Vulkan presentation validation errors

[JMS55]

Description
Bevy is seeing the following VK validation errors.

Repro steps
cargo run --example solari --features=... in the bevy repo (cargo will tell you what features you need)

Extra materials

        vkQueuePresentKHR(): pPresentInfo->pSwapchains[0] images passed to present must be in layout VK_IMAGE_LAYOUT_PRESENT_SRC_KHR or VK_IMAGE_LAYOUT_SHARED_PRESENT_KHR but VkImage 0x2190000000219 is in VK_IMAGE_LAYOUT_UNDEFINED.
The Vulkan spec states: Each element of pImageIndices must be the index of a presentable image acquired from the swapchain specified by the corresponding element of the pSwapchains array, and the presented image subresource must be in the VK_IMAGE_LAYOUT_PRESENT_SRC_KHR or VK_IMAGE_LAYOUT_SHARED_PRESENT_KHR layout at the time the operation is executed on a VkDevice (https://vulkan.lunarg.com/doc/view/1.4.328.1/windows/antora/spec/latest/chapters/VK_KHR_surface/wsi.html#VUID-VkPresentInfoKHR-pImageIndices-01430)
2026-03-11T18:15:08.472997Z ERROR wgpu_hal::vulkan::instance:   objects: (type: IMAGE, hndl: 0x2190000000219, name: ?)
2026-03-11T18:15:08.473341Z ERROR wgpu_hal::vulkan::instance: VALIDATION [VUID-VkPresentInfoKHR-pImageIndices-01430 (0x48ad24c6)]
        vkQueuePresentKHR(): pPresentInfo->pSwapchains[0] images passed to present must be in layout VK_IMAGE_LAYOUT_PRESENT_SRC_KHR or VK_IMAGE_LAYOUT_SHARED_PRESENT_KHR but VkImage 0x21a000000021a is in VK_IMAGE_LAYOUT_UNDEFINED.
The Vulkan spec states: Each element of pImageIndices must be the index of a presentable image acquired from the swapchain specified by the corresponding element of the pSwapchains array, and the presented image subresource must be in the VK_IMAGE_LAYOUT_PRESENT_SRC_KHR or VK_IMAGE_LAYOUT_SHARED_PRESENT_KHR layout at the time the operation is executed on a VkDevice (https://vulkan.lunarg.com/doc/view/1.4.328.1/windows/antora/spec/latest/chapters/VK_KHR_surface/wsi.html#VUID-VkPresentInfoKHR-pImageIndices-01430)
2026-03-11T18:15:08.473548Z ERROR wgpu_hal::vulkan::instance:   objects: (type: IMAGE, hndl: 0x21a000000021a, name: ?)    
2026-03-11T18:15:08.488725Z ERROR wgpu_hal::vulkan::instance: VALIDATION [VUID-vkAcquireNextImageKHR-semaphore-01286 (0xe9e4b2a9)]
        vkAcquireNextImageKHR(): Semaphore must not be currently signaled.
The Vulkan spec states: If semaphore is not VK_NULL_HANDLE, it must be unsignaled (https://vulkan.lunarg.com/doc/view/1.4.328.1/windows/antora/spec/latest/chapters/VK_KHR_surface/wsi.html#VUID-vkAcquireNextImageKHR-semaphore-01286)
2026-03-11T18:15:08.488929Z ERROR wgpu_hal::vulkan::instance:   objects: (type: SEMAPHORE, hndl: 0x21d000000021d, name: SwapchainImageSemaphore: Index 0 acquire)
2026-03-11T18:15:08.505208Z ERROR wgpu_hal::vulkan::instance: VALIDATION [VUID-vkAcquireNextImageKHR-semaphore-01286 (0xe9e4b2a9)]
        vkAcquireNextImageKHR(): Semaphore must not be currently signaled.
The Vulkan spec states: If semaphore is not VK_NULL_HANDLE, it must be unsignaled (https://vulkan.lunarg.com/doc/view/1.4.328.1/windows/antora/spec/latest/chapters/VK_KHR_surface/wsi.html#VUID-vkAcquireNextImageKHR-semaphore-01286)
2026-03-11T18:15:08.505455Z ERROR wgpu_hal::vulkan::instance:   objects: (type: SEMAPHORE, hndl: 0x21e000000021e, name: SwapchainImageSemaphore: Index 1 acquire)

Platform
Windows 11, wgpu 28

[inner-daemons]

@JMS55 What are the conditions under which this occurs? Are you not rendering anything to the texture, or are you resizing the window when this occurs?

[JMS55]

Not really sure, but iirc it only happens for the first few frames, so probably rendering nothing(?) while waiting for pipelines to compile? Idk really.

[inner-daemons]

@JMS55 Ok, can you try with #9222? That should fix errors with presenting after no work is done (there's even a manual regression test).

[inner-daemons]

Also just to note, this should be fixable if you manually clear the screen when nothing is being rendered. That should be the effect anyway, though currently the framebuffer will be completely uninitialized.

[beicause]

Is this bevyengine/bevy#22733? Not sure if this is bevy bug or wgpu bug.

[inner-daemons]

@beicause Yup, it should be the same, and yup, its us. Wgpu is doing incorrect things given correct input.

[ChristopherBiscardi]

@JMS55 Ok, can you try with #9222? That should fix errors with presenting after no work is done (there's even a manual regression test).

I tried this on the wgpu 29 branch. It seems to clear up the present issue

There's another validation that still exists. Not sure its relevant though since a naga-oil atomics test is currently failing on this branch and it seems possibly related.

the other validation

2026-03-15T07:17:09.298276Z ERROR wgpu_hal::vulkan::instance: VALIDATION [VUID-RuntimeSpirv-vulkanMemoryModel-06265 (0x7ab8899)]
        vkCreateShaderModule(): SPIR-V uses Device memory scope, but the vulkanMemoryModelDeviceScope feature was not enabled.
%420 = OpAtomicIAdd %9 %421 %422 %41 %42
The Vulkan spec states: If the vulkanMemoryModel feature is enabled and the vulkanMemoryModelDeviceScope feature is not enabled, Device memory scope must not be used (https://vulkan.lunarg.com/doc/view/1.4.328.1/windows/antora/spec/latest/appendices/spirvenv.html#VUID-RuntimeSpirv-vulkanMemoryModel-06265)

[inner-daemons]

@ChristopherBiscardi Looks different but worth its own issue. Are you by any chance using cooperative matrices there btw?

[ChristopherBiscardi]

@inner-daemons I'll file an issue if it still persists after I fix the naga-oil test. I don't know the answer to the coop matrices question. Its the solari functionality in bevy, so I'm not familiar enough to say definitively. @JMS55 would know better

[JMS55]

No co-op matrices used