sh: Revert lazy dcache writeback changes.

These ended up causing too many problems on older parts,
revert for now..

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

13 files changed, 194 insertions, 85 deletions

diff --git a/arch/sh/boards/renesas/r7780rp/io.c b/arch/sh/boards/renesas/r7780rp/io.c
index f74d2ffb385..86dfe85ec44 100644
--- a/arch/sh/boards/renesas/r7780rp/io.c
+++ b/arch/sh/boards/renesas/r7780rp/io.c
@@ -156,8 +156,6 @@ void r7780rp_insw(unsigned long port, void *dst, unsigned long count)
 
 	while (count--)
 		*buf++ = *p;
-
-	flush_dcache_all();
 }
 
 void r7780rp_insl(unsigned long port, void *dst, unsigned long count)
@@ -204,8 +202,6 @@ void r7780rp_outsw(unsigned long port, const void *src, unsigned long count)
 
 	while (count--)
 		*p = *buf++;
-
-	flush_dcache_all();
 }
 
 void r7780rp_outsl(unsigned long port, const void *src, unsigned long count)
diff --git a/arch/sh/kernel/io_generic.c b/arch/sh/kernel/io_generic.c
index 66626c03e1e..771ea423044 100644
--- a/arch/sh/kernel/io_generic.c
+++ b/arch/sh/kernel/io_generic.c
@@ -14,7 +14,6 @@
 #include <linux/module.h>
 #include <linux/io.h>
 #include <asm/machvec.h>
-#include <asm/cacheflush.h>
 
 #ifdef CONFIG_CPU_SH3
 /* SH3 has a PCMCIA bug that needs a dummy read from area 6 for a
@@ -96,7 +95,6 @@ void generic_insw(unsigned long port, void *dst, unsigned long count)
 	while (count--)
 		*buf++ = *port_addr;
 
-	flush_dcache_all();
 	dummy_read();
 }
 
@@ -171,7 +169,6 @@ void generic_outsw(unsigned long port, const void *src, unsigned long count)
 	while (count--)
 		*port_addr = *buf++;
 
-	flush_dcache_all();
 	dummy_read();
 }
 
diff --git a/arch/sh/mm/cache-sh4.c b/arch/sh/mm/cache-sh4.c
index e0cd4b7f4ae..981b0408905 100644
--- a/arch/sh/mm/cache-sh4.c
+++ b/arch/sh/mm/cache-sh4.c
@@ -237,20 +237,10 @@ static inline void flush_cache_4096(unsigned long start,
 /*
  * Write back & invalidate the D-cache of the page.
  * (To avoid "alias" issues)
- *
- * This uses a lazy write-back on UP, which is explicitly
- * disabled on SMP.
  */
 void flush_dcache_page(struct page *page)
 {
-#ifndef CONFIG_SMP
-	struct address_space *mapping = page_mapping(page);
-
-	if (mapping && !mapping_mapped(mapping))
-		set_bit(PG_dcache_dirty, &page->flags);
-	else
-#endif
-	{
+	if (test_bit(PG_mapped, &page->flags)) {
 		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 31f8deb7a15..4896d737692 100644
--- a/arch/sh/mm/cache-sh7705.c
+++ b/arch/sh/mm/cache-sh7705.c
@@ -3,11 +3,11 @@
  *
  * Copyright (C) 1999, 2000  Niibe Yutaka
  * Copyright (C) 2004  Alex Song
- * Copyright (C) 2006  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/init.h>
 #include <linux/mman.h>
@@ -51,6 +51,7 @@ static inline void cache_wback_all(void)
 
 			if ((data & v) == v)
 				ctrl_outl(data & ~v, addr);
+
 		}
 
 		addrstart += current_cpu_data.dcache.way_incr;
@@ -127,11 +128,7 @@ static void __flush_dcache_page(unsigned long phys)
  */
 void flush_dcache_page(struct page *page)
 {
-	struct address_space *mapping = page_mapping(page);
-
-	if (mapping && !mapping_mapped(mapping))
-		set_bit(PG_dcache_dirty, &page->flags);
-	else
+	if (test_bit(PG_mapped, &page->flags))
 		__flush_dcache_page(PHYSADDR(page_address(page)));
 }
 
diff --git a/arch/sh/mm/pg-sh4.c b/arch/sh/mm/pg-sh4.c
index 969efeceb92..df69da9ca69 100644
--- a/arch/sh/mm/pg-sh4.c
+++ b/arch/sh/mm/pg-sh4.c
@@ -23,6 +23,7 @@ extern struct mutex p3map_mutex[];
  */
 void clear_user_page(void *to, unsigned long address, struct page *page)
 {
+	__set_bit(PG_mapped, &page->flags);
 	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0)
 		clear_page(to);
 	else {
@@ -58,6 +59,7 @@ void clear_user_page(void *to, unsigned long address, struct page *page)
 void copy_user_page(void *to, void *from, unsigned long address,
 		    struct page *page)
 {
+	__set_bit(PG_mapped, &page->flags);
 	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0)
 		copy_page(to, from);
 	else {
@@ -82,3 +84,23 @@ void copy_user_page(void *to, void *from, unsigned long address,
 		mutex_unlock(&p3map_mutex[(address & CACHE_ALIAS)>>12]);
 	}
 }
+
+/*
+ * For SH-4, we have our own implementation for ptep_get_and_clear
+ */
+inline 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 887ab9d18cc..a4b015f95a3 100644
--- a/arch/sh/mm/pg-sh7705.c
+++ b/arch/sh/mm/pg-sh7705.c
@@ -7,7 +7,9 @@
  * 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/init.h>
 #include <linux/mman.h>
 #include <linux/mm.h>
@@ -74,6 +76,7 @@ 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);
@@ -92,11 +95,12 @@ void clear_user_page(void *to, unsigned long address, struct page *pg)
  * @from: P1 address
  * @address: U0 address to be mapped
  */
-void copy_user_page(void *to, void *from, 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);
@@ -108,3 +112,26 @@ void copy_user_page(void *to, void *from, unsigned long address,
 		__flush_wback_region(to, PAGE_SIZE);
 	}
 }
+
+/*
+ * For SH7705, we have our own implementation for ptep_get_and_clear
+ * Copied from pg-sh4.c
+ */
+inline 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/tlb-flush.c b/arch/sh/mm/tlb-flush.c
index d2f7b4a2eb0..6f45c1f8a7f 100644
--- a/arch/sh/mm/tlb-flush.c
+++ b/arch/sh/mm/tlb-flush.c
@@ -2,17 +2,15 @@
  * TLB flushing operations for SH with an MMU.
  *
  *  Copyright (C) 1999  Niibe Yutaka
- *  Copyright (C) 2003 - 2006  Paul Mundt
+ *  Copyright (C) 2003  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/mm.h>
-#include <linux/io.h>
 #include <asm/mmu_context.h>
 #include <asm/tlbflush.h>
-#include <asm/cacheflush.h>
 
