From 5417169026c3df151adf5a65eb061278b0a72e69 Mon Sep 17 00:00:00 2001 From: David Chinner Date: Thu, 19 Jul 2007 17:39:55 +1000 Subject: [FS] Implement block_page_mkwrite. Many filesystems need a ->page-mkwrite callout to correctly set up pages that have been written to by mmap. This is especially important when mmap is writing into holes as it allows filesystems to correctly account for and allocate space before the mmap write is allowed to proceed. Protection against truncate races is provided by locking the page and checking to see whether the page mapping is correct and whether it is beyond EOF so we don't end up allowing allocations beyond the current EOF or changing EOF as a result of a mmap write. SGI-PV: 940392 SGI-Modid: 2.6.x-xfs-melb:linux:29146a Signed-off-by: David Chinner Signed-off-by: Christoph Hellwig Signed-off-by: Tim Shimmin --- fs/buffer.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 47 insertions(+) (limited to 'fs/buffer.c') diff --git a/fs/buffer.c b/fs/buffer.c index 0f9006714230..02ebb1f1d3b0 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -2194,6 +2194,52 @@ int generic_commit_write(struct file *file, struct page *page, return 0; } +/* + * block_page_mkwrite() is not allowed to change the file size as it gets + * called from a page fault handler when a page is first dirtied. Hence we must + * be careful to check for EOF conditions here. We set the page up correctly + * for a written page which means we get ENOSPC checking when writing into + * holes and correct delalloc and unwritten extent mapping on filesystems that + * support these features. + * + * We are not allowed to take the i_mutex here so we have to play games to + * protect against truncate races as the page could now be beyond EOF. Because + * vmtruncate() writes the inode size before removing pages, once we have the + * page lock we can determine safely if the page is beyond EOF. If it is not + * beyond EOF, then the page is guaranteed safe against truncation until we + * unlock the page. + */ +int +block_page_mkwrite(struct vm_area_struct *vma, struct page *page, + get_block_t get_block) +{ + struct inode *inode = vma->vm_file->f_path.dentry->d_inode; + unsigned long end; + loff_t size; + int ret = -EINVAL; + + lock_page(page); + size = i_size_read(inode); + if ((page->mapping != inode->i_mapping) || + ((page->index << PAGE_CACHE_SHIFT) > size)) { + /* page got truncated out from underneath us */ + goto out_unlock; + } + + /* page is wholly or partially inside EOF */ + if (((page->index + 1) << PAGE_CACHE_SHIFT) > size) + end = size & ~PAGE_CACHE_MASK; + else + end = PAGE_CACHE_SIZE; + + ret = block_prepare_write(page, 0, end, get_block); + if (!ret) + ret = block_commit_write(page, 0, end); + +out_unlock: + unlock_page(page); + return ret; +} /* * nobh_prepare_write()'s prereads are special: the buffer_heads are freed @@ -2977,6 +3023,7 @@ EXPORT_SYMBOL(__brelse); EXPORT_SYMBOL(__wait_on_buffer); EXPORT_SYMBOL(block_commit_write); EXPORT_SYMBOL(block_prepare_write); +EXPORT_SYMBOL(block_page_mkwrite); EXPORT_SYMBOL(block_read_full_page); EXPORT_SYMBOL(block_sync_page); EXPORT_SYMBOL(block_truncate_page); -- cgit v1.2.3 From 1833633803c7ef4d8f09877d3f1549cbd252f477 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Fri, 20 Jul 2007 00:31:45 -0700 Subject: fix some conversion overflows Fix page index to offset conversion overflows in buffer layer, ecryptfs, and ocfs2. It would be nice to convert the whole tree to page_offset, but for now just fix the bugs. Signed-off-by: Nick Piggin Cc: Michael Halcrow Cc: Mark Fasheh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/buffer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/buffer.c') diff --git a/fs/buffer.c b/fs/buffer.c index 02ebb1f1d3b0..0e5ec371ce72 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -2221,7 +2221,7 @@ block_page_mkwrite(struct vm_area_struct *vma, struct page *page, lock_page(page); size = i_size_read(inode); if ((page->mapping != inode->i_mapping) || - ((page->index << PAGE_CACHE_SHIFT) > size)) { + (page_offset(page) > size)) { /* page got truncated out from underneath us */ goto out_unlock; } -- cgit v1.2.3 From 6712ecf8f648118c3363c142196418f89a510b90 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 27 Sep 2007 12:47:43 +0200 Subject: Drop 'size' argument from bio_endio and bi_end_io As bi_end_io is only called once when the reqeust is complete, the 'size' argument is now redundant. Remove it. Now there is no need for bio_endio to subtract the size completed from bi_size. So don't do that either. While we are at it, change bi_end_io to return void. Signed-off-by: Neil Brown Signed-off-by: Jens Axboe --- fs/buffer.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'fs/buffer.c') diff --git a/fs/buffer.c b/fs/buffer.c index 0e5ec371ce72..75b51dfa5e03 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -2634,13 +2634,10 @@ sector_t generic_block_bmap(struct address_space *mapping, sector_t block, return tmp.b_blocknr; } -static int end_bio_bh_io_sync(struct bio *bio, unsigned int bytes_done, int err) +static void end_bio_bh_io_sync(struct bio *bio, int err) { struct buffer_head *bh = bio->bi_private; - if (bio->bi_size) - return 1; - if (err == -EOPNOTSUPP) { set_bit(BIO_EOPNOTSUPP, &bio->bi_flags); set_bit(BH_Eopnotsupp, &bh->b_state); @@ -2648,7 +2645,6 @@ static int end_bio_bh_io_sync(struct bio *bio, unsigned int bytes_done, int err) bh->b_end_io(bh, test_bit(BIO_UPTODATE, &bio->bi_flags)); bio_put(bio); - return 0; } int submit_bh(int rw, struct buffer_head * bh) -- cgit v1.2.3