drm/i915/gen8+: Add RC6 CTX corruption WA

In some circumstances the RC6 context can get corrupted. We can detect
this and take the required action, that is disable RC6 and runtime PM.
The HW recovers from the corrupted state after a system suspend/resume
cycle, so detect the recovery and re-enable RC6 and runtime PM.

v2: rebase (Mika)
v3:
- Move intel_suspend_gt_powersave() to the end of the GEM suspend
  sequence.
- Add commit message.
v4:
- Rebased on intel_uncore_forcewake_put(i915->uncore, ...) API
  change.
v5:
- Rebased on latest upstream gt_pm refactoring.
v6:
- s/i915_rc6_/intel_rc6_/
- Don't return a value from i915_rc6_ctx_wa_check().
v7:
- Rebased on latest gt rc6 refactoring.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
[airlied: pull this later version of this patch into drm-next
to make resolving the conflict mess easier.]
Signed-off-by: Dave Airlie <airlied@redhat.com>

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

@@ -61,6 +61,9 @@ static int __gt_unpark(struct intel_wakeref *wf)
 	gt->awake = intel_display_power_get(i915, POWER_DOMAIN_GT_IRQ);
 	GEM_BUG_ON(!gt->awake);
 
+	if (NEEDS_RC6_CTX_CORRUPTION_WA(i915))
+		intel_uncore_forcewake_get(&i915->uncore, FORCEWAKE_ALL);
+
 	intel_rps_unpark(&gt->rps);
 	i915_pmu_gt_unparked(i915);
 
@@ -86,6 +89,11 @@ static int __gt_park(struct intel_wakeref *wf)
 	/* Everything switched off, flush any residual interrupt just in case */
 	intel_synchronize_irq(i915);
 
+	if (NEEDS_RC6_CTX_CORRUPTION_WA(i915)) {
+		intel_rc6_ctx_wa_check(&i915->gt.rc6);
+		intel_uncore_forcewake_put(&i915->uncore, FORCEWAKE_ALL);
+	}
+
 	GEM_BUG_ON(!wakeref);
 	intel_display_power_put(i915, POWER_DOMAIN_GT_IRQ, wakeref);
 

drivers/gpu/drm/i915/gt/intel_rc6.c

@@ -486,6 +486,66 @@ static void rpm_put(struct intel_rc6 *rc6)
 	rc6->wakeref = false;
 }
 
+static bool intel_rc6_ctx_corrupted(struct intel_rc6 *rc6)
+{
+	return !intel_uncore_read(rc6_to_uncore(rc6), GEN8_RC6_CTX_INFO);
+}
+
+static void intel_rc6_ctx_wa_init(struct intel_rc6 *rc6)
+{
+	struct drm_i915_private *i915 = rc6_to_i915(rc6);
+
+	if (!NEEDS_RC6_CTX_CORRUPTION_WA(i915))
+		return;
+
+	if (intel_rc6_ctx_corrupted(rc6)) {
+		DRM_INFO("RC6 context corrupted, disabling runtime power management\n");
+		rc6->ctx_corrupted = true;
+	}
+}
+
+/**
+ * intel_rc6_ctx_wa_resume - system resume sequence for the RC6 CTX WA
+ * @rc6: rc6 state
+ *
+ * Perform any steps needed to re-init the RC6 CTX WA after system resume.
+ */
+void intel_rc6_ctx_wa_resume(struct intel_rc6 *rc6)
+{
+	if (rc6->ctx_corrupted && !intel_rc6_ctx_corrupted(rc6)) {
+		DRM_INFO("RC6 context restored, re-enabling runtime power management\n");
+		rc6->ctx_corrupted = false;
+	}
+}
+
+/**
+ * intel_rc6_ctx_wa_check - check for a new RC6 CTX corruption
+ * @rc6: rc6 state
+ *
+ * Check if an RC6 CTX corruption has happened since the last check and if so
+ * disable RC6 and runtime power management.
+*/
+void intel_rc6_ctx_wa_check(struct intel_rc6 *rc6)
+{
+	struct drm_i915_private *i915 = rc6_to_i915(rc6);
+
+	if (!NEEDS_RC6_CTX_CORRUPTION_WA(i915))
+		return;
+
+	if (rc6->ctx_corrupted)
+		return;
+
+	if (!intel_rc6_ctx_corrupted(rc6))
+		return;
+
+	DRM_NOTE("RC6 context corruption, disabling runtime power management\n");
+
+	intel_rc6_disable(rc6);
+	rc6->ctx_corrupted = true;
+
+	return;
+}
+
 static void __intel_rc6_disable(struct intel_rc6 *rc6)
 {
 	struct drm_i915_private *i915 = rc6_to_i915(rc6);
@@ -510,6 +570,8 @@ void intel_rc6_init(struct intel_rc6 *rc6)
 	if (!rc6_supported(rc6))
 		return;
 
+	intel_rc6_ctx_wa_init(rc6);
+
 	if (IS_CHERRYVIEW(i915))
 		err = chv_rc6_init(rc6);
 	else if (IS_VALLEYVIEW(i915))
@@ -544,6 +606,9 @@ void intel_rc6_enable(struct intel_rc6 *rc6)
 
 	GEM_BUG_ON(rc6->enabled);
 
+	if (rc6->ctx_corrupted)
+		return;
+
 	intel_uncore_forcewake_get(uncore, FORCEWAKE_ALL);
 
 	if (IS_CHERRYVIEW(i915))

drivers/gpu/drm/i915/gt/intel_rc6.h

@@ -22,4 +22,7 @@ void intel_rc6_disable(struct intel_rc6 *rc6);
 u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, i915_reg_t reg);
 u64 intel_rc6_residency_us(struct intel_rc6 *rc6, i915_reg_t reg);
 
+void intel_rc6_ctx_wa_check(struct intel_rc6 *rc6);
+void intel_rc6_ctx_wa_resume(struct intel_rc6 *rc6);
+
 #endif /* INTEL_RC6_H */

drivers/gpu/drm/i915/gt/intel_rc6_types.h

@@ -23,6 +23,7 @@ struct intel_rc6 {
 	bool supported : 1;
 	bool enabled : 1;
 	bool wakeref : 1;
+	bool ctx_corrupted : 1;
 };
 
 #endif /* INTEL_RC6_TYPES_H */

drivers/gpu/drm/i915/i915_drv.c

@@ -63,6 +63,7 @@
 #include "gem/i915_gem_ioctls.h"
 #include "gt/intel_gt.h"
 #include "gt/intel_gt_pm.h"
+#include "gt/intel_rc6.h"
 
 #include "i915_debugfs.h"
 #include "i915_drv.h"
@@ -1819,6 +1820,8 @@ static int i915_drm_resume(struct drm_device *dev)
 
 	disable_rpm_wakeref_asserts(&dev_priv->runtime_pm);
 
+	intel_rc6_ctx_wa_resume(&dev_priv->gt.rc6);
+
 	intel_gt_sanitize(&dev_priv->gt, true);
 
 	ret = i915_ggtt_enable_hw(dev_priv);

drivers/gpu/drm/i915/i915_drv.h

@@ -1649,10 +1649,12 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915,
 /* Early gen2 have a totally busted CS tlb and require pinned batches. */
 #define HAS_BROKEN_CS_TLB(dev_priv)	(IS_I830(dev_priv) || IS_I845G(dev_priv))
 
+#define NEEDS_RC6_CTX_CORRUPTION_WA(dev_priv)	\
+	(IS_BROADWELL(dev_priv) || IS_GEN(dev_priv, 9))
+
 /* WaRsDisableCoarsePowerGating:skl,cnl */
 #define NEEDS_WaRsDisableCoarsePowerGating(dev_priv) \
-	(IS_CANNONLAKE(dev_priv) || \
-	 IS_SKL_GT3(dev_priv) || IS_SKL_GT4(dev_priv))
+	(IS_CANNONLAKE(dev_priv) || IS_GEN(dev_priv, 9))
 
 #define HAS_GMBUS_IRQ(dev_priv) (INTEL_GEN(dev_priv) >= 4)
 #define HAS_GMBUS_BURST_READ(dev_priv) (INTEL_GEN(dev_priv) >= 10 || \

drivers/gpu/drm/i915/i915_reg.h

@@ -474,6 +474,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define   ECOCHK_PPGTT_WT_HSW		(0x2 << 3)
 #define   ECOCHK_PPGTT_WB_HSW		(0x3 << 3)
 
+#define GEN8_RC6_CTX_INFO		_MMIO(0x8504)
+
 #define GAC_ECO_BITS			_MMIO(0x14090)
 #define   ECOBITS_SNB_BIT		(1 << 13)
 #define   ECOBITS_PPGTT_CACHE64B	(3 << 8)