Re: further IO-APIC oddities

From: William Lee Irwin III (wli@holomorphy.com)
Date: Thu Aug 08 2002 - 14:18:51 EST


At some point in the past, I wrote:
>> It's different from 2.5.29, I can follow up with that. 2.5.29 saw all 0's,
>> so whatever it was that was scribbling over the MPC table and making the
>> ID's all 0, it's scribbling on something else now (probably mem_map).

On Thu, Aug 08, 2002 at 11:14:06AM -0700, Martin J. Bligh wrote:
> I thought that was only if you reduced NR_CPUS? And you shouldn't
> need to read the IOAPIC tables to do that - the basic array was getting
> overwritten. Or am I confusing at least two different bugs?

Sorry, I'm being obtuse. What I had in mind here was:

(1) there's been a longstanding bug where the MPC table entries with
        the IO-APIC ID's gets zeroed out
(2) this bug mysteriously went away, even though we never tracked it down
(3) bugs that smell of mem_map getting stomped on are cropping up

... and I suspect these three things are related, e.g. the MPC stuff moved
or link order changed or some such Heisenbug-ish nonsense and now it's
hitting something else. The NR_CPUS bit isn't even bootable (the panic()
happens before console_init() IIRC) so it's not quite that one. =)

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



This archive was generated by hypermail 2b29 : Thu Aug 15 2002 - 22:00:17 EST