Re: [PATCH 1/2] tpm, tpm_tis: Handle interrupt storm

[Lino Sanfilippo]

On 24.05.23 17:30, Jarkko Sakkinen wrote:
> ATTENTION: This e-mail is from an external sender. Please check attachments and links before opening e.g. with mouseover.
> 
> 
> On Wed May 24, 2023 at 6:58 AM EEST, Jarkko Sakkinen wrote:
>>>     rc = tpm_tis_read32(priv, TPM_INT_STATUS(priv->locality), &interrupt);
>>>     if (rc < 0)
>>> -           return IRQ_NONE;
>>> +           goto unhandled;
>>>
>>>     if (interrupt == 0)
>>> -           return IRQ_NONE;
>>> +           goto unhandled;
>>>
>>>     set_bit(TPM_TIS_IRQ_TESTED, &priv->flags);
>>>     if (interrupt & TPM_INTF_DATA_AVAIL_INT)
>>> @@ -780,10 +829,14 @@ static irqreturn_t tis_int_handler(int dummy, void *dev_id)
>>>     rc = tpm_tis_write32(priv, TPM_INT_STATUS(priv->locality), interrupt);
>>>     tpm_tis_relinquish_locality(chip, 0);
>>>     if (rc < 0)
>>> -           return IRQ_NONE;
>>> +           goto unhandled;
>>>
>>>     tpm_tis_read32(priv, TPM_INT_STATUS(priv->locality), &interrupt);
>>>     return IRQ_HANDLED;
>>> +
>>> +unhandled:
>>> +   tpm_tis_process_unhandled_interrupt(chip);
>>> +   return IRQ_HANDLED;
>>>  }
> 
> Some minor glitches I noticed.
> 
> You could simplify the flow by making the helper to return IRQ_NONE.
> 
> E.g.
> 
>         tpm_tis_relinquish_locality(chip, 0);
>         if (rc < 0)
>                 return tpm_tis_process_unhandled_interrupt(chip);

Agreed, this way we could spare a few lines in the interrupt handler (but note
that the implementation only returns IRQ_HANDLED never IRQ_NONE. This is to prevent
the generic irq code from doing its own interrupt storm handling before the TPM driver
had a chance to fall back to polling).

> 
> I'd recommend changing the function name simply tpm_tis_rollback_interrupt().

> Also tpm_tis_handle_irq_storm() is a pretty bad function name
> because handle also can mean anything. You are resetting to the
> polling mode, right?
> 
> So perhaps that could be e.g. tpm_tis_reenable_polling? I'm open
> for any other name but it really needs to give a hint what the
> function does.

tpm_tis_reenable_polling() sounds good to me.


Regards,
Lino