[PATCH 3/3 v3] powerpc/mpic: FSL MPIC error interrupt support.

[Scott Wood]

On 08/03/2012 02:32 PM, Sethi Varun-B16395 wrote:
> 
> 
>> -----Original Message-----
>> From: Kumar Gala [mailto:galak at kernel.crashing.org]
>> Sent: Saturday, August 04, 2012 12:55 AM
>> To: Sethi Varun-B16395
>> Cc: linuxppc-dev at lists.ozlabs.org; Hamciuc Bogdan-BHAMCIU1
>> Subject: Re: [PATCH 3/3 v3] powerpc/mpic: FSL MPIC error interrupt
>> support.
>>
>>
>> On Aug 3, 2012, at 1:52 PM, Sethi Varun-B16395 wrote:
>>
>>>
>>>
>>>> -----Original Message-----
>>>> From: Kumar Gala [mailto:galak at kernel.crashing.org]
>>>> Sent: Friday, August 03, 2012 6:49 PM
>>>> To: Sethi Varun-B16395
>>>> Cc: linuxppc-dev at lists.ozlabs.org; Hamciuc Bogdan-BHAMCIU1
>>>> Subject: Re: [PATCH 3/3 v3] powerpc/mpic: FSL MPIC error interrupt
>>>> support.
>>>>> diff --git a/arch/powerpc/sysdev/mpic.c b/arch/powerpc/sysdev/mpic.c
>>>>> index 7e32db7..2a0b632 100644
>>>>> --- a/arch/powerpc/sysdev/mpic.c
>>>>> +++ b/arch/powerpc/sysdev/mpic.c
>>>>> @@ -1026,6 +1026,9 @@ static int mpic_host_map(struct irq_domain *h,
>>>> unsigned int virq,
>>>>> 		return 0;
>>>>> 	}
>>>>>
>>>>> +	if (mpic_map_error_int(mpic, virq, hw))
>>>>> +		return 0;
>>>>> +
>>>>> 	if (hw >= mpic->num_sources)
>>>>> 		return -EINVAL;
>>>>>
>>>>> @@ -1085,7 +1088,16 @@ static int mpic_host_xlate(struct irq_domain
>>>>> *h,
>>>> struct device_node *ct,
>>>>> 		 */
>>>>> 		switch (intspec[2]) {
>>>>> 		case 0:
>>>>> -		case 1: /* no EISR/EIMR support for now, treat as shared IRQ
>>>> */
>>>>> +			break;
>>>>> +		case 1:
>>>>> +			if (!mpic->err_int_config_done)
>>>>> +				break;
>>>>> +
>>>>
>>>> Under what case would we call mpic_host_xlate and have not called
>>>> mpic_init?
>>>>
>>> [Sethi Varun-B16395] Never, but we shouldn't translate the error
>>> interrupt specifier If mpic_err_int_init failed.
>>
>> Isnt that true of a 1000 other things.  If init failed we shouldn't even
>> call map for other reasons.  I don't think we need a special check here.
> [Sethi Varun-B16395] There is no specific check to see if mpic_init failed.
> In this particular case if we fail to register the error interrupt handler
> we cannot use the error sub interrupts.

As Kumar says, you could say the same thing about all the ioremaps it
does, or any other bit of init.  If mpic_init() fails, we're broken.
There's no point trying to hobble along by testing whether this bit or
that bit actually succeeded.

-Scott