NetBSD Problem Report #3536
Received: (qmail-queue invoked from smtpd); 25 Apr 1997 18:56:04 -0000
Message-Id: <199704251855.LAA13108@Cup.DSG.Stanford.EDU>
Date: Fri, 25 Apr 1997 11:55:34 -0700 (PDT)
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
Reply-To: jonathan@DSG.Stanford.EDU
To: gnats-bugs@gnats.netbsd.org
Subject: Not putting packet headers in a separate mbuf causes panic on mips?
X-Send-Pr-Version: 3.95
>Number: 3536
>Category: kern
>Synopsis: am7990 mbuf changes cause panic in ipintr()
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: kern-bug-people
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Fri Apr 25 12:05:02 +0000 1997
>Closed-Date: Mon May 12 22:32:50 +0000 1997
>Last-Modified: Mon May 12 22:33:08 +0000 1997
>Originator:
>Release: 1997-04-24
>Organization:
>Environment:
System: NetBSD Cup.DSG.Stanford.EDU 1.2D NetBSD 1.2D (DSG_4K) #0: Mon Mar 31 20:29:11 PST 1997 jonathan@Cup.DSG.Stanford.EDU:/aga/n1/src/NetBSD/IP-PLUS/src/sys/arch/i386/compile/DSG_4K i386
>Description:
am7990.c rev 1.33 causes a panic in ipintr() on a DECstation when the
network is started. I suspect that other CPUs that don't support
unaligned accesses will encounter the same problem.
>How-To-Repeat:
Build a -current kernel for a pmax as at 1997-04-24.
>Fix:
Reverting to rev 1.30 of am7990.c works around this problem for me. I
don't know whether it's better to do that, or to rework ipintr() to
handle the mbuf chains that am7990.c now produces.
>Release-Note:
>Audit-Trail:
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
To: gnats-bugs@gnats.netbsd.org
Cc: Subject: Re: kern/3536: Not putting packet headers in a separate mbuf causes
panic on mips?
Date: Fri, 25 Apr 1997 12:13:34 -0700
The following patch to am7990.c rev. 1.33 also works around this
problem, but effectively backs out rev. 1.32 and 1.33:
*** am7990.c.DIST Thu Apr 24 04:20:32 1997
--- am7990.c Fri Apr 25 12:00:09 1997
***************
*** 435,441 ****
}
len = MLEN;
}
! if (totlen >= MINCLSIZE) {
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
m_freem(top);
--- 435,441 ----
}
len = MLEN;
}
! if (top && totlen >= MINCLSIZE) {
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
m_freem(top);
From: Jason Thorpe <thorpej@nas.nasa.gov>
To: jonathan@dsg.stanford.edu
Cc: gnats-bugs@gnats.netbsd.org, netbsd-bugs@netbsd.org
Subject: Re: kern/3536: Not putting packet headers in a separate mbuf causes panic on mips?
Date: Fri, 25 Apr 1997 12:02:42 -0700
On Fri, 25 Apr 1997 11:55:34 -0700 (PDT)
Jonathan Stone <jonathan@DSG.Stanford.EDU> wrote:
> >Fix:
>
> Reverting to rev 1.30 of am7990.c works around this problem for me. I
> don't know whether it's better to do that, or to rework ipintr() to
> handle the mbuf chains that am7990.c now produces.
...I believe this is the correct solution... This is an issue on the
Alpha, as well. All of the network drivers on the Alpha currently
align the IP header in a separate mbuf (except for the LANCE, now,
apparently... I haven't updated my DEC 3000 yet).
I discussed this with Matt Thomas some months ago, and we came to
the conclusion that it would be best solved in ipintr(), using
the ALIGNED_POINTER() macro. If it is not aligned, copy the IP header
into a new mbuf, m_prepend() it, and continue.
Compilers would remove the dead code on systems where this isn't an
issue, and on systems where it _is_ an issue, you don't have to put
alignment crap into ever driver (esp since alignment issues might be
different for each layer 3).
Jason R. Thorpe thorpej@nas.nasa.gov
NASA Ames Research Center Home: 408.866.1912
NAS: M/S 258-6 Work: 415.604.0935
Moffett Field, CA 94035 Pager: 415.428.6939
State-Changed-From-To: open->closed
State-Changed-By: thorpej
State-Changed-When: Mon May 12 15:32:50 PDT 1997
State-Changed-Why:
Fixed some weeks ago.
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.