CODEWITHSUNDEEP
openglopengl-3
user815129
user815129

Reputation: 2314

glBindVertexArray random INVALID_OPERATION

I am facing a strange crash that randomly appears when trying to bind a VAO.

the VAO itself has previously been generated by glGenVertexArrays, which returns no error.

glGenVertexArrays(NUM_VAO,vao);
glGenBuffers(NUM_VAO,vbo);

glBindVertexArray(vao[2]);
glBindBuffer(GL_ARRAY_BUFFER,vbo[2]);
glBufferData(GL_ARRAY_BUFFER,1024*sizeof(GLfloat),NULL,GL_DYNAMIC_DRAW);
glVertexAttribPointer(...);
glEnableVertexAttribArray(...);

during execution, a call to:

glBindVertexArray(vao[2])

may or may not generate an INVALID_OPERATION. Note that the content of vao[2] is the same returned by glGenVertexArrays and the array is never deleted during execution.

According to the specs, an INVALID_OPERATION is issued only when a non-zero name is passed and the name itself was not returned by glGenVertexArrays, which is clearly not the case.

The problem occurs sporadically, approximately only 15% of the times the application is run, otherwise everything works as expected. The problem persists in debug mode.

I just upgraded my workstation, I used to develop on OpenSUSE 12.1 with NVidia drivers 290.10 and Qt 4.7.1 and never had this problem.

I just moved to OpenSUSE 12.2 with NVidia drivers 304.43 and Qt 4.8.1 and I have this issue.

Have someone encountered such a situation before? Am I missing something?

Upvotes: 1

Views: 1862

Answers (1)

user815129
user815129

Reputation: 2314

Ok ok my fault, there was a glDeleteVertexArrays() gone wild.. but it was really hard to spot.

Supposedly on the old OS and drivers the names returned, initial memory layout or whatsoever was different and it just deleted an inexistent VAO, while here it points to a valid one..

Changing linux version/distro or going BSD really helped me a lot in writing truly portable code..

Upvotes: 1

Related Questions