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 noinline_for_stack 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->d_short_name = utf16le_dupz(info->FileName,
 252                                                     info->FileNameLength);
 253                 if (!dentry->d_short_name)
 254                         return STATUS_NO_MEMORY;
 255                 dentry->d_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 noinline_for_stack 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 #define RP_FIXED        (-1)
 475
 476 static bool
 477 file_has_ino_and_dev(HANDLE h, u64 ino, u64 dev)
 478 {
 479         NTSTATUS status;
 480         IO_STATUS_BLOCK iosb;
 481         FILE_INTERNAL_INFORMATION int_info;
 482         FILE_FS_VOLUME_INFORMATION vol_info;
 483
 484         status = (*func_NtQueryInformationFile)(h, &iosb,
 485                                                 &int_info, sizeof(int_info),
 486                                                 FileInternalInformation);
 487         if (!NT_SUCCESS(status))
 488                 return false;
 489
 490         if (int_info.IndexNumber.QuadPart != ino)
 491                 return false;
 492
 493         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
 494                                                       &vol_info, sizeof(vol_info),
 495                                                       FileFsVolumeInformation);
 496         if (!(NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW))
 497                 return false;
 498
 499         if (iosb.Information <
 500              offsetof(FILE_FS_VOLUME_INFORMATION, VolumeSerialNumber) +
 501              sizeof(vol_info.VolumeSerialNumber))
 502                 return false;
 503
 504         return (vol_info.VolumeSerialNumber == dev);
 505 }
 506
 507 /*
 508  * This is the Windows equivalent of unix_relativize_link_target(); see there
 509  * for general details.  This version works with an "absolute" Windows link
 510  * target, specified from the root of the Windows kernel object namespace.  Note
 511  * that we have to open directories with a trailing slash when present because
 512  * \??\E: opens the E: device itself and not the filesystem root directory.
 513  */
 514 static const wchar_t *
 515 winnt_relativize_link_target(const wchar_t *target, size_t target_nbytes,
 516                              u64 ino, u64 dev)
 517 {
 518         UNICODE_STRING name;
 519         OBJECT_ATTRIBUTES attr;
 520         IO_STATUS_BLOCK iosb;
 521         NTSTATUS status;
 522         const wchar_t *target_end;
 523         const wchar_t *p;
 524
 525         target_end = target + (target_nbytes / sizeof(wchar_t));
 526
 527         /* Empty path??? */
 528         if (target_end == target)
 529                 return target;
 530
 531         /* No leading slash???  */
 532         if (target[0] != L'\\')
 533                 return target;
 534
 535         /* UNC path???  */
 536         if ((target_end - target) >= 2 &&
 537             target[0] == L'\\' && target[1] == L'\\')
 538                 return target;
 539
 540         attr.Length = sizeof(attr);
 541         attr.RootDirectory = NULL;
 542         attr.ObjectName = &name;
 543         attr.Attributes = 0;
 544         attr.SecurityDescriptor = NULL;
 545         attr.SecurityQualityOfService = NULL;
 546
 547         name.Buffer = (wchar_t *)target;
 548         name.Length = 0;
 549         p = target;
 550         do {
 551                 HANDLE h;
 552                 const wchar_t *orig_p = p;
 553
 554                 /* Skip non-backslashes  */
 555                 while (p != target_end && *p != L'\\')
 556                         p++;
 557
 558                 /* Skip backslashes  */
 559                 while (p != target_end && *p == L'\\')
 560                         p++;
 561
 562                 /* Append path component  */
 563                 name.Length += (p - orig_p) * sizeof(wchar_t);
 564                 name.MaximumLength = name.Length;
 565
 566                 /* Try opening the file  */
 567                 status = (*func_NtOpenFile) (&h,
 568                                              FILE_READ_ATTRIBUTES | FILE_TRAVERSE,
 569                                              &attr,
 570                                              &iosb,
 571                                              FILE_SHARE_VALID_FLAGS,
 572                                              FILE_OPEN_FOR_BACKUP_INTENT);
 573
 574                 if (NT_SUCCESS(status)) {
 575                         /* Reset root directory  */
 576                         if (attr.RootDirectory)
 577                                 (*func_NtClose)(attr.RootDirectory);
 578                         attr.RootDirectory = h;
 579                         name.Buffer = (wchar_t *)p;
 580                         name.Length = 0;
 581
 582                         if (file_has_ino_and_dev(h, ino, dev))
 583                                 goto out_close_root_dir;
 584                 }
 585         } while (p != target_end);
 586
 587         p = target;
 588
 589 out_close_root_dir:
 590         if (attr.RootDirectory)
 591                 (*func_NtClose)(attr.RootDirectory);
 592         while (p > target && *(p - 1) == L'\\')
 593                 p--;
 594         return p;
 595 }
 596
 597 static int
 598 winnt_rpfix_progress(struct capture_params *params, const wchar_t *path,
 599                      const struct link_reparse_point *link, int scan_status)
 600 {
 601         size_t print_name_nchars = link->print_name_nbytes / sizeof(wchar_t);
 602         wchar_t print_name0[print_name_nchars + 1];
 603
 604         wmemcpy(print_name0, link->print_name, print_name_nchars);
 605         print_name0[print_name_nchars] = L'\0';
 606
 607         params->progress.scan.cur_path = printable_path(path);
 608         params->progress.scan.symlink_target = print_name0;
 609         return do_capture_progress(params, scan_status, NULL);
 610 }
 611
 612 static int
 613 winnt_try_rpfix(struct reparse_buffer_disk *rpbuf, u16 *rpbuflen_p,
 614                 const wchar_t *path, struct capture_params *params)
 615 {
 616         struct link_reparse_point link;
 617         const wchar_t *rel_target;
 618         int ret;
 619
 620         if (parse_link_reparse_point(rpbuf, *rpbuflen_p, &link)) {
 621                 /* Couldn't understand the reparse data; don't do the fixup.  */
 622                 return 0;
 623         }
 624
 625         /*
 626          * Don't do reparse point fixups on relative symbolic links.
 627          *
 628          * On Windows, a relative symbolic link is supposed to be identifiable
 629          * by having reparse tag WIM_IO_REPARSE_TAG_SYMLINK and flags
 630          * SYMBOLIC_LINK_RELATIVE.  We will use this information, although this
 631          * may not always do what the user expects, since drive-relative
 632          * symbolic links such as "\Users\Public" have SYMBOLIC_LINK_RELATIVE
 633          * set, in addition to truely relative symbolic links such as "Users" or
 634          * "Users\Public".  However, WIMGAPI (as of Windows 8.1) has this same
 635          * behavior.
 636          *
 637          * Otherwise, as far as I can tell, the targets of symbolic links that
 638          * are NOT relative, as well as junctions (note: a mountpoint is the
 639          * sames thing as a junction), must be NT namespace paths, for example:
 640          *
 641          *     - \??\e:\Users\Public
 642          *     - \DosDevices\e:\Users\Public
 643          *     - \Device\HardDiskVolume4\Users\Public
 644          *     - \??\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 645          *     - \DosDevices\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 646          */
 647         if (link_is_relative_symlink(&link))
 648                 return 0;
 649
 650         rel_target = winnt_relativize_link_target(link.substitute_name,
 651                                                   link.substitute_name_nbytes,
 652                                                   params->capture_root_ino,
 653                                                   params->capture_root_dev);
 654
 655         if (rel_target == link.substitute_name) {
 656                 /* Target points outside of the tree being captured or had an
 657                  * unrecognized path format.  Don't adjust it.  */
 658                 return winnt_rpfix_progress(params, path, &link,
 659                                             WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK);
 660         }
 661
 662         /* We have an absolute target pointing within the directory being
 663          * captured. @rel_target is the suffix of the link target that is the
 664          * part relative to the directory being captured.
 665          *
 666          * We will cut off the prefix before this part (which is the path to the
 667          * directory being captured) and add a dummy prefix.  Since the process
 668          * will need to be reversed when applying the image, it doesn't matter
 669          * what exactly the prefix is, as long as it looks like an absolute
 670          * path.  */
 671
 672         static const wchar_t prefix[6] = L"\\??\\X:";
 673         static const size_t num_unprintable_chars = 4;
 674
 675         size_t rel_target_nbytes =
 676                 link.substitute_name_nbytes - ((const u8 *)rel_target -
 677                                                (const u8 *)link.substitute_name);
 678
 679         wchar_t tmp[(sizeof(prefix) + rel_target_nbytes) / sizeof(wchar_t)];
 680
 681         memcpy(tmp, prefix, sizeof(prefix));
 682         memcpy(tmp + ARRAY_LEN(prefix), rel_target, rel_target_nbytes);
 683
 684         link.substitute_name = tmp;
 685         link.substitute_name_nbytes = sizeof(tmp);
 686
 687         link.print_name = link.substitute_name + num_unprintable_chars;
 688         link.print_name_nbytes = link.substitute_name_nbytes -
 689                                  (num_unprintable_chars * sizeof(wchar_t));
 690
 691         if (make_link_reparse_point(&link, rpbuf, rpbuflen_p))
 692                 return 0;
 693
 694         ret = winnt_rpfix_progress(params, path, &link,
 695                                    WIMLIB_SCAN_DENTRY_FIXED_SYMLINK);
 696         if (ret)
 697                 return ret;
 698         return RP_FIXED;
 699 }
 700
 701 /* Load the reparse data of a file into the corresponding WIM inode.  If the
 702  * reparse point is a symbolic link or junction with an absolute target and
 703  * RPFIX mode is enabled, then also rewrite its target to be relative to the
 704  * capture root.  */
 705 static noinline_for_stack int
 706 winnt_load_reparse_data(HANDLE h, struct wim_inode *inode,
 707                         const wchar_t *full_path, struct capture_params *params)
 708 {
 709         struct reparse_buffer_disk rpbuf;
 710         DWORD bytes_returned;
 711         u16 rpbuflen;
 712         int ret;
 713
 714         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
 715                 /* See comment above assign_stream_types_encrypted()  */
 716                 WARNING("Ignoring reparse data of encrypted file \"%ls\"",
 717                         printable_path(full_path));
 718                 return 0;
 719         }
 720
 721         if (!DeviceIoControl(h, FSCTL_GET_REPARSE_POINT,
 722                              NULL, 0, &rpbuf, REPARSE_POINT_MAX_SIZE,
 723                              &bytes_returned, NULL))
 724         {
 725                 win32_error(GetLastError(), L"\"%ls\": Can't get reparse point",
 726                             printable_path(full_path));
 727                 return WIMLIB_ERR_READLINK;
 728         }
 729
 730         rpbuflen = bytes_returned;
 731
 732         if (unlikely(rpbuflen < REPARSE_DATA_OFFSET)) {
 733                 ERROR("\"%ls\": reparse point buffer is too short",
 734                       printable_path(full_path));
 735                 return WIMLIB_ERR_INVALID_REPARSE_DATA;
 736         }
 737
 738         if (params->add_flags & WIMLIB_ADD_FLAG_RPFIX) {
 739                 ret = winnt_try_rpfix(&rpbuf, &rpbuflen, full_path, params);
 740                 if (ret == RP_FIXED)
 741                         inode->i_rp_flags &= ~WIM_RP_FLAG_NOT_FIXED;
 742                 else if (ret)
 743                         return ret;
 744         }
 745
 746         inode->i_reparse_tag = le32_to_cpu(rpbuf.rptag);
 747         inode->i_rp_reserved = le16_to_cpu(rpbuf.rpreserved);
 748
 749         if (!inode_add_stream_with_data(inode,
 750                                         STREAM_TYPE_REPARSE_POINT,
 751                                         NO_STREAM_NAME,
 752                                         rpbuf.rpdata,
 753                                         rpbuflen - REPARSE_DATA_OFFSET,
 754                                         params->blob_table))
 755                 return WIMLIB_ERR_NOMEM;
 756
 757         return 0;
 758 }
 759
 760 static DWORD WINAPI
 761 win32_tally_encrypted_size_cb(unsigned char *_data, void *_size_ret,
 762                               unsigned long len)
 763 {
 764         *(u64*)_size_ret += len;
 765         return ERROR_SUCCESS;
 766 }
 767
 768 static int
 769 win32_get_encrypted_file_size(const wchar_t *path, bool is_dir, u64 *size_ret)
 770 {
 771         DWORD err;
 772         void *file_ctx;
 773         int ret;
 774         DWORD flags = 0;
 775
 776         if (is_dir)
 777                 flags |= CREATE_FOR_DIR;
 778
 779         err = OpenEncryptedFileRaw(path, flags, &file_ctx);
 780         if (err != ERROR_SUCCESS) {
 781                 win32_error(err,
 782                             L"Failed to open encrypted file \"%ls\" for raw read",
 783                             printable_path(path));
 784                 return WIMLIB_ERR_OPEN;
 785         }
 786         *size_ret = 0;
 787         err = ReadEncryptedFileRaw(win32_tally_encrypted_size_cb,
 788                                    size_ret, file_ctx);
 789         if (err != ERROR_SUCCESS) {
 790                 win32_error(err,
 791                             L"Failed to read raw encrypted data from \"%ls\"",
 792                             printable_path(path));
 793                 ret = WIMLIB_ERR_READ;
 794         } else {
 795                 ret = 0;
 796         }
 797         CloseEncryptedFileRaw(file_ctx);
 798         return ret;
 799 }
 800
 801 static int
 802 winnt_scan_efsrpc_raw_data(struct wim_inode *inode, const wchar_t *nt_path,
 803                            struct list_head *unhashed_blobs)
 804 {
 805         struct blob_descriptor *blob;
 806         struct wim_inode_stream *strm;
 807         int ret;
 808
 809         blob = new_blob_descriptor();
 810         if (!blob)
 811                 goto err_nomem;
 812
 813         blob->file_on_disk = WCSDUP(nt_path);
 814         if (!blob->file_on_disk)
 815                 goto err_nomem;
 816         blob->blob_location = BLOB_WIN32_ENCRYPTED;
 817
 818         /* OpenEncryptedFileRaw() expects a Win32 name.  */
 819         wimlib_assert(!wmemcmp(blob->file_on_disk, L"\\??\\", 4));
 820         blob->file_on_disk[1] = L'\\';
 821
 822         blob->file_inode = inode;
 823
 824         ret = win32_get_encrypted_file_size(blob->file_on_disk,
 825                                             (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY),
 826                                             &blob->size);
 827         if (ret)
 828                 goto err;
 829
 830         /* Empty EFSRPC data does not make sense  */
 831         wimlib_assert(blob->size != 0);
 832
 833         strm = inode_add_stream(inode, STREAM_TYPE_EFSRPC_RAW_DATA,
 834                                 NO_STREAM_NAME, blob);
 835         if (!strm)
 836                 goto err_nomem;
 837
 838         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 839         return 0;
 840
 841 err_nomem:
 842         ret = WIMLIB_ERR_NOMEM;
 843 err:
 844         free_blob_descriptor(blob);
 845         return ret;
 846 }
 847
 848 static bool
 849 get_data_stream_name(wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 850                      wchar_t **stream_name_ret, size_t *stream_name_nchars_ret)
 851 {
 852         const wchar_t *sep, *type, *end;
 853
 854         /* The stream name should be returned as :NAME:TYPE  */
 855         if (raw_stream_name_nchars < 1)
 856                 return false;
 857         if (raw_stream_name[0] != L':')
 858                 return false;
 859
 860         raw_stream_name++;
 861         raw_stream_name_nchars--;
 862
 863         end = raw_stream_name + raw_stream_name_nchars;
 864
 865         sep = wmemchr(raw_stream_name, L':', raw_stream_name_nchars);
 866         if (!sep)
 867                 return false;
 868
 869         type = sep + 1;
 870         if (end - type != 5)
 871                 return false;
 872
 873         if (wmemcmp(type, L"$DATA", 5))
 874                 return false;
 875
 876         *stream_name_ret = raw_stream_name;
 877         *stream_name_nchars_ret = sep - raw_stream_name;
 878         return true;
 879 }
 880
 881 /* Build the path to the data stream.  For unnamed streams, this is simply the
 882  * path to the file.  For named streams, this is the path to the file, followed
 883  * by a colon, followed by the stream name.  */
 884 static wchar_t *
 885 build_data_stream_path(const wchar_t *path, size_t path_nchars,
 886                        const wchar_t *stream_name, size_t stream_name_nchars)
 887 {
 888         size_t stream_path_nchars;
 889         wchar_t *stream_path;
 890         wchar_t *p;
 891
 892         stream_path_nchars = path_nchars;
 893         if (stream_name_nchars)
 894                 stream_path_nchars += 1 + stream_name_nchars;
 895
 896         stream_path = MALLOC((stream_path_nchars + 1) * sizeof(wchar_t));
 897         if (stream_path) {
 898                 p = wmempcpy(stream_path, path, path_nchars);
 899                 if (stream_name_nchars) {
 900                         *p++ = L':';
 901                         p = wmempcpy(p, stream_name, stream_name_nchars);
 902                 }
 903                 *p++ = L'\0';
 904         }
 905         return stream_path;
 906 }
 907
 908 static int
 909 winnt_scan_data_stream(const wchar_t *path, size_t path_nchars,
 910                        wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 911                        u64 stream_size,
 912                        struct wim_inode *inode, struct list_head *unhashed_blobs)
 913 {
 914         wchar_t *stream_name;
 915         size_t stream_name_nchars;
 916         struct blob_descriptor *blob;
 917         struct wim_inode_stream *strm;
 918
 919         /* Given the raw stream name (which is something like
 920          * :streamname:$DATA), extract just the stream name part (streamname).
 921          * Ignore any non-$DATA streams.  */
 922         if (!get_data_stream_name(raw_stream_name, raw_stream_name_nchars,
 923                                   &stream_name, &stream_name_nchars))
 924                 return 0;
 925
 926         stream_name[stream_name_nchars] = L'\0';
 927
 928         /* If the stream is non-empty, set up a blob descriptor for it.  */
 929         if (stream_size != 0) {
 930                 blob = new_blob_descriptor();
 931                 if (!blob)
 932                         goto err_nomem;
 933                 blob->file_on_disk = build_data_stream_path(path,
 934                                                             path_nchars,
 935                                                             stream_name,
 936                                                             stream_name_nchars);
 937                 if (!blob->file_on_disk)
 938                         goto err_nomem;
 939                 blob->blob_location = BLOB_IN_WINNT_FILE_ON_DISK;
 940                 blob->size = stream_size;
 941                 blob->file_inode = inode;
 942         } else {
 943                 blob = NULL;
 944         }
 945
 946         strm = inode_add_stream(inode, STREAM_TYPE_DATA, stream_name, blob);
 947         if (!strm)
 948                 goto err_nomem;
 949
 950         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 951         return 0;
 952
 953 err_nomem:
 954         free_blob_descriptor(blob);
 955         return WIMLIB_ERR_NOMEM;
 956 }
 957
 958 /*
 959  * Load information about the data streams of an open file into a WIM inode.
 960  *
 961  * We use the NtQueryInformationFile() system call instead of FindFirstStream()
 962  * and FindNextStream().  This is done for two reasons:
 963  *
 964  * - FindFirstStream() opens its own handle to the file or directory and
 965  *   apparently does so without specifying FILE_FLAG_BACKUP_SEMANTICS, thereby
 966  *   causing access denied errors on certain files (even when running as the
 967  *   Administrator).
 968  * - FindFirstStream() and FindNextStream() is only available on Windows Vista
 969  *   and later, whereas the stream support in NtQueryInformationFile() was
 970  *   already present in Windows XP.
 971  */
 972 static noinline_for_stack int
 973 winnt_scan_data_streams(HANDLE h, const wchar_t *path, size_t path_nchars,
 974                         struct wim_inode *inode, struct list_head *unhashed_blobs,
 975                         u64 file_size, u32 vol_flags)
 976 {
 977         int ret;
 978         u8 _buf[4096] _aligned_attribute(8);
 979         u8 *buf;
 980         size_t bufsize;
 981         IO_STATUS_BLOCK iosb;
 982         NTSTATUS status;
 983         FILE_STREAM_INFORMATION *info;
 984
 985         buf = _buf;
 986         bufsize = sizeof(_buf);
 987
 988         if (!(vol_flags & FILE_NAMED_STREAMS))
 989                 goto unnamed_only;
 990
 991         /* Get a buffer containing the stream information.  */
 992         while (!NT_SUCCESS(status = (*func_NtQueryInformationFile)(h,
 993                                                                    &iosb,
 994                                                                    buf,
 995                                                                    bufsize,
 996                                                                    FileStreamInformation)))
 997         {
 998
 999                 switch (status) {
1000                 case STATUS_BUFFER_OVERFLOW:
1001                         {
1002                                 u8 *newbuf;
1003
1004                                 bufsize *= 2;
1005                                 if (buf == _buf)
1006                                         newbuf = MALLOC(bufsize);
1007                                 else
1008                                         newbuf = REALLOC(buf, bufsize);
1009                                 if (!newbuf) {
1010                                         ret = WIMLIB_ERR_NOMEM;
1011                                         goto out_free_buf;
1012                                 }
1013                                 buf = newbuf;
1014                         }
1015                         break;
1016                 case STATUS_NOT_IMPLEMENTED:
1017                 case STATUS_NOT_SUPPORTED:
1018                 case STATUS_INVALID_INFO_CLASS:
1019                         goto unnamed_only;
1020                 default:
1021                         winnt_error(status,
1022                                     L"\"%ls\": Failed to query stream information",
1023                                     printable_path(path));
1024                         ret = WIMLIB_ERR_READ;
1025                         goto out_free_buf;
1026                 }
1027         }
1028
1029         if (iosb.Information == 0) {
1030                 /* No stream information.  */
1031                 ret = 0;
1032                 goto out_free_buf;
1033         }
1034
1035         /* Parse one or more stream information structures.  */
1036         info = (FILE_STREAM_INFORMATION *)buf;
1037         for (;;) {
1038                 /* Load the stream information.  */
1039                 ret = winnt_scan_data_stream(path, path_nchars,
1040                                              info->StreamName,
1041                                              info->StreamNameLength / 2,
1042                                              info->StreamSize.QuadPart,
1043                                              inode, unhashed_blobs);
1044                 if (ret)
1045                         goto out_free_buf;
1046
1047                 if (info->NextEntryOffset == 0) {
1048                         /* No more stream information.  */
1049                         break;
1050                 }
1051                 /* Advance to next stream information.  */
1052                 info = (FILE_STREAM_INFORMATION *)
1053                                 ((u8 *)info + info->NextEntryOffset);
1054         }
1055         ret = 0;
1056         goto out_free_buf;
1057
1058 unnamed_only:
1059         /* The volume does not support named streams.  Only capture the unnamed
1060          * data stream.  */
1061         if (inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
1062                                    FILE_ATTRIBUTE_REPARSE_POINT))
1063         {
1064                 ret = 0;
1065                 goto out_free_buf;
1066         }
1067
1068         {
1069                 wchar_t stream_name[] = L"::$DATA";
1070                 ret = winnt_scan_data_stream(path, path_nchars, stream_name, 7,
1071                                              file_size, inode, unhashed_blobs);
1072         }
1073 out_free_buf:
1074         /* Free buffer if allocated on heap.  */
1075         if (unlikely(buf != _buf))
1076                 FREE(buf);
1077         return ret;
1078 }
1079
1080 static noinline_for_stack u64
1081 get_sort_key(HANDLE h)
1082 {
1083         STARTING_VCN_INPUT_BUFFER in = { .StartingVcn.QuadPart = 0 };
1084         RETRIEVAL_POINTERS_BUFFER out;
1085         DWORD bytesReturned;
1086
1087         if (!DeviceIoControl(h, FSCTL_GET_RETRIEVAL_POINTERS,
1088                              &in, sizeof(in),
1089                              &out, sizeof(out),
1090                              &bytesReturned, NULL))
1091                 return 0;
1092
1093         if (out.ExtentCount < 1)
1094                 return 0;
1095
1096         return out.Extents[0].Lcn.QuadPart;
1097 }
1098
1099 static void
1100 set_sort_key(struct wim_inode *inode, u64 sort_key)
1101 {
1102         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1103                 struct wim_inode_stream *strm = &inode->i_streams[i];
1104                 struct blob_descriptor *blob = stream_blob_resolved(strm);
1105                 if (blob && (blob->blob_location == BLOB_IN_WINNT_FILE_ON_DISK ||
1106                              blob->blob_location == BLOB_WIN32_ENCRYPTED))
1107                         blob->sort_key = sort_key;
1108         }
1109 }
1110
1111 static noinline_for_stack u32
1112 get_volume_information(HANDLE h, const wchar_t *full_path,
1113                        struct capture_params *params)
1114 {
1115         FILE_FS_ATTRIBUTE_INFORMATION attr_info;
1116         FILE_FS_VOLUME_INFORMATION vol_info;
1117         IO_STATUS_BLOCK iosb;
1118         NTSTATUS status;
1119         u32 vol_flags;
1120
1121         /* Get volume flags  */
1122         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1123                                                       &attr_info,
1124                                                       sizeof(attr_info),
1125                                                       FileFsAttributeInformation);
1126         if (likely((NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW) &&
1127                    (iosb.Information >=
1128                         offsetof(FILE_FS_ATTRIBUTE_INFORMATION,
1129                                  FileSystemAttributes) +
1130                         sizeof(attr_info.FileSystemAttributes))))
1131         {
1132                 vol_flags = attr_info.FileSystemAttributes;
1133         } else {
1134                 winnt_warning(status, L"\"%ls\": Can't get volume attributes",
1135                               printable_path(full_path));
1136                 vol_flags = 0;
1137         }
1138
1139         /* Get volume ID.  */
1140         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1141                                                       &vol_info,
1142                                                       sizeof(vol_info),
1143                                                       FileFsVolumeInformation);
1144         if (likely((NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW) &&
1145                    (iosb.Information >=
1146                         offsetof(FILE_FS_VOLUME_INFORMATION,
1147                                  VolumeSerialNumber) +
1148                         sizeof(vol_info.VolumeSerialNumber))))
1149         {
1150                 params->capture_root_dev = vol_info.VolumeSerialNumber;
1151         } else {
1152                 winnt_warning(status, L"\"%ls\": Can't get volume ID",
1153                               printable_path(full_path));
1154                 params->capture_root_dev = 0;
1155         }
1156         return vol_flags;
1157 }
1158
1159 struct file_info {
1160         u32 attributes;
1161         u32 num_links;
1162         u64 creation_time;
1163         u64 last_write_time;
1164         u64 last_access_time;
1165         u64 ino;
1166         u64 end_of_file;
1167 };
1168
1169 static noinline_for_stack NTSTATUS
1170 get_file_info(HANDLE h, struct file_info *info)
1171 {
1172         IO_STATUS_BLOCK iosb;
1173         NTSTATUS status;
1174         FILE_ALL_INFORMATION all_info;
1175
1176         status = (*func_NtQueryInformationFile)(h, &iosb, &all_info,
1177                                                 sizeof(all_info),
1178                                                 FileAllInformation);
1179
1180         if (unlikely(!NT_SUCCESS(status) && status != STATUS_BUFFER_OVERFLOW))
1181                 return status;
1182
1183         info->attributes = all_info.BasicInformation.FileAttributes;
1184         info->num_links = all_info.StandardInformation.NumberOfLinks;
1185         info->creation_time = all_info.BasicInformation.CreationTime.QuadPart;
1186         info->last_write_time = all_info.BasicInformation.LastWriteTime.QuadPart;
1187         info->last_access_time = all_info.BasicInformation.LastAccessTime.QuadPart;
1188         info->ino = all_info.InternalInformation.IndexNumber.QuadPart;
1189         info->end_of_file = all_info.StandardInformation.EndOfFile.QuadPart;
1190         return STATUS_SUCCESS;
1191 }
1192
1193 static int
1194 winnt_build_dentry_tree_recursive(struct wim_dentry **root_ret,
1195                                   HANDLE cur_dir,
1196                                   wchar_t *full_path,
1197                                   size_t full_path_nchars,
1198                                   const wchar_t *filename,
1199                                   size_t filename_nchars,
1200                                   struct capture_params *params,
1201                                   struct winnt_scan_stats *stats,
1202                                   u32 vol_flags)
1203 {
1204         struct wim_dentry *root = NULL;
1205         struct wim_inode *inode = NULL;
1206         HANDLE h = NULL;
1207         int ret;
1208         NTSTATUS status;
1209         struct file_info file_info;
1210         ACCESS_MASK requestedPerms;
1211         u64 sort_key;
1212
1213         ret = try_exclude(full_path, params);
1214         if (unlikely(ret < 0)) /* Excluded? */
1215                 goto out_progress;
1216         if (unlikely(ret > 0)) /* Error? */
1217                 goto out;
1218
1219         /* Open the file.  */
1220         requestedPerms = FILE_READ_DATA |
1221                          FILE_READ_ATTRIBUTES |
1222                          READ_CONTROL |
1223                          ACCESS_SYSTEM_SECURITY |
1224                          SYNCHRONIZE;
1225 retry_open:
1226         status = winnt_openat(cur_dir,
1227                               (cur_dir ? filename : full_path),
1228                               (cur_dir ? filename_nchars : full_path_nchars),
1229                               requestedPerms,
1230                               &h);
1231         if (unlikely(!NT_SUCCESS(status))) {
1232                 if (status == STATUS_DELETE_PENDING) {
1233                         WARNING("\"%ls\": Deletion pending; skipping file",
1234                                 printable_path(full_path));
1235                         ret = 0;
1236                         goto out;
1237                 }
1238                 if (status == STATUS_ACCESS_DENIED &&
1239                     (requestedPerms & FILE_READ_DATA)) {
1240                         /* This happens on encrypted files.  */
1241                         requestedPerms &= ~FILE_READ_DATA;
1242                         goto retry_open;
1243                 }
1244
1245                 winnt_error(status, L"\"%ls\": Can't open file",
1246                             printable_path(full_path));
1247                 if (status == STATUS_FVE_LOCKED_VOLUME)
1248                         ret = WIMLIB_ERR_FVE_LOCKED_VOLUME;
1249                 else
1250                         ret = WIMLIB_ERR_OPEN;
1251                 goto out;
1252         }
1253
1254         /* Get information about the file.  */
1255         status = get_file_info(h, &file_info);
1256         if (!NT_SUCCESS(status)) {
1257                 winnt_error(status, L"\"%ls\": Can't get file information",
1258                             printable_path(full_path));
1259                 ret = WIMLIB_ERR_STAT;
1260                 goto out;
1261         }
1262
1263         if (unlikely(!(requestedPerms & FILE_READ_DATA)) &&
1264             !(file_info.attributes & FILE_ATTRIBUTE_ENCRYPTED))
1265         {
1266                 ERROR("\"%ls\": Permission to read data was denied",
1267                       printable_path(full_path));
1268                 ret = WIMLIB_ERR_OPEN;
1269                 goto out;
1270         }
1271
1272         if (unlikely(!cur_dir)) {
1273                 /* Root of tree being captured; get volume information.  */
1274                 vol_flags = get_volume_information(h, full_path, params);
1275                 params->capture_root_ino = file_info.ino;
1276         }
1277
1278
1279         /* Create a WIM dentry with an associated inode, which may be shared.
