In snd_skl_parse_uuids if allocation for module->instance_id fails, the
allocated memory for module shoulde be released. I changes the
allocation for module to use devm_kzalloc to be resource_managed
allocation and avoid the release in error path.
Signed-off-by: Navid Emamdoost <[email protected]>
---
Changes in v2:
- Changed the allocation for module from kzalloc to devm_kzalloc
---
sound/soc/intel/skylake/skl-sst-utils.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/sound/soc/intel/skylake/skl-sst-utils.c b/sound/soc/intel/skylake/skl-sst-utils.c
index d43cbf4a71ef..ac37f04b0eea 100644
--- a/sound/soc/intel/skylake/skl-sst-utils.c
+++ b/sound/soc/intel/skylake/skl-sst-utils.c
@@ -284,7 +284,7 @@ int snd_skl_parse_uuids(struct sst_dsp *ctx, const struct firmware *fw,
*/
for (i = 0; i < num_entry; i++, mod_entry++) {
- module = kzalloc(sizeof(*module), GFP_KERNEL);
+ module = devm_kzalloc(ctx->dev, sizeof(*module), GFP_KERNEL);
if (!module) {
ret = -ENOMEM;
goto free_uuid_list;
--
2.17.1
On 9/25/19 11:19 AM, Navid Emamdoost wrote:
> In snd_skl_parse_uuids if allocation for module->instance_id fails, the
> allocated memory for module shoulde be released. I changes the
> allocation for module to use devm_kzalloc to be resource_managed
> allocation and avoid the release in error path.
if you use devm_, don't you need to fix the error path as well then, I
see a kfree(uuid) in skl_freeup_uuid_list().
I am not very familiar with this code but the error seems to be that the
list_add_tail() is called after the module->instance_id is allocated, so
there is a risk that the module allocated earlier is not freed (since
it's not yet added to the list). Freeing the module as done in patch 1
works, using devm_ without fixing the error path does not seem correct
to me.
>
> Signed-off-by: Navid Emamdoost <[email protected]>
> ---
> Changes in v2:
> - Changed the allocation for module from kzalloc to devm_kzalloc
> ---
> sound/soc/intel/skylake/skl-sst-utils.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/sound/soc/intel/skylake/skl-sst-utils.c b/sound/soc/intel/skylake/skl-sst-utils.c
> index d43cbf4a71ef..ac37f04b0eea 100644
> --- a/sound/soc/intel/skylake/skl-sst-utils.c
> +++ b/sound/soc/intel/skylake/skl-sst-utils.c
> @@ -284,7 +284,7 @@ int snd_skl_parse_uuids(struct sst_dsp *ctx, const struct firmware *fw,
> */
>
> for (i = 0; i < num_entry; i++, mod_entry++) {
> - module = kzalloc(sizeof(*module), GFP_KERNEL);
> + module = devm_kzalloc(ctx->dev, sizeof(*module), GFP_KERNEL);
> if (!module) {
> ret = -ENOMEM;
> goto free_uuid_list;
>
On Wed, Sep 25, 2019 at 12:05:28PM -0500, Pierre-Louis Bossart wrote:
> On 9/25/19 11:19 AM, Navid Emamdoost wrote:
> > In snd_skl_parse_uuids if allocation for module->instance_id fails, the
> > allocated memory for module shoulde be released. I changes the
> > allocation for module to use devm_kzalloc to be resource_managed
> > allocation and avoid the release in error path.
>
> if you use devm_, don't you need to fix the error path as well then, I see a
> kfree(uuid) in skl_freeup_uuid_list().
>
> I am not very familiar with this code but the error seems to be that the
> list_add_tail() is called after the module->instance_id is allocated, so
> there is a risk that the module allocated earlier is not freed (since it's
> not yet added to the list). Freeing the module as done in patch 1 works,
> using devm_ without fixing the error path does not seem correct to me.
>
Thanks for the feedback, then it's your call if you can accept patch 1 as
fix.
Navid.
> >
> > Signed-off-by: Navid Emamdoost <[email protected]>
> > ---
> > Changes in v2:
> > - Changed the allocation for module from kzalloc to devm_kzalloc
> > ---
> > sound/soc/intel/skylake/skl-sst-utils.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/sound/soc/intel/skylake/skl-sst-utils.c b/sound/soc/intel/skylake/skl-sst-utils.c
> > index d43cbf4a71ef..ac37f04b0eea 100644
> > --- a/sound/soc/intel/skylake/skl-sst-utils.c
> > +++ b/sound/soc/intel/skylake/skl-sst-utils.c
> > @@ -284,7 +284,7 @@ int snd_skl_parse_uuids(struct sst_dsp *ctx, const struct firmware *fw,
> > */
> > for (i = 0; i < num_entry; i++, mod_entry++) {
> > - module = kzalloc(sizeof(*module), GFP_KERNEL);
> > + module = devm_kzalloc(ctx->dev, sizeof(*module), GFP_KERNEL);
> > if (!module) {
> > ret = -ENOMEM;
> > goto free_uuid_list;
> >
>
On 9/26/19 9:55 PM, Navid Emamdoost wrote:
> On Wed, Sep 25, 2019 at 12:05:28PM -0500, Pierre-Louis Bossart wrote:
>> On 9/25/19 11:19 AM, Navid Emamdoost wrote:
>>> In snd_skl_parse_uuids if allocation for module->instance_id fails, the
>>> allocated memory for module shoulde be released. I changes the
>>> allocation for module to use devm_kzalloc to be resource_managed
>>> allocation and avoid the release in error path.
>>
>> if you use devm_, don't you need to fix the error path as well then, I see a
>> kfree(uuid) in skl_freeup_uuid_list().
>>
>> I am not very familiar with this code but the error seems to be that the
>> list_add_tail() is called after the module->instance_id is allocated, so
>> there is a risk that the module allocated earlier is not freed (since it's
>> not yet added to the list). Freeing the module as done in patch 1 works,
>> using devm_ without fixing the error path does not seem correct to me.
>>
> Thanks for the feedback, then it's your call if you can accept patch 1 as
> fix.
Cezary, it's really your call.
> Navid.
>>>
>>> Signed-off-by: Navid Emamdoost <[email protected]>
>>> ---
>>> Changes in v2:
>>> - Changed the allocation for module from kzalloc to devm_kzalloc
>>> ---
>>> sound/soc/intel/skylake/skl-sst-utils.c | 2 +-
>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/sound/soc/intel/skylake/skl-sst-utils.c b/sound/soc/intel/skylake/skl-sst-utils.c
>>> index d43cbf4a71ef..ac37f04b0eea 100644
>>> --- a/sound/soc/intel/skylake/skl-sst-utils.c
>>> +++ b/sound/soc/intel/skylake/skl-sst-utils.c
>>> @@ -284,7 +284,7 @@ int snd_skl_parse_uuids(struct sst_dsp *ctx, const struct firmware *fw,
>>> */
>>> for (i = 0; i < num_entry; i++, mod_entry++) {
>>> - module = kzalloc(sizeof(*module), GFP_KERNEL);
>>> + module = devm_kzalloc(ctx->dev, sizeof(*module), GFP_KERNEL);
>>> if (!module) {
>>> ret = -ENOMEM;
>>> goto free_uuid_list;
>>>
>>
On 2019-09-27 15:14, Pierre-Louis Bossart wrote:
> On 9/26/19 9:55 PM, Navid Emamdoost wrote:
>> On Wed, Sep 25, 2019 at 12:05:28PM -0500, Pierre-Louis Bossart wrote:
>>> On 9/25/19 11:19 AM, Navid Emamdoost wrote:
>>>> In snd_skl_parse_uuids if allocation for module->instance_id fails, the
>>>> allocated memory for module shoulde be released. I changes the
>>>> allocation for module to use devm_kzalloc to be resource_managed
>>>> allocation and avoid the release in error path.
>>>
>>> if you use devm_, don't you need to fix the error path as well then,
>>> I see a
>>> kfree(uuid) in skl_freeup_uuid_list().
>>>
>>> I am not very familiar with this code but the error seems to be that the
>>> list_add_tail() is called after the module->instance_id is allocated, so
>>> there is a risk that the module allocated earlier is not freed (since
>>> it's
>>> not yet added to the list). Freeing the module as done in patch 1 works,
>>> using devm_ without fixing the error path does not seem correct to me.
>>>
Good catch, Pierre.
>> Thanks for the feedback, then it's your call if you can accept patch 1 as
>> fix.
>
> Cezary, it's really your call.
>
Actually, not the best person to ask about "objective decisions" here as
my vision is clouded by changes done internally. This code no longer
exists in our internal repo. It's better for host to send MODULE_INFO
request rather than understanding firmware binary structure and parse it
directly.
I'm fine with solution #1 as I guess asking to wait for refactor is not
an option. Code deployment is delayed due to range of administrative
decisions, some of which should be uncovered on alsa-devel soon enough.
Czarek
On Fri, Sep 27, 2019 at 05:10:18PM +0200, Cezary Rojewski wrote:
> I'm fine with solution #1 as I guess asking to wait for refactor is not an
> option. Code deployment is delayed due to range of administrative decisions,
> some of which should be uncovered on alsa-devel soon enough.
The problem with solution #1 is freeing orphaned pointer. It will work,
but it's simple is not okay from object life time prospective.
--
With Best Regards,
Andy Shevchenko
> The problem with solution #1 is freeing orphaned pointer. It will work,
> but it's simple is not okay from object life time prospective.
?? I don't get your point at all Andy.
Two allocations happens in a loop and if the second fails, you free the
first and then jump to free everything allocated in the previous
iterations. what am I missing?
On Fri, Sep 27, 2019 at 7:39 PM Pierre-Louis Bossart
<[email protected]> wrote:
> > The problem with solution #1 is freeing orphaned pointer. It will work,
> > but it's simple is not okay from object life time prospective.
>
> ?? I don't get your point at all Andy.
> Two allocations happens in a loop and if the second fails, you free the
> first and then jump to free everything allocated in the previous
> iterations. what am I missing?
Two things:
- one allocation is done with kzalloc(), while the other one with
devm_kcalloc()
- due to above the ordering of resources is reversed
--
With Best Regards,
Andy Shevchenko
On 9/27/19 3:39 PM, Andy Shevchenko wrote:
> On Fri, Sep 27, 2019 at 7:39 PM Pierre-Louis Bossart
> <[email protected]> wrote:
>>> The problem with solution #1 is freeing orphaned pointer. It will work,
>>> but it's simple is not okay from object life time prospective.
>>
>> ?? I don't get your point at all Andy.
>> Two allocations happens in a loop and if the second fails, you free the
>> first and then jump to free everything allocated in the previous
>> iterations. what am I missing?
>
> Two things:
> - one allocation is done with kzalloc(), while the other one with
> devm_kcalloc()
> - due to above the ordering of resources is reversed
Ah yes, I see your point now, sorry for being thick.
Indeed it'd make sense to use devm_ for both allocations, but then the
kfree needs to be removed in the error handling.