Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 23 Jan 2003 20:48:23 -0800
From:      Terry Lambert <tlambert2@mindspring.com>
To:        Dan Nelson <dnelson@allantgroup.com>
Cc:        Atte Peltomaki <atte.peltomaki@iki.fi>, freebsd-current@FreeBSD.ORG
Subject:   Re: Performance problems with 5.0-RELEASE
Message-ID:  <3E30C597.8FB82EC9@mindspring.com>
References:  <20030123162644.GA560@naama> <3E309E58.C0745816@mindspring.com> <20030124040705.GA7945@dan.emsphone.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Dan Nelson wrote:
> > If you lean on the keyboard, or if you set up the network adapters
> > as "entropy" sources, does the problem fix itself?
> 
> If you're thinking it's /dev/random blocking on him, 5.0's output never
> blocks.  Its output is a PRNG periodically seeded from random data,
> including interrupt timings and LAN traffic by default.

Mostly, I was thinking that the other suggestions (WITNESS,
INVARIANTS, malloc flags) didn't really have any chance of
being the cause of the lock-ups, and without some message on
the console, it's unlikely that it was a driver tiemout, either.

8-|.

If it's long enough to pause the console noticibly, the next
thing to try is breaking to the debugger -- which might require
an NMI card -- to see what code it's stuck in during the pause.

The lack of NMI cards in most people's tool boxes is what made
me not suggest that before (i.e. if the console is locked, then
using the console to get to the debugger is probably out of the
question).

-- Terry

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3E30C597.8FB82EC9>