Quoting Jani Nikula (2018-12-31 15:12:41) > On Mon, 31 Dec 2018, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote: > > Quoting Tvrtko Ursulin (2018-12-31 13:13:26) > >> > >> On 27/12/2018 14:33, Jani Nikula wrote: > >> > Seems like selftests is a better home for everything related to load > >> > failure injection, including the module parameter. > >> > >> Hm not sure I would immediately want to couple the two, however.. > >> > >> > The failure injection code gets moved from under DRM_I915_DEBUG to > >> > DRM_I915_SELFTEST config option. This should be of no consequence, as > >> > the former selects the latter. > >> > >> ... I also don't see why debug would auto-select self-tests. > > > > DRM_I915_DEBUG is now the list of things we use for igt > > s/DRM_I915_DEBUG/DRM_I915_IGT/ > > > > Just so we can have a single switch for "if you want to run igt, enable > > this CONFIG to enable all debug features used by igt". Handy for CI > > testing when more features are required. > > So what say you on the patch? I can keep it in i915_params.c too, it > just needs some extra juggling because it gets used before i915 gets > allocated. So not very easy as a "device param". I had no objections to the patch. When you raised the idea, I was thinking of doing the entire iterative fault injection inside a selftest and remove the kernel parameter entirely :) -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
