[PATCH 3/5] drm/virtio: track whether or not a context has been initiated

[Chia-I Wu]

On Fri, Feb 14, 2020 at 6:29 PM Gurchetan Singh
<gurchetansingh at chromium.org> wrote:
>
> On Fri, Feb 14, 2020 at 11:27 AM Chia-I Wu <olvaffe at gmail.com> wrote:
> >
> > On Thu, Feb 13, 2020 at 3:18 PM Gurchetan Singh
> > <gurchetansingh at chromium.org> wrote:
> > >
> > > Use an atomic variable to track whether a context has been
> > > initiated.
> > >
> > > v2: Fix possible race (@olv)
> > >
> > > Signed-off-by: Gurchetan Singh <gurchetansingh at chromium.org>
> > > ---
> > >  drivers/gpu/drm/virtio/virtgpu_drv.h   | 1 +
> > >  drivers/gpu/drm/virtio/virtgpu_ioctl.c | 3 +++
> > >  drivers/gpu/drm/virtio/virtgpu_kms.c   | 1 +
> > >  3 files changed, 5 insertions(+)
> > >
> > > diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h
> > > index 72c1d9b59dfe..ca505984f8ab 100644
> > > --- a/drivers/gpu/drm/virtio/virtgpu_drv.h
> > > +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h
> > > @@ -209,6 +209,7 @@ struct virtio_gpu_device {
> > >
> > >  struct virtio_gpu_fpriv {
> > >         uint32_t ctx_id;
> > > +       atomic_t context_initiated;
> > >  };
> > >
> > >  /* virtio_ioctl.c */
> > > diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c
> > > index 896c3f419a6d..a98884462944 100644
> > > --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c
> > > +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c
> > > @@ -44,6 +44,9 @@ void virtio_gpu_create_context(struct drm_device *dev,
> > >         if (!vgdev->has_virgl_3d)
> > >                 return;
> > >
> > > +       if (!atomic_add_unless(&vfpriv->context_initiated, 1, 1))
> > > +               return;
> > > +
> > How does this work?  When thread A and B enter this function at the
> > same time, and thread B returns early, it is possible that thread B
> > queues other commands before thread A has the chance to queue
> > virtio_gpu_cmd_context_create.
>
> Good catch, I'll add a spinlock in v3.
virtio_gpu_cmd_context_create can sleep.  You will need a mutex (or
figure out another way to do the lazy initialization).

>
> >
> > >         get_task_comm(dbgname, current);
> > >         virtio_gpu_cmd_context_create(vgdev, vfpriv->ctx_id,
> > >                                       strlen(dbgname), dbgname);
> > > diff --git a/drivers/gpu/drm/virtio/virtgpu_kms.c b/drivers/gpu/drm/virtio/virtgpu_kms.c
> > > index 282558576527..25248bad3fc4 100644
> > > --- a/drivers/gpu/drm/virtio/virtgpu_kms.c
> > > +++ b/drivers/gpu/drm/virtio/virtgpu_kms.c
> > > @@ -263,6 +263,7 @@ int virtio_gpu_driver_open(struct drm_device *dev, struct drm_file *file)
> > >         }
> > >
> > >         vfpriv->ctx_id = handle + 1;
> > > +       atomic_set(&vfpriv->context_initiated, 0);
> > >         file->driver_priv = vfpriv;
> > >         virtio_gpu_create_context(dev, file);
> > >         return 0;
> > > --
> > > 2.25.0.265.gbab2e86ba0-goog
> > >