267 lines · bash
1#!/bin/bash2##===- bolt/utils/bughunter.sh - Help locate BOLT bugs -------*- Script -*-===##3#4# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.5# See https://llvm.org/LICENSE.txt for license information.6# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception7# details.8#9##===----------------------------------------------------------------------===##10#11# This script attempts to narrow down llvm-bolt bug to a single function in the12# input binary.13#14# If such a function is found, llvm-bolt could be run just on this function15# to mitigate debugging process.16#17# The following envvars are used by this script:18#19# BOLT - path to llvm-bolt20#21# BOLT_OPTIONS - options to be used by llvm-bolt22#23# INPUT_BINARY - input for llvm-bolt24#25# PRE_COMMAND - command to execute prior to running optimized binary26#27# POST_COMMAND - command to filter results of running optimized binary28#29# TIMEOUT_OR_CMD - optional timeout or command on optimized binary command30# if the value is a number with an optional trailing letter31# [smhd] it is considered a parameter to "timeout",32# otherwise it's a shell command that wraps the optimized33# binary command.34#35# COMMAND_LINE - command line options to run optimized binary with36#37# IGNORE_ERROR - ignore error codes returned from optimized binary38#39# GOLD_FILE - file containing expected output from optimized binary40#41# FUNC_NAMES - if set, path to an initial list of function names to42# search. Otherwise, nm is used on the original binary.43#44# OFFLINE - if set, bughunter will produce the binaries but will not45# run them, and will depend on you telling whether it46# succeeded or not.47#48# MAX_FUNCS - if set, use -max-funcs to narrow down the offending49# function. if non-zero, start -max-funcs at $MAX_FUNCS50# otherwise, count the number of symbols in the binary.51#52# MAX_FUNCS_FLAG - BOLT command line option to use for MAX_FUNCS search.53# Default is -max-funcs. Can also be used for relocation54# debugging, e.g. -max-data-relocations.55#56# VERBOSE - if non-empty, set the script to echo mode.57#58##===----------------------------------------------------------------------===##59 60BOLT=${BOLT:=llvm-bolt}61 62ulimit -c 063set -o pipefail64 65if [[ -n "$VERBOSE" ]]; then66 set -x67fi68 69if [[ ! -x $INPUT_BINARY ]] ; then70 echo "INPUT_BINARY must be set to an executable file"71 exit 172fi73 74if [[ -z "$PRE_COMMAND" ]] ; then75 PRE_COMMAND=':'76fi77 78if [[ -z "$POST_COMMAND" ]] ; then79 POST_COMMAND='cat'80fi81 82if [[ -n "$TIMEOUT_OR_CMD" && $TIMEOUT_OR_CMD =~ ^[0-9]+[smhd]?$ ]] ; then83 TIMEOUT_OR_CMD="timeout -s KILL $TIMEOUT_OR_CMD"84fi85 86if [[ -z "$MAX_FUNCS_FLAG" ]] ; then87 MAX_FUNCS_FLAG="-max-funcs"88fi89 90OPTIMIZED_BINARY=$(mktemp -t -u --suffix=.bolt $(basename ${INPUT_BINARY}).XXX)91OUTPUT_FILE="${OPTIMIZED_BINARY}.out"92BOLT_LOG=$(mktemp -t -u --suffix=.log boltXXX)93 94if [[ -z $OFFLINE ]]; then95 echo "Verify input binary passes"96 echo " INPUT_BINARY: $PRE_COMMAND && $TIMEOUT_OR_CMD $INPUT_BINARY $COMMAND_LINE |& $POST_COMMAND >& $OUTPUT_FILE"97 ($PRE_COMMAND && $TIMEOUT_OR_CMD $INPUT_BINARY $COMMAND_LINE |& $POST_COMMAND >& $OUTPUT_FILE)98 STATUS=$?99 if [[ "$IGNORE_ERROR" == "1" ]]; then100 FAIL=0101 else102 FAIL=$STATUS103 fi104 if [[ -e "$GOLD_FILE" ]] ; then105 cmp -s "$OUTPUT_FILE" "$GOLD_FILE"106 FAIL=$?107 fi108 if [[ $FAIL -ne "0" ]] ; then109 echo " Warning: input binary failed"110 else111 echo " Input binary passes."112 fi113fi114 115echo "Verify optimized binary fails"116($BOLT $BOLT_OPTIONS $INPUT_BINARY -o $OPTIMIZED_BINARY >& $BOLT_LOG)117FAIL=$?118if [[ $FAIL -eq "0" ]]; then119 if [[ -z $OFFLINE ]]; then120 echo " OPTIMIZED_BINARY: $PRE_COMMAND && $TIMEOUT_OR_CMD $OPTIMIZED_BINARY $COMMAND_LINE |& $POST_COMMAND >& $OUTPUT_FILE"121 ($PRE_COMMAND && $TIMEOUT_OR_CMD $OPTIMIZED_BINARY $COMMAND_LINE |& $POST_COMMAND >& $OUTPUT_FILE)122 STATUS=$?123 if [[ "$IGNORE_ERROR" == "1" ]]; then124 FAIL=0125 else126 FAIL=$STATUS127 fi128 if [[ -e "$GOLD_FILE" ]] ; then129 cmp -s "$OUTPUT_FILE" "$GOLD_FILE"130 FAIL=$?131 fi132 else133 echo "Did it pass? Type the return code [0 = pass, 1 = fail]"134 read -n1 FAIL135 fi136 if [[ $FAIL -eq "0" ]] ; then137 echo " Warning: optimized binary passes."138 else139 echo " Optimized binary fails as expected."140 fi141else142 echo " Bolt crashes while generating optimized binary."143fi144 145# Collect function names146FF=$(mktemp -t -u --suffix=.txt func-names.XXX)147nm --defined-only -p $INPUT_BINARY | grep " [TtWw] " | cut -d ' ' -f 3 | egrep -v "\._" | egrep -v '^$' | sort -u > $FF148 149# Use function names or numbers150if [[ -z "$MAX_FUNCS" ]] ; then151 # Do binary search on function names152 if [[ -n "$FUNC_NAMES" ]]; then153 FF=$FUNC_NAMES154 fi155 NUM_FUNCS=$(wc -l $FF | cut -d ' ' -f 1)156 HALF=$(expr \( $NUM_FUNCS + 1 \) / 2)157 PREFIX=$(mktemp -t -u --suffix=.txt func-names.XXX)158 FF0=$PREFIX\aa159 FF1=$PREFIX\ab160 split -a 2 -l $HALF $FF $PREFIX161 FF=$FF0162 NUM_FUNCS=$(wc -l $FF | cut -d ' ' -f 1)163 CONTINUE=$(expr $NUM_FUNCS \> 0)164else165 P=0166 if [[ "$MAX_FUNCS" -eq "0" ]]; then167 Q=$(wc -l $FF | cut -d ' ' -f 1)168 else169 Q=$MAX_FUNCS170 fi171 I=$Q172 CONTINUE=$(expr \( $Q - $P \) \> 1)173fi174 175ITER=0176while [[ "$CONTINUE" -ne "0" ]] ; do177 rm -f $OPTIMIZED_BINARY178 if [[ -z "$MAX_FUNCS" ]] ; then179 echo "Iteration $ITER, trying $FF / $HALF functions"180 SEARCH_OPT="-funcs-file-no-regex=$FF"181 else182 I=$(expr \( $Q + $P \) / 2)183 echo "Iteration $ITER, P=$P, Q=$Q, I=$I"184 SEARCH_OPT="$MAX_FUNCS_FLAG=$I"185 fi186 echo " BOLT: $BOLT $BOLT_OPTIONS $INPUT_BINARY $SEARCH_OPT -o $OPTIMIZED_BINARY >& $BOLT_LOG"187 ($BOLT $BOLT_OPTIONS $INPUT_BINARY $SEARCH_OPT -o $OPTIMIZED_BINARY >& $BOLT_LOG)188 FAIL=$?189 echo " BOLT failure=$FAIL"190 rm -f $OUTPUT_FILE191 if [[ $FAIL -eq "0" ]] ; then192 if [[ -z $OFFLINE ]]; then193 echo " OPTIMIZED_BINARY: $PRE_COMMAND && $TIMEOUT_OR_CMD $OPTIMIZED_BINARY $COMMAND_LINE |& $POST_COMMAND >& $OUTPUT_FILE"194 ($PRE_COMMAND && $TIMEOUT_OR_CMD $OPTIMIZED_BINARY $COMMAND_LINE |& $POST_COMMAND >& $OUTPUT_FILE)195 STATUS=$?196 if [[ "$IGNORE_ERROR" == "1" ]]; then197 FAIL=0198 else199 FAIL=$STATUS200 fi201 if [[ -e "$GOLD_FILE" ]] ; then202 cmp -s "$OUTPUT_FILE" "$GOLD_FILE"203 FAIL=$?204 fi205 echo " OPTIMIZED_BINARY failure=$FAIL"206 else207 echo "Did it pass? Type the return code [0 = pass, 1 = fail]"208 read -n1 FAIL209 fi210 else211 FAIL=1212 fi213 214 if [[ -z "$MAX_FUNCS" ]] ; then215 if [[ $FAIL -eq "0" ]] ; then216 if [[ "$FF" == "$FF1" ]]; then217 NUM_FUNCS=0218 break;219 fi220 FF=$FF1221 NUM_FUNCS=$(wc -l $FF | cut -d ' ' -f 1)222 else223 HALF=$(expr \( $NUM_FUNCS + 1 \) / 2)224 PREFIX=$(mktemp -t -u --suffix=.txt func-names.XXX)225 split -a 2 -l $HALF $FF $PREFIX226 FF0=$PREFIX\aa227 FF1=$PREFIX\ab228 FF=$FF0229 NUM_FUNCS=$(wc -l $FF | cut -d ' ' -f 1)230 if [[ $NUM_FUNCS -eq "1" && ! -e $FF1 ]]; then231 break;232 fi233 fi234 CONTINUE=$(expr $NUM_FUNCS \> 0)235 else236 if [[ $FAIL -eq "0" ]] ; then237 P=$I238 else239 Q=$I240 fi241 FF=$I242 HALF=$I243 CONTINUE=$(expr \( $Q - $P \) \> 1)244 fi245 ITER=$(expr $ITER + 1)246done247 248if [[ -z "$MAX_FUNCS" ]] ; then249 if [[ "$NUM_FUNCS" -ne "0" ]] ; then250 FAILED="The function(s) that failed are in $FF"251 fi252else253 if [[ $P -ne $Q ]] ; then254 FF=$(grep "processing ending" $BOLT_LOG | sed -e "s/BOLT-INFO: processing ending on \(.*\)/\1/g" | tail -1)255 FAILED="The item that failed is $FF @ $Q"256 fi257fi258 259if [[ -n "$FAILED" ]] ; then260 echo "$FAILED"261 echo "To reproduce, run: $BOLT $BOLT_OPTIONS $INPUT_BINARY $SEARCH_OPT -o $OPTIMIZED_BINARY"262else263 echo "Unable to reproduce bug."264fi265 266rm $OPTIMIZED_BINARY $OUTPUT_FILE $BOLT_LOG267