diff --git a/src/GPU.cpp b/src/GPU.cpp
index 9ce8c314d..e492a0334 100644
--- a/src/GPU.cpp
+++ b/src/GPU.cpp
@@ -3127,7 +3127,9 @@ void GPU_ligne(NDS_Screen * screen, u16 l)
 		//SPP level 3-8 rotoscale room
 		//NSMB raster fx backdrops
 		//bubble bobble revolution classic mode
-		gpu->refreshAffineStartRegs();
+		//NOTE:
+		//I am REALLY unsatisfied with this logic now. But it seems to be working..
+		gpu->refreshAffineStartRegs(-1,-1);
 	}
 
 	//cache some parameters which are assumed to be stable throughout the rendering of the entire line
@@ -3212,7 +3214,7 @@ void GPU_ligne(NDS_Screen * screen, u16 l)
 void gpu_savestate(std::ostream* os)
 {
 	//version
-	write32le(0,os);
+	write32le(1,os);
 	
 	os->write((char*)GPU_screen,sizeof(GPU_screen));
 	
@@ -3230,21 +3232,37 @@ bool gpu_loadstate(std::istream* is, int size)
 {
 	//read version
 	int version;
-	if(read32le(&version,is) != 1) return false;
-	if(version != 0) return false;
+
+	//sigh.. shouldve used a new version number
+	if(size == 256*192*2*2)
+		version = 0;
+	else if(size== 0x30024)
+	{
+		read32le(&version,is);
+		version = 1;
+	}
+	else
+		if(read32le(&version,is) != 1) return false;
+		
+
+	if(version<0||version>1) return false;
 
 	is->read((char*)GPU_screen,sizeof(GPU_screen));
-	read32le(&MainScreen.gpu->affineInfo[0].x,is);
-	read32le(&MainScreen.gpu->affineInfo[0].y,is);
-	read32le(&MainScreen.gpu->affineInfo[1].x,is);
-	read32le(&MainScreen.gpu->affineInfo[1].y,is);
-	read32le(&SubScreen.gpu->affineInfo[0].x,is);
-	read32le(&SubScreen.gpu->affineInfo[0].y,is);
-	read32le(&SubScreen.gpu->affineInfo[1].x,is);
-	read32le(&SubScreen.gpu->affineInfo[1].y,is);
 
-	MainScreen.gpu->refreshAffineStartRegs();
-	SubScreen.gpu->refreshAffineStartRegs();
+	if(version==1)
+	{
+		read32le(&MainScreen.gpu->affineInfo[0].x,is);
+		read32le(&MainScreen.gpu->affineInfo[0].y,is);
+		read32le(&MainScreen.gpu->affineInfo[1].x,is);
+		read32le(&MainScreen.gpu->affineInfo[1].y,is);
+		read32le(&SubScreen.gpu->affineInfo[0].x,is);
+		read32le(&SubScreen.gpu->affineInfo[0].y,is);
+		read32le(&SubScreen.gpu->affineInfo[1].x,is);
+		read32le(&SubScreen.gpu->affineInfo[1].y,is);
+		MainScreen.gpu->refreshAffineStartRegs(-1,-1);
+		SubScreen.gpu->refreshAffineStartRegs(-1,-1);
+	}
+
 	MainScreen.gpu->updateBLDALPHA();
 	SubScreen.gpu->updateBLDALPHA();
 	return !is->fail();
@@ -3268,22 +3286,35 @@ void GPU::setAffineStart(int layer, int xy, u32 val)
 {
 	if(xy==0) affineInfo[layer-2].x = val;
 	else affineInfo[layer-2].y = val;
-	refreshAffineStartRegs();
+	refreshAffineStartRegs(layer,xy);
 }
 
-void GPU::refreshAffineStartRegs()
+void GPU::refreshAffineStartRegs(const int num, const int xy)
 {
-	for(int num=2;num<=3;num++)
+	if(num==-1)
 	{
-		BGxPARMS * parms;
-		if (num==2)
-			parms = &(dispx_st)->dispx_BG2PARMS;
-		else
-			parms = &(dispx_st)->dispx_BG3PARMS;		
-
-		parms->BGxX = affineInfo[num-2].x;
-		parms->BGxY = affineInfo[num-2].y;
+		refreshAffineStartRegs(2,xy);
+		refreshAffineStartRegs(3,xy);
+		return;
 	}
+
+	if(xy==-1)
+	{
+		refreshAffineStartRegs(num,0);
+		refreshAffineStartRegs(num,1);
+		return;
+	}
+
+	BGxPARMS * parms;
+	if (num==2)
+		parms = &(dispx_st)->dispx_BG2PARMS;
+	else
+		parms = &(dispx_st)->dispx_BG3PARMS;		
+
+	if(xy==0)
+		parms->BGxX = affineInfo[num-2].x;
+	else
+		parms->BGxY = affineInfo[num-2].y;
 }
 
 void gpu_UpdateRender()
diff --git a/src/GPU.h b/src/GPU.h
index c4279a39d..f77b8ba7e 100644
--- a/src/GPU.h
+++ b/src/GPU.h
@@ -775,7 +775,7 @@ struct GPU
 	void setAffineStart(int layer, int xy, u32 val);
 	void setAffineStartWord(int layer, int xy, u16 val, int word);
 	u32 getAffineStart(int layer, int xy);
-	void refreshAffineStartRegs();
+	void refreshAffineStartRegs(const int num, const int xy);
 
 	struct AffineInfo {
 		AffineInfo() : x(0), y(0) {}
diff --git a/src/NDSSystem.cpp b/src/NDSSystem.cpp
index f8f1fb1ed..c43718e45 100644
--- a/src/NDSSystem.cpp
+++ b/src/NDSSystem.cpp
@@ -845,7 +845,10 @@ int NDS_LoadROM( const char *filename, int bmtype, u32 bmsize,
 	return ret;
 }
 
-void MovieSRAM(int bmtype, u32 bmsize) {
+void MovieSRAM()
+{
+	int bmtype = MMU.bupmem.type;
+	u32 bmsize = MMU.bupmem.size;
 
 	char buf[MAX_PATH];
 
@@ -859,7 +862,6 @@ void MovieSRAM(int bmtype, u32 bmsize) {
 
 	mc_realloc(&MMU.bupmem, bmtype, bmsize);
 	mc_load_file(&MMU.bupmem, buf);
-
 }
 
 void NDS_FreeROM(void)
@@ -882,6 +884,12 @@ void NDS_Reset( void)
 
 	if (!header) return ;
 
+	if (MMU.bupmem.fp)
+	{
+		fclose(MMU.bupmem.fp);
+		MMU.bupmem.fp = NULL;
+	}
+
 	lagframecounter=0;
 	LagFrameFlag=0;
 	lastLag=0;
diff --git a/src/NDSSystem.h b/src/NDSSystem.h
index 83fca58e3..6be3edd1d 100644
--- a/src/NDSSystem.h
+++ b/src/NDSSystem.h
@@ -385,6 +385,7 @@ public:
 	virtual BOOL WIFI_Host_InitSystem() { return FALSE; }
 	virtual void WIFI_Host_ShutdownSystem() {}
 	virtual BOOL AVI_IsRecording() { return FALSE; }
+	virtual void USR_InfoMessage(const char *message) { printf("%s\n", message); }
 };
 extern Driver* driver;
 
@@ -392,7 +393,7 @@ extern std::string InputDisplayString;
 extern int LagFrameFlag;
 extern int lastLag, TotalLagFrames;
 
-void MovieSRAM(int bmtype, u32 bmsize);
+void MovieSRAM();
 
 void ClearAutoHold(void);
 
diff --git a/src/cocoa/sndOSX.mm b/src/cocoa/sndOSX.mm
index 112b274b9..43beb47a9 100644
--- a/src/cocoa/sndOSX.mm
+++ b/src/cocoa/sndOSX.mm
@@ -26,7 +26,7 @@
 
 #include <CoreServices/CoreServices.h>
 #include <AudioUnit/AudioUnit.h>
-#include <AudioToolbox/AudioToolBox.h>
+#include <AudioToolbox/AudioToolbox.h>
 
