Commit Graph

57 Commits

Author SHA1 Message Date
Dan Sinclair
d409210ee5 Move all .invalid shaders into no-opt folders. 2019-11-05 13:19:19 -05:00
Hans-Kristian Arntzen
a0c13e4ee8 Do not consider aliased struct types if the master is not a block.
It is possible for a shader to declare two plain struct types which
simply share the same OpName without there being an implicit
value/buffer alias relationship.

For to_member_name(), make sure to use the type alias master when
resolving member names. The member name may be different in a type alias
master if the SPIR-V is being intentionally difficult.
2019-10-07 10:52:16 +02:00
Chip Davis
2eff420d9a Support the SPV_EXT_fragment_shader_interlock extension.
This was straightforward to implement in GLSL. The
`ShadingRateInterlockOrderedEXT` and `ShadingRateInterlockUnorderedEXT`
modes aren't implemented yet, because we don't support
`SPV_NV_shading_rate` or `SPV_EXT_fragment_invocation_density` yet.

HLSL and MSL were more interesting. They don't support this directly,
but they do support marking resources as "rasterizer ordered," which
does roughly the same thing. So this implementation scans all accesses
inside the critical section and marks all storage resources found
therein as rasterizer ordered. They also don't support the fine-grained
controls on pixel- vs. sample-level interlock and disabling ordering
guarantees that GLSL and SPIR-V do, but that's OK. "Unordered" here
merely means the order is undefined; that it just so happens to be the
same as rasterizer order is immaterial. As for pixel- vs. sample-level
interlock, Vulkan explicitly states:

> With sample shading enabled, [the `PixelInterlockOrderedEXT` and
> `PixelInterlockUnorderedEXT`] execution modes are treated like
> `SampleInterlockOrderedEXT` or `SampleInterlockUnorderedEXT`
> respectively.

and:

> If [the `SampleInterlockOrderedEXT` or `SampleInterlockUnorderedEXT`]
> execution modes are used in single-sample mode they are treated like
> `PixelInterlockOrderedEXT` or `PixelInterlockUnorderedEXT`
> respectively.

So this will DTRT for MoltenVK and gfx-rs, at least.

MSL additionally supports multiple raster order groups; resources that
are not accessed together can be placed in different ROGs to allow them
to be synchronized separately. A more sophisticated analysis might be
able to place resources optimally, but that's outside the scope of this
change. For now, we assign all resources to group 0, which should do for
our purposes.

`glslang` doesn't support the `RasterizerOrdered` UAVs this
implementation produces for HLSL, so the test case needs `fxc.exe`.

It also insists on GLSL 4.50 for `GL_ARB_fragment_shader_interlock`,
even though the spec says it needs either 4.20 or
`GL_ARB_shader_image_load_store`; and it doesn't support the
`GL_NV_fragment_shader_interlock` extension at all. So I haven't been
able to test those code paths.

Fixes #1002.
2019-09-02 12:31:10 -05:00
Chip Davis
5fe1ecc324 GLSL: Fix post-depth coverage for ESSL.
ESSL does not support `GL_ARB_post_depth_coverage`. There, we must use
`GL_EXT_post_depth_coverage`. I've added this as a fallback for desktop
as well.

Note that `GL_EXT_post_depth_coverage` also requires the fragment shader
to set `early_fragment_tests` explicitly, while
`GL_ARB_post_depth_coverage` does not. It doesn't really matter either
way, since `SPV_KHR_post_depth_coverage` *also* requires both execution
modes to be explicitly set.
2019-08-28 13:40:13 -05:00
Chip Davis
343c6f4ff4 Update external repos.
Fix fallout from changes.

There's a bug in glslang that prevents `float16_t`, `[u]int16_t`, and
`[u]int8_t` constants from adding the corresponding SPIR-V capabilities.
SPIRV-Tools, meanwhile, tightened validation so that these constants are
only valid if the corresponding `Float16`, `Int16`, and `Int8` caps are
on. This affects the `16bit-constants.frag` test for GLSL and MSL.
2019-07-13 16:50:21 -05:00
Chip Davis
1df47db6ba Support the SPV_KHR_post_depth_coverage extension.
Using the `PostDepthCoverage` mode specifies that the `gl_SampleMaskIn`
variable is to contain the computed coverage mask following the early
fragment tests, which this mode requires and implicitly enables.

Note that unlike Vulkan and OpenGL, Metal places this on the sample mask
input itself, and furthermore does *not* implicitly enable early
fragment testing. If it isn't enabled explicitly with an
`[[early_fragment_tests]]` attribute, the compiler will error out. So we
have to enable that mode explicitly if `PostDepthCoverage` is enabled
but `EarlyFragmentTests` isn't.

For Metal, only iOS supports this; for some reason, Apple has yet to
implement it on macOS, even though many desktop cards support it.
2019-07-11 10:28:43 -05:00
Hans-Kristian Arntzen
50342966c0 Fall back to complex loop if non-trivial continue block is found.
There is a case where we can deduce a for/while loop, but the continue
block is actually very painful to deal with, so handle that case as
well. Removes an exceptional case.
2019-07-08 11:54:29 +02:00
Hans-Kristian Arntzen
041f103d44 MSL/HLSL: Support scalar reflect and refract. 2019-07-03 12:31:52 +02:00
Hans-Kristian Arntzen
fc9fe4e480 Fix variable scope when an if or else block dominates a variable.
Just like loops, we need complicated hoisting again to make this work.
2019-07-03 11:18:50 +02:00
Hans-Kristian Arntzen
707312b83a GLSL: Support NV barycentrics. 2019-06-19 09:52:35 +02:00
Hans-Kristian Arntzen
6b52b0fe8b Deal with nested loops.
Actually need to hoist out variable to outermost loop.
2019-06-06 14:37:02 +02:00
Hans-Kristian Arntzen
03d93abc1a Deal with case where a variable is dominated by inner part of a loop.
There is a risk that we try to preserve a loop variable through multiple
iterations, even though the dominating block is inside a loop.

Fix this by analyzing if a block starts off by writing to a variable. In
that case, there cannot be any preservation going on. If we don't, pretend the
loop header is reading the variable, which moves the variable to an
appropriate scope.
2019-06-06 11:11:44 +02:00
Hans-Kristian Arntzen
4ef51331b2 Always value-cast FP16 constants instead of using literals.
GL_NV_gpu_shader5 doesn't support "hf", so to avoid lots of complicated
workarounds, just value-cast the half literals.
2019-02-20 12:30:01 +01:00
Hans-Kristian Arntzen
3e584f2c3f Support LUTs in single-function CFGs on Private storage class.
Fairly common pattern in unoptimized SPIR-V. Support this case as well.
2019-02-06 10:38:59 +01:00
Hans-Kristian Arntzen
2ed171e525 GLSL/MSL: Implement 8-bit part of VK_KHR_shader_float16_int8.
Storage was in place already, so mostly just dealing with bitcasts and
constants.

Simplies some of the bitcasting logic, and this exposed some bugs in the
implementation. Refactor to use correct width integers with explicit bitcast opcodes.
2019-01-30 15:45:24 +01:00
Hans-Kristian Arntzen
d92de00cc1 Rewrite how IDs are iterated over.
This is a fairly fundamental change on how IDs are handled.
It serves many purposes:

- Improve performance. We only need to iterate over IDs which are
  relevant at any one time.
- Makes sure we iterate through IDs in SPIR-V module declaration order
  rather than ID space. IDs don't have to be monotonically increasing,
  which was an assumption SPIRV-Cross used to have. It has apparently
  never been a problem until now.
- Support LUTs of structs. We do this by interleaving declaration of
  constants and struct types in SPIR-V module order.

To support this, the ParsedIR interface needed to change slightly.
Before setting any ID with variant_set<T> we let ParsedIR know
that an ID with a specific type has been added. The surface for change
should be minimal.

ParsedIR will maintain a per-type list of IDs which the cross-compiler
will need to consider for later.

Instead of looping over ir.ids[] (which can be extremely large), we loop
over types now, using:

ir.for_each_typed_id<SPIRVariable>([&](uint32_t id, SPIRVariable &var) {
	handle_variable(var);
});

Now we make sure that we're never looking at irrelevant types.
2019-01-10 12:52:56 +01:00
Hans-Kristian Arntzen
acae607703 Register implied expression reads in OpLoad/OpAccessChain.
This is required to avoid relying on complex sub-expression elimination
in compilers, and generates cleaner code.

The problem case is if a complex expression is used in an access chain,
like:

Composite comp = buffer[texture(...)];
vec4 a = comp.a + comp.b + comp.c;

Before, we did not have common subexpression tracking for
OpLoad/OpAccessChain, so we easily ended up with code like:

vec4 a = buffer[texture(...)].a + buffer[texture(...)].b + buffer[texture(...)].c;

A good compiler will optimize this, but we should not rely on it, and
forcing texture(...) to a temporary also looks better.

