966: Immediate resource destruction and freeing r=cwfitzgerald a=kvark
**Connections**
Fixes#964
**Description**
We are making it so a buffer or a texture can have their native resources freed while they are still referenced, so without waiting for GC.
In addition, the PR adds a few missing cases where error IDs should have been handled, like at render pass encoding.
**Testing**
Tested on wgpu-rs examples, see https://github.com/gfx-rs/wgpu-rs/pull/591
Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
* Pass [u8] instead of [u32] for push constants
Use 2-instruction aligner
Fixed alignment
Fixed offsets
Added alignment checks and fixed size
Use semantic constants
* Replaced all magic number with constant
* Fixed type cast
968: CI playtests for Dx12, Vulkan r=kvark a=DevOrc
**Connections**
#950
**Description**
Implements CI testing for Vulkan and DX12. Also, cargo test stdout will no longer be captured so that people can verify that the playtests actually ran.
Note: If no adapters are found, nothing will run and the tests will pass. This might become a problem if the installation no longer works and no one notices.
**Testing**
Passed CI Tests
Co-authored-by: Noah Charlton <ncharlton002@gmail.com>
962: Refactor and optimize bind group creation r=cwfitzgerald a=kvark
**Connections**
Fixes#960
**Description**
There is a small bag of things all related in here:
1. It detects duplicate bindings in the `BindGroupDescriptor`. Previously, we compared the count, and formed the write array. If the expected bindings were [0, 1], and the given bindings were [0, 0], it would pass, erroneously. Now it properly errors out.
2. It defers the *actual* descriptor set creation to after we iterate the bindings. This means any error would not make use leak descriptors any more.
3. Finally, we form a single descriptor set write instead of having a write per binding. This is much more friendly to gfx-hal, which does the work per write, and should optimize bind group creation a bit.
**Testing**
Tested on wgpu-rs examples - https://github.com/gfx-rs/wgpu-rs/pull/589
Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
963: Add repr(C) on mapping structs r=trivial a=kvark
**Connections**
Required for https://phabricator.services.mozilla.com/D92636
**Description**
Allows C clients to avoid redefining those structs.
**Testing**
No need
Co-authored-by: Dzmitry Malyshau <dmalyshau@mozilla.com>
957: Enable RBA feature where available r=grovesNL a=kvark
**Connections**
Related to https://github.com/gpuweb/gpuweb/issues/955
**Description**
Until we have a better plan in place, enabling RBA is the least we can do.
**Testing**
Should work!
Co-authored-by: Dzmitry Malyshau <dmalyshau@mozilla.com>
940: Fixed typo in texture size check r=kvark a=Fruup
**Connections**
https://github.com/gfx-rs/wgpu/issues/937
**Description**
The texture's depth was omitted in the check.
**Testing**
No tests needed.
Co-authored-by: Leon Scherer <scherer-leon@web.de>
938: Remove erroneous extra feature check r=kvark a=cwfitzgerald
**Connections**
None, found during investigation of DI issue.
**Description**
This line of code just shows up twice in a row, checking the same feature and setting the same feature.
**Testing**
Untested, but shouldn't affect anything.
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
935: Properly implement compute pass usage scopes r=cwfitzgerald a=kvark
**Connections**
Fixes#934Closes#933
**Description**
The old barriers in a compute pass were totally wrong. They didn't combine usages across the bind groups, or with indirect buffer.
This change implements the *actual* semantics as seen by the spec. At dispatch time we gather the usages across bind groups needed by the active pipeline, possibly add the indirect buffer into the mix, and that combined used is what we transition into.
I think we can find a few interesting ideas on how to make this faster. In particular - how to avoid a separate `TrackerSet` per pass. But for now correctness is more important.
**Testing**
Tested on wgpu-rs examples
Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
936: Prevent an invalid texture from being registered in device_create_texture r=kvark a=kunalmohan
**Connections**
_Link to the issues addressed by this PR, or dependent PRs in other repositories_
**Description**
_Describe what problem this is solving, and how it's solved._
Since we checked the usage of a `Feature` in `device_create_texture()` after registering the new `Id`, there could be a case where it returns an error after registering the texture as a valid one, which further panics when the client tries to register it as an `error_id`.
To solve this, the check for the usage of a `Feature` is now done before registering the new `Id`.
**Testing**
_Explain how this change is tested._
Tested with wgpu-rs examples and CTS in Servo.
<!--
Non-trivial functional changes would need to be tested through:
- [wgpu-rs](https://github.com/gfx-rs/wgpu-rs) - test the examples.
- [wgpu-native](https://github.com/gfx-rs/wgpu-native/) - check the generated C header for sanity.
Ideally, a PR needs to link to the draft PRs in these projects with relevant modifications.
See https://github.com/gfx-rs/wgpu/pull/666 for an example.
If you can add a unit/integration test here in `wgpu`, that would be best.
-->
Co-authored-by: Kunal Mohan <kunalmohan99@gmail.com>
928: Reverts the addition of LabeledContextError r=kvark a=scoopr
Removed as discussed, moving them to wgpu-rs side
Co-authored-by: Mikko Lehtonen <scoopr@iki.fi>
* Allow dropping of error resources
* Unregister error resources without panic
Panic when removing Vacant element or when index is out of range
* address review comments
921: add `polygon_mode: PolyonMode` to `RasterizationStateDescriptor` to allow drawing wireframes r=kvark a=manuel-woelker
**Description**
This adds `polygon_mode: PolyonMode` to `RasterizationStateDescriptor` to allow drawing in wireframes and point mode
I added `hal::Features::NON_FILL_POLYGON_MODE` to the `wishful_features` in `instance.rs`. I don't know what side effects this might have.
**Testing**
Tested locally on my Windows machine with Vulkan backend.
Co-authored-by: Manuel Woelker <github@manuel.woelker.org>
922: Add LabeledContextError r=kvark a=scoopr
Purpose for this is to add more context to an error.
create_render_pipeline error is improved, by showing the action
(creating render pipeline) and the user provided label, to aid with
the investigation.
Example error
before:
```
wgpu error: Validation error
Caused by:
error in stage VERTEX: unable to find an entry point matching the Vertex execution model
thread 'main' panicked at 'Handling wgpu errors as fatal by default', src/backend/direct.rs:1457:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
```
after:
```
wgpu error: Validation error
Caused by:
Creating render pipeline (label: Some("hello-triangle pipeline"))
error in stage VERTEX
unable to find an entry point matching the Vertex execution model
thread 'main' panicked at 'Handling wgpu errors as fatal by default', src/backend/direct.rs:1457:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
```
Co-authored-by: Mikko Lehtonen <scoopr@iki.fi>
Purpose for this is to add more context to an error.
create_render_pipeline error is improved, by showing the action
(creating render pipeline) and the user provided label, to aid with
the investigation.
923: Fix invalid mip level count check r=kvark a=benfrankel
**Connections**
Fixes https://github.com/gfx-rs/wgpu/issues/894
Continues from https://github.com/gfx-rs/wgpu/pull/893
**Description**
Invalid mip level counts are allowed through `Device::create_texture`, triggering Vulkan validation errors. A fix was applied in https://github.com/gfx-rs/wgpu/pull/893, but the fix needs a minor correction to work properly.
Also added back the `MAX_MIP_LEVELS` check, because it's technically possible for `MAX_MIP_LEVELS` to be less than `kind.compute_num_levels()`, e.g. if one of the given dimensions is very big (2^16).
Error message is updated to reflect the change in behavior: there are now three ways for a mip level count to be invalid: it's equal to zero; it's greater than `MAX_MIP_LEVELS`; or it's greater than the largest mip level count allowed for the given size.
**Testing**
Tested by manually modifying the `mipmap` example in `wgpu-rs` to pass various edge case values for `size` and `mip_level_count`.
Co-authored-by: Ben Frankel <ben.frankel7@gmail.com>
920: Don't bind groups that aren't expected by the layout r=grovesNL a=kvark
**Connections**
Sibling of #919 for master.
**Description**
**Testing**
In addition to the fix itself, this comes with a new regression playtest!
Co-authored-by: Dzmitry Malyshau <dmalyshau@mozilla.com>
