Re: Wrong/strange TPM patches was Re: [PATCH 6.1 000/119] 6.1.31-rc1 review

[Lino Sanfilippo]

Hi,


On 30.05.23 15:02, Greg Kroah-Hartman wrote:
> ATTENTION: This e-mail is from an external sender. Please check attachments and links before opening e.g. with mouseover.
> 
> 
> On Tue, May 30, 2023 at 12:46:49PM +0200, Pavel Machek wrote:
>> Hi!
>>
>>> This is the start of the stable review cycle for the 6.1.31 release.
>>> There are 119 patches in this series, all will be posted as a response
>>> to this one.  If anyone has any issues with these being applied, please
>>> let me know.
>>
>>> Lino Sanfilippo <l.sanfilippo@xxxxxxxxxx>
>>>     tpm, tpm_tis: Avoid cache incoherency in test for interrupts
>>
>> Description on this one is wrong/confused. There's no cache problem in
>> the code. Plus test_bit and friend already use bit number, so
>>
>> -       bool itpm = priv->flags & TPM_TIS_ITPM_WORKAROUND;
>> +       bool itpm = test_bit(TPM_TIS_ITPM_WORKAROUND, &priv->flags);
>>
>> @@ -87,6 +87,7 @@ enum tpm_tis_flags {
>>         TPM_TIS_ITPM_WORKAROUND         = BIT(0),
>>         TPM_TIS_INVALID_STATUS          = BIT(1),
>>         TPM_TIS_DEFAULT_CANCELLATION    = BIT(2),
>> +       TPM_TIS_IRQ_TESTED              = BIT(3),
>>  };
>>
>> this enum needs to go from BIT() to raw numbers.
>>
>> You can just do return tpm_pm_resume();
>>
>>> Jarkko Sakkinen <jarkko@xxxxxxxxxx>
>>>     tpm: Prevent hwrng from activating during resume
>>
>> @@ -429,6 +431,14 @@ int tpm_pm_resume(struct device *dev)
>>         if (chip == NULL)
>>                 return -ENODEV;
>>
>> +       chip->flags &= ~TPM_CHIP_FLAG_SUSPENDED;
>> +
>> +       /*
>> +        * Guarantee that SUSPENDED is written last, so that hwrng does not
>> +        * activate before the chip has been fully resumed.
>> +        */
>> +       wmb();
>> +
>>         return 0;
>>  }
>>  EXPORT_SYMBOL_GPL(tpm_pm_resume);
>>
>> This code is confused. First, either you don't need memory barriers
>> here, or you need real locking. Second, if you want to guarantee flags
>> are written last, you need to put the barrier before the
>> assignment. (But ... get rid of that confusion, first).
> 
> Care to submit patches to resolve this?  It's this way in Linus's tree
> now from what I can tell, and these changes were needed for another
> stable-marked change, so I'll leave them in for now.
> 

First, thanks for the review Pavel and for spotting this.
I will send a patch to fix the enums.

Regards,
Lino