diff options
Diffstat (limited to 'src')
115 files changed, 5024 insertions, 942 deletions
diff --git a/src/common/mythread.h b/src/common/mythread.h index 476c2fc9e103..be22654240aa 100644 --- a/src/common/mythread.h +++ b/src/common/mythread.h @@ -1,7 +1,7 @@ /////////////////////////////////////////////////////////////////////////////// // /// \file mythread.h -/// \brief Wrappers for threads +/// \brief Some threading related helper macros and functions // // Author: Lasse Collin // @@ -10,33 +10,512 @@ // /////////////////////////////////////////////////////////////////////////////// +#ifndef MYTHREAD_H +#define MYTHREAD_H + #include "sysdefs.h" +// If any type of threading is enabled, #define MYTHREAD_ENABLED. +#if defined(MYTHREAD_POSIX) || defined(MYTHREAD_WIN95) \ + || defined(MYTHREAD_VISTA) +# define MYTHREAD_ENABLED 1 +#endif + + +#ifdef MYTHREAD_ENABLED + +//////////////////////////////////////// +// Shared between all threading types // +//////////////////////////////////////// + +// Locks a mutex for a duration of a block. +// +// Perform mythread_mutex_lock(&mutex) in the beginning of a block +// and mythread_mutex_unlock(&mutex) at the end of the block. "break" +// may be used to unlock the mutex and jump out of the block. +// mythread_sync blocks may be nested. +// +// Example: +// +// mythread_sync(mutex) { +// foo(); +// if (some_error) +// break; // Skips bar() +// bar(); +// } +// +// At least GCC optimizes the loops completely away so it doesn't slow +// things down at all compared to plain mythread_mutex_lock(&mutex) +// and mythread_mutex_unlock(&mutex) calls. +// +#define mythread_sync(mutex) mythread_sync_helper1(mutex, __LINE__) +#define mythread_sync_helper1(mutex, line) mythread_sync_helper2(mutex, line) +#define mythread_sync_helper2(mutex, line) \ + for (unsigned int mythread_i_ ## line = 0; \ + mythread_i_ ## line \ + ? (mythread_mutex_unlock(&(mutex)), 0) \ + : (mythread_mutex_lock(&(mutex)), 1); \ + mythread_i_ ## line = 1) \ + for (unsigned int mythread_j_ ## line = 0; \ + !mythread_j_ ## line; \ + mythread_j_ ## line = 1) +#endif + + +#if !defined(MYTHREAD_ENABLED) + +////////////////// +// No threading // +////////////////// + +// Calls the given function once. This isn't thread safe. +#define mythread_once(func) \ +do { \ + static bool once_ = false; \ + if (!once_) { \ + func(); \ + once_ = true; \ + } \ +} while (0) + + +#if !(defined(_WIN32) && !defined(__CYGWIN__)) +// Use sigprocmask() to set the signal mask in single-threaded programs. +#include <signal.h> + +static inline void +mythread_sigmask(int how, const sigset_t *restrict set, + sigset_t *restrict oset) +{ + int ret = sigprocmask(how, set, oset); + assert(ret == 0); + (void)ret; +} +#endif + + +#elif defined(MYTHREAD_POSIX) + +//////////////////// +// Using pthreads // +//////////////////// + +#include <sys/time.h> +#include <pthread.h> +#include <signal.h> +#include <time.h> +#include <errno.h> + +#define MYTHREAD_RET_TYPE void * +#define MYTHREAD_RET_VALUE NULL + +typedef pthread_t mythread; +typedef pthread_mutex_t mythread_mutex; + +typedef struct { + pthread_cond_t cond; +#ifdef HAVE_CLOCK_GETTIME + // Clock ID (CLOCK_REALTIME or CLOCK_MONOTONIC) associated with + // the condition variable. + clockid_t clk_id; +#endif +} mythread_cond; + +typedef struct timespec mythread_condtime; -#ifdef HAVE_PTHREAD -# include <pthread.h> -# define mythread_once(func) \ +// Calls the given function once in a thread-safe way. +#define mythread_once(func) \ do { \ static pthread_once_t once_ = PTHREAD_ONCE_INIT; \ pthread_once(&once_, &func); \ } while (0) -# define mythread_sigmask(how, set, oset) \ - pthread_sigmask(how, set, oset) +// Use pthread_sigmask() to set the signal mask in multi-threaded programs. +// Do nothing on OpenVMS since it lacks pthread_sigmask(). +static inline void +mythread_sigmask(int how, const sigset_t *restrict set, + sigset_t *restrict oset) +{ +#ifdef __VMS + (void)how; + (void)set; + (void)oset; +#else + int ret = pthread_sigmask(how, set, oset); + assert(ret == 0); + (void)ret; +#endif +} + + +// Creates a new thread with all signals blocked. Returns zero on success +// and non-zero on error. +static inline int +mythread_create(mythread *thread, void *(*func)(void *arg), void *arg) +{ + sigset_t old; + sigset_t all; + sigfillset(&all); + + mythread_sigmask(SIG_SETMASK, &all, &old); + const int ret = pthread_create(thread, NULL, func, arg); + mythread_sigmask(SIG_SETMASK, &old, NULL); + + return ret; +} + +// Joins a thread. Returns zero on success and non-zero on error. +static inline int +mythread_join(mythread thread) +{ + return pthread_join(thread, NULL); +} + + +// Initiatlizes a mutex. Returns zero on success and non-zero on error. +static inline int +mythread_mutex_init(mythread_mutex *mutex) +{ + return pthread_mutex_init(mutex, NULL); +} + +static inline void +mythread_mutex_destroy(mythread_mutex *mutex) +{ + int ret = pthread_mutex_destroy(mutex); + assert(ret == 0); + (void)ret; +} + +static inline void +mythread_mutex_lock(mythread_mutex *mutex) +{ + int ret = pthread_mutex_lock(mutex); + assert(ret == 0); + (void)ret; +} + +static inline void +mythread_mutex_unlock(mythread_mutex *mutex) +{ + int ret = pthread_mutex_unlock(mutex); + assert(ret == 0); + (void)ret; +} + + +// Initializes a condition variable. +// +// Using CLOCK_MONOTONIC instead of the default CLOCK_REALTIME makes the +// timeout in pthread_cond_timedwait() work correctly also if system time +// is suddenly changed. Unfortunately CLOCK_MONOTONIC isn't available +// everywhere while the default CLOCK_REALTIME is, so the default is +// used if CLOCK_MONOTONIC isn't available. +// +// If clock_gettime() isn't available at all, gettimeofday() will be used. +static inline int +mythread_cond_init(mythread_cond *mycond) +{ +#ifdef HAVE_CLOCK_GETTIME + // NOTE: HAVE_DECL_CLOCK_MONOTONIC is always defined to 0 or 1. +# if defined(HAVE_PTHREAD_CONDATTR_SETCLOCK) && HAVE_DECL_CLOCK_MONOTONIC + struct timespec ts; + pthread_condattr_t condattr; + + // POSIX doesn't seem to *require* that pthread_condattr_setclock() + // will fail if given an unsupported clock ID. Test that + // CLOCK_MONOTONIC really is supported using clock_gettime(). + if (clock_gettime(CLOCK_MONOTONIC, &ts) == 0 + && pthread_condattr_init(&condattr) == 0) { + int ret = pthread_condattr_setclock( + &condattr, CLOCK_MONOTONIC); + if (ret == 0) + ret = pthread_cond_init(&mycond->cond, &condattr); + + pthread_condattr_destroy(&condattr); + + if (ret == 0) { + mycond->clk_id = CLOCK_MONOTONIC; + return 0; + } + } + + // If anything above fails, fall back to the default CLOCK_REALTIME. + // POSIX requires that all implementations of clock_gettime() must + // support at least CLOCK_REALTIME. +# endif + + mycond->clk_id = CLOCK_REALTIME; +#endif + + return pthread_cond_init(&mycond->cond, NULL); +} + +static inline void +mythread_cond_destroy(mythread_cond *cond) +{ + int ret = pthread_cond_destroy(&cond->cond); + assert(ret == 0); + (void)ret; +} + +static inline void +mythread_cond_signal(mythread_cond *cond) +{ + int ret = pthread_cond_signal(&cond->cond); + assert(ret == 0); + (void)ret; +} + +static inline void +mythread_cond_wait(mythread_cond *cond, mythread_mutex *mutex) +{ + int ret = pthread_cond_wait(&cond->cond, mutex); + assert(ret == 0); + (void)ret; +} + +// Waits on a condition or until a timeout expires. If the timeout expires, +// non-zero is returned, otherwise zero is returned. +static inline int +mythread_cond_timedwait(mythread_cond *cond, mythread_mutex *mutex, + const mythread_condtime *condtime) +{ + int ret = pthread_cond_timedwait(&cond->cond, mutex, condtime); + assert(ret == 0 || ret == ETIMEDOUT); + return ret; +} + +// Sets condtime to the absolute time that is timeout_ms milliseconds +// in the future. The type of the clock to use is taken from cond. +static inline void +mythread_condtime_set(mythread_condtime *condtime, const mythread_cond *cond, + uint32_t timeout_ms) +{ + condtime->tv_sec = timeout_ms / 1000; + condtime->tv_nsec = (timeout_ms % 1000) * 1000000; + +#ifdef HAVE_CLOCK_GETTIME + struct timespec now; + int ret = clock_gettime(cond->clk_id, &now); + assert(ret == 0); + (void)ret; + + condtime->tv_sec += now.tv_sec; + condtime->tv_nsec += now.tv_nsec; #else + (void)cond; + + struct timeval now; + gettimeofday(&now, NULL); + + condtime->tv_sec += now.tv_sec; + condtime->tv_nsec += now.tv_usec * 1000L; +#endif + + // tv_nsec must stay in the range [0, 999_999_999]. + if (condtime->tv_nsec >= 1000000000L) { + condtime->tv_nsec -= 1000000000L; + ++condtime->tv_sec; + } +} + + +#elif defined(MYTHREAD_WIN95) || defined(MYTHREAD_VISTA) -# define mythread_once(func) \ +///////////////////// +// Windows threads // +///////////////////// + +#define WIN32_LEAN_AND_MEAN +#ifdef MYTHREAD_VISTA +# undef _WIN32_WINNT +# define _WIN32_WINNT 0x0600 +#endif +#include <windows.h> +#include <process.h> + +#define MYTHREAD_RET_TYPE unsigned int __stdcall +#define MYTHREAD_RET_VALUE 0 + +typedef HANDLE mythread; +typedef CRITICAL_SECTION mythread_mutex; + +#ifdef MYTHREAD_WIN95 +typedef HANDLE mythread_cond; +#else +typedef CONDITION_VARIABLE mythread_cond; +#endif + +typedef struct { + // Tick count (milliseconds) in the beginning of the timeout. + // NOTE: This is 32 bits so it wraps around after 49.7 days. + // Multi-day timeouts may not work as expected. + DWORD start; + + // Length of the timeout in milliseconds. The timeout expires + // when the current tick count minus "start" is equal or greater + // than "timeout". + DWORD timeout; +} mythread_condtime; + + +// mythread_once() is only available with Vista threads. +#ifdef MYTHREAD_VISTA +#define mythread_once(func) \ do { \ - static bool once_ = false; \ - if (!once_) { \ + static INIT_ONCE once_ = INIT_ONCE_STATIC_INIT; \ + BOOL pending_; \ + if (!InitOnceBeginInitialize(&once_, 0, &pending_, NULL)) \ + abort(); \ + if (pending_) \ func(); \ - once_ = true; \ - } \ + if (!InitOnceComplete(&once, 0, NULL)) \ + abort(); \ } while (0) +#endif + + +// mythread_sigmask() isn't available on Windows. Even a dummy version would +// make no sense because the other POSIX signal functions are missing anyway. + + +static inline int +mythread_create(mythread *thread, + unsigned int (__stdcall *func)(void *arg), void *arg) +{ + uintptr_t ret = _beginthreadex(NULL, 0, func, arg, 0, NULL); + if (ret == 0) + return -1; + + *thread = (HANDLE)ret; + return 0; +} + +static inline int +mythread_join(mythread thread) +{ + int ret = 0; + + if (WaitForSingleObject(thread, INFINITE) != WAIT_OBJECT_0) + ret = -1; + + if (!CloseHandle(thread)) + ret = -1; + + return ret; +} -# define mythread_sigmask(how, set, oset) \ - sigprocmask(how, set, oset) + +static inline int +mythread_mutex_init(mythread_mutex *mutex) +{ + InitializeCriticalSection(mutex); + return 0; +} + +static inline void +mythread_mutex_destroy(mythread_mutex *mutex) +{ + DeleteCriticalSection(mutex); +} + +static inline void +mythread_mutex_lock(mythread_mutex *mutex) +{ + EnterCriticalSection(mutex); +} + +static inline void +mythread_mutex_unlock(mythread_mutex *mutex) +{ + LeaveCriticalSection(mutex); +} + + +static inline int +mythread_cond_init(mythread_cond *cond) +{ +#ifdef MYTHREAD_WIN95 + *cond = CreateEvent(NULL, FALSE, FALSE, NULL); + return *cond == NULL ? -1 : 0; +#else + InitializeConditionVariable(cond); + return 0; +#endif +} + +static inline void +mythread_cond_destroy(mythread_cond *cond) +{ +#ifdef MYTHREAD_WIN95 + CloseHandle(*cond); +#else + (void)cond; +#endif +} + +static inline void +mythread_cond_signal(mythread_cond *cond) +{ +#ifdef MYTHREAD_WIN95 + SetEvent(*cond); +#else + WakeConditionVariable(cond); +#endif +} + +static inline void +mythread_cond_wait(mythread_cond *cond, mythread_mutex *mutex) +{ +#ifdef MYTHREAD_WIN95 + LeaveCriticalSection(mutex); + WaitForSingleObject(*cond, INFINITE); + EnterCriticalSection(mutex); +#else + BOOL ret = SleepConditionVariableCS(cond, mutex, INFINITE); + assert(ret); + (void)ret; +#endif +} + +static inline int +mythread_cond_timedwait(mythread_cond *cond, mythread_mutex *mutex, + const mythread_condtime *condtime) +{ +#ifdef MYTHREAD_WIN95 + LeaveCriticalSection(mutex); +#endif + + DWORD elapsed = GetTickCount() - condtime->start; + DWORD timeout = elapsed >= condtime->timeout + ? 0 : condtime->timeout - elapsed; + +#ifdef MYTHREAD_WIN95 + DWORD ret = WaitForSingleObject(*cond, timeout); + assert(ret == WAIT_OBJECT_0 || ret == WAIT_TIMEOUT); + + EnterCriticalSection(mutex); + + return ret == WAIT_TIMEOUT; +#else + BOOL ret = SleepConditionVariableCS(cond, mutex, timeout); + assert(ret || GetLastError() == ERROR_TIMEOUT); + return !ret; +#endif +} + +static inline void +mythread_condtime_set(mythread_condtime *condtime, const mythread_cond *cond, + uint32_t timeout) +{ + (void)cond; + condtime->start = GetTickCount(); + condtime->timeout = timeout; +} + +#endif #endif diff --git a/src/common/sysdefs.h b/src/common/sysdefs.h index 5ea6bdae4aed..e056ca4ac908 100644 --- a/src/common/sysdefs.h +++ b/src/common/sysdefs.h @@ -165,6 +165,16 @@ typedef unsigned char _Bool; # include <memory.h> #endif +// As of MSVC 2013, inline and restrict are supported with +// non-standard keywords. +#if defined(_WIN32) && defined(_MSC_VER) +# ifndef inline +# define inline __inline +# endif +# ifndef restrict +# define restrict __restrict +# endif +#endif //////////// // Macros // diff --git a/src/common/tuklib_cpucores.c b/src/common/tuklib_cpucores.c index 1da13df7f5f6..7574bc9c0645 100644 --- a/src/common/tuklib_cpucores.c +++ b/src/common/tuklib_cpucores.c @@ -12,7 +12,13 @@ #include "tuklib_cpucores.h" -#if defined(TUKLIB_CPUCORES_SYSCTL) +#if defined(_WIN32) || defined(__CYGWIN__) +# ifndef _WIN32_WINNT +# define _WIN32_WINNT 0x0500 +# endif +# include <windows.h> + +#elif defined(TUKLIB_CPUCORES_SYSCTL) # ifdef HAVE_SYS_PARAM_H # include <sys/param.h> # endif @@ -33,7 +39,12 @@ tuklib_cpucores(void) { uint32_t ret = 0; -#if defined(TUKLIB_CPUCORES_SYSCTL) +#if defined(_WIN32) || defined(__CYGWIN__) + SYSTEM_INFO sysinfo; + GetSystemInfo(&sysinfo); + ret = sysinfo.dwNumberOfProcessors; + +#elif defined(TUKLIB_CPUCORES_SYSCTL) int name[2] = { CTL_HW, HW_NCPU }; int cpus; size_t cpus_size = sizeof(cpus); diff --git a/src/common/tuklib_physmem.c b/src/common/tuklib_physmem.c index 623b6e70b7f2..3cc7d12a171f 100644 --- a/src/common/tuklib_physmem.c +++ b/src/common/tuklib_physmem.c @@ -33,6 +33,10 @@ # include <syidef.h> # include <ssdef.h> +#elif defined(AMIGA) || defined(__AROS__) +# define __USE_INLINE__ +# include <proto/exec.h> + // AIX #elif defined(TUKLIB_PHYSMEM_AIX) # include <sys/systemcfg.h> @@ -119,6 +123,9 @@ tuklib_physmem(void) if (LIB$GETSYI(&val, &vms_mem, 0, 0, 0, 0) == SS$_NORMAL) ret = (uint64_t)vms_mem * 8192; +#elif defined(AMIGA) || defined(__AROS__) + ret = AvailMem(MEMF_TOTAL); + #elif defined(TUKLIB_PHYSMEM_AIX) ret = _system_configuration.physmem; diff --git a/src/liblzma/api/lzma.h b/src/liblzma/api/lzma.h index fb874c3e1377..72c51b2411c3 100644 --- a/src/liblzma/api/lzma.h +++ b/src/liblzma/api/lzma.h @@ -286,7 +286,7 @@ extern "C" { #include "lzma/filter.h" #include "lzma/bcj.h" #include "lzma/delta.h" -#include "lzma/lzma.h" +#include "lzma/lzma12.h" /* Container formats */ #include "lzma/container.h" diff --git a/src/liblzma/api/lzma/base.h b/src/liblzma/api/lzma/base.h index 43dde8d60f3d..7a31a4205136 100644 --- a/src/liblzma/api/lzma/base.h +++ b/src/liblzma/api/lzma/base.h @@ -240,12 +240,12 @@ typedef enum { /** * \brief The `action' argument for lzma_code() * - * After the first use of LZMA_SYNC_FLUSH, LZMA_FULL_FLUSH, or LZMA_FINISH, - * the same `action' must is used until lzma_code() returns LZMA_STREAM_END. - * Also, the amount of input (that is, strm->avail_in) must not be modified - * by the application until lzma_code() returns LZMA_STREAM_END. Changing the - * `action' or modifying the amount of input will make lzma_code() return - * LZMA_PROG_ERROR. + * After the first use of LZMA_SYNC_FLUSH, LZMA_FULL_FLUSH, LZMA_FULL_BARRIER, + * or LZMA_FINISH, the same `action' must is used until lzma_code() returns + * LZMA_STREAM_END. Also, the amount of input (that is, strm->avail_in) must + * not be modified by the application until lzma_code() returns + * LZMA_STREAM_END. Changing the `action' or modifying the amount of input + * will make lzma_code() return LZMA_PROG_ERROR. */ typedef enum { LZMA_RUN = 0, @@ -293,7 +293,7 @@ typedef enum { * * All the input data going to the current Block must have * been given to the encoder (the last bytes can still be - * pending in* next_in). Call lzma_code() with LZMA_FULL_FLUSH + * pending in *next_in). Call lzma_code() with LZMA_FULL_FLUSH * until it returns LZMA_STREAM_END. Then continue normally * with LZMA_RUN or finish the Stream with LZMA_FINISH. * @@ -302,6 +302,29 @@ typedef enum { * no unfinished Block, no empty Block is created. */ + LZMA_FULL_BARRIER = 4, + /**< + * \brief Finish encoding of the current Block + * + * This is like LZMA_FULL_FLUSH except that this doesn't + * necessarily wait until all the input has been made + * available via the output buffer. That is, lzma_code() + * might return LZMA_STREAM_END as soon as all the input + * has been consumed (avail_in == 0). + * + * LZMA_FULL_BARRIER is useful with a threaded encoder if + * one wants to split the .xz Stream into Blocks at specific + * offsets but doesn't care if the output isn't flushed + * immediately. Using LZMA_FULL_BARRIER allows keeping + * the threads busy while LZMA_FULL_FLUSH would make + * lzma_code() wait until all the threads have finished + * until more data could be passed to the encoder. + * + * With a lzma_stream initialized with the single-threaded + * lzma_stream_encoder() or lzma_easy_encoder(), + * LZMA_FULL_BARRIER is an alias for LZMA_FULL_FLUSH. + */ + LZMA_FINISH = 3 /**< * \brief Finish the coding operation @@ -332,11 +355,19 @@ typedef enum { * malloc() and free(). C++ users should note that the custom memory * handling functions must not throw exceptions. * - * liblzma doesn't make an internal copy of lzma_allocator. Thus, it is - * OK to change these function pointers in the middle of the coding - * process, but obviously it must be done carefully to make sure that the - * replacement `free' can deallocate memory allocated by the earlier - * `alloc' function(s). + * Single-threaded mode only: liblzma doesn't make an internal copy of + * lzma_allocator. Thus, it is OK to change these function pointers in + * the middle of the coding process, but obviously it must be done + * carefully to make sure that the replacement `free' can deallocate + * memory allocated by the earlier `alloc' function(s). + * + * Multithreaded mode: liblzma might internally store pointers to the + * lzma_allocator given via the lzma_stream structure. The application + * must not change the allocator pointer in lzma_stream or the contents + * of the pointed lzma_allocator structure until lzma_end() has been used + * to free the memory associated with that lzma_stream. The allocation + * functions might be called simultaneously from multiple threads, and + * thus they must be thread safe. */ typedef struct { /** @@ -448,7 +479,8 @@ typedef struct lzma_internal_s lzma_internal; * * Application may modify the values of total_in and total_out as it wants. * They are updated by liblzma to match the amount of data read and - * written, but aren't used for anything else. + * written but aren't used for anything else except as a possible return + * values from lzma_get_progress(). */ typedef struct { const uint8_t *next_in; /**< Pointer to the next input byte. */ @@ -464,8 +496,10 @@ typedef struct { * * In most cases this is NULL which makes liblzma use * the standard malloc() and free(). + * + * \note In 5.0.x this is not a const pointer. */ - lzma_allocator *allocator; + const lzma_allocator *allocator; /** Internal state is not visible to applications. */ lzma_internal *internal; @@ -547,6 +581,25 @@ extern LZMA_API(void) lzma_end(lzma_stream *strm) lzma_nothrow; /** + * \brief Get progress information + * + * In single-threaded mode, applications can get progress information from + * strm->total_in and strm->total_out. In multi-threaded mode this is less + * useful because a significant amount of both input and output data gets + * buffered internally by liblzma. This makes total_in and total_out give + * misleading information and also makes the progress indicator updates + * non-smooth. + * + * This function gives realistic progress information also in multi-threaded + * mode by taking into account the progress made by each thread. In + * single-threaded mode *progress_in and *progress_out are set to + * strm->total_in and strm->total_out, respectively. + */ +extern LZMA_API(void) lzma_get_progress(lzma_stream *strm, + uint64_t *progress_in, uint64_t *progress_out) lzma_nothrow; + + +/** * \brief Get the memory usage of decoder filter chain * * This function is currently supported only when *strm has been initialized diff --git a/src/liblzma/api/lzma/block.h b/src/liblzma/api/lzma/block.h index e6710a7bc816..7bdcfd7cbe0c 100644 --- a/src/liblzma/api/lzma/block.h +++ b/src/liblzma/api/lzma/block.h @@ -31,11 +31,16 @@ typedef struct { /** * \brief Block format version * - * To prevent API and ABI breakages if new features are needed in - * the Block field, a version number is used to indicate which - * fields in this structure are in use. For now, version must always - * be zero. With non-zero version, most Block related functions will - * return LZMA_OPTIONS_ERROR. + * To prevent API and ABI breakages when new features are needed, + * a version number is used to indicate which fields in this + * structure are in use: + * - liblzma >= 5.0.0: version = 0 is supported. + * - liblzma >= 5.1.4beta: Support for version = 1 was added, + * which adds the ignore_check field. + * + * If version is greater than one, most Block related functions + * will return LZMA_OPTIONS_ERROR (lzma_block_header_decode() works + * with any version value). * * Read by: * - All functions that take pointer to lzma_block as argument, @@ -233,7 +238,28 @@ typedef struct { lzma_reserved_enum reserved_enum2; lzma_reserved_enum reserved_enum3; lzma_reserved_enum reserved_enum4; - lzma_bool reserved_bool1; + + /** + * \brief A flag to Block decoder to not verify the Check field + * + * This field is supported by liblzma >= 5.1.4beta if .version >= 1. + * + * If this is set to true, the integrity check won't be calculated + * and verified. Unless you know what you are doing, you should + * leave this to false. (A reason to set this to true is when the + * file integrity is verified externally anyway and you want to + * speed up the decompression, which matters mostly when using + * SHA-256 as the integrity check.) + * + * If .version >= 1, read by: + * - lzma_block_decoder() + * - lzma_block_buffer_decode() + * + * Written by (.version is ignored): + * - lzma_block_header_decode() always sets this to false + */ + lzma_bool ignore_check; + lzma_bool reserved_bool2; lzma_bool reserved_bool3; lzma_bool reserved_bool4; @@ -310,10 +336,14 @@ extern LZMA_API(lzma_ret) lzma_block_header_encode( /** * \brief Decode Block Header * - * block->version should be set to the highest value supported by the - * application; currently the only possible version is zero. This function - * will set version to the lowest value that still supports all the features - * required by the Block Header. + * block->version should (usually) be set to the highest value supported + * by the application. If the application sets block->version to a value + * higher than supported by the current liblzma version, this function will + * downgrade block->version to the highest value supported by it. Thus one + * should check the value of block->version after calling this function if + * block->version was set to a non-zero value and the application doesn't + * otherwise know that the liblzma version being used is new enough to + * support the specified block->version. * * The size of the Block Header must have already been decoded with * lzma_block_header_size_decode() macro and stored to block->header_size. @@ -344,7 +374,7 @@ extern LZMA_API(lzma_ret) lzma_block_header_encode( * block->header_size is invalid or block->filters is NULL. */ extern LZMA_API(lzma_ret) lzma_block_header_decode(lzma_block *block, - lzma_allocator *allocator, const uint8_t *in) + const lzma_allocator *allocator, const uint8_t *in) lzma_nothrow lzma_attr_warn_unused_result; @@ -493,7 +523,25 @@ extern LZMA_API(size_t) lzma_block_buffer_bound(size_t uncompressed_size) * - LZMA_PROG_ERROR */ extern LZMA_API(lzma_ret) lzma_block_buffer_encode( - lzma_block *block, lzma_allocator *allocator, + lzma_block *block, const lzma_allocator *allocator, + const uint8_t *in, size_t in_size, + uint8_t *out, size_t *out_pos, size_t out_size) + lzma_nothrow lzma_attr_warn_unused_result; + + +/** + * \brief Single-call uncompressed .xz Block encoder + * + * This is like lzma_block_buffer_encode() except this doesn't try to + * compress the data and instead encodes the data using LZMA2 uncompressed + * chunks. The required output buffer size can be determined with + * lzma_block_buffer_bound(). + * + * Since the data won't be compressed, this function ignores block->filters. + * This function doesn't take lzma_allocator because this function doesn't + * allocate any memory from the heap. + */ +extern LZMA_API(lzma_ret) lzma_block_uncomp_encode(lzma_block *block, const uint8_t *in, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow lzma_attr_warn_unused_result; @@ -527,7 +575,7 @@ extern LZMA_API(lzma_ret) lzma_block_buffer_encode( * - LZMA_PROG_ERROR */ extern LZMA_API(lzma_ret) lzma_block_buffer_decode( - lzma_block *block, lzma_allocator *allocator, + lzma_block *block, const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow; diff --git a/src/liblzma/api/lzma/container.h b/src/liblzma/api/lzma/container.h index 7a9ffc645788..86991add1967 100644 --- a/src/liblzma/api/lzma/container.h +++ b/src/liblzma/api/lzma/container.h @@ -61,6 +61,131 @@ /** + * \brief Multithreading options + */ +typedef struct { + /** + * \brief Flags + * + * Set this to zero if no flags are wanted. + * + * No flags are currently supported. + */ + uint32_t flags; + + /** + * \brief Number of worker threads to use + */ + uint32_t threads; + + /** + * \brief Maximum uncompressed size of a Block + * + * The encoder will start a new .xz Block every block_size bytes. + * Using LZMA_FULL_FLUSH or LZMA_FULL_BARRIER with lzma_code() + * the caller may tell liblzma to start a new Block earlier. + * + * With LZMA2, a recommended block size is 2-4 times the LZMA2 + * dictionary size. With very small dictionaries, it is recommended + * to use at least 1 MiB block size for good compression ratio, even + * if this is more than four times the dictionary size. Note that + * these are only recommendations for typical use cases; feel free + * to use other values. Just keep in mind that using a block size + * less than the LZMA2 dictionary size is waste of RAM. + * + * Set this to 0 to let liblzma choose the block size depending + * on the compression options. For LZMA2 it will be 3*dict_size + * or 1 MiB, whichever is more. + * + * For each thread, about 3 * block_size bytes of memory will be + * allocated. This may change in later liblzma versions. If so, + * the memory usage will probably be reduced, not increased. + */ + uint64_t block_size; + + /** + * \brief Timeout to allow lzma_code() to return early + * + * Multithreading can make liblzma to consume input and produce + * output in a very bursty way: it may first read a lot of input + * to fill internal buffers, then no input or output occurs for + * a while. + * + * In single-threaded mode, lzma_code() won't return until it has + * either consumed all the input or filled the output buffer. If + * this is done in multithreaded mode, it may cause a call + * lzma_code() to take even tens of seconds, which isn't acceptable + * in all applications. + * + * To avoid very long blocking times in lzma_code(), a timeout + * (in milliseconds) may be set here. If lzma_code() would block + * longer than this number of milliseconds, it will return with + * LZMA_OK. Reasonable values are 100 ms or more. The xz command + * line tool uses 300 ms. + * + * If long blocking times are fine for you, set timeout to a special + * value of 0, which will disable the timeout mechanism and will make + * lzma_code() block until all the input is consumed or the output + * buffer has been filled. + * + * \note Even with a timeout, lzma_code() might sometimes take + * somewhat long time to return. No timing guarantees + * are made. + */ + uint32_t timeout; + + /** + * \brief Compression preset (level and possible flags) + * + * The preset is set just like with lzma_easy_encoder(). + * The preset is ignored if filters below is non-NULL. + */ + uint32_t preset; + + /** + * \brief Filter chain (alternative to a preset) + * + * If this is NULL, the preset above is used. Otherwise the preset + * is ignored and the filter chain specified here is used. + */ + const lzma_filter *filters; + + /** + * \brief Integrity check type + * + * See check.h for available checks. The xz command line tool + * defaults to LZMA_CHECK_CRC64, which is a good choice if you + * are unsure. + */ + lzma_check check; + + /* + * Reserved space to allow possible future extensions without + * breaking the ABI. You should not touch these, because the names + * of these variables may change. These are and will never be used + * with the currently supported options, so it is safe to leave these + * uninitialized. + */ + lzma_reserved_enum reserved_enum1; + lzma_reserved_enum reserved_enum2; + lzma_reserved_enum reserved_enum3; + uint32_t reserved_int1; + uint32_t reserved_int2; + uint32_t reserved_int3; + uint32_t reserved_int4; + uint64_t reserved_int5; + uint64_t reserved_int6; + uint64_t reserved_int7; + uint64_t reserved_int8; + void *reserved_ptr1; + void *reserved_ptr2; + void *reserved_ptr3; + void *reserved_ptr4; + +} lzma_mt; + + +/** * \brief Calculate approximate memory usage of easy encoder * * This function is a wrapper for lzma_raw_encoder_memusage(). @@ -165,7 +290,8 @@ extern LZMA_API(lzma_ret) lzma_easy_encoder( */ extern LZMA_API(lzma_ret) lzma_easy_buffer_encode( uint32_t preset, lzma_check check, - lzma_allocator *allocator, const uint8_t *in, size_t in_size, + const lzma_allocator *allocator, + const uint8_t *in, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow; @@ -191,6 +317,49 @@ extern LZMA_API(lzma_ret) lzma_stream_encoder(lzma_stream *strm, /** + * \brief Calculate approximate memory usage of multithreaded .xz encoder + * + * Since doing the encoding in threaded mode doesn't affect the memory + * requirements of single-threaded decompressor, you can use + * lzma_easy_decoder_memusage(options->preset) or + * lzma_raw_decoder_memusage(options->filters) to calculate + * the decompressor memory requirements. + * + * \param options Compression options + * + * \return Number of bytes of memory required for encoding with the + * given options. If an error occurs, for example due to + * unsupported preset or filter chain, UINT64_MAX is returned. + */ +extern LZMA_API(uint64_t) lzma_stream_encoder_mt_memusage( + const lzma_mt *options) lzma_nothrow lzma_attr_pure; + + +/** + * \brief Initialize multithreaded .xz Stream encoder + * + * This provides the functionality of lzma_easy_encoder() and + * lzma_stream_encoder() as a single function for multithreaded use. + * + * The supported actions for lzma_code() are LZMA_RUN, LZMA_FULL_FLUSH, + * LZMA_FULL_BARRIER, and LZMA_FINISH. Support for LZMA_SYNC_FLUSH might be + * added in the future. + * + * \param strm Pointer to properly prepared lzma_stream + * \param options Pointer to multithreaded compression options + * + * \return - LZMA_OK + * - LZMA_MEM_ERROR + * - LZMA_UNSUPPORTED_CHECK + * - LZMA_OPTIONS_ERROR + * - LZMA_PROG_ERROR + */ +extern LZMA_API(lzma_ret) lzma_stream_encoder_mt( + lzma_stream *strm, const lzma_mt *options) + lzma_nothrow lzma_attr_warn_unused_result; + + +/** * \brief Initialize .lzma encoder (legacy file format) * * The .lzma format is sometimes called the LZMA_Alone format, which is the @@ -269,7 +438,8 @@ extern LZMA_API(size_t) lzma_stream_buffer_bound(size_t uncompressed_size) */ extern LZMA_API(lzma_ret) lzma_stream_buffer_encode( lzma_filter *filters, lzma_check check, - lzma_allocator *allocator, const uint8_t *in, size_t in_size, + const lzma_allocator *allocator, + const uint8_t *in, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow lzma_attr_warn_unused_result; @@ -305,6 +475,30 @@ extern LZMA_API(lzma_ret) lzma_stream_buffer_encode( /** + * This flag makes lzma_code() not calculate and verify the integrity check + * of the compressed data in .xz files. This means that invalid integrity + * check values won't be detected and LZMA_DATA_ERROR won't be returned in + * such cases. + * + * This flag only affects the checks of the compressed data itself; the CRC32 + * values in the .xz headers will still be verified normally. + * + * Don't use this flag unless you know what you are doing. Possible reasons + * to use this flag: + * + * - Trying to recover data from a corrupt .xz file. + * + * - Speeding up decompression, which matters mostly with SHA-256 + * or with files that have compressed extremely well. It's recommended + * to not use this flag for this purpose unless the file integrity is + * verified externally in some other way. + * + * Support for this flag was added in liblzma 5.1.4beta. + */ +#define LZMA_IGNORE_CHECK UINT32_C(0x10) + + +/** * This flag enables decoding of concatenated files with file formats that * allow concatenating compressed files as is. From the formats currently * supported by liblzma, only the .xz format allows concatenated files. @@ -418,7 +612,8 @@ extern LZMA_API(lzma_ret) lzma_alone_decoder( * - LZMA_PROG_ERROR */ extern LZMA_API(lzma_ret) lzma_stream_buffer_decode( - uint64_t *memlimit, uint32_t flags, lzma_allocator *allocator, + uint64_t *memlimit, uint32_t flags, + const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow lzma_attr_warn_unused_result; diff --git a/src/liblzma/api/lzma/filter.h b/src/liblzma/api/lzma/filter.h index e0bc163ad35a..4e78752b8f79 100644 --- a/src/liblzma/api/lzma/filter.h +++ b/src/liblzma/api/lzma/filter.h @@ -116,8 +116,9 @@ extern LZMA_API(lzma_bool) lzma_filter_decoder_is_supported(lzma_vli id) * is not NULL. * - LZMA_PROG_ERROR: src or dest is NULL. */ -extern LZMA_API(lzma_ret) lzma_filters_copy(const lzma_filter *src, - lzma_filter *dest, lzma_allocator *allocator) lzma_nothrow; +extern LZMA_API(lzma_ret) lzma_filters_copy( + const lzma_filter *src, lzma_filter *dest, + const lzma_allocator *allocator) lzma_nothrow; /** @@ -256,7 +257,7 @@ extern LZMA_API(lzma_ret) lzma_filters_update( * won't necessarily meet that bound.) */ extern LZMA_API(lzma_ret) lzma_raw_buffer_encode( - const lzma_filter *filters, lzma_allocator *allocator, + const lzma_filter *filters, const lzma_allocator *allocator, const uint8_t *in, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow; @@ -280,7 +281,7 @@ extern LZMA_API(lzma_ret) lzma_raw_buffer_encode( * which no data is written to is out[out_size]. */ extern LZMA_API(lzma_ret) lzma_raw_buffer_decode( - const lzma_filter *filters, lzma_allocator *allocator, + const lzma_filter *filters, const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow; @@ -356,7 +357,7 @@ extern LZMA_API(lzma_ret) lzma_properties_encode( * - LZMA_MEM_ERROR */ extern LZMA_API(lzma_ret) lzma_properties_decode( - lzma_filter *filter, lzma_allocator *allocator, + lzma_filter *filter, const lzma_allocator *allocator, const uint8_t *props, size_t props_size) lzma_nothrow; @@ -419,6 +420,6 @@ extern LZMA_API(lzma_ret) lzma_filter_flags_encode(const lzma_filter *filter, * - LZMA_PROG_ERROR */ extern LZMA_API(lzma_ret) lzma_filter_flags_decode( - lzma_filter *filter, lzma_allocator *allocator, + lzma_filter *filter, const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size) lzma_nothrow lzma_attr_warn_unused_result; diff --git a/src/liblzma/api/lzma/hardware.h b/src/liblzma/api/lzma/hardware.h index e7dd03c3e8dd..5321d9af8e46 100644 --- a/src/liblzma/api/lzma/hardware.h +++ b/src/liblzma/api/lzma/hardware.h @@ -48,3 +48,17 @@ * of RAM on the specific operating system. */ extern LZMA_API(uint64_t) lzma_physmem(void) lzma_nothrow; + + +/** + * \brief Get the number of processor cores or threads + * + * This function may be useful when determining how many threads to use. + * If the hardware supports more than one thread per CPU core, the number + * of hardware threads is returned if that information is available. + * + * \brief On success, the number of available CPU threads or cores is + * returned. If this information isn't available or an error + * occurs, zero is returned. + */ +extern LZMA_API(uint32_t) lzma_cputhreads(void) lzma_nothrow; diff --git a/src/liblzma/api/lzma/index.h b/src/liblzma/api/lzma/index.h index 16bacc287dfc..dda60ec1c185 100644 --- a/src/liblzma/api/lzma/index.h +++ b/src/liblzma/api/lzma/index.h @@ -303,7 +303,7 @@ extern LZMA_API(uint64_t) lzma_index_memused(const lzma_index *i) * \return On success, a pointer to an empty initialized lzma_index is * returned. If allocation fails, NULL is returned. */ -extern LZMA_API(lzma_index *) lzma_index_init(lzma_allocator *allocator) +extern LZMA_API(lzma_index *) lzma_index_init(const lzma_allocator *allocator) lzma_nothrow; @@ -312,8 +312,8 @@ extern LZMA_API(lzma_index *) lzma_index_init(lzma_allocator *allocator) * * If i is NULL, this does nothing. */ -extern LZMA_API(void) lzma_index_end(lzma_index *i, lzma_allocator *allocator) - lzma_nothrow; +extern LZMA_API(void) lzma_index_end( + lzma_index *i, const lzma_allocator *allocator) lzma_nothrow; /** @@ -341,7 +341,7 @@ extern LZMA_API(void) lzma_index_end(lzma_index *i, lzma_allocator *allocator) * - LZMA_PROG_ERROR */ extern LZMA_API(lzma_ret) lzma_index_append( - lzma_index *i, lzma_allocator *allocator, + lzma_index *i, const lzma_allocator *allocator, lzma_vli unpadded_size, lzma_vli uncompressed_size) lzma_nothrow lzma_attr_warn_unused_result; @@ -564,8 +564,8 @@ extern LZMA_API(lzma_bool) lzma_index_iter_locate( * - LZMA_MEM_ERROR * - LZMA_PROG_ERROR */ -extern LZMA_API(lzma_ret) lzma_index_cat( - lzma_index *dest, lzma_index *src, lzma_allocator *allocator) +extern LZMA_API(lzma_ret) lzma_index_cat(lzma_index *dest, lzma_index *src, + const lzma_allocator *allocator) lzma_nothrow lzma_attr_warn_unused_result; @@ -575,7 +575,7 @@ extern LZMA_API(lzma_ret) lzma_index_cat( * \return A copy of the lzma_index, or NULL if memory allocation failed. */ extern LZMA_API(lzma_index *) lzma_index_dup( - const lzma_index *i, lzma_allocator *allocator) + const lzma_index *i, const lzma_allocator *allocator) lzma_nothrow lzma_attr_warn_unused_result; @@ -677,6 +677,6 @@ extern LZMA_API(lzma_ret) lzma_index_buffer_encode(const lzma_index *i, * - LZMA_PROG_ERROR */ extern LZMA_API(lzma_ret) lzma_index_buffer_decode(lzma_index **i, - uint64_t *memlimit, lzma_allocator *allocator, + uint64_t *memlimit, const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size) lzma_nothrow; diff --git a/src/liblzma/api/lzma/index_hash.h b/src/liblzma/api/lzma/index_hash.h index fa2e048d552d..9287f1dfdb56 100644 --- a/src/liblzma/api/lzma/index_hash.h +++ b/src/liblzma/api/lzma/index_hash.h @@ -37,7 +37,7 @@ typedef struct lzma_index_hash_s lzma_index_hash; * pointer than the index_hash that was given as an argument. */ extern LZMA_API(lzma_index_hash *) lzma_index_hash_init( - lzma_index_hash *index_hash, lzma_allocator *allocator) + lzma_index_hash *index_hash, const lzma_allocator *allocator) lzma_nothrow lzma_attr_warn_unused_result; @@ -45,7 +45,7 @@ extern LZMA_API(lzma_index_hash *) lzma_index_hash_init( * \brief Deallocate lzma_index_hash structure */ extern LZMA_API(void) lzma_index_hash_end( - lzma_index_hash *index_hash, lzma_allocator *allocator) + lzma_index_hash *index_hash, const lzma_allocator *allocator) lzma_nothrow; diff --git a/src/liblzma/api/lzma/lzma.h b/src/liblzma/api/lzma/lzma12.h index 3f8e095f70e8..4e32fa3a214a 100644 --- a/src/liblzma/api/lzma/lzma.h +++ b/src/liblzma/api/lzma/lzma12.h @@ -1,5 +1,5 @@ /** - * \file lzma/lzma.h + * \file lzma/lzma12.h * \brief LZMA1 and LZMA2 filters */ diff --git a/src/liblzma/api/lzma/version.h b/src/liblzma/api/lzma/version.h index beec18e20440..d9614da76f7b 100644 --- a/src/liblzma/api/lzma/version.h +++ b/src/liblzma/api/lzma/version.h @@ -21,8 +21,8 @@ * Version number split into components */ #define LZMA_VERSION_MAJOR 5 -#define LZMA_VERSION_MINOR 0 -#define LZMA_VERSION_PATCH 7 +#define LZMA_VERSION_MINOR 2 +#define LZMA_VERSION_PATCH 0 #define LZMA_VERSION_STABILITY LZMA_VERSION_STABILITY_STABLE #ifndef LZMA_VERSION_COMMIT diff --git a/src/liblzma/check/check.h b/src/liblzma/check/check.h index e100d2b85303..0f96f65b360c 100644 --- a/src/liblzma/check/check.h +++ b/src/liblzma/check/check.h @@ -15,6 +15,43 @@ #include "common.h" +#if defined(HAVE_COMMONCRYPTO_COMMONDIGEST_H) +# include <CommonCrypto/CommonDigest.h> +#elif defined(HAVE_SHA256_H) +# include <sys/types.h> +# include <sha256.h> +#elif defined(HAVE_SHA2_H) +# include <sys/types.h> +# include <sha2.h> +#elif defined(HAVE_MINIX_SHA2_H) +# include <sys/types.h> +# include <minix/sha2.h> +#endif + +#if defined(HAVE_CC_SHA256_CTX) +typedef CC_SHA256_CTX lzma_sha256_state; +#elif defined(HAVE_SHA256_CTX) +typedef SHA256_CTX lzma_sha256_state; +#elif defined(HAVE_SHA2_CTX) +typedef SHA2_CTX lzma_sha256_state; +#else +/// State for the internal SHA-256 implementation +typedef struct { + /// Internal state + uint32_t state[8]; + + /// Size of the message excluding padding + uint64_t size; +} lzma_sha256_state; +#endif + +#if defined(HAVE_CC_SHA256_INIT) +# define LZMA_SHA256FUNC(x) CC_SHA256_ ## x +#elif defined(HAVE_SHA256_INIT) +# define LZMA_SHA256FUNC(x) SHA256_ ## x +#elif defined(HAVE_SHA256INIT) +# define LZMA_SHA256FUNC(x) SHA256 ## x +#endif // Index hashing needs the best possible hash function (preferably // a cryptographic hash) for maximum reliability. @@ -43,14 +80,7 @@ typedef struct { union { uint32_t crc32; uint64_t crc64; - - struct { - /// Internal state - uint32_t state[8]; - - /// Size of the message excluding padding - uint64_t size; - } sha256; + lzma_sha256_state sha256; } state; } lzma_check_state; @@ -82,6 +112,8 @@ extern void lzma_check_update(lzma_check_state *check, lzma_check type, extern void lzma_check_finish(lzma_check_state *check, lzma_check type); +#ifndef LZMA_SHA256FUNC + /// Prepare SHA-256 state for new input. extern void lzma_sha256_init(lzma_check_state *check); @@ -92,4 +124,39 @@ extern void lzma_sha256_update( /// Finish the SHA-256 calculation and store the result to check->buffer.u8. extern void lzma_sha256_finish(lzma_check_state *check); + +#else + +static inline void +lzma_sha256_init(lzma_check_state *check) +{ + LZMA_SHA256FUNC(Init)(&check->state.sha256); +} + + +static inline void +lzma_sha256_update(const uint8_t *buf, size_t size, lzma_check_state *check) +{ +#if defined(HAVE_CC_SHA256_INIT) && SIZE_MAX > UINT32_MAX + // Darwin's CC_SHA256_Update takes uint32_t as the buffer size, + // so use a loop to support size_t. + while (size > UINT32_MAX) { + LZMA_SHA256FUNC(Update)(&check->state.sha256, buf, UINT32_MAX); + buf += UINT32_MAX; + size -= UINT32_MAX; + } +#endif + + LZMA_SHA256FUNC(Update)(&check->state.sha256, buf, size); +} + + +static inline void +lzma_sha256_finish(lzma_check_state *check) +{ + LZMA_SHA256FUNC(Final)(check->buffer.u8, &check->state.sha256); +} + +#endif + #endif diff --git a/src/liblzma/check/sha256.c b/src/liblzma/check/sha256.c index f2cc0d71ac63..5eede5ce05bc 100644 --- a/src/liblzma/check/sha256.c +++ b/src/liblzma/check/sha256.c @@ -21,22 +21,22 @@ // /////////////////////////////////////////////////////////////////////////////// -// Avoid bogus warnings in transform(). -#if (__GNUC__ == 4 && __GNUC_MINOR__ >= 2) || __GNUC__ > 4 -# pragma GCC diagnostic ignored "-Wuninitialized" -#endif - #include "check.h" -// At least on x86, GCC is able to optimize this to a rotate instruction. -#define rotr_32(num, amount) ((num) >> (amount) | (num) << (32 - (amount))) +// Rotate a uint32_t. GCC can optimize this to a rotate instruction +// at least on x86. +static inline uint32_t +rotr_32(uint32_t num, unsigned amount) +{ + return (num >> amount) | (num << (32 - amount)); +} -#define blk0(i) (W[i] = data[i]) +#define blk0(i) (W[i] = conv32be(data[i])) #define blk2(i) (W[i & 15] += s1(W[(i - 2) & 15]) + W[(i - 7) & 15] \ + s0(W[(i - 15) & 15])) #define Ch(x, y, z) (z ^ (x & (y ^ z))) -#define Maj(x, y, z) ((x & y) | (z & (x | y))) +#define Maj(x, y, z) ((x & (y ^ z)) + (y & z)) #define a(i) T[(0 - i) & 7] #define b(i) T[(1 - i) & 7] @@ -47,16 +47,17 @@ #define g(i) T[(6 - i) & 7] #define h(i) T[(7 - i) & 7] -#define R(i) \ - h(i) += S1(e(i)) + Ch(e(i), f(i), g(i)) + SHA256_K[i + j] \ - + (j ? blk2(i) : blk0(i)); \ +#define R(i, j, blk) \ + h(i) += S1(e(i)) + Ch(e(i), f(i), g(i)) + SHA256_K[i + j] + blk; \ d(i) += h(i); \ h(i) += S0(a(i)) + Maj(a(i), b(i), c(i)) +#define R0(i) R(i, 0, blk0(i)) +#define R2(i) R(i, j, blk2(i)) -#define S0(x) (rotr_32(x, 2) ^ rotr_32(x, 13) ^ rotr_32(x, 22)) -#define S1(x) (rotr_32(x, 6) ^ rotr_32(x, 11) ^ rotr_32(x, 25)) -#define s0(x) (rotr_32(x, 7) ^ rotr_32(x, 18) ^ (x >> 3)) -#define s1(x) (rotr_32(x, 17) ^ rotr_32(x, 19) ^ (x >> 10)) +#define S0(x) rotr_32(x ^ rotr_32(x ^ rotr_32(x, 9), 11), 2) +#define S1(x) rotr_32(x ^ rotr_32(x ^ rotr_32(x, 14), 5), 6) +#define s0(x) (rotr_32(x ^ rotr_32(x, 11), 7) ^ (x >> 3)) +#define s1(x) (rotr_32(x ^ rotr_32(x, 2), 17) ^ (x >> 10)) static const uint32_t SHA256_K[64] = { @@ -88,12 +89,18 @@ transform(uint32_t state[8], const uint32_t data[16]) // Copy state[] to working vars. memcpy(T, state, sizeof(T)); - // 64 operations, partially loop unrolled - for (unsigned int j = 0; j < 64; j += 16) { - R( 0); R( 1); R( 2); R( 3); - R( 4); R( 5); R( 6); R( 7); - R( 8); R( 9); R(10); R(11); - R(12); R(13); R(14); R(15); + // The first 16 operations unrolled + R0( 0); R0( 1); R0( 2); R0( 3); + R0( 4); R0( 5); R0( 6); R0( 7); + R0( 8); R0( 9); R0(10); R0(11); + R0(12); R0(13); R0(14); R0(15); + + // The remaining 48 operations partially unrolled + for (unsigned int j = 16; j < 64; j += 16) { + R2( 0); R2( 1); R2( 2); R2( 3); + R2( 4); R2( 5); R2( 6); R2( 7); + R2( 8); R2( 9); R2(10); R2(11); + R2(12); R2(13); R2(14); R2(15); } // Add the working vars back into state[]. @@ -111,18 +118,7 @@ transform(uint32_t state[8], const uint32_t data[16]) static void process(lzma_check_state *check) { -#ifdef WORDS_BIGENDIAN transform(check->state.sha256.state, check->buffer.u32); - -#else - uint32_t data[16]; - - for (size_t i = 0; i < 16; ++i) - data[i] = bswap32(check->buffer.u32[i]); - - transform(check->state.sha256.state, data); -#endif - return; } diff --git a/src/liblzma/common/alone_decoder.c b/src/liblzma/common/alone_decoder.c index c25112e6875f..c1360ca1eb7a 100644 --- a/src/liblzma/common/alone_decoder.c +++ b/src/liblzma/common/alone_decoder.c @@ -51,7 +51,7 @@ struct lzma_coder_s { static lzma_ret alone_decode(lzma_coder *coder, - lzma_allocator *allocator lzma_attribute((__unused__)), + const lzma_allocator *allocator lzma_attribute((__unused__)), const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, @@ -166,7 +166,7 @@ alone_decode(lzma_coder *coder, static void -alone_decoder_end(lzma_coder *coder, lzma_allocator *allocator) +alone_decoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder, allocator); @@ -193,7 +193,7 @@ alone_decoder_memconfig(lzma_coder *coder, uint64_t *memusage, extern lzma_ret -lzma_alone_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_alone_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, uint64_t memlimit, bool picky) { lzma_next_coder_init(&lzma_alone_decoder_init, next, allocator); diff --git a/src/liblzma/common/alone_decoder.h b/src/liblzma/common/alone_decoder.h index f666fc3823e1..dfa031aa77dd 100644 --- a/src/liblzma/common/alone_decoder.h +++ b/src/liblzma/common/alone_decoder.h @@ -17,7 +17,7 @@ extern lzma_ret lzma_alone_decoder_init( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, uint64_t memlimit, bool picky); #endif diff --git a/src/liblzma/common/alone_encoder.c b/src/liblzma/common/alone_encoder.c index eb1697e99753..a2bc9eee1fa9 100644 --- a/src/liblzma/common/alone_encoder.c +++ b/src/liblzma/common/alone_encoder.c @@ -32,7 +32,7 @@ struct lzma_coder_s { static lzma_ret alone_encode(lzma_coder *coder, - lzma_allocator *allocator lzma_attribute((__unused__)), + const lzma_allocator *allocator lzma_attribute((__unused__)), const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, @@ -65,7 +65,7 @@ alone_encode(lzma_coder *coder, static void -alone_encoder_end(lzma_coder *coder, lzma_allocator *allocator) +alone_encoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder, allocator); @@ -75,7 +75,7 @@ alone_encoder_end(lzma_coder *coder, lzma_allocator *allocator) // At least for now, this is not used by any internal function. static lzma_ret -alone_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +alone_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_options_lzma *options) { lzma_next_coder_init(&alone_encoder_init, next, allocator); @@ -137,7 +137,7 @@ alone_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, /* extern lzma_ret -lzma_alone_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_alone_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_options_alone *options) { lzma_next_coder_init(&alone_encoder_init, next, allocator, options); diff --git a/src/liblzma/common/auto_decoder.c b/src/liblzma/common/auto_decoder.c index 35c895fd14c2..bf3550701fe6 100644 --- a/src/liblzma/common/auto_decoder.c +++ b/src/liblzma/common/auto_decoder.c @@ -30,7 +30,7 @@ struct lzma_coder_s { static lzma_ret -auto_decode(lzma_coder *coder, lzma_allocator *allocator, +auto_decode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -100,7 +100,7 @@ auto_decode(lzma_coder *coder, lzma_allocator *allocator, static void -auto_decoder_end(lzma_coder *coder, lzma_allocator *allocator) +auto_decoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder, allocator); @@ -143,7 +143,7 @@ auto_decoder_memconfig(lzma_coder *coder, uint64_t *memusage, static lzma_ret -auto_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +auto_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, uint64_t memlimit, uint32_t flags) { lzma_next_coder_init(&auto_decoder_init, next, allocator); diff --git a/src/liblzma/common/block_buffer_decoder.c b/src/liblzma/common/block_buffer_decoder.c index ff27a11ccfe6..b0ded90ddc3e 100644 --- a/src/liblzma/common/block_buffer_decoder.c +++ b/src/liblzma/common/block_buffer_decoder.c @@ -14,7 +14,7 @@ extern LZMA_API(lzma_ret) -lzma_block_buffer_decode(lzma_block *block, lzma_allocator *allocator, +lzma_block_buffer_decode(lzma_block *block, const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) { diff --git a/src/liblzma/common/block_buffer_encoder.c b/src/liblzma/common/block_buffer_encoder.c index 519c6a684d76..39e263aa4765 100644 --- a/src/liblzma/common/block_buffer_encoder.c +++ b/src/liblzma/common/block_buffer_encoder.c @@ -10,6 +10,7 @@ // /////////////////////////////////////////////////////////////////////////////// +#include "block_buffer_encoder.h" #include "block_encoder.h" #include "filter_encoder.h" #include "lzma2_encoder.h" @@ -28,8 +29,8 @@ + LZMA_CHECK_SIZE_MAX + 3) & ~3) -static lzma_vli -lzma2_bound(lzma_vli uncompressed_size) +static uint64_t +lzma2_bound(uint64_t uncompressed_size) { // Prevent integer overflow in overhead calculation. if (uncompressed_size > COMPRESSED_SIZE_MAX) @@ -39,7 +40,7 @@ lzma2_bound(lzma_vli uncompressed_size) // uncompressed_size up to the next multiple of LZMA2_CHUNK_MAX, // multiply by the size of per-chunk header, and add one byte for // the end marker. - const lzma_vli overhead = ((uncompressed_size + LZMA2_CHUNK_MAX - 1) + const uint64_t overhead = ((uncompressed_size + LZMA2_CHUNK_MAX - 1) / LZMA2_CHUNK_MAX) * LZMA2_HEADER_UNCOMPRESSED + 1; @@ -51,30 +52,36 @@ lzma2_bound(lzma_vli uncompressed_size) } -extern LZMA_API(size_t) -lzma_block_buffer_bound(size_t uncompressed_size) +extern uint64_t +lzma_block_buffer_bound64(uint64_t uncompressed_size) { - // For now, if the data doesn't compress, we always use uncompressed - // chunks of LZMA2. In future we may use Subblock filter too, but - // but for simplicity we probably will still use the same bound - // calculation even though Subblock filter would have slightly less - // overhead. - lzma_vli lzma2_size = lzma2_bound(uncompressed_size); + // If the data doesn't compress, we always use uncompressed + // LZMA2 chunks. + uint64_t lzma2_size = lzma2_bound(uncompressed_size); if (lzma2_size == 0) return 0; // Take Block Padding into account. - lzma2_size = (lzma2_size + 3) & ~LZMA_VLI_C(3); + lzma2_size = (lzma2_size + 3) & ~UINT64_C(3); -#if SIZE_MAX < LZMA_VLI_MAX - // Catch the possible integer overflow on 32-bit systems. There's no - // overflow on 64-bit systems, because lzma2_bound() already takes + // No risk of integer overflow because lzma2_bound() already takes // into account the size of the headers in the Block. - if (SIZE_MAX - HEADERS_BOUND < lzma2_size) + return HEADERS_BOUND + lzma2_size; +} + + +extern LZMA_API(size_t) +lzma_block_buffer_bound(size_t uncompressed_size) +{ + uint64_t ret = lzma_block_buffer_bound64(uncompressed_size); + +#if SIZE_MAX < UINT64_MAX + // Catch the possible integer overflow on 32-bit systems. + if (ret > SIZE_MAX) return 0; #endif - return HEADERS_BOUND + lzma2_size; + return ret; } @@ -82,9 +89,6 @@ static lzma_ret block_encode_uncompressed(lzma_block *block, const uint8_t *in, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) { - // TODO: Figure out if the last filter is LZMA2 or Subblock and use - // that filter to encode the uncompressed chunks. - // Use LZMA2 uncompressed chunks. We wouldn't need a dictionary at // all, but LZMA2 always requires a dictionary, so use the minimum // value to minimize memory usage of the decoder. @@ -160,16 +164,11 @@ block_encode_uncompressed(lzma_block *block, const uint8_t *in, size_t in_size, static lzma_ret -block_encode_normal(lzma_block *block, lzma_allocator *allocator, +block_encode_normal(lzma_block *block, const lzma_allocator *allocator, const uint8_t *in, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) { // Find out the size of the Block Header. - block->compressed_size = lzma2_bound(in_size); - if (block->compressed_size == 0) - return LZMA_DATA_ERROR; - - block->uncompressed_size = in_size; return_if_error(lzma_block_header_size(block)); // Reserve space for the Block Header and skip it for now. @@ -221,10 +220,11 @@ block_encode_normal(lzma_block *block, lzma_allocator *allocator, } -extern LZMA_API(lzma_ret) -lzma_block_buffer_encode(lzma_block *block, lzma_allocator *allocator, +static lzma_ret +block_buffer_encode(lzma_block *block, const lzma_allocator *allocator, const uint8_t *in, size_t in_size, - uint8_t *out, size_t *out_pos, size_t out_size) + uint8_t *out, size_t *out_pos, size_t out_size, + bool try_to_compress) { // Validate the arguments. if (block == NULL || (in == NULL && in_size != 0) || out == NULL @@ -233,11 +233,11 @@ lzma_block_buffer_encode(lzma_block *block, lzma_allocator *allocator, // The contents of the structure may depend on the version so // check the version before validating the contents of *block. - if (block->version != 0) + if (block->version > 1) return LZMA_OPTIONS_ERROR; if ((unsigned int)(block->check) > LZMA_CHECK_ID_MAX - || block->filters == NULL) + || (try_to_compress && block->filters == NULL)) return LZMA_PROG_ERROR; if (!lzma_check_is_supported(block->check)) @@ -258,9 +258,19 @@ lzma_block_buffer_encode(lzma_block *block, lzma_allocator *allocator, out_size -= check_size; + // Initialize block->uncompressed_size and calculate the worst-case + // value for block->compressed_size. + block->uncompressed_size = in_size; + block->compressed_size = lzma2_bound(in_size); + if (block->compressed_size == 0) + return LZMA_DATA_ERROR; + // Do the actual compression. - const lzma_ret ret = block_encode_normal(block, allocator, - in, in_size, out, out_pos, out_size); + lzma_ret ret = LZMA_BUF_ERROR; + if (try_to_compress) + ret = block_encode_normal(block, allocator, + in, in_size, out, out_pos, out_size); + if (ret != LZMA_OK) { // If the error was something else than output buffer // becoming full, return the error now. @@ -303,3 +313,25 @@ lzma_block_buffer_encode(lzma_block *block, lzma_allocator *allocator, return LZMA_OK; } + + +extern LZMA_API(lzma_ret) +lzma_block_buffer_encode(lzma_block *block, const lzma_allocator *allocator, + const uint8_t *in, size_t in_size, + uint8_t *out, size_t *out_pos, size_t out_size) +{ + return block_buffer_encode(block, allocator, + in, in_size, out, out_pos, out_size, true); +} + + +extern LZMA_API(lzma_ret) +lzma_block_uncomp_encode(lzma_block *block, + const uint8_t *in, size_t in_size, + uint8_t *out, size_t *out_pos, size_t out_size) +{ + // It won't allocate any memory from heap so no need + // for lzma_allocator. + return block_buffer_encode(block, NULL, + in, in_size, out, out_pos, out_size, false); +} diff --git a/src/liblzma/common/block_buffer_encoder.h b/src/liblzma/common/block_buffer_encoder.h new file mode 100644 index 000000000000..653207f73498 --- /dev/null +++ b/src/liblzma/common/block_buffer_encoder.h @@ -0,0 +1,24 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file block_buffer_encoder.h +/// \brief Single-call .xz Block encoder +// +// Author: Lasse Collin +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + +#ifndef LZMA_BLOCK_BUFFER_ENCODER_H +#define LZMA_BLOCK_BUFFER_ENCODER_H + +#include "common.h" + + +/// uint64_t version of lzma_block_buffer_bound(). It is used by +/// stream_encoder_mt.c. Probably the original lzma_block_buffer_bound() +/// should have been 64-bit, but fixing it would break the ABI. +extern uint64_t lzma_block_buffer_bound64(uint64_t uncompressed_size); + +#endif diff --git a/src/liblzma/common/block_decoder.c b/src/liblzma/common/block_decoder.c index a3ce6f49500c..685c3b038fc3 100644 --- a/src/liblzma/common/block_decoder.c +++ b/src/liblzma/common/block_decoder.c @@ -45,6 +45,9 @@ struct lzma_coder_s { /// Check of the uncompressed data lzma_check_state check; + + /// True if the integrity check won't be calculated and verified. + bool ignore_check; }; @@ -71,7 +74,7 @@ is_size_valid(lzma_vli size, lzma_vli reference) static lzma_ret -block_decode(lzma_coder *coder, lzma_allocator *allocator, +block_decode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -97,8 +100,9 @@ block_decode(lzma_coder *coder, lzma_allocator *allocator, coder->block->uncompressed_size)) return LZMA_DATA_ERROR; - lzma_check_update(&coder->check, coder->block->check, - out + out_start, out_used); + if (!coder->ignore_check) + lzma_check_update(&coder->check, coder->block->check, + out + out_start, out_used); if (ret != LZMA_STREAM_END) return ret; @@ -140,7 +144,9 @@ block_decode(lzma_coder *coder, lzma_allocator *allocator, if (coder->block->check == LZMA_CHECK_NONE) return LZMA_STREAM_END; - lzma_check_finish(&coder->check, coder->block->check); + if (!coder->ignore_check) + lzma_check_finish(&coder->check, coder->block->check); + coder->sequence = SEQ_CHECK; // Fall through @@ -155,7 +161,8 @@ block_decode(lzma_coder *coder, lzma_allocator *allocator, // Validate the Check only if we support it. // coder->check.buffer may be uninitialized // when the Check ID is not supported. - if (lzma_check_is_supported(coder->block->check) + if (!coder->ignore_check + && lzma_check_is_supported(coder->block->check) && memcmp(coder->block->raw_check, coder->check.buffer.u8, check_size) != 0) @@ -170,7 +177,7 @@ block_decode(lzma_coder *coder, lzma_allocator *allocator, static void -block_decoder_end(lzma_coder *coder, lzma_allocator *allocator) +block_decoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder, allocator); @@ -179,7 +186,7 @@ block_decoder_end(lzma_coder *coder, lzma_allocator *allocator) extern lzma_ret -lzma_block_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_block_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, lzma_block *block) { lzma_next_coder_init(&lzma_block_decoder_init, next, allocator); @@ -224,6 +231,9 @@ lzma_block_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, next->coder->check_pos = 0; lzma_check_init(&next->coder->check, block->check); + next->coder->ignore_check = block->version >= 1 + ? block->ignore_check : false; + // Initialize the filter chain. return lzma_raw_decoder_init(&next->coder->next, allocator, block->filters); diff --git a/src/liblzma/common/block_decoder.h b/src/liblzma/common/block_decoder.h index 7da9df63f767..718c5ced886c 100644 --- a/src/liblzma/common/block_decoder.h +++ b/src/liblzma/common/block_decoder.h @@ -17,6 +17,6 @@ extern lzma_ret lzma_block_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, lzma_block *block); + const lzma_allocator *allocator, lzma_block *block); #endif diff --git a/src/liblzma/common/block_encoder.c b/src/liblzma/common/block_encoder.c index 1eeb502b7f83..def586410d28 100644 --- a/src/liblzma/common/block_encoder.c +++ b/src/liblzma/common/block_encoder.c @@ -45,7 +45,7 @@ struct lzma_coder_s { static lzma_ret -block_encode(lzma_coder *coder, lzma_allocator *allocator, +block_encode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -134,7 +134,7 @@ block_encode(lzma_coder *coder, lzma_allocator *allocator, static void -block_encoder_end(lzma_coder *coder, lzma_allocator *allocator) +block_encoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder, allocator); @@ -143,7 +143,7 @@ block_encoder_end(lzma_coder *coder, lzma_allocator *allocator) static lzma_ret -block_encoder_update(lzma_coder *coder, lzma_allocator *allocator, +block_encoder_update(lzma_coder *coder, const lzma_allocator *allocator, const lzma_filter *filters lzma_attribute((__unused__)), const lzma_filter *reversed_filters) { @@ -156,7 +156,7 @@ block_encoder_update(lzma_coder *coder, lzma_allocator *allocator, extern lzma_ret -lzma_block_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_block_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, lzma_block *block) { lzma_next_coder_init(&lzma_block_encoder_init, next, allocator); @@ -166,7 +166,7 @@ lzma_block_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, // The contents of the structure may depend on the version so // check the version first. - if (block->version != 0) + if (block->version > 1) return LZMA_OPTIONS_ERROR; // If the Check ID is not supported, we cannot calculate the check and diff --git a/src/liblzma/common/block_encoder.h b/src/liblzma/common/block_encoder.h index b9eff0be2736..bd97c186e503 100644 --- a/src/liblzma/common/block_encoder.h +++ b/src/liblzma/common/block_encoder.h @@ -42,6 +42,6 @@ extern lzma_ret lzma_block_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, lzma_block *block); + const lzma_allocator *allocator, lzma_block *block); #endif diff --git a/src/liblzma/common/block_header_decoder.c b/src/liblzma/common/block_header_decoder.c index 2c9573ee204c..1dd982f6bd68 100644 --- a/src/liblzma/common/block_header_decoder.c +++ b/src/liblzma/common/block_header_decoder.c @@ -15,7 +15,7 @@ static void -free_properties(lzma_block *block, lzma_allocator *allocator) +free_properties(lzma_block *block, const lzma_allocator *allocator) { // Free allocated filter options. The last array member is not // touched after the initialization in the beginning of @@ -32,7 +32,7 @@ free_properties(lzma_block *block, lzma_allocator *allocator) extern LZMA_API(lzma_ret) lzma_block_header_decode(lzma_block *block, - lzma_allocator *allocator, const uint8_t *in) + const lzma_allocator *allocator, const uint8_t *in) { // NOTE: We consider the header to be corrupt not only when the // CRC32 doesn't match, but also when variable-length integers @@ -46,8 +46,16 @@ lzma_block_header_decode(lzma_block *block, block->filters[i].options = NULL; } - // Always zero for now. - block->version = 0; + // Versions 0 and 1 are supported. If a newer version was specified, + // we need to downgrade it. + if (block->version > 1) + block->version = 1; + + // This isn't a Block Header option, but since the decompressor will + // read it if version >= 1, it's better to initialize it here than + // to expect the caller to do it since in almost all cases this + // should be false. + block->ignore_check = false; // Validate Block Header Size and Check type. The caller must have // already set these, so it is a programming error if this test fails. diff --git a/src/liblzma/common/block_header_encoder.c b/src/liblzma/common/block_header_encoder.c index 707dd0cb14a2..5c5f5424ae85 100644 --- a/src/liblzma/common/block_header_encoder.c +++ b/src/liblzma/common/block_header_encoder.c @@ -17,7 +17,7 @@ extern LZMA_API(lzma_ret) lzma_block_header_size(lzma_block *block) { - if (block->version != 0) + if (block->version > 1) return LZMA_OPTIONS_ERROR; // Block Header Size + Block Flags + CRC32. diff --git a/src/liblzma/common/block_util.c b/src/liblzma/common/block_util.c index 62c934547c9e..00c7fe8d5196 100644 --- a/src/liblzma/common/block_util.c +++ b/src/liblzma/common/block_util.c @@ -51,7 +51,7 @@ lzma_block_unpadded_size(const lzma_block *block) // NOTE: This function is used for validation too, so it is // essential that these checks are always done even if // Compressed Size is unknown. - if (block == NULL || block->version != 0 + if (block == NULL || block->version > 1 || block->header_size < LZMA_BLOCK_HEADER_SIZE_MIN || block->header_size > LZMA_BLOCK_HEADER_SIZE_MAX || (block->header_size & 3) diff --git a/src/liblzma/common/common.c b/src/liblzma/common/common.c index b9e386027368..28aa2b7142f4 100644 --- a/src/liblzma/common/common.c +++ b/src/liblzma/common/common.c @@ -36,7 +36,7 @@ lzma_version_string(void) /////////////////////// extern void * lzma_attribute((__malloc__)) lzma_attr_alloc_size(1) -lzma_alloc(size_t size, lzma_allocator *allocator) +lzma_alloc(size_t size, const lzma_allocator *allocator) { // Some malloc() variants return NULL if called with size == 0. if (size == 0) @@ -53,8 +53,29 @@ lzma_alloc(size_t size, lzma_allocator *allocator) } +extern void * lzma_attribute((__malloc__)) lzma_attr_alloc_size(1) +lzma_alloc_zero(size_t size, const lzma_allocator *allocator) +{ + // Some calloc() variants return NULL if called with size == 0. + if (size == 0) + size = 1; + + void *ptr; + + if (allocator != NULL && allocator->alloc != NULL) { + ptr = allocator->alloc(allocator->opaque, 1, size); + if (ptr != NULL) + memzero(ptr, size); + } else { + ptr = calloc(1, size); + } + + return ptr; +} + + extern void -lzma_free(void *ptr, lzma_allocator *allocator) +lzma_free(void *ptr, const lzma_allocator *allocator) { if (allocator != NULL && allocator->free != NULL) allocator->free(allocator->opaque, ptr); @@ -88,7 +109,7 @@ lzma_bufcpy(const uint8_t *restrict in, size_t *restrict in_pos, extern lzma_ret -lzma_next_filter_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_next_filter_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { lzma_next_coder_init(filters[0].init, next, allocator); @@ -99,7 +120,7 @@ lzma_next_filter_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret -lzma_next_filter_update(lzma_next_coder *next, lzma_allocator *allocator, +lzma_next_filter_update(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *reversed_filters) { // Check that the application isn't trying to change the Filter ID. @@ -117,7 +138,7 @@ lzma_next_filter_update(lzma_next_coder *next, lzma_allocator *allocator, extern void -lzma_next_end(lzma_next_coder *next, lzma_allocator *allocator) +lzma_next_end(lzma_next_coder *next, const lzma_allocator *allocator) { if (next->init != (uintptr_t)(NULL)) { // To avoid tiny end functions that simply call @@ -156,10 +177,8 @@ lzma_strm_init(lzma_stream *strm) strm->internal->next = LZMA_NEXT_CODER_INIT; } - strm->internal->supported_actions[LZMA_RUN] = false; - strm->internal->supported_actions[LZMA_SYNC_FLUSH] = false; - strm->internal->supported_actions[LZMA_FULL_FLUSH] = false; - strm->internal->supported_actions[LZMA_FINISH] = false; + memzero(strm->internal->supported_actions, + sizeof(strm->internal->supported_actions)); strm->internal->sequence = ISEQ_RUN; strm->internal->allow_buf_error = false; @@ -178,7 +197,7 @@ lzma_code(lzma_stream *strm, lzma_action action) || (strm->next_out == NULL && strm->avail_out != 0) || strm->internal == NULL || strm->internal->next.code == NULL - || (unsigned int)(action) > LZMA_FINISH + || (unsigned int)(action) > LZMA_ACTION_MAX || !strm->internal->supported_actions[action]) return LZMA_PROG_ERROR; @@ -213,6 +232,10 @@ lzma_code(lzma_stream *strm, lzma_action action) case LZMA_FINISH: strm->internal->sequence = ISEQ_FINISH; break; + + case LZMA_FULL_BARRIER: + strm->internal->sequence = ISEQ_FULL_BARRIER; + break; } break; @@ -240,6 +263,13 @@ lzma_code(lzma_stream *strm, lzma_action action) break; + case ISEQ_FULL_BARRIER: + if (action != LZMA_FULL_BARRIER + || strm->internal->avail_in != strm->avail_in) + return LZMA_PROG_ERROR; + + break; + case ISEQ_END: return LZMA_STREAM_END; @@ -265,7 +295,9 @@ lzma_code(lzma_stream *strm, lzma_action action) strm->internal->avail_in = strm->avail_in; - switch (ret) { + // Cast is needed to silence a warning about LZMA_TIMED_OUT, which + // isn't part of lzma_ret enumeration. + switch ((unsigned int)(ret)) { case LZMA_OK: // Don't return LZMA_BUF_ERROR when it happens the first time. // This is to avoid returning LZMA_BUF_ERROR when avail_out @@ -281,9 +313,16 @@ lzma_code(lzma_stream *strm, lzma_action action) } break; + case LZMA_TIMED_OUT: + strm->internal->allow_buf_error = false; + ret = LZMA_OK; + break; + case LZMA_STREAM_END: if (strm->internal->sequence == ISEQ_SYNC_FLUSH - || strm->internal->sequence == ISEQ_FULL_FLUSH) + || strm->internal->sequence == ISEQ_FULL_FLUSH + || strm->internal->sequence + == ISEQ_FULL_BARRIER) strm->internal->sequence = ISEQ_RUN; else strm->internal->sequence = ISEQ_END; @@ -323,6 +362,22 @@ lzma_end(lzma_stream *strm) } +extern LZMA_API(void) +lzma_get_progress(lzma_stream *strm, + uint64_t *progress_in, uint64_t *progress_out) +{ + if (strm->internal->next.get_progress != NULL) { + strm->internal->next.get_progress(strm->internal->next.coder, + progress_in, progress_out); + } else { + *progress_in = strm->total_in; + *progress_out = strm->total_out; + } + + return; +} + + extern LZMA_API(lzma_check) lzma_get_check(const lzma_stream *strm) { diff --git a/src/liblzma/common/common.h b/src/liblzma/common/common.h index 45aba4f06b27..955d784a5b6a 100644 --- a/src/liblzma/common/common.h +++ b/src/liblzma/common/common.h @@ -49,6 +49,13 @@ #define LZMA_BUFFER_SIZE 4096 +/// Maximum number of worker threads within one multithreaded component. +/// The limit exists solely to make it simpler to prevent integer overflows +/// when allocating structures etc. This should be big enough for now... +/// the code won't scale anywhere close to this number anyway. +#define LZMA_THREADS_MAX 16384 + + /// Starting value for memory usage estimates. Instead of calculating size /// of _every_ structure and taking into account malloc() overhead etc., we /// add a base size to all memory usage estimates. It's not very accurate @@ -66,9 +73,21 @@ ( LZMA_TELL_NO_CHECK \ | LZMA_TELL_UNSUPPORTED_CHECK \ | LZMA_TELL_ANY_CHECK \ + | LZMA_IGNORE_CHECK \ | LZMA_CONCATENATED ) +/// Largest valid lzma_action value as unsigned integer. +#define LZMA_ACTION_MAX ((unsigned int)(LZMA_FULL_BARRIER)) + + +/// Special return value (lzma_ret) to indicate that a timeout was reached +/// and lzma_code() must not return LZMA_BUF_ERROR. This is converted to +/// LZMA_OK in lzma_code(). This is not in the lzma_ret enumeration because +/// there's no need to have it in the public API. +#define LZMA_TIMED_OUT 32 + + /// Type of encoder/decoder specific data; the actual structure is defined /// differently in different coders. typedef struct lzma_coder_s lzma_coder; @@ -80,7 +99,7 @@ typedef struct lzma_filter_info_s lzma_filter_info; /// Type of a function used to initialize a filter encoder or decoder typedef lzma_ret (*lzma_init_function)( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters); /// Type of a function to do some kind of coding work (filters, Stream, @@ -88,7 +107,7 @@ typedef lzma_ret (*lzma_init_function)( /// input and output buffers, but for simplicity they still use this same /// function prototype. typedef lzma_ret (*lzma_code_function)( - lzma_coder *coder, lzma_allocator *allocator, + lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, @@ -96,7 +115,7 @@ typedef lzma_ret (*lzma_code_function)( /// Type of a function to free the memory allocated for the coder typedef void (*lzma_end_function)( - lzma_coder *coder, lzma_allocator *allocator); + lzma_coder *coder, const lzma_allocator *allocator); /// Raw coder validates and converts an array of lzma_filter structures to @@ -139,6 +158,11 @@ struct lzma_next_coder_s { /// lzma_next_coder.coder. lzma_end_function end; + /// Pointer to a function to get progress information. If this is NULL, + /// lzma_stream.total_in and .total_out are used instead. + void (*get_progress)(lzma_coder *coder, + uint64_t *progress_in, uint64_t *progress_out); + /// Pointer to function to return the type of the integrity check. /// Most coders won't support this. lzma_check (*get_check)(const lzma_coder *coder); @@ -150,7 +174,7 @@ struct lzma_next_coder_s { /// Update the filter-specific options or the whole filter chain /// in the encoder. - lzma_ret (*update)(lzma_coder *coder, lzma_allocator *allocator, + lzma_ret (*update)(lzma_coder *coder, const lzma_allocator *allocator, const lzma_filter *filters, const lzma_filter *reversed_filters); }; @@ -164,6 +188,7 @@ struct lzma_next_coder_s { .id = LZMA_VLI_UNKNOWN, \ .code = NULL, \ .end = NULL, \ + .get_progress = NULL, \ .get_check = NULL, \ .memconfig = NULL, \ .update = NULL, \ @@ -185,6 +210,7 @@ struct lzma_internal_s { ISEQ_SYNC_FLUSH, ISEQ_FULL_FLUSH, ISEQ_FINISH, + ISEQ_FULL_BARRIER, ISEQ_END, ISEQ_ERROR, } sequence; @@ -195,7 +221,7 @@ struct lzma_internal_s { size_t avail_in; /// Indicates which lzma_action values are allowed by next.code. - bool supported_actions[4]; + bool supported_actions[LZMA_ACTION_MAX + 1]; /// If true, lzma_code will return LZMA_BUF_ERROR if no progress was /// made (no input consumed and no output produced by next.code). @@ -204,11 +230,17 @@ struct lzma_internal_s { /// Allocates memory -extern void *lzma_alloc(size_t size, lzma_allocator *allocator) +extern void *lzma_alloc(size_t size, const lzma_allocator *allocator) lzma_attribute((__malloc__)) lzma_attr_alloc_size(1); +/// Allocates memory and zeroes it (like calloc()). This can be faster +/// than lzma_alloc() + memzero() while being backward compatible with +/// custom allocators. +extern void * lzma_attribute((__malloc__)) lzma_attr_alloc_size(1) + lzma_alloc_zero(size_t size, const lzma_allocator *allocator); + /// Frees memory -extern void lzma_free(void *ptr, lzma_allocator *allocator); +extern void lzma_free(void *ptr, const lzma_allocator *allocator); /// Allocates strm->internal if it is NULL, and initializes *strm and @@ -220,17 +252,19 @@ extern lzma_ret lzma_strm_init(lzma_stream *strm); /// than the filter being initialized now. This way the actual filter /// initialization functions don't need to use lzma_next_coder_init macro. extern lzma_ret lzma_next_filter_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); /// Update the next filter in the chain, if any. This checks that /// the application is not trying to change the Filter IDs. extern lzma_ret lzma_next_filter_update( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *reversed_filters); /// Frees the memory allocated for next->coder either using next->end or, /// if next->end is NULL, using lzma_free. -extern void lzma_next_end(lzma_next_coder *next, lzma_allocator *allocator); +extern void lzma_next_end(lzma_next_coder *next, + const lzma_allocator *allocator); /// Copy as much data as possible from in[] to out[] and update *in_pos diff --git a/src/liblzma/common/easy_buffer_encoder.c b/src/liblzma/common/easy_buffer_encoder.c index c4be34ccfa27..48eb56f5cc91 100644 --- a/src/liblzma/common/easy_buffer_encoder.c +++ b/src/liblzma/common/easy_buffer_encoder.c @@ -15,8 +15,8 @@ extern LZMA_API(lzma_ret) lzma_easy_buffer_encode(uint32_t preset, lzma_check check, - lzma_allocator *allocator, const uint8_t *in, size_t in_size, - uint8_t *out, size_t *out_pos, size_t out_size) + const lzma_allocator *allocator, const uint8_t *in, + size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) { lzma_options_easy opt_easy; if (lzma_easy_preset(&opt_easy, preset)) diff --git a/src/liblzma/common/easy_encoder.c b/src/liblzma/common/easy_encoder.c index d13ccd7351f1..5cb492dd0681 100644 --- a/src/liblzma/common/easy_encoder.c +++ b/src/liblzma/common/easy_encoder.c @@ -11,7 +11,6 @@ /////////////////////////////////////////////////////////////////////////////// #include "easy_preset.h" -#include "stream_encoder.h" extern LZMA_API(lzma_ret) diff --git a/src/liblzma/common/filter_buffer_decoder.c b/src/liblzma/common/filter_buffer_decoder.c index 2d35ef8e0af4..6620986eea8a 100644 --- a/src/liblzma/common/filter_buffer_decoder.c +++ b/src/liblzma/common/filter_buffer_decoder.c @@ -14,7 +14,8 @@ extern LZMA_API(lzma_ret) -lzma_raw_buffer_decode(const lzma_filter *filters, lzma_allocator *allocator, +lzma_raw_buffer_decode( + const lzma_filter *filters, const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) { diff --git a/src/liblzma/common/filter_buffer_encoder.c b/src/liblzma/common/filter_buffer_encoder.c index 646e1b30374e..dda18e3d8e5e 100644 --- a/src/liblzma/common/filter_buffer_encoder.c +++ b/src/liblzma/common/filter_buffer_encoder.c @@ -14,9 +14,10 @@ extern LZMA_API(lzma_ret) -lzma_raw_buffer_encode(const lzma_filter *filters, lzma_allocator *allocator, - const uint8_t *in, size_t in_size, uint8_t *out, - size_t *out_pos, size_t out_size) +lzma_raw_buffer_encode( + const lzma_filter *filters, const lzma_allocator *allocator, + const uint8_t *in, size_t in_size, + uint8_t *out, size_t *out_pos, size_t out_size) { // Validate what isn't validated later in filter_common.c. if ((in == NULL && in_size != 0) || out == NULL diff --git a/src/liblzma/common/filter_common.c b/src/liblzma/common/filter_common.c index 7c95b05f2a3f..9ad5d5d8e2af 100644 --- a/src/liblzma/common/filter_common.c +++ b/src/liblzma/common/filter_common.c @@ -123,7 +123,7 @@ static const struct { extern LZMA_API(lzma_ret) lzma_filters_copy(const lzma_filter *src, lzma_filter *dest, - lzma_allocator *allocator) + const lzma_allocator *allocator) { if (src == NULL || dest == NULL) return LZMA_PROG_ERROR; @@ -239,7 +239,7 @@ validate_chain(const lzma_filter *filters, size_t *count) extern lzma_ret -lzma_raw_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_raw_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *options, lzma_filter_find coder_find, bool is_encoder) { diff --git a/src/liblzma/common/filter_common.h b/src/liblzma/common/filter_common.h index cd61fc0724f3..42a26a24a47f 100644 --- a/src/liblzma/common/filter_common.h +++ b/src/liblzma/common/filter_common.h @@ -36,7 +36,7 @@ typedef const lzma_filter_coder *(*lzma_filter_find)(lzma_vli id); extern lzma_ret lzma_raw_coder_init( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *filters, lzma_filter_find coder_find, bool is_encoder); diff --git a/src/liblzma/common/filter_decoder.c b/src/liblzma/common/filter_decoder.c index 1ebbe2afef0c..c75b0a89c30f 100644 --- a/src/liblzma/common/filter_decoder.c +++ b/src/liblzma/common/filter_decoder.c @@ -35,7 +35,8 @@ typedef struct { /// \return - LZMA_OK: Properties decoded successfully. /// - LZMA_OPTIONS_ERROR: Unsupported properties /// - LZMA_MEM_ERROR: Memory allocation failed. - lzma_ret (*props_decode)(void **options, lzma_allocator *allocator, + lzma_ret (*props_decode)( + void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size); } lzma_filter_decoder; @@ -136,7 +137,7 @@ lzma_filter_decoder_is_supported(lzma_vli id) extern lzma_ret -lzma_raw_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_raw_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *options) { return lzma_raw_coder_init(next, allocator, @@ -165,7 +166,7 @@ lzma_raw_decoder_memusage(const lzma_filter *filters) extern LZMA_API(lzma_ret) -lzma_properties_decode(lzma_filter *filter, lzma_allocator *allocator, +lzma_properties_decode(lzma_filter *filter, const lzma_allocator *allocator, const uint8_t *props, size_t props_size) { // Make it always NULL so that the caller can always safely free() it. diff --git a/src/liblzma/common/filter_decoder.h b/src/liblzma/common/filter_decoder.h index d5c68bdd4a68..a2e255fe5f06 100644 --- a/src/liblzma/common/filter_decoder.h +++ b/src/liblzma/common/filter_decoder.h @@ -17,7 +17,7 @@ extern lzma_ret lzma_raw_decoder_init( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *options); #endif diff --git a/src/liblzma/common/filter_encoder.c b/src/liblzma/common/filter_encoder.c index 635d81223472..c5d8f39721fe 100644 --- a/src/liblzma/common/filter_encoder.c +++ b/src/liblzma/common/filter_encoder.c @@ -30,11 +30,11 @@ typedef struct { /// invalid, UINT64_MAX is returned. uint64_t (*memusage)(const void *options); - /// Calculates the minimum sane size for Blocks (or other types of - /// chunks) to which the input data can be split to make - /// multithreaded encoding possible. If this is NULL, it is assumed - /// that the encoder is fast enough with single thread. - lzma_vli (*chunk_size)(const void *options); + /// Calculates the recommended Uncompressed Size for .xz Blocks to + /// which the input data can be split to make multithreaded + /// encoding possible. If this is NULL, it is assumed that + /// the encoder is fast enough with single thread. + uint64_t (*block_size)(const void *options); /// Tells the size of the Filter Properties field. If options are /// invalid, UINT32_MAX is returned. If this is NULL, props_size_fixed @@ -59,7 +59,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_LZMA1, .init = &lzma_lzma_encoder_init, .memusage = &lzma_lzma_encoder_memusage, - .chunk_size = NULL, // FIXME + .block_size = NULL, // FIXME .props_size_get = NULL, .props_size_fixed = 5, .props_encode = &lzma_lzma_props_encode, @@ -70,7 +70,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_LZMA2, .init = &lzma_lzma2_encoder_init, .memusage = &lzma_lzma2_encoder_memusage, - .chunk_size = NULL, // FIXME + .block_size = &lzma_lzma2_block_size, // FIXME .props_size_get = NULL, .props_size_fixed = 1, .props_encode = &lzma_lzma2_props_encode, @@ -81,7 +81,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_X86, .init = &lzma_simple_x86_encoder_init, .memusage = NULL, - .chunk_size = NULL, + .block_size = NULL, .props_size_get = &lzma_simple_props_size, .props_encode = &lzma_simple_props_encode, }, @@ -91,7 +91,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_POWERPC, .init = &lzma_simple_powerpc_encoder_init, .memusage = NULL, - .chunk_size = NULL, + .block_size = NULL, .props_size_get = &lzma_simple_props_size, .props_encode = &lzma_simple_props_encode, }, @@ -101,7 +101,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_IA64, .init = &lzma_simple_ia64_encoder_init, .memusage = NULL, - .chunk_size = NULL, + .block_size = NULL, .props_size_get = &lzma_simple_props_size, .props_encode = &lzma_simple_props_encode, }, @@ -111,7 +111,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_ARM, .init = &lzma_simple_arm_encoder_init, .memusage = NULL, - .chunk_size = NULL, + .block_size = NULL, .props_size_get = &lzma_simple_props_size, .props_encode = &lzma_simple_props_encode, }, @@ -121,7 +121,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_ARMTHUMB, .init = &lzma_simple_armthumb_encoder_init, .memusage = NULL, - .chunk_size = NULL, + .block_size = NULL, .props_size_get = &lzma_simple_props_size, .props_encode = &lzma_simple_props_encode, }, @@ -131,7 +131,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_SPARC, .init = &lzma_simple_sparc_encoder_init, .memusage = NULL, - .chunk_size = NULL, + .block_size = NULL, .props_size_get = &lzma_simple_props_size, .props_encode = &lzma_simple_props_encode, }, @@ -141,7 +141,7 @@ static const lzma_filter_encoder encoders[] = { .id = LZMA_FILTER_DELTA, .init = &lzma_delta_encoder_init, .memusage = &lzma_delta_coder_memusage, - .chunk_size = NULL, + .block_size = NULL, .props_size_get = NULL, .props_size_fixed = 1, .props_encode = &lzma_delta_props_encode, @@ -196,7 +196,7 @@ lzma_filters_update(lzma_stream *strm, const lzma_filter *filters) extern lzma_ret -lzma_raw_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_raw_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *options) { return lzma_raw_coder_init(next, allocator, @@ -226,20 +226,19 @@ lzma_raw_encoder_memusage(const lzma_filter *filters) } -/* -extern LZMA_API(lzma_vli) -lzma_chunk_size(const lzma_filter *filters) +extern uint64_t +lzma_mt_block_size(const lzma_filter *filters) { - lzma_vli max = 0; + uint64_t max = 0; for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i) { const lzma_filter_encoder *const fe = encoder_find(filters[i].id); - if (fe->chunk_size != NULL) { - const lzma_vli size - = fe->chunk_size(filters[i].options); - if (size == LZMA_VLI_UNKNOWN) - return LZMA_VLI_UNKNOWN; + if (fe->block_size != NULL) { + const uint64_t size + = fe->block_size(filters[i].options); + if (size == 0) + return 0; if (size > max) max = size; @@ -248,7 +247,6 @@ lzma_chunk_size(const lzma_filter *filters) return max; } -*/ extern LZMA_API(lzma_ret) diff --git a/src/liblzma/common/filter_encoder.h b/src/liblzma/common/filter_encoder.h index 5bc137f64584..f1d5683fe793 100644 --- a/src/liblzma/common/filter_encoder.h +++ b/src/liblzma/common/filter_encoder.h @@ -16,12 +16,12 @@ #include "common.h" -// FIXME: Might become a part of the public API once finished. -// extern lzma_vli lzma_chunk_size(const lzma_filter *filters); +// FIXME: Might become a part of the public API. +extern uint64_t lzma_mt_block_size(const lzma_filter *filters); extern lzma_ret lzma_raw_encoder_init( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *filters); #endif diff --git a/src/liblzma/common/filter_flags_decoder.c b/src/liblzma/common/filter_flags_decoder.c index caae10ce79a0..ddfb085943d0 100644 --- a/src/liblzma/common/filter_flags_decoder.c +++ b/src/liblzma/common/filter_flags_decoder.c @@ -15,7 +15,7 @@ extern LZMA_API(lzma_ret) lzma_filter_flags_decode( - lzma_filter *filter, lzma_allocator *allocator, + lzma_filter *filter, const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size) { // Set the pointer to NULL so the caller can always safely free it. diff --git a/src/liblzma/common/stream_encoder.h b/src/liblzma/common/hardware_cputhreads.c index 46a7aed72eaa..f468366a6045 100644 --- a/src/liblzma/common/stream_encoder.h +++ b/src/liblzma/common/hardware_cputhreads.c @@ -1,7 +1,7 @@ /////////////////////////////////////////////////////////////////////////////// // -/// \file stream_encoder.h -/// \brief Encodes .xz Streams +/// \file hardware_cputhreads.c +/// \brief Get the number of CPU threads or cores // // Author: Lasse Collin // @@ -10,14 +10,13 @@ // /////////////////////////////////////////////////////////////////////////////// -#ifndef LZMA_STREAM_ENCODER_H -#define LZMA_STREAM_ENCODER_H - #include "common.h" +#include "tuklib_cpucores.h" -extern lzma_ret lzma_stream_encoder_init( - lzma_next_coder *next, lzma_allocator *allocator, - const lzma_filter *filters, lzma_check check); -#endif +extern LZMA_API(uint32_t) +lzma_cputhreads(void) +{ + return tuklib_cpucores(); +} diff --git a/src/liblzma/common/index.c b/src/liblzma/common/index.c index 9af4bc19d1b5..11f45f4009ea 100644 --- a/src/liblzma/common/index.c +++ b/src/liblzma/common/index.c @@ -191,8 +191,8 @@ index_tree_init(index_tree *tree) /// Helper for index_tree_end() static void -index_tree_node_end(index_tree_node *node, lzma_allocator *allocator, - void (*free_func)(void *node, lzma_allocator *allocator)) +index_tree_node_end(index_tree_node *node, const lzma_allocator *allocator, + void (*free_func)(void *node, const lzma_allocator *allocator)) { // The tree won't ever be very huge, so recursion should be fine. // 20 levels in the tree is likely quite a lot already in practice. @@ -215,8 +215,8 @@ index_tree_node_end(index_tree_node *node, lzma_allocator *allocator, /// to free the Record groups from each index_stream before freeing /// the index_stream itself. static void -index_tree_end(index_tree *tree, lzma_allocator *allocator, - void (*free_func)(void *node, lzma_allocator *allocator)) +index_tree_end(index_tree *tree, const lzma_allocator *allocator, + void (*free_func)(void *node, const lzma_allocator *allocator)) { if (tree->root != NULL) index_tree_node_end(tree->root, allocator, free_func); @@ -340,7 +340,7 @@ index_tree_locate(const index_tree *tree, lzma_vli target) static index_stream * index_stream_init(lzma_vli compressed_base, lzma_vli uncompressed_base, lzma_vli stream_number, lzma_vli block_number_base, - lzma_allocator *allocator) + const lzma_allocator *allocator) { index_stream *s = lzma_alloc(sizeof(index_stream), allocator); if (s == NULL) @@ -368,7 +368,7 @@ index_stream_init(lzma_vli compressed_base, lzma_vli uncompressed_base, /// Free the memory allocated for a Stream and its Record groups. static void -index_stream_end(void *node, lzma_allocator *allocator) +index_stream_end(void *node, const lzma_allocator *allocator) { index_stream *s = node; index_tree_end(&s->groups, allocator, NULL); @@ -377,7 +377,7 @@ index_stream_end(void *node, lzma_allocator *allocator) static lzma_index * -index_init_plain(lzma_allocator *allocator) +index_init_plain(const lzma_allocator *allocator) { lzma_index *i = lzma_alloc(sizeof(lzma_index), allocator); if (i != NULL) { @@ -395,7 +395,7 @@ index_init_plain(lzma_allocator *allocator) extern LZMA_API(lzma_index *) -lzma_index_init(lzma_allocator *allocator) +lzma_index_init(const lzma_allocator *allocator) { lzma_index *i = index_init_plain(allocator); if (i == NULL) @@ -414,7 +414,7 @@ lzma_index_init(lzma_allocator *allocator) extern LZMA_API(void) -lzma_index_end(lzma_index *i, lzma_allocator *allocator) +lzma_index_end(lzma_index *i, const lzma_allocator *allocator) { // NOTE: If you modify this function, check also the bottom // of lzma_index_cat(). @@ -637,7 +637,7 @@ lzma_index_stream_padding(lzma_index *i, lzma_vli stream_padding) extern LZMA_API(lzma_ret) -lzma_index_append(lzma_index *i, lzma_allocator *allocator, +lzma_index_append(lzma_index *i, const lzma_allocator *allocator, lzma_vli unpadded_size, lzma_vli uncompressed_size) { // Validate. @@ -765,7 +765,7 @@ index_cat_helper(const index_cat_info *info, index_stream *this) extern LZMA_API(lzma_ret) lzma_index_cat(lzma_index *restrict dest, lzma_index *restrict src, - lzma_allocator *allocator) + const lzma_allocator *allocator) { const lzma_vli dest_file_size = lzma_index_file_size(dest); @@ -859,7 +859,7 @@ lzma_index_cat(lzma_index *restrict dest, lzma_index *restrict src, /// Duplicate an index_stream. static index_stream * -index_dup_stream(const index_stream *src, lzma_allocator *allocator) +index_dup_stream(const index_stream *src, const lzma_allocator *allocator) { // Catch a somewhat theoretical integer overflow. if (src->record_count > PREALLOC_MAX) @@ -919,7 +919,7 @@ index_dup_stream(const index_stream *src, lzma_allocator *allocator) extern LZMA_API(lzma_index *) -lzma_index_dup(const lzma_index *src, lzma_allocator *allocator) +lzma_index_dup(const lzma_index *src, const lzma_allocator *allocator) { // Allocate the base structure (no initial Stream). lzma_index *dest = index_init_plain(allocator); diff --git a/src/liblzma/common/index_decoder.c b/src/liblzma/common/index_decoder.c index 83c8a3af1db8..795d1834cc58 100644 --- a/src/liblzma/common/index_decoder.c +++ b/src/liblzma/common/index_decoder.c @@ -54,7 +54,7 @@ struct lzma_coder_s { static lzma_ret -index_decode(lzma_coder *coder, lzma_allocator *allocator, +index_decode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out lzma_attribute((__unused__)), @@ -207,7 +207,7 @@ out: static void -index_decoder_end(lzma_coder *coder, lzma_allocator *allocator) +index_decoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_index_end(coder->index, allocator); lzma_free(coder, allocator); @@ -234,7 +234,7 @@ index_decoder_memconfig(lzma_coder *coder, uint64_t *memusage, static lzma_ret -index_decoder_reset(lzma_coder *coder, lzma_allocator *allocator, +index_decoder_reset(lzma_coder *coder, const lzma_allocator *allocator, lzma_index **i, uint64_t memlimit) { // Remember the pointer given by the application. We will set it @@ -261,7 +261,7 @@ index_decoder_reset(lzma_coder *coder, lzma_allocator *allocator, static lzma_ret -index_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +index_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, lzma_index **i, uint64_t memlimit) { lzma_next_coder_init(&index_decoder_init, next, allocator); @@ -299,8 +299,8 @@ lzma_index_decoder(lzma_stream *strm, lzma_index **i, uint64_t memlimit) extern LZMA_API(lzma_ret) -lzma_index_buffer_decode( - lzma_index **i, uint64_t *memlimit, lzma_allocator *allocator, +lzma_index_buffer_decode(lzma_index **i, uint64_t *memlimit, + const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size) { // Sanity checks diff --git a/src/liblzma/common/index_encoder.c b/src/liblzma/common/index_encoder.c index 45919f094c42..d25ac7d3372b 100644 --- a/src/liblzma/common/index_encoder.c +++ b/src/liblzma/common/index_encoder.c @@ -42,7 +42,7 @@ struct lzma_coder_s { static lzma_ret index_encode(lzma_coder *coder, - lzma_allocator *allocator lzma_attribute((__unused__)), + const lzma_allocator *allocator lzma_attribute((__unused__)), const uint8_t *restrict in lzma_attribute((__unused__)), size_t *restrict in_pos lzma_attribute((__unused__)), size_t in_size lzma_attribute((__unused__)), @@ -159,7 +159,7 @@ out: static void -index_encoder_end(lzma_coder *coder, lzma_allocator *allocator) +index_encoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_free(coder, allocator); return; @@ -181,7 +181,7 @@ index_encoder_reset(lzma_coder *coder, const lzma_index *i) extern lzma_ret -lzma_index_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_index_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_index *i) { lzma_next_coder_init(&lzma_index_encoder_init, next, allocator); diff --git a/src/liblzma/common/index_encoder.h b/src/liblzma/common/index_encoder.h index a13c94dcd077..4d55cd104785 100644 --- a/src/liblzma/common/index_encoder.h +++ b/src/liblzma/common/index_encoder.h @@ -17,7 +17,7 @@ extern lzma_ret lzma_index_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_index *i); + const lzma_allocator *allocator, const lzma_index *i); #endif diff --git a/src/liblzma/common/index_hash.c b/src/liblzma/common/index_hash.c index e3e9386ae8e3..d7a0344b76c3 100644 --- a/src/liblzma/common/index_hash.c +++ b/src/liblzma/common/index_hash.c @@ -70,7 +70,8 @@ struct lzma_index_hash_s { extern LZMA_API(lzma_index_hash *) -lzma_index_hash_init(lzma_index_hash *index_hash, lzma_allocator *allocator) +lzma_index_hash_init(lzma_index_hash *index_hash, + const lzma_allocator *allocator) { if (index_hash == NULL) { index_hash = lzma_alloc(sizeof(lzma_index_hash), allocator); @@ -101,7 +102,8 @@ lzma_index_hash_init(lzma_index_hash *index_hash, lzma_allocator *allocator) extern LZMA_API(void) -lzma_index_hash_end(lzma_index_hash *index_hash, lzma_allocator *allocator) +lzma_index_hash_end(lzma_index_hash *index_hash, + const lzma_allocator *allocator) { lzma_free(index_hash, allocator); return; diff --git a/src/liblzma/common/memcmplen.h b/src/liblzma/common/memcmplen.h new file mode 100644 index 000000000000..f66e7cdb66c6 --- /dev/null +++ b/src/liblzma/common/memcmplen.h @@ -0,0 +1,170 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file memcmplen.h +/// \brief Optimized comparison of two buffers +// +// Author: Lasse Collin +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + +#ifndef LZMA_MEMCMPLEN_H +#define LZMA_MEMCMPLEN_H + +#include "common.h" + +#ifdef HAVE_IMMINTRIN_H +# include <immintrin.h> +#endif + +/// How many extra bytes lzma_memcmplen() may read. This depends on +/// the method but since it is just a few bytes the biggest possible +/// value is used here. +#define LZMA_MEMCMPLEN_EXTRA 16 + + +/// Find out how many equal bytes the two buffers have. +/// +/// \param buf1 First buffer +/// \param buf2 Second buffer +/// \param len How many bytes have already been compared and will +/// be assumed to match +/// \param limit How many bytes to compare at most, including the +/// already-compared bytes. This must be significantly +/// smaller than UINT32_MAX to avoid integer overflows. +/// Up to LZMA_MEMCMPLEN_EXTRA bytes may be read past +/// the specified limit from both buf1 and buf2. +/// +/// \return Number of equal bytes in the buffers is returned. +/// This is always at least len and at most limit. +static inline uint32_t lzma_attribute((__always_inline__)) +lzma_memcmplen(const uint8_t *buf1, const uint8_t *buf2, + uint32_t len, uint32_t limit) +{ + assert(len <= limit); + assert(limit <= UINT32_MAX / 2); + +#if defined(TUKLIB_FAST_UNALIGNED_ACCESS) \ + && ((TUKLIB_GNUC_REQ(3, 4) && defined(__x86_64__)) \ + || (defined(__INTEL_COMPILER) && defined(__x86_64__)) \ + || (defined(__INTEL_COMPILER) && defined(_M_X64)) \ + || (defined(_MSC_VER) && defined(_M_X64))) + // NOTE: This will use 64-bit unaligned access which + // TUKLIB_FAST_UNALIGNED_ACCESS wasn't meant to permit, but + // it's convenient here at least as long as it's x86-64 only. + // + // I keep this x86-64 only for now since that's where I know this + // to be a good method. This may be fine on other 64-bit CPUs too. + // On big endian one should use xor instead of subtraction and switch + // to __builtin_clzll(). + while (len < limit) { + const uint64_t x = *(const uint64_t *)(buf1 + len) + - *(const uint64_t *)(buf2 + len); + if (x != 0) { +# if defined(_M_X64) // MSVC or Intel C compiler on Windows + unsigned long tmp; + _BitScanForward64(&tmp, x); + len += (uint32_t)tmp >> 3; +# else // GCC, clang, or Intel C compiler + len += (uint32_t)__builtin_ctzll(x) >> 3; +# endif + return my_min(len, limit); + } + + len += 8; + } + + return limit; + +#elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) \ + && defined(HAVE__MM_MOVEMASK_EPI8) \ + && ((defined(__GNUC__) && defined(__SSE2_MATH__)) \ + || (defined(__INTEL_COMPILER) && defined(__SSE2__)) \ + || (defined(_MSC_VER) && defined(_M_IX86_FP) \ + && _M_IX86_FP >= 2)) + // NOTE: Like above, this will use 128-bit unaligned access which + // TUKLIB_FAST_UNALIGNED_ACCESS wasn't meant to permit. + // + // SSE2 version for 32-bit and 64-bit x86. On x86-64 the above + // version is sometimes significantly faster and sometimes + // slightly slower than this SSE2 version, so this SSE2 + // version isn't used on x86-64. + while (len < limit) { + const uint32_t x = 0xFFFF ^ _mm_movemask_epi8(_mm_cmpeq_epi8( + _mm_loadu_si128((const __m128i *)(buf1 + len)), + _mm_loadu_si128((const __m128i *)(buf2 + len)))); + + if (x != 0) { +# if defined(__INTEL_COMPILER) + len += _bit_scan_forward(x); +# elif defined(_MSC_VER) + unsigned long tmp; + _BitScanForward(&tmp, x); + len += tmp; +# else + len += __builtin_ctz(x); +# endif + return my_min(len, limit); + } + + len += 16; + } + + return limit; + +#elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) && !defined(WORDS_BIGENDIAN) + // Generic 32-bit little endian method + while (len < limit) { + uint32_t x = *(const uint32_t *)(buf1 + len) + - *(const uint32_t *)(buf2 + len); + if (x != 0) { + if ((x & 0xFFFF) == 0) { + len += 2; + x >>= 16; + } + + if ((x & 0xFF) == 0) + ++len; + + return my_min(len, limit); + } + + len += 4; + } + + return limit; + +#elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) && defined(WORDS_BIGENDIAN) + // Generic 32-bit big endian method + while (len < limit) { + uint32_t x = *(const uint32_t *)(buf1 + len) + ^ *(const uint32_t *)(buf2 + len); + if (x != 0) { + if ((x & 0xFFFF0000) == 0) { + len += 2; + x <<= 16; + } + + if ((x & 0xFF000000) == 0) + ++len; + + return my_min(len, limit); + } + + len += 4; + } + + return limit; + +#else + // Simple portable version that doesn't use unaligned access. + while (len < limit && buf1[len] == buf2[len]) + ++len; + + return len; +#endif +} + +#endif diff --git a/src/liblzma/common/outqueue.c b/src/liblzma/common/outqueue.c new file mode 100644 index 000000000000..2dc8a38d1be3 --- /dev/null +++ b/src/liblzma/common/outqueue.c @@ -0,0 +1,184 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file outqueue.c +/// \brief Output queue handling in multithreaded coding +// +// Author: Lasse Collin +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + +#include "outqueue.h" + + +/// This is to ease integer overflow checking: We may allocate up to +/// 2 * LZMA_THREADS_MAX buffers and we need some extra memory for other +/// data structures (that's the second /2). +#define BUF_SIZE_MAX (UINT64_MAX / LZMA_THREADS_MAX / 2 / 2) + + +static lzma_ret +get_options(uint64_t *bufs_alloc_size, uint32_t *bufs_count, + uint64_t buf_size_max, uint32_t threads) +{ + if (threads > LZMA_THREADS_MAX || buf_size_max > BUF_SIZE_MAX) + return LZMA_OPTIONS_ERROR; + + // The number of buffers is twice the number of threads. + // This wastes RAM but keeps the threads busy when buffers + // finish out of order. + // + // NOTE: If this is changed, update BUF_SIZE_MAX too. + *bufs_count = threads * 2; + *bufs_alloc_size = *bufs_count * buf_size_max; + + return LZMA_OK; +} + + +extern uint64_t +lzma_outq_memusage(uint64_t buf_size_max, uint32_t threads) +{ + uint64_t bufs_alloc_size; + uint32_t bufs_count; + + if (get_options(&bufs_alloc_size, &bufs_count, buf_size_max, threads) + != LZMA_OK) + return UINT64_MAX; + + return sizeof(lzma_outq) + bufs_count * sizeof(lzma_outbuf) + + bufs_alloc_size; +} + + +extern lzma_ret +lzma_outq_init(lzma_outq *outq, const lzma_allocator *allocator, + uint64_t buf_size_max, uint32_t threads) +{ + uint64_t bufs_alloc_size; + uint32_t bufs_count; + + // Set bufs_count and bufs_alloc_size. + return_if_error(get_options(&bufs_alloc_size, &bufs_count, + buf_size_max, threads)); + + // Allocate memory if needed. + if (outq->buf_size_max != buf_size_max + || outq->bufs_allocated != bufs_count) { + lzma_outq_end(outq, allocator); + +#if SIZE_MAX < UINT64_MAX + if (bufs_alloc_size > SIZE_MAX) + return LZMA_MEM_ERROR; +#endif + + outq->bufs = lzma_alloc(bufs_count * sizeof(lzma_outbuf), + allocator); + outq->bufs_mem = lzma_alloc((size_t)(bufs_alloc_size), + allocator); + + if (outq->bufs == NULL || outq->bufs_mem == NULL) { + lzma_outq_end(outq, allocator); + return LZMA_MEM_ERROR; + } + } + + // Initialize the rest of the main structure. Initialization of + // outq->bufs[] is done when they are actually needed. + outq->buf_size_max = (size_t)(buf_size_max); + outq->bufs_allocated = bufs_count; + outq->bufs_pos = 0; + outq->bufs_used = 0; + outq->read_pos = 0; + + return LZMA_OK; +} + + +extern void +lzma_outq_end(lzma_outq *outq, const lzma_allocator *allocator) +{ + lzma_free(outq->bufs, allocator); + outq->bufs = NULL; + + lzma_free(outq->bufs_mem, allocator); + outq->bufs_mem = NULL; + + return; +} + + +extern lzma_outbuf * +lzma_outq_get_buf(lzma_outq *outq) +{ + // Caller must have checked it with lzma_outq_has_buf(). + assert(outq->bufs_used < outq->bufs_allocated); + + // Initialize the new buffer. + lzma_outbuf *buf = &outq->bufs[outq->bufs_pos]; + buf->buf = outq->bufs_mem + outq->bufs_pos * outq->buf_size_max; + buf->size = 0; + buf->finished = false; + + // Update the queue state. + if (++outq->bufs_pos == outq->bufs_allocated) + outq->bufs_pos = 0; + + ++outq->bufs_used; + + return buf; +} + + +extern bool +lzma_outq_is_readable(const lzma_outq *outq) +{ + uint32_t i = outq->bufs_pos - outq->bufs_used; + if (outq->bufs_pos < outq->bufs_used) + i += outq->bufs_allocated; + + return outq->bufs[i].finished; +} + + +extern lzma_ret +lzma_outq_read(lzma_outq *restrict outq, uint8_t *restrict out, + size_t *restrict out_pos, size_t out_size, + lzma_vli *restrict unpadded_size, + lzma_vli *restrict uncompressed_size) +{ + // There must be at least one buffer from which to read. + if (outq->bufs_used == 0) + return LZMA_OK; + + // Get the buffer. + uint32_t i = outq->bufs_pos - outq->bufs_used; + if (outq->bufs_pos < outq->bufs_used) + i += outq->bufs_allocated; + + lzma_outbuf *buf = &outq->bufs[i]; + + // If it isn't finished yet, we cannot read from it. + if (!buf->finished) + return LZMA_OK; + + // Copy from the buffer to output. + lzma_bufcpy(buf->buf, &outq->read_pos, buf->size, + out, out_pos, out_size); + + // Return if we didn't get all the data from the buffer. + if (outq->read_pos < buf->size) + return LZMA_OK; + + // The buffer was finished. Tell the caller its size information. + *unpadded_size = buf->unpadded_size; + *uncompressed_size = buf->uncompressed_size; + + // Free this buffer for further use. + --outq->bufs_used; + outq->read_pos = 0; + + return LZMA_STREAM_END; +} diff --git a/src/liblzma/common/outqueue.h b/src/liblzma/common/outqueue.h new file mode 100644 index 000000000000..079634de4587 --- /dev/null +++ b/src/liblzma/common/outqueue.h @@ -0,0 +1,156 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file outqueue.h +/// \brief Output queue handling in multithreaded coding +// +// Author: Lasse Collin +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + +#include "common.h" + + +/// Output buffer for a single thread +typedef struct { + /// Pointer to the output buffer of lzma_outq.buf_size_max bytes + uint8_t *buf; + + /// Amount of data written to buf + size_t size; + + /// Additional size information + lzma_vli unpadded_size; + lzma_vli uncompressed_size; + + /// True when no more data will be written into this buffer. + /// + /// \note This is read by another thread and thus access + /// to this variable needs a mutex. + bool finished; + +} lzma_outbuf; + + +typedef struct { + /// Array of buffers that are used cyclically. + lzma_outbuf *bufs; + + /// Memory allocated for all the buffers + uint8_t *bufs_mem; + + /// Amount of buffer space available in each buffer + size_t buf_size_max; + + /// Number of buffers allocated + uint32_t bufs_allocated; + + /// Position in the bufs array. The next buffer to be taken + /// into use is bufs[bufs_pos]. + uint32_t bufs_pos; + + /// Number of buffers in use + uint32_t bufs_used; + + /// Position in the buffer in lzma_outq_read() + size_t read_pos; + +} lzma_outq; + + +/** + * \brief Calculate the memory usage of an output queue + * + * \return Approximate memory usage in bytes or UINT64_MAX on error. + */ +extern uint64_t lzma_outq_memusage(uint64_t buf_size_max, uint32_t threads); + + +/// \brief Initialize an output queue +/// +/// \param outq Pointer to an output queue. Before calling +/// this function the first time, *outq should +/// have been zeroed with memzero() so that this +/// function knows that there are no previous +/// allocations to free. +/// \param allocator Pointer to allocator or NULL +/// \param buf_size_max Maximum amount of data that a single buffer +/// in the queue may need to store. +/// \param threads Number of buffers that may be in use +/// concurrently. Note that more than this number +/// of buffers will actually get allocated to +/// improve performance when buffers finish +/// out of order. +/// +/// \return - LZMA_OK +/// - LZMA_MEM_ERROR +/// +extern lzma_ret lzma_outq_init( + lzma_outq *outq, const lzma_allocator *allocator, + uint64_t buf_size_max, uint32_t threads); + + +/// \brief Free the memory associated with the output queue +extern void lzma_outq_end(lzma_outq *outq, const lzma_allocator *allocator); + + +/// \brief Get a new buffer +/// +/// lzma_outq_has_buf() must be used to check that there is a buffer +/// available before calling lzma_outq_get_buf(). +/// +extern lzma_outbuf *lzma_outq_get_buf(lzma_outq *outq); + + +/// \brief Test if there is data ready to be read +/// +/// Call to this function must be protected with the same mutex that +/// is used to protect lzma_outbuf.finished. +/// +extern bool lzma_outq_is_readable(const lzma_outq *outq); + + +/// \brief Read finished data +/// +/// \param outq Pointer to an output queue +/// \param out Beginning of the output buffer +/// \param out_pos The next byte will be written to +/// out[*out_pos]. +/// \param out_size Size of the out buffer; the first byte into +/// which no data is written to is out[out_size]. +/// \param unpadded_size Unpadded Size from the Block encoder +/// \param uncompressed_size Uncompressed Size from the Block encoder +/// +/// \return - LZMA: All OK. Either no data was available or the buffer +/// being read didn't become empty yet. +/// - LZMA_STREAM_END: The buffer being read was finished. +/// *unpadded_size and *uncompressed_size were set. +/// +/// \note This reads lzma_outbuf.finished variables and thus call +/// to this function needs to be protected with a mutex. +/// +extern lzma_ret lzma_outq_read(lzma_outq *restrict outq, + uint8_t *restrict out, size_t *restrict out_pos, + size_t out_size, lzma_vli *restrict unpadded_size, + lzma_vli *restrict uncompressed_size); + + +/// \brief Test if there is at least one buffer free +/// +/// This must be used before getting a new buffer with lzma_outq_get_buf(). +/// +static inline bool +lzma_outq_has_buf(const lzma_outq *outq) +{ + return outq->bufs_used < outq->bufs_allocated; +} + + +/// \brief Test if the queue is completely empty +static inline bool +lzma_outq_is_empty(const lzma_outq *outq) +{ + return outq->bufs_used == 0; +} diff --git a/src/liblzma/common/stream_buffer_decoder.c b/src/liblzma/common/stream_buffer_decoder.c index ae753155180d..b9745b5dbe18 100644 --- a/src/liblzma/common/stream_buffer_decoder.c +++ b/src/liblzma/common/stream_buffer_decoder.c @@ -15,7 +15,7 @@ extern LZMA_API(lzma_ret) lzma_stream_buffer_decode(uint64_t *memlimit, uint32_t flags, - lzma_allocator *allocator, + const lzma_allocator *allocator, const uint8_t *in, size_t *in_pos, size_t in_size, uint8_t *out, size_t *out_pos, size_t out_size) { diff --git a/src/liblzma/common/stream_buffer_encoder.c b/src/liblzma/common/stream_buffer_encoder.c index 2450ee2e1fa0..af49554a6b0c 100644 --- a/src/liblzma/common/stream_buffer_encoder.c +++ b/src/liblzma/common/stream_buffer_encoder.c @@ -42,7 +42,8 @@ lzma_stream_buffer_bound(size_t uncompressed_size) extern LZMA_API(lzma_ret) lzma_stream_buffer_encode(lzma_filter *filters, lzma_check check, - lzma_allocator *allocator, const uint8_t *in, size_t in_size, + const lzma_allocator *allocator, + const uint8_t *in, size_t in_size, uint8_t *out, size_t *out_pos_ptr, size_t out_size) { // Sanity checks diff --git a/src/liblzma/common/stream_decoder.c b/src/liblzma/common/stream_decoder.c index 37ea71edbd71..3ab938c9f142 100644 --- a/src/liblzma/common/stream_decoder.c +++ b/src/liblzma/common/stream_decoder.c @@ -57,6 +57,10 @@ struct lzma_coder_s { /// If true, LZMA_GET_CHECK is returned after decoding Stream Header. bool tell_any_check; + /// If true, we will tell the Block decoder to skip calculating + /// and verifying the integrity check. + bool ignore_check; + /// If true, we will decode concatenated Streams that possibly have /// Stream Padding between or after them. LZMA_STREAM_END is returned /// once the application isn't giving us any new input, and we aren't @@ -80,7 +84,7 @@ struct lzma_coder_s { static lzma_ret -stream_decoder_reset(lzma_coder *coder, lzma_allocator *allocator) +stream_decoder_reset(lzma_coder *coder, const lzma_allocator *allocator) { // Initialize the Index hash used to verify the Index. coder->index_hash = lzma_index_hash_init(coder->index_hash, allocator); @@ -96,7 +100,7 @@ stream_decoder_reset(lzma_coder *coder, lzma_allocator *allocator) static lzma_ret -stream_decode(lzma_coder *coder, lzma_allocator *allocator, +stream_decode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -182,8 +186,8 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator, coder->pos = 0; - // Version 0 is currently the only possible version. - coder->block_options.version = 0; + // Version 1 is needed to support the .ignore_check option. + coder->block_options.version = 1; // Set up a buffer to hold the filter chain. Block Header // decoder will initialize all members of this array so @@ -195,6 +199,11 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator, return_if_error(lzma_block_header_decode(&coder->block_options, allocator, coder->buffer)); + // If LZMA_IGNORE_CHECK was used, this flag needs to be set. + // It has to be set after lzma_block_header_decode() because + // it always resets this to false. + coder->block_options.ignore_check = coder->ignore_check; + // Check the memory usage limit. const uint64_t memusage = lzma_raw_decoder_memusage(filters); lzma_ret ret; @@ -366,7 +375,7 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator, static void -stream_decoder_end(lzma_coder *coder, lzma_allocator *allocator) +stream_decoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->block_decoder, allocator); lzma_index_hash_end(coder->index_hash, allocator); @@ -401,7 +410,8 @@ stream_decoder_memconfig(lzma_coder *coder, uint64_t *memusage, extern lzma_ret -lzma_stream_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_stream_decoder_init( + lzma_next_coder *next, const lzma_allocator *allocator, uint64_t memlimit, uint32_t flags) { lzma_next_coder_init(&lzma_stream_decoder_init, next, allocator); @@ -432,6 +442,7 @@ lzma_stream_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, next->coder->tell_unsupported_check = (flags & LZMA_TELL_UNSUPPORTED_CHECK) != 0; next->coder->tell_any_check = (flags & LZMA_TELL_ANY_CHECK) != 0; + next->coder->ignore_check = (flags & LZMA_IGNORE_CHECK) != 0; next->coder->concatenated = (flags & LZMA_CONCATENATED) != 0; next->coder->first_stream = true; diff --git a/src/liblzma/common/stream_decoder.h b/src/liblzma/common/stream_decoder.h index e54ac28f44a5..c13c6ba12706 100644 --- a/src/liblzma/common/stream_decoder.h +++ b/src/liblzma/common/stream_decoder.h @@ -15,7 +15,8 @@ #include "common.h" -extern lzma_ret lzma_stream_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, uint64_t memlimit, uint32_t flags); +extern lzma_ret lzma_stream_decoder_init( + lzma_next_coder *next, const lzma_allocator *allocator, + uint64_t memlimit, uint32_t flags); #endif diff --git a/src/liblzma/common/stream_encoder.c b/src/liblzma/common/stream_encoder.c index 97a7a23a8135..a7663bc48db3 100644 --- a/src/liblzma/common/stream_encoder.c +++ b/src/liblzma/common/stream_encoder.c @@ -10,7 +10,6 @@ // /////////////////////////////////////////////////////////////////////////////// -#include "stream_encoder.h" #include "block_encoder.h" #include "index_encoder.h" @@ -26,7 +25,7 @@ struct lzma_coder_s { } sequence; /// True if Block encoder has been initialized by - /// lzma_stream_encoder_init() or stream_encoder_update() + /// stream_encoder_init() or stream_encoder_update() /// and thus doesn't need to be initialized in stream_encode(). bool block_encoder_is_initialized; @@ -60,7 +59,7 @@ struct lzma_coder_s { static lzma_ret -block_encoder_init(lzma_coder *coder, lzma_allocator *allocator) +block_encoder_init(lzma_coder *coder, const lzma_allocator *allocator) { // Prepare the Block options. Even though Block encoder doesn't need // compressed_size, uncompressed_size, and header_size to be @@ -79,7 +78,7 @@ block_encoder_init(lzma_coder *coder, lzma_allocator *allocator) static lzma_ret -stream_encode(lzma_coder *coder, lzma_allocator *allocator, +stream_encode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -126,7 +125,7 @@ stream_encode(lzma_coder *coder, lzma_allocator *allocator, } // Initialize the Block encoder unless it was already - // initialized by lzma_stream_encoder_init() or + // initialized by stream_encoder_init() or // stream_encoder_update(). if (!coder->block_encoder_is_initialized) return_if_error(block_encoder_init(coder, allocator)); @@ -147,11 +146,12 @@ stream_encode(lzma_coder *coder, lzma_allocator *allocator, } case SEQ_BLOCK_ENCODE: { - static const lzma_action convert[4] = { + static const lzma_action convert[LZMA_ACTION_MAX + 1] = { LZMA_RUN, LZMA_SYNC_FLUSH, LZMA_FINISH, LZMA_FINISH, + LZMA_FINISH, }; const lzma_ret ret = coder->block_encoder.code( @@ -209,7 +209,7 @@ stream_encode(lzma_coder *coder, lzma_allocator *allocator, static void -stream_encoder_end(lzma_coder *coder, lzma_allocator *allocator) +stream_encoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->block_encoder, allocator); lzma_next_end(&coder->index_encoder, allocator); @@ -224,7 +224,7 @@ stream_encoder_end(lzma_coder *coder, lzma_allocator *allocator) static lzma_ret -stream_encoder_update(lzma_coder *coder, lzma_allocator *allocator, +stream_encoder_update(lzma_coder *coder, const lzma_allocator *allocator, const lzma_filter *filters, const lzma_filter *reversed_filters) { @@ -262,11 +262,11 @@ stream_encoder_update(lzma_coder *coder, lzma_allocator *allocator, } -extern lzma_ret -lzma_stream_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +static lzma_ret +stream_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter *filters, lzma_check check) { - lzma_next_coder_init(&lzma_stream_encoder_init, next, allocator); + lzma_next_coder_init(&stream_encoder_init, next, allocator); if (filters == NULL) return LZMA_PROG_ERROR; @@ -320,11 +320,12 @@ extern LZMA_API(lzma_ret) lzma_stream_encoder(lzma_stream *strm, const lzma_filter *filters, lzma_check check) { - lzma_next_strm_init(lzma_stream_encoder_init, strm, filters, check); + lzma_next_strm_init(stream_encoder_init, strm, filters, check); strm->internal->supported_actions[LZMA_RUN] = true; strm->internal->supported_actions[LZMA_SYNC_FLUSH] = true; strm->internal->supported_actions[LZMA_FULL_FLUSH] = true; + strm->internal->supported_actions[LZMA_FULL_BARRIER] = true; strm->internal->supported_actions[LZMA_FINISH] = true; return LZMA_OK; diff --git a/src/liblzma/common/stream_encoder_mt.c b/src/liblzma/common/stream_encoder_mt.c new file mode 100644 index 000000000000..9780ed04cac4 --- /dev/null +++ b/src/liblzma/common/stream_encoder_mt.c @@ -0,0 +1,1131 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file stream_encoder_mt.c +/// \brief Multithreaded .xz Stream encoder +// +// Author: Lasse Collin +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + +#include "filter_encoder.h" +#include "easy_preset.h" +#include "block_encoder.h" +#include "block_buffer_encoder.h" +#include "index_encoder.h" +#include "outqueue.h" + + +/// Maximum supported block size. This makes it simpler to prevent integer +/// overflows if we are given unusually large block size. +#define BLOCK_SIZE_MAX (UINT64_MAX / LZMA_THREADS_MAX) + + +typedef enum { + /// Waiting for work. + THR_IDLE, + + /// Encoding is in progress. + THR_RUN, + + /// Encoding is in progress but no more input data will + /// be read. + THR_FINISH, + + /// The main thread wants the thread to stop whatever it was doing + /// but not exit. + THR_STOP, + + /// The main thread wants the thread to exit. We could use + /// cancellation but since there's stopped anyway, this is lazier. + THR_EXIT, + +} worker_state; + + +typedef struct worker_thread_s worker_thread; +struct worker_thread_s { + worker_state state; + + /// Input buffer of coder->block_size bytes. The main thread will + /// put new input into this and update in_size accordingly. Once + /// no more input is coming, state will be set to THR_FINISH. + uint8_t *in; + + /// Amount of data available in the input buffer. This is modified + /// only by the main thread. + size_t in_size; + + /// Output buffer for this thread. This is set by the main + /// thread every time a new Block is started with this thread + /// structure. + lzma_outbuf *outbuf; + + /// Pointer to the main structure is needed when putting this + /// thread back to the stack of free threads. + lzma_coder *coder; + + /// The allocator is set by the main thread. Since a copy of the + /// pointer is kept here, the application must not change the + /// allocator before calling lzma_end(). + const lzma_allocator *allocator; + + /// Amount of uncompressed data that has already been compressed. + uint64_t progress_in; + + /// Amount of compressed data that is ready. + uint64_t progress_out; + + /// Block encoder + lzma_next_coder block_encoder; + + /// Compression options for this Block + lzma_block block_options; + + /// Next structure in the stack of free worker threads. + worker_thread *next; + + mythread_mutex mutex; + mythread_cond cond; + + /// The ID of this thread is used to join the thread + /// when it's not needed anymore. + mythread thread_id; +}; + + +struct lzma_coder_s { + enum { + SEQ_STREAM_HEADER, + SEQ_BLOCK, + SEQ_INDEX, + SEQ_STREAM_FOOTER, + } sequence; + + /// Start a new Block every block_size bytes of input unless + /// LZMA_FULL_FLUSH or LZMA_FULL_BARRIER is used earlier. + size_t block_size; + + /// The filter chain currently in use + lzma_filter filters[LZMA_FILTERS_MAX + 1]; + + + /// Index to hold sizes of the Blocks + lzma_index *index; + + /// Index encoder + lzma_next_coder index_encoder; + + + /// Stream Flags for encoding the Stream Header and Stream Footer. + lzma_stream_flags stream_flags; + + /// Buffer to hold Stream Header and Stream Footer. + uint8_t header[LZMA_STREAM_HEADER_SIZE]; + + /// Read position in header[] + size_t header_pos; + + + /// Output buffer queue for compressed data + lzma_outq outq; + + + /// Maximum wait time if cannot use all the input and cannot + /// fill the output buffer. This is in milliseconds. + uint32_t timeout; + + + /// Error code from a worker thread + lzma_ret thread_error; + + /// Array of allocated thread-specific structures + worker_thread *threads; + + /// Number of structures in "threads" above. This is also the + /// number of threads that will be created at maximum. + uint32_t threads_max; + + /// Number of thread structures that have been initialized, and + /// thus the number of worker threads actually created so far. + uint32_t threads_initialized; + + /// Stack of free threads. When a thread finishes, it puts itself + /// back into this stack. This starts as empty because threads + /// are created only when actually needed. + worker_thread *threads_free; + + /// The most recent worker thread to which the main thread writes + /// the new input from the application. + worker_thread *thr; + + + /// Amount of uncompressed data in Blocks that have already + /// been finished. + uint64_t progress_in; + + /// Amount of compressed data in Stream Header + Blocks that + /// have already been finished. + uint64_t progress_out; + + + mythread_mutex mutex; + mythread_cond cond; +}; + + +/// Tell the main thread that something has gone wrong. +static void +worker_error(worker_thread *thr, lzma_ret ret) +{ + assert(ret != LZMA_OK); + assert(ret != LZMA_STREAM_END); + + mythread_sync(thr->coder->mutex) { + if (thr->coder->thread_error == LZMA_OK) + thr->coder->thread_error = ret; + + mythread_cond_signal(&thr->coder->cond); + } + + return; +} + + +static worker_state +worker_encode(worker_thread *thr, worker_state state) +{ + assert(thr->progress_in == 0); + assert(thr->progress_out == 0); + + // Set the Block options. + thr->block_options = (lzma_block){ + .version = 0, + .check = thr->coder->stream_flags.check, + .compressed_size = thr->coder->outq.buf_size_max, + .uncompressed_size = thr->coder->block_size, + + // TODO: To allow changing the filter chain, the filters + // array must be copied to each worker_thread. + .filters = thr->coder->filters, + }; + + // Calculate maximum size of the Block Header. This amount is + // reserved in the beginning of the buffer so that Block Header + // along with Compressed Size and Uncompressed Size can be + // written there. + lzma_ret ret = lzma_block_header_size(&thr->block_options); + if (ret != LZMA_OK) { + worker_error(thr, ret); + return THR_STOP; + } + + // Initialize the Block encoder. + ret = lzma_block_encoder_init(&thr->block_encoder, + thr->allocator, &thr->block_options); + if (ret != LZMA_OK) { + worker_error(thr, ret); + return THR_STOP; + } + + size_t in_pos = 0; + size_t in_size = 0; + + thr->outbuf->size = thr->block_options.header_size; + const size_t out_size = thr->coder->outq.buf_size_max; + + do { + mythread_sync(thr->mutex) { + // Store in_pos and out_pos into *thr so that + // an application may read them via + // lzma_get_progress() to get progress information. + // + // NOTE: These aren't updated when the encoding + // finishes. Instead, the final values are taken + // later from thr->outbuf. + thr->progress_in = in_pos; + thr->progress_out = thr->outbuf->size; + + while (in_size == thr->in_size + && thr->state == THR_RUN) + mythread_cond_wait(&thr->cond, &thr->mutex); + + state = thr->state; + in_size = thr->in_size; + } + + // Return if we were asked to stop or exit. + if (state >= THR_STOP) + return state; + + lzma_action action = state == THR_FINISH + ? LZMA_FINISH : LZMA_RUN; + + // Limit the amount of input given to the Block encoder + // at once. This way this thread can react fairly quickly + // if the main thread wants us to stop or exit. + static const size_t in_chunk_max = 16384; + size_t in_limit = in_size; + if (in_size - in_pos > in_chunk_max) { + in_limit = in_pos + in_chunk_max; + action = LZMA_RUN; + } + + ret = thr->block_encoder.code( + thr->block_encoder.coder, thr->allocator, + thr->in, &in_pos, in_limit, thr->outbuf->buf, + &thr->outbuf->size, out_size, action); + } while (ret == LZMA_OK && thr->outbuf->size < out_size); + + switch (ret) { + case LZMA_STREAM_END: + assert(state == THR_FINISH); + + // Encode the Block Header. By doing it after + // the compression, we can store the Compressed Size + // and Uncompressed Size fields. + ret = lzma_block_header_encode(&thr->block_options, + thr->outbuf->buf); + if (ret != LZMA_OK) { + worker_error(thr, ret); + return THR_STOP; + } + + break; + + case LZMA_OK: + // The data was incompressible. Encode it using uncompressed + // LZMA2 chunks. + // + // First wait that we have gotten all the input. + mythread_sync(thr->mutex) { + while (thr->state == THR_RUN) + mythread_cond_wait(&thr->cond, &thr->mutex); + + state = thr->state; + in_size = thr->in_size; + } + + if (state >= THR_STOP) + return state; + + // Do the encoding. This takes care of the Block Header too. + thr->outbuf->size = 0; + ret = lzma_block_uncomp_encode(&thr->block_options, + thr->in, in_size, thr->outbuf->buf, + &thr->outbuf->size, out_size); + + // It shouldn't fail. + if (ret != LZMA_OK) { + worker_error(thr, LZMA_PROG_ERROR); + return THR_STOP; + } + + break; + + default: + worker_error(thr, ret); + return THR_STOP; + } + + // Set the size information that will be read by the main thread + // to write the Index field. + thr->outbuf->unpadded_size + = lzma_block_unpadded_size(&thr->block_options); + assert(thr->outbuf->unpadded_size != 0); + thr->outbuf->uncompressed_size = thr->block_options.uncompressed_size; + + return THR_FINISH; +} + + +static MYTHREAD_RET_TYPE +worker_start(void *thr_ptr) +{ + worker_thread *thr = thr_ptr; + worker_state state = THR_IDLE; // Init to silence a warning + + while (true) { + // Wait for work. + mythread_sync(thr->mutex) { + while (true) { + // The thread is already idle so if we are + // requested to stop, just set the state. + if (thr->state == THR_STOP) { + thr->state = THR_IDLE; + mythread_cond_signal(&thr->cond); + } + + state = thr->state; + if (state != THR_IDLE) + break; + + mythread_cond_wait(&thr->cond, &thr->mutex); + } + } + + assert(state != THR_IDLE); + assert(state != THR_STOP); + + if (state <= THR_FINISH) + state = worker_encode(thr, state); + + if (state == THR_EXIT) + break; + + // Mark the thread as idle unless the main thread has + // told us to exit. Signal is needed for the case + // where the main thread is waiting for the threads to stop. + mythread_sync(thr->mutex) { + if (thr->state != THR_EXIT) { + thr->state = THR_IDLE; + mythread_cond_signal(&thr->cond); + } + } + + mythread_sync(thr->coder->mutex) { + // Mark the output buffer as finished if + // no errors occurred. + thr->outbuf->finished = state == THR_FINISH; + + // Update the main progress info. + thr->coder->progress_in + += thr->outbuf->uncompressed_size; + thr->coder->progress_out += thr->outbuf->size; + thr->progress_in = 0; + thr->progress_out = 0; + + // Return this thread to the stack of free threads. + thr->next = thr->coder->threads_free; + thr->coder->threads_free = thr; + + mythread_cond_signal(&thr->coder->cond); + } + } + + // Exiting, free the resources. + mythread_mutex_destroy(&thr->mutex); + mythread_cond_destroy(&thr->cond); + + lzma_next_end(&thr->block_encoder, thr->allocator); + lzma_free(thr->in, thr->allocator); + return MYTHREAD_RET_VALUE; +} + + +/// Make the threads stop but not exit. Optionally wait for them to stop. +static void +threads_stop(lzma_coder *coder, bool wait_for_threads) +{ + // Tell the threads to stop. + for (uint32_t i = 0; i < coder->threads_initialized; ++i) { + mythread_sync(coder->threads[i].mutex) { + coder->threads[i].state = THR_STOP; + mythread_cond_signal(&coder->threads[i].cond); + } + } + + if (!wait_for_threads) + return; + + // Wait for the threads to settle in the idle state. + for (uint32_t i = 0; i < coder->threads_initialized; ++i) { + mythread_sync(coder->threads[i].mutex) { + while (coder->threads[i].state != THR_IDLE) + mythread_cond_wait(&coder->threads[i].cond, + &coder->threads[i].mutex); + } + } + + return; +} + + +/// Stop the threads and free the resources associated with them. +/// Wait until the threads have exited. +static void +threads_end(lzma_coder *coder, const lzma_allocator *allocator) +{ + for (uint32_t i = 0; i < coder->threads_initialized; ++i) { + mythread_sync(coder->threads[i].mutex) { + coder->threads[i].state = THR_EXIT; + mythread_cond_signal(&coder->threads[i].cond); + } + } + + for (uint32_t i = 0; i < coder->threads_initialized; ++i) { + int ret = mythread_join(coder->threads[i].thread_id); + assert(ret == 0); + (void)ret; + } + + lzma_free(coder->threads, allocator); + return; +} + + +/// Initialize a new worker_thread structure and create a new thread. +static lzma_ret +initialize_new_thread(lzma_coder *coder, const lzma_allocator *allocator) +{ + worker_thread *thr = &coder->threads[coder->threads_initialized]; + + thr->in = lzma_alloc(coder->block_size, allocator); + if (thr->in == NULL) + return LZMA_MEM_ERROR; + + if (mythread_mutex_init(&thr->mutex)) + goto error_mutex; + + if (mythread_cond_init(&thr->cond)) + goto error_cond; + + thr->state = THR_IDLE; + thr->allocator = allocator; + thr->coder = coder; + thr->progress_in = 0; + thr->progress_out = 0; + thr->block_encoder = LZMA_NEXT_CODER_INIT; + + if (mythread_create(&thr->thread_id, &worker_start, thr)) + goto error_thread; + + ++coder->threads_initialized; + coder->thr = thr; + + return LZMA_OK; + +error_thread: + mythread_cond_destroy(&thr->cond); + +error_cond: + mythread_mutex_destroy(&thr->mutex); + +error_mutex: + lzma_free(thr->in, allocator); + return LZMA_MEM_ERROR; +} + + +static lzma_ret +get_thread(lzma_coder *coder, const lzma_allocator *allocator) +{ + // If there are no free output subqueues, there is no + // point to try getting a thread. + if (!lzma_outq_has_buf(&coder->outq)) + return LZMA_OK; + + // If there is a free structure on the stack, use it. + mythread_sync(coder->mutex) { + if (coder->threads_free != NULL) { + coder->thr = coder->threads_free; + coder->threads_free = coder->threads_free->next; + } + } + + if (coder->thr == NULL) { + // If there are no uninitialized structures left, return. + if (coder->threads_initialized == coder->threads_max) + return LZMA_OK; + + // Initialize a new thread. + return_if_error(initialize_new_thread(coder, allocator)); + } + + // Reset the parts of the thread state that have to be done + // in the main thread. + mythread_sync(coder->thr->mutex) { + coder->thr->state = THR_RUN; + coder->thr->in_size = 0; + coder->thr->outbuf = lzma_outq_get_buf(&coder->outq); + mythread_cond_signal(&coder->thr->cond); + } + + return LZMA_OK; +} + + +static lzma_ret +stream_encode_in(lzma_coder *coder, const lzma_allocator *allocator, + const uint8_t *restrict in, size_t *restrict in_pos, + size_t in_size, lzma_action action) +{ + while (*in_pos < in_size + || (coder->thr != NULL && action != LZMA_RUN)) { + if (coder->thr == NULL) { + // Get a new thread. + const lzma_ret ret = get_thread(coder, allocator); + if (coder->thr == NULL) + return ret; + } + + // Copy the input data to thread's buffer. + size_t thr_in_size = coder->thr->in_size; + lzma_bufcpy(in, in_pos, in_size, coder->thr->in, + &thr_in_size, coder->block_size); + + // Tell the Block encoder to finish if + // - it has got block_size bytes of input; or + // - all input was used and LZMA_FINISH, LZMA_FULL_FLUSH, + // or LZMA_FULL_BARRIER was used. + // + // TODO: LZMA_SYNC_FLUSH and LZMA_SYNC_BARRIER. + const bool finish = thr_in_size == coder->block_size + || (*in_pos == in_size && action != LZMA_RUN); + + bool block_error = false; + + mythread_sync(coder->thr->mutex) { + if (coder->thr->state == THR_IDLE) { + // Something has gone wrong with the Block + // encoder. It has set coder->thread_error + // which we will read a few lines later. + block_error = true; + } else { + // Tell the Block encoder its new amount + // of input and update the state if needed. + coder->thr->in_size = thr_in_size; + + if (finish) + coder->thr->state = THR_FINISH; + + mythread_cond_signal(&coder->thr->cond); + } + } + + if (block_error) { + lzma_ret ret; + + mythread_sync(coder->mutex) { + ret = coder->thread_error; + } + + return ret; + } + + if (finish) + coder->thr = NULL; + } + + return LZMA_OK; +} + + +/// Wait until more input can be consumed, more output can be read, or +/// an optional timeout is reached. +static bool +wait_for_work(lzma_coder *coder, mythread_condtime *wait_abs, + bool *has_blocked, bool has_input) +{ + if (coder->timeout != 0 && !*has_blocked) { + // Every time when stream_encode_mt() is called via + // lzma_code(), *has_blocked starts as false. We set it + // to true here and calculate the absolute time when + // we must return if there's nothing to do. + // + // The idea of *has_blocked is to avoid unneeded calls + // to mythread_condtime_set(), which may do a syscall + // depending on the operating system. + *has_blocked = true; + mythread_condtime_set(wait_abs, &coder->cond, coder->timeout); + } + + bool timed_out = false; + + mythread_sync(coder->mutex) { + // There are four things that we wait. If one of them + // becomes possible, we return. + // - If there is input left, we need to get a free + // worker thread and an output buffer for it. + // - Data ready to be read from the output queue. + // - A worker thread indicates an error. + // - Time out occurs. + while ((!has_input || coder->threads_free == NULL + || !lzma_outq_has_buf(&coder->outq)) + && !lzma_outq_is_readable(&coder->outq) + && coder->thread_error == LZMA_OK + && !timed_out) { + if (coder->timeout != 0) + timed_out = mythread_cond_timedwait( + &coder->cond, &coder->mutex, + wait_abs) != 0; + else + mythread_cond_wait(&coder->cond, + &coder->mutex); + } + } + + return timed_out; +} + + +static lzma_ret +stream_encode_mt(lzma_coder *coder, const lzma_allocator *allocator, + const uint8_t *restrict in, size_t *restrict in_pos, + size_t in_size, uint8_t *restrict out, + size_t *restrict out_pos, size_t out_size, lzma_action action) +{ + switch (coder->sequence) { + case SEQ_STREAM_HEADER: + lzma_bufcpy(coder->header, &coder->header_pos, + sizeof(coder->header), + out, out_pos, out_size); + if (coder->header_pos < sizeof(coder->header)) + return LZMA_OK; + + coder->header_pos = 0; + coder->sequence = SEQ_BLOCK; + + // Fall through + + case SEQ_BLOCK: { + // Initialized to silence warnings. + lzma_vli unpadded_size = 0; + lzma_vli uncompressed_size = 0; + lzma_ret ret = LZMA_OK; + + // These are for wait_for_work(). + bool has_blocked = false; + mythread_condtime wait_abs; + + while (true) { + mythread_sync(coder->mutex) { + // Check for Block encoder errors. + ret = coder->thread_error; + if (ret != LZMA_OK) { + assert(ret != LZMA_STREAM_END); + break; + } + + // Try to read compressed data to out[]. + ret = lzma_outq_read(&coder->outq, + out, out_pos, out_size, + &unpadded_size, + &uncompressed_size); + } + + if (ret == LZMA_STREAM_END) { + // End of Block. Add it to the Index. + ret = lzma_index_append(coder->index, + allocator, unpadded_size, + uncompressed_size); + + // If we didn't fill the output buffer yet, + // try to read more data. Maybe the next + // outbuf has been finished already too. + if (*out_pos < out_size) + continue; + } + + if (ret != LZMA_OK) { + // coder->thread_error was set or + // lzma_index_append() failed. + threads_stop(coder, false); + return ret; + } + + // Try to give uncompressed data to a worker thread. + ret = stream_encode_in(coder, allocator, + in, in_pos, in_size, action); + if (ret != LZMA_OK) { + threads_stop(coder, false); + return ret; + } + + // See if we should wait or return. + // + // TODO: LZMA_SYNC_FLUSH and LZMA_SYNC_BARRIER. + if (*in_pos == in_size) { + // LZMA_RUN: More data is probably coming + // so return to let the caller fill the + // input buffer. + if (action == LZMA_RUN) + return LZMA_OK; + + // LZMA_FULL_BARRIER: The same as with + // LZMA_RUN but tell the caller that the + // barrier was completed. + if (action == LZMA_FULL_BARRIER) + return LZMA_STREAM_END; + + // Finishing or flushing isn't completed until + // all input data has been encoded and copied + // to the output buffer. + if (lzma_outq_is_empty(&coder->outq)) { + // LZMA_FINISH: Continue to encode + // the Index field. + if (action == LZMA_FINISH) + break; + + // LZMA_FULL_FLUSH: Return to tell + // the caller that flushing was + // completed. + if (action == LZMA_FULL_FLUSH) + return LZMA_STREAM_END; + } + } + + // Return if there is no output space left. + // This check must be done after testing the input + // buffer, because we might want to use a different + // return code. + if (*out_pos == out_size) + return LZMA_OK; + + // Neither in nor out has been used completely. + // Wait until there's something we can do. + if (wait_for_work(coder, &wait_abs, &has_blocked, + *in_pos < in_size)) + return LZMA_TIMED_OUT; + } + + // All Blocks have been encoded and the threads have stopped. + // Prepare to encode the Index field. + return_if_error(lzma_index_encoder_init( + &coder->index_encoder, allocator, + coder->index)); + coder->sequence = SEQ_INDEX; + + // Update the progress info to take the Index and + // Stream Footer into account. Those are very fast to encode + // so in terms of progress information they can be thought + // to be ready to be copied out. + coder->progress_out += lzma_index_size(coder->index) + + LZMA_STREAM_HEADER_SIZE; + } + + // Fall through + + case SEQ_INDEX: { + // Call the Index encoder. It doesn't take any input, so + // those pointers can be NULL. + const lzma_ret ret = coder->index_encoder.code( + coder->index_encoder.coder, allocator, + NULL, NULL, 0, + out, out_pos, out_size, LZMA_RUN); + if (ret != LZMA_STREAM_END) + return ret; + + // Encode the Stream Footer into coder->buffer. + coder->stream_flags.backward_size + = lzma_index_size(coder->index); + if (lzma_stream_footer_encode(&coder->stream_flags, + coder->header) != LZMA_OK) + return LZMA_PROG_ERROR; + + coder->sequence = SEQ_STREAM_FOOTER; + } + + // Fall through + + case SEQ_STREAM_FOOTER: + lzma_bufcpy(coder->header, &coder->header_pos, + sizeof(coder->header), + out, out_pos, out_size); + return coder->header_pos < sizeof(coder->header) + ? LZMA_OK : LZMA_STREAM_END; + } + + assert(0); + return LZMA_PROG_ERROR; +} + + +static void +stream_encoder_mt_end(lzma_coder *coder, const lzma_allocator *allocator) +{ + // Threads must be killed before the output queue can be freed. + threads_end(coder, allocator); + lzma_outq_end(&coder->outq, allocator); + + for (size_t i = 0; coder->filters[i].id != LZMA_VLI_UNKNOWN; ++i) + lzma_free(coder->filters[i].options, allocator); + + lzma_next_end(&coder->index_encoder, allocator); + lzma_index_end(coder->index, allocator); + + mythread_cond_destroy(&coder->cond); + mythread_mutex_destroy(&coder->mutex); + + lzma_free(coder, allocator); + return; +} + + +/// Options handling for lzma_stream_encoder_mt_init() and +/// lzma_stream_encoder_mt_memusage() +static lzma_ret +get_options(const lzma_mt *options, lzma_options_easy *opt_easy, + const lzma_filter **filters, uint64_t *block_size, + uint64_t *outbuf_size_max) +{ + // Validate some of the options. + if (options == NULL) + return LZMA_PROG_ERROR; + + if (options->flags != 0 || options->threads == 0 + || options->threads > LZMA_THREADS_MAX) + return LZMA_OPTIONS_ERROR; + + if (options->filters != NULL) { + // Filter chain was given, use it as is. + *filters = options->filters; + } else { + // Use a preset. + if (lzma_easy_preset(opt_easy, options->preset)) + return LZMA_OPTIONS_ERROR; + + *filters = opt_easy->filters; + } + + // Block size + if (options->block_size > 0) { + if (options->block_size > BLOCK_SIZE_MAX) + return LZMA_OPTIONS_ERROR; + + *block_size = options->block_size; + } else { + // Determine the Block size from the filter chain. + *block_size = lzma_mt_block_size(*filters); + if (*block_size == 0) + return LZMA_OPTIONS_ERROR; + + assert(*block_size <= BLOCK_SIZE_MAX); + } + + // Calculate the maximum amount output that a single output buffer + // may need to hold. This is the same as the maximum total size of + // a Block. + *outbuf_size_max = lzma_block_buffer_bound64(*block_size); + if (*outbuf_size_max == 0) + return LZMA_MEM_ERROR; + + return LZMA_OK; +} + + +static void +get_progress(lzma_coder *coder, uint64_t *progress_in, uint64_t *progress_out) +{ + // Lock coder->mutex to prevent finishing threads from moving their + // progress info from the worker_thread structure to lzma_coder. + mythread_sync(coder->mutex) { + *progress_in = coder->progress_in; + *progress_out = coder->progress_out; + + for (size_t i = 0; i < coder->threads_initialized; ++i) { + mythread_sync(coder->threads[i].mutex) { + *progress_in += coder->threads[i].progress_in; + *progress_out += coder->threads[i] + .progress_out; + } + } + } + + return; +} + + +static lzma_ret +stream_encoder_mt_init(lzma_next_coder *next, const lzma_allocator *allocator, + const lzma_mt *options) +{ + lzma_next_coder_init(&stream_encoder_mt_init, next, allocator); + + // Get the filter chain. + lzma_options_easy easy; + const lzma_filter *filters; + uint64_t block_size; + uint64_t outbuf_size_max; + return_if_error(get_options(options, &easy, &filters, + &block_size, &outbuf_size_max)); + +#if SIZE_MAX < UINT64_MAX + if (block_size > SIZE_MAX) + return LZMA_MEM_ERROR; +#endif + + // Validate the filter chain so that we can give an error in this + // function instead of delaying it to the first call to lzma_code(). + // The memory usage calculation verifies the filter chain as + // a side effect so we take advatange of that. + if (lzma_raw_encoder_memusage(filters) == UINT64_MAX) + return LZMA_OPTIONS_ERROR; + + // Validate the Check ID. + if ((unsigned int)(options->check) > LZMA_CHECK_ID_MAX) + return LZMA_PROG_ERROR; + + if (!lzma_check_is_supported(options->check)) + return LZMA_UNSUPPORTED_CHECK; + + // Allocate and initialize the base structure if needed. + if (next->coder == NULL) { + next->coder = lzma_alloc(sizeof(lzma_coder), allocator); + if (next->coder == NULL) + return LZMA_MEM_ERROR; + + // For the mutex and condition variable initializations + // the error handling has to be done here because + // stream_encoder_mt_end() doesn't know if they have + // already been initialized or not. + if (mythread_mutex_init(&next->coder->mutex)) { + lzma_free(next->coder, allocator); + next->coder = NULL; + return LZMA_MEM_ERROR; + } + + if (mythread_cond_init(&next->coder->cond)) { + mythread_mutex_destroy(&next->coder->mutex); + lzma_free(next->coder, allocator); + next->coder = NULL; + return LZMA_MEM_ERROR; + } + + next->code = &stream_encode_mt; + next->end = &stream_encoder_mt_end; + next->get_progress = &get_progress; +// next->update = &stream_encoder_mt_update; + + next->coder->filters[0].id = LZMA_VLI_UNKNOWN; + next->coder->index_encoder = LZMA_NEXT_CODER_INIT; + next->coder->index = NULL; + memzero(&next->coder->outq, sizeof(next->coder->outq)); + next->coder->threads = NULL; + next->coder->threads_max = 0; + next->coder->threads_initialized = 0; + } + + // Basic initializations + next->coder->sequence = SEQ_STREAM_HEADER; + next->coder->block_size = (size_t)(block_size); + next->coder->thread_error = LZMA_OK; + next->coder->thr = NULL; + + // Allocate the thread-specific base structures. + assert(options->threads > 0); + if (next->coder->threads_max != options->threads) { + threads_end(next->coder, allocator); + + next->coder->threads = NULL; + next->coder->threads_max = 0; + + next->coder->threads_initialized = 0; + next->coder->threads_free = NULL; + + next->coder->threads = lzma_alloc( + options->threads * sizeof(worker_thread), + allocator); + if (next->coder->threads == NULL) + return LZMA_MEM_ERROR; + + next->coder->threads_max = options->threads; + } else { + // Reuse the old structures and threads. Tell the running + // threads to stop and wait until they have stopped. + threads_stop(next->coder, true); + } + + // Output queue + return_if_error(lzma_outq_init(&next->coder->outq, allocator, + outbuf_size_max, options->threads)); + + // Timeout + next->coder->timeout = options->timeout; + + // Free the old filter chain and copy the new one. + for (size_t i = 0; next->coder->filters[i].id != LZMA_VLI_UNKNOWN; ++i) + lzma_free(next->coder->filters[i].options, allocator); + + return_if_error(lzma_filters_copy( + filters, next->coder->filters, allocator)); + + // Index + lzma_index_end(next->coder->index, allocator); + next->coder->index = lzma_index_init(allocator); + if (next->coder->index == NULL) + return LZMA_MEM_ERROR; + + // Stream Header + next->coder->stream_flags.version = 0; + next->coder->stream_flags.check = options->check; + return_if_error(lzma_stream_header_encode( + &next->coder->stream_flags, next->coder->header)); + + next->coder->header_pos = 0; + + // Progress info + next->coder->progress_in = 0; + next->coder->progress_out = LZMA_STREAM_HEADER_SIZE; + + return LZMA_OK; +} + + +extern LZMA_API(lzma_ret) +lzma_stream_encoder_mt(lzma_stream *strm, const lzma_mt *options) +{ + lzma_next_strm_init(stream_encoder_mt_init, strm, options); + + strm->internal->supported_actions[LZMA_RUN] = true; +// strm->internal->supported_actions[LZMA_SYNC_FLUSH] = true; + strm->internal->supported_actions[LZMA_FULL_FLUSH] = true; + strm->internal->supported_actions[LZMA_FULL_BARRIER] = true; + strm->internal->supported_actions[LZMA_FINISH] = true; + + return LZMA_OK; +} + + +// This function name is a monster but it's consistent with the older +// monster names. :-( 31 chars is the max that C99 requires so in that +// sense it's not too long. ;-) +extern LZMA_API(uint64_t) +lzma_stream_encoder_mt_memusage(const lzma_mt *options) +{ + lzma_options_easy easy; + const lzma_filter *filters; + uint64_t block_size; + uint64_t outbuf_size_max; + + if (get_options(options, &easy, &filters, &block_size, + &outbuf_size_max) != LZMA_OK) + return UINT64_MAX; + + // Memory usage of the input buffers + const uint64_t inbuf_memusage = options->threads * block_size; + + // Memory usage of the filter encoders + uint64_t filters_memusage = lzma_raw_encoder_memusage(filters); + if (filters_memusage == UINT64_MAX) + return UINT64_MAX; + + filters_memusage *= options->threads; + + // Memory usage of the output queue + const uint64_t outq_memusage = lzma_outq_memusage( + outbuf_size_max, options->threads); + if (outq_memusage == UINT64_MAX) + return UINT64_MAX; + + // Sum them with overflow checking. + uint64_t total_memusage = LZMA_MEMUSAGE_BASE + sizeof(lzma_coder) + + options->threads * sizeof(worker_thread); + + if (UINT64_MAX - total_memusage < inbuf_memusage) + return UINT64_MAX; + + total_memusage += inbuf_memusage; + + if (UINT64_MAX - total_memusage < filters_memusage) + return UINT64_MAX; + + total_memusage += filters_memusage; + + if (UINT64_MAX - total_memusage < outq_memusage) + return UINT64_MAX; + + return total_memusage + outq_memusage; +} diff --git a/src/liblzma/delta/delta_common.c b/src/liblzma/delta/delta_common.c index 930ad215131c..13dd46828d87 100644 --- a/src/liblzma/delta/delta_common.c +++ b/src/liblzma/delta/delta_common.c @@ -15,7 +15,7 @@ static void -delta_coder_end(lzma_coder *coder, lzma_allocator *allocator) +delta_coder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder, allocator); @@ -24,7 +24,7 @@ delta_coder_end(lzma_coder *coder, lzma_allocator *allocator) extern lzma_ret -lzma_delta_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_delta_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { // Allocate memory for the decoder if needed. diff --git a/src/liblzma/delta/delta_decoder.c b/src/liblzma/delta/delta_decoder.c index 2cf60d5bdc7c..726d023991c0 100644 --- a/src/liblzma/delta/delta_decoder.c +++ b/src/liblzma/delta/delta_decoder.c @@ -27,7 +27,7 @@ decode_buffer(lzma_coder *coder, uint8_t *buffer, size_t size) static lzma_ret -delta_decode(lzma_coder *coder, lzma_allocator *allocator, +delta_decode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -47,7 +47,7 @@ delta_decode(lzma_coder *coder, lzma_allocator *allocator, extern lzma_ret -lzma_delta_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_delta_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { next->code = &delta_decode; @@ -56,7 +56,7 @@ lzma_delta_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret -lzma_delta_props_decode(void **options, lzma_allocator *allocator, +lzma_delta_props_decode(void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size) { if (props_size != 1) diff --git a/src/liblzma/delta/delta_decoder.h b/src/liblzma/delta/delta_decoder.h index ae89acc59f8c..ad89cc659764 100644 --- a/src/liblzma/delta/delta_decoder.h +++ b/src/liblzma/delta/delta_decoder.h @@ -16,10 +16,11 @@ #include "delta_common.h" extern lzma_ret lzma_delta_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_delta_props_decode( - void **options, lzma_allocator *allocator, + void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size); #endif diff --git a/src/liblzma/delta/delta_encoder.c b/src/liblzma/delta/delta_encoder.c index 15c7951e104a..5a842636ba1b 100644 --- a/src/liblzma/delta/delta_encoder.c +++ b/src/liblzma/delta/delta_encoder.c @@ -49,7 +49,7 @@ encode_in_place(lzma_coder *coder, uint8_t *buffer, size_t size) static lzma_ret -delta_encode(lzma_coder *coder, lzma_allocator *allocator, +delta_encode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -84,7 +84,7 @@ delta_encode(lzma_coder *coder, lzma_allocator *allocator, static lzma_ret -delta_encoder_update(lzma_coder *coder, lzma_allocator *allocator, +delta_encoder_update(lzma_coder *coder, const lzma_allocator *allocator, const lzma_filter *filters_null lzma_attribute((__unused__)), const lzma_filter *reversed_filters) { @@ -97,7 +97,7 @@ delta_encoder_update(lzma_coder *coder, lzma_allocator *allocator, extern lzma_ret -lzma_delta_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_delta_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { next->code = &delta_encode; diff --git a/src/liblzma/delta/delta_encoder.h b/src/liblzma/delta/delta_encoder.h index a447862f205b..4ab984785171 100644 --- a/src/liblzma/delta/delta_encoder.h +++ b/src/liblzma/delta/delta_encoder.h @@ -16,7 +16,8 @@ #include "delta_common.h" extern lzma_ret lzma_delta_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_delta_props_encode(const void *options, uint8_t *out); diff --git a/src/liblzma/delta/delta_private.h b/src/liblzma/delta/delta_private.h index 62b7fed86e45..46ce0c645091 100644 --- a/src/liblzma/delta/delta_private.h +++ b/src/liblzma/delta/delta_private.h @@ -31,7 +31,7 @@ struct lzma_coder_s { extern lzma_ret lzma_delta_coder_init( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters); #endif diff --git a/src/liblzma/liblzma.map b/src/liblzma/liblzma.map new file mode 100644 index 000000000000..f53a4ea30a3c --- /dev/null +++ b/src/liblzma/liblzma.map @@ -0,0 +1,108 @@ +XZ_5.0 { +global: + lzma_alone_decoder; + lzma_alone_encoder; + lzma_auto_decoder; + lzma_block_buffer_bound; + lzma_block_buffer_decode; + lzma_block_buffer_encode; + lzma_block_compressed_size; + lzma_block_decoder; + lzma_block_encoder; + lzma_block_header_decode; + lzma_block_header_encode; + lzma_block_header_size; + lzma_block_total_size; + lzma_block_unpadded_size; + lzma_check_is_supported; + lzma_check_size; + lzma_code; + lzma_crc32; + lzma_crc64; + lzma_easy_buffer_encode; + lzma_easy_decoder_memusage; + lzma_easy_encoder; + lzma_easy_encoder_memusage; + lzma_end; + lzma_filter_decoder_is_supported; + lzma_filter_encoder_is_supported; + lzma_filter_flags_decode; + lzma_filter_flags_encode; + lzma_filter_flags_size; + lzma_filters_copy; + lzma_filters_update; + lzma_get_check; + lzma_index_append; + lzma_index_block_count; + lzma_index_buffer_decode; + lzma_index_buffer_encode; + lzma_index_cat; + lzma_index_checks; + lzma_index_decoder; + lzma_index_dup; + lzma_index_encoder; + lzma_index_end; + lzma_index_file_size; + lzma_index_hash_append; + lzma_index_hash_decode; + lzma_index_hash_end; + lzma_index_hash_init; + lzma_index_hash_size; + lzma_index_init; + lzma_index_iter_init; + lzma_index_iter_locate; + lzma_index_iter_next; + lzma_index_iter_rewind; + lzma_index_memusage; + lzma_index_memused; + lzma_index_size; + lzma_index_stream_count; + lzma_index_stream_flags; + lzma_index_stream_padding; + lzma_index_stream_size; + lzma_index_total_size; + lzma_index_uncompressed_size; + lzma_lzma_preset; + lzma_memlimit_get; + lzma_memlimit_set; + lzma_memusage; + lzma_mf_is_supported; + lzma_mode_is_supported; + lzma_physmem; + lzma_properties_decode; + lzma_properties_encode; + lzma_properties_size; + lzma_raw_buffer_decode; + lzma_raw_buffer_encode; + lzma_raw_decoder; + lzma_raw_decoder_memusage; + lzma_raw_encoder; + lzma_raw_encoder_memusage; + lzma_stream_buffer_bound; + lzma_stream_buffer_decode; + lzma_stream_buffer_encode; + lzma_stream_decoder; + lzma_stream_encoder; + lzma_stream_flags_compare; + lzma_stream_footer_decode; + lzma_stream_footer_encode; + lzma_stream_header_decode; + lzma_stream_header_encode; + lzma_version_number; + lzma_version_string; + lzma_vli_decode; + lzma_vli_encode; + lzma_vli_size; +}; + +XZ_5.2 { +global: + lzma_block_uncomp_encode; + lzma_cputhreads; + lzma_get_progress; + lzma_stream_encoder_mt; + lzma_stream_encoder_mt_memusage; + +local: + *; +} XZ_5.0; diff --git a/src/liblzma/liblzma.pc.in b/src/liblzma/liblzma.pc.in index 7f11f1a2009e..9fa489115a0a 100644 --- a/src/liblzma/liblzma.pc.in +++ b/src/liblzma/liblzma.pc.in @@ -16,4 +16,4 @@ URL: @PACKAGE_URL@ Version: @PACKAGE_VERSION@ Cflags: -I${includedir} Libs: -L${libdir} -llzma -Libs.private: @PTHREAD_CFLAGS@ @PTHREAD_LIBS@ +Libs.private: @PTHREAD_CFLAGS@ @LIBS@ diff --git a/src/liblzma/lz/lz_decoder.c b/src/liblzma/lz/lz_decoder.c index d74085cf4471..2328a8e73f07 100644 --- a/src/liblzma/lz/lz_decoder.c +++ b/src/liblzma/lz/lz_decoder.c @@ -126,7 +126,7 @@ decode_buffer(lzma_coder *coder, static lzma_ret lz_decode(lzma_coder *coder, - lzma_allocator *allocator lzma_attribute((__unused__)), + const lzma_allocator *allocator lzma_attribute((__unused__)), const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, @@ -184,7 +184,7 @@ lz_decode(lzma_coder *coder, static void -lz_decoder_end(lzma_coder *coder, lzma_allocator *allocator) +lz_decoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder->dict.buf, allocator); @@ -200,10 +200,10 @@ lz_decoder_end(lzma_coder *coder, lzma_allocator *allocator) extern lzma_ret -lzma_lz_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_lz_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, lzma_ret (*lz_init)(lzma_lz_decoder *lz, - lzma_allocator *allocator, const void *options, + const lzma_allocator *allocator, const void *options, lzma_lz_options *lz_options)) { // Allocate the base structure if it isn't already allocated. diff --git a/src/liblzma/lz/lz_decoder.h b/src/liblzma/lz/lz_decoder.h index 7266e803165f..277900afb714 100644 --- a/src/liblzma/lz/lz_decoder.h +++ b/src/liblzma/lz/lz_decoder.h @@ -67,7 +67,7 @@ typedef struct { lzma_vli uncompressed_size); /// Free allocated resources - void (*end)(lzma_coder *coder, lzma_allocator *allocator); + void (*end)(lzma_coder *coder, const lzma_allocator *allocator); } lzma_lz_decoder; @@ -83,9 +83,10 @@ typedef struct { extern lzma_ret lzma_lz_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters, + const lzma_allocator *allocator, + const lzma_filter_info *filters, lzma_ret (*lz_init)(lzma_lz_decoder *lz, - lzma_allocator *allocator, const void *options, + const lzma_allocator *allocator, const void *options, lzma_lz_options *lz_options)); extern uint64_t lzma_lz_decoder_memusage(size_t dictionary_size); diff --git a/src/liblzma/lz/lz_encoder.c b/src/liblzma/lz/lz_encoder.c index e2406965882c..2033844f5d35 100644 --- a/src/liblzma/lz/lz_encoder.c +++ b/src/liblzma/lz/lz_encoder.c @@ -20,6 +20,8 @@ # include "lz_encoder_hash_table.h" #endif +#include "memcmplen.h" + struct lzma_coder_s { /// LZ-based encoder e.g. LZMA @@ -76,8 +78,9 @@ move_window(lzma_mf *mf) /// This function must not be called once it has returned LZMA_STREAM_END. /// static lzma_ret -fill_window(lzma_coder *coder, lzma_allocator *allocator, const uint8_t *in, - size_t *in_pos, size_t in_size, lzma_action action) +fill_window(lzma_coder *coder, const lzma_allocator *allocator, + const uint8_t *in, size_t *in_pos, size_t in_size, + lzma_action action) { assert(coder->mf.read_pos <= coder->mf.write_pos); @@ -148,7 +151,7 @@ fill_window(lzma_coder *coder, lzma_allocator *allocator, const uint8_t *in, static lzma_ret -lz_encode(lzma_coder *coder, lzma_allocator *allocator, +lz_encode(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, @@ -179,7 +182,7 @@ lz_encode(lzma_coder *coder, lzma_allocator *allocator, static bool -lz_encoder_prepare(lzma_mf *mf, lzma_allocator *allocator, +lz_encoder_prepare(lzma_mf *mf, const lzma_allocator *allocator, const lzma_lz_options *lz_options) { // For now, the dictionary size is limited to 1.5 GiB. This may grow @@ -325,25 +328,22 @@ lz_encoder_prepare(lzma_mf *mf, lzma_allocator *allocator, hs += HASH_4_SIZE; */ - // If the above code calculating hs is modified, make sure that - // this assertion stays valid (UINT32_MAX / 5 is not strictly the - // exact limit). If it doesn't, you need to calculate that - // hash_size_sum + sons_count cannot overflow. - assert(hs < UINT32_MAX / 5); - - const uint32_t old_count = mf->hash_size_sum + mf->sons_count; - mf->hash_size_sum = hs; + const uint32_t old_hash_count = mf->hash_count; + const uint32_t old_sons_count = mf->sons_count; + mf->hash_count = hs; mf->sons_count = mf->cyclic_size; if (is_bt) mf->sons_count *= 2; - const uint32_t new_count = mf->hash_size_sum + mf->sons_count; - // Deallocate the old hash array if it exists and has different size // than what is needed now. - if (old_count != new_count) { + if (old_hash_count != mf->hash_count + || old_sons_count != mf->sons_count) { lzma_free(mf->hash, allocator); mf->hash = NULL; + + lzma_free(mf->son, allocator); + mf->son = NULL; } // Maximum number of match finder cycles @@ -360,14 +360,23 @@ lz_encoder_prepare(lzma_mf *mf, lzma_allocator *allocator, static bool -lz_encoder_init(lzma_mf *mf, lzma_allocator *allocator, +lz_encoder_init(lzma_mf *mf, const lzma_allocator *allocator, const lzma_lz_options *lz_options) { // Allocate the history buffer. if (mf->buffer == NULL) { - mf->buffer = lzma_alloc(mf->size, allocator); + // lzma_memcmplen() is used for the dictionary buffer + // so we need to allocate a few extra bytes to prevent + // it from reading past the end of the buffer. + mf->buffer = lzma_alloc(mf->size + LZMA_MEMCMPLEN_EXTRA, + allocator); if (mf->buffer == NULL) return true; + + // Keep Valgrind happy with lzma_memcmplen() and initialize + // the extra bytes whose value may get read but which will + // effectively get ignored. + memzero(mf->buffer + mf->size, LZMA_MEMCMPLEN_EXTRA); } // Use cyclic_size as initial mf->offset. This allows @@ -381,43 +390,48 @@ lz_encoder_init(lzma_mf *mf, lzma_allocator *allocator, mf->write_pos = 0; mf->pending = 0; - // Allocate match finder's hash array. - const size_t alloc_count = mf->hash_size_sum + mf->sons_count; - #if UINT32_MAX >= SIZE_MAX / 4 // Check for integer overflow. (Huge dictionaries are not // possible on 32-bit CPU.) - if (alloc_count > SIZE_MAX / sizeof(uint32_t)) + if (mf->hash_count > SIZE_MAX / sizeof(uint32_t) + || mf->sons_count > SIZE_MAX / sizeof(uint32_t)) return true; #endif + // Allocate and initialize the hash table. Since EMPTY_HASH_VALUE + // is zero, we can use lzma_alloc_zero() or memzero() for mf->hash. + // + // We don't need to initialize mf->son, but not doing that may + // make Valgrind complain in normalization (see normalize() in + // lz_encoder_mf.c). Skipping the initialization is *very* good + // when big dictionary is used but only small amount of data gets + // actually compressed: most of the mf->son won't get actually + // allocated by the kernel, so we avoid wasting RAM and improve + // initialization speed a lot. if (mf->hash == NULL) { - mf->hash = lzma_alloc(alloc_count * sizeof(uint32_t), + mf->hash = lzma_alloc_zero(mf->hash_count * sizeof(uint32_t), + allocator); + mf->son = lzma_alloc(mf->sons_count * sizeof(uint32_t), allocator); - if (mf->hash == NULL) - return true; - } - mf->son = mf->hash + mf->hash_size_sum; - mf->cyclic_pos = 0; + if (mf->hash == NULL || mf->son == NULL) { + lzma_free(mf->hash, allocator); + mf->hash = NULL; + + lzma_free(mf->son, allocator); + mf->son = NULL; - // Initialize the hash table. Since EMPTY_HASH_VALUE is zero, we - // can use memset(). + return true; + } + } else { /* - for (uint32_t i = 0; i < hash_size_sum; ++i) - mf->hash[i] = EMPTY_HASH_VALUE; + for (uint32_t i = 0; i < mf->hash_count; ++i) + mf->hash[i] = EMPTY_HASH_VALUE; */ - memzero(mf->hash, (size_t)(mf->hash_size_sum) * sizeof(uint32_t)); + memzero(mf->hash, mf->hash_count * sizeof(uint32_t)); + } - // We don't need to initialize mf->son, but not doing that will - // make Valgrind complain in normalization (see normalize() in - // lz_encoder_mf.c). - // - // Skipping this initialization is *very* good when big dictionary is - // used but only small amount of data gets actually compressed: most - // of the mf->hash won't get actually allocated by the kernel, so - // we avoid wasting RAM and improve initialization speed a lot. - //memzero(mf->son, (size_t)(mf->sons_count) * sizeof(uint32_t)); + mf->cyclic_pos = 0; // Handle preset dictionary. if (lz_options->preset_dict != NULL @@ -445,7 +459,8 @@ lzma_lz_encoder_memusage(const lzma_lz_options *lz_options) lzma_mf mf = { .buffer = NULL, .hash = NULL, - .hash_size_sum = 0, + .son = NULL, + .hash_count = 0, .sons_count = 0, }; @@ -454,17 +469,17 @@ lzma_lz_encoder_memusage(const lzma_lz_options *lz_options) return UINT64_MAX; // Calculate the memory usage. - return (uint64_t)(mf.hash_size_sum + mf.sons_count) - * sizeof(uint32_t) - + (uint64_t)(mf.size) + sizeof(lzma_coder); + return ((uint64_t)(mf.hash_count) + mf.sons_count) * sizeof(uint32_t) + + mf.size + sizeof(lzma_coder); } static void -lz_encoder_end(lzma_coder *coder, lzma_allocator *allocator) +lz_encoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); + lzma_free(coder->mf.son, allocator); lzma_free(coder->mf.hash, allocator); lzma_free(coder->mf.buffer, allocator); @@ -479,7 +494,7 @@ lz_encoder_end(lzma_coder *coder, lzma_allocator *allocator) static lzma_ret -lz_encoder_update(lzma_coder *coder, lzma_allocator *allocator, +lz_encoder_update(lzma_coder *coder, const lzma_allocator *allocator, const lzma_filter *filters_null lzma_attribute((__unused__)), const lzma_filter *reversed_filters) { @@ -495,10 +510,10 @@ lz_encoder_update(lzma_coder *coder, lzma_allocator *allocator, extern lzma_ret -lzma_lz_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_lz_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, lzma_ret (*lz_init)(lzma_lz_encoder *lz, - lzma_allocator *allocator, const void *options, + const lzma_allocator *allocator, const void *options, lzma_lz_options *lz_options)) { #ifdef HAVE_SMALL @@ -522,7 +537,8 @@ lzma_lz_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, next->coder->mf.buffer = NULL; next->coder->mf.hash = NULL; - next->coder->mf.hash_size_sum = 0; + next->coder->mf.son = NULL; + next->coder->mf.hash_count = 0; next->coder->mf.sons_count = 0; next->coder->next = LZMA_NEXT_CODER_INIT; diff --git a/src/liblzma/lz/lz_encoder.h b/src/liblzma/lz/lz_encoder.h index 741c4532280d..dad9c6b29980 100644 --- a/src/liblzma/lz/lz_encoder.h +++ b/src/liblzma/lz/lz_encoder.h @@ -119,7 +119,7 @@ struct lzma_mf_s { lzma_action action; /// Number of elements in hash[] - uint32_t hash_size_sum; + uint32_t hash_count; /// Number of elements in son[] uint32_t sons_count; @@ -199,7 +199,7 @@ typedef struct { size_t *restrict out_pos, size_t out_size); /// Free allocated resources - void (*end)(lzma_coder *coder, lzma_allocator *allocator); + void (*end)(lzma_coder *coder, const lzma_allocator *allocator); /// Update the options in the middle of the encoding. lzma_ret (*options_update)(lzma_coder *coder, @@ -296,10 +296,10 @@ mf_read(lzma_mf *mf, uint8_t *out, size_t *out_pos, size_t out_size, extern lzma_ret lzma_lz_encoder_init( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, lzma_ret (*lz_init)(lzma_lz_encoder *lz, - lzma_allocator *allocator, const void *options, + const lzma_allocator *allocator, const void *options, lzma_lz_options *lz_options)); diff --git a/src/liblzma/lz/lz_encoder_mf.c b/src/liblzma/lz/lz_encoder_mf.c index f82a1c1d2959..78520779f1b6 100644 --- a/src/liblzma/lz/lz_encoder_mf.c +++ b/src/liblzma/lz/lz_encoder_mf.c @@ -13,6 +13,7 @@ #include "lz_encoder.h" #include "lz_encoder_hash.h" +#include "memcmplen.h" /// \brief Find matches starting from the current byte @@ -65,9 +66,7 @@ lzma_mf_find(lzma_mf *mf, uint32_t *count_ptr, lzma_match *matches) // here because the match distances are zero based. const uint8_t *p2 = p1 - matches[count - 1].dist - 1; - while (len_best < limit - && p1[len_best] == p2[len_best]) - ++len_best; + len_best = lzma_memcmplen(p1, p2, len_best, limit); } } @@ -116,24 +115,27 @@ normalize(lzma_mf *mf) = (MUST_NORMALIZE_POS - mf->cyclic_size); // & (~(UINT32_C(1) << 10) - 1); - const uint32_t count = mf->hash_size_sum + mf->sons_count; - uint32_t *hash = mf->hash; - - for (uint32_t i = 0; i < count; ++i) { + for (uint32_t i = 0; i < mf->hash_count; ++i) { // If the distance is greater than the dictionary size, // we can simply mark the hash element as empty. + if (mf->hash[i] <= subvalue) + mf->hash[i] = EMPTY_HASH_VALUE; + else + mf->hash[i] -= subvalue; + } + + for (uint32_t i = 0; i < mf->sons_count; ++i) { + // Do the same for mf->son. // - // NOTE: Only the first mf->hash_size_sum elements are - // initialized for sure. There may be uninitialized elements - // in mf->son. Since we go through both mf->hash and - // mf->son here in normalization, Valgrind may complain - // that the "if" below depends on uninitialized value. In - // this case it is safe to ignore the warning. See also the - // comments in lz_encoder_init() in lz_encoder.c. - if (hash[i] <= subvalue) - hash[i] = EMPTY_HASH_VALUE; + // NOTE: There may be uninitialized elements in mf->son. + // Valgrind may complain that the "if" below depends on + // an uninitialized value. In this case it is safe to ignore + // the warning. See also the comments in lz_encoder_init() + // in lz_encoder.c. + if (mf->son[i] <= subvalue) + mf->son[i] = EMPTY_HASH_VALUE; else - hash[i] -= subvalue; + mf->son[i] -= subvalue; } // Update offset to match the new locations. @@ -269,10 +271,7 @@ hc_find_func( + (delta > cyclic_pos ? cyclic_size : 0)]; if (pb[len_best] == cur[len_best] && pb[0] == cur[0]) { - uint32_t len = 0; - while (++len != len_limit) - if (pb[len] != cur[len]) - break; + uint32_t len = lzma_memcmplen(pb, cur, 1, len_limit); if (len_best < len) { len_best = len; @@ -318,9 +317,8 @@ lzma_mf_hc3_find(lzma_mf *mf, lzma_match *matches) uint32_t len_best = 2; if (delta2 < mf->cyclic_size && *(cur - delta2) == *cur) { - for ( ; len_best != len_limit; ++len_best) - if (*(cur + len_best - delta2) != cur[len_best]) - break; + len_best = lzma_memcmplen(cur - delta2, cur, + len_best, len_limit); matches[0].len = len_best; matches[0].dist = delta2 - 1; @@ -397,9 +395,8 @@ lzma_mf_hc4_find(lzma_mf *mf, lzma_match *matches) } if (matches_count != 0) { - for ( ; len_best != len_limit; ++len_best) - if (*(cur + len_best - delta2) != cur[len_best]) - break; + len_best = lzma_memcmplen(cur - delta2, cur, + len_best, len_limit); matches[matches_count - 1].len = len_best; @@ -484,9 +481,7 @@ bt_find_func( uint32_t len = my_min(len0, len1); if (pb[len] == cur[len]) { - while (++len != len_limit) - if (pb[len] != cur[len]) - break; + len = lzma_memcmplen(pb, cur, len + 1, len_limit); if (len_best < len) { len_best = len; @@ -549,9 +544,7 @@ bt_skip_func( uint32_t len = my_min(len0, len1); if (pb[len] == cur[len]) { - while (++len != len_limit) - if (pb[len] != cur[len]) - break; + len = lzma_memcmplen(pb, cur, len + 1, len_limit); if (len == len_limit) { *ptr1 = pair[0]; @@ -639,9 +632,8 @@ lzma_mf_bt3_find(lzma_mf *mf, lzma_match *matches) uint32_t len_best = 2; if (delta2 < mf->cyclic_size && *(cur - delta2) == *cur) { - for ( ; len_best != len_limit; ++len_best) - if (*(cur + len_best - delta2) != cur[len_best]) - break; + len_best = lzma_memcmplen( + cur, cur - delta2, len_best, len_limit); matches[0].len = len_best; matches[0].dist = delta2 - 1; @@ -712,9 +704,8 @@ lzma_mf_bt4_find(lzma_mf *mf, lzma_match *matches) } if (matches_count != 0) { - for ( ; len_best != len_limit; ++len_best) - if (*(cur + len_best - delta2) != cur[len_best]) - break; + len_best = lzma_memcmplen( + cur, cur - delta2, len_best, len_limit); matches[matches_count - 1].len = len_best; diff --git a/src/liblzma/lzma/fastpos.h b/src/liblzma/lzma/fastpos.h index 4aea23181ab6..a3feea58d8d3 100644 --- a/src/liblzma/lzma/fastpos.h +++ b/src/liblzma/lzma/fastpos.h @@ -14,15 +14,15 @@ #ifndef LZMA_FASTPOS_H #define LZMA_FASTPOS_H -// LZMA encodes match distances (positions) by storing the highest two -// bits using a six-bit value [0, 63], and then the missing lower bits. -// Dictionary size is also stored using this encoding in the new .lzma +// LZMA encodes match distances by storing the highest two bits using +// a six-bit value [0, 63], and then the missing lower bits. +// Dictionary size is also stored using this encoding in the .xz // file format header. // // fastpos.h provides a way to quickly find out the correct six-bit // values. The following table gives some examples of this encoding: // -// pos return +// dist return // 0 0 // 1 1 // 2 2 @@ -48,10 +48,10 @@ // Provided functions or macros // ---------------------------- // -// get_pos_slot(pos) is the basic version. get_pos_slot_2(pos) -// assumes that pos >= FULL_DISTANCES, thus the result is at least -// FULL_DISTANCES_BITS * 2. Using get_pos_slot(pos) instead of -// get_pos_slot_2(pos) would give the same result, but get_pos_slot_2(pos) +// get_dist_slot(dist) is the basic version. get_dist_slot_2(dist) +// assumes that dist >= FULL_DISTANCES, thus the result is at least +// FULL_DISTANCES_BITS * 2. Using get_dist_slot(dist) instead of +// get_dist_slot_2(dist) would give the same result, but get_dist_slot_2(dist) // should be tiny bit faster due to the assumption being made. // // @@ -76,13 +76,14 @@ // slightly faster, but sometimes it is a lot slower. #ifdef HAVE_SMALL -# define get_pos_slot(pos) ((pos) <= 4 ? (pos) : get_pos_slot_2(pos)) +# define get_dist_slot(dist) \ + ((dist) <= 4 ? (dist) : get_dist_slot_2(dist)) static inline uint32_t -get_pos_slot_2(uint32_t pos) +get_dist_slot_2(uint32_t dist) { - const uint32_t i = bsr32(pos); - return (i + i) + ((pos >> (i - 1)) & 1); + const uint32_t i = bsr32(dist); + return (i + i) + ((dist >> (i - 1)) & 1); } @@ -99,39 +100,39 @@ extern const uint8_t lzma_fastpos[1 << FASTPOS_BITS]; #define fastpos_limit(extra, n) \ (UINT32_C(1) << (FASTPOS_BITS + fastpos_shift(extra, n))) -#define fastpos_result(pos, extra, n) \ - lzma_fastpos[(pos) >> fastpos_shift(extra, n)] \ +#define fastpos_result(dist, extra, n) \ + lzma_fastpos[(dist) >> fastpos_shift(extra, n)] \ + 2 * fastpos_shift(extra, n) static inline uint32_t -get_pos_slot(uint32_t pos) +get_dist_slot(uint32_t dist) { // If it is small enough, we can pick the result directly from // the precalculated table. - if (pos < fastpos_limit(0, 0)) - return lzma_fastpos[pos]; + if (dist < fastpos_limit(0, 0)) + return lzma_fastpos[dist]; - if (pos < fastpos_limit(0, 1)) - return fastpos_result(pos, 0, 1); + if (dist < fastpos_limit(0, 1)) + return fastpos_result(dist, 0, 1); - return fastpos_result(pos, 0, 2); + return fastpos_result(dist, 0, 2); } #ifdef FULL_DISTANCES_BITS static inline uint32_t -get_pos_slot_2(uint32_t pos) +get_dist_slot_2(uint32_t dist) { - assert(pos >= FULL_DISTANCES); + assert(dist >= FULL_DISTANCES); - if (pos < fastpos_limit(FULL_DISTANCES_BITS - 1, 0)) - return fastpos_result(pos, FULL_DISTANCES_BITS - 1, 0); + if (dist < fastpos_limit(FULL_DISTANCES_BITS - 1, 0)) + return fastpos_result(dist, FULL_DISTANCES_BITS - 1, 0); - if (pos < fastpos_limit(FULL_DISTANCES_BITS - 1, 1)) - return fastpos_result(pos, FULL_DISTANCES_BITS - 1, 1); + if (dist < fastpos_limit(FULL_DISTANCES_BITS - 1, 1)) + return fastpos_result(dist, FULL_DISTANCES_BITS - 1, 1); - return fastpos_result(pos, FULL_DISTANCES_BITS - 1, 2); + return fastpos_result(dist, FULL_DISTANCES_BITS - 1, 2); } #endif diff --git a/src/liblzma/lzma/lzma2_decoder.c b/src/liblzma/lzma/lzma2_decoder.c index 3e42575d5b80..84982d2c4286 100644 --- a/src/liblzma/lzma/lzma2_decoder.c +++ b/src/liblzma/lzma/lzma2_decoder.c @@ -209,7 +209,7 @@ lzma2_decode(lzma_coder *restrict coder, lzma_dict *restrict dict, static void -lzma2_decoder_end(lzma_coder *coder, lzma_allocator *allocator) +lzma2_decoder_end(lzma_coder *coder, const lzma_allocator *allocator) { assert(coder->lzma.end == NULL); lzma_free(coder->lzma.coder, allocator); @@ -221,7 +221,7 @@ lzma2_decoder_end(lzma_coder *coder, lzma_allocator *allocator) static lzma_ret -lzma2_decoder_init(lzma_lz_decoder *lz, lzma_allocator *allocator, +lzma2_decoder_init(lzma_lz_decoder *lz, const lzma_allocator *allocator, const void *opt, lzma_lz_options *lz_options) { if (lz->coder == NULL) { @@ -248,7 +248,7 @@ lzma2_decoder_init(lzma_lz_decoder *lz, lzma_allocator *allocator, extern lzma_ret -lzma_lzma2_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_lzma2_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { // LZMA2 can only be the last filter in the chain. This is enforced @@ -269,7 +269,7 @@ lzma_lzma2_decoder_memusage(const void *options) extern lzma_ret -lzma_lzma2_props_decode(void **options, lzma_allocator *allocator, +lzma_lzma2_props_decode(void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size) { if (props_size != 1) diff --git a/src/liblzma/lzma/lzma2_decoder.h b/src/liblzma/lzma/lzma2_decoder.h index fac4ac487b07..ef2dcbfa76f0 100644 --- a/src/liblzma/lzma/lzma2_decoder.h +++ b/src/liblzma/lzma/lzma2_decoder.h @@ -17,12 +17,13 @@ #include "common.h" extern lzma_ret lzma_lzma2_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern uint64_t lzma_lzma2_decoder_memusage(const void *options); extern lzma_ret lzma_lzma2_props_decode( - void **options, lzma_allocator *allocator, + void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size); #endif diff --git a/src/liblzma/lzma/lzma2_encoder.c b/src/liblzma/lzma/lzma2_encoder.c index 992720ca6d5a..b6756bfc2b1c 100644 --- a/src/liblzma/lzma/lzma2_encoder.c +++ b/src/liblzma/lzma/lzma2_encoder.c @@ -262,7 +262,7 @@ lzma2_encode(lzma_coder *restrict coder, lzma_mf *restrict mf, static void -lzma2_encoder_end(lzma_coder *coder, lzma_allocator *allocator) +lzma2_encoder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_free(coder->lzma, allocator); lzma_free(coder, allocator); @@ -304,7 +304,7 @@ lzma2_encoder_options_update(lzma_coder *coder, const lzma_filter *filter) static lzma_ret -lzma2_encoder_init(lzma_lz_encoder *lz, lzma_allocator *allocator, +lzma2_encoder_init(lzma_lz_encoder *lz, const lzma_allocator *allocator, const void *options, lzma_lz_options *lz_options) { if (options == NULL) @@ -349,7 +349,7 @@ lzma2_encoder_init(lzma_lz_encoder *lz, lzma_allocator *allocator, extern lzma_ret -lzma_lzma2_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_lzma2_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { return lzma_lz_encoder_init( @@ -387,7 +387,17 @@ lzma_lzma2_props_encode(const void *options, uint8_t *out) if (d == UINT32_MAX) out[0] = 40; else - out[0] = get_pos_slot(d + 1) - 24; + out[0] = get_dist_slot(d + 1) - 24; return LZMA_OK; } + + +extern uint64_t +lzma_lzma2_block_size(const void *options) +{ + const lzma_options_lzma *const opt = options; + + // Use at least 1 MiB to keep compression ratio better. + return my_max((uint64_t)(opt->dict_size) * 3, UINT64_C(1) << 20); +} diff --git a/src/liblzma/lzma/lzma2_encoder.h b/src/liblzma/lzma/lzma2_encoder.h index ca19ef4691cc..515f1839347a 100644 --- a/src/liblzma/lzma/lzma2_encoder.h +++ b/src/liblzma/lzma/lzma2_encoder.h @@ -31,11 +31,13 @@ extern lzma_ret lzma_lzma2_encoder_init( - lzma_next_coder *next, lzma_allocator *allocator, + lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters); extern uint64_t lzma_lzma2_encoder_memusage(const void *options); extern lzma_ret lzma_lzma2_props_encode(const void *options, uint8_t *out); +extern uint64_t lzma_lzma2_block_size(const void *options); + #endif diff --git a/src/liblzma/lzma/lzma_common.h b/src/liblzma/lzma/lzma_common.h index e31e285f9a52..09efd3872917 100644 --- a/src/liblzma/lzma/lzma_common.h +++ b/src/liblzma/lzma/lzma_common.h @@ -171,53 +171,54 @@ literal_init(probability (*probs)[LITERAL_CODER_SIZE], // Match distance // //////////////////// -// Different set of probabilities is used for match distances that have very +// Different sets of probabilities are used for match distances that have very // short match length: Lengths of 2, 3, and 4 bytes have a separate set of // probabilities for each length. The matches with longer length use a shared // set of probabilities. -#define LEN_TO_POS_STATES 4 +#define DIST_STATES 4 // Macro to get the index of the appropriate probability array. -#define get_len_to_pos_state(len) \ - ((len) < LEN_TO_POS_STATES + MATCH_LEN_MIN \ +#define get_dist_state(len) \ + ((len) < DIST_STATES + MATCH_LEN_MIN \ ? (len) - MATCH_LEN_MIN \ - : LEN_TO_POS_STATES - 1) + : DIST_STATES - 1) -// The highest two bits of a match distance (pos slot) are encoded using six -// bits. See fastpos.h for more explanation. -#define POS_SLOT_BITS 6 -#define POS_SLOTS (1 << POS_SLOT_BITS) +// The highest two bits of a match distance (distance slot) are encoded +// using six bits. See fastpos.h for more explanation. +#define DIST_SLOT_BITS 6 +#define DIST_SLOTS (1 << DIST_SLOT_BITS) // Match distances up to 127 are fully encoded using probabilities. Since -// the highest two bits (pos slot) are always encoded using six bits, the -// distances 0-3 don't need any additional bits to encode, since the pos -// slot itself is the same as the actual distance. START_POS_MODEL_INDEX -// indicates the first pos slot where at least one additional bit is needed. -#define START_POS_MODEL_INDEX 4 +// the highest two bits (distance slot) are always encoded using six bits, +// the distances 0-3 don't need any additional bits to encode, since the +// distance slot itself is the same as the actual distance. DIST_MODEL_START +// indicates the first distance slot where at least one additional bit is +// needed. +#define DIST_MODEL_START 4 // Match distances greater than 127 are encoded in three pieces: -// - pos slot: the highest two bits +// - distance slot: the highest two bits // - direct bits: 2-26 bits below the highest two bits // - alignment bits: four lowest bits // // Direct bits don't use any probabilities. // -// The pos slot value of 14 is for distances 128-191 (see the table in +// The distance slot value of 14 is for distances 128-191 (see the table in // fastpos.h to understand why). -#define END_POS_MODEL_INDEX 14 +#define DIST_MODEL_END 14 -// Pos slots that indicate a distance <= 127. -#define FULL_DISTANCES_BITS (END_POS_MODEL_INDEX / 2) +// Distance slots that indicate a distance <= 127. +#define FULL_DISTANCES_BITS (DIST_MODEL_END / 2) #define FULL_DISTANCES (1 << FULL_DISTANCES_BITS) // For match distances greater than 127, only the highest two bits and the // lowest four bits (alignment) is encoded using probabilities. #define ALIGN_BITS 4 -#define ALIGN_TABLE_SIZE (1 << ALIGN_BITS) -#define ALIGN_MASK (ALIGN_TABLE_SIZE - 1) +#define ALIGN_SIZE (1 << ALIGN_BITS) +#define ALIGN_MASK (ALIGN_SIZE - 1) // LZMA remembers the four most recent match distances. Reusing these distances // tends to take less space than re-encoding the actual distance value. -#define REP_DISTANCES 4 +#define REPS 4 #endif diff --git a/src/liblzma/lzma/lzma_decoder.c b/src/liblzma/lzma/lzma_decoder.c index 9979bb4261b8..b8f931705bf9 100644 --- a/src/liblzma/lzma/lzma_decoder.c +++ b/src/liblzma/lzma/lzma_decoder.c @@ -193,15 +193,15 @@ struct lzma_coder_s { /// Probability tree for the highest two bits of the match distance. /// There is a separate probability tree for match lengths of /// 2 (i.e. MATCH_LEN_MIN), 3, 4, and [5, 273]. - probability pos_slot[LEN_TO_POS_STATES][POS_SLOTS]; + probability dist_slot[DIST_STATES][DIST_SLOTS]; /// Probability trees for additional bits for match distance when the /// distance is in the range [4, 127]. - probability pos_special[FULL_DISTANCES - END_POS_MODEL_INDEX]; + probability pos_special[FULL_DISTANCES - DIST_MODEL_END]; /// Probability tree for the lowest four bits of a match distance /// that is equal to or greater than 128. - probability pos_align[ALIGN_TABLE_SIZE]; + probability pos_align[ALIGN_SIZE]; /// Length of a normal match lzma_length_decoder match_len_decoder; @@ -245,8 +245,8 @@ struct lzma_coder_s { SEQ_LITERAL_WRITE, SEQ_IS_REP, seq_len(SEQ_MATCH_LEN), - seq_6(SEQ_POS_SLOT), - SEQ_POS_MODEL, + seq_6(SEQ_DIST_SLOT), + SEQ_DIST_MODEL, SEQ_DIRECT, seq_4(SEQ_ALIGN), SEQ_EOPM, @@ -289,8 +289,12 @@ lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr, // Initialization // //////////////////// - if (!rc_read_init(&coder->rc, in, in_pos, in_size)) - return LZMA_OK; + { + const lzma_ret ret = rc_read_init( + &coder->rc, in, in_pos, in_size); + if (ret != LZMA_STREAM_END) + return ret; + } /////////////// // Variables // @@ -502,28 +506,28 @@ lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr, // Prepare to decode the highest two bits of the // match distance. - probs = coder->pos_slot[get_len_to_pos_state(len)]; + probs = coder->dist_slot[get_dist_state(len)]; symbol = 1; #ifdef HAVE_SMALL - case SEQ_POS_SLOT: + case SEQ_DIST_SLOT: do { - rc_bit(probs[symbol], , , SEQ_POS_SLOT); - } while (symbol < POS_SLOTS); + rc_bit(probs[symbol], , , SEQ_DIST_SLOT); + } while (symbol < DIST_SLOTS); #else - rc_bit_case(probs[symbol], , , SEQ_POS_SLOT0); - rc_bit_case(probs[symbol], , , SEQ_POS_SLOT1); - rc_bit_case(probs[symbol], , , SEQ_POS_SLOT2); - rc_bit_case(probs[symbol], , , SEQ_POS_SLOT3); - rc_bit_case(probs[symbol], , , SEQ_POS_SLOT4); - rc_bit_case(probs[symbol], , , SEQ_POS_SLOT5); + rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT0); + rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT1); + rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT2); + rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT3); + rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT4); + rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT5); #endif // Get rid of the highest bit that was needed for // indexing of the probability array. - symbol -= POS_SLOTS; + symbol -= DIST_SLOTS; assert(symbol <= 63); - if (symbol < START_POS_MODEL_INDEX) { + if (symbol < DIST_MODEL_START) { // Match distances [0, 3] have only two bits. rep0 = symbol; } else { @@ -533,7 +537,7 @@ lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr, assert(limit >= 1 && limit <= 30); rep0 = 2 + (symbol & 1); - if (symbol < END_POS_MODEL_INDEX) { + if (symbol < DIST_MODEL_END) { // Prepare to decode the low bits for // a distance of [4, 127]. assert(limit <= 5); @@ -553,12 +557,12 @@ lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr, - symbol - 1; symbol = 1; offset = 0; - case SEQ_POS_MODEL: + case SEQ_DIST_MODEL: #ifdef HAVE_SMALL do { rc_bit(probs[symbol], , rep0 += 1 << offset, - SEQ_POS_MODEL); + SEQ_DIST_MODEL); } while (++offset < limit); #else switch (limit) { @@ -566,25 +570,25 @@ lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr, assert(offset == 0); rc_bit(probs[symbol], , rep0 += 1, - SEQ_POS_MODEL); + SEQ_DIST_MODEL); ++offset; --limit; case 4: rc_bit(probs[symbol], , rep0 += 1 << offset, - SEQ_POS_MODEL); + SEQ_DIST_MODEL); ++offset; --limit; case 3: rc_bit(probs[symbol], , rep0 += 1 << offset, - SEQ_POS_MODEL); + SEQ_DIST_MODEL); ++offset; --limit; case 2: rc_bit(probs[symbol], , rep0 += 1 << offset, - SEQ_POS_MODEL); + SEQ_DIST_MODEL); ++offset; --limit; case 1: @@ -596,7 +600,7 @@ lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr, // "symbol". rc_bit_last(probs[symbol], , rep0 += 1 << offset, - SEQ_POS_MODEL); + SEQ_DIST_MODEL); } #endif } else { @@ -637,7 +641,7 @@ lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr, rc_bit(coder->pos_align[symbol], , rep0 += 4, SEQ_ALIGN2); case SEQ_ALIGN3: - // Like in SEQ_POS_MODEL, we don't + // Like in SEQ_DIST_MODEL, we don't // need "symbol" for anything else // than indexing the probability array. rc_bit_last(coder->pos_align[symbol], , @@ -891,10 +895,10 @@ lzma_decoder_reset(lzma_coder *coder, const void *opt) bit_reset(coder->is_rep2[i]); } - for (uint32_t i = 0; i < LEN_TO_POS_STATES; ++i) - bittree_reset(coder->pos_slot[i], POS_SLOT_BITS); + for (uint32_t i = 0; i < DIST_STATES; ++i) + bittree_reset(coder->dist_slot[i], DIST_SLOT_BITS); - for (uint32_t i = 0; i < FULL_DISTANCES - END_POS_MODEL_INDEX; ++i) + for (uint32_t i = 0; i < FULL_DISTANCES - DIST_MODEL_END; ++i) bit_reset(coder->pos_special[i]); bittree_reset(coder->pos_align, ALIGN_BITS); @@ -933,7 +937,7 @@ lzma_decoder_reset(lzma_coder *coder, const void *opt) extern lzma_ret -lzma_lzma_decoder_create(lzma_lz_decoder *lz, lzma_allocator *allocator, +lzma_lzma_decoder_create(lzma_lz_decoder *lz, const lzma_allocator *allocator, const void *opt, lzma_lz_options *lz_options) { if (lz->coder == NULL) { @@ -961,7 +965,7 @@ lzma_lzma_decoder_create(lzma_lz_decoder *lz, lzma_allocator *allocator, /// initialization (lzma_lzma_decoder_init() passes function pointer to /// the LZ initialization). static lzma_ret -lzma_decoder_init(lzma_lz_decoder *lz, lzma_allocator *allocator, +lzma_decoder_init(lzma_lz_decoder *lz, const lzma_allocator *allocator, const void *options, lzma_lz_options *lz_options) { if (!is_lclppb_valid(options)) @@ -978,7 +982,7 @@ lzma_decoder_init(lzma_lz_decoder *lz, lzma_allocator *allocator, extern lzma_ret -lzma_lzma_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_lzma_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { // LZMA can only be the last filter in the chain. This is enforced @@ -1025,7 +1029,7 @@ lzma_lzma_decoder_memusage(const void *options) extern lzma_ret -lzma_lzma_props_decode(void **options, lzma_allocator *allocator, +lzma_lzma_props_decode(void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size) { if (props_size != 5) diff --git a/src/liblzma/lzma/lzma_decoder.h b/src/liblzma/lzma/lzma_decoder.h index a463a76fc694..fa8ecb23e43c 100644 --- a/src/liblzma/lzma/lzma_decoder.h +++ b/src/liblzma/lzma/lzma_decoder.h @@ -19,12 +19,13 @@ /// Allocates and initializes LZMA decoder extern lzma_ret lzma_lzma_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern uint64_t lzma_lzma_decoder_memusage(const void *options); extern lzma_ret lzma_lzma_props_decode( - void **options, lzma_allocator *allocator, + void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size); @@ -40,7 +41,7 @@ extern bool lzma_lzma_lclppb_decode( /// Allocate and setup function pointers only. This is used by LZMA1 and /// LZMA2 decoders. extern lzma_ret lzma_lzma_decoder_create( - lzma_lz_decoder *lz, lzma_allocator *allocator, + lzma_lz_decoder *lz, const lzma_allocator *allocator, const void *opt, lzma_lz_options *lz_options); /// Gets memory usage without validating lc/lp/pb. This is used by LZMA2 diff --git a/src/liblzma/lzma/lzma_encoder.c b/src/liblzma/lzma/lzma_encoder.c index 0b9ee9e15079..31cb4f044829 100644 --- a/src/liblzma/lzma/lzma_encoder.c +++ b/src/liblzma/lzma/lzma_encoder.c @@ -148,28 +148,28 @@ match(lzma_coder *coder, const uint32_t pos_state, length(&coder->rc, &coder->match_len_encoder, pos_state, len, coder->fast_mode); - const uint32_t pos_slot = get_pos_slot(distance); - const uint32_t len_to_pos_state = get_len_to_pos_state(len); - rc_bittree(&coder->rc, coder->pos_slot[len_to_pos_state], - POS_SLOT_BITS, pos_slot); - - if (pos_slot >= START_POS_MODEL_INDEX) { - const uint32_t footer_bits = (pos_slot >> 1) - 1; - const uint32_t base = (2 | (pos_slot & 1)) << footer_bits; - const uint32_t pos_reduced = distance - base; - - if (pos_slot < END_POS_MODEL_INDEX) { - // Careful here: base - pos_slot - 1 can be -1, but + const uint32_t dist_slot = get_dist_slot(distance); + const uint32_t dist_state = get_dist_state(len); + rc_bittree(&coder->rc, coder->dist_slot[dist_state], + DIST_SLOT_BITS, dist_slot); + + if (dist_slot >= DIST_MODEL_START) { + const uint32_t footer_bits = (dist_slot >> 1) - 1; + const uint32_t base = (2 | (dist_slot & 1)) << footer_bits; + const uint32_t dist_reduced = distance - base; + + if (dist_slot < DIST_MODEL_END) { + // Careful here: base - dist_slot - 1 can be -1, but // rc_bittree_reverse starts at probs[1], not probs[0]. rc_bittree_reverse(&coder->rc, - coder->pos_special + base - pos_slot - 1, - footer_bits, pos_reduced); + coder->dist_special + base - dist_slot - 1, + footer_bits, dist_reduced); } else { - rc_direct(&coder->rc, pos_reduced >> ALIGN_BITS, + rc_direct(&coder->rc, dist_reduced >> ALIGN_BITS, footer_bits - ALIGN_BITS); rc_bittree_reverse( - &coder->rc, coder->pos_align, - ALIGN_BITS, pos_reduced & ALIGN_MASK); + &coder->rc, coder->dist_align, + ALIGN_BITS, dist_reduced & ALIGN_MASK); ++coder->align_price_count; } } @@ -247,7 +247,7 @@ encode_symbol(lzma_coder *coder, lzma_mf *mf, rc_bit(&coder->rc, &coder->is_match[coder->state][pos_state], 1); - if (back < REP_DISTANCES) { + if (back < REPS) { // It's a repeated match i.e. the same distance // has been used earlier. rc_bit(&coder->rc, &coder->is_rep[coder->state], 1); @@ -255,7 +255,7 @@ encode_symbol(lzma_coder *coder, lzma_mf *mf, } else { // Normal match rc_bit(&coder->rc, &coder->is_rep[coder->state], 0); - match(coder, pos_state, back - REP_DISTANCES, len); + match(coder, pos_state, back - REPS, len); } } @@ -353,9 +353,9 @@ lzma_lzma_encode(lzma_coder *restrict coder, lzma_mf *restrict mf, // Get optimal match (repeat position and length). // Value ranges for pos: - // - [0, REP_DISTANCES): repeated match - // - [REP_DISTANCES, UINT32_MAX): - // match at (pos - REP_DISTANCES) + // - [0, REPS): repeated match + // - [REPS, UINT32_MAX): + // match at (pos - REPS) // - UINT32_MAX: not a match but a literal // Value ranges for len: // - [MATCH_LEN_MIN, MATCH_LEN_MAX] @@ -487,7 +487,7 @@ lzma_lzma_encoder_reset(lzma_coder *coder, const lzma_options_lzma *options) // State coder->state = STATE_LIT_LIT; - for (size_t i = 0; i < REP_DISTANCES; ++i) + for (size_t i = 0; i < REPS; ++i) coder->reps[i] = 0; literal_init(coder->literal, options->lc, options->lp); @@ -505,14 +505,14 @@ lzma_lzma_encoder_reset(lzma_coder *coder, const lzma_options_lzma *options) bit_reset(coder->is_rep2[i]); } - for (size_t i = 0; i < FULL_DISTANCES - END_POS_MODEL_INDEX; ++i) - bit_reset(coder->pos_special[i]); + for (size_t i = 0; i < FULL_DISTANCES - DIST_MODEL_END; ++i) + bit_reset(coder->dist_special[i]); // Bit tree encoders - for (size_t i = 0; i < LEN_TO_POS_STATES; ++i) - bittree_reset(coder->pos_slot[i], POS_SLOT_BITS); + for (size_t i = 0; i < DIST_STATES; ++i) + bittree_reset(coder->dist_slot[i], DIST_SLOT_BITS); - bittree_reset(coder->pos_align, ALIGN_BITS); + bittree_reset(coder->dist_align, ALIGN_BITS); // Length encoders length_encoder_reset(&coder->match_len_encoder, @@ -545,7 +545,8 @@ lzma_lzma_encoder_reset(lzma_coder *coder, const lzma_options_lzma *options) extern lzma_ret -lzma_lzma_encoder_create(lzma_coder **coder_ptr, lzma_allocator *allocator, +lzma_lzma_encoder_create(lzma_coder **coder_ptr, + const lzma_allocator *allocator, const lzma_options_lzma *options, lzma_lz_options *lz_options) { // Allocate lzma_coder if it wasn't already allocated. @@ -604,7 +605,7 @@ lzma_lzma_encoder_create(lzma_coder **coder_ptr, lzma_allocator *allocator, static lzma_ret -lzma_encoder_init(lzma_lz_encoder *lz, lzma_allocator *allocator, +lzma_encoder_init(lzma_lz_encoder *lz, const lzma_allocator *allocator, const void *options, lzma_lz_options *lz_options) { lz->code = &lzma_encode; @@ -614,7 +615,7 @@ lzma_encoder_init(lzma_lz_encoder *lz, lzma_allocator *allocator, extern lzma_ret -lzma_lzma_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_lzma_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters) { return lzma_lz_encoder_init( diff --git a/src/liblzma/lzma/lzma_encoder.h b/src/liblzma/lzma/lzma_encoder.h index 835e1f583304..cc9cc2f27ecb 100644 --- a/src/liblzma/lzma/lzma_encoder.h +++ b/src/liblzma/lzma/lzma_encoder.h @@ -18,7 +18,8 @@ extern lzma_ret lzma_lzma_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern uint64_t lzma_lzma_encoder_memusage(const void *options); @@ -35,7 +36,7 @@ extern bool lzma_lzma_lclppb_encode( /// Initializes raw LZMA encoder; this is used by LZMA2. extern lzma_ret lzma_lzma_encoder_create( - lzma_coder **coder_ptr, lzma_allocator *allocator, + lzma_coder **coder_ptr, const lzma_allocator *allocator, const lzma_options_lzma *options, lzma_lz_options *lz_options); diff --git a/src/liblzma/lzma/lzma_encoder_optimum_fast.c b/src/liblzma/lzma/lzma_encoder_optimum_fast.c index f835f69356bd..8922cbd91d3f 100644 --- a/src/liblzma/lzma/lzma_encoder_optimum_fast.c +++ b/src/liblzma/lzma/lzma_encoder_optimum_fast.c @@ -10,6 +10,7 @@ /////////////////////////////////////////////////////////////////////////////// #include "lzma_encoder_private.h" +#include "memcmplen.h" #define change_pair(small_dist, big_dist) \ @@ -46,7 +47,7 @@ lzma_lzma_optimum_fast(lzma_coder *restrict coder, lzma_mf *restrict mf, uint32_t rep_len = 0; uint32_t rep_index = 0; - for (uint32_t i = 0; i < REP_DISTANCES; ++i) { + for (uint32_t i = 0; i < REPS; ++i) { // Pointer to the beginning of the match candidate const uint8_t *const buf_back = buf - coder->reps[i] - 1; @@ -57,9 +58,8 @@ lzma_lzma_optimum_fast(lzma_coder *restrict coder, lzma_mf *restrict mf, // The first two bytes matched. // Calculate the length of the match. - uint32_t len; - for (len = 2; len < buf_avail - && buf[len] == buf_back[len]; ++len) ; + const uint32_t len = lzma_memcmplen( + buf, buf_back, 2, buf_avail); // If we have found a repeated match that is at least // nice_len long, return it immediately. @@ -79,8 +79,7 @@ lzma_lzma_optimum_fast(lzma_coder *restrict coder, lzma_mf *restrict mf, // We didn't find a long enough repeated match. Encode it as a normal // match if the match length is at least nice_len. if (len_main >= nice_len) { - *back_res = coder->matches[matches_count - 1].dist - + REP_DISTANCES; + *back_res = coder->matches[matches_count - 1].dist + REPS; *len_res = len_main; mf_skip(mf, len_main - 1); return; @@ -155,24 +154,15 @@ lzma_lzma_optimum_fast(lzma_coder *restrict coder, lzma_mf *restrict mf, const uint32_t limit = len_main - 1; - for (uint32_t i = 0; i < REP_DISTANCES; ++i) { - const uint8_t *const buf_back = buf - coder->reps[i] - 1; - - if (not_equal_16(buf, buf_back)) - continue; - - uint32_t len; - for (len = 2; len < limit - && buf[len] == buf_back[len]; ++len) ; - - if (len >= limit) { + for (uint32_t i = 0; i < REPS; ++i) { + if (memcmp(buf, buf - coder->reps[i] - 1, limit) == 0) { *back_res = UINT32_MAX; *len_res = 1; return; } } - *back_res = back_main + REP_DISTANCES; + *back_res = back_main + REPS; *len_res = len_main; mf_skip(mf, len_main - 2); return; diff --git a/src/liblzma/lzma/lzma_encoder_optimum_normal.c b/src/liblzma/lzma/lzma_encoder_optimum_normal.c index 7e856493c8cd..a360579885db 100644 --- a/src/liblzma/lzma/lzma_encoder_optimum_normal.c +++ b/src/liblzma/lzma/lzma_encoder_optimum_normal.c @@ -11,6 +11,7 @@ #include "lzma_encoder_private.h" #include "fastpos.h" +#include "memcmplen.h" //////////// @@ -108,18 +109,18 @@ get_rep_price(const lzma_coder *const coder, const uint32_t rep_index, static inline uint32_t -get_pos_len_price(const lzma_coder *const coder, const uint32_t pos, +get_dist_len_price(const lzma_coder *const coder, const uint32_t dist, const uint32_t len, const uint32_t pos_state) { - const uint32_t len_to_pos_state = get_len_to_pos_state(len); + const uint32_t dist_state = get_dist_state(len); uint32_t price; - if (pos < FULL_DISTANCES) { - price = coder->distances_prices[len_to_pos_state][pos]; + if (dist < FULL_DISTANCES) { + price = coder->dist_prices[dist_state][dist]; } else { - const uint32_t pos_slot = get_pos_slot_2(pos); - price = coder->pos_slot_prices[len_to_pos_state][pos_slot] - + coder->align_prices[pos & ALIGN_MASK]; + const uint32_t dist_slot = get_dist_slot_2(dist); + price = coder->dist_slot_prices[dist_state][dist_slot] + + coder->align_prices[dist & ALIGN_MASK]; } price += get_len_price(&coder->match_len_encoder, len, pos_state); @@ -129,55 +130,53 @@ get_pos_len_price(const lzma_coder *const coder, const uint32_t pos, static void -fill_distances_prices(lzma_coder *coder) +fill_dist_prices(lzma_coder *coder) { - for (uint32_t len_to_pos_state = 0; - len_to_pos_state < LEN_TO_POS_STATES; - ++len_to_pos_state) { + for (uint32_t dist_state = 0; dist_state < DIST_STATES; ++dist_state) { - uint32_t *const pos_slot_prices - = coder->pos_slot_prices[len_to_pos_state]; + uint32_t *const dist_slot_prices + = coder->dist_slot_prices[dist_state]; - // Price to encode the pos_slot. - for (uint32_t pos_slot = 0; - pos_slot < coder->dist_table_size; ++pos_slot) - pos_slot_prices[pos_slot] = rc_bittree_price( - coder->pos_slot[len_to_pos_state], - POS_SLOT_BITS, pos_slot); + // Price to encode the dist_slot. + for (uint32_t dist_slot = 0; + dist_slot < coder->dist_table_size; ++dist_slot) + dist_slot_prices[dist_slot] = rc_bittree_price( + coder->dist_slot[dist_state], + DIST_SLOT_BITS, dist_slot); // For matches with distance >= FULL_DISTANCES, add the price // of the direct bits part of the match distance. (Align bits // are handled by fill_align_prices()). - for (uint32_t pos_slot = END_POS_MODEL_INDEX; - pos_slot < coder->dist_table_size; ++pos_slot) - pos_slot_prices[pos_slot] += rc_direct_price( - ((pos_slot >> 1) - 1) - ALIGN_BITS); + for (uint32_t dist_slot = DIST_MODEL_END; + dist_slot < coder->dist_table_size; + ++dist_slot) + dist_slot_prices[dist_slot] += rc_direct_price( + ((dist_slot >> 1) - 1) - ALIGN_BITS); // Distances in the range [0, 3] are fully encoded with - // pos_slot, so they are used for coder->distances_prices + // dist_slot, so they are used for coder->dist_prices // as is. - for (uint32_t i = 0; i < START_POS_MODEL_INDEX; ++i) - coder->distances_prices[len_to_pos_state][i] - = pos_slot_prices[i]; + for (uint32_t i = 0; i < DIST_MODEL_START; ++i) + coder->dist_prices[dist_state][i] + = dist_slot_prices[i]; } - // Distances in the range [4, 127] depend on pos_slot and pos_special. - // We do this in a loop separate from the above loop to avoid - // redundant calls to get_pos_slot(). - for (uint32_t i = START_POS_MODEL_INDEX; i < FULL_DISTANCES; ++i) { - const uint32_t pos_slot = get_pos_slot(i); - const uint32_t footer_bits = ((pos_slot >> 1) - 1); - const uint32_t base = (2 | (pos_slot & 1)) << footer_bits; + // Distances in the range [4, 127] depend on dist_slot and + // dist_special. We do this in a loop separate from the above + // loop to avoid redundant calls to get_dist_slot(). + for (uint32_t i = DIST_MODEL_START; i < FULL_DISTANCES; ++i) { + const uint32_t dist_slot = get_dist_slot(i); + const uint32_t footer_bits = ((dist_slot >> 1) - 1); + const uint32_t base = (2 | (dist_slot & 1)) << footer_bits; const uint32_t price = rc_bittree_reverse_price( - coder->pos_special + base - pos_slot - 1, + coder->dist_special + base - dist_slot - 1, footer_bits, i - base); - for (uint32_t len_to_pos_state = 0; - len_to_pos_state < LEN_TO_POS_STATES; - ++len_to_pos_state) - coder->distances_prices[len_to_pos_state][i] - = price + coder->pos_slot_prices[ - len_to_pos_state][pos_slot]; + for (uint32_t dist_state = 0; dist_state < DIST_STATES; + ++dist_state) + coder->dist_prices[dist_state][i] + = price + coder->dist_slot_prices[ + dist_state][dist_slot]; } coder->match_price_count = 0; @@ -188,9 +187,9 @@ fill_distances_prices(lzma_coder *coder) static void fill_align_prices(lzma_coder *coder) { - for (uint32_t i = 0; i < ALIGN_TABLE_SIZE; ++i) + for (uint32_t i = 0; i < ALIGN_SIZE; ++i) coder->align_prices[i] = rc_bittree_reverse_price( - coder->pos_align, ALIGN_BITS, i); + coder->dist_align, ALIGN_BITS, i); coder->align_price_count = 0; return; @@ -296,10 +295,10 @@ helper1(lzma_coder *restrict coder, lzma_mf *restrict mf, const uint8_t *const buf = mf_ptr(mf) - 1; - uint32_t rep_lens[REP_DISTANCES]; + uint32_t rep_lens[REPS]; uint32_t rep_max_index = 0; - for (uint32_t i = 0; i < REP_DISTANCES; ++i) { + for (uint32_t i = 0; i < REPS; ++i) { const uint8_t *const buf_back = buf - coder->reps[i] - 1; if (not_equal_16(buf, buf_back)) { @@ -307,13 +306,9 @@ helper1(lzma_coder *restrict coder, lzma_mf *restrict mf, continue; } - uint32_t len_test; - for (len_test = 2; len_test < buf_avail - && buf[len_test] == buf_back[len_test]; - ++len_test) ; + rep_lens[i] = lzma_memcmplen(buf, buf_back, 2, buf_avail); - rep_lens[i] = len_test; - if (len_test > rep_lens[rep_max_index]) + if (rep_lens[i] > rep_lens[rep_max_index]) rep_max_index = i; } @@ -326,8 +321,7 @@ helper1(lzma_coder *restrict coder, lzma_mf *restrict mf, if (len_main >= nice_len) { - *back_res = coder->matches[matches_count - 1].dist - + REP_DISTANCES; + *back_res = coder->matches[matches_count - 1].dist + REPS; *len_res = len_main; mf_skip(mf, len_main - 1); return UINT32_MAX; @@ -381,7 +375,7 @@ helper1(lzma_coder *restrict coder, lzma_mf *restrict mf, coder->opts[1].pos_prev = 0; - for (uint32_t i = 0; i < REP_DISTANCES; ++i) + for (uint32_t i = 0; i < REPS; ++i) coder->opts[0].backs[i] = coder->reps[i]; uint32_t len = len_end; @@ -390,7 +384,7 @@ helper1(lzma_coder *restrict coder, lzma_mf *restrict mf, } while (--len >= 2); - for (uint32_t i = 0; i < REP_DISTANCES; ++i) { + for (uint32_t i = 0; i < REPS; ++i) { uint32_t rep_len = rep_lens[i]; if (rep_len < 2) continue; @@ -426,14 +420,13 @@ helper1(lzma_coder *restrict coder, lzma_mf *restrict mf, for(; ; ++len) { const uint32_t dist = coder->matches[i].dist; const uint32_t cur_and_len_price = normal_match_price - + get_pos_len_price(coder, + + get_dist_len_price(coder, dist, len, pos_state); if (cur_and_len_price < coder->opts[len].price) { coder->opts[len].price = cur_and_len_price; coder->opts[len].pos_prev = 0; - coder->opts[len].back_prev - = dist + REP_DISTANCES; + coder->opts[len].back_prev = dist + REPS; coder->opts[len].prev_1_is_literal = false; } @@ -463,7 +456,7 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf, if (coder->opts[cur].prev_2) { state = coder->opts[coder->opts[cur].pos_prev_2].state; - if (coder->opts[cur].back_prev_2 < REP_DISTANCES) + if (coder->opts[cur].back_prev_2 < REPS) update_long_rep(state); else update_match(state); @@ -492,33 +485,33 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf, update_long_rep(state); } else { pos = coder->opts[cur].back_prev; - if (pos < REP_DISTANCES) + if (pos < REPS) update_long_rep(state); else update_match(state); } - if (pos < REP_DISTANCES) { + if (pos < REPS) { reps[0] = coder->opts[pos_prev].backs[pos]; uint32_t i; for (i = 1; i <= pos; ++i) reps[i] = coder->opts[pos_prev].backs[i - 1]; - for (; i < REP_DISTANCES; ++i) + for (; i < REPS; ++i) reps[i] = coder->opts[pos_prev].backs[i]; } else { - reps[0] = pos - REP_DISTANCES; + reps[0] = pos - REPS; - for (uint32_t i = 1; i < REP_DISTANCES; ++i) + for (uint32_t i = 1; i < REPS; ++i) reps[i] = coder->opts[pos_prev].backs[i - 1]; } } coder->opts[cur].state = state; - for (uint32_t i = 0; i < REP_DISTANCES; ++i) + for (uint32_t i = 0; i < REPS; ++i) coder->opts[cur].backs[i] = reps[i]; const uint32_t cur_price = coder->opts[cur].price; @@ -572,11 +565,7 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf, const uint8_t *const buf_back = buf - reps[0] - 1; const uint32_t limit = my_min(buf_avail_full, nice_len + 1); - uint32_t len_test = 1; - while (len_test < limit && buf[len_test] == buf_back[len_test]) - ++len_test; - - --len_test; + const uint32_t len_test = lzma_memcmplen(buf, buf_back, 1, limit) - 1; if (len_test >= 2) { lzma_lzma_state state_2 = state; @@ -611,15 +600,12 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf, uint32_t start_len = 2; // speed optimization - for (uint32_t rep_index = 0; rep_index < REP_DISTANCES; ++rep_index) { + for (uint32_t rep_index = 0; rep_index < REPS; ++rep_index) { const uint8_t *const buf_back = buf - reps[rep_index] - 1; if (not_equal_16(buf, buf_back)) continue; - uint32_t len_test; - for (len_test = 2; len_test < buf_avail - && buf[len_test] == buf_back[len_test]; - ++len_test) ; + uint32_t len_test = lzma_memcmplen(buf, buf_back, 2, buf_avail); while (len_end < cur + len_test) coder->opts[++len_end].price = RC_INFINITY_PRICE; @@ -728,14 +714,14 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf, for (uint32_t len_test = start_len; ; ++len_test) { const uint32_t cur_back = coder->matches[i].dist; uint32_t cur_and_len_price = normal_match_price - + get_pos_len_price(coder, + + get_dist_len_price(coder, cur_back, len_test, pos_state); if (cur_and_len_price < coder->opts[cur + len_test].price) { coder->opts[cur + len_test].price = cur_and_len_price; coder->opts[cur + len_test].pos_prev = cur; coder->opts[cur + len_test].back_prev - = cur_back + REP_DISTANCES; + = cur_back + REPS; coder->opts[cur + len_test].prev_1_is_literal = false; } @@ -795,7 +781,7 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf, coder->opts[offset].prev_2 = true; coder->opts[offset].pos_prev_2 = cur; coder->opts[offset].back_prev_2 - = cur_back + REP_DISTANCES; + = cur_back + REPS; } //} } @@ -831,9 +817,9 @@ lzma_lzma_optimum_normal(lzma_coder *restrict coder, lzma_mf *restrict mf, // In liblzma they were moved into this single place. if (mf->read_ahead == 0) { if (coder->match_price_count >= (1 << 7)) - fill_distances_prices(coder); + fill_dist_prices(coder); - if (coder->align_price_count >= ALIGN_TABLE_SIZE) + if (coder->align_price_count >= ALIGN_SIZE) fill_align_prices(coder); } @@ -845,7 +831,7 @@ lzma_lzma_optimum_normal(lzma_coder *restrict coder, lzma_mf *restrict mf, if (len_end == UINT32_MAX) return; - uint32_t reps[REP_DISTANCES]; + uint32_t reps[REPS]; memcpy(reps, coder->reps, sizeof(reps)); uint32_t cur; diff --git a/src/liblzma/lzma/lzma_encoder_private.h b/src/liblzma/lzma/lzma_encoder_private.h index 684745236c82..2f62d6cba0b7 100644 --- a/src/liblzma/lzma/lzma_encoder_private.h +++ b/src/liblzma/lzma/lzma_encoder_private.h @@ -64,7 +64,7 @@ typedef struct { uint32_t pos_prev; // pos_next; uint32_t back_prev; - uint32_t backs[REP_DISTANCES]; + uint32_t backs[REPS]; } lzma_optimal; @@ -77,7 +77,7 @@ struct lzma_coder_s { lzma_lzma_state state; /// The four most recent match distances - uint32_t reps[REP_DISTANCES]; + uint32_t reps[REPS]; /// Array of match candidates lzma_match matches[MATCH_LEN_MAX + 1]; @@ -112,9 +112,9 @@ struct lzma_coder_s { probability is_rep1[STATES]; probability is_rep2[STATES]; probability is_rep0_long[STATES][POS_STATES_MAX]; - probability pos_slot[LEN_TO_POS_STATES][POS_SLOTS]; - probability pos_special[FULL_DISTANCES - END_POS_MODEL_INDEX]; - probability pos_align[ALIGN_TABLE_SIZE]; + probability dist_slot[DIST_STATES][DIST_SLOTS]; + probability dist_special[FULL_DISTANCES - DIST_MODEL_END]; + probability dist_align[ALIGN_SIZE]; // These are the same as in lzma_decoder.c except that the encoders // include also price tables. @@ -122,12 +122,12 @@ struct lzma_coder_s { lzma_length_encoder rep_len_encoder; // Price tables - uint32_t pos_slot_prices[LEN_TO_POS_STATES][POS_SLOTS]; - uint32_t distances_prices[LEN_TO_POS_STATES][FULL_DISTANCES]; + uint32_t dist_slot_prices[DIST_STATES][DIST_SLOTS]; + uint32_t dist_prices[DIST_STATES][FULL_DISTANCES]; uint32_t dist_table_size; uint32_t match_price_count; - uint32_t align_prices[ALIGN_TABLE_SIZE]; + uint32_t align_prices[ALIGN_SIZE]; uint32_t align_price_count; // Optimal diff --git a/src/liblzma/rangecoder/range_decoder.h b/src/liblzma/rangecoder/range_decoder.h index fb96180fb399..e0b051fac2d2 100644 --- a/src/liblzma/rangecoder/range_decoder.h +++ b/src/liblzma/rangecoder/range_decoder.h @@ -25,20 +25,26 @@ typedef struct { /// Reads the first five bytes to initialize the range decoder. -static inline bool +static inline lzma_ret rc_read_init(lzma_range_decoder *rc, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size) { while (rc->init_bytes_left > 0) { if (*in_pos == in_size) - return false; + return LZMA_OK; + + // The first byte is always 0x00. It could have been omitted + // in LZMA2 but it wasn't, so one byte is wasted in every + // LZMA2 chunk. + if (rc->init_bytes_left == 5 && in[*in_pos] != 0x00) + return LZMA_DATA_ERROR; rc->code = (rc->code << 8) | in[*in_pos]; ++*in_pos; --rc->init_bytes_left; } - return true; + return LZMA_STREAM_END; } diff --git a/src/liblzma/simple/arm.c b/src/liblzma/simple/arm.c index a84702ac62f2..258d870fead3 100644 --- a/src/liblzma/simple/arm.c +++ b/src/liblzma/simple/arm.c @@ -45,7 +45,7 @@ arm_code(lzma_simple *simple lzma_attribute((__unused__)), static lzma_ret -arm_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +arm_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, bool is_encoder) { return lzma_simple_coder_init(next, allocator, filters, @@ -54,7 +54,8 @@ arm_coder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret -lzma_simple_arm_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_simple_arm_encoder_init(lzma_next_coder *next, + const lzma_allocator *allocator, const lzma_filter_info *filters) { return arm_coder_init(next, allocator, filters, true); @@ -62,7 +63,8 @@ lzma_simple_arm_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret -lzma_simple_arm_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_simple_arm_decoder_init(lzma_next_coder *next, + const lzma_allocator *allocator, const lzma_filter_info *filters) { return arm_coder_init(next, allocator, filters, false); diff --git a/src/liblzma/simple/armthumb.c b/src/liblzma/simple/armthumb.c index 4b49175fec4d..06c21e4067a5 100644 --- a/src/liblzma/simple/armthumb.c +++ b/src/liblzma/simple/armthumb.c @@ -50,7 +50,7 @@ armthumb_code(lzma_simple *simple lzma_attribute((__unused__)), static lzma_ret -armthumb_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +armthumb_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, bool is_encoder) { return lzma_simple_coder_init(next, allocator, filters, @@ -60,7 +60,8 @@ armthumb_coder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret lzma_simple_armthumb_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return armthumb_coder_init(next, allocator, filters, true); } @@ -68,7 +69,8 @@ lzma_simple_armthumb_encoder_init(lzma_next_coder *next, extern lzma_ret lzma_simple_armthumb_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return armthumb_coder_init(next, allocator, filters, false); } diff --git a/src/liblzma/simple/ia64.c b/src/liblzma/simple/ia64.c index ce3692b98f1d..ba7249c00178 100644 --- a/src/liblzma/simple/ia64.c +++ b/src/liblzma/simple/ia64.c @@ -86,7 +86,7 @@ ia64_code(lzma_simple *simple lzma_attribute((__unused__)), static lzma_ret -ia64_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +ia64_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, bool is_encoder) { return lzma_simple_coder_init(next, allocator, filters, @@ -96,7 +96,8 @@ ia64_coder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret lzma_simple_ia64_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return ia64_coder_init(next, allocator, filters, true); } @@ -104,7 +105,8 @@ lzma_simple_ia64_encoder_init(lzma_next_coder *next, extern lzma_ret lzma_simple_ia64_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return ia64_coder_init(next, allocator, filters, false); } diff --git a/src/liblzma/simple/powerpc.c b/src/liblzma/simple/powerpc.c index 6f8351176c5e..46899196af50 100644 --- a/src/liblzma/simple/powerpc.c +++ b/src/liblzma/simple/powerpc.c @@ -49,7 +49,7 @@ powerpc_code(lzma_simple *simple lzma_attribute((__unused__)), static lzma_ret -powerpc_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +powerpc_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, bool is_encoder) { return lzma_simple_coder_init(next, allocator, filters, @@ -59,7 +59,8 @@ powerpc_coder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret lzma_simple_powerpc_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return powerpc_coder_init(next, allocator, filters, true); } @@ -67,7 +68,8 @@ lzma_simple_powerpc_encoder_init(lzma_next_coder *next, extern lzma_ret lzma_simple_powerpc_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return powerpc_coder_init(next, allocator, filters, false); } diff --git a/src/liblzma/simple/simple_coder.c b/src/liblzma/simple/simple_coder.c index a02b039aa0bb..dba5417b4264 100644 --- a/src/liblzma/simple/simple_coder.c +++ b/src/liblzma/simple/simple_coder.c @@ -18,7 +18,7 @@ /// Copied or encodes/decodes more data to out[]. static lzma_ret -copy_or_code(lzma_coder *coder, lzma_allocator *allocator, +copy_or_code(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -66,7 +66,7 @@ call_filter(lzma_coder *coder, uint8_t *buffer, size_t size) static lzma_ret -simple_code(lzma_coder *coder, lzma_allocator *allocator, +simple_code(lzma_coder *coder, const lzma_allocator *allocator, const uint8_t *restrict in, size_t *restrict in_pos, size_t in_size, uint8_t *restrict out, size_t *restrict out_pos, size_t out_size, lzma_action action) @@ -198,7 +198,7 @@ simple_code(lzma_coder *coder, lzma_allocator *allocator, static void -simple_coder_end(lzma_coder *coder, lzma_allocator *allocator) +simple_coder_end(lzma_coder *coder, const lzma_allocator *allocator) { lzma_next_end(&coder->next, allocator); lzma_free(coder->simple, allocator); @@ -208,7 +208,7 @@ simple_coder_end(lzma_coder *coder, lzma_allocator *allocator) static lzma_ret -simple_coder_update(lzma_coder *coder, lzma_allocator *allocator, +simple_coder_update(lzma_coder *coder, const lzma_allocator *allocator, const lzma_filter *filters_null lzma_attribute((__unused__)), const lzma_filter *reversed_filters) { @@ -219,7 +219,7 @@ simple_coder_update(lzma_coder *coder, lzma_allocator *allocator, extern lzma_ret -lzma_simple_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_simple_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, size_t (*filter)(lzma_simple *simple, uint32_t now_pos, bool is_encoder, uint8_t *buffer, size_t size), diff --git a/src/liblzma/simple/simple_coder.h b/src/liblzma/simple/simple_coder.h index 0952fad33b34..19c2ee03affd 100644 --- a/src/liblzma/simple/simple_coder.h +++ b/src/liblzma/simple/simple_coder.h @@ -17,44 +17,56 @@ extern lzma_ret lzma_simple_x86_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_x86_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_powerpc_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_powerpc_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_ia64_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_ia64_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_arm_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_arm_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_armthumb_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_armthumb_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_sparc_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); extern lzma_ret lzma_simple_sparc_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters); + const lzma_allocator *allocator, + const lzma_filter_info *filters); #endif diff --git a/src/liblzma/simple/simple_decoder.c b/src/liblzma/simple/simple_decoder.c index 0beccd32a7d8..1d864f2bf781 100644 --- a/src/liblzma/simple/simple_decoder.c +++ b/src/liblzma/simple/simple_decoder.c @@ -14,7 +14,7 @@ extern lzma_ret -lzma_simple_props_decode(void **options, lzma_allocator *allocator, +lzma_simple_props_decode(void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size) { if (props_size == 0) diff --git a/src/liblzma/simple/simple_decoder.h b/src/liblzma/simple/simple_decoder.h index b8bf590f7681..bed8d37a9653 100644 --- a/src/liblzma/simple/simple_decoder.h +++ b/src/liblzma/simple/simple_decoder.h @@ -16,7 +16,7 @@ #include "simple_coder.h" extern lzma_ret lzma_simple_props_decode( - void **options, lzma_allocator *allocator, + void **options, const lzma_allocator *allocator, const uint8_t *props, size_t props_size); #endif diff --git a/src/liblzma/simple/simple_private.h b/src/liblzma/simple/simple_private.h index fcf9f7c1980f..bb20cb419f8c 100644 --- a/src/liblzma/simple/simple_private.h +++ b/src/liblzma/simple/simple_private.h @@ -66,7 +66,8 @@ struct lzma_coder_s { extern lzma_ret lzma_simple_coder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters, + const lzma_allocator *allocator, + const lzma_filter_info *filters, size_t (*filter)(lzma_simple *simple, uint32_t now_pos, bool is_encoder, uint8_t *buffer, size_t size), size_t simple_size, size_t unfiltered_max, diff --git a/src/liblzma/simple/sparc.c b/src/liblzma/simple/sparc.c index 8270d6ab19f1..53ee49d154f1 100644 --- a/src/liblzma/simple/sparc.c +++ b/src/liblzma/simple/sparc.c @@ -57,7 +57,7 @@ sparc_code(lzma_simple *simple lzma_attribute((__unused__)), static lzma_ret -sparc_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +sparc_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, bool is_encoder) { return lzma_simple_coder_init(next, allocator, filters, @@ -67,7 +67,8 @@ sparc_coder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret lzma_simple_sparc_encoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return sparc_coder_init(next, allocator, filters, true); } @@ -75,7 +76,8 @@ lzma_simple_sparc_encoder_init(lzma_next_coder *next, extern lzma_ret lzma_simple_sparc_decoder_init(lzma_next_coder *next, - lzma_allocator *allocator, const lzma_filter_info *filters) + const lzma_allocator *allocator, + const lzma_filter_info *filters) { return sparc_coder_init(next, allocator, filters, false); } diff --git a/src/liblzma/simple/x86.c b/src/liblzma/simple/x86.c index 5d1509bb2278..3b2b4f852067 100644 --- a/src/liblzma/simple/x86.c +++ b/src/liblzma/simple/x86.c @@ -123,7 +123,7 @@ x86_code(lzma_simple *simple, uint32_t now_pos, bool is_encoder, static lzma_ret -x86_coder_init(lzma_next_coder *next, lzma_allocator *allocator, +x86_coder_init(lzma_next_coder *next, const lzma_allocator *allocator, const lzma_filter_info *filters, bool is_encoder) { const lzma_ret ret = lzma_simple_coder_init(next, allocator, filters, @@ -139,7 +139,8 @@ x86_coder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret -lzma_simple_x86_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_simple_x86_encoder_init(lzma_next_coder *next, + const lzma_allocator *allocator, const lzma_filter_info *filters) { return x86_coder_init(next, allocator, filters, true); @@ -147,7 +148,8 @@ lzma_simple_x86_encoder_init(lzma_next_coder *next, lzma_allocator *allocator, extern lzma_ret -lzma_simple_x86_decoder_init(lzma_next_coder *next, lzma_allocator *allocator, +lzma_simple_x86_decoder_init(lzma_next_coder *next, + const lzma_allocator *allocator, const lzma_filter_info *filters) { return x86_coder_init(next, allocator, filters, false); diff --git a/src/liblzma/validate_map.sh b/src/liblzma/validate_map.sh new file mode 100755 index 000000000000..3aee46687c38 --- /dev/null +++ b/src/liblzma/validate_map.sh @@ -0,0 +1,68 @@ +#!/bin/sh + +############################################################################### +# +# Check liblzma.map for certain types of errors +# +# Author: Lasse Collin +# +# This file has been put into the public domain. +# You can do whatever you want with this file. +# +############################################################################### + +LC_ALL=C +export LC_ALL + +STATUS=0 + +cd "$(dirname "$0")" + +# Get the list of symbols that aren't defined in liblzma.map. +SYMS=$(sed -n 's/^extern LZMA_API([^)]*) \([a-z0-9_]*\)(.*$/\1;/p' \ + api/lzma/*.h \ + | sort \ + | grep -Fve "$(sed '/[{}:*]/d;/^$/d;s/^ //' liblzma.map)") + +# Check that there are no old alpha or beta versions listed. +VER=$(cd ../.. && sh build-aux/version.sh) +NAMES= +case $VER in + *alpha | *beta) + NAMES=$(sed -n 's/^.*XZ_\([^ ]*\)\(alpha\|beta\) .*$/\1\2/p' \ + liblzma.map | grep -Fv "$VER") + ;; +esac + +# Check for duplicate lines. It can catch missing dependencies. +DUPS=$(sort liblzma.map | sed '/^$/d;/^global:$/d' | uniq -d) + +# Print error messages if needed. +if test -n "$SYMS$NAMES$DUPS"; then + echo + echo 'validate_map.sh found problems from liblzma.map:' + echo + + if test -n "$SYMS"; then + echo 'liblzma.map lacks the following symbols:' + echo "$SYMS" + echo + fi + + if test -n "$NAMES"; then + echo 'Obsolete alpha or beta version names:' + echo "$NAMES" + echo + fi + + if test -n "$DUPS"; then + echo 'Duplicate lines:' + echo "$DUPS" + echo + fi + + STATUS=1 +fi + +# Exit status is 1 if problems were found, 0 otherwise. +exit "$STATUS" diff --git a/src/xz/args.c b/src/xz/args.c index 34761d45d67e..041c80073e6d 100644 --- a/src/xz/args.c +++ b/src/xz/args.c @@ -22,6 +22,7 @@ bool opt_stdout = false; bool opt_force = false; bool opt_keep_original = false; bool opt_robot = false; +bool opt_ignore_check = false; // We don't modify or free() this, but we need to assign it in some // non-const pointers. @@ -55,6 +56,67 @@ parse_memlimit(const char *name, const char *name_percentage, char *str, static void +parse_block_list(char *str) +{ + // It must be non-empty and not begin with a comma. + if (str[0] == '\0' || str[0] == ',') + message_fatal(_("%s: Invalid argument to --block-list"), str); + + // Count the number of comma-separated strings. + size_t count = 1; + for (size_t i = 0; str[i] != '\0'; ++i) + if (str[i] == ',') + ++count; + + // Prevent an unlikely integer overflow. + if (count > SIZE_MAX / sizeof(uint64_t) - 1) + message_fatal(_("%s: Too many arguments to --block-list"), + str); + + // Allocate memory to hold all the sizes specified. + // If --block-list was specified already, its value is forgotten. + free(opt_block_list); + opt_block_list = xmalloc((count + 1) * sizeof(uint64_t)); + + for (size_t i = 0; i < count; ++i) { + // Locate the next comma and replace it with \0. + char *p = strchr(str, ','); + if (p != NULL) + *p = '\0'; + + if (str[0] == '\0') { + // There is no string, that is, a comma follows + // another comma. Use the previous value. + // + // NOTE: We checked earler that the first char + // of the whole list cannot be a comma. + assert(i > 0); + opt_block_list[i] = opt_block_list[i - 1]; + } else { + opt_block_list[i] = str_to_uint64("block-list", str, + 0, UINT64_MAX); + + // Zero indicates no more new Blocks. + if (opt_block_list[i] == 0) { + if (i + 1 != count) + message_fatal(_("0 can only be used " + "as the last element " + "in --block-list")); + + opt_block_list[i] = UINT64_MAX; + } + } + + str = p + 1; + } + + // Terminate the array. + opt_block_list[count] = 0; + return; +} + + +static void parse_real(args_info *args, int argc, char **argv) { enum { @@ -68,14 +130,19 @@ parse_real(args_info *args, int argc, char **argv) OPT_LZMA1, OPT_LZMA2, + OPT_SINGLE_STREAM, OPT_NO_SPARSE, OPT_FILES, OPT_FILES0, + OPT_BLOCK_SIZE, + OPT_BLOCK_LIST, OPT_MEM_COMPRESS, OPT_MEM_DECOMPRESS, OPT_NO_ADJUST, OPT_INFO_MEMORY, OPT_ROBOT, + OPT_FLUSH_TIMEOUT, + OPT_IGNORE_CHECK, }; static const char short_opts[] @@ -94,6 +161,7 @@ parse_real(args_info *args, int argc, char **argv) { "force", no_argument, NULL, 'f' }, { "stdout", no_argument, NULL, 'c' }, { "to-stdout", no_argument, NULL, 'c' }, + { "single-stream", no_argument, NULL, OPT_SINGLE_STREAM }, { "no-sparse", no_argument, NULL, OPT_NO_SPARSE }, { "suffix", required_argument, NULL, 'S' }, // { "recursive", no_argument, NULL, 'r' }, // TODO @@ -103,12 +171,16 @@ parse_real(args_info *args, int argc, char **argv) // Basic compression settings { "format", required_argument, NULL, 'F' }, { "check", required_argument, NULL, 'C' }, + { "ignore-check", no_argument, NULL, OPT_IGNORE_CHECK }, + { "block-size", required_argument, NULL, OPT_BLOCK_SIZE }, + { "block-list", required_argument, NULL, OPT_BLOCK_LIST }, { "memlimit-compress", required_argument, NULL, OPT_MEM_COMPRESS }, { "memlimit-decompress", required_argument, NULL, OPT_MEM_DECOMPRESS }, { "memlimit", required_argument, NULL, 'M' }, { "memory", required_argument, NULL, 'M' }, // Old alias { "no-adjust", no_argument, NULL, OPT_NO_ADJUST }, { "threads", required_argument, NULL, 'T' }, + { "flush-timeout", required_argument, NULL, OPT_FLUSH_TIMEOUT }, { "extreme", no_argument, NULL, 'e' }, { "fast", no_argument, NULL, '0' }, @@ -175,8 +247,9 @@ parse_real(args_info *args, int argc, char **argv) break; case 'T': - hardware_threadlimit_set(str_to_uint64( - "threads", optarg, 0, UINT32_MAX)); + // The max is from src/liblzma/common/common.h. + hardware_threads_set(str_to_uint64("threads", + optarg, 0, 16384)); break; // --version @@ -368,6 +441,24 @@ parse_real(args_info *args, int argc, char **argv) break; } + case OPT_IGNORE_CHECK: + opt_ignore_check = true; + break; + + case OPT_BLOCK_SIZE: + opt_block_size = str_to_uint64("block-size", optarg, + 0, LZMA_VLI_MAX); + break; + + case OPT_BLOCK_LIST: { + parse_block_list(optarg); + break; + } + + case OPT_SINGLE_STREAM: + opt_single_stream = true; + break; + case OPT_NO_SPARSE: io_no_sparse(); break; @@ -401,6 +492,11 @@ parse_real(args_info *args, int argc, char **argv) opt_auto_adjust = false; break; + case OPT_FLUSH_TIMEOUT: + opt_flush_timeout = str_to_uint64("flush-timeout", + optarg, 0, UINT64_MAX); + break; + default: message_try_help(); tuklib_exit(E_ERROR, E_ERROR, false); @@ -576,3 +672,13 @@ args_parse(args_info *args, int argc, char **argv) return; } + + +#ifndef NDEBUG +extern void +args_free(void) +{ + free(opt_block_list); + return; +} +#endif diff --git a/src/xz/args.h b/src/xz/args.h index b23f4ef12c92..1defad12e0dd 100644 --- a/src/xz/args.h +++ b/src/xz/args.h @@ -36,7 +36,9 @@ extern bool opt_force; extern bool opt_keep_original; // extern bool opt_recursive; extern bool opt_robot; +extern bool opt_ignore_check; extern const char stdin_filename[]; extern void args_parse(args_info *args, int argc, char **argv); +extern void args_free(void); diff --git a/src/xz/coder.c b/src/xz/coder.c index 017e04127e95..a94bdb83266f 100644 --- a/src/xz/coder.c +++ b/src/xz/coder.c @@ -24,6 +24,9 @@ enum coder_init_ret { enum operation_mode opt_mode = MODE_COMPRESS; enum format_type opt_format = FORMAT_AUTO; bool opt_auto_adjust = true; +bool opt_single_stream = false; +uint64_t opt_block_size = 0; +uint64_t *opt_block_list = NULL; /// Stream used to communicate with liblzma @@ -48,6 +51,14 @@ static lzma_check check; /// This becomes false if the --check=CHECK option is used. static bool check_default = true; +#ifdef MYTHREAD_ENABLED +static lzma_mt mt_options = { + .flags = 0, + .timeout = 300, + .filters = filters, +}; +#endif + extern void coder_set_check(lzma_check new_check) @@ -125,6 +136,15 @@ memlimit_too_small(uint64_t memory_usage) extern void coder_set_compression_settings(void) { + // The default check type is CRC64, but fallback to CRC32 + // if CRC64 isn't supported by the copy of liblzma we are + // using. CRC32 is always supported. + if (check_default) { + check = LZMA_CHECK_CRC64; + if (!lzma_check_is_supported(check)) + check = LZMA_CHECK_CRC32; + } + // Options for LZMA1 or LZMA2 in case we are using a preset. static lzma_options_lzma opt_lzma; @@ -175,15 +195,53 @@ coder_set_compression_settings(void) // Print the selected filter chain. message_filters_show(V_DEBUG, filters); - // If using --format=raw, we can be decoding. The memusage function - // also validates the filter chain and the options used for the - // filters. + // The --flush-timeout option requires LZMA_SYNC_FLUSH support + // from the filter chain. Currently threaded encoder doesn't support + // LZMA_SYNC_FLUSH so single-threaded mode must be used. + if (opt_mode == MODE_COMPRESS && opt_flush_timeout != 0) { + for (size_t i = 0; i < filters_count; ++i) { + switch (filters[i].id) { + case LZMA_FILTER_LZMA2: + case LZMA_FILTER_DELTA: + break; + + default: + message_fatal(_("The filter chain is " + "incompatible with --flush-timeout")); + } + } + + if (hardware_threads_get() > 1) { + message(V_WARNING, _("Switching to single-threaded " + "mode due to --flush-timeout")); + hardware_threads_set(1); + } + } + + // Get the memory usage. Note that if --format=raw was used, + // we can be decompressing. const uint64_t memory_limit = hardware_memlimit_get(opt_mode); uint64_t memory_usage; - if (opt_mode == MODE_COMPRESS) - memory_usage = lzma_raw_encoder_memusage(filters); - else + if (opt_mode == MODE_COMPRESS) { +#ifdef MYTHREAD_ENABLED + if (opt_format == FORMAT_XZ && hardware_threads_get() > 1) { + mt_options.threads = hardware_threads_get(); + mt_options.block_size = opt_block_size; + mt_options.check = check; + memory_usage = lzma_stream_encoder_mt_memusage( + &mt_options); + if (memory_usage != UINT64_MAX) + message(V_DEBUG, _("Using up to %" PRIu32 + " threads."), + mt_options.threads); + } else +#endif + { + memory_usage = lzma_raw_encoder_memusage(filters); + } + } else { memory_usage = lzma_raw_decoder_memusage(filters); + } if (memory_usage == UINT64_MAX) message_fatal(_("Unsupported filter chain or filter options")); @@ -199,90 +257,99 @@ coder_set_compression_settings(void) round_up_to_mib(decmem), 0)); } - if (memory_usage > memory_limit) { - // If --no-adjust was used or we didn't find LZMA1 or - // LZMA2 as the last filter, give an error immediately. - // --format=raw implies --no-adjust. - if (!opt_auto_adjust || opt_format == FORMAT_RAW) - memlimit_too_small(memory_usage); - - assert(opt_mode == MODE_COMPRESS); - - // Look for the last filter if it is LZMA2 or LZMA1, so - // we can make it use less RAM. With other filters we don't - // know what to do. - size_t i = 0; - while (filters[i].id != LZMA_FILTER_LZMA2 - && filters[i].id != LZMA_FILTER_LZMA1) { - if (filters[i].id == LZMA_VLI_UNKNOWN) - memlimit_too_small(memory_usage); - - ++i; - } + if (memory_usage <= memory_limit) + return; - // Decrease the dictionary size until we meet the memory - // usage limit. First round down to full mebibytes. - lzma_options_lzma *opt = filters[i].options; - const uint32_t orig_dict_size = opt->dict_size; - opt->dict_size &= ~((UINT32_C(1) << 20) - 1); - while (true) { - // If it is below 1 MiB, auto-adjusting failed. We - // could be more sophisticated and scale it down even - // more, but let's see if many complain about this - // version. - // - // FIXME: Displays the scaled memory usage instead - // of the original. - if (opt->dict_size < (UINT32_C(1) << 20)) + // If --no-adjust was used or we didn't find LZMA1 or + // LZMA2 as the last filter, give an error immediately. + // --format=raw implies --no-adjust. + if (!opt_auto_adjust || opt_format == FORMAT_RAW) + memlimit_too_small(memory_usage); + + assert(opt_mode == MODE_COMPRESS); + +#ifdef MYTHREAD_ENABLED + if (opt_format == FORMAT_XZ && mt_options.threads > 1) { + // Try to reduce the number of threads before + // adjusting the compression settings down. + do { + // FIXME? The real single-threaded mode has + // lower memory usage, but it's not comparable + // because it doesn't write the size info + // into Block Headers. + if (--mt_options.threads == 0) memlimit_too_small(memory_usage); - memory_usage = lzma_raw_encoder_memusage(filters); + memory_usage = lzma_stream_encoder_mt_memusage( + &mt_options); if (memory_usage == UINT64_MAX) message_bug(); - // Accept it if it is low enough. - if (memory_usage <= memory_limit) - break; + } while (memory_usage > memory_limit); - // Otherwise 1 MiB down and try again. I hope this - // isn't too slow method for cases where the original - // dict_size is very big. - opt->dict_size -= UINT32_C(1) << 20; - } + message(V_WARNING, _("Adjusted the number of threads " + "from %s to %s to not exceed " + "the memory usage limit of %s MiB"), + uint64_to_str(hardware_threads_get(), 0), + uint64_to_str(mt_options.threads, 1), + uint64_to_str(round_up_to_mib( + memory_limit), 2)); + } +#endif + + if (memory_usage <= memory_limit) + return; - // Tell the user that we decreased the dictionary size. - message(V_WARNING, _("Adjusted LZMA%c dictionary size " - "from %s MiB to %s MiB to not exceed " - "the memory usage limit of %s MiB"), - filters[i].id == LZMA_FILTER_LZMA2 - ? '2' : '1', - uint64_to_str(orig_dict_size >> 20, 0), - uint64_to_str(opt->dict_size >> 20, 1), - uint64_to_str(round_up_to_mib( - memory_limit), 2)); + // Look for the last filter if it is LZMA2 or LZMA1, so we can make + // it use less RAM. With other filters we don't know what to do. + size_t i = 0; + while (filters[i].id != LZMA_FILTER_LZMA2 + && filters[i].id != LZMA_FILTER_LZMA1) { + if (filters[i].id == LZMA_VLI_UNKNOWN) + memlimit_too_small(memory_usage); + + ++i; } -/* - // Limit the number of worker threads so that memory usage - // limit isn't exceeded. - assert(memory_usage > 0); - size_t thread_limit = memory_limit / memory_usage; - if (thread_limit == 0) - thread_limit = 1; + // Decrease the dictionary size until we meet the memory + // usage limit. First round down to full mebibytes. + lzma_options_lzma *opt = filters[i].options; + const uint32_t orig_dict_size = opt->dict_size; + opt->dict_size &= ~((UINT32_C(1) << 20) - 1); + while (true) { + // If it is below 1 MiB, auto-adjusting failed. We could be + // more sophisticated and scale it down even more, but let's + // see if many complain about this version. + // + // FIXME: Displays the scaled memory usage instead + // of the original. + if (opt->dict_size < (UINT32_C(1) << 20)) + memlimit_too_small(memory_usage); - if (opt_threads > thread_limit) - opt_threads = thread_limit; -*/ + memory_usage = lzma_raw_encoder_memusage(filters); + if (memory_usage == UINT64_MAX) + message_bug(); - if (check_default) { - // The default check type is CRC64, but fallback to CRC32 - // if CRC64 isn't supported by the copy of liblzma we are - // using. CRC32 is always supported. - check = LZMA_CHECK_CRC64; - if (!lzma_check_is_supported(check)) - check = LZMA_CHECK_CRC32; + // Accept it if it is low enough. + if (memory_usage <= memory_limit) + break; + + // Otherwise 1 MiB down and try again. I hope this + // isn't too slow method for cases where the original + // dict_size is very big. + opt->dict_size -= UINT32_C(1) << 20; } + // Tell the user that we decreased the dictionary size. + message(V_WARNING, _("Adjusted LZMA%c dictionary size " + "from %s MiB to %s MiB to not exceed " + "the memory usage limit of %s MiB"), + filters[i].id == LZMA_FILTER_LZMA2 + ? '2' : '1', + uint64_to_str(orig_dict_size >> 20, 0), + uint64_to_str(opt->dict_size >> 20, 1), + uint64_to_str(round_up_to_mib(memory_limit), 2)); + return; } @@ -364,7 +431,14 @@ coder_init(file_pair *pair) break; case FORMAT_XZ: - ret = lzma_stream_encoder(&strm, filters, check); +#ifdef MYTHREAD_ENABLED + if (hardware_threads_get() > 1) + ret = lzma_stream_encoder_mt( + &strm, &mt_options); + else +#endif + ret = lzma_stream_encoder( + &strm, filters, check); break; case FORMAT_LZMA: @@ -376,8 +450,17 @@ coder_init(file_pair *pair) break; } } else { - const uint32_t flags = LZMA_TELL_UNSUPPORTED_CHECK - | LZMA_CONCATENATED; + uint32_t flags = 0; + + // It seems silly to warn about unsupported check if the + // check won't be verified anyway due to --ignore-check. + if (opt_ignore_check) + flags |= LZMA_IGNORE_CHECK; + else + flags |= LZMA_TELL_UNSUPPORTED_CHECK; + + if (!opt_single_stream) + flags |= LZMA_CONCATENATED; // We abuse FORMAT_AUTO to indicate unknown file format, // for which we may consider passthru mode. @@ -408,7 +491,7 @@ coder_init(file_pair *pair) switch (init_format) { case FORMAT_AUTO: - // Uknown file format. If --decompress --stdout + // Unknown file format. If --decompress --stdout // --force have been given, then we copy the input // as is to stdout. Checking for MODE_DECOMPRESS // is needed, because we don't want to do use @@ -462,6 +545,56 @@ coder_init(file_pair *pair) } +/// Resolve conflicts between opt_block_size and opt_block_list in single +/// threaded mode. We want to default to opt_block_list, except when it is +/// larger than opt_block_size. If this is the case for the current Block +/// at *list_pos, then we break into smaller Blocks. Otherwise advance +/// to the next Block in opt_block_list, and break apart if needed. +static void +split_block(uint64_t *block_remaining, + uint64_t *next_block_remaining, + size_t *list_pos) +{ + if (*next_block_remaining > 0) { + // The Block at *list_pos has previously been split up. + assert(hardware_threads_get() == 1); + assert(opt_block_size > 0); + assert(opt_block_list != NULL); + + if (*next_block_remaining > opt_block_size) { + // We have to split the current Block at *list_pos + // into another opt_block_size length Block. + *block_remaining = opt_block_size; + } else { + // This is the last remaining split Block for the + // Block at *list_pos. + *block_remaining = *next_block_remaining; + } + + *next_block_remaining -= *block_remaining; + + } else { + // The Block at *list_pos has been finished. Go to the next + // entry in the list. If the end of the list has been reached, + // reuse the size of the last Block. + if (opt_block_list[*list_pos + 1] != 0) + ++*list_pos; + + *block_remaining = opt_block_list[*list_pos]; + + // If in single-threaded mode, split up the Block if needed. + // This is not needed in multi-threaded mode because liblzma + // will do this due to how threaded encoding works. + if (hardware_threads_get() == 1 && opt_block_size > 0 + && *block_remaining > opt_block_size) { + *next_block_remaining + = *block_remaining - opt_block_size; + *block_remaining = opt_block_size; + } + } +} + + /// Compress or decompress using liblzma. static bool coder_normal(file_pair *pair) @@ -469,8 +602,8 @@ coder_normal(file_pair *pair) // Encoder needs to know when we have given all the input to it. // The decoders need to know it too when we are using // LZMA_CONCATENATED. We need to check for src_eof here, because - // the first input chunk has been already read, and that may - // have been the only chunk we will read. + // the first input chunk has been already read if decompressing, + // and that may have been the only chunk we will read. lzma_action action = pair->src_eof ? LZMA_FINISH : LZMA_RUN; lzma_ret ret; @@ -478,22 +611,77 @@ coder_normal(file_pair *pair) // Assume that something goes wrong. bool success = false; + // block_remaining indicates how many input bytes to encode before + // finishing the current .xz Block. The Block size is set with + // --block-size=SIZE and --block-list. They have an effect only when + // compressing to the .xz format. If block_remaining == UINT64_MAX, + // only a single block is created. + uint64_t block_remaining = UINT64_MAX; + + // next_block_remining for when we are in single-threaded mode and + // the Block in --block-list is larger than the --block-size=SIZE. + uint64_t next_block_remaining = 0; + + // Position in opt_block_list. Unused if --block-list wasn't used. + size_t list_pos = 0; + + // Handle --block-size for single-threaded mode and the first step + // of --block-list. + if (opt_mode == MODE_COMPRESS && opt_format == FORMAT_XZ) { + // --block-size doesn't do anything here in threaded mode, + // because the threaded encoder will take care of splitting + // to fixed-sized Blocks. + if (hardware_threads_get() == 1 && opt_block_size > 0) + block_remaining = opt_block_size; + + // If --block-list was used, start with the first size. + // + // For threaded case, --block-size specifies how big Blocks + // the encoder needs to be prepared to create at maximum + // and --block-list will simultaneously cause new Blocks + // to be started at specified intervals. To keep things + // logical, the same is done in single-threaded mode. The + // output is still not identical because in single-threaded + // mode the size info isn't written into Block Headers. + if (opt_block_list != NULL) { + if (block_remaining < opt_block_list[list_pos]) { + assert(hardware_threads_get() == 1); + next_block_remaining = opt_block_list[list_pos] + - block_remaining; + } else { + block_remaining = opt_block_list[list_pos]; + } + } + } + strm.next_out = out_buf.u8; strm.avail_out = IO_BUFFER_SIZE; while (!user_abort) { - // Fill the input buffer if it is empty and we haven't reached - // end of file yet. - if (strm.avail_in == 0 && !pair->src_eof) { + // Fill the input buffer if it is empty and we aren't + // flushing or finishing. + if (strm.avail_in == 0 && action == LZMA_RUN) { strm.next_in = in_buf.u8; - strm.avail_in = io_read( - pair, &in_buf, IO_BUFFER_SIZE); + strm.avail_in = io_read(pair, &in_buf, + my_min(block_remaining, + IO_BUFFER_SIZE)); if (strm.avail_in == SIZE_MAX) break; - if (pair->src_eof) + if (pair->src_eof) { action = LZMA_FINISH; + + } else if (block_remaining != UINT64_MAX) { + // Start a new Block after every + // opt_block_size bytes of input. + block_remaining -= strm.avail_in; + if (block_remaining == 0) + action = LZMA_FULL_BARRIER; + } + + if (action == LZMA_RUN && flush_needed) + action = LZMA_SYNC_FLUSH; } // Let liblzma do the actual work. @@ -509,7 +697,39 @@ coder_normal(file_pair *pair) strm.avail_out = IO_BUFFER_SIZE; } - if (ret != LZMA_OK) { + if (ret == LZMA_STREAM_END && (action == LZMA_SYNC_FLUSH + || action == LZMA_FULL_BARRIER)) { + if (action == LZMA_SYNC_FLUSH) { + // Flushing completed. Write the pending data + // out immediatelly so that the reading side + // can decompress everything compressed so far. + if (io_write(pair, &out_buf, IO_BUFFER_SIZE + - strm.avail_out)) + break; + + strm.next_out = out_buf.u8; + strm.avail_out = IO_BUFFER_SIZE; + + // Set the time of the most recent flushing. + mytime_set_flush_time(); + } else { + // Start a new Block after LZMA_FULL_BARRIER. + if (opt_block_list == NULL) { + assert(hardware_threads_get() == 1); + assert(opt_block_size > 0); + block_remaining = opt_block_size; + } else { + split_block(&block_remaining, + &next_block_remaining, + &list_pos); + } + } + + // Start a new Block after LZMA_FULL_FLUSH or continue + // the same block after LZMA_SYNC_FLUSH. + action = LZMA_RUN; + + } else if (ret != LZMA_OK) { // Determine if the return value indicates that we // won't continue coding. const bool stop = ret != LZMA_NO_CHECK @@ -528,6 +748,12 @@ coder_normal(file_pair *pair) } if (ret == LZMA_STREAM_END) { + if (opt_single_stream) { + io_fix_src_pos(pair, strm.avail_in); + success = true; + break; + } + // Check that there is no trailing garbage. // This is needed for LZMA_Alone and raw // streams. @@ -630,10 +856,15 @@ coder_run(const char *filename) // Assume that something goes wrong. bool success = false; - // Read the first chunk of input data. This is needed to detect - // the input file type (for now, only for decompression). - strm.next_in = in_buf.u8; - strm.avail_in = io_read(pair, &in_buf, IO_BUFFER_SIZE); + if (opt_mode == MODE_COMPRESS) { + strm.next_in = NULL; + strm.avail_in = 0; + } else { + // Read the first chunk of input data. This is needed + // to detect the input file type. + strm.next_in = in_buf.u8; + strm.avail_in = io_read(pair, &in_buf, IO_BUFFER_SIZE); + } if (strm.avail_in != SIZE_MAX) { // Initialize the coder. This will detect the file format @@ -648,6 +879,11 @@ coder_run(const char *filename) // Don't open the destination file when --test // is used. if (opt_mode == MODE_TEST || !io_open_dest(pair)) { + // Remember the current time. It is needed + // for progress indicator and for timed + // flushing. + mytime_set_start_time(); + // Initialize the progress indicator. const uint64_t in_size = pair->src_st.st_size <= 0 @@ -671,3 +907,13 @@ coder_run(const char *filename) return; } + + +#ifndef NDEBUG +extern void +coder_free(void) +{ + lzma_end(&strm); + return; +} +#endif diff --git a/src/xz/coder.h b/src/xz/coder.h index 2d3add972745..583da8f68d50 100644 --- a/src/xz/coder.h +++ b/src/xz/coder.h @@ -41,6 +41,16 @@ extern enum format_type opt_format; /// they exceed the memory usage limit. extern bool opt_auto_adjust; +/// If true, stop after decoding the first stream. +extern bool opt_single_stream; + +/// If non-zero, start a new .xz Block after every opt_block_size bytes +/// of input. This has an effect only when compressing to the .xz format. +extern uint64_t opt_block_size; + +/// This is non-NULL if --block-list was used. This contains the Block sizes +/// as an array that is terminated with 0. +extern uint64_t *opt_block_list; /// Set the integrity check type used when compressing extern void coder_set_check(lzma_check check); @@ -59,3 +69,8 @@ extern void coder_set_compression_settings(void); /// Compress or decompress the given file extern void coder_run(const char *filename); + +#ifndef NDEBUG +/// Free the memory allocated for the coder and kill the worker threads. +extern void coder_free(void); +#endif diff --git a/src/xz/file_io.c b/src/xz/file_io.c index 871a099b6193..f135cf7cb6bd 100644 --- a/src/xz/file_io.c +++ b/src/xz/file_io.c @@ -17,6 +17,7 @@ #ifdef TUKLIB_DOSLIKE # include <io.h> #else +# include <poll.h> static bool warn_fchown; #endif @@ -37,14 +38,30 @@ static bool warn_fchown; #endif +typedef enum { + IO_WAIT_MORE, // Reading or writing is possible. + IO_WAIT_ERROR, // Error or user_abort + IO_WAIT_TIMEOUT, // poll() timed out +} io_wait_ret; + + /// If true, try to create sparse files when decompressing. static bool try_sparse = true; #ifndef TUKLIB_DOSLIKE +/// File status flags of standard input. This is used by io_open_src() +/// and io_close_src(). +static int stdin_flags; +static bool restore_stdin_flags = false; + /// Original file status flags of standard output. This is used by /// io_open_dest() and io_close_dest() to save and restore the flags. static int stdout_flags; static bool restore_stdout_flags = false; + +/// Self-pipe used together with the user_abort variable to avoid +/// race conditions with signal handling. +static int user_abort_pipe[2]; #endif @@ -64,19 +81,43 @@ io_init(void) // If fchown() fails setting the owner, we warn about it only if // we are root. warn_fchown = geteuid() == 0; + + if (pipe(user_abort_pipe) + || fcntl(user_abort_pipe[0], F_SETFL, O_NONBLOCK) + == -1 + || fcntl(user_abort_pipe[1], F_SETFL, O_NONBLOCK) + == -1) + message_fatal(_("Error creating a pipe: %s"), + strerror(errno)); #endif #ifdef __DJGPP__ // Avoid doing useless things when statting files. // This isn't important but doesn't hurt. - _djstat_flags = _STAT_INODE | _STAT_EXEC_EXT - | _STAT_EXEC_MAGIC | _STAT_DIRSIZE; + _djstat_flags = _STAT_EXEC_EXT | _STAT_EXEC_MAGIC | _STAT_DIRSIZE; #endif return; } +#ifndef TUKLIB_DOSLIKE +extern void +io_write_to_user_abort_pipe(void) +{ + // If the write() fails, it's probably due to the pipe being full. + // Failing in that case is fine. If the reason is something else, + // there's not much we can do since this is called in a signal + // handler. So ignore the errors and try to avoid warnings with + // GCC and glibc when _FORTIFY_SOURCE=2 is used. + uint8_t b = '\0'; + const int ret = write(user_abort_pipe[1], &b, 1); + (void)ret; + return; +} +#endif + + extern void io_no_sparse(void) { @@ -85,6 +126,63 @@ io_no_sparse(void) } +#ifndef TUKLIB_DOSLIKE +/// \brief Waits for input or output to become available or for a signal +/// +/// This uses the self-pipe trick to avoid a race condition that can occur +/// if a signal is caught after user_abort has been checked but before e.g. +/// read() has been called. In that situation read() could block unless +/// non-blocking I/O is used. With non-blocking I/O something like select() +/// or poll() is needed to avoid a busy-wait loop, and the same race condition +/// pops up again. There are pselect() (POSIX-1.2001) and ppoll() (not in +/// POSIX) but neither is portable enough in 2013. The self-pipe trick is +/// old and very portable. +static io_wait_ret +io_wait(file_pair *pair, int timeout, bool is_reading) +{ + struct pollfd pfd[2]; + + if (is_reading) { + pfd[0].fd = pair->src_fd; + pfd[0].events = POLLIN; + } else { + pfd[0].fd = pair->dest_fd; + pfd[0].events = POLLOUT; + } + + pfd[1].fd = user_abort_pipe[0]; + pfd[1].events = POLLIN; + + while (true) { + const int ret = poll(pfd, 2, timeout); + + if (user_abort) + return IO_WAIT_ERROR; + + if (ret == -1) { + if (errno == EINTR || errno == EAGAIN) + continue; + + message_error(_("%s: poll() failed: %s"), + is_reading ? pair->src_name + : pair->dest_name, + strerror(errno)); + return IO_WAIT_ERROR; + } + + if (ret == 0) { + assert(opt_flush_timeout != 0); + flush_needed = true; + return IO_WAIT_TIMEOUT; + } + + if (pfd[0].revents != 0) + return IO_WAIT_MORE; + } +} +#endif + + /// \brief Unlink a file /// /// This tries to verify that the file being unlinked really is the file that @@ -294,6 +392,31 @@ io_open_src_real(file_pair *pair) pair->src_fd = STDIN_FILENO; #ifdef TUKLIB_DOSLIKE setmode(STDIN_FILENO, O_BINARY); +#else + // Enable O_NONBLOCK for stdin. + stdin_flags = fcntl(STDIN_FILENO, F_GETFL); + if (stdin_flags == -1) { + message_error(_("Error getting the file status flags " + "from standard input: %s"), + strerror(errno)); + return true; + } + + if ((stdin_flags & O_NONBLOCK) == 0) { + if (fcntl(STDIN_FILENO, F_SETFL, + stdin_flags | O_NONBLOCK) == -1) { + message_error(_("Error setting O_NONBLOCK " + "on standard input: %s"), + strerror(errno)); + return true; + } + + restore_stdin_flags = true; + } +#endif +#ifdef HAVE_POSIX_FADVISE + // It will fail if stdin is a pipe and that's fine. + (void)posix_fadvise(STDIN_FILENO, 0, 0, POSIX_FADV_SEQUENTIAL); #endif return false; } @@ -311,13 +434,12 @@ io_open_src_real(file_pair *pair) int flags = O_RDONLY | O_BINARY | O_NOCTTY; #ifndef TUKLIB_DOSLIKE - // If we accept only regular files, we need to be careful to avoid - // problems with special files like devices and FIFOs. O_NONBLOCK - // prevents blocking when opening such files. When we want to accept - // special files, we must not use O_NONBLOCK, or otherwise we won't - // block waiting e.g. FIFOs to become readable. - if (reg_files_only) - flags |= O_NONBLOCK; + // Use non-blocking I/O: + // - It prevents blocking when opening FIFOs and some other + // special files, which is good if we want to accept only + // regular files. + // - It can help avoiding some race conditions with signal handling. + flags |= O_NONBLOCK; #endif #if defined(O_NOFOLLOW) @@ -345,30 +467,13 @@ io_open_src_real(file_pair *pair) (void)follow_symlinks; #endif - // Try to open the file. If we are accepting non-regular files, - // unblock the caught signals so that open() can be interrupted - // if it blocks e.g. due to a FIFO file. - if (!reg_files_only) - signals_unblock(); - - // Maybe this wouldn't need a loop, since all the signal handlers for - // which we don't use SA_RESTART set user_abort to true. But it - // doesn't hurt to have it just in case. - do { - pair->src_fd = open(pair->src_name, flags); - } while (pair->src_fd == -1 && errno == EINTR && !user_abort); - - if (!reg_files_only) - signals_block(); + // Try to open the file. Signals have been blocked so EINTR shouldn't + // be possible. + pair->src_fd = open(pair->src_name, flags); if (pair->src_fd == -1) { - // If we were interrupted, don't display any error message. - if (errno == EINTR) { - // All the signals that don't have SA_RESTART - // set user_abort. - assert(user_abort); - return true; - } + // Signals (that have a signal handler) have been blocked. + assert(errno != EINTR); #ifdef O_NOFOLLOW // Give an understandable error message if the reason @@ -427,26 +532,20 @@ io_open_src_real(file_pair *pair) return true; } -#ifndef TUKLIB_DOSLIKE - // Drop O_NONBLOCK, which is used only when we are accepting only - // regular files. After the open() call, we want things to block - // instead of giving EAGAIN. - if (reg_files_only) { - flags = fcntl(pair->src_fd, F_GETFL); - if (flags == -1) - goto error_msg; - - flags &= ~O_NONBLOCK; - - if (fcntl(pair->src_fd, F_SETFL, flags) == -1) - goto error_msg; - } -#endif - // Stat the source file. We need the result also when we copy // the permissions, and when unlinking. + // + // NOTE: Use stat() instead of fstat() with DJGPP, because + // then we have a better chance to get st_ino value that can + // be used in io_open_dest_real() to prevent overwriting the + // source file. +#ifdef __DJGPP__ + if (stat(pair->src_name, &pair->src_st)) + goto error_msg; +#else if (fstat(pair->src_fd, &pair->src_st)) goto error_msg; +#endif if (S_ISDIR(pair->src_st.st_mode)) { message_warning(_("%s: Is a directory, skipping"), @@ -492,6 +591,23 @@ io_open_src_real(file_pair *pair) goto error; } } + + // If it is something else than a regular file, wait until + // there is input available. This way reading from FIFOs + // will work when open() is used with O_NONBLOCK. + if (!S_ISREG(pair->src_st.st_mode)) { + signals_unblock(); + const io_wait_ret ret = io_wait(pair, -1, true); + signals_block(); + + if (ret != IO_WAIT_MORE) + goto error; + } +#endif + +#ifdef HAVE_POSIX_FADVISE + // It will fail with some special files like FIFOs but that is fine. + (void)posix_fadvise(pair->src_fd, 0, 0, POSIX_FADV_SEQUENTIAL); #endif return false; @@ -542,6 +658,19 @@ io_open_src(const char *src_name) static void io_close_src(file_pair *pair, bool success) { +#ifndef TUKLIB_DOSLIKE + if (restore_stdin_flags) { + assert(pair->src_fd == STDIN_FILENO); + + restore_stdin_flags = false; + + if (fcntl(STDIN_FILENO, F_SETFL, stdin_flags) == -1) + message_error(_("Error restoring the status flags " + "to standard input: %s"), + strerror(errno)); + } +#endif + if (pair->src_fd != STDIN_FILENO && pair->src_fd != -1) { #ifdef TUKLIB_DOSLIKE (void)close(pair->src_fd); @@ -575,12 +704,58 @@ io_open_dest_real(file_pair *pair) pair->dest_fd = STDOUT_FILENO; #ifdef TUKLIB_DOSLIKE setmode(STDOUT_FILENO, O_BINARY); +#else + // Set O_NONBLOCK if it isn't already set. + // + // NOTE: O_APPEND may be unset later in this function + // and it relies on stdout_flags being set here. + stdout_flags = fcntl(STDOUT_FILENO, F_GETFL); + if (stdout_flags == -1) { + message_error(_("Error getting the file status flags " + "from standard output: %s"), + strerror(errno)); + return true; + } + + if ((stdout_flags & O_NONBLOCK) == 0) { + if (fcntl(STDOUT_FILENO, F_SETFL, + stdout_flags | O_NONBLOCK) == -1) { + message_error(_("Error setting O_NONBLOCK " + "on standard output: %s"), + strerror(errno)); + return true; + } + + restore_stdout_flags = true; + } #endif } else { pair->dest_name = suffix_get_dest_name(pair->src_name); if (pair->dest_name == NULL) return true; +#ifdef __DJGPP__ + struct stat st; + if (stat(pair->dest_name, &st) == 0) { + // Check that it isn't a special file like "prn". + if (st.st_dev == -1) { + message_error("%s: Refusing to write to " + "a DOS special file", + pair->dest_name); + return true; + } + + // Check that we aren't overwriting the source file. + if (st.st_dev == pair->src_st.st_dev + && st.st_ino == pair->src_st.st_ino) { + message_error("%s: Output file is the same " + "as the input file", + pair->dest_name); + return true; + } + } +#endif + // If --force was used, unlink the target file first. if (opt_force && unlink(pair->dest_name) && errno != ENOENT) { message_error(_("%s: Cannot remove: %s"), @@ -590,8 +765,11 @@ io_open_dest_real(file_pair *pair) } // Open the file. - const int flags = O_WRONLY | O_BINARY | O_NOCTTY + int flags = O_WRONLY | O_BINARY | O_NOCTTY | O_CREAT | O_EXCL; +#ifndef TUKLIB_DOSLIKE + flags |= O_NONBLOCK; +#endif const mode_t mode = S_IRUSR | S_IWUSR; pair->dest_fd = open(pair->dest_name, flags, mode); @@ -603,17 +781,19 @@ io_open_dest_real(file_pair *pair) } } - // If this really fails... well, we have a safe fallback. +#ifndef TUKLIB_DOSLIKE + // dest_st isn't used on DOS-like systems except as a dummy + // argument to io_unlink(), so don't fstat() on such systems. if (fstat(pair->dest_fd, &pair->dest_st)) { -#if defined(__VMS) + // If fstat() really fails, we have a safe fallback here. +# if defined(__VMS) pair->dest_st.st_ino[0] = 0; pair->dest_st.st_ino[1] = 0; pair->dest_st.st_ino[2] = 0; -#elif !defined(TUKLIB_DOSLIKE) +# else pair->dest_st.st_dev = 0; pair->dest_st.st_ino = 0; -#endif -#ifndef TUKLIB_DOSLIKE +# endif } else if (try_sparse && opt_mode == MODE_DECOMPRESS) { // When writing to standard output, we need to be extra // careful: @@ -631,10 +811,6 @@ io_open_dest_real(file_pair *pair) if (!S_ISREG(pair->dest_st.st_mode)) return false; - stdout_flags = fcntl(STDOUT_FILENO, F_GETFL); - if (stdout_flags == -1) - return false; - if (stdout_flags & O_APPEND) { // Creating a sparse file is not possible // when O_APPEND is active (it's used by @@ -653,14 +829,23 @@ io_open_dest_real(file_pair *pair) if (lseek(STDOUT_FILENO, 0, SEEK_END) == -1) return false; + // O_NONBLOCK was set earlier in this function + // so it must be kept here too. If this + // fcntl() call fails, we continue but won't + // try to create sparse output. The original + // flags will still be restored if needed (to + // unset O_NONBLOCK) when the file is finished. if (fcntl(STDOUT_FILENO, F_SETFL, - stdout_flags & ~O_APPEND) - == -1) + (stdout_flags | O_NONBLOCK) + & ~O_APPEND) == -1) return false; // Disabling O_APPEND succeeded. Mark // that the flags should be restored - // in io_close_dest(). + // in io_close_dest(). This quite likely was + // already set when enabling O_NONBLOCK but + // just in case O_NONBLOCK was already set, + // set this again here. restore_stdout_flags = true; } else if (lseek(STDOUT_FILENO, 0, SEEK_CUR) @@ -673,8 +858,8 @@ io_open_dest_real(file_pair *pair) } pair->dest_try_sparse = true; -#endif } +#endif return false; } @@ -790,6 +975,21 @@ io_close(file_pair *pair, bool success) } +extern void +io_fix_src_pos(file_pair *pair, size_t rewind_size) +{ + assert(rewind_size <= IO_BUFFER_SIZE); + + if (rewind_size > 0) { + // This doesn't need to work on unseekable file descriptors, + // so just ignore possible errors. + (void)lseek(pair->src_fd, -(off_t)(rewind_size), SEEK_CUR); + } + + return; +} + + extern size_t io_read(file_pair *pair, io_buf *buf_union, size_t size) { @@ -815,12 +1015,30 @@ io_read(file_pair *pair, io_buf *buf_union, size_t size) continue; } +#ifndef TUKLIB_DOSLIKE + if (errno == EAGAIN || errno == EWOULDBLOCK) { + const io_wait_ret ret = io_wait(pair, + mytime_get_flush_timeout(), + true); + switch (ret) { + case IO_WAIT_MORE: + continue; + + case IO_WAIT_ERROR: + return SIZE_MAX; + + case IO_WAIT_TIMEOUT: + return size - left; + + default: + message_bug(); + } + } +#endif + message_error(_("%s: Read error: %s"), pair->src_name, strerror(errno)); - // FIXME Is this needed? - pair->src_eof = true; - return SIZE_MAX; } @@ -885,6 +1103,15 @@ io_write_buf(file_pair *pair, const uint8_t *buf, size_t size) continue; } +#ifndef TUKLIB_DOSLIKE + if (errno == EAGAIN || errno == EWOULDBLOCK) { + if (io_wait(pair, -1, false) == IO_WAIT_MORE) + continue; + + return true; + } +#endif + // Handle broken pipe specially. gzip and bzip2 // don't print anything on SIGPIPE. In addition, // gzip --quiet uses exit status 2 (warning) on diff --git a/src/xz/file_io.h b/src/xz/file_io.h index 967da868b079..2de3379238d6 100644 --- a/src/xz/file_io.h +++ b/src/xz/file_io.h @@ -68,6 +68,14 @@ typedef struct { extern void io_init(void); +#ifndef TUKLIB_DOSLIKE +/// \brief Write a byte to user_abort_pipe[1] +/// +/// This is called from a signal handler. +extern void io_write_to_user_abort_pipe(void); +#endif + + /// \brief Disable creation of sparse files when decompressing extern void io_no_sparse(void); @@ -102,6 +110,19 @@ extern void io_close(file_pair *pair, bool success); extern size_t io_read(file_pair *pair, io_buf *buf, size_t size); +/// \brief Fix the position in src_fd +/// +/// This is used when --single-thream has been specified and decompression +/// is successful. If the input file descriptor supports seeking, this +/// function fixes the input position to point to the next byte after the +/// decompressed stream. +/// +/// \param pair File pair having the source file open for reading +/// \param rewind_size How many bytes of extra have been read i.e. +/// how much to seek backwards. +extern void io_fix_src_pos(file_pair *pair, size_t rewind_size); + + /// \brief Read from source file from given offset to a buffer /// /// This is remotely similar to standard pread(). This uses lseek() though, diff --git a/src/xz/hardware.c b/src/xz/hardware.c index a4733c27e118..ff32f6d30148 100644 --- a/src/xz/hardware.c +++ b/src/xz/hardware.c @@ -11,12 +11,11 @@ /////////////////////////////////////////////////////////////////////////////// #include "private.h" -#include "tuklib_cpucores.h" -/// Maximum number of free *coder* threads. This can be set with +/// Maximum number of worker threads. This can be set with /// the --threads=NUM command line option. -static uint32_t threadlimit; +static uint32_t threads_max = 1; /// Memory usage limit for compression static uint64_t memlimit_compress; @@ -29,15 +28,23 @@ static uint64_t total_ram; extern void -hardware_threadlimit_set(uint32_t new_threadlimit) +hardware_threads_set(uint32_t n) { - if (new_threadlimit == 0) { - // The default is the number of available CPU cores. - threadlimit = tuklib_cpucores(); - if (threadlimit == 0) - threadlimit = 1; + if (n == 0) { + // Automatic number of threads was requested. + // If threading support was enabled at build time, + // use the number of available CPU cores. Otherwise + // use one thread since disabling threading support + // omits lzma_cputhreads() from liblzma. +#ifdef MYTHREAD_ENABLED + threads_max = lzma_cputhreads(); + if (threads_max == 0) + threads_max = 1; +#else + threads_max = 1; +#endif } else { - threadlimit = new_threadlimit; + threads_max = n; } return; @@ -45,9 +52,9 @@ hardware_threadlimit_set(uint32_t new_threadlimit) extern uint32_t -hardware_threadlimit_get(void) +hardware_threads_get(void) { - return threadlimit; + return threads_max; } @@ -139,6 +146,5 @@ hardware_init(void) // Set the defaults. hardware_memlimit_set(0, true, true, false); - hardware_threadlimit_set(0); return; } diff --git a/src/xz/hardware.h b/src/xz/hardware.h index ad526f260bc1..4fae61815656 100644 --- a/src/xz/hardware.h +++ b/src/xz/hardware.h @@ -15,12 +15,11 @@ extern void hardware_init(void); -/// Set custom value for maximum number of coder threads. -extern void hardware_threadlimit_set(uint32_t threadlimit); +/// Set the maximum number of worker threads. +extern void hardware_threads_set(uint32_t threadlimit); -/// Get the maximum number of coder threads. Some additional helper threads -/// are allowed on top of this). -extern uint32_t hardware_threadlimit_get(void); +/// Get the maximum number of worker threads. +extern uint32_t hardware_threads_get(void); /// Set the memory usage limit. There are separate limits for compression diff --git a/src/xz/list.c b/src/xz/list.c index 0e73d519ea4a..449c2bc4e02f 100644 --- a/src/xz/list.c +++ b/src/xz/list.c @@ -29,9 +29,12 @@ typedef struct { /// Uncompressed Size fields bool all_have_sizes; + /// Oldest XZ Utils version that will decompress the file + uint32_t min_version; + } xz_file_info; -#define XZ_FILE_INFO_INIT { NULL, 0, 0, true } +#define XZ_FILE_INFO_INIT { NULL, 0, 0, true, 50000002 } /// Information about a .xz Block @@ -104,8 +107,32 @@ static struct { uint64_t stream_padding; uint64_t memusage_max; uint32_t checks; + uint32_t min_version; bool all_have_sizes; -} totals = { 0, 0, 0, 0, 0, 0, 0, 0, true }; +} totals = { 0, 0, 0, 0, 0, 0, 0, 0, 0, true }; + + +/// Convert XZ Utils version number to a string. +static const char * +xz_ver_to_str(uint32_t ver) +{ + static char buf[32]; + + unsigned int major = ver / 10000000U; + ver -= major * 10000000U; + + unsigned int minor = ver / 10000U; + ver -= minor * 10000U; + + unsigned int patch = ver / 10U; + ver -= patch * 10U; + + const char *stability = ver == 0 ? "alpha" : ver == 1 ? "beta" : ""; + + snprintf(buf, sizeof(buf), "%u.%u.%u%s", + major, minor, patch, stability); + return buf; +} /// \brief Parse the Index(es) from the given .xz file @@ -478,6 +505,21 @@ parse_block_header(file_pair *pair, const lzma_index_iter *iter, if (xfi->memusage_max < bhi->memusage) xfi->memusage_max = bhi->memusage; + // Determine the minimum XZ Utils version that supports this Block. + // + // Currently the only thing that 5.0.0 doesn't support is empty + // LZMA2 Block. This decoder bug was fixed in 5.0.2. + { + size_t i = 0; + while (filters[i + 1].id != LZMA_VLI_UNKNOWN) + ++i; + + if (filters[i].id == LZMA_FILTER_LZMA2 + && iter->block.uncompressed_size == 0 + && xfi->min_version < 50000022U) + xfi->min_version = 50000022U; + } + // Convert the filter chain to human readable form. message_filters_to_str(bhi->filter_chain, filters, false); @@ -856,6 +898,8 @@ print_info_adv(xz_file_info *xfi, file_pair *pair) round_up_to_mib(xfi->memusage_max), 0)); printf(_(" Sizes in headers: %s\n"), xfi->all_have_sizes ? _("Yes") : _("No")); + printf(_(" Minimum XZ Utils version: %s\n"), + xz_ver_to_str(xfi->min_version)); } return false; @@ -938,9 +982,10 @@ print_info_robot(xz_file_info *xfi, file_pair *pair) } if (message_verbosity_get() >= V_DEBUG) - printf("summary\t%" PRIu64 "\t%s\n", + printf("summary\t%" PRIu64 "\t%s\t%" PRIu32 "\n", xfi->memusage_max, - xfi->all_have_sizes ? "yes" : "no"); + xfi->all_have_sizes ? "yes" : "no", + xfi->min_version); return false; } @@ -961,6 +1006,9 @@ update_totals(const xz_file_info *xfi) if (totals.memusage_max < xfi->memusage_max) totals.memusage_max = xfi->memusage_max; + if (totals.min_version < xfi->min_version) + totals.min_version = xfi->min_version; + totals.all_have_sizes &= xfi->all_have_sizes; return; @@ -1025,6 +1073,8 @@ print_totals_adv(void) round_up_to_mib(totals.memusage_max), 0)); printf(_(" Sizes in headers: %s\n"), totals.all_have_sizes ? _("Yes") : _("No")); + printf(_(" Minimum XZ Utils version: %s\n"), + xz_ver_to_str(totals.min_version)); } return; @@ -1050,9 +1100,10 @@ print_totals_robot(void) totals.files); if (message_verbosity_get() >= V_DEBUG) - printf("\t%" PRIu64 "\t%s", + printf("\t%" PRIu64 "\t%s\t%" PRIu32, totals.memusage_max, - totals.all_have_sizes ? "yes" : "no"); + totals.all_have_sizes ? "yes" : "no", + totals.min_version); putchar('\n'); diff --git a/src/xz/main.c b/src/xz/main.c index 8196c6e7e774..a8f0683a47bd 100644 --- a/src/xz/main.c +++ b/src/xz/main.c @@ -275,6 +275,11 @@ main(int argc, char **argv) list_totals(); } +#ifndef NDEBUG + coder_free(); + args_free(); +#endif + // If we have got a signal, raise it to kill the program instead // of calling tuklib_exit(). signals_exit(); diff --git a/src/xz/message.c b/src/xz/message.c index 0a7a522f7afa..8a31b00ed89c 100644 --- a/src/xz/message.c +++ b/src/xz/message.c @@ -12,10 +12,6 @@ #include "private.h" -#ifdef HAVE_SYS_TIME_H -# include <sys/time.h> -#endif - #include <stdarg.h> @@ -64,9 +60,6 @@ static lzma_stream *progress_strm; /// and estimate remaining time. static uint64_t expected_in_size; -/// Time when we started processing the file -static uint64_t start_time; - // Use alarm() and SIGALRM when they are supported. This has two minor // advantages over the alternative of polling gettimeofday(): @@ -112,16 +105,6 @@ static uint64_t progress_next_update; #endif -/// Get the current time as microseconds since epoch -static uint64_t -my_time(void) -{ - struct timeval tv; - gettimeofday(&tv, NULL); - return (uint64_t)(tv.tv_sec) * UINT64_C(1000000) + tv.tv_usec; -} - - extern void message_init(void) { @@ -264,11 +247,10 @@ message_progress_start(lzma_stream *strm, uint64_t in_size) // It is needed to find out the position in the stream. progress_strm = strm; - // Store the processing start time of the file and its expected size. - // If we aren't printing any statistics, then these are unused. But - // since it is possible that the user sends us a signal to show - // statistics, we need to have these available anyway. - start_time = my_time(); + // Store the expected size of the file. If we aren't printing any + // statistics, then is will be unused. But since it is possible + // that the user sends us a signal to show statistics, we need + // to have it available anyway. expected_in_size = in_size; // Indicate that progress info may need to be printed before @@ -290,7 +272,7 @@ message_progress_start(lzma_stream *strm, uint64_t in_size) alarm(1); #else progress_needs_updating = true; - progress_next_update = 1000000; + progress_next_update = 1000; #endif } @@ -364,7 +346,7 @@ progress_speed(uint64_t uncompressed_pos, uint64_t elapsed) { // Don't print the speed immediately, since the early values look // somewhat random. - if (elapsed < 3000000) + if (elapsed < 3000) return ""; static const char unit[][8] = { @@ -377,7 +359,7 @@ progress_speed(uint64_t uncompressed_pos, uint64_t elapsed) // Calculate the speed as KiB/s. double speed = (double)(uncompressed_pos) - / ((double)(elapsed) * (1024.0 / 1e6)); + / ((double)(elapsed) * (1024.0 / 1000.0)); // Adjust the unit of the speed if needed. while (speed > 999.0) { @@ -402,12 +384,12 @@ progress_speed(uint64_t uncompressed_pos, uint64_t elapsed) /// Make a string indicating elapsed or remaining time. The format is either /// M:SS or H:MM:SS depending on if the time is an hour or more. static const char * -progress_time(uint64_t useconds) +progress_time(uint64_t mseconds) { // 9999 hours = 416 days static char buf[sizeof("9999:59:59")]; - uint32_t seconds = useconds / 1000000; + uint32_t seconds = mseconds / 1000; // Don't show anything if the time is zero or ridiculously big. if (seconds == 0 || seconds > ((9999 * 60) + 59) * 60 + 59) @@ -445,14 +427,14 @@ progress_remaining(uint64_t in_pos, uint64_t elapsed) // - Only a few seconds has passed since we started (de)compressing, // so estimate would be too inaccurate. if (expected_in_size == 0 || in_pos > expected_in_size - || in_pos < (UINT64_C(1) << 19) || elapsed < 8000000) + || in_pos < (UINT64_C(1) << 19) || elapsed < 8000) return ""; // Calculate the estimate. Don't give an estimate of zero seconds, // since it is possible that all the input has been already passed // to the library, but there is still quite a bit of output pending. uint32_t remaining = (double)(expected_in_size - in_pos) - * ((double)(elapsed) / 1e6) / (double)(in_pos); + * ((double)(elapsed) / 1000.0) / (double)(in_pos); if (remaining < 1) remaining = 1; @@ -518,28 +500,26 @@ progress_remaining(uint64_t in_pos, uint64_t elapsed) } -/// Calculate the elapsed time as microseconds. -static uint64_t -progress_elapsed(void) -{ - return my_time() - start_time; -} - - -/// Get information about position in the stream. This is currently simple, -/// but it will become more complicated once we have multithreading support. +/// Get how much uncompressed and compressed data has been processed. static void progress_pos(uint64_t *in_pos, uint64_t *compressed_pos, uint64_t *uncompressed_pos) { - *in_pos = progress_strm->total_in; + uint64_t out_pos; + lzma_get_progress(progress_strm, in_pos, &out_pos); + + // It cannot have processed more input than it has been given. + assert(*in_pos <= progress_strm->total_in); + + // It cannot have produced more output than it claims to have ready. + assert(out_pos >= progress_strm->total_out); if (opt_mode == MODE_COMPRESS) { - *compressed_pos = progress_strm->total_out; - *uncompressed_pos = progress_strm->total_in; + *compressed_pos = out_pos; + *uncompressed_pos = *in_pos; } else { - *compressed_pos = progress_strm->total_in; - *uncompressed_pos = progress_strm->total_out; + *compressed_pos = *in_pos; + *uncompressed_pos = out_pos; } return; @@ -553,13 +533,13 @@ message_progress_update(void) return; // Calculate how long we have been processing this file. - const uint64_t elapsed = progress_elapsed(); + const uint64_t elapsed = mytime_get_elapsed(); #ifndef SIGALRM if (progress_next_update > elapsed) return; - progress_next_update = elapsed + 1000000; + progress_next_update = elapsed + 1000; #endif // Get our current position in the stream. @@ -652,7 +632,7 @@ progress_flush(bool finished) progress_active = false; - const uint64_t elapsed = progress_elapsed(); + const uint64_t elapsed = mytime_get_elapsed(); signals_block(); @@ -1122,7 +1102,10 @@ message_help(bool long_help) " -f, --force force overwrite of output file and (de)compress links\n" " -c, --stdout write to standard output and don't delete input files")); - if (long_help) + if (long_help) { + puts(_( +" --single-stream decompress only the first stream, and silently\n" +" ignore possible remaining input data")); puts(_( " --no-sparse do not create sparse files when decompressing\n" " -S, --suffix=.SUF use the suffix `.SUF' on compressed files\n" @@ -1130,6 +1113,7 @@ message_help(bool long_help) " omitted, filenames are read from the standard input;\n" " filenames must be terminated with the newline character\n" " --files0[=FILE] like --files but use the null character as terminator")); + } if (long_help) { puts(_("\n Basic file format and compression options:\n")); @@ -1138,6 +1122,8 @@ message_help(bool long_help) " `auto' (default), `xz', `lzma', and `raw'\n" " -C, --check=CHECK integrity check type: `none' (use with caution),\n" " `crc32', `crc64' (default), or `sha256'")); + puts(_( +" --ignore-check don't verify the integrity check when decompressing")); } puts(_( @@ -1148,7 +1134,25 @@ message_help(bool long_help) " -e, --extreme try to improve compression ratio by using more CPU time;\n" " does not affect decompressor memory requirements")); + puts(_( +" -T, --threads=NUM use at most NUM threads; the default is 1; set to 0\n" +" to use as many threads as there are processor cores")); + if (long_help) { + puts(_( +" --block-size=SIZE\n" +" start a new .xz block after every SIZE bytes of input;\n" +" use this to set the block size for threaded compression")); + puts(_( +" --block-list=SIZES\n" +" start a new .xz block after the given comma-separated\n" +" intervals of uncompressed data")); + puts(_( +" --flush-timeout=TIMEOUT\n" +" when compressing, if more than TIMEOUT milliseconds has\n" +" passed since the previous flush and reading more input\n" +" would block, all pending data is flushed out" + )); puts(_( // xgettext:no-c-format " --memlimit-compress=LIMIT\n" " --memlimit-decompress=LIMIT\n" @@ -1244,5 +1248,10 @@ message_help(bool long_help) PACKAGE_BUGREPORT); printf(_("%s home page: <%s>\n"), PACKAGE_NAME, PACKAGE_URL); +#if LZMA_VERSION_STABILITY != LZMA_VERSION_STABILITY_STABLE + puts(_( +"THIS IS A DEVELOPMENT VERSION NOT INTENDED FOR PRODUCTION USE.")); +#endif + tuklib_exit(E_SUCCESS, E_ERROR, verbosity != V_SILENT); } diff --git a/src/xz/mytime.c b/src/xz/mytime.c new file mode 100644 index 000000000000..4be184fd19da --- /dev/null +++ b/src/xz/mytime.c @@ -0,0 +1,89 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file mytime.c +/// \brief Time handling functions +// +// Author: Lasse Collin +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + +#include "private.h" + +#if !(defined(HAVE_CLOCK_GETTIME) && HAVE_DECL_CLOCK_MONOTONIC) +# include <sys/time.h> +#endif + +uint64_t opt_flush_timeout = 0; +bool flush_needed; + +static uint64_t start_time; +static uint64_t next_flush; + + +/// \brief Get the current time as milliseconds +/// +/// It's relative to some point but not necessarily to the UNIX Epoch. +static uint64_t +mytime_now(void) +{ + // NOTE: HAVE_DECL_CLOCK_MONOTONIC is always defined to 0 or 1. +#if defined(HAVE_CLOCK_GETTIME) && HAVE_DECL_CLOCK_MONOTONIC + // If CLOCK_MONOTONIC was available at compile time but for some + // reason isn't at runtime, fallback to CLOCK_REALTIME which + // according to POSIX is mandatory for all implementations. + static clockid_t clk_id = CLOCK_MONOTONIC; + struct timespec tv; + while (clock_gettime(clk_id, &tv)) + clk_id = CLOCK_REALTIME; + + return (uint64_t)(tv.tv_sec) * UINT64_C(1000) + tv.tv_nsec / 1000000; +#else + struct timeval tv; + gettimeofday(&tv, NULL); + return (uint64_t)(tv.tv_sec) * UINT64_C(1000) + tv.tv_usec / 1000; +#endif +} + + +extern void +mytime_set_start_time(void) +{ + start_time = mytime_now(); + next_flush = start_time + opt_flush_timeout; + flush_needed = false; + return; +} + + +extern uint64_t +mytime_get_elapsed(void) +{ + return mytime_now() - start_time; +} + + +extern void +mytime_set_flush_time(void) +{ + next_flush = mytime_now() + opt_flush_timeout; + flush_needed = false; + return; +} + + +extern int +mytime_get_flush_timeout(void) +{ + if (opt_flush_timeout == 0 || opt_mode != MODE_COMPRESS) + return -1; + + const uint64_t now = mytime_now(); + if (now >= next_flush) + return 0; + + const uint64_t remaining = next_flush - now; + return remaining > INT_MAX ? INT_MAX : (int)remaining; +} diff --git a/src/xz/mytime.h b/src/xz/mytime.h new file mode 100644 index 000000000000..ea291eed81c7 --- /dev/null +++ b/src/xz/mytime.h @@ -0,0 +1,47 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file mytime.h +/// \brief Time handling functions +// +// Author: Lasse Collin +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + + +/// \brief Number of milliseconds to between LZMA_SYNC_FLUSHes +/// +/// If 0, timed flushing is disabled. Otherwise if no more input is available +/// and not at the end of the file and at least opt_flush_timeout milliseconds +/// has elapsed since the start of compression or the previous flushing +/// (LZMA_SYNC_FLUSH or LZMA_FULL_FLUSH), set LZMA_SYNC_FLUSH to flush +/// the pending data. +extern uint64_t opt_flush_timeout; + + +/// \brief True when flushing is needed due to expired timeout +extern bool flush_needed; + + +/// \brief Store the time when (de)compression was started +/// +/// The start time is also stored as the time of the first flush. +extern void mytime_set_start_time(void); + + +/// \brief Get the number of milliseconds since the operation started +extern uint64_t mytime_get_elapsed(void); + + +/// \brief Store the time of when compressor was flushed +extern void mytime_set_flush_time(void); + + +/// \brief Get the number of milliseconds until the next flush +/// +/// This returns -1 if no timed flushing is used. +/// +/// The return value is inteded for use with poll(). +extern int mytime_get_flush_timeout(void); diff --git a/src/xz/options.c b/src/xz/options.c index f21a0ba51065..f9c7ab9e8535 100644 --- a/src/xz/options.c +++ b/src/xz/options.c @@ -31,8 +31,8 @@ typedef struct { } option_map; -/// Parses option=value pairs that are separated with colons, semicolons, -/// or commas: opt=val:opt=val;opt=val,opt=val +/// Parses option=value pairs that are separated with commas: +/// opt=val,opt=val,opt=val /// /// Each option is a string, that is converted to an integer using the /// index where the option string is in the array. diff --git a/src/xz/private.h b/src/xz/private.h index 6b01e51354e1..4acfa8dc4558 100644 --- a/src/xz/private.h +++ b/src/xz/private.h @@ -12,6 +12,7 @@ #include "sysdefs.h" #include "mythread.h" + #include "lzma.h" #include <sys/types.h> @@ -45,6 +46,7 @@ #endif #include "main.h" +#include "mytime.h" #include "coder.h" #include "message.h" #include "args.h" diff --git a/src/xz/signals.c b/src/xz/signals.c index 322811f472b5..5387c424e1a5 100644 --- a/src/xz/signals.c +++ b/src/xz/signals.c @@ -41,6 +41,11 @@ signal_handler(int sig) { exit_signal = sig; user_abort = true; + +#ifndef TUKLIB_DOSLIKE + io_write_to_user_abort_pipe(); +#endif + return; } diff --git a/src/xz/suffix.c b/src/xz/suffix.c index 8e331a7022a3..9d4fcd139b8f 100644 --- a/src/xz/suffix.c +++ b/src/xz/suffix.c @@ -12,6 +12,10 @@ #include "private.h" +#ifdef __DJGPP__ +# include <fcntl.h> +#endif + // For case-insensitive filename suffix on case-insensitive systems #if defined(TUKLIB_DOSLIKE) || defined(__VMS) # define strcmp strcasecmp @@ -45,6 +49,31 @@ has_dir_sep(const char *str) } +#ifdef __DJGPP__ +/// \brief Test for special suffix used for 8.3 short filenames (SFN) +/// +/// \return If str matches *.?- or *.??-, true is returned. Otherwise +/// false is returned. +static bool +has_sfn_suffix(const char *str, size_t len) +{ + if (len >= 4 && str[len - 1] == '-' && str[len - 2] != '.' + && !is_dir_sep(str[len - 2])) { + // *.?- + if (str[len - 3] == '.') + return !is_dir_sep(str[len - 4]); + + // *.??- + if (len >= 5 && !is_dir_sep(str[len - 3]) + && str[len - 4] == '.') + return !is_dir_sep(str[len - 5]); + } + + return false; +} +#endif + + /// \brief Checks if src_name has given compressed_suffix /// /// \param suffix Filename suffix to look for @@ -87,6 +116,9 @@ uncompressed_name(const char *src_name, const size_t src_len) { ".xz", "" }, { ".txz", ".tar" }, // .txz abbreviation for .txt.gz is rare. { ".lzma", "" }, +#ifdef __DJGPP__ + { ".lzm", "" }, +#endif { ".tlz", ".tar" }, // { ".gz", "" }, // { ".tgz", ".tar" }, @@ -112,6 +144,17 @@ uncompressed_name(const char *src_name, const size_t src_len) break; } } + +#ifdef __DJGPP__ + // Support also *.?- -> *.? and *.??- -> *.?? on DOS. + // This is done also when long filenames are available + // to keep it easy to decompress files created when + // long filename support wasn't available. + if (new_len == 0 && has_sfn_suffix(src_name, src_len)) { + new_suffix = ""; + new_len = src_len - 1; + } +#endif } if (new_len == 0 && custom_suffix != NULL) @@ -134,21 +177,35 @@ uncompressed_name(const char *src_name, const size_t src_len) } +/// This message is needed in multiple places in compressed_name(), +/// so the message has been put into its own function. +static void +msg_suffix(const char *src_name, const char *suffix) +{ + message_warning(_("%s: File already has `%s' suffix, skipping"), + src_name, suffix); + return; +} + + /// \brief Appends suffix to src_name /// /// In contrast to uncompressed_name(), we check only suffixes that are valid /// for the specified file format. static char * -compressed_name(const char *src_name, const size_t src_len) +compressed_name(const char *src_name, size_t src_len) { // The order of these must match the order in args.h. - static const char *const all_suffixes[][3] = { + static const char *const all_suffixes[][4] = { { ".xz", ".txz", NULL }, { ".lzma", +#ifdef __DJGPP__ + ".lzm", +#endif ".tlz", NULL /* @@ -170,20 +227,27 @@ compressed_name(const char *src_name, const size_t src_len) const size_t format = opt_format - 1; const char *const *suffixes = all_suffixes[format]; + // Look for known filename suffixes and refuse to compress them. for (size_t i = 0; suffixes[i] != NULL; ++i) { if (test_suffix(suffixes[i], src_name, src_len) != 0) { - message_warning(_("%s: File already has `%s' " - "suffix, skipping"), src_name, - suffixes[i]); + msg_suffix(src_name, suffixes[i]); return NULL; } } +#ifdef __DJGPP__ + // Recognize also the special suffix that is used when long + // filename (LFN) support isn't available. This suffix is + // recognized on LFN systems too. + if (opt_format == FORMAT_XZ && has_sfn_suffix(src_name, src_len)) { + msg_suffix(src_name, "-"); + return NULL; + } +#endif + if (custom_suffix != NULL) { if (test_suffix(custom_suffix, src_name, src_len) != 0) { - message_warning(_("%s: File already has `%s' " - "suffix, skipping"), src_name, - custom_suffix); + msg_suffix(src_name, custom_suffix); return NULL; } } @@ -199,7 +263,101 @@ compressed_name(const char *src_name, const size_t src_len) const char *suffix = custom_suffix != NULL ? custom_suffix : suffixes[0]; - const size_t suffix_len = strlen(suffix); + size_t suffix_len = strlen(suffix); + +#ifdef __DJGPP__ + if (!_use_lfn(src_name)) { + // Long filename (LFN) support isn't available and we are + // limited to 8.3 short filenames (SFN). + // + // Look for suffix separator from the filename, and make sure + // that it is in the filename, not in a directory name. + const char *sufsep = strrchr(src_name, '.'); + if (sufsep == NULL || sufsep[1] == '\0' + || has_dir_sep(sufsep)) { + // src_name has no filename extension. + // + // Examples: + // xz foo -> foo.xz + // xz -F lzma foo -> foo.lzm + // xz -S x foo -> foox + // xz -S x foo. -> foo.x + // xz -S x.y foo -> foox.y + // xz -S .x foo -> foo.x + // xz -S .x foo. -> foo.x + // + // Avoid double dots: + if (sufsep != NULL && sufsep[1] == '\0' + && suffix[0] == '.') + --src_len; + + } else if (custom_suffix == NULL + && strcasecmp(sufsep, ".tar") == 0) { + // ".tar" is handled specially. + // + // Examples: + // xz foo.tar -> foo.txz + // xz -F lzma foo.tar -> foo.tlz + static const char *const tar_suffixes[] = { + ".txz", + ".tlz", + // ".tgz", + }; + suffix = tar_suffixes[format]; + suffix_len = 4; + src_len -= 4; + + } else { + if (custom_suffix == NULL && opt_format == FORMAT_XZ) { + // Instead of the .xz suffix, use a single + // character at the end of the filename + // extension. This is to minimize name + // conflicts when compressing multiple files + // with the same basename. E.g. foo.txt and + // foo.exe become foo.tx- and foo.ex-. Dash + // is rare as the last character of the + // filename extension, so it seems to be + // quite safe choice and it stands out better + // in directory listings than e.g. x. For + // comparison, gzip uses z. + suffix = "-"; + suffix_len = 1; + } + + if (suffix[0] == '.') { + // The first character of the suffix is a dot. + // Throw away the original filename extension + // and replace it with the new suffix. + // + // Examples: + // xz -F lzma foo.txt -> foo.lzm + // xz -S .x foo.txt -> foo.x + src_len = sufsep - src_name; + + } else { + // The first character of the suffix is not + // a dot. Preserve the first 0-2 characters + // of the original filename extension. + // + // Examples: + // xz foo.txt -> foo.tx- + // xz -S x foo.c -> foo.cx + // xz -S ab foo.c -> foo.cab + // xz -S ab foo.txt -> foo.tab + // xz -S abc foo.txt -> foo.abc + // + // Truncate the suffix to three chars: + if (suffix_len > 3) + suffix_len = 3; + + // If needed, overwrite 1-3 characters. + if (strlen(sufsep) > 4 - suffix_len) + src_len = sufsep - src_name + + 4 - suffix_len; + } + } + } +#endif char *dest_name = xmalloc(src_len + suffix_len + 1); diff --git a/src/xz/xz.1 b/src/xz/xz.1 index 363b90cb4e55..75aead3d0439 100644 --- a/src/xz/xz.1 +++ b/src/xz/xz.1 @@ -5,7 +5,7 @@ .\" This file has been put into the public domain. .\" You can do whatever you want with this file. .\" -.TH XZ 1 "2013-06-21" "Tukaani" "XZ Utils" +.TH XZ 1 "2014-12-16" "Tukaani" "XZ Utils" . .SH NAME xz, unxz, xzcat, lzma, unlzma, lzcat \- Compress or decompress .xz and .lzma files @@ -435,6 +435,29 @@ standard output instead of a file. This implies .BR \-\-keep . .TP +.B \-\-single\-stream +Decompress only the first +.B .xz +stream, and +silently ignore possible remaining input data following the stream. +Normally such trailing garbage makes +.B xz +display an error. +.IP "" +.B xz +never decompresses more than one stream from +.B .lzma +files or raw streams, but this option still makes +.B xz +ignore the possible trailing data after the +.B .lzma +file or raw stream. +.IP "" +This option has no effect if the operation mode is not +.B \-\-decompress +or +.BR \-\-test . +.TP .B \-\-no\-sparse Disable creation of sparse files. By default, if decompressing into a regular file, @@ -586,6 +609,25 @@ Integrity of the headers is always verified with CRC32. It is not possible to change or disable it. .TP +.B \-\-ignore\-check +Don't verify the integrity check of the compressed data when decompressing. +The CRC32 values in the +.B .xz +headers will still be verified normally. +.IP "" +.B "Do not use this option unless you know what you are doing." +Possible reasons to use this option: +.RS +.IP \(bu 3 +Trying to recover data from a corrupt .xz file. +.IP \(bu 3 +Speeding up decompression. +This matters mostly with SHA-256 or +with files that have compressed extremely well. +It's recommended to not use this option for this purpose +unless the file integrity is verified externally in some other way. +.RE +.TP .BR \-0 " ... " \-9 Select a compression preset level. The default is @@ -778,6 +820,124 @@ These are provided only for backwards compatibility with LZMA Utils. Avoid using these options. .TP +.BI \-\-block\-size= size +When compressing to the +.B .xz +format, split the input data into blocks of +.I size +bytes. +The blocks are compressed independently from each other, +which helps with multi-threading and +makes limited random-access decompression possible. +This option is typically used to override the default +block size in multi-threaded mode, +but this option can be used in single-threaded mode too. +.IP "" +In multi-threaded mode about three times +.I size +bytes will be allocated in each thread for buffering input and output. +The default +.I size +is three times the LZMA2 dictionary size or 1 MiB, +whichever is more. +Typically a good value is 2\-4 times +the size of the LZMA2 dictionary or at least 1 MiB. +Using +.I size +less than the LZMA2 dictionary size is waste of RAM +because then the LZMA2 dictionary buffer will never get fully used. +The sizes of the blocks are stored in the block headers, +which a future version of +.B xz +will use for multi-threaded decompression. +.IP "" +In single-threaded mode no block splitting is done by default. +Setting this option doesn't affect memory usage. +No size information is stored in block headers, +thus files created in single-threaded mode +won't be identical to files created in multi-threaded mode. +The lack of size information also means that a future version of +.B xz +won't be able decompress the files in multi-threaded mode. +.TP +.BI \-\-block\-list= sizes +When compressing to the +.B .xz +format, start a new block after +the given intervals of uncompressed data. +.IP "" +The uncompressed +.I sizes +of the blocks are specified as a comma-separated list. +Omitting a size (two or more consecutive commas) is a shorthand +to use the size of the previous block. +.IP "" +If the input file is bigger than the sum of +.IR sizes , +the last value in +.I sizes +is repeated until the end of the file. +A special value of +.B 0 +may be used as the last value to indicate that +the rest of the file should be encoded as a single block. +.IP "" +If one specifies +.I sizes +that exceed the encoder's block size +(either the default value in threaded mode or +the value specified with \fB\-\-block\-size=\fIsize\fR), +the encoder will create additional blocks while +keeping the boundaries specified in +.IR sizes . +For example, if one specifies +.B \-\-block\-size=10MiB +.B \-\-block\-list=5MiB,10MiB,8MiB,12MiB,24MiB +and the input file is 80 MiB, +one will get 11 blocks: +5, 10, 8, 10, 2, 10, 10, 4, 10, 10, and 1 MiB. +.IP "" +In multi-threaded mode the sizes of the blocks +are stored in the block headers. +This isn't done in single-threaded mode, +so the encoded output won't be +identical to that of the multi-threaded mode. +.TP +.BI \-\-flush\-timeout= timeout +When compressing, if more than +.I timeout +milliseconds (a positive integer) has passed since the previous flush and +reading more input would block, +all the pending input data is flushed from the encoder and +made available in the output stream. +This can be useful if +.B xz +is used to compress data that is streamed over a network. +Small +.I timeout +values make the data available at the receiving end +with a small delay, but large +.I timeout +values give better compression ratio. +.IP "" +This feature is disabled by default. +If this option is specified more than once, the last one takes effect. +The special +.I timeout +value of +.B 0 +can be used to explicitly disable this feature. +.IP "" +This feature is not available on non-POSIX systems. +.IP "" +.\" FIXME +.B "This feature is still experimental." +Currently +.B xz +is unsuitable for decompressing the stream in real time due to how +.B xz +does buffering. +.TP .BI \-\-memlimit\-compress= limit Set a memory usage limit for compression. If this option is specified multiple times, @@ -876,24 +1036,25 @@ Automatic adjusting is always disabled when creating raw streams .TP \fB\-T\fR \fIthreads\fR, \fB\-\-threads=\fIthreads Specify the number of worker threads to use. +Setting +.I threads +to a special value +.B 0 +makes +.B xz +use as many threads as there are CPU cores on the system. The actual number of threads can be less than .I threads +if the input file is not big enough +for threading with the given settings or if using more threads would exceed the memory usage limit. .IP "" -.B "Multithreaded compression and decompression are not" -.B "implemented yet, so this option has no effect for now." -.IP "" -.B "As of writing (2010-09-27), it hasn't been decided" -.B "if threads will be used by default on multicore systems" -.B "once support for threading has been implemented." -.B "Comments are welcome." -The complicating factor is that using many threads -will increase the memory usage dramatically. -Note that if multithreading will be the default, -it will probably be done so that single-threaded and -multithreaded modes produce the same output, -so compression ratio won't be significantly affected -if threading will be enabled by default. +Currently the only threading method is to split the input into +blocks and compress them independently from each other. +The default block size depends on the compression level and +can be overriden with the +.BI \-\-block\-size= size +option. . .SS "Custom compressor filter chains" A custom filter chain allows specifying @@ -1863,6 +2024,14 @@ or .B no indicating if all block headers have both compressed size and uncompressed size stored in them +.PP +.I Since +.B xz +.I 5.1.2alpha: +.IP 4. 4 +Minimum +.B xz +version required to decompress the file .RE .PD .PP @@ -1913,6 +2082,14 @@ or .B no indicating if all block headers have both compressed size and uncompressed size stored in them +.PP +.I Since +.B xz +.I 5.1.2alpha: +.IP 12. 4 +Minimum +.B xz +version required to decompress the file .RE .PD .PP @@ -2173,7 +2350,9 @@ If there is data left after the first .B .lzma stream, .B xz -considers the file to be corrupt. +considers the file to be corrupt unless +.B \-\-single\-stream +was used. This may break obscure scripts which have assumed that trailing garbage is ignored. . |