On Thu, Aug 15, 2019 at 09:50:43AM -0700, Darrick J. Wong wrote: > From: Darrick J. Wong <darrick.wong@xxxxxxxxxx> > > While trawling through the dedupe file comparison code trying to fix > page deadlocking problems, Dave Chinner noticed that the reflink code > only takes shared IOLOCK/MMAPLOCKs on the source file. Because > page_mkwrite and directio writes do not take the EXCL versions of those > locks, this means that reflink can race with writer processes. > > For pure remapping this can lead to undefined behavior and file > corruption; for dedupe this means that we cannot be sure that the > contents are identical when we decide to go ahead with the remapping. > > Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx> I don't think this is quite right yet. The source inode locking change is good, but it doesn't break the layout on the source inode and so there is still they possibility that something has physical access and is directly modifying the source file. And with that, I suspect the locking algorithm changes substantially: order inodes restart: lock first inode break layout on first inode lock second inode break layout on second inode fail then unlock both inodes, goto restart Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx
