nagai/kernel
3
0
Fork 0
forked from mirrors/linux
Code Pull requests Activity

filemap: prepare find and delete operations for huge pages

Browse source 
For now, we would have HPAGE_PMD_NR entries in radix tree for every huge
page.  That's suboptimal and it will be changed to use Matthew's
multi-order entries later.

'add' operation is not changed, because we don't need it to implement
hugetmpfs: shmem uses its own implementation.

Link: http://lkml.kernel.org/r/1466021202-61880-25-git-send-email-kirill.shutemov@linux.intel.com
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This commit is contained in:
Kirill A. Shutemov 2016-07-26 15:26:04 -07:00 committed by Linus Torvalds
parent c78c66d1dd
commit 83929372f6
1 changed files with 122 additions and 56 deletions
repo.diff.show_diff_stats Download patch file Download diff file Expand all files Collapse all files

150
mm/filemap.c
View file

@ -114,14 +114,14 @@ static void page_cache_tree_delete(struct address_space *mapping,
struct page *page, void *shadow) struct page *page, void *shadow)
{ {
struct radix_tree_node *node; struct radix_tree_node *node;
int i, nr = PageHuge(page) ? 1 : hpage_nr_pages(page);
VM_BUG_ON(!PageLocked(page)); VM_BUG_ON_PAGE(!PageLocked(page), page);
VM_BUG_ON_PAGE(PageTail(page), page);
node = radix_tree_replace_clear_tags(&mapping->page_tree, page->index, VM_BUG_ON_PAGE(nr != 1 && shadow, page);
shadow);
if (shadow) { if (shadow) {
mapping->nrexceptional++; mapping->nrexceptional += nr;
/* /*
* Make sure the nrexceptional update is committed before * Make sure the nrexceptional update is committed before
* the nrpages update so that final truncate racing * the nrpages update so that final truncate racing
@ -130,31 +130,38 @@ static void page_cache_tree_delete(struct address_space *mapping,
*/ */
smp_wmb(); smp_wmb();
} }
mapping->nrpages--; mapping->nrpages -= nr;
if (!node) for (i = 0; i < nr; i++) {
node = radix_tree_replace_clear_tags(&mapping->page_tree,
page->index + i, shadow);
if (!node) {
VM_BUG_ON_PAGE(nr != 1, page);
return; return;
}
workingset_node_pages_dec(node); workingset_node_pages_dec(node);
if (shadow) if (shadow)
workingset_node_shadows_inc(node); workingset_node_shadows_inc(node);
else else
if (__radix_tree_delete_node(&mapping->page_tree, node)) if (__radix_tree_delete_node(&mapping->page_tree, node))
return; continue;
/* /*
* Track node that only contains shadow entries. DAX mappings contain * Track node that only contains shadow entries. DAX mappings
* no shadow entries and may contain other exceptional entries so skip * contain no shadow entries and may contain other exceptional
* those. * entries so skip those.
* *
* Avoid acquiring the list_lru lock if already tracked. The * Avoid acquiring the list_lru lock if already tracked.
* list_empty() test is safe as node->private_list is * The list_empty() test is safe as node->private_list is
* protected by mapping->tree_lock. * protected by mapping->tree_lock.
