Make generic TLB shootdown friendlier to non-x86 architecturesv2.5.17

3 files changed, 35 insertions, 10 deletions

diff --git a/include/asm-generic/tlb.h b/include/asm-generic/tlb.h
index 778990a36b41f8..6e1aabd52becc9 100644
--- a/include/asm-generic/tlb.h
+++ b/include/asm-generic/tlb.h
@@ -16,8 +16,17 @@
 #include <linux/config.h>
 #include <asm/tlbflush.h>
 
-/* aim for something that fits in the L1 cache */
-#define FREE_PTE_NR	508
+/*
+ * For UP we don't need to worry about TLB flush
+ * and page free order so much..
+ */
+#ifdef CONFIG_SMP
+  #define FREE_PTE_NR	507
+  #define tlb_fast_mode(tlb) ((tlb)->nr == ~0UL) 
+#else
+  #define FREE_PTE_NR	1
+  #define tlb_fast_mode(tlb) 1
+#endif
 
 /* mmu_gather_t is an opaque type used by the mm code for passing around any
  * data needed by arch specific code for tlb_remove_page.  This structure can
@@ -34,10 +43,6 @@ typedef struct free_pte_ctx {
 /* Users of the generic TLB shootdown code must declare this storage space. */
 extern mmu_gather_t	mmu_gathers[NR_CPUS];
 
-/* Do me later */
-#define tlb_start_vma(tlb, vma) do { } while (0)
-#define tlb_end_vma(tlb, vma) do { } while (0)
-
 /* tlb_gather_mmu
  *	Return a pointer to an initialized mmu_gather_t.
  */
@@ -57,9 +62,9 @@ static inline void tlb_flush_mmu(mmu_gather_t *tlb, unsigned long start, unsigne
 {
 	unsigned long nr;
 
-	flush_tlb_mm(tlb->mm);
+	tlb_flush(tlb);
 	nr = tlb->nr;
-	if (nr != ~0UL) {
+	if (!tlb_fast_mode(tlb)) {
 		unsigned long i;
 		tlb->nr = 0;
 		for (i=0; i < nr; i++)
@@ -91,8 +96,7 @@ static inline void tlb_finish_mmu(mmu_gather_t *tlb, unsigned long start, unsign
  */
 static inline void tlb_remove_page(mmu_gather_t *tlb, struct page *page)
 {
-	/* Handle the common case fast, first. */\
-	if (tlb->nr == ~0UL) {
+	if (tlb_fast_mode(tlb)) {
 		free_page_and_swap_cache(page);
 		return;
 	}
diff --git a/include/asm-i386/tlb.h b/include/asm-i386/tlb.h
index 69c0faa9319458..844c3d4c9aaaff 100644
--- a/include/asm-i386/tlb.h
+++ b/include/asm-i386/tlb.h
@@ -1 +1,20 @@
+#ifndef _I386_TLB_H
+#define _I386_TLB_H
+
+/*
+ * x86 doesn't need any special per-pte or
+ * per-vma handling..
+ */
+#define tlb_start_vma(tlb, vma) do { } while (0)
+#define tlb_end_vma(tlb, vma) do { } while (0)
+#define tlb_remove_tlb_entry(tlb, pte, address) do { } while (0)
+
+/*
+ * .. because we flush the whole mm when it
+ * fills up.
+ */
+#define tlb_flush(tlb) flush_tlb_mm((tlb)->mm)
+
 #include <asm-generic/tlb.h>
+
+#endif
diff --git a/mm/memory.c b/mm/memory.c
index 8de16cbed3d59d..c43c303c4b72b6 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -348,11 +348,13 @@ static void zap_pte_range(mmu_gather_t *tlb, pmd_t * pmd, unsigned long address,
 
 			pte_clear(ptep);
 			pfn = pte_pfn(pte);
+			tlb_remove_tlb_entry(tlb, pte, address+offset);
 			if (pfn_valid(pfn)) {
 				struct page *page = pfn_to_page(pfn);
 				if (!PageReserved(page)) {
 					if (pte_dirty(pte))
 						set_page_dirty(page);
+					tlb->freed++;
 					tlb_remove_page(tlb, page);
 				}
 			}