From mboxrd@z Thu Jan  1 00:00:00 1970
To: 9fans@9fans.net
Subject: Re: [9fans] 9vx problem with venti
From: "Russ Cox" <rsc@swtch.com>
Date: Mon, 30 Jun 2008 13:33:22 -0400
In-Reply-To: <13426df10806301003y68192e5xc9b18c5b4d4b5e7a@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Message-Id: <20080630173103.8B4CC1E8C51@holo.morphisms.net>
Topicbox-Message-UUID: cebc2316-ead3-11e9-9d60-3106f5b1d025

> on .10, I can run venti/venti. on .11, it locks up 9vx quite
> thoroughly after it prints init.... If you resize the window it is
> filled with garbage. Under strace I can see it taking the timer
> interrupts.
>
> Linux xcpu 2.6.25 #6 SMP Tue May 27 09:46:16 PDT 2008 i686 i686 i386 GNU/Linux
>
> Sorry I don't have much more. I'm still looking around.
>
>
> also, on .10, when I am doing a venti/copy, at some point vx dies with
> a sigsev on cpux, where x is 5 or 7 or 9 so far.

If you invoke it with the -A flag, then it will go into a
sleep loop on panic (-A stands for abort, but that didn't
work very well on OS X).  You can then attach with gdb
and get a stack trace or look at what the other threads
are doing.

The cpu numbers don't really mean anything except
for whether they're 0 (runs user code) or non-zero
(runs kprocs).

Russ