d94b005be2
This change alters DwarfReader and import-trace's ElfImporter so that they use unsigned int for static address and line numbers instead of signed int. Change-Id: I84ebbb500afd7cd4d93b137a35dcf736dc679fab
441 lines
12 KiB
C++
441 lines
12 KiB
C++
#include <sstream>
|
|
#include <iostream>
|
|
#include <fstream>
|
|
#include "util/Logger.hpp"
|
|
#include "ElfImporter.hpp"
|
|
#include "util/pstream.h"
|
|
#ifndef __puma
|
|
#include <boost/regex.hpp>
|
|
#include <boost/algorithm/string.hpp>
|
|
#endif
|
|
|
|
|
|
using namespace llvm;
|
|
using namespace llvm::object;
|
|
using namespace fail;
|
|
using namespace std;
|
|
|
|
static Logger LOG("ElfImporter");
|
|
|
|
/**
|
|
* Callback function that can be used to add command line options
|
|
* to the campaign
|
|
*/
|
|
bool ElfImporter::cb_commandline_init()
|
|
{
|
|
CommandLine &cmd = CommandLine::Inst();
|
|
|
|
OBJDUMP = cmd.addOption("", "objdump", Arg::Required,
|
|
"--objdump \tObjdump: location of objdump binary, otherwise LLVM Disassembler is used");
|
|
SOURCECODE = cmd.addOption("", "sources", Arg::None,
|
|
"--sources \timport all source files and the mapping of code line<->static instruction into the database");
|
|
return true;
|
|
}
|
|
|
|
bool ElfImporter::create_database()
|
|
{
|
|
CommandLine &cmd = CommandLine::Inst();
|
|
std::stringstream create_statement;
|
|
|
|
create_statement.str("");
|
|
create_statement << "CREATE TABLE IF NOT EXISTS objdump ("
|
|
" variant_id int(11) NOT NULL,"
|
|
" instr_address int(11) UNSIGNED NOT NULL,"
|
|
" opcode varchar(32) NOT NULL,"
|
|
" disassemble VARCHAR(64),"
|
|
" comment VARCHAR(128),"
|
|
" PRIMARY KEY (variant_id,instr_address)"
|
|
") engine=MyISAM ";
|
|
if (!db->query(create_statement.str().c_str())) {
|
|
return false;
|
|
}
|
|
|
|
if (cmd[SOURCECODE]) {
|
|
create_statement.str("");
|
|
create_statement << "CREATE TABLE IF NOT EXISTS dbg_filename ("
|
|
" file_id int(11) NOT NULL AUTO_INCREMENT,"
|
|
" variant_id int(11) NOT NULL,"
|
|
" path VARCHAR(1024),"
|
|
" PRIMARY KEY (file_id)"
|
|
") engine=MyISAM ";
|
|
if (!db->query(create_statement.str().c_str())) {
|
|
return false;
|
|
}
|
|
|
|
create_statement.str("");
|
|
create_statement << "CREATE TABLE IF NOT EXISTS dbg_source ("
|
|
" variant_id int(11) NOT NULL,"
|
|
" linenumber int(11) NOT NULL,"
|
|
" file_id int(11) NOT NULL,"
|
|
" line VARCHAR(1024),"
|
|
" PRIMARY KEY (variant_id, linenumber, file_id)"
|
|
") engine=MyISAM ";
|
|
if (!db->query(create_statement.str().c_str())) {
|
|
return false;
|
|
}
|
|
|
|
create_statement.str("");
|
|
create_statement << "CREATE TABLE IF NOT EXISTS dbg_mapping ("
|
|
" variant_id int(11) NOT NULL,"
|
|
" instr_absolute int(11) UNSIGNED NOT NULL,"
|
|
" linenumber int(11) UNSIGNED NOT NULL,"
|
|
" file_id int(11) NOT NULL,"
|
|
" PRIMARY KEY (variant_id, instr_absolute ,linenumber)"
|
|
") engine=MyISAM ";
|
|
if (!db->query(create_statement.str().c_str())) {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool ElfImporter::copy_to_database(ProtoIStream &ps)
|
|
{
|
|
if (!m_elf) {
|
|
LOG << "please give an elf binary as parameter (-e/--elf)" << std::endl;
|
|
return false;
|
|
}
|
|
|
|
CommandLine &cmd = CommandLine::Inst();
|
|
if (!cmd.parse()) {
|
|
std::cerr << "Error parsing arguments." << std::endl;
|
|
return false;
|
|
}
|
|
|
|
if (cmd[OBJDUMP]) { // using objdump
|
|
if (!import_with_objdump(std::string(cmd[OBJDUMP].first()->arg))) {
|
|
return false;
|
|
}
|
|
} else {
|
|
LOG << "importing an objdump with internal llvm dissassembler is not yet implemented" << std::endl;
|
|
}
|
|
|
|
if (cmd[SOURCECODE]) { // import sources
|
|
std::list<std::string> sourceFiles;
|
|
|
|
if (!import_source_files(m_elf->getFilename(), sourceFiles)) {
|
|
cerr << "unable to read dwarf2 debug info" << endl;
|
|
return false;
|
|
}
|
|
|
|
for (std::list<std::string>::iterator it = sourceFiles.begin(); it != sourceFiles.end(); ++it) {
|
|
if (!import_source_code(*it)) {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
// import debug information
|
|
if (!import_mapping(m_elf->getFilename())) {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool ElfImporter::import_with_objdump(const std::string &binary)
|
|
{
|
|
#ifndef __puma
|
|
|
|
LOG << "importing with " << binary << std::endl;
|
|
|
|
// -d: disassemble
|
|
// -C: demangle C++
|
|
std::string command = binary + " -C -d " + m_elf->getFilename();
|
|
LOG << "Executing: " << command << std::endl;
|
|
redi::ipstream objdump( command );
|
|
std::string str;
|
|
while (std::getline(objdump, str)) {
|
|
if (!evaluate_objdump_line(str)) {
|
|
objdump.close();
|
|
return false;
|
|
}
|
|
}
|
|
objdump.close();
|
|
if (objdump.rdbuf()->exited()) {
|
|
int ex = objdump.rdbuf()->status();
|
|
if (ex != 0) {
|
|
clear_database();
|
|
LOG << "Could not disassemble!" << std::endl;
|
|
return false;
|
|
}
|
|
}
|
|
|
|
#endif
|
|
return true;
|
|
}
|
|
|
|
bool ElfImporter::evaluate_objdump_line(const std::string& line)
|
|
{
|
|
#ifndef __puma
|
|
// Only read in real code lines:
|
|
// Code lines start with a leading whitespace! (hopefully in each objdump implementation!)
|
|
if (line.size() > 0 && isspace(line[0])) {
|
|
// a line looks like: 800156c:\tdd14 \tble.n 8001598 <_ZN2hw3hal7T32Term8PutBlockEPci+0x30>
|
|
static boost::regex expr("\\s+([A-Fa-f0-9]+):((?:\\s+[A-Fa-f0-9]{2})+)\\s+(.+?)(;.*)?$");
|
|
boost::smatch res;
|
|
if (boost::regex_search(line, res, expr)) {
|
|
std::string address = res[1];
|
|
std::stringstream ss;
|
|
ss << std::hex << address;
|
|
address_t addr = 0;
|
|
ss >> addr;
|
|
std::string opcode = res[2];
|
|
boost::trim(opcode);
|
|
std::string instruction = res[3];
|
|
boost::trim(instruction);
|
|
std::string comment = res[4];
|
|
boost::trim(comment);
|
|
|
|
// transform hex opcode to char array
|
|
char opcode_read[33];
|
|
std::stringstream opcode_stream(opcode);
|
|
unsigned long opcode_length = 0;
|
|
char c;
|
|
while (opcode_length < 16 && opcode_stream) {
|
|
opcode_stream << std::hex;
|
|
opcode_stream >> c;
|
|
if (!opcode_stream) break;
|
|
opcode_read[opcode_length++] = c;
|
|
}
|
|
opcode_read[opcode_length] = 0;
|
|
|
|
/* import instruction into database */
|
|
if (!import_instruction(addr, opcode_read, opcode_length,
|
|
instruction, comment))
|
|
return false;
|
|
}
|
|
}
|
|
#endif
|
|
return true;
|
|
}
|
|
|
|
|
|
|
|
bool ElfImporter::import_instruction(fail::address_t addr, char *opcode, int opcode_length,
|
|
const std::string &instruction, const std::string &comment)
|
|
{
|
|
/* Prepare a mysql statement if it was not done before */
|
|
static MYSQL_STMT *stmt = 0;
|
|
if (!stmt) {
|
|
std::string sql("INSERT INTO objdump (variant_id, instr_address, opcode, disassemble, comment) VALUES (?,?,?,?,?)");
|
|
stmt = mysql_stmt_init(db->getHandle());
|
|
if (mysql_stmt_prepare(stmt, sql.c_str(), sql.length())) {
|
|
LOG << "query '" << sql << "' failed: " << mysql_error(db->getHandle()) << std::endl;
|
|
return false;
|
|
}
|
|
}
|
|
MYSQL_BIND bind[5];
|
|
memset(bind, 0, sizeof(bind));
|
|
for (unsigned i = 0; i < 5; ++i) {
|
|
bind[i].buffer_type = MYSQL_TYPE_STRING;
|
|
bind[i].is_unsigned = 1;
|
|
switch (i) {
|
|
case 0:
|
|
bind[i].buffer = &m_variant_id;
|
|
bind[i].buffer_type = MYSQL_TYPE_LONG;
|
|
break;
|
|
case 1:
|
|
bind[i].buffer = &addr;
|
|
bind[i].buffer_type = MYSQL_TYPE_LONG;
|
|
break;
|
|
case 2:
|
|
bind[i].buffer_type = MYSQL_TYPE_BLOB;
|
|
bind[i].buffer = opcode;
|
|
bind[i].buffer_length = opcode_length;
|
|
break;
|
|
case 3:
|
|
bind[i].buffer = const_cast<char *>(instruction.c_str());
|
|
bind[i].buffer_length = instruction.length();
|
|
break;
|
|
case 4:
|
|
bind[i].buffer = const_cast<char *>(comment.c_str());
|
|
bind[i].buffer_length = comment.length();
|
|
break;
|
|
}
|
|
}
|
|
|
|
|
|
if (mysql_stmt_bind_param(stmt, bind)) {
|
|
LOG << "mysql_stmt_bind_param() failed: " << mysql_stmt_error(stmt) << std::endl;
|
|
return false;
|
|
}
|
|
if (mysql_stmt_execute(stmt)) {
|
|
LOG << "mysql_stmt_execute() failed: " << mysql_stmt_error(stmt) << std::endl;
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
static inline std::string rtrim(std::string str)
|
|
{
|
|
std::string whitespaces(" \t\f\v\n\r");
|
|
|
|
std::size_t found = str.find_last_not_of(whitespaces);
|
|
if (found != std::string::npos) {
|
|
str.erase(found+1);
|
|
} else {
|
|
str.clear();
|
|
}
|
|
return str;
|
|
}
|
|
|
|
bool ElfImporter::import_source_code(std::string fileName)
|
|
{
|
|
LOG << "Importing Sourcefile: " << fileName << endl;
|
|
|
|
ifstream in(fileName.c_str());
|
|
if (!in.is_open()) {
|
|
LOG << "Sourcefile not found: " << fileName << endl;
|
|
return true; // we can live with this
|
|
}
|
|
|
|
// retrieve file_id
|
|
std::stringstream ss;
|
|
ss << "SELECT file_id FROM dbg_filename "
|
|
<< "WHERE path = '" << fileName.c_str() << "' "
|
|
<< "AND variant_id = " << m_variant_id;
|
|
MYSQL_RES *res = db->query(ss.str().c_str(), true);
|
|
if (!res) {
|
|
return false;
|
|
}
|
|
MYSQL_ROW row;
|
|
if (!(row = mysql_fetch_row(res))) {
|
|
// this should not happen
|
|
LOG << "error: no entry for '" << fileName.c_str() << "' in dbg_filename, aborting" << std::endl;
|
|
return false;
|
|
}
|
|
std::string file_id = row[0];
|
|
|
|
// import lines from this file
|
|
for (unsigned lineNo = 1; !in.eof(); ++lineNo) {
|
|
// read and strip a line
|
|
string currentLine;
|
|
getline(in, currentLine);
|
|
if (!currentLine.length() && in.eof()) {
|
|
break;
|
|
}
|
|
currentLine = rtrim(currentLine);
|
|
|
|
// escape special characters for use in SQL
|
|
currentLine = db->escape_string(currentLine);
|
|
|
|
// insert line into dbg_source
|
|
std::stringstream sql;
|
|
sql << "(" << m_variant_id << "," << lineNo
|
|
<< "," << file_id << ",'" << currentLine << "')";
|
|
if (!db->insert_multiple("INSERT INTO dbg_source (variant_id, linenumber, file_id, line) VALUES ",
|
|
sql.str().c_str())) {
|
|
return false;
|
|
}
|
|
}
|
|
db->insert_multiple();
|
|
return true;
|
|
}
|
|
|
|
bool ElfImporter::import_source_files(const std::string& elf_filename, std::list<std::string>& filenames)
|
|
{
|
|
if (!dwReader.read_source_files(elf_filename, filenames)) {
|
|
return false;
|
|
}
|
|
for (std::list<std::string>::iterator it = filenames.begin(); it != filenames.end(); ++it) {
|
|
// INSERT group entry
|
|
std::stringstream sql;
|
|
sql << "(" << m_variant_id << ",'" << it->c_str() << "')";
|
|
|
|
if (!db->insert_multiple("INSERT INTO dbg_filename (variant_id, path) VALUES ",
|
|
sql.str().c_str())) {
|
|
return false;
|
|
}
|
|
}
|
|
db->insert_multiple();
|
|
return true;
|
|
}
|
|
|
|
bool ElfImporter::import_mapping(std::string fileName)
|
|
{
|
|
std::list<addrToLine> mapping;
|
|
if (!dwReader.read_mapping(fileName, mapping)) {
|
|
return false;
|
|
}
|
|
|
|
while (!mapping.empty()) {
|
|
struct addrToLine temp_addrToLine;
|
|
temp_addrToLine = mapping.front();
|
|
|
|
// FIXME reuse file_id from previous iteration if file name stays constant
|
|
std::stringstream ss;
|
|
ss << "SELECT file_id FROM dbg_filename "
|
|
<< "WHERE path = '" << temp_addrToLine.lineSource << "' "
|
|
<< "AND variant_id = " << m_variant_id;
|
|
|
|
MYSQL_RES *res = db->query(ss.str().c_str(), true);
|
|
if (!res) {
|
|
return false;
|
|
}
|
|
MYSQL_ROW row;
|
|
row = mysql_fetch_row(res);
|
|
|
|
// INSERT group entry
|
|
std::stringstream sql;
|
|
sql << "(" << m_variant_id << "," << temp_addrToLine.absoluteAddr << "," << temp_addrToLine.lineNumber << ",";
|
|
if (row != NULL) {
|
|
sql << row[0] << ")";
|
|
} else {
|
|
// this should not happen
|
|
LOG << "error: no entry for '" << temp_addrToLine.lineSource << "' in dbg_filename, aborting" << std::endl;
|
|
return false;
|
|
}
|
|
|
|
// TODO: Skip duplicated entries with ON DUPLICATE KEY UPDATE (?)
|
|
if (!db->insert_multiple("INSERT IGNORE INTO dbg_mapping (variant_id, instr_absolute, linenumber, file_id) VALUES ",
|
|
sql.str().c_str())) {
|
|
return false;
|
|
}
|
|
|
|
mapping.pop_front();
|
|
}
|
|
|
|
db->insert_multiple();
|
|
|
|
return true;
|
|
}
|
|
|
|
bool ElfImporter::clear_database()
|
|
{
|
|
CommandLine &cmd = CommandLine::Inst();
|
|
std::stringstream ss;
|
|
bool ret = true, result;
|
|
|
|
ss.str("");
|
|
ss << "DELETE FROM objdump WHERE variant_id = " << m_variant_id;
|
|
result = db->query(ss.str().c_str());
|
|
ret = ret && result;
|
|
LOG << "deleted " << db->affected_rows() << " rows from objdump table" << std::endl;
|
|
|
|
if (cmd[SOURCECODE]) {
|
|
ss.str("");
|
|
ss << "DELETE FROM dbg_source WHERE variant_id = " << m_variant_id;
|
|
result = db->query(ss.str().c_str());
|
|
ret = ret && result;
|
|
LOG << "deleted " << db->affected_rows() << " rows from dbg_source table" << std::endl;
|
|
|
|
ss.str("");
|
|
ss << "DELETE FROM dbg_filename WHERE variant_id = " << m_variant_id;
|
|
result = db->query(ss.str().c_str());
|
|
ret = ret && result;
|
|
LOG << "deleted " << db->affected_rows() << " rows from dbg_filename table" << std::endl;
|
|
|
|
//ToDo: Reset auto increment value to 1
|
|
ss.str("");
|
|
ss << "DELETE FROM dbg_mapping WHERE variant_id = " << m_variant_id;
|
|
result = db->query(ss.str().c_str());
|
|
ret = ret && result;
|
|
LOG << "deleted " << db->affected_rows() << " rows from source table" << std::endl;
|
|
}
|
|
|
|
return ret;
|
|
}
|