Merge branch 'trunk' into ad/i64-atomics

CHANGELOG.md

@@ -47,7 +47,13 @@ TODO(wumpf): This is still work in progress. Should write a bit more about it. A
 
 `wgpu::ComputePass` recording methods (e.g. `wgpu::ComputePass:set_render_pipeline`) no longer impose a lifetime constraint passed in resources.
 
-By @wumpf in [#5569](https://github.com/gfx-rs/wgpu/pull/5569), [#5575](https://github.com/gfx-rs/wgpu/pull/5575).
+Furthermore, `wgpu::ComputePass` no longer has a life time dependency on its parent `wgpu::CommandEncoder`.
+⚠️ As long as a `wgpu::ComputePass` is pending for a given `wgpu::CommandEncoder`, creation of a compute or render pass is an error and invalidates the `wgpu::CommandEncoder`.
+Previously, this was statically enforced by a lifetime constraint.
+TODO(wumpf): There was some discussion on whether to make this life time constraint opt-in or opt-out (entirely on `wgpu` side, no changes to `wgpu-core`).
+Lifting this lifetime dependencies is very useful for library authors, but opens up an easy way for incorrect use.
+
+By @wumpf in [#5569](https://github.com/gfx-rs/wgpu/pull/5569), [#5575](https://github.com/gfx-rs/wgpu/pull/5575), [#5620](https://github.com/gfx-rs/wgpu/pull/5620).
 
 #### Querying shader compilation errors
 
@@ -83,28 +89,47 @@ By @stefnotch in [#5410](https://github.com/gfx-rs/wgpu/pull/5410)
 #### Naga
 
 - Implement `WGSL`'s `unpack4xI8`,`unpack4xU8`,`pack4xI8` and `pack4xU8`. By @VlaDexa in [#5424](https://github.com/gfx-rs/wgpu/pull/5424)
+- Began work adding support for atomics to the SPIR-V frontend. Tracking issue is [here](https://github.com/gfx-rs/wgpu/issues/4489). By @schell in [#5702](https://github.com/gfx-rs/wgpu/pull/5702).
 
 ### Changes
 
 #### General
 
 - Avoid introducing spurious features for optional dependencies. By @bjorn3 in [#5691](https://github.com/gfx-rs/wgpu/pull/5691)
 
+#### Metal
+- Removed the `link` Cargo feature.
+
+  This was used to allow weakly linking frameworks. This can be achieved with putting something like the following in your `.cargo/config.toml` instead:
+  ```toml
+  [target.'cfg(target_vendor = "apple")']
+  rustflags = ["-C", "link-args=-weak_framework Metal -weak_framework QuartzCore -weak_framework CoreGraphics"]
+  ```
+
 ### Bug Fixes
 
-### General
+#### General
 
 - Ensure render pipelines have at least 1 target. By @ErichDonGubler in [#5715](https://github.com/gfx-rs/wgpu/pull/5715)
 
+#### Metal
+
+- Fix unrecognized selector crash on iOS 12. By @vladasz in [#5744](https://github.com/gfx-rs/wgpu/pull/5744).
+
 #### Vulkan
 
 - Fix enablement of subgroup ops extension on Vulkan devices that don't support Vulkan 1.3. By @cwfitzgerald in [#5624](https://github.com/gfx-rs/wgpu/pull/5624).
 
 #### GLES / OpenGL
 
--  Fix regression on OpenGL (EGL) where non-sRGB still used sRGB [#5642](https://github.com/gfx-rs/wgpu/pull/5642)
--  Fix `ClearColorF`, `ClearColorU` and `ClearColorI` commands being issued before `SetDrawColorBuffers` [#5666](https://github.com/gfx-rs/wgpu/pull/5666)
--  Replace `glClear` with `glClearBufferF` because `glDrawBuffers` requires that the ith buffer must be `COLOR_ATTACHMENTi` or `NONE` [#5666](https://github.com/gfx-rs/wgpu/pull/5666)
+- Fix regression on OpenGL (EGL) where non-sRGB still used sRGB [#5642](https://github.com/gfx-rs/wgpu/pull/5642)
+- Fix `ClearColorF`, `ClearColorU` and `ClearColorI` commands being issued before `SetDrawColorBuffers` [#5666](https://github.com/gfx-rs/wgpu/pull/5666)
+- Replace `glClear` with `glClearBufferF` because `glDrawBuffers` requires that the ith buffer must be `COLOR_ATTACHMENTi` or `NONE` [#5666](https://github.com/gfx-rs/wgpu/pull/5666)
+- Return the unmodified version in driver_info. By @Valaphee in [#5753](https://github.com/gfx-rs/wgpu/pull/5753)
+
+#### WebGPU
+
+- Added support for pipeline-overridable constants to the WebGPU backend by @DouglasDwyer in [#5688](https://github.com/gfx-rs/wgpu/pull/5688)
 
 ## v0.20.0 (2024-04-28)
 

deno_webgpu/command_encoder.rs

@@ -261,15 +261,14 @@ pub fn op_webgpu_command_encoder_begin_compute_pass(
         timestamp_writes: timestamp_writes.as_ref(),
     };
 
-    let compute_pass = gfx_select!(command_encoder => instance.command_encoder_create_compute_pass_dyn(*command_encoder, &descriptor));
-
+    let (compute_pass, error) = gfx_select!(command_encoder => instance.command_encoder_create_compute_pass_dyn(*command_encoder, &descriptor));
     let rid = state
         .resource_table
         .add(super::compute_pass::WebGpuComputePass(RefCell::new(
             compute_pass,
         )));
 
-    Ok(WebGpuResult::rid(rid))
+    Ok(WebGpuResult::rid_err(rid, error))
 }
 
 #[op2]

deno_webgpu/pipeline.rs

@@ -114,6 +114,7 @@ pub fn op_webgpu_create_compute_pipeline(
             entry_point: compute.entry_point.map(Cow::from),
             constants: Cow::Owned(compute.constants.unwrap_or_default()),
             zero_initialize_workgroup_memory: true,
+            vertex_pulling_transform: false,
         },
         cache: None,
     };
@@ -363,6 +364,7 @@ pub fn op_webgpu_create_render_pipeline(
                 constants: Cow::Owned(fragment.constants.unwrap_or_default()),
                 // Required to be true for WebGPU
                 zero_initialize_workgroup_memory: true,
+                vertex_pulling_transform: false,
             },
             targets: Cow::Owned(fragment.targets),
         })
@@ -388,6 +390,7 @@ pub fn op_webgpu_create_render_pipeline(
                 constants: Cow::Owned(args.vertex.constants.unwrap_or_default()),
                 // Required to be true for WebGPU
                 zero_initialize_workgroup_memory: true,
+                vertex_pulling_transform: false,
             },
             buffers: Cow::Owned(vertex_buffers),
         },

naga/CHANGELOG.md

@@ -79,6 +79,7 @@ For changelogs after v0.14, see [the wgpu changelog](../CHANGELOG.md).
 
 - Add and fix minimum Metal version checks for optional functionality. ([#2486](https://github.com/gfx-rs/naga/pull/2486)) **@teoxoy**
 - Make varyings' struct members unique. ([#2521](https://github.com/gfx-rs/naga/pull/2521)) **@evahop**
+- Add experimental vertex pulling transform flag. ([#5254](https://github.com/gfx-rs/wgpu/pull/5254)) **@bradwerth**
 
 #### GLSL-OUT
 

naga/src/back/msl/mod.rs

@@ -222,6 +222,113 @@ impl Default for Options {
     }
 }
 
+/// Corresponds to [WebGPU `GPUVertexFormat`](
+/// https://gpuweb.github.io/gpuweb/#enumdef-gpuvertexformat).
+#[repr(u32)]
+#[derive(Copy, Clone, Debug, Hash, Eq, PartialEq)]
+#[cfg_attr(feature = "serialize", derive(serde::Serialize))]
+#[cfg_attr(feature = "deserialize", derive(serde::Deserialize))]
+pub enum VertexFormat {
+    /// Two unsigned bytes (u8). `vec2<u32>` in shaders.
+    Uint8x2 = 0,
+    /// Four unsigned bytes (u8). `vec4<u32>` in shaders.
+    Uint8x4 = 1,
+    /// Two signed bytes (i8). `vec2<i32>` in shaders.
+    Sint8x2 = 2,
+    /// Four signed bytes (i8). `vec4<i32>` in shaders.
+    Sint8x4 = 3,
+    /// Two unsigned bytes (u8). [0, 255] converted to float [0, 1] `vec2<f32>` in shaders.
+    Unorm8x2 = 4,
+    /// Four unsigned bytes (u8). [0, 255] converted to float [0, 1] `vec4<f32>` in shaders.
+    Unorm8x4 = 5,
+    /// Two signed bytes (i8). [-127, 127] converted to float [-1, 1] `vec2<f32>` in shaders.
+    Snorm8x2 = 6,
+    /// Four signed bytes (i8). [-127, 127] converted to float [-1, 1] `vec4<f32>` in shaders.
+    Snorm8x4 = 7,
+    /// Two unsigned shorts (u16). `vec2<u32>` in shaders.
+    Uint16x2 = 8,
+    /// Four unsigned shorts (u16). `vec4<u32>` in shaders.
+    Uint16x4 = 9,
+    /// Two signed shorts (i16). `vec2<i32>` in shaders.
+    Sint16x2 = 10,
+    /// Four signed shorts (i16). `vec4<i32>` in shaders.
+    Sint16x4 = 11,
+    /// Two unsigned shorts (u16). [0, 65535] converted to float [0, 1] `vec2<f32>` in shaders.
+    Unorm16x2 = 12,
+    /// Four unsigned shorts (u16). [0, 65535] converted to float [0, 1] `vec4<f32>` in shaders.
+    Unorm16x4 = 13,
+    /// Two signed shorts (i16). [-32767, 32767] converted to float [-1, 1] `vec2<f32>` in shaders.
+    Snorm16x2 = 14,
+    /// Four signed shorts (i16). [-32767, 32767] converted to float [-1, 1] `vec4<f32>` in shaders.
+    Snorm16x4 = 15,
+    /// Two half-precision floats (no Rust equiv). `vec2<f32>` in shaders.
+    Float16x2 = 16,
+    /// Four half-precision floats (no Rust equiv). `vec4<f32>` in shaders.
+    Float16x4 = 17,
+    /// One single-precision float (f32). `f32` in shaders.
+    Float32 = 18,
+    /// Two single-precision floats (f32). `vec2<f32>` in shaders.
+    Float32x2 = 19,
+    /// Three single-precision floats (f32). `vec3<f32>` in shaders.
+    Float32x3 = 20,
+    /// Four single-precision floats (f32). `vec4<f32>` in shaders.
+    Float32x4 = 21,
+    /// One unsigned int (u32). `u32` in shaders.
+    Uint32 = 22,
+    /// Two unsigned ints (u32). `vec2<u32>` in shaders.
+    Uint32x2 = 23,
+    /// Three unsigned ints (u32). `vec3<u32>` in shaders.
+    Uint32x3 = 24,
+    /// Four unsigned ints (u32). `vec4<u32>` in shaders.
+    Uint32x4 = 25,
+    /// One signed int (i32). `i32` in shaders.
+    Sint32 = 26,
+    /// Two signed ints (i32). `vec2<i32>` in shaders.
+    Sint32x2 = 27,
+    /// Three signed ints (i32). `vec3<i32>` in shaders.
+    Sint32x3 = 28,
+    /// Four signed ints (i32). `vec4<i32>` in shaders.
+    Sint32x4 = 29,
+    /// Three unsigned 10-bit integers and one 2-bit integer, packed into a 32-bit integer (u32). [0, 1024] converted to float [0, 1] `vec4<f32>` in shaders.
+    #[cfg_attr(feature = "serde", serde(rename = "unorm10-10-10-2"))]
+    Unorm10_10_10_2 = 34,
+}
+
+/// A mapping of vertex buffers and their attributes to shader
+/// locations.
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+#[cfg_attr(feature = "serialize", derive(serde::Serialize))]
+#[cfg_attr(feature = "deserialize", derive(serde::Deserialize))]
+pub struct AttributeMapping {
+    /// Shader location associated with this attribute
+    pub shader_location: u32,
+    /// Offset in bytes from start of vertex buffer structure
+    pub offset: u32,
+    /// Format code to help us unpack the attribute into the type
+    /// used by the shader. Codes correspond to a 0-based index of
+    /// <https://gpuweb.github.io/gpuweb/#enumdef-gpuvertexformat>.
+    /// The conversion process is described by
+    /// <https://gpuweb.github.io/gpuweb/#vertex-processing>.
+    pub format: VertexFormat,
+}
+
+/// A description of a vertex buffer with all the information we
+/// need to address the attributes within it.
+#[derive(Debug, Default, Clone, PartialEq, Eq, Hash)]
+#[cfg_attr(feature = "serialize", derive(serde::Serialize))]
+#[cfg_attr(feature = "deserialize", derive(serde::Deserialize))]
+pub struct VertexBufferMapping {
+    /// Shader location associated with this buffer
+    pub id: u32,
+    /// Size of the structure in bytes
+    pub stride: u32,
+    /// True if the buffer is indexed by vertex, false if indexed
+    /// by instance.
+    pub indexed_by_vertex: bool,
+    /// Vec of the attributes within the structure
+    pub attributes: Vec<AttributeMapping>,
+}
+
 /// A subset of options that are meant to be changed per pipeline.
 #[derive(Debug, Default, Clone)]
 #[cfg_attr(feature = "serialize", derive(serde::Serialize))]
@@ -234,6 +341,17 @@ pub struct PipelineOptions {
     ///
     /// Enable this for vertex shaders with point primitive topologies.
     pub allow_and_force_point_size: bool,
+
+    /// If set, when generating the Metal vertex shader, transform it
+    /// to receive the vertex buffers, lengths, and vertex id as args,
+    /// and bounds-check the vertex id and use the index into the
+    /// vertex buffers to access attributes, rather than using Metal's
+    /// [[stage-in]] assembled attribute data.
+    pub vertex_pulling_transform: bool,
+
+    /// vertex_buffer_mappings are used during shader translation to
+    /// support vertex pulling.
+    pub vertex_buffer_mappings: Vec<VertexBufferMapping>,
 }
 
 impl Options {