1280          *
1281          * However, we need to explicitly check for directories and files with
1282          * only 1 link and refuse to hard link them.  This is because Windows
1283          * has a bug where it can return duplicate File IDs for files and
1284          * directories on the FAT filesystem.
1285          *
1286          * Since we don't follow mount points on Windows, we don't need to query
1287          * the volume ID per-file.  Just once, for the root, is enough.  But we
1288          * can't simply pass 0, because then there could be inode collisions
1289          * among multiple calls to win32_build_dentry_tree() that are scanning
1290          * files on different volumes.  */
1291         ret = inode_table_new_dentry(params->inode_table,
1292                                      filename,
1293                                      file_info.ino,
1294                                      params->capture_root_dev,
1295                                      (file_info.num_links <= 1 ||
1296                                         (file_info.attributes & FILE_ATTRIBUTE_DIRECTORY)),
1297                                      &root);
1298         if (ret)
1299                 goto out;
1300
1301         /* Get the short (DOS) name of the file.  */
1302         status = winnt_get_short_name(h, root);
1303
1304         /* If we can't read the short filename for any reason other than
1305          * out-of-memory, just ignore the error and assume the file has no short
1306          * name.  This shouldn't be an issue, since the short names are
1307          * essentially obsolete anyway.  */
1308         if (unlikely(status == STATUS_NO_MEMORY)) {
1309                 ret = WIMLIB_ERR_NOMEM;
1310                 goto out;
1311         }
1312
1313         inode = root->d_inode;
1314
1315         if (inode->i_nlink > 1) {
1316                 /* Shared inode (hard link); skip reading per-inode information.
1317                  */
1318                 goto out_progress;
1319         }
1320
1321         inode->i_attributes = file_info.attributes;
1322         inode->i_creation_time = file_info.creation_time;
1323         inode->i_last_write_time = file_info.last_write_time;
1324         inode->i_last_access_time = file_info.last_access_time;
1325
1326         /* Get the file's security descriptor, unless we are capturing in
1327          * NO_ACLS mode or the volume does not support security descriptors.  */
1328         if (!(params->add_flags & WIMLIB_ADD_FLAG_NO_ACLS)
1329             && (vol_flags & FILE_PERSISTENT_ACLS))
1330         {
1331                 status = winnt_get_security_descriptor(h, inode,
1332                                                        params->sd_set, stats,
1333                                                        params->add_flags);
1334                 if (!NT_SUCCESS(status)) {
1335                         winnt_error(status,
1336                                     L"\"%ls\": Can't read security descriptor",
1337                                     printable_path(full_path));
1338                         ret = WIMLIB_ERR_STAT;
1339                         goto out;
1340                 }
1341         }
1342
1343         /* If this is a reparse point, load the reparse data.  */
1344         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)) {
1345                 ret = winnt_load_reparse_data(h, inode, full_path, params);
1346                 if (ret)
1347                         goto out;
1348         }
1349
1350         sort_key = get_sort_key(h);
1351
1352         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED)) {
1353                 /* Load information about the raw encrypted data.  This is
1354                  * needed for any directory or non-directory that has
1355                  * FILE_ATTRIBUTE_ENCRYPTED set.
1356                  *
1357                  * Note: since OpenEncryptedFileRaw() fails with
1358                  * ERROR_SHARING_VIOLATION if there are any open handles to the
1359                  * file, we have to close the file and re-open it later if
1360                  * needed.  */
1361                 (*func_NtClose)(h);
1362                 h = NULL;
1363                 ret = winnt_scan_efsrpc_raw_data(inode, full_path,
1364                                                  params->unhashed_blobs);
1365                 if (ret)
1366                         goto out;
1367         } else {
1368                 /*
1369                  * Load information about data streams (unnamed and named).
1370                  *
1371                  * Skip this step for encrypted files, since the data from
1372                  * ReadEncryptedFileRaw() already contains all data streams (and
1373                  * they do in fact all get restored by WriteEncryptedFileRaw().)
1374                  *
1375                  * Note: WIMGAPI (as of Windows 8.1) gets wrong and stores both
1376                  * the EFSRPC data and the named data stream(s)...!
1377                  */
1378                 ret = winnt_scan_data_streams(h,
1379                                               full_path,
1380                                               full_path_nchars,
1381                                               inode,
1382                                               params->unhashed_blobs,
1383                                               file_info.end_of_file,
1384                                               vol_flags);
1385                 if (ret)
1386                         goto out;
1387         }
1388
1389         set_sort_key(inode, sort_key);
1390
1391         if (inode_is_directory(inode)) {
1392
1393                 /* Directory: recurse to children.  */
1394
1395                 if (unlikely(!h)) {
1396                         /* Re-open handle that was closed to read raw encrypted
1397                          * data.  */
1398                         status = winnt_openat(cur_dir,
1399                                               (cur_dir ?
1400                                                filename : full_path),
1401                                               (cur_dir ?
1402                                                filename_nchars : full_path_nchars),
1403                                               FILE_LIST_DIRECTORY | SYNCHRONIZE,
1404                                               &h);
1405                         if (!NT_SUCCESS(status)) {
1406                                 winnt_error(status,
1407                                             L"\"%ls\": Can't re-open file",
1408                                             printable_path(full_path));
1409                                 ret = WIMLIB_ERR_OPEN;
1410                                 goto out;
1411                         }
1412                 }
1413                 ret = winnt_recurse_directory(h,
1414                                               full_path,
1415                                               full_path_nchars,
1416                                               root,
1417                                               params,
1418                                               stats,
1419                                               vol_flags);
1420                 if (ret)
1421                         goto out;
1422         }
1423
1424 out_progress:
1425         params->progress.scan.cur_path = printable_path(full_path);
1426         if (likely(root))
1427                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
1428         else
1429                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
1430 out:
1431         if (likely(h))
1432                 (*func_NtClose)(h);
1433         if (unlikely(ret)) {
1434                 free_dentry_tree(root, params->blob_table);
1435                 root = NULL;
1436                 ret = report_capture_error(params, ret, full_path);
1437         }
1438         *root_ret = root;
1439         return ret;
1440 }
1441
1442 static void
1443 winnt_do_scan_warnings(const wchar_t *path, const struct winnt_scan_stats *stats)
1444 {
1445         if (likely(stats->num_get_sacl_priv_notheld == 0 &&
1446                    stats->num_get_sd_access_denied == 0))
1447                 return;
1448
1449         WARNING("Scan of \"%ls\" complete, but with one or more warnings:", path);
1450         if (stats->num_get_sacl_priv_notheld != 0) {
1451                 WARNING("- Could not capture SACL (System Access Control List)\n"
1452                         "            on %lu files or directories.",
1453                         stats->num_get_sacl_priv_notheld);
1454         }
1455         if (stats->num_get_sd_access_denied != 0) {
1456                 WARNING("- Could not capture security descriptor at all\n"
1457                         "            on %lu files or directories.",
1458                         stats->num_get_sd_access_denied);
1459         }
1460         WARNING("To fully capture all security descriptors, run the program\n"
1461                 "          with Administrator rights.");
1462 }
1463
1464 #define WINDOWS_NT_MAX_PATH 32768
1465
1466 /* Win32 version of capturing a directory tree.  */
1467 int
1468 win32_build_dentry_tree(struct wim_dentry **root_ret,
1469                         const wchar_t *root_disk_path,
1470                         struct capture_params *params)
1471 {
1472         wchar_t *path;
1473         int ret;
1474         UNICODE_STRING ntpath;
1475         struct winnt_scan_stats stats;
1476         size_t ntpath_nchars;
1477
1478         /* WARNING: There is no check for overflow later when this buffer is
1479          * being used!  But it's as long as the maximum path length understood
1480          * by Windows NT (which is NOT the same as MAX_PATH).  */
1481         path = MALLOC((WINDOWS_NT_MAX_PATH + 1) * sizeof(wchar_t));
1482         if (!path)
1483                 return WIMLIB_ERR_NOMEM;
1484
1485         ret = win32_path_to_nt_path(root_disk_path, &ntpath);
1486         if (ret)
1487                 goto out_free_path;
1488
1489         if (ntpath.Length < 4 * sizeof(wchar_t) ||
1490             ntpath.Length > WINDOWS_NT_MAX_PATH * sizeof(wchar_t) ||
1491             wmemcmp(ntpath.Buffer, L"\\??\\", 4))
1492         {
1493                 ERROR("\"%ls\": unrecognized path format", root_disk_path);
1494                 ret = WIMLIB_ERR_INVALID_PARAM;
1495         } else {
1496                 ntpath_nchars = ntpath.Length / sizeof(wchar_t);
1497                 wmemcpy(path, ntpath.Buffer, ntpath_nchars);
1498                 path[ntpath_nchars] = L'\0';
1499
1500                 params->capture_root_nchars = ntpath_nchars;
1501                 if (path[ntpath_nchars - 1] == L'\\')
1502                         params->capture_root_nchars--;
1503                 ret = 0;
1504         }
1505         HeapFree(GetProcessHeap(), 0, ntpath.Buffer);
1506         if (ret)
1507                 goto out_free_path;
1508
1509         memset(&stats, 0, sizeof(stats));
1510
1511         ret = winnt_build_dentry_tree_recursive(root_ret, NULL,
1512                                                 path, ntpath_nchars,
1513                                                 L"", 0, params, &stats, 0);
1514 out_free_path:
1515         FREE(path);
1516         if (ret == 0)
1517                 winnt_do_scan_warnings(root_disk_path, &stats);
1518         return ret;
1519 }
1520
1521 #endif /* __WIN32__ */