[PATCH] msync: check pte dirty earlier
authorAbhijit Karmarkar <abhijitk@veritas.com>
Wed, 22 Jun 2005 00:15:13 +0000 (17:15 -0700)
committerLinus Torvalds <torvalds@ppc970.osdl.org>
Wed, 22 Jun 2005 01:46:21 +0000 (18:46 -0700)
It's common practice to msync a large address range regularly, in which
often only a few ptes have actually been dirtied since the previous pass.

sync_pte_range then goes much faster if it tests whether pte is dirty
before locating and accessing each struct page cacheline; and it is hardly
slowed by ptep_clear_flush_dirty repeating that test in the opposite case,
when every pte actually is dirty.

But beware, s390's pte_dirty always says false, since its dirty bit is kept
in the storage key, located via the struct page address.  So skip this
optimization in its case: use a pte_maybe_dirty macro which just says true
if page_test_and_clear_dirty is implemented.

Signed-off-by: Abhijit Karmarkar <abhijitk@veritas.com>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
include/asm-generic/pgtable.h
mm/msync.c

index 1f4ec7b702706ed643c799e034c128595b3527a7..f40593565173939c5755adda6bee419d90a24635 100644 (file)
@@ -125,6 +125,9 @@ static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addres
 
 #ifndef __HAVE_ARCH_PAGE_TEST_AND_CLEAR_DIRTY
 #define page_test_and_clear_dirty(page) (0)
+#define pte_maybe_dirty(pte)           pte_dirty(pte)
+#else
+#define pte_maybe_dirty(pte)           (1)
 #endif
 
 #ifndef __HAVE_ARCH_PAGE_TEST_AND_CLEAR_YOUNG
index 090f426bca7d538d5f84c9a8c53a677f50a31434..d0f5a1bce7cb14ecd7b74b960e87c8b079024020 100644 (file)
@@ -34,6 +34,8 @@ static void sync_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
 
                if (!pte_present(*pte))
                        continue;
+               if (!pte_maybe_dirty(*pte))
+                       continue;
                pfn = pte_pfn(*pte);
                if (!pfn_valid(pfn))
                        continue;