diff --git a/libIRDB/test/fill_in_indtargs.cpp b/libIRDB/test/fill_in_indtargs.cpp
index 28ff48a8a4816f184a41a785f12298d5b6fcb792..648f8cf31b6c2a4cba3033888ff4f48b616061c0 100644
--- a/libIRDB/test/fill_in_indtargs.cpp
+++ b/libIRDB/test/fill_in_indtargs.cpp
@@ -43,8 +43,6 @@ using namespace libIRDB;
using namespace std;
using namespace EXEIO;
-#define HELLNODE_ID 0
-#define INDIRECT_CALLS_ID 1
int next_icfs_set_id = 2;
ICFS_t* hellnode_tgts = NULL;
@@ -248,6 +246,11 @@ void mark_jmptables(FileIR_t *firp)
Instruction_t* instr = it->first;
set<Instruction_t*> instruction_targets = it->second;
+ // ignore if instr already marked complete.
+ // FIXME: assert that fill_in_indtarg analysis matches already complete analysis.
+ if(instr->GetIBTargets() && instr->GetIBTargets()->IsComplete())
+ continue;
+
assert(instruction_targets.size() > 0);
ICFS_t* new_icfs = new ICFS_t(next_icfs_set_id++, true);
@@ -1421,11 +1424,25 @@ void check_for_nonPIC_switch_table(FileIR_t* firp, Instruction_t* insn, DISASM d
jmptables[IJ] = ibtargets;
}
+template <class T> T MAX(T a, T b)
+{
+ return a>b ? a : b;
+}
+
void icfs_init(FileIR_t* firp)
{
+
+
assert(firp);
- hellnode_tgts = new ICFS_t(HELLNODE_ID, false);
- indirect_calls = new ICFS_t(INDIRECT_CALLS_ID, false);
+ db_id_t max_id=0;
+ for(ICFSSet_t::iterator it=firp->GetAllICFS().begin(); it!=firp->GetAllICFS().end(); ++it)
+ {
+ max_id=MAX<db_id_t>(max_id, (*it)->GetBaseID());
+ }
+ next_icfs_set_id = max_id+1;
+ cerr<<"Found max ICFS id=="<<max_id<<endl;
+ hellnode_tgts = new ICFS_t(next_icfs_set_id++, false);
+ indirect_calls = new ICFS_t(next_icfs_set_id++, false);
firp->GetAllICFS().insert(hellnode_tgts);
firp->GetAllICFS().insert(indirect_calls);
}
@@ -1433,13 +1450,13 @@ void icfs_init(FileIR_t* firp)
void icfs_set_indirect_calls(FileIR_t* const firp, ICFS_t* const targets)
{
assert(firp && targets);
- for(
- FunctionSet_t::const_iterator it=firp->GetFunctions().begin();
- it!=firp->GetFunctions().end();
- ++it
- )
- {
- Function_t *func=*it;
+ for(
+ FunctionSet_t::const_iterator it=firp->GetFunctions().begin();
+ it!=firp->GetFunctions().end();
+ ++it
+ )
+ {
+ Function_t *func=*it;
if(!func->GetEntryPoint())
continue;
targets->insert(func->GetEntryPoint());
@@ -1472,6 +1489,10 @@ void check_for_ret(FileIR_t* const firp, Instruction_t* const insn)
if(strstr(d.Instruction.Mnemonic, "ret")==NULL)
return;
+ // already analysed by ida.
+ if(insn->GetIBTargets() && insn->GetIBTargets()->IsComplete())
+ return;
+
insn->SetIBTargets(hellnode_tgts);
}
@@ -1487,11 +1508,17 @@ void check_for_indirect_jmp(FileIR_t* const firp, Instruction_t* const insn)
if(d.Argument1.ArgType®ISTER_TYPE)
{
+ // already analysed by ida.
+ if(insn->GetIBTargets() && insn->GetIBTargets()->IsComplete())
+ return;
insn->SetIBTargets(hellnode_tgts);
}
else if(d.Argument1.ArgType&MEMORY_TYPE &&
(!d.Argument1.ArgType&RELATIVE_))
{
+ // already analysed by ida.
+ if(insn->GetIBTargets() && insn->GetIBTargets()->IsComplete())
+ return;
insn->SetIBTargets(hellnode_tgts);
}
}
@@ -1509,6 +1536,10 @@ void check_for_indirect_call(FileIR_t* const firp, Instruction_t* const insn)
if(d.Argument1.ArgType&CONSTANT_TYPE)
return;
+ // already analysed by ida.
