Add C++ code to make use of .pc_list

src/CodeGenerator.cpp

@@ -2,6 +2,8 @@
 #include "gen_context_struct.hpp"
 #include "settings.hpp"
 
+#include <algorithm>
+
 using namespace std;
 
 static const char* PRELUDE =
@@ -13,8 +15,12 @@ CodeGenerator::CodeGenerator(
         const SimpleDwarf& dwarf,
         std::ostream& os,
         NamingScheme naming_scheme):
-    dwarf(dwarf), os(os), naming_scheme(naming_scheme)
-{}
+    dwarf(dwarf), os(os), pc_list(nullptr), naming_scheme(naming_scheme)
+{
+    if(!settings::pc_list.empty()) {
+        pc_list = make_unique<PcListReader>(settings::pc_list);
+    }
+}
 
 void CodeGenerator::generate() {
     gen_of_dwarf();
@@ -95,8 +101,7 @@ void CodeGenerator::gen_of_row(
         const SimpleDwarf::DwRow& row,
         uintptr_t row_end)
 {
-    os << "\t\tcase " << std::hex << "0x" << row.ip
-       << " ... 0x" << row_end << ":" << std::dec << endl;
+    gen_case(row.ip, row_end);
 
     os << "\t\t\t" << "out_ctx.rsp = ";
     gen_of_reg(row.cfa);
@@ -113,6 +118,28 @@ void CodeGenerator::gen_of_row(
     os << "\t\t\treturn " << "out_ctx" << ";" << endl;
 }
 
+void CodeGenerator::gen_case(uintptr_t low_bound, uintptr_t high_bound) {
+    if(pc_list == nullptr) {
+        os << "\t\tcase " << std::hex << "0x" << low_bound
+           << " ... 0x" << high_bound << ":" << std::dec << endl;
+    }
+    else {
+        const auto& first_it = lower_bound(
+                pc_list->get_list().begin(),
+                pc_list->get_list().end(),
+                low_bound);
+        const auto& last_it = upper_bound(
+                pc_list->get_list().begin(),
+                pc_list->get_list().end(),
+                high_bound);
+
+        os << std::hex;
+        for(auto it = first_it; it != last_it; ++it)
+            os << "\t\tcase 0x" << *it << ":\n";
+        os << std::dec;
+    }
+}
+
 static const char* ctx_of_dw_name(SimpleDwarf::MachineRegister reg) {
     switch(reg) {
         case SimpleDwarf::REG_RIP:

src/CodeGenerator.hpp

@@ -4,8 +4,10 @@
 
 #include <ostream>
 #include <functional>
+#include <memory>
 
 #include "SimpleDwarf.hpp"
+#include "PcListReader.hpp"
 
 class CodeGenerator {
     public:
@@ -38,6 +40,7 @@ class CodeGenerator {
         void gen_of_row(
                 const SimpleDwarf::DwRow& row,
                 uintptr_t row_end);
+        void gen_case(uintptr_t low_bound, uintptr_t high_bound);
         void gen_of_reg(
                 const SimpleDwarf::DwRegister& reg);
 
@@ -46,6 +49,7 @@ class CodeGenerator {
     private:
         SimpleDwarf dwarf;
         std::ostream& os;
+        std::unique_ptr<PcListReader> pc_list;
 
         NamingScheme naming_scheme;
 };

src/Makefile

@@ -4,7 +4,13 @@ CXXFLAGS=$(CXXLOCS) -Wall -Wextra -std=c++14 -O2 -g
 CXXLIBS=-lelf -ldwarf -ldwarfpp -lsrk31c++ -lc++fileno
 
 TARGET=dwarf-assembly
-OBJS=DwarfReader.o SimpleDwarf.o CodeGenerator.o settings.o main.o
+OBJS=\
+	DwarfReader.o \
+	SimpleDwarf.o \
+	CodeGenerator.o \
+	PcListReader.o \
+	settings.o \
+	main.o
 
 ###############################################################################
 

src/PcListReader.cpp

@@ -0,0 +1,32 @@
+#include "PcListReader.hpp"
+
+#include <fstream>
+
+using namespace std;
+
+PcListReader::PcListReader(const std::string& path): path(path)
+{}
+
+void PcListReader::read() {
+    ifstream handle(path, ifstream::in | ifstream::binary);
+    if(!handle.good()) {
+        throw PcListReader::CannotReadFile();
+    }
+
+    uintptr_t last_pc;
+    unsigned char buffer[8];
+
+    while(!handle.eof()) {
+        handle.read((char*)buffer, 8);
+        if(handle.gcount() != 8)
+            throw PcListReader::BadFormat();
+
+        last_pc = 0;
+        for(int shift = 0; shift < 8; ++shift) {
+            uintptr_t c_byte = buffer[shift];
+            c_byte <<= 8*shift;
+            last_pc |= c_byte;
+        }
+        pc_list.push_back(last_pc);
+    }
+}

src/PcListReader.hpp

@@ -0,0 +1,30 @@
+/** Reads .pc_list files, containing a list of valid program counters for some
+ * elf file.
+ */
+
+#pragma once
+
+#include <vector>
+#include <string>
+#include <cstdint>
+
+class PcListReader {
+    public:
+        /// Thrown when the file is somehow not readable
+        class CannotReadFile: std::exception {};
+
+        /// Thrown when the file contains bad content (probably not aligned 8B)
+        class BadFormat: std::exception {};
+
+        PcListReader(const std::string& path);
+
+        /// Actually read and process the file
+        void read();
+
+        /// Access the PC list (filled iff `read` was called before)
+        std::vector<uintptr_t>& get_list() { return pc_list; }
+
+    private:
+        std::string path;
+        std::vector<uintptr_t> pc_list;
+};

src/main.cpp

@@ -45,6 +45,17 @@ MainOptions options_parse(int argc, char** argv) {
             settings::switch_generation_policy =
                 settings::SGP_GlobalSwitch;
         }
+
+        else if(option == "--pc-list") {
+            if(option_pos + 1 == argc) { // missing parameter
+                exit_status = 1;
+                print_helptext = true;
+            }
+            else {
+                ++option_pos;
+                settings::pc_list = argv[option_pos];
+            }
+        }
     }
 
     if(!seen_switch_gen_policy) {

src/settings.cpp

@@ -2,4 +2,5 @@
 
 namespace settings {
     SwitchGenerationPolicy switch_generation_policy = SGP_SwitchPerFunc;
+    std::string pc_list = "";
 }

src/settings.hpp

@@ -4,6 +4,8 @@
 
 #pragma once
 
+#include <string>
+
 namespace settings {
     /// Controls how the eh_elf switches are generated
     enum SwitchGenerationPolicy {
@@ -12,4 +14,5 @@ namespace settings {
     };
 
     extern SwitchGenerationPolicy switch_generation_policy;
+    extern std::string pc_list;
 }