drm/i915/selftests: Retarget igt_render_engine_reset_fallback()
authorChris Wilson <chris@chris-wilson.co.uk>
Fri, 28 Jul 2017 11:21:10 +0000 (12:21 +0100)
committerChris Wilson <chris@chris-wilson.co.uk>
Fri, 4 Aug 2017 18:08:30 +0000 (19:08 +0100)
The purpose of the test was to check per-engine resets would fallback to
the global reset when required, but first we actually need a test for a
basic i915_handle_error()!

Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170728112110.6464-1-chris@chris-wilson.co.uk
drivers/gpu/drm/i915/selftests/intel_hangcheck.c

index 7c55a4c..208b34e 100644 (file)
@@ -770,22 +770,20 @@ unlock:
        return err;
 }
 
-static int igt_render_engine_reset_fallback(void *arg)
+static int igt_handle_error(void *arg)
 {
        struct drm_i915_private *i915 = arg;
        struct intel_engine_cs *engine = i915->engine[RCS];
        struct hang h;
        struct drm_i915_gem_request *rq;
-       unsigned int reset_count, reset_engine_count;
-       int err = 0;
+       struct i915_gpu_state *error;
+       int err;
 
        /* Check that we can issue a global GPU and engine reset */
 
        if (!intel_has_reset_engine(i915))
                return 0;
 
-       global_reset_lock(i915);
-
        mutex_lock(&i915->drm.struct_mutex);
 
        err = hang_init(&h, i915);
@@ -801,69 +799,31 @@ static int igt_render_engine_reset_fallback(void *arg)
        i915_gem_request_get(rq);
        __i915_add_request(rq, true);
 
-       /* make reset engine fail */
-       rq->fence.error = -EIO;
-
        if (!wait_for_hang(&h, rq)) {
                pr_err("Failed to start request %x\n", rq->fence.seqno);
                err = -EIO;
                goto err_request;
        }
 
-       reset_engine_count = i915_reset_engine_count(&i915->gpu_error, engine);
-       reset_count = fake_hangcheck(rq);
-
-       /* unlock since we'll call handle_error */
        mutex_unlock(&i915->drm.struct_mutex);
-       global_reset_unlock(i915);
 
-       i915_handle_error(i915, intel_engine_flag(engine), "live test");
+       /* Temporarily disable error capture */
+       error = xchg(&i915->gpu_error.first_error, (void *)-1);
 
-       if (i915_reset_engine_count(&i915->gpu_error, engine) !=
-           reset_engine_count) {
-               pr_err("render engine reset recorded! (full reset expected)\n");
-               err = -EINVAL;
-               goto out_rq;
-       }
+       engine->hangcheck.stalled = true;
+       engine->hangcheck.seqno = intel_engine_get_seqno(engine);
 
-       if (i915_reset_count(&i915->gpu_error) == reset_count) {
-               pr_err("No full GPU reset recorded!\n");
-               err = -EINVAL;
-               goto out_rq;
-       }
+       i915_handle_error(i915, intel_engine_flag(engine), "%s", __func__);
 
-       /*
-        * by using fence.error = -EIO, full reset sets the wedged flag, do one
-        * more full reset to re-enable the hw.
-        */
-       if (i915_terminally_wedged(&i915->gpu_error)) {
-               global_reset_lock(i915);
-               rq->fence.error = 0;
-
-               mutex_lock(&i915->drm.struct_mutex);
-               set_bit(I915_RESET_HANDOFF, &i915->gpu_error.flags);
-               i915_reset(i915, I915_RESET_QUIET);
-               GEM_BUG_ON(test_bit(I915_RESET_HANDOFF,
-                                   &i915->gpu_error.flags));
-               mutex_unlock(&i915->drm.struct_mutex);
-
-               if (i915_reset_count(&i915->gpu_error) == reset_count) {
-                       pr_err("No full GPU reset recorded!\n");
-                       err = -EINVAL;
-                       goto out_rq;
-               }
-       }
-
-out_rq:
-       i915_gem_request_put(rq);
-       hang_fini(&h);
-out_backoff:
-       global_reset_unlock(i915);
+       xchg(&i915->gpu_error.first_error, error);
 
-       if (i915_terminally_wedged(&i915->gpu_error))
-               return -EIO;
+       mutex_lock(&i915->drm.struct_mutex);
 
-       return err;
+       if (rq->fence.error != -EIO) {
+               pr_err("Guilty request not identified!\n");
+               err = -EINVAL;
+               goto err_request;
+       }
 
 err_request:
        i915_gem_request_put(rq);
@@ -871,7 +831,7 @@ err_fini:
        hang_fini(&h);
 err_unlock:
        mutex_unlock(&i915->drm.struct_mutex);
-       goto out_backoff;
+       return err;
 }
 
 int intel_hangcheck_live_selftests(struct drm_i915_private *i915)
@@ -883,7 +843,7 @@ int intel_hangcheck_live_selftests(struct drm_i915_private *i915)
                SUBTEST(igt_reset_active_engines),
                SUBTEST(igt_wait_reset),
                SUBTEST(igt_reset_queue),
-               SUBTEST(igt_render_engine_reset_fallback),
+               SUBTEST(igt_handle_error),
        };
 
        if (!intel_has_gpu_reset(i915))