Commit Graph

511 Commits

Author SHA1 Message Date
Hans-Kristian Arntzen
480acdad18 Deal with OpSpecConstantOp used as array size.
When trying to validate buffer sizes, we usually need to bail out when
using SpecConstantOps, but for some very specific cases where we allow
unsized arrays currently, we can safely allow "unknown" sized arrays as
well.

This is probably the best we can do, when we have even more difficult
cases than this, we throw a more sensible error message.
2018-11-01 14:58:02 +01:00
Hans-Kristian Arntzen
d4ad1e266a Add test case for local_size_x_id and some complex cases. 2018-11-01 11:23:48 +01:00
Hans-Kristian Arntzen
62db535b3f Update tests. 2018-11-01 11:23:48 +01:00
Grigory Dzhavadyan
a5d82d1138 Alter the handling of spec consts in non-Vulkan GLSL
Previously, when generating non-Vulkan GLSL, each use of a spec constant
would be subsituted for its default value and the declaration of the constant
itself would be omitted completely.

This change slightly alters this behavior. The uses of the constant are kept,
as well as the declaration, although the latter is stripped of the layout
qualifier. The declaration is also prepended with the following code:

    #ifndef <constant name>_value
    #define <constant name> <default constant value>
    #endif

and the constant itself now looks like

    const <constant type> <constant name> = <constant name>_value;

The rationale for this change is that it gives the user a way to provide
custom values for specialization constants even when the target does not
support them.
2018-11-01 00:39:09 -07:00
Hans-Kristian Arntzen
6157bf3cae Add Windows support in Travis CI.
- Add new Windows support
- Use CMake/CTest instead of Make + shell scripts
- Use --parallel in CTest
- Fix CTest on Windows
- Cleanups in test_shaders.py
- Force specific commit for SPIRV-Headers
- Fix Inf/NaN odd-ball case by moving to ASM
2018-10-27 00:22:30 +02:00
Hans-Kristian Arntzen
226d837d7e HLSL: Use same logic as GLSL for picking cbuffer block name.
HLSL just picked the variable name which did not work as expected for
some users. Use the same logic as GLSL and set up declared_block_names,
so the actual name can be queried later.
2018-10-22 09:50:04 +02:00
Hans-Kristian Arntzen
519565b030
Merge pull request #718 from cdavis5e/op-image-sampled-image
MSL: Handle OpImage on OpSampledImage expressions.
2018-10-04 21:30:08 +02:00
Chip Davis
47089a48a0 Make the test case a lot simpler. 2018-10-04 11:26:46 -05:00
Chip Davis
9919fbbe0d MSL: Handle OpImage on OpSampledImage expressions.
I have seen this happen. The included test case is one such case.
2018-10-03 11:48:46 -05:00
Chip Davis
010fecc466 MSL: Swizzle gathers on depth textures as well.
Might as well.
2018-10-03 11:47:13 -05:00
Chip Davis
b7433c01ee Minor cleanups.
Throw an error for cases we don't support. Add a blank line after each
local array declaration.
2018-09-27 11:01:46 -05:00
Chip Davis
2506046cb4 Merge remote-tracking branch 'origin' into resource-arrays-msl 2018-09-27 10:50:16 -05:00
Hans-Kristian Arntzen
c07c303999 Use GL_EXT_samplerless_texture_functions in Vulkan GLSL. 2018-09-27 13:36:38 +02:00
Hans-Kristian Arntzen
af75ef005f Update glslang and SPIRV-Tools.
A lot of changes in spirv-opt output.
Some new invalid SPIR-V was found but most of them were not significant
for SPIRV-Cross, so just marked them as invalid.
2018-09-27 11:10:22 +02:00
Chip Davis
3a9af9681c MSL: Expand arrays of buffers passed as input.
Even as of Metal 2.1, MSL still doesn't support arrays of buffers
directly. Therefore, we must manually expand them. In the prologue, we
define arrays holding the argument pointers; these arrays are what the
transpiled code ends up referencing. We might be able to do similar
things for textures and samplers prior to MSL 2.0.

Speaking of which, also enable texture arrays on iOS MSL 1.2.
2018-09-26 20:48:09 -05:00
Hans-Kristian Arntzen
69b034f26e
Merge pull request #706 from cdavis5e/component-swizzle
MSL: Add an option to insert texture swizzles into generated shaders.
2018-09-25 10:06:03 +02:00
Chip Davis
7cb817e40e Add spvTexelBufferCoord for buffer image reads, too.
I should've caught this when I fixed this for writes.
2018-09-23 14:37:03 -05:00
Chip Davis
4302c5abfb Pass the swizzle constants as a buffer.
It'll be useful to have an "auxiliary buffer" for other builtins--e.g.
`DrawIndex` (which should be easier to implement now), or `ViewIndex`
when someone gets around to implementing multiview.

Pass this buffer to leaf functions as well.

Test that we handle this for integer textures as well.
2018-09-22 19:36:11 -05:00
Chip Davis
c793868417 Pack texture component swizzles by bytes. 2018-09-22 19:15:15 -05:00
Chip Davis
7fff65a811 Remove extraneous space in enum class decl. 2018-09-21 13:52:20 -05:00
Chip Davis
2583321657 MSL: Add an option to insert texture swizzles into generated shaders.
It's intended to be used with MoltenVK to support arbitrary
`VkComponentMapping` settings. The idea is that MoltenVK will pass a
buffer (which it set to some buffer index that isn't being used)
containing packed versions of the `VkComponentMapping` struct, one for
each sampled image.

Yes, this is horribly ugly. It is unfortunately necessary. Much of the
ugliness is to support swizzling gather operations, where we need to
alter the component that the gather operates on--something complicated
by the `gather()` method requiring the passed-in component to be a
constant expression. It doesn't even support swizzling gathers on depth
textures, though I could add that if it turns out we need it.
2018-09-19 22:32:24 -05:00
Chip Davis
ec857f6778 Cast uses of Layer and ViewportIndex to the expected type. 2018-09-19 09:13:30 -05:00
Chip Davis
0e9ad14ba6 MSL: Handle the ViewportIndex builtin.
This requires MSL 2.0+.

Also, force `ViewportIndex` and `Layer` to be defined as the correct
type, which is always `uint` in MSL.

Since Metal doesn't yet have geometry shaders, the vertex shader (or
tessellation evaluation shader == "post-tessellation vertex shader" in
Metal jargon) is the only kind of shader that can set this output. This
currently requires an extension to Vulkan, which causes validation of
the SPIR-V binaries for the test cases to fail. Therefore, the test
cases are marked "invalid", even though they're actually perfectly valid
SPIR-V--they just won't work without the
`SPV_EXT_shader_viewport_index_layer` extension.
2018-09-18 09:52:30 -05:00
Hans-Kristian Arntzen
a985ac9499 Add test case for continue out of switch default block. 2018-09-18 11:01:15 +02:00
Hans-Kristian Arntzen
3b5968bb26 Deal with switch cases which break out of a loop.
Need some pretty hideous ladder variable system, but high level
languages do not support breaking out of a loop. break in switch blocks
and break in loops alias each other.
2018-09-18 10:50:48 +02:00
Hans-Kristian Arntzen
439602dc6c Add missing reference files. 2018-09-18 09:32:53 +02:00
Chip Davis
7dcfed888a Use a hook to emit a local for the sample position.
That way, we don't have to handle it specially when constructing a call.
2018-09-17 11:51:09 -05:00
Chip Davis
72fc1cce53 Merge remote-tracking branch 'origin' into msl-sample-pos 2018-09-17 11:20:34 -05:00
Hans-Kristian Arntzen
737715214e Implement atomic increment/decrement in GLSL and HLSL. 2018-09-17 15:54:21 +02:00
Hans-Kristian Arntzen
a77880787d
Merge pull request #698 from KhronosGroup/fix-695
MSL: Support global I/O block and struct Input/Output usage.
2018-09-17 14:54:58 +02:00
Hans-Kristian Arntzen
49ac538a64 Remove maybe_assign_input_struct.
This is obsolete and wrong since we already unflatten I/O structs.
2018-09-17 13:51:02 +02:00
Chip Davis
39bc101e82 MSL: Handle the SamplePosition builtin.
This is somewhat tricky, because in MSL this value is obtained through a
function, `get_sample_position()`. Since the call expression is an
rvalue, it can't be passed by reference, so functions get a copy
instead.

This was the last piece preventing us from turning on sample-rate
shading support in MoltenVK.
2018-09-13 09:34:28 -05:00
Hans-Kristian Arntzen
1bbb4032c8
Merge pull request #693 from cdavis5e/msl-atomic-inc-dec
MSL: Fix OpAtomicIIncrement and OpAtomicIDecrement.
2018-09-13 16:19:27 +02:00
Hans-Kristian Arntzen
d310060f92 MSL: Support global I/O block and struct Input/Output usage.
Implement this by flattening outputs and unflattening inputs explicitly.
This allows us to pass down a single struct instead of dealing with the
insanity that would be passing down each flattened member separately.

Remove stage_uniforms_var_id.
Seems to be dead code. Naked uniforms do not exist in SPIR-V for Vulkan,
which this seems to have been intended for. It was also unused elsewhere.
2018-09-13 16:04:24 +02:00
Chip Davis
986345c754 Fix tests for changes to my last patch. 2018-09-12 09:43:12 -05:00
Hans-Kristian Arntzen
2f65a1583e MSL: Support array-of-arrays composite construction. 2018-09-12 10:25:51 +02:00
Hans-Kristian Arntzen
38d19821d4 MSL: Support copying array of arrays. 2018-09-12 09:54:55 +02:00
Hans-Kristian Arntzen
4d478316e5 Add missing reference files. 2018-09-12 09:45:08 +02:00
Hans-Kristian Arntzen
ecc94ccd1a Merge branch 'legacy-color-4comp' of git://github.com/crosire/SPIRV-Cross 2018-09-12 09:37:40 +02:00
Chip Davis
41eb5c43b5 MSL: Fix OpAtomicIIncrement and OpAtomicIDecrement.
We were passing a constant '1' to `emit_atomic_func_op()`--which caused
us to refer to SPIR-V value `%1`, which is almost certainly not what we
want! What we really want is to add/subtract the literal constant '1'
to/from the memory location.
2018-09-11 17:29:54 -05:00
crosire
3d39652853 Add tests for previous fix 2018-09-11 20:57:56 +02:00
Hans-Kristian Arntzen
403011e973
Merge pull request #684 from cdavis5e/msl-builtin-vector-cast
MSL: Cast uses of builtin vectors to their declared SPIR-V type.
2018-09-11 19:59:58 +02:00
Chip Davis
6757ef8512 Use bitcast_to_builtin_load() instead of hacking to_expression().
This only affects the builtin when it is used, and not when it's passed
to a function. It's a lot cleaner than the way I was doing it before.

Remove the `to_expression()` hack.
2018-09-11 11:15:17 -05:00
Hans-Kristian Arntzen
32a0d05e05 Bitcast loads from builtin compute variables. 2018-09-11 09:43:28 +02:00
Chip Davis
acb3fac747 Opt for a simple value cast in lieu of a bitcast. 2018-09-10 14:05:36 -05:00
Hans-Kristian Arntzen
63f6466065 Support Component decoration in GLSL. 2018-09-10 12:13:26 +02:00
Hans-Kristian Arntzen
af672b7a4b Add composite array test shaders for GLSL and HLSL as well. 2018-09-10 10:21:08 +02:00
Hans-Kristian Arntzen
aa17a02efb Add test shader for composite array initialization. 2018-09-10 10:05:00 +02:00
Hans-Kristian Arntzen
b114889102 Only declare typed initializer list for non-array types.
Also, cleanup now redundant constant_expression virtualization for MSL.
2018-09-10 10:04:17 +02:00
Chip Davis
f7dad9da66 MSL: Cast uses of builtin vectors to their declared SPIR-V type.
In SPIR-V, builtin integral vectors can be either signed or unsigned,
but in MSL they're always unsigned. Unfortunately, the MSL spec forbids
implicit conversions between vector types--even if the corresponding
scalar types would implicitly convert. If you try, the result is a
cryptic error message such as:

```
program_source:37:60: error: cannot convert between vector values of different size ('int4' (aka 'vector_int4') and 'vector_uint4' (vector of 4 'unsigned int' values))
            float4 r3 = as_type<float4>((as_type<int4>(r0) * gl_LocalInvocationID.xyyy) + as_type<int4>(r2));
                                         ~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~
```

Therefore, uses of these builtins must be explicitly cast, since the
rest of the binary likely assumes that the builtin is of its declared
type.
2018-09-08 21:17:54 -05:00