We were mapping the renderbuffer once, then walking over all the
buffers to map just the texture ones using the other texture mapping
function that handled the x/y offset to the image in the region. But
then we would go and overwrite *those* mappings with the original
mappings for depth/stencil, which was wrong.
Instead, just walk over the attachments once and map the attachments.
Wasn't that easy?
This is already pointing at 0 or Height - 1 and with an appropriate
pitch, so no need to recompute those values per customization of the
spans code. Cuts 3 out of 21kb of the compiled size.
Reviewed-by: Chad Versace <chad@chad-versace.us>
Since the stencil buffer is interleaved, the generic Mesa renderbuffer
accessors do not suffice. Custom span functions are necessary.
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Chad Versace <chad@chad-versace.us>
For 1 and 2-channel formats the hardware only supports rendering to R
and RG. To do I and L render targets we just call them R and
everything works out. For A, we would need to rewrite the CC to do
the alpha channel's blending on color instead, and send the fragment
alpha down the red channel. For LA, there doesn't seem to be any
hope, because we can't do independent color/alpha blending while
treating the LA surface as RG.
Reviewed-by: Brian Paul <brianp@vmware.com>
Since we're using GTT mappings now (no manual detiling), there's
really nothing special to accessing these buffers, other than needing
the new RowStride field of gl_renderbuffer to accomodate padding.
Reduces the driver size by 2.7kb, and improves glean depthStencil
performance 3-10x (!)
Reviewed-by: Brian Paul <brianp@vmware.com>
Shaves 60k off the driver from removing the broken spans code. This
means we now require 2.6.29, which seems fair given that it's a year
old and we've removed support for non-KMS already in the last release
of 2D.
This uses a stamp mechanisms to mark the DRI drawable as invalid.
Instead of immediately updating the buffers we just bump the drawable
stamp and call out to DRI2GetBuffers "later".
"Later" used to be at LOCK_HARDWARE time, and this patch brings back
callouts at the points where we used to call LOCK_HARDWARE. A new function,
intel_prepare_render(), is called where we used to call LOCK_HARDWARE,
and if the buffers are invalid, we call out to DRI2GetBuffers there.
This lets us invalidate buffers only when notified instead of on
every glViewport() call. If the loader calls the DRI invalidate
entrypoint, we disable viewport triggered buffer invalidation.
Additionally, we can clean up the old viewport mechanism a bit,
since we can just invalidate the buffers and not worry about
reentrancy and whatnot.
Before, if we just called glXMakeCurrent() and didn't render anything we'd
still trigger a flushFrontBuffer() call.
Now only set the intel->front_buffer_dirty field at state validation time
just before we draw something.
NOTE: additional calls to intel_check_front_buffer_rendering() might be
needed if I missed some rendering paths.
This improves piglit quick.tests runtime from 19:33 minutes to 6:06 on
my GM45. It should also hide most of the A17 swizzling issues, though
they'll still exist when swapping occurs (which is the kernel's problem
either way).
Pass a gl_format to intel_create_renderbuffer() instead of GLenum.
Add cases for MESA_FORMAT_XRGB8888 textures and renderbuffers.
However, we don't yet create any renderbuffers or textures with that
format. It seems the default alpha value is zero instead of one.
Need to investigate that first.
_ActualFormat is replaced by Format (MESA_FORMAT_x).
ColorEncoding, ComponentType, RedBits, GreenBits, BlueBits, etc. are
all replaced by MESA_FORMAT_x queries.
Since the TNL is all done in software anyway, it should be the same to
the user who's probably using ARB_vertex_program otherwise, but gives them
a nicer programming environment.
Now gl_texture_image::TexFormat is a simple MESA_FORMAT_x enum.
ctx->Driver.ChooseTexture format also returns a MESA_FORMAT_x.
gl_texture_format will go away next.
Also enable them all regardless of screen bpp, as 32 bpp what I've been
testing against, and haven't been able to detect any screen bpp-specific
troubles with them.
It seems that in this case the Mesa code is handing us x8z24 values instead
of z24s8 values, so we need to not do the rotation. Fixes half of OGLconform
depthrange.c.
Bug #19447.
This is insanity, but so is copying the same blocks containing the actual
interesting code in the file three times each for the different tile formats.
