fixed Game Crash (including server) via waypointing

[MartinCo]

Hi ED,

For a long time we've been having issues with "bad waypoints" causing server crashes, whilst this has always been intermittent and with many tens of units being moved at a given time over the course of a PvP campaign it can be hard to find out which unit was moved from where to where that may have caused the crash.

2023-07-02 22:41:59.407 INFO    EDCORE (8764): OS-Version: 10.0.20348 () 0x110-0x3
2023-07-02 22:41:59.416 INFO    EDCORE (8764): 0x000000000006cb67 (edterrain4): buildMaxEdgeLengthUVSet + 0x1607
2023-07-02 22:41:59.430 INFO    EDCORE (8764): 0x0000000000229e1f (edterrain4): landscape5::lsa5pureFile::Mesh::tinyCopy + 0x118F
2023-07-02 22:41:59.438 INFO    EDCORE (8764): 0x0000000000230181 (edterrain4): landscape5::navGraph5File::findPath + 0x961
2023-07-02 22:41:59.462 INFO    EDCORE (8764): 0x00000000002605a1 (edterrain4): navGraphPath2::save + 0x4471
--
2023-07-02 22:45:09.218 INFO    EDCORE (14644): OS-Version: 10.0.20348 () 0x110-0x3
2023-07-02 22:45:09.227 INFO    EDCORE (14644): 0x000000000006cb67 (edterrain4): buildMaxEdgeLengthUVSet + 0x1607
2023-07-02 22:45:09.235 INFO    EDCORE (14644): 0x0000000000229e1f (edterrain4): landscape5::lsa5pureFile::Mesh::tinyCopy + 0x118F
2023-07-02 22:45:09.244 INFO    EDCORE (14644): 0x0000000000230181 (edterrain4): landscape5::navGraph5File::findPath + 0x961
2023-07-02 22:45:09.267 INFO    EDCORE (14644): 0x00000000002605a1 (edterrain4): navGraphPath2::save + 0x4471
--
2023-07-09 22:44:40.522 INFO    EDCORE (15060): OS-Version: 10.0.20348 () 0x110-0x3
2023-07-09 22:44:40.530 INFO    EDCORE (15060): 0x000000000006cb67 (edterrain4): buildMaxEdgeLengthUVSet + 0x1607
2023-07-09 22:44:40.539 INFO    EDCORE (15060): 0x00000000002301bc (edterrain4): landscape5::navGraph5File::findPath + 0x99C
2023-07-09 22:44:40.549 INFO    EDCORE (15060): 0x000000000026055e (edterrain4): navGraphPath2::save + 0x442E
2023-07-09 22:44:40.560 INFO    EDCORE (15060): 0x000000000025ec62 (edterrain4): navGraphPath2::save + 0x2B32
--
2023-07-17 02:35:36.722 INFO    EDCORE (11376): OS-Version: 10.0.20348 () 0x110-0x3
2023-07-17 02:35:36.730 INFO    EDCORE (11376): 0x000000000006cb67 (edterrain4): buildMaxEdgeLengthUVSet + 0x1607
2023-07-17 02:35:36.739 INFO    EDCORE (11376): 0x00000000002301bc (edterrain4): landscape5::navGraph5File::findPath + 0x99C
2023-07-17 02:35:36.747 INFO    EDCORE (11376): 0x000000000026055e (edterrain4): navGraphPath2::save + 0x442E
2023-07-17 02:35:36.756 INFO    EDCORE (11376): 0x000000000025ec62 (edterrain4): navGraphPath2::save + 0x2B32
--
2023-07-20 19:08:45.377 INFO    EDCORE (6160): OS-Version: 10.0.20348 () 0x110-0x3
2023-07-20 19:08:45.388 INFO    EDCORE (6160): 0x000000000006cb67 (edterrain4): buildMaxEdgeLengthUVSet + 0x1607
2023-07-20 19:08:45.398 INFO    EDCORE (6160): 0x00000000002301bc (edterrain4): landscape5::navGraph5File::findPath + 0x99C
2023-07-20 19:08:45.407 INFO    EDCORE (6160): 0x000000000025e8a1 (edterrain4): navGraphPath2::save + 0x2771
2023-07-20 19:08:45.418 INFO    EDCORE (6160): 0x000000000025f032 (edterrain4): navGraphPath2::save + 0x2F02

However after the most recent one, we were fortunate the user knew exactly which unit and where it was moved to - enabling me to get an almost 100% reproducible setup in the ME

- Open Beta 2.8.6.41363
- No Mods

For a first hand, just jump into taccom, select the Leclerc and set a way point to +41666 +4712 and you should experience a game crash, if it succeeds in calculating a route, set the way point again in a similar area and it should fail to calculate and crash.

This isn't unique to Syria, or Bassel (though we have had 3 crashes with units around basal in particular, so seems to be more prone to occur around here)

Please find attached track replay.
 

dcs.log-20230720-200509.zip

Edited July 20, 2023 by MartinCo

[MartinCo]

This unit / route no longer seems to cause the crash under 2.8.7.42583 (nor the other locations that caused a crash under the previous release).

I'll wait for another unit / location to exhibit the issue and hopefully recreate the issue again should it recur.

[MartinCo]

This has reared it's head again in a similar location. Track attached, and once again, seems 100% reproducible 

DCS 2.9.0.47168

This occurs both Single player (as shown here) and on dedicated servers when a player issues a waypoint to the units. This is causing some disruption, but intermittent as it seems to depend on origin and destination, but once it happens, we have to jump in first person to move them, or re spawn them somewhere else but this is all reactionary. It does occur on other maps as it seems to be path evaluation based, but i had to wait for a player to specifically identify which unit they moved when the server crashed (I do a full DB backup when the server exits unexpectedly to be able to get the exact unit position to place the unit in the miz file for attached recreations) 

Simple miz file with a single unit and a taccom slot attached.

No scripts, fully repaired, non modded DCS open beta install with clean profile

Steps:
- Load Miz
- Select Gepard
- Set route to the grassy area NW of Bassal Runway (z: 42150 x: +5530 - pretty much anywhere in that area causes the crash)
- Crash

Crash dump, including track attached

Video: 

waypoint-crash.miz dcs.log-20231107-200016.zip

Edited November 7, 2023 by MartinCo

[Flappie]

Crash reproduced and reported. Thanks!

[MartinCo]

Hi @Flappie

This (I'm pretty sure) resolved at one point (or at least, the terrain in this area updated that it was not causing a problem) however, this looks to have regressed and we're back to the same stack trace crashing the dedicated server as well as SP (as shown in the track), kicking players out from MP sessions.

DCS: 2.9.8.1214

Steps:
- Load Miz
- Select Gepard
- Set route to the grassy area NW of Bassal Runway (z: 42150 x: +5530 - pretty much anywhere in that area causes the crash)
- Crash

Please find new track file attached: -20241012-134510.zip

2024-10-12 13:45:10.255 INFO    EDCORE (55720): # -------------- 20241012-134510 --------------
2024-10-12 13:45:10.257 INFO    EDCORE (55720): DCS/2.9.8.1214 (x86_64; MT; Windows NT 10.0.22631)
2024-10-12 13:45:10.259 INFO    EDCORE (55720): E:\DCS\DCS World\bin-mt\edterrain4.dll
2024-10-12 13:45:10.261 INFO    EDCORE (55720): # C0000005 ACCESS_VIOLATION at 00007ff923e07987 00:00000000
2024-10-12 13:45:10.262 INFO    EDCORE (55720): SymInit: Symbol-SearchPath: 'E:\DCS\DCS World\bin-mt;', symOptions: 532, UserName: 'Martin'
2024-10-12 13:45:10.264 INFO    EDCORE (55720): OS-Version: 10.0.22631 () 0x100-0x1
2024-10-12 13:45:10.266 INFO    EDCORE (55720): 0x0000000000077987 (edterrain4): buildMaxEdgeLengthUVSet + 0x1767
2024-10-12 13:45:10.268 INFO    EDCORE (55720): 0x000000000024001c (edterrain4): landscape5::navGraph5File::findPath + 0x97C
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x000000000026f8c1 (edterrain4): navGraphPath2::save + 0x2861
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x0000000000270042 (edterrain4): navGraphPath2::save + 0x2FE2
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x000000000002435c (edterrain4): ITerrainImpl4::createPathCB + 0x14C
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x0000000000206c21 (Transport): wcRoute::createRouteWORoad + 0x91
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x00000000000ac28c (Transport): RouteManager::clear + 0xC7C
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x0000000000033b58 (edCore): ed::SyncTaskQueue::push_task + 0x478
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x0000000000048b06 (edCore): ed::SyncTaskQueue::push_task + 0x15426
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x0000000000032451 (edCore): ed::thread::_get_current_thread_id + 0x71
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x0000000000029333 (ucrtbase): _recalloc + 0xA3
2024-10-12 13:45:10.270 INFO    EDCORE (55720): 0x000000000001257d (KERNEL32): BaseThreadInitThunk + 0x1D

 

I have not experienced this on Caucasus or PG however I do feel the underlying ed methods should be more graceful in the event of any topology issues (if that is the trigger / root cause)

Thanks,
Martin

[MartinCo]

This specific crash reproduction does not occur on DCS 2.9.9.2280 (I will monitor for more)

[Flappie]

Thank you. I'll tag this thread as "fixed".

Ping me here if the bug comes back someday.