 //desmume includes
 #define OBJ_C
diff --git a/src/ctrlssdl.cpp b/src/ctrlssdl.cpp
index d14436b5d..a07a8757a 100644
--- a/src/ctrlssdl.cpp
+++ b/src/ctrlssdl.cpp
@@ -292,8 +292,15 @@ void set_mouse_coord(signed long x,signed long y)
 /* Update NDS keypad */
 void update_keypad(u16 keys)
 {
+#ifdef WORDS_BIGENDIAN
+  ARM9Mem.ARM9_REG[0x130] = ~keys & 0xFF;
+  ARM9Mem.ARM9_REG[0x131] = (~keys >> 8) & 0x3;
+  MMU.ARM7_REG[0x130] = ~keys & 0xFF;
+  MMU.ARM7_REG[0x131] = (~keys >> 8) & 0x3;
+#else
   ((u16 *)ARM9Mem.ARM9_REG)[0x130>>1] = ~keys & 0x3FF;
   ((u16 *)MMU.ARM7_REG)[0x130>>1] = ~keys & 0x3FF;
+#endif
   /* Update X and Y buttons */
   MMU.ARM7_REG[0x136] = ( ~( keys >> 10) & 0x3 ) | (MMU.ARM7_REG[0x136] & ~0x3);
 }
@@ -304,7 +311,11 @@ u16 get_keypad( void)
   u16 keypad;
   keypad = ~MMU.ARM7_REG[0x136];
   keypad = (keypad & 0x3) << 10;
+#ifdef WORDS_BIGENDIAN
+  keypad |= ~(ARM9Mem.ARM9_REG[0x130] | (ARM9Mem.ARM9_REG[0x131] << 8)) & 0x3FF;
+#else
   keypad |= ~((u16 *)ARM9Mem.ARM9_REG)[0x130>>1] & 0x3FF;
+#endif
   return keypad;
 }
 
diff --git a/src/gfx3d.cpp b/src/gfx3d.cpp
index a924df224..c0f42dce6 100644
--- a/src/gfx3d.cpp
+++ b/src/gfx3d.cpp
@@ -28,6 +28,7 @@
 #include <algorithm>
 #include <assert.h>
 #include <math.h>
+#include <string.h>
 #include "armcpu.h"
 #include "debug.h"
 #include "gfx3d.h"
@@ -2364,4 +2365,4 @@ bool gfx3d_loadstate(std::istream* is, int size)
 	gfx3d.vertlist->count=0;
 
 	return true;
-}
\ No newline at end of file
+}
diff --git a/src/mc.cpp b/src/mc.cpp
index 69af93a48..abccf08b9 100644
--- a/src/mc.cpp
+++ b/src/mc.cpp
@@ -82,6 +82,7 @@ u8 *mc_alloc(memory_chip_t *mc, u32 size)
 {
 	u8 *buffer;
 	buffer = new u8[size];
+	memset(buffer,0,size);
 
 	mc->data = buffer;
 	if(!buffer) { return NULL; }
@@ -199,9 +200,9 @@ void mc_reset_com(memory_chip_t *mc)
 
 void mc_realloc(memory_chip_t *mc, int type, u32 size)
 {
-        if(mc->data) delete[] mc->data;
-        mc_init(mc, type);
-        mc_alloc(mc, size);     
+    if(mc->data) delete[] mc->data;
+    mc_init(mc, type);
+    mc_alloc(mc, size);     
 }
 
 void mc_load_file(memory_chip_t *mc, const char* filename)
@@ -213,7 +214,7 @@ void mc_load_file(memory_chip_t *mc, const char* filename)
 
    if(movieMode != MOVIEMODE_INACTIVE) {
 	    mc->filename = strdup(filename);
-      return;
+		return;
    }
    else
 	   file = fopen(filename, "rb+");
diff --git a/src/movie.cpp b/src/movie.cpp
index 471301658..e4dd0ab88 100644
--- a/src/movie.cpp
+++ b/src/movie.cpp
@@ -22,7 +22,6 @@
 #include <assert.h>
 #include <limits.h>
 #include <fstream>
-#include "main.h"
 #include "utils/guid.h"
 #include "utils/xstring.h"
 #include "movie.h"
@@ -355,7 +354,7 @@ static void closeRecordingMovie()
 /// Stop movie playback.
 static void StopPlayback()
 {
-	SetMessageToDisplay("Movie playback stopped.");
+	driver->USR_InfoMessage("Movie playback stopped.");
 	movieMode = MOVIEMODE_INACTIVE;
 }
 
