trace/dwarf.c

static u64
get_section_offset(u8 *file, char *name)
{
    struct elf_header_x64 header = { 0 };
    memcpy(&header, file, sizeof(header));
    
    for (int i = 0; i < header.header_table_entry_count; ++i) {
        struct elf_header_table_entry_x64 header_entry = { 0 };
        u64 offset = header.header_table_offset + header.header_table_entry_size * i;
        memcpy(&header_entry, file + offset, sizeof(header_entry));
        int a = 1;
    }
    
    struct elf_section_table_entry_x64 shstrtab_header = { 0 };
    u64 shstrtab_header_offset = header.section_table_offset + header.section_table_entry_size * header.section_names_table_index;
    memcpy(&shstrtab_header, file + shstrtab_header_offset, sizeof(shstrtab_header));
    
    u64 shstrtab_offset = shstrtab_header.offset_in_file;
    u64 debug_info_offset = 0;
    
    for (int i = 0; i < header.section_table_entry_count; ++i) {
        struct elf_section_table_entry_x64 section_entry = { 0 };
        u64 offset = header.section_table_offset + header.section_table_entry_size * i;
        memcpy(&section_entry, file + offset, sizeof(section_entry));
        u64 section_name_offset = shstrtab_offset + section_entry.name_offset;
        
        if (strncmp((char *) file + section_name_offset, name, strlen(name) + 1) == 0) {
            debug_info_offset = section_entry.offset_in_file;
        }
        
    }
    
    return(debug_info_offset);
}

static int
dwarf_section_contribution_is_64(u8 *file, u64 offset)
{
    u32 length;
    memcpy(&length, file + offset, 4);
    return(length >= 0xFFFFFFF0);
}

static int
decode_leb128(u8 *at, u32 *dest)
{
    int offset = 0;
    
    u64 result = 0;
    u64 shift = 0;
    
    while (1) {
        u8 byte = at[offset++];
        
        result |= ((byte & 127) << shift);
        
        if ((byte & 128) == 0) {
            break;
        }
        
        shift += 7;
    }
    
    if (dest) {
        *dest = result;
    }
    
    return(offset);
}

static int
decode_leb128s(u8 *at, s32 *dest)
{
    int offset = 0;
    
    s64 result = 0;
    u64 shift = 0;
    u32 size = 32;
    u8 byte;
    
    while (1) {
        byte = at[offset++];
        result |= ((byte & 127) << shift);
        shift += 7;
        if ((byte & 128) == 0) {
            break;
        }
    }
    
    if ((shift < size) && (byte & 64)) {
        result |= -(1 << shift);
    }
    
    if (dest) {
        *dest = result;
    }
    
    return(offset);
}

static u64
abbrev_entry_offset(u8 *file, u64 abbrev_offset, u32 requested_code)
{
    u32 code, tag;
    u32 offset = 0;
    
    do {
        offset += decode_leb128(file + abbrev_offset + offset, &code);
        offset += decode_leb128(file + abbrev_offset + offset, &tag);
        
        u32 has_children = file[abbrev_offset + offset++];
        
        if (code == requested_code) {
            return(abbrev_offset);
        }
        
        if (code == 0) {
            /* Abbreviation code not found, this should not happen */
            assert(0);
            return(0);
        }
        
        u32 attribute, form;
        
        do {
            offset += decode_leb128(file + abbrev_offset + offset, &form);
            offset += decode_leb128(file + abbrev_offset + offset, &attribute);
        } while (attribute != 0 || form != 0);
        
        abbrev_offset += offset;
        offset = 0;
    } while (code != 0);
    
    assert(0);
    
    return(0);
}

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)
{
    u32 increment = 0;
    
    switch (form) {
        case DW_FORM_sec_offset:
        case DW_FORM_strp: {
            u32 offset;
            memcpy(&offset, file + data_offset, 4);
            char *str = (char *) (file + string_offset + offset);
            *value = (u64) str;
            increment = 4; // 8 bytes for x64 DWARF!
            break;
        }
        
        case DW_FORM_addr: {
            memcpy(value, file + data_offset, address_size);
            increment = address_size;
            break;
        }
        
        case DW_FORM_string: {
            char *ptr = (char *) (file + data_offset);
            *value = (u64) ptr;
            increment = strlen(ptr) + 1;
            break;
        }
        
        case DW_FORM_flag_present: {
            *value = 1;
            break;
        }
        
        case DW_FORM_ref4: {
            u32 offset;
            memcpy(&offset, file + data_offset, 4);
            increment = 4;
            *value = file[base_data_offset + offset];
            break;
        }
        
        case DW_FORM_exprloc: {
            // TODO: return value to caller
            
            u32 length;
            increment = decode_leb128(file + data_offset, &length);
            increment += length;
            
            break;
        }
        
        case DW_FORM_data1: {
            *value = file[data_offset];
            increment = 1;
            break;
        }
        
        case DW_FORM_data2: {
            memcpy(value, file + data_offset, 2);
            increment = 2;
            break;
        }
        
        case DW_FORM_data4: {
            memcpy(value, file + data_offset, 4);
            increment = 4;
            break;
        }
        
        case DW_FORM_data8: {
            memcpy(value, file + data_offset, 8);
            increment = 8;
            break;
        }
        
        case DW_FORM_sdata: {
            increment = decode_leb128s(file + data_offset, (s32 *) value);
            break;
        }
        
        case DW_FORM_udata: {
            increment = decode_leb128(file + data_offset, (u32 *) value);
            break;
        }
        
        default: {
            if (form) {
                printf("unknown attribute form %d\n", form);
            }
        }
    }
    
    return(increment);
}

static void
parse_debug_info(u8 *file, struct mi_debuginfo *dest)
{
    u64 debug_info_offset = get_section_offset(file, ".debug_info");
    printf("Found .debug_info at offset %#lx\n", debug_info_offset);
    
    u64 debug_abbrev_offset = get_section_offset(file, ".debug_abbrev");
    printf("Found .debug_abbrev at offset %#lx\n", debug_abbrev_offset);
    
    u64 debug_str_offset = get_section_offset(file, ".debug_str");
    printf("Found .debug_str at offset %#lx\n", debug_str_offset);
    
    struct dwarf_debug_info_header_x32 di_header = { 0 };
    u32 header_size = sizeof(di_header);
    memcpy(&di_header, file + debug_info_offset, header_size);
    
    u64 abbrev_offset = debug_abbrev_offset + di_header.debug_abbrev_offset;
    u64 data_offset = debug_info_offset + header_size;
    
    u32 code, tag;
    u64 schema_offset;
    u32 depth = 0;
    u64 base_data_offset = data_offset - header_size;
    
    int found_sr = 0;
    
    struct mi_function *func = dest->functions;
    
    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);
        
        //printf("%d %s\n", code, tag_to_str(tag));
        
        u32 has_children = file[schema_offset++];
        if (has_children) {
            ++depth;
        }
        
        u32 attribute, form;
        
        do {
            schema_offset += decode_leb128(file + schema_offset, &attribute);
            schema_offset += decode_leb128(file + schema_offset, &form);
            
            u64 value;
            
            data_offset += read_actual_debug_data(file, debug_str_offset, di_header.address_size, base_data_offset, form, data_offset, &value);
            
            if (tag == DW_TAG_compile_unit) {
                if (attribute == DW_AT_comp_dir) {
                    dest->comp_dir = (char *) value;
                }
            } else if (tag == DW_TAG_subprogram) {
                if (attribute == DW_AT_name) {
                    func->name = (char *) value;
                } else if (attribute == DW_AT_low_pc) {
                    func->offset = value;
                }
            }
        } while (attribute != 0 || form != 0);
        
        if (tag == DW_TAG_subprogram) {
            ++func;
            ++dest->func_count;
        }
    }
}

static void
parse_debug_line(u8 *file, struct mi_sourcepoint *dest, char **dest_files, int *dest_size, int *dest_files_size)
{
    u64 dl_offset = get_section_offset(file, ".debug_line");
    
    struct dwarf_debug_line_header_v3_x32 header = { 0 };
    memcpy(&header, file + dl_offset, 15); /* all fixed-size info */
    
    dl_offset += 15;
    
    header.standard_opcode_lengths = malloc(header.opcode_base - 1);
    memcpy(header.standard_opcode_lengths, file + dl_offset, header.opcode_base - 1);
    
    dl_offset += header.opcode_base - 1;
    
    /* "Each entry is a null-terminated string containing a full path name. The last entry
is followed by a single null byte." */
    u8 ndirs = 0;
    u8 nfiles = 0;
    
    u8 *p = file + dl_offset;
    
    while (*p != 0) {
        ++ndirs;
        while (*p != 0) {
            ++p;
        }
        
        ++p;
    }
    
    header.ndirs = ndirs;
    header.include_directories = 0; // malloc(ndirs * sizeof(char *));
    
    ++p;
    
    while (*p != 0) {
        /* null-terminated string */
        if (dest_files) {
            dest_files[nfiles] = (char *) p;
        }
        
        ++nfiles;
        
        while (*p != 0) {
            ++p;
        }
        
        ++p;
        
        u64 offset = 0;
        u32 dummy = 0;
        
        offset += decode_leb128(p, &dummy);
        offset += decode_leb128(p, &dummy);
        offset += decode_leb128(p, &dummy);
        
        p += offset;
    }
    
    if (dest_files_size) {
        *dest_files_size = nfiles;
    }
    
    dl_offset = p - file + 1;
    
    u8 opcode;
    enum dwarf_lnp_opcode opcode_regular = 0;
    enum dwarf_lnp_opcode_extended opcode_extended = 0;
    
    p = file + dl_offset;
    
    struct dwarf_line_number_state state = { 0 };
    
    state.file = 1;
    state.line = 1;
    state.is_stmt = header.default_is_stmt;
    
    u64 pc_count = 0;
    
    do {
        opcode = *p;
        ++p;
        
        u8 nops = 0;
        if (opcode) {
            if (opcode <= header.opcode_base) {
                /* standart opcode */
                opcode_regular = opcode;
                
                switch (opcode_regular) {
                    case DW_LNS_copy: {
                        state.basic_block = 0;
                        state.prologue_end = 0;
                        state.epilogue_begin = 0;
                        break;
                    }
                    
                    case DW_LNS_advance_pc: {
                        u32 operand;
                        p += decode_leb128(p, &operand);
                        operand *= header.minimum_instruction_length;
                        state.pc += operand;
                        
                        if (dest) {
                            struct mi_sourcepoint *sp = dest + pc_count;
                            sp->pc = state.pc;
                            sp->line = state.line;
                            sp->column = state.column;
                            sp->file = state.file;
                        }
                        
                        ++pc_count;
                        
                        break;
                    }
                    
                    case DW_LNS_advance_line: {
                        s32 operand;
                        p += decode_leb128s(p, &operand);
                        state.line += operand;
                        break;
                    }
                    
                    case DW_LNS_set_file: {
                        u32 operand;
                        p += decode_leb128(p, &operand);
                        operand *= header.minimum_instruction_length;
                        state.file = operand;
                        
                        // printf("Switch to file %s in directory %d\n", header.files[state.file - 1].name, header.files[state.file - 1].directory_index);
                        
                        break;
                    }
                    
                    case DW_LNS_set_column: {
                        u32 operand;
                        p += decode_leb128(p, &operand);
                        operand *= header.minimum_instruction_length;
                        state.column = operand;
                        break;
                    }    
                    
                    case DW_LNS_negate_stmt: {
                        state.is_stmt = 1 - state.is_stmt;
                        break;
                    }            
                    
                    case DW_LNS_set_basic_block: {
                        state.basic_block = 1;
                        break;
                    }  
                    
                    case DW_LNS_const_add_pc: {
                        u8 adjusted = 255 - header.opcode_base;
                        s32 address_increment = (adjusted / header.line_range) * header.minimum_instruction_length;
                        state.pc += address_increment;
                        
                        if (dest) {
                            struct mi_sourcepoint *sp = dest + pc_count;
                            sp->pc = state.pc;
                            sp->line = state.line;
                            sp->column = state.column;
                            sp->file = state.file;
                        }
                        
                        ++pc_count;
                        
                        break;
                    } 
                    
                    case DW_LNS_fixed_advance_pc: {
                        u16 operand;
                        memcpy(&operand, p, 2);
                        p += 2;
                        state.pc += operand;
                        
                        if (dest) {
                            struct mi_sourcepoint *sp = dest + pc_count;
                            sp->pc = state.pc;
                            sp->line = state.line;
                            sp->column = state.column;
                            sp->file = state.file;
                        }
                        
                        ++pc_count;
                        
                        break;
                    }
                    
                    case DW_LNS_set_prologue_end: {
                        state.prologue_end = 1;
                        break;
                    }          
                    
                    case DW_LNS_set_epilogue_begin: {
                        state.epilogue_begin = 1;
                        break;
                    }        
                    
                    case DW_LNS_set_isa: {
                        u32 operand;
                        p += decode_leb128(p, &operand);
                        operand *= header.minimum_instruction_length;
                        state.isa = operand;
                        break;
                    }                  
                }
            } else {
                /* special opcode */
                u8 adjusted = opcode - header.opcode_base;
                s32 address_increment = (adjusted / header.line_range) * header.minimum_instruction_length;
                s32 line_increment = header.line_base + (adjusted % header.line_range);
                
                state.pc += address_increment;
                state.line += line_increment;
                state.basic_block = 0;
                state.prologue_end = 0;
                state.epilogue_begin = 0;
                state.discriminator = 0;
                
                if (dest) {
                    struct mi_sourcepoint *sp = dest + pc_count;
                    sp->pc = state.pc;
                    sp->line = state.line;
                    sp->column = state.column;
                    sp->file = state.file;
                }
                
                ++pc_count;
            }
        } else { 
            /* extended opcode */
            u32 instruction_length;
            p += decode_leb128(p, &instruction_length);
            opcode = *p;
            opcode_extended = opcode;
            
            ++p;
            
            switch (opcode_extended) {
                case DW_LNE_end_sequence: {
                    state.end_sequence = 1;
                    //printf("END: %lx -> %d\n", state.pc, state.line);
                    memset(&state, 0, sizeof(state));
                    state.file = 1;
                    state.line = 1;
                    state.is_stmt = header.default_is_stmt;
                    break;
                }
                
                case DW_LNE_set_address: {
                    u64 address;
                    memcpy(&address, p, 8);
                    state.pc = address;
                    p += 8;
                    
                    if (dest) {
                        struct mi_sourcepoint *sp = dest + pc_count;
                        sp->pc = state.pc;
                        sp->line = state.line;
                        sp->column = state.column;
                        sp->file = state.file;
                    }
                    
                    ++pc_count;
                    
                    break;
                }
                
                case DW_LNE_define_file: {
                    struct dwarf_debug_line_file_info f = { 0 };
                    f.name = (char *) p;
                    
                    while (*p != 0) {
                        ++p;
                    }
                    ++p;
                    
                    p += decode_leb128(p, &f.directory_index);
                    p += decode_leb128(p, &f.time_modified);
                    p += decode_leb128(p, &f.file_size);
                    
                    break;
                }
                
                case DW_LNE_set_discriminator: {
                    u32 operand;
                    p += decode_leb128(p, &operand);
                    operand *= header.minimum_instruction_length;
                    state.discriminator = operand;
                    break;
                    break;
                }
            }
        }
        
        //if (state.pc >= pc) {
        //return(last_line);
        //}
        
    } while (opcode_extended != DW_LNE_end_sequence);
    
    if (dest_size) {
        *dest_size = pc_count;
    }
}

static u64
get_executable_base_address(u8 *elf_file, int pid)
{
    char path[256] = { 0 };
    
    snprintf(path, 256, "/proc/%d/maps", pid);
    
    FILE *maps_file = fopen(path, "rb");
    if (!maps_file) {
        DIE("proc map not found\n");
    }
    
    struct elf_header_x64 header = { 0 };
    memcpy(&header, elf_file, sizeof(header));
    
    u64 elf_offset = 0;
    
    for (int i = 0; i < header.header_table_entry_count; ++i) {
        struct elf_header_table_entry_x64 header_entry = { 0 };
        u64 offset = header.header_table_offset + header.header_table_entry_size * i;
        memcpy(&header_entry, elf_file + offset, sizeof(header_entry));
        
        //printf("%#018lx %s\n", header_entry.segment_offset, header_entry.flags & PF_X ? "E" : "");
        
        if (header_entry.flags & PF_X) {
            elf_offset = header_entry.segment_offset;
            break;
        }
    }
    
    if (!elf_offset) {
        return(0);
    }
    
    size_t len;
    char *line = malloc(4096);
    
    while ((len = getline(&line, &len, maps_file)) != -1UL) {
        u64 base;
        u64 end;
        u64 offset;
        
        char *at = line;
        
        base = strtoll(at, &at, 16);
        
        return(base);
        
        end = strtoll(at + 1, &at, 16);
        (void) end;
        
        while (*at < '0' || '9' < *at) ++at;
        
        offset = strtoll(at, &at, 16);
        
        if (offset == elf_offset) {
            return(base);
        }
    }
    
    free(line);
    
    return(0);
}