Merge tag 'drm-intel-next-fixes-2015-02-11' of git://anongit.freedesktop.org/drm-intel into drm-next

Here's a batch of i915 fixes for drm-next, with more cc: stable material
than fixes specific to drm-next.

* tag 'drm-intel-next-fixes-2015-02-11' of git://anongit.freedesktop.org/drm-intel:
  drm/i915: Clamp efficient frequency to valid range
  drm/i915: Really ignore long HPD pulses on eDP
  drm/i915: Correct the base value while updating LP_OUTPUT_HOLD in MIPI_PORT_CTRL
  drm/i915: Insert a command barrier on BLT/BSD cache flushes
  drm/i915: Drop vblank wait from intel_dp_link_down
  drm/i915: Squelch overzealous uncore reset WARN_ON
  drm/i915: Take runtime pm reference on hangcheck_info
  drm/i915: Correct the IOSF Dev_FN field for IOSF transfers
  drm/i915: Prevent use-after-free in invalidate_range_start callback

9 files changed, 94 insertions, 55 deletions

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 211d494..96e811f 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -1223,8 +1223,11 @@ out:
 static int i915_hangcheck_info(struct seq_file *m, void *unused)
 {
 	struct drm_info_node *node = m->private;
-	struct drm_i915_private *dev_priv = to_i915(node->minor->dev);
+	struct drm_device *dev = node->minor->dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct intel_engine_cs *ring;
+	u64 acthd[I915_NUM_RINGS];
+	u32 seqno[I915_NUM_RINGS];
 	int i;
 
 	if (!i915.enable_hangcheck) {
@@ -1232,6 +1235,15 @@ static int i915_hangcheck_info(struct seq_file *m, void *unused)
 		return 0;
 	}
 
+	intel_runtime_pm_get(dev_priv);
+
+	for_each_ring(ring, dev_priv, i) {
+		seqno[i] = ring->get_seqno(ring, false);
+		acthd[i] = intel_ring_get_active_head(ring);
+	}
+
+	intel_runtime_pm_put(dev_priv);
+
 	if (delayed_work_pending(&dev_priv->gpu_error.hangcheck_work)) {
 		seq_printf(m, "Hangcheck active, fires in %dms\n",
 			   jiffies_to_msecs(dev_priv->gpu_error.hangcheck_work.timer.expires -
@@ -1242,14 +1254,14 @@ static int i915_hangcheck_info(struct seq_file *m, void *unused)
 	for_each_ring(ring, dev_priv, i) {
 		seq_printf(m, "%s:\n", ring->name);
 		seq_printf(m, "\tseqno = %x [current %x]\n",
-			   ring->hangcheck.seqno, ring->get_seqno(ring, false));
-		seq_printf(m, "\taction = %d\n", ring->hangcheck.action);
-		seq_printf(m, "\tscore = %d\n", ring->hangcheck.score);
+			   ring->hangcheck.seqno, seqno[i]);
 		seq_printf(m, "\tACTHD = 0x%08llx [current 0x%08llx]\n",
 			   (long long)ring->hangcheck.acthd,
-			   (long long)intel_ring_get_active_head(ring));
+			   (long long)acthd[i]);
 		seq_printf(m, "\tmax ACTHD = 0x%08llx\n",
 			   (long long)ring->hangcheck.max_acthd);
+		seq_printf(m, "\tscore = %d\n", ring->hangcheck.score);
+		seq_printf(m, "\taction = %d\n", ring->hangcheck.action);
 	}
 
 	return 0;
diff --git a/drivers/gpu/drm/i915/i915_gem_userptr.c b/drivers/gpu/drm/i915/i915_gem_userptr.c
index d182058..1719078 100644
--- a/drivers/gpu/drm/i915/i915_gem_userptr.c
+++ b/drivers/gpu/drm/i915/i915_gem_userptr.c
@@ -113,7 +113,10 @@ restart:
 			continue;
 
 		obj = mo->obj;
-		drm_gem_object_reference(&obj->base);
+
+		if (!kref_get_unless_zero(&obj->base.refcount))
+			continue;
+
 		spin_unlock(&mn->lock);
 
 		cancel_userptr(obj);
@@ -149,7 +152,20 @@ static void i915_gem_userptr_mn_invalidate_range_start(struct mmu_notifier *_mn,
 			it = interval_tree_iter_first(&mn->objects, start, end);
 		if (it != NULL) {
 			obj = container_of(it, struct i915_mmu_object, it)->obj;
-			drm_gem_object_reference(&obj->base);
+
+			/* The mmu_object is released late when destroying the
+			 * GEM object so it is entirely possible to gain a
+			 * reference on an object in the process of being freed
+			 * since our serialisation is via the spinlock and not
+			 * the struct_mutex - and consequently use it after it
+			 * is freed and then double free it.
+			 */
+			if (!kref_get_unless_zero(&obj->base.refcount)) {
+				spin_unlock(&mn->lock);
+				serial = 0;
+				continue;
+			}
+
 			serial = mn->serial;
 		}
 		spin_unlock(&mn->lock);
diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index eea9e36..a74aaf9 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -3521,8 +3521,6 @@ intel_dp_link_down(struct intel_dp *intel_dp)
 	enum port port = intel_dig_port->port;
 	struct drm_device *dev = intel_dig_port->base.base.dev;
 	struct drm_i915_private *dev_priv = dev->dev_private;
-	struct intel_crtc *intel_crtc =
-		to_intel_crtc(intel_dig_port->base.base.crtc);
 	uint32_t DP = intel_dp->DP;
 
 	if (WARN_ON(HAS_DDI(dev)))
@@ -3547,8 +3545,6 @@ intel_dp_link_down(struct intel_dp *intel_dp)
 
 	if (HAS_PCH_IBX(dev) &&
 	    I915_READ(intel_dp->output_reg) & DP_PIPEB_SELECT) {
-		struct drm_crtc *crtc = intel_dig_port->base.base.crtc;
-
 		/* Hardware workaround: leaving our transcoder select
 		 * set to transcoder B while it's off will prevent the
 		 * corresponding HDMI output on transcoder A.
@@ -3559,18 +3555,7 @@ intel_dp_link_down(struct intel_dp *intel_dp)
 		 */
 		DP &= ~DP_PIPEB_SELECT;
 		I915_WRITE(intel_dp->output_reg, DP);
-
-		/* Changes to enable or select take place the vblank
-		 * after being written.
-		 */
-		if (WARN_ON(crtc == NULL)) {
-			/* We should never try to disable a port without a crtc
-			 * attached. For paranoia keep the code around for a
-			 * bit. */
-			POSTING_READ(intel_dp->output_reg);
-			msleep(50);
-		} else
-			intel_wait_for_vblank(dev, intel_crtc->pipe);
+		POSTING_READ(intel_dp->output_reg);
 	}
 
 	DP &= ~DP_AUDIO_OUTPUT_ENABLE;
@@ -4446,7 +4431,7 @@ intel_dp_hpd_pulse(struct intel_digital_port *intel_dig_port, bool long_hpd)
 		 */
 		DRM_DEBUG_KMS("ignoring long hpd on eDP port %c\n",
 			      port_name(intel_dig_port->port));
-		return false;
+		return IRQ_HANDLED;
 	}
 
 	DRM_DEBUG_KMS("got hpd irq on port %c - %s\n",
diff --git a/drivers/gpu/drm/i915/intel_dsi.c b/drivers/gpu/drm/i915/intel_dsi.c
index ef3df5e..10ab684 100644
--- a/drivers/gpu/drm/i915/intel_dsi.c
+++ b/drivers/gpu/drm/i915/intel_dsi.c
@@ -360,12 +360,11 @@ static void intel_dsi_device_ready(struct intel_encoder *encoder)
 		I915_WRITE(MIPI_DEVICE_READY(port), ULPS_STATE_ENTER);
 		usleep_range(2500, 3000);
 
-		val = I915_READ(MIPI_PORT_CTRL(port));
-
 		/* Enable MIPI PHY transparent latch
 		 * Common bit for both MIPI Port A & MIPI Port C
 		 * No similar bit in MIPI Port C reg
 		 */
+		val = I915_READ(MIPI_PORT_CTRL(PORT_A));
 		I915_WRITE(MIPI_PORT_CTRL(PORT_A), val | LP_OUTPUT_HOLD);
 		usleep_range(1000, 1500);
 
@@ -543,10 +542,10 @@ static void intel_dsi_clear_device_ready(struct intel_encoder *encoder)
 							== 0x00000), 30))
 			DRM_ERROR("DSI LP not going Low\n");
 
-		val = I915_READ(MIPI_PORT_CTRL(port));
 		/* Disable MIPI PHY transparent latch
 		 * Common bit for both MIPI Port A & MIPI Port C
 		 */
+		val = I915_READ(MIPI_PORT_CTRL(PORT_A));
 		I915_WRITE(MIPI_PORT_CTRL(PORT_A), val & ~LP_OUTPUT_HOLD);
 		usleep_range(1000, 1500);
 
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index a94346f..0f358c5 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1211,15 +1211,17 @@ static int gen8_emit_flush(struct intel_ringbuffer *ringbuf,
 
 	cmd = MI_FLUSH_DW + 1;
 
-	if (ring == &dev_priv->ring[VCS]) {
-		if (invalidate_domains & I915_GEM_GPU_DOMAINS)
-			cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD |
-				MI_FLUSH_DW_STORE_INDEX |
-				MI_FLUSH_DW_OP_STOREDW;
-	} else {
-		if (invalidate_domains & I915_GEM_DOMAIN_RENDER)
-			cmd |= MI_INVALIDATE_TLB | MI_FLUSH_DW_STORE_INDEX |
-				MI_FLUSH_DW_OP_STOREDW;
+	/* We always require a command barrier so that subsequent
+	 * commands, such as breadcrumb interrupts, are strictly ordered
+	 * wrt the contents of the write cache being flushed to memory
+	 * (and thus being coherent from the CPU).
+	 */
+	cmd |= MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
+
+	if (invalidate_domains & I915_GEM_GPU_DOMAINS) {
+		cmd |= MI_INVALIDATE_TLB;
+		if (ring == &dev_priv->ring[VCS])
+			cmd |= MI_INVALIDATE_BSD;
 	}
 
 	intel_logical_ring_emit(ringbuf, cmd);
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 6ece663..24d77dd 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -4005,7 +4005,10 @@ static void gen6_init_rps_frequencies(struct drm_device *dev)
 					&ddcc_status);
 		if (0 == ret)
 			dev_priv->rps.efficient_freq =
-				(ddcc_status >> 8) & 0xff;
+				clamp_t(u8,
+					((ddcc_status >> 8) & 0xff),
+					dev_priv->rps.min_freq,
+					dev_priv->rps.max_freq);
 	}
 
 	/* Preserve min/max settings in case of re-init */
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 0bd3976..e5b3c6d 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -2240,6 +2240,14 @@ static int gen6_bsd_ring_flush(struct intel_engine_cs *ring,
 	cmd = MI_FLUSH_DW;
 	if (INTEL_INFO(ring->dev)->gen >= 8)
 		cmd += 1;
+
+	/* We always require a command barrier so that subsequent
+	 * commands, such as breadcrumb interrupts, are strictly ordered
+	 * wrt the contents of the write cache being flushed to memory
+	 * (and thus being coherent from the CPU).
+	 */
+	cmd |= MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
+
 	/*
 	 * Bspec vol 1c.5 - video engine command streamer:
 	 * "If ENABLED, all TLBs will be invalidated once the flush
@@ -2247,8 +2255,8 @@ static int gen6_bsd_ring_flush(struct intel_engine_cs *ring,
 	 * Post-Sync Operation field is a value of 1h or 3h."
 	 */
 	if (invalidate & I915_GEM_GPU_DOMAINS)
-		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD |
-			MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
+		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD;
+
 	intel_ring_emit(ring, cmd);
 	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
 	if (INTEL_INFO(ring->dev)->gen >= 8) {
@@ -2344,6 +2352,14 @@ static int gen6_ring_flush(struct intel_engine_cs *ring,
 	cmd = MI_FLUSH_DW;
 	if (INTEL_INFO(ring->dev)->gen >= 8)
 		cmd += 1;
+
+	/* We always require a command barrier so that subsequent
+	 * commands, such as breadcrumb interrupts, are strictly ordered
+	 * wrt the contents of the write cache being flushed to memory
+	 * (and thus being coherent from the CPU).
+	 */
+	cmd |= MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
+
 	/*
 	 * Bspec vol 1c.3 - blitter engine command streamer:
 	 * "If ENABLED, all TLBs will be invalidated once the flush
@@ -2351,8 +2367,7 @@ static int gen6_ring_flush(struct intel_engine_cs *ring,
 	 * Post-Sync Operation field is a value of 1h or 3h."
 	 */
 	if (invalidate & I915_GEM_DOMAIN_RENDER)
-		cmd |= MI_INVALIDATE_TLB | MI_FLUSH_DW_STORE_INDEX |
-			MI_FLUSH_DW_OP_STOREDW;
+		cmd |= MI_INVALIDATE_TLB;
 	intel_ring_emit(ring, cmd);
 	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
 	if (INTEL_INFO(ring->dev)->gen >= 8) {
diff --git a/drivers/gpu/drm/i915/intel_sideband.c b/drivers/gpu/drm/i915/intel_sideband.c
index 3c42eef..693ce82 100644
--- a/drivers/gpu/drm/i915/intel_sideband.c
+++ b/drivers/gpu/drm/i915/intel_sideband.c
@@ -82,7 +82,7 @@ u32 vlv_punit_read(struct drm_i915_private *dev_priv, u32 addr)
 	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
 
 	mutex_lock(&dev_priv->dpio_lock);
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_PUNIT,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_PUNIT,
 			SB_CRRDDA_NP, addr, &val);
 	mutex_unlock(&dev_priv->dpio_lock);
 
