wimlib.net Git - wimlib/blob - src/win32_capture.c

   1 /*
   2  * win32_capture.c - Windows-specific code for capturing files into a WIM image.
   3  */
   4
   5 /*
   6  * Copyright (C) 2013, 2014 Eric Biggers
   7  *
   8  * This file is part of wimlib, a library for working with WIM files.
   9  *
  10  * wimlib is free software; you can redistribute it and/or modify it under the
  11  * terms of the GNU General Public License as published by the Free
  12  * Software Foundation; either version 3 of the License, or (at your option)
  13  * any later version.
  14  *
  15  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  16  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  17  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  18  * details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  22  */
  23
  24 #ifdef __WIN32__
  25
  26 #ifdef HAVE_CONFIG_H
  27 #  include "config.h"
  28 #endif
  29
  30 #include "wimlib/win32_common.h"
  31
  32 #include "wimlib/capture.h"
  33 #include "wimlib/dentry.h"
  34 #include "wimlib/encoding.h"
  35 #include "wimlib/endianness.h"
  36 #include "wimlib/error.h"
  37 #include "wimlib/lookup_table.h"
  38 #include "wimlib/paths.h"
  39 #include "wimlib/reparse.h"
  40
  41 struct win32_capture_state {
  42         unsigned long num_get_sd_access_denied;
  43         unsigned long num_get_sacl_priv_notheld;
  44         unsigned long num_long_path_warnings;
  45 };
  46
  47 int
  48 read_win32_file_prefix(const struct wim_lookup_table_entry *lte,
  49                        u64 size,
  50                        consume_data_callback_t cb,
  51                        void *cb_ctx)
  52 {
  53         int ret = 0;
  54         u64 bytes_remaining;
  55         u8 buf[BUFFER_SIZE];
  56
  57         HANDLE hFile = win32_open_existing_file(lte->file_on_disk,
  58                                                 FILE_READ_DATA);
  59         if (hFile == INVALID_HANDLE_VALUE) {
  60                 set_errno_from_GetLastError();
  61                 ERROR_WITH_ERRNO("Failed to open \"%ls\"", lte->file_on_disk);
  62                 return WIMLIB_ERR_OPEN;
  63         }
  64
  65         bytes_remaining = size;
  66         while (bytes_remaining) {
  67                 DWORD bytesToRead, bytesRead;
  68
  69                 bytesToRead = min(sizeof(buf), bytes_remaining);
  70                 if (!ReadFile(hFile, buf, bytesToRead, &bytesRead, NULL) ||
  71                     bytesRead != bytesToRead)
  72                 {
  73                         set_errno_from_GetLastError();
  74                         ERROR_WITH_ERRNO("Failed to read data from \"%ls\"",
  75                                          lte->file_on_disk);
  76                         ret = WIMLIB_ERR_READ;
  77                         break;
  78                 }
  79                 bytes_remaining -= bytesRead;
  80                 ret = (*cb)(buf, bytesRead, cb_ctx);
  81                 if (ret)
  82                         break;
  83         }
  84         CloseHandle(hFile);
  85         return ret;
  86 }
  87
  88 struct win32_encrypted_read_ctx {
  89         consume_data_callback_t read_prefix_cb;
  90         void *read_prefix_ctx;
  91         int wimlib_err_code;
  92         u64 bytes_remaining;
  93 };
  94
  95 static DWORD WINAPI
  96 win32_encrypted_export_cb(unsigned char *data, void *_ctx, unsigned long len)
  97 {
  98         struct win32_encrypted_read_ctx *ctx = _ctx;
  99         int ret;
 100         size_t bytes_to_consume = min(len, ctx->bytes_remaining);
 101
 102         if (bytes_to_consume == 0)
 103                 return ERROR_SUCCESS;
 104
 105         ret = (*ctx->read_prefix_cb)(data, bytes_to_consume, ctx->read_prefix_ctx);
 106         if (ret) {
 107                 ctx->wimlib_err_code = ret;
 108                 /* Shouldn't matter what error code is returned here, as long as
 109                  * it isn't ERROR_SUCCESS.  */
 110                 return ERROR_READ_FAULT;
 111         }
 112         ctx->bytes_remaining -= bytes_to_consume;
 113         return ERROR_SUCCESS;
 114 }
 115
 116 int
 117 read_win32_encrypted_file_prefix(const struct wim_lookup_table_entry *lte,
 118                                  u64 size,
 119                                  consume_data_callback_t cb,
 120                                  void *cb_ctx)
 121 {
 122         struct win32_encrypted_read_ctx export_ctx;
 123         DWORD err;
 124         void *file_ctx;
 125         int ret;
 126
 127         DEBUG("Reading %"PRIu64" bytes from encrypted file \"%ls\"",
 128               size, lte->file_on_disk);
 129
 130         export_ctx.read_prefix_cb = cb;
 131         export_ctx.read_prefix_ctx = cb_ctx;
 132         export_ctx.wimlib_err_code = 0;
 133         export_ctx.bytes_remaining = size;
 134
 135         err = OpenEncryptedFileRaw(lte->file_on_disk, 0, &file_ctx);
 136         if (err != ERROR_SUCCESS) {
 137                 set_errno_from_win32_error(err);
 138                 ERROR_WITH_ERRNO("Failed to open encrypted file \"%ls\" "
 139                                  "for raw read", lte->file_on_disk);
 140                 return WIMLIB_ERR_OPEN;
 141         }
 142         err = ReadEncryptedFileRaw(win32_encrypted_export_cb,
 143                                    &export_ctx, file_ctx);
 144         if (err != ERROR_SUCCESS) {
 145                 set_errno_from_win32_error(err);
 146                 ERROR_WITH_ERRNO("Failed to read encrypted file \"%ls\"",
 147                                  lte->file_on_disk);
 148                 ret = export_ctx.wimlib_err_code;
 149                 if (ret == 0)
 150                         ret = WIMLIB_ERR_READ;
 151         } else if (export_ctx.bytes_remaining != 0) {
 152                 ERROR("Only could read %"PRIu64" of %"PRIu64" bytes from "
 153                       "encryted file \"%ls\"",
 154                       size - export_ctx.bytes_remaining, size,
 155                       lte->file_on_disk);
 156                 ret = WIMLIB_ERR_READ;
 157         } else {
 158                 ret = 0;
 159         }
 160         CloseEncryptedFileRaw(file_ctx);
 161         return ret;
 162 }
 163
 164
 165 static u64
 166 FILETIME_to_u64(const FILETIME *ft)
 167 {
 168         return ((u64)ft->dwHighDateTime << 32) | (u64)ft->dwLowDateTime;
 169 }
 170
 171 /* Load the short name of a file into a WIM dentry.
