From: Paolo 'Blaisorblade' Giarrusso <[EMAIL PROTECTED]>
This optimization avoid looking up pages for PROT_NONE mappings. The idea was
taken from the "wrong "historical" code for review - 1" one (the next one)
from mingo, but I fixed it, by adding another "detail" parameter. I've also
fixed the other callers to clear this parameter, and fixed madvise_dontneed()
to use memset(0) on its parameter - currently it's probably a bug.
Not even-compile tested, just written off the top of my head.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <[EMAIL PROTECTED]>
---
linux-2.6.git-paolo/include/linux/mm.h | 1 +
linux-2.6.git-paolo/mm/filemap.c | 18 ++++++++++++++++++
linux-2.6.git-paolo/mm/madvise.c | 10 ++++++----
linux-2.6.git-paolo/mm/memory.c | 11 ++++++++---
linux-2.6.git-paolo/mm/shmem.c | 11 +++++++++++
5 files changed, 44 insertions(+), 7 deletions(-)
diff -puN mm/filemap.c~rfp-avoid-lookup-pages-miss-mapping mm/filemap.c
--- linux-2.6.git/mm/filemap.c~rfp-avoid-lookup-pages-miss-mapping
2005-08-12 18:42:23.000000000 +0200
+++ linux-2.6.git-paolo/mm/filemap.c 2005-08-12 19:14:39.000000000 +0200
@@ -1495,6 +1495,24 @@ int filemap_populate(struct vm_area_stru
struct page *page;
int err;
+ /*
+ * mapping-removal fastpath:
+ */
+ if ((vma->vm_flags & VM_SHARED) &&
+ (pgprot_val(prot) == pgprot_val(PAGE_NONE))) {
+ struct zap_details details;
+
+ /* Still do error-checking! */
+ size = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
PAGE_CACHE_SHIFT;
+ if (pgoff + (len >> PAGE_CACHE_SHIFT) > size)
+ return -EINVAL;
+
+ memset(&details, 0, sizeof(details));
+ details.prot_none_ptes = 1;
+ zap_page_range(vma, addr, len, &details);
+ return 0;
+ }
+
if (!nonblock)
force_page_cache_readahead(mapping, vma->vm_file,
pgoff, len >> PAGE_CACHE_SHIFT);
diff -puN mm/shmem.c~rfp-avoid-lookup-pages-miss-mapping mm/shmem.c
--- linux-2.6.git/mm/shmem.c~rfp-avoid-lookup-pages-miss-mapping
2005-08-12 18:42:23.000000000 +0200
+++ linux-2.6.git-paolo/mm/shmem.c 2005-08-12 19:11:52.000000000 +0200
@@ -1186,6 +1186,17 @@ static int shmem_populate(struct vm_area
if (pgoff >= size || pgoff + (len >> PAGE_SHIFT) > size)
return -EINVAL;
+ /*
+ * mapping-removal fastpath:
+ */
+ if ((vma->vm_flags & VM_SHARED) &&
+ (pgprot_val(prot) == pgprot_val(PAGE_NONE))) {
+ memset(&details, 0, sizeof(details));
+ details.prot_none_ptes = 1;
+ zap_page_range(vma, addr, len, &details);
+ return 0;
+ }
+
while ((long) len > 0) {
struct page *page = NULL;
int err;
diff -puN mm/memory.c~rfp-avoid-lookup-pages-miss-mapping mm/memory.c
--- linux-2.6.git/mm/memory.c~rfp-avoid-lookup-pages-miss-mapping
2005-08-12 18:44:29.000000000 +0200
+++ linux-2.6.git-paolo/mm/memory.c 2005-08-12 19:09:50.000000000 +0200
@@ -575,11 +575,14 @@ static void zap_pte_range(struct mmu_gat
* If details->check_mapping, we leave swap entries;
* if details->nonlinear_vma, we leave file entries.
*/
- if (unlikely(details))
+ if (unlikely(details) && !details->prot_none_ptes)
continue;
if (!pte_file(ptent))
free_swap_and_cache(pte_to_swp_entry(ptent));
- pte_clear(tlb->mm, addr, pte);
+ if (unlikely(details->prot_none_ptes))
+ set_pte_at(mm, addr, pte, pfn_pte(0, __S000));
+ else
+ pte_clear(tlb->mm, addr, pte);
} while (pte++, addr += PAGE_SIZE, addr != end);
pte_unmap(pte - 1);
}
@@ -623,7 +626,8 @@ static void unmap_page_range(struct mmu_
pgd_t *pgd;
unsigned long next;
- if (details && !details->check_mapping && !details->nonlinear_vma)
+ if (details && !details->check_mapping && !details->nonlinear_vma &&
+ !details->prot_none_ptes)
details = NULL;
BUG_ON(addr >= end);
@@ -1499,6 +1503,7 @@ void unmap_mapping_range(struct address_
if (details.last_index < details.first_index)
details.last_index = ULONG_MAX;
details.i_mmap_lock = &mapping->i_mmap_lock;
+ details.prot_none_ptes = 0;
spin_lock(&mapping->i_mmap_lock);
diff -puN include/linux/mm.h~rfp-avoid-lookup-pages-miss-mapping
include/linux/mm.h
--- linux-2.6.git/include/linux/mm.h~rfp-avoid-lookup-pages-miss-mapping
2005-08-12 18:44:36.000000000 +0200
+++ linux-2.6.git-paolo/include/linux/mm.h 2005-08-12 19:01:52.000000000
+0200
@@ -680,6 +680,7 @@ struct zap_details {
pgoff_t last_index; /* Highest page->index to unmap
*/
spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
unsigned long truncate_count; /* Compare vm_truncate_count */
+ unsigned prot_none_ptes : 1; /* If 1, set all PTE's to
PROT_NONE ones.*/
};
unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
diff -puN mm/madvise.c~rfp-avoid-lookup-pages-miss-mapping mm/madvise.c
--- linux-2.6.git/mm/madvise.c~rfp-avoid-lookup-pages-miss-mapping
2005-08-12 18:45:58.000000000 +0200
+++ linux-2.6.git-paolo/mm/madvise.c 2005-08-12 19:20:45.000000000 +0200
@@ -127,10 +127,12 @@ static long madvise_dontneed(struct vm_a
return -EINVAL;
if (unlikely(vma->vm_flags & VM_NONLINEAR)) {
- struct zap_details details = {
- .nonlinear_vma = vma,
- .last_index = ULONG_MAX,
- };
+ struct zap_details details;
+ memset(&details, 0, sizeof(details));
+
+ details.nonlinear_vma = vma;
+ details.last_index = ULONG_MAX;
+
zap_page_range(vma, start, end - start, &details);
} else
zap_page_range(vma, start, end - start, NULL);
_
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
