View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001598 | unreal | ircd | public | 2004-02-27 15:12 | 2004-03-31 18:37 |
| Reporter | liverbugg | Assigned To | syzop | ||
| Priority | normal | Severity | crash | Reproducibility | always |
| Status | closed | Resolution | fixed | ||
| Product Version | 3.2-RC1 | ||||
| Summary | 0001598: Opteron crashes when linked | ||||
| Description | We are runing Unreal on a AMD Opteron and it runs fine as a single server. But when we link in a IA32 machine (Athlon XP) if a client quits, then reconnects from the same IP, ircd segfaults on the Opteron. The 32 bit server stays up and will relink with the Opteron when ircd is restarted, but as long as clients keep reconnecting to the Opteron, it keeps crashing. We tested with 2 32 bit servers to make sure the config was right. They don't crash. | ||||
| Steps To Reproduce | Link a Opteron to a 32 bit intel server. connect to the Opteron. disconnect. reconnect. | ||||
| Additional Information | Both servers have 2 NICs, a public and a private. The servers link over the private network and clients connect to each public. | ||||
| 3rd party modules | |||||
|
|
Could you do a gdb backtrace? -- gdb src/ircd name.of.core.file bt -- (core file is usually called 'core' 'core.<number>' 'ircd.core' 'ircd.core.<number>' you get the idea...) Thanks :) |
|
|
Unfortunately without a gdb backtrace we can't help. |
|
|
For what it's worth: I wasn't able to reproduce this on ia64<->ia64 (itanium 2.. so not an opteron, don't have access to such a machine), and I cannot test ia64<->ia32 due to firewall restrictions, so again.. a backtrace would be helpful. |
|
|
15 minutes after the bugreport was posted a request for more info was made, now 5 days ago still no reply... closing bug. |
|
|
Sorry the admin for the opteron was away for the weekend. Heres the backtrace # gdb src/ircd core GNU gdb 6.0 Copyright 2003 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. There is absolutely no warranty for GDB. Type "show warranty" for details. This GDB was configured as "x86_64-pc-linux-gnu"...Using host libthread_db library "/lib/libthread_db.so.1". Core was generated by `/root/Unreal3.2/src/ircd'. Program terminated with signal 11, Segmentation fault. Reading symbols from /usr/lib/libssl.so.0.9.7...done. Loaded symbols for /usr/lib/libssl.so.0.9.7 Reading symbols from /usr/lib/libcrypto.so.0.9.7...done. Loaded symbols for /usr/lib/libcrypto.so.0.9.7 Reading symbols from /lib/libcrypt.so.1...done. Loaded symbols for /lib/libcrypt.so.1 Reading symbols from /lib/libnsl.so.1...done. Loaded symbols for /lib/libnsl.so.1 Reading symbols from /lib/libz.so.1...done. Loaded symbols for /lib/libz.so.1 Reading symbols from /lib/libdl.so.2...done. Loaded symbols for /lib/libdl.so.2 Reading symbols from /lib/libc.so.6...done. Loaded symbols for /lib/libc.so.6 Reading symbols from /lib64/ld-linux-x86-64.so.2...done. Loaded symbols for /lib64/ld-linux-x86-64.so.2 Reading symbols from tmp/99174207.commands.so...done. Loaded symbols for tmp/99174207.commands.so #0 find_server_quick_search (name=0x6513ed "irc.lanchelms.com") at aln.c:146 146 if (!match(name, lp->value.cptr->name)) (gdb) bt #0 find_server_quick_search (name=0x6513ed "irc.lanchelms.com") at aln.c:146 #1 0x000000000040fbf1 in find_server_quickx ( name=0x6513ed "irc.lanchelms.com", cptr=0x0) at aln.c:167 #2 0x0000000000451d98 in sendto_serv_butone_token_opt (one=0x6cb060, opt=1280, prefix=0x6513ed "irc.lanchelms.com", command=0x466d76 "SJOIN", token=0x46c1f6 "~", pattern=0x466d69 "%B %s :%s%s ") at send.c:784 #3 0x0000000000416fb0 in join_channel (chptr=0x6cb6f0, cptr=0x6cb060, sptr=0x6cb060, flags=1) at channel.c:3676 #4 0x000000000041778d in do_join (cptr=0x6cb060, sptr=0x6cb060, parc=774778414, parv=0x61fcc0) at channel.c:3967 #5 0x0000000000416e5e in m_join (cptr=0x6cb060, sptr=0x6cb060, parc=2, parv=0x61fcc0) at channel.c:3631 #6 0x000000000042a614 in parse (cptr=0x6cb060, buffer=0x6cb18c "JOIN", bufend=0x69b1d0 "") at parse.c:453 #7 0x0000000000429206 in dopacket (cptr=0x6cb060, buffer=0x0, length=0) at packet.c:138 0000008 0x00000000004300ce in read_packet (cptr=0x6cb060, rfd=0xc) at s_bsd.c:1476 #9 0x0000000000430951 in read_message (delay=1, listp=0x652d60) at s_bsd.c:1937 #10 0x0000000000425c79 in main (argc=2, argv=0xbffff878) at ircd.c:1431 |
|
|
Ok. Could you type the following in gdb: -- p *lp p *lp->value.cptr -- Also, just to be sure, could you check if you still have this problem with latest CVS? (use: http://www.vulnscan.org/UnrealIrcd/cvs/Unr3.2-20040301.tar.gz ).. A few memory corruption bugs were fixed in it but I don't know if they could have affected you. Do you use any of the following features: SSL, ziplinks, spamfilter, remote includes? Did you have any previous unrealircd installed before on this machine? (I presume not but...) if so, did you have any problems? Thanks. |
|
|
