2.0 kernel locks on net activity

Chuck Mattern (cmattern@mindspring.com)
23 Jun 1996 02:47:41 GMT


I have a very stable RH 1.2.13 system that I attempted to upgrade to 2.0.0
today. I found that whenever a packets tries to hit the network via
ethernet the system hangs with nothing in the syslog.

I'm running a P133 PCI motherboard with 32 megs of ram, Adaptec 2940W
SCSI controller, STB 2meg PCI video based on S3 Trio32/Trio64 chipset,
LinkSys PCI ethernet card with DE435 chipset. For what it's worth, with
either kernel when the network comes up I get the following message:

eth0: Media is unconnected!

I suspect this is part of the cards auto detection of media, since I
have it on a BNC conector, but as long as I run 1.2.13 everything is
fine.

Stuff from proc that might be of interest:

[root@ronin /proc]# cat pci
PCI devices found :
Bus 0 Device 0 Function 0.
Host bridge : Intel 82437FX Triton (rev 2).
Bus 0 Device 7 Function 0.
ISA bridge : Intel 82371FB (rev 2).
Bus 0 Device 14 Function 0.
SCSI bus controller : Adaptec 2940 (rev 0). 8259's interrupt 9.
Bus 0 Device 16 Function 0.
VGA display controller : S3 Inc. Trio32/Trio64 (rev 0). 8259's
interrupt 11.

[root@ronin /proc]# cat ioports
0000-001f : dma1
0020-003f : pic1
0040-005f : timer
0060-0060 : kbd
0064-0064 : kbd
0070-007f : rtc
0080-009f : dma page reg
00a0-00bf : pic2
00c0-00df : dma2
00f0-00f1 : npu
00f8-00ff : npu
02f8-02ff : serial(auto)
0378-037a : lp
03bc-03be : lp
03d4-03d5 : ega+
03f8-03ff : serial(auto)
6000-60be : aic7xxx
6100-617f : DE435 (eth0)

[root@ronin /proc]# cat interrupts
0: 949397 timer
1: 13367 keyboard
2: 0 + cascade
3: 82903 + serial
9: 59956 + aic7xxx
10: 82 DE435 (eth0)
13: 1 math error

Anyone having suggestion, please send them my way either here or via
email.

Long narative research for cause of hangs:

To track this thing down I took everything out of runlevel 2 and made
two the initdefault. Then booted and started adding in process from
runleve three until it croaked. At first it looked like named was doing
it but when I went back to my isp's name servers and got far enough for
nfs to come up that triggered a name lookup and I hit the interface
again, locked tight. Normally this would have gone via ppp but I didn't
have diald running at the time. Eventually took out named and nfs and it
looked fine. I was working along and the machine locked tight. When I
got it backup I found three orphaned files in /var file system, all
remnants of my wifes pop mail run that I forgot to comment out of her
crontab. Network again. After another reboot I got diald and ppp back
up and cound get out but as soon as I took diald back down and hit that
card the system locked hard again. This system never locks up under
1.2.13 and I suspect that I may have been getting by with previosu
hardware problems that slipped through former system laxness, like the
xntpd business.

-- 
-------------------------------------------------------------------------------
|Chuck Mattern            | "That which does not kill us, makes us stronger.."|
|cmattern@mindspring.com  | -Friedrich Nietzsche-                             |
-------------------------------------------------------------------------------