[igt-dev] [V3] [PATCH i-g-t][V3]tests/i915/gem_exec_fence:Added __for_each_physical_engine to utilize all available engines.

Tvrtko Ursulin tvrtko.ursulin at linux.intel.com
Thu Feb 13 09:53:46 UTC 2020


On 11/02/2020 09:06, saigowth wrote:
> Replaced the legacy for_each_engine* defines with the ones implemented
> in the gem_engine_topology library.

1.
Do you also want to convert the test_long_history() subtest from 
for_each_physical_engine to __for_each_physical_engine?

2.
test_fence_flip() is weird - I don't see why we wouldn't just remove it?

3.
Not important to do straight away, but next time you encounter tests 
with a lot of repetitiveness around the dynamic subtests and engine 
iteration blocks, you can perhaps consider the approach taken in 
gem_ctx_persistence.c. Which is listing tests names, flags and vfuncs in 
an array and then just iterating once. All that is needed is for test 
functions prototypes to be compatible.

Regards,

Tvrtko

> 
> V2
> 
> Corrected subtest name in fast-feedback.testlist from,
> basic-busy-default --> basic-busy
> basic-wait-default --> basic-wait
> basi-await-deafult --> basic-await
> nb-await-default --> nb-await
> 
> Signed-off-by: saigowth <sai.gowtham.ch at intel.com>
> Reviewed-by: Katarzyna Dec <katarzyna.dec at intel.com>
> ---
>   tests/i915/gem_exec_fence.c           | 204 ++++++++++++++++----------
>   tests/intel-ci/fast-feedback.testlist |   8 +-
>   2 files changed, 127 insertions(+), 85 deletions(-)
> 
> diff --git a/tests/i915/gem_exec_fence.c b/tests/i915/gem_exec_fence.c
> index 6d369f58..36488ea7 100644
> --- a/tests/i915/gem_exec_fence.c
> +++ b/tests/i915/gem_exec_fence.c
> @@ -46,7 +46,8 @@ struct sync_merge_data {
>   #define SYNC_IOC_MERGE _IOWR(SYNC_IOC_MAGIC, 3, struct sync_merge_data)
>   #endif
>   
> -static void store(int fd, unsigned ring, int fence, uint32_t target, unsigned offset_value)
> +static void store(int fd, const struct intel_execution_engine2 *e,
> +		  int fence, uint32_t target, unsigned offset_value)
>   {
>   	const int SCRATCH = 0;
>   	const int BATCH = 1;
> @@ -60,7 +61,7 @@ static void store(int fd, unsigned ring, int fence, uint32_t target, unsigned of
>   	memset(&execbuf, 0, sizeof(execbuf));
>   	execbuf.buffers_ptr = to_user_pointer(obj);
>   	execbuf.buffer_count = 2;
> -	execbuf.flags = ring | I915_EXEC_FENCE_IN;
> +	execbuf.flags = e->flags | I915_EXEC_FENCE_IN;
>   	execbuf.rsvd2 = fence;
>   	if (gen < 6)
>   		execbuf.flags |= I915_EXEC_SECURE;
> @@ -108,7 +109,8 @@ static bool fence_busy(int fence)
>   #define NONBLOCK 0x2
>   #define WAIT 0x4
>   
> -static void test_fence_busy(int fd, unsigned ring, unsigned flags)
> +static void test_fence_busy(int fd, const struct intel_execution_engine2 *e,
> +			    unsigned flags)
>   {
>   	const int gen = intel_gen(intel_get_drm_devid(fd));
>   	struct drm_i915_gem_exec_object2 obj;
> @@ -119,14 +121,14 @@ static void test_fence_busy(int fd, unsigned ring, unsigned flags)
>   	int fence, i, timeout;
>   
>   	if ((flags & HANG) == 0)
> -		igt_require(gem_engine_has_mutable_submission(fd, ring));
> +		igt_require(gem_class_has_mutable_submission(fd, e->class));
>   
>   	gem_quiescent_gpu(fd);
>   
>   	memset(&execbuf, 0, sizeof(execbuf));
>   	execbuf.buffers_ptr = to_user_pointer(&obj);
>   	execbuf.buffer_count = 1;
> -	execbuf.flags = ring | I915_EXEC_FENCE_OUT;
> +	execbuf.flags = e->flags | I915_EXEC_FENCE_OUT;
>   
>   	memset(&obj, 0, sizeof(obj));
>   	obj.handle = gem_create(fd, 4096);
> @@ -205,6 +207,7 @@ static void test_fence_busy(int fd, unsigned ring, unsigned flags)
>   
>   static void test_fence_busy_all(int fd, unsigned flags)
>   {
> +	const struct intel_execution_engine2 *e;
>   	const int gen = intel_gen(intel_get_drm_devid(fd));
>   	struct drm_i915_gem_exec_object2 obj;
>   	struct drm_i915_gem_relocation_entry reloc;
> @@ -260,14 +263,14 @@ static void test_fence_busy_all(int fd, unsigned flags)
>   	i++;
>   
>   	all = -1;
> -	for_each_engine(e, fd) {
> +	__for_each_physical_engine(fd, e) {
>   		int fence, new;
>   
>   		if ((flags & HANG) == 0 &&
> -		    !gem_engine_has_mutable_submission(fd, eb_ring(e)))
> +		    !gem_class_has_mutable_submission(fd, e->class))
>   			continue;
>   
> -		execbuf.flags = eb_ring(e) | I915_EXEC_FENCE_OUT;
> +		execbuf.flags = e->flags | I915_EXEC_FENCE_OUT;
>   		execbuf.rsvd2 = -1;
>   		gem_execbuf_wr(fd, &execbuf);
>   		fence = execbuf.rsvd2 >> 32;
> @@ -316,39 +319,41 @@ static void test_fence_busy_all(int fd, unsigned flags)
>   	gem_quiescent_gpu(fd);
>   }
>   
> -static void test_fence_await(int fd, unsigned ring, unsigned flags)
> +static void test_fence_await(int fd, const struct intel_execution_engine2 *e,
> +			     unsigned flags)
>   {
> +	const struct intel_execution_engine2 *e2;
>   	uint32_t scratch = gem_create(fd, 4096);
>   	igt_spin_t *spin;
>   	uint32_t *out;
>   	int i;
>   
>   	if ((flags & HANG) == 0)
> -		igt_require(gem_engine_has_mutable_submission(fd, ring));
> +		igt_require(gem_class_has_mutable_submission(fd, e->class));
>   
> -	igt_require(gem_can_store_dword(fd, 0));
> +	igt_require(gem_class_can_store_dword(fd, 0));
>   
>   	out = gem_mmap__wc(fd, scratch, 0, 4096, PROT_WRITE);
>   	gem_set_domain(fd, scratch,
>   			I915_GEM_DOMAIN_GTT, I915_GEM_DOMAIN_GTT);
>   
>   	spin = igt_spin_new(fd,
> -			    .engine = ring,
> +			    .engine = e->flags,
>   			    .flags = (IGT_SPIN_FENCE_OUT |
>   				      IGT_SPIN_NO_PREEMPTION |
>   				      (flags & HANG ? IGT_SPIN_INVALID_CS : 0)));
>   	igt_assert(spin->out_fence != -1);
>   
>   	i = 0;
> -	for_each_physical_engine(e, fd) {
> -		if (!gem_can_store_dword(fd, eb_ring(e)))
> +	__for_each_physical_engine(fd, e2) {
> +		if (!gem_class_can_store_dword(fd, e->class))
>   			continue;
>   
>   		if (flags & NONBLOCK) {
> -			store(fd, eb_ring(e), spin->out_fence, scratch, i);
> +			store(fd, e2, spin->out_fence, scratch, i);
>   		} else {
>   			igt_fork(child, 1)
> -				store(fd, eb_ring(e), spin->out_fence, scratch, i);
> +				store(fd, e2, spin->out_fence, scratch, i);
>   		}
>   
>   		i++;
> @@ -376,13 +381,14 @@ static void test_fence_await(int fd, unsigned ring, unsigned flags)
>   	gem_close(fd, scratch);
>   }
>   
> -static void resubmit(int fd, uint32_t handle, unsigned int ring, int count)
> +static void resubmit(int fd, uint32_t handle,
> +		     const struct intel_execution_engine2 *e, int count)
>   {
>   	struct drm_i915_gem_exec_object2 obj = { .handle = handle };
>   	struct drm_i915_gem_execbuffer2 execbuf = {
>   		.buffers_ptr = to_user_pointer(&obj),
>   		.buffer_count = 1,
> -		.flags = ring,
> +		.flags = e->flags,
>   	};
>   	while (count--)
>   		gem_execbuf(fd, &execbuf);
> @@ -404,8 +410,9 @@ static int __execbuf(int fd, struct drm_i915_gem_execbuffer2 *execbuf)
>   	return err;
>   }
>   
> -static void test_parallel(int fd, unsigned int master)
> +static void test_parallel(int fd, const struct intel_execution_engine2 *e)
>   {
> +	const struct intel_execution_engine2 *e2;
>   	const int SCRATCH = 0;
>   	const int BATCH = 1;
>   	const int gen = intel_gen(intel_get_drm_devid(fd));
> @@ -426,14 +433,14 @@ static void test_parallel(int fd, unsigned int master)
>   	/* Fill the queue with many requests so that the next one has to
>   	 * wait before it can be executed by the hardware.
>   	 */
> -	spin = igt_spin_new(fd, .engine = master, .dependency = plug);
> -	resubmit(fd, spin->handle, master, 16);
> +	spin = igt_spin_new(fd, .engine = e->flags, .dependency = plug);
> +	resubmit(fd, spin->handle, e, 16);
>   
>   	/* Now queue the master request and its secondaries */
>   	memset(&execbuf, 0, sizeof(execbuf));
>   	execbuf.buffers_ptr = to_user_pointer(obj);
>   	execbuf.buffer_count = 2;
> -	execbuf.flags = master | I915_EXEC_FENCE_OUT;
> +	execbuf.flags = e->flags | I915_EXEC_FENCE_OUT;
>   	if (gen < 6)
>   		execbuf.flags |= I915_EXEC_SECURE;
>   
> @@ -502,11 +509,11 @@ static void test_parallel(int fd, unsigned int master)
>   	obj[BATCH].relocation_count = 1;
>   
>   	/* Queue all secondaries */
> -	for_each_physical_engine(e, fd) {
> -		if (eb_ring(e) == master)
> +	__for_each_physical_engine(fd, e2) {
> +		if (e2->flags == e->flags)
>   			continue;
>   
> -		execbuf.flags = eb_ring(e) | I915_EXEC_FENCE_SUBMIT;
> +		execbuf.flags = e2->flags | I915_EXEC_FENCE_SUBMIT;
>   		if (gen < 6)
>   			execbuf.flags |= I915_EXEC_SECURE;
>   
> @@ -593,7 +600,7 @@ static inline uint32_t upper_32_bits(uint64_t x)
>   	return x >> 32;
>   }
>   
> -static void test_keep_in_fence(int fd, unsigned int engine, unsigned int flags)
> +static void test_keep_in_fence(int fd, const struct intel_execution_engine2 *e)
>   {
>   	struct sigaction sa = { .sa_handler = alarm_handler };
>   	struct drm_i915_gem_exec_object2 obj = {
> @@ -602,14 +609,14 @@ static void test_keep_in_fence(int fd, unsigned int engine, unsigned int flags)
>   	struct drm_i915_gem_execbuffer2 execbuf = {
>   		.buffers_ptr = to_user_pointer(&obj),
>   		.buffer_count = 1,
> -		.flags = engine | I915_EXEC_FENCE_OUT,
> +		.flags = e->flags | I915_EXEC_FENCE_OUT,
>   	};
>   	unsigned long count, last;
>   	struct itimerval itv;
>   	igt_spin_t *spin;
>   	int fence;
>   
> -	spin = igt_spin_new(fd, .engine = engine);
> +	spin = igt_spin_new(fd, .engine = e->flags);
>   
>   	gem_execbuf_wr(fd, &execbuf);
>   	fence = upper_32_bits(execbuf.rsvd2);
> @@ -1373,7 +1380,7 @@ static void test_syncobj_channel(int fd)
>   
>   igt_main
>   {
> -	const struct intel_execution_engine *e;
> +	const struct intel_execution_engine2 *e;
>   	int i915 = -1;
>   
>   	igt_fixture {
> @@ -1412,69 +1419,105 @@ igt_main
>   		}
>   	}
>   
> -	for (e = intel_execution_engines; e->name; e++) {
> -		igt_subtest_group {
> +	igt_subtest_group {
> +		__for_each_physical_engine(i915, e) {
>   			igt_fixture {
> -				igt_require(gem_has_ring(i915, eb_ring(e)));
> -				igt_require(gem_can_store_dword(i915, eb_ring(e)));
> +				igt_require(gem_class_can_store_dword(i915, e->class));
>   			}
> +		}
> +		igt_subtest_group {
> +			igt_fixture {
> +				igt_fork_hang_detector(i915);
> +		}
> +			igt_subtest_with_dynamic("basic-busy") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
>   
> -			igt_subtest_group {
> -				igt_fixture {
> -					igt_fork_hang_detector(i915);
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_busy(i915, e, 0);
>   				}
> -
> -				igt_subtest_f("%sbusy-%s",
> -						e->exec_id == 0 ? "basic-" : "",
> -						e->name)
> -					test_fence_busy(i915, eb_ring(e), 0);
> -				igt_subtest_f("%swait-%s",
> -						e->exec_id == 0 ? "basic-" : "",
> -						e->name)
> -					test_fence_busy(i915, eb_ring(e), WAIT);
> -				igt_subtest_f("%sawait-%s",
> -						e->exec_id == 0 ? "basic-" : "",
> -						e->name)
> -					test_fence_await(i915, eb_ring(e), 0);
> -				igt_subtest_f("nb-await-%s", e->name)
> -					test_fence_await(i915, eb_ring(e), NONBLOCK);
> -
> -				igt_subtest_f("keep-in-fence-%s", e->name)
> -					test_keep_in_fence(i915, eb_ring(e), 0);
> -
> -				if (e->exec_id &&
> -				    !(e->exec_id == I915_EXEC_BSD && !e->flags)) {
> -					igt_subtest_f("parallel-%s", e->name) {
> -						igt_require(has_submit_fence(i915));
> -						igt_until_timeout(2)
> -							test_parallel(i915, eb_ring(e));
> -					}
> +			}
> +			igt_subtest_with_dynamic("basic-wait") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_busy(i915, e, WAIT);
>   				}
> -
> -				igt_fixture {
> -					igt_stop_hang_detector();
> +			}
> +			igt_subtest_with_dynamic("basic-await") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_await(i915, e, 0);
> +				}
> +			}
> +			igt_subtest_with_dynamic("nb-await") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_await(i915,
> +								 e, NONBLOCK);
> +				}
> +			}
> +			igt_subtest_with_dynamic("keep-in-fence") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_keep_in_fence(i915, e);
> +				}
> +			}
> +			igt_subtest_with_dynamic("parallel") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +					igt_require(has_submit_fence(i915));
> +					igt_until_timeout(2)
> +						test_parallel(i915, e);
>   				}
>   			}
>   
> -			igt_subtest_group {
> -				igt_hang_t hang;
> +			igt_fixture {
> +				igt_stop_hang_detector();
> +			}
> +		}
>   
> -				igt_fixture {
> -					hang = igt_allow_hang(i915, 0, 0);
> -				}
> +		igt_subtest_group {
> +			igt_hang_t hang;
>   
> -				igt_subtest_f("busy-hang-%s", e->name)
> -					test_fence_busy(i915, eb_ring(e), HANG);
> -				igt_subtest_f("wait-hang-%s", e->name)
> -					test_fence_busy(i915, eb_ring(e), HANG | WAIT);
> -				igt_subtest_f("await-hang-%s", e->name)
> -					test_fence_await(i915, eb_ring(e), HANG);
> -				igt_subtest_f("nb-await-hang-%s", e->name)
> -					test_fence_await(i915, eb_ring(e), NONBLOCK | HANG);
> -				igt_fixture {
> -					igt_disallow_hang(i915, hang);
> +			igt_fixture {
> +				hang = igt_allow_hang(i915, 0, 0);
> +			}
> +			igt_subtest_with_dynamic("busy-hang") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_busy(i915, e, HANG);
> +				}
> +			}
> +			igt_subtest_with_dynamic("wait-hang") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_busy(i915, e, HANG | WAIT);
> +				}
> +			}
> +			igt_subtest_with_dynamic("await-hang") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_await(i915, e, HANG);
> +				}
> +			}
> +			igt_subtest_with_dynamic("nb-await-hang") {
> +				__for_each_physical_engine(i915, e) {
> +				/* Requires master for STORE_DWORD on gen4/5 */
> +					igt_dynamic_f("%s", e->name)
> +						test_fence_await(i915, e, NONBLOCK | HANG);
>   				}
>   			}
> +			igt_fixture {
> +				igt_disallow_hang(i915, hang);
> +			}
>   		}
>   	}
>   
> @@ -1542,7 +1585,6 @@ igt_main
>   			igt_stop_hang_detector();
>   		}
>   	}
> -
>   	igt_fixture {
>   		close(i915);
>   	}
> diff --git a/tests/intel-ci/fast-feedback.testlist b/tests/intel-ci/fast-feedback.testlist
> index 40d273c1..8a191922 100644
> --- a/tests/intel-ci/fast-feedback.testlist
> +++ b/tests/intel-ci/fast-feedback.testlist
> @@ -16,10 +16,10 @@ igt at gem_ctx_param@basic
>   igt at gem_ctx_param@basic-default
>   igt at gem_exec_basic@basic
>   igt at gem_exec_create@basic
> -igt at gem_exec_fence@basic-busy-default
> -igt at gem_exec_fence@basic-wait-default
> -igt at gem_exec_fence@basic-await-default
> -igt at gem_exec_fence@nb-await-default
> +igt at gem_exec_fence@basic-busy
> +igt at gem_exec_fence@basic-wait
> +igt at gem_exec_fence@basic-await
> +igt at gem_exec_fence@nb-await
>   igt at gem_exec_gttfill@basic
>   igt at gem_exec_parallel@basic
>   igt at gem_exec_parallel@contexts
> 


More information about the igt-dev mailing list