From 411efa18e4b03840553ff58ad9b4621b82a30c04 Mon Sep 17 00:00:00 2001 From: Maxime Ripard Date: Tue, 25 May 2021 11:10:58 +0200 Subject: [PATCH 01/15] drm/vc4: hdmi: Move the HSM clock enable to runtime_pm In order to access the HDMI controller, we need to make sure the HSM clock is enabled. If we were to access it with the clock disabled, the CPU would completely hang, resulting in an hard crash. Since we have different code path that would require it, let's move that clock enable / disable to runtime_pm that will take care of the reference counting for us. Fixes: 4f6e3d66ac52 ("drm/vc4: Add runtime PM support to the HDMI encoder driver") Signed-off-by: Maxime Ripard Reviewed-by: Dave Stevenson Link: https://patchwork.freedesktop.org/patch/msgid/20210525091059.234116-3-maxime@cerno.tech --- drivers/gpu/drm/vc4/vc4_hdmi.c | 40 +++++++++++++++++++++++++--------- 1 file changed, 30 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c index 1fda574579af..84e218365045 100644 --- a/drivers/gpu/drm/vc4/vc4_hdmi.c +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c @@ -473,7 +473,6 @@ static void vc4_hdmi_encoder_post_crtc_powerdown(struct drm_encoder *encoder, HDMI_READ(HDMI_VID_CTL) & ~VC4_HD_VID_CTL_ENABLE); clk_disable_unprepare(vc4_hdmi->pixel_bvb_clock); - clk_disable_unprepare(vc4_hdmi->hsm_clock); clk_disable_unprepare(vc4_hdmi->pixel_clock); ret = pm_runtime_put(&vc4_hdmi->pdev->dev); @@ -784,13 +783,6 @@ static void vc4_hdmi_encoder_pre_crtc_configure(struct drm_encoder *encoder, return; } - ret = clk_prepare_enable(vc4_hdmi->hsm_clock); - if (ret) { - DRM_ERROR("Failed to turn on HSM clock: %d\n", ret); - clk_disable_unprepare(vc4_hdmi->pixel_clock); - return; - } - vc4_hdmi_cec_update_clk_div(vc4_hdmi); /* @@ -801,7 +793,6 @@ static void vc4_hdmi_encoder_pre_crtc_configure(struct drm_encoder *encoder, (hsm_rate > VC4_HSM_MID_CLOCK ? 150000000 : 75000000)); if (ret) { DRM_ERROR("Failed to set pixel bvb clock rate: %d\n", ret); - clk_disable_unprepare(vc4_hdmi->hsm_clock); clk_disable_unprepare(vc4_hdmi->pixel_clock); return; } @@ -809,7 +800,6 @@ static void vc4_hdmi_encoder_pre_crtc_configure(struct drm_encoder *encoder, ret = clk_prepare_enable(vc4_hdmi->pixel_bvb_clock); if (ret) { DRM_ERROR("Failed to turn on pixel bvb clock: %d\n", ret); - clk_disable_unprepare(vc4_hdmi->hsm_clock); clk_disable_unprepare(vc4_hdmi->pixel_clock); return; } @@ -1929,6 +1919,29 @@ static int vc5_hdmi_init_resources(struct vc4_hdmi *vc4_hdmi) return 0; } +#ifdef CONFIG_PM +static int vc4_hdmi_runtime_suspend(struct device *dev) +{ + struct vc4_hdmi *vc4_hdmi = dev_get_drvdata(dev); + + clk_disable_unprepare(vc4_hdmi->hsm_clock); + + return 0; +} + +static int vc4_hdmi_runtime_resume(struct device *dev) +{ + struct vc4_hdmi *vc4_hdmi = dev_get_drvdata(dev); + int ret; + + ret = clk_prepare_enable(vc4_hdmi->hsm_clock); + if (ret) + return ret; + + return 0; +} +#endif + static int vc4_hdmi_bind(struct device *dev, struct device *master, void *data) { const struct vc4_hdmi_variant *variant = of_device_get_match_data(dev); @@ -2165,11 +2178,18 @@ static const struct of_device_id vc4_hdmi_dt_match[] = { {} }; +static const struct dev_pm_ops vc4_hdmi_pm_ops = { + SET_RUNTIME_PM_OPS(vc4_hdmi_runtime_suspend, + vc4_hdmi_runtime_resume, + NULL) +}; + struct platform_driver vc4_hdmi_driver = { .probe = vc4_hdmi_dev_probe, .remove = vc4_hdmi_dev_remove, .driver = { .name = "vc4_hdmi", .of_match_table = vc4_hdmi_dt_match, + .pm = &vc4_hdmi_pm_ops, }, }; From 9984d6664ce9dcbbc713962539eaf7636ea246c2 Mon Sep 17 00:00:00 2001 From: Maxime Ripard Date: Tue, 25 May 2021 11:10:59 +0200 Subject: [PATCH 02/15] drm/vc4: hdmi: Make sure the controller is powered in detect If the HPD GPIO is not available and drm_probe_ddc fails, we end up reading the HDMI_HOTPLUG register, but the controller might be powered off resulting in a CPU hang. Make sure we have the power domain and the HSM clock powered during the detect cycle to prevent the hang from happening. Fixes: 4f6e3d66ac52 ("drm/vc4: Add runtime PM support to the HDMI encoder driver") Signed-off-by: Maxime Ripard Reviewed-by: Dave Stevenson Link: https://patchwork.freedesktop.org/patch/msgid/20210525091059.234116-4-maxime@cerno.tech --- drivers/gpu/drm/vc4/vc4_hdmi.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c index 84e218365045..8106b5634fe1 100644 --- a/drivers/gpu/drm/vc4/vc4_hdmi.c +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c @@ -159,6 +159,8 @@ vc4_hdmi_connector_detect(struct drm_connector *connector, bool force) struct vc4_hdmi *vc4_hdmi = connector_to_vc4_hdmi(connector); bool connected = false; + WARN_ON(pm_runtime_resume_and_get(&vc4_hdmi->pdev->dev)); + if (vc4_hdmi->hpd_gpio) { if (gpio_get_value_cansleep(vc4_hdmi->hpd_gpio) ^ vc4_hdmi->hpd_active_low) @@ -180,10 +182,12 @@ vc4_hdmi_connector_detect(struct drm_connector *connector, bool force) } } + pm_runtime_put(&vc4_hdmi->pdev->dev); return connector_status_connected; } cec_phys_addr_invalidate(vc4_hdmi->cec_adap); + pm_runtime_put(&vc4_hdmi->pdev->dev); return connector_status_disconnected; } From e484028bf39c0c87c499dc782dc9cd3bb72c0ab5 Mon Sep 17 00:00:00 2001 From: Dan Sneddon Date: Wed, 2 Jun 2021 09:08:45 -0700 Subject: [PATCH 03/15] drm: atmel_hlcdc: Enable the crtc vblank prior to crtc usage. 'commit eec44d44a3d2 ("drm/atmel: Use drm_atomic_helper_commit")' removed the home-grown handling of atomic commits and exposed an issue in the crtc atomic commit handling where vblank is expected to be enabled but hasn't yet, causing kernel warnings during boot. This patch cleans up the crtc vblank handling thus removing the warning on boot. Fixes: eec44d44a3d2 ("drm/atmel: Use drm_atomic_helper_commit") Signed-off-by: Dan Sneddon Signed-off-by: Sam Ravnborg Link: https://patchwork.freedesktop.org/patch/msgid/20210602160846.5013-1-dan.sneddon@microchip.com --- drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_crtc.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_crtc.c b/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_crtc.c index 05ad75d155e8..cfe4fc69277e 100644 --- a/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_crtc.c +++ b/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_crtc.c @@ -232,7 +232,6 @@ static void atmel_hlcdc_crtc_atomic_enable(struct drm_crtc *c, pm_runtime_put_sync(dev->dev); - drm_crtc_vblank_on(c); } #define ATMEL_HLCDC_RGB444_OUTPUT BIT(0) @@ -343,8 +342,17 @@ static int atmel_hlcdc_crtc_atomic_check(struct drm_crtc *c, static void atmel_hlcdc_crtc_atomic_begin(struct drm_crtc *c, struct drm_atomic_state *state) +{ + drm_crtc_vblank_on(c); +} + +static void atmel_hlcdc_crtc_atomic_flush(struct drm_crtc *c, + struct drm_atomic_state *state) { struct atmel_hlcdc_crtc *crtc = drm_crtc_to_atmel_hlcdc_crtc(c); + unsigned long flags; + + spin_lock_irqsave(&c->dev->event_lock, flags); if (c->state->event) { c->state->event->pipe = drm_crtc_index(c); @@ -354,12 +362,7 @@ static void atmel_hlcdc_crtc_atomic_begin(struct drm_crtc *c, crtc->event = c->state->event; c->state->event = NULL; } -} - -static void atmel_hlcdc_crtc_atomic_flush(struct drm_crtc *crtc, - struct drm_atomic_state *state) -{ - /* TODO: write common plane control register if available */ + spin_unlock_irqrestore(&c->dev->event_lock, flags); } static const struct drm_crtc_helper_funcs lcdc_crtc_helper_funcs = { From af42167f53ec18b0856387fc119b28c8c1ba98a1 Mon Sep 17 00:00:00 2001 From: Krzysztof Kozlowski Date: Wed, 26 May 2021 08:30:02 -0400 Subject: [PATCH 04/15] drm/panel: ld9040: reference spi_device_id table MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reference the spi_device_id table to silence W=1 warning: drivers/gpu/drm/panel/panel-samsung-ld9040.c:377:35: warning: ‘ld9040_ids’ defined but not used [-Wunused-const-variable=] This also would be needed for matching the driver if booted without CONFIG_OF (although it's not necessarily real case). Signed-off-by: Krzysztof Kozlowski Signed-off-by: Sam Ravnborg Link: https://patchwork.freedesktop.org/patch/msgid/20210526123002.12913-1-krzysztof.kozlowski@canonical.com --- drivers/gpu/drm/panel/panel-samsung-ld9040.