx_acked;
+
+ struct ldc_packet *rx_base;
+ unsigned long rx_head;
+ unsigned long rx_tail;
+ unsigned long rx_num_entries;
+ unsigned long rx_ra;
+
+ u32 rcv_nxt;
+ u32 snd_nxt;
+
+ unsigned long chan_state;
+
+ struct ldc_channel_config cfg;
+ void *event_arg;
+
+ const struct ldc_mode_ops *mops;
+
+ struct ldc_iommu iommu;
+
+ struct ldc_version ver;
+
+ u8 hs_state;
+#define LDC_HS_CLOSED 0x00
+#define LDC_HS_OPEN 0x01
+#define LDC_HS_GOTVERS 0x02
+#define LDC_HS_SENTRTR 0x03
+#define LDC_HS_GOTRTR 0x04
+#define LDC_HS_COMPLETE 0x10
+
+ u8 flags;
+#define LDC_FLAG_ALLOCED_QUEUES 0x01
+#define LDC_FLAG_REGISTERED_QUEUES 0x02
+#define LDC_FLAG_REGISTERED_IRQS 0x04
+#define LDC_FLAG_RESET 0x10
+
+ u8 mss;
+ u8 state;
+
+#define LDC_IRQ_NAME_MAX 32
+ char rx_irq_name[LDC_IRQ_NAME_MAX];
+ char tx_irq_name[LDC_IRQ_NAME_MAX];
+
+ struct hlist_head mh_list;
+
+ struct hlist_node list;
+};
+
+#define ldcdbg(TYPE, f, a...) \
+do { if (lp->cfg.debug & LDC_DEBUG_##TYPE) \
+ printk(KERN_INFO PFX "ID[%lu] " f, lp->id, ## a); \
+} while (0)
+
+static const char *state_to_str(u8 state)
+{
+ switch (state) {
+ case LDC_STATE_INVALID:
+ return "INVALID";
+ case LDC_STATE_INIT:
+ return "INIT";
+ case LDC_STATE_BOUND:
+ return "BOUND";
+ case LDC_STATE_READY:
+ return "READY";
+ case LDC_STATE_CONNECTED:
+ return "CONNECTED";
+ default:
+ return "<UNKNOWN>";
+ }
+}
+
+static void ldc_set_state(struct ldc_channel *lp, u8 state)
+{
+ ldcdbg(STATE, "STATE (%s) --> (%s)\n",
+ state_to_str(lp->state),
+ state_to_str(state));
+
+ lp->state = state;
+}
+
+static unsigned long __advance(unsigned long off, unsigned long num_entries)
+{
+ off += LDC_PACKET_SIZE;
+ if (off == (num_entries * LDC_PACKET_SIZE))
+ off = 0;
+
+ return off;
+}
+
+static unsigned long rx_advance(struct ldc_channel *lp, unsigned long off)
+{
+ return __advance(off, lp->rx_num_entries);
+}
+
+static unsigned long tx_advance(struct ldc_channel *lp, unsigned long off)
+{
+ return __advance(off, lp->tx_num_entries);
+}
+
+static struct ldc_packet *handshake_get_tx_packet(struct ldc_channel *lp,
+ unsigned long *new_tail)
+{
+ struct ldc_packet *p;
+ unsigned long t;
+
+ t = tx_advance(lp, lp->tx_tail);
+ if (t == lp->tx_head)
+ return NULL;
+
+ *new_tail = t;
+
+ p = lp->tx_base;
+ return p + (lp->tx_tail / LDC_PACKET_SIZE);
+}
+
+/* When we are in reliable or stream mode, have to track the next packet
+ * we haven't gotten an ACK for in the TX queue using tx_acked. We have
+ * to be careful not to stomp over the queue past that point. During
+ * the handshake, we don't have TX data packets pending in the queue
+ * and that's why handshake_get_tx_packet() need not be mindful of
+ * lp->tx_acked.
+ */
+static unsigned long head_for_data(struct ldc_channel *lp)
+{
+ if (lp->cfg.mode == LDC_MODE_STREAM)
+ return lp->tx_acked;
+ return lp->tx_head;
+}
+
+static int tx_has_space_for(struct ldc_channel *lp, unsigned int size)
+{
+ unsigned long limit, tail, new_tail, diff;
+ unsigned int mss;
+
+ limit = head_for_data(lp);
+ tail = lp->tx_tail;
+ new_tail = tx_advance(lp, tail);
+ if (new_tail == limit)
+ return 0;
+
+ if (limit > new_tail)
+ diff = limit - new_tail;
+ else
+ diff = (limit +
+ ((lp->tx_num_entries * LDC_PACKET_SIZE) - new_tail));
+ diff /= LDC_PACKET_SIZE;
+ mss = lp->mss;
+
+ if (diff * mss < size)
+ return 0;
+
+ return 1;
+}
+
+static struct ldc_packet *data_get_tx_packet(struct ldc_channel *lp,
+ unsigned long *new_tail)
+{
+ struct ldc_packet *p;
+ unsigned long h, t;
+
+ h = head_for_data(lp);
+ t = tx_advance(lp, lp->tx_tail);
+ if (t == h)
+ return NULL;
+
+ *new_tail = t;
+
+ p = lp->tx_base;
+ return p + (lp->tx_tail / LDC_PACKET_SIZE);
+}
+
+static int set_tx_tail(struct ldc_channel *lp, unsigned long tail)
+{
+ unsigned long orig_tail = lp->tx_tail;
+ int limit = 1000;
+
+ lp->tx_tail = tail;
+ while (limit-- > 0) {
+ unsigned long err;
+
+ err = sun4v_ldc_tx_set_qtail(lp->id, tail);
+ if (!err)
+ return 0;
+
+ if (err != HV_EWOULDBLOCK) {
+ lp->tx_tail = orig_tail;
+ return -EINVAL;
+ }
+ udelay(1);
+ }
+
+ lp->tx_tail = orig_tail;
+ return -EBUSY;
+}
+
+/* This just updates the head value in the hypervisor using
+ * a polling loop with a timeout. The caller takes care of
+ * upating software state representing the head change, if any.
+ */
+static int __set_rx_head(struct ldc_channel *lp, unsigned long head)
+{
+ int limit = 1000;
+
+ while (limit-- > 0) {
+ unsigned long err;
+
+ err = sun4v_ldc_rx_set_qhead(lp->id, head);
+ if (!err)
+ return 0;
+
+ if (err != HV_EWOULDBLOCK)
+ return -EINVAL;
+
+ udelay(1);
+ }
+
+ return -EBUSY;
+}
+
+static int send_tx_packet(struct ldc_channel *lp,
+ struct ldc_packet *p,
+ unsigned long new_tail)
+{
+ BUG_ON(p != (lp->tx_base + (lp->tx_tail / LDC_PACKET_SIZE)));
+
+ return set_tx_tail(lp, new_tail);
+}
+
+static struct ldc_packet *handshake_compose_ctrl(struct ldc_channel *lp,
+ u8 stype, u8 ctrl,
+ void *data, int dlen,
+ unsigned long *new_tail)
+{
+ struct ldc_packet *p = handshake_get_tx_packet(lp, new_tail);
+
+ if (p) {
+ memset(p, 0, sizeof(*p));
+ p->type = LDC_CTRL;
+ p->stype = stype;
+ p->ctrl = ctrl;
+ if (data)
+ memcpy(p->u.u_data, data, dlen);
+ }
+ return p;
+}
+
+static int start_handshake(struct ldc_channel *lp)
+{
+ struct ldc_packet *p;
+ struct ldc_version *ver;
+ unsigned long new_tail;
+
+ ver = &ver_arr[0];
+
+ ldcdbg(HS, "SEND VER INFO maj[%u] min[%u]\n",
+ ver->major, ver->minor);
+
+ p = handshake_compose_ctrl(lp, LDC_INFO, LDC_VERS,
+ ver, sizeof(*ver), &new_tail);
+ if (p) {
+ int err = send_tx_packet(lp, p, new_tail);
+ if (!err)
+ lp->flags &= ~LDC_FLAG_RESET;
+ return err;
+ }
+ return -EBUSY;
+}
+
+static int send_version_nack(struct ldc_channel *lp,
+ u16 major, u16 minor)
+{
+ struct ldc_packet *p;
+ struct ldc_version ver;
+ unsigned long new_tail;
+
+ ver.major = major;
+ ver.minor = minor;
+
+ p = handshake_compose_ctrl(lp, LDC_NACK, LDC_VERS,
+ &ver, sizeof(ver), &new_tail);
+ if (p) {
+ ldcdbg(HS, "SEND VER NACK maj[%u] min[%u]\n",
+ ver.major, ver.minor);
+
+ return send_tx_packet(lp, p, new_tail);
+ }
+ return -EBUSY;
+}
+
+static int send_version_ack(struct ldc_channel *lp,
+ struct ldc_version *vp)
+{
+ struct ldc_packet *p;
+ unsigned long new_tail;
+
+ p = handshake_compose_ctrl(lp, LDC_ACK, LDC_VERS,
+ vp, sizeof(*vp), &new_tail);
+ if (p) {
+ ldcdbg(HS, "SEND VER ACK maj[%u] min[%u]\n",
+ vp->major, vp->minor);
+
+ return send_tx_packet(lp, p, new_tail);
+ }
+ return -EBUSY;
+}
+
+static int send_rts(struct ldc_channel *lp)
+{
+ struct ldc_packet *p;
+ unsigned long new_tail;
+
+ p = handshake_compose_ctrl(lp, LDC_INFO, LDC_RTS, NULL, 0,
+ &new_tail);
+ if (p) {
+ p->env = lp->cfg.mode;
+ p->seqid = 0;
+ lp->rcv_nxt = 0;
+
+ ldcdbg(HS, "SEND RTS env[0x%x] seqid[0x%x]\n",
+ p->env, p->seqid);
+
+ return send_tx_packet(lp, p, new_tail);
+ }
+ return -EBUSY;
+}
+
+static int send_rtr(struct ldc_channel *lp)
+{
+ struct ldc_packet *p;
+ unsigned long new_tail;
+
+ p = handshake_compose_ctrl(lp, LDC_INFO, LDC_RTR, NULL, 0,
+ &new_tail);
+ if (p) {
+ p->env = lp->cfg.mode;
+ p->seqid = 0;
+
+ ldcdbg(HS, "SEND RTR env[0x%x] seqid[0x%x]\n",
+ p->env, p->seqid);
+
+ return send_tx_packet(lp, p, new_tail);
+ }
+ return -EBUSY;
+}
+
+static int send_rdx(struct ldc_channel *lp)
+{
+ struct ldc_packet *p;
+ unsigned long new_tail;
+
+ p = handshake_compose_ctrl(lp, LDC_INFO, LDC_RDX, NULL, 0,
+ &new_tail);
+ if (p) {
+ p->env = 0;
+ p->seqid = ++lp->snd_nxt;
+ p->u.r.ackid = lp->rcv_nxt;
+
+ ldcdbg(HS, "SEND RDX env[0x%x] seqid[0x%x] ackid[0x%x]\n",
+ p->env, p->seqid, p->u.r.ackid);
+
+ return send_tx_packet(lp, p, new_tail);
+ }
+ return -EBUSY;
+}
+
+static int send_data_nack(struct ldc_channel *lp, struct ldc_packet *data_pkt)
+{
+ struct ldc_packet *p;
+ unsigned long new_tail;
+ int err;
+
+ p = data_get_tx_packet(lp, &new_tail);
+ if (!p)
+ return -EBUSY;
+ memset(p, 0, sizeof(*p));
+ p->type = data_pkt->type;
+ p->stype = LDC_NACK;
+ p->ctrl = data_pkt->ctrl & LDC_CTRL_MSK;
+ p->seqid = lp->snd_nxt + 1;
+ p->u.r.ackid = lp->rcv_nxt;
+
+ ldcdbg(HS, "SEND DATA NACK type[0x%x] ctl[0x%x] seq[0x%x] ack[0x%x]\n",
+ p->type, p->ctrl, p->seqid, p->u.r.ackid);
+
+ err = send_tx_packet(lp, p, new_tail);
+ if (!err)
+ lp->snd_nxt++;
+
+ return err;
+}
+
+static int ldc_abort(struct ldc_channel *lp)
+{
+ unsigned long hv_err;
+
+ ldcdbg(STATE, "ABORT\n");
+
+ /* We report but do not act upon the hypervisor errors because
+ * there really isn't much we can do if they fail at this point.
+ */
+ hv_err = sun4v_ldc_tx_qconf(lp->id, lp->tx_ra, lp->tx_num_entries);
+ if (hv_err)
+ printk(KERN_ERR PFX "ldc_abort: "
+ "sun4v_ldc_tx_qconf(%lx,%lx,%lx) failed, err=%lu\n",
+ lp->id, lp->tx_ra, lp->tx_num_entries, hv_err);
+
+ hv_err = sun4v_ldc_tx_get_state(lp->id,
+ &lp->tx_head,
+ &lp->tx_tail,
+ &lp->chan_state);
+ if (hv_err)
+ printk(KERN_ERR PFX "ldc_abort: "
+ "sun4v_ldc_tx_get_state(%lx,...) failed, err=%lu\n",
+ lp->id, hv_err);
+
+ hv_err = sun4v_ldc_rx_qconf(lp->id, lp->rx_ra, lp->rx_num_entries);
+ if (hv_err)
+ printk(KERN_ERR PFX "ldc_abort: "
+ "sun4v_ldc_rx_qconf(%lx,%lx,%lx) failed, err=%lu\n",
+ lp->id, lp->rx_ra, lp->rx_num_entries, hv_err);
+
+ /* Refetch the RX queue state as well, because we could be invoked
+ * here in the queue processing context.
+ */
+ hv_err = sun4v_ldc_rx_get_state(lp->id,
+ &lp->rx_head,
+ &lp->rx_tail,
+ &lp->chan_state);
+ if (hv_err)
+ printk(KERN_ERR PFX "ldc_abort: "
+ "sun4v_ldc_rx_get_state(%lx,...) failed, err=%lu\n",
+ lp->id, hv_err);
+
+ return -ECONNRESET;
+}
+
+static struct ldc_version *find_by_major(u16 major)
+{
+ struct ldc_version *ret = NULL;
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(ver_arr); i++) {
+ struct ldc_version *v = &ver_arr[i];
+ if (v->major <= major) {
+ ret = v;
+ break;
+ }
+ }
+ return ret;
+}
+
+static int process_ver_info(struct ldc_channel *lp, struct ldc_version *vp)
+{
+ struct ldc_version *vap;
+ int err;
+
+ ldcdbg(HS, "GOT VERSION INFO major[%x] minor[%x]\n",
+ vp->major, vp->minor);
+
+ if (lp->hs_state == LDC_HS_GOTVERS) {
+ lp->hs_state = LDC_HS_OPEN;
+ memset(&lp->ver, 0, sizeof(lp->ver));
+ }
+
+ vap = find_by_major(vp->major);
+ if (!vap) {
+ err = send_version_nack(lp, 0, 0);
+ } else if (vap->major != vp->major) {
+ err = send_version_nack(lp, vap->major, vap->minor);
+ } else {
+ struct ldc_version ver = *vp;
+ if (ver.minor > vap->minor)
+ ver.minor = vap->minor;
+ err = send_version_ack(lp, &ver);
+ if (!err) {
+ lp->ver = ver;
+ lp->hs_state = LDC_HS_GOTVERS;
+ }
+ }
+ if (err)
+ return ldc_abort(lp);
+
+ return 0;
+}
+
+static int process_ver_ack(struct ldc_channel *lp, struct ldc_version *vp)
+{
+ ldcdbg(HS, "GOT VERSION ACK major[%x] minor[%x]\n",
+ vp->major, vp->minor);
+
+ if (lp->hs_state == LDC_HS_GOTVERS) {
+ if (lp->ver.major != vp->major ||
+ lp->ver.minor != vp->minor)
+ return ldc_abort(lp);
+ } else {
+ lp->ver = *vp;
+ lp->hs_state = LDC_HS_GOTVERS;
+ }
+ if (send_rts(lp))
+ return ldc_abort(lp);
+ return 0;
+}
+
+static int process_ver_nack(struct ldc_channel *lp, struct ldc_version *vp)
+{
+ struct ldc_version *vap;
+ struct ldc_packet *p;
+ unsigned long new_tail;
+
+ if (vp->major == 0 && vp->minor == 0)
+ return ldc_abort(lp);
+
+ vap = find_by_major(vp->major);
+ if (!vap)
+ return ldc_abort(lp);
+
+ p = handshake_compose_ctrl(lp, LDC_INFO, LDC_VERS,
+ vap, sizeof(*vap),
+ &new_tail);
+ if (!p)
+ return ldc_abort(lp);
+
+ return send_tx_packet(lp, p, new_tail);
+}
+
+static int process_version(struct ldc_channel *lp,
+ struct ldc_packet *p)
+{
+ struct ldc_version *vp;
+
+ vp = (struct ldc_version *) p->u.u_data;
+
+ switch (p->stype) {
+ case LDC_INFO:
+ return process_ver_info(lp, vp);
+
+ case LDC_ACK:
+ return process_ver_ack(lp, vp);
+
+ case LDC_NACK:
+ return process_ver_nack(lp, vp);
+
+ default:
+ return ldc_abort(lp);
+ }
+}
+
+static int process_rts(struct ldc_channel *lp,
+ struct ldc_packet *p)
+{
+ ldcdbg(HS, "GOT RTS stype[%x] seqid[%x] env[%x]\n",
+ p->stype, p->seqid, p->env);
+
+ if (p->stype != LDC_INFO ||
+ lp->hs_state != LDC_HS_GOTVERS ||
+ p->env != lp->cfg.mode)
+ return ldc_abort(lp);
+
+ lp->snd_nxt = p->seqid;
+ lp->rcv_nxt = p->seqid;
+ lp->hs_state = LDC_HS_SENTRTR;
+ if (send_rtr(lp))
+ return ldc_abort(lp);
+
+ return 0;
+}
+
+static int process_rtr(struct ldc_channel *lp,
+ struct ldc_packet *p)
+{
+ ldcdbg(HS, "GOT RTR stype[%x] seqid[%x] env[%x]\n",
+ p->stype, p->seqid, p->env);
+
+ if (p->stype != LDC_INFO ||
+ p->env != lp->cfg.mode)
+ return ldc_abort(lp);
+
+ lp->snd_nxt = p->seqid;
+ lp->hs_state = LDC_HS_COMPLETE;
+ ldc_set_state(lp, LDC_STATE_CONNECTED);
+ send_rdx(lp);
+
+ return LDC_EVENT_UP;
+}
+
+static int rx_seq_ok(struct ldc_channel *lp, u32 seqid)
+{
+ return lp->rcv_nxt + 1 == seqid;
+}
+
+static int process_rdx(struct ldc_channel *lp,
+ struct ldc_packet *p)
+{
+ ldcdbg(HS, "GOT RDX stype[%x] seqid[%x] env[%x] ackid[%x]\n",
+ p->stype, p->seqid, p->env, p->u.r.ackid);
+
+ if (p->stype != LDC_INFO ||
+ !(rx_seq_ok(lp, p->seqid)))
+ return ldc_abort(lp);
+
+ lp->rcv_nxt = p->seqid;
+
+ lp->hs_state = LDC_HS_COMPLETE;
+ ldc_set_state(lp, LDC_STATE_CONNECTED);
+
+ return LDC_EVENT_UP;
+}
+
+static int process_control_frame(struct ldc_channel *lp,
+ struct ldc_packet *p)
+{
+ switch (p->ctrl) {
+ case LDC_VERS:
+ return process_version(lp, p);
+
+ case LDC_RTS:
+ return process_rts(lp, p);
+
+ case LDC_RTR:
+ return process_rtr(lp, p);
+
+ case LDC_RDX:
+ return process_rdx(lp, p);
+
+ default:
+ return ldc_abort(lp);
+ }
+}
+
+static int process_error_frame(struct ldc_channel *lp,
+ struct ldc_packet *p)
+{
+ return ldc_abort(lp);
+}
+
+static int process_data_ack(struct ldc_channel *lp,
+ struct ldc_packet *ack)
+{
+ unsigned long head = lp->tx_acked;
+ u32 ackid = ack->u.r.ackid;
+
+ while (1) {
+ struct ldc_packet *p = lp->tx_base + (head / LDC_PACKET_SIZE);
+
+ head = tx_advance(lp, head);
+
+ if (p->seqid == ackid) {
+ lp->tx_acked = head;
+ return 0;
+ }
+ if (head == lp->tx_tail)
+ return ldc_abort(lp);
+ }
+
+ return 0;
+}
+
+static void send_events(struct ldc_channel *lp, unsigned int event_mask)
+{
+ if (event_mask & LDC_EVENT_RESET)
+ lp->cfg.event(lp->event_arg, LDC_EVENT_RESET);
+ if (event_mask & LDC_EVENT_UP)
+ lp->cfg.event(lp->event_arg, LDC_EVENT_UP);
+ if (event_mask & LDC_EVENT_DATA_READY)
+ lp->cfg.event(lp->event_arg, LDC_EVENT_DATA_READY);
+}
+
+static irqreturn_t ldc_rx(int irq, void *dev_id)
+{
+ struct ldc_channel *lp = dev_id;
+ unsigned long orig_state, flags;
+ unsigned int event_mask;
+
+ spin_lock_irqsave(&lp->lock, flags);
+
+ orig_state = lp->chan_state;
+
+ /* We should probably check for hypervisor errors here and
+ * reset the LDC channel if we get one.
+ */
+ sun4v_ldc_rx_get_state(lp->id,
+ &lp->rx_head,
+ &lp->rx_tail,
+ &lp->chan_state);
+
+ ldcdbg(RX, "RX state[0x%02lx:0x%02lx] head[0x%04lx] tail[0x%04lx]\n",
+ orig_state, lp->chan_state, lp->rx_head, lp->rx_tail);
+
+ event_mask = 0;
+
+ if (lp->cfg.mode == LDC_MODE_RAW &&
+ lp->chan_state == LDC_CHANNEL_UP) {
+ lp->hs_state = LDC_HS_COMPLETE;
+ ldc_set_state(lp, LDC_STATE_CONNECTED);
+
+ event_mask |= LDC_EVENT_UP;
+
+ orig_state = lp->chan_state;
+ }
+
+ /* If we are in reset state, flush the RX queue and ignore
+ * everything.
+ */
+ if (lp->flags & LDC_FLAG_RESET) {
+ (void) __set_rx_head(lp, lp->rx_tail);
+ goto out;
+ }
+
+ /* Once we finish the handshake, we let the ldc_read()
+ * paths do all of the control frame and state management.
+ * Just trigger the callback.
+ */
+ if (lp->hs_state == LDC_HS_COMPLETE) {
+handshake_complete:
+ if (lp->chan_state != orig_state) {
+ unsigned int event = LDC_EVENT_RESET;
+
+ if (lp->chan_state == LDC_CHANNEL_UP)
+ event = LDC_EVENT_UP;
+
+ event_mask |= event;
+ }
+ if (lp->rx_head != lp->rx_tail)
+ event_mask |= LDC_EVENT_DATA_READY;
+
+ goto out;
+ }
+
+ if (lp->chan_state != orig_state)
+ goto out;
+
+ while (lp->rx_head != lp->rx_tail) {
+ struct ldc_packet *p;
+ unsigned long new;
+ int err;
+
+ p = lp->rx_base + (lp->rx_head / LDC_PACKET_SIZE);
+
+ switch (p->type) {
+ case LDC_CTRL:
+ err = process_control_frame(lp, p);
+ if (err > 0)
+ event_mask |= err;
+ break;
+
+ case LDC_DATA:
+ event_mask |= LDC_EVENT_DATA_READY;
+ err = 0;
+ break;
+
+ case LDC_ERR:
+ err = process_error_frame(lp, p);
+ break;
+
+ default:
+ err = ldc_abort(lp);
+ break;
+ }
+
+ if (err < 0)
+ break;
+
+ new = lp->rx_head;
+ new += LDC_PACKET_SIZE;
+ if (new == (lp->rx_num_entries * LDC_PACKET_SIZE))
+ new = 0;
+ lp->rx_head = new;
+
+ err = __set_rx_head(lp, new);
+ if (err < 0) {
+ (void) ldc_abort(lp);
+ break;
+ }
+ if (lp->hs_state == LDC_HS_COMPLETE)
+ goto handshake_complete;
+ }
+
+out:
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ send_events(lp, event_mask);
+
+ return IRQ_HANDLED;
+}
+
+static irqreturn_t ldc_tx(int irq, void *dev_id)
+{
+ struct ldc_channel *lp = dev_id;
+ unsigned long flags, orig_state;
+ unsigned int event_mask = 0;
+
+ spin_lock_irqsave(&lp->lock, flags);
+
+ orig_state = lp->chan_state;
+
+ /* We should probably check for hypervisor errors here and
+ * reset the LDC channel if we get one.
+ */
+ sun4v_ldc_tx_get_state(lp->id,
+ &lp->tx_head,
+ &lp->tx_tail,
+ &lp->chan_state);
+
+ ldcdbg(TX, " TX state[0x%02lx:0x%02lx] head[0x%04lx] tail[0x%04lx]\n",
+ orig_state, lp->chan_state, lp->tx_head, lp->tx_tail);
+
+ if (lp->cfg.mode == LDC_MODE_RAW &&
+ lp->chan_state == LDC_CHANNEL_UP) {
+ lp->hs_state = LDC_HS_COMPLETE;
+ ldc_set_state(lp, LDC_STATE_CONNECTED);
+
+ event_mask |= LDC_EVENT_UP;
+ }
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ send_events(lp, event_mask);
+
+ return IRQ_HANDLED;
+}
+
+/* XXX ldc_alloc() and ldc_free() needs to run under a mutex so
+ * XXX that addition and removal from the ldc_channel_list has
+ * XXX atomicity, otherwise the __ldc_channel_exists() check is
+ * XXX totally pointless as another thread can slip into ldc_alloc()
+ * XXX and add a channel with the same ID. There also needs to be
+ * XXX a spinlock for ldc_channel_list.
+ */
+static HLIST_HEAD(ldc_channel_list);
+
+static int __ldc_channel_exists(unsigned long id)
+{
+ struct ldc_channel *lp;
+
+ hlist_for_each_entry(lp, &ldc_channel_list, list) {
+ if (lp->id == id)
+ return 1;
+ }
+ return 0;
+}
+
+static int alloc_queue(const char *name, unsigned long num_entries,
+ struct ldc_packet **base, unsigned long *ra)
+{
+ unsigned long size, order;
+ void *q;
+
+ size = num_entries * LDC_PACKET_SIZE;
+ order = get_order(size);
+
+ q = (void *) __get_free_pages(GFP_KERNEL, order);
+ if (!q) {
+ printk(KERN_ERR PFX "Alloc of %s queue failed with "
+ "size=%lu order=%lu\n", name, size, order);
+ return -ENOMEM;
+ }
+
+ memset(q, 0, PAGE_SIZE << order);
+
+ *base = q;
+ *ra = __pa(q);
+
+ return 0;
+}
+
+static void free_queue(unsigned long num_entries, struct ldc_packet *q)
+{
+ unsigned long size, order;
+
+ if (!q)
+ return;
+
+ size = num_entries * LDC_PACKET_SIZE;
+ order = get_order(size);
+
+ free_pages((unsigned long)q, order);
+}
+
+/* XXX Make this configurable... XXX */
+#define LDC_IOTABLE_SIZE (8 * 1024)
+
+static int ldc_iommu_init(struct ldc_channel *lp)
+{
+ unsigned long sz, num_tsb_entries, tsbsize, order;
+ struct ldc_iommu *iommu = &lp->iommu;
+ struct ldc_mtable_entry *table;
+ unsigned long hv_err;
+ int err;
+
+ num_tsb_entries = LDC_IOTABLE_SIZE;
+ tsbsize = num_tsb_entries * sizeof(struct ldc_mtable_entry);
+
+ spin_lock_init(&iommu->lock);
+
+ sz = num_tsb_entries / 8;
+ sz = (sz + 7UL) & ~7UL;
+ iommu->arena.map = kzalloc(sz, GFP_KERNEL);
+ if (!iommu->arena.map) {
+ printk(KERN_ERR PFX "Alloc of arena map failed, sz=%lu\n", sz);
+ return -ENOMEM;
+ }
+
+ iommu->arena.limit = num_tsb_entries;
+
+ order = get_order(tsbsize);
+
+ table = (struct ldc_mtable_entry *)
+ __get_free_pages(GFP_KERNEL, order);
+ err = -ENOMEM;
+ if (!table) {
+ printk(KERN_ERR PFX "Alloc of MTE table failed, "
+ "size=%lu order=%lu\n", tsbsize, order);
+ goto out_free_map;
+ }
+
+ memset(table, 0, PAGE_SIZE << order);
+
+ iommu->page_table = table;
+
+ hv_err = sun4v_ldc_set_map_table(lp->id, __pa(table),
+ num_tsb_entries);
+ err = -EINVAL;
+ if (hv_err)
+ goto out_free_table;
+
+ return 0;
+
+out_free_table:
+ free_pages((unsigned long) table, order);
+ iommu->page_table = NULL;
+
+out_free_map:
+ kfree(iommu->arena.map);
+ iommu->arena.map = NULL;
+
+ return err;
+}
+
+static void ldc_iommu_release(struct ldc_channel *lp)
+{
+ struct ldc_iommu *iommu = &lp->iommu;
+ unsigned long num_tsb_entries, tsbsize, order;
+
+ (void) sun4v_ldc_set_map_table(lp->id, 0, 0);
+
+ num_tsb_entries = iommu->arena.limit;
+ tsbsize = num_tsb_entries * sizeof(struct ldc_mtable_entry);
+ order = get_order(tsbsize);
+
+ free_pages((unsigned long) iommu->page_table, order);
+ iommu->page_table = NULL;
+
+ kfree(iommu->arena.map);
+ iommu->arena.map = NULL;
+}
+
+struct ldc_channel *ldc_alloc(unsigned long id,
+ const struct ldc_channel_config *cfgp,
+ void *event_arg)
+{
+ struct ldc_channel *lp;
+ const struct ldc_mode_ops *mops;
+ unsigned long dummy1, dummy2, hv_err;
+ u8 mss, *mssbuf;
+ int err;
+
+ err = -ENODEV;
+ if (!ldom_domaining_enabled)
+ goto out_err;
+
+ err = -EINVAL;
+ if (!cfgp)
+ goto out_err;
+
+ switch (cfgp->mode) {
+ case LDC_MODE_RAW:
+ mops = &raw_ops;
+ mss = LDC_PACKET_SIZE;
+ break;
+
+ case LDC_MODE_UNRELIABLE:
+ mops = &nonraw_ops;
+ mss = LDC_PACKET_SIZE - 8;
+ break;
+
+ case LDC_MODE_STREAM:
+ mops = &stream_ops;
+ mss = LDC_PACKET_SIZE - 8 - 8;
+ break;
+
+ default:
+ goto out_err;
+ }
+
+ if (!cfgp->event || !event_arg || !cfgp->rx_irq || !cfgp->tx_irq)
+ goto out_err;
+
+ hv_err = sun4v_ldc_tx_qinfo(id, &dummy1, &dummy2);
+ err = -ENODEV;
+ if (hv_err == HV_ECHANNEL)
+ goto out_err;
+
+ err = -EEXIST;
+ if (__ldc_channel_exists(id))
+ goto out_err;
+
+ mssbuf = NULL;
+
+ lp = kzalloc(sizeof(*lp), GFP_KERNEL);
+ err = -ENOMEM;
+ if (!lp)
+ goto out_err;
+
+ spin_lock_init(&lp->lock);
+
+ lp->id = id;
+
+ err = ldc_iommu_init(lp);
+ if (err)
+ goto out_free_ldc;
+
+ lp->mops = mops;
+ lp->mss = mss;
+
+ lp->cfg = *cfgp;
+ if (!lp->cfg.mtu)
+ lp->cfg.mtu = LDC_DEFAULT_MTU;
+
+ if (lp->cfg.mode == LDC_MODE_STREAM) {
+ mssbuf = kzalloc(lp->cfg.mtu, GFP_KERNEL);
+ if (!mssbuf) {
+ err = -ENOMEM;
+ goto out_free_iommu;
+ }
+ lp->mssbuf = mssbuf;
+ }
+
+ lp->event_arg = event_arg;
+
+ /* XXX allow setting via ldc_channel_config to override defaults
+ * XXX or use some formula based upon mtu
+ */
+ lp->tx_num_entries = LDC_DEFAULT_NUM_ENTRIES;
+ lp->rx_num_entries = LDC_DEFAULT_NUM_ENTRIES;
+
+ err = alloc_queue("TX", lp->tx_num_entries,
+ &lp->tx_base, &lp->tx_ra);
+ if (err)
+ goto out_free_mssbuf;
+
+ err = alloc_queue("RX", lp->rx_num_entries,
+ &lp->rx_base, &lp->rx_ra);
+ if (err)
+ goto out_free_txq;
+
+ lp->flags |= LDC_FLAG_ALLOCED_QUEUES;
+
+ lp->hs_state = LDC_HS_CLOSED;
+ ldc_set_state(lp, LDC_STATE_INIT);
+
+ INIT_HLIST_NODE(&lp->list);
+ hlist_add_head(&lp->list, &ldc_channel_list);
+
+ INIT_HLIST_HEAD(&lp->mh_list);
+
+ return lp;
+
+out_free_txq:
+ free_queue(lp->tx_num_entries, lp->tx_base);
+
+out_free_mssbuf:
+ kfree(mssbuf);
+
+out_free_iommu:
+ ldc_iommu_release(lp);
+
+out_free_ldc:
+ kfree(lp);
+
+out_err:
+ return ERR_PTR(err);
+}
+EXPORT_SYMBOL(ldc_alloc);
+
+void ldc_free(struct ldc_channel *lp)
+{
+ if (lp->flags & LDC_FLAG_REGISTERED_IRQS) {
+ free_irq(lp->cfg.rx_irq, lp);
+ free_irq(lp->cfg.tx_irq, lp);
+ }
+
+ if (lp->flags & LDC_FLAG_REGISTERED_QUEUES) {
+ sun4v_ldc_tx_qconf(lp->id, 0, 0);
+ sun4v_ldc_rx_qconf(lp->id, 0, 0);
+ lp->flags &= ~LDC_FLAG_REGISTERED_QUEUES;
+ }
+ if (lp->flags & LDC_FLAG_ALLOCED_QUEUES) {
+ free_queue(lp->tx_num_entries, lp->tx_base);
+ free_queue(lp->rx_num_entries, lp->rx_base);
+ lp->flags &= ~LDC_FLAG_ALLOCED_QUEUES;
+ }
+
+ hlist_del(&lp->list);
+
+ kfree(lp->mssbuf);
+
+ ldc_iommu_release(lp);
+
+ kfree(lp);
+}
+EXPORT_SYMBOL(ldc_free);
+
+/* Bind the channel. This registers the LDC queues with
+ * the hypervisor and puts the channel into a pseudo-listening
+ * state. This does not initiate a handshake, ldc_connect() does
+ * that.
+ */
+int ldc_bind(struct ldc_channel *lp, const char *name)
+{
+ unsigned long hv_err, flags;
+ int err = -EINVAL;
+
+ if (!name ||
+ (lp->state != LDC_STATE_INIT))
+ return -EINVAL;
+
+ snprintf(lp->rx_irq_name, LDC_IRQ_NAME_MAX, "%s RX", name);
+ snprintf(lp->tx_irq_name, LDC_IRQ_NAME_MAX, "%s TX", name);
+
+ err = request_irq(lp->cfg.rx_irq, ldc_rx, 0,
+ lp->rx_irq_name, lp);
+ if (err)
+ return err;
+
+ err = request_irq(lp->cfg.tx_irq, ldc_tx, 0,
+ lp->tx_irq_name, lp);
+ if (err) {
+ free_irq(lp->cfg.rx_irq, lp);
+ return err;
+ }
+
+
+ spin_lock_irqsave(&lp->lock, flags);
+
+ enable_irq(lp->cfg.rx_irq);
+ enable_irq(lp->cfg.tx_irq);
+
+ lp->flags |= LDC_FLAG_REGISTERED_IRQS;
+
+ err = -ENODEV;
+ hv_err = sun4v_ldc_tx_qconf(lp->id, 0, 0);
+ if (hv_err)
+ goto out_free_irqs;
+
+ hv_err = sun4v_ldc_tx_qconf(lp->id, lp->tx_ra, lp->tx_num_entries);
+ if (hv_err)
+ goto out_free_irqs;
+
+ hv_err = sun4v_ldc_rx_qconf(lp->id, 0, 0);
+ if (hv_err)
+ goto out_unmap_tx;
+
+ hv_err = sun4v_ldc_rx_qconf(lp->id, lp->rx_ra, lp->rx_num_entries);
+ if (hv_err)
+ goto out_unmap_tx;
+
+ lp->flags |= LDC_FLAG_REGISTERED_QUEUES;
+
+ hv_err = sun4v_ldc_tx_get_state(lp->id,
+ &lp->tx_head,
+ &lp->tx_tail,
+ &lp->chan_state);
+ err = -EBUSY;
+ if (hv_err)
+ goto out_unmap_rx;
+
+ lp->tx_acked = lp->tx_head;
+
+ lp->hs_state = LDC_HS_OPEN;
+ ldc_set_state(lp, LDC_STATE_BOUND);
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ return 0;
+
+out_unmap_rx:
+ lp->flags &= ~LDC_FLAG_REGISTERED_QUEUES;
+ sun4v_ldc_rx_qconf(lp->id, 0, 0);
+
+out_unmap_tx:
+ sun4v_ldc_tx_qconf(lp->id, 0, 0);
+
+out_free_irqs:
+ lp->flags &= ~LDC_FLAG_REGISTERED_IRQS;
+ free_irq(lp->cfg.tx_irq, lp);
+ free_irq(lp->cfg.rx_irq, lp);
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ return err;
+}
+EXPORT_SYMBOL(ldc_bind);
+
+int ldc_connect(struct ldc_channel *lp)
+{
+ unsigned long flags;
+ int err;
+
+ if (lp->cfg.mode == LDC_MODE_RAW)
+ return -EINVAL;
+
+ spin_lock_irqsave(&lp->lock, flags);
+
+ if (!(lp->flags & LDC_FLAG_ALLOCED_QUEUES) ||
+ !(lp->flags & LDC_FLAG_REGISTERED_QUEUES) ||
+ lp->hs_state != LDC_HS_OPEN)
+ err = -EINVAL;
+ else
+ err = start_handshake(lp);
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ return err;
+}
+EXPORT_SYMBOL(ldc_connect);
+
+int ldc_disconnect(struct ldc_channel *lp)
+{
+ unsigned long hv_err, flags;
+ int err;
+
+ if (lp->cfg.mode == LDC_MODE_RAW)
+ return -EINVAL;
+
+ if (!(lp->flags & LDC_FLAG_ALLOCED_QUEUES) ||
+ !(lp->flags & LDC_FLAG_REGISTERED_QUEUES))
+ return -EINVAL;
+
+ spin_lock_irqsave(&lp->lock, flags);
+
+ err = -ENODEV;
+ hv_err = sun4v_ldc_tx_qconf(lp->id, 0, 0);
+ if (hv_err)
+ goto out_err;
+
+ hv_err = sun4v_ldc_tx_qconf(lp->id, lp->tx_ra, lp->tx_num_entries);
+ if (hv_err)
+ goto out_err;
+
+ hv_err = sun4v_ldc_rx_qconf(lp->id, 0, 0);
+ if (hv_err)
+ goto out_err;
+
+ hv_err = sun4v_ldc_rx_qconf(lp->id, lp->rx_ra, lp->rx_num_entries);
+ if (hv_err)
+ goto out_err;
+
+ ldc_set_state(lp, LDC_STATE_BOUND);
+ lp->hs_state = LDC_HS_OPEN;
+ lp->flags |= LDC_FLAG_RESET;
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ return 0;
+
+out_err:
+ sun4v_ldc_tx_qconf(lp->id, 0, 0);
+ sun4v_ldc_rx_qconf(lp->id, 0, 0);
+ free_irq(lp->cfg.tx_irq, lp);
+ free_irq(lp->cfg.rx_irq, lp);
+ lp->flags &= ~(LDC_FLAG_REGISTERED_IRQS |
+ LDC_FLAG_REGISTERED_QUEUES);
+ ldc_set_state(lp, LDC_STATE_INIT);
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ return err;
+}
+EXPORT_SYMBOL(ldc_disconnect);
+
+int ldc_state(struct ldc_channel *lp)
+{
+ return lp->state;
+}
+EXPORT_SYMBOL(ldc_state);
+
+static int write_raw(struct ldc_channel *lp, const void *buf, unsigned int size)
+{
+ struct ldc_packet *p;
+ unsigned long new_tail;
+ int err;
+
+ if (size > LDC_PACKET_SIZE)
+ return -EMSGSIZE;
+
+ p = data_get_tx_packet(lp, &new_tail);
+ if (!p)
+ return -EAGAIN;
+
+ memcpy(p, buf, size);
+
+ err = send_tx_packet(lp, p, new_tail);
+ if (!err)
+ err = size;
+
+ return err;
+}
+
+static int read_raw(struct ldc_channel *lp, void *buf, unsigned int size)
+{
+ struct ldc_packet *p;
+ unsigned long hv_err, new;
+ int err;
+
+ if (size < LDC_PACKET_SIZE)
+ return -EINVAL;
+
+ hv_err = sun4v_ldc_rx_get_state(lp->id,
+ &lp->rx_head,
+ &lp->rx_tail,
+ &lp->chan_state);
+ if (hv_err)
+ return ldc_abort(lp);
+
+ if (lp->chan_state == LDC_CHANNEL_DOWN ||
+ lp->chan_state == LDC_CHANNEL_RESETTING)
+ return -ECONNRESET;
+
+ if (lp->rx_head == lp->rx_tail)
+ return 0;
+
+ p = lp->rx_base + (lp->rx_head / LDC_PACKET_SIZE);
+ memcpy(buf, p, LDC_PACKET_SIZE);
+
+ new = rx_advance(lp, lp->rx_head);
+ lp->rx_head = new;
+
+ err = __set_rx_head(lp, new);
+ if (err < 0)
+ err = -ECONNRESET;
+ else
+ err = LDC_PACKET_SIZE;
+
+ return err;
+}
+
+static const struct ldc_mode_ops raw_ops = {
+ .write = write_raw,
+ .read = read_raw,
+};
+
+static int write_nonraw(struct ldc_channel *lp, const void *buf,
+ unsigned int size)
+{
+ unsigned long hv_err, tail;
+ unsigned int copied;
+ u32 seq;
+ int err;
+
+ hv_err = sun4v_ldc_tx_get_state(lp->id, &lp->tx_head, &lp->tx_tail,
+ &lp->chan_state);
+ if (unlikely(hv_err))
+ return -EBUSY;
+
+ if (unlikely(lp->chan_state != LDC_CHANNEL_UP))
+ return ldc_abort(lp);
+
+ if (!tx_has_space_for(lp, size))
+ return -EAGAIN;
+
+ seq = lp->snd_nxt;
+ copied = 0;
+ tail = lp->tx_tail;
+ while (copied < size) {
+ struct ldc_packet *p = lp->tx_base + (tail / LDC_PACKET_SIZE);
+ u8 *data = ((lp->cfg.mode == LDC_MODE_UNRELIABLE) ?
+ p->u.u_data :
+ p->u.r.r_data);
+ int data_len;
+
+ p->type = LDC_DATA;
+ p->stype = LDC_INFO;
+ p->ctrl = 0;
+
+ data_len = size - copied;
+ if (data_len > lp->mss)
+ data_len = lp->mss;
+
+ BUG_ON(data_len > LDC_LEN);
+
+ p->env = (data_len |
+ (copied == 0 ? LDC_START : 0) |
+ (data_len == size - copied ? LDC_STOP : 0));
+
+ p->seqid = ++seq;
+
+ ldcdbg(DATA, "SENT DATA [%02x:%02x:%02x:%02x:%08x]\n",
+ p->type,
+ p->stype,
+ p->ctrl,
+ p->env,
+ p->seqid);
+
+ memcpy(data, buf, data_len);
+ buf += data_len;
+ copied += data_len;
+
+ tail = tx_advance(lp, tail);
+ }
+
+ err = set_tx_tail(lp, tail);
+ if (!err) {
+ lp->snd_nxt = seq;
+ err = size;
+ }
+
+ return err;
+}
+
+static int rx_bad_seq(struct ldc_channel *lp, struct ldc_packet *p,
+ struct ldc_packet *first_frag)
+{
+ int err;
+
+ if (first_frag)
+ lp->rcv_nxt = first_frag->seqid - 1;
+
+ err = send_data_nack(lp, p);
+ if (err)
+ return err;
+
+ err = __set_rx_head(lp, lp->rx_tail);
+ if (err < 0)
+ return ldc_abort(lp);
+
+ return 0;
+}
+
+static int data_ack_nack(struct ldc_channel *lp, struct ldc_packet *p)
+{
+ if (p->stype & LDC_ACK) {
+ int err = process_data_ack(lp, p);
+ if (err)
+ return err;
+ }
+ if (p->stype & LDC_NACK)
+ return ldc_abort(lp);
+
+ return 0;
+}
+
+static int rx_data_wait(struct ldc_channel *lp, unsigned long cur_head)
+{
+ unsigned long dummy;
+ int limit = 1000;
+
+ ldcdbg(DATA, "DATA WAIT cur_head[%lx] rx_head[%lx] rx_tail[%lx]\n",
+ cur_head, lp->rx_head, lp->rx_tail);
+ while (limit-- > 0) {
+ unsigned long hv_err;
+
+ hv_err = sun4v_ldc_rx_get_state(lp->id,
+ &dummy,
+ &lp->rx_tail,
+ &lp->chan_state);
+ if (hv_err)
+ return ldc_abort(lp);
+
+ if (lp->chan_state == LDC_CHANNEL_DOWN ||
+ lp->chan_state == LDC_CHANNEL_RESETTING)
+ return -ECONNRESET;
+
+ if (cur_head != lp->rx_tail) {
+ ldcdbg(DATA, "DATA WAIT DONE "
+ "head[%lx] tail[%lx] chan_state[%lx]\n",
+ dummy, lp->rx_tail, lp->chan_state);
+ return 0;
+ }
+
+ udelay(1);
+ }
+ return -EAGAIN;
+}
+
+static int rx_set_head(struct ldc_channel *lp, unsigned long head)
+{
+ int err = __set_rx_head(lp, head);
+
+ if (err < 0)
+ return ldc_abort(lp);
+
+ lp->rx_head = head;
+ return 0;
+}
+
+static void send_data_ack(struct ldc_channel *lp)
+{
+ unsigned long new_tail;
+ struct ldc_packet *p;
+
+ p = data_get_tx_packet(lp, &new_tail);
+ if (likely(p)) {
+ int err;
+
+ memset(p, 0, sizeof(*p));
+ p->type = LDC_DATA;
+ p->stype = LDC_ACK;
+ p->ctrl = 0;
+ p->seqid = lp->snd_nxt + 1;
+ p->u.r.ackid = lp->rcv_nxt;
+
+ err = send_tx_packet(lp, p, new_tail);
+ if (!err)
+ lp->snd_nxt++;
+ }
+}
+
+static int read_nonraw(struct ldc_channel *lp, void *buf, unsigned int size)
+{
+ struct ldc_packet *first_frag;
+ unsigned long hv_err, new;
+ int err, copied;
+
+ hv_err = sun4v_ldc_rx_get_state(lp->id,
+ &lp->rx_head,
+ &lp->rx_tail,
+ &lp->chan_state);
+ if (hv_err)
+ return ldc_abort(lp);
+
+ if (lp->chan_state == LDC_CHANNEL_DOWN ||
+ lp->chan_state == LDC_CHANNEL_RESETTING)
+ return -ECONNRESET;
+
+ if (lp->rx_head == lp->rx_tail)
+ return 0;
+
+ first_frag = NULL;
+ copied = err = 0;
+ new = lp->rx_head;
+ while (1) {
+ struct ldc_packet *p;
+ int pkt_len;
+
+ BUG_ON(new == lp->rx_tail);
+ p = lp->rx_base + (new / LDC_PACKET_SIZE);
+
+ ldcdbg(RX, "RX read pkt[%02x:%02x:%02x:%02x:%08x:%08x] "
+ "rcv_nxt[%08x]\n",
+ p->type,
+ p->stype,
+ p->ctrl,
+ p->env,
+ p->seqid,
+ p->u.r.ackid,
+ lp->rcv_nxt);
+
+ if (unlikely(!rx_seq_ok(lp, p->seqid))) {
+ err = rx_bad_seq(lp, p, first_frag);
+ copied = 0;
+ break;
+ }
+
+ if (p->type & LDC_CTRL) {
+ err = process_control_frame(lp, p);
+ if (err < 0)
+ break;
+ err = 0;
+ }
+
+ lp->rcv_nxt = p->seqid;
+
+ if (!(p->type & LDC_DATA)) {
+ new = rx_advance(lp, new);
+ goto no_data;
+ }
+ if (p->stype & (LDC_ACK | LDC_NACK)) {
+ err = data_ack_nack(lp, p);
+ if (err)
+ break;
+ }
+ if (!(p->stype & LDC_INFO)) {
+ new = rx_advance(lp, new);
+ err = rx_set_head(lp, new);
+ if (err)
+ break;
+ goto no_data;
+ }
+
+ pkt_len = p->env & LDC_LEN;
+
+ /* Every initial packet starts with the START bit set.
+ *
+ * Singleton packets will have both START+STOP set.
+ *
+ * Fragments will have START set in the first frame, STOP
+ * set in the last frame, and neither bit set in middle
+ * frames of the packet.
+ *
+ * Therefore if we are at the beginning of a packet and
+ * we don't see START, or we are in the middle of a fragmented
+ * packet and do see START, we are unsynchronized and should
+ * flush the RX queue.
+ */
+ if ((first_frag == NULL && !(p->env & LDC_START)) ||
+ (first_frag != NULL && (p->env & LDC_START))) {
+ if (!first_frag)
+ new = rx_advance(lp, new);
+
+ err = rx_set_head(lp, new);
+ if (err)
+ break;
+
+ if (!first_frag)
+ goto no_data;
+ }
+ if (!first_frag)
+ first_frag = p;
+
+ if (pkt_len > size - copied) {
+ /* User didn't give us a big enough buffer,
+ * what to do? This is a pretty serious error.
+ *
+ * Since we haven't updated the RX ring head to
+ * consume any of the packets, signal the error
+ * to the user and just leave the RX ring alone.
+ *
+ * This seems the best behavior because this allows
+ * a user of the LDC layer to start with a small
+ * RX buffer for ldc_read() calls and use -EMSGSIZE
+ * as a cue to enlarge it's read buffer.
+ */
+ err = -EMSGSIZE;
+ break;
+ }
+
+ /* Ok, we are gonna eat this one. */
+ new = rx_advance(lp, new);
+
+ memcpy(buf,
+ (lp->cfg.mode == LDC_MODE_UNRELIABLE ?
+ p->u.u_data : p->u.r.r_data), pkt_len);
+ buf += pkt_len;
+ copied += pkt_len;
+
+ if (p->env & LDC_STOP)
+ break;
+
+no_data:
+ if (new == lp->rx_tail) {
+ err = rx_data_wait(lp, new);
+ if (err)
+ break;
+ }
+ }
+
+ if (!err)
+ err = rx_set_head(lp, new);
+
+ if (err && first_frag)
+ lp->rcv_nxt = first_frag->seqid - 1;
+
+ if (!err) {
+ err = copied;
+ if (err > 0 && lp->cfg.mode != LDC_MODE_UNRELIABLE)
+ send_data_ack(lp);
+ }
+
+ return err;
+}
+
+static const struct ldc_mode_ops nonraw_ops = {
+ .write = write_nonraw,
+ .read = read_nonraw,
+};
+
+static int write_stream(struct ldc_channel *lp, const void *buf,
+ unsigned int size)
+{
+ if (size > lp->cfg.mtu)
+ size = lp->cfg.mtu;
+ return write_nonraw(lp, buf, size);
+}
+
+static int read_stream(struct ldc_channel *lp, void *buf, unsigned int size)
+{
+ if (!lp->mssbuf_len) {
+ int err = read_nonraw(lp, lp->mssbuf, lp->cfg.mtu);
+ if (err < 0)
+ return err;
+
+ lp->mssbuf_len = err;
+ lp->mssbuf_off = 0;
+ }
+
+ if (size > lp->mssbuf_len)
+ size = lp->mssbuf_len;
+ memcpy(buf, lp->mssbuf + lp->mssbuf_off, size);
+
+ lp->mssbuf_off += size;
+ lp->mssbuf_len -= size;
+
+ return size;
+}
+
+static const struct ldc_mode_ops stream_ops = {
+ .write = write_stream,
+ .read = read_stream,
+};
+
+int ldc_write(struct ldc_channel *lp, const void *buf, unsigned int size)
+{
+ unsigned long flags;
+ int err;
+
+ if (!buf)
+ return -EINVAL;
+
+ if (!size)
+ return 0;
+
+ spin_lock_irqsave(&lp->lock, flags);
+
+ if (lp->hs_state != LDC_HS_COMPLETE)
+ err = -ENOTCONN;
+ else
+ err = lp->mops->write(lp, buf, size);
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ return err;
+}
+EXPORT_SYMBOL(ldc_write);
+
+int ldc_read(struct ldc_channel *lp, void *buf, unsigned int size)
+{
+ unsigned long flags;
+ int err;
+
+ if (!buf)
+ return -EINVAL;
+
+ if (!size)
+ return 0;
+
+ spin_lock_irqsave(&lp->lock, flags);
+
+ if (lp->hs_state != LDC_HS_COMPLETE)
+ err = -ENOTCONN;
+ else
+ err = lp->mops->read(lp, buf, size);
+
+ spin_unlock_irqrestore(&lp->lock, flags);
+
+ return err;
+}
+EXPORT_SYMBOL(ldc_read);
+
+static long arena_alloc(struct ldc_iommu *iommu, unsigned long npages)
+{
+ struct iommu_arena *arena = &iommu->arena;
+ unsigned long n, start, end, limit;
+ int pass;
+
+ limit = arena->limit;
+ start = arena->hint;
+ pass = 0;
+
+again:
+ n = bitmap_find_next_zero_area(arena->map, limit, start, npages, 0);
+ end = n + npages;
+ if (unlikely(end >= limit)) {
+ if (likely(pass < 1)) {
+ limit = start;
+ start = 0;
+ pass++;
+ goto again;
+ } else {
+ /* Scanned the whole thing, give up. */
+ return -1;
+ }
+ }
+ bitmap_set(arena->map, n, npages);
+
+ arena->hint = end;
+
+ return n;
+}
+
+#define COOKIE_PGSZ_CODE 0xf000000000000000ULL
+#define COOKIE_PGSZ_CODE_SHIFT 60ULL
+
+static u64 pagesize_code(void)
+{
+ switch (PAGE_SIZE) {
+ default:
+ case (8ULL * 1024ULL):
+ return 0;
+ case (64ULL * 1024ULL):
+ return 1;
+ case (512ULL * 1024ULL):
+ return 2;
+ case (4ULL * 1024ULL * 1024ULL):
+ return 3;
+ case (32ULL * 1024ULL * 1024ULL):
+ return 4;
+ case (256ULL * 1024ULL * 1024ULL):
+ return 5;
+ }
+}
+
+static u64 make_cookie(u64 index, u64 pgsz_code, u64 page_offset)
+{
+ return ((pgsz_code << COOKIE_PGSZ_CODE_SHIFT) |
+ (index << PAGE_SHIFT) |
+ page_offset);
+}
+
+static u64 cookie_to_index(u64 cookie, unsigned long *shift)
+{
+ u64 szcode = cookie >> COOKIE_PGSZ_CODE_SHIFT;
+
+ cookie &= ~COOKIE_PGSZ_CODE;
+
+ *shift = szcode * 3;
+
+ return (cookie >> (13ULL + (szcode * 3ULL)));
+}
+
+static struct ldc_mtable_entry *alloc_npages(struct ldc_iommu *iommu,
+ unsigned long npages)
+{
+ long entry;
+
+ entry = arena_alloc(iommu, npages);
+ if (unlikely(entry < 0))
+ return NULL;
+
+ return iommu->page_table + entry;
+}
+
+static u64 perm_to_mte(unsigned int map_perm)
+{
+ u64 mte_base;
+
+ mte_base = pagesize_code();
+
+ if (map_perm & LDC_MAP_SHADOW) {
+ if (map_perm & LDC_MAP_R)
+ mte_base |= LDC_MTE_COPY_R;
+ if (map_perm & LDC_MAP_W)
+ mte_base |= LDC_MTE_COPY_W;
+ }
+ if (map_perm & LDC_MAP_DIRECT) {
+ if (map_perm & LDC_MAP_R)
+ mte_base |= LDC_MTE_READ;
+ if (map_perm & LDC_MAP_W)
+ mte_base |= LDC_MTE_WRITE;
+ if (map_perm & LDC_MAP_X)
+ mte_base |= LDC_MTE_EXEC;
+ }
+ if (map_perm & LDC_MAP_IO) {
+ if (map_perm & LDC_MAP_R)
+ mte_base |= LDC_MTE_IOMMU_R;
+ if (map_perm & LDC_MAP_W)
+ mte_base |= LDC_MTE_IOMMU_W;
+ }
+
+ return mte_base;
+}
+
+static int pages_in_region(unsigned long base, long len)
+{
+ int count = 0;
+
+ do {
+ unsigned long new = (base + PAGE_SIZE) & PAGE_MASK;
+
+ len -= (new - base);
+ base = new;
+ count++;
+ } while (len > 0);
+
+ return count;
+}
+
+struct cookie_state {
+ struct ldc_mtable_entry *page_table;
+ struct ldc_trans_cookie *cookies;
+ u64 mte_base;
+ u64 prev_cookie;
+ u32 pte_idx;
+ u32 nc;
+};
+
+static void fill_cookies(struct cookie_state *sp, unsigned long pa,
+ unsigned long off, unsigned long len)
+{
+ do {
+ unsigned long tlen, new = pa + PAGE_SIZE;
+ u64 this_cookie;
+
+ sp->page_table[sp->pte_idx].mte = sp->mte_base | pa;
+
+ tlen = PAGE_SIZE;
+ if (off)
+ tlen = PAGE_SIZE - off;
+ if (tlen > len)
+ tlen = len;
+
+ this_cookie = make_cookie(sp->pte_idx,
+ pagesize_code(), off);
+
+ off = 0;
+
+ if (this_cookie == sp->prev_cookie) {
+ sp->cookies[sp->nc - 1].cookie_size += tlen;
+ } else {
+ sp->cookies[sp->nc].cookie_addr = this_cookie;
+ sp->cookies[sp->nc].cookie_size = tlen;
+ sp->nc++;
+ }
+ sp->prev_cookie = this_cookie + tlen;
+
+ sp->pte_idx++;
+
+ len -= tlen;
+ pa = new;
+ } while (len > 0);
+}
+
+static int sg_count_one(struct scatterlist *sg)
+{
+ unsigned long base = page_to_pfn(sg_page(sg)) << PAGE_SHIFT;
+ long len = sg->length;
+
+ if ((sg->offset | len) & (8UL - 1))
+ return -EFAULT;
+
+ return pages_in_region(base + sg->offset, len);
+}
+
+static int sg_count_pages(struct scatterlist *sg, int num_sg)
+{
+ int count;
+ int i;
+
+ count = 0;
+ for (i = 0; i < num_sg; i++) {
+ int err = sg_count_one(sg + i);
+ if (err < 0)
+ return err;
+ count += err;
+ }
+
+ return count;
+}
+
+int ldc_map_sg(struct ldc_channel *lp,
+ struct scatterlist *sg, int num_sg,
+ struct ldc_trans_cookie *cookies, int ncookies,
+ unsigned int map_perm)
+{
+ unsigned long i, npages, flags;
+ struct ldc_mtable_entry *base;
+ struct cookie_state state;
+ struct ldc_iommu *iommu;
+ int err;
+
+ if (map_perm & ~LDC_MAP_ALL)
+ return -EINVAL;
+
+ err = sg_count_pages(sg, num_sg);
+ if (err < 0)
+ return err;
+
+ npages = err;
+ if (err > ncookies)
+ return -EMSGSIZE;
+
+ iommu = &lp->iommu;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ base = alloc_npages(iommu, npages);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (!base)
+ return -ENOMEM;
+
+ state.page_table = iommu->page_table;
+ state.cookies = cookies;
+ state.mte_base = perm_to_mte(map_perm);
+ state.prev_cookie = ~(u64)0;
+ state.pte_idx = (base - iommu->page_table);
+ state.nc = 0;
+
+ for (i = 0; i < num_sg; i++)
+ fill_cookies(&state, page_to_pfn(sg_page(&sg[i])) << PAGE_SHIFT,
+ sg[i].offset, sg[i].length);
+
+ return state.nc;
+}
+EXPORT_SYMBOL(ldc_map_sg);
+
+int ldc_map_single(struct ldc_channel *lp,
+ void *buf, unsigned int len,
+ struct ldc_trans_cookie *cookies, int ncookies,
+ unsigned int map_perm)
+{
+ unsigned long npages, pa, flags;
+ struct ldc_mtable_entry *base;
+ struct cookie_state state;
+ struct ldc_iommu *iommu;
+
+ if ((map_perm & ~LDC_MAP_ALL) || (ncookies < 1))
+ return -EINVAL;
+
+ pa = __pa(buf);
+ if ((pa | len) & (8UL - 1))
+ return -EFAULT;
+
+ npages = pages_in_region(pa, len);
+
+ iommu = &lp->iommu;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ base = alloc_npages(iommu, npages);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (!base)
+ return -ENOMEM;
+
+ state.page_table = iommu->page_table;
+ state.cookies = cookies;
+ state.mte_base = perm_to_mte(map_perm);
+ state.prev_cookie = ~(u64)0;
+ state.pte_idx = (base - iommu->page_table);
+ state.nc = 0;
+ fill_cookies(&state, (pa & PAGE_MASK), (pa & ~PAGE_MASK), len);
+ BUG_ON(state.nc != 1);
+
+ return state.nc;
+}
+EXPORT_SYMBOL(ldc_map_single);
+
+static void free_npages(unsigned long id, struct ldc_iommu *iommu,
+ u64 cookie, u64 size)
+{
+ struct iommu_arena *arena = &iommu->arena;
+ unsigned long i, shift, index, npages;
+ struct ldc_mtable_entry *base;
+
+ npages = PAGE_ALIGN(((cookie & ~PAGE_MASK) + size)) >> PAGE_SHIFT;
+ index = cookie_to_index(cookie, &shift);
+ base = iommu->page_table + index;
+
+ BUG_ON(index > arena->limit ||
+ (index + npages) > arena->limit);
+
+ for (i = 0; i < npages; i++) {
+ if (base->cookie)
+ sun4v_ldc_revoke(id, cookie + (i << shift),
+ base->cookie);
+ base->mte = 0;
+ __clear_bit(index + i, arena->map);
+ }
+}
+
+void ldc_unmap(struct ldc_channel *lp, struct ldc_trans_cookie *cookies,
+ int ncookies)
+{
+ struct ldc_iommu *iommu = &lp->iommu;
+ unsigned long flags;
+ int i;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ for (i = 0; i < ncookies; i++) {
+ u64 addr = cookies[i].cookie_addr;
+ u64 size = cookies[i].cookie_size;
+
+ free_npages(lp->id, iommu, addr, size);
+ }
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+EXPORT_SYMBOL(ldc_unmap);
+
+int ldc_copy(struct ldc_channel *lp, int copy_dir,
+ void *buf, unsigned int len, unsigned long offset,
+ struct ldc_trans_cookie *cookies, int ncookies)
+{
+ unsigned int orig_len;
+ unsigned long ra;
+ int i;
+
+ if (copy_dir != LDC_COPY_IN && copy_dir != LDC_COPY_OUT) {
+ printk(KERN_ERR PFX "ldc_copy: ID[%lu] Bad copy_dir[%d]\n",
+ lp->id, copy_dir);
+ return -EINVAL;
+ }
+
+ ra = __pa(buf);
+ if ((ra | len | offset) & (8UL - 1)) {
+ printk(KERN_ERR PFX "ldc_copy: ID[%lu] Unaligned buffer "
+ "ra[%lx] len[%x] offset[%lx]\n",
+ lp->id, ra, len, offset);
+ return -EFAULT;
+ }
+
+ if (lp->hs_state != LDC_HS_COMPLETE ||
+ (lp->flags & LDC_FLAG_RESET)) {
+ printk(KERN_ERR PFX "ldc_copy: ID[%lu] Link down hs_state[%x] "
+ "flags[%x]\n", lp->id, lp->hs_state, lp->flags);
+ return -ECONNRESET;
+ }
+
+ orig_len = len;
+ for (i = 0; i < ncookies; i++) {
+ unsigned long cookie_raddr = cookies[i].cookie_addr;
+ unsigned long this_len = cookies[i].cookie_size;
+ unsigned long actual_len;
+
+ if (unlikely(offset)) {
+ unsigned long this_off = offset;
+
+ if (this_off > this_len)
+ this_off = this_len;
+
+ offset -= this_off;
+ this_len -= this_off;
+ if (!this_len)
+ continue;
+ cookie_raddr += this_off;
+ }
+
+ if (this_len > len)
+ this_len = len;
+
+ while (1) {
+ unsigned long hv_err;
+
+ hv_err = sun4v_ldc_copy(lp->id, copy_dir,
+ cookie_raddr, ra,
+ this_len, &actual_len);
+ if (unlikely(hv_err)) {
+ printk(KERN_ERR PFX "ldc_copy: ID[%lu] "
+ "HV error %lu\n",
+ lp->id, hv_err);
+ if (lp->hs_state != LDC_HS_COMPLETE ||
+ (lp->flags & LDC_FLAG_RESET))
+ return -ECONNRESET;
+ else
+ return -EFAULT;
+ }
+
+ cookie_raddr += actual_len;
+ ra += actual_len;
+ len -= actual_len;
+ if (actual_len == this_len)
+ break;
+
+ this_len -= actual_len;
+ }
+
+ if (!len)
+ break;
+ }
+
+ /* It is caller policy what to do about short copies.
+ * For example, a networking driver can declare the
+ * packet a runt and drop it.
+ */
+
+ return orig_len - len;
+}
+EXPORT_SYMBOL(ldc_copy);
+
+void *ldc_alloc_exp_dring(struct ldc_channel *lp, unsigned int len,
+ struct ldc_trans_cookie *cookies, int *ncookies,
+ unsigned int map_perm)
+{
+ void *buf;
+ int err;
+
+ if (len & (8UL - 1))
+ return ERR_PTR(-EINVAL);
+
+ buf = kzalloc(len, GFP_KERNEL);
+ if (!buf)
+ return ERR_PTR(-ENOMEM);
+
+ err = ldc_map_single(lp, buf, len, cookies, *ncookies, map_perm);
+ if (err < 0) {
+ kfree(buf);
+ return ERR_PTR(err);
+ }
+ *ncookies = err;
+
+ return buf;
+}
+EXPORT_SYMBOL(ldc_alloc_exp_dring);
+
+void ldc_free_exp_dring(struct ldc_channel *lp, void *buf, unsigned int len,
+ struct ldc_trans_cookie *cookies, int ncookies)
+{
+ ldc_unmap(lp, cookies, ncookies);
+ kfree(buf);
+}
+EXPORT_SYMBOL(ldc_free_exp_dring);
+
+static int __init ldc_init(void)
+{
+ unsigned long major, minor;
+ struct mdesc_handle *hp;
+ const u64 *v;
+ int err;
+ u64 mp;
+
+ hp = mdesc_grab();
+ if (!hp)
+ return -ENODEV;
+
+ mp = mdesc_node_by_name(hp, MDESC_NODE_NULL, "platform");
+ err = -ENODEV;
+ if (mp == MDESC_NODE_NULL)
+ goto out;
+
+ v = mdesc_get_property(hp, mp, "domaining-enabled", NULL);
+ if (!v)
+ goto out;
+
+ major = 1;
+ minor = 0;
+ if (sun4v_hvapi_register(HV_GRP_LDOM, major, &minor)) {
+ printk(KERN_INFO PFX "Could not register LDOM hvapi.\n");
+ goto out;
+ }
+
+ printk(KERN_INFO "%s", version);
+
+ if (!*v) {
+ printk(KERN_INFO PFX "Domaining disabled.\n");
+ goto out;
+ }
+ ldom_domaining_enabled = 1;
+ err = 0;
+
+out:
+ mdesc_release(hp);
+ return err;
+}
+
+core_initcall(ldc_init);
diff --git a/arch/sparc/kernel/led.c b/arch/sparc/kernel/led.c
index 313d1620ae8..3ae36f36e75 100644
--- a/arch/sparc/kernel/led.c
+++ b/arch/sparc/kernel/led.c
@@ -2,7 +2,12 @@
#include <linux/module.h>
#include <linux/init.h>
#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
#include <linux/string.h>
+#include <linux/jiffies.h>
+#include <linux/timer.h>
+#include <linux/uaccess.h>
#include <asm/auxio.h>
@@ -42,21 +47,22 @@ static void led_blink(unsigned long timeout)
add_timer(&led_blink_timer);
}
-static int led_read_proc(char *buf, char **start, off_t offset, int count,
- int *eof, void *data)
+static int led_proc_show(struct seq_file *m, void *v)
{
- int len = 0;
-
if (get_auxio() & AUXIO_LED)
- len = sprintf(buf, "on\n");
+ seq_puts(m, "on\n");
else
- len = sprintf(buf, "off\n");
+ seq_puts(m, "off\n");
+ return 0;
+}
- return len;
+static int led_proc_open(struct inode *inode, struct file *file)
+{
+ return single_open(file, led_proc_show, NULL);
}
-static int led_write_proc(struct file *file, const char __user *buffer,
- unsigned long count, void *data)
+static ssize_t led_proc_write(struct file *file, const char __user *buffer,
+ size_t count, loff_t *ppos)
{
char *buf = NULL;
@@ -100,6 +106,15 @@ static int led_write_proc(struct file *file, const char __user *buffer,
return count;
}
+static const struct file_operations led_proc_fops = {
+ .owner = THIS_MODULE,
+ .open = led_proc_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = single_release,
+ .write = led_proc_write,
+};
+
static struct proc_dir_entry *led;
#define LED_VERSION "0.1"
@@ -109,14 +124,10 @@ static int __init led_init(void)
init_timer(&led_blink_timer);
led_blink_timer.function = led_blink;
- led = create_proc_entry("led", 0, NULL);
+ led = proc_create("led", 0, NULL, &led_proc_fops);
if (!led)
return -ENOMEM;
- led->read_proc = led_read_proc; /* reader function */
- led->write_proc = led_write_proc; /* writer function */
- led->owner = THIS_MODULE;
-
printk(KERN_INFO
"led: version %s, Lars Kotthoff <metalhead@metalhead.ws>\n",
LED_VERSION);
diff --git a/arch/sparc/kernel/leon_kernel.c b/arch/sparc/kernel/leon_kernel.c
new file mode 100644
index 00000000000..683c4af999d
--- /dev/null
+++ b/arch/sparc/kernel/leon_kernel.c
@@ -0,0 +1,499 @@
+/*
+ * Copyright (C) 2009 Daniel Hellstrom (daniel@gaisler.com) Aeroflex Gaisler AB
+ * Copyright (C) 2009 Konrad Eisele (konrad@gaisler.com) Aeroflex Gaisler AB
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/mutex.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/interrupt.h>
+#include <linux/of_device.h>
+#include <linux/clocksource.h>
+#include <linux/clockchips.h>
+
+#include <asm/oplib.h>
+#include <asm/timer.h>
+#include <asm/prom.h>
+#include <asm/leon.h>
+#include <asm/leon_amba.h>
+#include <asm/traps.h>
+#include <asm/cacheflush.h>
+#include <asm/smp.h>
+#include <asm/setup.h>
+
+#include "kernel.h"
+#include "prom.h"
+#include "irq.h"
+
+struct leon3_irqctrl_regs_map *leon3_irqctrl_regs; /* interrupt controller base address */
+struct leon3_gptimer_regs_map *leon3_gptimer_regs; /* timer controller base address */
+
+int leondebug_irq_disable;
+int leon_debug_irqout;
+static volatile u32 dummy_master_l10_counter;
+unsigned long amba_system_id;
+static DEFINE_SPINLOCK(leon_irq_lock);
+
+static unsigned long leon3_gptimer_idx; /* Timer Index (0..6) within Timer Core */
+unsigned long leon3_gptimer_irq; /* interrupt controller irq number */
+unsigned int sparc_leon_eirq;
+#define LEON_IMASK(cpu) (&leon3_irqctrl_regs->mask[cpu])
+#define LEON_IACK (&leon3_irqctrl_regs->iclear)
+#define LEON_DO_ACK_HW 1
+
+/* Return the last ACKed IRQ by the Extended IRQ controller. It has already
+ * been (automatically) ACKed when the CPU takes the trap.
+ */
+static inline unsigned int leon_eirq_get(int cpu)
+{
+ return LEON3_BYPASS_LOAD_PA(&leon3_irqctrl_regs->intid[cpu]) & 0x1f;
+}
+
+/* Handle one or multiple IRQs from the extended interrupt controller */
+static void leon_handle_ext_irq(unsigned int irq, struct irq_desc *desc)
+{
+ unsigned int eirq;
+ struct irq_bucket *p;
+ int cpu = sparc_leon3_cpuid();
+
+ eirq = leon_eirq_get(cpu);
+ p = irq_map[eirq];
+ if ((eirq & 0x10) && p && p->irq) /* bit4 tells if IRQ happened */
+ generic_handle_irq(p->irq);
+}
+
+/* The extended IRQ controller has been found, this function registers it */
+static void leon_eirq_setup(unsigned int eirq)
+{
+ unsigned long mask, oldmask;
+ unsigned int veirq;
+
+ if (eirq < 1 || eirq > 0xf) {
+ printk(KERN_ERR "LEON EXT IRQ NUMBER BAD: %d\n", eirq);
+ return;
+ }
+
+ veirq = leon_build_device_irq(eirq, leon_handle_ext_irq, "extirq", 0);
+
+ /*
+ * Unmask the Extended IRQ, the IRQs routed through the Ext-IRQ
+ * controller have a mask-bit of their own, so this is safe.
+ */
+ irq_link(veirq);
+ mask = 1 << eirq;
+ oldmask = LEON3_BYPASS_LOAD_PA(LEON_IMASK(boot_cpu_id));
+ LEON3_BYPASS_STORE_PA(LEON_IMASK(boot_cpu_id), (oldmask | mask));
+ sparc_leon_eirq = eirq;
+}
+
+unsigned long leon_get_irqmask(unsigned int irq)
+{
+ unsigned long mask;
+
+ if (!irq || ((irq > 0xf) && !sparc_leon_eirq)
+ || ((irq > 0x1f) && sparc_leon_eirq)) {
+ printk(KERN_ERR
+ "leon_get_irqmask: false irq number: %d\n", irq);
+ mask = 0;
+ } else {
+ mask = LEON_HARD_INT(irq);
+ }
+ return mask;
+}
+
+#ifdef CONFIG_SMP
+static int irq_choose_cpu(const struct cpumask *affinity)
+{
+ cpumask_t mask;
+
+ cpumask_and(&mask, cpu_online_mask, affinity);
+ if (cpumask_equal(&mask, cpu_online_mask) || cpumask_empty(&mask))
+ return boot_cpu_id;
+ else
+ return cpumask_first(&mask);
+}
+#else
+#define irq_choose_cpu(affinity) boot_cpu_id
+#endif
+
+static int leon_set_affinity(struct irq_data *data, const struct cpumask *dest,
+ bool force)
+{
+ unsigned long mask, oldmask, flags;
+ int oldcpu, newcpu;
+
+ mask = (unsigned long)data->chip_data;
+ oldcpu = irq_choose_cpu(data->affinity);
+ newcpu = irq_choose_cpu(dest);
+
+ if (oldcpu == newcpu)
+ goto out;
+
+ /* unmask on old CPU first before enabling on the selected CPU */
+ spin_lock_irqsave(&leon_irq_lock, flags);
+ oldmask = LEON3_BYPASS_LOAD_PA(LEON_IMASK(oldcpu));
+ LEON3_BYPASS_STORE_PA(LEON_IMASK(oldcpu), (oldmask & ~mask));
+ oldmask = LEON3_BYPASS_LOAD_PA(LEON_IMASK(newcpu));
+ LEON3_BYPASS_STORE_PA(LEON_IMASK(newcpu), (oldmask | mask));
+ spin_unlock_irqrestore(&leon_irq_lock, flags);
+out:
+ return IRQ_SET_MASK_OK;
+}
+
+static void leon_unmask_irq(struct irq_data *data)
+{
+ unsigned long mask, oldmask, flags;
+ int cpu;
+
+ mask = (unsigned long)data->chip_data;
+ cpu = irq_choose_cpu(data->affinity);
+ spin_lock_irqsave(&leon_irq_lock, flags);
+ oldmask = LEON3_BYPASS_LOAD_PA(LEON_IMASK(cpu));
+ LEON3_BYPASS_STORE_PA(LEON_IMASK(cpu), (oldmask | mask));
+ spin_unlock_irqrestore(&leon_irq_lock, flags);
+}
+
+static void leon_mask_irq(struct irq_data *data)
+{
+ unsigned long mask, oldmask, flags;
+ int cpu;
+
+ mask = (unsigned long)data->chip_data;
+ cpu = irq_choose_cpu(data->affinity);
+ spin_lock_irqsave(&leon_irq_lock, flags);
+ oldmask = LEON3_BYPASS_LOAD_PA(LEON_IMASK(cpu));
+ LEON3_BYPASS_STORE_PA(LEON_IMASK(cpu), (oldmask & ~mask));
+ spin_unlock_irqrestore(&leon_irq_lock, flags);
+}
+
+static unsigned int leon_startup_irq(struct irq_data *data)
+{
+ irq_link(data->irq);
+ leon_unmask_irq(data);
+ return 0;
+}
+
+static void leon_shutdown_irq(struct irq_data *data)
+{
+ leon_mask_irq(data);
+ irq_unlink(data->irq);
+}
+
+/* Used by external level sensitive IRQ handlers on the LEON: ACK IRQ ctrl */
+static void leon_eoi_irq(struct irq_data *data)
+{
+ unsigned long mask = (unsigned long)data->chip_data;
+
+ if (mask & LEON_DO_ACK_HW)
+ LEON3_BYPASS_STORE_PA(LEON_IACK, mask & ~LEON_DO_ACK_HW);
+}
+
+static struct irq_chip leon_irq = {
+ .name = "leon",
+ .irq_startup = leon_startup_irq,
+ .irq_shutdown = leon_shutdown_irq,
+ .irq_mask = leon_mask_irq,
+ .irq_unmask = leon_unmask_irq,
+ .irq_eoi = leon_eoi_irq,
+ .irq_set_affinity = leon_set_affinity,
+};
+
+/*
+ * Build a LEON IRQ for the edge triggered LEON IRQ controller:
+ * Edge (normal) IRQ - handle_simple_irq, ack=DONT-CARE, never ack
+ * Level IRQ (PCI|Level-GPIO) - handle_fasteoi_irq, ack=1, ack after ISR
+ * Per-CPU Edge - handle_percpu_irq, ack=0
+ */
+unsigned int leon_build_device_irq(unsigned int real_irq,
+ irq_flow_handler_t flow_handler,
+ const char *name, int do_ack)
+{
+ unsigned int irq;
+ unsigned long mask;
+ struct irq_desc *desc;
+
+ irq = 0;
+ mask = leon_get_irqmask(real_irq);
+ if (mask == 0)
+ goto out;
+
+ irq = irq_alloc(real_irq, real_irq);
+ if (irq == 0)
+ goto out;
+
+ if (do_ack)
+ mask |= LEON_DO_ACK_HW;
+
+ desc = irq_to_desc(irq);
+ if (!desc || !desc->handle_irq || desc->handle_irq == handle_bad_irq) {
+ irq_set_chip_and_handler_name(irq, &leon_irq,
+ flow_handler, name);
+ irq_set_chip_data(irq, (void *)mask);
+ }
+
+out:
+ return irq;
+}
+
+static unsigned int _leon_build_device_irq(struct platform_device *op,
+ unsigned int real_irq)
+{
+ return leon_build_device_irq(real_irq, handle_simple_irq, "edge", 0);
+}
+
+void leon_update_virq_handling(unsigned int virq,
+ irq_flow_handler_t flow_handler,
+ const char *name, int do_ack)
+{
+ unsigned long mask = (unsigned long)irq_get_chip_data(virq);
+
+ mask &= ~LEON_DO_ACK_HW;
+ if (do_ack)
+ mask |= LEON_DO_ACK_HW;
+
+ irq_set_chip_and_handler_name(virq, &leon_irq,
+ flow_handler, name);
+ irq_set_chip_data(virq, (void *)mask);
+}
+
+static u32 leon_cycles_offset(void)
+{
+ u32 rld, val, off;
+ rld = LEON3_BYPASS_LOAD_PA(&leon3_gptimer_regs->e[leon3_gptimer_idx].rld);
+ val = LEON3_BYPASS_LOAD_PA(&leon3_gptimer_regs->e[leon3_gptimer_idx].val);
+ off = rld - val;
+ return rld - val;
+}
+
+#ifdef CONFIG_SMP
+
+/* smp clockevent irq */
+static irqreturn_t leon_percpu_timer_ce_interrupt(int irq, void *unused)
+{
+ struct clock_event_device *ce;
+ int cpu = smp_processor_id();
+
+ leon_clear_profile_irq(cpu);
+
+ if (cpu == boot_cpu_id)
+ timer_interrupt(irq, NULL);
+
+ ce = &per_cpu(sparc32_clockevent, cpu);
+
+ irq_enter();
+ if (ce->event_handler)
+ ce->event_handler(ce);
+ irq_exit();
+
+ return IRQ_HANDLED;
+}
+
+#endif /* CONFIG_SMP */
+
+void __init leon_init_timers(void)
+{
+ int irq, eirq;
+ struct device_node *rootnp, *np, *nnp;
+ struct property *pp;
+ int len;
+ int icsel;
+ int ampopts;
+ int err;
+ u32 config;
+
+ sparc_config.get_cycles_offset = leon_cycles_offset;
+ sparc_config.cs_period = 1000000 / HZ;
+ sparc_config.features |= FEAT_L10_CLOCKSOURCE;
+
+#ifndef CONFIG_SMP
+ sparc_config.features |= FEAT_L10_CLOCKEVENT;
+#endif
+
+ leondebug_irq_disable = 0;
+ leon_debug_irqout = 0;
+ master_l10_counter = (u32 __iomem *)&dummy_master_l10_counter;
+ dummy_master_l10_counter = 0;
+
+ rootnp = of_find_node_by_path("/ambapp0");
+ if (!rootnp)
+ goto bad;
+
+ /* Find System ID: GRLIB build ID and optional CHIP ID */
+ pp = of_find_property(rootnp, "systemid", &len);
+ if (pp)
+ amba_system_id = *(unsigned long *)pp->value;
+
+ /* Find IRQMP IRQ Controller Registers base adr otherwise bail out */
+ np = of_find_node_by_name(rootnp, "GAISLER_IRQMP");
+ if (!np) {
+ np = of_find_node_by_name(rootnp, "01_00d");
+ if (!np)
+ goto bad;
+ }
+ pp = of_find_property(np, "reg", &len);
+ if (!pp)
+ goto bad;
+ leon3_irqctrl_regs = *(struct leon3_irqctrl_regs_map **)pp->value;
+
+ /* Find GPTIMER Timer Registers base address otherwise bail out. */
+ nnp = rootnp;
+ do {
+ np = of_find_node_by_name(nnp, "GAISLER_GPTIMER");
+ if (!np) {
+ np = of_find_node_by_name(nnp, "01_011");
+ if (!np)
+ goto bad;
+ }
+
+ ampopts = 0;
+ pp = of_find_property(np, "ampopts", &len);
+ if (pp) {
+ ampopts = *(int *)pp->value;
+ if (ampopts == 0) {
+ /* Skip this instance, resource already
+ * allocated by other OS */
+ nnp = np;
+ continue;
+ }
+ }
+
+ /* Select Timer-Instance on Timer Core. Default is zero */
+ leon3_gptimer_idx = ampopts & 0x7;
+
+ pp = of_find_property(np, "reg", &len);
+ if (pp)
+ leon3_gptimer_regs = *(struct leon3_gptimer_regs_map **)
+ pp->value;
+ pp = of_find_property(np, "interrupts", &len);
+ if (pp)
+ leon3_gptimer_irq = *(unsigned int *)pp->value;
+ } while (0);
+
+ if (!(leon3_gptimer_regs && leon3_irqctrl_regs && leon3_gptimer_irq))
+ goto bad;
+
+ LEON3_BYPASS_STORE_PA(&leon3_gptimer_regs->e[leon3_gptimer_idx].val, 0);
+ LEON3_BYPASS_STORE_PA(&leon3_gptimer_regs->e[leon3_gptimer_idx].rld,
+ (((1000000 / HZ) - 1)));
+ LEON3_BYPASS_STORE_PA(
+ &leon3_gptimer_regs->e[leon3_gptimer_idx].ctrl, 0);
+
+ /*
+ * The IRQ controller may (if implemented) consist of multiple
+ * IRQ controllers, each mapped on a 4Kb boundary.
+ * Each CPU may be routed to different IRQCTRLs, however
+ * we assume that all CPUs (in SMP system) is routed to the
+ * same IRQ Controller, and for non-SMP only one IRQCTRL is
+ * accessed anyway.
+ * In AMP systems, Linux must run on CPU0 for the time being.
+ */
+ icsel = LEON3_BYPASS_LOAD_PA(&leon3_irqctrl_regs->icsel[boot_cpu_id/8]);
+ icsel = (icsel >> ((7 - (boot_cpu_id&0x7)) * 4)) & 0xf;
+ leon3_irqctrl_regs += icsel;
+
+ /* Mask all IRQs on boot-cpu IRQ controller */
+ LEON3_BYPASS_STORE_PA(&leon3_irqctrl_regs->mask[boot_cpu_id], 0);
+
+ /* Probe extended IRQ controller */
+ eirq = (LEON3_BYPASS_LOAD_PA(&leon3_irqctrl_regs->mpstatus)
+ >> 16) & 0xf;
+ if (eirq != 0)
+ leon_eirq_setup(eirq);
+
+#ifdef CONFIG_SMP
+ {
+ unsigned long flags;
+
+ /*
+ * In SMP, sun4m adds a IPI handler to IRQ trap handler that
+ * LEON never must take, sun4d and LEON overwrites the branch
+ * with a NOP.
+ */
+ local_irq_save(flags);
+ patchme_maybe_smp_msg[0] = 0x01000000; /* NOP out the branch */
+ local_ops->cache_all();
+ local_irq_restore(flags);
+ }
+#endif
+
+ config = LEON3_BYPASS_LOAD_PA(&leon3_gptimer_regs->config);
+ if (config & (1 << LEON3_GPTIMER_SEPIRQ))
+ leon3_gptimer_irq += leon3_gptimer_idx;
+ else if ((config & LEON3_GPTIMER_TIMERS) > 1)
+ pr_warn("GPTIMER uses shared irqs, using other timers of the same core will fail.\n");
+
+#ifdef CONFIG_SMP
+ /* Install per-cpu IRQ handler for broadcasted ticker */
+ irq = leon_build_device_irq(leon3_gptimer_irq, handle_percpu_irq,
+ "per-cpu", 0);
+ err = request_irq(irq, leon_percpu_timer_ce_interrupt,
+ IRQF_PERCPU | IRQF_TIMER, "timer", NULL);
+#else
+ irq = _leon_build_device_irq(NULL, leon3_gptimer_irq);
+ err = request_irq(irq, timer_interrupt, IRQF_TIMER, "timer", NULL);
+#endif
+ if (err) {
+ pr_err("Unable to attach timer IRQ%d\n", irq);
+ prom_halt();
+ }
+ LEON3_BYPASS_STORE_PA(&leon3_gptimer_regs->e[leon3_gptimer_idx].ctrl,
+ LEON3_GPTIMER_EN |
+ LEON3_GPTIMER_RL |
+ LEON3_GPTIMER_LD |
+ LEON3_GPTIMER_IRQEN);
+ return;
+bad:
+ printk(KERN_ERR "No Timer/irqctrl found\n");
+ BUG();
+ return;
+}
+
+static void leon_clear_clock_irq(void)
+{
+}
+
+static void leon_load_profile_irq(int cpu, unsigned int limit)
+{
+}
+
+void __init leon_trans_init(struct device_node *dp)
+{
+ if (strcmp(dp->type, "cpu") == 0 && strcmp(dp->name, "<NULL>") == 0) {
+ struct property *p;
+ p = of_find_property(dp, "mid", (void *)0);
+ if (p) {
+ int mid;
+ dp->name = prom_early_alloc(5 + 1);
+ memcpy(&mid, p->value, p->length);
+ sprintf((char *)dp->name, "cpu%.2d", mid);
+ }
+ }
+}
+
+#ifdef CONFIG_SMP
+void leon_clear_profile_irq(int cpu)
+{
+}
+
+void leon_enable_irq_cpu(unsigned int irq_nr, unsigned int cpu)
+{
+ unsigned long mask, flags, *addr;
+ mask = leon_get_irqmask(irq_nr);
+ spin_lock_irqsave(&leon_irq_lock, flags);
+ addr = (unsigned long *)LEON_IMASK(cpu);
+ LEON3_BYPASS_STORE_PA(addr, (LEON3_BYPASS_LOAD_PA(addr) | mask));
+ spin_unlock_irqrestore(&leon_irq_lock, flags);
+}
+
+#endif
+
+void __init leon_init_IRQ(void)
+{
+ sparc_config.init_timers = leon_init_timers;
+ sparc_config.build_device_irq = _leon_build_device_irq;
+ sparc_config.clock_rate = 1000000;
+ sparc_config.clear_clock_irq = leon_clear_clock_irq;
+ sparc_config.load_profile_irq = leon_load_profile_irq;
+}
diff --git a/arch/sparc/kernel/leon_pci.c b/arch/sparc/kernel/leon_pci.c
new file mode 100644
index 00000000000..899b7203a4e
--- /dev/null
+++ b/arch/sparc/kernel/leon_pci.c
@@ -0,0 +1,100 @@
+/*
+ * leon_pci.c: LEON Host PCI support
+ *
+ * Copyright (C) 2011 Aeroflex Gaisler AB, Daniel Hellstrom
+ *
+ * Code is partially derived from pcic.c
+ */
+
+#include <linux/of_device.h>
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/export.h>
+#include <asm/leon.h>
+#include <asm/leon_pci.h>
+
+/* The LEON architecture does not rely on a BIOS or bootloader to setup
+ * PCI for us. The Linux generic routines are used to setup resources,
+ * reset values of configuration-space register settings are preserved.
+ *
+ * PCI Memory and Prefetchable Memory is direct-mapped. However I/O Space is
+ * accessed through a Window which is translated to low 64KB in PCI space, the
+ * first 4KB is not used so 60KB is available.
+ */
+void leon_pci_init(struct platform_device *ofdev, struct leon_pci_info *info)
+{
+ LIST_HEAD(resources);
+ struct pci_bus *root_bus;
+
+ pci_add_resource_offset(&resources, &info->io_space,
+ info->io_space.start - 0x1000);
+ pci_add_resource(&resources, &info->mem_space);
+ info->busn.flags = IORESOURCE_BUS;
+ pci_add_resource(&resources, &info->busn);
+
+ root_bus = pci_scan_root_bus(&ofdev->dev, 0, info->ops, info,
+ &resources);
+ if (root_bus) {
+ /* Setup IRQs of all devices using custom routines */
+ pci_fixup_irqs(pci_common_swizzle, info->map_irq);
+
+ /* Assign devices with resources */
+ pci_assign_unassigned_resources();
+ } else {
+ pci_free_resource_list(&resources);
+ }
+}
+
+void pcibios_fixup_bus(struct pci_bus *pbus)
+{
+ struct pci_dev *dev;
+ int i, has_io, has_mem;
+ u16 cmd;
+
+ list_for_each_entry(dev, &pbus->devices, bus_list) {
+ /*
+ * We can not rely on that the bootloader has enabled I/O
+ * or memory access to PCI devices. Instead we enable it here
+ * if the device has BARs of respective type.
+ */
+ has_io = has_mem = 0;
+ for (i = 0; i < PCI_ROM_RESOURCE; i++) {
+ unsigned long f = dev->resource[i].flags;
+ if (f & IORESOURCE_IO)
+ has_io = 1;
+ else if (f & IORESOURCE_MEM)
+ has_mem = 1;
+ }
+ /* ROM BARs are mapped into 32-bit memory space */
+ if (dev->resource[PCI_ROM_RESOURCE].end != 0) {
+ dev->resource[PCI_ROM_RESOURCE].flags |=
+ IORESOURCE_ROM_ENABLE;
+ has_mem = 1;
+ }
+ pci_bus_read_config_word(pbus, dev->devfn, PCI_COMMAND, &cmd);
+ if (has_io && !(cmd & PCI_COMMAND_IO)) {
+#ifdef CONFIG_PCI_DEBUG
+ printk(KERN_INFO "LEONPCI: Enabling I/O for dev %s\n",
+ pci_name(dev));
+#endif
+ cmd |= PCI_COMMAND_IO;
+ pci_bus_write_config_word(pbus, dev->devfn, PCI_COMMAND,
+ cmd);
+ }
+ if (has_mem && !(cmd & PCI_COMMAND_MEMORY)) {
+#ifdef CONFIG_PCI_DEBUG
+ printk(KERN_INFO "LEONPCI: Enabling MEMORY for dev"
+ "%s\n", pci_name(dev));
+#endif
+ cmd |= PCI_COMMAND_MEMORY;
+ pci_bus_write_config_word(pbus, dev->devfn, PCI_COMMAND,
+ cmd);
+ }
+ }
+}
+
+resource_size_t pcibios_align_resource(void *data, const struct resource *res,
+ resource_size_t size, resource_size_t align)
+{
+ return res->start;
+}
diff --git a/arch/sparc/kernel/leon_pci_grpci1.c b/arch/sparc/kernel/leon_pci_grpci1.c
new file mode 100644
index 00000000000..c8bf26edfa7
--- /dev/null
+++ b/arch/sparc/kernel/leon_pci_grpci1.c
@@ -0,0 +1,722 @@
+/*
+ * leon_pci_grpci1.c: GRPCI1 Host PCI driver
+ *
+ * Copyright (C) 2013 Aeroflex Gaisler AB
+ *
+ * This GRPCI1 driver does not support PCI interrupts taken from
+ * GPIO pins. Interrupt generation at PCI parity and system error
+ * detection is by default turned off since some GRPCI1 cores does
+ * not support detection. It can be turned on from the bootloader
+ * using the all_pci_errors property.
+ *
+ * Contributors: Daniel Hellstrom <daniel@gaisler.com>
+ */
+
+#include <linux/of_device.h>
+#include <linux/export.h>
+#include <linux/kernel.h>
+#include <linux/of_irq.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+
+#include <asm/leon_pci.h>
+#include <asm/sections.h>
+#include <asm/vaddrs.h>
+#include <asm/leon.h>
+#include <asm/io.h>
+
+#include "irq.h"
+
+/* Enable/Disable Debugging Configuration Space Access */
+#undef GRPCI1_DEBUG_CFGACCESS
+
+/*
+ * GRPCI1 APB Register MAP
+ */
+struct grpci1_regs {
+ unsigned int cfg_stat; /* 0x00 Configuration / Status */
+ unsigned int bar0; /* 0x04 BAR0 (RO) */
+ unsigned int page0; /* 0x08 PAGE0 (RO) */
+ unsigned int bar1; /* 0x0C BAR1 (RO) */
+ unsigned int page1; /* 0x10 PAGE1 */
+ unsigned int iomap; /* 0x14 IO Map */
+ unsigned int stat_cmd; /* 0x18 PCI Status & Command (RO) */
+ unsigned int irq; /* 0x1C Interrupt register */
+};
+
+#define REGLOAD(a) (be32_to_cpu(__raw_readl(&(a))))
+#define REGSTORE(a, v) (__raw_writel(cpu_to_be32(v), &(a)))
+
+#define PAGE0_BTEN_BIT 0
+#define PAGE0_BTEN (1 << PAGE0_BTEN_BIT)
+
+#define CFGSTAT_HOST_BIT 13
+#define CFGSTAT_CTO_BIT 8
+#define CFGSTAT_HOST (1 << CFGSTAT_HOST_BIT)
+#define CFGSTAT_CTO (1 << CFGSTAT_CTO_BIT)
+
+#define IRQ_DPE (1 << 9)
+#define IRQ_SSE (1 << 8)
+#define IRQ_RMA (1 << 7)
+#define IRQ_RTA (1 << 6)
+#define IRQ_STA (1 << 5)
+#define IRQ_DPED (1 << 4)
+#define IRQ_INTD (1 << 3)
+#define IRQ_INTC (1 << 2)
+#define IRQ_INTB (1 << 1)
+#define IRQ_INTA (1 << 0)
+#define IRQ_DEF_ERRORS (IRQ_RMA | IRQ_RTA | IRQ_STA)
+#define IRQ_ALL_ERRORS (IRQ_DPED | IRQ_DEF_ERRORS | IRQ_SSE | IRQ_DPE)
+#define IRQ_INTX (IRQ_INTA | IRQ_INTB | IRQ_INTC | IRQ_INTD)
+#define IRQ_MASK_BIT 16
+
+#define DEF_PCI_ERRORS (PCI_STATUS_SIG_TARGET_ABORT | \
+ PCI_STATUS_REC_TARGET_ABORT | \
+ PCI_STATUS_REC_MASTER_ABORT)
+#define ALL_PCI_ERRORS (PCI_STATUS_PARITY | PCI_STATUS_DETECTED_PARITY | \
+ PCI_STATUS_SIG_SYSTEM_ERROR | DEF_PCI_ERRORS)
+
+#define TGT 256
+
+struct grpci1_priv {
+ struct leon_pci_info info; /* must be on top of this structure */
+ struct grpci1_regs __iomem *regs; /* GRPCI register map */
+ struct device *dev;
+ int pci_err_mask; /* STATUS register error mask */
+ int irq; /* LEON irqctrl GRPCI IRQ */
+ unsigned char irq_map[4]; /* GRPCI nexus PCI INTX# IRQs */
+ unsigned int irq_err; /* GRPCI nexus Virt Error IRQ */
+
+ /* AHB PCI Windows */
+ unsigned long pci_area; /* MEMORY */
+ unsigned long pci_area_end;
+ unsigned long pci_io; /* I/O */
+ unsigned long pci_conf; /* CONFIGURATION */
+ unsigned long pci_conf_end;
+ unsigned long pci_io_va;
+};
+
+static struct grpci1_priv *grpci1priv;
+
+static int grpci1_cfg_w32(struct grpci1_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 val);
+
+static int grpci1_map_irq(const struct pci_dev *dev, u8 slot, u8 pin)
+{
+ struct grpci1_priv *priv = dev->bus->sysdata;
+ int irq_group;
+
+ /* Use default IRQ decoding on PCI BUS0 according slot numbering */
+ irq_group = slot & 0x3;
+ pin = ((pin - 1) + irq_group) & 0x3;
+
+ return priv->irq_map[pin];
+}
+
+static int grpci1_cfg_r32(struct grpci1_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 *val)
+{
+ u32 *pci_conf, tmp, cfg;
+
+ if (where & 0x3)
+ return -EINVAL;
+
+ if (bus == 0) {
+ devfn += (0x8 * 6); /* start at AD16=Device0 */
+ } else if (bus == TGT) {
+ bus = 0;
+ devfn = 0; /* special case: bridge controller itself */
+ }
+
+ /* Select bus */
+ cfg = REGLOAD(priv->regs->cfg_stat);
+ REGSTORE(priv->regs->cfg_stat, (cfg & ~(0xf << 23)) | (bus << 23));
+
+ /* do read access */
+ pci_conf = (u32 *) (priv->pci_conf | (devfn << 8) | (where & 0xfc));
+ tmp = LEON3_BYPASS_LOAD_PA(pci_conf);
+
+ /* check if master abort was received */
+ if (REGLOAD(priv->regs->cfg_stat) & CFGSTAT_CTO) {
+ *val = 0xffffffff;
+ /* Clear Master abort bit in PCI cfg space (is set) */
+ tmp = REGLOAD(priv->regs->stat_cmd);
+ grpci1_cfg_w32(priv, TGT, 0, PCI_COMMAND, tmp);
+ } else {
+ /* Bus always little endian (unaffected by byte-swapping) */
+ *val = swab32(tmp);
+ }
+
+ return 0;
+}
+
+static int grpci1_cfg_r16(struct grpci1_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 *val)
+{
+ u32 v;
+ int ret;
+
+ if (where & 0x1)
+ return -EINVAL;
+ ret = grpci1_cfg_r32(priv, bus, devfn, where & ~0x3, &v);
+ *val = 0xffff & (v >> (8 * (where & 0x3)));
+ return ret;
+}
+
+static int grpci1_cfg_r8(struct grpci1_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 *val)
+{
+ u32 v;
+ int ret;
+
+ ret = grpci1_cfg_r32(priv, bus, devfn, where & ~0x3, &v);
+ *val = 0xff & (v >> (8 * (where & 3)));
+
+ return ret;
+}
+
+static int grpci1_cfg_w32(struct grpci1_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 val)
+{
+ unsigned int *pci_conf;
+ u32 cfg;
+
+ if (where & 0x3)
+ return -EINVAL;
+
+ if (bus == 0) {
+ devfn += (0x8 * 6); /* start at AD16=Device0 */
+ } else if (bus == TGT) {
+ bus = 0;
+ devfn = 0; /* special case: bridge controller itself */
+ }
+
+ /* Select bus */
+ cfg = REGLOAD(priv->regs->cfg_stat);
+ REGSTORE(priv->regs->cfg_stat, (cfg & ~(0xf << 23)) | (bus << 23));
+
+ pci_conf = (unsigned int *) (priv->pci_conf |
+ (devfn << 8) | (where & 0xfc));
+ LEON3_BYPASS_STORE_PA(pci_conf, swab32(val));
+
+ return 0;
+}
+
+static int grpci1_cfg_w16(struct grpci1_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 val)
+{
+ int ret;
+ u32 v;
+
+ if (where & 0x1)
+ return -EINVAL;
+ ret = grpci1_cfg_r32(priv, bus, devfn, where&~3, &v);
+ if (ret)
+ return ret;
+ v = (v & ~(0xffff << (8 * (where & 0x3)))) |
+ ((0xffff & val) << (8 * (where & 0x3)));
+ return grpci1_cfg_w32(priv, bus, devfn, where & ~0x3, v);
+}
+
+static int grpci1_cfg_w8(struct grpci1_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 val)
+{
+ int ret;
+ u32 v;
+
+ ret = grpci1_cfg_r32(priv, bus, devfn, where & ~0x3, &v);
+ if (ret != 0)
+ return ret;
+ v = (v & ~(0xff << (8 * (where & 0x3)))) |
+ ((0xff & val) << (8 * (where & 0x3)));
+ return grpci1_cfg_w32(priv, bus, devfn, where & ~0x3, v);
+}
+
+/* Read from Configuration Space. When entering here the PCI layer has taken
+ * the pci_lock spinlock and IRQ is off.
+ */
+static int grpci1_read_config(struct pci_bus *bus, unsigned int devfn,
+ int where, int size, u32 *val)
+{
+ struct grpci1_priv *priv = grpci1priv;
+ unsigned int busno = bus->number;
+ int ret;
+
+ if (PCI_SLOT(devfn) > 15 || busno > 15) {
+ *val = ~0;
+ return 0;
+ }
+
+ switch (size) {
+ case 1:
+ ret = grpci1_cfg_r8(priv, busno, devfn, where, val);
+ break;
+ case 2:
+ ret = grpci1_cfg_r16(priv, busno, devfn, where, val);
+ break;
+ case 4:
+ ret = grpci1_cfg_r32(priv, busno, devfn, where, val);
+ break;
+ default:
+ ret = -EINVAL;
+ break;
+ }
+
+#ifdef GRPCI1_DEBUG_CFGACCESS
+ printk(KERN_INFO
+ "grpci1_read_config: [%02x:%02x:%x] ofs=%d val=%x size=%d\n",
+ busno, PCI_SLOT(devfn), PCI_FUNC(devfn), where, *val, size);
+#endif
+
+ return ret;
+}
+
+/* Write to Configuration Space. When entering here the PCI layer has taken
+ * the pci_lock spinlock and IRQ is off.
+ */
+static int grpci1_write_config(struct pci_bus *bus, unsigned int devfn,
+ int where, int size, u32 val)
+{
+ struct grpci1_priv *priv = grpci1priv;
+ unsigned int busno = bus->number;
+
+ if (PCI_SLOT(devfn) > 15 || busno > 15)
+ return 0;
+
+#ifdef GRPCI1_DEBUG_CFGACCESS
+ printk(KERN_INFO
+ "grpci1_write_config: [%02x:%02x:%x] ofs=%d size=%d val=%x\n",
+ busno, PCI_SLOT(devfn), PCI_FUNC(devfn), where, size, val);
+#endif
+
+ switch (size) {
+ default:
+ return -EINVAL;
+ case 1:
+ return grpci1_cfg_w8(priv, busno, devfn, where, val);
+ case 2:
+ return grpci1_cfg_w16(priv, busno, devfn, where, val);
+ case 4:
+ return grpci1_cfg_w32(priv, busno, devfn, where, val);
+ }
+}
+
+static struct pci_ops grpci1_ops = {
+ .read = grpci1_read_config,
+ .write = grpci1_write_config,
+};
+
+/* GENIRQ IRQ chip implementation for grpci1 irqmode=0..2. In configuration
+ * 3 where all PCI Interrupts has a separate IRQ on the system IRQ controller
+ * this is not needed and the standard IRQ controller can be used.
+ */
+
+static void grpci1_mask_irq(struct irq_data *data)
+{
+ u32 irqidx;
+ struct grpci1_priv *priv = grpci1priv;
+
+ irqidx = (u32)data->chip_data - 1;
+ if (irqidx > 3) /* only mask PCI interrupts here */
+ return;
+ irqidx += IRQ_MASK_BIT;
+
+ REGSTORE(priv->regs->irq, REGLOAD(priv->regs->irq) & ~(1 << irqidx));
+}
+
+static void grpci1_unmask_irq(struct irq_data *data)
+{
+ u32 irqidx;
+ struct grpci1_priv *priv = grpci1priv;
+
+ irqidx = (u32)data->chip_data - 1;
+ if (irqidx > 3) /* only unmask PCI interrupts here */
+ return;
+ irqidx += IRQ_MASK_BIT;
+
+ REGSTORE(priv->regs->irq, REGLOAD(priv->regs->irq) | (1 << irqidx));
+}
+
+static unsigned int grpci1_startup_irq(struct irq_data *data)
+{
+ grpci1_unmask_irq(data);
+ return 0;
+}
+
+static void grpci1_shutdown_irq(struct irq_data *data)
+{
+ grpci1_mask_irq(data);
+}
+
+static struct irq_chip grpci1_irq = {
+ .name = "grpci1",
+ .irq_startup = grpci1_startup_irq,
+ .irq_shutdown = grpci1_shutdown_irq,
+ .irq_mask = grpci1_mask_irq,
+ .irq_unmask = grpci1_unmask_irq,
+};
+
+/* Handle one or multiple IRQs from the PCI core */
+static void grpci1_pci_flow_irq(unsigned int irq, struct irq_desc *desc)
+{
+ struct grpci1_priv *priv = grpci1priv;
+ int i, ack = 0;
+ unsigned int irqreg;
+
+ irqreg = REGLOAD(priv->regs->irq);
+ irqreg = (irqreg >> IRQ_MASK_BIT) & irqreg;
+
+ /* Error Interrupt? */
+ if (irqreg & IRQ_ALL_ERRORS) {
+ generic_handle_irq(priv->irq_err);
+ ack = 1;
+ }
+
+ /* PCI Interrupt? */
+ if (irqreg & IRQ_INTX) {
+ /* Call respective PCI Interrupt handler */
+ for (i = 0; i < 4; i++) {
+ if (irqreg & (1 << i))
+ generic_handle_irq(priv->irq_map[i]);
+ }
+ ack = 1;
+ }
+
+ /*
+ * Call "first level" IRQ chip end-of-irq handler. It will ACK LEON IRQ
+ * Controller, this must be done after IRQ sources have been handled to
+ * avoid double IRQ generation
+ */
+ if (ack)
+ desc->irq_data.chip->irq_eoi(&desc->irq_data);
+}
+
+/* Create a virtual IRQ */
+static unsigned int grpci1_build_device_irq(unsigned int irq)
+{
+ unsigned int virq = 0, pil;
+
+ pil = 1 << 8;
+ virq = irq_alloc(irq, pil);
+ if (virq == 0)
+ goto out;
+
+ irq_set_chip_and_handler_name(virq, &grpci1_irq, handle_simple_irq,
+ "pcilvl");
+ irq_set_chip_data(virq, (void *)irq);
+
+out:
+ return virq;
+}
+
+/*
+ * Initialize mappings AMBA<->PCI, clear IRQ state, setup PCI interface
+ *
+ * Target BARs:
+ * BAR0: unused in this implementation
+ * BAR1: peripheral DMA to host's memory (size at least 256MByte)
+ * BAR2..BAR5: not implemented in hardware
+ */
+static void grpci1_hw_init(struct grpci1_priv *priv)
+{
+ u32 ahbadr, bar_sz, data, pciadr;
+ struct grpci1_regs __iomem *regs = priv->regs;
+
+ /* set 1:1 mapping between AHB -> PCI memory space */
+ REGSTORE(regs->cfg_stat, priv->pci_area & 0xf0000000);
+
+ /* map PCI accesses to target BAR1 to Linux kernel memory 1:1 */
+ ahbadr = 0xf0000000 & (u32)__pa(PAGE_ALIGN((unsigned long) &_end));
+ REGSTORE(regs->page1, ahbadr);
+
+ /* translate I/O accesses to 0, I/O Space always @ PCI low 64Kbytes */
+ REGSTORE(regs->iomap, REGLOAD(regs->iomap) & 0x0000ffff);
+
+ /* disable and clear pending interrupts */
+ REGSTORE(regs->irq, 0);
+
+ /* Setup BAR0 outside access range so that it does not conflict with
+ * peripheral DMA. There is no need to set up the PAGE0 register.
+ */
+ grpci1_cfg_w32(priv, TGT, 0, PCI_BASE_ADDRESS_0, 0xffffffff);
+ grpci1_cfg_r32(priv, TGT, 0, PCI_BASE_ADDRESS_0, &bar_sz);
+ bar_sz = ~bar_sz + 1;
+ pciadr = priv->pci_area - bar_sz;
+ grpci1_cfg_w32(priv, TGT, 0, PCI_BASE_ADDRESS_0, pciadr);
+
+ /*
+ * Setup the Host's PCI Target BAR1 for other peripherals to access,
+ * and do DMA to the host's memory.
+ */
+ grpci1_cfg_w32(priv, TGT, 0, PCI_BASE_ADDRESS_1, ahbadr);
+
+ /*
+ * Setup Latency Timer and cache line size. Default cache line
+ * size will result in poor performance (256 word fetches), 0xff
+ * will set it according to the max size of the PCI FIFO.
+ */
+ grpci1_cfg_w8(priv, TGT, 0, PCI_CACHE_LINE_SIZE, 0xff);
+ grpci1_cfg_w8(priv, TGT, 0, PCI_LATENCY_TIMER, 0x40);
+
+ /* set as bus master, enable pci memory responses, clear status bits */
+ grpci1_cfg_r32(priv, TGT, 0, PCI_COMMAND, &data);
+ data |= (PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER);
+ grpci1_cfg_w32(priv, TGT, 0, PCI_COMMAND, data);
+}
+
+static irqreturn_t grpci1_jump_interrupt(int irq, void *arg)
+{
+ struct grpci1_priv *priv = arg;
+ dev_err(priv->dev, "Jump IRQ happened\n");
+ return IRQ_NONE;
+}
+
+/* Handle GRPCI1 Error Interrupt */
+static irqreturn_t grpci1_err_interrupt(int irq, void *arg)
+{
+ struct grpci1_priv *priv = arg;
+ u32 status;
+
+ grpci1_cfg_r16(priv, TGT, 0, PCI_STATUS, &status);
+ status &= priv->pci_err_mask;
+
+ if (status == 0)
+ return IRQ_NONE;
+
+ if (status & PCI_STATUS_PARITY)
+ dev_err(priv->dev, "Data Parity Error\n");
+
+ if (status & PCI_STATUS_SIG_TARGET_ABORT)
+ dev_err(priv->dev, "Signalled Target Abort\n");
+
+ if (status & PCI_STATUS_REC_TARGET_ABORT)
+ dev_err(priv->dev, "Received Target Abort\n");
+
+ if (status & PCI_STATUS_REC_MASTER_ABORT)
+ dev_err(priv->dev, "Received Master Abort\n");
+
+ if (status & PCI_STATUS_SIG_SYSTEM_ERROR)
+ dev_err(priv->dev, "Signalled System Error\n");
+
+ if (status & PCI_STATUS_DETECTED_PARITY)
+ dev_err(priv->dev, "Parity Error\n");
+
+ /* Clear handled INT TYPE IRQs */
+ grpci1_cfg_w16(priv, TGT, 0, PCI_STATUS, status);
+
+ return IRQ_HANDLED;
+}
+
+static int grpci1_of_probe(struct platform_device *ofdev)
+{
+ struct grpci1_regs __iomem *regs;
+ struct grpci1_priv *priv;
+ int err, len;
+ const int *tmp;
+ u32 cfg, size, err_mask;
+ struct resource *res;
+
+ if (grpci1priv) {
+ dev_err(&ofdev->dev, "only one GRPCI1 supported\n");
+ return -ENODEV;
+ }
+
+ if (ofdev->num_resources < 3) {
+ dev_err(&ofdev->dev, "not enough APB/AHB resources\n");
+ return -EIO;
+ }
+
+ priv = devm_kzalloc(&ofdev->dev, sizeof(*priv), GFP_KERNEL);
+ if (!priv) {
+ dev_err(&ofdev->dev, "memory allocation failed\n");
+ return -ENOMEM;
+ }
+ platform_set_drvdata(ofdev, priv);
+ priv->dev = &ofdev->dev;
+
+ /* find device register base address */
+ res = platform_get_resource(ofdev, IORESOURCE_MEM, 0);
+ regs = devm_ioremap_resource(&ofdev->dev, res);
+ if (IS_ERR(regs))
+ return PTR_ERR(regs);
+
+ /*
+ * check that we're in Host Slot and that we can act as a Host Bridge
+ * and not only as target/peripheral.
+ */
+ cfg = REGLOAD(regs->cfg_stat);
+ if ((cfg & CFGSTAT_HOST) == 0) {
+ dev_err(&ofdev->dev, "not in host system slot\n");
+ return -EIO;
+ }
+
+ /* check that BAR1 support 256 MByte so that we can map kernel space */
+ REGSTORE(regs->page1, 0xffffffff);
+ size = ~REGLOAD(regs->page1) + 1;
+ if (size < 0x10000000) {
+ dev_err(&ofdev->dev, "BAR1 must be at least 256MByte\n");
+ return -EIO;
+ }
+
+ /* hardware must support little-endian PCI (byte-twisting) */
+ if ((REGLOAD(regs->page0) & PAGE0_BTEN) == 0) {
+ dev_err(&ofdev->dev, "byte-twisting is required\n");
+ return -EIO;
+ }
+
+ priv->regs = regs;
+ priv->irq = irq_of_parse_and_map(ofdev->dev.of_node, 0);
+ dev_info(&ofdev->dev, "host found at 0x%p, irq%d\n", regs, priv->irq);
+
+ /* Find PCI Memory, I/O and Configuration Space Windows */
+ priv->pci_area = ofdev->resource[1].start;
+ priv->pci_area_end = ofdev->resource[1].end+1;
+ priv->pci_io = ofdev->resource[2].start;
+ priv->pci_conf = ofdev->resource[2].start + 0x10000;
+ priv->pci_conf_end = priv->pci_conf + 0x10000;
+ priv->pci_io_va = (unsigned long)ioremap(priv->pci_io, 0x10000);
+ if (!priv->pci_io_va) {
+ dev_err(&ofdev->dev, "unable to map PCI I/O area\n");
+ return -EIO;
+ }
+
+ printk(KERN_INFO
+ "GRPCI1: MEMORY SPACE [0x%08lx - 0x%08lx]\n"
+ " I/O SPACE [0x%08lx - 0x%08lx]\n"
+ " CONFIG SPACE [0x%08lx - 0x%08lx]\n",
+ priv->pci_area, priv->pci_area_end-1,
+ priv->pci_io, priv->pci_conf-1,
+ priv->pci_conf, priv->pci_conf_end-1);
+
+ /*
+ * I/O Space resources in I/O Window mapped into Virtual Adr Space
+ * We never use low 4KB because some devices seem have problems using
+ * address 0.
+ */
+ priv->info.io_space.name = "GRPCI1 PCI I/O Space";
+ priv->info.io_space.start = priv->pci_io_va + 0x1000;
+ priv->info.io_space.end = priv->pci_io_va + 0x10000 - 1;
+ priv->info.io_space.flags = IORESOURCE_IO;
+
+ /*
+ * grpci1 has no prefetchable memory, map everything as
+ * non-prefetchable memory
+ */
+ priv->info.mem_space.name = "GRPCI1 PCI MEM Space";
+ priv->info.mem_space.start = priv->pci_area;
+ priv->info.mem_space.end = priv->pci_area_end - 1;
+ priv->info.mem_space.flags = IORESOURCE_MEM;
+
+ if (request_resource(&iomem_resource, &priv->info.mem_space) < 0) {
+ dev_err(&ofdev->dev, "unable to request PCI memory area\n");
+ err = -ENOMEM;
+ goto err1;
+ }
+
+ if (request_resource(&ioport_resource, &priv->info.io_space) < 0) {
+ dev_err(&ofdev->dev, "unable to request PCI I/O area\n");
+ err = -ENOMEM;
+ goto err2;
+ }
+
+ /* setup maximum supported PCI buses */
+ priv->info.busn.name = "GRPCI1 busn";
+ priv->info.busn.start = 0;
+ priv->info.busn.end = 15;
+
+ grpci1priv = priv;
+
+ /* Initialize hardware */
+ grpci1_hw_init(priv);
+
+ /*
+ * Get PCI Interrupt to System IRQ mapping and setup IRQ handling
+ * Error IRQ. All PCI and PCI-Error interrupts are shared using the
+ * same system IRQ.
+ */
+ leon_update_virq_handling(priv->irq, grpci1_pci_flow_irq, "pcilvl", 0);
+
+ priv->irq_map[0] = grpci1_build_device_irq(1);
+ priv->irq_map[1] = grpci1_build_device_irq(2);
+ priv->irq_map[2] = grpci1_build_device_irq(3);
+ priv->irq_map[3] = grpci1_build_device_irq(4);
+ priv->irq_err = grpci1_build_device_irq(5);
+
+ printk(KERN_INFO " PCI INTA..D#: IRQ%d, IRQ%d, IRQ%d, IRQ%d\n",
+ priv->irq_map[0], priv->irq_map[1], priv->irq_map[2],
+ priv->irq_map[3]);
+
+ /* Enable IRQs on LEON IRQ controller */
+ err = devm_request_irq(&ofdev->dev, priv->irq, grpci1_jump_interrupt, 0,
+ "GRPCI1_JUMP", priv);
+ if (err) {
+ dev_err(&ofdev->dev, "ERR IRQ request failed: %d\n", err);
+ goto err3;
+ }
+
+ /* Setup IRQ handler for access errors */
+ err = devm_request_irq(&ofdev->dev, priv->irq_err,
+ grpci1_err_interrupt, IRQF_SHARED, "GRPCI1_ERR",
+ priv);
+ if (err) {
+ dev_err(&ofdev->dev, "ERR VIRQ request failed: %d\n", err);
+ goto err3;
+ }
+
+ tmp = of_get_property(ofdev->dev.of_node, "all_pci_errors", &len);
+ if (tmp && (len == 4)) {
+ priv->pci_err_mask = ALL_PCI_ERRORS;
+ err_mask = IRQ_ALL_ERRORS << IRQ_MASK_BIT;
+ } else {
+ priv->pci_err_mask = DEF_PCI_ERRORS;
+ err_mask = IRQ_DEF_ERRORS << IRQ_MASK_BIT;
+ }
+
+ /*
+ * Enable Error Interrupts. PCI interrupts are unmasked once request_irq
+ * is called by the PCI Device drivers
+ */
+ REGSTORE(regs->irq, err_mask);
+
+ /* Init common layer and scan buses */
+ priv->info.ops = &grpci1_ops;
+ priv->info.map_irq = grpci1_map_irq;
+ leon_pci_init(ofdev, &priv->info);
+
+ return 0;
+
+err3:
+ release_resource(&priv->info.io_space);
+err2:
+ release_resource(&priv->info.mem_space);
+err1:
+ iounmap((void __iomem *)priv->pci_io_va);
+ grpci1priv = NULL;
+ return err;
+}
+
+static struct of_device_id grpci1_of_match[] = {
+ {
+ .name = "GAISLER_PCIFBRG",
+ },
+ {
+ .name = "01_014",
+ },
+ {},
+};
+
+static struct platform_driver grpci1_of_driver = {
+ .driver = {
+ .name = "grpci1",
+ .owner = THIS_MODULE,
+ .of_match_table = grpci1_of_match,
+ },
+ .probe = grpci1_of_probe,
+};
+
+static int __init grpci1_init(void)
+{
+ return platform_driver_register(&grpci1_of_driver);
+}
+
+subsys_initcall(grpci1_init);
diff --git a/arch/sparc/kernel/leon_pci_grpci2.c b/arch/sparc/kernel/leon_pci_grpci2.c
new file mode 100644
index 00000000000..e433a4d69fe
--- /dev/null
+++ b/arch/sparc/kernel/leon_pci_grpci2.c
@@ -0,0 +1,914 @@
+/*
+ * leon_pci_grpci2.c: GRPCI2 Host PCI driver
+ *
+ * Copyright (C) 2011 Aeroflex Gaisler AB, Daniel Hellstrom
+ *
+ */
+
+#include <linux/of_device.h>
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/export.h>
+#include <asm/io.h>
+#include <asm/leon.h>
+#include <asm/vaddrs.h>
+#include <asm/sections.h>
+#include <asm/leon_pci.h>
+
+#include "irq.h"
+
+struct grpci2_barcfg {
+ unsigned long pciadr; /* PCI Space Address */
+ unsigned long ahbadr; /* PCI Base address mapped to this AHB addr */
+};
+
+/* Device Node Configuration options:
+ * - barcfgs : Custom Configuration of Host's 6 target BARs
+ * - irq_mask : Limit which PCI interrupts are enabled
+ * - do_reset : Force PCI Reset on startup
+ *
+ * barcfgs
+ * =======
+ *
+ * Optional custom Target BAR configuration (see struct grpci2_barcfg). All
+ * addresses are physical. Array always contains 6 elements (len=2*4*6 bytes)
+ *
+ * -1 means not configured (let host driver do default setup).
+ *
+ * [i*2+0] = PCI Address of BAR[i] on target interface
+ * [i*2+1] = Accessing PCI address of BAR[i] result in this AMBA address
+ *
+ *
+ * irq_mask
+ * ========
+ *
+ * Limit which PCI interrupts are enabled. 0=Disable, 1=Enable. By default
+ * all are enabled. Use this when PCI interrupt pins are floating on PCB.
+ * int, len=4.
+ * bit0 = PCI INTA#
+ * bit1 = PCI INTB#
+ * bit2 = PCI INTC#
+ * bit3 = PCI INTD#
+ *
+ *
+ * reset
+ * =====
+ *
+ * Force PCI reset on startup. int, len=4
+ */
+
+/* Enable Debugging Configuration Space Access */
+#undef GRPCI2_DEBUG_CFGACCESS
+
+/*
+ * GRPCI2 APB Register MAP
+ */
+struct grpci2_regs {
+ unsigned int ctrl; /* 0x00 Control */
+ unsigned int sts_cap; /* 0x04 Status / Capabilities */
+ int res1; /* 0x08 */
+ unsigned int io_map; /* 0x0C I/O Map address */
+ unsigned int dma_ctrl; /* 0x10 DMA */
+ unsigned int dma_bdbase; /* 0x14 DMA */
+ int res2[2]; /* 0x18 */
+ unsigned int bars[6]; /* 0x20 read-only PCI BARs */
+ int res3[2]; /* 0x38 */
+ unsigned int ahbmst_map[16]; /* 0x40 AHB->PCI Map per AHB Master */
+
+ /* PCI Trace Buffer Registers (OPTIONAL) */
+ unsigned int t_ctrl; /* 0x80 */
+ unsigned int t_cnt; /* 0x84 */
+ unsigned int t_adpat; /* 0x88 */
+ unsigned int t_admask; /* 0x8C */
+ unsigned int t_sigpat; /* 0x90 */
+ unsigned int t_sigmask; /* 0x94 */
+ unsigned int t_adstate; /* 0x98 */
+ unsigned int t_sigstate; /* 0x9C */
+};
+
+#define REGLOAD(a) (be32_to_cpu(__raw_readl(&(a))))
+#define REGSTORE(a, v) (__raw_writel(cpu_to_be32(v), &(a)))
+
+#define CTRL_BUS_BIT 16
+
+#define CTRL_RESET (1<<31)
+#define CTRL_SI (1<<27)
+#define CTRL_PE (1<<26)
+#define CTRL_EI (1<<25)
+#define CTRL_ER (1<<24)
+#define CTRL_BUS (0xff<<CTRL_BUS_BIT)
+#define CTRL_HOSTINT 0xf
+
+#define STS_HOST_BIT 31
+#define STS_MST_BIT 30
+#define STS_TAR_BIT 29
+#define STS_DMA_BIT 28
+#define STS_DI_BIT 27
+#define STS_HI_BIT 26
+#define STS_IRQMODE_BIT 24
+#define STS_TRACE_BIT 23
+#define STS_CFGERRVALID_BIT 20
+#define STS_CFGERR_BIT 19
+#define STS_INTTYPE_BIT 12
+#define STS_INTSTS_BIT 8
+#define STS_FDEPTH_BIT 2
+#define STS_FNUM_BIT 0
+
+#define STS_HOST (1<<STS_HOST_BIT)
+#define STS_MST (1<<STS_MST_BIT)
+#define STS_TAR (1<<STS_TAR_BIT)
+#define STS_DMA (1<<STS_DMA_BIT)
+#define STS_DI (1<<STS_DI_BIT)
+#define STS_HI (1<<STS_HI_BIT)
+#define STS_IRQMODE (0x3<<STS_IRQMODE_BIT)
+#define STS_TRACE (1<<STS_TRACE_BIT)
+#define STS_CFGERRVALID (1<<STS_CFGERRVALID_BIT)
+#define STS_CFGERR (1<<STS_CFGERR_BIT)
+#define STS_INTTYPE (0x3f<<STS_INTTYPE_BIT)
+#define STS_INTSTS (0xf<<STS_INTSTS_BIT)
+#define STS_FDEPTH (0x7<<STS_FDEPTH_BIT)
+#define STS_FNUM (0x3<<STS_FNUM_BIT)
+
+#define STS_ISYSERR (1<<17)
+#define STS_IDMA (1<<16)
+#define STS_IDMAERR (1<<15)
+#define STS_IMSTABRT (1<<14)
+#define STS_ITGTABRT (1<<13)
+#define STS_IPARERR (1<<12)
+
+#define STS_ERR_IRQ (STS_ISYSERR | STS_IMSTABRT | STS_ITGTABRT | STS_IPARERR)
+
+struct grpci2_bd_chan {
+ unsigned int ctrl; /* 0x00 DMA Control */
+ unsigned int nchan; /* 0x04 Next DMA Channel Address */
+ unsigned int nbd; /* 0x08 Next Data Descriptor in chan */
+ unsigned int res; /* 0x0C Reserved */
+};
+
+#define BD_CHAN_EN 0x80000000
+#define BD_CHAN_TYPE 0x00300000
+#define BD_CHAN_BDCNT 0x0000ffff
+#define BD_CHAN_EN_BIT 31
+#define BD_CHAN_TYPE_BIT 20
+#define BD_CHAN_BDCNT_BIT 0
+
+struct grpci2_bd_data {
+ unsigned int ctrl; /* 0x00 DMA Data Control */
+ unsigned int pci_adr; /* 0x04 PCI Start Address */
+ unsigned int ahb_adr; /* 0x08 AHB Start address */
+ unsigned int next; /* 0x0C Next Data Descriptor in chan */
+};
+
+#define BD_DATA_EN 0x80000000
+#define BD_DATA_IE 0x40000000
+#define BD_DATA_DR 0x20000000
+#define BD_DATA_TYPE 0x00300000
+#define BD_DATA_ER 0x00080000
+#define BD_DATA_LEN 0x0000ffff
+#define BD_DATA_EN_BIT 31
+#define BD_DATA_IE_BIT 30
+#define BD_DATA_DR_BIT 29
+#define BD_DATA_TYPE_BIT 20
+#define BD_DATA_ER_BIT 19
+#define BD_DATA_LEN_BIT 0
+
+/* GRPCI2 Capability */
+struct grpci2_cap_first {
+ unsigned int ctrl;
+ unsigned int pci2ahb_map[6];
+ unsigned int ext2ahb_map;
+ unsigned int io_map;
+ unsigned int pcibar_size[6];
+};
+#define CAP9_CTRL_OFS 0
+#define CAP9_BAR_OFS 0x4
+#define CAP9_IOMAP_OFS 0x20
+#define CAP9_BARSIZE_OFS 0x24
+
+#define TGT 256
+
+struct grpci2_priv {
+ struct leon_pci_info info; /* must be on top of this structure */
+ struct grpci2_regs __iomem *regs;
+ char irq;
+ char irq_mode; /* IRQ Mode from CAPSTS REG */
+ char bt_enabled;
+ char do_reset;
+ char irq_mask;
+ u32 pciid; /* PCI ID of Host */
+ unsigned char irq_map[4];
+
+ /* Virtual IRQ numbers */
+ unsigned int virq_err;
+ unsigned int virq_dma;
+
+ /* AHB PCI Windows */
+ unsigned long pci_area; /* MEMORY */
+ unsigned long pci_area_end;
+ unsigned long pci_io; /* I/O */
+ unsigned long pci_conf; /* CONFIGURATION */
+ unsigned long pci_conf_end;
+ unsigned long pci_io_va;
+
+ struct grpci2_barcfg tgtbars[6];
+};
+
+static DEFINE_SPINLOCK(grpci2_dev_lock);
+static struct grpci2_priv *grpci2priv;
+
+static int grpci2_map_irq(const struct pci_dev *dev, u8 slot, u8 pin)
+{
+ struct grpci2_priv *priv = dev->bus->sysdata;
+ int irq_group;
+
+ /* Use default IRQ decoding on PCI BUS0 according slot numbering */
+ irq_group = slot & 0x3;
+ pin = ((pin - 1) + irq_group) & 0x3;
+
+ return priv->irq_map[pin];
+}
+
+static int grpci2_cfg_r32(struct grpci2_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 *val)
+{
+ unsigned int *pci_conf;
+ unsigned long flags;
+ u32 tmp;
+
+ if (where & 0x3)
+ return -EINVAL;
+
+ if (bus == 0) {
+ devfn += (0x8 * 6); /* start at AD16=Device0 */
+ } else if (bus == TGT) {
+ bus = 0;
+ devfn = 0; /* special case: bridge controller itself */
+ }
+
+ /* Select bus */
+ spin_lock_irqsave(&grpci2_dev_lock, flags);
+ REGSTORE(priv->regs->ctrl, (REGLOAD(priv->regs->ctrl) & ~(0xff << 16)) |
+ (bus << 16));
+ spin_unlock_irqrestore(&grpci2_dev_lock, flags);
+
+ /* clear old status */
+ REGSTORE(priv->regs->sts_cap, (STS_CFGERR | STS_CFGERRVALID));
+
+ pci_conf = (unsigned int *) (priv->pci_conf |
+ (devfn << 8) | (where & 0xfc));
+ tmp = LEON3_BYPASS_LOAD_PA(pci_conf);
+
+ /* Wait until GRPCI2 signals that CFG access is done, it should be
+ * done instantaneously unless a DMA operation is ongoing...
+ */
+ while ((REGLOAD(priv->regs->sts_cap) & STS_CFGERRVALID) == 0)
+ ;
+
+ if (REGLOAD(priv->regs->sts_cap) & STS_CFGERR) {
+ *val = 0xffffffff;
+ } else {
+ /* Bus always little endian (unaffected by byte-swapping) */
+ *val = swab32(tmp);
+ }
+
+ return 0;
+}
+
+static int grpci2_cfg_r16(struct grpci2_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 *val)
+{
+ u32 v;
+ int ret;
+
+ if (where & 0x1)
+ return -EINVAL;
+ ret = grpci2_cfg_r32(priv, bus, devfn, where & ~0x3, &v);
+ *val = 0xffff & (v >> (8 * (where & 0x3)));
+ return ret;
+}
+
+static int grpci2_cfg_r8(struct grpci2_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 *val)
+{
+ u32 v;
+ int ret;
+
+ ret = grpci2_cfg_r32(priv, bus, devfn, where & ~0x3, &v);
+ *val = 0xff & (v >> (8 * (where & 3)));
+
+ return ret;
+}
+
+static int grpci2_cfg_w32(struct grpci2_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 val)
+{
+ unsigned int *pci_conf;
+ unsigned long flags;
+
+ if (where & 0x3)
+ return -EINVAL;
+
+ if (bus == 0) {
+ devfn += (0x8 * 6); /* start at AD16=Device0 */
+ } else if (bus == TGT) {
+ bus = 0;
+ devfn = 0; /* special case: bridge controller itself */
+ }
+
+ /* Select bus */
+ spin_lock_irqsave(&grpci2_dev_lock, flags);
+ REGSTORE(priv->regs->ctrl, (REGLOAD(priv->regs->ctrl) & ~(0xff << 16)) |
+ (bus << 16));
+ spin_unlock_irqrestore(&grpci2_dev_lock, flags);
+
+ /* clear old status */
+ REGSTORE(priv->regs->sts_cap, (STS_CFGERR | STS_CFGERRVALID));
+
+ pci_conf = (unsigned int *) (priv->pci_conf |
+ (devfn << 8) | (where & 0xfc));
+ LEON3_BYPASS_STORE_PA(pci_conf, swab32(val));
+
+ /* Wait until GRPCI2 signals that CFG access is done, it should be
+ * done instantaneously unless a DMA operation is ongoing...
+ */
+ while ((REGLOAD(priv->regs->sts_cap) & STS_CFGERRVALID) == 0)
+ ;
+
+ return 0;
+}
+
+static int grpci2_cfg_w16(struct grpci2_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 val)
+{
+ int ret;
+ u32 v;
+
+ if (where & 0x1)
+ return -EINVAL;
+ ret = grpci2_cfg_r32(priv, bus, devfn, where&~3, &v);
+ if (ret)
+ return ret;
+ v = (v & ~(0xffff << (8 * (where & 0x3)))) |
+ ((0xffff & val) << (8 * (where & 0x3)));
+ return grpci2_cfg_w32(priv, bus, devfn, where & ~0x3, v);
+}
+
+static int grpci2_cfg_w8(struct grpci2_priv *priv, unsigned int bus,
+ unsigned int devfn, int where, u32 val)
+{
+ int ret;
+ u32 v;
+
+ ret = grpci2_cfg_r32(priv, bus, devfn, where & ~0x3, &v);
+ if (ret != 0)
+ return ret;
+ v = (v & ~(0xff << (8 * (where & 0x3)))) |
+ ((0xff & val) << (8 * (where & 0x3)));
+ return grpci2_cfg_w32(priv, bus, devfn, where & ~0x3, v);
+}
+
+/* Read from Configuration Space. When entering here the PCI layer has taken
+ * the pci_lock spinlock and IRQ is off.
+ */
+static int grpci2_read_config(struct pci_bus *bus, unsigned int devfn,
+ int where, int size, u32 *val)
+{
+ struct grpci2_priv *priv = grpci2priv;
+ unsigned int busno = bus->number;
+ int ret;
+
+ if (PCI_SLOT(devfn) > 15 || busno > 255) {
+ *val = ~0;
+ return 0;
+ }
+
+ switch (size) {
+ case 1:
+ ret = grpci2_cfg_r8(priv, busno, devfn, where, val);
+ break;
+ case 2:
+ ret = grpci2_cfg_r16(priv, busno, devfn, where, val);
+ break;
+ case 4:
+ ret = grpci2_cfg_r32(priv, busno, devfn, where, val);
+ break;
+ default:
+ ret = -EINVAL;
+ break;
+ }
+
+#ifdef GRPCI2_DEBUG_CFGACCESS
+ printk(KERN_INFO "grpci2_read_config: [%02x:%02x:%x] ofs=%d val=%x "
+ "size=%d\n", busno, PCI_SLOT(devfn), PCI_FUNC(devfn), where,
+ *val, size);
+#endif
+
+ return ret;
+}
+
+/* Write to Configuration Space. When entering here the PCI layer has taken
+ * the pci_lock spinlock and IRQ is off.
+ */
+static int grpci2_write_config(struct pci_bus *bus, unsigned int devfn,
+ int where, int size, u32 val)
+{
+ struct grpci2_priv *priv = grpci2priv;
+ unsigned int busno = bus->number;
+
+ if (PCI_SLOT(devfn) > 15 || busno > 255)
+ return 0;
+
+#ifdef GRPCI2_DEBUG_CFGACCESS
+ printk(KERN_INFO "grpci2_write_config: [%02x:%02x:%x] ofs=%d size=%d "
+ "val=%x\n", busno, PCI_SLOT(devfn), PCI_FUNC(devfn),
+ where, size, val);
+#endif
+
+ switch (size) {
+ default:
+ return -EINVAL;
+ case 1:
+ return grpci2_cfg_w8(priv, busno, devfn, where, val);
+ case 2:
+ return grpci2_cfg_w16(priv, busno, devfn, where, val);
+ case 4:
+ return grpci2_cfg_w32(priv, busno, devfn, where, val);
+ }
+}
+
+static struct pci_ops grpci2_ops = {
+ .read = grpci2_read_config,
+ .write = grpci2_write_config,
+};
+
+/* GENIRQ IRQ chip implementation for GRPCI2 irqmode=0..2. In configuration
+ * 3 where all PCI Interrupts has a separate IRQ on the system IRQ controller
+ * this is not needed and the standard IRQ controller can be used.
+ */
+
+static void grpci2_mask_irq(struct irq_data *data)
+{
+ unsigned long flags;
+ unsigned int irqidx;
+ struct grpci2_priv *priv = grpci2priv;
+
+ irqidx = (unsigned int)data->chip_data - 1;
+ if (irqidx > 3) /* only mask PCI interrupts here */
+ return;
+
+ spin_lock_irqsave(&grpci2_dev_lock, flags);
+ REGSTORE(priv->regs->ctrl, REGLOAD(priv->regs->ctrl) & ~(1 << irqidx));
+ spin_unlock_irqrestore(&grpci2_dev_lock, flags);
+}
+
+static void grpci2_unmask_irq(struct irq_data *data)
+{
+ unsigned long flags;
+ unsigned int irqidx;
+ struct grpci2_priv *priv = grpci2priv;
+
+ irqidx = (unsigned int)data->chip_data - 1;
+ if (irqidx > 3) /* only unmask PCI interrupts here */
+ return;
+
+ spin_lock_irqsave(&grpci2_dev_lock, flags);
+ REGSTORE(priv->regs->ctrl, REGLOAD(priv->regs->ctrl) | (1 << irqidx));
+ spin_unlock_irqrestore(&grpci2_dev_lock, flags);
+}
+
+static unsigned int grpci2_startup_irq(struct irq_data *data)
+{
+ grpci2_unmask_irq(data);
+ return 0;
+}
+
+static void grpci2_shutdown_irq(struct irq_data *data)
+{
+ grpci2_mask_irq(data);
+}
+
+static struct irq_chip grpci2_irq = {
+ .name = "grpci2",
+ .irq_startup = grpci2_startup_irq,
+ .irq_shutdown = grpci2_shutdown_irq,
+ .irq_mask = grpci2_mask_irq,
+ .irq_unmask = grpci2_unmask_irq,
+};
+
+/* Handle one or multiple IRQs from the PCI core */
+static void grpci2_pci_flow_irq(unsigned int irq, struct irq_desc *desc)
+{
+ struct grpci2_priv *priv = grpci2priv;
+ int i, ack = 0;
+ unsigned int ctrl, sts_cap, pci_ints;
+
+ ctrl = REGLOAD(priv->regs->ctrl);
+ sts_cap = REGLOAD(priv->regs->sts_cap);
+
+ /* Error Interrupt? */
+ if (sts_cap & STS_ERR_IRQ) {
+ generic_handle_irq(priv->virq_err);
+ ack = 1;
+ }
+
+ /* PCI Interrupt? */
+ pci_ints = ((~sts_cap) >> STS_INTSTS_BIT) & ctrl & CTRL_HOSTINT;
+ if (pci_ints) {
+ /* Call respective PCI Interrupt handler */
+ for (i = 0; i < 4; i++) {
+ if (pci_ints & (1 << i))
+ generic_handle_irq(priv->irq_map[i]);
+ }
+ ack = 1;
+ }
+
+ /*
+ * Decode DMA Interrupt only when shared with Err and PCI INTX#, when
+ * the DMA is a unique IRQ the DMA interrupts doesn't end up here, they
+ * goes directly to DMA ISR.
+ */
+ if ((priv->irq_mode == 0) && (sts_cap & (STS_IDMA | STS_IDMAERR))) {
+ generic_handle_irq(priv->virq_dma);
+ ack = 1;
+ }
+
+ /*
+ * Call "first level" IRQ chip end-of-irq handler. It will ACK LEON IRQ
+ * Controller, this must be done after IRQ sources have been handled to
+ * avoid double IRQ generation
+ */
+ if (ack)
+ desc->irq_data.chip->irq_eoi(&desc->irq_data);
+}
+
+/* Create a virtual IRQ */
+static unsigned int grpci2_build_device_irq(unsigned int irq)
+{
+ unsigned int virq = 0, pil;
+
+ pil = 1 << 8;
+ virq = irq_alloc(irq, pil);
+ if (virq == 0)
+ goto out;
+
+ irq_set_chip_and_handler_name(virq, &grpci2_irq, handle_simple_irq,
+ "pcilvl");
+ irq_set_chip_data(virq, (void *)irq);
+
+out:
+ return virq;
+}
+
+static void grpci2_hw_init(struct grpci2_priv *priv)
+{
+ u32 ahbadr, pciadr, bar_sz, capptr, io_map, data;
+ struct grpci2_regs __iomem *regs = priv->regs;
+ int i;
+ struct grpci2_barcfg *barcfg = priv->tgtbars;
+
+ /* Reset any earlier setup */
+ if (priv->do_reset) {
+ printk(KERN_INFO "GRPCI2: Resetting PCI bus\n");
+ REGSTORE(regs->ctrl, CTRL_RESET);
+ ssleep(1); /* Wait for boards to settle */
+ }
+ REGSTORE(regs->ctrl, 0);
+ REGSTORE(regs->sts_cap, ~0); /* Clear Status */
+ REGSTORE(regs->dma_ctrl, 0);
+ REGSTORE(regs->dma_bdbase, 0);
+
+ /* Translate I/O accesses to 0, I/O Space always @ PCI low 64Kbytes */
+ REGSTORE(regs->io_map, REGLOAD(regs->io_map) & 0x0000ffff);
+
+ /* set 1:1 mapping between AHB -> PCI memory space, for all Masters
+ * Each AHB master has it's own mapping registers. Max 16 AHB masters.
+ */
+ for (i = 0; i < 16; i++)
+ REGSTORE(regs->ahbmst_map[i], priv->pci_area);
+
+ /* Get the GRPCI2 Host PCI ID */
+ grpci2_cfg_r32(priv, TGT, 0, PCI_VENDOR_ID, &priv->pciid);
+
+ /* Get address to first (always defined) capability structure */
+ grpci2_cfg_r8(priv, TGT, 0, PCI_CAPABILITY_LIST, &capptr);
+
+ /* Enable/Disable Byte twisting */
+ grpci2_cfg_r32(priv, TGT, 0, capptr+CAP9_IOMAP_OFS, &io_map);
+ io_map = (io_map & ~0x1) | (priv->bt_enabled ? 1 : 0);
+ grpci2_cfg_w32(priv, TGT, 0, capptr+CAP9_IOMAP_OFS, io_map);
+
+ /* Setup the Host's PCI Target BARs for other peripherals to access,
+ * and do DMA to the host's memory. The target BARs can be sized and
+ * enabled individually.
+ *
+ * User may set custom target BARs, but default is:
+ * The first BARs is used to map kernel low (DMA is part of normal
+ * region on sparc which is SRMMU_MAXMEM big) main memory 1:1 to the
+ * PCI bus, the other BARs are disabled. We assume that the first BAR
+ * is always available.
+ */
+ for (i = 0; i < 6; i++) {
+ if (barcfg[i].pciadr != ~0 && barcfg[i].ahbadr != ~0) {
+ /* Target BARs must have the proper alignment */
+ ahbadr = barcfg[i].ahbadr;
+ pciadr = barcfg[i].pciadr;
+ bar_sz = ((pciadr - 1) & ~pciadr) + 1;
+ } else {
+ if (i == 0) {
+ /* Map main memory */
+ bar_sz = 0xf0000008; /* 256MB prefetchable */
+ ahbadr = 0xf0000000 & (u32)__pa(PAGE_ALIGN(
+ (unsigned long) &_end));
+ pciadr = ahbadr;
+ } else {
+ bar_sz = 0;
+ ahbadr = 0;
+ pciadr = 0;
+ }
+ }
+ grpci2_cfg_w32(priv, TGT, 0, capptr+CAP9_BARSIZE_OFS+i*4,
+ bar_sz);
+ grpci2_cfg_w32(priv, TGT, 0, PCI_BASE_ADDRESS_0+i*4, pciadr);
+ grpci2_cfg_w32(priv, TGT, 0, capptr+CAP9_BAR_OFS+i*4, ahbadr);
+ printk(KERN_INFO " TGT BAR[%d]: 0x%08x (PCI)-> 0x%08x\n",
+ i, pciadr, ahbadr);
+ }
+
+ /* set as bus master and enable pci memory responses */
+ grpci2_cfg_r32(priv, TGT, 0, PCI_COMMAND, &data);
+ data |= (PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER);
+ grpci2_cfg_w32(priv, TGT, 0, PCI_COMMAND, data);
+
+ /* Enable Error respone (CPU-TRAP) on illegal memory access. */
+ REGSTORE(regs->ctrl, CTRL_ER | CTRL_PE);
+}
+
+static irqreturn_t grpci2_jump_interrupt(int irq, void *arg)
+{
+ printk(KERN_ERR "GRPCI2: Jump IRQ happened\n");
+ return IRQ_NONE;
+}
+
+/* Handle GRPCI2 Error Interrupt */
+static irqreturn_t grpci2_err_interrupt(int irq, void *arg)
+{
+ struct grpci2_priv *priv = arg;
+ struct grpci2_regs __iomem *regs = priv->regs;
+ unsigned int status;
+
+ status = REGLOAD(regs->sts_cap);
+ if ((status & STS_ERR_IRQ) == 0)
+ return IRQ_NONE;
+
+ if (status & STS_IPARERR)
+ printk(KERN_ERR "GRPCI2: Parity Error\n");
+
+ if (status & STS_ITGTABRT)
+ printk(KERN_ERR "GRPCI2: Target Abort\n");
+
+ if (status & STS_IMSTABRT)
+ printk(KERN_ERR "GRPCI2: Master Abort\n");
+
+ if (status & STS_ISYSERR)
+ printk(KERN_ERR "GRPCI2: System Error\n");
+
+ /* Clear handled INT TYPE IRQs */
+ REGSTORE(regs->sts_cap, status & STS_ERR_IRQ);
+
+ return IRQ_HANDLED;
+}
+
+static int grpci2_of_probe(struct platform_device *ofdev)
+{
+ struct grpci2_regs __iomem *regs;
+ struct grpci2_priv *priv;
+ int err, i, len;
+ const int *tmp;
+ unsigned int capability;
+
+ if (grpci2priv) {
+ printk(KERN_ERR "GRPCI2: only one GRPCI2 core supported\n");
+ return -ENODEV;
+ }
+
+ if (ofdev->num_resources < 3) {
+ printk(KERN_ERR "GRPCI2: not enough APB/AHB resources\n");
+ return -EIO;
+ }
+
+ /* Find Device Address */
+ regs = of_ioremap(&ofdev->resource[0], 0,
+ resource_size(&ofdev->resource[0]),
+ "grlib-grpci2 regs");
+ if (regs == NULL) {
+ printk(KERN_ERR "GRPCI2: ioremap failed\n");
+ return -EIO;
+ }
+
+ /*
+ * Check that we're in Host Slot and that we can act as a Host Bridge
+ * and not only as target.
+ */
+ capability = REGLOAD(regs->sts_cap);
+ if ((capability & STS_HOST) || !(capability & STS_MST)) {
+ printk(KERN_INFO "GRPCI2: not in host system slot\n");
+ err = -EIO;
+ goto err1;
+ }
+
+ priv = grpci2priv = kzalloc(sizeof(struct grpci2_priv), GFP_KERNEL);
+ if (grpci2priv == NULL) {
+ err = -ENOMEM;
+ goto err1;
+ }
+ memset(grpci2priv, 0, sizeof(*grpci2priv));
+ priv->regs = regs;
+ priv->irq = ofdev->archdata.irqs[0]; /* BASE IRQ */
+ priv->irq_mode = (capability & STS_IRQMODE) >> STS_IRQMODE_BIT;
+
+ printk(KERN_INFO "GRPCI2: host found at %p, irq%d\n", regs, priv->irq);
+
+ /* Byte twisting should be made configurable from kernel command line */
+ priv->bt_enabled = 1;
+
+ /* Let user do custom Target BAR assignment */
+ tmp = of_get_property(ofdev->dev.of_node, "barcfg", &len);
+ if (tmp && (len == 2*4*6))
+ memcpy(priv->tgtbars, tmp, 2*4*6);
+ else
+ memset(priv->tgtbars, -1, 2*4*6);
+
+ /* Limit IRQ unmasking in irq_mode 2 and 3 */
+ tmp = of_get_property(ofdev->dev.of_node, "irq_mask", &len);
+ if (tmp && (len == 4))
+ priv->do_reset = *tmp;
+ else
+ priv->irq_mask = 0xf;
+
+ /* Optional PCI reset. Force PCI reset on startup */
+ tmp = of_get_property(ofdev->dev.of_node, "reset", &len);
+ if (tmp && (len == 4))
+ priv->do_reset = *tmp;
+ else
+ priv->do_reset = 0;
+
+ /* Find PCI Memory, I/O and Configuration Space Windows */
+ priv->pci_area = ofdev->resource[1].start;
+ priv->pci_area_end = ofdev->resource[1].end+1;
+ priv->pci_io = ofdev->resource[2].start;
+ priv->pci_conf = ofdev->resource[2].start + 0x10000;
+ priv->pci_conf_end = priv->pci_conf + 0x10000;
+ priv->pci_io_va = (unsigned long)ioremap(priv->pci_io, 0x10000);
+ if (!priv->pci_io_va) {
+ err = -EIO;
+ goto err2;
+ }
+
+ printk(KERN_INFO
+ "GRPCI2: MEMORY SPACE [0x%08lx - 0x%08lx]\n"
+ " I/O SPACE [0x%08lx - 0x%08lx]\n"
+ " CONFIG SPACE [0x%08lx - 0x%08lx]\n",
+ priv->pci_area, priv->pci_area_end-1,
+ priv->pci_io, priv->pci_conf-1,
+ priv->pci_conf, priv->pci_conf_end-1);
+
+ /*
+ * I/O Space resources in I/O Window mapped into Virtual Adr Space
+ * We never use low 4KB because some devices seem have problems using
+ * address 0.
+ */
+ memset(&priv->info.io_space, 0, sizeof(struct resource));
+ priv->info.io_space.name = "GRPCI2 PCI I/O Space";
+ priv->info.io_space.start = priv->pci_io_va + 0x1000;
+ priv->info.io_space.end = priv->pci_io_va + 0x10000 - 1;
+ priv->info.io_space.flags = IORESOURCE_IO;
+
+ /*
+ * GRPCI2 has no prefetchable memory, map everything as
+ * non-prefetchable memory
+ */
+ memset(&priv->info.mem_space, 0, sizeof(struct resource));
+ priv->info.mem_space.name = "GRPCI2 PCI MEM Space";
+ priv->info.mem_space.start = priv->pci_area;
+ priv->info.mem_space.end = priv->pci_area_end - 1;
+ priv->info.mem_space.flags = IORESOURCE_MEM;
+
+ if (request_resource(&iomem_resource, &priv->info.mem_space) < 0)
+ goto err3;
+ if (request_resource(&ioport_resource, &priv->info.io_space) < 0)
+ goto err4;
+
+ /* setup maximum supported PCI buses */
+ priv->info.busn.name = "GRPCI2 busn";
+ priv->info.busn.start = 0;
+ priv->info.busn.end = 255;
+
+ grpci2_hw_init(priv);
+
+ /*
+ * Get PCI Interrupt to System IRQ mapping and setup IRQ handling
+ * Error IRQ always on PCI INTA.
+ */
+ if (priv->irq_mode < 2) {
+ /* All PCI interrupts are shared using the same system IRQ */
+ leon_update_virq_handling(priv->irq, grpci2_pci_flow_irq,
+ "pcilvl", 0);
+
+ priv->irq_map[0] = grpci2_build_device_irq(1);
+ priv->irq_map[1] = grpci2_build_device_irq(2);
+ priv->irq_map[2] = grpci2_build_device_irq(3);
+ priv->irq_map[3] = grpci2_build_device_irq(4);
+
+ priv->virq_err = grpci2_build_device_irq(5);
+ if (priv->irq_mode & 1)
+ priv->virq_dma = ofdev->archdata.irqs[1];
+ else
+ priv->virq_dma = grpci2_build_device_irq(6);
+
+ /* Enable IRQs on LEON IRQ controller */
+ err = request_irq(priv->irq, grpci2_jump_interrupt, 0,
+ "GRPCI2_JUMP", priv);
+ if (err)
+ printk(KERN_ERR "GRPCI2: ERR IRQ request failed\n");
+ } else {
+ /* All PCI interrupts have an unique IRQ interrupt */
+ for (i = 0; i < 4; i++) {
+ /* Make LEON IRQ layer handle level IRQ by acking */
+ leon_update_virq_handling(ofdev->archdata.irqs[i],
+ handle_fasteoi_irq, "pcilvl",
+ 1);
+ priv->irq_map[i] = ofdev->archdata.irqs[i];
+ }
+ priv->virq_err = priv->irq_map[0];
+ if (priv->irq_mode & 1)
+ priv->virq_dma = ofdev->archdata.irqs[4];
+ else
+ priv->virq_dma = priv->irq_map[0];
+
+ /* Unmask all PCI interrupts, request_irq will not do that */
+ REGSTORE(regs->ctrl, REGLOAD(regs->ctrl)|(priv->irq_mask&0xf));
+ }
+
+ /* Setup IRQ handler for non-configuration space access errors */
+ err = request_irq(priv->virq_err, grpci2_err_interrupt, IRQF_SHARED,
+ "GRPCI2_ERR", priv);
+ if (err) {
+ printk(KERN_DEBUG "GRPCI2: ERR VIRQ request failed: %d\n", err);
+ goto err5;
+ }
+
+ /*
+ * Enable Error Interrupts. PCI interrupts are unmasked once request_irq
+ * is called by the PCI Device drivers
+ */
+ REGSTORE(regs->ctrl, REGLOAD(regs->ctrl) | CTRL_EI | CTRL_SI);
+
+ /* Init common layer and scan buses */
+ priv->info.ops = &grpci2_ops;
+ priv->info.map_irq = grpci2_map_irq;
+ leon_pci_init(ofdev, &priv->info);
+
+ return 0;
+
+err5:
+ release_resource(&priv->info.io_space);
+err4:
+ release_resource(&priv->info.mem_space);
+err3:
+ err = -ENOMEM;
+ iounmap((void __iomem *)priv->pci_io_va);
+err2:
+ kfree(priv);
+err1:
+ of_iounmap(&ofdev->resource[0], regs,
+ resource_size(&ofdev->resource[0]));
+ return err;
+}
+
+static struct of_device_id grpci2_of_match[] = {
+ {
+ .name = "GAISLER_GRPCI2",
+ },
+ {
+ .name = "01_07c",
+ },
+ {},
+};
+
+static struct platform_driver grpci2_of_driver = {
+ .driver = {
+ .name = "grpci2",
+ .owner = THIS_MODULE,
+ .of_match_table = grpci2_of_match,
+ },
+ .probe = grpci2_of_probe,
+};
+
+static int __init grpci2_init(void)
+{
+ return platform_driver_register(&grpci2_of_driver);
+}
+
+subsys_initcall(grpci2_init);
diff --git a/arch/sparc/kernel/leon_pmc.c b/arch/sparc/kernel/leon_pmc.c
new file mode 100644
index 00000000000..ddcf950282e
--- /dev/null
+++ b/arch/sparc/kernel/leon_pmc.c
@@ -0,0 +1,93 @@
+/* leon_pmc.c: LEON Power-down cpu_idle() handler
+ *
+ * Copyright (C) 2011 Daniel Hellstrom (daniel@gaisler.com) Aeroflex Gaisler AB
+ */
+
+#include <linux/init.h>
+#include <linux/pm.h>
+
+#include <asm/leon_amba.h>
+#include <asm/cpu_type.h>
+#include <asm/leon.h>
+#include <asm/processor.h>
+
+/* List of Systems that need fixup instructions around power-down instruction */
+static unsigned int pmc_leon_fixup_ids[] = {
+ AEROFLEX_UT699,
+ GAISLER_GR712RC,
+ LEON4_NEXTREME1,
+ 0
+};
+
+static int pmc_leon_need_fixup(void)
+{
+ unsigned int systemid = amba_system_id >> 16;
+ unsigned int *id;
+
+ id = &pmc_leon_fixup_ids[0];
+ while (*id != 0) {
+ if (*id == systemid)
+ return 1;
+ id++;
+ }
+
+ return 0;
+}
+
+/*
+ * CPU idle callback function for systems that need some extra handling
+ * See .../arch/sparc/kernel/process.c
+ */
+static void pmc_leon_idle_fixup(void)
+{
+ /* Prepare an address to a non-cachable region. APB is always
+ * none-cachable. One instruction is executed after the Sleep
+ * instruction, we make sure to read the bus and throw away the
+ * value by accessing a non-cachable area, also we make sure the
+ * MMU does not get a TLB miss here by using the MMU BYPASS ASI.
+ */
+ register unsigned int address = (unsigned int)leon3_irqctrl_regs;
+
+ /* Interrupts need to be enabled to not hang the CPU */
+ local_irq_enable();
+
+ __asm__ __volatile__ (
+ "wr %%g0, %%asr19\n"
+ "lda [%0] %1, %%g0\n"
+ :
+ : "r"(address), "i"(ASI_LEON_BYPASS));
+}
+
+/*
+ * CPU idle callback function
+ * See .../arch/sparc/kernel/process.c
+ */
+static void pmc_leon_idle(void)
+{
+ /* Interrupts need to be enabled to not hang the CPU */
+ local_irq_enable();
+
+ /* For systems without power-down, this will be no-op */
+ __asm__ __volatile__ ("wr %g0, %asr19\n\t");
+}
+
+/* Install LEON Power Down function */
+static int __init leon_pmc_install(void)
+{
+ if (sparc_cpu_model == sparc_leon) {
+ /* Assign power management IDLE handler */
+ if (pmc_leon_need_fixup())
+ sparc_idle = pmc_leon_idle_fixup;
+ else
+ sparc_idle = pmc_leon_idle;
+
+ printk(KERN_INFO "leon: power management initialized\n");
+ }
+
+ return 0;
+}
+
+/* This driver is not critical to the boot process, don't care
+ * if initialized late.
+ */
+late_initcall(leon_pmc_install);
diff --git a/arch/sparc/kernel/leon_smp.c b/arch/sparc/kernel/leon_smp.c
new file mode 100644
index 00000000000..018ef11f57d
--- /dev/null
+++ b/arch/sparc/kernel/leon_smp.c
@@ -0,0 +1,469 @@
+/* leon_smp.c: Sparc-Leon SMP support.
+ *
+ * based on sun4m_smp.c
+ * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 2009 Daniel Hellstrom (daniel@gaisler.com) Aeroflex Gaisler AB
+ * Copyright (C) 2009 Konrad Eisele (konrad@gaisler.com) Aeroflex Gaisler AB
+ */
+
+#include <asm/head.h>
+
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/threads.h>
+#include <linux/smp.h>
+#include <linux/interrupt.h>
+#include <linux/kernel_stat.h>
+#include <linux/of.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/mm.h>
+#include <linux/swap.h>
+#include <linux/profile.h>
+#include <linux/pm.h>
+#include <linux/delay.h>
+#include <linux/gfp.h>
+#include <linux/cpu.h>
+#include <linux/clockchips.h>
+
+#include <asm/cacheflush.h>
+#include <asm/tlbflush.h>
+
+#include <asm/ptrace.h>
+#include <linux/atomic.h>
+#include <asm/irq_regs.h>
+#include <asm/traps.h>
+
+#include <asm/delay.h>
+#include <asm/irq.h>
+#include <asm/page.h>
+#include <asm/pgalloc.h>
+#include <asm/pgtable.h>
+#include <asm/oplib.h>
+#include <asm/cpudata.h>
+#include <asm/asi.h>
+#include <asm/leon.h>
+#include <asm/leon_amba.h>
+#include <asm/timer.h>
+
+#include "kernel.h"
+
+#include "irq.h"
+
+extern ctxd_t *srmmu_ctx_table_phys;
+static int smp_processors_ready;
+extern volatile unsigned long cpu_callin_map[NR_CPUS];
+extern cpumask_t smp_commenced_mask;
+void leon_configure_cache_smp(void);
+static void leon_ipi_init(void);
+
+/* IRQ number of LEON IPIs */
+int leon_ipi_irq = LEON3_IRQ_IPI_DEFAULT;
+
+static inline unsigned long do_swap(volatile unsigned long *ptr,
+ unsigned long val)
+{
+ __asm__ __volatile__("swapa [%2] %3, %0\n\t" : "=&r"(val)
+ : "0"(val), "r"(ptr), "i"(ASI_LEON_DCACHE_MISS)
+ : "memory");
+ return val;
+}
+
+void leon_cpu_pre_starting(void *arg)
+{
+ leon_configure_cache_smp();
+}
+
+void leon_cpu_pre_online(void *arg)
+{
+ int cpuid = hard_smp_processor_id();
+
+ /* Allow master to continue. The master will then give us the
+ * go-ahead by setting the smp_commenced_mask and will wait without
+ * timeouts until our setup is completed fully (signified by
+ * our bit being set in the cpu_online_mask).
+ */
+ do_swap(&cpu_callin_map[cpuid], 1);
+
+ local_ops->cache_all();
+ local_ops->tlb_all();
+
+ /* Fix idle thread fields. */
+ __asm__ __volatile__("ld [%0], %%g6\n\t" : : "r"(&current_set[cpuid])
+ : "memory" /* paranoid */);
+
+ /* Attach to the address space of init_task. */
+ atomic_inc(&init_mm.mm_count);
+ current->active_mm = &init_mm;
+
+ while (!cpumask_test_cpu(cpuid, &smp_commenced_mask))
+ mb();
+}
+
+/*
+ * Cycle through the processors asking the PROM to start each one.
+ */
+
+extern struct linux_prom_registers smp_penguin_ctable;
+
+void leon_configure_cache_smp(void)
+{
+ unsigned long cfg = sparc_leon3_get_dcachecfg();
+ int me = smp_processor_id();
+
+ if (ASI_LEON3_SYSCTRL_CFG_SSIZE(cfg) > 4) {
+ printk(KERN_INFO "Note: SMP with snooping only works on 4k cache, found %dk(0x%x) on cpu %d, disabling caches\n",
+ (unsigned int)ASI_LEON3_SYSCTRL_CFG_SSIZE(cfg),
+ (unsigned int)cfg, (unsigned int)me);
+ sparc_leon3_disable_cache();
+ } else {
+ if (cfg & ASI_LEON3_SYSCTRL_CFG_SNOOPING) {
+ sparc_leon3_enable_snooping();
+ } else {
+ printk(KERN_INFO "Note: You have to enable snooping in the vhdl model cpu %d, disabling caches\n",
+ me);
+ sparc_leon3_disable_cache();
+ }
+ }
+
+ local_ops->cache_all();
+ local_ops->tlb_all();
+}
+
+static void leon_smp_setbroadcast(unsigned int mask)
+{
+ int broadcast =
+ ((LEON3_BYPASS_LOAD_PA(&(leon3_irqctrl_regs->mpstatus)) >>
+ LEON3_IRQMPSTATUS_BROADCAST) & 1);
+ if (!broadcast) {
+ prom_printf("######## !!!! The irqmp-ctrl must have broadcast enabled, smp wont work !!!!! ####### nr cpus: %d\n",
+ leon_smp_nrcpus());
+ if (leon_smp_nrcpus() > 1) {
+ BUG();
+ } else {
+ prom_printf("continue anyway\n");
+ return;
+ }
+ }
+ LEON_BYPASS_STORE_PA(&(leon3_irqctrl_regs->mpbroadcast), mask);
+}
+
+int leon_smp_nrcpus(void)
+{
+ int nrcpu =
+ ((LEON3_BYPASS_LOAD_PA(&(leon3_irqctrl_regs->mpstatus)) >>
+ LEON3_IRQMPSTATUS_CPUNR) & 0xf) + 1;
+ return nrcpu;
+}
+
+void __init leon_boot_cpus(void)
+{
+ int nrcpu = leon_smp_nrcpus();
+ int me = smp_processor_id();
+
+ /* Setup IPI */
+ leon_ipi_init();
+
+ printk(KERN_INFO "%d:(%d:%d) cpus mpirq at 0x%x\n", (unsigned int)me,
+ (unsigned int)nrcpu, (unsigned int)NR_CPUS,
+ (unsigned int)&(leon3_irqctrl_regs->mpstatus));
+
+ leon_enable_irq_cpu(LEON3_IRQ_CROSS_CALL, me);
+ leon_enable_irq_cpu(LEON3_IRQ_TICKER, me);
+ leon_enable_irq_cpu(leon_ipi_irq, me);
+
+ leon_smp_setbroadcast(1 << LEON3_IRQ_TICKER);
+
+ leon_configure_cache_smp();
+ local_ops->cache_all();
+
+}
+
+int leon_boot_one_cpu(int i, struct task_struct *idle)
+{
+ int timeout;
+
+ current_set[i] = task_thread_info(idle);
+
+ /* See trampoline.S:leon_smp_cpu_startup for details...
+ * Initialize the contexts table
+ * Since the call to prom_startcpu() trashes the structure,
+ * we need to re-initialize it for each cpu
+ */
+ smp_penguin_ctable.which_io = 0;
+ smp_penguin_ctable.phys_addr = (unsigned int)srmmu_ctx_table_phys;
+ smp_penguin_ctable.reg_size = 0;
+
+ /* whirrr, whirrr, whirrrrrrrrr... */
+ printk(KERN_INFO "Starting CPU %d : (irqmp: 0x%x)\n", (unsigned int)i,
+ (unsigned int)&leon3_irqctrl_regs->mpstatus);
+ local_ops->cache_all();
+
+ /* Make sure all IRQs are of from the start for this new CPU */
+ LEON_BYPASS_STORE_PA(&leon3_irqctrl_regs->mask[i], 0);
+
+ /* Wake one CPU */
+ LEON_BYPASS_STORE_PA(&(leon3_irqctrl_regs->mpstatus), 1 << i);
+
+ /* wheee... it's going... */
+ for (timeout = 0; timeout < 10000; timeout++) {
+ if (cpu_callin_map[i])
+ break;
+ udelay(200);
+ }
+ printk(KERN_INFO "Started CPU %d\n", (unsigned int)i);
+
+ if (!(cpu_callin_map[i])) {
+ printk(KERN_ERR "Processor %d is stuck.\n", i);
+ return -ENODEV;
+ } else {
+ leon_enable_irq_cpu(LEON3_IRQ_CROSS_CALL, i);
+ leon_enable_irq_cpu(LEON3_IRQ_TICKER, i);
+ leon_enable_irq_cpu(leon_ipi_irq, i);
+ }
+
+ local_ops->cache_all();
+ return 0;
+}
+
+void __init leon_smp_done(void)
+{
+
+ int i, first;
+ int *prev;
+
+ /* setup cpu list for irq rotation */
+ first = 0;
+ prev = &first;
+ for (i = 0; i < NR_CPUS; i++) {
+ if (cpu_online(i)) {
+ *prev = i;
+ prev = &cpu_data(i).next;
+ }
+ }
+ *prev = first;
+ local_ops->cache_all();
+
+ /* Free unneeded trap tables */
+ if (!cpu_present(1)) {
+ free_reserved_page(virt_to_page(&trapbase_cpu1));
+ }
+ if (!cpu_present(2)) {
+ free_reserved_page(virt_to_page(&trapbase_cpu2));
+ }
+ if (!cpu_present(3)) {
+ free_reserved_page(virt_to_page(&trapbase_cpu3));
+ }
+ /* Ok, they are spinning and ready to go. */
+ smp_processors_ready = 1;
+
+}
+
+struct leon_ipi_work {
+ int single;
+ int msk;
+ int resched;
+};
+
+static DEFINE_PER_CPU_SHARED_ALIGNED(struct leon_ipi_work, leon_ipi_work);
+
+/* Initialize IPIs on the LEON, in order to save IRQ resources only one IRQ
+ * is used for all three types of IPIs.
+ */
+static void __init leon_ipi_init(void)
+{
+ int cpu, len;
+ struct leon_ipi_work *work;
+ struct property *pp;
+ struct device_node *rootnp;
+ struct tt_entry *trap_table;
+ unsigned long flags;
+
+ /* Find IPI IRQ or stick with default value */
+ rootnp = of_find_node_by_path("/ambapp0");
+ if (rootnp) {
+ pp = of_find_property(rootnp, "ipi_num", &len);
+ if (pp && (*(int *)pp->value))
+ leon_ipi_irq = *(int *)pp->value;
+ }
+ printk(KERN_INFO "leon: SMP IPIs at IRQ %d\n", leon_ipi_irq);
+
+ /* Adjust so that we jump directly to smpleon_ipi */
+ local_irq_save(flags);
+ trap_table = &sparc_ttable[SP_TRAP_IRQ1 + (leon_ipi_irq - 1)];
+ trap_table->inst_three += smpleon_ipi - real_irq_entry;
+ local_ops->cache_all();
+ local_irq_restore(flags);
+
+ for_each_possible_cpu(cpu) {
+ work = &per_cpu(leon_ipi_work, cpu);
+ work->single = work->msk = work->resched = 0;
+ }
+}
+
+static void leon_send_ipi(int cpu, int level)
+{
+ unsigned long mask;
+ mask = leon_get_irqmask(level);
+ LEON3_BYPASS_STORE_PA(&leon3_irqctrl_regs->force[cpu], mask);
+}
+
+static void leon_ipi_single(int cpu)
+{
+ struct leon_ipi_work *work = &per_cpu(leon_ipi_work, cpu);
+
+ /* Mark work */
+ work->single = 1;
+
+ /* Generate IRQ on the CPU */
+ leon_send_ipi(cpu, leon_ipi_irq);
+}
+
+static void leon_ipi_mask_one(int cpu)
+{
+ struct leon_ipi_work *work = &per_cpu(leon_ipi_work, cpu);
+
+ /* Mark work */
+ work->msk = 1;
+
+ /* Generate IRQ on the CPU */
+ leon_send_ipi(cpu, leon_ipi_irq);
+}
+
+static void leon_ipi_resched(int cpu)
+{
+ struct leon_ipi_work *work = &per_cpu(leon_ipi_work, cpu);
+
+ /* Mark work */
+ work->resched = 1;
+
+ /* Generate IRQ on the CPU (any IRQ will cause resched) */
+ leon_send_ipi(cpu, leon_ipi_irq);
+}
+
+void leonsmp_ipi_interrupt(void)
+{
+ struct leon_ipi_work *work = &__get_cpu_var(leon_ipi_work);
+
+ if (work->single) {
+ work->single = 0;
+ smp_call_function_single_interrupt();
+ }
+ if (work->msk) {
+ work->msk = 0;
+ smp_call_function_interrupt();
+ }
+ if (work->resched) {
+ work->resched = 0;
+ smp_resched_interrupt();
+ }
+}
+
+static struct smp_funcall {
+ smpfunc_t func;
+ unsigned long arg1;
+ unsigned long arg2;
+ unsigned long arg3;
+ unsigned long arg4;
+ unsigned long arg5;
+ unsigned long processors_in[NR_CPUS]; /* Set when ipi entered. */
+ unsigned long processors_out[NR_CPUS]; /* Set when ipi exited. */
+} ccall_info;
+
+static DEFINE_SPINLOCK(cross_call_lock);
+
+/* Cross calls must be serialized, at least currently. */
+static void leon_cross_call(smpfunc_t func, cpumask_t mask, unsigned long arg1,
+ unsigned long arg2, unsigned long arg3,
+ unsigned long arg4)
+{
+ if (smp_processors_ready) {
+ register int high = NR_CPUS - 1;
+ unsigned long flags;
+
+ spin_lock_irqsave(&cross_call_lock, flags);
+
+ {
+ /* If you make changes here, make sure gcc generates proper code... */
+ register smpfunc_t f asm("i0") = func;
+ register unsigned long a1 asm("i1") = arg1;
+ register unsigned long a2 asm("i2") = arg2;
+ register unsigned long a3 asm("i3") = arg3;
+ register unsigned long a4 asm("i4") = arg4;
+ register unsigned long a5 asm("i5") = 0;
+
+ __asm__ __volatile__("std %0, [%6]\n\t"
+ "std %2, [%6 + 8]\n\t"
+ "std %4, [%6 + 16]\n\t" : :
+ "r"(f), "r"(a1), "r"(a2), "r"(a3),
+ "r"(a4), "r"(a5),
+ "r"(&ccall_info.func));
+ }
+
+ /* Init receive/complete mapping, plus fire the IPI's off. */
+ {
+ register int i;
+
+ cpumask_clear_cpu(smp_processor_id(), &mask);
+ cpumask_and(&mask, cpu_online_mask, &mask);
+ for (i = 0; i <= high; i++) {
+ if (cpumask_test_cpu(i, &mask)) {
+ ccall_info.processors_in[i] = 0;
+ ccall_info.processors_out[i] = 0;
+ leon_send_ipi(i, LEON3_IRQ_CROSS_CALL);
+
+ }
+ }
+ }
+
+ {
+ register int i;
+
+ i = 0;
+ do {
+ if (!cpumask_test_cpu(i, &mask))
+ continue;
+
+ while (!ccall_info.processors_in[i])
+ barrier();
+ } while (++i <= high);
+
+ i = 0;
+ do {
+ if (!cpumask_test_cpu(i, &mask))
+ continue;
+
+ while (!ccall_info.processors_out[i])
+ barrier();
+ } while (++i <= high);
+ }
+
+ spin_unlock_irqrestore(&cross_call_lock, flags);
+ }
+}
+
+/* Running cross calls. */
+void leon_cross_call_irq(void)
+{
+ int i = smp_processor_id();
+
+ ccall_info.processors_in[i] = 1;
+ ccall_info.func(ccall_info.arg1, ccall_info.arg2, ccall_info.arg3,
+ ccall_info.arg4, ccall_info.arg5);
+ ccall_info.processors_out[i] = 1;
+}
+
+static const struct sparc32_ipi_ops leon_ipi_ops = {
+ .cross_call = leon_cross_call,
+ .resched = leon_ipi_resched,
+ .single = leon_ipi_single,
+ .mask_one = leon_ipi_mask_one,
+};
+
+void __init leon_init_smp(void)
+{
+ /* Patch ipi15 trap table */
+ t_nmi[1] = t_nmi[1] + (linux_trap_ipi15_leon - linux_trap_ipi15_sun4m);
+
+ sparc32_ipi_ops = &leon_ipi_ops;
+}
diff --git a/arch/sparc/kernel/mdesc.c b/arch/sparc/kernel/mdesc.c
new file mode 100644
index 00000000000..a1a4400d402
--- /dev/null
+++ b/arch/sparc/kernel/mdesc.c
@@ -0,0 +1,974 @@
+/* mdesc.c: Sun4V machine description handling.
+ *
+ * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
+ */
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/memblock.h>
+#include <linux/log2.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/mm.h>
+#include <linux/miscdevice.h>
+#include <linux/bootmem.h>
+#include <linux/export.h>
+
+#include <asm/cpudata.h>
+#include <asm/hypervisor.h>
+#include <asm/mdesc.h>
+#include <asm/prom.h>
+#include <asm/uaccess.h>
+#include <asm/oplib.h>
+#include <asm/smp.h>
+
+/* Unlike the OBP device tree, the machine description is a full-on
+ * DAG. An arbitrary number of ARCs are possible from one
+ * node to other nodes and thus we can't use the OBP device_node
+ * data structure to represent these nodes inside of the kernel.
+ *
+ * Actually, it isn't even a DAG, because there are back pointers
+ * which create cycles in the graph.
+ *
+ * mdesc_hdr and mdesc_elem describe the layout of the data structure
+ * we get from the Hypervisor.
+ */
+struct mdesc_hdr {
+ u32 version; /* Transport version */
+ u32 node_sz; /* node block size */
+ u32 name_sz; /* name block size */
+ u32 data_sz; /* data block size */
+} __attribute__((aligned(16)));
+
+struct mdesc_elem {
+ u8 tag;
+#define MD_LIST_END 0x00
+#define MD_NODE 0x4e
+#define MD_NODE_END 0x45
+#define MD_NOOP 0x20
+#define MD_PROP_ARC 0x61
+#define MD_PROP_VAL 0x76
+#define MD_PROP_STR 0x73
+#define MD_PROP_DATA 0x64
+ u8 name_len;
+ u16 resv;
+ u32 name_offset;
+ union {
+ struct {
+ u32 data_len;
+ u32 data_offset;
+ } data;
+ u64 val;
+ } d;
+};
+
+struct mdesc_mem_ops {
+ struct mdesc_handle *(*alloc)(unsigned int mdesc_size);
+ void (*free)(struct mdesc_handle *handle);
+};
+
+struct mdesc_handle {
+ struct list_head list;
+ struct mdesc_mem_ops *mops;
+ void *self_base;
+ atomic_t refcnt;
+ unsigned int handle_size;
+ struct mdesc_hdr mdesc;
+};
+
+static void mdesc_handle_init(struct mdesc_handle *hp,
+ unsigned int handle_size,
+ void *base)
+{
+ BUG_ON(((unsigned long)&hp->mdesc) & (16UL - 1));
+
+ memset(hp, 0, handle_size);
+ INIT_LIST_HEAD(&hp->list);
+ hp->self_base = base;
+ atomic_set(&hp->refcnt, 1);
+ hp->handle_size = handle_size;
+}
+
+static struct mdesc_handle * __init mdesc_memblock_alloc(unsigned int mdesc_size)
+{
+ unsigned int handle_size, alloc_size;
+ struct mdesc_handle *hp;
+ unsigned long paddr;
+
+ handle_size = (sizeof(struct mdesc_handle) -
+ sizeof(struct mdesc_hdr) +
+ mdesc_size);
+ alloc_size = PAGE_ALIGN(handle_size);
+
+ paddr = memblock_alloc(alloc_size, PAGE_SIZE);
+
+ hp = NULL;
+ if (paddr) {
+ hp = __va(paddr);
+ mdesc_handle_init(hp, handle_size, hp);
+ }
+ return hp;
+}
+
+static void __init mdesc_memblock_free(struct mdesc_handle *hp)
+{
+ unsigned int alloc_size;
+ unsigned long start;
+
+ BUG_ON(atomic_read(&hp->refcnt) != 0);
+ BUG_ON(!list_empty(&hp->list));
+
+ alloc_size = PAGE_ALIGN(hp->handle_size);
+ start = __pa(hp);
+ free_bootmem_late(start, alloc_size);
+}
+
+static struct mdesc_mem_ops memblock_mdesc_ops = {
+ .alloc = mdesc_memblock_alloc,
+ .free = mdesc_memblock_free,
+};
+
+static struct mdesc_handle *mdesc_kmalloc(unsigned int mdesc_size)
+{
+ unsigned int handle_size;
+ void *base;
+
+ handle_size = (sizeof(struct mdesc_handle) -
+ sizeof(struct mdesc_hdr) +
+ mdesc_size);
+
+ base = kmalloc(handle_size + 15, GFP_KERNEL | __GFP_NOFAIL);
+ if (base) {
+ struct mdesc_handle *hp;
+ unsigned long addr;
+
+ addr = (unsigned long)base;
+ addr = (addr + 15UL) & ~15UL;
+ hp = (struct mdesc_handle *) addr;
+
+ mdesc_handle_init(hp, handle_size, base);
+ return hp;
+ }
+
+ return NULL;
+}
+
+static void mdesc_kfree(struct mdesc_handle *hp)
+{
+ BUG_ON(atomic_read(&hp->refcnt) != 0);
+ BUG_ON(!list_empty(&hp->list));
+
+ kfree(hp->self_base);
+}
+
+static struct mdesc_mem_ops kmalloc_mdesc_memops = {
+ .alloc = mdesc_kmalloc,
+ .free = mdesc_kfree,
+};
+
+static struct mdesc_handle *mdesc_alloc(unsigned int mdesc_size,
+ struct mdesc_mem_ops *mops)
+{
+ struct mdesc_handle *hp = mops->alloc(mdesc_size);
+
+ if (hp)
+ hp->mops = mops;
+
+ return hp;
+}
+
+static void mdesc_free(struct mdesc_handle *hp)
+{
+ hp->mops->free(hp);
+}
+
+static struct mdesc_handle *cur_mdesc;
+static LIST_HEAD(mdesc_zombie_list);
+static DEFINE_SPINLOCK(mdesc_lock);
+
+struct mdesc_handle *mdesc_grab(void)
+{
+ struct mdesc_handle *hp;
+ unsigned long flags;
+
+ spin_lock_irqsave(&mdesc_lock, flags);
+ hp = cur_mdesc;
+ if (hp)
+ atomic_inc(&hp->refcnt);
+ spin_unlock_irqrestore(&mdesc_lock, flags);
+
+ return hp;
+}
+EXPORT_SYMBOL(mdesc_grab);
+
+void mdesc_release(struct mdesc_handle *hp)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&mdesc_lock, flags);
+ if (atomic_dec_and_test(&hp->refcnt)) {
+ list_del_init(&hp->list);
+ hp->mops->free(hp);
+ }
+ spin_unlock_irqrestore(&mdesc_lock, flags);
+}
+EXPORT_SYMBOL(mdesc_release);
+
+static DEFINE_MUTEX(mdesc_mutex);
+static struct mdesc_notifier_client *client_list;
+
+void mdesc_register_notifier(struct mdesc_notifier_client *client)
+{
+ u64 node;
+
+ mutex_lock(&mdesc_mutex);
+ client->next = client_list;
+ client_list = client;
+
+ mdesc_for_each_node_by_name(cur_mdesc, node, client->node_name)
+ client->add(cur_mdesc, node);
+
+ mutex_unlock(&mdesc_mutex);
+}
+
+static const u64 *parent_cfg_handle(struct mdesc_handle *hp, u64 node)
+{
+ const u64 *id;
+ u64 a;
+
+ id = NULL;
+ mdesc_for_each_arc(a, hp, node, MDESC_ARC_TYPE_BACK) {
+ u64 target;
+
+ target = mdesc_arc_target(hp, a);
+ id = mdesc_get_property(hp, target,
+ "cfg-handle", NULL);
+ if (id)
+ break;
+ }
+
+ return id;
+}
+
+/* Run 'func' on nodes which are in A but not in B. */
+static void invoke_on_missing(const char *name,
+ struct mdesc_handle *a,
+ struct mdesc_handle *b,
+ void (*func)(struct mdesc_handle *, u64))
+{
+ u64 node;
+
+ mdesc_for_each_node_by_name(a, node, name) {
+ int found = 0, is_vdc_port = 0;
+ const char *name_prop;
+ const u64 *id;
+ u64 fnode;
+
+ name_prop = mdesc_get_property(a, node, "name", NULL);
+ if (name_prop && !strcmp(name_prop, "vdc-port")) {
+ is_vdc_port = 1;
+ id = parent_cfg_handle(a, node);
+ } else
+ id = mdesc_get_property(a, node, "id", NULL);
+
+ if (!id) {
+ printk(KERN_ERR "MD: Cannot find ID for %s node.\n",
+ (name_prop ? name_prop : name));
+ continue;
+ }
+
+ mdesc_for_each_node_by_name(b, fnode, name) {
+ const u64 *fid;
+
+ if (is_vdc_port) {
+ name_prop = mdesc_get_property(b, fnode,
+ "name", NULL);
+ if (!name_prop ||
+ strcmp(name_prop, "vdc-port"))
+ continue;
+ fid = parent_cfg_handle(b, fnode);
+ if (!fid) {
+ printk(KERN_ERR "MD: Cannot find ID "
+ "for vdc-port node.\n");
+ continue;
+ }
+ } else
+ fid = mdesc_get_property(b, fnode,
+ "id", NULL);
+
+ if (*id == *fid) {
+ found = 1;
+ break;
+ }
+ }
+ if (!found)
+ func(a, node);
+ }
+}
+
+static void notify_one(struct mdesc_notifier_client *p,
+ struct mdesc_handle *old_hp,
+ struct mdesc_handle *new_hp)
+{
+ invoke_on_missing(p->node_name, old_hp, new_hp, p->remove);
+ invoke_on_missing(p->node_name, new_hp, old_hp, p->add);
+}
+
+static void mdesc_notify_clients(struct mdesc_handle *old_hp,
+ struct mdesc_handle *new_hp)
+{
+ struct mdesc_notifier_client *p = client_list;
+
+ while (p) {
+ notify_one(p, old_hp, new_hp);
+ p = p->next;
+ }
+}
+
+void mdesc_update(void)
+{
+ unsigned long len, real_len, status;
+ struct mdesc_handle *hp, *orig_hp;
+ unsigned long flags;
+
+ mutex_lock(&mdesc_mutex);
+
+ (void) sun4v_mach_desc(0UL, 0UL, &len);
+
+ hp = mdesc_alloc(len, &kmalloc_mdesc_memops);
+ if (!hp) {
+ printk(KERN_ERR "MD: mdesc alloc fails\n");
+ goto out;
+ }
+
+ status = sun4v_mach_desc(__pa(&hp->mdesc), len, &real_len);
+ if (status != HV_EOK || real_len > len) {
+ printk(KERN_ERR "MD: mdesc reread fails with %lu\n",
+ status);
+ atomic_dec(&hp->refcnt);
+ mdesc_free(hp);
+ goto out;
+ }
+
+ spin_lock_irqsave(&mdesc_lock, flags);
+ orig_hp = cur_mdesc;
+ cur_mdesc = hp;
+ spin_unlock_irqrestore(&mdesc_lock, flags);
+
+ mdesc_notify_clients(orig_hp, hp);
+
+ spin_lock_irqsave(&mdesc_lock, flags);
+ if (atomic_dec_and_test(&orig_hp->refcnt))
+ mdesc_free(orig_hp);
+ else
+ list_add(&orig_hp->list, &mdesc_zombie_list);
+ spin_unlock_irqrestore(&mdesc_lock, flags);
+
+out:
+ mutex_unlock(&mdesc_mutex);
+}
+
+static struct mdesc_elem *node_block(struct mdesc_hdr *mdesc)
+{
+ return (struct mdesc_elem *) (mdesc + 1);
+}
+
+static void *name_block(struct mdesc_hdr *mdesc)
+{
+ return ((void *) node_block(mdesc)) + mdesc->node_sz;
+}
+
+static void *data_block(struct mdesc_hdr *mdesc)
+{
+ return ((void *) name_block(mdesc)) + mdesc->name_sz;
+}
+
+u64 mdesc_node_by_name(struct mdesc_handle *hp,
+ u64 from_node, const char *name)
+{
+ struct mdesc_elem *ep = node_block(&hp->mdesc);
+ const char *names = name_block(&hp->mdesc);
+ u64 last_node = hp->mdesc.node_sz / 16;
+ u64 ret;
+
+ if (from_node == MDESC_NODE_NULL) {
+ ret = from_node = 0;
+ } else if (from_node >= last_node) {
+ return MDESC_NODE_NULL;
+ } else {
+ ret = ep[from_node].d.val;
+ }
+
+ while (ret < last_node) {
+ if (ep[ret].tag != MD_NODE)
+ return MDESC_NODE_NULL;
+ if (!strcmp(names + ep[ret].name_offset, name))
+ break;
+ ret = ep[ret].d.val;
+ }
+ if (ret >= last_node)
+ ret = MDESC_NODE_NULL;
+ return ret;
+}
+EXPORT_SYMBOL(mdesc_node_by_name);
+
+const void *mdesc_get_property(struct mdesc_handle *hp, u64 node,
+ const char *name, int *lenp)
+{
+ const char *names = name_block(&hp->mdesc);
+ u64 last_node = hp->mdesc.node_sz / 16;
+ void *data = data_block(&hp->mdesc);
+ struct mdesc_elem *ep;
+
+ if (node == MDESC_NODE_NULL || node >= last_node)
+ return NULL;
+
+ ep = node_block(&hp->mdesc) + node;
+ ep++;
+ for (; ep->tag != MD_NODE_END; ep++) {
+ void *val = NULL;
+ int len = 0;
+
+ switch (ep->tag) {
+ case MD_PROP_VAL:
+ val = &ep->d.val;
+ len = 8;
+ break;
+
+ case MD_PROP_STR:
+ case MD_PROP_DATA:
+ val = data + ep->d.data.data_offset;
+ len = ep->d.data.data_len;
+ break;
+
+ default:
+ break;
+ }
+ if (!val)
+ continue;
+
+ if (!strcmp(names + ep->name_offset, name)) {
+ if (lenp)
+ *lenp = len;
+ return val;
+ }
+ }
+
+ return NULL;
+}
+EXPORT_SYMBOL(mdesc_get_property);
+
+u64 mdesc_next_arc(struct mdesc_handle *hp, u64 from, const char *arc_type)
+{
+ struct mdesc_elem *ep, *base = node_block(&hp->mdesc);
+ const char *names = name_block(&hp->mdesc);
+ u64 last_node = hp->mdesc.node_sz / 16;
+
+ if (from == MDESC_NODE_NULL || from >= last_node)
+ return MDESC_NODE_NULL;
+
+ ep = base + from;
+
+ ep++;
+ for (; ep->tag != MD_NODE_END; ep++) {
+ if (ep->tag != MD_PROP_ARC)
+ continue;
+
+ if (strcmp(names + ep->name_offset, arc_type))
+ continue;
+
+ return ep - base;
+ }
+
+ return MDESC_NODE_NULL;
+}
+EXPORT_SYMBOL(mdesc_next_arc);
+
+u64 mdesc_arc_target(struct mdesc_handle *hp, u64 arc)
+{
+ struct mdesc_elem *ep, *base = node_block(&hp->mdesc);
+
+ ep = base + arc;
+
+ return ep->d.val;
+}
+EXPORT_SYMBOL(mdesc_arc_target);
+
+const char *mdesc_node_name(struct mdesc_handle *hp, u64 node)
+{
+ struct mdesc_elem *ep, *base = node_block(&hp->mdesc);
+ const char *names = name_block(&hp->mdesc);
+ u64 last_node = hp->mdesc.node_sz / 16;
+
+ if (node == MDESC_NODE_NULL || node >= last_node)
+ return NULL;
+
+ ep = base + node;
+ if (ep->tag != MD_NODE)
+ return NULL;
+
+ return names + ep->name_offset;
+}
+EXPORT_SYMBOL(mdesc_node_name);
+
+static u64 max_cpus = 64;
+
+static void __init report_platform_properties(void)
+{
+ struct mdesc_handle *hp = mdesc_grab();
+ u64 pn = mdesc_node_by_name(hp, MDESC_NODE_NULL, "platform");
+ const char *s;
+ const u64 *v;
+
+ if (pn == MDESC_NODE_NULL) {
+ prom_printf("No platform node in machine-description.\n");
+ prom_halt();
+ }
+
+ s = mdesc_get_property(hp, pn, "banner-name", NULL);
+ printk("PLATFORM: banner-name [%s]\n", s);
+ s = mdesc_get_property(hp, pn, "name", NULL);
+ printk("PLATFORM: name [%s]\n", s);
+
+ v = mdesc_get_property(hp, pn, "hostid", NULL);
+ if (v)
+ printk("PLATFORM: hostid [%08llx]\n", *v);
+ v = mdesc_get_property(hp, pn, "serial#", NULL);
+ if (v)
+ printk("PLATFORM: serial# [%08llx]\n", *v);
+ v = mdesc_get_property(hp, pn, "stick-frequency", NULL);
+ printk("PLATFORM: stick-frequency [%08llx]\n", *v);
+ v = mdesc_get_property(hp, pn, "mac-address", NULL);
+ if (v)
+ printk("PLATFORM: mac-address [%llx]\n", *v);
+ v = mdesc_get_property(hp, pn, "watchdog-resolution", NULL);
+ if (v)
+ printk("PLATFORM: watchdog-resolution [%llu ms]\n", *v);
+ v = mdesc_get_property(hp, pn, "watchdog-max-timeout", NULL);
+ if (v)
+ printk("PLATFORM: watchdog-max-timeout [%llu ms]\n", *v);
+ v = mdesc_get_property(hp, pn, "max-cpus", NULL);
+ if (v) {
+ max_cpus = *v;
+ printk("PLATFORM: max-cpus [%llu]\n", max_cpus);
+ }
+
+#ifdef CONFIG_SMP
+ {
+ int max_cpu, i;
+
+ if (v) {
+ max_cpu = *v;
+ if (max_cpu > NR_CPUS)
+ max_cpu = NR_CPUS;
+ } else {
+ max_cpu = NR_CPUS;
+ }
+ for (i = 0; i < max_cpu; i++)
+ set_cpu_possible(i, true);
+ }
+#endif
+
+ mdesc_release(hp);
+}
+
+static void fill_in_one_cache(cpuinfo_sparc *c, struct mdesc_handle *hp, u64 mp)
+{
+ const u64 *level = mdesc_get_property(hp, mp, "level", NULL);
+ const u64 *size = mdesc_get_property(hp, mp, "size", NULL);
+ const u64 *line_size = mdesc_get_property(hp, mp, "line-size", NULL);
+ const char *type;
+ int type_len;
+
+ type = mdesc_get_property(hp, mp, "type", &type_len);
+
+ switch (*level) {
+ case 1:
+ if (of_find_in_proplist(type, "instn", type_len)) {
+ c->icache_size = *size;
+ c->icache_line_size = *line_size;
+ } else if (of_find_in_proplist(type, "data", type_len)) {
+ c->dcache_size = *size;
+ c->dcache_line_size = *line_size;
+ }
+ break;
+
+ case 2:
+ c->ecache_size = *size;
+ c->ecache_line_size = *line_size;
+ break;
+
+ default:
+ break;
+ }
+
+ if (*level == 1) {
+ u64 a;
+
+ mdesc_for_each_arc(a, hp, mp, MDESC_ARC_TYPE_FWD) {
+ u64 target = mdesc_arc_target(hp, a);
+ const char *name = mdesc_node_name(hp, target);
+
+ if (!strcmp(name, "cache"))
+ fill_in_one_cache(c, hp, target);
+ }
+ }
+}
+
+static void mark_core_ids(struct mdesc_handle *hp, u64 mp, int core_id)
+{
+ u64 a;
+
+ mdesc_for_each_arc(a, hp, mp, MDESC_ARC_TYPE_BACK) {
+ u64 t = mdesc_arc_target(hp, a);
+ const char *name;
+ const u64 *id;
+
+ name = mdesc_node_name(hp, t);
+ if (!strcmp(name, "cpu")) {
+ id = mdesc_get_property(hp, t, "id", NULL);
+ if (*id < NR_CPUS)
+ cpu_data(*id).core_id = core_id;
+ } else {
+ u64 j;
+
+ mdesc_for_each_arc(j, hp, t, MDESC_ARC_TYPE_BACK) {
+ u64 n = mdesc_arc_target(hp, j);
+ const char *n_name;
+
+ n_name = mdesc_node_name(hp, n);
+ if (strcmp(n_name, "cpu"))
+ continue;
+
+ id = mdesc_get_property(hp, n, "id", NULL);
+ if (*id < NR_CPUS)
+ cpu_data(*id).core_id = core_id;
+ }
+ }
+ }
+}
+
+static void set_core_ids(struct mdesc_handle *hp)
+{
+ int idx;
+ u64 mp;
+
+ idx = 1;
+ mdesc_for_each_node_by_name(hp, mp, "cache") {
+ const u64 *level;
+ const char *type;
+ int len;
+
+ level = mdesc_get_property(hp, mp, "level", NULL);
+ if (*level != 1)
+ continue;
+
+ type = mdesc_get_property(hp, mp, "type", &len);
+ if (!of_find_in_proplist(type, "instn", len))
+ continue;
+
+ mark_core_ids(hp, mp, idx);
+
+ idx++;
+ }
+}
+
+static void mark_proc_ids(struct mdesc_handle *hp, u64 mp, int proc_id)
+{
+ u64 a;
+
+ mdesc_for_each_arc(a, hp, mp, MDESC_ARC_TYPE_BACK) {
+ u64 t = mdesc_arc_target(hp, a);
+ const char *name;
+ const u64 *id;
+
+ name = mdesc_node_name(hp, t);
+ if (strcmp(name, "cpu"))
+ continue;
+
+ id = mdesc_get_property(hp, t, "id", NULL);
+ if (*id < NR_CPUS)
+ cpu_data(*id).proc_id = proc_id;
+ }
+}
+
+static void __set_proc_ids(struct mdesc_handle *hp, const char *exec_unit_name)
+{
+ int idx;
+ u64 mp;
+
+ idx = 0;
+ mdesc_for_each_node_by_name(hp, mp, exec_unit_name) {
+ const char *type;
+ int len;
+
+ type = mdesc_get_property(hp, mp, "type", &len);
+ if (!of_find_in_proplist(type, "int", len) &&
+ !of_find_in_proplist(type, "integer", len))
+ continue;
+
+ mark_proc_ids(hp, mp, idx);
+
+ idx++;
+ }
+}
+
+static void set_proc_ids(struct mdesc_handle *hp)
+{
+ __set_proc_ids(hp, "exec_unit");
+ __set_proc_ids(hp, "exec-unit");
+}
+
+static void get_one_mondo_bits(const u64 *p, unsigned int *mask,
+ unsigned long def, unsigned long max)
+{
+ u64 val;
+
+ if (!p)
+ goto use_default;
+ val = *p;
+
+ if (!val || val >= 64)
+ goto use_default;
+
+ if (val > max)
+ val = max;
+
+ *mask = ((1U << val) * 64U) - 1U;
+ return;
+
+use_default:
+ *mask = ((1U << def) * 64U) - 1U;
+}
+
+static void get_mondo_data(struct mdesc_handle *hp, u64 mp,
+ struct trap_per_cpu *tb)
+{
+ static int printed;
+ const u64 *val;
+
+ val = mdesc_get_property(hp, mp, "q-cpu-mondo-#bits", NULL);
+ get_one_mondo_bits(val, &tb->cpu_mondo_qmask, 7, ilog2(max_cpus * 2));
+
+ val = mdesc_get_property(hp, mp, "q-dev-mondo-#bits", NULL);
+ get_one_mondo_bits(val, &tb->dev_mondo_qmask, 7, 8);
+
+ val = mdesc_get_property(hp, mp, "q-resumable-#bits", NULL);
+ get_one_mondo_bits(val, &tb->resum_qmask, 6, 7);
+
+ val = mdesc_get_property(hp, mp, "q-nonresumable-#bits", NULL);
+ get_one_mondo_bits(val, &tb->nonresum_qmask, 2, 2);
+ if (!printed++) {
+ pr_info("SUN4V: Mondo queue sizes "
+ "[cpu(%u) dev(%u) r(%u) nr(%u)]\n",
+ tb->cpu_mondo_qmask + 1,
+ tb->dev_mondo_qmask + 1,
+ tb->resum_qmask + 1,
+ tb->nonresum_qmask + 1);
+ }
+}
+
+static void *mdesc_iterate_over_cpus(void *(*func)(struct mdesc_handle *, u64, int, void *), void *arg, cpumask_t *mask)
+{
+ struct mdesc_handle *hp = mdesc_grab();
+ void *ret = NULL;
+ u64 mp;
+
+ mdesc_for_each_node_by_name(hp, mp, "cpu") {
+ const u64 *id = mdesc_get_property(hp, mp, "id", NULL);
+ int cpuid = *id;
+
+#ifdef CONFIG_SMP
+ if (cpuid >= NR_CPUS) {
+ printk(KERN_WARNING "Ignoring CPU %d which is "
+ ">= NR_CPUS (%d)\n",
+ cpuid, NR_CPUS);
+ continue;
+ }
+ if (!cpumask_test_cpu(cpuid, mask))
+ continue;
+#endif
+
+ ret = func(hp, mp, cpuid, arg);
+ if (ret)
+ goto out;
+ }
+out:
+ mdesc_release(hp);
+ return ret;
+}
+
+static void *record_one_cpu(struct mdesc_handle *hp, u64 mp, int cpuid,
+ void *arg)
+{
+ ncpus_probed++;
+#ifdef CONFIG_SMP
+ set_cpu_present(cpuid, true);
+#endif
+ return NULL;
+}
+
+void mdesc_populate_present_mask(cpumask_t *mask)
+{
+ if (tlb_type != hypervisor)
+ return;
+
+ ncpus_probed = 0;
+ mdesc_iterate_over_cpus(record_one_cpu, NULL, mask);
+}
+
+static void * __init check_one_pgsz(struct mdesc_handle *hp, u64 mp, int cpuid, void *arg)
+{
+ const u64 *pgsz_prop = mdesc_get_property(hp, mp, "mmu-page-size-list", NULL);
+ unsigned long *pgsz_mask = arg;
+ u64 val;
+
+ val = (HV_PGSZ_MASK_8K | HV_PGSZ_MASK_64K |
+ HV_PGSZ_MASK_512K | HV_PGSZ_MASK_4MB);
+ if (pgsz_prop)
+ val = *pgsz_prop;
+
+ if (!*pgsz_mask)
+ *pgsz_mask = val;
+ else
+ *pgsz_mask &= val;
+ return NULL;
+}
+
+void __init mdesc_get_page_sizes(cpumask_t *mask, unsigned long *pgsz_mask)
+{
+ *pgsz_mask = 0;
+ mdesc_iterate_over_cpus(check_one_pgsz, pgsz_mask, mask);
+}
+
+static void *fill_in_one_cpu(struct mdesc_handle *hp, u64 mp, int cpuid,
+ void *arg)
+{
+ const u64 *cfreq = mdesc_get_property(hp, mp, "clock-frequency", NULL);
+ struct trap_per_cpu *tb;
+ cpuinfo_sparc *c;
+ u64 a;
+
+#ifndef CONFIG_SMP
+ /* On uniprocessor we only want the values for the
+ * real physical cpu the kernel booted onto, however
+ * cpu_data() only has one entry at index 0.
+ */
+ if (cpuid != real_hard_smp_processor_id())
+ return NULL;
+ cpuid = 0;
+#endif
+
+ c = &cpu_data(cpuid);
+ c->clock_tick = *cfreq;
+
+ tb = &trap_block[cpuid];
+ get_mondo_data(hp, mp, tb);
+
+ mdesc_for_each_arc(a, hp, mp, MDESC_ARC_TYPE_FWD) {
+ u64 j, t = mdesc_arc_target(hp, a);
+ const char *t_name;
+
+ t_name = mdesc_node_name(hp, t);
+ if (!strcmp(t_name, "cache")) {
+ fill_in_one_cache(c, hp, t);
+ continue;
+ }
+
+ mdesc_for_each_arc(j, hp, t, MDESC_ARC_TYPE_FWD) {
+ u64 n = mdesc_arc_target(hp, j);
+ const char *n_name;
+
+ n_name = mdesc_node_name(hp, n);
+ if (!strcmp(n_name, "cache"))
+ fill_in_one_cache(c, hp, n);
+ }
+ }
+
+ c->core_id = 0;
+ c->proc_id = -1;
+
+ return NULL;
+}
+
+void mdesc_fill_in_cpu_data(cpumask_t *mask)
+{
+ struct mdesc_handle *hp;
+
+ mdesc_iterate_over_cpus(fill_in_one_cpu, NULL, mask);
+
+ hp = mdesc_grab();
+
+ set_core_ids(hp);
+ set_proc_ids(hp);
+
+ mdesc_release(hp);
+
+ smp_fill_in_sib_core_maps();
+}
+
+static ssize_t mdesc_read(struct file *file, char __user *buf,
+ size_t len, loff_t *offp)
+{
+ struct mdesc_handle *hp = mdesc_grab();
+ int err;
+
+ if (!hp)
+ return -ENODEV;
+
+ err = hp->handle_size;
+ if (len < hp->handle_size)
+ err = -EMSGSIZE;
+ else if (copy_to_user(buf, &hp->mdesc, hp->handle_size))
+ err = -EFAULT;
+ mdesc_release(hp);
+
+ return err;
+}
+
+static const struct file_operations mdesc_fops = {
+ .read = mdesc_read,
+ .owner = THIS_MODULE,
+ .llseek = noop_llseek,
+};
+
+static struct miscdevice mdesc_misc = {
+ .minor = MISC_DYNAMIC_MINOR,
+ .name = "mdesc",
+ .fops = &mdesc_fops,
+};
+
+static int __init mdesc_misc_init(void)
+{
+ return misc_register(&mdesc_misc);
+}
+
+__initcall(mdesc_misc_init);
+
+void __init sun4v_mdesc_init(void)
+{
+ struct mdesc_handle *hp;
+ unsigned long len, real_len, status;
+
+ (void) sun4v_mach_desc(0UL, 0UL, &len);
+
+ printk("MDESC: Size is %lu bytes.\n", len);
+
+ hp = mdesc_alloc(len, &memblock_mdesc_ops);
+ if (hp == NULL) {
+ prom_printf("MDESC: alloc of %lu bytes failed.\n", len);
+ prom_halt();
+ }
+
+ status = sun4v_mach_desc(__pa(&hp->mdesc), len, &real_len);
+ if (status != HV_EOK || real_len > len) {
+ prom_printf("sun4v_mach_desc fails, err(%lu), "
+ "len(%lu), real_len(%lu)\n",
+ status, len, real_len);
+ mdesc_free(hp);
+ prom_halt();
+ }
+
+ cur_mdesc = hp;
+
+ report_platform_properties();
+}
diff --git a/arch/sparc/kernel/misctrap.S b/arch/sparc/kernel/misctrap.S
new file mode 100644
index 00000000000..753b4f031bf
--- /dev/null
+++ b/arch/sparc/kernel/misctrap.S
@@ -0,0 +1,97 @@
+#ifdef CONFIG_KGDB
+ .globl arch_kgdb_breakpoint
+ .type arch_kgdb_breakpoint,#function
+arch_kgdb_breakpoint:
+ ta 0x72
+ retl
+ nop
+ .size arch_kgdb_breakpoint,.-arch_kgdb_breakpoint
+#endif
+
+ .type __do_privact,#function
+__do_privact:
+ mov TLB_SFSR, %g3
+ stxa %g0, [%g3] ASI_DMMU ! Clear FaultValid bit
+ membar #Sync
+ sethi %hi(109f), %g7
+ ba,pt %xcc, etrap
+109: or %g7, %lo(109b), %g7
+ call do_privact
+ add %sp, PTREGS_OFF, %o0
+ ba,pt %xcc, rtrap
+ nop
+ .size __do_privact,.-__do_privact
+
+ .type do_mna,#function
+do_mna:
+ rdpr %tl, %g3
+ cmp %g3, 1
+
+ /* Setup %g4/%g5 now as they are used in the
+ * winfixup code.
+ */
+ mov TLB_SFSR, %g3
+ mov DMMU_SFAR, %g4
+ ldxa [%g4] ASI_DMMU, %g4
+ ldxa [%g3] ASI_DMMU, %g5
+ stxa %g0, [%g3] ASI_DMMU ! Clear FaultValid bit
+ membar #Sync
+ bgu,pn %icc, winfix_mna
+ rdpr %tpc, %g3
+
+1: sethi %hi(109f), %g7
+ ba,pt %xcc, etrap
+109: or %g7, %lo(109b), %g7
+ mov %l4, %o1
+ mov %l5, %o2
+ call mem_address_unaligned
+ add %sp, PTREGS_OFF, %o0
+ ba,pt %xcc, rtrap
+ nop
+ .size do_mna,.-do_mna
+
+ .type do_lddfmna,#function
+do_lddfmna:
+ sethi %hi(109f), %g7
+ mov TLB_SFSR, %g4
+ ldxa [%g4] ASI_DMMU, %g5
+ stxa %g0, [%g4] ASI_DMMU ! Clear FaultValid bit
+ membar #Sync
+ mov DMMU_SFAR, %g4
+ ldxa [%g4] ASI_DMMU, %g4
+ ba,pt %xcc, etrap
+109: or %g7, %lo(109b), %g7
+ mov %l4, %o1
+ mov %l5, %o2
+ call handle_lddfmna
+ add %sp, PTREGS_OFF, %o0
+ ba,pt %xcc, rtrap
+ nop
+ .size do_lddfmna,.-do_lddfmna
+
+ .type do_stdfmna,#function
+do_stdfmna:
+ sethi %hi(109f), %g7
+ mov TLB_SFSR, %g4
+ ldxa [%g4] ASI_DMMU, %g5
+ stxa %g0, [%g4] ASI_DMMU ! Clear FaultValid bit
+ membar #Sync
+ mov DMMU_SFAR, %g4
+ ldxa [%g4] ASI_DMMU, %g4
+ ba,pt %xcc, etrap
+109: or %g7, %lo(109b), %g7
+ mov %l4, %o1
+ mov %l5, %o2
+ call handle_stdfmna
+ add %sp, PTREGS_OFF, %o0
+ ba,pt %xcc, rtrap
+ nop
+ .size do_stdfmna,.-do_stdfmna
+
+ .type breakpoint_trap,#function
+breakpoint_trap:
+ call sparc_breakpoint
+ add %sp, PTREGS_OFF, %o0
+ ba,pt %xcc, rtrap
+ nop
+ .size breakpoint_trap,.-breakpoint_trap
diff --git a/arch/sparc/kernel/module.c b/arch/sparc/kernel/module.c
index 598682f31eb..97655e0fd24 100644
--- a/arch/sparc/kernel/module.c
+++ b/arch/sparc/kernel/module.c
@@ -1,4 +1,4 @@
-/* Kernel module help for sparc32.
+/* Kernel module help for sparc64.
*
* Copyright (C) 2001 Rusty Russell.
* Copyright (C) 2002 David S. Miller.
@@ -9,44 +9,55 @@
#include <linux/elf.h>
#include <linux/vmalloc.h>
#include <linux/fs.h>
+#include <linux/gfp.h>
#include <linux/string.h>
#include <linux/ctype.h>
+#include <linux/mm.h>
+
+#include <asm/processor.h>
+#include <asm/spitfire.h>
+#include <asm/cacheflush.h>
+
+#include "entry.h"
+
+#ifdef CONFIG_SPARC64
+
+#include <linux/jump_label.h>
+
+static void *module_map(unsigned long size)
+{
+ if (PAGE_ALIGN(size) > MODULES_LEN)
+ return NULL;
+ return __vmalloc_node_range(size, 1, MODULES_VADDR, MODULES_END,
+ GFP_KERNEL, PAGE_KERNEL, NUMA_NO_NODE,
+ __builtin_return_address(0));
+}
+#else
+static void *module_map(unsigned long size)
+{
+ return vmalloc(size);
+}
+#endif /* CONFIG_SPARC64 */
void *module_alloc(unsigned long size)
{
void *ret;
- /* We handle the zero case fine, unlike vmalloc */
- if (size == 0)
- return NULL;
-
- ret = vmalloc(size);
- if (!ret)
- ret = ERR_PTR(-ENOMEM);
- else
+ ret = module_map(size);
+ if (ret)
memset(ret, 0, size);
return ret;
}
-/* Free memory returned from module_core_alloc/module_init_alloc */
-void module_free(struct module *mod, void *module_region)
-{
- vfree(module_region);
- /* FIXME: If module_region == mod->init_region, trim exception
- table entries. */
-}
-
-/* Make generic code ignore STT_REGISTER dummy undefined symbols,
- * and replace references to .func with _Func
- */
+/* Make generic code ignore STT_REGISTER dummy undefined symbols. */
int module_frob_arch_sections(Elf_Ehdr *hdr,
Elf_Shdr *sechdrs,
char *secstrings,
struct module *mod)
{
unsigned int symidx;
- Elf32_Sym *sym;
+ Elf_Sym *sym;
char *strtab;
int i;
@@ -56,62 +67,79 @@ int module_frob_arch_sections(Elf_Ehdr *hdr,
return -ENOEXEC;
}
}
- sym = (Elf32_Sym *)sechdrs[symidx].sh_addr;
+ sym = (Elf_Sym *)sechdrs[symidx].sh_addr;
strtab = (char *)sechdrs[sechdrs[symidx].sh_link].sh_addr;
for (i = 1; i < sechdrs[symidx].sh_size / sizeof(Elf_Sym); i++) {
if (sym[i].st_shndx == SHN_UNDEF) {
- if (ELF32_ST_TYPE(sym[i].st_info) == STT_REGISTER)
+ if (ELF_ST_TYPE(sym[i].st_info) == STT_REGISTER)
sym[i].st_shndx = SHN_ABS;
- else {
- char *name = strtab + sym[i].st_name;
- if (name[0] == '.') {
- name[0] = '_';
- name[1] = toupper(name[1]);
- }
- }
}
}
return 0;
}
-int apply_relocate(Elf32_Shdr *sechdrs,
- const char *strtab,
- unsigned int symindex,
- unsigned int relsec,
- struct module *me)
-{
- printk(KERN_ERR "module %s: non-ADD RELOCATION unsupported\n",
- me->name);
- return -ENOEXEC;
-}
-
-int apply_relocate_add(Elf32_Shdr *sechdrs,
+int apply_relocate_add(Elf_Shdr *sechdrs,
const char *strtab,
unsigned int symindex,
unsigned int relsec,
struct module *me)
{
unsigned int i;
- Elf32_Rela *rel = (void *)sechdrs[relsec].sh_addr;
- Elf32_Sym *sym;
+ Elf_Rela *rel = (void *)sechdrs[relsec].sh_addr;
+ Elf_Sym *sym;
u8 *location;
u32 *loc32;
for (i = 0; i < sechdrs[relsec].sh_size / sizeof(*rel); i++) {
- Elf32_Addr v;
+ Elf_Addr v;
/* This is where to make the change */
location = (u8 *)sechdrs[sechdrs[relsec].sh_info].sh_addr
+ rel[i].r_offset;
loc32 = (u32 *) location;
+
+#ifdef CONFIG_SPARC64
+ BUG_ON(((u64)location >> (u64)32) != (u64)0);
+#endif /* CONFIG_SPARC64 */
+
/* This is the symbol it is referring to. Note that all
undefined symbols have been resolved. */
- sym = (Elf32_Sym *)sechdrs[symindex].sh_addr
- + ELF32_R_SYM(rel[i].r_info);
+ sym = (Elf_Sym *)sechdrs[symindex].sh_addr
+ + ELF_R_SYM(rel[i].r_info);
v = sym->st_value + rel[i].r_addend;
- switch (ELF32_R_TYPE(rel[i].r_info)) {
+ switch (ELF_R_TYPE(rel[i].r_info) & 0xff) {
+ case R_SPARC_DISP32:
+ v -= (Elf_Addr) location;
+ *loc32 = v;
+ break;
+#ifdef CONFIG_SPARC64
+ case R_SPARC_64:
+ location[0] = v >> 56;
+ location[1] = v >> 48;
+ location[2] = v >> 40;
+ location[3] = v >> 32;
+ location[4] = v >> 24;
+ location[5] = v >> 16;
+ location[6] = v >> 8;
+ location[7] = v >> 0;
+ break;
+
+ case R_SPARC_WDISP19:
+ v -= (Elf_Addr) location;
+ *loc32 = (*loc32 & ~0x7ffff) |
+ ((v >> 2) & 0x7ffff);
+ break;
+
+ case R_SPARC_OLO10:
+ *loc32 = (*loc32 & ~0x1fff) |
+ (((v & 0x3ff) +
+ (ELF_R_TYPE(rel[i].r_info) >> 8))
+ & 0x1fff);
+ break;
+#endif /* CONFIG_SPARC64 */
+
case R_SPARC_32:
case R_SPARC_UA32:
location[0] = v >> 24;
@@ -121,13 +149,13 @@ int apply_relocate_add(Elf32_Shdr *sechdrs,
break;
case R_SPARC_WDISP30:
- v -= (Elf32_Addr) location;
+ v -= (Elf_Addr) location;
*loc32 = (*loc32 & ~0x3fffffff) |
((v >> 2) & 0x3fffffff);
break;
case R_SPARC_WDISP22:
- v -= (Elf32_Addr) location;
+ v -= (Elf_Addr) location;
*loc32 = (*loc32 & ~0x3fffff) |
((v >> 2) & 0x3fffff);
break;
@@ -144,20 +172,56 @@ int apply_relocate_add(Elf32_Shdr *sechdrs,
default:
printk(KERN_ERR "module %s: Unknown relocation: %x\n",
me->name,
- (int) (ELF32_R_TYPE(rel[i].r_info) & 0xff));
+ (int) (ELF_R_TYPE(rel[i].r_info) & 0xff));
return -ENOEXEC;
- };
+ }
}
return 0;
}
+#ifdef CONFIG_SPARC64
+static void do_patch_sections(const Elf_Ehdr *hdr,
+ const Elf_Shdr *sechdrs)
+{
+ const Elf_Shdr *s, *sun4v_1insn = NULL, *sun4v_2insn = NULL;
+ char *secstrings = (void *)hdr + sechdrs[hdr->e_shstrndx].sh_offset;
+
+ for (s = sechdrs; s < sechdrs + hdr->e_shnum; s++) {
+ if (!strcmp(".sun4v_1insn_patch", secstrings + s->sh_name))
+ sun4v_1insn = s;
+ if (!strcmp(".sun4v_2insn_patch", secstrings + s->sh_name))
+ sun4v_2insn = s;
+ }
+
+ if (sun4v_1insn && tlb_type == hypervisor) {
+ void *p = (void *) sun4v_1insn->sh_addr;
+ sun4v_patch_1insn_range(p, p + sun4v_1insn->sh_size);
+ }
+ if (sun4v_2insn && tlb_type == hypervisor) {
+ void *p = (void *) sun4v_2insn->sh_addr;
+ sun4v_patch_2insn_range(p, p + sun4v_2insn->sh_size);
+ }
+}
+
int module_finalize(const Elf_Ehdr *hdr,
const Elf_Shdr *sechdrs,
struct module *me)
{
- return 0;
-}
+ /* make jump label nops */
+ jump_label_apply_nops(me);
-void module_arch_cleanup(struct module *mod)
-{
+ do_patch_sections(hdr, sechdrs);
+
+ /* Cheetah's I-cache is fully coherent. */
+ if (tlb_type == spitfire) {
+ unsigned long va;
+
+ flushw_all();
+ for (va = 0; va < (PAGE_SIZE << 1); va += 32)
+ spitfire_put_icache_tag(va, 0x0);
+ __asm__ __volatile__("flush %g6");
+ }
+
+ return 0;
}
+#endif /* CONFIG_SPARC64 */
diff --git a/arch/sparc/kernel/muldiv.c b/arch/sparc/kernel/muldiv.c
deleted file mode 100644
index 37b9a494223..00000000000
--- a/arch/sparc/kernel/muldiv.c
+++ /dev/null
@@ -1,240 +0,0 @@
-/* $Id: muldiv.c,v 1.5 1997/12/15 20:07:20 ecd Exp $
- * muldiv.c: Hardware multiply/division illegal instruction trap
- * for sun4c/sun4 (which do not have those instructions)
- *
- * Copyright (C) 1996 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
- * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
- *
- * 2004-12-25 Krzysztof Helt (krzysztof.h1@wp.pl)
- * - fixed registers constrains in inline assembly declarations
- */
-
-#include <linux/kernel.h>
-#include <linux/sched.h>
-#include <linux/mm.h>
-#include <asm/ptrace.h>
-#include <asm/processor.h>
-#include <asm/system.h>
-#include <asm/uaccess.h>
-
-/* #define DEBUG_MULDIV */
-
-static inline int has_imm13(int insn)
-{
- return (insn & 0x2000);
-}
-
-static inline int is_foocc(int insn)
-{
- return (insn & 0x800000);
-}
-
-static inline int sign_extend_imm13(int imm)
-{
- return imm << 19 >> 19;
-}
-
-static inline void advance(struct pt_regs *regs)
-{
- regs->pc = regs->npc;
- regs->npc += 4;
-}
-
-static inline void maybe_flush_windows(unsigned int rs1, unsigned int rs2,
- unsigned int rd)
-{
- if(rs2 >= 16 || rs1 >= 16 || rd >= 16) {
- /* Wheee... */
- __asm__ __volatile__("save %sp, -0x40, %sp\n\t"
- "save %sp, -0x40, %sp\n\t"
- "save %sp, -0x40, %sp\n\t"
- "save %sp, -0x40, %sp\n\t"
- "save %sp, -0x40, %sp\n\t"
- "save %sp, -0x40, %sp\n\t"
- "save %sp, -0x40, %sp\n\t"
- "restore; restore; restore; restore;\n\t"
- "restore; restore; restore;\n\t");
- }
-}
-
-#define fetch_reg(reg, regs) ({ \
- struct reg_window __user *win; \
- register unsigned long ret; \
- \
- if (!(reg)) ret = 0; \
- else if ((reg) < 16) { \
- ret = regs->u_regs[(reg)]; \
- } else { \
- /* Ho hum, the slightly complicated case. */ \
- win = (struct reg_window __user *)regs->u_regs[UREG_FP];\
- if (get_user (ret, &win->locals[(reg) - 16])) return -1;\
- } \
- ret; \
-})
-
-static inline int
-store_reg(unsigned int result, unsigned int reg, struct pt_regs *regs)
-{
- struct reg_window __user *win;
-
- if (!reg)
- return 0;
- if (reg < 16) {
- regs->u_regs[reg] = result;
- return 0;
- } else {
- /* need to use put_user() in this case: */
- win = (struct reg_window __user *) regs->u_regs[UREG_FP];
- return (put_user(result, &win->locals[reg - 16]));
- }
-}
-
-extern void handle_hw_divzero (struct pt_regs *regs, unsigned long pc,
- unsigned long npc, unsigned long psr);
-
-/* Should return 0 if mul/div emulation succeeded and SIGILL should
- * not be issued.
- */
-int do_user_muldiv(struct pt_regs *regs, unsigned long pc)
-{
- unsigned int insn;
- int inst;
- unsigned int rs1, rs2, rdv;
-
- if (!pc)
- return -1; /* This happens to often, I think */
- if (get_user (insn, (unsigned int __user *)pc))
- return -1;
- if ((insn & 0xc1400000) != 0x80400000)
- return -1;
- inst = ((insn >> 19) & 0xf);
- if ((inst & 0xe) != 10 && (inst & 0xe) != 14)
- return -1;
-
- /* Now we know we have to do something with umul, smul, udiv or sdiv */
- rs1 = (insn >> 14) & 0x1f;
- rs2 = insn & 0x1f;
- rdv = (insn >> 25) & 0x1f;
- if (has_imm13(insn)) {
- maybe_flush_windows(rs1, 0, rdv);
- rs2 = sign_extend_imm13(insn);
- } else {
- maybe_flush_windows(rs1, rs2, rdv);
- rs2 = fetch_reg(rs2, regs);
- }
- rs1 = fetch_reg(rs1, regs);
- switch (inst) {
- case 10: /* umul */
-#ifdef DEBUG_MULDIV
- printk ("unsigned muldiv: 0x%x * 0x%x = ", rs1, rs2);
-#endif
- __asm__ __volatile__ ("\n\t"
- "mov %0, %%o0\n\t"
- "call .umul\n\t"
- " mov %1, %%o1\n\t"
- "mov %%o0, %0\n\t"
- "mov %%o1, %1\n\t"
- : "=r" (rs1), "=r" (rs2)
- : "0" (rs1), "1" (rs2)
- : "o0", "o1", "o2", "o3", "o4", "o5", "o7", "cc");
-#ifdef DEBUG_MULDIV
- printk ("0x%x%08x\n", rs2, rs1);
-#endif
- if (store_reg(rs1, rdv, regs))
- return -1;
- regs->y = rs2;
- break;
- case 11: /* smul */
-#ifdef DEBUG_MULDIV
- printk ("signed muldiv: 0x%x * 0x%x = ", rs1, rs2);
-#endif
- __asm__ __volatile__ ("\n\t"
- "mov %0, %%o0\n\t"
- "call .mul\n\t"
- " mov %1, %%o1\n\t"
- "mov %%o0, %0\n\t"
- "mov %%o1, %1\n\t"
- : "=r" (rs1), "=r" (rs2)
- : "0" (rs1), "1" (rs2)
- : "o0", "o1", "o2", "o3", "o4", "o5", "o7", "cc");
-#ifdef DEBUG_MULDIV
- printk ("0x%x%08x\n", rs2, rs1);
-#endif
- if (store_reg(rs1, rdv, regs))
- return -1;
- regs->y = rs2;
- break;
- case 14: /* udiv */
-#ifdef DEBUG_MULDIV
- printk ("unsigned muldiv: 0x%x%08x / 0x%x = ", regs->y, rs1, rs2);
-#endif
- if (!rs2) {
-#ifdef DEBUG_MULDIV
- printk ("DIVISION BY ZERO\n");
-#endif
- handle_hw_divzero (regs, pc, regs->npc, regs->psr);
- return 0;
- }
- __asm__ __volatile__ ("\n\t"
- "mov %2, %%o0\n\t"
- "mov %0, %%o1\n\t"
- "mov %%g0, %%o2\n\t"
- "call __udivdi3\n\t"
- " mov %1, %%o3\n\t"
- "mov %%o1, %0\n\t"
- "mov %%o0, %1\n\t"
- : "=r" (rs1), "=r" (rs2)
- : "r" (regs->y), "0" (rs1), "1" (rs2)
- : "o0", "o1", "o2", "o3", "o4", "o5", "o7",
- "g1", "g2", "g3", "cc");
-#ifdef DEBUG_MULDIV
- printk ("0x%x\n", rs1);
-#endif
- if (store_reg(rs1, rdv, regs))
- return -1;
- break;
- case 15: /* sdiv */
-#ifdef DEBUG_MULDIV
- printk ("signed muldiv: 0x%x%08x / 0x%x = ", regs->y, rs1, rs2);
-#endif
- if (!rs2) {
-#ifdef DEBUG_MULDIV
- printk ("DIVISION BY ZERO\n");
-#endif
- handle_hw_divzero (regs, pc, regs->npc, regs->psr);
- return 0;
- }
- __asm__ __volatile__ ("\n\t"
- "mov %2, %%o0\n\t"
- "mov %0, %%o1\n\t"
- "mov %%g0, %%o2\n\t"
- "call __divdi3\n\t"
- " mov %1, %%o3\n\t"
- "mov %%o1, %0\n\t"
- "mov %%o0, %1\n\t"
- : "=r" (rs1), "=r" (rs2)
- : "r" (regs->y), "0" (rs1), "1" (rs2)
- : "o0", "o1", "o2", "o3", "o4", "o5", "o7",
- "g1", "g2", "g3", "cc");
-#ifdef DEBUG_MULDIV
- printk ("0x%x\n", rs1);
-#endif
- if (store_reg(rs1, rdv, regs))
- return -1;
- break;
- }
- if (is_foocc (insn)) {
- regs->psr &= ~PSR_ICC;
- if ((inst & 0xe) == 14) {
- /* ?div */
- if (rs2) regs->psr |= PSR_V;
- }
- if (!rs1) regs->psr |= PSR_Z;
- if (((int)rs1) < 0) regs->psr |= PSR_N;
-#ifdef DEBUG_MULDIV
- printk ("psr muldiv: %08x\n", regs->psr);
-#endif
- }
- advance(regs);
- return 0;
-}
diff --git a/arch/sparc/kernel/nmi.c b/arch/sparc/kernel/nmi.c
new file mode 100644
index 00000000000..33709455691
--- /dev/null
+++ b/arch/sparc/kernel/nmi.c
@@ -0,0 +1,273 @@
+/* Pseudo NMI support on sparc64 systems.
+ *
+ * Copyright (C) 2009 David S. Miller <davem@davemloft.net>
+ *
+ * The NMI watchdog support and infrastructure is based almost
+ * entirely upon the x86 NMI support code.
+ */
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/init.h>
+#include <linux/percpu.h>
+#include <linux/nmi.h>
+#include <linux/export.h>
+#include <linux/kprobes.h>
+#include <linux/kernel_stat.h>
+#include <linux/reboot.h>
+#include <linux/slab.h>
+#include <linux/kdebug.h>
+#include <linux/delay.h>
+#include <linux/smp.h>
+
+#include <asm/perf_event.h>
+#include <asm/ptrace.h>
+#include <asm/pcr.h>
+
+#include "kstack.h"
+
+/* We don't have a real NMI on sparc64, but we can fake one
+ * up using profiling counter overflow interrupts and interrupt
+ * levels.
+ *
+ * The profile overflow interrupts at level 15, so we use
+ * level 14 as our IRQ off level.
+ */
+
+static int panic_on_timeout;
+
+/* nmi_active:
+ * >0: the NMI watchdog is active, but can be disabled
+ * <0: the NMI watchdog has not been set up, and cannot be enabled
+ * 0: the NMI watchdog is disabled, but can be enabled
+ */
+atomic_t nmi_active = ATOMIC_INIT(0); /* oprofile uses this */
+EXPORT_SYMBOL(nmi_active);
+
+static unsigned int nmi_hz = HZ;
+static DEFINE_PER_CPU(short, wd_enabled);
+static int endflag __initdata;
+
+static DEFINE_PER_CPU(unsigned int, last_irq_sum);
+static DEFINE_PER_CPU(long, alert_counter);
+static DEFINE_PER_CPU(int, nmi_touch);
+
+void touch_nmi_watchdog(void)
+{
+ if (atomic_read(&nmi_active)) {
+ int cpu;
+
+ for_each_present_cpu(cpu) {
+ if (per_cpu(nmi_touch, cpu) != 1)
+ per_cpu(nmi_touch, cpu) = 1;
+ }
+ }
+
+ touch_softlockup_watchdog();
+}
+EXPORT_SYMBOL(touch_nmi_watchdog);
+
+static void die_nmi(const char *str, struct pt_regs *regs, int do_panic)
+{
+ int this_cpu = smp_processor_id();
+
+ if (notify_die(DIE_NMIWATCHDOG, str, regs, 0,
+ pt_regs_trap_type(regs), SIGINT) == NOTIFY_STOP)
+ return;
+
+ if (do_panic || panic_on_oops)
+ panic("Watchdog detected hard LOCKUP on cpu %d", this_cpu);
+ else
+ WARN(1, "Watchdog detected hard LOCKUP on cpu %d", this_cpu);
+}
+
+notrace __kprobes void perfctr_irq(int irq, struct pt_regs *regs)
+{
+ unsigned int sum, touched = 0;
+ void *orig_sp;
+
+ clear_softint(1 << irq);
+
+ local_cpu_data().__nmi_count++;
+
+ nmi_enter();
+
+ orig_sp = set_hardirq_stack();
+
+ if (notify_die(DIE_NMI, "nmi", regs, 0,
+ pt_regs_trap_type(regs), SIGINT) == NOTIFY_STOP)
+ touched = 1;
+ else
+ pcr_ops->write_pcr(0, pcr_ops->pcr_nmi_disable);
+
+ sum = local_cpu_data().irq0_irqs;
+ if (__get_cpu_var(nmi_touch)) {
+ __get_cpu_var(nmi_touch) = 0;
+ touched = 1;
+ }
+ if (!touched && __get_cpu_var(last_irq_sum) == sum) {
+ __this_cpu_inc(alert_counter);
+ if (__this_cpu_read(alert_counter) == 30 * nmi_hz)
+ die_nmi("BUG: NMI Watchdog detected LOCKUP",
+ regs, panic_on_timeout);
+ } else {
+ __get_cpu_var(last_irq_sum) = sum;
+ __this_cpu_write(alert_counter, 0);
+ }
+ if (__get_cpu_var(wd_enabled)) {
+ pcr_ops->write_pic(0, pcr_ops->nmi_picl_value(nmi_hz));
+ pcr_ops->write_pcr(0, pcr_ops->pcr_nmi_enable);
+ }
+
+ restore_hardirq_stack(orig_sp);
+
+ nmi_exit();
+}
+
+static inline unsigned int get_nmi_count(int cpu)
+{
+ return cpu_data(cpu).__nmi_count;
+}
+
+static __init void nmi_cpu_busy(void *data)
+{
+ local_irq_enable_in_hardirq();
+ while (endflag == 0)
+ mb();
+}
+
+static void report_broken_nmi(int cpu, int *prev_nmi_count)
+{
+ printk(KERN_CONT "\n");
+
+ printk(KERN_WARNING
+ "WARNING: CPU#%d: NMI appears to be stuck (%d->%d)!\n",
+ cpu, prev_nmi_count[cpu], get_nmi_count(cpu));
+
+ printk(KERN_WARNING
+ "Please report this to bugzilla.kernel.org,\n");
+ printk(KERN_WARNING
+ "and attach the output of the 'dmesg' command.\n");
+
+ per_cpu(wd_enabled, cpu) = 0;
+ atomic_dec(&nmi_active);
+}
+
+void stop_nmi_watchdog(void *unused)
+{
+ pcr_ops->write_pcr(0, pcr_ops->pcr_nmi_disable);
+ __get_cpu_var(wd_enabled) = 0;
+ atomic_dec(&nmi_active);
+}
+
+static int __init check_nmi_watchdog(void)
+{
+ unsigned int *prev_nmi_count;
+ int cpu, err;
+
+ if (!atomic_read(&nmi_active))
+ return 0;
+
+ prev_nmi_count = kmalloc(nr_cpu_ids * sizeof(unsigned int), GFP_KERNEL);
+ if (!prev_nmi_count) {
+ err = -ENOMEM;
+ goto error;
+ }
+
+ printk(KERN_INFO "Testing NMI watchdog ... ");
+
+ smp_call_function(nmi_cpu_busy, (void *)&endflag, 0);
+
+ for_each_possible_cpu(cpu)
+ prev_nmi_count[cpu] = get_nmi_count(cpu);
+ local_irq_enable();
+ mdelay((20 * 1000) / nmi_hz); /* wait 20 ticks */
+
+ for_each_online_cpu(cpu) {
+ if (!per_cpu(wd_enabled, cpu))
+ continue;
+ if (get_nmi_count(cpu) - prev_nmi_count[cpu] <= 5)
+ report_broken_nmi(cpu, prev_nmi_count);
+ }
+ endflag = 1;
+ if (!atomic_read(&nmi_active)) {
+ kfree(prev_nmi_count);
+ atomic_set(&nmi_active, -1);
+ err = -ENODEV;
+ goto error;
+ }
+ printk("OK.\n");
+
+ nmi_hz = 1;
+
+ kfree(prev_nmi_count);
+ return 0;
+error:
+ on_each_cpu(stop_nmi_watchdog, NULL, 1);
+ return err;
+}
+
+void start_nmi_watchdog(void *unused)
+{
+ __get_cpu_var(wd_enabled) = 1;
+ atomic_inc(&nmi_active);
+
+ pcr_ops->write_pcr(0, pcr_ops->pcr_nmi_disable);
+ pcr_ops->write_pic(0, pcr_ops->nmi_picl_value(nmi_hz));
+
+ pcr_ops->write_pcr(0, pcr_ops->pcr_nmi_enable);
+}
+
+static void nmi_adjust_hz_one(void *unused)
+{
+ if (!__get_cpu_var(wd_enabled))
+ return;
+
+ pcr_ops->write_pcr(0, pcr_ops->pcr_nmi_disable);
+ pcr_ops->write_pic(0, pcr_ops->nmi_picl_value(nmi_hz));
+
+ pcr_ops->write_pcr(0, pcr_ops->pcr_nmi_enable);
+}
+
+void nmi_adjust_hz(unsigned int new_hz)
+{
+ nmi_hz = new_hz;
+ on_each_cpu(nmi_adjust_hz_one, NULL, 1);
+}
+EXPORT_SYMBOL_GPL(nmi_adjust_hz);
+
+static int nmi_shutdown(struct notifier_block *nb, unsigned long cmd, void *p)
+{
+ on_each_cpu(stop_nmi_watchdog, NULL, 1);
+ return 0;
+}
+
+static struct notifier_block nmi_reboot_notifier = {
+ .notifier_call = nmi_shutdown,
+};
+
+int __init nmi_init(void)
+{
+ int err;
+
+ on_each_cpu(start_nmi_watchdog, NULL, 1);
+
+ err = check_nmi_watchdog();
+ if (!err) {
+ err = register_reboot_notifier(&nmi_reboot_notifier);
+ if (err) {
+ on_each_cpu(stop_nmi_watchdog, NULL, 1);
+ atomic_set(&nmi_active, -1);
+ }
+ }
+
+ return err;
+}
+
+static int __init setup_nmi_watchdog(char *str)
+{
+ if (!strncmp(str, "panic", 5))
+ panic_on_timeout = 1;
+
+ return 0;
+}
+__setup("nmi_watchdog=", setup_nmi_watchdog);
diff --git a/arch/sparc/kernel/of_device.c b/arch/sparc/kernel/of_device.c
deleted file mode 100644
index cc4c235c4f5..00000000000
--- a/arch/sparc/kernel/of_device.c
+++ /dev/null
@@ -1,586 +0,0 @@
-#include <linux/string.h>
-#include <linux/kernel.h>
-#include <linux/of.h>
-#include <linux/init.h>
-#include <linux/module.h>
-#include <linux/mod_devicetable.h>
-#include <linux/slab.h>
-#include <linux/errno.h>
-#include <linux/of_device.h>
-#include <linux/of_platform.h>
-
-static int node_match(struct device *dev, void *data)
-{
- struct of_device *op = to_of_device(dev);
- struct device_node *dp = data;
-
- return (op->node == dp);
-}
-
-struct of_device *of_find_device_by_node(struct device_node *dp)
-{
- struct device *dev = bus_find_device(&of_platform_bus_type, NULL,
- dp, node_match);
-
- if (dev)
- return to_of_device(dev);
-
- return NULL;
-}
-EXPORT_SYMBOL(of_find_device_by_node);
-
-#ifdef CONFIG_PCI
-struct bus_type ebus_bus_type;
-EXPORT_SYMBOL(ebus_bus_type);
-#endif
-
-#ifdef CONFIG_SBUS
-struct bus_type sbus_bus_type;
-EXPORT_SYMBOL(sbus_bus_type);
-#endif
-
-struct bus_type of_platform_bus_type;
-EXPORT_SYMBOL(of_platform_bus_type);
-
-static inline u64 of_read_addr(const u32 *cell, int size)
-{
- u64 r = 0;
- while (size--)
- r = (r << 32) | *(cell++);
- return r;
-}
-
-static void __init get_cells(struct device_node *dp,
- int *addrc, int *sizec)
-{
- if (addrc)
- *addrc = of_n_addr_cells(dp);
- if (sizec)
- *sizec = of_n_size_cells(dp);
-}
-
-/* Max address size we deal with */
-#define OF_MAX_ADDR_CELLS 4
-
-struct of_bus {
- const char *name;
- const char *addr_prop_name;
- int (*match)(struct device_node *parent);
- void (*count_cells)(struct device_node *child,
- int *addrc, int *sizec);
- int (*map)(u32 *addr, const u32 *range,
- int na, int ns, int pna);
- unsigned int (*get_flags)(const u32 *addr);
-};
-
-/*
- * Default translator (generic bus)
- */
-
-static void of_bus_default_count_cells(struct device_node *dev,
- int *addrc, int *sizec)
-{
- get_cells(dev, addrc, sizec);
-}
-
-/* Make sure the least significant 64-bits are in-range. Even
- * for 3 or 4 cell values it is a good enough approximation.
- */
-static int of_out_of_range(const u32 *addr, const u32 *base,
- const u32 *size, int na, int ns)
-{
- u64 a = of_read_addr(addr, na);
- u64 b = of_read_addr(base, na);
-
- if (a < b)
- return 1;
-
- b += of_read_addr(size, ns);
- if (a >= b)
- return 1;
-
- return 0;
-}
-
-static int of_bus_default_map(u32 *addr, const u32 *range,
- int na, int ns, int pna)
-{
- u32 result[OF_MAX_ADDR_CELLS];
- int i;
-
- if (ns > 2) {
- printk("of_device: Cannot handle size cells (%d) > 2.", ns);
- return -EINVAL;
- }
-
- if (of_out_of_range(addr, range, range + na + pna, na, ns))
- return -EINVAL;
-
- /* Start with the parent range base. */
- memcpy(result, range + na, pna * 4);
-
- /* Add in the child address offset. */
- for (i = 0; i < na; i++)
- result[pna - 1 - i] +=
- (addr[na - 1 - i] -
- range[na - 1 - i]);
-
- memcpy(addr, result, pna * 4);
-
- return 0;
-}
-
-static unsigned int of_bus_default_get_flags(const u32 *addr)
-{
- return IORESOURCE_MEM;
-}
-
-/*
- * PCI bus specific translator
- */
-
-static int of_bus_pci_match(struct device_node *np)
-{
- if (!strcmp(np->type, "pci") || !strcmp(np->type, "pciex")) {
- /* Do not do PCI specific frobbing if the
- * PCI bridge lacks a ranges property. We
- * want to pass it through up to the next
- * parent as-is, not with the PCI translate
- * method which chops off the top address cell.
- */
- if (!of_find_property(np, "ranges", NULL))
- return 0;
-
- return 1;
- }
-
- return 0;
-}
-
-static void of_bus_pci_count_cells(struct device_node *np,
- int *addrc, int *sizec)
-{
- if (addrc)
- *addrc = 3;
- if (sizec)
- *sizec = 2;
-}
-
-static int of_bus_pci_map(u32 *addr, const u32 *range,
- int na, int ns, int pna)
-{
- u32 result[OF_MAX_ADDR_CELLS];
- int i;
-
- /* Check address type match */
- if ((addr[0] ^ range[0]) & 0x03000000)
- return -EINVAL;
-
- if (of_out_of_range(addr + 1, range + 1, range + na + pna,
- na - 1, ns))
- return -EINVAL;
-
- /* Start with the parent range base. */
- memcpy(result, range + na, pna * 4);
-
- /* Add in the child address offset, skipping high cell. */
- for (i = 0; i < na - 1; i++)
- result[pna - 1 - i] +=
- (addr[na - 1 - i] -
- range[na - 1 - i]);
-
- memcpy(addr, result, pna * 4);
-
- return 0;
-}
-
-static unsigned int of_bus_pci_get_flags(const u32 *addr)
-{
- unsigned int flags = 0;
- u32 w = addr[0];
-
- switch((w >> 24) & 0x03) {
- case 0x01:
- flags |= IORESOURCE_IO;
- case 0x02: /* 32 bits */
- case 0x03: /* 64 bits */
- flags |= IORESOURCE_MEM;
- }
- if (w & 0x40000000)
- flags |= IORESOURCE_PREFETCH;
- return flags;
-}
-
-/*
- * SBUS bus specific translator
- */
-
-static int of_bus_sbus_match(struct device_node *np)
-{
- return !strcmp(np->name, "sbus") ||
- !strcmp(np->name, "sbi");
-}
-
-static void of_bus_sbus_count_cells(struct device_node *child,
- int *addrc, int *sizec)
-{
- if (addrc)
- *addrc = 2;
- if (sizec)
- *sizec = 1;
-}
-
-static int of_bus_sbus_map(u32 *addr, const u32 *range, int na, int ns, int pna)
-{
- return of_bus_default_map(addr, range, na, ns, pna);
-}
-
-static unsigned int of_bus_sbus_get_flags(const u32 *addr)
-{
- return IORESOURCE_MEM;
-}
-
-
-/*
- * Array of bus specific translators
- */
-
-static struct of_bus of_busses[] = {
- /* PCI */
- {
- .name = "pci",
- .addr_prop_name = "assigned-addresses",
- .match = of_bus_pci_match,
- .count_cells = of_bus_pci_count_cells,
- .map = of_bus_pci_map,
- .get_flags = of_bus_pci_get_flags,
- },
- /* SBUS */
- {
- .name = "sbus",
- .addr_prop_name = "reg",
- .match = of_bus_sbus_match,
- .count_cells = of_bus_sbus_count_cells,
- .map = of_bus_sbus_map,
- .get_flags = of_bus_sbus_get_flags,
- },
- /* Default */
- {
- .name = "default",
- .addr_prop_name = "reg",
- .match = NULL,
- .count_cells = of_bus_default_count_cells,
- .map = of_bus_default_map,
- .get_flags = of_bus_default_get_flags,
- },
-};
-
-static struct of_bus *of_match_bus(struct device_node *np)
-{
- int i;
-
- for (i = 0; i < ARRAY_SIZE(of_busses); i ++)
- if (!of_busses[i].match || of_busses[i].match(np))
- return &of_busses[i];
- BUG();
- return NULL;
-}
-
-static int __init build_one_resource(struct device_node *parent,
- struct of_bus *bus,
- struct of_bus *pbus,
- u32 *addr,
- int na, int ns, int pna)
-{
- const u32 *ranges;
- unsigned int rlen;
- int rone;
-
- ranges = of_get_property(parent, "ranges", &rlen);
- if (ranges == NULL || rlen == 0) {
- u32 result[OF_MAX_ADDR_CELLS];
- int i;
-
- memset(result, 0, pna * 4);
- for (i = 0; i < na; i++)
- result[pna - 1 - i] =
- addr[na - 1 - i];
-
- memcpy(addr, result, pna * 4);
- return 0;
- }
-
- /* Now walk through the ranges */
- rlen /= 4;
- rone = na + pna + ns;
- for (; rlen >= rone; rlen -= rone, ranges += rone) {
- if (!bus->map(addr, ranges, na, ns, pna))
- return 0;
- }
-
- return 1;
-}
-
-static int of_resource_verbose;
-
-static void __init build_device_resources(struct of_device *op,
- struct device *parent)
-{
- struct of_device *p_op;
- struct of_bus *bus;
- int na, ns;
- int index, num_reg;
- const void *preg;
-
- if (!parent)
- return;
-
- p_op = to_of_device(parent);
- bus = of_match_bus(p_op->node);
- bus->count_cells(op->node, &na, &ns);
-
- preg = of_get_property(op->node, bus->addr_prop_name, &num_reg);
- if (!preg || num_reg == 0)
- return;
-
- /* Convert to num-cells. */
- num_reg /= 4;
-
- /* Conver to num-entries. */
- num_reg /= na + ns;
-
- for (index = 0; index < num_reg; index++) {
- struct resource *r = &op->resource[index];
- u32 addr[OF_MAX_ADDR_CELLS];
- const u32 *reg = (preg + (index * ((na + ns) * 4)));
- struct device_node *dp = op->node;
- struct device_node *pp = p_op->node;
- struct of_bus *pbus, *dbus;
- u64 size, result = OF_BAD_ADDR;
- unsigned long flags;
- int dna, dns;
- int pna, pns;
-
- size = of_read_addr(reg + na, ns);
- flags = bus->get_flags(reg);
-
- memcpy(addr, reg, na * 4);
-
- /* If the immediate parent has no ranges property to apply,
- * just use a 1<->1 mapping.
- */
- if (of_find_property(pp, "ranges", NULL) == NULL) {
- result = of_read_addr(addr, na);
- goto build_res;
- }
-
- dna = na;
- dns = ns;
- dbus = bus;
-
- while (1) {
- dp = pp;
- pp = dp->parent;
- if (!pp) {
- result = of_read_addr(addr, dna);
- break;
- }
-
- pbus = of_match_bus(pp);
- pbus->count_cells(dp, &pna, &pns);
-
- if (build_one_resource(dp, dbus, pbus, addr,
- dna, dns, pna))
- break;
-
- dna = pna;
- dns = pns;
- dbus = pbus;
- }
-
- build_res:
- memset(r, 0, sizeof(*r));
-
- if (of_resource_verbose)
- printk("%s reg[%d] -> %llx\n",
- op->node->full_name, index,
- result);
-
- if (result != OF_BAD_ADDR) {
- r->start = result & 0xffffffff;
- r->end = result + size - 1;
- r->flags = flags | ((result >> 32ULL) & 0xffUL);
- }
- r->name = op->node->name;
- }
-}
-
-static struct of_device * __init scan_one_device(struct device_node *dp,
- struct device *parent)
-{
- struct of_device *op = kzalloc(sizeof(*op), GFP_KERNEL);
- const struct linux_prom_irqs *intr;
- struct dev_archdata *sd;
- int len, i;
-
- if (!op)
- return NULL;
-
- sd = &op->dev.archdata;
- sd->prom_node = dp;
- sd->op = op;
-
- op->node = dp;
-
- op->clock_freq = of_getintprop_default(dp, "clock-frequency",
- (25*1000*1000));
- op->portid = of_getintprop_default(dp, "upa-portid", -1);
- if (op->portid == -1)
- op->portid = of_getintprop_default(dp, "portid", -1);
-
- intr = of_get_property(dp, "intr", &len);
- if (intr) {
- op->num_irqs = len / sizeof(struct linux_prom_irqs);
- for (i = 0; i < op->num_irqs; i++)
- op->irqs[i] = intr[i].pri;
- } else {
- const unsigned int *irq =
- of_get_property(dp, "interrupts", &len);
-
- if (irq) {
- op->num_irqs = len / sizeof(unsigned int);
- for (i = 0; i < op->num_irqs; i++)
- op->irqs[i] = irq[i];
- } else {
- op->num_irqs = 0;
- }
- }
- if (sparc_cpu_model == sun4d) {
- static int pil_to_sbus[] = {
- 0, 0, 1, 2, 0, 3, 0, 4, 0, 5, 0, 6, 0, 7, 0, 0,
- };
- struct device_node *io_unit, *sbi = dp->parent;
- const struct linux_prom_registers *regs;
- int board, slot;
-
- while (sbi) {
- if (!strcmp(sbi->name, "sbi"))
- break;
-
- sbi = sbi->parent;
- }
- if (!sbi)
- goto build_resources;
-
- regs = of_get_property(dp, "reg", NULL);
- if (!regs)
- goto build_resources;
-
- slot = regs->which_io;
-
- /* If SBI's parent is not io-unit or the io-unit lacks
- * a "board#" property, something is very wrong.
- */
- if (!sbi->parent || strcmp(sbi->parent->name, "io-unit")) {
- printk("%s: Error, parent is not io-unit.\n",
- sbi->full_name);
- goto build_resources;
- }
- io_unit = sbi->parent;
- board = of_getintprop_default(io_unit, "board#", -1);
- if (board == -1) {
- printk("%s: Error, lacks board# property.\n",
- io_unit->full_name);
- goto build_resources;
- }
-
- for (i = 0; i < op->num_irqs; i++) {
- int this_irq = op->irqs[i];
- int sbusl = pil_to_sbus[this_irq];
-
- if (sbusl)
- this_irq = (((board + 1) << 5) +
- (sbusl << 2) +
- slot);
-
- op->irqs[i] = this_irq;
- }
- }
-
-build_resources:
- build_device_resources(op, parent);
-
- op->dev.parent = parent;
- op->dev.bus = &of_platform_bus_type;
- if (!parent)
- strcpy(op->dev.bus_id, "root");
- else
- sprintf(op->dev.bus_id, "%08x", dp->node);
-
- if (of_device_register(op)) {
- printk("%s: Could not register of device.\n",
- dp->full_name);
- kfree(op);
- op = NULL;
- }
-
- return op;
-}
-
-static void __init scan_tree(struct device_node *dp, struct device *parent)
-{
- while (dp) {
- struct of_device *op = scan_one_device(dp, parent);
-
- if (op)
- scan_tree(dp->child, &op->dev);
-
- dp = dp->sibling;
- }
-}
-
-static void __init scan_of_devices(void)
-{
- struct device_node *root = of_find_node_by_path("/");
- struct of_device *parent;
-
- parent = scan_one_device(root, NULL);
- if (!parent)
- return;
-
- scan_tree(root->child, &parent->dev);
-}
-
-static int __init of_bus_driver_init(void)
-{
- int err;
-
- err = of_bus_type_init(&of_platform_bus_type, "of");
-#ifdef CONFIG_PCI
- if (!err)
- err = of_bus_type_init(&ebus_bus_type, "ebus");
-#endif
-#ifdef CONFIG_SBUS
- if (!err)
- err = of_bus_type_init(&sbus_bus_type, "sbus");
-#endif
-
- if (!err)
- scan_of_devices();
-
- return err;
-}
-
-postcore_initcall(of_bus_driver_init);
-
-static int __init of_debug(char *str)
-{
- int val = 0;
-
- get_option(&str, &val);
- if (val & 1)
- of_resource_verbose = 1;
- return 1;
-}
-
-__setup("of_debug=", of_debug);
diff --git a/arch/sparc/kernel/of_device_32.c b/arch/sparc/kernel/of_device_32.c
new file mode 100644
index 00000000000..185aa96fa5b
--- /dev/null
+++ b/arch/sparc/kernel/of_device_32.c
@@ -0,0 +1,429 @@
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/of.h>
+#include <linux/init.h>
+#include <linux/mod_devicetable.h>
+#include <linux/slab.h>
+#include <linux/errno.h>
+#include <linux/irq.h>
+#include <linux/of_device.h>
+#include <linux/of_platform.h>
+#include <asm/leon.h>
+#include <asm/leon_amba.h>
+
+#include "of_device_common.h"
+#include "irq.h"
+
+/*
+ * PCI bus specific translator
+ */
+
+static int of_bus_pci_match(struct device_node *np)
+{
+ if (!strcmp(np->type, "pci") || !strcmp(np->type, "pciex")) {
+ /* Do not do PCI specific frobbing if the
+ * PCI bridge lacks a ranges property. We
+ * want to pass it through up to the next
+ * parent as-is, not with the PCI translate
+ * method which chops off the top address cell.
+ */
+ if (!of_find_property(np, "ranges", NULL))
+ return 0;
+
+ return 1;
+ }
+
+ return 0;
+}
+
+static void of_bus_pci_count_cells(struct device_node *np,
+ int *addrc, int *sizec)
+{
+ if (addrc)
+ *addrc = 3;
+ if (sizec)
+ *sizec = 2;
+}
+
+static int of_bus_pci_map(u32 *addr, const u32 *range,
+ int na, int ns, int pna)
+{
+ u32 result[OF_MAX_ADDR_CELLS];
+ int i;
+
+ /* Check address type match */
+ if ((addr[0] ^ range[0]) & 0x03000000)
+ return -EINVAL;
+
+ if (of_out_of_range(addr + 1, range + 1, range + na + pna,
+ na - 1, ns))
+ return -EINVAL;
+
+ /* Start with the parent range base. */
+ memcpy(result, range + na, pna * 4);
+
+ /* Add in the child address offset, skipping high cell. */
+ for (i = 0; i < na - 1; i++)
+ result[pna - 1 - i] +=
+ (addr[na - 1 - i] -
+ range[na - 1 - i]);
+
+ memcpy(addr, result, pna * 4);
+
+ return 0;
+}
+
+static unsigned long of_bus_pci_get_flags(const u32 *addr, unsigned long flags)
+{
+ u32 w = addr[0];
+
+ /* For PCI, we override whatever child busses may have used. */
+ flags = 0;
+ switch((w >> 24) & 0x03) {
+ case 0x01:
+ flags |= IORESOURCE_IO;
+ break;
+
+ case 0x02: /* 32 bits */
+ case 0x03: /* 64 bits */
+ flags |= IORESOURCE_MEM;
+ break;
+ }
+ if (w & 0x40000000)
+ flags |= IORESOURCE_PREFETCH;
+ return flags;
+}
+
+static unsigned long of_bus_sbus_get_flags(const u32 *addr, unsigned long flags)
+{
+ return IORESOURCE_MEM;
+}
+
+ /*
+ * AMBAPP bus specific translator
+ */
+
+static int of_bus_ambapp_match(struct device_node *np)
+{
+ return !strcmp(np->type, "ambapp");
+}
+
+static void of_bus_ambapp_count_cells(struct device_node *child,
+ int *addrc, int *sizec)
+{
+ if (addrc)
+ *addrc = 1;
+ if (sizec)
+ *sizec = 1;
+}
+
+static int of_bus_ambapp_map(u32 *addr, const u32 *range,
+ int na, int ns, int pna)
+{
+ return of_bus_default_map(addr, range, na, ns, pna);
+}
+
+static unsigned long of_bus_ambapp_get_flags(const u32 *addr,
+ unsigned long flags)
+{
+ return IORESOURCE_MEM;
+}
+
+/*
+ * Array of bus specific translators
+ */
+
+static struct of_bus of_busses[] = {
+ /* PCI */
+ {
+ .name = "pci",
+ .addr_prop_name = "assigned-addresses",
+ .match = of_bus_pci_match,
+ .count_cells = of_bus_pci_count_cells,
+ .map = of_bus_pci_map,
+ .get_flags = of_bus_pci_get_flags,
+ },
+ /* SBUS */
+ {
+ .name = "sbus",
+ .addr_prop_name = "reg",
+ .match = of_bus_sbus_match,
+ .count_cells = of_bus_sbus_count_cells,
+ .map = of_bus_default_map,
+ .get_flags = of_bus_sbus_get_flags,
+ },
+ /* AMBA */
+ {
+ .name = "ambapp",
+ .addr_prop_name = "reg",
+ .match = of_bus_ambapp_match,
+ .count_cells = of_bus_ambapp_count_cells,
+ .map = of_bus_ambapp_map,
+ .get_flags = of_bus_ambapp_get_flags,
+ },
+ /* Default */
+ {
+ .name = "default",
+ .addr_prop_name = "reg",
+ .match = NULL,
+ .count_cells = of_bus_default_count_cells,
+ .map = of_bus_default_map,
+ .get_flags = of_bus_default_get_flags,
+ },
+};
+
+static struct of_bus *of_match_bus(struct device_node *np)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(of_busses); i ++)
+ if (!of_busses[i].match || of_busses[i].match(np))
+ return &of_busses[i];
+ BUG();
+ return NULL;
+}
+
+static int __init build_one_resource(struct device_node *parent,
+ struct of_bus *bus,
+ struct of_bus *pbus,
+ u32 *addr,
+ int na, int ns, int pna)
+{
+ const u32 *ranges;
+ unsigned int rlen;
+ int rone;
+
+ ranges = of_get_property(parent, "ranges", &rlen);
+ if (ranges == NULL || rlen == 0) {
+ u32 result[OF_MAX_ADDR_CELLS];
+ int i;
+
+ memset(result, 0, pna * 4);
+ for (i = 0; i < na; i++)
+ result[pna - 1 - i] =
+ addr[na - 1 - i];
+
+ memcpy(addr, result, pna * 4);
+ return 0;
+ }
+
+ /* Now walk through the ranges */
+ rlen /= 4;
+ rone = na + pna + ns;
+ for (; rlen >= rone; rlen -= rone, ranges += rone) {
+ if (!bus->map(addr, ranges, na, ns, pna))
+ return 0;
+ }
+
+ return 1;
+}
+
+static int __init use_1to1_mapping(struct device_node *pp)
+{
+ /* If we have a ranges property in the parent, use it. */
+ if (of_find_property(pp, "ranges", NULL) != NULL)
+ return 0;
+
+ /* Some SBUS devices use intermediate nodes to express
+ * hierarchy within the device itself. These aren't
+ * real bus nodes, and don't have a 'ranges' property.
+ * But, we should still pass the translation work up
+ * to the SBUS itself.
+ */
+ if (!strcmp(pp->name, "dma") ||
+ !strcmp(pp->name, "espdma") ||
+ !strcmp(pp->name, "ledma") ||
+ !strcmp(pp->name, "lebuffer"))
+ return 0;
+
+ return 1;
+}
+
+static int of_resource_verbose;
+
+static void __init build_device_resources(struct platform_device *op,
+ struct device *parent)
+{
+ struct platform_device *p_op;
+ struct of_bus *bus;
+ int na, ns;
+ int index, num_reg;
+ const void *preg;
+
+ if (!parent)
+ return;
+
+ p_op = to_platform_device(parent);
+ bus = of_match_bus(p_op->dev.of_node);
+ bus->count_cells(op->dev.of_node, &na, &ns);
+
+ preg = of_get_property(op->dev.of_node, bus->addr_prop_name, &num_reg);
+ if (!preg || num_reg == 0)
+ return;
+
+ /* Convert to num-cells. */
+ num_reg /= 4;
+
+ /* Conver to num-entries. */
+ num_reg /= na + ns;
+
+ op->resource = op->archdata.resource;
+ op->num_resources = num_reg;
+ for (index = 0; index < num_reg; index++) {
+ struct resource *r = &op->resource[index];
+ u32 addr[OF_MAX_ADDR_CELLS];
+ const u32 *reg = (preg + (index * ((na + ns) * 4)));
+ struct device_node *dp = op->dev.of_node;
+ struct device_node *pp = p_op->dev.of_node;
+ struct of_bus *pbus, *dbus;
+ u64 size, result = OF_BAD_ADDR;
+ unsigned long flags;
+ int dna, dns;
+ int pna, pns;
+
+ size = of_read_addr(reg + na, ns);
+
+ memcpy(addr, reg, na * 4);
+
+ flags = bus->get_flags(reg, 0);
+
+ if (use_1to1_mapping(pp)) {
+ result = of_read_addr(addr, na);
+ goto build_res;
+ }
+
+ dna = na;
+ dns = ns;
+ dbus = bus;
+
+ while (1) {
+ dp = pp;
+ pp = dp->parent;
+ if (!pp) {
+ result = of_read_addr(addr, dna);
+ break;
+ }
+
+ pbus = of_match_bus(pp);
+ pbus->count_cells(dp, &pna, &pns);
+
+ if (build_one_resource(dp, dbus, pbus, addr,
+ dna, dns, pna))
+ break;
+
+ flags = pbus->get_flags(addr, flags);
+
+ dna = pna;
+ dns = pns;
+ dbus = pbus;
+ }
+
+ build_res:
+ memset(r, 0, sizeof(*r));
+
+ if (of_resource_verbose)
+ printk("%s reg[%d] -> %llx\n",
+ op->dev.of_node->full_name, index,
+ result);
+
+ if (result != OF_BAD_ADDR) {
+ r->start = result & 0xffffffff;
+ r->end = result + size - 1;
+ r->flags = flags | ((result >> 32ULL) & 0xffUL);
+ }
+ r->name = op->dev.of_node->name;
+ }
+}
+
+static struct platform_device * __init scan_one_device(struct device_node *dp,
+ struct device *parent)
+{
+ struct platform_device *op = kzalloc(sizeof(*op), GFP_KERNEL);
+ const struct linux_prom_irqs *intr;
+ struct dev_archdata *sd;
+ int len, i;
+
+ if (!op)
+ return NULL;
+
+ sd = &op->dev.archdata;
+ sd->op = op;
+
+ op->dev.of_node = dp;
+
+ intr = of_get_property(dp, "intr", &len);
+ if (intr) {
+ op->archdata.num_irqs = len / sizeof(struct linux_prom_irqs);
+ for (i = 0; i < op->archdata.num_irqs; i++)
+ op->archdata.irqs[i] =
+ sparc_config.build_device_irq(op, intr[i].pri);
+ } else {
+ const unsigned int *irq =
+ of_get_property(dp, "interrupts", &len);
+
+ if (irq) {
+ op->archdata.num_irqs = len / sizeof(unsigned int);
+ for (i = 0; i < op->archdata.num_irqs; i++)
+ op->archdata.irqs[i] =
+ sparc_config.build_device_irq(op, irq[i]);
+ } else {
+ op->archdata.num_irqs = 0;
+ }
+ }
+
+ build_device_resources(op, parent);
+
+ op->dev.parent = parent;
+ op->dev.bus = &platform_bus_type;
+ if (!parent)
+ dev_set_name(&op->dev, "root");
+ else
+ dev_set_name(&op->dev, "%08x", dp->phandle);
+
+ if (of_device_register(op)) {
+ printk("%s: Could not register of device.\n",
+ dp->full_name);
+ kfree(op);
+ op = NULL;
+ }
+
+ return op;
+}
+
+static void __init scan_tree(struct device_node *dp, struct device *parent)
+{
+ while (dp) {
+ struct platform_device *op = scan_one_device(dp, parent);
+
+ if (op)
+ scan_tree(dp->child, &op->dev);
+
+ dp = dp->sibling;
+ }
+}
+
+static int __init scan_of_devices(void)
+{
+ struct device_node *root = of_find_node_by_path("/");
+ struct platform_device *parent;
+
+ parent = scan_one_device(root, NULL);
+ if (!parent)
+ return 0;
+
+ scan_tree(root->child, &parent->dev);
+ return 0;
+}
+postcore_initcall(scan_of_devices);
+
+static int __init of_debug(char *str)
+{
+ int val = 0;
+
+ get_option(&str, &val);
+ if (val & 1)
+ of_resource_verbose = 1;
+ return 1;
+}
+
+__setup("of_debug=", of_debug);
diff --git a/arch/sparc/kernel/of_device_64.c b/arch/sparc/kernel/of_device_64.c
new file mode 100644
index 00000000000..7bbdc26d951
--- /dev/null
+++ b/arch/sparc/kernel/of_device_64.c
@@ -0,0 +1,726 @@
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/of.h>
+#include <linux/init.h>
+#include <linux/export.h>
+#include <linux/mod_devicetable.h>
+#include <linux/slab.h>
+#include <linux/errno.h>
+#include <linux/irq.h>
+#include <linux/of_device.h>
+#include <linux/of_platform.h>
+#include <asm/spitfire.h>
+
+#include "of_device_common.h"
+
+void __iomem *of_ioremap(struct resource *res, unsigned long offset, unsigned long size, char *name)
+{
+ unsigned long ret = res->start + offset;
+ struct resource *r;
+
+ if (res->flags & IORESOURCE_MEM)
+ r = request_mem_region(ret, size, name);
+ else
+ r = request_region(ret, size, name);
+ if (!r)
+ ret = 0;
+
+ return (void __iomem *) ret;
+}
+EXPORT_SYMBOL(of_ioremap);
+
+void of_iounmap(struct resource *res, void __iomem *base, unsigned long size)
+{
+ if (res->flags & IORESOURCE_MEM)
+ release_mem_region((unsigned long) base, size);
+ else
+ release_region((unsigned long) base, size);
+}
+EXPORT_SYMBOL(of_iounmap);
+
+/*
+ * PCI bus specific translator
+ */
+
+static int of_bus_pci_match(struct device_node *np)
+{
+ if (!strcmp(np->name, "pci")) {
+ const char *model = of_get_property(np, "model", NULL);
+
+ if (model && !strcmp(model, "SUNW,simba"))
+ return 0;
+
+ /* Do not do PCI specific frobbing if the
+ * PCI bridge lacks a ranges property. We
+ * want to pass it through up to the next
+ * parent as-is, not with the PCI translate
+ * method which chops off the top address cell.
+ */
+ if (!of_find_property(np, "ranges", NULL))
+ return 0;
+
+ return 1;
+ }
+
+ return 0;
+}
+
+static int of_bus_simba_match(struct device_node *np)
+{
+ const char *model = of_get_property(np, "model", NULL);
+
+ if (model && !strcmp(model, "SUNW,simba"))
+ return 1;
+
+ /* Treat PCI busses lacking ranges property just like
+ * simba.
+ */
+ if (!strcmp(np->name, "pci")) {
+ if (!of_find_property(np, "ranges", NULL))
+ return 1;
+ }
+
+ return 0;
+}
+
+static int of_bus_simba_map(u32 *addr, const u32 *range,
+ int na, int ns, int pna)
+{
+ return 0;
+}
+
+static void of_bus_pci_count_cells(struct device_node *np,
+ int *addrc, int *sizec)
+{
+ if (addrc)
+ *addrc = 3;
+ if (sizec)
+ *sizec = 2;
+}
+
+static int of_bus_pci_map(u32 *addr, const u32 *range,
+ int na, int ns, int pna)
+{
+ u32 result[OF_MAX_ADDR_CELLS];
+ int i;
+
+ /* Check address type match */
+ if (!((addr[0] ^ range[0]) & 0x03000000))
+ goto type_match;
+
+ /* Special exception, we can map a 64-bit address into
+ * a 32-bit range.
+ */
+ if ((addr[0] & 0x03000000) == 0x03000000 &&
+ (range[0] & 0x03000000) == 0x02000000)
+ goto type_match;
+
+ return -EINVAL;
+
+type_match:
+ if (of_out_of_range(addr + 1, range + 1, range + na + pna,
+ na - 1, ns))
+ return -EINVAL;
+
+ /* Start with the parent range base. */
+ memcpy(result, range + na, pna * 4);
+
+ /* Add in the child address offset, skipping high cell. */
+ for (i = 0; i < na - 1; i++)
+ result[pna - 1 - i] +=
+ (addr[na - 1 - i] -
+ range[na - 1 - i]);
+
+ memcpy(addr, result, pna * 4);
+
+ return 0;
+}
+
+static unsigned long of_bus_pci_get_flags(const u32 *addr, unsigned long flags)
+{
+ u32 w = addr[0];
+
+ /* For PCI, we override whatever child busses may have used. */
+ flags = 0;
+ switch((w >> 24) & 0x03) {
+ case 0x01:
+ flags |= IORESOURCE_IO;
+ break;
+
+ case 0x02: /* 32 bits */
+ case 0x03: /* 64 bits */
+ flags |= IORESOURCE_MEM;
+ break;
+ }
+ if (w & 0x40000000)
+ flags |= IORESOURCE_PREFETCH;
+ return flags;
+}
+
+/*
+ * FHC/Central bus specific translator.
+ *
+ * This is just needed to hard-code the address and size cell
+ * counts. 'fhc' and 'central' nodes lack the #address-cells and
+ * #size-cells properties, and if you walk to the root on such
+ * Enterprise boxes all you'll get is a #size-cells of 2 which is
+ * not what we want to use.
+ */
+static int of_bus_fhc_match(struct device_node *np)
+{
+ return !strcmp(np->name, "fhc") ||
+ !strcmp(np->name, "central");
+}
+
+#define of_bus_fhc_count_cells of_bus_sbus_count_cells
+
+/*
+ * Array of bus specific translators
+ */
+
+static struct of_bus of_busses[] = {
+ /* PCI */
+ {
+ .name = "pci",
+ .addr_prop_name = "assigned-addresses",
+ .match = of_bus_pci_match,
+ .count_cells = of_bus_pci_count_cells,
+ .map = of_bus_pci_map,
+ .get_flags = of_bus_pci_get_flags,
+ },
+ /* SIMBA */
+ {
+ .name = "simba",
+ .addr_prop_name = "assigned-addresses",
+ .match = of_bus_simba_match,
+ .count_cells = of_bus_pci_count_cells,
+ .map = of_bus_simba_map,
+ .get_flags = of_bus_pci_get_flags,
+ },
+ /* SBUS */
+ {
+ .name = "sbus",
+ .addr_prop_name = "reg",
+ .match = of_bus_sbus_match,
+ .count_cells = of_bus_sbus_count_cells,
+ .map = of_bus_default_map,
+ .get_flags = of_bus_default_get_flags,
+ },
+ /* FHC */
+ {
+ .name = "fhc",
+ .addr_prop_name = "reg",
+ .match = of_bus_fhc_match,
+ .count_cells = of_bus_fhc_count_cells,
+ .map = of_bus_default_map,
+ .get_flags = of_bus_default_get_flags,
+ },
+ /* Default */
+ {
+ .name = "default",
+ .addr_prop_name = "reg",
+ .match = NULL,
+ .count_cells = of_bus_default_count_cells,
+ .map = of_bus_default_map,
+ .get_flags = of_bus_default_get_flags,
+ },
+};
+
+static struct of_bus *of_match_bus(struct device_node *np)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(of_busses); i ++)
+ if (!of_busses[i].match || of_busses[i].match(np))
+ return &of_busses[i];
+ BUG();
+ return NULL;
+}
+
+static int __init build_one_resource(struct device_node *parent,
+ struct of_bus *bus,
+ struct of_bus *pbus,
+ u32 *addr,
+ int na, int ns, int pna)
+{
+ const u32 *ranges;
+ int rone, rlen;
+
+ ranges = of_get_property(parent, "ranges", &rlen);
+ if (ranges == NULL || rlen == 0) {
+ u32 result[OF_MAX_ADDR_CELLS];
+ int i;
+
+ memset(result, 0, pna * 4);
+ for (i = 0; i < na; i++)
+ result[pna - 1 - i] =
+ addr[na - 1 - i];
+
+ memcpy(addr, result, pna * 4);
+ return 0;
+ }
+
+ /* Now walk through the ranges */
+ rlen /= 4;
+ rone = na + pna + ns;
+ for (; rlen >= rone; rlen -= rone, ranges += rone) {
+ if (!bus->map(addr, ranges, na, ns, pna))
+ return 0;
+ }
+
+ /* When we miss an I/O space match on PCI, just pass it up
+ * to the next PCI bridge and/or controller.
+ */
+ if (!strcmp(bus->name, "pci") &&
+ (addr[0] & 0x03000000) == 0x01000000)
+ return 0;
+
+ return 1;
+}
+
+static int __init use_1to1_mapping(struct device_node *pp)
+{
+ /* If we have a ranges property in the parent, use it. */
+ if (of_find_property(pp, "ranges", NULL) != NULL)
+ return 0;
+
+ /* If the parent is the dma node of an ISA bus, pass
+ * the translation up to the root.
+ *
+ * Some SBUS devices use intermediate nodes to express
+ * hierarchy within the device itself. These aren't
+ * real bus nodes, and don't have a 'ranges' property.
+ * But, we should still pass the translation work up
+ * to the SBUS itself.
+ */
+ if (!strcmp(pp->name, "dma") ||
+ !strcmp(pp->name, "espdma") ||
+ !strcmp(pp->name, "ledma") ||
+ !strcmp(pp->name, "lebuffer"))
+ return 0;
+
+ /* Similarly for all PCI bridges, if we get this far
+ * it lacks a ranges property, and this will include
+ * cases like Simba.
+ */
+ if (!strcmp(pp->name, "pci"))
+ return 0;
+
+ return 1;
+}
+
+static int of_resource_verbose;
+
+static void __init build_device_resources(struct platform_device *op,
+ struct device *parent)
+{
+ struct platform_device *p_op;
+ struct of_bus *bus;
+ int na, ns;
+ int index, num_reg;
+ const void *preg;
+
+ if (!parent)
+ return;
+
+ p_op = to_platform_device(parent);
+ bus = of_match_bus(p_op->dev.of_node);
+ bus->count_cells(op->dev.of_node, &na, &ns);
+
+ preg = of_get_property(op->dev.of_node, bus->addr_prop_name, &num_reg);
+ if (!preg || num_reg == 0)
+ return;
+
+ /* Convert to num-cells. */
+ num_reg /= 4;
+
+ /* Convert to num-entries. */
+ num_reg /= na + ns;
+
+ /* Prevent overrunning the op->resources[] array. */
+ if (num_reg > PROMREG_MAX) {
+ printk(KERN_WARNING "%s: Too many regs (%d), "
+ "limiting to %d.\n",
+ op->dev.of_node->full_name, num_reg, PROMREG_MAX);
+ num_reg = PROMREG_MAX;
+ }
+
+ op->resource = op->archdata.resource;
+ op->num_resources = num_reg;
+ for (index = 0; index < num_reg; index++) {
+ struct resource *r = &op->resource[index];
+ u32 addr[OF_MAX_ADDR_CELLS];
+ const u32 *reg = (preg + (index * ((na + ns) * 4)));
+ struct device_node *dp = op->dev.of_node;
+ struct device_node *pp = p_op->dev.of_node;
+ struct of_bus *pbus, *dbus;
+ u64 size, result = OF_BAD_ADDR;
+ unsigned long flags;
+ int dna, dns;
+ int pna, pns;
+
+ size = of_read_addr(reg + na, ns);
+ memcpy(addr, reg, na * 4);
+
+ flags = bus->get_flags(addr, 0);
+
+ if (use_1to1_mapping(pp)) {
+ result = of_read_addr(addr, na);
+ goto build_res;
+ }
+
+ dna = na;
+ dns = ns;
+ dbus = bus;
+
+ while (1) {
+ dp = pp;
+ pp = dp->parent;
+ if (!pp) {
+ result = of_read_addr(addr, dna);
+ break;
+ }
+
+ pbus = of_match_bus(pp);
+ pbus->count_cells(dp, &pna, &pns);
+
+ if (build_one_resource(dp, dbus, pbus, addr,
+ dna, dns, pna))
+ break;
+
+ flags = pbus->get_flags(addr, flags);
+
+ dna = pna;
+ dns = pns;
+ dbus = pbus;
+ }
+
+ build_res:
+ memset(r, 0, sizeof(*r));
+
+ if (of_resource_verbose)
+ printk("%s reg[%d] -> %llx\n",
+ op->dev.of_node->full_name, index,
+ result);
+
+ if (result != OF_BAD_ADDR) {
+ if (tlb_type == hypervisor)
+ result &= 0x0fffffffffffffffUL;
+
+ r->start = result;
+ r->end = result + size - 1;
+ r->flags = flags;
+ }
+ r->name = op->dev.of_node->name;
+ }
+}
+
+static struct device_node * __init
+apply_interrupt_map(struct device_node *dp, struct device_node *pp,
+ const u32 *imap, int imlen, const u32 *imask,
+ unsigned int *irq_p)
+{
+ struct device_node *cp;
+ unsigned int irq = *irq_p;
+ struct of_bus *bus;
+ phandle handle;
+ const u32 *reg;
+ int na, num_reg, i;
+
+ bus = of_match_bus(pp);
+ bus->count_cells(dp, &na, NULL);
+
+ reg = of_get_property(dp, "reg", &num_reg);
+ if (!reg || !num_reg)
+ return NULL;
+
+ imlen /= ((na + 3) * 4);
+ handle = 0;
+ for (i = 0; i < imlen; i++) {
+ int j;
+
+ for (j = 0; j < na; j++) {
+ if ((reg[j] & imask[j]) != imap[j])
+ goto next;
+ }
+ if (imap[na] == irq) {
+ handle = imap[na + 1];
+ irq = imap[na + 2];
+ break;
+ }
+
+ next:
+ imap += (na + 3);
+ }
+ if (i == imlen) {
+ /* Psycho and Sabre PCI controllers can have 'interrupt-map'
+ * properties that do not include the on-board device
+ * interrupts. Instead, the device's 'interrupts' property
+ * is already a fully specified INO value.
+ *
+ * Handle this by deciding that, if we didn't get a
+ * match in the parent's 'interrupt-map', and the
+ * parent is an IRQ translator, then use the parent as
+ * our IRQ controller.
+ */
+ if (pp->irq_trans)
+ return pp;
+
+ return NULL;
+ }
+
+ *irq_p = irq;
+ cp = of_find_node_by_phandle(handle);
+
+ return cp;
+}
+
+static unsigned int __init pci_irq_swizzle(struct device_node *dp,
+ struct device_node *pp,
+ unsigned int irq)
+{
+ const struct linux_prom_pci_registers *regs;
+ unsigned int bus, devfn, slot, ret;
+
+ if (irq < 1 || irq > 4)
+ return irq;
+
+ regs = of_get_property(dp, "reg", NULL);
+ if (!regs)
+ return irq;
+
+ bus = (regs->phys_hi >> 16) & 0xff;
+ devfn = (regs->phys_hi >> 8) & 0xff;
+ slot = (devfn >> 3) & 0x1f;
+
+ if (pp->irq_trans) {
+ /* Derived from Table 8-3, U2P User's Manual. This branch
+ * is handling a PCI controller that lacks a proper set of
+ * interrupt-map and interrupt-map-mask properties. The
+ * Ultra-E450 is one example.
+ *
+ * The bit layout is BSSLL, where:
+ * B: 0 on bus A, 1 on bus B
+ * D: 2-bit slot number, derived from PCI device number as
+ * (dev - 1) for bus A, or (dev - 2) for bus B
+ * L: 2-bit line number
+ */
+ if (bus & 0x80) {
+ /* PBM-A */
+ bus = 0x00;
+ slot = (slot - 1) << 2;
+ } else {
+ /* PBM-B */
+ bus = 0x10;
+ slot = (slot - 2) << 2;
+ }
+ irq -= 1;
+
+ ret = (bus | slot | irq);
+ } else {
+ /* Going through a PCI-PCI bridge that lacks a set of
+ * interrupt-map and interrupt-map-mask properties.
+ */
+ ret = ((irq - 1 + (slot & 3)) & 3) + 1;
+ }
+
+ return ret;
+}
+
+static int of_irq_verbose;
+
+static unsigned int __init build_one_device_irq(struct platform_device *op,
+ struct device *parent,
+ unsigned int irq)
+{
+ struct device_node *dp = op->dev.of_node;
+ struct device_node *pp, *ip;
+ unsigned int orig_irq = irq;
+ int nid;
+
+ if (irq == 0xffffffff)
+ return irq;
+
+ if (dp->irq_trans) {
+ irq = dp->irq_trans->irq_build(dp, irq,
+ dp->irq_trans->data);
+
+ if (of_irq_verbose)
+ printk("%s: direct translate %x --> %x\n",
+ dp->full_name, orig_irq, irq);
+
+ goto out;
+ }
+
+ /* Something more complicated. Walk up to the root, applying
+ * interrupt-map or bus specific translations, until we hit
+ * an IRQ translator.
+ *
+ * If we hit a bus type or situation we cannot handle, we
+ * stop and assume that the original IRQ number was in a
+ * format which has special meaning to it's immediate parent.
+ */
+ pp = dp->parent;
+ ip = NULL;
+ while (pp) {
+ const void *imap, *imsk;
+ int imlen;
+
+ imap = of_get_property(pp, "interrupt-map", &imlen);
+ imsk = of_get_property(pp, "interrupt-map-mask", NULL);
+ if (imap && imsk) {
+ struct device_node *iret;
+ int this_orig_irq = irq;
+
+ iret = apply_interrupt_map(dp, pp,
+ imap, imlen, imsk,
+ &irq);
+
+ if (of_irq_verbose)
+ printk("%s: Apply [%s:%x] imap --> [%s:%x]\n",
+ op->dev.of_node->full_name,
+ pp->full_name, this_orig_irq,
+ of_node_full_name(iret), irq);
+
+ if (!iret)
+ break;
+
+ if (iret->irq_trans) {
+ ip = iret;
+ break;
+ }
+ } else {
+ if (!strcmp(pp->name, "pci")) {
+ unsigned int this_orig_irq = irq;
+
+ irq = pci_irq_swizzle(dp, pp, irq);
+ if (of_irq_verbose)
+ printk("%s: PCI swizzle [%s] "
+ "%x --> %x\n",
+ op->dev.of_node->full_name,
+ pp->full_name, this_orig_irq,
+ irq);
+
+ }
+
+ if (pp->irq_trans) {
+ ip = pp;
+ break;
+ }
+ }
+ dp = pp;
+ pp = pp->parent;
+ }
+ if (!ip)
+ return orig_irq;
+
+ irq = ip->irq_trans->irq_build(op->dev.of_node, irq,
+ ip->irq_trans->data);
+ if (of_irq_verbose)
+ printk("%s: Apply IRQ trans [%s] %x --> %x\n",
+ op->dev.of_node->full_name, ip->full_name, orig_irq, irq);
+
+out:
+ nid = of_node_to_nid(dp);
+ if (nid != -1) {
+ cpumask_t numa_mask;
+
+ cpumask_copy(&numa_mask, cpumask_of_node(nid));
+ irq_set_affinity(irq, &numa_mask);
+ }
+
+ return irq;
+}
+
+static struct platform_device * __init scan_one_device(struct device_node *dp,
+ struct device *parent)
+{
+ struct platform_device *op = kzalloc(sizeof(*op), GFP_KERNEL);
+ const unsigned int *irq;
+ struct dev_archdata *sd;
+ int len, i;
+
+ if (!op)
+ return NULL;
+
+ sd = &op->dev.archdata;
+ sd->op = op;
+
+ op->dev.of_node = dp;
+
+ irq = of_get_property(dp, "interrupts", &len);
+ if (irq) {
+ op->archdata.num_irqs = len / 4;
+
+ /* Prevent overrunning the op->irqs[] array. */
+ if (op->archdata.num_irqs > PROMINTR_MAX) {
+ printk(KERN_WARNING "%s: Too many irqs (%d), "
+ "limiting to %d.\n",
+ dp->full_name, op->archdata.num_irqs, PROMINTR_MAX);
+ op->archdata.num_irqs = PROMINTR_MAX;
+ }
+ memcpy(op->archdata.irqs, irq, op->archdata.num_irqs * 4);
+ } else {
+ op->archdata.num_irqs = 0;
+ }
+
+ build_device_resources(op, parent);
+ for (i = 0; i < op->archdata.num_irqs; i++)
+ op->archdata.irqs[i] = build_one_device_irq(op, parent, op->archdata.irqs[i]);
+
+ op->dev.parent = parent;
+ op->dev.bus = &platform_bus_type;
+ if (!parent)
+ dev_set_name(&op->dev, "root");
+ else
+ dev_set_name(&op->dev, "%08x", dp->phandle);
+
+ if (of_device_register(op)) {
+ printk("%s: Could not register of device.\n",
+ dp->full_name);
+ kfree(op);
+ op = NULL;
+ }
+
+ return op;
+}
+
+static void __init scan_tree(struct device_node *dp, struct device *parent)
+{
+ while (dp) {
+ struct platform_device *op = scan_one_device(dp, parent);
+
+ if (op)
+ scan_tree(dp->child, &op->dev);
+
+ dp = dp->sibling;
+ }
+}
+
+static int __init scan_of_devices(void)
+{
+ struct device_node *root = of_find_node_by_path("/");
+ struct platform_device *parent;
+
+ parent = scan_one_device(root, NULL);
+ if (!parent)
+ return 0;
+
+ scan_tree(root->child, &parent->dev);
+ return 0;
+}
+postcore_initcall(scan_of_devices);
+
+static int __init of_debug(char *str)
+{
+ int val = 0;
+
+ get_option(&str, &val);
+ if (val & 1)
+ of_resource_verbose = 1;
+ if (val & 2)
+ of_irq_verbose = 1;
+ return 1;
+}
+
+__setup("of_debug=", of_debug);
diff --git a/arch/sparc/kernel/of_device_common.c b/arch/sparc/kernel/of_device_common.c
new file mode 100644
index 00000000000..de0ee3971f0
--- /dev/null
+++ b/arch/sparc/kernel/of_device_common.c
@@ -0,0 +1,178 @@
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/of.h>
+#include <linux/export.h>
+#include <linux/mod_devicetable.h>
+#include <linux/errno.h>
+#include <linux/irq.h>
+#include <linux/of_platform.h>
+#include <linux/of_address.h>
+#include <linux/of_device.h>
+#include <linux/of_irq.h>
+
+#include "of_device_common.h"
+
+unsigned int irq_of_parse_and_map(struct device_node *node, int index)
+{
+ struct platform_device *op = of_find_device_by_node(node);
+
+ if (!op || index >= op->archdata.num_irqs)
+ return 0;
+
+ return op->archdata.irqs[index];
+}
+EXPORT_SYMBOL(irq_of_parse_and_map);
+
+int of_address_to_resource(struct device_node *node, int index,
+ struct resource *r)
+{
+ struct platform_device *op = of_find_device_by_node(node);
+
+ if (!op || index >= op->num_resources)
+ return -EINVAL;
+
+ memcpy(r, &op->archdata.resource[index], sizeof(*r));
+ return 0;
+}
+EXPORT_SYMBOL_GPL(of_address_to_resource);
+
+void __iomem *of_iomap(struct device_node *node, int index)
+{
+ struct platform_device *op = of_find_device_by_node(node);
+ struct resource *r;
+
+ if (!op || index >= op->num_resources)
+ return NULL;
+
+ r = &op->archdata.resource[index];
+
+ return of_ioremap(r, 0, resource_size(r), (char *) r->name);
+}
+EXPORT_SYMBOL(of_iomap);
+
+/* Take the archdata values for IOMMU, STC, and HOSTDATA found in
+ * BUS and propagate to all child platform_device objects.
+ */
+void of_propagate_archdata(struct platform_device *bus)
+{
+ struct dev_archdata *bus_sd = &bus->dev.archdata;
+ struct device_node *bus_dp = bus->dev.of_node;
+ struct device_node *dp;
+
+ for (dp = bus_dp->child; dp; dp = dp->sibling) {
+ struct platform_device *op = of_find_device_by_node(dp);
+
+ op->dev.archdata.iommu = bus_sd->iommu;
+ op->dev.archdata.stc = bus_sd->stc;
+ op->dev.archdata.host_controller = bus_sd->host_controller;
+ op->dev.archdata.numa_node = bus_sd->numa_node;
+
+ if (dp->child)
+ of_propagate_archdata(op);
+ }
+}
+
+static void get_cells(struct device_node *dp, int *addrc, int *sizec)
+{
+ if (addrc)
+ *addrc = of_n_addr_cells(dp);
+ if (sizec)
+ *sizec = of_n_size_cells(dp);
+}
+
+/*
+ * Default translator (generic bus)
+ */
+
+void of_bus_default_count_cells(struct device_node *dev, int *addrc, int *sizec)
+{
+ get_cells(dev, addrc, sizec);
+}
+
+/* Make sure the least significant 64-bits are in-range. Even
+ * for 3 or 4 cell values it is a good enough approximation.
+ */
+int of_out_of_range(const u32 *addr, const u32 *base,
+ const u32 *size, int na, int ns)
+{
+ u64 a = of_read_addr(addr, na);
+ u64 b = of_read_addr(base, na);
+
+ if (a < b)
+ return 1;
+
+ b += of_read_addr(size, ns);
+ if (a >= b)
+ return 1;
+
+ return 0;
+}
+
+int of_bus_default_map(u32 *addr, const u32 *range, int na, int ns, int pna)
+{
+ u32 result[OF_MAX_ADDR_CELLS];
+ int i;
+
+ if (ns > 2) {
+ printk("of_device: Cannot handle size cells (%d) > 2.", ns);
+ return -EINVAL;
+ }
+
+ if (of_out_of_range(addr, range, range + na + pna, na, ns))
+ return -EINVAL;
+
+ /* Start with the parent range base. */
+ memcpy(result, range + na, pna * 4);
+
+ /* Add in the child address offset. */
+ for (i = 0; i < na; i++)
+ result[pna - 1 - i] +=
+ (addr[na - 1 - i] -
+ range[na - 1 - i]);
+
+ memcpy(addr, result, pna * 4);
+
+ return 0;
+}
+
+unsigned long of_bus_default_get_flags(const u32 *addr, unsigned long flags)
+{
+ if (flags)
+ return flags;
+ return IORESOURCE_MEM;
+}
+
+/*
+ * SBUS bus specific translator
+ */
+
+int of_bus_sbus_match(struct device_node *np)
+{
+ struct device_node *dp = np;
+
+ while (dp) {
+ if (!strcmp(dp->name, "sbus") ||
+ !strcmp(dp->name, "sbi"))
+ return 1;
+
+ /* Have a look at use_1to1_mapping(). We're trying
+ * to match SBUS if that's the top-level bus and we
+ * don't have some intervening real bus that provides
+ * ranges based translations.
+ */
+ if (of_find_property(dp, "ranges", NULL) != NULL)
+ break;
+
+ dp = dp->parent;
+ }
+
+ return 0;
+}
+
+void of_bus_sbus_count_cells(struct device_node *child, int *addrc, int *sizec)
+{
+ if (addrc)
+ *addrc = 2;
+ if (sizec)
+ *sizec = 1;
+}
diff --git a/arch/sparc/kernel/of_device_common.h b/arch/sparc/kernel/of_device_common.h
new file mode 100644
index 00000000000..cdfd2399284
--- /dev/null
+++ b/arch/sparc/kernel/of_device_common.h
@@ -0,0 +1,36 @@
+#ifndef _OF_DEVICE_COMMON_H
+#define _OF_DEVICE_COMMON_H
+
+static inline u64 of_read_addr(const u32 *cell, int size)
+{
+ u64 r = 0;
+ while (size--)
+ r = (r << 32) | *(cell++);
+ return r;
+}
+
+void of_bus_default_count_cells(struct device_node *dev, int *addrc,
+ int *sizec);
+int of_out_of_range(const u32 *addr, const u32 *base,
+ const u32 *size, int na, int ns);
+int of_bus_default_map(u32 *addr, const u32 *range, int na, int ns, int pna);
+unsigned long of_bus_default_get_flags(const u32 *addr, unsigned long flags);
+
+int of_bus_sbus_match(struct device_node *np);
+void of_bus_sbus_count_cells(struct device_node *child, int *addrc, int *sizec);
+
+/* Max address size we deal with */
+#define OF_MAX_ADDR_CELLS 4
+
+struct of_bus {
+ const char *name;
+ const char *addr_prop_name;
+ int (*match)(struct device_node *parent);
+ void (*count_cells)(struct device_node *child,
+ int *addrc, int *sizec);
+ int (*map)(u32 *addr, const u32 *range,
+ int na, int ns, int pna);
+ unsigned long (*get_flags)(const u32 *addr, unsigned long);
+};
+
+#endif /* _OF_DEVICE_COMMON_H */
diff --git a/arch/sparc/kernel/pci.c b/arch/sparc/kernel/pci.c
new file mode 100644
index 00000000000..539babf00bb
--- /dev/null
+++ b/arch/sparc/kernel/pci.c
@@ -0,0 +1,1009 @@
+/* pci.c: UltraSparc PCI controller support.
+ *
+ * Copyright (C) 1997, 1998, 1999 David S. Miller (davem@redhat.com)
+ * Copyright (C) 1998, 1999 Eddie C. Dost (ecd@skynet.be)
+ * Copyright (C) 1999 Jakub Jelinek (jj@ultra.linux.cz)
+ *
+ * OF tree based PCI bus probing taken from the PowerPC port
+ * with minor modifications, see there for credits.
+ */
+
+#include <linux/export.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/sched.h>
+#include <linux/capability.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/msi.h>
+#include <linux/irq.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+
+#include <asm/uaccess.h>
+#include <asm/pgtable.h>
+#include <asm/irq.h>
+#include <asm/prom.h>
+#include <asm/apb.h>
+
+#include "pci_impl.h"
+#include "kernel.h"
+
+/* List of all PCI controllers found in the system. */
+struct pci_pbm_info *pci_pbm_root = NULL;
+
+/* Each PBM found gets a unique index. */
+int pci_num_pbms = 0;
+
+volatile int pci_poke_in_progress;
+volatile int pci_poke_cpu = -1;
+volatile int pci_poke_faulted;
+
+static DEFINE_SPINLOCK(pci_poke_lock);
+
+void pci_config_read8(u8 *addr, u8 *ret)
+{
+ unsigned long flags;
+ u8 byte;
+
+ spin_lock_irqsave(&pci_poke_lock, flags);
+ pci_poke_cpu = smp_processor_id();
+ pci_poke_in_progress = 1;
+ pci_poke_faulted = 0;
+ __asm__ __volatile__("membar #Sync\n\t"
+ "lduba [%1] %2, %0\n\t"
+ "membar #Sync"
+ : "=r" (byte)
+ : "r" (addr), "i" (ASI_PHYS_BYPASS_EC_E_L)
+ : "memory");
+ pci_poke_in_progress = 0;
+ pci_poke_cpu = -1;
+ if (!pci_poke_faulted)
+ *ret = byte;
+ spin_unlock_irqrestore(&pci_poke_lock, flags);
+}
+
+void pci_config_read16(u16 *addr, u16 *ret)
+{
+ unsigned long flags;
+ u16 word;
+
+ spin_lock_irqsave(&pci_poke_lock, flags);
+ pci_poke_cpu = smp_processor_id();
+ pci_poke_in_progress = 1;
+ pci_poke_faulted = 0;
+ __asm__ __volatile__("membar #Sync\n\t"
+ "lduha [%1] %2, %0\n\t"
+ "membar #Sync"
+ : "=r" (word)
+ : "r" (addr), "i" (ASI_PHYS_BYPASS_EC_E_L)
+ : "memory");
+ pci_poke_in_progress = 0;
+ pci_poke_cpu = -1;
+ if (!pci_poke_faulted)
+ *ret = word;
+ spin_unlock_irqrestore(&pci_poke_lock, flags);
+}
+
+void pci_config_read32(u32 *addr, u32 *ret)
+{
+ unsigned long flags;
+ u32 dword;
+
+ spin_lock_irqsave(&pci_poke_lock, flags);
+ pci_poke_cpu = smp_processor_id();
+ pci_poke_in_progress = 1;
+ pci_poke_faulted = 0;
+ __asm__ __volatile__("membar #Sync\n\t"
+ "lduwa [%1] %2, %0\n\t"
+ "membar #Sync"
+ : "=r" (dword)
+ : "r" (addr), "i" (ASI_PHYS_BYPASS_EC_E_L)
+ : "memory");
+ pci_poke_in_progress = 0;
+ pci_poke_cpu = -1;
+ if (!pci_poke_faulted)
+ *ret = dword;
+ spin_unlock_irqrestore(&pci_poke_lock, flags);
+}
+
+void pci_config_write8(u8 *addr, u8 val)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&pci_poke_lock, flags);
+ pci_poke_cpu = smp_processor_id();
+ pci_poke_in_progress = 1;
+ pci_poke_faulted = 0;
+ __asm__ __volatile__("membar #Sync\n\t"
+ "stba %0, [%1] %2\n\t"
+ "membar #Sync"
+ : /* no outputs */
+ : "r" (val), "r" (addr), "i" (ASI_PHYS_BYPASS_EC_E_L)
+ : "memory");
+ pci_poke_in_progress = 0;
+ pci_poke_cpu = -1;
+ spin_unlock_irqrestore(&pci_poke_lock, flags);
+}
+
+void pci_config_write16(u16 *addr, u16 val)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&pci_poke_lock, flags);
+ pci_poke_cpu = smp_processor_id();
+ pci_poke_in_progress = 1;
+ pci_poke_faulted = 0;
+ __asm__ __volatile__("membar #Sync\n\t"
+ "stha %0, [%1] %2\n\t"
+ "membar #Sync"
+ : /* no outputs */
+ : "r" (val), "r" (addr), "i" (ASI_PHYS_BYPASS_EC_E_L)
+ : "memory");
+ pci_poke_in_progress = 0;
+ pci_poke_cpu = -1;
+ spin_unlock_irqrestore(&pci_poke_lock, flags);
+}
+
+void pci_config_write32(u32 *addr, u32 val)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&pci_poke_lock, flags);
+ pci_poke_cpu = smp_processor_id();
+ pci_poke_in_progress = 1;
+ pci_poke_faulted = 0;
+ __asm__ __volatile__("membar #Sync\n\t"
+ "stwa %0, [%1] %2\n\t"
+ "membar #Sync"
+ : /* no outputs */
+ : "r" (val), "r" (addr), "i" (ASI_PHYS_BYPASS_EC_E_L)
+ : "memory");
+ pci_poke_in_progress = 0;
+ pci_poke_cpu = -1;
+ spin_unlock_irqrestore(&pci_poke_lock, flags);
+}
+
+static int ofpci_verbose;
+
+static int __init ofpci_debug(char *str)
+{
+ int val = 0;
+
+ get_option(&str, &val);
+ if (val)
+ ofpci_verbose = 1;
+ return 1;
+}
+
+__setup("ofpci_debug=", ofpci_debug);
+
+static unsigned long pci_parse_of_flags(u32 addr0)
+{
+ unsigned long flags = 0;
+
+ if (addr0 & 0x02000000) {
+ flags = IORESOURCE_MEM | PCI_BASE_ADDRESS_SPACE_MEMORY;
+ flags |= (addr0 >> 22) & PCI_BASE_ADDRESS_MEM_TYPE_64;
+ flags |= (addr0 >> 28) & PCI_BASE_ADDRESS_MEM_TYPE_1M;
+ if (addr0 & 0x40000000)
+ flags |= IORESOURCE_PREFETCH
+ | PCI_BASE_ADDRESS_MEM_PREFETCH;
+ } else if (addr0 & 0x01000000)
+ flags = IORESOURCE_IO | PCI_BASE_ADDRESS_SPACE_IO;
+ return flags;
+}
+
+/* The of_device layer has translated all of the assigned-address properties
+ * into physical address resources, we only have to figure out the register
+ * mapping.
+ */
+static void pci_parse_of_addrs(struct platform_device *op,
+ struct device_node *node,
+ struct pci_dev *dev)
+{
+ struct resource *op_res;
+ const u32 *addrs;
+ int proplen;
+
+ addrs = of_get_property(node, "assigned-addresses", &proplen);
+ if (!addrs)
+ return;
+ if (ofpci_verbose)
+ printk(" parse addresses (%d bytes) @ %p\n",
+ proplen, addrs);
+ op_res = &op->resource[0];
+ for (; proplen >= 20; proplen -= 20, addrs += 5, op_res++) {
+ struct resource *res;
+ unsigned long flags;
+ int i;
+
+ flags = pci_parse_of_flags(addrs[0]);
+ if (!flags)
+ continue;
+ i = addrs[0] & 0xff;
+ if (ofpci_verbose)
+ printk(" start: %llx, end: %llx, i: %x\n",
+ op_res->start, op_res->end, i);
+
+ if (PCI_BASE_ADDRESS_0 <= i && i <= PCI_BASE_ADDRESS_5) {
+ res = &dev->resource[(i - PCI_BASE_ADDRESS_0) >> 2];
+ } else if (i == dev->rom_base_reg) {
+ res = &dev->resource[PCI_ROM_RESOURCE];
+ flags |= IORESOURCE_READONLY | IORESOURCE_CACHEABLE
+ | IORESOURCE_SIZEALIGN;
+ } else {
+ printk(KERN_ERR "PCI: bad cfg reg num 0x%x\n", i);
+ continue;
+ }
+ res->start = op_res->start;
+ res->end = op_res->end;
+ res->flags = flags;
+ res->name = pci_name(dev);
+ }
+}
+
+static struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm,
+ struct device_node *node,
+ struct pci_bus *bus, int devfn)
+{
+ struct dev_archdata *sd;
+ struct pci_slot *slot;
+ struct platform_device *op;
+ struct pci_dev *dev;
+ const char *type;
+ u32 class;
+
+ dev = pci_alloc_dev(bus);
+ if (!dev)
+ return NULL;
+
+ sd = &dev->dev.archdata;
+ sd->iommu = pbm->iommu;
+ sd->stc = &pbm->stc;
+ sd->host_controller = pbm;
+ sd->op = op = of_find_device_by_node(node);
+ sd->numa_node = pbm->numa_node;
+
+ sd = &op->dev.archdata;
+ sd->iommu = pbm->iommu;
+ sd->stc = &pbm->stc;
+ sd->numa_node = pbm->numa_node;
+
+ if (!strcmp(node->name, "ebus"))
+ of_propagate_archdata(op);
+
+ type = of_get_property(node, "device_type", NULL);
+ if (type == NULL)
+ type = "";
+
+ if (ofpci_verbose)
+ printk(" create device, devfn: %x, type: %s\n",
+ devfn, type);
+
+ dev->sysdata = node;
+ dev->dev.parent = bus->bridge;
+ dev->dev.bus = &pci_bus_type;
+ dev->dev.of_node = of_node_get(node);
+ dev->devfn = devfn;
+ dev->multifunction = 0; /* maybe a lie? */
+ set_pcie_port_type(dev);
+
+ list_for_each_entry(slot, &dev->bus->slots, list)
+ if (PCI_SLOT(dev->devfn) == slot->number)
+ dev->slot = slot;
+
+ dev->vendor = of_getintprop_default(node, "vendor-id", 0xffff);
+ dev->device = of_getintprop_default(node, "device-id", 0xffff);
+ dev->subsystem_vendor =
+ of_getintprop_default(node, "subsystem-vendor-id", 0);
+ dev->subsystem_device =
+ of_getintprop_default(node, "subsystem-id", 0);
+
+ dev->cfg_size = pci_cfg_space_size(dev);
+
+ /* We can't actually use the firmware value, we have
+ * to read what is in the register right now. One
+ * reason is that in the case of IDE interfaces the
+ * firmware can sample the value before the the IDE
+ * interface is programmed into native mode.
+ */
+ pci_read_config_dword(dev, PCI_CLASS_REVISION, &class);
+ dev->class = class >> 8;
+ dev->revision = class & 0xff;
+
+ dev_set_name(&dev->dev, "%04x:%02x:%02x.%d", pci_domain_nr(bus),
+ dev->bus->number, PCI_SLOT(devfn), PCI_FUNC(devfn));
+
+ if (ofpci_verbose)
+ printk(" class: 0x%x device name: %s\n",
+ dev->class, pci_name(dev));
+
+ /* I have seen IDE devices which will not respond to
+ * the bmdma simplex check reads if bus mastering is
+ * disabled.
+ */
+ if ((dev->class >> 8) == PCI_CLASS_STORAGE_IDE)
+ pci_set_master(dev);
+
+ dev->current_state = PCI_UNKNOWN; /* unknown power state */
+ dev->error_state = pci_channel_io_normal;
+ dev->dma_mask = 0xffffffff;
+
+ if (!strcmp(node->name, "pci")) {
+ /* a PCI-PCI bridge */
+ dev->hdr_type = PCI_HEADER_TYPE_BRIDGE;
+ dev->rom_base_reg = PCI_ROM_ADDRESS1;
+ } else if (!strcmp(type, "cardbus")) {
+ dev->hdr_type = PCI_HEADER_TYPE_CARDBUS;
+ } else {
+ dev->hdr_type = PCI_HEADER_TYPE_NORMAL;
+ dev->rom_base_reg = PCI_ROM_ADDRESS;
+
+ dev->irq = sd->op->archdata.irqs[0];
+ if (dev->irq == 0xffffffff)
+ dev->irq = PCI_IRQ_NONE;
+ }
+
+ pci_parse_of_addrs(sd->op, node, dev);
+
+ if (ofpci_verbose)
+ printk(" adding to system ...\n");
+
+ pci_device_add(dev, bus);
+
+ return dev;
+}
+
+static void apb_calc_first_last(u8 map, u32 *first_p, u32 *last_p)
+{
+ u32 idx, first, last;
+
+ first = 8;
+ last = 0;
+ for (idx = 0; idx < 8; idx++) {
+ if ((map & (1 << idx)) != 0) {
+ if (first > idx)
+ first = idx;
+ if (last < idx)
+ last = idx;
+ }
+ }
+
+ *first_p = first;
+ *last_p = last;
+}
+
+/* Cook up fake bus resources for SUNW,simba PCI bridges which lack
+ * a proper 'ranges' property.
+ */
+static void apb_fake_ranges(struct pci_dev *dev,
+ struct pci_bus *bus,
+ struct pci_pbm_info *pbm)
+{
+ struct pci_bus_region region;
+ struct resource *res;
+ u32 first, last;
+ u8 map;
+
+ pci_read_config_byte(dev, APB_IO_ADDRESS_MAP, &map);
+ apb_calc_first_last(map, &first, &last);
+ res = bus->resource[0];
+ res->flags = IORESOURCE_IO;
+ region.start = (first << 21);
+ region.end = (last << 21) + ((1 << 21) - 1);
+ pcibios_bus_to_resource(dev->bus, res, &region);
+
+ pci_read_config_byte(dev, APB_MEM_ADDRESS_MAP, &map);
+ apb_calc_first_last(map, &first, &last);
+ res = bus->resource[1];
+ res->flags = IORESOURCE_MEM;
+ region.start = (first << 29);
+ region.end = (last << 29) + ((1 << 29) - 1);
+ pcibios_bus_to_resource(dev->bus, res, &region);
+}
+
+static void pci_of_scan_bus(struct pci_pbm_info *pbm,
+ struct device_node *node,
+ struct pci_bus *bus);
+
+#define GET_64BIT(prop, i) ((((u64) (prop)[(i)]) << 32) | (prop)[(i)+1])
+
+static void of_scan_pci_bridge(struct pci_pbm_info *pbm,
+ struct device_node *node,
+ struct pci_dev *dev)
+{
+ struct pci_bus *bus;
+ const u32 *busrange, *ranges;
+ int len, i, simba;
+ struct pci_bus_region region;
+ struct resource *res;
+ unsigned int flags;
+ u64 size;
+
+ if (ofpci_verbose)
+ printk("of_scan_pci_bridge(%s)\n", node->full_name);
+
+ /* parse bus-range property */
+ busrange = of_get_property(node, "bus-range", &len);
+ if (busrange == NULL || len != 8) {
+ printk(KERN_DEBUG "Can't get bus-range for PCI-PCI bridge %s\n",
+ node->full_name);
+ return;
+ }
+ ranges = of_get_property(node, "ranges", &len);
+ simba = 0;
+ if (ranges == NULL) {
+ const char *model = of_get_property(node, "model", NULL);
+ if (model && !strcmp(model, "SUNW,simba"))
+ simba = 1;
+ }
+
+ bus = pci_add_new_bus(dev->bus, dev, busrange[0]);
+ if (!bus) {
+ printk(KERN_ERR "Failed to create pci bus for %s\n",
+ node->full_name);
+ return;
+ }
+
+ bus->primary = dev->bus->number;
+ pci_bus_insert_busn_res(bus, busrange[0], busrange[1]);
+ bus->bridge_ctl = 0;
+
+ /* parse ranges property, or cook one up by hand for Simba */
+ /* PCI #address-cells == 3 and #size-cells == 2 always */
+ res = &dev->resource[PCI_BRIDGE_RESOURCES];
+ for (i = 0; i < PCI_NUM_RESOURCES - PCI_BRIDGE_RESOURCES; ++i) {
+ res->flags = 0;
+ bus->resource[i] = res;
+ ++res;
+ }
+ if (simba) {
+ apb_fake_ranges(dev, bus, pbm);
+ goto after_ranges;
+ } else if (ranges == NULL) {
+ pci_read_bridge_bases(bus);
+ goto after_ranges;
+ }
+ i = 1;
+ for (; len >= 32; len -= 32, ranges += 8) {
+ flags = pci_parse_of_flags(ranges[0]);
+ size = GET_64BIT(ranges, 6);
+ if (flags == 0 || size == 0)
+ continue;
+ if (flags & IORESOURCE_IO) {
+ res = bus->resource[0];
+ if (res->flags) {
+ printk(KERN_ERR "PCI: ignoring extra I/O range"
+ " for bridge %s\n", node->full_name);
+ continue;
+ }
+ } else {
+ if (i >= PCI_NUM_RESOURCES - PCI_BRIDGE_RESOURCES) {
+ printk(KERN_ERR "PCI: too many memory ranges"
+ " for bridge %s\n", node->full_name);
+ continue;
+ }
+ res = bus->resource[i];
+ ++i;
+ }
+
+ res->flags = flags;
+ region.start = GET_64BIT(ranges, 1);
+ region.end = region.start + size - 1;
+ pcibios_bus_to_resource(dev->bus, res, &region);
+ }
+after_ranges:
+ sprintf(bus->name, "PCI Bus %04x:%02x", pci_domain_nr(bus),
+ bus->number);
+ if (ofpci_verbose)
+ printk(" bus name: %s\n", bus->name);
+
+ pci_of_scan_bus(pbm, node, bus);
+}
+
+static void pci_of_scan_bus(struct pci_pbm_info *pbm,
+ struct device_node *node,
+ struct pci_bus *bus)
+{
+ struct device_node *child;
+ const u32 *reg;
+ int reglen, devfn, prev_devfn;
+ struct pci_dev *dev;
+
+ if (ofpci_verbose)
+ printk("PCI: scan_bus[%s] bus no %d\n",
+ node->full_name, bus->number);
+
+ child = NULL;
+ prev_devfn = -1;
+ while ((child = of_get_next_child(node, child)) != NULL) {
+ if (ofpci_verbose)
+ printk(" * %s\n", child->full_name);
+ reg = of_get_property(child, "reg", &reglen);
+ if (reg == NULL || reglen < 20)
+ continue;
+
+ devfn = (reg[0] >> 8) & 0xff;
+
+ /* This is a workaround for some device trees
+ * which list PCI devices twice. On the V100
+ * for example, device number 3 is listed twice.
+ * Once as "pm" and once again as "lomp".
+ */
+ if (devfn == prev_devfn)
+ continue;
+ prev_devfn = devfn;
+
+ /* create a new pci_dev for this device */
+ dev = of_create_pci_dev(pbm, child, bus, devfn);
+ if (!dev)
+ continue;
+ if (ofpci_verbose)
+ printk("PCI: dev header type: %x\n",
+ dev->hdr_type);
+
+ if (pci_is_bridge(dev))
+ of_scan_pci_bridge(pbm, child, dev);
+ }
+}
+
+static ssize_t
+show_pciobppath_attr(struct device * dev, struct device_attribute * attr, char * buf)
+{
+ struct pci_dev *pdev;
+ struct device_node *dp;
+
+ pdev = to_pci_dev(dev);
+ dp = pdev->dev.of_node;
+
+ return snprintf (buf, PAGE_SIZE, "%s\n", dp->full_name);
+}
+
+static DEVICE_ATTR(obppath, S_IRUSR | S_IRGRP | S_IROTH, show_pciobppath_attr, NULL);
+
+static void pci_bus_register_of_sysfs(struct pci_bus *bus)
+{
+ struct pci_dev *dev;
+ struct pci_bus *child_bus;
+ int err;
+
+ list_for_each_entry(dev, &bus->devices, bus_list) {
+ /* we don't really care if we can create this file or
+ * not, but we need to assign the result of the call
+ * or the world will fall under alien invasion and
+ * everybody will be frozen on a spaceship ready to be
+ * eaten on alpha centauri by some green and jelly
+ * humanoid.
+ */
+ err = sysfs_create_file(&dev->dev.kobj, &dev_attr_obppath.attr);
+ (void) err;
+ }
+ list_for_each_entry(child_bus, &bus->children, node)
+ pci_bus_register_of_sysfs(child_bus);
+}
+
+struct pci_bus *pci_scan_one_pbm(struct pci_pbm_info *pbm,
+ struct device *parent)
+{
+ LIST_HEAD(resources);
+ struct device_node *node = pbm->op->dev.of_node;
+ struct pci_bus *bus;
+
+ printk("PCI: Scanning PBM %s\n", node->full_name);
+
+ pci_add_resource_offset(&resources, &pbm->io_space,
+ pbm->io_space.start);
+ pci_add_resource_offset(&resources, &pbm->mem_space,
+ pbm->mem_space.start);
+ pbm->busn.start = pbm->pci_first_busno;
+ pbm->busn.end = pbm->pci_last_busno;
+ pbm->busn.flags = IORESOURCE_BUS;
+ pci_add_resource(&resources, &pbm->busn);
+ bus = pci_create_root_bus(parent, pbm->pci_first_busno, pbm->pci_ops,
+ pbm, &resources);
+ if (!bus) {
+ printk(KERN_ERR "Failed to create bus for %s\n",
+ node->full_name);
+ pci_free_resource_list(&resources);
+ return NULL;
+ }
+
+ pci_of_scan_bus(pbm, node, bus);
+ pci_bus_add_devices(bus);
+ pci_bus_register_of_sysfs(bus);
+
+ return bus;
+}
+
+void pcibios_fixup_bus(struct pci_bus *pbus)
+{
+}
+
+resource_size_t pcibios_align_resource(void *data, const struct resource *res,
+ resource_size_t size, resource_size_t align)
+{
+ return res->start;
+}
+
+int pcibios_enable_device(struct pci_dev *dev, int mask)
+{
+ u16 cmd, oldcmd;
+ int i;
+
+ pci_read_config_word(dev, PCI_COMMAND, &cmd);
+ oldcmd = cmd;
+
+ for (i = 0; i < PCI_NUM_RESOURCES; i++) {
+ struct resource *res = &dev->resource[i];
+
+ /* Only set up the requested stuff */
+ if (!(mask & (1<<i)))
+ continue;
+
+ if (res->flags & IORESOURCE_IO)
+ cmd |= PCI_COMMAND_IO;
+ if (res->flags & IORESOURCE_MEM)
+ cmd |= PCI_COMMAND_MEMORY;
+ }
+
+ if (cmd != oldcmd) {
+ printk(KERN_DEBUG "PCI: Enabling device: (%s), cmd %x\n",
+ pci_name(dev), cmd);
+ /* Enable the appropriate bits in the PCI command register. */
+ pci_write_config_word(dev, PCI_COMMAND, cmd);
+ }
+ return 0;
+}
+
+/* Platform support for /proc/bus/pci/X/Y mmap()s. */
+
+/* If the user uses a host-bridge as the PCI device, he may use
+ * this to perform a raw mmap() of the I/O or MEM space behind
+ * that controller.
+ *
+ * This can be useful for execution of x86 PCI bios initialization code
+ * on a PCI card, like the xfree86 int10 stuff does.
+ */
+static int __pci_mmap_make_offset_bus(struct pci_dev *pdev, struct vm_area_struct *vma,
+ enum pci_mmap_state mmap_state)
+{
+ struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
+ unsigned long space_size, user_offset, user_size;
+
+ if (mmap_state == pci_mmap_io) {
+ space_size = resource_size(&pbm->io_space);
+ } else {
+ space_size = resource_size(&pbm->mem_space);
+ }
+
+ /* Make sure the request is in range. */
+ user_offset = vma->vm_pgoff << PAGE_SHIFT;
+ user_size = vma->vm_end - vma->vm_start;
+
+ if (user_offset >= space_size ||
+ (user_offset + user_size) > space_size)
+ return -EINVAL;
+
+ if (mmap_state == pci_mmap_io) {
+ vma->vm_pgoff = (pbm->io_space.start +
+ user_offset) >> PAGE_SHIFT;
+ } else {
+ vma->vm_pgoff = (pbm->mem_space.start +
+ user_offset) >> PAGE_SHIFT;
+ }
+
+ return 0;
+}
+
+/* Adjust vm_pgoff of VMA such that it is the physical page offset
+ * corresponding to the 32-bit pci bus offset for DEV requested by the user.
+ *
+ * Basically, the user finds the base address for his device which he wishes
+ * to mmap. They read the 32-bit value from the config space base register,
+ * add whatever PAGE_SIZE multiple offset they wish, and feed this into the
+ * offset parameter of mmap on /proc/bus/pci/XXX for that device.
+ *
+ * Returns negative error code on failure, zero on success.
+ */
+static int __pci_mmap_make_offset(struct pci_dev *pdev,
+ struct vm_area_struct *vma,
+ enum pci_mmap_state mmap_state)
+{
+ unsigned long user_paddr, user_size;
+ int i, err;
+
+ /* First compute the physical address in vma->vm_pgoff,
+ * making sure the user offset is within range in the
+ * appropriate PCI space.
+ */
+ err = __pci_mmap_make_offset_bus(pdev, vma, mmap_state);
+ if (err)
+ return err;
+
+ /* If this is a mapping on a host bridge, any address
+ * is OK.
+ */
+ if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_HOST)
+ return err;
+
+ /* Otherwise make sure it's in the range for one of the
+ * device's resources.
+ */
+ user_paddr = vma->vm_pgoff << PAGE_SHIFT;
+ user_size = vma->vm_end - vma->vm_start;
+
+ for (i = 0; i <= PCI_ROM_RESOURCE; i++) {
+ struct resource *rp = &pdev->resource[i];
+ resource_size_t aligned_end;
+
+ /* Active? */
+ if (!rp->flags)
+ continue;
+
+ /* Same type? */
+ if (i == PCI_ROM_RESOURCE) {
+ if (mmap_state != pci_mmap_mem)
+ continue;
+ } else {
+ if ((mmap_state == pci_mmap_io &&
+ (rp->flags & IORESOURCE_IO) == 0) ||
+ (mmap_state == pci_mmap_mem &&
+ (rp->flags & IORESOURCE_MEM) == 0))
+ continue;
+ }
+
+ /* Align the resource end to the next page address.
+ * PAGE_SIZE intentionally added instead of (PAGE_SIZE - 1),
+ * because actually we need the address of the next byte
+ * after rp->end.
+ */
+ aligned_end = (rp->end + PAGE_SIZE) & PAGE_MASK;
+
+ if ((rp->start <= user_paddr) &&
+ (user_paddr + user_size) <= aligned_end)
+ break;
+ }
+
+ if (i > PCI_ROM_RESOURCE)
+ return -EINVAL;
+
+ return 0;
+}
+
+/* Set vm_page_prot of VMA, as appropriate for this architecture, for a pci
+ * device mapping.
+ */
+static void __pci_mmap_set_pgprot(struct pci_dev *dev, struct vm_area_struct *vma,
+ enum pci_mmap_state mmap_state)
+{
+ /* Our io_remap_pfn_range takes care of this, do nothing. */
+}
+
+/* Perform the actual remap of the pages for a PCI device mapping, as appropriate
+ * for this architecture. The region in the process to map is described by vm_start
+ * and vm_end members of VMA, the base physical address is found in vm_pgoff.
+ * The pci device structure is provided so that architectures may make mapping
+ * decisions on a per-device or per-bus basis.
+ *
+ * Returns a negative error code on failure, zero on success.
+ */
+int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
+ enum pci_mmap_state mmap_state,
+ int write_combine)
+{
+ int ret;
+
+ ret = __pci_mmap_make_offset(dev, vma, mmap_state);
+ if (ret < 0)
+ return ret;
+
+ __pci_mmap_set_pgprot(dev, vma, mmap_state);
+
+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+ ret = io_remap_pfn_range(vma, vma->vm_start,
+ vma->vm_pgoff,
+ vma->vm_end - vma->vm_start,
+ vma->vm_page_prot);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+#ifdef CONFIG_NUMA
+int pcibus_to_node(struct pci_bus *pbus)
+{
+ struct pci_pbm_info *pbm = pbus->sysdata;
+
+ return pbm->numa_node;
+}
+EXPORT_SYMBOL(pcibus_to_node);
+#endif
+
+/* Return the domain number for this pci bus */
+
+int pci_domain_nr(struct pci_bus *pbus)
+{
+ struct pci_pbm_info *pbm = pbus->sysdata;
+ int ret;
+
+ if (!pbm) {
+ ret = -ENXIO;
+ } else {
+ ret = pbm->index;
+ }
+
+ return ret;
+}
+EXPORT_SYMBOL(pci_domain_nr);
+
+#ifdef CONFIG_PCI_MSI
+int arch_setup_msi_irq(struct pci_dev *pdev, struct msi_desc *desc)
+{
+ struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
+ unsigned int irq;
+
+ if (!pbm->setup_msi_irq)
+ return -EINVAL;
+
+ return pbm->setup_msi_irq(&irq, pdev, desc);
+}
+
+void arch_teardown_msi_irq(unsigned int irq)
+{
+ struct msi_desc *entry = irq_get_msi_desc(irq);
+ struct pci_dev *pdev = entry->dev;
+ struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
+
+ if (pbm->teardown_msi_irq)
+ pbm->teardown_msi_irq(irq, pdev);
+}
+#endif /* !(CONFIG_PCI_MSI) */
+
+static void ali_sound_dma_hack(struct pci_dev *pdev, int set_bit)
+{
+ struct pci_dev *ali_isa_bridge;
+ u8 val;
+
+ /* ALI sound chips generate 31-bits of DMA, a special register
+ * determines what bit 31 is emitted as.
+ */
+ ali_isa_bridge = pci_get_device(PCI_VENDOR_ID_AL,
+ PCI_DEVICE_ID_AL_M1533,
+ NULL);
+
+ pci_read_config_byte(ali_isa_bridge, 0x7e, &val);
+ if (set_bit)
+ val |= 0x01;
+ else
+ val &= ~0x01;
+ pci_write_config_byte(ali_isa_bridge, 0x7e, val);
+ pci_dev_put(ali_isa_bridge);
+}
+
+int pci64_dma_supported(struct pci_dev *pdev, u64 device_mask)
+{
+ u64 dma_addr_mask;
+
+ if (pdev == NULL) {
+ dma_addr_mask = 0xffffffff;
+ } else {
+ struct iommu *iommu = pdev->dev.archdata.iommu;
+
+ dma_addr_mask = iommu->dma_addr_mask;
+
+ if (pdev->vendor == PCI_VENDOR_ID_AL &&
+ pdev->device == PCI_DEVICE_ID_AL_M5451 &&
+ device_mask == 0x7fffffff) {
+ ali_sound_dma_hack(pdev,
+ (dma_addr_mask & 0x80000000) != 0);
+ return 1;
+ }
+ }
+
+ if (device_mask >= (1UL << 32UL))
+ return 0;
+
+ return (device_mask & dma_addr_mask) == dma_addr_mask;
+}
+
+void pci_resource_to_user(const struct pci_dev *pdev, int bar,
+ const struct resource *rp, resource_size_t *start,
+ resource_size_t *end)
+{
+ struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
+ unsigned long offset;
+
+ if (rp->flags & IORESOURCE_IO)
+ offset = pbm->io_space.start;
+ else
+ offset = pbm->mem_space.start;
+
+ *start = rp->start - offset;
+ *end = rp->end - offset;
+}
+
+void pcibios_set_master(struct pci_dev *dev)
+{
+ /* No special bus mastering setup handling */
+}
+
+static int __init pcibios_init(void)
+{
+ pci_dfl_cache_line_size = 64 >> 2;
+ return 0;
+}
+subsys_initcall(pcibios_init);
+
+#ifdef CONFIG_SYSFS
+static void pci_bus_slot_names(struct device_node *node, struct pci_bus *bus)
+{
+ const struct pci_slot_names {
+ u32 slot_mask;
+ char names[0];
+ } *prop;
+ const char *sp;
+ int len, i;
+ u32 mask;
+
+ prop = of_get_property(node, "slot-names", &len);
+ if (!prop)
+ return;
+
+ mask = prop->slot_mask;
+ sp = prop->names;
+
+ if (ofpci_verbose)
+ printk("PCI: Making slots for [%s] mask[0x%02x]\n",
+ node->full_name, mask);
+
+ i = 0;
+ while (mask) {
+ struct pci_slot *pci_slot;
+ u32 this_bit = 1 << i;
+
+ if (!(mask & this_bit)) {
+ i++;
+ continue;
+ }
+
+ if (ofpci_verbose)
+ printk("PCI: Making slot [%s]\n", sp);
+
+ pci_slot = pci_create_slot(bus, i, sp, NULL);
+ if (IS_ERR(pci_slot))
+ printk(KERN_ERR "PCI: pci_create_slot returned %ld\n",
+ PTR_ERR(pci_slot));
+
+ sp += strlen(sp) + 1;
+ mask &= ~this_bit;
+ i++;
+ }
+}
+
+static int __init of_pci_slot_init(void)
+{
+ struct pci_bus *pbus = NULL;
+
+ while ((pbus = pci_find_next_bus(pbus)) != NULL) {
+ struct device_node *node;
+
+ if (pbus->self) {
+ /* PCI->PCI bridge */
+ node = pbus->self->dev.of_node;
+ } else {
+ struct pci_pbm_info *pbm = pbus->sysdata;
+
+ /* Host PCI controller */
+ node = pbm->op->dev.of_node;
+ }
+
+ pci_bus_slot_names(node, pbus);
+ }
+
+ return 0;
+}
+device_initcall(of_pci_slot_init);
+#endif
diff --git a/arch/sparc/kernel/pci_common.c b/arch/sparc/kernel/pci_common.c
new file mode 100644
index 00000000000..944a06536ec
--- /dev/null
+++ b/arch/sparc/kernel/pci_common.c
@@ -0,0 +1,553 @@
+/* pci_common.c: PCI controller common support.
+ *
+ * Copyright (C) 1999, 2007 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/string.h>
+#include <linux/slab.h>
+#include <linux/pci.h>
+#include <linux/device.h>
+#include <linux/of_device.h>
+
+#include <asm/prom.h>
+#include <asm/oplib.h>
+
+#include "pci_impl.h"
+#include "pci_sun4v.h"
+
+static int config_out_of_range(struct pci_pbm_info *pbm,
+ unsigned long bus,
+ unsigned long devfn,
+ unsigned long reg)
+{
+ if (bus < pbm->pci_first_busno ||
+ bus > pbm->pci_last_busno)
+ return 1;
+ return 0;
+}
+
+static void *sun4u_config_mkaddr(struct pci_pbm_info *pbm,
+ unsigned long bus,
+ unsigned long devfn,
+ unsigned long reg)
+{
+ unsigned long rbits = pbm->config_space_reg_bits;
+
+ if (config_out_of_range(pbm, bus, devfn, reg))
+ return NULL;
+
+ reg = (reg & ((1 << rbits) - 1));
+ devfn <<= rbits;
+ bus <<= rbits + 8;
+
+ return (void *) (pbm->config_space | bus | devfn | reg);
+}
+
+/* At least on Sabre, it is necessary to access all PCI host controller
+ * registers at their natural size, otherwise zeros are returned.
+ * Strange but true, and I see no language in the UltraSPARC-IIi
+ * programmer's manual that mentions this even indirectly.
+ */
+static int sun4u_read_pci_cfg_host(struct pci_pbm_info *pbm,
+ unsigned char bus, unsigned int devfn,
+ int where, int size, u32 *value)
+{
+ u32 tmp32, *addr;
+ u16 tmp16;
+ u8 tmp8;
+
+ addr = sun4u_config_mkaddr(pbm, bus, devfn, where);
+ if (!addr)
+ return PCIBIOS_SUCCESSFUL;
+
+ switch (size) {
+ case 1:
+ if (where < 8) {
+ unsigned long align = (unsigned long) addr;
+
+ align &= ~1;
+ pci_config_read16((u16 *)align, &tmp16);
+ if (where & 1)
+ *value = tmp16 >> 8;
+ else
+ *value = tmp16 & 0xff;
+ } else {
+ pci_config_read8((u8 *)addr, &tmp8);
+ *value = (u32) tmp8;
+ }
+ break;
+
+ case 2:
+ if (where < 8) {
+ pci_config_read16((u16 *)addr, &tmp16);
+ *value = (u32) tmp16;
+ } else {
+ pci_config_read8((u8 *)addr, &tmp8);
+ *value = (u32) tmp8;
+ pci_config_read8(((u8 *)addr) + 1, &tmp8);
+ *value |= ((u32) tmp8) << 8;
+ }
+ break;
+
+ case 4:
+ tmp32 = 0xffffffff;
+ sun4u_read_pci_cfg_host(pbm, bus, devfn,
+ where, 2, &tmp32);
+ *value = tmp32;
+
+ tmp32 = 0xffffffff;
+ sun4u_read_pci_cfg_host(pbm, bus, devfn,
+ where + 2, 2, &tmp32);
+ *value |= tmp32 << 16;
+ break;
+ }
+ return PCIBIOS_SUCCESSFUL;
+}
+
+static int sun4u_read_pci_cfg(struct pci_bus *bus_dev, unsigned int devfn,
+ int where, int size, u32 *value)
+{
+ struct pci_pbm_info *pbm = bus_dev->sysdata;
+ unsigned char bus = bus_dev->number;
+ u32 *addr;
+ u16 tmp16;
+ u8 tmp8;
+
+ switch (size) {
+ case 1:
+ *value = 0xff;
+ break;
+ case 2:
+ *value = 0xffff;
+ break;
+ case 4:
+ *value = 0xffffffff;
+ break;
+ }
+
+ if (!bus_dev->number && !PCI_SLOT(devfn))
+ return sun4u_read_pci_cfg_host(pbm, bus, devfn, where,
+ size, value);
+
+ addr = sun4u_config_mkaddr(pbm, bus, devfn, where);
+ if (!addr)
+ return PCIBIOS_SUCCESSFUL;
+
+ switch (size) {
+ case 1:
+ pci_config_read8((u8 *)addr, &tmp8);
+ *value = (u32) tmp8;
+ break;
+
+ case 2:
+ if (where & 0x01) {
+ printk("pci_read_config_word: misaligned reg [%x]\n",
+ where);
+ return PCIBIOS_SUCCESSFUL;
+ }
+ pci_config_read16((u16 *)addr, &tmp16);
+ *value = (u32) tmp16;
+ break;
+
+ case 4:
+ if (where & 0x03) {
+ printk("pci_read_config_dword: misaligned reg [%x]\n",
+ where);
+ return PCIBIOS_SUCCESSFUL;
+ }
+ pci_config_read32(addr, value);
+ break;
+ }
+ return PCIBIOS_SUCCESSFUL;
+}
+
+static int sun4u_write_pci_cfg_host(struct pci_pbm_info *pbm,
+ unsigned char bus, unsigned int devfn,
+ int where, int size, u32 value)
+{
+ u32 *addr;
+
+ addr = sun4u_config_mkaddr(pbm, bus, devfn, where);
+ if (!addr)
+ return PCIBIOS_SUCCESSFUL;
+
+ switch (size) {
+ case 1:
+ if (where < 8) {
+ unsigned long align = (unsigned long) addr;
+ u16 tmp16;
+
+ align &= ~1;
+ pci_config_read16((u16 *)align, &tmp16);
+ if (where & 1) {
+ tmp16 &= 0x00ff;
+ tmp16 |= value << 8;
+ } else {
+ tmp16 &= 0xff00;
+ tmp16 |= value;
+ }
+ pci_config_write16((u16 *)align, tmp16);
+ } else
+ pci_config_write8((u8 *)addr, value);
+ break;
+ case 2:
+ if (where < 8) {
+ pci_config_write16((u16 *)addr, value);
+ } else {
+ pci_config_write8((u8 *)addr, value & 0xff);
+ pci_config_write8(((u8 *)addr) + 1, value >> 8);
+ }
+ break;
+ case 4:
+ sun4u_write_pci_cfg_host(pbm, bus, devfn,
+ where, 2, value & 0xffff);
+ sun4u_write_pci_cfg_host(pbm, bus, devfn,
+ where + 2, 2, value >> 16);
+ break;
+ }
+ return PCIBIOS_SUCCESSFUL;
+}
+
+static int sun4u_write_pci_cfg(struct pci_bus *bus_dev, unsigned int devfn,
+ int where, int size, u32 value)
+{
+ struct pci_pbm_info *pbm = bus_dev->sysdata;
+ unsigned char bus = bus_dev->number;
+ u32 *addr;
+
+ if (!bus_dev->number && !PCI_SLOT(devfn))
+ return sun4u_write_pci_cfg_host(pbm, bus, devfn, where,
+ size, value);
+
+ addr = sun4u_config_mkaddr(pbm, bus, devfn, where);
+ if (!addr)
+ return PCIBIOS_SUCCESSFUL;
+
+ switch (size) {
+ case 1:
+ pci_config_write8((u8 *)addr, value);
+ break;
+
+ case 2:
+ if (where & 0x01) {
+ printk("pci_write_config_word: misaligned reg [%x]\n",
+ where);
+ return PCIBIOS_SUCCESSFUL;
+ }
+ pci_config_write16((u16 *)addr, value);
+ break;
+
+ case 4:
+ if (where & 0x03) {
+ printk("pci_write_config_dword: misaligned reg [%x]\n",
+ where);
+ return PCIBIOS_SUCCESSFUL;
+ }
+ pci_config_write32(addr, value);
+ }
+ return PCIBIOS_SUCCESSFUL;
+}
+
+struct pci_ops sun4u_pci_ops = {
+ .read = sun4u_read_pci_cfg,
+ .write = sun4u_write_pci_cfg,
+};
+
+static int sun4v_read_pci_cfg(struct pci_bus *bus_dev, unsigned int devfn,
+ int where, int size, u32 *value)
+{
+ struct pci_pbm_info *pbm = bus_dev->sysdata;
+ u32 devhandle = pbm->devhandle;
+ unsigned int bus = bus_dev->number;
+ unsigned int device = PCI_SLOT(devfn);
+ unsigned int func = PCI_FUNC(devfn);
+ unsigned long ret;
+
+ if (config_out_of_range(pbm, bus, devfn, where)) {
+ ret = ~0UL;
+ } else {
+ ret = pci_sun4v_config_get(devhandle,
+ HV_PCI_DEVICE_BUILD(bus, device, func),
+ where, size);
+ }
+ switch (size) {
+ case 1:
+ *value = ret & 0xff;
+ break;
+ case 2:
+ *value = ret & 0xffff;
+ break;
+ case 4:
+ *value = ret & 0xffffffff;
+ break;
+ }
+
+
+ return PCIBIOS_SUCCESSFUL;
+}
+
+static int sun4v_write_pci_cfg(struct pci_bus *bus_dev, unsigned int devfn,
+ int where, int size, u32 value)
+{
+ struct pci_pbm_info *pbm = bus_dev->sysdata;
+ u32 devhandle = pbm->devhandle;
+ unsigned int bus = bus_dev->number;
+ unsigned int device = PCI_SLOT(devfn);
+ unsigned int func = PCI_FUNC(devfn);
+
+ if (config_out_of_range(pbm, bus, devfn, where)) {
+ /* Do nothing. */
+ } else {
+ /* We don't check for hypervisor errors here, but perhaps
+ * we should and influence our return value depending upon
+ * what kind of error is thrown.
+ */
+ pci_sun4v_config_put(devhandle,
+ HV_PCI_DEVICE_BUILD(bus, device, func),
+ where, size, value);
+ }
+ return PCIBIOS_SUCCESSFUL;
+}
+
+struct pci_ops sun4v_pci_ops = {
+ .read = sun4v_read_pci_cfg,
+ .write = sun4v_write_pci_cfg,
+};
+
+void pci_get_pbm_props(struct pci_pbm_info *pbm)
+{
+ const u32 *val = of_get_property(pbm->op->dev.of_node, "bus-range", NULL);
+
+ pbm->pci_first_busno = val[0];
+ pbm->pci_last_busno = val[1];
+
+ val = of_get_property(pbm->op->dev.of_node, "ino-bitmap", NULL);
+ if (val) {
+ pbm->ino_bitmap = (((u64)val[1] << 32UL) |
+ ((u64)val[0] << 0UL));
+ }
+}
+
+static void pci_register_legacy_regions(struct resource *io_res,
+ struct resource *mem_res)
+{
+ struct resource *p;
+
+ /* VGA Video RAM. */
+ p = kzalloc(sizeof(*p), GFP_KERNEL);
+ if (!p)
+ return;
+
+ p->name = "Video RAM area";
+ p->start = mem_res->start + 0xa0000UL;
+ p->end = p->start + 0x1ffffUL;
+ p->flags = IORESOURCE_BUSY;
+ request_resource(mem_res, p);
+
+ p = kzalloc(sizeof(*p), GFP_KERNEL);
+ if (!p)
+ return;
+
+ p->name = "System ROM";
+ p->start = mem_res->start + 0xf0000UL;
+ p->end = p->start + 0xffffUL;
+ p->flags = IORESOURCE_BUSY;
+ request_resource(mem_res, p);
+
+ p = kzalloc(sizeof(*p), GFP_KERNEL);
+ if (!p)
+ return;
+
+ p->name = "Video ROM";
+ p->start = mem_res->start + 0xc0000UL;
+ p->end = p->start + 0x7fffUL;
+ p->flags = IORESOURCE_BUSY;
+ request_resource(mem_res, p);
+}
+
+static void pci_register_iommu_region(struct pci_pbm_info *pbm)
+{
+ const u32 *vdma = of_get_property(pbm->op->dev.of_node, "virtual-dma",
+ NULL);
+
+ if (vdma) {
+ struct resource *rp = kzalloc(sizeof(*rp), GFP_KERNEL);
+
+ if (!rp) {
+ pr_info("%s: Cannot allocate IOMMU resource.\n",
+ pbm->name);
+ return;
+ }
+ rp->name = "IOMMU";
+ rp->start = pbm->mem_space.start + (unsigned long) vdma[0];
+ rp->end = rp->start + (unsigned long) vdma[1] - 1UL;
+ rp->flags = IORESOURCE_BUSY;
+ if (request_resource(&pbm->mem_space, rp)) {
+ pr_info("%s: Unable to request IOMMU resource.\n",
+ pbm->name);
+ kfree(rp);
+ }
+ }
+}
+
+void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
+{
+ const struct linux_prom_pci_ranges *pbm_ranges;
+ int i, saw_mem, saw_io;
+ int num_pbm_ranges;
+
+ saw_mem = saw_io = 0;
+ pbm_ranges = of_get_property(pbm->op->dev.of_node, "ranges", &i);
+ if (!pbm_ranges) {
+ prom_printf("PCI: Fatal error, missing PBM ranges property "
+ " for %s\n",
+ pbm->name);
+ prom_halt();
+ }
+
+ num_pbm_ranges = i / sizeof(*pbm_ranges);
+
+ for (i = 0; i < num_pbm_ranges; i++) {
+ const struct linux_prom_pci_ranges *pr = &pbm_ranges[i];
+ unsigned long a, size;
+ u32 parent_phys_hi, parent_phys_lo;
+ u32 size_hi, size_lo;
+ int type;
+
+ parent_phys_hi = pr->parent_phys_hi;
+ parent_phys_lo = pr->parent_phys_lo;
+ if (tlb_type == hypervisor)
+ parent_phys_hi &= 0x0fffffff;
+
+ size_hi = pr->size_hi;
+ size_lo = pr->size_lo;
+
+ type = (pr->child_phys_hi >> 24) & 0x3;
+ a = (((unsigned long)parent_phys_hi << 32UL) |
+ ((unsigned long)parent_phys_lo << 0UL));
+ size = (((unsigned long)size_hi << 32UL) |
+ ((unsigned long)size_lo << 0UL));
+
+ switch (type) {
+ case 0:
+ /* PCI config space, 16MB */
+ pbm->config_space = a;
+ break;
+
+ case 1:
+ /* 16-bit IO space, 16MB */
+ pbm->io_space.start = a;
+ pbm->io_space.end = a + size - 1UL;
+ pbm->io_space.flags = IORESOURCE_IO;
+ saw_io = 1;
+ break;
+
+ case 2:
+ /* 32-bit MEM space, 2GB */
+ pbm->mem_space.start = a;
+ pbm->mem_space.end = a + size - 1UL;
+ pbm->mem_space.flags = IORESOURCE_MEM;
+ saw_mem = 1;
+ break;
+
+ case 3:
+ /* XXX 64-bit MEM handling XXX */
+
+ default:
+ break;
+ }
+ }
+
+ if (!saw_io || !saw_mem) {
+ prom_printf("%s: Fatal error, missing %s PBM range.\n",
+ pbm->name,
+ (!saw_io ? "IO" : "MEM"));
+ prom_halt();
+ }
+
+ printk("%s: PCI IO[%llx] MEM[%llx]\n",
+ pbm->name,
+ pbm->io_space.start,
+ pbm->mem_space.start);
+
+ pbm->io_space.name = pbm->mem_space.name = pbm->name;
+
+ request_resource(&ioport_resource, &pbm->io_space);
+ request_resource(&iomem_resource, &pbm->mem_space);
+
+ pci_register_legacy_regions(&pbm->io_space,
+ &pbm->mem_space);
+ pci_register_iommu_region(pbm);
+}
+
+/* Generic helper routines for PCI error reporting. */
+void pci_scan_for_target_abort(struct pci_pbm_info *pbm,
+ struct pci_bus *pbus)
+{
+ struct pci_dev *pdev;
+ struct pci_bus *bus;
+
+ list_for_each_entry(pdev, &pbus->devices, bus_list) {
+ u16 status, error_bits;
+
+ pci_read_config_word(pdev, PCI_STATUS, &status);
+ error_bits =
+ (status & (PCI_STATUS_SIG_TARGET_ABORT |
+ PCI_STATUS_REC_TARGET_ABORT));
+ if (error_bits) {
+ pci_write_config_word(pdev, PCI_STATUS, error_bits);
+ printk("%s: Device %s saw Target Abort [%016x]\n",
+ pbm->name, pci_name(pdev), status);
+ }
+ }
+
+ list_for_each_entry(bus, &pbus->children, node)
+ pci_scan_for_target_abort(pbm, bus);
+}
+
+void pci_scan_for_master_abort(struct pci_pbm_info *pbm,
+ struct pci_bus *pbus)
+{
+ struct pci_dev *pdev;
+ struct pci_bus *bus;
+
+ list_for_each_entry(pdev, &pbus->devices, bus_list) {
+ u16 status, error_bits;
+
+ pci_read_config_word(pdev, PCI_STATUS, &status);
+ error_bits =
+ (status & (PCI_STATUS_REC_MASTER_ABORT));
+ if (error_bits) {
+ pci_write_config_word(pdev, PCI_STATUS, error_bits);
+ printk("%s: Device %s received Master Abort [%016x]\n",
+ pbm->name, pci_name(pdev), status);
+ }
+ }
+
+ list_for_each_entry(bus, &pbus->children, node)
+ pci_scan_for_master_abort(pbm, bus);
+}
+
+void pci_scan_for_parity_error(struct pci_pbm_info *pbm,
+ struct pci_bus *pbus)
+{
+ struct pci_dev *pdev;
+ struct pci_bus *bus;
+
+ list_for_each_entry(pdev, &pbus->devices, bus_list) {
+ u16 status, error_bits;
+
+ pci_read_config_word(pdev, PCI_STATUS, &status);
+ error_bits =
+ (status & (PCI_STATUS_PARITY |
+ PCI_STATUS_DETECTED_PARITY));
+ if (error_bits) {
+ pci_write_config_word(pdev, PCI_STATUS, error_bits);
+ printk("%s: Device %s saw Parity Error [%016x]\n",
+ pbm->name, pci_name(pdev), status);
+ }
+ }
+
+ list_for_each_entry(bus, &pbus->children, node)
+ pci_scan_for_parity_error(pbm, bus);
+}
diff --git a/arch/sparc/kernel/pci_fire.c b/arch/sparc/kernel/pci_fire.c
new file mode 100644
index 00000000000..e60fc6a67e9
--- /dev/null
+++ b/arch/sparc/kernel/pci_fire.c
@@ -0,0 +1,522 @@
+/* pci_fire.c: Sun4u platform PCI-E controller support.
+ *
+ * Copyright (C) 2007 David S. Miller (davem@davemloft.net)
+ */
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+#include <linux/msi.h>
+#include <linux/export.h>
+#include <linux/irq.h>
+#include <linux/of_device.h>
+
+#include <asm/prom.h>
+#include <asm/irq.h>
+#include <asm/upa.h>
+
+#include "pci_impl.h"
+
+#define DRIVER_NAME "fire"
+#define PFX DRIVER_NAME ": "
+
+#define FIRE_IOMMU_CONTROL 0x40000UL
+#define FIRE_IOMMU_TSBBASE 0x40008UL
+#define FIRE_IOMMU_FLUSH 0x40100UL
+#define FIRE_IOMMU_FLUSHINV 0x40108UL
+
+static int pci_fire_pbm_iommu_init(struct pci_pbm_info *pbm)
+{
+ struct iommu *iommu = pbm->iommu;
+ u32 vdma[2], dma_mask;
+ u64 control;
+ int tsbsize, err;
+
+ /* No virtual-dma property on these guys, use largest size. */
+ vdma[0] = 0xc0000000; /* base */
+ vdma[1] = 0x40000000; /* size */
+ dma_mask = 0xffffffff;
+ tsbsize = 128;
+
+ /* Register addresses. */
+ iommu->iommu_control = pbm->pbm_regs + FIRE_IOMMU_CONTROL;
+ iommu->iommu_tsbbase = pbm->pbm_regs + FIRE_IOMMU_TSBBASE;
+ iommu->iommu_flush = pbm->pbm_regs + FIRE_IOMMU_FLUSH;
+ iommu->iommu_flushinv = pbm->pbm_regs + FIRE_IOMMU_FLUSHINV;
+
+ /* We use the main control/status register of FIRE as the write
+ * completion register.
+ */
+ iommu->write_complete_reg = pbm->controller_regs + 0x410000UL;
+
+ /*
+ * Invalidate TLB Entries.
+ */
+ upa_writeq(~(u64)0, iommu->iommu_flushinv);
+
+ err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask,
+ pbm->numa_node);
+ if (err)
+ return err;
+
+ upa_writeq(__pa(iommu->page_table) | 0x7UL, iommu->iommu_tsbbase);
+
+ control = upa_readq(iommu->iommu_control);
+ control |= (0x00000400 /* TSB cache snoop enable */ |
+ 0x00000300 /* Cache mode */ |
+ 0x00000002 /* Bypass enable */ |
+ 0x00000001 /* Translation enable */);
+ upa_writeq(control, iommu->iommu_control);
+
+ return 0;
+}
+
+#ifdef CONFIG_PCI_MSI
+struct pci_msiq_entry {
+ u64 word0;
+#define MSIQ_WORD0_RESV 0x8000000000000000UL
+#define MSIQ_WORD0_FMT_TYPE 0x7f00000000000000UL
+#define MSIQ_WORD0_FMT_TYPE_SHIFT 56
+#define MSIQ_WORD0_LEN 0x00ffc00000000000UL
+#define MSIQ_WORD0_LEN_SHIFT 46
+#define MSIQ_WORD0_ADDR0 0x00003fff00000000UL
+#define MSIQ_WORD0_ADDR0_SHIFT 32
+#define MSIQ_WORD0_RID 0x00000000ffff0000UL
+#define MSIQ_WORD0_RID_SHIFT 16
+#define MSIQ_WORD0_DATA0 0x000000000000ffffUL
+#define MSIQ_WORD0_DATA0_SHIFT 0
+
+#define MSIQ_TYPE_MSG 0x6
+#define MSIQ_TYPE_MSI32 0xb
+#define MSIQ_TYPE_MSI64 0xf
+
+ u64 word1;
+#define MSIQ_WORD1_ADDR1 0xffffffffffff0000UL
+#define MSIQ_WORD1_ADDR1_SHIFT 16
+#define MSIQ_WORD1_DATA1 0x000000000000ffffUL
+#define MSIQ_WORD1_DATA1_SHIFT 0
+
+ u64 resv[6];
+};
+
+/* All MSI registers are offset from pbm->pbm_regs */
+#define EVENT_QUEUE_BASE_ADDR_REG 0x010000UL
+#define EVENT_QUEUE_BASE_ADDR_ALL_ONES 0xfffc000000000000UL
+
+#define EVENT_QUEUE_CONTROL_SET(EQ) (0x011000UL + (EQ) * 0x8UL)
+#define EVENT_QUEUE_CONTROL_SET_OFLOW 0x0200000000000000UL
+#define EVENT_QUEUE_CONTROL_SET_EN 0x0000100000000000UL
+
+#define EVENT_QUEUE_CONTROL_CLEAR(EQ) (0x011200UL + (EQ) * 0x8UL)
+#define EVENT_QUEUE_CONTROL_CLEAR_OF 0x0200000000000000UL
+#define EVENT_QUEUE_CONTROL_CLEAR_E2I 0x0000800000000000UL
+#define EVENT_QUEUE_CONTROL_CLEAR_DIS 0x0000100000000000UL
+
+#define EVENT_QUEUE_STATE(EQ) (0x011400UL + (EQ) * 0x8UL)
+#define EVENT_QUEUE_STATE_MASK 0x0000000000000007UL
+#define EVENT_QUEUE_STATE_IDLE 0x0000000000000001UL
+#define EVENT_QUEUE_STATE_ACTIVE 0x0000000000000002UL
+#define EVENT_QUEUE_STATE_ERROR 0x0000000000000004UL
+
+#define EVENT_QUEUE_TAIL(EQ) (0x011600UL + (EQ) * 0x8UL)
+#define EVENT_QUEUE_TAIL_OFLOW 0x0200000000000000UL
+#define EVENT_QUEUE_TAIL_VAL 0x000000000000007fUL
+
+#define EVENT_QUEUE_HEAD(EQ) (0x011800UL + (EQ) * 0x8UL)
+#define EVENT_QUEUE_HEAD_VAL 0x000000000000007fUL
+
+#define MSI_MAP(MSI) (0x020000UL + (MSI) * 0x8UL)
+#define MSI_MAP_VALID 0x8000000000000000UL
+#define MSI_MAP_EQWR_N 0x4000000000000000UL
+#define MSI_MAP_EQNUM 0x000000000000003fUL
+
+#define MSI_CLEAR(MSI) (0x028000UL + (MSI) * 0x8UL)
+#define MSI_CLEAR_EQWR_N 0x4000000000000000UL
+
+#define IMONDO_DATA0 0x02C000UL
+#define IMONDO_DATA0_DATA 0xffffffffffffffc0UL
+
+#define IMONDO_DATA1 0x02C008UL
+#define IMONDO_DATA1_DATA 0xffffffffffffffffUL
+
+#define MSI_32BIT_ADDR 0x034000UL
+#define MSI_32BIT_ADDR_VAL 0x00000000ffff0000UL
+
+#define MSI_64BIT_ADDR 0x034008UL
+#define MSI_64BIT_ADDR_VAL 0xffffffffffff0000UL
+
+static int pci_fire_get_head(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long *head)
+{
+ *head = upa_readq(pbm->pbm_regs + EVENT_QUEUE_HEAD(msiqid));
+ return 0;
+}
+
+static int pci_fire_dequeue_msi(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long *head, unsigned long *msi)
+{
+ unsigned long type_fmt, type, msi_num;
+ struct pci_msiq_entry *base, *ep;
+
+ base = (pbm->msi_queues + ((msiqid - pbm->msiq_first) * 8192));
+ ep = &base[*head];
+
+ if ((ep->word0 & MSIQ_WORD0_FMT_TYPE) == 0)
+ return 0;
+
+ type_fmt = ((ep->word0 & MSIQ_WORD0_FMT_TYPE) >>
+ MSIQ_WORD0_FMT_TYPE_SHIFT);
+ type = (type_fmt >> 3);
+ if (unlikely(type != MSIQ_TYPE_MSI32 &&
+ type != MSIQ_TYPE_MSI64))
+ return -EINVAL;
+
+ *msi = msi_num = ((ep->word0 & MSIQ_WORD0_DATA0) >>
+ MSIQ_WORD0_DATA0_SHIFT);
+
+ upa_writeq(MSI_CLEAR_EQWR_N, pbm->pbm_regs + MSI_CLEAR(msi_num));
+
+ /* Clear the entry. */
+ ep->word0 &= ~MSIQ_WORD0_FMT_TYPE;
+
+ /* Go to next entry in ring. */
+ (*head)++;
+ if (*head >= pbm->msiq_ent_count)
+ *head = 0;
+
+ return 1;
+}
+
+static int pci_fire_set_head(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long head)
+{
+ upa_writeq(head, pbm->pbm_regs + EVENT_QUEUE_HEAD(msiqid));
+ return 0;
+}
+
+static int pci_fire_msi_setup(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long msi, int is_msi64)
+{
+ u64 val;
+
+ val = upa_readq(pbm->pbm_regs + MSI_MAP(msi));
+ val &= ~(MSI_MAP_EQNUM);
+ val |= msiqid;
+ upa_writeq(val, pbm->pbm_regs + MSI_MAP(msi));
+
+ upa_writeq(MSI_CLEAR_EQWR_N, pbm->pbm_regs + MSI_CLEAR(msi));
+
+ val = upa_readq(pbm->pbm_regs + MSI_MAP(msi));
+ val |= MSI_MAP_VALID;
+ upa_writeq(val, pbm->pbm_regs + MSI_MAP(msi));
+
+ return 0;
+}
+
+static int pci_fire_msi_teardown(struct pci_pbm_info *pbm, unsigned long msi)
+{
+ u64 val;
+
+ val = upa_readq(pbm->pbm_regs + MSI_MAP(msi));
+
+ val &= ~MSI_MAP_VALID;
+
+ upa_writeq(val, pbm->pbm_regs + MSI_MAP(msi));
+
+ return 0;
+}
+
+static int pci_fire_msiq_alloc(struct pci_pbm_info *pbm)
+{
+ unsigned long pages, order, i;
+
+ order = get_order(512 * 1024);
+ pages = __get_free_pages(GFP_KERNEL | __GFP_COMP, order);
+ if (pages == 0UL) {
+ printk(KERN_ERR "MSI: Cannot allocate MSI queues (o=%lu).\n",
+ order);
+ return -ENOMEM;
+ }
+ memset((char *)pages, 0, PAGE_SIZE << order);
+ pbm->msi_queues = (void *) pages;
+
+ upa_writeq((EVENT_QUEUE_BASE_ADDR_ALL_ONES |
+ __pa(pbm->msi_queues)),
+ pbm->pbm_regs + EVENT_QUEUE_BASE_ADDR_REG);
+
+ upa_writeq(pbm->portid << 6, pbm->pbm_regs + IMONDO_DATA0);
+ upa_writeq(0, pbm->pbm_regs + IMONDO_DATA1);
+
+ upa_writeq(pbm->msi32_start, pbm->pbm_regs + MSI_32BIT_ADDR);
+ upa_writeq(pbm->msi64_start, pbm->pbm_regs + MSI_64BIT_ADDR);
+
+ for (i = 0; i < pbm->msiq_num; i++) {
+ upa_writeq(0, pbm->pbm_regs + EVENT_QUEUE_HEAD(i));
+ upa_writeq(0, pbm->pbm_regs + EVENT_QUEUE_TAIL(i));
+ }
+
+ return 0;
+}
+
+static void pci_fire_msiq_free(struct pci_pbm_info *pbm)
+{
+ unsigned long pages, order;
+
+ order = get_order(512 * 1024);
+ pages = (unsigned long) pbm->msi_queues;
+
+ free_pages(pages, order);
+
+ pbm->msi_queues = NULL;
+}
+
+static int pci_fire_msiq_build_irq(struct pci_pbm_info *pbm,
+ unsigned long msiqid,
+ unsigned long devino)
+{
+ unsigned long cregs = (unsigned long) pbm->pbm_regs;
+ unsigned long imap_reg, iclr_reg, int_ctrlr;
+ unsigned int irq;
+ int fixup;
+ u64 val;
+
+ imap_reg = cregs + (0x001000UL + (devino * 0x08UL));
+ iclr_reg = cregs + (0x001400UL + (devino * 0x08UL));
+
+ /* XXX iterate amongst the 4 IRQ controllers XXX */
+ int_ctrlr = (1UL << 6);
+
+ val = upa_readq(imap_reg);
+ val |= (1UL << 63) | int_ctrlr;
+ upa_writeq(val, imap_reg);
+
+ fixup = ((pbm->portid << 6) | devino) - int_ctrlr;
+
+ irq = build_irq(fixup, iclr_reg, imap_reg);
+ if (!irq)
+ return -ENOMEM;
+
+ upa_writeq(EVENT_QUEUE_CONTROL_SET_EN,
+ pbm->pbm_regs + EVENT_QUEUE_CONTROL_SET(msiqid));
+
+ return irq;
+}
+
+static const struct sparc64_msiq_ops pci_fire_msiq_ops = {
+ .get_head = pci_fire_get_head,
+ .dequeue_msi = pci_fire_dequeue_msi,
+ .set_head = pci_fire_set_head,
+ .msi_setup = pci_fire_msi_setup,
+ .msi_teardown = pci_fire_msi_teardown,
+ .msiq_alloc = pci_fire_msiq_alloc,
+ .msiq_free = pci_fire_msiq_free,
+ .msiq_build_irq = pci_fire_msiq_build_irq,
+};
+
+static void pci_fire_msi_init(struct pci_pbm_info *pbm)
+{
+ sparc64_pbm_msi_init(pbm, &pci_fire_msiq_ops);
+}
+#else /* CONFIG_PCI_MSI */
+static void pci_fire_msi_init(struct pci_pbm_info *pbm)
+{
+}
+#endif /* !(CONFIG_PCI_MSI) */
+
+/* Based at pbm->controller_regs */
+#define FIRE_PARITY_CONTROL 0x470010UL
+#define FIRE_PARITY_ENAB 0x8000000000000000UL
+#define FIRE_FATAL_RESET_CTL 0x471028UL
+#define FIRE_FATAL_RESET_SPARE 0x0000000004000000UL
+#define FIRE_FATAL_RESET_MB 0x0000000002000000UL
+#define FIRE_FATAL_RESET_CPE 0x0000000000008000UL
+#define FIRE_FATAL_RESET_APE 0x0000000000004000UL
+#define FIRE_FATAL_RESET_PIO 0x0000000000000040UL
+#define FIRE_FATAL_RESET_JW 0x0000000000000004UL
+#define FIRE_FATAL_RESET_JI 0x0000000000000002UL
+#define FIRE_FATAL_RESET_JR 0x0000000000000001UL
+#define FIRE_CORE_INTR_ENABLE 0x471800UL
+
+/* Based at pbm->pbm_regs */
+#define FIRE_TLU_CTRL 0x80000UL
+#define FIRE_TLU_CTRL_TIM 0x00000000da000000UL
+#define FIRE_TLU_CTRL_QDET 0x0000000000000100UL
+#define FIRE_TLU_CTRL_CFG 0x0000000000000001UL
+#define FIRE_TLU_DEV_CTRL 0x90008UL
+#define FIRE_TLU_LINK_CTRL 0x90020UL
+#define FIRE_TLU_LINK_CTRL_CLK 0x0000000000000040UL
+#define FIRE_LPU_RESET 0xe2008UL
+#define FIRE_LPU_LLCFG 0xe2200UL
+#define FIRE_LPU_LLCFG_VC0 0x0000000000000100UL
+#define FIRE_LPU_FCTRL_UCTRL 0xe2240UL
+#define FIRE_LPU_FCTRL_UCTRL_N 0x0000000000000002UL
+#define FIRE_LPU_FCTRL_UCTRL_P 0x0000000000000001UL
+#define FIRE_LPU_TXL_FIFOP 0xe2430UL
+#define FIRE_LPU_LTSSM_CFG2 0xe2788UL
+#define FIRE_LPU_LTSSM_CFG3 0xe2790UL
+#define FIRE_LPU_LTSSM_CFG4 0xe2798UL
+#define FIRE_LPU_LTSSM_CFG5 0xe27a0UL
+#define FIRE_DMC_IENAB 0x31800UL
+#define FIRE_DMC_DBG_SEL_A 0x53000UL
+#define FIRE_DMC_DBG_SEL_B 0x53008UL
+#define FIRE_PEC_IENAB 0x51800UL
+
+static void pci_fire_hw_init(struct pci_pbm_info *pbm)
+{
+ u64 val;
+
+ upa_writeq(FIRE_PARITY_ENAB,
+ pbm->controller_regs + FIRE_PARITY_CONTROL);
+
+ upa_writeq((FIRE_FATAL_RESET_SPARE |
+ FIRE_FATAL_RESET_MB |
+ FIRE_FATAL_RESET_CPE |
+ FIRE_FATAL_RESET_APE |
+ FIRE_FATAL_RESET_PIO |
+ FIRE_FATAL_RESET_JW |
+ FIRE_FATAL_RESET_JI |
+ FIRE_FATAL_RESET_JR),
+ pbm->controller_regs + FIRE_FATAL_RESET_CTL);
+
+ upa_writeq(~(u64)0, pbm->controller_regs + FIRE_CORE_INTR_ENABLE);
+
+ val = upa_readq(pbm->pbm_regs + FIRE_TLU_CTRL);
+ val |= (FIRE_TLU_CTRL_TIM |
+ FIRE_TLU_CTRL_QDET |
+ FIRE_TLU_CTRL_CFG);
+ upa_writeq(val, pbm->pbm_regs + FIRE_TLU_CTRL);
+ upa_writeq(0, pbm->pbm_regs + FIRE_TLU_DEV_CTRL);
+ upa_writeq(FIRE_TLU_LINK_CTRL_CLK,
+ pbm->pbm_regs + FIRE_TLU_LINK_CTRL);
+
+ upa_writeq(0, pbm->pbm_regs + FIRE_LPU_RESET);
+ upa_writeq(FIRE_LPU_LLCFG_VC0, pbm->pbm_regs + FIRE_LPU_LLCFG);
+ upa_writeq((FIRE_LPU_FCTRL_UCTRL_N | FIRE_LPU_FCTRL_UCTRL_P),
+ pbm->pbm_regs + FIRE_LPU_FCTRL_UCTRL);
+ upa_writeq(((0xffff << 16) | (0x0000 << 0)),
+ pbm->pbm_regs + FIRE_LPU_TXL_FIFOP);
+ upa_writeq(3000000, pbm->pbm_regs + FIRE_LPU_LTSSM_CFG2);
+ upa_writeq(500000, pbm->pbm_regs + FIRE_LPU_LTSSM_CFG3);
+ upa_writeq((2 << 16) | (140 << 8),
+ pbm->pbm_regs + FIRE_LPU_LTSSM_CFG4);
+ upa_writeq(0, pbm->pbm_regs + FIRE_LPU_LTSSM_CFG5);
+
+ upa_writeq(~(u64)0, pbm->pbm_regs + FIRE_DMC_IENAB);
+ upa_writeq(0, pbm->pbm_regs + FIRE_DMC_DBG_SEL_A);
+ upa_writeq(0, pbm->pbm_regs + FIRE_DMC_DBG_SEL_B);
+
+ upa_writeq(~(u64)0, pbm->pbm_regs + FIRE_PEC_IENAB);
+}
+
+static int pci_fire_pbm_init(struct pci_pbm_info *pbm,
+ struct platform_device *op, u32 portid)
+{
+ const struct linux_prom64_registers *regs;
+ struct device_node *dp = op->dev.of_node;
+ int err;
+
+ pbm->numa_node = -1;
+
+ pbm->pci_ops = &sun4u_pci_ops;
+ pbm->config_space_reg_bits = 12;
+
+ pbm->index = pci_num_pbms++;
+
+ pbm->portid = portid;
+ pbm->op = op;
+ pbm->name = dp->full_name;
+
+ regs = of_get_property(dp, "reg", NULL);
+ pbm->pbm_regs = regs[0].phys_addr;
+ pbm->controller_regs = regs[1].phys_addr - 0x410000UL;
+
+ printk("%s: SUN4U PCIE Bus Module\n", pbm->name);
+
+ pci_determine_mem_io_space(pbm);
+
+ pci_get_pbm_props(pbm);
+
+ pci_fire_hw_init(pbm);
+
+ err = pci_fire_pbm_iommu_init(pbm);
+ if (err)
+ return err;
+
+ pci_fire_msi_init(pbm);
+
+ pbm->pci_bus = pci_scan_one_pbm(pbm, &op->dev);
+
+ /* XXX register error interrupt handlers XXX */
+
+ pbm->next = pci_pbm_root;
+ pci_pbm_root = pbm;
+
+ return 0;
+}
+
+static int fire_probe(struct platform_device *op)
+{
+ struct device_node *dp = op->dev.of_node;
+ struct pci_pbm_info *pbm;
+ struct iommu *iommu;
+ u32 portid;
+ int err;
+
+ portid = of_getintprop_default(dp, "portid", 0xff);
+
+ err = -ENOMEM;
+ pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+ if (!pbm) {
+ printk(KERN_ERR PFX "Cannot allocate pci_pbminfo.\n");
+ goto out_err;
+ }
+
+ iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+ if (!iommu) {
+ printk(KERN_ERR PFX "Cannot allocate PBM iommu.\n");
+ goto out_free_controller;
+ }
+
+ pbm->iommu = iommu;
+
+ err = pci_fire_pbm_init(pbm, op, portid);
+ if (err)
+ goto out_free_iommu;
+
+ dev_set_drvdata(&op->dev, pbm);
+
+ return 0;
+
+out_free_iommu:
+ kfree(pbm->iommu);
+
+out_free_controller:
+ kfree(pbm);
+
+out_err:
+ return err;
+}
+
+static const struct of_device_id fire_match[] = {
+ {
+ .name = "pci",
+ .compatible = "pciex108e,80f0",
+ },
+ {},
+};
+
+static struct platform_driver fire_driver = {
+ .driver = {
+ .name = DRIVER_NAME,
+ .owner = THIS_MODULE,
+ .of_match_table = fire_match,
+ },
+ .probe = fire_probe,
+};
+
+static int __init fire_init(void)
+{
+ return platform_driver_register(&fire_driver);
+}
+
+subsys_initcall(fire_init);
diff --git a/arch/sparc/kernel/pci_impl.h b/arch/sparc/kernel/pci_impl.h
new file mode 100644
index 00000000000..75803c780af
--- /dev/null
+++ b/arch/sparc/kernel/pci_impl.h
@@ -0,0 +1,186 @@
+/* pci_impl.h: Helper definitions for PCI controller support.
+ *
+ * Copyright (C) 1999, 2007 David S. Miller (davem@davemloft.net)
+ */
+
+#ifndef PCI_IMPL_H
+#define PCI_IMPL_H
+
+#include <linux/types.h>
+#include <linux/spinlock.h>
+#include <linux/pci.h>
+#include <linux/msi.h>
+#include <linux/of_device.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/iommu.h>
+
+/* The abstraction used here is that there are PCI controllers,
+ * each with one (Sabre) or two (PSYCHO/SCHIZO) PCI bus modules
+ * underneath. Each PCI bus module uses an IOMMU (shared by both
+ * PBMs of a controller, or per-PBM), and if a streaming buffer
+ * is present, each PCI bus module has it's own. (ie. the IOMMU
+ * might be shared between PBMs, the STC is never shared)
+ * Furthermore, each PCI bus module controls it's own autonomous
+ * PCI bus.
+ */
+
+#define PCI_STC_FLUSHFLAG_INIT(STC) \
+ (*((STC)->strbuf_flushflag) = 0UL)
+#define PCI_STC_FLUSHFLAG_SET(STC) \
+ (*((STC)->strbuf_flushflag) != 0UL)
+
+#ifdef CONFIG_PCI_MSI
+struct pci_pbm_info;
+struct sparc64_msiq_ops {
+ int (*get_head)(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long *head);
+ int (*dequeue_msi)(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long *head, unsigned long *msi);
+ int (*set_head)(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long head);
+ int (*msi_setup)(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long msi, int is_msi64);
+ int (*msi_teardown)(struct pci_pbm_info *pbm, unsigned long msi);
+ int (*msiq_alloc)(struct pci_pbm_info *pbm);
+ void (*msiq_free)(struct pci_pbm_info *pbm);
+ int (*msiq_build_irq)(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long devino);
+};
+
+void sparc64_pbm_msi_init(struct pci_pbm_info *pbm,
+ const struct sparc64_msiq_ops *ops);
+
+struct sparc64_msiq_cookie {
+ struct pci_pbm_info *pbm;
+ unsigned long msiqid;
+};
+#endif
+
+struct pci_pbm_info {
+ struct pci_pbm_info *next;
+ struct pci_pbm_info *sibling;
+ int index;
+
+ /* Physical address base of controller registers. */
+ unsigned long controller_regs;
+
+ /* Physical address base of PBM registers. */
+ unsigned long pbm_regs;
+
+ /* Physical address of DMA sync register, if any. */
+ unsigned long sync_reg;
+
+ /* Opaque 32-bit system bus Port ID. */
+ u32 portid;
+
+ /* Opaque 32-bit handle used for hypervisor calls. */
+ u32 devhandle;
+
+ /* Chipset version information. */
+ int chip_type;
+#define PBM_CHIP_TYPE_SABRE 1
+#define PBM_CHIP_TYPE_PSYCHO 2
+#define PBM_CHIP_TYPE_SCHIZO 3
+#define PBM_CHIP_TYPE_SCHIZO_PLUS 4
+#define PBM_CHIP_TYPE_TOMATILLO 5
+ int chip_version;
+ int chip_revision;
+
+ /* Name used for top-level resources. */
+ const char *name;
+
+ /* OBP specific information. */
+ struct platform_device *op;
+ u64 ino_bitmap;
+
+ /* PBM I/O and Memory space resources. */
+ struct resource io_space;
+ struct resource mem_space;
+ struct resource busn;
+
+ /* Base of PCI Config space, can be per-PBM or shared. */
+ unsigned long config_space;
+
+ /* This will be 12 on PCI-E controllers, 8 elsewhere. */
+ unsigned long config_space_reg_bits;
+
+ unsigned long pci_afsr;
+ unsigned long pci_afar;
+ unsigned long pci_csr;
+
+ /* State of 66MHz capabilities on this PBM. */
+ int is_66mhz_capable;
+ int all_devs_66mhz;
+
+#ifdef CONFIG_PCI_MSI
+ /* MSI info. */
+ u32 msiq_num;
+ u32 msiq_ent_count;
+ u32 msiq_first;
+ u32 msiq_first_devino;
+ u32 msiq_rotor;
+ struct sparc64_msiq_cookie *msiq_irq_cookies;
+ u32 msi_num;
+ u32 msi_first;
+ u32 msi_data_mask;
+ u32 msix_data_width;
+ u64 msi32_start;
+ u64 msi64_start;
+ u32 msi32_len;
+ u32 msi64_len;
+ void *msi_queues;
+ unsigned long *msi_bitmap;
+ unsigned int *msi_irq_table;
+ int (*setup_msi_irq)(unsigned int *irq_p, struct pci_dev *pdev,
+ struct msi_desc *entry);
+ void (*teardown_msi_irq)(unsigned int irq, struct pci_dev *pdev);
+ const struct sparc64_msiq_ops *msi_ops;
+#endif /* !(CONFIG_PCI_MSI) */
+
+ /* This PBM's streaming buffer. */
+ struct strbuf stc;
+
+ /* IOMMU state, potentially shared by both PBM segments. */
+ struct iommu *iommu;
+
+ /* Now things for the actual PCI bus probes. */
+ unsigned int pci_first_busno;
+ unsigned int pci_last_busno;
+ struct pci_bus *pci_bus;
+ struct pci_ops *pci_ops;
+
+ int numa_node;
+};
+
+extern struct pci_pbm_info *pci_pbm_root;
+
+extern int pci_num_pbms;
+
+/* PCI bus scanning and fixup support. */
+void pci_get_pbm_props(struct pci_pbm_info *pbm);
+struct pci_bus *pci_scan_one_pbm(struct pci_pbm_info *pbm,
+ struct device *parent);
+void pci_determine_mem_io_space(struct pci_pbm_info *pbm);
+
+/* Error reporting support. */
+void pci_scan_for_target_abort(struct pci_pbm_info *, struct pci_bus *);
+void pci_scan_for_master_abort(struct pci_pbm_info *, struct pci_bus *);
+void pci_scan_for_parity_error(struct pci_pbm_info *, struct pci_bus *);
+
+/* Configuration space access. */
+void pci_config_read8(u8 *addr, u8 *ret);
+void pci_config_read16(u16 *addr, u16 *ret);
+void pci_config_read32(u32 *addr, u32 *ret);
+void pci_config_write8(u8 *addr, u8 val);
+void pci_config_write16(u16 *addr, u16 val);
+void pci_config_write32(u32 *addr, u32 val);
+
+extern struct pci_ops sun4u_pci_ops;
+extern struct pci_ops sun4v_pci_ops;
+
+extern volatile int pci_poke_in_progress;
+extern volatile int pci_poke_cpu;
+extern volatile int pci_poke_faulted;
+
+#endif /* !(PCI_IMPL_H) */
diff --git a/arch/sparc/kernel/pci_msi.c b/arch/sparc/kernel/pci_msi.c
new file mode 100644
index 00000000000..580651af73f
--- /dev/null
+++ b/arch/sparc/kernel/pci_msi.c
@@ -0,0 +1,446 @@
+/* pci_msi.c: Sparc64 MSI support common layer.
+ *
+ * Copyright (C) 2007 David S. Miller (davem@davemloft.net)
+ */
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+#include <linux/slab.h>
+#include <linux/irq.h>
+
+#include "pci_impl.h"
+
+static irqreturn_t sparc64_msiq_interrupt(int irq, void *cookie)
+{
+ struct sparc64_msiq_cookie *msiq_cookie = cookie;
+ struct pci_pbm_info *pbm = msiq_cookie->pbm;
+ unsigned long msiqid = msiq_cookie->msiqid;
+ const struct sparc64_msiq_ops *ops;
+ unsigned long orig_head, head;
+ int err;
+
+ ops = pbm->msi_ops;
+
+ err = ops->get_head(pbm, msiqid, &head);
+ if (unlikely(err < 0))
+ goto err_get_head;
+
+ orig_head = head;
+ for (;;) {
+ unsigned long msi;
+
+ err = ops->dequeue_msi(pbm, msiqid, &head, &msi);
+ if (likely(err > 0)) {
+ unsigned int irq;
+
+ irq = pbm->msi_irq_table[msi - pbm->msi_first];
+ generic_handle_irq(irq);
+ }
+
+ if (unlikely(err < 0))
+ goto err_dequeue;
+
+ if (err == 0)
+ break;
+ }
+ if (likely(head != orig_head)) {
+ err = ops->set_head(pbm, msiqid, head);
+ if (unlikely(err < 0))
+ goto err_set_head;
+ }
+ return IRQ_HANDLED;
+
+err_get_head:
+ printk(KERN_EMERG "MSI: Get head on msiqid[%lu] gives error %d\n",
+ msiqid, err);
+ goto err_out;
+
+err_dequeue:
+ printk(KERN_EMERG "MSI: Dequeue head[%lu] from msiqid[%lu] "
+ "gives error %d\n",
+ head, msiqid, err);
+ goto err_out;
+
+err_set_head:
+ printk(KERN_EMERG "MSI: Set head[%lu] on msiqid[%lu] "
+ "gives error %d\n",
+ head, msiqid, err);
+ goto err_out;
+
+err_out:
+ return IRQ_NONE;
+}
+
+static u32 pick_msiq(struct pci_pbm_info *pbm)
+{
+ static DEFINE_SPINLOCK(rotor_lock);
+ unsigned long flags;
+ u32 ret, rotor;
+
+ spin_lock_irqsave(&rotor_lock, flags);
+
+ rotor = pbm->msiq_rotor;
+ ret = pbm->msiq_first + rotor;
+
+ if (++rotor >= pbm->msiq_num)
+ rotor = 0;
+ pbm->msiq_rotor = rotor;
+
+ spin_unlock_irqrestore(&rotor_lock, flags);
+
+ return ret;
+}
+
+
+static int alloc_msi(struct pci_pbm_info *pbm)
+{
+ int i;
+
+ for (i = 0; i < pbm->msi_num; i++) {
+ if (!test_and_set_bit(i, pbm->msi_bitmap))
+ return i + pbm->msi_first;
+ }
+
+ return -ENOENT;
+}
+
+static void free_msi(struct pci_pbm_info *pbm, int msi_num)
+{
+ msi_num -= pbm->msi_first;
+ clear_bit(msi_num, pbm->msi_bitmap);
+}
+
+static struct irq_chip msi_irq = {
+ .name = "PCI-MSI",
+ .irq_mask = mask_msi_irq,
+ .irq_unmask = unmask_msi_irq,
+ .irq_enable = unmask_msi_irq,
+ .irq_disable = mask_msi_irq,
+ /* XXX affinity XXX */
+};
+
+static int sparc64_setup_msi_irq(unsigned int *irq_p,
+ struct pci_dev *pdev,
+ struct msi_desc *entry)
+{
+ struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
+ const struct sparc64_msiq_ops *ops = pbm->msi_ops;
+ struct msi_msg msg;
+ int msi, err;
+ u32 msiqid;
+
+ *irq_p = irq_alloc(0, 0);
+ err = -ENOMEM;
+ if (!*irq_p)
+ goto out_err;
+
+ irq_set_chip_and_handler_name(*irq_p, &msi_irq, handle_simple_irq,
+ "MSI");
+
+ err = alloc_msi(pbm);
+ if (unlikely(err < 0))
+ goto out_irq_free;
+
+ msi = err;
+
+ msiqid = pick_msiq(pbm);
+
+ err = ops->msi_setup(pbm, msiqid, msi,
+ (entry->msi_attrib.is_64 ? 1 : 0));
+ if (err)
+ goto out_msi_free;
+
+ pbm->msi_irq_table[msi - pbm->msi_first] = *irq_p;
+
+ if (entry->msi_attrib.is_64) {
+ msg.address_hi = pbm->msi64_start >> 32;
+ msg.address_lo = pbm->msi64_start & 0xffffffff;
+ } else {
+ msg.address_hi = 0;
+ msg.address_lo = pbm->msi32_start;
+ }
+ msg.data = msi;
+
+ irq_set_msi_desc(*irq_p, entry);
+ write_msi_msg(*irq_p, &msg);
+
+ return 0;
+
+out_msi_free:
+ free_msi(pbm, msi);
+
+out_irq_free:
+ irq_set_chip(*irq_p, NULL);
+ irq_free(*irq_p);
+ *irq_p = 0;
+
+out_err:
+ return err;
+}
+
+static void sparc64_teardown_msi_irq(unsigned int irq,
+ struct pci_dev *pdev)
+{
+ struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
+ const struct sparc64_msiq_ops *ops = pbm->msi_ops;
+ unsigned int msi_num;
+ int i, err;
+
+ for (i = 0; i < pbm->msi_num; i++) {
+ if (pbm->msi_irq_table[i] == irq)
+ break;
+ }
+ if (i >= pbm->msi_num) {
+ printk(KERN_ERR "%s: teardown: No MSI for irq %u\n",
+ pbm->name, irq);
+ return;
+ }
+
+ msi_num = pbm->msi_first + i;
+ pbm->msi_irq_table[i] = ~0U;
+
+ err = ops->msi_teardown(pbm, msi_num);
+ if (err) {
+ printk(KERN_ERR "%s: teardown: ops->teardown() on MSI %u, "
+ "irq %u, gives error %d\n",
+ pbm->name, msi_num, irq, err);
+ return;
+ }
+
+ free_msi(pbm, msi_num);
+
+ irq_set_chip(irq, NULL);
+ irq_free(irq);
+}
+
+static int msi_bitmap_alloc(struct pci_pbm_info *pbm)
+{
+ unsigned long size, bits_per_ulong;
+
+ bits_per_ulong = sizeof(unsigned long) * 8;
+ size = (pbm->msi_num + (bits_per_ulong - 1)) & ~(bits_per_ulong - 1);
+ size /= 8;
+ BUG_ON(size % sizeof(unsigned long));
+
+ pbm->msi_bitmap = kzalloc(size, GFP_KERNEL);
+ if (!pbm->msi_bitmap)
+ return -ENOMEM;
+
+ return 0;
+}
+
+static void msi_bitmap_free(struct pci_pbm_info *pbm)
+{
+ kfree(pbm->msi_bitmap);
+ pbm->msi_bitmap = NULL;
+}
+
+static int msi_table_alloc(struct pci_pbm_info *pbm)
+{
+ int size, i;
+
+ size = pbm->msiq_num * sizeof(struct sparc64_msiq_cookie);
+ pbm->msiq_irq_cookies = kzalloc(size, GFP_KERNEL);
+ if (!pbm->msiq_irq_cookies)
+ return -ENOMEM;
+
+ for (i = 0; i < pbm->msiq_num; i++) {
+ struct sparc64_msiq_cookie *p;
+
+ p = &pbm->msiq_irq_cookies[i];
+ p->pbm = pbm;
+ p->msiqid = pbm->msiq_first + i;
+ }
+
+ size = pbm->msi_num * sizeof(unsigned int);
+ pbm->msi_irq_table = kzalloc(size, GFP_KERNEL);
+ if (!pbm->msi_irq_table) {
+ kfree(pbm->msiq_irq_cookies);
+ pbm->msiq_irq_cookies = NULL;
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+static void msi_table_free(struct pci_pbm_info *pbm)
+{
+ kfree(pbm->msiq_irq_cookies);
+ pbm->msiq_irq_cookies = NULL;
+
+ kfree(pbm->msi_irq_table);
+ pbm->msi_irq_table = NULL;
+}
+
+static int bringup_one_msi_queue(struct pci_pbm_info *pbm,
+ const struct sparc64_msiq_ops *ops,
+ unsigned long msiqid,
+ unsigned long devino)
+{
+ int irq = ops->msiq_build_irq(pbm, msiqid, devino);
+ int err, nid;
+
+ if (irq < 0)
+ return irq;
+
+ nid = pbm->numa_node;
+ if (nid != -1) {
+ cpumask_t numa_mask;
+
+ cpumask_copy(&numa_mask, cpumask_of_node(nid));
+ irq_set_affinity(irq, &numa_mask);
+ }
+ err = request_irq(irq, sparc64_msiq_interrupt, 0,
+ "MSIQ",
+ &pbm->msiq_irq_cookies[msiqid - pbm->msiq_first]);
+ if (err)
+ return err;
+
+ return 0;
+}
+
+static int sparc64_bringup_msi_queues(struct pci_pbm_info *pbm,
+ const struct sparc64_msiq_ops *ops)
+{
+ int i;
+
+ for (i = 0; i < pbm->msiq_num; i++) {
+ unsigned long msiqid = i + pbm->msiq_first;
+ unsigned long devino = i + pbm->msiq_first_devino;
+ int err;
+
+ err = bringup_one_msi_queue(pbm, ops, msiqid, devino);
+ if (err)
+ return err;
+ }
+
+ return 0;
+}
+
+void sparc64_pbm_msi_init(struct pci_pbm_info *pbm,
+ const struct sparc64_msiq_ops *ops)
+{
+ const u32 *val;
+ int len;
+
+ val = of_get_property(pbm->op->dev.of_node, "#msi-eqs", &len);
+ if (!val || len != 4)
+ goto no_msi;
+ pbm->msiq_num = *val;
+ if (pbm->msiq_num) {
+ const struct msiq_prop {
+ u32 first_msiq;
+ u32 num_msiq;
+ u32 first_devino;
+ } *mqp;
+ const struct msi_range_prop {
+ u32 first_msi;
+ u32 num_msi;
+ } *mrng;
+ const struct addr_range_prop {
+ u32 msi32_high;
+ u32 msi32_low;
+ u32 msi32_len;
+ u32 msi64_high;
+ u32 msi64_low;
+ u32 msi64_len;
+ } *arng;
+
+ val = of_get_property(pbm->op->dev.of_node, "msi-eq-size", &len);
+ if (!val || len != 4)
+ goto no_msi;
+
+ pbm->msiq_ent_count = *val;
+
+ mqp = of_get_property(pbm->op->dev.of_node,
+ "msi-eq-to-devino", &len);
+ if (!mqp)
+ mqp = of_get_property(pbm->op->dev.of_node,
+ "msi-eq-devino", &len);
+ if (!mqp || len != sizeof(struct msiq_prop))
+ goto no_msi;
+
+ pbm->msiq_first = mqp->first_msiq;
+ pbm->msiq_first_devino = mqp->first_devino;
+
+ val = of_get_property(pbm->op->dev.of_node, "#msi", &len);
+ if (!val || len != 4)
+ goto no_msi;
+ pbm->msi_num = *val;
+
+ mrng = of_get_property(pbm->op->dev.of_node, "msi-ranges", &len);
+ if (!mrng || len != sizeof(struct msi_range_prop))
+ goto no_msi;
+ pbm->msi_first = mrng->first_msi;
+
+ val = of_get_property(pbm->op->dev.of_node, "msi-data-mask", &len);
+ if (!val || len != 4)
+ goto no_msi;
+ pbm->msi_data_mask = *val;
+
+ val = of_get_property(pbm->op->dev.of_node, "msix-data-width", &len);
+ if (!val || len != 4)
+ goto no_msi;
+ pbm->msix_data_width = *val;
+
+ arng = of_get_property(pbm->op->dev.of_node, "msi-address-ranges",
+ &len);
+ if (!arng || len != sizeof(struct addr_range_prop))
+ goto no_msi;
+ pbm->msi32_start = ((u64)arng->msi32_high << 32) |
+ (u64) arng->msi32_low;
+ pbm->msi64_start = ((u64)arng->msi64_high << 32) |
+ (u64) arng->msi64_low;
+ pbm->msi32_len = arng->msi32_len;
+ pbm->msi64_len = arng->msi64_len;
+
+ if (msi_bitmap_alloc(pbm))
+ goto no_msi;
+
+ if (msi_table_alloc(pbm)) {
+ msi_bitmap_free(pbm);
+ goto no_msi;
+ }
+
+ if (ops->msiq_alloc(pbm)) {
+ msi_table_free(pbm);
+ msi_bitmap_free(pbm);
+ goto no_msi;
+ }
+
+ if (sparc64_bringup_msi_queues(pbm, ops)) {
+ ops->msiq_free(pbm);
+ msi_table_free(pbm);
+ msi_bitmap_free(pbm);
+ goto no_msi;
+ }
+
+ printk(KERN_INFO "%s: MSI Queue first[%u] num[%u] count[%u] "
+ "devino[0x%x]\n",
+ pbm->name,
+ pbm->msiq_first, pbm->msiq_num,
+ pbm->msiq_ent_count,
+ pbm->msiq_first_devino);
+ printk(KERN_INFO "%s: MSI first[%u] num[%u] mask[0x%x] "
+ "width[%u]\n",
+ pbm->name,
+ pbm->msi_first, pbm->msi_num, pbm->msi_data_mask,
+ pbm->msix_data_width);
+ printk(KERN_INFO "%s: MSI addr32[0x%llx:0x%x] "
+ "addr64[0x%llx:0x%x]\n",
+ pbm->name,
+ pbm->msi32_start, pbm->msi32_len,
+ pbm->msi64_start, pbm->msi64_len);
+ printk(KERN_INFO "%s: MSI queues at RA [%016lx]\n",
+ pbm->name,
+ __pa(pbm->msi_queues));
+
+ pbm->msi_ops = ops;
+ pbm->setup_msi_irq = sparc64_setup_msi_irq;
+ pbm->teardown_msi_irq = sparc64_teardown_msi_irq;
+ }
+ return;
+
+no_msi:
+ pbm->msiq_num = 0;
+ printk(KERN_INFO "%s: No MSI support.\n", pbm->name);
+}
diff --git a/arch/sparc/kernel/pci_psycho.c b/arch/sparc/kernel/pci_psycho.c
new file mode 100644
index 00000000000..c647634ead2
--- /dev/null
+++ b/arch/sparc/kernel/pci_psycho.c
@@ -0,0 +1,618 @@
+/* pci_psycho.c: PSYCHO/U2P specific PCI controller support.
+ *
+ * Copyright (C) 1997, 1998, 1999, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1998, 1999 Eddie C. Dost (ecd@skynet.be)
+ * Copyright (C) 1999 Jakub Jelinek (jakub@redhat.com)
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/init.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/of_device.h>
+
+#include <asm/iommu.h>
+#include <asm/irq.h>
+#include <asm/starfire.h>
+#include <asm/prom.h>
+#include <asm/upa.h>
+
+#include "pci_impl.h"
+#include "iommu_common.h"
+#include "psycho_common.h"
+
+#define DRIVER_NAME "psycho"
+#define PFX DRIVER_NAME ": "
+
+/* Misc. PSYCHO PCI controller register offsets and definitions. */
+#define PSYCHO_CONTROL 0x0010UL
+#define PSYCHO_CONTROL_IMPL 0xf000000000000000UL /* Implementation of this PSYCHO*/
+#define PSYCHO_CONTROL_VER 0x0f00000000000000UL /* Version of this PSYCHO */
+#define PSYCHO_CONTROL_MID 0x00f8000000000000UL /* UPA Module ID of PSYCHO */
+#define PSYCHO_CONTROL_IGN 0x0007c00000000000UL /* Interrupt Group Number */
+#define PSYCHO_CONTROL_RESV 0x00003ffffffffff0UL /* Reserved */
+#define PSYCHO_CONTROL_APCKEN 0x0000000000000008UL /* Address Parity Check Enable */
+#define PSYCHO_CONTROL_APERR 0x0000000000000004UL /* Incoming System Addr Parerr */
+#define PSYCHO_CONTROL_IAP 0x0000000000000002UL /* Invert UPA Parity */
+#define PSYCHO_CONTROL_MODE 0x0000000000000001UL /* PSYCHO clock mode */
+#define PSYCHO_PCIA_CTRL 0x2000UL
+#define PSYCHO_PCIB_CTRL 0x4000UL
+#define PSYCHO_PCICTRL_RESV1 0xfffffff000000000UL /* Reserved */
+#define PSYCHO_PCICTRL_SBH_ERR 0x0000000800000000UL /* Streaming byte hole error */
+#define PSYCHO_PCICTRL_SERR 0x0000000400000000UL /* SERR signal asserted */
+#define PSYCHO_PCICTRL_SPEED 0x0000000200000000UL /* PCI speed (1 is U2P clock) */
+#define PSYCHO_PCICTRL_RESV2 0x00000001ffc00000UL /* Reserved */
+#define PSYCHO_PCICTRL_ARB_PARK 0x0000000000200000UL /* PCI arbitration parking */
+#define PSYCHO_PCICTRL_RESV3 0x00000000001ff800UL /* Reserved */
+#define PSYCHO_PCICTRL_SBH_INT 0x0000000000000400UL /* Streaming byte hole int enab */
+#define PSYCHO_PCICTRL_WEN 0x0000000000000200UL /* Power Mgmt Wake Enable */
+#define PSYCHO_PCICTRL_EEN 0x0000000000000100UL /* PCI Error Interrupt Enable */
+#define PSYCHO_PCICTRL_RESV4 0x00000000000000c0UL /* Reserved */
+#define PSYCHO_PCICTRL_AEN 0x000000000000003fUL /* PCI DVMA Arbitration Enable */
+
+/* PSYCHO error handling support. */
+
+/* Helper function of IOMMU error checking, which checks out
+ * the state of the streaming buffers. The IOMMU lock is
+ * held when this is called.
+ *
+ * For the PCI error case we know which PBM (and thus which
+ * streaming buffer) caused the error, but for the uncorrectable
+ * error case we do not. So we always check both streaming caches.
+ */
+#define PSYCHO_STRBUF_CONTROL_A 0x2800UL
+#define PSYCHO_STRBUF_CONTROL_B 0x4800UL
+#define PSYCHO_STRBUF_CTRL_LPTR 0x00000000000000f0UL /* LRU Lock Pointer */
+#define PSYCHO_STRBUF_CTRL_LENAB 0x0000000000000008UL /* LRU Lock Enable */
+#define PSYCHO_STRBUF_CTRL_RRDIS 0x0000000000000004UL /* Rerun Disable */
+#define PSYCHO_STRBUF_CTRL_DENAB 0x0000000000000002UL /* Diagnostic Mode Enable */
+#define PSYCHO_STRBUF_CTRL_ENAB 0x0000000000000001UL /* Streaming Buffer Enable */
+#define PSYCHO_STRBUF_FLUSH_A 0x2808UL
+#define PSYCHO_STRBUF_FLUSH_B 0x4808UL
+#define PSYCHO_STRBUF_FSYNC_A 0x2810UL
+#define PSYCHO_STRBUF_FSYNC_B 0x4810UL
+#define PSYCHO_STC_DATA_A 0xb000UL
+#define PSYCHO_STC_DATA_B 0xc000UL
+#define PSYCHO_STC_ERR_A 0xb400UL
+#define PSYCHO_STC_ERR_B 0xc400UL
+#define PSYCHO_STC_TAG_A 0xb800UL
+#define PSYCHO_STC_TAG_B 0xc800UL
+#define PSYCHO_STC_LINE_A 0xb900UL
+#define PSYCHO_STC_LINE_B 0xc900UL
+
+/* When an Uncorrectable Error or a PCI Error happens, we
+ * interrogate the IOMMU state to see if it is the cause.
+ */
+#define PSYCHO_IOMMU_CONTROL 0x0200UL
+#define PSYCHO_IOMMU_CTRL_RESV 0xfffffffff9000000UL /* Reserved */
+#define PSYCHO_IOMMU_CTRL_XLTESTAT 0x0000000006000000UL /* Translation Error Status */
+#define PSYCHO_IOMMU_CTRL_XLTEERR 0x0000000001000000UL /* Translation Error encountered */
+#define PSYCHO_IOMMU_CTRL_LCKEN 0x0000000000800000UL /* Enable translation locking */
+#define PSYCHO_IOMMU_CTRL_LCKPTR 0x0000000000780000UL /* Translation lock pointer */
+#define PSYCHO_IOMMU_CTRL_TSBSZ 0x0000000000070000UL /* TSB Size */
+#define PSYCHO_IOMMU_TSBSZ_1K 0x0000000000000000UL /* TSB Table 1024 8-byte entries */
+#define PSYCHO_IOMMU_TSBSZ_2K 0x0000000000010000UL /* TSB Table 2048 8-byte entries */
+#define PSYCHO_IOMMU_TSBSZ_4K 0x0000000000020000UL /* TSB Table 4096 8-byte entries */
+#define PSYCHO_IOMMU_TSBSZ_8K 0x0000000000030000UL /* TSB Table 8192 8-byte entries */
+#define PSYCHO_IOMMU_TSBSZ_16K 0x0000000000040000UL /* TSB Table 16k 8-byte entries */
+#define PSYCHO_IOMMU_TSBSZ_32K 0x0000000000050000UL /* TSB Table 32k 8-byte entries */
+#define PSYCHO_IOMMU_TSBSZ_64K 0x0000000000060000UL /* TSB Table 64k 8-byte entries */
+#define PSYCHO_IOMMU_TSBSZ_128K 0x0000000000070000UL /* TSB Table 128k 8-byte entries */
+#define PSYCHO_IOMMU_CTRL_RESV2 0x000000000000fff8UL /* Reserved */
+#define PSYCHO_IOMMU_CTRL_TBWSZ 0x0000000000000004UL /* Assumed page size, 0=8k 1=64k */
+#define PSYCHO_IOMMU_CTRL_DENAB 0x0000000000000002UL /* Diagnostic mode enable */
+#define PSYCHO_IOMMU_CTRL_ENAB 0x0000000000000001UL /* IOMMU Enable */
+#define PSYCHO_IOMMU_TSBBASE 0x0208UL
+#define PSYCHO_IOMMU_FLUSH 0x0210UL
+#define PSYCHO_IOMMU_TAG 0xa580UL
+#define PSYCHO_IOMMU_DATA 0xa600UL
+
+/* Uncorrectable Errors. Cause of the error and the address are
+ * recorded in the UE_AFSR and UE_AFAR of PSYCHO. They are errors
+ * relating to UPA interface transactions.
+ */
+#define PSYCHO_UE_AFSR 0x0030UL
+#define PSYCHO_UEAFSR_PPIO 0x8000000000000000UL /* Primary PIO is cause */
+#define PSYCHO_UEAFSR_PDRD 0x4000000000000000UL /* Primary DVMA read is cause */
+#define PSYCHO_UEAFSR_PDWR 0x2000000000000000UL /* Primary DVMA write is cause */
+#define PSYCHO_UEAFSR_SPIO 0x1000000000000000UL /* Secondary PIO is cause */
+#define PSYCHO_UEAFSR_SDRD 0x0800000000000000UL /* Secondary DVMA read is cause */
+#define PSYCHO_UEAFSR_SDWR 0x0400000000000000UL /* Secondary DVMA write is cause*/
+#define PSYCHO_UEAFSR_RESV1 0x03ff000000000000UL /* Reserved */
+#define PSYCHO_UEAFSR_BMSK 0x0000ffff00000000UL /* Bytemask of failed transfer */
+#define PSYCHO_UEAFSR_DOFF 0x00000000e0000000UL /* Doubleword Offset */
+#define PSYCHO_UEAFSR_MID 0x000000001f000000UL /* UPA MID causing the fault */
+#define PSYCHO_UEAFSR_BLK 0x0000000000800000UL /* Trans was block operation */
+#define PSYCHO_UEAFSR_RESV2 0x00000000007fffffUL /* Reserved */
+#define PSYCHO_UE_AFAR 0x0038UL
+
+static irqreturn_t psycho_ue_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ unsigned long afsr_reg = pbm->controller_regs + PSYCHO_UE_AFSR;
+ unsigned long afar_reg = pbm->controller_regs + PSYCHO_UE_AFAR;
+ unsigned long afsr, afar, error_bits;
+ int reported;
+
+ /* Latch uncorrectable error status. */
+ afar = upa_readq(afar_reg);
+ afsr = upa_readq(afsr_reg);
+
+ /* Clear the primary/secondary error status bits. */
+ error_bits = afsr &
+ (PSYCHO_UEAFSR_PPIO | PSYCHO_UEAFSR_PDRD | PSYCHO_UEAFSR_PDWR |
+ PSYCHO_UEAFSR_SPIO | PSYCHO_UEAFSR_SDRD | PSYCHO_UEAFSR_SDWR);
+ if (!error_bits)
+ return IRQ_NONE;
+ upa_writeq(error_bits, afsr_reg);
+
+ /* Log the error. */
+ printk("%s: Uncorrectable Error, primary error type[%s]\n",
+ pbm->name,
+ (((error_bits & PSYCHO_UEAFSR_PPIO) ?
+ "PIO" :
+ ((error_bits & PSYCHO_UEAFSR_PDRD) ?
+ "DMA Read" :
+ ((error_bits & PSYCHO_UEAFSR_PDWR) ?
+ "DMA Write" : "???")))));
+ printk("%s: bytemask[%04lx] dword_offset[%lx] UPA_MID[%02lx] was_block(%d)\n",
+ pbm->name,
+ (afsr & PSYCHO_UEAFSR_BMSK) >> 32UL,
+ (afsr & PSYCHO_UEAFSR_DOFF) >> 29UL,
+ (afsr & PSYCHO_UEAFSR_MID) >> 24UL,
+ ((afsr & PSYCHO_UEAFSR_BLK) ? 1 : 0));
+ printk("%s: UE AFAR [%016lx]\n", pbm->name, afar);
+ printk("%s: UE Secondary errors [", pbm->name);
+ reported = 0;
+ if (afsr & PSYCHO_UEAFSR_SPIO) {
+ reported++;
+ printk("(PIO)");
+ }
+ if (afsr & PSYCHO_UEAFSR_SDRD) {
+ reported++;
+ printk("(DMA Read)");
+ }
+ if (afsr & PSYCHO_UEAFSR_SDWR) {
+ reported++;
+ printk("(DMA Write)");
+ }
+ if (!reported)
+ printk("(none)");
+ printk("]\n");
+
+ /* Interrogate both IOMMUs for error status. */
+ psycho_check_iommu_error(pbm, afsr, afar, UE_ERR);
+ if (pbm->sibling)
+ psycho_check_iommu_error(pbm->sibling, afsr, afar, UE_ERR);
+
+ return IRQ_HANDLED;
+}
+
+/* Correctable Errors. */
+#define PSYCHO_CE_AFSR 0x0040UL
+#define PSYCHO_CEAFSR_PPIO 0x8000000000000000UL /* Primary PIO is cause */
+#define PSYCHO_CEAFSR_PDRD 0x4000000000000000UL /* Primary DVMA read is cause */
+#define PSYCHO_CEAFSR_PDWR 0x2000000000000000UL /* Primary DVMA write is cause */
+#define PSYCHO_CEAFSR_SPIO 0x1000000000000000UL /* Secondary PIO is cause */
+#define PSYCHO_CEAFSR_SDRD 0x0800000000000000UL /* Secondary DVMA read is cause */
+#define PSYCHO_CEAFSR_SDWR 0x0400000000000000UL /* Secondary DVMA write is cause*/
+#define PSYCHO_CEAFSR_RESV1 0x0300000000000000UL /* Reserved */
+#define PSYCHO_CEAFSR_ESYND 0x00ff000000000000UL /* Syndrome Bits */
+#define PSYCHO_CEAFSR_BMSK 0x0000ffff00000000UL /* Bytemask of failed transfer */
+#define PSYCHO_CEAFSR_DOFF 0x00000000e0000000UL /* Double Offset */
+#define PSYCHO_CEAFSR_MID 0x000000001f000000UL /* UPA MID causing the fault */
+#define PSYCHO_CEAFSR_BLK 0x0000000000800000UL /* Trans was block operation */
+#define PSYCHO_CEAFSR_RESV2 0x00000000007fffffUL /* Reserved */
+#define PSYCHO_CE_AFAR 0x0040UL
+
+static irqreturn_t psycho_ce_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ unsigned long afsr_reg = pbm->controller_regs + PSYCHO_CE_AFSR;
+ unsigned long afar_reg = pbm->controller_regs + PSYCHO_CE_AFAR;
+ unsigned long afsr, afar, error_bits;
+ int reported;
+
+ /* Latch error status. */
+ afar = upa_readq(afar_reg);
+ afsr = upa_readq(afsr_reg);
+
+ /* Clear primary/secondary error status bits. */
+ error_bits = afsr &
+ (PSYCHO_CEAFSR_PPIO | PSYCHO_CEAFSR_PDRD | PSYCHO_CEAFSR_PDWR |
+ PSYCHO_CEAFSR_SPIO | PSYCHO_CEAFSR_SDRD | PSYCHO_CEAFSR_SDWR);
+ if (!error_bits)
+ return IRQ_NONE;
+ upa_writeq(error_bits, afsr_reg);
+
+ /* Log the error. */
+ printk("%s: Correctable Error, primary error type[%s]\n",
+ pbm->name,
+ (((error_bits & PSYCHO_CEAFSR_PPIO) ?
+ "PIO" :
+ ((error_bits & PSYCHO_CEAFSR_PDRD) ?
+ "DMA Read" :
+ ((error_bits & PSYCHO_CEAFSR_PDWR) ?
+ "DMA Write" : "???")))));
+
+ /* XXX Use syndrome and afar to print out module string just like
+ * XXX UDB CE trap handler does... -DaveM
+ */
+ printk("%s: syndrome[%02lx] bytemask[%04lx] dword_offset[%lx] "
+ "UPA_MID[%02lx] was_block(%d)\n",
+ pbm->name,
+ (afsr & PSYCHO_CEAFSR_ESYND) >> 48UL,
+ (afsr & PSYCHO_CEAFSR_BMSK) >> 32UL,
+ (afsr & PSYCHO_CEAFSR_DOFF) >> 29UL,
+ (afsr & PSYCHO_CEAFSR_MID) >> 24UL,
+ ((afsr & PSYCHO_CEAFSR_BLK) ? 1 : 0));
+ printk("%s: CE AFAR [%016lx]\n", pbm->name, afar);
+ printk("%s: CE Secondary errors [", pbm->name);
+ reported = 0;
+ if (afsr & PSYCHO_CEAFSR_SPIO) {
+ reported++;
+ printk("(PIO)");
+ }
+ if (afsr & PSYCHO_CEAFSR_SDRD) {
+ reported++;
+ printk("(DMA Read)");
+ }
+ if (afsr & PSYCHO_CEAFSR_SDWR) {
+ reported++;
+ printk("(DMA Write)");
+ }
+ if (!reported)
+ printk("(none)");
+ printk("]\n");
+
+ return IRQ_HANDLED;
+}
+
+/* PCI Errors. They are signalled by the PCI bus module since they
+ * are associated with a specific bus segment.
+ */
+#define PSYCHO_PCI_AFSR_A 0x2010UL
+#define PSYCHO_PCI_AFSR_B 0x4010UL
+#define PSYCHO_PCI_AFAR_A 0x2018UL
+#define PSYCHO_PCI_AFAR_B 0x4018UL
+
+/* XXX What about PowerFail/PowerManagement??? -DaveM */
+#define PSYCHO_ECC_CTRL 0x0020
+#define PSYCHO_ECCCTRL_EE 0x8000000000000000UL /* Enable ECC Checking */
+#define PSYCHO_ECCCTRL_UE 0x4000000000000000UL /* Enable UE Interrupts */
+#define PSYCHO_ECCCTRL_CE 0x2000000000000000UL /* Enable CE INterrupts */
+static void psycho_register_error_handlers(struct pci_pbm_info *pbm)
+{
+ struct platform_device *op = of_find_device_by_node(pbm->op->dev.of_node);
+ unsigned long base = pbm->controller_regs;
+ u64 tmp;
+ int err;
+
+ if (!op)
+ return;
+
+ /* Psycho interrupt property order is:
+ * 0: PCIERR INO for this PBM
+ * 1: UE ERR
+ * 2: CE ERR
+ * 3: POWER FAIL
+ * 4: SPARE HARDWARE
+ * 5: POWER MANAGEMENT
+ */
+
+ if (op->archdata.num_irqs < 6)
+ return;
+
+ /* We really mean to ignore the return result here. Two
+ * PCI controller share the same interrupt numbers and
+ * drive the same front-end hardware.
+ */
+ err = request_irq(op->archdata.irqs[1], psycho_ue_intr, IRQF_SHARED,
+ "PSYCHO_UE", pbm);
+ err = request_irq(op->archdata.irqs[2], psycho_ce_intr, IRQF_SHARED,
+ "PSYCHO_CE", pbm);
+
+ /* This one, however, ought not to fail. We can just warn
+ * about it since the system can still operate properly even
+ * if this fails.
+ */
+ err = request_irq(op->archdata.irqs[0], psycho_pcierr_intr, IRQF_SHARED,
+ "PSYCHO_PCIERR", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Could not register PCIERR, "
+ "err=%d\n", pbm->name, err);
+
+ /* Enable UE and CE interrupts for controller. */
+ upa_writeq((PSYCHO_ECCCTRL_EE |
+ PSYCHO_ECCCTRL_UE |
+ PSYCHO_ECCCTRL_CE), base + PSYCHO_ECC_CTRL);
+
+ /* Enable PCI Error interrupts and clear error
+ * bits for each PBM.
+ */
+ tmp = upa_readq(base + PSYCHO_PCIA_CTRL);
+ tmp |= (PSYCHO_PCICTRL_SERR |
+ PSYCHO_PCICTRL_SBH_ERR |
+ PSYCHO_PCICTRL_EEN);
+ tmp &= ~(PSYCHO_PCICTRL_SBH_INT);
+ upa_writeq(tmp, base + PSYCHO_PCIA_CTRL);
+
+ tmp = upa_readq(base + PSYCHO_PCIB_CTRL);
+ tmp |= (PSYCHO_PCICTRL_SERR |
+ PSYCHO_PCICTRL_SBH_ERR |
+ PSYCHO_PCICTRL_EEN);
+ tmp &= ~(PSYCHO_PCICTRL_SBH_INT);
+ upa_writeq(tmp, base + PSYCHO_PCIB_CTRL);
+}
+
+/* PSYCHO boot time probing and initialization. */
+static void pbm_config_busmastering(struct pci_pbm_info *pbm)
+{
+ u8 *addr;
+
+ /* Set cache-line size to 64 bytes, this is actually
+ * a nop but I do it for completeness.
+ */
+ addr = psycho_pci_config_mkaddr(pbm, pbm->pci_first_busno,
+ 0, PCI_CACHE_LINE_SIZE);
+ pci_config_write8(addr, 64 / sizeof(u32));
+
+ /* Set PBM latency timer to 64 PCI clocks. */
+ addr = psycho_pci_config_mkaddr(pbm, pbm->pci_first_busno,
+ 0, PCI_LATENCY_TIMER);
+ pci_config_write8(addr, 64);
+}
+
+static void psycho_scan_bus(struct pci_pbm_info *pbm,
+ struct device *parent)
+{
+ pbm_config_busmastering(pbm);
+ pbm->is_66mhz_capable = 0;
+ pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
+
+ /* After the PCI bus scan is complete, we can register
+ * the error interrupt handlers.
+ */
+ psycho_register_error_handlers(pbm);
+}
+
+#define PSYCHO_IRQ_RETRY 0x1a00UL
+#define PSYCHO_PCIA_DIAG 0x2020UL
+#define PSYCHO_PCIB_DIAG 0x4020UL
+#define PSYCHO_PCIDIAG_RESV 0xffffffffffffff80UL /* Reserved */
+#define PSYCHO_PCIDIAG_DRETRY 0x0000000000000040UL /* Disable retry limit */
+#define PSYCHO_PCIDIAG_DISYNC 0x0000000000000020UL /* Disable DMA wr / irq sync */
+#define PSYCHO_PCIDIAG_DDWSYNC 0x0000000000000010UL /* Disable DMA wr / PIO rd sync */
+#define PSYCHO_PCIDIAG_IDDPAR 0x0000000000000008UL /* Invert DMA data parity */
+#define PSYCHO_PCIDIAG_IPDPAR 0x0000000000000004UL /* Invert PIO data parity */
+#define PSYCHO_PCIDIAG_IPAPAR 0x0000000000000002UL /* Invert PIO address parity */
+#define PSYCHO_PCIDIAG_LPBACK 0x0000000000000001UL /* Enable loopback mode */
+
+static void psycho_controller_hwinit(struct pci_pbm_info *pbm)
+{
+ u64 tmp;
+
+ upa_writeq(5, pbm->controller_regs + PSYCHO_IRQ_RETRY);
+
+ /* Enable arbiter for all PCI slots. */
+ tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIA_CTRL);
+ tmp |= PSYCHO_PCICTRL_AEN;
+ upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIA_CTRL);
+
+ tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIB_CTRL);
+ tmp |= PSYCHO_PCICTRL_AEN;
+ upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIB_CTRL);
+
+ /* Disable DMA write / PIO read synchronization on
+ * both PCI bus segments.
+ * [ U2P Erratum 1243770, STP2223BGA data sheet ]
+ */
+ tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIA_DIAG);
+ tmp |= PSYCHO_PCIDIAG_DDWSYNC;
+ upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIA_DIAG);
+
+ tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIB_DIAG);
+ tmp |= PSYCHO_PCIDIAG_DDWSYNC;
+ upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIB_DIAG);
+}
+
+static void psycho_pbm_strbuf_init(struct pci_pbm_info *pbm,
+ int is_pbm_a)
+{
+ unsigned long base = pbm->controller_regs;
+ u64 control;
+
+ if (is_pbm_a) {
+ pbm->stc.strbuf_control = base + PSYCHO_STRBUF_CONTROL_A;
+ pbm->stc.strbuf_pflush = base + PSYCHO_STRBUF_FLUSH_A;
+ pbm->stc.strbuf_fsync = base + PSYCHO_STRBUF_FSYNC_A;
+ pbm->stc.strbuf_err_stat = base + PSYCHO_STC_ERR_A;
+ pbm->stc.strbuf_tag_diag = base + PSYCHO_STC_TAG_A;
+ pbm->stc.strbuf_line_diag= base + PSYCHO_STC_LINE_A;
+ } else {
+ pbm->stc.strbuf_control = base + PSYCHO_STRBUF_CONTROL_B;
+ pbm->stc.strbuf_pflush = base + PSYCHO_STRBUF_FLUSH_B;
+ pbm->stc.strbuf_fsync = base + PSYCHO_STRBUF_FSYNC_B;
+ pbm->stc.strbuf_err_stat = base + PSYCHO_STC_ERR_B;
+ pbm->stc.strbuf_tag_diag = base + PSYCHO_STC_TAG_B;
+ pbm->stc.strbuf_line_diag= base + PSYCHO_STC_LINE_B;
+ }
+ /* PSYCHO's streaming buffer lacks ctx flushing. */
+ pbm->stc.strbuf_ctxflush = 0;
+ pbm->stc.strbuf_ctxmatch_base = 0;
+
+ pbm->stc.strbuf_flushflag = (volatile unsigned long *)
+ ((((unsigned long)&pbm->stc.__flushflag_buf[0])
+ + 63UL)
+ & ~63UL);
+ pbm->stc.strbuf_flushflag_pa = (unsigned long)
+ __pa(pbm->stc.strbuf_flushflag);
+
+ /* Enable the streaming buffer. We have to be careful
+ * just in case OBP left it with LRU locking enabled.
+ *
+ * It is possible to control if PBM will be rerun on
+ * line misses. Currently I just retain whatever setting
+ * OBP left us with. All checks so far show it having
+ * a value of zero.
+ */
+#undef PSYCHO_STRBUF_RERUN_ENABLE
+#undef PSYCHO_STRBUF_RERUN_DISABLE
+ control = upa_readq(pbm->stc.strbuf_control);
+ control |= PSYCHO_STRBUF_CTRL_ENAB;
+ control &= ~(PSYCHO_STRBUF_CTRL_LENAB | PSYCHO_STRBUF_CTRL_LPTR);
+#ifdef PSYCHO_STRBUF_RERUN_ENABLE
+ control &= ~(PSYCHO_STRBUF_CTRL_RRDIS);
+#else
+#ifdef PSYCHO_STRBUF_RERUN_DISABLE
+ control |= PSYCHO_STRBUF_CTRL_RRDIS;
+#endif
+#endif
+ upa_writeq(control, pbm->stc.strbuf_control);
+
+ pbm->stc.strbuf_enabled = 1;
+}
+
+#define PSYCHO_IOSPACE_A 0x002000000UL
+#define PSYCHO_IOSPACE_B 0x002010000UL
+#define PSYCHO_IOSPACE_SIZE 0x00000ffffUL
+#define PSYCHO_MEMSPACE_A 0x100000000UL
+#define PSYCHO_MEMSPACE_B 0x180000000UL
+#define PSYCHO_MEMSPACE_SIZE 0x07fffffffUL
+
+static void psycho_pbm_init(struct pci_pbm_info *pbm,
+ struct platform_device *op, int is_pbm_a)
+{
+ psycho_pbm_init_common(pbm, op, "PSYCHO", PBM_CHIP_TYPE_PSYCHO);
+ psycho_pbm_strbuf_init(pbm, is_pbm_a);
+ psycho_scan_bus(pbm, &op->dev);
+}
+
+static struct pci_pbm_info *psycho_find_sibling(u32 upa_portid)
+{
+ struct pci_pbm_info *pbm;
+
+ for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
+ if (pbm->portid == upa_portid)
+ return pbm;
+ }
+ return NULL;
+}
+
+#define PSYCHO_CONFIGSPACE 0x001000000UL
+
+static int psycho_probe(struct platform_device *op)
+{
+ const struct linux_prom64_registers *pr_regs;
+ struct device_node *dp = op->dev.of_node;
+ struct pci_pbm_info *pbm;
+ struct iommu *iommu;
+ int is_pbm_a, err;
+ u32 upa_portid;
+
+ upa_portid = of_getintprop_default(dp, "upa-portid", 0xff);
+
+ err = -ENOMEM;
+ pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+ if (!pbm) {
+ printk(KERN_ERR PFX "Cannot allocate pci_pbm_info.\n");
+ goto out_err;
+ }
+
+ pbm->sibling = psycho_find_sibling(upa_portid);
+ if (pbm->sibling) {
+ iommu = pbm->sibling->iommu;
+ } else {
+ iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+ if (!iommu) {
+ printk(KERN_ERR PFX "Cannot allocate PBM iommu.\n");
+ goto out_free_controller;
+ }
+ }
+
+ pbm->iommu = iommu;
+ pbm->portid = upa_portid;
+
+ pr_regs = of_get_property(dp, "reg", NULL);
+ err = -ENODEV;
+ if (!pr_regs) {
+ printk(KERN_ERR PFX "No reg property.\n");
+ goto out_free_iommu;
+ }
+
+ is_pbm_a = ((pr_regs[0].phys_addr & 0x6000) == 0x2000);
+
+ pbm->controller_regs = pr_regs[2].phys_addr;
+ pbm->config_space = (pr_regs[2].phys_addr + PSYCHO_CONFIGSPACE);
+
+ if (is_pbm_a) {
+ pbm->pci_afsr = pbm->controller_regs + PSYCHO_PCI_AFSR_A;
+ pbm->pci_afar = pbm->controller_regs + PSYCHO_PCI_AFAR_A;
+ pbm->pci_csr = pbm->controller_regs + PSYCHO_PCIA_CTRL;
+ } else {
+ pbm->pci_afsr = pbm->controller_regs + PSYCHO_PCI_AFSR_B;
+ pbm->pci_afar = pbm->controller_regs + PSYCHO_PCI_AFAR_B;
+ pbm->pci_csr = pbm->controller_regs + PSYCHO_PCIB_CTRL;
+ }
+
+ psycho_controller_hwinit(pbm);
+ if (!pbm->sibling) {
+ err = psycho_iommu_init(pbm, 128, 0xc0000000,
+ 0xffffffff, PSYCHO_CONTROL);
+ if (err)
+ goto out_free_iommu;
+
+ /* If necessary, hook us up for starfire IRQ translations. */
+ if (this_is_starfire)
+ starfire_hookup(pbm->portid);
+ }
+
+ psycho_pbm_init(pbm, op, is_pbm_a);
+
+ pbm->next = pci_pbm_root;
+ pci_pbm_root = pbm;
+
+ if (pbm->sibling)
+ pbm->sibling->sibling = pbm;
+
+ dev_set_drvdata(&op->dev, pbm);
+
+ return 0;
+
+out_free_iommu:
+ if (!pbm->sibling)
+ kfree(pbm->iommu);
+
+out_free_controller:
+ kfree(pbm);
+
+out_err:
+ return err;
+}
+
+static const struct of_device_id psycho_match[] = {
+ {
+ .name = "pci",
+ .compatible = "pci108e,8000",
+ },
+ {},
+};
+
+static struct platform_driver psycho_driver = {
+ .driver = {
+ .name = DRIVER_NAME,
+ .owner = THIS_MODULE,
+ .of_match_table = psycho_match,
+ },
+ .probe = psycho_probe,
+};
+
+static int __init psycho_init(void)
+{
+ return platform_driver_register(&psycho_driver);
+}
+
+subsys_initcall(psycho_init);
diff --git a/arch/sparc/kernel/pci_sabre.c b/arch/sparc/kernel/pci_sabre.c
new file mode 100644
index 00000000000..6f00d27e8da
--- /dev/null
+++ b/arch/sparc/kernel/pci_sabre.c
@@ -0,0 +1,614 @@
+/* pci_sabre.c: Sabre specific PCI controller support.
