diff mbox series

ALSA: ac97: Fix possible NULL dereference in snd_ac97_mixer

Message ID 20230615021732.1972194-1-suhui@nfschina.com (mailing list archive)
State Accepted
Commit 79597c8bf64ca99eab385115743131d260339da5
Headers show
Series ALSA: ac97: Fix possible NULL dereference in snd_ac97_mixer | expand

Commit Message

Su Hui June 15, 2023, 2:17 a.m. UTC
smatch error:
sound/pci/ac97/ac97_codec.c:2354 snd_ac97_mixer() error:
we previously assumed 'rac97' could be null (see line 2072)

remove redundant assignment, return error if rac97 is NULL.

Fixes: da3cec35dd3c ("ALSA: Kill snd_assert() in sound/pci/*")
Signed-off-by: Su Hui <suhui@nfschina.com>
---
 sound/pci/ac97/ac97_codec.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

Christophe JAILLET Aug. 22, 2023, 8:07 p.m. UTC | #1
Le 15/06/2023 à 04:17, Su Hui a écrit :
> smatch error:
> sound/pci/ac97/ac97_codec.c:2354 snd_ac97_mixer() error:
> we previously assumed 'rac97' could be null (see line 2072)
> 
> remove redundant assignment, return error if rac97 is NULL.

Hi,

why is the assigment redundant?

Should an error occur, the 'struct snd_ac97 **' parameter was garanted 
to be set to NULL, now it is left as-is.

I've checked all callers and apparently this is fine because the probes 
fail if snd_ac97_mixer() returns an error.

However, some drivers with several mixers seem to rely on the value 
being NULL in case of error.

See [1] as an example of such code that forces a NULL value on its own, 
to be sure.

So, wouldn't it be safer to leave a "*rac97 = NULL;" just after the 
added sanity check?


CJ


[1]: 
https://elixir.bootlin.com/linux/v6.5-rc7/source/sound/pci/atiixp.c#L1438

> 
> Fixes: da3cec35dd3c ("ALSA: Kill snd_assert() in sound/pci/*")
> Signed-off-by: Su Hui <suhui@nfschina.com>
> ---
>   sound/pci/ac97/ac97_codec.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/sound/pci/ac97/ac97_codec.c b/sound/pci/ac97/ac97_codec.c
> index 9afc5906d662..80a65b8ad7b9 100644
> --- a/sound/pci/ac97/ac97_codec.c
> +++ b/sound/pci/ac97/ac97_codec.c
> @@ -2069,8 +2069,8 @@ int snd_ac97_mixer(struct snd_ac97_bus *bus, struct snd_ac97_template *template,
>   		.dev_disconnect =	snd_ac97_dev_disconnect,
>   	};
>   
> -	if (rac97)
> -		*rac97 = NULL;
> +	if (!rac97)
> +		return -EINVAL;
>   	if (snd_BUG_ON(!bus || !template))
>   		return -EINVAL;
>   	if (snd_BUG_ON(template->num >= 4))
Takashi Iwai Aug. 23, 2023, 2:37 p.m. UTC | #2
On Tue, 22 Aug 2023 22:07:40 +0200,
Christophe JAILLET wrote:
> 
> Le 15/06/2023 à 04:17, Su Hui a écrit :
> > smatch error:
> > sound/pci/ac97/ac97_codec.c:2354 snd_ac97_mixer() error:
> > we previously assumed 'rac97' could be null (see line 2072)
> > 
> > remove redundant assignment, return error if rac97 is NULL.
> 
> Hi,
> 
> why is the assigment redundant?

It's misleading, yeah.  Basically all callers are with non-NULL, hence
we took rather make it mandatory.  Maybe it should have been with
WARN_ON() to catch the NULL argument for an out-of-tree stuff. 

> Should an error occur, the 'struct snd_ac97 **' parameter was garanted
> to be set to NULL, now it is left as-is.
> 
> I've checked all callers and apparently this is fine because the
> probes fail if snd_ac97_mixer() returns an error.
> 
> However, some drivers with several mixers seem to rely on the value
> being NULL in case of error.
> 
> See [1] as an example of such code that forces a NULL value on its
> own, to be sure.
> 
> So, wouldn't it be safer to leave a "*rac97 = NULL;" just after the
> added sanity check?

