drm/i915: Don't check modeset state in the hw state force restore path
Subsystems (2)
PATCH v2 1/3 drm/i915: Don't check modeset state in the hw state force restore path
Since the force restore logic will restore the CRTCs state one at a time, it is possible that the state will be inconsistent until the whole operation finishes. A call to intel_modeset_check_state() is done once it's over, so don't check the state multiple times in between. This regression was introduced in:
commit 7f27126ea3db6ade886f18fd39caf0ff0cd1d37f
Author: Jesse Barnes <[email protected]> Date: Wed Nov 5 14:26:06 2014 -0800
drm/i915: factor out compute_config from __intel_set_mode v3
v2: Rename check parameter to force_restore. (Matt)
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94431
--- drivers/gpu/drm/i915/intel_display.c | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-)
drivers/gpu/drm/i915/intel_display.c +12 -9
@@ -87,7 +87,8 @@ static void ironlake_pch_clock_get(struct intel_crtc *crtc, struct intel_crtc_state *pipe_config);static int intel_set_mode (struct drm_crtc *crtc, -struct drm_atomic_state *state); +struct drm_atomic_state *state, +bool force_restore); static int intel_framebuffer_init(struct drm_device *dev, struct intel_framebuffer *ifb, struct drm_mode_fb_cmd2 *mode_cmd,
@@ -10096,7 +10097,7 @@ retry:drm_mode_copy (&crtc_state->base.mode, mode); -if (intel_set_mode (crtc, state)) { +if (intel_set_mode (crtc, state, true)) {DRM_DEBUG_KMS ("failed to set mode on load-detect pipe\n" );if (old->release_fb) old->release_fb->funcs->destroy (old->release_fb);
@@ -10170,7 +10171,7 @@ void intel_release_load_detect_pipe(struct drm_connector *connector,if (ret) goto fail; - ret =intel_set_mode (crtc, state); + ret =intel_set_mode (crtc, state, true);if (ret) goto fail;
@@ -12646,20 +12647,22 @@ static int __intel_set_mode(struct drm_crtc *modeset_crtc, }static int intel_set_mode_with_config (struct drm_crtc *crtc, -struct intel_crtc_state *pipe_config) +struct intel_crtc_state *pipe_config, +bool force_restore) {int ret; ret =__intel_set_mode (crtc, pipe_config); -if (ret ==0 ) +if (ret ==0 && force_restore)intel_modeset_check_state (crtc->dev);return ret; }static int intel_set_mode (struct drm_crtc *crtc, -struct drm_atomic_state *state) +struct drm_atomic_state *state, +bool force_restore) {struct intel_crtc_state *pipe_config;int ret =0 ;
@@ -12670,7 +12673,7 @@ static int intel_set_mode(struct drm_crtc *crtc, goto out; } - ret =intel_set_mode_with_config (crtc, pipe_config); + ret =intel_set_mode_with_config (crtc, pipe_config, force_restore);if (ret) goto out;
@@ -12747,7 +12750,7 @@ void intel_crtc_restore_mode(struct drm_crtc *crtc)intel_modeset_setup_plane_state (state, crtc, &crtc->mode, crtc->primary->fb, crtc->x, crtc->y); - ret =intel_set_mode (crtc, state); + ret =intel_set_mode (crtc, state, false);if (ret)drm_atomic_state_free (state); }
@@ -12947,7 +12950,7 @@ static int intel_crtc_set_config(struct drm_mode_set *set) primary_plane_was_visible =primary_plane_visible (set->crtc); - ret =intel_set_mode_with_config (set->crtc, pipe_config); + ret =intel_set_mode_with_config (set->crtc, pipe_config, true); if (ret == 0 && pipe_config->base.enable &&
Signature · 6 lines
2.1.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 2/3 drm/i915: Don't update staged config during force restore modesets
The force restore path relies on the staged config to preserve the configuration used before a suspend/resume cycle. The update done to it in intel_modeset_fixup_state() would cause that information to be lost after the first modeset, making it impossible to restore the modes for pipes B and C.
