Bug-NUM_FOR_EDICT Error: Bad Pointer

[Asurmen]

Sup all.

Thought I would start a new thread on this bug since I have now gotten this error typicaly on map change. Have not really pin pointed which maps are doing it yet.

Please if anyone else is getting this error and has some code to post to fix this that would be great. Again, only post code on this bug on this thread. Thanx

Lastly, could anyone who posts code post it in a way so anyone can understand it at to where the fix needs to be changed. Thanx to all that contribute.

Until next time..Cheers!

[sPlOrYgOn]

Asurmen, yes I know of this bug but.. I have no idea what that error means or what could be causing it because I have never experienced it.
Could you give some steps to reproduce it somehow?
make sure you can reproduce it a few times.
I'll be checking in the code where the server changes maps.
and what command did you use to change the map? "changelevel"/"map"
and also post some information about your setup.
and does this only happen in 1.6?

[Asurmen]

Spiorygon,

My setup is:

Steam CS 1.6 Listen Server
Metamod 1.17X3 from AMXX site-which seems to not matter which one I use wether it is the Filebase on this site or the above.
Latest Podbot MM here at Bots-United
AMXX 0.16

I do not use AMX anymore since the last update from Valve basicly made it un-usable.

I have done quite of testing with several bots in the past. I just haven't done much debugging. Is there any commands that I can add to my shorcut and or listenserver.cfg to get the information that would shed light on this? Thanx.

Until next time...Cheers!

I get this bug now as well on my dedicated mEAn PODbot server. The maps I experience this on are de_nuke , and as_oilrig. In the past I had this bug when I first started working on getting POD up and running on a 1.6 listen server. The error at that time would occur the instant the bomb was to drop out of the hand to the ground. I fixed this problem by reinstalling metamod, podbot, and amx. Anyway I haven't spent alot of time figuring out why its happening now, I know it doesn't happen on the map change. There is deffinately an action that triggers this kind of error. Wether its a sound that gets asked for or what I'm not sure.

[sPlOrYgOn]

Well the way I'm interpreting this error is...
There was a BAD Pointer (pointed to some wrong memory address) that was supposed to be pointed at an edict (entity).

[Lazy]

I think what your looking for can be found in the quake 1 source...

 int NUM_FOR_EDICT(edict_t *e)
 {
 	int		b;
 	
 	b = (byte *)e - (byte *)sv.edicts;
 	b = b / pr_edict_size;
 	
 	if (b < 0 || b >= sv.num_edicts)
 		Sys_Error ("NUM_FOR_EDICT: bad pointer");
 	return b;
 }

Basically make sure any calls to ENTINDEX are within range.

[sPlOrYgOn]

Well I was in the middle of a massive cs_office bot pwnage and it poped up
Well now I know it happens in 1.5 also.
I still don't know what is causing it because mine wasn't during changelevel but it did change round using amx's voting.

[>BKA< T Wrecks]

A friend of mine is running CS 1.6 on a Linux DS (just MM + PB), and he gets the error, too. Also on cs_office, among other maps I don't remember right now. So it really seems to be on all platforms, CS versions and OS'es...

[Pierre-Marie Baty]

Perhaps it happens only on a certain type of maps ?
for example cs_ maps ?

That would be a problem related to hostages...

[KWo]

Asurmen, try to delete all *.pxp and all *.pvi files from your WPTDEFAULT directory. After this test once again and describe us step by step, what did You do to see crash?
For example
1. I start to play on map de_dust.
2. Timelimit for map has reached.
3. New map cs_italy has began to load.
4. I saw the crash with this NUM_FOR_EDICT.

Write us , what is a date stamp of pwf file corresponding to map at which CS crashed. For example if You saw the crash at cs_office, write us date and time of creating the file cs_office.pwf in WPTDEFAULT directory.