Quoting Tvrtko Ursulin (2019-07-31 06:55:23) > > On 30/07/2019 16:48, Chris Wilson wrote: > > @@ -1183,11 +1189,16 @@ live_engine_reset_workarounds(void *arg) > > goto err; > > } > > > > - ret = igt_spinner_init(&spin, i915); > > - if (ret) > > - goto err; > > + ce = i915_gem_context_get_engine(ctx, engine->id); > > + if (IS_ERR(ce)) > > + continue; > > Can't instead use the for_each_gem_engine pattern as in live_sanitycheck? The problem here is that the verify_wa which then also uses for_each_gem_engine. Just needs to rejiggle the other callers so that the context is always locked. > > - spin->hws = i915_gem_object_create_internal(i915, PAGE_SIZE); > > + spin->hws = i915_gem_object_create_internal(gt->i915, PAGE_SIZE); > > I'd add the i915 local. Maybe, at some point we need to pass gt into [internal] object creation. At the moment, I like gt->i915 for being an eyesore. When there are enough of them, it indicates which API needs adjusting next :) -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
