X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Flookup_table.c;h=007cd9eae689d655ec4580b7d2a00bfcf9fb84f2;hb=ed828dc4cbf4eabc9f8b32cb4c6a86a04e087f7a;hp=0c6a55fcccb9463c8092fda7269ef09b048745a8;hpb=b51021c82376ce8d2aecb206c8cbfc4ffa6d4f3a;p=wimlib diff --git a/src/lookup_table.c b/src/lookup_table.c index 0c6a55fc..007cd9ea 100644 --- a/src/lookup_table.c +++ b/src/lookup_table.c @@ -52,7 +52,7 @@ err: return NULL; } -struct lookup_table_entry *new_lookup_table_entry(WIMStruct *wim) +struct lookup_table_entry *new_lookup_table_entry() { struct lookup_table_entry *lte; @@ -66,7 +66,6 @@ struct lookup_table_entry *new_lookup_table_entry(WIMStruct *wim) lte->part_number = 1; lte->refcnt = 1; - lte->wim = wim; INIT_LIST_HEAD(<e->lte_group_list); return lte; } @@ -197,7 +196,7 @@ int read_lookup_table(WIMStruct *w) while (num_entries--) { const u8 *p; - struct lookup_table_entry *cur_entry; + struct lookup_table_entry *cur_entry, *duplicate_entry; if (fread(buf, 1, sizeof(buf), w->fp) != sizeof(buf)) { if (feof(w->fp)) { @@ -209,7 +208,7 @@ int read_lookup_table(WIMStruct *w) ret = WIMLIB_ERR_READ; goto out; } - cur_entry = new_lookup_table_entry(w); + cur_entry = new_lookup_table_entry(); if (!cur_entry) { ret = WIMLIB_ERR_NOMEM; goto out; @@ -221,7 +220,40 @@ int read_lookup_table(WIMStruct *w) p = get_u16(p, &cur_entry->part_number); p = get_u32(p, &cur_entry->refcnt); p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash); + + if (is_zero_hash(cur_entry->hash)) { + ERROR("The WIM lookup table contains an entry with a " + "SHA1 message digest of all 0's"); + ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY; + FREE(cur_entry); + goto out; + } + + duplicate_entry = __lookup_resource(table, cur_entry->hash); + if (duplicate_entry) { + ERROR("The WIM lookup table contains two entries with the " + "same SHA1 message digest!"); + ERROR("The first entry is:"); + print_lookup_table_entry(duplicate_entry); + ERROR("The second entry is:"); + print_lookup_table_entry(cur_entry); + ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY; + FREE(cur_entry); + goto out; + } lookup_table_insert(table, cur_entry); + + if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED) + && (cur_entry->resource_entry.size != + cur_entry->resource_entry.original_size)) + { + ERROR("Found uncompressed resource with original size " + "not the same as compressed size"); + ERROR("The lookup table entry for the resource is as follows:"); + print_lookup_table_entry(cur_entry); + ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY; + goto out; + } } DEBUG("Done reading lookup table."); w->lookup_table = table; @@ -243,10 +275,6 @@ int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out) out = __out; - /* do not write lookup table entries for empty files */ - if (lte->output_resource_entry.original_size == 0) - return 0; - /* Don't write entries that have not had file resources or metadata * resources written for them. */ if (lte->out_refcnt == 0) @@ -316,6 +344,8 @@ void print_lookup_table_entry(const struct lookup_table_entry *lte) case RESOURCE_IN_STAGING_FILE: printf("Staging File = `%s'\n", lte->staging_file_name); break; + default: + break; } putchar('\n'); } @@ -369,7 +399,20 @@ int lookup_resource(WIMStruct *w, const char *path, struct dentry *dentry; struct lookup_table_entry *lte; unsigned stream_idx; + const char *stream_name = NULL; + char *p = NULL; + + if (lookup_flags & LOOKUP_FLAG_ADS_OK) { + stream_name = path_stream_name(path); + if (stream_name) { + p = (char*)stream_name - 1; + *p = '\0'; + } + } + dentry = get_dentry(w, path); + if (p) + *p = ':'; if (!dentry) return -ENOENT; @@ -380,22 +423,19 @@ int lookup_resource(WIMStruct *w, const char *path, && dentry_is_directory(dentry)) return -EISDIR; stream_idx = 0; - if (lookup_flags & LOOKUP_FLAG_ADS_OK) { - const char *stream_name = path_stream_name(path); - if (stream_name) { - size_t stream_name_len = strlen(stream_name); - for (u16 i = 0; i < dentry->num_ads; i++) { - if (ads_entry_has_name(&dentry->ads_entries[i], - stream_name, - stream_name_len)) - { - stream_idx = i + 1; - lte = dentry->ads_entries[i].lte; - goto out; - } + if (stream_name) { + size_t stream_name_len = strlen(stream_name); + for (u16 i = 0; i < dentry->num_ads; i++) { + if (ads_entry_has_name(&dentry->ads_entries[i], + stream_name, + stream_name_len)) + { + stream_idx = i + 1; + lte = dentry->ads_entries[i].lte; + goto out; } - return -ENOENT; } + return -ENOENT; } out: if (dentry_ret) @@ -452,12 +492,29 @@ int dentry_resolve_ltes(struct dentry *dentry, void *__table) return 0; } +/* Return the lookup table entry for the unnamed data stream of a dentry, or + * NULL if there is none. + * + * You'd think this would be easier than it actually is, since the unnamed data + * stream should be the one referenced from the dentry itself. Alas, if there + * are named data streams, Microsoft's "imagex.exe" program will put the unnamed + * data stream in one of the alternate data streams instead of inside the + * dentry. So we need to check the alternate data streams too. + * + * Also, note that a dentry may appear to have than one unnamed stream, but if + * the SHA1 message digest is all 0's then the corresponding stream does not + * really "count" (this is the case for the dentry's own file stream when the + * file stream that should be there is actually in one of the alternate stream + * entries.). This is despite the fact that we may need to extract such a + * missing entry as an empty file or empty named data stream. + */ struct lookup_table_entry * -dentry_first_lte(const struct dentry *dentry, const struct lookup_table *table) +dentry_unnamed_lte(const struct dentry *dentry, + const struct lookup_table *table) { if (dentry->resolved) - return dentry_first_lte_resolved(dentry); + return dentry_unnamed_lte_resolved(dentry); else - return dentry_first_lte_unresolved(dentry, table); + return dentry_unnamed_lte_unresolved(dentry, table); }