Closed marc0246 closed 1 month ago
cc @gfxstrand
I think this should be fine with the Mesa implementations. If @cubanismo can confirm it's okay on NVIDIA, we can make this change.
Thanks for the @. I've asked the engineer that implemented this to check.
We expect this will work fine on our implementation as-is, so fine with us to make the change.
Here's a the validation change: https://github.com/KhronosGroup/Vulkan-ValidationLayers/pull/7888 I'll make a spec MR and run that through the internal process.
This should be fixed in the 1.3.285 spec update.
VUID-VkMemoryMapInfoKHR-flags-09572
andVUID-VkMemoryMapInfoKHR-flags-09574
only acceptVK_WHOLE_SIZE
and not also the rest of the allocation size like the rest of the APIs that can takeVK_WHOLE_SIZE
. For example, without using theVK_EXT_map_memory_placed
extension, one can specify the rest of the allocation orVK_WHOLE_SIZE
. Same applies toVkMappedMemoryRange
, you can use both the rest of the allocation size orVK_WHOLE_SIZE
. It feels weird to have this special case only when using this extension. It means that, since we never used and don't intend on usingVK_WHOLE_SIZE
, we are currently special casing this such that the rest of the allocation size is translated toVK_WHOLE_SIZE
, only when using this extension. That feels rather strange, and I think it would be better if the rest of the allocation size was also permitted natively.