src/runtime/coreparse.c

   1 /*
   2  * A saved SBCL system is a .core file; the code here helps us accept
   3  * such a file as input.
   4  */
   5
   6 /*
   7  * This software is part of the SBCL system. See the README file for
   8  * more information.
   9  *
  10  * This software is derived from the CMU CL system, which was
  11  * written at Carnegie Mellon University and released into the
  12  * public domain. The software is in the public domain and is
  13  * provided with absolutely no warranty. See the COPYING and CREDITS
  14  * files for more information.
  15  */
  16
  17 #include <stdio.h>
  18 #include <stdlib.h>
  19 #include <string.h>
  20 #include <sys/file.h>
  21 #include <sys/types.h>
  22 #include <sys/stat.h>
  23 #include <fcntl.h>
  24 #include <unistd.h>
  25
  26 #ifdef irix
  27 #include <fcntl.h>
  28 #endif
  29
  30 #include "sbcl.h"
  31 #include "os.h"
  32 #include "runtime.h"
  33 #include "globals.h"
  34 #include "core.h"
  35 #include "arch.h"
  36 #include "interr.h"
  37 #include "thread.h"
  38
  39 #include "validate.h"
  40 #include "gc-internal.h"
  41
  42 unsigned char build_id[] =
  43 #include "../../output/build-id.tmp"
  44 ;
  45
  46 int
  47 open_binary(char *filename, int mode)
  48 {
  49 #ifdef LISP_FEATURE_WIN32
  50     mode |= O_BINARY;
  51 #endif
  52
  53     return open(filename, mode);
  54 }
  55
  56 /* Search 'filename' for an embedded core.  An SBCL core has, at the
  57  * end of the file, a trailer containing the size of the core (an
  58  * os_vm_offset_t) and a final signature word (the lispobj
  59  * CORE_MAGIC).  If this trailer is found at the end of the file, the
  60  * start of the core can be determined from the core size.
  61  *
  62  * If an embedded core is present, this returns the offset into the
  63  * file to load the core from, or -1 if no core is present. */
  64 os_vm_offset_t
  65 search_for_embedded_core(char *filename)
  66 {
  67     lispobj header;
  68     os_vm_offset_t lispobj_size = sizeof(lispobj);
  69     os_vm_offset_t trailer_size = lispobj_size + sizeof(os_vm_offset_t);
  70     os_vm_offset_t core_size, pos;
  71     int fd = -1;
  72
  73     if ((fd = open_binary(filename, O_RDONLY)) < 0)
  74         goto lose;
  75     if (lseek(fd, -lispobj_size, SEEK_END) < 0)
  76         goto lose;
  77     if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
  78         goto lose;
  79
  80     if (header == CORE_MAGIC) {
  81         if (lseek(fd, -trailer_size, SEEK_END) < 0)
  82             goto lose;
  83         if (read(fd, &core_size, sizeof(os_vm_offset_t)) < 0)
  84             goto lose;
  85
  86         if (lseek(fd, -(core_size + trailer_size), SEEK_END) < 0)
  87             goto lose;
  88         pos = lseek(fd, 0, SEEK_CUR);
  89
  90         if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
  91             goto lose;
  92
  93         if (header != CORE_MAGIC)
  94             goto lose;
  95
  96         close(fd);
  97         return pos;
  98     }
  99
 100 lose:
 101     if (fd != -1)
 102         close(fd);
 103
 104     return -1;
 105 }
 106
 107 static void
 108 process_directory(int fd, u32 *ptr, int count, os_vm_offset_t file_offset)
 109 {
 110     struct ndir_entry *entry;
 111
 112     FSHOW((stderr, "/process_directory(..), count=%d\n", count));
 113
 114     for (entry = (struct ndir_entry *) ptr; --count>= 0; ++entry) {
 115
 116         long id = entry->identifier;
 117         long offset = os_vm_page_size * (1 + entry->data_page);
 118         os_vm_address_t addr =
 119             (os_vm_address_t) (os_vm_page_size * entry->address);
 120         lispobj *free_pointer = (lispobj *) addr + entry->nwords;
 121         long len = os_vm_page_size * entry->page_count;
 122
 123         if (len != 0) {
 124             os_vm_address_t real_addr;
 125             FSHOW((stderr, "/mapping %ld(0x%lx) bytes at 0x%lx\n",
 126                    (long)len, (long)len, (unsigned long)addr));
 127             real_addr = os_map(fd, offset + file_offset, addr, len);
 128             if (real_addr != addr) {
 129                 lose("file mapped in wrong place! "
 130                      "(0x%08x != 0x%08lx)\n",
 131                      real_addr,
 132                      addr);
 133             }
 134         }
 135
 136         FSHOW((stderr, "/space id = %ld, free pointer = 0x%lx\n",
 137                id, (unsigned long)free_pointer));
 138
 139         switch (id) {
 140         case DYNAMIC_CORE_SPACE_ID:
 141 #ifdef LISP_FEATURE_GENCGC
 142             if (addr != (os_vm_address_t)DYNAMIC_SPACE_START) {
 143                 fprintf(stderr, "in core: 0x%lx; in runtime: 0x%lx \n",
 144                         (long)addr, (long)DYNAMIC_SPACE_START);
 145                 lose("core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 146             }
 147 #else
 148             if ((addr != (os_vm_address_t)DYNAMIC_0_SPACE_START) &&
 149                 (addr != (os_vm_address_t)DYNAMIC_1_SPACE_START)) {
 150                 fprintf(stderr, "in core: 0x%lx; in runtime: 0x%lx or 0x%lx\n",
 151                         (long)addr,
 152                         (long)DYNAMIC_0_SPACE_START,
 153                         (long)DYNAMIC_1_SPACE_START);
 154                 lose("warning: core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 155             }
 156 #endif
 157 #if defined(ALLOCATION_POINTER)
 158             SetSymbolValue(ALLOCATION_POINTER, (lispobj)free_pointer,0);
 159 #else
 160             dynamic_space_free_pointer = free_pointer;
 161 #endif
 162             /* For stop-and-copy GC, this will be whatever the GC was
 163              * using at the time. With GENCGC, this will always be
 164              * space 0. (We checked above that for GENCGC,
 165              * addr==DYNAMIC_SPACE_START.) */
 166             current_dynamic_space = (lispobj *)addr;
 167             break;
 168         case STATIC_CORE_SPACE_ID:
 169             if (addr != (os_vm_address_t)STATIC_SPACE_START) {
 170                 fprintf(stderr, "in core: 0x%lx - in runtime: 0x%lx\n",
 171                         (long)addr, (long)STATIC_SPACE_START);
 172                 lose("core/runtime address mismatch: STATIC_SPACE_START\n");
 173             }
 174             break;
 175         case READ_ONLY_CORE_SPACE_ID:
 176             if (addr != (os_vm_address_t)READ_ONLY_SPACE_START) {
 177                 fprintf(stderr, "in core: 0x%lx - in runtime: 0x%lx\n",
 178                         (long)addr, (long)READ_ONLY_SPACE_START);
 179                 lose("core/runtime address mismatch: READ_ONLY_SPACE_START\n");
 180             }
 181             break;
 182         default:
 183             lose("unknown space ID %ld addr 0x%lx\n", id, (long)addr);
 184         }
 185     }
 186 }
 187
 188 lispobj
 189 load_core_file(char *file, os_vm_offset_t file_offset)
 190 {
 191     lispobj *header, val, len, *ptr, remaining_len;
 192     int fd = open_binary(file, O_RDONLY), count;
 193
 194     lispobj initial_function = NIL;
 195     FSHOW((stderr, "/entering load_core_file(%s)\n", file));
 196     if (fd < 0) {
 197         fprintf(stderr, "could not open file \"%s\"\n", file);
 198         perror("open");
 199         exit(1);
 200     }
 201
 202     lseek(fd, file_offset, SEEK_SET);
 203     header = calloc(os_vm_page_size / sizeof(u32), sizeof(u32));
 204
 205     count = read(fd, header, os_vm_page_size);
 206     if (count < os_vm_page_size) {
 207         lose("premature end of core file\n");
 208     }
 209     SHOW("successfully read first page of core");
 210
 211     ptr = header;
 212     val = *ptr++;
 213
 214     if (val != CORE_MAGIC) {
 215         lose("invalid magic number in core: 0x%lx should have been 0x%x.\n",
 216              val,
 217              CORE_MAGIC);
 218     }
 219     SHOW("found CORE_MAGIC");
 220
 221     while (val != END_CORE_ENTRY_TYPE_CODE) {
 222         val = *ptr++;
 223         len = *ptr++;
 224         remaining_len = len - 2; /* (-2 to cancel the two ++ operations) */
 225         FSHOW((stderr, "/val=0x%ld, remaining_len=0x%ld\n",
 226                (long)val, (long)remaining_len));
 227
 228         switch (val) {
 229
 230         case END_CORE_ENTRY_TYPE_CODE:
 231             SHOW("END_CORE_ENTRY_TYPE_CODE case");
 232             break;
 233
 234         case VERSION_CORE_ENTRY_TYPE_CODE:
 235             SHOW("VERSION_CORE_ENTRY_TYPE_CODE case");
 236             if (*ptr != SBCL_CORE_VERSION_INTEGER) {
 237                 lose("core file version (%d) != runtime library version (%d)\n",
 238                      *ptr,
 239                      SBCL_CORE_VERSION_INTEGER);
 240             }
 241             break;
 242
 243         case BUILD_ID_CORE_ENTRY_TYPE_CODE:
 244             SHOW("BUILD_ID_CORE_ENTRY_TYPE_CODE case");
 245             {
 246                 int i;
 247
 248                 FSHOW((stderr, "build_id[]=\"%s\"\n", build_id));
 249                 FSHOW((stderr, "remaining_len = %d\n", remaining_len));
 250                 if (remaining_len != strlen((const char *)build_id))
 251                     goto losing_build_id;
 252                 for (i = 0; i < remaining_len; ++i) {
 253                     FSHOW((stderr, "ptr[%d] = char = %d, expected=%d\n",
 254                            i, ptr[i], build_id[i]));
 255                     if (ptr[i] != build_id[i])
 256                         goto losing_build_id;
 257                 }
 258                 break;
 259             losing_build_id:
 260                 /* .core files are not binary-compatible between
 261                  * builds because we can't easily detect whether the
 262                  * sources were patched between the time the
 263                  * dumping-the-.core runtime was built and the time
 264                  * that the loading-the-.core runtime was built.
 265                  *
 266                  * (We could easily detect whether version.lisp-expr
 267                  * was changed, but people experimenting with patches
 268                  * don't necessarily update version.lisp-expr.) */
 269
 270                 lose("can't load .core for different runtime, sorry\n");
 271             }
 272
 273         case NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE:
 274             SHOW("NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE case");
 275             process_directory(fd,
 276                               ptr,
 277 #ifndef LISP_FEATURE_ALPHA
 278                               remaining_len / (sizeof(struct ndir_entry) /
 279                                                sizeof(long)),
 280 #else
 281                               remaining_len / (sizeof(struct ndir_entry) /
 282                                                sizeof(u32)),
 283 #endif
 284                               file_offset);
 285             break;
 286
 287         case INITIAL_FUN_CORE_ENTRY_TYPE_CODE:
 288             SHOW("INITIAL_FUN_CORE_ENTRY_TYPE_CODE case");
 289             initial_function = (lispobj)*ptr;
 290             break;
 291
 292 #ifdef LISP_FEATURE_GENCGC
 293         case PAGE_TABLE_CORE_ENTRY_TYPE_CODE:
 294         {
 295             size_t size = *ptr;
 296             size_t fdoffset = (*(ptr+1) + 1) * (os_vm_page_size);
 297             size_t offset = 0;
 298             long bytes_read;
 299             long data[4096];
 300             lseek(fd, fdoffset + file_offset, SEEK_SET);
 301             while ((bytes_read = read(fd, data, (size < 4096 ? size : 4096 )))
 302                     > 0)
 303             {
 304                 int i = 0;
 305                 size -= bytes_read;
 306                 while (bytes_read) {
 307                     bytes_read -= sizeof(long);
 308                     page_table[offset++].first_object_offset = data[i++];
 309                 }
 310             }
 311
 312             gencgc_partial_pickup = 1;
 313             break;
 314         }
 315 #endif
 316         default:
 317             lose("unknown core file entry: %ld\n", (long)val);
 318         }
 319
 320         ptr += remaining_len;
 321         FSHOW((stderr, "/new ptr=%lx\n", (unsigned long)ptr));
 322     }
 323     SHOW("about to free(header)");
 324     free(header);
 325     SHOW("returning from load_core_file(..)");
 326     return initial_function;
 327 }
 328