[Bug] Bot crashed for some reason

teq · March 11, 2019, 4:20pm

hi ryan,

i use the new 21.7.7 with 5 chars on LegionSRO. 2 of them are leveling. 3 of them are stalling.

the stalling ones work best.

the leveling ones crash right when botting is started. it doesnt matter where they start (trainplace, hotan, donwhang …) or if they got started manually using the start bot button or the auto start botting after (re)connect.

i use wine-staging on a linux server.

here is the crash dump wine provides me (cutted at some point due to login-data as command-line arguments; let me know when u require more):

Unhandled exception: page fault on read access to 0xc63ed9ff in 32-bit code (0x004a807e).
Register dump:
 CS:0023 SS:002b DS:002b ES:002b FS:0063 GS:006b
 EIP:004a807e ESP:0ecc636c EBP:0ecc63d8 EFLAGS:00010292(  R- --  I S -A- - )
 EAX:be390a5f EBX:05449548 ECX:00000000 EDX:0805cfa0
 ESI:0ecc64d8 EDI:0ecc64cc
Stack dump:
0x0ecc636c:  dc968cd0 05447f70 0ecc6560 05447f70
0x0ecc637c:  0ecc6504 00000001 00183a4c 0ecc6390
0x0ecc638c:  0ecc63d4 0ecc64cc 0805cfa0 006f7d11
0x0ecc639c:  dc968cdc 05447f70 be390a63 0805cfa0
0x0ecc63ac:  0ecc6504 00000000 00000000 00000000
0x0ecc63bc:  be390a5f 00000000 00000000 dc968cd0
Backtrace:
=>0 0x004a807e EntryPoint+0xffffffff() in phbot (0x0ecc63d8)
  1 0x004b581a EntryPoint+0xffffffff() in phbot (0x0ecc653c)
  2 0x006f1f93 EntryPoint+0xffffffff() in phbot (0x0ecc66e8)
  3 0x006f1087 EntryPoint+0xffffffff() in phbot (0x0eccfc68)
  4 0x01890de2 EntryPoint+0xffffffff() in phbot (0x0eccfe70)
  5 0x00d013be EntryPoint+0xffffffff() in phbot (0x0eccfe98)
  6 0x1002c01d $I10_OUTPUT+0x10f9a() in msvcr120 (0x0eccfed0)
  7 0x1002c001 $I10_OUTPUT+0x10f7e() in msvcr120 (0x0eccfedc)
  8 0x7bcae194 call_thread_func_wrapper+0xb() in ntdll (0x0eccfeec)
  9 0x7bcb1ab7 RtlCaptureStackBackTrace+0x246() in ntdll (0x0eccffdc)
  10 0x7bcae186 call_thread_exit_func+0x31() in ntdll (0x0eccffec)
0x004a807e EntryPoint+0xffffffff in phbot: movl	0x0(%eax,%edx,1),%ecx
Modules:
Module	Address			Debug info	Name (107 modules)
PE	  330000-  3a1000	Deferred        msvcp120
PE	  3b0000-  3cd000	Deferred        vcomp120
PE	  400000- 41d7000	Export          phbot
PE	10000000-100ee000	Export          msvcr120
PE	1e000000-1e2c4000	Deferred        python34
ELF	7b400000-7b804000	Deferred        kernel32<elf>
  \-PE	7b420000-7b804000	\               kernel32
ELF	7bc00000-7bd38000	Dwarf           ntdll<elf>
  \-PE	7bc40000-7bd38000	\               ntdll
ELF	7c000000-7c004000	Deferred        <wine-loader>
ELF	7cf62000-7cfa5000	Deferred        uxtheme<elf>
  \-PE	7cf70000-7cfa5000	\               uxtheme
ELF	7cfa5000-7cfac000	Deferred        libxfixes.so.3
...

got some idea what triggers this issue? already tried to uncheck “use attack pet”. no change.

teq · March 11, 2019, 4:26pm

fyi i dont use any conditions with both leveling chars

Ryan · March 11, 2019, 5:30pm

I’m not sure how to find the cause since it isn’t crashing for me while training. The only change was PK2 reading and I doubt that could be it.

Delirus · March 11, 2019, 6:06pm

I dont got such problems too. My 24 (Training)bots are running fine with wine

Ryan · March 11, 2019, 6:08pm

The latest version though?

Delirus · March 11, 2019, 6:09pm

Jup v21.7.7 and v1.7.9 Manager

teq · March 11, 2019, 6:38pm

did this with both leveling chars where i have the mentioned issue:

removed phBot.exe, downloaded v21.7.6, pressed “skip that version”, started the bot => works
removed phBot.exe again, downloaded v21.7.7, removed “UpdateSkip=21.7.7” from the phBot.ini file, started the bot => crashed

Ryan · March 11, 2019, 6:43pm

Try turning off the collision stuff. If it’s a PK2 problem then those options could be it. When you start it in your training area it first loads the navmesh data out of the PK2. You could also delete your Data folder and see if it reloads the info without crashing.

teq · March 11, 2019, 6:51pm

well, yeah. u are right. when i disable collision detection it works. when i dont have a stall, it cares about the trainingplace and crashes for some reason. i removed all db files from data directory, let the bot regenerate the file but the issue still exists. maybe you can just check it for yourself using LegionSRO (https://lsro.eu/). its probably working with iSRO, i havent checked that yet

Ryan · March 11, 2019, 6:53pm

Might be your Wine version since it’s fine for me under iSRO and vSRO.

Which version of Wine are you using? @Delirus @teq

teq · March 11, 2019, 6:53pm

nope, i did the check a few minutes ago on my windows 10 machine without any wine or unix environment on top > same issue. its not related to wine. its related to LegionSRO for some reason. maybe some incompatibilty with older versions of vSRO or some

Delirus · March 11, 2019, 6:59pm

4.1 but as I said, I havn’t any problems. All bots are working fine until now^^ So I think its not wine too

Ryan · March 11, 2019, 7:09pm

I tested Legion and it is not crashing for me.

teq · March 11, 2019, 7:12pm

have u used some trainplace walkscript + enabled collision detection?

Ryan · March 11, 2019, 7:12pm

No, I’m testing data loading which was crashing for you, right?

teq · March 11, 2019, 7:13pm

nope. its crashing when i have collision detection enabled, some walkscript to some trainplace and starting the bot once im logged in. then it crashes

Ryan · March 11, 2019, 7:15pm

OK that’s crashing for me which is good. Hopefully I’ll have a fix out soon.

teq · March 11, 2019, 7:15pm

thank you mate! strange right?

teq · March 11, 2019, 7:16pm

but keep relaxed, i can bot with that version now while having collision detection disabled. all good

Ryan · March 11, 2019, 7:46pm

Not sure why but I think some of their data in the PK2 is broken so it’s loading bad data. It’s not happening on my vSRO server and everything loads correctly there.

Do you know if collision detection actually did anything on that server previously? It’s not able to load any of the objects from it and it seems fine on my server, so I don’t think it’s the new PK2 reading I rewrote.

Edit: I tested vSRO 2 job and it does not crash so I think it’s just something with that server and the previous version not crashing was just sheer luck.