Vulkan renderer segfault on window resize - nvidia 440.44 (SDL+bgfx) [linux]

[TheBeardOfTruth]

bgfx will throw some pipeline validation errors/warnings/notices/whatever.

Resizing the window throws further errors and segfaults the application. I'd assume BGFX's Vulkan implementation doesn't want to play nice with SDL2, though it's a relatively thin wrapper over native x11 windowing so I don't know, this would seem a fairly common use-case however (since cross-platform windowing is pretty tedious)

../../../src/bgfx.cpp (3400): BGFX Init complete.
../../../src/renderer_vk.cpp (2789): BGFX present mode: 0
../../../src/renderer_vk.cpp (2822): BGFX refreshed swapchain: 1920 x 1080
../../../src/renderer_vk.cpp (544): BGFX ---E- Pipeline, Validation, 0: [ UNASSIGNED-CoreValidation-DrawState-NoActiveRenderpass ] Object: 0x520000000052 (Type = 19) | No active render pass found at draw-time in VkPipeline 0x520000000052[]!
../../../src/renderer_vk.cpp (544): BGFX ---E- CommandBuffer, Validation, 0: [ VUID-vkCmdDrawIndexed-renderpass ] Object: 0x7f775897edb8 (Type = 6) | vkCmdDrawIndexed(): This call must be issued inside an active render pass. The Vulkan spec states: This command must only be called inside of a render pass instance (https://www.khronos.org/registry/vulkan/specs/1.1-extensions/html/vkspec.html#VUID-vkCmdDrawIndexed-renderpass)
../../../src/renderer_vk.cpp (544): BGFX ---E- SwapchainKHR, Validation, 0: [ VUID-vkAcquireNextImageKHR-swapchain-01802 ] Object: 0x470000000047 (Type = 27) | vkAcquireNextImageKHR: Application has already previously acquired 1 image from swapchain. Only 1 is available to be acquired using a timeout of UINT64_MAX (given the swapchain has 2, and VkSurfaceCapabilitiesKHR::minImageCount is 2). The Vulkan spec states: If the number of currently acquired images is greater than the difference between the number of images in swapchain and the value of VkSurfaceCapabilitiesKHR::minImageCount as returned by a call to vkGetPhysicalDeviceSurfaceCapabilities2KHR with the surface used to create swapchain, timeout must not be UINT64_MAX (https://www.khronos.org/registry/vulkan/specs/1.1-extensions/html/vkspec.html#VUID-vkAcquireNextImageKHR-swapchain-01802)
../../../src/renderer_vk.cpp (544): BGFX ---E- Semaphore, Validation, 0: [ UNASSIGNED-CoreValidation-DrawState-QueueForwardProgress ] Object: 0xc000000000c (Type = 5) | VkQueue 0x7f7758253ea8[] is waiting on VkSemaphore 0xc000000000c[] that has no way to be signaled.
../../../src/renderer_vk.cpp (4018): BGFX CHECK vkQueueSubmit(m_queueGraphics, 1, &si, 0); VK error 0xc4650b07: VK_ERROR_VALIDATION_FAILED_EXT
[1] 415567 trace trap (core dumped) ./bin/a.out --vulkan

The relevant code snippet (SDL2+bgfx, basic boilerplate -- works fine with OpenGL 2.1 and 4.6)

Relevant code:

SDL_Event event;
while(!quit)
{
	bgfx::setViewRect(0,0,0, uint16_t(width), uint16_t(height));
	bgfx::touch(0);
	bgfx::frame();
	while(SDL_PollEvent(&event))
	{
		switch (event.type)
		{
			case SDL_WINDOWEVENT:
			{
				const SDL_WindowEvent& wev = event.window;
				switch (wev.event)
				{
					//we crash here, with and without the bgfx::reset() call
					//framebuffer bugs with reset, no bugs without but still crash
					//If we want to be strict it's technically just the window size changing
					//i.e even if we don't check for the event we still crash, 
					//which means we never get the chance to call bgfx::reset()
					//before we crash, since we can't call it without knowing our new size
					case SDL_WINDOWEVENT_SIZE_CHANGED:
						SDL_GetWindowSize(window, &width, &height);
						bgfx::reset(width, height, BGFX_RESET_MSAA_X16 | BGFX_RESET_MAXANISOTROPY);
						break;
				}
			}
		} 
		break;
	}
}```

EDIT: typo

[Hexlord]

Having the same issue on more up-to-date bgfx. SDL2 + vulkan + bgfx seems to not support window resize well.

[Hexlord]

Found a workaround: call SDL_SetWindowResizable(false) from time to time to enforce vulkan rendering so that fences get released and not just hang.

[bkaradzic]

@Hexlord Can you reproduce the issue in provided examples?