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 #include "sbcl.h"
  27
  28 #ifndef LISP_FEATURE_WIN32
  29 #ifdef LISP_FEATURE_LINUX
  30 /* For madvise */
  31 # define _BSD_SOURCE
  32 #endif
  33 #include <sys/mman.h>
  34 #endif
  35
  36 #include "os.h"
  37 #include "runtime.h"
  38 #include "globals.h"
  39 #include "core.h"
  40 #include "arch.h"
  41 #include "interr.h"
  42 #include "thread.h"
  43
  44 #include "validate.h"
  45 #include "gc-internal.h"
  46
  47 #include <errno.h>
  48
  49 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
  50 # include <zlib.h>
  51 #endif
  52
  53 unsigned char build_id[] =
  54 #include "../../output/build-id.tmp"
  55 ;
  56
  57 int
  58 open_binary(char *filename, int mode)
  59 {
  60 #ifdef LISP_FEATURE_WIN32
  61     mode |= O_BINARY;
  62 #endif
  63
  64     return open(filename, mode);
  65 }
  66
  67
  68 static struct runtime_options *
  69 read_runtime_options(int fd)
  70 {
  71     size_t optarray[RUNTIME_OPTIONS_WORDS];
  72     struct runtime_options *options = NULL;
  73
  74     if (read(fd, optarray, RUNTIME_OPTIONS_WORDS * sizeof(size_t)) !=
  75         RUNTIME_OPTIONS_WORDS * sizeof(size_t)) {
  76         return NULL;
  77     }
  78
  79     if ((RUNTIME_OPTIONS_MAGIC != optarray[0]) || (0 == optarray[1])) {
  80         return NULL;
  81     }
  82
  83     options = successful_malloc(sizeof(struct runtime_options));
  84
  85     options->dynamic_space_size = optarray[2];
  86     options->thread_control_stack_size = optarray[3];
  87
  88     return options;
  89 }
  90
  91 void
  92 maybe_initialize_runtime_options(int fd)
  93 {
  94     struct runtime_options *new_runtime_options;
  95     off_t end_offset = sizeof(lispobj) +
  96         sizeof(os_vm_offset_t) +
  97         (RUNTIME_OPTIONS_WORDS * sizeof(size_t));
  98
  99     lseek(fd, -end_offset, SEEK_END);
 100
 101     if ((new_runtime_options = read_runtime_options(fd))) {
 102         runtime_options = new_runtime_options;
 103     }
 104 }
 105
 106 /* Search 'filename' for an embedded core.  An SBCL core has, at the
 107  * end of the file, a trailer containing optional saved runtime
 108  * options, the start of the core (an os_vm_offset_t), and a final
 109  * signature word (the lispobj CORE_MAGIC).  If this trailer is found
 110  * at the end of the file, the start of the core can be determined
 111  * from the core size.
 112  *
 113  * If an embedded core is present, this returns the offset into the
 114  * file to load the core from, or -1 if no core is present. */
 115 os_vm_offset_t
 116 search_for_embedded_core(char *filename)
 117 {
 118     lispobj header;
 119     os_vm_offset_t lispobj_size = sizeof(lispobj);
 120     os_vm_offset_t trailer_size = lispobj_size + sizeof(os_vm_offset_t);
 121     os_vm_offset_t core_start, pos;
 122     int fd = -1;
 123
 124     if ((fd = open_binary(filename, O_RDONLY)) < 0)
 125         goto lose;
 126
 127     if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
 128         goto lose;
 129     if (header == CORE_MAGIC) {
 130         /* This file is a real core, not an embedded core.  Return 0 to
 131          * indicate where the core starts, and do not look for runtime
 132          * options in this case. */
 133         return 0;
 134     }
 135
 136     if (lseek(fd, -lispobj_size, SEEK_END) < 0)
 137         goto lose;
 138     if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
 139         goto lose;
 140
 141     if (header == CORE_MAGIC) {
 142         if (lseek(fd, -trailer_size, SEEK_END) < 0)
 143             goto lose;
 144         if (read(fd, &core_start, sizeof(os_vm_offset_t)) < 0)
 145             goto lose;
 146
 147         if (lseek(fd, core_start, SEEK_SET) < 0)
 148             goto lose;
 149         pos = lseek(fd, 0, SEEK_CUR);
 150
 151         if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
 152             goto lose;
 153
 154         if (header != CORE_MAGIC)
 155             goto lose;
 156
 157         maybe_initialize_runtime_options(fd);
 158
 159         close(fd);
 160         return pos;
 161     }
 162
 163 lose:
 164     if (fd != -1)
 165         close(fd);
 166
 167     return -1;
 168 }
 169
 170 /* If more platforms doesn't support overlapping mmap rename this
 171  * def to something like ifdef nommapoverlap */
 172 /* currently hpux only */
 173 #ifdef LISP_FEATURE_HPUX
 174 os_vm_address_t copy_core_bytes(int fd, os_vm_offset_t offset,
 175                                 os_vm_address_t addr, int len)
 176 {
 177   unsigned char buf[4096];
 178   int c,x;
 179   int old_fd = lseek(fd, 0, SEEK_CUR);
 180
 181   if(len & (4096-1)){
 182     fprintf(stderr, "cant copy a slice of core because slice-length is not of page size(4096)\n");
 183     exit(-1);
 184   }
 185   if(old_fd < 0){
 186     fprintf(stderr, "cant perform lseek() on corefile\n");
 187   }
 188   lseek(fd, offset, SEEK_SET);
 189   if(fd < 0){
 190     fprintf(stderr, "cant perform lseek(%u,%lu,SEEK_SET) on corefile\n", fd, offset);
 191   }
 192   for(x = 0; x < len; x += 4096){
 193     c = read(fd, buf, 4096);
 194     if(c != 4096){
 195       fprintf(stderr, "cant read memory area from corefile at position %lu, got %d\n", offset + x, c);
 196       exit(-1);
 197     }
 198     memcpy(addr+x, buf, 4096);
 199   }
 200   os_flush_icache(addr, len);
 201   return addr;
 202 }
 203 #endif
 204
 205 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
 206 # define ZLIB_BUFFER_SIZE (1u<<16)
 207 os_vm_address_t inflate_core_bytes(int fd, os_vm_offset_t offset,
 208                                    os_vm_address_t addr, int len)
 209 {
 210     z_stream stream;
 211     unsigned char buf[ZLIB_BUFFER_SIZE];
 212     int ret;
 213
 214     if (-1 == lseek(fd, offset, SEEK_SET)) {
 215         lose("Unable to lseek() on corefile\n");
 216     }
 217
 218     stream.zalloc = NULL;
 219     stream.zfree = NULL;
 220     stream.opaque = NULL;
 221     stream.avail_in = 0;
 222     stream.next_in = buf;
 223
 224     ret = inflateInit(&stream);
 225     if (ret != Z_OK)
 226         lose("zlib error %i\n", ret);
 227
 228     stream.next_out  = (void*)addr;
 229     stream.avail_out = len;
 230     do {
 231         ssize_t count = read(fd, buf, sizeof(buf));
 232         if (count < 0)
 233             lose("unable to read core file (errno = %i)\n", errno);
 234         stream.next_in = buf;
 235         stream.avail_in = count;
 236         if (count == 0) break;
 237         ret = inflate(&stream, Z_NO_FLUSH);
 238         switch (ret) {
 239         case Z_STREAM_END:
 240             break;
 241         case Z_OK:
 242             if (stream.avail_out == 0)
 243                 lose("Runaway gzipped core directory... aborting\n");
 244             if (stream.avail_in > 0)
 245                 lose("zlib inflate returned without fully"
 246                      "using up input buffer... aborting\n");
 247             break;
 248         default:
 249             lose("zlib inflate error: %i\n", ret);
 250             break;
 251         }
 252     } while (ret != Z_STREAM_END);
 253
 254     if (stream.avail_out > 0) {
 255         if (stream.avail_out >= os_vm_page_size)
 256             fprintf(stderr, "Warning: gzipped core directory significantly"
 257                     "shorter than expected (%lu bytes)", (unsigned long)stream.avail_out);
 258         /* Is this needed? */
 259         memset(stream.next_out, 0, stream.avail_out);
 260     }
 261
 262     inflateEnd(&stream);
 263     return addr;
 264 }
 265 # undef ZLIB_BUFFER_SIZE
 266 #endif
 267
 268 int merge_core_pages = -1;
 269
 270 static void
 271 process_directory(int fd, lispobj *ptr, int count, os_vm_offset_t file_offset)
 272 {
 273     struct ndir_entry *entry;
 274     int compressed;
 275
 276     FSHOW((stderr, "/process_directory(..), count=%d\n", count));
 277
 278     for (entry = (struct ndir_entry *) ptr; --count>= 0; ++entry) {
 279         compressed = 0;
 280         long id = entry->identifier;
 281         if (id <= (MAX_CORE_SPACE_ID | DEFLATED_CORE_SPACE_ID_FLAG)) {
 282             if (id & DEFLATED_CORE_SPACE_ID_FLAG)
 283                 compressed = 1;
 284             id &= ~(DEFLATED_CORE_SPACE_ID_FLAG);
 285         }
 286         long offset = os_vm_page_size * (1 + entry->data_page);
 287         os_vm_address_t addr =
 288             (os_vm_address_t) (os_vm_page_size * entry->address);
 289         lispobj *free_pointer = (lispobj *) addr + entry->nwords;
 290         unsigned long len = os_vm_page_size * entry->page_count;
 291         if (len != 0) {
 292             os_vm_address_t real_addr;
 293             FSHOW((stderr, "/mapping %ld(0x%lx) bytes at 0x%lx\n",
 294                    (long)len, (long)len, (unsigned long)addr));
 295             if (compressed) {
 296 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
 297                 real_addr = inflate_core_bytes(fd, offset + file_offset, addr, len);
 298 #else
 299                 lose("This runtime was not built with zlib-compressed core support... aborting\n");
 300 #endif
 301             } else {
 302 #ifdef LISP_FEATURE_HPUX
 303                 real_addr = copy_core_bytes(fd, offset + file_offset, addr, len);
 304 #else
 305                 real_addr = os_map(fd, offset + file_offset, addr, len);
 306 #endif
 307             }
 308             if (real_addr != addr) {
 309                 lose("file mapped in wrong place! "
 310                      "(0x%08x != 0x%08lx)\n",
 311                      real_addr,
 312                      addr);
 313             }
 314         }
 315
 316 #ifdef MADV_MERGEABLE
 317         if ((merge_core_pages == 1)
 318             || ((merge_core_pages == -1) && compressed)) {
 319                 madvise(addr, len, MADV_MERGEABLE);
 320         }
 321 #endif
 322
 323         FSHOW((stderr, "/space id = %ld, free pointer = 0x%lx\n",
 324                id, (unsigned long)free_pointer));
 325
 326         switch (id) {
 327         case DYNAMIC_CORE_SPACE_ID:
 328             if (len > dynamic_space_size) {
 329                 fprintf(stderr,
 330                         "dynamic space too small for core: %ldKiB required, %ldKiB available.\n",
 331                         len >> 10,
 332                         (long)dynamic_space_size >> 10);
 333                 exit(1);
 334             }
 335 #ifdef LISP_FEATURE_GENCGC
 336             if (addr != (os_vm_address_t)DYNAMIC_SPACE_START) {
 337                 fprintf(stderr, "in core: 0x%lx; in runtime: 0x%lx \n",
 338                         (long)addr, (long)DYNAMIC_SPACE_START);
 339                 lose("core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 340             }
