From: Stephen Warren swarren@nvidia.com

ext4fs_allocate_blocks() always allocates at least one block for a file. If the file size is zero, this causes total_remaining_blocks to underflow, which then causes an apparent hang while 2^32 blocks are allocated.

To solve this, check that total_remaining_blocks is non-zero as part of the loop condition (i.e. before each loop) rather than at the end of the loop.

Signed-off-by: Stephen Warren swarren@nvidia.com --- fs/ext4/ext4_common.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/fs/ext4/ext4_common.c b/fs/ext4/ext4_common.c index 1c1172163c09..33d69c9c71f0 100644 --- a/fs/ext4/ext4_common.c +++ b/fs/ext4/ext4_common.c @@ -1380,7 +1380,7 @@ void ext4fs_allocate_blocks(struct ext2_inode *file_inode, unsigned int no_blks_reqd = 0;

/* allocation of direct blocks */ - for (i = 0; i < INDIRECT_BLOCKS; i++) { + for (i = 0; total_remaining_blocks && i < INDIRECT_BLOCKS; i++) { direct_blockno = ext4fs_get_new_blk_no(); if (direct_blockno == -1) { printf("no block left to assign\n"); @@ -1390,8 +1390,6 @@ void ext4fs_allocate_blocks(struct ext2_inode *file_inode, debug("DB %ld: %u\n", direct_blockno, total_remaining_blocks);

total_remaining_blocks--; - if (total_remaining_blocks == 0) - break; }

alloc_single_indirect_block(file_inode, &total_remaining_blocks,