On Thu, May 05, 2016 at 11:18:41AM +0100, Tvrtko Ursulin wrote: > > Hi, > > On 05/05/16 10:15, Chris Wilson wrote: > >Move all of the constant assignments up front and into a common > >function. This is primarily to ensure the backpointers are set as early > >as possible for later use during initialisation. > > > >v2: Use a constant struct so that all the similar values are set > >together. > >v3: Sanitize the engine's IMR to disable any potential interrupt before > >we are ready (enabled in init_hw). > > Same as before - I don't like hardware access in this code path > since we otherwise have it split into a later init_hw phase. And I > don't like engine->dev being used for intel_engine_initialized. I think you raised a good point on the last round! It is an oversight that we have not explicitly sanitized the per-engine registers as is our mo. This gives us the symmetry with the init_hw phase where they are enabled. > On retrospect, interrupt vs engine->irq_queue race is already there > now, for the render ring at least. So maybe just drop the IMR bit > which would make the patch pure refactoring and can have my R-b > then. And this closes a race with a potential interrupt pending from takeover. -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
