NetBSD Problem Report #38229
From martin@duskware.de Wed Mar 12 13:49:03 2008
Return-Path: <martin@duskware.de>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
by narn.NetBSD.org (Postfix) with ESMTP id 48BD963B8E3
for <gnats-bugs@gnats.netbsd.org>; Wed, 12 Mar 2008 13:49:03 +0000 (UTC)
Message-Id: <20080312134605.7FD8663B8E3@narn.NetBSD.org>
Date: Wed, 12 Mar 2008 13:46:05 +0000 (UTC)
From: netbsd@luedke.org
Reply-To: netbsd@luedke.org
To: netbsd-bugs-owner@NetBSD.org
Subject: Realtek 8139C+ problems with more than 2 NICs
X-Send-Pr-Version: www-1.0
>Number: 38229
>Category: kern
>Synopsis: Realtek 8139C+ problems with more than 2 NICs
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: kern-bug-people
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Mar 12 13:50:00 +0000 2008
>Last-Modified: Thu Mar 13 10:00:03 +0000 2008
>Originator: Frank Luedke
>Release: 4.0 RELEASE
>Organization:
>Environment:
NetBSD 4.0 NetBSD 4.0 (6420) #0: Fri Feb 22 14:08:14 CET 2008 root@:/usr/src/sys/arch/i386/compile/6420 i386
>Description:
On an appliance with four 8139C+ chipsets on-board only re0 and re1 are working without problems. re2 is running in 10 Mbit only, re3 is not working at all. Port re3 can be configured using "ifconfig", but Link LEDs stay dark and no signal gets transferred.
Console is always reporting "re3: watchdog timeout" when trying to send data through that interface.
I have already tried to block the IRQ re3 is using after I found this as a possible solution in the net, but with another IRQ the problem remains.
------- output of dmesg ----------
Realtek Semiconductor 8139 10/100 Ethernet (ethernet network, revision 0x20) at ? dev 12 function 0 (tag 0x80006000, intrtag 0x80006000, intrswiz 0, intrpin 0x1, i/o on, mem on, no quirks): RealTek 8139C+ 10/100BaseTX (rev. 0x20)
re1: interrupting at irq 12
re1: Ethernet address 00:90:0b:0a:54:4c
re1: using 64 tx descriptors
rlphy1 at re1 phy 0: Realtek internal PHY
rlphy1: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
------- output of dmesg ----------
Realtek Semiconductor 8139 10/100 Ethernet (ethernet network, revision 0x20) at ? dev 14 function 0 (tag 0x80007000, intrtag 0x80007000, intrswiz 0, intrpin 0x1, i/o on, mem on, no quirks): RealTek 8139C+ 10/100BaseTX (rev. 0x20)
re3: interrupting at irq 11
re3: Ethernet address 00:90:0b:0a:54:4a
re3: using 64 tx descriptors
rlphy3 at re3 phy 0: Realtek internal PHY
rlphy3: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
---------------------------------
>How-To-Repeat:
Use a system with more than 2 8139C+ Chipsets. Problem occurs on Kernel 4 with or without Patch from bug kern/38172 included.
>Fix:
Unknown.
>Audit-Trail:
From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/38229: Realtek 8139C+ problems with more than 2 NICs
Date: Wed, 12 Mar 2008 14:56:23 +0100
On Wed, Mar 12, 2008 at 01:50:00PM +0000, netbsd@luedke.org wrote:
> Realtek Semiconductor 8139 10/100 Ethernet (ethernet network, revision 0x20) at ? dev 14 function 0 (tag 0x80007000, intrtag 0x80007000, intrswiz 0, intrpin 0x1, i/o on, mem on, no quirks): RealTek 8139C+ 10/100BaseTX (rev. 0x20)
> re3: interrupting at irq 11
> re3: Ethernet address 00:90:0b:0a:54:4a
Do you get that irq at all (more than once)?
You can check with vmstat -i
Martin
From: Frank Luedke <frank@landitec.de>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/38229: Realtek 8139C+ problems with more than 2 NICs
Date: Thu, 13 Mar 2008 10:58:59 +0100
Here the requested outputs.
1. With BIOS set to PCI/PNP auto (ESCD)
--------------------
# vmstat -i
interrupt total rate
cpu0 softclock 68909 100
cpu0 softnet 1284 1
pic0 pin 5 1495 2
pic0 pin 15 1100 1
pic0 pin 0 68904 100
Total 141692 206
# dmesg | grep -i irq
re0: interrupting at irq 5
re1: interrupting at irq 12
re2: interrupting at irq 10
re3: interrupting at irq 11
viaide0: primary channel interrupting at irq 14
viaide0: secondary channel interrupting at irq 15
lpt0 at isa0 port 0x378-0x37b irq 7
com0 at isa0 port 0x3f8-0x3ff irq 4: ns16550a, working fifo
====================
2. With ESCD disabled, IRQ 10 and 11 blocked.
--------------------
# vmstat -i
interrupt total rate
cpu0 softclock 3560 114
cpu0 softnet 131 4
pic0 pin 9 174 5
pic0 pin 15 1027 33
pic0 pin 0 3556 114
Total 8448 272
# dmesg | grep -i irq
re0: interrupting at irq 9
re1: interrupting at irq 3
re2: interrupting at irq 5
re3: interrupting at irq 12
viaide0: primary channel interrupting at irq 14
viaide0: secondary channel interrupting at irq 15
lpt0 at isa0 port 0x378-0x37b irq 7
com0 at isa0 port 0x3f8-0x3ff irq 4: ns16550a, working fifo
====================
Results are the same. re3 reports "watchdog timeout", no
data could be sent.
This Hardware is working without problems when BIOS set to
ESCD when using Linux (RHEL 4), FreeBSD (m0n0wall) or WinXP.
So this must be a NetBSD issue.
.fl
(Contact us)
$NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2007
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.