Tool to benchmark tar with reflnk support

tar-reflink-bench.sh

#!/bin/sh
# Benchmark tar --reflink extraction against a regular one.
#
# usage: bench.sh <workdir-on-target-fs> <archive>...
#
# Each archive is extracted RUNS times (default 3) in both modes, at
# cold cache when possible, timing tar plus the sync of the extracted
# data; the best run is reported.  Contents and extent sharing are
# verified once per archive.
#
# TAR=/path/to/tar overrides the tar binary (default ~/src/tar/src/tar).

set -e

TAR=${TAR:-$HOME/src/tar/src/tar}
RUNS=${RUNS:-3}

dir=${1:?usage: $0 <workdir-on-target-fs> <archive>...}
shift
test $# -gt 0 || { echo "no archives given" >&2; exit 1; }
dir=$(cd "$dir" && pwd) || exit 1

uname -r
stat -f -c 'fstype=%T bsize=%S' "$dir"
"$TAR" --version | head -1
echo 3 | sudo -n tee /proc/sys/vm/drop_caches >/dev/null 2>&1 ||
	echo "note: no sudo, evicting the archive from cache with dd iflag=nocache instead"
echo

# evict the archive ($1) from the page cache; global drop with sudo,
# posix_fadvise(DONTNEED) on the archive alone otherwise, via dd
drop_caches()
{
	sync
	echo 3 | sudo -n tee /proc/sys/vm/drop_caches >/dev/null 2>&1 ||
		dd if="$1" iflag=nocache count=0 status=none 2>/dev/null || true
}

now()
{
	date +%s.%N
}

# bench_one <archive> <mode: normal|reflink> -> prints best seconds
bench_one()
{
	best=""
	i=0
	while [ $i -lt "$RUNS" ]; do
		rm -rf "$dir/x"
		mkdir "$dir/x"
		drop_caches "$1"
		t0=$(now)
		case $2 in
		reflink) (cd "$dir/x" && "$TAR" -xf "$1" --reflink) ;;
		*)       (cd "$dir/x" && "$TAR" -xf "$1") ;;
		esac
		sync -f "$dir/x"
		t1=$(now)
		t=$(awk "BEGIN { printf \"%.3f\", $t1 - $t0 }")
		best=$(awk "BEGIN { b = \"$best\"; print (b == \"\" || $t < b) ? $t : b }")
		i=$((i + 1))
	done
	echo "$best"
}

printf '%-28s %10s %10s %9s\n' archive normal reflink speedup

for a in "$@"; do
	# downloaded archives are not aligned: extract the content
	# once and repack it with --reflink on the target fs
	case $a in /*) ;; *) a=$PWD/$a ;; esac
	rl=$dir/$(basename "$a" .tar).reflink.tar
	if [ ! -f "$rl" ]; then
		echo "$(basename "$a"): repacking with --reflink..." >&2
		rm -rf "$dir/unpack"
		mkdir "$dir/unpack"
		(cd "$dir/unpack" && "$TAR" -xf "$a")
		(cd "$dir/unpack" && "$TAR" -cf "$rl" --reflink .)
		rm -rf "$dir/unpack"
	fi
	a=$rl

	# correctness: contents must match and extents must be shared
	# (skip with VERIFY=0)
	if [ "${VERIFY:-1}" != 0 ]; then
		echo "$(basename "$a"): verifying contents and sharing..." >&2
		rm -rf "$dir/x" "$dir/ref"
		mkdir "$dir/x" "$dir/ref"
		(cd "$dir/ref" && "$TAR" -xf "$a")
		(cd "$dir/x" && "$TAR" -xf "$a" --reflink)
		diff -r "$dir/ref" "$dir/x" > /dev/null || { echo "FAIL: content mismatch for $a" >&2; exit 1; }
		sync
		f=$(find "$dir/x" -type f -size +4k | head -1)
		if [ -n "$f" ] && command -v filefrag > /dev/null; then
			filefrag -v "$f" | grep -qi shared || {
				echo "FAIL: no shared extents for $a: the reflink" \
				     "extraction fell back to a copy, timings would" \
				     "be meaningless" >&2
				exit 1
			}
		fi
		rm -rf "$dir/ref"
	fi

	echo "$(basename "$a"): benchmarking ($RUNS+$RUNS runs)..." >&2
	n=$(bench_one "$a" normal)
	r=$(bench_one "$a" reflink)
	printf '%-28s %9ss %9ss %8sx\n' "$(basename "$a")" "$n" "$r" \
		"$(awk "BEGIN { printf \"%.1f\", $n / $r }")"
done

rm -rf "$dir/x"