*/ */
if (!dax_mapping(mapping) && !workingset_node_pages(node) && if (!dax_mapping(mapping) && !workingset_node_pages(node) &&
list_empty(&node->private_list)) { list_empty(&node->private_list)) {
node->private_data = mapping; node->private_data = mapping;
list_lru_add(&workingset_shadow_nodes, &node->private_list); list_lru_add(&workingset_shadow_nodes,
&node->private_list);
}
} }
} }
@ -166,6 +173,7 @@ static void page_cache_tree_delete(struct address_space *mapping,
void __delete_from_page_cache(struct page *page, void *shadow) void __delete_from_page_cache(struct page *page, void *shadow)
{ {
struct address_space *mapping = page->mapping; struct address_space *mapping = page->mapping;
int nr = hpage_nr_pages(page);
trace_mm_filemap_delete_from_page_cache(page); trace_mm_filemap_delete_from_page_cache(page);
/* /*
@ -178,6 +186,7 @@ void __delete_from_page_cache(struct page *page, void *shadow)
else else
cleancache_invalidate_page(mapping, page); cleancache_invalidate_page(mapping, page);
VM_BUG_ON_PAGE(PageTail(page), page);
VM_BUG_ON_PAGE(page_mapped(page), page); VM_BUG_ON_PAGE(page_mapped(page), page);
if (!IS_ENABLED(CONFIG_DEBUG_VM) && unlikely(page_mapped(page))) { if (!IS_ENABLED(CONFIG_DEBUG_VM) && unlikely(page_mapped(page))) {
int mapcount; int mapcount;
@ -209,9 +218,9 @@ void __delete_from_page_cache(struct page *page, void *shadow)
/* hugetlb pages do not participate in page cache accounting. */ /* hugetlb pages do not participate in page cache accounting. */
if (!PageHuge(page)) if (!PageHuge(page))
__dec_zone_page_state(page, NR_FILE_PAGES); __mod_zone_page_state(page_zone(page), NR_FILE_PAGES, -nr);
if (PageSwapBacked(page)) if (PageSwapBacked(page))
__dec_zone_page_state(page, NR_SHMEM); __mod_zone_page_state(page_zone(page), NR_SHMEM, -nr);
/* /*
* At this point page must be either written or cleaned by truncate. * At this point page must be either written or cleaned by truncate.
@ -235,9 +244,8 @@ void __delete_from_page_cache(struct page *page, void *shadow)
*/ */
void delete_from_page_cache(struct page *page) void delete_from_page_cache(struct page *page)
{ {
struct address_space *mapping = page->mapping; struct address_space *mapping = page_mapping(page);
unsigned long flags; unsigned long flags;
void (*freepage)(struct page *); void (*freepage)(struct page *);
BUG_ON(!PageLocked(page)); BUG_ON(!PageLocked(page));
@ -250,8 +258,14 @@ void delete_from_page_cache(struct page *page)
if (freepage) if (freepage)
freepage(page); freepage(page);
if (PageTransHuge(page) && !PageHuge(page)) {
page_ref_sub(page, HPAGE_PMD_NR);
VM_BUG_ON_PAGE(page_count(page) <= 0, page);
} else {
put_page(page); put_page(page);
} }
}
EXPORT_SYMBOL(delete_from_page_cache); EXPORT_SYMBOL(delete_from_page_cache);
static int filemap_check_errors(struct address_space *mapping) static int filemap_check_errors(struct address_space *mapping)
@ -1053,7 +1067,7 @@ EXPORT_SYMBOL(page_cache_prev_hole);
struct page *find_get_entry(struct address_space *mapping, pgoff_t offset) struct page *find_get_entry(struct address_space *mapping, pgoff_t offset)
{ {
void **pagep; void **pagep;
struct page *page; struct page *head, *page;
rcu_read_lock(); rcu_read_lock();
repeat: repeat:
@ -1073,16 +1087,24 @@ struct page *find_get_entry(struct address_space *mapping, pgoff_t offset)
*/ */
goto out; goto out;
} }
if (!page_cache_get_speculative(page))
head = compound_head(page);
if (!page_cache_get_speculative(head))
goto repeat; goto repeat;
/* The page was split under us? */
if (compound_head(page) != head) {
put_page(head);
goto repeat;
}
/* /*
* Has the page moved? * Has the page moved?
* This is part of the lockless pagecache protocol. See * This is part of the lockless pagecache protocol. See
* include/linux/pagemap.h for details. * include/linux/pagemap.h for details.
