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 # ifdef LISP_FEATURE_WIN32
 219     /* Ensure the memory is committed so zlib doesn't segfault trying to
 220        inflate. */
 221     os_validate_recommit(addr, len);
 222 # endif
 223
 224     if (-1 == lseek(fd, offset, SEEK_SET)) {
 225         lose("Unable to lseek() on corefile\n");
 226     }
 227
 228     stream.zalloc = NULL;
 229     stream.zfree = NULL;
 230     stream.opaque = NULL;
 231     stream.avail_in = 0;
 232     stream.next_in = buf;
 233
 234     ret = inflateInit(&stream);
 235     if (ret != Z_OK)
 236         lose("zlib error %i\n", ret);
 237
 238     stream.next_out  = (void*)addr;
 239     stream.avail_out = len;
 240     do {
 241         ssize_t count = read(fd, buf, sizeof(buf));
 242         if (count < 0)
 243             lose("unable to read core file (errno = %i)\n", errno);
 244         stream.next_in = buf;
 245         stream.avail_in = count;
 246         if (count == 0) break;
 247         ret = inflate(&stream, Z_NO_FLUSH);
 248         switch (ret) {
 249         case Z_STREAM_END:
 250             break;
 251         case Z_OK:
 252             if (stream.avail_out == 0)
 253                 lose("Runaway gzipped core directory... aborting\n");
 254             if (stream.avail_in > 0)
 255                 lose("zlib inflate returned without fully"
 256                      "using up input buffer... aborting\n");
 257             break;
 258         default:
 259             lose("zlib inflate error: %i\n", ret);
 260             break;
 261         }
 262     } while (ret != Z_STREAM_END);
 263
 264     if (stream.avail_out > 0) {
 265         if (stream.avail_out >= os_vm_page_size)
 266             fprintf(stderr, "Warning: gzipped core directory significantly"
 267                     "shorter than expected (%lu bytes)", (unsigned long)stream.avail_out);
 268         /* Is this needed? */
 269         memset(stream.next_out, 0, stream.avail_out);
 270     }
 271
 272     inflateEnd(&stream);
 273     return addr;
 274 }
 275 # undef ZLIB_BUFFER_SIZE
 276 #endif
 277
 278 int merge_core_pages = -1;
 279
 280 static void
 281 process_directory(int fd, lispobj *ptr, int count, os_vm_offset_t file_offset)
 282 {
 283     extern void immobile_space_coreparse(uword_t,uword_t);
 284     struct ndir_entry *entry;
 285     int compressed;
 286
 287     FSHOW((stderr, "/process_directory(..), count=%d\n", count));
 288
 289     for (entry = (struct ndir_entry *) ptr; --count>= 0; ++entry) {
 290
 291         compressed = 0;
 292         sword_t id = entry->identifier;
 293         if (id <= (MAX_CORE_SPACE_ID | DEFLATED_CORE_SPACE_ID_FLAG)) {
 294             if (id & DEFLATED_CORE_SPACE_ID_FLAG)
 295                 compressed = 1;
 296             id &= ~(DEFLATED_CORE_SPACE_ID_FLAG);
 297         }
 298         sword_t offset = os_vm_page_size * (1 + entry->data_page);
 299         os_vm_address_t addr =
 300             (os_vm_address_t) (os_vm_page_size * entry->address);
 301         lispobj *free_pointer = (lispobj *) addr + entry->nwords;
 302         uword_t len = os_vm_page_size * entry->page_count;
 303         if (len != 0) {
 304             os_vm_address_t real_addr;
 305             FSHOW((stderr, "/mapping %ld(0x%lx) bytes at 0x%lx\n",
 306                    len, len, (uword_t)addr));
 307             if (compressed) {
 308 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
 309                 real_addr = inflate_core_bytes(fd, offset + file_offset, addr, len);
 310 #else
 311                 lose("This runtime was not built with zlib-compressed core support... aborting\n");
 312 #endif
 313             } else {
 314 #ifdef LISP_FEATURE_HPUX
 315                 real_addr = copy_core_bytes(fd, offset + file_offset, addr, len);
 316 #else
 317                 real_addr = os_map(fd, offset + file_offset, addr, len);
 318 #endif
 319             }
 320             if (real_addr != addr) {
 321                 lose("file mapped in wrong place! "
 322                      "(0x%08x != 0x%08lx)\n",
 323                      real_addr,
 324                      addr);
 325             }
 326         }
 327
 328 #ifdef MADV_MERGEABLE
 329         if ((merge_core_pages == 1)
 330             || ((merge_core_pages == -1) && compressed)) {
 331                 madvise(addr, len, MADV_MERGEABLE);
 332         }
 333 #endif
 334         FSHOW((stderr, "/space id = %ld, free pointer = %p\n",
 335                id, (uword_t)free_pointer));
 336
 337         switch (id) {
 338         case DYNAMIC_CORE_SPACE_ID:
 339             if (len > dynamic_space_size) {
 340                 fprintf(stderr,
 341                         "dynamic space too small for core: %luKiB required, %luKiB available.\n",
 342                         (unsigned long)len >> 10,
 343                         (unsigned long)dynamic_space_size >> 10);
 344                 exit(1);
 345             }
 346 #ifdef LISP_FEATURE_GENCGC
 347             if (addr != (os_vm_address_t)DYNAMIC_SPACE_START) {
 348                 fprintf(stderr, "in core: %p; in runtime: %p \n",
 349                         (void*)addr, (void*)DYNAMIC_SPACE_START);
 350                 lose("core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 351             }
 352 #else
 353             if ((addr != (os_vm_address_t)DYNAMIC_0_SPACE_START) &&
 354                 (addr != (os_vm_address_t)DYNAMIC_1_SPACE_START)) {
 355                 fprintf(stderr, "in core: %p; in runtime: %p or %p\n",
 356                         (void*)addr,
 357                         (void*)DYNAMIC_0_SPACE_START,
 358                         (void*)DYNAMIC_1_SPACE_START);
 359                 lose("warning: core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 360             }
 361 #endif
 362 #if defined(ALLOCATION_POINTER)
 363             SetSymbolValue(ALLOCATION_POINTER, (lispobj)free_pointer,0);
 364 #else
 365             dynamic_space_free_pointer = free_pointer;
 366 #endif
 367             /* For stop-and-copy GC, this will be whatever the GC was
 368              * using at the time. With GENCGC, this will always be
 369              * space 0. (We checked above that for GENCGC,
 370              * addr==DYNAMIC_SPACE_START.) */
 371             current_dynamic_space = (lispobj *)addr;
 372             break;
 373         case STATIC_CORE_SPACE_ID:
 374             if (addr != (os_vm_address_t)STATIC_SPACE_START) {
 375                 fprintf(stderr, "in core: %p - in runtime: %p\n",
 376                         (void*)addr, (void*)STATIC_SPACE_START);
 377                 lose("core/runtime address mismatch: STATIC_SPACE_START\n");
 378             }
 379             break;
 380         case READ_ONLY_CORE_SPACE_ID:
 381             if (addr != (os_vm_address_t)READ_ONLY_SPACE_START) {
 382                 fprintf(stderr, "in core: %p - in runtime: %p\n",
 383                         (void*)addr, (void*)READ_ONLY_SPACE_START);
 384                 lose("core/runtime address mismatch: READ_ONLY_SPACE_START\n");
 385             }
 386             break;
 387 #ifdef LISP_FEATURE_IMMOBILE_SPACE
 388          // Immobile space is subdivided into fixed-size and variable-size.
 389          // There is no margin between the two, though for efficiency
 390          // they are written separately to eliminate waste in the core file.
