extzstd 0.3.2 → 0.4

data/contrib/zstd/lib/common/pool.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -10,9 +10,9 @@
 
 
 /* ======   Dependencies   ======= */
+#include "../common/allocations.h"  /* ZSTD_customCalloc, ZSTD_customFree */
 #include "zstd_deps.h" /* size_t */
 #include "debug.h"     /* assert */
-#include "zstd_internal.h"  /* ZSTD_customMalloc, ZSTD_customFree */
 #include "pool.h"
 
 /* ======   Compiler specifics   ====== */
@@ -86,7 +86,7 @@ static void* POOL_thread(void* opaque) {
         {   POOL_job const job = ctx->queue[ctx->queueHead];
             ctx->queueHead = (ctx->queueHead + 1) % ctx->queueSize;
             ctx->numThreadsBusy++;
-            ctx->queueEmpty = ctx->queueHead == ctx->queueTail;
+            ctx->queueEmpty = (ctx->queueHead == ctx->queueTail);
             /* Unlock the mutex, signal a pusher, and run the job */
             ZSTD_pthread_cond_signal(&ctx->queuePushCond);
             ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
@@ -96,15 +96,14 @@ static void* POOL_thread(void* opaque) {
             /* If the intended queue size was 0, signal after finishing job */
             ZSTD_pthread_mutex_lock(&ctx->queueMutex);
             ctx->numThreadsBusy--;
-            if (ctx->queueSize == 1) {
-                ZSTD_pthread_cond_signal(&ctx->queuePushCond);
-            }
+            ZSTD_pthread_cond_signal(&ctx->queuePushCond);
             ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
         }
     }  /* for (;;) */
     assert(0);  /* Unreachable */
 }
 
+/* ZSTD_createThreadPool() : public access point */
 POOL_ctx* ZSTD_createThreadPool(size_t numThreads) {
     return POOL_create (numThreads, 0);
 }
@@ -114,7 +113,8 @@ POOL_ctx* POOL_create(size_t numThreads, size_t queueSize) {
 }
 
 POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize,
-                               ZSTD_customMem customMem) {
+                               ZSTD_customMem customMem)
+{
     POOL_ctx* ctx;
     /* Check parameters */
     if (!numThreads) { return NULL; }
@@ -126,7 +126,7 @@ POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize,
      * empty and full queues.
      */
     ctx->queueSize = queueSize + 1;
-    ctx->queue = (POOL_job*)ZSTD_customMalloc(ctx->queueSize * sizeof(POOL_job), customMem);
+    ctx->queue = (POOL_job*)ZSTD_customCalloc(ctx->queueSize * sizeof(POOL_job), customMem);
     ctx->queueHead = 0;
     ctx->queueTail = 0;
     ctx->numThreadsBusy = 0;
@@ -140,7 +140,7 @@ POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize,
     }
     ctx->shutdown = 0;
     /* Allocate space for the thread handles */
-    ctx->threads = (ZSTD_pthread_t*)ZSTD_customMalloc(numThreads * sizeof(ZSTD_pthread_t), customMem);
+    ctx->threads = (ZSTD_pthread_t*)ZSTD_customCalloc(numThreads * sizeof(ZSTD_pthread_t), customMem);
     ctx->threadCapacity = 0;
     ctx->customMem = customMem;
     /* Check for errors */
@@ -173,7 +173,7 @@ static void POOL_join(POOL_ctx* ctx) {
     /* Join all of the threads */
     {   size_t i;
         for (i = 0; i < ctx->threadCapacity; ++i) {
-            ZSTD_pthread_join(ctx->threads[i], NULL);  /* note : could fail */
+            ZSTD_pthread_join(ctx->threads[i]);  /* note : could fail */
     }   }
 }
 
@@ -188,11 +188,22 @@ void POOL_free(POOL_ctx *ctx) {
     ZSTD_customFree(ctx, ctx->customMem);
 }
 
+/*! POOL_joinJobs() :
+ *  Waits for all queued jobs to finish executing.
+ */
+void POOL_joinJobs(POOL_ctx* ctx) {
+    ZSTD_pthread_mutex_lock(&ctx->queueMutex);
+    while(!ctx->queueEmpty || ctx->numThreadsBusy > 0) {
+        ZSTD_pthread_cond_wait(&ctx->queuePushCond, &ctx->queueMutex);
+    }
+    ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
+}
+
 void ZSTD_freeThreadPool (ZSTD_threadPool* pool) {
   POOL_free (pool);
 }
 
-size_t POOL_sizeof(POOL_ctx *ctx) {
+size_t POOL_sizeof(const POOL_ctx* ctx) {
     if (ctx==NULL) return 0;  /* supports sizeof NULL */
     return sizeof(*ctx)
         + ctx->queueSize * sizeof(POOL_job)
@@ -209,10 +220,10 @@ static int POOL_resize_internal(POOL_ctx* ctx, size_t numThreads)
         return 0;
     }
     /* numThreads > threadCapacity */
-    {   ZSTD_pthread_t* const threadPool = (ZSTD_pthread_t*)ZSTD_customMalloc(numThreads * sizeof(ZSTD_pthread_t), ctx->customMem);
+    {   ZSTD_pthread_t* const threadPool = (ZSTD_pthread_t*)ZSTD_customCalloc(numThreads * sizeof(ZSTD_pthread_t), ctx->customMem);
         if (!threadPool) return 1;
         /* replace existing thread pool */
-        ZSTD_memcpy(threadPool, ctx->threads, ctx->threadCapacity * sizeof(*threadPool));
+        ZSTD_memcpy(threadPool, ctx->threads, ctx->threadCapacity * sizeof(ZSTD_pthread_t));
         ZSTD_customFree(ctx->threads, ctx->customMem);
         ctx->threads = threadPool;
         /* Initialize additional threads */
@@ -257,9 +268,12 @@ static int isQueueFull(POOL_ctx const* ctx) {
 }
 
 
-static void POOL_add_internal(POOL_ctx* ctx, POOL_function function, void *opaque)
+static void
+POOL_add_internal(POOL_ctx* ctx, POOL_function function, void *opaque)
 {
-    POOL_job const job = {function, opaque};
+    POOL_job job;
+    job.function = function;
+    job.opaque = opaque;
     assert(ctx != NULL);
     if (ctx->shutdown) return;
 
@@ -313,7 +327,9 @@ POOL_ctx* POOL_create(size_t numThreads, size_t queueSize) {
     return POOL_create_advanced(numThreads, queueSize, ZSTD_defaultCMem);
 }
 
-POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize, ZSTD_customMem customMem) {
+POOL_ctx*
+POOL_create_advanced(size_t numThreads, size_t queueSize, ZSTD_customMem customMem)
+{
     (void)numThreads;
     (void)queueSize;
     (void)customMem;
@@ -325,6 +341,11 @@ void POOL_free(POOL_ctx* ctx) {
     (void)ctx;
 }
 
+void POOL_joinJobs(POOL_ctx* ctx){
+    assert(!ctx || ctx == &g_poolCtx);
+    (void)ctx;
+}
+
 int POOL_resize(POOL_ctx* ctx, size_t numThreads) {
     (void)ctx; (void)numThreads;
     return 0;
@@ -341,7 +362,7 @@ int POOL_tryAdd(POOL_ctx* ctx, POOL_function function, void* opaque) {
     return 1;
 }
 
-size_t POOL_sizeof(POOL_ctx* ctx) {
+size_t POOL_sizeof(const POOL_ctx* ctx) {
     if (ctx==NULL) return 0;  /* supports sizeof NULL */
     assert(ctx == &g_poolCtx);
     return sizeof(*ctx);

data/contrib/zstd/lib/common/pool.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -38,10 +38,16 @@ POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize,
  */
 void POOL_free(POOL_ctx* ctx);
 
+
+/*! POOL_joinJobs() :
+ *  Waits for all queued jobs to finish executing.
+ */
+void POOL_joinJobs(POOL_ctx* ctx);
+
 /*! POOL_resize() :
  *  Expands or shrinks pool's number of threads.
  *  This is more efficient than releasing + creating a new context,
- *  since it tries to preserve and re-use existing threads.
+ *  since it tries to preserve and reuse existing threads.
  * `numThreads` must be at least 1.
  * @return : 0 when resize was successful,
  *           !0 (typically 1) if there is an error.
@@ -53,7 +59,7 @@ int POOL_resize(POOL_ctx* ctx, size_t numThreads);
  * @return threadpool memory usage
  *  note : compatible with NULL (returns 0 in this case)
  */
-size_t POOL_sizeof(POOL_ctx* ctx);
+size_t POOL_sizeof(const POOL_ctx* ctx);
 
 /*! POOL_function :
  *  The function type that can be added to a thread pool.
@@ -70,7 +76,7 @@ void POOL_add(POOL_ctx* ctx, POOL_function function, void* opaque);
 
 
 /*! POOL_tryAdd() :
- *  Add the job `function(opaque)` to thread pool _if_ a worker is available.
+ *  Add the job `function(opaque)` to thread pool _if_ a queue slot is available.
  *  Returns immediately even if not (does not block).
  * @return : 1 if successful, 0 if not.
  */

data/contrib/zstd/lib/common/portability_macros.h

@@ -0,0 +1,158 @@
+/*
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_PORTABILITY_MACROS_H
+#define ZSTD_PORTABILITY_MACROS_H
+
+/**
+ * This header file contains macro definitions to support portability.
+ * This header is shared between C and ASM code, so it MUST only
+ * contain macro definitions. It MUST not contain any C code.
+ *
+ * This header ONLY defines macros to detect platforms/feature support.
+ *
+ */
+
+
+/* compat. with non-clang compilers */
+#ifndef __has_attribute
+  #define __has_attribute(x) 0
+#endif
+
+/* compat. with non-clang compilers */
+#ifndef __has_builtin
+#  define __has_builtin(x) 0
+#endif
+
+/* compat. with non-clang compilers */
+#ifndef __has_feature
+#  define __has_feature(x) 0
+#endif
+
+/* detects whether we are being compiled under msan */
+#ifndef ZSTD_MEMORY_SANITIZER
+#  if __has_feature(memory_sanitizer)
+#    define ZSTD_MEMORY_SANITIZER 1
+#  else
+#    define ZSTD_MEMORY_SANITIZER 0
+#  endif
+#endif
+
+/* detects whether we are being compiled under asan */
+#ifndef ZSTD_ADDRESS_SANITIZER
+#  if __has_feature(address_sanitizer)
+#    define ZSTD_ADDRESS_SANITIZER 1
+#  elif defined(__SANITIZE_ADDRESS__)
+#    define ZSTD_ADDRESS_SANITIZER 1
+#  else
+#    define ZSTD_ADDRESS_SANITIZER 0
+#  endif
+#endif
+
+/* detects whether we are being compiled under dfsan */
+#ifndef ZSTD_DATAFLOW_SANITIZER
+# if __has_feature(dataflow_sanitizer)
+#  define ZSTD_DATAFLOW_SANITIZER 1
+# else
+#  define ZSTD_DATAFLOW_SANITIZER 0
+# endif
+#endif
+
+/* Mark the internal assembly functions as hidden  */
+#ifdef __ELF__
+# define ZSTD_HIDE_ASM_FUNCTION(func) .hidden func
+#elif defined(__APPLE__)
+# define ZSTD_HIDE_ASM_FUNCTION(func) .private_extern func
+#else
+# define ZSTD_HIDE_ASM_FUNCTION(func)
+#endif
+
+/* Enable runtime BMI2 dispatch based on the CPU.
+ * Enabled for clang & gcc >=4.8 on x86 when BMI2 isn't enabled by default.
+ */
+#ifndef DYNAMIC_BMI2
+  #if ((defined(__clang__) && __has_attribute(__target__)) \
+      || (defined(__GNUC__) \
+          && (__GNUC__ >= 5 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)))) \
+      && (defined(__x86_64__) || defined(_M_X64)) \
+      && !defined(__BMI2__)
+  #  define DYNAMIC_BMI2 1
+  #else
+  #  define DYNAMIC_BMI2 0
+  #endif
+#endif
+
+/**
+ * Only enable assembly for GNUC compatible compilers,
+ * because other platforms may not support GAS assembly syntax.
+ *
+ * Only enable assembly for Linux / MacOS, other platforms may
+ * work, but they haven't been tested. This could likely be
+ * extended to BSD systems.
+ *
+ * Disable assembly when MSAN is enabled, because MSAN requires
+ * 100% of code to be instrumented to work.
+ */
+#if defined(__GNUC__)
+#  if defined(__linux__) || defined(__linux) || defined(__APPLE__)
+#    if ZSTD_MEMORY_SANITIZER
+#      define ZSTD_ASM_SUPPORTED 0
+#    elif ZSTD_DATAFLOW_SANITIZER
+#      define ZSTD_ASM_SUPPORTED 0
+#    else
+#      define ZSTD_ASM_SUPPORTED 1
+#    endif
+#  else
+#    define ZSTD_ASM_SUPPORTED 0
+#  endif
+#else
+#  define ZSTD_ASM_SUPPORTED 0
+#endif
+
+/**
+ * Determines whether we should enable assembly for x86-64
+ * with BMI2.
+ *
+ * Enable if all of the following conditions hold:
+ * - ASM hasn't been explicitly disabled by defining ZSTD_DISABLE_ASM
+ * - Assembly is supported
+ * - We are compiling for x86-64 and either:
+ *   - DYNAMIC_BMI2 is enabled
+ *   - BMI2 is supported at compile time
+ */
+#if !defined(ZSTD_DISABLE_ASM) &&                                 \
+    ZSTD_ASM_SUPPORTED &&                                         \
+    defined(__x86_64__) &&                                        \
+    (DYNAMIC_BMI2 || defined(__BMI2__))
+# define ZSTD_ENABLE_ASM_X86_64_BMI2 1
+#else
+# define ZSTD_ENABLE_ASM_X86_64_BMI2 0
+#endif
+
+/*
+ * For x86 ELF targets, add .note.gnu.property section for Intel CET in
+ * assembly sources when CET is enabled.
+ *
+ * Additionally, any function that may be called indirectly must begin
+ * with ZSTD_CET_ENDBRANCH.
+ */
+#if defined(__ELF__) && (defined(__x86_64__) || defined(__i386__)) \
+    && defined(__has_include)
+# if __has_include(<cet.h>)
+#  include <cet.h>
+#  define ZSTD_CET_ENDBRANCH _CET_ENDBR
+# endif
+#endif
+
+#ifndef ZSTD_CET_ENDBRANCH
+# define ZSTD_CET_ENDBRANCH
+#endif
+
+#endif /* ZSTD_PORTABILITY_MACROS_H */

data/contrib/zstd/lib/common/threading.c

@@ -23,8 +23,7 @@ int g_ZSTD_threading_useless_symbol;
 #if defined(ZSTD_MULTITHREAD) && defined(_WIN32)
 
 /**
- * Windows minimalist Pthread Wrapper, based on :
- * http://www.cse.wustl.edu/~schmidt/win32-cv-1.html
+ * Windows minimalist Pthread Wrapper
  */
 
 
@@ -35,37 +34,94 @@ int g_ZSTD_threading_useless_symbol;
 
 /* ===  Implementation  === */
 
+typedef struct {
+    void* (*start_routine)(void*);
+    void* arg;
+    int initialized;
+    ZSTD_pthread_cond_t initialized_cond;
+    ZSTD_pthread_mutex_t initialized_mutex;
+} ZSTD_thread_params_t;
+
 static unsigned __stdcall worker(void *arg)
 {
-    ZSTD_pthread_t* const thread = (ZSTD_pthread_t*) arg;
-    thread->arg = thread->start_routine(thread->arg);
+    void* (*start_routine)(void*);
+    void* thread_arg;
+
+    /* Initialized thread_arg and start_routine and signal main thread that we don't need it
+     * to wait any longer.
+     */
+    {
+        ZSTD_thread_params_t*  thread_param = (ZSTD_thread_params_t*)arg;
+        thread_arg = thread_param->arg;
+        start_routine = thread_param->start_routine;
+
+        /* Signal main thread that we are running and do not depend on its memory anymore */
+        ZSTD_pthread_mutex_lock(&thread_param->initialized_mutex);
+        thread_param->initialized = 1;
+        ZSTD_pthread_cond_signal(&thread_param->initialized_cond);
+        ZSTD_pthread_mutex_unlock(&thread_param->initialized_mutex);
+    }
+
+    start_routine(thread_arg);
+
     return 0;
 }
 
 int ZSTD_pthread_create(ZSTD_pthread_t* thread, const void* unused,
             void* (*start_routine) (void*), void* arg)
 {
+    ZSTD_thread_params_t thread_param;
     (void)unused;
-    thread->arg = arg;
-    thread->start_routine = start_routine;
-    thread->handle = (HANDLE) _beginthreadex(NULL, 0, worker, thread, 0, NULL);
 
-    if (!thread->handle)
+    if (thread==NULL) return -1;
+    *thread = NULL;
+
+    thread_param.start_routine = start_routine;
+    thread_param.arg = arg;
+    thread_param.initialized = 0;
+
+    /* Setup thread initialization synchronization */
+    if(ZSTD_pthread_cond_init(&thread_param.initialized_cond, NULL)) {
+        /* Should never happen on Windows */
+        return -1;
+    }
+    if(ZSTD_pthread_mutex_init(&thread_param.initialized_mutex, NULL)) {
+        /* Should never happen on Windows */
+        ZSTD_pthread_cond_destroy(&thread_param.initialized_cond);
+        return -1;
+    }
+
+    /* Spawn thread */
+    *thread = (HANDLE)_beginthreadex(NULL, 0, worker, &thread_param, 0, NULL);
+    if (*thread==NULL) {
+        ZSTD_pthread_mutex_destroy(&thread_param.initialized_mutex);
+        ZSTD_pthread_cond_destroy(&thread_param.initialized_cond);
         return errno;
-    else
-        return 0;
+    }
+
+    /* Wait for thread to be initialized */
+    ZSTD_pthread_mutex_lock(&thread_param.initialized_mutex);
+    while(!thread_param.initialized) {
+        ZSTD_pthread_cond_wait(&thread_param.initialized_cond, &thread_param.initialized_mutex);
+    }
+    ZSTD_pthread_mutex_unlock(&thread_param.initialized_mutex);
+    ZSTD_pthread_mutex_destroy(&thread_param.initialized_mutex);
+    ZSTD_pthread_cond_destroy(&thread_param.initialized_cond);
+
+    return 0;
 }
 
-int ZSTD_pthread_join(ZSTD_pthread_t thread, void **value_ptr)
+int ZSTD_pthread_join(ZSTD_pthread_t thread)
 {
     DWORD result;
 
-    if (!thread.handle) return 0;
+    if (!thread) return 0;
+
+    result = WaitForSingleObject(thread, INFINITE);
+    CloseHandle(thread);
 
-    result = WaitForSingleObject(thread.handle, INFINITE);
     switch (result) {
     case WAIT_OBJECT_0:
-        if (value_ptr) *value_ptr = thread.arg;
         return 0;
     case WAIT_ABANDONED:
         return EINVAL;
@@ -83,6 +139,7 @@ int ZSTD_pthread_join(ZSTD_pthread_t thread, void **value_ptr)
 
 int ZSTD_pthread_mutex_init(ZSTD_pthread_mutex_t* mutex, pthread_mutexattr_t const* attr)
 {
+    assert(mutex != NULL);
     *mutex = (pthread_mutex_t*)ZSTD_malloc(sizeof(pthread_mutex_t));
     if (!*mutex)
         return 1;
@@ -91,6 +148,7 @@ int ZSTD_pthread_mutex_init(ZSTD_pthread_mutex_t* mutex, pthread_mutexattr_t con
 
 int ZSTD_pthread_mutex_destroy(ZSTD_pthread_mutex_t* mutex)
 {
+    assert(mutex != NULL);
     if (!*mutex)
         return 0;
     {
@@ -102,6 +160,7 @@ int ZSTD_pthread_mutex_destroy(ZSTD_pthread_mutex_t* mutex)
 
 int ZSTD_pthread_cond_init(ZSTD_pthread_cond_t* cond, pthread_condattr_t const* attr)
 {
+    assert(cond != NULL);
     *cond = (pthread_cond_t*)ZSTD_malloc(sizeof(pthread_cond_t));
     if (!*cond)
         return 1;
@@ -110,6 +169,7 @@ int ZSTD_pthread_cond_init(ZSTD_pthread_cond_t* cond, pthread_condattr_t const*
 
 int ZSTD_pthread_cond_destroy(ZSTD_pthread_cond_t* cond)
 {
+    assert(cond != NULL);
     if (!*cond)
         return 0;
     {

data/contrib/zstd/lib/common/threading.h

@@ -23,8 +23,7 @@ extern "C" {
 #if defined(ZSTD_MULTITHREAD) && defined(_WIN32)
 
 /**
- * Windows minimalist Pthread Wrapper, based on :
- * http://www.cse.wustl.edu/~schmidt/win32-cv-1.html
+ * Windows minimalist Pthread Wrapper
  */
 #ifdef WINVER
 #  undef WINVER
@@ -62,16 +61,12 @@ extern "C" {
 #define ZSTD_pthread_cond_broadcast(a)  WakeAllConditionVariable((a))
 
 /* ZSTD_pthread_create() and ZSTD_pthread_join() */
-typedef struct {
-    HANDLE handle;
-    void* (*start_routine)(void*);
-    void* arg;
-} ZSTD_pthread_t;
+typedef HANDLE ZSTD_pthread_t;
 
 int ZSTD_pthread_create(ZSTD_pthread_t* thread, const void* unused,
                    void* (*start_routine) (void*), void* arg);
 
-int ZSTD_pthread_join(ZSTD_pthread_t thread, void** value_ptr);
+int ZSTD_pthread_join(ZSTD_pthread_t thread);
 
 /**
  * add here more wrappers as required
@@ -99,7 +94,7 @@ int ZSTD_pthread_join(ZSTD_pthread_t thread, void** value_ptr);
 
 #define ZSTD_pthread_t                  pthread_t
 #define ZSTD_pthread_create(a, b, c, d) pthread_create((a), (b), (c), (d))
-#define ZSTD_pthread_join(a, b)         pthread_join((a),(b))
+#define ZSTD_pthread_join(a)         pthread_join((a),NULL)
 
 #else /* DEBUGLEVEL >= 1 */
 
@@ -124,7 +119,7 @@ int ZSTD_pthread_cond_destroy(ZSTD_pthread_cond_t* cond);
 
 #define ZSTD_pthread_t                  pthread_t
 #define ZSTD_pthread_create(a, b, c, d) pthread_create((a), (b), (c), (d))
-#define ZSTD_pthread_join(a, b)         pthread_join((a),(b))
+#define ZSTD_pthread_join(a)         pthread_join((a),NULL)
 
 #endif