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     struct ndir_entry *entry;
 284     int compressed;
 285
 286     FSHOW((stderr, "/process_directory(..), count=%d\n", count));
 287
 288     for (entry = (struct ndir_entry *) ptr; --count>= 0; ++entry) {
 289
 290         compressed = 0;
 291         sword_t id = entry->identifier;
 292         if (id <= (MAX_CORE_SPACE_ID | DEFLATED_CORE_SPACE_ID_FLAG)) {
 293             if (id & DEFLATED_CORE_SPACE_ID_FLAG)
 294                 compressed = 1;
 295             id &= ~(DEFLATED_CORE_SPACE_ID_FLAG);
 296         }
 297         sword_t offset = os_vm_page_size * (1 + entry->data_page);
 298         os_vm_address_t addr =
 299             (os_vm_address_t) (os_vm_page_size * entry->address);
 300         lispobj *free_pointer = (lispobj *) addr + entry->nwords;
 301         uword_t len = os_vm_page_size * entry->page_count;
 302         if (len != 0) {
 303             os_vm_address_t real_addr;
 304             FSHOW((stderr, "/mapping %ld(0x%lx) bytes at 0x%lx\n",
 305                    len, len, (uword_t)addr));
 306             if (compressed) {
 307 #ifdef LISP_FEATURE_SB_CORE_COMPRESSION
 308                 real_addr = inflate_core_bytes(fd, offset + file_offset, addr, len);
 309 #else
 310                 lose("This runtime was not built with zlib-compressed core support... aborting\n");
 311 #endif
 312             } else {
 313 #ifdef LISP_FEATURE_HPUX
 314                 real_addr = copy_core_bytes(fd, offset + file_offset, addr, len);
 315 #else
 316                 real_addr = os_map(fd, offset + file_offset, addr, len);
 317 #endif
 318             }
 319             if (real_addr != addr) {
 320                 lose("file mapped in wrong place! "
 321                      "(0x%08x != 0x%08lx)\n",
 322                      real_addr,
 323                      addr);
 324             }
 325         }
 326
 327 #ifdef MADV_MERGEABLE
 328         if ((merge_core_pages == 1)
 329             || ((merge_core_pages == -1) && compressed)) {
 330                 madvise(addr, len, MADV_MERGEABLE);
 331         }
 332 #endif
 333         FSHOW((stderr, "/space id = %ld, free pointer = %p\n",
 334                id, (uword_t)free_pointer));
 335
 336         switch (id) {
 337         case DYNAMIC_CORE_SPACE_ID:
 338             if (len > dynamic_space_size) {
 339                 fprintf(stderr,
 340                         "dynamic space too small for core: %ldKiB required, %ldKiB available.\n",
 341                         len >> 10,
 342                         (uword_t)dynamic_space_size >> 10);
 343                 exit(1);
 344             }
 345 #ifdef LISP_FEATURE_GENCGC
 346             if (addr != (os_vm_address_t)DYNAMIC_SPACE_START) {
 347                 fprintf(stderr, "in core: %p; in runtime: %p \n",
 348                         (void*)addr, (void*)DYNAMIC_SPACE_START);
 349                 lose("core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 350             }
 351 #else
 352             if ((addr != (os_vm_address_t)DYNAMIC_0_SPACE_START) &&
 353                 (addr != (os_vm_address_t)DYNAMIC_1_SPACE_START)) {
 354                 fprintf(stderr, "in core: %p; in runtime: %p or %p\n",
 355                         (void*)addr,
 356                         (void*)DYNAMIC_0_SPACE_START,
 357                         (void*)DYNAMIC_1_SPACE_START);
 358                 lose("warning: core/runtime address mismatch: DYNAMIC_SPACE_START\n");
 359             }
 360 #endif
 361 #if defined(ALLOCATION_POINTER)
 362             SetSymbolValue(ALLOCATION_POINTER, (lispobj)free_pointer,0);
 363 #else
 364             dynamic_space_free_pointer = free_pointer;
 365 #endif
 366             /* For stop-and-copy GC, this will be whatever the GC was
 367              * using at the time. With GENCGC, this will always be
