summaryrefslogtreecommitdiff
path: root/fs/ext2
diff options
context:
space:
mode:
authorJan Kara <jack@suse.cz>2009-04-13 14:40:14 -0700
committerLinus Torvalds <torvalds@linux-foundation.org>2009-04-13 15:04:33 -0700
commit316cb4ef3eb2ad6e35e15cc56d39c6cda58c093a (patch)
tree23084bf762a3dc0ec91c299c5f10206b132df621 /fs/ext2
parentcaefba1740d8016e6dfe8fda84f85bdcb8f8c85d (diff)
ext2: fix data corruption for racing writes
If two writers allocating blocks to file race with each other (e.g. because writepages races with ordinary write or two writepages race with each other), ext2_getblock() can be called on the same inode in parallel. Before we are going to allocate new blocks, we have to recheck the block chain we have obtained so far without holding truncate_mutex. Otherwise we could overwrite the indirect block pointer set by the other writer leading to data loss. The below test program by Ying is able to reproduce the data loss with ext2 on in BRD in a few minutes if the machine is under memory pressure: long kMemSize = 50 << 20; int kPageSize = 4096; int main(int argc, char **argv) { int status; int count = 0; int i; char *fname = "/mnt/test.mmap"; char *mem; unlink(fname); int fd = open(fname, O_CREAT | O_EXCL | O_RDWR, 0600); status = ftruncate(fd, kMemSize); mem = mmap(0, kMemSize, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0); // Fill the memory with 1s. memset(mem, 1, kMemSize); sleep(2); for (i = 0; i < kMemSize; i++) { int byte_good = mem[i] != 0; if (!byte_good && ((i % kPageSize) == 0)) { //printf("%d ", i / kPageSize); count++; } } munmap(mem, kMemSize); close(fd); unlink(fname); if (count > 0) { printf("Running %d bad page\n", count); return 1; } return 0; } Cc: Ying Han <yinghan@google.com> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Jan Kara <jack@suse.cz> Cc: Mingming Cao <cmm@us.ibm.com> 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/ext2')
-rw-r--r--fs/ext2/inode.c44
1 files changed, 33 insertions, 11 deletions
diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c
index b43b95563663..acf678831103 100644
--- a/fs/ext2/inode.c
+++ b/fs/ext2/inode.c
@@ -590,9 +590,8 @@ static int ext2_get_blocks(struct inode *inode,
if (depth == 0)
return (err);
-reread:
- partial = ext2_get_branch(inode, depth, offsets, chain, &err);
+ partial = ext2_get_branch(inode, depth, offsets, chain, &err);
/* Simplest case - block found, no allocation needed */
if (!partial) {
first_block = le32_to_cpu(chain[depth - 1].key);
@@ -602,15 +601,16 @@ reread:
while (count < maxblocks && count <= blocks_to_boundary) {
ext2_fsblk_t blk;
- if (!verify_chain(chain, partial)) {
+ if (!verify_chain(chain, chain + depth - 1)) {
/*
* Indirect block might be removed by
* truncate while we were reading it.
* Handling of that case: forget what we've
* got now, go to reread.
*/
+ err = -EAGAIN;
count = 0;
- goto changed;
+ break;
}
blk = le32_to_cpu(*(chain[depth-1].p + count));
if (blk == first_block + count)
@@ -618,7 +618,8 @@ reread:
else
break;
}
- goto got_it;
+ if (err != -EAGAIN)
+ goto got_it;
}
/* Next simple case - plain lookup or failed read of indirect block */
@@ -626,6 +627,33 @@ reread:
goto cleanup;
mutex_lock(&ei->truncate_mutex);
+ /*
+ * If the indirect block is missing while we are reading
+ * the chain(ext3_get_branch() returns -EAGAIN err), or
+ * if the chain has been changed after we grab the semaphore,
+ * (either because another process truncated this branch, or
+ * another get_block allocated this branch) re-grab the chain to see if
+ * the request block has been allocated or not.
+ *
+ * Since we already block the truncate/other get_block
+ * at this point, we will have the current copy of the chain when we
+ * splice the branch into the tree.
+ */
+ if (err == -EAGAIN || !verify_chain(chain, partial)) {
+ while (partial > chain) {
+ brelse(partial->bh);
+ partial--;
+ }
+ partial = ext2_get_branch(inode, depth, offsets, chain, &err);
+ if (!partial) {
+ count++;
+ mutex_unlock(&ei->truncate_mutex);
+ if (err)
+ goto cleanup;
+ clear_buffer_new(bh_result);
+ goto got_it;
+ }
+ }
/*
* Okay, we need to do block allocation. Lazily initialize the block
@@ -683,12 +711,6 @@ cleanup:
partial--;
}
return err;
-changed:
- while (partial > chain) {
- brelse(partial->bh);
- partial--;
- }
- goto reread;
}
int ext2_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create)