From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Sterba Subject: [PATCH 2/4] btrfs: add LZ4 compression method Date: Mon, 13 Feb 2012 20:05:35 +0100 Message-ID: <1329159937-23976-2-git-send-email-dsterba@suse.cz> References: <20120213190325.GJ15350@twin.jikos.cz> <1329159937-23976-1-git-send-email-dsterba@suse.cz> Cc: David Sterba To: linux-btrfs@vger.kernel.org Return-path: In-Reply-To: <1329159937-23976-1-git-send-email-dsterba@suse.cz> List-ID: Adjusted a few defines to compile. The on-stack context allocation is disabled and either LZ4_compress64kCtx or LZ4_compressCtx must be used. Origin: http://lz4.googlecode.com/svn/trunk Revision: 55 Signed-off-by: David Sterba --- fs/btrfs/Makefile | 2 +- fs/btrfs/lz4.c | 810 +++++++++++++++++++++++++++++++++++++++++++++++++++++ fs/btrfs/lz4.h | 107 +++++++ 3 files changed, 918 insertions(+), 1 deletions(-) create mode 100644 fs/btrfs/lz4.c create mode 100644 fs/btrfs/lz4.h diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile index f22fe03..11f8c4e 100644 --- a/fs/btrfs/Makefile +++ b/fs/btrfs/Makefile @@ -8,7 +8,7 @@ btrfs-y += super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \ extent_io.o volumes.o async-thread.o ioctl.o locking.o orphan.o \ export.o tree-log.o free-space-cache.o zlib.o lzo.o \ compression.o delayed-ref.o relocation.o delayed-inode.o scrub.o \ - reada.o backref.o ulist.o snappy.o + reada.o backref.o ulist.o snappy.o lz4.o btrfs-$(CONFIG_BTRFS_FS_POSIX_ACL) += acl.o btrfs-$(CONFIG_BTRFS_FS_CHECK_INTEGRITY) += check-integrity.o diff --git a/fs/btrfs/lz4.c b/fs/btrfs/lz4.c new file mode 100644 index 0000000..e41d0cf --- /dev/null +++ b/fs/btrfs/lz4.c @@ -0,0 +1,810 @@ +/* + LZ4 - Fast LZ compression algorithm + Copyright (C) 2011-2012, Yann Collet. + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ +/* + * With authors permission dual licensed as BSD/GPL for linux kernel + * + * Origin: http://lz4.googlecode.com/svn/trunk + * Revision: 55 + */ + +//************************************** +// Tuning parameters +//************************************** +// COMPRESSIONLEVEL : +// Increasing this value improves compression ratio +// Lowering this value reduces memory usage +// Reduced memory usage typically improves speed, due to cache effect (ex : L1 32KB for Intel, L1 64KB for AMD) +// Memory usage formula : N->2^(N+2) Bytes (examples : 12 -> 16KB ; 17 -> 512KB) +#define COMPRESSIONLEVEL 12 + +// NONCOMPRESSIBLE_CONFIRMATION : +// Decreasing this value will make the algorithm skip faster data segments considered "incompressible" +// This may decrease compression ratio dramatically, but will be faster on incompressible data +// Increasing this value will make the algorithm search more before declaring a segment "incompressible" +// This could improve compression a bit, but will be slower on incompressible data +// The default value (6) is recommended +#define NONCOMPRESSIBLE_CONFIRMATION 6 + +// BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE : +// This will provide a boost to performance for big endian cpu, but the resulting compressed stream will be incompatible with little-endian CPU. +// You can set this option to 1 in situations where data will stay within closed environment +// This option is useless on Little_Endian CPU (such as x86) +//#define BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE 1 + + + +//************************************** +// CPU Feature Detection +//************************************** +// 32 or 64 bits ? +#if (__x86_64__ || __x86_64 || __amd64__ || __amd64 || __ppc64__ || _WIN64 || __LP64__ || _LP64) // Detects 64 bits mode +#define ARCH64 1 +#else +#define ARCH64 0 +#endif + +// Little Endian or Big Endian ? +#if (__BIG_ENDIAN__ || __BIG_ENDIAN || _BIG_ENDIAN || _ARCH_PPC || __PPC__ || __PPC || PPC || __powerpc__ || __powerpc || powerpc || ((defined(__BYTE_ORDER__)&&(__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__))) ) +#define CPU_BIG_ENDIAN 1 +#else +// Little Endian assumed. PDP Endian and other very rare endian format are unsupported. +#endif + +// Unaligned memory access ? +// This feature is automatically enabled for "common" CPU, such as x86. +// For others CPU, you may want to force this option manually to improve performance if your target CPU supports unaligned memory access +#if (__ARM_FEATURE_UNALIGNED) +#define CPU_UNALIGNED_ACCESS 1 +#endif + +// Uncomment this parameter if your target system does not support hardware bit count +//#define _FORCE_SW_BITCOUNT + + + +//************************************** +// Compiler Options +//************************************** +#if __STDC_VERSION__ >= 199901L // C99 + /* "restrict" is a known keyword */ +#else +#define restrict // Disable restrict +#endif + +#ifdef _MSC_VER +#define inline __forceinline // Visual is not C99, but supports inline +#endif + +#if (defined(__GNUC__) && (!(CPU_UNALIGNED_ACCESS))) +#define _PACKED __attribute__ ((packed)) +#else +#define _PACKED +#endif + +#ifdef _MSC_VER // Visual Studio +#define bswap16(i) _byteswap_ushort(i) +#else +#define bswap16(i) (((i)>>8) | ((i)<<8)) +#endif + + +//************************************** +// Includes +//************************************** +#ifdef __KERNEL__ +#include +#include +#define malloc(size) ({ BUG(); (void*)0; }) +#define free(ptr) ({ BUG(); (void*)0; }) +#else +#include // for malloc +#include // for memset +#include "lz4.h" +#endif + + +//************************************** +// Basic Types +//************************************** +#if defined(_MSC_VER) // Visual Studio does not support 'stdint' natively +#define BYTE unsigned __int8 +#define U16 unsigned __int16 +#define U32 unsigned __int32 +#define S32 __int32 +#define U64 unsigned __int64 +#else +#ifdef __KERNEL__ +#include +#include +#define BYTE u8 +#define U16 u16 +#define U32 u32 +#define S32 s32 +#define U64 u64 + +#else +#include +#define BYTE uint8_t +#define U16 uint16_t +#define U32 uint32_t +#define S32 int32_t +#define U64 uint64_t +#endif // __KERNEL__ +#endif + + +//************************************** +// Constants +//************************************** +#define MINMATCH 4 +#define SKIPSTRENGTH (NONCOMPRESSIBLE_CONFIRMATION>2?NONCOMPRESSIBLE_CONFIRMATION:2) +#define STACKLIMIT 13 +#define HEAPMODE (HASH_LOG>STACKLIMIT) // Defines if memory is allocated into the stack (local variable), or into the heap (malloc()). +#define COPYLENGTH 8 +#define LASTLITERALS 5 +#define MFLIMIT (COPYLENGTH+MINMATCH) +#define MINLENGTH (MFLIMIT+1) + +#define MAXD_LOG 16 +#define MAX_DISTANCE ((1 << MAXD_LOG) - 1) + +#define HASH_LOG COMPRESSIONLEVEL +#define HASHTABLESIZE (1 << HASH_LOG) +#define HASH_MASK (HASHTABLESIZE - 1) + +#define ML_BITS 4 +#define ML_MASK ((1U<v) +#define A32(x) (((U32_S *)(x))->v) +#define A16(x) (((U16_S *)(x))->v) + + +//************************************** +// Macros +//************************************** +#define LZ4_HASH_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8)-HASH_LOG)) +#define LZ4_HASH_VALUE(p) LZ4_HASH_FUNCTION(A32(p)) +#define LZ4_WILDCOPY(s,d,e) do { LZ4_COPYPACKET(s,d) } while (d>3); + #elif defined(__GNUC__) && !defined(_FORCE_SW_BITCOUNT) + return (__builtin_clzll(val) >> 3); + #else + int r; + if (!(val>>32)) { r=4; } else { r=0; val>>=32; } + if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; } + r += (!val); + return r; + #endif +#else + #if defined(_MSC_VER) && !defined(_FORCE_SW_BITCOUNT) + unsigned long r = 0; + _BitScanForward64( &r, val ); + return (int)(r>>3); + #elif defined(__GNUC__) && !defined(_FORCE_SW_BITCOUNT) + return (__builtin_ctzll(val) >> 3); + #else + static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2, 0, 3, 1, 3, 1, 4, 2, 7, 0, 2, 3, 6, 1, 5, 3, 5, 1, 3, 4, 4, 2, 5, 6, 7, 7, 0, 1, 2, 3, 3, 4, 6, 2, 6, 5, 5, 3, 4, 5, 6, 7, 1, 2, 4, 6, 4, 4, 5, 7, 2, 6, 5, 7, 6, 7, 7 }; + return DeBruijnBytePos[((U64)((val & -val) * 0x0218A392CDABBD3F)) >> 58]; + #endif +#endif +} + +#else + +inline static int LZ4_NbCommonBytes (register U32 val) +{ +#if CPU_BIG_ENDIAN + #if defined(_MSC_VER) && !defined(_FORCE_SW_BITCOUNT) + unsigned long r = 0; + _BitScanReverse( &r, val ); + return (int)(r>>3); + #elif defined(__GNUC__) && !defined(_FORCE_SW_BITCOUNT) + return (__builtin_clz(val) >> 3); + #else + int r; + if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; } + r += (!val); + return r; + #endif +#else + #if defined(_MSC_VER) && !defined(_FORCE_SW_BITCOUNT) + unsigned long r = 0; + _BitScanForward( &r, val ); + return (int)(r>>3); + #elif defined(__GNUC__) && !defined(_FORCE_SW_BITCOUNT) + return (__builtin_ctz(val) >> 3); + #else + static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0, 3, 2, 2, 1, 3, 2, 0, 1, 3, 3, 1, 2, 2, 2, 2, 0, 3, 1, 2, 0, 1, 0, 1, 1 }; + return DeBruijnBytePos[((U32)((val & -val) * 0x077CB531U)) >> 27]; + #endif +#endif +} + +#endif + + +//****************************** +// Public Compression functions +//****************************** + +int LZ4_compressCtx(void** ctx, + const char* source, + char* dest, + int isize) +{ +#if HEAPMODE + struct refTables *srt = (struct refTables *) (*ctx); + HTYPE* HashTable; +#else + HTYPE HashTable[HASHTABLESIZE] = {0}; +#endif + + const BYTE* ip = (BYTE*) source; + INITBASE(base); + const BYTE* anchor = ip; + const BYTE* const iend = ip + isize; + const BYTE* const mflimit = iend - MFLIMIT; +#define matchlimit (iend - LASTLITERALS) + + BYTE* op = (BYTE*) dest; + + int len, length; + const int skipStrength = SKIPSTRENGTH; + U32 forwardH; + + + // Init + if (isizehashTable); + memset((void*)HashTable, 0, sizeof(srt->hashTable)); +#else + (void) ctx; +#endif + + + // First Byte + HashTable[LZ4_HASH_VALUE(ip)] = ip - base; + ip++; forwardH = LZ4_HASH_VALUE(ip); + + // Main Loop + for ( ; ; ) + { + int findMatchAttempts = (1U << skipStrength) + 3; + const BYTE* forwardIp = ip; + const BYTE* ref; + BYTE* token; + + // Find a match + do { + U32 h = forwardH; + int step = findMatchAttempts++ >> skipStrength; + ip = forwardIp; + forwardIp = ip + step; + + if (forwardIp > mflimit) { goto _last_literals; } + + forwardH = LZ4_HASH_VALUE(forwardIp); + ref = base + HashTable[h]; + HashTable[h] = ip - base; + + } while ((ref < ip - MAX_DISTANCE) || (A32(ref) != A32(ip))); + + // Catch up + while ((ip>anchor) && (ref>(BYTE*)source) && (ip[-1]==ref[-1])) { ip--; ref--; } + + // Encode Literal length + length = ip - anchor; + token = op++; + if (length>=(int)RUN_MASK) { *token=(RUN_MASK< 254 ; len-=255) *op++ = 255; *op++ = (BYTE)len; } + else *token = (length<=(int)ML_MASK) { *token+=ML_MASK; len-=ML_MASK; for(; len > 509 ; len-=510) { *op++ = 255; *op++ = 255; } if (len > 254) { len-=255; *op++ = 255; } *op++ = (BYTE)len; } + else *token += len; + + // Test end of chunk + if (ip > mflimit) { anchor = ip; break; } + + // Fill table + HashTable[LZ4_HASH_VALUE(ip-2)] = ip - 2 - base; + + // Test next position + ref = base + HashTable[LZ4_HASH_VALUE(ip)]; + HashTable[LZ4_HASH_VALUE(ip)] = ip - base; + if ((ref > ip - (MAX_DISTANCE + 1)) && (A32(ref) == A32(ip))) { token = op++; *token=0; goto _next_match; } + + // Prepare next loop + anchor = ip++; + forwardH = LZ4_HASH_VALUE(ip); + } + +_last_literals: + // Encode Last Literals + { + int lastRun = iend - anchor; + if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK< 254 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; } + else *op++ = (lastRun<> ((MINMATCH*8)-HASHLOG64K)) +#define LZ4_HASH64K_VALUE(p) LZ4_HASH64K_FUNCTION(A32(p)) +int LZ4_compress64kCtx(void** ctx, + const char* source, + char* dest, + int isize) +{ +#if HEAPMODE + struct refTables *srt = (struct refTables *) (*ctx); + U16* HashTable; +#else + U16 HashTable[HASH64KTABLESIZE] = {0}; +#endif + + const BYTE* ip = (BYTE*) source; + const BYTE* anchor = ip; + const BYTE* const base = ip; + const BYTE* const iend = ip + isize; + const BYTE* const mflimit = iend - MFLIMIT; +#define matchlimit (iend - LASTLITERALS) + + BYTE* op = (BYTE*) dest; + + int len, length; + const int skipStrength = SKIPSTRENGTH; + U32 forwardH; + + + // Init + if (isizehashTable); + memset((void*)HashTable, 0, sizeof(srt->hashTable)); +#else + (void) ctx; +#endif + + + // First Byte + ip++; forwardH = LZ4_HASH64K_VALUE(ip); + + // Main Loop + for ( ; ; ) + { + int findMatchAttempts = (1U << skipStrength) + 3; + const BYTE* forwardIp = ip; + const BYTE* ref; + BYTE* token; + + // Find a match + do { + U32 h = forwardH; + int step = findMatchAttempts++ >> skipStrength; + ip = forwardIp; + forwardIp = ip + step; + + if (forwardIp > mflimit) { goto _last_literals; } + + forwardH = LZ4_HASH64K_VALUE(forwardIp); + ref = base + HashTable[h]; + HashTable[h] = ip - base; + + } while (A32(ref) != A32(ip)); + + // Catch up + while ((ip>anchor) && (ref>(BYTE*)source) && (ip[-1]==ref[-1])) { ip--; ref--; } + + // Encode Literal length + length = ip - anchor; + token = op++; + if (length>=(int)RUN_MASK) { *token=(RUN_MASK< 254 ; len-=255) *op++ = 255; *op++ = (BYTE)len; } + else *token = (length<=(int)ML_MASK) { *token+=ML_MASK; len-=ML_MASK; for(; len > 509 ; len-=510) { *op++ = 255; *op++ = 255; } if (len > 254) { len-=255; *op++ = 255; } *op++ = (BYTE)len; } + else *token += len; + + // Test end of chunk + if (ip > mflimit) { anchor = ip; break; } + + // Fill table + HashTable[LZ4_HASH64K_VALUE(ip-2)] = ip - 2 - base; + + // Test next position + ref = base + HashTable[LZ4_HASH64K_VALUE(ip)]; + HashTable[LZ4_HASH64K_VALUE(ip)] = ip - base; + if (A32(ref) == A32(ip)) { token = op++; *token=0; goto _next_match; } + + // Prepare next loop + anchor = ip++; + forwardH = LZ4_HASH64K_VALUE(ip); + } + +_last_literals: + // Encode Last Literals + { + int lastRun = iend - anchor; + if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK< 254 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; } + else *op++ = (lastRun<>ML_BITS)) == RUN_MASK) { for (;(len=*ip++)==255;length+=255){} length += len; } + + // copy literals + cpy = op+length; + if (cpy>oend-COPYLENGTH) + { + if (cpy > oend) goto _output_error; + memcpy(op, ip, length); + ip += length; + break; // Necessarily EOF + } + LZ4_WILDCOPY(ip, op, cpy); ip -= (op-cpy); op = cpy; + + // get offset + LZ4_READ_LITTLEENDIAN_16(ref,cpy,ip); ip+=2; + if (ref < (BYTE* const)dest) goto _output_error; + + // get matchlength + if ((length=(token&ML_MASK)) == ML_MASK) { for (;*ip==255;length+=255) {ip++;} length += *ip++; } + + // copy repeated sequence + if (op-refoend-COPYLENGTH) + { + if (cpy > oend) goto _output_error; + LZ4_SECURECOPY(ref, op, (oend-COPYLENGTH)); + while(op>ML_BITS)) == RUN_MASK) { for (;(len=*ip++)==255;length+=255){} length += len; } + + // copy literals + cpy = op+length; + if (cpy>oend-COPYLENGTH) + { + if (cpy > oend) goto _output_error; + memcpy(op, ip, length); + op += length; + break; // Necessarily EOF + } + LZ4_WILDCOPY(ip, op, cpy); ip -= (op-cpy); op = cpy; + if (ip>=iend) break; // check EOF + + // get offset + LZ4_READ_LITTLEENDIAN_16(ref,cpy,ip); ip+=2; + if (ref < (BYTE* const)dest) goto _output_error; + + // get matchlength + if ((length=(token&ML_MASK)) == ML_MASK) { for (;(len=*ip++)==255;length+=255){} length += len; } + + // copy repeated sequence + if (op-refoend-COPYLENGTH) + { + if (cpy > oend) goto _output_error; + LZ4_SECURECOPY(ref, op, (oend-COPYLENGTH)); + while(op