]> wimlib.net Git - wimlib/blobdiff - src/lookup_table.c
Refactoring/rewrites
[wimlib] / src / lookup_table.c
index 776ad94b41a72769768e161971208ffd903435ac..67ce92a3f37b79617ac2a9c370e5980c0fe18f9a 100644 (file)
@@ -26,7 +26,7 @@
 
 #include "wimlib_internal.h"
 #include "lookup_table.h"
-#include "io.h"
+#include "buffer_io.h"
 #include <errno.h>
 
 #ifdef WITH_FUSE
@@ -39,47 +39,100 @@ struct lookup_table *new_lookup_table(size_t capacity)
        struct hlist_head *array;
 
        table = MALLOC(sizeof(struct lookup_table));
-       if (!table)
-               goto err;
-       array = CALLOC(capacity, sizeof(array[0]));
-       if (!array) {
-               FREE(table);
-               goto err;
+       if (table) {
+               array = CALLOC(capacity, sizeof(array[0]));
+               if (array) {
+                       table->num_entries = 0;
+                       table->capacity = capacity;
+                       table->array = array;
+               } else {
+                       FREE(table);
+                       table = NULL;
+                       ERROR("Failed to allocate memory for lookup table with capacity %zu",
+                             capacity);
+               }
        }
-       table->num_entries = 0;
-       table->capacity = capacity;
-       table->array = array;
        return table;
-err:
-       ERROR("Failed to allocate memory for lookup table with capacity %zu",
-             capacity);
-       return NULL;
 }
 
 struct lookup_table_entry *new_lookup_table_entry()
 {
        struct lookup_table_entry *lte;
-       
+
        lte = CALLOC(1, sizeof(struct lookup_table_entry));
-       if (!lte) {
+       if (lte) {
+               lte->part_number  = 1;
+               lte->refcnt       = 1;
+       } else {
                ERROR("Out of memory (tried to allocate %zu bytes for "
                      "lookup table entry)",
                      sizeof(struct lookup_table_entry));
-               return NULL;
        }
-
-       lte->part_number  = 1;
-       lte->refcnt       = 1;
        return lte;
 }
 
+struct lookup_table_entry *
+clone_lookup_table_entry(const struct lookup_table_entry *old)
+{
+       struct lookup_table_entry *new;
+
+       new = MALLOC(sizeof(*new));
+       if (!new)
+               return NULL;
+
+       memcpy(new, old, sizeof(*old));
+       new->extracted_file = NULL;
+       switch (new->resource_location) {
+       case RESOURCE_IN_STAGING_FILE:
+       case RESOURCE_IN_FILE_ON_DISK:
+               wimlib_assert((void*)&old->file_on_disk ==
+                             (void*)&old->staging_file_name);
+               new->staging_file_name = STRDUP(old->staging_file_name);
+               if (!new->staging_file_name)
+                       goto out_free;
+               break;
+       case RESOURCE_IN_ATTACHED_BUFFER:
+               new->attached_buffer = MALLOC(wim_resource_size(old));
+               if (!new->attached_buffer)
+                       goto out_free;
+               memcpy(new->attached_buffer, old->attached_buffer,
+                      wim_resource_size(old));
+               break;
+#ifdef WITH_NTFS_3G
+       case RESOURCE_IN_NTFS_VOLUME:
+               if (old->ntfs_loc) {
+                       struct ntfs_location *loc;
+                       loc = MALLOC(sizeof(*loc));
+                       if (!loc)
+                               goto out_free;
+                       memcpy(loc, old->ntfs_loc, sizeof(*loc));
+                       loc->path_utf8 = NULL;
+                       loc->stream_name_utf16 = NULL;
+                       new->ntfs_loc = loc;
+                       loc->path_utf8 = STRDUP(old->ntfs_loc->path_utf8);
+                       if (!loc->path_utf8)
+                               goto out_free;
+                       loc->stream_name_utf16 = MALLOC(loc->stream_name_utf16_num_chars * 2);
+                       if (!loc->stream_name_utf16)
+                               goto out_free;
+                       memcpy(loc->stream_name_utf16,
+                              old->ntfs_loc->stream_name_utf16,
+                              loc->stream_name_utf16_num_chars * 2);
+               }
+               break;
+#endif
+       default:
+               break;
+       }
+       return new;
+out_free:
+       free_lookup_table_entry(new);
+       return NULL;
+}
+
 void free_lookup_table_entry(struct lookup_table_entry *lte)
 {
        if (lte) {
-#ifdef WITH_FUSE
-               if (lte->staging_list.next)
-                       list_del(&lte->staging_list);
-#endif
                switch (lte->resource_location) {
                case RESOURCE_IN_STAGING_FILE:
                case RESOURCE_IN_ATTACHED_BUFFER:
@@ -102,7 +155,6 @@ void free_lookup_table_entry(struct lookup_table_entry *lte)
                default:
                        break;
                }
-               FREE(lte->extracted_file);
                FREE(lte);
        }
 }
@@ -135,7 +187,7 @@ void free_lookup_table(struct lookup_table *table)
  * @table:     A pointer to the lookup table.
  * @entry:     A pointer to the entry to insert.
  */
-void lookup_table_insert(struct lookup_table *table, 
+void lookup_table_insert(struct lookup_table *table,
                         struct lookup_table_entry *lte)
 {
        size_t i = lte->hash_short % table->capacity;
@@ -148,8 +200,12 @@ void lookup_table_insert(struct lookup_table *table,
 static void finalize_lte(struct lookup_table_entry *lte)
 {
        #ifdef WITH_FUSE
-       if (lte->resource_location == RESOURCE_IN_STAGING_FILE)
+       if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
                unlink(lte->staging_file_name);
+               wimlib_assert(lte->staging_list.next);
+               wimlib_assert(lte->staging_list.prev);
+               list_del(&lte->staging_list);
+       }
        #endif
        free_lookup_table_entry(lte);
 }
@@ -158,47 +214,36 @@ static void finalize_lte(struct lookup_table_entry *lte)
  * reference count reaches 0, it is unlinked from the lookup table.  If,
  * furthermore, the entry has no opened file descriptors associated with it, the
  * entry is freed.  */
-struct lookup_table_entry *
-lte_decrement_refcnt(struct lookup_table_entry *lte, struct lookup_table *table)
+void lte_decrement_refcnt(struct lookup_table_entry *lte,
+                         struct lookup_table *table)
 {
-       if (lte) {
-               wimlib_assert(lte->refcnt);
-               if (--lte->refcnt == 0) {
-                       lookup_table_unlink(table, lte);
-               #ifdef WITH_FUSE
-                       if (lte->num_opened_fds == 0)
-               #endif
-                       {
-                               finalize_lte(lte);
-                               lte = NULL;
-                       }
-               }
+       wimlib_assert(lte != NULL);
+       wimlib_assert(lte->refcnt != 0);
+       if (--lte->refcnt == 0) {
+               lookup_table_unlink(table, lte);
+       #ifdef WITH_FUSE
+               if (lte->num_opened_fds == 0)
+       #endif
+                       finalize_lte(lte);
        }
-       return lte;
 }
 
 #ifdef WITH_FUSE
-struct lookup_table_entry *
-lte_decrement_num_opened_fds(struct lookup_table_entry *lte,
-                            struct lookup_table *table)
+void lte_decrement_num_opened_fds(struct lookup_table_entry *lte)
 {
-       if (lte) {
-               wimlib_assert(lte->num_opened_fds);
-               if (--lte->num_opened_fds == 0 && lte->refcnt == 0) {
-                       lookup_table_unlink(table, lte);
+       wimlib_assert(lte != NULL);
+       if (lte->num_opened_fds != 0) {
+               if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
                        finalize_lte(lte);
-                       lte = NULL;
-               }
        }
-       return lte;
 }
 #endif
 
-/* 
+/*
  * Calls a function on all the entries in the lookup table.  Stop early and
  * return nonzero if any call to the function returns nonzero.
  */
-int for_lookup_table_entry(struct lookup_table *table, 
+int for_lookup_table_entry(struct lookup_table *table,
                           int (*visitor)(struct lookup_table_entry *, void *),
                           void *arg)
 {
@@ -224,17 +269,24 @@ int for_lookup_table_entry(struct lookup_table *table,
  */
 int read_lookup_table(WIMStruct *w)
 {
-       u64    num_entries;
-       u8     buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
-       int    ret;
+       u64 num_entries;
+       u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
+       int ret;
        struct lookup_table *table;
        struct lookup_table_entry *cur_entry = NULL, *duplicate_entry;
 
+       if (resource_is_compressed(&w->hdr.lookup_table_res_entry)) {
+               ERROR("Didn't expect a compressed lookup table!");
+               ERROR("Ask the author to implement support for this.");
+               return WIMLIB_ERR_COMPRESSED_LOOKUP_TABLE;
+       }
+
        DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
              w->hdr.lookup_table_res_entry.offset,
              w->hdr.lookup_table_res_entry.original_size);
 
-       if (fseeko(w->fp, w->hdr.lookup_table_res_entry.offset, SEEK_SET) != 0) {
+       if (fseeko(w->fp, w->hdr.lookup_table_res_entry.offset, SEEK_SET) != 0)
+       {
                ERROR_WITH_ERRNO("Failed to seek to byte %"PRIu64" to read "
                                 "lookup table",
                                 w->hdr.lookup_table_res_entry.offset);
@@ -267,7 +319,7 @@ int read_lookup_table(WIMStruct *w)
                }
                cur_entry->wim = w;
                cur_entry->resource_location = RESOURCE_IN_WIM;
-                        
+
                p = get_resource_entry(buf, &cur_entry->resource_entry);
                p = get_u16(p, &cur_entry->part_number);
                p = get_u32(p, &cur_entry->refcnt);
@@ -279,7 +331,6 @@ int read_lookup_table(WIMStruct *w)
                              w->hdr.part_number, cur_entry->part_number);
                        ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
                        goto out_free_cur_entry;
-                       
                }
 
                if (is_zero_hash(cur_entry->hash)) {
@@ -289,8 +340,15 @@ int read_lookup_table(WIMStruct *w)
                        goto out_free_cur_entry;
                }
 
+               /* Ordinarily, no two streams should share the same SHA1 message
+                * digest.  However, this constraint can be broken for metadata
+                * resources--- two identical images will have the same metadata
+                * resource, but their lookup table entries are not shared. */
                duplicate_entry = __lookup_resource(table, cur_entry->hash);
-               if (duplicate_entry) {
+               if (duplicate_entry
+                   && !((duplicate_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
+                         && cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA))
+               {
                        ERROR("The WIM lookup table contains two entries with the "
                              "same SHA1 message digest!");
                        ERROR("The first entry is:");
@@ -303,7 +361,7 @@ int read_lookup_table(WIMStruct *w)
 
                if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
                    && (cur_entry->resource_entry.size !=
-                     cur_entry->resource_entry.original_size))
+                       cur_entry->resource_entry.original_size))
                {
                        ERROR("Found uncompressed resource with original size "
                              "not the same as compressed size");
@@ -312,6 +370,14 @@ int read_lookup_table(WIMStruct *w)
                        ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
                        goto out_free_cur_entry;
                }
+               if ((cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
+                   && cur_entry->refcnt != 1)
+               {
+                       ERROR("Found metadata resource with refcnt != 1:");
+                       print_lookup_table_entry(cur_entry);
+                       ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
+                       goto out_free_cur_entry;
+               }
                lookup_table_insert(table, cur_entry);
 
        }
@@ -326,7 +392,7 @@ out:
 }
 
 
-/* 
+/*
  * Writes a lookup table entry to the output file.
  */
 int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out)
@@ -357,11 +423,58 @@ int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out)
        return 0;
 }
 
+/* Writes the lookup table to the output file. */
+int write_lookup_table(struct lookup_table *table, FILE *out,
+                      struct resource_entry *out_res_entry)
+{
+       off_t start_offset, end_offset;
+       int ret;
+
+       start_offset = ftello(out);
+       if (start_offset == -1)
+               return WIMLIB_ERR_WRITE;
+
+       ret = for_lookup_table_entry(table, write_lookup_table_entry, out);
+       if (ret != 0)
+               return ret;
+
+       end_offset = ftello(out);
+       if (end_offset == -1)
+               return WIMLIB_ERR_WRITE;
+
+       out_res_entry->offset        = start_offset;
+       out_res_entry->size          = end_offset - start_offset;
+       out_res_entry->original_size = end_offset - start_offset;
+       out_res_entry->flags         = WIM_RESHDR_FLAG_METADATA;
+
+       return 0;
+}
+
+
+int lte_zero_real_refcnt(struct lookup_table_entry *lte, void *ignore)
+{
+       lte->real_refcnt = 0;
+       return 0;
+}
+
+int lte_zero_out_refcnt(struct lookup_table_entry *lte, void *ignore)
+{
+       lte->out_refcnt = 0;
+       return 0;
+}
 
+int lte_zero_extracted_file(struct lookup_table_entry *lte, void *ignore)
+{
+       lte->extracted_file = NULL;
+       return 0;
+}
 
-int zero_out_refcnts(struct lookup_table_entry *entry, void *ignore)
+int lte_free_extracted_file(struct lookup_table_entry *lte, void *ignore)
 {
-       entry->out_refcnt = 0;
+       if (lte->extracted_file != NULL) {
+               FREE(lte->extracted_file);
+               lte->extracted_file = NULL;
+       }
        return 0;
 }
 
@@ -371,11 +484,11 @@ void print_lookup_table_entry(const struct lookup_table_entry *lte)
                putchar('\n');
                return;
        }
-       printf("Offset            = %"PRIu64" bytes\n", 
+       printf("Offset            = %"PRIu64" bytes\n",
               lte->resource_entry.offset);
-       printf("Size              = %"PRIu64" bytes\n", 
+       printf("Size              = %"PRIu64" bytes\n",
               (u64)lte->resource_entry.size);
-       printf("Original size     = %"PRIu64" bytes\n", 
+       printf("Original size     = %"PRIu64" bytes\n",
               lte->resource_entry.original_size);
        printf("Part Number       = %hu\n", lte->part_number);
        printf("Reference Count   = %u\n", lte->refcnt);
@@ -420,16 +533,16 @@ static int do_print_lookup_table_entry(struct lookup_table_entry *lte,
 }
 
 /*
- * Prints the lookup table of a WIM file. 
+ * Prints the lookup table of a WIM file.
  */
 WIMLIBAPI void wimlib_print_lookup_table(WIMStruct *w)
 {
-       for_lookup_table_entry(w->lookup_table, 
+       for_lookup_table_entry(w->lookup_table,
                               do_print_lookup_table_entry,
                               NULL);
 }
 
-/* 
+/*
  * Looks up an entry in the lookup table.
  */
 struct lookup_table_entry *
@@ -439,6 +552,9 @@ __lookup_resource(const struct lookup_table *table, const u8 hash[])
        struct lookup_table_entry *lte;
        struct hlist_node *pos;
 
+       wimlib_assert(table != NULL);
+       wimlib_assert(hash != NULL);
+
        i = *(size_t*)hash % table->capacity;
        hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
                if (hashes_equal(hash, lte->hash))
@@ -446,7 +562,8 @@ __lookup_resource(const struct lookup_table *table, const u8 hash[])
        return NULL;
 }
 
-/* 
+#ifdef WITH_FUSE
+/*
  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
  * given a path name.
  *
@@ -479,7 +596,7 @@ int lookup_resource(WIMStruct *w, const char *path,
        if (!dentry)
                return -ENOENT;
 
-       inode = dentry->inode;
+       inode = dentry->d_inode;
 
        wimlib_assert(inode->resolved);
 
@@ -512,27 +629,47 @@ out:
                *stream_idx_ret = stream_idx;
        return 0;
 }
+#endif
 
-static int inode_resolve_ltes(struct inode *inode, struct lookup_table *table)
+void inode_resolve_ltes(struct inode *inode, struct lookup_table *table)
 {
-       struct lookup_table_entry *lte;
 
-       /* Resolve the default file stream */
-       lte = __lookup_resource(table, inode->hash);
-       inode->lte = lte;
-       inode->resolved = true;
-
-       /* Resolve the alternate data streams */
-       for (u16 i = 0; i < inode->num_ads; i++) {
-               struct ads_entry *cur_entry = inode->ads_entries[i];
+       if (!inode->resolved) {
+               struct lookup_table_entry *lte;
+               /* Resolve the default file stream */
+               lte = __lookup_resource(table, inode->hash);
+               inode->lte = lte;
+               inode->resolved = 1;
+
+               /* Resolve the alternate data streams */
+               for (u16 i = 0; i < inode->num_ads; i++) {
+                       struct ads_entry *cur_entry = &inode->ads_entries[i];
+                       lte = __lookup_resource(table, cur_entry->hash);
+                       cur_entry->lte = lte;
+               }
+       }
+}
 
-               lte = __lookup_resource(table, cur_entry->hash);
-               cur_entry->lte = lte;
+void inode_unresolve_ltes(struct inode *inode)
+{
+       if (inode->resolved) {
+               if (inode->lte)
+                       copy_hash(inode->hash, inode->lte->hash);
+               else
+                       zero_out_hash(inode->hash);
+
+               for (u16 i = 0; i < inode->num_ads; i++) {
+                       if (inode->ads_entries[i].lte)
+                               copy_hash(inode->ads_entries[i].hash,
+                                         inode->ads_entries[i].lte->hash);
+                       else
+                               zero_out_hash(inode->ads_entries[i].hash);
+               }
+               inode->resolved = 0;
        }
-       return 0;
 }
 
-/* Resolve a dentry's lookup table entries 
+/* Resolve a dentry's lookup table entries
  *
  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
  * lookup table) with pointers directly to the lookup table entries.  A circular
@@ -543,26 +680,46 @@ static int inode_resolve_ltes(struct inode *inode, struct lookup_table *table)
  */
 int dentry_resolve_ltes(struct dentry *dentry, void *table)
 {
-       if (dentry->inode->resolved)
-               return 0;
-       else
-               return inode_resolve_ltes(dentry->inode, table);
+       wimlib_assert(dentry->refcnt == 1);
+       inode_resolve_ltes(dentry->d_inode, table);
+       return 0;
 }
 
+int dentry_unresolve_ltes(struct dentry *dentry, void *ignore)
+{
+       inode_unresolve_ltes(dentry->d_inode);
+       return 0;
+}
 
+/*
+ * Returns the lookup table entry for stream @stream_idx of the inode, where
+ * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
+ * corresponds to an alternate data stream.
+ *
+ * This works for both resolved and un-resolved dentries.
+ */
+struct lookup_table_entry *
+inode_stream_lte(const struct inode *inode, unsigned stream_idx,
+                const struct lookup_table *table)
+{
+       if (inode->resolved)
+               return inode_stream_lte_resolved(inode, stream_idx);
+       else
+               return inode_stream_lte_unresolved(inode, stream_idx, table);
+}
 
 
-/* Return the lookup table entry for the unnamed data stream of a inode, or
+/* Return the lookup table entry for the unnamed data stream of an inode, 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 inode 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
- * inode.  So we need to check the alternate data streams too.
+ * data stream in one of the alternate data streams instead of inside the WIM
+ * dentry itself.  So we need to check the alternate data streams too.
  *
- * Also, note that a inode may appear to have than one unnamed stream, but if
- * the SHA1 message digest is all 0's then the corresponding stream does not
+ * Also, note that a dentry may appear to have more 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 inode'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
@@ -570,7 +727,7 @@ int dentry_resolve_ltes(struct dentry *dentry, void *table)
  */
 struct lookup_table_entry *
 inode_unnamed_lte(const struct inode *inode,
-                  const struct lookup_table *table)
+                 const struct lookup_table *table)
 {
        if (inode->resolved)
                return inode_unnamed_lte_resolved(inode);
@@ -578,3 +735,16 @@ inode_unnamed_lte(const struct inode *inode,
                return inode_unnamed_lte_unresolved(inode, table);
 }
 
+static int lte_add_stream_size(struct lookup_table_entry *lte,
+                              void *total_bytes_p)
+{
+       *(u64*)total_bytes_p += lte->resource_entry.size;
+       return 0;
+}
+
+u64 lookup_table_total_stream_size(struct lookup_table *table)
+{
+       u64 total_size = 0;
+       for_lookup_table_entry(table, lte_add_stream_size, &total_size);
+       return total_size;
+}