I'm running Fedora Core 1 with kernel 2.6.7 with the latest patches installed. I have two problems.

First of all I had to disable Kudzu to be able to get networking up at all... I have two 3com-cards.

Here are the errors before disabling Kudzu:

3c59x: Donald Becker and others.
See Documentation/networking/vortex.txt
00:0d.0: 3Com PCI 3c905 Boomerang 100baseTx at 0xe400. Vers LK1.1.18-ac
00:10:4b:ad:95:a8, IRQ 10
product code 4d4e rev 00.0 date 07-05-98
8K word-wide RAM 3:5 Rx:Tx split, autoselect/MII interface.
MII transceiver found at address 24, status 786f.
Enabling bus-master transmits and whole-frame receives.
00:0d.0: scatter/gather enabled. h/w checksums disabled
eth0: Dropping NETIF_F_SG since no checksum feature.
PCI: Found IRQ 11 for device 00:0f.0
See Documentation/networking/vortex.txt
00:0f.0: 3Com PCI 3c905B Cyclone 100baseTx at 0xe800. Vers LK1.1.18-ac
00:10:5a:1e:52:f2, IRQ 11
product code 5152 rev 00.12 date 08-21-98
Full duplex capable
8K byte-wide RAM 5:3 Rx:Tx split, 10baseT interface.
Enabling bus-master transmits and whole-frame receives.
00:0f.0: scatter/gather enabled. h/w checksums enabled
ip_tables: (C) 2000-2002 Netfilter core team
PCI: Found IRQ 10 for device 00:0d.0
3c59x: Donald Becker and others.
See Documentation/networking/vortex.txt
00:0d.0: 3Com PCI 3c905 Boomerang 100baseTx at 0xe400. Vers LK1.1.18-ac
00:10:4b:ad:95:a8, IRQ 10
product code 4d4e rev 00.0 date 07-05-98
64K word-wide RAM 1:1 Rx:Tx split, autoselect/10baseT interface.
Enabling bus-master transmits and whole-frame receives.
00:0d.0: scatter/gather enabled. h/w checksums disabled
eth0: Dropping NETIF_F_SG since no checksum feature.
PCI: Found IRQ 11 for device 00:0f.0
See Documentation/networking/vortex.txt
00:0f.0: 3Com PCI 3c905B Cyclone 100baseTx at 0xe800. Vers LK1.1.18-ac
00:10:5a:1e:52:f2, IRQ 11
product code 5152 rev 00.12 date 08-21-98
Full duplex capable
8K byte-wide RAM 5:3 Rx:Tx split, autoselect/10baseT interface.
Enabling bus-master transmits and whole-frame receives.
00:0f.0: scatter/gather enabled. h/w checksums enabled
eth0: Host error, FIFO diagnostic register 2000.
eth0: PCI bus error, bus status 00a00029
eth0: Transmit error, Tx status register d0.
Flags; bus-master 1, dirty 1(1) current 1(1)"
Transmit list 00000000 vs. cd289240.
0: @cd289200 length 8000002a status 8000002a
1: @cd289240 length 00000000 status 00000000
2: @cd289280 length 00000000 status 00000000
3: @cd2892c0 length 00000000 status 00000000
4: @cd289300 length 00000000 status 00000000
5: @cd289340 length 00000000 status 00000000
6: @cd289380 length 00000000 status 00000000
7: @cd2893c0 length 00000000 status 00000000
8: @cd289400 length 00000000 status 00000000
9: @cd289440 length 00000000 status 00000000
10: @cd289480 length 00000000 status 00000000
11: @cd2894c0 length 00000000 status 00000000
12: @cd289500 length 00000000 status 00000000
13: @cd289540 length 00000000 status 00000000
14: @cd289580 length 00000000 status 00000000
15: @cd2895c0 length 00000000 status 00000000
eth0: Host error, FIFO diagnostic register 2000.
eth0: PCI bus error, bus status 00a00029
NETDEV WATCHDOG: eth0: transmit timed out
eth0: transmit timed out, tx_status 00 status e000.
diagnostics: net 0cc0 media 8802 dma 00000000.
Flags; bus-master 1, dirty 1(1) current 17(1)"
Transmit list 00000000 vs. cd289240.
0: @cd289200 length 8000002a status 8000002a
eth0: PCI bus error, bus status 00a00029



Second problem is that I get a lot of RX- and Frame errors on my external interface, which is connected to a cable modem, when I try to download anything from the internet from my internal network. Speed should be about 200kb/s max, but it's as low as 10k/s.

The two cards:
Index #1: Found a 3c905 Boomerang 100baseTx adapter at 0xe400.
Index #2: Found a 3c905B Cyclone 100baseTx adapter at 0xe800.

Operating at:
eth0: negotiated 100baseTx-FD, link ok (connected to internal 100mbit/FD switch)
eth1: 10 Mbit, full duplex, link ok (connected to cable modem)


# cat /proc/net/dev
Inter-| Receive | Transmit
face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed
lo: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
eth0: 3946842 46949 0 0 0 0 0 0 74136658 55016 0 0 0 0 0 0
eth1:93230696 275806 5237 0 0 5237 0 0 4605859 56282 0 0 0 0 0 0

# netstat -i
Kernel Interface table
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flg
eth0 1500 0 48324 0 0 0 56692 0 0 0 BMRU
eth1 1500 0 281617 5415 0 0 57589 0 0 0 BMRU
lo 16436 0 0 0 0 0 0 0 0 0 LRU

eth0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:49522 errors:0 dropped:0 overruns:0 frame:0
TX packets:58252 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:4137230 (3.9 Mb) TX bytes:78870168 (75.2 Mb)
Interrupt:10 Base address:0xe400

eth1 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:286272 errors:5585 dropped:0 overruns:0 frame:5585
TX packets:58766 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:98395258 (93.8 Mb) TX bytes:4790111 (4.5 Mb)
Interrupt:11 Base address:0xe800

# cat /proc/interrupts
CPU0
...
10: 108488 XT-PIC eth0
11: 352204 XT-PIC eth1
...

# grep 3c /etc/modules.conf
alias eth0 3c59x
alias eth1 3c59x




As you see, frame and RX errors are the same.. and keeps going up when I transfer files from my 100mbit net out through 10mbit NIC.

All my internal machines work fine over 100mbit localnet, except when they connect to the internet, speed is terrible.

My switch detects all NIC:s as 100mbit autoneg. The cable modem isn't connected to the switch, but to the external interface of my box, eth1. The cable modem doesn't support autoneg.

Any ideas? If I set 10mbit on the internal net, it would probably work? I rather have 100mbit.

Appreciate any help !

Thanks

I don't know if this will help or not, but I'd suggest 2 things: first, be sure the 3c59x modules are actually being loaded, which can be confirmed by running the lsmod command as root. If they aren't there, as I expect you know you can manually load them via the insmod or modprobe command. I've got a Cyclone card, and at least for me, it wasn't being loaded automatically at the start, which caused the same kind of problem as you described.

Second, what are the results of running ifconfig? I don't see that there's a INET address value being shown for either eth0 and eth1 (which of course would be the case if they weren't working) and therefore I'd also suggest trying to just bounce the interface:
Let me emphasize that these may not solve the problem, but if nothing else, they could be 2 factors that you could eliminate from consideration as being the cause. In any event, good luck with it. -- J.W.

Thanks for your suggestions.

I removed the inet to not expose my IP:s. Sorry for the confusion

lsmod output:

I suspect my first problem is caused by an old bug, which apparently wasn't fixed correctly. I think the problem lies in Kudzu having a conflict with the 3com driver.

Doing a quick search on '"FIFO diagnostic register 2000" + kudzu' gives a couple of references to this old bug (note posts on Redhat.com forum), and the posted solutions are to disable Kudzu until this has been fixed.

It's weird that no patches came out since this is an old bug.

My next real problem is the frame and RX errors:

Note 'errs' and 'frame'. Anyone know what causes these errors?? Very little information on google on this.


Thanks