Re: [PATCH] RAS: Add a tracepoint for reporting memory controllerevents

[Mauro Carvalho Chehab]

Em 31-05-2012 11:22, Borislav Petkov escreveu:
> On Thu, May 31, 2012 at 10:56:28AM -0300, Mauro Carvalho Chehab wrote:
>> Em 31-05-2012 09:17, Borislav Petkov escreveu:
>>> On Thu, May 31, 2012 at 07:33:25AM -0300, Mauro Carvalho Chehab wrote:
>>>> Huh? Tony said that some errors report at 4K granularity while, others
>>>> are at cache line size.
>>>
>>> Cacheline size is from MCE. This tracepoint is for memory controller
>>> errors where granularity should be pretty stable.
>>
>> Both sb_edac and i7core_edac get errors via MCE. They're currently providing
>> a fake granularity, but I'm planning to add proper grain support on them,
>> by retrieving such information from the MCE registers.
> 
> Those are only 2 drivers and not _all_ edac drivers. For the remaining
> ones grain is a pretty stable, seldomly changing value. Which means it
> should be part of const char driver_detail only on those two drivers and
> not a driver-global field in the tracepoint.

Grain is a global property: all drivers have it; userspace tools need it.

It doesn't matter if some drivers use statically-defined grain, while others
are dynamic: at the reported event, the grain should be there for both types
of drivers.

Regards,
Mauro


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/