sh: Migrate from PG_mapped to PG_dcache_dirty.

This inverts the delayed dcache flush a bit to be more in line with other
platforms. At the same time this also gives us the ability to do some
more optimizations and cleanup. Now that the update_mmu_cache() callsite
only tests for the bit, the implementation can gradually be split out and
made generic, rather than relying on special implementations for each of
the peculiar CPU types.

SH7705 in 32kB mode and SH-4 still need slightly different handling, but
this is something that can remain isolated in the varying page copy/clear
routines. On top of that, SH-X3 is dcache coherent, so there is no need
to bother with any of these tests in the PTEAEX version of
update_mmu_cache(), so we kill that off too.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>

11 files changed, 81 insertions, 142 deletions

diff --git a/arch/sh/include/asm/page.h b/arch/sh/include/asm/page.h
index 49592c780a6e..a31ab40040f0 100644
--- a/arch/sh/include/asm/page.h
+++ b/arch/sh/include/asm/page.h
@@ -50,6 +50,12 @@ extern unsigned long shm_align_mask;
 extern unsigned long max_low_pfn, min_low_pfn;
 extern unsigned long memory_start, memory_end;
 
+static inline unsigned long
+pages_do_alias(unsigned long addr1, unsigned long addr2)
+{
+	return (addr1 ^ addr2) & shm_align_mask;
+}
+
 extern void clear_page(void *to);
 extern void copy_page(void *to, void *from);
 
diff --git a/arch/sh/include/asm/pgtable.h b/arch/sh/include/asm/pgtable.h
index 2a011b18090b..d9f68f9c3cb3 100644
--- a/arch/sh/include/asm/pgtable.h
+++ b/arch/sh/include/asm/pgtable.h
@@ -133,13 +133,6 @@ typedef pte_t *pte_addr_t;
  */
 #define pgtable_cache_init()	do { } while (0)
 
-#if !defined(CONFIG_CACHE_OFF) && (defined(CONFIG_CPU_SH4) || \
-	defined(CONFIG_SH7705_CACHE_32KB))
-struct mm_struct;
-#define __HAVE_ARCH_PTEP_GET_AND_CLEAR
-pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep);
-#endif
-
 struct vm_area_struct;
 extern void update_mmu_cache(struct vm_area_struct * vma,
 			     unsigned long address, pte_t pte);
diff --git a/arch/sh/include/cpu-sh3/cpu/cacheflush.h b/arch/sh/include/cpu-sh3/cpu/cacheflush.h
index 1ac27aae6700..6485ad5649ad 100644
--- a/arch/sh/include/cpu-sh3/cpu/cacheflush.h
+++ b/arch/sh/include/cpu-sh3/cpu/cacheflush.h
@@ -15,10 +15,7 @@
  * SH4. Unlike the SH4 this is a unified cache so we need to do some work
  * in mmap when 'exec'ing a new binary
  */
- /* 32KB cache, 4kb PAGE sizes need to check bit 12 */
-#define CACHE_ALIAS 0x00001000
-
-#define PG_mapped	PG_arch_1
+#define PG_dcache_dirty	PG_arch_1
 
 void flush_cache_all(void);
 void flush_cache_mm(struct mm_struct *mm);
diff --git a/arch/sh/include/cpu-sh4/cpu/cacheflush.h b/arch/sh/include/cpu-sh4/cpu/cacheflush.h
index 065306d376eb..3564f1722195 100644
--- a/arch/sh/include/cpu-sh4/cpu/cacheflush.h
+++ b/arch/sh/include/cpu-sh4/cpu/cacheflush.h
@@ -38,6 +38,6 @@ void flush_icache_user_range(struct vm_area_struct *vma, struct page *page,
 /* Initialization of P3 area for copy_user_page */
 void p3_cache_init(void);
 
-#define PG_mapped	PG_arch_1
+#define PG_dcache_dirty	PG_arch_1
 
 #endif /* __ASM_CPU_SH4_CACHEFLUSH_H */
diff --git a/arch/sh/mm/cache-sh4.c b/arch/sh/mm/cache-sh4.c
index 5cfe08dbb59e..c3a09b27f8d5 100644
--- a/arch/sh/mm/cache-sh4.c
+++ b/arch/sh/mm/cache-sh4.c
@@ -14,6 +14,7 @@
 #include <linux/mm.h>
 #include <linux/io.h>
 #include <linux/mutex.h>
+#include <linux/fs.h>
 #include <asm/mmu_context.h>
 #include <asm/cacheflush.h>
 
@@ -246,7 +247,14 @@ static inline void flush_cache_4096(unsigned long start,
  */
 void flush_dcache_page(struct page *page)
 {
-	if (test_bit(PG_mapped, &page->flags)) {
+	struct address_space *mapping = page_mapping(page);
+
+#ifndef CONFIG_SMP
+	if (mapping && !mapping_mapped(mapping))
+		set_bit(PG_dcache_dirty, &page->flags);
+	else
+#endif
+	{
 		unsigned long phys = PHYSADDR(page_address(page));
 		unsigned long addr = CACHE_OC_ADDRESS_ARRAY;
 		int i, n;
diff --git a/arch/sh/mm/cache-sh7705.c b/arch/sh/mm/cache-sh7705.c
index 22dacc778823..fa37bff306b9 100644
--- a/arch/sh/mm/cache-sh7705.c
+++ b/arch/sh/mm/cache-sh7705.c
@@ -12,6 +12,7 @@
 #include <linux/init.h>
 #include <linux/mman.h>
 #include <linux/mm.h>
+#include <linux/fs.h>
 #include <linux/threads.h>
 #include <asm/addrspace.h>
 #include <asm/page.h>
@@ -128,7 +129,11 @@ static void __uses_jump_to_uncached __flush_dcache_page(unsigned long phys)
  */
 void flush_dcache_page(struct page *page)
 {
-	if (test_bit(PG_mapped, &page->flags))
+	struct address_space *mapping = page_mapping(page);
+
+	if (mapping && !mapping_mapped(mapping))
+		set_bit(PG_dcache_dirty, &page->flags);
+	else
 		__flush_dcache_page(PHYSADDR(page_address(page)));
 }
 
diff --git a/arch/sh/mm/pg-sh4.c b/arch/sh/mm/pg-sh4.c
index 2fe14da1f839..f3c4b2a54fc7 100644
--- a/arch/sh/mm/pg-sh4.c
+++ b/arch/sh/mm/pg-sh4.c
@@ -15,8 +15,6 @@
 #include <asm/mmu_context.h>
 #include <asm/cacheflush.h>
 
-#define CACHE_ALIAS (current_cpu_data.dcache.alias_mask)
-
 #define kmap_get_fixmap_pte(vaddr)                                     \
 	pte_offset_kernel(pmd_offset(pud_offset(pgd_offset_k(vaddr), (vaddr)), (vaddr)), (vaddr))
 
@@ -68,10 +66,9 @@ static inline void kunmap_coherent(struct page *page)
  */
 void clear_user_page(void *to, unsigned long address, struct page *page)
 {
-	__set_bit(PG_mapped, &page->flags);
-
 	clear_page(to);
-	if ((((address & PAGE_MASK) ^ (unsigned long)to) & CACHE_ALIAS))
+
+	if (pages_do_alias((unsigned long)to, address & PAGE_MASK))
 		__flush_wback_region(to, PAGE_SIZE);
 }
 
@@ -79,13 +76,14 @@ void copy_to_user_page(struct vm_area_struct *vma, struct page *page,
 		       unsigned long vaddr, void *dst, const void *src,
 		       unsigned long len)
 {
-	void *vto;
-
-	__set_bit(PG_mapped, &page->flags);
-
-	vto = kmap_coherent(page, vaddr) + (vaddr & ~PAGE_MASK);
-	memcpy(vto, src, len);
-	kunmap_coherent(vto);
+	if (page_mapped(page) && !test_bit(PG_dcache_dirty, &page->flags)) {
+		void *vto = kmap_coherent(page, vaddr) + (vaddr & ~PAGE_MASK);
+		memcpy(vto, src, len);
+		kunmap_coherent(vto);
+	} else {
+		memcpy(dst, src, len);
+		set_bit(PG_dcache_dirty, &page->flags);
+	}
 
 	if (vma->vm_flags & VM_EXEC)
 		flush_cache_page(vma, vaddr, page_to_pfn(page));
@@ -95,13 +93,14 @@ void copy_from_user_page(struct vm_area_struct *vma, struct page *page,
 			 unsigned long vaddr, void *dst, const void *src,
 			 unsigned long len)
 {
-	void *vfrom;
-
-	__set_bit(PG_mapped, &page->flags);
-
-	vfrom = kmap_coherent(page, vaddr) + (vaddr & ~PAGE_MASK);
-	memcpy(dst, vfrom, len);
-	kunmap_coherent(vfrom);
+	if (page_mapped(page) && !test_bit(PG_dcache_dirty, &page->flags)) {
+		void *vfrom = kmap_coherent(page, vaddr) + (vaddr & ~PAGE_MASK);
+		memcpy(dst, vfrom, len);
+		kunmap_coherent(vfrom);
+	} else {
+		memcpy(dst, src, len);
+		set_bit(PG_dcache_dirty, &page->flags);
+	}
 }
 
 void copy_user_highpage(struct page *to, struct page *from,
@@ -109,14 +108,19 @@ void copy_user_highpage(struct page *to, struct page *from,
 {
 	void *vfrom, *vto;
 
