drm/msm: Remove dead code in msm_ioctl_gem_submit()

[RFC PATCH] drm/msm: Remove dead code in msm_ioctl_gem_submit()

Posted by I Hsin Cheng 7 months, 2 weeks ago

According to the report of Coverity Scan [1], "sync_file" is going to be
NULL when entering the "if" section after "out_post_unlock", so
"fput(sync_file->file)" is never going to be exected in this block.

[1]: https://scan5.scan.coverity.com/#/project-view/10074/10063?selectedIssue=1655089
Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
---
 drivers/gpu/drm/msm/msm_gem_submit.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c
index d4f71bb54e84..cba1dc6fe6c6 100644
--- a/drivers/gpu/drm/msm/msm_gem_submit.c
+++ b/drivers/gpu/drm/msm/msm_gem_submit.c
@@ -904,8 +904,6 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data,
 out_post_unlock:
 	if (ret && (out_fence_fd >= 0)) {
 		put_unused_fd(out_fence_fd);
-		if (sync_file)
-			fput(sync_file->file);
 	}
 
 	if (!IS_ERR_OR_NULL(submit)) {
-- 
2.43.0

Re: [RFC PATCH] drm/msm: Remove dead code in msm_ioctl_gem_submit()

Posted by Shuah Khan 7 months, 1 week ago

On 6/23/25 12:47, I Hsin Cheng wrote:
> According to the report of Coverity Scan [1], "sync_file" is going to be
> NULL when entering the "if" section after "out_post_unlock", so
> "fput(sync_file->file)" is never going to be exected in this block.
> 
> [1]: https://scan5.scan.coverity.com/#/project-view/10074/10063?selectedIssue=1655089
> Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
> ---
>   drivers/gpu/drm/msm/msm_gem_submit.c | 2 --
>   1 file changed, 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c
> index d4f71bb54e84..cba1dc6fe6c6 100644
> --- a/drivers/gpu/drm/msm/msm_gem_submit.c
> +++ b/drivers/gpu/drm/msm/msm_gem_submit.c
> @@ -904,8 +904,6 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data,
>   out_post_unlock:
>   	if (ret && (out_fence_fd >= 0)) {
>   		put_unused_fd(out_fence_fd);
> -		if (sync_file)
> -			fput(sync_file->file);

Are you sure you want delete these two lines? It might not make
sense to check sync_file inside if (ret && (out_fence_fd >= 0)),
but it is ncecessary to fput.

>   	}
>   
>   	if (!IS_ERR_OR_NULL(submit)) {

Check the error paths carefully to see if this is indeed the right fix.

thanks,
-- Shuah

Re: [RFC PATCH] drm/msm: Remove dead code in msm_ioctl_gem_submit()

Posted by Rob Clark 7 months, 1 week ago

On Tue, Jul 1, 2025 at 12:38 PM Shuah Khan <skhan@linuxfoundation.org> wrote:
>
> On 6/23/25 12:47, I Hsin Cheng wrote:
> > According to the report of Coverity Scan [1], "sync_file" is going to be
> > NULL when entering the "if" section after "out_post_unlock", so
> > "fput(sync_file->file)" is never going to be exected in this block.
> >
> > [1]: https://scan5.scan.coverity.com/#/project-view/10074/10063?selectedIssue=1655089
> > Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
> > ---
> >   drivers/gpu/drm/msm/msm_gem_submit.c | 2 --
> >   1 file changed, 2 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c
> > index d4f71bb54e84..cba1dc6fe6c6 100644
> > --- a/drivers/gpu/drm/msm/msm_gem_submit.c
> > +++ b/drivers/gpu/drm/msm/msm_gem_submit.c
> > @@ -904,8 +904,6 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data,
> >   out_post_unlock:
> >       if (ret && (out_fence_fd >= 0)) {
> >               put_unused_fd(out_fence_fd);
> > -             if (sync_file)
> > -                     fput(sync_file->file);
>
> Are you sure you want delete these two lines? It might not make
> sense to check sync_file inside if (ret && (out_fence_fd >= 0)),
> but it is ncecessary to fput.

fwiw, there is at least about to be a code path where this error
handling is not dead, once the VM_BIND series is merged

BR,
-R

> >       }
> >
> >       if (!IS_ERR_OR_NULL(submit)) {
>
> Check the error paths carefully to see if this is indeed the right fix.
>
> thanks,
> -- Shuah