 368              * space 0. (We checked above that for GENCGC,
 369              * addr==DYNAMIC_SPACE_START.) */
 370             current_dynamic_space = (lispobj *)addr;
 371             break;
 372         case STATIC_CORE_SPACE_ID:
 373             if (addr != (os_vm_address_t)STATIC_SPACE_START) {
 374                 fprintf(stderr, "in core: %p - in runtime: %p\n",
 375                         (void*)addr, (void*)STATIC_SPACE_START);
 376                 lose("core/runtime address mismatch: STATIC_SPACE_START\n");
 377             }
 378             break;
 379         case READ_ONLY_CORE_SPACE_ID:
 380             if (addr != (os_vm_address_t)READ_ONLY_SPACE_START) {
 381                 fprintf(stderr, "in core: %p - in runtime: %p\n",
 382                         (void*)addr, (void*)READ_ONLY_SPACE_START);
 383                 lose("core/runtime address mismatch: READ_ONLY_SPACE_START\n");
 384             }
 385             break;
 386         default:
 387             lose("unknown space ID %ld addr %p\n", id, addr);
 388         }
 389     }
 390 }
 391
 392 lispobj
 393 load_core_file(char *file, os_vm_offset_t file_offset)
 394 {
 395     void *header;
 396 #ifndef LISP_FEATURE_ALPHA
 397     word_t val, *ptr;
 398 #else
 399     u32 val, *ptr;
 400 #endif
 401     os_vm_size_t len, remaining_len;
 402     int fd = open_binary(file, O_RDONLY);
 403     ssize_t count;
 404     lispobj initial_function = NIL;
 405
 406     FSHOW((stderr, "/entering load_core_file(%s)\n", file));
 407     if (fd < 0) {
 408         fprintf(stderr, "could not open file \"%s\"\n", file);
 409         perror("open");
 410         exit(1);
 411     }
 412
 413     lseek(fd, file_offset, SEEK_SET);
 414     header = calloc(os_vm_page_size, 1);
 415
 416     count = read(fd, header, os_vm_page_size);
 417     if (count < (ssize_t) os_vm_page_size) {
 418         lose("premature end of core file\n");
 419     }
 420     SHOW("successfully read first page of core");
 421
 422     ptr = header;
 423     val = *ptr++;
 424
 425     if (val != CORE_MAGIC) {
 426         lose("invalid magic number in core: 0x%lx should have been 0x%x.\n",
 427              val,
 428              CORE_MAGIC);
 429     }
 430     SHOW("found CORE_MAGIC");
 431
 432     while (val != END_CORE_ENTRY_TYPE_CODE) {
 433         val = *ptr++;
 434         len = *ptr++;
 435         remaining_len = len - 2; /* (-2 to cancel the two ++ operations) */
 436         FSHOW((stderr, "/val=0x%"WORD_FMTX", remaining_len=0x%"WORD_FMTX"\n",
 437                val, remaining_len));
 438
 439         switch (val) {
 440
 441         case END_CORE_ENTRY_TYPE_CODE:
 442             SHOW("END_CORE_ENTRY_TYPE_CODE case");
 443             break;
 444
 445         case BUILD_ID_CORE_ENTRY_TYPE_CODE:
 446             SHOW("BUILD_ID_CORE_ENTRY_TYPE_CODE case");
 447             {
 448                 os_vm_size_t i;
 449
 450                 FSHOW((stderr, "build_id[]=\"%s\"\n", build_id));
 451                 FSHOW((stderr, "remaining_len = %d\n", remaining_len));
 452                 if (remaining_len != strlen((const char *)build_id))
 453                     goto losing_build_id;
 454                 for (i = 0; i < remaining_len; ++i) {
 455                     FSHOW((stderr, "ptr[%d] = char = %d, expected=%d\n",
 456                            i, ptr[i], build_id[i]));
 457                     if (ptr[i] != build_id[i])
 458                         goto losing_build_id;
 459                 }
 460                 break;
 461             losing_build_id:
 462                 /* .core files are not binary-compatible between
 463                  * builds because we can't easily detect whether the
 464                  * sources were patched between the time the
 465                  * dumping-the-.core runtime was built and the time
 466                  * that the loading-the-.core runtime was built.
 467                  *
 468                  * (We could easily detect whether version.lisp-expr
 469                  * was changed, but people experimenting with patches
 470                  * don't necessarily update version.lisp-expr.) */
 471
 472                 lose("can't load .core for different runtime, sorry\n");
 473             }
 474
 475         case NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE:
 476             SHOW("NEW_DIRECTORY_CORE_ENTRY_TYPE_CODE case");
 477             process_directory(fd,
 478                               ptr,
 479 #ifndef LISP_FEATURE_ALPHA
 480                               remaining_len / (sizeof(struct ndir_entry) /
 481                                                sizeof(lispobj)),
 482 #else
 483                               remaining_len / (sizeof(struct ndir_entry) /
 484                                                sizeof(u32)),
 485 #endif
 486                               file_offset);
 487             break;
 488
 489         case INITIAL_FUN_CORE_ENTRY_TYPE_CODE:
 490             SHOW("INITIAL_FUN_CORE_ENTRY_TYPE_CODE case");
 491             initial_function = (lispobj)*ptr;
 492             break;
 493
 494 #ifdef LISP_FEATURE_GENCGC
 495         case PAGE_TABLE_CORE_ENTRY_TYPE_CODE:
 496         {
 497             os_vm_size_t size = *ptr;
 498             os_vm_size_t fdoffset = (*(ptr+1) + 1) * (os_vm_page_size);
 499             page_index_t offset = 0;
 500             ssize_t bytes_read;
 501             word_t data[4096];
 502             word_t word;
 503             lseek(fd, fdoffset + file_offset, SEEK_SET);
 504             while ((bytes_read = read(fd, data, (size < 4096 ? size : 4096 )))
 505                     > 0)
 506             {
 507                 int i = 0;
 508                 size -= bytes_read;
 509                 while (bytes_read) {
 510                     bytes_read -= sizeof(word_t);
 511                     /* Ignore all zeroes. The size of the page table
 512                      * core entry was rounded up to os_vm_page_size
 513                      * during the save, and might now have more
 514                      * elements than the page table.
 515                      *
 516                      * The low bits of each word are allocation flags.
 517                      */
 518                     if ((word=data[i])) {
 519                         page_table[offset].scan_start_offset = word & ~0x03;
 520                         page_table[offset].allocated = word & 0x03;
 521                     }
 522                     i++;
 523                     offset++;
 524                 }
 525             }
 526
 527             gencgc_partial_pickup = 1;
 528             break;
 529         }
 530 #endif
 531         default:
 532             lose("unknown core file entry: 0x%"WORD_FMTX"\n", val);
 533         }
 534
 535         ptr += remaining_len;
 536         FSHOW((stderr, "/new ptr=0x%"WORD_FMTX"\n", ptr));
 537     }
 538     SHOW("about to free(header)");
 539     free(header);
 540     SHOW("returning from load_core_file(..)");
 541     return initial_function;
 542 }
 543
 544 #include "genesis/hash-table.h"
 545 #include "genesis/vector.h"
 546 os_vm_address_t get_asm_routine_by_name(const char* name)
 547 {
 548     lispobj routines = SYMBOL(ASSEMBLER_ROUTINES)->value;
 549     if (lowtag_of(routines) == INSTANCE_POINTER_LOWTAG) {
 550         struct hash_table* ht = (struct hash_table*)native_pointer(routines);
 551         struct vector* table = (struct vector*)native_pointer(ht->table);
 552         lispobj sym;
 553         int i;
 554         for (i=2 ; i < fixnum_value(table->length) ; i += 2) {
 555           sym = table->data[i];
 556           if (lowtag_of(sym) == OTHER_POINTER_LOWTAG
 557               && widetag_of(SYMBOL(sym)->header) == SYMBOL_HEADER_WIDETAG
 558               && !strcmp(name,
 559                          (char*)((struct vector*)
 560                                  native_pointer(SYMBOL(sym)->name))->data))
 561               return (os_vm_address_t)fixnum_value(table->data[i+1]);
 562         }
 563         // Something is wrong if we have a hashtable but find nothing.
 564         fprintf(stderr, "WARNING: get_asm_routine_by_name(%s) failed\n",
 565                 name);
 566     }
 567     return NULL;
 568 }
 569
 570 void asm_routine_poke(const char* routine, int offset, char byte)
 571 {
 572     os_vm_address_t address = get_asm_routine_by_name(routine);
 573     if (address)
 574         address[offset] = byte;
 575 }