Skip to contentSkip to conversation

drm/i915: Fix all pipe->plane mappings before sanitizing crtc

PATCH In review Ander Conselvan de Oliveira 5 participants 6 messages Jan 13, 2015 – Jan 14, 2015

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/i915: Fix all pipe->plane mappings before sanitizing crtc

If QUIRK_PIPEA_FORCE is necessary, intel_sanitize_crtc() might trigger a mode set. In that case, if pipe A is disabled and pipe B is mapped to plane B, that mode set happens before the mapping is fixed. Due to the wrong state, the call to disable pipe B disables plane A (which is already disabled) and later an assertion for plane B being enabled (while it should have been disabled) is triggered.

References: https://bugs.freedesktop.org/show_bug.cgi?id=72782

---
 drivers/gpu/drm/i915/intel_display.c | 50 +++++++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/intel_display.c +32 -18
index d01db1b..cf87528 100644
@@ -13127,30 +13127,22 @@ intel_check_plane_mapping(struct intel_crtc *crtc)
 	return true;
 }
 
-static void intel_sanitize_crtc(struct intel_crtc *crtc)
+static void intel_fix_plane_mapping(struct drm_device *dev)
 {
-	struct drm_device *dev = crtc->base.dev;
-	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 reg;
-
-	/* Clear any frame start delays used for debugging left by the BIOS */
-	reg = PIPECONF(crtc->config.cpu_transcoder);
-	I915_WRITE(reg, I915_READ(reg) & ~PIPECONF_FRAME_START_DELAY_MASK);
+	struct drm_i915_private *dev_priv = to_i915(dev);
+	struct intel_crtc *crtc;
 
-	/* restore vblank interrupts to correct state */
-	if (crtc->active) {
-		update_scanline_offset(crtc);
-		drm_vblank_on(dev, crtc->pipe);
-	} else
-		drm_vblank_off(dev, crtc->pipe);
+	/* gen4+ has a fixed plane -> pipe mapping.*/
+	if (INTEL_INFO(dev)->gen >= 4)
+		return;
 
-	/* We need to sanitize the plane -> pipe mapping first because this will
-	 * disable the crtc (and hence change the state) if it is wrong. Note
-	 * that gen4+ has a fixed plane -> pipe mapping.  */
-	if (INTEL_INFO(dev)->gen < 4 && !intel_check_plane_mapping(crtc)) {
+	for_each_intel_crtc(dev, crtc) {
 		struct intel_connector *connector;
 		bool plane;
 
+		if (intel_check_plane_mapping(crtc))
+			continue;
+
 		DRM_DEBUG_KMS("[CRTC:%d] wrong plane connection detected!\n",
 			      crtc->base.base.id);
 
@@ -13184,6 +13176,24 @@ static void intel_sanitize_crtc(struct intel_crtc *crtc)
 		WARN_ON(crtc->active);
 		crtc->base.enabled = false;
 	}
+}
+
+static void intel_sanitize_crtc(struct intel_crtc *crtc)
+{
+	struct drm_device *dev = crtc->base.dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 reg;
+
+	/* Clear any frame start delays used for debugging left by the BIOS */
+	reg = PIPECONF(crtc->config.cpu_transcoder);
+	I915_WRITE(reg, I915_READ(reg) & ~PIPECONF_FRAME_START_DELAY_MASK);
+
+	/* restore vblank interrupts to correct state */
+	if (crtc->active) {
+		update_scanline_offset(crtc);
+		drm_vblank_on(dev, crtc->pipe);
+	} else
+		drm_vblank_off(dev, crtc->pipe);
 
 	if (dev_priv->quirks & QUIRK_PIPEA_FORCE &&
 	    crtc->pipe == PIPE_A && !crtc->active) {
@@ -13442,6 +13452,10 @@ void intel_modeset_setup_hw_state(struct drm_device *dev,
 		intel_sanitize_encoder(encoder);
 	}
 
+	/* We need to sanitize the plane -> pipe mapping first because this will
+	 * disable the crtc (and hence change the state) if it is wrong. */
+	intel_fix_plane_mapping(dev);
+
 	for_each_pipe(dev_priv, pipe) {
 		crtc = to_intel_crtc(dev_priv->pipe_to_crtc_mapping[pipe]);
 		intel_sanitize_crtc(crtc);
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Tested-by on v1 1/1
#1 lore raw

-------------------------------------Summary-------------------------------------
Platform Delta drm-intel-nightly Series Applied

PNV                                  354/354              354/354
ILK                                  201/201              201/201
SNB              +1-1              401/424              401/424
IVB                                  488/488              488/488
BYT                                  278/278              278/278
HSW                                  529/529              529/529
BDW                 -1              405/405              404/405
-------------------------------------Detailed-------------------------------------
Platform  Test                                drm-intel-nightly          Series Applied
 SNB  igt_kms_flip_flip-vs-dpms-off-vs-modeset-interruptible      NSPT(1, M35)PASS(10, M35M22)      PASS(1, M35)
*SNB  igt_gem_concurrent_blit_gtt-rcs-early-read-interruptible      PASS(10, M35M22)      DMESG_WARN(1, M35)
*BDW  igt_gem_concurrent_blit_gtt-bcs-gpu-read-after-write-interruptible      PASS(9, M30M28)      DMESG_WARN(1, M30)

Note: You need to pay more attention to line start with '*'
_______________________________________________
Intel-gfx mailing list [email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer
#2 lore raw
In reply to Ander Conselvan de Oliveira
Quoted 8 lines · Ander Conselvan de Oliveira

If QUIRK_PIPEA_FORCE is necessary, intel_sanitize_crtc() might trigger a mode set. In that case, if pipe A is disabled and pipe B is mapped to plane B, that mode set happens before the mapping is fixed. Due to the wrong state, the call to disable pipe B disables plane A (which is already disabled) and later an assertion for plane B being enabled (while it should have been disabled) is triggered.

References: https://bugs.freedesktop.org/show_bug.cgi?id=72782

If the patch is expected to fix the bug, please use Bugzilla:. If you're referencing a bug because it's otherwise relevant, please use References:.

Which one is this?

BR,
Jani.

Quoted 93 lines
---
 drivers/gpu/drm/i915/intel_display.c | 50 +++++++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/intel_display.c +32 -18
index d01db1b..cf87528 100644
@@ -13127,30 +13127,22 @@ intel_check_plane_mapping(struct intel_crtc *crtc)
 	return true;
 }
 
-static void intel_sanitize_crtc(struct intel_crtc *crtc)
+static void intel_fix_plane_mapping(struct drm_device *dev)
 {
-	struct drm_device *dev = crtc->base.dev;
-	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 reg;
-
-	/* Clear any frame start delays used for debugging left by the BIOS */
-	reg = PIPECONF(crtc->config.cpu_transcoder);
-	I915_WRITE(reg, I915_READ(reg) & ~PIPECONF_FRAME_START_DELAY_MASK);
+	struct drm_i915_private *dev_priv = to_i915(dev);
+	struct intel_crtc *crtc;
 
-	/* restore vblank interrupts to correct state */
-	if (crtc->active) {
-		update_scanline_offset(crtc);
-		drm_vblank_on(dev, crtc->pipe);
-	} else
-		drm_vblank_off(dev, crtc->pipe);
+	/* gen4+ has a fixed plane -> pipe mapping.*/
+	if (INTEL_INFO(dev)->gen >= 4)
+		return;
 
-	/* We need to sanitize the plane -> pipe mapping first because this will
-	 * disable the crtc (and hence change the state) if it is wrong. Note
-	 * that gen4+ has a fixed plane -> pipe mapping.  */
-	if (INTEL_INFO(dev)->gen < 4 && !intel_check_plane_mapping(crtc)) {
+	for_each_intel_crtc(dev, crtc) {
 		struct intel_connector *connector;
 		bool plane;
 
+		if (intel_check_plane_mapping(crtc))
+			continue;
+
 		DRM_DEBUG_KMS("[CRTC:%d] wrong plane connection detected!\n",
 			      crtc->base.base.id);
 
@@ -13184,6 +13176,24 @@ static void intel_sanitize_crtc(struct intel_crtc *crtc)
 		WARN_ON(crtc->active);
 		crtc->base.enabled = false;
 	}
+}
+
+static void intel_sanitize_crtc(struct intel_crtc *crtc)
+{
+	struct drm_device *dev = crtc->base.dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 reg;
+
+	/* Clear any frame start delays used for debugging left by the BIOS */
+	reg = PIPECONF(crtc->config.cpu_transcoder);
+	I915_WRITE(reg, I915_READ(reg) & ~PIPECONF_FRAME_START_DELAY_MASK);
+
+	/* restore vblank interrupts to correct state */
+	if (crtc->active) {
+		update_scanline_offset(crtc);
+		drm_vblank_on(dev, crtc->pipe);
+	} else
+		drm_vblank_off(dev, crtc->pipe);
 
 	if (dev_priv->quirks & QUIRK_PIPEA_FORCE &&
 	    crtc->pipe == PIPE_A && !crtc->active) {
@@ -13442,6 +13452,10 @@ void intel_modeset_setup_hw_state(struct drm_device *dev,
 		intel_sanitize_encoder(encoder);
 	}
 
+	/* We need to sanitize the plane -> pipe mapping first because this will
+	 * disable the crtc (and hence change the state) if it is wrong. */
+	intel_fix_plane_mapping(dev);
+
 	for_each_pipe(dev_priv, pipe) {
 		crtc = to_intel_crtc(dev_priv->pipe_to_crtc_mapping[pipe]);
 		intel_sanitize_crtc(crtc);
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw
Quoted 8 lines · Jani Nikula
Quoted 7 lines · Ander Conselvan de Oliveira

If QUIRK_PIPEA_FORCE is necessary, intel_sanitize_crtc() might trigger a mode set. In that case, if pipe A is disabled and pipe B is mapped to plane B, that mode set happens before the mapping is fixed. Due to the wrong state, the call to disable pipe B disables plane A (which is already disabled) and later an assertion for plane B being enabled (while it should have been disabled) is triggered.

Quoted 7 lines
Quoted 1 line

References: https://bugs.freedesktop.org/show_bug.cgi?id=72782

If the patch is expected to fix the bug, please use Bugzilla:. If you're referencing a bug because it's otherwise relevant, please use References:.

Which one is this?

This showed up in the dmesg attached to the referenced bug, but is not the root cause of the reported problem. As far as I know, there is no open bug for this particular issue.

Ander

Quoted 94 lines

BR,
Jani.

Quoted 89 lines
---
  drivers/gpu/drm/i915/intel_display.c | 50 +++++++++++++++++++++++-------------
  1 file changed, 32 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/intel_display.c +32 -19
index d01db1b..cf87528 100644
@@ -13127,30 +13127,22 @@ intel_check_plane_mapping(struct intel_crtc *crtc)
  	return true;
  }
 
-static void intel_sanitize_crtc(struct intel_crtc *crtc)
+static void intel_fix_plane_mapping(struct drm_device *dev)
  {
-	struct drm_device *dev = crtc->base.dev;
-	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 reg;
-
-	/* Clear any frame start delays used for debugging left by the BIOS */
-	reg = PIPECONF(crtc->config.cpu_transcoder);
-	I915_WRITE(reg, I915_READ(reg) & ~PIPECONF_FRAME_START_DELAY_MASK);
+	struct drm_i915_private *dev_priv = to_i915(dev);
+	struct intel_crtc *crtc;
 
-	/* restore vblank interrupts to correct state */
-	if (crtc->active) {
-		update_scanline_offset(crtc);
-		drm_vblank_on(dev, crtc->pipe);
-	} else
-		drm_vblank_off(dev, crtc->pipe);
+	/* gen4+ has a fixed plane -> pipe mapping.*/
+	if (INTEL_INFO(dev)->gen >= 4)
+		return;
 
-	/* We need to sanitize the plane -> pipe mapping first because this will
-	 * disable the crtc (and hence change the state) if it is wrong. Note
-	 * that gen4+ has a fixed plane -> pipe mapping.  */
-	if (INTEL_INFO(dev)->gen < 4 && !intel_check_plane_mapping(crtc)) {
+	for_each_intel_crtc(dev, crtc) {
  		struct intel_connector *connector;
  		bool plane;
 
+		if (intel_check_plane_mapping(crtc))
+			continue;
+
  		DRM_DEBUG_KMS("[CRTC:%d] wrong plane connection detected!\n",
  			      crtc->base.base.id);
 
@@ -13184,6 +13176,24 @@ static void intel_sanitize_crtc(struct intel_crtc *crtc)
  		WARN_ON(crtc->active);
  		crtc->base.enabled = false;
  	}
+}
+
+static void intel_sanitize_crtc(struct intel_crtc *crtc)
+{
+	struct drm_device *dev = crtc->base.dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 reg;
+
+	/* Clear any frame start delays used for debugging left by the BIOS */
+	reg = PIPECONF(crtc->config.cpu_transcoder);
+	I915_WRITE(reg, I915_READ(reg) & ~PIPECONF_FRAME_START_DELAY_MASK);
+
+	/* restore vblank interrupts to correct state */
+	if (crtc->active) {
+		update_scanline_offset(crtc);
+		drm_vblank_on(dev, crtc->pipe);
+	} else
+		drm_vblank_off(dev, crtc->pipe);
 
  	if (dev_priv->quirks & QUIRK_PIPEA_FORCE &&
  	    crtc->pipe == PIPE_A && !crtc->active) {
@@ -13442,6 +13452,10 @@ void intel_modeset_setup_hw_state(struct drm_device *dev,
  		intel_sanitize_encoder(encoder);
  	}
 
+	/* We need to sanitize the plane -> pipe mapping first because this will
+	 * disable the crtc (and hence change the state) if it is wrong. */
+	intel_fix_plane_mapping(dev);
+
  	for_each_pipe(dev_priv, pipe) {
  		crtc = to_intel_crtc(dev_priv->pipe_to_crtc_mapping[pipe]);
  		intel_sanitize_crtc(crtc);
--

1.9.1

Quoted 5 lines
Quoted 4 lines

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw
Quoted 15 lines · Ander Conselvan de Oliveira
Quoted 14 lines · Jani Nikula
Quoted 8 lines · Ander Conselvan de Oliveira

If QUIRK_PIPEA_FORCE is necessary, intel_sanitize_crtc() might trigger a mode set. In that case, if pipe A is disabled and pipe B is mapped to plane B, that mode set happens before the mapping is fixed. Due to the wrong state, the call to disable pipe B disables plane A (which is already disabled) and later an assertion for plane B being enabled (while it should have been disabled) is triggered.

References: https://bugs.freedesktop.org/show_bug.cgi?id=72782

If the patch is expected to fix the bug, please use Bugzilla:. If you're referencing a bug because it's otherwise relevant, please use References:.

Quoted 5 lines
Quoted 1 line

Which one is this?

This showed up in the dmesg attached to the referenced bug, but is not the root cause of the reported problem. As far as I know, there is no open bug for this particular issue.

But that machine should not be using QUIRK_PIPEA_FORCE... What I expect is happening is that the BIOS sets up the LVDS on the "wrong" pipe. -Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer
#5 lore raw
In reply to Ander Conselvan de Oliveira
Quoted 15 lines · Ander Conselvan de Oliveira
Quoted 14 lines · Jani Nikula
Quoted 8 lines · Ander Conselvan de Oliveira

If QUIRK_PIPEA_FORCE is necessary, intel_sanitize_crtc() might trigger a mode set. In that case, if pipe A is disabled and pipe B is mapped to plane B, that mode set happens before the mapping is fixed. Due to the wrong state, the call to disable pipe B disables plane A (which is already disabled) and later an assertion for plane B being enabled (while it should have been disabled) is triggered.

References: https://bugs.freedesktop.org/show_bug.cgi?id=72782

If the patch is expected to fix the bug, please use Bugzilla:. If you're referencing a bug because it's otherwise relevant, please use References:.

Quoted 5 lines
Quoted 1 line

Which one is this?

This showed up in the dmesg attached to the referenced bug, but is not the root cause of the reported problem. As far as I know, there is no open bug for this particular issue.

Ah, so it's correct as it is. Sorry for the noise. Please carry on.

BR,
Jani.

Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx