Re: [Xen-devel] [PATCH] xen/apic: use xen_apic for all pv guests, not just dom0.

[Ian Campbell <Ian.Campbell@xxxxxxxxxxxxx>]

On Wed, 2011-06-01 at 17:03 +0100, Stefano Stabellini wrote:
> On Wed, 1 Jun 2011, Ian Campbell wrote:
> > c3dd7941354f "xen/apic: Provide an 'apic_xen' to set the override the
> > apic->[read|write] for all cases" caused my 32 bit domU to panic with:
> > [    0.000000] Kernel panic - not syncing: Didn't find an APIC driver
> > [    0.000000] Pid: 0, comm: swapper Not tainted 
> > 2.6.32.41-x86_32p-xenU-01177-g22982d8-dirty #233
> > [    0.000000] Call Trace:
> > [    0.000000]  [<c11f8284>] ? printk+0xf/0x13
> > [    0.000000]  [<c11f81b8>] panic+0x39/0xf6
> > [    0.000000]  [<c1308f60>] generic_apic_probe+0x72/0x8a
> > [    0.000000]  [<c1301cd3>] setup_arch+0x90f/0x966
> > [    0.000000]  [<c102e5c1>] ? vprintk+0x368/0x38a
> > [    0.000000]  [<c104b9bb>] ? lockdep_init_map+0x87/0x40f
> > [    0.000000]  [<c104aed9>] ? debug_check_no_locks_freed+0x122/0x147
> > [    0.000000]  [<c12fc5ca>] start_kernel+0x90/0x30e
> > [    0.000000]  [<c12fc0a8>] i386_start_kernel+0x97/0x9e
> > [    0.000000]  [<c12ff372>] xen_start_kernel+0x4ce/0x4d6
> > 
> > Add the APIC override for PV guests as well as dom0.
> > 
> > Signed-off-by: Ian Campbell <ian.campbell@xxxxxxxxxx>
> > Cc: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
> > ---
> >  arch/x86/xen/enlighten.c |    2 +-
> >  1 files changed, 1 insertions(+), 1 deletions(-)
> > 
> > diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
> > index 0493093..7cb12e0 100644
> > --- a/arch/x86/xen/enlighten.c
> > +++ b/arch/x86/xen/enlighten.c
> > @@ -761,7 +761,7 @@ static __init int 
> > xen_safe_flat_acpi_madt_oem_check(char *oem_id,
> >  
> >  static __init int xen_safe_probe(void) {
> >  
> > -   if (!xen_initial_domain())
> > +   if (!xen_pv_domain())
> >             return 0;
> >  
> >     return 1;
>  
> why not remove xen_safe_probe?
> see alpine.DEB.2.00.1105271616390.12963@kaball-desktop

I knew I'd seen a discussion around this area somewhere, I just couldn't
find it.

If you think your solution is preferable please test and submit it
properly. The important thing is that the kernel has been broken for
ages (last success seems to be 13 May) and the automated tests have been
failing.

Ian.




_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel