On Fri, 2016-03-18 at 14:32 +0100, Peter Zijlstra wrote:
> On Fri, Mar 18, 2016 at 01:39:16PM +0100, Mike Galbraith wrote:
> > On Fri, 2016-03-18 at 11:15 +0100, Peter Zijlstra wrote:
>
> > > Ah, did you actually disable HT in the BIOS, or just skip the HT
> > > enumeration by saying nr_cpus=64 (knowing that all the siblings are
> > > last)?
> >
> > It's disabled in BIOS.
>
> OK, so I disabled HT in the BIOS too, on my IVB-EX
>
> > > In any case, Thomas has a clue and I'm going to test, but 4 socket
> > > machine takes forever to boot, so might be a few minutes :/
> >
> > His one-liner made my DL980 all better.
>
> My machine is profoundly unhappy though; and since I have no nr_cpus=
> nr_cpu_ids == total_cpus and his patch wouldn't do anything anyway.
>
> [ 0.286838] max_cores: 15, cpu_ids: 60, num_siblings: 2, coreid_bits: 5
> [ 0.293463] smpboot: Max logical packages: 2
> [ 0.297733] smpboot: APIC(0) Converting physical 0 to logical package 0
> [ 0.304346] smpboot: APIC(20) Converting physical 1 to logical package 1
> [ 0.311047] smpboot: APIC(40) Package 2 exceeds logical package map
> [ 0.317309] smpboot: CPU 30 APICId 40 disabled
> [ 0.321757] smpboot: APIC(60) Package 3 exceeds logical package map
> [ 0.328022] smpboot: CPU 45 APICId 60 disabled
>
> This machine does exactly what I suspected yours did.
Yup, that looks very familiar. I just booted without nr_cpus=64 to
make sure it's a happy camper both w/wo, and it says all is peachy.
-Mike
