[igt-dev] [PATCH V4] i915/gem_exec_nop:Adjusted test to utilize all available engines
Chris Wilson
chris at chris-wilson.co.uk
Thu Jan 23 09:27:36 UTC 2020
Quoting Arjun Melkaveri (2020-01-23 09:18:24)
> Added __for_each_physical_engine to utilize all available engines.
> Moved single, signal, preempt, poll and headless test cases
> from static to dynamic group.
>
> Cc: Dec Katarzyna <katarzyna.dec at intel.com>
> Cc: Kempczynski Zbigniew <zbigniew.kempczynski at intel.com>
> Cc: Tahvanainen Jari <jari.tahvanainen at intel.com>
> Cc: Ursulin Tvrtko <tvrtko.ursulin at intel.com>
> Signed-off-by: Arjun Melkaveri <arjun.melkaveri at intel.com>
> ---
> V1 -> V2
>
> Addressed Tvrtko review comments
> 1) removed gem_require_ring
> 2) replaced gem_can_store_dword with gem_class_can_store_dword
> 3) Fixed WhiteSpace issues.
> ---
> V2 -> V3
>
> Added back missing code. i.e. MIN_PRIO
> ---
> V3 -> V4
>
> 1) Added gem_context_set_all_engines , that was deleted accidentally
> 2) Removed gem_require_ring from fence_signal
> 3) Passing NULL in fence_signal to run test for all engines.
> ---
> tests/i915/gem_exec_nop.c | 164 ++++++++++++++++++++++----------------
> 1 file changed, 94 insertions(+), 70 deletions(-)
>
> diff --git a/tests/i915/gem_exec_nop.c b/tests/i915/gem_exec_nop.c
> index dbedb356..c2d45e5e 100644
> --- a/tests/i915/gem_exec_nop.c
> +++ b/tests/i915/gem_exec_nop.c
> @@ -66,8 +66,9 @@ static double elapsed(const struct timespec *start, const struct timespec *end)
> (end->tv_nsec - start->tv_nsec)*1e-9);
> }
>
> -static double nop_on_ring(int fd, uint32_t handle, unsigned ring_id,
> - int timeout, unsigned long *out)
> +static double nop_on_ring(int fd, uint32_t handle,
> + const struct intel_execution_engine2 *e, int timeout,
> + unsigned long *out)
> {
> struct drm_i915_gem_execbuffer2 execbuf;
> struct drm_i915_gem_exec_object2 obj;
> @@ -80,11 +81,11 @@ static double nop_on_ring(int fd, uint32_t handle, unsigned ring_id,
> memset(&execbuf, 0, sizeof(execbuf));
> execbuf.buffers_ptr = to_user_pointer(&obj);
> execbuf.buffer_count = 1;
> - execbuf.flags = ring_id;
> + execbuf.flags = e->flags;
> execbuf.flags |= LOCAL_I915_EXEC_HANDLE_LUT;
> execbuf.flags |= LOCAL_I915_EXEC_NO_RELOC;
> if (__gem_execbuf(fd, &execbuf)) {
> - execbuf.flags = ring_id;
> + execbuf.flags = e->flags;
> gem_execbuf(fd, &execbuf);
> }
> intel_detect_and_clear_missed_interrupts(fd);
> @@ -104,7 +105,8 @@ static double nop_on_ring(int fd, uint32_t handle, unsigned ring_id,
> return elapsed(&start, &now);
> }
>
> -static void poll_ring(int fd, unsigned engine, const char *name, int timeout)
> +static void poll_ring(int fd, const struct intel_execution_engine2 *e,
> + int timeout)
> {
> const int gen = intel_gen(intel_get_drm_devid(fd));
> const uint32_t MI_ARB_CHK = 0x5 << 23;
> @@ -121,9 +123,8 @@ static void poll_ring(int fd, unsigned engine, const char *name, int timeout)
> if (gen == 4 || gen == 5)
> flags |= I915_EXEC_SECURE;
>
> - gem_require_ring(fd, engine);
> - igt_require(gem_can_store_dword(fd, engine));
> - igt_require(gem_engine_has_mutable_submission(fd, engine));
> + igt_require(gem_class_can_store_dword(fd, e->class));
> + igt_require(gem_class_has_mutable_submission(fd, e->class));
>
> memset(&obj, 0, sizeof(obj));
> obj.handle = gem_create(fd, 4096);
> @@ -187,7 +188,7 @@ static void poll_ring(int fd, unsigned engine, const char *name, int timeout)
> memset(&execbuf, 0, sizeof(execbuf));
> execbuf.buffers_ptr = to_user_pointer(&obj);
> execbuf.buffer_count = 1;
> - execbuf.flags = engine | flags;
> + execbuf.flags = e->flags | flags;
>
> cycles = 0;
> do {
> @@ -209,7 +210,7 @@ static void poll_ring(int fd, unsigned engine, const char *name, int timeout)
> gem_sync(fd, obj.handle);
>
> igt_info("%s completed %ld cycles: %.3f us\n",
> - name, cycles, elapsed*1e-3/cycles);
> + e->name, cycles, elapsed*1e-3/cycles);
>
> munmap(batch, 4096);
> gem_close(fd, obj.handle);
> @@ -218,6 +219,7 @@ static void poll_ring(int fd, unsigned engine, const char *name, int timeout)
> static void poll_sequential(int fd, const char *name, int timeout)
> {
> const int gen = intel_gen(intel_get_drm_devid(fd));
> + const struct intel_execution_engine2 *e;
> const uint32_t MI_ARB_CHK = 0x5 << 23;
> struct drm_i915_gem_execbuffer2 execbuf;
> struct drm_i915_gem_exec_object2 obj[2];
> @@ -234,13 +236,14 @@ static void poll_sequential(int fd, const char *name, int timeout)
> flags |= I915_EXEC_SECURE;
>
> nengine = 0;
> - for_each_physical_engine(e, fd) {
> - if (!gem_can_store_dword(fd, eb_ring(e)) ||
> - !gem_engine_has_mutable_submission(fd, eb_ring(e)))
> + __for_each_physical_engine(fd, e) {
> + if (!gem_class_can_store_dword(fd, e->class) ||
> + !gem_class_has_mutable_submission(fd, e->class))
> continue;
>
> - engines[nengine++] = eb_ring(e);
> + engines[nengine++] = e->flags;
> }
> +
> igt_require(nengine);
>
> memset(obj, 0, sizeof(obj));
> @@ -344,21 +347,20 @@ static void poll_sequential(int fd, const char *name, int timeout)
> }
>
> static void single(int fd, uint32_t handle,
> - unsigned ring_id, const char *ring_name)
> + const struct intel_execution_engine2 *e)
> {
> double time;
> unsigned long count;
>
> - gem_require_ring(fd, ring_id);
> -
> - time = nop_on_ring(fd, handle, ring_id, 20, &count);
> + time = nop_on_ring(fd, handle, e, 20, &count);
> igt_info("%s: %'lu cycles: %.3fus\n",
> - ring_name, count, time*1e6 / count);
> + e->name, count, time*1e6 / count);
> }
>
> static double
> -stable_nop_on_ring(int fd, uint32_t handle, unsigned int engine,
> - int timeout, int reps)
> +stable_nop_on_ring(int fd, uint32_t handle,
> + const struct intel_execution_engine2 *e, int timeout,
> + int reps)
> {
> igt_stats_t s;
> double n;
> @@ -372,7 +374,7 @@ stable_nop_on_ring(int fd, uint32_t handle, unsigned int engine,
> unsigned long count;
> double time;
>
> - time = nop_on_ring(fd, handle, engine, timeout, &count);
> + time = nop_on_ring(fd, handle, e, timeout, &count);
> igt_stats_push_float(&s, time / count);
> }
>
> @@ -388,7 +390,8 @@ stable_nop_on_ring(int fd, uint32_t handle, unsigned int engine,
> "'%s' != '%s' (%f not within %f%% tolerance of %f)\n",\
> #x, #ref, x, tolerance * 100.0, ref)
>
> -static void headless(int fd, uint32_t handle)
> +static void headless(int fd, uint32_t handle,
> + const struct intel_execution_engine2 *e)
> {
> unsigned int nr_connected = 0;
> drmModeConnector *connector;
> @@ -411,7 +414,7 @@ static void headless(int fd, uint32_t handle)
> kmstest_set_vt_graphics_mode();
>
> /* benchmark nops */
> - n_display = stable_nop_on_ring(fd, handle, I915_EXEC_DEFAULT, 1, 5);
> + n_display = stable_nop_on_ring(fd, handle, e, 1, 5);
> igt_info("With one display connected: %.2fus\n",
> n_display * 1e6);
>
> @@ -419,7 +422,7 @@ static void headless(int fd, uint32_t handle)
> kmstest_unset_all_crtcs(fd, res);
>
> /* benchmark nops again */
> - n_headless = stable_nop_on_ring(fd, handle, I915_EXEC_DEFAULT, 1, 5);
> + n_headless = stable_nop_on_ring(fd, handle, e, 1, 5);
> igt_info("Without a display connected (headless): %.2fus\n",
> n_headless * 1e6);
>
> @@ -429,6 +432,7 @@ static void headless(int fd, uint32_t handle)
>
> static void parallel(int fd, uint32_t handle, int timeout)
> {
> + const struct intel_execution_engine2 *e;
> struct drm_i915_gem_execbuffer2 execbuf;
> struct drm_i915_gem_exec_object2 obj;
> unsigned engines[16];
> @@ -439,12 +443,11 @@ static void parallel(int fd, uint32_t handle, int timeout)
>
> sum = 0;
> nengine = 0;
> - for_each_physical_engine(e, fd) {
> - engines[nengine] = eb_ring(e);
> - names[nengine] = e->name;
> - nengine++;
> + __for_each_physical_engine(fd, e) {
> + engines[nengine] = e->flags;
> + names[nengine++] = e->name;
>
> - time = nop_on_ring(fd, handle, eb_ring(e), 1, &count) / count;
> + time = nop_on_ring(fd, handle, e, 1, &count) / count;
> sum += time;
> igt_debug("%s: %.3fus\n", e->name, 1e6*time);
> }
> @@ -490,6 +493,7 @@ static void parallel(int fd, uint32_t handle, int timeout)
>
> static void series(int fd, uint32_t handle, int timeout)
> {
> + const struct intel_execution_engine2 *e;
> struct drm_i915_gem_execbuffer2 execbuf;
> struct drm_i915_gem_exec_object2 obj;
> struct timespec start, now, sync;
> @@ -500,8 +504,8 @@ static void series(int fd, uint32_t handle, int timeout)
> const char *name;
>
> nengine = 0;
> - for_each_physical_engine(e, fd) {
> - time = nop_on_ring(fd, handle, eb_ring(e), 1, &count) / count;
> + __for_each_physical_engine(fd, e) {
> + time = nop_on_ring(fd, handle, e, 1, &count) / count;
> if (time > max) {
> name = e->name;
> max = time;
> @@ -509,7 +513,7 @@ static void series(int fd, uint32_t handle, int timeout)
> if (time < min)
> min = time;
> sum += time;
> - engines[nengine++] = eb_ring(e);
> + engines[nengine++] = e->flags;
> }
> igt_require(nengine);
> igt_info("Maximum execution latency on %s, %.3fus, min %.3fus, total %.3fus per cycle, average %.3fus\n",
> @@ -580,6 +584,7 @@ static void xchg(void *array, unsigned i, unsigned j)
> static void sequential(int fd, uint32_t handle, unsigned flags, int timeout)
> {
> const int ncpus = flags & FORKED ? sysconf(_SC_NPROCESSORS_ONLN) : 1;
> + const struct intel_execution_engine2 *e;
> struct drm_i915_gem_execbuffer2 execbuf;
> struct drm_i915_gem_exec_object2 obj[2];
> unsigned engines[16];
> @@ -595,14 +600,14 @@ static void sequential(int fd, uint32_t handle, unsigned flags, int timeout)
>
> nengine = 0;
> sum = 0;
> - for_each_physical_engine(e, fd) {
> + __for_each_physical_engine(fd, e) {
> unsigned long count;
>
> - time = nop_on_ring(fd, handle, eb_ring(e), 1, &count) / count;
> + time = nop_on_ring(fd, handle, e, 1, &count) / count;
> sum += time;
> igt_debug("%s: %.3fus\n", e->name, 1e6*time);
>
> - engines[nengine++] = eb_ring(e);
> + engines[nengine++] = e->flags;
> }
> igt_require(nengine);
> igt_info("Total (individual) execution latency %.3fus per cycle\n",
> @@ -625,6 +630,7 @@ static void sequential(int fd, uint32_t handle, unsigned flags, int timeout)
>
> igt_require(__gem_context_create(fd, &id) == 0);
> execbuf.rsvd1 = id;
> + gem_context_set_all_engines(fd, execbuf.rsvd1);
> }
>
> for (n = 0; n < nengine; n++) {
> @@ -642,8 +648,10 @@ static void sequential(int fd, uint32_t handle, unsigned flags, int timeout)
> obj[0].handle = gem_create(fd, 4096);
> gem_execbuf(fd, &execbuf);
>
> - if (flags & CONTEXT)
> + if (flags & CONTEXT) {
> execbuf.rsvd1 = gem_context_create(fd);
> + gem_context_set_all_engines(fd, execbuf.rsvd1);
> + }
I do not like gem_context_set_all_engines(). It's magical inference
about the state of the iterator and should be derived or at least
informed by the iterator / iterated-upon-context. I've suggested maybe
we should use a gem_context_clone_with_engines() but it's still an
implicit link to the iterator as to why... And the storm on horizon
suggests we need to keep v4.19 in mind as a target for igt.
-Chris
More information about the igt-dev
mailing list