The following has be reported by IBM LTC:
fake_ll not workingfor QETH device
Software Environment:RHEL-3 RC1
Steps to Reproduce:
1. Add the fake_ll to the parm in chandev.conf
Followed the directions in Linux For zSeries and S/390 Device Drivers and
Installation Commands July 25, 2003 Chapter 19, Enabling OSA-Express QDIO
devices in Linux for DHCP and tcpdump.
eth0 is the interface used in this case.
2. Reboot the system
3. Start "ethereal" and capture packets for eth0
Capturing on eth0
0.000000 40:00:40:06:3e:d1 -> 45:10:00:4c:20:99 0x090c Ethernet II
0.132967 40:00:7d:06:52:a3 -> 45:00:00:28:cf:fa 0x0938 Ethernet II
0.159949 40:00:40:06:3e:58 -> 45:10:00:c4:20:9a 0x090c Ethernet II
0.333547 40:00:7d:06:52:a2 -> 45:00:00:28:cf:fb 0x0938 Ethernet II
0.333580 40:00:40:06:3e:9f -> 45:10:00:7c:20:9b 0x090c Ethernet II
Packets should have been valid TCP packets. It appears that the LLC headers
are still not on the packets.
Additional Information: I get the same results on both 31 and 64 bit systems.
It's Boeblingen's fault, but I fixed it (*beaming* :-)
See the attached patch.
Created attachment 95089 [details]
Fully working, but with P3s
Peter, Not sure what you mean by P3's. Is that a 390 box. Also, will this be
picked up in the next release of Red Hat or do I need to go back to Boeblingen.
Duane, please look at the patch - "P3" is a tag for debugging messages.
Final version won't have them, naturally.
I think I can get this into U1.
I also sumbitted it to Utz Bacher for inclusion into DeveloperWorks.
If you are interested helping it to happen, you might want to poke
him for opinion/rejection/acception.
I observe that Utz made himself an account in RH Bugzilla,
let me try to add him to cc: for this bug...
Thanks for the update, I applied the patch and it is working fine. I'll nudge
Ernie approved, I'm resending for commit now.
I dropped all unrelated fixes (such as dst_link_failure removal).
Duane, please test and reopen or close, according to results.