19 files changed, 1202 insertions, 375 deletions

diff --git a/arch/sh/kernel/cpu/Makefile b/arch/sh/kernel/cpu/Makefile
index 3d6b9312dc4..d97c803719e 100644
--- a/arch/sh/kernel/cpu/Makefile
+++ b/arch/sh/kernel/cpu/Makefile
@@ -15,7 +15,6 @@ obj-$(CONFIG_ARCH_SHMOBILE)	+= shmobile/
 
 # Common interfaces.
 
-obj-$(CONFIG_UBC_WAKEUP)	+= ubc.o
 obj-$(CONFIG_SH_ADC)		+= adc.o
 obj-$(CONFIG_SH_CLK_CPG)	+= clock-cpg.o
 
diff --git a/arch/sh/kernel/cpu/init.c b/arch/sh/kernel/cpu/init.c
index e932ebef473..89b4b76c0d7 100644
--- a/arch/sh/kernel/cpu/init.c
+++ b/arch/sh/kernel/cpu/init.c
@@ -75,16 +75,11 @@ static void __init expmask_init(void)
 	/*
 	 * Future proofing.
 	 *
-	 * Disable support for slottable sleep instruction
-	 * and non-nop instructions in the rte delay slot.
+	 * Disable support for slottable sleep instruction, non-nop
+	 * instructions in the rte delay slot, and associative writes to
+	 * the memory-mapped cache array.
 	 */
-	expmask &= ~(EXPMASK_RTEDS | EXPMASK_BRDSSLP);
-
-	/*
-	 * Enable associative writes to the memory-mapped cache array
-	 * until the cache flush ops have been rewritten.
-	 */
-	expmask |= EXPMASK_MMCAW;
+	expmask &= ~(EXPMASK_RTEDS | EXPMASK_BRDSSLP | EXPMASK_MMCAW);
 
 	__raw_writel(expmask, EXPMASK);
 	ctrl_barrier();
@@ -311,12 +306,12 @@ asmlinkage void __init sh_cpu_init(void)
 	if (fpu_disabled) {
 		printk("FPU Disabled\n");
 		current_cpu_data.flags &= ~CPU_HAS_FPU;
-		disable_fpu();
 	}
 
 	/* FPU initialization */
+	disable_fpu();
 	if ((current_cpu_data.flags & CPU_HAS_FPU)) {
-		clear_thread_flag(TIF_USEDFPU);
+		current_thread_info()->status &= ~TS_USEDFPU;
 		clear_used_math();
 	}
 
@@ -338,17 +333,6 @@ asmlinkage void __init sh_cpu_init(void)
 	}
 #endif
 
-	/*
-	 * Some brain-damaged loaders decided it would be a good idea to put
-	 * the UBC to sleep. This causes some issues when it comes to things
-	 * like PTRACE_SINGLESTEP or doing hardware watchpoints in GDB.  So ..
-	 * we wake it up and hope that all is well.
-	 */
-#ifdef CONFIG_SUPERH32
-	if (raw_smp_processor_id() == 0)
-		ubc_wakeup();
-#endif
-
 	speculative_execution_init();
 	expmask_init();
 }
diff --git a/arch/sh/kernel/cpu/sh2a/fpu.c b/arch/sh/kernel/cpu/sh2a/fpu.c
index 6df2fb98eb3..d395ce5740e 100644
--- a/arch/sh/kernel/cpu/sh2a/fpu.c
+++ b/arch/sh/kernel/cpu/sh2a/fpu.c
@@ -25,14 +25,12 @@
 
 /*
  * Save FPU registers onto task structure.
- * Assume called with FPU enabled (SR.FD=0).
  */
 void
-save_fpu(struct task_struct *tsk, struct pt_regs *regs)
+save_fpu(struct task_struct *tsk)
 {
 	unsigned long dummy;
 
-	clear_tsk_thread_flag(tsk, TIF_USEDFPU);
 	enable_fpu();
 	asm volatile("sts.l	fpul, @-%0\n\t"
 		     "sts.l	fpscr, @-%0\n\t"
@@ -60,7 +58,6 @@ save_fpu(struct task_struct *tsk, struct pt_regs *regs)
 		     : "memory");
 
 	disable_fpu();
-	release_fpu(regs);
 }
 
 static void
@@ -598,31 +595,31 @@ BUILD_TRAP_HANDLER(fpu_error)
 	struct task_struct *tsk = current;
 	TRAP_HANDLER_DECL;
 
-	save_fpu(tsk, regs);
+	__unlazy_fpu(tsk, regs);
 	if (ieee_fpe_handler(regs)) {
 		tsk->thread.fpu.hard.fpscr &=
 			~(FPSCR_CAUSE_MASK | FPSCR_FLAG_MASK);
 		grab_fpu(regs);
 		restore_fpu(tsk);
-		set_tsk_thread_flag(tsk, TIF_USEDFPU);
+		task_thread_info(tsk)->status |= TS_USEDFPU;
 		return;
 	}
 
 	force_sig(SIGFPE, tsk);
 }
 
-BUILD_TRAP_HANDLER(fpu_state_restore)
+void fpu_state_restore(struct pt_regs *regs)
 {
 	struct task_struct *tsk = current;
-	TRAP_HANDLER_DECL;
 
 	grab_fpu(regs);
-	if (!user_mode(regs)) {
+	if (unlikely(!user_mode(regs))) {
 		printk(KERN_ERR "BUG: FPU is used in kernel mode.\n");
+		BUG();
 		return;
 	}
 
-	if (used_math()) {
+	if (likely(used_math())) {
 		/* Using the FPU again.  */
 		restore_fpu(tsk);
 	} else	{
@@ -630,5 +627,13 @@ BUILD_TRAP_HANDLER(fpu_state_restore)
 		fpu_init();
 		set_used_math();
 	}
-	set_tsk_thread_flag(tsk, TIF_USEDFPU);
+	task_thread_info(tsk)->status |= TS_USEDFPU;
+	tsk->fpu_counter++;
+}
+
+BUILD_TRAP_HANDLER(fpu_state_restore)
+{
+	TRAP_HANDLER_DECL;
+
+	fpu_state_restore(regs);
 }
diff --git a/arch/sh/kernel/cpu/sh3/entry.S b/arch/sh/kernel/cpu/sh3/entry.S
index bb407ef0b91..3f7e2a22c7c 100644
--- a/arch/sh/kernel/cpu/sh3/entry.S
+++ b/arch/sh/kernel/cpu/sh3/entry.S
@@ -297,41 +297,8 @@ ENTRY(vbr_base)
 !
 	.balign 	256,0,256
 general_exception:
-#ifndef CONFIG_CPU_SUBTYPE_SHX3
 	bra	handle_exception
 	 sts	pr, k3		! save original pr value in k3
-#else
-	mov.l	1f, k4
-	mov.l	@k4, k4
-
-	! Is EXPEVT larger than 0x800?
-	mov	#0x8, k0
-	shll8	k0
-	cmp/hs	k0, k4
-	bf	0f
-
-	! then add 0x580 (k2 is 0xd80 or 0xda0)
-	mov	#0x58, k0
-	shll2	k0
-	shll2	k0
-	add	k0, k4
-0:
-	! Setup stack and save DSP context (k0 contains original r15 on return)
-	bsr	prepare_stack
-	 nop
-
-	! Save registers / Switch to bank 0
-	mov		k4, k2		! keep vector in k2
-	mov.l	1f, k4		! SR bits to clear in k4
-	bsr	save_regs	! needs original pr value in k3
-	 nop
-
-	bra	handle_exception_special
-	 nop
-
-	.align	2
-1:	.long	EXPEVT
-#endif
 
 ! prepare_stack()
 ! - roll back gRB
diff --git a/arch/sh/kernel/cpu/sh4/Makefile b/arch/sh/kernel/cpu/sh4/Makefile
index 203b18347b8..3a1dbc70983 100644
--- a/arch/sh/kernel/cpu/sh4/Makefile
+++ b/arch/sh/kernel/cpu/sh4/Makefile
@@ -9,6 +9,11 @@ obj-$(CONFIG_HIBERNATION)		+= $(addprefix ../sh3/, swsusp.o)
 obj-$(CONFIG_SH_FPU)			+= fpu.o softfloat.o
 obj-$(CONFIG_SH_STORE_QUEUES)		+= sq.o
 
+# Perf events
+perf-$(CONFIG_CPU_SUBTYPE_SH7750)	:= perf_event.o
+perf-$(CONFIG_CPU_SUBTYPE_SH7750S)	:= perf_event.o
+perf-$(CONFIG_CPU_SUBTYPE_SH7091)	:= perf_event.o
+
 # CPU subtype setup
 obj-$(CONFIG_CPU_SUBTYPE_SH7750)	+= setup-sh7750.o
 obj-$(CONFIG_CPU_SUBTYPE_SH7750R)	+= setup-sh7750.o
@@ -27,4 +32,5 @@ endif
 # Additional clocks by subtype
 clock-$(CONFIG_CPU_SUBTYPE_SH4_202)	+= clock-sh4-202.o
 
