drm/i915: Improve record of hung engines in error state
authorTvrtko Ursulin <tvrtko.ursulin@intel.com>
Wed, 4 Nov 2020 13:47:42 +0000 (13:47 +0000)
committerTvrtko Ursulin <tvrtko.ursulin@intel.com>
Mon, 9 Nov 2020 11:59:43 +0000 (11:59 +0000)
commitbda3002485a3dcf975b4ecf3fab9dd53f228db6d
treea0368c3483d02ded18849335ee5e2bb317745390
parentad18fa0f5f052046cad96fee762b5c64f42dd86a
drm/i915: Improve record of hung engines in error state

Between events which trigger engine and GPU resets and capturing the error
state we lose information on which engine triggered the reset. Improve
this by passing in the hung engine mask down to error capture.

Result is that the list of engines in user visible "GPU HANG: ecode
<gen>:<engines>:<ecode>, <process>" is now a list of hanging and not just
active engines. Most importantly the displayed process is now the one
which was actually hung.

v2:
 * Stub prototype. (Chris)

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20201104134743.916027-1-tvrtko.ursulin@linux.intel.com
drivers/gpu/drm/i915/gt/intel_lrc.c
drivers/gpu/drm/i915/gt/intel_reset.c
drivers/gpu/drm/i915/i915_debugfs.c
drivers/gpu/drm/i915/i915_gpu_error.c
drivers/gpu/drm/i915/i915_gpu_error.h