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/panel/panel-samsung-ld9040.c b/drivers/gpu/drm/panel/panel-samsung-ld9040.c index f484147fc3a6..c4b388850a13 100644 --- a/drivers/gpu/drm/panel/panel-samsung-ld9040.c +++ b/drivers/gpu/drm/panel/panel-samsung-ld9040.c @@ -383,6 +383,7 @@ MODULE_DEVICE_TABLE(spi, ld9040_ids); static struct spi_driver ld9040_driver = { .probe = ld9040_probe, .remove = ld9040_remove, + .id_table = ld9040_ids, .driver = { .name = "panel-samsung-ld9040", .of_match_table = ld9040_of_match, From e541845ae0858616c52dd97df4bf91568c7a7a1b Mon Sep 17 00:00:00 2001 From: Dan Sneddon Date: Tue, 30 Mar 2021 08:17:20 -0700 Subject: [PATCH 05/15] drm/atmel-hlcdc: Allow async page flips The driver is capable of doing async page flips so we need to tell the core to allow them. Signed-off-by: Dan Sneddon Tested-by: Ludovic Desroches Signed-off-by: Sam Ravnborg Link: https://patchwork.freedesktop.org/patch/msgid/20210330151721.6616-1-dan.sneddon@microchip.com --- drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_dc.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_dc.c b/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_dc.c index 65af56e47129..f09b6dd8754c 100644 --- a/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_dc.c +++ b/drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_dc.c @@ -593,6 +593,7 @@ static int atmel_hlcdc_dc_modeset_init(struct drm_device *dev) dev->mode_config.max_width = dc->desc->max_width; dev->mode_config.max_height = dc->desc->max_height; dev->mode_config.funcs = &mode_config_funcs; + dev->mode_config.async_page_flip = true; return 0; } From 1815d9c86e3090477fbde066ff314a7e9721ee0f Mon Sep 17 00:00:00 2001 From: Desmond Cheong Zhi Xi Date: Sun, 20 Jun 2021 19:03:26 +0800 Subject: [PATCH 06/15] drm: add a locked version of drm_is_current_master While checking the master status of the DRM file in drm_is_current_master(), the device's master mutex should be held. Without the mutex, the pointer fpriv->master may be freed concurrently by another process calling drm_setmaster_ioctl(). This could lead to use-after-free errors when the pointer is subsequently dereferenced in drm_lease_owner(). The callers of drm_is_current_master() from drm_auth.c hold the device's master mutex, but external callers do not. Hence, we implement drm_is_current_master_locked() to be used within drm_auth.c, and modify drm_is_current_master() to grab the device's master mutex before checking the master status. Reported-by: Daniel Vetter Signed-off-by: Desmond Cheong Zhi Xi Reviewed-by: Emil Velikov Cc: stable@vger.kernel.org Signed-off-by: Daniel Vetter Link: https://patchwork.freedesktop.org/patch/msgid/20210620110327.4964-2-desmondcheongzx@gmail.com --- drivers/gpu/drm/drm_auth.c | 51 ++++++++++++++++++++++++-------------- 1 file changed, 32 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/drm_auth.c b/drivers/gpu/drm/drm_auth.c index 232abbba3686..86d4b72e95cb 100644 --- a/drivers/gpu/drm/drm_auth.c +++ b/drivers/gpu/drm/drm_auth.c @@ -61,6 +61,35 @@ * trusted clients. */ +static bool drm_is_current_master_locked(struct drm_file *fpriv) +{ + lockdep_assert_held_once(&fpriv->master->dev->master_mutex); + + return fpriv->is_master && drm_lease_owner(fpriv->master) == fpriv->minor->dev->master; +} + +/** + * drm_is_current_master - checks whether @priv is the current master + * @fpriv: DRM file private + * + * Checks whether @fpriv is current master on its device. This decides whether a + * client is allowed to run DRM_MASTER IOCTLs. + * + * Most of the modern IOCTL which require DRM_MASTER are for kernel modesetting + * - the current master is assumed to own the non-shareable display hardware. + */ +bool drm_is_current_master(struct drm_file *fpriv) +{ + bool ret; + + mutex_lock(&fpriv->master->dev->master_mutex); + ret = drm_is_current_master_locked(fpriv); + mutex_unlock(&fpriv->master->dev->master_mutex); + + return