+ if(insn->GetIBTargets() && insn->GetIBTargets()->IsComplete())
+ return;
+
insn->SetIBTargets(indirect_calls);
}
diff --git a/tools/meds2pdb/meds2pdb.cpp b/tools/meds2pdb/meds2pdb.cpp
index bed63cfbd406a9ce00546fb84faffcdf9feaee90..e12b6098972ba0fbede16079e497b16042a11481 100644
--- a/tools/meds2pdb/meds2pdb.cpp
+++ b/tools/meds2pdb/meds2pdb.cpp
@@ -41,6 +41,8 @@ string functionTable;
string addressTable;
string instructionTable;
string typesTable;
+string icfsTable;
+string icfsMapTable;
static const int STRIDE = 50;
@@ -56,6 +58,7 @@ inline std::string my_to_string (const T& t)
int next_address_id=0;
map<app_iaddr_t,int> address_to_instructionid_map;
+map<wahoo::Instruction*,int> instruction_to_addressid_map;
// extract the file id from the md5 hash and the program name
int get_file_id(char *progName, char *md5hash)
@@ -81,8 +84,63 @@ int get_file_id(char *progName, char *md5hash)
}
+void insert_icfs(int fileID, const vector<wahoo::Instruction*>& instructions)
+{
+ using namespace wahoo;
+
+ connection conn;
+ work txn(conn);
+ txn.exec("SET client_encoding='LATIN1';");
+
+ int next_icfs_id=0;
+
+ for (int i = 0; i < instructions.size(); i ++)
+ {
+ wahoo::Instruction *instruction = instructions[i];
+ assert(instruction);
+
+ const std::set<Instruction*> &ibts=instruction->getIBTs();
+
+ if(ibts.size()==0)
+ continue;
+ cerr<<"Found fromIB=="<<instruction->getAddress()<<endl;
+
+ string query = "INSERT INTO " + icfsTable;
+ query += " (icfs_id,is_complete) VALUES ";
+ query += "(";
+ query += txn.quote(next_icfs_id) + ",";
+ query += txn.quote(1);
+ query += ")";
+
+ string query2 = "INSERT INTO " + icfsMapTable;
+ query2 += " (icfs_id,address_id) VALUES ";
+
+ for(set<Instruction*>::iterator it=ibts.begin(); it!=ibts.end(); it++)
+ {
+ cerr<<"Found toIBT=="<<(*it)->getAddress()<<endl;
+ if(it!=ibts.begin())
+ query2+=",";
+ int target_address_id=instruction_to_addressid_map[*it];
+ query2 += "(";
+ query2 += txn.quote(next_icfs_id) + ",";
+ query2 += txn.quote(target_address_id);
+ query2 += ")";
+
+ }
+
+ query+=";";
+ query2+=";";
+ txn.exec(query+query2);
+ next_icfs_id++;
+ }
+ cerr<<"Finished inserting ICFS into IR."<<endl;
+ txn.commit();
+
+}
+
+
// insert addresses & instructions into DB
-void insert_instructions(int fileID, vector<wahoo::Instruction*> instructions, vector<wahoo::Function*> functions)
+void insert_instructions(int fileID, const vector<wahoo::Instruction*> &instructions, const vector<wahoo::Function*> &functions)
{
cerr << "Inserting instructions in the DB"<<endl;
connection conn;
@@ -115,6 +173,7 @@ void insert_instructions(int fileID, vector<wahoo::Instruction*> instructions, v
address_to_instructionid_map[addr]=j;
int address_id = next_address_id++;
+ instruction_to_addressid_map[instruction]=address_id;
// insert into address table
if (j != i) query += ",";
@@ -329,7 +388,7 @@ void populate_predefined_types()
txn.commit();
}
-void update_function_prototype(vector<wahoo::Function*> functions, char* annotFile)
+void update_function_prototype(const vector<wahoo::Function*> &functions, char* annotFile)
{
populate_predefined_types();
@@ -456,9 +515,9 @@ void update_function_prototype(vector<wahoo::Function*> functions, char* annotFi
int main(int argc, char **argv)
{
- if (argc != 10)
+ if (argc != 12)
{
- cerr << "usage: " << argv[0] << " <annotations file> <info annotation file> <file id> <func tab name> <insn tab name> <addr tab name> <types tab name> <elf file> <STARSxref file>" << endl;
+ cerr << "usage: " << argv[0] << " <annotations file> <info annotation file> <file id> <func tab name> <insn tab name> <addr tab name> <types tab name> <icfs table name> <icfs map table name> <elf file> <STARSxref file>" << endl;
return 1;
}
@@ -469,22 +528,30 @@ int main(int argc, char **argv)
char *myInstructionTable=argv[5];
char *myAddressTable=argv[6];
char *myTypesTable=argv[7];
- char *elfFile=argv[8];
- char *starsXrefFile=argv[9];
+ char *myicfsTable=argv[8];
+ char *myicfsMapTable=argv[9];
+ char *elfFile=argv[10];
+ char *starsXrefFile=argv[11];
cout<<"Annotation file: "<< annotFile<<endl;
cout<<"Info annotation file: "<< infoAnnotFile<<endl;
cout<<"File ID: "<< fid<<endl;
- cout<<"FTN:: "<< myFunctionTable<<endl;
+ cout<<"FTN: "<< myFunctionTable<<endl;
cout<<"ITN: "<< myInstructionTable<<endl;
cout<<"ATN: "<< myAddressTable<<endl;
cout<<"TYP: "<< myTypesTable<<endl;
+ cout<<"ICFSTab: "<< myicfsTable<<endl;
+ cout<<"ICFSMapTab: "<< myicfsMapTable<<endl;
+ cout<<"elfFile: "<< elfFile<<endl;
+ cout<<"xrefFile: "<< starsXrefFile<<endl;
// set global vars for importing.
functionTable=myFunctionTable;
addressTable=myAddressTable;
instructionTable=myInstructionTable;
typesTable=myTypesTable;
+ icfsTable=myicfsTable;
+ icfsMapTable=myicfsMapTable;
Rewriter *rewriter = new Rewriter(elfFile, annotFile, starsXrefFile);
@@ -504,6 +571,7 @@ int main(int argc, char **argv)
insert_functions(fileID, functions);
insert_instructions(fileID, instructions, functions);
update_functions(fileID, functions);
+ insert_icfs(fileID, instructions);
// add function prototype information to the IRDB
update_function_prototype(functions, infoAnnotFile);
diff --git a/xform/instruction_descriptor.h b/xform/instruction_descriptor.h
index e09c6bfd7705f4bab2b9402c48feef8c9aa7e9b1..420332a4f631f82f4f4afc95ff19f751f07dfc57 100644
--- a/xform/instruction_descriptor.h
+++ b/xform/instruction_descriptor.h
@@ -2,6 +2,7 @@
#define _instruction_h_
#include <string>
+#include <set>
#include "targ-config.h"
@@ -46,6 +47,10 @@ class Instruction {
void setVisited() { m_isVisited = true; }
bool isVisited() const { return m_isVisited; }
+ // include Indirect branch targets for insructions.
+ void addIBT(Instruction* insn) { ibts.insert(insn); }
+ const std::set<Instruction*>& getIBTs() { return ibts; }
+
private:
app_iaddr_t m_address;
app_iaddr_t m_ibt_address;
@@ -61,6 +66,9 @@ class Instruction {
bool m_varStackRef;
bool m_isVisited;
+
+ std::set<Instruction*> ibts;
+
};
}
diff --git a/xform/rewriter.cpp b/xform/rewriter.cpp
index 7c7a45db44137ee222cdb1227f9992030e135483..6e80e709c2535532d10ae41a9f9625adee439d39 100644
--- a/xform/rewriter.cpp
+++ b/xform/rewriter.cpp
@@ -652,16 +652,20 @@ void Rewriter::readXrefsFile(char p_filename[])
do
{
-
+ addr=0;
fscanf(fin, "%p %d\n", (void**)&addr, &size_type_u.size);
if(feof(fin)) // deal with blank lines at the EOF
break;
fscanf(fin, "%s%s", type,scope);
+ if(feof(fin)) // deal with blank lines at the EOF
+ break;
assert(strcmp(type,"INSTR")==0);
assert(strcmp(scope,"XREF")==0);
- fscanf(fin, "%s%s%s", ibt,fromib,dest);
+ fscanf(fin, "%s%s", ibt,fromib);
+ if(feof(fin)) // deal with blank lines at the EOF
+ break;
assert(strcmp(ibt,"IBT")==0);
assert(strcmp(fromib,"FROMIB")==0 || strcmp(fromib,"FROMDATA")==0
|| strcmp(fromib,"FROMUNKNOWN")==0);
@@ -672,11 +676,29 @@ void Rewriter::readXrefsFile(char p_filename[])
{
// cout<<"Setting IBT for addr "<<std::hex<<addr<<std::dec<<endl;
instr->setIBTAddress(addr);
+ if(strcmp(fromib,"FROMIB")==0)
+ {
+ // get the from point into memory.
+ app_iaddr_t from_addr = 0;
+ fscanf(fin, "%p\n", (void**)&from_addr);
+ if(feof(fin)) // deal with blank lines at the EOF
+ break;
+
+ // find that instruction
+ wahoo::Instruction *from_instr = addr_to_insn_map[from_addr];
+ assert(from_instr);
+
+ // record in the IR listing.
+ from_instr->addIBT(instr);
+ }
}
else
{
- cerr<<"Warning, instruction at "<<std::hex<<addr<<std::dec<<" not in db?"<<endl;
+ cerr<<"Error: instruction at "<<std::hex<<addr<<std::dec<<" not in db?"<<endl;
+ // assert(0);
}
+
+
char remainder[2000];
fgets(remainder, sizeof(remainder), fin);
diff --git a/xform/rewriter.h b/xform/rewriter.h
index 417c2950117b819f6860419aa48df71331f364a8..c035835215fa86a001a66380de4dcd5f89a35584 100644
--- a/xform/rewriter.h
+++ b/xform/rewriter.h
@@ -1,4 +1,5 @@
#include <map>
+#include <set>
#include "targ-config.h"
#include "elfio/elfio.hpp"
#include "elfio/elfio_dump.hpp"
@@ -6,6 +7,7 @@
#include "elfreader.h"
#include "function_descriptor.h"
+
using namespace std;
class Rewriter