Re: [RFC][PATCH 3/3] kselftest: Add drm_syncobj API test tool

[Daniel Vetter]

On Tue, Jul 12, 2022 at 08:52:53AM -0700, John Stultz wrote:
> On Tue, Jul 12, 2022 at 12:43 AM Christian König
> <christian.koenig@xxxxxxx> wrote:
> > Am 12.07.22 um 06:22 schrieb John Stultz:
> > > An initial pass at a drm_syncobj API test.
> > >
> > > Currently covers trivial use of:
> > >    DRM_IOCTL_SYNCOBJ_CREATE
> > >    DRM_IOCTL_SYNCOBJ_DESTROY
> > >    DRM_IOCTL_SYNCOBJ_HANDLE_TO_FD
> > >    DRM_IOCTL_SYNCOBJ_FD_TO_HANDLE
> > >    DRM_IOCTL_SYNCOBJ_WAIT
> > >    DRM_IOCTL_SYNCOBJ_RESET
> > >    DRM_IOCTL_SYNCOBJ_SIGNAL
> > >    DRM_IOCTL_SYNCOBJ_TIMELINE_WAIT
> > >    DRM_IOCTL_SYNCOBJ_TIMELINE_SIGNAL
> > >
> > > And demonstrates how the userspace API can be used, along with
> > > some fairly simple bad parameter checking.
> > >
> > > The patch includes a few helpers taken from libdrm, as at least
> > > on the VM I was testing with, I didn't have a new enough libdrm
> > > to support the *_TIMELINE_* ioctls. Ideally the ioctl-helper bits
> > > can be dropped at a later time.
> > >
> > > Feedback would be appreciated!
> >
> > DRM userspace selftests usually go either into libdrm or igt and not
> > into the kernel source.
> 
> Appreciate the pointer, I'll rework and submit to one of those projects.

https://dri.freedesktop.org/docs/drm/gpu/drm-uapi.html#testing-and-validation

There should be already a ton of syncobj tests, so probably more just work
needed to make them work on vgem so we can test them all without a
suitable hw driver loaded.

Cheers, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch