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  * This now uses the native Windows NT API a lot and not just Win32.
   5  */
   6
   7 /*
   8  * Copyright (C) 2013, 2014, 2015 Eric Biggers
   9  *
  10  * This file is free software; you can redistribute it and/or modify it under
  11  * the terms of the GNU Lesser General Public License as published by the Free
  12  * Software Foundation; either version 3 of the License, or (at your option) any
  13  * later version.
  14  *
  15  * This file is distributed in the hope that it will be useful, but WITHOUT
  16  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  17  * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  18  * details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public License
  21  * along with this file; 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/assert.h"
  33 #include "wimlib/blob_table.h"
  34 #include "wimlib/capture.h"
  35 #include "wimlib/dentry.h"
  36 #include "wimlib/encoding.h"
  37 #include "wimlib/endianness.h"
  38 #include "wimlib/error.h"
  39 #include "wimlib/paths.h"
  40 #include "wimlib/reparse.h"
  41
  42 struct winnt_scan_stats {
  43         unsigned long num_get_sd_access_denied;
  44         unsigned long num_get_sacl_priv_notheld;
  45 };
  46
  47 static inline const wchar_t *
  48 printable_path(const wchar_t *full_path)
  49 {
  50         /* Skip over \\?\ or \??\  */
  51         return full_path + 4;
  52 }
  53
  54 /*
  55  * If cur_dir is not NULL, open an existing file relative to the already-open
  56  * directory cur_dir.
  57  *
  58  * Otherwise, open the file specified by @path, which must be a Windows NT
  59  * namespace path.
  60  */
  61 static NTSTATUS
  62 winnt_openat(HANDLE cur_dir, const wchar_t *path, size_t path_nchars,
  63              ACCESS_MASK perms, HANDLE *h_ret)
  64 {
  65         UNICODE_STRING name;
  66         OBJECT_ATTRIBUTES attr;
  67         IO_STATUS_BLOCK iosb;
  68         NTSTATUS status;
  69
  70         name.Length = path_nchars * sizeof(wchar_t);
  71         name.MaximumLength = name.Length + sizeof(wchar_t);
  72         name.Buffer = (wchar_t *)path;
  73
  74         attr.Length = sizeof(attr);
  75         attr.RootDirectory = cur_dir;
  76         attr.ObjectName = &name;
  77         attr.Attributes = 0;
  78         attr.SecurityDescriptor = NULL;
  79         attr.SecurityQualityOfService = NULL;
  80
  81 retry:
  82         status = (*func_NtOpenFile)(h_ret, perms, &attr, &iosb,
  83                                     FILE_SHARE_VALID_FLAGS,
  84                                     FILE_OPEN_REPARSE_POINT |
  85                                             FILE_OPEN_FOR_BACKUP_INTENT |
  86                                             FILE_SYNCHRONOUS_IO_NONALERT |
  87                                             FILE_SEQUENTIAL_ONLY);
  88         if (!NT_SUCCESS(status)) {
  89                 /* Try requesting fewer permissions  */
  90                 if (status == STATUS_ACCESS_DENIED ||
  91                     status == STATUS_PRIVILEGE_NOT_HELD) {
  92                         if (perms & ACCESS_SYSTEM_SECURITY) {
  93                                 perms &= ~ACCESS_SYSTEM_SECURITY;
  94                                 goto retry;
  95                         }
  96                         if (perms & READ_CONTROL) {
  97                                 perms &= ~READ_CONTROL;
  98                                 goto retry;
  99                         }
 100                 }
 101         }
 102         return status;
 103 }
 104
 105 /* Read the first @size bytes from the file, or named data stream of a file,
 106  * described by @blob.  */
 107 int
 108 read_winnt_stream_prefix(const struct blob_descriptor *blob, u64 size,
 109                          const struct read_blob_callbacks *cbs)
 110 {
 111         const wchar_t *path;
 112         HANDLE h;
 113         NTSTATUS status;
 114         u8 buf[BUFFER_SIZE];
 115         u64 bytes_remaining;
 116         int ret;
 117
 118         /* This is an NT namespace path.  */
 119         path = blob->file_on_disk;
 120
 121         status = winnt_openat(NULL, path, wcslen(path),
 122                               FILE_READ_DATA | SYNCHRONIZE, &h);
 123         if (!NT_SUCCESS(status)) {
 124                 winnt_error(status, L"\"%ls\": Can't open for reading",
 125                             printable_path(path));
 126                 return WIMLIB_ERR_OPEN;
 127         }
 128
 129         ret = 0;
 130         bytes_remaining = size;
 131         while (bytes_remaining) {
 132                 IO_STATUS_BLOCK iosb;
 133                 ULONG count;
 134                 ULONG bytes_read;
 135
 136                 count = min(sizeof(buf), bytes_remaining);
 137
 138                 status = (*func_NtReadFile)(h, NULL, NULL, NULL,
 139                                             &iosb, buf, count, NULL, NULL);
 140                 if (unlikely(!NT_SUCCESS(status))) {
 141                         if (status == STATUS_END_OF_FILE) {
 142                                 ERROR("\"%ls\": File was concurrently truncated",
 143                                       printable_path(path));
 144                                 ret = WIMLIB_ERR_CONCURRENT_MODIFICATION_DETECTED;
 145                         } else {
 146                                 winnt_error(status, L"\"%ls\": Error reading data",
 147                                             printable_path(path));
 148                                 ret = WIMLIB_ERR_READ;
 149                         }
 150                         break;
 151                 }
 152
 153                 bytes_read = iosb.Information;
 154
 155                 bytes_remaining -= bytes_read;
 156                 ret = call_consume_chunk(buf, bytes_read, cbs);
 157                 if (ret)
 158                         break;
 159         }
 160         (*func_NtClose)(h);
 161         return ret;
 162 }
 163
 164 struct win32_encrypted_read_ctx {
 165         const struct read_blob_callbacks *cbs;
 166         int wimlib_err_code;
 167         u64 bytes_remaining;
 168 };
 169
 170 static DWORD WINAPI
 171 win32_encrypted_export_cb(unsigned char *data, void *_ctx, unsigned long len)
 172 {
 173         struct win32_encrypted_read_ctx *ctx = _ctx;
 174         int ret;
 175         size_t bytes_to_consume = min(len, ctx->bytes_remaining);
 176
 177         if (bytes_to_consume == 0)
 178                 return ERROR_SUCCESS;
 179
 180         ret = call_consume_chunk(data, bytes_to_consume, ctx->cbs);
 181         if (ret) {
 182                 ctx->wimlib_err_code = ret;
 183                 /* It doesn't matter what error code is returned here, as long
 184                  * as it isn't ERROR_SUCCESS.  */
 185                 return ERROR_READ_FAULT;
 186         }
 187         ctx->bytes_remaining -= bytes_to_consume;
 188         return ERROR_SUCCESS;
 189 }
 190
 191 int
 192 read_win32_encrypted_file_prefix(const struct blob_descriptor *blob,
 193                                  u64 size,
 194                                  const struct read_blob_callbacks *cbs)
 195 {
 196         struct win32_encrypted_read_ctx export_ctx;
 197         DWORD err;
 198         void *file_ctx;
 199         int ret;
 200         DWORD flags = 0;
 201
 202         if (blob->file_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
 203                 flags |= CREATE_FOR_DIR;
 204
 205         export_ctx.cbs = cbs;
 206         export_ctx.wimlib_err_code = 0;
 207         export_ctx.bytes_remaining = size;
 208
 209         err = OpenEncryptedFileRaw(blob->file_on_disk, flags, &file_ctx);
 210         if (err != ERROR_SUCCESS) {
 211                 win32_error(err,
 212                             L"Failed to open encrypted file \"%ls\" for raw read",
 213                             printable_path(blob->file_on_disk));
 214                 return WIMLIB_ERR_OPEN;
 215         }
 216         err = ReadEncryptedFileRaw(win32_encrypted_export_cb,
 217                                    &export_ctx, file_ctx);
 218         if (err != ERROR_SUCCESS) {
 219                 ret = export_ctx.wimlib_err_code;
 220                 if (ret == 0) {
 221                         win32_error(err,
 222                                     L"Failed to read encrypted file \"%ls\"",
 223                                     printable_path(blob->file_on_disk));
 224                         ret = WIMLIB_ERR_READ;
 225                 }
 226         } else if (export_ctx.bytes_remaining != 0) {
 227                 ERROR("Only could read %"PRIu64" of %"PRIu64" bytes from "
 228                       "encrypted file \"%ls\"",
 229                       size - export_ctx.bytes_remaining, size,
 230                       printable_path(blob->file_on_disk));
 231                 ret = WIMLIB_ERR_READ;
 232         } else {
 233                 ret = 0;
 234         }
 235         CloseEncryptedFileRaw(file_ctx);
 236         return ret;
 237 }
 238
 239 /*
 240  * Load the short name of a file into a WIM dentry.
 241  */
 242 static noinline_for_stack NTSTATUS
 243 winnt_get_short_name(HANDLE h, struct wim_dentry *dentry)
 244 {
 245         /* It's not any harder to just make the NtQueryInformationFile() system
 246          * call ourselves, and it saves a dumb call to FindFirstFile() which of
 247          * course has to create its own handle.  */
 248         NTSTATUS status;
 249         IO_STATUS_BLOCK iosb;
 250         u8 buf[128] _aligned_attribute(8);
 251         const FILE_NAME_INFORMATION *info;
 252
 253         status = (*func_NtQueryInformationFile)(h, &iosb, buf, sizeof(buf),
 254                                                 FileAlternateNameInformation);
 255         info = (const FILE_NAME_INFORMATION *)buf;
 256         if (NT_SUCCESS(status) && info->FileNameLength != 0) {
 257                 dentry->d_short_name = utf16le_dupz(info->FileName,
 258                                                     info->FileNameLength);
 259                 if (!dentry->d_short_name)
 260                         return STATUS_NO_MEMORY;
 261                 dentry->d_short_name_nbytes = info->FileNameLength;
 262         }
 263         return status;
 264 }
 265
 266 /*
 267  * Load the security descriptor of a file into the corresponding inode and the
 268  * WIM image's security descriptor set.
 269  */
 270 static noinline_for_stack NTSTATUS
 271 winnt_get_security_descriptor(HANDLE h, struct wim_inode *inode,
 272                               struct wim_sd_set *sd_set,
 273                               struct winnt_scan_stats *stats, int add_flags)
 274 {
 275         SECURITY_INFORMATION requestedInformation;
 276         u8 _buf[4096] _aligned_attribute(8);
 277         u8 *buf;
 278         ULONG bufsize;
 279         ULONG len_needed;
 280         NTSTATUS status;
 281
 282         /*
 283          * LABEL_SECURITY_INFORMATION is needed on Windows Vista and 7 because
 284          * Microsoft decided to add mandatory integrity labels to the SACL but
 285          * not have them returned by SACL_SECURITY_INFORMATION.
 286          *
 287          * BACKUP_SECURITY_INFORMATION is needed on Windows 8 because Microsoft
 288          * decided to add even more stuff to the SACL and still not have it
 289          * returned by SACL_SECURITY_INFORMATION; but they did remember that
 290          * backup applications exist and simply want to read the stupid thing
 291          * once and for all, so they added a flag to read the entire security
 292          * descriptor.
 293          *
 294          * Older versions of Windows tolerate these new flags being passed in.
 295          */
 296         requestedInformation = OWNER_SECURITY_INFORMATION |
 297                                GROUP_SECURITY_INFORMATION |
 298                                DACL_SECURITY_INFORMATION |
 299                                SACL_SECURITY_INFORMATION |
 300                                LABEL_SECURITY_INFORMATION |
 301                                BACKUP_SECURITY_INFORMATION;
 302
 303         buf = _buf;
 304         bufsize = sizeof(_buf);
 305
 306         /*
 307          * We need the file's security descriptor in
 308          * SECURITY_DESCRIPTOR_RELATIVE format, and we currently have a handle
 309          * opened with as many relevant permissions as possible.  At this point,
 310          * on Windows there are a number of options for reading a file's
 311          * security descriptor:
 312          *
 313          * GetFileSecurity():  This takes in a path and returns the
 314          * SECURITY_DESCRIPTOR_RELATIVE.  Problem: this uses an internal handle,
 315          * not ours, and the handle created internally doesn't specify
 316          * FILE_FLAG_BACKUP_SEMANTICS.  Therefore there can be access denied
 317          * errors on some files and directories, even when running as the
 318          * Administrator.
 319          *
 320          * GetSecurityInfo():  This takes in a handle and returns the security
 321          * descriptor split into a bunch of different parts.  This should work,
 322          * but it's dumb because we have to put the security descriptor back
 323          * together again.
 324          *
 325          * BackupRead():  This can read the security descriptor, but this is a
 326          * difficult-to-use API, probably only works as the Administrator, and
 327          * the format of the returned data is not well documented.
 328          *
 329          * NtQuerySecurityObject():  This is exactly what we need, as it takes
 330          * in a handle and returns the security descriptor in
 331          * SECURITY_DESCRIPTOR_RELATIVE format.  Only problem is that it's a
 332          * ntdll function and therefore not officially part of the Win32 API.
 333          * Oh well.
