Re: [Xen-devel] [PATCH v2] xen/x86: clear per cpu stub page information in cpu_smpboot_free()

[Jürgen Groß <jgross@xxxxxxxx>]

On 08.01.20 16:21, Jan Beulich wrote:
> On 08.01.2020 15:34, Juergen Gross wrote:
> > cpu_smpboot_free() removes the stubs for the cpu going offline, but it
> > isn't clearing the related percpu variables. This will result in
> > crashes when a stub page is released due to all related cpus gone
> > offline and one of those cpus going online later.
> >
> > Fix that by clearing stubs.addr and stubs.mfn in order to allocate a
> > new stub page when needed.
> I was really hoping for you to mention CPU parking here. How about
>
> "Fix that by clearing stubs.mfn (and also stubs.addr just to be on
>   the safe side) in order to allocate a new stub page when needed,
>   irrespective of whether the CPU gets parked or removed."
>
> > --- a/xen/arch/x86/smpboot.c
> > +++ b/xen/arch/x86/smpboot.c
> > @@ -945,6 +945,8 @@ static void cpu_smpboot_free(unsigned int cpu, bool remove)
> >                                (per_cpu(stubs.addr, cpu) | ~PAGE_MASK) + 1);
> >           if ( i == STUBS_PER_PAGE )
> >               free_domheap_page(mfn_to_page(mfn));
> > +        per_cpu(stubs.addr, cpu) = 0;
> > +        per_cpu(stubs.mfn, cpu) = 0;
> Looking more closely, I think I'd prefer these two lines (of which
> the addr one isn't strictly needed anyway) to move ahead of the
> if().
>
> If you agree, I'll be happy to do both while committing.
I agree.

I'm not sure the addr clearing can be omitted. This might result in
problems when during onlining an early error happens in
cpu_smpboot_alloc() and thus skipping the call of alloc_stub_page().
The subsequent call of cpu_smpboot_free() will then overwrite mfn 0.


Juergen


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel