[PATCH v3 11/12] drm/virtio: implement context init: add virtio_gpu_fence_event

Fri Nov 19 17:43:28 UTC 2021

On Thu, Nov 18, 2021 at 12:53 AM Daniel Vetter <daniel at ffwll.ch> wrote:
>
> On Tue, Nov 16, 2021 at 06:31:10PM -0800, Gurchetan Singh wrote:
> > On Tue, Nov 16, 2021 at 7:43 AM Daniel Vetter <daniel at ffwll.ch> wrote:
> >
> > > On Mon, Nov 15, 2021 at 07:26:14PM +0000, Kasireddy, Vivek wrote:
> > > > Hi Daniel, Greg,
> > > >
> > > > If it is the same or a similar crash reported here:
> > > >
> > > https://lists.freedesktop.org/archives/dri-devel/2021-November/330018.html
> > > > and here:
> > > https://lists.freedesktop.org/archives/dri-devel/2021-November/330212.html
> > > > then the fix is already merged:
> > > >
> > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d89c0c8322ecdc9a2ec84b959b6f766be082da76
> >
> > Yeah but that still leaves the problem of why exaxtly virtgpu is
> > > reinventing drm_poll here?
> >
> >
> > > Can you please replace it with drm_poll like all other drivers, including
> > > the ones that have private events?
> > >
> >
> > Hi Daniel,
> >
> > Allow me to explain the use case a bit.  It's for when virtgpu KMS is not
> > used, but a special Wayland compositor does wayland passthrough instead:
> >
> > https://www.youtube.com/watch?v=WwrXqDERFm8https://www.youtube.com/watch?v=EkNBsBx501Q
> >
> > This technique has gained much popularity in the virtualized laptop
> > space, where it offers better performance/user experience than virtgpu
> > KMS.  The relevant paravirtualized userspace is "Sommelier":
> >
> > https://chromium.googlesource.com/chromiumos/platform2/+/main/vm_tools/sommelier/
> > https://chromium.googlesource.com/chromiumos/platform2/+/main/vm_tools/sommelier/virtualization/virtgpu_channel.cc
> >
> > Previously, we were using the out-of-tree virtio-wl device and there
> > were many discussions on how we could get this upstream:
> >
> > https://lists.freedesktop.org/archives/dri-devel/2017-December/160309.html
> > https://lists.oasis-open.org/archives/virtio-dev/202002/msg00005.html
> >
> > Extending virtgpu was deemed the least intrusive option:
> >
> > https://www.spinics.net/lists/kvm/msg159206.html
> >
> > We ultimately settled on the context type abstraction and used
> > virtio_gpu_poll to tell the guest "hey, we have a Wayland event".  The
> > host response is actually in a buffer of type BLOB_MEM_GUEST.
> >
> > It is possible to use drm_poll(..), but that would have to be
> > accompanied by a drm_read(..).  You'll need to define a dummy
> > VIRTGPU_EVENT_FENCE_SIGNALED in the uapi too.
> >
> > That's originally how I did it, but some pointed out that's
> > unnecessary since the host response is in the BLOB_MEM_GUEST buffer
> > and virtgpu event is a dummy event.  So we decided just to modify
> > virtio_gpu_poll(..) to have the desired semantics in that case.
> >
> > For the regular virtio-gpu KMS path, things remain unchanged.
> >
> > There are of course other ways to do it (perhaps polling a dma_fence),
> > but that was the cleanest way we could find.
> >
> > It's not rare for virtio to "special things" (see virtio_dma_buf_ops,
> > virtio_dma_ops), since they are in fake devices.
>
> These are all internal interfaces, not uapi.
>
> > We're open to other ideas, but hopefully that answers some of your
> > questions.
>
> Well for one, why does the commit message not explain any of this. You're
> building uapi, which is forever, it's paramount all considerations are
> properly explained.
>
> Second, I really don't like that youre redefining poll semantics in
> incompatible ways from all other drm drivers. If you want special poll
> semantics then just create a sperate fd for that (or a dma_fence or
> whatever, maybe that saves some typing), but bending the drm fd semantics
> is no good at all. We have tons of different fd with their dedicated
> semantics in drm, trying to shoehorn it all into one just isn't very good
> design.
>
> Or do the dummy event which is just the event code, but does not contain
> any data. Either is fine with me.
>
> Can you pls do this asap? I really don't want to bake this in as uapi
> which we then have to quirk and support forever. I'd say revert for -rc2
> of these two and then maybe sort it out properly in -next.

