2001-12-04 00:12:11

by Francois Romieu

[permalink] [raw]
Subject: Re: Are we going too fast? [upatch]

PinkFreud <[email protected]> :
[...]
> >From Documentation/nmi_watchdog.txt:
> NOTE: currently the NMI-oopser is enabled unconditionally on x86 SMP
> boxes.

My 2.4.8 tree seems to disagree with the doc:
find . -name \*[ch] | xargs fgrep nmi_watchdog | grep =
./arch/i386/kernel/traps.c:int nmi_watchdog = 0;
./arch/i386/kernel/traps.c:__setup("nmi_watchdog=", setup_nmi_watchdog);
./arch/i386/kernel/io_apic.c: nmi_watchdog = 0;

> I'm not specifically enabling it in LILO, but according to the docs, it's
> enabled already. Unfortunately, the lockup happens when switching between
> virtual consoles, so even if something WERE printed to the screen, I'm unlikely
> to see it.

Let's hope Mr Murphy took some vacation and please give a try at
append="nmi_watchdog=1".

> Side note: The lockup does *NOT* occur on 2.2.19 with SMP.

The console code differs a lot. Trace would help.

--
Ueimor