Files
codex/.github/scripts/run-bazel-ci.sh
Michael Bolin 466798aa83 ci: cross-compile Windows Bazel tests (#20585)
## Status

This is the Bazel PR-CI cross-compilation follow-up to #20485. It is
intentionally split from the Cargo/cargo-xwin release-build PoC so
#20485 can stay as the historical release-build exploration. The
unrelated async-utils test cleanup has been moved to #20686, so this PR
is focused on the Windows Bazel CI path.

The intended tradeoff is now explicit in `.github/workflows/bazel.yml`:
pull requests get the fast Windows cross-compiled Bazel test leg, while
post-merge pushes to `main` run both that fast cross leg and a fully
native Windows Bazel test leg. The native main-only job keeps full
V8/code-mode coverage and gets a 40-minute timeout because it is less
latency-sensitive than PR CI. All other Bazel jobs remain at 30 minutes.

## Why

Windows Bazel PR CI currently does the expensive part of the build on
Windows. A native Windows Bazel test job on `main` completed in about
28m12s, leaving very little headroom under the 30-minute job timeout and
making Windows the slowest PR signal.

#20485 showed that Windows cross-compilation can be materially faster
for Cargo release builds, but PR CI needs Bazel because Bazel owns our
test sharding, flaky-test retries, and integration-test layout. This PR
applies the same high-level shape we already use for macOS Bazel CI:
compile with remote Linux execution, then run platform-specific tests on
the platform runner.

The compromise is deliberately signal-aware: code-mode/V8 changes are
rare enough that PR CI can accept losing the direct V8/code-mode
smoke-test signal temporarily, while `main` still runs the native
Windows job post-merge to catch that class of regression. A follow-up PR
should investigate making the cross-built Windows gnullvm V8 archive
pass the direct V8/code-mode tests so this tradeoff can eventually go
away.

## What Changed

- Adds a `ci-windows-cross` Bazel config that targets
`x86_64-pc-windows-gnullvm`, uses Linux RBE for build actions, and keeps
`TestRunner` actions local on the Windows runner.
- Adds explicit Windows platform definitions for
`windows_x86_64_gnullvm`, `windows_x86_64_msvc`, and a bridge toolchain
that lets gnullvm test targets execute under the Windows MSVC host
platform.
- Updates the Windows Bazel PR test leg to opt into the cross-compile
path via `--windows-cross-compile` and `--remote-download-toplevel`.
- Adds a `test-windows-native-main` job that runs only for `push` events
on `refs/heads/main`, uses the native Windows Bazel path, includes
V8/code-mode smoke tests, and has `timeout-minutes: 40`.
- Keeps fork/community PRs without `BUILDBUDDY_API_KEY` on the previous
local Windows MSVC-host fallback, including
`--host_platform=//:local_windows_msvc` and `--jobs=8`.
- Preserves the existing integration-test shape on non-gnullvm
platforms, while generating Windows-cross wrapper targets only for
`windows_gnullvm`.
- Resolves `CARGO_BIN_EXE_*` values from runfiles at test runtime,
avoiding hard-coded Cargo paths and duplicate test runfiles.
- Extends the V8 Bazel patches enough for the
`x86_64-pc-windows-gnullvm` target and Linux remote execution path.
- Makes the Windows sandbox test cwd derive from `INSTA_WORKSPACE_ROOT`
at runtime when Bazel provides it, because cross-compiled binaries may
contain Linux compile-time paths.
- Keeps the direct V8/code-mode unit smoke tests out of the Windows
cross PR path for now while native Windows CI continues to cover them
post-merge.

## Command Shape

The fast Windows PR test leg invokes the normal Bazel CI wrapper like
this:

```shell
./.github/scripts/run-bazel-ci.sh \
  --print-failed-action-summary \
  --print-failed-test-logs \
  --windows-cross-compile \
  --remote-download-toplevel \
  -- \
  test \
  --test_tag_filters=-argument-comment-lint \
  --test_verbose_timeout_warnings \
  --build_metadata=COMMIT_SHA=${GITHUB_SHA} \
  -- \
  //... \
  -//third_party/v8:all \
  -//codex-rs/code-mode:code-mode-unit-tests \
  -//codex-rs/v8-poc:v8-poc-unit-tests
```

With the BuildBuddy secret available on Windows, the wrapper selects
`--config=ci-windows-cross` and appends the important Windows-cross
overrides after rc expansion:

```shell
--host_platform=//:rbe
--shell_executable=/bin/bash
--action_env=PATH=/usr/bin:/bin
--host_action_env=PATH=/usr/bin:/bin
--test_env=PATH=${CODEX_BAZEL_WINDOWS_PATH}
```

The native post-merge Windows job intentionally omits
`--windows-cross-compile` and does not exclude the V8/code-mode unit
targets:

```shell
./.github/scripts/run-bazel-ci.sh \
  --print-failed-action-summary \
  --print-failed-test-logs \
  -- \
  test \
  --test_tag_filters=-argument-comment-lint \
  --test_verbose_timeout_warnings \
  --build_metadata=COMMIT_SHA=${GITHUB_SHA} \
  --build_metadata=TAG_windows_native_main=true \
  -- \
  //... \
  -//third_party/v8:all
```

## Research Notes

The existing macOS Bazel CI config already uses the model we want here:
build actions run remotely with `--strategy=remote`, but `TestRunner`
actions execute on the macOS runner. This PR mirrors that pattern for
Windows with `--strategy=TestRunner=local`.

The important Bazel detail is that `rules_rs` is already targeting
`x86_64-pc-windows-gnullvm` for Windows Bazel PR tests. This PR changes
where the build actions execute; it does not switch the Bazel PR test
target to Cargo, `cargo-nextest`, or the MSVC release target.

Cargo release builds differ from this Bazel path for V8: the normal
Windows Cargo release target is MSVC, and `rusty_v8` publishes prebuilt
Windows MSVC `.lib.gz` archives. The Bazel PR path targets
`windows-gnullvm`; `rusty_v8` does not publish a prebuilt Windows
GNU/gnullvm archive, so this PR builds that archive in-tree. That
Linux-RBE-built gnullvm archive currently crashes in direct V8/code-mode
smoke tests, which is why the workflow keeps native Windows coverage on
`main`.

The less obvious Bazel detail is test wrapper selection. Bazel chooses
the Windows test wrapper (`tw.exe`) from the test action execution
platform, not merely from the Rust target triple. The outer
`workspace_root_test` therefore declares the default test toolchain and
uses the bridge toolchain above so the test action executes on Windows
while its inner Rust binary is built for gnullvm.

The V8 investigation exposed a Windows-client gotcha: even when an
action execution platform is Linux RBE, Bazel can still derive the
genrule shell path from the Windows client. That produced remote
commands trying to run `C:\Program Files\Git\usr\bin\bash.exe` on Linux
workers. The wrapper now passes `--shell_executable=/bin/bash` with
`--host_platform=//:rbe` for the Windows cross path.

The same Windows-client/Linux-RBE boundary also affected
`third_party/v8:binding_cc`: a multiline genrule command can carry CRLF
line endings into Linux remote bash, which failed as `$'\r'`. That
genrule now keeps the `sed` command on one physical shell line while
using an explicit Starlark join so the shell arguments stay readable.

## Verification

Local checks included:

```shell
bash -n .github/scripts/run-bazel-ci.sh
bash -n workspace_root_test_launcher.sh.tpl
ruby -e "require %q{yaml}; YAML.load_file(%q{.github/workflows/bazel.yml}); puts %q{ok}"
RUNNER_OS=Linux ./scripts/list-bazel-clippy-targets.sh
RUNNER_OS=Windows ./scripts/list-bazel-clippy-targets.sh
RUNNER_OS=Linux ./tools/argument-comment-lint/list-bazel-targets.sh
RUNNER_OS=Windows ./tools/argument-comment-lint/list-bazel-targets.sh
```

The Linux clippy and argument-comment target lists contain zero
`*-windows-cross-bin` labels, while the Windows lists still include 47
Windows-cross internal test binaries.

CI evidence:

- Baseline native Windows Bazel test on `main`: success in about 28m12s,
https://github.com/openai/codex/actions/runs/25206257208/job/73907325959
- Green Windows-cross Bazel run on the split PR before adding the
main-only native leg: Windows test 9m16s, Windows release verify 5m10s,
Windows clippy 4m43s,
https://github.com/openai/codex/actions/runs/25231890068
- The latest SHA adds the explicit PR-vs-main tradeoff in `bazel.yml`;
CI is rerunning on that focused diff.

## Follow-Up

A subsequent PR should investigate making a cross-built Windows binary
work with V8/code-mode enabled. Likely options are either making the
Linux-RBE-built `windows-gnullvm` V8 archive correct at runtime, or
evaluating whether a Bazel MSVC target/toolchain can reuse the same
prebuilt MSVC `rusty_v8` archive shape that Cargo release builds already
use.
2026-05-01 15:55:28 -07:00

454 lines
14 KiB
Bash
Executable File

#!/usr/bin/env bash
set -euo pipefail
print_failed_bazel_test_logs=0
print_failed_bazel_action_summary=0
remote_download_toplevel=0
windows_msvc_host_platform=0
windows_cross_compile=0
while [[ $# -gt 0 ]]; do
case "$1" in
--print-failed-test-logs)
print_failed_bazel_test_logs=1
shift
;;
--print-failed-action-summary)
print_failed_bazel_action_summary=1
shift
;;
--remote-download-toplevel)
remote_download_toplevel=1
shift
;;
--windows-msvc-host-platform)
windows_msvc_host_platform=1
shift
;;
--windows-cross-compile)
windows_cross_compile=1
shift
;;
--)
shift
break
;;
*)
echo "Unknown option: $1" >&2
exit 1
;;
esac
done
if [[ $# -eq 0 ]]; then
echo "Usage: $0 [--print-failed-test-logs] [--print-failed-action-summary] [--remote-download-toplevel] [--windows-msvc-host-platform] [--windows-cross-compile] -- <bazel args> -- <targets>" >&2
exit 1
fi
bazel_startup_args=()
if [[ -n "${BAZEL_OUTPUT_USER_ROOT:-}" ]]; then
bazel_startup_args+=("--output_user_root=${BAZEL_OUTPUT_USER_ROOT}")
fi
run_bazel() {
if [[ "${RUNNER_OS:-}" == "Windows" ]]; then
MSYS2_ARG_CONV_EXCL='*' bazel "$@"
return
fi
bazel "$@"
}
ci_config=ci-linux
case "${RUNNER_OS:-}" in
macOS)
ci_config=ci-macos
;;
Windows)
if [[ $windows_cross_compile -eq 1 ]]; then
ci_config=ci-windows-cross
else
ci_config=ci-windows
fi
;;
esac
print_bazel_test_log_tails() {
local console_log="$1"
local testlogs_dir
local -a bazel_info_cmd=(bazel)
local -a bazel_info_args=(info)
if (( ${#bazel_startup_args[@]} > 0 )); then
bazel_info_cmd+=("${bazel_startup_args[@]}")
fi
# `bazel info` needs the same CI config as the failed test invocation so
# platform-specific output roots match. On Windows, omitting `ci-windows`
# would point at `local_windows-fastbuild` even when the test ran with the
# MSVC host platform under `local_windows_msvc-fastbuild`.
if [[ -n "${BUILDBUDDY_API_KEY:-}" ]]; then
bazel_info_args+=(
"--config=${ci_config}"
"--remote_header=x-buildbuddy-api-key=${BUILDBUDDY_API_KEY}"
)
fi
# Only pass flags that affect Bazel's output-root selection or repository
# lookup. Test/build-only flags such as execution logs or remote download
# mode can make `bazel info` fail, which would hide the real test log path.
for arg in "${post_config_bazel_args[@]}"; do
case "$arg" in
--host_platform=* | --repo_contents_cache=* | --repository_cache=*)
bazel_info_args+=("$arg")
;;
esac
done
testlogs_dir="$(run_bazel "${bazel_info_cmd[@]:1}" \
--noexperimental_remote_repo_contents_cache \
"${bazel_info_args[@]}" \
bazel-testlogs 2>/dev/null || echo bazel-testlogs)"
local failed_targets=()
while IFS= read -r target; do
failed_targets+=("$target")
done < <(
grep -E '^(FAIL: //|ERROR: .* Testing //)' "$console_log" \
| sed -E 's#^FAIL: (//[^ ]+).*#\1#; s#^ERROR: .* Testing (//[^ ]+) failed:.*#\1#' \
| sort -u
)
if [[ ${#failed_targets[@]} -eq 0 ]]; then
