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 (!NT_SUCCESS(status)) {
 141                         winnt_error(status, L"\"%ls\": Error reading data",
 142                                     printable_path(path));
 143                         ret = WIMLIB_ERR_READ;
 144                         break;
 145                 }
 146
 147                 bytes_read = iosb.Information;
 148
 149                 bytes_remaining -= bytes_read;
 150                 ret = call_consume_chunk(buf, bytes_read, cbs);
 151                 if (ret)
 152                         break;
 153         }
 154         (*func_NtClose)(h);
 155         return ret;
 156 }
 157
 158 struct win32_encrypted_read_ctx {
 159         const struct read_blob_callbacks *cbs;
 160         int wimlib_err_code;
 161         u64 bytes_remaining;
 162 };
 163
 164 static DWORD WINAPI
 165 win32_encrypted_export_cb(unsigned char *data, void *_ctx, unsigned long len)
 166 {
 167         struct win32_encrypted_read_ctx *ctx = _ctx;
 168         int ret;
 169         size_t bytes_to_consume = min(len, ctx->bytes_remaining);
 170
 171         if (bytes_to_consume == 0)
 172                 return ERROR_SUCCESS;
 173
 174         ret = call_consume_chunk(data, bytes_to_consume, ctx->cbs);
 175         if (ret) {
 176                 ctx->wimlib_err_code = ret;
 177                 /* It doesn't matter what error code is returned here, as long
 178                  * as it isn't ERROR_SUCCESS.  */
 179                 return ERROR_READ_FAULT;
 180         }
 181         ctx->bytes_remaining -= bytes_to_consume;
 182         return ERROR_SUCCESS;
 183 }
 184
 185 int
 186 read_win32_encrypted_file_prefix(const struct blob_descriptor *blob,
 187                                  u64 size,
 188                                  const struct read_blob_callbacks *cbs)
 189 {
 190         struct win32_encrypted_read_ctx export_ctx;
 191         DWORD err;
 192         void *file_ctx;
 193         int ret;
 194         DWORD flags = 0;
 195
 196         if (blob->file_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
 197                 flags |= CREATE_FOR_DIR;
 198
 199         export_ctx.cbs = cbs;
 200         export_ctx.wimlib_err_code = 0;
 201         export_ctx.bytes_remaining = size;
 202
 203         err = OpenEncryptedFileRaw(blob->file_on_disk, flags, &file_ctx);
 204         if (err != ERROR_SUCCESS) {
 205                 win32_error(err,
 206                             L"Failed to open encrypted file \"%ls\" for raw read",
 207                             printable_path(blob->file_on_disk));
 208                 return WIMLIB_ERR_OPEN;
 209         }
 210         err = ReadEncryptedFileRaw(win32_encrypted_export_cb,
 211                                    &export_ctx, file_ctx);
 212         if (err != ERROR_SUCCESS) {
 213                 ret = export_ctx.wimlib_err_code;
 214                 if (ret == 0) {
 215                         win32_error(err,
 216                                     L"Failed to read encrypted file \"%ls\"",
 217                                     printable_path(blob->file_on_disk));
 218                         ret = WIMLIB_ERR_READ;
 219                 }
 220         } else if (export_ctx.bytes_remaining != 0) {
 221                 ERROR("Only could read %"PRIu64" of %"PRIu64" bytes from "
 222                       "encrypted file \"%ls\"",
 223                       size - export_ctx.bytes_remaining, size,
 224                       printable_path(blob->file_on_disk));
 225                 ret = WIMLIB_ERR_READ;
 226         } else {
 227                 ret = 0;
 228         }
 229         CloseEncryptedFileRaw(file_ctx);
 230         return ret;
 231 }
 232
 233 /*
 234  * Load the short name of a file into a WIM dentry.
 235  */
 236 static NTSTATUS
 237 winnt_get_short_name(HANDLE h, struct wim_dentry *dentry)
 238 {
 239         /* It's not any harder to just make the NtQueryInformationFile() system
 240          * call ourselves, and it saves a dumb call to FindFirstFile() which of
 241          * course has to create its own handle.  */
 242         NTSTATUS status;
 243         IO_STATUS_BLOCK iosb;
 244         u8 buf[128] _aligned_attribute(8);
 245         const FILE_NAME_INFORMATION *info;
 246
 247         status = (*func_NtQueryInformationFile)(h, &iosb, buf, sizeof(buf),
 248                                                 FileAlternateNameInformation);
 249         info = (const FILE_NAME_INFORMATION *)buf;
 250         if (NT_SUCCESS(status) && info->FileNameLength != 0) {
 251                 dentry->short_name = utf16le_dupz(info->FileName,
 252                                                   info->FileNameLength);
 253                 if (!dentry->short_name)
 254                         return STATUS_NO_MEMORY;
 255                 dentry->short_name_nbytes = info->FileNameLength;
 256         }
 257         return status;
 258 }
 259
 260 /*
 261  * Load the security descriptor of a file into the corresponding inode and the
 262  * WIM image's security descriptor set.
 263  */
 264 static NTSTATUS
 265 winnt_get_security_descriptor(HANDLE h, struct wim_inode *inode,
 266                               struct wim_sd_set *sd_set,
 267                               struct winnt_scan_stats *stats, int add_flags)
 268 {
 269         SECURITY_INFORMATION requestedInformation;
 270         u8 _buf[4096] _aligned_attribute(8);
 271         u8 *buf;
 272         ULONG bufsize;
 273         ULONG len_needed;
 274         NTSTATUS status;
 275
 276         /*
 277          * LABEL_SECURITY_INFORMATION is needed on Windows Vista and 7 because
 278          * Microsoft decided to add mandatory integrity labels to the SACL but
 279          * not have them returned by SACL_SECURITY_INFORMATION.
 280          *
 281          * BACKUP_SECURITY_INFORMATION is needed on Windows 8 because Microsoft
 282          * decided to add even more stuff to the SACL and still not have it
 283          * returned by SACL_SECURITY_INFORMATION; but they did remember that
 284          * backup applications exist and simply want to read the stupid thing
 285          * once and for all, so they added a flag to read the entire security
 286          * descriptor.
 287          *
 288          * Older versions of Windows tolerate these new flags being passed in.
 289          */
 290         requestedInformation = OWNER_SECURITY_INFORMATION |
 291                                GROUP_SECURITY_INFORMATION |
 292                                DACL_SECURITY_INFORMATION |
 293                                SACL_SECURITY_INFORMATION |
 294                                LABEL_SECURITY_INFORMATION |
 295                                BACKUP_SECURITY_INFORMATION;
 296
 297         buf = _buf;
 298         bufsize = sizeof(_buf);
 299
 300         /*
 301          * We need the file's security descriptor in
 302          * SECURITY_DESCRIPTOR_RELATIVE format, and we currently have a handle
 303          * opened with as many relevant permissions as possible.  At this point,
 304          * on Windows there are a number of options for reading a file's
 305          * security descriptor:
 306          *
 307          * GetFileSecurity():  This takes in a path and returns the
 308          * SECURITY_DESCRIPTOR_RELATIVE.  Problem: this uses an internal handle,
 309          * not ours, and the handle created internally doesn't specify
 310          * FILE_FLAG_BACKUP_SEMANTICS.  Therefore there can be access denied
 311          * errors on some files and directories, even when running as the
 312          * Administrator.
 313          *
 314          * GetSecurityInfo():  This takes in a handle and returns the security
 315          * descriptor split into a bunch of different parts.  This should work,
 316          * but it's dumb because we have to put the security descriptor back
 317          * together again.
 318          *
 319          * BackupRead():  This can read the security descriptor, but this is a
 320          * difficult-to-use API, probably only works as the Administrator, and
 321          * the format of the returned data is not well documented.
 322          *
 323          * NtQuerySecurityObject():  This is exactly what we need, as it takes
 324          * in a handle and returns the security descriptor in
 325          * SECURITY_DESCRIPTOR_RELATIVE format.  Only problem is that it's a
 326          * ntdll function and therefore not officially part of the Win32 API.
 327          * Oh well.
 328          */
 329         while (!(NT_SUCCESS(status = (*func_NtQuerySecurityObject)(h,
 330                                                                    requestedInformation,
 331                                                                    (PSECURITY_DESCRIPTOR)buf,
 332                                                                    bufsize,
 333                                                                    &len_needed))))
 334         {
 335                 switch (status) {
 336                 case STATUS_BUFFER_TOO_SMALL:
 337                         wimlib_assert(buf == _buf);
 338                         buf = MALLOC(len_needed);
 339                         if (!buf)
 340                                 return STATUS_NO_MEMORY;
 341                         bufsize = len_needed;
 342                         break;
 343                 case STATUS_PRIVILEGE_NOT_HELD:
 344                 case STATUS_ACCESS_DENIED:
 345                         if (add_flags & WIMLIB_ADD_FLAG_STRICT_ACLS) {
 346                 default:
 347                                 /* Permission denied in STRICT_ACLS mode, or
 348                                  * unknown error.  */
 349                                 goto out_free_buf;
 350                         }
 351                         if (requestedInformation & SACL_SECURITY_INFORMATION) {
 352                                 /* Try again without the SACL.  */
 353                                 stats->num_get_sacl_priv_notheld++;
 354                                 requestedInformation &= ~(SACL_SECURITY_INFORMATION |
 355                                                           LABEL_SECURITY_INFORMATION |
 356                                                           BACKUP_SECURITY_INFORMATION);
 357                                 break;
 358                         }
 359                         /* Fake success (useful when capturing as
 360                          * non-Administrator).  */
 361                         stats->num_get_sd_access_denied++;
 362                         status = STATUS_SUCCESS;
 363                         goto out_free_buf;
 364                 }
 365         }
 366
 367         /* Add the security descriptor to the WIM image, and save its ID in
 368          * file's inode.  */
 369         inode->i_security_id = sd_set_add_sd(sd_set, buf, len_needed);
 370         if (unlikely(inode->i_security_id < 0))
 371                 status = STATUS_NO_MEMORY;
 372 out_free_buf:
 373         if (unlikely(buf != _buf))
 374                 FREE(buf);
 375         return status;
 376 }
 377
 378 static int
 379 winnt_build_dentry_tree_recursive(struct wim_dentry **root_ret,
 380                                   HANDLE cur_dir,
 381                                   wchar_t *full_path,
 382                                   size_t full_path_nchars,
 383                                   const wchar_t *filename,
 384                                   size_t filename_nchars,
 385                                   struct capture_params *params,
 386                                   struct winnt_scan_stats *stats,
 387                                   u32 vol_flags);
 388
 389 static int
 390 winnt_recurse_directory(HANDLE h,
 391                         wchar_t *full_path,
 392                         size_t full_path_nchars,
 393                         struct wim_dentry *parent,
 394                         struct capture_params *params,
 395                         struct winnt_scan_stats *stats,
 396                         u32 vol_flags)
 397 {
 398         void *buf;
 399         const size_t bufsize = 8192;
 400         IO_STATUS_BLOCK iosb;
 401         NTSTATUS status;
 402         int ret;
 403
 404         buf = MALLOC(bufsize);
 405         if (!buf)
 406                 return WIMLIB_ERR_NOMEM;
 407
 408         /* Using NtQueryDirectoryFile() we can re-use the same open handle,
 409          * which we opened with FILE_FLAG_BACKUP_SEMANTICS.  */
 410
 411         while (NT_SUCCESS(status = (*func_NtQueryDirectoryFile)(h, NULL, NULL, NULL,
 412                                                                 &iosb, buf, bufsize,
 413                                                                 FileNamesInformation,
 414                                                                 FALSE, NULL, FALSE)))
 415         {
 416                 const FILE_NAMES_INFORMATION *info = buf;
 417                 for (;;) {
 418                         if (!(info->FileNameLength == 2 && info->FileName[0] == L'.') &&
 419                             !(info->FileNameLength == 4 && info->FileName[0] == L'.' &&
 420                                                            info->FileName[1] == L'.'))
 421                         {
 422                                 wchar_t *p;
 423                                 wchar_t *filename;
 424                                 struct wim_dentry *child;
 425
 426                                 p = full_path + full_path_nchars;
 427                                 /* Only add a backslash if we don't already have
 428                                  * one.  This prevents a duplicate backslash
 429                                  * from being added when the path to the capture
 430                                  * dir had a trailing backslash.  */
 431                                 if (*(p - 1) != L'\\')
 432                                         *p++ = L'\\';
 433                                 filename = p;
 434                                 p = wmempcpy(filename, info->FileName,
 435                                              info->FileNameLength / 2);
 436                                 *p = '\0';
 437
 438                                 ret = winnt_build_dentry_tree_recursive(
 439                                                         &child,
 440                                                         h,
 441                                                         full_path,
 442                                                         p - full_path,
 443                                                         filename,
 444                                                         info->FileNameLength / 2,
 445                                                         params,
 446                                                         stats,
 447                                                         vol_flags);
 448
 449                                 full_path[full_path_nchars] = L'\0';
 450
 451                                 if (ret)
 452                                         goto out_free_buf;
 453                                 if (child)
 454                                         dentry_add_child(parent, child);
 455                         }
 456                         if (info->NextEntryOffset == 0)
 457                                 break;
 458                         info = (const FILE_NAMES_INFORMATION *)
 459                                         ((const u8 *)info + info->NextEntryOffset);
 460                 }
 461         }
 462
 463         if (unlikely(status != STATUS_NO_MORE_FILES)) {
 464                 winnt_error(status, L"\"%ls\": Can't read directory",
 465                             printable_path(full_path));
 466                 ret = WIMLIB_ERR_READ;
 467         }
 468 out_free_buf:
 469         FREE(buf);
 470         return ret;
 471 }
 472
 473 /* Reparse point fixup status code  */
 474 enum rp_status {
 475         /* Reparse point will be captured literally (no fixup)  */
 476         RP_NOT_FIXED    = -1,
 477
 478         /* Reparse point will be captured with fixup  */
 479         RP_FIXED        = -2,
 480 };
 481
 482 static bool
 483 file_has_ino_and_dev(HANDLE h, u64 ino, u64 dev)
 484 {
 485         NTSTATUS status;
 486         IO_STATUS_BLOCK iosb;
 487         FILE_INTERNAL_INFORMATION int_info;
 488         FILE_FS_VOLUME_INFORMATION vol_info;
 489
 490         status = (*func_NtQueryInformationFile)(h, &iosb,
 491                                                 &int_info, sizeof(int_info),
 492                                                 FileInternalInformation);
 493         if (!NT_SUCCESS(status))
 494                 return false;
 495
 496         if (int_info.IndexNumber.QuadPart != ino)
 497                 return false;
 498
 499         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
 500                                                       &vol_info, sizeof(vol_info),
 501                                                       FileFsVolumeInformation);
 502         if (!(NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW))
 503                 return false;
 504
 505         if (iosb.Information <
 506              offsetof(FILE_FS_VOLUME_INFORMATION, VolumeSerialNumber) +
 507              sizeof(vol_info.VolumeSerialNumber))
 508                 return false;
 509
 510         return (vol_info.VolumeSerialNumber == dev);
 511 }
 512
 513 /*
 514  * Given an (expected) NT namespace symbolic link or junction target @target of
 515  * length @target_nbytes, determine if a prefix of the target points to a file
 516  * identified by @capture_root_ino and @capture_root_dev.
 517  *
 518  * If yes, return a pointer to the portion of the link following this prefix.
 519  *
 520  * If no, return NULL.
 521  *
 522  * If the link target does not appear to be a valid NT namespace path, return
 523  * @target itself.
 524  */
 525 static const wchar_t *
 526 winnt_get_root_relative_target(const wchar_t *target, size_t target_nbytes,
 527                                u64 capture_root_ino, u64 capture_root_dev)
 528 {
 529         UNICODE_STRING name;
 530         OBJECT_ATTRIBUTES attr;
 531         IO_STATUS_BLOCK iosb;
 532         NTSTATUS status;
 533         const wchar_t *target_end;
 534         const wchar_t *p;
 535
 536         target_end = target + (target_nbytes / sizeof(wchar_t));
 537
 538         /* Empty path??? */
 539         if (target_end == target)
 540                 return target;
 541
 542         /* No leading slash???  */
 543         if (target[0] != L'\\')
 544                 return target;
 545
 546         /* UNC path???  */
 547         if ((target_end - target) >= 2 &&
 548             target[0] == L'\\' && target[1] == L'\\')
 549                 return target;
 550
 551         attr.Length = sizeof(attr);
 552         attr.RootDirectory = NULL;
 553         attr.ObjectName = &name;
 554         attr.Attributes = 0;
 555         attr.SecurityDescriptor = NULL;
 556         attr.SecurityQualityOfService = NULL;
 557
 558         name.Buffer = (wchar_t *)target;
 559         name.Length = 0;
 560         p = target;
 561         do {
 562                 HANDLE h;
 563                 const wchar_t *orig_p = p;
 564
 565                 /* Skip non-backslashes  */
 566                 while (p != target_end && *p != L'\\')
 567                         p++;
 568
 569                 /* Skip backslashes  */
 570                 while (p != target_end && *p == L'\\')
 571                         p++;
 572
 573                 /* Append path component  */
 574                 name.Length += (p - orig_p) * sizeof(wchar_t);
 575                 name.MaximumLength = name.Length;
 576
 577                 /* Try opening the file  */
 578                 status = (*func_NtOpenFile) (&h,
 579                                              FILE_READ_ATTRIBUTES | FILE_TRAVERSE,
 580                                              &attr,
 581                                              &iosb,
 582                                              FILE_SHARE_VALID_FLAGS,
 583                                              FILE_OPEN_FOR_BACKUP_INTENT);
 584
 585                 if (NT_SUCCESS(status)) {
 586                         /* Reset root directory  */
 587                         if (attr.RootDirectory)
 588                                 (*func_NtClose)(attr.RootDirectory);
 589                         attr.RootDirectory = h;
 590                         name.Buffer = (wchar_t *)p;
 591                         name.Length = 0;
 592
 593                         if (file_has_ino_and_dev(h, capture_root_ino,
 594                                                  capture_root_dev))
 595                                 goto out_close_root_dir;
 596                 }
 597         } while (p != target_end);
 598
 599         p = NULL;
 600
 601 out_close_root_dir:
 602         if (attr.RootDirectory)
 603                 (*func_NtClose)(attr.RootDirectory);
 604         return p;
 605 }
 606
 607 static int
 608 winnt_rpfix_progress(struct capture_params *params, const wchar_t *path,
 609                      const struct reparse_data *rpdata, int scan_status)
 610 {
 611         size_t print_name_nchars = rpdata->print_name_nbytes / sizeof(wchar_t);
 612         wchar_t print_name0[print_name_nchars + 1];
 613
 614         wmemcpy(print_name0, rpdata->print_name, print_name_nchars);
 615         print_name0[print_name_nchars] = L'\0';
 616
 617         params->progress.scan.cur_path = printable_path(path);
 618         params->progress.scan.symlink_target = print_name0;
 619         return do_capture_progress(params, scan_status, NULL);
 620 }
 621
 622 static int
 623 winnt_try_rpfix(u8 *rpbuf, u16 *rpbuflen_p,
 624                 u64 capture_root_ino, u64 capture_root_dev,
 625                 const wchar_t *path, struct capture_params *params)
 626 {
 627         struct reparse_data rpdata;
 628         const wchar_t *rel_target;
 629         int ret;
 630
 631         if (parse_reparse_data(rpbuf, *rpbuflen_p, &rpdata)) {
 632                 /* Couldn't even understand the reparse data.  Don't try the
 633                  * fixup.  */
 634                 return RP_NOT_FIXED;
 635         }
 636
 637         /*
 638          * Don't do reparse point fixups on relative symbolic links.
 639          *
 640          * On Windows, a relative symbolic link is supposed to be identifiable
 641          * by having reparse tag WIM_IO_REPARSE_TAG_SYMLINK and flags
 642          * SYMBOLIC_LINK_RELATIVE.  We will use this information, although this
 643          * may not always do what the user expects, since drive-relative
 644          * symbolic links such as "\Users\Public" have SYMBOLIC_LINK_RELATIVE
 645          * set, in addition to truely relative symbolic links such as "Users" or
 646          * "Users\Public".  However, WIMGAPI (as of Windows 8.1) has this same
 647          * behavior.
 648          *
 649          * Otherwise, as far as I can tell, the targets of symbolic links that
 650          * are NOT relative, as well as junctions (note: a mountpoint is the
 651          * sames thing as a junction), must be NT namespace paths, for example:
 652          *
 653          *     - \??\e:\Users\Public
 654          *     - \DosDevices\e:\Users\Public
 655          *     - \Device\HardDiskVolume4\Users\Public
 656          *     - \??\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 657          *     - \DosDevices\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 658          */
 659         if (rpdata.rptag == WIM_IO_REPARSE_TAG_SYMLINK &&
 660             (rpdata.rpflags & SYMBOLIC_LINK_RELATIVE))
 661                 return RP_NOT_FIXED;
 662
 663         rel_target = winnt_get_root_relative_target(rpdata.substitute_name,
 664                                                     rpdata.substitute_name_nbytes,
 665                                                     capture_root_ino,
 666                                                     capture_root_dev);
 667         if (!rel_target) {
 668                 /* Target points outside of the tree being captured.  Don't
 669                  * adjust it.  */
 670                 ret = winnt_rpfix_progress(params, path, &rpdata,
 671                                            WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK);
 672                 if (ret)
 673                         return ret;
 674                 return RP_NOT_FIXED;
 675         }
 676
 677         if (rel_target == rpdata.substitute_name) {
 678                 /* Weird target --- keep the reparse point and don't mess with
 679                  * it.  */
 680                 return RP_NOT_FIXED;
 681         }
 682
 683         /* We have an absolute target pointing within the directory being
 684          * captured. @rel_target is the suffix of the link target that is the
 685          * part relative to the directory being captured.
 686          *
 687          * We will cut off the prefix before this part (which is the path to the
 688          * directory being captured) and add a dummy prefix.  Since the process
 689          * will need to be reversed when applying the image, it shouldn't matter
 690          * what exactly the prefix is, as long as it looks like an absolute
 691          * path.
 692          */
 693
 694         {
 695                 size_t rel_target_nbytes =
 696                         rpdata.substitute_name_nbytes - ((const u8 *)rel_target -
 697                                                          (const u8 *)rpdata.substitute_name);
 698                 size_t rel_target_nchars = rel_target_nbytes / sizeof(wchar_t);
 699
 700                 wchar_t tmp[rel_target_nchars + 7];
 701
 702                 wmemcpy(tmp, L"\\??\\X:\\", 7);
 703                 wmemcpy(tmp + 7, rel_target, rel_target_nchars);
 704
 705                 rpdata.substitute_name = tmp;
 706                 rpdata.substitute_name_nbytes = rel_target_nbytes + (7 * sizeof(wchar_t));
 707                 rpdata.print_name = tmp + 4;
 708                 rpdata.print_name_nbytes = rel_target_nbytes + (3 * sizeof(wchar_t));
 709
 710                 if (make_reparse_buffer(&rpdata, rpbuf, rpbuflen_p))
 711                         return RP_NOT_FIXED;
 712         }
 713         ret = winnt_rpfix_progress(params, path, &rpdata,
 714                                    WIMLIB_SCAN_DENTRY_FIXED_SYMLINK);
 715         if (ret)
 716                 return ret;
 717         return RP_FIXED;
 718 }
 719
 720 /*
 721  * Loads the reparse point data from a reparse point into memory, optionally
 722  * fixing the targets of absolute symbolic links and junction points to be
 723  * relative to the root of capture.
 724  *
 725  * @h:
 726  *      Open handle to the reparse point file.
 727  * @path:
 728  *      Path to the reparse point file.
 729  * @params:
 730  *      Capture parameters.  add_flags, capture_root_ino, capture_root_dev,
 731  *      progfunc, progctx, and progress are used.
 732  * @rpbuf:
 733  *      Buffer of length at least REPARSE_POINT_MAX_SIZE bytes into which the
 734  *      reparse point buffer will be loaded.
 735  * @rpbuflen_ret:
 736  *      On success, the length of the reparse point buffer in bytes is written
 737  *      to this location.
 738  *
 739  * On success, returns a negative `enum rp_status' value.
 740  * On failure, returns a positive error code.
 741  */
 742 static int
 743 winnt_get_reparse_data(HANDLE h, const wchar_t *path,
 744                        struct capture_params *params,
 745                        u8 *rpbuf, u16 *rpbuflen_ret)
 746 {
 747         DWORD bytes_returned;
 748         u32 reparse_tag;
 749         int ret;
 750         u16 rpbuflen;
 751
 752         if (!DeviceIoControl(h, FSCTL_GET_REPARSE_POINT,
 753                              NULL, 0, rpbuf, REPARSE_POINT_MAX_SIZE,
 754                              &bytes_returned, NULL))
 755         {
 756                 win32_error(GetLastError(), L"\"%ls\": Can't get reparse data",
 757                             printable_path(path));
 758                 return WIMLIB_ERR_READ;
 759         }
 760
 761         if (unlikely(bytes_returned < REPARSE_DATA_OFFSET)) {
 762                 ERROR("\"%ls\": Reparse point data is invalid",
 763                       printable_path(path));
 764                 return WIMLIB_ERR_INVALID_REPARSE_DATA;
 765         }
 766
 767         rpbuflen = bytes_returned;
 768         reparse_tag = le32_to_cpu(*(le32*)rpbuf);
 769         ret = RP_NOT_FIXED;
 770         if (params->add_flags & WIMLIB_ADD_FLAG_RPFIX &&
 771             (reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK ||
 772              reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT))
 773         {
 774                 ret = winnt_try_rpfix(rpbuf, &rpbuflen,
 775                                       params->capture_root_ino,
 776                                       params->capture_root_dev,
 777                                       path, params);
 778         }
 779         *rpbuflen_ret = rpbuflen;
 780         return ret;
 781 }
 782
 783 static DWORD WINAPI
 784 win32_tally_encrypted_size_cb(unsigned char *_data, void *_size_ret,
 785                               unsigned long len)
 786 {
 787         *(u64*)_size_ret += len;
 788         return ERROR_SUCCESS;
 789 }
 790
 791 static int
 792 win32_get_encrypted_file_size(const wchar_t *path, bool is_dir, u64 *size_ret)
 793 {
 794         DWORD err;
 795         void *file_ctx;
 796         int ret;
 797         DWORD flags = 0;
 798
 799         if (is_dir)
 800                 flags |= CREATE_FOR_DIR;
 801
 802         err = OpenEncryptedFileRaw(path, flags, &file_ctx);
 803         if (err != ERROR_SUCCESS) {
 804                 win32_error(err,
 805                             L"Failed to open encrypted file \"%ls\" for raw read",
 806                             printable_path(path));
 807                 return WIMLIB_ERR_OPEN;
 808         }
 809         *size_ret = 0;
 810         err = ReadEncryptedFileRaw(win32_tally_encrypted_size_cb,
 811                                    size_ret, file_ctx);
 812         if (err != ERROR_SUCCESS) {
 813                 win32_error(err,
 814                             L"Failed to read raw encrypted data from \"%ls\"",
 815                             printable_path(path));
 816                 ret = WIMLIB_ERR_READ;
 817         } else {
 818                 ret = 0;
 819         }
 820         CloseEncryptedFileRaw(file_ctx);
 821         return ret;
 822 }
 823
 824 static int
 825 winnt_scan_efsrpc_raw_data(struct wim_inode *inode, const wchar_t *nt_path,
 826                            struct list_head *unhashed_blobs)
 827 {
 828         struct blob_descriptor *blob;
 829         struct wim_inode_stream *strm;
 830         int ret;
 831
 832         blob = new_blob_descriptor();
 833         if (!blob)
 834                 goto err_nomem;
 835
 836         blob->file_on_disk = WCSDUP(nt_path);
 837         if (!blob->file_on_disk)
 838                 goto err_nomem;
 839         blob->blob_location = BLOB_WIN32_ENCRYPTED;
 840
 841         /* OpenEncryptedFileRaw() expects a Win32 name.  */
 842         wimlib_assert(!wmemcmp(blob->file_on_disk, L"\\??\\", 4));
 843         blob->file_on_disk[1] = L'\\';
 844
 845         blob->file_inode = inode;
 846
 847         ret = win32_get_encrypted_file_size(blob->file_on_disk,
 848                                             (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY),
 849                                             &blob->size);
 850         if (ret)
 851                 goto err;
 852
 853         /* Empty EFSRPC data does not make sense  */
 854         wimlib_assert(blob->size != 0);
 855
 856         strm = inode_add_stream(inode, STREAM_TYPE_EFSRPC_RAW_DATA,
 857                                 NO_STREAM_NAME, blob);
 858         if (!strm)
 859                 goto err_nomem;
 860
 861         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 862         return 0;
 863
 864 err_nomem:
 865         ret = WIMLIB_ERR_NOMEM;
 866 err:
 867         free_blob_descriptor(blob);
 868         return ret;
 869 }
 870
 871 static bool
 872 get_data_stream_name(wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 873                      wchar_t **stream_name_ret, size_t *stream_name_nchars_ret)
 874 {
 875         const wchar_t *sep, *type, *end;
 876
 877         /* The stream name should be returned as :NAME:TYPE  */
 878         if (raw_stream_name_nchars < 1)
 879                 return false;
 880         if (raw_stream_name[0] != L':')
 881                 return false;
 882
 883         raw_stream_name++;
 884         raw_stream_name_nchars--;
 885
 886         end = raw_stream_name + raw_stream_name_nchars;
 887
 888         sep = wmemchr(raw_stream_name, L':', raw_stream_name_nchars);
 889         if (!sep)
 890                 return false;
 891
 892         type = sep + 1;
 893         if (end - type != 5)
 894                 return false;
 895
 896         if (wmemcmp(type, L"$DATA", 5))
 897                 return false;
 898
 899         *stream_name_ret = raw_stream_name;
 900         *stream_name_nchars_ret = sep - raw_stream_name;
 901         return true;
 902 }
 903
 904 /* Build the path to the data stream.  For unnamed streams, this is simply the
 905  * path to the file.  For named streams, this is the path to the file, followed
 906  * by a colon, followed by the stream name.  */
 907 static wchar_t *
 908 build_data_stream_path(const wchar_t *path, size_t path_nchars,
 909                        const wchar_t *stream_name, size_t stream_name_nchars)
 910 {
 911         size_t stream_path_nchars;
 912         wchar_t *stream_path;
 913         wchar_t *p;
 914
 915         stream_path_nchars = path_nchars;
 916         if (stream_name_nchars)
 917                 stream_path_nchars += 1 + stream_name_nchars;
 918
 919         stream_path = MALLOC((stream_path_nchars + 1) * sizeof(wchar_t));
 920         if (stream_path) {
 921                 p = wmempcpy(stream_path, path, path_nchars);
 922                 if (stream_name_nchars) {
 923                         *p++ = L':';
 924                         p = wmempcpy(p, stream_name, stream_name_nchars);
 925                 }
 926                 *p++ = L'\0';
 927         }
 928         return stream_path;
 929 }
 930
 931 static int
 932 winnt_scan_data_stream(const wchar_t *path, size_t path_nchars,
 933                        wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 934                        u64 stream_size,
 935                        struct wim_inode *inode, struct list_head *unhashed_blobs)
 936 {
 937         wchar_t *stream_name;
 938         size_t stream_name_nchars;
 939         struct blob_descriptor *blob;
 940         struct wim_inode_stream *strm;
 941
 942         /* Given the raw stream name (which is something like
 943          * :streamname:$DATA), extract just the stream name part (streamname).
 944          * Ignore any non-$DATA streams.  */
 945         if (!get_data_stream_name(raw_stream_name, raw_stream_name_nchars,
 946                                   &stream_name, &stream_name_nchars))
 947                 return 0;
 948
 949         stream_name[stream_name_nchars] = L'\0';
 950
 951         /* If the stream is non-empty, set up a blob descriptor for it.  */
 952         if (stream_size != 0) {
 953                 blob = new_blob_descriptor();
 954                 if (!blob)
 955                         goto err_nomem;
 956                 blob->file_on_disk = build_data_stream_path(path,
 957                                                             path_nchars,
 958                                                             stream_name,
 959                                                             stream_name_nchars);
 960                 if (!blob->file_on_disk)
 961                         goto err_nomem;
 962                 blob->blob_location = BLOB_IN_WINNT_FILE_ON_DISK;
 963                 blob->size = stream_size;
 964                 blob->file_inode = inode;
 965         } else {
 966                 blob = NULL;
 967         }
 968
 969         strm = inode_add_stream(inode, STREAM_TYPE_DATA, stream_name, blob);
 970         if (!strm)
 971                 goto err_nomem;
 972
 973         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 974         return 0;
 975
 976 err_nomem:
 977         free_blob_descriptor(blob);
 978         return WIMLIB_ERR_NOMEM;
 979 }
 980
 981 /*
 982  * Load information about the data streams of an open file into a WIM inode.
 983  *
 984  * We use the NtQueryInformationFile() system call instead of FindFirstStream()
 985  * and FindNextStream().  This is done for two reasons:
 986  *
 987  * - FindFirstStream() opens its own handle to the file or directory and
 988  *   apparently does so without specifying FILE_FLAG_BACKUP_SEMANTICS, thereby
 989  *   causing access denied errors on certain files (even when running as the
 990  *   Administrator).
 991  * - FindFirstStream() and FindNextStream() is only available on Windows Vista
 992  *   and later, whereas the stream support in NtQueryInformationFile() was
 993  *   already present in Windows XP.
