This fixes the
dEQP-GLES3.functional.shaders.invariance.{low,medium,high}p.loop_4 tests when
run in a VM with virgl on a host with iris. virgl mangles the GLSL shaders and
emits shader code for the host driver that contains vec4 arrays. As such, the
test did not fail when running directly on the host.
The test also did not fail if the host was using i965. Disabling
PIPE_CAP_GLSL_OPTIMIZE_CONSERVATIVELY in iris was sufficient to work around it,
so I believe that i965 didn't show the problem because after arrays were split
by optimize_split_arrays(), even though the invariant/precise qualifiers were
lost, do_common_optimization() would be called again and thus
propagate_invariance() would propagate the qualifiers to the new variables
produced by optimize_split_arrays().
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10292>
While it produces functioning code the pass creates worse code
for arrays of arrays. See the comment added in this patch for more
detail.
V2: skip splitting of AoA of matrices too.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Memory/format layout qualifiers shouldn't be lost when arrays
of images are splitted by this pass.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Previously, we failed to split constant arrays. Code such as
int[2] numbers = int[](1, 2);
would generates a whole-array assignment:
(assign () (var_ref numbers)
(constant (array int 4) (constant int 1) (constant int 2)))
opt_array_splitting generally tried to visit ir_dereference_array nodes,
and avoid recursing into the inner ir_dereference_variable. So if it
ever saw a ir_dereference_variable, it assumed this was a whole-array
read and bailed. However, in the above case, there's no array deref,
and we can totally handle it - we just have to "unroll" the assignment,
creating assignments for each element.
This was mitigated by the fact that we constant propagate whole arrays,
so a dereference of a single component would usually get the desired
single value anyway. However, I plan to stop doing that shortly;
early experiments with disabling constant propagation of arrays
revealed this shortcoming.
This patch causes some arrays in Gl32GSCloth's geometry shaders to be
split, which allows other optimizations to eliminate unused GS inputs.
The VS then doesn't have to write them, which eliminates the entire VS
(5 -> 2 instructions). It still renders correctly.
No other change in shader-db.
v2: Drop !AOA check and improve a comment (feedback from Tim Arceri).
Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
When we find indirect indexing into an array, the current implementation
of the array spliiting optimization pass does not look further into the
expression tree. However, if the variable expression involves variable
indexing into other arrays, we can miss that these other arrays also have
variable indexing. If that happens, the pass will crash later on after
hitting an assertion put there to ensure that split arrays are in fact
always indexed via constants:
shader_runner: opt_array_splitting.cpp:296:
void ir_array_splitting_visitor::split_deref(ir_dereference**): Assertion `constant' failed.
This patch fixes the problem by letting the pass step into the variable
index expression to identify these cases properly.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89607
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
