On Wed 18-10-23 14:45:49, Joseph Qi wrote: > On 10/13/23 8:13 PM, Jan Kara wrote: > > Gao Xiang has reported that on ext4 O_SYNC direct IO does not properly > > sync file size update and thus if we crash at unfortunate moment, the > > file can have smaller size although O_SYNC IO has reported successful > > completion. The problem happens because update of on-disk inode size is > > handled in ext4_dio_write_iter() *after* iomap_dio_rw() (and thus > > dio_complete() in particular) has returned and generic_file_sync() gets > > called by dio_complete(). Fix the problem by handling on-disk inode size > > update directly in our ->end_io completion handler. > > > > References: https://lore.kernel.org/all/02d18236-26ef-09b0-90ad-030c4fe3ee20@xxxxxxxxxxxxxxxxx > > Reported-by: Gao Xiang <hsiangkao@xxxxxxxxxxxxxxxxx> > > CC: stable@xxxxxxxxxxxxxxx > > Fixes: 378f32bab371 ("ext4: introduce direct I/O write using iomap infrastructure") > > Signed-off-by: Jan Kara <jack@xxxxxxx> > > Tested with the reproducer after applying to 6.6-rc5, > Tested-by: Joseph Qi <joseph.qi@xxxxxxxxxxxxxxxxx> Thanks for testing! > BTW, once backported to older kernel like 5.10, it seems that it depends > on the following commit: > 936e114a245b iomap: update ki_pos a little later in iomap_dio_complete > > Otherwise, it will fail the following xfstests cases: > generic/091 generic/094 generic/225 generic/263 generic/311 generic/617 That is kind of curious because that commit should not influence how the ext4 fix behaves. It only influences what is in iocb->ki_pos when we are invalidating pagecache pages... Honza -- Jan Kara <jack@xxxxxxxx> SUSE Labs, CR
