diff options
author | Nick Piggin <npiggin@suse.de> | 2007-10-16 01:24:57 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@woody.linux-foundation.org> | 2007-10-16 09:42:54 -0700 |
commit | eb2be189317d031895b5ca534fbf735eb546158b (patch) | |
tree | 8f1eda7af3be7285244a6f1ad77682e90d403c7b /mm/filemap.c | |
parent | 64649a58919e66ec21792dbb6c48cb3da22cbd7f (diff) | |
download | linux-stable-eb2be189317d031895b5ca534fbf735eb546158b.tar.gz linux-stable-eb2be189317d031895b5ca534fbf735eb546158b.tar.bz2 linux-stable-eb2be189317d031895b5ca534fbf735eb546158b.zip |
mm: buffered write cleanup
Quite a bit of code is used in maintaining these "cached pages" that are
probably pretty unlikely to get used. It would require a narrow race where
the page is inserted concurrently while this process is allocating a page
in order to create the spare page. Then a multi-page write into an uncached
part of the file, to make use of it.
Next, the buffered write path (and others) uses its own LRU pagevec when it
should be just using the per-CPU LRU pagevec (which will cut down on both data
and code size cacheline footprint). Also, these private LRU pagevecs are
emptied after just a very short time, in contrast with the per-CPU pagevecs
that are persistent. Net result: 7.3 times fewer lru_lock acquisitions required
to add the pages to pagecache for a bulk write (in 4K chunks).
[this gets rid of some cond_resched() calls in readahead.c and mpage.c due
to clashes in -mm. What put them there, and why? ]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'mm/filemap.c')
-rw-r--r-- | mm/filemap.c | 143 |
1 files changed, 55 insertions, 88 deletions
diff --git a/mm/filemap.c b/mm/filemap.c index 73b98c6a3389..fb4c1c0792e7 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -666,27 +666,22 @@ EXPORT_SYMBOL(find_lock_page); struct page *find_or_create_page(struct address_space *mapping, pgoff_t index, gfp_t gfp_mask) { - struct page *page, *cached_page = NULL; + struct page *page; int err; repeat: page = find_lock_page(mapping, index); if (!page) { - if (!cached_page) { - cached_page = - __page_cache_alloc(gfp_mask); - if (!cached_page) - return NULL; + page = __page_cache_alloc(gfp_mask); + if (!page) + return NULL; + err = add_to_page_cache_lru(page, mapping, index, gfp_mask); + if (unlikely(err)) { + page_cache_release(page); + page = NULL; + if (err == -EEXIST) + goto repeat; } - err = add_to_page_cache_lru(cached_page, mapping, - index, gfp_mask); - if (!err) { - page = cached_page; - cached_page = NULL; - } else if (err == -EEXIST) - goto repeat; } - if (cached_page) - page_cache_release(cached_page); return page; } EXPORT_SYMBOL(find_or_create_page); @@ -872,10 +867,8 @@ void do_generic_mapping_read(struct address_space *mapping, pgoff_t prev_index; unsigned long offset; /* offset into pagecache page */ unsigned int prev_offset; - struct page *cached_page; int error; - cached_page = NULL; index = *ppos >> PAGE_CACHE_SHIFT; prev_index = ra->prev_pos >> PAGE_CACHE_SHIFT; prev_offset = ra->prev_pos & (PAGE_CACHE_SIZE-1); @@ -1031,23 +1024,20 @@ no_cached_page: * Ok, it wasn't cached, so we need to create a new * page.. */ - if (!cached_page) { - cached_page = page_cache_alloc_cold(mapping); - if (!cached_page) { - desc->error = -ENOMEM; - goto out; - } + page = page_cache_alloc_cold(mapping); + if (!page) { + desc->error = -ENOMEM; + goto out; } - error = add_to_page_cache_lru(cached_page, mapping, + error = add_to_page_cache_lru(page, mapping, index, GFP_KERNEL); if (error) { + page_cache_release(page); if (error == -EEXIST) goto find_page; desc->error = error; goto out; } - page = cached_page; - cached_page = NULL; goto readpage; } @@ -1057,8 +1047,6 @@ out: ra->prev_pos |= prev_offset; *ppos = ((loff_t)index << PAGE_CACHE_SHIFT) + offset; - if (cached_page) - page_cache_release(cached_page); if (filp) file_accessed(filp); } @@ -1502,35 +1490,28 @@ static struct page *__read_cache_page(struct address_space *mapping, int (*filler)(void *,struct page*), void *data) { - struct page *page, *cached_page = NULL; + struct page *page; int err; repeat: page = find_get_page(mapping, index); if (!page) { - if (!cached_page) { - cached_page = page_cache_alloc_cold(mapping); - if (!cached_page) - return ERR_PTR(-ENOMEM); - } - err = add_to_page_cache_lru(cached_page, mapping, - index, GFP_KERNEL); - if (err == -EEXIST) - goto repeat; - if (err < 0) { + page = page_cache_alloc_cold(mapping); + if (!page) + return ERR_PTR(-ENOMEM); + err = add_to_page_cache_lru(page, mapping, index, GFP_KERNEL); + if (unlikely(err)) { + page_cache_release(page); + if (err == -EEXIST) + goto repeat; /* Presumably ENOMEM for radix tree node */ - page_cache_release(cached_page); return ERR_PTR(err); } - page = cached_page; - cached_page = NULL; err = filler(data, page); if (err < 0) { page_cache_release(page); page = ERR_PTR(err); } } - if (cached_page) - page_cache_release(cached_page); return page; } @@ -1607,40 +1588,6 @@ struct page *read_cache_page(struct address_space *mapping, EXPORT_SYMBOL(read_cache_page); /* - * If the page was newly created, increment its refcount and add it to the - * caller's lru-buffering pagevec. This function is specifically for - * generic_file_write(). - */ -static inline struct page * -__grab_cache_page(struct address_space *mapping, unsigned long index, - struct page **cached_page, struct pagevec *lru_pvec) -{ - int err; - struct page *page; -repeat: - page = find_lock_page(mapping, index); - if (!page) { - if (!*cached_page) { - *cached_page = page_cache_alloc(mapping); - if (!*cached_page) - return NULL; - } - err = add_to_page_cache(*cached_page, mapping, - index, GFP_KERNEL); - if (err == -EEXIST) - goto repeat; - if (err == 0) { - page = *cached_page; - page_cache_get(page); - if (!pagevec_add(lru_pvec, page)) - __pagevec_lru_add(lru_pvec); - *cached_page = NULL; - } - } - return page; -} - -/* * The logic we want is * * if suid or (sgid and xgrp) @@ -1832,6 +1779,33 @@ generic_file_direct_write(struct kiocb *iocb, const struct iovec *iov, } EXPORT_SYMBOL(generic_file_direct_write); +/* + * Find or create a page at the given pagecache position. Return the locked + * page. This function is specifically for buffered writes. + */ +static struct page *__grab_cache_page(struct address_space *mapping, + pgoff_t index) +{ + int status; + struct page *page; +repeat: + page = find_lock_page(mapping, index); + if (likely(page)) + return page; + + page = page_cache_alloc(mapping); + if (!page) + return NULL; + status = add_to_page_cache_lru(page, mapping, index, GFP_KERNEL); + if (unlikely(status)) { + page_cache_release(page); + if (status == -EEXIST) + goto repeat; + return NULL; + } + return page; +} + ssize_t generic_file_buffered_write(struct kiocb *iocb, const struct iovec *iov, unsigned long nr_segs, loff_t pos, loff_t *ppos, @@ -1842,15 +1816,10 @@ generic_file_buffered_write(struct kiocb *iocb, const struct iovec *iov, const struct address_space_operations *a_ops = mapping->a_ops; struct inode *inode = mapping->host; long status = 0; - struct page *page; - struct page *cached_page = NULL; - struct pagevec lru_pvec; const struct iovec *cur_iov = iov; /* current iovec */ size_t iov_offset = 0; /* offset in the current iovec */ char __user *buf; - pagevec_init(&lru_pvec, 0); - /* * handle partial DIO write. Adjust cur_iov if needed. */ @@ -1862,6 +1831,7 @@ generic_file_buffered_write(struct kiocb *iocb, const struct iovec *iov, } do { + struct page *page; pgoff_t index; /* Pagecache index for current page */ unsigned long offset; /* Offset into pagecache page */ unsigned long maxlen; /* Bytes remaining in current iovec */ @@ -1888,7 +1858,8 @@ generic_file_buffered_write(struct kiocb *iocb, const struct iovec *iov, fault_in_pages_readable(buf, maxlen); #endif - page = __grab_cache_page(mapping,index,&cached_page,&lru_pvec); + + page = __grab_cache_page(mapping, index); if (!page) { status = -ENOMEM; break; @@ -1956,9 +1927,6 @@ fs_write_aop_error: } while (count); *ppos = pos; - if (cached_page) - page_cache_release(cached_page); - /* * For now, when the user asks for O_SYNC, we'll actually give O_DSYNC */ @@ -1978,7 +1946,6 @@ fs_write_aop_error: if (unlikely(file->f_flags & O_DIRECT) && written) status = filemap_write_and_wait(mapping); - pagevec_lru_add(&lru_pvec); return written ? written : status; } EXPORT_SYMBOL(generic_file_buffered_write); |