Nouveau - Jun 2024 - [PATCH v2] drm/nouveau: don't attempt to schedule hpd_work on headless cards

If the card doesn't have display hardware, hpd_work and hpd_lock are
left uninitialized which causes BUG when attempting to schedule hpd_work
on runtime PM resume.

Fix it by adding headless flag to DRM and skip any hpd if it's set.

Fixes: ae1aadb1eb8d ("nouveau: don't fail driver load if no display hw
present.")
Link: https://gitlab.freedesktop.org/drm/nouveau/-/issues/337
Signed-off-by: Vasily Khoruzhick <anarsoul at gmail.com>
---
v2: drop extra checks in nouveau_display_hpd_work() and
nouveau_connector_hpd()

 drivers/gpu/drm/nouveau/dispnv04/disp.c   | 2 +-
 drivers/gpu/drm/nouveau/dispnv50/disp.c   | 2 +-
 drivers/gpu/drm/nouveau/nouveau_display.c | 6 +++++-
 drivers/gpu/drm/nouveau/nouveau_drv.h     | 1 +
 4 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/dispnv04/disp.c
b/drivers/gpu/drm/nouveau/dispnv04/disp.c
index 13705c5f1497..4b7497a8755c 100644
--- a/drivers/gpu/drm/nouveau/dispnv04/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv04/disp.c
@@ -68,7 +68,7 @@ nv04_display_fini(struct drm_device *dev, bool runtime, bool
suspend)
 	if (nv_two_heads(dev))
 		NVWriteCRTC(dev, 1, NV_PCRTC_INTR_EN_0, 0);
 
-	if (!runtime)
+	if (!runtime && !drm->headless)
 		cancel_work_sync(&drm->hpd_work);
 
 	if (!suspend)
diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c
b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index 88728a0b2c25..674dc567e179 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -2680,7 +2680,7 @@ nv50_display_fini(struct drm_device *dev, bool runtime,
bool suspend)
 			nv50_mstm_fini(nouveau_encoder(encoder));
 	}
 
-	if (!runtime)
+	if (!runtime && !drm->headless)
 		cancel_work_sync(&drm->hpd_work);
 }
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c
b/drivers/gpu/drm/nouveau/nouveau_display.c
index aed5d5b51b43..d4725a968827 100644
--- a/drivers/gpu/drm/nouveau/nouveau_display.c
+++ b/drivers/gpu/drm/nouveau/nouveau_display.c
@@ -450,6 +450,9 @@ nouveau_display_hpd_resume(struct drm_device *dev)
 {
 	struct nouveau_drm *drm = nouveau_drm(dev);
 
+	if (drm->headless)
+		return;
+
 	spin_lock_irq(&drm->hpd_lock);
 	drm->hpd_pending = ~0;
 	spin_unlock_irq(&drm->hpd_lock);
@@ -635,7 +638,7 @@ nouveau_display_fini(struct drm_device *dev, bool suspend,
bool runtime)
 	}
 	drm_connector_list_iter_end(&conn_iter);
 
-	if (!runtime)
+	if (!runtime && !drm->headless)
 		cancel_work_sync(&drm->hpd_work);
 
 	drm_kms_helper_poll_disable(dev);
@@ -729,6 +732,7 @@ nouveau_display_create(struct drm_device *dev)
 		/* no display hw */
 		if (ret == -ENODEV) {
 			ret = 0;
+			drm->headless = true;
 			goto disp_create_err;
 		}
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h
b/drivers/gpu/drm/nouveau/nouveau_drv.h
index e239c6bf4afa..25fca98a20bc 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.h
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
@@ -276,6 +276,7 @@ struct nouveau_drm {
 	/* modesetting */
 	struct nvbios vbios;
 	struct nouveau_display *display;
+	bool headless;
 	struct work_struct hpd_work;
 	spinlock_t hpd_lock;
 	u32 hpd_pending;
-- 
2.45.2

On 8/6/24 08:09, Vasily Khoruzhick wrote:
> If the card doesn't have display hardware, hpd_work and hpd_lock are
> left uninitialized which causes BUG when attempting to schedule hpd_work
> on runtime PM resume.
>
> Fix it by adding headless flag to DRM and skip any hpd if it's set.
>
> Fixes: ae1aadb1eb8d ("nouveau: don't fail driver load if no
display hw present.")
> Link: https://gitlab.freedesktop.org/drm/nouveau/-/issues/337
> Signed-off-by: Vasily Khoruzhick <anarsoul at gmail.com>
Reviewed-by: Ben Skeggs <bskeggs at nvidia.com>

> ---
> v2: drop extra checks in nouveau_display_hpd_work() and
> nouveau_connector_hpd()
>
>   drivers/gpu/drm/nouveau/dispnv04/disp.c   | 2 +-
>   drivers/gpu/drm/nouveau/dispnv50/disp.c   | 2 +-
>   drivers/gpu/drm/nouveau/nouveau_display.c | 6 +++++-
>   drivers/gpu/drm/nouveau/nouveau_drv.h     | 1 +
>   4 files changed, 8 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/nouveau/dispnv04/disp.c
b/drivers/gpu/drm/nouveau/dispnv04/disp.c
> index 13705c5f1497..4b7497a8755c 100644
> --- a/drivers/gpu/drm/nouveau/dispnv04/disp.c
> +++ b/drivers/gpu/drm/nouveau/dispnv04/disp.c
> @@ -68,7 +68,7 @@ nv04_display_fini(struct drm_device *dev, bool runtime,
bool suspend)
>   	if (nv_two_heads(dev))
>   		NVWriteCRTC(dev, 1, NV_PCRTC_INTR_EN_0, 0);
>   
> -	if (!runtime)
> +	if (!runtime && !drm->headless)
>   		cancel_work_sync(&drm->hpd_work);
>   
>   	if (!suspend)
> diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c
b/drivers/gpu/drm/nouveau/dispnv50/disp.c
> index 88728a0b2c25..674dc567e179 100644
> --- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
> +++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
> @@ -2680,7 +2680,7 @@ nv50_display_fini(struct drm_device *dev, bool
runtime, bool suspend)
>   			nv50_mstm_fini(nouveau_encoder(encoder));
>   	}
>   
> -	if (!runtime)
> +	if (!runtime && !drm->headless)
>   		cancel_work_sync(&drm->hpd_work);
>   }
>   
> diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c
b/drivers/gpu/drm/nouveau/nouveau_display.c
> index aed5d5b51b43..d4725a968827 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_display.c
> +++ b/drivers/gpu/drm/nouveau/nouveau_display.c
> @@ -450,6 +450,9 @@ nouveau_display_hpd_resume(struct drm_device *dev)
>   {
>   	struct nouveau_drm *drm = nouveau_drm(dev);
>   
> +	if (drm->headless)
> +		return;
> +
>   	spin_lock_irq(&drm->hpd_lock);
>   	drm->hpd_pending = ~0;
>   	spin_unlock_irq(&drm->hpd_lock);
> @@ -635,7 +638,7 @@ nouveau_display_fini(struct drm_device *dev, bool
suspend, bool runtime)
>   	}
>   	drm_connector_list_iter_end(&conn_iter);
>   
> -	if (!runtime)
> +	if (!runtime && !drm->headless)
>   		cancel_work_sync(&drm->hpd_work);
>   
>   	drm_kms_helper_poll_disable(dev);
> @@ -729,6 +732,7 @@ nouveau_display_create(struct drm_device *dev)
>   		/* no display hw */
>   		if (ret == -ENODEV) {
>   			ret = 0;
> +			drm->headless = true;
>   			goto disp_create_err;
>   		}
>   
> diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h
b/drivers/gpu/drm/nouveau/nouveau_drv.h
> index e239c6bf4afa..25fca98a20bc 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_drv.h
> +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
> @@ -276,6 +276,7 @@ struct nouveau_drm {
>   	/* modesetting */
>   	struct nvbios vbios;
>   	struct nouveau_display *display;
> +	bool headless;
>   	struct work_struct hpd_work;
>   	spinlock_t hpd_lock;
>   	u32 hpd_pending;

On 6/8/24 00:09, Vasily Khoruzhick wrote:
> If the card doesn't have display hardware, hpd_work and hpd_lock are
> left uninitialized which causes BUG when attempting to schedule hpd_work
> on runtime PM resume.
> 
> Fix it by adding headless flag to DRM and skip any hpd if it's set.
> 
> Fixes: ae1aadb1eb8d ("nouveau: don't fail driver load if no
display hw present.")
> Link: https://gitlab.freedesktop.org/drm/nouveau/-/issues/337
> Signed-off-by: Vasily Khoruzhick <anarsoul at gmail.com>
Applied to drm-misc-fixes, thanks!
> ---
> v2: drop extra checks in nouveau_display_hpd_work() and
> nouveau_connector_hpd()
> 
>   drivers/gpu/drm/nouveau/dispnv04/disp.c   | 2 +-
>   drivers/gpu/drm/nouveau/dispnv50/disp.c   | 2 +-
>   drivers/gpu/drm/nouveau/nouveau_display.c | 6 +++++-
>   drivers/gpu/drm/nouveau/nouveau_drv.h     | 1 +
>   4 files changed, 8 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/nouveau/dispnv04/disp.c
b/drivers/gpu/drm/nouveau/dispnv04/disp.c
> index 13705c5f1497..4b7497a8755c 100644
> --- a/drivers/gpu/drm/nouveau/dispnv04/disp.c
> +++ b/drivers/gpu/drm/nouveau/dispnv04/disp.c
> @@ -68,7 +68,7 @@ nv04_display_fini(struct drm_device *dev, bool runtime,
bool suspend)
>   	if (nv_two_heads(dev))
>   		NVWriteCRTC(dev, 1, NV_PCRTC_INTR_EN_0, 0);
>   
> -	if (!runtime)
> +	if (!runtime && !drm->headless)
>   		cancel_work_sync(&drm->hpd_work);
>   
>   	if (!suspend)
> diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c
b/drivers/gpu/drm/nouveau/dispnv50/disp.c
> index 88728a0b2c25..674dc567e179 100644
> --- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
> +++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
> @@ -2680,7 +2680,7 @@ nv50_display_fini(struct drm_device *dev, bool
runtime, bool suspend)
>   			nv50_mstm_fini(nouveau_encoder(encoder));
>   	}
>   
> -	if (!runtime)
> +	if (!runtime && !drm->headless)
>   		cancel_work_sync(&drm->hpd_work);
>   }
>   
> diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c
b/drivers/gpu/drm/nouveau/nouveau_display.c
> index aed5d5b51b43..d4725a968827 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_display.c
> +++ b/drivers/gpu/drm/nouveau/nouveau_display.c
> @@ -450,6 +450,9 @@ nouveau_display_hpd_resume(struct drm_device *dev)
>   {
>   	struct nouveau_drm *drm = nouveau_drm(dev);
>   
> +	if (drm->headless)
> +		return;
> +
>   	spin_lock_irq(&drm->hpd_lock);
>   	drm->hpd_pending = ~0;
>   	spin_unlock_irq(&drm->hpd_lock);
> @@ -635,7 +638,7 @@ nouveau_display_fini(struct drm_device *dev, bool
suspend, bool runtime)
>   	}
>   	drm_connector_list_iter_end(&conn_iter);
>   
> -	if (!runtime)
> +	if (!runtime && !drm->headless)
>   		cancel_work_sync(&drm->hpd_work);
>   
>   	drm_kms_helper_poll_disable(dev);
> @@ -729,6 +732,7 @@ nouveau_display_create(struct drm_device *dev)
>   		/* no display hw */
>   		if (ret == -ENODEV) {
>   			ret = 0;
> +			drm->headless = true;
>   			goto disp_create_err;
>   		}
>   
> diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h
b/drivers/gpu/drm/nouveau/nouveau_drv.h
> index e239c6bf4afa..25fca98a20bc 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_drv.h
> +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
> @@ -276,6 +276,7 @@ struct nouveau_drm {
>   	/* modesetting */
>   	struct nvbios vbios;
>   	struct nouveau_display *display;
> +	bool headless;
>   	struct work_struct hpd_work;
>   	spinlock_t hpd_lock;
>   	u32 hpd_pending;