Getting BASE type info, printing variable type and pretty value

.gitignore

@@ -4,3 +4,4 @@ res/001-simple
 res/002-compilation-units
 res/003-recursion
 res/004-scoped-variables
+res/005-base-types

command.c

@@ -147,13 +147,88 @@ command_print(struct mi_process proc, char *name, int name_length)
     struct mi_registers regs = get_process_registers(proc);
     u64 pc = regs.rip - proc.base_address;
     u64 cfa = get_cfa_at_pc(proc, pc);
+    
     struct mi_variable *variable = get_variable(proc, name, name_length, pc);
+    struct mi_type type = proc.debug.types[variable->type];
     
     if (cfa && variable) {
         u64 address = cfa + variable->location;
-        long value = ptrace(PTRACE_PEEKDATA, proc.pid, address, NULL);
-        int val4 = value & 0x00000000FFFFFFFF;
-        printf("%.*s = %d\n", name_length, name, val4);
+        u64 raw_value = ptrace(PTRACE_PEEKDATA, proc.pid, address, NULL);
+        
+        u8  raw_value1;
+        u16 raw_value2;
+        u32 raw_value4;
+        
+        if (type.size == 1) {
+            raw_value1 = (raw_value & 0x00000000000000ffUL);
+        } else if (type.size == 2) {
+            raw_value2 = (raw_value & 0x000000000000ffffUL);
+        } else if (type.size == 4) {
+            raw_value4 = (raw_value & 0x00000000ffffffffUL);
+        }
+        
+        switch (type.encoding) {
+            case MI_ADDRESS: {
+                printf("(address) %.*s = %#lx\n", name_length, name, raw_value); 
+                break;
+            }
+            
+            case MI_BOOLEAN: {
+                if (raw_value) {
+                    printf("(boolean) %.*s = true\n", name_length, name);
+                } else {
+                    printf("(boolean) %.*s = false\n", name_length, name);
+                }
+                break;
+            }
+            
+            case MI_FLOAT: {
+                if (type.size == 4) { 
+                    f32 float_val;
+                    memcpy(&float_val, &raw_value4, 4);
+                    printf("(float) %.*s = %f\n", name_length, name, float_val);
+                } else if (type.size == 8) {
+                    f64 float_val;
+                    memcpy(&float_val, &raw_value, 8);
+                    printf("(double) %.*s = %f\n", name_length, name, float_val);
+                }
+                break;
+            }
+            
+            case MI_SIGNED: {
+                if (type.size == 1) {
+                    s8 signed_value1;
+                    memcpy(&signed_value1, &raw_value1, 1);
+                    printf("(s8) %.*s = %d\n", name_length, name, signed_value1);
+                } else if (type.size == 2) {
+                    s16 signed_value2;
+                    memcpy(&signed_value2, &raw_value2, 2);
+                    printf("(s16) %.*s = %d\n", name_length, name, signed_value2);
+                } else if (type.size == 4) {
+                    s32 signed_value4;
+                    memcpy(&signed_value4, &raw_value4, 4);
+                    printf("(s32) %.*s = %d\n", name_length, name, signed_value4);
+                } else if (type.size == 8) {
+                    s64 signed_value8;
+                    memcpy(&signed_value8, &raw_value, 8);
+                    printf("(s64) %.*s = %ld\n", name_length, name, signed_value8);
+                }
+                break;
+            }
+            
+            case MI_UNSIGNED: {
+                if (type.size == 1) {
+                    printf("(u8) %.*s = %u\n", name_length, name, raw_value1);
+                } else if (type.size == 2) {
+                    printf("(u16) %.*s = %u\n", name_length, name, raw_value2);
+                } else if (type.size == 4) {
+                    printf("(u32) %.*s = %u\n", name_length, name, raw_value4);
+                } else if (type.size == 8) {
+                    printf("(u64) %.*s = %lu\n", name_length, name, raw_value);
+                }
+                break;
+            }
+        }
     } else {
         printf("Variable %.*s not found\n", name_length, name);
     }

common.h

@@ -29,8 +29,19 @@ typedef int16_t s16;
 typedef int32_t s32;
 typedef int64_t s64;
 
+typedef float  f32;
+typedef double f64;
+
 #include "elf_dwarf.h"
 
+enum mi_type_encoding {
+    MI_ADDRESS,
+    MI_BOOLEAN,
+    MI_FLOAT,
+    MI_SIGNED,
+    MI_UNSIGNED
+};
+
 struct mi_buffer {
     u8 *data;
     u64 size;
@@ -49,9 +60,20 @@ struct mi_sourcefile {
     struct mi_buffer file;
 };
 
+struct mi_type {
+    char *name;
+    int size;
+    enum mi_type_encoding encoding;
+    
+    // used during parsing so that variables (which only have offsets) could find their
+    // corresponding parsed type record
+    s64 _offset;
+};
+
 struct mi_variable {
     char *name;
     s64 location;
+    int type;
 };
 
 struct mi_function {
@@ -89,10 +111,12 @@ struct mi_debuginfo {
     struct mi_compunit compilation_units[16];
     struct mi_function functions[64]; 
     struct mi_variable variables[64];
+    struct mi_type types[64];
     
     int cu_count;
     int func_count;
     int var_count;
+    int type_count;
 };
 
 struct mi_process {

dwarf.c

@@ -119,6 +119,7 @@ abbrev_entry_offset(u8 *file, u64 abbrev_offset, u32 requested_code)
         
         u32 has_children = file[abbrev_offset + offset++];
         (void) has_children;
+        
         if (code == requested_code) {
             return(abbrev_offset);
         }
@@ -654,7 +655,7 @@ TODO: The DW_OP_form_tls_address operation pops a value from the stack, translat
 }
 
 static u32
-read_actual_debug_data(u8 *file, u64 string_offset, u32 address_size, u64 base_data_offset, u32 form, u64 data_offset, u64 *value)
+read_actual_debug_data(u8 *file, u64 string_offset, u32 address_size, u32 form, u64 data_offset, u64 *value)
 {
     u32 increment = 0;
     
@@ -691,7 +692,8 @@ read_actual_debug_data(u8 *file, u64 string_offset, u32 address_size, u64 base_d
             u32 offset;
             memcpy(&offset, file + data_offset, 4);
             increment = 4;
-            *value = file[base_data_offset + offset];
+            //*value = file[base_data_offset + offset];
+            *value = offset;
             break;
         }
         
@@ -765,15 +767,21 @@ read_debug_info_for_compilation_unit(u8 *file, struct mi_debuginfo *dest,
     u32 code, tag;
     u64 schema_offset;
     u32 depth = 0;
-    u64 base_data_offset = data_offset - header_size;
+    //u64 base_data_offset = data_offset - header_size;
+    
+    int vars_from = dest->var_count;
+    int types_from = dest->type_count;
     
     struct mi_compunit *comp_unit = dest->compilation_units + dest->cu_count;
     struct mi_function *func = dest->functions + dest->func_count;
     struct mi_variable *variable = dest->variables + dest->var_count;
+    struct mi_type     *type = dest->types + dest->type_count;
     
     comp_unit->functions_from = dest->func_count;
     
     for (;;) {
+        s64 record_offset = data_offset - debug_info_offset;
+        
         data_offset += decode_leb128(file + data_offset, &code);
         
         if (code == 0) {
@@ -804,7 +812,7 @@ read_debug_info_for_compilation_unit(u8 *file, struct mi_debuginfo *dest,
             
             u64 value;
             
-            data_offset += read_actual_debug_data(file, debug_str_offset, di_header.address_size, base_data_offset, form, data_offset, &value);
+            data_offset += read_actual_debug_data(file, debug_str_offset, di_header.address_size, form, data_offset, &value);
             
             if (tag == DW_TAG_compile_unit) {
                 if (attribute == DW_AT_low_pc) {
@@ -827,6 +835,49 @@ read_debug_info_for_compilation_unit(u8 *file, struct mi_debuginfo *dest,
                     variable->name = (char *) value;
                 } else if (attribute == DW_AT_location) {
                     variable->location = value;
+                } else if (attribute == DW_AT_type) {
+                    variable->type = value;
+                }
+            } else if (tag == DW_TAG_base_type) {
+                type->_offset = record_offset;
+                if (attribute == DW_AT_name) {
+                    type->name = (char *) value;
+                } else if (attribute == DW_AT_byte_size) {
+                    type->size = value;
+                } else if (attribute == DW_AT_encoding) {
+                    enum dwarf_type_encoding encoding = value;
+                    switch (encoding) {
+                        case DW_ATE_address: {
+                            type->encoding = MI_ADDRESS;
+                            break;
+                        }
+                        
+                        case DW_ATE_boolean: {
+                            type->encoding = MI_BOOLEAN;
+                            break;
+                        }
+                        
+                        case DW_ATE_float: {
+                            type->encoding = MI_FLOAT;
+                            break;
+                        }
+                        
+                        case DW_ATE_signed_char:
+                        case DW_ATE_signed: {
+                            type->encoding = MI_SIGNED;
+                            break;
+                        }
+                        
+                        case DW_ATE_unsigned_char:
+                        case DW_ATE_unsigned: {
+                            type->encoding = MI_UNSIGNED;
+                            break;
+                        }
+                        
+                        default: {
+                            DIE("unexpected type encoding!\n");
+                        }
+                    }
                 }
             }
         } while (attribute != 0 || form != 0);
@@ -843,6 +894,23 @@ read_debug_info_for_compilation_unit(u8 *file, struct mi_debuginfo *dest,
             ++parent->variables_count;
             ++dest->var_count;
             ++variable;
+        } else if (tag == DW_TAG_base_type) {
+            ++dest->type_count;
+            ++type;
+        }
+    }
+    
+    // Resolve types
+    for (int v = vars_from; v < dest->var_count; ++v) {
+        struct mi_variable *variable = dest->variables + v;
+        
+        for (int t = types_from; t < dest->type_count; ++t) {
+            struct mi_type *type = dest->types + t;
+            
+            if (type->_offset == variable->type) {
+                variable->type = t;
+                break;
+            }
         }
     }
     

elf_dwarf.h

@@ -478,6 +478,28 @@ enum dwarf_expression_op {
     DW_OP_hi_user = 0xff
 };
 
+enum dwarf_type_encoding {
+    DW_ATE_address = 0x01,
+    DW_ATE_boolean = 0x02,
+    DW_ATE_complex_float = 0x03,
+    DW_ATE_float = 0x04,
+    DW_ATE_signed = 0x05,
+    DW_ATE_signed_char = 0x06,
+    DW_ATE_unsigned = 0x07,
+    DW_ATE_unsigned_char = 0x08,
+    DW_ATE_imaginary_float = 0x09,
+    DW_ATE_packed_decimal = 0x0a,
+    DW_ATE_numeric_string = 0x0b,
+    DW_ATE_edited = 0x0c,
+    DW_ATE_signed_fixed = 0x0d,
+    DW_ATE_unsigned_fixed = 0x0e,
+    DW_ATE_decimal_float = 0x0f,
+    DW_ATE_UTF = 0x10,
+    
+    DW_ATE_lo_user = 0x80,
+    DW_ATE_hi_user = 0xff,
+};
+
 enum dwarf_cfa_op_base {
     DW_CFA_advance_loc = 0x1,
     DW_CFA_offset = 0x2,

main.c

@@ -21,9 +21,11 @@ main(int argc, char *argv[])
     char *command = malloc(max_command_length + 1);
     char *last_command = malloc(max_command_length + 1);
     
+    process.base_address = 0x555555554000UL; // get_executable_base_address(file, proc.pid);
+    
     parse_debug_info(process.elf, &process.debug);
+    parse_debug_line(process.elf, &process.debug);
     
-    process.base_address = 0x555555554000UL; // get_executable_base_address(file, proc.pid);
     process.main_address = get_address_of_subroutine(process, "main");
     
     printf("Base address: %#lx\n", process.base_address);
@@ -32,8 +34,6 @@ main(int argc, char *argv[])
     printf("> ");
     fflush(stdout);
     
-    parse_debug_line(process.elf, &process.debug);
-    
 #if 0
     command_start(process);
     command_step(process);

res/005-base-types.c

@@ -0,0 +1,13 @@
+int
+main(int argc, char **argv)
+{
+	int var_int = -123;
+	unsigned int var_unsigned = 1;
+	long var_long = -999;
+	unsigned long var_unsigned_long = 3;
+	char var_char = '4';
+	unsigned char var_unsigned_char = '5';
+	float var_float = 3.14f;
+	double var_double = 3.14 * 2.0;
+	void *var_void = (void *) 0xdeadbeef;
+}

res/Makefile

@@ -3,3 +3,4 @@ all:
 	gcc -g -o 002-compilation-units 002-compilation-units.c 002-compilation-units-impl.c
 	gcc -g -o 003-recursion 003-recursion.c
 	gcc -g -o 004-scoped-variables 004-scoped-variables.c
+	gcc -g -o 005-base-types 005-base-types.c