-obj-y	+= $(clock-y)
+obj-y					+= $(clock-y)
+obj-$(CONFIG_PERF_EVENTS)		+= $(perf-y)
diff --git a/arch/sh/kernel/cpu/sh4/fpu.c b/arch/sh/kernel/cpu/sh4/fpu.c
index e3ea5411da6..e97857aec8a 100644
--- a/arch/sh/kernel/cpu/sh4/fpu.c
+++ b/arch/sh/kernel/cpu/sh4/fpu.c
@@ -41,13 +41,11 @@ static unsigned int fpu_exception_flags;
 
 /*
  * Save FPU registers onto task structure.
- * Assume called with FPU enabled (SR.FD=0).
  */
-void save_fpu(struct task_struct *tsk, struct pt_regs *regs)
+void save_fpu(struct task_struct *tsk)
 {
 	unsigned long dummy;
 
-	clear_tsk_thread_flag(tsk, TIF_USEDFPU);
 	enable_fpu();
 	asm volatile ("sts.l	fpul, @-%0\n\t"
 		      "sts.l	fpscr, @-%0\n\t"
@@ -92,7 +90,6 @@ void save_fpu(struct task_struct *tsk, struct pt_regs *regs)
 		      :"memory");
 
 	disable_fpu();
-	release_fpu(regs);
 }
 
 static void restore_fpu(struct task_struct *tsk)
@@ -285,7 +282,6 @@ static int ieee_fpe_handler(struct pt_regs *regs)
 		/* fcnvsd */
 		struct task_struct *tsk = current;
 
-		save_fpu(tsk, regs);
 		if ((tsk->thread.fpu.hard.fpscr & FPSCR_CAUSE_ERROR))
 			/* FPU error */
 			denormal_to_double(&tsk->thread.fpu.hard,
@@ -462,7 +458,7 @@ BUILD_TRAP_HANDLER(fpu_error)
 	struct task_struct *tsk = current;
 	TRAP_HANDLER_DECL;
 
-	save_fpu(tsk, regs);
+	__unlazy_fpu(tsk, regs);
 	fpu_exception_flags = 0;
 	if (ieee_fpe_handler(regs)) {
 		tsk->thread.fpu.hard.fpscr &=
@@ -473,7 +469,7 @@ BUILD_TRAP_HANDLER(fpu_error)
 		tsk->thread.fpu.hard.fpscr |= (fpu_exception_flags >> 10);
 		grab_fpu(regs);
 		restore_fpu(tsk);
-		set_tsk_thread_flag(tsk, TIF_USEDFPU);
+		task_thread_info(tsk)->status |= TS_USEDFPU;
 		if ((((tsk->thread.fpu.hard.fpscr & FPSCR_ENABLE_MASK) >> 7) &
 		     (fpu_exception_flags >> 2)) == 0) {
 			return;
@@ -483,18 +479,18 @@ BUILD_TRAP_HANDLER(fpu_error)
 	force_sig(SIGFPE, tsk);
 }
 
-BUILD_TRAP_HANDLER(fpu_state_restore)
+void fpu_state_restore(struct pt_regs *regs)
 {
 	struct task_struct *tsk = current;
-	TRAP_HANDLER_DECL;
 
 	grab_fpu(regs);
-	if (!user_mode(regs)) {
+	if (unlikely(!user_mode(regs))) {
 		printk(KERN_ERR "BUG: FPU is used in kernel mode.\n");
+		BUG();
 		return;
 	}
 
-	if (used_math()) {
+	if (likely(used_math())) {
 		/* Using the FPU again.  */
 		restore_fpu(tsk);
 	} else {
@@ -502,5 +498,13 @@ BUILD_TRAP_HANDLER(fpu_state_restore)
 		fpu_init();
 		set_used_math();
 	}
-	set_tsk_thread_flag(tsk, TIF_USEDFPU);
+	task_thread_info(tsk)->status |= TS_USEDFPU;
+	tsk->fpu_counter++;
+}
+
+BUILD_TRAP_HANDLER(fpu_state_restore)
+{
+	TRAP_HANDLER_DECL;
+
+	fpu_state_restore(regs);
 }
diff --git a/arch/sh/kernel/cpu/sh4/perf_event.c b/arch/sh/kernel/cpu/sh4/perf_event.c
new file mode 100644
index 00000000000..7f9ecc9c2d0
--- /dev/null
+++ b/arch/sh/kernel/cpu/sh4/perf_event.c
@@ -0,0 +1,253 @@
+/*
+ * Performance events support for SH7750-style performance counters
+ *
+ *  Copyright (C) 2009  Paul Mundt
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ */
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/perf_event.h>
+#include <asm/processor.h>
+
+#define PM_CR_BASE	0xff000084	/* 16-bit */
+#define PM_CTR_BASE	0xff100004	/* 32-bit */
+
+#define PMCR(n)		(PM_CR_BASE + ((n) * 0x04))
+#define PMCTRH(n)	(PM_CTR_BASE + 0x00 + ((n) * 0x08))
+#define PMCTRL(n)	(PM_CTR_BASE + 0x04 + ((n) * 0x08))
+
+#define PMCR_PMM_MASK	0x0000003f
+
+#define PMCR_CLKF	0x00000100
+#define PMCR_PMCLR	0x00002000
+#define PMCR_PMST	0x00004000
+#define PMCR_PMEN	0x00008000
+
+static struct sh_pmu sh7750_pmu;
+
+/*
+ * There are a number of events supported by each counter (33 in total).
+ * Since we have 2 counters, each counter will take the event code as it
+ * corresponds to the PMCR PMM setting. Each counter can be configured
+ * independently.
+ *
+ *	Event Code	Description
+ *	----------	-----------
+ *
+ *	0x01		Operand read access
+ *	0x02		Operand write access
+ *	0x03		UTLB miss
+ *	0x04		Operand cache read miss
+ *	0x05		Operand cache write miss
+ *	0x06		Instruction fetch (w/ cache)
+ *	0x07		Instruction TLB miss
+ *	0x08		Instruction cache miss
+ *	0x09		All operand accesses
+ *	0x0a		All instruction accesses
+ *	0x0b		OC RAM operand access
+ *	0x0d		On-chip I/O space access
+ *	0x0e		Operand access (r/w)
+ *	0x0f		Operand cache miss (r/w)
+ *	0x10		Branch instruction
+ *	0x11		Branch taken
+ *	0x12		BSR/BSRF/JSR
+ *	0x13		Instruction execution
+ *	0x14		Instruction execution in parallel
+ *	0x15		FPU Instruction execution
+ *	0x16		Interrupt
+ *	0x17		NMI
+ *	0x18		trapa instruction execution
+ *	0x19		UBCA match
+ *	0x1a		UBCB match
+ *	0x21		Instruction cache fill
+ *	0x22		Operand cache fill
+ *	0x23		Elapsed time
+ *	0x24		Pipeline freeze by I-cache miss
+ *	0x25		Pipeline freeze by D-cache miss
+ *	0x27		Pipeline freeze by branch instruction
+ *	0x28		Pipeline freeze by CPU register
+ *	0x29		Pipeline freeze by FPU
+ */
+
+static const int sh7750_general_events[] = {
+	[PERF_COUNT_HW_CPU_CYCLES]		= 0x0023,
+	[PERF_COUNT_HW_INSTRUCTIONS]		= 0x000a,
+	[PERF_COUNT_HW_CACHE_REFERENCES]	= 0x0006,	/* I-cache */
+	[PERF_COUNT_HW_CACHE_MISSES]		= 0x0008,	/* I-cache */
+	[PERF_COUNT_HW_BRANCH_INSTRUCTIONS]	= 0x0010,
+	[PERF_COUNT_HW_BRANCH_MISSES]		= -1,
+	[PERF_COUNT_HW_BUS_CYCLES]		= -1,
+};
+
+#define C(x)	PERF_COUNT_HW_CACHE_##x
+
+static const int sh7750_cache_events
+			[PERF_COUNT_HW_CACHE_MAX]
+			[PERF_COUNT_HW_CACHE_OP_MAX]
+			[PERF_COUNT_HW_CACHE_RESULT_MAX] =
+{
+	[ C(L1D) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0001,
+			[ C(RESULT_MISS)   ] = 0x0004,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0002,
+			[ C(RESULT_MISS)   ] = 0x0005,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(L1I) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0006,
+			[ C(RESULT_MISS)   ] = 0x0008,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(LL) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(DTLB) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0x0003,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(ITLB) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0x0007,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+	},
+
+	[ C(BPU) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+	},
+};
+
+static int sh7750_event_map(int event)
+{
+	return sh7750_general_events[event];
+}
+
+static u64 sh7750_pmu_read(int idx)
+{
+	return (u64)((u64)(__raw_readl(PMCTRH(idx)) & 0xffff) << 32) |
+			   __raw_readl(PMCTRL(idx));
+}
+
+static void sh7750_pmu_disable(struct hw_perf_event *hwc, int idx)
+{
+	unsigned int tmp;
+
+	tmp = __raw_readw(PMCR(idx));
+	tmp &= ~(PMCR_PMM_MASK | PMCR_PMEN);
+	__raw_writew(tmp, PMCR(idx));
+}
+
+static void sh7750_pmu_enable(struct hw_perf_event *hwc, int idx)
+{
+	__raw_writew(__raw_readw(PMCR(idx)) | PMCR_PMCLR, PMCR(idx));
+	__raw_writew(hwc->config | PMCR_PMEN | PMCR_PMST, PMCR(idx));
+}
+
+static void sh7750_pmu_disable_all(void)
+{
+	int i;
+
+	for (i = 0; i < sh7750_pmu.num_events; i++)
+		__raw_writew(__raw_readw(PMCR(i)) & ~PMCR_PMEN, PMCR(i));
+}
+
+static void sh7750_pmu_enable_all(void)
+{
+	int i;
+
+	for (i = 0; i < sh7750_pmu.num_events; i++)
+		__raw_writew(__raw_readw(PMCR(i)) | PMCR_PMEN, PMCR(i));
+}
+
+static struct sh_pmu sh7750_pmu = {
+	.name		= "SH7750",
+	.num_events	= 2,
+	.event_map	= sh7750_event_map,
+	.max_events	= ARRAY_SIZE(sh7750_general_events),
+	.raw_event_mask	= PMCR_PMM_MASK,
+	.cache_events	= &sh7750_cache_events,
+	.read		= sh7750_pmu_read,
+	.disable	= sh7750_pmu_disable,
+	.enable		= sh7750_pmu_enable,
+	.disable_all	= sh7750_pmu_disable_all,
+	.enable_all	= sh7750_pmu_enable_all,
+};
+
+static int __init sh7750_pmu_init(void)
+{
+	/*
+	 * Make sure this CPU actually has perf counters.
+	 */
+	if (!(boot_cpu_data.flags & CPU_HAS_PERF_COUNTER)) {
+		pr_notice("HW perf events unsupported, software events only.\n");
+		return -ENODEV;
+	}
+
+	return register_sh_pmu(&sh7750_pmu);
+}
+arch_initcall(sh7750_pmu_init);
diff --git a/arch/sh/kernel/cpu/sh4a/Makefile b/arch/sh/kernel/cpu/sh4a/Makefile
index 490d5dc9e37..33bab477d2e 100644
--- a/arch/sh/kernel/cpu/sh4a/Makefile
+++ b/arch/sh/kernel/cpu/sh4a/Makefile
@@ -44,3 +44,4 @@ pinmux-$(CONFIG_CPU_SUBTYPE_SH7786)	:= pinmux-sh7786.o
 obj-y				+= $(clock-y)
 obj-$(CONFIG_SMP)		+= $(smp-y)
 obj-$(CONFIG_GENERIC_GPIO)	+= $(pinmux-y)
+obj-$(CONFIG_PERF_EVENTS)	+= perf_event.o
diff --git a/arch/sh/kernel/cpu/sh4a/clock-sh7724.c b/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
index dfe9192be63..9db743802f0 100644
--- a/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
+++ b/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
@@ -152,7 +152,7 @@ struct clk div6_clks[] = {
 	SH_CLK_DIV6("fsia_clk", &div3_clk, FCLKACR, 0),
 	SH_CLK_DIV6("fsib_clk", &div3_clk, FCLKBCR, 0),
 	SH_CLK_DIV6("irda_clk", &div3_clk, IRDACLKCR, 0),
-	SH_CLK_DIV6("spu_clk", &div3_clk, SPUCLKCR, 0),
+	SH_CLK_DIV6("spu_clk", &div3_clk, SPUCLKCR, CLK_ENABLE_ON_INIT),
 };
 
 #define R_CLK (&r_clk)
diff --git a/arch/sh/kernel/cpu/sh4a/perf_event.c b/arch/sh/kernel/cpu/sh4a/perf_event.c
new file mode 100644
index 00000000000..eddc21973fa
--- /dev/null
+++ b/arch/sh/kernel/cpu/sh4a/perf_event.c
@@ -0,0 +1,269 @@
+/*
+ * Performance events support for SH-4A performance counters
+ *
+ *  Copyright (C) 2009  Paul Mundt
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ */
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/perf_event.h>
+#include <asm/processor.h>
+
+#define PPC_CCBR(idx)	(0xff200800 + (sizeof(u32) * idx))
+#define PPC_PMCTR(idx)	(0xfc100000 + (sizeof(u32) * idx))
+
+#define CCBR_CIT_MASK	(0x7ff << 6)
+#define CCBR_DUC	(1 << 3)
+#define CCBR_CMDS	(1 << 1)
+#define CCBR_PPCE	(1 << 0)
+
+#define PPC_PMCAT	0xfc100080
+
+#define PMCAT_OVF3	(1 << 27)
+#define PMCAT_CNN3	(1 << 26)
+#define PMCAT_CLR3	(1 << 25)
+#define PMCAT_OVF2	(1 << 19)
+#define PMCAT_CLR2	(1 << 17)
+#define PMCAT_OVF1	(1 << 11)
+#define PMCAT_CNN1	(1 << 10)
+#define PMCAT_CLR1	(1 << 9)
+#define PMCAT_OVF0	(1 << 3)
+#define PMCAT_CLR0	(1 << 1)
+
+static struct sh_pmu sh4a_pmu;
+
+/*
+ * Supported raw event codes:
+ *
+ *	Event Code	Description
+ *	----------	-----------
+ *
+ *	0x0000		number of elapsed cycles
+ *	0x0200		number of elapsed cycles in privileged mode
+ *	0x0280		number of elapsed cycles while SR.BL is asserted
+ *	0x0202		instruction execution
+ *	0x0203		instruction execution in parallel
+ *	0x0204		number of unconditional branches
+ *	0x0208		number of exceptions
+ *	0x0209		number of interrupts
+ *	0x0220		UTLB miss caused by instruction fetch
+ *	0x0222		UTLB miss caused by operand access
+ *	0x02a0		number of ITLB misses
+ *	0x0028		number of accesses to instruction memories
+ *	0x0029		number of accesses to instruction cache
+ *	0x002a		instruction cache miss
+ *	0x022e		number of access to instruction X/Y memory
+ *	0x0030		number of reads to operand memories
+ *	0x0038		number of writes to operand memories
+ *	0x0031		number of operand cache read accesses
+ *	0x0039		number of operand cache write accesses
+ *	0x0032		operand cache read miss
+ *	0x003a		operand cache write miss
+ *	0x0236		number of reads to operand X/Y memory
+ *	0x023e		number of writes to operand X/Y memory
+ *	0x0237		number of reads to operand U memory
+ *	0x023f		number of writes to operand U memory
+ *	0x0337		number of U memory read buffer misses
+ *	0x02b4		number of wait cycles due to operand read access
+ *	0x02bc		number of wait cycles due to operand write access
+ *	0x0033		number of wait cycles due to operand cache read miss
+ *	0x003b		number of wait cycles due to operand cache write miss
+ */
+
+/*
+ * Special reserved bits used by hardware emulators, read values will
+ * vary, but writes must always be 0.
+ */
+#define PMCAT_EMU_CLR_MASK	((1 << 24) | (1 << 16) | (1 << 8) | (1 << 0))
+
+static const int sh4a_general_events[] = {
+	[PERF_COUNT_HW_CPU_CYCLES]		= 0x0000,
+	[PERF_COUNT_HW_INSTRUCTIONS]		= 0x0202,
+	[PERF_COUNT_HW_CACHE_REFERENCES]	= 0x0029,	/* I-cache */
+	[PERF_COUNT_HW_CACHE_MISSES]		= 0x002a,	/* I-cache */
+	[PERF_COUNT_HW_BRANCH_INSTRUCTIONS]	= 0x0204,
+	[PERF_COUNT_HW_BRANCH_MISSES]		= -1,
+	[PERF_COUNT_HW_BUS_CYCLES]		= -1,
+};
+
+#define C(x)	PERF_COUNT_HW_CACHE_##x
+
+static const int sh4a_cache_events
+			[PERF_COUNT_HW_CACHE_MAX]
+			[PERF_COUNT_HW_CACHE_OP_MAX]
+			[PERF_COUNT_HW_CACHE_RESULT_MAX] =
+{
+	[ C(L1D) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0031,
+			[ C(RESULT_MISS)   ] = 0x0032,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0039,
+			[ C(RESULT_MISS)   ] = 0x003a,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(L1I) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0029,
