[crash] Waypoints

[balacas]

I hope that you indeed have to find the problem!

[Schmoozerd]

Thanks guys!

[MrGhost]

Compiling and will test afterwards! Lets keep our fingers / hooves crossed

[ankso]

Good work ^^

[MrGhost]

45 minutes uptime with 300+ players and still counting. Still having my fingers crossed though! Will try to give you an update tomorrow

[DanqeR]

Updated to last revision 9447 but the server crashed again.

Core was generated by `/home/DangeR/server/bin/mangos-worldd -c ../etc/mangosd.conf'.
Program terminated with signal 11, Segmentation fault.
[New process 20272]
[New process 20276]
[New process 20275]
[New process 20274]
[New process 20273]
[New process 20177]
[New process 20180]
[New process 20179]
[New process 20178]
#0  Map::UnloadGrid (this=0x4137c10, x=@0x42777f98, y=@0x42777f9c, 
   pForce=<value optimized out>)
   at ../../../src/game/../framework/GameSystem/NGrid.h:39
39        void setUnloadReferenceLock( bool on ) { i_unloadReferenceLock = on; }
#0  Map::UnloadGrid (this=0x4137c10, x=@0x42777f98, y=@0x42777f9c, 
   pForce=<value optimized out>)
   at ../../../src/game/../framework/GameSystem/NGrid.h:39
   grid = <value optimized out>
   gx = 26
   gy = 31
   __PRETTY_FUNCTION__ = "bool Map::UnloadGrid(const uint32&, const uint32&, bool)"
#1  0x0000000000605bca in ~Map (this=0x4137c10)
   at ../../../src/game/Map.cpp:1096
No locals.
#2  0x000000000060601d in ~InstanceMap (this=0x4137c10)
   at ../../../src/game/Map.cpp:2367
No locals.
#3  0x000000000083527f in MapInstanced::DestroyInstance (this=0x411d940, 
   itr=@0x42778010) at ../../../src/game/MapInstanced.cpp:242
No locals.
#4  0x00000000008359b2 in MapInstanced::Update (this=0x411d940, t=@0x4277804c)
   at ../../../src/game/MapInstanced.cpp:58
   i = {<std::tr1::__detail::_Hashtable_iterator_base<std::Pair<const unsigned int, Map*>, false>> = {_M_cur_node = 0x40d6b40, 
   _M_cur_bucket = 0x40eb840}, <No data fields>}
#5  0x000000000061f54f in MapManager::Update (this=0x1dce590, 
   diff=<value optimized out>) at ../../../src/game/MapManager.cpp:265
No locals.
#6  0x00000000007ccf92 in World::Update (this=0x7fc8a4006d00, diff=48)
   at ../../../src/game/World.cpp:1405
No locals.
#7  0x00000000004e619c in WorldRunnable::run (this=<value optimized out>)
   at ../../../src/mangosd/WorldRunnable.cpp:60
   diff = 48
   realCurrTime = 44789191
   realPrevTime = <value optimized out>
   prevSleepTime = 48
#8  0x0000000000884f3c in ACE_Based::Thread::ThreadTask (param=0x411d940)
   at ../../../src/shared/Threading.cpp:183
   _task = (class ACE_Based::Runnable *) 0x7fc89adf41b0
#9  0x00007fc8ab5c1fc7 in start_thread () from /lib/libpthread.so.0
No symbol table info available.
#10 0x00007fc8aab9159d in clone () from /lib/libc.so.6
No symbol table info available.
#11 0x0000000000000000 in ?? ()
No symbol table info available.

[VladimirMangos]

And how this crash related to discussed problem? Please avoid report in thread where discussed specific problem your random problems...

[silvermcd]

And how this crash related to discussed problem? Please avoid report in thread where discussed specific problem your random problems...

now we must pay attention to the other cases, because they're anoying too, b4 waypoints, there was anyoing crashes already

[przemratajczak]

MaNGOS 9452 based on insider42 + mtmaps + external ACE 5.7.6

UDB 387

SD2 1605

http://pastebin.ca/1811636

http://pastebin.ca/1811644

[The_Game_Master]

Try reverting to 9453 and try again. I think it's related to 9454 or all the patches you added. You should use clean cores to test. I use a modifyed version myself, but not when testing.

Edit: Ok, i can confirm, bug still exists in latest revision.

[VladimirMangos]

now we must pay attention to the other cases, because they're anoying too, b4 waypoints, there was anyoing crashes already

Then pay attention in _another_ thread. Posting unrelated crashes to thread is just useless spam. I cleary will ignore like crashes as any spam and thread hijacking attempt

[VladimirMangos]

Edit: Ok, i can confirm, bug still exists in latest revision.

Posted crashes not fit my code line numbers... So as i understand it more seldom now but happens at same line?

[The_Game_Master]

Vladimir, any chance of the problem continuing here?

   if (i_destinationHolder.UpdateTraveller(traveller, time_diff, false))
   {
       i_destinationHolder.ResetUpdate(50);
       if(i_nextCheckTime.Passed() && i_destinationHolder.HasArrived())
       {
           _setTargetLocation(owner);
           return true;
       }
   }
   return true;

FleeingMovementGenerator.cpp

"So as i understand it more seldom now but happens at same line" looks like it. I'll repost when i'll have a full backtrace.

[fdb_]

The fix seems to work fine, but I still have some crashes with WaypointGen. Very few crashes I admit, one or two per day.

Here is a sample of the backtrace:

http://pastebin.mozilla.org/705126

Perhaps triggered inside an instance I dunno.

[The_Game_Master]

It's crashing every 5 to 25 minutes now. A little progress from 2min. Lol

[VladimirMangos]

I add some more checks in [9461] but but possible crashes have one more hidden source

[Quriq14]

I'll try again to explain my idea.

A bug is here (MotionMaster.cpp, line 40):

MotionMaster::Initialize()
{
   // clear ALL movement generators (including default)
   while(!empty())
   {
       MovementGenerator *curr = top();
       pop();
       curr->Finalize(*i_owner);
       if( !isStatic( curr ) )
           delete curr;
   }

There aren't checks about the flag MMCF_UPDATE, so the reset of MotionMaster is never delayed.

Instead of "delete curr;" it should be like:

if (m_cleanFlag & MMCF_UPDATE) // need delayed?
{
   if (!m_expList)
       m_expList = new ExpireList();

   m_expList->push_back(curr);
}
else
     delete curr;

[fdb_]

Still one crash related to WaypointMoveGen, using cherrypicked 9461

http://pastebin.mozilla.org/705308

[VladimirMangos]

I'll try again to explain my idea.

A bug is here (MotionMaster.cpp, line 40):

Thank you! Add in to [9479]. This is can possible source for not fixed part crashes. So will look at result

[fdb_]

All fixes related to the Waypoints and MovGens seem to reduce the number of crashes drastically. Good work guys!!!

Here is another dump ( http://pastebin.mozilla.org/705378 ).

I do not think it is linked to the Waypoint but there is a reference to TargetedMovementGeneratorMedium, so perhaps there is another missing check.

By the way, it seems that the following patch fix the problem:

diff --git a/src/game/CellImpl.h b/src/game/CellImpl.h
index 2c8048d..67c5ae5 100644
--- a/src/game/CellImpl.h
+++ b/src/game/CellImpl.h
@@ -182,6 +182,10 @@ Cell::Visit(const CellPair &standing_cell, TypeContainerVisitor<T, CONTAINER> &v
    if(radius > 333.0f)
        radius = 333.0f;

+    // fdb avoid null ptr being processed
+    if (&obj == NULL)
+        return;
+
    //lets calculate object coord offsets from cell borders.
    CellArea area = Cell::CalculateCellArea(obj, radius);
    //if radius fits inside standing cell

Vladimir, do you think it is a good way to handle this kind of crashes ?

[TommyAnders]

I still have crash after MangosZero rev.[z0230], if this isn't related with waypoints, then sorry. Problem appeared at boss Jandica Barov, when she summed adds Illusions

http://pastebin.org/98930

[The_Game_Master]

No, Tommy Anders, it's not mangos core related. It's caused by a script problem.

Thank you fdb.

[VladimirMangos]

if (&obj == NULL)

Ofc, no. Like crash only meaning that greadsearch function called with NULL somewhere (look in crashdump for search where. And in BUT chance this is some broken script.

[VladimirMangos]

In any case last reported crashes not related to problem. I will add like posts in independent from it usefulness or will just close thread...