diff options
author | Richard Purdie <rpurdie@rpsys.net> | 2006-06-22 17:47:34 -0400 |
---|---|---|
committer | Linus Torvalds <torvalds@g5.osdl.org> | 2006-06-22 18:05:58 -0400 |
commit | 4f3865fb57a04db7cca068fed1c15badc064a302 (patch) | |
tree | 4c923c72b6ac9b633c87cc73b55a75c7cfd0f044 /include/linux/zlib.h | |
parent | 4f1bcaf094ccc512c23e10104c05a6f8e5b7a9e4 (diff) |
[PATCH] zlib_inflate: Upgrade library code to a recent version
Upgrade the zlib_inflate implementation in the kernel from a patched
version 1.1.3/4 to a patched 1.2.3.
The code in the kernel is about seven years old and I noticed that the
external zlib library's inflate performance was significantly faster (~50%)
than the code in the kernel on ARM (and faster again on x86_32).
For comparison the newer deflate code is 20% slower on ARM and 50% slower
on x86_32 but gives an approx 1% compression ratio improvement. I don't
consider this to be an improvement for kernel use so have no plans to
change the zlib_deflate code.
Various changes have been made to the zlib code in the kernel, the most
significant being the extra functions/flush option used by ppp_deflate.
This update reimplements the features PPP needs to ensure it continues to
work.
This code has been tested on ARM under both JFFS2 (with zlib compression
enabled) and ppp_deflate and on x86_32. JFFS2 sees an approx. 10% real
world file read speed improvement.
This patch also removes ZLIB_VERSION as it no longer has a correct value.
We don't need version checks anyway as the kernel's module handling will
take care of that for us. This removal is also more in keeping with the
zlib author's wishes (http://www.zlib.net/zlib_faq.html#faq24) and I've
added something to the zlib.h header to note its a modified version.
Signed-off-by: Richard Purdie <rpurdie@rpsys.net>
Acked-by: Joern Engel <joern@wh.fh-wedel.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Diffstat (limited to 'include/linux/zlib.h')
-rw-r--r-- | include/linux/zlib.h | 209 |
1 files changed, 129 insertions, 80 deletions
diff --git a/include/linux/zlib.h b/include/linux/zlib.h index 4fa32f0d4df8..9e3192a7dc6f 100644 --- a/include/linux/zlib.h +++ b/include/linux/zlib.h | |||
@@ -1,7 +1,6 @@ | |||
1 | /* zlib.h -- interface of the 'zlib' general purpose compression library | 1 | /* zlib.h -- interface of the 'zlib' general purpose compression library |
2 | version 1.1.3, July 9th, 1998 | ||
3 | 2 | ||
4 | Copyright (C) 1995-1998 Jean-loup Gailly and Mark Adler | 3 | Copyright (C) 1995-2005 Jean-loup Gailly and Mark Adler |
5 | 4 | ||
6 | This software is provided 'as-is', without any express or implied | 5 | This software is provided 'as-is', without any express or implied |
7 | warranty. In no event will the authors be held liable for any damages | 6 | warranty. In no event will the authors be held liable for any damages |
@@ -24,7 +23,7 @@ | |||
24 | 23 | ||
25 | 24 | ||
26 | The data format used by the zlib library is described by RFCs (Request for | 25 | The data format used by the zlib library is described by RFCs (Request for |
27 | Comments) 1950 to 1952 in the files ftp://ds.internic.net/rfc/rfc1950.txt | 26 | Comments) 1950 to 1952 in the files http://www.ietf.org/rfc/rfc1950.txt |
28 | (zlib format), rfc1951.txt (deflate format) and rfc1952.txt (gzip format). | 27 | (zlib format), rfc1951.txt (deflate format) and rfc1952.txt (gzip format). |
29 | */ | 28 | */ |
30 | 29 | ||
@@ -33,7 +32,22 @@ | |||
33 | 32 | ||
34 | #include <linux/zconf.h> | 33 | #include <linux/zconf.h> |
35 | 34 | ||
36 | #define ZLIB_VERSION "1.1.3" | 35 | /* zlib deflate based on ZLIB_VERSION "1.1.3" */ |
36 | /* zlib inflate based on ZLIB_VERSION "1.2.3" */ | ||
37 | |||
38 | /* | ||
39 | This is a modified version of zlib for use inside the Linux kernel. | ||
40 | The main changes are to perform all memory allocation in advance. | ||
41 | |||
42 | Inflation Changes: | ||
43 | * Z_PACKET_FLUSH is added and used by ppp_deflate. Before returning | ||
44 | this checks there is no more input data available and the next data | ||
45 | is a STORED block. It also resets the mode to be read for the next | ||
46 | data, all as per PPP requirements. | ||
47 | * Addition of zlib_inflateIncomp which copies incompressible data into | ||
48 | the history window and adjusts the accoutning without calling | ||
49 | zlib_inflate itself to inflate the data. | ||
50 | */ | ||
37 | 51 | ||
38 | /* | 52 | /* |
39 | The 'zlib' compression library provides in-memory compression and | 53 | The 'zlib' compression library provides in-memory compression and |
@@ -48,9 +62,18 @@ | |||
48 | application must provide more input and/or consume the output | 62 | application must provide more input and/or consume the output |
49 | (providing more output space) before each call. | 63 | (providing more output space) before each call. |
50 | 64 | ||
65 | The compressed data format used by default by the in-memory functions is | ||
66 | the zlib format, which is a zlib wrapper documented in RFC 1950, wrapped | ||
67 | around a deflate stream, which is itself documented in RFC 1951. | ||
68 | |||
51 | The library also supports reading and writing files in gzip (.gz) format | 69 | The library also supports reading and writing files in gzip (.gz) format |
52 | with an interface similar to that of stdio. | 70 | with an interface similar to that of stdio. |
53 | 71 | ||
72 | The zlib format was designed to be compact and fast for use in memory | ||
73 | and on communications channels. The gzip format was designed for single- | ||
74 | file compression on file systems, has a larger header than zlib to maintain | ||
75 | directory information, and uses a different, slower check method than zlib. | ||
76 | |||
54 | The library does not install any signal handler. The decoder checks | 77 | The library does not install any signal handler. The decoder checks |
55 | the consistency of the compressed data, so the library should never | 78 | the consistency of the compressed data, so the library should never |
56 | crash even in case of corrupted input. | 79 | crash even in case of corrupted input. |
@@ -119,7 +142,8 @@ typedef z_stream *z_streamp; | |||
119 | #define Z_SYNC_FLUSH 3 | 142 | #define Z_SYNC_FLUSH 3 |
120 | #define Z_FULL_FLUSH 4 | 143 | #define Z_FULL_FLUSH 4 |
121 | #define Z_FINISH 5 | 144 | #define Z_FINISH 5 |
122 | /* Allowed flush values; see deflate() below for details */ | 145 | #define Z_BLOCK 6 /* Only for inflate at present */ |
146 | /* Allowed flush values; see deflate() and inflate() below for details */ | ||
123 | 147 | ||
124 | #define Z_OK 0 | 148 | #define Z_OK 0 |
125 | #define Z_STREAM_END 1 | 149 | #define Z_STREAM_END 1 |
@@ -155,13 +179,6 @@ typedef z_stream *z_streamp; | |||
155 | 179 | ||
156 | /* basic functions */ | 180 | /* basic functions */ |
157 | 181 | ||
158 | extern const char * zlib_zlibVersion (void); | ||
159 | /* The application can compare zlibVersion and ZLIB_VERSION for consistency. | ||
160 | If the first character differs, the library code actually used is | ||
161 | not compatible with the zlib.h header file used by the application. | ||
162 | This check is automatically made by deflateInit and inflateInit. | ||
163 | */ | ||
164 | |||
165 | extern int zlib_deflate_workspacesize (void); | 182 | extern int zlib_deflate_workspacesize (void); |
166 | /* | 183 | /* |
167 | Returns the number of bytes that needs to be allocated for a per- | 184 | Returns the number of bytes that needs to be allocated for a per- |
@@ -315,9 +332,9 @@ extern int zlib_inflateInit (z_streamp strm); | |||
315 | extern int zlib_inflate (z_streamp strm, int flush); | 332 | extern int zlib_inflate (z_streamp strm, int flush); |
316 | /* | 333 | /* |
317 | inflate decompresses as much data as possible, and stops when the input | 334 | inflate decompresses as much data as possible, and stops when the input |
318 | buffer becomes empty or the output buffer becomes full. It may some | 335 | buffer becomes empty or the output buffer becomes full. It may introduce |
319 | introduce some output latency (reading input without producing any output) | 336 | some output latency (reading input without producing any output) except when |
320 | except when forced to flush. | 337 | forced to flush. |
321 | 338 | ||
322 | The detailed semantics are as follows. inflate performs one or both of the | 339 | The detailed semantics are as follows. inflate performs one or both of the |
323 | following actions: | 340 | following actions: |
@@ -341,11 +358,26 @@ extern int zlib_inflate (z_streamp strm, int flush); | |||
341 | must be called again after making room in the output buffer because there | 358 | must be called again after making room in the output buffer because there |
342 | might be more output pending. | 359 | might be more output pending. |
343 | 360 | ||
344 | If the parameter flush is set to Z_SYNC_FLUSH, inflate flushes as much | 361 | The flush parameter of inflate() can be Z_NO_FLUSH, Z_SYNC_FLUSH, |
345 | output as possible to the output buffer. The flushing behavior of inflate is | 362 | Z_FINISH, or Z_BLOCK. Z_SYNC_FLUSH requests that inflate() flush as much |
346 | not specified for values of the flush parameter other than Z_SYNC_FLUSH | 363 | output as possible to the output buffer. Z_BLOCK requests that inflate() stop |
347 | and Z_FINISH, but the current implementation actually flushes as much output | 364 | if and when it gets to the next deflate block boundary. When decoding the |
348 | as possible anyway. | 365 | zlib or gzip format, this will cause inflate() to return immediately after |
366 | the header and before the first block. When doing a raw inflate, inflate() | ||
367 | will go ahead and process the first block, and will return when it gets to | ||
368 | the end of that block, or when it runs out of data. | ||
369 | |||
370 | The Z_BLOCK option assists in appending to or combining deflate streams. | ||
371 | Also to assist in this, on return inflate() will set strm->data_type to the | ||
372 | number of unused bits in the last byte taken from strm->next_in, plus 64 | ||
373 | if inflate() is currently decoding the last block in the deflate stream, | ||
374 | plus 128 if inflate() returned immediately after decoding an end-of-block | ||
375 | code or decoding the complete header up to just before the first byte of the | ||
376 | deflate stream. The end-of-block will not be indicated until all of the | ||
377 | uncompressed data from that block has been written to strm->next_out. The | ||
378 | number of unused bits may in general be greater than seven, except when | ||
379 | bit 7 of data_type is set, in which case the number of unused bits will be | ||
380 | less than eight. | ||
349 | 381 | ||
350 | inflate() should normally be called until it returns Z_STREAM_END or an | 382 | inflate() should normally be called until it returns Z_STREAM_END or an |
351 | error. However if all decompression is to be performed in a single step | 383 | error. However if all decompression is to be performed in a single step |
@@ -355,29 +387,44 @@ extern int zlib_inflate (z_streamp strm, int flush); | |||
355 | uncompressed data. (The size of the uncompressed data may have been saved | 387 | uncompressed data. (The size of the uncompressed data may have been saved |
356 | by the compressor for this purpose.) The next operation on this stream must | 388 | by the compressor for this purpose.) The next operation on this stream must |
357 | be inflateEnd to deallocate the decompression state. The use of Z_FINISH | 389 | be inflateEnd to deallocate the decompression state. The use of Z_FINISH |
358 | is never required, but can be used to inform inflate that a faster routine | 390 | is never required, but can be used to inform inflate that a faster approach |
359 | may be used for the single inflate() call. | 391 | may be used for the single inflate() call. |
360 | 392 | ||
361 | If a preset dictionary is needed at this point (see inflateSetDictionary | 393 | In this implementation, inflate() always flushes as much output as |
362 | below), inflate sets strm-adler to the adler32 checksum of the | 394 | possible to the output buffer, and always uses the faster approach on the |
363 | dictionary chosen by the compressor and returns Z_NEED_DICT; otherwise | 395 | first call. So the only effect of the flush parameter in this implementation |
364 | it sets strm->adler to the adler32 checksum of all output produced | 396 | is on the return value of inflate(), as noted below, or when it returns early |
365 | so far (that is, total_out bytes) and returns Z_OK, Z_STREAM_END or | 397 | because Z_BLOCK is used. |
366 | an error code as described below. At the end of the stream, inflate() | 398 | |
367 | checks that its computed adler32 checksum is equal to that saved by the | 399 | If a preset dictionary is needed after this call (see inflateSetDictionary |
368 | compressor and returns Z_STREAM_END only if the checksum is correct. | 400 | below), inflate sets strm->adler to the adler32 checksum of the dictionary |
401 | chosen by the compressor and returns Z_NEED_DICT; otherwise it sets | ||
402 | strm->adler to the adler32 checksum of all output produced so far (that is, | ||
403 | total_out bytes) and returns Z_OK, Z_STREAM_END or an error code as described | ||
404 | below. At the end of the stream, inflate() checks that its computed adler32 | ||
405 | checksum is equal to that saved by the compressor and returns Z_STREAM_END | ||
406 | only if the checksum is correct. | ||
407 | |||
408 | inflate() will decompress and check either zlib-wrapped or gzip-wrapped | ||
409 | deflate data. The header type is detected automatically. Any information | ||
410 | contained in the gzip header is not retained, so applications that need that | ||
411 | information should instead use raw inflate, see inflateInit2() below, or | ||
412 | inflateBack() and perform their own processing of the gzip header and | ||
413 | trailer. | ||
369 | 414 | ||
370 | inflate() returns Z_OK if some progress has been made (more input processed | 415 | inflate() returns Z_OK if some progress has been made (more input processed |
371 | or more output produced), Z_STREAM_END if the end of the compressed data has | 416 | or more output produced), Z_STREAM_END if the end of the compressed data has |
372 | been reached and all uncompressed output has been produced, Z_NEED_DICT if a | 417 | been reached and all uncompressed output has been produced, Z_NEED_DICT if a |
373 | preset dictionary is needed at this point, Z_DATA_ERROR if the input data was | 418 | preset dictionary is needed at this point, Z_DATA_ERROR if the input data was |
374 | corrupted (input stream not conforming to the zlib format or incorrect | 419 | corrupted (input stream not conforming to the zlib format or incorrect check |
375 | adler32 checksum), Z_STREAM_ERROR if the stream structure was inconsistent | 420 | value), Z_STREAM_ERROR if the stream structure was inconsistent (for example |
376 | (for example if next_in or next_out was NULL), Z_MEM_ERROR if there was not | 421 | if next_in or next_out was NULL), Z_MEM_ERROR if there was not enough memory, |
377 | enough memory, Z_BUF_ERROR if no progress is possible or if there was not | 422 | Z_BUF_ERROR if no progress is possible or if there was not enough room in the |
378 | enough room in the output buffer when Z_FINISH is used. In the Z_DATA_ERROR | 423 | output buffer when Z_FINISH is used. Note that Z_BUF_ERROR is not fatal, and |
379 | case, the application may then call inflateSync to look for a good | 424 | inflate() can be called again with more input and more output space to |
380 | compression block. | 425 | continue decompressing. If Z_DATA_ERROR is returned, the application may then |
426 | call inflateSync() to look for a good compression block if a partial recovery | ||
427 | of the data is desired. | ||
381 | */ | 428 | */ |
382 | 429 | ||
383 | 430 | ||
@@ -547,16 +594,36 @@ extern int inflateInit2 (z_streamp strm, int windowBits); | |||
547 | The windowBits parameter is the base two logarithm of the maximum window | 594 | The windowBits parameter is the base two logarithm of the maximum window |
548 | size (the size of the history buffer). It should be in the range 8..15 for | 595 | size (the size of the history buffer). It should be in the range 8..15 for |
549 | this version of the library. The default value is 15 if inflateInit is used | 596 | this version of the library. The default value is 15 if inflateInit is used |
550 | instead. If a compressed stream with a larger window size is given as | 597 | instead. windowBits must be greater than or equal to the windowBits value |
551 | input, inflate() will return with the error code Z_DATA_ERROR instead of | 598 | provided to deflateInit2() while compressing, or it must be equal to 15 if |
552 | trying to allocate a larger window. | 599 | deflateInit2() was not used. If a compressed stream with a larger window |
553 | 600 | size is given as input, inflate() will return with the error code | |
554 | inflateInit2 returns Z_OK if success, Z_MEM_ERROR if there was not enough | 601 | Z_DATA_ERROR instead of trying to allocate a larger window. |
555 | memory, Z_STREAM_ERROR if a parameter is invalid (such as a negative | 602 | |
556 | memLevel). msg is set to null if there is no error message. inflateInit2 | 603 | windowBits can also be -8..-15 for raw inflate. In this case, -windowBits |
557 | does not perform any decompression apart from reading the zlib header if | 604 | determines the window size. inflate() will then process raw deflate data, |
558 | present: this will be done by inflate(). (So next_in and avail_in may be | 605 | not looking for a zlib or gzip header, not generating a check value, and not |
559 | modified, but next_out and avail_out are unchanged.) | 606 | looking for any check values for comparison at the end of the stream. This |
607 | is for use with other formats that use the deflate compressed data format | ||
608 | such as zip. Those formats provide their own check values. If a custom | ||
609 | format is developed using the raw deflate format for compressed data, it is | ||
610 | recommended that a check value such as an adler32 or a crc32 be applied to | ||
611 | the uncompressed data as is done in the zlib, gzip, and zip formats. For | ||
612 | most applications, the zlib format should be used as is. Note that comments | ||
613 | above on the use in deflateInit2() applies to the magnitude of windowBits. | ||
614 | |||
615 | windowBits can also be greater than 15 for optional gzip decoding. Add | ||
616 | 32 to windowBits to enable zlib and gzip decoding with automatic header | ||
617 | detection, or add 16 to decode only the gzip format (the zlib format will | ||
618 | return a Z_DATA_ERROR). If a gzip stream is being decoded, strm->adler is | ||
619 | a crc32 instead of an adler32. | ||
620 | |||
621 | inflateInit2 returns Z_OK if success, Z_MEM_ERROR if there was not enough | ||
622 | memory, Z_STREAM_ERROR if a parameter is invalid (such as a null strm). msg | ||
623 | is set to null if there is no error message. inflateInit2 does not perform | ||
624 | any decompression apart from reading the zlib header if present: this will | ||
625 | be done by inflate(). (So next_in and avail_in may be modified, but next_out | ||
626 | and avail_out are unchanged.) | ||
560 | */ | 627 | */ |
561 | 628 | ||
562 | extern int zlib_inflateSetDictionary (z_streamp strm, | 629 | extern int zlib_inflateSetDictionary (z_streamp strm, |
@@ -564,16 +631,19 @@ extern int zlib_inflateSetDictionary (z_streamp strm, | |||
564 | uInt dictLength); | 631 | uInt dictLength); |
565 | /* | 632 | /* |
566 | Initializes the decompression dictionary from the given uncompressed byte | 633 | Initializes the decompression dictionary from the given uncompressed byte |
567 | sequence. This function must be called immediately after a call of inflate | 634 | sequence. This function must be called immediately after a call of inflate, |
568 | if this call returned Z_NEED_DICT. The dictionary chosen by the compressor | 635 | if that call returned Z_NEED_DICT. The dictionary chosen by the compressor |
569 | can be determined from the Adler32 value returned by this call of | 636 | can be determined from the adler32 value returned by that call of inflate. |
570 | inflate. The compressor and decompressor must use exactly the same | 637 | The compressor and decompressor must use exactly the same dictionary (see |
571 | dictionary (see deflateSetDictionary). | 638 | deflateSetDictionary). For raw inflate, this function can be called |
639 | immediately after inflateInit2() or inflateReset() and before any call of | ||
640 | inflate() to set the dictionary. The application must insure that the | ||
641 | dictionary that was used for compression is provided. | ||
572 | 642 | ||
573 | inflateSetDictionary returns Z_OK if success, Z_STREAM_ERROR if a | 643 | inflateSetDictionary returns Z_OK if success, Z_STREAM_ERROR if a |
574 | parameter is invalid (such as NULL dictionary) or the stream state is | 644 | parameter is invalid (such as NULL dictionary) or the stream state is |
575 | inconsistent, Z_DATA_ERROR if the given dictionary doesn't match the | 645 | inconsistent, Z_DATA_ERROR if the given dictionary doesn't match the |
576 | expected one (incorrect Adler32 value). inflateSetDictionary does not | 646 | expected one (incorrect adler32 value). inflateSetDictionary does not |
577 | perform any decompression: this will be done by subsequent calls of | 647 | perform any decompression: this will be done by subsequent calls of |
578 | inflate(). | 648 | inflate(). |
579 | */ | 649 | */ |
@@ -614,40 +684,19 @@ extern int zlib_inflateIncomp (z_stream *strm); | |||
614 | containing the data at next_in (except that the data is not output). | 684 | containing the data at next_in (except that the data is not output). |
615 | */ | 685 | */ |
616 | 686 | ||
617 | /* various hacks, don't look :) */ | ||
618 | |||
619 | /* deflateInit and inflateInit are macros to allow checking the zlib version | ||
620 | * and the compiler's view of z_stream: | ||
621 | */ | ||
622 | extern int zlib_deflateInit_ (z_streamp strm, int level, | ||
623 | const char *version, int stream_size); | ||
624 | extern int zlib_inflateInit_ (z_streamp strm, | ||
625 | const char *version, int stream_size); | ||
626 | extern int zlib_deflateInit2_ (z_streamp strm, int level, int method, | ||
627 | int windowBits, int memLevel, | ||
628 | int strategy, const char *version, | ||
629 | int stream_size); | ||
630 | extern int zlib_inflateInit2_ (z_streamp strm, int windowBits, | ||
631 | const char *version, int stream_size); | ||
632 | #define zlib_deflateInit(strm, level) \ | 687 | #define zlib_deflateInit(strm, level) \ |
633 | zlib_deflateInit_((strm), (level), ZLIB_VERSION, sizeof(z_stream)) | 688 | zlib_deflateInit2((strm), (level), Z_DEFLATED, MAX_WBITS, \ |
689 | DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY) | ||
634 | #define zlib_inflateInit(strm) \ | 690 | #define zlib_inflateInit(strm) \ |
635 | zlib_inflateInit_((strm), ZLIB_VERSION, sizeof(z_stream)) | 691 | zlib_inflateInit2((strm), DEF_WBITS) |
636 | #define zlib_deflateInit2(strm, level, method, windowBits, memLevel, strategy) \ | ||
637 | zlib_deflateInit2_((strm),(level),(method),(windowBits),(memLevel),\ | ||
638 | (strategy), ZLIB_VERSION, sizeof(z_stream)) | ||
639 | #define zlib_inflateInit2(strm, windowBits) \ | ||
640 | zlib_inflateInit2_((strm), (windowBits), ZLIB_VERSION, sizeof(z_stream)) | ||
641 | 692 | ||
693 | extern int zlib_deflateInit2(z_streamp strm, int level, int method, | ||
694 | int windowBits, int memLevel, | ||
695 | int strategy); | ||
696 | extern int zlib_inflateInit2(z_streamp strm, int windowBits); | ||
642 | 697 | ||
643 | #if !defined(_Z_UTIL_H) && !defined(NO_DUMMY_DECL) | 698 | #if !defined(_Z_UTIL_H) && !defined(NO_DUMMY_DECL) |
644 | struct internal_state {int dummy;}; /* hack for buggy compilers */ | 699 | struct internal_state {int dummy;}; /* hack for buggy compilers */ |
645 | #endif | 700 | #endif |
646 | 701 | ||
647 | extern const char * zlib_zError (int err); | ||
648 | #if 0 | ||
649 | extern int zlib_inflateSyncPoint (z_streamp z); | ||
650 | #endif | ||
651 | extern const uLong * zlib_get_crc_table (void); | ||
652 | |||
653 | #endif /* _ZLIB_H */ | 702 | #endif /* _ZLIB_H */ |