Merge branch 'bazel-contrib:main' into aarch64Support

.bazelversion

@@ -1 +1 @@
-6.3.2
+6.4.0

.github/workflows/github-pages.yaml

@@ -16,6 +16,7 @@ jobs:
         docs:
           # { ref: <branch/tag name>, name: <docs drop down entry> }
           - { ref: main, name: latest }
+          - { ref: v0.2.1, name: 0.2.1 }
     steps:
       - uses: actions/checkout@v4
         if: ${{ matrix.docs.ref == 'main' }}
@@ -31,7 +32,7 @@ jobs:
           path: ~/.cache/bazel
           key: bazel-gen-docs-${{ matrix.docs.ref }}
 
-      - uses: actions/setup-python@v4
+      - uses: actions/setup-python@v5
         with:
           python-version: "3.10"
           cache: pip
@@ -44,7 +45,7 @@ jobs:
 
       - run: bazelisk shutdown
 
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: "${{ matrix.docs.name }}"
           path: ${{ github.workspace }}/docs/site/
@@ -58,13 +59,13 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
           path: ${{ github.workspace }}/docs/generated
       - name: Inspect docs site directory structure
         run: find ${{ github.workspace }}/docs/generated -maxdepth 2
 
-      - uses: actions/setup-python@v4
+      - uses: actions/setup-python@v5
         with:
           python-version: "3.10"
       - run: |

.github/workflows/integration-tests.yaml

@@ -148,7 +148,7 @@ jobs:
           status: ${{ job.status }}
 
       - name: Add comment to PR
-        uses: actions/github-script@v6
+        uses: actions/github-script@v7
         if: always()
         with:
           script: |

.github/workflows/release.yaml

@@ -45,7 +45,7 @@ jobs:
           echo "ARCHIVE_SHA256=$(shasum -a 256 ${{ github.workspace }}/.github/rules_cuda.tar.gz | cut -d ' ' -f 1)" >> $GITHUB_ENV
           echo "RELEASE_VERSION=${GITHUB_REF_NAME}" >> $GITHUB_ENV
       # Upload the artifact in case creating a release fails so all artifacts can then be manually recovered.
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: "rules_cuda.tar.gz"
           path: ${{ github.workspace }}/.github/rules_cuda.tar.gz

cuda/defs.bzl

@@ -6,6 +6,7 @@ load("//cuda/private:providers.bzl", _CudaArchsInfo = "CudaArchsInfo", _cuda_arc
 load("//cuda/private:os_helpers.bzl", _cc_import_versioned_sos = "cc_import_versioned_sos", _if_linux = "if_linux", _if_windows = "if_windows")
 load("//cuda/private:rules/cuda_objects.bzl", _cuda_objects = "cuda_objects")
 load("//cuda/private:rules/cuda_library.bzl", _cuda_library = "cuda_library")
+load("//cuda/private:macros/cuda_binary.bzl", _cuda_binary = "cuda_binary")
 load("//cuda/private:macros/cuda_test.bzl", _cuda_test = "cuda_test")
 load("//cuda/private:rules/cuda_toolkit.bzl", _cuda_toolkit = "cuda_toolkit")
 load(
@@ -29,8 +30,12 @@ cuda_toolchain_config_nvcc = _cuda_toolchain_config_nvcc
 cuda_archs = _cuda_archs
 CudaArchsInfo = _CudaArchsInfo
 
+# rules
 cuda_objects = _cuda_objects
 cuda_library = _cuda_library
+
+# macros
+cuda_binary = _cuda_binary
 cuda_test = _cuda_test
 
 if_linux = _if_linux

cuda/private/cuda_helper.bzl

@@ -3,6 +3,7 @@
 load("@bazel_skylib//lib:paths.bzl", "paths")
 load("@bazel_skylib//lib:types.bzl", "types")
 load("@bazel_skylib//rules:common_settings.bzl", "BuildSettingInfo")
+load("@bazel_tools//tools/cpp:toolchain_utils.bzl", "find_cpp_toolchain")
 load("//cuda/private:action_names.bzl", "ACTION_NAMES")
 load("//cuda/private:artifact_categories.bzl", "ARTIFACT_CATEGORIES")
 load("//cuda/private:providers.bzl", "ArchSpecInfo", "CudaArchsInfo", "CudaInfo", "Stage2ArchInfo", "cuda_archs")
@@ -242,6 +243,8 @@ def _create_common(ctx):
 
     merged_cc_info = cc_common.merge_cc_infos(cc_infos = [dep[CcInfo] for dep in all_cc_deps])
 
+    cc_toolchain = find_cpp_toolchain(ctx)
+
     # gather include info
     includes = merged_cc_info.compilation_context.includes.to_list()
     system_includes = []
@@ -276,6 +279,8 @@ def _create_common(ctx):
     host_defines = []
     host_local_defines = [i for i in attr.host_local_defines]
     host_compile_flags = attr._default_host_copts[BuildSettingInfo].value + [i for i in attr.host_copts]
+    if cc_toolchain.sysroot:
+        host_compile_flags.append("--sysroot={}".format(cc_toolchain.sysroot))
     host_link_flags = []
     if hasattr(attr, "host_linkopts"):
         host_link_flags.extend([i for i in attr.host_linkopts])

cuda/private/macros/cuda_binary.bzl

@@ -0,0 +1,21 @@
+load("//cuda/private:rules/cuda_library.bzl", _cuda_library = "cuda_library")
+
+def cuda_binary(name, **attrs):
+    """Wrapper to ensure the binary is compiled with the CUDA compiler."""
+    cuda_library_only_attrs = ["deps", "srcs", "hdrs"]
+
+    # https://bazel.build/reference/be/common-definitions?hl=en#common-attributes-binaries
+    cc_binary_only_attrs = ["args", "env", "output_licenses"]
+
+    cuda_library_name = "_" + name
+
+    _cuda_library(
+        name = cuda_library_name,
+        **{k: v for k, v in attrs.items() if k not in cc_binary_only_attrs}
+    )
+
+    native.cc_binary(
+        name = name,
+        deps = [cuda_library_name],
+        **{k: v for k, v in attrs.items() if k not in cuda_library_only_attrs}
+    )

cuda/private/macros/cuda_test.bzl

@@ -2,16 +2,21 @@ load("//cuda/private:rules/cuda_library.bzl", _cuda_library = "cuda_library")
 
 def cuda_test(name, **attrs):
     """Wrapper to ensure the test is compiled with the CUDA compiler."""
+    cuda_library_only_attrs = ["deps", "srcs", "hdrs"]
+
+    # https://bazel.build/reference/be/common-definitions?hl=en#common-attributes-tests
+    cc_test_only_attrs = ["args", "env", "env_inherit", "size", "timeout", "flaky", "shard_count", "local"]
+
     cuda_library_name = "_" + name
 
     _cuda_library(
         name = cuda_library_name,
         testonly = True,
-        **attrs
+        **{k: v for k, v in attrs.items() if k not in cc_test_only_attrs}
     )
 
     native.cc_test(
         name = name,
         deps = [cuda_library_name],
-        **{k: v for k, v in attrs.items() if k != "deps"}
+        **{k: v for k, v in attrs.items() if k not in cuda_library_only_attrs}
     )

cuda/runtime/BUILD.local_cuda

@@ -8,18 +8,24 @@ filegroup(
     name = "compiler_deps",
     srcs = [
         ":_cuda_header_files",
-    ] + glob([
-        "cuda/version.txt",
-        "cuda/version.json",
-        "cuda/bin/**",
-        "cuda/lib64/**",
-        "cuda/nvvm/**",
-    ]),
+    ] + glob(
+        [
+            "cuda/version.json",
+            "cuda/version.txt",
+            "cuda/bin/**",
+            "cuda/lib64/**",
+            "cuda/nvvm/**",
+        ],
+        allow_empty = True,
+    ),
 )
 
 filegroup(
     name = "_cuda_header_files",
-    srcs = glob(["cuda/include/**"]),
+    srcs = glob(
+        ["cuda/include/**"],
+        allow_empty = True,
+    ),
     visibility = ["//visibility:private"],
 )
 
@@ -174,15 +180,21 @@ cc_library(
 
 cc_library(
     name = "cub",
-    hdrs = glob(["cuda/include/cub/**"]),
+    hdrs = glob(
+        ["cuda/include/cub/**"],
+        allow_empty = True,
+    ),
     includes = [
         "cuda/include",
     ],
 )
 
 cc_library(
     name = "thrust",
-    hdrs = glob(["cuda/include/thrust/**"]),
+    hdrs = glob(
+        ["cuda/include/thrust/**"],
+        allow_empty = True,
+    ),
     includes = [
         "cuda/include",
     ],
@@ -204,7 +216,10 @@ cc_import(
 
 cc_library(
     name = "cupti_headers",
-    hdrs = glob(["cuda/extras/CUPTI/include/*.h"]),
+    hdrs = glob(
+        ["cuda/extras/CUPTI/include/*.h"],
+        allow_empty = True,
+    ),
     includes = ["cuda/extras/CUPTI/include"],
 )
 
@@ -337,11 +352,14 @@ cc_import(
 cc_library(
     name = "nvptxcompiler",
     srcs = [],
-    hdrs = glob([
-        "cuda/include/fatbinary_section.h",
-        "cuda/include/nvPTXCompiler.h",
-        "cuda/include/crt/*",
-    ]),
+    hdrs = glob(
+        [
+            "cuda/include/fatbinary_section.h",
+            "cuda/include/nvPTXCompiler.h",
+            "cuda/include/crt/*",
+        ],
+        allow_empty = True,
+    ),
     includes = [
         "cuda/include",
     ],

docs/requirements.txt

@@ -1 +1 @@
-mkdocs-material==9.4.2
+mkdocs-material==9.5.3

docs/user_docs.bzl

@@ -1,9 +1,11 @@
-load("@rules_cuda//cuda:defs.bzl", _cuda_library = "cuda_library", _cuda_objects = "cuda_objects", _cuda_test = "cuda_test")
+load("@rules_cuda//cuda:defs.bzl", _cuda_binary = "cuda_binary", _cuda_library = "cuda_library", _cuda_objects = "cuda_objects", _cuda_test = "cuda_test")
 load("@rules_cuda//cuda:repositories.bzl", _register_detected_cuda_toolchains = "register_detected_cuda_toolchains", _rules_cuda_dependencies = "rules_cuda_dependencies")
 load("@rules_cuda//cuda/private:rules/flags.bzl", _cuda_archs_flag = "cuda_archs_flag")
 
 cuda_library = _cuda_library
 cuda_objects = _cuda_objects
+
+cuda_binary = _cuda_binary
 cuda_test = _cuda_test
 
 cuda_archs = _cuda_archs_flag

examples/WORKSPACE.bazel

@@ -27,9 +27,9 @@ http_archive(
     name = "nccl",
     add_prefix = "nccl",
     build_file = "@rules_cuda_examples//nccl:nccl.BUILD",
-    sha256 = "0de8aed51bf95c96c14bc801ba840eb5b5df527e7105b1f42a58ad5d4d617678",
-    strip_prefix = "nccl-4365458757e4107ecbf629b2fd6e0e19a5d237c2",
-    urls = ["https://github.com/nvidia/nccl/archive/4365458757e4107ecbf629b2fd6e0e19a5d237c2.tar.gz"],
+    sha256 = "83b299cfc2dfe63887dadf3590b3ac2b8b2fd68ec5515b6878774eda39a697d2",
+    strip_prefix = "nccl-9814c75eea18fc7374cde884592233b6b7dc055b",
+    urls = ["https://github.com/nvidia/nccl/archive/9814c75eea18fc7374cde884592233b6b7dc055b.tar.gz"],
 )
 
 http_archive(
@@ -41,7 +41,7 @@ http_archive(
         "-p1",
     ],
     patches = ["@rules_cuda_examples//nccl:nccl-tests-clang.patch"],
-    sha256 = "fe0bdc0fdfba75a9326b32381ce486b99aa5e6189181fbf92baa4f6153961c3f",
-    strip_prefix = "nccl-tests-6c46206a478203b6453035fe0d40dc6418acd089",
-    urls = ["https://github.com/nvidia/nccl-tests/archive/6c46206a478203b6453035fe0d40dc6418acd089.tar.gz"],
+    sha256 = "946adb84f63aec66aea7aab9739d41df81c24f783e85fba6328ba243cfc057e0",
+    strip_prefix = "nccl-tests-1a5f551ffd6e3271982b03a9d5653a3f6ba545fa",
+    urls = ["https://github.com/nvidia/nccl-tests/archive/1a5f551ffd6e3271982b03a9d5653a3f6ba545fa.tar.gz"],
 )

examples/basic_macros/BUILD.bazel

@@ -0,0 +1,13 @@
+load("@rules_cuda//cuda:defs.bzl", "cuda_binary", "cuda_test")
+
+package(default_visibility = ["//visibility:public"])
+
+cuda_binary(
+    name = "main",
+    srcs = ["main.cu"],
+)
+
+cuda_test(
+    name = "test",
+    srcs = ["main.cu"],
+)

examples/basic_macros/main.cu

@@ -0,0 +1,26 @@
+#include <cstdio>
+
+#define CUDA_CHECK(expr)                                                \
+  do {                                                                  \
+    cudaError_t err = (expr);                                           \
+    if (err != cudaSuccess) {                                           \
+      fprintf(stderr, "CUDA Error Code  : %d\n     Error String: %s\n", \
+              err, cudaGetErrorString(err));                            \
+      exit(err);                                                        \
+    }                                                                   \
+  } while (0)
+
+__global__ void kernel() {
+  printf("cuda kernel called!\n");
+}
+
+void launch() {
+  kernel<<<1, 1>>>();
+  CUDA_CHECK(cudaGetLastError());
+  CUDA_CHECK(cudaDeviceSynchronize());
+}
+
+int main() {
+  launch();
+  return 0;
+}

renovate.json

@@ -2,5 +2,11 @@
   "$schema": "https://docs.renovatebot.com/renovate-schema.json",
   "extends": ["config:base"],
   "ignorePaths": ["MODULE.bazel"],
+  "packageRules": [
+    {
+      "matchManagers": ["bazel", "bazel-module"],
+      "excludePackageNames": ["nccl", "nccl-tests"]
+    }
+  ],
   "schedule": ["on the first day of the month"]
 }