On Tuesday 29 June 2004 03:31 pm, Daniel Lang wrote: > Hi again, > > I installed gdb6 from ports, which works. > It just confirms the addr2line result: > > (kgdb) l *0xc053932b > 0xc053932b is in witness_checkorder (/usr/src/sys/kern/subr_witness.c:898). > > > Brian Fundakowski Feldman wrote on Tue, Jun 29, 2004 at 02:49:46PM -0400: > [..] > > > > (What hurts most, is, that in one occasion I had a ddb prompt > > > and could call doadump() successfully. But after reboot, damn > > > /var was full, so savecore could not write it to disk, argl!). > > > > You can make /var/crash a symlink to a directory with more space. > > Yes, I usually have enough space there. It was an ill coincidence. :-/ > > Don: > Thanks for the hint to run savecore later on. I should have guessed > as long as there is not much going on and swap did not need to used > yet, I would have gotten a good core. Unfortunately it is to late > for now. I'll keep it in mind, though. > > John: > Ok, mutex/witness confusion is maybe going on. How would you suggest > to proceed. I am a little reluctant to boot the machine to multi-user > and wait for the next crash (which most of the time does not > produce a dump or ddb prompt), since I would have to hit the > reset button and right now, I'm at home now and have only remote > access. Unfortunately, I don't have any good ideas off the top of my head. If you have a crashdump and a working gdb binary of some sort then you can try looking at the sleepq_chains[] hash table in the core file to look for a mutex that is corrupted. -- John Baldwin <jhb_at_FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve" = http://www.FreeBSD.orgReceived on Tue Jun 29 2004 - 23:38:27 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:59 UTC