[PATCH i-g-t] i915/gem_ctx_persistence: Check that we cannot hide hangs on old engines

[Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>]

As the kernel loses track of the context's old engines, if we request
that the context is non-persistent then any request on the untracked
engines must be cancelled.

Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
Cc: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
---
 tests/i915/gem_ctx_persistence.c | 62 +++++++++++++++++++++++++++++++-
 1 file changed, 61 insertions(+), 1 deletion(-)

diff --git a/tests/i915/gem_ctx_persistence.c b/tests/i915/gem_ctx_persistence.c
index c54797e9b..0db5c31c2 100644
--- a/tests/i915/gem_ctx_persistence.c
+++ b/tests/i915/gem_ctx_persistence.c
@@ -761,6 +761,51 @@ static void smoketest(int i915)
 	gem_quiescent_gpu(i915);
 }
 
+static void replace_engines_hostile(int i915,
+				    const struct intel_execution_engine2 *e)
+{
+	I915_DEFINE_CONTEXT_PARAM_ENGINES(engines, 1) = {
+		.engines = {{ e->class, e->instance }}
+	};
+	struct drm_i915_gem_context_param param = {
+		.ctx_id = gem_context_create(i915),
+		.param = I915_CONTEXT_PARAM_ENGINES,
+		.value = to_user_pointer(&engines),
+		.size = sizeof(engines),
+	};
+	igt_spin_t *spin[2];
+	int64_t timeout;
+
+	/*
+	 * Suppose the user tries to hide a hanging batch by replacing
+	 * the set of engines on the context so that it's not visible
+	 * at the time of closure? Then we must act when they replace
+	 * the engines!
+	 */
+
+	gem_context_set_persistence(i915, param.ctx_id, false);
+
+	gem_context_set_param(i915, &param);
+	spin[0] = igt_spin_new(i915, param.ctx_id,
+			       .flags = IGT_SPIN_NO_PREEMPTION);
+
+	gem_context_set_param(i915, &param);
+	spin[1] = igt_spin_new(i915, param.ctx_id,
+			       .flags = IGT_SPIN_NO_PREEMPTION);
+
+	gem_context_destroy(i915, param.ctx_id);
+
+	timeout = reset_timeout_ms * NSEC_PER_MSEC;
+	igt_assert_eq(gem_wait(i915, spin[1]->handle, &timeout), 0);
+
+	timeout = reset_timeout_ms * NSEC_PER_MSEC;
+	igt_assert_eq(gem_wait(i915, spin[0]->handle, &timeout), 0);
+
+	igt_spin_free(i915, spin[1]);
+	igt_spin_free(i915, spin[0]);
+	gem_quiescent_gpu(i915);
+}
+
 int i915;
 
 static void exit_handler(int sig)
@@ -793,10 +838,10 @@ igt_main
 		igt_assert(igt_sysfs_set_parameter
 			   (i915, "reset", "%d", -1 /* any [default] reset */));
 
-		igt_require(has_persistence(i915));
 		enable_hangcheck(i915);
 		igt_install_exit_handler(exit_handler);
 
+		igt_require(has_persistence(i915));
 		igt_allow_hang(i915, 0, 0);
 	}
 
@@ -861,6 +906,21 @@ igt_main
 			smoketest(i915);
 	}
 
+	/* Check interactions with set-engines */
+	igt_subtest_group {
+		const struct intel_execution_engine2 *e;
+
+		igt_fixture
+			gem_require_contexts(i915);
+
+		igt_subtest_with_dynamic("replace-hostile") {
+			__for_each_physical_engine(i915, e) {
+				igt_dynamic_f("%s", e->name)
+					replace_engines_hostile(i915, e);
+			}
+		}
+	}
+
 	igt_fixture {
 		close(i915);
 	}
-- 
2.25.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx