1 /* hivex - Windows Registry "hive" extraction library.
2 * Copyright (C) 2009-2010 Red Hat Inc.
3 * Derived from code by Petter Nordahl-Hagen under a compatible license:
4 * Copyright (c) 1997-2007 Petter Nordahl-Hagen.
5 * Derived from code by Markus Stephany under a compatible license:
6 * Copyright (c) 2000-2004, Markus Stephany.
8 * This library is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation;
11 * version 2.1 of the License.
13 * This library is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
18 * See file LICENSE for the full license.
38 #include "full-read.h"
39 #include "full-write.h"
45 #define STREQ(a,b) (strcmp((a),(b)) == 0)
46 #define STRCASEEQ(a,b) (strcasecmp((a),(b)) == 0)
47 //#define STRNEQ(a,b) (strcmp((a),(b)) != 0)
48 //#define STRCASENEQ(a,b) (strcasecmp((a),(b)) != 0)
49 #define STREQLEN(a,b,n) (strncmp((a),(b),(n)) == 0)
50 //#define STRCASEEQLEN(a,b,n) (strncasecmp((a),(b),(n)) == 0)
51 //#define STRNEQLEN(a,b,n) (strncmp((a),(b),(n)) != 0)
52 //#define STRCASENEQLEN(a,b,n) (strncasecmp((a),(b),(n)) != 0)
53 #define STRPREFIX(a,b) (strncmp((a),(b),strlen((b))) == 0)
56 #include "byte_conversions.h"
58 static char *windows_utf16_to_utf8 (/* const */ char *input, size_t len);
67 /* Registry file, memory mapped if read-only, or malloc'd if writing. */
70 struct ntreg_header *hdr;
73 /* Use a bitmap to store which file offsets are valid (point to a
74 * used block). We only need to store 1 bit per 32 bits of the file
75 * (because blocks are 4-byte aligned). We found that the average
76 * block size in a registry file is ~50 bytes. So roughly 1 in 12
77 * bits in the bitmap will be set, making it likely a more efficient
78 * structure than a hash table.
81 #define BITMAP_SET(bitmap,off) (bitmap[(off)>>5] |= 1 << (((off)>>2)&7))
82 #define BITMAP_CLR(bitmap,off) (bitmap[(off)>>5] &= ~ (1 << (((off)>>2)&7)))
83 #define BITMAP_TST(bitmap,off) (bitmap[(off)>>5] & (1 << (((off)>>2)&7)))
84 #define IS_VALID_BLOCK(h,off) \
85 (((off) & 3) == 0 && \
87 (off) < (h)->size && \
88 BITMAP_TST((h)->bitmap,(off)))
90 /* Fields from the header, extracted from little-endianness hell. */
91 size_t rootoffs; /* Root key offset (always an nk-block). */
92 size_t endpages; /* Offset of end of pages. */
95 size_t endblocks; /* Offset to next block allocation (0
96 if not allocated anything yet). */
99 /* NB. All fields are little endian. */
100 struct ntreg_header {
101 char magic[4]; /* "regf" */
104 char last_modified[8];
105 uint32_t major_ver; /* 1 */
106 uint32_t minor_ver; /* 3 */
107 uint32_t unknown5; /* 0 */
108 uint32_t unknown6; /* 1 */
109 uint32_t offset; /* offset of root key record - 4KB */
110 uint32_t blocks; /* pointer AFTER last hbin in file - 4KB */
111 uint32_t unknown7; /* 1 */
113 char name[64]; /* original file name of hive */
114 char unknown_guid1[16];
115 char unknown_guid2[16];
118 char unknown_guid3[16];
123 uint32_t csum; /* checksum: xor of dwords 0-0x1fb. */
125 char unknown11[3528];
127 char unknown_guid4[16];
128 char unknown_guid5[16];
129 char unknown_guid6[16];
133 } __attribute__((__packed__));
135 struct ntreg_hbin_page {
136 char magic[4]; /* "hbin" */
137 uint32_t offset_first; /* offset from 1st block */
138 uint32_t page_size; /* size of this page (multiple of 4KB) */
140 /* Linked list of blocks follows here. */
141 } __attribute__((__packed__));
143 struct ntreg_hbin_block {
144 int32_t seg_len; /* length of this block (-ve for used block) */
145 char id[2]; /* the block type (eg. "nk" for nk record) */
146 /* Block data follows here. */
147 } __attribute__((__packed__));
149 #define BLOCK_ID_EQ(h,offs,eqid) \
150 (STREQLEN (((struct ntreg_hbin_block *)((h)->addr + (offs)))->id, (eqid), 2))
153 block_len (hive_h *h, size_t blkoff, int *used)
155 struct ntreg_hbin_block *block;
156 block = (struct ntreg_hbin_block *) (h->addr + blkoff);
158 int32_t len = le32toh (block->seg_len);
169 struct ntreg_nk_record {
170 int32_t seg_len; /* length (always -ve because used) */
171 char id[2]; /* "nk" */
175 uint32_t parent; /* offset of owner/parent */
176 uint32_t nr_subkeys; /* number of subkeys */
177 uint32_t nr_subkeys_volatile;
178 uint32_t subkey_lf; /* lf record containing list of subkeys */
179 uint32_t subkey_lf_volatile;
180 uint32_t nr_values; /* number of values */
181 uint32_t vallist; /* value-list record */
182 uint32_t sk; /* offset of sk-record */
183 uint32_t classname; /* offset of classname record */
184 uint16_t max_subkey_name_len; /* maximum length of a subkey name in bytes
185 if the subkey was reencoded as UTF-16LE */
188 uint32_t max_vk_name_len; /* maximum length of any vk name in bytes
189 if the name was reencoded as UTF-16LE */
190 uint32_t max_vk_data_len; /* maximum length of any vk data in bytes */
192 uint16_t name_len; /* length of name */
193 uint16_t classname_len; /* length of classname */
194 char name[1]; /* name follows here */
195 } __attribute__((__packed__));
197 struct ntreg_lf_record {
199 char id[2]; /* "lf"|"lh" */
200 uint16_t nr_keys; /* number of keys in this record */
202 uint32_t offset; /* offset of nk-record for this subkey */
203 char hash[4]; /* hash of subkey name */
205 } __attribute__((__packed__));
207 struct ntreg_ri_record {
209 char id[2]; /* "ri" */
210 uint16_t nr_offsets; /* number of pointers to lh records */
211 uint32_t offset[1]; /* list of pointers to lh records */
212 } __attribute__((__packed__));
214 /* This has no ID header. */
215 struct ntreg_value_list {
217 uint32_t offset[1]; /* list of pointers to vk records */
218 } __attribute__((__packed__));
220 struct ntreg_vk_record {
221 int32_t seg_len; /* length (always -ve because used) */
222 char id[2]; /* "vk" */
223 uint16_t name_len; /* length of name */
224 /* length of the data:
225 * If data_len is <= 4, then it's stored inline.
226 * If data_len is 0x80000000, then it's an inline dword.
227 * Top bit may be set or not set at random.
230 uint32_t data_offset; /* pointer to the data (or data if inline) */
231 uint32_t data_type; /* type of the data */
232 uint16_t flags; /* bit 0 set => key name ASCII,
233 bit 0 clr => key name UTF-16.
