s_CurMgr fall down, go <poof>

[Sillyboy72]

So... let's just ignore that fact that I don't just steal wow's ObjectFromGuid hotness...

Here is mine:

WowObject *WowObjectFromGuid(ULONGLONG Guid)
{
	WowObject *pCurrent = *(WowObject**) (Mgr + 0xAC);	
	
	while ( pCurrent && ((DWORD)pCurrent & 1) == 0 )
	{
		if (pCurrent->guid == Guid)
		{
			return pCurrent;
		}

		pCurrent = pCurrent->Next;
	}

	return NULL;
}

But here's the thing... while I call this a zillion million jillion times, just fine... every once in awhile it goes coo-coo for coco puffs.

btw, I call this via hooked EndScene. Looking at other running threads, I certainly do see anything that should be stomping my guts.

Yet, what I call "Mgr" sometimes points to la-la land.

Static address 0x11CA310 points to s_CurMgr.

Consulting windbg during a crash...
0:000> dd 0x11CA310 L1
011ca310 0ae7a578

0:000> dd 0ae7a578+0x28A4 L1
0ae7ce1c 00000000

Uhhh... wha? Tis null? wtf?

This was literally while flying around in the game.

Theories?

[Cypher]

Null pointer checks are out of the question or something?

[kynox]

Static address 0x11CA310 points to s_CurMgr.

Consulting windbg during a crash...
0:000> dd 0x11CA310 L1
011ca310 0ae7a578

0:000> dd 0ae7a578+0x28A4 L1
0ae7ce1c 00000000

points to s_curConnection*.

Obviously something major is happening if the structure is being recreated. Possibly zoning/changing battlegrounds?

[Sillyboy72]

Null pointer checks are out of the question or something?

There is only a pointer to check for null, if you are attempting to grab the current manager ... over and over. I was holding onto it, assuming it was stayin' valid for the entire session. Apparently a bad assumption, which ... I find weird.

BYTE *GetMgr()
{
	BYTE **ppClientConnection = (BYTE**) 0x11CA310;
	BYTE *pClientConnection = *ppClientConnection;
	BYTE **ppMgr = (BYTE**) &pClientConnection[0x28A4];
	return *ppMgr;
}

Apparently I would be better off holding (what I was calling) ppMgr instead, and checking for null before a deref.

Curious if others have seen similiar. I see others use s_CurMgr, implying they only grab it once as well...

Hrm, might be time to review some code and seeing if I am overflowing a buffer, cuz I am crazy skeptical this pointer suddenly goes to null...

[kynox]

Ugly:

BYTE *GetMgr()
{
    BYTE **ppClientConnection = (BYTE**) 0x11CA310;
    BYTE *pClientConnection = *ppClientConnection;
    BYTE **ppMgr = (BYTE**) &pClientConnection[0x28A4];
    return *ppMgr;
}

Clean:

DWORD* dwGetMgr()
{
    DWORD dwClientConnection = *(DWORD*)0x11CA310;
    if ( pClientConnection )
        return *(DWORD*)( dwClientConnection + 0x0x28A4 );
    return NULL;
}

[Cypher]

Ugly:

BYTE *GetMgr()
{
    BYTE **ppClientConnection = (BYTE**) 0x11CA310;
    BYTE *pClientConnection = *ppClientConnection;
    BYTE **ppMgr = (BYTE**) &pClientConnection[0x28A4];
    return *ppMgr;
}

Clean:

DWORD* dwGetMgr()
{
    DWORD dwClientConnection = *(DWORD*)0x11CA310;
    if ( pClientConnection )
        return *(DWORD*)( dwClientConnection + 0x0x28A4 );
    return NULL;
}

You fail. Did you even read that "clean" code before you posted it? >_>

[Sillyboy72]

Ugly:

LOL. No doubt. Isn't that shweet?

I think I know what's going on here. I think I am getting DC'd... as a result of jamming results from a atan2() call directly into my "facing". I need to wow-ize my angle (0 to 2pi) since atan2 returns -pi to pi.

So, client connection goes bye-bye when I DC, and I crash. While I could watch for the null, seems even more useful to avoid the DC

Still, your clean codez is heading to the pasta factory :wave:

edit: LOL, man in green 1, kynox 0. I really thought there were only 2 issues with the code. More like 4

[kynox]

LOL. No doubt. Isn't that shweet?

I think I know what's going on here. I think I am getting DC'd... as a result of jamming results from a atan2() call directly into my "facing". I need to wow-ize my angle (0 to 2pi) since atan2 returns -pi to pi.

So, client connection goes bye-bye when I DC, and I crash. While I could watch for the null, seems even more useful to avoid the DC

Still, your clean codez is heading to the pasta factory :wave:

edit: LOL, man in green 1, kynox 0. I really thought there were only 2 issues with the code. More like 4

Thats what i get for writing code in a small forum box after 2 hours of wiping to 3 drakes and extremely tired.

[Cursed]

Thats what i get for writing code in a small forum box after 2 hours of wiping to 3 drakes and extremely tired.

Dont be sad kynox Take this +Rep for all the contributions you have done

Edit: I hate to tell you, I have to spread first-.- But you can remind me!