@@ -363,7 +362,7 @@ static void StopPlayback()
 /// Stop movie recording
 static void StopRecording()
 {
-	SetMessageToDisplay("Movie recording stopped.");
+	driver->USR_InfoMessage("Movie recording stopped.");
 	movieMode = MOVIEMODE_INACTIVE;
 	
 	closeRecordingMovie();
@@ -436,14 +435,14 @@ void FCEUI_LoadMovie(const char *fname, bool _read_only, bool tasedit, int _paus
 	movieMode = MOVIEMODE_PLAY;
 	currRerecordCount = currMovieData.rerecordCount;
 	InitMovieTime();
-	MovieSRAM(backupmemorytype, backupmemorysize);
+	MovieSRAM();
 	freshMovie = true;
 	ClearAutoHold();
 
 	if(movie_readonly)
-		SetMessageToDisplay("Replay started Read-Only.");
+		driver->USR_InfoMessage("Replay started Read-Only.");
 	else
-		SetMessageToDisplay("Replay started Read+Write.");
+		driver->USR_InfoMessage("Replay started Read+Write.");
 }
 
 static void openRecordingMovie(const char* fname)
@@ -492,9 +491,9 @@ static void openRecordingMovie(const char* fname)
 	movie_readonly = false;
 	currRerecordCount = 0;
 	InitMovieTime();
-	MovieSRAM(backupmemorytype, backupmemorysize);
-	
-	SetMessageToDisplay("Movie recording started.");
+	MovieSRAM();
+
+	driver->USR_InfoMessage("Movie recording started.");
 }
 
  void NDS_setTouchFromMovie(void) {
@@ -609,6 +608,9 @@ static void FCEUMOV_AddCommand(int cmd)
 	//DoEncode((cmd>>3)&0x3,cmd&0x7,1);
 }
 
+//little endian 4-byte cookies
+static const int kMOVI = 0x49564F4D;
+static const int kNOMO = 0x4F4D4F4E;
 
 void mov_savestate(std::ostream* os)
 {
@@ -618,12 +620,12 @@ void mov_savestate(std::ostream* os)
 	//else return 0;
 	if(movieMode == MOVIEMODE_RECORD || movieMode == MOVIEMODE_PLAY)
 	{
-		write32le('MOVI',os);
+		write32le(kMOVI,os);
 		currMovieData.dump(os, true);
 	}
 	else
 	{
-		write32le('NOMO',os);
+		write32le(kNOMO,os);
 	}
 }
 
@@ -637,9 +639,9 @@ bool mov_loadstate(std::istream* is, int size)
 
 	int cookie;
 	if(read32le(&cookie,is) != 1) return false;
-	if(cookie == 'NOMO')
+	if(cookie == kNOMO)
 		return true;
-	else if(cookie != 'MOVI')
+	else if(cookie != kMOVI)
 		return false;
 
 	size -= 4;
@@ -851,7 +853,7 @@ void LoadFM2_binarychunk(MovieData& movieData, std::istream* fp, int size)
 
