Hi Matthew, could you do me a favor? For the next (and hopefully final :-) spin of this patch series, could you base it on the ext4.git's master branch. Then pull in Darrick's iomap-for-next branch, and then apply your patches on top of that. I attempted to do this with the v6 patch series --- see the tt/mb-dio branch --- and I described on another e-mail thread, I appear to have screwed up that patch conflicts, since it's causing a failure with diroead-nolock using a 1k block size. Since this wasn't something that worked when you were first working on the patch set, this isn't something I'm going to consider blocking, especially since a flay test failure which happens 7% of the time, and using dioread_nolock with a sub-page blocksize isn't something that is going to be all that common (since it wasn't working at all up until now). Still, I'm hoping that either Ritesh or you can figure out how my simple-minded handling of the patch conflict between your and his patch series can be addressed properly. Thanks!! - Ted
