No longer used.

The memory / immediate source should already be in the only valid
position.

The CLIP_RECTs always affect dedicated clears, and it's nicer than
having to mark the viewport or scissor state dirty after it.

Early-Z pass raises the occlusion counter.

We may want to put the converted vertex buffer in persistent
storage instead, but these are rare corner cases.

Fixes piglit/fbo-srgb.

pointed out by Marek on irc.

This reverts commit 6e7d782d.

Oops, I just had this locally for testing and forgot to remove it before pushing.

this fixes the piglit mipmap generation sRGB on my rv730.
Signed-off-by: Dave Airlie <airlied@redhat.com>

This is a workaround for a bug in libtxc_dxtn.

Fixes:
- piglit/GL_EXT_texture_compression_s3tc/fbo-generatemipmap-formats
Signed-off-by: Dave Airlie <airlied@redhat.com>

this fixes piglit fbo-generatemipmap-formats on my rv730.
Signed-off-by: Dave Airlie <airlied@redhat.com>

we need to translate the destination box as well.

fixes piglit's s3tc-texsubimage test.
Signed-off-by: Dave Airlie <airlied@redhat.com>

Arrays are zero based.  If the highest element accessed is 6, the
array needs to have 7 elements.

Fixes piglit test glsl-fs-implicit-array-size-03 and bugzilla #34198.

NOTE: This is a candidate for the 7.9 and 7.10 branches.

This is a follow-up to commit eafb7f23.

Fixes Linux SCons build.

Fixes regression: https://bugs.freedesktop.org/show_bug.cgi?id=34160

Commit e7c1f058 disabled constant-folding
when division-by-zero occured. This was a mistake, because the spec does
allow division by zero. (From section 5.9 of the GLSL 1.20 spec: Dividing
by zero does not cause an exception but does result in an unspecified
value.)

For floating-point division, the original pre-e7c1f058 behavior is
reinstated.

For integer division, constant-fold 1/0 to 0.

This reverts commit b3cf92aa.

The reverted commit prevented constant-folding of reciprocal expressions
when the reciprocated expression was 0. However, since the spec allows
division by zero, constant-folding *is* permissible in this case.

From Section 5.9 of the GLSL 1.20 spec:
    Dividing by zero does not cause an exception but does result in an
    unspecified value.

Before populating the vertex buffer attribute pointer (VB->AttribPtr[]),
convert vertex data in GL_FIXED format to GL_FLOAT.

Fixes bug: http://bugs.freedesktop.org/show_bug.cgi?id=34047

NOTE: This is a candidate for the 7.9 and 7.10 branches.

And set a reasonable buffer usage flag everywhere instead of
just PIPE_USAGE_DEFAULT.

I typod this when copy-pasting.

Fixes this GCC warning.
nvc0_tgsi_to_nc.c: In function 'bld_tex':
nvc0_tgsi_to_nc.c:1392: warning: 'dim' may be used uninitialized in this function

This is a multi-threading optimization which hides the kernel overhead
behind a thread. It improves performance in CPU-limited apps by 2-15%.
Of course you must have at least 2 cores for it to make any difference.

It can be disabled with:

export RADEON_THREAD=0

On r600, s3tc formats require a 1D tiled texture format,
so we have to do uploads using a blit, via the 64-bit and 128-bit formats

Based on the r600c code we use a 64 and 128-bit type to do the
blits.

Still requires R600_ENABLE_S3TC until the kernel fixes are in,
this has only been tested on evergreen where the kernel doesn't
yet get in the way.

the miptree setup and pitch storing didn't work so well for block
based things like compressed textures. The CB takes blocks, where
the texture sampler takes pixels, and transfers need bytes,

So now we store blocks/bytes and translate to pixels in the sampler.

This is necessary for s3tc to work properly.