Re: [PATCH] 2.6.1-mm2: Get irq_vector size right for generic subarch UP installer kernels

[William Lee Irwin III]

On Thu, 15 Jan 2004, James Cleverdon wrote:
>> No, I haven't exceeded the available vectors, but wli has on a large
>> NUMA-Q box.

On Thu, Jan 15, 2004 at 05:40:54PM -0500, Zwane Mwaikambo wrote:
> Yes i believe the 8 node NUMA-Qs do this easily.

The mainline kernel panics before console_init() on the 9th IO-APIC
(the first IO-APIC on the 5th node) and this has been an extreme
annoyance essentially since Linux was ported to the beasts.

It won't be long before currently shipping ia32 boxen catch up with
that blast from the 4+ -year-old past, and x86-64 or whatever will have
similar issues if/when it ever gets chipsets allowing similar numbers
of cpus and devices (presumably they're correlated) to be attached,
since AFAIK it didn't change the number of IDT entries.

One thing that's particularly asinine is that since the things use
physical destinations in the RTE's and use a serial APIC bus or
whatever per node (i.e. can only interrupt cpus in their node), they
really are just a tabulation method away from functioning properly.

-- wli
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/