Merge tag 'drm-intel-gt-next-2023-05-24' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

UAPI Changes:

- New getparam for querying PXP support and load status

Cross-subsystem Changes:

- GSC/MEI proxy driver

Driver Changes:

Fixes/improvements/new stuff:

- Avoid clearing pre-allocated framebuffers with the TTM backend (Nirmoy Das)
- Implement framebuffer mmap support (Nirmoy Das)
- Disable sampler indirect state in bindless heap (Lionel Landwerlin)
- Avoid out-of-bounds access when loading HuC (Lucas De Marchi)
- Actually return an error if GuC version range check fails (John Harrison)
- Get mutex and rpm ref just once in hwm_power_max_write (Ashutosh Dixit)
- Disable PL1 power limit when loading GuC firmware (Ashutosh Dixit)
- Block in hwmon while waiting for GuC reset to complete (Ashutosh Dixit)
- Provide sysfs for SLPC efficient freq (Vinay Belgaumkar)
- Add support for total context runtime for GuC back-end (Umesh Nerlige Ramappa)
- Enable fdinfo for GuC backends (Umesh Nerlige Ramappa)
- Don't capture Gen8 regs on Xe devices (John Harrison)
- Fix error capture for virtual engines (John Harrison)
- Track patch level versions on reduced version firmware files (John Harrison)
- Decode another GuC load failure case (John Harrison)
- GuC loading and firmware table handling fixes (John Harrison)
- Fix confused register capture list creation (John Harrison)
- Dump error capture to kernel log (John Harrison)
- Dump error capture to dmesg on CTB error (John Harrison)
- Disable rps_boost debugfs when SLPC is used (Vinay Belgaumkar)

Future platform enablement:

- Disable stolen memory backed FB for A0 [mtl] (Nirmoy Das)
- Various refactors for multi-tile enablement (Andi Shyti, Tejas Upadhyay)
- Extend Wa_22011802037 to MTL A-step (Madhumitha Tolakanahalli Pradeep)
- WA to clear RDOP clock gating [mtl] (Haridhar Kalvala)
- Set has_llc=0 [mtl] (Fei Yang)
- Define MOCS and PAT tables for MTL (Madhumitha Tolakanahalli Pradeep)
- Add PTE encode function [mtl] (Fei Yang)
- fix mocs selftest [mtl] (Fei Yang)
- Workaround coherency issue for Media [mtl] (Fei Yang)
- Add workaround 14018778641 [mtl] (Tejas Upadhyay)
- Implement Wa_14019141245 [mtl] (Radhakrishna Sripada)
- Fix the wa number for Wa_22016670082 [mtl] (Radhakrishna Sripada)
- Use correct huge page manager for MTL (Jonathan Cavitt)
- GSC/MEI support for Meteorlake (Alexander Usyskin, Daniele Ceraolo Spurio)
- Define GuC firmware version for MTL (John Harrison)
- Drop FLAT CCS check [mtl] (Pallavi Mishra)
- Add MTL for remapping CCS FBs [mtl] (Clint Taylor)
- Meteorlake PXP enablement (Alan Previn)
- Do not enable render power-gating on MTL (Andrzej Hajda)
- Add MTL performance tuning changes (Radhakrishna Sripada)
- Extend Wa_16014892111 to MTL A-step (Radhakrishna Sripada)
- PMU multi-tile support (Tvrtko Ursulin)
- End support for set caching ioctl [mtl] (Fei Yang)

Driver refactors:

- Use i915 instead of dev_priv insied the file_priv structure (Andi Shyti)
- Use proper parameter naming in for_each_engine() (Andi Shyti)
- Use gt_err for GT info (Tejas Upadhyay)
- Consolidate duplicated capture list code (John Harrison)
- Capture list naming clean up (John Harrison)
- Use kernel-doc -Werror when CONFIG_DRM_I915_WERROR=y (Jani Nikula)
- Preparation for using PAT index (Fei Yang)
- Use pat_index instead of cache_level (Fei Yang)

Miscellaneous:

- Fix memory leaks in i915 selftests (Cong Liu)
- Record GT error for gt failure (Tejas Upadhyay)
- Migrate platform-dependent mock hugepage selftests to live (Jonathan Cavitt)
- Update the SLPC selftest (Vinay Belgaumkar)
- Throw out set() wrapper (Jani Nikula)
- Large driver kernel doc cleanup (Jani Nikula)
- Fix probe injection CI failures after recent change (John Harrison)
- Make unexpected firmware versions an error in debug builds (John Harrison)
- Silence UBSAN uninitialized bool variable warning (Ashutosh Dixit)
- Fix memory leaks in function live_nop_switch (Cong Liu)

Merges:

- Merge drm/drm-next into drm-intel-gt-next (Joonas Lahtinen)

Signed-off-by: Dave Airlie <airlied@redhat.com>

# Conflicts:
#	drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZG5SxCWRSkZhTDtY@tursulin-desk

1 files changed, 65 insertions, 10 deletions

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
index 88e881b100cf..a0e3ef1c65d2 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
@@ -1402,13 +1402,34 @@ static void __update_guc_busyness_stats(struct intel_guc *guc)
 	spin_unlock_irqrestore(&guc->timestamp.lock, flags);
 }
 
+static void __guc_context_update_stats(struct intel_context *ce)
+{
+	struct intel_guc *guc = ce_to_guc(ce);
+	unsigned long flags;
+
+	spin_lock_irqsave(&guc->timestamp.lock, flags);
+	lrc_update_runtime(ce);
+	spin_unlock_irqrestore(&guc->timestamp.lock, flags);
+}
+
+static void guc_context_update_stats(struct intel_context *ce)
+{
+	if (!intel_context_pin_if_active(ce))
+		return;
+
+	__guc_context_update_stats(ce);
+	intel_context_unpin(ce);
+}
+
 static void guc_timestamp_ping(struct work_struct *wrk)
 {
 	struct intel_guc *guc = container_of(wrk, typeof(*guc),
 					     timestamp.work.work);
 	struct intel_uc *uc = container_of(guc, typeof(*uc), guc);
 	struct intel_gt *gt = guc_to_gt(guc);
+	struct intel_context *ce;
 	intel_wakeref_t wakeref;
+	unsigned long index;
 	int srcu, ret;
 
 	/*
@@ -1424,6 +1445,10 @@ static void guc_timestamp_ping(struct work_struct *wrk)
 	with_intel_runtime_pm(&gt->i915->runtime_pm, wakeref)
 		__update_guc_busyness_stats(guc);
 
+	/* adjust context stats for overflow */
+	xa_for_each(&guc->context_lookup, index, ce)
+		guc_context_update_stats(ce);
+
 	intel_gt_reset_unlock(gt, srcu);
 
 	guc_enable_busyness_worker(guc);
@@ -1629,16 +1654,16 @@ static void guc_reset_state(struct intel_context *ce, u32 head, bool scrub)
 
 static void guc_engine_reset_prepare(struct intel_engine_cs *engine)
 {
-	if (!IS_GRAPHICS_VER(engine->i915, 11, 12))
-		return;
-
-	intel_engine_stop_cs(engine);
-
 	/*
 	 * Wa_22011802037: In addition to stopping the cs, we need
 	 * to wait for any pending mi force wakeups
 	 */
-	intel_engine_wait_for_pending_mi_fw(engine);
+	if (IS_MTL_GRAPHICS_STEP(engine->i915, M, STEP_A0, STEP_B0) ||
+	    (GRAPHICS_VER(engine->i915) >= 11 &&
+	     GRAPHICS_VER_FULL(engine->i915) < IP_VER(12, 70))) {
+		intel_engine_stop_cs(engine);
+		intel_engine_wait_for_pending_mi_fw(engine);
+	}
 }
 
 static void guc_reset_nop(struct intel_engine_cs *engine)
@@ -2774,6 +2799,7 @@ static void guc_context_unpin(struct intel_context *ce)
 {
 	struct intel_guc *guc = ce_to_guc(ce);
 
+	__guc_context_update_stats(ce);
 	unpin_guc_id(guc, ce);
 	lrc_unpin(ce);
 
@@ -3455,6 +3481,7 @@ static void remove_from_context(struct i915_request *rq)
 }
 
 static const struct intel_context_ops guc_context_ops = {
+	.flags = COPS_RUNTIME_CYCLES,
 	.alloc = guc_context_alloc,
 
 	.close = guc_context_close,
@@ -3473,6 +3500,8 @@ static const struct intel_context_ops guc_context_ops = {
 
 	.sched_disable = guc_context_sched_disable,
 
+	.update_stats = guc_context_update_stats,
+
 	.reset = lrc_reset,
 	.destroy = guc_context_destroy,
 
@@ -3728,6 +3757,7 @@ static int guc_virtual_context_alloc(struct intel_context *ce)
 }
 
 static const struct intel_context_ops virtual_guc_context_ops = {
+	.flags = COPS_RUNTIME_CYCLES,
 	.alloc = guc_virtual_context_alloc,
 
 	.close = guc_context_close,
@@ -3745,6 +3775,7 @@ static const struct intel_context_ops virtual_guc_context_ops = {
 	.exit = guc_virtual_context_exit,
 
 	.sched_disable = guc_context_sched_disable,
+	.update_stats = guc_context_update_stats,
 
 	.destroy = guc_context_destroy,
 
@@ -4697,13 +4728,37 @@ static void capture_error_state(struct intel_guc *guc,
 {
 	struct intel_gt *gt = guc_to_gt(guc);
 	struct drm_i915_private *i915 = gt->i915;
-	struct intel_engine_cs *engine = __context_to_physical_engine(ce);
 	intel_wakeref_t wakeref;
+	intel_engine_mask_t engine_mask;
+
+	if (intel_engine_is_virtual(ce->engine)) {
+		struct intel_engine_cs *e;
+		intel_engine_mask_t tmp, virtual_mask = ce->engine->mask;
+
+		engine_mask = 0;
+		for_each_engine_masked(e, ce->engine->gt, virtual_mask, tmp) {
+			bool match = intel_guc_capture_is_matching_engine(gt, ce, e);
+
+			if (match) {
+				intel_engine_set_hung_context(e, ce);
+				engine_mask |= e->mask;
+				atomic_inc(&i915->gpu_error.reset_engine_count[e->uabi_class]);
+			}
+		}
+
+		if (!engine_mask) {
+			guc_warn(guc, "No matching physical engine capture for virtual engine context 0x%04X / %s",
+				 ce->guc_id.id, ce->engine->name);
+			engine_mask = ~0U;
+		}
+	} else {
+		intel_engine_set_hung_context(ce->engine, ce);
+		engine_mask = ce->engine->mask;
+		atomic_inc(&i915->gpu_error.reset_engine_count[ce->engine->uabi_class]);
+	}
 
-	intel_engine_set_hung_context(engine, ce);
 	with_intel_runtime_pm(&i915->runtime_pm, wakeref)
-		i915_capture_error_state(gt, engine->mask, CORE_DUMP_FLAG_IS_GUC_CAPTURE);
-	atomic_inc(&i915->gpu_error.reset_engine_count[engine->uabi_class]);
+		i915_capture_error_state(gt, engine_mask, CORE_DUMP_FLAG_IS_GUC_CAPTURE);
 }
 
 static void guc_context_replay(struct intel_context *ce)