text-io.c

   1 #include <fcntl.h>
   2 #include <unistd.h>
   3 #include <stdlib.h>
   4 #include <libgen.h>
   5 #include <stdio.h>
   6 #include <errno.h>
   7 #include <string.h>
   8 #include <limits.h>
   9 #include <sys/mman.h>
  10 #if CONFIG_ACL
  11 #include <sys/acl.h>
  12 #endif
  13 #if CONFIG_SELINUX
  14 #include <selinux/selinux.h>
  15 #endif
  16
  17 #include "text.h"
  18 #include "text-internal.h"
  19 #include "text-util.h"
  20 #include "util.h"
  21
  22 struct TextSave {                  /* used to hold context between text_save_{begin,commit} calls */
  23         Text *txt;                 /* text to operate on */
  24         char *filename;            /* filename to save to as given to text_save_begin */
  25         char *tmpname;             /* temporary name used for atomic rename(2) */
  26         int fd;                    /* file descriptor to write data to using text_save_write */
  27         int dirfd;                 /* directory file descriptor, relative to which we save */
  28         enum TextSaveMethod type;  /* method used to save file */
  29 };
  30
  31 /* Allocate blocks holding the actual file content in chunks of size: */
  32 #ifndef BLOCK_SIZE
  33 #define BLOCK_SIZE (1 << 20)
  34 #endif
  35 /* Files smaller than this value are copied on load, larger ones are mmap(2)-ed
  36  * directly. Hence the former can be truncated, while doing so on the latter
  37  * results in havoc. */
  38 #define BLOCK_MMAP_SIZE (1 << 26)
  39
  40 /* allocate a new block of MAX(size, BLOCK_SIZE) bytes */
  41 Block *block_alloc(size_t size) {
  42         Block *blk = calloc(1, sizeof *blk);
  43         if (!blk)
  44                 return NULL;
  45         if (BLOCK_SIZE > size)
  46                 size = BLOCK_SIZE;
  47         if (!(blk->data = malloc(size))) {
  48                 free(blk);
  49                 return NULL;
  50         }
  51         blk->type = BLOCK_TYPE_MALLOC;
  52         blk->size = size;
  53         return blk;
  54 }
  55
  56 Block *block_read(size_t size, int fd) {
  57         Block *blk = block_alloc(size);
  58         if (!blk)
  59                 return NULL;
  60         char *data = blk->data;
  61         size_t rem = size;
  62         while (rem > 0) {
  63                 ssize_t len = read(fd, data, rem);
  64                 if (len == -1) {
  65                         block_free(blk);
  66                         return NULL;
  67                 } else if (len == 0) {
  68                         break;
  69                 } else {
  70                         data += len;
  71                         rem -= len;
  72                 }
  73         }
  74         blk->len = size - rem;
  75         return blk;
  76 }
  77
  78 Block *block_mmap(size_t size, int fd, off_t offset) {
  79         Block *blk = calloc(1, sizeof *blk);
  80         if (!blk)
  81                 return NULL;
  82         if (size) {
  83                 blk->data = mmap(NULL, size, PROT_READ, MAP_SHARED, fd, offset);
  84                 if (blk->data == MAP_FAILED) {
  85                         free(blk);
  86                         return NULL;
  87                 }
  88         }
  89         blk->type = BLOCK_TYPE_MMAP_ORIG;
  90         blk->size = size;
  91         blk->len = size;
  92         return blk;
  93 }
  94
  95 Block *block_load(int dirfd, const char *filename, enum TextLoadMethod method, struct stat *info) {
  96         Block *block = NULL;
  97         int fd = openat(dirfd, filename, O_RDONLY);
  98         if (fd == -1)
  99                 goto out;
 100         if (fstat(fd, info) == -1)
 101                 goto out;
 102         if (!S_ISREG(info->st_mode)) {
 103                 errno = S_ISDIR(info->st_mode) ? EISDIR : ENOTSUP;
 104                 goto out;
 105         }
 106
 107         // XXX: use lseek(fd, 0, SEEK_END); instead?
 108         size_t size = info->st_size;
 109         if (size == 0)
 110                 goto out;
 111         if (method == TEXT_LOAD_READ || (method == TEXT_LOAD_AUTO && size < BLOCK_MMAP_SIZE))
 112                 block = block_read(size, fd);
 113         else
 114                 block = block_mmap(size, fd, 0);
 115 out:
 116         if (fd != -1)
 117                 close(fd);
 118         return block;
 119 }
 120
 121 void block_free(Block *blk) {
 122         if (!blk)
 123                 return;
 124         if (blk->type == BLOCK_TYPE_MALLOC)
 125                 free(blk->data);
 126         else if ((blk->type == BLOCK_TYPE_MMAP_ORIG || blk->type == BLOCK_TYPE_MMAP) && blk->data)
 127                 munmap(blk->data, blk->size);
 128         free(blk);
 129 }
 130
 131 /* check whether block has enough free space to store len bytes */
 132 bool block_capacity(Block *blk, size_t len) {
 133         return blk->size - blk->len >= len;
 134 }
 135
 136 /* append data to block, assumes there is enough space available */
 137 const char *block_append(Block *blk, const char *data, size_t len) {
 138         char *dest = memcpy(blk->data + blk->len, data, len);
 139         blk->len += len;
 140         return dest;
 141 }
 142
 143 /* insert data into block at an arbitrary position, this should only be used with
 144  * data of the most recently created piece. */
 145 bool block_insert(Block *blk, size_t pos, const char *data, size_t len) {
 146         if (pos > blk->len || !block_capacity(blk, len))
 147                 return false;
 148         if (blk->len == pos)
 149                 return block_append(blk, data, len);
 150         char *insert = blk->data + pos;
 151         memmove(insert + len, insert, blk->len - pos);
 152         memcpy(insert, data, len);
 153         blk->len += len;
 154         return true;
 155 }
 156
 157 /* delete data from a block at an arbitrary position, this should only be used with
 158  * data of the most recently created piece. */
 159 bool block_delete(Block *blk, size_t pos, size_t len) {
 160         size_t end;
 161         if (!addu(pos, len, &end) || end > blk->len)
 162                 return false;
 163         if (blk->len == pos) {
 164                 blk->len -= len;
 165                 return true;
 166         }
 167         char *delete = blk->data + pos;
 168         memmove(delete, delete + len, blk->len - pos - len);
 169         blk->len -= len;
 170         return true;
 171 }
 172
 173 Text *text_load(const char *filename) {
 174         return text_load_method(filename, TEXT_LOAD_AUTO);
 175 }
 176
 177 Text *text_loadat(int dirfd, const char *filename) {
 178         return text_loadat_method(dirfd, filename, TEXT_LOAD_AUTO);
 179 }
 180
 181 Text *text_load_method(const char *filename, enum TextLoadMethod method) {
 182         return text_loadat_method(AT_FDCWD, filename, method);
 183 }
 184
 185 static ssize_t write_all(int fd, const char *buf, size_t count) {
 186         size_t rem = count;
 187         while (rem > 0) {
 188                 ssize_t written = write(fd, buf, rem > INT_MAX ? INT_MAX : rem);
 189                 if (written < 0) {
 190                         if (errno == EAGAIN || errno == EINTR)
 191                                 continue;
 192                         return -1;
 193                 } else if (written == 0) {
 194                         break;
 195                 }
 196                 rem -= written;
 197                 buf += written;
 198         }
 199         return count - rem;
 200 }
 201
 202 static bool preserve_acl(int src, int dest) {
 203 #if CONFIG_ACL
 204         acl_t acl = acl_get_fd(src);
 205         if (!acl)
 206                 return errno == ENOTSUP ? true : false;
 207         if (acl_set_fd(dest, acl) == -1) {
 208                 acl_free(acl);
 209                 return false;
 210         }
 211         acl_free(acl);
 212 #endif /* CONFIG_ACL */
 213         return true;
 214 }
 215
 216 static bool preserve_selinux_context(int src, int dest) {
 217 #if CONFIG_SELINUX
 218         char *context = NULL;
 219         if (!is_selinux_enabled())
 220                 return true;
 221         if (fgetfilecon(src, &context) == -1)
 222                 return errno == ENOTSUP ? true : false;
 223         if (fsetfilecon(dest, context) == -1) {
 224                 freecon(context);
 225                 return false;
 226         }
 227         freecon(context);
 228 #endif /* CONFIG_SELINUX */
 229         return true;
 230 }
 231
 232 static int mkstempat(int dirfd, char *template) {
 233         if (dirfd == AT_FDCWD)
 234                 return mkstemp(template);
 235         // FIXME: not thread safe
 236         int fd = -1;
 237         int cwd = open(".", O_RDONLY|O_DIRECTORY);
 238         if (cwd == -1)
 239                 goto err;
 240         if (fchdir(dirfd) == -1)
 241                 goto err;
 242         fd = mkstemp(template);
 243 err:
 244         if (cwd != -1) {
 245                 fchdir(cwd);
 246                 close(cwd);
 247         }
 248         return fd;
 249 }
 250
 251 /* Create a new file named `.filename.vis.XXXXXX` (where `XXXXXX` is a
 252  * randomly generated, unique suffix) and try to preserve all important
 253  * meta data. After the file content has been written to this temporary
 254  * file, text_save_commit_atomic will atomically move it to  its final
 255  * (possibly already existing) destination using rename(2).
 256  *
 257  * This approach does not work if:
 258  *
 259  *   - the file is a symbolic link
 260  *   - the file is a hard link
 261  *   - file ownership can not be preserved
 262  *   - file group can not be preserved
 263  *   - directory permissions do not allow creation of a new file
 264  *   - POSIX ACL can not be preserved (if enabled)
 265  *   - SELinux security context can not be preserved (if enabled)
 266  */
 267 static bool text_save_begin_atomic(TextSave *ctx) {
 268         int oldfd, saved_errno;
 269         if ((oldfd = openat(ctx->dirfd, ctx->filename, O_RDONLY)) == -1 && errno != ENOENT)
 270                 goto err;
 271         struct stat oldmeta = { 0 };
 272         if (oldfd != -1 && fstatat(ctx->dirfd, ctx->filename, &oldmeta, AT_SYMLINK_NOFOLLOW) == -1)
 273                 goto err;
 274         if (oldfd != -1) {
 275                 if (S_ISLNK(oldmeta.st_mode)) /* symbolic link */
 276                         goto err;
 277                 if (oldmeta.st_nlink > 1) /* hard link */
 278                         goto err;
 279         }
 280
 281         char suffix[] = ".vis.XXXXXX";
 282         size_t len = strlen(ctx->filename) + sizeof("./.") + sizeof(suffix);
 283         char *dir = strdup(ctx->filename);
 284         char *base = strdup(ctx->filename);
 285
 286         if (!(ctx->tmpname = malloc(len)) || !dir || !base) {
 287                 free(dir);
 288                 free(base);
 289                 goto err;
 290         }
 291
 292         snprintf(ctx->tmpname, len, "%s/.%s%s", dirname(dir), basename(base), suffix);
 293         free(dir);
 294         free(base);
 295
 296         if ((ctx->fd = mkstempat(ctx->dirfd, ctx->tmpname)) == -1)
 297                 goto err;
 298
 299         if (oldfd == -1) {
 300                 mode_t mask = umask(0);
 301                 umask(mask);
 302                 if (fchmod(ctx->fd, 0666 & ~mask) == -1)
 303                         goto err;
 304         } else {
 305                 if (fchmod(ctx->fd, oldmeta.st_mode) == -1)
 306                         goto err;
 307                 if (!preserve_acl(oldfd, ctx->fd) || !preserve_selinux_context(oldfd, ctx->fd))
 308                         goto err;
 309                 /* change owner if necessary */
 310                 if (oldmeta.st_uid != getuid() && fchown(ctx->fd, oldmeta.st_uid, (uid_t)-1) == -1)
 311                         goto err;
 312                 /* change group if necessary, in case of failure some editors reset
 313                  * the group permissions to the same as for others */
 314                 if (oldmeta.st_gid != getgid() && fchown(ctx->fd, (uid_t)-1, oldmeta.st_gid) == -1)
 315                         goto err;
 316                 close(oldfd);
 317         }
 318
 319         ctx->type = TEXT_SAVE_ATOMIC;
 320         return true;
 321 err:
 322         saved_errno = errno;
 323         if (oldfd != -1)
 324                 close(oldfd);
 325         if (ctx->fd != -1)
 326                 close(ctx->fd);
 327         ctx->fd = -1;
 328         free(ctx->tmpname);
 329         ctx->tmpname = NULL;
 330         errno = saved_errno;
 331         return false;
 332 }
 333
 334 static bool text_save_commit_atomic(TextSave *ctx) {
 335         if (fsync(ctx->fd) == -1)
 336                 return false;
 337
 338         struct stat meta = { 0 };
 339         if (fstat(ctx->fd, &meta) == -1)
 340                 return false;
 341
 342         bool close_failed = (close(ctx->fd) == -1);
 343         ctx->fd = -1;
 344         if (close_failed)
 345                 return false;
 346
 347         if (renameat(ctx->dirfd, ctx->tmpname, ctx->dirfd, ctx->filename) == -1)
 348                 return false;
 349
 350         free(ctx->tmpname);
 351         ctx->tmpname = NULL;
 352
 353         int dir = openat(ctx->dirfd, dirname(ctx->filename), O_DIRECTORY|O_RDONLY);
 354         if (dir == -1)
 355                 return false;
 356
 357         if (fsync(dir) == -1 && errno != EINVAL) {
 358                 close(dir);
 359                 return false;
 360         }
 361
 362         if (close(dir) == -1)
 363                 return false;
 364
 365         text_saved(ctx->txt, &meta);
 366         return true;
 367 }
 368
 369 static bool text_save_begin_inplace(TextSave *ctx) {
 370         Text *txt = ctx->txt;
 371         struct stat now = { 0 };
 372         int newfd = -1, saved_errno;
 373         if ((ctx->fd = openat(ctx->dirfd, ctx->filename, O_CREAT|O_WRONLY, 0666)) == -1)
 374                 goto err;
 375         if (fstat(ctx->fd, &now) == -1)
 376                 goto err;
 377         struct stat loaded = text_stat(txt);
 378         Block *block = text_block_mmaped(txt);
 379         if (block && now.st_dev == loaded.st_dev && now.st_ino == loaded.st_ino) {
 380                 /* The file we are going to overwrite is currently mmap-ed from
 381                  * text_load, therefore we copy the mmap-ed block to a temporary
 382                  * file and remap it at the same position such that all pointers
 383                  * from the various pieces are still valid.
 384                  */
 385                 size_t size = block->size;
 386                 char tmpname[32] = "/tmp/vis-XXXXXX";
 387                 newfd = mkstemp(tmpname);
 388                 if (newfd == -1)
 389                         goto err;
 390                 if (unlink(tmpname) == -1)
 391                         goto err;
 392                 ssize_t written = write_all(newfd, block->data, size);
 393                 if (written == -1 || (size_t)written != size)
 394                         goto err;
 395                 void *data = mmap(block->data, size, PROT_READ, MAP_SHARED|MAP_FIXED, newfd, 0);
 396                 if (data == MAP_FAILED)
 397                         goto err;
 398                 bool close_failed = (close(newfd) == -1);
 399                 newfd = -1;
 400                 if (close_failed)
 401                         goto err;
 402                 block->type = BLOCK_TYPE_MMAP;
 403         }
 404         /* overwrite the existing file content, if something goes wrong
 405          * here we are screwed, TODO: make a backup before? */
 406         if (ftruncate(ctx->fd, 0) == -1)
 407                 goto err;
 408         ctx->type = TEXT_SAVE_INPLACE;
 409         return true;
 410 err:
 411         saved_errno = errno;
 412         if (newfd != -1)
 413                 close(newfd);
 414         if (ctx->fd != -1)
 415                 close(ctx->fd);
 416         ctx->fd = -1;
 417         errno = saved_errno;
 418         return false;
 419 }
 420
 421 static bool text_save_commit_inplace(TextSave *ctx) {
 422         if (fsync(ctx->fd) == -1)
 423                 return false;
 424         struct stat meta = { 0 };
 425         if (fstat(ctx->fd, &meta) == -1)
 426                 return false;
 427         if (close(ctx->fd) == -1)
 428                 return false;
 429         text_saved(ctx->txt, &meta);
 430         return true;
 431 }
 432
 433 TextSave *text_save_begin(Text *txt, int dirfd, const char *filename, enum TextSaveMethod type) {
 434         if (!filename)
 435                 return NULL;
 436         TextSave *ctx = calloc(1, sizeof *ctx);
 437         if (!ctx)
 438                 return NULL;
 439         ctx->txt = txt;
 440         ctx->fd = -1;
 441         ctx->dirfd = dirfd;
 442         if (!(ctx->filename = strdup(filename)))
 443                 goto err;
 444         errno = 0;
 445         if ((type == TEXT_SAVE_AUTO || type == TEXT_SAVE_ATOMIC) && text_save_begin_atomic(ctx))
 446                 return ctx;
