archive-tar.c

   1 /*
   2  * Copyright (c) 2005, 2006 Rene Scharfe
   3  */
   4 #include "git-compat-util.h"
   5 #include "config.h"
   6 #include "gettext.h"
   7 #include "git-zlib.h"
   8 #include "hex.h"
   9 #include "tar.h"
  10 #include "archive.h"
  11 #include "object-store-ll.h"
  12 #include "strbuf.h"
  13 #include "streaming.h"
  14 #include "run-command.h"
  15 #include "write-or-die.h"
  16
  17 #define RECORDSIZE      (512)
  18 #define BLOCKSIZE       (RECORDSIZE * 20)
  19
  20 static char block[BLOCKSIZE];
  21 static unsigned long offset;
  22
  23 static int tar_umask = 002;
  24
  25 static int write_tar_filter_archive(const struct archiver *ar,
  26                                     struct archiver_args *args);
  27
  28 /*
  29  * This is the max value that a ustar size header can specify, as it is fixed
  30  * at 11 octal digits. POSIX specifies that we switch to extended headers at
  31  * this size.
  32  *
  33  * Likewise for the mtime (which happens to use a buffer of the same size).
  34  */
  35 #if ULONG_MAX == 0xFFFFFFFF
  36 #define USTAR_MAX_SIZE ULONG_MAX
  37 #else
  38 #define USTAR_MAX_SIZE 077777777777UL
  39 #endif
  40 #if TIME_MAX == 0xFFFFFFFF
  41 #define USTAR_MAX_MTIME TIME_MAX
  42 #else
  43 #define USTAR_MAX_MTIME 077777777777ULL
  44 #endif
  45
  46 static void tar_write_block(const void *buf)
  47 {
  48         write_or_die(1, buf, BLOCKSIZE);
  49 }
  50
  51 static void (*write_block)(const void *) = tar_write_block;
  52
  53 /* writes out the whole block, but only if it is full */
  54 static void write_if_needed(void)
  55 {
  56         if (offset == BLOCKSIZE) {
  57                 write_block(block);
  58                 offset = 0;
  59         }
  60 }
  61
  62 /*
  63  * queues up writes, so that all our write(2) calls write exactly one
  64  * full block; pads writes to RECORDSIZE
  65  */
  66 static void do_write_blocked(const void *data, unsigned long size)
  67 {
  68         const char *buf = data;
  69
  70         if (offset) {
  71                 unsigned long chunk = BLOCKSIZE - offset;
  72                 if (size < chunk)
  73                         chunk = size;
  74                 memcpy(block + offset, buf, chunk);
  75                 size -= chunk;
  76                 offset += chunk;
  77                 buf += chunk;
  78                 write_if_needed();
  79         }
  80         while (size >= BLOCKSIZE) {
  81                 write_block(buf);
  82                 size -= BLOCKSIZE;
  83                 buf += BLOCKSIZE;
  84         }
  85         if (size) {
  86                 memcpy(block + offset, buf, size);
  87                 offset += size;
  88         }
  89 }
  90
  91 static void finish_record(void)
  92 {
  93         unsigned long tail;
  94         tail = offset % RECORDSIZE;
  95         if (tail)  {
  96                 memset(block + offset, 0, RECORDSIZE - tail);
  97                 offset += RECORDSIZE - tail;
  98         }
  99         write_if_needed();
 100 }
 101
 102 static void write_blocked(const void *data, unsigned long size)
 103 {
 104         do_write_blocked(data, size);
 105         finish_record();
 106 }
 107
 108 /*
 109  * The end of tar archives is marked by 2*512 nul bytes and after that
 110  * follows the rest of the block (if any).
