Re: [PATCH] tpm: Fix memory leak in tpmm_chip_alloc

["Guozihua (Scott)" <guozihua@huawei.com>]

On 2022/3/7 21:45, Jarkko Sakkinen wrote:
> On Mon, Mar 07, 2022 at 06:48:27PM +0800, GUO Zihua wrote:
>> Fix a memory leak in tpmm_chip_alloc. devm_add_action_or_reset would
>> call put_device on error, while tpm->devs is left untouched. Call
>> put_device on tpm->devs as well if devm_add_action_or_reset returns an
>> error.
>>
>> Fixes: fdc915f7f719 ("tpm: expose spaces via a device link /dev/tpmrm")
>> Signed-off-by: GUO Zihua <guozihua@huawei.com>
>> ---
>>   drivers/char/tpm/tpm-chip.c | 13 +++++++++----
>>   1 file changed, 9 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/char/tpm/tpm-chip.c b/drivers/char/tpm/tpm-chip.c
>> index b009e7479b70..0a92334e8c40 100644
>> --- a/drivers/char/tpm/tpm-chip.c
>> +++ b/drivers/char/tpm/tpm-chip.c
>> @@ -308,6 +308,12 @@ static int tpm_class_shutdown(struct device *dev)
>>   	return 0;
>>   }
>>   
>> +static void tpm_chip_free(struct tpm_chip *chip)
>> +{
>> +	put_device(&chip->devs);
>> +	put_device(&chip->dev);
>> +}
>> +
>>   /**
>>    * tpm_chip_alloc() - allocate a new struct tpm_chip instance
>>    * @pdev: device to which the chip is associated
>> @@ -396,8 +402,7 @@ struct tpm_chip *tpm_chip_alloc(struct device *pdev,
>>   	return chip;
>>   
>>   out:
>> -	put_device(&chip->devs);
>> -	put_device(&chip->dev);
>> +	tpm_chip_free(chip);
>>   	return ERR_PTR(rc);
>>   }
>>   EXPORT_SYMBOL_GPL(tpm_chip_alloc);
>> @@ -420,8 +425,8 @@ struct tpm_chip *tpmm_chip_alloc(struct device *pdev,
>>   		return chip;
>>   
>>   	rc = devm_add_action_or_reset(pdev,
>> -				      (void (*)(void *)) put_device,
>> -				      &chip->dev);
>> +				      (void (*)(void *)) tpm_chip_free,
>> +				      chip);
>>   	if (rc)
>>   		return ERR_PTR(rc);
>>   
>> -- 
>> 2.17.1
>>
> 
> Please test against the latest in
> 
> git://git.kernel.org/pub/scm/linux/kernel/git/jarkko/linux-tpmdd.git
> 
> and share your results.
> 
> BR, Jarkko
> .

Hi Jarkko,

I'll do that. Do we have a test set for TPM? Or do we just build and run 
it and see if everything works as expected?

This is an error handling optimization BTW.

-- 
Best
GUO Zihua