| 1 | #!/usr/bin/env bash
|
| 2 | #
|
| 3 | # Measure the number of syscalls that shells use.
|
| 4 | #
|
| 5 | # Usage:
|
| 6 | # test/syscall.sh <function name>
|
| 7 |
|
| 8 | set -o nounset
|
| 9 | set -o pipefail
|
| 10 | set -o errexit
|
| 11 |
|
| 12 | source build/dev-shell.sh
|
| 13 |
|
| 14 | readonly -a SHELLS=(dash bash mksh zsh ash yash osh)
|
| 15 |
|
| 16 | readonly BASE_DIR='_tmp/syscall' # What we'll publish
|
| 17 | readonly RAW_DIR='_tmp/syscall-raw' # Raw data
|
| 18 |
|
| 19 | # Run it against the dev version of OSH
|
| 20 | REPO_ROOT=$(cd "$(dirname $0)/.."; pwd)
|
| 21 |
|
| 22 | count-procs() {
|
| 23 | local out_prefix=$1
|
| 24 | local sh=$2
|
| 25 | shift 2
|
| 26 |
|
| 27 | case $sh in
|
| 28 | # avoid the extra processes that bin/osh starts!
|
| 29 | # relies on word splitting
|
| 30 | #(X) # to compare against osh 0.8.pre3 installed
|
| 31 | (osh)
|
| 32 | sh="env PYTHONPATH=$REPO_ROOT:$REPO_ROOT/vendor $REPO_ROOT/bin/oil.py osh"
|
| 33 | ;;
|
| 34 | esac
|
| 35 |
|
| 36 | strace -ff -o $out_prefix -- $sh "$@"
|
| 37 | }
|
| 38 |
|
| 39 | run-case() {
|
| 40 | ### Run a test case with many shells
|
| 41 |
|
| 42 | local num=$1
|
| 43 | local code_str=$2
|
| 44 |
|
| 45 | for sh in "${SHELLS[@]}"; do
|
| 46 | local out_prefix=$RAW_DIR/$num-$sh
|
| 47 | echo "--- $sh"
|
| 48 | count-procs $out_prefix $sh -c "$code_str"
|
| 49 | done
|
| 50 | }
|
| 51 |
|
| 52 | run-case-file() {
|
| 53 | ### Like the above, but the shell reads from a file
|
| 54 |
|
| 55 | local num=$1
|
| 56 | local code_str=$2
|
| 57 |
|
| 58 | echo -n "$code_str" > _tmp/$num.sh
|
| 59 |
|
| 60 | for sh in "${SHELLS[@]}"; do
|
| 61 | local out_prefix=$RAW_DIR/$num-$sh
|
| 62 | echo "--- $sh"
|
| 63 | count-procs $out_prefix $sh _tmp/$num.sh
|
| 64 | done
|
| 65 | }
|
| 66 |
|
| 67 | run-case-stdin() {
|
| 68 | ### Like the above, but read from a pipe
|
| 69 |
|
| 70 | local num=$1
|
| 71 | local code_str=$2
|
| 72 |
|
| 73 | for sh in "${SHELLS[@]}"; do
|
| 74 | local out_prefix=$RAW_DIR/$num-$sh
|
| 75 | echo "--- $sh"
|
| 76 | echo -n "$code_str" | count-procs $out_prefix $sh
|
| 77 | done
|
| 78 | }
|
| 79 |
|
| 80 |
|
| 81 | print-cases() {
|
| 82 | # format: number, whitespace, then an arbitrary code string
|
| 83 | egrep -v '^[[:space:]]*(#|$)' <<EOF
|
| 84 |
|
| 85 | # builtin
|
| 86 | echo hi
|
| 87 |
|
| 88 | # external command
|
| 89 | date
|
| 90 |
|
| 91 | # Oil sentence
|
| 92 | date ;
|
| 93 |
|
| 94 | # external then builtin
|
| 95 | date; echo hi
|
| 96 |
|
| 97 | # builtin then external
|
| 98 | echo hi; date
|
| 99 |
|
| 100 | # two external commands
|
| 101 | date; date
|
| 102 |
|
| 103 | # does a brace group make a difference?
|
| 104 | { date; date; }
|
| 105 |
|
| 106 | # singleton brace group
|
| 107 | date; { date; }
|
| 108 |
|
| 109 | # does it behave differently if sourced?
|
| 110 | . _tmp/sourced.sh
|
| 111 |
|
| 112 | # dash and zsh somehow optimize this to 1
|
| 113 | (echo hi)
|
| 114 |
|
| 115 | (date)
|
| 116 |
|
| 117 | ( ( date ) )
|
| 118 |
|
| 119 | ( ( date ) ); echo hi
|
| 120 |
|
| 121 | echo hi; (date)
|
| 122 |
|
| 123 | # Sentence in Oil
|
| 124 | (date;) > /tmp/out.txt
|
| 125 |
|
| 126 | (date; echo hi)
|
| 127 |
|
| 128 | # command sub
|
| 129 | echo \$(date)
|
| 130 |
|
| 131 | # command sub with builtin
|
| 132 | echo \$(echo hi)
|
| 133 |
|
| 134 | # command sub with useless subshell (some scripts use this)
|
| 135 | echo \$( ( date ) )
|
| 136 |
|
| 137 | # command sub with other subshell
|
| 138 | echo \$( ( date ); echo hi )
|
| 139 |
|
| 140 | # 2 processes for all shells
|
| 141 | ( echo hi ); echo done
|
| 142 |
|
| 143 | # simple pipeline
|
| 144 | date | wc -l
|
| 145 |
|
| 146 | # every shell does 3
|
| 147 | echo a | wc -l
|
| 148 |
|
| 149 | # every shell does 3
|
| 150 | command echo a | wc -l
|
| 151 |
|
| 152 | # bash does 4 here!
|
| 153 | command date | wc -l
|
| 154 |
|
| 155 | # 3 processes for all?
|
| 156 | # osh gives FIVE??? But others give 3. That's bad.
|
| 157 | ( date ) | wc -l
|
| 158 |
|
| 159 | # 3 processes for all shells except zsh and osh, which have shopt -s lastpipe!
|
| 160 | date | read x
|
| 161 |
|
| 162 | # osh has 3, but should be 2 like zsh?
|
| 163 | # hm how can zsh do 2 here? That seems impossible.
|
| 164 | # oh it's lastpipe turns the shell process into wc -l ??? wow.
|
| 165 | { echo a; echo b; } | wc -l
|
| 166 |
|
| 167 | # zsh behaves normally here. That is a crazy optimization. I guess it's
|
| 168 | # nice when you have SH -c 'mypipeline | wc-l'
|
| 169 | { echo a; echo b; } | wc -l; echo done
|
| 170 |
|
| 171 | # this is all over the map too. 3 4 4 2.
|
| 172 | { echo a; date; } | wc -l
|
| 173 |
|
| 174 | # osh does 4 when others do 3. So every shell optimizes this extra pipeline.
|
| 175 | ( echo a; echo b ) | wc -l
|
| 176 |
|
| 177 | # osh does 5 when others do 3.
|
| 178 | ( echo a; echo b ) | ( wc -l )
|
| 179 | EOF
|
| 180 |
|
| 181 | # Discarded because they're identical
|
| 182 | # pipeline with redirect last
|
| 183 | #date | wc -l > /tmp/out.txt
|
| 184 |
|
| 185 | # pipeline with redirect first
|
| 186 | #date 2>&1 | wc -l
|
| 187 |
|
| 188 | }
|
| 189 |
|
| 190 | number-cases() {
|
| 191 | # Right justified, leading zeros, with 2
|
| 192 | # Wish this was %02d
|
| 193 | print-cases | nl --number-format rz --number-width 2
|
| 194 | }
|
| 195 |
|
| 196 | by-input() {
|
| 197 | ### Run cases that vary by input reader
|
| 198 | if ! strace true; then
|
| 199 | echo "Aborting because we couldn't run strace"
|
| 200 | return
|
| 201 | fi
|
| 202 |
|
| 203 | local suite='by-input'
|
| 204 |
|
| 205 | rm -r -f -v $RAW_DIR
|
| 206 | mkdir -p $RAW_DIR
|
| 207 |
|
| 208 | # Wow this newline makes a difference in shells!
|
| 209 |
|
| 210 | # This means that Id.Eof_Real is different than Id.Op_Newline?
|
| 211 | # Should we create a Sentence for it too then?
|
| 212 | # That is possible in _ParseCommandLine
|
| 213 |
|
| 214 | zero=$'date; date'
|
| 215 | one=$'date; date\n'
|
| 216 | two=$'date; date\n#comment\n'
|
| 217 | comment=$'# comment\ndate;date'
|
| 218 | newline=$'date\n\ndate'
|
| 219 | newline2=$'date\n\ndate\n#comment'
|
| 220 |
|
| 221 | # zsh is the only shell to optimize all 6 cases! 2 processes instead of 3.
|
| 222 | run-case 30 "$zero"
|
| 223 | run-case 31 "$one"
|
| 224 | run-case 32 "$two"
|
| 225 | run-case 33 "$comment"
|
| 226 | run-case 34 "$newline"
|
| 227 | run-case 35 "$newline2"
|
| 228 |
|
| 229 | run-case-file 40 "$zero"
|
| 230 | run-case-file 41 "$one"
|
| 231 | run-case-file 42 "$two"
|
| 232 | run-case-file 43 "$comment"
|
| 233 | run-case-file 44 "$newline2"
|
| 234 | run-case-file 45 "$newline2"
|
| 235 |
|
| 236 | # yash is the only shell to optimize the stdin case at all!
|
| 237 | # it looks for a lack of trailing newline.
|
| 238 | run-case-stdin 50 "$zero"
|
| 239 | run-case-stdin 51 "$one"
|
| 240 | run-case-stdin 52 "$two"
|
| 241 | run-case-stdin 53 "$comment"
|
| 242 | run-case-stdin 54 "$newline2"
|
| 243 | run-case-stdin 55 "$newline2"
|
| 244 |
|
| 245 | # This is identical for all shells
|
| 246 | #run-case 32 $'date; date\n#comment\n'
|
| 247 |
|
| 248 | cat >$BASE_DIR/${suite}-cases.txt <<EOF
|
| 249 | 30 -c: zero lines
|
| 250 | 31 -c: one line
|
| 251 | 32 -c: one line and comment
|
| 252 | 33 -c: comment first
|
| 253 | 34 -c: newline
|
| 254 | 35 -c: newline2
|
| 255 | 40 file: zero lines
|
| 256 | 41 file: one line
|
| 257 | 42 file: one line and comment
|
| 258 | 43 file: comment first
|
| 259 | 44 file: newline
|
| 260 | 45 file: newline2
|
| 261 | 50 stdin: zero lines
|
| 262 | 51 stdin: one line
|
| 263 | 52 stdin: one line and comment
|
| 264 | 53 stdin: comment first
|
| 265 | 54 stdin: newline
|
| 266 | 55 stdin: newline2
|
| 267 | EOF
|
| 268 |
|
| 269 | count-lines $suite
|
| 270 | summarize $suite 3 0
|
| 271 |
|
| 272 | }
|
| 273 |
|
| 274 | # Quick hack: every shell uses 2 processes for this... doesn't illuminate much.
|
| 275 | weird-command-sub() {
|
| 276 | shopt -s nullglob
|
| 277 | rm -r -f -v $RAW_DIR/*
|
| 278 |
|
| 279 | local tmp=_tmp/cs
|
| 280 | echo FOO > $tmp
|
| 281 | run-case 60 "echo $(< $tmp)"
|
| 282 | run-case 61 "echo $(< $tmp; echo hi)"
|
| 283 |
|
| 284 | local suite=weird-command-sub
|
| 285 |
|
| 286 | cat >$BASE_DIR/${suite}-cases.txt <<EOF
|
| 287 | 60 \$(< file)
|
| 288 | 61 \$(< file; echo hi)
|
| 289 | EOF
|
| 290 |
|
| 291 | count-lines $suite
|
| 292 | summarize $suite 0 0
|
| 293 | }
|
| 294 |
|
| 295 | readonly MAX_CASES=100
|
| 296 | #readonly MAX_CASES=3
|
| 297 |
|
| 298 | by-code() {
|
| 299 | ### Run cases that vary by code snippet
|
| 300 |
|
| 301 | if ! strace true; then
|
| 302 | echo "Aborting because we couldn't run strace"
|
| 303 | return
|
| 304 | fi
|
| 305 |
|
| 306 | local max_cases=${1:-$MAX_CASES}
|
| 307 |
|
| 308 | rm -r -f -v $RAW_DIR
|
| 309 | mkdir -p $RAW_DIR $BASE_DIR
|
| 310 |
|
| 311 | write-sourced
|
| 312 |
|
| 313 | local suite='by-code'
|
| 314 | local cases=$BASE_DIR/${suite}-cases.txt
|
| 315 |
|
| 316 | number-cases > $cases
|
| 317 | head -n $max_cases $cases | while read -r num code_str; do
|
| 318 | echo
|
| 319 | echo '==='
|
| 320 | echo "$num $code_str"
|
| 321 | echo
|
| 322 |
|
| 323 | run-case $num "$code_str"
|
| 324 | done
|
| 325 |
|
| 326 | # omit total line
|
| 327 | count-lines $suite
|
| 328 | summarize $suite 3 0
|
| 329 | }
|
| 330 |
|
| 331 | syscall-py() {
|
| 332 | PYTHONPATH=. test/syscall.py "$@"
|
| 333 | }
|
| 334 |
|
| 335 | write-sourced() {
|
| 336 | echo -n 'date; date' > _tmp/sourced.sh
|
| 337 | }
|
| 338 |
|
| 339 | count-lines() {
|
| 340 | local suite=${1:-by-code}
|
| 341 | ( cd $RAW_DIR && wc -l * ) | head -n -1 > $BASE_DIR/${suite}-counts.txt
|
| 342 | }
|
| 343 |
|
| 344 | summarize() {
|
| 345 | local suite=${1:-by-code}
|
| 346 | local not_minimum=${2:-0}
|
| 347 | local more_than_bash=${3:-0}
|
| 348 |
|
| 349 | local out=$BASE_DIR/${suite}.txt
|
| 350 | set +o errexit
|
| 351 | cat $BASE_DIR/${suite}-counts.txt \
|
| 352 | | syscall-py --not-minimum $not_minimum --more-than-bash $more_than_bash \
|
| 353 | $BASE_DIR/${suite}-cases.txt \
|
| 354 | > $out
|
| 355 | local status=$?
|
| 356 | set -o errexit
|
| 357 |
|
| 358 | echo "Wrote $out"
|
| 359 | if test $status -eq 0; then
|
| 360 | echo 'OK'
|
| 361 | else
|
| 362 | echo 'FAIL'
|
| 363 | fi
|
| 364 | }
|
| 365 |
|
| 366 | run-for-release() {
|
| 367 | ### Run the two syscall suites
|
| 368 |
|
| 369 | # Invoked as one of the "other" tests. Soil runs by-code and by-input
|
| 370 | # separately.
|
| 371 |
|
| 372 | # Note: Only $BASE_DIR/*.txt is included in the release/$VERSION/other.wwz
|
| 373 | by-code
|
| 374 | by-input
|
| 375 |
|
| 376 | echo 'OK'
|
| 377 | }
|
| 378 |
|
| 379 | #
|
| 380 | # Real World
|
| 381 | #
|
| 382 | # $ ls|grep dash|wc -l
|
| 383 | # 6098
|
| 384 | # $ ls|grep bash|wc -l
|
| 385 | # 6102
|
| 386 | # $ ls|grep osh|wc -l
|
| 387 | # 6098
|
| 388 | #
|
| 389 | # So Oil is already at dash level for CPython's configure, and bash isn't
|
| 390 | # far off. So autoconf-generated scripts probably already use constructs
|
| 391 | # that are already "optimal" in most shells.
|
| 392 |
|
| 393 | readonly PY27_DIR=$PWD/Python-2.7.13
|
| 394 |
|
| 395 | cpython-configure() {
|
| 396 | local raw_dir=$PWD/$RAW_DIR/real
|
| 397 | mkdir -p $raw_dir
|
| 398 |
|
| 399 | pushd $PY27_DIR
|
| 400 | #for sh in "${SHELLS[@]}"; do
|
| 401 | for sh in bash dash osh; do
|
| 402 | local out_prefix=$raw_dir/cpython-$sh
|
| 403 | echo "--- $sh"
|
| 404 |
|
| 405 | # TODO: Use a different dir
|
| 406 | count-procs $out_prefix $sh -c './configure'
|
| 407 | done
|
| 408 | popd
|
| 409 | }
|
| 410 |
|
| 411 | "$@"
|