linux-mips
[Top] [All Lists]

Re: R5000 Linux

To: Guido Guenther <agx@sigxcpu.org>, linux-mips@linux-mips.org
Subject: Re: R5000 Linux
From: ilya@theIlya.com
Date: Thu, 17 Apr 2003 08:10:38 -0700
In-reply-to: <20030417145650.GJ22768@bogon.ms20.nix>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <200304170912.55096.donath@hks.com> <20030417145650.GJ22768@bogon.ms20.nix>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.4i
On Thu, Apr 17, 2003 at 04:56:50PM +0200, Guido Guenther wrote:
> You'll also find a cross toolchain there. The major show stopper for me
> is currently the "do_cpu invoked from kernel context!" problem which I
> didn't find any time to look into yet. I wonder wether I'm the only
> person seeing this?
I was able to reproduce it with following simple sequence of events:
#(while true; do true; done)>/dev/zero
As soon as I interrupt this, do_cpu oops happened. Problem occurs
somewhere in signeal handlin code. I didn't investoigate further yet, as
for what I am doing at the moment it is not a problem.

What is a problem, is memory leak in 2.5.47. Every time *any* process is forked,
system permenently looses few K of memory. Modifying above loop to run
cat</proc/meminfo will bring down system to its knees in about 5-10 minutes.

Attachment: pgpp4RMA62vix.pgp
Description: PGP signature

<Prev in Thread] Current Thread [Next in Thread>