The solution is to add a vector "implied_expression_reads", which works
similarly to expression_dependencies. We also need an extra mechanism in
to_expression which lets us skip expression read checking and do it
later. E.g. for expr -> access chain -> load, we should only trigger
a read of expr when using the loaded expression.
2019-01-04 14:56:12 +01:00
Hans-Kristian Arntzen
61f1d8b2cf Support gl_HelperInvocation on GLSL and MSL.
There is no obvious builtin for this on HLSL.
2018-11-28 15:18:43 +01:00
Hans-Kristian Arntzen
04f410d35c Fix unsigned switch case selectors. 2018-11-26 10:36:50 +01:00
Chip Davis
ca4744ab72 Support constants of 16-bit integral type in GLSL and MSL.
Constants of 8-bit type aren't supported in GLSL, since there's no
extension letting you use them.
2018-11-02 14:39:55 -05:00
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
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
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
eee290a029 MSL: Fix support for texelFetchOffset.
Just apply the offset directly, MSL has no immediate offset parameter.
2018-08-07 15:28:04 +02:00
Hans-Kristian Arntzen
5582523d9a Add some tests for LUT promotion.
Also, update other tests.
2018-07-05 14:14:18 +02:00
Hans-Kristian Arntzen
d29f48ef06 Deduce constant LUTs from read-write variables. 2018-07-05 13:25:57 +02:00
Pascal Muetschard
aced6058b4 Don't limit GLSL identifiers with HLSL keywords.
- The HLSL compiler now has its own list of keywords in addition to
   the ones from GLSL.
 - Added "buffer", "precise", and "shared" to the GLSL keywords.
2018-05-07 10:58:52 -07:00
Hans-Kristian Arntzen
47081f810a Fix GatherDref on GLSL. 2018-04-30 12:45:23 +02:00
Hans-Kristian Arntzen
e8e58844d4 Rewrite everything to use Bitset rather than uint64_t. 2018-03-12 13:24:14 +01:00
Hans-Kristian Arntzen
922420e346 Disallow arrays and structs from becoming loop variables.
Fixes awkward code-gen issue.
2018-03-07 14:54:11 +01:00
Hans-Kristian Arntzen
5fe79eb59c Update tests.
Adds an earlier reported shader packing failure into regression suite.
2018-03-05 16:34:42 +01:00
Hans-Kristian Arntzen
047ad7df0f Support special float constants (NaN/Inf). 2018-02-23 13:06:20 +01:00
Hans-Kristian Arntzen
843e34b604 Add IsFrontFace support to HLSL. 2018-02-15 12:42:56 +01:00
Hans-Kristian Arntzen
636cc30088 Fix case where hoisted temporaries were used before being declared. 2018-02-15 10:52:56 +01:00
Hans-Kristian Arntzen
5d9df6a31c Do not declare constant composites inline in HLSL.
Move arrays and structs out to their own global static constants.

Also, replace illegal names in HLSL as well.
2018-02-02 10:12:26 +01:00
Hans-Kristian Arntzen
af0a887997 Add test for false loop init.
Clean up how for loop variables are declared.
2018-01-23 21:15:09 +01:00
Hans-Kristian Arntzen
3c52771aee Make sure image integer coords are int, not uint.
HLSL can emit uint here.
2017-12-01 15:02:50 +01:00
Hans-Kristian Arntzen
9091eadb0d Support FrexpStruct/ModfStruct. 2017-09-04 10:27:08 +02:00
Hans-Kristian Arntzen
744d0405b0 Preserve arguments with inout unless complete writes are made. 2017-08-09 17:06:41 +02:00
Hans-Kristian Arntzen
6ff9007311 Fix unary enclosures. 2017-07-24 10:17:19 +02:00
Hans-Kristian Arntzen
c8d60914c4 Add support for SampleId/SampleMask/SamplePosition builtins. 2017-07-24 10:07:31 +02:00
David Srbecky
77b5b4446b Always make a copy when handling OpCompositeInsert
The modified object might not be mutable (e.g. shader input).

Added a test for the case when this happens.
2017-06-26 18:32:53 +01:00
Hans-Kristian Arntzen
de33d89074 Add explicit in/out locations everywhere.
Needed for newer glslang. With Vulkan semantics for SPIR-V, all
locations must be explicitly defined.
2017-06-21 09:39:08 +02:00
Hans-Kristian Arntzen
851acf3712 Avoid boolean mix in HLSL.
Update glslang travis checkout as boolean mix support was broken on that commit.
2017-05-04 10:28:30 +02:00
Hans-Kristian Arntzen
036b9b73f5 Emit storage qualifier in the correct place.
Need to emit qualifiers in a specific order for legacy GLSL.

<interpolation> <storage> <precision> <type>.
2017-02-24 09:56:17 +01:00
Hans-Kristian Arntzen
0e7c33ff81 Update glslang to latest on Travis.
Fix a lingering bug with OpSelect and boolean mix().
2017-02-11 10:52:34 +01:00
Hans-Kristian Arntzen
44b3216611 Sort loop variables to make sure Travis CI runs are reproducable. 2016-12-16 14:01:09 +01:00
Hans-Kristian Arntzen
45c797d54c Improve debuggability of Travis CI when things go wrong. 2016-12-16 13:48:30 +01:00
Hans-Kristian Arntzen
d11b8aa3ef Optimize += 1, -= 1 to ++, --.
Purely cosmetic, but easier to read.
2016-12-16 13:24:49 +01:00
Hans-Kristian Arntzen
62613df5a5 Optimize for read-modify-writes.
Required for legacy loop increments.
2016-12-16 13:14:22 +01:00