Get actual DIE _tree_. No useful info extracted from the tree for now. Nodes malloc()-ed all over the place

src/dwarf.c

@@ -967,8 +967,8 @@ read_die_structure_type(u8 *file, u64 debug_str_offset, u64 *schema_offset, u64
 }
 
 static void
-read_die_member(u8 *file, u64 debug_str_offset, u64 *schema_offset, u64 *data_offset, 
-                u8 address_size, struct dwarf_die_type *dest)
+read_die_structure_member(u8 *file, u64 debug_str_offset, u64 *schema_offset, u64 *data_offset, 
+                          u8 address_size, struct dwarf_die_type *dest)
 {
     u64 data = *data_offset;
     u64 schema = *schema_offset;
@@ -979,6 +979,18 @@ read_die_member(u8 *file, u64 debug_str_offset, u64 *schema_offset, u64 *data_of
         schema += decode_leb128(file + schema, &attribute);
         schema += decode_leb128(file + schema, &form);
         data += read_actual_debug_data(file, debug_str_offset, address_size, form, data, &value);
+        
+        switch (attribute) {
+            case DW_AT_name: {
+                dest->name = (char *) value;
+                break;
+            }
+            
+            case DW_AT_type: {
+                dest->pointer = value;
+                break;
+            }
+        }
     } while (attribute != 0 || form != 0);
     
     *data_offset = data;
@@ -986,53 +998,155 @@ read_die_member(u8 *file, u64 debug_str_offset, u64 *schema_offset, u64 *data_of
 }
 
 static void
-read_one_die(u8 *file, u64 debug_info_offset, u64 debug_str_offset, u64 *schema_offset, u64 *data_offset, 
-             enum dwarf_die_tag tag, u8 address_size, struct dwarf_die *dest)
+read_die_pointer_type(u8 *file, u64 debug_str_offset, u64 *schema_offset, u64 *data_offset, 
+                      u8 address_size, struct dwarf_die_type *dest)
 {
+    u64 data = *data_offset;
     u64 schema = *schema_offset;
+    
+    u32 attribute, form;
+    do {
+        u64 value;
+        schema += decode_leb128(file + schema, &attribute);
+        schema += decode_leb128(file + schema, &form);
+        data += read_actual_debug_data(file, debug_str_offset, address_size, form, data, &value);
+        
+        switch (attribute) {
+            case DW_AT_byte_size: {
+                dest->size = value;
+                break;
+            }
+            
+            case DW_AT_type: {
+                dest->pointer = value;
+                break;
+            }
+        }
+    } while (attribute != 0 || form != 0);
+    
+    *data_offset = data;
+    *schema_offset = schema;
+}
+
+static void
+read_die_const_type(u8 *file, u64 debug_str_offset, u64 *schema_offset, u64 *data_offset, 
+                    u8 address_size, struct dwarf_die_type *dest)
+{
     u64 data = *data_offset;
+    u64 schema = *schema_offset;
     
-    dest->offset = data - debug_info_offset;
-    dest->tag = tag;
+    u32 attribute, form;
+    do {
+        u64 value;
+        schema += decode_leb128(file + schema, &attribute);
+        schema += decode_leb128(file + schema, &form);
+        data += read_actual_debug_data(file, debug_str_offset, address_size, form, data, &value);
+        
+        switch (attribute) {
+            case DW_AT_type: {
+                dest->pointer = value;
+                break;
+            }
+        }
+    } while (attribute != 0 || form != 0);
     
+    *data_offset = data;
+    *schema_offset = schema;
+}
+
+// NOTE(aolo2): parse one DIE and all of its children
+static struct dwarf_die *
+read_one_die(u8 *file, u64 string_offset, u64 info_offset, u64 abbrev_offset, u8 address_size, u64 *data_offset, u32 depth)
+{
+    u64 data = *data_offset;
+    u64 record_offset = data - info_offset;
+    
+    u32 code;
+    u32 tag;
+    
+    data += decode_leb128(file + data, &code);
+    
+    if (code == 0) {
+        *data_offset = data;
+        return(0);
+    }
+    
+    // TODO: use a custom allocator / preallocated memory for this
+    struct dwarf_die *die = calloc(1, sizeof(struct dwarf_die));
+    
+    u64 schema_offset = abbrev_entry_offset(file, abbrev_offset, code);
+    schema_offset += decode_leb128(file + schema_offset, NULL);
+    schema_offset += decode_leb128(file + schema_offset, &tag);
+    
+    u8 has_children = file[schema_offset++];
+    
+    die->tag = tag;
+    die->offset = record_offset;
+    
+    // TODO: remove code duplication for prologue/epilogue of all these functions
     switch (tag) {
         case DW_TAG_compile_unit: {
-            read_die_compilation_unit(file, debug_str_offset, &schema, &data, address_size, &dest->compilation_unit);
+            read_die_compilation_unit(file, string_offset, &schema_offset, &data, address_size, &die->compilation_unit);
             break;
         }
         
         case DW_TAG_subprogram: {
-            read_die_subprogram(file, debug_str_offset, &schema, &data, address_size, &dest->subprogram);
+            read_die_subprogram(file, string_offset, &schema_offset, &data, address_size, &die->subprogram);
             break;
         }
         
         case DW_TAG_variable: {
-            read_die_variable(file, debug_str_offset, &schema, &data, address_size, &dest->variable);
+            read_die_variable(file, string_offset, &schema_offset, &data, address_size, &die->variable);
             break;
         }
         
         case DW_TAG_base_type: {
-            read_die_base_type(file, debug_str_offset, &schema, &data, address_size, &dest->type);
+            read_die_base_type(file, string_offset, &schema_offset, &data, address_size, &die->type);
             break;
         }
         
         case DW_TAG_structure_type: {
-            read_die_structure_type(file, debug_str_offset, &schema, &data, address_size, &dest->type);
+            read_die_structure_type(file, string_offset, &schema_offset, &data, address_size, &die->type);
             break;
         }
         
         case DW_TAG_member: {
-            read_die_member(file, debug_str_offset, &schema, &data, address_size, &dest->type);
+            read_die_structure_member(file, string_offset, &schema_offset, &data, address_size, &die->type);
+            break;
+        }
+        
+        case DW_TAG_pointer_type: {
+            read_die_pointer_type(file, string_offset, &schema_offset, &data, address_size, &die->type);
+            break;
+        }
+        
+        case DW_TAG_const_type: {
+            read_die_const_type(file, string_offset, &schema_offset, &data, address_size, &die->type);
             break;
         }
         
         default: {
-            read_die_nop(file, debug_str_offset, &schema, &data, address_size);
+            read_die_nop(file, string_offset, &schema_offset, &data, address_size);
         }
-    } 
+    }
+    
+    if (has_children) {
+        die->first_child = read_one_die(file, string_offset, info_offset, abbrev_offset, address_size, &data, depth + 1);
+    }
+    
+    if (depth > 0) {
+        die->next = read_one_die(file, string_offset, info_offset, abbrev_offset, address_size, &data, depth);
+    }
     
-    *schema_offset = schema;
     *data_offset = data;
+    
+    return(die);
+}
+
+static void
+decrap_die_tree(struct mi_debuginfo *dest, struct dwarf_die *die_root)
+{
+    
 }
 
 static u64
@@ -1045,49 +1159,11 @@ read_debug_info_for_compilation_unit(u8 *file, struct mi_debuginfo *dest,
     
     u64 abbrev_offset = debug_abbrev_offset + di_header.debug_abbrev_offset;
     u64 data_offset = debug_info_offset + header_size;
-    u8 address_size = di_header.address_size;
-    
-    u32 code, tag;
-    u64 schema_offset;
-    u32 depth = 0;
-    //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;
-    
-    struct dwarf_die d_die = { 0 };
-    struct dwarf_die *die = &d_die;
+    u8  address_size = di_header.address_size;
     
-    comp_unit->functions_from = dest->func_count;
+    struct dwarf_die *root_die = read_one_die(file, debug_str_offset, debug_info_offset, abbrev_offset, address_size, &data_offset, 0);
     
-    for (;;) {
-        data_offset += decode_leb128(file + data_offset, &code);
-        
-        if (code == 0) {
-            if (depth > 1) {
-                --depth;
-                continue;
-            } else {
-                break;
-            }
-        }
-        
-        schema_offset = abbrev_entry_offset(file, abbrev_offset, code);
-        schema_offset += decode_leb128(file + schema_offset, NULL);
-        schema_offset += decode_leb128(file + schema_offset, &tag);
-        
-        u32 has_children = file[schema_offset++];
-        if (has_children) {
-            ++depth;
-        }
-        
-        read_one_die(file, debug_info_offset, debug_str_offset, &schema_offset, &data_offset, tag, address_size, die);
-    }
+    decrap_die_tree(dest, root_die);
     
     return(di_header.length + 4);
 }

src/elf_dwarf.h

@@ -753,13 +753,15 @@ struct dwarf_die_type {
     int is_pointer;
     int is_restrict;
     
-    struct dwarf_die_type *next;
-    struct dwarf_die_type *first_child;
+    u64 pointer;
 };
 
 struct dwarf_die {
     u64 offset;
-    struct dwarf_die *parent;
+    
+    struct dwarf_die *first_child;
+    struct dwarf_die *next;
+    
     enum dwarf_die_tag tag;
     union {
         struct dwarf_die_compilation_unit compilation_unit;