@@ -94,7 +94,7 @@ void vlv_punit_write(struct drm_i915_private *dev_priv, u32 addr, u32 val)
 	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
 
 	mutex_lock(&dev_priv->dpio_lock);
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_PUNIT,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_PUNIT,
 			SB_CRWRDA_NP, addr, &val);
 	mutex_unlock(&dev_priv->dpio_lock);
 }
@@ -103,7 +103,7 @@ u32 vlv_bunit_read(struct drm_i915_private *dev_priv, u32 reg)
 {
 	u32 val = 0;
 
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_BUNIT,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_BUNIT,
 			SB_CRRDDA_NP, reg, &val);
 
 	return val;
@@ -111,7 +111,7 @@ u32 vlv_bunit_read(struct drm_i915_private *dev_priv, u32 reg)
 
 void vlv_bunit_write(struct drm_i915_private *dev_priv, u32 reg, u32 val)
 {
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_BUNIT,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_BUNIT,
 			SB_CRWRDA_NP, reg, &val);
 }
 
@@ -122,7 +122,7 @@ u32 vlv_nc_read(struct drm_i915_private *dev_priv, u8 addr)
 	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
 
 	mutex_lock(&dev_priv->dpio_lock);
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_NC,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_NC,
 			SB_CRRDDA_NP, addr, &val);
 	mutex_unlock(&dev_priv->dpio_lock);
 
@@ -132,56 +132,56 @@ u32 vlv_nc_read(struct drm_i915_private *dev_priv, u8 addr)
 u32 vlv_gpio_nc_read(struct drm_i915_private *dev_priv, u32 reg)
 {
 	u32 val = 0;
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_GPIO_NC,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_GPIO_NC,
 			SB_CRRDDA_NP, reg, &val);
 	return val;
 }
 
 void vlv_gpio_nc_write(struct drm_i915_private *dev_priv, u32 reg, u32 val)
 {
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_GPIO_NC,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_GPIO_NC,
 			SB_CRWRDA_NP, reg, &val);
 }
 
 u32 vlv_cck_read(struct drm_i915_private *dev_priv, u32 reg)
 {
 	u32 val = 0;
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_CCK,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_CCK,
 			SB_CRRDDA_NP, reg, &val);
 	return val;
 }
 
 void vlv_cck_write(struct drm_i915_private *dev_priv, u32 reg, u32 val)
 {
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_CCK,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_CCK,
 			SB_CRWRDA_NP, reg, &val);
 }
 
 u32 vlv_ccu_read(struct drm_i915_private *dev_priv, u32 reg)
 {
 	u32 val = 0;
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_CCU,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_CCU,
 			SB_CRRDDA_NP, reg, &val);
 	return val;
 }
 
 void vlv_ccu_write(struct drm_i915_private *dev_priv, u32 reg, u32 val)
 {
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_CCU,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_CCU,
 			SB_CRWRDA_NP, reg, &val);
 }
 
 u32 vlv_gps_core_read(struct drm_i915_private *dev_priv, u32 reg)
 {
 	u32 val = 0;
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_GPS_CORE,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_GPS_CORE,
 			SB_CRRDDA_NP, reg, &val);
 	return val;
 }
 
 void vlv_gps_core_write(struct drm_i915_private *dev_priv, u32 reg, u32 val)
 {
-	vlv_sideband_rw(dev_priv, PCI_DEVFN(2, 0), IOSF_PORT_GPS_CORE,
+	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_GPS_CORE,
 			SB_CRWRDA_NP, reg, &val);
 }
 
diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
index 76b60a3..c47a3ba 100644
--- a/drivers/gpu/drm/i915/intel_uncore.c
+++ b/drivers/gpu/drm/i915/intel_uncore.c
@@ -166,7 +166,8 @@ fw_domains_reset(struct drm_i915_private *dev_priv, enum forcewake_domains fw_do
 	struct intel_uncore_forcewake_domain *d;
 	enum forcewake_domain_id id;
 
-	WARN_ON(dev_priv->uncore.fw_domains == 0);
+	if (dev_priv->uncore.fw_domains == 0)
+		return;
 
 	for_each_fw_domain_mask(d, fw_domains, dev_priv, id)
 		fw_domain_reset(d);
@@ -997,6 +998,9 @@ static void intel_uncore_fw_domains_init(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
+	if (INTEL_INFO(dev_priv->dev)->gen <= 5)
+		return;
+
 	if (IS_GEN9(dev)) {
 		dev_priv->uncore.funcs.force_wake_get = fw_domains_get;
 		dev_priv->uncore.funcs.force_wake_put = fw_domains_put;
@@ -1069,6 +1073,9 @@ static void intel_uncore_fw_domains_init(struct drm_device *dev)
 		fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
 			       FORCEWAKE, FORCEWAKE_ACK);
 	}
+
+	/* All future platforms are expected to require complex power gating */
+	WARN_ON(dev_priv->uncore.fw_domains == 0);
 }
 
 void intel_uncore_init(struct drm_device *dev)