Re: [Qemu-devel] scsi-hd with discard_granularity and unmap results in Aborted Commands

[Stefan Priebe - Profihost AG]

Am 19.11.2012 15:41, schrieb Paolo Bonzini:
> Il 19/11/2012 15:28, Stefan Priebe - Profihost AG ha scritto:
> >   typedef struct RADOSCB {
> > @@ -376,6 +377,10 @@ static void qemu_rbd_complete_aio(RADOSCB *rcb)
> >       RBDAIOCB *acb = rcb->acb;
> >       int64_t r;
> >
> > +    if (acb->bh) {
> > +        return;
> > +    }
> > +
> >       r = rcb->ret;
> >
> >       if (acb->cmd == RBD_AIO_WRITE ||
> > @@ -560,6 +565,20 @@ static void qemu_rbd_close(BlockDriverState *bs)
> >       rados_shutdown(s->cluster);
> >   }
> >
> > +static void qemu_rbd_aio_abort(void *private_data)
> > +{
> > +    RBDAIOCB *acb = (RBDAIOCB *) private_data;
> > +
> > +    acb->status = -ECANCELED;
> > +
> > +    if (acb->bh) {
> > +        return;
> > +    }
> > +
> > +    acb->bh = qemu_bh_new(rbd_aio_bh_cb, acb);
> > +    qemu_bh_schedule(acb->bh);
> > +}
>
> I think this is all unneeded.  Just store rcb->ret into
> rcb->acb->status, and your version of qemu_rbd_aio_cancel should just work.
>
> Also, I think the acb->cancelled field is not necessary anymore after
> these changes.

The iscsi driver still relies on canceled that's why i left it here in too.

So you mean in qemu_rbd_complete_aio i should remove the check for 
cancelled and then just overwrite acb->status to that it changes from 
-EINPROGRESS to something else?

And qemu_rbd_aio_cancel should just wait for status != -EINPROGRESS?

Stefan