Use _mesa_reference_vert/fragprog() wherever we assign program pointers.
Fixes a memory corruption bug found with glean/api2 test.
Another memory bug involving shaders yet to be fixed...
Picked from gallium-0.1
The bitmap cache attempts to accumulate a series of glBitmap calls in a
buffer to effectively render a whole bunch of bitmaps at once.
The cache can be disabled, if needed, by setting UseBitmapCache=GL_FALSE.
Use the code in cso_context.c rather than st_cache.c.
Basically, binding of state objects now goes through the CSO module.
But Vertex/fragment shaders go through pipe->bind_fs/vs_state() since they're
not cached by the CSO module at this time.
Also, update softpipe driver to handle NULL state objects in various places.
This happens during context destruction. May need to update other drivers...
Update the Makefiles and includes for the new paths.
Note that there hasn't been no separation of the Makefiles yet, and make is
jumping all over the place. That will be taken care shortly. But for now, make
should work. It was tested with linux and linux-dri. Linux-cell and linux-llvm
might require some minor tweaks.
Provide an actual definition of the pipe_buffer struct, containing
the parameters used to create the buffer, and its refcount.
Shift refcounting buffers out of the winsys interface, similar to
surfaces & textures.
Rework pipebuffer/ to reflect the fact these changes, and also Michel's
reworking of the buffer interface.
If an app never called glViewport, the viewport size was always 0 by 0 pixels.
Now pass initial size to st_create_framebuffer() and initialize the viewport
and scissor bounds in st_make_current().
This could also be fixed by ensuring the gl_framebuffers passed to
_mesa_make_current() were initialized to the right size. But that involves
allocating the renderbuffers/pipe_surfaces earlier and that runs into some
other issues ATM.
Also remove obsolete createRenderbuffers param to st_create_framebuffer().
The state tracker calls pipe->get_param() to determine the GL limits and
which OpenGL extensions are supported.
This is an initial implementation that'll probably change...
For some drivers (like Xlib) it's not possible to treat the front/back color
buffers as pipe_regions. So pipe->region_copy() won't work. Added a new
state tracker field indicating if we can use regions for colorbuffer accesses.
This should probably be re-considered someday...
Replacing mesa's main hash with one that handles collisions, moving state_tracker
related caching to the state tracker to keep cso cache independent of it. Cleanups.
We want our state objects to be immutable, handled via the
create/bind/delete calls instead of struct propagation.
Only implementing the blend state to see how it would look like
and work.
Not quite finished yet.
Selection/feedback are done with a private instance of the 'draw' module
in the state tracker. Not quite all the draw context's state is set yet,
namely vertex format info. Hold off on that for a bit...
