[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] 4.11.0 RC1 panic

On 25/04/2018 07:58, Jan Beulich wrote:
>>>> On 24.04.18 at 18:06, <bouyer@xxxxxxxxxxxxxxx> wrote:
>> Hello,
>> I tested xen 4.11.0 rc1 with NetBSD as dom0.
>> I could boot a NetBSD PV domU without problem, but at shutdown time 
>> (poweroff
>> in the domU), I got a Xen panic:
>> (XEN) Assertion 'cpu < nr_cpu_ids' failed at 
>> ...1/work/xen-4.11.0-rc1/xen/include/xen/cpumask.h:97
>> A xl destroy instead of poweroff gives the same result.
>> This happens with both 32bitsPAE and 64bits domU. This doens't seem to
>> happen with HVM domUs.
>> Attached are a cut-n-paste of the panic, and the output of xl demsg.
> Without line numbers associated with at least the top stack trace entry
> I can only guess what it might be - could you give the patch below a try?
> (This may not be the final patch, as I'm afraid there may be some race
> here, but I'd have to work this out later.)
> Jan
> --- unstable.orig/xen/arch/x86/mm.c
> +++ unstable/xen/arch/x86/mm.c
> @@ -1255,7 +1255,7 @@ void put_page_from_l1e(l1_pgentry_t l1e,
>          {
>              for_each_vcpu ( pg_owner, v )
>              {
> -                if ( pv_destroy_ldt(v) )
> +                if ( pv_destroy_ldt(v) && v->dirty_cpu != VCPU_CPU_CLEAN )
>                      flush_tlb_mask(cpumask_of(v->dirty_cpu));
>              }
>          }

Manuel: As a tangentially related question, does NetBSD ever try to page
out its LDT?

I'm fairly sure this particular bit of code exists solely for the
Windows XP port to PV guests.  The code itself is broken as far as the
"feature" goes (as it only works on present => not present PTE changes,
and not for other PTE permissions changes which would also drop the
segdesc typeref), and dropping it would remove one vcpu scalability
limitation for PV guests.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.