[PATCH] drm/omapdrm: avoid double drm_mode_config_cleanup()

[PATCH] drm/omapdrm: avoid double drm_mode_config_cleanup()

[Hongyan Xu]

omapdrm_init() unwinds drm_vblank_init() and drm_dev_register()
failures through omap_modeset_fini() and then falls through to
err_gem_deinit, which calls drm_mode_config_cleanup() again.

omapdrm_cleanup() has the same problem during the normal teardown path:
it calls omap_modeset_fini() and then calls drm_mode_config_cleanup()
again afterwards.

omap_modeset_fini() should only undo omap_modeset_init(). The
drm_mode_config_init()/drm_mode_config_cleanup() pairing already lives
in the top-level init and cleanup paths, so keeping another
drm_mode_config_cleanup() inside omap_modeset_fini() causes repeated
cleanup of the same mode_config state.

Remove drm_mode_config_cleanup() from omap_modeset_fini() so each path
cleans mode_config exactly once.

Fixes: 2f95bc6d324a ("drm: omapdrm: Perform initialization/cleanup at probe/remove time")
Issue found using a prototype static analysis tool
and confirmed by code review.

Signed-off-by: Hongyan Xu <getshell@seu.edu.cn>
Signed-off-by: Slavin Liu <220245772@seu.edu.cn>
---
 drivers/gpu/drm/omapdrm/omap_drv.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/gpu/drm/omapdrm/omap_drv.c b/drivers/gpu/drm/omapdrm/omap_drv.c
index bf0bad8c8cf1..838a1a236160 100644
--- a/drivers/gpu/drm/omapdrm/omap_drv.c
+++ b/drivers/gpu/drm/omapdrm/omap_drv.c
@@ -538,8 +538,6 @@ static int omap_modeset_init(struct drm_device *dev)
 static void omap_modeset_fini(struct drm_device *ddev)
 {
 	omap_drm_irq_uninstall(ddev);
-
-	drm_mode_config_cleanup(ddev);
 }
 
 /*
-- 
2.50.1.windows.1

Claude review: drm/omapdrm: avoid double drm_mode_config_cleanup()

[Claude Code Review Bot]

Overall Series Review

Subject: drm/omapdrm: avoid double drm_mode_config_cleanup()
Author: Hongyan Xu <getshell@seu.edu.cn>
Patches: 1
Reviewed: 2026-05-16T12:06:25.913209

---

This is a single-patch series that fixes a real double-cleanup bug in the omapdrm driver. The analysis in the commit message is correct: `drm_mode_config_cleanup()` is called twice in both error and normal teardown paths. The fix is minimal, correct, and well-targeted. **Recommend: Accept with minor commit message nits.**

---
Generated by Claude Code Patch Reviewer

Claude review: drm/omapdrm: avoid double drm_mode_config_cleanup()

[Claude Code Review Bot]

Patch Review

**Verdict: The bug is real and the fix is correct.**

The code structure is:

- `drm_mode_config_init()` is called in `omapdrm_init()` at line 715
- `drm_mode_config_cleanup()` is called in the `err_gem_deinit` label (line 761) and in `omapdrm_cleanup()` (line 785)
- `omap_modeset_init()` calls `omap_drm_irq_install()` — it does NOT call `drm_mode_config_init()`

So the init/cleanup pairing should be:
- `omap_modeset_init()` ↔ `omap_modeset_fini()` — IRQ install/uninstall only
- `drm_mode_config_init()` ↔ `drm_mode_config_cleanup()` — handled at the `omapdrm_init()`/`omapdrm_cleanup()` level

**Error path double-cleanup (confirmed):** When `drm_vblank_init()` or `drm_dev_register()` fail, the error path hits `err_cleanup_modeset` → `omap_modeset_fini()` (which before this patch called `drm_mode_config_cleanup()`), then falls through to `err_gem_deinit` → `drm_mode_config_cleanup()` again.

**Normal teardown double-cleanup (confirmed):** `omapdrm_cleanup()` calls `omap_modeset_fini()` at line 782 (which before the patch called `drm_mode_config_cleanup()`), then calls `drm_mode_config_cleanup()` again at line 785.

After this patch, `omap_modeset_fini()` only calls `omap_drm_irq_uninstall()`, which correctly mirrors `omap_modeset_init()` calling `omap_drm_irq_install()`. The `drm_mode_config_cleanup()` is left solely in the top-level paths where it belongs.

**Minor nits:**

1. The commit message line `Issue found using a prototype static analysis tool` lacks a `Signed-off-by`-style trailer separator — it reads as part of the body but feels like a tagline. Convention is to either drop it or phrase it as part of the description.

2. The `Signed-off-by` from Slavin Liu (`220245772@seu.edu.cn`) — this person is listed as `Cc:` in the mail headers. If they co-developed the patch, a `Co-developed-by:` tag should precede their `Signed-off-by:` per kernel conventions (`Documentation/process/submitting-patches.rst`). If they only reviewed, `Reviewed-by:` would be more appropriate.

3. After the patch, `omap_modeset_fini()` becomes a trivial one-line wrapper around `omap_drm_irq_uninstall()`. This is fine — keeping the `_init`/`_fini` symmetry aids readability — but a reviewer could reasonably suggest inlining it. Not a blocker.

---
Generated by Claude Code Patch Reviewer