shader_prepass example crashes with opaque CustomMaterial

[TesseractCat]

Bevy version

release-0.10.0 (also get the same crash on main).

Relevant system information

`AdapterInfo { name: "NVIDIA GeForce RTX 2060 with Max-Q Design", vendor: 4318, device: 7954, device_type: DiscreteGpu, driver: "NVIDIA", driver_info: "526.86", backend: Vulkan }`

What you did

I modified the shader_prepass example so that the CustomMaterial was AlphaMode::Opaque rather than AlphaMode::Blend.

diff --git a/examples/shader/shader_prepass.rs b/examples/shader/shader_prepass.rs
index 688152c5..ae85d681 100644
--- a/examples/shader/shader_prepass.rs
+++ b/examples/shader/shader_prepass.rs
@@ -109,7 +109,7 @@ fn setup(
         material: materials.add(CustomMaterial {
             color: Color::WHITE,
             color_texture: Some(asset_server.load("branding/icon.png")),
-            alpha_mode: AlphaMode::Blend,
+            alpha_mode: AlphaMode::Opaque,
         }),
         transform: Transform::from_xyz(1.0, 0.5, 0.0),
         ..default()

What went wrong

Bevy crashes. With other testing it appears the issue only happens when the camera is looking at the object with the custom material.

Additional information

2023-03-31T07:35:36.907647Z  INFO bevy_winit::system: Creating new window "Bevy App" (0v0)
2023-03-31T07:35:39.202777Z  INFO bevy_render::renderer: AdapterInfo { name: "NVIDIA GeForce RTX 2060 with Max-Q Design", vendor: 4318, device: 7954, device_type: DiscreteGpu, driver: "NVIDIA", driver_info: "526.86", backend: Vulkan }
2023-03-31T07:35:40.722993Z  INFO bevy_diagnostic::system_information_diagnostics_plugin::internal: SystemInfo { os: "Windows 10 Home", kernel: "18363", cpu: "AMD Ryzen 9 4900HS with Radeon Graphics", core_count: "8", memory: "15.4 GiB" }
2023-03-31T07:35:41.512883Z  INFO naga::back::spv::writer: Skip function Some("calculate_view")
2023-03-31T07:35:41.553574Z ERROR wgpu::backend::direct: Handling wgpu errors as fatal by default
thread 'Compute Task Pool (7)' panicked at 'wgpu error: Validation Error

Caused by:
    In Device::create_render_pipeline
      note: label = `prepass_pipeline`
    error matching FRAGMENT shader requirements against the pipeline
    location[0] is provided by the previous stage output but is not consumed as input by this stage.

', C:\Users\Tesse\scoop\persist\rustup\.cargo\registry\src\github.com-1ecc6299db9ec823\wgpu-0.15.1\src\backend\direct.rs:3024:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
thread 'Compute Task Pool (7)' panicked at 'A system has panicked so the executor cannot continue.: RecvError', crates\bevy_ecs\src\schedule\executor\multi_threaded.rs:194:60
thread '<unnamed>' panicked at 'called `Option::unwrap()` on a `None` value', C:\Users\Tesse\Desktop\Bevy Games\bevy\crates\bevy_tasks\src\task_pool.rs:376:49
thread 'Compute Task Pool (7)' panicked at 'called `Result::unwrap()` on an `Err` value: RecvError', crates\bevy_render\src\pipelined_rendering.rs:136:45
thread 'main' panicked at 'called `Option::unwrap()` on a `None` value', C:\Users\Tesse\Desktop\Bevy Games\bevy\crates\bevy_tasks\src\task_pool.rs:376:49

[TesseractCat]

I found this interaction on the discord: https://discord.com/channels/691052431525675048/866787577687310356/1077982198331867166. Not sure why this is necessary though.

[TesseractCat]

Okay, I'm pretty sure the issue is that the default vertex prepass shader isn't consuming location[0]

bevy/crates/bevy_pbr/src/prepass/prepass.wgsl

Lines 86 to 95 in f0f5d79

	struct FragmentInput {
	#ifdef NORMAL_PREPASS
	@location(1) world_normal: vec3<f32>,
	#endif // NORMAL_PREPASS
	#ifdef MOTION_VECTOR_PREPASS
	@location(3) world_position: vec4<f32>,
	@location(4) previous_world_position: vec4<f32>,
	#endif // MOTION_VECTOR_PREPASS
	}

, which is why this happens with custom materials. The pbr_prepass shader does consume it though,

bevy/crates/bevy_pbr/src/render/pbr_prepass.wgsl

Lines 7 to 12 in f0f5d79

	struct FragmentInput {
	@builtin(front_facing) is_front: bool,
	@builtin(position) frag_coord: vec4<f32>,
	#ifdef VERTEX_UVS
	@location(0) uv: vec2<f32>,
	#endif // VERTEX_UVS

. I assume the UV location isn't defined when AlphaMode is blend, which causes this problem only with opaque materials. Not sure if this is intentional.

[josfeenstra]

Thanks for the fix!