[PATCH 0/1] MR3033: d3d11/tests: Use more aligned offsets in test_clear_buffer_unordered_access_view().

[Zebediah Figura (＠zfigura)]

Creating a view with an unaligned offset is legal in d3d and apparently works, both on Windows and Linux, but may violate the Vulkan specification:

VUID-VkBufferViewCreateInfo-offset-00926(ERROR / SPEC): msgNum: -833749292 - Validation Error: [ VUID-VkBufferViewCreateInfo-offset-00926 ] Object 0: handle = 0x3c000000003c, type = VK_OBJECT_TYPE_BUFFER; | MessageID = 0xce4dfed4 | vkCreateBufferView(): VkBufferViewCreateInfo offset (36) must be a multiple of VkPhysicalDeviceLimits::minTexelBufferOffsetAlignment (16). The Vulkan spec states: offset must be a multiple of VkPhysicalDeviceLimits::minTexelBufferOffsetAlignment (https://www.khronos.org/registry/vulkan/specs/1.2-khr-extensions/html/vkspec...)

Without any reason to test unaligned buffer views here, just resolve this by changing the test.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/3033

[PATCH 1/1] d3d11/tests: Use more aligned offsets in test_clear_buffer_unordered_access_view().

[Zebediah Figura]

From: Zebediah Figura <zfigura(a)codeweavers.com>

Creating a view with an unaligned offset is legal in d3d and apparently works, both on Windows and Linux, but may violate the Vulkan specification:

VUID-VkBufferViewCreateInfo-offset-00926(ERROR / SPEC): msgNum: -833749292 - Validation Error: [ VUID-VkBufferViewCreateInfo-offset-00926 ] Object 0: handle = 0x3c000000003c, type = VK_OBJECT_TYPE_BUFFER; | MessageID = 0xce4dfed4 | vkCreateBufferView(): VkBufferViewCreateInfo offset (36) must be a multiple of VkPhysicalDeviceLimits::minTexelBufferOffsetAlignment (16). The Vulkan spec states: offset must be a multiple of VkPhysicalDeviceLimits::minTexelBufferOffsetAlignment (https://www.khronos.org/registry/vulkan/specs/1.2-khr-extensions/html/vkspec...)

Without any reason to test unaligned buffer views here, just resolve this by changing the test.
---
 dlls/d3d11/tests/d3d11.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/dlls/d3d11/tests/d3d11.c b/dlls/d3d11/tests/d3d11.c
index f6623614583..c54fa67e5b1 100644
--- a/dlls/d3d11/tests/d3d11.c
+++ b/dlls/d3d11/tests/d3d11.c
@@ -16638,8 +16638,8 @@ static void test_clear_buffer_unordered_access_view(void)
     U(uav_desc).Buffer.Flags = 0;
     hr = ID3D11Device_CreateUnorderedAccessView(device, (ID3D11Resource *)buffer, &uav_desc, &uav);
     ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr);
-    U(uav_desc).Buffer.FirstElement = 9;
-    U(uav_desc).Buffer.NumElements = 7;
+    U(uav_desc).Buffer.FirstElement = 8;
+    U(uav_desc).Buffer.NumElements = 8;
     hr = ID3D11Device_CreateUnorderedAccessView(device, (ID3D11Resource *)buffer, &uav_desc, &uav2);
     ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr);
 
-- 
GitLab

https://gitlab.winehq.org/wine/wine/-/merge_requests/3033

[Henri Verbeet (＠hverbeet)]

> Without any reason to test unaligned buffer views here, just resolve this by changing the test.

I think this was deliberately testing unaligned views because it seemed like an interesting case. It probably makes sense to just drop the test though.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/3033#note_35437

[Zebediah Figura (＠zfigura)]

On Mon Jun 12 17:46:39 2023 +0000, Henri Verbeet wrote:
> > Without any reason to test unaligned buffer views here, just resolve
> > this by changing the test.
> > I think this was deliberately testing unaligned views because it seemed
> > like an interesting case. It probably makes sense to just drop the test though.
> > I figured as much, but yeah, it doesn't seem interesting enough to justify violating the spec without any evidence that programs care.

I figure we might as well keep the partial clear test, though.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/3033#note_35492

[Chip Davis (＠cdavis5e)]

> Without any reason to test unaligned buffer views here, just resolve this by changing the test.

Or, you could enable `VK_EXT_texel_buffer_alignment`, or use a 1.3 device, where the functionality is always enabled--then, in [`VkPhysicalDeviceTexelBufferAlignmentProperties`](https://registry.khronos.org/vulkan/specs/1.3-extensions/html/chap46.html#Vk...), if `storageTexelBufferOffsetSingleTexelAlignment` is `VK_TRUE`, which it should be on any GPU that works with D3D, this *won't* violate the spec. (But then what do you do about devices that don't support the extension and/or don't support single-texel alignment...?)

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/3033#note_35578

[PATCH 0/1] MR3033: d3d11/tests: Use more aligned offsets in test_clear_buffer_unordered_access_view(). - approved

[Jan Sikorski (＠jsikorski)]

This merge request was approved by Jan Sikorski.
-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/3033