 334          */
 335         while (!(NT_SUCCESS(status = (*func_NtQuerySecurityObject)(h,
 336                                                                    requestedInformation,
 337                                                                    (PSECURITY_DESCRIPTOR)buf,
 338                                                                    bufsize,
 339                                                                    &len_needed))))
 340         {
 341                 switch (status) {
 342                 case STATUS_BUFFER_TOO_SMALL:
 343                         wimlib_assert(buf == _buf);
 344                         buf = MALLOC(len_needed);
 345                         if (!buf)
 346                                 return STATUS_NO_MEMORY;
 347                         bufsize = len_needed;
 348                         break;
 349                 case STATUS_PRIVILEGE_NOT_HELD:
 350                 case STATUS_ACCESS_DENIED:
 351                         if (add_flags & WIMLIB_ADD_FLAG_STRICT_ACLS) {
 352                 default:
 353                                 /* Permission denied in STRICT_ACLS mode, or
 354                                  * unknown error.  */
 355                                 goto out_free_buf;
 356                         }
 357                         if (requestedInformation & SACL_SECURITY_INFORMATION) {
 358                                 /* Try again without the SACL.  */
 359                                 stats->num_get_sacl_priv_notheld++;
 360                                 requestedInformation &= ~(SACL_SECURITY_INFORMATION |
 361                                                           LABEL_SECURITY_INFORMATION |
 362                                                           BACKUP_SECURITY_INFORMATION);
 363                                 break;
 364                         }
 365                         /* Fake success (useful when capturing as
 366                          * non-Administrator).  */
 367                         stats->num_get_sd_access_denied++;
 368                         status = STATUS_SUCCESS;
 369                         goto out_free_buf;
 370                 }
 371         }
 372
 373         /* Add the security descriptor to the WIM image, and save its ID in
 374          * file's inode.  */
 375         inode->i_security_id = sd_set_add_sd(sd_set, buf, len_needed);
 376         if (unlikely(inode->i_security_id < 0))
 377                 status = STATUS_NO_MEMORY;
 378 out_free_buf:
 379         if (unlikely(buf != _buf))
 380                 FREE(buf);
 381         return status;
 382 }
 383
 384 static int
 385 winnt_build_dentry_tree_recursive(struct wim_dentry **root_ret,
 386                                   HANDLE cur_dir,
 387                                   wchar_t *full_path,
 388                                   size_t full_path_nchars,
 389                                   const wchar_t *filename,
 390                                   size_t filename_nchars,
 391                                   struct capture_params *params,
 392                                   struct winnt_scan_stats *stats,
 393                                   u32 vol_flags);
 394
 395 static int
 396 winnt_recurse_directory(HANDLE h,
 397                         wchar_t *full_path,
 398                         size_t full_path_nchars,
 399                         struct wim_dentry *parent,
 400                         struct capture_params *params,
 401                         struct winnt_scan_stats *stats,
 402                         u32 vol_flags)
 403 {
 404         void *buf;
 405         const size_t bufsize = 8192;
 406         IO_STATUS_BLOCK iosb;
 407         NTSTATUS status;
 408         int ret;
 409
 410         buf = MALLOC(bufsize);
 411         if (!buf)
 412                 return WIMLIB_ERR_NOMEM;
 413
 414         /* Using NtQueryDirectoryFile() we can re-use the same open handle,
 415          * which we opened with FILE_FLAG_BACKUP_SEMANTICS.  */
 416
 417         while (NT_SUCCESS(status = (*func_NtQueryDirectoryFile)(h, NULL, NULL, NULL,
 418                                                                 &iosb, buf, bufsize,
 419                                                                 FileNamesInformation,
 420                                                                 FALSE, NULL, FALSE)))
 421         {
 422                 const FILE_NAMES_INFORMATION *info = buf;
 423                 for (;;) {
 424                         if (!should_ignore_filename(info->FileName,
 425                                                     info->FileNameLength / 2))
 426                         {
 427                                 wchar_t *p;
 428                                 wchar_t *filename;
 429                                 struct wim_dentry *child;
 430
 431                                 p = full_path + full_path_nchars;
 432                                 /* Only add a backslash if we don't already have
 433                                  * one.  This prevents a duplicate backslash
 434                                  * from being added when the path to the capture
 435                                  * dir had a trailing backslash.  */
 436                                 if (*(p - 1) != L'\\')
 437                                         *p++ = L'\\';
 438                                 filename = p;
 439                                 p = wmempcpy(filename, info->FileName,
 440                                              info->FileNameLength / 2);
 441                                 *p = '\0';
 442
 443                                 ret = winnt_build_dentry_tree_recursive(
 444                                                         &child,
 445                                                         h,
 446                                                         full_path,
 447                                                         p - full_path,
 448                                                         filename,
 449                                                         info->FileNameLength / 2,
 450                                                         params,
 451                                                         stats,
 452                                                         vol_flags);
 453
 454                                 full_path[full_path_nchars] = L'\0';
 455
 456                                 if (ret)
 457                                         goto out_free_buf;
 458                                 attach_scanned_tree(parent, child, params->blob_table);
 459                         }
 460                         if (info->NextEntryOffset == 0)
 461                                 break;
 462                         info = (const FILE_NAMES_INFORMATION *)
 463                                         ((const u8 *)info + info->NextEntryOffset);
 464                 }
 465         }
 466
 467         if (unlikely(status != STATUS_NO_MORE_FILES)) {
 468                 winnt_error(status, L"\"%ls\": Can't read directory",
 469                             printable_path(full_path));
 470                 ret = WIMLIB_ERR_READ;
 471         }
 472 out_free_buf:
 473         FREE(buf);
 474         return ret;
 475 }
 476
 477 /* Reparse point fixup status code  */
 478 #define RP_FIXED        (-1)
 479
 480 static bool
 481 file_has_ino_and_dev(HANDLE h, u64 ino, u64 dev)
 482 {
 483         NTSTATUS status;
 484         IO_STATUS_BLOCK iosb;
 485         FILE_INTERNAL_INFORMATION int_info;
 486         FILE_FS_VOLUME_INFORMATION vol_info;
 487
 488         status = (*func_NtQueryInformationFile)(h, &iosb,
 489                                                 &int_info, sizeof(int_info),
 490                                                 FileInternalInformation);
 491         if (!NT_SUCCESS(status))
 492                 return false;
 493
 494         if (int_info.IndexNumber.QuadPart != ino)
 495                 return false;
 496
 497         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
 498                                                       &vol_info, sizeof(vol_info),
 499                                                       FileFsVolumeInformation);
 500         if (!(NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW))
 501                 return false;
 502
 503         if (iosb.Information <
 504              offsetof(FILE_FS_VOLUME_INFORMATION, VolumeSerialNumber) +
 505              sizeof(vol_info.VolumeSerialNumber))
 506                 return false;
 507
 508         return (vol_info.VolumeSerialNumber == dev);
 509 }
 510
 511 /*
 512  * This is the Windows equivalent of unix_relativize_link_target(); see there
 513  * for general details.  This version works with an "absolute" Windows link
 514  * target, specified from the root of the Windows kernel object namespace.  Note
 515  * that we have to open directories with a trailing slash when present because
 516  * \??\E: opens the E: device itself and not the filesystem root directory.