 172  *
 173  * If we can't read the short filename for some reason, we just ignore the error
 174  * and assume the file has no short name.  This shouldn't be an issue, since the
 175  * short names are essentially obsolete anyway.
 176  */
 177 static int
 178 win32_get_short_name(HANDLE hFile, struct wim_dentry *dentry)
 179 {
 180
 181         /* It's not any harder to just make the NtQueryInformationFile() system
 182          * call ourselves, and it saves a dumb call to FindFirstFile() which of
 183          * course has to create its own handle.  */
 184         NTSTATUS status;
 185         IO_STATUS_BLOCK io_status;
 186         u8 buf[128] _aligned_attribute(8);
 187         const FILE_NAME_INFORMATION *info;
 188
 189         status = (*func_NtQueryInformationFile)(hFile, &io_status, buf, sizeof(buf),
 190                                                 FileAlternateNameInformation);
 191         info = (const FILE_NAME_INFORMATION*)buf;
 192         if (NT_SUCCESS(status) && info->FileNameLength != 0) {
 193                 dentry->short_name = utf16le_dupz(info->FileName,
 194                                                   info->FileNameLength);
 195                 if (!dentry->short_name)
 196                         return WIMLIB_ERR_NOMEM;
 197                 dentry->short_name_nbytes = info->FileNameLength;
 198         }
 199         return 0;
 200 }
 201
 202 static int
 203 win32_get_security_descriptor(HANDLE hFile,
 204                               const wchar_t *path,
 205                               struct wim_inode *inode,
 206                               struct wim_sd_set *sd_set,
 207                               struct win32_capture_state *state,
 208                               int add_flags)
 209 {
 210         SECURITY_INFORMATION requestedInformation;
 211         u8 _buf[4096];
 212         u8 *buf;
 213         size_t bufsize;
 214         DWORD lenNeeded;
 215         NTSTATUS status;
 216         int ret;
 217
 218         requestedInformation = DACL_SECURITY_INFORMATION |
 219                                SACL_SECURITY_INFORMATION |
 220                                OWNER_SECURITY_INFORMATION |
 221                                GROUP_SECURITY_INFORMATION;
 222         buf = _buf;
 223         bufsize = sizeof(_buf);
 224
 225         /*
 226          * We need the file's security descriptor in SECURITY_DESCRIPTOR_RELATIVE
 227          * format, and we currently have a handle opened with as many relevant
 228          * permissions as possible.  At this point, on Windows there are a number of
 229          * options for reading a file's security descriptor:
 230          *
 231          * GetFileSecurity():  This takes in a path and returns the
 232          * SECURITY_DESCRIPTOR_RELATIVE.  Problem: this uses an internal handle, not
 233          * ours, and the handle created internally doesn't specify
 234          * FILE_FLAG_BACKUP_SEMANTICS.  Therefore there can be access denied errors on
 235          * some files and directories, even when running as the Administrator.
 236          *
 237          * GetSecurityInfo():  This takes in a handle and returns the security
 238          * descriptor split into a bunch of different parts.  This should work, but it's
 239          * dumb because we have to put the security descriptor back together again.
 240          *
 241          * BackupRead():  This can read the security descriptor, but this is a
 242          * difficult-to-use API, probably only works as the Administrator, and the
 243          * format of the returned data is not well documented.
 244          *
 245          * NtQuerySecurityObject():  This is exactly what we need, as it takes in a
 246          * handle and returns the security descriptor in SECURITY_DESCRIPTOR_RELATIVE
 247          * format.  Only problem is that it's a ntdll function and therefore not
 248          * officially part of the Win32 API.  Oh well.
 249          */
 250         while (!(NT_SUCCESS(status = (*func_NtQuerySecurityObject)(hFile,
 251                                                                    requestedInformation,
 252                                                                    (PSECURITY_DESCRIPTOR)buf,
 253                                                                    bufsize,
 254                                                                    &lenNeeded))))
 255         {
 256                 switch (status) {
 257                 case STATUS_BUFFER_OVERFLOW:
 258                         wimlib_assert(buf == _buf);
 259                         buf = MALLOC(lenNeeded);
 260                         if (!buf)
 261                                 return WIMLIB_ERR_NOMEM;
 262                         bufsize = lenNeeded;
 263                         break;
 264                 case STATUS_PRIVILEGE_NOT_HELD:
 265                 case STATUS_ACCESS_DENIED:
 266                         if (add_flags & WIMLIB_ADD_FLAG_STRICT_ACLS) {
 267                 default:
 268                                 set_errno_from_nt_status(status);
 269                                 ERROR_WITH_ERRNO("\"%ls\": Failed to "
 270                                                  "read security descriptor", path);
 271                                 ret = WIMLIB_ERR_READ;
 272                                 goto out_free_buf;
 273                         }
 274                         if (requestedInformation & SACL_SECURITY_INFORMATION) {
 275                                 state->num_get_sacl_priv_notheld++;
 276                                 requestedInformation &= ~SACL_SECURITY_INFORMATION;
 277                                 break;
 278                         }
 279                         state->num_get_sd_access_denied++;
 280                         ret = 0;
 281                         goto out_free_buf;
 282                 }
 283         }
 284
 285         inode->i_security_id = sd_set_add_sd(sd_set, buf, lenNeeded);
 286         if (inode->i_security_id < 0)
 287                 ret = WIMLIB_ERR_NOMEM;
 288         else
 289                 ret = 0;
 290 out_free_buf:
 291         if (buf != _buf)
 292                 FREE(buf);
 293         return ret;
 294 }
 295
 296 static int
 297 win32_build_dentry_tree_recursive(struct wim_dentry **root_ret,
 298                                   wchar_t *path,
 299                                   size_t path_num_chars,
 300                                   struct add_image_params *params,
 301                                   struct win32_capture_state *state,
 302                                   unsigned vol_flags);
 303
 304 /* Reads the directory entries of directory and recursively calls
 305  * win32_build_dentry_tree() on them.  */
 306 static int
 307 win32_recurse_directory(HANDLE hDir,
 308                         wchar_t *dir_path,
 309                         size_t dir_path_num_chars,
 310                         struct wim_dentry *root,
 311                         struct add_image_params *params,
 312                         struct win32_capture_state *state,
 313                         unsigned vol_flags)
 314 {
 315         int ret;
 316
 317         DEBUG("Recurse to directory \"%ls\"", dir_path);
 318
 319         /* Using NtQueryDirectoryFile() we can re-use the same open handle,
 320          * which we opened with FILE_FLAG_BACKUP_SEMANTICS.  */
 321
 322         NTSTATUS status;
 323         IO_STATUS_BLOCK io_status;
 324         const size_t bufsize = 8192;
 325         void *buf;
 326
 327         buf = MALLOC(bufsize);
 328         if (!buf)
 329                 return WIMLIB_ERR_NOMEM;
 330
 331         while (NT_SUCCESS(status = (*func_NtQueryDirectoryFile)(hDir, NULL, NULL, NULL,
 332                                                                 &io_status, buf, bufsize,
 333                                                                 FileNamesInformation,
 334                                                                 FALSE, NULL, FALSE)))
 335         {
 336                 const FILE_NAMES_INFORMATION *info = buf;
 337                 for (;;) {
 338                         if (!(info->FileNameLength == 2 && info->FileName[0] == L'.') &&
 339                             !(info->FileNameLength == 4 && info->FileName[0] == L'.' &&
 340                                                            info->FileName[1] == L'.'))
 341                         {
 342                                 wchar_t *p;
 343                                 struct wim_dentry *child;
 344
 345                                 p = dir_path + dir_path_num_chars;
 346                                 *p++ = L'\\';
 347                                 p = wmempcpy(p, info->FileName,
 348                                              info->FileNameLength / 2);
 349                                 *p = '\0';
 350
 351                                 ret = win32_build_dentry_tree_recursive(
 352                                                                 &child,
 353                                                                 dir_path,
 354                                                                 p - dir_path,
 355                                                                 params,
 356                                                                 state,
 357                                                                 vol_flags);
 358
 359                                 dir_path[dir_path_num_chars] = L'\0';
 360
 361                                 if (ret)
 362                                         goto out_free_buf;
 363                                 if (child)
 364                                         dentry_add_child(root, child);
 365                         }
 366                         if (info->NextEntryOffset == 0)
 367                                 break;
 368                         info = (const FILE_NAMES_INFORMATION *)
 369                                         ((const u8 *)info + info->NextEntryOffset);
 370                 }
 371         }
 372
 373         if (status != STATUS_NO_MORE_FILES) {
 374                 set_errno_from_nt_status(status);
 375                 ERROR_WITH_ERRNO("Failed to read directory \"%ls\"", dir_path);
 376                 ret = WIMLIB_ERR_READ;
 377         }
 378 out_free_buf:
 379         FREE(buf);
 380         return ret;
 381 }
 382
 383 /* Reparse point fixup status code */
 384 enum rp_status {
 385         /* Reparse point corresponded to an absolute symbolic link or junction
 386          * point that pointed outside the directory tree being captured, and
 387          * therefore was excluded. */
 388         RP_EXCLUDED       = 0x0,
 389
 390         /* Reparse point was not fixed as it was either a relative symbolic
 391          * link, a mount point, or something else we could not understand. */
 392         RP_NOT_FIXED      = 0x1,
 393
 394         /* Reparse point corresponded to an absolute symbolic link or junction
 395          * point that pointed inside the directory tree being captured, where
 396          * the target was specified by a "full" \??\ prefixed path, and
 397          * therefore was fixed to be relative to the root of the directory tree
 398          * being captured. */
 399         RP_FIXED_FULLPATH = 0x2,
 400
 401         /* Same as RP_FIXED_FULLPATH, except the absolute link target did not
 402          * have the \??\ prefix.  It may have begun with a drive letter though.
 403          * */
 404         RP_FIXED_ABSPATH  = 0x4,
 405
 406         /* Either RP_FIXED_FULLPATH or RP_FIXED_ABSPATH. */
 407         RP_FIXED          = RP_FIXED_FULLPATH | RP_FIXED_ABSPATH,
 408 };
 409
 410 /* Given the "substitute name" target of a Windows reparse point, try doing a
 411  * fixup where we change it to be absolute relative to the root of the directory
 412  * tree being captured.
 413  *
 414  * Note that this is only executed when WIMLIB_ADD_FLAG_RPFIX has been
 415  * set.
 416  *
 417  * @capture_root_ino and @capture_root_dev indicate the inode number and device
 418  * of the root of the directory tree being captured.  They are meant to identify
 419  * this directory (as an alternative to its actual path, which could potentially
 420  * be reached via multiple destinations due to other symbolic links).  This may
 421  * not work properly on FAT, which doesn't seem to supply proper inode numbers
 422  * or file IDs.  However, FAT doesn't support reparse points so this function
 423  * wouldn't even be called anyway.