 391         case IMMOBILE_FIXEDOBJ_CORE_SPACE_ID:
 392             if (addr != (os_vm_address_t)IMMOBILE_SPACE_START) {
 393                 fprintf(stderr, "in core: %p - in runtime: %p\n",
 394                         (void*)addr, (void*)IMMOBILE_SPACE_START);
 395                 lose("core/runtime address mismatch: IMMOBILE_SPACE_START\n");
 396             }
 397             immobile_space_coreparse(IMMOBILE_SPACE_START, len);
 398             break;
 399         case IMMOBILE_VARYOBJ_CORE_SPACE_ID:
 400             if (addr != (os_vm_address_t)IMMOBILE_VARYOBJ_SUBSPACE_START) {
 401                 fprintf(stderr, "in core: %p - in runtime: %p\n",
 402                         (void*)addr, (void*)IMMOBILE_VARYOBJ_SUBSPACE_START);
 403                 lose("core/runtime address mismatch: IMMOBILE_VARYOBJ_SUBSPACE_START\n");
 404             }
 405             immobile_space_coreparse(IMMOBILE_VARYOBJ_SUBSPACE_START, len);
 406             break;
 407 #endif
 408         default:
 409             lose("unknown space ID %ld addr %p\n", id, addr);
 410         }
 411     }
 412 }
 413
 414 lispobj
 415 load_core_file(char *file, os_vm_offset_t file_offset)
 416 {
 417     void *header;
 418 #ifndef LISP_FEATURE_ALPHA
 419     word_t val, *ptr;
 420 #else
 421     u32 val, *ptr;
 422 #endif
 423     os_vm_size_t len, remaining_len;
 424     int fd = open_binary(file, O_RDONLY);
 425     ssize_t count;
 426     lispobj initial_function = NIL;
 427
 428     FSHOW((stderr, "/entering load_core_file(%s)\n", file));
 429     if (fd < 0) {
 430         fprintf(stderr, "could not open file \"%s\"\n", file);
 431         perror("open");
 432         exit(1);
 433     }
 434
 435     lseek(fd, file_offset, SEEK_SET);
 436     header = calloc(os_vm_page_size, 1);
 437
 438     count = read(fd, header, os_vm_page_size);
 439     if (count < (ssize_t) os_vm_page_size) {
 440         lose("premature end of core file\n");
 441     }
 442     SHOW("successfully read first page of core");
 443
 444     ptr = header;
 445     val = *ptr++;
 446
 447     if (val != CORE_MAGIC) {
 448         lose("invalid magic number in core: 0x%lx should have been 0x%x.\n",
 449              val,
 450              CORE_MAGIC);
 451     }
 452     SHOW("found CORE_MAGIC");
 453
 454     while (val != END_CORE_ENTRY_TYPE_CODE) {
 455         val = *ptr++;
 456         len = *ptr++;
 457         remaining_len = len - 2; /* (-2 to cancel the two ++ operations) */
 458         FSHOW((stderr, "/val=0x%"WORD_FMTX", remaining_len=0x%"WORD_FMTX"\n",
 459                val, remaining_len));
 460
 461         switch (val) {
 462
 463         case END_CORE_ENTRY_TYPE_CODE:
 464             SHOW("END_CORE_ENTRY_TYPE_CODE case");
 465             break;
 466
 467         case BUILD_ID_CORE_ENTRY_TYPE_CODE:
 468             SHOW("BUILD_ID_CORE_ENTRY_TYPE_CODE case");
 469             {
 470                 os_vm_size_t i;
 471
 472                 FSHOW((stderr, "build_id[]=\"%s\"\n", build_id));
 473                 FSHOW((stderr, "remaining_len = %d\n", remaining_len));
 474                 if (remaining_len != strlen((const char *)build_id))
 475                     goto losing_build_id;
 476                 for (i = 0; i < remaining_len; ++i) {
 477                     FSHOW((stderr, "ptr[%d] = char = %d, expected=%d\n",
 478                            i, ptr[i], build_id[i]));
 479                     if (ptr[i] != build_id[i])
 480                         goto losing_build_id;
 481                 }
 482                 break;
 483             losing_build_id:
 484                 /* .core files are not binary-compatible between
 485                  * builds because we can't easily detect whether the
 486                  * sources were patched between the time the
 487                  * dumping-the-.core runtime was built and the time
 488                  * that the loading-the-.core runtime was built.
 489                  *
 490                  * (We could easily detect whether version.lisp-expr
 491                  * was changed, but people experimenting with patches
 492                  * don't necessarily update version.lisp-expr.) */
 493
 494                 lose("can't load .core for different runtime, sorry\n");
 495             }
 496
 497         case NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE:
 498             SHOW("NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE case");
 499             process_directory(fd,
 500                               ptr,
 501 #ifndef LISP_FEATURE_ALPHA
 502                               remaining_len / (sizeof(struct ndir_entry) /
 503                                                sizeof(lispobj)),
 504 #else
 505                               remaining_len / (sizeof(struct ndir_entry) /
 506                                                sizeof(u32)),
 507 #endif
 508                               file_offset);
 509             break;
 510
 511         case INITIAL_FUN_CORE_ENTRY_TYPE_CODE:
 512             SHOW("INITIAL_FUN_CORE_ENTRY_TYPE_CODE case");
 513             initial_function = (lispobj)*ptr;
 514             break;
 515
 516 #ifdef LISP_FEATURE_GENCGC
 517         case PAGE_TABLE_CORE_ENTRY_TYPE_CODE:
 518         {
 519             os_vm_size_t size = *ptr;
 520             os_vm_size_t fdoffset = (*(ptr+1) + 1) * (os_vm_page_size);
 521             page_index_t offset = 0;
 522             ssize_t bytes_read;
 523             word_t data[4096];
 524             word_t word;
 525             lseek(fd, fdoffset + file_offset, SEEK_SET);
 526             while ((bytes_read = read(fd, data, (size < 4096 ? size : 4096 )))
 527                     > 0)
 528             {
 529                 int i = 0;
 530                 size -= bytes_read;
 531                 while (bytes_read) {
 532                     bytes_read -= sizeof(word_t);
 533                     /* Ignore all zeroes. The size of the page table
 534                      * core entry was rounded up to os_vm_page_size
 535                      * during the save, and might now have more
 536                      * elements than the page table.
 537                      *
 538                      * The low bits of each word are allocation flags.
 539                      */
 540                     if ((word=data[i])) {
 541                         page_table[offset].scan_start_offset = word & ~0x03;
 542                         page_table[offset].allocated = word & 0x03;
 543                     }
 544                     i++;
 545                     offset++;
 546                 }
 547             }
 548
 549             gencgc_partial_pickup = 1;
 550             break;
 551         }
 552 #endif
 553         default:
 554             lose("unknown core file entry: 0x%"WORD_FMTX"\n", val);
 555         }
 556
 557         ptr += remaining_len;
 558         FSHOW((stderr, "/new ptr=0x%"WORD_FMTX"\n", ptr));
 559     }
 560     SHOW("about to free(header)");
 561     free(header);
 562     close(fd);
 563     SHOW("returning from load_core_file(..)");
 564     return initial_function;
 565 }
 566
 567 #include "genesis/hash-table.h"
 568 #include "genesis/vector.h"
 569 os_vm_address_t get_asm_routine_by_name(const char* name)
 570 {
 571     lispobj routines = SYMBOL(ASSEMBLER_ROUTINES)->value;
 572     if (lowtag_of(routines) == INSTANCE_POINTER_LOWTAG) {
 573         struct hash_table* ht = (struct hash_table*)native_pointer(routines);
 574         struct vector* table = (struct vector*)native_pointer(ht->table);
 575         lispobj sym;
 576         int i;
 577         for (i=2 ; i < fixnum_value(table->length) ; i += 2) {
 578           sym = table->data[i];
 579           if (lowtag_of(sym) == OTHER_POINTER_LOWTAG
 580               && widetag_of(SYMBOL(sym)->header) == SYMBOL_HEADER_WIDETAG
 581               && !strcmp(name,
 582                          (char*)((struct vector*)
 583                                  native_pointer(SYMBOL(sym)->name))->data))
 584               return (os_vm_address_t)fixnum_value(table->data[i+1]);
 585         }
 586         // Something is wrong if we have a hashtable but find nothing.
 587         fprintf(stderr, "WARNING: get_asm_routine_by_name(%s) failed\n",
 588                 name);
 589     }
 590     return NULL;
 591 }
 592
 593 void asm_routine_poke(const char* routine, int offset, char byte)
 594 {
 595     char *address = (char *)get_asm_routine_by_name(routine);
 596     if (address)
 597         address[offset] = byte;
 598 }