On Wednesday 01 September 2010 18:02:13 Yang, Sheng wrote:
> On Wednesday 01 September 2010 17:14:52 Jan Beulich wrote:
> > >>> On 01.09.10 at 05:39, "Yang, Sheng" <sheng.yang@xxxxxxxxx> wrote:
> > > Yes, here is the patch with modification of other variants.
> >
> > If indeed an adjustment like this is needed, then this (and other similar
> > instances)
> >
> > >@@ -71,6 +72,11 @@
> > >
> > > unsigned int cpu_mask_to_apicid_phys(cpumask_t cpumask)
> > > {
> > >
> > >+ int cpu;
> > >
> > > /* As we are using single CPU as destination, pick only one CPU here
> > > */
> > >
> > >- return cpu_physical_id(first_cpu(cpumask));
> > >+ for_each_cpu_mask(cpu, cpumask) {
> > >+ if (cpu_online(cpu))
> > >+ break;
> > >+ }
> > >+ return cpu_physical_id(cpu);
> > >
> > > }
> >
> > is both insufficient: You need to handle the case where you don't
> > find any online CPU in the mask (at least by adding a respective
> > BUG_ON()).
>
> Yes, BUG_ON() is needed.
>
> > But I tend to agree with Keir that this shouldn't be done here -
> > these functions are simple accessors, which shouldn't enforce
> > any policy. Higher level code, if it doesn't already, should be
> > adjusted to never allow offline CPUs to slip through.
>
> Well, I think it's acceptable to add a wrap function for it. So how about
> this one?
Keir & Jan, how do you think about this patchset?
If you still think we should never allow offline CPUs in the cpu_mask, then at
least
we need one patch to fix serial port IRQ's cpu_mask which was CPU_MASK_ALL(this
fix
would also result the serial interrupt delivery to CPU0).
--
regards
Yang, Sheng
diff --git a/xen/arch/x86/smpboot.c b/xen/arch/x86/smpboot.c
--- a/xen/arch/x86/smpboot.c
+++ b/xen/arch/x86/smpboot.c
@@ -1015,7 +1015,7 @@
irq_vector[irq] = FIRST_HIPRIORITY_VECTOR + seridx + 1;
per_cpu(vector_irq, cpu)[FIRST_HIPRIORITY_VECTOR + seridx + 1] = irq;
irq_cfg[irq].vector = FIRST_HIPRIORITY_VECTOR + seridx + 1;
- irq_cfg[irq].cpu_mask = (cpumask_t)CPU_MASK_ALL;
+ irq_cfg[irq].cpu_mask = cpu_online_map;
}
/* IPI for cleanuping vectors after irq move */
dest_fix.patch
Description: Text Data
serial_fix.patch
Description: Text Data
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home