Skip to contentSkip to conversation

drm/i915: Nuke debug messages from the pipe update critical section

PATCH In review [email protected] 5 participants 8 messages Mar 7, 2017 – Mar 8, 2017

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

PATCH 1/1 drm/i915: Nuke debug messages from the pipe update critical section

From: Ville Syrjälä <[email protected]>

printks are slow so we should not be doing them from the vblank evade critical section. These could explain why we sometimes seem to blow past our 100 usec deadline.

The problem has been there ever since commit bfd16b2a23dc ("drm/i915:
Make updating pipe without modeset atomic.") but it may not have been readily visible until commit e1edbd44e23b ("drm/i915: Complain if we take too long under vblank evasion.") increased our chances of noticing it.

Cc Maarten Lankhorst <[email protected]>
Fixes bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.")
---
 drivers/gpu/drm/i915/intel_display.c | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)
drivers/gpu/drm/i915/intel_display.c +1 -11
index e77ca7dfa44d..726ae191076b 100644
@@ -3650,10 +3650,6 @@ static void intel_update_pipe_config(struct intel_crtc *crtc,
 	/* drm_atomic_helper_update_legacy_modeset_state might not be called. */
 	crtc->base.mode = crtc->base.state->mode;
 
-	DRM_DEBUG_KMS("Updating pipe size %ix%i -> %ix%i\n",
-		      old_crtc_state->pipe_src_w, old_crtc_state->pipe_src_h,
-		      pipe_config->pipe_src_w, pipe_config->pipe_src_h);
-
 	/*
 	 * Update pipe size and adjust fitter if needed: the reason for this is
 	 * that in compute_mode_changes we check the native mode (not the pfit
@@ -4775,23 +4771,17 @@ static void skylake_pfit_enable(struct intel_crtc *crtc)
 	struct intel_crtc_scaler_state *scaler_state =
 		&crtc->config->scaler_state;
 
-	DRM_DEBUG_KMS("for crtc_state = %p\n", crtc->config);
-
 	if (crtc->config->pch_pfit.enabled) {
 		int id;
 
-		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0)) {
-			DRM_ERROR("Requesting pfit without getting a scaler first\n");
+		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0))
 			return;
-		}
 
 		id = scaler_state->scaler_id;
 		I915_WRITE(SKL_PS_CTRL(pipe, id), PS_SCALER_EN |
 			PS_FILTER_MEDIUM | scaler_state->scalers[id].mode);
 		I915_WRITE(SKL_PS_WIN_POS(pipe, id), crtc->config->pch_pfit.pos);
 		I915_WRITE(SKL_PS_WIN_SZ(pipe, id), crtc->config->pch_pfit.size);
-
-		DRM_DEBUG_KMS("for crtc_state = %p scaler_id = %d\n", crtc->config, id);
 	}
 }
 
Signature · 6 lines
2.10.2

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

✓ Fi.CI.BAT: success for drm/i915: Nuke debug messages from the pipe update critical section

== Series Details ==

== Series Details ==

Series: drm/i915: Nuke debug messages from the pipe update critical section
URL : https://patchwork.freedesktop.org/series/20853/
State : success

== Summary ==

Series 20853v1 drm/i915: Nuke debug messages from the pipe update critical section
https://patchwork.freedesktop.org/api/1.0/series/20853/revisions/1/mbox/

Test gem_exec_flush:

        Subgroup basic-batch-kernel-default-uc:
                fail       -> PASS       (fi-snb-2600) fdo#100007

