Re: [Qemu-devel] [PATCH v4 2/3] linux-aio: implement io plug, unplug and flush io queue

[Kevin Wolf]

Am 03.07.2014 um 11:51 hat Ming Lei geschrieben:
> Hi Kevin,
> 
> On Thu, Jul 3, 2014 at 5:40 PM, Kevin Wolf <address@hidden> wrote:
> > Am 02.07.2014 um 14:18 hat Ming Lei geschrieben:
> >> This patch implements .bdrv_io_plug, .bdrv_io_unplug and
> >> .bdrv_flush_io_queue callbacks for linux-aio Block Drivers,
> >> so that submitting I/O as a batch can be supported on linux-aio.
> >>
> >> Signed-off-by: Ming Lei <address@hidden>
> >
> > Just a couple of minor comments, see inline.
> 
> Thanks for your review.
> 
> >> +void laio_io_plug(BlockDriverState *bs, void *aio_ctx)
> >> +{
> >> +    struct qemu_laio_state *s = aio_ctx;
> >> +
> >> +    s->io_q.plugged++;
> >> +}
> >> +
> >> +int laio_io_unplug(BlockDriverState *bs, void *aio_ctx, bool unplug)
> >> +{
> >> +    struct qemu_laio_state *s = aio_ctx;
> >> +    int ret = 0;
> >> +
> >
> > How about an assert(s->io_q.plugged > 0); here?
> 
> how about just adding a warning because flush io queue uses
> the function too?

Good point, this is what the assertion should look like then:

    assert(s->io_q.plugged > 0 || !unplug);

> Also that is why 'plugged' is defined as signed.

I don't understand. The flush function leaves s->io_q.plugged alone
(otherwise it would be buggy), so how can it ever become negative? And
if you say that a negative value is valid, what would it even mean?

> >
> >> +    if (unplug && --s->io_q.plugged > 0)
> >> +        return 0;
> >
> > Missing braces.
> >
> >> +
> >> +    if (s->io_q.idx > 0) {
> >> +        ret = ioq_submit(s);
> >> +    }
> >> +
> >> +    return ret;
> >> +}

Kevin