+ *
+ * Copyright (C) 1997, 1998, 1999, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1998, 1999 Eddie C. Dost (ecd@skynet.be)
+ * Copyright (C) 1999 Jakub Jelinek (jakub@redhat.com)
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/init.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/of_device.h>
+
+#include <asm/apb.h>
+#include <asm/iommu.h>
+#include <asm/irq.h>
+#include <asm/prom.h>
+#include <asm/upa.h>
+
+#include "pci_impl.h"
+#include "iommu_common.h"
+#include "psycho_common.h"
+
+#define DRIVER_NAME "sabre"
+#define PFX DRIVER_NAME ": "
+
+/* SABRE PCI controller register offsets and definitions. */
+#define SABRE_UE_AFSR 0x0030UL
+#define SABRE_UEAFSR_PDRD 0x4000000000000000UL /* Primary PCI DMA Read */
+#define SABRE_UEAFSR_PDWR 0x2000000000000000UL /* Primary PCI DMA Write */
+#define SABRE_UEAFSR_SDRD 0x0800000000000000UL /* Secondary PCI DMA Read */
+#define SABRE_UEAFSR_SDWR 0x0400000000000000UL /* Secondary PCI DMA Write */
+#define SABRE_UEAFSR_SDTE 0x0200000000000000UL /* Secondary DMA Translation Error */
+#define SABRE_UEAFSR_PDTE 0x0100000000000000UL /* Primary DMA Translation Error */
+#define SABRE_UEAFSR_BMSK 0x0000ffff00000000UL /* Bytemask */
+#define SABRE_UEAFSR_OFF 0x00000000e0000000UL /* Offset (AFAR bits [5:3] */
+#define SABRE_UEAFSR_BLK 0x0000000000800000UL /* Was block operation */
+#define SABRE_UECE_AFAR 0x0038UL
+#define SABRE_CE_AFSR 0x0040UL
+#define SABRE_CEAFSR_PDRD 0x4000000000000000UL /* Primary PCI DMA Read */
+#define SABRE_CEAFSR_PDWR 0x2000000000000000UL /* Primary PCI DMA Write */
+#define SABRE_CEAFSR_SDRD 0x0800000000000000UL /* Secondary PCI DMA Read */
+#define SABRE_CEAFSR_SDWR 0x0400000000000000UL /* Secondary PCI DMA Write */
+#define SABRE_CEAFSR_ESYND 0x00ff000000000000UL /* ECC Syndrome */
+#define SABRE_CEAFSR_BMSK 0x0000ffff00000000UL /* Bytemask */
+#define SABRE_CEAFSR_OFF 0x00000000e0000000UL /* Offset */
+#define SABRE_CEAFSR_BLK 0x0000000000800000UL /* Was block operation */
+#define SABRE_UECE_AFAR_ALIAS 0x0048UL /* Aliases to 0x0038 */
+#define SABRE_IOMMU_CONTROL 0x0200UL
+#define SABRE_IOMMUCTRL_ERRSTS 0x0000000006000000UL /* Error status bits */
+#define SABRE_IOMMUCTRL_ERR 0x0000000001000000UL /* Error present in IOTLB */
+#define SABRE_IOMMUCTRL_LCKEN 0x0000000000800000UL /* IOTLB lock enable */
+#define SABRE_IOMMUCTRL_LCKPTR 0x0000000000780000UL /* IOTLB lock pointer */
+#define SABRE_IOMMUCTRL_TSBSZ 0x0000000000070000UL /* TSB Size */
+#define SABRE_IOMMU_TSBSZ_1K 0x0000000000000000
+#define SABRE_IOMMU_TSBSZ_2K 0x0000000000010000
+#define SABRE_IOMMU_TSBSZ_4K 0x0000000000020000
+#define SABRE_IOMMU_TSBSZ_8K 0x0000000000030000
+#define SABRE_IOMMU_TSBSZ_16K 0x0000000000040000
+#define SABRE_IOMMU_TSBSZ_32K 0x0000000000050000
+#define SABRE_IOMMU_TSBSZ_64K 0x0000000000060000
+#define SABRE_IOMMU_TSBSZ_128K 0x0000000000070000
+#define SABRE_IOMMUCTRL_TBWSZ 0x0000000000000004UL /* TSB assumed page size */
+#define SABRE_IOMMUCTRL_DENAB 0x0000000000000002UL /* Diagnostic Mode Enable */
+#define SABRE_IOMMUCTRL_ENAB 0x0000000000000001UL /* IOMMU Enable */
+#define SABRE_IOMMU_TSBBASE 0x0208UL
+#define SABRE_IOMMU_FLUSH 0x0210UL
+#define SABRE_IMAP_A_SLOT0 0x0c00UL
+#define SABRE_IMAP_B_SLOT0 0x0c20UL
+#define SABRE_IMAP_SCSI 0x1000UL
+#define SABRE_IMAP_ETH 0x1008UL
+#define SABRE_IMAP_BPP 0x1010UL
+#define SABRE_IMAP_AU_REC 0x1018UL
+#define SABRE_IMAP_AU_PLAY 0x1020UL
+#define SABRE_IMAP_PFAIL 0x1028UL
+#define SABRE_IMAP_KMS 0x1030UL
+#define SABRE_IMAP_FLPY 0x1038UL
+#define SABRE_IMAP_SHW 0x1040UL
+#define SABRE_IMAP_KBD 0x1048UL
+#define SABRE_IMAP_MS 0x1050UL
+#define SABRE_IMAP_SER 0x1058UL
+#define SABRE_IMAP_UE 0x1070UL
+#define SABRE_IMAP_CE 0x1078UL
+#define SABRE_IMAP_PCIERR 0x1080UL
+#define SABRE_IMAP_GFX 0x1098UL
+#define SABRE_IMAP_EUPA 0x10a0UL
+#define SABRE_ICLR_A_SLOT0 0x1400UL
+#define SABRE_ICLR_B_SLOT0 0x1480UL
+#define SABRE_ICLR_SCSI 0x1800UL
+#define SABRE_ICLR_ETH 0x1808UL
+#define SABRE_ICLR_BPP 0x1810UL
+#define SABRE_ICLR_AU_REC 0x1818UL
+#define SABRE_ICLR_AU_PLAY 0x1820UL
+#define SABRE_ICLR_PFAIL 0x1828UL
+#define SABRE_ICLR_KMS 0x1830UL
+#define SABRE_ICLR_FLPY 0x1838UL
+#define SABRE_ICLR_SHW 0x1840UL
+#define SABRE_ICLR_KBD 0x1848UL
+#define SABRE_ICLR_MS 0x1850UL
+#define SABRE_ICLR_SER 0x1858UL
+#define SABRE_ICLR_UE 0x1870UL
+#define SABRE_ICLR_CE 0x1878UL
+#define SABRE_ICLR_PCIERR 0x1880UL
+#define SABRE_WRSYNC 0x1c20UL
+#define SABRE_PCICTRL 0x2000UL
+#define SABRE_PCICTRL_MRLEN 0x0000001000000000UL /* Use MemoryReadLine for block loads/stores */
+#define SABRE_PCICTRL_SERR 0x0000000400000000UL /* Set when SERR asserted on PCI bus */
+#define SABRE_PCICTRL_ARBPARK 0x0000000000200000UL /* Bus Parking 0=Ultra-IIi 1=prev-bus-owner */
+#define SABRE_PCICTRL_CPUPRIO 0x0000000000100000UL /* Ultra-IIi granted every other bus cycle */
+#define SABRE_PCICTRL_ARBPRIO 0x00000000000f0000UL /* Slot which is granted every other bus cycle */
+#define SABRE_PCICTRL_ERREN 0x0000000000000100UL /* PCI Error Interrupt Enable */
+#define SABRE_PCICTRL_RTRYWE 0x0000000000000080UL /* DMA Flow Control 0=wait-if-possible 1=retry */
+#define SABRE_PCICTRL_AEN 0x000000000000000fUL /* Slot PCI arbitration enables */
+#define SABRE_PIOAFSR 0x2010UL
+#define SABRE_PIOAFSR_PMA 0x8000000000000000UL /* Primary Master Abort */
+#define SABRE_PIOAFSR_PTA 0x4000000000000000UL /* Primary Target Abort */
+#define SABRE_PIOAFSR_PRTRY 0x2000000000000000UL /* Primary Excessive Retries */
+#define SABRE_PIOAFSR_PPERR 0x1000000000000000UL /* Primary Parity Error */
+#define SABRE_PIOAFSR_SMA 0x0800000000000000UL /* Secondary Master Abort */
+#define SABRE_PIOAFSR_STA 0x0400000000000000UL /* Secondary Target Abort */
+#define SABRE_PIOAFSR_SRTRY 0x0200000000000000UL /* Secondary Excessive Retries */
+#define SABRE_PIOAFSR_SPERR 0x0100000000000000UL /* Secondary Parity Error */
+#define SABRE_PIOAFSR_BMSK 0x0000ffff00000000UL /* Byte Mask */
+#define SABRE_PIOAFSR_BLK 0x0000000080000000UL /* Was Block Operation */
+#define SABRE_PIOAFAR 0x2018UL
+#define SABRE_PCIDIAG 0x2020UL
+#define SABRE_PCIDIAG_DRTRY 0x0000000000000040UL /* Disable PIO Retry Limit */
+#define SABRE_PCIDIAG_IPAPAR 0x0000000000000008UL /* Invert PIO Address Parity */
+#define SABRE_PCIDIAG_IPDPAR 0x0000000000000004UL /* Invert PIO Data Parity */
+#define SABRE_PCIDIAG_IDDPAR 0x0000000000000002UL /* Invert DMA Data Parity */
+#define SABRE_PCIDIAG_ELPBK 0x0000000000000001UL /* Loopback Enable - not supported */
+#define SABRE_PCITASR 0x2028UL
+#define SABRE_PCITASR_EF 0x0000000000000080UL /* Respond to 0xe0000000-0xffffffff */
+#define SABRE_PCITASR_CD 0x0000000000000040UL /* Respond to 0xc0000000-0xdfffffff */
+#define SABRE_PCITASR_AB 0x0000000000000020UL /* Respond to 0xa0000000-0xbfffffff */
+#define SABRE_PCITASR_89 0x0000000000000010UL /* Respond to 0x80000000-0x9fffffff */
+#define SABRE_PCITASR_67 0x0000000000000008UL /* Respond to 0x60000000-0x7fffffff */
+#define SABRE_PCITASR_45 0x0000000000000004UL /* Respond to 0x40000000-0x5fffffff */
+#define SABRE_PCITASR_23 0x0000000000000002UL /* Respond to 0x20000000-0x3fffffff */
+#define SABRE_PCITASR_01 0x0000000000000001UL /* Respond to 0x00000000-0x1fffffff */
+#define SABRE_PIOBUF_DIAG 0x5000UL
+#define SABRE_DMABUF_DIAGLO 0x5100UL
+#define SABRE_DMABUF_DIAGHI 0x51c0UL
+#define SABRE_IMAP_GFX_ALIAS 0x6000UL /* Aliases to 0x1098 */
+#define SABRE_IMAP_EUPA_ALIAS 0x8000UL /* Aliases to 0x10a0 */
+#define SABRE_IOMMU_VADIAG 0xa400UL
+#define SABRE_IOMMU_TCDIAG 0xa408UL
+#define SABRE_IOMMU_TAG 0xa580UL
+#define SABRE_IOMMUTAG_ERRSTS 0x0000000001800000UL /* Error status bits */
+#define SABRE_IOMMUTAG_ERR 0x0000000000400000UL /* Error present */
+#define SABRE_IOMMUTAG_WRITE 0x0000000000200000UL /* Page is writable */
+#define SABRE_IOMMUTAG_STREAM 0x0000000000100000UL /* Streamable bit - unused */
+#define SABRE_IOMMUTAG_SIZE 0x0000000000080000UL /* 0=8k 1=16k */
+#define SABRE_IOMMUTAG_VPN 0x000000000007ffffUL /* Virtual Page Number [31:13] */
+#define SABRE_IOMMU_DATA 0xa600UL
+#define SABRE_IOMMUDATA_VALID 0x0000000040000000UL /* Valid */
+#define SABRE_IOMMUDATA_USED 0x0000000020000000UL /* Used (for LRU algorithm) */
+#define SABRE_IOMMUDATA_CACHE 0x0000000010000000UL /* Cacheable */
+#define SABRE_IOMMUDATA_PPN 0x00000000001fffffUL /* Physical Page Number [33:13] */
+#define SABRE_PCI_IRQSTATE 0xa800UL
+#define SABRE_OBIO_IRQSTATE 0xa808UL
+#define SABRE_FFBCFG 0xf000UL
+#define SABRE_FFBCFG_SPRQS 0x000000000f000000 /* Slave P_RQST queue size */
+#define SABRE_FFBCFG_ONEREAD 0x0000000000004000 /* Slave supports one outstanding read */
+#define SABRE_MCCTRL0 0xf010UL
+#define SABRE_MCCTRL0_RENAB 0x0000000080000000 /* Refresh Enable */
+#define SABRE_MCCTRL0_EENAB 0x0000000010000000 /* Enable all ECC functions */
+#define SABRE_MCCTRL0_11BIT 0x0000000000001000 /* Enable 11-bit column addressing */
+#define SABRE_MCCTRL0_DPP 0x0000000000000f00 /* DIMM Pair Present Bits */
+#define SABRE_MCCTRL0_RINTVL 0x00000000000000ff /* Refresh Interval */
+#define SABRE_MCCTRL1 0xf018UL
+#define SABRE_MCCTRL1_AMDC 0x0000000038000000 /* Advance Memdata Clock */
+#define SABRE_MCCTRL1_ARDC 0x0000000007000000 /* Advance DRAM Read Data Clock */
+#define SABRE_MCCTRL1_CSR 0x0000000000e00000 /* CAS to RAS delay for CBR refresh */
+#define SABRE_MCCTRL1_CASRW 0x00000000001c0000 /* CAS length for read/write */
+#define SABRE_MCCTRL1_RCD 0x0000000000038000 /* RAS to CAS delay */
+#define SABRE_MCCTRL1_CP 0x0000000000007000 /* CAS Precharge */
+#define SABRE_MCCTRL1_RP 0x0000000000000e00 /* RAS Precharge */
+#define SABRE_MCCTRL1_RAS 0x00000000000001c0 /* Length of RAS for refresh */
+#define SABRE_MCCTRL1_CASRW2 0x0000000000000038 /* Must be same as CASRW */
+#define SABRE_MCCTRL1_RSC 0x0000000000000007 /* RAS after CAS hold time */
+#define SABRE_RESETCTRL 0xf020UL
+
+#define SABRE_CONFIGSPACE 0x001000000UL
+#define SABRE_IOSPACE 0x002000000UL
+#define SABRE_IOSPACE_SIZE 0x000ffffffUL
+#define SABRE_MEMSPACE 0x100000000UL
+#define SABRE_MEMSPACE_SIZE 0x07fffffffUL
+
+static int hummingbird_p;
+static struct pci_bus *sabre_root_bus;
+
+static irqreturn_t sabre_ue_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ unsigned long afsr_reg = pbm->controller_regs + SABRE_UE_AFSR;
+ unsigned long afar_reg = pbm->controller_regs + SABRE_UECE_AFAR;
+ unsigned long afsr, afar, error_bits;
+ int reported;
+
+ /* Latch uncorrectable error status. */
+ afar = upa_readq(afar_reg);
+ afsr = upa_readq(afsr_reg);
+
+ /* Clear the primary/secondary error status bits. */
+ error_bits = afsr &
+ (SABRE_UEAFSR_PDRD | SABRE_UEAFSR_PDWR |
+ SABRE_UEAFSR_SDRD | SABRE_UEAFSR_SDWR |
+ SABRE_UEAFSR_SDTE | SABRE_UEAFSR_PDTE);
+ if (!error_bits)
+ return IRQ_NONE;
+ upa_writeq(error_bits, afsr_reg);
+
+ /* Log the error. */
+ printk("%s: Uncorrectable Error, primary error type[%s%s]\n",
+ pbm->name,
+ ((error_bits & SABRE_UEAFSR_PDRD) ?
+ "DMA Read" :
+ ((error_bits & SABRE_UEAFSR_PDWR) ?
+ "DMA Write" : "???")),
+ ((error_bits & SABRE_UEAFSR_PDTE) ?
+ ":Translation Error" : ""));
+ printk("%s: bytemask[%04lx] dword_offset[%lx] was_block(%d)\n",
+ pbm->name,
+ (afsr & SABRE_UEAFSR_BMSK) >> 32UL,
+ (afsr & SABRE_UEAFSR_OFF) >> 29UL,
+ ((afsr & SABRE_UEAFSR_BLK) ? 1 : 0));
+ printk("%s: UE AFAR [%016lx]\n", pbm->name, afar);
+ printk("%s: UE Secondary errors [", pbm->name);
+ reported = 0;
+ if (afsr & SABRE_UEAFSR_SDRD) {
+ reported++;
+ printk("(DMA Read)");
+ }
+ if (afsr & SABRE_UEAFSR_SDWR) {
+ reported++;
+ printk("(DMA Write)");
+ }
+ if (afsr & SABRE_UEAFSR_SDTE) {
+ reported++;
+ printk("(Translation Error)");
+ }
+ if (!reported)
+ printk("(none)");
+ printk("]\n");
+
+ /* Interrogate IOMMU for error status. */
+ psycho_check_iommu_error(pbm, afsr, afar, UE_ERR);
+
+ return IRQ_HANDLED;
+}
+
+static irqreturn_t sabre_ce_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ unsigned long afsr_reg = pbm->controller_regs + SABRE_CE_AFSR;
+ unsigned long afar_reg = pbm->controller_regs + SABRE_UECE_AFAR;
+ unsigned long afsr, afar, error_bits;
+ int reported;
+
+ /* Latch error status. */
+ afar = upa_readq(afar_reg);
+ afsr = upa_readq(afsr_reg);
+
+ /* Clear primary/secondary error status bits. */
+ error_bits = afsr &
+ (SABRE_CEAFSR_PDRD | SABRE_CEAFSR_PDWR |
+ SABRE_CEAFSR_SDRD | SABRE_CEAFSR_SDWR);
+ if (!error_bits)
+ return IRQ_NONE;
+ upa_writeq(error_bits, afsr_reg);
+
+ /* Log the error. */
+ printk("%s: Correctable Error, primary error type[%s]\n",
+ pbm->name,
+ ((error_bits & SABRE_CEAFSR_PDRD) ?
+ "DMA Read" :
+ ((error_bits & SABRE_CEAFSR_PDWR) ?
+ "DMA Write" : "???")));
+
+ /* XXX Use syndrome and afar to print out module string just like
+ * XXX UDB CE trap handler does... -DaveM
+ */
+ printk("%s: syndrome[%02lx] bytemask[%04lx] dword_offset[%lx] "
+ "was_block(%d)\n",
+ pbm->name,
+ (afsr & SABRE_CEAFSR_ESYND) >> 48UL,
+ (afsr & SABRE_CEAFSR_BMSK) >> 32UL,
+ (afsr & SABRE_CEAFSR_OFF) >> 29UL,
+ ((afsr & SABRE_CEAFSR_BLK) ? 1 : 0));
+ printk("%s: CE AFAR [%016lx]\n", pbm->name, afar);
+ printk("%s: CE Secondary errors [", pbm->name);
+ reported = 0;
+ if (afsr & SABRE_CEAFSR_SDRD) {
+ reported++;
+ printk("(DMA Read)");
+ }
+ if (afsr & SABRE_CEAFSR_SDWR) {
+ reported++;
+ printk("(DMA Write)");
+ }
+ if (!reported)
+ printk("(none)");
+ printk("]\n");
+
+ return IRQ_HANDLED;
+}
+
+static void sabre_register_error_handlers(struct pci_pbm_info *pbm)
+{
+ struct device_node *dp = pbm->op->dev.of_node;
+ struct platform_device *op;
+ unsigned long base = pbm->controller_regs;
+ u64 tmp;
+ int err;
+
+ if (pbm->chip_type == PBM_CHIP_TYPE_SABRE)
+ dp = dp->parent;
+
+ op = of_find_device_by_node(dp);
+ if (!op)
+ return;
+
+ /* Sabre/Hummingbird IRQ property layout is:
+ * 0: PCI ERR
+ * 1: UE ERR
+ * 2: CE ERR
+ * 3: POWER FAIL
+ */
+ if (op->archdata.num_irqs < 4)
+ return;
+
+ /* We clear the error bits in the appropriate AFSR before
+ * registering the handler so that we don't get spurious
+ * interrupts.
+ */
+ upa_writeq((SABRE_UEAFSR_PDRD | SABRE_UEAFSR_PDWR |
+ SABRE_UEAFSR_SDRD | SABRE_UEAFSR_SDWR |
+ SABRE_UEAFSR_SDTE | SABRE_UEAFSR_PDTE),
+ base + SABRE_UE_AFSR);
+
+ err = request_irq(op->archdata.irqs[1], sabre_ue_intr, 0, "SABRE_UE", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Couldn't register UE, err=%d.\n",
+ pbm->name, err);
+
+ upa_writeq((SABRE_CEAFSR_PDRD | SABRE_CEAFSR_PDWR |
+ SABRE_CEAFSR_SDRD | SABRE_CEAFSR_SDWR),
+ base + SABRE_CE_AFSR);
+
+
+ err = request_irq(op->archdata.irqs[2], sabre_ce_intr, 0, "SABRE_CE", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Couldn't register CE, err=%d.\n",
+ pbm->name, err);
+ err = request_irq(op->archdata.irqs[0], psycho_pcierr_intr, 0,
+ "SABRE_PCIERR", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Couldn't register PCIERR, err=%d.\n",
+ pbm->name, err);
+
+ tmp = upa_readq(base + SABRE_PCICTRL);
+ tmp |= SABRE_PCICTRL_ERREN;
+ upa_writeq(tmp, base + SABRE_PCICTRL);
+}
+
+static void apb_init(struct pci_bus *sabre_bus)
+{
+ struct pci_dev *pdev;
+
+ list_for_each_entry(pdev, &sabre_bus->devices, bus_list) {
+ if (pdev->vendor == PCI_VENDOR_ID_SUN &&
+ pdev->device == PCI_DEVICE_ID_SUN_SIMBA) {
+ u16 word16;
+
+ pci_read_config_word(pdev, PCI_COMMAND, &word16);
+ word16 |= PCI_COMMAND_SERR | PCI_COMMAND_PARITY |
+ PCI_COMMAND_MASTER | PCI_COMMAND_MEMORY |
+ PCI_COMMAND_IO;
+ pci_write_config_word(pdev, PCI_COMMAND, word16);
+
+ /* Status register bits are "write 1 to clear". */
+ pci_write_config_word(pdev, PCI_STATUS, 0xffff);
+ pci_write_config_word(pdev, PCI_SEC_STATUS, 0xffff);
+
+ /* Use a primary/seconday latency timer value
+ * of 64.
+ */
+ pci_write_config_byte(pdev, PCI_LATENCY_TIMER, 64);
+ pci_write_config_byte(pdev, PCI_SEC_LATENCY_TIMER, 64);
+
+ /* Enable reporting/forwarding of master aborts,
+ * parity, and SERR.
+ */
+ pci_write_config_byte(pdev, PCI_BRIDGE_CONTROL,
+ (PCI_BRIDGE_CTL_PARITY |
+ PCI_BRIDGE_CTL_SERR |
+ PCI_BRIDGE_CTL_MASTER_ABORT));
+ }
+ }
+}
+
+static void sabre_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
+{
+ static int once;
+
+ /* The APB bridge speaks to the Sabre host PCI bridge
+ * at 66Mhz, but the front side of APB runs at 33Mhz
+ * for both segments.
+ *
+ * Hummingbird systems do not use APB, so they run
+ * at 66MHZ.
+ */
+ if (hummingbird_p)
+ pbm->is_66mhz_capable = 1;
+ else
+ pbm->is_66mhz_capable = 0;
+
+ /* This driver has not been verified to handle
+ * multiple SABREs yet, so trap this.
+ *
+ * Also note that the SABRE host bridge is hardwired
+ * to live at bus 0.
+ */
+ if (once != 0) {
+ printk(KERN_ERR PFX "Multiple controllers unsupported.\n");
+ return;
+ }
+ once++;
+
+ pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
+ if (!pbm->pci_bus)
+ return;
+
+ sabre_root_bus = pbm->pci_bus;
+
+ apb_init(pbm->pci_bus);
+
+ sabre_register_error_handlers(pbm);
+}
+
+static void sabre_pbm_init(struct pci_pbm_info *pbm,
+ struct platform_device *op)
+{
+ psycho_pbm_init_common(pbm, op, "SABRE", PBM_CHIP_TYPE_SABRE);
+ pbm->pci_afsr = pbm->controller_regs + SABRE_PIOAFSR;
+ pbm->pci_afar = pbm->controller_regs + SABRE_PIOAFAR;
+ pbm->pci_csr = pbm->controller_regs + SABRE_PCICTRL;
+ sabre_scan_bus(pbm, &op->dev);
+}
+
+static const struct of_device_id sabre_match[];
+static int sabre_probe(struct platform_device *op)
+{
+ const struct of_device_id *match;
+ const struct linux_prom64_registers *pr_regs;
+ struct device_node *dp = op->dev.of_node;
+ struct pci_pbm_info *pbm;
+ u32 upa_portid, dma_mask;
+ struct iommu *iommu;
+ int tsbsize, err;
+ const u32 *vdma;
+ u64 clear_irq;
+
+ match = of_match_device(sabre_match, &op->dev);
+ hummingbird_p = match && (match->data != NULL);
+ if (!hummingbird_p) {
+ struct device_node *cpu_dp;
+
+ /* Of course, Sun has to encode things a thousand
+ * different ways, inconsistently.
+ */
+ for_each_node_by_type(cpu_dp, "cpu") {
+ if (!strcmp(cpu_dp->name, "SUNW,UltraSPARC-IIe"))
+ hummingbird_p = 1;
+ }
+ }
+
+ err = -ENOMEM;
+ pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+ if (!pbm) {
+ printk(KERN_ERR PFX "Cannot allocate pci_pbm_info.\n");
+ goto out_err;
+ }
+
+ iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
+ if (!iommu) {
+ printk(KERN_ERR PFX "Cannot allocate PBM iommu.\n");
+ goto out_free_controller;
+ }
+
+ pbm->iommu = iommu;
+
+ upa_portid = of_getintprop_default(dp, "upa-portid", 0xff);
+
+ pbm->portid = upa_portid;
+
+ /*
+ * Map in SABRE register set and report the presence of this SABRE.
+ */
+
+ pr_regs = of_get_property(dp, "reg", NULL);
+ err = -ENODEV;
+ if (!pr_regs) {
+ printk(KERN_ERR PFX "No reg property\n");
+ goto out_free_iommu;
+ }
+
+ /*
+ * First REG in property is base of entire SABRE register space.
+ */
+ pbm->controller_regs = pr_regs[0].phys_addr;
+
+ /* Clear interrupts */
+
+ /* PCI first */
+ for (clear_irq = SABRE_ICLR_A_SLOT0; clear_irq < SABRE_ICLR_B_SLOT0 + 0x80; clear_irq += 8)
+ upa_writeq(0x0UL, pbm->controller_regs + clear_irq);
+
+ /* Then OBIO */
+ for (clear_irq = SABRE_ICLR_SCSI; clear_irq < SABRE_ICLR_SCSI + 0x80; clear_irq += 8)
+ upa_writeq(0x0UL, pbm->controller_regs + clear_irq);
+
+ /* Error interrupts are enabled later after the bus scan. */
+ upa_writeq((SABRE_PCICTRL_MRLEN | SABRE_PCICTRL_SERR |
+ SABRE_PCICTRL_ARBPARK | SABRE_PCICTRL_AEN),
+ pbm->controller_regs + SABRE_PCICTRL);
+
+ /* Now map in PCI config space for entire SABRE. */
+ pbm->config_space = pbm->controller_regs + SABRE_CONFIGSPACE;
+
+ vdma = of_get_property(dp, "virtual-dma", NULL);
+ if (!vdma) {
+ printk(KERN_ERR PFX "No virtual-dma property\n");
+ goto out_free_iommu;
+ }
+
+ dma_mask = vdma[0];
+ switch(vdma[1]) {
+ case 0x20000000:
+ dma_mask |= 0x1fffffff;
+ tsbsize = 64;
+ break;
+ case 0x40000000:
+ dma_mask |= 0x3fffffff;
+ tsbsize = 128;
+ break;
+
+ case 0x80000000:
+ dma_mask |= 0x7fffffff;
+ tsbsize = 128;
+ break;
+ default:
+ printk(KERN_ERR PFX "Strange virtual-dma size.\n");
+ goto out_free_iommu;
+ }
+
+ err = psycho_iommu_init(pbm, tsbsize, vdma[0], dma_mask, SABRE_WRSYNC);
+ if (err)
+ goto out_free_iommu;
+
+ /*
+ * Look for APB underneath.
+ */
+ sabre_pbm_init(pbm, op);
+
+ pbm->next = pci_pbm_root;
+ pci_pbm_root = pbm;
+
+ dev_set_drvdata(&op->dev, pbm);
+
+ return 0;
+
+out_free_iommu:
+ kfree(pbm->iommu);
+
+out_free_controller:
+ kfree(pbm);
+
+out_err:
+ return err;
+}
+
+static const struct of_device_id sabre_match[] = {
+ {
+ .name = "pci",
+ .compatible = "pci108e,a001",
+ .data = (void *) 1,
+ },
+ {
+ .name = "pci",
+ .compatible = "pci108e,a000",
+ },
+ {},
+};
+
+static struct platform_driver sabre_driver = {
+ .driver = {
+ .name = DRIVER_NAME,
+ .owner = THIS_MODULE,
+ .of_match_table = sabre_match,
+ },
+ .probe = sabre_probe,
+};
+
+static int __init sabre_init(void)
+{
+ return platform_driver_register(&sabre_driver);
+}
+
+subsys_initcall(sabre_init);
diff --git a/arch/sparc/kernel/pci_schizo.c b/arch/sparc/kernel/pci_schizo.c
new file mode 100644
index 00000000000..8f76f23dac3
--- /dev/null
+++ b/arch/sparc/kernel/pci_schizo.c
@@ -0,0 +1,1509 @@
+/* pci_schizo.c: SCHIZO/TOMATILLO specific PCI controller support.
+ *
+ * Copyright (C) 2001, 2002, 2003, 2007, 2008 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/export.h>
+#include <linux/interrupt.h>
+#include <linux/of_device.h>
+
+#include <asm/iommu.h>
+#include <asm/irq.h>
+#include <asm/pstate.h>
+#include <asm/prom.h>
+#include <asm/upa.h>
+
+#include "pci_impl.h"
+#include "iommu_common.h"
+
+#define DRIVER_NAME "schizo"
+#define PFX DRIVER_NAME ": "
+
+/* This is a convention that at least Excalibur and Merlin
+ * follow. I suppose the SCHIZO used in Starcat and friends
+ * will do similar.
+ *
+ * The only way I could see this changing is if the newlink
+ * block requires more space in Schizo's address space than
+ * they predicted, thus requiring an address space reorg when
+ * the newer Schizo is taped out.
+ */
+
+/* Streaming buffer control register. */
+#define SCHIZO_STRBUF_CTRL_LPTR 0x00000000000000f0UL /* LRU Lock Pointer */
+#define SCHIZO_STRBUF_CTRL_LENAB 0x0000000000000008UL /* LRU Lock Enable */
+#define SCHIZO_STRBUF_CTRL_RRDIS 0x0000000000000004UL /* Rerun Disable */
+#define SCHIZO_STRBUF_CTRL_DENAB 0x0000000000000002UL /* Diagnostic Mode Enable */
+#define SCHIZO_STRBUF_CTRL_ENAB 0x0000000000000001UL /* Streaming Buffer Enable */
+
+/* IOMMU control register. */
+#define SCHIZO_IOMMU_CTRL_RESV 0xfffffffff9000000UL /* Reserved */
+#define SCHIZO_IOMMU_CTRL_XLTESTAT 0x0000000006000000UL /* Translation Error Status */
+#define SCHIZO_IOMMU_CTRL_XLTEERR 0x0000000001000000UL /* Translation Error encountered */
+#define SCHIZO_IOMMU_CTRL_LCKEN 0x0000000000800000UL /* Enable translation locking */
+#define SCHIZO_IOMMU_CTRL_LCKPTR 0x0000000000780000UL /* Translation lock pointer */
+#define SCHIZO_IOMMU_CTRL_TSBSZ 0x0000000000070000UL /* TSB Size */
+#define SCHIZO_IOMMU_TSBSZ_1K 0x0000000000000000UL /* TSB Table 1024 8-byte entries */
+#define SCHIZO_IOMMU_TSBSZ_2K 0x0000000000010000UL /* TSB Table 2048 8-byte entries */
+#define SCHIZO_IOMMU_TSBSZ_4K 0x0000000000020000UL /* TSB Table 4096 8-byte entries */
+#define SCHIZO_IOMMU_TSBSZ_8K 0x0000000000030000UL /* TSB Table 8192 8-byte entries */
+#define SCHIZO_IOMMU_TSBSZ_16K 0x0000000000040000UL /* TSB Table 16k 8-byte entries */
+#define SCHIZO_IOMMU_TSBSZ_32K 0x0000000000050000UL /* TSB Table 32k 8-byte entries */
+#define SCHIZO_IOMMU_TSBSZ_64K 0x0000000000060000UL /* TSB Table 64k 8-byte entries */
+#define SCHIZO_IOMMU_TSBSZ_128K 0x0000000000070000UL /* TSB Table 128k 8-byte entries */
+#define SCHIZO_IOMMU_CTRL_RESV2 0x000000000000fff8UL /* Reserved */
+#define SCHIZO_IOMMU_CTRL_TBWSZ 0x0000000000000004UL /* Assumed page size, 0=8k 1=64k */
+#define SCHIZO_IOMMU_CTRL_DENAB 0x0000000000000002UL /* Diagnostic mode enable */
+#define SCHIZO_IOMMU_CTRL_ENAB 0x0000000000000001UL /* IOMMU Enable */
+
+/* Schizo config space address format is nearly identical to
+ * that of PSYCHO:
+ *
+ * 32 24 23 16 15 11 10 8 7 2 1 0
+ * ---------------------------------------------------------
+ * |0 0 0 0 0 0 0 0 0| bus | device | function | reg | 0 0 |
+ * ---------------------------------------------------------
+ */
+#define SCHIZO_CONFIG_BASE(PBM) ((PBM)->config_space)
+#define SCHIZO_CONFIG_ENCODE(BUS, DEVFN, REG) \
+ (((unsigned long)(BUS) << 16) | \
+ ((unsigned long)(DEVFN) << 8) | \
+ ((unsigned long)(REG)))
+
+static void *schizo_pci_config_mkaddr(struct pci_pbm_info *pbm,
+ unsigned char bus,
+ unsigned int devfn,
+ int where)
+{
+ if (!pbm)
+ return NULL;
+ bus -= pbm->pci_first_busno;
+ return (void *)
+ (SCHIZO_CONFIG_BASE(pbm) |
+ SCHIZO_CONFIG_ENCODE(bus, devfn, where));
+}
+
+/* SCHIZO error handling support. */
+enum schizo_error_type {
+ UE_ERR, CE_ERR, PCI_ERR, SAFARI_ERR
+};
+
+static DEFINE_SPINLOCK(stc_buf_lock);
+static unsigned long stc_error_buf[128];
+static unsigned long stc_tag_buf[16];
+static unsigned long stc_line_buf[16];
+
+#define SCHIZO_UE_INO 0x30 /* Uncorrectable ECC error */
+#define SCHIZO_CE_INO 0x31 /* Correctable ECC error */
+#define SCHIZO_PCIERR_A_INO 0x32 /* PBM A PCI bus error */
+#define SCHIZO_PCIERR_B_INO 0x33 /* PBM B PCI bus error */
+#define SCHIZO_SERR_INO 0x34 /* Safari interface error */
+
+#define SCHIZO_STC_ERR 0xb800UL /* --> 0xba00 */
+#define SCHIZO_STC_TAG 0xba00UL /* --> 0xba80 */
+#define SCHIZO_STC_LINE 0xbb00UL /* --> 0xbb80 */
+
+#define SCHIZO_STCERR_WRITE 0x2UL
+#define SCHIZO_STCERR_READ 0x1UL
+
+#define SCHIZO_STCTAG_PPN 0x3fffffff00000000UL
+#define SCHIZO_STCTAG_VPN 0x00000000ffffe000UL
+#define SCHIZO_STCTAG_VALID 0x8000000000000000UL
+#define SCHIZO_STCTAG_READ 0x4000000000000000UL
+
+#define SCHIZO_STCLINE_LINDX 0x0000000007800000UL
+#define SCHIZO_STCLINE_SPTR 0x000000000007e000UL
+#define SCHIZO_STCLINE_LADDR 0x0000000000001fc0UL
+#define SCHIZO_STCLINE_EPTR 0x000000000000003fUL
+#define SCHIZO_STCLINE_VALID 0x0000000000600000UL
+#define SCHIZO_STCLINE_FOFN 0x0000000000180000UL
+
+static void __schizo_check_stc_error_pbm(struct pci_pbm_info *pbm,
+ enum schizo_error_type type)
+{
+ struct strbuf *strbuf = &pbm->stc;
+ unsigned long regbase = pbm->pbm_regs;
+ unsigned long err_base, tag_base, line_base;
+ u64 control;
+ int i;
+
+ err_base = regbase + SCHIZO_STC_ERR;
+ tag_base = regbase + SCHIZO_STC_TAG;
+ line_base = regbase + SCHIZO_STC_LINE;
+
+ spin_lock(&stc_buf_lock);
+
+ /* This is __REALLY__ dangerous. When we put the
+ * streaming buffer into diagnostic mode to probe
+ * it's tags and error status, we _must_ clear all
+ * of the line tag valid bits before re-enabling
+ * the streaming buffer. If any dirty data lives
+ * in the STC when we do this, we will end up
+ * invalidating it before it has a chance to reach
+ * main memory.
+ */
+ control = upa_readq(strbuf->strbuf_control);
+ upa_writeq((control | SCHIZO_STRBUF_CTRL_DENAB),
+ strbuf->strbuf_control);
+ for (i = 0; i < 128; i++) {
+ unsigned long val;
+
+ val = upa_readq(err_base + (i * 8UL));
+ upa_writeq(0UL, err_base + (i * 8UL));
+ stc_error_buf[i] = val;
+ }
+ for (i = 0; i < 16; i++) {
+ stc_tag_buf[i] = upa_readq(tag_base + (i * 8UL));
+ stc_line_buf[i] = upa_readq(line_base + (i * 8UL));
+ upa_writeq(0UL, tag_base + (i * 8UL));
+ upa_writeq(0UL, line_base + (i * 8UL));
+ }
+
+ /* OK, state is logged, exit diagnostic mode. */
+ upa_writeq(control, strbuf->strbuf_control);
+
+ for (i = 0; i < 16; i++) {
+ int j, saw_error, first, last;
+
+ saw_error = 0;
+ first = i * 8;
+ last = first + 8;
+ for (j = first; j < last; j++) {
+ unsigned long errval = stc_error_buf[j];
+ if (errval != 0) {
+ saw_error++;
+ printk("%s: STC_ERR(%d)[wr(%d)rd(%d)]\n",
+ pbm->name,
+ j,
+ (errval & SCHIZO_STCERR_WRITE) ? 1 : 0,
+ (errval & SCHIZO_STCERR_READ) ? 1 : 0);
+ }
+ }
+ if (saw_error != 0) {
+ unsigned long tagval = stc_tag_buf[i];
+ unsigned long lineval = stc_line_buf[i];
+ printk("%s: STC_TAG(%d)[PA(%016lx)VA(%08lx)V(%d)R(%d)]\n",
+ pbm->name,
+ i,
+ ((tagval & SCHIZO_STCTAG_PPN) >> 19UL),
+ (tagval & SCHIZO_STCTAG_VPN),
+ ((tagval & SCHIZO_STCTAG_VALID) ? 1 : 0),
+ ((tagval & SCHIZO_STCTAG_READ) ? 1 : 0));
+
+ /* XXX Should spit out per-bank error information... -DaveM */
+ printk("%s: STC_LINE(%d)[LIDX(%lx)SP(%lx)LADDR(%lx)EP(%lx)"
+ "V(%d)FOFN(%d)]\n",
+ pbm->name,
+ i,
+ ((lineval & SCHIZO_STCLINE_LINDX) >> 23UL),
+ ((lineval & SCHIZO_STCLINE_SPTR) >> 13UL),
+ ((lineval & SCHIZO_STCLINE_LADDR) >> 6UL),
+ ((lineval & SCHIZO_STCLINE_EPTR) >> 0UL),
+ ((lineval & SCHIZO_STCLINE_VALID) ? 1 : 0),
+ ((lineval & SCHIZO_STCLINE_FOFN) ? 1 : 0));
+ }
+ }
+
+ spin_unlock(&stc_buf_lock);
+}
+
+/* IOMMU is per-PBM in Schizo, so interrogate both for anonymous
+ * controller level errors.
+ */
+
+#define SCHIZO_IOMMU_TAG 0xa580UL
+#define SCHIZO_IOMMU_DATA 0xa600UL
+
+#define SCHIZO_IOMMU_TAG_CTXT 0x0000001ffe000000UL
+#define SCHIZO_IOMMU_TAG_ERRSTS 0x0000000001800000UL
+#define SCHIZO_IOMMU_TAG_ERR 0x0000000000400000UL
+#define SCHIZO_IOMMU_TAG_WRITE 0x0000000000200000UL
+#define SCHIZO_IOMMU_TAG_STREAM 0x0000000000100000UL
+#define SCHIZO_IOMMU_TAG_SIZE 0x0000000000080000UL
+#define SCHIZO_IOMMU_TAG_VPAGE 0x000000000007ffffUL
+
+#define SCHIZO_IOMMU_DATA_VALID 0x0000000100000000UL
+#define SCHIZO_IOMMU_DATA_CACHE 0x0000000040000000UL
+#define SCHIZO_IOMMU_DATA_PPAGE 0x000000003fffffffUL
+
+static void schizo_check_iommu_error_pbm(struct pci_pbm_info *pbm,
+ enum schizo_error_type type)
+{
+ struct iommu *iommu = pbm->iommu;
+ unsigned long iommu_tag[16];
+ unsigned long iommu_data[16];
+ unsigned long flags;
+ u64 control;
+ int i;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ control = upa_readq(iommu->iommu_control);
+ if (control & SCHIZO_IOMMU_CTRL_XLTEERR) {
+ unsigned long base;
+ char *type_string;
+
+ /* Clear the error encountered bit. */
+ control &= ~SCHIZO_IOMMU_CTRL_XLTEERR;
+ upa_writeq(control, iommu->iommu_control);
+
+ switch((control & SCHIZO_IOMMU_CTRL_XLTESTAT) >> 25UL) {
+ case 0:
+ type_string = "Protection Error";
+ break;
+ case 1:
+ type_string = "Invalid Error";
+ break;
+ case 2:
+ type_string = "TimeOut Error";
+ break;
+ case 3:
+ default:
+ type_string = "ECC Error";
+ break;
+ }
+ printk("%s: IOMMU Error, type[%s]\n",
+ pbm->name, type_string);
+
+ /* Put the IOMMU into diagnostic mode and probe
+ * it's TLB for entries with error status.
+ *
+ * It is very possible for another DVMA to occur
+ * while we do this probe, and corrupt the system
+ * further. But we are so screwed at this point
+ * that we are likely to crash hard anyways, so
+ * get as much diagnostic information to the
+ * console as we can.
+ */
+ upa_writeq(control | SCHIZO_IOMMU_CTRL_DENAB,
+ iommu->iommu_control);
+
+ base = pbm->pbm_regs;
+
+ for (i = 0; i < 16; i++) {
+ iommu_tag[i] =
+ upa_readq(base + SCHIZO_IOMMU_TAG + (i * 8UL));
+ iommu_data[i] =
+ upa_readq(base + SCHIZO_IOMMU_DATA + (i * 8UL));
+
+ /* Now clear out the entry. */
+ upa_writeq(0, base + SCHIZO_IOMMU_TAG + (i * 8UL));
+ upa_writeq(0, base + SCHIZO_IOMMU_DATA + (i * 8UL));
+ }
+
+ /* Leave diagnostic mode. */
+ upa_writeq(control, iommu->iommu_control);
+
+ for (i = 0; i < 16; i++) {
+ unsigned long tag, data;
+
+ tag = iommu_tag[i];
+ if (!(tag & SCHIZO_IOMMU_TAG_ERR))
+ continue;
+
+ data = iommu_data[i];
+ switch((tag & SCHIZO_IOMMU_TAG_ERRSTS) >> 23UL) {
+ case 0:
+ type_string = "Protection Error";
+ break;
+ case 1:
+ type_string = "Invalid Error";
+ break;
+ case 2:
+ type_string = "TimeOut Error";
+ break;
+ case 3:
+ default:
+ type_string = "ECC Error";
+ break;
+ }
+ printk("%s: IOMMU TAG(%d)[error(%s) ctx(%x) wr(%d) str(%d) "
+ "sz(%dK) vpg(%08lx)]\n",
+ pbm->name, i, type_string,
+ (int)((tag & SCHIZO_IOMMU_TAG_CTXT) >> 25UL),
+ ((tag & SCHIZO_IOMMU_TAG_WRITE) ? 1 : 0),
+ ((tag & SCHIZO_IOMMU_TAG_STREAM) ? 1 : 0),
+ ((tag & SCHIZO_IOMMU_TAG_SIZE) ? 64 : 8),
+ (tag & SCHIZO_IOMMU_TAG_VPAGE) << IOMMU_PAGE_SHIFT);
+ printk("%s: IOMMU DATA(%d)[valid(%d) cache(%d) ppg(%016lx)]\n",
+ pbm->name, i,
+ ((data & SCHIZO_IOMMU_DATA_VALID) ? 1 : 0),
+ ((data & SCHIZO_IOMMU_DATA_CACHE) ? 1 : 0),
+ (data & SCHIZO_IOMMU_DATA_PPAGE) << IOMMU_PAGE_SHIFT);
+ }
+ }
+ if (pbm->stc.strbuf_enabled)
+ __schizo_check_stc_error_pbm(pbm, type);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static void schizo_check_iommu_error(struct pci_pbm_info *pbm,
+ enum schizo_error_type type)
+{
+ schizo_check_iommu_error_pbm(pbm, type);
+ if (pbm->sibling)
+ schizo_check_iommu_error_pbm(pbm->sibling, type);
+}
+
+/* Uncorrectable ECC error status gathering. */
+#define SCHIZO_UE_AFSR 0x10030UL
+#define SCHIZO_UE_AFAR 0x10038UL
+
+#define SCHIZO_UEAFSR_PPIO 0x8000000000000000UL /* Safari */
+#define SCHIZO_UEAFSR_PDRD 0x4000000000000000UL /* Safari/Tomatillo */
+#define SCHIZO_UEAFSR_PDWR 0x2000000000000000UL /* Safari */
+#define SCHIZO_UEAFSR_SPIO 0x1000000000000000UL /* Safari */
+#define SCHIZO_UEAFSR_SDMA 0x0800000000000000UL /* Safari/Tomatillo */
+#define SCHIZO_UEAFSR_ERRPNDG 0x0300000000000000UL /* Safari */
+#define SCHIZO_UEAFSR_BMSK 0x000003ff00000000UL /* Safari */
+#define SCHIZO_UEAFSR_QOFF 0x00000000c0000000UL /* Safari/Tomatillo */
+#define SCHIZO_UEAFSR_AID 0x000000001f000000UL /* Safari/Tomatillo */
+#define SCHIZO_UEAFSR_PARTIAL 0x0000000000800000UL /* Safari */
+#define SCHIZO_UEAFSR_OWNEDIN 0x0000000000400000UL /* Safari */
+#define SCHIZO_UEAFSR_MTAGSYND 0x00000000000f0000UL /* Safari */
+#define SCHIZO_UEAFSR_MTAG 0x000000000000e000UL /* Safari */
+#define SCHIZO_UEAFSR_ECCSYND 0x00000000000001ffUL /* Safari */
+
+static irqreturn_t schizo_ue_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ unsigned long afsr_reg = pbm->controller_regs + SCHIZO_UE_AFSR;
+ unsigned long afar_reg = pbm->controller_regs + SCHIZO_UE_AFAR;
+ unsigned long afsr, afar, error_bits;
+ int reported, limit;
+
+ /* Latch uncorrectable error status. */
+ afar = upa_readq(afar_reg);
+
+ /* If either of the error pending bits are set in the
+ * AFSR, the error status is being actively updated by
+ * the hardware and we must re-read to get a clean value.
+ */
+ limit = 1000;
+ do {
+ afsr = upa_readq(afsr_reg);
+ } while ((afsr & SCHIZO_UEAFSR_ERRPNDG) != 0 && --limit);
+
+ /* Clear the primary/secondary error status bits. */
+ error_bits = afsr &
+ (SCHIZO_UEAFSR_PPIO | SCHIZO_UEAFSR_PDRD | SCHIZO_UEAFSR_PDWR |
+ SCHIZO_UEAFSR_SPIO | SCHIZO_UEAFSR_SDMA);
+ if (!error_bits)
+ return IRQ_NONE;
+ upa_writeq(error_bits, afsr_reg);
+
+ /* Log the error. */
+ printk("%s: Uncorrectable Error, primary error type[%s]\n",
+ pbm->name,
+ (((error_bits & SCHIZO_UEAFSR_PPIO) ?
+ "PIO" :
+ ((error_bits & SCHIZO_UEAFSR_PDRD) ?
+ "DMA Read" :
+ ((error_bits & SCHIZO_UEAFSR_PDWR) ?
+ "DMA Write" : "???")))));
+ printk("%s: bytemask[%04lx] qword_offset[%lx] SAFARI_AID[%02lx]\n",
+ pbm->name,
+ (afsr & SCHIZO_UEAFSR_BMSK) >> 32UL,
+ (afsr & SCHIZO_UEAFSR_QOFF) >> 30UL,
+ (afsr & SCHIZO_UEAFSR_AID) >> 24UL);
+ printk("%s: partial[%d] owned_in[%d] mtag[%lx] mtag_synd[%lx] ecc_sync[%lx]\n",
+ pbm->name,
+ (afsr & SCHIZO_UEAFSR_PARTIAL) ? 1 : 0,
+ (afsr & SCHIZO_UEAFSR_OWNEDIN) ? 1 : 0,
+ (afsr & SCHIZO_UEAFSR_MTAG) >> 13UL,
+ (afsr & SCHIZO_UEAFSR_MTAGSYND) >> 16UL,
+ (afsr & SCHIZO_UEAFSR_ECCSYND) >> 0UL);
+ printk("%s: UE AFAR [%016lx]\n", pbm->name, afar);
+ printk("%s: UE Secondary errors [", pbm->name);
+ reported = 0;
+ if (afsr & SCHIZO_UEAFSR_SPIO) {
+ reported++;
+ printk("(PIO)");
+ }
+ if (afsr & SCHIZO_UEAFSR_SDMA) {
+ reported++;
+ printk("(DMA)");
+ }
+ if (!reported)
+ printk("(none)");
+ printk("]\n");
+
+ /* Interrogate IOMMU for error status. */
+ schizo_check_iommu_error(pbm, UE_ERR);
+
+ return IRQ_HANDLED;
+}
+
+#define SCHIZO_CE_AFSR 0x10040UL
+#define SCHIZO_CE_AFAR 0x10048UL
+
+#define SCHIZO_CEAFSR_PPIO 0x8000000000000000UL
+#define SCHIZO_CEAFSR_PDRD 0x4000000000000000UL
+#define SCHIZO_CEAFSR_PDWR 0x2000000000000000UL
+#define SCHIZO_CEAFSR_SPIO 0x1000000000000000UL
+#define SCHIZO_CEAFSR_SDMA 0x0800000000000000UL
+#define SCHIZO_CEAFSR_ERRPNDG 0x0300000000000000UL
+#define SCHIZO_CEAFSR_BMSK 0x000003ff00000000UL
+#define SCHIZO_CEAFSR_QOFF 0x00000000c0000000UL
+#define SCHIZO_CEAFSR_AID 0x000000001f000000UL
+#define SCHIZO_CEAFSR_PARTIAL 0x0000000000800000UL
+#define SCHIZO_CEAFSR_OWNEDIN 0x0000000000400000UL
+#define SCHIZO_CEAFSR_MTAGSYND 0x00000000000f0000UL
+#define SCHIZO_CEAFSR_MTAG 0x000000000000e000UL
+#define SCHIZO_CEAFSR_ECCSYND 0x00000000000001ffUL
+
+static irqreturn_t schizo_ce_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ unsigned long afsr_reg = pbm->controller_regs + SCHIZO_CE_AFSR;
+ unsigned long afar_reg = pbm->controller_regs + SCHIZO_CE_AFAR;
+ unsigned long afsr, afar, error_bits;
+ int reported, limit;
+
+ /* Latch error status. */
+ afar = upa_readq(afar_reg);
+
+ /* If either of the error pending bits are set in the
+ * AFSR, the error status is being actively updated by
+ * the hardware and we must re-read to get a clean value.
+ */
+ limit = 1000;
+ do {
+ afsr = upa_readq(afsr_reg);
+ } while ((afsr & SCHIZO_UEAFSR_ERRPNDG) != 0 && --limit);
+
+ /* Clear primary/secondary error status bits. */
+ error_bits = afsr &
+ (SCHIZO_CEAFSR_PPIO | SCHIZO_CEAFSR_PDRD | SCHIZO_CEAFSR_PDWR |
+ SCHIZO_CEAFSR_SPIO | SCHIZO_CEAFSR_SDMA);
+ if (!error_bits)
+ return IRQ_NONE;
+ upa_writeq(error_bits, afsr_reg);
+
+ /* Log the error. */
+ printk("%s: Correctable Error, primary error type[%s]\n",
+ pbm->name,
+ (((error_bits & SCHIZO_CEAFSR_PPIO) ?
+ "PIO" :
+ ((error_bits & SCHIZO_CEAFSR_PDRD) ?
+ "DMA Read" :
+ ((error_bits & SCHIZO_CEAFSR_PDWR) ?
+ "DMA Write" : "???")))));
+
+ /* XXX Use syndrome and afar to print out module string just like
+ * XXX UDB CE trap handler does... -DaveM
+ */
+ printk("%s: bytemask[%04lx] qword_offset[%lx] SAFARI_AID[%02lx]\n",
+ pbm->name,
+ (afsr & SCHIZO_UEAFSR_BMSK) >> 32UL,
+ (afsr & SCHIZO_UEAFSR_QOFF) >> 30UL,
+ (afsr & SCHIZO_UEAFSR_AID) >> 24UL);
+ printk("%s: partial[%d] owned_in[%d] mtag[%lx] mtag_synd[%lx] ecc_sync[%lx]\n",
+ pbm->name,
+ (afsr & SCHIZO_UEAFSR_PARTIAL) ? 1 : 0,
+ (afsr & SCHIZO_UEAFSR_OWNEDIN) ? 1 : 0,
+ (afsr & SCHIZO_UEAFSR_MTAG) >> 13UL,
+ (afsr & SCHIZO_UEAFSR_MTAGSYND) >> 16UL,
+ (afsr & SCHIZO_UEAFSR_ECCSYND) >> 0UL);
+ printk("%s: CE AFAR [%016lx]\n", pbm->name, afar);
+ printk("%s: CE Secondary errors [", pbm->name);
+ reported = 0;
+ if (afsr & SCHIZO_CEAFSR_SPIO) {
+ reported++;
+ printk("(PIO)");
+ }
+ if (afsr & SCHIZO_CEAFSR_SDMA) {
+ reported++;
+ printk("(DMA)");
+ }
+ if (!reported)
+ printk("(none)");
+ printk("]\n");
+
+ return IRQ_HANDLED;
+}
+
+#define SCHIZO_PCI_AFSR 0x2010UL
+#define SCHIZO_PCI_AFAR 0x2018UL
+
+#define SCHIZO_PCIAFSR_PMA 0x8000000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_PTA 0x4000000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_PRTRY 0x2000000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_PPERR 0x1000000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_PTTO 0x0800000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_PUNUS 0x0400000000000000UL /* Schizo */
+#define SCHIZO_PCIAFSR_SMA 0x0200000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_STA 0x0100000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_SRTRY 0x0080000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_SPERR 0x0040000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_STTO 0x0020000000000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_SUNUS 0x0010000000000000UL /* Schizo */
+#define SCHIZO_PCIAFSR_BMSK 0x000003ff00000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_BLK 0x0000000080000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_CFG 0x0000000040000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_MEM 0x0000000020000000UL /* Schizo/Tomatillo */
+#define SCHIZO_PCIAFSR_IO 0x0000000010000000UL /* Schizo/Tomatillo */
+
+#define SCHIZO_PCI_CTRL (0x2000UL)
+#define SCHIZO_PCICTRL_BUS_UNUS (1UL << 63UL) /* Safari */
+#define SCHIZO_PCICTRL_DTO_INT (1UL << 61UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_ARB_PRIO (0x1ff << 52UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_ESLCK (1UL << 51UL) /* Safari */
+#define SCHIZO_PCICTRL_ERRSLOT (7UL << 48UL) /* Safari */
+#define SCHIZO_PCICTRL_TTO_ERR (1UL << 38UL) /* Safari/Tomatillo */
+#define SCHIZO_PCICTRL_RTRY_ERR (1UL << 37UL) /* Safari/Tomatillo */
+#define SCHIZO_PCICTRL_DTO_ERR (1UL << 36UL) /* Safari/Tomatillo */
+#define SCHIZO_PCICTRL_SBH_ERR (1UL << 35UL) /* Safari */
+#define SCHIZO_PCICTRL_SERR (1UL << 34UL) /* Safari/Tomatillo */
+#define SCHIZO_PCICTRL_PCISPD (1UL << 33UL) /* Safari */
+#define SCHIZO_PCICTRL_MRM_PREF (1UL << 30UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_RDO_PREF (1UL << 29UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_RDL_PREF (1UL << 28UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_PTO (3UL << 24UL) /* Safari/Tomatillo */
+#define SCHIZO_PCICTRL_PTO_SHIFT 24UL
+#define SCHIZO_PCICTRL_TRWSW (7UL << 21UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_F_TGT_A (1UL << 20UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_S_DTO_INT (1UL << 19UL) /* Safari */
+#define SCHIZO_PCICTRL_F_TGT_RT (1UL << 19UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_SBH_INT (1UL << 18UL) /* Safari */
+#define SCHIZO_PCICTRL_T_DTO_INT (1UL << 18UL) /* Tomatillo */
+#define SCHIZO_PCICTRL_EEN (1UL << 17UL) /* Safari/Tomatillo */
+#define SCHIZO_PCICTRL_PARK (1UL << 16UL) /* Safari/Tomatillo */
+#define SCHIZO_PCICTRL_PCIRST (1UL << 8UL) /* Safari */
+#define SCHIZO_PCICTRL_ARB_S (0x3fUL << 0UL) /* Safari */
+#define SCHIZO_PCICTRL_ARB_T (0xffUL << 0UL) /* Tomatillo */
+
+static irqreturn_t schizo_pcierr_intr_other(struct pci_pbm_info *pbm)
+{
+ unsigned long csr_reg, csr, csr_error_bits;
+ irqreturn_t ret = IRQ_NONE;
+ u16 stat;
+
+ csr_reg = pbm->pbm_regs + SCHIZO_PCI_CTRL;
+ csr = upa_readq(csr_reg);
+ csr_error_bits =
+ csr & (SCHIZO_PCICTRL_BUS_UNUS |
+ SCHIZO_PCICTRL_TTO_ERR |
+ SCHIZO_PCICTRL_RTRY_ERR |
+ SCHIZO_PCICTRL_DTO_ERR |
+ SCHIZO_PCICTRL_SBH_ERR |
+ SCHIZO_PCICTRL_SERR);
+ if (csr_error_bits) {
+ /* Clear the errors. */
+ upa_writeq(csr, csr_reg);
+
+ /* Log 'em. */
+ if (csr_error_bits & SCHIZO_PCICTRL_BUS_UNUS)
+ printk("%s: Bus unusable error asserted.\n",
+ pbm->name);
+ if (csr_error_bits & SCHIZO_PCICTRL_TTO_ERR)
+ printk("%s: PCI TRDY# timeout error asserted.\n",
+ pbm->name);
+ if (csr_error_bits & SCHIZO_PCICTRL_RTRY_ERR)
+ printk("%s: PCI excessive retry error asserted.\n",
+ pbm->name);
+ if (csr_error_bits & SCHIZO_PCICTRL_DTO_ERR)
+ printk("%s: PCI discard timeout error asserted.\n",
+ pbm->name);
+ if (csr_error_bits & SCHIZO_PCICTRL_SBH_ERR)
+ printk("%s: PCI streaming byte hole error asserted.\n",
+ pbm->name);
+ if (csr_error_bits & SCHIZO_PCICTRL_SERR)
+ printk("%s: PCI SERR signal asserted.\n",
+ pbm->name);
+ ret = IRQ_HANDLED;
+ }
+ pci_read_config_word(pbm->pci_bus->self, PCI_STATUS, &stat);
+ if (stat & (PCI_STATUS_PARITY |
+ PCI_STATUS_SIG_TARGET_ABORT |
+ PCI_STATUS_REC_TARGET_ABORT |
+ PCI_STATUS_REC_MASTER_ABORT |
+ PCI_STATUS_SIG_SYSTEM_ERROR)) {
+ printk("%s: PCI bus error, PCI_STATUS[%04x]\n",
+ pbm->name, stat);
+ pci_write_config_word(pbm->pci_bus->self, PCI_STATUS, 0xffff);
+ ret = IRQ_HANDLED;
+ }
+ return ret;
+}
+
+static irqreturn_t schizo_pcierr_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ unsigned long afsr_reg, afar_reg, base;
+ unsigned long afsr, afar, error_bits;
+ int reported;
+
+ base = pbm->pbm_regs;
+
+ afsr_reg = base + SCHIZO_PCI_AFSR;
+ afar_reg = base + SCHIZO_PCI_AFAR;
+
+ /* Latch error status. */
+ afar = upa_readq(afar_reg);
+ afsr = upa_readq(afsr_reg);
+
+ /* Clear primary/secondary error status bits. */
+ error_bits = afsr &
+ (SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_PTA |
+ SCHIZO_PCIAFSR_PRTRY | SCHIZO_PCIAFSR_PPERR |
+ SCHIZO_PCIAFSR_PTTO | SCHIZO_PCIAFSR_PUNUS |
+ SCHIZO_PCIAFSR_SMA | SCHIZO_PCIAFSR_STA |
+ SCHIZO_PCIAFSR_SRTRY | SCHIZO_PCIAFSR_SPERR |
+ SCHIZO_PCIAFSR_STTO | SCHIZO_PCIAFSR_SUNUS);
+ if (!error_bits)
+ return schizo_pcierr_intr_other(pbm);
+ upa_writeq(error_bits, afsr_reg);
+
+ /* Log the error. */
+ printk("%s: PCI Error, primary error type[%s]\n",
+ pbm->name,
+ (((error_bits & SCHIZO_PCIAFSR_PMA) ?
+ "Master Abort" :
+ ((error_bits & SCHIZO_PCIAFSR_PTA) ?
+ "Target Abort" :
+ ((error_bits & SCHIZO_PCIAFSR_PRTRY) ?
+ "Excessive Retries" :
+ ((error_bits & SCHIZO_PCIAFSR_PPERR) ?
+ "Parity Error" :
+ ((error_bits & SCHIZO_PCIAFSR_PTTO) ?
+ "Timeout" :
+ ((error_bits & SCHIZO_PCIAFSR_PUNUS) ?
+ "Bus Unusable" : "???"))))))));
+ printk("%s: bytemask[%04lx] was_block(%d) space(%s)\n",
+ pbm->name,
+ (afsr & SCHIZO_PCIAFSR_BMSK) >> 32UL,
+ (afsr & SCHIZO_PCIAFSR_BLK) ? 1 : 0,
+ ((afsr & SCHIZO_PCIAFSR_CFG) ?
+ "Config" :
+ ((afsr & SCHIZO_PCIAFSR_MEM) ?
+ "Memory" :
+ ((afsr & SCHIZO_PCIAFSR_IO) ?
+ "I/O" : "???"))));
+ printk("%s: PCI AFAR [%016lx]\n",
+ pbm->name, afar);
+ printk("%s: PCI Secondary errors [",
+ pbm->name);
+ reported = 0;
+ if (afsr & SCHIZO_PCIAFSR_SMA) {
+ reported++;
+ printk("(Master Abort)");
+ }
+ if (afsr & SCHIZO_PCIAFSR_STA) {
+ reported++;
+ printk("(Target Abort)");
+ }
+ if (afsr & SCHIZO_PCIAFSR_SRTRY) {
+ reported++;
+ printk("(Excessive Retries)");
+ }
+ if (afsr & SCHIZO_PCIAFSR_SPERR) {
+ reported++;
+ printk("(Parity Error)");
+ }
+ if (afsr & SCHIZO_PCIAFSR_STTO) {
+ reported++;
+ printk("(Timeout)");
+ }
+ if (afsr & SCHIZO_PCIAFSR_SUNUS) {
+ reported++;
+ printk("(Bus Unusable)");
+ }
+ if (!reported)
+ printk("(none)");
+ printk("]\n");
+
+ /* For the error types shown, scan PBM's PCI bus for devices
+ * which have logged that error type.
+ */
+
+ /* If we see a Target Abort, this could be the result of an
+ * IOMMU translation error of some sort. It is extremely
+ * useful to log this information as usually it indicates
+ * a bug in the IOMMU support code or a PCI device driver.
+ */
+ if (error_bits & (SCHIZO_PCIAFSR_PTA | SCHIZO_PCIAFSR_STA)) {
+ schizo_check_iommu_error(pbm, PCI_ERR);
+ pci_scan_for_target_abort(pbm, pbm->pci_bus);
+ }
+ if (error_bits & (SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_SMA))
+ pci_scan_for_master_abort(pbm, pbm->pci_bus);
+
+ /* For excessive retries, PSYCHO/PBM will abort the device
+ * and there is no way to specifically check for excessive
+ * retries in the config space status registers. So what
+ * we hope is that we'll catch it via the master/target
+ * abort events.
+ */
+
+ if (error_bits & (SCHIZO_PCIAFSR_PPERR | SCHIZO_PCIAFSR_SPERR))
+ pci_scan_for_parity_error(pbm, pbm->pci_bus);
+
+ return IRQ_HANDLED;
+}
+
+#define SCHIZO_SAFARI_ERRLOG 0x10018UL
+
+#define SAFARI_ERRLOG_ERROUT 0x8000000000000000UL
+
+#define BUS_ERROR_BADCMD 0x4000000000000000UL /* Schizo/Tomatillo */
+#define BUS_ERROR_SSMDIS 0x2000000000000000UL /* Safari */
+#define BUS_ERROR_BADMA 0x1000000000000000UL /* Safari */
+#define BUS_ERROR_BADMB 0x0800000000000000UL /* Safari */
+#define BUS_ERROR_BADMC 0x0400000000000000UL /* Safari */
+#define BUS_ERROR_SNOOP_GR 0x0000000000200000UL /* Tomatillo */
+#define BUS_ERROR_SNOOP_PCI 0x0000000000100000UL /* Tomatillo */
+#define BUS_ERROR_SNOOP_RD 0x0000000000080000UL /* Tomatillo */
+#define BUS_ERROR_SNOOP_RDS 0x0000000000020000UL /* Tomatillo */
+#define BUS_ERROR_SNOOP_RDSA 0x0000000000010000UL /* Tomatillo */
+#define BUS_ERROR_SNOOP_OWN 0x0000000000008000UL /* Tomatillo */
+#define BUS_ERROR_SNOOP_RDO 0x0000000000004000UL /* Tomatillo */
+#define BUS_ERROR_CPU1PS 0x0000000000002000UL /* Safari */
+#define BUS_ERROR_WDATA_PERR 0x0000000000002000UL /* Tomatillo */
+#define BUS_ERROR_CPU1PB 0x0000000000001000UL /* Safari */
+#define BUS_ERROR_CTRL_PERR 0x0000000000001000UL /* Tomatillo */
+#define BUS_ERROR_CPU0PS 0x0000000000000800UL /* Safari */
+#define BUS_ERROR_SNOOP_ERR 0x0000000000000800UL /* Tomatillo */
+#define BUS_ERROR_CPU0PB 0x0000000000000400UL /* Safari */
+#define BUS_ERROR_JBUS_ILL_B 0x0000000000000400UL /* Tomatillo */
+#define BUS_ERROR_CIQTO 0x0000000000000200UL /* Safari */
+#define BUS_ERROR_LPQTO 0x0000000000000100UL /* Safari */
+#define BUS_ERROR_JBUS_ILL_C 0x0000000000000100UL /* Tomatillo */
+#define BUS_ERROR_SFPQTO 0x0000000000000080UL /* Safari */
+#define BUS_ERROR_UFPQTO 0x0000000000000040UL /* Safari */
+#define BUS_ERROR_RD_PERR 0x0000000000000040UL /* Tomatillo */
+#define BUS_ERROR_APERR 0x0000000000000020UL /* Safari/Tomatillo */
+#define BUS_ERROR_UNMAP 0x0000000000000010UL /* Safari/Tomatillo */
+#define BUS_ERROR_BUSERR 0x0000000000000004UL /* Safari/Tomatillo */
+#define BUS_ERROR_TIMEOUT 0x0000000000000002UL /* Safari/Tomatillo */
+#define BUS_ERROR_ILL 0x0000000000000001UL /* Safari */
+
+/* We only expect UNMAP errors here. The rest of the Safari errors
+ * are marked fatal and thus cause a system reset.
+ */
+static irqreturn_t schizo_safarierr_intr(int irq, void *dev_id)
+{
+ struct pci_pbm_info *pbm = dev_id;
+ u64 errlog;
+
+ errlog = upa_readq(pbm->controller_regs + SCHIZO_SAFARI_ERRLOG);
+ upa_writeq(errlog & ~(SAFARI_ERRLOG_ERROUT),
+ pbm->controller_regs + SCHIZO_SAFARI_ERRLOG);
+
+ if (!(errlog & BUS_ERROR_UNMAP)) {
+ printk("%s: Unexpected Safari/JBUS error interrupt, errlog[%016llx]\n",
+ pbm->name, errlog);
+
+ return IRQ_HANDLED;
+ }
+
+ printk("%s: Safari/JBUS interrupt, UNMAPPED error, interrogating IOMMUs.\n",
+ pbm->name);
+ schizo_check_iommu_error(pbm, SAFARI_ERR);
+
+ return IRQ_HANDLED;
+}
+
+/* Nearly identical to PSYCHO equivalents... */
+#define SCHIZO_ECC_CTRL 0x10020UL
+#define SCHIZO_ECCCTRL_EE 0x8000000000000000UL /* Enable ECC Checking */
+#define SCHIZO_ECCCTRL_UE 0x4000000000000000UL /* Enable UE Interrupts */
+#define SCHIZO_ECCCTRL_CE 0x2000000000000000UL /* Enable CE INterrupts */
+
+#define SCHIZO_SAFARI_ERRCTRL 0x10008UL
+#define SCHIZO_SAFERRCTRL_EN 0x8000000000000000UL
+#define SCHIZO_SAFARI_IRQCTRL 0x10010UL
+#define SCHIZO_SAFIRQCTRL_EN 0x8000000000000000UL
+
+static int pbm_routes_this_ino(struct pci_pbm_info *pbm, u32 ino)
+{
+ ino &= IMAP_INO;
+
+ if (pbm->ino_bitmap & (1UL << ino))
+ return 1;
+
+ return 0;
+}
+
+/* How the Tomatillo IRQs are routed around is pure guesswork here.
+ *
+ * All the Tomatillo devices I see in prtconf dumps seem to have only
+ * a single PCI bus unit attached to it. It would seem they are separate
+ * devices because their PortID (ie. JBUS ID) values are all different
+ * and thus the registers are mapped to totally different locations.
+ *
+ * However, two Tomatillo's look "similar" in that the only difference
+ * in their PortID is the lowest bit.
+ *
+ * So if we were to ignore this lower bit, it certainly looks like two
+ * PCI bus units of the same Tomatillo. I still have not really
+ * figured this out...
+ */
+static void tomatillo_register_error_handlers(struct pci_pbm_info *pbm)
+{
+ struct platform_device *op = of_find_device_by_node(pbm->op->dev.of_node);
+ u64 tmp, err_mask, err_no_mask;
+ int err;
+
+ /* Tomatillo IRQ property layout is:
+ * 0: PCIERR
+ * 1: UE ERR
+ * 2: CE ERR
+ * 3: SERR
+ * 4: POWER FAIL?
+ */
+
+ if (pbm_routes_this_ino(pbm, SCHIZO_UE_INO)) {
+ err = request_irq(op->archdata.irqs[1], schizo_ue_intr, 0,
+ "TOMATILLO_UE", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Could not register UE, "
+ "err=%d\n", pbm->name, err);
+ }
+ if (pbm_routes_this_ino(pbm, SCHIZO_CE_INO)) {
+ err = request_irq(op->archdata.irqs[2], schizo_ce_intr, 0,
+ "TOMATILLO_CE", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Could not register CE, "
+ "err=%d\n", pbm->name, err);
+ }
+ err = 0;
+ if (pbm_routes_this_ino(pbm, SCHIZO_PCIERR_A_INO)) {
+ err = request_irq(op->archdata.irqs[0], schizo_pcierr_intr, 0,
+ "TOMATILLO_PCIERR", pbm);
+ } else if (pbm_routes_this_ino(pbm, SCHIZO_PCIERR_B_INO)) {
+ err = request_irq(op->archdata.irqs[0], schizo_pcierr_intr, 0,
+ "TOMATILLO_PCIERR", pbm);
+ }
+ if (err)
+ printk(KERN_WARNING "%s: Could not register PCIERR, "
+ "err=%d\n", pbm->name, err);
+
+ if (pbm_routes_this_ino(pbm, SCHIZO_SERR_INO)) {
+ err = request_irq(op->archdata.irqs[3], schizo_safarierr_intr, 0,
+ "TOMATILLO_SERR", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Could not register SERR, "
+ "err=%d\n", pbm->name, err);
+ }
+
+ /* Enable UE and CE interrupts for controller. */
+ upa_writeq((SCHIZO_ECCCTRL_EE |
+ SCHIZO_ECCCTRL_UE |
+ SCHIZO_ECCCTRL_CE), pbm->controller_regs + SCHIZO_ECC_CTRL);
+
+ /* Enable PCI Error interrupts and clear error
+ * bits.
+ */
+ err_mask = (SCHIZO_PCICTRL_BUS_UNUS |
+ SCHIZO_PCICTRL_TTO_ERR |
+ SCHIZO_PCICTRL_RTRY_ERR |
+ SCHIZO_PCICTRL_SERR |
+ SCHIZO_PCICTRL_EEN);
+
+ err_no_mask = SCHIZO_PCICTRL_DTO_ERR;
+
+ tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_CTRL);
+ tmp |= err_mask;
+ tmp &= ~err_no_mask;
+ upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_CTRL);
+
+ err_mask = (SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_PTA |
+ SCHIZO_PCIAFSR_PRTRY | SCHIZO_PCIAFSR_PPERR |
+ SCHIZO_PCIAFSR_PTTO |
+ SCHIZO_PCIAFSR_SMA | SCHIZO_PCIAFSR_STA |
+ SCHIZO_PCIAFSR_SRTRY | SCHIZO_PCIAFSR_SPERR |
+ SCHIZO_PCIAFSR_STTO);
+
+ upa_writeq(err_mask, pbm->pbm_regs + SCHIZO_PCI_AFSR);
+
+ err_mask = (BUS_ERROR_BADCMD | BUS_ERROR_SNOOP_GR |
+ BUS_ERROR_SNOOP_PCI | BUS_ERROR_SNOOP_RD |
+ BUS_ERROR_SNOOP_RDS | BUS_ERROR_SNOOP_RDSA |
+ BUS_ERROR_SNOOP_OWN | BUS_ERROR_SNOOP_RDO |
+ BUS_ERROR_WDATA_PERR | BUS_ERROR_CTRL_PERR |
+ BUS_ERROR_SNOOP_ERR | BUS_ERROR_JBUS_ILL_B |
+ BUS_ERROR_JBUS_ILL_C | BUS_ERROR_RD_PERR |
+ BUS_ERROR_APERR | BUS_ERROR_UNMAP |
+ BUS_ERROR_BUSERR | BUS_ERROR_TIMEOUT);
+
+ upa_writeq((SCHIZO_SAFERRCTRL_EN | err_mask),
+ pbm->controller_regs + SCHIZO_SAFARI_ERRCTRL);
+
+ upa_writeq((SCHIZO_SAFIRQCTRL_EN | (BUS_ERROR_UNMAP)),
+ pbm->controller_regs + SCHIZO_SAFARI_IRQCTRL);
+}
+
+static void schizo_register_error_handlers(struct pci_pbm_info *pbm)
+{
+ struct platform_device *op = of_find_device_by_node(pbm->op->dev.of_node);
+ u64 tmp, err_mask, err_no_mask;
+ int err;
+
+ /* Schizo IRQ property layout is:
+ * 0: PCIERR
+ * 1: UE ERR
+ * 2: CE ERR
+ * 3: SERR
+ * 4: POWER FAIL?
+ */
+
+ if (pbm_routes_this_ino(pbm, SCHIZO_UE_INO)) {
+ err = request_irq(op->archdata.irqs[1], schizo_ue_intr, 0,
+ "SCHIZO_UE", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Could not register UE, "
+ "err=%d\n", pbm->name, err);
+ }
+ if (pbm_routes_this_ino(pbm, SCHIZO_CE_INO)) {
+ err = request_irq(op->archdata.irqs[2], schizo_ce_intr, 0,
+ "SCHIZO_CE", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Could not register CE, "
+ "err=%d\n", pbm->name, err);
+ }
+ err = 0;
+ if (pbm_routes_this_ino(pbm, SCHIZO_PCIERR_A_INO)) {
+ err = request_irq(op->archdata.irqs[0], schizo_pcierr_intr, 0,
+ "SCHIZO_PCIERR", pbm);
+ } else if (pbm_routes_this_ino(pbm, SCHIZO_PCIERR_B_INO)) {
+ err = request_irq(op->archdata.irqs[0], schizo_pcierr_intr, 0,
+ "SCHIZO_PCIERR", pbm);
+ }
+ if (err)
+ printk(KERN_WARNING "%s: Could not register PCIERR, "
+ "err=%d\n", pbm->name, err);
+
+ if (pbm_routes_this_ino(pbm, SCHIZO_SERR_INO)) {
+ err = request_irq(op->archdata.irqs[3], schizo_safarierr_intr, 0,
+ "SCHIZO_SERR", pbm);
+ if (err)
+ printk(KERN_WARNING "%s: Could not register SERR, "
+ "err=%d\n", pbm->name, err);
+ }
+
+ /* Enable UE and CE interrupts for controller. */
+ upa_writeq((SCHIZO_ECCCTRL_EE |
+ SCHIZO_ECCCTRL_UE |
+ SCHIZO_ECCCTRL_CE), pbm->controller_regs + SCHIZO_ECC_CTRL);
+
+ err_mask = (SCHIZO_PCICTRL_BUS_UNUS |
+ SCHIZO_PCICTRL_ESLCK |
+ SCHIZO_PCICTRL_TTO_ERR |
+ SCHIZO_PCICTRL_RTRY_ERR |
+ SCHIZO_PCICTRL_SBH_ERR |
+ SCHIZO_PCICTRL_SERR |
+ SCHIZO_PCICTRL_EEN);
+
+ err_no_mask = (SCHIZO_PCICTRL_DTO_ERR |
+ SCHIZO_PCICTRL_SBH_INT);
+
+ /* Enable PCI Error interrupts and clear error
+ * bits for each PBM.
+ */
+ tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_CTRL);
+ tmp |= err_mask;
+ tmp &= ~err_no_mask;
+ upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_CTRL);
+
+ upa_writeq((SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_PTA |
+ SCHIZO_PCIAFSR_PRTRY | SCHIZO_PCIAFSR_PPERR |
+ SCHIZO_PCIAFSR_PTTO | SCHIZO_PCIAFSR_PUNUS |
+ SCHIZO_PCIAFSR_SMA | SCHIZO_PCIAFSR_STA |
+ SCHIZO_PCIAFSR_SRTRY | SCHIZO_PCIAFSR_SPERR |
+ SCHIZO_PCIAFSR_STTO | SCHIZO_PCIAFSR_SUNUS),
+ pbm->pbm_regs + SCHIZO_PCI_AFSR);
+
+ /* Make all Safari error conditions fatal except unmapped
+ * errors which we make generate interrupts.
+ */
+ err_mask = (BUS_ERROR_BADCMD | BUS_ERROR_SSMDIS |
+ BUS_ERROR_BADMA | BUS_ERROR_BADMB |
+ BUS_ERROR_BADMC |
+ BUS_ERROR_CPU1PS | BUS_ERROR_CPU1PB |
+ BUS_ERROR_CPU0PS | BUS_ERROR_CPU0PB |
+ BUS_ERROR_CIQTO |
+ BUS_ERROR_LPQTO | BUS_ERROR_SFPQTO |
+ BUS_ERROR_UFPQTO | BUS_ERROR_APERR |
+ BUS_ERROR_BUSERR | BUS_ERROR_TIMEOUT |
+ BUS_ERROR_ILL);
+#if 1
+ /* XXX Something wrong with some Excalibur systems
+ * XXX Sun is shipping. The behavior on a 2-cpu
+ * XXX machine is that both CPU1 parity error bits
+ * XXX are set and are immediately set again when
+ * XXX their error status bits are cleared. Just
+ * XXX ignore them for now. -DaveM
+ */
+ err_mask &= ~(BUS_ERROR_CPU1PS | BUS_ERROR_CPU1PB |
+ BUS_ERROR_CPU0PS | BUS_ERROR_CPU0PB);
+#endif
+
+ upa_writeq((SCHIZO_SAFERRCTRL_EN | err_mask),
+ pbm->controller_regs + SCHIZO_SAFARI_ERRCTRL);
+}
+
+static void pbm_config_busmastering(struct pci_pbm_info *pbm)
+{
+ u8 *addr;
+
+ /* Set cache-line size to 64 bytes, this is actually
+ * a nop but I do it for completeness.
+ */
+ addr = schizo_pci_config_mkaddr(pbm, pbm->pci_first_busno,
+ 0, PCI_CACHE_LINE_SIZE);
+ pci_config_write8(addr, 64 / sizeof(u32));
+
+ /* Set PBM latency timer to 64 PCI clocks. */
+ addr = schizo_pci_config_mkaddr(pbm, pbm->pci_first_busno,
+ 0, PCI_LATENCY_TIMER);
+ pci_config_write8(addr, 64);
+}
+
+static void schizo_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
+{
+ pbm_config_busmastering(pbm);
+ pbm->is_66mhz_capable =
+ (of_find_property(pbm->op->dev.of_node, "66mhz-capable", NULL)
+ != NULL);
+
+ pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
+
+ if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO)
+ tomatillo_register_error_handlers(pbm);
+ else
+ schizo_register_error_handlers(pbm);
+}
+
+#define SCHIZO_STRBUF_CONTROL (0x02800UL)
+#define SCHIZO_STRBUF_FLUSH (0x02808UL)
+#define SCHIZO_STRBUF_FSYNC (0x02810UL)
+#define SCHIZO_STRBUF_CTXFLUSH (0x02818UL)
+#define SCHIZO_STRBUF_CTXMATCH (0x10000UL)
+
+static void schizo_pbm_strbuf_init(struct pci_pbm_info *pbm)
+{
+ unsigned long base = pbm->pbm_regs;
+ u64 control;
+
+ if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO) {
+ /* TOMATILLO lacks streaming cache. */
+ return;
+ }
+
+ /* SCHIZO has context flushing. */
+ pbm->stc.strbuf_control = base + SCHIZO_STRBUF_CONTROL;
+ pbm->stc.strbuf_pflush = base + SCHIZO_STRBUF_FLUSH;
+ pbm->stc.strbuf_fsync = base + SCHIZO_STRBUF_FSYNC;
+ pbm->stc.strbuf_ctxflush = base + SCHIZO_STRBUF_CTXFLUSH;
+ pbm->stc.strbuf_ctxmatch_base = base + SCHIZO_STRBUF_CTXMATCH;
+
+ pbm->stc.strbuf_flushflag = (volatile unsigned long *)
+ ((((unsigned long)&pbm->stc.__flushflag_buf[0])
+ + 63UL)
+ & ~63UL);
+ pbm->stc.strbuf_flushflag_pa = (unsigned long)
+ __pa(pbm->stc.strbuf_flushflag);
+
+ /* Turn off LRU locking and diag mode, enable the
+ * streaming buffer and leave the rerun-disable
+ * setting however OBP set it.
+ */
+ control = upa_readq(pbm->stc.strbuf_control);
+ control &= ~(SCHIZO_STRBUF_CTRL_LPTR |
+ SCHIZO_STRBUF_CTRL_LENAB |
+ SCHIZO_STRBUF_CTRL_DENAB);
+ control |= SCHIZO_STRBUF_CTRL_ENAB;
+ upa_writeq(control, pbm->stc.strbuf_control);
+
+ pbm->stc.strbuf_enabled = 1;
+}
+
+#define SCHIZO_IOMMU_CONTROL (0x00200UL)
+#define SCHIZO_IOMMU_TSBBASE (0x00208UL)
+#define SCHIZO_IOMMU_FLUSH (0x00210UL)
+#define SCHIZO_IOMMU_CTXFLUSH (0x00218UL)
+
+static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm)
+{
+ static const u32 vdma_default[] = { 0xc0000000, 0x40000000 };
+ unsigned long i, tagbase, database;
+ struct iommu *iommu = pbm->iommu;
+ int tsbsize, err;
+ const u32 *vdma;
+ u32 dma_mask;
+ u64 control;
+
+ vdma = of_get_property(pbm->op->dev.of_node, "virtual-dma", NULL);
+ if (!vdma)
+ vdma = vdma_default;
+
+ dma_mask = vdma[0];
+ switch (vdma[1]) {
+ case 0x20000000:
+ dma_mask |= 0x1fffffff;
+ tsbsize = 64;
+ break;
+
+ case 0x40000000:
+ dma_mask |= 0x3fffffff;
+ tsbsize = 128;
+ break;
+
+ case 0x80000000:
+ dma_mask |= 0x7fffffff;
+ tsbsize = 128;
+ break;
+
+ default:
+ printk(KERN_ERR PFX "Strange virtual-dma size.\n");
+ return -EINVAL;
+ }
+
+ /* Register addresses, SCHIZO has iommu ctx flushing. */
+ iommu->iommu_control = pbm->pbm_regs + SCHIZO_IOMMU_CONTROL;
+ iommu->iommu_tsbbase = pbm->pbm_regs + SCHIZO_IOMMU_TSBBASE;
+ iommu->iommu_flush = pbm->pbm_regs + SCHIZO_IOMMU_FLUSH;
+ iommu->iommu_tags = iommu->iommu_flush + (0xa580UL - 0x0210UL);
+ iommu->iommu_ctxflush = pbm->pbm_regs + SCHIZO_IOMMU_CTXFLUSH;
+
+ /* We use the main control/status register of SCHIZO as the write
+ * completion register.
+ */
+ iommu->write_complete_reg = pbm->controller_regs + 0x10000UL;
+
+ /*
+ * Invalidate TLB Entries.
+ */
+ control = upa_readq(iommu->iommu_control);
+ control |= SCHIZO_IOMMU_CTRL_DENAB;
+ upa_writeq(control, iommu->iommu_control);
+
+ tagbase = SCHIZO_IOMMU_TAG, database = SCHIZO_IOMMU_DATA;
+
+ for (i = 0; i < 16; i++) {
+ upa_writeq(0, pbm->pbm_regs + tagbase + (i * 8UL));
+ upa_writeq(0, pbm->pbm_regs + database + (i * 8UL));
+ }
+
+ /* Leave diag mode enabled for full-flushing done
+ * in pci_iommu.c
+ */
+ err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask,
+ pbm->numa_node);
+ if (err) {
+ printk(KERN_ERR PFX "iommu_table_init() fails with %d\n", err);
+ return err;
+ }
+
+ upa_writeq(__pa(iommu->page_table), iommu->iommu_tsbbase);
+
+ control = upa_readq(iommu->iommu_control);
+ control &= ~(SCHIZO_IOMMU_CTRL_TSBSZ | SCHIZO_IOMMU_CTRL_TBWSZ);
+ switch (tsbsize) {
+ case 64:
+ control |= SCHIZO_IOMMU_TSBSZ_64K;
+ break;
+ case 128:
+ control |= SCHIZO_IOMMU_TSBSZ_128K;
+ break;
+ }
+
+ control |= SCHIZO_IOMMU_CTRL_ENAB;
+ upa_writeq(control, iommu->iommu_control);
+
+ return 0;
+}
+
+#define SCHIZO_PCI_IRQ_RETRY (0x1a00UL)
+#define SCHIZO_IRQ_RETRY_INF 0xffUL
+
+#define SCHIZO_PCI_DIAG (0x2020UL)
+#define SCHIZO_PCIDIAG_D_BADECC (1UL << 10UL) /* Disable BAD ECC errors (Schizo) */
+#define SCHIZO_PCIDIAG_D_BYPASS (1UL << 9UL) /* Disable MMU bypass mode (Schizo/Tomatillo) */
+#define SCHIZO_PCIDIAG_D_TTO (1UL << 8UL) /* Disable TTO errors (Schizo/Tomatillo) */
+#define SCHIZO_PCIDIAG_D_RTRYARB (1UL << 7UL) /* Disable retry arbitration (Schizo) */
+#define SCHIZO_PCIDIAG_D_RETRY (1UL << 6UL) /* Disable retry limit (Schizo/Tomatillo) */
+#define SCHIZO_PCIDIAG_D_INTSYNC (1UL << 5UL) /* Disable interrupt/DMA synch (Schizo/Tomatillo) */
+#define SCHIZO_PCIDIAG_I_DMA_PARITY (1UL << 3UL) /* Invert DMA parity (Schizo/Tomatillo) */
+#define SCHIZO_PCIDIAG_I_PIOD_PARITY (1UL << 2UL) /* Invert PIO data parity (Schizo/Tomatillo) */
+#define SCHIZO_PCIDIAG_I_PIOA_PARITY (1UL << 1UL) /* Invert PIO address parity (Schizo/Tomatillo) */
+
+#define TOMATILLO_PCI_IOC_CSR (0x2248UL)
+#define TOMATILLO_IOC_PART_WPENAB 0x0000000000080000UL
+#define TOMATILLO_IOC_RDMULT_PENAB 0x0000000000040000UL
+#define TOMATILLO_IOC_RDONE_PENAB 0x0000000000020000UL
+#define TOMATILLO_IOC_RDLINE_PENAB 0x0000000000010000UL
+#define TOMATILLO_IOC_RDMULT_PLEN 0x000000000000c000UL
+#define TOMATILLO_IOC_RDMULT_PLEN_SHIFT 14UL
+#define TOMATILLO_IOC_RDONE_PLEN 0x0000000000003000UL
+#define TOMATILLO_IOC_RDONE_PLEN_SHIFT 12UL
+#define TOMATILLO_IOC_RDLINE_PLEN 0x0000000000000c00UL
+#define TOMATILLO_IOC_RDLINE_PLEN_SHIFT 10UL
+#define TOMATILLO_IOC_PREF_OFF 0x00000000000003f8UL
+#define TOMATILLO_IOC_PREF_OFF_SHIFT 3UL
+#define TOMATILLO_IOC_RDMULT_CPENAB 0x0000000000000004UL
+#define TOMATILLO_IOC_RDONE_CPENAB 0x0000000000000002UL
+#define TOMATILLO_IOC_RDLINE_CPENAB 0x0000000000000001UL
+
+#define TOMATILLO_PCI_IOC_TDIAG (0x2250UL)
+#define TOMATILLO_PCI_IOC_DDIAG (0x2290UL)
+
+static void schizo_pbm_hw_init(struct pci_pbm_info *pbm)
+{
+ u64 tmp;
+
+ upa_writeq(5, pbm->pbm_regs + SCHIZO_PCI_IRQ_RETRY);
+
+ tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_CTRL);
+
+ /* Enable arbiter for all PCI slots. */
+ tmp |= 0xff;
+
+ if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO &&
+ pbm->chip_version >= 0x2)
+ tmp |= 0x3UL << SCHIZO_PCICTRL_PTO_SHIFT;
+
+ if (!of_find_property(pbm->op->dev.of_node, "no-bus-parking", NULL))
+ tmp |= SCHIZO_PCICTRL_PARK;
+ else
+ tmp &= ~SCHIZO_PCICTRL_PARK;
+
+ if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO &&
+ pbm->chip_version <= 0x1)
+ tmp |= SCHIZO_PCICTRL_DTO_INT;
+ else
+ tmp &= ~SCHIZO_PCICTRL_DTO_INT;
+
+ if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO)
+ tmp |= (SCHIZO_PCICTRL_MRM_PREF |
+ SCHIZO_PCICTRL_RDO_PREF |
+ SCHIZO_PCICTRL_RDL_PREF);
+
+ upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_CTRL);
+
+ tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_DIAG);
+ tmp &= ~(SCHIZO_PCIDIAG_D_RTRYARB |
+ SCHIZO_PCIDIAG_D_RETRY |
+ SCHIZO_PCIDIAG_D_INTSYNC);
+ upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_DIAG);
+
+ if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO) {
+ /* Clear prefetch lengths to workaround a bug in
+ * Jalapeno...
+ */
+ tmp = (TOMATILLO_IOC_PART_WPENAB |
+ (1 << TOMATILLO_IOC_PREF_OFF_SHIFT) |
+ TOMATILLO_IOC_RDMULT_CPENAB |
+ TOMATILLO_IOC_RDONE_CPENAB |
+ TOMATILLO_IOC_RDLINE_CPENAB);
+
+ upa_writeq(tmp, pbm->pbm_regs + TOMATILLO_PCI_IOC_CSR);
+ }
+}
+
+static int schizo_pbm_init(struct pci_pbm_info *pbm,
+ struct platform_device *op, u32 portid,
+ int chip_type)
+{
+ const struct linux_prom64_registers *regs;
+ struct device_node *dp = op->dev.of_node;
+ const char *chipset_name;
+ int err;
+
+ switch (chip_type) {
+ case PBM_CHIP_TYPE_TOMATILLO:
+ chipset_name = "TOMATILLO";
+ break;
+
+ case PBM_CHIP_TYPE_SCHIZO_PLUS:
+ chipset_name = "SCHIZO+";
+ break;
+
+ case PBM_CHIP_TYPE_SCHIZO:
+ default:
+ chipset_name = "SCHIZO";
+ break;
+ }
+
+ /* For SCHIZO, three OBP regs:
+ * 1) PBM controller regs
+ * 2) Schizo front-end controller regs (same for both PBMs)
+ * 3) PBM PCI config space
+ *
+ * For TOMATILLO, four OBP regs:
+ * 1) PBM controller regs
+ * 2) Tomatillo front-end controller regs
+ * 3) PBM PCI config space
+ * 4) Ichip regs
+ */
+ regs = of_get_property(dp, "reg", NULL);
+
+ pbm->next = pci_pbm_root;
+ pci_pbm_root = pbm;
+
+ pbm->numa_node = -1;
+
+ pbm->pci_ops = &sun4u_pci_ops;
+ pbm->config_space_reg_bits = 8;
+
+ pbm->index = pci_num_pbms++;
+
+ pbm->portid = portid;
+ pbm->op = op;
+
+ pbm->chip_type = chip_type;
+ pbm->chip_version = of_getintprop_default(dp, "version#", 0);
+ pbm->chip_revision = of_getintprop_default(dp, "module-version#", 0);
+
+ pbm->pbm_regs = regs[0].phys_addr;
+ pbm->controller_regs = regs[1].phys_addr - 0x10000UL;
+
+ if (chip_type == PBM_CHIP_TYPE_TOMATILLO)
+ pbm->sync_reg = regs[3].phys_addr + 0x1a18UL;
+
+ pbm->name = dp->full_name;
+
+ printk("%s: %s PCI Bus Module ver[%x:%x]\n",
+ pbm->name, chipset_name,
+ pbm->chip_version, pbm->chip_revision);
+
+ schizo_pbm_hw_init(pbm);
+
+ pci_determine_mem_io_space(pbm);
+
+ pci_get_pbm_props(pbm);
+
+ err = schizo_pbm_iommu_init(pbm);
+ if (err)
+ return err;
+
+ schizo_pbm_strbuf_init(pbm);
+
+ schizo_scan_bus(pbm, &op->dev);
+
+ return 0;
+}
+
+static inline int portid_compare(u32 x, u32 y, int chip_type)
+{
+ if (chip_type == PBM_CHIP_TYPE_TOMATILLO) {
+ if (x == (y ^ 1))
+ return 1;
+ return 0;
+ }
+ return (x == y);
+}
+
+static struct pci_pbm_info *schizo_find_sibling(u32 portid, int chip_type)
+{
+ struct pci_pbm_info *pbm;
+
+ for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
+ if (portid_compare(pbm->portid, portid, chip_type))
+ return pbm;
+ }
+ return NULL;
+}
+
+static int __schizo_init(struct platform_device *op, unsigned long chip_type)
+{
+ struct device_node *dp = op->dev.of_node;
+ struct pci_pbm_info *pbm;
+ struct iommu *iommu;
+ u32 portid;
+ int err;
+
+ portid = of_getintprop_default(dp, "portid", 0xff);
+
+ err = -ENOMEM;
+ pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+ if (!pbm) {
+ printk(KERN_ERR PFX "Cannot allocate pci_pbm_info.\n");
+ goto out_err;
+ }
+
+ pbm->sibling = schizo_find_sibling(portid, chip_type);
+
+ iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+ if (!iommu) {
+ printk(KERN_ERR PFX "Cannot allocate PBM A iommu.\n");
+ goto out_free_pbm;
+ }
+
+ pbm->iommu = iommu;
+
+ if (schizo_pbm_init(pbm, op, portid, chip_type))
+ goto out_free_iommu;
+
+ if (pbm->sibling)
+ pbm->sibling->sibling = pbm;
+
+ dev_set_drvdata(&op->dev, pbm);
+
+ return 0;
+
+out_free_iommu:
+ kfree(pbm->iommu);
+
+out_free_pbm:
+ kfree(pbm);
+
+out_err:
+ return err;
+}
+
+static const struct of_device_id schizo_match[];
+static int schizo_probe(struct platform_device *op)
+{
+ const struct of_device_id *match;
+
+ match = of_match_device(schizo_match, &op->dev);
+ if (!match)
+ return -EINVAL;
+ return __schizo_init(op, (unsigned long)match->data);
+}
+
+/* The ordering of this table is very important. Some Tomatillo
+ * nodes announce that they are compatible with both pci108e,a801
+ * and pci108e,8001. So list the chips in reverse chronological
+ * order.
+ */
+static const struct of_device_id schizo_match[] = {
+ {
+ .name = "pci",
+ .compatible = "pci108e,a801",
+ .data = (void *) PBM_CHIP_TYPE_TOMATILLO,
+ },
+ {
+ .name = "pci",
+ .compatible = "pci108e,8002",
+ .data = (void *) PBM_CHIP_TYPE_SCHIZO_PLUS,
+ },
+ {
+ .name = "pci",
+ .compatible = "pci108e,8001",
+ .data = (void *) PBM_CHIP_TYPE_SCHIZO,
+ },
+ {},
+};
+
+static struct platform_driver schizo_driver = {
+ .driver = {
+ .name = DRIVER_NAME,
+ .owner = THIS_MODULE,
+ .of_match_table = schizo_match,
+ },
+ .probe = schizo_probe,
+};
+
+static int __init schizo_init(void)
+{
+ return platform_driver_register(&schizo_driver);
+}
+
+subsys_initcall(schizo_init);
diff --git a/arch/sparc/kernel/pci_sun4v.c b/arch/sparc/kernel/pci_sun4v.c
new file mode 100644
index 00000000000..d07f6b29aed
--- /dev/null
+++ b/arch/sparc/kernel/pci_sun4v.c
@@ -0,0 +1,1024 @@
+/* pci_sun4v.c: SUN4V specific PCI controller support.
+ *
+ * Copyright (C) 2006, 2007, 2008 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/percpu.h>
+#include <linux/irq.h>
+#include <linux/msi.h>
+#include <linux/export.h>
+#include <linux/log2.h>
+#include <linux/of_device.h>
+
+#include <asm/iommu.h>
+#include <asm/irq.h>
+#include <asm/hypervisor.h>
+#include <asm/prom.h>
+
+#include "pci_impl.h"
+#include "iommu_common.h"
+
+#include "pci_sun4v.h"
+
+#define DRIVER_NAME "pci_sun4v"
+#define PFX DRIVER_NAME ": "
+
+static unsigned long vpci_major = 1;
+static unsigned long vpci_minor = 1;
+
+#define PGLIST_NENTS (PAGE_SIZE / sizeof(u64))
+
+struct iommu_batch {
+ struct device *dev; /* Device mapping is for. */
+ unsigned long prot; /* IOMMU page protections */
+ unsigned long entry; /* Index into IOTSB. */
+ u64 *pglist; /* List of physical pages */
+ unsigned long npages; /* Number of pages in list. */
+};
+
+static DEFINE_PER_CPU(struct iommu_batch, iommu_batch);
+static int iommu_batch_initialized;
+
+/* Interrupts must be disabled. */
+static inline void iommu_batch_start(struct device *dev, unsigned long prot, unsigned long entry)
+{
+ struct iommu_batch *p = &__get_cpu_var(iommu_batch);
+
+ p->dev = dev;
+ p->prot = prot;
+ p->entry = entry;
+ p->npages = 0;
+}
+
+/* Interrupts must be disabled. */
+static long iommu_batch_flush(struct iommu_batch *p)
+{
+ struct pci_pbm_info *pbm = p->dev->archdata.host_controller;
+ unsigned long devhandle = pbm->devhandle;
+ unsigned long prot = p->prot;
+ unsigned long entry = p->entry;
+ u64 *pglist = p->pglist;
+ unsigned long npages = p->npages;
+
+ while (npages != 0) {
+ long num;
+
+ num = pci_sun4v_iommu_map(devhandle, HV_PCI_TSBID(0, entry),
+ npages, prot, __pa(pglist));
+ if (unlikely(num < 0)) {
+ if (printk_ratelimit())
+ printk("iommu_batch_flush: IOMMU map of "
+ "[%08lx:%08llx:%lx:%lx:%lx] failed with "
+ "status %ld\n",
+ devhandle, HV_PCI_TSBID(0, entry),
+ npages, prot, __pa(pglist), num);
+ return -1;
+ }
+
+ entry += num;
+ npages -= num;
+ pglist += num;
+ }
+
+ p->entry = entry;
+ p->npages = 0;
+
+ return 0;
+}
+
+static inline void iommu_batch_new_entry(unsigned long entry)
+{
+ struct iommu_batch *p = &__get_cpu_var(iommu_batch);
+
+ if (p->entry + p->npages == entry)
+ return;
+ if (p->entry != ~0UL)
+ iommu_batch_flush(p);
+ p->entry = entry;
+}
+
+/* Interrupts must be disabled. */
+static inline long iommu_batch_add(u64 phys_page)
+{
+ struct iommu_batch *p = &__get_cpu_var(iommu_batch);
+
+ BUG_ON(p->npages >= PGLIST_NENTS);
+
+ p->pglist[p->npages++] = phys_page;
+ if (p->npages == PGLIST_NENTS)
+ return iommu_batch_flush(p);
+
+ return 0;
+}
+
+/* Interrupts must be disabled. */
+static inline long iommu_batch_end(void)
+{
+ struct iommu_batch *p = &__get_cpu_var(iommu_batch);
+
+ BUG_ON(p->npages >= PGLIST_NENTS);
+
+ return iommu_batch_flush(p);
+}
+
+static void *dma_4v_alloc_coherent(struct device *dev, size_t size,
+ dma_addr_t *dma_addrp, gfp_t gfp,
+ struct dma_attrs *attrs)
+{
+ unsigned long flags, order, first_page, npages, n;
+ struct iommu *iommu;
+ struct page *page;
+ void *ret;
+ long entry;
+ int nid;
+
+ size = IO_PAGE_ALIGN(size);
+ order = get_order(size);
+ if (unlikely(order >= MAX_ORDER))
+ return NULL;
+
+ npages = size >> IO_PAGE_SHIFT;
+
+ nid = dev->archdata.numa_node;
+ page = alloc_pages_node(nid, gfp, order);
+ if (unlikely(!page))
+ return NULL;
+
+ first_page = (unsigned long) page_address(page);
+ memset((char *)first_page, 0, PAGE_SIZE << order);
+
+ iommu = dev->archdata.iommu;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ entry = iommu_range_alloc(dev, iommu, npages, NULL);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (unlikely(entry == DMA_ERROR_CODE))
+ goto range_alloc_fail;
+
+ *dma_addrp = (iommu->page_table_map_base +
+ (entry << IO_PAGE_SHIFT));
+ ret = (void *) first_page;
+ first_page = __pa(first_page);
+
+ local_irq_save(flags);
+
+ iommu_batch_start(dev,
+ (HV_PCI_MAP_ATTR_READ |
+ HV_PCI_MAP_ATTR_WRITE),
+ entry);
+
+ for (n = 0; n < npages; n++) {
+ long err = iommu_batch_add(first_page + (n * PAGE_SIZE));
+ if (unlikely(err < 0L))
+ goto iommu_map_fail;
+ }
+
+ if (unlikely(iommu_batch_end() < 0L))
+ goto iommu_map_fail;
+
+ local_irq_restore(flags);
+
+ return ret;
+
+iommu_map_fail:
+ /* Interrupts are disabled. */
+ spin_lock(&iommu->lock);
+ iommu_range_free(iommu, *dma_addrp, npages);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+range_alloc_fail:
+ free_pages(first_page, order);
+ return NULL;
+}
+
+static void dma_4v_free_coherent(struct device *dev, size_t size, void *cpu,
+ dma_addr_t dvma, struct dma_attrs *attrs)
+{
+ struct pci_pbm_info *pbm;
+ struct iommu *iommu;
+ unsigned long flags, order, npages, entry;
+ u32 devhandle;
+
+ npages = IO_PAGE_ALIGN(size) >> IO_PAGE_SHIFT;
+ iommu = dev->archdata.iommu;
+ pbm = dev->archdata.host_controller;
+ devhandle = pbm->devhandle;
+ entry = ((dvma - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ iommu_range_free(iommu, dvma, npages);
+
+ do {
+ unsigned long num;
+
+ num = pci_sun4v_iommu_demap(devhandle, HV_PCI_TSBID(0, entry),
+ npages);
+ entry += num;
+ npages -= num;
+ } while (npages != 0);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ order = get_order(size);
+ if (order < 10)
+ free_pages((unsigned long)cpu, order);
+}
+
+static dma_addr_t dma_4v_map_page(struct device *dev, struct page *page,
+ unsigned long offset, size_t sz,
+ enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ struct iommu *iommu;
+ unsigned long flags, npages, oaddr;
+ unsigned long i, base_paddr;
+ u32 bus_addr, ret;
+ unsigned long prot;
+ long entry;
+
+ iommu = dev->archdata.iommu;
+
+ if (unlikely(direction == DMA_NONE))
+ goto bad;
+
+ oaddr = (unsigned long)(page_address(page) + offset);
+ npages = IO_PAGE_ALIGN(oaddr + sz) - (oaddr & IO_PAGE_MASK);
+ npages >>= IO_PAGE_SHIFT;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ entry = iommu_range_alloc(dev, iommu, npages, NULL);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (unlikely(entry == DMA_ERROR_CODE))
+ goto bad;
+
+ bus_addr = (iommu->page_table_map_base +
+ (entry << IO_PAGE_SHIFT));
+ ret = bus_addr | (oaddr & ~IO_PAGE_MASK);
+ base_paddr = __pa(oaddr & IO_PAGE_MASK);
+ prot = HV_PCI_MAP_ATTR_READ;
+ if (direction != DMA_TO_DEVICE)
+ prot |= HV_PCI_MAP_ATTR_WRITE;
+
+ local_irq_save(flags);
+
+ iommu_batch_start(dev, prot, entry);
+
+ for (i = 0; i < npages; i++, base_paddr += IO_PAGE_SIZE) {
+ long err = iommu_batch_add(base_paddr);
+ if (unlikely(err < 0L))
+ goto iommu_map_fail;
+ }
+ if (unlikely(iommu_batch_end() < 0L))
+ goto iommu_map_fail;
+
+ local_irq_restore(flags);
+
+ return ret;
+
+bad:
+ if (printk_ratelimit())
+ WARN_ON(1);
+ return DMA_ERROR_CODE;
+
+iommu_map_fail:
+ /* Interrupts are disabled. */
+ spin_lock(&iommu->lock);
+ iommu_range_free(iommu, bus_addr, npages);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ return DMA_ERROR_CODE;
+}
+
+static void dma_4v_unmap_page(struct device *dev, dma_addr_t bus_addr,
+ size_t sz, enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ struct pci_pbm_info *pbm;
+ struct iommu *iommu;
+ unsigned long flags, npages;
+ long entry;
+ u32 devhandle;
+
+ if (unlikely(direction == DMA_NONE)) {
+ if (printk_ratelimit())
+ WARN_ON(1);
+ return;
+ }
+
+ iommu = dev->archdata.iommu;
+ pbm = dev->archdata.host_controller;
+ devhandle = pbm->devhandle;
+
+ npages = IO_PAGE_ALIGN(bus_addr + sz) - (bus_addr & IO_PAGE_MASK);
+ npages >>= IO_PAGE_SHIFT;
+ bus_addr &= IO_PAGE_MASK;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ iommu_range_free(iommu, bus_addr, npages);
+
+ entry = (bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT;
+ do {
+ unsigned long num;
+
+ num = pci_sun4v_iommu_demap(devhandle, HV_PCI_TSBID(0, entry),
+ npages);
+ entry += num;
+ npages -= num;
+ } while (npages != 0);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
+ int nelems, enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ struct scatterlist *s, *outs, *segstart;
+ unsigned long flags, handle, prot;
+ dma_addr_t dma_next = 0, dma_addr;
+ unsigned int max_seg_size;
+ unsigned long seg_boundary_size;
+ int outcount, incount, i;
+ struct iommu *iommu;
+ unsigned long base_shift;
+ long err;
+
+ BUG_ON(direction == DMA_NONE);
+
+ iommu = dev->archdata.iommu;
+ if (nelems == 0 || !iommu)
+ return 0;
+
+ prot = HV_PCI_MAP_ATTR_READ;
+ if (direction != DMA_TO_DEVICE)
+ prot |= HV_PCI_MAP_ATTR_WRITE;
+
+ outs = s = segstart = &sglist[0];
+ outcount = 1;
+ incount = nelems;
+ handle = 0;
+
+ /* Init first segment length for backout at failure */
+ outs->dma_length = 0;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ iommu_batch_start(dev, prot, ~0UL);
+
+ max_seg_size = dma_get_max_seg_size(dev);
+ seg_boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
+ IO_PAGE_SIZE) >> IO_PAGE_SHIFT;
+ base_shift = iommu->page_table_map_base >> IO_PAGE_SHIFT;
+ for_each_sg(sglist, s, nelems, i) {
+ unsigned long paddr, npages, entry, out_entry = 0, slen;
+
+ slen = s->length;
+ /* Sanity check */
+ if (slen == 0) {
+ dma_next = 0;
+ continue;
+ }
+ /* Allocate iommu entries for that segment */
+ paddr = (unsigned long) SG_ENT_PHYS_ADDRESS(s);
+ npages = iommu_num_pages(paddr, slen, IO_PAGE_SIZE);
+ entry = iommu_range_alloc(dev, iommu, npages, &handle);
+
+ /* Handle failure */
+ if (unlikely(entry == DMA_ERROR_CODE)) {
+ if (printk_ratelimit())
+ printk(KERN_INFO "iommu_alloc failed, iommu %p paddr %lx"
+ " npages %lx\n", iommu, paddr, npages);
+ goto iommu_map_failed;
+ }
+
+ iommu_batch_new_entry(entry);
+
+ /* Convert entry to a dma_addr_t */
+ dma_addr = iommu->page_table_map_base +
+ (entry << IO_PAGE_SHIFT);
+ dma_addr |= (s->offset & ~IO_PAGE_MASK);
+
+ /* Insert into HW table */
+ paddr &= IO_PAGE_MASK;
+ while (npages--) {
+ err = iommu_batch_add(paddr);
+ if (unlikely(err < 0L))
+ goto iommu_map_failed;
+ paddr += IO_PAGE_SIZE;
+ }
+
+ /* If we are in an open segment, try merging */
+ if (segstart != s) {
+ /* We cannot merge if:
+ * - allocated dma_addr isn't contiguous to previous allocation
+ */
+ if ((dma_addr != dma_next) ||
+ (outs->dma_length + s->length > max_seg_size) ||
+ (is_span_boundary(out_entry, base_shift,
+ seg_boundary_size, outs, s))) {
+ /* Can't merge: create a new segment */
+ segstart = s;
+ outcount++;
+ outs = sg_next(outs);
+ } else {
+ outs->dma_length += s->length;
+ }
+ }
+
+ if (segstart == s) {
+ /* This is a new segment, fill entries */
+ outs->dma_address = dma_addr;
+ outs->dma_length = slen;
+ out_entry = entry;
+ }
+
+ /* Calculate next page pointer for contiguous check */
+ dma_next = dma_addr + slen;
+ }
+
+ err = iommu_batch_end();
+
+ if (unlikely(err < 0L))
+ goto iommu_map_failed;
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if (outcount < incount) {
+ outs = sg_next(outs);
+ outs->dma_address = DMA_ERROR_CODE;
+ outs->dma_length = 0;
+ }
+
+ return outcount;
+
+iommu_map_failed:
+ for_each_sg(sglist, s, nelems, i) {
+ if (s->dma_length != 0) {
+ unsigned long vaddr, npages;
+
+ vaddr = s->dma_address & IO_PAGE_MASK;
+ npages = iommu_num_pages(s->dma_address, s->dma_length,
+ IO_PAGE_SIZE);
+ iommu_range_free(iommu, vaddr, npages);
+ /* XXX demap? XXX */
+ s->dma_address = DMA_ERROR_CODE;
+ s->dma_length = 0;
+ }
+ if (s == outs)
+ break;
+ }
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ return 0;
+}
+
+static void dma_4v_unmap_sg(struct device *dev, struct scatterlist *sglist,
+ int nelems, enum dma_data_direction direction,
+ struct dma_attrs *attrs)
+{
+ struct pci_pbm_info *pbm;
+ struct scatterlist *sg;
+ struct iommu *iommu;
+ unsigned long flags;
+ u32 devhandle;
+
+ BUG_ON(direction == DMA_NONE);
+
+ iommu = dev->archdata.iommu;
+ pbm = dev->archdata.host_controller;
+ devhandle = pbm->devhandle;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ sg = sglist;
+ while (nelems--) {
+ dma_addr_t dma_handle = sg->dma_address;
+ unsigned int len = sg->dma_length;
+ unsigned long npages, entry;
+
+ if (!len)
+ break;
+ npages = iommu_num_pages(dma_handle, len, IO_PAGE_SIZE);
+ iommu_range_free(iommu, dma_handle, npages);
+
+ entry = ((dma_handle - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
+ while (npages) {
+ unsigned long num;
+
+ num = pci_sun4v_iommu_demap(devhandle, HV_PCI_TSBID(0, entry),
+ npages);
+ entry += num;
+ npages -= num;
+ }
+
+ sg = sg_next(sg);
+ }
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static struct dma_map_ops sun4v_dma_ops = {
+ .alloc = dma_4v_alloc_coherent,
+ .free = dma_4v_free_coherent,
+ .map_page = dma_4v_map_page,
+ .unmap_page = dma_4v_unmap_page,
+ .map_sg = dma_4v_map_sg,
+ .unmap_sg = dma_4v_unmap_sg,
+};
+
+static void pci_sun4v_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
+{
+ struct property *prop;
+ struct device_node *dp;
+
+ dp = pbm->op->dev.of_node;
+ prop = of_find_property(dp, "66mhz-capable", NULL);
+ pbm->is_66mhz_capable = (prop != NULL);
+ pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
+
+ /* XXX register error interrupt handlers XXX */
+}
+
+static unsigned long probe_existing_entries(struct pci_pbm_info *pbm,
+ struct iommu *iommu)
+{
+ struct iommu_arena *arena = &iommu->arena;
+ unsigned long i, cnt = 0;
+ u32 devhandle;
+
+ devhandle = pbm->devhandle;
+ for (i = 0; i < arena->limit; i++) {
+ unsigned long ret, io_attrs, ra;
+
+ ret = pci_sun4v_iommu_getmap(devhandle,
+ HV_PCI_TSBID(0, i),
+ &io_attrs, &ra);
+ if (ret == HV_EOK) {
+ if (page_in_phys_avail(ra)) {
+ pci_sun4v_iommu_demap(devhandle,
+ HV_PCI_TSBID(0, i), 1);
+ } else {
+ cnt++;
+ __set_bit(i, arena->map);
+ }
+ }
+ }
+
+ return cnt;
+}
+
+static int pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
+{
+ static const u32 vdma_default[] = { 0x80000000, 0x80000000 };
+ struct iommu *iommu = pbm->iommu;
+ unsigned long num_tsb_entries, sz;
+ u32 dma_mask, dma_offset;
+ const u32 *vdma;
+
+ vdma = of_get_property(pbm->op->dev.of_node, "virtual-dma", NULL);
+ if (!vdma)
+ vdma = vdma_default;
+
+ if ((vdma[0] | vdma[1]) & ~IO_PAGE_MASK) {
+ printk(KERN_ERR PFX "Strange virtual-dma[%08x:%08x].\n",
+ vdma[0], vdma[1]);
+ return -EINVAL;
+ }
+
+ dma_mask = (roundup_pow_of_two(vdma[1]) - 1UL);
+ num_tsb_entries = vdma[1] / IO_PAGE_SIZE;
+
+ dma_offset = vdma[0];
+
+ /* Setup initial software IOMMU state. */
+ spin_lock_init(&iommu->lock);
+ iommu->ctx_lowest_free = 1;
+ iommu->page_table_map_base = dma_offset;
+ iommu->dma_addr_mask = dma_mask;
+
+ /* Allocate and initialize the free area map. */
+ sz = (num_tsb_entries + 7) / 8;
+ sz = (sz + 7UL) & ~7UL;
+ iommu->arena.map = kzalloc(sz, GFP_KERNEL);
+ if (!iommu->arena.map) {
+ printk(KERN_ERR PFX "Error, kmalloc(arena.map) failed.\n");
+ return -ENOMEM;
+ }
+ iommu->arena.limit = num_tsb_entries;
+
+ sz = probe_existing_entries(pbm, iommu);
+ if (sz)
+ printk("%s: Imported %lu TSB entries from OBP\n",
+ pbm->name, sz);
+
+ return 0;
+}
+
+#ifdef CONFIG_PCI_MSI
+struct pci_sun4v_msiq_entry {
+ u64 version_type;
+#define MSIQ_VERSION_MASK 0xffffffff00000000UL
+#define MSIQ_VERSION_SHIFT 32
+#define MSIQ_TYPE_MASK 0x00000000000000ffUL
+#define MSIQ_TYPE_SHIFT 0
+#define MSIQ_TYPE_NONE 0x00
+#define MSIQ_TYPE_MSG 0x01
+#define MSIQ_TYPE_MSI32 0x02
+#define MSIQ_TYPE_MSI64 0x03
+#define MSIQ_TYPE_INTX 0x08
+#define MSIQ_TYPE_NONE2 0xff
+
+ u64 intx_sysino;
+ u64 reserved1;
+ u64 stick;
+ u64 req_id; /* bus/device/func */
+#define MSIQ_REQID_BUS_MASK 0xff00UL
+#define MSIQ_REQID_BUS_SHIFT 8
+#define MSIQ_REQID_DEVICE_MASK 0x00f8UL
+#define MSIQ_REQID_DEVICE_SHIFT 3
+#define MSIQ_REQID_FUNC_MASK 0x0007UL
+#define MSIQ_REQID_FUNC_SHIFT 0
+
+ u64 msi_address;
+
+ /* The format of this value is message type dependent.
+ * For MSI bits 15:0 are the data from the MSI packet.
+ * For MSI-X bits 31:0 are the data from the MSI packet.
+ * For MSG, the message code and message routing code where:
+ * bits 39:32 is the bus/device/fn of the msg target-id
+ * bits 18:16 is the message routing code
+ * bits 7:0 is the message code
+ * For INTx the low order 2-bits are:
+ * 00 - INTA
+ * 01 - INTB
+ * 10 - INTC
+ * 11 - INTD
+ */
+ u64 msi_data;
+
+ u64 reserved2;
+};
+
+static int pci_sun4v_get_head(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long *head)
+{
+ unsigned long err, limit;
+
+ err = pci_sun4v_msiq_gethead(pbm->devhandle, msiqid, head);
+ if (unlikely(err))
+ return -ENXIO;
+
+ limit = pbm->msiq_ent_count * sizeof(struct pci_sun4v_msiq_entry);
+ if (unlikely(*head >= limit))
+ return -EFBIG;
+
+ return 0;
+}
+
+static int pci_sun4v_dequeue_msi(struct pci_pbm_info *pbm,
+ unsigned long msiqid, unsigned long *head,
+ unsigned long *msi)
+{
+ struct pci_sun4v_msiq_entry *ep;
+ unsigned long err, type;
+
+ /* Note: void pointer arithmetic, 'head' is a byte offset */
+ ep = (pbm->msi_queues + ((msiqid - pbm->msiq_first) *
+ (pbm->msiq_ent_count *
+ sizeof(struct pci_sun4v_msiq_entry))) +
+ *head);
+
+ if ((ep->version_type & MSIQ_TYPE_MASK) == 0)
+ return 0;
+
+ type = (ep->version_type & MSIQ_TYPE_MASK) >> MSIQ_TYPE_SHIFT;
+ if (unlikely(type != MSIQ_TYPE_MSI32 &&
+ type != MSIQ_TYPE_MSI64))
+ return -EINVAL;
+
+ *msi = ep->msi_data;
+
+ err = pci_sun4v_msi_setstate(pbm->devhandle,
+ ep->msi_data /* msi_num */,
+ HV_MSISTATE_IDLE);
+ if (unlikely(err))
+ return -ENXIO;
+
+ /* Clear the entry. */
+ ep->version_type &= ~MSIQ_TYPE_MASK;
+
+ (*head) += sizeof(struct pci_sun4v_msiq_entry);
+ if (*head >=
+ (pbm->msiq_ent_count * sizeof(struct pci_sun4v_msiq_entry)))
+ *head = 0;
+
+ return 1;
+}
+
+static int pci_sun4v_set_head(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long head)
+{
+ unsigned long err;
+
+ err = pci_sun4v_msiq_sethead(pbm->devhandle, msiqid, head);
+ if (unlikely(err))
+ return -EINVAL;
+
+ return 0;
+}
+
+static int pci_sun4v_msi_setup(struct pci_pbm_info *pbm, unsigned long msiqid,
+ unsigned long msi, int is_msi64)
+{
+ if (pci_sun4v_msi_setmsiq(pbm->devhandle, msi, msiqid,
+ (is_msi64 ?
+ HV_MSITYPE_MSI64 : HV_MSITYPE_MSI32)))
+ return -ENXIO;
+ if (pci_sun4v_msi_setstate(pbm->devhandle, msi, HV_MSISTATE_IDLE))
+ return -ENXIO;
+ if (pci_sun4v_msi_setvalid(pbm->devhandle, msi, HV_MSIVALID_VALID))
+ return -ENXIO;
+ return 0;
+}
+
+static int pci_sun4v_msi_teardown(struct pci_pbm_info *pbm, unsigned long msi)
+{
+ unsigned long err, msiqid;
+
+ err = pci_sun4v_msi_getmsiq(pbm->devhandle, msi, &msiqid);
+ if (err)
+ return -ENXIO;
+
+ pci_sun4v_msi_setvalid(pbm->devhandle, msi, HV_MSIVALID_INVALID);
+
+ return 0;
+}
+
+static int pci_sun4v_msiq_alloc(struct pci_pbm_info *pbm)
+{
+ unsigned long q_size, alloc_size, pages, order;
+ int i;
+
+ q_size = pbm->msiq_ent_count * sizeof(struct pci_sun4v_msiq_entry);
+ alloc_size = (pbm->msiq_num * q_size);
+ order = get_order(alloc_size);
+ pages = __get_free_pages(GFP_KERNEL | __GFP_COMP, order);
+ if (pages == 0UL) {
+ printk(KERN_ERR "MSI: Cannot allocate MSI queues (o=%lu).\n",
+ order);
+ return -ENOMEM;
+ }
+ memset((char *)pages, 0, PAGE_SIZE << order);
+ pbm->msi_queues = (void *) pages;
+
+ for (i = 0; i < pbm->msiq_num; i++) {
+ unsigned long err, base = __pa(pages + (i * q_size));
+ unsigned long ret1, ret2;
+
+ err = pci_sun4v_msiq_conf(pbm->devhandle,
+ pbm->msiq_first + i,
+ base, pbm->msiq_ent_count);
+ if (err) {
+ printk(KERN_ERR "MSI: msiq register fails (err=%lu)\n",
+ err);
+ goto h_error;
+ }
+
+ err = pci_sun4v_msiq_info(pbm->devhandle,
+ pbm->msiq_first + i,
+ &ret1, &ret2);
+ if (err) {
+ printk(KERN_ERR "MSI: Cannot read msiq (err=%lu)\n",
+ err);
+ goto h_error;
+ }
+ if (ret1 != base || ret2 != pbm->msiq_ent_count) {
+ printk(KERN_ERR "MSI: Bogus qconf "
+ "expected[%lx:%x] got[%lx:%lx]\n",
+ base, pbm->msiq_ent_count,
+ ret1, ret2);
+ goto h_error;
+ }
+ }
+
+ return 0;
+
+h_error:
+ free_pages(pages, order);
+ return -EINVAL;
+}
+
+static void pci_sun4v_msiq_free(struct pci_pbm_info *pbm)
+{
+ unsigned long q_size, alloc_size, pages, order;
+ int i;
+
+ for (i = 0; i < pbm->msiq_num; i++) {
+ unsigned long msiqid = pbm->msiq_first + i;
+
+ (void) pci_sun4v_msiq_conf(pbm->devhandle, msiqid, 0UL, 0);
+ }
+
+ q_size = pbm->msiq_ent_count * sizeof(struct pci_sun4v_msiq_entry);
+ alloc_size = (pbm->msiq_num * q_size);
+ order = get_order(alloc_size);
+
+ pages = (unsigned long) pbm->msi_queues;
+
+ free_pages(pages, order);
+
+ pbm->msi_queues = NULL;
+}
+
+static int pci_sun4v_msiq_build_irq(struct pci_pbm_info *pbm,
+ unsigned long msiqid,
+ unsigned long devino)
+{
+ unsigned int irq = sun4v_build_irq(pbm->devhandle, devino);
+
+ if (!irq)
+ return -ENOMEM;
+
+ if (pci_sun4v_msiq_setvalid(pbm->devhandle, msiqid, HV_MSIQ_VALID))
+ return -EINVAL;
+ if (pci_sun4v_msiq_setstate(pbm->devhandle, msiqid, HV_MSIQSTATE_IDLE))
+ return -EINVAL;
+
+ return irq;
+}
+
+static const struct sparc64_msiq_ops pci_sun4v_msiq_ops = {
+ .get_head = pci_sun4v_get_head,
+ .dequeue_msi = pci_sun4v_dequeue_msi,
+ .set_head = pci_sun4v_set_head,
+ .msi_setup = pci_sun4v_msi_setup,
+ .msi_teardown = pci_sun4v_msi_teardown,
+ .msiq_alloc = pci_sun4v_msiq_alloc,
+ .msiq_free = pci_sun4v_msiq_free,
+ .msiq_build_irq = pci_sun4v_msiq_build_irq,
+};
+
+static void pci_sun4v_msi_init(struct pci_pbm_info *pbm)
+{
+ sparc64_pbm_msi_init(pbm, &pci_sun4v_msiq_ops);
+}
+#else /* CONFIG_PCI_MSI */
+static void pci_sun4v_msi_init(struct pci_pbm_info *pbm)
+{
+}
+#endif /* !(CONFIG_PCI_MSI) */
+
+static int pci_sun4v_pbm_init(struct pci_pbm_info *pbm,
+ struct platform_device *op, u32 devhandle)
+{
+ struct device_node *dp = op->dev.of_node;
+ int err;
+
+ pbm->numa_node = of_node_to_nid(dp);
+
+ pbm->pci_ops = &sun4v_pci_ops;
+ pbm->config_space_reg_bits = 12;
+
+ pbm->index = pci_num_pbms++;
+
+ pbm->op = op;
+
+ pbm->devhandle = devhandle;
+
+ pbm->name = dp->full_name;
+
+ printk("%s: SUN4V PCI Bus Module\n", pbm->name);
+ printk("%s: On NUMA node %d\n", pbm->name, pbm->numa_node);
+
+ pci_determine_mem_io_space(pbm);
+
+ pci_get_pbm_props(pbm);
+
+ err = pci_sun4v_iommu_init(pbm);
+ if (err)
+ return err;
+
+ pci_sun4v_msi_init(pbm);
+
+ pci_sun4v_scan_bus(pbm, &op->dev);
+
+ pbm->next = pci_pbm_root;
+ pci_pbm_root = pbm;
+
+ return 0;
+}
+
+static int pci_sun4v_probe(struct platform_device *op)
+{
+ const struct linux_prom64_registers *regs;
+ static int hvapi_negotiated = 0;
+ struct pci_pbm_info *pbm;
+ struct device_node *dp;
+ struct iommu *iommu;
+ u32 devhandle;
+ int i, err;
+
+ dp = op->dev.of_node;
+
+ if (!hvapi_negotiated++) {
+ err = sun4v_hvapi_register(HV_GRP_PCI,
+ vpci_major,
+ &vpci_minor);
+
+ if (err) {
+ printk(KERN_ERR PFX "Could not register hvapi, "
+ "err=%d\n", err);
+ return err;
+ }
+ printk(KERN_INFO PFX "Registered hvapi major[%lu] minor[%lu]\n",
+ vpci_major, vpci_minor);
+
+ dma_ops = &sun4v_dma_ops;
+ }
+
+ regs = of_get_property(dp, "reg", NULL);
+ err = -ENODEV;
+ if (!regs) {
+ printk(KERN_ERR PFX "Could not find config registers\n");
+ goto out_err;
+ }
+ devhandle = (regs->phys_addr >> 32UL) & 0x0fffffff;
+
+ err = -ENOMEM;
+ if (!iommu_batch_initialized) {
+ for_each_possible_cpu(i) {
+ unsigned long page = get_zeroed_page(GFP_KERNEL);
+
+ if (!page)
+ goto out_err;
+
+ per_cpu(iommu_batch, i).pglist = (u64 *) page;
+ }
+ iommu_batch_initialized = 1;
+ }
+
+ pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+ if (!pbm) {
+ printk(KERN_ERR PFX "Could not allocate pci_pbm_info\n");
+ goto out_err;
+ }
+
+ iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+ if (!iommu) {
+ printk(KERN_ERR PFX "Could not allocate pbm iommu\n");
+ goto out_free_controller;
+ }
+
+ pbm->iommu = iommu;
+
+ err = pci_sun4v_pbm_init(pbm, op, devhandle);
+ if (err)
+ goto out_free_iommu;
+
+ dev_set_drvdata(&op->dev, pbm);
+
+ return 0;
+
+out_free_iommu:
+ kfree(pbm->iommu);
+
+out_free_controller:
+ kfree(pbm);
+
+out_err:
+ return err;
+}
+
+static const struct of_device_id pci_sun4v_match[] = {
+ {
+ .name = "pci",
+ .compatible = "SUNW,sun4v-pci",
+ },
+ {},
+};
+
+static struct platform_driver pci_sun4v_driver = {
+ .driver = {
+ .name = DRIVER_NAME,
+ .owner = THIS_MODULE,
+ .of_match_table = pci_sun4v_match,
+ },
+ .probe = pci_sun4v_probe,
+};
+
+static int __init pci_sun4v_init(void)
+{
+ return platform_driver_register(&pci_sun4v_driver);
+}
+
+subsys_initcall(pci_sun4v_init);
diff --git a/arch/sparc/kernel/pci_sun4v.h b/arch/sparc/kernel/pci_sun4v.h
new file mode 100644
index 00000000000..5642212390b
--- /dev/null
+++ b/arch/sparc/kernel/pci_sun4v.h
@@ -0,0 +1,92 @@
+/* pci_sun4v.h: SUN4V specific PCI controller support.
+ *
+ * Copyright (C) 2006 David S. Miller (davem@davemloft.net)
+ */
+
+#ifndef _PCI_SUN4V_H
+#define _PCI_SUN4V_H
+
+long pci_sun4v_iommu_map(unsigned long devhandle,
+ unsigned long tsbid,
+ unsigned long num_ttes,
+ unsigned long io_attributes,
+ unsigned long io_page_list_pa);
+unsigned long pci_sun4v_iommu_demap(unsigned long devhandle,
+ unsigned long tsbid,
+ unsigned long num_ttes);
+unsigned long pci_sun4v_iommu_getmap(unsigned long devhandle,
+ unsigned long tsbid,
+ unsigned long *io_attributes,
+ unsigned long *real_address);
+unsigned long pci_sun4v_config_get(unsigned long devhandle,
+ unsigned long pci_device,
+ unsigned long config_offset,
+ unsigned long size);
+int pci_sun4v_config_put(unsigned long devhandle,
+ unsigned long pci_device,
+ unsigned long config_offset,
+ unsigned long size,
+ unsigned long data);
+
+unsigned long pci_sun4v_msiq_conf(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long msiq_paddr,
+ unsigned long num_entries);
+unsigned long pci_sun4v_msiq_info(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long *msiq_paddr,
+ unsigned long *num_entries);
+unsigned long pci_sun4v_msiq_getvalid(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long *valid);
+unsigned long pci_sun4v_msiq_setvalid(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long valid);
+unsigned long pci_sun4v_msiq_getstate(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long *state);
+unsigned long pci_sun4v_msiq_setstate(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long state);
+unsigned long pci_sun4v_msiq_gethead(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long *head);
+unsigned long pci_sun4v_msiq_sethead(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long head);
+unsigned long pci_sun4v_msiq_gettail(unsigned long devhandle,
+ unsigned long msiqid,
+ unsigned long *head);
+unsigned long pci_sun4v_msi_getvalid(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long *valid);
+unsigned long pci_sun4v_msi_setvalid(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long valid);
+unsigned long pci_sun4v_msi_getmsiq(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long *msiq);
+unsigned long pci_sun4v_msi_setmsiq(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long msiq,
+ unsigned long msitype);
+unsigned long pci_sun4v_msi_getstate(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long *state);
+unsigned long pci_sun4v_msi_setstate(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long state);
+unsigned long pci_sun4v_msg_getmsiq(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long *msiq);
+unsigned long pci_sun4v_msg_setmsiq(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long msiq);
+unsigned long pci_sun4v_msg_getvalid(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long *valid);
+unsigned long pci_sun4v_msg_setvalid(unsigned long devhandle,
+ unsigned long msinum,
+ unsigned long valid);
+
+#endif /* !(_PCI_SUN4V_H) */
diff --git a/arch/sparc/kernel/pci_sun4v_asm.S b/arch/sparc/kernel/pci_sun4v_asm.S
new file mode 100644
index 00000000000..e606d46c681
--- /dev/null
+++ b/arch/sparc/kernel/pci_sun4v_asm.S
@@ -0,0 +1,362 @@
+/* pci_sun4v_asm: Hypervisor calls for PCI support.
+ *
+ * Copyright (C) 2006, 2008 David S. Miller <davem@davemloft.net>
+ */
+
+#include <linux/linkage.h>
+#include <asm/hypervisor.h>
+
+ /* %o0: devhandle
+ * %o1: tsbid
+ * %o2: num ttes
+ * %o3: io_attributes
+ * %o4: io_page_list phys address
+ *
+ * returns %o0: -status if status was non-zero, else
+ * %o0: num pages mapped
+ */
+ENTRY(pci_sun4v_iommu_map)
+ mov %o5, %g1
+ mov HV_FAST_PCI_IOMMU_MAP, %o5
+ ta HV_FAST_TRAP
+ brnz,pn %o0, 1f
+ sub %g0, %o0, %o0
+ mov %o1, %o0
+1: retl
+ nop
+ENDPROC(pci_sun4v_iommu_map)
+
+ /* %o0: devhandle
+ * %o1: tsbid
+ * %o2: num ttes
+ *
+ * returns %o0: num ttes demapped
+ */
+ENTRY(pci_sun4v_iommu_demap)
+ mov HV_FAST_PCI_IOMMU_DEMAP, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o1, %o0
+ENDPROC(pci_sun4v_iommu_demap)
+
+ /* %o0: devhandle
+ * %o1: tsbid
+ * %o2: &io_attributes
+ * %o3: &real_address
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_iommu_getmap)
+ mov %o2, %o4
+ mov HV_FAST_PCI_IOMMU_GETMAP, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ stx %o2, [%o3]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_iommu_getmap)
+
+ /* %o0: devhandle
+ * %o1: pci_device
+ * %o2: pci_config_offset
+ * %o3: size
+ *
+ * returns %o0: data
+ *
+ * If there is an error, the data will be returned
+ * as all 1's.
+ */
+ENTRY(pci_sun4v_config_get)
+ mov HV_FAST_PCI_CONFIG_GET, %o5
+ ta HV_FAST_TRAP
+ brnz,a,pn %o1, 1f
+ mov -1, %o2
+1: retl
+ mov %o2, %o0
+ENDPROC(pci_sun4v_config_get)
+
+ /* %o0: devhandle
+ * %o1: pci_device
+ * %o2: pci_config_offset
+ * %o3: size
+ * %o4: data
+ *
+ * returns %o0: status
+ *
+ * status will be zero if the operation completed
+ * successfully, else -1 if not
+ */
+ENTRY(pci_sun4v_config_put)
+ mov HV_FAST_PCI_CONFIG_PUT, %o5
+ ta HV_FAST_TRAP
+ brnz,a,pn %o1, 1f
+ mov -1, %o1
+1: retl
+ mov %o1, %o0
+ENDPROC(pci_sun4v_config_put)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: msiq phys address
+ * %o3: num entries
+ *
+ * returns %o0: status
+ *
+ * status will be zero if the operation completed
+ * successfully, else -1 if not
+ */
+ENTRY(pci_sun4v_msiq_conf)
+ mov HV_FAST_PCI_MSIQ_CONF, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_conf)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: &msiq_phys_addr
+ * %o3: &msiq_num_entries
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_info)
+ mov %o2, %o4
+ mov HV_FAST_PCI_MSIQ_INFO, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o4]
+ stx %o2, [%o3]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_info)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: &valid
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_getvalid)
+ mov HV_FAST_PCI_MSIQ_GETVALID, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_getvalid)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: valid
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_setvalid)
+ mov HV_FAST_PCI_MSIQ_SETVALID, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_setvalid)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: &state
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_getstate)
+ mov HV_FAST_PCI_MSIQ_GETSTATE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_getstate)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: state
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_setstate)
+ mov HV_FAST_PCI_MSIQ_SETSTATE, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_setstate)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: &head
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_gethead)
+ mov HV_FAST_PCI_MSIQ_GETHEAD, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_gethead)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: head
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_sethead)
+ mov HV_FAST_PCI_MSIQ_SETHEAD, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_sethead)
+
+ /* %o0: devhandle
+ * %o1: msiqid
+ * %o2: &tail
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msiq_gettail)
+ mov HV_FAST_PCI_MSIQ_GETTAIL, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msiq_gettail)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: &valid
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msi_getvalid)
+ mov HV_FAST_PCI_MSI_GETVALID, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msi_getvalid)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: valid
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msi_setvalid)
+ mov HV_FAST_PCI_MSI_SETVALID, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msi_setvalid)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: &msiq
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msi_getmsiq)
+ mov HV_FAST_PCI_MSI_GETMSIQ, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msi_getmsiq)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: msitype
+ * %o3: msiq
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msi_setmsiq)
+ mov HV_FAST_PCI_MSI_SETMSIQ, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msi_setmsiq)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: &state
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msi_getstate)
+ mov HV_FAST_PCI_MSI_GETSTATE, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msi_getstate)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: state
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msi_setstate)
+ mov HV_FAST_PCI_MSI_SETSTATE, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msi_setstate)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: &msiq
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msg_getmsiq)
+ mov HV_FAST_PCI_MSG_GETMSIQ, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msg_getmsiq)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: msiq
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msg_setmsiq)
+ mov HV_FAST_PCI_MSG_SETMSIQ, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msg_setmsiq)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: &valid
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msg_getvalid)
+ mov HV_FAST_PCI_MSG_GETVALID, %o5
+ ta HV_FAST_TRAP
+ stx %o1, [%o2]
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msg_getvalid)
+
+ /* %o0: devhandle
+ * %o1: msinum
+ * %o2: valid
+ *
+ * returns %o0: status
+ */
+ENTRY(pci_sun4v_msg_setvalid)
+ mov HV_FAST_PCI_MSG_SETVALID, %o5
+ ta HV_FAST_TRAP
+ retl
+ mov %o0, %o0
+ENDPROC(pci_sun4v_msg_setvalid)
+
diff --git a/arch/sparc/kernel/pcic.c b/arch/sparc/kernel/pcic.c
index a6a6f982337..6cc78c213c0 100644
--- a/arch/sparc/kernel/pcic.c
+++ b/arch/sparc/kernel/pcic.c
@@ -17,8 +17,6 @@
#include <linux/slab.h>
#include <linux/jiffies.h>
-#include <asm/ebus.h>
-#include <asm/sbus.h> /* for sanity check... */
#include <asm/swift.h> /* for cache flushing. */
#include <asm/io.h>
@@ -27,15 +25,18 @@
#include <linux/time.h>
#include <linux/timex.h>
#include <linux/interrupt.h>
+#include <linux/export.h>
#include <asm/irq.h>
#include <asm/oplib.h>
#include <asm/prom.h>
#include <asm/pcic.h>
+#include <asm/timex.h>
#include <asm/timer.h>
#include <asm/uaccess.h>
#include <asm/irq_regs.h>
+#include "kernel.h"
#include "irq.h"
/*
@@ -162,11 +163,12 @@ static int pcic0_up;
static struct linux_pcic pcic0;
void __iomem *pcic_regs;
-volatile int pcic_speculative;
-volatile int pcic_trapped;
+static volatile int pcic_speculative;
+static volatile int pcic_trapped;
-static void pci_do_gettimeofday(struct timeval *tv);
-static int pci_do_settimeofday(struct timespec *tv);
+/* forward */
+unsigned int pcic_build_device_irq(struct platform_device *op,
+ unsigned int real_irq);
#define CONFIG_CMD(bus, device_fn, where) (0x80000000 | (((unsigned int)bus) << 16) | (((unsigned int)device_fn) << 8) | (where & ~3))
@@ -287,7 +289,7 @@ int __init pcic_probe(void)
struct linux_prom_registers regs[PROMREG_MAX];
struct linux_pbm_info* pbm;
char namebuf[64];
- int node;
+ phandle node;
int err;
if (pcic0_up) {
@@ -328,7 +330,7 @@ int __init pcic_probe(void)
pcic->pcic_res_cfg_addr.name = "pcic_cfg_addr";
if ((pcic->pcic_config_space_addr =
- ioremap(regs[2].phys_addr, regs[2].reg_size * 2)) == 0) {
+ ioremap(regs[2].phys_addr, regs[2].reg_size * 2)) == NULL) {
prom_printf("PCIC: Error, cannot map "
"PCI Configuration Space Address.\n");
prom_halt();
@@ -340,7 +342,7 @@ int __init pcic_probe(void)
*/
pcic->pcic_res_cfg_data.name = "pcic_cfg_data";
if ((pcic->pcic_config_space_data =
- ioremap(regs[3].phys_addr, regs[3].reg_size * 2)) == 0) {
+ ioremap(regs[3].phys_addr, regs[3].reg_size * 2)) == NULL) {
prom_printf("PCIC: Error, cannot map "
"PCI Configuration Space Data.\n");
prom_halt();
@@ -352,8 +354,7 @@ int __init pcic_probe(void)
strcpy(pbm->prom_name, namebuf);
{
- extern volatile int t_nmi[1];
- extern int pcic_nmi_trap_patch[1];
+ extern int pcic_nmi_trap_patch[4];
t_nmi[0] = pcic_nmi_trap_patch[0];
t_nmi[1] = pcic_nmi_trap_patch[1];
@@ -430,7 +431,6 @@ static int __init pcic_init(void)
pcic_pbm_scan_bus(pcic);
- ebus_init();
return 0;
}
@@ -439,12 +439,11 @@ int pcic_present(void)
return pcic0_up;
}
-static int __init pdev_to_pnode(struct linux_pbm_info *pbm,
- struct pci_dev *pdev)
+static int pdev_to_pnode(struct linux_pbm_info *pbm, struct pci_dev *pdev)
{
struct linux_prom_pci_registers regs[PROMREG_MAX];
int err;
- int node = prom_getchild(pbm->prom_node);
+ phandle node = prom_getchild(pbm->prom_node);
while(node) {
err = prom_getproperty(node, "reg",
@@ -493,10 +492,6 @@ static void pcic_map_pci_device(struct linux_pcic *pcic,
* do ioremap() before accessing PC-style I/O,
* we supply virtual, ready to access address.
*
- * Ebus devices do not come here even if
- * CheerIO makes a similar conversion.
- * See ebus.c for details.
- *
* Note that request_region()
* works for these devices.
*
@@ -531,6 +526,7 @@ static void
pcic_fill_irq(struct linux_pcic *pcic, struct pci_dev *dev, int node)
{
struct pcic_ca2irq *p;
+ unsigned int real_irq;
int i, ivec;
char namebuf[64];
@@ -540,7 +536,7 @@ pcic_fill_irq(struct linux_pcic *pcic, struct pci_dev *dev, int node)
prom_getstring(node, "name", namebuf, sizeof(namebuf));
}
- if ((p = pcic->pcic_imap) == 0) {
+ if ((p = pcic->pcic_imap) == NULL) {
dev->irq = 0;
return;
}
@@ -559,26 +555,25 @@ pcic_fill_irq(struct linux_pcic *pcic, struct pci_dev *dev, int node)
i = p->pin;
if (i >= 0 && i < 4) {
ivec = readw(pcic->pcic_regs+PCI_INT_SELECT_LO);
- dev->irq = ivec >> (i << 2) & 0xF;
+ real_irq = ivec >> (i << 2) & 0xF;
} else if (i >= 4 && i < 8) {
ivec = readw(pcic->pcic_regs+PCI_INT_SELECT_HI);
- dev->irq = ivec >> ((i-4) << 2) & 0xF;
+ real_irq = ivec >> ((i-4) << 2) & 0xF;
} else { /* Corrupted map */
printk("PCIC: BAD PIN %d\n", i); for (;;) {}
}
/* P3 */ /* printk("PCIC: device %s pin %d ivec 0x%x irq %x\n", namebuf, i, ivec, dev->irq); */
- /*
- * dev->irq=0 means PROM did not bother to program the upper
+ /* real_irq means PROM did not bother to program the upper
* half of PCIC. This happens on JS-E with PROM 3.11, for instance.
*/
- if (dev->irq == 0 || p->force) {
+ if (real_irq == 0 || p->force) {
if (p->irq == 0 || p->irq >= 15) { /* Corrupted map */
printk("PCIC: BAD IRQ %d\n", p->irq); for (;;) {}
}
printk("PCIC: setting irq %d at pin %d for device %02x:%02x\n",
p->irq, p->pin, dev->bus->number, dev->devfn);
- dev->irq = p->irq;
+ real_irq = p->irq;
i = p->pin;
if (i >= 4) {
@@ -592,15 +587,14 @@ pcic_fill_irq(struct linux_pcic *pcic, struct pci_dev *dev, int node)
ivec |= p->irq << (i << 2);
writew(ivec, pcic->pcic_regs+PCI_INT_SELECT_LO);
}
- }
-
- return;
+ }
+ dev->irq = pcic_build_device_irq(NULL, real_irq);
}
/*
* Normally called from {do_}pci_scan_bus...
*/
-void __devinit pcibios_fixup_bus(struct pci_bus *bus)
+void pcibios_fixup_bus(struct pci_bus *bus)
{
struct pci_dev *dev;
int i, has_io, has_mem;
@@ -676,30 +670,6 @@ void __devinit pcibios_fixup_bus(struct pci_bus *bus)
}
}
-/*
- * pcic_pin_to_irq() is exported to ebus.c.
- */
-unsigned int
-pcic_pin_to_irq(unsigned int pin, const char *name)
-{
- struct linux_pcic *pcic = &pcic0;
- unsigned int irq;
- unsigned int ivec;
-
- if (pin < 4) {
- ivec = readw(pcic->pcic_regs+PCI_INT_SELECT_LO);
- irq = ivec >> (pin << 2) & 0xF;
- } else if (pin < 8) {
- ivec = readw(pcic->pcic_regs+PCI_INT_SELECT_HI);
- irq = ivec >> ((pin-4) << 2) & 0xF;
- } else { /* Corrupted map */
- printk("PCIC: BAD PIN %d FOR %s\n", pin, name);
- for (;;) {} /* XXX Cannot panic properly in case of PROLL */
- }
-/* P3 */ /* printk("PCIC: dev %s pin %d ivec 0x%x irq %x\n", name, pin, ivec, irq); */
- return irq;
-}
-
/* Makes compiler happy */
static volatile int pcic_timer_dummy;
@@ -708,35 +678,46 @@ static void pcic_clear_clock_irq(void)
pcic_timer_dummy = readl(pcic0.pcic_regs+PCI_SYS_LIMIT);
}
-static irqreturn_t pcic_timer_handler (int irq, void *h)
+/* CPU frequency is 100 MHz, timer increments every 4 CPU clocks */
+#define USECS_PER_JIFFY (1000000 / HZ)
+#define TICK_TIMER_LIMIT ((100 * 1000000 / 4) / HZ)
+
+static unsigned int pcic_cycles_offset(void)
{
- write_seqlock(&xtime_lock); /* Dummy, to show that we remember */
- pcic_clear_clock_irq();
- do_timer(1);
- write_sequnlock(&xtime_lock);
-#ifndef CONFIG_SMP
- update_process_times(user_mode(get_irq_regs()));
-#endif
- return IRQ_HANDLED;
-}
+ u32 value, count;
-#define USECS_PER_JIFFY 10000 /* We have 100HZ "standard" timer for sparc */
-#define TICK_TIMER_LIMIT ((100*1000000/4)/100)
+ value = readl(pcic0.pcic_regs + PCI_SYS_COUNTER);
+ count = value & ~PCI_SYS_COUNTER_OVERFLOW;
+
+ if (value & PCI_SYS_COUNTER_OVERFLOW)
+ count += TICK_TIMER_LIMIT;
+ /*
+ * We divide all by HZ
+ * to have microsecond resolution and to avoid overflow
+ */
+ count = ((count / HZ) * USECS_PER_JIFFY) / (TICK_TIMER_LIMIT / HZ);
+
+ /* Coordinate with the sparc_config.clock_rate setting */
+ return count * 2;
+}
void __init pci_time_init(void)
{
struct linux_pcic *pcic = &pcic0;
unsigned long v;
int timer_irq, irq;
+ int err;
- /* A hack until do_gettimeofday prototype is moved to arch specific headers
- and btfixupped. Patch do_gettimeofday with ba pci_do_gettimeofday; nop */
- ((unsigned int *)do_gettimeofday)[0] =
- 0x10800000 | ((((unsigned long)pci_do_gettimeofday -
- (unsigned long)do_gettimeofday) >> 2) & 0x003fffff);
- ((unsigned int *)do_gettimeofday)[1] = 0x01000000;
- BTFIXUPSET_CALL(bus_do_settimeofday, pci_do_settimeofday, BTFIXUPCALL_NORM);
- btfixup();
+#ifndef CONFIG_SMP
+ /*
+ * The clock_rate is in SBUS dimension.
+ * We take into account this in pcic_cycles_offset()
+ */
+ sparc_config.clock_rate = SBUS_CLOCK_RATE / HZ;
+ sparc_config.features |= FEAT_L10_CLOCKEVENT;
+#endif
+ sparc_config.features |= FEAT_L10_CLOCKSOURCE;
+ sparc_config.get_cycles_offset = pcic_cycles_offset;
writel (TICK_TIMER_LIMIT, pcic->pcic_regs+PCI_SYS_LIMIT);
/* PROM should set appropriate irq */
@@ -744,93 +725,16 @@ void __init pci_time_init(void)
timer_irq = PCI_COUNTER_IRQ_SYS(v);
writel (PCI_COUNTER_IRQ_SET(timer_irq, 0),
pcic->pcic_regs+PCI_COUNTER_IRQ);
- irq = request_irq(timer_irq, pcic_timer_handler,
- (IRQF_DISABLED | SA_STATIC_ALLOC), "timer", NULL);
- if (irq) {
+ irq = pcic_build_device_irq(NULL, timer_irq);
+ err = request_irq(irq, timer_interrupt,
+ IRQF_TIMER, "timer", NULL);
+ if (err) {
prom_printf("time_init: unable to attach IRQ%d\n", timer_irq);
prom_halt();
}
local_irq_enable();
}
-static inline unsigned long do_gettimeoffset(void)
-{
- /*
- * We divide all by 100
- * to have microsecond resolution and to avoid overflow
- */
- unsigned long count =
- readl(pcic0.pcic_regs+PCI_SYS_COUNTER) & ~PCI_SYS_COUNTER_OVERFLOW;
- count = ((count/100)*USECS_PER_JIFFY) / (TICK_TIMER_LIMIT/100);
- return count;
-}
-
-static void pci_do_gettimeofday(struct timeval *tv)
-{
- unsigned long flags;
- unsigned long seq;
- unsigned long usec, sec;
- unsigned long max_ntp_tick = tick_usec - tickadj;
-
- do {
- seq = read_seqbegin_irqsave(&xtime_lock, flags);
- usec = do_gettimeoffset();
-
- /*
- * If time_adjust is negative then NTP is slowing the clock
- * so make sure not to go into next possible interval.
- * Better to lose some accuracy than have time go backwards..
- */
- if (unlikely(time_adjust < 0))
- usec = min(usec, max_ntp_tick);
-
- sec = xtime.tv_sec;
- usec += (xtime.tv_nsec / 1000);
- } while (read_seqretry_irqrestore(&xtime_lock, seq, flags));
-
- while (usec >= 1000000) {
- usec -= 1000000;
- sec++;
- }
-
- tv->tv_sec = sec;
- tv->tv_usec = usec;
-}
-
-static int pci_do_settimeofday(struct timespec *tv)
-{
- if ((unsigned long)tv->tv_nsec >= NSEC_PER_SEC)
- return -EINVAL;
-
- /*
- * This is revolting. We need to set "xtime" correctly. However, the
- * value in this location is the value at the most recent update of
- * wall time. Discover what correction gettimeofday() would have
- * made, and then undo it!
- */
- tv->tv_nsec -= 1000 * do_gettimeoffset();
- while (tv->tv_nsec < 0) {
- tv->tv_nsec += NSEC_PER_SEC;
- tv->tv_sec--;
- }
-
- wall_to_monotonic.tv_sec += xtime.tv_sec - tv->tv_sec;
- wall_to_monotonic.tv_nsec += xtime.tv_nsec - tv->tv_nsec;
-
- if (wall_to_monotonic.tv_nsec > NSEC_PER_SEC) {
- wall_to_monotonic.tv_nsec -= NSEC_PER_SEC;
- wall_to_monotonic.tv_sec++;
- }
- if (wall_to_monotonic.tv_nsec < 0) {
- wall_to_monotonic.tv_nsec += NSEC_PER_SEC;
- wall_to_monotonic.tv_sec--;
- }
-
- xtime.tv_sec = tv->tv_sec;
- xtime.tv_nsec = tv->tv_nsec;
- ntp_clear();
- return 0;
-}
#if 0
static void watchdog_reset() {
@@ -838,17 +742,10 @@ static void watchdog_reset() {
}
#endif
-/*
- * Other archs parse arguments here.
- */
-char * __devinit pcibios_setup(char *str)
-{
- return str;
-}
-
-void pcibios_align_resource(void *data, struct resource *res,
- resource_size_t size, resource_size_t align)
+resource_size_t pcibios_align_resource(void *data, const struct resource *res,
+ resource_size_t size, resource_size_t align)
{
+ return res->start;
}
int pcibios_enable_device(struct pci_dev *pdev, int mask)
@@ -862,7 +759,7 @@ int pcibios_enable_device(struct pci_dev *pdev, int mask)
void pcic_nmi(unsigned int pend, struct pt_regs *regs)
{
- pend = flip_dword(pend);
+ pend = swab32(pend);
if (!pcic_speculative || (pend & PCI_SYS_INT_PENDING_PIO) == 0) {
/*
@@ -884,143 +781,74 @@ static inline unsigned long get_irqmask(int irq_nr)
return 1 << irq_nr;
}
-static void pcic_disable_irq(unsigned int irq_nr)
+static void pcic_mask_irq(struct irq_data *data)
{
unsigned long mask, flags;
- mask = get_irqmask(irq_nr);
+ mask = (unsigned long)data->chip_data;
local_irq_save(flags);
writel(mask, pcic0.pcic_regs+PCI_SYS_INT_TARGET_MASK_SET);
local_irq_restore(flags);
}
-static void pcic_enable_irq(unsigned int irq_nr)
+static void pcic_unmask_irq(struct irq_data *data)
{
unsigned long mask, flags;
- mask = get_irqmask(irq_nr);
+ mask = (unsigned long)data->chip_data;
local_irq_save(flags);
writel(mask, pcic0.pcic_regs+PCI_SYS_INT_TARGET_MASK_CLEAR);
local_irq_restore(flags);
}
-static void pcic_clear_profile_irq(int cpu)
-{
- printk("PCIC: unimplemented code: FILE=%s LINE=%d", __FILE__, __LINE__);
-}
-
-static void pcic_load_profile_irq(int cpu, unsigned int limit)
-{
- printk("PCIC: unimplemented code: FILE=%s LINE=%d", __FILE__, __LINE__);
-}
-
-/* We assume the caller has disabled local interrupts when these are called,
- * or else very bizarre behavior will result.
- */
-static void pcic_disable_pil_irq(unsigned int pil)
-{
- writel(get_irqmask(pil), pcic0.pcic_regs+PCI_SYS_INT_TARGET_MASK_SET);
-}
-
-static void pcic_enable_pil_irq(unsigned int pil)
-{
- writel(get_irqmask(pil), pcic0.pcic_regs+PCI_SYS_INT_TARGET_MASK_CLEAR);
-}
-
-void __init sun4m_pci_init_IRQ(void)
+static unsigned int pcic_startup_irq(struct irq_data *data)
{
- BTFIXUPSET_CALL(enable_irq, pcic_enable_irq, BTFIXUPCALL_NORM);
- BTFIXUPSET_CALL(disable_irq, pcic_disable_irq, BTFIXUPCALL_NORM);
- BTFIXUPSET_CALL(enable_pil_irq, pcic_enable_pil_irq, BTFIXUPCALL_NORM);
- BTFIXUPSET_CALL(disable_pil_irq, pcic_disable_pil_irq, BTFIXUPCALL_NORM);
- BTFIXUPSET_CALL(clear_clock_irq, pcic_clear_clock_irq, BTFIXUPCALL_NORM);
- BTFIXUPSET_CALL(clear_profile_irq, pcic_clear_profile_irq, BTFIXUPCALL_NORM);
- BTFIXUPSET_CALL(load_profile_irq, pcic_load_profile_irq, BTFIXUPCALL_NORM);
+ irq_link(data->irq);
+ pcic_unmask_irq(data);
+ return 0;
}
-int pcibios_assign_resource(struct pci_dev *pdev, int resource)
-{
- return -ENXIO;
-}
+static struct irq_chip pcic_irq = {
+ .name = "pcic",
+ .irq_startup = pcic_startup_irq,
+ .irq_mask = pcic_mask_irq,
+ .irq_unmask = pcic_unmask_irq,
+};
-struct device_node *pci_device_to_OF_node(struct pci_dev *pdev)
+unsigned int pcic_build_device_irq(struct platform_device *op,
+ unsigned int real_irq)
{
- struct pcidev_cookie *pc = pdev->sysdata;
+ unsigned int irq;
+ unsigned long mask;
- return pc->prom_node;
-}
-EXPORT_SYMBOL(pci_device_to_OF_node);
+ irq = 0;
+ mask = get_irqmask(real_irq);
+ if (mask == 0)
+ goto out;
-/*
- * This probably belongs here rather than ioport.c because
- * we do not want this crud linked into SBus kernels.
- * Also, think for a moment about likes of floppy.c that
- * include architecture specific parts. They may want to redefine ins/outs.
- *
- * We do not use horrible macros here because we want to
- * advance pointer by sizeof(size).
- */
-void outsb(unsigned long addr, const void *src, unsigned long count)
-{
- while (count) {
- count -= 1;
- outb(*(const char *)src, addr);
- src += 1;
- /* addr += 1; */
- }
-}
+ irq = irq_alloc(real_irq, real_irq);
+ if (irq == 0)
+ goto out;
-void outsw(unsigned long addr, const void *src, unsigned long count)
-{
- while (count) {
- count -= 2;
- outw(*(const short *)src, addr);
- src += 2;
- /* addr += 2; */
- }
-}
+ irq_set_chip_and_handler_name(irq, &pcic_irq,
+ handle_level_irq, "PCIC");
+ irq_set_chip_data(irq, (void *)mask);
-void outsl(unsigned long addr, const void *src, unsigned long count)
-{
- while (count) {
- count -= 4;
- outl(*(const long *)src, addr);
- src += 4;
- /* addr += 4; */
- }
+out:
+ return irq;
}
-void insb(unsigned long addr, void *dst, unsigned long count)
-{
- while (count) {
- count -= 1;
- *(unsigned char *)dst = inb(addr);
- dst += 1;
- /* addr += 1; */
- }
-}
-void insw(unsigned long addr, void *dst, unsigned long count)
+static void pcic_load_profile_irq(int cpu, unsigned int limit)
{
- while (count) {
- count -= 2;
- *(unsigned short *)dst = inw(addr);
- dst += 2;
- /* addr += 2; */
- }
+ printk("PCIC: unimplemented code: FILE=%s LINE=%d", __FILE__, __LINE__);
}
-void insl(unsigned long addr, void *dst, unsigned long count)
+void __init sun4m_pci_init_IRQ(void)
{
- while (count) {
- count -= 4;
- /*
- * XXX I am sure we are in for an unaligned trap here.
- */
- *(unsigned long *)dst = inl(addr);
- dst += 4;
- /* addr += 4; */
- }
+ sparc_config.build_device_irq = pcic_build_device_irq;
+ sparc_config.clear_clock_irq = pcic_clear_clock_irq;
+ sparc_config.load_profile_irq = pcic_load_profile_irq;
}
subsys_initcall(pcic_init);
diff --git a/arch/sparc/kernel/pcr.c b/arch/sparc/kernel/pcr.c
new file mode 100644
index 00000000000..269af58497a
--- /dev/null
+++ b/arch/sparc/kernel/pcr.c
@@ -0,0 +1,300 @@
+/* pcr.c: Generic sparc64 performance counter infrastructure.
+ *
+ * Copyright (C) 2009 David S. Miller (davem@davemloft.net)
+ */
+#include <linux/kernel.h>
+#include <linux/export.h>
+#include <linux/init.h>
+#include <linux/irq.h>
+
+#include <linux/irq_work.h>
+#include <linux/ftrace.h>
+
+#include <asm/pil.h>
+#include <asm/pcr.h>
+#include <asm/nmi.h>
+#include <asm/asi.h>
+#include <asm/spitfire.h>
+
+/* This code is shared between various users of the performance
+ * counters. Users will be oprofile, pseudo-NMI watchdog, and the
+ * perf_event support layer.
+ */
+
+/* Performance counter interrupts run unmasked at PIL level 15.
+ * Therefore we can't do things like wakeups and other work
+ * that expects IRQ disabling to be adhered to in locking etc.
+ *
+ * Therefore in such situations we defer the work by signalling
+ * a lower level cpu IRQ.
+ */
+void __irq_entry deferred_pcr_work_irq(int irq, struct pt_regs *regs)
+{
+ struct pt_regs *old_regs;
+
+ clear_softint(1 << PIL_DEFERRED_PCR_WORK);
+
+ old_regs = set_irq_regs(regs);
+ irq_enter();
+#ifdef CONFIG_IRQ_WORK
+ irq_work_run();
+#endif
+ irq_exit();
+ set_irq_regs(old_regs);
+}
+
+void arch_irq_work_raise(void)
+{
+ set_softint(1 << PIL_DEFERRED_PCR_WORK);
+}
+
+const struct pcr_ops *pcr_ops;
+EXPORT_SYMBOL_GPL(pcr_ops);
+
+static u64 direct_pcr_read(unsigned long reg_num)
+{
+ u64 val;
+
+ WARN_ON_ONCE(reg_num != 0);
+ __asm__ __volatile__("rd %%pcr, %0" : "=r" (val));
+ return val;
+}
+
+static void direct_pcr_write(unsigned long reg_num, u64 val)
+{
+ WARN_ON_ONCE(reg_num != 0);
+ __asm__ __volatile__("wr %0, 0x0, %%pcr" : : "r" (val));
+}
+
+static u64 direct_pic_read(unsigned long reg_num)
+{
+ u64 val;
+
+ WARN_ON_ONCE(reg_num != 0);
+ __asm__ __volatile__("rd %%pic, %0" : "=r" (val));
+ return val;
+}
+
+static void direct_pic_write(unsigned long reg_num, u64 val)
+{
+ WARN_ON_ONCE(reg_num != 0);
+
+ /* Blackbird errata workaround. See commentary in
+ * arch/sparc64/kernel/smp.c:smp_percpu_timer_interrupt()
+ * for more information.