fdo#100007 https://bugs.freedesktop.org/show_bug.cgi?id=100007

fi-bdw-5557u     total:278  pass:267  dwarn:0   dfail:0   fail:0   skip:11  time: 471s
fi-bsw-n3050     total:278  pass:239  dwarn:0   dfail:0   fail:0   skip:39  time: 613s
fi-bxt-j4205     total:278  pass:259  dwarn:0   dfail:0   fail:0   skip:19  time: 534s
fi-bxt-t5700     total:278  pass:258  dwarn:0   dfail:0   fail:0   skip:20  time: 620s
fi-byt-j1900     total:278  pass:251  dwarn:0   dfail:0   fail:0   skip:27  time: 505s
fi-byt-n2820     total:278  pass:247  dwarn:0   dfail:0   fail:0   skip:31  time: 501s
fi-hsw-4770      total:278  pass:262  dwarn:0   dfail:0   fail:0   skip:16  time: 436s
fi-hsw-4770r     total:278  pass:262  dwarn:0   dfail:0   fail:0   skip:16  time: 432s
fi-ilk-650       total:278  pass:228  dwarn:0   dfail:0   fail:0   skip:50  time: 441s
fi-ivb-3520m     total:278  pass:260  dwarn:0   dfail:0   fail:0   skip:18  time: 506s
fi-ivb-3770      total:278  pass:260  dwarn:0   dfail:0   fail:0   skip:18  time: 490s
fi-kbl-7500u     total:278  pass:259  dwarn:1   dfail:0   fail:0   skip:18  time: 475s
fi-skl-6260u     total:278  pass:268  dwarn:0   dfail:0   fail:0   skip:10  time: 505s
fi-skl-6700hq    total:278  pass:261  dwarn:0   dfail:0   fail:0   skip:17  time: 603s
fi-skl-6700k     total:278  pass:256  dwarn:4   dfail:0   fail:0   skip:18  time: 493s
fi-skl-6770hq    total:278  pass:268  dwarn:0   dfail:0   fail:0   skip:10  time: 552s
fi-snb-2520m     total:278  pass:250  dwarn:0   dfail:0   fail:0   skip:28  time: 549s
fi-snb-2600      total:278  pass:249  dwarn:0   dfail:0   fail:0   skip:29  time: 415s

dcca4ca8923adc61254f23eb66ca18a4c9e1ffd3 drm-tip: 2017y-03m-07d-18h-17m-03s UTC integration manifest 92f5123 drm/i915: Nuke debug messages from the pipe update critical section

== Logs ==

For more details see: https://intel-gfx-ci.01.org/CI/Patchwork_4089/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer
#2 lore raw
In reply to [email protected]
Quoted 6 lines · On Tue, 07 Mar 2017, [email protected] wrote:

From: Ville Syrjälä <[email protected]>

printks are slow so we should not be doing them from the vblank evade critical section. These could explain why we sometimes seem to blow past our 100 usec deadline.

Quoted 8 lines

The problem has been there ever since commit bfd16b2a23dc ("drm/i915:
Make updating pipe without modeset atomic.") but it may not have been readily visible until commit e1edbd44e23b ("drm/i915: Complain if we take too long under vblank evasion.") increased our chances of noticing it.

Cc Maarten Lankhorst <[email protected]>
Fixes bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.")

Is this not worth it for cc: stable, because e1edbd44e23b is not in Linus' tree?

BR,
Jani.

Quoted 44 lines
---
 drivers/gpu/drm/i915/intel_display.c | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)
drivers/gpu/drm/i915/intel_display.c +1 -11
index e77ca7dfa44d..726ae191076b 100644
@@ -3650,10 +3650,6 @@ static void intel_update_pipe_config(struct intel_crtc *crtc,
 	/* drm_atomic_helper_update_legacy_modeset_state might not be called. */
 	crtc->base.mode = crtc->base.state->mode;
 
-	DRM_DEBUG_KMS("Updating pipe size %ix%i -> %ix%i\n",
-		      old_crtc_state->pipe_src_w, old_crtc_state->pipe_src_h,
-		      pipe_config->pipe_src_w, pipe_config->pipe_src_h);
-
 	/*
 	 * Update pipe size and adjust fitter if needed: the reason for this is
 	 * that in compute_mode_changes we check the native mode (not the pfit
@@ -4775,23 +4771,17 @@ static void skylake_pfit_enable(struct intel_crtc *crtc)
 	struct intel_crtc_scaler_state *scaler_state =
 		&crtc->config->scaler_state;
 
-	DRM_DEBUG_KMS("for crtc_state = %p\n", crtc->config);
-
 	if (crtc->config->pch_pfit.enabled) {
 		int id;
 
