On Sat, Dec 08, 2018 at 10:25:58PM +0800, zhangyi (F) wrote: > Current ext4 will call ext4_ext_convert_to_initialized() to split and > initialize an unwritten extent if someone write something to it. It may > also zeroout the nearby blocks and expand the split extent if the > allocated extent is fully inside i_size or new_size. But it may lead to > inode inconsistency when system crash or the power fails. > > Consider the following case: > - Create an empty file and buffer write from block A to D (with delay > allocate). It will update the i_size to D. > - Zero range from part of block B to D. It will allocate an unwritten > extent from B to D. > - The write back worker write block B and initialize the unwritten > extent from B to D, and then update the i_disksize to B. > - System crash. > - Remount and fsck complain about the extent size exceeds the inode > size. > > This patch add checking i_disksize and chose the small one between > i_size to make sure it's safe to convert extent to initialized. > > --------------------- > > This problem can reproduce by xfstests generic/482 with fsstress seed > 1544025012. Hmm, your explanation is great and the patch makes sense. I haven't been able to reproduce the problem by adding -s 1544025012 to the fsstress arguments. This may be because fsstress being run with two processes (-p 2) and the failure may be timing dependent? How easily can you replicate the problem? - Ted
