RE: x86_mce: mce_start uses number of phsical cores instead oflogical cores

From: Luck, Tony
Date: Fri May 10 2013 - 15:09:55 EST


> With hyperthread turns on, the num_online_cpus reports the number of all logical cores.
> What I found in testing is only half the cores receives the mce broadcast, so I assume only the physical cores get broadcast.

See Intel Software Developer Manual Volume 3B Section 15.10.4.1, 3rd bullet:

o For processors on which CPUID reports DisplayFamily_DisplayModel as 06H_0EH and onward, an MCA signal is
broadcast to all logical processors in the system

Your E-5645 processors are a lot newer than this cut-off version - so they should broadcast
to all your threads.

You are seeing something very strange. It would be interesting to know *which* 12 cpus
show up for your machine check. Perhaps you are seeing all the hyperthreads from one
socket and none from the other?

I still suspect that something is strange in the EDAC error injection side of this problem
and that you are not getting a h/w initiated INT#18 event.

-Tony

--
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/