Re: [PATCH][POKE] Skip looking for ioapic overrides when ioapics are not present

[ebiederm@xmission.com (Eric W. Biederman)]

Sedat Dilek <sedat.dilek@googlemail.com> writes:

> On Thu, Mar 31, 2011 at 12:53 PM, Florian Mickler <florian@mickler.org> wrote:
>> On Thu, 31 Mar 2011 10:48:43 +0200
>> Paul Bolle <pebolle@tiscali.nl> wrote:
>>
>>> On Thu, 2011-03-31 at 10:01 +0200, Florian Mickler wrote:
>>> > diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
>>> > index 68df09b..3940103 100644
>>> > --- a/arch/x86/kernel/apic/io_apic.c
>>> > +++ b/arch/x86/kernel/apic/io_apic.c
>>> > @@ -3789,6 +3789,9 @@ int acpi_get_override_irq(u32 gsi, int *trigger, int *polarity)
>>> >  {
>>> >     int ioapic, pin, idx;
>>> >
>>> > +   if (acpi_irq_model != ACPI_IRQ_MODEL_IOAPIC)
>>> > +           return -1;
>>> > +
>>> >     if (skip_ioapic_setup)
>>> >             return -1;
>>> >
>>>
>>> Seems to have the same goal as commit
>>> 678301ecadec24ff77ab310eebf8a32ccddb1850 ("x86, ioapic: Don't warn about
>>> non-existing IOAPICs if we have none"), which got merged in the v2.6.38
>>> cycle (authored by me, signed off by Ingo Molnar). Maybe Eric's patch is
>>> more correct. I can't say as I was happy with the effect of my patch
>>> (ie, make an uninteresting error disappear) and didn't investigate any
>>> further. I have also no desire to dive into this matter again.

Yes. My patch is more correct.  We really do want the warning if we have
0 ioapics and we expect to be using ioapics.  It doesn't make sense to
suppress the warning unless we aren't in ioapic mode.

I don't have a clue why my patch got lost, but can we please get it
applied?

>> Thanks for letting me know. Sedat, did you actually test with 2.6.38?
>>
>> Regards,
>> Flo

> I have and had this patch in my own patch-series *before* Debian
> included it (IIRC right after Eric committed it to LKML).
> To answer your question: Yes.
> I have the patch also in my current linux-next kernels
> (next-20110331).

Eric