On Fri, Oct 02, 2020 at 09:29:58AM -0700, Darrick J. Wong wrote: > > Instead of coming up with our own inode unlocking and release schemes, > > can't we just require that the inode is joinged by passing the lock > > flags to xfs_trans_ijoin, and piggy back on xfs_trans_commit unlocking > > it in that case? > > Yes, and let's also xfs_iget(capture_ip->i_ino) to increase the incore > inode's refcount, which would make it so that the caller would still > unlock and rele the reference that they got. Please use ihold(VFS_I(capture_ip)) as that is a lot more efficient. Can you resend the whole 2 series? I'm lost with all the incremental updates for individual patches.
