Merge tag 'drm-intel-gt-next-2024-12-18' of https://gitlab.freedesktop.org/drm/i915/kernel into drm-next

Driver Changes:

- More accurate engine busyness metrics with GuC submission (Umesh)
- Ensure partial BO segment offset never exceeds allowed max (Krzysztof)
- Flush GuC CT receive tasklet during reset preparation (Zhanjun)

- Code cleanups and refactoring (David, Lucas)
- Debugging improvements (Jesus)
- Selftest improvements (Sk)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/Z2KadNXgumx1aQMP@jlahtine-mobl.ger.corp.intel.com

Author: airlied

drivers/gpu/drm/i915/gt/intel_engine.h

@@ -126,9 +126,6 @@ execlists_active(const struct intel_engine_execlists *execlists)
 	return active;
 }
 
-struct i915_request *
-execlists_unwind_incomplete_requests(struct intel_engine_execlists *execlists);
-
 static inline u32
 intel_read_status_page(const struct intel_engine_cs *engine, int reg)
 {

drivers/gpu/drm/i915/gt/intel_engine_types.h

@@ -343,6 +343,11 @@ struct intel_engine_guc_stats {
 	 * @start_gt_clk: GT clock time of last idle to active transition.
 	 */
 	u64 start_gt_clk;
+
+	/**
+	 * @total: The last value of total returned
+	 */
+	u64 total;
 };
 
 union intel_engine_tlb_inv_reg {

drivers/gpu/drm/i915/gt/intel_execlists_submission.c

@@ -405,15 +405,6 @@ __unwind_incomplete_requests(struct intel_engine_cs *engine)
 	return active;
 }
 
-struct i915_request *
-execlists_unwind_incomplete_requests(struct intel_engine_execlists *execlists)
-{
-	struct intel_engine_cs *engine =
-		container_of(execlists, typeof(*engine), execlists);
-
-	return __unwind_incomplete_requests(engine);
-}
-
 static void
 execlists_context_status_change(struct i915_request *rq, unsigned long status)
 {

drivers/gpu/drm/i915/gt/selftest_rps.c

@@ -1125,6 +1125,7 @@ static u64 measure_power(struct intel_rps *rps, int *freq)
 static u64 measure_power_at(struct intel_rps *rps, int *freq)
 {
 	*freq = rps_set_check(rps, *freq);
+	msleep(100);
 	return measure_power(rps, freq);
 }
 

drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c

@@ -1243,6 +1243,21 @@ static void __get_engine_usage_record(struct intel_engine_cs *engine,
 	} while (++i < 6);
 }
 
+static void __set_engine_usage_record(struct intel_engine_cs *engine,
+				      u32 last_in, u32 id, u32 total)
+{
+	struct iosys_map rec_map = intel_guc_engine_usage_record_map(engine);
+
+#define record_write(map_, field_, val_) \
+	iosys_map_wr_field(map_, 0, struct guc_engine_usage_record, field_, val_)
+
+	record_write(&rec_map, last_switch_in_stamp, last_in);
+	record_write(&rec_map, current_context_index, id);
+	record_write(&rec_map, total_runtime, total);
+
+#undef record_write
+}
+
 static void guc_update_engine_gt_clks(struct intel_engine_cs *engine)
 {
 	struct intel_engine_guc_stats *stats = &engine->stats.guc;
@@ -1363,9 +1378,12 @@ static ktime_t guc_engine_busyness(struct intel_engine_cs *engine, ktime_t *now)
 		total += intel_gt_clock_interval_to_ns(gt, clk);
 	}
 
+	if (total > stats->total)
+		stats->total = total;
+
 	spin_unlock_irqrestore(&guc->timestamp.lock, flags);
 
-	return ns_to_ktime(total);
+	return ns_to_ktime(stats->total);
 }
 
 static void guc_enable_busyness_worker(struct intel_guc *guc)
@@ -1431,8 +1449,21 @@ static void __reset_guc_busyness_stats(struct intel_guc *guc)
 
 	guc_update_pm_timestamp(guc, &unused);
 	for_each_engine(engine, gt, id) {
+		struct intel_engine_guc_stats *stats = &engine->stats.guc;
+
 		guc_update_engine_gt_clks(engine);
-		engine->stats.guc.prev_total = 0;
+
+		/*
+		 * If resetting a running context, accumulate the active
+		 * time as well since there will be no context switch.
+		 */
+		if (stats->running) {
+			u64 clk = guc->timestamp.gt_stamp - stats->start_gt_clk;
+
+			stats->total_gt_clks += clk;
+		}
+		stats->prev_total = 0;
+		stats->running = 0;
 	}
 
 	spin_unlock_irqrestore(&guc->timestamp.lock, flags);
@@ -1543,13 +1574,19 @@ static void guc_timestamp_ping(struct work_struct *wrk)
 
 static int guc_action_enable_usage_stats(struct intel_guc *guc)
 {
+	struct intel_gt *gt = guc_to_gt(guc);
+	struct intel_engine_cs *engine;
+	enum intel_engine_id id;
 	u32 offset = intel_guc_engine_usage_offset(guc);
 	u32 action[] = {
 		INTEL_GUC_ACTION_SET_ENG_UTIL_BUFF,
 		offset,
 		0,
 	};
 
+	for_each_engine(engine, gt, id)
+		__set_engine_usage_record(engine, 0, 0xffffffff, 0);
+
 	return intel_guc_send(guc, action, ARRAY_SIZE(action));
 }
 
@@ -1688,6 +1725,10 @@ void intel_guc_submission_reset_prepare(struct intel_guc *guc)
 	spin_lock_irq(guc_to_gt(guc)->irq_lock);
 	spin_unlock_irq(guc_to_gt(guc)->irq_lock);
 
+	/* Flush tasklet */
+	tasklet_disable(&guc->ct.receive_tasklet);
+	tasklet_enable(&guc->ct.receive_tasklet);
+
 	guc_flush_submissions(guc);
 	guc_flush_destroyed_contexts(guc);
 	flush_work(&guc->ct.requests.worker);
@@ -2005,6 +2046,8 @@ void intel_guc_submission_cancel_requests(struct intel_guc *guc)
 
 void intel_guc_submission_reset_finish(struct intel_guc *guc)
 {
+	int outstanding;
+
 	/* Reset called during driver load or during wedge? */
 	if (unlikely(!guc_submission_initialized(guc) ||
 		     !intel_guc_is_fw_running(guc) ||
@@ -2018,8 +2061,10 @@ void intel_guc_submission_reset_finish(struct intel_guc *guc)
 	 * see in CI if this happens frequently / a precursor to taking down the
 	 * machine.
 	 */
-	if (atomic_read(&guc->outstanding_submission_g2h))
-		guc_err(guc, "Unexpected outstanding GuC to Host in reset finish\n");
+	outstanding = atomic_read(&guc->outstanding_submission_g2h);
+	if (outstanding)
+		guc_err(guc, "Unexpected outstanding GuC to Host response(s) in reset finish: %d\n",
+			outstanding);
 	atomic_set(&guc->outstanding_submission_g2h, 0);
 
 	intel_guc_global_policies_update(guc);

drivers/gpu/drm/i915/i915_mm.c

@@ -143,8 +143,8 @@ int remap_io_sg(struct vm_area_struct *vma,
 	/* We rely on prevalidation of the io-mapping to skip track_pfn(). */
 	GEM_BUG_ON((vma->vm_flags & EXPECTED_FLAGS) != EXPECTED_FLAGS);
 
-	while (offset >= sg_dma_len(r.sgt.sgp) >> PAGE_SHIFT) {
-		offset -= sg_dma_len(r.sgt.sgp) >> PAGE_SHIFT;
+	while (offset >= r.sgt.max >> PAGE_SHIFT) {
+		offset -= r.sgt.max >> PAGE_SHIFT;
 		r.sgt = __sgt_iter(__sg_next(r.sgt.sgp), use_dma(iobase));
 		if (!r.sgt.sgp)
 			return -EINVAL;

drivers/gpu/drm/i915/i915_pmu.c

@@ -302,7 +302,7 @@ void i915_pmu_gt_parked(struct intel_gt *gt)
 {
 	struct i915_pmu *pmu = &gt->i915->pmu;
 
-	if (!pmu->base.event_init)
+	if (!pmu->registered)
 		return;
 
 	spin_lock_irq(&pmu->lock);
@@ -324,7 +324,7 @@ void i915_pmu_gt_unparked(struct intel_gt *gt)
 {
 	struct i915_pmu *pmu = &gt->i915->pmu;
 
-	if (!pmu->base.event_init)
+	if (!pmu->registered)
 		return;
 
 	spin_lock_irq(&pmu->lock);
@@ -626,7 +626,7 @@ static int i915_pmu_event_init(struct perf_event *event)
 	struct drm_i915_private *i915 = pmu_to_i915(pmu);
 	int ret;
 
-	if (pmu->closed)
+	if (!pmu->registered)
 		return -ENODEV;
 
 	if (event->attr.type != event->pmu->type)
@@ -724,7 +724,7 @@ static void i915_pmu_event_read(struct perf_event *event)
 	struct hw_perf_event *hwc = &event->hw;
 	u64 prev, new;
 
-	if (pmu->closed) {
+	if (!pmu->registered) {
 		event->hw.state = PERF_HES_STOPPED;
 		return;
 	}
@@ -850,7 +850,7 @@ static void i915_pmu_event_start(struct perf_event *event, int flags)
 {
 	struct i915_pmu *pmu = event_to_pmu(event);
 
-	if (pmu->closed)
+	if (!pmu->registered)
 		return;
 
 	i915_pmu_enable(event);
@@ -861,7 +861,7 @@ static void i915_pmu_event_stop(struct perf_event *event, int flags)
 {
 	struct i915_pmu *pmu = event_to_pmu(event);
 
-	if (pmu->closed)
+	if (!pmu->registered)
 		goto out;
 
 	if (flags & PERF_EF_UPDATE)
@@ -877,7 +877,7 @@ static int i915_pmu_event_add(struct perf_event *event, int flags)
 {
 	struct i915_pmu *pmu = event_to_pmu(event);
 
-	if (pmu->closed)
+	if (!pmu->registered)
 		return -ENODEV;
 
 	if (flags & PERF_EF_START)
@@ -1177,8 +1177,6 @@ static int i915_pmu_cpu_online(unsigned int cpu, struct hlist_node *node)
 {
 	struct i915_pmu *pmu = hlist_entry_safe(node, typeof(*pmu), cpuhp.node);
 
-	GEM_BUG_ON(!pmu->base.event_init);
-
 	/* Select the first online CPU as a designated reader. */
 	if (cpumask_empty(&i915_pmu_cpumask))
 		cpumask_set_cpu(cpu, &i915_pmu_cpumask);
@@ -1191,13 +1189,11 @@ static int i915_pmu_cpu_offline(unsigned int cpu, struct hlist_node *node)
 	struct i915_pmu *pmu = hlist_entry_safe(node, typeof(*pmu), cpuhp.node);
 	unsigned int target = i915_pmu_target_cpu;
 
-	GEM_BUG_ON(!pmu->base.event_init);
-
 	/*
 	 * Unregistering an instance generates a CPU offline event which we must
 	 * ignore to avoid incorrectly modifying the shared i915_pmu_cpumask.
 	 */
-	if (pmu->closed)
+	if (!pmu->registered)
 		return 0;
 
 	if (cpumask_test_and_clear_cpu(cpu, &i915_pmu_cpumask)) {
@@ -1218,7 +1214,7 @@ static int i915_pmu_cpu_offline(unsigned int cpu, struct hlist_node *node)
 	return 0;
 }
 
-static enum cpuhp_state cpuhp_slot = CPUHP_INVALID;
+static enum cpuhp_state cpuhp_state = CPUHP_INVALID;
 
 int i915_pmu_init(void)
 {
@@ -1232,28 +1228,28 @@ int i915_pmu_init(void)
 		pr_notice("Failed to setup cpuhp state for i915 PMU! (%d)\n",
 			  ret);
 	else
-		cpuhp_slot = ret;
+		cpuhp_state = ret;
 
 	return 0;
 }
 
 void i915_pmu_exit(void)
 {
-	if (cpuhp_slot != CPUHP_INVALID)
-		cpuhp_remove_multi_state(cpuhp_slot);
+	if (cpuhp_state != CPUHP_INVALID)
+		cpuhp_remove_multi_state(cpuhp_state);
 }
 
 static int i915_pmu_register_cpuhp_state(struct i915_pmu *pmu)
 {
-	if (cpuhp_slot == CPUHP_INVALID)
+	if (cpuhp_state == CPUHP_INVALID)
 		return -EINVAL;
 
-	return cpuhp_state_add_instance(cpuhp_slot, &pmu->cpuhp.node);
+	return cpuhp_state_add_instance(cpuhp_state, &pmu->cpuhp.node);
 }
 
 static void i915_pmu_unregister_cpuhp_state(struct i915_pmu *pmu)
 {
-	cpuhp_state_remove_instance(cpuhp_slot, &pmu->cpuhp.node);
+	cpuhp_state_remove_instance(cpuhp_state, &pmu->cpuhp.node);
 }
 
 void i915_pmu_register(struct drm_i915_private *i915)
@@ -1265,7 +1261,6 @@ void i915_pmu_register(struct drm_i915_private *i915)
 		&i915_pmu_cpumask_attr_group,
 		NULL
 	};
-
 	int ret = -ENOMEM;
 
 	spin_lock_init(&pmu->lock);
@@ -1316,14 +1311,15 @@ void i915_pmu_register(struct drm_i915_private *i915)
 	if (ret)
 		goto err_unreg;
 
+	pmu->registered = true;
+
 	return;
 
 err_unreg:
 	perf_pmu_unregister(&pmu->base);
 err_groups:
 	kfree(pmu->base.attr_groups);
 err_attr:
-	pmu->base.event_init = NULL;
 	free_event_attributes(pmu);
 err_name:
 	if (IS_DGFX(i915))
@@ -1336,23 +1332,17 @@ void i915_pmu_unregister(struct drm_i915_private *i915)
 {
 	struct i915_pmu *pmu = &i915->pmu;
 
-	if (!pmu->base.event_init)
+	if (!pmu->registered)
 		return;
 
-	/*
-	 * "Disconnect" the PMU callbacks - since all are atomic synchronize_rcu
-	 * ensures all currently executing ones will have exited before we
-	 * proceed with unregistration.
-	 */
-	pmu->closed = true;
-	synchronize_rcu();
+	/* Disconnect the PMU callbacks */
+	pmu->registered = false;
 
 	hrtimer_cancel(&pmu->timer);
 
 	i915_pmu_unregister_cpuhp_state(pmu);
 
 	perf_pmu_unregister(&pmu->base);
-	pmu->base.event_init = NULL;
 	kfree(pmu->base.attr_groups);
 	if (IS_DGFX(i915))
 		kfree(pmu->name);

drivers/gpu/drm/i915/i915_pmu.h

@@ -68,9 +68,9 @@ struct i915_pmu {
 	 */
 	struct pmu base;
 	/**
-	 * @closed: i915 is unregistering.
+	 * @registered: PMU is registered and not in the unregistering process.
 	 */
-	bool closed;
+	bool registered;
 	/**
 	 * @name: Name as registered with perf core.
 	 */