234 Only seen ASCII here in the wild.
235 NB: this is CLEAR for default key. */
237 char name[1]; /* key name follows here */
238 } __attribute__((__packed__));
240 struct ntreg_sk_record {
241 int32_t seg_len; /* length (always -ve because used) */
242 char id[2]; /* "sk" */
244 uint32_t sk_next; /* linked into a circular list */
246 uint32_t refcount; /* reference count */
247 uint32_t sec_len; /* length of security info */
248 char sec_desc[1]; /* security info follows */
249 } __attribute__((__packed__));
252 header_checksum (const hive_h *h)
254 uint32_t *daddr = (uint32_t *) h->addr;
258 for (i = 0; i < 0x1fc / 4; ++i) {
259 sum ^= le32toh (*daddr);
267 hivex_open (const char *filename, int flags)
271 assert (sizeof (struct ntreg_header) == 0x1000);
272 assert (offsetof (struct ntreg_header, csum) == 0x1fc);
274 h = calloc (1, sizeof *h);
278 h->msglvl = flags & HIVEX_OPEN_MSGLVL_MASK;
280 const char *debug = getenv ("HIVEX_DEBUG");
281 if (debug && STREQ (debug, "1"))
285 fprintf (stderr, "hivex_open: created handle %p\n", h);
287 h->writable = !!(flags & HIVEX_OPEN_WRITE);
288 h->filename = strdup (filename);
289 if (h->filename == NULL)
292 h->fd = open (filename, O_RDONLY | O_CLOEXEC);
297 if (fstat (h->fd, &statbuf) == -1)
300 h->size = statbuf.st_size;
303 h->addr = mmap (NULL, h->size, PROT_READ, MAP_SHARED, h->fd, 0);
304 if (h->addr == MAP_FAILED)
308 fprintf (stderr, "hivex_open: mapped file at %p\n", h->addr);
310 h->addr = malloc (h->size);
314 if (full_read (h->fd, h->addr, h->size) < h->size)
319 if (h->hdr->magic[0] != 'r' ||
320 h->hdr->magic[1] != 'e' ||
321 h->hdr->magic[2] != 'g' ||
322 h->hdr->magic[3] != 'f') {
323 fprintf (stderr, "hivex: %s: not a Windows NT Registry hive file\n",
329 /* Check major version. */
330 uint32_t major_ver = le32toh (h->hdr->major_ver);
331 if (major_ver != 1) {
333 "hivex: %s: hive file major version %" PRIu32 " (expected 1)\n",
334 filename, major_ver);
339 h->bitmap = calloc (1 + h->size / 32, 1);
340 if (h->bitmap == NULL)
343 /* Header checksum. */
344 uint32_t sum = header_checksum (h);
345 if (sum != le32toh (h->hdr->csum)) {
346 fprintf (stderr, "hivex: %s: bad checksum in hive header\n", filename);
351 if (h->msglvl >= 2) {
352 char *name = windows_utf16_to_utf8 (h->hdr->name, 64);
355 "hivex_open: header fields:\n"
356 " file version %" PRIu32 ".%" PRIu32 "\n"
357 " sequence nos %" PRIu32 " %" PRIu32 "\n"
358 " (sequences nos should match if hive was synched at shutdown)\n"
359 " original file name %s\n"
360 " (only 32 chars are stored, name is probably truncated)\n"
361 " root offset 0x%x + 0x1000\n"
362 " end of last page 0x%x + 0x1000 (total file size 0x%zx)\n"
363 " checksum 0x%x (calculated 0x%x)\n",
364 major_ver, le32toh (h->hdr->minor_ver),
365 le32toh (h->hdr->sequence1), le32toh (h->hdr->sequence2),
366 name ? name : "(conversion failed)",
367 le32toh (h->hdr->offset),
368 le32toh (h->hdr->blocks), h->size,
369 le32toh (h->hdr->csum), sum);
373 h->rootoffs = le32toh (h->hdr->offset) + 0x1000;
374 h->endpages = le32toh (h->hdr->blocks) + 0x1000;
377 fprintf (stderr, "hivex_open: root offset = 0x%zx\n", h->rootoffs);
379 /* We'll set this flag when we see a block with the root offset (ie.
382 int seen_root_block = 0, bad_root_block = 0;
384 /* Collect some stats. */
385 size_t pages = 0; /* Number of hbin pages read. */
386 size_t smallest_page = SIZE_MAX, largest_page = 0;
387 size_t blocks = 0; /* Total number of blocks found. */
388 size_t smallest_block = SIZE_MAX, largest_block = 0, blocks_bytes = 0;
389 size_t used_blocks = 0; /* Total number of used blocks found. */
390 size_t used_size = 0; /* Total size (bytes) of used blocks. */
392 /* Read the pages and blocks. The aim here is to be robust against
393 * corrupt or malicious registries. So we make sure the loops
394 * always make forward progress. We add the address of each block
395 * we read to a hash table so pointers will only reference the start
399 struct ntreg_hbin_page *page;
400 for (off = 0x1000; off < h->size; off += le32toh (page->page_size)) {
401 if (off >= h->endpages)
404 page = (struct ntreg_hbin_page *) (h->addr + off);
405 if (page->magic[0] != 'h' ||
406 page->magic[1] != 'b' ||
407 page->magic[2] != 'i' ||
408 page->magic[3] != 'n') {
409 fprintf (stderr, "hivex: %s: trailing garbage at end of file (at 0x%zx, after %zu pages)\n",
410 filename, off, pages);
415 size_t page_size = le32toh (page->page_size);
417 fprintf (stderr, "hivex_open: page at 0x%zx, size %zu\n", off, page_size);
419 if (page_size < smallest_page) smallest_page = page_size;
420 if (page_size > largest_page) largest_page = page_size;
422 if (page_size <= sizeof (struct ntreg_hbin_page) ||
423 (page_size & 0x0fff) != 0) {
424 fprintf (stderr, "hivex: %s: page size %zu at 0x%zx, bad registry\n",
425 filename, page_size, off);
430 /* Read the blocks in this page. */
432 struct ntreg_hbin_block *block;
434 for (blkoff = off + 0x20;
435 blkoff < off + page_size;
439 int is_root = blkoff == h->rootoffs;
443 block = (struct ntreg_hbin_block *) (h->addr + blkoff);
445 seg_len = block_len (h, blkoff, &used);
446 if (seg_len <= 4 || (seg_len & 3) != 0) {
447 fprintf (stderr, "hivex: %s: block size %" PRIu32 " at 0x%zx, bad registry\n",
448 filename, le32toh (block->seg_len), blkoff);
454 fprintf (stderr, "hivex_open: %s block id %d,%d at 0x%zx size %zu%s\n",
455 used ? "used" : "free", block->id[0], block->id[1], blkoff,
456 seg_len, is_root ? " (root)" : "");
458 blocks_bytes += seg_len;
459 if (seg_len < smallest_block) smallest_block = seg_len;
460 if (seg_len > largest_block) largest_block = seg_len;
462 if (is_root && !used)
467 used_size += seg_len;
469 /* Root block must be an nk-block. */
470 if (is_root && (block->id[0] != 'n' || block->id[1] != 'k'))
473 /* Note this blkoff is a valid address. */
474 BITMAP_SET (h->bitmap, blkoff);
479 if (!seen_root_block) {
480 fprintf (stderr, "hivex: %s: no root block found\n", filename);
485 if (bad_root_block) {
486 fprintf (stderr, "hivex: %s: bad root block (free or not nk)\n", filename);
493 "hivex_open: successfully read Windows Registry hive file:\n"
494 " pages: %zu [sml: %zu, lge: %zu]\n"
495 " blocks: %zu [sml: %zu, avg: %zu, lge: %zu]\n"
496 " blocks used: %zu\n"
497 " bytes used: %zu\n",
498 pages, smallest_page, largest_page,
499 blocks, smallest_block, blocks_bytes / blocks, largest_block,
500 used_blocks, used_size);
508 if (h->addr && h->size && h->addr != MAP_FAILED) {
510 munmap (h->addr, h->size);
524 hivex_close (hive_h *h)
530 munmap (h->addr, h->size);
540 /*----------------------------------------------------------------------
545 hivex_root (hive_h *h)
547 hive_node_h ret = h->rootoffs;
548 if (!IS_VALID_BLOCK (h, ret)) {
556 hivex_node_name (hive_h *h, hive_node_h node)
558 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
563 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
565 /* AFAIK the node name is always plain ASCII, so no conversion
566 * to UTF-8 is necessary. However we do need to nul-terminate
570 /* nk->name_len is unsigned, 16 bit, so this is safe ... However
571 * we have to make sure the length doesn't exceed the block length.
573 size_t len = le16toh (nk->name_len);
574 size_t seg_len = block_len (h, node, NULL);
575 if (sizeof (struct ntreg_nk_record) + len - 1 > seg_len) {
577 fprintf (stderr, "hivex_node_name: returning EFAULT because node name is too long (%zu, %zu)\n",
583 char *ret = malloc (len + 1);
586 memcpy (ret, nk->name, len);
592 /* I think the documentation for the sk and classname fields in the nk
593 * record is wrong, or else the offset field is in the wrong place.
594 * Otherwise this makes no sense. Disabled this for now -- it's not
595 * useful for reading the registry anyway.
599 hivex_node_security (hive_h *h, hive_node_h node)
601 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
606 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
608 hive_node_h ret = le32toh (nk->sk);
610 if (!IS_VALID_BLOCK (h, ret)) {
618 hivex_node_classname (hive_h *h, hive_node_h node)
620 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
625 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
627 hive_node_h ret = le32toh (nk->classname);
629 if (!IS_VALID_BLOCK (h, ret)) {
637 /* Structure for returning 0-terminated lists of offsets (nodes,
647 init_offset_list (struct offset_list *list)
651 list->offsets = NULL;
654 #define INIT_OFFSET_LIST(name) \
655 struct offset_list name; \
656 init_offset_list (&name)
658 /* Preallocates the offset_list, but doesn't make the contents longer. */
660 grow_offset_list (struct offset_list *list, size_t alloc)
662 assert (alloc >= list->len);
663 size_t *p = realloc (list->offsets, alloc * sizeof (size_t));
672 add_to_offset_list (struct offset_list *list, size_t offset)
674 if (list->len >= list->alloc) {
675 if (grow_offset_list (list, list->alloc ? list->alloc * 2 : 4) == -1)
678 list->offsets[list->len] = offset;
684 free_offset_list (struct offset_list *list)
686 free (list->offsets);
690 return_offset_list (struct offset_list *list)
692 if (add_to_offset_list (list, 0) == -1)
694 return list->offsets; /* caller frees */
697 /* Iterate over children, returning child nodes and intermediate blocks. */
698 #define GET_CHILDREN_NO_CHECK_NK 1
701 get_children (hive_h *h, hive_node_h node,
702 hive_node_h **children_ret, size_t **blocks_ret,
705 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
710 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
712 size_t nr_subkeys_in_nk = le32toh (nk->nr_subkeys);
714 INIT_OFFSET_LIST (children);
715 INIT_OFFSET_LIST (blocks);
717 /* Deal with the common "no subkeys" case quickly. */
718 if (nr_subkeys_in_nk == 0)
721 /* Arbitrarily limit the number of subkeys we will ever deal with. */
722 if (nr_subkeys_in_nk > 1000000) {
727 /* Preallocate space for the children. */
728 if (grow_offset_list (&children, nr_subkeys_in_nk) == -1)
731 /* The subkey_lf field can point either to an lf-record, which is
732 * the common case, or if there are lots of subkeys, to an
735 size_t subkey_lf = le32toh (nk->subkey_lf);
737 if (!IS_VALID_BLOCK (h, subkey_lf)) {
739 fprintf (stderr, "hivex_node_children: returning EFAULT because subkey_lf is not a valid block (0x%zx)\n",
745 if (add_to_offset_list (&blocks, subkey_lf) == -1)
748 struct ntreg_hbin_block *block =
749 (struct ntreg_hbin_block *) (h->addr + subkey_lf);
751 /* Points to lf-record? (Note, also "lh" but that is basically the
752 * same as "lf" as far as we are concerned here).
754 if (block->id[0] == 'l' && (block->id[1] == 'f' || block->id[1] == 'h')) {
755 struct ntreg_lf_record *lf = (struct ntreg_lf_record *) block;
757 /* Check number of subkeys in the nk-record matches number of subkeys
760 size_t nr_subkeys_in_lf = le16toh (lf->nr_keys);
763 fprintf (stderr, "hivex_node_children: nr_subkeys_in_nk = %zu, nr_subkeys_in_lf = %zu\n",
764 nr_subkeys_in_nk, nr_subkeys_in_lf);
766 if (nr_subkeys_in_nk != nr_subkeys_in_lf) {
771 size_t len = block_len (h, subkey_lf, NULL);
772 if (8 + nr_subkeys_in_lf * 8 > len) {
774 fprintf (stderr, "hivex_node_children: returning EFAULT because too many subkeys (%zu, %zu)\n",
775 nr_subkeys_in_lf, len);
781 for (i = 0; i < nr_subkeys_in_lf; ++i) {
782 hive_node_h subkey = le32toh (lf->keys[i].offset);
784 if (!(flags & GET_CHILDREN_NO_CHECK_NK)) {
785 if (!IS_VALID_BLOCK (h, subkey)) {
787 fprintf (stderr, "hivex_node_children: returning EFAULT because subkey is not a valid block (0x%zx)\n",
793 if (add_to_offset_list (&children, subkey) == -1)
798 /* Points to ri-record? */
799 else if (block->id[0] == 'r' && block->id[1] == 'i') {
800 struct ntreg_ri_record *ri = (struct ntreg_ri_record *) block;
802 size_t nr_offsets = le16toh (ri->nr_offsets);
804 /* Count total number of children. */
806 for (i = 0; i < nr_offsets; ++i) {
807 hive_node_h offset = le32toh (ri->offset[i]);
809 if (!IS_VALID_BLOCK (h, offset)) {
811 fprintf (stderr, "hivex_node_children: returning EFAULT because ri-offset is not a valid block (0x%zx)\n",
816 if (!BLOCK_ID_EQ (h, offset, "lf") && !BLOCK_ID_EQ (h, offset, "lh")) {
818 fprintf (stderr, "get_children: returning ENOTSUP because ri-record offset does not point to lf/lh (0x%zx)\n",
824 if (add_to_offset_list (&blocks, offset) == -1)
827 struct ntreg_lf_record *lf =
828 (struct ntreg_lf_record *) (h->addr + offset);
830 count += le16toh (lf->nr_keys);
834 fprintf (stderr, "hivex_node_children: nr_subkeys_in_nk = %zu, counted = %zu\n",
835 nr_subkeys_in_nk, count);
837 if (nr_subkeys_in_nk != count) {
842 /* Copy list of children. Note nr_subkeys_in_nk is limited to
843 * something reasonable above.
845 for (i = 0; i < nr_offsets; ++i) {
846 hive_node_h offset = le32toh (ri->offset[i]);
848 if (!IS_VALID_BLOCK (h, offset)) {
850 fprintf (stderr, "hivex_node_children: returning EFAULT because ri-offset is not a valid block (0x%zx)\n",
855 if (!BLOCK_ID_EQ (h, offset, "lf") && !BLOCK_ID_EQ (h, offset, "lh")) {
857 fprintf (stderr, "get_children: returning ENOTSUP because ri-record offset does not point to lf/lh (0x%zx)\n",
863 struct ntreg_lf_record *lf =
864 (struct ntreg_lf_record *) (h->addr + offset);
867 for (j = 0; j < le16toh (lf->nr_keys); ++j) {
868 hive_node_h subkey = le32toh (lf->keys[j].offset);
870 if (!(flags & GET_CHILDREN_NO_CHECK_NK)) {
871 if (!IS_VALID_BLOCK (h, subkey)) {
873 fprintf (stderr, "hivex_node_children: returning EFAULT because indirect subkey is not a valid block (0x%zx)\n",
879 if (add_to_offset_list (&children, subkey) == -1)
885 /* else not supported, set errno and fall through */
887 fprintf (stderr, "get_children: returning ENOTSUP because subkey block is not lf/lh/ri (0x%zx, %d, %d)\n",
888 subkey_lf, block->id[0], block->id[1]);
891 free_offset_list (&children);
892 free_offset_list (&blocks);
896 *children_ret = return_offset_list (&children);
897 *blocks_ret = return_offset_list (&blocks);
898 if (!*children_ret || !*blocks_ret)
904 hivex_node_children (hive_h *h, hive_node_h node)
906 hive_node_h *children;
909 if (get_children (h, node, &children, &blocks, 0) == -1)
916 /* Very inefficient, but at least having a separate API call
917 * allows us to make it more efficient in future.
920 hivex_node_get_child (hive_h *h, hive_node_h node, const char *nname)
922 hive_node_h *children = NULL;
926 children = hivex_node_children (h, node);
927 if (!children) goto error;
930 for (i = 0; children[i] != 0; ++i) {
931 name = hivex_node_name (h, children[i]);
932 if (!name) goto error;
933 if (STRCASEEQ (name, nname)) {
937 free (name); name = NULL;
947 hivex_node_parent (hive_h *h, hive_node_h node)
949 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
954 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
956 hive_node_h ret = le32toh (nk->parent);
958 if (!IS_VALID_BLOCK (h, ret)) {
960 fprintf (stderr, "hivex_node_parent: returning EFAULT because parent is not a valid block (0x%zx)\n",
969 get_values (hive_h *h, hive_node_h node,
970 hive_value_h **values_ret, size_t **blocks_ret)
972 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
977 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
979 size_t nr_values = le32toh (nk->nr_values);
982 fprintf (stderr, "hivex_node_values: nr_values = %zu\n", nr_values);
984 INIT_OFFSET_LIST (values);
985 INIT_OFFSET_LIST (blocks);
987 /* Deal with the common "no values" case quickly. */
991 /* Arbitrarily limit the number of values we will ever deal with. */
992 if (nr_values > 100000) {
997 /* Preallocate space for the values. */
998 if (grow_offset_list (&values, nr_values) == -1)
1001 /* Get the value list and check it looks reasonable. */
1002 size_t vlist_offset = le32toh (nk->vallist);
1003 vlist_offset += 0x1000;
1004 if (!IS_VALID_BLOCK (h, vlist_offset)) {
1006 fprintf (stderr, "hivex_node_values: returning EFAULT because value list is not a valid block (0x%zx)\n",
1012 if (add_to_offset_list (&blocks, vlist_offset) == -1)
1015 struct ntreg_value_list *vlist =
1016 (struct ntreg_value_list *) (h->addr + vlist_offset);
1018 size_t len = block_len (h, vlist_offset, NULL);
1019 if (4 + nr_values * 4 > len) {
1021 fprintf (stderr, "hivex_node_values: returning EFAULT because value list is too long (%zu, %zu)\n",
1028 for (i = 0; i < nr_values; ++i) {
1029 hive_node_h value = vlist->offset[i];
1031 if (!IS_VALID_BLOCK (h, value)) {
1033 fprintf (stderr, "hivex_node_values: returning EFAULT because value is not a valid block (0x%zx)\n",
1038 if (add_to_offset_list (&values, value) == -1)
1043 *values_ret = return_offset_list (&values);
1044 *blocks_ret = return_offset_list (&blocks);
1045 if (!*values_ret || !*blocks_ret)
1050 free_offset_list (&values);
1051 free_offset_list (&blocks);
1056 hivex_node_values (hive_h *h, hive_node_h node)
1058 hive_value_h *values;
1061 if (get_values (h, node, &values, &blocks) == -1)
1068 /* Very inefficient, but at least having a separate API call
1069 * allows us to make it more efficient in future.
1072 hivex_node_get_value (hive_h *h, hive_node_h node, const char *key)
1074 hive_value_h *values = NULL;
1076 hive_value_h ret = 0;
1078 values = hivex_node_values (h, node);
1079 if (!values) goto error;
1082 for (i = 0; values[i] != 0; ++i) {
1083 name = hivex_value_key (h, values[i]);
1084 if (!name) goto error;
1085 if (STRCASEEQ (name, key)) {
1089 free (name); name = NULL;
1099 hivex_value_key (hive_h *h, hive_value_h value)
1101 if (!IS_VALID_BLOCK (h, value) || !BLOCK_ID_EQ (h, value, "vk")) {
1106 struct ntreg_vk_record *vk = (struct ntreg_vk_record *) (h->addr + value);
1108 /* AFAIK the key is always plain ASCII, so no conversion to UTF-8 is
1109 * necessary. However we do need to nul-terminate the string.
1112 /* vk->name_len is unsigned, 16 bit, so this is safe ... However
1113 * we have to make sure the length doesn't exceed the block length.
1115 size_t len = le16toh (vk->name_len);
1116 size_t seg_len = block_len (h, value, NULL);
1117 if (sizeof (struct ntreg_vk_record) + len - 1 > seg_len) {
1119 fprintf (stderr, "hivex_value_key: returning EFAULT because key length is too long (%zu, %zu)\n",
1125 char *ret = malloc (len + 1);
1128 memcpy (ret, vk->name, len);
1134 hivex_value_type (hive_h *h, hive_value_h value, hive_type *t, size_t *len)
1136 if (!IS_VALID_BLOCK (h, value) || !BLOCK_ID_EQ (h, value, "vk")) {
1141 struct ntreg_vk_record *vk = (struct ntreg_vk_record *) (h->addr + value);
1144 *t = le32toh (vk->data_type);
1147 *len = le32toh (vk->data_len);
1148 if (*len == 0x80000000) { /* special case */
1150 if (t) *t = hive_t_dword;
1159 hivex_value_value (hive_h *h, hive_value_h value,
1160 hive_type *t_rtn, size_t *len_rtn)
1162 if (!IS_VALID_BLOCK (h, value) || !BLOCK_ID_EQ (h, value, "vk")) {
1167 struct ntreg_vk_record *vk = (struct ntreg_vk_record *) (h->addr + value);
1172 t = le32toh (vk->data_type);
1174 len = le32toh (vk->data_len);
1175 if (len == 0x80000000) { /* special case */
1182 fprintf (stderr, "hivex_value_value: value=0x%zx, t=%d, len=%zu\n",
1190 /* Arbitrarily limit the length that we will read. */
1191 if (len > 1000000) {
1196 char *ret = malloc (len);
1200 /* If length is <= 4 it's always stored inline. */
1202 memcpy (ret, (char *) &vk->data_offset, len);
1206 size_t data_offset = le32toh (vk->data_offset);
1207 data_offset += 0x1000;
1208 if (!IS_VALID_BLOCK (h, data_offset)) {
1210 fprintf (stderr, "hivex_value_value: returning EFAULT because data offset is not a valid block (0x%zx)\n",
1217 /* Check that the declared size isn't larger than the block its in.
1219 * XXX Some apparently valid registries are seen to have this,
1220 * so turn this into a warning and substitute the smaller length
1223 size_t blen = block_len (h, data_offset, NULL);
1224 if (len > blen - 4 /* subtract 4 for block header */) {
1226 fprintf (stderr, "hivex_value_value: warning: declared data length is longer than the block it is in (data 0x%zx, data len %zu, block len %zu)\n",
1227 data_offset, len, blen);
1231 char *data = h->addr + data_offset + 4;
1232 memcpy (ret, data, len);
1237 windows_utf16_to_utf8 (/* const */ char *input, size_t len)
1239 iconv_t ic = iconv_open ("UTF-8", "UTF-16");
1240 if (ic == (iconv_t) -1)
1243 /* iconv(3) has an insane interface ... */
1245 /* Mostly UTF-8 will be smaller, so this is a good initial guess. */
1246 size_t outalloc = len;
1250 size_t outlen = outalloc;
1251 char *out = malloc (outlen + 1);
1261 size_t r = iconv (ic, &inp, &inlen, &outp, &outlen);
1262 if (r == (size_t) -1) {
1263 if (errno == E2BIG) {
1264 size_t prev = outalloc;
1265 /* Try again with a larger output buffer. */
1268 if (outalloc < prev)
1273 /* Else some conversion failure, eg. EILSEQ, EINVAL. */
1289 hivex_value_string (hive_h *h, hive_value_h value)
1293 char *data = hivex_value_value (h, value, &t, &len);
1298 if (t != hive_t_string && t != hive_t_expand_string && t != hive_t_link) {
1304 char *ret = windows_utf16_to_utf8 (data, len);
1313 free_strings (char **argv)
1318 for (i = 0; argv[i] != NULL; ++i)
1324 /* Get the length of a UTF-16 format string. Handle the string as
1325 * pairs of bytes, looking for the first \0\0 pair.
1328 utf16_string_len_in_bytes (const char *str)
1332 while (str[0] || str[1]) {
1340 /* http://blogs.msdn.com/oldnewthing/archive/2009/10/08/9904646.aspx */
1342 hivex_value_multiple_strings (hive_h *h, hive_value_h value)
1346 char *data = hivex_value_value (h, value, &t, &len);
1351 if (t != hive_t_multiple_strings) {
1357 size_t nr_strings = 0;
1358 char **ret = malloc ((1 + nr_strings) * sizeof (char *));
1368 while (p < data + len && (plen = utf16_string_len_in_bytes (p)) > 0) {
1370 char **ret2 = realloc (ret, (1 + nr_strings) * sizeof (char *));
1378 ret[nr_strings-1] = windows_utf16_to_utf8 (p, plen);
1379 ret[nr_strings] = NULL;
1380 if (ret[nr_strings-1] == NULL) {
1386 p += plen + 2 /* skip over UTF-16 \0\0 at the end of this string */;
1394 hivex_value_dword (hive_h *h, hive_value_h value)
1398 char *data = hivex_value_value (h, value, &t, &len);
1403 if ((t != hive_t_dword && t != hive_t_dword_be) || len != 4) {
1409 int32_t ret = *(int32_t*)data;
1411 if (t == hive_t_dword) /* little endian */
1412 ret = le32toh (ret);
1414 ret = be32toh (ret);
1420 hivex_value_qword (hive_h *h, hive_value_h value)
1424 char *data = hivex_value_value (h, value, &t, &len);
1429 if (t != hive_t_qword || len != 8) {
1435 int64_t ret = *(int64_t*)data;
1437 ret = le64toh (ret); /* always little endian */
1442 /*----------------------------------------------------------------------
1447 hivex_visit (hive_h *h, const struct hivex_visitor *visitor, size_t len,
1448 void *opaque, int flags)
1450 return hivex_visit_node (h, hivex_root (h), visitor, len, opaque, flags);
1453 static int hivex__visit_node (hive_h *h, hive_node_h node, const struct hivex_visitor *vtor, char *unvisited, void *opaque, int flags);
1456 hivex_visit_node (hive_h *h, hive_node_h node,
1457 const struct hivex_visitor *visitor, size_t len, void *opaque,
1460 struct hivex_visitor vtor;
1461 memset (&vtor, 0, sizeof vtor);
1463 /* Note that len might be larger *or smaller* than the expected size. */
1464 size_t copysize = len <= sizeof vtor ? len : sizeof vtor;
1465 memcpy (&vtor, visitor, copysize);
1467 /* This bitmap records unvisited nodes, so we don't loop if the
1468 * registry contains cycles.
1470 char *unvisited = malloc (1 + h->size / 32);
1471 if (unvisited == NULL)
1473 memcpy (unvisited, h->bitmap, 1 + h->size / 32);
1475 int r = hivex__visit_node (h, node, &vtor, unvisited, opaque, flags);
1481 hivex__visit_node (hive_h *h, hive_node_h node,
1482 const struct hivex_visitor *vtor, char *unvisited,
1483 void *opaque, int flags)
1485 int skip_bad = flags & HIVEX_VISIT_SKIP_BAD;
1487 hive_value_h *values = NULL;
1488 hive_node_h *children = NULL;
1494 /* Return -1 on all callback errors. However on internal errors,
1495 * check if skip_bad is set and suppress those errors if so.
1499 if (!BITMAP_TST (unvisited, node)) {
1501 fprintf (stderr, "hivex__visit_node: contains cycle: visited node 0x%zx already\n",
1505 return skip_bad ? 0 : -1;
1507 BITMAP_CLR (unvisited, node);
1509 name = hivex_node_name (h, node);
1510 if (!name) return skip_bad ? 0 : -1;
1511 if (vtor->node_start && vtor->node_start (h, opaque, node, name) == -1)
1514 values = hivex_node_values (h, node);
1516 ret = skip_bad ? 0 : -1;
1520 for (i = 0; values[i] != 0; ++i) {
1524 if (hivex_value_type (h, values[i], &t, &len) == -1) {
1525 ret = skip_bad ? 0 : -1;
1529 key = hivex_value_key (h, values[i]);
1531 ret = skip_bad ? 0 : -1;
1535 if (vtor->value_any) {
1536 str = hivex_value_value (h, values[i], &t, &len);
1538 ret = skip_bad ? 0 : -1;
1541 if (vtor->value_any (h, opaque, node, values[i], t, len, key, str) == -1)
1543 free (str); str = NULL;
1548 str = hivex_value_value (h, values[i], &t, &len);
1550 ret = skip_bad ? 0 : -1;
1553 if (t != hive_t_none) {
1554 ret = skip_bad ? 0 : -1;
1557 if (vtor->value_none &&
1558 vtor->value_none (h, opaque, node, values[i], t, len, key, str) == -1)
1560 free (str); str = NULL;
1564 case hive_t_expand_string:
1566 str = hivex_value_string (h, values[i]);
1568 if (errno != EILSEQ && errno != EINVAL) {
1569 ret = skip_bad ? 0 : -1;
1572 if (vtor->value_string_invalid_utf16) {
1573 str = hivex_value_value (h, values[i], &t, &len);
1574 if (vtor->value_string_invalid_utf16 (h, opaque, node, values[i], t, len, key, str) == -1)
1576 free (str); str = NULL;
1580 if (vtor->value_string &&
1581 vtor->value_string (h, opaque, node, values[i], t, len, key, str) == -1)
1583 free (str); str = NULL;
1587 case hive_t_dword_be: {
1588 int32_t i32 = hivex_value_dword (h, values[i]);
1589 if (vtor->value_dword &&
1590 vtor->value_dword (h, opaque, node, values[i], t, len, key, i32) == -1)
1595 case hive_t_qword: {
1596 int64_t i64 = hivex_value_qword (h, values[i]);
1597 if (vtor->value_qword &&
1598 vtor->value_qword (h, opaque, node, values[i], t, len, key, i64) == -1)
1604 str = hivex_value_value (h, values[i], &t, &len);
1606 ret = skip_bad ? 0 : -1;
1609 if (t != hive_t_binary) {
1610 ret = skip_bad ? 0 : -1;
1613 if (vtor->value_binary &&
1614 vtor->value_binary (h, opaque, node, values[i], t, len, key, str) == -1)
1616 free (str); str = NULL;
1619 case hive_t_multiple_strings:
1620 strs = hivex_value_multiple_strings (h, values[i]);
1622 if (errno != EILSEQ && errno != EINVAL) {
1623 ret = skip_bad ? 0 : -1;
1626 if (vtor->value_string_invalid_utf16) {
1627 str = hivex_value_value (h, values[i], &t, &len);
1628 if (vtor->value_string_invalid_utf16 (h, opaque, node, values[i], t, len, key, str) == -1)
1630 free (str); str = NULL;
1634 if (vtor->value_multiple_strings &&
1635 vtor->value_multiple_strings (h, opaque, node, values[i], t, len, key, strs) == -1)
1637 free_strings (strs); strs = NULL;
1640 case hive_t_resource_list:
1641 case hive_t_full_resource_description:
1642 case hive_t_resource_requirements_list:
1644 str = hivex_value_value (h, values[i], &t, &len);
1646 ret = skip_bad ? 0 : -1;
1649 if (vtor->value_other &&
1650 vtor->value_other (h, opaque, node, values[i], t, len, key, str) == -1)
1652 free (str); str = NULL;
1657 free (key); key = NULL;
1660 children = hivex_node_children (h, node);
1661 if (children == NULL) {
1662 ret = skip_bad ? 0 : -1;
1666 for (i = 0; children[i] != 0; ++i) {
1668 fprintf (stderr, "hivex__visit_node: %s: visiting subkey %d (0x%zx)\n",
1669 name, i, children[i]);
1671 if (hivex__visit_node (h, children[i], vtor, unvisited, opaque, flags) == -1)
1675 if (vtor->node_end && vtor->node_end (h, opaque, node, name) == -1)
1686 free_strings (strs);
1690 /*----------------------------------------------------------------------
1694 /* Allocate an hbin (page), extending the malloc'd space if necessary,
1695 * and updating the hive handle fields (but NOT the hive disk header
1696 * -- the hive disk header is updated when we commit). This function
1697 * also extends the bitmap if necessary.
1699 * 'allocation_hint' is the size of the block allocation we would like
1700 * to make. Normally registry blocks are very small (avg 50 bytes)
1701 * and are contained in standard-sized pages (4KB), but the registry
1702 * can support blocks which are larger than a standard page, in which
1703 * case it creates a page of 8KB, 12KB etc.
1706 * > 0 : offset of first usable byte of new page (after page header)
1707 * 0 : error (errno set)
1710 allocate_page (hive_h *h, size_t allocation_hint)
1712 /* In almost all cases this will be 1. */
1713 size_t nr_4k_pages =
1714 1 + (allocation_hint + sizeof (struct ntreg_hbin_page) - 1) / 4096;
1715 assert (nr_4k_pages >= 1);
1717 /* 'extend' is the number of bytes to extend the file by. Note that
1718 * hives found in the wild often contain slack between 'endpages'
1719 * and the actual end of the file, so we don't always need to make
1722 ssize_t extend = h->endpages + nr_4k_pages * 4096 - h->size;
1724 if (h->msglvl >= 2) {
1725 fprintf (stderr, "allocate_page: current endpages = 0x%zx, current size = 0x%zx\n",
1726 h->endpages, h->size);
1727 fprintf (stderr, "allocate_page: extending file by %zd bytes (<= 0 if no extension)\n",
1732 size_t oldsize = h->size;
1733 size_t newsize = h->size + extend;
1734 char *newaddr = realloc (h->addr, newsize);
1735 if (newaddr == NULL)
1738 size_t oldbitmapsize = 1 + oldsize / 32;
1739 size_t newbitmapsize = 1 + newsize / 32;
1740 char *newbitmap = realloc (h->bitmap, newbitmapsize);
1741 if (newbitmap == NULL) {
1748 h->bitmap = newbitmap;
1750 memset (h->addr + oldsize, 0, newsize - oldsize);
1751 memset (h->bitmap + oldbitmapsize, 0, newbitmapsize - oldbitmapsize);
1754 size_t offset = h->endpages;
1755 h->endpages += nr_4k_pages * 4096;
1758 fprintf (stderr, "allocate_page: new endpages = 0x%zx, new size = 0x%zx\n",
1759 h->endpages, h->size);
1761 /* Write the hbin header. */
1762 struct ntreg_hbin_page *page =
1763 (struct ntreg_hbin_page *) (h->addr + offset);
1764 page->magic[0] = 'h';
1765 page->magic[1] = 'b';
1766 page->magic[2] = 'i';
1767 page->magic[3] = 'n';
1768 page->offset_first = htole32 (offset - 0x1000);
1769 page->page_size = htole32 (nr_4k_pages * 4096);
1770 memset (page->unknown, 0, sizeof (page->unknown));
1773 fprintf (stderr, "allocate_page: new page at 0x%zx\n", offset);
1775 /* Offset of first usable byte after the header. */
1776 return offset + sizeof (struct ntreg_hbin_page);
1779 /* Allocate a single block, first allocating an hbin (page) at the end
1780 * of the current file if necessary. NB. To keep the implementation
1781 * simple and more likely to be correct, we do not reuse existing free
1784 * seg_len is the size of the block (this INCLUDES the block header).
1785 * The header of the block is initialized to -seg_len (negative to
1786 * indicate used). id[2] is the block ID (type), eg. "nk" for nk-
1787 * record. The block bitmap is updated to show this block as valid.
1788 * The rest of the contents of the block will be zero.
1791 * > 0 : offset of new block
1792 * 0 : error (errno set)
1795 allocate_block (hive_h *h, size_t seg_len, const char id[2])
1803 /* The caller probably forgot to include the header. Note that
1804 * value lists have no ID field, so seg_len == 4 would be possible
1805 * for them, albeit unusual.
1808 fprintf (stderr, "allocate_block: refusing too small allocation (%zu), returning ERANGE\n",
1814 /* Refuse really large allocations. */
1815 if (seg_len > 1000000) {
1817 fprintf (stderr, "allocate_block: refusing large allocation (%zu), returning ERANGE\n",
1823 /* Round up allocation to multiple of 8 bytes. All blocks must be
1824 * on an 8 byte boundary.
1826 seg_len = (seg_len + 7) & ~7;
1828 /* Allocate a new page if necessary. */
1829 if (h->endblocks == 0 || h->endblocks + seg_len > h->endpages) {
1830 size_t newendblocks = allocate_page (h, seg_len);
1831 if (newendblocks == 0)
1833 h->endblocks = newendblocks;
1836 size_t offset = h->endblocks;
1839 fprintf (stderr, "allocate_block: new block at 0x%zx, size %zu\n",
1842 struct ntreg_hbin_block *blockhdr =
1843 (struct ntreg_hbin_block *) (h->addr + offset);
1845 blockhdr->seg_len = htole32 (- (int32_t) seg_len);
1846 if (id[0] && id[1] && seg_len >= 6) {
1847 blockhdr->id[0] = id[0];
1848 blockhdr->id[1] = id[1];
1851 BITMAP_SET (h->bitmap, offset);
1853 h->endblocks += seg_len;
1855 /* If there is space after the last block in the last page, then we
1856 * have to put a dummy free block header here to mark the rest of
1859 ssize_t rem = h->endpages - h->endblocks;
1862 fprintf (stderr, "allocate_block: marking remainder of page free starting at 0x%zx, size %zd\n",
1867 blockhdr = (struct ntreg_hbin_block *) (h->addr + h->endblocks);
1868 blockhdr->seg_len = htole32 ((int32_t) rem);
1874 /* 'offset' must point to a valid, used block. This function marks
1875 * the block unused (by updating the seg_len field) and invalidates
1876 * the bitmap. It does NOT do this recursively, so to avoid creating
1877 * unreachable used blocks, callers may have to recurse over the hive
1878 * structures. Also callers must ensure there are no references to
1879 * this block from other parts of the hive.
1882 mark_block_unused (hive_h *h, size_t offset)
1884 assert (h->writable);
1885 assert (IS_VALID_BLOCK (h, offset));
1888 fprintf (stderr, "mark_block_unused: marking 0x%zx unused\n", offset);
1890 struct ntreg_hbin_block *blockhdr =
1891 (struct ntreg_hbin_block *) (h->addr + offset);
1893 size_t seg_len = block_len (h, offset, NULL);
1894 blockhdr->seg_len = htole32 (seg_len);
1896 BITMAP_CLR (h->bitmap, offset);
1899 /* Delete all existing values at this node. */
1901 delete_values (hive_h *h, hive_node_h node)
1903 assert (h->writable);
1905 hive_value_h *values;
1907 if (get_values (h, node, &values, &blocks) == -1)
1911 for (i = 0; blocks[i] != 0; ++i)
1912 mark_block_unused (h, blocks[i]);
1916 for (i = 0; values[i] != 0; ++i) {
1917 struct ntreg_vk_record *vk =
1918 (struct ntreg_vk_record *) (h->addr + values[i]);
1921 len = le32toh (vk->data_len);
1922 if (len == 0x80000000) /* special case */
1926 if (len > 4) { /* non-inline, so remove data block */
1927 size_t data_offset = le32toh (vk->data_offset);
1928 data_offset += 0x1000;
1929 mark_block_unused (h, data_offset);
1932 /* remove vk record */
1933 mark_block_unused (h, values[i]);
1938 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
1939 nk->nr_values = htole32 (0);
1940 nk->vallist = htole32 (0xffffffff);
1946 hivex_commit (hive_h *h, const char *filename, int flags)
1958 filename = filename ? : h->filename;
1959 int fd = open (filename, O_WRONLY|O_CREAT|O_TRUNC|O_NOCTTY, 0666);
1963 /* Update the header fields. */
1964 uint32_t sequence = le32toh (h->hdr->sequence1);
1966 h->hdr->sequence1 = htole32 (sequence);
1967 h->hdr->sequence2 = htole32 (sequence);
1968 /* XXX Ought to update h->hdr->last_modified. */
1969 h->hdr->blocks = htole32 (h->endpages - 0x1000);
1971 /* Recompute header checksum. */
1972 uint32_t sum = header_checksum (h);
1973 h->hdr->csum = htole32 (sum);
1976 fprintf (stderr, "hivex_commit: new header checksum: 0x%x\n", sum);
1978 if (full_write (fd, h->addr, h->size) != h->size) {
1985 if (close (fd) == -1)
1991 /* Calculate the hash for a lf or lh record offset.
1994 calc_hash (const char *type, const char *name, char *ret)
1996 size_t len = strlen (name);
1998 if (STRPREFIX (type, "lf"))
1999 /* Old-style, not used in current registries. */
2000 memcpy (ret, name, len < 4 ? len : 4);
2002 /* New-style for lh-records. */
2005 for (i = 0; i < len; ++i) {
2006 c = c_toupper (name[i]);
2010 *((uint32_t *) ret) = htole32 (h);
2014 /* Create a completely new lh-record containing just the single node. */
2016 new_lh_record (hive_h *h, const char *name, hive_node_h node)
2018 static const char id[2] = { 'l', 'h' };
2019 size_t seg_len = sizeof (struct ntreg_lf_record);
2020 size_t offset = allocate_block (h, seg_len, id);
2024 struct ntreg_lf_record *lh = (struct ntreg_lf_record *) (h->addr + offset);
2025 lh->nr_keys = htole16 (1);
2026 lh->keys[0].offset = htole32 (node - 0x1000);
2027 calc_hash ("lh", name, lh->keys[0].hash);
2032 /* Insert node into existing lf/lh-record at position.
2033 * This allocates a new record and marks the old one as unused.
2036 insert_lf_record (hive_h *h, size_t old_offs, size_t posn,
2037 const char *name, hive_node_h node)
2039 assert (IS_VALID_BLOCK (h, old_offs));
2041 /* Work around C stupidity.
2042 * http://www.redhat.com/archives/libguestfs/2010-February/msg00056.html
2044 int test = BLOCK_ID_EQ (h, old_offs, "lf") || BLOCK_ID_EQ (h, old_offs, "lh");
2047 struct ntreg_lf_record *old_lf =
2048 (struct ntreg_lf_record *) (h->addr + old_offs);
2049 size_t nr_keys = le16toh (old_lf->nr_keys);
2051 nr_keys++; /* in new record ... */
2053 size_t seg_len = sizeof (struct ntreg_lf_record) + (nr_keys-1) * 8;
2054 size_t new_offs = allocate_block (h, seg_len, old_lf->id);
2058 struct ntreg_lf_record *new_lf =
2059 (struct ntreg_lf_record *) (h->addr + new_offs);
2060 new_lf->nr_keys = htole16 (nr_keys);
2062 /* Copy the keys until we reach posn, insert the new key there, then
2063 * copy the remaining keys.
2066 for (i = 0; i < posn; ++i)
2067 new_lf->keys[i] = old_lf->keys[i];
2069 new_lf->keys[i].offset = htole32 (node - 0x1000);
2070 calc_hash (new_lf->id, name, new_lf->keys[i].hash);
2072 for (i = posn+1; i < nr_keys; ++i)
2073 new_lf->keys[i] = old_lf->keys[i-1];
2075 /* Old block is unused, return new block. */
2076 mark_block_unused (h, old_offs);
2080 /* Compare name with name in nk-record. */
2082 compare_name_with_nk_name (hive_h *h, const char *name, hive_node_h nk_offs)
2084 assert (IS_VALID_BLOCK (h, nk_offs));
2085 assert (BLOCK_ID_EQ (h, nk_offs, "nk"));
2087 /* Name in nk is not necessarily nul-terminated. */
2088 char *nname = hivex_node_name (h, nk_offs);
2090 /* Unfortunately we don't have a way to return errors here. */
2092 perror ("compare_name_with_nk_name");
2096 int r = strcasecmp (name, nname);
2103 hivex_node_add_child (hive_h *h, hive_node_h parent, const char *name)
2110 if (!IS_VALID_BLOCK (h, parent) || !BLOCK_ID_EQ (h, parent, "nk")) {
2115 if (name == NULL || strlen (name) == 0) {
2120 if (hivex_node_get_child (h, parent, name) != 0) {
2125 /* Create the new nk-record. */
2126 static const char nk_id[2] = { 'n', 'k' };
2127 size_t seg_len = sizeof (struct ntreg_nk_record) + strlen (name);
2128 hive_node_h node = allocate_block (h, seg_len, nk_id);
2133 fprintf (stderr, "hivex_node_add_child: allocated new nk-record for child at 0x%zx\n", node);
2135 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
2136 nk->flags = htole16 (0x0020); /* key is ASCII. */
2137 nk->parent = htole32 (parent - 0x1000);
2138 nk->subkey_lf = htole32 (0xffffffff);
2139 nk->subkey_lf_volatile = htole32 (0xffffffff);
2140 nk->vallist = htole32 (0xffffffff);
2141 nk->classname = htole32 (0xffffffff);
2142 nk->name_len = htole16 (strlen (name));
2143 strcpy (nk->name, name);
2145 /* Inherit parent sk. */
2146 struct ntreg_nk_record *parent_nk =
2147 (struct ntreg_nk_record *) (h->addr + parent);
2148 size_t parent_sk_offset = le32toh (parent_nk->sk);
2149 parent_sk_offset += 0x1000;
2150 if (!IS_VALID_BLOCK (h, parent_sk_offset) ||
2151 !BLOCK_ID_EQ (h, parent_sk_offset, "sk")) {
2153 fprintf (stderr, "hivex_node_add_child: returning EFAULT because parent sk is not a valid block (%zu)\n",
2158 struct ntreg_sk_record *sk =
2159 (struct ntreg_sk_record *) (h->addr + parent_sk_offset);
2160 sk->refcount = htole32 (le32toh (sk->refcount) + 1);
2161 nk->sk = htole32 (parent_sk_offset - 0x1000);
2163 /* Inherit parent timestamp. */
2164 memcpy (nk->timestamp, parent_nk->timestamp, sizeof (parent_nk->timestamp));
2166 /* What I found out the hard way (not documented anywhere): the
2167 * subkeys in lh-records must be kept sorted. If you just add a
2168 * subkey in a non-sorted position (eg. just add it at the end) then
2169 * Windows won't see the subkey _and_ Windows will corrupt the hive
2170 * itself when it modifies or saves it.
2172 * So use get_children() to get a list of intermediate
2173 * lf/lh-records. get_children() returns these in reading order
2174 * (which is sorted), so we look for the lf/lh-records in sequence
2175 * until we find the key name just after the one we are inserting,
2176 * and we insert the subkey just before it.
2178 * The only other case is the no-subkeys case, where we have to
2179 * create a brand new lh-record.
2181 hive_node_h *unused;
2184 if (get_children (h, parent, &unused, &blocks, 0) == -1)
2189 size_t nr_subkeys_in_parent_nk = le32toh (parent_nk->nr_subkeys);
2190 if (nr_subkeys_in_parent_nk == 0) { /* No subkeys case. */
2191 /* Free up any existing intermediate blocks. */
2192 for (i = 0; blocks[i] != 0; ++i)
2193 mark_block_unused (h, blocks[i]);
2194 size_t lh_offs = new_lh_record (h, name, node);
2201 fprintf (stderr, "hivex_node_add_child: no keys, allocated new lh-record at 0x%zx\n", lh_offs);
2203 parent_nk->subkey_lf = htole32 (lh_offs - 0x1000);
2205 else { /* Insert subkeys case. */
2206 size_t old_offs = 0, new_offs = 0;
2207 struct ntreg_lf_record *old_lf = NULL;
2209 /* Find lf/lh key name just after the one we are inserting. */
2210 for (i = 0; blocks[i] != 0; ++i) {
2211 if (BLOCK_ID_EQ (h, blocks[i], "lf") ||
2212 BLOCK_ID_EQ (h, blocks[i], "lh")) {
2213 old_offs = blocks[i];
2214 old_lf = (struct ntreg_lf_record *) (h->addr + old_offs);
2215 for (j = 0; j < le16toh (old_lf->nr_keys); ++j) {
2216 hive_node_h nk_offs = le32toh (old_lf->keys[j].offset);
2218 if (compare_name_with_nk_name (h, name, nk_offs) < 0)
2224 /* Insert it at the end.
2225 * old_offs points to the last lf record, set j.
2227 assert (old_offs != 0); /* should never happen if nr_subkeys > 0 */
2228 j = le16toh (old_lf->nr_keys);
2233 fprintf (stderr, "hivex_node_add_child: insert key in existing lh-record at 0x%zx, posn %zu\n", old_offs, j);
2235 new_offs = insert_lf_record (h, old_offs, j, name, node);
2236 if (new_offs == 0) {
2242 fprintf (stderr, "hivex_node_add_child: new lh-record at 0x%zx\n",
2245 /* If the lf/lh-record was directly referenced by the parent nk,
2246 * then update the parent nk.
2248 if (le32toh (parent_nk->subkey_lf) + 0x1000 == old_offs)
2249 parent_nk->subkey_lf = htole32 (new_offs - 0x1000);
2250 /* Else we have to look for the intermediate ri-record and update
2254 for (i = 0; blocks[i] != 0; ++i) {
2255 if (BLOCK_ID_EQ (h, blocks[i], "ri")) {
2256 struct ntreg_ri_record *ri =
2257 (struct ntreg_ri_record *) (h->addr + blocks[i]);
2258 for (j = 0; j < le16toh (ri->nr_offsets); ++j)
2259 if (le32toh (ri->offset[j] + 0x1000) == old_offs) {
2260 ri->offset[j] = htole32 (new_offs - 0x1000);
2266 /* Not found .. This is an internal error. */
2268 fprintf (stderr, "hivex_node_add_child: returning ENOTSUP because could not find ri->lf link\n");
2280 /* Update nr_subkeys in parent nk. */
2281 nr_subkeys_in_parent_nk++;
2282 parent_nk->nr_subkeys = htole32 (nr_subkeys_in_parent_nk);
2284 /* Update max_subkey_name_len in parent nk. */
2285 uint16_t max = le16toh (parent_nk->max_subkey_name_len);
2286 if (max < strlen (name) * 2) /* *2 because "recoded" in UTF16-LE. */
2287 parent_nk->max_subkey_name_len = htole16 (strlen (name) * 2);
2292 /* Decrement the refcount of an sk-record, and if it reaches zero,
2293 * unlink it from the chain and delete it.
2296 delete_sk (hive_h *h, size_t sk_offset)
2298 if (!IS_VALID_BLOCK (h, sk_offset) || !BLOCK_ID_EQ (h, sk_offset, "sk")) {
2300 fprintf (stderr, "delete_sk: not an sk record: 0x%zx\n", sk_offset);
2305 struct ntreg_sk_record *sk = (struct ntreg_sk_record *) (h->addr + sk_offset);
2307 if (sk->refcount == 0) {
2309 fprintf (stderr, "delete_sk: sk record already has refcount 0: 0x%zx\n",
2317 if (sk->refcount == 0) {
2318 size_t sk_prev_offset = sk->sk_prev;
2319 sk_prev_offset += 0x1000;
2321 size_t sk_next_offset = sk->sk_next;
2322 sk_next_offset += 0x1000;
2324 /* Update sk_prev/sk_next SKs, unless they both point back to this
2325 * cell in which case we are deleting the last SK.
2327 if (sk_prev_offset != sk_offset && sk_next_offset != sk_offset) {
2328 struct ntreg_sk_record *sk_prev =
2329 (struct ntreg_sk_record *) (h->addr + sk_prev_offset);
2330 struct ntreg_sk_record *sk_next =
2331 (struct ntreg_sk_record *) (h->addr + sk_next_offset);
2333 sk_prev->sk_next = htole32 (sk_next_offset - 0x1000);
2334 sk_next->sk_prev = htole32 (sk_prev_offset - 0x1000);
2337 /* Refcount is zero so really delete this block. */
2338 mark_block_unused (h, sk_offset);
2344 /* Callback from hivex_node_delete_child which is called to delete a
2345 * node AFTER its subnodes have been visited. The subnodes have been
2346 * deleted but we still have to delete any lf/lh/li/ri records and the
2347 * value list block and values, followed by deleting the node itself.
2350 delete_node (hive_h *h, void *opaque, hive_node_h node, const char *name)
2352 /* Get the intermediate blocks. The subkeys have already been
2353 * deleted by this point, so tell get_children() not to check for
2354 * validity of the nk-records.
2356 hive_node_h *unused;
2358 if (get_children (h, node, &unused, &blocks, GET_CHILDREN_NO_CHECK_NK) == -1)
2362 /* We don't care what's in these intermediate blocks, so we can just
2363 * delete them unconditionally.
2366 for (i = 0; blocks[i] != 0; ++i)
2367 mark_block_unused (h, blocks[i]);
2371 /* Delete the values in the node. */
2372 if (delete_values (h, node) == -1)
2375 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
2377 /* If the NK references an SK, delete it. */
2378 size_t sk_offs = le32toh (nk->sk);
2379 if (sk_offs != 0xffffffff) {
2381 if (delete_sk (h, sk_offs) == -1)
2383 nk->sk = htole32 (0xffffffff);
2386 /* If the NK references a classname, delete it. */
2387 size_t cl_offs = le32toh (nk->classname);
2388 if (cl_offs != 0xffffffff) {
2390 mark_block_unused (h, cl_offs);
2391 nk->classname = htole32 (0xffffffff);
2394 /* Delete the node itself. */
2395 mark_block_unused (h, node);
2401 hivex_node_delete_child (hive_h *h, hive_node_h node)
2408 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
2413 if (node == hivex_root (h)) {
2415 fprintf (stderr, "hivex_node_delete_child: cannot delete root node\n");
2420 hive_node_h parent = hivex_node_parent (h, node);
2424 /* Delete node and all its children and values recursively. */
2425 static const struct hivex_visitor visitor = { .node_end = delete_node };
2426 if (hivex_visit_node (h, node, &visitor, sizeof visitor, NULL, 0) == -1)
2429 /* Delete the link from parent to child. We need to find the lf/lh
2430 * record which contains the offset and remove the offset from that
2431 * record, then decrement the element count in that record, and
2432 * decrement the overall number of subkeys stored in the parent
2435 hive_node_h *unused;
2437 if (get_children (h, parent, &unused, &blocks, GET_CHILDREN_NO_CHECK_NK)== -1)
2442 for (i = 0; blocks[i] != 0; ++i) {
2443 struct ntreg_hbin_block *block =
2444 (struct ntreg_hbin_block *) (h->addr + blocks[i]);
2446 if (block->id[0] == 'l' && (block->id[1] == 'f' || block->id[1] == 'h')) {
2447 struct ntreg_lf_record *lf = (struct ntreg_lf_record *) block;
2449 size_t nr_subkeys_in_lf = le16toh (lf->nr_keys);
2451 for (j = 0; j < nr_subkeys_in_lf; ++j)
2452 if (le32toh (lf->keys[j].offset) + 0x1000 == node) {
2453 for (; j < nr_subkeys_in_lf - 1; ++j)
2454 memcpy (&lf->keys[j], &lf->keys[j+1], sizeof (lf->keys[j]));
2455 lf->nr_keys = htole16 (nr_subkeys_in_lf - 1);
2461 fprintf (stderr, "hivex_node_delete_child: could not find parent to child link\n");
2466 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + parent);
2467 size_t nr_subkeys_in_nk = le32toh (nk->nr_subkeys);
2468 nk->nr_subkeys = htole32 (nr_subkeys_in_nk - 1);
2471 fprintf (stderr, "hivex_node_delete_child: updating nr_subkeys in parent 0x%zx to %zu\n",
2472 parent, nr_subkeys_in_nk);
2478 hivex_node_set_values (hive_h *h, hive_node_h node,
2479 size_t nr_values, const hive_set_value *values,
2487 if (!IS_VALID_BLOCK (h, node) || !BLOCK_ID_EQ (h, node, "nk")) {
2492 /* Delete all existing values. */
2493 if (delete_values (h, node) == -1)
2499 /* Allocate value list node. Value lists have no id field. */
2500 static const char nul_id[2] = { 0, 0 };
2502 sizeof (struct ntreg_value_list) + (nr_values - 1) * sizeof (uint32_t);
2503 size_t vallist_offs = allocate_block (h, seg_len, nul_id);
2504 if (vallist_offs == 0)
2507 struct ntreg_nk_record *nk = (struct ntreg_nk_record *) (h->addr + node);
2508 nk->nr_values = htole32 (nr_values);
2509 nk->vallist = htole32 (vallist_offs - 0x1000);
2511 struct ntreg_value_list *vallist =
2512 (struct ntreg_value_list *) (h->addr + vallist_offs);
2515 for (i = 0; i < nr_values; ++i) {
2516 /* Allocate vk record to store this (key, value) pair. */
2517 static const char vk_id[2] = { 'v', 'k' };
2518 seg_len = sizeof (struct ntreg_vk_record) + strlen (values[i].key);
2519 size_t vk_offs = allocate_block (h, seg_len, vk_id);
2523 vallist->offset[i] = htole32 (vk_offs - 0x1000);
2525 struct ntreg_vk_record *vk = (struct ntreg_vk_record *) (h->addr + vk_offs);
2526 size_t name_len = strlen (values[i].key);
2527 vk->name_len = htole16 (name_len);
2528 strcpy (vk->name, values[i].key);
2529 vk->data_type = htole32 (values[i].t);
2530 vk->data_len = htole16 (values[i].len);
2531 vk->flags = name_len == 0 ? 0 : 1;
2533 if (values[i].len <= 4) /* Store data inline. */
2534 memcpy (&vk->data_offset, values[i].value, values[i].len);
2536 size_t offs = allocate_block (h, values[i].len + 4, nul_id);
2539 memcpy (h->addr + offs + 4, values[i].value, values[i].len);
2540 vk->data_offset = htole32 (offs - 0x1000);
2543 if (name_len * 2 > le32toh (nk->max_vk_name_len))
2544 nk->max_vk_name_len = htole32 (name_len * 2);
2545 if (values[i].len > le32toh (nk->max_vk_data_len))
2546 nk->max_vk_data_len = htole32 (values[i].len);