 111  */
 112 static void write_trailer(void)
 113 {
 114         int tail = BLOCKSIZE - offset;
 115         memset(block + offset, 0, tail);
 116         write_block(block);
 117         if (tail < 2 * RECORDSIZE) {
 118                 memset(block, 0, offset);
 119                 write_block(block);
 120         }
 121 }
 122
 123 /*
 124  * queues up writes, so that all our write(2) calls write exactly one
 125  * full block; pads writes to RECORDSIZE
 126  */
 127 static int stream_blocked(struct repository *r, const struct object_id *oid)
 128 {
 129         struct git_istream *st;
 130         enum object_type type;
 131         unsigned long sz;
 132         char buf[BLOCKSIZE];
 133         ssize_t readlen;
 134
 135         st = open_istream(r, oid, &type, &sz, NULL);
 136         if (!st)
 137                 return error(_("cannot stream blob %s"), oid_to_hex(oid));
 138         for (;;) {
 139                 readlen = read_istream(st, buf, sizeof(buf));
 140                 if (readlen <= 0)
 141                         break;
 142                 do_write_blocked(buf, readlen);
 143         }
 144         close_istream(st);
 145         if (!readlen)
 146                 finish_record();
 147         return readlen;
 148 }
 149
 150 /*
 151  * pax extended header records have the format "%u %s=%s\n".  %u contains
 152  * the size of the whole string (including the %u), the first %s is the
 153  * keyword, the second one is the value.  This function constructs such a
 154  * string and appends it to a struct strbuf.
 155  */
 156 static void strbuf_append_ext_header(struct strbuf *sb, const char *keyword,
 157                                      const char *value, size_t valuelen)
 158 {
 159         size_t orig_len = sb->len;
 160         size_t len, tmp;
 161
 162         /* "%u %s=%s\n" */
 163         len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 164         for (tmp = 1; len / 10 >= tmp; tmp *= 10)
 165                 len++;
 166
 167         strbuf_grow(sb, len);
 168         strbuf_addf(sb, "%"PRIuMAX" %s=", (uintmax_t)len, keyword);
 169         strbuf_add(sb, value, valuelen);
 170         strbuf_addch(sb, '\n');
 171
 172         if (len != sb->len - orig_len)
 173                 BUG("pax extended header length miscalculated as %"PRIuMAX
 174                     ", should be %"PRIuMAX,
 175                     (uintmax_t)len, (uintmax_t)(sb->len - orig_len));
 176 }
 177
 178 /*
 179  * Like strbuf_append_ext_header, but for numeric values.
 180  */
 181 static void strbuf_append_ext_header_uint(struct strbuf *sb,
 182                                           const char *keyword,
 183                                           uintmax_t value)
 184 {
 185         char buf[40]; /* big enough for 2^128 in decimal, plus NUL */
 186         int len;
 187
 188         len = xsnprintf(buf, sizeof(buf), "%"PRIuMAX, value);
 189         strbuf_append_ext_header(sb, keyword, buf, len);
 190 }
 191
 192 static unsigned int ustar_header_chksum(const struct ustar_header *header)
 193 {
 194         const unsigned char *p = (const unsigned char *)header;
 195         unsigned int chksum = 0;
 196         while (p < (const unsigned char *)header->chksum)
 197                 chksum += *p++;
 198         chksum += sizeof(header->chksum) * ' ';
 199         p += sizeof(header->chksum);
 200         while (p < (const unsigned char *)header + sizeof(struct ustar_header))
 201                 chksum += *p++;
 202         return chksum;
 203 }
 204
 205 static size_t get_path_prefix(const char *path, size_t pathlen, size_t maxlen)
 206 {
 207         size_t i = pathlen;
 208         if (i > 1 && path[i - 1] == '/')
 209                 i--;
 210         if (i > maxlen)
 211                 i = maxlen;
 212         do {
 213                 i--;
 214         } while (i > 0 && path[i] != '/');
 215         return i;
 216 }
 217
 218 static void prepare_header(struct archiver_args *args,
 219                            struct ustar_header *header,
 220                            unsigned int mode, unsigned long size)
 221 {
 222         xsnprintf(header->mode, sizeof(header->mode), "%07o", mode & 07777);
 223         xsnprintf(header->size, sizeof(header->size), "%011"PRIoMAX , S_ISREG(mode) ? (uintmax_t)size : (uintmax_t)0);
 224         xsnprintf(header->mtime, sizeof(header->mtime), "%011lo", (unsigned long) args->time);
