Skip to contentSkip to conversation

drm/msm/dpu: avoid uninitialized variable use

PATCH In review Arnd Bergmann 4 participants 6 messages Aug 7, 2025 – Aug 10, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/msm/dpu: avoid uninitialized variable use

From: Arnd Bergmann <[email protected]>

clang-21 points out a variable that is conditionally initialized but then dereferenced:

drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1138:6: error: variable 'crtc_state' is used uninitialized whenever 'if' condition is false [-Werror,-Wsometimes-uninitialized]
1138 | if (plane_state->crtc)

      |             ^~~~~~~~~~~~~~~~~

drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1142:58: note: uninitialized use occurs here
1142 | ret = dpu_plane_atomic_check_nosspp(plane, plane_state, crtc_state);
| ^~~~~~~~~~
drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1138:2: note: remove the 'if' if its condition is always true

 1138 |         if (plane_state->crtc)
      |         ^~~~~~~~~~~~~~~~~~~~~~
 1139 |                 crtc_state = drm_atomic_get_new_crtc_state(state,
drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1132:35: note: initialize the variable 'crtc_state' to silence this warning
 1132 |         struct drm_crtc_state *crtc_state;
      |                                          ^
      |                                           = NULL

The bug is real, but the suggestion from clang to set it to NULL is unfortunately just as harmful as dereferencing a NULL pointer is little better than uninitialized data.

Change the function to return an error in this case.

Fixes 774bcfb73176 ("drm/msm/dpu: add support for virtual planes")
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c +3 -3
index 01171c535a27..4987f2f2fee0 100644
@@ -1135,10 +1135,10 @@ static int dpu_plane_virtual_atomic_check(struct drm_plane *plane,
 	if (IS_ERR(plane_state))
 		return PTR_ERR(plane_state);
 
-	if (plane_state->crtc)
-		crtc_state = drm_atomic_get_new_crtc_state(state,
-							   plane_state->crtc);
+	if (!plane_state->crtc)
+		return -ENXIO;
 
+	crtc_state = drm_atomic_get_new_crtc_state(state, plane_state->crtc);
 	ret = dpu_plane_atomic_check_nosspp(plane, plane_state, crtc_state);
 	if (ret)
 		return ret;
Signature · 1 line
2.39.5
#1 lore raw
Quoted 20 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

clang-21 points out a variable that is conditionally initialized but then dereferenced:

drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1138:6: error: variable 'crtc_state' is used uninitialized whenever 'if' condition is false [-Werror,-Wsometimes-uninitialized]
1138 | if (plane_state->crtc)

      |             ^~~~~~~~~~~~~~~~~

drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1142:58: note: uninitialized use occurs here
1142 | ret = dpu_plane_atomic_check_nosspp(plane, plane_state, crtc_state);
| ^~~~~~~~~~
drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1138:2: note: remove the 'if' if its condition is always true

 1138 |         if (plane_state->crtc)
      |         ^~~~~~~~~~~~~~~~~~~~~~
 1139 |                 crtc_state = drm_atomic_get_new_crtc_state(state,
drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c:1132:35: note: initialize the variable 'crtc_state' to silence this warning
 1132 |         struct drm_crtc_state *crtc_state;
      |                                          ^
      |                                           = NULL
Quoted 3 lines

The bug is real, but the suggestion from clang to set it to NULL is unfortunately just as harmful as dereferencing a NULL pointer is little better than uninitialized data.

Having no plane->crtc is a valid setting and it is handled inside drm_atomic_helper_check_plane_state() by setting plane_state->visible = false and returning early. Setting crtc_state to NULL is a correct fix. Could you please send it?

Quoted 9 lines

Change the function to return an error in this case.

Fixes 774bcfb73176 ("drm/msm/dpu: add support for virtual planes")
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
Signature · 2 lines
With best wishes
Dmitry
#3 lore raw
Quoted 7 lines · Nathan Chancellor
Quoted 4 lines · Dmitry Baryshkov

Having no plane->crtc is a valid setting and it is handled inside drm_atomic_helper_check_plane_state() by setting plane_state->visible = false and returning early. Setting crtc_state to NULL is a correct fix. Could you please send it?

I sent this fix three weeks ago, could this be applied?

It will be picked up for -rc2 (hopefully)

Quoted 5 lines

https://lore.kernel.org/20250715-drm-msm-fix-const-uninit-warning-v1-1-d6a366fd9a32@kernel.org/

Cheers,
Nathan

Signature · 2 lines
With best wishes
Dmitry
#4 lore raw
Quoted 10 lines · Dmitry Baryshkov
Quoted 7 lines · Nathan Chancellor
Quoted 4 lines · Dmitry Baryshkov

Having no plane->crtc is a valid setting and it is handled inside drm_atomic_helper_check_plane_state() by setting plane_state->visible = false and returning early. Setting crtc_state to NULL is a correct fix. Could you please send it?

I sent this fix three weeks ago, could this be applied?

It will be picked up for -rc2 (hopefully)

Great, thank you! Even if it slips to -rc3, no worries, as long it is in the pipeline to get there.

Cheers,
Nathan

#5 lore raw
In reply to Dmitry Baryshkov
Quoted 11 lines · Dmitry Baryshkov
Quoted 4 lines · Arnd Bergmann

The bug is real, but the suggestion from clang to set it to NULL is unfortunately just as harmful as dereferencing a NULL pointer is little better than uninitialized data.

Having no plane->crtc is a valid setting and it is handled inside drm_atomic_helper_check_plane_state() by setting plane_state->visible = false and returning early. Setting crtc_state to NULL is a correct fix. Could you please send it?

Ah, I see. I saw the crtc_state dereference in

  WARN_ON(plane_state->crtc && plane_state->crtc != crtc_state->crtc);

but that is indeed guarded by the plane_state->crtc check.

Nathan's patch is sufficient then.

Arnd