[benchmarks/time] Implement time-helper.c

To fix the "bash invokes dash" problem, and the "/usr/bin/time has bad
precision" problem.

(Although maybe that is fixed with a newer version?)

asdl/run.sh

@@ -49,7 +49,6 @@ osh-both() { py-cpp osh/osh.asdl; }
 #
 
 cxx() {
-  #local CXX=c++ 
   local CXX=$CLANGXX
   local opt_flag='-O2'
   #local opt_flag='-O0'

benchmarks/time-helper.c

@@ -0,0 +1,166 @@
+#define _GNU_SOURCE  // for timersub()
+#include <assert.h>
+#include <getopt.h>
+#include <errno.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>  // exit()
+#include <sys/resource.h>  // getrusage()
+#include <sys/time.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+void die_errno(const char *message) {
+  perror(message);
+  exit(1);
+}
+
+void die(const char *message) {
+  fprintf(stderr, "time-helper: %s\n", message);
+  exit(1);
+}
+
+typedef struct Spec_t {
+  char* out_path;
+  bool append;
+
+  char delimiter;  // delimiter should be tab or ,
+  bool verbose;    // whether to show verbose logging
+
+  bool x;  // %x status
+  bool e;  // %e elapsed
+  bool U;  // %U user time
+  bool S;  // %S system time
+  bool M;  // %M maxrss
+  int argc;
+  char **argv;
+} Spec;
+
+
+// Write CSV/TSV cells of different types
+void int_cell(FILE* f, char delimiter, int val) {
+  if (delimiter != 0) {  // NUL is invalid delimiter
+    fprintf(f, "%c%d", delimiter, val);
+  } else {
+    fprintf(f, "%d", val);
+  }
+}
+
+void time_cell(FILE* f, char delimiter, struct timeval *val) {
+  fprintf(f, "%c%ld.%06ld", delimiter, val->tv_sec, val->tv_usec);
+}
+
+int time_helper(Spec *spec, FILE* f) {
+  char *prog = spec->argv[0];
+
+  struct timeval start;
+  struct timeval end;
+
+  int status = 0;
+  switch (fork()) {
+  case -1:
+    die_errno("fork");
+    break;
+
+  case 0:  // child exec
+    if (execvp(prog, spec->argv) < 0) {
+      fprintf(stderr, "time-helper: error executing '%s'\n", prog);
+      die_errno("execvp");
+    }
+    assert(0);  // execvp() never returns
+
+  default:  // parent measures elapsed time of child
+    if (gettimeofday(&start, NULL) < 0) {
+      die_errno("gettimeofday");
+    }
+    wait(&status);
+    if (gettimeofday(&end, NULL) < 0) {
+      die_errno("gettimeofday");
+    }
+    break;
+  }
+  //fprintf(stderr, "done waiting\n");
+
+  struct timeval elapsed;
+  timersub(&end, &start, &elapsed);
+
+  struct rusage usage;
+  getrusage(RUSAGE_CHILDREN, &usage);
+
+  struct timeval *user = &usage.ru_utime;
+  struct timeval *sys  = &usage.ru_stime;
+
+  char d = spec->delimiter;
+  // NO delimiter at first!
+  if (spec->x) { int_cell(f, 0, WEXITSTATUS(status)); }
+  if (spec->e) { time_cell(f, d, &elapsed); }
+  if (spec->U) { time_cell(f, d, &usage.ru_utime); }
+  if (spec->S) { time_cell(f, d, &usage.ru_stime); }
+  if (spec->M) { int_cell(f, d, usage.ru_maxrss); }
+
+  return status;
+}
+
+int main(int argc, char **argv) {
+  Spec spec = {0};
+
+  spec.out_path = "/dev/null";  // default value
+
+  // http://www.gnu.org/software/libc/manual/html_node/Example-of-Getopt.html
+  // + means to be strict about flag parsing.
+  char c;
+  while ((c = getopt(argc, argv, "+o:ad:vxeUSM")) != -1) {
+    switch (c) {
+    case 'o':
+      spec.out_path = optarg;
+      break;
+    case 'a':
+      spec.append = true;
+      break;
+
+    case 'd':
+      spec.delimiter = optarg[0];
+      break;
+    case 'v':
+      spec.verbose = true;
+      break;
+
+    case 'x':
+      spec.x = true;
+      break;
+    case 'e':
+      spec.e = true;
+      break;
+    case 'U':
+      spec.U = true;
+      break;
+    case 'S':
+      spec.S = true;
+      break;
+    case 'M':
+      spec.M = true;
+      break;
+
+    case '?':  // getopt library will print error
+      return 2;
+
+    default:
+      abort();  // should never happen
+    }
+  }
+
+  int a = optind;  // index into argv
+  if (a == argc) {
+    die("expected a command to run");
+  }
+
+  spec.argv = argv + a;
+  spec.argc = argc - a;
+
+  char* mode = spec.append ? "a" : "w";
+  FILE* f = fopen(spec.out_path, mode);
+  int status = time_helper(&spec, f);
+  fclose(f);
+
+  return status;
+}

build/dev.sh

@@ -321,12 +321,49 @@ demo-grammar() {
   oil_lang/grammar_gen.py marshal mycpp/examples/arith.pgen2 _devbuild/gen
 }
 
+time-helper() {
+  local out=_devbuild/bin
+  mkdir -p $out
+  cc -std=c99 -o $out/time-helper benchmarks/time-helper.c
+  set +o errexit
+
+  local tmp=_tmp/time-helper.txt
+
+  # Make some work show up
+  local cmd='{ md5sum */*.md; sleep 0.15; exit 42; } > /dev/null'
+
+  echo 'will be overwritten' > $tmp
+  cat $tmp
+
+  $out/time-helper
+  echo status=$?
+
+  $out/time-helper /bad
+  echo status=$?
+
+  $out/time-helper -o $tmp -d $'\t' -x -e -- sh -c "$cmd"
+  echo status=$?
+  cat $tmp
+  echo
+
+  # Now append
+  $out/time-helper -o $tmp -a -d , -x -e -U -S -M -- sh -c "$cmd"
+  echo status=$?
+  cat $tmp
+  echo
+
+  # Error case
+  $out/time-helper -z
+  echo status=$?
+}
+
 # Prerequisites: build/codegen.sh {download,install}-re2c
 all() {
   rm -f *.so  # 12/2019: to clear old symlinks, maybe get rid of
 
   _minimal
   fastlex
+  time-helper
   build/doc.sh all-help
 }
 

test/lint.sh

@@ -47,7 +47,9 @@ clang-format() {
 
 cpp-files() {
   shopt -s nullglob
-  for file in asdl/*.cc cpp/*.{cc,h} mycpp/*.{cc,h} mycpp/demo/*.{cc,h}; do
+  for file in asdl/*.cc benchmarks/*.c cpp/*.{cc,h} mycpp/*.{cc,h} \
+    mycpp/demo/*.{cc,h}; do
+
     echo $file
   done
 }