[PATCH] drm/msm/adreno: fix autosuspend cleanup during teardown

[PATCH] drm/msm/adreno: fix autosuspend cleanup during teardown

[Guangshuo Li]

adreno_gpu_init() calls pm_runtime_use_autosuspend(), but
adreno_gpu_cleanup() does not call the matching
pm_runtime_dont_use_autosuspend() during teardown.

If the autosuspend delay is set to a negative value while autosuspend
is enabled, the runtime PM core increments usage_count to prevent
runtime suspend. Without calling pm_runtime_dont_use_autosuspend()
during teardown, this reference is not dropped and usage_count remains
unbalanced.

The documentation for pm_runtime_use_autosuspend() also notes that it
is important to undo it with pm_runtime_dont_use_autosuspend() at
driver exit time, unless runtime PM was initially enabled with
devm_pm_runtime_enable().

Add the missing pm_runtime_dont_use_autosuspend() call to
adreno_gpu_cleanup().

This issue was found by manual code inspection.

Fixes: eeb754746b14 ("drm/msm/gpu: use pm-runtime")
Cc: stable@vger.kernel.org
Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
---
 drivers/gpu/drm/msm/adreno/adreno_gpu.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
index c62c45bb0ddb..29c59899fdef 100644
--- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
@@ -1252,6 +1252,8 @@ void adreno_gpu_cleanup(struct adreno_gpu *adreno_gpu)
 	for (i = 0; i < ARRAY_SIZE(adreno_gpu->info->fw); i++)
 		release_firmware(adreno_gpu->fw[i]);
 
+	pm_runtime_dont_use_autosuspend(&gpu->pdev->dev);
+
 	if (priv && pm_runtime_enabled(&priv->gpu_pdev->dev))
 		pm_runtime_disable(&priv->gpu_pdev->dev);
 
-- 
2.43.0

Re: [PATCH] drm/msm/adreno: fix autosuspend cleanup during teardown

[sashiko-bot]

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

Pre-existing issues:
- [High] The GPU object and its associated resources are leaked if `dev_pm_opp_of_find_icc_paths()` fails in `adreno_bind()`.
--

--- Patch [1]: [PATCH] drm/msm/adreno: fix autosuspend cleanup during teardown ---
commit d01531efa3b8df1b3ab17728c644cfbadd605a8f
Author: Guangshuo Li <lgs201920130244@gmail.com>

drm/msm/adreno: fix autosuspend cleanup during teardown

This commit adds a missing pm_runtime_dont_use_autosuspend() call to
adreno_gpu_cleanup() to properly drop the usage count during teardown.

> diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
> index 3370cd44382fe..f4b22e04f9135 100644
> --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
> +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
> @@ -1261,6 +1261,8 @@ void adreno_gpu_cleanup(struct adreno_gpu *adreno_gpu)
>  	for (i = 0; i < ARRAY_SIZE(adreno_gpu->info->fw); i++)
>  		release_firmware(adreno_gpu->fw[i]);
>  
> +	pm_runtime_dont_use_autosuspend(&gpu->pdev->dev);
> +
>  	if (priv && pm_runtime_enabled(&priv->gpu_pdev->dev))
>  		pm_runtime_disable(&priv->gpu_pdev->dev);

[Severity: High]
This isn't a regression introduced by this patch, but I noticed a
potential leak in adreno_bind().

Looking at adreno_bind() in drivers/gpu/drm/msm/adreno/adreno_device.c:

drivers/gpu/drm/msm/adreno/adreno_device.c:adreno_bind() {
    ...
	gpu = info->funcs->init(drm);
	if (IS_ERR(gpu)) {
		dev_warn(drm->dev, "failed to load adreno gpu\n");
		return PTR_ERR(gpu);
	}

	ret = dev_pm_opp_of_find_icc_paths(dev, NULL);
	if (ret)
		return ret;
    ...
}

During component binding in adreno_bind(), if dev_pm_opp_of_find_icc_paths()
fails (which can happen with -EPROBE_DEFER), it returns immediately without
calling gpu->funcs->destroy(gpu).

Because component binding failed, the component framework will not call
adreno_unbind(), which leaves resources allocated during
info->funcs->init(drm) (like a6xx_gpu_init()) leaked. This happens on
every probe deferral.

Does this code leak the GPU object and its associated resources if
dev_pm_opp_of_find_icc_paths() fails?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260808131624.2854412-1-lgs201920130244@gmail.com?part=1

Re: [PATCH] drm/msm/adreno: fix autosuspend cleanup during teardown

[Konrad Dybcio]

On 8/8/26 3:16 PM, Guangshuo Li wrote:
> adreno_gpu_init() calls pm_runtime_use_autosuspend(), but
> adreno_gpu_cleanup() does not call the matching
> pm_runtime_dont_use_autosuspend() during teardown.
> 
> If the autosuspend delay is set to a negative value while autosuspend
> is enabled, the runtime PM core increments usage_count to prevent
> runtime suspend. Without calling pm_runtime_dont_use_autosuspend()
> during teardown, this reference is not dropped and usage_count remains
> unbalanced.
> 
> The documentation for pm_runtime_use_autosuspend() also notes that it
> is important to undo it with pm_runtime_dont_use_autosuspend() at
> driver exit time, unless runtime PM was initially enabled with
> devm_pm_runtime_enable().
> 
> Add the missing pm_runtime_dont_use_autosuspend() call to
> adreno_gpu_cleanup().
> 
> This issue was found by manual code inspection.
> 
> Fixes: eeb754746b14 ("drm/msm/gpu: use pm-runtime")
> Cc: stable@vger.kernel.org
> Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
> ---

I think we can use devm_pm_runtime_enable() instead, which will take
care of cleaning up both the autosuspend and runpm halves

Konrad

Re: [PATCH] drm/msm/adreno: fix autosuspend cleanup during teardown

[Krzysztof Kozlowski]

On 08/08/2026 15:16, Guangshuo Li wrote:
> adreno_gpu_init() calls pm_runtime_use_autosuspend(), but
> adreno_gpu_cleanup() does not call the matching
> pm_runtime_dont_use_autosuspend() during teardown.
> 
> If the autosuspend delay is set to a negative value while autosuspend
> is enabled, the runtime PM core increments usage_count to prevent
> runtime suspend. Without calling pm_runtime_dont_use_autosuspend()
> during teardown, this reference is not dropped and usage_count remains
> unbalanced.
> 
> The documentation for pm_runtime_use_autosuspend() also notes that it
> is important to undo it with pm_runtime_dont_use_autosuspend() at
> driver exit time, unless runtime PM was initially enabled with
> devm_pm_runtime_enable().
> 
> Add the missing pm_runtime_dont_use_autosuspend() call to
> adreno_gpu_cleanup().
> 
> This issue was found by manual code inspection.
> 
> Fixes: eeb754746b14 ("drm/msm/gpu: use pm-runtime")
> Cc: stable@vger.kernel.org
> Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
> ---

You sent vast amount of patches, all separate, making it very difficult
to track and respond in efficient way. Do not do that.

Group your work per subsystem.

You were asked to clarify and respond to incorrect fixes statement. I do
not see how you clarified and responded at all.


Best regards,
Krzysztof