On Thu, 2023-05-18 at 13:43 +0000, Chuck Lever III wrote: > > > On May 18, 2023, at 7:47 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote: > > > > If getattr fails, then nfsd can end up scraping the time values directly > > out of the inode for pre and post-op attrs. This may or may not be the > > right thing to do, but for now make it at least use ctime_peek in this > > situation to ensure that the QUERIED flag is masked. > > That code comes from: > > commit 39ca1bf624b6b82cc895b0217889eaaf572a7913 > Author: Amir Goldstein <amir73il@xxxxxxxxx> > AuthorDate: Wed Jan 3 17:14:35 2018 +0200 > Commit: J. Bruce Fields <bfields@xxxxxxxxxx> > CommitDate: Thu Feb 8 13:40:17 2018 -0500 > > nfsd: store stat times in fill_pre_wcc() instead of inode times > > The time values in stat and inode may differ for overlayfs and stat time > values are the correct ones to use. This is also consistent with the fact > that fill_post_wcc() also stores stat time values. > > This means introducing a stat call that could fail, where previously we > were just copying values out of the inode. To be conservative about > changing behavior, we fall back to copying values out of the inode in > the error case. It might be better just to clear fh_pre_saved (though > note the BUG_ON in set_change_info). > > Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx> > Signed-off-by: J. Bruce Fields <bfields@xxxxxxxxxx> > > I was thinking it might have been added to handle odd corner > cases around re-exporting NFS mounts, but that does not seem > to be the case. > > The fh_getattr() can fail for legitimate reasons -- like the > file is in the middle of being deleted or renamed over -- I > would think. This code should really deal with that by not > adding pre-op attrs, since they are optional. > That sounds fine to me. I'll plan to drop this patch from the series and I'll send a separate patch to just remove those branches altogether (which should DTRT). -- Jeff Layton <jlayton@xxxxxxxxxx>