I think as a general rule, if there is not consensus about uabi
change, even if it is just a semantic change, then revert and reland
later is ok..

As far as dummy VIRTGPU_EVENT_FENCE_SIGNALED.. that doesn't entirely
sound like a bad thing to me.  Actually, it sounds like a good thing..
it makes it more explicit what is going on.  And would avoid confusing
a different userspace polling for kms related events expecting to be
able to do a read.

BR,
-R

> Cheers, Daniel
> >
> >
> > > Thanks, Daniel
> > >
> > > >
> > > > Thanks,
> > > > Vivek
> > > >
> > > > > On Sat, Nov 13, 2021 at 03:51:48PM +0100, Greg KH wrote:
> > > > > > On Tue, Sep 21, 2021 at 04:20:23PM -0700, Gurchetan Singh wrote:
> > > > > > > Similar to DRM_VMW_EVENT_FENCE_SIGNALED.  Sends a pollable event
> > > > > > > to the DRM file descriptor when a fence on a specific ring is
> > > > > > > signaled.
> > > > > > >
> > > > > > > One difference is the event is not exposed via the UAPI -- this is
> > > > > > > because host responses are on a shared memory buffer of type
> > > > > > > BLOB_MEM_GUEST [this is the common way to receive responses with
> > > > > > > virtgpu].  As such, there is no context specific read(..)
> > > > > > > implementation either -- just a poll(..) implementation.
> > > > > > >
> > > > > > > Signed-off-by: Gurchetan Singh <gurchetansingh at chromium.org>
> > > > > > > Acked-by: Nicholas Verne <nverne at chromium.org>
> > > > > > > ---
> > > > > > >  drivers/gpu/drm/virtio/virtgpu_drv.c   | 43
> > > +++++++++++++++++++++++++-
> > > > > > >  drivers/gpu/drm/virtio/virtgpu_drv.h   |  7 +++++
> > > > > > >  drivers/gpu/drm/virtio/virtgpu_fence.c | 10 ++++++
> > > > > > >  drivers/gpu/drm/virtio/virtgpu_ioctl.c | 34 ++++++++++++++++++++
> > > > > > >  4 files changed, 93 insertions(+), 1 deletion(-)
> > > > > >
> > > > > > This commit seems to cause a crash in a virtual drm gpu driver for
> > > > > > Android.  I have reverted this, and the next commit in the series
> > > from
> > > > > > Linus's tree and all is good again.
> > > > > >
> > > > > > Any ideas?
> > > > >
> > > > > Well no, but also this patch looks very questionable of hand-rolling
> > > > > drm_poll. Yes you can do driver private events like
> > > > > DRM_VMW_EVENT_FENCE_SIGNALED, that's fine. But you really should not
> > > need
> > > > > to hand-roll the poll callback. vmwgfx (which generally is a very old
> > > > > driver which has lots of custom stuff, so not a great example) doesn't
> > > do
> > > > > that either.
> > > > >
> > > > > So that part should go no matter what I think.
> > > > > -Daniel
> > > > > --
> > > > > Daniel Vetter
> > > > > Software Engineer, Intel Corporation
> > > > > http://blog.ffwll.ch
> > >
> > > --
> > > Daniel Vetter
> > > Software Engineer, Intel Corporation
> > > http://blog.ffwll.ch
> > >
>
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch