Add support for gaps in bind group layouts & update relevant validation (gfx-rs#9034)

* [wgpu-core] add support for optional BG & BGL entries

* [wgpu] allow gaps in bind group layouts

And allow unbinding in the web backend.

* adapt tests

* add changelog entry

* [wgpu] remove generic from `PipelineLayoutDescriptor`

Author: teoxoy

CHANGELOG.md

@@ -44,6 +44,25 @@ Bottom level categories:
 
 ### Major Changes
 
+#### Bind group layouts now optional in `PipelineLayoutDescriptor`
+
+Allow gaps in bind group layouts and added full support for unbinding. As a result of this `PipelineLayoutDescriptor`'s `bind_group_layouts` field now has type of `&[Option<&BindGroupLayout>]`, making this a breaking change. To migrate wrap bind group layout references in `Some`:
+
+```diff
+  let pl_desc = wgpu::PipelineLayoutDescriptor {
+      label: None,
+      bind_group_layouts: &[
+-         &bind_group_layout
++         Some(&bind_group_layout)
+      ],
+      immediate_size: 0,
+  });
+```
+
+By @teoxoy in [#9034](https://github.com/gfx-rs/wgpu/pull/9034).
+
+#### MSRV update
+
 `wgpu` now has a new MSRV policy. This release has an MSRV of **1.87**. This is lower than v27's 1.88 and v28's 1.92. Going forward, we will only bump wgpu's MSRV if it has tangible benefits for the code, and we will never bump to an MSRV higher than `stable - 3`. So if stable is at 1.97 and 1.94 brought benefit to our code, we could bump it no higher than 1.94. As before, MSRV bumps will always be breaking changes. 
 
 By @cwfitzgerald in [#8999](https://github.com/gfx-rs/wgpu/pull/8999).

benches/benches/wgpu-benchmark/computepass.rs

@@ -250,7 +250,7 @@ impl ComputepassState {
                 .device
                 .create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
                     label: None,
-                    bind_group_layouts: &[&bind_group_layout],
+                    bind_group_layouts: &[Some(&bind_group_layout)],
                     immediate_size: 0,
                 });
 
@@ -345,7 +345,7 @@ impl ComputepassState {
                     .device
                     .create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
                         label: None,
-                        bind_group_layouts: &[&bindless_bind_group_layout],
+                        bind_group_layouts: &[Some(&bindless_bind_group_layout)],
                         immediate_size: 0,
                     });
 

benches/benches/wgpu-benchmark/renderpass.rs

@@ -150,7 +150,7 @@ impl RenderpassState {
                 .device
                 .create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
                     label: None,
-                    bind_group_layouts: &[&bind_group_layout],
+                    bind_group_layouts: &[Some(&bind_group_layout)],
                     immediate_size: 0,
                 });
 
@@ -286,7 +286,7 @@ impl RenderpassState {
                     .device
                     .create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
                         label: None,
-                        bind_group_layouts: &[&bindless_bind_group_layout],
+                        bind_group_layouts: &[Some(&bindless_bind_group_layout)],
                         immediate_size: 0,
                     });
 

cts_runner/fail.lst

@@ -27,7 +27,7 @@ webgpu:api,validation,createBindGroupLayout:max_resources_per_stage,in_pipeline_
 webgpu:api,validation,createBindGroupLayout:visibility,VERTEX_shader_stage_buffer_type:* // 25%, writable storage buffers not allowed in VERTEX
 webgpu:api,validation,createBindGroupLayout:visibility,VERTEX_shader_stage_storage_texture_access:* // 25%, write-access storage textures not allowed in VERTEX
 webgpu:api,validation,createBindGroupLayout:visibility:* // 25%, missing per-stage storage limits
-webgpu:api,validation,createPipelineLayout:* // 33%, https://github.com/gfx-rs/wgpu/issues/4738, and at least 1 more issue
+webgpu:api,validation,createPipelineLayout:number_of_dynamic_buffers_exceeds_the_maximum_value:* // limits issue
 webgpu:api,validation,createView:texture_state:* // 0%, https://github.com/gfx-rs/wgpu/issues/7881
 webgpu:api,validation,encoding,cmds,copyTextureToTexture:copy_ranges:* // ***, https://github.com/gfx-rs/wgpu/issues/8118
 webgpu:api,validation,encoding,cmds,debug:* // 92%, https://github.com/gfx-rs/wgpu/issues/8039
@@ -40,13 +40,8 @@ webgpu:api,validation,encoding,cmds,setBindGroup:u32array_start_and_length:* //
 webgpu:api,validation,encoding,cmds,setImmediates:* // 0%, feature not implemented
 webgpu:api,validation,encoding,createRenderBundleEncoder:* // 26%, empty attachments, format compatibility
 webgpu:api,validation,encoding,encoder_open_state:* // https://github.com/gfx-rs/wgpu/issues/7857
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:default_bind_group_layouts_never_match,compute_pass:* // 75%, empty bind group layouts compatibility
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:default_bind_group_layouts_never_match,render_pass:* // 75%, empty bind group layouts compatibility
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:empty_bind_group_layouts_never_requires_empty_bind_groups,compute_pass:* // 17%, null bind groups, https://github.com/gfx-rs/wgpu/issues/4738
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:empty_bind_group_layouts_never_requires_empty_bind_groups,render_pass:* // 17%, null bind groups, https://github.com/gfx-rs/wgpu/issues/4738
 webgpu:api,validation,encoding,queries,resolveQuerySet:* // 93%, https://github.com/gfx-rs/wgpu/issues/7881
 webgpu:api,validation,encoding,render_bundle:* // 81%, readonly flag normalization mismatch
-webgpu:api,validation,getBindGroupLayout:* // 29%, rejects valid indices beyond defined layouts
 webgpu:api,validation,image_copy,buffer_texture_copies:* // https://github.com/gfx-rs/wgpu/issues/7946
 webgpu:api,validation,layout_shader_compat:pipeline_layout_shader_exact_match:* // 99%, storage texture read-write bindings not compatible with write-only shaders
 webgpu:api,validation,non_filterable_texture:non_filterable_texture_with_filtering_sampler:* // 80%, depth textures with filtering samplers

cts_runner/test.lst

@@ -107,6 +107,11 @@ webgpu:api,validation,createBindGroupLayout:maximum_binding_limit:*
 webgpu:api,validation,createBindGroupLayout:multisampled_validation:*
 webgpu:api,validation,createBindGroupLayout:storage_texture,formats:*
 webgpu:api,validation,createBindGroupLayout:storage_texture,layout_dimension:*
+webgpu:api,validation,createPipelineLayout:bind_group_layouts,create_pipeline_with_null_bind_group_layouts:*
+webgpu:api,validation,createPipelineLayout:bind_group_layouts,device_mismatch:*
+webgpu:api,validation,createPipelineLayout:bind_group_layouts,null_bind_group_layouts:*
+webgpu:api,validation,createPipelineLayout:bind_group_layouts,set_pipeline_with_null_bind_group_layouts:*
+webgpu:api,validation,createPipelineLayout:number_of_bind_group_layouts_exceeds_the_maximum_value:*
 webgpu:api,validation,createSampler:*
 webgpu:api,validation,createTexture:*
 webgpu:api,validation,createView:array_layers:*
@@ -169,15 +174,11 @@ webgpu:api,validation,encoding,encoder_open_state:non_pass_commands:*
 // https://github.com/gfx-rs/wgpu/issues/7857
 webgpu:api,validation,encoding,encoder_open_state:render_pass_commands:*
 webgpu:api,validation,encoding,encoder_state:*
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:bgl_binding_mismatch:*
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:bgl_resource_type_mismatch:*
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:bgl_visibility_mismatch:*
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:bind_groups_and_pipeline_layout_mismatch:*
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:buffer_binding,render_pipeline:*
-webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:sampler_binding,render_pipeline:*
+webgpu:api,validation,encoding,programmable,pipeline_bind_group_compat:*
 webgpu:api,validation,encoding,queries,begin_end:occlusion_query,*
 webgpu:api,validation,encoding,queries,general:occlusion_query,*
 webgpu:api,validation,error_scope:*
+webgpu:api,validation,getBindGroupLayout:*
 webgpu:api,validation,image_copy,buffer_related:*
 // image_copy depth/stencil failures on dx12: https://github.com/gfx-rs/wgpu/issues/8133
 fails-if(dx12) webgpu:api,validation,image_copy,buffer_texture_copies:depth_stencil_format,copy_buffer_offset:format="depth24plus-stencil8";aspect="stencil-only";copyType="CopyB2T"

deno_webgpu/device.rs

@@ -379,7 +379,11 @@ impl GPUDevice {
     let bind_group_layouts = descriptor
       .bind_group_layouts
       .into_iter()
-      .map(|bind_group_layout| bind_group_layout.id)
+      .map(|bind_group_layout| {
+        bind_group_layout
+          .into_option()
+          .map(|bind_group_layout| bind_group_layout.id)
+      })
       .collect();
 
     let wgpu_descriptor = wgpu_core::binding_model::PipelineLayoutDescriptor {

deno_webgpu/pipeline_layout.rs

@@ -2,6 +2,7 @@
 
 use deno_core::cppgc::Ptr;
 use deno_core::op2;
+use deno_core::webidl::Nullable;
 use deno_core::webidl::WebIdlInterfaceConverter;
 use deno_core::GarbageCollected;
 use deno_core::WebIDL;
@@ -58,5 +59,5 @@ pub(crate) struct GPUPipelineLayoutDescriptor {
   pub label: String,
 
   pub bind_group_layouts:
-    Vec<Ptr<super::bind_group_layout::GPUBindGroupLayout>>,
+    Vec<Nullable<Ptr<super::bind_group_layout::GPUBindGroupLayout>>>,
 }

examples/features/src/big_compute_buffers/mod.rs

@@ -126,7 +126,7 @@ fn setup_pipeline(
 ) -> wgpu::ComputePipeline {
     let pipeline_layout = device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
         label: Some("Compute Pipeline Layout"),
-        bind_group_layouts: &[&bind_group_layout],
+        bind_group_layouts: &[Some(&bind_group_layout)],
         immediate_size: 0,
     });
 

examples/features/src/boids/mod.rs

@@ -106,7 +106,7 @@ impl crate::framework::Example for Example {
         let compute_pipeline_layout =
             device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
                 label: Some("compute"),
-                bind_group_layouts: &[&compute_bind_group_layout],
+                bind_group_layouts: &[Some(&compute_bind_group_layout)],
                 immediate_size: 0,
             });
 

examples/features/src/bunnymark/mod.rs

@@ -200,7 +200,10 @@ impl crate::framework::Example for Example {
             });
         let pipeline_layout = device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
             label: None,
-            bind_group_layouts: &[&global_bind_group_layout, &local_bind_group_layout],
+            bind_group_layouts: &[
+                Some(&global_bind_group_layout),
+                Some(&local_bind_group_layout),
+            ],
             immediate_size: 0,
         });