 424  */
 425 static enum rp_status
 426 win32_capture_maybe_rpfix_target(wchar_t *target, u16 *target_nbytes_p,
 427                                  u64 capture_root_ino, u64 capture_root_dev,
 428                                  u32 rptag)
 429 {
 430         u16 target_nchars = *target_nbytes_p / 2;
 431         size_t stripped_chars;
 432         wchar_t *orig_target;
 433         int ret;
 434
 435         ret = parse_substitute_name(target, *target_nbytes_p, rptag);
 436         if (ret < 0)
 437                 return RP_NOT_FIXED;
 438         stripped_chars = ret;
 439         if (stripped_chars)
 440                 stripped_chars -= 2;
 441         target[target_nchars] = L'\0';
 442         orig_target = target;
 443         target = capture_fixup_absolute_symlink(target + stripped_chars,
 444                                                 capture_root_ino, capture_root_dev);
 445         if (!target)
 446                 return RP_EXCLUDED;
 447         target_nchars = wcslen(target);
 448         wmemmove(orig_target + stripped_chars, target, target_nchars + 1);
 449         *target_nbytes_p = (target_nchars + stripped_chars) * sizeof(wchar_t);
 450         DEBUG("Fixed reparse point (new target: \"%ls\")", orig_target);
 451         if (stripped_chars)
 452                 return RP_FIXED_FULLPATH;
 453         else
 454                 return RP_FIXED_ABSPATH;
 455 }
 456
 457 /* Returns: `enum rp_status' value on success; negative WIMLIB_ERR_* value on
 458  * failure. */
 459 static int
 460 win32_capture_try_rpfix(u8 *rpbuf, u16 *rpbuflen_p,
 461                         u64 capture_root_ino, u64 capture_root_dev,
 462                         const wchar_t *path, struct add_image_params *params)
 463 {
 464         struct reparse_data rpdata;
 465         int ret;
 466         enum rp_status rp_status;
 467
 468         ret = parse_reparse_data(rpbuf, *rpbuflen_p, &rpdata);
 469         if (ret)
 470                 return -ret;
 471
 472         rp_status = win32_capture_maybe_rpfix_target(rpdata.substitute_name,
 473                                                      &rpdata.substitute_name_nbytes,
 474                                                      capture_root_ino,
 475                                                      capture_root_dev,
 476                                                      le32_to_cpu(*(le32*)rpbuf));
 477         if (rp_status & RP_FIXED) {
 478                 wimlib_assert(rpdata.substitute_name_nbytes % 2 == 0);
 479                 utf16lechar substitute_name_copy[rpdata.substitute_name_nbytes / 2];
 480                 wmemcpy(substitute_name_copy, rpdata.substitute_name,
 481                         rpdata.substitute_name_nbytes / 2);
 482                 rpdata.substitute_name = substitute_name_copy;
 483                 rpdata.print_name = substitute_name_copy;
 484                 rpdata.print_name_nbytes = rpdata.substitute_name_nbytes;
 485                 if (rp_status == RP_FIXED_FULLPATH) {
 486                         /* "full path", meaning \??\ prefixed.  We should not
 487                          * include this prefix in the print name, as it is
 488                          * apparently meant for the filesystem driver only. */
 489                         rpdata.print_name += 4;
 490                         rpdata.print_name_nbytes -= 8;
 491                 }
 492                 ret = make_reparse_buffer(&rpdata, rpbuf, rpbuflen_p);
 493                 if (ret == 0)
 494                         ret = rp_status;
 495                 else
 496                         ret = -ret;
 497         } else {
 498                 if (rp_status == RP_EXCLUDED) {
 499                         /* Ignoring absolute symbolic link or junction point
 500                          * that points out of the tree to be captured.  */
 501                         size_t print_name_nchars = rpdata.print_name_nbytes / 2;
 502                         wchar_t print_name0[print_name_nchars + 1];
 503                         print_name0[print_name_nchars] = L'\0';
 504                         wmemcpy(print_name0, rpdata.print_name, print_name_nchars);
 505
 506                         params->progress.scan.cur_path = path;
 507                         params->progress.scan.symlink_target = print_name0;
 508                         do_capture_progress(params,
 509                                             WIMLIB_SCAN_DENTRY_EXCLUDED_SYMLINK,
 510                                             NULL);
 511                 }
 512                 ret = rp_status;
 513         }
 514         return ret;
 515 }
 516
 517 /*
 518  * Loads the reparse point data from a reparse point into memory, optionally
 519  * fixing the targets of absolute symbolic links and junction points to be
 520  * relative to the root of capture.
 521  *
 522  * @hFile:  Open handle to the reparse point.
 523  * @path:   Path to the reparse point.  Used for error messages only.
 524  * @params: Additional parameters, including whether to do reparse point fixups
 525  *          or not.
 526  * @rpbuf:  Buffer of length at least REPARSE_POINT_MAX_SIZE bytes into which
 527  *          the reparse point buffer will be loaded.
 528  * @rpbuflen_ret:  On success, the length of the reparse point buffer in bytes
 529  *                 is written to this location.
 530  *
 531  * Returns:
 532  *      On success, returns an `enum rp_status' value that indicates if and/or
 533  *      how the reparse point fixup was done.
 534  *
 535  *      On failure, returns a negative value that is a negated WIMLIB_ERR_*
 536  *      code.
 537  */
 538 static int
 539 win32_get_reparse_data(HANDLE hFile, const wchar_t *path,
 540                        struct add_image_params *params,
 541                        u8 *rpbuf, u16 *rpbuflen_ret)
 542 {
 543         DWORD bytesReturned;
 544         u32 reparse_tag;
 545         int ret;
 546         u16 rpbuflen;
 547
 548         DEBUG("Loading reparse data from \"%ls\"", path);
 549         if (!DeviceIoControl(hFile, FSCTL_GET_REPARSE_POINT,
 550                              NULL, /* "Not used with this operation; set to NULL" */
 551                              0, /* "Not used with this operation; set to 0" */
 552                              rpbuf, /* "A pointer to a buffer that
 553                                                    receives the reparse point data */
 554                              REPARSE_POINT_MAX_SIZE, /* "The size of the output
 555                                                         buffer, in bytes */
 556                              &bytesReturned,
 557                              NULL))
 558         {
 559                 set_errno_from_GetLastError();
 560                 ERROR_WITH_ERRNO("Failed to get reparse data of \"%ls\"", path);
 561                 return -WIMLIB_ERR_READ;
 562         }
 563         if (bytesReturned < 8 || bytesReturned > REPARSE_POINT_MAX_SIZE) {
 564                 ERROR("Reparse data on \"%ls\" is invalid", path);
 565                 return -WIMLIB_ERR_INVALID_REPARSE_DATA;
 566         }
 567
 568         rpbuflen = bytesReturned;
 569         reparse_tag = le32_to_cpu(*(le32*)rpbuf);
 570         if (params->add_flags & WIMLIB_ADD_FLAG_RPFIX &&
 571             (reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK ||
 572              reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT))
 573         {
 574                 /* Try doing reparse point fixup */
 575                 ret = win32_capture_try_rpfix(rpbuf,
 576                                               &rpbuflen,
 577                                               params->capture_root_ino,
 578                                               params->capture_root_dev,
 579                                               path,
 580                                               params);
 581         } else {
 582                 ret = RP_NOT_FIXED;
 583         }
 584         *rpbuflen_ret = rpbuflen;
 585         return ret;
 586 }
 587
 588 static DWORD WINAPI
 589 win32_tally_encrypted_size_cb(unsigned char *_data, void *_size_ret,
 590                               unsigned long len)
 591 {
 592         *(u64*)_size_ret += len;
 593         return ERROR_SUCCESS;
 594 }
 595
 596 static int
 597 win32_get_encrypted_file_size(const wchar_t *path, u64 *size_ret)
 598 {
 599         DWORD err;
 600         void *file_ctx;
 601         int ret;
 602
 603         err = OpenEncryptedFileRaw(path, 0, &file_ctx);
 604         if (err != ERROR_SUCCESS) {
 605                 set_errno_from_win32_error(err);
 606                 ERROR_WITH_ERRNO("Failed to open encrypted file \"%ls\" "
 607                                  "for raw read", path);
 608                 return WIMLIB_ERR_OPEN;
 609         }
 610         *size_ret = 0;
 611         err = ReadEncryptedFileRaw(win32_tally_encrypted_size_cb,
 612                                    size_ret, file_ctx);
 613         if (err != ERROR_SUCCESS) {
 614                 set_errno_from_win32_error(err);
 615                 ERROR_WITH_ERRNO("Failed to read raw encrypted data from "
 616                                  "\"%ls\"", path);
 617                 ret = WIMLIB_ERR_READ;
 618         } else {
 619                 ret = 0;
 620         }
 621         CloseEncryptedFileRaw(file_ctx);
 622         return ret;
 623 }
 624
 625 /* Scans an unnamed or named stream of a Win32 file (not a reparse point
 626  * stream); calculates its SHA1 message digest and either creates a `struct
 627  * wim_lookup_table_entry' in memory for it, or uses an existing 'struct
 628  * wim_lookup_table_entry' for an identical stream.
 629  *
 630  * @path:               Path to the file (UTF-16LE).
 631  *
 632  * @path_num_chars:     Number of 2-byte characters in @path.
 633  *
 634  * @inode:              WIM inode to save the stream into.
 635  *
 636  * @unhashed_streams:   List of unhashed streams that have been added to the WIM
 637  *                      image.
 638  *
 639  * @dat:                A `WIN32_FIND_STREAM_DATA' structure that specifies the
 640  *                      stream name.
 641  *
 642  * Returns 0 on success; nonzero on failure.
 643  */
 644 static int
 645 win32_capture_stream(const wchar_t *path,
 646                      size_t path_num_chars,
 647                      struct wim_inode *inode,
 648                      struct list_head *unhashed_streams,
 649                      WIN32_FIND_STREAM_DATA *dat)
 650 {
 651         struct wim_ads_entry *ads_entry;
 652         struct wim_lookup_table_entry *lte;
 653         int ret;
 654         wchar_t *stream_name, *colon;
 655         size_t stream_name_nchars;
 656         bool is_named_stream;
 657         wchar_t *spath;
 658         size_t spath_nchars;
 659         size_t spath_buf_nbytes;
 660         const wchar_t *relpath_prefix;
 661         const wchar_t *colonchar;
 662
 663         DEBUG("Capture \"%ls\" stream \"%ls\"", path, dat->cStreamName);
 664
 665         /* The stream name should be returned as :NAME:TYPE */
 666         stream_name = dat->cStreamName;
 667         if (*stream_name != L':')
 668                 goto out_invalid_stream_name;
 669         stream_name += 1;
 670         colon = wcschr(stream_name, L':');
 671         if (colon == NULL)
 672                 goto out_invalid_stream_name;
 673
 674         if (wcscmp(colon + 1, L"$DATA")) {
 675                 /* Not a DATA stream */
 676                 ret = 0;
 677                 goto out;
 678         }
 679
 680         *colon = '\0';
 681
 682         stream_name_nchars = colon - stream_name;
 683         is_named_stream = (stream_name_nchars != 0);
 684
 685         if (is_named_stream) {
 686                 /* Allocate an ADS entry for the named stream. */
 687                 ads_entry = inode_add_ads_utf16le(inode, stream_name,
 688                                                   stream_name_nchars * sizeof(wchar_t));
 689                 if (!ads_entry) {
 690                         ret = WIMLIB_ERR_NOMEM;
 691                         goto out;
 692                 }
 693         }
 694
 695         /* If zero length stream, no lookup table entry needed. */
 696         if ((u64)dat->StreamSize.QuadPart == 0) {
 697                 ret = 0;
 698                 goto out;
 699         }
 700
 701         /* Create a UTF-16LE string @spath that gives the filename, then a
 702          * colon, then the stream name.  Or, if it's an unnamed stream, just the
 703          * filename.  It is MALLOC()'ed so that it can be saved in the
 704          * wim_lookup_table_entry if needed.
 705          *
 706          * As yet another special case, relative paths need to be changed to
 707          * begin with an explicit "./" so that, for example, a file t:ads, where
 708          * :ads is the part we added, is not interpreted as a file on the t:
 709          * drive. */
 710         spath_nchars = path_num_chars;
 711         relpath_prefix = L"";
 712         colonchar = L"";
 713         if (is_named_stream) {
 714                 spath_nchars += 1 + stream_name_nchars;
 715                 colonchar = L":";
 716                 if (path_num_chars == 1 && !is_any_path_separator(path[0])) {
 717                         spath_nchars += 2;
 718                         static const wchar_t _relpath_prefix[] =
 719                                 {L'.', OS_PREFERRED_PATH_SEPARATOR, L'\0'};
 720                         relpath_prefix = _relpath_prefix;
 721                 }
 722         }
 723
 724         spath_buf_nbytes = (spath_nchars + 1) * sizeof(wchar_t);
 725         spath = MALLOC(spath_buf_nbytes);
 726
 727         tsprintf(spath, L"%ls%ls%ls%ls",
 728                  relpath_prefix, path, colonchar, stream_name);
 729
 730         /* Make a new wim_lookup_table_entry */
 731         lte = new_lookup_table_entry();
 732         if (!lte) {
 733                 ret = WIMLIB_ERR_NOMEM;
 734                 goto out_free_spath;
 735         }
 736         lte->file_on_disk = spath;
 737         spath = NULL;
 738         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED && !is_named_stream) {
 739                 u64 encrypted_size;
 740                 lte->resource_location = RESOURCE_WIN32_ENCRYPTED;
 741                 ret = win32_get_encrypted_file_size(path, &encrypted_size);
 742                 if (ret)
 743                         goto out_free_spath;
 744                 lte->size = encrypted_size;
 745         } else {
 746                 lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 747                 lte->size = (u64)dat->StreamSize.QuadPart;
 748         }
 749
 750         u32 stream_id;
 751         if (is_named_stream) {
 752                 stream_id = ads_entry->stream_id;
 753                 ads_entry->lte = lte;
 754         } else {
 755                 stream_id = 0;
 756                 inode->i_lte = lte;
 757         }
 758         add_unhashed_stream(lte, inode, stream_id, unhashed_streams);
 759         ret = 0;
 760 out_free_spath:
 761         FREE(spath);
 762 out:
 763         return ret;
 764 out_invalid_stream_name:
 765         ERROR("Invalid stream name: \"%ls:%ls\"", path, dat->cStreamName);
 766         ret = WIMLIB_ERR_READ;
 767         goto out;
 768 }
 769
 770 /* Load information about the streams of an open file into a WIM inode.
 771  *
 772  * We use the NtQueryInformationFile() system call instead of FindFirstStream()
 773  * and FindNextStream().  This is done for two reasons:
 774  *
 775  * - FindFirstStream() opens its own handle to the file or directory and
 776  *   apparently does so without specifying FILE_FLAG_BACKUP_SEMANTICS, thereby
 777  *   causing access denied errors on certain files (even when running as the
 778  *   Administrator).
 779  * - FindFirstStream() and FindNextStream() is only available on Windows Vista
 780  *   and later, whereas the stream support in NtQueryInformationFile() was
 781  *   already present in Windows XP.
 782  */
 783 static int
 784 win32_capture_streams(HANDLE *hFile_p,
 785                       const wchar_t *path,
 786                       size_t path_num_chars,
 787                       struct wim_inode *inode,
 788                       struct list_head *unhashed_streams,
 789                       u64 file_size,
 790                       unsigned vol_flags)
 791 {
 792         int ret;
 793         u8 _buf[8192] _aligned_attribute(8);
 794         u8 *buf;
 795         size_t bufsize;
 796         IO_STATUS_BLOCK io_status;
 797         NTSTATUS status;
 798         const FILE_STREAM_INFORMATION *info;
 799
 800         DEBUG("Capturing streams from \"%ls\"", path);
 801
 802         buf = _buf;
 803         bufsize = sizeof(_buf);
 804
 805         if (!(vol_flags & FILE_NAMED_STREAMS))
 806                 goto unnamed_only;
 807
 808         /* Get a buffer containing the stream information.  */
 809         while (!NT_SUCCESS(status = (*func_NtQueryInformationFile)(*hFile_p,
 810                                                                    &io_status,
 811                                                                    buf,
 812                                                                    bufsize,
 813                                                                    FileStreamInformation)))
 814         {
 815
 816                 switch (status) {
 817                 case STATUS_BUFFER_OVERFLOW:
 818                         {
 819                                 u8 *newbuf;
 820
 821                                 bufsize *= 2;
 822                                 if (buf == _buf)
 823                                         newbuf = MALLOC(bufsize);
 824                                 else
 825                                         newbuf = REALLOC(buf, bufsize);
 826                                 if (!newbuf) {
 827                                         ret = WIMLIB_ERR_NOMEM;
 828                                         goto out_free_buf;
 829                                 }
 830                                 buf = newbuf;
 831                         }
 832                         break;
 833                 case STATUS_NOT_IMPLEMENTED:
 834                 case STATUS_NOT_SUPPORTED:
 835                 case STATUS_INVALID_INFO_CLASS:
 836                         goto unnamed_only;
 837                 default:
 838                         set_errno_from_nt_status(status);
 839                         ERROR_WITH_ERRNO("\"%ls\": Failed to query "
 840                                          "stream information", path);
 841                         ret = WIMLIB_ERR_READ;
 842                         goto out_free_buf;
 843                 }
 844         }
 845
 846         if (io_status.Information == 0) {
 847                 /* No stream information.  */
 848                 ret = 0;
 849                 goto out_free_buf;
 850         }
 851
 852         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
 853                 /* OpenEncryptedFileRaw() seems to fail with
 854                  * ERROR_SHARING_VIOLATION if there are any handles opened to
 855                  * the file.  */
 856                 CloseHandle(*hFile_p);
 857                 *hFile_p = INVALID_HANDLE_VALUE;
 858         }
 859
 860         /* Parse one or more stream information structures.  */
 861         info = (const FILE_STREAM_INFORMATION *)buf;
 862         for (;;) {
 863                 WIN32_FIND_STREAM_DATA dat;
 864
 865                 if (info->StreamNameLength <= sizeof(dat.cStreamName) - 2) {
 866                         dat.StreamSize = info->StreamSize;
 867                         memcpy(dat.cStreamName, info->StreamName, info->StreamNameLength);
 868                         dat.cStreamName[info->StreamNameLength / 2] = L'\0';
 869
 870                         /* Capture the stream.  */
 871                         ret = win32_capture_stream(path, path_num_chars, inode,
 872                                                    unhashed_streams, &dat);
 873                         if (ret)
 874                                 goto out_free_buf;
 875                 }
 876                 if (info->NextEntryOffset == 0) {
 877                         /* No more stream information.  */
 878                         break;
 879                 }
 880                 /* Advance to next stream information.  */
 881                 info = (const FILE_STREAM_INFORMATION *)
 882                                 ((const u8 *)info + info->NextEntryOffset);
 883         }
 884         ret = 0;
 885         goto out_free_buf;
 886
 887 unnamed_only:
 888         /* The volume does not support named streams.  Only capture the unnamed
 889          * data stream. */
 890         DEBUG("Only capturing unnamed data stream");
 891         if (inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
 892                                    FILE_ATTRIBUTE_REPARSE_POINT))
 893         {
 894                 ret = 0;
 895                 goto out_free_buf;
 896         }
 897
 898         {
 899                 WIN32_FIND_STREAM_DATA dat;
 900
 901                 wcscpy(dat.cStreamName, L"::$DATA");
 902                 dat.StreamSize.QuadPart = file_size;
 903                 ret = win32_capture_stream(path, path_num_chars,
 904                                            inode, unhashed_streams, &dat);
 905         }
 906 out_free_buf:
 907         /* Free buffer if allocated on heap.  */
 908         if (buf != _buf)
 909                 FREE(buf);
 910         return ret;
 911 }
 912
 913 static int
 914 win32_build_dentry_tree_recursive(struct wim_dentry **root_ret,
 915                                   wchar_t *path,
 916                                   size_t path_num_chars,
 917                                   struct add_image_params *params,
 918                                   struct win32_capture_state *state,
 919                                   unsigned vol_flags)
 920 {
 921         struct wim_dentry *root = NULL;
 922         struct wim_inode *inode = NULL;
 923         DWORD err;
 924         u64 file_size;
 925         int ret;
 926         u8 *rpbuf;
 927         u16 rpbuflen;
 928         u16 not_rpfixed;
 929         HANDLE hFile = INVALID_HANDLE_VALUE;
 930         DWORD desiredAccess;
 931
 932
 933         if (should_exclude_path(path + params->capture_root_nchars,
 934                                 path_num_chars - params->capture_root_nchars,
 935                                 params->config))
 936         {
 937                 ret = 0;
 938                 goto out_progress;
 939         }
 940
 941         desiredAccess = FILE_READ_DATA | FILE_READ_ATTRIBUTES |
 942                         READ_CONTROL | ACCESS_SYSTEM_SECURITY;
 943 again:
 944         hFile = win32_open_existing_file(path, desiredAccess);
 945         if (hFile == INVALID_HANDLE_VALUE) {
 946                 err = GetLastError();
 947                 if (err == ERROR_ACCESS_DENIED || err == ERROR_PRIVILEGE_NOT_HELD) {
 948                         if (desiredAccess & ACCESS_SYSTEM_SECURITY) {
 949                                 desiredAccess &= ~ACCESS_SYSTEM_SECURITY;
 950                                 goto again;
 951                         }
 952                         if (desiredAccess & READ_CONTROL) {
 953                                 desiredAccess &= ~READ_CONTROL;
 954                                 goto again;
 955                         }
 956                 }
 957                 set_errno_from_GetLastError();
 958                 ERROR_WITH_ERRNO("Failed to open \"%ls\" for reading", path);
 959                 ret = WIMLIB_ERR_OPEN;
 960                 goto out;
 961         }
 962
 963         BY_HANDLE_FILE_INFORMATION file_info;
 964         if (!GetFileInformationByHandle(hFile, &file_info)) {
 965                 set_errno_from_GetLastError();
 966                 ERROR_WITH_ERRNO("Failed to get file information for \"%ls\"",
 967                                  path);
 968                 ret = WIMLIB_ERR_STAT;
 969                 goto out;
 970         }
 971
 972         if (file_info.dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 973                 rpbuf = alloca(REPARSE_POINT_MAX_SIZE);
 974                 ret = win32_get_reparse_data(hFile, path, params,
 975                                              rpbuf, &rpbuflen);
 976                 if (ret < 0) {
 977                         /* WIMLIB_ERR_* (inverted) */
 978                         ret = -ret;
 979                         goto out;
 980                 } else if (ret & RP_FIXED) {
 981                         not_rpfixed = 0;
 982                 } else if (ret == RP_EXCLUDED) {
 983                         ret = 0;
 984                         goto out;
 985                 } else {
 986                         not_rpfixed = 1;
 987                 }
 988         }
 989
 990         /* Create a WIM dentry with an associated inode, which may be shared.
 991          *
 992          * However, we need to explicitly check for directories and files with
 993          * only 1 link and refuse to hard link them.  This is because Windows
 994          * has a bug where it can return duplicate File IDs for files and
 995          * directories on the FAT filesystem. */
 996         ret = inode_table_new_dentry(params->inode_table,
 997                                      path_basename_with_len(path, path_num_chars),
 998                                      ((u64)file_info.nFileIndexHigh << 32) |
 999                                          (u64)file_info.nFileIndexLow,
1000                                      file_info.dwVolumeSerialNumber,
1001                                      (file_info.nNumberOfLinks <= 1 ||
1002                                         (file_info.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY)),
1003                                      &root);
1004         if (ret)
1005                 goto out;
1006
1007         ret = win32_get_short_name(hFile, root);
1008         if (ret)
1009                 goto out;
1010
1011         inode = root->d_inode;
1012
1013         if (inode->i_nlink > 1) {
1014                 /* Shared inode; nothing more to do */
1015                 goto out_progress;
1016         }
1017
1018         inode->i_attributes = file_info.dwFileAttributes;
1019         inode->i_creation_time = FILETIME_to_u64(&file_info.ftCreationTime);
1020         inode->i_last_write_time = FILETIME_to_u64(&file_info.ftLastWriteTime);
1021         inode->i_last_access_time = FILETIME_to_u64(&file_info.ftLastAccessTime);
1022         inode->i_resolved = 1;
1023
1024         params->add_flags &= ~WIMLIB_ADD_FLAG_ROOT;
1025
1026         if (!(params->add_flags & WIMLIB_ADD_FLAG_NO_ACLS)
1027             && (vol_flags & FILE_PERSISTENT_ACLS))
1028         {
1029                 ret = win32_get_security_descriptor(hFile, path, inode,
1030                                                     params->sd_set, state,
1031                                                     params->add_flags);
1032                 if (ret)
1033                         goto out;
1034         }
1035
1036         file_size = ((u64)file_info.nFileSizeHigh << 32) |
1037                      (u64)file_info.nFileSizeLow;
1038
1039
1040         /* Capture the unnamed data stream (only should be present for regular
1041          * files) and any alternate data streams. */
1042         ret = win32_capture_streams(&hFile,
1043                                     path,
1044                                     path_num_chars,
1045                                     inode,
1046                                     params->unhashed_streams,
1047                                     file_size,
1048                                     vol_flags);
1049         if (ret)
1050                 goto out;
1051
1052         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1053                 /* Reparse point: set the reparse data (which we read already)
1054                  * */
1055                 inode->i_not_rpfixed = not_rpfixed;
1056                 inode->i_reparse_tag = le32_to_cpu(*(le32*)rpbuf);
1057                 ret = inode_set_unnamed_stream(inode, rpbuf + 8, rpbuflen - 8,
1058                                                params->lookup_table);
1059         } else if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
1060                 /* Directory (not a reparse point) --- recurse to children */
1061
1062                 if (hFile == INVALID_HANDLE_VALUE) {
1063                         /* Re-open handle that was closed to read raw encrypted
1064                          * data.  */
1065                         hFile = win32_open_existing_file(path, FILE_READ_DATA);
1066                         if (hFile == INVALID_HANDLE_VALUE) {
1067                                 set_errno_from_GetLastError();
1068                                 ERROR_WITH_ERRNO("Failed to reopen \"%ls\"",
1069                                                  path);
1070                                 ret = WIMLIB_ERR_OPEN;
1071                                 goto out;
1072                         }
1073                 }
1074                 ret = win32_recurse_directory(hFile,
1075                                               path,
1076                                               path_num_chars,
1077                                               root,
1078                                               params,
1079                                               state,
1080                                               vol_flags);
1081         }
1082         if (ret)
1083                 goto out;
1084
1085         path[path_num_chars] = '\0';
1086 out_progress:
1087         params->progress.scan.cur_path = path;
1088         if (root == NULL)
1089                 do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
1090         else
1091                 do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
1092 out:
1093         if (hFile != INVALID_HANDLE_VALUE)
1094                 CloseHandle(hFile);
1095         if (ret == 0)
1096                 *root_ret = root;
1097         else
1098                 free_dentry_tree(root, params->lookup_table);
1099         return ret;
1100 }
1101
1102 static void
1103 win32_do_capture_warnings(const wchar_t *path,
1104                           const struct win32_capture_state *state,
1105                           int add_flags)
1106 {
1107         if (state->num_get_sacl_priv_notheld == 0 &&
1108             state->num_get_sd_access_denied == 0)
1109                 return;
1110
1111         WARNING("Scan of \"%ls\" complete, but with one or more warnings:", path);
1112         if (state->num_get_sacl_priv_notheld != 0) {
1113                 WARNING("- Could not capture SACL (System Access Control List)\n"
1114                         "            on %lu files or directories.",
1115                         state->num_get_sacl_priv_notheld);
1116         }
1117         if (state->num_get_sd_access_denied != 0) {
1118                 WARNING("- Could not capture security descriptor at all\n"
1119                         "            on %lu files or directories.",
1120                         state->num_get_sd_access_denied);
1121         }
1122         WARNING("To fully capture all security descriptors, run the program\n"
1123                 "          with Administrator rights.");
1124 }
1125
1126 #define WINDOWS_NT_MAX_PATH 32768
1127
1128 /* Win32 version of capturing a directory tree */
1129 int
1130 win32_build_dentry_tree(struct wim_dentry **root_ret,
1131                         const wchar_t *root_disk_path,
1132                         struct add_image_params *params)
1133 {
1134         size_t path_nchars;
1135         wchar_t *path;
1136         int ret;
1137         struct win32_capture_state state;
1138         unsigned vol_flags;
1139         DWORD dret;
1140
1141         path_nchars = wcslen(root_disk_path);
1142         if (path_nchars > WINDOWS_NT_MAX_PATH)
1143                 return WIMLIB_ERR_INVALID_PARAM;
1144
1145         ret = win32_get_file_and_vol_ids(root_disk_path,
1146                                          &params->capture_root_ino,
1147                                          &params->capture_root_dev);
1148         if (ret) {
1149                 ERROR_WITH_ERRNO("Can't open %ls", root_disk_path);
1150                 return ret;
1151         }
1152
1153         win32_get_vol_flags(root_disk_path, &vol_flags, NULL);
1154
1155         /* WARNING: There is no check for overflow later when this buffer is
1156          * being used!  But it's as long as the maximum path length understood
1157          * by Windows NT (which is NOT the same as MAX_PATH). */
1158         path = MALLOC((WINDOWS_NT_MAX_PATH + 1) * sizeof(wchar_t));
1159         if (path == NULL)
1160                 return WIMLIB_ERR_NOMEM;
1161
1162         /* Work around defective behavior in Windows where paths longer than 260
1163          * characters are not supported by default; instead they need to be
1164          * turned into absolute paths and prefixed with "\\?\".  */
1165
1166         if (wcsncmp(root_disk_path, L"\\\\?\\", 4)) {
1167                 dret = GetFullPathName(root_disk_path, WINDOWS_NT_MAX_PATH - 3,
1168                                        &path[4], NULL);
1169
1170                 if (dret == 0 || dret >= WINDOWS_NT_MAX_PATH - 3) {
1171                         WARNING("Can't get full path name for \"%ls\"", root_disk_path);
1172                         wmemcpy(path, root_disk_path, path_nchars + 1);
1173                 } else {
1174                         wmemcpy(path, L"\\\\?\\", 4);
1175                         path_nchars = 4 + dret;
1176                 }
1177         } else {
1178                 wmemcpy(path, root_disk_path, path_nchars + 1);
1179         }
1180
1181        /* Strip trailing slashes.  If we don't do this, we may create a path
1182         * with multiple consecutive backslashes, which for some reason causes
1183         * Windows to report that the file cannot be found.  */
1184         while (path_nchars >= 2 &&
1185                path[path_nchars - 1] == L'\\' &&
1186                path[path_nchars - 2] != L':')
1187         {
1188                 path[--path_nchars] = L'\0';
1189         }
1190
1191         params->capture_root_nchars = path_nchars;
1192
1193         memset(&state, 0, sizeof(state));
1194         ret = win32_build_dentry_tree_recursive(root_ret, path,
1195                                                 path_nchars, params,
1196                                                 &state, vol_flags);
1197         FREE(path);
1198         if (ret == 0)
1199                 win32_do_capture_warnings(root_disk_path, &state, params->add_flags);
1200         return ret;
1201 }
1202
1203 #endif /* __WIN32__ */