View Issue Details

IDProjectCategoryView StatusLast Update
0006538unrealircdpublic2025-07-31 18:27
Reportervectr0n Assigned Tosyzop  
PrioritynormalSeveritycrashReproducibilityalways
Status resolvedResolutionfixed 
Product Version6.1.10 
Fixed in Version6.2.0-beta3 
Summary0006538: IRCd Crash on REHASH
DescriptionI think I might have found a crash bug in 6.1.10.
When I use `/rehash -global -dns` the IRCd will crash.
Seems to happen every time on multiple servers.
Steps To Reproduce/rehash -global -dns
TagsNo tags attached.
3rd party modules

Activities

syzop

2025-07-31 09:19

administrator   ~0023458

Thanks for the report, and setting it privately, crash reports tend to lead to quite a bit of speculation, and you never know how easy it is to trigger or how far reaching some bug can be, so it is good not to have this in public :).

Let's see...

1) Do you have a crash report on any of the crashed UnrealIRCd servers? Because I didn't see any and usually you would be prompted on ./unrealircd start if you want to send a crash report.
But of course, your IRCds may be restarted via cron, then it would not prompt. You can force checking (now) with:
/home/ircd/unrealircd/bin/unrealircd -R
(change path accordingly, of course)
If you answer yes to send the crash report, it would hopefully show me why/where it crashed exactly. That would help a lot.

2) If some servers crashed, but others did not crash, can you check with '/QUOTE DNS i' (i for info) if there is a pattern in those who crashed and those who didn't? E.g. them all being a certain c-ares version (c-ares is the resolver library we use).

3) Of course, the first thing I tried was to crash my own server (test network) but that didn't work. But don't let that distract us, that is not uncommon.

syzop

2025-07-31 17:19

administrator   ~0023470

I see the attempt to crash.unrealircd.org now from this morning. Possibly something I screwed up on our end, resulting in a http 400 error code. If i test-report a crash myself via the crash reporter it fails as well. I will look at that later. We want the crash reporter to work.. obviously :D.

syzop

2025-07-31 17:50

administrator   ~0023474

Thanks, got it :)

And now I see what is wrong and how to reproduce it. You have to do "REHASH -dns" twice.

syzop

2025-07-31 17:53

administrator   ~0023475

I think we can make this "public" since it is only triggerable by IRCOps (or trusted servers).

syzop

2025-07-31 17:56

administrator   ~0023476

Commited to git. The fix will be in 6.2.0-beta3 and later in the 6.2.0 stable release. As it is now 100% clear that only IRCOps (or trusted server traffic) can trigger this, I have changed the view status from private to public and put the commit and the release notes without any censoring or delay. Thanks again for the report!

https://github.com/unrealircd/unrealircd/commit/24fde4f889bce17f47f87e0af16511032ac7984f

commit 24fde4f889bce17f47f87e0af16511032ac7984f (HEAD -> unreal60_dev)
Author: Bram Matthys <[email protected]>
Date: Thu Jul 31 17:51:56 2025 +0200

    Fix crash on "REHASH -dns" (IRCOp only)
    Reported by vectr0n in https://bugs.unrealircd.org/view.php?id=6538

Issue History

Date Modified Username Field Change
2025-07-31 05:55 vectr0n New Issue
2025-07-31 09:19 syzop Note Added: 0023458
2025-07-31 09:19 syzop Assigned To => syzop
2025-07-31 09:19 syzop Status new => feedback
2025-07-31 17:19 syzop Note Added: 0023470
2025-07-31 17:50 syzop Note Added: 0023474
2025-07-31 17:53 syzop View Status private => public
2025-07-31 17:53 syzop Note Added: 0023475
2025-07-31 17:56 syzop Status feedback => resolved
2025-07-31 17:56 syzop Resolution open => fixed
2025-07-31 17:56 syzop Fixed in Version => 6.2.0-beta3
2025-07-31 17:56 syzop Note Added: 0023476
2025-07-31 17:58 syzop Summary IRCd Crash => IRCd Crash on REHASH
2025-07-31 18:15 syzop Priority urgent => normal
2025-07-31 18:25 syzop File Deleted: crash.report.core.1753972524.txt
2025-07-31 18:25 syzop File Deleted: crash.report.core.1753972536.txt
2025-07-31 18:26 syzop File Deleted: crash.report.core.1753933434.txt