Simple gizmo usage might trigger vulkan validation errors

[torsteingrindvik]

Bevy version

eb3c813

What you did

Gizmos might trigger vulkan validation errors.

Minimal:

use bevy::prelude::*;

fn main() {
    App::new()
        .add_plugins(DefaultPlugins)
        .add_systems(Startup, |mut commands: Commands| {
            commands.spawn(Camera3dBundle::default());
        })
        .add_systems(Update, |mut gizmos: Gizmos| {
            gizmos.arrow(Vec3::ZERO, Vec3::ONE, Color::default());
        })
        .run();
}

What went wrong

Validation spam, but not always, ~30% of the time I start the app I get spammed, other times not at all.

2024-06-15T12:12:38.018757Z  INFO bevy_diagnostic::system_information_diagnostics_plugin::internal: SystemInfo { os: "Linux rolling EndeavourOS", kernel: "6.9.3-arch1-1", cpu: "AMD Ryzen 9 3900X 12-Core Processor", core_count: "12", memory: "31.3 GiB" }
2024-06-15T12:12:38.230219Z  INFO bevy_render::renderer: AdapterInfo { name: "NVIDIA GeForce RTX 2080 Ti", vendor: 4318, device: 7687, device_type: DiscreteGpu, driver: "NVIDIA", driver_info: "555.52.04", backend: Vulkan }
2024-06-15T12:12:38.942473Z  INFO gilrs_core::platform::platform::gamepad: Gamepad /dev/input/event7 (Microsoft X-Box One S pad) connected.
2024-06-15T12:12:39.087499Z  INFO bevy_winit::system: Creating new window "App" (Entity { index: 0, generation: 1 })
2024-06-15T12:12:39.088085Z  INFO winit::platform_impl::platform::x11::window: Guessed window scale factor: 1.75
2024-06-15T12:12:39.097405Z  INFO bevy_input::gamepad: Gamepad { id: 0 } Connected
2024-06-15T12:12:39.300020Z  WARN bevy_render::view::window: Couldn't get swap chain texture. This often happens with the NVIDIA drivers on Linux. It can be safely ignored.
2024-06-15T12:12:39.590875Z ERROR wgpu_hal::vulkan::instance: VALIDATION [SYNC-HAZARD-READ-AFTER-WRITE (0xe4d96472)]
        Validation Error: [ SYNC-HAZARD-READ-AFTER-WRITE ] Object 0: handle = 0x5647b5cda570, type = VK_OBJECT_TYPE_QUEUE; | MessageID = 0xe4d96472 | vkQueueSubmit():  Hazard READ_AFTER_WRITE for entry 9, VkCommandBuffer 0x7c72e0039d20[main_transparent_pass_3d], Submitted
 access info (submitted_usage: SYNC_VERTEX_ATTRIBUTE_INPUT_VERTEX_ATTRIBUTE_READ, command: vkCmdDraw, seq_no: 2, reset_no: 8). Access info (prior_usage: SYNC_COPY_TRANSFER_WRITE, write_barriers: SYNC_FRAGMENT_SHADER_COLOR_ATTACHMENT_READ|SYNC_FRAGMENT_SHADER_DEPTH_STENCIL
_ATTACHMENT_READ|SYNC_FRAGMENT_SHADER_INPUT_ATTACHMENT_READ|SYNC_EARLY_FRAGMENT_TESTS_DEPTH_STENCIL_ATTACHMENT_READ|SYNC_EARLY_FRAGMENT_TESTS_DEPTH_STENCIL_ATTACHMENT_WRITE|SYNC_LATE_FRAGMENT_TESTS_DEPTH_STENCIL_ATTACHMENT_READ|SYNC_LATE_FRAGMENT_TESTS_DEPTH_STENCIL_ATTAC
HMENT_WRITE|SYNC_COLOR_ATTACHMENT_OUTPUT_COLOR_ATTACHMENT_READ|SYNC_COLOR_ATTACHMENT_OUTPUT_COLOR_ATTACHMENT_WRITE|SYNC_SUBPASS_SHADER_HUAWEI_INPUT_ATTACHMENT_READ, queue: VkQueue 0x5647b5cda570[], submit: 7, batch: 0, batch_tag: 276, command: vkCmdCopyBuffer, command_buf
fer: VkCommandBuffer 0x5647b5f1d4f0[(wgpu internal) PendingWrites], seq_no: 6, reset_no: 1).
2024-06-15T12:12:39.590956Z ERROR wgpu_hal::vulkan::instance:   objects: (type: QUEUE, hndl: 0x5647b5cda570, name: ?)
2024-06-15T12:12:39.626466Z ERROR wgpu_hal::vulkan::instance: VALIDATION [SYNC-HAZARD-READ-AFTER-WRITE (0xe4d96472)]
        Validation Error: [ SYNC-HAZARD-READ-AFTER-WRITE ] Object 0: handle = 0x5647b5cda570, type = VK_OBJECT_TYPE_QUEUE; | MessageID = 0xe4d96472 | vkQueueSubmit():  Hazard READ_AFTER_WRITE for entry 9, VkCommandBuffer 0x7c728456a7b0[main_transparent_pass_3d], Submitted
 access info (submitted_usage: SYNC_VERTEX_ATTRIBUTE_INPUT_VERTEX_ATTRIBUTE_READ, command: vkCmdDraw, seq_no: 2, reset_no: 11). Access info (prior_usage: SYNC_COPY_TRANSFER_WRITE, write_barriers: SYNC_FRAGMENT_SHADER_COLOR_ATTACHMENT_READ|SYNC_FRAGMENT_SHADER_DEPTH_STENCI
L_ATTACHMENT_READ|SYNC_FRAGMENT_SHADER_INPUT_ATTACHMENT_READ|SYNC_EARLY_FRAGMENT_TESTS_DEPTH_STENCIL_ATTACHMENT_READ|SYNC_EARLY_FRAGMENT_TESTS_DEPTH_STENCIL_ATTACHMENT_WRITE|SYNC_LATE_FRAGMENT_TESTS_DEPTH_STENCIL_ATTACHMENT_READ|SYNC_LATE_FRAGMENT_TESTS_DEPTH_STENCIL_ATTA
CHMENT_WRITE|SYNC_COLOR_ATTACHMENT_OUTPUT_COLOR_ATTACHMENT_READ|SYNC_COLOR_ATTACHMENT_OUTPUT_COLOR_ATTACHMENT_WRITE|SYNC_SUBPASS_SHADER_HUAWEI_INPUT_ATTACHMENT_READ, queue: VkQueue 0x5647b5cda570[], submit: 3, batch: 0, batch_tag: 184, command: vkCmdCopyBuffer, command_bu
ffer: VkCommandBuffer 0x5647b5f25790[(wgpu internal) PendingWrites], seq_no: 6, reset_no: 1).

Additional information

Output screenshot for readability:

[torsteingrindvik]

Never seen this before just now doing some work on main.
I'm suspecting the recent wgpu update is the cause: 2825ac8

[torsteingrindvik]

I confirmed locally that the wgpu upgrade is the first commit this happens on.

[robojeb]

This occurs for me on 0.14.0-rc.3 but only if there is another Non-gizmo mesh rendering.
If the scene consists of entirely gizmos (which I have for some math tests) it renders without validation errors.

[Azorlogh]

Happening for me too in 0.14.0-rc.3.
Linux/X11 + NVIDIA proprietary drivers
AdapterInfo { name: "NVIDIA GeForce RTX 2060", vendor: 4318, device: 7817, device_type: DiscreteGpu, driver: "NVIDIA", driver_info: "550.90.07", backend: Vulkan }

[torsteingrindvik]

I just tried an example in a 0.14.0-rc.4 context and I still get problems.

gfx-rs/wgpu#5231

This WGPU issue seems relevant

[torsteingrindvik]

https://discord.com/channels/691052431525675048/692572690833473578/1251537174181052436

Here's some discussion on discord about this issue from a couple of weeks ago.