References: https://bugs.freedesktop.org/show_bug.cgi?id=90468
--- drivers/gpu/drm/i915/intel_display.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)
drivers/gpu/drm/i915/intel_display.c +3 -5
@@ -11386,10 +11386,6 @@ static void intel_modeset_fixup_state(struct drm_atomic_state *state) crtc->base.enabled = crtc->base.state->enable; crtc->config =to_intel_crtc_state (crtc->base.state); } - -/* Copy the new configuration to the staged state, to keep the few -* pieces of code that haven't been converted yet happy */ -intel_modeset_update_staged_output_state (state->dev); } static void
@@ -12654,8 +12650,10 @@ static int intel_set_mode_with_config(struct drm_crtc *crtc, ret =__intel_set_mode (crtc, pipe_config); -if (ret ==0 && force_restore) +if (ret ==0 && force_restore) { +intel_modeset_update_staged_output_state (crtc->dev);intel_modeset_check_state (crtc->dev); + }return ret; }
Signature · 6 lines
2.1.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 3/3 drm/i915: Don't set enabled value of all CRTCs when restoring the mode
In reply to Ander Conselvan de OliveiraThe code in intel_crtc_restore_mode() sets the enabled value of all the CRTCs when restoring the mode after a suspend/resume cycle. When more than one CRTC is enabled, that causes drm_atomic_helper_check_modeset() to fail if there is more than one pipe enabled, since all but one CRTC has valid connector data. Instead, set only the enabled value for the CRTC passed as an argument.
v2: Don't leak atomic state. (Matt)
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90468
References: https://bugs.freedesktop.org/show_bug.cgi?id=90396
--- drivers/gpu/drm/i915/intel_display.c | 27 ++++++++++----------------- 1 file changed, 10 insertions(+), 17 deletions(-)
drivers/gpu/drm/i915/intel_display.c +10 -17
@@ -12683,7 +12683,6 @@ void intel_crtc_restore_mode(struct drm_crtc *crtc) {struct drm_device *dev = crtc->dev;struct drm_atomic_state *state; -struct intel_crtc *intel_crtc;struct intel_encoder *encoder;struct intel_connector *connector;struct drm_connector_state *connector_state;
@@ -12726,24 +12725,18 @@ void intel_crtc_restore_mode(struct drm_crtc *crtc) } } -for_each_intel_crtc (dev, intel_crtc) { -if (intel_crtc->new_enabled == intel_crtc->base.enabled) -continue ; - - crtc_state =intel_atomic_get_crtc_state (state, intel_crtc); -if (IS_ERR (crtc_state)) { -DRM_DEBUG_KMS ("Failed to add [CRTC:%d] to state: %ld\n" , - intel_crtc->base.base.id, -PTR_ERR (crtc_state)); -continue ; - } + crtc_state =intel_atomic_get_crtc_state (state,to_intel_crtc (crtc)); +if (IS_ERR (crtc_state)) { +DRM_DEBUG_KMS ("Failed to add [CRTC:%d] to state: %ld\n" , + crtc->base.id,PTR_ERR (crtc_state)); +drm_atomic_state_free (state); +return ; + } - crtc_state->base.active = crtc_state->base.enable = - intel_crtc->new_enabled; + crtc_state->base.active = crtc_state->base.enable = +to_intel_crtc (crtc)->new_enabled; -if (&intel_crtc->base == crtc) -drm_mode_copy (&crtc_state->base.mode, &crtc->mode); - } +drm_mode_copy (&crtc_state->base.mode, &crtc->mode);intel_modeset_setup_plane_state (state, crtc, &crtc->mode, crtc->primary->fb, crtc->x, crtc->y);
Signature · 6 lines
2.1.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerQuoted 10 lines · Ander Conselvan de Oliveira
Since the force restore logic will restore the CRTCs state one at a time, it is possible that the state will be inconsistent until the whole operation finishes. A call to
intel_modeset_check_state()is done once it's over, so don't check the state multiple times in between. This regression was introduced in:commit
7f27126ea3db6ade886f18fd39caf0ff0cd1d37fAuthor: Jesse Barnes <[email protected]> Date: Wed Nov 5 14:26:06 2014 -0800
Quoted 8 lines
drm/i915: factor out
compute_configfrom__intel_set_modev3v2: Rename check parameter to
force_restore. (Matt)Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94431
Cc Jesse Barnes <[email protected]>Signed-off-by Ander Conselvan de Oliveira <[email protected]>Reviewed-by Matt Roper <[email protected]>
All three patches applied to drm-intel-next-fixes, aiming for v4.2 merge window. Thanks for the patches and review.
For drm-intel-nightly, I resolved the conflicts by ignoring these changes and favoring what's in drm-intel-next-queued. Fingers crossed I didn't botch it up!
BR,
Jani.
Quoted 96 lines
--- drivers/gpu/drm/i915/intel_display.c | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-)drivers/gpu/drm/i915/intel_display.c +12
-9@@ -87,7 +87,8 @@ static void ironlake_pch_clock_get(struct intel_crtc *crtc, struct intel_crtc_state *pipe_config);static int intel_set_mode (struct drm_crtc *crtc, -struct drm_atomic_state *state); +struct drm_atomic_state *state, +bool force_restore); static int intel_framebuffer_init(struct drm_device *dev, struct intel_framebuffer *ifb, struct drm_mode_fb_cmd2 *mode_cmd,@@ -10096,7 +10097,7 @@ retry:drm_mode_copy (&crtc_state->base.mode, mode); -if (intel_set_mode (crtc, state)) { +if (intel_set_mode (crtc, state, true)) {DRM_DEBUG_KMS ("failed to set mode on load-detect pipe\n" );if (old->release_fb) old->release_fb->funcs->destroy (old->release_fb);@@ -10170,7 +10171,7 @@ void intel_release_load_detect_pipe(struct drm_connector *connector,if (ret) goto fail; - ret =intel_set_mode (crtc, state); + ret =intel_set_mode (crtc, state, true);if (ret) goto fail;@@ -12646,20 +12647,22 @@ static int __intel_set_mode(struct drm_crtc *modeset_crtc, }static int intel_set_mode_with_config (struct drm_crtc *crtc, -struct intel_crtc_state *pipe_config) +struct intel_crtc_state *pipe_config, +bool force_restore) {int ret; ret =__intel_set_mode (crtc, pipe_config); -if (ret ==0 ) +if (ret ==0 && force_restore)intel_modeset_check_state (crtc->dev);return ret; }static int intel_set_mode (struct drm_crtc *crtc, -struct drm_atomic_state *state) +struct drm_atomic_state *state, +bool force_restore) {struct intel_crtc_state *pipe_config;int ret =0 ;@@ -12670,7 +12673,7 @@ static int intel_set_mode(struct drm_crtc *crtc, goto out; } - ret =intel_set_mode_with_config (crtc, pipe_config); + ret =intel_set_mode_with_config (crtc, pipe_config, force_restore);if (ret) goto out;@@ -12747,7 +12750,7 @@ void intel_crtc_restore_mode(struct drm_crtc *crtc)intel_modeset_setup_plane_state (state, crtc, &crtc->mode, crtc->primary->fb, crtc->x, crtc->y); - ret =intel_set_mode (crtc, state); + ret =intel_set_mode (crtc, state, false);if (ret)drm_atomic_state_free (state); }@@ -12947,7 +12950,7 @@ static int intel_crtc_set_config(struct drm_mode_set *set) primary_plane_was_visible =primary_plane_visible (set->crtc); - ret =intel_set_mode_with_config (set->crtc, pipe_config); + ret =intel_set_mode_with_config (set->crtc, pipe_config, true); if (ret == 0 && pipe_config->base.enable &&Signature · 6 lines
2.1.0 _______________________________________________ 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
Jani Nikula
maintainerQuoted 20 lines · Jani Nikula
Quoted 10 lines · Ander Conselvan de Oliveira
Since the force restore logic will restore the CRTCs state one at a time, it is possible that the state will be inconsistent until the whole operation finishes. A call to
intel_modeset_check_state()is done once it's over, so don't check the state multiple times in between. This regression was introduced in:commit
7f27126ea3db6ade886f18fd39caf0ff0cd1d37fAuthor: Jesse Barnes <[email protected]> Date: Wed Nov 5 14:26:06 2014 -0800Quoted 8 lines
drm/i915: factor out
compute_configfrom__intel_set_modev3v2: Rename check parameter to
force_restore. (Matt)Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94431
Cc Jesse Barnes <[email protected]>Signed-off-by Ander Conselvan de Oliveira <[email protected]>Reviewed-by Matt Roper <[email protected]>
Quoted 6 lines
All three patches applied to drm-intel-next-fixes, aiming for v4.2 merge window. Thanks for the patches and review.
For drm-intel-nightly, I resolved the conflicts by ignoring these changes and favoring what's in drm-intel-next-queued. Fingers crossed I didn't botch it up!
Also, do we need some of these in v4.1 or earlier as well? Should we backport once these are upstream?