 void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
 {
@@ -140,54 +138,3 @@ void local_flush_tlb_all(void)
 	ctrl_barrier();
 	local_irq_restore(flags);
 }
-
-void update_mmu_cache(struct vm_area_struct *vma,
-		      unsigned long address, pte_t pte)
-{
-	unsigned long flags;
-	unsigned long pteval;
-	unsigned long vpn;
-	struct page *page;
-	unsigned long pfn = pte_pfn(pte);
-	struct address_space *mapping;
-
-	if (!pfn_valid(pfn))
-		return;
-
-	page = pfn_to_page(pfn);
-	mapping = page_mapping(page);
-	if (mapping) {
-		unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
-		int dirty = test_and_clear_bit(PG_dcache_dirty, &page->flags);
-
-		if (dirty)
-			__flush_wback_region((void *)P1SEGADDR(phys),
-					     PAGE_SIZE);
-	}
-
-	local_irq_save(flags);
-
-	/* Set PTEH register */
-	vpn = (address & MMU_VPN_MASK) | get_asid();
-	ctrl_outl(vpn, MMU_PTEH);
-
-	pteval = pte_val(pte);
-
-#ifdef CONFIG_CPU_HAS_PTEA
-	/* Set PTEA register */
-	/* TODO: make this look less hacky */
-	ctrl_outl(((pteval >> 28) & 0xe) | (pteval & 0x1), MMU_PTEA);
-#endif
-
-	/* Set PTEL register */
-	pteval &= _PAGE_FLAGS_HARDWARE_MASK; /* drop software flags */
-#if defined(CONFIG_SH_WRITETHROUGH) && defined(CONFIG_CPU_SH4)
-	pteval |= _PAGE_WT;
-#endif
-	/* conveniently, we want all the software flags to be 0 anyway */
-	ctrl_outl(pteval, MMU_PTEL);
-
-	/* Load the TLB */
-	asm volatile("ldtlb": /* no output */ : /* no input */ : "memory");
-	local_irq_restore(flags);
-}
diff --git a/arch/sh/mm/tlb-sh3.c b/arch/sh/mm/tlb-sh3.c
index e5e76eb7ee0..7fbfd5a11ff 100644
--- a/arch/sh/mm/tlb-sh3.c
+++ b/arch/sh/mm/tlb-sh3.c
@@ -8,9 +8,69 @@
  *
  * Released under the terms of the GNU GPL v2.0.
  */
-#include <linux/io.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+
 #include <asm/system.h>
+#include <asm/io.h>
+#include <asm/uaccess.h>
+#include <asm/pgalloc.h>
 #include <asm/mmu_context.h>
+#include <asm/cacheflush.h>
+
+void update_mmu_cache(struct vm_area_struct * vma,
+		      unsigned long address, pte_t pte)
+{
+	unsigned long flags;
+	unsigned long pteval;
+	unsigned long vpn;
+
+	/* Ptrace may call this routine. */
+	if (vma && current->active_mm != vma->vm_mm)
+		return;
+
+#if defined(CONFIG_SH7705_CACHE_32KB)
+	{
+		struct page *page = pte_page(pte);
+		unsigned long pfn = pte_pfn(pte);
+
+		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);
+		}
+	}
+#endif
+
+	local_irq_save(flags);
+
+	/* Set PTEH register */
+	vpn = (address & MMU_VPN_MASK) | get_asid();
+	ctrl_outl(vpn, MMU_PTEH);
+
+	pteval = pte_val(pte);
+
+	/* Set PTEL register */
+	pteval &= _PAGE_FLAGS_HARDWARE_MASK; /* drop software flags */
+	/* conveniently, we want all the software flags to be 0 anyway */
+	ctrl_outl(pteval, MMU_PTEL);
+
+	/* Load the TLB */
+	asm volatile("ldtlb": /* no output */ : /* no input */ : "memory");
+	local_irq_restore(flags);
+}
 
 void local_flush_tlb_one(unsigned long asid, unsigned long page)
 {
@@ -34,3 +94,4 @@ void local_flush_tlb_one(unsigned long asid, unsigned long page)
 	for (i = 0; i < ways; i++)
 		ctrl_outl(data, addr + (i << 8));
 }
+
diff --git a/arch/sh/mm/tlb-sh4.c b/arch/sh/mm/tlb-sh4.c
index 221e7095473..f74cf667c8f 100644
--- a/arch/sh/mm/tlb-sh4.c
+++ b/arch/sh/mm/tlb-sh4.c
@@ -8,9 +8,74 @@
  *
  * Released under the terms of the GNU GPL v2.0.
  */
-#include <linux/io.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+
 #include <asm/system.h>
+#include <asm/io.h>
+#include <asm/uaccess.h>
+#include <asm/pgalloc.h>
 #include <asm/mmu_context.h>
+#include <asm/cacheflush.h>
+
+void update_mmu_cache(struct vm_area_struct * vma,
+		      unsigned long address, pte_t pte)
+{
+	unsigned long flags;
+	unsigned long pteval;
+	unsigned long vpn;
+	struct page *page;
+	unsigned long pfn;
+
+	/* Ptrace may call this routine. */
+	if (vma && current->active_mm != vma->vm_mm)
+		return;
+
+	pfn = pte_pfn(pte);
+	if (pfn_valid(pfn)) {
+		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);
+		}
+	}
+
+	local_irq_save(flags);
+
+	/* Set PTEH register */
+	vpn = (address & MMU_VPN_MASK) | get_asid();
+	ctrl_outl(vpn, MMU_PTEH);
+
+	pteval = pte_val(pte);
+
+	/* Set PTEA register */
+	if (cpu_data->flags & CPU_HAS_PTEA)
+		/* TODO: make this look less hacky */
+		ctrl_outl(((pteval >> 28) & 0xe) | (pteval & 0x1), MMU_PTEA);
+
+	/* Set PTEL register */
+	pteval &= _PAGE_FLAGS_HARDWARE_MASK; /* drop software flags */
+#ifdef CONFIG_SH_WRITETHROUGH
+	pteval |= _PAGE_WT;
+#endif
+	/* conveniently, we want all the software flags to be 0 anyway */
+	ctrl_outl(pteval, MMU_PTEL);
+
+	/* Load the TLB */
+	asm volatile("ldtlb": /* no output */ : /* no input */ : "memory");
+	local_irq_restore(flags);
+}
 
 void local_flush_tlb_one(unsigned long asid, unsigned long page)
 {
@@ -28,3 +93,4 @@ void local_flush_tlb_one(unsigned long asid, unsigned long page)
 	ctrl_outl(data, addr);
 	back_to_P1();
 }
+
diff --git a/include/asm-sh/cacheflush.h b/include/asm-sh/cacheflush.h
index 22f12634975..07f62ec9ff0 100644
--- a/include/asm-sh/cacheflush.h
+++ b/include/asm-sh/cacheflush.h
@@ -30,8 +30,5 @@ extern void __flush_invalidate_region(void *start, int size);
 
 #define HAVE_ARCH_UNMAPPED_AREA
 
-/* Page flag for lazy dcache write-back for the aliasing UP caches */
-#define PG_dcache_dirty	PG_arch_1
-
 #endif /* __KERNEL__ */
 #endif /* __ASM_SH_CACHEFLUSH_H */
diff --git a/include/asm-sh/cpu-sh3/cacheflush.h b/include/asm-sh/cpu-sh3/cacheflush.h
index 6fabbba228d..f70d8ef76a1 100644
--- a/include/asm-sh/cpu-sh3/cacheflush.h
+++ b/include/asm-sh/cpu-sh3/cacheflush.h
@@ -36,6 +36,8 @@
  /* 32KB cache, 4kb PAGE sizes need to check bit 12 */
 #define CACHE_ALIAS 0x00001000
 
+#define PG_mapped	PG_arch_1
+
 void flush_cache_all(void);
 void flush_cache_mm(struct mm_struct *mm);
 #define flush_cache_dup_mm(mm) flush_cache_mm(mm)
diff --git a/include/asm-sh/cpu-sh4/cacheflush.h b/include/asm-sh/cpu-sh4/cacheflush.h
index b3746a936a0..5fd5c89ef86 100644
--- a/include/asm-sh/cpu-sh4/cacheflush.h
+++ b/include/asm-sh/cpu-sh4/cacheflush.h
@@ -39,4 +39,6 @@ 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
+
 #endif /* __ASM_CPU_SH4_CACHEFLUSH_H */
diff --git a/include/asm-sh/pgtable.h b/include/asm-sh/pgtable.h
index 9214c015fe1..184d7fcaaf1 100644
--- a/include/asm-sh/pgtable.h
+++ b/include/asm-sh/pgtable.h
@@ -583,6 +583,11 @@ struct mm_struct;
 extern unsigned int kobjsize(const void *objp);
 #endif /* !CONFIG_MMU */
 
+#if defined(CONFIG_CPU_SH4) || defined(CONFIG_SH7705_CACHE_32KB)
+#define __HAVE_ARCH_PTEP_GET_AND_CLEAR
+extern pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep);
+#endif
+
 extern pgd_t swapper_pg_dir[PTRS_PER_PGD];
 extern void paging_init(void);