Multiple compilation units

.gitignore

@@ -0,0 +1,4 @@
+.gdb_history
+build/
+res/001-simple
+res/002-compilation-units

command.c

@@ -15,15 +15,22 @@ command_regs(struct mi_process proc)
 static void
 command_step(struct mi_process proc)
 {
+    int comp_unit;
+    
+    struct mi_registers regs = get_process_registers(proc);
+    struct mi_sourcepoint *sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address, &comp_unit);
+    struct mi_sourcepoint *next_sp;
+    
+    do {
         ptrace(PTRACE_SINGLESTEP, proc.pid, 0, 0);
         waitpid(proc.pid, 0, 0);
+        regs = get_process_registers(proc);
+        long instruction = ptrace(PTRACE_PEEKDATA, proc.pid, regs.rip, NULL);
+        next_sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address, &comp_unit);
+        if (!next_sp) break;
+    } while (next_sp->line == sp->line);
     
-#if 1
-    print_current_instruction(proc);
-    struct mi_registers regs = get_process_registers(proc);
-    struct mi_sourcepoint *sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address);
-    print_sourcepoint(proc, sp);
-#endif
+    print_sourcepoint(proc, comp_unit, next_sp);
 }
 
 static void
@@ -37,9 +44,10 @@ command_start(struct mi_process proc)
 static void
 command_next(struct mi_process proc)
 {
-    struct mi_registers regs = get_process_registers(proc);
+    int comp_unit;
     
-    struct mi_sourcepoint *sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address);
+    struct mi_registers regs = get_process_registers(proc);
+    struct mi_sourcepoint *sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address, &comp_unit);
     struct mi_sourcepoint *next_sp;
     
     do {
@@ -63,23 +71,24 @@ command_next(struct mi_process proc)
             
             regs = get_process_registers(proc);
         }
-        // TODO: skip call, repXXX
+        // TODO: repXXX
         
         //printf("%#lx\n", regs.rip - proc.base_address);
-        next_sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address);
+        next_sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address, &comp_unit);
         
         if (!next_sp) break;
     } while (next_sp->line == sp->line);
     
-    print_sourcepoint(proc, next_sp);
+    print_sourcepoint(proc, comp_unit, next_sp);
 }
 
 static void
 command_list(struct mi_process proc)
 {
+    int comp_unit;
     struct mi_registers regs = get_process_registers(proc);
-    struct mi_sourcepoint *sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address);
-    print_sourcepoint(proc, sp);
+    struct mi_sourcepoint *sp = pc_to_sourcepoint(proc, regs.rip - proc.base_address, &comp_unit);
+    print_sourcepoint(proc, comp_unit, sp);
 }
 
 static void

common.h

@@ -45,23 +45,39 @@ struct mi_sourcepoint {
 
 struct mi_function {
     char *name;
-    u64 offset;
+    int comp_unit;
+    u64 low_pc;
+    u64 high_pc;
 };
 
-struct mi_debuginfo {
+struct mi_sourcefile {
+    char *filename;
+    char *dir;
+    struct mi_buffer file;
+};
+
+struct mi_compunit {
+    u64 low_pc;
+    u64 high_pc;
+    
     struct mi_sourcepoint *sp_table;
     int sp_count;
     
-    struct mi_function functions[64]; // TODO
-    int func_count;
+    int source_file_count;
+    struct mi_sourcefile *source_files;
+    
+    char **source_directories;
+    int source_dirs_count;
     
     char *comp_dir;
 };
 
-struct mi_sourcefile {
-    char *filename;
-    char *dir;
-    struct mi_buffer file;
+struct mi_debuginfo {
+    struct mi_compunit compilation_units[16]; // TODO
+    int cu_count;
+    
+    struct mi_function functions[64]; // TODO
+    int func_count;
 };
 
 struct mi_process {
@@ -74,11 +90,6 @@ struct mi_process {
     u64 main_address;
     
     struct mi_debuginfo debug;
-    
-    int source_file_count;
-    int source_dirs_count;
-    struct mi_sourcefile *source_files;
-    char **source_directories;
 };
 
 struct mi_registers {

dwarf.c

@@ -1,5 +1,5 @@
-static u64
-get_section_offset(u8 *file, char *name)
+static struct elf_section_table_entry_x64
+get_section_entry(u8 *file, char *name)
 {
     struct elf_header_x64 header = { 0 };
     memcpy(&header, file, sizeof(header));
@@ -16,7 +16,8 @@ get_section_offset(u8 *file, char *name)
     memcpy(&shstrtab_header, file + shstrtab_header_offset, sizeof(shstrtab_header));
     
     u64 shstrtab_offset = shstrtab_header.offset_in_file;
-    u64 debug_info_offset = 0;
+    
+    struct elf_section_table_entry_x64 result = { 0 };
     
     for (int i = 0; i < header.section_table_entry_count; ++i) {
         struct elf_section_table_entry_x64 section_entry = { 0 };
@@ -25,12 +26,21 @@ get_section_offset(u8 *file, char *name)
         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;
+            result = section_entry;
+            break;
         }
         
     }
     
-    return(debug_info_offset);
+    return(result);
+}
+
+static u64
+get_section_offset(u8 *file, char *name)
+{
+    struct elf_section_table_entry_x64 entry = get_section_entry(file, name);
+    u64 result = entry.offset_in_file;
+    return(result);
 }
 
 static int
@@ -232,18 +242,10 @@ read_actual_debug_data(u8 *file, u64 string_offset, u32 address_size, u64 base_d
     return(increment);
 }
 
-static void
-parse_debug_info(u8 *file, struct mi_debuginfo *dest)
+static u64
+read_debug_info_for_compilation_unit(u8 *file, struct mi_debuginfo *dest,
+                                     u64 debug_info_offset, u64 debug_abbrev_offset, u64 debug_str_offset)
 {
-    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);
@@ -258,7 +260,8 @@ parse_debug_info(u8 *file, struct mi_debuginfo *dest)
     
     int found_sr = 0;
     
-    struct mi_function *func = dest->functions;
+    struct mi_function *func = dest->functions + dest->func_count;
+    struct mi_compunit *comp_unit = dest->compilation_units + dest->cu_count;
     
     for (;;) {
         data_offset += decode_leb128(file + data_offset, &code);
@@ -294,31 +297,77 @@ parse_debug_info(u8 *file, struct mi_debuginfo *dest)
             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;
+                if (attribute == DW_AT_low_pc) {
+                    comp_unit->low_pc = value;
+                } else if (attribute == DW_AT_high_pc) {
+                    comp_unit->high_pc = value;
+                } else if (attribute == DW_AT_comp_dir) {
+                    comp_unit->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;
+                    func->low_pc = value;
+                } else if (attribute == DW_AT_high_pc) {
+                    func->high_pc = value;
                 }
             }
         } while (attribute != 0 || form != 0);
         
+        if (tag == DW_TAG_compile_unit) {
+            comp_unit->high_pc = comp_unit->low_pc + comp_unit->high_pc;
+            ++comp_unit;
+            ++dest->cu_count;
+        }
+        
         if (tag == DW_TAG_subprogram) {
+            func->high_pc = func->low_pc + func->high_pc;
+            func->comp_unit = dest->cu_count;
             ++func;
             ++dest->func_count;
         }
     }
+    
+    return(di_header.length + 4);
 }
 
 static void
-parse_debug_line(u8 *file, 
-                 struct mi_sourcepoint *dest, struct mi_sourcefile *dest_files, char **dest_directories,
-                 int *dest_size, int *dest_files_size, int *dest_directories_size)
+parse_debug_info(u8 *file, struct mi_debuginfo *dest)
+{
+    struct elf_section_table_entry_x64 debug_info = get_section_entry(file, ".debug_info");
+    printf("Found .debug_info at offset %#lx\n", debug_info.offset_in_file);
+    
+    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);
+    
+    u64 at = debug_info.offset_in_file;
+    u64 read = 0;
+    
+    for (;;) {
+        u64 size = read_debug_info_for_compilation_unit(file, dest, at, debug_abbrev_offset, debug_str_offset);
+        read += size;
+        at += size;
+        if (read >= debug_info.size) {
+            break;
+        }
+    }
+}
+
+static u64
+read_debug_line_for_compilation_unit(u8 *file, u64 dl_offset, struct mi_compunit *unit)
 {
-    u64 dl_offset = get_section_offset(file, ".debug_line");
+    struct mi_sourcepoint *dest = unit->sp_table;
+    int *dest_size = &unit->sp_count;
+    
+    struct mi_sourcefile *dest_files = unit->source_files; 
+    int *dest_files_size = &unit->source_file_count;
+    
+    char **dest_directories = unit->source_directories;
+    int *dest_directories_size = &unit->source_dirs_count;
     
     struct dwarf_debug_line_header_v3_x32 header = { 0 };
     memcpy(&header, file + dl_offset, 15); /* all fixed-size info */
@@ -640,6 +689,48 @@ is followed by a single null byte." */
     if (dest_directories_size) {
         *dest_directories_size = ndirs;
     }
+    
+    return(header.length + 4);
+}
+
+static void
+parse_debug_line(u8 *file, struct mi_debuginfo *debug)
+{
+    struct elf_section_table_entry_x64 debug_line = get_section_entry(file, ".debug_line");
+    
+    u64 at = debug_line.offset_in_file;
+    u64 read = 0;
+    
+    struct mi_compunit *unit = debug->compilation_units;
+    
+    // count
+    for (;;) {
+        u64 size = read_debug_line_for_compilation_unit(file, at, unit);
+        unit->sp_table = calloc(1, unit->sp_count * sizeof(struct mi_sourcepoint));
+        unit->source_directories = calloc(1, unit->source_dirs_count * sizeof(char *));
+        unit->source_files = calloc(1, unit->source_file_count * sizeof(struct mi_sourcefile));
+        read += size;
+        at += size;
+        ++unit;
+        if (read >= debug_line.size) {
+            break;
+        }
+    }
+    
+    unit = debug->compilation_units;
+    at = debug_line.offset_in_file;
+    read = 0;
+    
+    // fill
+    for (;;) {
+        u64 size = read_debug_line_for_compilation_unit(file, at, unit);
+        read += size;
+        at += size;
+        ++unit;
+        if (read >= debug_line.size) {
+            break;
+        }
+    }
 }
 
 static u64

main.c

@@ -31,11 +31,7 @@ main(int argc, char *argv[])
     printf("> ");
     fflush(stdout);
     
-    parse_debug_line(process.elf, 0, 0, 0, &process.debug.sp_count, &process.source_file_count, &process.source_dirs_count);
-    process.debug.sp_table = calloc(1, process.debug.sp_count * sizeof(struct mi_sourcepoint));
-    process.source_directories = calloc(1, process.source_dirs_count * sizeof(char *));
-    process.source_files = calloc(1, process.source_file_count * sizeof(struct mi_sourcefile));
-    parse_debug_line(process.elf, process.debug.sp_table, process.source_files, process.source_directories, 0, 0, 0);
+    parse_debug_line(process.elf, &process.debug);
     
     while ((command_length = getline(&command, &max_command_length, stdin))) {
         if (command_length == 1) {

res/001-simple.c

@@ -1,7 +1,7 @@
 #include <stdio.h>
 #include <unistd.h>
 
-#include "trace_include.c"
+#include "001-simple-inc.c"
 
 int main()
 {   

res/002-compilation-units-impl.c

@@ -0,0 +1,8 @@
+#include "002-compilation-units.h"
+
+int foo(int arg)
+{
+	int sq = arg * arg;
+	int sq2 = sq * 2;
+	return(sq2);
+}

res/002-compilation-units.c

@@ -0,0 +1,9 @@
+#include "002-compilation-units.h"
+
+int
+main(void)
+{
+	int a = 3;
+	int b = foo(a);
+	return(b);
+}

res/002-compilation-units.h

@@ -0,0 +1 @@
+int foo(int arg);

res/Makefile

@@ -0,0 +1,3 @@
+all:
+	gcc -g -o 001-simple 001-simple.c
+	gcc -g -o 002-compilation-units 002-compilation-units.c 002-compilation-units-impl.c

util.c

@@ -98,7 +98,7 @@ get_address_of_subroutine(struct mi_process proc, char *sr)
     for (int i = 0; i < proc.debug.func_count; ++i) {
         struct mi_function *func = proc.debug.functions + i;
         if (0 == strcmp(func->name, sr)) {
-            return(func->offset);
+            return(func->low_pc);
         }
     }
     
@@ -106,14 +106,21 @@ get_address_of_subroutine(struct mi_process proc, char *sr)
 }
 
 static struct mi_sourcepoint *
-pc_to_sourcepoint(struct mi_process proc, u64 pc)
+pc_to_sourcepoint(struct mi_process proc, u64 pc, int *comp_unit)
 {
     // NOTE: find first point BIGGER that pc, return the sourcepoint just before that
     // TODO: binary search
-    for (int i = 0; i < proc.debug.sp_count; ++i) {
-        struct mi_sourcepoint *point = proc.debug.sp_table + i;
+    
+    for (int c = 0; c < proc.debug.cu_count; ++c) {
+        struct mi_compunit unit = proc.debug.compilation_units[c];
+        if (unit.low_pc <= pc && pc < unit.high_pc) {
+            for (int i = 0; i < unit.sp_count; ++i) {
+                struct mi_sourcepoint *point = unit.sp_table + i;
                 if (point->pc > pc) {
-            return(proc.debug.sp_table + i - 1);
+                    *comp_unit = c;
+                    return(unit.sp_table + i - 1);
+                }
+            }
         }
     }
     
@@ -155,27 +162,29 @@ read_file_mmap(char *path)
 }
 
 static void
-print_sourcepoint(struct mi_process proc, struct mi_sourcepoint *sp)
+print_sourcepoint(struct mi_process proc, int comp_unit, struct mi_sourcepoint *sp)
 {
     if (!sp) {
         printf("Unknown location\n");
         return;
     }
     
+    struct mi_compunit unit = proc.debug.compilation_units[comp_unit];
+    
     // NOTE: sourcepoint file indices are 1-based
-    if (proc.source_files[sp->file - 1].file.data == 0) {
-        char *file_filename = proc.source_files[sp->file - 1].filename;
-        char *file_dirname = proc.source_files[sp->file - 1].dir;
-        char *comp_dir = proc.debug.comp_dir;
+    if (unit.source_files[sp->file - 1].file.data == 0) {
+        char *file_filename = unit.source_files[sp->file - 1].filename;
+        char *file_dirname = unit.source_files[sp->file - 1].dir;
+        char *comp_dir = unit.comp_dir;
         
         char full_path[512] = { 0 };
         snprintf(full_path, 511, "%s/%s/%s", comp_dir, file_dirname, file_filename);
         
         struct mi_buffer file = read_file_mmap(full_path);
-        proc.source_files[sp->file - 1].file = file;
+        unit.source_files[sp->file - 1].file = file;
     }
     
-    struct mi_buffer file = proc.source_files[sp->file - 1].file;
+    struct mi_buffer file = unit.source_files[sp->file - 1].file;
     
     char *source = (char *) file.data;
     u64 size = file.size;