On 8 May 2015 at 18:06, Axel Davy <d...@clipper.ens.fr> wrote: > On Fri, 8 May 2015, Emil Velikov wrote: > >> Shouldn't we authenticate with the correct gpu or master/render node ? >> This implementation will auth with GPU1, and then use GPU2 which seems >> a bit odd. I might be missing something ? >> >> > > The original patches did do differently: when GPU1 was discovered to not be > the wanted gpu, it was not authenticating to it. > > However loader_get_user_preferred_fd was introduced and thus that leads to > something different (which I think is cleaner): > > We authenticate to GPU1, so we have one node we can render to. > > loader_get_user_preferred_fd takes the GPU1 fd, and eventually replaces > it by GPU2 fd if user wants it and it is possible. > > Given this way of doing, we it makes sense to auth to GPU1, even if we end > up rendering to GPU2. No objections against loader_get_user_preferred_fd. I'm thinking that if we push it into drm_handle_device() we can avoid a needless (in some cases) auth. Not sure how much it's worth, but it seems like a good idea imho.
-Emil _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev