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 "sbcl.h"
  18
  19 #ifndef LISP_FEATURE_WIN32
  20 #ifdef LISP_FEATURE_LINUX
  21 /* For madvise */
  22 #define _BSD_SOURCE
  23 #include <sys/mman.h>
  24 #undef _BSD_SOURCE
  25 #else
  26 #include <sys/mman.h>
  27 #endif
  28 #endif
  29
  30 #include <stdio.h>
  31 #include <stdlib.h>
  32 #include <string.h>
  33 #include <sys/file.h>
  34 #include <sys/types.h>
  35 #include <sys/stat.h>
  36 #include <fcntl.h>
  37 #include <unistd.h>
  38
  39 #include "os.h"
  40 #include "runtime.h"
  41 #include "globals.h"
  42 #include "core.h"
  43 #include "arch.h"
  44 #include "interr.h"
  45 #include "thread.h"
  46
  47 #include "validate.h"
  48 #include "gc-internal.h"
  49 #include "runtime-options.h"
  50
  51 #include <errno.h>
  52
  53 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
  54 # include <zlib.h>
  55 #endif
  56
  57 unsigned char build_id[] =
  58 #include "../../output/build-id.tmp"
  59 ;
  60
  61 int
  62 open_binary(char *filename, int mode)
  63 {
  64 #ifdef LISP_FEATURE_WIN32
  65     mode |= O_BINARY;
  66 #endif
  67
  68     return open(filename, mode);
  69 }
  70
  71
  72 static struct runtime_options *
  73 read_runtime_options(int fd)
  74 {
  75     os_vm_size_t optarray[RUNTIME_OPTIONS_WORDS];
  76     struct runtime_options *options = NULL;
  77
  78     if (read(fd, optarray, RUNTIME_OPTIONS_WORDS * sizeof(os_vm_size_t)) !=
  79         RUNTIME_OPTIONS_WORDS * sizeof(size_t)) {
  80         return NULL;
  81     }
  82
  83     if ((RUNTIME_OPTIONS_MAGIC != optarray[0]) || (0 == optarray[1])) {
  84         return NULL;
  85     }
  86
  87     options = successful_malloc(sizeof(struct runtime_options));
  88
  89     options->dynamic_space_size = optarray[2];
  90     options->thread_control_stack_size = optarray[3];
  91
  92     return options;
  93 }
  94
  95 void
  96 maybe_initialize_runtime_options(int fd)
  97 {
  98     struct runtime_options *new_runtime_options;
  99     off_t end_offset = sizeof(lispobj) +
 100         sizeof(os_vm_offset_t) +
 101         (RUNTIME_OPTIONS_WORDS * sizeof(size_t));
 102
 103     lseek(fd, -end_offset, SEEK_END);
 104
 105     if ((new_runtime_options = read_runtime_options(fd))) {
 106         runtime_options = new_runtime_options;
 107     }
 108 }
 109
 110 /* Search 'filename' for an embedded core.  An SBCL core has, at the
 111  * end of the file, a trailer containing optional saved runtime
 112  * options, the start of the core (an os_vm_offset_t), and a final
 113  * signature word (the lispobj CORE_MAGIC).  If this trailer is found
 114  * at the end of the file, the start of the core can be determined
 115  * from the core size.
