X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flookup_table.c;h=2e45f8f51299fd25e464c4bd20f35d15ea96d639;hp=6370f03873440edf83e634bc7264a0a5b1d4ca67;hb=2e44f90c21db693058037f83f92ad136c818ce9d;hpb=f8698b9c814a62a117982701b9551f699553b2a4 diff --git a/src/lookup_table.c b/src/lookup_table.c index 6370f038..2e45f8f5 100644 --- a/src/lookup_table.c +++ b/src/lookup_table.c @@ -24,13 +24,23 @@ * along with wimlib; if not, see http://www.gnu.org/licenses/. */ -#include "wimlib_internal.h" -#include "lookup_table.h" -#include "buffer_io.h" -#include +#ifdef HAVE_CONFIG_H +# include "config.h" +#endif + +#include "wimlib/buffer_io.h" +#include "wimlib/error.h" +#include "wimlib/file_io.h" +#include "wimlib/lookup_table.h" +#include "wimlib/metadata.h" +#include "wimlib/paths.h" +#include "wimlib/resource.h" +#include "wimlib/util.h" +#include +#include #ifdef WITH_FUSE -#include +# include /* for unlink() */ #endif struct wim_lookup_table * @@ -39,7 +49,7 @@ new_lookup_table(size_t capacity) struct wim_lookup_table *table; struct hlist_head *array; - table = MALLOC(sizeof(struct wim_lookup_table)); + table = CALLOC(1, sizeof(struct wim_lookup_table)); if (table) { array = CALLOC(capacity, sizeof(array[0])); if (array) { @@ -49,15 +59,15 @@ new_lookup_table(size_t capacity) } else { FREE(table); table = NULL; - ERROR("Failed to allocate memory for lookup table with capacity %zu", - capacity); + ERROR("Failed to allocate memory for lookup table " + "with capacity %zu", capacity); } } return table; } struct wim_lookup_table_entry * -new_lookup_table_entry() +new_lookup_table_entry(void) { struct wim_lookup_table_entry *lte; @@ -78,51 +88,52 @@ clone_lookup_table_entry(const struct wim_lookup_table_entry *old) { struct wim_lookup_table_entry *new; - new = MALLOC(sizeof(*new)); + new = memdup(old, sizeof(struct wim_lookup_table_entry)); if (!new) return NULL; - memcpy(new, old, sizeof(*old)); new->extracted_file = NULL; switch (new->resource_location) { #ifdef __WIN32__ case RESOURCE_WIN32: + case RESOURCE_WIN32_ENCRYPTED: +#else + case RESOURCE_IN_FILE_ON_DISK: #endif +#ifdef WITH_FUSE case RESOURCE_IN_STAGING_FILE: - case RESOURCE_IN_FILE_ON_DISK: BUILD_BUG_ON((void*)&old->file_on_disk != (void*)&old->staging_file_name); - new->staging_file_name = TSTRDUP(old->staging_file_name); - if (!new->staging_file_name) +#endif + new->file_on_disk = TSTRDUP(old->file_on_disk); + if (!new->file_on_disk) goto out_free; break; case RESOURCE_IN_ATTACHED_BUFFER: - new->attached_buffer = MALLOC(wim_resource_size(old)); + new->attached_buffer = memdup(old->attached_buffer, + 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)); + loc = memdup(old->ntfs_loc, sizeof(struct ntfs_location)); if (!loc) goto out_free; - memcpy(loc, old->ntfs_loc, sizeof(*loc)); loc->path = NULL; loc->stream_name = NULL; new->ntfs_loc = loc; loc->path = STRDUP(old->ntfs_loc->path); if (!loc->path) goto out_free; - loc->stream_name = MALLOC((loc->stream_name_nchars + 1) * 2); - if (!loc->stream_name) - goto out_free; - memcpy(loc->stream_name, - old->ntfs_loc->stream_name, - (loc->stream_name_nchars + 1) * 2); + if (loc->stream_name_nchars) { + loc->stream_name = memdup(old->ntfs_loc->stream_name, + loc->stream_name_nchars * 2); + if (!loc->stream_name) + goto out_free; + } } break; #endif @@ -140,14 +151,18 @@ free_lookup_table_entry(struct wim_lookup_table_entry *lte) { if (lte) { switch (lte->resource_location) { - case RESOURCE_IN_STAGING_FILE: - case RESOURCE_IN_ATTACHED_BUFFER: - case RESOURCE_IN_FILE_ON_DISK: -#ifdef __WIN32__ + #ifdef __WIN32__ case RESOURCE_WIN32: -#endif + case RESOURCE_WIN32_ENCRYPTED: + #else + case RESOURCE_IN_FILE_ON_DISK: + #endif + #ifdef WITH_FUSE + case RESOURCE_IN_STAGING_FILE: BUILD_BUG_ON((void*)<e->file_on_disk != (void*)<e->staging_file_name); + #endif + case RESOURCE_IN_ATTACHED_BUFFER: BUILD_BUG_ON((void*)<e->file_on_disk != (void*)<e->attached_buffer); FREE(lte->file_on_disk); @@ -214,7 +229,7 @@ finalize_lte(struct wim_lookup_table_entry *lte) #ifdef WITH_FUSE if (lte->resource_location == RESOURCE_IN_STAGING_FILE) { unlink(lte->staging_file_name); - list_del(<e->staging_list); + list_del(<e->unhashed_list); } #endif free_lookup_table_entry(lte); @@ -231,7 +246,10 @@ lte_decrement_refcnt(struct wim_lookup_table_entry *lte, wimlib_assert(lte != NULL); wimlib_assert(lte->refcnt != 0); if (--lte->refcnt == 0) { - lookup_table_unlink(table, lte); + if (lte->unhashed) + list_del(<e->unhashed_list); + else + lookup_table_unlink(table, lte); #ifdef WITH_FUSE if (lte->num_opened_fds == 0) #endif @@ -266,13 +284,73 @@ for_lookup_table_entry(struct wim_lookup_table *table, { wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)); ret = visitor(lte, arg); - if (ret != 0) + if (ret) return ret; } } return 0; } +int +cmp_streams_by_wim_position(const void *p1, const void *p2) +{ + const struct wim_lookup_table_entry *lte1, *lte2; + lte1 = *(const struct wim_lookup_table_entry**)p1; + lte2 = *(const struct wim_lookup_table_entry**)p2; + if (lte1->resource_entry.offset < lte2->resource_entry.offset) + return -1; + else if (lte1->resource_entry.offset > lte2->resource_entry.offset) + return 1; + else + return 0; +} + + +static int +add_lte_to_array(struct wim_lookup_table_entry *lte, + void *_pp) +{ + struct wim_lookup_table_entry ***pp = _pp; + *(*pp)++ = lte; + return 0; +} + +/* Iterate through the lookup table entries, but first sort them by stream + * offset in the WIM. Caution: this is intended to be used when the stream + * offset field has actually been set. */ +int +for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table, + int (*visitor)(struct wim_lookup_table_entry *, + void *), + void *arg) +{ + struct wim_lookup_table_entry **lte_array, **p; + size_t num_streams = table->num_entries; + int ret; + + lte_array = MALLOC(num_streams * sizeof(lte_array[0])); + if (!lte_array) + return WIMLIB_ERR_NOMEM; + p = lte_array; + for_lookup_table_entry(table, add_lte_to_array, &p); + + wimlib_assert(p == lte_array + num_streams); + + qsort(lte_array, num_streams, sizeof(lte_array[0]), + cmp_streams_by_wim_position); + ret = 0; + for (size_t i = 0; i < num_streams; i++) { + ret = visitor(lte_array[i], arg); + if (ret) + break; + } + FREE(lte_array); + return ret; +} + + +/* Size of each lookup table entry in the WIM file. */ +#define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50 /* * Reads the lookup table from a WIM file. @@ -284,49 +362,55 @@ for_lookup_table_entry(struct wim_lookup_table *table, int read_lookup_table(WIMStruct *w) { - u64 num_entries; - u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE]; int ret; + size_t num_entries; struct wim_lookup_table *table; struct wim_lookup_table_entry *cur_entry, *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; - } + u8 table_buf[(BUFFER_SIZE / WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE) * + WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE]; + const u8 *p; + off_t offset; + size_t buf_entries_remaining; 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) - { - ERROR_WITH_ERRNO("Failed to seek to byte %"PRIu64" to read " - "lookup table", - w->hdr.lookup_table_res_entry.offset); - return WIMLIB_ERR_READ; + 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; } - num_entries = w->hdr.lookup_table_res_entry.original_size / + num_entries = w->hdr.lookup_table_res_entry.size / WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE; table = new_lookup_table(num_entries * 2 + 1); if (!table) return WIMLIB_ERR_NOMEM; w->current_image = 0; - while (num_entries--) { - const u8 *p; - - if (fread(buf, 1, sizeof(buf), w->fp) != sizeof(buf)) { - if (feof(w->fp)) { - ERROR("Unexpected EOF in WIM lookup table!"); - } else { - ERROR_WITH_ERRNO("Error reading WIM lookup " - "table"); + offset = w->hdr.lookup_table_res_entry.offset; + buf_entries_remaining = 0; + for (; num_entries != 0; num_entries--, buf_entries_remaining--) { + if (buf_entries_remaining == 0) { + size_t entries_to_read, bytes_to_read; + + entries_to_read = min(sizeof(table_buf) / + WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE, + num_entries); + bytes_to_read = entries_to_read * + WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE; + if (full_pread(w->in_fd, table_buf, + bytes_to_read, offset) != bytes_to_read) + { + ERROR_WITH_ERRNO("Error reading lookup table " + "(offset=%"PRIu64")", offset); + ret = WIMLIB_ERR_READ; + goto out_free_lookup_table; } - ret = WIMLIB_ERR_READ; - goto out_free_lookup_table; + offset += bytes_to_read; + p = table_buf; + buf_entries_remaining = entries_to_read; } cur_entry = new_lookup_table_entry(); if (!cur_entry) { @@ -336,7 +420,7 @@ 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_resource_entry(p, &cur_entry->resource_entry); 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); @@ -407,7 +491,7 @@ read_lookup_table(WIMStruct *w) w->current_image + 1, cur_entry->resource_entry.offset); w->image_metadata[ - w->current_image++].metadata_lte = cur_entry; + w->current_image++]->metadata_lte = cur_entry; } else { /* Lookup table entry for a stream that is not a * metadata resource */ @@ -428,8 +512,7 @@ read_lookup_table(WIMStruct *w) } } - if (w->hdr.part_number == 1 && - w->current_image != w->hdr.image_count) + if (w->hdr.part_number == 1 && w->current_image != w->hdr.image_count) { ERROR("The WIM header says there are %u images " "in the WIM, but we only found %d metadata " @@ -451,37 +534,72 @@ out: } -/* - * Writes a lookup table entry to the output file. - */ -int -write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_out) +static u8 * +write_lookup_table_entry(const struct wim_lookup_table_entry *lte, u8 *buf_p) { - FILE *out; - u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE]; - u8 *p; - - out = _out; + buf_p = put_resource_entry(buf_p, <e->output_resource_entry); + buf_p = put_u16(buf_p, lte->part_number); + buf_p = put_u32(buf_p, lte->out_refcnt); + buf_p = put_bytes(buf_p, SHA1_HASH_SIZE, lte->hash); + return buf_p; +} - /* Don't write entries that have not had file resources or metadata - * resources written for them. */ - if (lte->out_refcnt == 0) - return 0; +int +write_lookup_table_from_stream_list(struct list_head *stream_list, + int out_fd, + struct resource_entry *out_res_entry) +{ + int ret; + off_t start_offset; + u8 table_buf[(BUFFER_SIZE / WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE) * + WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE]; + u8 *buf_p; + size_t table_size; + size_t bytes_to_write; + struct wim_lookup_table_entry *lte; - if (lte->output_resource_entry.flags & WIM_RESHDR_FLAG_METADATA) { - DEBUG("Writing metadata entry at %"PRIu64" " - "(orig size = %"PRIu64")", - ftello(out), lte->output_resource_entry.original_size); + start_offset = filedes_offset(out_fd); + if (start_offset == -1) + goto write_error; + + buf_p = table_buf; + table_size = 0; + list_for_each_entry(lte, stream_list, lookup_table_list) { + if (buf_p == table_buf + sizeof(table_buf)) { + bytes_to_write = sizeof(table_buf); + if (full_write(out_fd, table_buf, + bytes_to_write) != bytes_to_write) + goto write_error; + table_size += bytes_to_write; + buf_p = table_buf; + } + buf_p = write_lookup_table_entry(lte, buf_p); } - - p = put_resource_entry(buf, <e->output_resource_entry); - p = put_u16(p, lte->part_number); - p = put_u32(p, lte->out_refcnt); - p = put_bytes(p, SHA1_HASH_SIZE, lte->hash); - if (fwrite(buf, 1, sizeof(buf), out) != sizeof(buf)) { - ERROR_WITH_ERRNO("Failed to write lookup table entry"); - return WIMLIB_ERR_WRITE; + bytes_to_write = buf_p - table_buf; + if (bytes_to_write != 0) { + if (full_write(out_fd, table_buf, + bytes_to_write) != bytes_to_write) + goto write_error; + table_size += bytes_to_write; } + out_res_entry->offset = start_offset; + out_res_entry->size = table_size; + out_res_entry->original_size = table_size; + out_res_entry->flags = WIM_RESHDR_FLAG_METADATA; + ret = 0; +out: + return ret; +write_error: + ERROR_WITH_ERRNO("Failed to write lookup table"); + ret = WIMLIB_ERR_WRITE; + goto out; +} + +static int +append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list) +{ + if (lte->out_refcnt != 0) + list_add_tail(<e->lookup_table_list, (struct list_head*)_list); return 0; } @@ -489,14 +607,9 @@ write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_out) int write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry) { - FILE *out = w->out_fp; - off_t start_offset, end_offset; - int ret; - int start_image, end_image; - - start_offset = ftello(out); - if (start_offset == -1) - return WIMLIB_ERR_WRITE; + LIST_HEAD(stream_list); + int start_image; + int end_image; if (image == WIMLIB_ALL_IMAGES) { start_image = 1; @@ -505,49 +618,39 @@ write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry start_image = image; end_image = image; } + for (int i = start_image; i <= end_image; i++) { struct wim_lookup_table_entry *metadata_lte; - metadata_lte = w->image_metadata[i - 1].metadata_lte; + metadata_lte = w->image_metadata[i - 1]->metadata_lte; metadata_lte->out_refcnt = 1; metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA; - ret = write_lookup_table_entry(metadata_lte, out); - if (ret) - return ret; + append_lookup_table_entry(metadata_lte, &stream_list); } - - ret = for_lookup_table_entry(w->lookup_table, write_lookup_table_entry, out); - if (ret) - 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; + for_lookup_table_entry(w->lookup_table, + append_lookup_table_entry, + &stream_list); + return write_lookup_table_from_stream_list(&stream_list, + w->out_fd, + out_res_entry); } - int -lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *ignore) +lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore) { lte->real_refcnt = 0; return 0; } int -lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *ignore) +lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore) { lte->out_refcnt = 0; return 0; } int -lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *ignore) +lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore) { if (lte->extracted_file != NULL) { FREE(lte->extracted_file); @@ -575,9 +678,14 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out) tfprintf(out, T("Part Number = %hu\n"), lte->part_number); tfprintf(out, T("Reference Count = %u\n"), lte->refcnt); - tfprintf(out, T("Hash = 0x")); - print_hash(lte->hash, out); - tputc(T('\n'), out); + if (lte->unhashed) { + tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"), + lte->back_inode, lte->back_stream_id); + } else { + tfprintf(out, T("Hash = 0x")); + print_hash(lte->hash, out); + tputc(T('\n'), out); + } tfprintf(out, T("Flags = ")); u8 flags = lte->resource_entry.flags; @@ -599,15 +707,19 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out) break; #ifdef __WIN32__ case RESOURCE_WIN32: -#endif + case RESOURCE_WIN32_ENCRYPTED: +#else case RESOURCE_IN_FILE_ON_DISK: +#endif tfprintf(out, T("File on Disk = `%"TS"'\n"), lte->file_on_disk); break; +#ifdef WITH_FUSE case RESOURCE_IN_STAGING_FILE: tfprintf(out, T("Staging File = `%"TS"'\n"), lte->staging_file_name); break; +#endif default: break; } @@ -722,6 +834,19 @@ out: } #endif +/* + * XXX Probably should store the compression type directly in the lookup table + * entry + */ +int +wim_resource_compression_type(const struct wim_lookup_table_entry *lte) +{ + if (!(lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED) + || lte->resource_location != RESOURCE_IN_WIM) + return WIMLIB_COMPRESSION_TYPE_NONE; + return wimlib_get_compression_type(lte->wim); +} + /* Resolve an inode's lookup table entries * * This replaces the SHA1 hash fields (which are used to lookup an entry in the @@ -788,6 +913,34 @@ inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx, return inode_stream_lte_unresolved(inode, stream_idx, table); } +struct wim_lookup_table_entry * +inode_unnamed_lte_resolved(const struct wim_inode *inode) +{ + wimlib_assert(inode->i_resolved); + for (unsigned i = 0; i <= inode->i_num_ads; i++) { + if (inode_stream_name_nbytes(inode, i) == 0 && + !is_zero_hash(inode_stream_hash_resolved(inode, i))) + { + return inode_stream_lte_resolved(inode, i); + } + } + return NULL; +} + +struct wim_lookup_table_entry * +inode_unnamed_lte_unresolved(const struct wim_inode *inode, + const struct wim_lookup_table *table) +{ + wimlib_assert(!inode->i_resolved); + for (unsigned i = 0; i <= inode->i_num_ads; i++) { + if (inode_stream_name_nbytes(inode, i) == 0 && + !is_zero_hash(inode_stream_hash_unresolved(inode, i))) + { + return inode_stream_lte_unresolved(inode, i, table); + } + } + return NULL; +} /* Return the lookup table entry for the unnamed data stream of an inode, or * NULL if there is none. @@ -829,3 +982,79 @@ lookup_table_total_stream_size(struct wim_lookup_table *table) for_lookup_table_entry(table, lte_add_stream_size, &total_size); return total_size; } + +struct wim_lookup_table_entry ** +retrieve_lte_pointer(struct wim_lookup_table_entry *lte) +{ + wimlib_assert(lte->unhashed); + struct wim_inode *inode = lte->back_inode; + u32 stream_id = lte->back_stream_id; + if (stream_id == 0) + return &inode->i_lte; + else + for (u16 i = 0; i < inode->i_num_ads; i++) + if (inode->i_ads_entries[i].stream_id == stream_id) + return &inode->i_ads_entries[i].lte; + wimlib_assert(0); + return NULL; +} + +/* Calculate the SHA1 message digest of a stream and move it from the list of + * unhashed streams to the stream lookup table, possibly joining it with an + * existing lookup table entry for an identical stream. + * + * @lte: An unhashed lookup table entry. + * @lookup_table: Lookup table for the WIM. + * @lte_ret: On success, write a pointer to the resulting lookup table + * entry to this location. This will be the same as @lte + * if it was inserted into the lookup table, or different if + * a duplicate stream was found. + * + * Returns 0 on success; nonzero if there is an error reading the stream. + */ +int +hash_unhashed_stream(struct wim_lookup_table_entry *lte, + struct wim_lookup_table *lookup_table, + struct wim_lookup_table_entry **lte_ret) +{ + int ret; + struct wim_lookup_table_entry *duplicate_lte; + struct wim_lookup_table_entry **back_ptr; + + wimlib_assert(lte->unhashed); + + /* back_ptr must be saved because @back_inode and @back_stream_id are in + * union with the SHA1 message digest and will no longer be valid once + * the SHA1 has been calculated. */ + back_ptr = retrieve_lte_pointer(lte); + + ret = sha1_resource(lte); + if (ret) + return ret; + + /* Look for a duplicate stream */ + duplicate_lte = __lookup_resource(lookup_table, lte->hash); + list_del(<e->unhashed_list); + if (duplicate_lte) { + /* We have a duplicate stream. Transfer the reference counts + * from this stream to the duplicate, update the reference to + * this stream (in an inode or ads_entry) to point to the + * duplicate, then free this stream. */ + wimlib_assert(!(duplicate_lte->unhashed)); + duplicate_lte->refcnt += lte->refcnt; + duplicate_lte->out_refcnt += lte->refcnt; + *back_ptr = duplicate_lte; + free_lookup_table_entry(lte); + lte = duplicate_lte; + } else { + /* No duplicate stream, so we need to insert + * this stream into the lookup table and treat + * it as a hashed stream. */ + lookup_table_insert(lookup_table, lte); + lte->unhashed = 0; + } + if (lte_ret) + *lte_ret = lte; + return 0; +} +