archive-tar.c

   1 /*
   2  * Copyright (c) 2005, 2006 Rene Scharfe
   3  */
   4 #include "cache.h"
   5 #include "tar.h"
   6 #include "archive.h"
   7 #include "streaming.h"
   8 #include "run-command.h"
   9
  10 #define RECORDSIZE      (512)
  11 #define BLOCKSIZE       (RECORDSIZE * 20)
  12
  13 static char block[BLOCKSIZE];
  14 static unsigned long offset;
  15
  16 static int tar_umask = 002;
  17
  18 static int write_tar_filter_archive(const struct archiver *ar,
  19                                     struct archiver_args *args);
  20
  21 /* writes out the whole block, but only if it is full */
  22 static void write_if_needed(void)
  23 {
  24         if (offset == BLOCKSIZE) {
  25                 write_or_die(1, block, BLOCKSIZE);
  26                 offset = 0;
  27         }
  28 }
  29
  30 /*
  31  * queues up writes, so that all our write(2) calls write exactly one
  32  * full block; pads writes to RECORDSIZE
  33  */
  34 static void do_write_blocked(const void *data, unsigned long size)
  35 {
  36         const char *buf = data;
  37
  38         if (offset) {
  39                 unsigned long chunk = BLOCKSIZE - offset;
  40                 if (size < chunk)
  41                         chunk = size;
  42                 memcpy(block + offset, buf, chunk);
  43                 size -= chunk;
  44                 offset += chunk;
  45                 buf += chunk;
  46                 write_if_needed();
  47         }
  48         while (size >= BLOCKSIZE) {
  49                 write_or_die(1, buf, BLOCKSIZE);
  50                 size -= BLOCKSIZE;
  51                 buf += BLOCKSIZE;
  52         }
  53         if (size) {
  54                 memcpy(block + offset, buf, size);
  55                 offset += size;
  56         }
  57 }
  58
  59 static void finish_record(void)
  60 {
  61         unsigned long tail;
  62         tail = offset % RECORDSIZE;
  63         if (tail)  {
  64                 memset(block + offset, 0, RECORDSIZE - tail);
  65                 offset += RECORDSIZE - tail;
  66         }
  67         write_if_needed();
  68 }
  69
  70 static void write_blocked(const void *data, unsigned long size)
  71 {
  72         do_write_blocked(data, size);
  73         finish_record();
  74 }
  75
  76 /*
  77  * The end of tar archives is marked by 2*512 nul bytes and after that
  78  * follows the rest of the block (if any).
  79  */
  80 static void write_trailer(void)
  81 {
  82         int tail = BLOCKSIZE - offset;
  83         memset(block + offset, 0, tail);
  84         write_or_die(1, block, BLOCKSIZE);
  85         if (tail < 2 * RECORDSIZE) {
  86                 memset(block, 0, offset);
  87                 write_or_die(1, block, BLOCKSIZE);
  88         }
  89 }
  90
  91 /*
  92  * queues up writes, so that all our write(2) calls write exactly one
  93  * full block; pads writes to RECORDSIZE
  94  */
  95 static int stream_blocked(const unsigned char *sha1)
  96 {
  97         struct git_istream *st;
  98         enum object_type type;
  99         unsigned long sz;
 100         char buf[BLOCKSIZE];
 101         ssize_t readlen;
 102
 103         st = open_istream(sha1, &type, &sz, NULL);
 104         if (!st)
 105                 return error("cannot stream blob %s", sha1_to_hex(sha1));
 106         for (;;) {
 107                 readlen = read_istream(st, buf, sizeof(buf));
 108                 if (readlen <= 0)
 109                         break;
 110                 do_write_blocked(buf, readlen);
 111         }
 112         close_istream(st);
 113         if (!readlen)
 114                 finish_record();
 115         return readlen;
 116 }
 117
 118 /*
 119  * pax extended header records have the format "%u %s=%s\n".  %u contains
 120  * the size of the whole string (including the %u), the first %s is the
 121  * keyword, the second one is the value.  This function constructs such a
 122  * string and appends it to a struct strbuf.
 123  */
 124 static void strbuf_append_ext_header(struct strbuf *sb, const char *keyword,
 125                                      const char *value, unsigned int valuelen)
 126 {
 127         int len, tmp;
 128
 129         /* "%u %s=%s\n" */
 130         len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 131         for (tmp = len; tmp > 9; tmp /= 10)
 132                 len++;
 133
 134         strbuf_grow(sb, len);
 135         strbuf_addf(sb, "%u %s=", len, keyword);
 136         strbuf_add(sb, value, valuelen);
 137         strbuf_addch(sb, '\n');
 138 }
 139
 140 static unsigned int ustar_header_chksum(const struct ustar_header *header)
 141 {
 142         const unsigned char *p = (const unsigned char *)header;
 143         unsigned int chksum = 0;
 144         while (p < (const unsigned char *)header->chksum)
 145                 chksum += *p++;
 146         chksum += sizeof(header->chksum) * ' ';
 147         p += sizeof(header->chksum);
 148         while (p < (const unsigned char *)header + sizeof(struct ustar_header))
 149                 chksum += *p++;
 150         return chksum;
 151 }
 152
 153 static size_t get_path_prefix(const char *path, size_t pathlen, size_t maxlen)
 154 {
 155         size_t i = pathlen;
 156         if (i > maxlen)
 157                 i = maxlen;
 158         do {
 159                 i--;
 160         } while (i > 0 && path[i] != '/');
 161         return i;
 162 }
 163
 164 static void prepare_header(struct archiver_args *args,
 165                            struct ustar_header *header,
 166                            unsigned int mode, unsigned long size)
 167 {
 168         sprintf(header->mode, "%07o", mode & 07777);
 169         sprintf(header->size, "%011lo", S_ISREG(mode) ? size : 0);
 170         sprintf(header->mtime, "%011lo", (unsigned long) args->time);
 171
 172         sprintf(header->uid, "%07o", 0);
 173         sprintf(header->gid, "%07o", 0);
 174         strlcpy(header->uname, "root", sizeof(header->uname));
 175         strlcpy(header->gname, "root", sizeof(header->gname));
 176         sprintf(header->devmajor, "%07o", 0);
 177         sprintf(header->devminor, "%07o", 0);
 178
 179         memcpy(header->magic, "ustar", 6);
 180         memcpy(header->version, "00", 2);
 181
 182         sprintf(header->chksum, "%07o", ustar_header_chksum(header));
 183 }
 184
 185 static int write_extended_header(struct archiver_args *args,
 186                                  const unsigned char *sha1,
 187                                  const void *buffer, unsigned long size)
 188 {
 189         struct ustar_header header;
 190         unsigned int mode;
 191         memset(&header, 0, sizeof(header));
 192         *header.typeflag = TYPEFLAG_EXT_HEADER;
 193         mode = 0100666;
 194         sprintf(header.name, "%s.paxheader", sha1_to_hex(sha1));
 195         prepare_header(args, &header, mode, size);
 196         write_blocked(&header, sizeof(header));
 197         write_blocked(buffer, size);
 198         return 0;
 199 }
 200
 201 static int write_tar_entry(struct archiver_args *args,
 202                            const unsigned char *sha1,
 203                            const char *path, size_t pathlen,
 204                            unsigned int mode)
 205 {
 206         struct ustar_header header;
 207         struct strbuf ext_header = STRBUF_INIT;
 208         unsigned int old_mode = mode;
 209         unsigned long size;
 210         void *buffer;
 211         int err = 0;
 212
 213         memset(&header, 0, sizeof(header));
 214
 215         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 216                 *header.typeflag = TYPEFLAG_DIR;
 217                 mode = (mode | 0777) & ~tar_umask;
 218         } else if (S_ISLNK(mode)) {
 219                 *header.typeflag = TYPEFLAG_LNK;
 220                 mode |= 0777;
 221         } else if (S_ISREG(mode)) {
 222                 *header.typeflag = TYPEFLAG_REG;
 223                 mode = (mode | ((mode & 0100) ? 0777 : 0666)) & ~tar_umask;
 224         } else {
 225                 return error("unsupported file mode: 0%o (SHA1: %s)",
 226                              mode, sha1_to_hex(sha1));
 227         }
 228         if (pathlen > sizeof(header.name)) {