ret; +} +EXPORT_SYMBOL(drm_is_current_master); + int drm_getmagic(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_auth *auth = data; @@ -223,7 +252,7 @@ int drm_setmaster_ioctl(struct drm_device *dev, void *data, if (ret) goto out_unlock; - if (drm_is_current_master(file_priv)) + if (drm_is_current_master_locked(file_priv)) goto out_unlock; if (dev->master) { @@ -272,7 +301,7 @@ int drm_dropmaster_ioctl(struct drm_device *dev, void *data, if (ret) goto out_unlock; - if (!drm_is_current_master(file_priv)) { + if (!drm_is_current_master_locked(file_priv)) { ret = -EINVAL; goto out_unlock; } @@ -321,7 +350,7 @@ void drm_master_release(struct drm_file *file_priv) if (file_priv->magic) idr_remove(&file_priv->master->magic_map, file_priv->magic); - if (!drm_is_current_master(file_priv)) + if (!drm_is_current_master_locked(file_priv)) goto out; drm_legacy_lock_master_cleanup(dev, master); @@ -342,22 +371,6 @@ out: mutex_unlock(&dev->master_mutex); } -/** - * drm_is_current_master - checks whether @priv is the current master - * @fpriv: DRM file private - * - * Checks whether @fpriv is current master on its device. This decides whether a - * client is allowed to run DRM_MASTER IOCTLs. - * - * Most of the modern IOCTL which require DRM_MASTER are for kernel modesetting - * - the current master is assumed to own the non-shareable display hardware. - */ -bool drm_is_current_master(struct drm_file *fpriv) -{ - return fpriv->is_master && drm_lease_owner(fpriv->master) == fpriv->minor->dev->master; -} -EXPORT_SYMBOL(drm_is_current_master); - /** * drm_master_get - reference a master pointer * @master: &struct drm_master From 4c6a23188e26339fd3dbc78e6ce6fe0fc4009553 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Michel=20D=C3=A4nzer?= Date: Wed, 16 Jun 2021 12:46:51 +0200 Subject: [PATCH 07/15] drm/amdgpu: Call drm_framebuffer_init last for framebuffer init MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Once drm_framebuffer_init has returned 0, the framebuffer is hooked up to the reference counting machinery and can no longer be destroyed with a simple kfree. Therefore, it must be called last. If drm_framebuffer_init returns 0 but its caller then returns non-0, there will likely be memory corruption fireworks down the road. The following lead me to this fix: [ 12.891228] kernel BUG at lib/list_debug.c:25! [...] [ 12.891263] RIP: 0010:__list_add_valid+0x4b/0x70 [...] [ 12.891324] Call Trace: [ 12.891330] drm_framebuffer_init+0xb5/0x100 [drm] [ 12.891378] amdgpu_display_gem_fb_verify_and_init+0x47/0x120 [amdgpu] [ 12.891592] ? amdgpu_display_user_framebuffer_create+0x10d/0x1f0 [amdgpu] [ 12.891794] amdgpu_display_user_framebuffer_create+0x126/0x1f0 [amdgpu] [ 12.891995] drm_internal_framebuffer_create+0x378/0x3f0 [drm] [ 12.892036] ? drm_internal_framebuffer_create+0x3f0/0x3f0 [drm] [ 12.892075] drm_mode_addfb2+0x34/0xd0 [drm] [ 12.892115] ? drm_internal_framebuffer_create+0x3f0/0x3f0 [drm] [ 12.892153] drm_ioctl_kernel+0xe2/0x150 [drm] [ 12.892193] drm_ioctl+0x3da/0x460 [drm] [ 12.892232] ? drm_internal_framebuffer_create+0x3f0/0x3f0 [drm] [ 12.892274] amdgpu_drm_ioctl+0x43/0x80 [amdgpu] [ 12.892475] __se_sys_ioctl+0x72/0xc0 [ 12.892483] do_syscall_64+0x33/0x40 [ 12.892491] entry_SYSCALL_64_after_hwframe+0x44/0xae Fixes: f258907fdd835e "drm/amdgpu: Verify bo size can fit framebuffer size on init." Signed-off-by: Michel Dänzer Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index c13985fb35be..2a4cd7d377bf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -1047,11 +1047,12 @@ int amdgpu_display_gem_fb_init(struct drm_device *dev, rfb->base.obj[0] = obj; drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd); - ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); + + ret = amdgpu_display_framebuffer_init(dev, rfb, mode_cmd, obj); if (ret) goto err; - ret = amdgpu_display_framebuffer_init(dev, rfb, mode_cmd, obj); + ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); if (ret) goto err; @@ -1071,9 +1072,6 @@ int amdgpu_display_gem_fb_verify_and_init( rfb->base.obj[0] = obj; drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd); - ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); - if (ret) - goto err; /* Verify that the modifier is supported. */ if (!drm_any_plane_has_format(dev, mode_cmd->pixel_format, mode_cmd->modifier[0])) { @@ -1092,6 +1090,10 @@ int amdgpu_display_gem_fb_verify_and_init( if (ret) goto err; + ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); + if (ret) + goto err; + return 0; err: drm_dbg_kms(dev, "Failed to verify and init gem fb: %d\n", ret); From baacf52a473b24e10322b67757ddb92ab8d86717 Mon Sep 17 00:00:00 2001 From: Yifan Zhang Date: Sat, 19 Jun 2021 11:39:43 +0800 Subject: [PATCH 08/15] Revert "drm/amdgpu/gfx10: enlarge CP_MEC_DOORBELL_RANGE_UPPER to cover full doorbell." This reverts commit 1c0b0efd148d5b24c4932ddb3fa03c8edd6097b3. Reason for revert: Side effect of enlarging CP_MEC_DOORBELL_RANGE may cause some APUs fail to enter gfxoff in certain user cases. Signed-off-by: Yifan Zhang Acked-by: Alex Deucher Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 327b1f8213a8..0597aeb5f0e8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -6871,12 +6871,8 @@ static int gfx_v10_0_kiq_init_register(struct amdgpu_ring *ring) if (ring->use_doorbell) { WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_LOWER, (adev->doorbell_index.kiq * 2) << 2); - /* If GC has entered CGPG, ringing doorbell > first page doesn't - * wakeup GC. Enlarge CP_MEC_DOORBELL_RANGE_UPPER to workaround - * this issue. - */ WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER, - (adev->doorbell.size - 4)); + (adev->doorbell_index.userqueue_end * 2) << 2); } WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, From ee5468b9f1d3bf48082eed351dace14598e8ca39 Mon Sep 17 00:00:00 2001 From: Yifan Zhang Date: Sat, 19 Jun 2021 11:40:54 +0800 Subject: [PATCH 09/15] Revert "drm/amdgpu/gfx9: fix the doorbell missing when in CGPG issue." This reverts commit 4cbbe34807938e6e494e535a68d5ff64edac3f20. Reason for revert: side effect of enlarging CP_MEC_DOORBELL_RANGE may cause some APUs fail to enter gfxoff in certain user cases. Signed-off-by: Yifan Zhang Acked-by: Alex Deucher Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index c09225d065c2..516467e962b7 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -3673,12 +3673,8 @@ static int gfx_v9_0_kiq_init_register(struct amdgpu_ring *ring) if (ring->use_doorbell) { WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_LOWER, (adev->doorbell_index.kiq * 2) << 2); - /* If GC has entered CGPG, ringing doorbell > first page doesn't - * wakeup GC. Enlarge CP_MEC_DOORBELL_RANGE_UPPER to workaround - * this issue. - */ WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER, - (adev->doorbell.size - 4)); + (adev->doorbell_index.userqueue_end * 2) << 2); } WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, From f54b3ca7ea1e5e02f481cf4ca54568e57bd66086 Mon Sep 17 00:00:00 2001 From: Daniel Vetter Date: Tue, 22 Jun 2021 09:54:09 +0200 Subject: [PATCH 10/15] Revert "drm: add a locked version of drm_is_current_master" This reverts commit 1815d9c86e3090477fbde066ff314a7e9721ee0f. Unfortunately this inverts the locking hierarchy, so back to the drawing board. Full lockdep splat below: ====================================================== WARNING: possible circular locking dependency detected 5.13.0-rc7-CI-CI_DRM_10254+ #1 Not tainted ------------------------------------------------------ kms_frontbuffer/1087 is trying to acquire lock: ffff88810dcd01a8 (&dev->master_mutex){+.+.}-{3:3}, at: drm_is_current_master+0x1b/0x40 but task is already holding lock: ffff88810dcd0488 (&dev->mode_config.mutex){+.+.}-{3:3}, at: drm_mode_getconnector+0x1c6/0x4a0 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (&dev->mode_config.mutex){+.+.}-{3:3}: __mutex_lock+0xab/0x970 drm_client_modeset_probe+0x22e/0xca0 __drm_fb_helper_initial_config_and_unlock+0x42/0x540 intel_fbdev_initial_config+0xf/0x20 [i915] async_run_entry_fn+0x28/0x130 process_one_work+0x26d/0x5c0 worker_thread+0x37/0x380 kthread+0x144/0x170 ret_from_fork+0x1f/0x30 -> #1 (&client->modeset_mutex){+.+.}-{3:3}: __mutex_lock+0xab/0x970 drm_client_modeset_commit_locked+0x1c/0x180 drm_client_modeset_commit+0x1c/0x40 __drm_fb_helper_restore_fbdev_mode_unlocked+0x88/0xb0 drm_fb_helper_set_par+0x34/0x40 intel_fbdev_set_par+0x11/0x40 [i915] fbcon_init+0x270/0x4f0 visual_init+0xc6/0x130 do_bind_con_driver+0x1e5/0x2d0 do_take_over_console+0x10e/0x180 do_fbcon_takeover+0x53/0xb0 register_framebuffer+0x22d/0x310 __drm_fb_helper_initial_config_and_unlock+0x36c/0x540 intel_fbdev_initial_config+0xf/0x20 [i915] async_run_entry_fn+0x28/0x130 process_one_work+0x26d/0x5c0 worker_thread+0x37/0x380 kthread+0x144/0x170 ret_from_fork+0x1f/0x30 -> #0 (&dev->master_mutex){+.+.}-{3:3}: __lock_acquire+0x151e/0x2590 lock_acquire+0xd1/0x3d0 __mutex_lock+0xab/0x970 drm_is_current_master+0x1b/0x40 drm_mode_getconnector+0x37e/0x4a0 drm_ioctl_kernel+0xa8/0xf0 drm_ioctl+0x1e8/0x390 __x64_sys_ioctl+0x6a/0xa0 do_syscall_64+0x39/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae other info that might help us debug this: Chain exists of: &dev->master_mutex --> &client->modeset_mutex --> &dev->mode_config.mutex Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&dev->mode_config.mutex); lock(&client->modeset_mutex); lock(&dev->mode_config.mutex); lock(&dev->master_mutex); *** DEADLOCK *** 1 lock held by kms_frontbuffer/1087: #0: ffff88810dcd0488 (&dev->mode_config.mutex){+.+.}-{3:3}, at: drm_mode_getconnector+0x1c6/0x4a0 stack backtrace: CPU: 7 PID: 1087 Comm: kms_frontbuffer Not tainted 5.13.0-rc7-CI-CI_DRM_10254+ #1 Hardware name: Intel Corporation Ice Lake Client Platform/IceLake U DDR4 SODIMM PD RVP TLC, BIOS ICLSFWR1.R00.3234.A01.1906141750 06/14/2019 Call Trace: dump_stack+0x7f/0xad check_noncircular+0x12e/0x150 __lock_acquire+0x151e/0x2590 lock_acquire+0xd1/0x3d0 __mutex_lock+0xab/0x970 drm_is_current_master+0x1b/0x40 drm_mode_getconnector+0x37e/0x4a0 drm_ioctl_kernel+0xa8/0xf0 drm_ioctl+0x1e8/0x390 __x64_sys_ioctl+0x6a/0xa0 do_syscall_64+0x39/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae Note that this broke the intel-gfx CI pretty much across the board because it has to reboot machines after it hits a lockdep splat. Testcase: igt/debugfs_test/read_all_entries Acked-by: Petri Latvala Fixes: 1815d9c86e30 ("drm: add a locked version of drm_is_current_master") Cc: Desmond Cheong Zhi Xi Cc: Emil Velikov Cc: stable@vger.kernel.org Signed-off-by: Daniel Vetter Cc: Maarten Lankhorst Cc: Maxime Ripard Cc: Thomas Zimmermann Cc: David Airlie Cc: Daniel Vetter Link: https://patchwork.freedesktop.org/patch/msgid/20210622075409.2673805-1-daniel.vetter@ffwll.ch --- drivers/gpu/drm/drm_auth.c | 51 ++++++++++++++------------------------ 1 file changed, 19 insertions(+), 32 deletions(-) diff --git a/drivers/gpu/drm/drm_auth.c b/drivers/gpu/drm/drm_auth.c index 86d4b72e95cb..232abbba3686 100644 --- a/drivers/gpu/drm/drm_auth.c +++ b/drivers/gpu/drm/drm_auth.c @@ -61,35 +61,6 @@ * trusted clients. */ -static bool drm_is_current_master_locked(struct drm_file *fpriv) -{ - lockdep_assert_held_once(&fpriv->master->dev->master_mutex); - - return fpriv->is_master && drm_lease_owner(fpriv->master) == fpriv->minor->dev->master; -} - -/** - * drm_is_current_master - checks whether @priv is the current master - * @fpriv: DRM file private - * - * Checks whether @fpriv is current master on its device. This decides whether a - * client is allowed to run DRM_MASTER IOCTLs. - * - * Most of the modern IOCTL which require DRM_MASTER are for kernel modesetting - * - the current master is assumed to own the non-shareable display hardware. - */ -bool drm_is_current_master(struct drm_file *fpriv) -{ - bool ret; - - mutex_lock(&fpriv->master->dev->master_mutex); - ret = drm_is_current_master_locked(fpriv); - mutex_unlock(&fpriv->master->dev->master_mutex); - - return ret; -} -EXPORT_SYMBOL(drm_is_current_master); - int