Jump to content
  • Advertisement
Sign in to follow this  
Amazing Retardo

OpenGL Really strange extension problem

This topic is 3773 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi Everyone! Just posting about a really weird problem I'm facing with opengl extensions, particularly deleting extension objects (like shaders, vbos, fbos.. etc) Setting up the extensions and using them works fine, the problem happens when I try to delete these objects. (I'm using ARB extensions, but I've tested with the OpenGL 2.0 alternatives to no avail)
HFGLShader::~HFGLShader()
{
	for (list<HFVirtualShader*>::iterator i = mShaderList.begin(); i != mShaderList.end(); i ++)
	{
		HFBaseShader * shader = (HFBaseShader*)(*i);
		glDetachObjectARB(mShaderProgId, shader->GetShader());
		shader->UnloadAssets();
		delete shader;
	}
	if (mShaderProgId)
		glDeleteObjectARB(mShaderProgId);
	mShaderProgId = 0;
}



The program crashes at glDetachObjectARB() with: "Unhandled exception at 0x69638b16 in HFTestingPad.exe: 0xC0000005: Access violation reading location 0x00000864.". This happens with other related extension deletion code as well (e.g. glDeleteObjectARB(), glDeleteBuffersARB()) I've tried implementing extensions using glew first, and then glee later to see if it was my extension library that was causing it, but sadly, it wasn't. (I even obtained the address myself using wglGetProcAddress()). Lastly, this problem only seems to happen on computers using nvidia cards. (Testing on a quadrofx and a 8600 GT caused the crash) ATI cards do not seem to have a problem with the extension. (Tested using X1400 Mobility and 2900 HD XT) Any help solving this strange bug is greatly appreciated! Thanks! A. Retardo

Share this post


Link to post
Share on other sites
Advertisement
Hmmm strange indeed that it works on some cards but not others!

My first thought is that perhaps there's some kind of memory corruption problem in your code that is causing you to pass garbage into glDetachObjectARB. Perhaps ATI's driver is more robust (for once) and silently ignores the garbage, while NVidia's driver trusts you more and tries to use the garbage but crashes.


What is this cast for:
HFBaseShader * shader = (HFBaseShader*)(*i);

Is HFVirtualShader a sub-class of HFBaseShader?

If so, then the cast is not needed, and if not... bad stuff... Anyway, if you do need to cast you should be using a C++ one, not a C one ;)

Share this post


Link to post
Share on other sites
Yup HFVirtualShader is a sub-class, however to access the UnloadAssets() function it has to be type-cast to the HFBaseShader (a little hack I did, since my alternative (Cg) doesn't require the UnloadAssets() function and is deleted in a different way, in it's own inherited class. It's bad program design, but it shouldn't be the cause of the problem?)

I'm quite sure its not a memory corruption problem, because I placed a break point at glDetachObjectARB and the first time it calls (before any other actual deletion is done) it crashes. (p.s. I'm not calling glDetachObjectARB anywhere else in the code).

I know I should be using the C++ cast.. but I got lazy :P (will probably implement it later once I solve this bug though.)

Share this post


Link to post
Share on other sites
Quote:
Original post by Amazing Retardo
...
I'm quite sure its not a memory corruption problem, because I placed a break point at glDetachObjectARB and the first time it calls (before any other actual deletion is done) it crashes. (p.s. I'm not calling glDetachObjectARB anywhere else in the code).
...


Your interpretation about memory corruption is incorrect. When you have memory corruption, the behavior will become unpredictable. It may or may not cause a access violation depending on the content of the memory. However, when your program end up with a access violation, it is almost saying your program have memory corruption.

Share this post


Link to post
Share on other sites
Whoops, I'll admit that I'm not an expert in C++. Alright, so what is the likely cause of memory corruption then? and why does it only seem to be affecting deletion of extension objects?

Share this post


Link to post
Share on other sites
The location of the memory corruption can be anywhere of your source code. Although it causing a access violation during deleting a extension, it doesn't imply the memory corruption is located in the routine about the extension. For memory corruption, all outcomes are possible anyway.

I suggest you debug your application as a whole, especially for the memory allocation routine and the destructor of all your C++ class. Although this is going to be a very difficult debugging process, it doesn't sound all that bad actually. At least, you knew your program has a bug as it cause a access violation occasionally. The situation can be even worst if the bug just causing the result incorrect occasionally. In this case, you will not alerted of the bug until it cause a problem to your customer, painful isn't it?

Share this post


Link to post
Share on other sites
It's strange though, how it only breaks at that line on computers with nvidia cards. Strangely enough, on all the computers I tested on it crashed on the exact same line only on nvidia cards. Ah well, could be like what Hodgman described it. At any rate, I'll start debugging the code the moment I get access to my workstation.

Thanks!

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

Participate in the game development conversation and more when you create an account on GameDev.net!

Sign me up!