 229                 size_t plen = get_path_prefix(path, pathlen,
 230                                               sizeof(header.prefix));
 231                 size_t rest = pathlen - plen - 1;
 232                 if (plen > 0 && rest <= sizeof(header.name)) {
 233                         memcpy(header.prefix, path, plen);
 234                                 memcpy(header.name, path + plen + 1, rest);
 235                 } else {
 236                         sprintf(header.name, "%s.data",
 237                                 sha1_to_hex(sha1));
 238                         strbuf_append_ext_header(&ext_header, "path",
 239                                                  path, pathlen);
 240                 }
 241         } else
 242                 memcpy(header.name, path, pathlen);
 243
 244         if (S_ISREG(mode) && !args->convert &&
 245             sha1_object_info(sha1, &size) == OBJ_BLOB &&
 246             size > big_file_threshold)
 247                 buffer = NULL;
 248         else if (S_ISLNK(mode) || S_ISREG(mode)) {
 249                 enum object_type type;
 250                 buffer = sha1_file_to_archive(args, path, sha1, old_mode, &type, &size);
 251                 if (!buffer)
 252                         return error("cannot read %s", sha1_to_hex(sha1));
 253         } else {
 254                 buffer = NULL;
 255                 size = 0;
 256         }
 257
 258         if (S_ISLNK(mode)) {
 259                 if (size > sizeof(header.linkname)) {
 260                         sprintf(header.linkname, "see %s.paxheader",
 261                                 sha1_to_hex(sha1));
 262                         strbuf_append_ext_header(&ext_header, "linkpath",
 263                                                  buffer, size);
 264                 } else
 265                         memcpy(header.linkname, buffer, size);
 266         }
 267
 268         prepare_header(args, &header, mode, size);
 269
 270         if (ext_header.len > 0) {
 271                 err = write_extended_header(args, sha1, ext_header.buf,
 272                                             ext_header.len);
 273                 if (err) {
 274                         free(buffer);
 275                         return err;
 276                 }
 277         }
 278         strbuf_release(&ext_header);
 279         write_blocked(&header, sizeof(header));
 280         if (S_ISREG(mode) && size > 0) {
 281                 if (buffer)
 282                         write_blocked(buffer, size);
 283                 else
 284                         err = stream_blocked(sha1);
 285         }
 286         free(buffer);
 287         return err;
 288 }
 289
 290 static int write_global_extended_header(struct archiver_args *args)
 291 {
 292         const unsigned char *sha1 = args->commit_sha1;
 293         struct strbuf ext_header = STRBUF_INIT;
 294         struct ustar_header header;
 295         unsigned int mode;
 296         int err = 0;
 297
 298         strbuf_append_ext_header(&ext_header, "comment", sha1_to_hex(sha1), 40);
 299         memset(&header, 0, sizeof(header));
 300         *header.typeflag = TYPEFLAG_GLOBAL_HEADER;
 301         mode = 0100666;
 302         strcpy(header.name, "pax_global_header");
 303         prepare_header(args, &header, mode, ext_header.len);
 304         write_blocked(&header, sizeof(header));
 305         write_blocked(ext_header.buf, ext_header.len);
 306         strbuf_release(&ext_header);
 307         return err;
 308 }
 309
 310 static struct archiver **tar_filters;
 311 static int nr_tar_filters;
 312 static int alloc_tar_filters;
 313
 314 static struct archiver *find_tar_filter(const char *name, int len)
 315 {
 316         int i;
 317         for (i = 0; i < nr_tar_filters; i++) {
 318                 struct archiver *ar = tar_filters[i];
 319                 if (!strncmp(ar->name, name, len) && !ar->name[len])
 320                         return ar;
 321         }
 322         return NULL;
 323 }
 324
 325 static int tar_filter_config(const char *var, const char *value, void *data)
 326 {
 327         struct archiver *ar;
 328         const char *dot;
 329         const char *name;
 330         const char *type;
 331         int namelen;
 332
 333         if (prefixcmp(var, "tar."))
 334                 return 0;
 335         dot = strrchr(var, '.');
 336         if (dot == var + 9)
 337                 return 0;
 338
 339         name = var + 4;
 340         namelen = dot - name;
 341         type = dot + 1;
 342
 343         ar = find_tar_filter(name, namelen);
 344         if (!ar) {
 345                 ar = xcalloc(1, sizeof(*ar));
 346                 ar->name = xmemdupz(name, namelen);
 347                 ar->write_archive = write_tar_filter_archive;
 348                 ar->flags = ARCHIVER_WANT_COMPRESSION_LEVELS;
 349                 ALLOC_GROW(tar_filters, nr_tar_filters + 1, alloc_tar_filters);
 350                 tar_filters[nr_tar_filters++] = ar;
 351         }
 352
 353         if (!strcmp(type, "command")) {
 354                 if (!value)
 355                         return config_error_nonbool(var);
 356                 free(ar->data);
 357                 ar->data = xstrdup(value);
 358                 return 0;
 359         }
 360         if (!strcmp(type, "remote")) {
 361                 if (git_config_bool(var, value))
 362                         ar->flags |= ARCHIVER_REMOTE;
 363                 else
 364                         ar->flags &= ~ARCHIVER_REMOTE;
 365                 return 0;
 366         }
 367
 368         return 0;
 369 }
 370
 371 static int git_tar_config(const char *var, const char *value, void *cb)
 372 {
 373         if (!strcmp(var, "tar.umask")) {
 374                 if (value && !strcmp(value, "user")) {
 375                         tar_umask = umask(0);
 376                         umask(tar_umask);
 377                 } else {
 378                         tar_umask = git_config_int(var, value);
 379                 }
 380                 return 0;
 381         }
 382
 383         return tar_filter_config(var, value, cb);
 384 }
 385
 386 static int write_tar_archive(const struct archiver *ar,
 387                              struct archiver_args *args)
 388 {
 389         int err = 0;
 390
 391         if (args->commit_sha1)
 392                 err = write_global_extended_header(args);
 393         if (!err)
 394                 err = write_archive_entries(args, write_tar_entry);
 395         if (!err)
 396                 write_trailer();
 397         return err;
 398 }
 399
 400 static int write_tar_filter_archive(const struct archiver *ar,
 401                                     struct archiver_args *args)
 402 {
 403         struct strbuf cmd = STRBUF_INIT;
 404         struct child_process filter;
 405         const char *argv[2];
 406         int r;
 407
 408         if (!ar->data)
 409                 die("BUG: tar-filter archiver called with no filter defined");
 410
 411         strbuf_addstr(&cmd, ar->data);
 412         if (args->compression_level >= 0)
 413                 strbuf_addf(&cmd, " -%d", args->compression_level);
 414
 415         memset(&filter, 0, sizeof(filter));
 416         argv[0] = cmd.buf;
 417         argv[1] = NULL;
 418         filter.argv = argv;
 419         filter.use_shell = 1;
 420         filter.in = -1;
 421
 422         if (start_command(&filter) < 0)
 423                 die_errno("unable to start '%s' filter", argv[0]);
 424         close(1);
 425         if (dup2(filter.in, 1) < 0)
 426                 die_errno("unable to redirect descriptor");
 427         close(filter.in);
 428
 429         r = write_tar_archive(ar, args);
 430
 431         close(1);
 432         if (finish_command(&filter) != 0)
 433                 die("'%s' filter reported error", argv[0]);
 434
 435         strbuf_release(&cmd);
 436         return r;
 437 }
 438
 439 static struct archiver tar_archiver = {
 440         "tar",
 441         write_tar_archive,
 442         ARCHIVER_REMOTE
 443 };
 444
 445 void init_tar_archiver(void)
 446 {
 447         int i;
 448         register_archiver(&tar_archiver);
 449
 450         tar_filter_config("tar.tgz.command", "gzip -cn", NULL);
 451         tar_filter_config("tar.tgz.remote", "true", NULL);
 452         tar_filter_config("tar.tar.gz.command", "gzip -cn", NULL);
 453         tar_filter_config("tar.tar.gz.remote", "true", NULL);
 454         git_config(git_tar_config, NULL);
 455         for (i = 0; i < nr_tar_filters; i++) {
 456                 /* omit any filters that never had a command configured */
 457                 if (tar_filters[i]->data)
 458                         register_archiver(tar_filters[i]);
 459         }
 460 }