+ */
+ __asm__ __volatile__("ba,pt %%xcc, 99f\n\t"
+ " nop\n\t"
+ ".align 64\n"
+ "99:wr %0, 0x0, %%pic\n\t"
+ "rd %%pic, %%g0" : : "r" (val));
+}
+
+static u64 direct_picl_value(unsigned int nmi_hz)
+{
+ u32 delta = local_cpu_data().clock_tick / nmi_hz;
+
+ return ((u64)((0 - delta) & 0xffffffff)) << 32;
+}
+
+static const struct pcr_ops direct_pcr_ops = {
+ .read_pcr = direct_pcr_read,
+ .write_pcr = direct_pcr_write,
+ .read_pic = direct_pic_read,
+ .write_pic = direct_pic_write,
+ .nmi_picl_value = direct_picl_value,
+ .pcr_nmi_enable = (PCR_PIC_PRIV | PCR_STRACE | PCR_UTRACE),
+ .pcr_nmi_disable = PCR_PIC_PRIV,
+};
+
+static void n2_pcr_write(unsigned long reg_num, u64 val)
+{
+ unsigned long ret;
+
+ WARN_ON_ONCE(reg_num != 0);
+ if (val & PCR_N2_HTRACE) {
+ ret = sun4v_niagara2_setperf(HV_N2_PERF_SPARC_CTL, val);
+ if (ret != HV_EOK)
+ direct_pcr_write(reg_num, val);
+ } else
+ direct_pcr_write(reg_num, val);
+}
+
+static u64 n2_picl_value(unsigned int nmi_hz)
+{
+ u32 delta = local_cpu_data().clock_tick / (nmi_hz << 2);
+
+ return ((u64)((0 - delta) & 0xffffffff)) << 32;
+}
+
+static const struct pcr_ops n2_pcr_ops = {
+ .read_pcr = direct_pcr_read,
+ .write_pcr = n2_pcr_write,
+ .read_pic = direct_pic_read,
+ .write_pic = direct_pic_write,
+ .nmi_picl_value = n2_picl_value,
+ .pcr_nmi_enable = (PCR_PIC_PRIV | PCR_STRACE | PCR_UTRACE |
+ PCR_N2_TOE_OV1 |
+ (2 << PCR_N2_SL1_SHIFT) |
+ (0xff << PCR_N2_MASK1_SHIFT)),
+ .pcr_nmi_disable = PCR_PIC_PRIV,
+};
+
+static u64 n4_pcr_read(unsigned long reg_num)
+{
+ unsigned long val;
+
+ (void) sun4v_vt_get_perfreg(reg_num, &val);
+
+ return val;
+}
+
+static void n4_pcr_write(unsigned long reg_num, u64 val)
+{
+ (void) sun4v_vt_set_perfreg(reg_num, val);
+}
+
+static u64 n4_pic_read(unsigned long reg_num)
+{
+ unsigned long val;
+
+ __asm__ __volatile__("ldxa [%1] %2, %0"
+ : "=r" (val)
+ : "r" (reg_num * 0x8UL), "i" (ASI_PIC));
+
+ return val;
+}
+
+static void n4_pic_write(unsigned long reg_num, u64 val)
+{
+ __asm__ __volatile__("stxa %0, [%1] %2"
+ : /* no outputs */
+ : "r" (val), "r" (reg_num * 0x8UL), "i" (ASI_PIC));
+}
+
+static u64 n4_picl_value(unsigned int nmi_hz)
+{
+ u32 delta = local_cpu_data().clock_tick / (nmi_hz << 2);
+
+ return ((u64)((0 - delta) & 0xffffffff));
+}
+
+static const struct pcr_ops n4_pcr_ops = {
+ .read_pcr = n4_pcr_read,
+ .write_pcr = n4_pcr_write,
+ .read_pic = n4_pic_read,
+ .write_pic = n4_pic_write,
+ .nmi_picl_value = n4_picl_value,
+ .pcr_nmi_enable = (PCR_N4_PICNPT | PCR_N4_STRACE |
+ PCR_N4_UTRACE | PCR_N4_TOE |
+ (26 << PCR_N4_SL_SHIFT)),
+ .pcr_nmi_disable = PCR_N4_PICNPT,
+};
+
+static unsigned long perf_hsvc_group;
+static unsigned long perf_hsvc_major;
+static unsigned long perf_hsvc_minor;
+
+static int __init register_perf_hsvc(void)
+{
+ if (tlb_type == hypervisor) {
+ switch (sun4v_chip_type) {
+ case SUN4V_CHIP_NIAGARA1:
+ perf_hsvc_group = HV_GRP_NIAG_PERF;
+ break;
+
+ case SUN4V_CHIP_NIAGARA2:
+ perf_hsvc_group = HV_GRP_N2_CPU;
+ break;
+
+ case SUN4V_CHIP_NIAGARA3:
+ perf_hsvc_group = HV_GRP_KT_CPU;
+ break;
+
+ case SUN4V_CHIP_NIAGARA4:
+ perf_hsvc_group = HV_GRP_VT_CPU;
+ break;
+
+ default:
+ return -ENODEV;
+ }
+
+
+ perf_hsvc_major = 1;
+ perf_hsvc_minor = 0;
+ if (sun4v_hvapi_register(perf_hsvc_group,
+ perf_hsvc_major,
+ &perf_hsvc_minor)) {
+ printk("perfmon: Could not register hvapi.\n");
+ return -ENODEV;
+ }
+ }
+ return 0;
+}
+
+static void __init unregister_perf_hsvc(void)
+{
+ if (tlb_type != hypervisor)
+ return;
+ sun4v_hvapi_unregister(perf_hsvc_group);
+}
+
+static int __init setup_sun4v_pcr_ops(void)
+{
+ int ret = 0;
+
+ switch (sun4v_chip_type) {
+ case SUN4V_CHIP_NIAGARA1:
+ case SUN4V_CHIP_NIAGARA2:
+ case SUN4V_CHIP_NIAGARA3:
+ pcr_ops = &n2_pcr_ops;
+ break;
+
+ case SUN4V_CHIP_NIAGARA4:
+ pcr_ops = &n4_pcr_ops;
+ break;
+
+ default:
+ ret = -ENODEV;
+ break;
+ }
+
+ return ret;
+}
+
+int __init pcr_arch_init(void)
+{
+ int err = register_perf_hsvc();
+
+ if (err)
+ return err;
+
+ switch (tlb_type) {
+ case hypervisor:
+ err = setup_sun4v_pcr_ops();
+ if (err)
+ goto out_unregister;
+ break;
+
+ case cheetah:
+ case cheetah_plus:
+ pcr_ops = &direct_pcr_ops;
+ break;
+
+ case spitfire:
+ /* UltraSPARC-I/II and derivatives lack a profile
+ * counter overflow interrupt so we can't make use of
+ * their hardware currently.
+ */
+ /* fallthrough */
+ default:
+ err = -ENODEV;
+ goto out_unregister;
+ }
+
+ return nmi_init();
+
+out_unregister:
+ unregister_perf_hsvc();
+ return err;
+}
diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c
new file mode 100644
index 00000000000..8efd33753ad
--- /dev/null
+++ b/arch/sparc/kernel/perf_event.c
@@ -0,0 +1,1804 @@
+/* Performance event support for sparc64.
+ *
+ * Copyright (C) 2009, 2010 David S. Miller <davem@davemloft.net>
+ *
+ * This code is based almost entirely upon the x86 perf event
+ * code, which is:
+ *
+ * Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de>
+ * Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar
+ * Copyright (C) 2009 Jaswinder Singh Rajput
+ * Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter
+ * Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
+ */
+
+#include <linux/perf_event.h>
+#include <linux/kprobes.h>
+#include <linux/ftrace.h>
+#include <linux/kernel.h>
+#include <linux/kdebug.h>
+#include <linux/mutex.h>
+
+#include <asm/stacktrace.h>
+#include <asm/cpudata.h>
+#include <asm/uaccess.h>
+#include <linux/atomic.h>
+#include <asm/nmi.h>
+#include <asm/pcr.h>
+#include <asm/cacheflush.h>
+
+#include "kernel.h"
+#include "kstack.h"
+
+/* Two classes of sparc64 chips currently exist. All of which have
+ * 32-bit counters which can generate overflow interrupts on the
+ * transition from 0xffffffff to 0.
+ *
+ * All chips upto and including SPARC-T3 have two performance
+ * counters. The two 32-bit counters are accessed in one go using a
+ * single 64-bit register.
+ *
+ * On these older chips both counters are controlled using a single
+ * control register. The only way to stop all sampling is to clear
+ * all of the context (user, supervisor, hypervisor) sampling enable
+ * bits. But these bits apply to both counters, thus the two counters
+ * can't be enabled/disabled individually.
+ *
+ * Furthermore, the control register on these older chips have two
+ * event fields, one for each of the two counters. It's thus nearly
+ * impossible to have one counter going while keeping the other one
+ * stopped. Therefore it is possible to get overflow interrupts for
+ * counters not currently "in use" and that condition must be checked
+ * in the overflow interrupt handler.
+ *
+ * So we use a hack, in that we program inactive counters with the
+ * "sw_count0" and "sw_count1" events. These count how many times
+ * the instruction "sethi %hi(0xfc000), %g0" is executed. It's an
+ * unusual way to encode a NOP and therefore will not trigger in
+ * normal code.
+ *
+ * Starting with SPARC-T4 we have one control register per counter.
+ * And the counters are stored in individual registers. The registers
+ * for the counters are 64-bit but only a 32-bit counter is
+ * implemented. The event selections on SPARC-T4 lack any
+ * restrictions, therefore we can elide all of the complicated
+ * conflict resolution code we have for SPARC-T3 and earlier chips.
+ */
+
+#define MAX_HWEVENTS 4
+#define MAX_PCRS 4
+#define MAX_PERIOD ((1UL << 32) - 1)
+
+#define PIC_UPPER_INDEX 0
+#define PIC_LOWER_INDEX 1
+#define PIC_NO_INDEX -1
+
+struct cpu_hw_events {
+ /* Number of events currently scheduled onto this cpu.
+ * This tells how many entries in the arrays below
+ * are valid.
+ */
+ int n_events;
+
+ /* Number of new events added since the last hw_perf_disable().
+ * This works because the perf event layer always adds new
+ * events inside of a perf_{disable,enable}() sequence.
+ */
+ int n_added;
+
+ /* Array of events current scheduled on this cpu. */
+ struct perf_event *event[MAX_HWEVENTS];
+
+ /* Array of encoded longs, specifying the %pcr register
+ * encoding and the mask of PIC counters this even can
+ * be scheduled on. See perf_event_encode() et al.
+ */
+ unsigned long events[MAX_HWEVENTS];
+
+ /* The current counter index assigned to an event. When the
+ * event hasn't been programmed into the cpu yet, this will
+ * hold PIC_NO_INDEX. The event->hw.idx value tells us where
+ * we ought to schedule the event.
+ */
+ int current_idx[MAX_HWEVENTS];
+
+ /* Software copy of %pcr register(s) on this cpu. */
+ u64 pcr[MAX_HWEVENTS];
+
+ /* Enabled/disable state. */
+ int enabled;
+
+ unsigned int group_flag;
+};
+static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = { .enabled = 1, };
+
+/* An event map describes the characteristics of a performance
+ * counter event. In particular it gives the encoding as well as
+ * a mask telling which counters the event can be measured on.
+ *
+ * The mask is unused on SPARC-T4 and later.
+ */
+struct perf_event_map {
+ u16 encoding;
+ u8 pic_mask;
+#define PIC_NONE 0x00
+#define PIC_UPPER 0x01
+#define PIC_LOWER 0x02
+};
+
+/* Encode a perf_event_map entry into a long. */
+static unsigned long perf_event_encode(const struct perf_event_map *pmap)
+{
+ return ((unsigned long) pmap->encoding << 16) | pmap->pic_mask;
+}
+
+static u8 perf_event_get_msk(unsigned long val)
+{
+ return val & 0xff;
+}
+
+static u64 perf_event_get_enc(unsigned long val)
+{
+ return val >> 16;
+}
+
+#define C(x) PERF_COUNT_HW_CACHE_##x
+
+#define CACHE_OP_UNSUPPORTED 0xfffe
+#define CACHE_OP_NONSENSE 0xffff
+
+typedef struct perf_event_map cache_map_t
+ [PERF_COUNT_HW_CACHE_MAX]
+ [PERF_COUNT_HW_CACHE_OP_MAX]
+ [PERF_COUNT_HW_CACHE_RESULT_MAX];
+
+struct sparc_pmu {
+ const struct perf_event_map *(*event_map)(int);
+ const cache_map_t *cache_map;
+ int max_events;
+ u32 (*read_pmc)(int);
+ void (*write_pmc)(int, u64);
+ int upper_shift;
+ int lower_shift;
+ int event_mask;
+ int user_bit;
+ int priv_bit;
+ int hv_bit;
+ int irq_bit;
+ int upper_nop;
+ int lower_nop;
+ unsigned int flags;
+#define SPARC_PMU_ALL_EXCLUDES_SAME 0x00000001
+#define SPARC_PMU_HAS_CONFLICTS 0x00000002
+ int max_hw_events;
+ int num_pcrs;
+ int num_pic_regs;
+};
+
+static u32 sparc_default_read_pmc(int idx)
+{
+ u64 val;
+
+ val = pcr_ops->read_pic(0);
+ if (idx == PIC_UPPER_INDEX)
+ val >>= 32;
+
+ return val & 0xffffffff;
+}
+
+static void sparc_default_write_pmc(int idx, u64 val)
+{
+ u64 shift, mask, pic;
+
+ shift = 0;
+ if (idx == PIC_UPPER_INDEX)
+ shift = 32;
+
+ mask = ((u64) 0xffffffff) << shift;
+ val <<= shift;
+
+ pic = pcr_ops->read_pic(0);
+ pic &= ~mask;
+ pic |= val;
+ pcr_ops->write_pic(0, pic);
+}
+
+static const struct perf_event_map ultra3_perfmon_event_map[] = {
+ [PERF_COUNT_HW_CPU_CYCLES] = { 0x0000, PIC_UPPER | PIC_LOWER },
+ [PERF_COUNT_HW_INSTRUCTIONS] = { 0x0001, PIC_UPPER | PIC_LOWER },
+ [PERF_COUNT_HW_CACHE_REFERENCES] = { 0x0009, PIC_LOWER },
+ [PERF_COUNT_HW_CACHE_MISSES] = { 0x0009, PIC_UPPER },
+};
+
+static const struct perf_event_map *ultra3_event_map(int event_id)
+{
+ return &ultra3_perfmon_event_map[event_id];
+}
+
+static const cache_map_t ultra3_cache_map = {
+[C(L1D)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x09, PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x09, PIC_UPPER, },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { 0x0a, PIC_LOWER },
+ [C(RESULT_MISS)] = { 0x0a, PIC_UPPER },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(L1I)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x09, PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x09, PIC_UPPER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_NONSENSE },
+ [ C(RESULT_MISS) ] = { CACHE_OP_NONSENSE },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(LL)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x0c, PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x0c, PIC_UPPER, },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { 0x0c, PIC_LOWER },
+ [C(RESULT_MISS)] = { 0x0c, PIC_UPPER },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(DTLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x12, PIC_UPPER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(ITLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x11, PIC_UPPER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(BPU)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(NODE)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+};
+
+static const struct sparc_pmu ultra3_pmu = {
+ .event_map = ultra3_event_map,
+ .cache_map = &ultra3_cache_map,
+ .max_events = ARRAY_SIZE(ultra3_perfmon_event_map),
+ .read_pmc = sparc_default_read_pmc,
+ .write_pmc = sparc_default_write_pmc,
+ .upper_shift = 11,
+ .lower_shift = 4,
+ .event_mask = 0x3f,
+ .user_bit = PCR_UTRACE,
+ .priv_bit = PCR_STRACE,
+ .upper_nop = 0x1c,
+ .lower_nop = 0x14,
+ .flags = (SPARC_PMU_ALL_EXCLUDES_SAME |
+ SPARC_PMU_HAS_CONFLICTS),
+ .max_hw_events = 2,
+ .num_pcrs = 1,
+ .num_pic_regs = 1,
+};
+
+/* Niagara1 is very limited. The upper PIC is hard-locked to count
+ * only instructions, so it is free running which creates all kinds of
+ * problems. Some hardware designs make one wonder if the creator
+ * even looked at how this stuff gets used by software.
+ */
+static const struct perf_event_map niagara1_perfmon_event_map[] = {
+ [PERF_COUNT_HW_CPU_CYCLES] = { 0x00, PIC_UPPER },
+ [PERF_COUNT_HW_INSTRUCTIONS] = { 0x00, PIC_UPPER },
+ [PERF_COUNT_HW_CACHE_REFERENCES] = { 0, PIC_NONE },
+ [PERF_COUNT_HW_CACHE_MISSES] = { 0x03, PIC_LOWER },
+};
+
+static const struct perf_event_map *niagara1_event_map(int event_id)
+{
+ return &niagara1_perfmon_event_map[event_id];
+}
+
+static const cache_map_t niagara1_cache_map = {
+[C(L1D)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x03, PIC_LOWER, },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x03, PIC_LOWER, },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(L1I)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x00, PIC_UPPER },
+ [C(RESULT_MISS)] = { 0x02, PIC_LOWER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_NONSENSE },
+ [ C(RESULT_MISS) ] = { CACHE_OP_NONSENSE },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(LL)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x07, PIC_LOWER, },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x07, PIC_LOWER, },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(DTLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x05, PIC_LOWER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(ITLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x04, PIC_LOWER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(BPU)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(NODE)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+};
+
+static const struct sparc_pmu niagara1_pmu = {
+ .event_map = niagara1_event_map,
+ .cache_map = &niagara1_cache_map,
+ .max_events = ARRAY_SIZE(niagara1_perfmon_event_map),
+ .read_pmc = sparc_default_read_pmc,
+ .write_pmc = sparc_default_write_pmc,
+ .upper_shift = 0,
+ .lower_shift = 4,
+ .event_mask = 0x7,
+ .user_bit = PCR_UTRACE,
+ .priv_bit = PCR_STRACE,
+ .upper_nop = 0x0,
+ .lower_nop = 0x0,
+ .flags = (SPARC_PMU_ALL_EXCLUDES_SAME |
+ SPARC_PMU_HAS_CONFLICTS),
+ .max_hw_events = 2,
+ .num_pcrs = 1,
+ .num_pic_regs = 1,
+};
+
+static const struct perf_event_map niagara2_perfmon_event_map[] = {
+ [PERF_COUNT_HW_CPU_CYCLES] = { 0x02ff, PIC_UPPER | PIC_LOWER },
+ [PERF_COUNT_HW_INSTRUCTIONS] = { 0x02ff, PIC_UPPER | PIC_LOWER },
+ [PERF_COUNT_HW_CACHE_REFERENCES] = { 0x0208, PIC_UPPER | PIC_LOWER },
+ [PERF_COUNT_HW_CACHE_MISSES] = { 0x0302, PIC_UPPER | PIC_LOWER },
+ [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = { 0x0201, PIC_UPPER | PIC_LOWER },
+ [PERF_COUNT_HW_BRANCH_MISSES] = { 0x0202, PIC_UPPER | PIC_LOWER },
+};
+
+static const struct perf_event_map *niagara2_event_map(int event_id)
+{
+ return &niagara2_perfmon_event_map[event_id];
+}
+
+static const cache_map_t niagara2_cache_map = {
+[C(L1D)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x0208, PIC_UPPER | PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x0302, PIC_UPPER | PIC_LOWER, },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { 0x0210, PIC_UPPER | PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x0302, PIC_UPPER | PIC_LOWER, },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(L1I)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x02ff, PIC_UPPER | PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x0301, PIC_UPPER | PIC_LOWER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_NONSENSE },
+ [ C(RESULT_MISS) ] = { CACHE_OP_NONSENSE },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(LL)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x0208, PIC_UPPER | PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x0330, PIC_UPPER | PIC_LOWER, },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { 0x0210, PIC_UPPER | PIC_LOWER, },
+ [C(RESULT_MISS)] = { 0x0320, PIC_UPPER | PIC_LOWER, },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(DTLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0x0b08, PIC_UPPER | PIC_LOWER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(ITLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { 0xb04, PIC_UPPER | PIC_LOWER, },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(BPU)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(NODE)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+};
+
+static const struct sparc_pmu niagara2_pmu = {
+ .event_map = niagara2_event_map,
+ .cache_map = &niagara2_cache_map,
+ .max_events = ARRAY_SIZE(niagara2_perfmon_event_map),
+ .read_pmc = sparc_default_read_pmc,
+ .write_pmc = sparc_default_write_pmc,
+ .upper_shift = 19,
+ .lower_shift = 6,
+ .event_mask = 0xfff,
+ .user_bit = PCR_UTRACE,
+ .priv_bit = PCR_STRACE,
+ .hv_bit = PCR_N2_HTRACE,
+ .irq_bit = 0x30,
+ .upper_nop = 0x220,
+ .lower_nop = 0x220,
+ .flags = (SPARC_PMU_ALL_EXCLUDES_SAME |
+ SPARC_PMU_HAS_CONFLICTS),
+ .max_hw_events = 2,
+ .num_pcrs = 1,
+ .num_pic_regs = 1,
+};
+
+static const struct perf_event_map niagara4_perfmon_event_map[] = {
+ [PERF_COUNT_HW_CPU_CYCLES] = { (26 << 6) },
+ [PERF_COUNT_HW_INSTRUCTIONS] = { (3 << 6) | 0x3f },
+ [PERF_COUNT_HW_CACHE_REFERENCES] = { (3 << 6) | 0x04 },
+ [PERF_COUNT_HW_CACHE_MISSES] = { (16 << 6) | 0x07 },
+ [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = { (4 << 6) | 0x01 },
+ [PERF_COUNT_HW_BRANCH_MISSES] = { (25 << 6) | 0x0f },
+};
+
+static const struct perf_event_map *niagara4_event_map(int event_id)
+{
+ return &niagara4_perfmon_event_map[event_id];
+}
+
+static const cache_map_t niagara4_cache_map = {
+[C(L1D)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { (3 << 6) | 0x04 },
+ [C(RESULT_MISS)] = { (16 << 6) | 0x07 },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { (3 << 6) | 0x08 },
+ [C(RESULT_MISS)] = { (16 << 6) | 0x07 },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(L1I)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { (3 << 6) | 0x3f },
+ [C(RESULT_MISS)] = { (11 << 6) | 0x03 },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_NONSENSE },
+ [ C(RESULT_MISS) ] = { CACHE_OP_NONSENSE },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(LL)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { (3 << 6) | 0x04 },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { (3 << 6) | 0x08 },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(DTLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { (17 << 6) | 0x3f },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(ITLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { (6 << 6) | 0x3f },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(BPU)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+[C(NODE)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
+ [C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_WRITE) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+ [ C(OP_PREFETCH) ] = {
+ [ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
+ [ C(RESULT_MISS) ] = { CACHE_OP_UNSUPPORTED },
+ },
+},
+};
+
+static u32 sparc_vt_read_pmc(int idx)
+{
+ u64 val = pcr_ops->read_pic(idx);
+
+ return val & 0xffffffff;
+}
+
+static void sparc_vt_write_pmc(int idx, u64 val)
+{
+ u64 pcr;
+
+ /* There seems to be an internal latch on the overflow event
+ * on SPARC-T4 that prevents it from triggering unless you
+ * update the PIC exactly as we do here. The requirement
+ * seems to be that you have to turn off event counting in the
+ * PCR around the PIC update.
+ *
+ * For example, after the following sequence:
+ *
+ * 1) set PIC to -1
+ * 2) enable event counting and overflow reporting in PCR
+ * 3) overflow triggers, softint 15 handler invoked
+ * 4) clear OV bit in PCR
+ * 5) write PIC to -1
+ *
+ * a subsequent overflow event will not trigger. This
+ * sequence works on SPARC-T3 and previous chips.
+ */
+ pcr = pcr_ops->read_pcr(idx);
+ pcr_ops->write_pcr(idx, PCR_N4_PICNPT);
+
+ pcr_ops->write_pic(idx, val & 0xffffffff);
+
+ pcr_ops->write_pcr(idx, pcr);
+}
+
+static const struct sparc_pmu niagara4_pmu = {
+ .event_map = niagara4_event_map,
+ .cache_map = &niagara4_cache_map,
+ .max_events = ARRAY_SIZE(niagara4_perfmon_event_map),
+ .read_pmc = sparc_vt_read_pmc,
+ .write_pmc = sparc_vt_write_pmc,
+ .upper_shift = 5,
+ .lower_shift = 5,
+ .event_mask = 0x7ff,
+ .user_bit = PCR_N4_UTRACE,
+ .priv_bit = PCR_N4_STRACE,
+
+ /* We explicitly don't support hypervisor tracing. The T4
+ * generates the overflow event for precise events via a trap
+ * which will not be generated (ie. it's completely lost) if
+ * we happen to be in the hypervisor when the event triggers.
+ * Essentially, the overflow event reporting is completely
+ * unusable when you have hypervisor mode tracing enabled.
+ */
+ .hv_bit = 0,
+
+ .irq_bit = PCR_N4_TOE,
+ .upper_nop = 0,
+ .lower_nop = 0,
+ .flags = 0,
+ .max_hw_events = 4,
+ .num_pcrs = 4,
+ .num_pic_regs = 4,
+};
+
+static const struct sparc_pmu *sparc_pmu __read_mostly;
+
+static u64 event_encoding(u64 event_id, int idx)
+{
+ if (idx == PIC_UPPER_INDEX)
+ event_id <<= sparc_pmu->upper_shift;
+ else
+ event_id <<= sparc_pmu->lower_shift;
+ return event_id;
+}
+
+static u64 mask_for_index(int idx)
+{
+ return event_encoding(sparc_pmu->event_mask, idx);
+}
+
+static u64 nop_for_index(int idx)
+{
+ return event_encoding(idx == PIC_UPPER_INDEX ?
+ sparc_pmu->upper_nop :
+ sparc_pmu->lower_nop, idx);
+}
+
+static inline void sparc_pmu_enable_event(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc, int idx)
+{
+ u64 enc, val, mask = mask_for_index(idx);
+ int pcr_index = 0;
+
+ if (sparc_pmu->num_pcrs > 1)
+ pcr_index = idx;
+
+ enc = perf_event_get_enc(cpuc->events[idx]);
+
+ val = cpuc->pcr[pcr_index];
+ val &= ~mask;
+ val |= event_encoding(enc, idx);
+ cpuc->pcr[pcr_index] = val;
+
+ pcr_ops->write_pcr(pcr_index, cpuc->pcr[pcr_index]);
+}
+
+static inline void sparc_pmu_disable_event(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc, int idx)
+{
+ u64 mask = mask_for_index(idx);
+ u64 nop = nop_for_index(idx);
+ int pcr_index = 0;
+ u64 val;
+
+ if (sparc_pmu->num_pcrs > 1)
+ pcr_index = idx;
+
+ val = cpuc->pcr[pcr_index];
+ val &= ~mask;
+ val |= nop;
+ cpuc->pcr[pcr_index] = val;
+
+ pcr_ops->write_pcr(pcr_index, cpuc->pcr[pcr_index]);
+}
+
+static u64 sparc_perf_event_update(struct perf_event *event,
+ struct hw_perf_event *hwc, int idx)
+{
+ int shift = 64 - 32;
+ u64 prev_raw_count, new_raw_count;
+ s64 delta;
+
+again:
+ prev_raw_count = local64_read(&hwc->prev_count);
+ new_raw_count = sparc_pmu->read_pmc(idx);
+
+ if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+ new_raw_count) != prev_raw_count)
+ goto again;
+
+ delta = (new_raw_count << shift) - (prev_raw_count << shift);
+ delta >>= shift;
+
+ local64_add(delta, &event->count);
+ local64_sub(delta, &hwc->period_left);
+
+ return new_raw_count;
+}
+
+static int sparc_perf_event_set_period(struct perf_event *event,
+ struct hw_perf_event *hwc, int idx)
+{
+ s64 left = local64_read(&hwc->period_left);
+ s64 period = hwc->sample_period;
+ int ret = 0;
+
+ if (unlikely(left <= -period)) {
+ left = period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ ret = 1;
+ }
+
+ if (unlikely(left <= 0)) {
+ left += period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ ret = 1;
+ }
+ if (left > MAX_PERIOD)
+ left = MAX_PERIOD;
+
+ local64_set(&hwc->prev_count, (u64)-left);
+
+ sparc_pmu->write_pmc(idx, (u64)(-left) & 0xffffffff);
+
+ perf_event_update_userpage(event);
+
+ return ret;
+}
+
+static void read_in_all_counters(struct cpu_hw_events *cpuc)
+{
+ int i;
+
+ for (i = 0; i < cpuc->n_events; i++) {
+ struct perf_event *cp = cpuc->event[i];
+
+ if (cpuc->current_idx[i] != PIC_NO_INDEX &&
+ cpuc->current_idx[i] != cp->hw.idx) {
+ sparc_perf_event_update(cp, &cp->hw,
+ cpuc->current_idx[i]);
+ cpuc->current_idx[i] = PIC_NO_INDEX;
+ }
+ }
+}
+
+/* On this PMU all PICs are programmed using a single PCR. Calculate
+ * the combined control register value.
+ *
+ * For such chips we require that all of the events have the same
+ * configuration, so just fetch the settings from the first entry.
+ */
+static void calculate_single_pcr(struct cpu_hw_events *cpuc)
+{
+ int i;
+
+ if (!cpuc->n_added)
+ goto out;
+
+ /* Assign to counters all unassigned events. */
+ for (i = 0; i < cpuc->n_events; i++) {
+ struct perf_event *cp = cpuc->event[i];
+ struct hw_perf_event *hwc = &cp->hw;
+ int idx = hwc->idx;
+ u64 enc;
+
+ if (cpuc->current_idx[i] != PIC_NO_INDEX)
+ continue;
+
+ sparc_perf_event_set_period(cp, hwc, idx);
+ cpuc->current_idx[i] = idx;
+
+ enc = perf_event_get_enc(cpuc->events[i]);
+ cpuc->pcr[0] &= ~mask_for_index(idx);
+ if (hwc->state & PERF_HES_STOPPED)
+ cpuc->pcr[0] |= nop_for_index(idx);
+ else
+ cpuc->pcr[0] |= event_encoding(enc, idx);
+ }
+out:
+ cpuc->pcr[0] |= cpuc->event[0]->hw.config_base;
+}
+
+/* On this PMU each PIC has it's own PCR control register. */
+static void calculate_multiple_pcrs(struct cpu_hw_events *cpuc)
+{
+ int i;
+
+ if (!cpuc->n_added)
+ goto out;
+
+ for (i = 0; i < cpuc->n_events; i++) {
+ struct perf_event *cp = cpuc->event[i];
+ struct hw_perf_event *hwc = &cp->hw;
+ int idx = hwc->idx;
+ u64 enc;
+
+ if (cpuc->current_idx[i] != PIC_NO_INDEX)
+ continue;
+
+ sparc_perf_event_set_period(cp, hwc, idx);
+ cpuc->current_idx[i] = idx;
+
+ enc = perf_event_get_enc(cpuc->events[i]);
+ cpuc->pcr[idx] &= ~mask_for_index(idx);
+ if (hwc->state & PERF_HES_STOPPED)
+ cpuc->pcr[idx] |= nop_for_index(idx);
+ else
+ cpuc->pcr[idx] |= event_encoding(enc, idx);
+ }
+out:
+ for (i = 0; i < cpuc->n_events; i++) {
+ struct perf_event *cp = cpuc->event[i];
+ int idx = cp->hw.idx;
+
+ cpuc->pcr[idx] |= cp->hw.config_base;
+ }
+}
+
+/* If performance event entries have been added, move existing events
+ * around (if necessary) and then assign new entries to counters.
+ */
+static void update_pcrs_for_enable(struct cpu_hw_events *cpuc)
+{
+ if (cpuc->n_added)
+ read_in_all_counters(cpuc);
+
+ if (sparc_pmu->num_pcrs == 1) {
+ calculate_single_pcr(cpuc);
+ } else {
+ calculate_multiple_pcrs(cpuc);
+ }
+}
+
+static void sparc_pmu_enable(struct pmu *pmu)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int i;
+
+ if (cpuc->enabled)
+ return;
+
+ cpuc->enabled = 1;
+ barrier();
+
+ if (cpuc->n_events)
+ update_pcrs_for_enable(cpuc);
+
+ for (i = 0; i < sparc_pmu->num_pcrs; i++)
+ pcr_ops->write_pcr(i, cpuc->pcr[i]);
+}
+
+static void sparc_pmu_disable(struct pmu *pmu)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int i;
+
+ if (!cpuc->enabled)
+ return;
+
+ cpuc->enabled = 0;
+ cpuc->n_added = 0;
+
+ for (i = 0; i < sparc_pmu->num_pcrs; i++) {
+ u64 val = cpuc->pcr[i];
+
+ val &= ~(sparc_pmu->user_bit | sparc_pmu->priv_bit |
+ sparc_pmu->hv_bit | sparc_pmu->irq_bit);
+ cpuc->pcr[i] = val;
+ pcr_ops->write_pcr(i, cpuc->pcr[i]);
+ }
+}
+
+static int active_event_index(struct cpu_hw_events *cpuc,
+ struct perf_event *event)
+{
+ int i;
+
+ for (i = 0; i < cpuc->n_events; i++) {
+ if (cpuc->event[i] == event)
+ break;
+ }
+ BUG_ON(i == cpuc->n_events);
+ return cpuc->current_idx[i];
+}
+
+static void sparc_pmu_start(struct perf_event *event, int flags)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int idx = active_event_index(cpuc, event);
+
+ if (flags & PERF_EF_RELOAD) {
+ WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE));
+ sparc_perf_event_set_period(event, &event->hw, idx);
+ }
+
+ event->hw.state = 0;
+
+ sparc_pmu_enable_event(cpuc, &event->hw, idx);
+}
+
+static void sparc_pmu_stop(struct perf_event *event, int flags)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int idx = active_event_index(cpuc, event);
+
+ if (!(event->hw.state & PERF_HES_STOPPED)) {
+ sparc_pmu_disable_event(cpuc, &event->hw, idx);
+ event->hw.state |= PERF_HES_STOPPED;
+ }
+
+ if (!(event->hw.state & PERF_HES_UPTODATE) && (flags & PERF_EF_UPDATE)) {
+ sparc_perf_event_update(event, &event->hw, idx);
+ event->hw.state |= PERF_HES_UPTODATE;
+ }
+}
+
+static void sparc_pmu_del(struct perf_event *event, int _flags)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ unsigned long flags;
+ int i;
+
+ local_irq_save(flags);
+ perf_pmu_disable(event->pmu);
+
+ for (i = 0; i < cpuc->n_events; i++) {
+ if (event == cpuc->event[i]) {
+ /* Absorb the final count and turn off the
+ * event.
+ */
+ sparc_pmu_stop(event, PERF_EF_UPDATE);
+
+ /* Shift remaining entries down into
+ * the existing slot.
+ */
+ while (++i < cpuc->n_events) {
+ cpuc->event[i - 1] = cpuc->event[i];
+ cpuc->events[i - 1] = cpuc->events[i];
+ cpuc->current_idx[i - 1] =
+ cpuc->current_idx[i];
+ }
+
+ perf_event_update_userpage(event);
+
+ cpuc->n_events--;
+ break;
+ }
+ }
+
+ perf_pmu_enable(event->pmu);
+ local_irq_restore(flags);
+}
+
+static void sparc_pmu_read(struct perf_event *event)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int idx = active_event_index(cpuc, event);
+ struct hw_perf_event *hwc = &event->hw;
+
+ sparc_perf_event_update(event, hwc, idx);
+}
+
+static atomic_t active_events = ATOMIC_INIT(0);
+static DEFINE_MUTEX(pmc_grab_mutex);
+
+static void perf_stop_nmi_watchdog(void *unused)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int i;
+
+ stop_nmi_watchdog(NULL);
+ for (i = 0; i < sparc_pmu->num_pcrs; i++)
+ cpuc->pcr[i] = pcr_ops->read_pcr(i);
+}
+
+static void perf_event_grab_pmc(void)
+{
+ if (atomic_inc_not_zero(&active_events))
+ return;
+
+ mutex_lock(&pmc_grab_mutex);
+ if (atomic_read(&active_events) == 0) {
+ if (atomic_read(&nmi_active) > 0) {
+ on_each_cpu(perf_stop_nmi_watchdog, NULL, 1);
+ BUG_ON(atomic_read(&nmi_active) != 0);
+ }
+ atomic_inc(&active_events);
+ }
+ mutex_unlock(&pmc_grab_mutex);
+}
+
+static void perf_event_release_pmc(void)
+{
+ if (atomic_dec_and_mutex_lock(&active_events, &pmc_grab_mutex)) {
+ if (atomic_read(&nmi_active) == 0)
+ on_each_cpu(start_nmi_watchdog, NULL, 1);
+ mutex_unlock(&pmc_grab_mutex);
+ }
+}
+
+static const struct perf_event_map *sparc_map_cache_event(u64 config)
+{
+ unsigned int cache_type, cache_op, cache_result;
+ const struct perf_event_map *pmap;
+
+ if (!sparc_pmu->cache_map)
+ return ERR_PTR(-ENOENT);
+
+ cache_type = (config >> 0) & 0xff;
+ if (cache_type >= PERF_COUNT_HW_CACHE_MAX)
+ return ERR_PTR(-EINVAL);
+
+ cache_op = (config >> 8) & 0xff;
+ if (cache_op >= PERF_COUNT_HW_CACHE_OP_MAX)
+ return ERR_PTR(-EINVAL);
+
+ cache_result = (config >> 16) & 0xff;
+ if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
+ return ERR_PTR(-EINVAL);
+
+ pmap = &((*sparc_pmu->cache_map)[cache_type][cache_op][cache_result]);
+
+ if (pmap->encoding == CACHE_OP_UNSUPPORTED)
+ return ERR_PTR(-ENOENT);
+
+ if (pmap->encoding == CACHE_OP_NONSENSE)
+ return ERR_PTR(-EINVAL);
+
+ return pmap;
+}
+
+static void hw_perf_event_destroy(struct perf_event *event)
+{
+ perf_event_release_pmc();
+}
+
+/* Make sure all events can be scheduled into the hardware at
+ * the same time. This is simplified by the fact that we only
+ * need to support 2 simultaneous HW events.
+ *
+ * As a side effect, the evts[]->hw.idx values will be assigned
+ * on success. These are pending indexes. When the events are
+ * actually programmed into the chip, these values will propagate
+ * to the per-cpu cpuc->current_idx[] slots, see the code in
+ * maybe_change_configuration() for details.
+ */
+static int sparc_check_constraints(struct perf_event **evts,
+ unsigned long *events, int n_ev)
+{
+ u8 msk0 = 0, msk1 = 0;
+ int idx0 = 0;
+
+ /* This case is possible when we are invoked from
+ * hw_perf_group_sched_in().
+ */
+ if (!n_ev)
+ return 0;
+
+ if (n_ev > sparc_pmu->max_hw_events)
+ return -1;
+
+ if (!(sparc_pmu->flags & SPARC_PMU_HAS_CONFLICTS)) {
+ int i;
+
+ for (i = 0; i < n_ev; i++)
+ evts[i]->hw.idx = i;
+ return 0;
+ }
+
+ msk0 = perf_event_get_msk(events[0]);
+ if (n_ev == 1) {
+ if (msk0 & PIC_LOWER)
+ idx0 = 1;
+ goto success;
+ }
+ BUG_ON(n_ev != 2);
+ msk1 = perf_event_get_msk(events[1]);
+
+ /* If both events can go on any counter, OK. */
+ if (msk0 == (PIC_UPPER | PIC_LOWER) &&
+ msk1 == (PIC_UPPER | PIC_LOWER))
+ goto success;
+
+ /* If one event is limited to a specific counter,
+ * and the other can go on both, OK.
+ */
+ if ((msk0 == PIC_UPPER || msk0 == PIC_LOWER) &&
+ msk1 == (PIC_UPPER | PIC_LOWER)) {
+ if (msk0 & PIC_LOWER)
+ idx0 = 1;
+ goto success;
+ }
+
+ if ((msk1 == PIC_UPPER || msk1 == PIC_LOWER) &&
+ msk0 == (PIC_UPPER | PIC_LOWER)) {
+ if (msk1 & PIC_UPPER)
+ idx0 = 1;
+ goto success;
+ }
+
+ /* If the events are fixed to different counters, OK. */
+ if ((msk0 == PIC_UPPER && msk1 == PIC_LOWER) ||
+ (msk0 == PIC_LOWER && msk1 == PIC_UPPER)) {
+ if (msk0 & PIC_LOWER)
+ idx0 = 1;
+ goto success;
+ }
+
+ /* Otherwise, there is a conflict. */
+ return -1;
+
+success:
+ evts[0]->hw.idx = idx0;
+ if (n_ev == 2)
+ evts[1]->hw.idx = idx0 ^ 1;
+ return 0;
+}
+
+static int check_excludes(struct perf_event **evts, int n_prev, int n_new)
+{
+ int eu = 0, ek = 0, eh = 0;
+ struct perf_event *event;
+ int i, n, first;
+
+ if (!(sparc_pmu->flags & SPARC_PMU_ALL_EXCLUDES_SAME))
+ return 0;
+
+ n = n_prev + n_new;
+ if (n <= 1)
+ return 0;
+
+ first = 1;
+ for (i = 0; i < n; i++) {
+ event = evts[i];
+ if (first) {
+ eu = event->attr.exclude_user;
+ ek = event->attr.exclude_kernel;
+ eh = event->attr.exclude_hv;
+ first = 0;
+ } else if (event->attr.exclude_user != eu ||
+ event->attr.exclude_kernel != ek ||
+ event->attr.exclude_hv != eh) {
+ return -EAGAIN;
+ }
+ }
+
+ return 0;
+}
+
+static int collect_events(struct perf_event *group, int max_count,
+ struct perf_event *evts[], unsigned long *events,
+ int *current_idx)
+{
+ struct perf_event *event;
+ int n = 0;
+
+ if (!is_software_event(group)) {
+ if (n >= max_count)
+ return -1;
+ evts[n] = group;
+ events[n] = group->hw.event_base;
+ current_idx[n++] = PIC_NO_INDEX;
+ }
+ list_for_each_entry(event, &group->sibling_list, group_entry) {
+ if (!is_software_event(event) &&
+ event->state != PERF_EVENT_STATE_OFF) {
+ if (n >= max_count)
+ return -1;
+ evts[n] = event;
+ events[n] = event->hw.event_base;
+ current_idx[n++] = PIC_NO_INDEX;
+ }
+ }
+ return n;
+}
+
+static int sparc_pmu_add(struct perf_event *event, int ef_flags)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int n0, ret = -EAGAIN;
+ unsigned long flags;
+
+ local_irq_save(flags);
+ perf_pmu_disable(event->pmu);
+
+ n0 = cpuc->n_events;
+ if (n0 >= sparc_pmu->max_hw_events)
+ goto out;
+
+ cpuc->event[n0] = event;
+ cpuc->events[n0] = event->hw.event_base;
+ cpuc->current_idx[n0] = PIC_NO_INDEX;
+
+ event->hw.state = PERF_HES_UPTODATE;
+ if (!(ef_flags & PERF_EF_START))
+ event->hw.state |= PERF_HES_STOPPED;
+
+ /*
+ * If group events scheduling transaction was started,
+ * skip the schedulability test here, it will be performed
+ * at commit time(->commit_txn) as a whole
+ */
+ if (cpuc->group_flag & PERF_EVENT_TXN)
+ goto nocheck;
+
+ if (check_excludes(cpuc->event, n0, 1))
+ goto out;
+ if (sparc_check_constraints(cpuc->event, cpuc->events, n0 + 1))
+ goto out;
+
+nocheck:
+ cpuc->n_events++;
+ cpuc->n_added++;
+
+ ret = 0;
+out:
+ perf_pmu_enable(event->pmu);
+ local_irq_restore(flags);
+ return ret;
+}
+
+static int sparc_pmu_event_init(struct perf_event *event)
+{
+ struct perf_event_attr *attr = &event->attr;
+ struct perf_event *evts[MAX_HWEVENTS];
+ struct hw_perf_event *hwc = &event->hw;
+ unsigned long events[MAX_HWEVENTS];
+ int current_idx_dmy[MAX_HWEVENTS];
+ const struct perf_event_map *pmap;
+ int n;
+
+ if (atomic_read(&nmi_active) < 0)
+ return -ENODEV;
+
+ /* does not support taken branch sampling */
+ if (has_branch_stack(event))
+ return -EOPNOTSUPP;
+
+ switch (attr->type) {
+ case PERF_TYPE_HARDWARE:
+ if (attr->config >= sparc_pmu->max_events)
+ return -EINVAL;
+ pmap = sparc_pmu->event_map(attr->config);
+ break;
+
+ case PERF_TYPE_HW_CACHE:
+ pmap = sparc_map_cache_event(attr->config);
+ if (IS_ERR(pmap))
+ return PTR_ERR(pmap);
+ break;
+
+ case PERF_TYPE_RAW:
+ pmap = NULL;
+ break;
+
+ default:
+ return -ENOENT;
+
+ }
+
+ if (pmap) {
+ hwc->event_base = perf_event_encode(pmap);
+ } else {
+ /*
+ * User gives us "(encoding << 16) | pic_mask" for
+ * PERF_TYPE_RAW events.
+ */
+ hwc->event_base = attr->config;
+ }
+
+ /* We save the enable bits in the config_base. */
+ hwc->config_base = sparc_pmu->irq_bit;
+ if (!attr->exclude_user)
+ hwc->config_base |= sparc_pmu->user_bit;
+ if (!attr->exclude_kernel)
+ hwc->config_base |= sparc_pmu->priv_bit;
+ if (!attr->exclude_hv)
+ hwc->config_base |= sparc_pmu->hv_bit;
+
+ n = 0;
+ if (event->group_leader != event) {
+ n = collect_events(event->group_leader,
+ sparc_pmu->max_hw_events - 1,
+ evts, events, current_idx_dmy);
+ if (n < 0)
+ return -EINVAL;
+ }
+ events[n] = hwc->event_base;
+ evts[n] = event;
+
+ if (check_excludes(evts, n, 1))
+ return -EINVAL;
+
+ if (sparc_check_constraints(evts, events, n + 1))
+ return -EINVAL;
+
+ hwc->idx = PIC_NO_INDEX;
+
+ /* Try to do all error checking before this point, as unwinding
+ * state after grabbing the PMC is difficult.
+ */
+ perf_event_grab_pmc();
+ event->destroy = hw_perf_event_destroy;
+
+ if (!hwc->sample_period) {
+ hwc->sample_period = MAX_PERIOD;
+ hwc->last_period = hwc->sample_period;
+ local64_set(&hwc->period_left, hwc->sample_period);
+ }
+
+ return 0;
+}
+
+/*
+ * Start group events scheduling transaction
+ * Set the flag to make pmu::enable() not perform the
+ * schedulability test, it will be performed at commit time
+ */
+static void sparc_pmu_start_txn(struct pmu *pmu)
+{
+ struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events);
+
+ perf_pmu_disable(pmu);
+ cpuhw->group_flag |= PERF_EVENT_TXN;
+}
+
+/*
+ * Stop group events scheduling transaction
+ * Clear the flag and pmu::enable() will perform the
+ * schedulability test.
+ */
+static void sparc_pmu_cancel_txn(struct pmu *pmu)
+{
+ struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events);
+
+ cpuhw->group_flag &= ~PERF_EVENT_TXN;
+ perf_pmu_enable(pmu);
+}
+
+/*
+ * Commit group events scheduling transaction
+ * Perform the group schedulability test as a whole
+ * Return 0 if success
+ */
+static int sparc_pmu_commit_txn(struct pmu *pmu)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ int n;
+
+ if (!sparc_pmu)
+ return -EINVAL;
+
+ cpuc = &__get_cpu_var(cpu_hw_events);
+ n = cpuc->n_events;
+ if (check_excludes(cpuc->event, 0, n))
+ return -EINVAL;
+ if (sparc_check_constraints(cpuc->event, cpuc->events, n))
+ return -EAGAIN;
+
+ cpuc->group_flag &= ~PERF_EVENT_TXN;
+ perf_pmu_enable(pmu);
+ return 0;
+}
+
+static struct pmu pmu = {
+ .pmu_enable = sparc_pmu_enable,
+ .pmu_disable = sparc_pmu_disable,
+ .event_init = sparc_pmu_event_init,
+ .add = sparc_pmu_add,
+ .del = sparc_pmu_del,
+ .start = sparc_pmu_start,
+ .stop = sparc_pmu_stop,
+ .read = sparc_pmu_read,
+ .start_txn = sparc_pmu_start_txn,
+ .cancel_txn = sparc_pmu_cancel_txn,
+ .commit_txn = sparc_pmu_commit_txn,
+};
+
+void perf_event_print_debug(void)
+{
+ unsigned long flags;
+ int cpu, i;
+
+ if (!sparc_pmu)
+ return;
+
+ local_irq_save(flags);
+
+ cpu = smp_processor_id();
+
+ pr_info("\n");
+ for (i = 0; i < sparc_pmu->num_pcrs; i++)
+ pr_info("CPU#%d: PCR%d[%016llx]\n",
+ cpu, i, pcr_ops->read_pcr(i));
+ for (i = 0; i < sparc_pmu->num_pic_regs; i++)
+ pr_info("CPU#%d: PIC%d[%016llx]\n",
+ cpu, i, pcr_ops->read_pic(i));
+
+ local_irq_restore(flags);
+}
+
+static int __kprobes perf_event_nmi_handler(struct notifier_block *self,
+ unsigned long cmd, void *__args)
+{
+ struct die_args *args = __args;
+ struct perf_sample_data data;
+ struct cpu_hw_events *cpuc;
+ struct pt_regs *regs;
+ int i;
+
+ if (!atomic_read(&active_events))
+ return NOTIFY_DONE;
+
+ switch (cmd) {
+ case DIE_NMI:
+ break;
+
+ default:
+ return NOTIFY_DONE;
+ }
+
+ regs = args->regs;
+
+ cpuc = &__get_cpu_var(cpu_hw_events);
+
+ /* If the PMU has the TOE IRQ enable bits, we need to do a
+ * dummy write to the %pcr to clear the overflow bits and thus
+ * the interrupt.
+ *
+ * Do this before we peek at the counters to determine
+ * overflow so we don't lose any events.
+ */
+ if (sparc_pmu->irq_bit &&
+ sparc_pmu->num_pcrs == 1)
+ pcr_ops->write_pcr(0, cpuc->pcr[0]);
+
+ for (i = 0; i < cpuc->n_events; i++) {
+ struct perf_event *event = cpuc->event[i];
+ int idx = cpuc->current_idx[i];
+ struct hw_perf_event *hwc;
+ u64 val;
+
+ if (sparc_pmu->irq_bit &&
+ sparc_pmu->num_pcrs > 1)
+ pcr_ops->write_pcr(idx, cpuc->pcr[idx]);
+
+ hwc = &event->hw;
+ val = sparc_perf_event_update(event, hwc, idx);
+ if (val & (1ULL << 31))
+ continue;
+
+ perf_sample_data_init(&data, 0, hwc->last_period);
+ if (!sparc_perf_event_set_period(event, hwc, idx))
+ continue;
+
+ if (perf_event_overflow(event, &data, regs))
+ sparc_pmu_stop(event, 0);
+ }
+
+ return NOTIFY_STOP;
+}
+
+static __read_mostly struct notifier_block perf_event_nmi_notifier = {
+ .notifier_call = perf_event_nmi_handler,
+};
+
+static bool __init supported_pmu(void)
+{
+ if (!strcmp(sparc_pmu_type, "ultra3") ||
+ !strcmp(sparc_pmu_type, "ultra3+") ||
+ !strcmp(sparc_pmu_type, "ultra3i") ||
+ !strcmp(sparc_pmu_type, "ultra4+")) {
+ sparc_pmu = &ultra3_pmu;
+ return true;
+ }
+ if (!strcmp(sparc_pmu_type, "niagara")) {
+ sparc_pmu = &niagara1_pmu;
+ return true;
+ }
+ if (!strcmp(sparc_pmu_type, "niagara2") ||
+ !strcmp(sparc_pmu_type, "niagara3")) {
+ sparc_pmu = &niagara2_pmu;
+ return true;
+ }
+ if (!strcmp(sparc_pmu_type, "niagara4")) {
+ sparc_pmu = &niagara4_pmu;
+ return true;
+ }
+ return false;
+}
+
+static int __init init_hw_perf_events(void)
+{
+ pr_info("Performance events: ");
+
+ if (!supported_pmu()) {
+ pr_cont("No support for PMU type '%s'\n", sparc_pmu_type);
+ return 0;
+ }
+
+ pr_cont("Supported PMU type is '%s'\n", sparc_pmu_type);
+
+ perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
+ register_die_notifier(&perf_event_nmi_notifier);
+
+ return 0;
+}
+early_initcall(init_hw_perf_events);
+
+void perf_callchain_kernel(struct perf_callchain_entry *entry,
+ struct pt_regs *regs)
+{
+ unsigned long ksp, fp;
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+ int graph = 0;
+#endif
+
+ stack_trace_flush();
+
+ perf_callchain_store(entry, regs->tpc);
+
+ ksp = regs->u_regs[UREG_I6];
+ fp = ksp + STACK_BIAS;
+ do {
+ struct sparc_stackf *sf;
+ struct pt_regs *regs;
+ unsigned long pc;
+
+ if (!kstack_valid(current_thread_info(), fp))
+ break;
+
+ sf = (struct sparc_stackf *) fp;
+ regs = (struct pt_regs *) (sf + 1);
+
+ if (kstack_is_trap_frame(current_thread_info(), regs)) {
+ if (user_mode(regs))
+ break;
+ pc = regs->tpc;
+ fp = regs->u_regs[UREG_I6] + STACK_BIAS;
+ } else {
+ pc = sf->callers_pc;
+ fp = (unsigned long)sf->fp + STACK_BIAS;
+ }
+ perf_callchain_store(entry, pc);
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+ if ((pc + 8UL) == (unsigned long) &return_to_handler) {
+ int index = current->curr_ret_stack;
+ if (current->ret_stack && index >= graph) {
+ pc = current->ret_stack[index - graph].ret;
+ perf_callchain_store(entry, pc);
+ graph++;
+ }
+ }
+#endif
+ } while (entry->nr < PERF_MAX_STACK_DEPTH);
+}
+
+static void perf_callchain_user_64(struct perf_callchain_entry *entry,
+ struct pt_regs *regs)
+{
+ unsigned long ufp;
+
+ ufp = regs->u_regs[UREG_I6] + STACK_BIAS;
+ do {
+ struct sparc_stackf __user *usf;
+ struct sparc_stackf sf;
+ unsigned long pc;
+
+ usf = (struct sparc_stackf __user *)ufp;
+ if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
+ break;
+
+ pc = sf.callers_pc;
+ ufp = (unsigned long)sf.fp + STACK_BIAS;
+ perf_callchain_store(entry, pc);
+ } while (entry->nr < PERF_MAX_STACK_DEPTH);
+}
+
+static void perf_callchain_user_32(struct perf_callchain_entry *entry,
+ struct pt_regs *regs)
+{
+ unsigned long ufp;
+
+ ufp = regs->u_regs[UREG_I6] & 0xffffffffUL;
+ do {
+ unsigned long pc;
+
+ if (thread32_stack_is_64bit(ufp)) {
+ struct sparc_stackf __user *usf;
+ struct sparc_stackf sf;
+
+ ufp += STACK_BIAS;
+ usf = (struct sparc_stackf __user *)ufp;
+ if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
+ break;
+ pc = sf.callers_pc & 0xffffffff;
+ ufp = ((unsigned long) sf.fp) & 0xffffffff;
+ } else {
+ struct sparc_stackf32 __user *usf;
+ struct sparc_stackf32 sf;
+ usf = (struct sparc_stackf32 __user *)ufp;
+ if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
+ break;
+ pc = sf.callers_pc;
+ ufp = (unsigned long)sf.fp;
+ }
+ perf_callchain_store(entry, pc);
+ } while (entry->nr < PERF_MAX_STACK_DEPTH);
+}
+
+void
+perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs)
+{
+ perf_callchain_store(entry, regs->tpc);
+
+ if (!current->mm)
+ return;
+
+ flushw_user();
+ if (test_thread_flag(TIF_32BIT))
+ perf_callchain_user_32(entry, regs);
+ else
+ perf_callchain_user_64(entry, regs);
+}
diff --git a/arch/sparc/kernel/pmc.c b/arch/sparc/kernel/pmc.c
index 7eca8871ff4..8b7297faca7 100644
--- a/arch/sparc/kernel/pmc.c
+++ b/arch/sparc/kernel/pmc.c
@@ -8,14 +8,16 @@
#include <linux/fs.h>
#include <linux/errno.h>
#include <linux/init.h>
-#include <linux/miscdevice.h>
#include <linux/pm.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/module.h>
#include <asm/io.h>
-#include <asm/sbus.h>
#include <asm/oplib.h>
#include <asm/uaccess.h>
#include <asm/auxio.h>
+#include <asm/processor.h>
/* Debug
*
@@ -23,61 +25,38 @@
* #define PMC_NO_IDLE
*/
-#define PMC_MINOR MISC_DYNAMIC_MINOR
#define PMC_OBPNAME "SUNW,pmc"
-#define PMC_DEVNAME "pmc"
+#define PMC_DEVNAME "pmc"
#define PMC_IDLE_REG 0x00
-#define PMC_IDLE_ON 0x01
+#define PMC_IDLE_ON 0x01
-volatile static u8 __iomem *regs;
-static int pmc_regsize;
+static u8 __iomem *regs;
-#define pmc_readb(offs) (sbus_readb(regs+offs))
-#define pmc_writeb(val, offs) (sbus_writeb(val, regs+offs))
+#define pmc_readb(offs) (sbus_readb(regs+offs))
+#define pmc_writeb(val, offs) (sbus_writeb(val, regs+offs))
-/*
+/*
* CPU idle callback function
* See .../arch/sparc/kernel/process.c
*/
-void pmc_swift_idle(void)
+static void pmc_swift_idle(void)
{
#ifdef PMC_DEBUG_LED
- set_auxio(0x00, AUXIO_LED);
+ set_auxio(0x00, AUXIO_LED);
#endif
pmc_writeb(pmc_readb(PMC_IDLE_REG) | PMC_IDLE_ON, PMC_IDLE_REG);
#ifdef PMC_DEBUG_LED
- set_auxio(AUXIO_LED, 0x00);
+ set_auxio(AUXIO_LED, 0x00);
#endif
-}
-
-static inline void pmc_free(void)
-{
- sbus_iounmap(regs, pmc_regsize);
}
-static int __init pmc_probe(void)
+static int pmc_probe(struct platform_device *op)
{
- struct sbus_bus *sbus = NULL;
- struct sbus_dev *sdev = NULL;
- for_each_sbus(sbus) {
- for_each_sbusdev(sdev, sbus) {
- if (!strcmp(sdev->prom_name, PMC_OBPNAME)) {
- goto sbus_done;
- }
- }
- }
-
-sbus_done:
- if (!sdev) {
- return -ENODEV;
- }
-
- pmc_regsize = sdev->reg_addrs[0].reg_size;
- regs = sbus_ioremap(&sdev->resource[0], 0,
- pmc_regsize, PMC_OBPNAME);
+ regs = of_ioremap(&op->resource[0], 0,
+ resource_size(&op->resource[0]), PMC_OBPNAME);
if (!regs) {
printk(KERN_ERR "%s: unable to map registers\n", PMC_DEVNAME);
return -ENODEV;
@@ -85,15 +64,37 @@ sbus_done:
#ifndef PMC_NO_IDLE
/* Assign power management IDLE handler */
- pm_idle = pmc_swift_idle;
+ sparc_idle = pmc_swift_idle;
#endif
printk(KERN_INFO "%s: power management initialized\n", PMC_DEVNAME);
return 0;
}
+static struct of_device_id pmc_match[] = {
+ {
+ .name = PMC_OBPNAME,
+ },
+ {},
+};
+MODULE_DEVICE_TABLE(of, pmc_match);
+
+static struct platform_driver pmc_driver = {
+ .driver = {
+ .name = "pmc",
+ .owner = THIS_MODULE,
+ .of_match_table = pmc_match,
+ },
+ .probe = pmc_probe,
+};
+
+static int __init pmc_init(void)
+{
+ return platform_driver_register(&pmc_driver);
+}
+
/* This driver is not critical to the boot process
* and is easiest to ioremap when SBus is already
* initialized, so we install ourselves thusly:
*/
-__initcall(pmc_probe);
+__initcall(pmc_init);
diff --git a/arch/sparc/kernel/power.c b/arch/sparc/kernel/power.c
new file mode 100644
index 00000000000..4cb23c41553
--- /dev/null
+++ b/arch/sparc/kernel/power.c
@@ -0,0 +1,76 @@
+/* power.c: Power management driver.
+ *
+ * Copyright (C) 1999, 2007, 2008 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/kernel.h>
+#include <linux/export.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/reboot.h>
+#include <linux/of_device.h>
+
+#include <asm/prom.h>
+#include <asm/io.h>
+
+static void __iomem *power_reg;
+
+static irqreturn_t power_handler(int irq, void *dev_id)
+{
+ orderly_poweroff(true);
+
+ /* FIXME: Check registers for status... */
+ return IRQ_HANDLED;
+}
+
+static int has_button_interrupt(unsigned int irq, struct device_node *dp)
+{
+ if (irq == 0xffffffff)
+ return 0;
+ if (!of_find_property(dp, "button", NULL))
+ return 0;
+
+ return 1;
+}
+
+static int power_probe(struct platform_device *op)
+{
+ struct resource *res = &op->resource[0];
+ unsigned int irq = op->archdata.irqs[0];
+
+ power_reg = of_ioremap(res, 0, 0x4, "power");
+
+ printk(KERN_INFO "%s: Control reg at %llx\n",
+ op->dev.of_node->name, res->start);
+
+ if (has_button_interrupt(irq, op->dev.of_node)) {
+ if (request_irq(irq,
+ power_handler, 0, "power", NULL) < 0)
+ printk(KERN_ERR "power: Cannot setup IRQ handler.\n");
+ }
+
+ return 0;
+}
+
+static const struct of_device_id power_match[] = {
+ {
+ .name = "power",
+ },
+ {},
+};
+
+static struct platform_driver power_driver = {
+ .probe = power_probe,
+ .driver = {
+ .name = "power",
+ .owner = THIS_MODULE,
+ .of_match_table = power_match,
+ },
+};
+
+static int __init power_init(void)
+{
+ return platform_driver_register(&power_driver);
+}
+
+device_initcall(power_init);
diff --git a/arch/sparc/kernel/process.c b/arch/sparc/kernel/process_32.c
index 0bd69d0b5cd..50e7b626afe 100644
--- a/arch/sparc/kernel/process.c
+++ b/arch/sparc/kernel/process_32.c
@@ -1,6 +1,6 @@
/* linux/arch/sparc/kernel/process.c
*
- * Copyright (C) 1995 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1995, 2008 David S. Miller (davem@davemloft.net)
* Copyright (C) 1996 Eddie C. Dost (ecd@skynet.be)
*/
@@ -10,26 +10,25 @@
#include <stdarg.h>
+#include <linux/elfcore.h>
#include <linux/errno.h>
#include <linux/module.h>
#include <linux/sched.h>
#include <linux/kernel.h>
-#include <linux/kallsyms.h>
#include <linux/mm.h>
#include <linux/stddef.h>
#include <linux/ptrace.h>
-#include <linux/slab.h>
#include <linux/user.h>
#include <linux/smp.h>
#include <linux/reboot.h>
#include <linux/delay.h>
#include <linux/pm.h>
-#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/cpu.h>
#include <asm/auxio.h>
#include <asm/oplib.h>
#include <asm/uaccess.h>
-#include <asm/system.h>
#include <asm/page.h>
#include <asm/pgalloc.h>
#include <asm/pgtable.h>
@@ -39,12 +38,15 @@
#include <asm/elf.h>
#include <asm/prom.h>
#include <asm/unistd.h>
+#include <asm/setup.h>
+
+#include "kernel.h"
/*
* Power management idle function
* Set in pm platform drivers (apc.c and pmc.c)
*/
-void (*pm_idle)(void);
+void (*sparc_idle)(void);
/*
* Power-off handler instantiation for pm.h compliance
@@ -65,82 +67,14 @@ extern void fpsave(unsigned long *, unsigned long *, void *, unsigned long *);
struct task_struct *last_task_used_math = NULL;
struct thread_info *current_set[NR_CPUS];
-#ifndef CONFIG_SMP
-
-#define SUN4C_FAULT_HIGH 100
-
-/*
- * the idle loop on a Sparc... ;)
- */
-void cpu_idle(void)
-{
- /* endless idle loop with no priority at all */
- for (;;) {
- if (ARCH_SUN4C_SUN4) {
- static int count = HZ;
- static unsigned long last_jiffies;
- static unsigned long last_faults;
- static unsigned long fps;
- unsigned long now;
- unsigned long faults;
-
- extern unsigned long sun4c_kernel_faults;
- extern void sun4c_grow_kernel_ring(void);
-
- local_irq_disable();
- now = jiffies;
- count -= (now - last_jiffies);
- last_jiffies = now;
- if (count < 0) {
- count += HZ;
- faults = sun4c_kernel_faults;
- fps = (fps + (faults - last_faults)) >> 1;
- last_faults = faults;
-#if 0
- printk("kernel faults / second = %ld\n", fps);
-#endif
- if (fps >= SUN4C_FAULT_HIGH) {
- sun4c_grow_kernel_ring();
- }
- }
- local_irq_enable();
- }
-
- if (pm_idle) {
- while (!need_resched())
- (*pm_idle)();
- } else {
- while (!need_resched())
- cpu_relax();
- }
- preempt_enable_no_resched();
- schedule();
- preempt_disable();
- check_pgt_cache();
- }
-}
-
-#else
-
-/* This is being executed in task 0 'user space'. */
-void cpu_idle(void)
+/* Idle loop support. */
+void arch_cpu_idle(void)
{
- set_thread_flag(TIF_POLLING_NRFLAG);
- /* endless idle loop with no priority at all */
- while(1) {
- while (!need_resched())
- cpu_relax();
- preempt_enable_no_resched();
- schedule();
- preempt_disable();
- check_pgt_cache();
- }
+ if (sparc_idle)
+ (*sparc_idle)();
+ local_irq_enable();
}
-#endif
-
-extern char reboot_command [];
-
/* XXX cli/sti -> local_irq_xxx here, check this works once SMP is fixed. */
void machine_halt(void)
{
@@ -171,109 +105,32 @@ void machine_restart(char * cmd)
void machine_power_off(void)
{
-#ifdef CONFIG_SUN_AUXIO
if (auxio_power_register &&
- (strcmp(of_console_device->type, "serial") || scons_pwroff))
- *auxio_power_register |= AUXIO_POWER_OFF;
-#endif
- machine_halt();
-}
-
-static DEFINE_SPINLOCK(sparc_backtrace_lock);
-
-void __show_backtrace(unsigned long fp)
-{
- struct reg_window *rw;
- unsigned long flags;
- int cpu = smp_processor_id();
-
- spin_lock_irqsave(&sparc_backtrace_lock, flags);
-
- rw = (struct reg_window *)fp;
- while(rw && (((unsigned long) rw) >= PAGE_OFFSET) &&
- !(((unsigned long) rw) & 0x7)) {
- printk("CPU[%d]: ARGS[%08lx,%08lx,%08lx,%08lx,%08lx,%08lx] "
- "FP[%08lx] CALLER[%08lx]: ", cpu,
- rw->ins[0], rw->ins[1], rw->ins[2], rw->ins[3],
- rw->ins[4], rw->ins[5],
- rw->ins[6],
- rw->ins[7]);
- print_symbol("%s\n", rw->ins[7]);
- rw = (struct reg_window *) rw->ins[6];
+ (strcmp(of_console_device->type, "serial") || scons_pwroff)) {
+ u8 power_register = sbus_readb(auxio_power_register);
+ power_register |= AUXIO_POWER_OFF;
+ sbus_writeb(power_register, auxio_power_register);
}
- spin_unlock_irqrestore(&sparc_backtrace_lock, flags);
-}
-
-#define __SAVE __asm__ __volatile__("save %sp, -0x40, %sp\n\t")
-#define __RESTORE __asm__ __volatile__("restore %g0, %g0, %g0\n\t")
-#define __GET_FP(fp) __asm__ __volatile__("mov %%i6, %0" : "=r" (fp))
-
-void show_backtrace(void)
-{
- unsigned long fp;
-
- __SAVE; __SAVE; __SAVE; __SAVE;
- __SAVE; __SAVE; __SAVE; __SAVE;
- __RESTORE; __RESTORE; __RESTORE; __RESTORE;
- __RESTORE; __RESTORE; __RESTORE; __RESTORE;
-
- __GET_FP(fp);
- __show_backtrace(fp);
-}
-
-#ifdef CONFIG_SMP
-void smp_show_backtrace_all_cpus(void)
-{
- xc0((smpfunc_t) show_backtrace);
- show_backtrace();
-}
-#endif
-
-#if 0
-void show_stackframe(struct sparc_stackf *sf)
-{
- unsigned long size;
- unsigned long *stk;
- int i;
-
- printk("l0: %08lx l1: %08lx l2: %08lx l3: %08lx "
- "l4: %08lx l5: %08lx l6: %08lx l7: %08lx\n",
- sf->locals[0], sf->locals[1], sf->locals[2], sf->locals[3],
- sf->locals[4], sf->locals[5], sf->locals[6], sf->locals[7]);
- printk("i0: %08lx i1: %08lx i2: %08lx i3: %08lx "
- "i4: %08lx i5: %08lx fp: %08lx i7: %08lx\n",
- sf->ins[0], sf->ins[1], sf->ins[2], sf->ins[3],
- sf->ins[4], sf->ins[5], (unsigned long)sf->fp, sf->callers_pc);
- printk("sp: %08lx x0: %08lx x1: %08lx x2: %08lx "
- "x3: %08lx x4: %08lx x5: %08lx xx: %08lx\n",
- (unsigned long)sf->structptr, sf->xargs[0], sf->xargs[1],
- sf->xargs[2], sf->xargs[3], sf->xargs[4], sf->xargs[5],
- sf->xxargs[0]);
- size = ((unsigned long)sf->fp) - ((unsigned long)sf);
- size -= STACKFRAME_SZ;
- stk = (unsigned long *)((unsigned long)sf + STACKFRAME_SZ);
- i = 0;
- do {
- printk("s%d: %08lx\n", i++, *stk++);
- } while ((size -= sizeof(unsigned long)));
+ machine_halt();
}
-#endif
void show_regs(struct pt_regs *r)
{
- struct reg_window *rw = (struct reg_window *) r->u_regs[14];
+ struct reg_window32 *rw = (struct reg_window32 *) r->u_regs[14];
+
+ show_regs_print_info(KERN_DEFAULT);
printk("PSR: %08lx PC: %08lx NPC: %08lx Y: %08lx %s\n",
r->psr, r->pc, r->npc, r->y, print_tainted());
- print_symbol("PC: <%s>\n", r->pc);
+ printk("PC: <%pS>\n", (void *) r->pc);
printk("%%G: %08lx %08lx %08lx %08lx %08lx %08lx %08lx %08lx\n",
r->u_regs[0], r->u_regs[1], r->u_regs[2], r->u_regs[3],
r->u_regs[4], r->u_regs[5], r->u_regs[6], r->u_regs[7]);
printk("%%O: %08lx %08lx %08lx %08lx %08lx %08lx %08lx %08lx\n",
r->u_regs[8], r->u_regs[9], r->u_regs[10], r->u_regs[11],
r->u_regs[12], r->u_regs[13], r->u_regs[14], r->u_regs[15]);
- print_symbol("RPC: <%s>\n", r->u_regs[15]);
+ printk("RPC: <%pS>\n", (void *) r->u_regs[15]);
printk("%%L: %08lx %08lx %08lx %08lx %08lx %08lx %08lx %08lx\n",
rw->locals[0], rw->locals[1], rw->locals[2], rw->locals[3],
@@ -291,40 +148,31 @@ void show_stack(struct task_struct *tsk, unsigned long *_ksp)
{
unsigned long pc, fp;
unsigned long task_base;
- struct reg_window *rw;
+ struct reg_window32 *rw;
int count = 0;
- if (tsk != NULL)
- task_base = (unsigned long) task_stack_page(tsk);
- else
- task_base = (unsigned long) current_thread_info();
+ if (!tsk)
+ tsk = current;
+
+ if (tsk == current && !_ksp)
+ __asm__ __volatile__("mov %%fp, %0" : "=r" (_ksp));
+ task_base = (unsigned long) task_stack_page(tsk);
fp = (unsigned long) _ksp;
do {
/* Bogus frame pointer? */
if (fp < (task_base + sizeof(struct thread_info)) ||
fp >= (task_base + (PAGE_SIZE << 1)))
break;
- rw = (struct reg_window *) fp;
+ rw = (struct reg_window32 *) fp;
pc = rw->ins[7];
printk("[%08lx : ", pc);
- print_symbol("%s ] ", pc);
+ printk("%pS ] ", (void *) pc);
fp = rw->ins[6];
} while (++count < 16);
printk("\n");
}
-void dump_stack(void)
-{
- unsigned long *ksp;
-
- __asm__ __volatile__("mov %%fp, %0"
- : "=r" (ksp));
- show_stack(current, ksp);
-}
-
-EXPORT_SYMBOL(dump_stack);
-
/*
* Note: sparc64 has a pretty intricated thread_saved_pc, check it out.
*/
@@ -359,8 +207,6 @@ void flush_thread(void)
{
current_thread_info()->w_saved = 0;
- /* No new signal delivery by default */
- current->thread.new_signal = 0;
#ifndef CONFIG_SMP
if(last_task_used_math == current) {
#else
@@ -377,8 +223,7 @@ void flush_thread(void)
#endif
}
- /* Now, this task is no longer a kernel thread. */
- current->thread.current_ds = USER_DS;
+ /* This task is no longer a kernel thread. */
if (current->thread.flags & SPARC_FLAG_KTHREAD) {
current->thread.flags &= ~SPARC_FLAG_KTHREAD;
@@ -423,14 +268,25 @@ asmlinkage int sparc_do_fork(unsigned long clone_flags,
unsigned long stack_size)
{
unsigned long parent_tid_ptr, child_tid_ptr;
+ unsigned long orig_i1 = regs->u_regs[UREG_I1];
+ long ret;
parent_tid_ptr = regs->u_regs[UREG_I2];
child_tid_ptr = regs->u_regs[UREG_I4];
- return do_fork(clone_flags, stack_start,
- regs, stack_size,
- (int __user *) parent_tid_ptr,
- (int __user *) child_tid_ptr);
+ ret = do_fork(clone_flags, stack_start, stack_size,
+ (int __user *) parent_tid_ptr,
+ (int __user *) child_tid_ptr);
+
+ /* If we get an error and potentially restart the system
+ * call, we're screwed because copy_thread() clobbered
+ * the parent's %o1. So detect that case and restore it
+ * here.
+ */
+ if ((unsigned long)ret >= -ERESTART_RESTARTBLOCK)
+ regs->u_regs[UREG_I1] = orig_i1;
+
+ return ret;
}
/* Copy a Sparc thread. The fork() return value conventions
@@ -447,13 +303,13 @@ asmlinkage int sparc_do_fork(unsigned long clone_flags,
* XXX See comment above sys_vfork in sparc64. todo.
*/
extern void ret_from_fork(void);
+extern void ret_from_kernel_thread(void);
-int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
- unsigned long unused,
- struct task_struct *p, struct pt_regs *regs)
+int copy_thread(unsigned long clone_flags, unsigned long sp,
+ unsigned long arg, struct task_struct *p)
{
struct thread_info *ti = task_thread_info(p);
- struct pt_regs *childregs;
+ struct pt_regs *childregs, *regs = current_pt_regs();
char *new_stack;
#ifndef CONFIG_SMP
@@ -464,22 +320,16 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
put_psr(get_psr() | PSR_EF);
fpsave(&p->thread.float_regs[0], &p->thread.fsr,
&p->thread.fpqueue[0], &p->thread.fpqdepth);
-#ifdef CONFIG_SMP
- clear_thread_flag(TIF_USEDFPU);
-#endif
}
/*
- * p->thread_info new_stack childregs
- * ! ! ! {if(PSR_PS) }
- * V V (stk.fr.) V (pt_regs) { (stk.fr.) }
- * +----- - - - - - ------+===========+============={+==========}+
+ * p->thread_info new_stack childregs stack bottom
+ * ! ! ! !
+ * V V (stk.fr.) V (pt_regs) V
+ * +----- - - - - - ------+===========+=============+
*/
new_stack = task_stack_page(p) + THREAD_SIZE;
- if (regs->psr & PSR_PS)
- new_stack -= STACKFRAME_SZ;
new_stack -= STACKFRAME_SZ + TRACEREG_SZ;
- memcpy(new_stack, (char *)regs - STACKFRAME_SZ, STACKFRAME_SZ + TRACEREG_SZ);
childregs = (struct pt_regs *) (new_stack + STACKFRAME_SZ);
/*
@@ -490,60 +340,64 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
* Thus, kpsr|=PSR_PIL.
*/
ti->ksp = (unsigned long) new_stack;
+ p->thread.kregs = childregs;
+
+ if (unlikely(p->flags & PF_KTHREAD)) {
+ extern int nwindows;
+ unsigned long psr;
+ memset(new_stack, 0, STACKFRAME_SZ + TRACEREG_SZ);
+ p->thread.flags |= SPARC_FLAG_KTHREAD;
+ p->thread.current_ds = KERNEL_DS;
+ ti->kpc = (((unsigned long) ret_from_kernel_thread) - 0x8);
+ childregs->u_regs[UREG_G1] = sp; /* function */
+ childregs->u_regs[UREG_G2] = arg;
+ psr = childregs->psr = get_psr();
+ ti->kpsr = psr | PSR_PIL;
+ ti->kwim = 1 << (((psr & PSR_CWP) + 1) % nwindows);
+ return 0;
+ }
+ memcpy(new_stack, (char *)regs - STACKFRAME_SZ, STACKFRAME_SZ + TRACEREG_SZ);
+ childregs->u_regs[UREG_FP] = sp;
+ p->thread.flags &= ~SPARC_FLAG_KTHREAD;
+ p->thread.current_ds = USER_DS;
ti->kpc = (((unsigned long) ret_from_fork) - 0x8);
ti->kpsr = current->thread.fork_kpsr | PSR_PIL;
ti->kwim = current->thread.fork_kwim;
- if(regs->psr & PSR_PS) {
- extern struct pt_regs fake_swapper_regs;
+ if (sp != regs->u_regs[UREG_FP]) {
+ struct sparc_stackf __user *childstack;
+ struct sparc_stackf __user *parentstack;
- p->thread.kregs = &fake_swapper_regs;
- new_stack += STACKFRAME_SZ + TRACEREG_SZ;
- childregs->u_regs[UREG_FP] = (unsigned long) new_stack;
- p->thread.flags |= SPARC_FLAG_KTHREAD;
- p->thread.current_ds = KERNEL_DS;
- memcpy(new_stack, (void *)regs->u_regs[UREG_FP], STACKFRAME_SZ);
- childregs->u_regs[UREG_G6] = (unsigned long) ti;
- } else {
- p->thread.kregs = childregs;
- childregs->u_regs[UREG_FP] = sp;
- p->thread.flags &= ~SPARC_FLAG_KTHREAD;
- p->thread.current_ds = USER_DS;
-
- if (sp != regs->u_regs[UREG_FP]) {
- struct sparc_stackf __user *childstack;
- struct sparc_stackf __user *parentstack;
-
- /*
- * This is a clone() call with supplied user stack.
- * Set some valid stack frames to give to the child.
- */
- childstack = (struct sparc_stackf __user *)
- (sp & ~0x7UL);
- parentstack = (struct sparc_stackf __user *)
- regs->u_regs[UREG_FP];
+ /*
+ * This is a clone() call with supplied user stack.
+ * Set some valid stack frames to give to the child.
+ */
+ childstack = (struct sparc_stackf __user *)
+ (sp & ~0xfUL);
+ parentstack = (struct sparc_stackf __user *)
+ regs->u_regs[UREG_FP];
#if 0
- printk("clone: parent stack:\n");
- show_stackframe(parentstack);
+ printk("clone: parent stack:\n");
+ show_stackframe(parentstack);
#endif
- childstack = clone_stackframe(childstack, parentstack);
- if (!childstack)
- return -EFAULT;
+ childstack = clone_stackframe(childstack, parentstack);
+ if (!childstack)
+ return -EFAULT;
#if 0
- printk("clone: child stack:\n");
- show_stackframe(childstack);
+ printk("clone: child stack:\n");
+ show_stackframe(childstack);
#endif
- childregs->u_regs[UREG_FP] = (unsigned long)childstack;
- }
+ childregs->u_regs[UREG_FP] = (unsigned long)childstack;
}
#ifdef CONFIG_SMP
/* FPU must be disabled on SMP. */
childregs->psr &= ~PSR_EF;
+ clear_tsk_thread_flag(p, TIF_USEDFPU);
#endif
/* Set the return value for the child. */
@@ -608,77 +462,12 @@ int dump_fpu (struct pt_regs * regs, elf_fpregset_t * fpregs)
return 1;
}
-/*
- * sparc_execve() executes a new program after the asm stub has set
- * things up for us. This should basically do what I want it to.
- */
-asmlinkage int sparc_execve(struct pt_regs *regs)
-{
- int error, base = 0;
- char *filename;
-
- /* Check for indirect call. */
- if(regs->u_regs[UREG_G1] == 0)
- base = 1;
-
- filename = getname((char __user *)regs->u_regs[base + UREG_I0]);
- error = PTR_ERR(filename);
- if(IS_ERR(filename))
- goto out;
- error = do_execve(filename,
- (char __user * __user *)regs->u_regs[base + UREG_I1],
- (char __user * __user *)regs->u_regs[base + UREG_I2],
- regs);
- putname(filename);
- if (error == 0) {
- task_lock(current);
- current->ptrace &= ~PT_DTRACE;
- task_unlock(current);
- }
-out:
- return error;
-}
-
-/*
- * This is the mechanism for creating a new kernel thread.
- *
- * NOTE! Only a kernel-only process(ie the swapper or direct descendants
- * who haven't done an "execve()") should use this: it will work within
- * a system call from a "real" process, but the process memory space will
- * not be freed until both the parent and the child have exited.
- */
-pid_t kernel_thread(int (*fn)(void *), void * arg, unsigned long flags)
-{
- long retval;
-
- __asm__ __volatile__("mov %4, %%g2\n\t" /* Set aside fn ptr... */
- "mov %5, %%g3\n\t" /* and arg. */
- "mov %1, %%g1\n\t"
- "mov %2, %%o0\n\t" /* Clone flags. */
- "mov 0, %%o1\n\t" /* usp arg == 0 */
- "t 0x10\n\t" /* Linux/Sparc clone(). */
- "cmp %%o1, 0\n\t"
- "be 1f\n\t" /* The parent, just return. */
- " nop\n\t" /* Delay slot. */
- "jmpl %%g2, %%o7\n\t" /* Call the function. */
- " mov %%g3, %%o0\n\t" /* Get back the arg in delay. */
- "mov %3, %%g1\n\t"
- "t 0x10\n\t" /* Linux/Sparc exit(). */
- /* Notreached by child. */
- "1: mov %%o0, %0\n\t" :
- "=r" (retval) :
- "i" (__NR_clone), "r" (flags | CLONE_VM | CLONE_UNTRACED),
- "i" (__NR_exit), "r" (fn), "r" (arg) :
- "g1", "g2", "g3", "o0", "o1", "memory", "cc");
- return retval;
-}
-
unsigned long get_wchan(struct task_struct *task)
{
unsigned long pc, fp, bias = 0;
unsigned long task_base = (unsigned long) task;
unsigned long ret = 0;
- struct reg_window *rw;
+ struct reg_window32 *rw;
int count = 0;
if (!task || task == current ||
@@ -691,7 +480,7 @@ unsigned long get_wchan(struct task_struct *task)
if (fp < (task_base + sizeof(struct thread_info)) ||
fp >= (task_base + (2 * PAGE_SIZE)))
break;
- rw = (struct reg_window *) fp;
+ rw = (struct reg_window32 *) fp;
pc = rw->ins[7];
if (!in_sched_functions(pc)) {
ret = pc;
diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c
new file mode 100644