From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; fr-FR; rv:1.6) Gecko/20040113 Description of problem: I can't get the network to work on my Dell Latitude CPx laptop. Some information on this machine. It's a Dell Latitude CPx, model PPX: http://docs.us.dell.com/docs/Systems/pcpxhj/en/ug_a02/specs.htm # lspci -v [...] 0000:06:00.0 Ethernet controller: 3Com Corporation 3CCFE575CT Cyclone CardBus (rev 10) Subsystem: 3Com Corporation FE575C-3Com 10/100 LAN CardBus-Fast Ethernet Flags: bus master, medium devsel, latency 64, IRQ 11 I/O ports at 4800 [size=268M] Memory at 11000000 (32-bit, non-prefetchable) [size=128] Memory at 11000080 (32-bit, non-prefetchable) [size=128] Expansion ROM at 00020000 [disabled] Capabilities: [50] Power Management version 2 [...] # Usually the card just doesn't work: # mii-tool -v No MII transceiver present!. # Once in a while I boot the laptop and the card does work: # mii-tool -v eth0: autonegotiation failed, link ok product info: TDK 78Q2120 rev 11 basic mode: autonegotiation enabled basic status: autonegotiation complete, link ok capabilities: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD advertising: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD # # mii-tool -F 10baseT-HD # # mii-tool -v eth0: 10 Mbit, half duplex, link ok product info: TDK 78Q2120 rev 11 basic mode: 10 Mbit, half duplex basic status: link ok capabilities: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD advertising: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD # Note that autonegotiation is disabled on our network (don't ask me why). This is why I have to force a 10baseT-HD connection. But anyway, in this case the network card does work. Please tell me if you need additional information such as boot messages. Version-Release number of selected component (if applicable): kernel-2.6.1-1.65 How reproducible: Sometimes
Mmmh... Actually this could be a duplicate of bug 102685. Disabling kudzu at startup makes the card work better as suggested: at least mii-tools doesn't report "No MII transceiver present!.". However many packets are still dropped: # ping 172.16.4.36 PING 172.16.4.36 (172.16.4.36) 56(84) bytes of data. From 172.16.4.23 icmp_seq=0 Destination Host Unreachable [...] 64 bytes from 172.16.4.36: icmp_seq=54 ttl=255 time=9245 ms [...] ping: sendmsg: No buffer space available [...] --- 172.16.4.36 ping statistics --- 100 packets transmitted, 66 received, +12 errors, 34% packet loos, time 142982ms rtt [...] # # ping 172.16.4.36 PING 172.16.4.36 (172.16.4.36) 56(84) bytes of data. From 172.16.4.23 icmp_seq=0 Destination Host Unreachable [...] --- 172.16.4.36 ping statistics --- 25 packets transmitted, 0 received, +21 errors, 100% packet loos, time 23989ms rtt, pipe 5 #
Created attachment 97877 [details] dmesg
please try the latest kernel, there have been a *lot* of network driver fixes in the last few days
I've installed kernel-2.6.3-1.97 and it fixes both problems: * the interface is not messed up when running kudzu * no more packet loss Thanks!
I have upgraded to kernel-2.6.3-1.106 and kernel-2.6.3-1.118 and now I experience network problems again. Is it OK to reopen this bug report? Do you need additional information? This time the device isn't even visible to the system: # lspci | fgrep 3Com # # kudzu --probe | fgrep 3Com # Of course mii-tool reports no device: # mii-tool -v no MII interfaces found # Module 3c59x is not loaded: # lsmod Module Size Used by ext3 115240 2 jbd 72472 1 ext3 #
I have reinstalled Fedora Core Test 2 from scratch and directly updated to kernel-2.6.3-1.164. Now the network runs just fine. Either this was a problem with kernel-2.6.3-1.106 and kernel-2.6.3-1.118, or the successive updates had messed up the system. I noticed some errors in the post-install scripts or translated descriptions of some packages, which may account for the problem...
I'm terribly sorry to have opened and closed this issue so many times, but the problem appears to be transient, so the network sometimes works and usually doesn't. So I still have problems, both with kernel 2.6.3-1.146 and 2.6.3-1.153. At least the newtork card seems to be detected correctly: # lspci | fgrep 3Com 0000:06:00.0 Ethernet controller: 3Com Corporation 3CCFE575CT Cyclone CardBus (rev 10) # mii-tool eth0: 10 Mbit, half duplex, link ok # The network configuration looks OK to me: # ifconfig eth0 eth0 Lien encap:Ethernet HWaddr 00:01:03:A8:6C:AD inet adr:172.16.4.23 BCast:172.16.7.255 Masque:255.255.252.0 adr inet6: fe80::201:3ff:fea8:6cad/64 Scope:Lien UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:653 errors:0 dropped:0 overruns:1040 frame:0 TX packets:142 errors:7 dropped:0 overruns:0 carrier:2 collisions:11 lg file transmission:1000 RX bytes:56271 (54.9 Kb) TX bytes:11243 (10.9 Kb) Interruption:11 Adresse de base:0x4800 # but there are lots of packets dropped (when pinging a machine on the same network) between 100% (most often) and 0% (sometimes).
Ouch! My current kernel is 2.6.3-2.1.253, not 2.6.3-1.153. Sorry.
I've upgraded to FC2 test2 and the network has been running for days. DNS lookups, updates, and large file transfers work just fine. I guess something used to go wrong with FC2 test1 updates, or something like that. Case closed.