]> wimlib.net Git - wimlib/blob - src/write.c
22b1c9682c08e2509d072e907fa8abc0c6e522d2
[wimlib] / src / write.c
1 /*
2  * write.c
3  *
4  * Support for writing WIM files; write a WIM file, overwrite a WIM file, write
5  * compressed file resources, etc.
6  */
7
8 /*
9  * Copyright (C) 2012, 2013 Eric Biggers
10  *
11  * This file is part of wimlib, a library for working with WIM files.
12  *
13  * wimlib is free software; you can redistribute it and/or modify it under the
14  * terms of the GNU General Public License as published by the Free
15  * Software Foundation; either version 3 of the License, or (at your option)
16  * any later version.
17  *
18  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
19  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
20  * A PARTICULAR PURPOSE. See the GNU General Public License for more
21  * details.
22  *
23  * You should have received a copy of the GNU General Public License
24  * along with wimlib; if not, see http://www.gnu.org/licenses/.
25  */
26
27 #ifdef HAVE_CONFIG_H
28 #  include "config.h"
29 #endif
30
31 #if defined(HAVE_SYS_FILE_H) && defined(HAVE_FLOCK)
32 /* On BSD, this should be included before "wimlib/list.h" so that "wimlib/list.h" can
33  * overwrite the LIST_HEAD macro. */
34 #  include <sys/file.h>
35 #endif
36
37 #include "wimlib/chunk_compressor.h"
38 #include "wimlib/endianness.h"
39 #include "wimlib/error.h"
40 #include "wimlib/file_io.h"
41 #include "wimlib/header.h"
42 #include "wimlib/integrity.h"
43 #include "wimlib/lookup_table.h"
44 #include "wimlib/metadata.h"
45 #include "wimlib/resource.h"
46 #ifdef __WIN32__
47 #  include "wimlib/win32.h" /* win32_rename_replacement() */
48 #endif
49 #include "wimlib/write.h"
50 #include "wimlib/xml.h"
51
52 #include <errno.h>
53 #include <fcntl.h>
54 #include <stdlib.h>
55 #include <unistd.h>
56
57 #ifdef HAVE_ALLOCA_H
58 #  include <alloca.h>
59 #endif
60
61 /* wimlib internal flags used when writing resources.  */
62 #define WRITE_RESOURCE_FLAG_RECOMPRESS          0x00000001
63 #define WRITE_RESOURCE_FLAG_PIPABLE             0x00000002
64 #define WRITE_RESOURCE_FLAG_PACK_STREAMS        0x00000004
65
66 static inline int
67 write_flags_to_resource_flags(int write_flags)
68 {
69         int write_resource_flags = 0;
70
71         if (write_flags & WIMLIB_WRITE_FLAG_RECOMPRESS)
72                 write_resource_flags |= WRITE_RESOURCE_FLAG_RECOMPRESS;
73         if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
74                 write_resource_flags |= WRITE_RESOURCE_FLAG_PIPABLE;
75         if (write_flags & WIMLIB_WRITE_FLAG_PACK_STREAMS)
76                 write_resource_flags |= WRITE_RESOURCE_FLAG_PACK_STREAMS;
77         return write_resource_flags;
78 }
79
80 struct filter_context {
81         int write_flags;
82         WIMStruct *wim;
83 };
84
85 /* Determine specified stream should be filtered out from the write.
86  *
87  * Return values:
88  *
89  *  < 0 : The stream should be hard-filtered; that is, not included in the
90  *        output WIM at all.
91  *    0 : The stream should not be filtered out.
92  *  > 0 : The stream should be soft-filtered; that is, it already exists in the
93  *        WIM file and may not need to be written again.
94  */
95 static int
96 stream_filtered(const struct wim_lookup_table_entry *lte,
97                 const struct filter_context *ctx)
98 {
99         int write_flags = ctx->write_flags;
100         WIMStruct *wim = ctx->wim;
101
102         if (ctx == NULL)
103                 return 0;
104
105         if (write_flags & WIMLIB_WRITE_FLAG_OVERWRITE &&
106             lte->resource_location == RESOURCE_IN_WIM &&
107             lte->rspec->wim == wim)
108                 return 1;
109
110         if (write_flags & WIMLIB_WRITE_FLAG_SKIP_EXTERNAL_WIMS &&
111             lte->resource_location == RESOURCE_IN_WIM &&
112             lte->rspec->wim != wim)
113                 return -1;
114
115         return 0;
116 }
117
118 static bool
119 stream_hard_filtered(const struct wim_lookup_table_entry *lte,
120                      struct filter_context *ctx)
121 {
122         return stream_filtered(lte, ctx) < 0;
123 }
124
125 static inline int
126 may_soft_filter_streams(const struct filter_context *ctx)
127 {
128         if (ctx == NULL)
129                 return 0;
130         return ctx->write_flags & WIMLIB_WRITE_FLAG_OVERWRITE;
131 }
132
133 static inline int
134 may_hard_filter_streams(const struct filter_context *ctx)
135 {
136         if (ctx == NULL)
137                 return 0;
138         return ctx->write_flags & WIMLIB_WRITE_FLAG_SKIP_EXTERNAL_WIMS;
139 }
140
141 static inline int
142 may_filter_streams(const struct filter_context *ctx)
143 {
144         return (may_soft_filter_streams(ctx) ||
145                 may_hard_filter_streams(ctx));
146 }
147
148
149 /* Return true if the specified resource is compressed and the compressed data
150  * can be reused with the specified output parameters.  */
151 static bool
152 can_raw_copy(const struct wim_lookup_table_entry *lte,
153              int write_resource_flags, int out_ctype, u32 out_chunk_size)
154 {
155         const struct wim_resource_spec *rspec;
156
157         if (write_resource_flags & WRITE_RESOURCE_FLAG_RECOMPRESS)
158                 return false;
159
160         if (out_ctype == WIMLIB_COMPRESSION_TYPE_NONE)
161                 return false;
162
163         if (lte->resource_location != RESOURCE_IN_WIM)
164                 return false;
165
166         rspec = lte->rspec;
167
168         if (rspec->is_pipable != !!(write_resource_flags & WRITE_RESOURCE_FLAG_PIPABLE))
169                 return false;
170
171
172         if (rspec->flags & WIM_RESHDR_FLAG_COMPRESSED) {
173                 /* Normal compressed resource: Must use same compression type
174                  * and chunk size.  */
175                 return (rspec->wim->compression_type == out_ctype &&
176                         rspec->wim->chunk_size == out_chunk_size);
177         }
178
179         /* XXX: For compatibility, we can't allow multiple packed resources per
180          * WIM.  */
181 #if 0
182         if ((rspec->flags & WIM_RESHDR_FLAG_PACKED_STREAMS) &&
183             (write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS))
184         {
185                 /* Packed resource: Such resources may contain multiple streams,
186                  * and in general only a subset of them need to be written.  As
187                  * a heuristic, re-use the raw data if at least half the
188                  * uncompressed size is being written.  */
189
190                 /* Note: packed resources contain a header that specifies the
191                  * compression type and chunk size; therefore we don't need to
192                  * check if they are compatible with @out_ctype and
193                  * @out_chunk_size.  */
194
195                 struct wim_lookup_table_entry *res_stream;
196                 u64 write_size = 0;
197
198                 list_for_each_entry(res_stream, &rspec->stream_list, rspec_node)
199                         if (res_stream->will_be_in_output_wim)
200                                 write_size += res_stream->size;
201
202                 return (write_size > rspec->uncompressed_size / 2);
203         }
204 #endif
205
206         return false;
207 }
208
209 static u8
210 filter_resource_flags(u8 flags)
211 {
212         return (flags & ~(WIM_RESHDR_FLAG_PACKED_STREAMS |
213                           WIM_RESHDR_FLAG_COMPRESSED |
214                           WIM_RESHDR_FLAG_SPANNED |
215                           WIM_RESHDR_FLAG_FREE));
216 }
217
218 static void
219 stream_set_out_reshdr_for_reuse(struct wim_lookup_table_entry *lte)
220 {
221         const struct wim_resource_spec *rspec;
222
223         wimlib_assert(lte->resource_location == RESOURCE_IN_WIM);
224         rspec = lte->rspec;
225
226         if (rspec->flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
227
228                 wimlib_assert(lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS);
229
230                 lte->out_reshdr.offset_in_wim = lte->offset_in_res;
231                 lte->out_reshdr.uncompressed_size = 0;
232                 lte->out_reshdr.size_in_wim = lte->size;
233
234                 lte->out_res_offset_in_wim = rspec->offset_in_wim;
235                 lte->out_res_size_in_wim = rspec->size_in_wim;
236                 /*lte->out_res_uncompressed_size = rspec->uncompressed_size;*/
237         } else {
238                 wimlib_assert(!(lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS));
239
240                 lte->out_reshdr.offset_in_wim = rspec->offset_in_wim;
241                 lte->out_reshdr.uncompressed_size = rspec->uncompressed_size;
242                 lte->out_reshdr.size_in_wim = rspec->size_in_wim;
243         }
244         lte->out_reshdr.flags = lte->flags;
245 }
246
247
248 /* Write the header for a stream in a pipable WIM.  */
249 static int
250 write_pwm_stream_header(const struct wim_lookup_table_entry *lte,
251                         struct filedes *out_fd,
252                         int additional_reshdr_flags)
253 {
254         struct pwm_stream_hdr stream_hdr;
255         u32 reshdr_flags;
256         int ret;
257
258         stream_hdr.magic = cpu_to_le64(PWM_STREAM_MAGIC);
259         stream_hdr.uncompressed_size = cpu_to_le64(lte->size);
260         if (additional_reshdr_flags & PWM_RESHDR_FLAG_UNHASHED) {
261                 zero_out_hash(stream_hdr.hash);
262         } else {
263                 wimlib_assert(!lte->unhashed);
264                 copy_hash(stream_hdr.hash, lte->hash);
265         }
266
267         reshdr_flags = filter_resource_flags(lte->flags);
268         reshdr_flags |= additional_reshdr_flags;
269         stream_hdr.flags = cpu_to_le32(reshdr_flags);
270         ret = full_write(out_fd, &stream_hdr, sizeof(stream_hdr));
271         if (ret)
272                 ERROR_WITH_ERRNO("Write error");
273         return ret;
274 }
275
276 struct write_streams_progress_data {
277         wimlib_progress_func_t progress_func;
278         union wimlib_progress_info progress;
279         uint64_t next_progress;
280         WIMStruct *prev_wim_part;
281 };
282
283 static void
284 do_write_streams_progress(struct write_streams_progress_data *progress_data,
285                           u64 size,
286                           bool discarded,
287                           struct wim_lookup_table_entry *cur_stream)
288 {
289         union wimlib_progress_info *progress = &progress_data->progress;
290         bool new_wim_part;
291
292         if (discarded) {
293                 progress->write_streams.total_bytes -= size;
294                 if (progress_data->next_progress != ~(uint64_t)0 &&
295                     progress_data->next_progress > progress->write_streams.total_bytes)
296                 {
297                         progress_data->next_progress = progress->write_streams.total_bytes;
298                 }
299         } else {
300                 progress->write_streams.completed_bytes += size;
301         }
302         new_wim_part = false;
303         if (cur_stream->resource_location == RESOURCE_IN_WIM &&
304             cur_stream->rspec->wim != progress_data->prev_wim_part)
305         {
306                 if (progress_data->prev_wim_part) {
307                         new_wim_part = true;
308                         progress->write_streams.completed_parts++;
309                 }
310                 progress_data->prev_wim_part = cur_stream->rspec->wim;
311         }
312         progress->write_streams.completed_streams++;
313         if (progress_data->progress_func
314             && (progress->write_streams.completed_bytes >= progress_data->next_progress
315                 || new_wim_part))
316         {
317                 progress_data->progress_func(WIMLIB_PROGRESS_MSG_WRITE_STREAMS,
318                                              progress);
319                 if (progress_data->next_progress == progress->write_streams.total_bytes) {
320                         progress_data->next_progress = ~(uint64_t)0;
321                 } else {
322                         progress_data->next_progress =
323                                 min(progress->write_streams.total_bytes,
324                                     progress->write_streams.completed_bytes +
325                                         progress->write_streams.total_bytes / 100);
326                 }
327         }
328 }
329
330 struct write_streams_ctx {
331         /* File descriptor the streams are being written to.  */
332         struct filedes *out_fd;
333
334         /* Lookup table for the WIMStruct on whose behalf the streams are being
335          * written.  */
336         struct wim_lookup_table *lookup_table;
337
338         /* Compression format to use.  */
339         int out_ctype;
340
341         /* Maximum uncompressed chunk size in compressed resources to use.  */
342         u32 out_chunk_size;
343
344         /* Flags that affect how the streams will be written.  */
345         int write_resource_flags;
346
347         /* Data used for issuing WRITE_STREAMS progress.  */
348         struct write_streams_progress_data progress_data;
349
350         struct filter_context *filter_ctx;
351
352         /* Upper bound on the total number of bytes that need to be compressed.
353          * */
354         u64 num_bytes_to_compress;
355
356         /* Pointer to the chunk_compressor implementation being used for
357          * compressing chunks of data, or NULL if chunks are being written
358          * uncompressed.  */
359         struct chunk_compressor *compressor;
360
361         /* Buffer for dividing the read data into chunks of size
362          * @out_chunk_size.  */
363         u8 *chunk_buf;
364
365         /* Number of bytes in @chunk_buf that are currently filled.  */
366         size_t chunk_buf_filled;
367
368         /* List of streams that currently have chunks being compressed.  */
369         struct list_head pending_streams;
370
371         /* Set to true if the stream currently being read was a duplicate, and
372          * therefore the corresponding stream entry needs to be freed once the
373          * read finishes.  (In this case we add the duplicate entry to
374          * pending_streams rather than the entry being read.)  */
375         bool stream_was_duplicate;
376
377         /* Current uncompressed offset in the resource being read.  */
378         u64 cur_read_res_offset;
379
380         /* Uncompressed size of the resource currently being read.  */
381         u64 cur_read_res_size;
382
383         /* Current uncompressed offset in the resource being written.  */
384         u64 cur_write_res_offset;
385
386         /* Uncompressed size of resource currently being written.  */
387         u64 cur_write_res_size;
388
389         /* Array that is filled in with compressed chunk sizes as a resource is
390          * being written.  */
391         u64 *chunk_csizes;
392
393         /* Index of next entry in @chunk_csizes to fill in.  */
394         size_t chunk_index;
395
396         /* Number of entries in @chunk_csizes currently allocated.  */
397         size_t num_alloc_chunks;
398
399         /* Offset in the output file of the start of the chunks of the resource
400          * currently being written.  */
401         u64 chunks_start_offset;
402 };
403
404 static u64
405 get_chunk_entry_size(u64 res_size, int write_resource_flags)
406 {
407         if (res_size <= UINT32_MAX ||
408             (write_resource_flags & WIM_RESHDR_FLAG_PACKED_STREAMS))
409                 return 4;
410         else
411                 return 8;
412 }
413
414 /* Reserve space for the chunk table and prepare to accumulate the chunk table
415  * in memory.  */
416 static int
417 begin_chunk_table(struct write_streams_ctx *ctx, u64 res_expected_size)
418 {
419         u64 expected_num_chunks;
420         u64 expected_num_chunk_entries;
421         size_t reserve_size;
422         int ret;
423
424         /* Calculate the number of chunks and chunk entries that should be
425          * needed for the resource.  These normally will be the final values,
426          * but in PACKED_STREAMS mode some of the streams we're planning to
427          * write into the resource may be duplicates, and therefore discarded,
428          * potentially decreasing the number of chunk entries needed.  */
429         expected_num_chunks = DIV_ROUND_UP(res_expected_size, ctx->out_chunk_size);
430         expected_num_chunk_entries = expected_num_chunks;
431         if (!(ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS))
432                 expected_num_chunk_entries--;
433
434         /* Make sure the chunk_csizes array is long enough to store the
435          * compressed size of each chunk.  */
436         if (expected_num_chunks > ctx->num_alloc_chunks) {
437                 u64 new_length = expected_num_chunks + 50;
438
439                 if ((size_t)new_length != new_length) {
440                         ERROR("Resource size too large (%"PRIu64" bytes!",
441                               res_expected_size);
442                         return WIMLIB_ERR_NOMEM;
443                 }
444
445                 FREE(ctx->chunk_csizes);
446                 ctx->chunk_csizes = MALLOC(new_length * sizeof(ctx->chunk_csizes[0]));
447                 if (ctx->chunk_csizes == NULL) {
448                         ctx->num_alloc_chunks = 0;
449                         return WIMLIB_ERR_NOMEM;
450                 }
451                 ctx->num_alloc_chunks = new_length;
452         }
453
454         ctx->chunk_index = 0;
455
456         if (!(ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PIPABLE)) {
457                 /* Reserve space for the chunk table in the output file.  In the
458                  * case of packed resources this reserves the upper bound for
459                  * the needed space, not necessarily the exact space which will
460                  * prove to be needed.  At this point, we just use @chunk_csizes
461                  * for a buffer of 0's because the actual compressed chunk sizes
462                  * are unknown.  */
463                 reserve_size = expected_num_chunk_entries *
464                                get_chunk_entry_size(res_expected_size,
465                                                     ctx->write_resource_flags);
466                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS)
467                         reserve_size += sizeof(struct alt_chunk_table_header_disk);
468                 memset(ctx->chunk_csizes, 0, reserve_size);
469                 ret = full_write(ctx->out_fd, ctx->chunk_csizes, reserve_size);
470                 if (ret)
471                         return ret;
472         }
473         return 0;
474 }
475
476 static int
477 begin_write_resource(struct write_streams_ctx *ctx, u64 res_expected_size)
478 {
479         int ret;
480
481         wimlib_assert(res_expected_size != 0);
482
483         if (ctx->compressor != NULL) {
484                 ret = begin_chunk_table(ctx, res_expected_size);
485                 if (ret)
486                         return ret;
487         }
488
489         /* Output file descriptor is now positioned at the offset at which to
490          * write the first chunk of the resource.  */
491         ctx->chunks_start_offset = ctx->out_fd->offset;
492         ctx->cur_write_res_offset = 0;
493         ctx->cur_write_res_size = res_expected_size;
494         return 0;
495 }
496
497 static int
498 end_chunk_table(struct write_streams_ctx *ctx, u64 res_actual_size,
499                 u64 *res_start_offset_ret, u64 *res_store_size_ret)
500 {
501         size_t actual_num_chunks;
502         size_t actual_num_chunk_entries;
503         size_t chunk_entry_size;
504         int ret;
505
506         actual_num_chunks = ctx->chunk_index;
507         actual_num_chunk_entries = actual_num_chunks;
508         if (!(ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS))
509                 actual_num_chunk_entries--;
510
511         chunk_entry_size = get_chunk_entry_size(res_actual_size,
512                                                 ctx->write_resource_flags);
513
514         typedef le64 __attribute__((may_alias)) aliased_le64_t;
515         typedef le32 __attribute__((may_alias)) aliased_le32_t;
516
517         if (chunk_entry_size == 4) {
518                 aliased_le32_t *entries = (aliased_le32_t*)ctx->chunk_csizes;
519
520                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS) {
521                         for (size_t i = 0; i < actual_num_chunk_entries; i++)
522                                 entries[i] = cpu_to_le32(ctx->chunk_csizes[i]);
523                 } else {
524                         u32 offset = ctx->chunk_csizes[0];
525                         for (size_t i = 0; i < actual_num_chunk_entries; i++) {
526                                 u32 next_size = ctx->chunk_csizes[i + 1];
527                                 entries[i] = cpu_to_le32(offset);
528                                 offset += next_size;
529                         }
530                 }
531         } else {
532                 aliased_le64_t *entries = (aliased_le64_t*)ctx->chunk_csizes;
533
534                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS) {
535                         for (size_t i = 0; i < actual_num_chunk_entries; i++)
536                                 entries[i] = cpu_to_le64(ctx->chunk_csizes[i]);
537                 } else {
538                         u64 offset = ctx->chunk_csizes[0];
539                         for (size_t i = 0; i < actual_num_chunk_entries; i++) {
540                                 u64 next_size = ctx->chunk_csizes[i + 1];
541                                 entries[i] = cpu_to_le64(offset);
542                                 offset += next_size;
543                         }
544                 }
545         }
546
547         size_t chunk_table_size = actual_num_chunk_entries * chunk_entry_size;
548         u64 res_start_offset;
549         u64 res_end_offset;
550
551         if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PIPABLE) {
552                 ret = full_write(ctx->out_fd, ctx->chunk_csizes, chunk_table_size);
553                 if (ret)
554                         goto error;
555                 res_end_offset = ctx->out_fd->offset;
556                 res_start_offset = ctx->chunks_start_offset;
557         } else {
558                 res_end_offset = ctx->out_fd->offset;
559
560                 u64 chunk_table_offset;
561
562                 chunk_table_offset = ctx->chunks_start_offset - chunk_table_size;
563
564                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS) {
565                         struct alt_chunk_table_header_disk hdr;
566
567                         hdr.res_usize = cpu_to_le64(res_actual_size);
568                         hdr.chunk_size = cpu_to_le32(ctx->out_chunk_size);
569                         hdr.compression_format = cpu_to_le32(ctx->out_ctype);
570
571                         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_LZX != 1);
572                         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_XPRESS != 2);
573                         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_LZMS != 3);
574
575                         ret = full_pwrite(ctx->out_fd, &hdr, sizeof(hdr),
576                                           chunk_table_offset - sizeof(hdr));
577                         if (ret)
578                                 goto error;
579                         res_start_offset = chunk_table_offset - sizeof(hdr);
580                 } else {
581                         res_start_offset = chunk_table_offset;
582                 }
583
584                 ret = full_pwrite(ctx->out_fd, ctx->chunk_csizes,
585                                   chunk_table_size, chunk_table_offset);
586                 if (ret)
587                         goto error;
588         }
589
590         *res_start_offset_ret = res_start_offset;
591         *res_store_size_ret = res_end_offset - res_start_offset;
592
593         return 0;
594
595 error:
596         ERROR_WITH_ERRNO("Write error");
597         return ret;
598 }
599
600 /* Finish writing a WIM resource by writing or updating the chunk table (if not
601  * writing the data uncompressed) and loading its metadata into @out_reshdr.  */
602 static int
603 end_write_resource(struct write_streams_ctx *ctx, struct wim_reshdr *out_reshdr)
604 {
605         int ret;
606         u64 res_size_in_wim;
607         u64 res_uncompressed_size;
608         u64 res_offset_in_wim;
609
610         wimlib_assert(ctx->cur_write_res_size == ctx->cur_write_res_offset);
611         res_uncompressed_size = ctx->cur_write_res_size;
612
613         if (ctx->compressor) {
614                 ret = end_chunk_table(ctx, res_uncompressed_size,
615                                       &res_offset_in_wim, &res_size_in_wim);
616                 if (ret)
617                         return ret;
618         } else {
619                 res_offset_in_wim = ctx->chunks_start_offset;
620                 res_size_in_wim = ctx->out_fd->offset - res_offset_in_wim;
621         }
622         out_reshdr->uncompressed_size = res_uncompressed_size;
623         out_reshdr->size_in_wim = res_size_in_wim;
624         out_reshdr->offset_in_wim = res_offset_in_wim;
625         DEBUG("Finished writing resource: %"PRIu64" => %"PRIu64" @ %"PRIu64"",
626               res_uncompressed_size, res_size_in_wim, res_offset_in_wim);
627         return 0;
628 }
629
630 /* Begin processing a stream for writing.  */
631 static int
632 write_stream_begin_read(struct wim_lookup_table_entry *lte,
633                         bool is_partial_res, void *_ctx)
634 {
635         struct write_streams_ctx *ctx = _ctx;
636         int ret;
637
638         wimlib_assert(lte->size > 0);
639
640         ctx->cur_read_res_offset = 0;
641         ctx->cur_read_res_size = lte->size;
642
643         /* As an optimization, we allow some streams to be "unhashed", meaning
644          * their SHA1 message digests are unknown.  This is the case with
645          * streams that are added by scanning a directry tree with
646          * wimlib_add_image(), for example.  Since WIM uses single-instance
647          * streams, we don't know whether such each such stream really need to
648          * written until it is actually checksummed, unless it has a unique
649          * size.  In such cases we read and checksum the stream in this
650          * function, thereby advancing ahead of read_stream_list(), which will
651          * still provide the data again to write_stream_process_chunk().  This
652          * is okay because an unhashed stream cannot be in a WIM resource, which
653          * might be costly to decompress.  */
654         ctx->stream_was_duplicate = false;
655         if (ctx->lookup_table != NULL && lte->unhashed && !lte->unique_size) {
656
657                 wimlib_assert(!is_partial_res);
658
659                 struct wim_lookup_table_entry *lte_new;
660
661                 ret = hash_unhashed_stream(lte, ctx->lookup_table, &lte_new);
662                 if (ret)
663                         return ret;
664                 if (lte_new != lte) {
665                         /* Duplicate stream detected.  */
666
667                         if (lte_new->will_be_in_output_wim ||
668                             stream_filtered(lte_new, ctx->filter_ctx))
669                         {
670                                 /* The duplicate stream is already being
671                                  * included in the output WIM, or it would be
672                                  * filtered out if it had been.  Skip writing
673                                  * this stream (and reading it again) entirely,
674                                  * passing its output reference count to the
675                                  * duplicate stream in the former case.  */
676                                 DEBUG("Discarding duplicate stream of "
677                                       "length %"PRIu64, lte->size);
678                                 do_write_streams_progress(&ctx->progress_data,
679                                                           lte->size, true, lte);
680                                 list_del(&lte->write_streams_list);
681                                 list_del(&lte->lookup_table_list);
682                                 if (lte_new->will_be_in_output_wim)
683                                         lte_new->out_refcnt += lte->out_refcnt;
684                                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS)
685                                         ctx->cur_write_res_size -= lte->size;
686                                 free_lookup_table_entry(lte);
687                                 return BEGIN_STREAM_STATUS_SKIP_STREAM;
688                         } else {
689                                 /* The duplicate stream can validly be written,
690                                  * but was not marked as such.  Discard the
691                                  * current stream entry and use the duplicate,
692                                  * but actually freeing the current entry must
693                                  * wait until read_stream_list() has finished
694                                  * reading its data.  */
695                                 DEBUG("Stream duplicate, but not already "
696                                       "selected for writing.");
697                                 list_replace(&lte->write_streams_list,
698                                              &lte_new->write_streams_list);
699                                 list_replace(&lte->lookup_table_list,
700                                              &lte_new->lookup_table_list);
701                                 lte_new->out_refcnt = lte->out_refcnt;
702                                 lte_new->will_be_in_output_wim = 1;
703                                 ctx->stream_was_duplicate = true;
704                                 lte = lte_new;
705                         }
706                 }
707         }
708         list_move_tail(&lte->write_streams_list, &ctx->pending_streams);
709         return 0;
710 }
711
712 static int
713 write_stream_uncompressed(struct wim_lookup_table_entry *lte,
714                           struct filedes *out_fd)
715 {
716         int ret;
717
718         if (-1 == lseek(out_fd->fd, lte->out_reshdr.offset_in_wim, SEEK_SET) ||
719             0 != ftruncate(out_fd->fd, lte->out_reshdr.offset_in_wim))
720         {
721                 ERROR_WITH_ERRNO("Can't truncate output file to "
722                                  "offset %"PRIu64, lte->out_reshdr.offset_in_wim);
723                 return WIMLIB_ERR_WRITE;
724         }
725
726         out_fd->offset = lte->out_reshdr.offset_in_wim;
727
728         ret = extract_stream_to_fd(lte, out_fd, lte->size);
729         if (ret)
730                 return ret;
731
732         wimlib_assert(out_fd->offset - lte->out_reshdr.offset_in_wim == lte->size);
733         lte->out_reshdr.size_in_wim = lte->size;
734         lte->out_reshdr.flags &= ~(WIM_RESHDR_FLAG_COMPRESSED |
735                                    WIM_RESHDR_FLAG_PACKED_STREAMS);
736         return 0;
737 }
738
739 /* Write the next chunk of (typically compressed) data to the output WIM,
740  * handling the writing of the chunk table.  */
741 static int
742 write_chunk(struct write_streams_ctx *ctx, const void *cchunk,
743             size_t csize, size_t usize)
744 {
745         int ret;
746
747         struct wim_lookup_table_entry *lte;
748
749         lte = list_entry(ctx->pending_streams.next,
750                          struct wim_lookup_table_entry, write_streams_list);
751
752         if (ctx->cur_write_res_offset == 0 &&
753             !(ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS))
754         {
755                 /* Starting to write a new stream in non-packed mode.  */
756
757                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PIPABLE) {
758                         int additional_reshdr_flags = 0;
759                         if (ctx->compressor != NULL)
760                                 additional_reshdr_flags |= WIM_RESHDR_FLAG_COMPRESSED;
761
762                         DEBUG("Writing pipable WIM stream header "
763                               "(offset=%"PRIu64")", ctx->out_fd->offset);
764
765                         ret = write_pwm_stream_header(lte, ctx->out_fd,
766                                                       additional_reshdr_flags);
767                         if (ret)
768                                 return ret;
769                 }
770
771                 ret = begin_write_resource(ctx, lte->size);
772                 if (ret)
773                         return ret;
774         }
775
776         if (ctx->compressor != NULL) {
777                 /* Record the compresed chunk size.  */
778                 wimlib_assert(ctx->chunk_index < ctx->num_alloc_chunks);
779                 ctx->chunk_csizes[ctx->chunk_index++] = csize;
780
781                /* If writing a pipable WIM, before the chunk data write a chunk
782                 * header that provides the compressed chunk size.  */
783                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PIPABLE) {
784                         struct pwm_chunk_hdr chunk_hdr = {
785                                 .compressed_size = cpu_to_le32(csize),
786                         };
787                         ret = full_write(ctx->out_fd, &chunk_hdr,
788                                          sizeof(chunk_hdr));
789                         if (ret)
790                                 goto error;
791                 }
792         }
793
794         /* Write the chunk data.  */
795         ret = full_write(ctx->out_fd, cchunk, csize);
796         if (ret)
797                 goto error;
798
799         ctx->cur_write_res_offset += usize;
800
801         do_write_streams_progress(&ctx->progress_data,
802                                   usize, false, lte);
803
804         if (ctx->cur_write_res_offset == ctx->cur_write_res_size &&
805             !(ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS))
806         {
807                 wimlib_assert(ctx->cur_write_res_offset == lte->size);
808
809                 /* Finished writing a stream in non-packed mode.  */
810
811                 ret = end_write_resource(ctx, &lte->out_reshdr);
812                 if (ret)
813                         return ret;
814
815                 lte->out_reshdr.flags = filter_resource_flags(lte->flags);
816                 if (ctx->compressor != NULL)
817                         lte->out_reshdr.flags |= WIM_RESHDR_FLAG_COMPRESSED;
818
819                 if (ctx->compressor != NULL &&
820                     lte->out_reshdr.size_in_wim >= lte->out_reshdr.uncompressed_size &&
821                     !(ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PIPABLE) &&
822                     !(lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS))
823                 {
824                         /* Stream did not compress to less than its original
825                          * size.  If we're not writing a pipable WIM (which
826                          * could mean the output file descriptor is
827                          * non-seekable), and the stream isn't located in a
828                          * resource pack (which would make reading it again
829                          * costly), truncate the file to the start of the stream
830                          * and write it uncompressed instead.  */
831                         DEBUG("Stream of size %"PRIu64" did not compress to "
832                               "less than original size; writing uncompressed.",
833                               lte->size);
834                         ret = write_stream_uncompressed(lte, ctx->out_fd);
835                         if (ret)
836                                 return ret;
837                 }
838
839                 wimlib_assert(lte->out_reshdr.uncompressed_size == lte->size);
840
841                 list_del(&lte->write_streams_list);
842                 ctx->cur_write_res_offset = 0;
843         }
844
845         return 0;
846
847 error:
848         ERROR_WITH_ERRNO("Write error");
849         return ret;
850 }
851
852 static int
853 submit_chunk_for_compression(struct write_streams_ctx *ctx,
854                              const void *chunk, size_t size)
855 {
856         /* While we are unable to submit the chunk for compression (due to too
857          * many chunks already outstanding), retrieve and write the next
858          * compressed chunk.  */
859         while (!ctx->compressor->submit_chunk(ctx->compressor, chunk, size)) {
860                 const void *cchunk;
861                 unsigned csize;
862                 unsigned usize;
863                 bool bret;
864                 int ret;
865
866                 bret = ctx->compressor->get_chunk(ctx->compressor,
867                                                   &cchunk, &csize, &usize);
868
869                 wimlib_assert(bret);
870
871                 ret = write_chunk(ctx, cchunk, csize, usize);
872                 if (ret)
873                         return ret;
874         }
875         return 0;
876 }
877
878 /* Process the next chunk of data to be written to a WIM resource.  */
879 static int
880 write_stream_process_chunk(const void *chunk, size_t size, void *_ctx)
881 {
882         struct write_streams_ctx *ctx = _ctx;
883         int ret;
884         const u8 *chunkptr, *chunkend;
885
886         wimlib_assert(size != 0);
887
888         if (ctx->compressor == NULL) {
889                 /* Write chunk uncompressed.  */
890                  ret = write_chunk(ctx, chunk, size, size);
891                  if (ret)
892                          return ret;
893                  ctx->cur_read_res_offset += size;
894                  return 0;
895         }
896
897         /* Submit the chunk for compression, but take into account that the
898          * @size the chunk was provided in may not correspond to the
899          * @out_chunk_size being used for compression.  */
900         chunkptr = chunk;
901         chunkend = chunkptr + size;
902         do {
903                 const u8 *resized_chunk;
904                 size_t needed_chunk_size;
905
906                 if (ctx->write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS) {
907                         needed_chunk_size = ctx->out_chunk_size;
908                 } else {
909                         u64 res_bytes_remaining;
910
911                         res_bytes_remaining = ctx->cur_read_res_size -
912                                               ctx->cur_read_res_offset;
913                         needed_chunk_size = min(ctx->out_chunk_size,
914                                                 ctx->chunk_buf_filled +
915                                                         res_bytes_remaining);
916                 }
917
918                 if (ctx->chunk_buf_filled == 0 &&
919                     chunkend - chunkptr >= needed_chunk_size)
920                 {
921                         /* No intermediate buffering needed.  */
922                         resized_chunk = chunkptr;
923                         chunkptr += needed_chunk_size;
924                         ctx->cur_read_res_offset += needed_chunk_size;
925                 } else {
926                         /* Intermediate buffering needed.  */
927                         size_t bytes_consumed;
928
929                         bytes_consumed = min(chunkend - chunkptr,
930                                              needed_chunk_size - ctx->chunk_buf_filled);
931
932                         memcpy(&ctx->chunk_buf[ctx->chunk_buf_filled],
933                                chunkptr, bytes_consumed);
934
935                         resized_chunk = ctx->chunk_buf;
936
937                         chunkptr += bytes_consumed;
938                         ctx->cur_read_res_offset += bytes_consumed;
939                         ctx->chunk_buf_filled += bytes_consumed;
940                         if (ctx->chunk_buf_filled == needed_chunk_size) {
941                                 resized_chunk = ctx->chunk_buf;
942                                 ctx->chunk_buf_filled = 0;
943                         } else {
944                                 break;
945                         }
946
947                 }
948
949                 ret = submit_chunk_for_compression(ctx, resized_chunk,
950                                                    needed_chunk_size);
951                 if (ret)
952                         return ret;
953
954         } while (chunkptr != chunkend);
955         return 0;
956 }
957
958 /* Finish processing a stream for writing.  It may not have been completely
959  * written yet, as the chunk_compressor implementation may still have chunks
960  * buffered or being compressed.  */
961 static int
962 write_stream_end_read(struct wim_lookup_table_entry *lte, int status, void *_ctx)
963 {
964         struct write_streams_ctx *ctx = _ctx;
965         if (status == 0)
966                 wimlib_assert(ctx->cur_read_res_offset == ctx->cur_read_res_size);
967         if (ctx->stream_was_duplicate) {
968                 free_lookup_table_entry(lte);
969         } else if (lte->unhashed && ctx->lookup_table != NULL) {
970                 list_del(&lte->unhashed_list);
971                 lookup_table_insert(ctx->lookup_table, lte);
972                 lte->unhashed = 0;
973         }
974         return status;
975 }
976
977 /* Compute statistics about a list of streams that will be written.
978  *
979  * Assumes the streams are sorted such that all streams located in each distinct
980  * WIM (specified by WIMStruct) are together.  */
981 static void
982 compute_stream_list_stats(struct list_head *stream_list,
983                           struct write_streams_ctx *ctx)
984 {
985         struct wim_lookup_table_entry *lte;
986         u64 total_bytes = 0;
987         u64 num_streams = 0;
988         u64 total_parts = 0;
989         WIMStruct *prev_wim_part = NULL;
990
991         list_for_each_entry(lte, stream_list, write_streams_list) {
992                 num_streams++;
993                 total_bytes += lte->size;
994                 if (lte->resource_location == RESOURCE_IN_WIM) {
995                         if (prev_wim_part != lte->rspec->wim) {
996                                 prev_wim_part = lte->rspec->wim;
997                                 total_parts++;
998                         }
999                 }
1000         }
1001         ctx->progress_data.progress.write_streams.total_bytes       = total_bytes;
1002         ctx->progress_data.progress.write_streams.total_streams     = num_streams;
1003         ctx->progress_data.progress.write_streams.completed_bytes   = 0;
1004         ctx->progress_data.progress.write_streams.completed_streams = 0;
1005         ctx->progress_data.progress.write_streams.compression_type  = ctx->out_ctype;
1006         ctx->progress_data.progress.write_streams.total_parts       = total_parts;
1007         ctx->progress_data.progress.write_streams.completed_parts   = 0;
1008         ctx->progress_data.next_progress = 0;
1009         ctx->progress_data.prev_wim_part = NULL;
1010 }
1011
1012 /* Find streams in @stream_list that can be copied to the output WIM in raw form
1013  * rather than compressed.  Delete these streams from @stream_list, and move one
1014  * per resource to @raw_copy_resources.  Return the total uncompressed size of
1015  * the streams that need to be compressed.  */
1016 static u64
1017 find_raw_copy_resources(struct list_head *stream_list,
1018                         int write_resource_flags,
1019                         int out_ctype,
1020                         u32 out_chunk_size,
1021                         struct list_head *raw_copy_resources)
1022 {
1023         struct wim_lookup_table_entry *lte, *tmp;
1024         u64 num_bytes_to_compress = 0;
1025
1026         INIT_LIST_HEAD(raw_copy_resources);
1027
1028         /* Initialize temporary raw_copy_ok flag.  */
1029         list_for_each_entry(lte, stream_list, write_streams_list)
1030                 if (lte->resource_location == RESOURCE_IN_WIM)
1031                         lte->rspec->raw_copy_ok = 0;
1032
1033         list_for_each_entry_safe(lte, tmp, stream_list, write_streams_list) {
1034                 if (lte->resource_location == RESOURCE_IN_WIM &&
1035                     lte->rspec->raw_copy_ok)
1036                 {
1037                         list_del(&lte->write_streams_list);
1038                 } else if (can_raw_copy(lte, write_resource_flags,
1039                                  out_ctype, out_chunk_size))
1040                 {
1041                         lte->rspec->raw_copy_ok = 1;
1042                         list_move_tail(&lte->write_streams_list,
1043                                        raw_copy_resources);
1044                 } else {
1045                         num_bytes_to_compress += lte->size;
1046                 }
1047         }
1048
1049         return num_bytes_to_compress;
1050 }
1051
1052 /* Copy a raw compressed resource located in another WIM file to the WIM file
1053  * being written.  */
1054 static int
1055 write_raw_copy_resource(struct wim_resource_spec *in_rspec,
1056                         struct filedes *out_fd)
1057 {
1058         u64 cur_read_offset;
1059         u64 end_read_offset;
1060         u8 buf[BUFFER_SIZE];
1061         size_t bytes_to_read;
1062         int ret;
1063         struct filedes *in_fd;
1064         struct wim_lookup_table_entry *lte;
1065         u64 out_offset_in_wim;
1066
1067         DEBUG("Copying raw compressed data (size_in_wim=%"PRIu64", "
1068               "uncompressed_size=%"PRIu64")",
1069               in_rspec->size_in_wim, in_rspec->uncompressed_size);
1070
1071         /* Copy the raw data.  */
1072         cur_read_offset = in_rspec->offset_in_wim;
1073         end_read_offset = cur_read_offset + in_rspec->size_in_wim;
1074
1075         out_offset_in_wim = out_fd->offset;
1076
1077         if (in_rspec->is_pipable) {
1078                 if (cur_read_offset < sizeof(struct pwm_stream_hdr))
1079                         return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1080                 cur_read_offset -= sizeof(struct pwm_stream_hdr);
1081                 out_offset_in_wim += sizeof(struct pwm_stream_hdr);
1082         }
1083         in_fd = &in_rspec->wim->in_fd;
1084         wimlib_assert(cur_read_offset != end_read_offset);
1085         do {
1086
1087                 bytes_to_read = min(sizeof(buf), end_read_offset - cur_read_offset);
1088
1089                 ret = full_pread(in_fd, buf, bytes_to_read, cur_read_offset);
1090                 if (ret)
1091                         return ret;
1092
1093                 ret = full_write(out_fd, buf, bytes_to_read);
1094                 if (ret)
1095                         return ret;
1096
1097                 cur_read_offset += bytes_to_read;
1098
1099         } while (cur_read_offset != end_read_offset);
1100
1101         list_for_each_entry(lte, &in_rspec->stream_list, rspec_node) {
1102                 if (lte->will_be_in_output_wim) {
1103                         stream_set_out_reshdr_for_reuse(lte);
1104                         if (in_rspec->flags & WIM_RESHDR_FLAG_PACKED_STREAMS)
1105                                 lte->out_res_offset_in_wim = out_offset_in_wim;
1106                         else
1107                                 lte->out_reshdr.offset_in_wim = out_offset_in_wim;
1108
1109                 }
1110         }
1111         return 0;
1112 }
1113
1114 /* Copy a list of raw compressed resources located other WIM file(s) to the WIM
1115  * file being written.  */
1116 static int
1117 write_raw_copy_resources(struct list_head *raw_copy_resources,
1118                          struct filedes *out_fd,
1119                          struct write_streams_progress_data *progress_data)
1120 {
1121         struct wim_lookup_table_entry *lte;
1122         int ret;
1123
1124         list_for_each_entry(lte, raw_copy_resources, write_streams_list) {
1125                 ret = write_raw_copy_resource(lte->rspec, out_fd);
1126                 if (ret)
1127                         return ret;
1128                 do_write_streams_progress(progress_data, lte->size, false, lte);
1129         }
1130         return 0;
1131 }
1132
1133 /* Wait for and write all chunks pending in the compressor.  */
1134 static int
1135 finish_remaining_chunks(struct write_streams_ctx *ctx)
1136 {
1137         const void *cdata;
1138         unsigned csize;
1139         unsigned usize;
1140         int ret;
1141
1142         if (ctx->compressor == NULL)
1143                 return 0;
1144
1145         if (ctx->chunk_buf_filled != 0) {
1146                 ret = submit_chunk_for_compression(ctx, ctx->chunk_buf,
1147                                                    ctx->chunk_buf_filled);
1148                 if (ret)
1149                         return ret;
1150         }
1151
1152         while (ctx->compressor->get_chunk(ctx->compressor, &cdata, &csize, &usize)) {
1153                 ret = write_chunk(ctx, cdata, csize, usize);
1154                 if (ret)
1155                         return ret;
1156         }
1157         return 0;
1158 }
1159
1160 static void
1161 remove_zero_length_streams(struct list_head *stream_list)
1162 {
1163         struct wim_lookup_table_entry *lte, *tmp;
1164
1165         list_for_each_entry_safe(lte, tmp, stream_list, write_streams_list) {
1166                 wimlib_assert(lte->will_be_in_output_wim);
1167                 if (lte->size == 0) {
1168                         list_del(&lte->write_streams_list);
1169                         lte->out_reshdr.offset_in_wim = 0;
1170                         lte->out_reshdr.size_in_wim = 0;
1171                         lte->out_reshdr.uncompressed_size = 0;
1172                         lte->out_reshdr.flags = filter_resource_flags(lte->flags);
1173                 }
1174         }
1175 }
1176
1177 /*
1178  * Write a list of streams to the output WIM file.
1179  *
1180  * @stream_list
1181  *      The list of streams to write, specifies a list of `struct
1182  *      wim_lookup_table_entry's linked by the 'write_streams_list' member.
1183  *
1184  * @out_fd
1185  *      The file descriptor, opened for writing, to which to write the streams.
1186  *
1187  * @write_resource_flags
1188  *      Flags to modify how the streams are written:
1189  *
1190  *      WRITE_RESOURCE_FLAG_RECOMPRESS:
1191  *              Force compression of all resources, even if they could otherwise
1192  *              be re-used by caping the raw data, due to being located in a WIM
1193  *              file with compatible compression parameters.
1194  *
1195  *      WRITE_RESOURCE_FLAG_PIPABLE:
1196  *              Write the resources in the wimlib-specific pipable format, and
1197  *              furthermore do so in such a way that no seeking backwards in
1198  *              @out_fd will be performed (so it may be a pipe, contrary to the
1199  *              default behavior).
1200  *
1201  *      WRITE_RESOURCE_FLAG_PACK_STREAMS:
1202  *              Pack all the streams into a single resource rather than writing
1203  *              them in separate resources.  This format is only valid if the
1204  *              WIM version number is WIM_VERSION_PACKED_STREAMS.  This flag
1205  *              currently may not be combined with WRITE_RESOURCE_FLAG_PIPABLE.
1206  *
1207  * @out_ctype
1208  *      Compression format to use to write the output streams, specified as one
1209  *      of the WIMLIB_COMPRESSION_TYPE_* constants, excepting
1210  *      WIMLIB_COMPRESSION_TYPE_INVALID but including
1211  *      WIMLIB_COMPRESSION_TYPE_NONE.
1212  *
1213  * @out_chunk_size
1214  *      Chunk size to use to write the streams.  It must be a valid chunk size
1215  *      for the specified compression format @out_ctype, unless @out_ctype is
1216  *      WIMLIB_COMPRESSION_TYPE_NONE, in which case this parameter is ignored.
1217  *
1218  * @num_threads
1219  *      Number of threads to use to compress data.  If 0, a default number of
1220  *      threads will be chosen.  The number of threads still may be decreased
1221  *      from the specified value if insufficient memory is detected.
1222  *
1223  * @lookup_table
1224  *      If on-the-fly deduplication of unhashed streams is desired, this
1225  *      parameter must be pointer to the lookup table for the WIMStruct on whose
1226  *      behalf the streams are being written.  Otherwise, this parameter can be
1227  *      NULL.
1228  *
1229  * @filter_ctx
1230  *      If on-the-fly deduplication of unhashed streams is desired, this
1231  *      parameter can be a pointer to a context for stream filtering used to
1232  *      detect whether the duplicate stream has been hard-filtered or not.  If
1233  *      no streams are hard-filtered or no streams are unhashed, this parameter
1234  *      can be NULL.
1235  *
1236  * @progress_func
1237  *      If non-NULL, a progress function that will be called periodically with
1238  *      WIMLIB_PROGRESS_MSG_WRITE_STREAMS messages.  Note that on-the-fly
1239  *      deduplication of unhashed streams may result in the total bytes provided
1240  *      in the progress data to decrease from one message to the next.
1241  *
1242  * This function will write the streams in @stream_list to resources in
1243  * consecutive positions in the output WIM file, or to a single packed resource
1244  * if WRITE_RESOURCE_FLAG_PACK_STREAMS was specified in @write_resource_flags.
1245  * In both cases, the @out_reshdr of the `struct wim_lookup_table_entry' for
1246  * each stream written will be updated to specify its location, size, and flags
1247  * in the output WIM.  In the packed resource case,
1248  * WIM_RESHDR_FLAG_PACKED_STREAMS shall be set in the @flags field of the
1249  * @out_reshdr, and @out_res_offset_in_wim and @out_res_size_in_wim will also
1250  * be set to the offset and size, respectively, in the output WIM of the full
1251  * packed resource containing the corresponding stream.
1252  *
1253  * Each of the streams to write may be in any location supported by the
1254  * resource-handling code (specifically, read_stream_list()), such as the
1255  * contents of external file that has been logically added to the output WIM, or
1256  * a stream in another WIM file that has been imported, or even stream in the
1257  * "same" WIM file of which a modified copy is being written.  In the case that
1258  * a stream is already in a WIM file and uses compatible compression parameters,
1259  * by default this function will re-use the raw data instead of decompressing
1260  * it, then recompressing it; however, with WRITE_RESOURCE_FLAG_RECOMPRESS
1261  * specified in @write_resource_flags, this is not done.
1262  *
1263  * As a further requirement, this function requires that the
1264  * @will_be_in_output_wim member be set on all streams in @stream_list as well
1265  * as any other streams not in @stream_list that will be in the output WIM file,
1266  * but not on any other streams in the output WIM's lookup table or sharing a
1267  * packed resource with a stream in @stream_list.  Still furthermore, if
1268  * on-the-fly deduplication of streams is possible, then all streams in
1269  * @stream_list must also be linked by @lookup_table_list along with any other
1270  * streams that have @will_be_in_output_wim set.
1271  *
1272  * This function handles on-the-fly deduplication of streams for which SHA1
1273  * message digests have not yet been calculated and it is therefore known
1274  * whether such streams are already in @stream_list or in the WIM's lookup table
1275  * at all.  If @lookup_table is non-NULL, then each stream in @stream_list that
1276  * has @unhashed set but not @unique_size set is checksummed immediately before
1277  * it would otherwise be read for writing in order to determine if it is
1278  * identical to another stream already being written or one that would be
1279  * filtered out of the output WIM using stream_filtered() with the context
1280  * @filter_ctx.  Each such duplicate stream will be removed from @stream_list, its
1281  * reference count transfered to the pre-existing duplicate stream, its memory
1282  * freed, and will not be written.  Alternatively, if a stream in @stream_list
1283  * is a duplicate with any stream in @lookup_table that has not been marked for
1284  * writing or would not be hard-filtered, it is freed and the pre-existing
1285  * duplicate is written instead, taking ownership of the reference count and
1286  * slot in the @lookup_table_list.
1287  *
1288  * Returns 0 if all streams were written successfully (or did not need to be
1289  * written); otherwise a non-zero error code.
1290  */
1291 static int
1292 write_stream_list(struct list_head *stream_list,
1293                   struct filedes *out_fd,
1294                   int write_resource_flags,
1295                   int out_ctype,
1296                   u32 out_chunk_size,
1297                   unsigned num_threads,
1298                   struct wim_lookup_table *lookup_table,
1299                   struct filter_context *filter_ctx,
1300                   wimlib_progress_func_t progress_func)
1301 {
1302         int ret;
1303         struct write_streams_ctx ctx;
1304         struct list_head raw_copy_resources;
1305
1306         wimlib_assert((write_resource_flags &
1307                        (WRITE_RESOURCE_FLAG_PACK_STREAMS |
1308                         WRITE_RESOURCE_FLAG_PIPABLE)) !=
1309                                 (WRITE_RESOURCE_FLAG_PACK_STREAMS |
1310                                  WRITE_RESOURCE_FLAG_PIPABLE));
1311
1312         remove_zero_length_streams(stream_list);
1313
1314         if (list_empty(stream_list)) {
1315                 DEBUG("No streams to write.");
1316                 return 0;
1317         }
1318
1319         memset(&ctx, 0, sizeof(ctx));
1320
1321         /* Pre-sorting the streams is required for compute_stream_list_stats().
1322          * Afterwards, read_stream_list() need not sort them again.  */
1323         ret = sort_stream_list_by_sequential_order(stream_list,
1324                                                    offsetof(struct wim_lookup_table_entry,
1325                                                             write_streams_list));
1326         if (ret)
1327                 return ret;
1328
1329         ctx.out_fd = out_fd;
1330         ctx.lookup_table = lookup_table;
1331         ctx.out_ctype = out_ctype;
1332         ctx.out_chunk_size = out_chunk_size;
1333         ctx.write_resource_flags = write_resource_flags;
1334         ctx.filter_ctx = filter_ctx;
1335
1336         if (out_chunk_size != 0) {
1337                 if (out_chunk_size <= STACK_MAX) {
1338                         ctx.chunk_buf = alloca(out_chunk_size);
1339                 } else {
1340                         ctx.chunk_buf = MALLOC(out_chunk_size);
1341                         if (ctx.chunk_buf == NULL) {
1342                                 ret = WIMLIB_ERR_NOMEM;
1343                                 goto out_destroy_context;
1344                         }
1345                 }
1346         }
1347         ctx.chunk_buf_filled = 0;
1348
1349         compute_stream_list_stats(stream_list, &ctx);
1350
1351         ctx.progress_data.progress_func = progress_func;
1352
1353         ctx.num_bytes_to_compress = find_raw_copy_resources(stream_list,
1354                                                             write_resource_flags,
1355                                                             out_ctype,
1356                                                             out_chunk_size,
1357                                                             &raw_copy_resources);
1358
1359         DEBUG("Writing stream list "
1360               "(offset = %"PRIu64", write_resource_flags=0x%08x, "
1361               "out_ctype=%d, out_chunk_size=%u, num_threads=%u, "
1362               "total_bytes=%"PRIu64", num_bytes_to_compress=%"PRIu64")",
1363               out_fd->offset, write_resource_flags,
1364               out_ctype, out_chunk_size, num_threads,
1365               ctx.progress_data.progress.write_streams.total_bytes,
1366               ctx.num_bytes_to_compress);
1367
1368         if (ctx.num_bytes_to_compress == 0) {
1369                 DEBUG("No compression needed; skipping to raw copy!");
1370                 goto out_write_raw_copy_resources;
1371         }
1372
1373         /* Unless uncompressed output was required, allocate a chunk_compressor
1374          * to do compression.  There are serial and parallel implementations of
1375          * the chunk_compressor interface.  We default to parallel using the
1376          * specified number of threads, unless the upper bound on the number
1377          * bytes needing to be compressed is less 2000000 (heuristic value).  */
1378         if (out_ctype != WIMLIB_COMPRESSION_TYPE_NONE) {
1379
1380         #ifdef ENABLE_MULTITHREADED_COMPRESSION
1381                 if (ctx.num_bytes_to_compress >= 2000000) {
1382                         ret = new_parallel_chunk_compressor(out_ctype,
1383                                                             out_chunk_size,
1384                                                             num_threads, 0,
1385                                                             &ctx.compressor);
1386                         if (ret) {
1387                                 DEBUG("Couldn't create parallel chunk compressor "
1388                                       "(status %d)", ret);
1389                         }
1390                 }
1391         #endif
1392
1393                 if (ctx.compressor == NULL) {
1394                         ret = new_serial_chunk_compressor(out_ctype, out_chunk_size,
1395                                                           &ctx.compressor);
1396                         if (ret)
1397                                 goto out_destroy_context;
1398                 }
1399         }
1400
1401         if (ctx.compressor)
1402                 ctx.progress_data.progress.write_streams.num_threads = ctx.compressor->num_threads;
1403         else
1404                 ctx.progress_data.progress.write_streams.num_threads = 1;
1405
1406         DEBUG("Actually using %u threads",
1407               ctx.progress_data.progress.write_streams.num_threads);
1408
1409         INIT_LIST_HEAD(&ctx.pending_streams);
1410
1411         if (ctx.progress_data.progress_func) {
1412                 (*ctx.progress_data.progress_func)(WIMLIB_PROGRESS_MSG_WRITE_STREAMS,
1413                                                    &ctx.progress_data.progress);
1414         }
1415
1416         if (write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS) {
1417                 ret = begin_write_resource(&ctx, ctx.num_bytes_to_compress);
1418                 if (ret)
1419                         goto out_destroy_context;
1420         }
1421
1422         /* Read the list of streams needing to be compressed, using the
1423          * specified callbacks to execute processing of the data.  */
1424
1425         struct read_stream_list_callbacks cbs = {
1426                 .begin_stream           = write_stream_begin_read,
1427                 .begin_stream_ctx       = &ctx,
1428                 .consume_chunk          = write_stream_process_chunk,
1429                 .consume_chunk_ctx      = &ctx,
1430                 .end_stream             = write_stream_end_read,
1431                 .end_stream_ctx         = &ctx,
1432         };
1433
1434         ret = read_stream_list(stream_list,
1435                                offsetof(struct wim_lookup_table_entry, write_streams_list),
1436                                &cbs,
1437                                STREAM_LIST_ALREADY_SORTED |
1438                                         VERIFY_STREAM_HASHES |
1439                                         COMPUTE_MISSING_STREAM_HASHES);
1440
1441         if (ret)
1442                 goto out_destroy_context;
1443
1444         ret = finish_remaining_chunks(&ctx);
1445         if (ret)
1446                 goto out_destroy_context;
1447
1448         if (write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS) {
1449                 struct wim_reshdr reshdr;
1450                 struct wim_lookup_table_entry *lte;
1451                 u64 offset_in_res;
1452
1453                 ret = end_write_resource(&ctx, &reshdr);
1454                 if (ret)
1455                         goto out_destroy_context;
1456
1457                 DEBUG("Ending packed resource: %lu %lu %lu.",
1458                       reshdr.offset_in_wim,
1459                       reshdr.size_in_wim,
1460                       reshdr.uncompressed_size);
1461
1462                 offset_in_res = 0;
1463                 list_for_each_entry(lte, &ctx.pending_streams, write_streams_list) {
1464                         lte->out_reshdr.size_in_wim = lte->size;
1465                         lte->out_reshdr.flags = filter_resource_flags(lte->flags);
1466                         lte->out_reshdr.flags |= WIM_RESHDR_FLAG_PACKED_STREAMS;
1467                         lte->out_reshdr.uncompressed_size = 0;
1468                         lte->out_reshdr.offset_in_wim = offset_in_res;
1469                         lte->out_res_offset_in_wim = reshdr.offset_in_wim;
1470                         lte->out_res_size_in_wim = reshdr.size_in_wim;
1471                         /*lte->out_res_uncompressed_size = reshdr.uncompressed_size;*/
1472                         offset_in_res += lte->size;
1473                 }
1474                 wimlib_assert(offset_in_res == reshdr.uncompressed_size);
1475         }
1476
1477 out_write_raw_copy_resources:
1478         /* Copy any compressed resources for which the raw data can be reused
1479          * without decompression.  */
1480         ret = write_raw_copy_resources(&raw_copy_resources, ctx.out_fd,
1481                                        &ctx.progress_data);
1482
1483 out_destroy_context:
1484         if (out_chunk_size > STACK_MAX)
1485                 FREE(ctx.chunk_buf);
1486         FREE(ctx.chunk_csizes);
1487         if (ctx.compressor)
1488                 ctx.compressor->destroy(ctx.compressor);
1489         DEBUG("Done (ret=%d)", ret);
1490         return ret;
1491 }
1492
1493 static int
1494 wim_write_stream_list(WIMStruct *wim,
1495                       struct list_head *stream_list,
1496                       int write_flags,
1497                       unsigned num_threads,
1498                       struct filter_context *filter_ctx,
1499                       wimlib_progress_func_t progress_func)
1500 {
1501         int out_ctype;
1502         u32 out_chunk_size;
1503         int write_resource_flags;
1504
1505         write_resource_flags = write_flags_to_resource_flags(write_flags);
1506
1507         if (write_resource_flags & WRITE_RESOURCE_FLAG_PACK_STREAMS) {
1508                 out_chunk_size = wim->out_pack_chunk_size;
1509                 out_ctype = wim->out_pack_compression_type;
1510         } else {
1511                 out_chunk_size = wim->out_chunk_size;
1512                 out_ctype = wim->out_compression_type;
1513         }
1514
1515         return write_stream_list(stream_list,
1516                                  &wim->out_fd,
1517                                  write_resource_flags,
1518                                  out_ctype,
1519                                  out_chunk_size,
1520                                  num_threads,
1521                                  wim->lookup_table,
1522                                  filter_ctx,
1523                                  progress_func);
1524 }
1525
1526 static int
1527 write_wim_resource(struct wim_lookup_table_entry *lte,
1528                    struct filedes *out_fd,
1529                    int out_ctype,
1530                    u32 out_chunk_size,
1531                    int write_resource_flags)
1532 {
1533         LIST_HEAD(stream_list);
1534         list_add(&lte->write_streams_list, &stream_list);
1535         lte->will_be_in_output_wim = 1;
1536         return write_stream_list(&stream_list,
1537                                  out_fd,
1538                                  write_resource_flags & ~WRITE_RESOURCE_FLAG_PACK_STREAMS,
1539                                  out_ctype,
1540                                  out_chunk_size,
1541                                  1,
1542                                  NULL,
1543                                  NULL,
1544                                  NULL);
1545 }
1546
1547 int
1548 write_wim_resource_from_buffer(const void *buf, size_t buf_size,
1549                                int reshdr_flags, struct filedes *out_fd,
1550                                int out_ctype,
1551                                u32 out_chunk_size,
1552                                struct wim_reshdr *out_reshdr,
1553                                u8 *hash,
1554                                int write_resource_flags)
1555 {
1556         int ret;
1557         struct wim_lookup_table_entry *lte;
1558
1559         /* Set up a temporary lookup table entry to provide to
1560          * write_wim_resource().  */
1561
1562         lte = new_lookup_table_entry();
1563         if (lte == NULL)
1564                 return WIMLIB_ERR_NOMEM;
1565
1566         lte->resource_location  = RESOURCE_IN_ATTACHED_BUFFER;
1567         lte->attached_buffer    = (void*)buf;
1568         lte->size               = buf_size;
1569         lte->flags              = reshdr_flags;
1570
1571         if (write_resource_flags & WRITE_RESOURCE_FLAG_PIPABLE) {
1572                 sha1_buffer(buf, buf_size, lte->hash);
1573                 lte->unhashed = 0;
1574         } else {
1575                 lte->unhashed = 1;
1576         }
1577
1578         ret = write_wim_resource(lte, out_fd, out_ctype, out_chunk_size,
1579                                  write_resource_flags);
1580         if (ret)
1581                 goto out_free_lte;
1582
1583         copy_reshdr(out_reshdr, &lte->out_reshdr);
1584
1585         if (hash)
1586                 copy_hash(hash, lte->hash);
1587         ret = 0;
1588 out_free_lte:
1589         lte->resource_location = RESOURCE_NONEXISTENT;
1590         free_lookup_table_entry(lte);
1591         return ret;
1592 }
1593
1594 struct stream_size_table {
1595         struct hlist_head *array;
1596         size_t num_entries;
1597         size_t capacity;
1598 };
1599
1600 static int
1601 init_stream_size_table(struct stream_size_table *tab, size_t capacity)
1602 {
1603         tab->array = CALLOC(capacity, sizeof(tab->array[0]));
1604         if (tab->array == NULL)
1605                 return WIMLIB_ERR_NOMEM;
1606         tab->num_entries = 0;
1607         tab->capacity = capacity;
1608         return 0;
1609 }
1610
1611 static void
1612 destroy_stream_size_table(struct stream_size_table *tab)
1613 {
1614         FREE(tab->array);
1615 }
1616
1617 static int
1618 stream_size_table_insert(struct wim_lookup_table_entry *lte, void *_tab)
1619 {
1620         struct stream_size_table *tab = _tab;
1621         size_t pos;
1622         struct wim_lookup_table_entry *same_size_lte;
1623         struct hlist_node *tmp;
1624
1625         pos = hash_u64(lte->size) % tab->capacity;
1626         lte->unique_size = 1;
1627         hlist_for_each_entry(same_size_lte, tmp, &tab->array[pos], hash_list_2) {
1628                 if (same_size_lte->size == lte->size) {
1629                         lte->unique_size = 0;
1630                         same_size_lte->unique_size = 0;
1631                         break;
1632                 }
1633         }
1634
1635         hlist_add_head(&lte->hash_list_2, &tab->array[pos]);
1636         tab->num_entries++;
1637         return 0;
1638 }
1639
1640 struct find_streams_ctx {
1641         WIMStruct *wim;
1642         int write_flags;
1643         struct list_head stream_list;
1644         struct stream_size_table stream_size_tab;
1645 };
1646
1647 static void
1648 reference_stream_for_write(struct wim_lookup_table_entry *lte,
1649                            struct list_head *stream_list, u32 nref)
1650 {
1651         if (!lte->will_be_in_output_wim) {
1652                 lte->out_refcnt = 0;
1653                 list_add_tail(&lte->write_streams_list, stream_list);
1654                 lte->will_be_in_output_wim = 1;
1655         }
1656         lte->out_refcnt += nref;
1657 }
1658
1659 static int
1660 fully_reference_stream_for_write(struct wim_lookup_table_entry *lte,
1661                                  void *_stream_list)
1662 {
1663         struct list_head *stream_list = _stream_list;
1664         lte->will_be_in_output_wim = 0;
1665         reference_stream_for_write(lte, stream_list, lte->refcnt);
1666         return 0;
1667 }
1668
1669 static int
1670 inode_find_streams_to_reference(const struct wim_inode *inode,
1671                                 const struct wim_lookup_table *table,
1672                                 struct list_head *stream_list)
1673 {
1674         struct wim_lookup_table_entry *lte;
1675         unsigned i;
1676
1677         wimlib_assert(inode->i_nlink > 0);
1678
1679         for (i = 0; i <= inode->i_num_ads; i++) {
1680                 lte = inode_stream_lte(inode, i, table);
1681                 if (lte)
1682                         reference_stream_for_write(lte, stream_list,
1683                                                    inode->i_nlink);
1684                 else if (!is_zero_hash(inode_stream_hash(inode, i)))
1685                         return WIMLIB_ERR_RESOURCE_NOT_FOUND;
1686         }
1687         return 0;
1688 }
1689
1690 static int
1691 do_stream_set_not_in_output_wim(struct wim_lookup_table_entry *lte, void *_ignore)
1692 {
1693         lte->will_be_in_output_wim = 0;
1694         return 0;
1695 }
1696
1697 static int
1698 image_find_streams_to_reference(WIMStruct *wim)
1699 {
1700         struct wim_image_metadata *imd;
1701         struct wim_inode *inode;
1702         struct wim_lookup_table_entry *lte;
1703         struct list_head *stream_list;
1704         int ret;
1705
1706         imd = wim_get_current_image_metadata(wim);
1707
1708         image_for_each_unhashed_stream(lte, imd)
1709                 lte->will_be_in_output_wim = 0;
1710
1711         stream_list = wim->private;
1712         image_for_each_inode(inode, imd) {
1713                 ret = inode_find_streams_to_reference(inode,
1714                                                       wim->lookup_table,
1715                                                       stream_list);
1716                 if (ret)
1717                         return ret;
1718         }
1719         return 0;
1720 }
1721
1722 static int
1723 prepare_unfiltered_list_of_streams_in_output_wim(WIMStruct *wim,
1724                                                  int image,
1725                                                  int streams_ok,
1726                                                  struct list_head *stream_list_ret)
1727 {
1728         int ret;
1729
1730         INIT_LIST_HEAD(stream_list_ret);
1731
1732         if (streams_ok && (image == WIMLIB_ALL_IMAGES ||
1733                            (image == 1 && wim->hdr.image_count == 1)))
1734         {
1735                 /* Fast case:  Assume that all streams are being written and
1736                  * that the reference counts are correct.  */
1737                 struct wim_lookup_table_entry *lte;
1738                 struct wim_image_metadata *imd;
1739                 unsigned i;
1740
1741                 for_lookup_table_entry(wim->lookup_table,
1742                                        fully_reference_stream_for_write,
1743                                        stream_list_ret);
1744
1745                 for (i = 0; i < wim->hdr.image_count; i++) {
1746                         imd = wim->image_metadata[i];
1747                         image_for_each_unhashed_stream(lte, imd)
1748                                 fully_reference_stream_for_write(lte, stream_list_ret);
1749                 }
1750         } else {
1751                 /* Slow case:  Walk through the images being written and
1752                  * determine the streams referenced.  */
1753                 for_lookup_table_entry(wim->lookup_table,
1754                                        do_stream_set_not_in_output_wim, NULL);
1755                 wim->private = stream_list_ret;
1756                 ret = for_image(wim, image, image_find_streams_to_reference);
1757                 if (ret)
1758                         return ret;
1759         }
1760
1761         return 0;
1762 }
1763
1764 struct insert_other_if_hard_filtered_ctx {
1765         struct stream_size_table *tab;
1766         struct filter_context *filter_ctx;
1767 };
1768
1769 static int
1770 insert_other_if_hard_filtered(struct wim_lookup_table_entry *lte, void *_ctx)
1771 {
1772         struct insert_other_if_hard_filtered_ctx *ctx = _ctx;
1773
1774         if (!lte->will_be_in_output_wim &&
1775             stream_hard_filtered(lte, ctx->filter_ctx))
1776                 stream_size_table_insert(lte, ctx->tab);
1777         return 0;
1778 }
1779
1780 static int
1781 determine_stream_size_uniquity(struct list_head *stream_list,
1782                                struct wim_lookup_table *lt,
1783                                struct filter_context *filter_ctx)
1784 {
1785         int ret;
1786         struct stream_size_table tab;
1787         struct wim_lookup_table_entry *lte;
1788
1789         ret = init_stream_size_table(&tab, lt->capacity);
1790         if (ret)
1791                 return ret;
1792
1793         if (may_hard_filter_streams(filter_ctx)) {
1794                 struct insert_other_if_hard_filtered_ctx ctx = {
1795                         .tab = &tab,
1796                         .filter_ctx = filter_ctx,
1797                 };
1798                 for_lookup_table_entry(lt, insert_other_if_hard_filtered, &ctx);
1799         }
1800
1801         list_for_each_entry(lte, stream_list, write_streams_list)
1802                 stream_size_table_insert(lte, &tab);
1803
1804         destroy_stream_size_table(&tab);
1805         return 0;
1806 }
1807
1808 static void
1809 filter_stream_list_for_write(struct list_head *stream_list,
1810                              struct filter_context *filter_ctx)
1811 {
1812         struct wim_lookup_table_entry *lte, *tmp;
1813
1814         list_for_each_entry_safe(lte, tmp,
1815                                  stream_list, write_streams_list)
1816         {
1817                 int status = stream_filtered(lte, filter_ctx);
1818
1819                 if (status == 0) {
1820                         /* Not filtered.  */
1821                         continue;
1822                 } else {
1823                         if (status > 0) {
1824                                 /* Soft filtered.  */
1825                         } else {
1826                                 /* Hard filtered.  */
1827                                 lte->will_be_in_output_wim = 0;
1828                                 list_del(&lte->lookup_table_list);
1829                         }
1830                         list_del(&lte->write_streams_list);
1831                 }
1832         }
1833 }
1834
1835 /*
1836  * prepare_stream_list_for_write() -
1837  *
1838  * Prepare the list of streams to write for writing a WIM containing the
1839  * specified image(s) with the specified write flags.
1840  *
1841  * @wim
1842  *      The WIMStruct on whose behalf the write is occurring.
1843  *
1844  * @image
1845  *      Image(s) from the WIM to write; may be WIMLIB_ALL_IMAGES.
1846  *
1847  * @write_flags
1848  *      WIMLIB_WRITE_FLAG_* flags for the write operation:
1849  *
1850  *      STREAMS_OK:  For writes of all images, assume that all streams in the
1851  *      lookup table of @wim and the per-image lists of unhashed streams should
1852  *      be taken as-is, and image metadata should not be searched for
1853  *      references.  This does not exclude filtering with OVERWRITE and
1854  *      SKIP_EXTERNAL_WIMS, below.
1855  *
1856  *      OVERWRITE:  Streams already present in @wim shall not be returned in
1857  *      @stream_list_ret.
1858  *
1859  *      SKIP_EXTERNAL_WIMS:  Streams already present in a WIM file, but not
1860  *      @wim, shall be be returned in neither @stream_list_ret nor
1861  *      @lookup_table_list_ret.
1862  *
1863  * @stream_list_ret
1864  *      List of streams, linked by write_streams_list, that need to be written
1865  *      will be returned here.
1866  *
1867  *      Note that this function assumes that unhashed streams will be written;
1868  *      it does not take into account that they may become duplicates when
1869  *      actually hashed.
1870  *
1871  * @lookup_table_list_ret
1872  *      List of streams, linked by lookup_table_list, that need to be included
1873  *      in the WIM's lookup table will be returned here.  This will be a
1874  *      superset of the streams in @stream_list_ret.
1875  *
1876  *      This list will be a proper superset of @stream_list_ret if and only if
1877  *      WIMLIB_WRITE_FLAG_OVERWRITE was specified in @write_flags and some of
1878  *      the streams that would otherwise need to be written were already located
1879  *      in the WIM file.
1880  *
1881  *      All streams in this list will have @out_refcnt set to the number of
1882  *      references to the stream in the output WIM.  If
1883  *      WIMLIB_WRITE_FLAG_STREAMS_OK was specified in @write_flags, @out_refcnt
1884  *      may be as low as 0.
1885  *
1886  * @filter_ctx_ret
1887  *      A context for queries of stream filter status with stream_filtered() is
1888  *      returned in this location.
1889  *
1890  * In addition, @will_be_in_output_wim will be set to 1 in all stream entries
1891  * inserted into @lookup_table_list_ret and to 0 in all stream entries in the
1892  * lookup table of @wim not inserted into @lookup_table_list_ret.
1893  *
1894  * Still furthermore, @unique_size will be set to 1 on all stream entries in
1895  * @stream_list_ret that have unique size among all stream entries in
1896  * @stream_list_ret and among all stream entries in the lookup table of @wim
1897  * that are ineligible for being written due to filtering.
1898  *
1899  * Returns 0 on success; nonzero on read error, memory allocation error, or
1900  * otherwise.
1901  */
1902 static int
1903 prepare_stream_list_for_write(WIMStruct *wim, int image,
1904                               int write_flags,
1905                               struct list_head *stream_list_ret,
1906                               struct list_head *lookup_table_list_ret,
1907                               struct filter_context *filter_ctx_ret)
1908 {
1909         int ret;
1910         struct wim_lookup_table_entry *lte;
1911
1912         filter_ctx_ret->write_flags = write_flags;
1913         filter_ctx_ret->wim = wim;
1914
1915         ret = prepare_unfiltered_list_of_streams_in_output_wim(
1916                                 wim,
1917                                 image,
1918                                 write_flags & WIMLIB_WRITE_FLAG_STREAMS_OK,
1919                                 stream_list_ret);
1920         if (ret)
1921                 return ret;
1922
1923         INIT_LIST_HEAD(lookup_table_list_ret);
1924         list_for_each_entry(lte, stream_list_ret, write_streams_list)
1925                 list_add_tail(&lte->lookup_table_list, lookup_table_list_ret);
1926
1927         ret = determine_stream_size_uniquity(stream_list_ret, wim->lookup_table,
1928                                              filter_ctx_ret);
1929         if (ret)
1930                 return ret;
1931
1932         if (may_filter_streams(filter_ctx_ret))
1933                 filter_stream_list_for_write(stream_list_ret, filter_ctx_ret);
1934
1935         return 0;
1936 }
1937
1938 static int
1939 write_wim_streams(WIMStruct *wim, int image, int write_flags,
1940                   unsigned num_threads,
1941                   wimlib_progress_func_t progress_func,
1942                   struct list_head *stream_list_override,
1943                   struct list_head *lookup_table_list_ret)
1944 {
1945         int ret;
1946         struct list_head _stream_list;
1947         struct list_head *stream_list;
1948         struct wim_lookup_table_entry *lte;
1949         struct filter_context _filter_ctx;
1950         struct filter_context *filter_ctx;
1951
1952         if (stream_list_override == NULL) {
1953                 /* Normal case: prepare stream list from image(s) being written.
1954                  */
1955                 stream_list = &_stream_list;
1956                 filter_ctx = &_filter_ctx;
1957                 ret = prepare_stream_list_for_write(wim, image, write_flags,
1958                                                     stream_list,
1959                                                     lookup_table_list_ret,
1960                                                     filter_ctx);
1961                 if (ret)
1962                         return ret;
1963         } else {
1964                 /* Currently only as a result of wimlib_split() being called:
1965                  * use stream list already explicitly provided.  Use existing
1966                  * reference counts.  */
1967                 stream_list = stream_list_override;
1968                 filter_ctx = NULL;
1969                 INIT_LIST_HEAD(lookup_table_list_ret);
1970                 list_for_each_entry(lte, stream_list, write_streams_list) {
1971                         lte->out_refcnt = lte->refcnt;
1972                         lte->will_be_in_output_wim = 1;
1973                         lte->unique_size = 0;
1974                         list_add_tail(&lte->lookup_table_list, lookup_table_list_ret);
1975                 }
1976         }
1977
1978         return wim_write_stream_list(wim,
1979                                      stream_list,
1980                                      write_flags,
1981                                      num_threads,
1982                                      filter_ctx,
1983                                      progress_func);
1984 }
1985
1986 static int
1987 write_wim_metadata_resources(WIMStruct *wim, int image, int write_flags,
1988                              wimlib_progress_func_t progress_func)
1989 {
1990         int ret;
1991         int start_image;
1992         int end_image;
1993         int write_resource_flags;
1994
1995         if (write_flags & WIMLIB_WRITE_FLAG_NO_METADATA) {
1996                 DEBUG("Not writing any metadata resources.");
1997                 return 0;
1998         }
1999
2000         write_resource_flags = write_flags_to_resource_flags(write_flags);
2001
2002         write_resource_flags &= ~WRITE_RESOURCE_FLAG_PACK_STREAMS;
2003
2004         DEBUG("Writing metadata resources (offset=%"PRIu64")",
2005               wim->out_fd.offset);
2006
2007         if (progress_func)
2008                 progress_func(WIMLIB_PROGRESS_MSG_WRITE_METADATA_BEGIN, NULL);
2009
2010         if (image == WIMLIB_ALL_IMAGES) {
2011                 start_image = 1;
2012                 end_image = wim->hdr.image_count;
2013         } else {
2014                 start_image = image;
2015                 end_image = image;
2016         }
2017
2018         for (int i = start_image; i <= end_image; i++) {
2019                 struct wim_image_metadata *imd;
2020
2021                 imd = wim->image_metadata[i - 1];
2022                 /* Build a new metadata resource only if image was modified from
2023                  * the original (or was newly added).  Otherwise just copy the
2024                  * existing one.  */
2025                 if (imd->modified) {
2026                         DEBUG("Image %u was modified; building and writing new "
2027                               "metadata resource", i);
2028                         ret = write_metadata_resource(wim, i,
2029                                                       write_resource_flags);
2030                 } else if (write_flags & WIMLIB_WRITE_FLAG_OVERWRITE) {
2031                         DEBUG("Image %u was not modified; re-using existing "
2032                               "metadata resource.", i);
2033                         stream_set_out_reshdr_for_reuse(imd->metadata_lte);
2034                         ret = 0;
2035                 } else {
2036                         DEBUG("Image %u was not modified; copying existing "
2037                               "metadata resource.", i);
2038                         ret = write_wim_resource(imd->metadata_lte,
2039                                                  &wim->out_fd,
2040                                                  wim->out_compression_type,
2041                                                  wim->out_chunk_size,
2042                                                  write_resource_flags);
2043                 }
2044                 if (ret)
2045                         return ret;
2046         }
2047         if (progress_func)
2048                 progress_func(WIMLIB_PROGRESS_MSG_WRITE_METADATA_END, NULL);
2049         return 0;
2050 }
2051
2052 static int
2053 open_wim_writable(WIMStruct *wim, const tchar *path, int open_flags)
2054 {
2055         int raw_fd;
2056         DEBUG("Opening \"%"TS"\" for writing.", path);
2057
2058         raw_fd = topen(path, open_flags | O_BINARY, 0644);
2059         if (raw_fd < 0) {
2060                 ERROR_WITH_ERRNO("Failed to open \"%"TS"\" for writing", path);
2061                 return WIMLIB_ERR_OPEN;
2062         }
2063         filedes_init(&wim->out_fd, raw_fd);
2064         return 0;
2065 }
2066
2067 static int
2068 close_wim_writable(WIMStruct *wim, int write_flags)
2069 {
2070         int ret = 0;
2071
2072         if (!(write_flags & WIMLIB_WRITE_FLAG_FILE_DESCRIPTOR)) {
2073                 DEBUG("Closing WIM file.");
2074                 if (filedes_valid(&wim->out_fd))
2075                         if (filedes_close(&wim->out_fd))
2076                                 ret = WIMLIB_ERR_WRITE;
2077         }
2078         filedes_invalidate(&wim->out_fd);
2079         return ret;
2080 }
2081
2082 static int
2083 cmp_streams_by_out_rspec(const void *p1, const void *p2)
2084 {
2085         const struct wim_lookup_table_entry *lte1, *lte2;
2086
2087         lte1 = *(const struct wim_lookup_table_entry**)p1;
2088         lte2 = *(const struct wim_lookup_table_entry**)p2;
2089
2090         if (lte1->out_reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
2091                 if (lte2->out_reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
2092                         if (lte1->out_res_offset_in_wim != lte2->out_res_offset_in_wim)
2093                                 return cmp_u64(lte1->out_res_offset_in_wim,
2094                                                lte2->out_res_offset_in_wim);
2095                 } else {
2096                         return 1;
2097                 }
2098         } else {
2099                 if (lte2->out_reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS)
2100                         return -1;
2101         }
2102         return cmp_u64(lte1->out_reshdr.offset_in_wim,
2103                        lte2->out_reshdr.offset_in_wim);
2104 }
2105
2106 static int
2107 write_wim_lookup_table(WIMStruct *wim, int image, int write_flags,
2108                        struct wim_reshdr *out_reshdr,
2109                        struct list_head *lookup_table_list)
2110 {
2111         int ret;
2112
2113         /* Set output resource metadata for streams already present in WIM.  */
2114         if (write_flags & WIMLIB_WRITE_FLAG_OVERWRITE) {
2115                 struct wim_lookup_table_entry *lte;
2116                 list_for_each_entry(lte, lookup_table_list, lookup_table_list)
2117                 {
2118                         if (lte->resource_location == RESOURCE_IN_WIM &&
2119                             lte->rspec->wim == wim)
2120                         {
2121                                 stream_set_out_reshdr_for_reuse(lte);
2122                         }
2123                 }
2124         }
2125
2126         ret = sort_stream_list(lookup_table_list,
2127                                offsetof(struct wim_lookup_table_entry, lookup_table_list),
2128                                cmp_streams_by_out_rspec);
2129         if (ret)
2130                 return ret;
2131
2132         /* Add entries for metadata resources.  */
2133         if (!(write_flags & WIMLIB_WRITE_FLAG_NO_METADATA)) {
2134                 int start_image;
2135                 int end_image;
2136
2137                 if (image == WIMLIB_ALL_IMAGES) {
2138                         start_image = 1;
2139                         end_image = wim->hdr.image_count;
2140                 } else {
2141                         start_image = image;
2142                         end_image = image;
2143                 }
2144
2145                 /* Push metadata resource lookup table entries onto the front of
2146                  * the list in reverse order, so that they're written in order.
2147                  */
2148                 for (int i = end_image; i >= start_image; i--) {
2149                         struct wim_lookup_table_entry *metadata_lte;
2150
2151                         metadata_lte = wim->image_metadata[i - 1]->metadata_lte;
2152                         wimlib_assert(metadata_lte->out_reshdr.flags & WIM_RESHDR_FLAG_METADATA);
2153                         metadata_lte->out_refcnt = 1;
2154                         list_add(&metadata_lte->lookup_table_list, lookup_table_list);
2155                 }
2156         }
2157
2158         return write_wim_lookup_table_from_stream_list(lookup_table_list,
2159                                                        &wim->out_fd,
2160                                                        wim->hdr.part_number,
2161                                                        out_reshdr,
2162                                                        write_flags_to_resource_flags(write_flags));
2163 }
2164
2165 /*
2166  * finish_write():
2167  *
2168  * Finish writing a WIM file: write the lookup table, xml data, and integrity
2169  * table, then overwrite the WIM header.  By default, closes the WIM file
2170  * descriptor (@wim->out_fd) if successful.
2171  *
2172  * write_flags is a bitwise OR of the following:
2173  *
2174  *      (public) WIMLIB_WRITE_FLAG_CHECK_INTEGRITY:
2175  *              Include an integrity table.
2176  *
2177  *      (public) WIMLIB_WRITE_FLAG_FSYNC:
2178  *              fsync() the output file before closing it.
2179  *
2180  *      (public) WIMLIB_WRITE_FLAG_PIPABLE:
2181  *              Writing a pipable WIM, possibly to a pipe; include pipable WIM
2182  *              stream headers before the lookup table and XML data, and also
2183  *              write the WIM header at the end instead of seeking to the
2184  *              beginning.  Can't be combined with
2185  *              WIMLIB_WRITE_FLAG_CHECK_INTEGRITY.
2186  *
2187  *      (private) WIMLIB_WRITE_FLAG_NO_LOOKUP_TABLE:
2188  *              Don't write the lookup table.
2189  *
2190  *      (private) WIMLIB_WRITE_FLAG_REUSE_INTEGRITY_TABLE:
2191  *              When (if) writing the integrity table, re-use entries from the
2192  *              existing integrity table, if possible.
2193  *
2194  *      (private) WIMLIB_WRITE_FLAG_CHECKPOINT_AFTER_XML:
2195  *              After writing the XML data but before writing the integrity
2196  *              table, write a temporary WIM header and flush the stream so that
2197  *              the WIM is less likely to become corrupted upon abrupt program
2198  *              termination.
2199  *      (private) WIMLIB_WRITE_FLAG_HEADER_AT_END:
2200  *              Instead of overwriting the WIM header at the beginning of the
2201  *              file, simply append it to the end of the file.  (Used when
2202  *              writing to pipe.)
2203  *      (private) WIMLIB_WRITE_FLAG_FILE_DESCRIPTOR:
2204  *              Do not close the file descriptor @wim->out_fd on either success
2205  *              on failure.
2206  *      (private) WIMLIB_WRITE_FLAG_USE_EXISTING_TOTALBYTES:
2207  *              Use the existing <TOTALBYTES> stored in the in-memory XML
2208  *              information, rather than setting it to the offset of the XML
2209  *              data being written.
2210  */
2211 static int
2212 finish_write(WIMStruct *wim, int image, int write_flags,
2213              wimlib_progress_func_t progress_func,
2214              struct list_head *lookup_table_list)
2215 {
2216         int ret;
2217         off_t hdr_offset;
2218         int write_resource_flags;
2219         off_t old_lookup_table_end;
2220         off_t new_lookup_table_end;
2221         u64 xml_totalbytes;
2222
2223         DEBUG("image=%d, write_flags=%08x", image, write_flags);
2224
2225         write_resource_flags = write_flags_to_resource_flags(write_flags);
2226
2227         /* In the WIM header, there is room for the resource entry for a
2228          * metadata resource labeled as the "boot metadata".  This entry should
2229          * be zeroed out if there is no bootable image (boot_idx 0).  Otherwise,
2230          * it should be a copy of the resource entry for the image that is
2231          * marked as bootable.  This is not well documented...  */
2232         if (wim->hdr.boot_idx == 0) {
2233                 zero_reshdr(&wim->hdr.boot_metadata_reshdr);
2234         } else {
2235                 copy_reshdr(&wim->hdr.boot_metadata_reshdr,
2236                             &wim->image_metadata[
2237                                 wim->hdr.boot_idx - 1]->metadata_lte->out_reshdr);
2238         }
2239
2240         /* Write lookup table.  (Save old position first.)  */
2241         old_lookup_table_end = wim->hdr.lookup_table_reshdr.offset_in_wim +
2242                                wim->hdr.lookup_table_reshdr.size_in_wim;
2243         if (!(write_flags & WIMLIB_WRITE_FLAG_NO_LOOKUP_TABLE)) {
2244                 ret = write_wim_lookup_table(wim, image, write_flags,
2245                                              &wim->hdr.lookup_table_reshdr,
2246                                              lookup_table_list);
2247                 if (ret)
2248                         return ret;
2249         }
2250
2251         /* Write XML data.  */
2252         xml_totalbytes = wim->out_fd.offset;
2253         if (write_flags & WIMLIB_WRITE_FLAG_USE_EXISTING_TOTALBYTES)
2254                 xml_totalbytes = WIM_TOTALBYTES_USE_EXISTING;
2255         ret = write_wim_xml_data(wim, image, xml_totalbytes,
2256                                  &wim->hdr.xml_data_reshdr,
2257                                  write_resource_flags);
2258         if (ret)
2259                 return ret;
2260
2261         /* Write integrity table (optional).  */
2262         if (write_flags & WIMLIB_WRITE_FLAG_CHECK_INTEGRITY) {
2263                 if (write_flags & WIMLIB_WRITE_FLAG_CHECKPOINT_AFTER_XML) {
2264                         struct wim_header checkpoint_hdr;
2265                         memcpy(&checkpoint_hdr, &wim->hdr, sizeof(struct wim_header));
2266                         zero_reshdr(&checkpoint_hdr.integrity_table_reshdr);
2267                         checkpoint_hdr.flags |= WIM_HDR_FLAG_WRITE_IN_PROGRESS;
2268                         ret = write_wim_header_at_offset(&checkpoint_hdr,
2269                                                          &wim->out_fd, 0);
2270                         if (ret)
2271                                 return ret;
2272                 }
2273
2274                 if (!(write_flags & WIMLIB_WRITE_FLAG_REUSE_INTEGRITY_TABLE))
2275                         old_lookup_table_end = 0;
2276
2277                 new_lookup_table_end = wim->hdr.lookup_table_reshdr.offset_in_wim +
2278                                        wim->hdr.lookup_table_reshdr.size_in_wim;
2279
2280                 ret = write_integrity_table(wim,
2281                                             new_lookup_table_end,
2282                                             old_lookup_table_end,
2283                                             progress_func);
2284                 if (ret)
2285                         return ret;
2286         } else {
2287                 /* No integrity table.  */
2288                 zero_reshdr(&wim->hdr.integrity_table_reshdr);
2289         }
2290
2291         /* Now that all information in the WIM header has been determined, the
2292          * preliminary header written earlier can be overwritten, the header of
2293          * the existing WIM file can be overwritten, or the final header can be
2294          * written to the end of the pipable WIM.  */
2295         wim->hdr.flags &= ~WIM_HDR_FLAG_WRITE_IN_PROGRESS;
2296         hdr_offset = 0;
2297         if (write_flags & WIMLIB_WRITE_FLAG_HEADER_AT_END)
2298                 hdr_offset = wim->out_fd.offset;
2299         DEBUG("Writing new header @ %"PRIu64".", hdr_offset);
2300         ret = write_wim_header_at_offset(&wim->hdr, &wim->out_fd, hdr_offset);
2301         if (ret)
2302                 return ret;
2303
2304         /* Possibly sync file data to disk before closing.  On POSIX systems, it
2305          * is necessary to do this before using rename() to overwrite an
2306          * existing file with a new file.  Otherwise, data loss would occur if
2307          * the system is abruptly terminated when the metadata for the rename
2308          * operation has been written to disk, but the new file data has not.
2309          */
2310         if (write_flags & WIMLIB_WRITE_FLAG_FSYNC) {
2311                 DEBUG("Syncing WIM file.");
2312                 if (fsync(wim->out_fd.fd)) {
2313                         ERROR_WITH_ERRNO("Error syncing data to WIM file");
2314                         return WIMLIB_ERR_WRITE;
2315                 }
2316         }
2317
2318         if (close_wim_writable(wim, write_flags)) {
2319                 ERROR_WITH_ERRNO("Failed to close the output WIM file");
2320                 return WIMLIB_ERR_WRITE;
2321         }
2322
2323         return 0;
2324 }
2325
2326 #if defined(HAVE_SYS_FILE_H) && defined(HAVE_FLOCK)
2327 int
2328 lock_wim(WIMStruct *wim, int fd)
2329 {
2330         int ret = 0;
2331         if (fd != -1 && !wim->wim_locked) {
2332                 ret = flock(fd, LOCK_EX | LOCK_NB);
2333                 if (ret != 0) {
2334                         if (errno == EWOULDBLOCK) {
2335                                 ERROR("`%"TS"' is already being modified or has been "
2336                                       "mounted read-write\n"
2337                                       "        by another process!", wim->filename);
2338                                 ret = WIMLIB_ERR_ALREADY_LOCKED;
2339                         } else {
2340                                 WARNING_WITH_ERRNO("Failed to lock `%"TS"'",
2341                                                    wim->filename);
2342                                 ret = 0;
2343                         }
2344                 } else {
2345                         wim->wim_locked = 1;
2346                 }
2347         }
2348         return ret;
2349 }
2350 #endif
2351
2352 /*
2353  * write_pipable_wim():
2354  *
2355  * Perform the intermediate stages of creating a "pipable" WIM (i.e. a WIM
2356  * capable of being applied from a pipe).
2357  *
2358  * Pipable WIMs are a wimlib-specific modification of the WIM format such that
2359  * images can be applied from them sequentially when the file data is sent over
2360  * a pipe.  In addition, a pipable WIM can be written sequentially to a pipe.
2361  * The modifications made to the WIM format for pipable WIMs are:
2362  *
2363  * - Magic characters in header are "WLPWM\0\0\0" (wimlib pipable WIM) instead
2364  *   of "MSWIM\0\0\0".  This lets wimlib know that the WIM is pipable and also
2365  *   stops other software from trying to read the file as a normal WIM.
2366  *
2367  * - The header at the beginning of the file does not contain all the normal
2368  *   information; in particular it will have all 0's for the lookup table and
2369  *   XML data resource entries.  This is because this information cannot be
2370  *   determined until the lookup table and XML data have been written.
2371  *   Consequently, wimlib will write the full header at the very end of the
2372  *   file.  The header at the end, however, is only used when reading the WIM
2373  *   from a seekable file (not a pipe).
2374  *
2375  * - An extra copy of the XML data is placed directly after the header.  This
2376  *   allows image names and sizes to be determined at an appropriate time when
2377  *   reading the WIM from a pipe.  This copy of the XML data is ignored if the
2378  *   WIM is read from a seekable file (not a pipe).
2379  *
2380  * - The format of resources, or streams, has been modified to allow them to be
2381  *   used before the "lookup table" has been read.  Each stream is prefixed with
2382  *   a `struct pwm_stream_hdr' that is basically an abbreviated form of `struct
2383  *   wim_lookup_table_entry_disk' that only contains the SHA1 message digest,
2384  *   uncompressed stream size, and flags that indicate whether the stream is
2385  *   compressed.  The data of uncompressed streams then follows literally, while
2386  *   the data of compressed streams follows in a modified format.  Compressed
2387  *   streams do not begin with a chunk table, since the chunk table cannot be
2388  *   written until all chunks have been compressed.  Instead, each compressed
2389  *   chunk is prefixed by a `struct pwm_chunk_hdr' that gives its size.
2390  *   Furthermore, the chunk table is written at the end of the resource instead
2391  *   of the start.  Note: chunk offsets are given in the chunk table as if the
2392  *   `struct pwm_chunk_hdr's were not present; also, the chunk table is only
2393  *   used if the WIM is being read from a seekable file (not a pipe).
2394  *
2395  * - Metadata resources always come before other file resources (streams).
2396  *   (This does not by itself constitute an incompatibility with normal WIMs,
2397  *   since this is valid in normal WIMs.)
2398  *
2399  * - At least up to the end of the file resources, all components must be packed
2400  *   as tightly as possible; there cannot be any "holes" in the WIM.  (This does
2401  *   not by itself consititute an incompatibility with normal WIMs, since this
2402  *   is valid in normal WIMs.)
2403  *
2404  * Note: the lookup table, XML data, and header at the end are not used when
2405  * applying from a pipe.  They exist to support functionality such as image
2406  * application and export when the WIM is *not* read from a pipe.
2407  *
2408  *   Layout of pipable WIM:
2409  *
2410  * ---------+----------+--------------------+----------------+--------------+-----------+--------+
2411  * | Header | XML data | Metadata resources | File resources | Lookup table | XML data  | Header |
2412  * ---------+----------+--------------------+----------------+--------------+-----------+--------+
2413  *
2414  *   Layout of normal WIM:
2415  *
2416  * +--------+-----------------------------+-------------------------+
2417  * | Header | File and metadata resources | Lookup table | XML data |
2418  * +--------+-----------------------------+-------------------------+
2419  *
2420  * An optional integrity table can follow the final XML data in both normal and
2421  * pipable WIMs.  However, due to implementation details, wimlib currently can
2422  * only include an integrity table in a pipable WIM when writing it to a
2423  * seekable file (not a pipe).
2424  *
2425  * Do note that since pipable WIMs are not supported by Microsoft's software,
2426  * wimlib does not create them unless explicitly requested (with
2427  * WIMLIB_WRITE_FLAG_PIPABLE) and as stated above they use different magic
2428  * characters to identify the file.
2429  */
2430 static int
2431 write_pipable_wim(WIMStruct *wim, int image, int write_flags,
2432                   unsigned num_threads, wimlib_progress_func_t progress_func,
2433                   struct list_head *stream_list_override,
2434                   struct list_head *lookup_table_list_ret)
2435 {
2436         int ret;
2437         struct wim_reshdr xml_reshdr;
2438
2439         WARNING("Creating a pipable WIM, which will "
2440                 "be incompatible\n"
2441                 "          with Microsoft's software (wimgapi/imagex/Dism).");
2442
2443         /* At this point, the header at the beginning of the file has already
2444          * been written.  */
2445
2446         /* For efficiency, when wimlib adds an image to the WIM with
2447          * wimlib_add_image(), the SHA1 message digests of files is not
2448          * calculated; instead, they are calculated while the files are being
2449          * written.  However, this does not work when writing a pipable WIM,
2450          * since when writing a stream to a pipable WIM, its SHA1 message digest
2451          * needs to be known before the stream data is written.  Therefore,
2452          * before getting much farther, we need to pre-calculate the SHA1
2453          * message digests of all streams that will be written.  */
2454         ret = wim_checksum_unhashed_streams(wim);
2455         if (ret)
2456                 return ret;
2457
2458         /* Write extra copy of the XML data.  */
2459         ret = write_wim_xml_data(wim, image, WIM_TOTALBYTES_OMIT,
2460                                  &xml_reshdr,
2461                                  WRITE_RESOURCE_FLAG_PIPABLE);
2462         if (ret)
2463                 return ret;
2464
2465         /* Write metadata resources for the image(s) being included in the
2466          * output WIM.  */
2467         ret = write_wim_metadata_resources(wim, image, write_flags,
2468                                            progress_func);
2469         if (ret)
2470                 return ret;
2471
2472         /* Write streams needed for the image(s) being included in the output
2473          * WIM, or streams needed for the split WIM part.  */
2474         return write_wim_streams(wim, image, write_flags, num_threads,
2475                                  progress_func, stream_list_override,
2476                                  lookup_table_list_ret);
2477
2478         /* The lookup table, XML data, and header at end are handled by
2479          * finish_write().  */
2480 }
2481
2482 /* Write a standalone WIM or split WIM (SWM) part to a new file or to a file
2483  * descriptor.  */
2484 int
2485 write_wim_part(WIMStruct *wim,
2486                const void *path_or_fd,
2487                int image,
2488                int write_flags,
2489                unsigned num_threads,
2490                wimlib_progress_func_t progress_func,
2491                unsigned part_number,
2492                unsigned total_parts,
2493                struct list_head *stream_list_override,
2494                const u8 *guid)
2495 {
2496         int ret;
2497         struct wim_header hdr_save;
2498         struct list_head lookup_table_list;
2499
2500         if (total_parts == 1)
2501                 DEBUG("Writing standalone WIM.");
2502         else
2503                 DEBUG("Writing split WIM part %u/%u", part_number, total_parts);
2504         if (image == WIMLIB_ALL_IMAGES)
2505                 DEBUG("Including all images.");
2506         else
2507                 DEBUG("Including image %d only.", image);
2508         if (write_flags & WIMLIB_WRITE_FLAG_FILE_DESCRIPTOR)
2509                 DEBUG("File descriptor: %d", *(const int*)path_or_fd);
2510         else
2511                 DEBUG("Path: \"%"TS"\"", (const tchar*)path_or_fd);
2512         DEBUG("Write flags: 0x%08x", write_flags);
2513
2514         if (write_flags & WIMLIB_WRITE_FLAG_CHECK_INTEGRITY)
2515                 DEBUG("\tCHECK_INTEGRITY");
2516
2517         if (write_flags & WIMLIB_WRITE_FLAG_REBUILD)
2518                 DEBUG("\tREBUILD");
2519
2520         if (write_flags & WIMLIB_WRITE_FLAG_RECOMPRESS)
2521                 DEBUG("\tRECOMPRESS");
2522
2523         if (write_flags & WIMLIB_WRITE_FLAG_FSYNC)
2524                 DEBUG("\tFSYNC");
2525
2526         if (write_flags & WIMLIB_WRITE_FLAG_SOFT_DELETE)
2527                 DEBUG("\tFSYNC");
2528
2529         if (write_flags & WIMLIB_WRITE_FLAG_IGNORE_READONLY_FLAG)
2530                 DEBUG("\tIGNORE_READONLY_FLAG");
2531
2532         if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
2533                 DEBUG("\tPIPABLE");
2534
2535         if (write_flags & WIMLIB_WRITE_FLAG_NOT_PIPABLE)
2536                 DEBUG("\tNOT_PIPABLE");
2537
2538         if (write_flags & WIMLIB_WRITE_FLAG_PACK_STREAMS)
2539                 DEBUG("\tPACK_STREAMS");
2540
2541         if (write_flags & WIMLIB_WRITE_FLAG_FILE_DESCRIPTOR)
2542                 DEBUG("\tFILE_DESCRIPTOR");
2543
2544         if (write_flags & WIMLIB_WRITE_FLAG_NO_METADATA)
2545                 DEBUG("\tNO_METADATA");
2546
2547         if (write_flags & WIMLIB_WRITE_FLAG_USE_EXISTING_TOTALBYTES)
2548                 DEBUG("\tUSE_EXISTING_TOTALBYTES");
2549
2550         if (num_threads == 0)
2551                 DEBUG("Number of threads: autodetect");
2552         else
2553                 DEBUG("Number of threads: %u", num_threads);
2554         DEBUG("Progress function: %s", (progress_func ? "yes" : "no"));
2555         DEBUG("Stream list:       %s", (stream_list_override ? "specified" : "autodetect"));
2556         DEBUG("GUID:              %s", ((guid || wim->guid_set_explicitly) ?
2557                                         "specified" : "generate new"));
2558
2559         /* Internally, this is always called with a valid part number and total
2560          * parts.  */
2561         wimlib_assert(total_parts >= 1);
2562         wimlib_assert(part_number >= 1 && part_number <= total_parts);
2563
2564         /* A valid image (or all images) must be specified.  */
2565         if (image != WIMLIB_ALL_IMAGES &&
2566              (image < 1 || image > wim->hdr.image_count))
2567                 return WIMLIB_ERR_INVALID_IMAGE;
2568
2569         /* If we need to write metadata resources, make sure the ::WIMStruct has
2570          * the needed information attached (e.g. is not a resource-only WIM,
2571          * such as a non-first part of a split WIM).  */
2572         if (!wim_has_metadata(wim) &&
2573             !(write_flags & WIMLIB_WRITE_FLAG_NO_METADATA))
2574                 return WIMLIB_ERR_METADATA_NOT_FOUND;
2575
2576         /* Check for contradictory flags.  */
2577         if ((write_flags & (WIMLIB_WRITE_FLAG_CHECK_INTEGRITY |
2578                             WIMLIB_WRITE_FLAG_NO_CHECK_INTEGRITY))
2579                                 == (WIMLIB_WRITE_FLAG_CHECK_INTEGRITY |
2580                                     WIMLIB_WRITE_FLAG_NO_CHECK_INTEGRITY))
2581                 return WIMLIB_ERR_INVALID_PARAM;
2582
2583         if ((write_flags & (WIMLIB_WRITE_FLAG_PIPABLE |
2584                             WIMLIB_WRITE_FLAG_NOT_PIPABLE))
2585                                 == (WIMLIB_WRITE_FLAG_PIPABLE |
2586                                     WIMLIB_WRITE_FLAG_NOT_PIPABLE))
2587                 return WIMLIB_ERR_INVALID_PARAM;
2588
2589         /* Save previous header, then start initializing the new one.  */
2590         memcpy(&hdr_save, &wim->hdr, sizeof(struct wim_header));
2591
2592         /* Set default integrity, pipable, and packed stream flags.  */
2593         if (!(write_flags & (WIMLIB_WRITE_FLAG_PIPABLE |
2594                              WIMLIB_WRITE_FLAG_NOT_PIPABLE)))
2595                 if (wim_is_pipable(wim)) {
2596                         DEBUG("WIM is pipable; default to PIPABLE.");
2597                         write_flags |= WIMLIB_WRITE_FLAG_PIPABLE;
2598                 }
2599
2600         if (!(write_flags & (WIMLIB_WRITE_FLAG_CHECK_INTEGRITY |
2601                              WIMLIB_WRITE_FLAG_NO_CHECK_INTEGRITY)))
2602                 if (wim_has_integrity_table(wim)) {
2603                         DEBUG("Integrity table present; default to CHECK_INTEGRITY.");
2604                         write_flags |= WIMLIB_WRITE_FLAG_CHECK_INTEGRITY;
2605                 }
2606
2607         if ((write_flags & (WIMLIB_WRITE_FLAG_PIPABLE |
2608                             WIMLIB_WRITE_FLAG_PACK_STREAMS))
2609                                     == (WIMLIB_WRITE_FLAG_PIPABLE |
2610                                         WIMLIB_WRITE_FLAG_PACK_STREAMS))
2611         {
2612                 ERROR("Cannot specify both PIPABLE and PACK_STREAMS!");
2613                 return WIMLIB_ERR_INVALID_PARAM;
2614         }
2615
2616         /* Set appropriate magic number.  */
2617         if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
2618                 wim->hdr.magic = PWM_MAGIC;
2619         else
2620                 wim->hdr.magic = WIM_MAGIC;
2621
2622         /* Set appropriate version number.  */
2623         if ((write_flags & WIMLIB_WRITE_FLAG_PACK_STREAMS) ||
2624             wim->out_compression_type == WIMLIB_COMPRESSION_TYPE_LZMS)
2625                 wim->hdr.wim_version = WIM_VERSION_PACKED_STREAMS;
2626         else
2627                 wim->hdr.wim_version = WIM_VERSION_DEFAULT;
2628
2629         /* Clear header flags that will be set automatically.  */
2630         wim->hdr.flags &= ~(WIM_HDR_FLAG_METADATA_ONLY          |
2631                             WIM_HDR_FLAG_RESOURCE_ONLY          |
2632                             WIM_HDR_FLAG_SPANNED                |
2633                             WIM_HDR_FLAG_WRITE_IN_PROGRESS);
2634
2635         /* Set SPANNED header flag if writing part of a split WIM.  */
2636         if (total_parts != 1)
2637                 wim->hdr.flags |= WIM_HDR_FLAG_SPANNED;
2638
2639         /* Set part number and total parts of split WIM.  This will be 1 and 1
2640          * if the WIM is standalone.  */
2641         wim->hdr.part_number = part_number;
2642         wim->hdr.total_parts = total_parts;
2643
2644         /* Set compression type if different.  */
2645         if (wim->compression_type != wim->out_compression_type) {
2646                 ret = set_wim_hdr_cflags(wim->out_compression_type, &wim->hdr);
2647                 wimlib_assert(ret == 0);
2648         }
2649
2650         /* Set chunk size if different.  */
2651         wim->hdr.chunk_size = wim->out_chunk_size;
2652
2653         /* Use GUID if specified; otherwise generate a new one.  */
2654         if (guid)
2655                 memcpy(wim->hdr.guid, guid, WIMLIB_GUID_LEN);
2656         else if (!wim->guid_set_explicitly)
2657                 randomize_byte_array(wim->hdr.guid, WIMLIB_GUID_LEN);
2658
2659         /* Clear references to resources that have not been written yet.  */
2660         zero_reshdr(&wim->hdr.lookup_table_reshdr);
2661         zero_reshdr(&wim->hdr.xml_data_reshdr);
2662         zero_reshdr(&wim->hdr.boot_metadata_reshdr);
2663         zero_reshdr(&wim->hdr.integrity_table_reshdr);
2664
2665         /* Set image count and boot index correctly for single image writes.  */
2666         if (image != WIMLIB_ALL_IMAGES) {
2667                 wim->hdr.image_count = 1;
2668                 if (wim->hdr.boot_idx == image)
2669                         wim->hdr.boot_idx = 1;
2670                 else
2671                         wim->hdr.boot_idx = 0;
2672         }
2673
2674         /* Split WIMs can't be bootable.  */
2675         if (total_parts != 1)
2676                 wim->hdr.boot_idx = 0;
2677
2678         /* Initialize output file descriptor.  */
2679         if (write_flags & WIMLIB_WRITE_FLAG_FILE_DESCRIPTOR) {
2680                 /* File descriptor was explicitly provided.  Return error if
2681                  * file descriptor is not seekable, unless writing a pipable WIM
2682                  * was requested.  */
2683                 wim->out_fd.fd = *(const int*)path_or_fd;
2684                 wim->out_fd.offset = 0;
2685                 if (!filedes_is_seekable(&wim->out_fd)) {
2686                         ret = WIMLIB_ERR_INVALID_PARAM;
2687                         if (!(write_flags & WIMLIB_WRITE_FLAG_PIPABLE))
2688                                 goto out_restore_hdr;
2689                         if (write_flags & WIMLIB_WRITE_FLAG_CHECK_INTEGRITY) {
2690                                 ERROR("Can't include integrity check when "
2691                                       "writing pipable WIM to pipe!");
2692                                 goto out_restore_hdr;
2693                         }
2694                 }
2695
2696         } else {
2697                 /* Filename of WIM to write was provided; open file descriptor
2698                  * to it.  */
2699                 ret = open_wim_writable(wim, (const tchar*)path_or_fd,
2700                                         O_TRUNC | O_CREAT | O_RDWR);
2701                 if (ret)
2702                         goto out_restore_hdr;
2703         }
2704
2705         /* Write initial header.  This is merely a "dummy" header since it
2706          * doesn't have all the information yet, so it will be overwritten later
2707          * (unless writing a pipable WIM).  */
2708         if (!(write_flags & WIMLIB_WRITE_FLAG_PIPABLE))
2709                 wim->hdr.flags |= WIM_HDR_FLAG_WRITE_IN_PROGRESS;
2710         ret = write_wim_header(&wim->hdr, &wim->out_fd);
2711         wim->hdr.flags &= ~WIM_HDR_FLAG_WRITE_IN_PROGRESS;
2712         if (ret)
2713                 goto out_restore_hdr;
2714
2715         /* Write metadata resources and streams.  */
2716         if (!(write_flags & WIMLIB_WRITE_FLAG_PIPABLE)) {
2717                 /* Default case: create a normal (non-pipable) WIM.  */
2718                 ret = write_wim_streams(wim, image, write_flags, num_threads,
2719                                         progress_func, stream_list_override,
2720                                         &lookup_table_list);
2721                 if (ret)
2722                         goto out_restore_hdr;
2723
2724                 ret = write_wim_metadata_resources(wim, image, write_flags,
2725                                                    progress_func);
2726                 if (ret)
2727                         goto out_restore_hdr;
2728         } else {
2729                 /* Non-default case: create pipable WIM.  */
2730                 ret = write_pipable_wim(wim, image, write_flags, num_threads,
2731                                         progress_func, stream_list_override,
2732                                         &lookup_table_list);
2733                 if (ret)
2734                         goto out_restore_hdr;
2735                 write_flags |= WIMLIB_WRITE_FLAG_HEADER_AT_END;
2736         }
2737
2738
2739         /* Write lookup table, XML data, and (optional) integrity table.  */
2740         ret = finish_write(wim, image, write_flags, progress_func,
2741                            &lookup_table_list);
2742 out_restore_hdr:
2743         memcpy(&wim->hdr, &hdr_save, sizeof(struct wim_header));
2744         (void)close_wim_writable(wim, write_flags);
2745         DEBUG("ret=%d", ret);
2746         return ret;
2747 }
2748
2749 /* Write a standalone WIM to a file or file descriptor.  */
2750 static int
2751 write_standalone_wim(WIMStruct *wim, const void *path_or_fd,
2752                      int image, int write_flags, unsigned num_threads,
2753                      wimlib_progress_func_t progress_func)
2754 {
2755         return write_wim_part(wim, path_or_fd, image, write_flags,
2756                               num_threads, progress_func, 1, 1, NULL, NULL);
2757 }
2758
2759 /* API function documented in wimlib.h  */
2760 WIMLIBAPI int
2761 wimlib_write(WIMStruct *wim, const tchar *path,
2762              int image, int write_flags, unsigned num_threads,
2763              wimlib_progress_func_t progress_func)
2764 {
2765         if (!path)
2766                 return WIMLIB_ERR_INVALID_PARAM;
2767
2768         write_flags &= WIMLIB_WRITE_MASK_PUBLIC;
2769
2770         return write_standalone_wim(wim, path, image, write_flags,
2771                                     num_threads, progress_func);
2772 }
2773
2774 /* API function documented in wimlib.h  */
2775 WIMLIBAPI int
2776 wimlib_write_to_fd(WIMStruct *wim, int fd,
2777                    int image, int write_flags, unsigned num_threads,
2778                    wimlib_progress_func_t progress_func)
2779 {
2780         if (fd < 0)
2781                 return WIMLIB_ERR_INVALID_PARAM;
2782
2783         write_flags &= WIMLIB_WRITE_MASK_PUBLIC;
2784         write_flags |= WIMLIB_WRITE_FLAG_FILE_DESCRIPTOR;
2785
2786         return write_standalone_wim(wim, &fd, image, write_flags,
2787                                     num_threads, progress_func);
2788 }
2789
2790 static bool
2791 any_images_modified(WIMStruct *wim)
2792 {
2793         for (int i = 0; i < wim->hdr.image_count; i++)
2794                 if (wim->image_metadata[i]->modified)
2795                         return true;
2796         return false;
2797 }
2798
2799 static int
2800 check_resource_offset(struct wim_lookup_table_entry *lte, void *_wim)
2801 {
2802         const WIMStruct *wim = _wim;
2803         off_t end_offset = *(const off_t*)wim->private;
2804
2805         if (lte->resource_location == RESOURCE_IN_WIM && lte->rspec->wim == wim &&
2806             lte->rspec->offset_in_wim + lte->rspec->size_in_wim > end_offset)
2807                 return WIMLIB_ERR_RESOURCE_ORDER;
2808         return 0;
2809 }
2810
2811 /* Make sure no file or metadata resources are located after the XML data (or
2812  * integrity table if present)--- otherwise we can't safely overwrite the WIM in
2813  * place and we return WIMLIB_ERR_RESOURCE_ORDER.  */
2814 static int
2815 check_resource_offsets(WIMStruct *wim, off_t end_offset)
2816 {
2817         int ret;
2818         unsigned i;
2819
2820         wim->private = &end_offset;
2821         ret = for_lookup_table_entry(wim->lookup_table, check_resource_offset, wim);
2822         if (ret)
2823                 return ret;
2824
2825         for (i = 0; i < wim->hdr.image_count; i++) {
2826                 ret = check_resource_offset(wim->image_metadata[i]->metadata_lte, wim);
2827                 if (ret)
2828                         return ret;
2829         }
2830         return 0;
2831 }
2832
2833 /*
2834  * Overwrite a WIM, possibly appending streams to it.
2835  *
2836  * A WIM looks like (or is supposed to look like) the following:
2837  *
2838  *                   Header (212 bytes)
2839  *                   Streams and metadata resources (variable size)
2840  *                   Lookup table (variable size)
2841  *                   XML data (variable size)
2842  *                   Integrity table (optional) (variable size)
2843  *
2844  * If we are not adding any streams or metadata resources, the lookup table is
2845  * unchanged--- so we only need to overwrite the XML data, integrity table, and
2846  * header.  This operation is potentially unsafe if the program is abruptly
2847  * terminated while the XML data or integrity table are being overwritten, but
2848  * before the new header has been written.  To partially alleviate this problem,
2849  * a special flag (WIMLIB_WRITE_FLAG_CHECKPOINT_AFTER_XML) is passed to
2850  * finish_write() to cause a temporary WIM header to be written after the XML
2851  * data has been written.  This may prevent the WIM from becoming corrupted if
2852  * the program is terminated while the integrity table is being calculated (but
2853  * no guarantees, due to write re-ordering...).
2854  *
2855  * If we are adding new streams or images (metadata resources), the lookup table
2856  * needs to be changed, and those streams need to be written.  In this case, we
2857  * try to perform a safe update of the WIM file by writing the streams *after*
2858  * the end of the previous WIM, then writing the new lookup table, XML data, and
2859  * (optionally) integrity table following the new streams.  This will produce a
2860  * layout like the following:
2861  *
2862  *                   Header (212 bytes)
2863  *                   (OLD) Streams and metadata resources (variable size)
2864  *                   (OLD) Lookup table (variable size)
2865  *                   (OLD) XML data (variable size)
2866  *                   (OLD) Integrity table (optional) (variable size)
2867  *                   (NEW) Streams and metadata resources (variable size)
2868  *                   (NEW) Lookup table (variable size)
2869  *                   (NEW) XML data (variable size)
2870  *                   (NEW) Integrity table (optional) (variable size)
2871  *
2872  * At all points, the WIM is valid as nothing points to the new data yet.  Then,
2873  * the header is overwritten to point to the new lookup table, XML data, and
2874  * integrity table, to produce the following layout:
2875  *
2876  *                   Header (212 bytes)
2877  *                   Streams and metadata resources (variable size)
2878  *                   Nothing (variable size)
2879  *                   More Streams and metadata resources (variable size)
2880  *                   Lookup table (variable size)
2881  *                   XML data (variable size)
2882  *                   Integrity table (optional) (variable size)
2883  *
2884  * This method allows an image to be appended to a large WIM very quickly, and
2885  * is is crash-safe except in the case of write re-ordering, but the
2886  * disadvantage is that a small hole is left in the WIM where the old lookup
2887  * table, xml data, and integrity table were.  (These usually only take up a
2888  * small amount of space compared to the streams, however.)
2889  */
2890 static int
2891 overwrite_wim_inplace(WIMStruct *wim, int write_flags,
2892                       unsigned num_threads,
2893                       wimlib_progress_func_t progress_func)
2894 {
2895         int ret;
2896         off_t old_wim_end;
2897         u64 old_lookup_table_end, old_xml_begin, old_xml_end;
2898         struct wim_header hdr_save;
2899         struct list_head stream_list;
2900         struct list_head lookup_table_list;
2901         struct filter_context filter_ctx;
2902
2903         DEBUG("Overwriting `%"TS"' in-place", wim->filename);
2904
2905         /* Save original header so it can be restored in case of error  */
2906         memcpy(&hdr_save, &wim->hdr, sizeof(struct wim_header));
2907
2908         /* Set default integrity flag.  */
2909         if (!(write_flags & (WIMLIB_WRITE_FLAG_CHECK_INTEGRITY |
2910                              WIMLIB_WRITE_FLAG_NO_CHECK_INTEGRITY)))
2911                 if (wim_has_integrity_table(wim))
2912                         write_flags |= WIMLIB_WRITE_FLAG_CHECK_INTEGRITY;
2913
2914         /* Set WIM version if adding packed streams.  */
2915         if (write_flags & WIMLIB_WRITE_FLAG_PACK_STREAMS)
2916                 wim->hdr.wim_version = WIM_VERSION_PACKED_STREAMS;
2917
2918         /* Set additional flags for overwrite.  */
2919         write_flags |= WIMLIB_WRITE_FLAG_OVERWRITE |
2920                        WIMLIB_WRITE_FLAG_STREAMS_OK;
2921
2922         /* Make sure that the integrity table (if present) is after the XML
2923          * data, and that there are no stream resources, metadata resources, or
2924          * lookup tables after the XML data.  Otherwise, these data would be
2925          * overwritten. */
2926         old_xml_begin = wim->hdr.xml_data_reshdr.offset_in_wim;
2927         old_xml_end = old_xml_begin + wim->hdr.xml_data_reshdr.size_in_wim;
2928         old_lookup_table_end = wim->hdr.lookup_table_reshdr.offset_in_wim +
2929                                wim->hdr.lookup_table_reshdr.size_in_wim;
2930         if (wim->hdr.integrity_table_reshdr.offset_in_wim != 0 &&
2931             wim->hdr.integrity_table_reshdr.offset_in_wim < old_xml_end) {
2932                 WARNING("Didn't expect the integrity table to be before the XML data");
2933                 ret = WIMLIB_ERR_RESOURCE_ORDER;
2934                 goto out_restore_memory_hdr;
2935         }
2936
2937         if (old_lookup_table_end > old_xml_begin) {
2938                 WARNING("Didn't expect the lookup table to be after the XML data");
2939                 ret = WIMLIB_ERR_RESOURCE_ORDER;
2940                 goto out_restore_memory_hdr;
2941         }
2942
2943         /* Set @old_wim_end, which indicates the point beyond which we don't
2944          * allow any file and metadata resources to appear without returning
2945          * WIMLIB_ERR_RESOURCE_ORDER (due to the fact that we would otherwise
2946          * overwrite these resources). */
2947         if (!wim->deletion_occurred && !any_images_modified(wim)) {
2948                 /* If no images have been modified and no images have been
2949                  * deleted, a new lookup table does not need to be written.  We
2950                  * shall write the new XML data and optional integrity table
2951                  * immediately after the lookup table.  Note that this may
2952                  * overwrite an existing integrity table. */
2953                 DEBUG("Skipping writing lookup table "
2954                       "(no images modified or deleted)");
2955                 old_wim_end = old_lookup_table_end;
2956                 write_flags |= WIMLIB_WRITE_FLAG_NO_LOOKUP_TABLE |
2957                                WIMLIB_WRITE_FLAG_CHECKPOINT_AFTER_XML;
2958         } else if (wim->hdr.integrity_table_reshdr.offset_in_wim != 0) {
2959                 /* Old WIM has an integrity table; begin writing new streams
2960                  * after it. */
2961                 old_wim_end = wim->hdr.integrity_table_reshdr.offset_in_wim +
2962                               wim->hdr.integrity_table_reshdr.size_in_wim;
2963         } else {
2964                 /* No existing integrity table; begin writing new streams after
2965                  * the old XML data. */
2966                 old_wim_end = old_xml_end;
2967         }
2968
2969         ret = check_resource_offsets(wim, old_wim_end);
2970         if (ret)
2971                 goto out_restore_memory_hdr;
2972
2973         ret = prepare_stream_list_for_write(wim, WIMLIB_ALL_IMAGES, write_flags,
2974                                             &stream_list, &lookup_table_list,
2975                                             &filter_ctx);
2976         if (ret)
2977                 goto out_restore_memory_hdr;
2978
2979         ret = open_wim_writable(wim, wim->filename, O_RDWR);
2980         if (ret)
2981                 goto out_restore_memory_hdr;
2982
2983         ret = lock_wim(wim, wim->out_fd.fd);
2984         if (ret)
2985                 goto out_close_wim;
2986
2987         /* Set WIM_HDR_FLAG_WRITE_IN_PROGRESS flag in header. */
2988         wim->hdr.flags |= WIM_HDR_FLAG_WRITE_IN_PROGRESS;
2989         ret = write_wim_header_flags(wim->hdr.flags, &wim->out_fd);
2990         if (ret) {
2991                 ERROR_WITH_ERRNO("Error updating WIM header flags");
2992                 goto out_unlock_wim;
2993         }
2994
2995         if (filedes_seek(&wim->out_fd, old_wim_end) == -1) {
2996                 ERROR_WITH_ERRNO("Can't seek to end of WIM");
2997                 ret = WIMLIB_ERR_WRITE;
2998                 goto out_restore_physical_hdr;
2999         }
3000
3001         ret = wim_write_stream_list(wim,
3002                                     &stream_list,
3003                                     write_flags,
3004                                     num_threads,
3005                                     &filter_ctx,
3006                                     progress_func);
3007         if (ret)
3008                 goto out_truncate;
3009
3010         ret = write_wim_metadata_resources(wim, WIMLIB_ALL_IMAGES,
3011                                            write_flags, progress_func);
3012         if (ret)
3013                 goto out_truncate;
3014
3015         write_flags |= WIMLIB_WRITE_FLAG_REUSE_INTEGRITY_TABLE;
3016         ret = finish_write(wim, WIMLIB_ALL_IMAGES, write_flags,
3017                            progress_func, &lookup_table_list);
3018         if (ret)
3019                 goto out_truncate;
3020
3021         wim->wim_locked = 0;
3022         return 0;
3023
3024 out_truncate:
3025         if (!(write_flags & WIMLIB_WRITE_FLAG_NO_LOOKUP_TABLE)) {
3026                 WARNING("Truncating `%"TS"' to its original size (%"PRIu64" bytes)",
3027                         wim->filename, old_wim_end);
3028                 /* Return value of ftruncate() is ignored because this is
3029                  * already an error path.  */
3030                 (void)ftruncate(wim->out_fd.fd, old_wim_end);
3031         }
3032 out_restore_physical_hdr:
3033         (void)write_wim_header_flags(hdr_save.flags, &wim->out_fd);
3034 out_unlock_wim:
3035         wim->wim_locked = 0;
3036 out_close_wim:
3037         (void)close_wim_writable(wim, write_flags);
3038 out_restore_memory_hdr:
3039         memcpy(&wim->hdr, &hdr_save, sizeof(struct wim_header));
3040         return ret;
3041 }
3042
3043 static int
3044 overwrite_wim_via_tmpfile(WIMStruct *wim, int write_flags,
3045                           unsigned num_threads,
3046                           wimlib_progress_func_t progress_func)
3047 {
3048         size_t wim_name_len;
3049         int ret;
3050
3051         DEBUG("Overwriting `%"TS"' via a temporary file", wim->filename);
3052
3053         /* Write the WIM to a temporary file in the same directory as the
3054          * original WIM. */
3055         wim_name_len = tstrlen(wim->filename);
3056         tchar tmpfile[wim_name_len + 10];
3057         tmemcpy(tmpfile, wim->filename, wim_name_len);
3058         randomize_char_array_with_alnum(tmpfile + wim_name_len, 9);
3059         tmpfile[wim_name_len + 9] = T('\0');
3060
3061         ret = wimlib_write(wim, tmpfile, WIMLIB_ALL_IMAGES,
3062                            write_flags | WIMLIB_WRITE_FLAG_FSYNC,
3063                            num_threads, progress_func);
3064         if (ret) {
3065                 tunlink(tmpfile);
3066                 return ret;
3067         }
3068
3069         close_wim(wim);
3070
3071         /* Rename the new WIM file to the original WIM file.  Note: on Windows
3072          * this actually calls win32_rename_replacement(), not _wrename(), so
3073          * that removing the existing destination file can be handled.  */
3074         DEBUG("Renaming `%"TS"' to `%"TS"'", tmpfile, wim->filename);
3075         ret = trename(tmpfile, wim->filename);
3076         if (ret) {
3077                 ERROR_WITH_ERRNO("Failed to rename `%"TS"' to `%"TS"'",
3078                                  tmpfile, wim->filename);
3079         #ifdef __WIN32__
3080                 if (ret < 0)
3081         #endif
3082                 {
3083                         tunlink(tmpfile);
3084                 }
3085                 return WIMLIB_ERR_RENAME;
3086         }
3087
3088         if (progress_func) {
3089                 union wimlib_progress_info progress;
3090                 progress.rename.from = tmpfile;
3091                 progress.rename.to = wim->filename;
3092                 progress_func(WIMLIB_PROGRESS_MSG_RENAME, &progress);
3093         }
3094         return 0;
3095 }
3096
3097 /* Determine if the specified WIM file may be updated by appending in-place
3098  * rather than writing and replacing it with an entirely new file.  */
3099 static bool
3100 can_overwrite_wim_inplace(const WIMStruct *wim, int write_flags)
3101 {
3102         /* REBUILD flag forces full rebuild.  */
3103         if (write_flags & WIMLIB_WRITE_FLAG_REBUILD)
3104                 return false;
3105
3106         /* Deletions cause full rebuild by default.  */
3107         if (wim->deletion_occurred && !(write_flags & WIMLIB_WRITE_FLAG_SOFT_DELETE))
3108                 return false;
3109
3110         /* Pipable WIMs cannot be updated in place, nor can a non-pipable WIM be
3111          * turned into a pipable WIM in-place.  */
3112         if (wim_is_pipable(wim) || (write_flags & WIMLIB_WRITE_FLAG_PIPABLE))
3113                 return false;
3114
3115         /* wimlib allows multiple packs in a single WIM, but they don't seem to
3116          * be compatible with WIMGAPI, so force all streams to be repacked if
3117          * the WIM already may have contained a pack and PACK_STREAMS was
3118          * requested.  */
3119         if (write_flags & WIMLIB_WRITE_FLAG_PACK_STREAMS &&
3120             wim->hdr.wim_version == WIM_VERSION_PACKED_STREAMS)
3121                 return false;
3122
3123         /* The default compression type and compression chunk size selected for
3124          * the output WIM must be the same as those currently used for the WIM.
3125          */
3126         if (wim->compression_type != wim->out_compression_type)
3127                 return false;
3128         if (wim->chunk_size != wim->out_chunk_size)
3129                 return false;
3130
3131         return true;
3132 }
3133
3134 /* API function documented in wimlib.h  */
3135 WIMLIBAPI int
3136 wimlib_overwrite(WIMStruct *wim, int write_flags,
3137                  unsigned num_threads,
3138                  wimlib_progress_func_t progress_func)
3139 {
3140         int ret;
3141         u32 orig_hdr_flags;
3142
3143         write_flags &= WIMLIB_WRITE_MASK_PUBLIC;
3144
3145         if (write_flags & WIMLIB_WRITE_FLAG_FILE_DESCRIPTOR)
3146                 return WIMLIB_ERR_INVALID_PARAM;
3147
3148         if (!wim->filename)
3149                 return WIMLIB_ERR_NO_FILENAME;
3150
3151         orig_hdr_flags = wim->hdr.flags;
3152         if (write_flags & WIMLIB_WRITE_FLAG_IGNORE_READONLY_FLAG)
3153                 wim->hdr.flags &= ~WIM_HDR_FLAG_READONLY;
3154         ret = can_modify_wim(wim);
3155         wim->hdr.flags = orig_hdr_flags;
3156         if (ret)
3157                 return ret;
3158
3159         if (can_overwrite_wim_inplace(wim, write_flags)) {
3160                 ret = overwrite_wim_inplace(wim, write_flags, num_threads,
3161                                             progress_func);
3162                 if (ret != WIMLIB_ERR_RESOURCE_ORDER)
3163                         return ret;
3164                 WARNING("Falling back to re-building entire WIM");
3165         }
3166         return overwrite_wim_via_tmpfile(wim, write_flags, num_threads,
3167                                          progress_func);
3168 }