opengl binding of target in thread fails - opengl

Question

I am trying to update the content of a vertex buffer by a thread.
Inside the thread function it is not possible to bind the buffer
to a target, or retrieve a pointer via glMapBufferRange.
I always get the error 1282. Outside thge thread function everything
works fine.
while( t_con->out_loop )
{
Sleep( t_con->sleep_time );
if( t_con->in_loop )
{
t_con->count++; // thread counter up
EnterCriticalSection( &__cr_sec );
/*******************************************/
/* create the strings for the variables */
/*******************************************/
str_run = sfs_p->str_ptr;
sprintf( str_run, "%4.0f / %4.0f", __mou.pos.x, __mou.pos.y ); // actual mouse position
str_run += sfs_p->str_size;
sprintf( str_run, "%4.0f / %4.0f", __mou.old_pos.x, __mou.old_pos.y ); // old mouse position
str_run += sfs_p->str_size;
sprintf( str_run, "%4.0f / %4.0f", __mou.travel.x, __mou.travel.y ); // mouse travel delta
str_run += sfs_p->str_size;
sprintf( str_run, "%d", __mou.but_l ); // mouse button left
str_run += sfs_p->str_size;
sprintf( str_run, "%d", __mou.but_m ); // mouse button center
str_run += sfs_p->str_size;
sprintf( str_run, "%d", __mou.but_r ); // mouse buttom right
str_run += sfs_p->str_size;
sprintf( str_run, "%f", __mou.speed ); // mouse speed
/******************************************************************/
/* create a Vertex_Buffer_Struct with the chars geometry data */
/******************************************************************/
vx = SFS_to_Vx_Buf( sfs_p );
glBindBuffer( GL_ARRAY_BUFFER, m_dbg->vao->vx_b->id );
G_E_C;
// bind the buffer
gl_ptr = glMapBufferRange( GL_ARRAY_BUFFER, 0, 300, GL_MAP_WRITE_BIT | GL_MAP_UNSYNCHRONIZED_BIT ); // get the pointer
G_E_C;
Destroy_Vertex_Buffer_Struct( vx );
LeaveCriticalSection( &__cr_sec );
}
}
Any suggestions?

Answer 1

In addition to #Ripi2's answe it must be noted that a GL context can be current to at most on thread at any time. One has actually to release the context on the main thread before any worker thread can make it current. As a consequence, using a single OpenGL context in multiple threads is neither convenient nor efficient in most cases.
I am trying to update the content of a vertex buffer by a thread. Inside the thread function it is not possible to bind the buffer to a target, or retrieve a pointer via glMapBufferRange.
There are much better options for that, notably:
Just keep all GL calls in a single thread, and do the glMapBufferRange in the GL thread. While the buffer is mapped, it is writeable by the whole process, so any other thread can fill it.
Use multiple GL contexts. Each thread can have it's own context and keep it current all the time. By using shared context, the actual data objects like textures, buffers are shared between the context, so you can access the data the GL context in the other thread will be using.
In any case, you will need some proper means of synchronization between those threads, and OpenGL Sync Objects might be helpful there.

Answer 2

Before any gl-command is executed the gl-context must be set as current to the thread that uses those gl-commands.
Just call the SetCurrent() command or similar in your thread.
And remember that the main thread (where the app starts) is also a thread. Thus, it also needs SetCurrent() if it uses any gl-command.