bench: add codex help e2e macrobenchmark (#31295)

## Why

Bazel-backed end-to-end macrobenchmark plumbing needs a small,
deterministic first consumer that does not couple the shared
infrastructure to the remote-skill scenario.

## What

- add `codex_e2e_benchmark`, a small macro for Bazel-only Divan
benchmarks and runtime binary runfiles
- keep benchmark sources under `e2e_benches/` so Cargo does not
auto-discover them
- add a CLI example that resolves the real `codex` binary and measures
`codex --help`
- assert the spawned command succeeds

## Validation

- `bazel test --compilation_mode=fastbuild
--@rules_rust//rust/settings:extra_rustc_flag=-Cdebug-assertions=no
--@rules_rust//rust/settings:extra_exec_rustc_flag=-Cdebug-assertions=no
--cache_test_results=no --test_output=errors --test_arg=--test
//codex-rs/cli:codex-help-bench`

## Stack

1. [#31295 bench: add codex help e2e
macrobenchmark](https://github.com/openai/codex/pull/31295)
2. [#31428 bench: add e2e benchmark
entrypoints](https://github.com/openai/codex/pull/31428)
3. [#31429 ci: smoke Bazel e2e
benchmarks](https://github.com/openai/codex/pull/31429)

bazel/rules/BUILD.bazel

@@ -0,0 +1,5 @@
+package(default_visibility = ["//visibility:public"])
+
+exports_files([
+    "e2e_benchmark.bzl",
+])

bazel/rules/e2e_benchmark.bzl

@@ -0,0 +1,56 @@
+load("@crates//:defs.bzl", "all_crate_deps")
+load("@rules_rust//rust:defs.bzl", "rust_binary")
+load("//:defs.bzl", "workspace_root_test")
+
+_WORKSPACE_ROOT_MARKER = "//codex-rs/utils/cargo-bin:repo_root.marker"
+
+def codex_e2e_benchmark(name, binaries = [], data = [], deps = []):
+    """Defines a Bazel-only Divan end-to-end benchmark.
+
+    The benchmark source lives at `e2e_benches/<name>.rs`, with hyphens in
+    `name` replaced by underscores. `binaries` are runtime executables made
+    available through the same `CARGO_BIN_EXE_*` bridge used by Rust tests.
+
+    Args:
+        name: Stem for the generated `<name>-bench` target and benchmark.
+        binaries: Runtime executable labels that the benchmark spawns.
+        data: Additional runtime files needed by the benchmark.
+        deps: Additional Rust dependencies beyond the crate's Cargo deps.
+    """
+    benchmark_name = name.replace("-", "_")
+    source = "e2e_benches/{}.rs".format(benchmark_name)
+    binary_name = name + "-bench-bin"
+    runfile_env = {
+        binary: "CARGO_BIN_EXE_" + native.package_relative_label(binary).name
+        for binary in binaries
+    }
+
+    rust_binary(
+        name = binary_name,
+        testonly = True,
+        srcs = [source],
+        crate_name = benchmark_name + "_bench",
+        crate_root = source,
+        deps = all_crate_deps(
+            normal = True,
+            normal_dev = True,
+        ) + [
+            "@crates//:divan",
+        ] + deps,
+    )
+
+    workspace_root_test(
+        name = name + "-bench",
+        args = [
+            "--bench",
+            benchmark_name,
+        ],
+        data = data,
+        # Keep path resolution inside the wrapper so manifest-only runfiles
+        # work on every supported host platform.
+        runfile_env = runfile_env,
+        tags = ["manual"],
+        test_bin = ":" + binary_name,
+        visibility = ["//codex-rs:__pkg__"],
+        workspace_root_marker = _WORKSPACE_ROOT_MARKER,
+    )

codex-rs/cli/BUILD.bazel

@@ -1,5 +1,6 @@
 load("//:defs.bzl", "MACOS_WEBRTC_RUSTC_LINK_FLAGS", "codex_rust_crate")
 load("//bazel/platforms:release_binaries.bzl", "multiplatform_binaries")
+load("//bazel/rules:e2e_benchmark.bzl", "codex_e2e_benchmark")
 
 codex_rust_crate(
     name = "cli",
@@ -13,3 +14,8 @@ codex_rust_crate(
 multiplatform_binaries(
     name = "codex",
 )
+
+codex_e2e_benchmark(
+    name = "codex-help",
+    binaries = [":codex"],
+)

codex-rs/cli/Cargo.toml

@@ -104,3 +104,8 @@ predicates = { workspace = true }
 pretty_assertions = { workspace = true }
 sqlx = { workspace = true }
 wiremock = { workspace = true }
+
+[package.metadata.cargo-shear]
+# These Rust sources are intentionally Bazel-only macrobenchmarks rather than
+# Cargo targets.
+ignored-paths = ["e2e_benches/*.rs"]

codex-rs/cli/e2e_benches/codex_help.rs

@@ -0,0 +1,26 @@
+#![allow(clippy::expect_used)]
+
+use std::process::Command;
+
+use divan::Bencher;
+
+fn main() {
+    divan::main();
+}
+
+/// Exercises the Bazel-backed end-to-end benchmark path with a cheap,
+/// deterministic Codex invocation. Richer scenarios can add separate
+/// benchmark binaries without making the shared harness depend on them.
+#[divan::bench(sample_count = 20, sample_size = 1)]
+fn codex_help(bencher: Bencher) {
+    let codex = codex_utils_cargo_bin::cargo_bin("codex")
+        .expect("codex binary should be available through Bazel runfiles");
+
+    bencher.bench_local(move || {
+        let output = Command::new(&codex)
+            .arg("--help")
+            .output()
+            .expect("codex --help should run");
+        assert!(output.status.success(), "codex --help should succeed");
+    });
+}