-		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0)) {
-			DRM_ERROR("Requesting pfit without getting a scaler first\n");
+		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0))
 			return;
-		}
 
 		id = scaler_state->scaler_id;
 		I915_WRITE(SKL_PS_CTRL(pipe, id), PS_SCALER_EN |
 			PS_FILTER_MEDIUM | scaler_state->scalers[id].mode);
 		I915_WRITE(SKL_PS_WIN_POS(pipe, id), crtc->config->pch_pfit.pos);
 		I915_WRITE(SKL_PS_WIN_SZ(pipe, id), crtc->config->pch_pfit.size);
-
-		DRM_DEBUG_KMS("for crtc_state = %p scaler_id = %d\n", crtc->config, id);
 	}
 }
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw
Quoted 13 lines · Jani Nikula
Quoted 12 lines · On Tue, 07 Mar 2017, [email protected] wrote:

From: Ville Syrjälä <[email protected]>

printks are slow so we should not be doing them from the vblank evade critical section. These could explain why we sometimes seem to blow past our 100 usec deadline.

The problem has been there ever since commit bfd16b2a23dc ("drm/i915:
Make updating pipe without modeset atomic.") but it may not have been readily visible until commit e1edbd44e23b ("drm/i915: Complain if we take too long under vblank evasion.") increased our chances of noticing it.

Quoted 5 lines
Quoted 2 lines
Cc Maarten Lankhorst <[email protected]>
Fixes bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.")

Is this not worth it for cc: stable, because e1edbd44e23b is not in Linus' tree?

Hmm. We did have the other error message about exceeding the deadline already before, so it's possible that this might have caused some noise even before the e1edbd44e23b. Also removing printk()s can do harm (famous last words, right?) so we might as well cc:stable I suppose.

Quoted 52 lines

BR,
Jani.

Quoted 44 lines
---
 drivers/gpu/drm/i915/intel_display.c | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)
drivers/gpu/drm/i915/intel_display.c +1 -11
index e77ca7dfa44d..726ae191076b 100644
@@ -3650,10 +3650,6 @@ static void intel_update_pipe_config(struct intel_crtc *crtc,
 	/* drm_atomic_helper_update_legacy_modeset_state might not be called. */
 	crtc->base.mode = crtc->base.state->mode;
 
-	DRM_DEBUG_KMS("Updating pipe size %ix%i -> %ix%i\n",
-		      old_crtc_state->pipe_src_w, old_crtc_state->pipe_src_h,
-		      pipe_config->pipe_src_w, pipe_config->pipe_src_h);
-
 	/*
 	 * Update pipe size and adjust fitter if needed: the reason for this is
 	 * that in compute_mode_changes we check the native mode (not the pfit
@@ -4775,23 +4771,17 @@ static void skylake_pfit_enable(struct intel_crtc *crtc)
 	struct intel_crtc_scaler_state *scaler_state =
 		&crtc->config->scaler_state;
 
-	DRM_DEBUG_KMS("for crtc_state = %p\n", crtc->config);
-
 	if (crtc->config->pch_pfit.enabled) {
 		int id;
 
-		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0)) {
-			DRM_ERROR("Requesting pfit without getting a scaler first\n");
+		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0))
 			return;
-		}
 
 		id = scaler_state->scaler_id;
 		I915_WRITE(SKL_PS_CTRL(pipe, id), PS_SCALER_EN |
 			PS_FILTER_MEDIUM | scaler_state->scalers[id].mode);
 		I915_WRITE(SKL_PS_WIN_POS(pipe, id), crtc->config->pch_pfit.pos);
 		I915_WRITE(SKL_PS_WIN_SZ(pipe, id), crtc->config->pch_pfit.size);
-
-		DRM_DEBUG_KMS("for crtc_state = %p scaler_id = %d\n", crtc->config, id);
 	}
 }
Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
Signature · 6 lines
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer
#4 lore raw
Quoted 17 lines · Ville Syrjälä
Quoted 16 lines · Jani Nikula
Quoted 14 lines · On Tue, 07 Mar 2017, [email protected] wrote:

From: Ville Syrjälä <[email protected]>

printks are slow so we should not be doing them from the vblank evade critical section. These could explain why we sometimes seem to blow past our 100 usec deadline.

The problem has been there ever since commit bfd16b2a23dc ("drm/i915:
Make updating pipe without modeset atomic.") but it may not have been readily visible until commit e1edbd44e23b ("drm/i915: Complain if we take too long under vblank evasion.") increased our chances of noticing it.

Cc Maarten Lankhorst <[email protected]>
Fixes bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.")
Quoted 7 lines
Quoted 2 lines

Is this not worth it for cc: stable, because e1edbd44e23b is not in Linus' tree?

Hmm. We did have the other error message about exceeding the deadline already before, so it's possible that this might have caused some noise even before the e1edbd44e23b. Also removing printk()s can do harm (famous last words, right?) so we might as well cc:stable I suppose.

Nice typo there! ;) But make it so.

BR,
Jani.

Quoted 53 lines
Quoted 53 lines

BR,
Jani.

Quoted 44 lines
---
 drivers/gpu/drm/i915/intel_display.c | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)
drivers/gpu/drm/i915/intel_display.c +1 -11
index e77ca7dfa44d..726ae191076b 100644
@@ -3650,10 +3650,6 @@ static void intel_update_pipe_config(struct intel_crtc *crtc,
 	/* drm_atomic_helper_update_legacy_modeset_state might not be called. */
 	crtc->base.mode = crtc->base.state->mode;
 
-	DRM_DEBUG_KMS("Updating pipe size %ix%i -> %ix%i\n",
-		      old_crtc_state->pipe_src_w, old_crtc_state->pipe_src_h,
-		      pipe_config->pipe_src_w, pipe_config->pipe_src_h);
-
 	/*
 	 * Update pipe size and adjust fitter if needed: the reason for this is
 	 * that in compute_mode_changes we check the native mode (not the pfit
@@ -4775,23 +4771,17 @@ static void skylake_pfit_enable(struct intel_crtc *crtc)
 	struct intel_crtc_scaler_state *scaler_state =
 		&crtc->config->scaler_state;
 
-	DRM_DEBUG_KMS("for crtc_state = %p\n", crtc->config);
-
 	if (crtc->config->pch_pfit.enabled) {
 		int id;
 
-		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0)) {
-			DRM_ERROR("Requesting pfit without getting a scaler first\n");
+		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0))
 			return;
-		}
 
 		id = scaler_state->scaler_id;
 		I915_WRITE(SKL_PS_CTRL(pipe, id), PS_SCALER_EN |
 			PS_FILTER_MEDIUM | scaler_state->scalers[id].mode);
 		I915_WRITE(SKL_PS_WIN_POS(pipe, id), crtc->config->pch_pfit.pos);
 		I915_WRITE(SKL_PS_WIN_SZ(pipe, id), crtc->config->pch_pfit.size);
-
-		DRM_DEBUG_KMS("for crtc_state = %p scaler_id = %d\n", crtc->config, id);
 	}
 }
Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#5 lore raw
In reply to Ville Syrjälä
Quoted 17 lines · Ville Syrjälä
Quoted 16 lines · Jani Nikula
Quoted 14 lines · On Tue, 07 Mar 2017, [email protected] wrote:

From: Ville Syrjälä <[email protected]>

printks are slow so we should not be doing them from the vblank evade critical section. These could explain why we sometimes seem to blow past our 100 usec deadline.

The problem has been there ever since commit bfd16b2a23dc ("drm/i915:
Make updating pipe without modeset atomic.") but it may not have been readily visible until commit e1edbd44e23b ("drm/i915: Complain if we take too long under vblank evasion.") increased our chances of noticing it.

Cc Maarten Lankhorst <[email protected]>
Fixes bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.")
Quoted 7 lines
Quoted 2 lines

Is this not worth it for cc: stable, because e1edbd44e23b is not in Linus' tree?

Hmm. We did have the other error message about exceeding the deadline already before, so it's possible that this might have caused some noise even before the e1edbd44e23b. Also removing printk()s can do harm (famous last words, right?) so we might as well cc:stable I suppose.

+1 on cc: stable from me.

Also, is there any way we could try to prevent this? One idea that crossed my mind is to create a set of forbid/allow_printk() functions, which simple fake-take the printk buffer lock using lockdep (but not take the lock for real, because that's where the stalls are from). Since we already disable irqs there's no problem with recursion into irq handlers which might call printk (and upset the lockdep rules), we just need to nest forbid/allow_printk within the irq disabled section.

Can I volunteer to try this out pls?

Thanks, Daniel

Quoted 61 lines
Quoted 53 lines

BR,
Jani.

Quoted 44 lines
---
 drivers/gpu/drm/i915/intel_display.c | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)
drivers/gpu/drm/i915/intel_display.c +1 -11
index e77ca7dfa44d..726ae191076b 100644
@@ -3650,10 +3650,6 @@ static void intel_update_pipe_config(struct intel_crtc *crtc,
 	/* drm_atomic_helper_update_legacy_modeset_state might not be called. */
 	crtc->base.mode = crtc->base.state->mode;
 
-	DRM_DEBUG_KMS("Updating pipe size %ix%i -> %ix%i\n",
-		      old_crtc_state->pipe_src_w, old_crtc_state->pipe_src_h,
-		      pipe_config->pipe_src_w, pipe_config->pipe_src_h);
-
 	/*
 	 * Update pipe size and adjust fitter if needed: the reason for this is
 	 * that in compute_mode_changes we check the native mode (not the pfit
@@ -4775,23 +4771,17 @@ static void skylake_pfit_enable(struct intel_crtc *crtc)
 	struct intel_crtc_scaler_state *scaler_state =
 		&crtc->config->scaler_state;
 
-	DRM_DEBUG_KMS("for crtc_state = %p\n", crtc->config);
-
 	if (crtc->config->pch_pfit.enabled) {
 		int id;
 
-		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0)) {
-			DRM_ERROR("Requesting pfit without getting a scaler first\n");
+		if (WARN_ON(crtc->config->scaler_state.scaler_id < 0))
 			return;
-		}
 
 		id = scaler_state->scaler_id;
 		I915_WRITE(SKL_PS_CTRL(pipe, id), PS_SCALER_EN |
 			PS_FILTER_MEDIUM | scaler_state->scalers[id].mode);
 		I915_WRITE(SKL_PS_WIN_POS(pipe, id), crtc->config->pch_pfit.pos);
 		I915_WRITE(SKL_PS_WIN_SZ(pipe, id), crtc->config->pch_pfit.size);
-
-		DRM_DEBUG_KMS("for crtc_state = %p scaler_id = %d\n", crtc->config, id);
 	}
 }
Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
Signature · 6 lines
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 1/1
#6 lore raw
In reply to [email protected]

Op 07-03-17 om 21:54 schreef [email protected]:

Quoted 12 lines

From: Ville Syrjälä <[email protected]>

printks are slow so we should not be doing them from the vblank evade critical section. These could explain why we sometimes seem to blow past our 100 usec deadline.

The problem has been there ever since commit bfd16b2a23dc ("drm/i915:
Make updating pipe without modeset atomic.") but it may not have been readily visible until commit e1edbd44e23b ("drm/i915: Complain if we take too long under vblank evasion.") increased our chances of noticing it.

Quoted 3 lines
Cc Maarten Lankhorst <[email protected]>
Fixes bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.")

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

author Applied
#7 lore raw
Quoted 13 lines · Maarten Lankhorst

Op 07-03-17 om 21:54 schreef [email protected]:

Quoted 12 lines

From: Ville Syrjälä <[email protected]>

printks are slow so we should not be doing them from the vblank evade critical section. These could explain why we sometimes seem to blow past our 100 usec deadline.

The problem has been there ever since commit bfd16b2a23dc ("drm/i915:
Make updating pipe without modeset atomic.") but it may not have been readily visible until commit e1edbd44e23b ("drm/i915: Complain if we take too long under vblank evasion.") increased our chances of noticing it.

Quoted 5 lines
Quoted 3 lines
Cc Maarten Lankhorst <[email protected]>
Fixes bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.")

Pushed to dinq with cc:stable. Thanls for the review.

Signature · 6 lines
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx