Azerothcore-wotlk: World Crash (SIGSEGV, Segmentation fault)

Created on 13 May 2020 · 37Comments · Source: azerothcore/azerothcore-wotlk

Crash log:

https://gist.github.com/CosmeF/c8b4427be4eb5b105723ee8b2d01425a

SMALL DESCRIPTION:

Server crash, gdb generated the log above.

STEPS TO REPRODUCE THE PROBLEM:

Not a clue, sorry

BRANCH(ES):

master

AC HASH/COMMIT:

f97d494ffc90c3b3dde041564d6b2ca20ec22905

OPERATING SYSTEM:

Ubuntu 18.04

MODULES:

mod-anticheat-master
mod-boss-announcer
mod-eluna-lua-engine-master
mod-emblem-transfer
mod-premium
mod-reward-played-time-master
mod-transmog-master

OTHER CUSTOMIZATIONS:

I've lowered conflarate delay to 0 on https://github.com/azerothcore/azerothcore-wotlk/blob/master/src/server/game/Entities/Unit/Unit.cpp#L3928

Source

CosmeF

Most helpful comment

Anyways.
Possible crash fix:

        if (!me->GetMap()->IsDungeon())
        {

change to:

        if (me && me->FindMap() && !me->GetMap()->IsDungeon())
        {

But I dont think that this smart action should actually work for gameobjects ????

We can restrict it from loading on gameobjects like the other actions that are limited creatures.

P-Kito on 23 Jun 2020

👍2

All 37 comments

@CosmeF Looks like its related to SAI but its a useless crash log without symbols being loaded, cause the server is compiled in release.

Need a log in Debug or RelWithDebInfo.

Meltie2013 on 14 May 2020

@Meltie2013 Thanks a lot. I've checked the flag for aditional debug code, so hopefuly I'll catch some extra info next time, so I can share it on this issue.

CosmeF on 14 May 2020

This server error appears periodically. Is there a solution? How to output the error log, my logs are empty. in the logs this error "Segmentation fault"

hlainin on 26 May 2020

@hlainin You need to build in Debug if you want a useful log

Kitzunu on 30 May 2020

@CosmeF, did you solve? is there any progress?

Helias on 14 Jun 2020

Server did crash but haven't been able to register a more detailed crash log. Now i'm still waiting for it to crash again and check if the crash log works with some extra flags enabled.

CosmeF on 19 Jun 2020

👍1

Did you build in RelWithDebInfo or Debug this time?

Kitzunu on 19 Jun 2020

👍1

Yes, the build is now running whith coredebug flag. Still waiting for the crash though.

CosmeF on 19 Jun 2020

Crashed again, segmentation fault. Log appears to be the same than last time. Am I doing something wrong? I use "-WITH_COREDEBUG=1" option on cmake, using GDB. The modules are the same since last time.

https://gist.github.com/CosmeF/4566730670e4170e42efe167a195f86d

CosmeF on 20 Jun 2020

I have no idea what system you use, but could you paste the entire line of the Worldserver please?

Like this
```
AzerothCore rev. ebec48e6fd93+ 2020-06-14 23:12:24 +0200 (master branch) (Win64, RelWithDebInfo) (worldserver-daemon) ready...

Kitzunu on 20 Jun 2020

SmartScriptHolder

SMART_EVENT

Map::VisitNearbyCellsOf(WorldObject*, TypeContainerVisitor > > > > >&, TypeContainerVisitor > > > > > >&, TypeContainerVisitor > > > > >&, TypeContainerVisitor > > > > > >&) ()

tatense on 20 Jun 2020

@Kitzunu Sure, here it is. Sorry for not pointing out the current commit the system is running on.
AzerothCore rev. 34077115a5de+ 2020-06-18 12:04:56 +0200 (master branch) (Unix, Release) (worldserver-daemon) ready...

CosmeF on 21 Jun 2020

You have compiled in Release that is why you dont get info from the crash log

I suggest you compile in RelWithDebInfo

Kitzunu on 21 Jun 2020

Thanks. Now it shows (Unix, RelWithDebInfo). I'll post the log as soon as I can, when the crash shows up again.

CosmeF on 21 Jun 2020

Sorry for the previous logs. Hopefully this helps now.

https://gist.github.com/CosmeF/436a0f273c3c56a967fc64dd79ccf597

Aparently is linked to game object id 124371 (Uldaman "Keystone"?)

CosmeF on 22 Jun 2020

@P-Kito

smart_scripts

SELECT * FROM smart_scripts WHERE entryorguid = 124371

| entryorguid | source_type | id | link | event_type | event_phase_mask | event_chance | event_flags | event_param1 | event_param2 | event_param3 | event_param4 | event_param5 | action_type | action_param1 | action_param2 | action_param3 | action_param4 | action_param5 | action_param6 | target_type | target_param1 | target_param2 | target_param3 | target_param4 | target_x | target_y | target_z | target_o | comment |
| ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- |
| 124371 | 1 | 4 | 0 | 61 | 0 | 100 | 0 | 0 | 0 | 0 | 0 | 0 | 38 | 0 | 0 | 0 | 0 | 0 | 0 | 19 | 7228 | 100 | 0 | 0 | 0 | 0 | 0 | 0 | Keystone - On Gossip Hello - Set In Combat With Zone |

Could this crash have to do with your changes of Combat with zone?

Kitzunu on 22 Jun 2020

@CosmeF Try reverting this commit and come back in a day or two to see if you got a new crash https://github.com/azerothcore/azerothcore-wotlk/commit/f381bd47cb1a1b5c0f4a1e2acab4327e86193e90

Kitzunu on 22 Jun 2020

@Kitzunu I've conducted a small test on the test realm. With f381bd4, on AzerothCore rev. 435de8a1355b, the activation of the Keystone on Uldaman started the door unlock sequence, and just after the door shows a blue glow, the server crashes, whith the same error log showed before.

On the same server, but doing a revert to commit f381bd4, the same test showed no crash at all. So I think you were right on pointing this commit as troublesome for this crash.

CosmeF on 22 Jun 2020

since when action 38 is allowed for gameobjects?

P-Kito on 23 Jun 2020

Is this custom SAI? Or blizzlike?

P-Kito on 23 Jun 2020

It is in the Main repo, so it is nothing custom that he's done

Kitzunu on 23 Jun 2020

But I think gameobjects cannot be SetInCombatWithZone O_o

We should limit this action to creatures only, right?

P-Kito on 23 Jun 2020

Is this SAI executed in a dungeon?
If it is, then it had nothing to do with my changes, since I only touched it when
IsDungeon() is true ;-)

P-Kito on 23 Jun 2020

Anyways.
Possible crash fix:

        if (!me->GetMap()->IsDungeon())
        {

change to:

        if (me && me->FindMap() && !me->GetMap()->IsDungeon())
        {

But I dont think that this smart action should actually work for gameobjects ????

We can restrict it from loading on gameobjects like the other actions that are limited creatures.

P-Kito on 23 Jun 2020

👍2

Yea sure we could restrict it @P-Kito

smart_scripts

SELECT * FROM smart_scripts WHERE source_type = 1 AND action_type = 38

Only that gameobject uses it, but then we'd need to find a workaround (Maybe this is not even blizzlike and not needed?)

Kitzunu on 23 Jun 2020

Can a gameobejct be in combat with something?
If no, than this sai can be deleted.

P-Kito on 23 Jun 2020

If I crossreference to TC, they have no SAI with this Action for Gobs.

Kitzunu on 24 Jun 2020

There you have your answer...

P-Kito on 24 Jun 2020

I've got a different segmentation fault crash. Here is the log. If is not related to the previous one, please ignore and I'll post this on another issue. Running: Worldserver crash AzerothCore rev. 34077115a5de+ (master branch) (Unix, RelWithDebInfo)

https://gist.github.com/CosmeF/35b7e8a6697cf44bd96e9dcc6c4930c7

CosmeF on 24 Jun 2020

not related

Kitzunu on 24 Jun 2020

Hi, everybody. I have a "Segmentation fault" error. I made a RelWithDebInfo build, but I don't get a report file when the server crashes. Please can you tell me how to get the report file. Debian 9

hlainin on 6 Jul 2020

@hlainin https://www.azerothcore.org/wiki/How-to-ask-for-help

FrancescoBorzi on 6 Jul 2020

I didn't ask how to create a theme. I can't get a crash log file

hlainin on 6 Jul 2020

Can a gameobejct be in combat with something?
If no, than this sai can be deleted.

Since I've deleted the conflicting SAI entry, no more crashes have ocurred. I've conducted tests on Uldaman using the keystone, and it shows no problems. So thank you guys.

CosmeF on 7 Jul 2020

👍1

This never got resolved on the core @Helias

Kitzunu on 7 Jul 2020

👍1

@CosmeF could you open a PR about this?

Helias on 8 Jul 2020

See, just a false SAI... Gameobjects shouldnt "SetInCombatWithZone" :P

P-Kito on 17 Jul 2020

Was this page helpful?

0 / 5 - 0 ratings