From dbcc913071d9ee52f9151d14e1c8b667ca5628b5 Mon Sep 17 00:00:00 2001
From: jdh8d <jdh8d@git.zephyr-software.com>
Date: Thu, 18 Aug 2011 20:00:24 +0000
Subject: [PATCH] Fixes to record the function entry point
---
Makefile | 1 +
libIRDB/include/function.hpp | 9 +-
libIRDB/include/variantir.hpp | 19 ++--
libIRDB/src/function.cpp | 22 ++---
libIRDB/src/variantid.cpp | 4 +-
libIRDB/src/variantir.cpp | 59 +++++++++---
tools/meds2pdb/meds2pdb.cpp | 169 ++++++++++++++++++++++------------
xform/elfreader.cpp | 3 +-
8 files changed, 189 insertions(+), 97 deletions(-)
diff --git a/Makefile b/Makefile
index 3a6fad0d4..4f8beac89 100644
--- a/Makefile
+++ b/Makefile
@@ -3,6 +3,7 @@ all:
cd beaengine; make all
cd libIRDB; make all
cd xform; make all
+ cd libIRDB; make all
cd tools; make all
clean:
diff --git a/libIRDB/include/function.hpp b/libIRDB/include/function.hpp
index a8ec98381..c1dd15b72 100644
--- a/libIRDB/include/function.hpp
+++ b/libIRDB/include/function.hpp
@@ -5,7 +5,8 @@ class Function_t : public BaseObj_t
Function_t() : BaseObj_t(NULL) {} // create a new function not in the db
- Function_t(db_id_t id, std::string name, int size, int oa_size, bool use_fp, File_t *file); // create a function that's already in the DB
+ // create a function that's already in the DB
+ Function_t(db_id_t id, std::string name, int size, int oa_size, bool use_fp, Instruction_t *entry);
std::set<Instruction_t*>& GetInstructions() { return my_insns; }
@@ -17,8 +18,8 @@ class Function_t : public BaseObj_t
void SetName(std::string newname) { name=newname; }
void SetOutArgsRegionSize(int oa_size) {out_args_region_size=oa_size;}
- void SetFile(File_t* file) {my_file=file;}
- File_t* GetFile() { return my_file;}
+ void SetEntryPoint(Instruction_t *insn) {entry_point=insn;}
+ Instruction_t* GetEntryPoint() { return entry_point;}
void WriteToDB(); // we need the variant ID to write into a program.
std::string WriteToDB(VariantID_t *vid, db_id_t newid);
@@ -28,7 +29,7 @@ class Function_t : public BaseObj_t
private:
- File_t *my_file;
+ Instruction_t *entry_point;
std::set<Instruction_t*> my_insns;
int stack_frame_size;
std::string name;
diff --git a/libIRDB/include/variantir.hpp b/libIRDB/include/variantir.hpp
index bf7bf2ce6..cedbb8839 100644
--- a/libIRDB/include/variantir.hpp
+++ b/libIRDB/include/variantir.hpp
@@ -41,13 +41,18 @@ class VariantIR_t : public BaseObj_t
VariantID_t progid;
std::map<db_id_t,File_t*> ReadFilesFromDB();
- std::map<db_id_t,Function_t*> ReadFuncsFromDB ( std::map<db_id_t,File_t*> fileMap) ;
- std::map<db_id_t,AddressID_t*> ReadAddrsFromDB ( std::map<db_id_t,File_t*> fileMap,
- std::map<db_id_t,Function_t*> funcMap) ;
- std::map<db_id_t,Instruction_t*> ReadInsnsFromDB ( std::map<db_id_t,File_t*> fileMap,
- std::map<db_id_t,Function_t*> funcMap,
- std::map<db_id_t,AddressID_t*> addrMap
- ) ;
+ std::map<db_id_t,AddressID_t*> ReadAddrsFromDB(std::map<db_id_t,File_t*> fileMap);
+ std::map<db_id_t,Function_t*> ReadFuncsFromDB
+ (
+ std::map<db_id_t,File_t*> fileMap,
+ std::map<db_id_t,AddressID_t*> addrMap
+ );
+ std::map<db_id_t,Instruction_t*> ReadInsnsFromDB
+ (
+ std::map<db_id_t,File_t*> fileMap,
+ std::map<db_id_t,Function_t*> funcMap,
+ std::map<db_id_t,AddressID_t*> addrMap
+ ) ;
};
diff --git a/libIRDB/src/function.cpp b/libIRDB/src/function.cpp
index 2bc9d8dff..bc27e40f0 100644
--- a/libIRDB/src/function.cpp
+++ b/libIRDB/src/function.cpp
@@ -6,8 +6,8 @@ using namespace std;
-Function_t::Function_t(db_id_t id, std::string myname, int size, int oa_size, bool useFP, File_t* file)
- : BaseObj_t(NULL), my_file(file)
+Function_t::Function_t(db_id_t id, std::string myname, int size, int oa_size, bool useFP, Instruction_t* entry)
+ : BaseObj_t(NULL), entry_point(entry)
{
SetBaseID(id);
name=myname;
@@ -25,21 +25,21 @@ void Function_t::WriteToDB()
string Function_t::WriteToDB(VariantID_t *vid, db_id_t newid)
{
assert(vid);
- assert(my_file);
+ assert(entry_point);
if(GetBaseID()==NOT_IN_DATABASE)
SetBaseID(newid);
string q=string("insert into ")+vid->function_table_name +
- string(" (function_id, file_id, name, stack_frame_size, out_args_region_size, use_frame_pointer, doip_id) ")+
+ string(" (function_id, entry_point_id, name, stack_frame_size, out_args_region_size, use_frame_pointer, doip_id) ")+
string(" VALUES (") +
- string("'") + to_string(GetBaseID()) + string("', ") +
- string("'") + to_string(my_file->GetBaseID()) + string("', ") +
- string("'") + name + string("', ") +
- string("'") + to_string(stack_frame_size) + string("', ") +
- string("'") + to_string(out_args_region_size) + string("', ") +
- string("'") + to_string(use_fp) + string("', ") +
- string("'") + to_string(GetDoipID()) + string("') ; ") ;
+ string("'") + to_string(GetBaseID()) + string("', ") +
+ string("'") + to_string(entry_point->GetBaseID()) + string("', ") +
+ string("'") + name + string("', ") +
+ string("'") + to_string(stack_frame_size) + string("', ") +
+ string("'") + to_string(out_args_region_size) + string("', ") +
+ string("'") + to_string(use_fp) + string("', ") +
+ string("'") + to_string(GetDoipID()) + string("') ; ") ;
return q;
}
diff --git a/libIRDB/src/variantid.cpp b/libIRDB/src/variantid.cpp
index 28c98e2ae..6019ffa10 100644
--- a/libIRDB/src/variantid.cpp
+++ b/libIRDB/src/variantid.cpp
@@ -58,8 +58,8 @@ void VariantID_t::CreateTables()
"address_id integer REFERENCES " + address_table_name + ", " +
"parent_function_id integer, "
"orig_address_id integer, "
- "fallthrough_address_id integer, "
- "target_address_id integer, "
+ "fallthrough_address_id integer DEFAULT -1, "
+ "target_address_id integer DEFAULT -1, "
"data bytea, "
"callback text, "
"comment text, "
diff --git a/libIRDB/src/variantir.cpp b/libIRDB/src/variantir.cpp
index 5ca4f6d8a..1eb8050cb 100644
--- a/libIRDB/src/variantir.cpp
+++ b/libIRDB/src/variantir.cpp
@@ -6,6 +6,26 @@
using namespace libIRDB;
using namespace std;
+static map<Function_t*,db_id_t> entry_points;
+
+
+static void UpdateEntryPoints(std::map<db_id_t,Instruction_t*> &insnMap)
+{
+ /* for each function, look up the instruction that's the entry point */
+ for( static map<Function_t*,db_id_t>::const_iterator it=entry_points.begin();
+ it!=entry_points.end();
+ ++it
+ )
+ {
+ Function_t* func=(*it).first;
+ db_id_t func_entry_id=(*it).second;
+
+ assert(insnMap[func_entry_id]);
+ func->SetEntryPoint(insnMap[func_entry_id]);
+ }
+
+}
+
// Create a Variant from the database
VariantIR_t::VariantIR_t(VariantID_t newprogid) : BaseObj_t(NULL)
{
@@ -19,10 +39,14 @@ VariantIR_t::VariantIR_t(VariantID_t newprogid) : BaseObj_t(NULL)
// DB operations
void VariantIR_t::ReadFromDB()
{
+ entry_points.clear();
+
std::map<db_id_t,File_t*> fileMap=ReadFilesFromDB();
- std::map<db_id_t,Function_t*> funcMap=ReadFuncsFromDB(fileMap);
- std::map<db_id_t,AddressID_t*> addrMap=ReadAddrsFromDB(fileMap,funcMap);
+ std::map<db_id_t,AddressID_t*> addrMap=ReadAddrsFromDB(fileMap);
+ std::map<db_id_t,Function_t*> funcMap=ReadFuncsFromDB(fileMap, addrMap);
std::map<db_id_t,Instruction_t*> insnMap=ReadInsnsFromDB(fileMap,funcMap,addrMap);
+
+ UpdateEntryPoints(insnMap);
}
std::map<db_id_t,File_t*> VariantIR_t::ReadFilesFromDB()
@@ -64,14 +88,14 @@ std::map<db_id_t,File_t*> VariantIR_t::ReadFilesFromDB()
std::map<db_id_t,Function_t*> VariantIR_t::ReadFuncsFromDB
(
- std::map<db_id_t,File_t*> fileMap
+ std::map<db_id_t,File_t*> fileMap,
+ std::map<db_id_t,AddressID_t*> addrMap
)
{
std::map<db_id_t,Function_t*> idMap;
std::string q= "select * from " + progid.function_table_name + " ; ";
-
dbintr->IssueQuery(q);
while(!dbintr->IsDone())
@@ -79,7 +103,7 @@ std::map<db_id_t,Function_t*> VariantIR_t::ReadFuncsFromDB
// function_id | file_id | name | stack_frame_size | out_args_region_size | use_frame_pointer | doip_id
db_id_t fid=atoi(dbintr->GetResultColumn("function_id").c_str());
- db_id_t file_id=atoi(dbintr->GetResultColumn("file_id").c_str());
+ db_id_t entry_point_id=atoi(dbintr->GetResultColumn("entry_point_id").c_str());
std::string name=dbintr->GetResultColumn("name");
int sfsize=atoi(dbintr->GetResultColumn("stack_frame_size").c_str());
int oasize=atoi(dbintr->GetResultColumn("out_args_region_size").c_str());
@@ -94,8 +118,9 @@ std::map<db_id_t,Function_t*> VariantIR_t::ReadFuncsFromDB
db_id_t doipid=atoi(dbintr->GetResultColumn("doip_id").c_str());
- Function_t *newfunc=new Function_t(fid,name,sfsize,oasize,useFP,fileMap[file_id]);
-
+ Function_t *newfunc=new Function_t(fid,name,sfsize,oasize,useFP,NULL);
+ entry_points[newfunc]=entry_point_id;
+
//std::cout<<"Found function "<<name<<"."<<std::endl;
idMap[fid]=newfunc;
@@ -108,8 +133,11 @@ std::map<db_id_t,Function_t*> VariantIR_t::ReadFuncsFromDB
return idMap;
}
-std::map<db_id_t,AddressID_t*> VariantIR_t::ReadAddrsFromDB ( std::map<db_id_t,File_t*> fileMap,
- std::map<db_id_t,Function_t*> funcMap)
+
+std::map<db_id_t,AddressID_t*> VariantIR_t::ReadAddrsFromDB
+ (
+ std::map<db_id_t,File_t*> fileMap
+ )
{
std::map<db_id_t,AddressID_t*> idMap;
@@ -146,10 +174,12 @@ std::map<db_id_t,AddressID_t*> VariantIR_t::ReadAddrsFromDB ( std::map<
}
-std::map<db_id_t,Instruction_t*> VariantIR_t::ReadInsnsFromDB ( std::map<db_id_t,File_t*> fileMap,
- std::map<db_id_t,Function_t*> funcMap,
- std::map<db_id_t,AddressID_t*> addrMap
- )
+std::map<db_id_t,Instruction_t*> VariantIR_t::ReadInsnsFromDB
+ (
+ std::map<db_id_t,File_t*> fileMap,
+ std::map<db_id_t,Function_t*> funcMap,
+ std::map<db_id_t,AddressID_t*> addrMap
+ )
{
std::map<db_id_t,Instruction_t*> idMap;
std::map<db_id_t,db_id_t> fallthroughs;
@@ -273,6 +303,9 @@ void VariantIR_t::SetBaseIDS()
for(std::set<File_t*>::const_iterator i=files.begin(); i!=files.end(); ++i)
j=MAX(j,(*i)->GetBaseID());
+ /* increment past the max ID so we don't duplicate */
+ j++;
+
/* for anything that's not yet in the DB, assign an ID to it */
for(std::set<Function_t*>::const_iterator i=funcs.begin(); i!=funcs.end(); ++i)
if((*i)->GetBaseID()==NOT_IN_DATABASE)
diff --git a/tools/meds2pdb/meds2pdb.cpp b/tools/meds2pdb/meds2pdb.cpp
index 93e5b5e46..b3f25136c 100644
--- a/tools/meds2pdb/meds2pdb.cpp
+++ b/tools/meds2pdb/meds2pdb.cpp
@@ -8,6 +8,22 @@
using namespace std;
using namespace pqxx;
+#include <sstream>
+
+template <class T>
+inline std::string my_to_string (const T& t)
+{
+ std::stringstream ss;
+ ss << t;
+ return ss.str();
+}
+
+
+int next_address_id=0;
+
+map<app_iaddr_t,int> address_to_instructionid_map;
+
+
// extract the file id from the md5 hash and the program name
int get_file_id(char *progName, char *md5hash)
{
@@ -29,20 +45,20 @@ int get_file_id(char *progName, char *md5hash)
return -1; // error
}
+
// insert addresses & instructions into DB
void insert_instructions(string programName, int fileID, vector<wahoo::Instruction*> instructions, vector<wahoo::Function*> functions)
{
- cerr << "Inserting instructions in the DB";
+ cerr << "Inserting instructions in the DB"<<endl;
connection conn;
work txn(conn);
// for each instruction:
// (1) get address, insert into address table
// (2) populate instruction table
- const int STRIDE = 1000;
int count = 0;
- for (int i = 0; i < instructions.size(); i += STRIDE)
+ for (int i = 0; i < instructions.size(); i ++ )
{
char buf[128];
string addressTable = programName + "_" + "address";
@@ -51,29 +67,24 @@ void insert_instructions(string programName, int fileID, vector<wahoo::Instructi
string instructionTable = programName + "_" + "instruction";
string query2 = "INSERT INTO " + instructionTable;
- query2 += " (address_id, parent_function_id, orig_address_id, data, comment) VALUES ";
+ query2 += " (instruction_id,address_id, parent_function_id, orig_address_id, data, comment) VALUES ";
- for (int j = i; j < i + STRIDE; ++j)
- {
- if (j >= instructions.size()) break;
- count++;
-
- wahoo::Instruction *instruction = instructions[j];
+ wahoo::Instruction *instruction = instructions[i];
app_iaddr_t addr = instruction->getAddress();
+ address_to_instructionid_map[addr]=i;
+
+ int address_id = next_address_id++;
+
// insert into address table
- if (j != i) query += ",";
query += "(";
- query += txn.quote(j) + ",";
+ query += txn.quote(address_id) + ",";
query += txn.quote(fileID) + ",";
sprintf(buf,"%d", addr);
query += txn.quote(string(buf));
query += ")";
- // insert into instruction table
- if (j != i) query2 += ",";
- int address_id = j;
int parent_function_id = -1;
if (instruction->getFunction())
{
@@ -83,7 +94,8 @@ void insert_instructions(string programName, int fileID, vector<wahoo::Instructi
string asmData = instruction->getAsm();
query2 += "(";
- query2 += txn.quote(address_id) + ","; // j is the address id
+ query2 += txn.quote(my_to_string(i)) + ",";
+ query2 += txn.quote(address_id) + ","; // i is the address id
query2 += txn.quote(parent_function_id) + ",";
query2 += txn.quote(orig_address_id) + ",";
@@ -109,7 +121,6 @@ void insert_instructions(string programName, int fileID, vector<wahoo::Instructi
query2 += string(buf) + ",";
query2 += txn.quote(asmData);
query2 += ")";
- }
// cerr << "Query: " << query << endl;
// cerr << "Query2: " << query2 << endl;
@@ -118,9 +129,87 @@ void insert_instructions(string programName, int fileID, vector<wahoo::Instructi
txn.exec(query2);
}
- cerr << "Committing all instructions - this may take a while";
+ cerr << "Committing all instructions - this may take a while"<<endl;
txn.commit();
- cerr << "Done inserting instructions in the DB";
+ cerr << "Done inserting instructions in the DB"<<endl;
+}
+
+
+void insert_functions( string programName, int fileID, const vector<wahoo::Function*> &functions )
+{
+ connection conn;
+ work txn(conn);
+
+ // bulk insert of function information into the DB
+ const int STRIDE = 25;
+ int count = 0;
+ for (int i = 0; i < functions.size(); i += STRIDE)
+ {
+ string functionTable = string(programName) + "_" + "function";
+ string query = "INSERT INTO " + functionTable;
+ query += " (function_id, name, stack_frame_size, out_args_region_size, use_frame_pointer) VALUES ";
+
+
+ for (int j = i; j < i + STRIDE; ++j)
+ {
+ if (j >= functions.size()) break;
+ wahoo::Function *f = functions[j];
+ string functionName = f->getName();
+ app_iaddr_t functionAddress = f->getAddress();
+ int functionSize = f->getSize();
+
+ int function_id = j;
+ f->setFunctionID(function_id);
+
+ int outArgsRegionSize = f->getOutArgsRegionSize();
+ bool useFP = f->getUseFramePointer();
+
+ if (j != i) query += ",";
+ query += "(";