*/ */
if (unlikely(page != *pagep)) { if (unlikely(page != *pagep)) {
put_page(page); put_page(head);
goto repeat; goto repeat;
} }
} }
@ -1118,12 +1140,12 @@ struct page *find_lock_entry(struct address_space *mapping, pgoff_t offset)
if (page && !radix_tree_exception(page)) { if (page && !radix_tree_exception(page)) {
lock_page(page); lock_page(page);
/* Has the page been truncated? */ /* Has the page been truncated? */
if (unlikely(page->mapping != mapping)) { if (unlikely(page_mapping(page) != mapping)) {
unlock_page(page); unlock_page(page);
put_page(page); put_page(page);
goto repeat; goto repeat;
} }
VM_BUG_ON_PAGE(page->index != offset, page); VM_BUG_ON_PAGE(page_to_pgoff(page) != offset, page);
} }
return page; return page;
} }
@ -1255,7 +1277,7 @@ unsigned find_get_entries(struct address_space *mapping,
rcu_read_lock(); rcu_read_lock();
radix_tree_for_each_slot(slot, &mapping->page_tree, &iter, start) { radix_tree_for_each_slot(slot, &mapping->page_tree, &iter, start) {
struct page *page; struct page *head, *page;
repeat: repeat:
page = radix_tree_deref_slot(slot); page = radix_tree_deref_slot(slot);
if (unlikely(!page)) if (unlikely(!page))
@ -1272,12 +1294,20 @@ unsigned find_get_entries(struct address_space *mapping,
*/ */
goto export; goto export;
} }
if (!page_cache_get_speculative(page))
head = compound_head(page);
if (!page_cache_get_speculative(head))
goto repeat; goto repeat;
/* The page was split under us? */
if (compound_head(page) != head) {
put_page(head);
goto repeat;
}
/* Has the page moved? */ /* Has the page moved? */
if (unlikely(page != *slot)) { if (unlikely(page != *slot)) {
put_page(page); put_page(head);
goto repeat; goto repeat;
} }
export: export:
@ -1318,7 +1348,7 @@ unsigned find_get_pages(struct address_space *mapping, pgoff_t start,
rcu_read_lock(); rcu_read_lock();
radix_tree_for_each_slot(slot, &mapping->page_tree, &iter, start) { radix_tree_for_each_slot(slot, &mapping->page_tree, &iter, start) {
struct page *page; struct page *head, *page;
repeat: repeat:
page = radix_tree_deref_slot(slot); page = radix_tree_deref_slot(slot);
if (unlikely(!page)) if (unlikely(!page))
@ -1337,12 +1367,19 @@ unsigned find_get_pages(struct address_space *mapping, pgoff_t start,
continue; continue;
} }
if (!page_cache_get_speculative(page)) head = compound_head(page);
if (!page_cache_get_speculative(head))
goto repeat; goto repeat;
/* The page was split under us? */
if (compound_head(page) != head) {
put_page(head);
goto repeat;
}
/* Has the page moved? */ /* Has the page moved? */
if (unlikely(page != *slot)) { if (unlikely(page != *slot)) {
put_page(page); put_page(head);
goto repeat; goto repeat;
} }
@ -1379,7 +1416,7 @@ unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t index,
rcu_read_lock(); rcu_read_lock();
radix_tree_for_each_contig(slot, &mapping->page_tree, &iter, index) { radix_tree_for_each_contig(slot, &mapping->page_tree, &iter, index) {
struct page *page; struct page *head, *page;
repeat: repeat:
page = radix_tree_deref_slot(slot); page = radix_tree_deref_slot(slot);
/* The hole, there no reason to continue */ /* The hole, there no reason to continue */
@ -1399,12 +1436,19 @@ unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t index,
break; break;
} }
if (!page_cache_get_speculative(page)) head = compound_head(page);
if (!page_cache_get_speculative(head))
goto repeat; goto repeat;
/* The page was split under us? */
if (compound_head(page) != head) {
put_page(head);
goto repeat;
}
/* Has the page moved? */ /* Has the page moved? */
if (unlikely(page != *slot)) { if (unlikely(page != *slot)) {
put_page(page); put_page(head);
goto repeat; goto repeat;
} }
@ -1413,7 +1457,7 @@ unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t index,
* otherwise we can get both false positives and false * otherwise we can get both false positives and false
* negatives, which is just confusing to the caller. * negatives, which is just confusing to the caller.
*/ */
if (page->mapping == NULL || page->index != iter.index) { if (page->mapping == NULL || page_to_pgoff(page) != iter.index) {
put_page(page); put_page(page);
break; break;
} }
@ -1451,7 +1495,7 @@ unsigned find_get_pages_tag(struct address_space *mapping, pgoff_t *index,
rcu_read_lock(); rcu_read_lock();
radix_tree_for_each_tagged(slot, &mapping->page_tree, radix_tree_for_each_tagged(slot, &mapping->page_tree,
&iter, *index, tag) { &iter, *index, tag) {
struct page *page; struct page *head, *page;
repeat: repeat:
page = radix_tree_deref_slot(slot); page = radix_tree_deref_slot(slot);
if (unlikely(!page)) if (unlikely(!page))
@ -1476,12 +1520,19 @@ unsigned find_get_pages_tag(struct address_space *mapping, pgoff_t *index,
continue; continue;
} }
if (!page_cache_get_speculative(page)) head = compound_head(page);
if (!page_cache_get_speculative(head))
goto repeat; goto repeat;
/* The page was split under us? */
if (compound_head(page) != head) {
put_page(head);
goto repeat;
}
/* Has the page moved? */ /* Has the page moved? */
if (unlikely(page != *slot)) { if (unlikely(page != *slot)) {
put_page(page); put_page(head);
goto repeat; goto repeat;
} }
@ -1525,7 +1576,7 @@ unsigned find_get_entries_tag(struct address_space *mapping, pgoff_t start,
rcu_read_lock(); rcu_read_lock();
radix_tree_for_each_tagged(slot, &mapping->page_tree, radix_tree_for_each_tagged(slot, &mapping->page_tree,
&iter, start, tag) { &iter, start, tag) {
struct page *page; struct page *head, *page;
repeat: repeat:
page = radix_tree_deref_slot(slot); page = radix_tree_deref_slot(slot);
if (unlikely(!page)) if (unlikely(!page))
@ -1543,12 +1594,20 @@ unsigned find_get_entries_tag(struct address_space *mapping, pgoff_t start,
*/ */
goto export; goto export;
} }
if (!page_cache_get_speculative(page))
head = compound_head(page);
if (!page_cache_get_speculative(head))
goto repeat; goto repeat;
/* The page was split under us? */
if (compound_head(page) != head) {
put_page(head);
goto repeat;
}
/* Has the page moved? */ /* Has the page moved? */
if (unlikely(page != *slot)) { if (unlikely(page != *slot)) {
put_page(page); put_page(head);
goto repeat; goto repeat;
} }
export: export:
@ -2137,7 +2196,7 @@ void filemap_map_pages(struct fault_env *fe,
struct address_space *mapping = file->f_mapping; struct address_space *mapping = file->f_mapping;
pgoff_t last_pgoff = start_pgoff; pgoff_t last_pgoff = start_pgoff;
loff_t size; loff_t size;
struct page *page; struct page *head, *page;
rcu_read_lock(); rcu_read_lock();
radix_tree_for_each_slot(slot, &mapping->page_tree, &iter, radix_tree_for_each_slot(slot, &mapping->page_tree, &iter,
@ -2156,12 +2215,19 @@ void filemap_map_pages(struct fault_env *fe,
goto next; goto next;
} }
if (!page_cache_get_speculative(page)) head = compound_head(page);
if (!page_cache_get_speculative(head))
goto repeat; goto repeat;
/* The page was split under us? */
if (compound_head(page) != head) {
put_page(head);
goto repeat;
}
/* Has the page moved? */ /* Has the page moved? */
if (unlikely(page != *slot)) { if (unlikely(page != *slot)) {
put_page(page); put_page(head);
goto repeat; goto repeat;
} }