[Qemu-devel] Re: [PATCHv4 09/12] vhost: vhost net support

[Amit Shah <amit.shah@redhat.com>]

On (Tue) Mar 16 2010 [17:37:39], Michael S. Tsirkin wrote:
> On Mon, Mar 08, 2010 at 11:50:23AM +0530, Amit Shah wrote:
> > On (Sat) Mar 06 2010 [21:06:35], Michael S. Tsirkin wrote:
> > > 
> > > > > +    r = vhost_virtqueue_set_addr(dev, vq, idx, dev->log_enabled);
> > > > > +    if (r < 0) {
> > > > > +        r = -errno;
> > > > > +        goto fail_alloc;
> > > > > +    }
> > > > > +    if (!vdev->binding->guest_notifier || !vdev->binding->host_notifier) {
> > > > > +        fprintf(stderr, "binding does not support irqfd/queuefd\n");
> > > > > +        r = -ENOSYS;
> > > > > +        goto fail_alloc;
> > > > > +    }
> > > > 
> > > > This could be checked much earlier on in the function; so that we avoid
> > > > doing all that stuff above and the cleanup.
> > > 
> > > Whatever order we put checks in, we'll have to undo stuff
> > > done beforehand on error.
> > 
> > Not if you do this check before any ioctls or allocations.
> > !vdev->binding->guest_notifier is not dependent on anything you do above
> > it in this function, so just checking for this first thing in the
> > function will not need any cleanup.
> 
> Yes, but I think it's clearer to do check function just before
> calling it. No?

The good thing about doing it before is that we can check for whatever
conditions are not met first and exit instead of doing N things and
undoing them later and then exiting -- faster and more efficient in the
case when vhost is not being used.

		Amit
-- 
http://log.amitshah.net/