freebsd stable - Jun 2010 - disable RAM parity error

Hello List,

In continuation to my earlier mail, I am now convinced(nearly) that
the freezes faced by me are "RAM parity error" indeed.
http://www.mail-archive.com/freebsd-hackers@freebsd.org/msg70730.html

when using internet via wpi0, FreeBSD freezes are very prone to occur
compared to when I am using bfe0. But they do occur, due to some
reason it feels stable when connected via bfe0.
Over course of time I have accumulated some core.txt.<n> files, which
I have uploaded to pastebin and their links are below. My question is
why only FreeBSD minds about RAM parity error check, while latest
offering from Redmond does not ?
is there a way to IGNORE RAM parity check ? is it possible to get hold
of offending address from these core files ?
while I observe "stack pointer" and "frame pointer" have
repeated
values in core files, I cannot conclude anything from this
observation. Anybody here care ?

core.txt.1  ------ http://pastebin.com/HDji1Xvs
core.txt.2  ------ http://pastebin.com/yXg1iXAN
core.txt.3  ------ http://pastebin.com/xEfcgtQf
core.txt.4  ------ http://pastebin.com/zVQspP4g
core.txt.5  ------ http://pastebin.com/7RjL2san
core.txt.6  ------ http://pastebin.com/q4yjVefB

Masoom Shaikh

on 15/06/2010 08:34 Masoom Shaikh said the following:
> Hello List,
> 
> In continuation to my earlier mail, I am now convinced(nearly) that
> the freezes faced by me are "RAM parity error" indeed.
> http://www.mail-archive.com/freebsd-hackers@freebsd.org/msg70730.html
> 
> when using internet via wpi0, FreeBSD freezes are very prone to occur
> compared to when I am using bfe0. But they do occur, due to some
> reason it feels stable when connected via bfe0.
> Over course of time I have accumulated some core.txt.<n> files, which
> I have uploaded to pastebin and their links are below. My question is
> why only FreeBSD minds about RAM parity error check, while latest
> offering from Redmond does not ?
> is there a way to IGNORE RAM parity check ? is it possible to get hold
> of offending address from these core files ?
> while I observe "stack pointer" and "frame pointer"
have repeated
> values in core files, I cannot conclude anything from this
> observation. Anybody here care ?
If I were you I'd be concerned with finding and replacing the flaky hardware
instead of wasting time before disaster happens.
As to why "other OS" doesn't report errors - it could have been
the other way
around, different OSes have different usage/load patterns and hit different
hardware problems with higher probability.

-- 
Andriy Gapon