 116  *
 117  * If an embedded core is present, this returns the offset into the
 118  * file to load the core from, or -1 if no core is present. */
 119 os_vm_offset_t
 120 search_for_embedded_core(char *filename)
 121 {
 122     lispobj header;
 123     os_vm_offset_t lispobj_size = sizeof(lispobj);
 124     os_vm_offset_t trailer_size = lispobj_size + sizeof(os_vm_offset_t);
 125     os_vm_offset_t core_start, pos;
 126     int fd = -1;
 127
 128     if ((fd = open_binary(filename, O_RDONLY)) < 0)
 129         goto lose;
 130
 131     if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
 132         goto lose;
 133     if (header == CORE_MAGIC) {
 134         /* This file is a real core, not an embedded core.  Return 0 to
 135          * indicate where the core starts, and do not look for runtime
 136          * options in this case. */
 137         return 0;
 138     }
 139
 140     if (lseek(fd, -lispobj_size, SEEK_END) < 0)
 141         goto lose;
 142     if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
 143         goto lose;
 144
 145     if (header == CORE_MAGIC) {
 146         if (lseek(fd, -trailer_size, SEEK_END) < 0)
 147             goto lose;
 148         if (read(fd, &core_start, sizeof(os_vm_offset_t)) < 0)
 149             goto lose;
 150
 151         if (lseek(fd, core_start, SEEK_SET) < 0)
 152             goto lose;
 153         pos = lseek(fd, 0, SEEK_CUR);
 154
 155         if (read(fd, &header, (size_t)lispobj_size) < lispobj_size)
 156             goto lose;
 157
 158         if (header != CORE_MAGIC)
 159             goto lose;
 160
 161         maybe_initialize_runtime_options(fd);
 162
 163         close(fd);
 164         return pos;
 165     }
 166
 167 lose:
 168     if (fd != -1)
 169         close(fd);
 170
 171     return -1;
 172 }
 173
 174 /* If more platforms doesn't support overlapping mmap rename this
 175  * def to something like ifdef nommapoverlap */
 176 /* currently hpux only */
 177 #ifdef LISP_FEATURE_HPUX
 178 os_vm_address_t copy_core_bytes(int fd, os_vm_offset_t offset,
 179                                 os_vm_address_t addr, int len)
 180 {
 181   unsigned char buf[4096];
 182   int c,x;
 183   int old_fd = lseek(fd, 0, SEEK_CUR);
 184
 185   if(len & (4096-1)){
 186     fprintf(stderr, "cant copy a slice of core because slice-length is not of page size(4096)\n");
 187     exit(-1);
 188   }
 189   if(old_fd < 0){
 190     fprintf(stderr, "cant perform lseek() on corefile\n");
 191   }
 192   lseek(fd, offset, SEEK_SET);
 193   if(fd < 0){
 194     fprintf(stderr, "cant perform lseek(%u,%lu,SEEK_SET) on corefile\n", fd, offset);
 195   }
 196   for(x = 0; x < len; x += 4096){
 197     c = read(fd, buf, 4096);
 198     if(c != 4096){
 199       fprintf(stderr, "cant read memory area from corefile at position %lu, got %d\n", offset + x, c);
 200       exit(-1);
 201     }
 202     memcpy(addr+x, buf, 4096);
 203   }
 204   os_flush_icache(addr, len);
 205   return addr;
 206 }
 207 #endif
 208
 209 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
 210 # define ZLIB_BUFFER_SIZE (1u<<16)
 211 os_vm_address_t inflate_core_bytes(int fd, os_vm_offset_t offset,
 212                                    os_vm_address_t addr, int len)
 213 {
 214     z_stream stream;
 215     unsigned char buf[ZLIB_BUFFER_SIZE];
 216     int ret;
 217
 218     if (-1 == lseek(fd, offset, SEEK_SET)) {
 219         lose("Unable to lseek() on corefile\n");
 220     }
 221
 222     stream.zalloc = NULL;
 223     stream.zfree = NULL;
 224     stream.opaque = NULL;
 225     stream.avail_in = 0;
 226     stream.next_in = buf;
 227
 228     ret = inflateInit(&stream);
 229     if (ret != Z_OK)
 230         lose("zlib error %i\n", ret);
 231
 232     stream.next_out  = (void*)addr;
 233     stream.avail_out = len;
 234     do {
 235         ssize_t count = read(fd, buf, sizeof(buf));
 236         if (count < 0)
 237             lose("unable to read core file (errno = %i)\n", errno);
 238         stream.next_in = buf;
 239         stream.avail_in = count;
 240         if (count == 0) break;
 241         ret = inflate(&stream, Z_NO_FLUSH);
 242         switch (ret) {
 243         case Z_STREAM_END:
 244             break;
 245         case Z_OK:
 246             if (stream.avail_out == 0)
 247                 lose("Runaway gzipped core directory... aborting\n");
 248             if (stream.avail_in > 0)
 249                 lose("zlib inflate returned without fully"
 250                      "using up input buffer... aborting\n");
 251             break;
 252         default:
 253             lose("zlib inflate error: %i\n", ret);
 254             break;
 255         }
 256     } while (ret != Z_STREAM_END);
 257
 258     if (stream.avail_out > 0) {
 259         if (stream.avail_out >= os_vm_page_size)
 260             fprintf(stderr, "Warning: gzipped core directory significantly"
 261                     "shorter than expected (%lu bytes)", (unsigned long)stream.avail_out);
 262         /* Is this needed? */
 263         memset(stream.next_out, 0, stream.avail_out);
 264     }
 265
 266     inflateEnd(&stream);
 267     return addr;
 268 }
 269 # undef ZLIB_BUFFER_SIZE
 270 #endif
 271
 272 int merge_core_pages = -1;
 273
 274 static void
 275 process_directory(int fd, lispobj *ptr, int count, os_vm_offset_t file_offset)
 276 {
 277     struct ndir_entry *entry;
 278     int compressed;
 279
 280     FSHOW((stderr, "/process_directory(..), count=%d\n", count));
 281
 282     for (entry = (struct ndir_entry *) ptr; --count>= 0; ++entry) {
 283         compressed = 0;
 284         long id = entry->identifier;
 285         if (id <= (MAX_CORE_SPACE_ID | DEFLATED_CORE_SPACE_ID_FLAG)) {
 286             if (id & DEFLATED_CORE_SPACE_ID_FLAG)
 287                 compressed = 1;
 288             id &= ~(DEFLATED_CORE_SPACE_ID_FLAG);
 289         }
 290         long offset = os_vm_page_size * (1 + entry->data_page);
 291         os_vm_address_t addr =
 292             (os_vm_address_t) (os_vm_page_size * entry->address);
 293         lispobj *free_pointer = (lispobj *) addr + entry->nwords;
 294         unsigned long len = os_vm_page_size * entry->page_count;
 295         if (len != 0) {
 296             os_vm_address_t real_addr;
 297             FSHOW((stderr, "/mapping %ld(0x%lx) bytes at 0x%lx\n",
 298                    (long)len, (long)len, (unsigned long)addr));
 299             if (compressed) {
 300 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
 301                 real_addr = inflate_core_bytes(fd, offset + file_offset, addr, len);
 302 #else
 303                 lose("This runtime was not built with zlib-compressed core support... aborting\n");
 304 #endif
 305             } else {
 306 #ifdef LISP_FEATURE_HPUX
 307                 real_addr = copy_core_bytes(fd, offset + file_offset, addr, len);
 308 #else
 309                 real_addr = os_map(fd, offset + file_offset, addr, len);
 310 #endif
 311             }
 312             if (real_addr != addr) {
 313                 lose("file mapped in wrong place! "
 314                      "(0x%08x != 0x%08lx)\n",
 315                      real_addr,
 316                      addr);
 317             }
 318         }
 319
 320 #ifdef MADV_MERGEABLE
 321         if ((merge_core_pages == 1)
 322             || ((merge_core_pages == -1) && compressed)) {
 323                 madvise(addr, len, MADV_MERGEABLE);
 324         }
 325 #endif
 326
 327         FSHOW((stderr, "/space id = %ld, free pointer = 0x%lx\n",
 328                id, (unsigned long)free_pointer));
 329
 330         switch (id) {
 331         case DYNAMIC_CORE_SPACE_ID:
 332             if (len > dynamic_space_size) {
 333                 fprintf(stderr,
 334                         "dynamic space too small for core: %ldKiB required, %ldKiB available.\n",
 335                         len >> 10,
 336                         (long)dynamic_space_size >> 10);
 337                 exit(1);
 338             }
 339 #ifdef LISP_FEATURE_GENCGC
 340             if (addr != (os_vm_address_t)DYNAMIC_SPACE_START) {
 341                 fprintf(stderr, "in core: 0x%lx; in runtime: 0x%lx \n",
 342                         (long)addr, (long)DYNAMIC_SPACE_START);
 343                 lose("core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 344             }
 345 #else
 346             if ((addr != (os_vm_address_t)DYNAMIC_0_SPACE_START) &&
 347                 (addr != (os_vm_address_t)DYNAMIC_1_SPACE_START)) {
 348                 fprintf(stderr, "in core: 0x%lx; in runtime: 0x%lx or 0x%lx\n",
 349                         (long)addr,
 350                         (long)DYNAMIC_0_SPACE_START,
 351                         (long)DYNAMIC_1_SPACE_START);
 352                 lose("warning: core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 353             }
 354 #endif
 355 #if defined(ALLOCATION_POINTER)
 356             SetSymbolValue(ALLOCATION_POINTER, (lispobj)free_pointer,0);
 357 #else
 358             dynamic_space_free_pointer = free_pointer;
 359 #endif
 360             /* For stop-and-copy GC, this will be whatever the GC was
 361              * using at the time. With GENCGC, this will always be
 362              * space 0. (We checked above that for GENCGC,
 363              * addr==DYNAMIC_SPACE_START.) */
 364             current_dynamic_space = (lispobj *)addr;
 365             break;
 366         case STATIC_CORE_SPACE_ID:
 367             if (addr != (os_vm_address_t)STATIC_SPACE_START) {
 368                 fprintf(stderr, "in core: 0x%lx - in runtime: 0x%lx\n",
 369                         (long)addr, (long)STATIC_SPACE_START);
 370                 lose("core/runtime address mismatch: STATIC_SPACE_START\n");
 371             }
 372             break;
 373         case READ_ONLY_CORE_SPACE_ID:
 374             if (addr != (os_vm_address_t)READ_ONLY_SPACE_START) {
 375                 fprintf(stderr, "in core: 0x%lx - in runtime: 0x%lx\n",
 376                         (long)addr, (long)READ_ONLY_SPACE_START);
 377                 lose("core/runtime address mismatch: READ_ONLY_SPACE_START\n");
 378             }
 379             break;
 380         default:
 381             lose("unknown space ID %ld addr 0x%lx\n", id, (long)addr);
 382         }
 383     }
 384 }
 385
 386 lispobj
 387 load_core_file(char *file, os_vm_offset_t file_offset)
 388 {
 389     void *header;
 390     word_t val, *ptr;
 391     os_vm_size_t len, remaining_len;
 392     int fd = open_binary(file, O_RDONLY);
 393     ssize_t count;
 394     lispobj initial_function = NIL;
 395
 396     FSHOW((stderr, "/entering load_core_file(%s)\n", file));
 397     if (fd < 0) {
 398         fprintf(stderr, "could not open file \"%s\"\n", file);
 399         perror("open");
 400         exit(1);
 401     }
 402
 403     lseek(fd, file_offset, SEEK_SET);
 404     header = calloc(os_vm_page_size, 1);
 405
 406     count = read(fd, header, os_vm_page_size);
 407     if (count < os_vm_page_size) {
 408         lose("premature end of core file\n");
 409     }
 410     SHOW("successfully read first page of core");
 411
 412     ptr = header;
 413     val = *ptr++;
 414
 415     if (val != CORE_MAGIC) {
 416         lose("invalid magic number in core: 0x%lx should have been 0x%x.\n",
 417              val,
 418              CORE_MAGIC);
 419     }
 420     SHOW("found CORE_MAGIC");
 421
 422     while (val != END_CORE_ENTRY_TYPE_CODE) {
 423         val = *ptr++;
 424         len = *ptr++;
 425         remaining_len = len - 2; /* (-2 to cancel the two ++ operations) */
 426         FSHOW((stderr, "/val=0x%"WORD_FMTX", remaining_len=0x%"WORD_FMTX"\n",
 427                val, remaining_len));
 428
 429         switch (val) {
 430
 431         case END_CORE_ENTRY_TYPE_CODE:
 432             SHOW("END_CORE_ENTRY_TYPE_CODE case");
 433             break;
 434
 435         case VERSION_CORE_ENTRY_TYPE_CODE:
 436             SHOW("VERSION_CORE_ENTRY_TYPE_CODE case");
 437             if (*ptr != SBCL_CORE_VERSION_INTEGER) {
 438                 lose("core file version (%d) != runtime library version (%d)\n",
 439                      *ptr,
 440                      SBCL_CORE_VERSION_INTEGER);
 441             }
 442             break;
 443
 444         case BUILD_ID_CORE_ENTRY_TYPE_CODE:
 445             SHOW("BUILD_ID_CORE_ENTRY_TYPE_CODE case");
 446             {
 447                 os_vm_size_t i;
 448