More info: It crashes on other client commands after they reconnect, not just /join. /oper and /whois crash. If the opteron server is started first, and then the other server started, clients can connect to the opteron once and have to disconnect and reconnect before it crashes like I said before. But if the opteron is restarted after a crash and relinks with the other server, then any client connecting and giving a command makes it crash, no disconnecting and reconnecting required. |
|
|
(gdb) p *lp Cannot access memory at address 0x536b4b37006ba320 (gdb) p *lp->value.cptr Cannot access memory at address 0x536b4b37006ba330 We use ziplinks but the crashes happen without them also. This is the first Unreal install on here. I'll give the cvs a try. |
|
|
hmm fun. dunnow if you can still do gdb (I hope you are installing in a different dir ;p... else nevermind and we'll wait at cvs results), could you do: p *Servers p *Servers->next p *Servers->next->next |
|
|
(gdb) p *Servers Cannot access memory at address 0x536b4b37006ba320 (gdb) p *Servers->next Cannot access memory at address 0x536b4b37006ba320 (gdb) p *Servers->next->next Cannot access memory at address 0x536b4b37006ba320 CVS crashes the same. Also, I don't see the user who joins to the opteron as being connected if I'm on the other server. |
|
|
Hm strange :|. Looks quite hard to debug too (memory corruption or something, perhaps related to make_link or add_server_to_table, but seems weird)... Could I get access to this machine (or another opteron) to debug it? If so, mail me at [email protected] or msg 'Syzop[AWAY]'/'Syzop' at irc.unrealircd.org. |
|
|
I'll talk to the admin about giving you access and email with the responce. |
|
|
Got a shell (thanks!) and I can confirm this an unreal bug... it's because of mixed int/long use, and since int is 32 bits and long 64 bits on opteron this causes problems (it also seems to be a bit different in paramter passing to functions). Found at least 1 crashbug caused by sendto_nickserv_cmd (or something), but I'm still having problems (I get randomly killed), judging from the gcc warnings I get there are quite some places left too (eg: /stats s where %d is used for sizeof() output). I'll have a look at this issue again on Sunday, so unfortunately after RC2.. it's not worth committing the bug I found right now anyway since there's more to fix too. |
|
|
Fixed several things/bugs now, let's see if that was enough :). |
|
|
[merged dup 0001651] |
|
|
I'll keep the report open for a few days to see if any opteron users still experience problems :). |
|
|
Haven't had a chance to test this because the server got powered off and no one has access to it until Sunday. Will test asap. |
|
|
ok, thanks :). |
|
|
Finaly had the chance to test. It looks like its fixed. If I run in to any other commands or actions that cause a similar crash I'll update this bug. Thanks for the fix. |
|
|
Ok, thanks for your help :). I'll just close the bug for now, if you have any crash again mail [email protected] or re-report it here (whatever you prefer). |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2004-02-27 15:12 | liverbugg | New Issue | |
| 2004-02-27 15:33 | syzop | Note Added: 0005232 | |
| 2004-03-01 21:54 | syzop | Note Added: 0005264 | |
| 2004-03-01 22:32 | syzop | Note Added: 0005266 | |
| 2004-03-02 15:42 | syzop | Status | new => closed |
| 2004-03-02 15:42 | syzop | Note Added: 0005270 | |
| 2004-03-02 19:24 | liverbugg | Status | closed => feedback |
| 2004-03-02 19:24 | liverbugg | Resolution | open => reopened |
| 2004-03-02 19:24 | liverbugg | Note Added: 0005274 | |
| 2004-03-02 19:41 | syzop | Note Added: 0005275 | |
| 2004-03-02 19:48 | liverbugg | Note Added: 0005276 | |
| 2004-03-02 19:51 | liverbugg | Note Added: 0005277 | |
| 2004-03-02 19:56 | syzop | Note Added: 0005278 | |
| 2004-03-02 19:58 | syzop | Status | feedback => new |
| 2004-03-02 19:58 | syzop | Resolution | reopened => open |
| 2004-03-02 20:06 | liverbugg | Note Added: 0005279 | |
| 2004-03-02 20:19 | syzop | Note Added: 0005280 | |
| 2004-03-02 20:23 | liverbugg | Note Added: 0005281 | |
| 2004-03-05 22:49 | syzop | Note Added: 0005315 | |
| 2004-03-05 22:49 | syzop | Assigned To | => syzop |
| 2004-03-05 22:49 | syzop | Status | new => assigned |
| 2004-03-05 22:49 | syzop | Projection | none => minor fix |
| 2004-03-05 22:49 | syzop | ETA | none => < 1 month |
| 2004-03-11 17:29 | syzop | Note Added: 0005429 | |
| 2004-03-15 10:18 | syzop | Note Added: 0005498 | |
| 2004-03-15 10:21 | syzop | Note Added: 0005499 | |
| 2004-03-15 10:21 | syzop | Status | assigned => feedback |
| 2004-03-15 10:21 | syzop | Resolution | open => fixed |
| 2004-03-15 10:21 | syzop | Projection | minor fix => none |
| 2004-03-15 10:21 | syzop | ETA | < 1 month => none |
| 2004-03-15 21:57 | liverbugg | Note Added: 0005504 | |
| 2004-03-15 23:16 | syzop | Note Added: 0005505 | |
| 2004-03-31 18:07 | liverbugg | Note Added: 0005698 | |
| 2004-03-31 18:37 | syzop | Status | feedback => closed |
| 2004-03-31 18:37 | syzop | Note Added: 0005701 |
