breakpad/android/common-functions.sh
digit@chromium.org fa064e215b Fix 'make check' for Android
This patch allows 'make check' to work when performing
an Automake-based build of Breakpad for Android. This
requires to have an Android device connected, and the
'adb' tool in your path.

You can test that with something like:

    configure --host=arm-linux-androideabi
    make check

This is achieved by adding a new small shell script
under android/test-shell.sh, which is invoked by the
Makefile (see TESTS_ENVIRONMENT definition in
Makefile.am).

By default, this runs all unit tests, including those
for the processor and tools (which normally never run
on an Android device).

Note that the test suites fails (e.g. 11 failing tests
for the client library). This will be addressed in later
patches.

+ Modify android/run-checks.sh to run the client library
  test suite on the device by default.

+ Add a new option (--all-tests) to android/run-checks.sh
  which forces it to run the unit test suite for the host
  binaries, and the full suite on the Android device.

+ Update README.ANDROID appropriately.
Review URL: https://breakpad.appspot.com/441002

git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@1023 4c0a9323-5329-0410-9bdc-e9ce6186880e
2012-08-31 17:07:25 +00:00

373 lines
9.4 KiB
Bash
Executable File

# Copyright (c) 2012 Google Inc.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met:
#
# * Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
# * Redistributions in binary form must reproduce the above
# copyright notice, this list of conditions and the following disclaimer
# in the documentation and/or other materials provided with the
# distribution.
# * Neither the name of Google Inc. nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
# Collection of common shell functions for 'run-checks.sh' et 'test-shell.sh'
# All internal variables and functions use an underscore as a prefix
# (e.g. _VERBOSE, _ALL_CLEANUPS, etc..).
# Sanitize the environment
export LANG=C
export LC_ALL=C
if [ "$BASH_VERSION" ]; then
set -o posix
fi
# Utility functions
_ALL_CLEANUPS=
# Register a function to be called when the script exits, even in case of
# Ctrl-C, logout, etc.
# $1: function name.
atexit () {
if [ -z "$_ALL_CLEANUPS" ]; then
_ALL_CLEANUPS=$1
# Ensure a clean exit when the script is:
# - Exiting normally (EXIT)
# - Interrupted by Ctrl-C (INT)
# - Interrupted by log out (HUP)
# - Being asked to quit nicely (TERM)
# - Being asked to quit and dump core (QUIT)
trap "_exit_cleanups \$?" EXIT INT HUP QUIT TERM
else
_ALL_CLEANUPS="$_ALL_CLEANUPS $1"
fi
}
# Called on exit if at least one function was registered with atexit
# $1: final exit status code
_exit_cleanups () {
local CLEANUP CLEANUPS
# Ignore calls to atexit during cleanups
CLEANUPS=$_ALL_CLEANUPS
_ALL_CLEANUPS=
for CLEANUP in $CLEANUPS; do
($CLEANUP)
done
exit "$@"
}
# Dump a panic message then exit.
# $1+: message
panic () {
echo "ERROR: $@" >&2
exit 1
}
# If the previous command failed, dump a panic message then exit.
# $1+: message.
fail_panic () {
if [ $? != 0 ]; then
panic "$@"
fi;
}
_VERBOSE=0
# Increase verbosity for dump/log/run/run2 functions
increase_verbosity () {
_VERBOSE=$(( $_VERBOSE + 1 ))
}
# Decrease verbosity
decrease_verbosity () {
_VERBOSE=$(( $_VERBOSE - 1 ))
}
# Returns success iff verbosity level is higher than a specific value
# $1: verbosity level
verbosity_is_higher_than () {
[ "$_VERBOSE" -gt "$1" ]
}
# Returns success iff verbosity level is lower than a specific value
# $1: verbosity level
verbosity_is_lower_than () {
[ "$_VERBOSE" -le "$1" ]
}
# Dump message to stdout, unless verbosity is < 0, i.e. --quiet was called
# $1+: message
dump () {
if [ "$_VERBOSE" -ge 0 ]; then
printf "%s\n" "$*"
fi
}
# If --verbose was used, dump a message to stdout.
# $1+: message
log () {
if [ "$_VERBOSE" -ge 1 ]; then
printf "%s\n" "$*"
fi
}
_RUN_LOG=
# Set a run log file that can be used to collect the output of commands that
# are not displayed.
set_run_log () {
_RUN_LOG=$1
}
# Run a command. Output depends on $_VERBOSE:
# $_VERBOSE <= 0: Run command, store output into the run log
# $_VERBOSE >= 1: Dump command, run it, output goest to stdout
# Note: Ideally, the command's output would go to the run log for $_VERBOSE >= 1
# but the 'tee' tool doesn't preserve the status code of its input pipe
# in case of error.
run () {
local LOGILE
if [ "$_RUN_LOG" ]; then
LOGFILE=$_RUN_LOG
else
LOGFILE=/dev/null
fi
if [ "$_VERBOSE" -ge 1 ]; then
echo "COMMAND: $@"
"$@"
else
"$@" >>$LOGFILE 2>&1
fi
}
# Same as run(), but only dump command output for $_VERBOSE >= 2
run2 () {
local LOGILE
if [ "$_RUN_LOG" ]; then
LOGFILE=$_RUN_LOG
else
LOGFILE=/dev/null
fi
if [ "$_VERBOSE" -ge 1 ]; then
echo "COMMAND: $@"
fi
if [ "$_VERBOSE" -ge 2 ]; then
"$@"
else
"$@" >>$LOGFILE 2>&1
fi
}
# Extract number of cores to speed up the builds
# Out: number of CPU cores
get_core_count () {
case $(uname -s) in
Linux)
grep -c -e '^processor' /proc/cpuinfo
;;
Darwin)
sysctl -n hw.ncpu
;;
CYGWIN*|*_NT-*)
echo $NUMBER_OF_PROCESSORS
;;
*)
echo 1
;;
esac
}
# Check for the Android ADB program.
#
# On success, return nothing, but updates internal variables so later calls to
# adb_shell, adb_push, etc.. will work. You can get the path to the ADB program
# with adb_get_program if needed.
#
# On failure, returns 1, and updates the internal adb error message, which can
# be retrieved with adb_get_error.
#
# $1: optional ADB program path.
# Return: success or failure.
_ADB=
_ADB_STATUS=
_ADB_ERROR=
adb_check () {
# First, try to find the executable in the path, or the SDK install dir.
_ADB=$1
if [ -z "$_ADB" ]; then
_ADB=$(which adb 2>/dev/null)
if [ -z "$_ADB" -a "$ANDROID_SDK_ROOT" ]; then
_ADB=$ANDROID_SDK_ROOT/platform-tools/adb
if [ ! -f "$_ADB" ]; then
_ADB=
fi
fi
if [ -z "$_ADB" ]; then
_ADB_STATUS=1
_ADB_ERROR="The Android 'adb' tool is not in your path."
return 1
fi
fi
log "Found ADB program: $_ADB"
# Check that it works correctly
local ADB_VERSION
ADB_VERSION=$("$_ADB" version 2>/dev/null)
case $ADB_VERSION in
"Android Debug Bridge "*) # Pass
log "Found ADB version: $ADB_VERSION"
;;
*) # Fail
_ADB_ERROR="Your ADB binary reports a bad version ($ADB_VERSION): $_ADB"
_ADB_STATUS=1
return 1
esac
_ADB_STATUS=0
return 0
}
# Return the path to the Android ADB program, if correctly detected.
# On failure, return the empty string.
# Out: ADB program path (or empty on failure)
# Return: success or failure.
adb_get_program () {
# Return cached value as soon as possible.
if [ -z "$_ADB_STATUS" ]; then
adb_check $1
fi
echo "$_ADB"
return $_ADB_STATUS
}
# Return the error corresponding to the last ADB function failure.
adb_get_error () {
echo "$_ADB_ERROR"
}
# Check that there is one device connected through ADB.
# In case of failure, use adb_get_error to know why this failed.
# $1: Optional adb program path
# Return: success or failure.
_ADB_DEVICE=
_ADB_DEVICE_STATUS=
adb_check_device () {
if [ "$_ADB_DEVICE_STATUS" ]; then
return $_ADB_DEVICE_STATUS
fi
# Check for ADB.
if ! adb_check $1; then
_ADB_DEVICE_STATUS=$_ADB_STATUS
return 1
fi
local ADB_DEVICES NUM_DEVICES FINGERPRINT
# Count the number of connected devices.
ADB_DEVICES=$("$_ADB" devices 2>/dev/null | awk '$2 == "device" { print $1; }')
NUM_DEVICES=$(echo "$ADB_DEVICES" | wc -l)
case $NUM_DEVICES in
0)
_ADB_ERROR="No Android device connected. Please connect one to your machine."
_ADB_DEVICE_STATUS=1
return 1
;;
1) # Pass
# Ensure the same device will be called in later adb_shell calls.
export ANDROID_SERIAL=$ADB_DEVICES
;;
*) # 2 or more devices.
if [ "$ANDROID_SERIAL" ]; then
ADB_DEVICES=$ANDROID_SERIAL
NUM_DEVICES=1
else
_ADB_ERROR="More than one Android device connected. \
Please define ANDROID_SERIAL in your environment"
_ADB_DEVICE_STATUS=1
return 1
fi
;;
esac
_ADB_DEVICE_STATUS=0
_ADB_DEVICE=$ADB_DEVICES
FINGERPRINT=$(adb_shell getprop ro.build.fingerprint)
log "Using ADB device: $ANDROID_SERIAL ($FINGERPRINT)"
return 0
}
# The 'adb shell' command is pretty hopeless, try to make sense of it by:
# 1/ Removing trailing \r from line endings.
# 2/ Ensuring the function returns the command's status code.
#
# $1+: Command
# Out: command output (stdout + stderr combined)
# Return: command exit status
adb_shell () {
local RET ADB_LOG
# Check for ADB device.
adb_check_device || return 1
ADB_LOG=$(mktemp "${TMPDIR:-/tmp}/adb-XXXXXXXX")
"$_ADB" shell "$@" ";" echo \$? > "$ADB_LOG" 2>&1
sed -i -e 's![[:cntrl:]]!!g' "$ADB_LOG" # Remove \r.
RET=$(sed -e '$!d' "$ADB_LOG") # Last line contains status code.
sed -e '$d' "$ADB_LOG" # Print everything except last line.
rm -f "$ADB_LOG"
return $RET
}
# Push a file to a device.
# $1: source file path
# $2: device target file path
# Return: success or failure.
adb_push () {
adb_check_device || return 1
run "$_ADB" push "$1" "$2"
}
# Pull a file from a device
# $1: device file path
# $2: target host file path
# Return: success or failure.
adb_pull () {
adb_check_device || return 1
run "$_ADB" pull "$1" "$2"
}
# Same as adb_push, but will panic if the operations didn't succeed.
adb_install () {
adb_push "$@"
fail_panic "Failed to install $1 to the Android device at $2"
}