Trinitycore: [3.3.5] Crash worldserver related to FlightPathMovementGenerator

[TheWinchesters]

Using rev ce46a0d53c3dbd6d2837020686e3880d98c545ed I had this crash:

Thread 7 "worldserver" received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x7fff2d5ff700 (LWP 31846)] 0x00000000028ebf9a in FlightPathMovementGenerator::InitEndGridInfo (this=0x7fff2d7b0360) at /TrinityCore/src/server/game/Movement/MovementGenerators/FlightPathMovementGenerator.cpp:252 252 _endMapId = _path[nodeCount - 1]->MapID; //! MapId of last node

I think that this may be related to a3697052c3860b9bd5524b932fbda9c8690e9e72
I reverted that commit and I had no more crashes so far.

Branch: 3.3.5
Rev: ce46a0d53c3dbd6d2837020686e3880d98c545ed
OS: Windows 8.1 / Debian 9

[jackpoz]

please post the whole crashlog on https://gist.github.com/ , a single line is not enough for us to do anything

[TheWinchesters]

https://gist.github.com/TheWinchesters/7bb51b4335d343081c042199be1215f2

[illfated]

https://github.com/Winchester1987 does not exist anymore (and not the gist pages either), so you need to re-upload the gist to an existing page.

[TheWinchesters]

@illfated
Thank you. Updated.

[jackpoz]

please try after https://github.com/TrinityCore/TrinityCore/commit/2a06923879ccbcfff35cba4f35427eaf8c79dffd

[TheWinchesters]

I would love to try, but since I still dont get along with crash logs I would not know what to do to replicate what happened in the crashlog I posted here.

[Faq]

@TheWinchesters just update your core to revision what @jackpoz linked, there with it should be more info in crashlog to solve this.

[Treeston]

tagged issue has repro

[jackpoz]

what was in the end the cause of the crash ? a quest triggering 2 flight paths to a player one after another ?

[Killyana]

The spell 45883 was casted twice: Effet #1 Send Taxi : (792)
I suppose if you start a taxi when the char is already into a fly it will crash

[Treeston]

Trying to start a flight path while a flight path was already in progress.

Starting the second one would first setup m_taxi, then Remove any pre-existing flight movement generators, which would cause the logic added in a369705 to reset m_taxi.

You ended up with a flight path movement generator without any flight path data, which would crash.

The aforementioned Remove call was changed to an assertion in ead439f, which should catch stuff like this during testing.