From d21a3962d3042e6f56ad324cf18bdd64a1e6ecfa Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Tue, 21 Nov 2023 07:43:15 +0200 Subject: drm/i915: Call intel_pre_plane_updates() also for pipes getting enabled MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We used to call intel_pre_plane_updates() for any pipe going through a modeset whether the pipe was previously enabled or not. This in fact needed to apply all the necessary clock gating workarounds/etc. Restore the correct behaviour. Fixes: 39919997322f ("drm/i915: Disable all planes before modesetting any pipes") Reviewed-by: Jani Nikula Signed-off-by: Ville Syrjälä Link: https://patchwork.freedesktop.org/patch/msgid/20231121054324.9988-3-ville.syrjala@linux.intel.com (cherry picked from commit e0d5ce11ed0a21bb2bf328ad82fd261783c7ad88) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_display.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/i915/display/intel_display.c') diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index 28d85e1e858e..a2a806262c9e 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -6853,10 +6853,11 @@ static void intel_commit_modeset_disables(struct intel_atomic_state *state) if (!intel_crtc_needs_modeset(new_crtc_state)) continue; + intel_pre_plane_update(state, crtc); + if (!old_crtc_state->hw.active) continue; - intel_pre_plane_update(state, crtc); intel_crtc_disable_planes(state, crtc); } -- cgit v1.2.3 From 96d7e79401364c6e9a63af5f74f76792b03cb832 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Tue, 21 Nov 2023 07:43:14 +0200 Subject: drm/i915: Check pipe active state in {planes,vrr}_{enabling,disabling}() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit {planes,vrr}_{enabling,disabling}() are supposed to indicate whether the specific hardware feature is supposed to be enabling or disabling. That can only makes sense if the pipe is active overall. So check for that before we go poking at the hardware. I think we're semi-safe currently on due to: - intel_pre_plane_update() doesn't get called when the pipe was not-active prior to the commit, but this is actually a bug. This saves vrr_disabling(), and vrr_enabling() is called from deeper down where we have already checked hw.active. - active_planes mirrors the crtc's hw.active Reviewed-by: Jani Nikula Signed-off-by: Ville Syrjälä Link: https://patchwork.freedesktop.org/patch/msgid/20231121054324.9988-2-ville.syrjala@linux.intel.com (cherry picked from commit bc53c4d56eb24dbe56cd2c66ef4e9fc9393b1533) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_display.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'drivers/gpu/drm/i915/display/intel_display.c') diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index a2a806262c9e..94330108145e 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -906,12 +906,18 @@ static bool needs_async_flip_vtd_wa(const struct intel_crtc_state *crtc_state) static bool planes_enabling(const struct intel_crtc_state *old_crtc_state, const struct intel_crtc_state *new_crtc_state) { + if (!new_crtc_state->hw.active) + return false; + return is_enabling(active_planes, old_crtc_state, new_crtc_state); } static bool planes_disabling(const struct intel_crtc_state *old_crtc_state, const struct intel_crtc_state *new_crtc_state) { + if (!old_crtc_state->hw.active) + return false; + return is_disabling(active_planes, old_crtc_state, new_crtc_state); } @@ -928,6 +934,9 @@ static bool vrr_params_changed(const struct intel_crtc_state *old_crtc_state, static bool vrr_enabling(const struct intel_crtc_state *old_crtc_state, const struct intel_crtc_state *new_crtc_state) { + if (!new_crtc_state->hw.active) + return false; + return is_enabling(vrr.enable, old_crtc_state, new_crtc_state) || (new_crtc_state->vrr.enable && (new_crtc_state->update_m_n || new_crtc_state->update_lrr || @@ -937,6 +946,9 @@ static bool vrr_enabling(const struct intel_crtc_state *old_crtc_state, static bool vrr_disabling(const struct intel_crtc_state *old_crtc_state, const struct intel_crtc_state *new_crtc_state) { + if (!old_crtc_state->hw.active) + return false; + return is_disabling(vrr.enable, old_crtc_state, new_crtc_state) || (old_crtc_state->vrr.enable && (new_crtc_state->update_m_n || new_crtc_state->update_lrr || -- cgit v1.2.3 From 20c2dbff342aec13bf93c2f6c951da198916a455 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Mon, 27 Nov 2023 16:50:25 +0200 Subject: drm/i915: Skip some timing checks on BXT/GLK DSI transcoders MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Apparently some BXT/GLK systems have DSI panels whose timings don't agree with the normal cpu transcoder hblank>=32 limitation. This is perhaps fine as there are no specific hblank/etc. limits listed for the BXT/GLK DSI transcoders. Move those checks out from the global intel_mode_valid() into into connector specific .mode_valid() hooks, skipping BXT/GLK DSI connectors. We'll leave the basic [hv]display/[hv]total checks in intel_mode_valid() as those seem like sensible upper limits regardless of the transcoder used. Cc: stable@vger.kernel.org Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9720 Fixes: 8f4b1068e7fc ("drm/i915: Check some transcoder timing minimum limits") Signed-off-by: Ville Syrjälä Link: https://patchwork.freedesktop.org/patch/msgid/20231127145028.4899-1-ville.syrjala@linux.intel.com Reviewed-by: Jani Nikula (cherry picked from commit e0ef2daa8ca8ce4dbc2fd0959e383b753a87fd7d) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_display.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'drivers/gpu/drm/i915/display/intel_display.c') diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index 94330108145e..4ee7d569b379 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -7869,6 +7869,16 @@ enum drm_mode_status intel_mode_valid(struct drm_device *dev, mode->vtotal > vtotal_max) return MODE_V_ILLEGAL; + return MODE_OK; +} + +enum drm_mode_status intel_cpu_transcoder_mode_valid(struct drm_i915_private *dev_priv, + const struct drm_display_mode *mode) +{ + /* + * Additional transcoder timing limits, + * excluding BXT/GLK DSI transcoders. + */ if (DISPLAY_VER(dev_priv) >= 5) { if (mode->hdisplay < 64 || mode->htotal - mode->hdisplay < 32) -- cgit v1.2.3 From e0f04e41e8eedd4e5a1275f2318df7e1841855f2 Mon Sep 17 00:00:00 2001 From: Thomas Zimmermann Date: Mon, 4 Dec 2023 09:32:33 +0100 Subject: drm/atomic-helpers: Invoke end_fb_access while owning plane state Invoke drm_plane_helper_funcs.end_fb_access before drm_atomic_helper_commit_hw_done(). The latter function hands over ownership of the plane state to the following commit, which might free it. Releasing resources in end_fb_access then operates on undefined state. This bug has been observed with non-blocking commits when they are being queued up quickly. Here is an example stack trace from the bug report. The plane state has been free'd already, so the pages for drm_gem_fb_vunmap() are gone. Unable to handle kernel paging request at virtual address 0000000100000049 [...] drm_gem_fb_vunmap+0x18/0x74 drm_gem_end_shadow_fb_access+0x1c/0x2c drm_atomic_helper_cleanup_planes+0x58/0xd8 drm_atomic_helper_commit_tail+0x90/0xa0 commit_tail+0x15c/0x188 commit_work+0x14/0x20 Fix this by running end_fb_access immediately after updating all planes in drm_atomic_helper_commit_planes(). The existing clean-up helper drm_atomic_helper_cleanup_planes() now only handles cleanup_fb. For aborted commits, roll back from drm_atomic_helper_prepare_planes() in the new helper drm_atomic_helper_unprepare_planes(). This case is different from regular cleanup, as we have to release the new state; regular cleanup releases the old state. The new helper also invokes cleanup_fb for all planes. The changes mostly involve DRM's atomic helpers. Only two drivers, i915 and nouveau, implement their own commit function. Update them to invoke drm_atomic_helper_unprepare_planes(). Drivers with custom commit_tail function do not require changes. v4: * fix documentation (kernel test robot) v3: * add drm_atomic_helper_unprepare_planes() for rolling back * use correct state for end_fb_access v2: * fix test in drm_atomic_helper_cleanup_planes() Reported-by: Alyssa Ross Closes: https://lore.kernel.org/dri-devel/87leazm0ya.fsf@alyssa.is/ Suggested-by: Daniel Vetter Fixes: 94d879eaf7fb ("drm/atomic-helper: Add {begin,end}_fb_access to plane helpers") Tested-by: Alyssa Ross Reviewed-by: Alyssa Ross Signed-off-by: Thomas Zimmermann Cc: # v6.2+ Link: https://patchwork.freedesktop.org/patch/msgid/20231204083247.22006-1-tzimmermann@suse.de --- drivers/gpu/drm/i915/display/intel_display.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/i915/display/intel_display.c') diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index 28d85e1e858e..02f873738905 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -7475,7 +7475,7 @@ int intel_atomic_commit(struct drm_device *dev, struct drm_atomic_state *_state, for_each_new_intel_crtc_in_state(state, crtc, new_crtc_state, i) intel_color_cleanup_commit(new_crtc_state); - drm_atomic_helper_cleanup_planes(dev, &state->base); + drm_atomic_helper_unprepare_planes(dev, &state->base); intel_runtime_pm_put(&dev_priv->runtime_pm, state->wakeref); return ret; } -- cgit v1.2.3