archive-tar.c

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