echo "No failed Bazel test targets were found in console output."
return
fi
for target in "${failed_targets[@]}"; do
local rel_path="${target#//}"
rel_path="${rel_path/://}"
local test_log="${testlogs_dir}/${rel_path}/test.log"
local reported_test_log
reported_test_log="$(grep -F "FAIL: ${target} " "$console_log" | sed -nE 's#.* \(see (.*[\\/]test\.log)\).*#\1#p' | head -n 1 || true)"
if [[ -n "$reported_test_log" ]]; then
reported_test_log="${reported_test_log//\\//}"
test_log="$reported_test_log"
fi
echo "::group::Bazel test log tail for ${target}"
if [[ -f "$test_log" ]]; then
tail -n 200 "$test_log"
else
echo "Missing test log: $test_log"
fi
echo "::endgroup::"
done
}
print_bazel_action_failure_summary() {
local console_log="$1"
local escaped_summary
local summary
summary="$(
awk '
function clean(line) {
gsub(sprintf("%c", 27) "\\[[0-9;]*m", "", line)
sub(/^.*\t[^\t]*\t[0-9TZ:._-]+ /, "", line)
return line
}
function is_diagnostic(line) {
return line ~ /^(error(\[[^]]+\])?:|warning:|note:|help:)/ ||
line ~ /^[[:space:]]+-->/ ||
line ~ /^[[:space:]]*[0-9]+[[:space:]]+\|/ ||
line ~ /^[[:space:]]*\|/ ||
line ~ /^[[:space:]]+= (note|help):/ ||
line ~ /^[[:space:]]*\^[[:space:]^~-]*$/ ||
line ~ /^For more information/ ||
line ~ /^error: aborting/
}
{
line = clean($0)
}
line ~ /^ERROR: .* failed:/ {
if (printed) {
print ""
}
print line
in_failure = 1
seen_diagnostic = 0
printed = 1
next
}
in_failure && is_diagnostic(line) {
print line
seen_diagnostic = 1
next
}
in_failure && seen_diagnostic && line == "" {
print ""
next
}
in_failure && seen_diagnostic {
in_failure = 0
seen_diagnostic = 0
next
}
' "$console_log"
)"
if [[ -z "$summary" ]]; then
summary="$(grep -E '^ERROR: |^FAILED: ' "$console_log" | tail -n 50 || true)"
fi
if [[ -z "$summary" ]]; then
echo "No Bazel action failures were found in the captured console output."
return
fi
if [[ "${GITHUB_ACTIONS:-}" == "true" ]]; then
escaped_summary="$(
printf '%s' "$summary" \
| awk 'BEGIN { ORS = "" } {
gsub(/%/, "%25")
gsub(/\r/, "%0D")
print sep $0
sep = "%0A"
}'
)"
echo "::error title=Bazel failed action diagnostics::${escaped_summary}"
fi
echo
echo "Bazel failed action diagnostics:"
echo "--------------------------------"
printf '%s\n' "$summary"
echo "--------------------------------"
}
bazel_args=()
bazel_targets=()
found_target_separator=0
for arg in "$@"; do
if [[ "$arg" == "--" && $found_target_separator -eq 0 ]]; then
found_target_separator=1
continue
fi
if [[ $found_target_separator -eq 0 ]]; then
bazel_args+=("$arg")
else
bazel_targets+=("$arg")
fi
done
if [[ ${#bazel_args[@]} -eq 0 || ${#bazel_targets[@]} -eq 0 ]]; then
echo "Expected Bazel args and targets separated by --" >&2
exit 1
fi
if [[ "${RUNNER_OS:-}" == "Windows" && $windows_cross_compile -eq 1 && -z "${BUILDBUDDY_API_KEY:-}" ]]; then
# Fork PRs do not receive the BuildBuddy secret needed for the remote
# cross-compile config. Preserve the previous local Windows build shape.
windows_msvc_host_platform=1
fi
post_config_bazel_args=()
if [[ "${RUNNER_OS:-}" == "Windows" && $windows_msvc_host_platform -eq 1 ]]; then
has_host_platform_override=0
for arg in "${bazel_args[@]}"; do
if [[ "$arg" == --host_platform=* ]]; then
has_host_platform_override=1
break
fi
done
