https://github.com/akkartik/mu/blob/main/039debug.cc
  1 //:: Some helpers for debugging.
  2 
  3 //: Load the 'map' file generated during 'bootstrap --debug translate' when running
  4 //: 'bootstrap --trace run'.
  5 //: (It'll only affect the trace.)
  6 
  7 :(before "End Globals")
  8 map</*address*/uint32_t, string> Symbol_name;  // used only by 'bootstrap run'
  9 map</*address*/uint32_t, string> Source_line;  // used only by 'bootstrap run'
 10 :(before "End --trace Settings")
 11 load_labels();
 12 load_source_lines();
 13 :(code)
 14 void load_labels() {
 15   ifstream fin("labels");
 16   if (fin.fail()) return;
 17   fin >> std::hex;
 18   while (has_data(fin)) {
 19     uint32_t addr = 0;
 20     fin >> addr;
 21     string name;
 22     fin >> name;
 23     put(Symbol_name, addr, name);
 24   }
 25 }
 26 
 27 void load_source_lines() {
 28   ifstream fin("source_lines");
 29   if (fin.fail()) return;
 30   fin >> std::hex;
 31   while (has_data(fin)) {
 32     uint32_t addr = 0;
 33     fin >> addr;
 34     string line;
 35     getline(fin, line);
 36     put(Source_line, addr, hacky_squeeze_out_whitespace(line));
 37   }
 38 }
 39 
 40 :(after "Run One Instruction")
 41 if (contains_key(Symbol_name, EIP))
 42   trace(Callstack_depth, "run") << "== label " << get(Symbol_name, EIP) << end();
 43 if (contains_key(Source_line, EIP))
 44   trace(Callstack_depth, "run") << "inst: " << get(Source_line, EIP) << end();
 45 else
 46   // no source line info; do what you can
 47   trace(Callstack_depth, "run") << "inst: " << debug_info(EIP) << end();
 48 
 49 :(code)
 50 string debug_info(uint32_t inst_address) {
 51   uint8_t op = read_mem_u8(inst_address);
 52   if (op != 0xe8) {
 53     ostringstream out;
 54     out << HEXBYTE << NUM(op);
 55     return out.str();
 56   }
 57   int32_t offset = read_mem_i32(inst_address+/*skip op*/1);
 58   uint32_t next_eip = inst_address+/*inst length*/5+offset;
 59   if (contains_key(Symbol_name, next_eip))
 60     return "e8/call "+get(Symbol_name, next_eip);
 61   ostringstream out;
 62   out << "e8/call 0x" << HEXWORD << next_eip;
 63   return out.str();
 64 }
 65 
 66 //: If a label starts with '$watch-', make a note of the effective address
 67 //: computed by the next instruction. Start dumping out its contents to the
 68 //: trace after every subsequent instruction.
 69 
 70 :(after "Run One Instruction")
 71 dump_watch_points();
 72 :(before "End Globals")
 73 map<string, uint32_t> Watch_points;
 74 :(before "End Reset")
 75 Watch_points.clear();
 76 :(code)
 77 void dump_watch_points() {
 78   if (Watch_points.empty()) return;
 79   trace(Callstack_depth, "dbg") << "watch points:" << end();
 80   for (map<string, uint32_t>::iterator p = Watch_points.begin();  p != Watch_points.end();  ++p)
 81     trace(Callstack_depth, "dbg") << "  " << p->first << ": " << HEXWORD << p->second << " -> " << HEXWORD << read_mem_u32(p->second) << end();
 82 }
 83 
 84 :(before "End Globals")
 85 string Watch_this_effective_address;
 86 :(after "Run One Instruction")
 87 Watch_this_effective_address = "";
 88 if (contains_key(Symbol_name, EIP) && starts_with(get(Symbol_name, EIP), "$watch-"))
 89   Watch_this_effective_address = get(Symbol_name, EIP);
 90 :(after "Found effective_address(addr)")
 91 if (!Watch_this_effective_address.empty()) {
 92   dbg << "now watching " << HEXWORD << addr << " for " << Watch_this_effective_address << end();
 93   put(Watch_points, Watch_this_effective_address, addr);
 94 }
 95 
 96 //: If a label starts with '$dump-stack', dump out to the trace n bytes on
 97 //: either side of ESP.
 98 
 99 :(after "Run One Instruction")
100 if (contains_key(Symbol_name, EIP) && starts_with(get(Symbol_name, EIP), "$dump-stack")) {
101   dump_stack(64);
102 }
103 :(code)
104 void dump_stack(int n) {
105   uint32_t stack_pointer = Reg[ESP].u;
106   uint32_t start = ((stack_pointer-n)&0xfffffff0);
107   dbg << "stack:" << end();
108   for (uint32_t addr = start;  addr < start+n*2;  addr+=16) {
109     if (addr >= AFTER_STACK) break;
110     ostringstream out;
111     out << HEXWORD << addr << ":";
112     for (int i = 0;  i < 16;  i+=4) {
113       out << ' ';
114       out << ((addr+i == stack_pointer) ? '[' : ' ');
115       out << HEXWORD << read_mem_u32(addr+i);
116       out << ((addr+i == stack_pointer) ? ']' : ' ');
117     }
118     dbg << out.str() << end();
119   }
120 }
121 
122 //: Special label that dumps regions of memory.
123 //: Not a general mechanism; by the time you get here you're willing to hack
124 //: on the emulator.
125 :(after "Run One Instruction")
126 if (contains_key(Symbol_name, EIP) && get(Symbol_name, EIP) == "$dump-stream-at-EAX")
127   dump_stream_at(Reg[EAX].u);
128 :(code)
129 void dump_stream_at(uint32_t stream_start) {
130   int32_t stream_length = read_mem_i32(stream_start + 8);
131   dbg << "stream length: " << std::dec << stream_length << end();
132   for (int i = 0;  i < stream_length + 12;  ++i)
133     dbg << "0x" << HEXWORD << (stream_start+i) << ": " << HEXBYTE << NUM(read_mem_u8(stream_start+i)) << end();
134 }
135 
136 //: helpers
137 
138 :(code)
139 string hacky_squeeze_out_whitespace(const string& s) {
140   // strip whitespace at start
141   string::const_iterator first = s.begin();
142   while (first != s.end() && isspace(*first))
143     ++first;
144   if (first == s.end()) return "";
145 
146   // strip whitespace at end
147   string::const_iterator last = --s.end();
148   while (last != s.begin() && isspace(*last))
149     --last;
150   ++last;
151 
152   // replace runs of spaces/dots with single space until comment or string
153   // TODO:
154   //   leave alone dots not surrounded by whitespace
155   //   leave alone '#' within word
156   //   leave alone '"' within word
157   //   squeeze spaces after end of string
158   ostringstream out;
159   bool previous_was_space = false;
160   bool in_comment_or_string = false;
161   for (string::const_iterator curr = first;  curr != last;  ++curr) {
162     if (in_comment_or_string)
163       out << *curr;
164     else if (isspace(*curr) || *curr == '.')
165       previous_was_space = true;
166     else {
167       if (previous_was_space)
168         out << ' ';
169       out << *curr;
170       previous_was_space = false;
171       if (*curr == '#' || *curr == '"') in_comment_or_string = true;
172     }
173   }
174   return out.str();
175 }