+			[ C(RESULT_MISS)   ] = 0x002a,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(LL) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0030,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0038,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(DTLB) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0x0222,
+			[ C(RESULT_MISS)   ] = 0x0220,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0,
+		},
+	},
+
+	[ C(ITLB) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = 0,
+			[ C(RESULT_MISS)   ] = 0x02a0,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+	},
+
+	[ C(BPU) ] = {
+		[ C(OP_READ) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_WRITE) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+		[ C(OP_PREFETCH) ] = {
+			[ C(RESULT_ACCESS) ] = -1,
+			[ C(RESULT_MISS)   ] = -1,
+		},
+	},
+};
+
+static int sh4a_event_map(int event)
+{
+	return sh4a_general_events[event];
+}
+
+static u64 sh4a_pmu_read(int idx)
+{
+	return __raw_readl(PPC_PMCTR(idx));
+}
+
+static void sh4a_pmu_disable(struct hw_perf_event *hwc, int idx)
+{
+	unsigned int tmp;
+
+	tmp = __raw_readl(PPC_CCBR(idx));
+	tmp &= ~(CCBR_CIT_MASK | CCBR_DUC);
+	__raw_writel(tmp, PPC_CCBR(idx));
+}
+
+static void sh4a_pmu_enable(struct hw_perf_event *hwc, int idx)
+{
+	unsigned int tmp;
+
+	tmp = __raw_readl(PPC_PMCAT);
+	tmp &= ~PMCAT_EMU_CLR_MASK;
+	tmp |= idx ? PMCAT_CLR1 : PMCAT_CLR0;
+	__raw_writel(tmp, PPC_PMCAT);
+
+	tmp = __raw_readl(PPC_CCBR(idx));
+	tmp |= (hwc->config << 6) | CCBR_CMDS | CCBR_PPCE;
+	__raw_writel(tmp, PPC_CCBR(idx));
+
+	__raw_writel(__raw_readl(PPC_CCBR(idx)) | CCBR_DUC, PPC_CCBR(idx));
+}
+
+static void sh4a_pmu_disable_all(void)
+{
+	int i;
+
+	for (i = 0; i < sh4a_pmu.num_events; i++)
+		__raw_writel(__raw_readl(PPC_CCBR(i)) & ~CCBR_DUC, PPC_CCBR(i));
+}
+
+static void sh4a_pmu_enable_all(void)
+{
+	int i;
+
+	for (i = 0; i < sh4a_pmu.num_events; i++)
+		__raw_writel(__raw_readl(PPC_CCBR(i)) | CCBR_DUC, PPC_CCBR(i));
+}
+
+static struct sh_pmu sh4a_pmu = {
+	.name		= "SH-4A",
+	.num_events	= 2,
+	.event_map	= sh4a_event_map,
+	.max_events	= ARRAY_SIZE(sh4a_general_events),
+	.raw_event_mask	= 0x3ff,
+	.cache_events	= &sh4a_cache_events,
+	.read		= sh4a_pmu_read,
+	.disable	= sh4a_pmu_disable,
+	.enable		= sh4a_pmu_enable,
+	.disable_all	= sh4a_pmu_disable_all,
+	.enable_all	= sh4a_pmu_enable_all,
+};
+
+static int __init sh4a_pmu_init(void)
+{
+	/*
+	 * Make sure this CPU actually has perf counters.
+	 */
+	if (!(boot_cpu_data.flags & CPU_HAS_PERF_COUNTER)) {
+		pr_notice("HW perf events unsupported, software events only.\n");
+		return -ENODEV;
+	}
+
+	return register_sh_pmu(&sh4a_pmu);
+}
+arch_initcall(sh4a_pmu_init);
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7724.c b/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
index f3851fd757e..845e89c936e 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
@@ -20,6 +20,8 @@
 #include <linux/uio_driver.h>
 #include <linux/sh_timer.h>
 #include <linux/io.h>
+#include <linux/notifier.h>
+#include <asm/suspend.h>
 #include <asm/clock.h>
 #include <asm/mmzone.h>
 #include <cpu/sh7724.h>
@@ -202,7 +204,7 @@ static struct resource veu0_resources[] = {
 	[0] = {
 		.name	= "VEU3F0",
 		.start	= 0xfe920000,
-		.end	= 0xfe9200cb - 1,
+		.end	= 0xfe9200cb,
 		.flags	= IORESOURCE_MEM,
 	},
 	[1] = {
@@ -234,7 +236,7 @@ static struct resource veu1_resources[] = {
 	[0] = {
 		.name	= "VEU3F1",
 		.start	= 0xfe924000,
-		.end	= 0xfe9240cb - 1,
+		.end	= 0xfe9240cb,
 		.flags	= IORESOURCE_MEM,
 	},
 	[1] = {
@@ -523,6 +525,70 @@ static struct platform_device jpu_device = {
 	},
 };
 
+/* SPU2DSP0 */
+static struct uio_info spu0_platform_data = {
+	.name = "SPU2DSP0",
+	.version = "0",
+	.irq = 86,
+};
+
+static struct resource spu0_resources[] = {
+	[0] = {
+		.name	= "SPU2DSP0",
+		.start	= 0xFE200000,
+		.end	= 0xFE2FFFFF,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1] = {
+		/* place holder for contiguous memory */
+	},
+};
+
+static struct platform_device spu0_device = {
+	.name		= "uio_pdrv_genirq",
+	.id		= 4,
+	.dev = {
+		.platform_data	= &spu0_platform_data,
+	},
+	.resource	= spu0_resources,
+	.num_resources	= ARRAY_SIZE(spu0_resources),
+	.archdata = {
+		.hwblk_id = HWBLK_SPU,
+	},
+};
+
+/* SPU2DSP1 */
+static struct uio_info spu1_platform_data = {
+	.name = "SPU2DSP1",
+	.version = "0",
+	.irq = 87,
+};
+
+static struct resource spu1_resources[] = {
+	[0] = {
+		.name	= "SPU2DSP1",
+		.start	= 0xFE300000,
+		.end	= 0xFE3FFFFF,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1] = {
+		/* place holder for contiguous memory */
+	},
+};
+
+static struct platform_device spu1_device = {
+	.name		= "uio_pdrv_genirq",
+	.id		= 5,
+	.dev = {
+		.platform_data	= &spu1_platform_data,
+	},
+	.resource	= spu1_resources,
+	.num_resources	= ARRAY_SIZE(spu1_resources),
+	.archdata = {
+		.hwblk_id = HWBLK_SPU,
+	},
+};
+
 static struct platform_device *sh7724_devices[] __initdata = {
 	&cmt_device,
 	&tmu0_device,
@@ -539,6 +605,8 @@ static struct platform_device *sh7724_devices[] __initdata = {
 	&veu0_device,
 	&veu1_device,
 	&jpu_device,
+	&spu0_device,
+	&spu1_device,
 };
 
