In particular, with foreach_list_safe, one can remove and free the current
node without crashes; if new nodes are added after the current node,
they will be properly visited as well.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
The following instruction sequence will no longer be emitted in separate
TEX blocks:
0: TEX temp[0].xyz, temp[1].xy__, 2D[0];
1: TEX temp[1].xyz, temp[2].xy__, 2D[0];
This fixes fdo bug #25109
The code to emit an array of OpenGL state vars lacked the code
to handle the gl_TextureMatrix[] array.
Fixes fd.o bug 28967
NOTE: this is a candidate for the 7.8 branch.
Most places in the code simply use a static name, which works because
names are never used to look up an ir_variable. generate_temporary is
simply unnecessary (and looks like it would leak memory, and isn't
thread safe...)
When EU executes 'wait' instruction, it stalls and sets notification
register state. Host can issue MMIO write to clear notification
register state to allow EU continue on executing again.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Fixes fd.o bug 27216. May also be the root cause of fd.o bug 28950.
We weren't propogating the storage info for the x=foo() expression up
through the IR tree to the inequality expression.
NOTE: This is a candidate for the Mesa 7.8 branch.
Previously, a register defined at main scope and used in a loop in a
loop could end up getting marked as needed only from the definition
outside of the loops to the end of the inner loop, and we would
cleverly slot in something else in its register in the end of the
outer loop.
Fixes glsl-vs-loop-nested and glsl-fs-loop-nested on glsl2. This
doesn't happen much on master because the original compiler does its
own register allocation, so we find little we can do with linear scan
register (re)allocation.
First, this undoes commit e503af4baa
so we use iround() in lp_build_nearest_mip_level().
Second, in lp_build_sample_general() we need to check if we're sampling
a cube map before anything else. Choose the cube face and then recompute
the partial derivatives of (S,T) with respect to the chosen cube face.
Before, we were using the directional (S,T,R) derivatives to compute
the LOD.
Third, work around an apparent bug in LLVM 2.7 where setting the lod
variable to a const(0) value results in bad x86 code. See comments in
the code.
Uses of the bits for allocation are offset by 16, and
VERT_BIT_GENERIC0 already has the 16 offset. As a result, it was
preventing the wrong thing from being allocated.
rcp of an integer value did not produce the result you're looking for.
Instead, do the a * rcp(b) as float and truncate after. This mostly
fixes glsl-fs-loop-nested.
