aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorJan Kara <jack@suse.cz>2009-06-17 19:26:23 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2009-06-18 16:03:45 -0400
commite8ef7aaea79a899be4d7f50e829900c0ce15e52f (patch)
tree1cc2370f081264e506d85e954110d363961e37bd /fs
parentce05b2a9db1d86635a906f14427deff97eeb6183 (diff)
ext3: fix chain verification in ext3_get_blocks()
Chain verification in ext3_get_blocks() has been hosed since it called verify_chain(chain, NULL) which always returns success. As a result readers could in theory race with truncate. On the other hand the race probably cannot happen with the current locking scheme, since by the time ext3_truncate() is called all the pages are already removed and hence get_block() shouldn't be called on such pages... Signed-off-by: Jan Kara <jack@suse.cz> Cc: <linux-ext4@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'fs')
-rw-r--r--fs/ext3/inode.c2
1 files changed, 1 insertions, 1 deletions
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index b0248c6d5d4c..253c2cdc8d04 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -820,7 +820,7 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
820 while (count < maxblocks && count <= blocks_to_boundary) { 820 while (count < maxblocks && count <= blocks_to_boundary) {
821 ext3_fsblk_t blk; 821 ext3_fsblk_t blk;
822 822
823 if (!verify_chain(chain, partial)) { 823 if (!verify_chain(chain, chain + depth - 1)) {
824 /* 824 /*
825 * Indirect block might be removed by 825 * Indirect block might be removed by
826 * truncate while we were reading it. 826 * truncate while we were reading it.