|
From: Bob D. <bd...@sr...> - 2005-07-25 14:57:19
|
Thank you, Sergey. That does seem to explain it. You're right though, it may be hard to find the signal offender. I'll try shutting down a couple of apps to see if I can find the culprit. I'm downloading 2.6.12, right now. Thanks again. I appreciate it. Bob On Sat, 2005-07-23 at 15:54 +0400, Sergey Vlasov wrote: > On Fri, 22 Jul 2005 15:25:22 -0400 Bob Dusek wrote: > > > ==14005== Signal 11 (SIGSEGV) appears to have lost its siginfo; I can't go on. > > ==14005== This may be because one of your programs has consumed your > > ==14005== ration of siginfo structures. > > This looks like a known problem: > > http://kerneltrap.org/mailarchive/1/message/25599/thread > > Basically, some program on your system is building up a large queue of > pending signals, and this causes the siginfo data for other signals to > be dropped (the number of queued signals is limited per user; some older > kernels had only a system-wide limit for it). However, Valgrind > absolutely needs siginfo for SIGSEGV. > > A workaround is to track down the offending program and avoid running it > while using Valgrind (but this may be hard, given the fact that the > kernel does not report the number of queued signals anywhere in /proc). > > Seems that the problem is finally solved in kernel 2.6.12: > > http://linux.bkbits.net:8080/linux-2.6/cset@422dec58vwDdDoiI1M3rszWXpJubSw?nav=index.html|src/|src/kernel|related/kernel/signal.c |