C++11 patchset 3: remove Atomic/GenericAtomic and use std::atomic (#5906)
authorLoïc Blot <nerzhul@users.noreply.github.com>
Tue, 6 Jun 2017 12:34:14 +0000 (14:34 +0200)
committerGitHub <noreply@github.com>
Tue, 6 Jun 2017 12:34:14 +0000 (14:34 +0200)
src/environment.h
src/threading/atomic.h [deleted file]
src/threading/thread.h
src/unittest/test_threading.cpp

index 5a40ff31ab57332854a9d8be1ea4c5ae066507fd..ff3942599a34f0f0ff7bfd8dba91f1978ec4fad1 100644 (file)
@@ -33,11 +33,11 @@ with this program; if not, write to the Free Software Foundation, Inc.,
 #include <list>
 #include <queue>
 #include <map>
+#include <atomic>
 #include "irr_v3d.h"
 #include "activeobject.h"
 #include "util/numeric.h"
 #include "threading/mutex.h"
-#include "threading/atomic.h"
 #include "network/networkprotocol.h" // for AccessDeniedCode
 
 class IGameDef;
@@ -81,7 +81,7 @@ public:
        IGameDef *getGameDef() { return m_gamedef; }
 
 protected:
-       GenericAtomic<float> m_time_of_day_speed;
+       std::atomic<float> m_time_of_day_speed;
 
        /*
         * Below: values managed by m_time_lock
@@ -98,7 +98,7 @@ protected:
        u32 m_day_night_ratio_override;
        // Days from the server start, accounts for time shift
        // in game (e.g. /time or bed usage)
-       Atomic<u32> m_day_count;
+       std::atomic<u32> m_day_count;
        /*
         * Above: values managed by m_time_lock
        */
diff --git a/src/threading/atomic.h b/src/threading/atomic.h
deleted file mode 100644 (file)
index c9a454f..0000000
+++ /dev/null
@@ -1,139 +0,0 @@
-/*
-Minetest
-Copyright (C) 2015 ShadowNinja <shadowninja@minetest.net>
-
-This program is free software; you can redistribute it and/or modify
-it under the terms of the GNU Lesser General Public License as published by
-the Free Software Foundation; either version 2.1 of the License, or
-(at your option) any later version.
-
-This program is distributed in the hope that it will be useful,
-but WITHOUT ANY WARRANTY; without even the implied warranty of
-MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-GNU Lesser General Public License for more details.
-
-You should have received a copy of the GNU Lesser General Public License along
-with this program; if not, write to the Free Software Foundation, Inc.,
-51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
-*/
-
-#ifndef THREADING_ATOMIC_H
-#define THREADING_ATOMIC_H
-
-
-#if __cplusplus >= 201103L
-       #include <atomic>
-       template<typename T> using Atomic = std::atomic<T>;
-       template<typename T> using GenericAtomic = std::atomic<T>;
-#else
-
-#define GCC_VERSION   (__GNUC__        * 100 + __GNUC_MINOR__)
-#define CLANG_VERSION (__clang_major__ * 100 + __clang_minor__)
-#if GCC_VERSION >= 407 || CLANG_VERSION >= 302
-       #define ATOMIC_LOAD_GENERIC(T, v) do {               \
-               T _val;                                          \
-                __atomic_load(&(v), &(_val), __ATOMIC_SEQ_CST); \
-               return _val;                                     \
-       } while(0)
-       #define ATOMIC_LOAD(T, v)        return __atomic_load_n    (&(v),       __ATOMIC_SEQ_CST)
-       #define ATOMIC_STORE(T, v, x)           __atomic_store     (&(v), &(x), __ATOMIC_SEQ_CST); return x
-       #define ATOMIC_EXCHANGE(T, v, x) return __atomic_exchange  (&(v), &(x), __ATOMIC_SEQ_CST)
-       #define ATOMIC_ADD_EQ(T, v, x)   return __atomic_add_fetch (&(v), (x),  __ATOMIC_SEQ_CST)
-       #define ATOMIC_SUB_EQ(T, v, x)   return __atomic_sub_fetch (&(v), (x),  __ATOMIC_SEQ_CST)
-       #define ATOMIC_POST_INC(T, v)    return __atomic_fetch_add (&(v), 1,    __ATOMIC_SEQ_CST)
-       #define ATOMIC_POST_DEC(T, v)    return __atomic_fetch_sub (&(v), 1,    __ATOMIC_SEQ_CST)
-       #define ATOMIC_CAS(T, v, e, d)   return __atomic_compare_exchange(&(v), &(e), &(d), \
-               false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)
-#else
-       #define ATOMIC_USE_LOCK
-       #include "threading/mutex.h"
-
-       #define ATOMIC_LOCK_OP(T, op) do { \
-                       m_mutex.lock();            \
-                       T _val = (op);             \
-                       m_mutex.unlock();          \
-                       return _val;               \
-               } while (0)
-       #define ATOMIC_LOCK_CAS(T, v, e, d) do { \
-                       m_mutex.lock();                  \
-                       bool _eq = (v == e);             \
-                       if (_eq)                         \
-                               v = d;                       \
-                       m_mutex.unlock();                \
-                       return _eq;                      \
-               } while (0)
-       #define ATOMIC_LOAD(T, v) ATOMIC_LOCK_OP(T, v)
-       #define ATOMIC_LOAD_GENERIC(T, v) ATOMIC_LOAD(T, v)
-       #define ATOMIC_STORE(T, v, x) ATOMIC_LOCK_OP(T, v = x)
-       #define ATOMIC_EXCHANGE(T, v, x) do { \
-                       m_mutex.lock();               \
-                       T _val = v;                   \
-                       v = x;                        \
-                       m_mutex.unlock();             \
-                       return _val;                  \
-               } while (0)
-       #if GCC_VERSION >= 401
-               #define ATOMIC_ADD_EQ(T, v, x) return __sync_add_and_fetch(&(v), (x))
-               #define ATOMIC_SUB_EQ(T, v, x) return __sync_sub_and_fetch(&(v), (x))
-               #define ATOMIC_POST_INC(T, v)  return __sync_fetch_and_add(&(v), 1)
-               #define ATOMIC_POST_DEC(T, v)  return __sync_fetch_and_sub(&(v), 1)
-               #define ATOMIC_CAS(T, v, e, d) return __sync_bool_compare_and_swap(&(v), &(e), (d))
-       #else
-               #define ATOMIC_ADD_EQ(T, v, x) ATOMIC_LOCK_OP(T, v += x)
-               #define ATOMIC_SUB_EQ(T, v, x) ATOMIC_LOCK_OP(T, v -= x)
-               #define ATOMIC_POST_INC(T, v)  ATOMIC_LOCK_OP(T, v++)
-               #define ATOMIC_POST_DEC(T, v)  ATOMIC_LOCK_OP(T, v--)
-               #define ATOMIC_CAS(T, v, e, d) ATOMIC_LOCK_CAS(T, v, e, d)
-       #endif
-#endif
-
-// For usage with integral types.
-template<typename T>
-class Atomic {
-public:
-       Atomic(const T &v = 0) : m_val(v) {}
-
-       operator T () { ATOMIC_LOAD(T, m_val); }
-
-       T exchange(T x)  { ATOMIC_EXCHANGE(T, m_val, x); }
-       bool compare_exchange_strong(T &expected, T desired) { ATOMIC_CAS(T, m_val, expected, desired); }
-
-       T operator =  (T x) { ATOMIC_STORE(T, m_val, x); }
-       T operator += (T x) { ATOMIC_ADD_EQ(T, m_val, x); }
-       T operator -= (T x) { ATOMIC_SUB_EQ(T, m_val, x); }
-       T operator ++ ()    { return *this += 1; }
-       T operator -- ()    { return *this -= 1; }
-       T operator ++ (int) { ATOMIC_POST_INC(T, m_val); }
-       T operator -- (int) { ATOMIC_POST_DEC(T, m_val); }
-private:
-       T m_val;
-#ifdef ATOMIC_USE_LOCK
-       Mutex m_mutex;
-#endif
-};
-
-// For usage with non-integral types like float for example.
-// Needed because the other operations aren't provided by gcc
-// for non-integral types:
-// https://gcc.gnu.org/onlinedocs/gcc-4.7.0/gcc/_005f_005fatomic-Builtins.html
-template<typename T>
-class GenericAtomic {
-public:
-       GenericAtomic(const T &v = 0) : m_val(v) {}
-
-       operator T () { ATOMIC_LOAD_GENERIC(T, m_val); }
-
-       T exchange(T x)  { ATOMIC_EXCHANGE(T, m_val, x); }
-       bool compare_exchange_strong(T &expected, T desired) { ATOMIC_CAS(T, m_val, expected, desired); }
-
-       T operator = (T x) { ATOMIC_STORE(T, m_val, x); }
-private:
-       T m_val;
-#ifdef ATOMIC_USE_LOCK
-       Mutex m_mutex;
-#endif
-};
-
-#endif  // C++11
-
-#endif
index 4785d3e039477cbe29109e533443c35c5761ceb5..671a9be0b973eb8ddc8e2b163b8d4e43523f2545 100644 (file)
@@ -27,11 +27,12 @@ DEALINGS IN THE SOFTWARE.
 #define THREADING_THREAD_H
 
 #include "util/basic_macros.h"
-#include "threading/atomic.h"
 #include "threading/mutex.h"
 #include "threads.h"
 
 #include <string>
+#include <atomic>
+
 #ifdef _AIX
        #include <sys/thread.h> // for tid_t
 #endif
@@ -150,8 +151,8 @@ protected:
 private:
        void *m_retval;
        bool m_joinable;
-       Atomic<bool> m_request_stop;
-       Atomic<bool> m_running;
+       std::atomic<bool> m_request_stop;
+       std::atomic<bool> m_running;
        Mutex m_mutex;
        Mutex m_start_finished_mutex;
 
index e1e1d366014c1e45654a03d42509b7d70ef4ee98..03d7765394a0f49004562e5815f94c50508163cd 100644 (file)
@@ -19,7 +19,7 @@ with this program; if not, write to the Free Software Foundation, Inc.,
 
 #include "test.h"
 
-#include "threading/atomic.h"
+#include <atomic>
 #include "threading/semaphore.h"
 #include "threading/thread.h"
 
@@ -137,7 +137,7 @@ void TestThreading::testThreadKill()
 
 class AtomicTestThread : public Thread {
 public:
-       AtomicTestThread(Atomic<u32> &v, Semaphore &trigger) :
+       AtomicTestThread(std::atomic<u32> &v, Semaphore &trigger) :
                Thread("AtomicTest"),
                val(v),
                trigger(trigger)
@@ -153,14 +153,14 @@ private:
                return NULL;
        }
 
-       Atomic<u32> &val;
+       std::atomic<u32> &val;
        Semaphore &trigger;
 };
 
 
 void TestThreading::testAtomicSemaphoreThread()
 {
-       Atomic<u32> val;
+       std::atomic<u32> val;
        val = 0;
        Semaphore trigger;
        static const u8 num_threads = 4;