Re: [GIT PULL] RAS update for 3.20 (one more thing)

[Borislav Petkov <bp@suse.de>]

On Mon, Feb 09, 2015 at 11:13:16AM +0100, Ingo Molnar wrote:
> This kind of vendor specific hard coding is really ugly:
> 
>         if (c->x86_vendor == X86_VENDOR_AMD && ret == NOTIFY_STOP)
>                 return;

Yeah, I raised the same concern but we already have a bunch of vendor
checks in mce.c anyway so one more shouldn't hurt.

> Instead we should fix the Intel side to do a proper decode 
> as well - by the time mcelog is running it might be too 
> late, attempting an intelligent printk is way better...
> 
> So what would we need to make the Intel side just as good 
> as the AMD side?

The only thing that needs to be done is actually code it, methinks.
Once that's done, we would finally carry ready and decoded information
through a tracepoint to userspace instead of the /dev/mcelog thing.

I'm afraid this won't happen quickly though so we probably would need
this fix for the interim while we're working on the decoding.

Both sides (AMD/Intel) would need to be taught to funnel the decoded
strings through a tracepoint - something which we've talked about in the
past but keeps getting pushed down the TODO list. :-\

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--