Expose sysroots as single directory artifacts

bazel/MODULE.bazel

@@ -86,9 +86,9 @@ use_repo(reachability_test_ext, "apparent_excluded_repo", "apparent_excluded_tra
 http_archive = use_repo_rule("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
 
 http_archive(
-    dev_dependency = True,
     name = "llvm_source",
     build_file_content = """filegroup(name = "all", srcs = glob(["**"]), visibility = ["//visibility:public"])""",
+    dev_dependency = True,
     sha256 = "ad18b70e287954c3d62bc7e0b86e7b7af2adf87bcfce21c15fe717f101d7aace",
     strip_prefix = "llvm-project-llvmorg-22.1.8",
     url = "https://github.com/llvm/llvm-project/archive/llvmorg-22.1.8.tar.gz",
@@ -107,6 +107,14 @@ sysroot_ext = use_extension("//sysroot:extensions.bzl", "sysroot_extension", dev
 sysroot_ext.setup()
 use_repo(sysroot_ext, "sysroot_linux_amd64", "sysroot_linux_arm64")
 
+macos_sysroot_test_repo = use_repo_rule("//sysroot:sysroot.bzl", "macos_sysroot")
+
+macos_sysroot_test_repo(
+    name = "sysroot_macos_empty",
+    dev_dependency = True,
+    version = "test",
+)
+
 macos_sysroot_build_ext = use_extension("//sysroot:extensions.bzl", "macos_sysroot_build_extension", dev_dependency = True)
 use_repo(macos_sysroot_build_ext, "macos_sdk_pkg", "pkgutil_linux_amd64")
 
@@ -192,12 +200,12 @@ llvm.toolchain_root(
 )
 llvm.sysroot(
     name = "llvm_toolchain",
-    label = "@sysroot_linux_amd64//:sysroot",
+    label = "@sysroot_linux_amd64//sysroot:sysroot",
     targets = ["linux-x86_64"],
 )
 llvm.sysroot(
     name = "llvm_toolchain",
-    label = "@sysroot_linux_arm64//:sysroot",
+    label = "@sysroot_linux_arm64//sysroot:sysroot",
     targets = ["linux-aarch64"],
 )
 llvm.extra_exec_compatible_with(

bazel/sysroot/README.md

@@ -39,6 +39,33 @@ This ensures that even older glibc sysroots can compile code using modern kernel
 
 ## Usage in Bazel
 
+### Repository layout
+
+Archives are extracted into the repository's `sysroot/` directory. The public
+`:sysroot` target is an alias to `//sysroot:sysroot`, a single source-directory
+artifact rather than a filegroup expanding every file. This gives compile actions
+one sandbox symlink and one merkle-tree node, avoiding Bazel 9's multi-file sysroot
+warning. Files named with `:` and `*.pl` files are removed after extraction because
+a directory artifact cannot exclude individual files.
+
+The Linux `:headers`, `:libs`, and `:toolchain_sysroot` targets remain public:
+headers and libraries still provide individual files, while `:toolchain_sysroot`
+wraps the directory artifact. Their filegroups live inside the `sysroot` package
+because Bazel globs cannot cross package boundaries.
+
+Raw file labels now use the `sysroot` package, for example
+`@sysroot_linux_amd64//sysroot:usr/include/stdio.h` instead of
+`@sysroot_linux_amd64//:usr/include/stdio.h`. Downstream Envoy's
+`source/extensions/dynamic_modules/sdk/rust/BUILD` must update its raw `stdio.h`
+label accordingly; `@sysroot_linux_amd64//:headers` is unchanged.
+
+Configure LLVM's `llvm.sysroot` with
+`@sysroot_linux_amd64//sysroot:sysroot` (and the arm64 equivalent).
+LLVM derives its sysroot path from the label's package, not the alias's resolved
+artifact, so the direct label is required to get
+`external/<canonical-repository-name>/sysroot`. macOS repositories use the same
+directory-and-alias layout, including when the SDK is empty.
+
 ### Direct `.bzl` load
 
 To use these sysroots via a direct `.bzl` load:

bazel/sysroot/sysroot.bzl

@@ -1,7 +1,22 @@
-"""Repository rules for sysroots."""
+"""Repository rules exposing sysroots as single source-directory artifacts."""
 
 load("//:versions.bzl", "VERSIONS")
 
+_SYSROOT_ALIAS_BUILD = """\
+package(default_visibility = ["//visibility:public"])
+
+alias(
+    name = "sysroot",
+    actual = "//sysroot",
+)
+"""
+
+def _prune_sysroot(ctx):
+    # Source directories cannot filter inputs, unlike the previous glob.
+    result = ctx.execute(["find", "sysroot", "-depth", "(", "-name", "*:*", "-o", "-name", "*.pl", ")", "-delete"])
+    if result.return_code:
+        fail("Failed to prune sysroot: {}".format(result.stderr))
+
 def _get_platform_arch(ctx):
     """Get the platform architecture for sysroot selection."""
     arch = ctx.os.arch
@@ -13,7 +28,7 @@ def _get_platform_arch(ctx):
         fail("Unsupported architecture: {}".format(arch))
 
 def _sysroot_impl(ctx):
-    """Implementation for sysroot repository rule."""
+    """Download a sysroot with one sandbox symlink and one merkle-tree node."""
     arch = ctx.attr.arch or _get_platform_arch(ctx)
     glibc_version = ctx.attr.glibc_version
     stdcc_version = ctx.attr.stdcc_version
@@ -37,19 +52,16 @@ def _sysroot_impl(ctx):
         url = url,
         sha256 = ctx.attr.sha256,
         stripPrefix = "",
+        output = "sysroot",
     )
-    ctx.file("BUILD.bazel", """
+    _prune_sysroot(ctx)
+    ctx.file("sysroot/BUILD.bazel", """
 package(default_visibility = ["//visibility:public"])
 
+# A single source directory avoids thousands of individual action inputs.
 filegroup(
     name = "sysroot",
-    srcs = glob(
-        ["**"],
-        exclude = [
-            "**/*:*",
-            "**/*.pl",
-        ],
-    ),
+    srcs = ["."],
 )
 
 filegroup(
@@ -72,6 +84,19 @@ filegroup(
         exclude = ["**/*:*"],
     ),
 )
+""")
+
+    # Globs cannot cross the sysroot subpackage boundary.
+    ctx.file("BUILD.bazel", _SYSROOT_ALIAS_BUILD + """
+filegroup(
+    name = "headers",
+    srcs = ["//sysroot:headers"],
+)
+
+filegroup(
+    name = "libs",
+    srcs = ["//sysroot:libs"],
+)
 
 filegroup(
     name = "toolchain_sysroot",
@@ -161,35 +186,33 @@ package(default_visibility = ["//visibility:public"])
 
 filegroup(
     name = "sysroot",
-    srcs = glob(
-        ["**"],
-        exclude = ["**/*:*"],
-    ),
+    srcs = ["."],
 )
 """
 
 def _macos_sysroot_impl(ctx):
-    """Implementation for macOS SDK sysroot repository rule."""
-    if ctx.os.name.startswith("mac"):
-        # On macOS, no cross-compilation sysroot needed.
-        ctx.file("BUILD.bazel", _MACOS_SYSROOT_BUILD)
-        return
-
+    """Expose the macOS SDK as one source directory, including empty SDKs."""
     sha256 = ctx.attr.sha256
-    if sha256:
+    if sha256 and not ctx.os.name.startswith("mac"):
         url = "https://github.com/envoyproxy/toolshed/releases/download/bins-v{version}/sysroot-macos-{arch}.tar.xz".format(
             version = ctx.attr.version,
             arch = ctx.attr.arch,
         )
         ctx.download_and_extract(
             url,
             sha256 = sha256,
+            output = "sysroot",
         )
+        _prune_sysroot(ctx)
     else:
-        # No hash available yet — create an empty sysroot.
-        ctx.execute(["mkdir", "-p", "usr/include"])
-
-    ctx.file("BUILD.bazel", _MACOS_SYSROOT_BUILD)
+        # Native macOS or no hash available yet — create an empty sysroot.
+        result = ctx.execute(["mkdir", "-p", "sysroot/usr/include"])
+        if result.return_code:
+            fail("Failed to create empty sysroot: {}".format(result.stderr))
+
+    # One sandbox symlink and one merkle-tree node, just like the Linux SDK.
+    ctx.file("sysroot/BUILD.bazel", _MACOS_SYSROOT_BUILD)
+    ctx.file("BUILD.bazel", _SYSROOT_ALIAS_BUILD)
 
 macos_sysroot = repository_rule(
     implementation = _macos_sysroot_impl,

bazel/sysroot/test/BUILD

@@ -0,0 +1,21 @@
+load(":sysroot_test.bzl", "sysroot_test")
+
+[
+    sysroot_test(
+        name = "linux_%s_test" % arch,
+        headers = "@sysroot_linux_%s//:headers" % arch,
+        libs = "@sysroot_linux_%s//:libs" % arch,
+        stdio = "@sysroot_linux_%s//sysroot:usr/include/stdio.h" % arch,
+        sysroot = "@sysroot_linux_%s//:sysroot" % arch,
+        toolchain_sysroot = "@sysroot_linux_%s//:toolchain_sysroot" % arch,
+    )
+    for arch in [
+        "amd64",
+        "arm64",
+    ]
+]
+
+sysroot_test(
+    name = "macos_empty_test",
+    sysroot = "@sysroot_macos_empty//:sysroot",
+)

bazel/sysroot/test/sysroot_test.bzl

@@ -0,0 +1,34 @@
+"""Regression tests for directory sysroots and their file-level targets."""
+
+load("@bazel_skylib//lib:unittest.bzl", "asserts", "unittest")
+
+def _sysroot_test_impl(ctx):
+    env = unittest.begin(ctx)
+    files = ctx.attr.sysroot[DefaultInfo].files.to_list()
+    asserts.equals(env, 1, len(files))
+    if len(files) == 1:
+        root = files[0]
+        asserts.equals(env, "sysroot", root.basename)
+        if ctx.attr.toolchain_sysroot:
+            asserts.equals(env, files, ctx.attr.toolchain_sysroot[DefaultInfo].files.to_list())
+        if ctx.attr.headers:
+            headers = ctx.attr.headers[DefaultInfo].files.to_list()
+            libs = ctx.attr.libs[DefaultInfo].files.to_list()
+            asserts.true(env, len(headers) > 0)
+            asserts.true(env, len(libs) > 0)
+            asserts.true(env, ctx.file.stdio in headers)
+            for file in headers + libs:
+                asserts.true(env, file.path.startswith(root.path + "/"))
+                asserts.false(env, ":" in file.basename or file.basename.endswith(".pl"))
+    return unittest.end(env)
+
+sysroot_test = unittest.make(
+    _sysroot_test_impl,
+    attrs = {
+        "sysroot": attr.label(mandatory = True),
+        "toolchain_sysroot": attr.label(),
+        "headers": attr.label(),
+        "libs": attr.label(),
+        "stdio": attr.label(allow_single_file = True),
+    },
+)