Re: [PATCH] xen/kexec: Prevent deadlock because of one_cpu_only()

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 08/04/13 08:48, Jan Beulich wrote:
>>>> On 06.04.13 at 17:03, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
>> I can probably fix the acpi_dmar_reinstate() and kexec_crash_save_cpu()
>> issues without too much problem and will look into those in due course,
>> but my next task is to see whether there was a programatic reason as to
>> why the wait descriptor was not executed successfully.
> Was that perhaps still without the Intel chipset interrupt remapping
> errata workaround? These errata can - afaict - have exactly that
> panic as a symptom.
>
> Jan
>

This is not from an affected system; It is Sandy-Bridge EP (And we have
the workaround in place anyway), although we too saw these symptoms for
the errata-affected systems.

I am double checking the errata docs, given the similarity.

~Andrew