Re: interrupts on APs

[yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi)]

hi,

> On Mon, Dec 12, 2011 at 03:44:59AM +0000, YAMAMOTO Takashi wrote:
>> hi,
>> 
>> > On Wed, Dec 07, 2011 at 04:05:34AM +0000, YAMAMOTO Takashi wrote:
>> >> only think i'm aware of is the "no_irq_affinity" quirk for 
>> >> E7520/7320/7525.
>> >> from their comment:
>> >> 
>> >>         /* BIOS may enable hardware IRQ balancing for
>> >>          * E7520/E7320/E7525(revision ID 0x9 and below)
>> >>          * based platforms.
>> >>          * Disable SW irqbalance/affinity on those platforms.
>> >>          */
>> >> 
>> >> can anyone test on those hardware?
>> > 
>> > Tested on a Dell poweredge 2850 with 2 Xen 3,2Ghz (single-core, 
>> > hyperthread)
>> > with:
>> > pchb0 at pci0 dev 0 function 0
>> > pchb0: Intel E7525 Memory Controller Hub (rev. 0x09)
>> > ppb0 at pci0 dev 2 function 0: Intel E7525 PCI Express Port A (rev. 0x09)
>> > 
>> > Boot fails with obvious interrupts issues (messages from ehci controller
>> > which looks like timeouts; 'lost interrupts' from cmdide).
>> > With plain GENERIC from HEAD, there's no such issues.
>> > 
>> > So this is one platform where it fails.
>> 
>> does a dynamic reconfiguration (cpuctl nointr) fail similarly?
> 
> I'm not sure what you ask for here. The kernel does't reach
> single-user, so I've no chances to run cpuctl.
> Or should I run cpuctlagainst the non-patched kernel ? If so, with what
> arguments ?

i meant eg.
        # cpuctl list
        (check #Intr column)
        # cpuctl nointr 0
        # cpuctl list
        (check #Intr column again)

but probably it isn't worth trying as it doesn't always work
even on my box. :)

YAMAMOTO Takashi

> 
> -- 
> Manuel Bouyer <bouyer%antioche.eu.org@localhost>
>      NetBSD: 26 ans d'experience feront toujours la difference
> --