liblzma: CRC CLMUL: Omit is_arch_extension_supported() when not needed
[xz/debian.git] / src / liblzma / common / file_info.c
blob7c85084a706e2c71c177a6b944b119dbae67eac0
1 // SPDX-License-Identifier: 0BSD
3 ///////////////////////////////////////////////////////////////////////////////
4 //
5 /// \file file_info.c
6 /// \brief Decode .xz file information into a lzma_index structure
7 //
8 // Author: Lasse Collin
9 //
10 ///////////////////////////////////////////////////////////////////////////////
12 #include "index_decoder.h"
15 typedef struct {
16 enum {
17 SEQ_MAGIC_BYTES,
18 SEQ_PADDING_SEEK,
19 SEQ_PADDING_DECODE,
20 SEQ_FOOTER,
21 SEQ_INDEX_INIT,
22 SEQ_INDEX_DECODE,
23 SEQ_HEADER_DECODE,
24 SEQ_HEADER_COMPARE,
25 } sequence;
27 /// Absolute position of in[*in_pos] in the file. All code that
28 /// modifies *in_pos also updates this. seek_to_pos() needs this
29 /// to determine if we need to request the application to seek for
30 /// us or if we can do the seeking internally by adjusting *in_pos.
31 uint64_t file_cur_pos;
33 /// This refers to absolute positions of interesting parts of the
34 /// input file. Sometimes it points to the *beginning* of a specific
35 /// field and sometimes to the *end* of a field. The current target
36 /// position at each moment is explained in the comments.
37 uint64_t file_target_pos;
39 /// Size of the .xz file (from the application).
40 uint64_t file_size;
42 /// Index decoder
43 lzma_next_coder index_decoder;
45 /// Number of bytes remaining in the Index field that is currently
46 /// being decoded.
47 lzma_vli index_remaining;
49 /// The Index decoder will store the decoded Index in this pointer.
50 lzma_index *this_index;
52 /// Amount of Stream Padding in the current Stream.
53 lzma_vli stream_padding;
55 /// The final combined index is collected here.
56 lzma_index *combined_index;
58 /// Pointer from the application where to store the index information
59 /// after successful decoding.
60 lzma_index **dest_index;
62 /// Pointer to lzma_stream.seek_pos to be used when returning
63 /// LZMA_SEEK_NEEDED. This is set by seek_to_pos() when needed.
64 uint64_t *external_seek_pos;
66 /// Memory usage limit
67 uint64_t memlimit;
69 /// Stream Flags from the very beginning of the file.
70 lzma_stream_flags first_header_flags;
72 /// Stream Flags from Stream Header of the current Stream.
73 lzma_stream_flags header_flags;
75 /// Stream Flags from Stream Footer of the current Stream.
76 lzma_stream_flags footer_flags;
78 size_t temp_pos;
79 size_t temp_size;
80 uint8_t temp[8192];
82 } lzma_file_info_coder;
85 /// Copies data from in[*in_pos] into coder->temp until
86 /// coder->temp_pos == coder->temp_size. This also keeps coder->file_cur_pos
87 /// in sync with *in_pos. Returns true if more input is needed.
88 static bool
89 fill_temp(lzma_file_info_coder *coder, const uint8_t *restrict in,
90 size_t *restrict in_pos, size_t in_size)
92 coder->file_cur_pos += lzma_bufcpy(in, in_pos, in_size,
93 coder->temp, &coder->temp_pos, coder->temp_size);
94 return coder->temp_pos < coder->temp_size;
98 /// Seeks to the absolute file position specified by target_pos.
99 /// This tries to do the seeking by only modifying *in_pos, if possible.
100 /// The main benefit of this is that if one passes the whole file at once
101 /// to lzma_code(), the decoder will never need to return LZMA_SEEK_NEEDED
102 /// as all the seeking can be done by adjusting *in_pos in this function.
104 /// Returns true if an external seek is needed and the caller must return
105 /// LZMA_SEEK_NEEDED.
106 static bool
107 seek_to_pos(lzma_file_info_coder *coder, uint64_t target_pos,
108 size_t in_start, size_t *in_pos, size_t in_size)
110 // The input buffer doesn't extend beyond the end of the file.
111 // This has been checked by file_info_decode() already.
112 assert(coder->file_size - coder->file_cur_pos >= in_size - *in_pos);
114 const uint64_t pos_min = coder->file_cur_pos - (*in_pos - in_start);
115 const uint64_t pos_max = coder->file_cur_pos + (in_size - *in_pos);
117 bool external_seek_needed;
119 if (target_pos >= pos_min && target_pos <= pos_max) {
120 // The requested position is available in the current input
121 // buffer or right after it. That is, in a corner case we
122 // end up setting *in_pos == in_size and thus will immediately
123 // need new input bytes from the application.
124 *in_pos += (size_t)(target_pos - coder->file_cur_pos);
125 external_seek_needed = false;
126 } else {
127 // Ask the application to seek the input file.
128 *coder->external_seek_pos = target_pos;
129 external_seek_needed = true;
131 // Mark the whole input buffer as used. This way
132 // lzma_stream.total_in will have a better estimate
133 // of the amount of data read. It still won't be perfect
134 // as the value will depend on the input buffer size that
135 // the application uses, but it should be good enough for
136 // those few who want an estimate.
137 *in_pos = in_size;
140 // After seeking (internal or external) the current position
141 // will match the requested target position.
142 coder->file_cur_pos = target_pos;
144 return external_seek_needed;
148 /// The caller sets coder->file_target_pos so that it points to the *end*
149 /// of the desired file position. This function then determines how far
150 /// backwards from that position we can seek. After seeking fill_temp()
151 /// can be used to read data into coder->temp. When fill_temp() has finished,
152 /// coder->temp[coder->temp_size] will match coder->file_target_pos.
154 /// This also validates that coder->target_file_pos is sane in sense that
155 /// we aren't trying to seek too far backwards (too close or beyond the
156 /// beginning of the file).
157 static lzma_ret
158 reverse_seek(lzma_file_info_coder *coder,
159 size_t in_start, size_t *in_pos, size_t in_size)
161 // Check that there is enough data before the target position
162 // to contain at least Stream Header and Stream Footer. If there
163 // isn't, the file cannot be valid.
164 if (coder->file_target_pos < 2 * LZMA_STREAM_HEADER_SIZE)
165 return LZMA_DATA_ERROR;
167 coder->temp_pos = 0;
169 // The Stream Header at the very beginning of the file gets handled
170 // specially in SEQ_MAGIC_BYTES and thus we will never need to seek
171 // there. By not seeking to the first LZMA_STREAM_HEADER_SIZE bytes
172 // we avoid a useless external seek after SEQ_MAGIC_BYTES if the
173 // application uses an extremely small input buffer and the input
174 // file is very small.
175 if (coder->file_target_pos - LZMA_STREAM_HEADER_SIZE
176 < sizeof(coder->temp))
177 coder->temp_size = (size_t)(coder->file_target_pos
178 - LZMA_STREAM_HEADER_SIZE);
179 else
180 coder->temp_size = sizeof(coder->temp);
182 // The above if-statements guarantee this. This is important because
183 // the Stream Header/Footer decoders assume that there's at least
184 // LZMA_STREAM_HEADER_SIZE bytes in coder->temp.
185 assert(coder->temp_size >= LZMA_STREAM_HEADER_SIZE);
187 if (seek_to_pos(coder, coder->file_target_pos - coder->temp_size,
188 in_start, in_pos, in_size))
189 return LZMA_SEEK_NEEDED;
191 return LZMA_OK;
195 /// Gets the number of zero-bytes at the end of the buffer.
196 static size_t
197 get_padding_size(const uint8_t *buf, size_t buf_size)
199 size_t padding = 0;
200 while (buf_size > 0 && buf[--buf_size] == 0x00)
201 ++padding;
203 return padding;
207 /// With the Stream Header at the very beginning of the file, LZMA_FORMAT_ERROR
208 /// is used to tell the application that Magic Bytes didn't match. In other
209 /// Stream Header/Footer fields (in the middle/end of the file) it could be
210 /// a bit confusing to return LZMA_FORMAT_ERROR as we already know that there
211 /// is a valid Stream Header at the beginning of the file. For those cases
212 /// this function is used to convert LZMA_FORMAT_ERROR to LZMA_DATA_ERROR.
213 static lzma_ret
214 hide_format_error(lzma_ret ret)
216 if (ret == LZMA_FORMAT_ERROR)
217 ret = LZMA_DATA_ERROR;
219 return ret;
223 /// Calls the Index decoder and updates coder->index_remaining.
224 /// This is a separate function because the input can be either directly
225 /// from the application or from coder->temp.
226 static lzma_ret
227 decode_index(lzma_file_info_coder *coder, const lzma_allocator *allocator,
228 const uint8_t *restrict in, size_t *restrict in_pos,
229 size_t in_size, bool update_file_cur_pos)
231 const size_t in_start = *in_pos;
233 const lzma_ret ret = coder->index_decoder.code(
234 coder->index_decoder.coder,
235 allocator, in, in_pos, in_size,
236 NULL, NULL, 0, LZMA_RUN);
238 coder->index_remaining -= *in_pos - in_start;
240 if (update_file_cur_pos)
241 coder->file_cur_pos += *in_pos - in_start;
243 return ret;
247 static lzma_ret
248 file_info_decode(void *coder_ptr, const lzma_allocator *allocator,
249 const uint8_t *restrict in, size_t *restrict in_pos,
250 size_t in_size,
251 uint8_t *restrict out lzma_attribute((__unused__)),
252 size_t *restrict out_pos lzma_attribute((__unused__)),
253 size_t out_size lzma_attribute((__unused__)),
254 lzma_action action lzma_attribute((__unused__)))
256 lzma_file_info_coder *coder = coder_ptr;
257 const size_t in_start = *in_pos;
259 // If the caller provides input past the end of the file, trim
260 // the extra bytes from the buffer so that we won't read too far.
261 assert(coder->file_size >= coder->file_cur_pos);
262 if (coder->file_size - coder->file_cur_pos < in_size - in_start)
263 in_size = in_start
264 + (size_t)(coder->file_size - coder->file_cur_pos);
266 while (true)
267 switch (coder->sequence) {
268 case SEQ_MAGIC_BYTES:
269 // Decode the Stream Header at the beginning of the file
270 // first to check if the Magic Bytes match. The flags
271 // are stored in coder->first_header_flags so that we
272 // don't need to seek to it again.
274 // Check that the file is big enough to contain at least
275 // Stream Header.
276 if (coder->file_size < LZMA_STREAM_HEADER_SIZE)
277 return LZMA_FORMAT_ERROR;
279 // Read the Stream Header field into coder->temp.
280 if (fill_temp(coder, in, in_pos, in_size))
281 return LZMA_OK;
283 // This is the only Stream Header/Footer decoding where we
284 // want to return LZMA_FORMAT_ERROR if the Magic Bytes don't
285 // match. Elsewhere it will be converted to LZMA_DATA_ERROR.
286 return_if_error(lzma_stream_header_decode(
287 &coder->first_header_flags, coder->temp));
289 // Now that we know that the Magic Bytes match, check the
290 // file size. It's better to do this here after checking the
291 // Magic Bytes since this way we can give LZMA_FORMAT_ERROR
292 // instead of LZMA_DATA_ERROR when the Magic Bytes don't
293 // match in a file that is too big or isn't a multiple of
294 // four bytes.
295 if (coder->file_size > LZMA_VLI_MAX || (coder->file_size & 3))
296 return LZMA_DATA_ERROR;
298 // Start looking for Stream Padding and Stream Footer
299 // at the end of the file.
300 coder->file_target_pos = coder->file_size;
302 // Fall through
304 case SEQ_PADDING_SEEK:
305 coder->sequence = SEQ_PADDING_DECODE;
306 return_if_error(reverse_seek(
307 coder, in_start, in_pos, in_size));
309 // Fall through
311 case SEQ_PADDING_DECODE: {
312 // Copy to coder->temp first. This keeps the code simpler if
313 // the application only provides input a few bytes at a time.
