Re: [PATCH] ACPI, APEI, Add APEI _OSC support

[Huang Ying <ying.huang@intel.com>]

On 06/17/2011 09:34 AM, Matthew Garrett wrote:
> On Fri, Jun 17, 2011 at 08:57:09AM +0800, Huang Ying wrote:
>> On 06/16/2011 09:57 AM, Matthew Garrett wrote:
>>> Yeah, this is going to be a problem. We have the HEST available at this 
>>> point so we ought to be able to parse it, though. I'll take a look 
>>> tomorrow.
>>
>> We can check the HEST table before _OSC evaluating.  But it is much
>> harder to check software part, because we have implemented GHES support
>> (Generic Hardware Error Source, the handler of firmware first mode
>> hardware error notification) as device driver and module.
> 
> If the kernel has been configured with support for the feature then I 
> think we ought to be able to assume that the kernel will support it at 
> runtime.

There may be error during driver initialization.  That is what I am
concerned.

>> So I think we can do that in 2 steps.  At first, we just enable WHEA
>> UUID, because that is easier to do.  Then we find a way to implement
>> "APEI bit" in generic _OSC call.  Do you think that is a good idea?
> 
> I'm fine with that, providing that GHES isn't disabled purely because 
> the WHEA UUID call wasn't successful.

Because we have not added the code to make generic _OSC call with "APEI
bit" now, so if WHEA UUID call failed, we have no firmware first mode
enabled.  So I think it is safe to disable GHES if WHEA UUID call
failed.  But in another hand, keeping GHES has no harm too.  So I am OK
to keep GHES if WHEA UUID call failed.

Best Regards,
Huang Ying