diff mbox series

[v12,4/9] drm/panic: Add debugfs entry to test without triggering panic.

Message ID 20240409163432.352518-5-jfalempe@redhat.com (mailing list archive)
State New, archived
Headers show
Series drm/panic: Add a drm panic handler | expand

Commit Message

Jocelyn Falempe April 9, 2024, 4:30 p.m. UTC
Add a debugfs file, so you can test drm_panic without freezing
your machine. This is unsafe, and should be enabled only for
developer or tester.

To display the drm_panic screen on the device 0:
echo 1 > /sys/kernel/debug/dri/0/drm_panic_plane_0

v9:
 * Create a debugfs file for each plane in the device's debugfs
   directory. This allows to test for each plane of each GPU
   independently.

Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
---
 drivers/gpu/drm/Kconfig     |  9 ++++++++
 drivers/gpu/drm/drm_panic.c | 43 ++++++++++++++++++++++++++++++++++++-
 2 files changed, 51 insertions(+), 1 deletion(-)

Comments

Daniel Vetter April 10, 2024, 8:01 a.m. UTC | #1
On Tue, Apr 09, 2024 at 06:30:43PM +0200, Jocelyn Falempe wrote:
> Add a debugfs file, so you can test drm_panic without freezing
> your machine. This is unsafe, and should be enabled only for
> developer or tester.
> 
> To display the drm_panic screen on the device 0:
> echo 1 > /sys/kernel/debug/dri/0/drm_panic_plane_0
> 
> v9:
>  * Create a debugfs file for each plane in the device's debugfs
>    directory. This allows to test for each plane of each GPU
>    independently.
> 
> Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>

I was pondering whether this guarantees that the debugfs file disappears
before drm_dev_unregister finishes (otherwise we have a bit a problem),
and looks like we're good.

Maybe add a todo that it would be nice to simulate nmi context, not sure
lockdept can help here ...

Anyway Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> ---
>  drivers/gpu/drm/Kconfig     |  9 ++++++++
>  drivers/gpu/drm/drm_panic.c | 43 ++++++++++++++++++++++++++++++++++++-
>  2 files changed, 51 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig
> index f8a26423369e..959b19a04101 100644
> --- a/drivers/gpu/drm/Kconfig
> +++ b/drivers/gpu/drm/Kconfig
> @@ -127,6 +127,15 @@ config DRM_PANIC_BACKGROUND_COLOR
>  	depends on DRM_PANIC
>  	default 0x000000
>  
> +config DRM_PANIC_DEBUG
> +	bool "Add a debug fs entry to trigger drm_panic"
> +	depends on DRM_PANIC && DEBUG_FS
> +	help
> +	  Add dri/[device]/drm_panic_plane_x in the kernel debugfs, to force the
> +	  panic handler to write the panic message to this plane scanout buffer.
> +	  This is unsafe and should not be enabled on a production build.
> +	  If in doubt, say "N".
> +
>  config DRM_DEBUG_DP_MST_TOPOLOGY_REFS
>          bool "Enable refcount backtrace history in the DP MST helpers"
>  	depends on STACKTRACE_SUPPORT
> diff --git a/drivers/gpu/drm/drm_panic.c b/drivers/gpu/drm/drm_panic.c
> index e1ec30b6c04a..78fd6d5d7adc 100644
> --- a/drivers/gpu/drm/drm_panic.c
> +++ b/drivers/gpu/drm/drm_panic.c
> @@ -495,6 +495,45 @@ static void drm_panic(struct kmsg_dumper *dumper, enum kmsg_dump_reason reason)
>  		draw_panic_plane(plane);
>  }
>  
> +
> +/*
> + * DEBUG FS, This is currently unsafe.
> + * Create one file per plane, so it's possible to debug one plane at a time.
> + */
> +#ifdef CONFIG_DRM_PANIC_DEBUG
> +#include <linux/debugfs.h>
> +
> +static ssize_t debugfs_trigger_write(struct file *file, const char __user *user_buf,
> +				     size_t count, loff_t *ppos)
> +{
> +	bool run;
> +
> +	if (kstrtobool_from_user(user_buf, count, &run) == 0 && run) {
> +		struct drm_plane *plane = file->private_data;
> +
> +		draw_panic_plane(plane);
> +	}
> +	return count;
> +}
> +
> +static const struct file_operations dbg_drm_panic_ops = {
> +	.owner = THIS_MODULE,
> +	.write = debugfs_trigger_write,
> +	.open = simple_open,
> +};
> +
> +static void debugfs_register_plane(struct drm_plane *plane, int index)
> +{
> +	char fname[32];
> +
> +	snprintf(fname, 32, "drm_panic_plane_%d", index);
> +	debugfs_create_file(fname, 0200, plane->dev->debugfs_root,
> +			    plane, &dbg_drm_panic_ops);
> +}
> +#else
> +static void debugfs_register_plane(struct drm_plane *plane, int index) {}
> +#endif /* CONFIG_DRM_PANIC_DEBUG */
> +
>  /**
>   * drm_panic_register() - Initialize DRM panic for a device
>   * @dev: the drm device on which the panic screen will be displayed.
> @@ -514,8 +553,10 @@ void drm_panic_register(struct drm_device *dev)
>  		plane->kmsg_panic.max_reason = KMSG_DUMP_PANIC;
>  		if (kmsg_dump_register(&plane->kmsg_panic))
>  			drm_warn(dev, "Failed to register panic handler\n");
> -		else
> +		else {
> +			debugfs_register_plane(plane, registered_plane);
>  			registered_plane++;
> +		}
>  	}
>  	if (registered_plane)
>  		drm_info(dev, "Registered %d planes with drm panic\n", registered_plane);
> -- 
> 2.44.0
>
diff mbox series

