Adding NVIDIA 'Threaded optimization' workaround/fix at initializatio…

…n on Windows for javaws/applets.

It has been observed that for some combinations, eg:
     - Windows 7 64bit (other variants may apply too)
     - NVIDIA 8600M GT
     - 260.99

the NVIDIA setting of 'Threaded optimization' := 'auto' (default) causes the JVM to simply crash
in case of javaws and [jnlp] applets.

'Threaded Optimization' := 'off' works reliable

'Threaded Optimization' := 'on'  never works with javaws and applets on the above configuration

A user could workaround this by setting 'Threaded Optimization' := 'off',
however, this would disable many users on the spot,
since you cannot ask the average user for such a task, if she only wants to see a web page.

This patch 'fixes' the 'auto' mode by running the eager GL profile initialization
within a block of single CPU affinity:

  SetProcessAffinityMask(pid, 1);
  try {
    initProfilesForDeviceImpl(device);
  } finally {
    SetProcessAffinityMask(pid, sysValue);
  }

Hopefully we can remove this hack with a driver fix.
However this workaround is as little invasive as possible.

make/stub_includes/win32/windows.h

@@ -12,7 +12,7 @@
 typedef int                 BOOL;
 typedef unsigned char       BYTE;
 typedef char                CHAR;
-typedef unsigned int        DWORD;
+typedef unsigned __int32    DWORD;
 typedef int                 INT;
 typedef __int32             INT32;
 typedef __int64             INT64;
@@ -29,12 +29,15 @@ typedef HANDLE              HWND;
 typedef __int32             LONG;
 typedef const char*         LPCSTR;
 typedef void*               LPVOID;
+typedef unsigned __int64    ULONG_PTR;
 typedef struct _proc*       PROC;
 typedef unsigned int*       PUINT;
 typedef unsigned int        UINT;
 typedef unsigned short      USHORT;
 typedef unsigned short      WORD;
 typedef unsigned short      ATOM;
+typedef intptr_t            DWORD_PTR;
+typedef intptr_t*           PDWORD_PTR;
 
 /* Necessary handle typedefs for parsing wglext.h */
 

make/stub_includes/win32/wingdi.h

@@ -194,6 +194,11 @@ WINUSERAPI int  WINAPI ReleaseDC(HWND hWnd, HDC hDC);
 WINUSERAPI BOOL WINAPI DestroyWindow(HWND hWnd);
 WINUSERAPI DWORD WINAPI GetObjectType(HGDIOBJ h);
 
+WINUSERAPI HANDLE WINAPI GetCurrentProcess(void);
+WINUSERAPI BOOL WINAPI GetProcessAffinityMask(HANDLE hProcess,PDWORD_PTR lpProcessAffinityMask,PDWORD_PTR lpSystemAffinityMask);
+WINUSERAPI BOOL WINAPI SetProcessAffinityMask(HANDLE hProcess,DWORD_PTR dwProcessAffinityMask);
+
+
 // Routines for changing gamma ramp of display device
 WINGDIAPI BOOL        WINAPI GetDeviceGammaRamp(HDC,LPVOID);
 WINGDIAPI BOOL        WINAPI SetDeviceGammaRamp(HDC,LPVOID);

src/jogl/classes/com/jogamp/opengl/impl/windows/wgl/WindowsWGLDrawableFactory.java

@@ -60,6 +60,7 @@
 import javax.media.opengl.GLProfile;
 
 import com.jogamp.common.JogampRuntimeException;
+import com.jogamp.common.nio.PointerBuffer;
 import com.jogamp.common.util.ReflectionUtil;
 import com.jogamp.nativewindow.impl.ProxySurface;
 import com.jogamp.nativewindow.impl.windows.GDI;
@@ -125,6 +126,42 @@ public WindowsWGLDrawableFactory() {
   Thread sharedResourceThread;
   HashMap/*<connection, SharedResource>*/ sharedMap = new HashMap();
 
+  long processAffinityChanges = 0;
+  PointerBuffer procMask = PointerBuffer.allocateDirect(1);
+  PointerBuffer sysMask = PointerBuffer.allocateDirect(1);
+
+  protected void enterThreadCriticalZone() {
+    synchronized (sysMask) {
+        if( 0 == processAffinityChanges) {
+            long pid = GDI.GetCurrentProcess();
+            if ( GDI.GetProcessAffinityMask(pid, procMask, sysMask) ) {
+                if(DEBUG) {
+                    System.err.println("WindowsWGLDrawableFactory.enterThreadCriticalZone() - 0x" + Long.toHexString(pid) + " - " + Thread.currentThread().getName());
+                    Thread.dumpStack();
+                }
+                processAffinityChanges = pid;
+                GDI.SetProcessAffinityMask(pid, 1);
+            }
+        }
+    }
+  }
+
+  protected void leaveThreadCriticalZone() {
+    synchronized (sysMask) {
+        if( 0 != processAffinityChanges) {
+            long pid = GDI.GetCurrentProcess();
+            if( pid != processAffinityChanges) {
+                throw new GLException("PID doesn't match: set PID 0x" + Long.toHexString(processAffinityChanges) +
+                                                       " this PID 0x" + Long.toHexString(pid) );
+            }
+            if(DEBUG) {
+                System.err.println("WindowsWGLDrawableFactory.leaveThreadCriticalZone() - 0x" + Long.toHexString(pid) + " - " + Thread.currentThread().getName());
+            }
+            GDI.SetProcessAffinityMask(pid, sysMask.get(0));
+        }
+    }
+  }
+
   static class SharedResource implements SharedResourceRunner.Resource {
       private WindowsGraphicsDevice device;
       private AbstractGraphicsScreen screen;

src/jogl/classes/javax/media/opengl/GLDrawableFactory.java

@@ -220,6 +220,9 @@ protected GLDrawableFactory() {
     }
   }
 
+  protected void enterThreadCriticalZone() {};
+  protected void leaveThreadCriticalZone() {};
+
   protected abstract void shutdownInstance();
 
   /**

src/jogl/classes/javax/media/opengl/GLProfile.java

@@ -1218,6 +1218,15 @@ private static synchronized boolean initProfilesForDevice(AbstractGraphicsDevice
         if(null == device) {
             return false;
         }
+        GLDrawableFactory factory = GLDrawableFactory.getFactoryImpl(device);
+        factory.enterThreadCriticalZone();
+        try {
+            return initProfilesForDeviceImpl(device);
+        } finally {
+            factory.leaveThreadCriticalZone();
+        }
+    }
+    private static synchronized boolean initProfilesForDeviceImpl(AbstractGraphicsDevice device) {
         boolean isSet = GLContext.getAvailableGLVersionsSet(device);
 
         if(DEBUG) {