View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000301 | unreal | ircd | public | 2002-08-28 14:24 | 2003-02-10 22:04 |
| Reporter | jollino | Assigned To | syzop | ||
| Priority | normal | Severity | crash | Reproducibility | random |
| Status | closed | Resolution | fixed | ||
| Product Version | 3.2-beta12 | ||||
| Summary | 0000301: Random crashes w/ b11 and b12(gdb complaining about kill(), with traces of pthreads) | ||||
| Description | Beta12 crashes at random times, leaving no traces of errors - my bnc won't show me error messages right before I get knocked out, but other opers on the net say they've just seen my server splitting away. (gdb) bt #0 0x4007a1a1 in kill () from /lib/libc.so.6 #1 0x80aaa29 in default_action () #2 0x80aabcc in handle_thread_signal () #3 0x80ab0f7 in handle_one_signal () #4 0x80ab26e in pthread_handle_one_process_signal () #5 0x80b5006 in pthread_signal_sched () #6 0x80b44e0 in pthread_sched_wrapper () #7 0x80ab57b in sighandler () 0000008 0x4007a0b8 in sigaction () from /lib/libc.so.6 #9 0x807ef70 in find_cache_number (rptr=0x821f9e0, numb=0x821fac0 "") at res.c:1468 #10 0x807efd2 in make_cache (rptr=0x821f9e0) at res.c:1497 #11 0x807e904 in get_res (lp=0x81c36d8 "ÿÿÿÿ") at res.c:1029 #12 0x8083756 in do_dns_async () at s_bsd.c:2469 #13 0x8082bb8 in read_message (delay=1, listp=0x81d4f40) at s_bsd.c:1674 #14 0x8079fa8 in main (argc=1, argv=0xbffffae4) at ircd.c:1329 The string provided in frame #11 seems interesting, as it contains four times the character 0xff, which looks like a bug somewhere. (gdb) frame 11 #11 0x807e904 in get_res (lp=0x81c36d8 "ÿÿÿÿ") at res.c:1029 1029 cp = make_cache(rptr); (gdb) list 1024 1025 if (a > 0) 1026 { 1027 if (lp) 1028 bcopy((char *)&rptr->cinfo, lp, sizeof(Link)); 1029 cp = make_cache(rptr); 1030 #ifdef DEBUGMODE 1031 Debug((DEBUG_INFO, "get_res:cp=%#x rptr=%#x (made)", cp, rptr)); 1032 #endif 1033 | ||||
| Additional Information | This happened a couple of times with Unreal 3.2 beta11, and I was told on #unreal-support that it could be possibly pthreads-related; I waited for Unreal beta12 and compiled in the pthreads libraries provided with the package. It happens nevertheless. With beta11 it once crashed two servers at the same time, with about the same backtrace (same functions). | ||||
| Tags | No tags attached. | ||||
| 3rd party modules | |||||
|
|
The problem likely occurs because you are using the pthreads that comes with unreal. Rerun ./Config and when it asks if you want to use the system default threads say "yes". Hopefully that will correct the problem, if not, let us know. |
|
|
I recompiled using my system's pthreads (which was what b11 used too), and it crashed after around one hour. (gdb) bt #0 0x400c97f2 in memcmp () from /lib/libc.so.6 #1 0x8238dd8 in ?? () #2 0x807c6f0 in find_cache_number (rptr=0x8238970, numb=0x8238a50 "") at res.c:1468 #3 0x807c752 in make_cache (rptr=0x8238970) at res.c:1497 #4 0x807c084 in get_res (lp=0x81ae098 "ÿÿÿÿ") at res.c:1029 #5 0x8080ed6 in do_dns_async () at s_bsd.c:2469 #6 0x8080338 in read_message (delay=1, listp=0x81bee40) at s_bsd.c:1674 #7 0x8077728 in main (argc=1, argv=0xbffffae4) at ircd.c:1329 (gdb) frame 4 #4 0x807c084 in get_res (lp=0x81ae098 "ÿÿÿÿ") at res.c:1029 1029 cp = make_cache(rptr); (gdb) list 1024 1025 if (a > 0) 1026 { 1027 if (lp) 1028 bcopy((char *)&rptr->cinfo, lp, sizeof(Link)); 1029 cp = make_cache(rptr); 1030 #ifdef DEBUGMODE 1031 Debug((DEBUG_INFO, "get_res:cp=%#x rptr=%#x (made)", cp, rptr)); 1032 #endif 1033 I'm pretty confused. No oper on the net (not even the two other net-admins) saw anything about it, only 'Exiting ssl client Chieti.IT.Discussioni.Org: SSL_read(): Underlying syscall error', but I think that this error message is just a consequence to the crash, not its cause. |
|
|
I think it might be the cause, I know stskeeps has fixed a bunch of SSL errors. Would you be able to upgrade to the current devel version? cvs login (just press enter when it asks for a password) cvs update -r devel then ./Config -q && make Hopefully that will fix the problem. Also you didn't mention the OS, I assume it's Linux, but what distro? |
|
|
After upgrading to the latest devel like codemastr told me to, the crashes are happening again, but at a different location: (gdb) bt #0 0x400c97f2 in memcmp () from /lib/libc.so.6 #1 0x8231518 in ?? () #2 0x807c700 in find_cache_number (rptr=0x821d7f0, numb=0x821d8d0 "") at res.c:1468 #3 0x807c762 in make_cache (rptr=0x821d7f0) at res.c:1497 #4 0x807c094 in get_res (lp=0x81ae1b8 "ÿÿÿÿ") at res.c:1029 #5 0x8080ef6 in do_dns_async () at s_bsd.c:2478 #6 0x8080348 in read_message (delay=1, listp=0x81bef60) at s_bsd.c:1676 #7 0x8077728 in main (argc=1, argv=0xbffffae4) at ircd.c:1348 If this is unfixable, then any advice would be really appreciated. (like using another version, or whatever). |
|
|
This was probably fixed in beta14. |
