[PATCH 5/5] drm: drop DRM_AUTH from PRIME_TO/FROM_HANDLE ioctls

Emil Velikov emil.l.velikov at gmail.com
Wed Nov 27 18:32:56 UTC 2019


On Wed, 27 Nov 2019 at 18:04, Daniel Vetter <daniel at ffwll.ch> wrote:
>
> On Wed, Nov 27, 2019 at 04:27:29PM +0000, Emil Velikov wrote:
> > On Wed, 27 Nov 2019 at 07:41, Boris Brezillon
> > <boris.brezillon at collabora.com> wrote:
> > >
> > > Hi Emil,
> > >
> > > On Fri,  1 Nov 2019 13:03:13 +0000
> > > Emil Velikov <emil.l.velikov at gmail.com> wrote:
> > >
> > > > From: Emil Velikov <emil.velikov at collabora.com>
> > > >
> > > > As mentioned by Christian, for drivers which support only primary nodes
> > > > this changes the returned error from -EACCES into -EOPNOTSUPP/-ENOSYS.
> > >
> > > Are you sure this is true for MODESET-only nodes (those that do not
> > > have the RENDER cap set) implementing ->{fd_to_handle,handle_to_fd}()?
> > > Shouldn't the is_authenticated() check still be done in that case?
> > >
> > Thanks for catching this. Just sent out v2, which I should address the concern.
>
> Why do we need this additional check in v2? What can go wrong on modeset
> drivers if non-authenticated legacy things can use this? modeset-only
> drivers have all their resources segregated by the drm core (drm_fb,
> mmaps, buffer lists), so there's really no access limitations that can go
> wrong here.

Welcome back Daniel.

I haven't audited the core drm code, so wasn't sure if there's any
issues that may arise.
Hence the conservative approach in v2.

If you think this is fine as-is a formal Reviewed-by would be highly
appreciated.

Thanks
Emil


More information about the amd-gfx mailing list