On Wed, Mar 25, 2020 at 05:27:38AM +0000, Damien Le Moal wrote:

quoted

At least for a normal file system that is absolutely not true.  If
zonefs is so special it might be better of just using a slightly tweaked
copy of blkdev_direct_IO rather than using iomap.

It would be very nice to not have to add this direct BIO use case in zonefs
since that would be only for writes to sequential zones while all other
operations use iomap. So instead of this, what about using a flag as Dave
suggested (see below comment too) ?

Given how special the use case is I'm not sure overloading iomap
is a good idea.  Think of how a "normal" zone aware file system would
use iomap and not of this will apply.  OTOH the "simple" single bio
code in __blkdev_direct_IO_simple is less than 100 lines of code.  I
think having a specialized code base for a specialized use case
might be better than overloading generic code with tons of flags.

quoted

I don't think the iocb is the right interface for passing this
kind of information.  We currently pass a bool wait to iomap_dio_rw
which really should be flags.  I have a pending patch for that.

Is that patch queued in iomap or xfs tree ? Could you point us to it please ?

It isn't queued up anywhere yet.