|
Hello Dave,
Here we go again, this time with 3.1.5. I get Oopses that seem to have no real "reason", other than "sometimes when I insert a card"... I tried to trace some. Kernel 2.2.13, pcmcia 3.1.5, 3com cardbus 575TX on a 100Mbit hub, Sony Vaio PCG-F304 with Ricoh stuff (IIRC), eh... that's about it I guess. I boot with "init=/bin/sh" to not kill my filesystem all the time, then mount /tmp and /var to be able to use PCMCIA. While ejecting the card, I always seem to get a:
vortex_detach(eth0) called nonatomically from interrupt c01742e1 called nonatomically from interrupt c015a329 (This is not the problem). (FYI: System.map shows that c01742d4 is rtmsg_ifa and the next thing is c0174354 (inet_forward_change). c015a329 is c015a31c rtmsg_ifinfo. Now "sometimes" after reinserting the card I get a freeze at c015adb3 (dev_do_watchdog at c015ada0), and if my Vaio is in a bad mood, I can look very briefly at something that seems to start at c015adb3 and then keeps running with oopses or whatever, the only thing i can figure out is that address "c011127e" is coming by all the time. This is "c0111014 T schedule" I think and I cannot see what it tries to say (it says many, many things and does not wait for me to understand anything :) Just for the sake of it, I tried a cardctl reset. No errors. I repeated this two times. Still no errors. Unfortunately, eject did not work after that and an "ioctl() busy" or alike came by. Then... (sorry)... an oops, at c015ace6. ab38 is eth_header_cache. I hope I did enough to keep you busy some more hours. Since the bug addresses seemed so strange (laying in between to functions), I looked at System.map and vmlinuz. Their timestamps are identical, so I guess the above values are right. The kernel is a rather big one, with all types of networking stuff in it. Still, as this happens during PCMCIA insertion (or removal) I think this is the right place to start complaining? ;-) (Thanks for your help and prompt responses so far) Valentijn
|