Set type to r-value resulting from indexing vector, to prevent
float->uint conversion when source is already uint. Resulting
OpConvertFToU would otherwise fail validation because source is
already uint.
For LoadN, incorrect uint->float->uint can be avoided; fixing
potential truncation of big integer values.
Also, only emit this XFB information where the SPIR-V spec says
it should be emitted: essentially, on objects.
This and the previous commit together fix#1185.
Some stage (e.g, hull shaders) have arrayed builtin outputs (e.g, position).
When copying from the internal structure to the split form, it is necessary
to propagate that indirection to the actual arrayed outputs. This was not
happening.
Addresses #1181
This makes struct returns from functions work, but breaks
structs containing arrays, due to limitations in subsequent
transforms in spirv-opt. This is expected to be fixed soon.
Adds command line options:
--invert-y
--iy
(synonyms) which invert position.Y on vertex shader output. Handles these cases:
* Direct single variable return
* Member of direct returned struct
* Single variable output parameter
* Member of struct output parameter
API:
// Enables position.Y output negation in vertex shader
void TShader::setInvertY(bool invert);
Fixes#1173
This continues to prevent writing output buffers (out from a function),
but fixes the problem where the copy-in/out was not getting done.
Making everything work will require knowing both in/out-ness and bufferness,
but these are currently mutually exclusive, because both are storage
qualifiers.
Issue #791 was partially fixed by PR #1161 (the mat mul implicit
truncations were its main point), but it still wouldn't compile due to
the use of ConstantBuffer as an identifier. Apparently those fall into
the same class as "float float", where float is both a type and an
identifier.
This allows struct definitions with such keyword-identifiers,
and adds ConstantBuffer to the set. 'cbuffer int' is legal in HLSL,
and 'struct int' appears to only be rejected due to the redefinition
of the 'int' type.
Fixes#791
HLSL truncates the vector, or one of the two matrix dimensions if there is a
dimensional mismatch in m*v, v*m, or m*m.
This PR adds that ability. Conversion constructors are added as required.
If a shader includes a mixture of several stages, such as HS and GS,
the non-stage output geometry should be ignored, lest it conflict
with the stage output.
Per feedback on PR #1111, this reverses the order of the parameters for the setShiftBinding API.
It is now:
void TShader::setShiftBindingForSet(TResourceType res, unsigned int base, unsigned int set);
This script will crank a supplied set of glslang test shaders through the
spirv-val tool, reporting on the results.
There are some important things to note:
* Like 'runtests', this must be invoked from the 'Test' subdirectory.
* This is mostly useful on the hlsl.* tests, although it is not strictly
limited to those. The reason is that most of the glsl tests either contain
validation error cases, and so fail to compile, or are not using a #version
compatible with producing SPIR-V modules.
* Some tests, such as negative tests, or most of the glsl tests, have
intentional compilation errors. This script treats that as OK. Failures
are successfully compiling shaders which proceed to fail spirv-val.
* spirv-val is looked for in either the External directory, or if not
found there, in a sibling directory of glslang, and if not found there
either, in /usr/local/bin.
* There are a bunch of command line options. ./validate-shaders.sh --help
will describe them.
Some examples to try:
./validate-shaders.sh hlsl.* # exercise all hlsl.* tests.
./validate-shaders.sh --terse hlsl.* # same, but tersely.
# dump validator results for problems in something.frag:
./validate-shaders.sh --quiet --dump-val something.frag