if [[ $has_host_platform_override -eq 0 ]]; then
# Use the MSVC Windows platform for jobs that need helper binaries like
# Rust test wrappers and V8 generators to resolve a compatible toolchain.
# Callers that need a different Windows target platform should pass an
# explicit `--platforms=...` flag.
post_config_bazel_args+=("--host_platform=//:local_windows_msvc")
fi
fi
if [[ $remote_download_toplevel -eq 1 ]]; then
# Override the CI config's remote_download_minimal setting when callers need
# the built artifact to exist on disk after the command completes.
post_config_bazel_args+=(--remote_download_toplevel)
fi
if [[ "${RUNNER_OS:-}" == "Windows" && $windows_cross_compile -eq 1 && -n "${BUILDBUDDY_API_KEY:-}" ]]; then
# `--enable_platform_specific_config` expands `common:windows` on Windows
# hosts after ordinary rc configs, which can override `ci-windows-cross`'s
# RBE host platform. Repeat the host platform on the command line so V8 and
# other genrules execute on Linux RBE workers instead of Git Bash locally.
#
# Bazel also derives the default genrule shell from the client host. Without
# an explicit shell executable, remote Linux actions can be asked to run
# `C:\Program Files\Git\usr\bin\bash.exe`.
post_config_bazel_args+=(--host_platform=//:rbe --shell_executable=/bin/bash)
fi
if [[ "${RUNNER_OS:-}" == "Windows" && $windows_cross_compile -eq 1 && -z "${BUILDBUDDY_API_KEY:-}" ]]; then
# The Windows cross-compile config depends on remote execution. Fork PRs do
# not receive the BuildBuddy secret, so fall back to the existing local build
# shape and keep its lower concurrency cap.
post_config_bazel_args+=(--jobs=8)
fi
if [[ -n "${BAZEL_REPO_CONTENTS_CACHE:-}" ]]; then
# Windows self-hosted runners can run multiple Bazel jobs concurrently. Give
# each job its own repo contents cache so they do not fight over the shared
# path configured in `ci-windows`.
post_config_bazel_args+=("--repo_contents_cache=${BAZEL_REPO_CONTENTS_CACHE}")
fi
if [[ -n "${BAZEL_REPOSITORY_CACHE:-}" ]]; then
post_config_bazel_args+=("--repository_cache=${BAZEL_REPOSITORY_CACHE}")
fi
if [[ -n "${CODEX_BAZEL_EXECUTION_LOG_COMPACT_DIR:-}" ]]; then
post_config_bazel_args+=(
"--execution_log_compact_file=${CODEX_BAZEL_EXECUTION_LOG_COMPACT_DIR}/execution-log-${bazel_args[0]}-${GITHUB_JOB:-local}-$$.zst"
)
fi
if [[ "${RUNNER_OS:-}" == "Windows" ]]; then
pass_windows_build_env=1
if [[ $windows_cross_compile -eq 1 && -n "${BUILDBUDDY_API_KEY:-}" ]]; then
# Remote build actions execute on Linux RBE workers. Passing the Windows
# runner's build environment there makes Bazel genrules try to execute
# C:\Program Files\Git\usr\bin\bash.exe on Linux.
pass_windows_build_env=0
fi
if [[ $pass_windows_build_env -eq 1 ]]; then
windows_action_env_vars=(
INCLUDE
LIB
LIBPATH
UCRTVersion
UniversalCRTSdkDir
VCINSTALLDIR
VCToolsInstallDir
WindowsLibPath
WindowsSdkBinPath
WindowsSdkDir
WindowsSDKLibVersion
WindowsSDKVersion
)
for env_var in "${windows_action_env_vars[@]}"; do
if [[ -n "${!env_var:-}" ]]; then
post_config_bazel_args+=("--action_env=${env_var}" "--host_action_env=${env_var}")
fi
done
fi
if [[ -z "${CODEX_BAZEL_WINDOWS_PATH:-}" ]]; then
echo "CODEX_BAZEL_WINDOWS_PATH must be set for Windows Bazel CI." >&2
exit 1
fi
if [[ $pass_windows_build_env -eq 1 ]]; then
post_config_bazel_args+=(
"--action_env=PATH=${CODEX_BAZEL_WINDOWS_PATH}"
"--host_action_env=PATH=${CODEX_BAZEL_WINDOWS_PATH}"
)
elif [[ $windows_cross_compile -eq 1 ]]; then
# Remote build actions run on Linux RBE workers. Give their shell snippets
# a Linux PATH while preserving CODEX_BAZEL_WINDOWS_PATH below for local
# Windows test execution.
post_config_bazel_args+=(
"--action_env=PATH=/usr/bin:/bin"
"--host_action_env=PATH=/usr/bin:/bin"
)
fi
post_config_bazel_args+=("--test_env=PATH=${CODEX_BAZEL_WINDOWS_PATH}")
fi
bazel_console_log="$(mktemp)"
trap 'rm -f "$bazel_console_log"' EXIT
bazel_cmd=(bazel)
if (( ${#bazel_startup_args[@]} > 0 )); then
bazel_cmd+=("${bazel_startup_args[@]}")
fi
if [[ -n "${BUILDBUDDY_API_KEY:-}" ]]; then
echo "BuildBuddy API key is available; using remote Bazel configuration."
# Work around Bazel 9 remote repo contents cache / overlay materialization failures
# seen in CI (for example "is not a symlink" or permission errors while
# materializing external repos such as rules_perl). We still use BuildBuddy for
# remote execution/cache; this only disables the startup-level repo contents cache.
bazel_run_args=(
"${bazel_args[@]}"
"--config=${ci_config}"
"--remote_header=x-buildbuddy-api-key=${BUILDBUDDY_API_KEY}"
)
if (( ${#post_config_bazel_args[@]} > 0 )); then
bazel_run_args+=("${post_config_bazel_args[@]}")
fi
set +e
run_bazel "${bazel_cmd[@]:1}" \
--noexperimental_remote_repo_contents_cache \
"${bazel_run_args[@]}" \
-- \
"${bazel_targets[@]}" \
2>&1 | tee "$bazel_console_log"
bazel_status=${PIPESTATUS[0]}
set -e
else
echo "BuildBuddy API key is not available; using local Bazel configuration."
# Keep fork/community PRs on Bazel but disable remote services that are
# configured in .bazelrc and require auth.
#
# Flag docs:
# - Command-line reference: https://bazel.build/reference/command-line-reference
# - Remote caching overview: https://bazel.build/remote/caching
# - Remote execution overview: https://bazel.build/remote/rbe
# - Build Event Protocol overview: https://bazel.build/remote/bep
#
# --noexperimental_remote_repo_contents_cache:
# disable remote repo contents cache enabled in .bazelrc startup options.
# https://bazel.build/reference/command-line-reference#startup_options-flag--experimental_remote_repo_contents_cache
# --remote_cache= and --remote_executor=:
# clear remote cache/execution endpoints configured in .bazelrc.
# https://bazel.build/reference/command-line-reference#common_options-flag--remote_cache
# https://bazel.build/reference/command-line-reference#common_options-flag--remote_executor
bazel_run_args=(
"${bazel_args[@]}"
--remote_cache=
--remote_executor=
)
if (( ${#post_config_bazel_args[@]} > 0 )); then
bazel_run_args+=("${post_config_bazel_args[@]}")
fi
set +e
run_bazel "${bazel_cmd[@]:1}" \
--noexperimental_remote_repo_contents_cache \
"${bazel_run_args[@]}" \
-- \
"${bazel_targets[@]}" \
2>&1 | tee "$bazel_console_log"
bazel_status=${PIPESTATUS[0]}
set -e
fi
if [[ ${bazel_status:-0} -ne 0 ]]; then
if [[ $print_failed_bazel_action_summary -eq 1 ]]; then
print_bazel_action_failure_summary "$bazel_console_log"
fi
if [[ $print_failed_bazel_test_logs -eq 1 ]]; then
print_bazel_test_log_tails "$bazel_console_log"
fi
exit "$bazel_status"
fi