vmaMapMemory triggers validation layer warning on Intel HD 620 about vulkanmemoryallocator

[victorbush]

You nailed it - I used VMA_ALLOCATION_CREATE_DEDICATED_MEMORY_BIT when allocating my depth buffer images and no longer got the validation warning. Thank you!

[adam-sawicki-a]

I can imagine how this happened. Long story short: I think it's neither bug in VMA nor in driver, or Vulkan validation layer. It's just a warning that can be safely ignored.

Full story:

Intel GPU has unified memory, with just one memory heap and two memory types. Objects so different as depth-stencil image GPU_ONLY and staging buffer CPU_ONLY end up in the same memory type, and possibly in the same VkDeviceMemory block.

Now comes the mapping. You want to map just your buffer, but Vulkan (as opposite to DX12) doesn't allow to map same memory block twice, so VMA always maps entire VkDeviceMemory block, just in case you want to map different object inside of it simultaneously. Validation layer detects that there is an image with LAYOUT_DEPTH_STENCIL_ATTACHMENT_OPTIMAL inside of it and issues a warning.

The warning says can result in undefined behavior if this memory is used by the device - but it won't be, because you really just access your buffer with mapped pointer. So the warning can be safely ignored.

I added this warning to ignores and documented it, in "development" branch.

Alternatively, if you want to get rid of this warning, you can try to create your depth-stencil images with VMA_ALLOCATION_CREATE_DEDICATED_MEMORY_BIT flag (which is recommended anyway), or you can create custom VmaPool to place your buffers which you want to map separate from GPU-only resources.

[devshgraphicsprogramming]

I thought normally in Vulkan you create the image and map it with a general or preinitialized layout and then transition it to the *_OPTIMAL layout later either via explicit transition or implicit transition (layouts at start and end in renderpasses).

…

On 17/08/18 19:45, victorbush wrote: When mapping a buffer with |vmaMapMemory|, a validation layer warning is triggered: |validation layer: Object: 0xb (Type = 8) | Mapping an image with layout VK_IMAGE_LAYOUT_DEPTH_STENCIL_ATTACHMENT_OPTIMAL can result in undefined behavior if this memory is used by the device. Only GENERAL or PREINITIALIZED should be used. | The buffer is created with |vmaCreateBuffer|: |VkBufferCreateInfo bufferInfo = {}; bufferInfo.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO; bufferInfo.size = size; bufferInfo.usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT; bufferInfo.sharingMode = VK_SHARING_MODE_EXCLUSIVE; VmaAllocationCreateInfo allocInfo = {}; allocInfo.usage = VMA_MEMORY_USAGE_CPU_ONLY; auto result = vmaCreateBuffer(_allocator, &bufferInfo, &allocInfo, &buffer, &allocation, nullptr); | Then I map the memory with |vmaMapMemory|: |void* data; vmaMapMemory(_device.GetAllocator(), allocation, &data); memcpy(data, RenderMesh::GetVertices().data(), (size_t)bufferSize); vmaUnmapMemory(_device.GetAllocator(), allocation); | When |vmaMapMemory| is called, the validation layer warning is triggered 3 times. The call to |vmaMapMemory| succeeds and works as expected, it just triggers validation warnings every time its called. When I use |vkCreateBuffer|, |vkAllocateMemory|, and |vkMapMemory| to handle the buffer creation and memory mapping instead of the vma library, the warning is not triggered. This happens on a Surface Laptop with Intel HD Graphics 620. Windows 10 Pro x64 Driver version: 24.20.100.6136 This does not happen on my discrete NVIDIA GPU, so probably related to the Intel HD 620? I'm still new to Vulkan, so I apologize if I missed something. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#34>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGky8fXgyxTf-PerQ-WI9hI9wzncSl-rks5uRw9lgaJpZM4WCB34>.

[victorbush]

The strange thing is I'm not mapping an image, it's just a buffer for vertex data (VkBuffer as opposed to VkImage). So I was confused why the validation layer was talking about mapping an image.

[devshgraphicsprogramming]

Then validation layer has a bug. But again these are Intel drivers on Windows :')

…

On 17/08/18 20:31, victorbush wrote: The strange thing is I'm not mapping an image, it's just a buffer for vertex data (VkBuffer as opposed to VkImage). So I was confused why the validation layer was talking about mapping an image. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#34 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGky8aNTuSY42xaatIbiIiqq1yrdvfn7ks5uRxn8gaJpZM4WCB34>.

[victorbush]

Ok thanks. I also found it happens with the sample exe provided with the vma library as well, so its not limited to my program. So potentially a driver issue.