ARM exclusive monitor and multicore fixes (#589)

* Implement ARM exclusive load/store with compare exchange insts, and enable multicore by default

* Fix comment typo

* Support Linux and OSX on MemoryAlloc and CompareExchange128, some cleanup

* Use intel syntax on assembly code

* Adjust identation

* Add CPUID check and fix exclusive reservation granule size

* Update schema multicore scheduling default value

* Make the cpu id check code lower case aswell

4 files changed, 41 insertions, 109 deletions

diff --git a/Ryujinx.HLE/HOS/Kernel/Threading/HleScheduler.cs b/Ryujinx.HLE/HOS/Kernel/Threading/HleScheduler.cs
index 835c2a2f..d5dbb4d8 100644
--- a/Ryujinx.HLE/HOS/Kernel/Threading/HleScheduler.cs
+++ b/Ryujinx.HLE/HOS/Kernel/Threading/HleScheduler.cs
@@ -92,8 +92,6 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
                         if (coreContext.CurrentThread != null)
                         {
-                            coreContext.CurrentThread.ClearExclusive();
-
                             CoreManager.Set(coreContext.CurrentThread.Context.Work);
 
                             coreContext.CurrentThread.Context.Execute();
diff --git a/Ryujinx.HLE/HOS/Kernel/Threading/KAddressArbiter.cs b/Ryujinx.HLE/HOS/Kernel/Threading/KAddressArbiter.cs
index faeea5c5..b11df61e 100644
--- a/Ryujinx.HLE/HOS/Kernel/Threading/KAddressArbiter.cs
+++ b/Ryujinx.HLE/HOS/Kernel/Threading/KAddressArbiter.cs
@@ -228,43 +228,31 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
             KProcess currentProcess = _system.Scheduler.GetCurrentProcess();
 
-            currentProcess.CpuMemory.SetExclusive(0, (long)address);
+            int mutexValue, newMutexValue;
 
-            if (!KernelTransfer.UserToKernelInt32(_system, address, out int mutexValue))
+            do
             {
-                //Invalid address.
-                currentProcess.CpuMemory.ClearExclusive(0);
-
-                requester.SignaledObj   = null;
-                requester.ObjSyncResult = KernelResult.InvalidMemState;
-
-                return null;
-            }
-
-            while (true)
-            {
-                if (currentProcess.CpuMemory.TestExclusive(0, (long)address))
+                if (!KernelTransfer.UserToKernelInt32(_system, address, out mutexValue))
                 {
-                    if (mutexValue != 0)
-                    {
-                        //Update value to indicate there is a mutex waiter now.
-                        currentProcess.CpuMemory.WriteInt32((long)address, mutexValue | HasListenersMask);
-                    }
-                    else
-                    {
-                        //No thread owning the mutex, assign to requesting thread.
-                        currentProcess.CpuMemory.WriteInt32((long)address, requester.ThreadHandleForUserMutex);
-                    }
-
-                    currentProcess.CpuMemory.ClearExclusiveForStore(0);
+                    //Invalid address.
+                    requester.SignaledObj   = null;
+                    requester.ObjSyncResult = KernelResult.InvalidMemState;
 
-                    break;
+                    return null;
                 }
 
-                currentProcess.CpuMemory.SetExclusive(0, (long)address);
-
-                mutexValue = currentProcess.CpuMemory.ReadInt32((long)address);
+                if (mutexValue != 0)
+                {
+                    //Update value to indicate there is a mutex waiter now.
+                    newMutexValue = mutexValue | HasListenersMask;
+                }
+                else
+                {
+                    //No thread owning the mutex, assign to requesting thread.
+                    newMutexValue = requester.ThreadHandleForUserMutex;
+                }
             }
+            while (!currentProcess.CpuMemory.AtomicCompareExchangeInt32((long)address, mutexValue, newMutexValue));
 
             if (mutexValue == 0)
             {
@@ -392,9 +380,6 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
             KProcess currentProcess = _system.Scheduler.GetCurrentProcess();
 
-            //If ShouldDecrement is true, do atomic decrement of the value at Address.
-            currentProcess.CpuMemory.SetExclusive(0, (long)address);
-
             if (!KernelTransfer.UserToKernelInt32(_system, address, out int currentValue))
             {
                 _system.CriticalSection.Leave();
@@ -404,25 +389,9 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
             if (shouldDecrement)
             {
-                while (currentValue < value)
-                {
-                    if (currentProcess.CpuMemory.TestExclusive(0, (long)address))
-                    {
-                        currentProcess.CpuMemory.WriteInt32((long)address, currentValue - 1);
-
-                        currentProcess.CpuMemory.ClearExclusiveForStore(0);
-
-                        break;
-                    }
-
-                    currentProcess.CpuMemory.SetExclusive(0, (long)address);
-
-                    currentValue = currentProcess.CpuMemory.ReadInt32((long)address);
-                }
+                currentValue = currentProcess.CpuMemory.AtomicDecrementInt32((long)address) + 1;
             }
 
-            currentProcess.CpuMemory.ClearExclusive(0);
-
             if (currentValue < value)
             {
                 if (timeout == 0)
@@ -511,39 +480,25 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
             KProcess currentProcess = _system.Scheduler.GetCurrentProcess();
 
-            currentProcess.CpuMemory.SetExclusive(0, (long)address);
+            int currentValue;
 
-            if (!KernelTransfer.UserToKernelInt32(_system, address, out int currentValue))
+            do
             {
-                _system.CriticalSection.Leave();
-
-                return KernelResult.InvalidMemState;
-            }
-
-            while (currentValue == value)
-            {
-                if (currentProcess.CpuMemory.TestExclusive(0, (long)address))
+                if (!KernelTransfer.UserToKernelInt32(_system, address, out currentValue))
                 {
-                    currentProcess.CpuMemory.WriteInt32((long)address, currentValue + 1);
-
-                    currentProcess.CpuMemory.ClearExclusiveForStore(0);
+                    _system.CriticalSection.Leave();
 
-                    break;
+                    return KernelResult.InvalidMemState;
                 }
 
-                currentProcess.CpuMemory.SetExclusive(0, (long)address);
-
-                currentValue = currentProcess.CpuMemory.ReadInt32((long)address);
-            }
-
-            currentProcess.CpuMemory.ClearExclusive(0);
-
-            if (currentValue != value)
-            {
-                _system.CriticalSection.Leave();
+                if (currentValue != value)
+                {
+                    _system.CriticalSection.Leave();
 
-                return KernelResult.InvalidState;
+                    return KernelResult.InvalidState;
+                }
             }
+            while (!currentProcess.CpuMemory.AtomicCompareExchangeInt32((long)address, currentValue, currentValue + 1));
 
             WakeArbiterThreads(address, count);
 
@@ -582,39 +537,25 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
             KProcess currentProcess = _system.Scheduler.GetCurrentProcess();
 
-            currentProcess.CpuMemory.SetExclusive(0, (long)address);
+            int currentValue;
 
-            if (!KernelTransfer.UserToKernelInt32(_system, address, out int currentValue))
+            do
             {
-                _system.CriticalSection.Leave();
-
-                return KernelResult.InvalidMemState;
-            }
-
-            while (currentValue == value)
-            {
-                if (currentProcess.CpuMemory.TestExclusive(0, (long)address))
+                if (!KernelTransfer.UserToKernelInt32(_system, address, out currentValue))
                 {
-                    currentProcess.CpuMemory.WriteInt32((long)address, currentValue + offset);
-
-                    currentProcess.CpuMemory.ClearExclusiveForStore(0);
+                    _system.CriticalSection.Leave();
 
-                    break;
+                    return KernelResult.InvalidMemState;
                 }
 
-                currentProcess.CpuMemory.SetExclusive(0, (long)address);
-
-                currentValue = currentProcess.CpuMemory.ReadInt32((long)address);
-            }
-
-            currentProcess.CpuMemory.ClearExclusive(0);
-
-            if (currentValue != value)
-            {
-                _system.CriticalSection.Leave();
+                if (currentValue != value)
+                {
+                    _system.CriticalSection.Leave();
 
-                return KernelResult.InvalidState;
+                    return KernelResult.InvalidState;
+                }
             }
+            while (!currentProcess.CpuMemory.AtomicCompareExchangeInt32((long)address, currentValue, currentValue + offset));
 
             WakeArbiterThreads(address, count);
 
diff --git a/Ryujinx.HLE/HOS/Kernel/Threading/KCoreContext.cs b/Ryujinx.HLE/HOS/Kernel/Threading/KCoreContext.cs
index 81cd8883..97907177 100644
--- a/Ryujinx.HLE/HOS/Kernel/Threading/KCoreContext.cs
+++ b/Ryujinx.HLE/HOS/Kernel/Threading/KCoreContext.cs
@@ -70,8 +70,6 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
                 CurrentThread.TotalTimeRunning += currentTime - CurrentThread.LastScheduledTime;
                 CurrentThread.LastScheduledTime = currentTime;
 
-                CurrentThread.ClearExclusive();
-
                 _coreManager.Set(CurrentThread.Context.Work);
 
                 CurrentThread.Context.Execute();
diff --git a/Ryujinx.HLE/HOS/Kernel/Threading/KThread.cs b/Ryujinx.HLE/HOS/Kernel/Threading/KThread.cs
index 7eb27efc..17e0f3c3 100644
--- a/Ryujinx.HLE/HOS/Kernel/Threading/KThread.cs
+++ b/Ryujinx.HLE/HOS/Kernel/Threading/KThread.cs
@@ -1004,11 +1004,6 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
             Context.ThreadState.X1 = (ulong)threadHandle;
         }
 
-        public void ClearExclusive()
-        {
-            Owner.CpuMemory.ClearExclusive(CurrentCore);
-        }
-
         public void TimeUp()
         {
             ReleaseAndResume();