2015-04-27 15:37:13

by Oleg Drokin

Subject: [PATCH v2] drm: fix a memleak on mutex failure path

From: Oleg Drokin <[email protected]>

Need to free just allocated ctx allocation if we cannot
get our config mutex.

This one has been flagged by kbuild bot all the way back in August,
but somehow nobody picked it up:
https://lists.01.org/pipermail/kbuild/2014-August/001691.html

In addition there is another failure path that leaks the same
ctx reference that is fixed.

Found with smatch.

Signed-off-by: Oleg Drokin <[email protected]>
CC: Daniel Vetter <[email protected]>
---
drivers/gpu/drm/drm_modeset_lock.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/drm_modeset_lock.c b/drivers/gpu/drm/drm_modeset_lock.c
index 51cc47d..c0a5cd8 100644
--- a/drivers/gpu/drm/drm_modeset_lock.c
+++ b/drivers/gpu/drm/drm_modeset_lock.c
@@ -80,8 +80,10 @@ int __drm_modeset_lock_all(struct drm_device *dev,
return -ENOMEM;

if (trylock) {
- if (!mutex_trylock(&config->mutex))
- return -EBUSY;
+ if (!mutex_trylock(&config->mutex)) {
+ ret = -EBUSY;
+ goto out;
+ }
} else {
mutex_lock(&config->mutex);
}
@@ -114,6 +116,8 @@ fail:
goto retry;
}

+out:
+ kfree(ctx);
return ret;
}
EXPORT_SYMBOL(__drm_modeset_lock_all);
--
2.1.0

2015-04-28 07:23:52

by Jani Nikula

Subject: Re: [PATCH v2] drm: fix a memleak on mutex failure path

On Mon, 27 Apr 2015, [email protected] wrote:
> From: Oleg Drokin <[email protected]>
>
> Need to free just allocated ctx allocation if we cannot
> get our config mutex.
>
> This one has been flagged by kbuild bot all the way back in August,
> but somehow nobody picked it up:
> https://lists.01.org/pipermail/kbuild/2014-August/001691.html
>
> In addition there is another failure path that leaks the same
> ctx reference that is fixed.
>
> Found with smatch.
>
> Signed-off-by: Oleg Drokin <[email protected]>
> CC: Daniel Vetter <[email protected]>

Reviewed-by: Jani Nikula <[email protected]>


> ---
> drivers/gpu/drm/drm_modeset_lock.c | 8 ++++++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/drm_modeset_lock.c b/drivers/gpu/drm/drm_modeset_lock.c
> index 51cc47d..c0a5cd8 100644
> --- a/drivers/gpu/drm/drm_modeset_lock.c
> +++ b/drivers/gpu/drm/drm_modeset_lock.c
> @@ -80,8 +80,10 @@ int __drm_modeset_lock_all(struct drm_device *dev,
> return -ENOMEM;
>
> if (trylock) {
> - if (!mutex_trylock(&config->mutex))
> - return -EBUSY;
> + if (!mutex_trylock(&config->mutex)) {
> + ret = -EBUSY;
> + goto out;
> + }
> } else {
> mutex_lock(&config->mutex);
> }
> @@ -114,6 +116,8 @@ fail:
> goto retry;
> }
>
> +out:
> + kfree(ctx);
> return ret;
> }
> EXPORT_SYMBOL(__drm_modeset_lock_all);
> --
> 2.1.0
>

--
Jani Nikula, Intel Open Source Technology Center

2015-05-05 07:25:07

by Daniel Vetter

Subject: Re: [PATCH v2] drm: fix a memleak on mutex failure path

On Tue, Apr 28, 2015 at 10:25:46AM +0300, Jani Nikula wrote:
> On Mon, 27 Apr 2015, [email protected] wrote:
> > From: Oleg Drokin <[email protected]>
> >
> > Need to free just allocated ctx allocation if we cannot
> > get our config mutex.
> >
> > This one has been flagged by kbuild bot all the way back in August,
> > but somehow nobody picked it up:
> > https://lists.01.org/pipermail/kbuild/2014-August/001691.html
> >
> > In addition there is another failure path that leaks the same
> > ctx reference that is fixed.
> >
> > Found with smatch.
> >
> > Signed-off-by: Oleg Drokin <[email protected]>
> > CC: Daniel Vetter <[email protected]>
>
> Reviewed-by: Jani Nikula <[email protected]>

Applied to topic/drm-misc, thanks.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch