From owner-freebsd-bugs  Mon Jun 12 11:16:20 2000
Delivered-To: freebsd-bugs@freebsd.org
Received: by hub.freebsd.org (Postfix, from userid 618)
	id 87A1037B9F8; Mon, 12 Jun 2000 11:16:18 -0700 (PDT)
Subject: Re: kern/19106: Corruption of xl driver's TX chain during startup
In-Reply-To: <200006111123.EAA20756@freefall.freebsd.org> from "asmodai@FreeBSD.org" at "Jun 11, 2000 04:23:45 am"
To: asmodai@FreeBSD.org
Date: Mon, 12 Jun 2000 11:16:18 -0700 (PDT)
Cc: gibbs@FreeBSD.org, freebsd-bugs@FreeBSD.org
X-Mailer: ELM [version 2.4ME+ PL54 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Message-Id: <20000612181618.87A1037B9F8@hub.freebsd.org>
From: wpaul@FreeBSD.ORG (Bill Paul)
Sender: owner-freebsd-bugs@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org

Hm. In most drivers, I bzero() the softc struct at the start of foo_attach()
just to make sure there isn't any garbage lying around. For some reason,
I'm not doing this in if_xl. It could be a coincidence that this has
never bitten anyone before. Anyway, try the following patch:


*** if_xl.c.orig	Mon Jun 12 11:07:12 2000
--- if_xl.c	Mon Jun 12 11:06:47 2000
***************
*** 1185,1190 ****
--- 1185,1191 ----
  
  	sc = device_get_softc(dev);
  	unit = device_get_unit(dev);
+ 	bzero(sc, sizeof(struct xl_softc));
  
  	/*
  	 * If this is a 3c905B, we have to check one extra thing.

If this fixes the problem, then the memory allocated for the
softc still has crap in it and I'm a doofus for not remembering to
clear it. If the problem persists, something else is corrupting the
softc space between xl_attach() and xl_init(), though offhand I
don't know what that could be.

-Bill


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