Bump wgpu-native to v29.0.1.1

README.md

@@ -120,7 +120,7 @@ This code is distributed under the 2-clause BSD license.
 * Install devtools using `pip install -e .[dev]`.
 * Using `pip install -e .` will also download the upstream wgpu-native
   binaries.
-  * You can use `python tools/download_wgpu_native.py` when needed.
+  * You can use `python tools/download_wgpu_native.py` when needed. And run `python codegen` once to obtain the combined headerfile.
   * Or point the `WGPU_LIB_PATH` environment variable to a custom build of `wgpu-native`.
 * Use `ruff format` to apply autoformatting.
 * Use `ruff check` to check for linting errors.

codegen/wgpu_native_patcher.py

@@ -95,7 +95,7 @@ def write_mappings():
         if native_name in hp.enums:
             for key, val in hp.enums[native_name].items():
                 key = key.lower()
-                if key in hp_enum:
+                if key in hp_enum and val != hp_enum.get(key):
                     print(f"Enum field {name}.{key} overridden by wgpu.h")
                     hp_enum[key] = val
         for ikey in idl.enums[name].values():

tests/test_wgpu_native_texture.py

@@ -5,7 +5,7 @@
 import numpy as np
 
 from testutils import run_tests, can_use_wgpu_lib, iters_equal
-from pytest import mark, raises
+from pytest import mark, raises, skip
 
 
 @mark.skipif(not can_use_wgpu_lib, reason="Needs wgpu lib")
@@ -281,5 +281,48 @@ def test_write_texture2():
     assert iters_equal(data0, data2)
 
 
+def test_16bit_norm():
+    """
+    Test creating, writing and making a view wiht a 16bit norm texture formats, if the native only feature is available.
+    """
+    # can't use the default device because this need an additional feature
+    feature_16bit_norm = "texture-format16bit-norm"  # in the future it will be tier1
+    adapter = wgpu.gpu.request_adapter_sync()
+    if feature_16bit_norm not in adapter.features:
+        skip(
+            f"{adapter.summary} does not support {feature_16bit_norm}, skipping test for 16bit norm textures"
+        )
+    device = adapter.request_device_sync(required_features=[feature_16bit_norm])
+
+    shape = (64, 64, 1)  # 1 channel should do it
+
+    data = np.empty(
+        shape=shape, dtype=np.uint16
+    )  # TODO: maybe use actual data or rand ints?
+
+    texture = device.create_texture(
+        label="test 16bit norm texture",
+        size=shape,
+        format=wgpu.enums.TextureFormat.r16unorm,
+        usage=wgpu.flags.TextureUsage.COPY_DST,  # not yet possible: wgpu.TextureUsage.RENDER_ATTACHMENT
+    )
+
+    device.queue.write_texture(
+        destination=wgpu.structs.TexelCopyTextureInfo(
+            texture=texture,
+        ),
+        data=data,
+        data_layout=wgpu.structs.TexelCopyBufferLayout(
+            bytes_per_row=shape[0] * 2,
+            rows_per_image=shape[1],
+        ),
+        size=shape,
+    )
+
+    texture_view = texture.create_view(format=wgpu.enums.TextureFormat.r16unorm)
+    assert texture_view
+    assert "r16unorm" in texture.format
+
+
 if __name__ == "__main__":
     run_tests(globals())

tools/download_wgpu_native.py

@@ -150,7 +150,7 @@ def main(version=None, os_string=None, arch=None, upstream=None):
     current_version = get_current_version()
     if version != current_version:
         print(
-            f"Version changed, updating {VERSION_FILE}, diff: https://github.com/{upstream}/compare/v{current_version}...v{version}"
+            f"Version changed, updating {VERSION_FILE}, diff: https://github.com/{upstream}/compare/v{current_version}...v{version}\nPlese run codegen to obtain the combined headerfile"
         )
         filename = "commit-sha"
         url = f"https://github.com/{upstream}/releases/download/v{version}/{filename}"

wgpu/backends/wgpu_native/__init__.py

@@ -11,8 +11,8 @@
 
 
 # The wgpu-native version that we target/expect
-__version__ = "29.0.0.0"
-__commit_sha__ = "d2e3330ade4ae1bb238d76b485926f067e7ee64c"
+__version__ = "29.0.1.1"
+__commit_sha__ = "6aed50955d934ac36049ba8d002034841633ae02"
 version_info = tuple(map(int, __version__.split(".")))  # noqa: RUF048
 _check_expected_version(version_info)  # produces a warning on mismatch
 

wgpu/backends/wgpu_native/_api.py

@@ -323,9 +323,7 @@ def _get_limits(id: int, device: bool = False, adapter: bool = False):
     """Gets the limits for a device or an adapter"""
     assert device + adapter == 1  # exactly one is set
 
-    # NOTE: this will get even simpler in the near future: https://github.com/gfx-rs/wgpu-native/pull/575 as maxImmediateSize shouldn't exist in both
-
-    # H: chain: WGPUChainedStruct, maxImmediateSize: int, maxNonSamplerBindings: int, maxBindingArrayElementsPerShaderStage: int
+    # H: chain: WGPUChainedStruct, maxNonSamplerBindings: int, maxBindingArrayElementsPerShaderStage: int, maxBindingArraySamplerElementsPerShaderStage: int, maxMultiviewViewCount: int
     c_limits_native = new_struct(
         "WGPUNativeLimits",
         # H: next: WGPUChainedStruct *, sType: WGPUSType
@@ -334,9 +332,10 @@ def _get_limits(id: int, device: bool = False, adapter: bool = False):
             # not used: next
             sType=lib.WGPUSType_NativeLimits,
         ),
-        # not used: maxImmediateSize
         # not used: maxNonSamplerBindings
         # not used: maxBindingArrayElementsPerShaderStage
+        # not used: maxBindingArraySamplerElementsPerShaderStage
+        # not used: maxMultiviewViewCount
     )
 
     # Note that the object returned by ffi.cast() does not own the memory, so we must keep a ref to the uncast object, until wgpu-native has consumed it.
@@ -1229,7 +1228,6 @@ def canonicalize_limit_name(name):
             # Skip the  pointers
             if snake_key in (
                 "next-in-chain",
-                "max-immediate-size",
                 "max-non-sampler-bindings",
             ):
                 # Skip the chain and the native limits as they are handled in their own
@@ -1241,18 +1239,18 @@ def canonicalize_limit_name(name):
                 value = self._limits[snake_key]
             setattr(c_required_limits, key, value)
 
+        # TODO: handle new native limits, or find a programmatic solution for it
         #  the native only limits are passed in via the next-in-chain struct
-        # H: chain: WGPUChainedStruct, maxImmediateSize: int, maxNonSamplerBindings: int, maxBindingArrayElementsPerShaderStage: int
+        # H: chain: WGPUChainedStruct, maxNonSamplerBindings: int, maxBindingArrayElementsPerShaderStage: int, maxBindingArraySamplerElementsPerShaderStage: int, maxMultiviewViewCount: int
         c_required_limits_native = new_struct_p(
             "WGPUNativeLimits *",
-            maxImmediateSize=required_limits.get(
-                "max-immediate-size", self._limits["max-immediate-size"]
-            ),
             maxNonSamplerBindings=required_limits.get(
                 "max-non-sampler-bindings", self._limits["max-non-sampler-bindings"]
             ),
             # not used: chain
             # not used: maxBindingArrayElementsPerShaderStage
+            # not used: maxBindingArraySamplerElementsPerShaderStage
+            # not used: maxMultiviewViewCount
         )
         c_required_limits_native.chain.next = ffi.NULL
         c_required_limits_native.chain.sType = lib.WGPUSType_NativeLimits
@@ -1805,29 +1803,14 @@ def create_pipeline_layout(
         bind_group_layouts_ids = [x._internal for x in bind_group_layouts]
         c_layout_array = new_array("WGPUBindGroupLayout[]", bind_group_layouts_ids)
 
-        c_pipeline_layout_next_in_chain = ffi.NULL
-        if immediate_size:
-            # H: chain: WGPUChainedStruct, immediateDataSize: int
-            c_pipeline_layout_extras = new_struct_p(
-                "WGPUPipelineLayoutExtras *",
-                # not used: chain
-                immediateDataSize=immediate_size,
-            )
-            c_pipeline_layout_extras.chain.sType = lib.WGPUSType_PipelineLayoutExtras
-            # Note that the object returned by ffi.cast() does not own the memory, so we must keep a ref to the uncast object, until wgpu-native has consumed it.
-            c_pipeline_layout_next_in_chain = ffi.cast(
-                "WGPUChainedStruct *", c_pipeline_layout_extras
-            )
-
-        # TODO: there is an immediateSize field in this struct too. so do we even need the extras?
         # H: nextInChain: WGPUChainedStruct *, label: WGPUStringView, bindGroupLayoutCount: int, bindGroupLayouts: WGPUBindGroupLayout *, immediateSize: int
         struct = new_struct_p(
             "WGPUPipelineLayoutDescriptor *",
-            nextInChain=c_pipeline_layout_next_in_chain,
             label=to_c_string_view(label),
             bindGroupLayouts=c_layout_array,
             bindGroupLayoutCount=len(bind_group_layouts),
-            # not used: immediateSize
+            immediateSize=immediate_size,
+            # not used: nextInChain
         )
 
         # H: WGPUPipelineLayout f(WGPUDevice device, WGPUPipelineLayoutDescriptor const * descriptor)
@@ -2990,13 +2973,13 @@ def set_immediates(
             raise ValueError("data_size + data_offset is too large")
 
         c_data = ffi.cast("void *", address)  # do we want to add data_offset?
-        # H: void wgpuComputePassEncoderSetImmediates(WGPUComputePassEncoder encoder, uint32_t offset, uint32_t sizeBytes, void const *data)
-        # H: void wgpuRenderPassEncoderSetImmediates(WGPURenderPassEncoder encoder, uint32_t offset, uint32_t sizeBytes, void const *data)
-        # H: void wgpuRenderBundleEncoderSetImmediates(WGPURenderBundleEncoder encoder, uint32_t offset, uint32_t sizeBytes, void const *data)
+        # H: void wgpuComputePassEncoderSetImmediates(WGPUComputePassEncoder computePassEncoder, uint32_t offset, void const * data, size_t size)
+        # H: void wgpuRenderPassEncoderSetImmediates(WGPURenderPassEncoder renderPassEncoder, uint32_t offset, void const * data, size_t size)
+        # H: void wgpuRenderBundleEncoderSetImmediates(WGPURenderBundleEncoder renderBundleEncoder, uint32_t offset, void const * data, size_t size)
         function = type(self)._set_immediates_function
         if function is None:
             self._not_implemented("set_immediates")
-        function(self._internal, range_offset, size, c_data + data_offset)
+        function(self._internal, range_offset, c_data + data_offset, size)
 
 
 class GPUDebugCommandsMixin(classes.GPUDebugCommandsMixin):

wgpu/backends/wgpu_native/_mappings.py

@@ -180,9 +180,9 @@
     "TextureFormat.etc2-rgba8unorm-srgb": 69,
     "TextureFormat.r16float": 9,
     "TextureFormat.r16sint": 8,
-    "TextureFormat.r16snorm": 196610,
+    "TextureFormat.r16snorm": 6,
     "TextureFormat.r16uint": 7,
-    "TextureFormat.r16unorm": 196609,
+    "TextureFormat.r16unorm": 5,
     "TextureFormat.r32float": 14,
     "TextureFormat.r32sint": 16,
     "TextureFormat.r32uint": 15,
@@ -193,9 +193,9 @@
     "TextureFormat.rg11b10ufloat": 31,
     "TextureFormat.rg16float": 21,
     "TextureFormat.rg16sint": 20,
-    "TextureFormat.rg16snorm": 196612,
+    "TextureFormat.rg16snorm": 18,
     "TextureFormat.rg16uint": 19,
-    "TextureFormat.rg16unorm": 196611,
+    "TextureFormat.rg16unorm": 17,
     "TextureFormat.rg32float": 33,
     "TextureFormat.rg32sint": 35,
     "TextureFormat.rg32uint": 34,
@@ -208,9 +208,9 @@
     "TextureFormat.rgb9e5ufloat": 32,
     "TextureFormat.rgba16float": 40,
     "TextureFormat.rgba16sint": 39,
-    "TextureFormat.rgba16snorm": 196614,
+    "TextureFormat.rgba16snorm": 37,
     "TextureFormat.rgba16uint": 38,
-    "TextureFormat.rgba16unorm": 196613,
+    "TextureFormat.rgba16unorm": 36,
     "TextureFormat.rgba32float": 41,
     "TextureFormat.rgba32sint": 43,
     "TextureFormat.rgba32uint": 42,
@@ -275,7 +275,7 @@
     "VertexStepMode.vertex": 1,
 }
 
-# There are 48 struct-field enum mappings
+# There are 49 struct-field enum mappings
 
 cstructfield2enum = {
     "BlendComponent.dstFactor": "BlendFactor",
@@ -286,6 +286,7 @@
     "CompilationMessage.type": "CompilationMessageType",
     "DepthStencilState.depthCompare": "CompareFunction",
     "DepthStencilState.format": "TextureFormat",
+    "ImageSubresourceRange.aspect": "TextureAspect",
     "PrimitiveState.cullMode": "CullMode",
     "PrimitiveState.frontFace": "FrontFace",
     "PrimitiveState.stripIndexFormat": "IndexFormat",
@@ -354,11 +355,14 @@
         "texture-compression-astc-hdr": 196620,
         "mappable-primary-buffers": 196622,
         "buffer-binding-array": 196623,
-        "uniform-buffer-and-storage-texture-array-non-uniform-indexing": 196624,
+        "storage-texture-array-non-uniform-indexing": 196624,
+        "address-mode-clamp-to-zero": 196625,
+        "address-mode-clamp-to-border": 196626,
         "polygon-mode-line": 196627,
         "polygon-mode-point": 196628,
         "conservative-rasterization": 196629,
-        "spirv-shader-passthrough": 196631,
+        "clear-texture": 196630,
+        "multiview": 196632,
         "vertex-attribute64bit": 196633,
         "texture-format-nv12": 196634,
         "ray-query": 196636,
@@ -371,6 +375,22 @@
         "timestamp-query-inside-encoders": 196644,
         "timestamp-query-inside-passes": 196645,
         "shader-int64": 196646,
+        "shader-float32-atomic": 196647,
+        "texture-atomic": 196648,
+        "texture-format-p010": 196649,
+        "pipeline-cache": 196651,
+        "shader-int64-atomic-min-max": 196652,
+        "shader-int64-atomic-all-ops": 196653,
+        "texture-int64-atomic": 196656,
+        "shader-barycentrics": 196663,
+        "selective-multiview": 196664,
+        "multisample-array": 196666,
+        "cooperative-matrix": 196667,
+        "shader-per-vertex": 196668,
+        "shader-draw-index": 196669,
+        "acceleration-structure-binding-array": 196670,
+        "memory-decoration-coherent": 196671,
+        "memory-decoration-volatile": 196672,
     },
     "PipelineStatisticName": {
         "vertex-shader-invocations": 0,

wgpu/resources/codegen_report.md

@@ -2,8 +2,8 @@
 ## Preparing
 * The webgpu.idl defines 37 classes with 77 functions
 * The webgpu.idl defines 5 flags, 34 enums, 60 structs
-* webgpu.h/wgpu.h define 226 functions
-* webgpu.h/wgpu.h define 7 flags, 68 enums, 114 structs
+* webgpu.h/wgpu.h define 228 functions
+* webgpu.h/wgpu.h define 8 flags, 70 enums, 115 structs
 ## Updating API
 * Wrote 5 flags to flags.py
 * Wrote 34 enums to enums.py
@@ -24,19 +24,12 @@
 ### Patching API for backends/wgpu_native/_api.py
 * Validated 38 classes, 116 methods, 0 properties
 ## Validating backends/wgpu_native/_api.py
-* Enum field TextureFormat.r16unorm overridden by wgpu.h
-* Enum field TextureFormat.r16snorm overridden by wgpu.h
-* Enum field TextureFormat.rg16unorm overridden by wgpu.h
-* Enum field TextureFormat.rg16snorm overridden by wgpu.h
-* Enum field TextureFormat.rgba16unorm overridden by wgpu.h
-* Enum field TextureFormat.rgba16snorm overridden by wgpu.h
 * Enum PipelineErrorReason missing in webgpu.h
 * Enum AutoLayoutMode missing in webgpu.h
 * Enum field VertexFormat.unorm10-10-10-2 missing in webgpu.h
-* Enum field QueryType.force32 overridden by wgpu.h
 * Enum CanvasAlphaMode missing in webgpu.h
 * Enum CanvasToneMappingMode missing in webgpu.h
-* Wrote 267 enum mappings and 48 struct-field mappings to wgpu_native/_mappings.py
+* Wrote 267 enum mappings and 49 struct-field mappings to wgpu_native/_mappings.py
 * Validated 154 C function calls
-* Not using 83 C functions
-* Validated 97 C structs
+* Not using 85 C functions
+* Validated 96 C structs