[vortex] Problems with Redhat 7.1 and 3c095c
Bill MacAllister
bill@macallister.grass-valley.ca.us
Wed, 10 Oct 2001 13:31:15 -0700
--On Wednesday, October 10, 2001 11:19 AM -0400 Donald Becker
<becker@scyld.com> wrote:
> On Mon, 8 Oct 2001, Bill MacAllister wrote:
>
>> I have a dual boot system, Windows 2000 Server and Redhat 7.1. I will
> ..
>> with Linux. This system originally had a Netgear card in it.
>
> Which netgear card, and what were the problems?
I don't have the box/docs in front of me, but as I remember it, it used the
fa311 driver. I'll get the details to you in a follow up message tonight.
The symptoms where pretty much the same, but worse. At first blush network
connectivity seems fine. Can ping anywhere, can telnet. But large transfers
just stall after a bit. Then I started seeing telnet hang. In particular
there are some VMS systems that I manage and one time I could not successfully
get though the login sequence without hanging. But, it is not consistent.
Rebooting the system with the Netgear card seemed to fix telnet for a while.
I know that this is not a network problem because just about when I was to get
in the car to go see what was wrong with my VMS systems I tried telneting to
them from my Alpha Linux system. All of the VMS systems were accessible from a
Alpha Redhat 6.1 system.
>> But, when I
>> starting having problems I decided to replace it with a more well-known-to-
>> the-Linux-community card, a 3com card, 3C905C-TX-M. Initially I thought
>> that this fixed my problem, but while things are a bit better the problem
>> still exists.
>>
> ...
>> When I look with netstat -s the only thing that I see that looks amiss is
>> in the TCP section:
>
> What does /proc/net/dev report?
Inter-| Receive | Transmit
face |bytes packets errs drop fifo frame compressed multicast|bytes
lo: 420 6 0 0 0 0 0 0 420
eth0: 2483234 1895 0 0 0 0 0 0 173693
packets errs drop fifo colls carrier compressed
6 0 0 0 0 0 0
1400 0 0 0 0 0 0
> [[ Netstat was written for BSD-style systems, which don't have network
> error counters. Thus netstat doesn't report the full information set. ]]
>
>> I did build a copy of mii-diag and vortex-diag, but this really is getting
>> over my head and I could use some help understanding what is going on.
>
> Both of the outputs look fine.
>
> What driver version are you using?
Hmmmm, not sure how to look that up. Here is what I see in /lib/modules:
$ls -l /lib/modules/2.4.3-12/kernel/drivers/net/3c59x.o
-rw-r--r-- 1 root root 33716 Jun 8 12:33
/lib/modules/2.4.3-12/kernel/drivers/net/3c59x.o
Note, in response to some of the other messages I have seen I upgraded from the
2.4.2 kernel to 2.4.3-12. There is no change in the problem, but the driver is
a bit different:
$ls -l /lib/modules/2.4.2-2/kernel/drivers/net/3c59x.o
-rw-r--r-- 1 root root 34908 Apr 8 2001
/lib/modules/2.4.2-2/kernel/drivers/net/3c59x.o
I guess I will put the latest and greatest from your site and see if I can move
the problem, but I don't expect it will make a lot of difference. I also tried
an SMC card that I had, the box calls is an EZNET 10/100 with, the same results.
> What is the detection message?
>From /var/log/messages
Oct 10 00:10:18 wp kernel: PCI: Found IRQ 11 for device 00:0c.0
Oct 10 00:10:18 wp kernel: 3c59x.c:LK1.1.13 27 Jan 2001 Donald Becker and
others. http://www.scyld.com/network/vortex.html
Oct 10 00:10:18 wp rpc.statd[485]: Version 0.3.1 Starting
Oct 10 00:10:18 wp kernel: See Documentation/networking/vortex.txt
Oct 10 00:10:18 wp kernel: eth0: 3Com PCI 3c905C Tornado at 0x6800,
00:01:03:e9:7a:f9, IRQ 11
Oct 10 00:10:18 wp keytable: Loading keymap:
Oct 10 00:10:18 wp kernel: product code 4a45 rev 00.12 date 07-16-01
Oct 10 00:10:18 wp kernel: 8K byte-wide RAM 5:3 Rx:Tx split,
autoselect/Autonegotiate interface.
Oct 10 00:10:18 wp kernel: MII transceiver found at address 24, status 786d.
Oct 10 00:10:18 wp kernel: Enabling bus-master transmits and whole-frame
receives.
Oct 10 00:10:18 wp kernel: eth0: scatter/gather disabled. h/w checksums enabled
Oct 10 00:10:18 wp kernel: eth0: using NWAY device table, not 8
>> $ vortex-diag -e
>> Index #1: Found a 3c905C Tornado 100baseTx adapter at 0x6800.
> ...
>> Vortex format checksum is incorrect (0090 vs. 10b7).
>> Cyclone format checksum is incorrect (00 vs. 0x06).
>> Hurricane format checksum is incorrect (0x29 vs. 0x06).
>> Bad checksums? Is this the issue? If it is I don't have a clue what to do
>> about it.
>
> No. The EEPROM checksums are frequently reported as wrong on newer 3com
> boards. 3Com keeps changing the fields that the checksum covers, and
> some of 3Com's own boot-ROM code seems to update EEPROM parameters
> related to network booting without correcting the checksum.
>
> Donald Becker becker@scyld.com
> Scyld Computing Corporation http://www.scyld.com
> 410 Severn Ave. Suite 210 Second Generation Beowulf Clusters
> Annapolis MD 21403 410-990-9993
>
I saw a recommendation on the list to upgrade to at least a 2.4.9 kernel. That
the problem seems to be independant of the network card and driver seems to
indicate that I should start there. The 2.4.3-12 kernel is the latest that I
see available from Redhat so I guess I will figure out how to build kernels to
do that. It will be a bit of a pain getting the source there since I am having
problems with large transfers.
It also occurs to me that it might be some PCI problem that I don't understand.
Since this motherboard has some ISA slots I was thinking of trying a really old
ISA 3com card just to see if that is the problem.
Thanks for thinking about this for me. If you have any suggestions or if there
is anything information that you need, let me know.
Bill
+------------------------------------------------------------
| Bill MacAllister
| 14219 Auburn Road
| Grass Valley, CA 95949
| +1 530.272.8555