 341 #else
 342             if ((addr != (os_vm_address_t)DYNAMIC_0_SPACE_START) &&
 343                 (addr != (os_vm_address_t)DYNAMIC_1_SPACE_START)) {
 344                 fprintf(stderr, "in core: 0x%lx; in runtime: 0x%lx or 0x%lx\n",
 345                         (long)addr,
 346                         (long)DYNAMIC_0_SPACE_START,
 347                         (long)DYNAMIC_1_SPACE_START);
 348                 lose("warning: core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 349             }
 350 #endif
 351 #if defined(ALLOCATION_POINTER)
 352             SetSymbolValue(ALLOCATION_POINTER, (lispobj)free_pointer,0);
 353 #else
 354             dynamic_space_free_pointer = free_pointer;
 355 #endif
 356             /* For stop-and-copy GC, this will be whatever the GC was
 357              * using at the time. With GENCGC, this will always be
 358              * space 0. (We checked above that for GENCGC,
 359              * addr==DYNAMIC_SPACE_START.) */
 360             current_dynamic_space = (lispobj *)addr;
 361             break;
 362         case STATIC_CORE_SPACE_ID:
 363             if (addr != (os_vm_address_t)STATIC_SPACE_START) {
 364                 fprintf(stderr, "in core: 0x%lx - in runtime: 0x%lx\n",
 365                         (long)addr, (long)STATIC_SPACE_START);
 366                 lose("core/runtime address mismatch: STATIC_SPACE_START\n");
 367             }
 368             break;
 369         case READ_ONLY_CORE_SPACE_ID:
 370             if (addr != (os_vm_address_t)READ_ONLY_SPACE_START) {
 371                 fprintf(stderr, "in core: 0x%lx - in runtime: 0x%lx\n",
 372                         (long)addr, (long)READ_ONLY_SPACE_START);
 373                 lose("core/runtime address mismatch: READ_ONLY_SPACE_START\n");
 374             }
 375             break;
 376         default:
 377             lose("unknown space ID %ld addr 0x%lx\n", id, (long)addr);
 378         }
 379     }
 380 }
 381
 382 lispobj
 383 load_core_file(char *file, os_vm_offset_t file_offset)
 384 {
 385     lispobj *header, val, len, *ptr, remaining_len;
 386     int fd = open_binary(file, O_RDONLY);
 387     unsigned int count;
 388
 389     lispobj initial_function = NIL;
 390     FSHOW((stderr, "/entering load_core_file(%s)\n", file));
 391     if (fd < 0) {
 392         fprintf(stderr, "could not open file \"%s\"\n", file);
 393         perror("open");
 394         exit(1);
 395     }
 396
 397     lseek(fd, file_offset, SEEK_SET);
 398     header = calloc(os_vm_page_size / sizeof(u32), sizeof(u32));
 399
 400     count = read(fd, header, os_vm_page_size);
 401     if (count < os_vm_page_size) {
 402         lose("premature end of core file\n");
 403     }
 404     SHOW("successfully read first page of core");
 405
 406     ptr = header;
 407     val = *ptr++;
 408
 409     if (val != CORE_MAGIC) {
 410         lose("invalid magic number in core: 0x%lx should have been 0x%x.\n",
 411              val,
 412              CORE_MAGIC);
 413     }
 414     SHOW("found CORE_MAGIC");
 415
 416     while (val != END_CORE_ENTRY_TYPE_CODE) {
 417         val = *ptr++;
 418         len = *ptr++;
 419         remaining_len = len - 2; /* (-2 to cancel the two ++ operations) */
 420         FSHOW((stderr, "/val=0x%ld, remaining_len=0x%ld\n",
 421                (long)val, (long)remaining_len));
 422
 423         switch (val) {
 424
 425         case END_CORE_ENTRY_TYPE_CODE:
 426             SHOW("END_CORE_ENTRY_TYPE_CODE case");
 427             break;
 428
 429         case VERSION_CORE_ENTRY_TYPE_CODE:
 430             SHOW("VERSION_CORE_ENTRY_TYPE_CODE case");
 431             if (*ptr != SBCL_CORE_VERSION_INTEGER) {
 432                 lose("core file version (%d) != runtime library version (%d)\n",
 433                      *ptr,
 434                      SBCL_CORE_VERSION_INTEGER);
 435             }
 436             break;
 437
 438         case BUILD_ID_CORE_ENTRY_TYPE_CODE:
 439             SHOW("BUILD_ID_CORE_ENTRY_TYPE_CODE case");
 440             {
 441                 unsigned int i;
 442
 443                 FSHOW((stderr, "build_id[]=\"%s\"\n", build_id));
 444                 FSHOW((stderr, "remaining_len = %d\n", remaining_len));
 445                 if (remaining_len != strlen((const char *)build_id))
 446                     goto losing_build_id;
 447                 for (i = 0; i < remaining_len; ++i) {
 448                     FSHOW((stderr, "ptr[%d] = char = %d, expected=%d\n",
 449                            i, ptr[i], build_id[i]));
 450                     if (ptr[i] != build_id[i])
 451                         goto losing_build_id;
 452                 }
 453                 break;
 454             losing_build_id:
 455                 /* .core files are not binary-compatible between
 456                  * builds because we can't easily detect whether the
 457                  * sources were patched between the time the
 458                  * dumping-the-.core runtime was built and the time
 459                  * that the loading-the-.core runtime was built.
 460                  *
 461                  * (We could easily detect whether version.lisp-expr
 462                  * was changed, but people experimenting with patches
 463                  * don't necessarily update version.lisp-expr.) */
 464
 465                 lose("can't load .core for different runtime, sorry\n");
 466             }
 467
 468         case NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE:
 469             SHOW("NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE case");
 470             process_directory(fd,
 471                               ptr,
 472 #ifndef LISP_FEATURE_ALPHA
 473                               remaining_len / (sizeof(struct ndir_entry) /
 474                                                sizeof(long)),
 475 #else
 476                               remaining_len / (sizeof(struct ndir_entry) /
 477                                                sizeof(u32)),
 478 #endif
 479                               file_offset);
 480             break;
 481
 482         case INITIAL_FUN_CORE_ENTRY_TYPE_CODE:
 483             SHOW("INITIAL_FUN_CORE_ENTRY_TYPE_CODE case");
 484             initial_function = (lispobj)*ptr;
 485             break;
 486
 487 #ifdef LISP_FEATURE_GENCGC
 488         case PAGE_TABLE_CORE_ENTRY_TYPE_CODE:
 489         {
 490             size_t size = *ptr;
 491             size_t fdoffset = (*(ptr+1) + 1) * (os_vm_page_size);
 492             size_t offset = 0;
 493             long bytes_read;
 494             unsigned long data[4096];
 495             unsigned long word;
 496             lseek(fd, fdoffset + file_offset, SEEK_SET);
 497             while ((bytes_read = read(fd, data, (size < 4096 ? size : 4096 )))
 498                     > 0)
 499             {
 500                 int i = 0;
 501                 size -= bytes_read;
 502                 while (bytes_read) {
 503                     bytes_read -= sizeof(long);
 504                     /* Ignore all zeroes. The size of the page table
 505                      * core entry was rounded up to os_vm_page_size
 506                      * during the save, and might now have more
 507                      * elements than the page table.
 508                      *
 509                      * The low bits of each word are allocation flags.
 510                      */
 511                     if (word=data[i]) {
 512                         page_table[offset].region_start_offset = word & ~0x03;
 513                         page_table[offset].allocated = word & 0x03;
 514                     }
 515                     i++;
 516                     offset++;
 517                 }
 518             }
 519
 520             gencgc_partial_pickup = 1;
 521             break;
 522         }
 523 #endif
 524         default:
 525             lose("unknown core file entry: %ld\n", (long)val);
 526         }
 527
 528         ptr += remaining_len;
 529         FSHOW((stderr, "/new ptr=%lx\n", (unsigned long)ptr));
 530     }
 531     SHOW("about to free(header)");
 532     free(header);
 533     SHOW("returning from load_core_file(..)");
 534     return initial_function;
 535 }
 536