 static int __init sh7724_devices_setup(void)
@@ -547,6 +615,8 @@ static int __init sh7724_devices_setup(void)
 	platform_resource_setup_memory(&veu0_device, "veu0", 2 << 20);
 	platform_resource_setup_memory(&veu1_device, "veu1", 2 << 20);
 	platform_resource_setup_memory(&jpu_device,  "jpu",  2 << 20);
+	platform_resource_setup_memory(&spu0_device, "spu0", 2 << 20);
+	platform_resource_setup_memory(&spu1_device, "spu1", 2 << 20);
 
 	return platform_add_devices(sh7724_devices,
 				    ARRAY_SIZE(sh7724_devices));
@@ -827,3 +897,193 @@ void __init plat_irq_setup(void)
 {
 	register_intc_controller(&intc_desc);
 }
+
+static struct {
+	/* BSC */
+	unsigned long mmselr;
+	unsigned long cs0bcr;
+	unsigned long cs4bcr;
+	unsigned long cs5abcr;
+	unsigned long cs5bbcr;
+	unsigned long cs6abcr;
+	unsigned long cs6bbcr;
+	unsigned long cs4wcr;
+	unsigned long cs5awcr;
+	unsigned long cs5bwcr;
+	unsigned long cs6awcr;
+	unsigned long cs6bwcr;
+	/* INTC */
+	unsigned short ipra;
+	unsigned short iprb;
+	unsigned short iprc;
+	unsigned short iprd;
+	unsigned short ipre;
+	unsigned short iprf;
+	unsigned short iprg;
+	unsigned short iprh;
+	unsigned short ipri;
+	unsigned short iprj;
+	unsigned short iprk;
+	unsigned short iprl;
+	unsigned char imr0;
+	unsigned char imr1;
+	unsigned char imr2;
+	unsigned char imr3;
+	unsigned char imr4;
+	unsigned char imr5;
+	unsigned char imr6;
+	unsigned char imr7;
+	unsigned char imr8;
+	unsigned char imr9;
+	unsigned char imr10;
+	unsigned char imr11;
+	unsigned char imr12;
+	/* RWDT */
+	unsigned short rwtcnt;
+	unsigned short rwtcsr;
+	/* CPG */
+	unsigned long irdaclk;