| Message ID | 20170307205419.19447-1-ville.syrjala@linux.intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, 07 Mar 2017, ville.syrjala@linux.intel.com wrote: > From: Ville Syrjälä <ville.syrjala@linux.intel.com> > > 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 <maarten.lankhorst@linux.intel.com> > 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. > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> > --- > drivers/gpu/drm/i915/intel_display.c | 12 +----------- > 1 file changed, 1 insertion(+), 11 deletions(-) > > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c > index e77ca7dfa44d..726ae191076b 100644 > --- a/drivers/gpu/drm/i915/intel_display.c > +++ b/drivers/gpu/drm/i915/intel_display.c > @@ -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); > } > }
Op 07-03-17 om 21:54 schreef ville.syrjala@linux.intel.com: > From: Ville Syrjälä <ville.syrjala@linux.intel.com> > > 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 <maarten.lankhorst@linux.intel.com> > Fixes: bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.") > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
On Wed, Mar 08, 2017 at 10:19:39AM +0200, Jani Nikula wrote: > On Tue, 07 Mar 2017, ville.syrjala@linux.intel.com wrote: > > From: Ville Syrjälä <ville.syrjala@linux.intel.com> > > > > 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 <maarten.lankhorst@linux.intel.com> > > 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. > > BR, > Jani. > > > > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> > > --- > > drivers/gpu/drm/i915/intel_display.c | 12 +----------- > > 1 file changed, 1 insertion(+), 11 deletions(-) > > > > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c > > index e77ca7dfa44d..726ae191076b 100644 > > --- a/drivers/gpu/drm/i915/intel_display.c > > +++ b/drivers/gpu/drm/i915/intel_display.c > > @@ -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); > > } > > } > > -- > Jani Nikula, Intel Open Source Technology Center
On Wed, 08 Mar 2017, Ville Syrjälä <ville.syrjala@linux.intel.com> wrote: > On Wed, Mar 08, 2017 at 10:19:39AM +0200, Jani Nikula wrote: >> On Tue, 07 Mar 2017, ville.syrjala@linux.intel.com wrote: >> > From: Ville Syrjälä <ville.syrjala@linux.intel.com> >> > >> > 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 <maarten.lankhorst@linux.intel.com> >> > 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. Nice typo there! ;) But make it so. BR, Jani. > >> >> BR, >> Jani. >> >> >> > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> >> > --- >> > drivers/gpu/drm/i915/intel_display.c | 12 +----------- >> > 1 file changed, 1 insertion(+), 11 deletions(-) >> > >> > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c >> > index e77ca7dfa44d..726ae191076b 100644 >> > --- a/drivers/gpu/drm/i915/intel_display.c >> > +++ b/drivers/gpu/drm/i915/intel_display.c >> > @@ -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); >> > } >> > } >> >> -- >> Jani Nikula, Intel Open Source Technology Center
On Wed, Mar 08, 2017 at 12:20:53PM +0200, Ville Syrjälä wrote: > On Wed, Mar 08, 2017 at 10:19:39AM +0200, Jani Nikula wrote: > > On Tue, 07 Mar 2017, ville.syrjala@linux.intel.com wrote: > > > From: Ville Syrjälä <ville.syrjala@linux.intel.com> > > > > > > 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 <maarten.lankhorst@linux.intel.com> > > > 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. +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 > > > > > BR, > > Jani. > > > > > > > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> > > > --- > > > drivers/gpu/drm/i915/intel_display.c | 12 +----------- > > > 1 file changed, 1 insertion(+), 11 deletions(-) > > > > > > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c > > > index e77ca7dfa44d..726ae191076b 100644 > > > --- a/drivers/gpu/drm/i915/intel_display.c > > > +++ b/drivers/gpu/drm/i915/intel_display.c > > > @@ -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); > > > } > > > } > > > > -- > > Jani Nikula, Intel Open Source Technology Center > > -- > Ville Syrjälä > Intel OTC > _______________________________________________ > Intel-gfx mailing list > Intel-gfx@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
On Wed, Mar 08, 2017 at 10:36:15AM +0100, Maarten Lankhorst wrote: > Op 07-03-17 om 21:54 schreef ville.syrjala@linux.intel.com: > > From: Ville Syrjälä <ville.syrjala@linux.intel.com> > > > > 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 <maarten.lankhorst@linux.intel.com> > > Fixes: bfd16b2a23dc ("drm/i915: Make updating pipe without modeset atomic.") > > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> > > Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Pushed to dinq with cc:stable. Thanls for the review.
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index e77ca7dfa44d..726ae191076b 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -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); } }