314 if (fill_temp(coder, in, in_pos, in_size))
315 return LZMA_OK;
317 // Scan the buffer backwards to get the size of the
318 // Stream Padding field (if any).
319 const size_t new_padding = get_padding_size(
320 coder->temp, coder->temp_size);
321 coder->stream_padding += new_padding;
323 // Set the target position to the beginning of Stream Padding
324 // that has been observed so far. If all Stream Padding has
325 // been seen, then the target position will be at the end
326 // of the Stream Footer field.
327 coder->file_target_pos -= new_padding;
329 if (new_padding == coder->temp_size) {
330 // The whole buffer was padding. Seek backwards in
331 // the file to get more input.
332 coder->sequence = SEQ_PADDING_SEEK;
333 break;
336 // Size of Stream Padding must be a multiple of 4 bytes.
337 if (coder->stream_padding & 3)
338 return LZMA_DATA_ERROR;
340 coder->sequence = SEQ_FOOTER;
342 // Calculate the amount of non-padding data in coder->temp.
343 coder->temp_size -= new_padding;
344 coder->temp_pos = coder->temp_size;
346 // We can avoid an external seek if the whole Stream Footer
347 // is already in coder->temp. In that case SEQ_FOOTER won't
348 // read more input and will find the Stream Footer from
349 // coder->temp[coder->temp_size - LZMA_STREAM_HEADER_SIZE].
351 // Otherwise we will need to seek. The seeking is done so
352 // that Stream Footer will be at the end of coder->temp.
353 // This way it's likely that we also get a complete Index
354 // field into coder->temp without needing a separate seek
355 // for that (unless the Index field is big).
356 if (coder->temp_size < LZMA_STREAM_HEADER_SIZE)
357 return_if_error(reverse_seek(
358 coder, in_start, in_pos, in_size));
361 // Fall through
363 case SEQ_FOOTER:
364 // Copy the Stream Footer field into coder->temp.
365 // If Stream Footer was already available in coder->temp
366 // in SEQ_PADDING_DECODE, then this does nothing.
367 if (fill_temp(coder, in, in_pos, in_size))
368 return LZMA_OK;
370 // Make coder->file_target_pos and coder->temp_size point
371 // to the beginning of Stream Footer and thus to the end
372 // of the Index field. coder->temp_pos will be updated
373 // a bit later.
374 coder->file_target_pos -= LZMA_STREAM_HEADER_SIZE;
375 coder->temp_size -= LZMA_STREAM_HEADER_SIZE;
377 // Decode Stream Footer.
378 return_if_error(hide_format_error(lzma_stream_footer_decode(
379 &coder->footer_flags,
380 coder->temp + coder->temp_size)));
382 // Check that we won't seek past the beginning of the file.
384 // LZMA_STREAM_HEADER_SIZE is added because there must be
385 // space for Stream Header too even though we won't seek
386 // there before decoding the Index field.
388 // There's no risk of integer overflow here because
389 // Backward Size cannot be greater than 2^34.
390 if (coder->file_target_pos < coder->footer_flags.backward_size
391 + LZMA_STREAM_HEADER_SIZE)
392 return LZMA_DATA_ERROR;
394 // Set the target position to the beginning of the Index field.
395 coder->file_target_pos -= coder->footer_flags.backward_size;
396 coder->sequence = SEQ_INDEX_INIT;
398 // We can avoid an external seek if the whole Index field is
399 // already available in coder->temp.
400 if (coder->temp_size >= coder->footer_flags.backward_size) {
401 // Set coder->temp_pos to point to the beginning
402 // of the Index.
403 coder->temp_pos = coder->temp_size
404 - coder->footer_flags.backward_size;
405 } else {
406 // These are set to zero to indicate that there's no
407 // useful data (Index or anything else) in coder->temp.
408 coder->temp_pos = 0;
409 coder->temp_size = 0;
411 // Seek to the beginning of the Index field.
412 if (seek_to_pos(coder, coder->file_target_pos,
413 in_start, in_pos, in_size))
414 return LZMA_SEEK_NEEDED;
417 // Fall through
419 case SEQ_INDEX_INIT: {
420 // Calculate the amount of memory already used by the earlier
421 // Indexes so that we know how big memory limit to pass to
422 // the Index decoder.
424 // NOTE: When there are multiple Streams, the separate
425 // lzma_index structures can use more RAM (as measured by
426 // lzma_index_memused()) than the final combined lzma_index.
427 // Thus memlimit may need to be slightly higher than the final
428 // calculated memory usage will be. This is perhaps a bit
429 // confusing to the application, but I think it shouldn't
430 // cause problems in practice.
431 uint64_t memused = 0;
432 if (coder->combined_index != NULL) {
433 memused = lzma_index_memused(coder->combined_index);
434 assert(memused <= coder->memlimit);
435 if (memused > coder->memlimit) // Extra sanity check
436 return LZMA_PROG_ERROR;
439 // Initialize the Index decoder.
440 return_if_error(lzma_index_decoder_init(
441 &coder->index_decoder, allocator,
442 &coder->this_index,
443 coder->memlimit - memused));
445 coder->index_remaining = coder->footer_flags.backward_size;
446 coder->sequence = SEQ_INDEX_DECODE;
449 // Fall through
451 case SEQ_INDEX_DECODE: {
452 // Decode (a part of) the Index. If the whole Index is already
453 // in coder->temp, read it from there. Otherwise read from
454 // in[*in_pos] onwards. Note that index_decode() updates
455 // coder->index_remaining and optionally coder->file_cur_pos.
456 lzma_ret ret;
457 if (coder->temp_size != 0) {
458 assert(coder->temp_size - coder->temp_pos
459 == coder->index_remaining);
460 ret = decode_index(coder, allocator, coder->temp,
461 &coder->temp_pos, coder->temp_size,
462 false);
463 } else {
464 // Don't give the decoder more input than the known
465 // remaining size of the Index field.
466 size_t in_stop = in_size;
467 if (in_size - *in_pos > coder->index_remaining)
468 in_stop = *in_pos
469 + (size_t)(coder->index_remaining);
471 ret = decode_index(coder, allocator,
472 in, in_pos, in_stop, true);
475 switch (ret) {
476 case LZMA_OK:
477 // If the Index docoder asks for more input when we
478 // have already given it as much input as Backward Size
479 // indicated, the file is invalid.
480 if (coder->index_remaining == 0)
481 return LZMA_DATA_ERROR;
483 // We cannot get here if we were reading Index from
484 // coder->temp because when reading from coder->temp
485 // we give the Index decoder exactly
486 // coder->index_remaining bytes of input.
487 assert(coder->temp_size == 0);
489 return LZMA_OK;
491 case LZMA_STREAM_END:
492 // If the decoding seems to be successful, check also
493 // that the Index decoder consumed as much input as
494 // indicated by the Backward Size field.
495 if (coder->index_remaining != 0)
496 return LZMA_DATA_ERROR;
498 break;
500 default:
501 return ret;
504 // Calculate how much the Index tells us to seek backwards
505 // (relative to the beginning of the Index): Total size of
506 // all Blocks plus the size of the Stream Header field.
507 // No integer overflow here because lzma_index_total_size()
508 // cannot return a value greater than LZMA_VLI_MAX.
509 const uint64_t seek_amount
510 = lzma_index_total_size(coder->this_index)
511 + LZMA_STREAM_HEADER_SIZE;
513 // Check that Index is sane in sense that seek_amount won't
514 // make us seek past the beginning of the file when locating
515 // the Stream Header.
517 // coder->file_target_pos still points to the beginning of
518 // the Index field.
519 if (coder->file_target_pos < seek_amount)
520 return LZMA_DATA_ERROR;
522 // Set the target to the beginning of Stream Header.
523 coder->file_target_pos -= seek_amount;
525 if (coder->file_target_pos == 0) {
526 // We would seek to the beginning of the file, but
527 // since we already decoded that Stream Header in
528 // SEQ_MAGIC_BYTES, we can use the cached value from
529 // coder->first_header_flags to avoid the seek.
530 coder->header_flags = coder->first_header_flags;
531 coder->sequence = SEQ_HEADER_COMPARE;
532 break;
535 coder->sequence = SEQ_HEADER_DECODE;
537 // Make coder->file_target_pos point to the end of
538 // the Stream Header field.
539 coder->file_target_pos += LZMA_STREAM_HEADER_SIZE;
541 // If coder->temp_size is non-zero, it points to the end
542 // of the Index field. Then the beginning of the Index
543 // field is at coder->temp[coder->temp_size
544 // - coder->footer_flags.backward_size].
545 assert(coder->temp_size == 0 || coder->temp_size
546 >= coder->footer_flags.backward_size);
548 // If coder->temp contained the whole Index, see if it has
549 // enough data to contain also the Stream Header. If so,
550 // we avoid an external seek.
552 // NOTE: This can happen only with small .xz files and only
553 // for the non-first Stream as the Stream Flags of the first
554 // Stream are cached and already handled a few lines above.
555 // So this isn't as useful as the other seek-avoidance cases.
556 if (coder->temp_size != 0 && coder->temp_size
557 - coder->footer_flags.backward_size
558 >= seek_amount) {
559 // Make temp_pos and temp_size point to the *end* of
560 // Stream Header so that SEQ_HEADER_DECODE will find
561 // the start of Stream Header from coder->temp[
562 // coder->temp_size - LZMA_STREAM_HEADER_SIZE].
563 coder->temp_pos = coder->temp_size
564 - coder->footer_flags.backward_size
565 - seek_amount
566 + LZMA_STREAM_HEADER_SIZE;
567 coder->temp_size = coder->temp_pos;
568 } else {
569 // Seek so that Stream Header will be at the end of
570 // coder->temp. With typical multi-Stream files we
571 // will usually also get the Stream Footer and Index
572 // of the *previous* Stream in coder->temp and thus
573 // won't need a separate seek for them.
574 return_if_error(reverse_seek(coder,
575 in_start, in_pos, in_size));
579 // Fall through
581 case SEQ_HEADER_DECODE:
582 // Copy the Stream Header field into coder->temp.
583 // If Stream Header was already available in coder->temp
584 // in SEQ_INDEX_DECODE, then this does nothing.
585 if (fill_temp(coder, in, in_pos, in_size))
586 return LZMA_OK;
588 // Make all these point to the beginning of Stream Header.
589 coder->file_target_pos -= LZMA_STREAM_HEADER_SIZE;
590 coder->temp_size -= LZMA_STREAM_HEADER_SIZE;
591 coder->temp_pos = coder->temp_size;
593 // Decode the Stream Header.
594 return_if_error(hide_format_error(lzma_stream_header_decode(
595 &coder->header_flags,
596 coder->temp + coder->temp_size)));
598 coder->sequence = SEQ_HEADER_COMPARE;
600 // Fall through
602 case SEQ_HEADER_COMPARE:
603 // Compare Stream Header against Stream Footer. They must
604 // match.
605 return_if_error(lzma_stream_flags_compare(
606 &coder->header_flags, &coder->footer_flags));
608 // Store the decoded Stream Flags into the Index. Use the
609 // Footer Flags because it contains Backward Size, although
610 // it shouldn't matter in practice.
611 if (lzma_index_stream_flags(coder->this_index,
612 &coder->footer_flags) != LZMA_OK)
613 return LZMA_PROG_ERROR;
615 // Store also the size of the Stream Padding field. It is
616 // needed to calculate the offsets of the Streams correctly.
617 if (lzma_index_stream_padding(coder->this_index,
618 coder->stream_padding) != LZMA_OK)
619 return LZMA_PROG_ERROR;
621 // Reset it so that it's ready for the next Stream.
622 coder->stream_padding = 0;
624 // Append the earlier decoded Indexes after this_index.
625 if (coder->combined_index != NULL)
626 return_if_error(lzma_index_cat(coder->this_index,
627 coder->combined_index, allocator));
629 coder->combined_index = coder->this_index;
630 coder->this_index = NULL;
632 // If the whole file was decoded, tell the caller that we
633 // are finished.
634 if (coder->file_target_pos == 0) {
635 // The combined index must indicate the same file
636 // size as was told to us at initialization.
637 assert(lzma_index_file_size(coder->combined_index)
638 == coder->file_size);
640 // Make the combined index available to
641 // the application.
642 *coder->dest_index = coder->combined_index;
643 coder->combined_index = NULL;
645 // Mark the input buffer as used since we may have
646 // done internal seeking and thus don't know how
647 // many input bytes were actually used. This way
648 // lzma_stream.total_in gets a slightly better
649 // estimate of the amount of input used.
650 *in_pos = in_size;
651 return LZMA_STREAM_END;
654 // We didn't hit the beginning of the file yet, so continue
655 // reading backwards in the file. If we have unprocessed
656 // data in coder->temp, use it before requesting more data
657 // from the application.
659 // coder->file_target_pos, coder->temp_size, and
660 // coder->temp_pos all point to the beginning of Stream Header
661 // and thus the end of the previous Stream in the file.
662 coder->sequence = coder->temp_size > 0
663 ? SEQ_PADDING_DECODE : SEQ_PADDING_SEEK;
664 break;
666 default:
667 assert(0);
668 return LZMA_PROG_ERROR;
673 static lzma_ret
674 file_info_decoder_memconfig(void *coder_ptr, uint64_t *memusage,
675 uint64_t *old_memlimit, uint64_t new_memlimit)
677 lzma_file_info_coder *coder = coder_ptr;
679 // The memory usage calculation comes from three things:
681 // (1) The Indexes that have already been decoded and processed into
682 // coder->combined_index.
684 // (2) The latest Index in coder->this_index that has been decoded but
685 // not yet put into coder->combined_index.
687 // (3) The latest Index that we have started decoding but haven't
688 // finished and thus isn't available in coder->this_index yet.
689 // Memory usage and limit information needs to be communicated
690 // from/to coder->index_decoder.
692 // Care has to be taken to not do both (2) and (3) when calculating
693 // the memory usage.
694 uint64_t combined_index_memusage = 0;
695 uint64_t this_index_memusage = 0;
697 // (1) If we have already successfully decoded one or more Indexes,
698 // get their memory usage.
699 if (coder->combined_index != NULL)
700 combined_index_memusage = lzma_index_memused(
701 coder->combined_index);
703 // Choose between (2), (3), or neither.
704 if (coder->this_index != NULL) {
705 // (2) The latest Index is available. Use its memory usage.
706 this_index_memusage = lzma_index_memused(coder->this_index);
708 } else if (coder->sequence == SEQ_INDEX_DECODE) {
709 // (3) The Index decoder is activate and hasn't yet stored
710 // the new index in coder->this_index. Get the memory usage
711 // information from the Index decoder.
713 // NOTE: If the Index decoder doesn't yet know how much memory
714 // it will eventually need, it will return a tiny value here.
715 uint64_t dummy;
716 if (coder->index_decoder.memconfig(coder->index_decoder.coder,
717 &this_index_memusage, &dummy, 0)
718 != LZMA_OK) {
719 assert(0);
720 return LZMA_PROG_ERROR;
724 // Now we know the total memory usage/requirement. If we had neither
725 // old Indexes nor a new Index, this will be zero which isn't
726 // acceptable as lzma_memusage() has to return non-zero on success
727 // and even with an empty .xz file we will end up with a lzma_index
728 // that takes some memory.
729 *memusage = combined_index_memusage + this_index_memusage;
730 if (*memusage == 0)
731 *memusage = lzma_index_memusage(1, 0);
733 *old_memlimit = coder->memlimit;
735 // If requested, set a new memory usage limit.
736 if (new_memlimit != 0) {
737 if (new_memlimit < *memusage)
738 return LZMA_MEMLIMIT_ERROR;
740 // In the condition (3) we need to tell the Index decoder
741 // its new memory usage limit.
742 if (coder->this_index == NULL
743 && coder->sequence == SEQ_INDEX_DECODE) {
744 const uint64_t idec_new_memlimit = new_memlimit
745 - combined_index_memusage;
747 assert(this_index_memusage > 0);
748 assert(idec_new_memlimit > 0);
750 uint64_t dummy1;
751 uint64_t dummy2;
753 if (coder->index_decoder.memconfig(
754 coder->index_decoder.coder,
755 &dummy1, &dummy2, idec_new_memlimit)
756 != LZMA_OK) {
757 assert(0);
758 return LZMA_PROG_ERROR;
762 coder->memlimit = new_memlimit;
765 return LZMA_OK;
769 static void
770 file_info_decoder_end(void *coder_ptr, const lzma_allocator *allocator)
772 lzma_file_info_coder *coder = coder_ptr;
774 lzma_next_end(&coder->index_decoder, allocator);
775 lzma_index_end(coder->this_index, allocator);
776 lzma_index_end(coder->combined_index, allocator);
778 lzma_free(coder, allocator);
779 return;
783 static lzma_ret
784 lzma_file_info_decoder_init(lzma_next_coder *next,
785 const lzma_allocator *allocator, uint64_t *seek_pos,
786 lzma_index **dest_index,
787 uint64_t memlimit, uint64_t file_size)
789 lzma_next_coder_init(&lzma_file_info_decoder_init, next, allocator);
791 if (dest_index == NULL)
792 return LZMA_PROG_ERROR;
794 lzma_file_info_coder *coder = next->coder;
795 if (coder == NULL) {
796 coder = lzma_alloc(sizeof(lzma_file_info_coder), allocator);
797 if (coder == NULL)
798 return LZMA_MEM_ERROR;
800 next->coder = coder;
801 next->code = &file_info_decode;
802 next->end = &file_info_decoder_end;
803 next->memconfig = &file_info_decoder_memconfig;
805 coder->index_decoder = LZMA_NEXT_CODER_INIT;
806 coder->this_index = NULL;
807 coder->combined_index = NULL;
810 coder->sequence = SEQ_MAGIC_BYTES;
811 coder->file_cur_pos = 0;
812 coder->file_target_pos = 0;
813 coder->file_size = file_size;
815 lzma_index_end(coder->this_index, allocator);
816 coder->this_index = NULL;
818 lzma_index_end(coder->combined_index, allocator);
819 coder->combined_index = NULL;
821 coder->stream_padding = 0;
823 coder->dest_index = dest_index;
824 coder->external_seek_pos = seek_pos;
826 // If memlimit is 0, make it 1 to ensure that lzma_memlimit_get()
827 // won't return 0 (which would indicate an error).
828 coder->memlimit = my_max(1, memlimit);
830 // Prepare these for reading the first Stream Header into coder->temp.
831 coder->temp_pos = 0;
832 coder->temp_size = LZMA_STREAM_HEADER_SIZE;
834 return LZMA_OK;
838 extern LZMA_API(lzma_ret)
839 lzma_file_info_decoder(lzma_stream *strm, lzma_index **dest_index,
840 uint64_t memlimit, uint64_t file_size)
842 lzma_next_strm_init(lzma_file_info_decoder_init, strm, &strm->seek_pos,
843 dest_index, memlimit, file_size);
845 // We allow LZMA_FINISH in addition to LZMA_RUN for convenience.
846 // lzma_code() is able to handle the LZMA_FINISH + LZMA_SEEK_NEEDED
847 // combination in a sane way. Applications still need to be careful
848 // if they use LZMA_FINISH so that they remember to reset it back
849 // to LZMA_RUN after seeking if needed.
850 strm->internal->supported_actions[LZMA_RUN] = true;
851 strm->internal->supported_actions[LZMA_FINISH] = true;
853 return LZMA_OK;