Rewrite ethread library

Large parts of the ethread library have been rewritten. The
ethread library is an Erlang runtime system internal, portable
thread library used by the runtime system itself.

Most notable improvement is a reader optimized rwlock
implementation which dramatically improve the performance of
read-lock/read-unlock operations on multi processor systems by
avoiding ping-ponging of the rwlock cache lines. The reader
optimized rwlock implementation is used by miscellaneous
rwlocks in the runtime system that are known to be read-locked
frequently, and can be enabled on ETS tables by passing the
`{read_concurrency, true}' option upon table creation. See the
documentation of `ets:new/2' for more information.

The ethread library can now also use the libatomic_ops library
for atomic memory accesses. This makes it possible for the
Erlang runtime system to utilize optimized atomic operations
on more platforms than before. Use the
`--with-libatomic_ops=PATH' configure command line argument
when specifying where the libatomic_ops installation is
located. The libatomic_ops library can be downloaded from:
http://www.hpl.hp.com/research/linux/atomic_ops/

The changed API of the ethread library has also caused
modifications in the Erlang runtime system. Preparations for
the to come "delayed deallocation" feature has also been done
since it depends on the ethread library.

Note: When building for x86, the ethread library will now use
instructions that first appeared on the pentium 4 processor. If
you want the runtime system to be compatible with older
processors (back to 486) you need to pass the
`--enable-ethread-pre-pentium4-compatibility' configure command
line argument when configuring the system.

3 files changed, 337 insertions, 0 deletions

diff --git a/erts/include/internal/win/ethr_atomic.h b/erts/include/internal/win/ethr_atomic.h
new file mode 100644
index 0000000000..500459dd6c
--- /dev/null
+++ b/erts/include/internal/win/ethr_atomic.h
@@ -0,0 +1,244 @@
+/*
+ * %CopyrightBegin%
+ *
+ * Copyright Ericsson AB 2010. All Rights Reserved.
+ *
+ * The contents of this file are subject to the Erlang Public License,
+ * Version 1.1, (the "License"); you may not use this file except in
+ * compliance with the License. You should have received a copy of the
+ * Erlang Public License along with this software. If not, it can be
+ * retrieved online at http://www.erlang.org/.
+ *
+ * Software distributed under the License is distributed on an "AS IS"
+ * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
+ * the License for the specific language governing rights and limitations
+ * under the License.
+ *
+ * %CopyrightEnd%
+ */
+
+/*
+ * Description: Native atomics ethread support when using VC++
+ * Author: Rickard Green
+ */
+
+#ifndef ETHR_WIN_ATOMIC_H__
+#define ETHR_WIN_ATOMIC_H__
+
+#ifdef _MSC_VER
+#  if _MSC_VER < 1300
+#    define ETHR_IMMED_ATOMIC_SET_GET_SAFE__ 0 /* Dont trust really old compilers */
+#  else
+#    if defined(_M_IX86)
+#      define ETHR_IMMED_ATOMIC_SET_GET_SAFE__ 1
+#    else /* I.e. IA64 */
+#      if _MSC_VER >= 1400
+#        define ETHR_IMMED_ATOMIC_SET_GET_SAFE__ 1
+#      else
+#        define ETHR_IMMED_ATOMIC_SET_GET_SAFE__ 0
+#      endif
+#    endif
+#  endif
+#  if _MSC_VER >= 1400
+#    include <intrin.h>
+#    undef ETHR_COMPILER_BARRIER
+#    define ETHR_COMPILER_BARRIER _ReadWriteBarrier()
+#  endif
+#pragma intrinsic(_ReadWriteBarrier)
+#pragma intrinsic(_InterlockedAnd)
+#pragma intrinsic(_InterlockedOr)
+#else
+#    define ETHR_IMMED_ATOMIC_SET_GET_SAFE__ 0
+#endif
+
+/*
+ * No configure test checking for _Interlocked*_{acq,rel} and
+ * Interlocked*{Acquire,Release} have been written yet...
+ *
+ * Note, that these are pure optimizations for the itanium
+ * processor.
+ */
+
+#ifdef ETHR_HAVE_INTERLOCKEDCOMPAREEXCHANGE_ACQ
+#pragma intrinsic(_InterlockedCompareExchange_acq)
+#endif
+#ifdef ETHR_HAVE_INTERLOCKEDCOMPAREEXCHANGE_REL
+#pragma intrinsic(_InterlockedCompareExchange_rel)
+#endif
+
+
+typedef struct {
+    volatile LONG value;
+} ethr_native_atomic_t;
+
+#define ETHR_MEMORY_BARRIER \
+do { \
+    volatile LONG x___ = 0; \
+    _InterlockedCompareExchange(&x___, (LONG) 1, (LONG) 0); \
+} while (0)
+
+#if defined(ETHR_TRY_INLINE_FUNCS) || defined(ETHR_AUX_IMPL__)
+
+static ETHR_INLINE void
+ethr_native_atomic_init(ethr_native_atomic_t *var, long i)
+{
+    var->value = (LONG) i;
+}
+
+static ETHR_INLINE void
+ethr_native_atomic_set(ethr_native_atomic_t *var, long i)
+{
+#if ETHR_IMMED_ATOMIC_SET_GET_SAFE__
+    var->value = (LONG) i;
+#else
+    (void) InterlockedExchange(&var->value, (LONG) i);
+#endif
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_read(ethr_native_atomic_t *var)
+{
+#if ETHR_IMMED_ATOMIC_SET_GET_SAFE__
+    return var->value;
+#else
+    return InterlockedExchangeAdd(&var->value, (LONG) 0);
+#endif
+}
+
+static ETHR_INLINE void
+ethr_native_atomic_add(ethr_native_atomic_t *var, long incr)
+{
+    (void) InterlockedExchangeAdd(&var->value, (LONG) incr);
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_add_return(ethr_native_atomic_t *var, long i)
+{
+    LONG tmp = InterlockedExchangeAdd(&var->value, (LONG) i);
+    return tmp + i;
+}
+
+static ETHR_INLINE void
+ethr_native_atomic_inc(ethr_native_atomic_t *var)
+{
+    (void) InterlockedIncrement(&var->value);
+}
+
+static ETHR_INLINE void
+ethr_native_atomic_dec(ethr_native_atomic_t *var)
+{
+    (void) InterlockedDecrement(&var->value);
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_inc_return(ethr_native_atomic_t *var)
+{
+    return (long) InterlockedIncrement(&var->value);
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_dec_return(ethr_native_atomic_t *var)
+{
+    return (long) InterlockedDecrement(&var->value);
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_and_retold(ethr_native_atomic_t *var, long mask)
+{
+    return (long) _InterlockedAnd(&var->value, mask);
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_or_retold(ethr_native_atomic_t *var, long mask)
+{
+    return (long) _InterlockedOr(&var->value, mask);
+}
+
+
+static ETHR_INLINE long
+ethr_native_atomic_cmpxchg(ethr_native_atomic_t *var, long new, long old)
+{
+    return (long) _InterlockedCompareExchange(&var->value, (LONG) new, (LONG) old);
+}
+
+
+static ETHR_INLINE long
+ethr_native_atomic_xchg(ethr_native_atomic_t *var, long new)
+{
+    return (long) InterlockedExchange(&var->value, (LONG) new);
+}
+
+/*
+ * Atomic ops with at least specified barriers.
+ */
+
+static ETHR_INLINE long
+ethr_native_atomic_read_acqb(ethr_native_atomic_t *var)
+{
+#ifdef ETHR_HAVE_INTERLOCKEDEXCHANGEADDACQUIRE
+    return (long) InterlockedExchangeAddAcquire(&var->value, (LONG) 0);
+#else
+    return (long) InterlockedExchangeAdd(&var->value, (LONG) 0);
+#endif
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_inc_return_acqb(ethr_native_atomic_t *var)
+{
+#ifdef ETHR_HAVE_INTERLOCKEDINCREMENTACQUIRE
+    return (long) InterlockedIncrementAcquire(&var->value);
+#else
+    return (long) InterlockedIncrement(&var->value);
+#endif
+}
+
+static ETHR_INLINE void
+ethr_native_atomic_set_relb(ethr_native_atomic_t *var, long i)
+{
+    (void) InterlockedExchange(&var->value, (LONG) i);
+}
+
+static ETHR_INLINE void
+ethr_native_atomic_dec_relb(ethr_native_atomic_t *var)
+{
+#ifdef ETHR_HAVE_INTERLOCKEDDECREMENTRELEASE
+    (void) InterlockedDecrementRelease(&var->value);
+#else
+    (void) InterlockedDecrement(&var->value);
+#endif
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_dec_return_relb(ethr_native_atomic_t *var)
+{
+#ifdef ETHR_HAVE_INTERLOCKEDDECREMENTRELEASE
+    return (long) InterlockedDecrementRelease(&var->value);
+#else
+    return (long) InterlockedDecrement(&var->value);
+#endif
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_cmpxchg_acqb(ethr_native_atomic_t *var, long new, long old)
+{
+#ifdef ETHR_HAVE_INTERLOCKEDCOMPAREEXCHANGE_ACQ
+    return (long) _InterlockedCompareExchange_acq(&var->value, (LONG) new, (LONG) old);
+#else
+    return (long) _InterlockedCompareExchange(&var->value, (LONG) new, (LONG) old);
+#endif
+}
+
+static ETHR_INLINE long
+ethr_native_atomic_cmpxchg_relb(ethr_native_atomic_t *var, long new, long old)
+{
+
+#ifdef ETHR_HAVE_INTERLOCKEDCOMPAREEXCHANGE_REL
+    return (long) _InterlockedCompareExchange_rel(&var->value, (LONG) new, (LONG) old);
+#else
+    return (long) _InterlockedCompareExchange(&var->value, (LONG) new, (LONG) old);
+#endif
+}
+
+#endif
+
+#endif
diff --git a/erts/include/internal/win/ethr_event.h b/erts/include/internal/win/ethr_event.h
new file mode 100644
index 0000000000..af57c20f91
--- /dev/null
+++ b/erts/include/internal/win/ethr_event.h
@@ -0,0 +1,62 @@
+/*
+ * %CopyrightBegin%
+ *
+ * Copyright Ericsson AB 2009-2010. All Rights Reserved.
+ *
+ * The contents of this file are subject to the Erlang Public License,
+ * Version 1.1, (the "License"); you may not use this file except in
+ * compliance with the License. You should have received a copy of the
+ * Erlang Public License along with this software. If not, it can be
+ * retrieved online at http://www.erlang.org/.
+ *
+ * Software distributed under the License is distributed on an "AS IS"
+ * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
+ * the License for the specific language governing rights and limitations
+ * under the License.
+ *
+ * %CopyrightEnd%
+ */
+
+/*
+ * Author: Rickard Green
+ */
+
+#define ETHR_EVENT_OFF_WAITER__		((LONG) -1)
+#define ETHR_EVENT_OFF__		((LONG) 1)
+#define ETHR_EVENT_ON__ 		((LONG) 0)
+
+typedef struct {
+    volatile LONG state;
+    HANDLE handle;
+} ethr_event;
+
+#if defined(ETHR_TRY_INLINE_FUNCS) || defined(ETHR_EVENT_IMPL__)
+
+static ETHR_INLINE void
+ETHR_INLINE_FUNC_NAME_(ethr_event_set)(ethr_event *e)
+{
+    /* InterlockedExchange() imply a full memory barrier which is important */
+    LONG state = InterlockedExchange(&e->state, ETHR_EVENT_ON__);
+    if (state == ETHR_EVENT_OFF_WAITER__) {
+	if (!SetEvent(e->handle))
+	    ETHR_FATAL_ERROR__(ethr_win_get_errno__());
+    }
+}
+
+static ETHR_INLINE void
+ETHR_INLINE_FUNC_NAME_(ethr_event_reset)(ethr_event *e)
+{
+    /* InterlockedExchange() imply a full memory barrier which is important */
+    InterlockedExchange(&e->state, ETHR_EVENT_OFF__);
+}
+
+#endif
+
+int ethr_event_init(ethr_event *e);
+int ethr_event_destroy(ethr_event *e);
+int ethr_event_wait(ethr_event *e);
+int ethr_event_swait(ethr_event *e, int spincount);
+#if !defined(ETHR_TRY_INLINE_FUNCS) || defined(ETHR_EVENT_IMPL__)
+void ethr_event_set(ethr_event *e);
+void ethr_event_reset(ethr_event *e);
+#endif
diff --git a/erts/include/internal/win/ethread.h b/erts/include/internal/win/ethread.h
new file mode 100644
index 0000000000..b52710f6a3
--- /dev/null
+++ b/erts/include/internal/win/ethread.h
@@ -0,0 +1,31 @@
+/*
+ * %CopyrightBegin%
+ *
+ * Copyright Ericsson AB 2010. All Rights Reserved.
+ *
+ * The contents of this file are subject to the Erlang Public License,
+ * Version 1.1, (the "License"); you may not use this file except in
+ * compliance with the License. You should have received a copy of the
+ * Erlang Public License along with this software. If not, it can be
+ * retrieved online at http://www.erlang.org/.
+ *
+ * Software distributed under the License is distributed on an "AS IS"
+ * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
+ * the License for the specific language governing rights and limitations
+ * under the License.
+ *
+ * %CopyrightEnd%
+ */
+
+/*
+ * Description: Native atomic and spinlock ethread support when using VC++
+ * Author: Rickard Green
+ */
+
+#ifndef ETHREAD_WIN_H__
+#define ETHREAD_WIN_H__
+
+#include "ethr_atomic.h"
+#define ETHR_HAVE_NATIVE_ATOMICS 1
+
+#endif