From efe1af294fd9f6c0c05178be5bcbfb1be565a1f5 Mon Sep 17 00:00:00 2001
From: Sam Lantinga <slouken@libsdl.org>
Date: Thu, 17 Feb 2011 09:13:12 -0800
Subject: [PATCH] Fixed bug 1128

 Patrick Baggett      2011-02-16 22:58:33 PST

This enhancement is for both x86/x64 Windows.

The SDL implementation of mutexes uses the Win32 API interprocess
synchronization primitive called a "Mutex". This implementation is subpar
because it has a much higher overhead than an intraprocess mutex. The exact
technical details are below, but my tests have shown that for reasonably high
contention (10 threads on 4 physical cores), it has 13x higher overhead than
the Win32 CriticalSection API.

If this enhancement is accepted, I will write a patch to implement SDL mutexes
using the critical section API, which should dramatically reduce overhead and
improve scalability.


-- Tech details --
Normally, Win32 Mutexes are used across process boundaries to synchronize
separate processes. In order to lock or unlock them, a user->kernel space
transition is necessary, even in the uncontented case on a single CPU machine.
Win32 CriticalSection objects can only be used within the same process virtual
address space and thus to lock one, does not require a user->kernel space
transition for the uncontended case, and additionally may spin a short while
before going into kernel wait. This small spin allows a thread to obtain the
lock if the mutex is released shortly after the thread starts spinning, in
effect bypassing the overhead of user->kernel space transition which has higher
overhead than the spinning itself.
---
 src/thread/windows/SDL_sysmutex.c | 45 +++++++++++++++++--------------
 1 file changed, 25 insertions(+), 20 deletions(-)

diff --git a/src/thread/windows/SDL_sysmutex.c b/src/thread/windows/SDL_sysmutex.c
index 8440ae7dc..fbcb1d3e1 100644
--- a/src/thread/windows/SDL_sysmutex.c
+++ b/src/thread/windows/SDL_sysmutex.c
@@ -30,7 +30,7 @@
 
 struct SDL_mutex
 {
-    HANDLE id;
+    CRITICAL_SECTION cs;
 };
 
 /* Create a mutex */
@@ -38,17 +38,29 @@ SDL_mutex *
 SDL_CreateMutex(void)
 {
     SDL_mutex *mutex;
+    static DWORD (WINAPI*pf_SetCriticalSectionSpinCount)(LPCRITICAL_SECTION, DWORD) = NULL;
+    static HMODULE kernel32 = NULL;
+
+    /* One time logic - detect WinNT */
+    if(kernel32 == NULL) {
+        kernel32 = GetModuleHandleA("kernel32.dll");
+		if(kernel32) {
+            /* Attempt to resolve symbol -- Win9x gets NULL */
+            pf_SetCriticalSectionSpinCount = (DWORD (WINAPI*)(LPCRITICAL_SECTION, DWORD))GetProcAddress(kernel32, "SetCriticalSectionSpinCount");
+        }
+		else
+			kernel32 = (HMODULE)0x01; /* don't try to init again */
+	}
+
 
     /* Allocate mutex memory */
     mutex = (SDL_mutex *) SDL_malloc(sizeof(*mutex));
     if (mutex) {
-        /* Create the mutex, with initial value signaled */
-        mutex->id = CreateMutex(NULL, FALSE, NULL);
-        if (!mutex->id) {
-            SDL_SetError("Couldn't create mutex");
-            SDL_free(mutex);
-            mutex = NULL;
-        }
+        /* Initialize */
+        InitializeCriticalSection(&mutex->cs);
+
+        /* On SMP systems, a non-zero spin count generally helps performance */
+        if(pf_SetCriticalSectionSpinCount) pf_SetCriticalSectionSpinCount(&mutex->cs, 2000);
     } else {
         SDL_OutOfMemory();
     }
@@ -60,10 +72,7 @@ void
 SDL_DestroyMutex(SDL_mutex * mutex)
 {
     if (mutex) {
-        if (mutex->id) {
-            CloseHandle(mutex->id);
-            mutex->id = 0;
-        }
+        DeleteCriticalSection(&mutex->cs);
         SDL_free(mutex);
     }
 }
@@ -76,10 +85,8 @@ SDL_mutexP(SDL_mutex * mutex)
         SDL_SetError("Passed a NULL mutex");
         return -1;
     }
-    if (WaitForSingleObject(mutex->id, INFINITE) == WAIT_FAILED) {
-        SDL_SetError("Couldn't wait on mutex");
-        return -1;
-    }
+
+    EnterCriticalSection(&mutex->cs);
     return (0);
 }
 
@@ -91,10 +98,8 @@ SDL_mutexV(SDL_mutex * mutex)
         SDL_SetError("Passed a NULL mutex");
         return -1;
     }
-    if (ReleaseMutex(mutex->id) == FALSE) {
-        SDL_SetError("Couldn't release mutex");
-        return -1;
-    }
+
+    LeaveCriticalSection(&mutex->cs);
     return (0);
 }