 517  */
 518 static const wchar_t *
 519 winnt_relativize_link_target(const wchar_t *target, size_t target_nbytes,
 520                              u64 ino, u64 dev)
 521 {
 522         UNICODE_STRING name;
 523         OBJECT_ATTRIBUTES attr;
 524         IO_STATUS_BLOCK iosb;
 525         NTSTATUS status;
 526         const wchar_t *target_end;
 527         const wchar_t *p;
 528
 529         target_end = target + (target_nbytes / sizeof(wchar_t));
 530
 531         /* Empty path??? */
 532         if (target_end == target)
 533                 return target;
 534
 535         /* No leading slash???  */
 536         if (target[0] != L'\\')
 537                 return target;
 538
 539         /* UNC path???  */
 540         if ((target_end - target) >= 2 &&
 541             target[0] == L'\\' && target[1] == L'\\')
 542                 return target;
 543
 544         attr.Length = sizeof(attr);
 545         attr.RootDirectory = NULL;
 546         attr.ObjectName = &name;
 547         attr.Attributes = 0;
 548         attr.SecurityDescriptor = NULL;
 549         attr.SecurityQualityOfService = NULL;
 550
 551         name.Buffer = (wchar_t *)target;
 552         name.Length = 0;
 553         p = target;
 554         do {
 555                 HANDLE h;
 556                 const wchar_t *orig_p = p;
 557
 558                 /* Skip non-backslashes  */
 559                 while (p != target_end && *p != L'\\')
 560                         p++;
 561
 562                 /* Skip backslashes  */
 563                 while (p != target_end && *p == L'\\')
 564                         p++;
 565
 566                 /* Append path component  */
 567                 name.Length += (p - orig_p) * sizeof(wchar_t);
 568                 name.MaximumLength = name.Length;
 569
 570                 /* Try opening the file  */
 571                 status = (*func_NtOpenFile) (&h,
 572                                              FILE_READ_ATTRIBUTES | FILE_TRAVERSE,
 573                                              &attr,
 574                                              &iosb,
 575                                              FILE_SHARE_VALID_FLAGS,
 576                                              FILE_OPEN_FOR_BACKUP_INTENT);
 577
 578                 if (NT_SUCCESS(status)) {
 579                         /* Reset root directory  */
 580                         if (attr.RootDirectory)
 581                                 (*func_NtClose)(attr.RootDirectory);
 582                         attr.RootDirectory = h;
 583                         name.Buffer = (wchar_t *)p;
 584                         name.Length = 0;
 585
 586                         if (file_has_ino_and_dev(h, ino, dev))
 587                                 goto out_close_root_dir;
 588                 }
 589         } while (p != target_end);
 590
 591         p = target;
 592
 593 out_close_root_dir:
 594         if (attr.RootDirectory)
 595                 (*func_NtClose)(attr.RootDirectory);
 596         while (p > target && *(p - 1) == L'\\')
 597                 p--;
 598         return p;
 599 }
 600
 601 static int
 602 winnt_rpfix_progress(struct capture_params *params, const wchar_t *path,
 603                      const struct link_reparse_point *link, int scan_status)
 604 {
 605         size_t print_name_nchars = link->print_name_nbytes / sizeof(wchar_t);
 606         wchar_t print_name0[print_name_nchars + 1];
 607
 608         wmemcpy(print_name0, link->print_name, print_name_nchars);
 609         print_name0[print_name_nchars] = L'\0';
 610
 611         params->progress.scan.cur_path = printable_path(path);
 612         params->progress.scan.symlink_target = print_name0;
 613         return do_capture_progress(params, scan_status, NULL);
 614 }
 615
 616 static int
 617 winnt_try_rpfix(struct reparse_buffer_disk *rpbuf, u16 *rpbuflen_p,
 618                 const wchar_t *path, struct capture_params *params)
 619 {
 620         struct link_reparse_point link;
 621         const wchar_t *rel_target;
 622         int ret;
 623
 624         if (parse_link_reparse_point(rpbuf, *rpbuflen_p, &link)) {
 625                 /* Couldn't understand the reparse data; don't do the fixup.  */
 626                 return 0;
 627         }
 628
 629         /*
 630          * Don't do reparse point fixups on relative symbolic links.
 631          *
 632          * On Windows, a relative symbolic link is supposed to be identifiable
 633          * by having reparse tag WIM_IO_REPARSE_TAG_SYMLINK and flags
 634          * SYMBOLIC_LINK_RELATIVE.  We will use this information, although this
 635          * may not always do what the user expects, since drive-relative
 636          * symbolic links such as "\Users\Public" have SYMBOLIC_LINK_RELATIVE
 637          * set, in addition to truly relative symbolic links such as "Users" or
 638          * "Users\Public".  However, WIMGAPI (as of Windows 8.1) has this same
 639          * behavior.
 640          *
 641          * Otherwise, as far as I can tell, the targets of symbolic links that
 642          * are NOT relative, as well as junctions (note: a mountpoint is the
 643          * sames thing as a junction), must be NT namespace paths, for example:
 644          *
 645          *     - \??\e:\Users\Public
 646          *     - \DosDevices\e:\Users\Public
 647          *     - \Device\HardDiskVolume4\Users\Public
 648          *     - \??\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 649          *     - \DosDevices\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 650          */
 651         if (link_is_relative_symlink(&link))
 652                 return 0;
 653
 654         rel_target = winnt_relativize_link_target(link.substitute_name,
 655                                                   link.substitute_name_nbytes,
 656                                                   params->capture_root_ino,
 657                                                   params->capture_root_dev);
 658
 659         if (rel_target == link.substitute_name) {
 660                 /* Target points outside of the tree being captured or had an
 661                  * unrecognized path format.  Don't adjust it.  */
 662                 return winnt_rpfix_progress(params, path, &link,
 663                                             WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK);
 664         }
 665
 666         /* We have an absolute target pointing within the directory being
 667          * captured. @rel_target is the suffix of the link target that is the
 668          * part relative to the directory being captured.
 669          *
 670          * We will cut off the prefix before this part (which is the path to the
 671          * directory being captured) and add a dummy prefix.  Since the process
 672          * will need to be reversed when applying the image, it doesn't matter
 673          * what exactly the prefix is, as long as it looks like an absolute
 674          * path.  */
 675
 676         static const wchar_t prefix[6] = L"\\??\\X:";
 677         static const size_t num_unprintable_chars = 4;
 678
 679         size_t rel_target_nbytes =
 680                 link.substitute_name_nbytes - ((const u8 *)rel_target -
 681                                                (const u8 *)link.substitute_name);
 682
 683         wchar_t tmp[(sizeof(prefix) + rel_target_nbytes) / sizeof(wchar_t)];
 684
 685         memcpy(tmp, prefix, sizeof(prefix));
 686         memcpy(tmp + ARRAY_LEN(prefix), rel_target, rel_target_nbytes);
 687
 688         link.substitute_name = tmp;
 689         link.substitute_name_nbytes = sizeof(tmp);
 690
 691         link.print_name = link.substitute_name + num_unprintable_chars;
 692         link.print_name_nbytes = link.substitute_name_nbytes -
 693                                  (num_unprintable_chars * sizeof(wchar_t));
 694
 695         if (make_link_reparse_point(&link, rpbuf, rpbuflen_p))
 696                 return 0;
 697
 698         ret = winnt_rpfix_progress(params, path, &link,
 699                                    WIMLIB_SCAN_DENTRY_FIXED_SYMLINK);
 700         if (ret)
 701                 return ret;
 702         return RP_FIXED;
 703 }
 704
 705 /* Load the reparse data of a file into the corresponding WIM inode.  If the
 706  * reparse point is a symbolic link or junction with an absolute target and
 707  * RPFIX mode is enabled, then also rewrite its target to be relative to the
 708  * capture root.  */
 709 static noinline_for_stack int
 710 winnt_load_reparse_data(HANDLE h, struct wim_inode *inode,
 711                         const wchar_t *full_path, struct capture_params *params)
 712 {
 713         struct reparse_buffer_disk rpbuf;
 714         DWORD bytes_returned;
 715         u16 rpbuflen;
 716         int ret;
 717
 718         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
 719                 /* See comment above assign_stream_types_encrypted()  */
 720                 WARNING("Ignoring reparse data of encrypted file \"%ls\"",
 721                         printable_path(full_path));
 722                 return 0;
 723         }
 724
 725         if (!DeviceIoControl(h, FSCTL_GET_REPARSE_POINT,
 726                              NULL, 0, &rpbuf, REPARSE_POINT_MAX_SIZE,
 727                              &bytes_returned, NULL))
 728         {
 729                 win32_error(GetLastError(), L"\"%ls\": Can't get reparse point",
 730                             printable_path(full_path));
 731                 return WIMLIB_ERR_READLINK;
 732         }
 733
 734         rpbuflen = bytes_returned;
 735
 736         if (unlikely(rpbuflen < REPARSE_DATA_OFFSET)) {
 737                 ERROR("\"%ls\": reparse point buffer is too short",
 738                       printable_path(full_path));
 739                 return WIMLIB_ERR_INVALID_REPARSE_DATA;
 740         }
 741
 742         if (params->add_flags & WIMLIB_ADD_FLAG_RPFIX) {
 743                 ret = winnt_try_rpfix(&rpbuf, &rpbuflen, full_path, params);
 744                 if (ret == RP_FIXED)
 745                         inode->i_rp_flags &= ~WIM_RP_FLAG_NOT_FIXED;
 746                 else if (ret)
 747                         return ret;
 748         }
 749
 750         inode->i_reparse_tag = le32_to_cpu(rpbuf.rptag);
 751         inode->i_rp_reserved = le16_to_cpu(rpbuf.rpreserved);
 752
 753         if (!inode_add_stream_with_data(inode,
 754                                         STREAM_TYPE_REPARSE_POINT,
 755                                         NO_STREAM_NAME,
 756                                         rpbuf.rpdata,
 757                                         rpbuflen - REPARSE_DATA_OFFSET,
 758                                         params->blob_table))
 759                 return WIMLIB_ERR_NOMEM;
 760
 761         return 0;
 762 }
 763
 764 static DWORD WINAPI
 765 win32_tally_encrypted_size_cb(unsigned char *_data, void *_size_ret,
 766                               unsigned long len)
 767 {
 768         *(u64*)_size_ret += len;
 769         return ERROR_SUCCESS;
 770 }
 771
 772 static int
 773 win32_get_encrypted_file_size(const wchar_t *path, bool is_dir, u64 *size_ret)
 774 {
 775         DWORD err;
 776         void *file_ctx;
 777         int ret;
 778         DWORD flags = 0;
 779
 780         if (is_dir)
 781                 flags |= CREATE_FOR_DIR;
 782
 783         err = OpenEncryptedFileRaw(path, flags, &file_ctx);
 784         if (err != ERROR_SUCCESS) {
 785                 win32_error(err,
 786                             L"Failed to open encrypted file \"%ls\" for raw read",
 787                             printable_path(path));
 788                 return WIMLIB_ERR_OPEN;
 789         }
 790         *size_ret = 0;
 791         err = ReadEncryptedFileRaw(win32_tally_encrypted_size_cb,
 792                                    size_ret, file_ctx);
 793         if (err != ERROR_SUCCESS) {
 794                 win32_error(err,
 795                             L"Failed to read raw encrypted data from \"%ls\"",
 796                             printable_path(path));
 797                 ret = WIMLIB_ERR_READ;
 798         } else {
 799                 ret = 0;
 800         }
 801         CloseEncryptedFileRaw(file_ctx);
 802         return ret;
 803 }
 804
 805 static int
 806 winnt_scan_efsrpc_raw_data(struct wim_inode *inode, const wchar_t *nt_path,
 807                            struct list_head *unhashed_blobs)
 808 {
 809         struct blob_descriptor *blob;
 810         struct wim_inode_stream *strm;
 811         int ret;
 812
 813         blob = new_blob_descriptor();
 814         if (!blob)
 815                 goto err_nomem;
 816
 817         blob->file_on_disk = WCSDUP(nt_path);
 818         if (!blob->file_on_disk)
 819                 goto err_nomem;
 820         blob->blob_location = BLOB_WIN32_ENCRYPTED;
 821
 822         /* OpenEncryptedFileRaw() expects a Win32 name.  */
 823         wimlib_assert(!wmemcmp(blob->file_on_disk, L"\\??\\", 4));
 824         blob->file_on_disk[1] = L'\\';
 825
 826         blob->file_inode = inode;
 827
 828         ret = win32_get_encrypted_file_size(blob->file_on_disk,
 829                                             (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY),
 830                                             &blob->size);
 831         if (ret)
 832                 goto err;
 833
 834         /* Empty EFSRPC data does not make sense  */
 835         wimlib_assert(blob->size != 0);
 836
 837         strm = inode_add_stream(inode, STREAM_TYPE_EFSRPC_RAW_DATA,
 838                                 NO_STREAM_NAME, blob);
 839         if (!strm)
 840                 goto err_nomem;
 841
 842         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 843         return 0;
 844
 845 err_nomem:
 846         ret = WIMLIB_ERR_NOMEM;
 847 err:
 848         free_blob_descriptor(blob);
 849         return ret;
 850 }
 851
 852 static bool
 853 get_data_stream_name(wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 854                      wchar_t **stream_name_ret, size_t *stream_name_nchars_ret)
 855 {
 856         const wchar_t *sep, *type, *end;
 857
 858         /* The stream name should be returned as :NAME:TYPE  */
 859         if (raw_stream_name_nchars < 1)
 860                 return false;
 861         if (raw_stream_name[0] != L':')
 862                 return false;
 863
 864         raw_stream_name++;
 865         raw_stream_name_nchars--;
 866
 867         end = raw_stream_name + raw_stream_name_nchars;
 868
 869         sep = wmemchr(raw_stream_name, L':', raw_stream_name_nchars);
 870         if (!sep)
 871                 return false;
 872
 873         type = sep + 1;
 874         if (end - type != 5)
 875                 return false;
 876
 877         if (wmemcmp(type, L"$DATA", 5))
 878                 return false;
 879
 880         *stream_name_ret = raw_stream_name;
 881         *stream_name_nchars_ret = sep - raw_stream_name;
 882         return true;
 883 }
 884
 885 /* Build the path to the data stream.  For unnamed streams, this is simply the
 886  * path to the file.  For named streams, this is the path to the file, followed
 887  * by a colon, followed by the stream name.  */
 888 static wchar_t *
 889 build_data_stream_path(const wchar_t *path, size_t path_nchars,
 890                        const wchar_t *stream_name, size_t stream_name_nchars)
 891 {
 892         size_t stream_path_nchars;
 893         wchar_t *stream_path;
 894         wchar_t *p;
 895
 896         stream_path_nchars = path_nchars;
 897         if (stream_name_nchars)
 898                 stream_path_nchars += 1 + stream_name_nchars;
 899
 900         stream_path = MALLOC((stream_path_nchars + 1) * sizeof(wchar_t));
 901         if (stream_path) {
 902                 p = wmempcpy(stream_path, path, path_nchars);
 903                 if (stream_name_nchars) {
 904                         *p++ = L':';
 905                         p = wmempcpy(p, stream_name, stream_name_nchars);
 906                 }
 907                 *p++ = L'\0';
 908         }
 909         return stream_path;
 910 }
 911
 912 static int
 913 winnt_scan_data_stream(const wchar_t *path, size_t path_nchars,
 914                        wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 915                        u64 stream_size,
 916                        struct wim_inode *inode, struct list_head *unhashed_blobs)
 917 {
 918         wchar_t *stream_name;
 919         size_t stream_name_nchars;
 920         struct blob_descriptor *blob;
 921         struct wim_inode_stream *strm;
 922
 923         /* Given the raw stream name (which is something like
 924          * :streamname:$DATA), extract just the stream name part (streamname).
 925          * Ignore any non-$DATA streams.  */
 926         if (!get_data_stream_name(raw_stream_name, raw_stream_name_nchars,
 927                                   &stream_name, &stream_name_nchars))
 928                 return 0;
 929
 930         stream_name[stream_name_nchars] = L'\0';
 931
 932         /* If the stream is non-empty, set up a blob descriptor for it.  */
 933         if (stream_size != 0) {
 934                 blob = new_blob_descriptor();
 935                 if (!blob)
 936                         goto err_nomem;
 937                 blob->file_on_disk = build_data_stream_path(path,
 938                                                             path_nchars,
 939                                                             stream_name,
 940                                                             stream_name_nchars);
 941                 if (!blob->file_on_disk)
 942                         goto err_nomem;
 943                 blob->blob_location = BLOB_IN_WINNT_FILE_ON_DISK;
 944                 blob->size = stream_size;
 945                 blob->file_inode = inode;
 946         } else {
 947                 blob = NULL;
 948         }
 949
 950         strm = inode_add_stream(inode, STREAM_TYPE_DATA, stream_name, blob);
 951         if (!strm)
 952                 goto err_nomem;
 953
 954         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 955         return 0;
 956
 957 err_nomem:
 958         free_blob_descriptor(blob);
 959         return WIMLIB_ERR_NOMEM;
 960 }
 961
 962 /*
 963  * Load information about the data streams of an open file into a WIM inode.
 964  *
 965  * We use the NtQueryInformationFile() system call instead of FindFirstStream()
 966  * and FindNextStream().  This is done for two reasons:
 967  *
 968  * - FindFirstStream() opens its own handle to the file or directory and
 969  *   apparently does so without specifying FILE_FLAG_BACKUP_SEMANTICS, thereby
 970  *   causing access denied errors on certain files (even when running as the
 971  *   Administrator).
 972  * - FindFirstStream() and FindNextStream() is only available on Windows Vista
 973  *   and later, whereas the stream support in NtQueryInformationFile() was
 974  *   already present in Windows XP.
 975  */
 976 static noinline_for_stack int
 977 winnt_scan_data_streams(HANDLE h, const wchar_t *path, size_t path_nchars,
 978                         struct wim_inode *inode, struct list_head *unhashed_blobs,
 979                         u64 file_size, u32 vol_flags)
 980 {
 981         int ret;
 982         u8 _buf[4096] _aligned_attribute(8);
 983         u8 *buf;
 984         size_t bufsize;
 985         IO_STATUS_BLOCK iosb;
 986         NTSTATUS status;
 987         FILE_STREAM_INFORMATION *info;
 988
 989         buf = _buf;
 990         bufsize = sizeof(_buf);
 991
 992         if (!(vol_flags & FILE_NAMED_STREAMS))
 993                 goto unnamed_only;
 994
 995         /* Get a buffer containing the stream information.  */
 996         while (!NT_SUCCESS(status = (*func_NtQueryInformationFile)(h,
 997                                                                    &iosb,
 998                                                                    buf,
 999                                                                    bufsize,
1000                                                                    FileStreamInformation)))
1001         {
1002
1003                 switch (status) {
1004                 case STATUS_BUFFER_OVERFLOW:
1005                         {
1006                                 u8 *newbuf;
1007
1008                                 bufsize *= 2;
1009                                 if (buf == _buf)
1010                                         newbuf = MALLOC(bufsize);
1011                                 else
1012                                         newbuf = REALLOC(buf, bufsize);
1013                                 if (!newbuf) {
1014                                         ret = WIMLIB_ERR_NOMEM;
1015                                         goto out_free_buf;
1016                                 }
1017                                 buf = newbuf;
1018                         }
1019                         break;
1020                 case STATUS_NOT_IMPLEMENTED:
1021                 case STATUS_NOT_SUPPORTED:
1022                 case STATUS_INVALID_INFO_CLASS:
1023                         goto unnamed_only;
1024                 default:
1025                         winnt_error(status,
1026                                     L"\"%ls\": Failed to query stream information",
1027                                     printable_path(path));
1028                         ret = WIMLIB_ERR_READ;
1029                         goto out_free_buf;
1030                 }
1031         }
1032
1033         if (iosb.Information == 0) {
1034                 /* No stream information.  */
1035                 ret = 0;
1036                 goto out_free_buf;
1037         }
1038
1039         /* Parse one or more stream information structures.  */
1040         info = (FILE_STREAM_INFORMATION *)buf;
1041         for (;;) {
1042                 /* Load the stream information.  */
1043                 ret = winnt_scan_data_stream(path, path_nchars,
1044                                              info->StreamName,
1045                                              info->StreamNameLength / 2,
1046                                              info->StreamSize.QuadPart,
1047                                              inode, unhashed_blobs);
1048                 if (ret)
1049                         goto out_free_buf;
1050
1051                 if (info->NextEntryOffset == 0) {
1052                         /* No more stream information.  */
1053                         break;
1054                 }
1055                 /* Advance to next stream information.  */
1056                 info = (FILE_STREAM_INFORMATION *)
1057                                 ((u8 *)info + info->NextEntryOffset);
1058         }
1059         ret = 0;
1060         goto out_free_buf;
1061
1062 unnamed_only:
1063         /* The volume does not support named streams.  Only capture the unnamed
1064          * data stream.  */
1065         if (inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
1066                                    FILE_ATTRIBUTE_REPARSE_POINT))
1067         {
1068                 ret = 0;
1069                 goto out_free_buf;
1070         }
1071
1072         {
1073                 wchar_t stream_name[] = L"::$DATA";
1074                 ret = winnt_scan_data_stream(path, path_nchars, stream_name, 7,
1075                                              file_size, inode, unhashed_blobs);
1076         }
1077 out_free_buf:
1078         /* Free buffer if allocated on heap.  */
1079         if (unlikely(buf != _buf))
1080                 FREE(buf);
1081         return ret;
1082 }
1083
1084 static noinline_for_stack u64
1085 get_sort_key(HANDLE h)
1086 {
1087         STARTING_VCN_INPUT_BUFFER in = { .StartingVcn.QuadPart = 0 };
1088         RETRIEVAL_POINTERS_BUFFER out;
1089         DWORD bytesReturned;
1090
1091         if (!DeviceIoControl(h, FSCTL_GET_RETRIEVAL_POINTERS,
1092                              &in, sizeof(in),
1093                              &out, sizeof(out),
1094                              &bytesReturned, NULL))
1095                 return 0;
1096
1097         if (out.ExtentCount < 1)
1098                 return 0;
1099
1100         return out.Extents[0].Lcn.QuadPart;
1101 }
1102
1103 static void
1104 set_sort_key(struct wim_inode *inode, u64 sort_key)
1105 {
1106         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1107                 struct wim_inode_stream *strm = &inode->i_streams[i];
1108                 struct blob_descriptor *blob = stream_blob_resolved(strm);
1109                 if (blob && (blob->blob_location == BLOB_IN_WINNT_FILE_ON_DISK ||
1110                              blob->blob_location == BLOB_WIN32_ENCRYPTED))
1111                         blob->sort_key = sort_key;
1112         }
1113 }
1114
1115 static noinline_for_stack u32
1116 get_volume_information(HANDLE h, const wchar_t *full_path,
1117                        struct capture_params *params)
1118 {
1119         FILE_FS_ATTRIBUTE_INFORMATION attr_info;
1120         FILE_FS_VOLUME_INFORMATION vol_info;
1121         IO_STATUS_BLOCK iosb;
1122         NTSTATUS status;
1123         u32 vol_flags;
1124
1125         /* Get volume flags  */
1126         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1127                                                       &attr_info,
1128                                                       sizeof(attr_info),
1129                                                       FileFsAttributeInformation);
1130         if (likely((NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW) &&
1131                    (iosb.Information >=
1132                         offsetof(FILE_FS_ATTRIBUTE_INFORMATION,
1133                                  FileSystemAttributes) +
1134                         sizeof(attr_info.FileSystemAttributes))))
1135         {
1136                 vol_flags = attr_info.FileSystemAttributes;
1137         } else {
1138                 winnt_warning(status, L"\"%ls\": Can't get volume attributes",
1139                               printable_path(full_path));
1140                 vol_flags = 0;
1141         }
1142
1143         /* Get volume ID.  */
1144         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1145                                                       &vol_info,
1146                                                       sizeof(vol_info),
1147                                                       FileFsVolumeInformation);
1148         if (likely((NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW) &&
1149                    (iosb.Information >=
1150                         offsetof(FILE_FS_VOLUME_INFORMATION,
1151                                  VolumeSerialNumber) +
1152                         sizeof(vol_info.VolumeSerialNumber))))
1153         {
1154                 params->capture_root_dev = vol_info.VolumeSerialNumber;
1155         } else {
1156                 winnt_warning(status, L"\"%ls\": Can't get volume ID",
1157                               printable_path(full_path));
1158                 params->capture_root_dev = 0;
1159         }
1160         return vol_flags;
1161 }
1162
1163 struct file_info {
1164         u32 attributes;
1165         u32 num_links;
1166         u64 creation_time;
1167         u64 last_write_time;
1168         u64 last_access_time;
1169         u64 ino;
1170         u64 end_of_file;
1171 };
1172
1173 static noinline_for_stack NTSTATUS
1174 get_file_info(HANDLE h, struct file_info *info)
1175 {
1176         IO_STATUS_BLOCK iosb;
1177         NTSTATUS status;
1178         FILE_ALL_INFORMATION all_info;
1179
1180         status = (*func_NtQueryInformationFile)(h, &iosb, &all_info,
1181                                                 sizeof(all_info),
1182                                                 FileAllInformation);
1183
1184         if (unlikely(!NT_SUCCESS(status) && status != STATUS_BUFFER_OVERFLOW))
1185                 return status;
1186
1187         info->attributes = all_info.BasicInformation.FileAttributes;
1188         info->num_links = all_info.StandardInformation.NumberOfLinks;
1189         info->creation_time = all_info.BasicInformation.CreationTime.QuadPart;
1190         info->last_write_time = all_info.BasicInformation.LastWriteTime.QuadPart;
1191         info->last_access_time = all_info.BasicInformation.LastAccessTime.QuadPart;
1192         info->ino = all_info.InternalInformation.IndexNumber.QuadPart;
1193         info->end_of_file = all_info.StandardInformation.EndOfFile.QuadPart;
1194         return STATUS_SUCCESS;
1195 }
1196
1197 static int
1198 winnt_build_dentry_tree_recursive(struct wim_dentry **root_ret,
1199                                   HANDLE cur_dir,
1200                                   wchar_t *full_path,
1201                                   size_t full_path_nchars,
1202                                   const wchar_t *filename,
1203                                   size_t filename_nchars,
1204                                   struct capture_params *params,
1205                                   struct winnt_scan_stats *stats,
1206                                   u32 vol_flags)
1207 {
1208         struct wim_dentry *root = NULL;
1209         struct wim_inode *inode = NULL;
1210         HANDLE h = NULL;
1211         int ret;
1212         NTSTATUS status;
1213         struct file_info file_info;
1214         ACCESS_MASK requestedPerms;
1215         u64 sort_key;
1216
1217         ret = try_exclude(full_path, params);
1218         if (unlikely(ret < 0)) /* Excluded? */
1219                 goto out_progress;
1220         if (unlikely(ret > 0)) /* Error? */
1221                 goto out;
1222
1223         /* Open the file.  */
1224         requestedPerms = FILE_READ_DATA |
1225                          FILE_READ_ATTRIBUTES |
1226                          READ_CONTROL |
1227                          ACCESS_SYSTEM_SECURITY |
1228                          SYNCHRONIZE;
1229 retry_open:
1230         status = winnt_openat(cur_dir,
1231                               (cur_dir ? filename : full_path),
1232                               (cur_dir ? filename_nchars : full_path_nchars),
1233                               requestedPerms,
1234                               &h);
1235         if (unlikely(!NT_SUCCESS(status))) {
1236                 if (status == STATUS_DELETE_PENDING) {
1237                         WARNING("\"%ls\": Deletion pending; skipping file",
1238                                 printable_path(full_path));
1239                         ret = 0;
1240                         goto out;
1241                 }
1242                 if (status == STATUS_ACCESS_DENIED &&
1243                     (requestedPerms & FILE_READ_DATA)) {
1244                         /* This happens on encrypted files.  */
1245                         requestedPerms &= ~FILE_READ_DATA;
1246                         goto retry_open;
1247                 }
1248
1249                 winnt_error(status, L"\"%ls\": Can't open file",
1250                             printable_path(full_path));
1251                 if (status == STATUS_FVE_LOCKED_VOLUME)
1252                         ret = WIMLIB_ERR_FVE_LOCKED_VOLUME;
1253                 else
1254                         ret = WIMLIB_ERR_OPEN;
1255                 goto out;
1256         }
1257
1258         /* Get information about the file.  */
1259         status = get_file_info(h, &file_info);
1260         if (!NT_SUCCESS(status)) {
1261                 winnt_error(status, L"\"%ls\": Can't get file information",
1262                             printable_path(full_path));
1263                 ret = WIMLIB_ERR_STAT;
1264                 goto out;
1265         }
1266
1267         if (unlikely(!(requestedPerms & FILE_READ_DATA)) &&
1268             !(file_info.attributes & FILE_ATTRIBUTE_ENCRYPTED))
1269         {
1270                 ERROR("\"%ls\": Permission to read data was denied",
1271                       printable_path(full_path));
1272                 ret = WIMLIB_ERR_OPEN;
1273                 goto out;
1274         }
1275
1276         if (unlikely(!cur_dir)) {
1277                 /* Root of tree being captured; get volume information.  */
1278                 vol_flags = get_volume_information(h, full_path, params);
1279                 params->capture_root_ino = file_info.ino;
1280         }
1281
1282
1283         /* Create a WIM dentry with an associated inode, which may be shared.
1284          *
1285          * However, we need to explicitly check for directories and files with
1286          * only 1 link and refuse to hard link them.  This is because Windows
1287          * has a bug where it can return duplicate File IDs for files and
1288          * directories on the FAT filesystem.
1289          *
1290          * Since we don't follow mount points on Windows, we don't need to query
1291          * the volume ID per-file.  Just once, for the root, is enough.  But we
1292          * can't simply pass 0, because then there could be inode collisions
1293          * among multiple calls to win32_build_dentry_tree() that are scanning
1294          * files on different volumes.  */
1295         ret = inode_table_new_dentry(params->inode_table,
1296                                      filename,
1297                                      file_info.ino,
1298                                      params->capture_root_dev,
1299                                      (file_info.num_links <= 1),
1300                                      &root);
1301         if (ret)
1302                 goto out;
1303
1304         /* Get the short (DOS) name of the file.  */
1305         status = winnt_get_short_name(h, root);
1306
1307         /* If we can't read the short filename for any reason other than
1308          * out-of-memory, just ignore the error and assume the file has no short
1309          * name.  This shouldn't be an issue, since the short names are
1310          * essentially obsolete anyway.  */
1311         if (unlikely(status == STATUS_NO_MEMORY)) {
1312                 ret = WIMLIB_ERR_NOMEM;
1313                 goto out;
1314         }
1315
1316         inode = root->d_inode;
1317
1318         if (inode->i_nlink > 1) {
1319                 /* Shared inode (hard link); skip reading per-inode information.
1320                  */
1321                 goto out_progress;
1322         }
1323
1324         inode->i_attributes = file_info.attributes;
1325         inode->i_creation_time = file_info.creation_time;
1326         inode->i_last_write_time = file_info.last_write_time;
1327         inode->i_last_access_time = file_info.last_access_time;
1328
1329         /* Get the file's security descriptor, unless we are capturing in
1330          * NO_ACLS mode or the volume does not support security descriptors.  */
1331         if (!(params->add_flags & WIMLIB_ADD_FLAG_NO_ACLS)
1332             && (vol_flags & FILE_PERSISTENT_ACLS))
1333         {
1334                 status = winnt_get_security_descriptor(h, inode,
1335                                                        params->sd_set, stats,
1336                                                        params->add_flags);
1337                 if (!NT_SUCCESS(status)) {
1338                         winnt_error(status,
1339                                     L"\"%ls\": Can't read security descriptor",
1340                                     printable_path(full_path));
1341                         ret = WIMLIB_ERR_STAT;
1342                         goto out;
1343                 }
1344         }
1345
1346         /* If this is a reparse point, load the reparse data.  */
1347         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)) {
1348                 ret = winnt_load_reparse_data(h, inode, full_path, params);
1349                 if (ret)
1350                         goto out;
1351         }
1352
1353         sort_key = get_sort_key(h);
1354
1355         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED)) {
1356                 /* Load information about the raw encrypted data.  This is
1357                  * needed for any directory or non-directory that has
1358                  * FILE_ATTRIBUTE_ENCRYPTED set.
1359                  *
1360                  * Note: since OpenEncryptedFileRaw() fails with
1361                  * ERROR_SHARING_VIOLATION if there are any open handles to the
1362                  * file, we have to close the file and re-open it later if
1363                  * needed.  */
1364                 (*func_NtClose)(h);
1365                 h = NULL;
1366                 ret = winnt_scan_efsrpc_raw_data(inode, full_path,
1367                                                  params->unhashed_blobs);
1368                 if (ret)
1369                         goto out;
1370         } else {
1371                 /*
1372                  * Load information about data streams (unnamed and named).
1373                  *
1374                  * Skip this step for encrypted files, since the data from
1375                  * ReadEncryptedFileRaw() already contains all data streams (and
1376                  * they do in fact all get restored by WriteEncryptedFileRaw().)
1377                  *
1378                  * Note: WIMGAPI (as of Windows 8.1) gets wrong and stores both
1379                  * the EFSRPC data and the named data stream(s)...!
1380                  */
1381                 ret = winnt_scan_data_streams(h,
1382                                               full_path,
1383                                               full_path_nchars,
1384                                               inode,
1385                                               params->unhashed_blobs,
1386                                               file_info.end_of_file,
1387                                               vol_flags);
1388                 if (ret)
1389                         goto out;
1390         }
1391
1392         set_sort_key(inode, sort_key);
1393
1394         if (inode_is_directory(inode)) {
1395
1396                 /* Directory: recurse to children.  */
1397
1398                 if (unlikely(!h)) {
1399                         /* Re-open handle that was closed to read raw encrypted
1400                          * data.  */
1401                         status = winnt_openat(cur_dir,
1402                                               (cur_dir ?
1403                                                filename : full_path),
1404                                               (cur_dir ?
1405                                                filename_nchars : full_path_nchars),
1406                                               FILE_LIST_DIRECTORY | SYNCHRONIZE,
1407                                               &h);
1408                         if (!NT_SUCCESS(status)) {
1409                                 winnt_error(status,
1410                                             L"\"%ls\": Can't re-open file",
1411                                             printable_path(full_path));
1412                                 ret = WIMLIB_ERR_OPEN;
1413                                 goto out;
1414                         }
1415                 }
1416                 ret = winnt_recurse_directory(h,
1417                                               full_path,
1418                                               full_path_nchars,
1419                                               root,
1420                                               params,
1421                                               stats,
1422                                               vol_flags);
1423                 if (ret)
1424                         goto out;
1425         }
1426
1427 out_progress:
1428         params->progress.scan.cur_path = printable_path(full_path);
1429         if (likely(root))
1430                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
1431         else
1432                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
1433 out:
1434         if (likely(h))
1435                 (*func_NtClose)(h);
1436         if (unlikely(ret)) {
1437                 free_dentry_tree(root, params->blob_table);
1438                 root = NULL;
1439                 ret = report_capture_error(params, ret, full_path);
1440         }
1441         *root_ret = root;
1442         return ret;
1443 }
1444
1445 static void
1446 winnt_do_scan_warnings(const wchar_t *path, const struct winnt_scan_stats *stats)
1447 {
1448         if (likely(stats->num_get_sacl_priv_notheld == 0 &&
1449                    stats->num_get_sd_access_denied == 0))
1450                 return;
1451
1452         WARNING("Scan of \"%ls\" complete, but with one or more warnings:", path);
1453         if (stats->num_get_sacl_priv_notheld != 0) {
1454                 WARNING("- Could not capture SACL (System Access Control List)\n"
1455                         "            on %lu files or directories.",
1456                         stats->num_get_sacl_priv_notheld);
1457         }
1458         if (stats->num_get_sd_access_denied != 0) {
1459                 WARNING("- Could not capture security descriptor at all\n"
1460                         "            on %lu files or directories.",
1461                         stats->num_get_sd_access_denied);
1462         }
1463         WARNING("To fully capture all security descriptors, run the program\n"
1464                 "          with Administrator rights.");
1465 }
1466
1467 #define WINDOWS_NT_MAX_PATH 32768
1468
1469 /* Win32 version of capturing a directory tree.  */
1470 int
1471 win32_build_dentry_tree(struct wim_dentry **root_ret,
1472                         const wchar_t *root_disk_path,
1473                         struct capture_params *params)
1474 {
1475         wchar_t *path;
1476         int ret;
1477         UNICODE_STRING ntpath;
1478         struct winnt_scan_stats stats;
1479         size_t ntpath_nchars;
1480
1481         /* WARNING: There is no check for overflow later when this buffer is
1482          * being used!  But it's as long as the maximum path length understood
1483          * by Windows NT (which is NOT the same as MAX_PATH).  */
1484         path = MALLOC((WINDOWS_NT_MAX_PATH + 1) * sizeof(wchar_t));
1485         if (!path)
1486                 return WIMLIB_ERR_NOMEM;
1487
1488         ret = win32_path_to_nt_path(root_disk_path, &ntpath);
1489         if (ret)
1490                 goto out_free_path;
1491
1492         if (ntpath.Length < 4 * sizeof(wchar_t) ||
1493             ntpath.Length > WINDOWS_NT_MAX_PATH * sizeof(wchar_t) ||
1494             wmemcmp(ntpath.Buffer, L"\\??\\", 4))
1495         {
1496                 ERROR("\"%ls\": unrecognized path format", root_disk_path);
1497                 ret = WIMLIB_ERR_INVALID_PARAM;
1498         } else {
1499                 ntpath_nchars = ntpath.Length / sizeof(wchar_t);
1500                 wmemcpy(path, ntpath.Buffer, ntpath_nchars);
1501                 path[ntpath_nchars] = L'\0';
1502
1503                 params->capture_root_nchars = ntpath_nchars;
1504                 if (path[ntpath_nchars - 1] == L'\\')
1505                         params->capture_root_nchars--;
1506                 ret = 0;
1507         }
1508         HeapFree(GetProcessHeap(), 0, ntpath.Buffer);
1509         if (ret)
1510                 goto out_free_path;
1511
1512         memset(&stats, 0, sizeof(stats));
1513
1514         ret = winnt_build_dentry_tree_recursive(root_ret, NULL,
1515                                                 path, ntpath_nchars,
1516                                                 L"", 0, params, &stats, 0);
1517 out_free_path:
1518         FREE(path);
1519         if (ret == 0)
1520                 winnt_do_scan_warnings(root_disk_path, &stats);
1521         return ret;
1522 }
1523
1524 #endif /* __WIN32__ */