[PATCH] greybus: audio: ensure module is set to avoid crash on dev_err message

[PATCH] greybus: audio: ensure module is set to avoid crash on dev_err message

[Colin King]

From: Colin Ian King <colin.king@canonical.com>

Currently, if info is null, the dev_err message is dereferencing an
uninitialized module pointer.  Instead, initialize module before the
dev_err call to fix this issue.

Found using static analysis with cppcheck:
[drivers/staging/greybus/audio_topology.c:175]: (error)
  Uninitialized variable: module

Signed-off-by: Colin Ian King <colin.king@canonical.com>
---
 drivers/staging/greybus/audio_topology.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/greybus/audio_topology.c b/drivers/staging/greybus/audio_topology.c
index 5eef536..c43a959 100644
--- a/drivers/staging/greybus/audio_topology.c
+++ b/drivers/staging/greybus/audio_topology.c
@@ -171,6 +171,9 @@ static int gbcodec_mixer_ctl_info(struct snd_kcontrol *kcontrol,
 	data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
 	info = (struct gb_audio_ctl_elem_info *)data->info;
 
+	module = find_gb_module(gbcodec, kcontrol->id.name);
+	if (!module)
+		return -EINVAL;
 	if (!info) {
 		dev_err(module->dev, "NULL info for %s\n", uinfo->id.name);
 		return -EINVAL;
@@ -192,9 +195,6 @@ static int gbcodec_mixer_ctl_info(struct snd_kcontrol *kcontrol,
 		uinfo->value.enumerated.items = max;
 		if (uinfo->value.enumerated.item > max - 1)
 			uinfo->value.enumerated.item = max - 1;
-		module = find_gb_module(gbcodec, kcontrol->id.name);
-		if (!module)
-			return -EINVAL;
 		name = gbaudio_map_controlid(module, data->ctl_id,
 					     uinfo->value.enumerated.item);
 		strlcpy(uinfo->value.enumerated.name, name, NAME_SIZE);
-- 
2.9.3

Re: [PATCH] greybus: audio: ensure module is set to avoid crash on dev_err message

[Greg Kroah-Hartman]

On Fri, Sep 23, 2016 at 11:25:40AM +0100, Colin King wrote:
> From: Colin Ian King <colin.king@canonical.com>
> 
> Currently, if info is null, the dev_err message is dereferencing an
> uninitialized module pointer.  Instead, initialize module before the
> dev_err call to fix this issue.
> 
> Found using static analysis with cppcheck:
> [drivers/staging/greybus/audio_topology.c:175]: (error)
>   Uninitialized variable: module
> 
> Signed-off-by: Colin Ian King <colin.king@canonical.com>
> ---
>  drivers/staging/greybus/audio_topology.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/staging/greybus/audio_topology.c b/drivers/staging/greybus/audio_topology.c
> index 5eef536..c43a959 100644
> --- a/drivers/staging/greybus/audio_topology.c
> +++ b/drivers/staging/greybus/audio_topology.c
> @@ -171,6 +171,9 @@ static int gbcodec_mixer_ctl_info(struct snd_kcontrol *kcontrol,
>  	data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
>  	info = (struct gb_audio_ctl_elem_info *)data->info;
>  
> +	module = find_gb_module(gbcodec, kcontrol->id.name);
> +	if (!module)
> +		return -EINVAL;

How do you know you can get a module at this point in time, you haven't
looked at the type of info to know that?

I agree that there is an issue here, but I don't think this is the
correct fix.  Vaibhav?

Re: [PATCH] greybus: audio: ensure module is set to avoid crash on dev_err message

[Vaibhav Agarwal]

On Fri, Sep 23, 2016 at 10:28 PM, Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
> On Fri, Sep 23, 2016 at 11:25:40AM +0100, Colin King wrote:
>> From: Colin Ian King <colin.king@canonical.com>
>>
>> Currently, if info is null, the dev_err message is dereferencing an
>> uninitialized module pointer.  Instead, initialize module before the
>> dev_err call to fix this issue.
>>
>> Found using static analysis with cppcheck:
>> [drivers/staging/greybus/audio_topology.c:175]: (error)
>>   Uninitialized variable: module
>>
>> Signed-off-by: Colin Ian King <colin.king@canonical.com>
>> ---
>>  drivers/staging/greybus/audio_topology.c | 6 +++---
>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/staging/greybus/audio_topology.c b/drivers/staging/greybus/audio_topology.c
>> index 5eef536..c43a959 100644
>> --- a/drivers/staging/greybus/audio_topology.c
>> +++ b/drivers/staging/greybus/audio_topology.c
>> @@ -171,6 +171,9 @@ static int gbcodec_mixer_ctl_info(struct snd_kcontrol *kcontrol,
>>       data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
>>       info = (struct gb_audio_ctl_elem_info *)data->info;
>>
>> +     module = find_gb_module(gbcodec, kcontrol->id.name);
>> +     if (!module)
>> +             return -EINVAL;
>
> How do you know you can get a module at this point in time, you haven't
> looked at the type of info to know that?
>
> I agree that there is an issue here, but I don't think this is the
> correct fix.  Vaibhav?

Actually, fetching module is not related to info->type. However it is
only required in case of ENUMERATED element_type, thus used in
specific case. Also, I think it's better to use codec->dev in the err
message and align with other err_msg in this function.

Hi Colin, thanks for sharing this patch. Do you mind updating this
patch including my above suggestion? Otherwise, I can share a separate
patch.

--
thanks,
vaibhav
>

Re: [PATCH] greybus: audio: ensure module is set to avoid crash on dev_err message

[Colin Ian King]

On 23/09/16 19:20, Vaibhav Agarwal wrote:
> On Fri, Sep 23, 2016 at 10:28 PM, Greg Kroah-Hartman
> <gregkh@linuxfoundation.org> wrote:
>> On Fri, Sep 23, 2016 at 11:25:40AM +0100, Colin King wrote:
>>> From: Colin Ian King <colin.king@canonical.com>
>>>
>>> Currently, if info is null, the dev_err message is dereferencing an
>>> uninitialized module pointer.  Instead, initialize module before the
>>> dev_err call to fix this issue.
>>>
>>> Found using static analysis with cppcheck:
>>> [drivers/staging/greybus/audio_topology.c:175]: (error)
>>>   Uninitialized variable: module
>>>
>>> Signed-off-by: Colin Ian King <colin.king@canonical.com>
>>> ---
>>>  drivers/staging/greybus/audio_topology.c | 6 +++---
>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/staging/greybus/audio_topology.c b/drivers/staging/greybus/audio_topology.c
>>> index 5eef536..c43a959 100644
>>> --- a/drivers/staging/greybus/audio_topology.c
>>> +++ b/drivers/staging/greybus/audio_topology.c
>>> @@ -171,6 +171,9 @@ static int gbcodec_mixer_ctl_info(struct snd_kcontrol *kcontrol,
>>>       data = (struct gbaudio_ctl_pvt *)kcontrol->private_value;
>>>       info = (struct gb_audio_ctl_elem_info *)data->info;
>>>
>>> +     module = find_gb_module(gbcodec, kcontrol->id.name);
>>> +     if (!module)
>>> +             return -EINVAL;
>>
>> How do you know you can get a module at this point in time, you haven't
>> looked at the type of info to know that?
>>
>> I agree that there is an issue here, but I don't think this is the
>> correct fix.  Vaibhav?
> 
> Actually, fetching module is not related to info->type. However it is
> only required in case of ENUMERATED element_type, thus used in
> specific case. Also, I think it's better to use codec->dev in the err
> message and align with other err_msg in this function.
> 
> Hi Colin, thanks for sharing this patch. Do you mind updating this
> patch including my above suggestion? Otherwise, I can share a separate
> patch.

Vaibhav, I'm currently traveling, so won't be able to check this until
next week, so if you would rather send a correct fix sooner rather me
re-sending a fix then I'm fine with that.

Thanks, Colin
> 
> --
> thanks,
> vaibhav
>>