Re: -current amd64 does not boot on huge machine (80 cores, RAM 1TB)

[Nicolas Joly <njoly%pasteur.fr@localhost>]

On Thu, Oct 06, 2011 at 07:32:30AM +0300, Jukka Ruohonen wrote:
> On Wed, Oct 05, 2011 at 11:56:09PM +0200, Nicolas Joly wrote:
> > (and coretemp via userconf which paniced the machine)
> 
> Can you try the following patch? I re-enabled one read from a register via the
> recently added rdmsr_safe(), but it could be that it still does not work
> (although the code is now pretty much identical with FreeBSD and Linux).
[...]
>               if (rdmsr_safe(MSR_TEMPERATURE_TARGET, &msr) == EFAULT)

Still no luck ... but that's the previous one (MSR_IA32_EXT_CONFIG
still from coretemp_tjmax()) the culprit:

        /*
         * On some Core 2 CPUs,there is an undocumented
         * MSR that tells if Tj(max) is 100 or 85. Note
         * that MSR_IA32_EXT_CONFIG is not safe on all CPUs.
         */
        if ((model == 0x0F && stepping >= 2) ||
            (model == 0x0E && extmodel != 1)) {

                msr = rdmsr(MSR_IA32_EXT_CONFIG);

                if ((msr & __BIT(30)) != 0) {
                        sc->sc_tjmax = 85;
                        return;
                }
        }

Commenting that part of code make the panic vanish ...

-- 
Nicolas Joly

Projects and Developments in Bioinformatics
Institut Pasteur, Paris.