Patch

diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig
index f8a26423369e..959b19a04101 100644
--- a/drivers/gpu/drm/Kconfig
+++ b/drivers/gpu/drm/Kconfig
@@ -127,6 +127,15 @@  config DRM_PANIC_BACKGROUND_COLOR
 	depends on DRM_PANIC
 	default 0x000000
 
+config DRM_PANIC_DEBUG
+	bool "Add a debug fs entry to trigger drm_panic"
+	depends on DRM_PANIC && DEBUG_FS
+	help
+	  Add dri/[device]/drm_panic_plane_x in the kernel debugfs, to force the
+	  panic handler to write the panic message to this plane scanout buffer.
+	  This is unsafe and should not be enabled on a production build.
+	  If in doubt, say "N".
+
 config DRM_DEBUG_DP_MST_TOPOLOGY_REFS
         bool "Enable refcount backtrace history in the DP MST helpers"
 	depends on STACKTRACE_SUPPORT
diff --git a/drivers/gpu/drm/drm_panic.c b/drivers/gpu/drm/drm_panic.c
index e1ec30b6c04a..78fd6d5d7adc 100644
--- a/drivers/gpu/drm/drm_panic.c
+++ b/drivers/gpu/drm/drm_panic.c
@@ -495,6 +495,45 @@  static void drm_panic(struct kmsg_dumper *dumper, enum kmsg_dump_reason reason)
 		draw_panic_plane(plane);
 }
 
+
+/*
+ * DEBUG FS, This is currently unsafe.
+ * Create one file per plane, so it's possible to debug one plane at a time.
+ */
+#ifdef CONFIG_DRM_PANIC_DEBUG
+#include <linux/debugfs.h>
+
+static ssize_t debugfs_trigger_write(struct file *file, const char __user *user_buf,
+				     size_t count, loff_t *ppos)
+{
+	bool run;
+
+	if (kstrtobool_from_user(user_buf, count, &run) == 0 && run) {
+		struct drm_plane *plane = file->private_data;
+
+		draw_panic_plane(plane);
+	}
+	return count;
+}
+
+static const struct file_operations dbg_drm_panic_ops = {
+	.owner = THIS_MODULE,
+	.write = debugfs_trigger_write,
+	.open = simple_open,
+};
+
+static void debugfs_register_plane(struct drm_plane *plane, int index)
+{
+	char fname[32];
+
+	snprintf(fname, 32, "drm_panic_plane_%d", index);
+	debugfs_create_file(fname, 0200, plane->dev->debugfs_root,
+			    plane, &dbg_drm_panic_ops);
+}
+#else
+static void debugfs_register_plane(struct drm_plane *plane, int index) {}
+#endif /* CONFIG_DRM_PANIC_DEBUG */
+
 /**
  * drm_panic_register() - Initialize DRM panic for a device
  * @dev: the drm device on which the panic screen will be displayed.
@@ -514,8 +553,10 @@  void drm_panic_register(struct drm_device *dev)
 		plane->kmsg_panic.max_reason = KMSG_DUMP_PANIC;
 		if (kmsg_dump_register(&plane->kmsg_panic))
 			drm_warn(dev, "Failed to register panic handler\n");
-		else
+		else {
+			debugfs_register_plane(plane, registered_plane);
 			registered_plane++;
+		}
 	}
 	if (registered_plane)
 		drm_info(dev, "Registered %d planes with drm panic\n", registered_plane);