Part 2 of 3: Moving tools/llvm to pnacl/

tools/llvm/utman-test.sh

Index: tools/llvm/utman-test.sh
===================================================================
--- tools/llvm/utman-test.sh	(revision 7025)
+++ tools/llvm/utman-test.sh	(working copy)
@@ -1,339 +0,0 @@
-#!/bin/bash
-# Copyright (c) 2011 The Native Client Authors. All rights reserved.
-# Use of this source code is governed by a BSD-style license that can be
-# found in the LICENSE file.
-#
-#@                   PNaCl Test Helper
-#@-------------------------------------------------------------------
-#@
-#@ SCons Test Usage:
-#@
-#@   utman-test.sh test-<arch>-<mode> [extra_arguments_to_scons]
-#@
-#@      Runs the SCons tests with selected arch and mode.
-#@      Valid arches:
-#@          x86-32
-#@          x86-64
-#@          arm
-#@      Available modes:
-#@          newlib (same as omitting mode)
-#@          pic
-#@          sbtc
-#@          glibc
-#@          browser
-#@          browser-glibc
-#@
-#@      For example: tools/llvm/utman-test.sh test-x86-32-glibc
-#@
-#@ The env variables: UTMAN_CONCURRENCY, UTMAN_BUILDBOT, UTMAN_DEBUG
-#@ control behavior of this script
-#@
-#@
-
-######################################################################
-# Config
-######################################################################
-
-set -o nounset
-set -o errexit
-
-# The script is located in "native_client/tools/llvm".
-# Set pwd to native_client/
-cd "$(dirname "$0")"/../..
-if [[ $(basename "$(pwd)") != "native_client" ]] ; then
-  echo "ERROR: cannot find native_client/ directory"
-  exit -1
-fi
-readonly NACL_ROOT="$(pwd)"
-
-readonly DRYRUN=${DRYRUN:-false}
-
-# This is only used by Spec2K test scripts.
-readonly PNACL_LIBMODE=${LIBMODE:-newlib}
-export PNACL_LIBMODE
-
-source tools/llvm/common-tools.sh
-SetScriptPath "${NACL_ROOT}/tools/llvm/utman-test.sh"
-SetLogDirectory "${NACL_ROOT}/toolchain/test-log"
-
-# For different levels of make parallelism change this in your env
-readonly UTMAN_CONCURRENCY=${UTMAN_CONCURRENCY:-8}
-
-readonly OTHER_TEST_SCRIPT="${NACL_ROOT}/buildbot/buildbot_pnacl.sh"
-
-######################################################################
-######################################################################
-#
-#   < TESTING >
-#
-######################################################################
-######################################################################
-
-# TODO(robertm): figure out what to do about concurrency in debug mode.
-# Perhaps it is fine just tweaking that via UTMAN_CONCURRENCY.
-if ${UTMAN_DEBUG} || ${UTMAN_BUILDBOT}; then
-  readonly SCONS_ARGS=(MODE=nacl,opt-host
-                       bitcode=1
-                       --verbose
-                       -j${UTMAN_CONCURRENCY})
-else
-  readonly SCONS_ARGS=(MODE=nacl,opt-host
-                       bitcode=1
-                       naclsdk_validate=0
-                       sysinfo=0
-                       -j${UTMAN_CONCURRENCY})
-fi
-
-#@ show-tests            - see what tests can be run
-show-tests() {
-  StepBanner "SHOWING TESTS"
-  cat $(find tests -name nacl.scons) | grep -o 'run_[A-Za-z_-]*' | sort | uniq
-}
-
-Run() {
-  if ${DRYRUN}; then
-    echo "$@"
-  else
-    "$@"
-  fi
-}
-
-RunScons() {
-  local arch="$1"
-  shift 1
-  Run ./scons "${SCONS_ARGS[@]}" platform=${arch} "$@"
-}
-
-# Returns true if the arguments specify a test
-# or target name to SCons.
-has-target-name() {
-  while [ $# -gt 0 ]; do
-    # Skip arguments of the form -foo, --foo, or foo=bar.
-    if [[ "$1" =~ ^-.* ]] || [[ "$1" =~ = ]]; then
-      shift 1
-      continue
-    fi
-    return 0
-  done
-  return 1
-}
-
-scons-clean () {
-  local arch=$1
-  local mode=$2
-  local frontend=clang
-  if [ "${mode}" == "newlib" ] ; then
-    Run rm -rf scons-out/nacl-${arch}-pnacl-${frontend}
-  else
-    Run rm -rf scons-out/nacl-${arch}-pnacl-${mode}-${frontend}
-  fi
-}
-
-build-sbtc-prerequisites() {
-  local arch=$1
-  # Sandboxed translators currently only require irt_core since they do not
-  # use PPAPI.
-  RunScons ${arch} sel_ldr sel_universal irt_core
-}
-
-# TODO(pdox):
-# "mode" is presently a combination of multiple bits:
-# Newlib vs. GLibC, sandboxed vs unsandboxed, PIC vs non-PIC
-# These options should be isolated and kept separate.
-get-mode-flags() {
-  local mode="$1"
-  local modeflags=""
-  case ${mode} in
-  newlib)
-    ;;
-  sbtc)
-    modeflags="use_sandboxed_translator=1"
-    ;;
-  pic)
-    modeflags="nacl_pic=1"
-    ;;
-  glibc)
-    modeflags="--nacl_glibc"
-    ;;
-  *)
-    echo "Unknown mode" 1>&2
-    exit -1
-  esac
-  echo ${modeflags}
-}
-
-scons-tests () {
-  local arch="$1"
-  local mode="$2"
-  shift 2
-  scons-clean ${arch} ${mode}
-
-  if [ ${mode} == "sbtc" ]; then
-    build-sbtc-prerequisites "${arch}"
-  fi
-
-  local modeflags=$(get-mode-flags ${mode})
-
-  if has-target-name "$@" ; then
-    RunScons ${arch} ${modeflags} "$@"
-  else
-    RunScons ${arch} ${modeflags} "$@"
-    RunScons ${arch} ${modeflags} "$@" smoke_tests
-  fi
-}
-
-browser-tests() {
-  local arch="$1"
-  local extra="$2"
-  # This calls out to the other buildbot test script. We should have
-  # buildbot_toolchain_arm_untrusted.sh use the same tests directly.
-  # TODO(jvoung): remove these when unified.
-  # scons browser-tests are currently broken with concurrency, so use -j1
-  # BUG= http://code.google.com/p/nativeclient/issues/detail?id=2019
-  Run ${OTHER_TEST_SCRIPT} browser-tests \
-    "${arch}" "--mode=opt-host,nacl -j1 ${extra}"
-}
-
-test-arm()        { scons-tests arm newlib "$@" ; }
-test-x86-32()     { scons-tests x86-32 newlib "$@" ; }
-test-x86-64()     { scons-tests x86-64 newlib "$@" ; }
-
-test-arm-newlib()    { scons-tests arm newlib "$@" ; }
-test-x86-32-newlib() { scons-tests x86-32 newlib "$@" ; }
-test-x86-64-newlib() { scons-tests x86-64 newlib "$@" ; }
-
-test-arm-pic()    { scons-tests arm pic "$@" ; }
-test-x86-32-pic() { scons-tests x86-32 pic "$@" ; }
-test-x86-64-pic() { scons-tests x86-64 pic "$@" ; }
-
-test-arm-sbtc()    { scons-tests arm sbtc "$@" ; }
-test-x86-32-sbtc() { scons-tests x86-32 sbtc "$@" ; }
-test-x86-64-sbtc() { scons-tests x86-64 sbtc "$@" ; }
-
-test-arm-glibc()    { scons-tests arm glibc "$@" ; }
-test-x86-32-glibc() { scons-tests x86-32 glibc "$@" ; }
-test-x86-64-glibc() { scons-tests x86-64 glibc "$@" ; }
-
-test-arm-browser()    { browser-tests "arm" "" ; }
-test-x86-32-browser() { browser-tests "x86-32" "" ; }
-test-x86-64-browser() { browser-tests "x86-64" "" ; }
-
-test-arm-browser-glibc() { browser-tests "arm" "--nacl_glibc" ; }
-test-x86-32-browser-glibc() { browser-tests "x86-32" "--nacl_glibc" ; }
-test-x86-64-browser-glibc() { browser-tests "x86-64" "--nacl_glibc" ; }
-
-#@
-#@ test-all  - Run arm, x86-32, and x86-64 tests. (all should pass)
-test-all() {
-  if [ $# -ne 0 ]; then
-    echo "test-all does not take any arguments"
-    exit -1
-  fi
-
-  FAIL_FAST=true ${OTHER_TEST_SCRIPT} mode-test-all ${UTMAN_CONCURRENCY}
-}
-
-#@
-#@ test-spec <official-spec-dir> <setup> [ref|train] [<benchmarks>]*
-#@                       - run spec tests
-test-spec() {
-  if [[ $# -lt 2 ]]; then
-    echo "not enough arguments for test-spec"
-    exit 1
-  fi;
-  official=$(GetAbsolutePath $1)
-  setup=$2
-  shift 2
-  spushd tests/spec2k
-  ./run_all.sh BuildPrerequisitesSetupBased ${setup}
-  ./run_all.sh CleanBenchmarks "$@"
-  ./run_all.sh PopulateFromSpecHarness ${official} "$@"
-  ./run_all.sh BuildAndRunBenchmarks ${setup} "$@"
-  spopd
-}
-
-#+ CollectTimingInfo <directory> <timing_result_file> <tagtype...>
-#+  CD's into the directory in a subshell and collects all the
-#+  relevant timed run info
-#+  tagtype just gets printed out.
-CollectTimingInfo() {
-  wd=$1
-  result_file=$2
-  setup=$3
-  (cd ${wd};
-   mkdir -p $(dirname ${result_file})
-   echo "##################################################" >>${result_file}
-   date +"# Completed at %F %H:%M:%S %A ${result_file}" >> ${result_file}
-   echo "# " ${wd}
-   echo "#" $(uname -a) >> ${result_file}
-   echo "# SETUP: ${setup}" >>${result_file}
-   echo "##################################################" >>${result_file}
-   echo "# COMPILE " >> ${result_file}
-   for ff in $(find . -name "*.compile_time"); do
-     cat ${ff} >> ${result_file}
-   done
-   echo "# RUN " >> ${result_file}
-   for ff in $(find . -name "*.run_time"); do
-     cat ${ff} >> ${result_file}
-   done
-   cat ${result_file}
-  )
-}
-
-#@
-#@ timed-test-spec <result-file> <official-spec-dir> <setup> ... - run spec and
-#@  measure time / size data. Data is emitted to stdout, but also collected
-#@  in <result-file>. <result-file> is not cleared across runs (but temp files
-#@  are cleared on each run).
-#@  Note that the VERIFY variable effects the timing!
-timed-test-spec() {
-  if ${BUILD_PLATFORM_MAC} ; then
-    echo "Timed-test-spec is not currently supported on MacOS"
-    exit -1
-  fi
-  if [ "$#" -lt "3" ]; then
-    echo "timed-test-spec {result-file} {spec2krefdir} {setupfunc}" \
-         "[ref|train] [benchmark]*"
-    exit 1
-  fi
-  result_file=$1
-  official=$(GetAbsolutePath $2)
-  setup=$3
-  shift 3
-  spushd tests/spec2k
-  ./run_all.sh BuildPrerequisitesSetupBased ${setup}
-  ./run_all.sh CleanBenchmarks "$@"
-  ./run_all.sh PopulateFromSpecHarness ${official} "$@"
-  ./run_all.sh TimedBuildAndRunBenchmarks ${setup} "$@"
-  CollectTimingInfo $(pwd) ${result_file} ${setup}
-  spopd
-}
-
-#@ help                  - Usage information.
-help() {
-  Usage
-}
-
-#@ help-full             - Usage information including internal functions.
-help-full() {
-  Usage2
-}
-
-######################################################################
-######################################################################
-#
-#                               < MAIN >
-#
-######################################################################
-######################################################################
-
-[ $# = 0 ] && set -- help  # Avoid reference to undefined $1.
-if [ "$(type -t $1)" != "function" ]; then
-  #Usage
-  echo "ERROR: unknown function '$1'." >&2
-  echo "For help, try:"
-  echo "    $0 help"
-  exit 1
-fi
-
-"$@"