aboutsummaryrefslogtreecommitdiff
path: root/arch/mips/mm/c-r4k.c
diff options
context:
space:
mode:
Diffstat (limited to 'arch/mips/mm/c-r4k.c')
-rw-r--r--arch/mips/mm/c-r4k.c1188
1 files changed, 809 insertions, 379 deletions
diff --git a/arch/mips/mm/c-r4k.c b/arch/mips/mm/c-r4k.c
index 38223b44d96..f2e8302fa70 100644
--- a/arch/mips/mm/c-r4k.c
+++ b/arch/mips/mm/c-r4k.c
@@ -3,15 +3,21 @@
* License. See the file "COPYING" in the main directory of this archive
* for more details.
*
- * Copyright (C) 1996 David S. Miller (dm@engr.sgi.com)
+ * Copyright (C) 1996 David S. Miller (davem@davemloft.net)
* Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002 Ralf Baechle (ralf@gnu.org)
* Copyright (C) 1999, 2000 Silicon Graphics, Inc.
*/
-#include <linux/config.h>
+#include <linux/cpu_pm.h>
+#include <linux/hardirq.h>
#include <linux/init.h>
+#include <linux/highmem.h>
#include <linux/kernel.h>
+#include <linux/linkage.h>
+#include <linux/preempt.h>
#include <linux/sched.h>
+#include <linux/smp.h>
#include <linux/mm.h>
+#include <linux/module.h>
#include <linux/bitops.h>
#include <asm/bcache.h>
@@ -20,14 +26,43 @@
#include <asm/cacheops.h>
#include <asm/cpu.h>
#include <asm/cpu-features.h>
+#include <asm/cpu-type.h>
#include <asm/io.h>
#include <asm/page.h>
#include <asm/pgtable.h>
#include <asm/r4kcache.h>
-#include <asm/system.h>
+#include <asm/sections.h>
#include <asm/mmu_context.h>
#include <asm/war.h>
#include <asm/cacheflush.h> /* for run_uncached() */
+#include <asm/traps.h>
+#include <asm/dma-coherence.h>
+
+/*
+ * Special Variant of smp_call_function for use by cache functions:
+ *
+ * o No return value
+ * o collapses to normal function call on UP kernels
+ * o collapses to normal function call on systems with a single shared
+ * primary cache.
+ * o doesn't disable interrupts on the local CPU
+ */
+static inline void r4k_on_each_cpu(void (*func) (void *info), void *info)
+{
+ preempt_disable();
+
+#ifndef CONFIG_MIPS_MT_SMP
+ smp_call_function(func, info, 1);
+#endif
+ func(info);
+ preempt_enable();
+}
+
+#if defined(CONFIG_MIPS_CMP) || defined(CONFIG_MIPS_CPS)
+#define cpu_has_safe_index_cacheops 0
+#else
+#define cpu_has_safe_index_cacheops 1
+#endif
/*
* Must die.
@@ -39,13 +74,13 @@ static unsigned long scache_size __read_mostly;
/*
* Dummy cache handling routines for machines without boardcaches
*/
-static void no_sc_noop(void) {}
+static void cache_noop(void) {}
static struct bcache_ops no_sc_ops = {
- .bc_enable = (void *)no_sc_noop,
- .bc_disable = (void *)no_sc_noop,
- .bc_wback_inv = (void *)no_sc_noop,
- .bc_inv = (void *)no_sc_noop
+ .bc_enable = (void *)cache_noop,
+ .bc_disable = (void *)cache_noop,
+ .bc_wback_inv = (void *)cache_noop,
+ .bc_inv = (void *)cache_noop
};
struct bcache_ops *bcops = &no_sc_ops;
@@ -69,38 +104,98 @@ static inline void r4k_blast_dcache_page_dc32(unsigned long addr)
blast_dcache32_page(addr);
}
-static inline void r4k_blast_dcache_page_setup(void)
+static inline void r4k_blast_dcache_page_dc64(unsigned long addr)
+{
+ blast_dcache64_page(addr);
+}
+
+static inline void r4k_blast_dcache_page_dc128(unsigned long addr)
+{
+ blast_dcache128_page(addr);
+}
+
+static void r4k_blast_dcache_page_setup(void)
{
unsigned long dc_lsize = cpu_dcache_line_size();
- if (dc_lsize == 16)
+ switch (dc_lsize) {
+ case 0:
+ r4k_blast_dcache_page = (void *)cache_noop;
+ break;
+ case 16:
r4k_blast_dcache_page = blast_dcache16_page;
- else if (dc_lsize == 32)
+ break;
+ case 32:
r4k_blast_dcache_page = r4k_blast_dcache_page_dc32;
+ break;
+ case 64:
+ r4k_blast_dcache_page = r4k_blast_dcache_page_dc64;
+ break;
+ case 128:
+ r4k_blast_dcache_page = r4k_blast_dcache_page_dc128;
+ break;
+ default:
+ break;
+ }
}
+#ifndef CONFIG_EVA
+#define r4k_blast_dcache_user_page r4k_blast_dcache_page
+#else
+
+static void (*r4k_blast_dcache_user_page)(unsigned long addr);
+
+static void r4k_blast_dcache_user_page_setup(void)
+{
+ unsigned long dc_lsize = cpu_dcache_line_size();
+
+ if (dc_lsize == 0)
+ r4k_blast_dcache_user_page = (void *)cache_noop;
+ else if (dc_lsize == 16)
+ r4k_blast_dcache_user_page = blast_dcache16_user_page;
+ else if (dc_lsize == 32)
+ r4k_blast_dcache_user_page = blast_dcache32_user_page;
+ else if (dc_lsize == 64)
+ r4k_blast_dcache_user_page = blast_dcache64_user_page;
+}
+
+#endif
+
static void (* r4k_blast_dcache_page_indexed)(unsigned long addr);
-static inline void r4k_blast_dcache_page_indexed_setup(void)
+static void r4k_blast_dcache_page_indexed_setup(void)
{
unsigned long dc_lsize = cpu_dcache_line_size();
- if (dc_lsize == 16)
+ if (dc_lsize == 0)
+ r4k_blast_dcache_page_indexed = (void *)cache_noop;
+ else if (dc_lsize == 16)
r4k_blast_dcache_page_indexed = blast_dcache16_page_indexed;
else if (dc_lsize == 32)
r4k_blast_dcache_page_indexed = blast_dcache32_page_indexed;
+ else if (dc_lsize == 64)
+ r4k_blast_dcache_page_indexed = blast_dcache64_page_indexed;
+ else if (dc_lsize == 128)
+ r4k_blast_dcache_page_indexed = blast_dcache128_page_indexed;
}
-static void (* r4k_blast_dcache)(void);
+void (* r4k_blast_dcache)(void);
+EXPORT_SYMBOL(r4k_blast_dcache);
-static inline void r4k_blast_dcache_setup(void)
+static void r4k_blast_dcache_setup(void)
{
unsigned long dc_lsize = cpu_dcache_line_size();
- if (dc_lsize == 16)
+ if (dc_lsize == 0)
+ r4k_blast_dcache = (void *)cache_noop;
+ else if (dc_lsize == 16)
r4k_blast_dcache = blast_dcache16;
else if (dc_lsize == 32)
r4k_blast_dcache = blast_dcache32;
+ else if (dc_lsize == 64)
+ r4k_blast_dcache = blast_dcache64;
+ else if (dc_lsize == 128)
+ r4k_blast_dcache = blast_dcache128;
}
/* force code alignment (used for TX49XX_ICACHE_INDEX_INV_WAR) */
@@ -111,7 +206,7 @@ static inline void r4k_blast_dcache_setup(void)
"1:\n\t" \
)
#define CACHE32_UNROLL32_ALIGN JUMP_TO_ALIGN(10) /* 32 * 32 = 1024 */
-#define CACHE32_UNROLL32_ALIGN2 JUMP_TO_ALIGN(11)
+#define CACHE32_UNROLL32_ALIGN2 JUMP_TO_ALIGN(11)
static inline void blast_r4600_v1_icache32(void)
{
@@ -128,19 +223,19 @@ static inline void tx49_blast_icache32(void)
unsigned long end = start + current_cpu_data.icache.waysize;
unsigned long ws_inc = 1UL << current_cpu_data.icache.waybit;
unsigned long ws_end = current_cpu_data.icache.ways <<
- current_cpu_data.icache.waybit;
+ current_cpu_data.icache.waybit;
unsigned long ws, addr;
CACHE32_UNROLL32_ALIGN2;
/* I'm in even chunk. blast odd chunks */
for (ws = 0; ws < ws_end; ws += ws_inc)
for (addr = start + 0x400; addr < end; addr += 0x400 * 2)
- cache32_unroll32(addr|ws,Index_Invalidate_I);
+ cache32_unroll32(addr|ws, Index_Invalidate_I);
CACHE32_UNROLL32_ALIGN;
/* I'm in odd chunk. blast even chunks */
for (ws = 0; ws < ws_end; ws += ws_inc)
for (addr = start; addr < end; addr += 0x400 * 2)
- cache32_unroll32(addr|ws,Index_Invalidate_I);
+ cache32_unroll32(addr|ws, Index_Invalidate_I);
}
static inline void blast_icache32_r4600_v1_page_indexed(unsigned long page)
@@ -154,47 +249,77 @@ static inline void blast_icache32_r4600_v1_page_indexed(unsigned long page)
static inline void tx49_blast_icache32_page_indexed(unsigned long page)
{
- unsigned long start = page;
+ unsigned long indexmask = current_cpu_data.icache.waysize - 1;
+ unsigned long start = INDEX_BASE + (page & indexmask);
unsigned long end = start + PAGE_SIZE;
unsigned long ws_inc = 1UL << current_cpu_data.icache.waybit;
unsigned long ws_end = current_cpu_data.icache.ways <<
- current_cpu_data.icache.waybit;
+ current_cpu_data.icache.waybit;
unsigned long ws, addr;
CACHE32_UNROLL32_ALIGN2;
/* I'm in even chunk. blast odd chunks */
for (ws = 0; ws < ws_end; ws += ws_inc)
for (addr = start + 0x400; addr < end; addr += 0x400 * 2)
- cache32_unroll32(addr|ws,Index_Invalidate_I);
+ cache32_unroll32(addr|ws, Index_Invalidate_I);
CACHE32_UNROLL32_ALIGN;
/* I'm in odd chunk. blast even chunks */
for (ws = 0; ws < ws_end; ws += ws_inc)
for (addr = start; addr < end; addr += 0x400 * 2)
- cache32_unroll32(addr|ws,Index_Invalidate_I);
+ cache32_unroll32(addr|ws, Index_Invalidate_I);
}
static void (* r4k_blast_icache_page)(unsigned long addr);
-static inline void r4k_blast_icache_page_setup(void)
+static void r4k_blast_icache_page_setup(void)
{
unsigned long ic_lsize = cpu_icache_line_size();
- if (ic_lsize == 16)
+ if (ic_lsize == 0)
+ r4k_blast_icache_page = (void *)cache_noop;
+ else if (ic_lsize == 16)
r4k_blast_icache_page = blast_icache16_page;
+ else if (ic_lsize == 32 && current_cpu_type() == CPU_LOONGSON2)
+ r4k_blast_icache_page = loongson2_blast_icache32_page;
else if (ic_lsize == 32)
r4k_blast_icache_page = blast_icache32_page;
else if (ic_lsize == 64)
r4k_blast_icache_page = blast_icache64_page;
+ else if (ic_lsize == 128)
+ r4k_blast_icache_page = blast_icache128_page;
}
+#ifndef CONFIG_EVA
+#define r4k_blast_icache_user_page r4k_blast_icache_page
+#else
+
+static void (*r4k_blast_icache_user_page)(unsigned long addr);
+
+static void __cpuinit r4k_blast_icache_user_page_setup(void)
+{
+ unsigned long ic_lsize = cpu_icache_line_size();
+
+ if (ic_lsize == 0)
+ r4k_blast_icache_user_page = (void *)cache_noop;
+ else if (ic_lsize == 16)
+ r4k_blast_icache_user_page = blast_icache16_user_page;
+ else if (ic_lsize == 32)
+ r4k_blast_icache_user_page = blast_icache32_user_page;
+ else if (ic_lsize == 64)
+ r4k_blast_icache_user_page = blast_icache64_user_page;
+}
+
+#endif
static void (* r4k_blast_icache_page_indexed)(unsigned long addr);
-static inline void r4k_blast_icache_page_indexed_setup(void)
+static void r4k_blast_icache_page_indexed_setup(void)
{
unsigned long ic_lsize = cpu_icache_line_size();
- if (ic_lsize == 16)
+ if (ic_lsize == 0)
+ r4k_blast_icache_page_indexed = (void *)cache_noop;
+ else if (ic_lsize == 16)
r4k_blast_icache_page_indexed = blast_icache16_page_indexed;
else if (ic_lsize == 32) {
if (R4600_V1_INDEX_ICACHEOP_WAR && cpu_is_r4600_v1_x())
@@ -203,6 +328,9 @@ static inline void r4k_blast_icache_page_indexed_setup(void)
else if (TX49XX_ICACHE_INDEX_INV_WAR)
r4k_blast_icache_page_indexed =
tx49_blast_icache32_page_indexed;
+ else if (current_cpu_type() == CPU_LOONGSON2)
+ r4k_blast_icache_page_indexed =
+ loongson2_blast_icache32_page_indexed;
else
r4k_blast_icache_page_indexed =
blast_icache32_page_indexed;
@@ -210,32 +338,41 @@ static inline void r4k_blast_icache_page_indexed_setup(void)
r4k_blast_icache_page_indexed = blast_icache64_page_indexed;
}
-static void (* r4k_blast_icache)(void);
+void (* r4k_blast_icache)(void);
+EXPORT_SYMBOL(r4k_blast_icache);
-static inline void r4k_blast_icache_setup(void)
+static void r4k_blast_icache_setup(void)
{
unsigned long ic_lsize = cpu_icache_line_size();
- if (ic_lsize == 16)
+ if (ic_lsize == 0)
+ r4k_blast_icache = (void *)cache_noop;
+ else if (ic_lsize == 16)
r4k_blast_icache = blast_icache16;
else if (ic_lsize == 32) {
if (R4600_V1_INDEX_ICACHEOP_WAR && cpu_is_r4600_v1_x())
r4k_blast_icache = blast_r4600_v1_icache32;
else if (TX49XX_ICACHE_INDEX_INV_WAR)
r4k_blast_icache = tx49_blast_icache32;
+ else if (current_cpu_type() == CPU_LOONGSON2)
+ r4k_blast_icache = loongson2_blast_icache32;
else
r4k_blast_icache = blast_icache32;
} else if (ic_lsize == 64)
r4k_blast_icache = blast_icache64;
+ else if (ic_lsize == 128)
+ r4k_blast_icache = blast_icache128;
}
static void (* r4k_blast_scache_page)(unsigned long addr);
-static inline void r4k_blast_scache_page_setup(void)
+static void r4k_blast_scache_page_setup(void)
{
unsigned long sc_lsize = cpu_scache_line_size();
- if (sc_lsize == 16)
+ if (scache_size == 0)
+ r4k_blast_scache_page = (void *)cache_noop;
+ else if (sc_lsize == 16)
r4k_blast_scache_page = blast_scache16_page;
else if (sc_lsize == 32)
r4k_blast_scache_page = blast_scache32_page;
@@ -247,11 +384,13 @@ static inline void r4k_blast_scache_page_setup(void)
static void (* r4k_blast_scache_page_indexed)(unsigned long addr);
-static inline void r4k_blast_scache_page_indexed_setup(void)
+static void r4k_blast_scache_page_indexed_setup(void)
{
unsigned long sc_lsize = cpu_scache_line_size();
- if (sc_lsize == 16)
+ if (scache_size == 0)
+ r4k_blast_scache_page_indexed = (void *)cache_noop;
+ else if (sc_lsize == 16)
r4k_blast_scache_page_indexed = blast_scache16_page_indexed;
else if (sc_lsize == 32)
r4k_blast_scache_page_indexed = blast_scache32_page_indexed;
@@ -263,11 +402,13 @@ static inline void r4k_blast_scache_page_indexed_setup(void)
static void (* r4k_blast_scache)(void);
-static inline void r4k_blast_scache_setup(void)
+static void r4k_blast_scache_setup(void)
{
unsigned long sc_lsize = cpu_scache_line_size();
- if (sc_lsize == 16)
+ if (scache_size == 0)
+ r4k_blast_scache = (void *)cache_noop;
+ else if (sc_lsize == 16)
r4k_blast_scache = blast_scache16;
else if (sc_lsize == 32)
r4k_blast_scache = blast_scache32;
@@ -277,56 +418,72 @@ static inline void r4k_blast_scache_setup(void)
r4k_blast_scache = blast_scache128;
}
-/*
- * This is former mm's flush_cache_all() which really should be
- * flush_cache_vunmap these days ...
- */
-static inline void local_r4k_flush_cache_all(void * args)
-{
- r4k_blast_dcache();
- r4k_blast_icache();
-}
-
-static void r4k_flush_cache_all(void)
-{
- if (!cpu_has_dc_aliases)
- return;
-
- on_each_cpu(local_r4k_flush_cache_all, NULL, 1, 1);
-}
-
static inline void local_r4k___flush_cache_all(void * args)
{
- r4k_blast_dcache();
- r4k_blast_icache();
-
- switch (current_cpu_data.cputype) {
+ switch (current_cpu_type()) {
+ case CPU_LOONGSON2:
+ case CPU_LOONGSON3:
case CPU_R4000SC:
case CPU_R4000MC:
case CPU_R4400SC:
case CPU_R4400MC:
case CPU_R10000:
case CPU_R12000:
+ case CPU_R14000:
+ /*
+ * These caches are inclusive caches, that is, if something
+ * is not cached in the S-cache, we know it also won't be
+ * in one of the primary caches.
+ */
r4k_blast_scache();
+ break;
+
+ default:
+ r4k_blast_dcache();
+ r4k_blast_icache();
+ break;
}
}
static void r4k___flush_cache_all(void)
{
- on_each_cpu(local_r4k___flush_cache_all, NULL, 1, 1);
+ r4k_on_each_cpu(local_r4k___flush_cache_all, NULL);
+}
+
+static inline int has_valid_asid(const struct mm_struct *mm)
+{
+#ifdef CONFIG_MIPS_MT_SMP
+ int i;
+
+ for_each_online_cpu(i)
+ if (cpu_context(i, mm))
+ return 1;
+
+ return 0;
+#else
+ return cpu_context(smp_processor_id(), mm);
+#endif
+}
+
+static void r4k__flush_cache_vmap(void)
+{
+ r4k_blast_dcache();
+}
+
+static void r4k__flush_cache_vunmap(void)
+{
+ r4k_blast_dcache();
}
static inline void local_r4k_flush_cache_range(void * args)
{
struct vm_area_struct *vma = args;
- int exec;
+ int exec = vma->vm_flags & VM_EXEC;
- if (!(cpu_context(smp_processor_id(), vma->vm_mm)))
+ if (!(has_valid_asid(vma->vm_mm)))
return;
- exec = vma->vm_flags & VM_EXEC;
- if (cpu_has_dc_aliases || exec)
- r4k_blast_dcache();
+ r4k_blast_dcache();
if (exec)
r4k_blast_icache();
}
@@ -334,28 +491,34 @@ static inline void local_r4k_flush_cache_range(void * args)
static void r4k_flush_cache_range(struct vm_area_struct *vma,
unsigned long start, unsigned long end)
{
- on_each_cpu(local_r4k_flush_cache_range, vma, 1, 1);
+ int exec = vma->vm_flags & VM_EXEC;
+
+ if (cpu_has_dc_aliases || (exec && !cpu_has_ic_fills_f_dc))
+ r4k_on_each_cpu(local_r4k_flush_cache_range, vma);
}
static inline void local_r4k_flush_cache_mm(void * args)
{
struct mm_struct *mm = args;
- if (!cpu_context(smp_processor_id(), mm))
+ if (!has_valid_asid(mm))
return;
- r4k_blast_dcache();
- r4k_blast_icache();
-
/*
* Kludge alert. For obscure reasons R4000SC and R4400SC go nuts if we
* only flush the primary caches but R10000 and R12000 behave sane ...
+ * R4000SC and R4400SC indexed S-cache ops also invalidate primary
+ * caches, so we can bail out early.
*/
- if (current_cpu_data.cputype == CPU_R4000SC ||
- current_cpu_data.cputype == CPU_R4000MC ||
- current_cpu_data.cputype == CPU_R4400SC ||
- current_cpu_data.cputype == CPU_R4400MC)
+ if (current_cpu_type() == CPU_R4000SC ||
+ current_cpu_type() == CPU_R4000MC ||
+ current_cpu_type() == CPU_R4400SC ||
+ current_cpu_type() == CPU_R4400MC) {
r4k_blast_scache();
+ return;
+ }
+
+ r4k_blast_dcache();
}
static void r4k_flush_cache_mm(struct mm_struct *mm)
@@ -363,12 +526,13 @@ static void r4k_flush_cache_mm(struct mm_struct *mm)
if (!cpu_has_dc_aliases)
return;
- on_each_cpu(local_r4k_flush_cache_mm, mm, 1, 1);
+ r4k_on_each_cpu(local_r4k_flush_cache_mm, mm);
}
struct flush_cache_page_args {
struct vm_area_struct *vma;
unsigned long addr;
+ unsigned long pfn;
};
static inline void local_r4k_flush_cache_page(void *args)
@@ -376,18 +540,21 @@ static inline void local_r4k_flush_cache_page(void *args)
struct flush_cache_page_args *fcp_args = args;
struct vm_area_struct *vma = fcp_args->vma;
unsigned long addr = fcp_args->addr;
+ struct page *page = pfn_to_page(fcp_args->pfn);
int exec = vma->vm_flags & VM_EXEC;
struct mm_struct *mm = vma->vm_mm;
+ int map_coherent = 0;
pgd_t *pgdp;
pud_t *pudp;
pmd_t *pmdp;
pte_t *ptep;
+ void *vaddr;
/*
* If ownes no valid ASID yet, cannot possibly have gotten
* this page into the cache.
*/
- if (cpu_context(smp_processor_id(), mm) == 0)
+ if (!has_valid_asid(mm))
return;
addr &= PAGE_MASK;
@@ -400,45 +567,47 @@ static inline void local_r4k_flush_cache_page(void *args)
* If the page isn't marked valid, the page cannot possibly be
* in the cache.
*/
- if (!(pte_val(*ptep) & _PAGE_PRESENT))
+ if (!(pte_present(*ptep)))
return;
- /*
- * Doing flushes for another ASID than the current one is
- * too difficult since stupid R4k caches do a TLB translation
- * for every cache flush operation. So we do indexed flushes
- * in that case, which doesn't overly flush the cache too much.
- */
- if ((mm == current->active_mm) && (pte_val(*ptep) & _PAGE_VALID)) {
- if (cpu_has_dc_aliases || (exec && !cpu_has_ic_fills_f_dc)) {
- r4k_blast_dcache_page(addr);
- if (exec && !cpu_icache_snoops_remote_store)
- r4k_blast_scache_page(addr);
- }
- if (exec)
- r4k_blast_icache_page(addr);
-
- return;
+ if ((mm == current->active_mm) && (pte_val(*ptep) & _PAGE_VALID))
+ vaddr = NULL;
+ else {
+ /*
+ * Use kmap_coherent or kmap_atomic to do flushes for
+ * another ASID than the current one.
+ */
+ map_coherent = (cpu_has_dc_aliases &&
+ page_mapped(page) && !Page_dcache_dirty(page));
+ if (map_coherent)
+ vaddr = kmap_coherent(page, addr);
+ else
+ vaddr = kmap_atomic(page);
+ addr = (unsigned long)vaddr;
}
- /*
- * Do indexed flush, too much work to get the (possible) TLB refills
- * to work correctly.
- */
- addr = INDEX_BASE + (addr & (dcache_size - 1));
if (cpu_has_dc_aliases || (exec && !cpu_has_ic_fills_f_dc)) {
- r4k_blast_dcache_page_indexed(addr);
+ vaddr ? r4k_blast_dcache_page(addr) :
+ r4k_blast_dcache_user_page(addr);
if (exec && !cpu_icache_snoops_remote_store)
- r4k_blast_scache_page_indexed(addr);
+ r4k_blast_scache_page(addr);
}
if (exec) {
- if (cpu_has_vtag_icache) {
+ if (vaddr && cpu_has_vtag_icache && mm == current->active_mm) {
int cpu = smp_processor_id();
if (cpu_context(cpu, mm) != 0)
drop_mmu_context(mm, cpu);
} else
- r4k_blast_icache_page_indexed(addr);
+ vaddr ? r4k_blast_icache_page(addr) :
+ r4k_blast_icache_user_page(addr);
+ }
+
+ if (vaddr) {
+ if (map_coherent)
+ kunmap_coherent();
+ else
+ kunmap_atomic(vaddr);
}
}
@@ -449,8 +618,9 @@ static void r4k_flush_cache_page(struct vm_area_struct *vma,
args.vma = vma;
args.addr = addr;
+ args.pfn = pfn;
- on_each_cpu(local_r4k_flush_cache_page, &args, 1, 1);
+ r4k_on_each_cpu(local_r4k_flush_cache_page, &args);
}
static inline void local_r4k_flush_data_cache_page(void * addr)
@@ -460,187 +630,89 @@ static inline void local_r4k_flush_data_cache_page(void * addr)
static void r4k_flush_data_cache_page(unsigned long addr)
{
- on_each_cpu(local_r4k_flush_data_cache_page, (void *) addr, 1, 1);
+ if (in_atomic())
+ local_r4k_flush_data_cache_page((void *)addr);
+ else
+ r4k_on_each_cpu(local_r4k_flush_data_cache_page, (void *) addr);
}
struct flush_icache_range_args {
- unsigned long __user start;
- unsigned long __user end;
+ unsigned long start;
+ unsigned long end;
};
-static inline void local_r4k_flush_icache_range(void *args)
+static inline void local_r4k_flush_icache_range(unsigned long start, unsigned long end)
{
- struct flush_icache_range_args *fir_args = args;
- unsigned long dc_lsize = cpu_dcache_line_size();
- unsigned long ic_lsize = cpu_icache_line_size();
- unsigned long sc_lsize = cpu_scache_line_size();
- unsigned long start = fir_args->start;
- unsigned long end = fir_args->end;
- unsigned long addr, aend;
-
if (!cpu_has_ic_fills_f_dc) {
- if (end - start > dcache_size) {
+ if (end - start >= dcache_size) {
r4k_blast_dcache();
} else {
R4600_HIT_CACHEOP_WAR_IMPL;
- addr = start & ~(dc_lsize - 1);
- aend = (end - 1) & ~(dc_lsize - 1);
-
- while (1) {
- /* Hit_Writeback_Inv_D */
- protected_writeback_dcache_line(addr);
- if (addr == aend)
- break;
- addr += dc_lsize;
- }
- }
-
- if (!cpu_icache_snoops_remote_store) {
- if (end - start > scache_size) {
- r4k_blast_scache();
- } else {
- addr = start & ~(sc_lsize - 1);
- aend = (end - 1) & ~(sc_lsize - 1);
-
- while (1) {
- /* Hit_Writeback_Inv_SD */
- protected_writeback_scache_line(addr);
- if (addr == aend)
- break;
- addr += sc_lsize;
- }
- }
+ protected_blast_dcache_range(start, end);
}
}
if (end - start > icache_size)
r4k_blast_icache();
else {
- addr = start & ~(ic_lsize - 1);
- aend = (end - 1) & ~(ic_lsize - 1);
- while (1) {
- /* Hit_Invalidate_I */
- protected_flush_icache_line(addr);
- if (addr == aend)
- break;
- addr += ic_lsize;
+ switch (boot_cpu_type()) {
+ case CPU_LOONGSON2:
+ protected_loongson2_blast_icache_range(start, end);
+ break;
+
+ default:
+ protected_blast_icache_range(start, end);
+ break;
}
}
-}
-
-static void r4k_flush_icache_range(unsigned long __user start,
- unsigned long __user end)
-{
- struct flush_icache_range_args args;
-
- args.start = start;
- args.end = end;
-
- on_each_cpu(local_r4k_flush_icache_range, &args, 1, 1);
- instruction_hazard();
-}
-
-/*
- * Ok, this seriously sucks. We use them to flush a user page but don't
- * know the virtual address, so we have to blast away the whole icache
- * which is significantly more expensive than the real thing. Otoh we at
- * least know the kernel address of the page so we can flush it
- * selectivly.
- */
-
-struct flush_icache_page_args {
- struct vm_area_struct *vma;
- struct page *page;
-};
-
-static inline void local_r4k_flush_icache_page(void *args)
-{
- struct flush_icache_page_args *fip_args = args;
- struct vm_area_struct *vma = fip_args->vma;
- struct page *page = fip_args->page;
-
+#ifdef CONFIG_EVA
/*
- * Tricky ... Because we don't know the virtual address we've got the
- * choice of either invalidating the entire primary and secondary
- * caches or invalidating the secondary caches also. With the subset
- * enforcment on R4000SC, R4400SC, R10000 and R12000 invalidating the
- * secondary cache will result in any entries in the primary caches
- * also getting invalidated which hopefully is a bit more economical.
+ * Due to all possible segment mappings, there might cache aliases
+ * caused by the bootloader being in non-EVA mode, and the CPU switching
+ * to EVA during early kernel init. It's best to flush the scache
+ * to avoid having secondary cores fetching stale data and lead to
+ * kernel crashes.
*/
- if (cpu_has_subset_pcaches) {
- unsigned long addr = (unsigned long) page_address(page);
-
- r4k_blast_scache_page(addr);
- ClearPageDcacheDirty(page);
-
- return;
- }
-
- if (!cpu_has_ic_fills_f_dc) {
- unsigned long addr = (unsigned long) page_address(page);
- r4k_blast_dcache_page(addr);
- if (!cpu_icache_snoops_remote_store)
- r4k_blast_scache_page(addr);
- ClearPageDcacheDirty(page);
- }
+ bc_wback_inv(start, (end - start));
+ __sync();
+#endif
+}
- /*
- * We're not sure of the virtual address(es) involved here, so
- * we have to flush the entire I-cache.
- */
- if (cpu_has_vtag_icache) {
- int cpu = smp_processor_id();
+static inline void local_r4k_flush_icache_range_ipi(void *args)
+{
+ struct flush_icache_range_args *fir_args = args;
+ unsigned long start = fir_args->start;
+ unsigned long end = fir_args->end;
- if (cpu_context(cpu, vma->vm_mm) != 0)
- drop_mmu_context(vma->vm_mm, cpu);
- } else
- r4k_blast_icache();
+ local_r4k_flush_icache_range(start, end);
}
-static void r4k_flush_icache_page(struct vm_area_struct *vma,
- struct page *page)
+static void r4k_flush_icache_range(unsigned long start, unsigned long end)
{
- struct flush_icache_page_args args;
-
- /*
- * If there's no context yet, or the page isn't executable, no I-cache
- * flush is needed.
- */
- if (!(vma->vm_flags & VM_EXEC))
- return;
+ struct flush_icache_range_args args;
- args.vma = vma;
- args.page = page;
+ args.start = start;
+ args.end = end;
- on_each_cpu(local_r4k_flush_icache_page, &args, 1, 1);
+ r4k_on_each_cpu(local_r4k_flush_icache_range_ipi, &args);
+ instruction_hazard();
}
-
-#ifdef CONFIG_DMA_NONCOHERENT
+#if defined(CONFIG_DMA_NONCOHERENT) || defined(CONFIG_DMA_MAYBE_COHERENT)
static void r4k_dma_cache_wback_inv(unsigned long addr, unsigned long size)
{
- unsigned long end, a;
-
/* Catch bad driver code */
BUG_ON(size == 0);
- if (cpu_has_subset_pcaches) {
- unsigned long sc_lsize = cpu_scache_line_size();
-
- if (size >= scache_size) {
+ preempt_disable();
+ if (cpu_has_inclusive_pcaches) {
+ if (size >= scache_size)
r4k_blast_scache();
- return;
- }
-
- a = addr & ~(sc_lsize - 1);
- end = (addr + size - 1) & ~(sc_lsize - 1);
- while (1) {
- flush_scache_line(a); /* Hit_Writeback_Inv_SD */
- if (a == end)
- break;
- a += sc_lsize;
- }
+ else
+ blast_scache_range(addr, addr + size);
+ preempt_enable();
+ __sync();
return;
}
@@ -649,70 +721,55 @@ static void r4k_dma_cache_wback_inv(unsigned long addr, unsigned long size)
* subset property so we have to flush the primary caches
* explicitly
*/
- if (size >= dcache_size) {
+ if (cpu_has_safe_index_cacheops && size >= dcache_size) {
r4k_blast_dcache();
} else {
- unsigned long dc_lsize = cpu_dcache_line_size();
-
R4600_HIT_CACHEOP_WAR_IMPL;
- a = addr & ~(dc_lsize - 1);
- end = (addr + size - 1) & ~(dc_lsize - 1);
- while (1) {
- flush_dcache_line(a); /* Hit_Writeback_Inv_D */
- if (a == end)
- break;
- a += dc_lsize;
- }
+ blast_dcache_range(addr, addr + size);
}
+ preempt_enable();
bc_wback_inv(addr, size);
+ __sync();
}
static void r4k_dma_cache_inv(unsigned long addr, unsigned long size)
{
- unsigned long end, a;
-
/* Catch bad driver code */
BUG_ON(size == 0);
- if (cpu_has_subset_pcaches) {
- unsigned long sc_lsize = cpu_scache_line_size();
-
- if (size >= scache_size) {
+ preempt_disable();
+ if (cpu_has_inclusive_pcaches) {
+ if (size >= scache_size)
r4k_blast_scache();
- return;
- }
-
- a = addr & ~(sc_lsize - 1);
- end = (addr + size - 1) & ~(sc_lsize - 1);
- while (1) {
- flush_scache_line(a); /* Hit_Writeback_Inv_SD */
- if (a == end)
- break;
- a += sc_lsize;
+ else {
+ /*
+ * There is no clearly documented alignment requirement
+ * for the cache instruction on MIPS processors and
+ * some processors, among them the RM5200 and RM7000
+ * QED processors will throw an address error for cache
+ * hit ops with insufficient alignment. Solved by
+ * aligning the address to cache line size.
+ */
+ blast_inv_scache_range(addr, addr + size);
}
+ preempt_enable();
+ __sync();
return;
}
- if (size >= dcache_size) {
+ if (cpu_has_safe_index_cacheops && size >= dcache_size) {
r4k_blast_dcache();
} else {
- unsigned long dc_lsize = cpu_dcache_line_size();
-
R4600_HIT_CACHEOP_WAR_IMPL;
- a = addr & ~(dc_lsize - 1);
- end = (addr + size - 1) & ~(dc_lsize - 1);
- while (1) {
- flush_dcache_line(a); /* Hit_Writeback_Inv_D */
- if (a == end)
- break;
- a += dc_lsize;
- }
+ blast_inv_dcache_range(addr, addr + size);
}
+ preempt_enable();
bc_inv(addr, size);
+ __sync();
}
-#endif /* CONFIG_DMA_NONCOHERENT */
+#endif /* CONFIG_DMA_NONCOHERENT || CONFIG_DMA_MAYBE_COHERENT */
/*
* While we're protected against bad userland addresses we don't care
@@ -727,10 +784,12 @@ static void local_r4k_flush_cache_sigtramp(void * arg)
unsigned long addr = (unsigned long) arg;
R4600_HIT_CACHEOP_WAR_IMPL;
- protected_writeback_dcache_line(addr & ~(dc_lsize - 1));
- if (!cpu_icache_snoops_remote_store)
+ if (dc_lsize)
+ protected_writeback_dcache_line(addr & ~(dc_lsize - 1));
+ if (!cpu_icache_snoops_remote_store && scache_size)
protected_writeback_scache_line(addr & ~(sc_lsize - 1));
- protected_flush_icache_line(addr & ~(ic_lsize - 1));
+ if (ic_lsize)
+ protected_flush_icache_line(addr & ~(ic_lsize - 1));
if (MIPS4K_ICACHE_REFILL_WAR) {
__asm__ __volatile__ (
".set push\n\t"
@@ -755,7 +814,7 @@ static void local_r4k_flush_cache_sigtramp(void * arg)
static void r4k_flush_cache_sigtramp(unsigned long addr)
{
- on_each_cpu(local_r4k_flush_cache_sigtramp, (void *) addr, 1, 1);
+ r4k_on_each_cpu(local_r4k_flush_cache_sigtramp, (void *) addr);
}
static void r4k_flush_icache_all(void)
@@ -764,6 +823,39 @@ static void r4k_flush_icache_all(void)
r4k_blast_icache();
}
+struct flush_kernel_vmap_range_args {
+ unsigned long vaddr;
+ int size;
+};
+
+static inline void local_r4k_flush_kernel_vmap_range(void *args)
+{
+ struct flush_kernel_vmap_range_args *vmra = args;
+ unsigned long vaddr = vmra->vaddr;
+ int size = vmra->size;
+
+ /*
+ * Aliases only affect the primary caches so don't bother with
+ * S-caches or T-caches.
+ */
+ if (cpu_has_safe_index_cacheops && size >= dcache_size)
+ r4k_blast_dcache();
+ else {
+ R4600_HIT_CACHEOP_WAR_IMPL;
+ blast_dcache_range(vaddr, vaddr + size);
+ }
+}
+
+static void r4k_flush_kernel_vmap_range(unsigned long vaddr, int size)
+{
+ struct flush_kernel_vmap_range_args args;
+
+ args.vaddr = (unsigned long) vaddr;
+ args.size = size;
+
+ r4k_on_each_cpu(local_r4k_flush_kernel_vmap_range, &args);
+}
+
static inline void rm7k_erratum31(void)
{
const unsigned long ic_lsize = 32;
@@ -796,11 +888,40 @@ static inline void rm7k_erratum31(void)
}
}
-static char *way_string[] __initdata = { NULL, "direct mapped", "2-way",
+static inline void alias_74k_erratum(struct cpuinfo_mips *c)
+{
+ unsigned int imp = c->processor_id & PRID_IMP_MASK;
+ unsigned int rev = c->processor_id & PRID_REV_MASK;
+
+ /*
+ * Early versions of the 74K do not update the cache tags on a
+ * vtag miss/ptag hit which can occur in the case of KSEG0/KUSEG
+ * aliases. In this case it is better to treat the cache as always
+ * having aliases.
+ */
+ switch (imp) {
+ case PRID_IMP_74K:
+ if (rev <= PRID_REV_ENCODE_332(2, 4, 0))
+ c->dcache.flags |= MIPS_CACHE_VTAG;
+ if (rev == PRID_REV_ENCODE_332(2, 4, 0))
+ write_c0_config6(read_c0_config6() | MIPS_CONF6_SYND);
+ break;
+ case PRID_IMP_1074K:
+ if (rev <= PRID_REV_ENCODE_332(1, 1, 0)) {
+ c->dcache.flags |= MIPS_CACHE_VTAG;
+ write_c0_config6(read_c0_config6() | MIPS_CONF6_SYND);
+ }
+ break;
+ default:
+ BUG();
+ }
+}
+
+static char *way_string[] = { NULL, "direct mapped", "2-way",
"3-way", "4-way", "5-way", "6-way", "7-way", "8-way"
};
-static void __init probe_pcache(void)
+static void probe_pcache(void)
{
struct cpuinfo_mips *c = &current_cpu_data;
unsigned int config = read_c0_config();
@@ -808,7 +929,7 @@ static void __init probe_pcache(void)
unsigned long config1;
unsigned int lsize;
- switch (c->cputype) {
+ switch (current_cpu_type()) {
case CPU_R4600: /* QED style two way caches? */
case CPU_R4700:
case CPU_R5000:
@@ -816,12 +937,12 @@ static void __init probe_pcache(void)
icache_size = 1 << (12 + ((config & CONF_IC) >> 9));
c->icache.linesz = 16 << ((config & CONF_IB) >> 5);
c->icache.ways = 2;
- c->icache.waybit = ffs(icache_size/2) - 1;
+ c->icache.waybit = __ffs(icache_size/2);
dcache_size = 1 << (12 + ((config & CONF_DC) >> 6));
c->dcache.linesz = 16 << ((config & CONF_DB) >> 4);
c->dcache.ways = 2;
- c->dcache.waybit= ffs(dcache_size/2) - 1;
+ c->dcache.waybit= __ffs(dcache_size/2);
c->options |= MIPS_CPU_CACHE_CDEX_P;
break;
@@ -838,7 +959,7 @@ static void __init probe_pcache(void)
c->dcache.ways = 2;
c->dcache.waybit = 0;
- c->options |= MIPS_CPU_CACHE_CDEX_P;
+ c->options |= MIPS_CPU_CACHE_CDEX_P | MIPS_CPU_PREFETCH;
break;
case CPU_TX49XX:
@@ -853,6 +974,7 @@ static void __init probe_pcache(void)
c->dcache.waybit = 0;
c->options |= MIPS_CPU_CACHE_CDEX_P;
+ c->options |= MIPS_CPU_PREFETCH;
break;
case CPU_R4000PC:
@@ -865,7 +987,7 @@ static void __init probe_pcache(void)
icache_size = 1 << (12 + ((config & CONF_IC) >> 9));
c->icache.linesz = 16 << ((config & CONF_IB) >> 5);
c->icache.ways = 1;
- c->icache.waybit = 0; /* doesn't matter */
+ c->icache.waybit = 0; /* doesn't matter */
dcache_size = 1 << (12 + ((config & CONF_DC) >> 6));
c->dcache.linesz = 16 << ((config & CONF_DB) >> 4);
@@ -877,6 +999,7 @@ static void __init probe_pcache(void)
case CPU_R10000:
case CPU_R12000:
+ case CPU_R14000:
icache_size = 1 << (12 + ((config & R10K_CONF_IC) >> 29));
c->icache.linesz = 64;
c->icache.ways = 2;
@@ -891,26 +1014,27 @@ static void __init probe_pcache(void)
break;
case CPU_VR4133:
- write_c0_config(config & ~CONF_EB);
+ write_c0_config(config & ~VR41_CONF_P4K);
case CPU_VR4131:
/* Workaround for cache instruction bug of VR4131 */
if (c->processor_id == 0x0c80U || c->processor_id == 0x0c81U ||
c->processor_id == 0x0c82U) {
- config &= ~0x00000030U;
- config |= 0x00410000U;
+ config |= 0x00400000U;
+ if (c->processor_id == 0x0c80U)
+ config |= VR41_CONF_BP;
write_c0_config(config);
- }
+ } else
+ c->options |= MIPS_CPU_CACHE_CDEX_P;
+
icache_size = 1 << (10 + ((config & CONF_IC) >> 9));
c->icache.linesz = 16 << ((config & CONF_IB) >> 5);
c->icache.ways = 2;
- c->icache.waybit = ffs(icache_size/2) - 1;
+ c->icache.waybit = __ffs(icache_size/2);
dcache_size = 1 << (10 + ((config & CONF_DC) >> 6));
c->dcache.linesz = 16 << ((config & CONF_DB) >> 4);
c->dcache.ways = 2;
- c->dcache.waybit = ffs(dcache_size/2) - 1;
-
- c->options |= MIPS_CPU_CACHE_CDEX_P;
+ c->dcache.waybit = __ffs(dcache_size/2);
break;
case CPU_VR41XX:
@@ -922,7 +1046,7 @@ static void __init probe_pcache(void)
icache_size = 1 << (10 + ((config & CONF_IC) >> 9));
c->icache.linesz = 16 << ((config & CONF_IB) >> 5);
c->icache.ways = 1;
- c->icache.waybit = 0; /* doesn't matter */
+ c->icache.waybit = 0; /* doesn't matter */
dcache_size = 1 << (10 + ((config & CONF_DC) >> 6));
c->dcache.linesz = 16 << ((config & CONF_DB) >> 4);
@@ -935,20 +1059,77 @@ static void __init probe_pcache(void)
case CPU_RM7000:
rm7k_erratum31();
- case CPU_RM9000:
icache_size = 1 << (12 + ((config & CONF_IC) >> 9));
c->icache.linesz = 16 << ((config & CONF_IB) >> 5);
c->icache.ways = 4;
- c->icache.waybit = ffs(icache_size / c->icache.ways) - 1;
+ c->icache.waybit = __ffs(icache_size / c->icache.ways);
dcache_size = 1 << (12 + ((config & CONF_DC) >> 6));
c->dcache.linesz = 16 << ((config & CONF_DB) >> 4);
c->dcache.ways = 4;
- c->dcache.waybit = ffs(dcache_size / c->dcache.ways) - 1;
+ c->dcache.waybit = __ffs(dcache_size / c->dcache.ways);
-#if !defined(CONFIG_SMP) || !defined(RM9000_CDEX_SMP_WAR)
c->options |= MIPS_CPU_CACHE_CDEX_P;
-#endif
+ c->options |= MIPS_CPU_PREFETCH;
+ break;
+
+ case CPU_LOONGSON2:
+ icache_size = 1 << (12 + ((config & CONF_IC) >> 9));
+ c->icache.linesz = 16 << ((config & CONF_IB) >> 5);
+ if (prid & 0x3)
+ c->icache.ways = 4;
+ else
+ c->icache.ways = 2;
+ c->icache.waybit = 0;
+
+ dcache_size = 1 << (12 + ((config & CONF_DC) >> 6));
+ c->dcache.linesz = 16 << ((config & CONF_DB) >> 4);
+ if (prid & 0x3)
+ c->dcache.ways = 4;
+ else
+ c->dcache.ways = 2;
+ c->dcache.waybit = 0;
+ break;
+
+ case CPU_LOONGSON3:
+ config1 = read_c0_config1();
+ lsize = (config1 >> 19) & 7;
+ if (lsize)
+ c->icache.linesz = 2 << lsize;
+ else
+ c->icache.linesz = 0;
+ c->icache.sets = 64 << ((config1 >> 22) & 7);
+ c->icache.ways = 1 + ((config1 >> 16) & 7);
+ icache_size = c->icache.sets *
+ c->icache.ways *
+ c->icache.linesz;
+ c->icache.waybit = 0;
+
+ lsize = (config1 >> 10) & 7;
+ if (lsize)
+ c->dcache.linesz = 2 << lsize;
+ else
+ c->dcache.linesz = 0;
+ c->dcache.sets = 64 << ((config1 >> 13) & 7);
+ c->dcache.ways = 1 + ((config1 >> 7) & 7);
+ dcache_size = c->dcache.sets *
+ c->dcache.ways *
+ c->dcache.linesz;
+ c->dcache.waybit = 0;
+ break;
+
+ case CPU_CAVIUM_OCTEON3:
+ /* For now lie about the number of ways. */
+ c->icache.linesz = 128;
+ c->icache.sets = 16;
+ c->icache.ways = 8;
+ c->icache.flags |= MIPS_CACHE_VTAG;
+ icache_size = c->icache.sets * c->icache.ways * c->icache.linesz;
+
+ c->dcache.linesz = 128;
+ c->dcache.ways = 8;
+ c->dcache.sets = 8;
+ dcache_size = c->dcache.sets * c->dcache.ways * c->dcache.linesz;
c->options |= MIPS_CPU_PREFETCH;
break;
@@ -962,17 +1143,21 @@ static void __init probe_pcache(void)
*/
config1 = read_c0_config1();
- if ((lsize = ((config1 >> 19) & 7)))
- c->icache.linesz = 2 << lsize;
- else
- c->icache.linesz = lsize;
- c->icache.sets = 64 << ((config1 >> 22) & 7);
+ lsize = (config1 >> 19) & 7;
+
+ /* IL == 7 is reserved */
+ if (lsize == 7)
+ panic("Invalid icache line size");
+
+ c->icache.linesz = lsize ? 2 << lsize : 0;
+
+ c->icache.sets = 32 << (((config1 >> 22) + 1) & 7);
c->icache.ways = 1 + ((config1 >> 16) & 7);
icache_size = c->icache.sets *
- c->icache.ways *
- c->icache.linesz;
- c->icache.waybit = ffs(icache_size/c->icache.ways) - 1;
+ c->icache.ways *
+ c->icache.linesz;
+ c->icache.waybit = __ffs(icache_size/c->icache.ways);
if (config & 0x8) /* VI bit */
c->icache.flags |= MIPS_CACHE_VTAG;
@@ -982,17 +1167,21 @@ static void __init probe_pcache(void)
*/
c->dcache.flags = 0;
- if ((lsize = ((config1 >> 10) & 7)))
- c->dcache.linesz = 2 << lsize;
- else
- c->dcache.linesz= lsize;
- c->dcache.sets = 64 << ((config1 >> 13) & 7);
+ lsize = (config1 >> 10) & 7;
+
+ /* DL == 7 is reserved */
+ if (lsize == 7)
+ panic("Invalid dcache line size");
+
+ c->dcache.linesz = lsize ? 2 << lsize : 0;
+
+ c->dcache.sets = 32 << (((config1 >> 13) + 1) & 7);
c->dcache.ways = 1 + ((config1 >> 7) & 7);
dcache_size = c->dcache.sets *
- c->dcache.ways *
- c->dcache.linesz;
- c->dcache.waybit = ffs(dcache_size/c->dcache.ways) - 1;
+ c->dcache.ways *
+ c->dcache.linesz;
+ c->dcache.waybit = __ffs(dcache_size/c->dcache.ways);
c->options |= MIPS_CPU_PREFETCH;
break;
@@ -1000,13 +1189,14 @@ static void __init probe_pcache(void)
/*
* Processor configuration sanity check for the R4000SC erratum
- * #5. With page sizes larger than 32kB there is no possibility
+ * #5. With page sizes larger than 32kB there is no possibility
* to get a VCE exception anymore so we don't care about this
* misconfiguration. The case is rather theoretical anyway;
* presumably no vendor is shipping his hardware in the "bad"
* configuration.
*/
- if ((prid & 0xff00) == PRID_IMP_R4000 && (prid & 0xff) < 0x40 &&
+ if ((prid & PRID_IMP_MASK) == PRID_IMP_R4000 &&
+ (prid & PRID_REV_MASK) < PRID_REV_R4400 &&
!(config & CONF_SC) && c->icache.linesz != 16 &&
PAGE_SIZE <= 0x8000)
panic("Improper R4000SC processor configuration detected");
@@ -1015,8 +1205,10 @@ static void __init probe_pcache(void)
c->icache.waysize = icache_size / c->icache.ways;
c->dcache.waysize = dcache_size / c->dcache.ways;
- c->icache.sets = icache_size / (c->icache.linesz * c->icache.ways);
- c->dcache.sets = dcache_size / (c->dcache.linesz * c->dcache.ways);
+ c->icache.sets = c->icache.linesz ?
+ icache_size / (c->icache.linesz * c->icache.ways) : 0;
+ c->dcache.sets = c->dcache.linesz ?
+ dcache_size / (c->dcache.linesz * c->dcache.ways) : 0;
/*
* R10000 and R12000 P-caches are odd in a positive way. They're 32kB
@@ -1024,21 +1216,50 @@ static void __init probe_pcache(void)
* normally they'd suffer from aliases but magic in the hardware deals
* with that for us so we don't need to take care ourselves.
*/
- switch (c->cputype) {
+ switch (current_cpu_type()) {
case CPU_20KC:
case CPU_25KF:
+ case CPU_SB1:
+ case CPU_SB1A:
+ case CPU_XLR:
+ c->dcache.flags |= MIPS_CACHE_PINDEX;
+ break;
+
case CPU_R10000:
case CPU_R12000:
- case CPU_SB1:
+ case CPU_R14000:
break;
+
+ case CPU_M14KC:
+ case CPU_M14KEC:
case CPU_24K:
- if (!(read_c0_config7() & (1 << 16)))
+ case CPU_34K:
+ case CPU_74K:
+ case CPU_1004K:
+ case CPU_1074K:
+ case CPU_INTERAPTIV:
+ case CPU_P5600:
+ case CPU_PROAPTIV:
+ case CPU_M5150:
+ if ((c->cputype == CPU_74K) || (c->cputype == CPU_1074K))
+ alias_74k_erratum(c);
+ if (!(read_c0_config7() & MIPS_CONF7_IAR) &&
+ (c->icache.waysize > PAGE_SIZE))
+ c->icache.flags |= MIPS_CACHE_ALIASES;
+ if (read_c0_config7() & MIPS_CONF7_AR) {
+ /*
+ * Effectively physically indexed dcache,
+ * thus no virtual aliases.
+ */
+ c->dcache.flags |= MIPS_CACHE_PINDEX;
+ break;
+ }
default:
- if (c->dcache.waysize > PAGE_SIZE)
- c->dcache.flags |= MIPS_CACHE_ALIASES;
+ if (c->dcache.waysize > PAGE_SIZE)
+ c->dcache.flags |= MIPS_CACHE_ALIASES;
}
- switch (c->cputype) {
+ switch (current_cpu_type()) {
case CPU_20KC:
/*
* Some older 20Kc chips doesn't have the 'VI' bit in
@@ -1047,22 +1268,29 @@ static void __init probe_pcache(void)
c->icache.flags |= MIPS_CACHE_VTAG;
break;
- case CPU_AU1000:
- case CPU_AU1500:
- case CPU_AU1100:
- case CPU_AU1550:
- case CPU_AU1200:
+ case CPU_ALCHEMY:
c->icache.flags |= MIPS_CACHE_IC_F_DC;
break;
+
+ case CPU_LOONGSON2:
+ /*
+ * LOONGSON2 has 4 way icache, but when using indexed cache op,
+ * one op will act on all 4 ways
+ */
+ c->icache.ways = 1;
}
printk("Primary instruction cache %ldkB, %s, %s, linesize %d bytes.\n",
icache_size >> 10,
- cpu_has_vtag_icache ? "virtually tagged" : "physically tagged",
+ c->icache.flags & MIPS_CACHE_VTAG ? "VIVT" : "VIPT",
way_string[c->icache.ways], c->icache.linesz);
- printk("Primary data cache %ldkB, %s, linesize %d bytes.\n",
- dcache_size >> 10, way_string[c->dcache.ways], c->dcache.linesz);
+ printk("Primary data cache %ldkB, %s, %s, %s, linesize %d bytes\n",
+ dcache_size >> 10, way_string[c->dcache.ways],
+ (c->dcache.flags & MIPS_CACHE_PINDEX) ? "PIPT" : "VIPT",
+ (c->dcache.flags & MIPS_CACHE_ALIASES) ?
+ "cache aliases" : "no aliases",
+ c->dcache.linesz);
}
/*
@@ -1071,18 +1299,16 @@ static void __init probe_pcache(void)
* executes in KSEG1 space or else you will crash and burn badly. You have
* been warned.
*/
-static int __init probe_scache(void)
+static int probe_scache(void)
{
- extern unsigned long stext;
unsigned long flags, addr, begin, end, pow2;
unsigned int config = read_c0_config();
struct cpuinfo_mips *c = &current_cpu_data;
- int tmp;
if (config & CONF_SC)
return 0;
- begin = (unsigned long) &stext;
+ begin = (unsigned long) &_stext;
begin &= ~((4 * 1024 * 1024) - 1);
end = begin + (4 * 1024 * 1024);
@@ -1110,7 +1336,6 @@ static int __init probe_scache(void)
/* Now search for the wrap around point. */
pow2 = (128 * 1024);
- tmp = 0;
for (addr = begin + (128 * 1024); addr < end; addr = begin + pow2) {
cache_op(Index_Load_Tag_SD, addr);
__asm__ __volatile__("nop; nop; nop; nop;"); /* hazard... */
@@ -1129,10 +1354,54 @@ static int __init probe_scache(void)
return 1;
}
+static void __init loongson2_sc_init(void)
+{
+ struct cpuinfo_mips *c = &current_cpu_data;
+
+ scache_size = 512*1024;
+ c->scache.linesz = 32;
+ c->scache.ways = 4;
+ c->scache.waybit = 0;
+ c->scache.waysize = scache_size / (c->scache.ways);
+ c->scache.sets = scache_size / (c->scache.linesz * c->scache.ways);
+ pr_info("Unified secondary cache %ldkB %s, linesize %d bytes.\n",
+ scache_size >> 10, way_string[c->scache.ways], c->scache.linesz);
+
+ c->options |= MIPS_CPU_INCLUSIVE_CACHES;
+}
+
+static void __init loongson3_sc_init(void)
+{
+ struct cpuinfo_mips *c = &current_cpu_data;
+ unsigned int config2, lsize;
+
+ config2 = read_c0_config2();
+ lsize = (config2 >> 4) & 15;
+ if (lsize)
+ c->scache.linesz = 2 << lsize;
+ else
+ c->scache.linesz = 0;
+ c->scache.sets = 64 << ((config2 >> 8) & 15);
+ c->scache.ways = 1 + (config2 & 15);
+
+ scache_size = c->scache.sets *
+ c->scache.ways *
+ c->scache.linesz;
+ /* Loongson-3 has 4 cores, 1MB scache for each. scaches are shared */
+ scache_size *= 4;
+ c->scache.waybit = 0;
+ pr_info("Unified secondary cache %ldkB %s, linesize %d bytes.\n",
+ scache_size >> 10, way_string[c->scache.ways], c->scache.linesz);
+ if (scache_size)
+ c->options |= MIPS_CPU_INCLUSIVE_CACHES;
+ return;
+}
+
extern int r5k_sc_init(void);
extern int rm7k_sc_init(void);
+extern int mips_sc_init(void);
-static void __init setup_scache(void)
+static void setup_scache(void)
{
struct cpuinfo_mips *c = &current_cpu_data;
unsigned int config = read_c0_config();
@@ -1141,9 +1410,9 @@ static void __init setup_scache(void)
/*
* Do the probing thing on R4000SC and R4400SC processors. Other
* processors don't have a S-cache that would be relevant to the
- * Linux memory managment.
+ * Linux memory management.
*/
- switch (c->cputype) {
+ switch (current_cpu_type()) {
case CPU_R4000SC:
case CPU_R4000MC:
case CPU_R4400SC:
@@ -1155,6 +1424,7 @@ static void __init setup_scache(void)
case CPU_R10000:
case CPU_R12000:
+ case CPU_R14000:
scache_size = 0x80000 << ((config & R10K_CONF_SS) >> 16);
c->scache.linesz = 64 << ((config >> 13) & 1);
c->scache.ways = 2;
@@ -1167,27 +1437,49 @@ static void __init setup_scache(void)
#ifdef CONFIG_R5000_CPU_SCACHE
r5k_sc_init();
#endif
- return;
+ return;
case CPU_RM7000:
- case CPU_RM9000:
#ifdef CONFIG_RM7000_CPU_SCACHE
rm7k_sc_init();
#endif
return;
+ case CPU_LOONGSON2:
+ loongson2_sc_init();
+ return;
+
+ case CPU_LOONGSON3:
+ loongson3_sc_init();
+ return;
+
+ case CPU_CAVIUM_OCTEON3:
+ case CPU_XLP:
+ /* don't need to worry about L2, fully coherent */
+ return;
+
default:
+ if (c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M32R2 |
+ MIPS_CPU_ISA_M64R1 | MIPS_CPU_ISA_M64R2)) {
+#ifdef CONFIG_MIPS_CPU_SCACHE
+ if (mips_sc_init ()) {
+ scache_size = c->scache.ways * c->scache.sets * c->scache.linesz;
+ printk("MIPS secondary cache %ldkB, %s, linesize %d bytes.\n",
+ scache_size >> 10,
+ way_string[c->scache.ways], c->scache.linesz);
+ }
+#else
+ if (!(c->scache.flags & MIPS_CACHE_NOT_PRESENT))
+ panic("Dunno how to handle MIPS32 / MIPS64 second level cache");
+#endif
+ return;
+ }
sc_present = 0;
}
if (!sc_present)
return;
- if ((c->isa_level == MIPS_CPU_ISA_M32 ||
- c->isa_level == MIPS_CPU_ISA_M64) &&
- !(c->scache.flags & MIPS_CACHE_NOT_PRESENT))
- panic("Dunno how to handle MIPS32 / MIPS64 second level cache");
-
/* compute a couple of other cache variables */
c->scache.waysize = scache_size / c->scache.ways;
@@ -1196,12 +1488,75 @@ static void __init setup_scache(void)
printk("Unified secondary cache %ldkB %s, linesize %d bytes.\n",
scache_size >> 10, way_string[c->scache.ways], c->scache.linesz);
- c->options |= MIPS_CPU_SUBSET_CACHES;
+ c->options |= MIPS_CPU_INCLUSIVE_CACHES;
+}
+
+void au1x00_fixup_config_od(void)
+{
+ /*
+ * c0_config.od (bit 19) was write only (and read as 0)
+ * on the early revisions of Alchemy SOCs. It disables the bus
+ * transaction overlapping and needs to be set to fix various errata.
+ */
+ switch (read_c0_prid()) {
+ case 0x00030100: /* Au1000 DA */
+ case 0x00030201: /* Au1000 HA */
+ case 0x00030202: /* Au1000 HB */
+ case 0x01030200: /* Au1500 AB */
+ /*
+ * Au1100 errata actually keeps silence about this bit, so we set it
+ * just in case for those revisions that require it to be set according
+ * to the (now gone) cpu table.
+ */
+ case 0x02030200: /* Au1100 AB */
+ case 0x02030201: /* Au1100 BA */
+ case 0x02030202: /* Au1100 BC */
+ set_c0_config(1 << 19);
+ break;
+ }
}
-static inline void coherency_setup(void)
+/* CP0 hazard avoidance. */
+#define NXP_BARRIER() \
+ __asm__ __volatile__( \
+ ".set noreorder\n\t" \
+ "nop; nop; nop; nop; nop; nop;\n\t" \
+ ".set reorder\n\t")
+
+static void nxp_pr4450_fixup_config(void)
{
- change_c0_config(CONF_CM_CMASK, CONF_CM_DEFAULT);
+ unsigned long config0;
+
+ config0 = read_c0_config();
+
+ /* clear all three cache coherency fields */
+ config0 &= ~(0x7 | (7 << 25) | (7 << 28));
+ config0 |= (((_page_cachable_default >> _CACHE_SHIFT) << 0) |
+ ((_page_cachable_default >> _CACHE_SHIFT) << 25) |
+ ((_page_cachable_default >> _CACHE_SHIFT) << 28));
+ write_c0_config(config0);
+ NXP_BARRIER();
+}
+
+static int cca = -1;
+
+static int __init cca_setup(char *str)
+{
+ get_option(&str, &cca);
+
+ return 0;
+}
+
+early_param("cca", cca_setup);
+
+static void coherency_setup(void)
+{
+ if (cca < 0 || cca > 7)
+ cca = read_c0_config() & CONF_CM_CMASK;
+ _page_cachable_default = cca << _CACHE_SHIFT;
+
+ pr_debug("Using cache attribute %d\n", cca);
+ change_c0_config(CONF_CM_CMASK, cca);
/*
* c0_status.cu=0 specifies that updates by the sc instruction use
@@ -1210,7 +1565,7 @@ static inline void coherency_setup(void)
* this bit and; some wire it to zero, others like Toshiba had the
* silly idea of putting something else there ...
*/
- switch (current_cpu_data.cputype) {
+ switch (current_cpu_type()) {
case CPU_R4000PC:
case CPU_R4000SC:
case CPU_R4000MC:
@@ -1219,19 +1574,44 @@ static inline void coherency_setup(void)
case CPU_R4400MC:
clear_c0_config(CONF_CU);
break;
+ /*
+ * We need to catch the early Alchemy SOCs with
+ * the write-only co_config.od bit and set it back to one on:
+ * Au1000 rev DA, HA, HB; Au1100 AB, BA, BC, Au1500 AB
+ */
+ case CPU_ALCHEMY:
+ au1x00_fixup_config_od();
+ break;
+
+ case PRID_IMP_PR4450:
+ nxp_pr4450_fixup_config();
+ break;
+ }
+}
+
+static void r4k_cache_error_setup(void)
+{
+ extern char __weak except_vec2_generic;
+ extern char __weak except_vec2_sb1;
+
+ switch (current_cpu_type()) {
+ case CPU_SB1:
+ case CPU_SB1A:
+ set_uncached_handler(0x100, &except_vec2_sb1, 0x80);
+ break;
+
+ default:
+ set_uncached_handler(0x100, &except_vec2_generic, 0x80);
+ break;
}
}
-void __init r4k_cache_init(void)
+void r4k_cache_init(void)
{
extern void build_clear_page(void);
extern void build_copy_page(void);
- extern char except_vec2_generic;
struct cpuinfo_mips *c = &current_cpu_data;
- /* Default cache error handler for R4000 and R5000 family */
- set_uncached_handler (0x100, &except_vec2_generic, 0x80);
-
probe_pcache();
setup_scache();
@@ -1244,36 +1624,86 @@ void __init r4k_cache_init(void)
r4k_blast_scache_page_setup();
r4k_blast_scache_page_indexed_setup();
r4k_blast_scache_setup();
+#ifdef CONFIG_EVA
+ r4k_blast_dcache_user_page_setup();
+ r4k_blast_icache_user_page_setup();
+#endif
/*
* Some MIPS32 and MIPS64 processors have physically indexed caches.
* This code supports virtually indexed processors and will be
* unnecessarily inefficient on physically indexed processors.
*/
- shm_align_mask = max_t( unsigned long,
- c->dcache.sets * c->dcache.linesz - 1,
- PAGE_SIZE - 1);
+ if (c->dcache.linesz)
+ shm_align_mask = max_t( unsigned long,
+ c->dcache.sets * c->dcache.linesz - 1,
+ PAGE_SIZE - 1);
+ else
+ shm_align_mask = PAGE_SIZE-1;
- flush_cache_all = r4k_flush_cache_all;
+ __flush_cache_vmap = r4k__flush_cache_vmap;
+ __flush_cache_vunmap = r4k__flush_cache_vunmap;
+
+ flush_cache_all = cache_noop;
__flush_cache_all = r4k___flush_cache_all;
flush_cache_mm = r4k_flush_cache_mm;
flush_cache_page = r4k_flush_cache_page;
- flush_icache_page = r4k_flush_icache_page;
flush_cache_range = r4k_flush_cache_range;
+ __flush_kernel_vmap_range = r4k_flush_kernel_vmap_range;
+
flush_cache_sigtramp = r4k_flush_cache_sigtramp;
flush_icache_all = r4k_flush_icache_all;
+ local_flush_data_cache_page = local_r4k_flush_data_cache_page;
flush_data_cache_page = r4k_flush_data_cache_page;
flush_icache_range = r4k_flush_icache_range;
+ local_flush_icache_range = local_r4k_flush_icache_range;
-#ifdef CONFIG_DMA_NONCOHERENT
- _dma_cache_wback_inv = r4k_dma_cache_wback_inv;
- _dma_cache_wback = r4k_dma_cache_wback_inv;
- _dma_cache_inv = r4k_dma_cache_inv;
+#if defined(CONFIG_DMA_NONCOHERENT) || defined(CONFIG_DMA_MAYBE_COHERENT)
+ if (coherentio) {
+ _dma_cache_wback_inv = (void *)cache_noop;
+ _dma_cache_wback = (void *)cache_noop;
+ _dma_cache_inv = (void *)cache_noop;
+ } else {
+ _dma_cache_wback_inv = r4k_dma_cache_wback_inv;
+ _dma_cache_wback = r4k_dma_cache_wback_inv;
+ _dma_cache_inv = r4k_dma_cache_inv;
+ }
#endif
build_clear_page();
build_copy_page();
+
+ /*
+ * We want to run CMP kernels on core with and without coherent
+ * caches. Therefore, do not use CONFIG_MIPS_CMP to decide whether
+ * or not to flush caches.
+ */
local_r4k___flush_cache_all(NULL);
+
coherency_setup();
+ board_cache_error_setup = r4k_cache_error_setup;
+}
+
+static int r4k_cache_pm_notifier(struct notifier_block *self, unsigned long cmd,
+ void *v)
+{
+ switch (cmd) {
+ case CPU_PM_ENTER_FAILED:
+ case CPU_PM_EXIT:
+ coherency_setup();
+ break;
+ }
+
+ return NOTIFY_OK;
+}
+
+static struct notifier_block r4k_cache_pm_notifier_block = {
+ .notifier_call = r4k_cache_pm_notifier,
+};
+
+int __init r4k_cache_init_pm(void)
+{
+ return cpu_pm_register_notifier(&r4k_cache_pm_notifier_block);
}
+arch_initcall(r4k_cache_init_pm);
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-12-05 18:47:19 +0000