diff options
author | Tvrtko Ursulin <tvrtko.ursulin@intel.com> | 2017-05-16 10:27:45 +0100 |
---|---|---|
committer | Tvrtko Ursulin <tvrtko.ursulin@intel.com> | 2017-05-16 10:27:45 +0100 |
commit | a47419f95c67665ae67b22de70346ec0a99c7b84 (patch) | |
tree | 70406cafe91868124646fef81e363e8b00b482a5 /benchmarks | |
parent | 26a5e655883f2b0a2ff903cb24e40d6e2570dbe3 (diff) |
gem_wsim: Fence support
Add sync fence dependency support to workload steps.
Only one sync fence dependency per step is supported at the
moment.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Diffstat (limited to 'benchmarks')
-rw-r--r-- | benchmarks/gem_wsim.c | 161 | ||||
-rw-r--r-- | benchmarks/wsim/README | 28 |
2 files changed, 154 insertions, 35 deletions
diff --git a/benchmarks/gem_wsim.c b/benchmarks/gem_wsim.c index 5cf59a68..840353e5 100644 --- a/benchmarks/gem_wsim.c +++ b/benchmarks/gem_wsim.c @@ -76,6 +76,12 @@ enum w_type QD_THROTTLE }; +struct deps +{ + int nr; + int *list; +}; + struct w_step { /* Workload step metadata */ @@ -83,8 +89,9 @@ struct w_step unsigned int context; unsigned int engine; struct duration duration; - int nr_deps; - int *dep; + struct deps data_deps; + struct deps fence_deps; + int emit_fence; int wait; /* Implementation details */ @@ -179,28 +186,41 @@ parse_dependencies(unsigned int nr_steps, struct w_step *w, char *_desc) { char *desc = strdup(_desc); char *token, *tctx = NULL, *tstart = desc; - int dep; igt_assert(desc); - - w->nr_deps = 0; - w->dep = NULL; + igt_assert(!w->data_deps.nr && w->data_deps.nr == w->fence_deps.nr); + igt_assert(!w->data_deps.list && + w->data_deps.list == w->fence_deps.list); while ((token = strtok_r(tstart, "/", &tctx)) != NULL) { + char *str = token; + struct deps *deps; + int dep; + tstart = NULL; - dep = atoi(token); + if (strlen(token) > 1 && token[0] == 'f') { + deps = &w->fence_deps; + str++; + } else { + deps = &w->data_deps; + } + + dep = atoi(str); if (dep > 0 || ((int)nr_steps + dep) < 0) { - if (w->dep) - free(w-dep); + if (deps->list) + free(deps->list); return -1; } if (dep < 0) { - w->nr_deps++; - w->dep = realloc(w->dep, sizeof(*w->dep) * w->nr_deps); - igt_assert(w->dep); - w->dep[w->nr_deps - 1] = dep; + deps->nr++; + /* Multiple fences not yet supported. */ + igt_assert(deps->nr == 1 || deps != &w->fence_deps); + deps->list = realloc(deps->list, + sizeof(*deps->list) * deps->nr); + igt_assert(deps->list); + deps->list[deps->nr - 1] = dep; } } @@ -220,7 +240,7 @@ parse_workload(char *_desc, unsigned int flags, struct workload *app_w) struct w_step step, *steps = NULL; bool bcs_used = false; unsigned int valid; - int tmp; + int i, j, tmp; igt_assert(desc); @@ -332,7 +352,7 @@ parse_workload(char *_desc, unsigned int flags, struct workload *app_w) } if ((field = strtok_r(fstart, ".", &fctx)) != NULL) { - unsigned int i, old_valid = valid; + unsigned int old_valid = valid; fstart = NULL; @@ -441,8 +461,6 @@ add_step: } if (app_w) { - unsigned int i; - steps = realloc(steps, sizeof(step) * (nr_steps + app_w->nr_steps)); igt_assert(steps); @@ -464,6 +482,19 @@ add_step: free(desc); + /* + * Tag all steps which need to emit a sync fence if another step is + * referencing them as a sync fence dependency. + */ + for (i = 0; i < nr_steps; i++) { + for (j = 0; j < steps[i].fence_deps.nr; j++) { + tmp = steps[i].idx + steps[i].fence_deps.list[j]; + igt_assert(tmp >= 0 && tmp < i); + igt_assert(steps[tmp].type == BATCH); + steps[tmp].emit_fence = -1; + } + } + if (bcs_used && verbose) printf("BCS usage in workload with VCS2 remapping enabled!\n"); @@ -608,6 +639,10 @@ eb_update_flags(struct w_step *w, enum intel_engine_id engine, w->eb.flags |= I915_EXEC_HANDLE_LUT; w->eb.flags |= I915_EXEC_NO_RELOC; + + igt_assert(w->emit_fence <= 0); + if (w->emit_fence) + w->eb.flags |= I915_EXEC_FENCE_OUT; } static void @@ -615,7 +650,7 @@ alloc_step_batch(struct workload *wrk, struct w_step *w, unsigned int flags) { enum intel_engine_id engine = w->engine; unsigned int j = 0; - unsigned int nr_obj = 3 + w->nr_deps; + unsigned int nr_obj = 3 + w->data_deps.nr; unsigned int i; w->obj = calloc(nr_obj, sizeof(*w->obj)); @@ -631,10 +666,10 @@ alloc_step_batch(struct workload *wrk, struct w_step *w, unsigned int flags) igt_assert(j < nr_obj); } - for (i = 0; i < w->nr_deps; i++) { - igt_assert(w->dep[i] <= 0); - if (w->dep[i]) { - int dep_idx = w->idx + w->dep[i]; + for (i = 0; i < w->data_deps.nr; i++) { + igt_assert(w->data_deps.list[i] <= 0); + if (w->data_deps.list[i]) { + int dep_idx = w->idx + w->data_deps.list[i]; igt_assert(dep_idx >= 0 && dep_idx < wrk->nr_steps); igt_assert(wrk->steps[dep_idx].type == BATCH); @@ -1231,6 +1266,65 @@ static void init_status_page(struct workload *wrk, unsigned int flags) } } +#define LOCAL_IOCTL_I915_GEM_EXECBUFFER2_WR DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GEM_EXECBUFFER2, struct drm_i915_gem_execbuffer2) + +static int __gem_execbuf_wr(int gemfd, struct drm_i915_gem_execbuffer2 *execbuf) +{ + int err = 0; + if (igt_ioctl(gemfd, LOCAL_IOCTL_I915_GEM_EXECBUFFER2_WR, execbuf)) + err = -errno; + errno = 0; + return err; +} + +static void gem_execbuf_wr(int gemfd, struct drm_i915_gem_execbuffer2 *execbuf) +{ + igt_assert_eq(__gem_execbuf_wr(gemfd, execbuf), 0); +} + +static void +do_eb(struct workload *wrk, struct w_step *w, enum intel_engine_id engine, + unsigned int flags) +{ + unsigned int i; + + eb_update_flags(w, engine, flags); + + wrk->seqno[engine]++; + + if (flags & SEQNO) + update_bb_seqno(w, engine, wrk->seqno[engine]); + if (flags & RT) + update_bb_rt(w, engine, wrk->seqno[engine]); + + w->eb.batch_start_offset = + ALIGN(w->bb_sz - get_bb_sz(get_duration(w)), + 2 * sizeof(uint32_t)); + + for (i = 0; i < w->fence_deps.nr; i++) { + int tgt = w->idx + w->fence_deps.list[i]; + + /* TODO: fence merging needed to support multiple inputs */ + igt_assert(i == 0); + igt_assert(tgt >= 0 && tgt < w->idx); + igt_assert(wrk->steps[tgt].emit_fence > 0); + + w->eb.flags |= I915_EXEC_FENCE_IN; + w->eb.rsvd2 = wrk->steps[tgt].emit_fence; + } + + if ((w->eb.flags & I915_EXEC_FENCE_IN) || + (w->eb.flags & I915_EXEC_FENCE_OUT)) + gem_execbuf_wr(fd, &w->eb); + else + gem_execbuf(fd, &w->eb); + + if (w->eb.flags & I915_EXEC_FENCE_OUT) { + w->emit_fence = w->eb.rsvd2 >> 32; + igt_assert(w->emit_fence > 0); + } +} + static void run_workload(unsigned int id, struct workload *wrk, bool background, int pipe_fd, @@ -1300,22 +1394,12 @@ run_workload(unsigned int id, struct workload *wrk, engine = balancer->balance(balancer, wrk, w); wrk->nr_bb[engine]++; } - eb_update_flags(w, engine, flags); - - wrk->seqno[engine]++; - if (flags & SEQNO) - update_bb_seqno(w, engine, wrk->seqno[engine]); - if (flags & RT) - update_bb_rt(w, engine, wrk->seqno[engine]); - - w->eb.batch_start_offset = - ALIGN(w->bb_sz - get_bb_sz(get_duration(w)), - 2 * sizeof(uint32_t)); if (throttle > 0) w_sync_to(wrk, w, i - throttle); - gem_execbuf(fd, &w->eb); + do_eb(wrk, w, engine, flags); + if (w->request != -1) { igt_list_del(&w->rq_link); wrk->nrequest[w->request]--; @@ -1356,6 +1440,15 @@ run_workload(unsigned int id, struct workload *wrk, } } } + + /* Cleanup all fences instantiated in this iteration. */ + for (i = 0, w = wrk->steps; run && (i < wrk->nr_steps); + i++, w++) { + if (w->emit_fence > 0) { + close(w->emit_fence); + w->emit_fence = -1; + } + } } for (i = 0; i < NUM_ENGINES; i++) { diff --git a/benchmarks/wsim/README b/benchmarks/wsim/README index 5d2a90b2..617b1338 100644 --- a/benchmarks/wsim/README +++ b/benchmarks/wsim/README @@ -55,4 +55,30 @@ The above workload described in human language works like this: When workload descriptors are provided on the command line, commas must be used instead of new lines. -Multiple dependencies can be given separated by forward slashes.
\ No newline at end of file +Multiple dependencies can be given separated by forward slashes. + +Example: + + 1.VCS1.3000.0.1 + 1.RCS.3700.0.0 + 1.VCS2.2300.-1/-2.0 + +I this case the last step has a data dependency on both first and second steps. + +Sync (fd) fences +---------------- + +Sync fences are also supported as dependencies. + +To use them put a "f<N>" token in the step dependecy list. N is this case the +same relative step offset to the dependee batch, but instead of the data +dependency an output fence will be emitted at the dependee step, and passed in +as a dependency in the current step. + +Example: + + 1.VCS1.3000.0.1 + 1.RCS.500-1000.-1/f-1.0 + +In this case the second step will have both a data dependency and a sync fence +dependency on the previous step. |