-	__set_bit(PG_mapped, &to->flags);
-
 	vto = kmap_atomic(to, KM_USER1);
-	vfrom = kmap_coherent(from, vaddr);
-	copy_page(vto, vfrom);
-	kunmap_coherent(vfrom);
 
-	if (((vaddr ^ (unsigned long)vto) & CACHE_ALIAS))
+	if (page_mapped(from) && !test_bit(PG_dcache_dirty, &from->flags)) {
+		vfrom = kmap_coherent(from, vaddr);
+		copy_page(vto, vfrom);
+		kunmap_coherent(vfrom);
+	} else {
+		vfrom = kmap_atomic(from, KM_USER0);
+		copy_page(vto, vfrom);
+		kunmap_atomic(vfrom, KM_USER0);
+	}
+
+	if (pages_do_alias((unsigned long)vto, vaddr & PAGE_MASK))
 		__flush_wback_region(vto, PAGE_SIZE);
 
 	kunmap_atomic(vto, KM_USER1);
@@ -124,23 +128,3 @@ void copy_user_highpage(struct page *to, struct page *from,
 	smp_wmb();
 }
 EXPORT_SYMBOL(copy_user_highpage);
-
-/*
- * For SH-4, we have our own implementation for ptep_get_and_clear
- */
-pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
-{
-	pte_t pte = *ptep;
-
-	pte_clear(mm, addr, ptep);
-	if (!pte_not_present(pte)) {
-		unsigned long pfn = pte_pfn(pte);
-		if (pfn_valid(pfn)) {
-			struct page *page = pfn_to_page(pfn);
-			struct address_space *mapping = page_mapping(page);
-			if (!mapping || !mapping_writably_mapped(mapping))
-				__clear_bit(PG_mapped, &page->flags);
-		}
-	}
-	return pte;
-}
diff --git a/arch/sh/mm/pg-sh7705.c b/arch/sh/mm/pg-sh7705.c
index eaf25147194c..684891b5c8c0 100644
--- a/arch/sh/mm/pg-sh7705.c
+++ b/arch/sh/mm/pg-sh7705.c
@@ -26,7 +26,7 @@
 #include <asm/mmu_context.h>
 #include <asm/cacheflush.h>
 
-static inline void __flush_purge_virtual_region(void *p1, void *virt, int size)
+static void __flush_purge_virtual_region(void *p1, void *virt, int size)
 {
 	unsigned long v;
 	unsigned long begin, end;
@@ -75,19 +75,13 @@ static inline void __flush_purge_virtual_region(void *p1, void *virt, int size)
  */
 void clear_user_page(void *to, unsigned long address, struct page *pg)
 {
-	struct page *page = virt_to_page(to);
-
-	__set_bit(PG_mapped, &page->flags);
-	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0) {
-		clear_page(to);
-		__flush_wback_region(to, PAGE_SIZE);
-	} else {
+	if (pages_do_alias(address, (unsigned long)to))
 		__flush_purge_virtual_region(to,
 					     (void *)(address & 0xfffff000),
 					     PAGE_SIZE);
-		clear_page(to);
-		__flush_wback_region(to, PAGE_SIZE);
-	}
+
+	clear_page(to);
+	__flush_wback_region(to, PAGE_SIZE);
 }
 
 /*
@@ -98,41 +92,11 @@ void clear_user_page(void *to, unsigned long address, struct page *pg)
  */
 void copy_user_page(void *to, void *from, unsigned long address, struct page *pg)
 {
-	struct page *page = virt_to_page(to);
-
-
-	__set_bit(PG_mapped, &page->flags);
-	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0) {
-		copy_page(to, from);
-		__flush_wback_region(to, PAGE_SIZE);
-	} else {
+	if (pages_do_alias(address, (unsigned long)to))
 		__flush_purge_virtual_region(to,
 					     (void *)(address & 0xfffff000),
 					     PAGE_SIZE);
-		copy_page(to, from);
-		__flush_wback_region(to, PAGE_SIZE);
-	}
-}
 
-/*
- * For SH7705, we have our own implementation for ptep_get_and_clear
- * Copied from pg-sh4.c
- */
-pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
-{
-	pte_t pte = *ptep;
-
-	pte_clear(mm, addr, ptep);
-	if (!pte_not_present(pte)) {
-		unsigned long pfn = pte_pfn(pte);
-		if (pfn_valid(pfn)) {
-			struct page *page = pfn_to_page(pfn);
-			struct address_space *mapping = page_mapping(page);
-			if (!mapping || !mapping_writably_mapped(mapping))
-				__clear_bit(PG_mapped, &page->flags);
-		}
-	}
-
-	return pte;
+	copy_page(to, from);
+	__flush_wback_region(to, PAGE_SIZE);
 }
-
diff --git a/arch/sh/mm/tlb-pteaex.c b/arch/sh/mm/tlb-pteaex.c
index 2aab3ea934d7..c39b77363352 100644
--- a/arch/sh/mm/tlb-pteaex.c
+++ b/arch/sh/mm/tlb-pteaex.c
@@ -27,23 +27,6 @@ void update_mmu_cache(struct vm_area_struct * vma,
 	if (vma && current->active_mm != vma->vm_mm)
 		return;
 
-#ifndef CONFIG_CACHE_OFF
-	{
-		unsigned long pfn = pte_pfn(pte);
-
-		if (pfn_valid(pfn)) {
-			struct page *page = pfn_to_page(pfn);
-
-			if (!test_bit(PG_mapped, &page->flags)) {
-				unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
-				__flush_wback_region((void *)P1SEGADDR(phys),
-						     PAGE_SIZE);
-				__set_bit(PG_mapped, &page->flags);
-			}
-		}
-	}
-#endif
-
 	local_irq_save(flags);
 
 	/* Set PTEH register */
diff --git a/arch/sh/mm/tlb-sh3.c b/arch/sh/mm/tlb-sh3.c
index 17cb7c3adf22..9b8459c74abd 100644
--- a/arch/sh/mm/tlb-sh3.c
+++ b/arch/sh/mm/tlb-sh3.c
@@ -33,25 +33,25 @@ void update_mmu_cache(struct vm_area_struct * vma,
 	unsigned long flags;
 	unsigned long pteval;
 	unsigned long vpn;
+	unsigned long pfn = pte_pfn(pte);
+	struct page *page;
 
 	/* Ptrace may call this routine. */
 	if (vma && current->active_mm != vma->vm_mm)
 		return;
 
+	page = pfn_to_page(pfn);
+	if (pfn_valid(pfn) && page_mapping(page)) {
 #if defined(CONFIG_SH7705_CACHE_32KB)
-	{
-		struct page *page = pte_page(pte);
-		unsigned long pfn = pte_pfn(pte);
+		int dirty = test_and_clear_bit(PG_dcache_dirty, &page->flags);
+		if (dirty) {
+			unsigned long addr = (unsigned long)page_address(page);
 
-		if (pfn_valid(pfn) && !test_bit(PG_mapped, &page->flags)) {
-			unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
-
-			__flush_wback_region((void *)P1SEGADDR(phys),
-					     PAGE_SIZE);
-			__set_bit(PG_mapped, &page->flags);
+			if (pages_do_alias(addr, address & PAGE_MASK))
+				__flush_wback_region((void *)addr, PAGE_SIZE);
 		}
-	}
 #endif
+	}
 
 	local_irq_save(flags);
 
diff --git a/arch/sh/mm/tlb-sh4.c b/arch/sh/mm/tlb-sh4.c
index f0c7b7397fa6..cf50082d2435 100644
--- a/arch/sh/mm/tlb-sh4.c
+++ b/arch/sh/mm/tlb-sh4.c
@@ -21,27 +21,26 @@ void update_mmu_cache(struct vm_area_struct * vma,
 	unsigned long flags;
 	unsigned long pteval;
 	unsigned long vpn;
+	unsigned long pfn = pte_pfn(pte);
+	struct page *page;
 
 	/* Ptrace may call this routine. */
 	if (vma && current->active_mm != vma->vm_mm)
 		return;
 
-#ifndef CONFIG_CACHE_OFF
-	{
-		unsigned long pfn = pte_pfn(pte);
+	page = pfn_to_page(pfn);
+	if (pfn_valid(pfn) && page_mapping(page)) {
+#ifndef CONFIG_SMP
+		int dirty = test_and_clear_bit(PG_dcache_dirty, &page->flags);
+		if (dirty) {
 
-		if (pfn_valid(pfn)) {
-			struct page *page = pfn_to_page(pfn);
+			unsigned long addr = (unsigned long)page_address(page);
 
-			if (!test_bit(PG_mapped, &page->flags)) {
-				unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
-				__flush_wback_region((void *)P1SEGADDR(phys),
-						     PAGE_SIZE);
-				__set_bit(PG_mapped, &page->flags);
-			}
+			if (pages_do_alias(addr, address & PAGE_MASK))
+				__flush_wback_region((void *)addr, PAGE_SIZE);
 		}
-	}
 #endif
+	}
 
 	local_irq_save(flags);