BR,
Jani.
Quoted 105 lines
BR,
Jani.Quoted 96 lines
--- drivers/gpu/drm/i915/intel_display.c | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-)drivers/gpu/drm/i915/intel_display.c +12
-9@@ -87,7 +87,8 @@ static void ironlake_pch_clock_get(struct intel_crtc *crtc, struct intel_crtc_state *pipe_config);static int intel_set_mode (struct drm_crtc *crtc, -struct drm_atomic_state *state); +struct drm_atomic_state *state, +bool force_restore); static int intel_framebuffer_init(struct drm_device *dev, struct intel_framebuffer *ifb, struct drm_mode_fb_cmd2 *mode_cmd,@@ -10096,7 +10097,7 @@ retry:drm_mode_copy (&crtc_state->base.mode, mode); -if (intel_set_mode (crtc, state)) { +if (intel_set_mode (crtc, state, true)) {DRM_DEBUG_KMS ("failed to set mode on load-detect pipe\n" );if (old->release_fb) old->release_fb->funcs->destroy (old->release_fb);@@ -10170,7 +10171,7 @@ void intel_release_load_detect_pipe(struct drm_connector *connector,if (ret) goto fail; - ret =intel_set_mode (crtc, state); + ret =intel_set_mode (crtc, state, true);if (ret) goto fail;@@ -12646,20 +12647,22 @@ static int __intel_set_mode(struct drm_crtc *modeset_crtc, }static int intel_set_mode_with_config (struct drm_crtc *crtc, -struct intel_crtc_state *pipe_config) +struct intel_crtc_state *pipe_config, +bool force_restore) {int ret; ret =__intel_set_mode (crtc, pipe_config); -if (ret ==0 ) +if (ret ==0 && force_restore)intel_modeset_check_state (crtc->dev);return ret; }static int intel_set_mode (struct drm_crtc *crtc, -struct drm_atomic_state *state) +struct drm_atomic_state *state, +bool force_restore) {struct intel_crtc_state *pipe_config;int ret =0 ;@@ -12670,7 +12673,7 @@ static int intel_set_mode(struct drm_crtc *crtc, goto out; } - ret =intel_set_mode_with_config (crtc, pipe_config); + ret =intel_set_mode_with_config (crtc, pipe_config, force_restore);if (ret) goto out;@@ -12747,7 +12750,7 @@ void intel_crtc_restore_mode(struct drm_crtc *crtc)intel_modeset_setup_plane_state (state, crtc, &crtc->mode, crtc->primary->fb, crtc->x, crtc->y); - ret =intel_set_mode (crtc, state); + ret =intel_set_mode (crtc, state, false);if (ret)drm_atomic_state_free (state); }@@ -12947,7 +12950,7 @@ static int intel_crtc_set_config(struct drm_mode_set *set) primary_plane_was_visible =primary_plane_visible (set->crtc); - ret =intel_set_mode_with_config (set->crtc, pipe_config); + ret =intel_set_mode_with_config (set->crtc, pipe_config, true); if (ret == 0 && pipe_config->base.enable &&Signature · 6 lines
2.1.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfxSignature · 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] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 24 lines · Jani Nikula
Quoted 23 lines · Jani Nikula
Quoted 10 lines · Ander Conselvan de Oliveira
Since the force restore logic will restore the CRTCs state one at a time, it is possible that the state will be inconsistent until the whole operation finishes. A call to
intel_modeset_check_state()is done once it's over, so don't check the state multiple times in between. This regression was introduced in:commit
7f27126ea3db6ade886f18fd39caf0ff0cd1d37fAuthor: Jesse Barnes <[email protected]> Date: Wed Nov 5 14:26:06 2014 -0800Quoted 8 lines
drm/i915: factor out
compute_configfrom__intel_set_modev3v2: Rename check parameter to
force_restore. (Matt)Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94431
Cc Jesse Barnes <[email protected]>Signed-off-by Ander Conselvan de Oliveira <[email protected]>Reviewed-by Matt Roper <[email protected]>All three patches applied to drm-intel-next-fixes, aiming for v4.2 merge window. Thanks for the patches and review.
Quoted 6 lines
Quoted 3 lines
For drm-intel-nightly, I resolved the conflicts by ignoring these changes and favoring what's in drm-intel-next-queued. Fingers crossed I didn't botch it up!
Also, do we need some of these in v4.1 or earlier as well? Should we backport once these are upstream?
The regression 1/3 solves goes back to v3.19. The other patches are only relevant for 4.2 and later.
Thanks,
Ander
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 20 lines · Jani Nikula
Quoted 10 lines · Ander Conselvan de Oliveira
Since the force restore logic will restore the CRTCs state one at a time, it is possible that the state will be inconsistent until the whole operation finishes. A call to
intel_modeset_check_state()is done once it's over, so don't check the state multiple times in between. This regression was introduced in:commit
7f27126ea3db6ade886f18fd39caf0ff0cd1d37fAuthor: Jesse Barnes <[email protected]> Date: Wed Nov 5 14:26:06 2014 -0800Quoted 8 lines
drm/i915: factor out
compute_configfrom__intel_set_modev3v2: Rename check parameter to
force_restore. (Matt)Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94431
Cc Jesse Barnes <[email protected]>Signed-off-by Ander Conselvan de Oliveira <[email protected]>Reviewed-by Matt Roper <[email protected]>
Quoted 6 lines
All three patches applied to drm-intel-next-fixes, aiming for v4.2 merge window. Thanks for the patches and review.
For drm-intel-nightly, I resolved the conflicts by ignoring these changes and favoring what's in drm-intel-next-queued. Fingers crossed I didn't botch it up!
In the end, that means only the content of 3/3 is in -nightly. It won't fix any of the issues, but shouldn't cause any problems.
Ander
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerQuoted 24 lines · Ander Conselvan De Oliveira
Quoted 23 lines · Jani Nikula
Quoted 10 lines · Ander Conselvan de Oliveira
Since the force restore logic will restore the CRTCs state one at a time, it is possible that the state will be inconsistent until the whole operation finishes. A call to
intel_modeset_check_state()is done once it's over, so don't check the state multiple times in between. This regression was introduced in:commit
7f27126ea3db6ade886f18fd39caf0ff0cd1d37fAuthor: Jesse Barnes <[email protected]> Date: Wed Nov 5 14:26:06 2014 -0800Quoted 8 lines
drm/i915: factor out
compute_configfrom__intel_set_modev3v2: Rename check parameter to
force_restore. (Matt)Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94431
Cc Jesse Barnes <[email protected]>Signed-off-by Ander Conselvan de Oliveira <[email protected]>Reviewed-by Matt Roper <[email protected]>All three patches applied to drm-intel-next-fixes, aiming for v4.2 merge window. Thanks for the patches and review.
Quoted 6 lines
Quoted 3 lines
For drm-intel-nightly, I resolved the conflicts by ignoring these changes and favoring what's in drm-intel-next-queued. Fingers crossed I didn't botch it up!
In the end, that means only the content of 3/3 is in -nightly. It won't fix any of the issues, but shouldn't cause any problems.
But it is expected that Maarten's atomic work will eventually fix this, right?
BR,
Jani.
Quoted 4 lines
Ander
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 29 lines · Jani Nikula
Quoted 28 lines · Ander Conselvan De Oliveira
Quoted 26 lines · Jani Nikula
Quoted 10 lines · Ander Conselvan de Oliveira
Since the force restore logic will restore the CRTCs state one at a time, it is possible that the state will be inconsistent until the whole operation finishes. A call to
intel_modeset_check_state()is done once it's over, so don't check the state multiple times in between. This regression was introduced in:commit
7f27126ea3db6ade886f18fd39caf0ff0cd1d37fAuthor: Jesse Barnes <[email protected]> Date: Wed Nov 5 14:26:06 2014 -0800Quoted 8 lines
drm/i915: factor out
compute_configfrom__intel_set_modev3v2: Rename check parameter to
force_restore. (Matt)Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94431
Cc Jesse Barnes <[email protected]>Signed-off-by Ander Conselvan de Oliveira <[email protected]>Reviewed-by Matt Roper <[email protected]>All three patches applied to drm-intel-next-fixes, aiming for v4.2 merge window. Thanks for the patches and review.
For drm-intel-nightly, I resolved the conflicts by ignoring these changes and favoring what's in drm-intel-next-queued. Fingers crossed I didn't botch it up!
Quoted 5 lines
Quoted 2 lines
In the end, that means only the content of 3/3 is in -nightly. It won't fix any of the issues, but shouldn't cause any problems.
But it is expected that Maarten's atomic work will eventually fix this, right?
Yes, that's correct.
Ander
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx