diff options
author | Andy Whitcroft <apw@shadowen.org> | 2008-11-06 15:53:26 -0500 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2008-11-06 18:41:18 -0500 |
commit | 69d177c2fc702d402b17fdca2190d5a7e3ca55c5 (patch) | |
tree | 2040e0a84b7c07c29ac6fb6e51e125de52256f5d /mm/hugetlb.c | |
parent | 22bece00dc1f28dd3374c55e464c9f02eb642876 (diff) |
hugetlbfs: handle pages higher order than MAX_ORDER
When working with hugepages, hugetlbfs assumes that those hugepages are
smaller than MAX_ORDER. Specifically it assumes that the mem_map is
contigious and uses that to optimise access to the elements of the mem_map
that represent the hugepage. Gigantic pages (such as 16GB pages on
powerpc) by definition are of greater order than MAX_ORDER (larger than
MAX_ORDER_NR_PAGES in size). This means that we can no longer make use of
the buddy alloctor guarentees for the contiguity of the mem_map, which
ensures that the mem_map is at least contigious for maximmally aligned
areas of MAX_ORDER_NR_PAGES pages.
This patch adds new mem_map accessors and iterator helpers which handle
any discontiguity at MAX_ORDER_NR_PAGES boundaries. It then uses these to
implement gigantic page versions of copy_huge_page and clear_huge_page,
and to allow follow_hugetlb_page handle gigantic pages.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Cc: Jon Tollefson <kniht@linux.vnet.ibm.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: <stable@kernel.org> [2.6.27.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'mm/hugetlb.c')
-rw-r--r-- | mm/hugetlb.c | 37 |
1 files changed, 36 insertions, 1 deletions
diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 421aee99b84..e6afe527bd0 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c | |||
@@ -354,11 +354,26 @@ static int vma_has_reserves(struct vm_area_struct *vma) | |||
354 | return 0; | 354 | return 0; |
355 | } | 355 | } |
356 | 356 | ||
357 | static void clear_gigantic_page(struct page *page, | ||
358 | unsigned long addr, unsigned long sz) | ||
359 | { | ||
360 | int i; | ||
361 | struct page *p = page; | ||
362 | |||
363 | might_sleep(); | ||
364 | for (i = 0; i < sz/PAGE_SIZE; i++, p = mem_map_next(p, page, i)) { | ||
365 | cond_resched(); | ||
366 | clear_user_highpage(p, addr + i * PAGE_SIZE); | ||
367 | } | ||
368 | } | ||
357 | static void clear_huge_page(struct page *page, | 369 | static void clear_huge_page(struct page *page, |
358 | unsigned long addr, unsigned long sz) | 370 | unsigned long addr, unsigned long sz) |
359 | { | 371 | { |
360 | int i; | 372 | int i; |
361 | 373 | ||
374 | if (unlikely(sz > MAX_ORDER_NR_PAGES)) | ||
375 | return clear_gigantic_page(page, addr, sz); | ||
376 | |||
362 | might_sleep(); | 377 | might_sleep(); |
363 | for (i = 0; i < sz/PAGE_SIZE; i++) { | 378 | for (i = 0; i < sz/PAGE_SIZE; i++) { |
364 | cond_resched(); | 379 | cond_resched(); |
@@ -366,12 +381,32 @@ static void clear_huge_page(struct page *page, | |||
366 | } | 381 | } |
367 | } | 382 | } |
368 | 383 | ||
384 | static void copy_gigantic_page(struct page *dst, struct page *src, | ||
385 | unsigned long addr, struct vm_area_struct *vma) | ||
386 | { | ||
387 | int i; | ||
388 | struct hstate *h = hstate_vma(vma); | ||
389 | struct page *dst_base = dst; | ||
390 | struct page *src_base = src; | ||
391 | might_sleep(); | ||
392 | for (i = 0; i < pages_per_huge_page(h); ) { | ||
393 | cond_resched(); | ||
394 | copy_user_highpage(dst, src, addr + i*PAGE_SIZE, vma); | ||
395 | |||
396 | i++; | ||
397 | dst = mem_map_next(dst, dst_base, i); | ||
398 | src = mem_map_next(src, src_base, i); | ||
399 | } | ||
400 | } | ||
369 | static void copy_huge_page(struct page *dst, struct page *src, | 401 | static void copy_huge_page(struct page *dst, struct page *src, |
370 | unsigned long addr, struct vm_area_struct *vma) | 402 | unsigned long addr, struct vm_area_struct *vma) |
371 | { | 403 | { |
372 | int i; | 404 | int i; |
373 | struct hstate *h = hstate_vma(vma); | 405 | struct hstate *h = hstate_vma(vma); |
374 | 406 | ||
407 | if (unlikely(pages_per_huge_page(h) > MAX_ORDER_NR_PAGES)) | ||
408 | return copy_gigantic_page(dst, src, addr, vma); | ||
409 | |||
375 | might_sleep(); | 410 | might_sleep(); |
376 | for (i = 0; i < pages_per_huge_page(h); i++) { | 411 | for (i = 0; i < pages_per_huge_page(h); i++) { |
377 | cond_resched(); | 412 | cond_resched(); |
@@ -2130,7 +2165,7 @@ same_page: | |||
2130 | if (zeropage_ok) | 2165 | if (zeropage_ok) |
2131 | pages[i] = ZERO_PAGE(0); | 2166 | pages[i] = ZERO_PAGE(0); |
2132 | else | 2167 | else |
2133 | pages[i] = page + pfn_offset; | 2168 | pages[i] = mem_map_offset(page, pfn_offset); |
2134 | get_page(pages[i]); | 2169 | get_page(pages[i]); |
2135 | } | 2170 | } |
2136 | 2171 | ||