drm_getmagic(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_auth *auth = data; @@ -252,7 +223,7 @@ int drm_setmaster_ioctl(struct drm_device *dev, void *data, if (ret) goto out_unlock; - if (drm_is_current_master_locked(file_priv)) + if (drm_is_current_master(file_priv)) goto out_unlock; if (dev->master) { @@ -301,7 +272,7 @@ int drm_dropmaster_ioctl(struct drm_device *dev, void *data, if (ret) goto out_unlock; - if (!drm_is_current_master_locked(file_priv)) { + if (!drm_is_current_master(file_priv)) { ret = -EINVAL; goto out_unlock; } @@ -350,7 +321,7 @@ void drm_master_release(struct drm_file *file_priv) if (file_priv->magic) idr_remove(&file_priv->master->magic_map, file_priv->magic); - if (!drm_is_current_master_locked(file_priv)) + if (!drm_is_current_master(file_priv)) goto out; drm_legacy_lock_master_cleanup(dev, master); @@ -371,6 +342,22 @@ out: mutex_unlock(&dev->master_mutex); } +/** + * drm_is_current_master - checks whether @priv is the current master + * @fpriv: DRM file private + * + * Checks whether @fpriv is current master on its device. This decides whether a + * client is allowed to run DRM_MASTER IOCTLs. + * + * Most of the modern IOCTL which require DRM_MASTER are for kernel modesetting + * - the current master is assumed to own the non-shareable display hardware. + */ +bool drm_is_current_master(struct drm_file *fpriv) +{ + return fpriv->is_master && drm_lease_owner(fpriv->master) == fpriv->minor->dev->master; +} +EXPORT_SYMBOL(drm_is_current_master); + /** * drm_master_get - reference a master pointer * @master: &struct drm_master From 17b11f71795abdce46f62a808f906857e525cea8 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Mon, 21 Jun 2021 13:36:35 +0200 Subject: [PATCH 11/15] drm/nouveau: wait for moving fence after pinning v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We actually need to wait for the moving fence after pinning the BO to make sure that the pin is completed. v2: grab the lock while waiting Signed-off-by: Christian König Reviewed-by: Daniel Vetter References: https://lore.kernel.org/dri-devel/20210621151758.2347474-1-daniel.vetter@ffwll.ch/ CC: stable@kernel.org Link: https://patchwork.freedesktop.org/patch/msgid/20210622114506.106349-1-christian.koenig@amd.com --- drivers/gpu/drm/nouveau/nouveau_prime.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_prime.c b/drivers/gpu/drm/nouveau/nouveau_prime.c index 347488685f74..60019d0532fc 100644 --- a/drivers/gpu/drm/nouveau/nouveau_prime.c +++ b/drivers/gpu/drm/nouveau/nouveau_prime.c @@ -93,7 +93,22 @@ int nouveau_gem_prime_pin(struct drm_gem_object *obj) if (ret) return -EINVAL; - return 0; + ret = ttm_bo_reserve(&nvbo->bo, false, false, NULL); + if (ret) + goto error; + + if (nvbo->bo.moving) + ret = dma_fence_wait(nvbo->bo.moving, true); + + ttm_bo_unreserve(&nvbo->bo); + if (ret) + goto error; + + return ret; + +error: + nouveau_bo_unpin(nvbo); + return ret; } void nouveau_gem_prime_unpin(struct drm_gem_object *obj) From 4b41726aae563273bb4b4a9462ba51ce4d372f78 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Mon, 21 Jun 2021 13:43:05 +0200 Subject: [PATCH 12/15] drm/radeon: wait for moving fence after pinning MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We actually need to wait for the moving fence after pinning the BO to make sure that the pin is completed. Signed-off-by: Christian König Reviewed-by: Daniel Vetter References: https://lore.kernel.org/dri-devel/20210621151758.2347474-1-daniel.vetter@ffwll.ch/ CC: stable@kernel.org Link: https://patchwork.freedesktop.org/patch/msgid/20210622114506.106349-2-christian.koenig@amd.com --- drivers/gpu/drm/radeon/radeon_prime.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/radeon/radeon_prime.c b/drivers/gpu/drm/radeon/radeon_prime.c index 42a87948e28c..4a90807351e7 100644 --- a/drivers/gpu/drm/radeon/radeon_prime.c +++ b/drivers/gpu/drm/radeon/radeon_prime.c @@ -77,9 +77,19 @@ int radeon_gem_prime_pin(struct drm_gem_object *obj) /* pin buffer into GTT */ ret = radeon_bo_pin(bo, RADEON_GEM_DOMAIN_GTT, NULL); - if (likely(ret == 0)) - bo->prime_shared_count++; + if (unlikely(ret)) + goto error; + if (bo->tbo.moving) { + ret = dma_fence_wait(bo->tbo.moving, false); + if (unlikely(ret)) { + radeon_bo_unpin(bo); + goto error; + } + } + + bo->prime_shared_count++; +error: radeon_bo_unreserve(bo); return ret; } From 8ddf5b9bb479570a3825d70fecfb9399bc15700c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Mon, 21 Jun 2021 14:29:14 +0200 Subject: [PATCH 13/15] drm/amdgpu: wait for moving fence after pinning MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We actually need to wait for the moving fence after pinning the BO to make sure that the pin is completed. Signed-off-by: Christian König Reviewed-by: Daniel Vetter References: https://lore.kernel.org/dri-devel/20210621151758.2347474-1-daniel.vetter@ffwll.ch/ CC: stable@kernel.org Link: https://patchwork.freedesktop.org/patch/msgid/20210622114506.106349-3-christian.koenig@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index baa980a477d9..37ec59365080 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -214,9 +214,21 @@ static int amdgpu_dma_buf_pin(struct dma_buf_attachment *attach) { struct drm_gem_object *obj = attach->dmabuf->priv; struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); + int r; /* pin buffer into GTT */ - return amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT); + r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT); + if (r) + return r; + + if (bo->tbo.moving) { + r = dma_fence_wait(bo->tbo.moving, true); + if (r) { + amdgpu_bo_unpin(bo); + return r; + } + } + return 0; } /** From 6fd8f323b3e4e5290d02174559308669507c00dd Mon Sep 17 00:00:00 2001 From: Zhen Lei Date: Thu, 13 May 2021 21:46:38 +0800 Subject: [PATCH 14/15] drm/kmb: Fix error return code in kmb_hw_init() When the call to platform_get_irq() to obtain the IRQ of the lcd fails, the returned error code should be propagated. However, we currently do not explicitly assign this error code to 'ret'. As a result, 0 was incorrectly returned. Fixes: 7f7b96a8a0a1 ("drm/kmb: Add support for KeemBay Display") Reported-by: Hulk Robot Signed-off-by: Zhen Lei Signed-off-by: Anitha Chrisanthus Link: https://patchwork.freedesktop.org/patch/msgid/20210513134639.6541-1-thunder.leizhen@huawei.com --- drivers/gpu/drm/kmb/kmb_drv.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/kmb/kmb_drv.c b/drivers/gpu/drm/kmb/kmb_drv.c index f64e06e1067d..96ea1a2c11dd 100644 --- a/drivers/gpu/drm/kmb/kmb_drv.c +++ b/drivers/gpu/drm/kmb/kmb_drv.c @@ -137,6 +137,7 @@ static int kmb_hw_init(struct drm_device *drm, unsigned long flags) /* Allocate LCD interrupt resources */ irq_lcd = platform_get_irq(pdev, 0); if (irq_lcd < 0) { + ret = irq_lcd; drm_err(&kmb->drm, "irq_lcd not found"); goto setup_fail; } From d330099115597bbc238d6758a4930e72b49ea9ba Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Fri, 11 Jun 2021 14:34:50 +0200 Subject: [PATCH 15/15] drm/nouveau: fix dma_address check for CPU/GPU sync MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit AGP for example doesn't have a dma_address array. Signed-off-by: Christian König Acked-by: Alex Deucher Link: https://patchwork.freedesktop.org/patch/msgid/20210614110517.1624-1-christian.koenig@amd.com --- drivers/gpu/drm/nouveau/nouveau_bo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 3e09df0472ce..170aba99a110 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -546,7 +546,7 @@ nouveau_bo_sync_for_device(struct nouveau_bo *nvbo) struct ttm_tt *ttm_dma = (struct ttm_tt *)nvbo->bo.ttm; int i, j; - if (!ttm_dma) + if (!ttm_dma || !ttm_dma->dma_address) return; if (!ttm_dma->pages) { NV_DEBUG(drm, "ttm_dma 0x%p: pages NULL\n", ttm_dma); @@ -582,7 +582,7 @@ nouveau_bo_sync_for_cpu(struct nouveau_bo *nvbo) struct ttm_tt *ttm_dma = (struct ttm_tt *)nvbo->bo.ttm; int i, j; - if (!ttm_dma) + if (!ttm_dma || !ttm_dma->dma_address) return; if (!ttm_dma->pages) { NV_DEBUG(drm, "ttm_dma 0x%p: pages NULL\n", ttm_dma);