diff options
Diffstat (limited to 'ext')
33 files changed, 4082 insertions, 1930 deletions
diff --git a/ext/lz4/bench.c b/ext/lz4/bench.c deleted file mode 100644 index d1858116..00000000 --- a/ext/lz4/bench.c +++ /dev/null @@ -1,477 +0,0 @@ -/*
- bench.c - Demo program to benchmark open-source compression algorithm
- Copyright (C) Yann Collet 2012
- GPL v2 License
-
- This program is free software; you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation; either version 2 of the License, or
- (at your option) any later version.
-
- This program is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License along
- with this program; if not, write to the Free Software Foundation, Inc.,
- 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
-
- You can contact the author at :
- - LZ4 homepage : http://fastcompression.blogspot.com/p/lz4.html
- - LZ4 source repository : http://code.google.com/p/lz4/
-*/
-
-//**************************************
-// Compiler Options
-//**************************************
-// Disable some Visual warning messages
-#define _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_DEPRECATE // VS2005
-
-// Unix Large Files support (>4GB)
-#if (defined(__sun__) && (!defined(__LP64__))) // Sun Solaris 32-bits requires specific definitions
-# define _LARGEFILE_SOURCE
-# define FILE_OFFSET_BITS=64
-#elif ! defined(__LP64__) // No point defining Large file for 64 bit
-# define _LARGEFILE64_SOURCE
-#endif
-
-// S_ISREG & gettimeofday() are not supported by MSVC
-#if defined(_MSC_VER)
-# define S_ISREG(x) (((x) & S_IFMT) == S_IFREG)
-# define BMK_LEGACY_TIMER 1
-#endif
-
-// GCC does not support _rotl outside of Windows
-#if !defined(_WIN32)
-# define _rotl(x,r) ((x << r) | (x >> (32 - r)))
-#endif
-
-
-//**************************************
-// Includes
-//**************************************
-#include <stdlib.h> // malloc
-#include <stdio.h> // fprintf, fopen, ftello64
-#include <sys/types.h> // stat64
-#include <sys/stat.h> // stat64
-
-// Use ftime() if gettimeofday() is not available on your target
-#if defined(BMK_LEGACY_TIMER)
-# include <sys/timeb.h> // timeb, ftime
-#else
-# include <sys/time.h> // gettimeofday
-#endif
-
-#include "lz4.h"
-#define COMPRESSOR0 LZ4_compress
-#include "lz4hc.h"
-#define COMPRESSOR1 LZ4_compressHC
-#define DEFAULTCOMPRESSOR LZ4_compress
-
-
-
-//**************************************
-// 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
-#include <stdint.h>
-#define BYTE uint8_t
-#define U16 uint16_t
-#define U32 uint32_t
-#define S32 int32_t
-#define U64 uint64_t
-#endif
-
-
-//**************************************
-// Constants
-//**************************************
-#define NBLOOPS 3
-#define TIMELOOP 2000
-
-#define KNUTH 2654435761U
-#define MAX_MEM (1984<<20)
-#define DEFAULT_CHUNKSIZE (8<<20)
-
-
-//**************************************
-// Local structures
-//**************************************
-struct chunkParameters
-{
- U32 id;
- char* inputBuffer;
- char* outputBuffer;
- int inputSize;
- int outputSize;
-};
-
-struct compressionParameters
-{
- int (*compressionFunction)(const char*, char*, int);
- int (*decompressionFunction)(const char*, char*, int);
-};
-
-
-//**************************************
-// MACRO
-//**************************************
-#define DISPLAY(...) fprintf(stderr, __VA_ARGS__)
-
-
-
-//**************************************
-// Benchmark Parameters
-//**************************************
-static int chunkSize = DEFAULT_CHUNKSIZE;
-static int nbIterations = NBLOOPS;
-static int BMK_pause = 0;
-
-void BMK_SetBlocksize(int bsize)
-{
- chunkSize = bsize;
- DISPLAY("-Using Block Size of %i KB-", chunkSize>>10);
-}
-
-void BMK_SetNbIterations(int nbLoops)
-{
- nbIterations = nbLoops;
- DISPLAY("- %i iterations-", nbIterations);
-}
-
-void BMK_SetPause()
-{
- BMK_pause = 1;
-}
-
-//*********************************************************
-// Private functions
-//*********************************************************
-
-#if defined(BMK_LEGACY_TIMER)
-
-static int BMK_GetMilliStart()
-{
- // Based on Legacy ftime()
- // Rolls over every ~ 12.1 days (0x100000/24/60/60)
- // Use GetMilliSpan to correct for rollover
- struct timeb tb;
- int nCount;
- ftime( &tb );
- nCount = (int) (tb.millitm + (tb.time & 0xfffff) * 1000);
- return nCount;
-}
-
-#else
-
-static int BMK_GetMilliStart()
-{
- // Based on newer gettimeofday()
- // Use GetMilliSpan to correct for rollover
- struct timeval tv;
- int nCount;
- gettimeofday(&tv, NULL);
- nCount = (int) (tv.tv_usec/1000 + (tv.tv_sec & 0xfffff) * 1000);
- return nCount;
-}
-
-#endif
-
-
-static int BMK_GetMilliSpan( int nTimeStart )
-{
- int nSpan = BMK_GetMilliStart() - nTimeStart;
- if ( nSpan < 0 )
- nSpan += 0x100000 * 1000;
- return nSpan;
-}
-
-
-static U32 BMK_checksum_MMH3A (char* buff, U32 length)
-{
- const BYTE* data = (const BYTE*)buff;
- const int nblocks = length >> 2;
-
- U32 h1 = KNUTH;
- U32 c1 = 0xcc9e2d51;
- U32 c2 = 0x1b873593;
-
- const U32* blocks = (const U32*)(data + nblocks*4);
- int i;
-
- for(i = -nblocks; i; i++)
- {
- U32 k1 = blocks[i];
-
- k1 *= c1;
- k1 = _rotl(k1,15);
- k1 *= c2;
-
- h1 ^= k1;
- h1 = _rotl(h1,13);
- h1 = h1*5+0xe6546b64;
- }
-
- {
- const BYTE* tail = (const BYTE*)(data + nblocks*4);
- U32 k1 = 0;
-
- switch(length & 3)
- {
- case 3: k1 ^= tail[2] << 16;
- case 2: k1 ^= tail[1] << 8;
- case 1: k1 ^= tail[0];
- k1 *= c1; k1 = _rotl(k1,15); k1 *= c2; h1 ^= k1;
- };
- }
-
- h1 ^= length;
- h1 ^= h1 >> 16;
- h1 *= 0x85ebca6b;
- h1 ^= h1 >> 13;
- h1 *= 0xc2b2ae35;
- h1 ^= h1 >> 16;
-
- return h1;
-}
-
-
-static size_t BMK_findMaxMem(U64 requiredMem)
-{
- size_t step = (64U<<20); // 64 MB
- BYTE* testmem=NULL;
-
- requiredMem = (((requiredMem >> 25) + 1) << 26);
- if (requiredMem > MAX_MEM) requiredMem = MAX_MEM;
-
- requiredMem += 2*step;
- while (!testmem)
- {
- requiredMem -= step;
- testmem = malloc ((size_t)requiredMem);
- }
-
- free (testmem);
- return (size_t) (requiredMem - step);
-}
-
-
-static U64 BMK_GetFileSize(char* infilename)
-{
- int r;
-#if defined(_MSC_VER)
- struct _stat64 statbuf;
- r = _stat64(infilename, &statbuf);
-#else
- struct stat statbuf;
- r = stat(infilename, &statbuf);
-#endif
- if (r || !S_ISREG(statbuf.st_mode)) return 0; // No good...
- return (U64)statbuf.st_size;
-}
-
-
-//*********************************************************
-// Public function
-//*********************************************************
-
-int BMK_benchFile(char** fileNamesTable, int nbFiles, int cLevel)
-{
- int fileIdx=0;
- FILE* fileIn;
- char* infilename;
- U64 largefilesize;
- size_t benchedsize;
- int nbChunks;
- int maxCChunkSize;
- size_t readSize;
- char* in_buff;
- char* out_buff; int out_buff_size;
- struct chunkParameters* chunkP;
- U32 crcc, crcd=0;
- struct compressionParameters compP;
-
- U64 totals = 0;
- U64 totalz = 0;
- double totalc = 0.;
- double totald = 0.;
-
-
- // Init
- switch (cLevel)
- {
-#ifdef COMPRESSOR0
- case 0 : compP.compressionFunction = COMPRESSOR0; break;
-#endif
-#ifdef COMPRESSOR1
- case 1 : compP.compressionFunction = COMPRESSOR1; break;
-#endif
- default : compP.compressionFunction = DEFAULTCOMPRESSOR;
- }
- compP.decompressionFunction = LZ4_uncompress;
-
- // Loop for each file
- while (fileIdx<nbFiles)
- {
- // Check file existence
- infilename = fileNamesTable[fileIdx++];
- fileIn = fopen( infilename, "rb" );
- if (fileIn==NULL)
- {
- DISPLAY( "Pb opening %s\n", infilename);
- return 11;
- }
-
- // Memory allocation & restrictions
- largefilesize = BMK_GetFileSize(infilename);
- benchedsize = (size_t) BMK_findMaxMem(largefilesize) / 2;
- if ((U64)benchedsize > largefilesize) benchedsize = (size_t)largefilesize;
- if (benchedsize < largefilesize)
- {
- DISPLAY("Not enough memory for '%s' full size; testing %i MB only...\n", infilename, (int)(benchedsize>>20));
- }
-
- // Alloc
- chunkP = (struct chunkParameters*) malloc(((benchedsize / chunkSize)+1) * sizeof(struct chunkParameters));
- in_buff = malloc((size_t )benchedsize);
- nbChunks = (int) (benchedsize / chunkSize) + 1;
- maxCChunkSize = LZ4_compressBound(chunkSize);
- out_buff_size = nbChunks * maxCChunkSize;
- out_buff = malloc((size_t )out_buff_size);
-
-
- if(!in_buff || !out_buff)
- {
- DISPLAY("\nError: not enough memory!\n");
- free(in_buff);
- free(out_buff);
- fclose(fileIn);
- return 12;
- }
-
- // Init chunks data
- {
- int i;
- size_t remaining = benchedsize;
- char* in = in_buff;
- char* out = out_buff;
- for (i=0; i<nbChunks; i++)
- {
- chunkP[i].id = i;
- chunkP[i].inputBuffer = in; in += chunkSize;
- if ((int)remaining > chunkSize) { chunkP[i].inputSize = chunkSize; remaining -= chunkSize; } else { chunkP[i].inputSize = (int)remaining; remaining = 0; }
- chunkP[i].outputBuffer = out; out += maxCChunkSize;
- chunkP[i].outputSize = 0;
- }
- }
-
- // Fill input buffer
- DISPLAY("Loading %s... \r", infilename);
- readSize = fread(in_buff, 1, benchedsize, fileIn);
- fclose(fileIn);
-
- if(readSize != benchedsize)
- {
- DISPLAY("\nError: problem reading file '%s' !! \n", infilename);
- free(in_buff);
- free(out_buff);
- return 13;
- }
-
- // Calculating input Checksum
- crcc = BMK_checksum_MMH3A(in_buff, (unsigned int)benchedsize);
-
-
- // Bench
- {
- int loopNb, nb_loops, chunkNb;
- size_t cSize=0;
- int milliTime;
- double fastestC = 100000000., fastestD = 100000000.;
- double ratio=0.;
-
- DISPLAY("\r%79s\r", "");
- for (loopNb = 1; loopNb <= nbIterations; loopNb++)
- {
- // Compression
- DISPLAY("%1i-%-14.14s : %9i ->\r", loopNb, infilename, (int)benchedsize);
- { size_t i; for (i=0; i<benchedsize; i++) out_buff[i]=(char)i; } // warmimg up memory
-
- nb_loops = 0;
- milliTime = BMK_GetMilliStart();
- while(BMK_GetMilliStart() == milliTime);
- milliTime = BMK_GetMilliStart();
- while(BMK_GetMilliSpan(milliTime) < TIMELOOP)
- {
- for (chunkNb=0; chunkNb<nbChunks; chunkNb++)
- chunkP[chunkNb].outputSize = compP.compressionFunction(chunkP[chunkNb].inputBuffer, chunkP[chunkNb].outputBuffer, chunkP[chunkNb].inputSize);
- nb_loops++;
- }
- milliTime = BMK_GetMilliSpan(milliTime);
-
- if ((double)milliTime < fastestC*nb_loops) fastestC = (double)milliTime/nb_loops;
- cSize=0; for (chunkNb=0; chunkNb<nbChunks; chunkNb++) cSize += chunkP[chunkNb].outputSize;
- ratio = (double)cSize/(double)benchedsize*100.;
-
- DISPLAY("%1i-%-14.14s : %9i -> %9i (%5.2f%%),%7.1f MB/s\r", loopNb, infilename, (int)benchedsize, (int)cSize, ratio, (double)benchedsize / fastestC / 1000.);
-
- // Decompression
- { size_t i; for (i=0; i<benchedsize; i++) in_buff[i]=0; } // zeroing area, for CRC checking
-
- nb_loops = 0;
- milliTime = BMK_GetMilliStart();
- while(BMK_GetMilliStart() == milliTime);
- milliTime = BMK_GetMilliStart();
- while(BMK_GetMilliSpan(milliTime) < TIMELOOP)
- {
- for (chunkNb=0; chunkNb<nbChunks; chunkNb++)
- chunkP[chunkNb].outputSize = LZ4_uncompress(chunkP[chunkNb].outputBuffer, chunkP[chunkNb].inputBuffer, chunkP[chunkNb].inputSize);
- //chunkP[chunkNb].inputSize = LZ4_uncompress_unknownOutputSize(chunkP[chunkNb].outputBuffer, chunkP[chunkNb].inputBuffer, chunkP[chunkNb].outputSize, chunkSize);
- nb_loops++;
- }
- milliTime = BMK_GetMilliSpan(milliTime);
-
- if ((double)milliTime < fastestD*nb_loops) fastestD = (double)milliTime/nb_loops;
- DISPLAY("%1i-%-14.14s : %9i -> %9i (%5.2f%%),%7.1f MB/s ,%7.1f MB/s\r", loopNb, infilename, (int)benchedsize, (int)cSize, ratio, (double)benchedsize / fastestC / 1000., (double)benchedsize / fastestD / 1000.);
-
- // CRC Checking
- crcd = BMK_checksum_MMH3A(in_buff, (unsigned int)benchedsize);
- if (crcc!=crcd) { DISPLAY("\n!!! WARNING !!! %14s : Invalid Checksum : %x != %x\n", infilename, (unsigned)crcc, (unsigned)crcd); break; }
- }
-
- if (crcc==crcd)
- {
- if (ratio<100.)
- DISPLAY("%-16.16s : %9i -> %9i (%5.2f%%),%7.1f MB/s ,%7.1f MB/s\n", infilename, (int)benchedsize, (int)cSize, ratio, (double)benchedsize / fastestC / 1000., (double)benchedsize / fastestD / 1000.);
- else
- DISPLAY("%-16.16s : %9i -> %9i (%5.1f%%),%7.1f MB/s ,%7.1f MB/s \n", infilename, (int)benchedsize, (int)cSize, ratio, (double)benchedsize / fastestC / 1000., (double)benchedsize / fastestD / 1000.);
- }
- totals += benchedsize;
- totalz += cSize;
- totalc += fastestC;
- totald += fastestD;
- }
-
- free(in_buff);
- free(out_buff);
- free(chunkP);
- }
-
- if (nbFiles > 1)
- printf("%-16.16s :%10llu ->%10llu (%5.2f%%), %6.1f MB/s , %6.1f MB/s\n", " TOTAL", (long long unsigned int)totals, (long long unsigned int)totalz, (double)totalz/(double)totals*100., (double)totals/totalc/1000., (double)totals/totald/1000.);
-
- if (BMK_pause) { printf("press enter...\n"); getchar(); }
-
- return 0;
-}
-
-
-
diff --git a/ext/lz4/bench.h b/ext/lz4/bench.h deleted file mode 100644 index a55c3ce2..00000000 --- a/ext/lz4/bench.h +++ /dev/null @@ -1,41 +0,0 @@ -/*
- bench.h - Demo program to benchmark open-source compression algorithm
- Copyright (C) Yann Collet 2012
-
- This program is free software; you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation; either version 2 of the License, or
- (at your option) any later version.
-
- This program is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License along
- with this program; if not, write to the Free Software Foundation, Inc.,
- 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
-
- You can contact the author at :
- - LZ4 homepage : http://fastcompression.blogspot.com/p/lz4.html
- - LZ4 source repository : http://code.google.com/p/lz4/
-*/
-#pragma once
-
-#if defined (__cplusplus)
-extern "C" {
-#endif
-
-
-int BMK_benchFile(char** fileNamesTable, int nbFiles, int cLevel);
-
-// Parameters
-void BMK_SetBlocksize(int bsize);
-void BMK_SetNbIterations(int nbLoops);
-void BMK_SetPause();
-
-
-
-#if defined (__cplusplus)
-}
-#endif
diff --git a/ext/lz4/fuzzer.c b/ext/lz4/fuzzer.c deleted file mode 100644 index 11697fd0..00000000 --- a/ext/lz4/fuzzer.c +++ /dev/null @@ -1,227 +0,0 @@ -/*
- fuzzer.c - Fuzzer test tool for LZ4
- Copyright (C) Andrew Mahone - Yann Collet 2012
- Original code by Andrew Mahone / Modified by Yann Collet
- GPL v2 License
-
- This program is free software; you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation; either version 2 of the License, or
- (at your option) any later version.
-
- This program is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License along
- with this program; if not, write to the Free Software Foundation, Inc.,
- 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
-
- You can contact the author at :
- - LZ4 homepage : http://fastcompression.blogspot.com/p/lz4.html
- - LZ4 source repository : http://code.google.com/p/lz4/
-*/
-
-//**************************************
-// Remove Visual warning messages
-//**************************************
-#define _CRT_SECURE_NO_WARNINGS // fgets
-
-
-//**************************************
-// Includes
-//**************************************
-#include <stdlib.h>
-#include <stdio.h> // fgets, sscanf
-#include <sys/timeb.h> // timeb
-#include "lz4.h"
-
-
-//**************************************
-// Constants
-//**************************************
-#define NB_ATTEMPTS (1<<18)
-#define LEN ((1<<15))
-#define SEQ_POW 2
-#define NUM_SEQ (1 << SEQ_POW)
-#define SEQ_MSK ((NUM_SEQ) - 1)
-#define MOD_SEQ(x) ((((x) >> 8) & 255) == 0)
-#define NEW_SEQ(x) ((((x) >> 10) %10) == 0)
-#define PAGE_SIZE 4096
-#define ROUND_PAGE(x) (((x) + PAGE_SIZE - 1) & ~(PAGE_SIZE - 1))
-#define PRIME1 2654435761U
-#define PRIME2 2246822519U
-#define PRIME3 3266489917U
-
-
-
-//*********************************************************
-// Functions
-//*********************************************************
-static int FUZ_GetMilliStart()
-{
- struct timeb tb;
- int nCount;
- ftime( &tb );
- nCount = (int) (tb.millitm + (tb.time & 0xfffff) * 1000);
- return nCount;
-}
-
-static int FUZ_GetMilliSpan( int nTimeStart )
-{
- int nSpan = FUZ_GetMilliStart() - nTimeStart;
- if ( nSpan < 0 )
- nSpan += 0x100000 * 1000;
- return nSpan;
-}
-
-
-unsigned int FUZ_rand(unsigned int* src)
-{
- *src = ((*src) * PRIME1) + PRIME2;
- return *src;
-}
-
-
-int test_canary(unsigned char *buf) {
- int i;
- for (i = 0; i < 2048; i++)
- if (buf[i] != buf[i + 2048])
- return 0;
- return 1;
-}
-
-int FUZ_SecurityTest()
-{
- char* output;
- char* input;
- int i, r;
-
- printf("Starting security tests...");
- input = (char*) malloc (20<<20);
- output = (char*) malloc (20<<20);
- input[0] = 0x0F;
- input[1] = 0x00;
- input[2] = 0x00;
- for(i = 3; i < 16840000; i++)
- input[i] = 0xff;
- r = LZ4_uncompress(input, output, 20<<20);
-
- free(input);
- free(output);
- printf(" Completed (r=%i)\n",r);
- return 0;
-}
-
-
-//int main(int argc, char *argv[]) {
-int main() {
- unsigned long long bytes = 0;
- unsigned long long cbytes = 0;
- unsigned char buf[LEN];
- unsigned char testOut[LEN+1];
-# define FUZ_max LZ4_COMPRESSBOUND(LEN)
-# define FUZ_avail ROUND_PAGE(FUZ_max)
- const int off_full = FUZ_avail - FUZ_max;
- unsigned char cbuf[FUZ_avail + PAGE_SIZE];
- unsigned int seed, cur_seq=PRIME3, seeds[NUM_SEQ], timestamp=FUZ_GetMilliStart();
- int i, j, k, ret, len;
- char userInput[30] = {0};
-
- printf("starting LZ4 fuzzer\n");
- printf("Select an Initialisation number (default : random) : ");
- fflush(stdout);
- if ( fgets(userInput, sizeof userInput, stdin) )
- {
- if ( sscanf(userInput, "%d", &seed) == 1 ) {}
- else seed = FUZ_GetMilliSpan(timestamp);
- }
- printf("Seed = %u\n", seed);
-
- FUZ_SecurityTest();
-
- for (i = 0; i < 2048; i++)
- cbuf[FUZ_avail + i] = cbuf[FUZ_avail + 2048 + i] = FUZ_rand(&seed) >> 16;
-
- for (i = 0; i < NB_ATTEMPTS; i++)
- {
- printf("\r%7i /%7i\r", i, NB_ATTEMPTS);
-
- FUZ_rand(&seed);
- for (j = 0; j < NUM_SEQ; j++) {
- seeds[j] = FUZ_rand(&seed) << 8;
- seeds[j] ^= (FUZ_rand(&seed) >> 8) & 65535;
- }
- for (j = 0; j < LEN; j++) {
- k = FUZ_rand(&seed);
- if (j == 0 || NEW_SEQ(k))
- cur_seq = seeds[(FUZ_rand(&seed) >> 16) & SEQ_MSK];
- if (MOD_SEQ(k)) {
- k = (FUZ_rand(&seed) >> 16) & SEQ_MSK;
- seeds[k] = FUZ_rand(&seed) << 8;
- seeds[k] ^= (FUZ_rand(&seed) >> 8) & 65535;
- }
- buf[j] = FUZ_rand(&cur_seq) >> 16;
- }
-
- // Test compression
- ret = LZ4_compress_limitedOutput((const char*)buf, (char*)&cbuf[off_full], LEN, FUZ_max);
- if (ret == 0) { printf("compression failed despite sufficient space: seed %u, len %d\n", seed, LEN); goto _output_error; }
- len = ret;
-
- // Test decoding with output size being exactly what's necessary => must work
- ret = LZ4_uncompress((char*)&cbuf[off_full], (char*)testOut, LEN);
- if (ret<0) { printf("decompression failed despite correct space: seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test decoding with one byte missing => must fail
- ret = LZ4_uncompress((char*)&cbuf[off_full], (char*)testOut, LEN-1);
- if (ret>=0) { printf("decompression should have failed, due to Output Size being too small : seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test decoding with one byte too much => must fail
- ret = LZ4_uncompress((char*)&cbuf[off_full], (char*)testOut, LEN+1);
- if (ret>=0) { printf("decompression should have failed, due to Output Size being too large : seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test decoding with enough output size => must work
- ret = LZ4_uncompress_unknownOutputSize((char*)&cbuf[off_full], (char*)testOut, len, LEN+1);
- if (ret<0) { printf("decompression failed despite sufficient space: seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test decoding with output size being exactly what's necessary => must work
- ret = LZ4_uncompress_unknownOutputSize((char*)&cbuf[off_full], (char*)testOut, len, LEN);
- if (ret<0) { printf("decompression failed despite sufficient space: seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test decoding with output size being one byte too short => must fail
- ret = LZ4_uncompress_unknownOutputSize((char*)&cbuf[off_full], (char*)testOut, len, LEN-1);
- if (ret>=0) { printf("decompression should have failed, due to Output Size being too small : seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test decoding with input size being one byte too short => must fail
- ret = LZ4_uncompress_unknownOutputSize((char*)&cbuf[off_full], (char*)testOut, len-1, LEN);
- if (ret>=0) { printf("decompression should have failed, due to input size being too small : seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test decoding with input size being one byte too large => must fail
- ret = LZ4_uncompress_unknownOutputSize((char*)&cbuf[off_full], (char*)testOut, len+1, LEN);
- if (ret>=0) { printf("decompression should have failed, due to input size being too large : seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test compression with output size being exactly what's necessary (should work)
- ret = LZ4_compress_limitedOutput((const char*)buf, (char*)&cbuf[FUZ_avail-len], LEN, len);
- if (!test_canary(&cbuf[FUZ_avail])) { printf("compression overran output buffer: seed %u, len %d, olen %d\n", seed, LEN, len); goto _output_error; }
- if (ret == 0) { printf("compression failed despite sufficient space: seed %u, len %d\n", seed, LEN); goto _output_error; }
-
- // Test compression with just one missing byte into output buffer => must fail
- ret = LZ4_compress_limitedOutput((const char*)buf, (char*)&cbuf[FUZ_avail-(len-1)], LEN, len-1);
- if (ret) { printf("compression overran output buffer: seed %u, len %d, olen %d => ret %d", seed, LEN, len-1, ret); goto _output_error; }
- if (!test_canary(&cbuf[FUZ_avail])) { printf("compression overran output buffer: seed %u, len %d, olen %d", seed, LEN, len-1); goto _output_error; }
-
- bytes += LEN;
- cbytes += len;
- }
-
- printf("all tests completed successfully \n");
- printf("compression ratio: %0.3f%%\n", (double)cbytes/bytes*100);
- getchar();
- return 0;
-
-_output_error:
- getchar();
- return 1;
-}
diff --git a/ext/lz4/lz4.c b/ext/lz4/lz4.c index 1f2eafde..5668521f 100644 --- a/ext/lz4/lz4.c +++ b/ext/lz4/lz4.c @@ -1,6 +1,6 @@ /*
LZ4 - Fast LZ compression algorithm
- Copyright (C) 2011-2012, Yann Collet.
+ Copyright (C) 2011-2013, 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
@@ -27,8 +27,8 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
You can contact the author at :
- - LZ4 homepage : http://fastcompression.blogspot.com/p/lz4.html
- LZ4 source repository : http://code.google.com/p/lz4/
+ - LZ4 public forum : https://groups.google.com/forum/#!forum/lz4c
*/
//**************************************
@@ -41,19 +41,20 @@ // Default value is 14, for 16KB, which nicely fits into Intel x86 L1 cache
#define MEMORY_USAGE 14
-// BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE :
-// This will provide a small 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 remain within closed environment
-// This option is useless on Little_Endian CPU (such as x86)
-//#define BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE 1
-
+// HEAPMODE :
+// Select how default compression functions will allocate memory for their hash table,
+// in memory stack (0:default, fastest), or in memory heap (1:requires memory allocation (malloc)).
+#define HEAPMODE 0
//**************************************
// CPU Feature Detection
//**************************************
// 32 or 64 bits ?
-#if (defined(__x86_64__) || defined(__x86_64) || defined(__amd64__) || defined(__amd64) || defined(__ppc64__) || defined(_WIN64) || defined(__LP64__) || defined(_LP64) ) // Detects 64 bits mode
+#if (defined(__x86_64__) || defined(_M_X64) || defined(_WIN64) \
+ || defined(__powerpc64__) || defined(__ppc64__) || defined(__PPC64__) \
+ || defined(__64BIT__) || defined(_LP64) || defined(__LP64__) \
+ || defined(__ia64) || defined(__itanium__) || defined(_M_IA64) ) // Detects 64 bits mode
# define LZ4_ARCH64 1
#else
# define LZ4_ARCH64 0
@@ -69,7 +70,7 @@ #elif (defined(__BIG_ENDIAN__) || defined(__BIG_ENDIAN) || defined(_BIG_ENDIAN)) && !(defined(__LITTLE_ENDIAN__) || defined(__LITTLE_ENDIAN) || defined(_LITTLE_ENDIAN))
# define LZ4_BIG_ENDIAN 1
#elif defined(__sparc) || defined(__sparc__) \
- || defined(__ppc__) || defined(_POWER) || defined(__powerpc__) || defined(_ARCH_PPC) || defined(__PPC__) || defined(__PPC) || defined(PPC) || defined(__powerpc__) || defined(__powerpc) || defined(powerpc) \
+ || defined(__powerpc__) || defined(__ppc__) || defined(__PPC__) \
|| defined(__hpux) || defined(__hppa) \
|| defined(_MIPSEB) || defined(__s390__)
# define LZ4_BIG_ENDIAN 1
@@ -78,7 +79,7 @@ #endif
// Unaligned memory access is automatically enabled for "common" CPU, such as x86.
-// For others CPU, the compiler will be more cautious, and insert extra code to ensure aligned access is respected
+// For others CPU, such as ARM, the compiler may be more cautious, inserting unnecessary extra code to ensure aligned access property
// If you know your target CPU supports unaligned memory access, you want to force this option manually to improve performance
#if defined(__ARM_FEATURE_UNALIGNED)
# define LZ4_FORCE_UNALIGNED_ACCESS 1
@@ -89,27 +90,39 @@ # define LZ4_FORCE_SW_BITCOUNT
#endif
+// BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE :
+// This option may provide a small boost to performance for some big endian cpu, although probably modest.
+// You may set this option to 1 if data will remain within closed environment.
+// This option is useless on Little_Endian CPU (such as x86)
+//#define BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE 1
+
//**************************************
// Compiler Options
//**************************************
-#if __STDC_VERSION__ >= 199901L // C99
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) // C99
/* "restrict" is a known keyword */
#else
# define restrict // Disable restrict
#endif
-#define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)
-
-#ifdef _MSC_VER // Visual Studio
-# include <intrin.h> // For Visual 2005
-# if LZ4_ARCH64 // 64-bit
+#ifdef _MSC_VER // Visual Studio
+# define FORCE_INLINE static __forceinline
+# include <intrin.h> // For Visual 2005
+# if LZ4_ARCH64 // 64-bits
# pragma intrinsic(_BitScanForward64) // For Visual 2005
# pragma intrinsic(_BitScanReverse64) // For Visual 2005
-# else
+# else // 32-bits
# pragma intrinsic(_BitScanForward) // For Visual 2005
# pragma intrinsic(_BitScanReverse) // For Visual 2005
# endif
+# pragma warning(disable : 4127) // disable: C4127: conditional expression is constant
+#else
+# ifdef __GNUC__
+# define FORCE_INLINE static inline __attribute__((always_inline))
+# else
+# define FORCE_INLINE static inline
+# endif
#endif
#ifdef _MSC_VER
@@ -118,6 +131,8 @@ # define lz4_bswap16(x) ((unsigned short int) ((((x) >> 8) & 0xffu) | (((x) & 0xffu) << 8)))
#endif
+#define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)
+
#if (GCC_VERSION >= 302) || (__INTEL_COMPILER >= 800) || defined(__clang__)
# define expect(expr,value) (__builtin_expect ((expr),(value)) )
#else
@@ -129,71 +144,88 @@ //**************************************
+// Memory routines
+//**************************************
+#include <stdlib.h> // malloc, calloc, free
+#define ALLOCATOR(n,s) calloc(n,s)
+#define FREEMEM free
+#include <string.h> // memset, memcpy
+#define MEM_INIT memset
+
+
+//**************************************
// Includes
//**************************************
-#include <stdlib.h> // for malloc
-#include <string.h> // for memset
#include "lz4.h"
//**************************************
// 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
+#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L // C99
+# include <stdint.h>
+ typedef uint8_t BYTE;
+ typedef uint16_t U16;
+ typedef uint32_t U32;
+ typedef int32_t S32;
+ typedef uint64_t U64;
#else
-# include <stdint.h>
-# define BYTE uint8_t
-# define U16 uint16_t
-# define U32 uint32_t
-# define S32 int32_t
-# define U64 uint64_t
+ typedef unsigned char BYTE;
+ typedef unsigned short U16;
+ typedef unsigned int U32;
+ typedef signed int S32;
+ typedef unsigned long long U64;
#endif
-#ifndef LZ4_FORCE_UNALIGNED_ACCESS
-# pragma pack(push, 1)
+#if defined(__GNUC__) && !defined(LZ4_FORCE_UNALIGNED_ACCESS)
+# define _PACKED __attribute__ ((packed))
+#else
+# define _PACKED
+#endif
+
+#if !defined(LZ4_FORCE_UNALIGNED_ACCESS) && !defined(__GNUC__)
+# if defined(__IBMC__) || defined(__SUNPRO_C) || defined(__SUNPRO_CC)
+# pragma pack(1)
+# else
+# pragma pack(push, 1)
+# endif
#endif
-typedef struct _U16_S { U16 v; } U16_S;
-typedef struct _U32_S { U32 v; } U32_S;
-typedef struct _U64_S { U64 v; } U64_S;
+typedef struct { U16 v; } _PACKED U16_S;
+typedef struct { U32 v; } _PACKED U32_S;
+typedef struct { U64 v; } _PACKED U64_S;
+typedef struct {size_t v;} _PACKED size_t_S;
-#ifndef LZ4_FORCE_UNALIGNED_ACCESS
-# pragma pack(pop)
+#if !defined(LZ4_FORCE_UNALIGNED_ACCESS) && !defined(__GNUC__)
+# if defined(__SUNPRO_C) || defined(__SUNPRO_CC)
+# pragma pack(0)
+# else
+# pragma pack(pop)
+# endif
#endif
-#define A64(x) (((U64_S *)(x))->v)
-#define A32(x) (((U32_S *)(x))->v)
-#define A16(x) (((U16_S *)(x))->v)
+#define A16(x) (((U16_S *)(x))->v)
+#define A32(x) (((U32_S *)(x))->v)
+#define A64(x) (((U64_S *)(x))->v)
+#define AARCH(x) (((size_t_S *)(x))->v)
//**************************************
// Constants
//**************************************
+#define LZ4_HASHLOG (MEMORY_USAGE-2)
+#define HASHTABLESIZE (1 << MEMORY_USAGE)
+#define HASHNBCELLS4 (1 << LZ4_HASHLOG)
+
#define MINMATCH 4
-#define HASH_LOG (MEMORY_USAGE-2)
-#define HASHTABLESIZE (1 << HASH_LOG)
-#define HASH_MASK (HASHTABLESIZE - 1)
-
-// NOTCOMPRESSIBLE_DETECTIONLEVEL :
-// 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 NOTCOMPRESSIBLE_DETECTIONLEVEL 6
-#define SKIPSTRENGTH (NOTCOMPRESSIBLE_DETECTIONLEVEL>2?NOTCOMPRESSIBLE_DETECTIONLEVEL: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)
+const int LZ4_minLength = (MFLIMIT+1);
+
+#define LZ4_64KLIMIT ((1<<16) + (MFLIMIT-1))
+#define SKIPSTRENGTH 6 // Increasing this value will make the compression run slower on incompressible data
#define MAXD_LOG 16
#define MAX_DISTANCE ((1 << MAXD_LOG) - 1)
@@ -203,26 +235,43 @@ typedef struct _U64_S { U64 v; } U64_S; #define RUN_BITS (8-ML_BITS)
#define RUN_MASK ((1U<<RUN_BITS)-1)
+#define KB *(1U<<10)
+#define MB *(1U<<20)
+#define GB *(1U<<30)
+
+
+//**************************************
+// Structures and local types
+//**************************************
+
+typedef struct {
+ U32 hashTable[HASHNBCELLS4];
+ const BYTE* bufferStart;
+ const BYTE* base;
+ const BYTE* nextBlock;
+} LZ4_Data_Structure;
+
+typedef enum { notLimited = 0, limited = 1 } limitedOutput_directive;
+typedef enum { byPtr, byU32, byU16 } tableType_t;
+
+typedef enum { noPrefix = 0, withPrefix = 1 } prefix64k_directive;
+
+typedef enum { endOnOutputSize = 0, endOnInputSize = 1 } endCondition_directive;
+typedef enum { full = 0, partial = 1 } earlyEnd_directive;
+
//**************************************
// Architecture-specific macros
//**************************************
-#if LZ4_ARCH64 // 64-bit
-# define STEPSIZE 8
-# define UARCH U64
-# define AARCH A64
-# define LZ4_COPYSTEP(s,d) A64(d) = A64(s); d+=8; s+=8;
-# define LZ4_COPYPACKET(s,d) LZ4_COPYSTEP(s,d)
-# define LZ4_SECURECOPY(s,d,e) if (d<e) LZ4_WILDCOPY(s,d,e)
+#define STEPSIZE sizeof(size_t)
+#define LZ4_COPYSTEP(d,s) { AARCH(d) = AARCH(s); d+=STEPSIZE; s+=STEPSIZE; }
+#define LZ4_COPY8(d,s) { LZ4_COPYSTEP(d,s); if (STEPSIZE<8) LZ4_COPYSTEP(d,s); }
+#define LZ4_SECURECOPY(d,s,e) { if ((STEPSIZE==4)||(d<e)) LZ4_WILDCOPY(d,s,e); }
+
+#if LZ4_ARCH64 // 64-bit
# define HTYPE U32
# define INITBASE(base) const BYTE* const base = ip
-#else // 32-bit
-# define STEPSIZE 4
-# define UARCH U32
-# define AARCH A32
-# define LZ4_COPYSTEP(s,d) A32(d) = A32(s); d+=4; s+=4;
-# define LZ4_COPYPACKET(s,d) LZ4_COPYSTEP(s,d); LZ4_COPYSTEP(s,d);
-# define LZ4_SECURECOPY LZ4_WILDCOPY
+#else // 32-bit
# define HTYPE const BYTE*
# define INITBASE(base) const int base = 0
#endif
@@ -230,28 +279,16 @@ typedef struct _U64_S { U64 v; } U64_S; #if (defined(LZ4_BIG_ENDIAN) && !defined(BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE))
# define LZ4_READ_LITTLEENDIAN_16(d,s,p) { U16 v = A16(p); v = lz4_bswap16(v); d = (s) - v; }
# define LZ4_WRITE_LITTLEENDIAN_16(p,i) { U16 v = (U16)(i); v = lz4_bswap16(v); A16(p) = v; p+=2; }
-#else // Little Endian
+#else // Little Endian
# define LZ4_READ_LITTLEENDIAN_16(d,s,p) { d = (s) - A16(p); }
# define LZ4_WRITE_LITTLEENDIAN_16(p,v) { A16(p) = v; p+=2; }
#endif
//**************************************
-// Local structures
-//**************************************
-struct refTables
-{
- HTYPE hashTable[HASHTABLESIZE];
-};
-
-
-//**************************************
// 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<e);
-#define LZ4_BLINDCOPY(s,d,l) { BYTE* e=(d)+l; LZ4_WILDCOPY(s,d,e); d=e; }
+#define LZ4_WILDCOPY(d,s,e) { do { LZ4_COPY8(d,s) } while (d<e); } // at the end, d>=e;
//****************************
@@ -259,100 +296,131 @@ struct refTables //****************************
#if LZ4_ARCH64
-static inline int LZ4_NbCommonBytes (register U64 val)
+FORCE_INLINE int LZ4_NbCommonBytes (register U64 val)
{
-#if defined(LZ4_BIG_ENDIAN)
- #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# if defined(LZ4_BIG_ENDIAN)
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r = 0;
_BitScanReverse64( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_clzll(val) >> 3);
- #else
+# 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(LZ4_FORCE_SW_BITCOUNT)
+# endif
+# else
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r = 0;
_BitScanForward64( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_ctzll(val) >> 3);
- #else
+# 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
+ return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58];
+# endif
+# endif
}
#else
-static inline int LZ4_NbCommonBytes (register U32 val)
+FORCE_INLINE int LZ4_NbCommonBytes (register U32 val)
{
-#if defined(LZ4_BIG_ENDIAN)
- #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# if defined(LZ4_BIG_ENDIAN)
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r = 0;
_BitScanReverse( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_clz(val) >> 3);
- #else
+# 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(LZ4_FORCE_SW_BITCOUNT)
+# endif
+# else
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r;
_BitScanForward( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && (GCC_VERSION >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_ctz(val) >> 3);
- #else
+# 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 & -(S32)val) * 0x077CB531U)) >> 27];
- #endif
-#endif
+# endif
+# endif
}
#endif
-
-//******************************
+//****************************
// Compression functions
-//******************************
+//****************************
+FORCE_INLINE int LZ4_hashSequence(U32 sequence, tableType_t tableType)
+{
+ if (tableType == byU16)
+ return (((sequence) * 2654435761U) >> ((MINMATCH*8)-(LZ4_HASHLOG+1)));
+ else
+ return (((sequence) * 2654435761U) >> ((MINMATCH*8)-LZ4_HASHLOG));
+}
-// LZ4_compressCtx :
-// -----------------
-// Compress 'isize' bytes from 'source' into an output buffer 'dest' of maximum size 'maxOutputSize'.
-// If it cannot achieve it, compression will stop, and result of the function will be zero.
-// return : the number of bytes written in buffer 'dest', or 0 if the compression fails
+FORCE_INLINE int LZ4_hashPosition(const BYTE* p, tableType_t tableType) { return LZ4_hashSequence(A32(p), tableType); }
-static inline int LZ4_compressCtx(void** ctx,
+FORCE_INLINE void LZ4_putPositionOnHash(const BYTE* p, U32 h, void* tableBase, tableType_t tableType, const BYTE* srcBase)
+{
+ switch (tableType)
+ {
+ case byPtr: { const BYTE** hashTable = (const BYTE**) tableBase; hashTable[h] = p; break; }
+ case byU32: { U32* hashTable = (U32*) tableBase; hashTable[h] = (U32)(p-srcBase); break; }
+ case byU16: { U16* hashTable = (U16*) tableBase; hashTable[h] = (U16)(p-srcBase); break; }
+ }
+}
+
+FORCE_INLINE void LZ4_putPosition(const BYTE* p, void* tableBase, tableType_t tableType, const BYTE* srcBase)
+{
+ U32 h = LZ4_hashPosition(p, tableType);
+ LZ4_putPositionOnHash(p, h, tableBase, tableType, srcBase);
+}
+
+FORCE_INLINE const BYTE* LZ4_getPositionOnHash(U32 h, void* tableBase, tableType_t tableType, const BYTE* srcBase)
+{
+ if (tableType == byPtr) { const BYTE** hashTable = (const BYTE**) tableBase; return hashTable[h]; }
+ if (tableType == byU32) { U32* hashTable = (U32*) tableBase; return hashTable[h] + srcBase; }
+ { U16* hashTable = (U16*) tableBase; return hashTable[h] + srcBase; } // default, to ensure a return
+}
+
+FORCE_INLINE const BYTE* LZ4_getPosition(const BYTE* p, void* tableBase, tableType_t tableType, const BYTE* srcBase)
+{
+ U32 h = LZ4_hashPosition(p, tableType);
+ return LZ4_getPositionOnHash(h, tableBase, tableType, srcBase);
+}
+
+
+FORCE_INLINE int LZ4_compress_generic(
+ void* ctx,
const char* source,
char* dest,
- int isize,
- int maxOutputSize)
-{
-#if HEAPMODE
- struct refTables *srt = (struct refTables *) (*ctx);
- HTYPE* HashTable;
-#else
- HTYPE HashTable[HASHTABLESIZE] = {0};
-#endif
+ int inputSize,
+ int maxOutputSize,
- const BYTE* ip = (BYTE*) source;
- INITBASE(base);
- const BYTE* anchor = ip;
- const BYTE* const iend = ip + isize;
+ limitedOutput_directive limitedOutput,
+ tableType_t tableType,
+ prefix64k_directive prefix)
+{
+ const BYTE* ip = (const BYTE*) source;
+ const BYTE* const base = (prefix==withPrefix) ? ((LZ4_Data_Structure*)ctx)->base : (const BYTE*) source;
+ const BYTE* const lowLimit = ((prefix==withPrefix) ? ((LZ4_Data_Structure*)ctx)->bufferStart : (const BYTE*)source);
+ const BYTE* anchor = (const BYTE*) source;
+ const BYTE* const iend = ip + inputSize;
const BYTE* const mflimit = iend - MFLIMIT;
-#define matchlimit (iend - LASTLITERALS)
+ const BYTE* const matchlimit = iend - LASTLITERALS;
BYTE* op = (BYTE*) dest;
BYTE* const oend = op + maxOutputSize;
@@ -361,25 +429,16 @@ static inline int LZ4_compressCtx(void** ctx, const int skipStrength = SKIPSTRENGTH;
U32 forwardH;
-
- // Init
- if (isize<MINLENGTH) goto _last_literals;
-#if HEAPMODE
- if (*ctx == NULL)
- {
- srt = (struct refTables *) malloc ( sizeof(struct refTables) );
- *ctx = (void*) srt;
- }
- HashTable = (HTYPE*)(srt->hashTable);
- memset((void*)HashTable, 0, sizeof(srt->hashTable));
-#else
- (void) ctx;
-#endif
-
+ // Init conditions
+ if ((U32)inputSize > (U32)LZ4_MAX_INPUT_SIZE) return 0; // Unsupported input size, too large (or negative)
+ if ((prefix==withPrefix) && (ip != ((LZ4_Data_Structure*)ctx)->nextBlock)) return 0; // must continue from end of previous block
+ if (prefix==withPrefix) ((LZ4_Data_Structure*)ctx)->nextBlock=iend; // do it now, due to potential early exit
+ if ((tableType == byU16) && (inputSize>=LZ4_64KLIMIT)) return 0; // Size too large (not within 64K limit)
+ if (inputSize<LZ4_minLength) goto _last_literals; // Input too small, no compression (all literals)
// First Byte
- HashTable[LZ4_HASH_VALUE(ip)] = ip - base;
- ip++; forwardH = LZ4_HASH_VALUE(ip);
+ LZ4_putPosition(ip, ctx, tableType, base);
+ ip++; forwardH = LZ4_hashPosition(ip, tableType);
// Main Loop
for ( ; ; )
@@ -398,50 +457,30 @@ static inline int LZ4_compressCtx(void** ctx, if unlikely(forwardIp > mflimit) { goto _last_literals; }
- forwardH = LZ4_HASH_VALUE(forwardIp);
- ref = base + HashTable[h];
- HashTable[h] = ip - base;
+ forwardH = LZ4_hashPosition(forwardIp, tableType);
+ ref = LZ4_getPositionOnHash(h, ctx, tableType, base);
+ LZ4_putPositionOnHash(ip, h, ctx, tableType, base);
- } while ((ref < ip - MAX_DISTANCE) || (A32(ref) != A32(ip)));
+ } while ((ref + MAX_DISTANCE < ip) || (A32(ref) != A32(ip)));
// Catch up
- while ((ip>anchor) && (ref>(BYTE*)source) && unlikely(ip[-1]==ref[-1])) { ip--; ref--; }
+ while ((ip>anchor) && (ref > lowLimit) && unlikely(ip[-1]==ref[-1])) { ip--; ref--; }
// Encode Literal length
length = (int)(ip - anchor);
token = op++;
- if unlikely(op + length + (2 + 1 + LASTLITERALS) + (length>>8) > oend) return 0; // Check output limit
-#ifdef _MSC_VER
+ if ((limitedOutput) && unlikely(op + length + (2 + 1 + LASTLITERALS) + (length>>8) > oend)) return 0; // Check output limit
if (length>=(int)RUN_MASK)
{
int len = length-RUN_MASK;
*token=(RUN_MASK<<ML_BITS);
- if (len>254)
- {
- do { *op++ = 255; len -= 255; } while (len>254);
- *op++ = (BYTE)len;
- memcpy(op, anchor, length);
- op += length;
- goto _next_match;
- }
- else
+ for(; len >= 255 ; len-=255) *op++ = 255;
*op++ = (BYTE)len;
}
- else *token = (length<<ML_BITS);
-#else
- if (length>=(int)RUN_MASK)
- {
- int len;
- *token=(RUN_MASK<<ML_BITS);
- len = length-RUN_MASK;
- for(; len > 254 ; len-=255) *op++ = 255;
- *op++ = (BYTE)len;
- }
- else *token = (length<<ML_BITS);
-#endif
+ else *token = (BYTE)(length<<ML_BITS);
// Copy Literals
- LZ4_BLINDCOPY(anchor, op, length);
+ { BYTE* end=(op)+(length); LZ4_WILDCOPY(op,anchor,end); op=end; }
_next_match:
// Encode Offset
@@ -452,7 +491,7 @@ _next_match: anchor = ip;
while likely(ip<matchlimit-(STEPSIZE-1))
{
- UARCH diff = AARCH(ref) ^ AARCH(ip);
+ size_t diff = AARCH(ref) ^ AARCH(ip);
if (!diff) { ip+=STEPSIZE; ref+=STEPSIZE; continue; }
ip += LZ4_NbCommonBytes(diff);
goto _endCount;
@@ -464,40 +503,40 @@ _endCount: // Encode MatchLength
length = (int)(ip - anchor);
- if unlikely(op + (1 + LASTLITERALS) + (length>>8) > oend) return 0; // Check output limit
+ if ((limitedOutput) && unlikely(op + (1 + LASTLITERALS) + (length>>8) > oend)) return 0; // Check output limit
if (length>=(int)ML_MASK)
{
*token += ML_MASK;
length -= ML_MASK;
for (; length > 509 ; length-=510) { *op++ = 255; *op++ = 255; }
- if (length > 254) { length-=255; *op++ = 255; }
+ if (length >= 255) { length-=255; *op++ = 255; }
*op++ = (BYTE)length;
}
- else *token += length;
+ else *token += (BYTE)(length);
// Test end of chunk
if (ip > mflimit) { anchor = ip; break; }
// Fill table
- HashTable[LZ4_HASH_VALUE(ip-2)] = ip - 2 - base;
+ LZ4_putPosition(ip-2, ctx, tableType, 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; }
+ ref = LZ4_getPosition(ip, ctx, tableType, base);
+ LZ4_putPosition(ip, ctx, tableType, base);
+ if ((ref + MAX_DISTANCE >= ip) && (A32(ref) == A32(ip))) { token = op++; *token=0; goto _next_match; }
// Prepare next loop
anchor = ip++;
- forwardH = LZ4_HASH_VALUE(ip);
+ forwardH = LZ4_hashPosition(ip, tableType);
}
_last_literals:
// Encode Last Literals
{
int lastRun = (int)(iend - anchor);
- if (((char*)op - dest) + lastRun + 1 + ((lastRun+255-RUN_MASK)/255) > (U32)maxOutputSize) return 0;
- if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK<<ML_BITS); lastRun-=RUN_MASK; for(; lastRun > 254 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; }
- else *op++ = (lastRun<<ML_BITS);
+ if ((limitedOutput) && (((char*)op - dest) + lastRun + 1 + ((lastRun+255-RUN_MASK)/255) > (U32)maxOutputSize)) return 0; // Check output limit
+ if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK<<ML_BITS); lastRun-=RUN_MASK; for(; lastRun >= 255 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; }
+ else *op++ = (BYTE)(lastRun<<ML_BITS);
memcpy(op, anchor, iend - anchor);
op += iend-anchor;
}
@@ -507,372 +546,217 @@ _last_literals: }
-
-// Note : this function is valid only if isize < LZ4_64KLIMIT
-#define LZ4_64KLIMIT ((1<<16) + (MFLIMIT-1))
-#define HASHLOG64K (HASH_LOG+1)
-#define HASH64KTABLESIZE (1U<<HASHLOG64K)
-#define LZ4_HASH64K_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8)-HASHLOG64K))
-#define LZ4_HASH64K_VALUE(p) LZ4_HASH64K_FUNCTION(A32(p))
-static inline int LZ4_compress64kCtx(void** ctx,
- const char* source,
- char* dest,
- int isize,
- int maxOutputSize)
+int LZ4_compress(const char* source, char* dest, int inputSize)
{
-#if HEAPMODE
- struct refTables *srt = (struct refTables *) (*ctx);
- U16* HashTable;
+#if (HEAPMODE)
+ void* ctx = ALLOCATOR(HASHNBCELLS4, 4); // Aligned on 4-bytes boundaries
#else
- U16 HashTable[HASH64KTABLESIZE] = {0};
+ U32 ctx[1U<<(MEMORY_USAGE-2)] = {0}; // Ensure data is aligned on 4-bytes boundaries
#endif
+ int result;
- 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;
- BYTE* const oend = op + maxOutputSize;
-
- int len, length;
- const int skipStrength = SKIPSTRENGTH;
- U32 forwardH;
-
+ if (inputSize < (int)LZ4_64KLIMIT)
+ result = LZ4_compress_generic((void*)ctx, source, dest, inputSize, 0, notLimited, byU16, noPrefix);
+ else
+ result = LZ4_compress_generic((void*)ctx, source, dest, inputSize, 0, notLimited, (sizeof(void*)==8) ? byU32 : byPtr, noPrefix);
- // Init
- if (isize<MINLENGTH) goto _last_literals;
-#if HEAPMODE
- if (*ctx == NULL)
- {
- srt = (struct refTables *) malloc ( sizeof(struct refTables) );
- *ctx = (void*) srt;
- }
- HashTable = (U16*)(srt->hashTable);
- memset((void*)HashTable, 0, sizeof(srt->hashTable));
-#else
- (void) ctx;
+#if (HEAPMODE)
+ FREEMEM(ctx);
#endif
+ return result;
+}
+int LZ4_compress_continue (void* LZ4_Data, const char* source, char* dest, int inputSize)
+{
+ return LZ4_compress_generic(LZ4_Data, source, dest, inputSize, 0, notLimited, byU32, withPrefix);
+}
- // 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] = (U16)(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 = (int)(ip - anchor);
- token = op++;
- if unlikely(op + length + (2 + 1 + LASTLITERALS) + (length>>8) > oend) return 0; // Check output limit
-#ifdef _MSC_VER
- if (length>=(int)RUN_MASK)
- {
- int len = length-RUN_MASK;
- *token=(RUN_MASK<<ML_BITS);
- if (len>254)
- {
- do { *op++ = 255; len -= 255; } while (len>254);
- *op++ = (BYTE)len;
- memcpy(op, anchor, length);
- op += length;
- goto _next_match;
- }
- else
- *op++ = (BYTE)len;
- }
- else *token = (length<<ML_BITS);
+int LZ4_compress_limitedOutput(const char* source, char* dest, int inputSize, int maxOutputSize)
+{
+#if (HEAPMODE)
+ void* ctx = ALLOCATOR(HASHNBCELLS4, 4); // Aligned on 4-bytes boundaries
#else
- if (length>=(int)RUN_MASK) { *token=(RUN_MASK<<ML_BITS); len = length-RUN_MASK; for(; len > 254 ; len-=255) *op++ = 255; *op++ = (BYTE)len; }
- else *token = (length<<ML_BITS);
+ U32 ctx[1U<<(MEMORY_USAGE-2)] = {0}; // Ensure data is aligned on 4-bytes boundaries
#endif
+ int result;
- // Copy Literals
- LZ4_BLINDCOPY(anchor, op, length);
-
-_next_match:
- // Encode Offset
- LZ4_WRITE_LITTLEENDIAN_16(op,(U16)(ip-ref));
-
- // Start Counting
- ip+=MINMATCH; ref+=MINMATCH; // MinMatch verified
- anchor = ip;
- while (ip<matchlimit-(STEPSIZE-1))
- {
- UARCH diff = AARCH(ref) ^ AARCH(ip);
- if (!diff) { ip+=STEPSIZE; ref+=STEPSIZE; continue; }
- ip += LZ4_NbCommonBytes(diff);
- goto _endCount;
- }
- if (LZ4_ARCH64) if ((ip<(matchlimit-3)) && (A32(ref) == A32(ip))) { ip+=4; ref+=4; }
- if ((ip<(matchlimit-1)) && (A16(ref) == A16(ip))) { ip+=2; ref+=2; }
- if ((ip<matchlimit) && (*ref == *ip)) ip++;
-_endCount:
-
- // Encode MatchLength
- len = (int)(ip - anchor);
- if unlikely(op + (1 + LASTLITERALS) + (len>>8) > oend) return 0; // Check output limit
- if (len>=(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)] = (U16)(ip - 2 - base);
-
- // Test next position
- ref = base + HashTable[LZ4_HASH64K_VALUE(ip)];
- HashTable[LZ4_HASH64K_VALUE(ip)] = (U16)(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 = (int)(iend - anchor);
- if (op + lastRun + 1 + (lastRun-RUN_MASK+255)/255 > oend) return 0;
- if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK<<ML_BITS); lastRun-=RUN_MASK; for(; lastRun > 254 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; }
- else *op++ = (lastRun<<ML_BITS);
- memcpy(op, anchor, iend - anchor);
- op += iend-anchor;
- }
-
- // End
- return (int) (((char*)op)-dest);
-}
-
+ if (inputSize < (int)LZ4_64KLIMIT)
+ result = LZ4_compress_generic((void*)ctx, source, dest, inputSize, maxOutputSize, limited, byU16, noPrefix);
+ else
+ result = LZ4_compress_generic((void*)ctx, source, dest, inputSize, maxOutputSize, limited, (sizeof(void*)==8) ? byU32 : byPtr, noPrefix);
-int LZ4_compress_limitedOutput(const char* source,
- char* dest,
- int isize,
- int maxOutputSize)
-{
-#if HEAPMODE
- void* ctx = malloc(sizeof(struct refTables));
- int result;
- if (isize < LZ4_64KLIMIT)
- result = LZ4_compress64kCtx(&ctx, source, dest, isize, maxOutputSize);
- else result = LZ4_compressCtx(&ctx, source, dest, isize, maxOutputSize);
- free(ctx);
- return result;
-#else
- if (isize < (int)LZ4_64KLIMIT) return LZ4_compress64kCtx(NULL, source, dest, isize, maxOutputSize);
- return LZ4_compressCtx(NULL, source, dest, isize, maxOutputSize);
+#if (HEAPMODE)
+ FREEMEM(ctx);
#endif
+ return result;
}
-
-int LZ4_compress(const char* source,
- char* dest,
- int isize)
+int LZ4_compress_limitedOutput_continue (void* LZ4_Data, const char* source, char* dest, int inputSize, int maxOutputSize)
{
- return LZ4_compress_limitedOutput(source, dest, isize, LZ4_compressBound(isize));
+ return LZ4_compress_generic(LZ4_Data, source, dest, inputSize, maxOutputSize, limited, byU32, withPrefix);
}
-
-
//****************************
-// Decompression functions
+// Stream functions
//****************************
-// Note : The decoding functions LZ4_uncompress() and LZ4_uncompress_unknownOutputSize()
-// are safe against "buffer overflow" attack type.
-// They will never write nor read outside of the provided output buffers.
-// LZ4_uncompress_unknownOutputSize() also insures that it will never read outside of the input buffer.
-// A corrupted input will produce an error result, a negative int, indicating the position of the error within input stream.
-
-int LZ4_uncompress(const char* source,
- char* dest,
- int osize)
+FORCE_INLINE void LZ4_init(LZ4_Data_Structure* lz4ds, const BYTE* base)
{
- // Local Variables
- const BYTE* restrict ip = (const BYTE*) source;
- const BYTE* ref;
-
- BYTE* op = (BYTE*) dest;
- BYTE* const oend = op + osize;
- BYTE* cpy;
-
- unsigned token;
-
- size_t dec32table[] = {0, 3, 2, 3, 0, 0, 0, 0};
-#if LZ4_ARCH64
- size_t dec64table[] = {0, 0, 0, -1, 0, 1, 2, 3};
-#endif
+ MEM_INIT(lz4ds->hashTable, 0, sizeof(lz4ds->hashTable));
+ lz4ds->bufferStart = base;
+ lz4ds->base = base;
+ lz4ds->nextBlock = base;
+}
- // Main Loop
- while (1)
- {
- size_t length;
+void* LZ4_create (const char* inputBuffer)
+{
+ void* lz4ds = ALLOCATOR(1, sizeof(LZ4_Data_Structure));
+ LZ4_init ((LZ4_Data_Structure*)lz4ds, (const BYTE*)inputBuffer);
+ return lz4ds;
+}
- // get runlength
- token = *ip++;
- if ((length=(token>>ML_BITS)) == RUN_MASK) { size_t len; for (;(len=*ip++)==255;length+=255){} length += len; }
- // copy literals
- cpy = op+length;
- if (cpy>oend-COPYLENGTH)
- {
- if (cpy != oend) goto _output_error; // Error : not enough place for another match (min 4) + 5 literals
- memcpy(op, ip, length);
- ip += length;
- break; // EOF
- }
- LZ4_WILDCOPY(ip, op, cpy); ip -= (op-cpy); op = cpy;
+int LZ4_free (void* LZ4_Data)
+{
+ FREEMEM(LZ4_Data);
+ return (0);
+}
- // get offset
- LZ4_READ_LITTLEENDIAN_16(ref,cpy,ip); ip+=2;
- if unlikely(ref < (BYTE* const)dest) goto _output_error; // Error : offset outside destination buffer
- // get matchlength
- if ((length=(token&ML_MASK)) == ML_MASK) { for (;*ip==255;length+=255) {ip++;} length += *ip++; }
+char* LZ4_slideInputBuffer (void* LZ4_Data)
+{
+ LZ4_Data_Structure* lz4ds = (LZ4_Data_Structure*)LZ4_Data;
+ size_t delta = lz4ds->nextBlock - (lz4ds->bufferStart + 64 KB);
- // copy repeated sequence
- if unlikely((op-ref)<STEPSIZE)
- {
-#if LZ4_ARCH64
- size_t dec64 = dec64table[op-ref];
-#else
- const int dec64 = 0;
-#endif
- op[0] = ref[0];
- op[1] = ref[1];
- op[2] = ref[2];
- op[3] = ref[3];
- op += 4, ref += 4; ref -= dec32table[op-ref];
- A32(op) = A32(ref);
- op += STEPSIZE-4; ref -= dec64;
- } else { LZ4_COPYSTEP(ref,op); }
- cpy = op + length - (STEPSIZE-4);
+ if ( (lz4ds->base - delta > lz4ds->base) // underflow control
+ || ((size_t)(lz4ds->nextBlock - lz4ds->base) > 0xE0000000) ) // close to 32-bits limit
+ {
+ size_t deltaLimit = (lz4ds->nextBlock - 64 KB) - lz4ds->base;
+ int nH;
- if unlikely(cpy>oend-(COPYLENGTH)-(STEPSIZE-4))
+ for (nH=0; nH < HASHNBCELLS4; nH++)
{
- if (cpy > oend-LASTLITERALS) goto _output_error; // Error : last 5 bytes must be literals
- LZ4_SECURECOPY(ref, op, (oend-COPYLENGTH));
- while(op<cpy) *op++=*ref++;
- op=cpy;
- continue;
+ if ((size_t)(lz4ds->hashTable[nH]) < deltaLimit) lz4ds->hashTable[nH] = 0;
+ else lz4ds->hashTable[nH] -= (U32)deltaLimit;
}
-
- LZ4_WILDCOPY(ref, op, cpy);
- op=cpy; // correction
+ memcpy((void*)(lz4ds->bufferStart), (const void*)(lz4ds->nextBlock - 64 KB), 64 KB);
+ lz4ds->base = lz4ds->bufferStart;
+ lz4ds->nextBlock = lz4ds->base + 64 KB;
}
+ else
+ {
+ memcpy((void*)(lz4ds->bufferStart), (const void*)(lz4ds->nextBlock - 64 KB), 64 KB);
+ lz4ds->nextBlock -= delta;
+ lz4ds->base -= delta;
+ }
- // end of decoding
- return (int) (((char*)ip)-source);
-
- // write overflow error detected
-_output_error:
- return (int) (-(((char*)ip)-source));
+ return (char*)(lz4ds->nextBlock);
}
-int LZ4_uncompress_unknownOutputSize(
- const char* source,
- char* dest,
- int isize,
- int maxOutputSize)
+//****************************
+// Decompression functions
+//****************************
+
+// This generic decompression function cover all use cases.
+// It shall be instanciated several times, using different sets of directives
+// Note that it is essential this generic function is really inlined,
+// in order to remove useless branches during compilation optimisation.
+FORCE_INLINE int LZ4_decompress_generic(
+ const char* source,
+ char* dest,
+ int inputSize, //
+ int outputSize, // If endOnInput==endOnInputSize, this value is the max size of Output Buffer.
+
+ int endOnInput, // endOnOutputSize, endOnInputSize
+ int prefix64k, // noPrefix, withPrefix
+ int partialDecoding, // full, partial
+ int targetOutputSize // only used if partialDecoding==partial
+ )
{
// Local Variables
const BYTE* restrict ip = (const BYTE*) source;
- const BYTE* const iend = ip + isize;
const BYTE* ref;
+ const BYTE* const iend = ip + inputSize;
BYTE* op = (BYTE*) dest;
- BYTE* const oend = op + maxOutputSize;
+ BYTE* const oend = op + outputSize;
BYTE* cpy;
+ BYTE* oexit = op + targetOutputSize;
- size_t dec32table[] = {0, 3, 2, 3, 0, 0, 0, 0};
-#if LZ4_ARCH64
- size_t dec64table[] = {0, 0, 0, -1, 0, 1, 2, 3};
-#endif
+ const size_t dec32table[] = {0, 3, 2, 3, 0, 0, 0, 0}; // static reduces speed for LZ4_decompress_safe() on GCC64
+ static const size_t dec64table[] = {0, 0, 0, (size_t)-1, 0, 1, 2, 3};
- // Special case
- if unlikely(ip==iend) goto _output_error; // A correctly formed null-compressed LZ4 must have at least one byte (token=0)
+ // Special cases
+ if ((partialDecoding) && (oexit> oend-MFLIMIT)) oexit = oend-MFLIMIT; // targetOutputSize too high => decode everything
+ if ((endOnInput) && unlikely(outputSize==0)) return ((inputSize==1) && (*ip==0)) ? 0 : -1; // Empty output buffer
+ if ((!endOnInput) && unlikely(outputSize==0)) return (*ip==0?1:-1);
+
// Main Loop
- while (1)
+ while (1)
{
unsigned token;
size_t length;
// get runlength
token = *ip++;
- if ((length=(token>>ML_BITS)) == RUN_MASK)
+ if ((length=(token>>ML_BITS)) == RUN_MASK)
{
- int s=255;
- while (likely(ip<iend) && (s==255)) { s=*ip++; length += s; }
+ unsigned s=255;
+ while (((endOnInput)?ip<iend:1) && (s==255))
+ {
+ s = *ip++;
+ length += s;
+ }
}
// copy literals
cpy = op+length;
- if ((cpy>oend-MFLIMIT) || (ip+length>iend-(2+1+LASTLITERALS)))
+ if (((endOnInput) && ((cpy>(partialDecoding?oexit:oend-MFLIMIT)) || (ip+length>iend-(2+1+LASTLITERALS))) )
+ || ((!endOnInput) && (cpy>oend-COPYLENGTH)))
{
- if (cpy > oend) goto _output_error; // Error : writes beyond output buffer
- if (ip+length != iend) goto _output_error; // Error : LZ4 format requires to consume all input at this stage (no match within the last 11 bytes, and at least 8 remaining input bytes for another match+literals)
+ if (partialDecoding)
+ {
+ if (cpy > oend) goto _output_error; // Error : write attempt beyond end of output buffer
+ if ((endOnInput) && (ip+length > iend)) goto _output_error; // Error : read attempt beyond end of input buffer
+ }
+ else
+ {
+ if ((!endOnInput) && (cpy != oend)) goto _output_error; // Error : block decoding must stop exactly there
+ if ((endOnInput) && ((ip+length != iend) || (cpy > oend))) goto _output_error; // Error : input must be consumed
+ }
memcpy(op, ip, length);
+ ip += length;
op += length;
break; // Necessarily EOF, due to parsing restrictions
}
- LZ4_WILDCOPY(ip, op, cpy); ip -= (op-cpy); op = cpy;
+ LZ4_WILDCOPY(op, ip, cpy); ip -= (op-cpy); op = cpy;
// get offset
LZ4_READ_LITTLEENDIAN_16(ref,cpy,ip); ip+=2;
- if unlikely(ref < (BYTE* const)dest) goto _output_error; // Error : offset outside of destination buffer
+ if ((prefix64k==noPrefix) && unlikely(ref < (BYTE* const)dest)) goto _output_error; // Error : offset outside destination buffer
// get matchlength
- if ((length=(token&ML_MASK)) == ML_MASK)
+ if ((length=(token&ML_MASK)) == ML_MASK)
{
- while likely(ip<iend-(LASTLITERALS+1)) // Error : a minimum input bytes must remain for LASTLITERALS + token
- {
- int s = *ip++;
- length +=s;
+ while ((!endOnInput) || (ip<iend-(LASTLITERALS+1))) // Ensure enough bytes remain for LASTLITERALS + token
+ {
+ unsigned s = *ip++;
+ length += s;
if (s==255) continue;
break;
- }
+ }
}
// copy repeated sequence
- if unlikely(op-ref<STEPSIZE)
+ if unlikely((op-ref)<(int)STEPSIZE)
{
-#if LZ4_ARCH64
- size_t dec64 = dec64table[op-ref];
-#else
- const int dec64 = 0;
-#endif
+ const size_t dec64 = dec64table[(sizeof(void*)==4) ? 0 : op-ref];
op[0] = ref[0];
op[1] = ref[1];
op[2] = ref[2];
@@ -880,27 +764,59 @@ int LZ4_uncompress_unknownOutputSize( op += 4, ref += 4; ref -= dec32table[op-ref];
A32(op) = A32(ref);
op += STEPSIZE-4; ref -= dec64;
- } else { LZ4_COPYSTEP(ref,op); }
+ } else { LZ4_COPYSTEP(op,ref); }
cpy = op + length - (STEPSIZE-4);
- if unlikely(cpy>oend-(COPYLENGTH+(STEPSIZE-4)))
+ if unlikely(cpy>oend-COPYLENGTH-(STEPSIZE-4))
{
if (cpy > oend-LASTLITERALS) goto _output_error; // Error : last 5 bytes must be literals
- LZ4_SECURECOPY(ref, op, (oend-COPYLENGTH));
+ LZ4_SECURECOPY(op, ref, (oend-COPYLENGTH));
while(op<cpy) *op++=*ref++;
op=cpy;
continue;
}
-
- LZ4_WILDCOPY(ref, op, cpy);
- op=cpy; // correction
+ LZ4_WILDCOPY(op, ref, cpy);
+ op=cpy; // correction
}
// end of decoding
- return (int) (((char*)op)-dest);
+ if (endOnInput)
+ return (int) (((char*)op)-dest); // Nb of output bytes decoded
+ else
+ return (int) (((char*)ip)-source); // Nb of input bytes read
- // write overflow error detected
+ // Overflow error detected
_output_error:
- return (int) (-(((char*)ip)-source));
+ return (int) (-(((char*)ip)-source))-1;
+}
+
+
+int LZ4_decompress_safe(const char* source, char* dest, int inputSize, int maxOutputSize)
+{
+ return LZ4_decompress_generic(source, dest, inputSize, maxOutputSize, endOnInputSize, noPrefix, full, 0);
+}
+
+int LZ4_decompress_safe_withPrefix64k(const char* source, char* dest, int inputSize, int maxOutputSize)
+{
+ return LZ4_decompress_generic(source, dest, inputSize, maxOutputSize, endOnInputSize, withPrefix, full, 0);
+}
+
+int LZ4_decompress_safe_partial(const char* source, char* dest, int inputSize, int targetOutputSize, int maxOutputSize)
+{
+ return LZ4_decompress_generic(source, dest, inputSize, maxOutputSize, endOnInputSize, noPrefix, partial, targetOutputSize);
+}
+
+int LZ4_decompress_fast_withPrefix64k(const char* source, char* dest, int outputSize)
+{
+ return LZ4_decompress_generic(source, dest, 0, outputSize, endOnOutputSize, withPrefix, full, 0);
+}
+
+int LZ4_decompress_fast(const char* source, char* dest, int outputSize)
+{
+#ifdef _MSC_VER // This version is faster with Visual
+ return LZ4_decompress_generic(source, dest, 0, outputSize, endOnOutputSize, noPrefix, full, 0);
+#else
+ return LZ4_decompress_generic(source, dest, 0, outputSize, endOnOutputSize, withPrefix, full, 0);
+#endif
}
diff --git a/ext/lz4/lz4.h b/ext/lz4/lz4.h index 4897eb28..9ef58862 100644 --- a/ext/lz4/lz4.h +++ b/ext/lz4/lz4.h @@ -1,7 +1,7 @@ /*
LZ4 - Fast LZ compression algorithm
Header File
- Copyright (C) 2011-2012, Yann Collet.
+ Copyright (C) 2011-2013, 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
@@ -42,7 +42,7 @@ extern "C" { // Compiler Options
//**************************************
#if defined(_MSC_VER) && !defined(__cplusplus) // Visual Studio
-# define inline __inline // Visual is not C99, but supports some kind of inline
+# define inline __inline // Visual C is not C99, but supports some kind of inline
#endif
@@ -50,79 +50,156 @@ extern "C" { // Simple Functions
//****************************
-int LZ4_compress (const char* source, char* dest, int isize);
-int LZ4_uncompress (const char* source, char* dest, int osize);
+int LZ4_compress (const char* source, char* dest, int inputSize);
+int LZ4_decompress_safe (const char* source, char* dest, int inputSize, int maxOutputSize);
/*
LZ4_compress() :
- Compresses 'isize' bytes from 'source' into 'dest'.
+ Compresses 'inputSize' bytes from 'source' into 'dest'.
Destination buffer must be already allocated,
and must be sized to handle worst cases situations (input data not compressible)
Worst case size evaluation is provided by function LZ4_compressBound()
-
- isize : is the input size. Max supported value is ~1.9GB
+ inputSize : Max supported value is LZ4_MAX_INPUT_VALUE
return : the number of bytes written in buffer dest
+ or 0 if the compression fails
-
-LZ4_uncompress() :
- osize : is the output size, therefore the original size
- return : the number of bytes read in the source buffer
- If the source stream is malformed, the function will stop decoding and return a negative result, indicating the byte position of the faulty instruction
- This function never writes outside of provided buffers, and never modifies input buffer.
- note : destination buffer must be already allocated.
- its size must be a minimum of 'osize' bytes.
+LZ4_decompress_safe() :
+ maxOutputSize : is the size of the destination buffer (which must be already allocated)
+ return : the number of bytes decoded in the destination buffer (necessarily <= maxOutputSize)
+ If the source stream is detected malformed, the function will stop decoding and return a negative result.
+ This function is protected against buffer overflow exploits (never writes outside of output buffer, and never reads outside of input buffer). Therefore, it is protected against malicious data packets
*/
//****************************
// Advanced Functions
//****************************
-
-static inline int LZ4_compressBound(int isize) { return ((isize) + ((isize)/255) + 16); }
-#define LZ4_COMPRESSBOUND( isize) ((isize) + ((isize)/255) + 16)
+#define LZ4_MAX_INPUT_SIZE 0x7E000000 // 2 113 929 216 bytes
+#define LZ4_COMPRESSBOUND(isize) ((unsigned int)(isize) > (unsigned int)LZ4_MAX_INPUT_SIZE ? 0 : (isize) + ((isize)/255) + 16)
+static inline int LZ4_compressBound(int isize) { return LZ4_COMPRESSBOUND(isize); }
/*
LZ4_compressBound() :
Provides the maximum size that LZ4 may output in a "worst case" scenario (input data not compressible)
primarily useful for memory allocation of output buffer.
- inline function is recommended for the general case,
- but macro is also provided when results need to be evaluated at compile time (such as table size allocation).
+ inline function is recommended for the general case,
+ macro is also provided when result needs to be evaluated at compilation (such as stack memory allocation).
- isize : is the input size. Max supported value is ~1.9GB
+ isize : is the input size. Max supported value is LZ4_MAX_INPUT_SIZE
return : maximum output size in a "worst case" scenario
- note : this function is limited by "int" range (2^31-1)
+ or 0, if input size is too large ( > LZ4_MAX_INPUT_SIZE)
*/
-int LZ4_compress_limitedOutput (const char* source, char* dest, int isize, int maxOutputSize);
+int LZ4_compress_limitedOutput (const char* source, char* dest, int inputSize, int maxOutputSize);
/*
LZ4_compress_limitedOutput() :
- Compress 'isize' bytes from 'source' into an output buffer 'dest' of maximum size 'maxOutputSize'.
+ Compress 'inputSize' bytes from 'source' into an output buffer 'dest' of maximum size 'maxOutputSize'.
If it cannot achieve it, compression will stop, and result of the function will be zero.
This function never writes outside of provided output buffer.
- isize : is the input size. Max supported value is ~1.9GB
+ inputSize : Max supported value is LZ4_MAX_INPUT_VALUE
maxOutputSize : is the size of the destination buffer (which must be already allocated)
return : the number of bytes written in buffer 'dest'
or 0 if the compression fails
*/
-int LZ4_uncompress_unknownOutputSize (const char* source, char* dest, int isize, int maxOutputSize);
+int LZ4_decompress_fast (const char* source, char* dest, int outputSize);
/*
-LZ4_uncompress_unknownOutputSize() :
- isize : is the input size, therefore the compressed size
- maxOutputSize : is the size of the destination buffer (which must be already allocated)
+LZ4_decompress_fast() :
+ outputSize : is the original (uncompressed) size
+ return : the number of bytes read from the source buffer (in other words, the compressed size)
+ If the source stream is malformed, the function will stop decoding and return a negative result.
+ note : This function is a bit faster than LZ4_decompress_safe()
+ This function never writes outside of output buffers, but may read beyond input buffer in case of malicious data packet.
+ Use this function preferably into a trusted environment (data to decode comes from a trusted source).
+ Destination buffer must be already allocated. Its size must be a minimum of 'outputSize' bytes.
+*/
+
+int LZ4_decompress_safe_partial (const char* source, char* dest, int inputSize, int targetOutputSize, int maxOutputSize);
+
+/*
+LZ4_decompress_safe_partial() :
+ This function decompress a compressed block of size 'inputSize' at position 'source'
+ into output buffer 'dest' of size 'maxOutputSize'.
+ The function tries to stop decompressing operation as soon as 'targetOutputSize' has been reached,
+ reducing decompression time.
return : the number of bytes decoded in the destination buffer (necessarily <= maxOutputSize)
- If the source stream is malformed, the function will stop decoding and return a negative result, indicating the byte position of the faulty instruction
- This function never writes beyond dest + maxOutputSize, and is therefore protected against malicious data packets
- note : Destination buffer must be already allocated.
- This version is slightly slower than LZ4_uncompress()
+ Note : this number can be < 'targetOutputSize' should the compressed block to decode be smaller.
+ Always control how many bytes were decoded.
+ If the source stream is detected malformed, the function will stop decoding and return a negative result.
+ This function never writes outside of output buffer, and never reads outside of input buffer. It is therefore protected against malicious data packets
*/
+//****************************
+// Stream Functions
+//****************************
+
+void* LZ4_create (const char* inputBuffer);
+int LZ4_compress_continue (void* LZ4_Data, const char* source, char* dest, int inputSize);
+int LZ4_compress_limitedOutput_continue (void* LZ4_Data, const char* source, char* dest, int inputSize, int maxOutputSize);
+char* LZ4_slideInputBuffer (void* LZ4_Data);
+int LZ4_free (void* LZ4_Data);
+
+/*
+These functions allow the compression of dependent blocks, where each block benefits from prior 64 KB within preceding blocks.
+In order to achieve this, it is necessary to start creating the LZ4 Data Structure, thanks to the function :
+
+void* LZ4_create (const char* inputBuffer);
+The result of the function is the (void*) pointer on the LZ4 Data Structure.
+This pointer will be needed in all other functions.
+If the pointer returned is NULL, then the allocation has failed, and compression must be aborted.
+The only parameter 'const char* inputBuffer' must, obviously, point at the beginning of input buffer.
+The input buffer must be already allocated, and size at least 192KB.
+'inputBuffer' will also be the 'const char* source' of the first block.
+
+All blocks are expected to lay next to each other within the input buffer, starting from 'inputBuffer'.
+To compress each block, use either LZ4_compress_continue() or LZ4_compress_limitedOutput_continue().
+Their behavior are identical to LZ4_compress() or LZ4_compress_limitedOutput(),
+but require the LZ4 Data Structure as their first argument, and check that each block starts right after the previous one.
+If next block does not begin immediately after the previous one, the compression will fail (return 0).
+
+When it's no longer possible to lay the next block after the previous one (not enough space left into input buffer), a call to :
+char* LZ4_slideInputBuffer(void* LZ4_Data);
+must be performed. It will typically copy the latest 64KB of input at the beginning of input buffer.
+Note that, for this function to work properly, minimum size of an input buffer must be 192KB.
+==> The memory position where the next input data block must start is provided as the result of the function.
+
+Compression can then resume, using LZ4_compress_continue() or LZ4_compress_limitedOutput_continue(), as usual.
+
+When compression is completed, a call to LZ4_free() will release the memory used by the LZ4 Data Structure.
+*/
+
+
+int LZ4_decompress_safe_withPrefix64k (const char* source, char* dest, int inputSize, int maxOutputSize);
+int LZ4_decompress_fast_withPrefix64k (const char* source, char* dest, int outputSize);
+
+/*
+*_withPrefix64k() :
+ These decoding functions work the same as their "normal name" versions,
+ but can use up to 64KB of data in front of 'char* dest'.
+ These functions are necessary to decode inter-dependant blocks.
+*/
+
+
+//****************************
+// Obsolete Functions
+//****************************
+
+static inline int LZ4_uncompress (const char* source, char* dest, int outputSize) { return LZ4_decompress_fast(source, dest, outputSize); }
+static inline int LZ4_uncompress_unknownOutputSize (const char* source, char* dest, int isize, int maxOutputSize) { return LZ4_decompress_safe(source, dest, isize, maxOutputSize); }
+
+/*
+These functions are deprecated and should no longer be used.
+They are provided here for compatibility with existing user programs.
+*/
+
+
+
#if defined (__cplusplus)
}
#endif
diff --git a/ext/lz4/lz4_format_description.txt b/ext/lz4/lz4_format_description.txt deleted file mode 100644 index a170ddef..00000000 --- a/ext/lz4/lz4_format_description.txt +++ /dev/null @@ -1,121 +0,0 @@ -LZ4 Format Description
-Last revised: 2012-02-27
-Author : Y. Collet
-
-
-
-This small specification intents to provide enough information
-to anyone willing to produce LZ4-compatible compressed streams
-using any programming language.
-
-LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
-The most important design principle behind LZ4 is simplicity.
-It helps to create an easy to read and maintain source code.
-It also helps later on for optimisations, compactness, and speed.
-There is no entropy encoder backend nor framing layer.
-The latter is assumed to be handled by other parts of the system.
-
-This document only describes the format,
-not how the LZ4 compressor nor decompressor actually work.
-The correctness of the decompressor should not depend
-on implementation details of the compressor, and vice versa.
-
-
-
--- Compressed stream format --
-
-An LZ4 compressed stream is composed of sequences.
-Schematically, a sequence is a suite of literals, followed by a match copy.
-
-Each sequence starts with a token.
-The token is a one byte value, separated into two 4-bits fields.
-Therefore each field ranges from 0 to 15.
-
-
-The first field uses the 4 high-bits of the token.
-It provides the length of literals to follow.
-(Note : a literal is a not-compressed byte).
-If the field value is 0, then there is no literal.
-If it is 15, then we need to add some more bytes to indicate the full length.
-Each additionnal byte then represent a value from 0 to 255,
-which is added to the previous value to produce a total length.
-When the byte value is 255, another byte is output.
-There can be any number of bytes following the token. There is no "size limit".
-(Sidenote this is why a not-compressible input stream is expanded by 0.4%).
-
-Example 1 : A length of 48 will be represented as :
-- 15 : value for the 4-bits High field
-- 33 : (=48-15) remaining length to reach 48
-
-Example 2 : A length of 280 will be represented as :
-- 15 : value for the 4-bits High field
-- 255 : following byte is maxed, since 280-15 >= 255
-- 10 : (=280 - 15 - 255) ) remaining length to reach 280
-
-Example 3 : A length of 15 will be represented as :
-- 15 : value for the 4-bits High field
-- 0 : (=15-15) yes, the zero must be output
-
-Following the token and optional length bytes, are the literals themselves.
-They are exactly as numerous as previously decoded (length of literals).
-It's possible that there are zero literal.
-
-
-Following the literals is the match copy operation.
-
-It starts by the offset.
-This is a 2 bytes value, in little endian format :
-the lower byte is the first one in the stream.
-
-The offset represents the position of the match to be copied from.
-1 means "current position - 1 byte".
-The maximum offset value is 65535, 65536 cannot be coded.
-Note that 0 is an invalid value, not used.
-
-Then we need to extract the match length.
-For this, we use the second token field, the low 4-bits.
-Value, obviously, ranges from 0 to 15.
-However here, 0 means that the copy operation will be minimal.
-The minimum length of a match, called minmatch, is 4.
-As a consequence, a 0 value means 4 bytes, and a value of 15 means 19+ bytes.
-Similar to literal length, on reaching the highest possible value (15),
-we output additional bytes, one at a time, with values ranging from 0 to 255.
-They are added to total to provide the final match length.
-A 255 value means there is another byte to read and add.
-There is no limit to the number of optional bytes that can be output this way.
-(This points towards a maximum achievable compression ratio of ~250).
-
-With the offset and the matchlength,
-the decoder can now proceed to copy the data from the already decoded buffer.
-On decoding the matchlength, we reach the end of the compressed sequence,
-and therefore start another one.
-
-
--- Parsing restrictions --
-
-There are specific parsing rules to respect in order to remain compatible
-with assumptions made by the decoder :
-1) The last 5 bytes are always literals
-2) The last match must start at least 12 bytes before end of stream
-Consequently, a file with less than 13 bytes cannot be compressed.
-These rules are in place to ensure that the decoder
-will never read beyond the input buffer, nor write beyond the output buffer.
-
-Note that the last sequence is also incomplete,
-and stops right after literals.
-
-
--- Additional notes --
-
-There is no assumption nor limits to the way the compressor
-searches and selects matches within the source stream.
-It could be a fast scan, a multi-probe, a full search using BST,
-standard hash chains or MMC, well whatever.
-
-Advanced parsing strategies can also be implemented, such as lazy match,
-or full optimal parsing.
-
-All these trade-off offer distinctive speed/memory/compression advantages.
-Whatever the method used by the compressor, its result will be decodable
-by any LZ4 decoder if it follows the format specification described above.
-
diff --git a/ext/lz4/lz4demo.c b/ext/lz4/lz4demo.c deleted file mode 100644 index 85edcf00..00000000 --- a/ext/lz4/lz4demo.c +++ /dev/null @@ -1,402 +0,0 @@ -/*
- LZ4Demo - Demo CLI program using LZ4 compression
- Copyright (C) Yann Collet 2011-2012
- GPL v2 License
-
- This program is free software; you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation; either version 2 of the License, or
- (at your option) any later version.
-
- This program is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License along
- with this program; if not, write to the Free Software Foundation, Inc.,
- 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
-
- You can contact the author at :
- - LZ4 homepage : http://fastcompression.blogspot.com/p/lz4.html
- - LZ4 source repository : http://code.google.com/p/lz4/
-*/
-/*
- Note : this is *only* a demo program, an example to show how LZ4 can be used.
- It is not considered part of LZ4 compression library.
- The license of LZ4 is BSD.
- The license of the demo program is GPL.
-*/
-
-//**************************************
-// Compiler Options
-//**************************************
-// Disable some Visual warning messages
-#define _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_DEPRECATE // VS2005
-
-
-//****************************
-// Includes
-//****************************
-#include <stdio.h> // fprintf, fopen, fread, _fileno(?)
-#include <stdlib.h> // malloc
-#include <string.h> // strcmp
-#include <time.h> // clock
-#ifdef _WIN32
-#include <io.h> // _setmode
-#include <fcntl.h> // _O_BINARY
-#endif
-#include "lz4.h"
-#include "lz4hc.h"
-#include "bench.h"
-
-
-//**************************************
-// Compiler-specific functions
-//**************************************
-#define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)
-
-#if defined(_MSC_VER) // Visual Studio
-#define swap32 _byteswap_ulong
-#elif GCC_VERSION >= 403
-#define swap32 __builtin_bswap32
-#else
-static inline unsigned int swap32(unsigned int x) {
- return ((x << 24) & 0xff000000 ) |
- ((x << 8) & 0x00ff0000 ) |
- ((x >> 8) & 0x0000ff00 ) |
- ((x >> 24) & 0x000000ff );
- }
-#endif
-
-
-//****************************
-// Constants
-//****************************
-#define COMPRESSOR_NAME "Compression CLI using LZ4 algorithm"
-#define COMPRESSOR_VERSION ""
-#define COMPILED __DATE__
-#define AUTHOR "Yann Collet"
-#define EXTENSION ".lz4"
-#define WELCOME_MESSAGE "*** %s %s, by %s (%s) ***\n", COMPRESSOR_NAME, COMPRESSOR_VERSION, AUTHOR, COMPILED
-
-#define CHUNKSIZE (8<<20) // 8 MB
-#define CACHELINE 64
-#define ARCHIVE_MAGICNUMBER 0x184C2102
-#define ARCHIVE_MAGICNUMBER_SIZE 4
-
-
-//**************************************
-// Architecture Macros
-//**************************************
-static const int one = 1;
-#define CPU_LITTLE_ENDIAN (*(char*)(&one))
-#define CPU_BIG_ENDIAN (!CPU_LITTLE_ENDIAN)
-#define LITTLE_ENDIAN32(i) if (CPU_BIG_ENDIAN) { i = swap32(i); }
-
-
-//**************************************
-// Macros
-//**************************************
-#define DISPLAY(...) fprintf(stderr, __VA_ARGS__)
-
-
-//****************************
-// Functions
-//****************************
-int usage(char* exename)
-{
- DISPLAY( "Usage :\n");
- DISPLAY( " %s [arg] input output\n", exename);
- DISPLAY( "Arguments :\n");
- DISPLAY( " -c0: Fast compression (default) \n");
- DISPLAY( " -c1: High compression \n");
- DISPLAY( " -d : decompression \n");
- DISPLAY( " -b#: Benchmark files, using # compression level\n");
- DISPLAY( " -t : check compressed file \n");
- DISPLAY( " -h : help (this text)\n");
- DISPLAY( "input : can be 'stdin' (pipe) or a filename\n");
- DISPLAY( "output : can be 'stdout'(pipe) or a filename or 'null'\n");
- return 0;
-}
-
-
-int badusage(char* exename)
-{
- DISPLAY("Wrong parameters\n");
- usage(exename);
- return 0;
-}
-
-
-
-int get_fileHandle(char* input_filename, char* output_filename, FILE** pfinput, FILE** pfoutput)
-{
- char stdinmark[] = "stdin";
- char stdoutmark[] = "stdout";
-
- if (!strcmp (input_filename, stdinmark)) {
- DISPLAY( "Using stdin for input\n");
- *pfinput = stdin;
-#ifdef _WIN32 // Need to set stdin/stdout to binary mode specifically for windows
- _setmode( _fileno( stdin ), _O_BINARY );
-#endif
- } else {
- *pfinput = fopen( input_filename, "rb" );
- }
-
- if (!strcmp (output_filename, stdoutmark)) {
- DISPLAY( "Using stdout for output\n");
- *pfoutput = stdout;
-#ifdef _WIN32 // Need to set stdin/stdout to binary mode specifically for windows
- _setmode( _fileno( stdout ), _O_BINARY );
-#endif
- } else {
- *pfoutput = fopen( output_filename, "wb" );
- }
-
- if ( *pfinput==0 ) { DISPLAY( "Pb opening %s\n", input_filename); return 2; }
- if ( *pfoutput==0) { DISPLAY( "Pb opening %s\n", output_filename); return 3; }
-
- return 0;
-}
-
-
-
-int compress_file(char* input_filename, char* output_filename, int compressionlevel)
-{
- int (*compressionFunction)(const char*, char*, int);
- unsigned long long filesize = 0;
- unsigned long long compressedfilesize = ARCHIVE_MAGICNUMBER_SIZE;
- unsigned int u32var;
- char* in_buff;
- char* out_buff;
- FILE* finput;
- FILE* foutput;
- int r;
- int displayLevel = (compressionlevel>0);
- clock_t start, end;
- size_t sizeCheck;
-
-
- // Init
- switch (compressionlevel)
- {
- case 0 : compressionFunction = LZ4_compress; break;
- case 1 : compressionFunction = LZ4_compressHC; break;
- default : compressionFunction = LZ4_compress;
- }
- start = clock();
- r = get_fileHandle(input_filename, output_filename, &finput, &foutput);
- if (r) return r;
-
- // Allocate Memory
- in_buff = (char*)malloc(CHUNKSIZE);
- out_buff = (char*)malloc(LZ4_compressBound(CHUNKSIZE));
- if (!in_buff || !out_buff) { DISPLAY("Allocation error : not enough memory\n"); return 8; }
-
- // Write Archive Header
- u32var = ARCHIVE_MAGICNUMBER;
- LITTLE_ENDIAN32(u32var);
- *(unsigned int*)out_buff = u32var;
- sizeCheck = fwrite(out_buff, 1, ARCHIVE_MAGICNUMBER_SIZE, foutput);
- if (sizeCheck!=ARCHIVE_MAGICNUMBER_SIZE) { DISPLAY("write error\n"); return 10; }
-
- // Main Loop
- while (1)
- {
- int outSize;
- // Read Block
- int inSize = (int) fread(in_buff, (size_t)1, (size_t)CHUNKSIZE, finput);
- if( inSize<=0 ) break;
- filesize += inSize;
- if (displayLevel) DISPLAY("Read : %i MB \r", (int)(filesize>>20));
-
- // Compress Block
- outSize = compressionFunction(in_buff, out_buff+4, inSize);
- compressedfilesize += outSize+4;
- if (displayLevel) DISPLAY("Read : %i MB ==> %.2f%%\r", (int)(filesize>>20), (double)compressedfilesize/filesize*100);
-
- // Write Block
- LITTLE_ENDIAN32(outSize);
- * (unsigned int*) out_buff = outSize;
- LITTLE_ENDIAN32(outSize);
- sizeCheck = fwrite(out_buff, 1, outSize+4, foutput);
- if (sizeCheck!=(size_t)(outSize+4)) { DISPLAY("write error\n"); return 11; }
- }
-
- // Status
- end = clock();
- DISPLAY( "Compressed %llu bytes into %llu bytes ==> %.2f%%\n",
- (unsigned long long) filesize, (unsigned long long) compressedfilesize, (double)compressedfilesize/filesize*100);
- {
- double seconds = (double)(end - start)/CLOCKS_PER_SEC;
- DISPLAY( "Done in %.2f s ==> %.2f MB/s\n", seconds, (double)filesize / seconds / 1024 / 1024);
- }
-
- // Close & Free
- free(in_buff);
- free(out_buff);
- fclose(finput);
- fclose(foutput);
-
- return 0;
-}
-
-
-int decode_file(char* input_filename, char* output_filename)
-{
- unsigned long long filesize = 0;
- char* in_buff;
- char* out_buff;
- size_t uselessRet;
- int sinkint;
- unsigned int chunkSize;
- FILE* finput;
- FILE* foutput;
- clock_t start, end;
- int r;
- size_t sizeCheck;
-
-
- // Init
- start = clock();
- r = get_fileHandle(input_filename, output_filename, &finput, &foutput);
- if (r) return r;
-
- // Allocate Memory
- in_buff = (char*)malloc(LZ4_compressBound(CHUNKSIZE));
- out_buff = (char*)malloc(CHUNKSIZE);
- if (!in_buff || !out_buff) { DISPLAY("Allocation error : not enough memory\n"); return 7; }
-
- // Check Archive Header
- chunkSize = 0;
- uselessRet = fread(&chunkSize, 1, ARCHIVE_MAGICNUMBER_SIZE, finput);
- LITTLE_ENDIAN32(chunkSize);
- if (chunkSize != ARCHIVE_MAGICNUMBER) { DISPLAY("Unrecognized header : file cannot be decoded\n"); return 6; }
-
- // Main Loop
- while (1)
- {
- // Block Size
- uselessRet = fread(&chunkSize, 1, 4, finput);
- if( uselessRet==0 ) break; // Nothing to read : file read is completed
- LITTLE_ENDIAN32(chunkSize);
- if (chunkSize == ARCHIVE_MAGICNUMBER)
- continue; // appended compressed stream
-
- // Read Block
- uselessRet = fread(in_buff, 1, chunkSize, finput);
-
- // Decode Block
- sinkint = LZ4_uncompress_unknownOutputSize(in_buff, out_buff, chunkSize, CHUNKSIZE);
- if (sinkint < 0) { DISPLAY("Decoding Failed ! Corrupted input !\n"); return 9; }
- filesize += sinkint;
-
- // Write Block
- sizeCheck = fwrite(out_buff, 1, sinkint, foutput);
- if (sizeCheck != (size_t)sinkint) { DISPLAY("write error\n"); return 12; }
- }
-
- // Status
- end = clock();
- DISPLAY( "Successfully decoded %llu bytes \n", (unsigned long long)filesize);
- {
- double seconds = (double)(end - start)/CLOCKS_PER_SEC;
- DISPLAY( "Done in %.2f s ==> %.2f MB/s\n", seconds, (double)filesize / seconds / 1024 / 1024);
- }
-
- // Close & Free
- free(in_buff);
- free(out_buff);
- fclose(finput);
- fclose(foutput);
-
- return 0;
-}
-
-
-int main(int argc, char** argv)
-{
- int i,
- cLevel=0,
- decode=0,
- bench=0,
- filenamesStart=2;
- char* exename=argv[0];
- char* input_filename=0;
- char* output_filename=0;
-#ifdef _WIN32
- char nulmark[] = "nul";
-#else
- char nulmark[] = "/dev/null";
-#endif
- char nullinput[] = "null";
-
- // Welcome message
- DISPLAY( WELCOME_MESSAGE);
-
- if (argc<2) { badusage(exename); return 1; }
-
- for(i=1; i<argc; i++)
- {
- char* argument = argv[i];
-
- if(!argument) continue; // Protection if argument empty
-
- // Select command
- if (argument[0]=='-')
- {
- argument ++;
-
- // Display help on usage
- if ( argument[0] =='h' ) { usage(exename); return 0; }
-
- // Compression (default)
- if ( argument[0] =='c' ) { if (argument[1] >='0') cLevel=argument[1] - '0'; continue; }
-
- // Decoding
- if ( argument[0] =='d' ) { decode=1; continue; }
-
- // Bench
- if ( argument[0] =='b' ) { bench=1; if (argument[1] >= '0') cLevel=argument[1] - '0'; continue; }
-
- // Modify Block Size (benchmark only)
- if ( argument[0] =='B' ) { int B = argument[1] - '0'; int S = 1 << (10 + 2*B); BMK_SetBlocksize(S); continue; }
-
- // Modify Nb Iterations (benchmark only)
- if ( argument[0] =='i' ) { int iters = argument[1] - '0'; BMK_SetNbIterations(iters); continue; }
-
- // Pause at the end (benchmark only)
- if ( argument[0] =='p' ) { BMK_SetPause(); continue; }
-
- // Test
- if ( argument[0] =='t' ) { decode=1; output_filename=nulmark; continue; }
- }
-
- // first provided filename is input
- if (!input_filename) { input_filename=argument; filenamesStart=i; continue; }
-
- // second provided filename is output
- if (!output_filename)
- {
- output_filename=argument;
- if (!strcmp (output_filename, nullinput)) output_filename = nulmark;
- continue;
- }
- }
-
- // No input filename ==> Error
- if(!input_filename) { badusage(exename); return 1; }
-
- if (bench) return BMK_benchFile(argv+filenamesStart, argc-filenamesStart, cLevel);
-
- // No output filename ==> Error
- if (!output_filename) { badusage(exename); return 1; }
-
- if (decode) return decode_file(input_filename, output_filename);
-
- return compress_file(input_filename, output_filename, cLevel); // Compression is 'default' action
-
-}
diff --git a/ext/lz4/lz4hc.c b/ext/lz4/lz4hc.c index 7324492e..db6b484a 100644 --- a/ext/lz4/lz4hc.c +++ b/ext/lz4/lz4hc.c @@ -1,6 +1,6 @@ /*
LZ4 HC - High Compression Mode of LZ4
- Copyright (C) 2011-2012, Yann Collet.
+ Copyright (C) 2011-2013, 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
@@ -31,12 +31,24 @@ - LZ4 source repository : http://code.google.com/p/lz4/
*/
+//**************************************
+// Memory routines
+//**************************************
+#include <stdlib.h> // calloc, free
+#define ALLOCATOR(s) calloc(1,s)
+#define FREEMEM free
+#include <string.h> // memset, memcpy
+#define MEM_INIT memset
+
//**************************************
// CPU Feature Detection
//**************************************
// 32 or 64 bits ?
-#if (defined(__x86_64__) || defined(__x86_64) || defined(__amd64__) || defined(__amd64) || defined(__ppc64__) || defined(_WIN64) || defined(__LP64__) || defined(_LP64) ) // Detects 64 bits mode
+#if (defined(__x86_64__) || defined(_M_X64) || defined(_WIN64) \
+ || defined(__powerpc64__) || defined(__ppc64__) || defined(__PPC64__) \
+ || defined(__64BIT__) || defined(_LP64) || defined(__LP64__) \
+ || defined(__ia64) || defined(__itanium__) || defined(_M_IA64) ) // Detects 64 bits mode
# define LZ4_ARCH64 1
#else
# define LZ4_ARCH64 0
@@ -52,7 +64,7 @@ #elif (defined(__BIG_ENDIAN__) || defined(__BIG_ENDIAN) || defined(_BIG_ENDIAN)) && !(defined(__LITTLE_ENDIAN__) || defined(__LITTLE_ENDIAN) || defined(_LITTLE_ENDIAN))
# define LZ4_BIG_ENDIAN 1
#elif defined(__sparc) || defined(__sparc__) \
- || defined(__ppc__) || defined(_POWER) || defined(__powerpc__) || defined(_ARCH_PPC) || defined(__PPC__) || defined(__PPC) || defined(PPC) || defined(__powerpc__) || defined(__powerpc) || defined(powerpc) \
+ || defined(__powerpc__) || defined(__ppc__) || defined(__PPC__) \
|| defined(__hpux) || defined(__hppa) \
|| defined(_MIPSEB) || defined(__s390__)
# define LZ4_BIG_ENDIAN 1
@@ -76,78 +88,84 @@ //**************************************
// Compiler Options
//**************************************
-#if __STDC_VERSION__ >= 199901L // C99
+#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L // C99
/* "restrict" is a known keyword */
#else
# define restrict // Disable restrict
#endif
-#ifdef _MSC_VER
-# define inline __inline // Visual is not C99, but supports some kind of inline
-# define forceinline __forceinline
-# include <intrin.h> // For Visual 2005
-# if LZ4_ARCH64 // 64-bit
+#ifdef _MSC_VER // Visual Studio
+# define FORCE_INLINE static __forceinline
+# include <intrin.h> // For Visual 2005
+# if LZ4_ARCH64 // 64-bits
# pragma intrinsic(_BitScanForward64) // For Visual 2005
# pragma intrinsic(_BitScanReverse64) // For Visual 2005
-# else
+# else // 32-bits
# pragma intrinsic(_BitScanForward) // For Visual 2005
# pragma intrinsic(_BitScanReverse) // For Visual 2005
# endif
+# pragma warning(disable : 4127) // disable: C4127: conditional expression is constant
+# pragma warning(disable : 4701) // disable: C4701: potentially uninitialized local variable used
#else
# ifdef __GNUC__
-# define forceinline inline __attribute__((always_inline))
+# define FORCE_INLINE static inline __attribute__((always_inline))
# else
-# define forceinline inline
+# define FORCE_INLINE static inline
# endif
#endif
#ifdef _MSC_VER // Visual Studio
-#define lz4_bswap16(x) _byteswap_ushort(x)
+# define lz4_bswap16(x) _byteswap_ushort(x)
#else
-#define lz4_bswap16(x) ((unsigned short int) ((((x) >> 8) & 0xffu) | (((x) & 0xffu) << 8)))
+# define lz4_bswap16(x) ((unsigned short int) ((((x) >> 8) & 0xffu) | (((x) & 0xffu) << 8)))
#endif
//**************************************
// Includes
//**************************************
-#include <stdlib.h> // calloc, free
-#include <string.h> // memset, memcpy
#include "lz4hc.h"
-
-#define ALLOCATOR(s) calloc(1,s)
-#define FREEMEM free
-#define MEM_INIT memset
+#include "lz4.h"
//**************************************
// 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
+#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L // C99
+# include <stdint.h>
+ typedef uint8_t BYTE;
+ typedef uint16_t U16;
+ typedef uint32_t U32;
+ typedef int32_t S32;
+ typedef uint64_t U64;
#else
-#include <stdint.h>
-#define BYTE uint8_t
-#define U16 uint16_t
-#define U32 uint32_t
-#define S32 int32_t
-#define U64 uint64_t
+ typedef unsigned char BYTE;
+ typedef unsigned short U16;
+ typedef unsigned int U32;
+ typedef signed int S32;
+ typedef unsigned long long U64;
#endif
-#ifndef LZ4_FORCE_UNALIGNED_ACCESS
-#pragma pack(push, 1)
+#if defined(__GNUC__) && !defined(LZ4_FORCE_UNALIGNED_ACCESS)
+# define _PACKED __attribute__ ((packed))
+#else
+# define _PACKED
#endif
-typedef struct _U16_S { U16 v; } U16_S;
-typedef struct _U32_S { U32 v; } U32_S;
-typedef struct _U64_S { U64 v; } U64_S;
+#if !defined(LZ4_FORCE_UNALIGNED_ACCESS) && !defined(__GNUC__)
+# ifdef __IBMC__
+# pragma pack(1)
+# else
+# pragma pack(push, 1)
+# endif
+#endif
-#ifndef LZ4_FORCE_UNALIGNED_ACCESS
-#pragma pack(pop)
+typedef struct _U16_S { U16 v; } _PACKED U16_S;
+typedef struct _U32_S { U32 v; } _PACKED U32_S;
+typedef struct _U64_S { U64 v; } _PACKED U64_S;
+
+#if !defined(LZ4_FORCE_UNALIGNED_ACCESS) && !defined(__GNUC__)
+# pragma pack(pop)
#endif
#define A64(x) (((U64_S *)(x))->v)
@@ -182,34 +200,40 @@ typedef struct _U64_S { U64 v; } U64_S; #define MINLENGTH (MFLIMIT+1)
#define OPTIMAL_ML (int)((ML_MASK-1)+MINMATCH)
+#define KB *(1U<<10)
+#define MB *(1U<<20)
+#define GB *(1U<<30)
+
//**************************************
// Architecture-specific macros
//**************************************
-#if LZ4_ARCH64 // 64-bit
-#define STEPSIZE 8
-#define LZ4_COPYSTEP(s,d) A64(d) = A64(s); d+=8; s+=8;
-#define LZ4_COPYPACKET(s,d) LZ4_COPYSTEP(s,d)
-#define UARCH U64
-#define AARCH A64
-#define HTYPE U32
-#define INITBASE(b,s) const BYTE* const b = s
-#else // 32-bit
-#define STEPSIZE 4
-#define LZ4_COPYSTEP(s,d) A32(d) = A32(s); d+=4; s+=4;
-#define LZ4_COPYPACKET(s,d) LZ4_COPYSTEP(s,d); LZ4_COPYSTEP(s,d);
-#define UARCH U32
-#define AARCH A32
-#define HTYPE const BYTE*
-#define INITBASE(b,s) const int b = 0
+#if LZ4_ARCH64 // 64-bit
+# define STEPSIZE 8
+# define LZ4_COPYSTEP(s,d) A64(d) = A64(s); d+=8; s+=8;
+# define LZ4_COPYPACKET(s,d) LZ4_COPYSTEP(s,d)
+# define UARCH U64
+# define AARCH A64
+# define HTYPE U32
+# define INITBASE(b,s) const BYTE* const b = s
+#else // 32-bit
+# define STEPSIZE 4
+# define LZ4_COPYSTEP(s,d) A32(d) = A32(s); d+=4; s+=4;
+# define LZ4_COPYPACKET(s,d) LZ4_COPYSTEP(s,d); LZ4_COPYSTEP(s,d);
+# define UARCH U32
+# define AARCH A32
+//# define HTYPE const BYTE*
+//# define INITBASE(b,s) const int b = 0
+# define HTYPE U32
+# define INITBASE(b,s) const BYTE* const b = s
#endif
#if defined(LZ4_BIG_ENDIAN)
-#define LZ4_READ_LITTLEENDIAN_16(d,s,p) { U16 v = A16(p); v = lz4_bswap16(v); d = (s) - v; }
-#define LZ4_WRITE_LITTLEENDIAN_16(p,i) { U16 v = (U16)(i); v = lz4_bswap16(v); A16(p) = v; p+=2; }
-#else // Little Endian
-#define LZ4_READ_LITTLEENDIAN_16(d,s,p) { d = (s) - A16(p); }
-#define LZ4_WRITE_LITTLEENDIAN_16(p,v) { A16(p) = v; p+=2; }
+# define LZ4_READ_LITTLEENDIAN_16(d,s,p) { U16 v = A16(p); v = lz4_bswap16(v); d = (s) - v; }
+# define LZ4_WRITE_LITTLEENDIAN_16(p,i) { U16 v = (U16)(i); v = lz4_bswap16(v); A16(p) = v; p+=2; }
+#else // Little Endian
+# define LZ4_READ_LITTLEENDIAN_16(d,s,p) { d = (s) - A16(p); }
+# define LZ4_WRITE_LITTLEENDIAN_16(p,v) { A16(p) = v; p+=2; }
#endif
@@ -218,7 +242,9 @@ typedef struct _U64_S { U64 v; } U64_S; //************************************************************
typedef struct
{
+ const BYTE* inputBuffer;
const BYTE* base;
+ const BYTE* end;
HTYPE hashTable[HASHTABLESIZE];
U16 chainTable[MAXD];
const BYTE* nextToUpdate;
@@ -230,11 +256,11 @@ typedef struct //**************************************
#define LZ4_WILDCOPY(s,d,e) do { LZ4_COPYPACKET(s,d) } while (d<e);
#define LZ4_BLINDCOPY(s,d,l) { BYTE* e=d+l; LZ4_WILDCOPY(s,d,e); d=e; }
-#define HASH_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8)-HASH_LOG))
-#define HASH_VALUE(p) HASH_FUNCTION(A32(p))
-#define HASH_POINTER(p) (HashTable[HASH_VALUE(p)] + base)
-#define DELTANEXT(p) chainTable[(size_t)(p) & MAXD_MASK]
-#define GETNEXT(p) ((p) - (size_t)DELTANEXT(p))
+#define HASH_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8)-HASH_LOG))
+#define HASH_VALUE(p) HASH_FUNCTION(A32(p))
+#define HASH_POINTER(p) (HashTable[HASH_VALUE(p)] + base)
+#define DELTANEXT(p) chainTable[(size_t)(p) & MAXD_MASK]
+#define GETNEXT(p) ((p) - (size_t)DELTANEXT(p))
//**************************************
@@ -242,99 +268,98 @@ typedef struct //**************************************
#if LZ4_ARCH64
-inline static int LZ4_NbCommonBytes (register U64 val)
+FORCE_INLINE int LZ4_NbCommonBytes (register U64 val)
{
#if defined(LZ4_BIG_ENDIAN)
- #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r = 0;
_BitScanReverse64( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_clzll(val) >> 3);
- #else
+# 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
+# endif
#else
- #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r = 0;
_BitScanForward64( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_ctzll(val) >> 3);
- #else
+# 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
#endif
}
#else
-inline static int LZ4_NbCommonBytes (register U32 val)
+FORCE_INLINE int LZ4_NbCommonBytes (register U32 val)
{
#if defined(LZ4_BIG_ENDIAN)
- #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r;
_BitScanReverse( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_clz(val) >> 3);
- #else
+# else
int r;
if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; }
r += (!val);
return r;
- #endif
+# endif
#else
- #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
unsigned long r;
_BitScanForward( &r, val );
return (int)(r>>3);
- #elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
+# elif defined(__GNUC__) && ((__GNUC__ * 100 + __GNUC_MINOR__) >= 304) && !defined(LZ4_FORCE_SW_BITCOUNT)
return (__builtin_ctz(val) >> 3);
- #else
+# 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 & -(S32)val) * 0x077CB531U)) >> 27];
- #endif
+# endif
#endif
}
#endif
-inline static int LZ4HC_Init (LZ4HC_Data_Structure* hc4, const BYTE* base)
+FORCE_INLINE void LZ4_initHC (LZ4HC_Data_Structure* hc4, const BYTE* base)
{
MEM_INIT((void*)hc4->hashTable, 0, sizeof(hc4->hashTable));
MEM_INIT(hc4->chainTable, 0xFF, sizeof(hc4->chainTable));
- hc4->nextToUpdate = base + LZ4_ARCH64;
+ hc4->nextToUpdate = base + 1;
hc4->base = base;
- return 1;
+ hc4->inputBuffer = base;
+ hc4->end = base;
}
-inline static void* LZ4HC_Create (const BYTE* base)
+void* LZ4_createHC (const char* inputBuffer)
{
void* hc4 = ALLOCATOR(sizeof(LZ4HC_Data_Structure));
-
- LZ4HC_Init ((LZ4HC_Data_Structure*)hc4, base);
+ LZ4_initHC ((LZ4HC_Data_Structure*)hc4, (const BYTE*)inputBuffer);
return hc4;
}
-inline static int LZ4HC_Free (void** LZ4HC_Data)
+int LZ4_freeHC (void* LZ4HC_Data)
{
- FREEMEM(*LZ4HC_Data);
- *LZ4HC_Data = NULL;
- return (1);
+ FREEMEM(LZ4HC_Data);
+ return (0);
}
// Update chains up to ip (excluded)
-forceinline static void LZ4HC_Insert (LZ4HC_Data_Structure* hc4, const BYTE* ip)
+FORCE_INLINE void LZ4HC_Insert (LZ4HC_Data_Structure* hc4, const BYTE* ip)
{
U16* chainTable = hc4->chainTable;
HTYPE* HashTable = hc4->hashTable;
@@ -342,17 +367,37 @@ forceinline static void LZ4HC_Insert (LZ4HC_Data_Structure* hc4, const BYTE* ip) while(hc4->nextToUpdate < ip)
{
- const BYTE* p = hc4->nextToUpdate;
+ const BYTE* const p = hc4->nextToUpdate;
size_t delta = (p) - HASH_POINTER(p);
if (delta>MAX_DISTANCE) delta = MAX_DISTANCE;
DELTANEXT(p) = (U16)delta;
- HashTable[HASH_VALUE(p)] = (p) - base;
+ HashTable[HASH_VALUE(p)] = (HTYPE)((p) - base);
hc4->nextToUpdate++;
}
}
-forceinline static size_t LZ4HC_CommonLength (const BYTE* p1, const BYTE* p2, const BYTE* const matchlimit)
+char* LZ4_slideInputBufferHC(void* LZ4HC_Data)
+{
+ LZ4HC_Data_Structure* hc4 = (LZ4HC_Data_Structure*)LZ4HC_Data;
+ U32 distance = (U32)(hc4->end - hc4->inputBuffer) - 64 KB;
+ distance = (distance >> 16) << 16; // Must be a multiple of 64 KB
+ LZ4HC_Insert(hc4, hc4->end - MINMATCH);
+ memcpy((void*)(hc4->end - 64 KB - distance), (const void*)(hc4->end - 64 KB), 64 KB);
+ hc4->nextToUpdate -= distance;
+ hc4->base -= distance;
+ if ((U32)(hc4->inputBuffer - hc4->base) > 1 GB + 64 KB) // Avoid overflow
+ {
+ int i;
+ hc4->base += 1 GB;
+ for (i=0; i<HASHTABLESIZE; i++) hc4->hashTable[i] -= 1 GB;
+ }
+ hc4->end -= distance;
+ return (char*)(hc4->end);
+}
+
+
+FORCE_INLINE size_t LZ4HC_CommonLength (const BYTE* p1, const BYTE* p2, const BYTE* const matchlimit)
{
const BYTE* p1t = p1;
@@ -370,7 +415,7 @@ forceinline static size_t LZ4HC_CommonLength (const BYTE* p1, const BYTE* p2, co }
-forceinline static int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* const matchlimit, const BYTE** matchpos)
+FORCE_INLINE int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* const matchlimit, const BYTE** matchpos)
{
U16* const chainTable = hc4->chainTable;
HTYPE* const HashTable = hc4->hashTable;
@@ -378,7 +423,7 @@ forceinline static int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, INITBASE(base,hc4->base);
int nbAttempts=MAX_NB_ATTEMPTS;
size_t repl=0, ml=0;
- U16 delta;
+ U16 delta=0; // useless assignment, to remove an uninitialization warning
// HC4 match finder
LZ4HC_Insert(hc4, ip);
@@ -387,7 +432,7 @@ forceinline static int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, #define REPEAT_OPTIMIZATION
#ifdef REPEAT_OPTIMIZATION
// Detect repetitive sequences of length <= 4
- if (ref >= ip-4) // potential repetition
+ if ((U32)(ip-ref) <= 4) // potential repetition
{
if (A32(ref) == A32(ip)) // confirmed
{
@@ -399,7 +444,7 @@ forceinline static int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, }
#endif
- while ((ref >= ip-MAX_DISTANCE) && (nbAttempts))
+ while (((U32)(ip-ref) <= MAX_DISTANCE) && (nbAttempts))
{
nbAttempts--;
if (*(ref+ml) == *(ip+ml))
@@ -427,7 +472,7 @@ forceinline static int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, do
{
DELTANEXT(ptr) = delta;
- HashTable[HASH_VALUE(ptr)] = (ptr) - base; // Head of chain
+ HashTable[HASH_VALUE(ptr)] = (HTYPE)((ptr) - base); // Head of chain
ptr++;
} while(ptr < end);
hc4->nextToUpdate = end;
@@ -438,7 +483,7 @@ forceinline static int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, }
-forceinline static int LZ4HC_InsertAndGetWiderMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* startLimit, const BYTE* matchlimit, int longest, const BYTE** matchpos, const BYTE** startpos)
+FORCE_INLINE int LZ4HC_InsertAndGetWiderMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* startLimit, const BYTE* matchlimit, int longest, const BYTE** matchpos, const BYTE** startpos)
{
U16* const chainTable = hc4->chainTable;
HTYPE* const HashTable = hc4->hashTable;
@@ -451,7 +496,7 @@ forceinline static int LZ4HC_InsertAndGetWiderMatch (LZ4HC_Data_Structure* hc4, LZ4HC_Insert(hc4, ip);
ref = HASH_POINTER(ip);
- while ((ref >= ip-MAX_DISTANCE) && (nbAttempts))
+ while (((U32)(ip-ref) <= MAX_DISTANCE) && (nbAttempts))
{
nbAttempts--;
if (*(startLimit + longest) == *(ref - delta + longest))
@@ -481,7 +526,7 @@ _endCount: const BYTE* ipt = ip + MINMATCH + LZ4HC_CommonLength(ip+MINMATCH, ref+MINMATCH, matchlimit);
#endif
- while ((startt>startLimit) && (reft > hc4->base) && (startt[-1] == reft[-1])) {startt--; reft--;}
+ while ((startt>startLimit) && (reft > hc4->inputBuffer) && (startt[-1] == reft[-1])) {startt--; reft--;}
if ((ipt-startt) > longest)
{
@@ -497,16 +542,26 @@ _endCount: }
-forceinline static int LZ4_encodeSequence(const BYTE** ip, BYTE** op, const BYTE** anchor, int ml, const BYTE* ref)
+typedef enum { noLimit = 0, limitedOutput = 1 } limitedOutput_directive;
+
+FORCE_INLINE int LZ4HC_encodeSequence (
+ const BYTE** ip,
+ BYTE** op,
+ const BYTE** anchor,
+ int matchLength,
+ const BYTE* ref,
+ limitedOutput_directive limitedOutputBuffer,
+ BYTE* oend)
{
- int length, len;
+ int length;
BYTE* token;
// Encode Literal length
length = (int)(*ip - *anchor);
token = (*op)++;
- if (length>=(int)RUN_MASK) { *token=(RUN_MASK<<ML_BITS); len = length-RUN_MASK; for(; len > 254 ; len-=255) *(*op)++ = 255; *(*op)++ = (BYTE)len; }
- else *token = (length<<ML_BITS);
+ if ((limitedOutputBuffer) && ((*op + length + (2 + 1 + LASTLITERALS) + (length>>8)) > oend)) return 1; // Check output limit
+ if (length>=(int)RUN_MASK) { int len; *token=(RUN_MASK<<ML_BITS); len = length-RUN_MASK; for(; len > 254 ; len-=255) *(*op)++ = 255; *(*op)++ = (BYTE)len; }
+ else *token = (BYTE)(length<<ML_BITS);
// Copy Literals
LZ4_BLINDCOPY(*anchor, *op, length);
@@ -515,36 +570,39 @@ forceinline static int LZ4_encodeSequence(const BYTE** ip, BYTE** op, const BYTE LZ4_WRITE_LITTLEENDIAN_16(*op,(U16)(*ip-ref));
// Encode MatchLength
- len = (int)(ml-MINMATCH);
- if (len>=(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;
+ length = (int)(matchLength-MINMATCH);
+ if ((limitedOutputBuffer) && (*op + (1 + LASTLITERALS) + (length>>8) > oend)) return 1; // Check output limit
+ if (length>=(int)ML_MASK) { *token+=ML_MASK; length-=ML_MASK; for(; length > 509 ; length-=510) { *(*op)++ = 255; *(*op)++ = 255; } if (length > 254) { length-=255; *(*op)++ = 255; } *(*op)++ = (BYTE)length; }
+ else *token += (BYTE)(length);
// Prepare next loop
- *ip += ml;
+ *ip += matchLength;
*anchor = *ip;
return 0;
}
-//****************************
-// Compression CODE
-//****************************
-
-int LZ4_compressHCCtx(LZ4HC_Data_Structure* ctx,
+static int LZ4HC_compress_generic (
+ void* ctxvoid,
const char* source,
char* dest,
- int isize)
-{
+ int inputSize,
+ int maxOutputSize,
+ limitedOutput_directive limit
+ )
+{
+ LZ4HC_Data_Structure* ctx = (LZ4HC_Data_Structure*) ctxvoid;
const BYTE* ip = (const BYTE*) source;
const BYTE* anchor = ip;
- const BYTE* const iend = ip + isize;
+ const BYTE* const iend = ip + inputSize;
const BYTE* const mflimit = iend - MFLIMIT;
const BYTE* const matchlimit = (iend - LASTLITERALS);
BYTE* op = (BYTE*) dest;
+ BYTE* const oend = op + maxOutputSize;
- int ml, ml2, ml3, ml0;
+ int ml, ml2, ml3, ml0;
const BYTE* ref=NULL;
const BYTE* start2=NULL;
const BYTE* ref2=NULL;
@@ -553,6 +611,11 @@ int LZ4_compressHCCtx(LZ4HC_Data_Structure* ctx, const BYTE* start0;
const BYTE* ref0;
+
+ // Ensure blocks follow each other
+ if (ip != ctx->end) return 0;
+ ctx->end += inputSize;
+
ip++;
// Main Loop
@@ -573,7 +636,7 @@ _Search2: if (ml2 == ml) // No better match
{
- LZ4_encodeSequence(&ip, &op, &anchor, ml, ref);
+ if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml, ref, limit, oend)) return 0;
continue;
}
@@ -625,9 +688,9 @@ _Search3: // ip & ref are known; Now for ml
if (start2 < ip+ml) ml = (int)(start2 - ip);
// Now, encode 2 sequences
- LZ4_encodeSequence(&ip, &op, &anchor, ml, ref);
+ if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml, ref, limit, oend)) return 0;
ip = start2;
- LZ4_encodeSequence(&ip, &op, &anchor, ml2, ref2);
+ if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml2, ref2, limit, oend)) return 0;
continue;
}
@@ -649,7 +712,7 @@ _Search3: }
}
- LZ4_encodeSequence(&ip, &op, &anchor, ml, ref);
+ if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml, ref, limit, oend)) return 0;
ip = start3;
ref = ref3;
ml = ml3;
@@ -688,7 +751,7 @@ _Search3: ml = (int)(start2 - ip);
}
}
- LZ4_encodeSequence(&ip, &op, &anchor, ml, ref);
+ if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml, ref, limit, oend)) return 0;
ip = start2;
ref = ref2;
@@ -705,8 +768,9 @@ _Search3: // Encode Last Literals
{
int lastRun = (int)(iend - anchor);
+ if ((limit) && (((char*)op - dest) + lastRun + 1 + ((lastRun+255-RUN_MASK)/255) > (U32)maxOutputSize)) return 0; // Check output limit
if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK<<ML_BITS); lastRun-=RUN_MASK; for(; lastRun > 254 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; }
- else *op++ = (lastRun<<ML_BITS);
+ else *op++ = (BYTE)(lastRun<<ML_BITS);
memcpy(op, anchor, iend - anchor);
op += iend-anchor;
}
@@ -716,15 +780,38 @@ _Search3: }
-int LZ4_compressHC(const char* source,
- char* dest,
- int isize)
+int LZ4_compressHC(const char* source, char* dest, int inputSize)
+{
+ void* ctx = LZ4_createHC(source);
+ int result;
+ if (ctx==NULL) return 0;
+
+ result = LZ4HC_compress_generic (ctx, source, dest, inputSize, 0, noLimit);
+
+ LZ4_freeHC(ctx);
+ return result;
+}
+
+int LZ4_compressHC_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize)
+{
+ return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, 0, noLimit);
+}
+
+
+int LZ4_compressHC_limitedOutput(const char* source, char* dest, int inputSize, int maxOutputSize)
{
- void* ctx = LZ4HC_Create((const BYTE*)source);
- int result = LZ4_compressHCCtx(ctx, source, dest, isize);
- LZ4HC_Free (&ctx);
+ void* ctx = LZ4_createHC(source);
+ int result;
+ if (ctx==NULL) return 0;
+ result = LZ4HC_compress_generic (ctx, source, dest, inputSize, maxOutputSize, limitedOutput);
+
+ LZ4_freeHC(ctx);
return result;
}
+int LZ4_compressHC_limitedOutput_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize, int maxOutputSize)
+{
+ return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, maxOutputSize, limitedOutput);
+}
diff --git a/ext/lz4/lz4hc.h b/ext/lz4/lz4hc.h index cb74689f..107fd0fc 100644 --- a/ext/lz4/lz4hc.h +++ b/ext/lz4/lz4hc.h @@ -1,7 +1,7 @@ /*
LZ4 HC - High Compression Mode of LZ4
Header File
- Copyright (C) 2011-2012, Yann Collet.
+ Copyright (C) 2011-2013, 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
@@ -39,19 +39,70 @@ extern "C" { #endif
-int LZ4_compressHC (const char* source, char* dest, int isize);
-
+int LZ4_compressHC (const char* source, char* dest, int inputSize);
/*
LZ4_compressHC :
- return : the number of bytes in compressed buffer dest
- note : destination buffer must be already allocated.
- To avoid any problem, size it to handle worst cases situations (input data not compressible)
- Worst case size evaluation is provided by function LZ4_compressBound() (see "lz4.h")
+ return : the number of bytes in compressed buffer dest
+ or 0 if compression fails.
+ note : destination buffer must be already allocated.
+ To avoid any problem, size it to handle worst cases situations (input data not compressible)
+ Worst case size evaluation is provided by function LZ4_compressBound() (see "lz4.h")
+*/
+
+int LZ4_compressHC_limitedOutput (const char* source, char* dest, int inputSize, int maxOutputSize);
+/*
+LZ4_compress_limitedOutput() :
+ Compress 'inputSize' bytes from 'source' into an output buffer 'dest' of maximum size 'maxOutputSize'.
+ If it cannot achieve it, compression will stop, and result of the function will be zero.
+ This function never writes outside of provided output buffer.
+
+ inputSize : Max supported value is 1 GB
+ maxOutputSize : is maximum allowed size into the destination buffer (which must be already allocated)
+ return : the number of output bytes written in buffer 'dest'
+ or 0 if compression fails.
*/
/* Note :
-Decompression functions are provided within regular LZ4 source code (see "lz4.h") (BSD license)
+Decompression functions are provided within LZ4 source code (see "lz4.h") (BSD license)
+*/
+
+
+/* Advanced Functions */
+
+void* LZ4_createHC (const char* inputBuffer);
+int LZ4_compressHC_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize);
+int LZ4_compressHC_limitedOutput_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize, int maxOutputSize);
+char* LZ4_slideInputBufferHC (void* LZ4HC_Data);
+int LZ4_freeHC (void* LZ4HC_Data);
+
+/*
+These functions allow the compression of dependent blocks, where each block benefits from prior 64 KB within preceding blocks.
+In order to achieve this, it is necessary to start creating the LZ4HC Data Structure, thanks to the function :
+
+void* LZ4_createHC (const char* inputBuffer);
+The result of the function is the (void*) pointer on the LZ4HC Data Structure.
+This pointer will be needed in all other functions.
+If the pointer returned is NULL, then the allocation has failed, and compression must be aborted.
+The only parameter 'const char* inputBuffer' must, obviously, point at the beginning of input buffer.
+The input buffer must be already allocated, and size at least 192KB.
+'inputBuffer' will also be the 'const char* source' of the first block.
+
+All blocks are expected to lay next to each other within the input buffer, starting from 'inputBuffer'.
+To compress each block, use either LZ4_compressHC_continue() or LZ4_compressHC_limitedOutput_continue().
+Their behavior are identical to LZ4_compressHC() or LZ4_compressHC_limitedOutput(),
+but require the LZ4HC Data Structure as their first argument, and check that each block starts right after the previous one.
+If next block does not begin immediately after the previous one, the compression will fail (return 0).
+
+When it's no longer possible to lay the next block after the previous one (not enough space left into input buffer), a call to :
+char* LZ4_slideInputBufferHC(void* LZ4HC_Data);
+must be performed. It will typically copy the latest 64KB of input at the beginning of input buffer.
+Note that, for this function to work properly, minimum size of an input buffer must be 192KB.
+==> The memory position where the next input data block must start is provided as the result of the function.
+
+Compression can then resume, using LZ4_compressHC_continue() or LZ4_compressHC_limitedOutput_continue(), as usual.
+
+When compression is completed, a call to LZ4_freeHC() will release the memory used by the LZ4HC Data Structure.
*/
diff --git a/ext/tap-mac/README.txt b/ext/tap-mac/README.txt new file mode 100644 index 00000000..177b936f --- /dev/null +++ b/ext/tap-mac/README.txt @@ -0,0 +1,19 @@ +This is a hack of tuntaposx. It's here for two reasons: + +1) There seem to be issues with large MTUs in the original tuntap code, + so we set up our zt0 tap with the correct ZeroTier MTU as the default. + +2) Lots of other mac products (VPNs, etc.) ship their own tap device + drivers that like to conflict with one another. This gives us no + choice but to play along. But we call our tap device zt0, which means + it won't conflict with everyone else's tap0. + +3) It's nice to call the device zt0, same as Linux, for consistency across + *nix platforms. Mac does not seem to support interface renaming. + +This will be placed in the ZeroTier home as a kext and is auto-loaded by the +ZeroTier One binary if /dev/zt0 is not found. It can also be auto-updated. + +See this page for the original: + +http://tuntaposx.sourceforge.net diff --git a/ext/tap-mac/tuntap/Changelog b/ext/tap-mac/tuntap/Changelog new file mode 100644 index 00000000..4599a132 --- /dev/null +++ b/ext/tap-mac/tuntap/Changelog @@ -0,0 +1,141 @@ + +November 1, 2011: + o Make the netmask address family fix work without knowledge of the struct + ifaddr definition. This fixes a crash on Lion, where the layout of the + structure has been changed, but at the cost of the fix no longer working for + IPv6. I think this is OK though, since mDNSResponder has been fixed on + Leopard and beyond to no longer require the hack. + o Proper multicast address checking for tun; multicast should now work reliably + with IP and IPv6 on tun also. + o A quite comprehensive test suite has been added that allows for quick release + testing. + o PPC support has been dropped due to XCode 4 no longer supporting PPC arch. + +September 13, 2009: + o Change linker options to produce 64 bit kext bundle for Snow Leopard. + o Switch from kmem_alloc and friends to OSAlloc for memory allocation and + avoid the delay() call. Respective symbols are not available on 64 bit + kernels anymore. + +September 5, 2009: + o Initial Snow Leopard port. Thanks to various people contributing patches in + the bugtracker. The new official version can only be compiled on Snow + Leopard but has been tested to work on all Tiger, Leopard and Snow Leopard + o Clean up unused locking code and switch to rwlocks even for simple mutexes, + which avoids a symbol incompatibility for Tiger and Leopard. + o Clean up compilation flags in the Makefiles. + +July 4, 2008: + o Adapt the former Leopard package to also be installable on Tiger systems. + This obsoletes the Tiger version, both Leopard and Tiger are now supported + by a single package. + +June 7, 2008: + o Protect the selwakeup() call by the lock. This fixes incorrect select() + behaviour, thanks to Roland Wendelin for reporting this. + o Fix tuntap_mbuf_queue::size initialization + o Use a proper wait condition for synchronization when detaching the network + interface. The old code would crash if the if_detached() handler was called + from a different thread than unregister_interface(). + +January 21, 2008: + o Work around an issue in the Darwin kernel. When unregistering an interface, + addresses are not properly removed from the interface. This leads to + crashes and other problems when reusing the interface. Introduce an ugly + hack that tries to remove all interface addresses when shutting the + interface down. + o Fix a small mbuf leak that could occur when the output queue was full. + Thanks to Oleg Dolgov for reporting this. + +December 21, 2007: + o Fix paths in the startup item postflight scripts + o Check if_ioctl arguments more defensively after a report of a panic after + receiving a NULL arg. + +November 14, 2007: + o I have done a complete rework of the installer package generation. The + package is now edited in PackageMaker. The distribution package can still + be built from the commandline though. + o Fix incorrect permission & ownership of the installed files. + +Oktober 11, 2007: + o Fix the permissions of the postflight scripts. Installer packages should work + again. + o Drop the kmod and kmodc++ in the linker command, they seem to be unneeded + with Leopard. + +September 20, 2007: + o Initial Leopard port, it's basically the latest Tiger version with some + Leopard-related fixes and s/Tiger/Leopard/g + o I have switched to a proper version management system (git) and could + remove some of the CVS hacks subsequently. + o The installation packages have been reworked a bit, they now install into + /System/Extensions and /System/StartupItems directly by using + DestinationPaths. + +May 13, 2006: + o This version is not stable, it may crash, sorry. + o Universal binaries that run on ppc and intel macs. + o Adds tap MAC address randomization + o Redesigned locking. + o Better multicast support + o mDNSResponder workaround, so that the tap interfaces should get picked up + now. Note that we are fixing ifconfig/kernel behaviour here. + o All tapX and tunX devices are visible in /dev at all times, network + interfaces still created dynamically, though. + o Startup items moved to /Library/StartupItems + +May 17, 2005: + o Initial Tiger port. We now have KPI-style interfaces. I guess the Tiger + version is little slower than the Panther version because of all the + wrapping and hiding in the kernel. + o The kernel extensions moved to /Library/Extensions. That is the place where + non-Apple kexts are supposed to live. + +April 21, 2005: + o I added support in tun for AF prepending like some BSDs do. Thanks to Dennis + kSchneider for mailing the initial patch. You can also set the value of + AF_INET6 to be used. + o I finally found that major bug causing crashes (especially on multiprocessor + machines). It also caused a memory leak (lost mbufs), and might have caused + performance/througput/data-loss problems. Everyone is recommended to upgrade. + +April 6, 2005: + o I rewrote the common part concerning the tun and tap initialization and + cleanup. This should make the code more maintainable (less duplication). + o The devices now reinitialize to the state they were started in when they + are closed by an application. This concerns IP addresses for example. + o I changed the package building system to use PackageMaker.app in batch + mode. The packages also check for version 10.3 now, so nobody should be + able to install tun/tap on 10.2 using installer packages. Furthermore I + have sprinkled some warnings telling you not to use tun/tap on SMP machines + over the installation process ;-) + o Some minor locking fixes. + +November 19, 2004: + o Jamie Wood reported that the packet queue in the driver could be considered + empty even if there were packets in it. This was probably caused by a + synchronization problem that should be fixed now. People encountering + timeouts etc. should try the new version. + o I finally implemented support for changing the interface MTU. The driver + enforces the MTU when writing packets to the character device now. However, + packets coming from the kernel are not checked. + +September 9, 2004: + o Marcello Teodori told me that the tun driver wasn't working with openvpn. + The problem was the fcntl call, fixed that. Should work now. Thanks + Marcello! + o changed the tun driver not to prepend the address family field before each + and every packet (which is the behaviour of OpenBSD). As there is currently + only IPv4 and IPv6 support there is no problem with the standard tun + approach used on other OSes. This should make the driver much more + compatible. + o Did a script and makefile support so that the installer packages can now be + built from the command prompt. Unfortunately this might break things + someday as I am not using the 'official' way to build the packages + o Cleaned up installer packages a little. + +August 24, 2004: + o initial version put online + o basic tun/tap support, tap working with qemu + diff --git a/ext/tap-mac/tuntap/Makefile b/ext/tap-mac/tuntap/Makefile new file mode 100644 index 00000000..8d79577e --- /dev/null +++ b/ext/tap-mac/tuntap/Makefile @@ -0,0 +1,12 @@ +TUNTAP_VERSION = 20131028 +BASE= + +all: tap.kext + +clean: + cd src/tap && make -f Makefile clean + +tap.kext: + cd src/tap && make TUNTAP_VERSION=$(TUNTAP_VERSION) -f Makefile all + +.PHONY: test diff --git a/ext/tap-mac/tuntap/README b/ext/tap-mac/tuntap/README new file mode 100644 index 00000000..6a9e526b --- /dev/null +++ b/ext/tap-mac/tuntap/README @@ -0,0 +1,85 @@ + +tun/tap driver for Mac OS X +=========================== + +This is an experimental IP tunnel/ethertap driver for Mac OS X/Darwin. It +provides /dev/tunX and /dev/tapX devices. The maximum number of devices can be +configured at compile time, it is currently set to 16. That should be enough in +most cases. + +The driver ships as two kernel extensions, one for tap and one for tun. They are +located in /Library/Extensions and can also be loaded and unloaded by hand. If +you install the startup item, the system will load them automatically at +startup (tun and tap startup items get installed in /Library/StartupItems). + +Operation & Programming notes +============================= + +tapX are ethertap devices which provide an interface to the kernel's ethernet +layer. Packets can be read from and written to the /dev/tapX character devices +one at a time (same name as the interface that shows up in ifconfig). + +tunX are IP tunnel devices. These can be used to exchange IP packets with the +kernel. You will get single packets for each read() and should write() packets +one at a time to /dev/tunX. + +There are some special ioctls with the tun devices that allow you to have them +prepend the address family of the packet when reading it from /dev/tunX. Using +this mode the driver also expects you put this 4-byte address family field +(network byte order) in front of the packets you write to /dev/tunX. + +Here are the ioctls to setup up address prepending mode (for convenience there +also is a header called tun_ioctls.h in the source package that you can use) +Set the int argument to one if you want to have AF prepending, use 0 if you want +to switch it off. + +#define TUNSIFHEAD _IOW('t', 96, int) +#define TUNGIFHEAD _IOR('t', 97, int) + +Prepending mode is off by default. Currently it is not recommended to switch the +mode while packets are in flight on the device. + +The character devices are always visible in the filesystem as /dev/tunX and +/dev/tapX. The number of available character devices is a compile time constant +and is currently fixed to 16. Each character devices is associated with a +network interface of the same name. The network interfaces are only created when +the corresponding character device is opened by a program and will be removed +when the character device is closed. + +The character devices currently provide a pretty minimal interface. Whole +packets are read and written using a singe read/write call. File descriptors +opened on the devices can also be select()ed and support O_NONBLOCK. +Asynchronous i/o and some ioctls are currently unimplemented, but implementing +them shouldn't be very hard. Do it yourself or contact me if you can't live +without. + +There is another limitation imposed by the Darwin 8 kernel. It concerns the +poll() system call; Darwin currently does *not* support that for (character) +devices. Use select() instead. + +The interfaces can be configured using ifconfig, the tap devices also support +setting the MAC address to be used. Both tun and tap should be ready for IPv6. +Just setup addresses and routing as you would do with other interfaces. + +Please contact me if you find any bugs or have suggestions. + +Enjoy! + +Mattias +<mattias.nissler@gmx.de> + + +Uninstalling +============ + +The installer packages for OS X currently don't have support for uninstall as +the installer doesn't provide it. Remove the following directories if you want +to completely remove the files installed: + +/Library/Extensions/tap.kext +/Library/Extensions/tun.kext +/Library/StartupItems/tap +/Library/StartupItems/tun + +Unload the the kernel extensions or reboot and you're done. + diff --git a/ext/tap-mac/tuntap/README.zerotier-build b/ext/tap-mac/tuntap/README.zerotier-build new file mode 100644 index 00000000..20d6d461 --- /dev/null +++ b/ext/tap-mac/tuntap/README.zerotier-build @@ -0,0 +1,18 @@ +Building the tap for both x86_64 and i386 requires an older version of the +Xcode tools than what now ships for Mavericks (10.9). The newer version +does not support creating i386 kernel images. + +These can be obtained from: + +https://developer.apple.com/downloads + +It requires a bit of a dance to unpack the package and obtain an unpacked +tree, but once it's there you can change the line in tap/Makefile and +build for both architectures. + +This will go on until i386 is thoroughly legacy, at which point we'll +probably start just supporting x86_64. But that might be a while. We want +to support old Macs through their entire useful life. + +Since this build is irritating, a pre-built copy is packaged in ext/ and +is installed by 'make install'. So users shouldn't have to build this. diff --git a/ext/tap-mac/tuntap/src/lock.cc b/ext/tap-mac/tuntap/src/lock.cc new file mode 100644 index 00000000..0da48be2 --- /dev/null +++ b/ext/tap-mac/tuntap/src/lock.cc @@ -0,0 +1,201 @@ +/* + * ip tunnel/ethertap device for MacOSX. + * + * Locking implementation. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#include "lock.h" + +extern "C" { + +#include <sys/syslog.h> +#include <sys/proc.h> + +} + +#if 0 +#define dprintf(...) log(LOG_INFO, __VA_ARGS__) +#else +#define dprintf(...) +#endif + +/* class tt_lock */ +lck_grp_t *tt_lock::tt_lck_grp = NULL; + +bool +tt_lock::initialize() +{ + /* init if necessary */ + if (tt_lck_grp == NULL) { + dprintf("initing lock group\n"); + tt_lck_grp = lck_grp_alloc_init("tuntap locks", LCK_GRP_ATTR_NULL); + + if (tt_lck_grp == NULL) { + /* if something fails, the lock won't work */ + log(LOG_ERR, "tuntap: could not allocate locking group\n"); + return false; + } + } + + return true; +} + +void +tt_lock::shutdown() +{ + /* free the locking group */ + if (tt_lck_grp != NULL) { + dprintf("freeing lock group\n"); + lck_grp_free(tt_lck_grp); + tt_lck_grp = NULL; + } +} + +/* tt_mutex */ +tt_mutex::tt_mutex() +{ + /* fail if locking group not initialized */ + if (tt_lck_grp == NULL) + return; + + /* allocate the lock */ + lck = lck_rw_alloc_init(tt_lck_grp, NULL); + + if (lck == NULL) + log(LOG_ERR, "tuntap: could not allocate mutex\n"); +} + +tt_mutex::~tt_mutex() +{ + /* if the lock doesn't exist, this will be a no-op */ + if (lck == NULL) + return; + + /* free the lock */ + lck_rw_free(lck, tt_lck_grp); +} + +void +tt_mutex::lock() +{ + if (lck != NULL) + lck_rw_lock_exclusive(lck); +} + +void +tt_mutex::unlock() +{ + if (lck != NULL) + lck_rw_unlock_exclusive(lck); +} + +void +tt_mutex::sleep(void *cond) +{ + if (lck != NULL) + lck_rw_sleep(lck, LCK_SLEEP_DEFAULT, cond, THREAD_INTERRUPTIBLE); +} + +void +tt_mutex::sleep(void *cond, uint64_t timeout) +{ + if (lck != NULL) + lck_rw_sleep_deadline(lck, LCK_SLEEP_DEFAULT, cond, THREAD_INTERRUPTIBLE, timeout); +} + +void +tt_mutex::wakeup(void *cond) +{ + if (lck != NULL) + ::wakeup(cond); +} + +/* tt_gate */ +tt_gate::tt_gate() + : ticket_number(0), + population(0) +{ +} + +void +tt_gate::enter() +{ + /* just try to grab the lock, increase the ticket number and the population */ + auto_lock l(&slock); + ticket_number++; + population++; +} + +void +tt_gate::exit() +{ + auto_lock l(&slock); + ticket_number--; + population--; +} + +bool +tt_gate::is_anyone_in() +{ + return population != 0; +} + +unsigned int +tt_gate::get_ticket_number() +{ + return ticket_number; +} + +void +tt_gate::lock() +{ + slock.lock(); +} + +void +tt_gate::unlock() +{ + slock.unlock(); +} + +void +tt_gate::sleep(void* cond) +{ + slock.sleep(cond); +} + +void +tt_gate::sleep(void* cond, uint64_t timeout) +{ + slock.sleep(cond, timeout); +} + +void +tt_gate::wakeup(void* cond) +{ + slock.wakeup(cond); +} + diff --git a/ext/tap-mac/tuntap/src/lock.h b/ext/tap-mac/tuntap/src/lock.h new file mode 100644 index 00000000..51d3299a --- /dev/null +++ b/ext/tap-mac/tuntap/src/lock.h @@ -0,0 +1,160 @@ +/* + * ip tunnel/ethertap device for MacOSX. + * + * Locking is not as straightforward for Tiger. So declare our own locking class. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#ifndef __LOCK_H__ +#define __LOCK_H__ + +extern "C" { + +#include <kern/locks.h> +#include <sys/param.h> + +} + +/* our own locking class. declares the common interface of the locking primitives. */ +class tt_lock { + + protected: + /* locking group */ + static lck_grp_t *tt_lck_grp; + + public: + /* be virtual */ + virtual ~tt_lock() { }; + + /* static intialization (inits the locking group) */ + static bool initialize(); + static void shutdown(); + + /* locking */ + virtual void lock() = 0; + virtual void unlock() = 0; + + /* monitor primitives */ + virtual void sleep(void* cond) = 0; + virtual void sleep(void* cond, uint64_t) = 0; + virtual void wakeup(void* cond) = 0; +}; + +/* simple mutex */ +class tt_mutex : public tt_lock { + + private: + /* underlying darwin lock */ + lck_rw_t *lck; + + public: + tt_mutex(); + virtual ~tt_mutex(); + + void lock(); + void unlock(); + + /* monitor primitives */ + void sleep(void* cond); + void sleep(void* cond, uint64_t); + void wakeup(void* cond); +}; + +/* A very special locking class that we use to track threads that enter and leave the character + * device service functions. They call enter() before entering the actual service routinge and + * exit() when done. enter() only permits them to pass when the gate isn't locked. Furthermore, the + * gate assigns ticket numbers to everyone that passes the gate, so you can check whether more + * threads came through. See tuntap_mgr::shutdown() for how we use that stuff. + */ +class tt_gate : public tt_lock { + + private: + /* synchronization lock */ + tt_mutex slock; + /* ticket number */ + unsigned int ticket_number; + /* count of threads that are in */ + unsigned int population; + + public: + /* construct a new gate */ + tt_gate(); + + /* enter - pass the gate */ + void enter(); + /* exit - pass the gate */ + void exit(); + + /* check whether anyone is in */ + bool is_anyone_in(); + /* gets the next ticket number */ + unsigned int get_ticket_number(); + + /* lock the gate */ + void lock(); + /* unlock the gate */ + void unlock(); + + /* monitor primitives */ + void sleep(void* cond); + void sleep(void* cond, uint64_t); + void wakeup(void* cond); +}; + +/* auto_lock and auto_rwlock serve as automatic lock managers: Create an object, passing the + * tt_[rw]lock you want to lock to have it grab the lock. When the object goes out of scope, the + * destructor of the class will release the lock. + */ +class auto_lock { + + protected: + /* the lock we hold */ + tt_lock *l; + + public: + auto_lock(tt_lock *m) + : l(m) + { + lock(); + } + + ~auto_lock() + { + unlock(); + } + + void lock() + { + l->lock(); + } + + void unlock() + { + l->unlock(); + } +}; + +#endif /* __LOCK_H__ */ + diff --git a/ext/tap-mac/tuntap/src/mem.cc b/ext/tap-mac/tuntap/src/mem.cc new file mode 100644 index 00000000..cd3264fa --- /dev/null +++ b/ext/tap-mac/tuntap/src/mem.cc @@ -0,0 +1,76 @@ +/* + * ip tunnel/ethertap device for MacOSX. Common functionality of tap_interface and tun_interface. + * + * Memory management implementation. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#include "mem.h" + +extern "C" { + +#include <libkern/OSMalloc.h> + +} + +#if 0 +#define dprintf(...) log(LOG_INFO, __VA_ARGS__) +#else +#define dprintf(...) +#endif + +static int inited = 0; +static OSMallocTag tag; + +void +mem_initialize(const char* name) { + + if (!inited) { + tag = OSMalloc_Tagalloc(name, OSMT_DEFAULT); + inited = 1; + } +} + +void +mem_shutdown() { + + if (inited) { + OSMalloc_Tagfree(tag); + inited = 0; + } +} + +void * +mem_alloc(uint32_t size) { + + return OSMalloc(size, tag); +} + +void +mem_free(void *addr, uint32_t size) { + + OSFree(addr, size, tag); +} + diff --git a/ext/tap-mac/tuntap/src/mem.h b/ext/tap-mac/tuntap/src/mem.h new file mode 100644 index 00000000..4d06fd8c --- /dev/null +++ b/ext/tap-mac/tuntap/src/mem.h @@ -0,0 +1,48 @@ +/* + * ip tunnel/ethertap device for MacOSX. Common functionality of tap_interface and tun_interface. + * + * Memory management. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#ifndef __MEM_H__ +#define __MEM_H__ + +extern "C" { + +#include <stdint.h> + +} + +/* Memory manager initalization and shutdown */ +void mem_initialize(const char *name); +void mem_shutdown(); + +/* Memory allocation functions */ +void *mem_alloc(uint32_t size); +void mem_free(void *addr, uint32_t size); + +#endif /* __MEM_H__ */ + diff --git a/ext/tap-mac/tuntap/src/tap/Info.plist b/ext/tap-mac/tuntap/src/tap/Info.plist new file mode 100644 index 00000000..bb9b03fd --- /dev/null +++ b/ext/tap-mac/tuntap/src/tap/Info.plist @@ -0,0 +1,36 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd"> +<plist version="1.0"> +<dict> + <key>CFBundleDevelopmentRegion</key> + <string>@@CFBUNDLEDEVELOPMENTREGION@@</string> + <key>CFBundleExecutable</key> + <string>@@CFBUNDLEEXECUTABLE@@</string> + <key>CFBundleIdentifier</key> + <string>@@CFBUNDLEIDENTIFIER@@</string> + <key>CFBundleInfoDictionaryVersion</key> + <string>6.0</string> + <key>CFBundleName</key> + <string>@@CFBUNDLEEXECUTABLE@@</string> + <key>CFBundlePackageType</key> + <string>@@CFBUNDLEPACKAGETYPE@@</string> + <key>CFBundleShortVersionString</key> + <string>@@CFBUNDLEVERSION@@</string> + <key>CFBundleSignature</key> + <string>@@CFBUNDLESIGNATURE@@</string> + <key>CFBundleVersion</key> + <string>1.0</string> + <key>OSBundleLibraries</key> + <dict> + <key>com.apple.kpi.mach</key> + <string>8.0</string> + <key>com.apple.kpi.bsd</key> + <string>8.0</string> + <key>com.apple.kpi.libkern</key> + <string>8.0</string> + <key>com.apple.kpi.unsupported</key> + <string>8.0</string> + </dict> +</dict> +</plist> + diff --git a/ext/tap-mac/tuntap/src/tap/Makefile b/ext/tap-mac/tuntap/src/tap/Makefile new file mode 100644 index 00000000..58e582da --- /dev/null +++ b/ext/tap-mac/tuntap/src/tap/Makefile @@ -0,0 +1,60 @@ +# +# ethertap driver for MacOSX +# +# Makefile +# +# (c) 2004, 2005, 2006, 2007, 2008 Mattias Nissler +# + +OBJS = ../tuntap.o ../tuntap_mgr.o ../lock.o ../mem.o kmod.o tap.o +KMOD_BIN = tap +BUNDLE_DIR = ../.. +BUNDLE_NAME = tap.kext + +TAP_KEXT_VERSION = $(TUNTAP_VERSION) + +BUNDLE_REGION = English +BUNDLE_IDENTIFIER = com.zerotier.tap +BUNDLE_SIGNATURE = ???? +BUNDLE_PACKAGETYPE = KEXT +BUNDLE_VERSION = $(TAP_KEXT_VERSION) + +INCLUDE = -I.. -I/System/Library/Frameworks/Kernel.framework/Headers -I/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.8.sdk/System/Library/Frameworks/Kernel.framework/Headers +CFLAGS = -Wall -mkernel -force_cpusubtype_ALL \ + -fno-builtin -fno-stack-protector -arch i386 -arch x86_64 \ + -DKERNEL -D__APPLE__ -DKERNEL_PRIVATE -DTUNTAP_VERSION=\"$(TUNTAP_VERSION)\" \ + -DTAP_KEXT_VERSION=\"$(TAP_KEXT_VERSION)\" +CCFLAGS = $(CFLAGS) +LDFLAGS = -Wall -mkernel -nostdlib -r -lcc_kext -arch i386 -arch x86_64 -Xlinker -kext + +#CCP = g++ +#CC = gcc +CCP = $(HOME)/Code/llvm-g++-Xcode4.6.2/bin/llvm-g++ +CC = $(HOME)/Code/llvm-g++-Xcode4.6.2/bin/llvm-gcc + +all: $(KMOD_BIN) bundle + +.c.o: + $(CC) $(CFLAGS) $(INCLUDE) -c $< -o $@ +.cc.o: + $(CCP) $(CCFLAGS) $(INCLUDE) -c $< -o $@ + +$(KMOD_BIN): $(OBJS) + $(CCP) $(LDFLAGS) -o $(KMOD_BIN) $(OBJS) + +bundle: $(KMOD_BIN) + rm -rf $(BUNDLE_DIR)/$(BUNDLE_NAME) + mkdir -p $(BUNDLE_DIR)/$(BUNDLE_NAME)/Contents/MacOS + cp $(KMOD_BIN) $(BUNDLE_DIR)/$(BUNDLE_NAME)/Contents/MacOS + sed -e "s/@@CFBUNDLEEXECUTABLE@@/$(KMOD_BIN)/" \ + -e "s/@@CFBUNDLEDEVELOPMENTREGION@@/$(BUNDLE_REGION)/" \ + -e "s/@@CFBUNDLEIDENTIFIER@@/$(BUNDLE_IDENTIFIER)/" \ + -e "s/@@CFBUNDLESIGNATURE@@/$(BUNDLE_SIGNATURE)/" \ + -e "s/@@CFBUNDLEPACKAGETYPE@@/$(BUNDLE_PACKAGETYPE)/" \ + -e "s/@@CFBUNDLEVERSION@@/$(BUNDLE_VERSION)/" \ + Info.plist > $(BUNDLE_DIR)/$(BUNDLE_NAME)/Contents/Info.plist + +clean: + -rm -f $(OBJS) $(KMOD_BIN) + -rm -rf $(BUNDLE_DIR)/$(BUNDLE_NAME) + diff --git a/ext/tap-mac/tuntap/src/tap/kmod.cc b/ext/tap-mac/tuntap/src/tap/kmod.cc new file mode 100644 index 00000000..f9c4a40e --- /dev/null +++ b/ext/tap-mac/tuntap/src/tap/kmod.cc @@ -0,0 +1,93 @@ +/* + * ethertap device for MacOSX. + * + * Kext definition (it is a mach kmod really...) + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#include "tap.h" +#include "mem.h" + +extern "C" { + +#include <sys/param.h> + +#include <mach/kmod.h> + +static tap_manager *mgr; + +/* + * start function. called when the kext gets loaded. + */ +static kern_return_t tap_module_start(struct kmod_info *ki, void *data) +{ + mem_initialize(TAP_FAMILY_NAME); + + /* initialize locking */ + if (!tt_lock::initialize()) + return KMOD_RETURN_FAILURE; + + /* create a tap manager that will handle the rest */ + mgr = new tap_manager(); + + if (mgr != NULL) { + if (mgr->initialize(TAP_IF_COUNT, (char *) TAP_FAMILY_NAME)) + return KMOD_RETURN_SUCCESS; + + delete mgr; + mgr = NULL; + /* clean up locking */ + tt_lock::shutdown(); + } + + return KMOD_RETURN_FAILURE; +} + +/* + * stop function. called when the kext should be unloaded. unloading can be prevented by + * returning failure + */ +static kern_return_t tap_module_stop(struct kmod_info *ki, void *data) +{ + if (mgr != NULL) { + if (!mgr->shutdown()) + return KMOD_RETURN_FAILURE; + + delete mgr; + mgr = NULL; + } + + /* clean up locking */ + tt_lock::shutdown(); + + mem_shutdown(); + + return KMOD_RETURN_SUCCESS; +} + +KMOD_DECL(tap, TAP_KEXT_VERSION) + +} + diff --git a/ext/tap-mac/tuntap/src/tap/tap.cc b/ext/tap-mac/tuntap/src/tap/tap.cc new file mode 100644 index 00000000..149f1e71 --- /dev/null +++ b/ext/tap-mac/tuntap/src/tap/tap.cc @@ -0,0 +1,452 @@ +/* + * ethertap device for macosx. + * + * tap_interface class definition + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#include "tap.h" + +extern "C" { + +#include <sys/systm.h> +#include <sys/syslog.h> +#include <sys/param.h> +#include <sys/sockio.h> +#include <sys/random.h> +#include <sys/kern_event.h> + +#include <net/if_types.h> +#include <net/if_arp.h> +#include <net/if_dl.h> +#include <net/if_media.h> +#include <net/dlil.h> +#include <net/ethernet.h> + +} + +#if 0 +#define dprintf(...) log(LOG_INFO, __VA_ARGS__) +#else +#define dprintf(...) +#endif + +// These declarations are missing in the Kernel.framework headers, put present in userspace :-/ +#pragma pack(4) +struct ifmediareq { + char ifm_name[IFNAMSIZ]; /* if name, e.g. "en0" */ + int ifm_current; /* current media options */ + int ifm_mask; /* don't care mask */ + int ifm_status; /* media status */ + int ifm_active; /* active options */ + int ifm_count; /* # entries in ifm_ulist array */ + int *ifm_ulist; /* media words */ +}; + +struct ifmediareq64 { + char ifm_name[IFNAMSIZ]; /* if name, e.g. "en0" */ + int ifm_current; /* current media options */ + int ifm_mask; /* don't care mask */ + int ifm_status; /* media status */ + int ifm_active; /* active options */ + int ifm_count; /* # entries in ifm_ulist array */ + user64_addr_t ifmu_ulist __attribute__((aligned(8))); +}; + +struct ifmediareq32 { + char ifm_name[IFNAMSIZ]; /* if name, e.g. "en0" */ + int ifm_current; /* current media options */ + int ifm_mask; /* don't care mask */ + int ifm_status; /* media status */ + int ifm_active; /* active options */ + int ifm_count; /* # entries in ifm_ulist array */ + user32_addr_t ifmu_ulist; /* 32-bit pointer */ +}; +#pragma pack() + +#define SIOCGIFMEDIA32 _IOWR('i', 56, struct ifmediareq32) /* get net media */ +#define SIOCGIFMEDIA64 _IOWR('i', 56, struct ifmediareq64) /* get net media (64-bit) */ + + +static unsigned char ETHER_BROADCAST_ADDR[] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; + +/* members */ +bool +tap_interface::initialize(unsigned short major, unsigned short unit) +{ + this->unit = unit; + this->family_name = TAP_FAMILY_NAME; + this->family = IFNET_FAMILY_ETHERNET; + this->type = IFT_ETHER; + bzero(unique_id, UIDLEN); + snprintf(unique_id, UIDLEN, "%s%d", family_name, unit); + + dprintf("tap: starting interface %s%d\n", TAP_FAMILY_NAME, unit); + + /* register character device */ + if (!tuntap_interface::register_chardev(major)) + return false; + + return true; +} + +void +tap_interface::shutdown() +{ + dprintf("tap: shutting down tap interface %s%d\n", TAP_FAMILY_NAME, unit); + + unregister_chardev(); +} + +int +tap_interface::initialize_interface() +{ + struct sockaddr_dl lladdr; + lladdr.sdl_len = sizeof(lladdr); + lladdr.sdl_family = AF_LINK; + lladdr.sdl_alen = ETHER_ADDR_LEN; + lladdr.sdl_nlen = lladdr.sdl_slen = 0; + + /* generate a random MAC address */ + read_random(LLADDR(&lladdr), ETHER_ADDR_LEN); + + /* clear multicast bit and set local assignment bit (see IEEE 802) */ + (LLADDR(&lladdr))[0] &= 0xfe; + (LLADDR(&lladdr))[0] |= 0x02; + + dprintf("tap: random tap address: %02x:%02x:%02x:%02x:%02x:%02x\n", + (LLADDR(&lladdr))[0] & 0xff, + (LLADDR(&lladdr))[1] & 0xff, + (LLADDR(&lladdr))[2] & 0xff, + (LLADDR(&lladdr))[3] & 0xff, + (LLADDR(&lladdr))[4] & 0xff, + (LLADDR(&lladdr))[5] & 0xff); + + /* register interface */ + if (!tuntap_interface::register_interface(&lladdr, ETHER_BROADCAST_ADDR, ETHER_ADDR_LEN)) + return EIO; + + /* Set link level address. Yes, we need to do that again. Darwin sucks. */ + errno_t err = ifnet_set_lladdr(ifp, LLADDR(&lladdr), ETHER_ADDR_LEN); + if (err) + dprintf("tap: failed to set lladdr on %s%d: %d\n", family_name, unit, err); + + /* set mtu */ + ifnet_set_mtu(ifp, TAP_MTU); + /* set header length */ + ifnet_set_hdrlen(ifp, sizeof(struct ether_header)); + /* add the broadcast flag */ + ifnet_set_flags(ifp, IFF_BROADCAST, IFF_BROADCAST); + + /* we must call bpfattach(). Otherwise we deadlock BPF while unloading. Seems to be a bug in + * the kernel, see bpfdetach() in net/bpf.c, it will return without releasing the lock if + * the interface wasn't attached. I wonder what they were smoking while writing it ;-) + */ + bpfattach(ifp, DLT_EN10MB, ifnet_hdrlen(ifp)); + + return 0; +} + +void +tap_interface::shutdown_interface() +{ + dprintf("tap: shutting down network interface of device %s%d\n", TAP_FAMILY_NAME, unit); + + /* detach all protocols */ + for (unsigned int i = 0; i < MAX_ATTACHED_PROTOS; i++) { + if (attached_protos[i].used) { + errno_t err = ifnet_detach_protocol(ifp, attached_protos[i].proto); + if (err) + log(LOG_WARNING, "tap: could not detach protocol %d from %s%d\n", + attached_protos[i].proto, TAP_FAMILY_NAME, unit); + } + } + + cleanup_interface(); + unregister_interface(); +} + +errno_t +tap_interface::if_ioctl(u_int32_t cmd, void *arg) +{ + dprintf("tap: if_ioctl cmd: %d (%x)\n", cmd & 0xff, cmd); + + switch (cmd) { + case SIOCSIFLLADDR: + { + /* set ethernet address */ + struct sockaddr *ea = &(((struct ifreq *) arg)->ifr_addr); + + dprintf("tap: SIOCSIFLLADDR family %d len %d\n", + ea->sa_family, ea->sa_len); + + /* check if it is really an ethernet address */ + if (ea->sa_family != AF_LINK || ea->sa_len != ETHER_ADDR_LEN) + return EINVAL; + + /* ok, copy */ + errno_t err = ifnet_set_lladdr(ifp, ea->sa_data, ETHER_ADDR_LEN); + if (err) { + dprintf("tap: failed to set lladdr on %s%d: %d\n", + family_name, unit, err); + return err; + } + + /* Generate a LINK_ON event. This necessary for configd to re-read + * the interface data and refresh the MAC address. Not doing so + * would result in the DHCP client using a stale MAC address... + */ + generate_link_event(KEV_DL_LINK_ON); + + return 0; + } + + case SIOCGIFMEDIA32: + case SIOCGIFMEDIA64: + { + struct ifmediareq *ifmr = (struct ifmediareq*) arg; + user_addr_t list = USER_ADDR_NULL; + + ifmr->ifm_current = IFM_ETHER; + ifmr->ifm_mask = 0; + ifmr->ifm_status = IFM_AVALID | IFM_ACTIVE; + ifmr->ifm_active = IFM_ETHER; + ifmr->ifm_count = 1; + + if (cmd == SIOCGIFMEDIA64) + list = ((struct ifmediareq64*) ifmr)->ifmu_ulist; + else + list = CAST_USER_ADDR_T( + ((struct ifmediareq32*) ifmr)->ifmu_ulist); + + if (list != USER_ADDR_NULL) + return copyout(&ifmr->ifm_current, list, sizeof(int)); + + return 0; + } + + default: + /* let our superclass handle it */ + return tuntap_interface::if_ioctl(cmd, arg); + } + + return EOPNOTSUPP; +} + +errno_t +tap_interface::if_demux(mbuf_t m, char *header, protocol_family_t *proto) +{ + struct ether_header *eh = (struct ether_header *) header; + unsigned char lladdr[ETHER_ADDR_LEN]; + + dprintf("tap: if_demux\n"); + + /* size check */ + if (mbuf_len(m) < sizeof(struct ether_header)) + return ENOENT; + + /* catch broadcast and multicast (stolen from bsd/net/ether_if_module.c) */ + if (eh->ether_dhost[0] & 1) { + if (memcmp(ETHER_BROADCAST_ADDR, eh->ether_dhost, ETHER_ADDR_LEN) == 0) { + /* broadcast */ + dprintf("tap: broadcast packet.\n"); + mbuf_setflags_mask(m, MBUF_BCAST, MBUF_BCAST); + } else { + /* multicast */ + dprintf("tap: multicast packet.\n"); + mbuf_setflags_mask(m, MBUF_MCAST, MBUF_MCAST); + } + } else { + /* check wether the packet has our address */ + ifnet_lladdr_copy_bytes(ifp, lladdr, ETHER_ADDR_LEN); + if (memcmp(lladdr, eh->ether_dhost, ETHER_ADDR_LEN) != 0) + mbuf_setflags_mask(m, MBUF_PROMISC, MBUF_PROMISC); + } + + /* find the protocol */ + for (unsigned int i = 0; i < MAX_ATTACHED_PROTOS; i++) { + if (attached_protos[i].used && attached_protos[i].type == eh->ether_type) { + *proto = attached_protos[i].proto; + return 0; + } + } + + dprintf("tap: if_demux() failed to find proto.\n"); + + /* no matching proto found */ + return ENOENT; +} + +errno_t +tap_interface::if_framer(mbuf_t *m, const struct sockaddr *dest, const char *dest_linkaddr, + const char *frame_type) +{ + struct ether_header *eh; + mbuf_t nm = *m; + errno_t err; + + dprintf("tap: if_framer\n"); + + /* prepend the ethernet header */ + err = mbuf_prepend(&nm, sizeof (struct ether_header), MBUF_WAITOK); + if (err) { + dprintf("tap: could not prepend data to mbuf: %d\n", err); + return err; + } + *m = nm; + + /* fill the header */ + eh = (struct ether_header *) mbuf_data(*m); + memcpy(eh->ether_dhost, dest_linkaddr, ETHER_ADDR_LEN); + ifnet_lladdr_copy_bytes(ifp, eh->ether_shost, ETHER_ADDR_LEN); + eh->ether_type = *((u_int16_t *) frame_type); + + return 0; +} + +errno_t +tap_interface::if_add_proto(protocol_family_t proto, const struct ifnet_demux_desc *desc, + u_int32_t ndesc) +{ + errno_t err; + + dprintf("tap: if_add_proto proto %d\n", proto); + + for (unsigned int i = 0; i < ndesc; i++) { + /* try to add the protocol */ + err = add_one_proto(proto, desc[i]); + if (err != 0) { + /* if that fails, remove everything stored so far */ + if_del_proto(proto); + return err; + } + } + + return 0; +} + +errno_t +tap_interface::if_del_proto(protocol_family_t proto) +{ + dprintf("tap: if_del_proto proto %d\n", proto); + + /* delete all matching entries in attached_protos */ + for (unsigned int i = 0; i < MAX_ATTACHED_PROTOS; i++) { + if (attached_protos[i].proto == proto) + attached_protos[i].used = false; + } + + return 0; +} + +errno_t +tap_interface::if_check_multi(const struct sockaddr *maddr) +{ + dprintf("tap: if_check_multi family %d\n", maddr->sa_family); + + /* see whether it is a ethernet address with the multicast bit set */ + if (maddr->sa_family == AF_LINK) { + struct sockaddr_dl *dlmaddr = (struct sockaddr_dl *) maddr; + if (LLADDR(dlmaddr)[0] & 0x01) + return 0; + else + return EADDRNOTAVAIL; + } + + return EOPNOTSUPP; +} + +errno_t +tap_interface::add_one_proto(protocol_family_t proto, const struct ifnet_demux_desc &dd) +{ + int free = -1; + u_int16_t dt; + + /* we only support DLIL_DESC_ETYPE2 */ + if (dd.type != DLIL_DESC_ETYPE2 || dd.datalen != 2) { + log(LOG_WARNING, "tap: tap only supports DLIL_DESC_ETYPE2 protocols.\n"); + return EINVAL; + } + + dt = *((u_int16_t *) (dd.data)); + + /* see if the protocol is already registered */ + for (unsigned int i = 0; i < MAX_ATTACHED_PROTOS; i++) { + if (attached_protos[i].used) { + if (dt == attached_protos[i].type) { + /* already registered */ + if (attached_protos[i].proto == proto) { + /* matches the old entry */ + return 0; + } else + return EEXIST; + } + } else if (free == -1) + free = i; + } + + /* did we find a free entry? */ + if (free == -1) + /* is ENOBUFS correct? */ + return ENOBUFS; + + /* ok, save information */ + attached_protos[free].used = true; + attached_protos[free].type = dt; + attached_protos[free].proto = proto; + + return 0; +} + +/* This code is shamelessly stolen from if_bond.c */ +void +tap_interface::generate_link_event(u_int32_t code) +{ + struct { + struct kern_event_msg header; + u_int32_t unit; + char if_name[IFNAMSIZ]; + } event; + + bzero(&event, sizeof(event)); + event.header.total_size = sizeof(event); + event.header.vendor_code = KEV_VENDOR_APPLE; + event.header.kev_class = KEV_NETWORK_CLASS; + event.header.kev_subclass = KEV_DL_SUBCLASS; + event.header.event_code = code; + event.header.event_data[0] = family; + event.unit = (u_int32_t) unit; + strncpy(event.if_name, ifnet_name(ifp), IFNAMSIZ); + + ifnet_event(ifp, &event.header); +} + +/* tap_manager members */ +tuntap_interface * +tap_manager::create_interface() +{ + return new tap_interface(); +} + diff --git a/ext/tap-mac/tuntap/src/tap/tap.h b/ext/tap-mac/tuntap/src/tap/tap.h new file mode 100644 index 00000000..0ef9e159 --- /dev/null +++ b/ext/tap-mac/tuntap/src/tap/tap.h @@ -0,0 +1,103 @@ +/* + * ethertap device for MacOSX. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#ifndef __TAP_H__ +#define __TAP_H__ + +#include "tuntap.h" + +#define TAP_FAMILY_NAME ((char *) "zt") + +#define TAP_IF_COUNT 16 /* max number of tap interfaces */ + +#define TAP_MTU 2800 + +#define TAP_LLADDR tap_lladdr + +/* the mac address of our interfaces. note that the last byte will be replaced by the unit number */ +extern u_char tap_lladdr[]; + +/* tap manager */ +class tap_manager : public tuntap_manager { + + protected: + /* just define the interface creation method */ + virtual tuntap_interface *create_interface(); + +}; + +/* the tap network interface */ +class tap_interface : public tuntap_interface { + + protected: + /* maximum number of protocols that can be attached */ + static const unsigned int MAX_ATTACHED_PROTOS = 8; + + /* information about attached protocols for demuxing is stored here */ + struct { + /* whether this entry is used */ + bool used; + /* type in the ethernet header */ + u_int16_t type; + /* protocol passed to add_proto */ + protocol_family_t proto; + } attached_protos[MAX_ATTACHED_PROTOS]; + + /* initializes the interface */ + virtual bool initialize(unsigned short major, unsigned short unit); + + /* shuts the interface down */ + virtual void shutdown(); + + /* called when the character device is opened in order to intialize the network + * interface. + */ + virtual int initialize_interface(); + /* called when the character device is closed to shutdown the network interface */ + virtual void shutdown_interface(); + + /* override interface routines */ + virtual errno_t if_ioctl(u_int32_t cmd, void *arg); + virtual errno_t if_demux(mbuf_t m, char *header, protocol_family_t *proto); + virtual errno_t if_framer(mbuf_t *m, const struct sockaddr *dest, + const char *dest_linkaddr, const char *frame_type); + virtual errno_t if_add_proto(protocol_family_t proto, + const struct ifnet_demux_desc *ddesc, u_int32_t ndesc); + virtual errno_t if_del_proto(protocol_family_t proto); + virtual errno_t if_check_multi(const struct sockaddr *maddr); + + /* if_add_proto helper */ + errno_t add_one_proto(protocol_family_t proto, const struct ifnet_demux_desc &dd); + + /* generates a kernel event */ + void generate_link_event(u_int32_t code); + + friend class tap_manager; +}; + +#endif /* __TAP_H__ */ + diff --git a/ext/tap-mac/tuntap/src/tuntap.cc b/ext/tap-mac/tuntap/src/tuntap.cc new file mode 100644 index 00000000..7fdbb795 --- /dev/null +++ b/ext/tap-mac/tuntap/src/tuntap.cc @@ -0,0 +1,966 @@ +/* + * ip tunnel/ethertap device for MacOSX. Common functionality of tap_interface and tun_interface. + * + * tuntap_interface class definition + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#include "tuntap.h" + +#if 0 +#define dprintf(...) log(LOG_INFO, __VA_ARGS__) +#else +#define dprintf(...) +#endif + +extern "C" { + +#include <sys/conf.h> +#include <sys/syslog.h> +#include <sys/param.h> +#include <sys/filio.h> +#include <sys/sockio.h> +#include <sys/fcntl.h> +#include <sys/kpi_socket.h> + +#include <vm/vm_kern.h> + +#include <net/if_types.h> +#include <net/if_var.h> +#include <net/if_dl.h> +#include <net/if_arp.h> + +#include <miscfs/devfs/devfs.h> + +} + +extern "C" { + +/* interface service functions that delegate to the appropriate tuntap_interface instance */ +errno_t +tuntap_if_output(ifnet_t ifp, mbuf_t m) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_output(m); + } + + if (m != NULL) + mbuf_freem_list(m); + + return ENODEV; +} + +errno_t +tuntap_if_ioctl(ifnet_t ifp, long unsigned int cmd, void *arg) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_ioctl(cmd, arg); + } + + return ENODEV; +} + +errno_t +tuntap_if_set_bpf_tap(ifnet_t ifp, bpf_tap_mode mode, int (*cb)(ifnet_t, mbuf_t)) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_set_bpf_tap(mode, cb); + } + + return ENODEV; +} + +errno_t +tuntap_if_demux(ifnet_t ifp, mbuf_t m, char *header, protocol_family_t *proto) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_demux(m, header, proto); + } + + return ENODEV; +} + +errno_t +tuntap_if_framer(ifnet_t ifp, mbuf_t *m, const struct sockaddr *dest, const char *dest_linkaddr, + const char *frame_type) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_framer(m, dest, dest_linkaddr, frame_type); + } + + return ENODEV; +} + +errno_t +tuntap_if_add_proto(ifnet_t ifp, protocol_family_t proto, const struct ifnet_demux_desc *ddesc, + u_int32_t ndesc) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_add_proto(proto, ddesc, ndesc); + } + + return ENODEV; +} + +errno_t +tuntap_if_del_proto(ifnet_t ifp, protocol_family_t proto) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_del_proto(proto); + } + + return ENODEV; +} + +errno_t +tuntap_if_check_multi(ifnet_t ifp, const struct sockaddr* maddr) +{ + if (ifp != NULL) + { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + return ttif->if_check_multi(maddr); + } + + return ENODEV; +} + +void +tuntap_if_detached(ifnet_t ifp) +{ + if (ifp != NULL) { + tuntap_interface *ttif = (tuntap_interface *) ifnet_softc(ifp); + if (ttif != NULL) + ttif->if_detached(); + } +} + +errno_t +tuntap_if_noop_output(ifnet_t, mbuf_t) +{ + return ENODEV; +} + +errno_t +tuntap_if_noop_demux(ifnet_t, mbuf_t, char*, protocol_family_t*) +{ + return ENODEV; +} + +errno_t +tuntap_if_noop_add_proto(ifnet_t, protocol_family_t, const struct ifnet_demux_desc*, u_int32_t) +{ + return ENODEV; +} + +errno_t +tuntap_if_noop_del_proto(ifnet_t, protocol_family_t) +{ + return ENODEV; +} + +} /* extern "C" */ + +/* tuntap_mbuf_queue */ +tuntap_mbuf_queue::tuntap_mbuf_queue() +{ + head = tail = NULL; + size = 0; +} + +tuntap_mbuf_queue::~tuntap_mbuf_queue() +{ + clear(); +} + +bool +tuntap_mbuf_queue::enqueue(mbuf_t mb) +{ + if (size == QUEUE_SIZE) + return false; + + mbuf_setnextpkt(mb, NULL); + + if (head == NULL) + head = tail = mb; + else { + mbuf_setnextpkt(tail, mb); + tail = mb; + } + size++; + + return true; +} + +mbuf_t +tuntap_mbuf_queue::dequeue() +{ + mbuf_t ret; + + /* check wether there is a packet in the queue */ + if (head == NULL) + return NULL; + + /* fetch it */ + ret = head; + head = mbuf_nextpkt(head); + mbuf_setnextpkt(ret, NULL); + size--; + + return ret; +} + +void +tuntap_mbuf_queue::clear() +{ + /* free mbufs that are in the queue */ + if (head != NULL) + mbuf_freem_list(head); + + head = NULL; + tail = NULL; + size = 0; +} + +/* tuntap_interface members */ +tuntap_interface::tuntap_interface() +{ + /* initialize the members */ + ifp = NULL; + open = false; + block_io = true; + dev_handle = NULL; + pid = 0; + selthreadclear(&rsel); + bpf_mode = BPF_MODE_DISABLED; + bpf_callback = NULL; + bzero(unique_id, UIDLEN); + in_ioctl = false; +} + +tuntap_interface::~tuntap_interface() +{ +} + +bool +tuntap_interface::register_chardev(unsigned short major) +{ + /* register character device */ + dev_handle = devfs_make_node(makedev(major, unit), DEVFS_CHAR, 0, 0, 0660, "%s%d", + family_name, (int) unit); + + if (dev_handle == NULL) { + log(LOG_ERR, "tuntap: could not make /dev/%s%d\n", family_name, (int) unit); + return false; + } + + return true; +} + +void +tuntap_interface::unregister_chardev() +{ + dprintf("unregistering character device\n"); + + /* unregister character device */ + if (dev_handle != NULL) + devfs_remove(dev_handle); + dev_handle = NULL; +} + +bool +tuntap_interface::register_interface(const struct sockaddr_dl* lladdr, void *bcaddr, + u_int32_t bcaddrlen) +{ + struct ifnet_init_params ip; + errno_t err; + + dprintf("register_interface\n"); + + /* initialize an initialization info struct */ + ip.uniqueid_len = UIDLEN; + ip.uniqueid = unique_id; + ip.name = family_name; + ip.unit = unit; + ip.family = family; + ip.type = type; + ip.output = tuntap_if_output; + ip.demux = tuntap_if_demux; + ip.add_proto = tuntap_if_add_proto; + ip.del_proto = tuntap_if_del_proto; + ip.check_multi = tuntap_if_check_multi; + ip.framer = tuntap_if_framer; + ip.softc = this; + ip.ioctl = tuntap_if_ioctl; + ip.set_bpf_tap = tuntap_if_set_bpf_tap; + ip.detach = tuntap_if_detached; + ip.event = NULL; + ip.broadcast_addr = bcaddr; + ip.broadcast_len = bcaddrlen; + + dprintf("tuntap: tuntap_if_check_multi is at 0x%08x\n", (void*) tuntap_if_check_multi); + + /* allocate the interface */ + err = ifnet_allocate(&ip, &ifp); + if (err) { + log(LOG_ERR, "tuntap: could not allocate interface for %s%d: %d\n", family_name, + (int) unit, err); + ifp = NULL; + return false; + } + + /* activate the interface */ + err = ifnet_attach(ifp, lladdr); + if (err) { + log(LOG_ERR, "tuntap: could not attach interface %s%d: %d\n", family_name, + (int) unit, err); + ifnet_release(ifp); + ifp = NULL; + return false; + } + + dprintf("setting interface flags\n"); + + /* set interface flags */ + ifnet_set_flags(ifp, IFF_RUNNING | IFF_MULTICAST | IFF_SIMPLEX, (u_int16_t) ~0UL); + + dprintf("flags: %x\n", ifnet_flags(ifp)); + + return true; +} + +void +tuntap_interface::unregister_interface() +{ + errno_t err; + + dprintf("unregistering network interface\n"); + + if (ifp != NULL) { + interface_detached = false; + + /* detach interface */ + err = ifnet_detach(ifp); + if (err) + log(LOG_ERR, "tuntap: error detaching interface %s%d: %d\n", + family_name, unit, err); + + dprintf("interface detaching\n"); + + /* Wait until the interface has completely been detached. */ + detach_lock.lock(); + while (!interface_detached) + detach_lock.sleep(&interface_detached); + detach_lock.unlock(); + + dprintf("interface detached\n"); + + /* release the interface */ + ifnet_release(ifp); + + ifp = NULL; + } + + dprintf("network interface unregistered\n"); +} + +void +tuntap_interface::cleanup_interface() +{ + errno_t err; + ifaddr_t *addrs; + ifaddr_t *a; + struct ifreq ifr; + + /* mark the interface down */ + ifnet_set_flags(ifp, 0, IFF_UP | IFF_RUNNING); + + /* Unregister all interface addresses. This works around a deficiency in the Darwin kernel. + * If we don't remove all IP addresses that are attached to the interface it can happen that + * the IP code fails to clean them up itself. When the interface is recycled, the IP code + * might then think some addresses are still attached to the interface... + */ + + err = ifnet_get_address_list(ifp, &addrs); + if (!err) { + + /* Execute a SIOCDIFADDR ioctl for each address. For technical reasons, we can only + * do that with a socket of the appropriate family. So try to create a dummy socket. + * I know this is a little expensive, but better than crashing... + * + * This really sucks. + */ + for (a = addrs; *a != NULL; a++) { + /* initialize the request parameters */ + snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s%d", + ifnet_name(ifp), ifnet_unit(ifp)); + ifaddr_address(*a, &(ifr.ifr_addr), sizeof(ifr.ifr_addr)); + if (ifr.ifr_addr.sa_family != AF_INET) + continue; + + dprintf("trying to delete address of family %d\n", ifr.ifr_addr.sa_family); + + do_sock_ioctl(ifr.ifr_addr.sa_family, SIOCDIFADDR, &ifr); + } + + /* release the address list */ + ifnet_free_address_list(addrs); + } +} + +bool +tuntap_interface::idle() +{ + return !(open); +} + +void +tuntap_interface::notify_bpf(mbuf_t mb, bool out) +{ + auto_lock l(&bpf_lock); + + if ((out && bpf_mode == BPF_MODE_OUTPUT) + || (!out && bpf_mode == BPF_MODE_INPUT) + || (bpf_mode == BPF_MODE_INPUT_OUTPUT)) + (*bpf_callback)(ifp, mb); +} + +void +tuntap_interface::do_sock_ioctl(sa_family_t af, unsigned long cmd, void* arg) { + if (in_ioctl) { + log(LOG_ERR, "tuntap: ioctl recursion detected, aborting.\n"); + return; + } + + socket_t sock; + errno_t err = sock_socket(af, SOCK_RAW, 0, NULL, NULL, &sock); + if (err) { + log(LOG_ERR, "tuntap: failed to create socket: %d\n", err); + return; + } + + in_ioctl = true; + + /* issue the ioctl */ + err = sock_ioctl(sock, cmd, arg); + if (err) + log(LOG_ERR, "tuntap: socket ioctl %d failed: %d\n", cmd, err); + + in_ioctl = false; + + /* get rid of the socket */ + sock_close(sock); +} + +/* character device service methods */ +int +tuntap_interface::cdev_open(int flags, int devtype, proc_t p) +{ + dprintf("tuntap: cdev_open()\n"); + + /* grab the lock so that there can only be one thread inside */ + auto_lock l(&lock); + + /* check wether it is already open */ + if (open) + return EBUSY; + + /* bring the network interface up */ + int error = initialize_interface(); + if (error) + return error; + + open = true; + pid = proc_pid(p); + + return 0; +} + +int +tuntap_interface::cdev_close(int flags, int devtype, proc_t p) +{ + dprintf("tuntap: cdev_close()\n"); + + auto_lock l(&lock); + + if (open) { + open = false; + + /* shut down the network interface */ + shutdown_interface(); + + /* clear the queue */ + send_queue.clear(); + + /* wakeup the cdev thread and notify selects */ + wakeup(this); + selwakeup(&rsel); + + return 0; + } + + return EBADF; +} + +int +tuntap_interface::cdev_read(uio_t uio, int ioflag) +{ + auto_lock l(&lock); + + unsigned int nb = 0; + int error; + + dprintf("tuntap: cdev read\n"); + + if (!open || ifp == NULL || !(ifnet_flags(ifp) & IFF_UP)) + return EIO; + + /* fetch a new mbuf from the queue if necessary */ + mbuf_t cur_mbuf = NULL; + while (cur_mbuf == NULL) { + dprintf("tuntap: fetching new mbuf\n"); + + cur_mbuf = send_queue.dequeue(); + if (cur_mbuf == NULL) { + /* nothing in queue, block or return */ + if (!block_io) { + dprintf("tuntap: aborting (nbio)\n"); + return EWOULDBLOCK; + } else { + /* block */ + dprintf("tuntap: waiting\n"); + /* release the lock while waiting */ + l.unlock(); + error = msleep(this, NULL, PZERO | PCATCH, "tuntap", NULL); + + l.lock(); + + if (error) + return error; + + /* see whether the device was closed in the meantime */ + if (!open || ifp == NULL || !(ifnet_flags(ifp) & IFF_UP)) + return EIO; + + } + } + } + + /* notify bpf */ + notify_bpf(cur_mbuf, true); + + /* output what we have */ + do { + dprintf("tuntap: got new mbuf: %p uio_resid: %d\n", cur_mbuf, uio_resid(uio)); + + /* now we have an mbuf */ + int chunk_len = min(mbuf_len(cur_mbuf), uio_resid(uio)); + error = uiomove((char *) mbuf_data(cur_mbuf), chunk_len, uio); + if (error) { + mbuf_freem(cur_mbuf); + return error; + } + nb += chunk_len; + + dprintf("tuntap: moved %d bytes to userspace uio_resid: %d\n", chunk_len, + uio_resid(uio)); + + /* update cur_mbuf */ + cur_mbuf = mbuf_free(cur_mbuf); + + } while (uio_resid(uio) > 0 && cur_mbuf != NULL); + + /* update statistics */ + ifnet_stat_increment_out(ifp, 1, nb, 0); + + /* still data left? forget about that ;-) */ + if (cur_mbuf != NULL) + mbuf_freem(cur_mbuf); + + dprintf("tuntap: read done\n"); + + return 0; +} + +int +tuntap_interface::cdev_write(uio_t uio, int ioflag) +{ + auto_lock l(&lock); + + if (!open || ifp == NULL || !(ifnet_flags(ifp) & IFF_UP)) + return EIO; + + dprintf("tuntap: cdev write. uio_resid: %d\n", uio_resid(uio)); + + /* pack the data into an mbuf chain */ + mbuf_t first, mb; + + /* first we need an mbuf having a header */ + mbuf_gethdr(MBUF_WAITOK, MBUF_TYPE_DATA, &first); + if (first == NULL) { + log(LOG_ERR, "tuntap: could not get mbuf.\n"); + return ENOMEM; + } + mbuf_setlen(first, 0); + + unsigned int mlen = mbuf_maxlen(first); + unsigned int chunk_len; + unsigned int copied = 0; + int error; + + /* stuff the data into the mbuf(s) */ + mb = first; + while (uio_resid(uio) > 0) { + /* copy a chunk. enforce mtu (don't know if this is correct behaviour) */ + // ... evidently not :) -- Adam Ierymenko <adam.ierymenko@zerotier.com> + //chunk_len = min(ifnet_mtu(ifp), min(uio_resid(uio), mlen)); + chunk_len = min(uio_resid(uio),mlen); + error = uiomove((caddr_t) mbuf_data(mb), chunk_len, uio); + if (error) { + log(LOG_ERR, "tuntap: could not copy data from userspace: %d\n", error); + mbuf_freem(first); + return error; + } + + dprintf("tuntap: copied %d bytes, uio_resid %d\n", chunk_len, + uio_resid(uio)); + + mlen -= chunk_len; + mbuf_setlen(mb, mbuf_len(mb) + chunk_len); + copied += chunk_len; + + /* if done, break the loop */ + //if (uio_resid(uio) <= 0 || copied >= ifnet_mtu(ifp)) + // break; + if (uio_resid(uio) <= 0) + break; + + /* allocate a new mbuf if the current is filled */ + if (mlen == 0) { + mbuf_t next; + mbuf_get(MBUF_WAITOK, MBUF_TYPE_DATA, &next); + if (next == NULL) { + log(LOG_ERR, "tuntap: could not get mbuf.\n"); + mbuf_freem(first); + return ENOMEM; + } + mbuf_setnext(mb, next); + mb = next; + mbuf_setlen(mb, 0); + mlen = mbuf_maxlen(mb); + } + } + + /* fill in header info */ + mbuf_pkthdr_setrcvif(first, ifp); + mbuf_pkthdr_setlen(first, copied); + mbuf_pkthdr_setheader(first, mbuf_data(first)); + mbuf_set_csum_performed(first, 0, 0); + + /* update statistics */ + ifnet_stat_increment_in(ifp, 1, copied, 0); + + dprintf("tuntap: mbuf chain constructed. first: %p mb: %p len: %d data: %p\n", + first, mb, mbuf_len(first), mbuf_data(first)); + + /* notify bpf */ + notify_bpf(first, false); + + /* need to adjust the data pointer to point directly behind the linklevel header. The header + * itself is later accessed via m_pkthdr.header. Well, if something is ugly, here is it. + */ + mbuf_adj(first, ifnet_hdrlen(ifp)); + + /* pass the packet over to the network stack */ + error = ifnet_input(ifp, first, NULL); + + if (error) { + log(LOG_ERR, "tuntap: could not input packet into network stack.\n"); + mbuf_freem(first); + return error; + } + + return 0; +} + +int +tuntap_interface::cdev_ioctl(u_long cmd, caddr_t data, int fflag, proc_t p) +{ + auto_lock l(&lock); + + dprintf("tuntap: cdev ioctl: %d\n", (int) (cmd & 0xff)); + + switch (cmd) { + case FIONBIO: + /* set i/o mode */ + block_io = *((int *) data) ? false : true; + return 0; + case FIOASYNC: + /* don't allow switching it on */ + if (*((int *) data)) + return ENOTTY; + return 0; + } + + return ENOTTY; +} + +int +tuntap_interface::cdev_select(int which, void *wql, proc_t p) +{ + auto_lock l(&lock); + + int ret = 0; + + dprintf("tuntap: select. which: %d\n", which); + + switch (which) { + case FREAD: + /* check wether data is available */ + { + if (!send_queue.empty()) + ret = 1; + else { + dprintf("tuntap: select: waiting\n"); + selrecord(p, &rsel, wql); + } + } + break; + case FWRITE: + /* we are always writeable */ + ret = 1; + } + + return ret; +} + +/* interface service methods */ +errno_t +tuntap_interface::if_output(mbuf_t m) +{ + mbuf_t pkt; + + dprintf("tuntap: if output\n"); + + /* just to be sure */ + if (m == NULL) + return 0; + + if (!open || ifp == NULL || !(ifnet_flags(ifp) & IFF_UP)) { + mbuf_freem_list(m); + return EHOSTDOWN; + } + + /* check whether packet has a header */ + if ((mbuf_flags(m) & MBUF_PKTHDR) == 0) { + log(LOG_ERR, "tuntap: packet to be output has no mbuf header.\n"); + mbuf_freem_list(m); + return EINVAL; + } + + /* put the packet(s) into the output queue */ + while (m != NULL) { + /* keep pointer, iterate */ + pkt = m; + m = mbuf_nextpkt(m); + mbuf_setnextpkt(pkt, NULL); + + auto_lock l(&lock); + + if (!send_queue.enqueue(pkt)) { + mbuf_freem(pkt); + mbuf_freem_list(m); + return ENOBUFS; + } + } + + /* protect the wakeup calls with the lock, not sure they are safe. */ + { + auto_lock l(&lock); + + /* wakeup the cdev thread and notify selects */ + wakeup(this); + selwakeup(&rsel); + } + + return 0; +} + +errno_t +tuntap_interface::if_ioctl(u_int32_t cmd, void *arg) +{ + dprintf("tuntap: if ioctl: %d\n", (int) (cmd & 0xff)); + + switch (cmd) { + case SIOCSIFADDR: + { + dprintf("tuntap: if_ioctl: SIOCSIFADDR\n"); + + /* Unfortunately, ifconfig sets the address family field of an INET + * netmask to zero, which makes early mDNSresponder versions ignore + * the interface. Fix that here. This one is of the category "ugly + * workaround". Dumb Darwin... + * + * Meanwhile, Apple has fixed mDNSResponder, and recent versions of + * Leopard don't need this hack anymore. However, Tiger still has a + * broken version so we leave the hack in for now. + * + * TODO: Revisit when dropping Tiger support. + * + * Btw. If you configure other network interfaces using ifconfig, + * you run into the same problem. I still don't know how to make the + * tap devices show up in the network configuration panel... + */ + ifaddr_t ifa = (ifaddr_t) arg; + if (ifa == NULL) + return 0; + + sa_family_t af = ifaddr_address_family(ifa); + if (af != AF_INET) + return 0; + + struct ifaliasreq ifra; + int sa_size = sizeof(struct sockaddr); + if (ifaddr_address(ifa, &ifra.ifra_addr, sa_size) + || ifaddr_dstaddress(ifa, &ifra.ifra_broadaddr, sa_size) + || ifaddr_netmask(ifa, &ifra.ifra_mask, sa_size)) { + log(LOG_WARNING, + "tuntap: failed to parse interface address.\n"); + return 0; + } + + // Check that the address family fields match. If not, issue another + // SIOCAIFADDR to fix the entry. + if (ifra.ifra_addr.sa_family != af + || ifra.ifra_broadaddr.sa_family != af + || ifra.ifra_mask.sa_family != af) { + log(LOG_INFO, "tuntap: Fixing address family for %s%d\n", + family_name, unit); + + snprintf(ifra.ifra_name, sizeof(ifra.ifra_name), "%s%d", + family_name, unit); + ifra.ifra_addr.sa_family = af; + ifra.ifra_broadaddr.sa_family = af; + ifra.ifra_mask.sa_family = af; + + do_sock_ioctl(af, SIOCAIFADDR, &ifra); + } + + return 0; + } + + case SIOCSIFFLAGS: + return 0; + + case SIOCGIFSTATUS: + { + struct ifstat *stat = (struct ifstat *) arg; + int len; + char *p; + + if (stat == NULL) + return EINVAL; + + /* print status */ + len = strlen(stat->ascii); + p = stat->ascii + len; + if (open) { + snprintf(p, IFSTATMAX - len, "\topen (pid %u)\n", pid); + } else { + snprintf(p, IFSTATMAX - len, "\tclosed\n"); + } + + return 0; + } + + case SIOCSIFMTU: + { + struct ifreq *ifr = (struct ifreq *) arg; + + if (ifr == NULL) + return EINVAL; + + ifnet_set_mtu(ifp, ifr->ifr_mtu); + + return 0; + } + + case SIOCDIFADDR: + return 0; + + } + + return EOPNOTSUPP; +} + +errno_t +tuntap_interface::if_set_bpf_tap(bpf_tap_mode mode, int (*cb)(ifnet_t, mbuf_t)) +{ + dprintf("tuntap: mode %d\n", mode); + + auto_lock l(&bpf_lock); + + bpf_callback = cb; + bpf_mode = mode; + + return 0; +} + +errno_t +tuntap_interface::if_check_multi(const struct sockaddr *maddr) +{ + dprintf("tuntap: if_check_multi\n"); + + return EOPNOTSUPP; +} + +void +tuntap_interface::if_detached() +{ + dprintf("tuntap: if_detached\n"); + + /* wake unregister_interface() */ + detach_lock.lock(); + interface_detached = true; + detach_lock.wakeup(&interface_detached); + detach_lock.unlock(); + + dprintf("if_detached done\n"); +} + diff --git a/ext/tap-mac/tuntap/src/tuntap.h b/ext/tap-mac/tuntap/src/tuntap.h new file mode 100644 index 00000000..f10d4a06 --- /dev/null +++ b/ext/tap-mac/tuntap/src/tuntap.h @@ -0,0 +1,301 @@ +/* + * ip tunnel/ethertap device for MacOSX. + * + * The class tuntaptap_interface contains the common functionality of tuntap_interface and + * tap_interface. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#ifndef __TUNTAP_H__ +#define __TUNTAP_H__ + +#include "util.h" +#include "lock.h" + +extern "C" { + +#include <sys/types.h> +#include <sys/socket.h> +#include <sys/select.h> +#include <sys/systm.h> +#include <sys/kpi_mbuf.h> + +#include <kern/locks.h> + +#include <net/if.h> +#include <net/bpf.h> +#include <net/kpi_interface.h> + +} + +extern "C" { + +errno_t tuntap_if_output(ifnet_t ifp, mbuf_t m); +errno_t tuntap_if_ioctl(ifnet_t ifp, long unsigned int cmd, void *arg); +errno_t tuntap_if_set_bpf_tap(ifnet_t ifp, bpf_tap_mode mode, int (*cb)(ifnet_t, mbuf_t)); +errno_t tuntap_if_demux(ifnet_t ifp, mbuf_t m, char *header, protocol_family_t *proto); +errno_t tuntap_if_framer(ifnet_t ifp, mbuf_t *m, const struct sockaddr *dest, + const char *dest_linkaddr, const char *frame_type); +errno_t tuntap_if_add_proto(ifnet_t ifp, protocol_family_t proto, + const struct ifnet_demux_desc *ddesc, u_int32_t ndesc); +errno_t tuntap_if_del_proto(ifnet_t ifp, protocol_family_t proto); +errno_t tuntap_if_check_multi(ifnet_t ifp, const struct sockaddr *maddr); +void tuntap_if_detached(ifnet_t ifp); + +} + +/* forward declaration */ +class tuntap_interface; + +/* both interface families have their manager object that will create, initialize, shutdown and + * delete interfaces. This is (mostly) generic so it can be used both for tun and tap. The only + * exception is the interface creation, therefore this class is abstract. tun and tap have their own + * versions that simply fill in create_interface(). + */ +class tuntap_manager { + + protected: + /* manager cdev gate */ + tt_gate cdev_gate; + /* interface count */ + unsigned int count; + /* an array holding all the interface instances */ + tuntap_interface **tuntaps; + /* the major device number */ + int dev_major; + /* family name */ + char *family; + + /* wether static members are initialized */ + static bool statics_initialized; + + /* major-to-manager-map */ + static const int MAX_CDEV = 256; + static tuntap_manager *mgr_map[MAX_CDEV]; + + /* initializes static members */ + void initialize_statics(); + + public: + /* sets major device number, allocates the interface table. */ + bool initialize(unsigned int count, char *family); + + /* tries to shutdown the family. returns true if successful. the manager object may + * not be deleted if this wasn't called successfully. + */ + bool shutdown(); + + /* the destructor deletes allocated memory and unregisters the character device + * switch */ + virtual ~tuntap_manager(); + + /* here are the cdev routines for the class. They will figure out the manager object + * and call the service methods declared below. + */ + static int cdev_open(dev_t dev, int flags, int devtype, proc_t p); + static int cdev_close(dev_t dev, int flags, int devtype, proc_t p); + static int cdev_read(dev_t dev, uio_t uio, int ioflag); + static int cdev_write(dev_t dev, uio_t uio, int ioflag); + static int cdev_ioctl(dev_t dev, u_long cmd, caddr_t data, int fflag, + proc_t p); + static int cdev_select(dev_t dev, int which, void *wql, proc_t p); + + protected: + /* Here are the actual service routines that will do the required things (creating + * interfaces and such) and forward to the interface's implementation. + */ + int do_cdev_open(dev_t dev, int flags, int devtype, proc_t p); + int do_cdev_close(dev_t dev, int flags, int devtype, proc_t p); + int do_cdev_read(dev_t dev, uio_t uio, int ioflag); + int do_cdev_write(dev_t dev, uio_t uio, int ioflag); + int do_cdev_ioctl(dev_t dev, u_long cmd, caddr_t data, int fflag, proc_t p); + int do_cdev_select(dev_t dev, int which, void *wql, proc_t p); + + /* abstract method that will create an interface. Implemented by tun and tap */ + virtual tuntap_interface *create_interface() = 0; + + /* makes sure there is one idle interface available (if nothing fails */ + void ensure_idle_device(); + +}; + +/* a class implementing a mbuf packet queue. On Darwin 7 we had struct ifqueue, but that is now + * internal to the kernel for Darwin 8. So lets have our own. + */ +class tuntap_mbuf_queue { + + private: + /* output end of the queue. dequeueing takes mbufs from here */ + mbuf_t head; + /* input end. new mbufs are appended here. */ + mbuf_t tail; + + /* size */ + unsigned int size; + + /* maximum queue size */ + static const unsigned int QUEUE_SIZE = 128; + + public: + /* initialize new empty queue */ + tuntap_mbuf_queue(); + ~tuntap_mbuf_queue(); + + /* is the queue full? */ + bool full() { return size == QUEUE_SIZE; } + /* is it emtpy? */ + bool empty() { return size == 0; } + + /* enqueue an mbuf. returns true if there was space left, so the mbuf could be + * queued, false otherwise */ + bool enqueue(mbuf_t mb); + + /* tries to dequeue the next mbuf. If the queue is empty, NULL is returned */ + mbuf_t dequeue(); + + /* makes the queue empty, discarding any queue packets */ + void clear(); +}; + +class tuntap_interface { + + protected: + /* interface number */ + unsigned int unit; + /* family name */ + char *family_name; + /* family identifier */ + ifnet_family_t family; + /* interface type */ + u_int32_t type; + /* id string */ + static const unsigned int UIDLEN = 20; + char unique_id[UIDLEN]; + + /* synchronization */ + tt_mutex lock; + tt_mutex bpf_lock; + tt_mutex detach_lock; + + /* the interface structure registered */ + ifnet_t ifp; + /* whether the device has been opened */ + bool open; + /* whether we are doing blocking i/o */ + bool block_io; + /* whether the interface has properly been detached */ + bool interface_detached; + /* handle to the devfs node for the character device */ + void *dev_handle; + /* the pid of the process that opened the cdev, if any */ + pid_t pid; + /* read select info */ + struct selinfo rsel; + /* bpf mode, wether filtering is on or off */ + bpf_tap_mode bpf_mode; + /* bpf callback. called when packet arrives/leaves */ + int (*bpf_callback)(ifnet_t, mbuf_t); + /* pending packets queue (for output), must be accessed with the lock held */ + tuntap_mbuf_queue send_queue; + /* whether an ioctl that we issued is currently being processed */ + bool in_ioctl; + + /* protected constructor. initializes most of the members */ + tuntap_interface(); + virtual ~tuntap_interface(); + + /* initialize the device */ + virtual bool initialize(unsigned short major, unsigned short unit) = 0; + + /* character device management */ + virtual bool register_chardev(unsigned short major); + virtual void unregister_chardev(); + + /* network interface management */ + virtual bool register_interface(const struct sockaddr_dl *lladdr, + void *bcaddr, u_int32_t bcaddrlen); + virtual void unregister_interface(); + virtual void cleanup_interface(); + + /* called when the character device is opened in order to intialize the network + * interface. + */ + virtual int initialize_interface() = 0; + /* called when the character device is closed to shutdown the network interface */ + virtual void shutdown_interface() = 0; + + /* check wether the interface is idle (so it can be brought down) */ + virtual bool idle(); + + /* shut it down */ + virtual void shutdown() = 0; + + /* notifies BPF of a packet coming through */ + virtual void notify_bpf(mbuf_t mb, bool out); + + /* executes a socket ioctl through a temporary socket */ + virtual void do_sock_ioctl(sa_family_t af, unsigned long cmd, void* arg); + + /* character device service methods. Called by the manager */ + virtual int cdev_open(int flags, int devtype, proc_t p); + virtual int cdev_close(int flags, int devtype, proc_t p); + virtual int cdev_read(uio_t uio, int ioflag); + virtual int cdev_write(uio_t uio, int ioflag); + virtual int cdev_ioctl(u_long cmd, caddr_t data, int fflag, proc_t p); + virtual int cdev_select(int which, void *wql, proc_t p); + + /* interface functions. friends and implementation methods */ + friend errno_t tuntap_if_output(ifnet_t ifp, mbuf_t m); + friend errno_t tuntap_if_ioctl(ifnet_t ifp, long unsigned int cmd, void *arg); + friend errno_t tuntap_if_set_bpf_tap(ifnet_t ifp, bpf_tap_mode mode, + int (*cb)(ifnet_t, mbuf_t)); + friend errno_t tuntap_if_demux(ifnet_t ifp, mbuf_t m, char *header, + protocol_family_t *proto); + friend errno_t tuntap_if_framer(ifnet_t ifp, mbuf_t *m, const struct sockaddr *dest, + const char *dest_linkaddr, const char *frame_type); + friend errno_t tuntap_if_add_proto(ifnet_t ifp, protocol_family_t proto, + const struct ifnet_demux_desc *ddesc, u_int32_t ndesc); + friend errno_t tuntap_if_del_proto(ifnet_t ifp, protocol_family_t proto); + friend errno_t tuntap_if_check_multi(ifnet_t ifp, const struct sockaddr *maddr); + friend void tuntap_if_detached(ifnet_t ifp); + + virtual errno_t if_output(mbuf_t m); + virtual errno_t if_ioctl(u_int32_t cmd, void *arg); + virtual errno_t if_set_bpf_tap(bpf_tap_mode mode, int (*cb)(ifnet_t, mbuf_t)); + virtual errno_t if_demux(mbuf_t m, char *header, protocol_family_t *proto) = 0; + virtual errno_t if_framer(mbuf_t *m, const struct sockaddr *dest, + const char *dest_linkaddr, const char *frame_type) = 0; + virtual errno_t if_add_proto(protocol_family_t proto, + const struct ifnet_demux_desc *ddesc, u_int32_t ndesc) = 0; + virtual errno_t if_del_proto(protocol_family_t proto) = 0; + virtual errno_t if_check_multi(const struct sockaddr *maddr); + virtual void if_detached(); + + /* tuntap_manager feeds us with cdev input, so it is our friend */ + friend class tuntap_manager; +}; + +#endif /* __TUNTAP_H__ */ + diff --git a/ext/tap-mac/tuntap/src/tuntap_mgr.cc b/ext/tap-mac/tuntap/src/tuntap_mgr.cc new file mode 100644 index 00000000..f41394e9 --- /dev/null +++ b/ext/tap-mac/tuntap/src/tuntap_mgr.cc @@ -0,0 +1,372 @@ +/* + * ip tunnel/ethertap device for MacOSX. + * + * tuntap_manager definition. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#include "tuntap.h" +#include "mem.h" + +extern "C" { + +#include <sys/conf.h> +#include <sys/param.h> +#include <sys/syslog.h> +#include <sys/systm.h> + +#include <vm/vm_kern.h> + +#include <miscfs/devfs/devfs.h> + +} + +#if 0 +#define dprintf(...) log(LOG_INFO, __VA_ARGS__) +#else +#define dprintf(...) +#endif + +/* cdevsw for tuntap_manager */ +static struct cdevsw mgr_cdevsw = +{ + tuntap_manager::cdev_open, + tuntap_manager::cdev_close, + tuntap_manager::cdev_read, + tuntap_manager::cdev_write, + tuntap_manager::cdev_ioctl, + eno_stop, + eno_reset, + NULL, + tuntap_manager::cdev_select, + eno_mmap, + eno_strat, + eno_getc, + eno_putc, + 0 +}; + +/* tuntap_manager members */ +tuntap_manager *tuntap_manager::mgr_map[MAX_CDEV]; + +bool tuntap_manager::statics_initialized = false; + +/* static initializer */ +void +tuntap_manager::initialize_statics() +{ + dprintf("initializing mgr_map\n"); + + /* initialize the major-to-manager map */ + for (int i = 0; i < MAX_CDEV; i++) + mgr_map[i] = NULL; + + statics_initialized = true; +} + +bool +tuntap_manager::initialize(unsigned int count, char *family) +{ + this->count = count; + this->family = family; + this->tuntaps = NULL; + + if (!statics_initialized) + initialize_statics(); + + /* make sure noone can access the character devices until we are done */ + auto_lock l(&cdev_gate); + + /* register the switch for the tap character devices */ + dev_major = cdevsw_add(-1, &mgr_cdevsw); + if (dev_major == -1) { + log(LOG_ERR, "%s: could not register character device switch.\n", family); + return false; + } + + /* allocate memory for the interface instance table */ + tuntaps = (tuntap_interface **) mem_alloc(count * sizeof(tuntap_interface *)); + if (tuntaps == NULL) + { + log(LOG_ERR, "%s: no memory!\n", family); + return false; + } + + bzero(tuntaps, count * sizeof(tuntap_interface *)); + + /* Create the interfaces. This will only add the character devices. The network devices will + * be created upon open()ing the corresponding character devices. + */ + for (int i = 0; i < (int) count; i++) + { + tuntaps[i] = create_interface(); + + if (tuntaps[i] != NULL) + { + if (tuntaps[i]->initialize(dev_major, i)) + { + continue; + } + + /* error here. current interface needs to be shut down */ + i++; + } + + /* something went wrong. clean up. */ + while (--i >= 0) + { + tuntaps[i]->shutdown(); + delete tuntaps[i]; + } + + return false; + } + + /* register the new family in the mgr switch */ + mgr_map[dev_major] = this; + + log(LOG_INFO, "%s kernel extension version %s <mattias.nissler@gmx.de>\n", + family, TUNTAP_VERSION); + + return true; +} + +bool +tuntap_manager::shutdown() +{ + bool ok = true; + + /* we halt the whole thing while we check whether we can shutdown */ + auto_lock l(&cdev_gate); + + /* anyone in? */ + if (cdev_gate.is_anyone_in()) { + dprintf("tuntap_mgr: won't shutdown, threads still behind the gate."); + ok = false; + } else { + /* query the interfaces to see if shutting down is ok */ + if (tuntaps != NULL) { + for (unsigned int i = 0; i < count; i++) { + if (tuntaps[i] != NULL) + ok &= tuntaps[i]->idle(); + } + + /* if yes, do it now */ + if (ok) { + for (unsigned int i = 0; i < count; i++) { + if (tuntaps[i] != NULL) { + tuntaps[i]->shutdown(); + delete tuntaps[i]; + tuntaps[i] = NULL; + } + } + } + } + } + + /* unregister the character device switch */ + if (ok) { + if (dev_major != -1 && cdevsw_remove(dev_major, &mgr_cdevsw) == -1) { + log(LOG_WARNING, + "%s: character device switch got lost. strange.\n", family); + } + mgr_map[dev_major] = NULL; + dev_major = -1; + + /* at this point there is still a chance that some thread hangs at the cdev_gate in + * one of the cdev service functions. I can't imagine any way that would aviod this. + * So lets unblock the gate such that they fail. + */ + unsigned int old_number; + do { + old_number = cdev_gate.get_ticket_number(); + + dprintf("tuntap_manager: waiting for other threads to give up.\n"); + + /* wait one second */ + cdev_gate.sleep(&cdev_gate, 1000000); + + } while (cdev_gate.get_ticket_number() != old_number); + + /* I hope it is safe to unload now. */ + + } else { + log(LOG_WARNING, "%s: won't unload, at least one interface is busy.\n", family); + } + + dprintf("tuntap manager: shutdown %s\n", ok ? "ok" : "failed"); + + return ok; +} + +tuntap_manager::~tuntap_manager() +{ + dprintf("freeing interface table\n"); + + /* free memory */ + if (tuntaps != NULL) + mem_free(tuntaps, count * sizeof(tuntap_interface *)); +} + +/* service method dispatchers */ +int +tuntap_manager::cdev_open(dev_t dev, int flags, int devtype, proc_t p) +{ + return (mgr_map[major(dev)] == NULL ? ENOENT + : mgr_map[major(dev)]->do_cdev_open(dev, flags, devtype, p)); +} + +int +tuntap_manager::cdev_close(dev_t dev, int flags, int devtype, proc_t p) +{ + return (mgr_map[major(dev)] == NULL ? EBADF + : mgr_map[major(dev)]->do_cdev_close(dev, flags, devtype, p)); +} + +int +tuntap_manager::cdev_read(dev_t dev, uio_t uio, int ioflag) +{ + return (mgr_map[major(dev)] == NULL ? EBADF + : mgr_map[major(dev)]->do_cdev_read(dev, uio, ioflag)); +} + +int +tuntap_manager::cdev_write(dev_t dev, uio_t uio, int ioflag) +{ + return (mgr_map[major(dev)] == NULL ? EBADF + : mgr_map[major(dev)]->do_cdev_write(dev, uio, ioflag)); +} + +int +tuntap_manager::cdev_ioctl(dev_t dev, u_long cmd, caddr_t data, int fflag, proc_t p) +{ + return (mgr_map[major(dev)] == NULL ? EBADF + : mgr_map[major(dev)]->do_cdev_ioctl(dev, cmd, data, fflag, p)); +} + +int +tuntap_manager::cdev_select(dev_t dev, int which, void *wql, proc_t p) +{ + return (mgr_map[major(dev)] == NULL ? EBADF + : mgr_map[major(dev)]->do_cdev_select(dev, which, wql, p)); +} + +/* character device service methods */ +int +tuntap_manager::do_cdev_open(dev_t dev, int flags, int devtype, proc_t p) +{ + int dmin = minor(dev); + int error = ENOENT; + + cdev_gate.enter(); + + if (dmin < (int) count && dmin >= 0 && tuntaps[dmin] != NULL) + error = tuntaps[dmin]->cdev_open(flags, devtype, p); + + cdev_gate.exit(); + + return error; +} + +int +tuntap_manager::do_cdev_close(dev_t dev, int flags, int devtype, proc_t p) +{ + int dmin = minor(dev); + int error = EBADF; + + cdev_gate.enter(); + + if (dmin < (int) count && dmin >= 0 && tuntaps[dmin] != NULL) + error = tuntaps[dmin]->cdev_close(flags, devtype, p); + + cdev_gate.exit(); + + return error; +} + +int +tuntap_manager::do_cdev_read(dev_t dev, uio_t uio, int ioflag) +{ + int dmin = minor(dev); + int error = EBADF; + + cdev_gate.enter(); + + if (dmin < (int) count && dmin >= 0 && tuntaps[dmin] != NULL) + error = tuntaps[dmin]->cdev_read(uio, ioflag); + + cdev_gate.exit(); + + return error; +} + +int +tuntap_manager::do_cdev_write(dev_t dev, uio_t uio, int ioflag) +{ + int dmin = minor(dev); + int error = EBADF; + + cdev_gate.enter(); + + if (dmin < (int) count && dmin >= 0 && tuntaps[dmin] != NULL) + error = tuntaps[dmin]->cdev_write(uio, ioflag); + + cdev_gate.exit(); + + return error; +} + +int +tuntap_manager::do_cdev_ioctl(dev_t dev, u_long cmd, caddr_t data, int fflag, proc_t p) +{ + int dmin = minor(dev); + int error = EBADF; + + cdev_gate.enter(); + + if (dmin < (int) count && dmin >= 0 && tuntaps[dmin] != NULL) + error = tuntaps[dmin]->cdev_ioctl(cmd, data, fflag, p); + + cdev_gate.exit(); + + return error; +} + +int +tuntap_manager::do_cdev_select(dev_t dev, int which, void *wql, proc_t p) +{ + int dmin = minor(dev); + int error = EBADF; + + cdev_gate.enter(); + + if (dmin < (int) count && dmin >= 0 && tuntaps[dmin] != NULL) + error = tuntaps[dmin]->cdev_select(which, wql, p); + + cdev_gate.exit(); + + return error; +} + diff --git a/ext/tap-mac/tuntap/src/util.h b/ext/tap-mac/tuntap/src/util.h new file mode 100644 index 00000000..0f6955e8 --- /dev/null +++ b/ext/tap-mac/tuntap/src/util.h @@ -0,0 +1,46 @@ +/* + * ip tunnel/ethertap device for MacOSX. + * + * Some utilities and misc stuff. + */ +/* + * Copyright (c) 2011 Mattias Nissler <mattias.nissler@gmx.de> + * + * Redistribution and use in source and binary forms, with or without modification, are permitted + * provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, this list of + * conditions and the following disclaimer. + * 2. 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. + * 3. The name of the author may not be used to endorse or promote products derived from this + * software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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. + */ + +#ifndef __UTIL_H__ +#define __UTIL_H__ + +extern "C" { + +/* In Darwin 8 (OS X Tiger) there is a problem with struct selinfo. It was made `private' to the + * kernel, so its definition is not available from the headers in Kernel.framework. However, we need + * to declare something :-( + */ +struct selinfo { + char data[128]; /* should be enough... */ +}; + +} /* extern "C" */ + +#endif /* __UTIL_H__ */ + diff --git a/ext/tap-mac/tuntap/startup_item/tap/Resources/English.lproj/Localizable.strings b/ext/tap-mac/tuntap/startup_item/tap/Resources/English.lproj/Localizable.strings new file mode 100644 index 00000000..4d3b29ca --- /dev/null +++ b/ext/tap-mac/tuntap/startup_item/tap/Resources/English.lproj/Localizable.strings @@ -0,0 +1,11 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!DOCTYPE plist SYSTEM "file://localhost/System/Library/DTDs/PropertyList.dtd"> +<plist version="0.9"> +<dict> + <key>ethertap device kernel extension</key> + <string>ethertap kernel extension</string> + <key>Initializing tap devices</key> + <string>Initializing tap devices</string> +</dict> +</plist> + diff --git a/ext/tap-mac/tuntap/startup_item/tap/StartupParameters.plist b/ext/tap-mac/tuntap/startup_item/tap/StartupParameters.plist new file mode 100644 index 00000000..fe9cf001 --- /dev/null +++ b/ext/tap-mac/tuntap/startup_item/tap/StartupParameters.plist @@ -0,0 +1,6 @@ +{ + Description = "ethertap device kernel extension"; + Provides = ("ethertap"); + Requires = ("Network Configuration"); + OrderPreference = "None"; +} diff --git a/ext/tap-mac/tuntap/startup_item/tap/tap b/ext/tap-mac/tuntap/startup_item/tap/tap new file mode 100755 index 00000000..d7600991 --- /dev/null +++ b/ext/tap-mac/tuntap/startup_item/tap/tap @@ -0,0 +1,33 @@ +#!/bin/sh + +## +# load the tap kext +## + +. /etc/rc.common + +StartService () +{ + ConsoleMessage "Initializing tap devices" + + if [ -d /Library/Extensions/tap.kext ]; then + kextload /Library/Extensions/tap.kext + fi +} + +StopService () +{ + if [ -d /Library/Extensions/tap.kext ]; then + kextunload /Library/Extensions/tap.kext + fi +} + +RestartService () +{ + if [ -d /Library/Extensions/tap.kext ]; then + kextunload /Library/Extensions/tap.kext + kextload /Library/Extensions/tap.kext + fi +} + +RunService "$1" diff --git a/ext/tap-mac/tuntap/startup_item/tun/Resources/English.lproj/Localizable.strings b/ext/tap-mac/tuntap/startup_item/tun/Resources/English.lproj/Localizable.strings new file mode 100644 index 00000000..054790e7 --- /dev/null +++ b/ext/tap-mac/tuntap/startup_item/tun/Resources/English.lproj/Localizable.strings @@ -0,0 +1,11 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!DOCTYPE plist SYSTEM "file://localhost/System/Library/DTDs/PropertyList.dtd"> +<plist version="0.9"> +<dict> + <key>ip tunnel device kernel extension</key> + <string>ip tunnel kernel extension</string> + <key>Initializing tun devices</key> + <string>Initializing tun devices</string> +</dict> +</plist> + diff --git a/ext/tap-mac/tuntap/startup_item/tun/StartupParameters.plist b/ext/tap-mac/tuntap/startup_item/tun/StartupParameters.plist new file mode 100644 index 00000000..71a92626 --- /dev/null +++ b/ext/tap-mac/tuntap/startup_item/tun/StartupParameters.plist @@ -0,0 +1,6 @@ +{ + Description = "ip tunnel device kernel extension"; + Provides = ("tun"); + Requires = ("Network Configuration"); + OrderPreference = "None"; +} diff --git a/ext/tap-mac/tuntap/startup_item/tun/tun b/ext/tap-mac/tuntap/startup_item/tun/tun new file mode 100755 index 00000000..40b1b411 --- /dev/null +++ b/ext/tap-mac/tuntap/startup_item/tun/tun @@ -0,0 +1,33 @@ +#!/bin/sh + +## +# load the tun kext +## + +. /etc/rc.common + +StartService () +{ + ConsoleMessage "Initializing tun devices" + + if [ -d /Library/Extensions/tun.kext ]; then + kextload /Library/Extensions/tun.kext + fi +} + +StopService () +{ + if [ -d /Library/Extensions/tun.kext ]; then + kextunload /Library/Extensions/tun.kext + fi +} + +RestartService () +{ + if [ -d /Library/Extensions/tun.kext ]; then + kextunload /Library/Extensions/tun.kext + kextload /Library/Extensions/tun.kext + fi +} + +RunService "$1" |