 449                 FSHOW((stderr, "build_id[]=\"%s\"\n", build_id));
 450                 FSHOW((stderr, "remaining_len = %d\n", remaining_len));
 451                 if (remaining_len != strlen((const char *)build_id))
 452                     goto losing_build_id;
 453                 for (i = 0; i < remaining_len; ++i) {
 454                     FSHOW((stderr, "ptr[%d] = char = %d, expected=%d\n",
 455                            i, ptr[i], build_id[i]));
 456                     if (ptr[i] != build_id[i])
 457                         goto losing_build_id;
 458                 }
 459                 break;
 460             losing_build_id:
 461                 /* .core files are not binary-compatible between
 462                  * builds because we can't easily detect whether the
 463                  * sources were patched between the time the
 464                  * dumping-the-.core runtime was built and the time
 465                  * that the loading-the-.core runtime was built.
 466                  *
 467                  * (We could easily detect whether version.lisp-expr
 468                  * was changed, but people experimenting with patches
 469                  * don't necessarily update version.lisp-expr.) */
 470
 471                 lose("can't load .core for different runtime, sorry\n");
 472             }
 473
 474         case NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE:
 475             SHOW("NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE case");
 476             process_directory(fd,
 477                               ptr,
 478 #ifndef LISP_FEATURE_ALPHA
 479                               remaining_len / (sizeof(struct ndir_entry) /
 480                                                sizeof(long)),
 481 #else
 482                               remaining_len / (sizeof(struct ndir_entry) /
 483                                                sizeof(u32)),
 484 #endif
 485                               file_offset);
 486             break;
 487
 488         case INITIAL_FUN_CORE_ENTRY_TYPE_CODE:
 489             SHOW("INITIAL_FUN_CORE_ENTRY_TYPE_CODE case");
 490             initial_function = (lispobj)*ptr;
 491             break;
 492
 493 #ifdef LISP_FEATURE_GENCGC
 494         case PAGE_TABLE_CORE_ENTRY_TYPE_CODE:
 495         {
 496             os_vm_size_t size = *ptr;
 497             os_vm_size_t fdoffset = (*(ptr+1) + 1) * (os_vm_page_size);
 498             page_index_t offset = 0;
 499             ssize_t bytes_read;
 500             word_t data[4096];
 501             word_t word;
 502             lseek(fd, fdoffset + file_offset, SEEK_SET);
 503             while ((bytes_read = read(fd, data, (size < 4096 ? size : 4096 )))
 504                     > 0)
 505             {
 506                 int i = 0;
 507                 size -= bytes_read;
 508                 while (bytes_read) {
 509                     bytes_read -= sizeof(word_t);
 510                     /* Ignore all zeroes. The size of the page table
 511                      * core entry was rounded up to os_vm_page_size
 512                      * during the save, and might now have more
 513                      * elements than the page table.
 514                      *
 515                      * The low bits of each word are allocation flags.
 516                      */
 517                     if ((word=data[i])) {
 518                         page_table[offset].region_start_offset = word & ~0x03;
 519                         page_table[offset].allocated = word & 0x03;
 520                     }
 521                     i++;
 522                     offset++;
 523                 }
 524             }
 525
 526             gencgc_partial_pickup = 1;
 527             break;
 528         }
 529 #endif
 530         default:
 531             lose("unknown core file entry: 0x%"WORD_FMTX"\n", val);
 532         }
 533
 534         ptr += remaining_len;
 535         FSHOW((stderr, "/new ptr=0x%"WORD_FMTX"\n", ptr));
 536     }
 537     SHOW("about to free(header)");
 538     free(header);
 539     SHOW("returning from load_core_file(..)");
 540     return initial_function;
 541 }
 542