[PATCH] drm/vmwgfx: Add blend mode property

[PATCH] drm/vmwgfx: Add blend mode property

[Ian Forbes]

This stops the log spamming introduced by the referenced fixes commit.

Fixes: 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed")
Signed-off-by: Ian Forbes <ian.forbes@broadcom.com>
---
 drivers/gpu/drm/vmwgfx/vmwgfx_kms.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
index 1b407b61f683..c06548960008 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
@@ -20,10 +20,17 @@
 #include <drm/drm_rect.h>
 #include <drm/drm_sysfs.h>
 #include <drm/drm_edid.h>
+#include <drm/drm_blend.h>
 
 void vmw_du_init(struct vmw_display_unit *du)
 {
 	vmw_vkms_crtc_init(&du->crtc);
+	drm_plane_create_blend_mode_property(&du->primary,
+					     BIT(DRM_MODE_BLEND_PIXEL_NONE));
+	// Check that the cursor plane was actually initialized
+	if (du->crtc.cursor == &du->cursor.base)
+		drm_plane_create_blend_mode_property(&du->cursor.base,
+						     BIT(DRM_MODE_BLEND_PREMULTI));
 }
 
 void vmw_du_cleanup(struct vmw_display_unit *du)
-- 
2.55.0

Re: [PATCH] drm/vmwgfx: Add blend mode property

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Missing drm_plane_cleanup() on error paths in display unit initialization functions leads to list corruption and use-after-free.
--

commit dea10d2954fcbf7f9f6713f8f44088d76a62295e
Author: Ian Forbes <ian.forbes@broadcom.com>

drm/vmwgfx: Add blend mode property

This commit adds blend mode properties to the primary and cursor planes to 
stop log spamming. It initializes the properties in vmw_du_init() for both
planes if they are present.

> diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
> --- a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
> +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
[ ... ]
>  void vmw_du_init(struct vmw_display_unit *du)
>  {
>  	vmw_vkms_crtc_init(&du->crtc);
> +	drm_plane_create_blend_mode_property(&du->primary,
> +					     BIT(DRM_MODE_BLEND_PIXEL_NONE));
> +	// Check that the cursor plane was actually initialized
> +	if (du->crtc.cursor == &du->cursor.base)
> +		drm_plane_create_blend_mode_property(&du->cursor.base,
> +						     BIT(DRM_MODE_BLEND_PREMULTI));
>  }

[Severity: High]
This isn't a bug introduced by this patch, but I noticed a pre-existing
issue regarding the cleanup of these planes on error paths during display unit
initialization.

In functions like vmw_ldu_init(), vmw_sou_init(), and vmw_stdu_init(), the
primary and cursor planes are initialized, which links them into the DRM
core plane list. If a subsequent step like drm_connector_init() fails, the
code jumps to an error label and frees the display unit structure without
cleaning up the planes:

drivers/gpu/drm/vmwgfx/vmwgfx_ldu.c:vmw_ldu_init() {
...
	ret = drm_connector_init(dev, connector, &vmw_legacy_connector_funcs,
				 DRM_MODE_CONNECTOR_VIRTUAL);
	if (ret) {
		DRM_ERROR("Failed to initialize connector\n");
		goto err_free;
	}
...
err_free:
	kfree(ldu);
	return ret;
}

Similar error paths exist in vmw_sou_init() for the sou structure, and
vmw_stdu_init() for the stdu structure.

Since the display unit structure containing the planes is freed with kfree()
but drm_plane_cleanup() hasn't been called, could this leave dangling
pointers in the DRM core plane list? If the list is later traversed during
driver teardown in drm_mode_config_cleanup(), wouldn't this result in a
use-after-free? 

Should these error paths be updated to call drm_plane_cleanup() or a similar
teardown helper before freeing the memory?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260911175607.1127474-1-ian.forbes@broadcom.com?part=1