Re: [dm-devel] block: properly handle flush/fua requests in blk_insert_cloned_request

[Jeff Moyer <jmoyer redhat com>]

Vivek Goyal <vgoyal redhat com> writes:

> On Tue, Aug 09, 2011 at 02:55:31PM -0400, Mike Snitzer wrote:
>
> [..]
>> > > +	/*
>> > > +	 * All FLUSH/FUA requests are expected to have gone through the
>> > > +	 * flush machinary.  If a request's cmd_flags doesn't match the
>> > > +	 * flush_flags of the underlying request_queue it is a bug.
>> > > +	 */
>> > > +	BUG_ON((rq->cmd_flags & REQ_FLUSH) && !(q->flush_flags & REQ_FLUSH));
>> > > +	BUG_ON((rq->cmd_flags & REQ_FUA) && !(q->flush_flags & REQ_FUA));
>> > > +
>> > 
>> > Actually this makes sense and is simple. :-) Is BUG_ON() too harsh, how
>> > about WARN_ONCE() variants? To me system continues to work so warning 
>> > is probably good enough.
>> 
>> Sure, WARN_ONCE() is fine by me.
>> 
>> Seems Tejun wants a more involved fix though.
>
> Fixing it properly doesn't hurt. Makes it more future proof. In fact I am
> thinking what happens to blk_execute_rq() variants where one can prepare a
> request and send it down. What if caller sets FLUSH/FUA flags there.

Callers of blk_execute_rq are special.  Those aren't REQ_TYPE_FS
requests, and so the callers are responsible for doing their own
sequencing.

Cheers,
Jeff