 225
 226         xsnprintf(header->uid, sizeof(header->uid), "%07o", 0);
 227         xsnprintf(header->gid, sizeof(header->gid), "%07o", 0);
 228         strlcpy(header->uname, "root", sizeof(header->uname));
 229         strlcpy(header->gname, "root", sizeof(header->gname));
 230         xsnprintf(header->devmajor, sizeof(header->devmajor), "%07o", 0);
 231         xsnprintf(header->devminor, sizeof(header->devminor), "%07o", 0);
 232
 233         memcpy(header->magic, "ustar", 6);
 234         memcpy(header->version, "00", 2);
 235
 236         xsnprintf(header->chksum, sizeof(header->chksum), "%07o", ustar_header_chksum(header));
 237 }
 238
 239 static void write_extended_header(struct archiver_args *args,
 240                                   const struct object_id *oid,
 241                                   const void *buffer, unsigned long size)
 242 {
 243         struct ustar_header header;
 244         unsigned int mode;
 245         memset(&header, 0, sizeof(header));
 246         *header.typeflag = TYPEFLAG_EXT_HEADER;
 247         mode = 0100666;
 248         xsnprintf(header.name, sizeof(header.name), "%s.paxheader", oid_to_hex(oid));
 249         prepare_header(args, &header, mode, size);
 250         write_blocked(&header, sizeof(header));
 251         write_blocked(buffer, size);
 252 }
 253
 254 static int write_tar_entry(struct archiver_args *args,
 255                            const struct object_id *oid,
 256                            const char *path, size_t pathlen,
 257                            unsigned int mode,
 258                            void *buffer, unsigned long size)
 259 {
 260         struct ustar_header header;
 261         struct strbuf ext_header = STRBUF_INIT;
 262         unsigned long size_in_header;
 263         int err = 0;
 264
 265         memset(&header, 0, sizeof(header));
 266
 267         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 268                 *header.typeflag = TYPEFLAG_DIR;
 269                 mode = (mode | 0777) & ~tar_umask;
 270         } else if (S_ISLNK(mode)) {
 271                 *header.typeflag = TYPEFLAG_LNK;
 272                 mode |= 0777;
 273         } else if (S_ISREG(mode)) {
 274                 *header.typeflag = TYPEFLAG_REG;
 275                 mode = (mode | ((mode & 0100) ? 0777 : 0666)) & ~tar_umask;
 276         } else {
 277                 return error(_("unsupported file mode: 0%o (SHA1: %s)"),
 278                              mode, oid_to_hex(oid));
 279         }
 280         if (pathlen > sizeof(header.name)) {
 281                 size_t plen = get_path_prefix(path, pathlen,
 282                                               sizeof(header.prefix));
 283                 size_t rest = pathlen - plen - 1;
 284                 if (plen > 0 && rest <= sizeof(header.name)) {
 285                         memcpy(header.prefix, path, plen);
 286                         memcpy(header.name, path + plen + 1, rest);
 287                 } else {
 288                         xsnprintf(header.name, sizeof(header.name), "%s.data",
 289                                   oid_to_hex(oid));
 290                         strbuf_append_ext_header(&ext_header, "path",
 291                                                  path, pathlen);
 292                 }
 293         } else
 294                 memcpy(header.name, path, pathlen);
 295
 296         if (S_ISLNK(mode)) {
 297                 if (size > sizeof(header.linkname)) {
 298                         xsnprintf(header.linkname, sizeof(header.linkname),
 299                                   "see %s.paxheader", oid_to_hex(oid));
 300                         strbuf_append_ext_header(&ext_header, "linkpath",
 301                                                  buffer, size);
 302                 } else
 303                         memcpy(header.linkname, buffer, size);
 304         }
 305
 306         size_in_header = size;
 307         if (S_ISREG(mode) && size > USTAR_MAX_SIZE) {
 308                 size_in_header = 0;
 309                 strbuf_append_ext_header_uint(&ext_header, "size", size);
 310         }
 311
 312         prepare_header(args, &header, mode, size_in_header);
 313
 314         if (ext_header.len > 0) {
 315                 write_extended_header(args, oid, ext_header.buf,
 316                                       ext_header.len);