 994  */
 995 static int
 996 winnt_scan_data_streams(HANDLE h, const wchar_t *path, size_t path_nchars,
 997                         struct wim_inode *inode, struct list_head *unhashed_blobs,
 998                         u64 file_size, u32 vol_flags)
 999 {
1000         int ret;
1001         u8 _buf[1024] _aligned_attribute(8);
1002         u8 *buf;
1003         size_t bufsize;
1004         IO_STATUS_BLOCK iosb;
1005         NTSTATUS status;
1006         FILE_STREAM_INFORMATION *info;
1007
1008         buf = _buf;
1009         bufsize = sizeof(_buf);
1010
1011         if (!(vol_flags & FILE_NAMED_STREAMS))
1012                 goto unnamed_only;
1013
1014         /* Get a buffer containing the stream information.  */
1015         while (!NT_SUCCESS(status = (*func_NtQueryInformationFile)(h,
1016                                                                    &iosb,
1017                                                                    buf,
1018                                                                    bufsize,
1019                                                                    FileStreamInformation)))
1020         {
1021
1022                 switch (status) {
1023                 case STATUS_BUFFER_OVERFLOW:
1024                         {
1025                                 u8 *newbuf;
1026
1027                                 bufsize *= 2;
1028                                 if (buf == _buf)
1029                                         newbuf = MALLOC(bufsize);
1030                                 else
1031                                         newbuf = REALLOC(buf, bufsize);
1032                                 if (!newbuf) {
1033                                         ret = WIMLIB_ERR_NOMEM;
1034                                         goto out_free_buf;
1035                                 }
1036                                 buf = newbuf;
1037                         }
1038                         break;
1039                 case STATUS_NOT_IMPLEMENTED:
1040                 case STATUS_NOT_SUPPORTED:
1041                 case STATUS_INVALID_INFO_CLASS:
1042                         goto unnamed_only;
1043                 default:
1044                         winnt_error(status,
1045                                     L"\"%ls\": Failed to query stream information",
1046                                     printable_path(path));
1047                         ret = WIMLIB_ERR_READ;
1048                         goto out_free_buf;
1049                 }
1050         }
1051
1052         if (iosb.Information == 0) {
1053                 /* No stream information.  */
1054                 ret = 0;
1055                 goto out_free_buf;
1056         }
1057
1058         /* Parse one or more stream information structures.  */
1059         info = (FILE_STREAM_INFORMATION *)buf;
1060         for (;;) {
1061                 /* Load the stream information.  */
1062                 ret = winnt_scan_data_stream(path, path_nchars,
1063                                              info->StreamName,
1064                                              info->StreamNameLength / 2,
1065                                              info->StreamSize.QuadPart,
1066                                              inode, unhashed_blobs);
1067                 if (ret)
1068                         goto out_free_buf;
1069
1070                 if (info->NextEntryOffset == 0) {
1071                         /* No more stream information.  */
1072                         break;
1073                 }
1074                 /* Advance to next stream information.  */
1075                 info = (FILE_STREAM_INFORMATION *)
1076                                 ((u8 *)info + info->NextEntryOffset);
1077         }
1078         ret = 0;
1079         goto out_free_buf;
1080
1081 unnamed_only:
1082         /* The volume does not support named streams.  Only capture the unnamed
1083          * data stream.  */
1084         if (inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
1085                                    FILE_ATTRIBUTE_REPARSE_POINT))
1086         {
1087                 ret = 0;
1088                 goto out_free_buf;
1089         }
1090
1091         {
1092                 wchar_t stream_name[] = L"::$DATA";
1093                 ret = winnt_scan_data_stream(path, path_nchars, stream_name, 7,
1094                                              file_size, inode, unhashed_blobs);
1095         }
1096 out_free_buf:
1097         /* Free buffer if allocated on heap.  */
1098         if (unlikely(buf != _buf))
1099                 FREE(buf);
1100         return ret;
1101 }
1102
1103 static u64
1104 get_sort_key(HANDLE h)
1105 {
1106         STARTING_VCN_INPUT_BUFFER in = { .StartingVcn.QuadPart = 0 };
1107         RETRIEVAL_POINTERS_BUFFER out;
1108         DWORD bytesReturned;
1109
1110         if (!DeviceIoControl(h, FSCTL_GET_RETRIEVAL_POINTERS,
1111                              &in, sizeof(in),
1112                              &out, sizeof(out),
1113                              &bytesReturned, NULL))
1114                 return 0;
1115
1116         if (out.ExtentCount < 1)
1117                 return 0;
1118
1119         return out.Extents[0].Lcn.QuadPart;
1120 }
1121
1122 static void
1123 set_sort_key(struct wim_inode *inode, u64 sort_key)
1124 {
1125         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1126                 struct wim_inode_stream *strm = &inode->i_streams[i];
1127                 struct blob_descriptor *blob = stream_blob_resolved(strm);
1128                 if (blob && (blob->blob_location == BLOB_IN_WINNT_FILE_ON_DISK ||
1129                              blob->blob_location == BLOB_WIN32_ENCRYPTED))
1130                         blob->sort_key = sort_key;
1131         }
1132 }
1133
1134 static int
1135 winnt_build_dentry_tree_recursive(struct wim_dentry **root_ret,
1136                                   HANDLE cur_dir,
1137                                   wchar_t *full_path,
1138                                   size_t full_path_nchars,
1139                                   const wchar_t *filename,
1140                                   size_t filename_nchars,
1141                                   struct capture_params *params,
1142                                   struct winnt_scan_stats *stats,
1143                                   u32 vol_flags)
1144 {
1145         struct wim_dentry *root = NULL;
1146         struct wim_inode *inode = NULL;
1147         HANDLE h = NULL;
1148         int ret;
1149         NTSTATUS status;
1150         FILE_ALL_INFORMATION file_info;
1151         ACCESS_MASK requestedPerms;
1152         u64 sort_key;
1153
1154         ret = try_exclude(full_path, full_path_nchars, params);
1155         if (ret < 0) /* Excluded? */
1156                 goto out_progress;
1157         if (ret > 0) /* Error? */
1158                 goto out;
1159
1160         /* Open the file.  */
1161         requestedPerms = FILE_READ_DATA |
1162                          FILE_READ_ATTRIBUTES |
1163                          READ_CONTROL |
1164                          ACCESS_SYSTEM_SECURITY |
1165                          SYNCHRONIZE;
1166 retry_open:
1167         status = winnt_openat(cur_dir,
1168                               (cur_dir ? filename : full_path),
1169                               (cur_dir ? filename_nchars : full_path_nchars),
1170                               requestedPerms,
1171                               &h);
1172         if (unlikely(!NT_SUCCESS(status))) {
1173                 if (status == STATUS_DELETE_PENDING) {
1174                         WARNING("\"%ls\": Deletion pending; skipping file",
1175                                 printable_path(full_path));
1176                         ret = 0;
1177                         goto out;
1178                 }
1179                 if (status == STATUS_ACCESS_DENIED &&
1180                     (requestedPerms & FILE_READ_DATA)) {
1181                         /* This happens on encrypted files.  */
1182                         requestedPerms &= ~FILE_READ_DATA;
1183                         goto retry_open;
1184                 }
1185
1186                 winnt_error(status, L"\"%ls\": Can't open file",
1187                             printable_path(full_path));
1188                 if (status == STATUS_FVE_LOCKED_VOLUME)
1189                         ret = WIMLIB_ERR_FVE_LOCKED_VOLUME;
1190                 else
1191                         ret = WIMLIB_ERR_OPEN;
1192                 goto out;
1193         }
1194
1195         /* Get information about the file.  */
1196         {
1197                 IO_STATUS_BLOCK iosb;
1198
1199                 status = (*func_NtQueryInformationFile)(h, &iosb,
1200                                                         &file_info,
1201                                                         sizeof(file_info),
1202                                                         FileAllInformation);
1203
1204                 if (unlikely(!NT_SUCCESS(status) &&
1205                              status != STATUS_BUFFER_OVERFLOW))
1206                 {
1207                         winnt_error(status,
1208                                     L"\"%ls\": Can't get file information",
1209                                     printable_path(full_path));
1210                         ret = WIMLIB_ERR_STAT;
1211                         goto out;
1212                 }
1213         }
1214
1215         if (unlikely(!(requestedPerms & FILE_READ_DATA)) &&
1216             !(file_info.BasicInformation.FileAttributes & FILE_ATTRIBUTE_ENCRYPTED))
1217         {
1218                 ERROR("\"%ls\": Permission to read data was denied",
1219                       printable_path(full_path));
1220                 ret = WIMLIB_ERR_OPEN;
1221                 goto out;
1222         }
1223
1224         if (unlikely(!cur_dir)) {
1225
1226                 /* Root of tree being captured; get volume information.  */
1227
1228                 FILE_FS_ATTRIBUTE_INFORMATION attr_info;
1229                 FILE_FS_VOLUME_INFORMATION vol_info;
1230                 IO_STATUS_BLOCK iosb;
1231
1232                 /* Get volume flags  */
1233                 status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1234                                                               &attr_info,
1235                                                               sizeof(attr_info),
1236                                                               FileFsAttributeInformation);
1237                 if (likely((NT_SUCCESS(status) ||
1238                             (status == STATUS_BUFFER_OVERFLOW)) &&
1239                            (iosb.Information >=
1240                                 offsetof(FILE_FS_ATTRIBUTE_INFORMATION,
1241                                          FileSystemAttributes) +
1242                                 sizeof(attr_info.FileSystemAttributes))))
1243                 {
1244                         vol_flags = attr_info.FileSystemAttributes;
1245                 } else {
1246                         winnt_warning(status,
1247                                       L"\"%ls\": Can't get volume attributes",
1248                                       printable_path(full_path));
1249                         vol_flags = 0;
1250                 }
1251
1252                 /* Set inode number of root directory  */
1253                 params->capture_root_ino =
1254                         file_info.InternalInformation.IndexNumber.QuadPart;
1255
1256                 /* Get volume ID.  */
1257                 status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1258                                                               &vol_info,
1259                                                               sizeof(vol_info),
1260                                                               FileFsVolumeInformation);
1261                 if (likely((NT_SUCCESS(status) ||
1262                             (status == STATUS_BUFFER_OVERFLOW)) &&
1263                            (iosb.Information >=
1264                                 offsetof(FILE_FS_VOLUME_INFORMATION,
1265                                          VolumeSerialNumber) +
1266                                 sizeof(vol_info.VolumeSerialNumber))))
1267                 {
1268                         params->capture_root_dev = vol_info.VolumeSerialNumber;
1269                 } else {
1270                         winnt_warning(status, L"\"%ls\": Can't get volume ID",
1271                                       printable_path(full_path));
1272                         params->capture_root_dev = 0;
1273                 }
1274         }
1275
1276         /* Create a WIM dentry with an associated inode, which may be shared.
1277          *
1278          * However, we need to explicitly check for directories and files with
1279          * only 1 link and refuse to hard link them.  This is because Windows
1280          * has a bug where it can return duplicate File IDs for files and
1281          * directories on the FAT filesystem.
1282          *
1283          * Since we don't follow mount points on Windows, we don't need to query
1284          * the volume ID per-file.  Just once, for the root, is enough.  But we
1285          * can't simply pass 0, because then there could be inode collisions
1286          * among multiple calls to win32_build_dentry_tree() that are scanning
1287          * files on different volumes.  */
1288         ret = inode_table_new_dentry(params->inode_table,
1289                                      filename,
1290                                      file_info.InternalInformation.IndexNumber.QuadPart,
1291                                      params->capture_root_dev,
1292                                      (file_info.StandardInformation.NumberOfLinks <= 1 ||
1293                                         (file_info.BasicInformation.FileAttributes &
1294                                          FILE_ATTRIBUTE_DIRECTORY)),
1295                                      &root);
1296         if (ret)
1297                 goto out;
1298
1299         /* Get the short (DOS) name of the file.  */
1300         status = winnt_get_short_name(h, root);
1301
1302         /* If we can't read the short filename for any reason other than
1303          * out-of-memory, just ignore the error and assume the file has no short
1304          * name.  This shouldn't be an issue, since the short names are
1305          * essentially obsolete anyway.  */
1306         if (unlikely(status == STATUS_NO_MEMORY)) {
1307                 ret = WIMLIB_ERR_NOMEM;
1308                 goto out;
1309         }
1310
1311         inode = root->d_inode;
1312
1313         if (inode->i_nlink > 1) {
1314                 /* Shared inode (hard link); skip reading per-inode information.
1315                  */
1316                 goto out_progress;
1317         }
1318
1319         inode->i_attributes = file_info.BasicInformation.FileAttributes;
1320         inode->i_creation_time = file_info.BasicInformation.CreationTime.QuadPart;
1321         inode->i_last_write_time = file_info.BasicInformation.LastWriteTime.QuadPart;
1322         inode->i_last_access_time = file_info.BasicInformation.LastAccessTime.QuadPart;
1323
1324         /* Get the file's security descriptor, unless we are capturing in
1325          * NO_ACLS mode or the volume does not support security descriptors.  */
1326         if (!(params->add_flags & WIMLIB_ADD_FLAG_NO_ACLS)
1327             && (vol_flags & FILE_PERSISTENT_ACLS))
1328         {
1329                 status = winnt_get_security_descriptor(h, inode,
1330                                                        params->sd_set, stats,
1331                                                        params->add_flags);
1332                 if (!NT_SUCCESS(status)) {
1333                         winnt_error(status,
1334                                     L"\"%ls\": Can't read security descriptor",
1335                                     printable_path(full_path));
1336                         ret = WIMLIB_ERR_STAT;
1337                         goto out;
1338                 }
1339         }
1340
1341         /* If this is a reparse point, load the reparse data.  */
1342         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)) {
1343                 if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1344                         /* See comment above assign_stream_types_encrypted()  */
1345                         WARNING("Ignoring reparse data of encrypted file \"%ls\"",
1346                                 printable_path(full_path));
1347                 } else {
1348                         u8 rpbuf[REPARSE_POINT_MAX_SIZE] _aligned_attribute(8);
1349                         u16 rpbuflen;
1350
1351                         ret = winnt_get_reparse_data(h, full_path, params,
1352                                                      rpbuf, &rpbuflen);
1353                         switch (ret) {
1354                         case RP_FIXED:
1355                                 inode->i_not_rpfixed = 0;
1356                                 break;
1357                         case RP_NOT_FIXED:
1358                                 inode->i_not_rpfixed = 1;
1359                                 break;
1360                         default:
1361                                 goto out;
1362                         }
1363                         inode->i_reparse_tag = le32_to_cpu(*(le32*)rpbuf);
1364                         if (!inode_add_stream_with_data(inode,
1365                                                         STREAM_TYPE_REPARSE_POINT,
1366                                                         NO_STREAM_NAME,
1367                                                         rpbuf + REPARSE_DATA_OFFSET,
1368                                                         rpbuflen - REPARSE_DATA_OFFSET,
1369                                                         params->blob_table))
1370                         {
1371                                 ret = WIMLIB_ERR_NOMEM;
1372                                 goto out;
1373                         }
1374                 }
1375         }
1376
1377         sort_key = get_sort_key(h);
1378
1379         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED)) {
1380                 /* Load information about the raw encrypted data.  This is
1381                  * needed for any directory or non-directory that has
1382                  * FILE_ATTRIBUTE_ENCRYPTED set.
1383                  *
1384                  * Note: since OpenEncryptedFileRaw() fails with
1385                  * ERROR_SHARING_VIOLATION if there are any open handles to the
1386                  * file, we have to close the file and re-open it later if
1387                  * needed.  */
1388                 (*func_NtClose)(h);
1389                 h = NULL;
1390                 ret = winnt_scan_efsrpc_raw_data(inode, full_path,
1391                                                  params->unhashed_blobs);
1392                 if (ret)
1393                         goto out;
1394         } else {
1395                 /*
1396                  * Load information about data streams (unnamed and named).
1397                  *
1398                  * Skip this step for encrypted files, since the data from
1399                  * ReadEncryptedFileRaw() already contains all data streams (and
1400                  * they do in fact all get restored by WriteEncryptedFileRaw().)
1401                  *
1402                  * Note: WIMGAPI (as of Windows 8.1) gets wrong and stores both
1403                  * the EFSRPC data and the named data stream(s)...!
1404                  */
1405                 ret = winnt_scan_data_streams(h,
1406                                               full_path,
1407                                               full_path_nchars,
1408                                               inode,
1409                                               params->unhashed_blobs,
1410                                               file_info.StandardInformation.EndOfFile.QuadPart,
1411                                               vol_flags);
1412                 if (ret)
1413                         goto out;
1414         }
1415
1416         set_sort_key(inode, sort_key);
1417
1418         if (inode_is_directory(inode)) {
1419
1420                 /* Directory: recurse to children.  */
1421
1422                 if (unlikely(!h)) {
1423                         /* Re-open handle that was closed to read raw encrypted
1424                          * data.  */
1425                         status = winnt_openat(cur_dir,
1426                                               (cur_dir ?
1427                                                filename : full_path),
1428                                               (cur_dir ?
1429                                                filename_nchars : full_path_nchars),
1430                                               FILE_LIST_DIRECTORY | SYNCHRONIZE,
1431                                               &h);
1432                         if (!NT_SUCCESS(status)) {
1433                                 winnt_error(status,
1434                                             L"\"%ls\": Can't re-open file",
1435                                             printable_path(full_path));
1436                                 ret = WIMLIB_ERR_OPEN;
1437                                 goto out;
1438                         }
1439                 }
1440                 ret = winnt_recurse_directory(h,
1441                                               full_path,
1442                                               full_path_nchars,
1443                                               root,
1444                                               params,
1445                                               stats,
1446                                               vol_flags);
1447                 if (ret)
1448                         goto out;
1449         }
1450
1451 out_progress:
1452         params->progress.scan.cur_path = printable_path(full_path);
1453         if (likely(root))
1454                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
1455         else
1456                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
1457 out:
1458         if (likely(h))
1459                 (*func_NtClose)(h);
1460         if (unlikely(ret)) {
1461                 free_dentry_tree(root, params->blob_table);
1462                 root = NULL;
1463                 ret = report_capture_error(params, ret, full_path);
1464         }
1465         *root_ret = root;
1466         return ret;
1467 }
1468
1469 static void
1470 winnt_do_scan_warnings(const wchar_t *path, const struct winnt_scan_stats *stats)
1471 {
1472         if (likely(stats->num_get_sacl_priv_notheld == 0 &&
1473                    stats->num_get_sd_access_denied == 0))
1474                 return;
1475
1476         WARNING("Scan of \"%ls\" complete, but with one or more warnings:", path);
1477         if (stats->num_get_sacl_priv_notheld != 0) {
1478                 WARNING("- Could not capture SACL (System Access Control List)\n"
1479                         "            on %lu files or directories.",
1480                         stats->num_get_sacl_priv_notheld);
1481         }
1482         if (stats->num_get_sd_access_denied != 0) {
1483                 WARNING("- Could not capture security descriptor at all\n"
1484                         "            on %lu files or directories.",
1485                         stats->num_get_sd_access_denied);
1486         }
1487         WARNING("To fully capture all security descriptors, run the program\n"
1488                 "          with Administrator rights.");
1489 }
1490
1491 #define WINDOWS_NT_MAX_PATH 32768
1492
1493 /* Win32 version of capturing a directory tree.  */
1494 int
1495 win32_build_dentry_tree(struct wim_dentry **root_ret,
1496                         const wchar_t *root_disk_path,
1497                         struct capture_params *params)
1498 {
1499         wchar_t *path;
1500         int ret;
1501         UNICODE_STRING ntpath;
1502         struct winnt_scan_stats stats;
1503         size_t ntpath_nchars;
1504
1505         /* WARNING: There is no check for overflow later when this buffer is
1506          * being used!  But it's as long as the maximum path length understood
1507          * by Windows NT (which is NOT the same as MAX_PATH).  */
1508         path = MALLOC((WINDOWS_NT_MAX_PATH + 1) * sizeof(wchar_t));
1509         if (!path)
1510                 return WIMLIB_ERR_NOMEM;
1511
1512         ret = win32_path_to_nt_path(root_disk_path, &ntpath);
1513         if (ret)
1514                 goto out_free_path;
1515
1516         if (ntpath.Length < 4 * sizeof(wchar_t) ||
1517             ntpath.Length > WINDOWS_NT_MAX_PATH * sizeof(wchar_t) ||
1518             wmemcmp(ntpath.Buffer, L"\\??\\", 4))
1519         {
1520                 ERROR("\"%ls\": unrecognized path format", root_disk_path);
1521                 ret = WIMLIB_ERR_INVALID_PARAM;
1522         } else {
1523                 ntpath_nchars = ntpath.Length / sizeof(wchar_t);
1524                 wmemcpy(path, ntpath.Buffer, ntpath_nchars);
1525                 path[ntpath_nchars] = L'\0';
1526
1527                 params->capture_root_nchars = ntpath_nchars;
1528                 if (path[ntpath_nchars - 1] == L'\\')
1529                         params->capture_root_nchars--;
1530                 ret = 0;
1531         }
1532         HeapFree(GetProcessHeap(), 0, ntpath.Buffer);
1533         if (ret)
1534                 goto out_free_path;
1535
1536         memset(&stats, 0, sizeof(stats));
1537
1538         ret = winnt_build_dentry_tree_recursive(root_ret, NULL,
1539                                                 path, ntpath_nchars,
1540                                                 L"", 0, params, &stats, 0);
1541 out_free_path:
1542         FREE(path);
1543         if (ret == 0)
1544                 winnt_do_scan_warnings(root_disk_path, &stats);
1545         return ret;
1546 }
1547
1548 #endif /* __WIN32__ */