Details
-
Bug
-
Resolution: Fixed
-
P2: Important
-
None
-
6.8
-
None
Description
The easiest way to reproduce is to run the lights example on vulkan with QSG_RHI_DEBUG_LAYER=1 and activate shadows casting/receiving bounds from the DebugView.
Then errors like this will be spammed:
vkDebug: Validation Error: [ VUID-vkCmdDrawIndexed-multisampledRenderToSingleSampled-07284 ] Object 0: handle = 0x7fffd0f1a900, type = VK_OBJECT_TYPE_COMMAND_BUFFER; Object 1: handle = 0xb262000000001461, type = VK_OBJECT_TYPE_PIPELINE; Object 2: handle = 0xb9181f0000000029, type = VK_OBJECT_TYPE_RENDER_PASS; | MessageID = 0x576454a4 | vkCmdDrawIndexed(): In VkPipeline 0xb262000000001461[] the sample count is VK_SAMPLE_COUNT_1_BIT while the current VkRenderPass 0xb9181f0000000029[] has VK_SAMPLE_COUNT_4_BIT and they need to be the same. The Vulkan spec states: If rasterization is not disabled in the bound graphics pipeline, and none of the following is enabled: the VK_AMD_mixed_attachment_samples extension the VK_NV_framebuffer_mixed_samples extension the multisampledRenderToSingleSampled feature then rasterizationSamples for the currently bound graphics pipeline must be the same as the current subpass color and/or depth/stencil attachments (https://www.khronos.org/registry/vulkan/specs/1.3-extensions/html/vkspec.html#VUID-vkCmdDrawIndexed-multisampledRenderToSingleSampled-07284)
Might be similar to https://bugreports.qt.io/browse/QTBUG-122890