On Fri, Jun 28, 2019 at 3:05 AM Gerd Hoffmann <kraxel@xxxxxxxxxx> wrote:
>
> On Wed, Jun 26, 2019 at 04:55:20PM -0700, Chia-I Wu wrote:
> > On Wed, Jun 19, 2019 at 11:07 PM Gerd Hoffmann <kraxel@xxxxxxxxxx> wrote:
> > >
> > > Use drm_gem_reservation_object_wait() in virtio_gpu_wait_ioctl().
> > > This also makes the ioctl run lockless.
> > The userspace has a BO cache to avoid freeing BOs immediately but to
> > reuse them on next allocations. The BO cache checks if a BO is busy
> > before reuse, and I am seeing a big negative perf impact because of
> > slow virtio_gpu_wait_ioctl. I wonder if this helps.
>
> Could help indeed (assuming it checks with NOWAIT).
Yeah, that is the case.
>
> How many objects does userspace check in one go typically? Maybe it
> makes sense to add an ioctl which checks a list, to reduce the system
> call overhead.
One. The cache sorts BOs by the time they are freed, and checks only
the first (compatible) BO. If it is idle, cache hit. Otherwise,
cache miss. A new ioctl probably won't help.
>
> > > + if (args->flags & VIRTGPU_WAIT_NOWAIT) {
> > > + obj = drm_gem_object_lookup(file, args->handle);
> > Don't we need a NULL check here?
>
> Yes, we do. Will fix.
>
> thanks,
> Gerd
>
_______________________________________________
Virtualization mailing list
Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linuxfoundation.org/mailman/listinfo/virtualization
