lei.chen/myxce
1
0
Fork 1
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix(tools): one xce-build.sh run per build dir at a time (bug-1604)

Browse source 
Two xce-build.sh runs against the same build dir raced: one run's
setup_build rm -rf hit files the other was writing ('Directory not
empty'), or deleted a binary the other had just built and tested.

xce-build.sh now takes an exclusive per-build-dir lock right after cd-ing
into src/ and holds it until it exits; a second run against the same dir
waits, runs against different dirs are unaffected. The lock
(src/tools/build-lock.sh) is an mkdir lock holding the owner PID, outside
the build dir, keyed by a hash of its absolute path; it needs only
coreutils (the nix dev shell has no flock) and takes over a lock whose
owner died. src/tools/test-build-lock.sh tests it.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Rongsong Shen 2026-09-30 03:43:48 +00:00
parent e9698c3f54
commit 980a0054d1
Signed by untrusted user: rshen
GPG key ID: 80533536B0EB651C
3 changed files with 183 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

59
src/tools/build-lock.sh Normal file
View file

@ -0,0 +1,59 @@
# build-lock.sh — exclusive per-build-dir lock for xce-build.sh (bug-1604).
#
# Sourced, not executed. Two xce-build.sh runs against the SAME build dir used
# to race: one run's `rm -rf $BUILD_DIR` hit files the other was writing
# ("Directory not empty") or silently deleted a binary the other had just
# built and tested. The lock serializes them for the whole run (wipe, setup,
# compile, tests); runs against DIFFERENT build dirs stay concurrent.
#
# The lock is a directory created with mkdir (atomic; needs nothing beyond
# coreutils, unlike flock(1), which the nix dev shell does not provide),
# holding the owner's PID. It lives outside the build dir -- the wipe would
# delete it -- under $XCE_BUILD_LOCK_ROOT (default $TMPDIR or /tmp), named
# after a hash of the build dir's absolute path, so separate checkouts never
# share a lock. A lock whose owner is no longer running (a build killed with
# SIGKILL) is taken over.
#
# build_lock_acquire <build_dir> blocks until held
# build_lock_release releases it (safe to call twice)
#
# Tunables: XCE_BUILD_LOCK_ROOT, XCE_BUILD_LOCK_POLL_SECS (default 2).
BUILD_LOCK_DIR=""
build_lock_path() {
local abs sum
abs="$(realpath -m -- "$1")"
sum="$(printf '%s' "$abs" | cksum | cut -d' ' -f1)"
printf '%s/xce-build-%s.lock' "${XCE_BUILD_LOCK_ROOT:-${TMPDIR:-/tmp}}" "$sum"
}
build_lock_acquire() {
local lock owner waited=0 poll="${XCE_BUILD_LOCK_POLL_SECS:-2}"
lock="$(build_lock_path "$1")"
while ! mkdir "$lock" 2>/dev/null; do
owner="$(cat "$lock/pid" 2>/dev/null || true)"
if [ -n "$owner" ] && ! kill -0 "$owner" 2>/dev/null; then
# The owner died holding it. If two waiters take over the same
# stale lock at once, mkdir below still lets only one of them in.
echo "=== Build lock for $1 was left by PID $owner, which is gone -- taking it over ===" >&2
rm -rf -- "$lock"
continue
fi
if [ "$waited" = 0 ]; then
echo "=== Waiting for another xce-build.sh run (PID ${owner:-?}) to finish with $1 ... ===" >&2
waited=1
fi
sleep "$poll"
done
echo "${BASHPID:-$$}" > "$lock/pid"
BUILD_LOCK_DIR="$lock"
}
build_lock_release() {
if [ -n "$BUILD_LOCK_DIR" ]; then
rm -rf -- "$BUILD_LOCK_DIR"
BUILD_LOCK_DIR=""
fi
}