Commit Graph
8323 Commits
Author SHA1 Message Date
João Capucho d3fe1c978b [glsl-in] Add support for precision qualifiers 2021-06-21 21:37:08 +01:00
João Capucho 898c2e7c68 [glsl-in] Use macro for qualifier match 2021-06-21 21:37:08 +01:00
João Capucho 93436047f3 [glsl-in] Fix builtin types 2021-06-21 21:37:08 +01:00
Igor Shaposhnik 3a4d6fa295 Reduce code duplication between backends 2021-06-21 15:33:56 -04:00
Jim Blandy 548cde4701 Clear Validator::valid_expression_list before validating each function.
If validation fails, `Validator` may be left with entries in
`valid_expression_list` and `valid_expression_set`. The validator does clear the
set before examining a function's body, but not the list, which means a
long-lived `Validator` could accumulate an unbounded amount of garbage in the
list.

(Minor cleanup: use structural update syntax where appropriate.)
2021-06-21 14:53:20 -04:00
Dzmitry Malyshau da38b8b077 Make spirv an optional feature 2021-06-21 12:09:12 -04:00
João Capucho cae71a788a [glsl-in] Support for vector scalar operations 2021-06-21 10:44:26 -04:00
bors[bot]andgrovesNL 56d584ad92 Merge #1534
1534: web: Restore WebGPU CI r=kvark a=grovesNL

**Connections**
Depends on/currently includes #1533

**Description**
Even though the WebGL backend needs to be temporarily disabled, we should be able to continue checking WebGPU on wasm.

**Testing**
Manually ran the CI command `cargo check --package wgpu --examples --target=wasm32-unknown-unknown` locally which seems to catch the mistake in #1533 without building wgpu-core/wgpu-hal.

Co-authored-by: grovesNL <josh@joshgroves.com>
2021-06-21 06:04:34 +00:00
bors[bot]andgrovesNL 88b11e28b4 Merge #1533
1533: web: add empty `device_create_shader_module_spirv` r=kvark a=grovesNL

**Description**
We added a SPIR-V passthrough function to the `Context` trait shared between web/native backends, but the web backend also needs an implementation of this function to fully implement the trait.

**Testing**
Compiling the web backend for WebGPU.

Co-authored-by: grovesNL <josh@joshgroves.com>
2021-06-21 05:57:47 +00:00
grovesNL ee254291c7 web: restore CI for WebGPU 2021-06-21 01:32:24 -02:30
grovesNL fdd30916db web: add empty device_create_shader_module_spirv 2021-06-21 01:19:25 -02:30
João Capucho ffe67139cd [glsl-in] Add support for mix with bool selectors 2021-06-20 23:11:16 -04:00
João Capucho c152205c87 [glsl-in] arguments of opaque types are constant 2021-06-20 23:11:16 -04:00
João Capucho 01036e8f42 [glsl-in] Use crate::BOOL_WIDTH for booleans width 2021-06-20 23:11:16 -04:00
João Capucho e7d712e608 [glsl-in] Apply constant binary ops to composites 2021-06-20 23:11:16 -04:00
João Capucho 60644504d8 [glsl-in] Implicit cast stores with ValuePointer 2021-06-20 23:11:16 -04:00
João Capucho 68609cdff8 [glsl-in] Implement textureSize 2021-06-20 23:11:16 -04:00
João Capucho fd042e6f31 [glsl-in] Implement default for options 2021-06-20 23:11:16 -04:00
João Capucho 16d7c0e428 [glsl-in] Implement texelFetch 2021-06-20 23:11:16 -04:00
Jim Blandy 696ca340e0 Improve docs for TypeInner, valid::TypeFlags, and some internal types. 2021-06-20 12:18:50 -04:00
Jim Blandy 70016ce6e3 Address warnings about links from cargo doc --package naga. 2021-06-20 12:18:50 -04:00
bors[bot]andAlex S 96dbd06a1f Merge #1525
1525: Add Naga bypass to allow feeding raw SPIR-V shader data to the backend. r=kvark a=ElectronicRU

**Connections**
Fixes #1520 .

**Description**
While Naga checking is undoubtedly very useful, it currently lags behind
what is possible in SPIR-V and even what is promised by WGPU (ie binding
arrays). This adds an unsafe method to wgpu::Device to allow feeding
raw SPIR-V data into the backend, and adds a feature flag to request a
backend supporting this operation.

**Testing**
`texture_arrays` example is runnable now, which uncovers an additional bug in Vulkan backend - binding arrays are instead treated like sole bindings, and indices shift, too. Lots of errors from validation layer ensue.


Co-authored-by: Alex S <alex0player@gmail.com>
2021-06-20 16:13:53 +00:00
bors[bot]andDzmitry Malyshau 55c7b7975b Merge #1530
1530: Fix wgpu downlevel check r=kvark a=kvark

**Connections**
Fixes #1529

**Description**
The anisotropic flag is not in the compliant set, but the check was made for equality of the flags (instead of inclusion).

**Testing**
Untested


Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-06-20 16:07:56 +00:00
Alex S 7cd7b18560 Code review:
- Remove Cow from wgpu-hal API surface. You graze on in our hearts.
- Rename feature flag to SPIRV_SHADER_PASSTHROUGH in anticipation
  of crate feature `spirv` to reduce confusion.
- Add some more documentation about behaviour and intended avenues of
  use to the feature.
2021-06-20 15:22:04 +03:00
bors[bot]andAlex S cfda54f108 Merge #1528
1528: Support TEXTURE_SPECIFIC_FORMAT_CAPABILITIES for rendertargets. r=kvark a=ElectronicRU

**Description**
With TEXTURE_SPECIFIC_FORMAT_FEATURES enabled, one would assume that render pipeline creation would use extended features
for color/depth-stencil render targets. However, it does not. This aims to fix that discrepancy.

Co-authored-by: Alex S <alex0player@gmail.com>
2021-06-20 06:14:42 +00:00
Dzmitry Malyshau 70071228a3 [spv-out] don't OpConvert the same width 2021-06-20 01:27:58 -04:00
Dzmitry Malyshau e72bf3e10c Fix wgpu downlevel check 2021-06-20 01:24:44 -04:00
bors[bot]andAlex S 6cacfca10b Merge #1527
1527: Fix wgpu-hal incorrectly ignoring arrays of bindings. r=kvark a=ElectronicRU

**Connections**
Fixes #1526.

**Description**
Even though binding sizes could potentially be derived from resource array lengths and binding indices, current implementation
flat out ignored that. Metal did a more sensible (?) thing and bound only the first resource in each array, Vulkan did a less
sensible thing and bound as many resources as there are bindings, obtaining them from unwrapped arrays.
(IE on Vulkan, if you bound two arrays, [view1, view2] and [view3, view4], it would actually bind [view1, <invalid>] and [view2, <invalid>]).

This aims to fix the issue.

**Testing**
After #1525 is merged, `cargo run --example texture-arrays` would be working as intended for Vulkan. Unfortunately, I can think of no easy way to test this on Metal - the shader support isn't there yet.


Co-authored-by: Alex S <alex0player@gmail.com>
2021-06-20 05:08:12 +00:00
Alex S a82062595a Code review:
Remove .size field from BindGroupEntry, since this information
can be obtained from the layout.
2021-06-20 06:53:03 +03:00
Alex S 8bfa949261 Code review:
- rename ShaderInput variants
- rename feature flag
- hard error on Metal backend trying to compile SPIR-V
- fix test failing because of feature flag bits changing
2021-06-20 06:22:59 +03:00
Alex S 066bd7e647 Support TEXTURE_SPECIFIC_FORMAT_CAPABILITIES for rendertargets. 2021-06-20 05:57:55 +03:00
Alex S b0ba01dba6 ...And add the feature flag to Vulkan, which previously worked by accident. 2021-06-20 05:12:19 +03:00
Alex S 3d6cda100a Fix unfortunate flag duplication. (I wish bitflags would catch that). 2021-06-20 05:05:33 +03:00
Alex S 7e6d6a33ed Fix (?) texture array binding for Metal.
Potentially needs more testing - Metal's binding model is quite
different from Vulkan and WGPU, but as far as I understand from the
documentation, arrays of textures occupy a contiguous range of binding
slots, so unwrapping and binding one-by-one should be fine.

Someone should actually go and test this on a Mac, maybe.
2021-06-19 23:35:48 +03:00
Alex S dd4c681a2f Fix HAL binding for Vulkan. 2021-06-19 23:35:18 +03:00
Alex S 6d2e6e5a56 Add Naga bypass to allow feeding raw SPIR-V shader data to the backend.
While Naga checking is undoubtedly very useful, it currently lags behind
what is possible in SPIR-V and even what is promised by WGPU (ie binding
arrays). This adds an unsafe method to wgpu::Device to allow feeding
raw SPIR-V data into the backend, and adds a feature flag to request a
backend supporting this operation.
2021-06-19 23:32:44 +03:00
bors[bot]andGordon-F 616b489061 Merge #1524
1524: Remove MPL 2.0 header in files r=kvark a=Gordon-F

This is no more required since the repository is licensed under `MIT` and `Apache 2.0`.

Co-authored-by: Gordon-F <ishaposhnik@icloud.com>
2021-06-19 18:17:10 +00:00
Gordon-F 2873ac32d9 Remove MPL 2.0 header in files 2021-06-19 21:06:08 +03:00
João Capucho 4bc4c60663 [glsl-in] Implicit casts for function calls 2021-06-19 13:58:21 -04:00
bors[bot]andAlex S a8be371acf Merge #1521
1521: Better compile error for not using resolver=2. r=kvark a=ElectronicRU

**Connections**
Usability fix proposed in #1517 with some re-wording to make the intent clearer.

**Description**
Provide a useful compile error in case a downstream project is using old Cargo resolver, instead of the deluge of obscure errors they get r/n if not on MacOS.

**Limitations**
On MacOS, vulkan feature is totally valid, so we cannot detect resolver=1 on Mac.


Co-authored-by: Alex S <alex0player@gmail.com>
2021-06-19 15:08:13 +00:00
Alex S 1dea443340 Better compile error for not using resolver=2. 2021-06-19 17:36:50 +03:00
bors[bot]andAlex S 77e57ebcc6 Merge #1519
1519: Add TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES to supported for Vulkan and Metal. r=kvark a=ElectronicRU

Adapter callbacks are already there, so this is likely an oversight.

**Description**
Adapter callbacks for texture capabilities were there, but feature flags weren't enabled, which prevented use of them.


Co-authored-by: Alex S <alex0player@gmail.com>
2021-06-19 14:29:45 +00:00
Alex S 70168ee0f5 Query optimal tiling features for Vulkan.
Images are always created with vk::ImageTiling::OPTIMAL,
so it doesn't make much sense querying linear tiling features.
2021-06-19 17:25:39 +03:00
Alex S 1b96f1a27e Add TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES to supported for Vulkan and Metal.
Adapter callbacks are already there, so this is likely an oversight.
2021-06-19 17:25:39 +03:00
Gordon-F 6cdd332cef Reorganize test snapshot folder 2021-06-19 10:19:04 -04:00
Jim Blandy c16f2097ad [spv-out]: Ensure array subscripts are in bounds. 2021-06-18 20:44:07 -04:00
Jim Blandy a7cacab276 [spv-out] Move Writer type to back/spv/mod.rs.
The `impl` blocks for `Writer` and its associated types remain in spv/writer.rs.

There should be no substantive code changes in this commit, only moving code
around and adjusting `use` declarations to match.
2021-06-18 20:44:07 -04:00
Jim Blandy 9146942412 [spv-out]: Abstract out null constant creation. 2021-06-18 20:44:07 -04:00
Jim Blandy ac92f09a48 [spv-out] Document is_intermediate. 2021-06-18 20:44:07 -04:00
Jim Blandy 1d0b3f3a75 Functions can take pointers to unsized structs, but no other unsized types. 2021-06-18 20:44:07 -04:00