Yes, we need the NULL initialization.
Care to submit an additional fix patch?


thanks,

Takashi

> 
> 
> CJ
> 
> 
> [1]:
> https://elixir.bootlin.com/linux/v6.5-rc7/source/sound/pci/atiixp.c#L1438
> 
> > 
> > Fixes: da3cec35dd3c ("ALSA: Kill snd_assert() in sound/pci/*")
> > Signed-off-by: Su Hui <suhui@nfschina.com>
> > ---
> >   sound/pci/ac97/ac97_codec.c | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/sound/pci/ac97/ac97_codec.c b/sound/pci/ac97/ac97_codec.c
> > index 9afc5906d662..80a65b8ad7b9 100644
> > --- a/sound/pci/ac97/ac97_codec.c
> > +++ b/sound/pci/ac97/ac97_codec.c
> > @@ -2069,8 +2069,8 @@ int snd_ac97_mixer(struct snd_ac97_bus *bus, struct snd_ac97_template *template,
> >   		.dev_disconnect =	snd_ac97_dev_disconnect,
> >   	};
> >   -	if (rac97)
> > -		*rac97 = NULL;
> > +	if (!rac97)
> > +		return -EINVAL;
> >   	if (snd_BUG_ON(!bus || !template))
> >   		return -EINVAL;
> >   	if (snd_BUG_ON(template->num >= 4))
>
Christophe JAILLET Aug. 23, 2023, 5:18 p.m. UTC | #3
Le 23/08/2023 à 16:37, Takashi Iwai a écrit :
> On Tue, 22 Aug 2023 22:07:40 +0200,
> Christophe JAILLET wrote:
>>
>> Le 15/06/2023 à 04:17, Su Hui a écrit :
>>> smatch error:
>>> sound/pci/ac97/ac97_codec.c:2354 snd_ac97_mixer() error:
>>> we previously assumed 'rac97' could be null (see line 2072)
>>>
>>> remove redundant assignment, return error if rac97 is NULL.
>>
>> Hi,
>>
>> why is the assigment redundant?
> 
> It's misleading, yeah.  Basically all callers are with non-NULL, hence
> we took rather make it mandatory.  Maybe it should have been with
> WARN_ON() to catch the NULL argument for an out-of-tree stuff.
> 
>> Should an error occur, the 'struct snd_ac97 **' parameter was garanted
>> to be set to NULL, now it is left as-is.
>>
>> I've checked all callers and apparently this is fine because the
>> probes fail if snd_ac97_mixer() returns an error.
>>
>> However, some drivers with several mixers seem to rely on the value
>> being NULL in case of error.
>>
>> See [1] as an example of such code that forces a NULL value on its
>> own, to be sure.
>>
>> So, wouldn't it be safer to leave a "*rac97 = NULL;" just after the
>> added sanity check?
> 
> Yes, we need the NULL initialization.
> Care to submit an additional fix patch?

Hi,

Su Hui already did.

CJ

> 
> 
> thanks,
> 
> Takashi
> 
>>
>>
>> CJ
>>
>>
>> [1]:
>> https://elixir.bootlin.com/linux/v6.5-rc7/source/sound/pci/atiixp.c#L1438
>>
>>>
>>> Fixes: da3cec35dd3c ("ALSA: Kill snd_assert() in sound/pci/*")
>>> Signed-off-by: Su Hui <suhui@nfschina.com>
>>> ---
>>>    sound/pci/ac97/ac97_codec.c | 4 ++--
>>>    1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/sound/pci/ac97/ac97_codec.c b/sound/pci/ac97/ac97_codec.c
>>> index 9afc5906d662..80a65b8ad7b9 100644
>>> --- a/sound/pci/ac97/ac97_codec.c
>>> +++ b/sound/pci/ac97/ac97_codec.c
>>> @@ -2069,8 +2069,8 @@ int snd_ac97_mixer(struct snd_ac97_bus *bus, struct snd_ac97_template *template,
>>>    		.dev_disconnect =	snd_ac97_dev_disconnect,
>>>    	};
>>>    -	if (rac97)
>>> -		*rac97 = NULL;
>>> +	if (!rac97)
>>> +		return -EINVAL;
>>>    	if (snd_BUG_ON(!bus || !template))
>>>    		return -EINVAL;
>>>    	if (snd_BUG_ON(template->num >= 4))
>>
>
diff mbox series

Patch

diff --git a/sound/pci/ac97/ac97_codec.c b/sound/pci/ac97/ac97_codec.c
index 9afc5906d662..80a65b8ad7b9 100644
--- a/sound/pci/ac97/ac97_codec.c
+++ b/sound/pci/ac97/ac97_codec.c
@@ -2069,8 +2069,8 @@  int snd_ac97_mixer(struct snd_ac97_bus *bus, struct snd_ac97_template *template,
 		.dev_disconnect =	snd_ac97_dev_disconnect,
 	};
 
-	if (rac97)
-		*rac97 = NULL;
+	if (!rac97)
+		return -EINVAL;
 	if (snd_BUG_ON(!bus || !template))
 		return -EINVAL;
 	if (snd_BUG_ON(template->num >= 4))