Quoting Tvrtko Ursulin (2018-12-07 10:45:17) > > @@ -1941,15 +1948,9 @@ void i915_capture_error_state(struct > > drm_i915_private *i915, > > if (!i915_modparams.error_capture) > > return; > > > > - if (READ_ONCE(i915->gpu_error.first_error)) > > - return; > > - > > Will it now overwrite the first error on subsequent hangs? > i915_capture_gpu_state only does early exit on stored error.
No, after we capture the state, we do another check under the spinlock to make sure we don't overwrite existing state. However, that early check is still useful to avoid capturing subsequent errors and throwing them away. -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
