[mono][interp] Fix concurrency issues with publishing transfomed imethod (#87555)

* [mono][utils] Move all membar code to single file

* [mono][utils] Redefine write and read memory barriers

Before this change they were doing a full memory barrier, regardless of architecture. We have a beginning of more precise implementation via the *_FENCE defines. Implement mono_memory_write_barrier and mono_memory_read_barrier reusing these defines instead.

The only consequence of this change is that, on x86 and amd64, `mono_memory_write_barrier` and `mono_memory_read_barrier` become a compiler barrier instead of a full mfence.

* [mono][interp] Fix concurrency issues with publishing transfomed imethod

When publishing a transformed InterpMethod*, we first set all relevant fields (like `code`, `alloca_size` etc), we execute a write barrier and finally we set the `transformed` flag. On relaxed memory arches we need to have a read barrier on the consumer, since there is no data dependency between `transformed` and the other fields of `InterpMethod`.

On arm this change does a full barrier (we could get away with just a load acquire but we haven't yet added support for emitting this in the runtime). Still, this doesn't seem to introduce a heavy perf penalty (on my arm64 M1) but we can revisit if necessary. On x86/amd64 this is a compiler barrier so it should have no impact. WASM is single threaded for now.

Author: BrzVlad

src/mono/mono/metadata/domain.c

@@ -22,7 +22,6 @@
 #include <mono/utils/atomic.h>
 #include <mono/utils/mono-compiler.h>
 #include <mono/utils/mono-logger-internals.h>
-#include <mono/utils/mono-membar.h>
 #include <mono/utils/mono-counters.h>
 #include <mono/utils/hazard-pointer.h>
 #include <mono/utils/mono-tls.h>

src/mono/mono/metadata/icall.c

@@ -38,7 +38,6 @@
 #endif
 
 #include "mono/metadata/icall-internals.h"
-#include "mono/utils/mono-membar.h"
 #include <mono/metadata/object.h>
 #include <mono/metadata/threads.h>
 #include <mono/metadata/threads-types.h>

src/mono/mono/metadata/jit-info.c

@@ -23,7 +23,7 @@
 #include <mono/utils/atomic.h>
 #include <mono/utils/mono-compiler.h>
 #include <mono/utils/mono-logger-internals.h>
-#include <mono/utils/mono-membar.h>
+#include <mono/utils/mono-memory-model.h>
 #include <mono/utils/hazard-pointer.h>
 #include <mono/utils/mono-tls.h>
 #include <mono/utils/mono-mmap.h>

src/mono/mono/metadata/loader.c

@@ -46,7 +46,7 @@
 #include <mono/metadata/jit-info.h>
 #include <mono/utils/mono-logger-internals.h>
 #include <mono/utils/mono-dl.h>
-#include <mono/utils/mono-membar.h>
+#include <mono/utils/mono-memory-model.h>
 #include <mono/utils/mono-counters.h>
 #include <mono/utils/mono-error-internals.h>
 #include <mono/utils/mono-tls.h>

src/mono/mono/metadata/property-bag.c

@@ -9,7 +9,7 @@
  */
 #include <mono/metadata/property-bag.h>
 #include <mono/utils/atomic.h>
-#include <mono/utils/mono-membar.h>
+#include <mono/utils/mono-memory-model.h>
 
 /*
  * mono_property_bag_get:

src/mono/mono/metadata/reflection.c

@@ -13,7 +13,7 @@
  * Licensed under the MIT license. See LICENSE file in the project root for full license information.
  */
 #include <config.h>
-#include "mono/utils/mono-membar.h"
+#include "mono/utils/mono-memory-model.h"
 #include "mono/metadata/assembly-internals.h"
 #include "mono/metadata/reflection-internals.h"
 #include "mono/metadata/tabledefs.h"

src/mono/mono/metadata/string-icalls.c

@@ -14,7 +14,7 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
-#include "mono/utils/mono-membar.h"
+#include "mono/utils/mono-memory-model.h"
 #include <mono/metadata/string-icalls.h>
 #include <mono/metadata/class-internals.h>
 #include <mono/metadata/appdomain.h>

src/mono/mono/metadata/threads-types.h

@@ -19,7 +19,7 @@
 #include <mono/metadata/object.h>
 #include "mono/metadata/handle.h"
 #include "mono/utils/mono-compiler.h"
-#include "mono/utils/mono-membar.h"
+#include "mono/utils/mono-memory-model.h"
 #include "mono/utils/mono-threads.h"
 #include "mono/metadata/class-internals.h"
 #include <mono/metadata/icalls.h>

src/mono/mono/metadata/threads.c

@@ -36,7 +36,7 @@
 #include <mono/utils/monobitset.h>
 #include <mono/utils/mono-compiler.h>
 #include <mono/utils/mono-mmap.h>
-#include <mono/utils/mono-membar.h>
+#include <mono/utils/mono-memory-model.h>
 #include <mono/utils/mono-time.h>
 #include <mono/utils/mono-threads.h>
 #include <mono/utils/mono-threads-coop.h>

src/mono/mono/mini/interp/interp.c

@@ -27,7 +27,7 @@
 #include <mono/utils/mono-logger-internals.h>
 #include <mono/utils/mono-tls-inline.h>
 #include <mono/utils/mono-threads.h>
-#include <mono/utils/mono-membar.h>
+#include <mono/utils/mono-memory-model.h>
 
 #ifdef HAVE_ALLOCA_H
 #   include <alloca.h>
@@ -3606,6 +3606,8 @@ method_entry (ThreadContext *context, InterpFrame *frame,
 			frame->stack = (stackval*)context->stack_pointer;
 			return slow;
 		}
+	} else {
+		mono_memory_read_barrier ();
 	}
 
 	return slow;