diff options
Diffstat (limited to 'drivers/gpu/drm/i915/gt/intel_engine_cs.c')
| -rw-r--r-- | drivers/gpu/drm/i915/gt/intel_engine_cs.c | 160 |
1 files changed, 124 insertions, 36 deletions
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c b/drivers/gpu/drm/i915/gt/intel_engine_cs.c index 8691eb61e185..dd1a42c4d344 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c +++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c @@ -370,7 +370,7 @@ static void __setup_engine_capabilities(struct intel_engine_cs *engine) * instances. */ if ((INTEL_GEN(i915) >= 11 && - RUNTIME_INFO(i915)->vdbox_sfc_access & engine->mask) || + engine->gt->info.vdbox_sfc_access & engine->mask) || (INTEL_GEN(i915) >= 9 && engine->instance == 0)) engine->uabi_capabilities |= I915_VIDEO_AND_ENHANCE_CLASS_CAPABILITY_SFC; @@ -414,12 +414,12 @@ void intel_engines_release(struct intel_gt *gt) /* Decouple the backend; but keep the layout for late GPU resets */ for_each_engine(engine, gt, id) { - intel_wakeref_wait_for_idle(&engine->wakeref); - GEM_BUG_ON(intel_engine_pm_is_awake(engine)); - if (!engine->release) continue; + intel_wakeref_wait_for_idle(&engine->wakeref); + GEM_BUG_ON(intel_engine_pm_is_awake(engine)); + engine->release(engine); engine->release = NULL; @@ -450,6 +450,80 @@ void intel_engines_free(struct intel_gt *gt) } } +/* + * Determine which engines are fused off in our particular hardware. + * Note that we have a catch-22 situation where we need to be able to access + * the blitter forcewake domain to read the engine fuses, but at the same time + * we need to know which engines are available on the system to know which + * forcewake domains are present. We solve this by intializing the forcewake + * domains based on the full engine mask in the platform capabilities before + * calling this function and pruning the domains for fused-off engines + * afterwards. + */ +static intel_engine_mask_t init_engine_mask(struct intel_gt *gt) +{ + struct drm_i915_private *i915 = gt->i915; + struct intel_gt_info *info = >->info; + struct intel_uncore *uncore = gt->uncore; + unsigned int logical_vdbox = 0; + unsigned int i; + u32 media_fuse; + u16 vdbox_mask; + u16 vebox_mask; + + info->engine_mask = INTEL_INFO(i915)->platform_engine_mask; + + if (INTEL_GEN(i915) < 11) + return info->engine_mask; + + media_fuse = ~intel_uncore_read(uncore, GEN11_GT_VEBOX_VDBOX_DISABLE); + + vdbox_mask = media_fuse & GEN11_GT_VDBOX_DISABLE_MASK; + vebox_mask = (media_fuse & GEN11_GT_VEBOX_DISABLE_MASK) >> + GEN11_GT_VEBOX_DISABLE_SHIFT; + + for (i = 0; i < I915_MAX_VCS; i++) { + if (!HAS_ENGINE(gt, _VCS(i))) { + vdbox_mask &= ~BIT(i); + continue; + } + + if (!(BIT(i) & vdbox_mask)) { + info->engine_mask &= ~BIT(_VCS(i)); + drm_dbg(&i915->drm, "vcs%u fused off\n", i); + continue; + } + + /* + * In Gen11, only even numbered logical VDBOXes are + * hooked up to an SFC (Scaler & Format Converter) unit. + * In TGL each VDBOX has access to an SFC. + */ + if (INTEL_GEN(i915) >= 12 || logical_vdbox++ % 2 == 0) + gt->info.vdbox_sfc_access |= BIT(i); + } + drm_dbg(&i915->drm, "vdbox enable: %04x, instances: %04lx\n", + vdbox_mask, VDBOX_MASK(gt)); + GEM_BUG_ON(vdbox_mask != VDBOX_MASK(gt)); + + for (i = 0; i < I915_MAX_VECS; i++) { + if (!HAS_ENGINE(gt, _VECS(i))) { + vebox_mask &= ~BIT(i); + continue; + } + + if (!(BIT(i) & vebox_mask)) { + info->engine_mask &= ~BIT(_VECS(i)); + drm_dbg(&i915->drm, "vecs%u fused off\n", i); + } + } + drm_dbg(&i915->drm, "vebox enable: %04x, instances: %04lx\n", + vebox_mask, VEBOX_MASK(gt)); + GEM_BUG_ON(vebox_mask != VEBOX_MASK(gt)); + + return info->engine_mask; +} + /** * intel_engines_init_mmio() - allocate and prepare the Engine Command Streamers * @gt: pointer to struct intel_gt @@ -459,8 +533,7 @@ void intel_engines_free(struct intel_gt *gt) int intel_engines_init_mmio(struct intel_gt *gt) { struct drm_i915_private *i915 = gt->i915; - struct intel_device_info *device_info = mkwrite_device_info(i915); - const unsigned int engine_mask = INTEL_INFO(i915)->engine_mask; + const unsigned int engine_mask = init_engine_mask(gt); unsigned int mask = 0; unsigned int i; int err; @@ -473,7 +546,7 @@ int intel_engines_init_mmio(struct intel_gt *gt) return -ENODEV; for (i = 0; i < ARRAY_SIZE(intel_engines); i++) { - if (!HAS_ENGINE(i915, i)) + if (!HAS_ENGINE(gt, i)) continue; err = intel_engine_setup(gt, i); @@ -489,14 +562,16 @@ int intel_engines_init_mmio(struct intel_gt *gt) * engines. */ if (drm_WARN_ON(&i915->drm, mask != engine_mask)) - device_info->engine_mask = mask; + gt->info.engine_mask = mask; - RUNTIME_INFO(i915)->num_engines = hweight32(mask); + gt->info.num_engines = hweight32(mask); intel_gt_check_and_clear_faults(gt); intel_setup_engine_capabilities(gt); + intel_uncore_prune_engine_fw_domains(gt->uncore, gt); + return 0; cleanup: @@ -634,7 +709,7 @@ static int engine_setup_common(struct intel_engine_cs *engine) /* Use the whole device by default */ engine->sseu = - intel_sseu_from_device_info(&RUNTIME_INFO(engine->i915)->sseu); + intel_sseu_from_device_info(&engine->gt->info.sseu); intel_engine_init_workarounds(engine); intel_engine_init_whitelist(engine); @@ -661,7 +736,6 @@ static int measure_breadcrumb_dw(struct intel_context *ce) if (!frame) return -ENOMEM; - frame->rq.i915 = engine->i915; frame->rq.engine = engine; frame->rq.context = ce; rcu_assign_pointer(frame->rq.timeline, ce->timeline); @@ -1001,7 +1075,7 @@ void intel_engine_get_instdone(const struct intel_engine_cs *engine, struct intel_instdone *instdone) { struct drm_i915_private *i915 = engine->i915; - const struct sseu_dev_info *sseu = &RUNTIME_INFO(i915)->sseu; + const struct sseu_dev_info *sseu = &engine->gt->info.sseu; struct intel_uncore *uncore = engine->uncore; u32 mmio_base = engine->mmio_base; int slice; @@ -1095,19 +1169,21 @@ void intel_engine_flush_submission(struct intel_engine_cs *engine) { struct tasklet_struct *t = &engine->execlists.tasklet; - if (__tasklet_is_scheduled(t)) { - local_bh_disable(); - if (tasklet_trylock(t)) { - /* Must wait for any GPU reset in progress. */ - if (__tasklet_is_enabled(t)) - t->func(t->data); - tasklet_unlock(t); - } - local_bh_enable(); - } + if (!t->func) + return; - /* Otherwise flush the tasklet if it was running on another cpu */ - tasklet_unlock_wait(t); + /* Synchronise and wait for the tasklet on another CPU */ + tasklet_kill(t); + + /* Having cancelled the tasklet, ensure that is run */ + local_bh_disable(); + if (tasklet_trylock(t)) { + /* Must wait for any GPU reset in progress. */ + if (__tasklet_is_enabled(t)) + t->func(t->data); + tasklet_unlock(t); + } + local_bh_enable(); } /** @@ -1194,8 +1270,7 @@ bool intel_engine_can_store_dword(struct intel_engine_cs *engine) } } -static int print_sched_attr(struct drm_i915_private *i915, - const struct i915_sched_attr *attr, +static int print_sched_attr(const struct i915_sched_attr *attr, char *buf, int x, int len) { if (attr->priority == I915_PRIORITY_INVALID) @@ -1215,7 +1290,7 @@ static void print_request(struct drm_printer *m, char buf[80] = ""; int x = 0; - x = print_sched_attr(rq->i915, &rq->sched.attr, buf, x, sizeof(buf)); + x = print_sched_attr(&rq->sched.attr, buf, x, sizeof(buf)); drm_printf(m, "%s %llx:%llx%s%s %s @ %dms: %s\n", prefix, @@ -1423,9 +1498,11 @@ static void intel_engine_print_registers(struct intel_engine_cs *engine, int len; len = scnprintf(hdr, sizeof(hdr), - "\t\tActive[%d]: ccid:%08x, ", + "\t\tActive[%d]: ccid:%08x%s%s, ", (int)(port - execlists->active), - rq->context->lrc.ccid); + rq->context->lrc.ccid, + intel_context_is_closed(rq->context) ? "!" : "", + intel_context_is_banned(rq->context) ? "*" : ""); len += print_ring(hdr + len, sizeof(hdr) - len, rq); scnprintf(hdr + len, sizeof(hdr) - len, "rq: "); print_request(m, rq, hdr); @@ -1435,9 +1512,11 @@ static void intel_engine_print_registers(struct intel_engine_cs *engine, int len; len = scnprintf(hdr, sizeof(hdr), - "\t\tPending[%d]: ccid:%08x, ", + "\t\tPending[%d]: ccid:%08x%s%s, ", (int)(port - execlists->pending), - rq->context->lrc.ccid); + rq->context->lrc.ccid, + intel_context_is_closed(rq->context) ? "!" : "", + intel_context_is_banned(rq->context) ? "*" : ""); len += print_ring(hdr + len, sizeof(hdr) - len, rq); scnprintf(hdr + len, sizeof(hdr) - len, "rq: "); print_request(m, rq, hdr); @@ -1506,6 +1585,7 @@ void intel_engine_dump(struct intel_engine_cs *engine, struct i915_request *rq; intel_wakeref_t wakeref; unsigned long flags; + ktime_t dummy; if (header) { va_list ap; @@ -1523,6 +1603,12 @@ void intel_engine_dump(struct intel_engine_cs *engine, yesno(!llist_empty(&engine->barrier_tasks))); drm_printf(m, "\tLatency: %luus\n", ewma__engine_latency_read(&engine->latency)); + if (intel_engine_supports_stats(engine)) + drm_printf(m, "\tRuntime: %llums\n", + ktime_to_ms(intel_engine_get_busy_time(engine, + &dummy))); + drm_printf(m, "\tForcewake: %x domains, %d active\n", + engine->fw_domain, atomic_read(&engine->fw_active)); rcu_read_lock(); rq = READ_ONCE(engine->heartbeat.systole); @@ -1589,7 +1675,8 @@ void intel_engine_dump(struct intel_engine_cs *engine, intel_engine_print_breadcrumbs(engine, m); } -static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine) +static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine, + ktime_t *now) { ktime_t total = engine->stats.total; @@ -1597,9 +1684,9 @@ static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine) * If the engine is executing something at the moment * add it to the total. */ + *now = ktime_get(); if (atomic_read(&engine->stats.active)) - total = ktime_add(total, - ktime_sub(ktime_get(), engine->stats.start)); + total = ktime_add(total, ktime_sub(*now, engine->stats.start)); return total; } @@ -1607,17 +1694,18 @@ static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine) /** * intel_engine_get_busy_time() - Return current accumulated engine busyness * @engine: engine to report on + * @now: monotonic timestamp of sampling * * Returns accumulated time @engine was busy since engine stats were enabled. */ -ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine) +ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine, ktime_t *now) { unsigned int seq; ktime_t total; do { seq = read_seqbegin(&engine->stats.lock); - total = __intel_engine_get_busy_time(engine); + total = __intel_engine_get_busy_time(engine, now); } while (read_seqretry(&engine->stats.lock, seq)); return total; |