xtask: Add musl-sysroot-url subcommand

Add a submcommand that prints the full URL of a musl sysroot tarball
from Gentoo[0][1] for the given architecture.

These tarballs are convenient to use for building bpf-linker for the
`*-unknown-linux-musl` targets in GitHub actions. They contain static
libstdc++ that's linked to musl.

We need to link libstdc++ becuase libLLVM provided by Rust CI are
linked dynamically to it. Producing a static binary requires a static
libstdc++ to provide missing symbols.

Building bpf-linker for `*-unknown-linux-musl` targets using Ubuntu's
libstdc++ does not work because it's linked dynamically to glibc,
causing linking errors like:

```
(.text._ZNSt6locale13_S_initializeEv+0x0): undefined reference to `__libc_single_threaded'
```

Using Gentoo's tarball and referencing it in `CXXSTDLIB` variable
solves the problem. After unpacking the tarball to `sysroot/`, one can
build a static binary using the following command on any distro:

```
RUSTFLAGS="-C target-feature=+crt-static" \
    CXXSTDLIB_PATH="$(dirname $(find sysroot/ -name libstdc++.a))" \
    cargo +nightly test \
    --features llvm-21,llvm-link-static --no-default-features \
    --target x86_64-unknown-linux-musl
```

Why Gentoo? It's more convenient than the considered alternatives:

- Alpine Linux minirootfs[2] does not contain libstdc++. Neither does
  Void Linux rootfs[3]. One would need to chroot into it the sysroot and
  install it, which requires more work.
- musl.cc[4] did not provide any update update since 2021.
- Chimera Linux[5] uses LLVM libc++.

[0] https://distfiles.gentoo.org/releases/amd64/autobuilds/current-stage3-amd64-musl/
[1] https://distfiles.gentoo.org/releases/arm64/autobuilds/current-stage3-arm64-musl/
[2] https://www.alpinelinux.org/downloads/
[3] https://voidlinux.org/download/#download-installable-base-live-images-and-rootfs-tarballs
[4] https://musl.cc/
[5] https://chimera-linux.org/

Author: vadorovsky

Cargo.lock

@@ -52,7 +52,7 @@ object = { version = "0.37", default-features = false, features = [
 
 [dev-dependencies]
 compiletest_rs = { version = "0.11.0" }
-regex = { version = "1.11.1", default-features = false }
+regex = { workspace = true }
 rustc-build-sysroot = { workspace = true }
 which = { version = "8.0.0", default-features = false, features = ["real-sys", "regex"] }
 
@@ -101,6 +101,7 @@ edition = "2024"
 anyhow = { version = "1.0.100", default-features = false }
 clap = { version = "4.5.53", features = ["derive", "env"] }
 # dev deps
+regex = { version = "1.11.1", default-features = false }
 reqwest = { version = "0.12.24", default-features = false }
 rustc-build-sysroot = { version = "0.5.11", default-features = false }
 serde = { version = "1.0.228", default-features = false }

Cargo.toml

@@ -6,6 +6,7 @@ edition.workspace = true
 [dependencies]
 anyhow = { workspace = true }
 clap = { workspace = true }
+regex = { workspace = true }
 reqwest = { workspace = true, features = ["blocking", "json", "rustls-tls"] }
 rustc-build-sysroot = { workspace = true }
 serde = { workspace = true, features = ["derive"] }

xtask/Cargo.toml

@@ -70,6 +70,9 @@ enum XtaskSubcommand {
     BuildStd(BuildStd),
     /// Manages and builds LLVM.
     BuildLlvm(BuildLlvm),
+    /// Finds the full URL of a musl-based sysroot that contains necessary
+    /// system libraries for static linking.
+    MuslSysrootUrl,
     /// Finds the commit in github.com/rust-lang/rust that can be used for
     /// downloading LLVM for the current Rust toolchain.
     RustcLlvmCommit(RustcLlvmCommitOptions),
@@ -202,6 +205,53 @@ fn build_llvm(options: BuildLlvm) -> Result<()> {
     Ok(())
 }
 
+fn musl_sysroot_url() -> Result<()> {
+    let arch = if cfg!(target_arch = "aarch64") {
+        "arm64"
+    } else if cfg!(target_arch = "x86_64") {
+        "amd64"
+    } else {
+        anyhow::bail!("unsupported architecture: {}", env::consts::ARCH);
+    };
+    let base_url = format!(
+        "https://distfiles.gentoo.org/releases/{arch}/autobuilds/current-stage3-{arch}-musl"
+    );
+    let latest_file = format!("latest-stage3-{arch}-musl.txt");
+
+    // Parse the the `latest-*.txt` file that contains the full tarball name.
+    let latest_url = format!("{base_url}/{latest_file}");
+    let content = reqwest::blocking::get(&latest_url)
+        .with_context(|| format!("failed to fetch latest version info from {latest_url}"))?
+        .text()
+        .with_context(|| format!("failed to read response text from {latest_url}"))?;
+    let re = regex::Regex::new(r"^(stage3-[a-z0-9]+-musl-[^.]+\.tar\.xz) +[0-9]+$")
+        .expect("could not compile the regular expression");
+    let tarballs: Vec<_> = content
+        .lines()
+        .filter_map(|line| {
+            re.captures(line).map(|cap| {
+                let (_, [tarball]) = cap.extract();
+                tarball
+            })
+        })
+        .collect();
+    let tarball = match tarballs.as_slice() {
+        [] => {
+            anyhow::bail!("could not find stage3 tarball filename in {latest_url}: {content}");
+        }
+        [tarball] => tarball,
+        tarballs => {
+            anyhow::bail!(
+                "found multiple stage3 tarball filenames in {latest_url}: {tarballs:?}: {content}"
+            );
+        }
+    };
+
+    println!("{base_url}/{tarball}");
+
+    Ok(())
+}
+
 #[derive(Deserialize)]
 struct SearchIssuesResponse {
     items: Vec<IssueItem>,
@@ -329,6 +379,7 @@ fn main() -> Result<()> {
     match subcommand {
         XtaskSubcommand::BuildStd(options) => build_std(options),
         XtaskSubcommand::BuildLlvm(options) => build_llvm(options),
+        XtaskSubcommand::MuslSysrootUrl => musl_sysroot_url(),
         XtaskSubcommand::RustcLlvmCommit(options) => rustc_llvm_commit(options),
     }
 }