 317         }
 318         strbuf_release(&ext_header);
 319         write_blocked(&header, sizeof(header));
 320         if (S_ISREG(mode) && size > 0) {
 321                 if (buffer)
 322                         write_blocked(buffer, size);
 323                 else
 324                         err = stream_blocked(args->repo, oid);
 325         }
 326         return err;
 327 }
 328
 329 static void write_global_extended_header(struct archiver_args *args)
 330 {
 331         const struct object_id *oid = args->commit_oid;
 332         struct strbuf ext_header = STRBUF_INIT;
 333         struct ustar_header header;
 334         unsigned int mode;
 335
 336         if (oid)
 337                 strbuf_append_ext_header(&ext_header, "comment",
 338                                          oid_to_hex(oid),
 339                                          the_hash_algo->hexsz);
 340         if (args->time > USTAR_MAX_MTIME) {
 341                 strbuf_append_ext_header_uint(&ext_header, "mtime",
 342                                               args->time);
 343                 args->time = USTAR_MAX_MTIME;
 344         }
 345
 346         if (!ext_header.len)
 347                 return;
 348
 349         memset(&header, 0, sizeof(header));
 350         *header.typeflag = TYPEFLAG_GLOBAL_HEADER;
 351         mode = 0100666;
 352         xsnprintf(header.name, sizeof(header.name), "pax_global_header");
 353         prepare_header(args, &header, mode, ext_header.len);
 354         write_blocked(&header, sizeof(header));
 355         write_blocked(ext_header.buf, ext_header.len);
 356         strbuf_release(&ext_header);
 357 }
 358
 359 static struct archiver **tar_filters;
 360 static int nr_tar_filters;
 361 static int alloc_tar_filters;
 362
 363 static struct archiver *find_tar_filter(const char *name, size_t len)
 364 {
 365         int i;
 366         for (i = 0; i < nr_tar_filters; i++) {
 367                 struct archiver *ar = tar_filters[i];
 368                 if (!strncmp(ar->name, name, len) && !ar->name[len])
 369                         return ar;
 370         }
 371         return NULL;
 372 }
 373
 374 static int tar_filter_config(const char *var, const char *value,
 375                              void *data UNUSED)
 376 {
 377         struct archiver *ar;
 378         const char *name;
 379         const char *type;
 380         size_t namelen;
 381
 382         if (parse_config_key(var, "tar", &name, &namelen, &type) < 0 || !name)
 383                 return 0;
 384
 385         ar = find_tar_filter(name, namelen);
 386         if (!ar) {
 387                 CALLOC_ARRAY(ar, 1);
 388                 ar->name = xmemdupz(name, namelen);
 389                 ar->write_archive = write_tar_filter_archive;
 390                 ar->flags = ARCHIVER_WANT_COMPRESSION_LEVELS |
 391                             ARCHIVER_HIGH_COMPRESSION_LEVELS;
 392                 ALLOC_GROW(tar_filters, nr_tar_filters + 1, alloc_tar_filters);
 393                 tar_filters[nr_tar_filters++] = ar;
 394         }
 395
 396         if (!strcmp(type, "command")) {
 397                 if (!value)
 398                         return config_error_nonbool(var);
 399                 free(ar->filter_command);
 400                 ar->filter_command = xstrdup(value);
 401                 return 0;
 402         }
 403         if (!strcmp(type, "remote")) {
 404                 if (git_config_bool(var, value))
 405                         ar->flags |= ARCHIVER_REMOTE;
 406                 else
 407                         ar->flags &= ~ARCHIVER_REMOTE;
 408                 return 0;
 409         }
 410
 411         return 0;
 412 }
 413
 414 static int git_tar_config(const char *var, const char *value,
 415                           const struct config_context *ctx, void *cb)
 416 {
 417         if (!strcmp(var, "tar.umask")) {
 418                 if (value && !strcmp(value, "user")) {
 419                         tar_umask = umask(0);
 420                         umask(tar_umask);
 421                 } else {
 422                         tar_umask = git_config_int(var, value, ctx->kvi);
 423                 }
 424                 return 0;
 425         }
 426
 427         return tar_filter_config(var, value, cb);
 428 }
 429
 430 static int write_tar_archive(const struct archiver *ar UNUSED,
 431                              struct archiver_args *args)
 432 {
 433         int err = 0;
 434
 435         write_global_extended_header(args);
 436         err = write_archive_entries(args, write_tar_entry);
 437         if (!err)
 438                 write_trailer();
 439         return err;
 440 }
 441
 442 static git_zstream gzstream;
 443 static unsigned char outbuf[16384];
 444
 445 static void tgz_deflate(int flush)
 446 {
 447         while (gzstream.avail_in || flush == Z_FINISH) {
 448                 int status = git_deflate(&gzstream, flush);
 449                 if (!gzstream.avail_out || status == Z_STREAM_END) {
 450                         write_or_die(1, outbuf, gzstream.next_out - outbuf);
 451                         gzstream.next_out = outbuf;
 452                         gzstream.avail_out = sizeof(outbuf);
 453                         if (status == Z_STREAM_END)
 454                                 break;
 455                 }
 456                 if (status != Z_OK && status != Z_BUF_ERROR)
 457                         die(_("deflate error (%d)"), status);
 458         }
 459 }
 460
 461 static void tgz_write_block(const void *data)
 462 {
 463         gzstream.next_in = (void *)data;
 464         gzstream.avail_in = BLOCKSIZE;
 465         tgz_deflate(Z_NO_FLUSH);
 466 }
 467
 468 static const char internal_gzip_command[] = "git archive gzip";
 469
 470 static int write_tar_filter_archive(const struct archiver *ar,
 471                                     struct archiver_args *args)
 472 {
 473 #if ZLIB_VERNUM >= 0x1221
 474         struct gz_header_s gzhead = { .os = 3 }; /* Unix, for reproducibility */
 475 #endif
 476         struct strbuf cmd = STRBUF_INIT;
 477         struct child_process filter = CHILD_PROCESS_INIT;
 478         int r;
 479
 480         if (!ar->filter_command)
 481                 BUG("tar-filter archiver called with no filter defined");
 482
 483         if (!strcmp(ar->filter_command, internal_gzip_command)) {
 484                 write_block = tgz_write_block;
 485                 git_deflate_init_gzip(&gzstream, args->compression_level);
 486 #if ZLIB_VERNUM >= 0x1221
 487                 if (deflateSetHeader(&gzstream.z, &gzhead) != Z_OK)
 488                         BUG("deflateSetHeader() called too late");
 489 #endif
 490                 gzstream.next_out = outbuf;
 491                 gzstream.avail_out = sizeof(outbuf);
 492
 493                 r = write_tar_archive(ar, args);
 494
 495                 tgz_deflate(Z_FINISH);
 496                 git_deflate_end(&gzstream);
 497                 return r;
 498         }
 499
 500         strbuf_addstr(&cmd, ar->filter_command);
 501         if (args->compression_level >= 0)
 502                 strbuf_addf(&cmd, " -%d", args->compression_level);
 503
 504         strvec_push(&filter.args, cmd.buf);
 505         filter.use_shell = 1;
 506         filter.in = -1;
 507         filter.silent_exec_failure = 1;
 508
 509         if (start_command(&filter) < 0)
 510                 die_errno(_("unable to start '%s' filter"), cmd.buf);
 511         close(1);
 512         if (dup2(filter.in, 1) < 0)
 513                 die_errno(_("unable to redirect descriptor"));
 514         close(filter.in);
 515
 516         r = write_tar_archive(ar, args);
 517
 518         close(1);
 519         if (finish_command(&filter) != 0)
 520                 die(_("'%s' filter reported error"), cmd.buf);
 521
 522         strbuf_release(&cmd);
 523         return r;
 524 }
 525
 526 static struct archiver tar_archiver = {
 527         .name = "tar",
 528         .write_archive = write_tar_archive,
 529         .flags = ARCHIVER_REMOTE,
 530 };
 531
 532 void init_tar_archiver(void)
 533 {
 534         int i;
 535         register_archiver(&tar_archiver);
 536
 537         tar_filter_config("tar.tgz.command", internal_gzip_command, NULL);
 538         tar_filter_config("tar.tgz.remote", "true", NULL);
 539         tar_filter_config("tar.tar.gz.command", internal_gzip_command, NULL);
 540         tar_filter_config("tar.tar.gz.remote", "true", NULL);
 541         git_config(git_tar_config, NULL);
 542         for (i = 0; i < nr_tar_filters; i++) {
 543                 /* omit any filters that never had a command configured */
 544                 if (tar_filters[i]->filter_command)
 545                         register_archiver(tar_filters[i]);
 546         }
 547 }