From 51e499b3a0e4cd847172357b84d9a01479ddd558 Mon Sep 17 00:00:00 2001
From: aolo2 <aolo2@yandex.ru>
Date: Sun, 11 Jul 2021 21:00:42 +0300
Subject: [PATCH] Extract debug info (compilation dir and function list) ahead
 of time

---
 common.h |  18 +++-
 dwarf.c  | 290 +++++++++++++++++++++++++++----------------------------
 main.c   |  10 +-
 util.c   |  28 +++++-
 4 files changed, 185 insertions(+), 161 deletions(-)

diff --git a/common.h b/common.h
index 93206ff..1ebf049 100644
--- a/common.h
+++ b/common.h
@@ -43,6 +43,21 @@ struct mi_sourcepoint {
     int file;
 };
 
+struct mi_function {
+    char *name;
+    u64 offset;
+};
+
+struct mi_debuginfo {
+    struct mi_sourcepoint *sp_table;
+    int sp_count;
+    
+    struct mi_function functions[64]; // TODO
+    int func_count;
+    
+    char *comp_dir;
+};
+
 struct mi_process {
     pid_t pid;
     
@@ -52,8 +67,7 @@ struct mi_process {
     u64 base_address;
     u64 main_address;
     
-    struct mi_sourcepoint *sp_table;
-    int sp_count;
+    struct mi_debuginfo debug;
     
     char **source_file_names;
     struct mi_buffer *source_files;
diff --git a/dwarf.c b/dwarf.c
index b3034d3..e5a0cff 100644
--- a/dwarf.c
+++ b/dwarf.c
@@ -139,19 +139,131 @@ abbrev_entry_offset(u8 *file, u64 abbrev_offset, u32 requested_code)
     return(0);
 }
 
-static u64
-find_subroutine_offset(u8 *file, u64 header_size, u8 address_size, 
-                       u64 string_offset, u64 abbrev_offset, u64 data_offset,
-                       char *subroutine)
+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 original_data_offset = data_offset;
+    u64 base_data_offset = data_offset - header_size;
     
     int found_sr = 0;
     
-    do {
+    struct mi_function *func = dest->functions;
+    
+    for (;;) {
         data_offset += decode_leb128(file + data_offset, &code);
         
         if (code == 0) {
@@ -180,127 +292,32 @@ find_subroutine_offset(u8 *file, u64 header_size, u8 address_size,
             schema_offset += decode_leb128(file + schema_offset, &attribute);
             schema_offset += decode_leb128(file + schema_offset, &form);
             
-            if (attribute) {
-                //printf("\t%s ", attribute_to_str(attribute));
-            }
+            u64 value;
             
-            switch (form) {
-                case DW_FORM_sec_offset:
-                case DW_FORM_strp: {
-                    u32 data;
-                    memcpy(&data, file + data_offset, 4);
-                    data_offset += 4; // 8 bytes for x64 DWARF!
-                    
-                    if (form == DW_FORM_strp) {
-                        char *str = (char *) file + string_offset + data;
-                        //printf("(indirect string, offset: %#x): %s\n", data, str);
-                        if (tag == DW_TAG_subprogram) {
-                            if (strcmp(str, subroutine) == 0) {
-                                found_sr = 1;
-                            }
-                        }
-                    } else {
-                        //printf("%#x\n", data);
-                    }
-                    
-                    break;
-                }
-                
-                case DW_FORM_addr: {
-                    u64 data = 0;
-                    memcpy(&data, file + data_offset, address_size);
-                    data_offset += address_size;
-                    //printf("%#lx\n", data);
-                    
-                    if (tag == DW_TAG_subprogram && found_sr == 1 && attribute == DW_AT_low_pc) {
-                        return(data);
-                    }
-                    
-                    break;
-                };
-                
-                case DW_FORM_string: {
-                    char *data = (char *) file + data_offset;
-                    data_offset += strlen(data) + 1;
-                    //printf("%s\n", data);
-                    break;
-                }
-                
-                case DW_FORM_flag_present: {
-                    int data = 1;
-                    //printf("Flag = 1\n");
-                    break;
-                }
-                
-                case DW_FORM_ref4: {
-                    u32 data;
-                    memcpy(&data, file + data_offset, 4);
-                    data_offset += 4;
-                    //printf("%#x\n", data);
-                    u32 referenced_data = file[original_data_offset - header_size + data];
-                    break;
+            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;
                 }
-                
-                case DW_FORM_exprloc: {
-                    u32 length;
-                    data_offset += decode_leb128(file + data_offset, &length);
-                    //printf("%d byte block:", length);
-                    
-                    for (u32 i = 0; i < length; ++i) {
-                        //printf(" %x", file[data_offset + i]);
-                    }
-                    //printf("\n");
-                    
-                    data_offset += length;
-                    
-                    break;
-                }
-                
-                case DW_FORM_data1: {
-                    u8 data = file[data_offset];
-                    data_offset += 1;
-                    //printf("%#x\n", data);
-                    break;
-                };
-                
-                case DW_FORM_data2: {
-                    u16 data;
-                    memcpy(&data, file + data_offset, 2);
-                    data_offset += 2;
-                    //printf("%#x\n", data);
-                    break;
-                };
-                
-                case DW_FORM_data4: {
-                    u32 data;
-                    memcpy(&data, file + data_offset, 4);
-                    data_offset += 4;
-                    //printf("%#x\n", data);
-                    break;
-                };
-                
-                case DW_FORM_data8: {
-                    u64 data;
-                    memcpy(&data, file + data_offset, 8);
-                    data_offset += 8;
-                    //printf("%#lx\n", data);
-                    break;
-                };
-                
-                default: {
-                    if (form) {
-                        printf("unknown attribute form %d\n", form);
-                    }
+            } 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);
-    } while (1);
-    
-    return(0);
+        
+        if (tag == DW_TAG_subprogram) {
+            ++func;
+            ++dest->func_count;
+        }
+    }
 }
 
 static void
-construct_line_number_table(u8 *file, struct mi_sourcepoint *dest, char **dest_files, int *dest_size, int *dest_files_size)
+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");
     
@@ -326,15 +343,15 @@ is followed by a single null byte." */
         while (*p != 0) {
             ++p;
         }
+        
+        ++p;
     }
     
     header.ndirs = ndirs;
     header.include_directories = 0; // malloc(ndirs * sizeof(char *));
     
-    dl_offset += (p - (file + dl_offset)) + 1;
-    
+    ++p;
     
-    p = file + dl_offset;
     while (*p != 0) {
         /* null-terminated string */
         if (dest_files) {
@@ -658,7 +675,7 @@ get_executable_base_address(u8 *elf_file, int pid)
         end = strtoll(at + 1, &at, 16);
         (void) end;
         
-        while (*at < '0' || *at > '9') ++at;
+        while (*at < '0' || '9' < *at) ++at;
         
         offset = strtoll(at, &at, 16);
         
@@ -671,30 +688,3 @@ get_executable_base_address(u8 *elf_file, int pid)
     
     return(0);
 }
-
-static u64
-get_address_of_subroutine(u8 *file, char *sr)
-{
-    u64 debug_info_offset = get_section_offset(file, ".debug_info");
-    printf("Found .debug_info at offset %#lx\n", debug_info_offset);
-    
-    u64 debug_line_offset = get_section_offset(file, ".debug_line");
-    printf("Found .debug_line at offset %#lx\n", debug_line_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 };
-    memcpy(&di_header, file + debug_info_offset, sizeof(di_header));
-    
-    u64 abbrev_offset = debug_abbrev_offset + di_header.debug_abbrev_offset;
-    u64 data_offset = debug_info_offset + sizeof(di_header);
-    
-    u64 result = find_subroutine_offset(file, sizeof(di_header), di_header.address_size,
-                                        debug_str_offset, abbrev_offset, data_offset, sr);
-    
-    return(result);
-}
\ No newline at end of file
diff --git a/main.c b/main.c
index 9b84e7a..6d1985b 100644
--- a/main.c
+++ b/main.c
@@ -20,8 +20,10 @@ main(int argc, char *argv[])
     char *command = malloc(max_command_length + 1);
     char *last_command = malloc(max_command_length + 1);
     
+    parse_debug_info(process.elf, &process.debug);
+    
     process.base_address = 0x555555554000UL; // get_executable_base_address(file, proc.pid);
-    process.main_address = get_address_of_subroutine(process.elf, "main");
+    process.main_address = get_address_of_subroutine(process, "main");
     
     printf("Base address: %#lx\n", process.base_address);
     printf("Main address: %#lx\n", process.main_address);
@@ -29,11 +31,11 @@ main(int argc, char *argv[])
     printf("> ");
     fflush(stdout);
     
-    construct_line_number_table(process.elf, 0, 0, &process.sp_count, &process.source_file_count);
-    process.sp_table = malloc(process.sp_count * sizeof(struct mi_sourcepoint));
+    parse_debug_line(process.elf, 0, 0, &process.debug.sp_count, &process.source_file_count);
+    process.debug.sp_table = malloc(process.debug.sp_count * sizeof(struct mi_sourcepoint));
     process.source_file_names = malloc(process.source_file_count * sizeof(char *));
     process.source_files = malloc(process.source_file_count * sizeof(char *));
-    construct_line_number_table(process.elf, process.sp_table, process.source_file_names, 0, 0);
+    parse_debug_line(process.elf, process.debug.sp_table, process.source_file_names, 0, 0);
     
     while ((command_length = getline(&command, &max_command_length, stdin))) {
         if (command_length == 1) {
diff --git a/util.c b/util.c
index 762ec65..a2e25bd 100644
--- a/util.c
+++ b/util.c
@@ -1,12 +1,25 @@
+static u64
+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(0);
+}
+
 static struct mi_sourcepoint *
 pc_to_sourcepoint(struct mi_process proc, u64 pc)
 {
     // NOTE: find first point BIGGER that pc, return the sourcepoint just before that
     // TODO: binary search
-    for (int i = 0; i < proc.sp_count; ++i) {
-        struct mi_sourcepoint *point = proc.sp_table + i;
+    for (int i = 0; i < proc.debug.sp_count; ++i) {
+        struct mi_sourcepoint *point = proc.debug.sp_table + i;
         if (point->pc > pc) {
-            return(proc.sp_table + i - 1);
+            return(proc.debug.sp_table + i - 1);
         }
     }
     
@@ -47,8 +60,13 @@ print_sourcepoint(struct mi_process proc, struct mi_sourcepoint *sp)
 {
     // NOTE: sourcepoint file indices are 1-based
     if (proc.source_files[sp->file - 1].data == 0) {
-        char *path = proc.source_file_names[sp->file - 1];
-        struct mi_buffer file = read_file_mmap(path);
+        char *filename = proc.source_file_names[sp->file - 1];
+        char *dir = proc.debug.comp_dir;
+        
+        char full_path[512] = { 0 };
+        snprintf(full_path, 511, "%s/%s", dir, filename);
+        
+        struct mi_buffer file = read_file_mmap(full_path);
         proc.source_files[sp->file - 1] = file;
     }