 447         if (errno == ENOSPC)
 448                 goto err;
 449         if ((type == TEXT_SAVE_AUTO || type == TEXT_SAVE_INPLACE) && text_save_begin_inplace(ctx))
 450                 return ctx;
 451 err:
 452         text_save_cancel(ctx);
 453         return NULL;
 454 }
 455
 456 bool text_save_commit(TextSave *ctx) {
 457         if (!ctx)
 458                 return true;
 459         bool ret;
 460         switch (ctx->type) {
 461         case TEXT_SAVE_ATOMIC:
 462                 ret = text_save_commit_atomic(ctx);
 463                 break;
 464         case TEXT_SAVE_INPLACE:
 465                 ret = text_save_commit_inplace(ctx);
 466                 break;
 467         default:
 468                 ret = false;
 469                 break;
 470         }
 471
 472         text_save_cancel(ctx);
 473         return ret;
 474 }
 475
 476 void text_save_cancel(TextSave *ctx) {
 477         if (!ctx)
 478                 return;
 479         int saved_errno = errno;
 480         if (ctx->fd != -1)
 481                 close(ctx->fd);
 482         if (ctx->tmpname && ctx->tmpname[0])
 483                 unlinkat(ctx->dirfd, ctx->tmpname, 0);
 484         free(ctx->tmpname);
 485         free(ctx->filename);
 486         free(ctx);
 487         errno = saved_errno;
 488 }
 489
 490 /* First try to save the file atomically using rename(2) if this does not
 491  * work overwrite the file in place. However if something goes wrong during
 492  * this overwrite the original file is permanently damaged.
 493  */
 494 bool text_save(Text *txt, const char *filename) {
 495         return text_saveat(txt, AT_FDCWD, filename);
 496 }
 497
 498 bool text_saveat(Text *txt, int dirfd, const char *filename) {
 499         return text_saveat_method(txt, dirfd, filename, TEXT_SAVE_AUTO);
 500 }
 501
 502 bool text_save_method(Text *txt, const char *filename, enum TextSaveMethod method) {
 503         return text_saveat_method(txt, AT_FDCWD, filename, method);
 504 }
 505
 506 bool text_saveat_method(Text *txt, int dirfd, const char *filename, enum TextSaveMethod method) {
 507         if (!filename) {
 508                 text_saved(txt, NULL);
 509                 return true;
 510         }
 511         TextSave *ctx = text_save_begin(txt, dirfd, filename, method);
 512         if (!ctx)
 513                 return false;
 514         Filerange range = (Filerange){ .start = 0, .end = text_size(txt) };
 515         ssize_t written = text_save_write_range(ctx, &range);
 516         if (written == -1 || (size_t)written != text_range_size(&range)) {
 517                 text_save_cancel(ctx);
 518                 return false;
 519         }
 520         return text_save_commit(ctx);
 521 }
 522
 523 ssize_t text_save_write_range(TextSave *ctx, const Filerange *range) {
 524         return text_write_range(ctx->txt, range, ctx->fd);
 525 }
 526
 527 ssize_t text_write(const Text *txt, int fd) {
 528         Filerange r = (Filerange){ .start = 0, .end = text_size(txt) };
 529         return text_write_range(txt, &r, fd);
 530 }
 531
 532 ssize_t text_write_range(const Text *txt, const Filerange *range, int fd) {
 533         size_t size = text_range_size(range), rem = size;
 534         for (Iterator it = text_iterator_get(txt, range->start);
 535              rem > 0 && text_iterator_valid(&it);
 536              text_iterator_next(&it)) {
 537                 size_t prem = it.end - it.text;
 538                 if (prem > rem)
 539                         prem = rem;
 540                 ssize_t written = write_all(fd, it.text, prem);
 541                 if (written == -1)
 542                         return -1;
 543                 rem -= written;
 544                 if ((size_t)written != prem)
 545                         break;
 546         }
 547         return size - rem;
 548 }