View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001476 | unreal | ircd | public | 2004-01-06 20:17 | 2004-05-09 14:00 |
| Reporter | Praetorian_ | Assigned To | |||
| Priority | normal | Severity | crash | Reproducibility | always |
| Status | closed | Resolution | open | ||
| Product Version | 3.2-beta19 | ||||
| Summary | 0001476: OpenBSD: Crash on /rehash | ||||
| Description | ircd crashes on any /rehash. Tested with modules loaded and no modules (besides commands.so) in the configuration file | ||||
| Steps To Reproduce | 1. start 2. oper 3. /rehash 4. rehashes then crashes seconds, or milliseconds after | ||||
| Additional Information | SSL is enabled. Remote includes most disabled. | ||||
| 3rd party modules | |||||
|
|
Could you do: gdb src/ircd <corefile> bt |
|
|
#0 0x4025d7d0 in ?? () #1 0x1a311 in main (argc=1, argv=0xcfbfdc5c) at ircd.c:1315 [Removed Core File] edited on: 01-07-04 20:14 |
|
|
did you do any source code modifications? |
|
|
no, there shouldn't be any source code modifications in it. just downloaded latest cvs to try and see if it still exists in that.. will try non-ssl then ssl incase its that.. let you know in a bit |
|
|
Correction to what I said before.. There was one modification (remembered on doing make again), and that was to get OpenBSD 3.3 to compile with SSL on the default openssl version as per this bugreport: http://bugs.unrealircd.org/bug_view_advanced_page.php?bug_id=0001079 Yes, understand the downside (exploitable) but its restricted internal only atm.. I understand that source modifications to the ircd code when counting as a bug aren't considered because of variable factors, but i do believe this shouldn't matter to cause this problem - especially when i have previously /rehash 'd while using the same build.. sorry for the hassle if this is discarded for that reason Meanwhile .. with the latest CVS .2033: With SSL (and made to work with obsd3.3): Same Error. Without SSL (and no code edit): Same Error. (gdb) bt #0 0x401527d0 in ?? () #1 0x19c9d in main (argc=1, argv=0xcfbfdac0) at ircd.c:1317 (gdb) atleast that proves it wasn't the change ... but still funny its occurring .. same error (sorry for delay in btw posts - machine is slow) |
|
|
Note: line 1317 is DoEvents(); My best guess is that somehow the HTM module isn't freeing up its events (which just doesn't seem possible looking at the code). Unfortunately though, I don't have access to any OpenBSD systems to try and play around with this on. |
|
|
I'd offer the machine for it to be used to test it, but its laggy (dialup) and machine is slow (p100 - 40mb ram ;p) Edit: possibly make this unprivate and see if someone else has an openbsd machine with same problems/similar .. or can test with? ;) edited on: 01-07-04 20:18 |
|
|
codemaster/syzop, Irish has agreed to set you up a shell on his OpenBSD box for you to use to test/debug with. Get in contact with him @ irc.pimpwar.com in the admin channel. |
|
|
Based on doing some serious testing, the conclusion is that OpenBSD is broken, not Unreal. We might possibly be able to work around this, but I'm not making any guarantees because, to be quite honest, I'm getting rather tired of having to add hacks for OpenBSD. OpenBSD seems to do everything wrong, or everything in it is broken. Honestly, I can't for the life of me understand why people seem to think OpenBSD is such a great, stable, secure OS. I'm just looking at the OpenBSD man page for the dl*() functions (which is what modules uses), and it lists a bug that has been in the dlopen() function since OpenBSD 2.2, the current is 3.4! 13 releases and it still isn't fixed. OpenBSD has decided to ignore all standards, do things their own way, lie in the documentation about how things are done, and then expect coders to deal with it. I'm getting rather tired of dealing with it. |
|
|
I have this exact problem on OSF4. Any workaround yet? Appreciate it. |
|
|
[12:24:55] < Yes! I'd love to! This looks like the same exact bug and affects me also. OSF1 endor V4.0 1229 alpha alpha unknown Tru64 [email protected] |
|
|
Digital UNIX V4.0F (Rev. 1229); Wed Feb 18 15:15:26 EST 2004 |
|
|
as codemastr says, dlsym() seems seriously broken on openbsd.. or at least on the openbsd versions we tested (might be fixed in latest, but perhaps I'm just too optimistic). I never worked with your OS so I don't know if it really is affected by the same issue, but if it is.. then there's not much we can do... Well, you could statically link the ircd (=no modules support) but that requires you to write a new makefile + probably X modifications. We are not the only one that is enjoying these OS bugs (or should I say.. libc/dl), anope has the same issue.. And they too decided not to support modules on OpenBSD. *edit* we worked >10 hours on this bug and it was really f$#&@cked as hell.. so at least I personally am not thinking about working on this bug ever again, sorry. */edit* edited on: 2004-04-23 13:47 |
|
|
This appears to be fixed with the current build of OpenBSD 3.5 (I've tried to crash it with /rehash but it just won't :) I'm currently using the latest snapshot as the final wont be released until Saturday, May 1st. Cheers! |
|
|
So this bug is fixed or not?:P |
|
|
Confirmed to be fixed in openbsd 3.5. Also fixed the openssl compile error by using the proper openssl include files when I was at it. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2004-01-06 20:17 | Praetorian_ | New Issue | |
| 2004-01-06 20:24 | syzop | Note Added: 0004562 | |
| 2004-01-06 20:29 | Praetorian_ | Note Added: 0004563 | |
| 2004-01-06 20:44 | syzop | Note Added: 0004565 | |
| 2004-01-06 20:57 | Praetorian_ | Note Added: 0004567 | |
| 2004-01-06 22:47 | Praetorian_ | Note Added: 0004569 | |
| 2004-01-07 04:13 |
|
Note Added: 0004573 | |
| 2004-01-07 12:51 | syzop | Summary | Crash on /rehash => OpenBSD: Crash on /rehash |
| 2004-01-07 20:14 | Praetorian_ | Note Edited: 0004563 | |
| 2004-01-07 20:17 | Praetorian_ | Note Added: 0004576 | |
| 2004-01-07 20:18 | Praetorian_ | Note Edited: 0004576 | |
| 2004-01-07 22:40 | syzop | View Status | private => public |
| 2004-01-07 22:40 | syzop | Status | new => feedback |
| 2004-02-23 17:09 | Tony | Note Added: 0005168 | |
| 2004-02-23 19:46 |
|
Note Added: 0005169 | |
| 2004-04-23 12:17 | x | Note Added: 0005914 | |
| 2004-04-23 12:25 | x | Note Added: 0005915 | |
| 2004-04-23 12:25 | x | Note Added: 0005916 | |
| 2004-04-23 13:42 | syzop | Note Added: 0005917 | |
| 2004-04-23 13:47 | syzop | Note Edited: 0005917 | |
| 2004-04-27 22:43 | schantzk | Note Added: 0005935 | |
| 2004-04-30 09:11 | vonitsanet | Note Added: 0005962 | |
| 2004-05-09 14:00 | syzop | Status | feedback => closed |
| 2004-05-09 14:00 | syzop | Note Added: 0006142 |
