From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4.1) Gecko/20031003 Description of problem: The 3Com 3c575 PCMCIA adapter is not initiated properly during rhgb, although all the lines when showing extra information pass with service status OK. After nfslock service the status display changes from "[ OK ]" to "Ã OK Ã", I don't know if that's related though. Version-Release number of selected component (if applicable): rhgb-0.10.2-1 How reproducible: Always Steps to Reproduce: 1. Boot machine with rhgb 2. Try to access network (ping, www, ...) Actual Results: Network not working although the adapter seems to be up (module loaded and link light on) Expected Results: Working network adapter Additional info: A manual restart of pcmcia and networks services usually brings the adapter back and I have a working network. There is no problem with the network if I boot without rhgb with the nogui option passed to the kernel. I'm also not sure how this relates to the problem of other 3c59x based cards: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=100470 I've tried both the current 2.4.22-1.2087.nptl and arjan's 2.6.0-0.test7.1.52 with the same result (although with 2.6 I can't even manually restart the network service)
Created attachment 95116 [details] dmesg output with rhgb
Created attachment 95117 [details] dmesg output with nogui Oh yes, and this adapter worked without a hitch with RHL 7.3, 8.0, 9, Null, Phoebe etc. - before rhgb appeared anyway.
Hm, no actual difference between the ethernet messages there. The character set changing is a red herring; that's the keytable init script.
I don't know if it's only an ethernet problem: I have to restart pcmcia before I can get the card working, so it might be a cs problem only? Anything I can do to help debug this?
There is more rhgb+kernel oddity: I have an IBM Thikpad 600E with cs423x sound, and to get it working I used a workaround suggested by IBM http://www-1.ibm.com/support/docview.wss?rs=0&q=linux+600e&uid=psg1MIGR-4BP6Q6&loc=en_US&cs=utf-8&cc=us&lang=en which has worked great until Severn+rhgb, and works as expected w/o rhgb. If I boot with rhgb, I have to remove and load the cs4232 module once again after rhgb in order to get the sound working properly. And again there is no difference in dmesg output. So there seems to be a general problem with loading kernel modules (or at least 3c59x and cs4232) when rghb is up during init. Now running: kernel-2.4.22-1.2097.nptl initscripts-7.38-1 rhgb-0.11-1
Another thing, if I don't restart pcmcia manually after boot, the machine freezes when shutting down with Stopping pcmcia:
I've had that problem too, albeit very intermittently (i.e., it happens, but very rarely). When it does happen, the machine "unfreezes" and continues shutting down if I yank the 3com card out.
Problem still persists in Yarrow.
I had the same problem with a DELL Latitude CPx and 3COM PCMCIA NIC. After trying to figure out what the problem is for hours, I found this bugzilla post, disabled rhgb and everything works well.
Fedora Core w/ 2.6.1 ( and 2.6.0/2.4.x kernels) boot to GUI with rhgb causes /etc/rc.d/pcmcia/network NOT to be run. if either of above is NOT met ( either boot to 3, or removal of rhgb ) pcmcia/network starts fine. My machine: Dell C800 Aironet 350 Wireless Card I have edited rc.sysinit to search for some special variables, and then cp a special /etc/rc.d/pcmcia/network file depending upon what I enter. This file has several debug outputs, consisting of simple echo's that I don't see with rhgb in place
Edit the about to read : /etc/pcmcia/network NOT /etc/rc.d/pcmcia/network And yes, Im going to bed now :)
There is another workaround reported at bug 100920 I guess this one can be closed as a dup
Working as expected again in FC2 test2, closing.