+ query += txn.quote(function_id) + ",";
+ query += txn.quote(functionName) + ",";
+ query += txn.quote(functionSize) + ",";
+ query += txn.quote(outArgsRegionSize) + ",";
+ query += txn.quote(useFP) + ")";
+
+ }
+
+ txn.exec(query);
+ }
+
+ txn.commit(); // must commit o/w everything will be rolled back
+}
+
+void update_functions( string programName, int fileID, const vector<wahoo::Function*> &functions )
+{
+ connection conn;
+ work txn(conn);
+
+ // bulk insert of function information into the DB
+ int count = 0;
+ string query;
+ for (int i = 0; i < functions.size(); i++ )
+ {
+ wahoo::Function *f = functions[i];
+ string functionName = f->getName();
+ app_iaddr_t functionAddress = f->getAddress();
+ int functionSize = f->getSize();
+ int function_id = f->getFunctionID();
+ int outArgsRegionSize = f->getOutArgsRegionSize();
+ bool useFP = f->getUseFramePointer();
+ int insnid=address_to_instructionid_map[functionAddress];
+
+ string functionTable = string(programName) + "_" + "function";
+ query += "update " + functionTable;
+ query += " set entry_point_id = " + txn.quote(my_to_string(insnid));
+ query += " where function_id = " + txn.quote(my_to_string(function_id));
+ query += ";";
+
+
+ }
+
+ txn.exec(query);
+ txn.commit(); // must commit o/w everything will be rolled back
}
int main(int argc, char **argv)
@@ -141,8 +230,6 @@ int main(int argc, char **argv)
cerr << "hash-md5:" << md5hash << endl;
cerr << "annotation file:" << annotFile << endl;
- connection conn;
- work txn(conn);
Rewriter *rewriter = new Rewriter(elfFile, annotFile);
@@ -162,44 +249,8 @@ int main(int argc, char **argv)
cerr << "Number of functions: " << functions.size() << endl;
cerr << "Number of instructions: " << instructions.size() << endl;
- // bulk insert of function information into the DB
- const int STRIDE = 25;
- int count = 0;
- for (int i = 0; i < functions.size(); i += STRIDE)
- {
- string functionTable = string(programName) + "_" + "function";
- string query = "INSERT INTO " + functionTable;
- query += " (function_id, file_id, name, stack_frame_size, out_args_region_size, use_frame_pointer) VALUES ";//*
-
- for (int j = i; j < i + STRIDE; ++j)
- {
- if (j >= functions.size()) break;
-
- wahoo::Function *f = functions[j];
- string functionName = f->getName();
- app_iaddr_t functionAddress = f->getAddress();
- int functionSize = f->getSize();
- int function_id = j;
- int outArgsRegionSize = f->getOutArgsRegionSize();//*
- bool useFP = f->getUseFramePointer();//*
-
- if (j != i) query += ",";
- query += "(";
- query += txn.quote(function_id) + ",";
- query += txn.quote(fileID) + ",";
- query += txn.quote(functionName) + ",";
- query += txn.quote(functionSize) + ",";
- query += txn.quote(outArgsRegionSize) + ",";
- query += txn.quote(useFP) + ")";
-
- f->setFunctionID(function_id);
- }
-
- txn.exec(query);
- }
-
- txn.commit(); // must commit o/w everything will be rolled back
+ insert_functions(programName, fileID, functions);
insert_instructions(programName, fileID, instructions, functions);
+ update_functions(programName, fileID, functions);
}
-
diff --git a/xform/elfreader.cpp b/xform/elfreader.cpp
index 7a354fef0..a9f15b624 100644
--- a/xform/elfreader.cpp
+++ b/xform/elfreader.cpp
@@ -21,10 +21,11 @@ ElfReader::ElfReader(char *p_elfFile)
{ // For all sections
const IELFISection* pSec = m_reader->GetSection( i );
m_sections.push_back(pSec);
-
+#if 0
std::cout << "Sec. name: " << pSec->GetName()
<< " Sec. offset: " << pSec->GetOffset()
<< " Sec. size: " << pSec->GetSize() << std::endl;
+#endif
}
std::cout << std::endl;
--
GitLab