 #include <sstream>
 
-bool CheckFileExists(const char* filename)
+static bool CheckFileExists(const char* filename)
 {
 	//This function simply checks to see if the given filename exists
 	string checkFilename; 
diff --git a/src/rasterize.cpp b/src/rasterize.cpp
index d04e1468c..f5fd52982 100644
--- a/src/rasterize.cpp
+++ b/src/rasterize.cpp
@@ -1145,6 +1145,7 @@ static void SoftRastRender()
 {
 	Fragment clearFragment;
 	FragmentColor clearFragmentColor;
+	clearFragment.pad = 0;
 	clearFragmentColor.r = gfx3d.clearColor&0x1F;
 	clearFragmentColor.g = (gfx3d.clearColor>>5)&0x1F;
 	clearFragmentColor.b = (gfx3d.clearColor>>10)&0x1F;
diff --git a/src/saves.cpp b/src/saves.cpp
index 5255144a8..084993354 100644
--- a/src/saves.cpp
+++ b/src/saves.cpp
@@ -37,7 +37,9 @@
 #include "readwrite.h"
 #include "gfx3d.h"
 #include "movie.h"
+#ifdef _MSC_VER
 #include "windows/main.h"
+#endif
 
 
 //void*v is actually a void** which will be indirected before reading
@@ -49,6 +51,9 @@ savestates_t savestates[NB_STATES];
 #define SAVESTATE_VERSION       11
 static const char* magic = "DeSmuME SState\0";
 
+//a savestate chunk loader can set this if it wants to permit a silent failure (for compatibility)
+static bool SAV_silent_fail_flag;
+
 #ifndef MAX_PATH
 #define MAX_PATH 256
 #endif
@@ -145,7 +150,9 @@ SFORMAT SF_MEM[]={
 
 	{ "VMEM", 1, sizeof(ARM9Mem.ARM9_VMEM),    ARM9Mem.ARM9_VMEM},
 	{ "OAMS", 1, sizeof(ARM9Mem.ARM9_OAM),    ARM9Mem.ARM9_OAM},
-	{ "LCDM", 1, sizeof(ARM9Mem.ARM9_LCD),		ARM9Mem.ARM9_LCD},
+
+	//this size is specially chosen to avoid saving the blank space at the end
+	{ "LCDM", 1, 0xA4000,		ARM9Mem.ARM9_LCD},
 	{ 0 }
 };
 
@@ -248,7 +255,7 @@ SFORMAT SF_MOVIE[]={
 static void mmu_savestate(std::ostream* os)
 {
 	//version
-	write32le(0,os);
+	write32le(1,os);
 
 	write32le(MMU.bupmem.type,os);
 	write32le(MMU.bupmem.size,os);
@@ -260,14 +267,27 @@ static bool mmu_loadstate(std::istream* is, int size)
 	//read version
 	int version;
 	if(read32le(&version,is) != 1) return false;
-	if(version != 0) return false;
-
-	int bupmem_type;
-	u32 bupmem_size;
-	if(read32le(&bupmem_type,is) != 1) return false;
-	if(read32le(&bupmem_size,is) != 1) return false;
 	
-	mc_realloc(&MMU.bupmem,bupmem_type,bupmem_size);
+
+	u32 bupmem_size;
+
+	if(version == 0)
+	{
+		//version 0 was buggy and didnt save the type. 
+		//it would silently fail if there was a size mismatch
+		SAV_silent_fail_flag = true;
+		if(read32le(&bupmem_size,is) != 1) return false;
+		//if(bupmem_size != MMU.bupmem.size) return false; //mismatch between current initialized and saved size
+		mc_realloc(&MMU.bupmem,MC_TYPE_AUTODETECT,bupmem_size);
+	}
+	else
+	{
+		//version 1 reinitializes the save system with the type that was saved
+		int bupmem_type;
+		if(read32le(&bupmem_type,is) != 1) return false;
+		if(read32le(&bupmem_size,is) != 1) return false;
+		mc_realloc(&MMU.bupmem,bupmem_type,bupmem_size);
+	}
 
 	is->read((char*)MMU.bupmem.data,bupmem_size);
 	if(is->fail()) return false;
@@ -593,7 +613,7 @@ static int SubWrite(std::ostream* os, SFORMAT *sf)
 		int size = sf->size;
 
         //add size of current node to the accumulator
-		acc += sizeof(sf->desc) + sizeof(sf->size) + sizeof(sf->count);
+		acc += 4 + sizeof(sf->size) + sizeof(sf->count);
 		acc += count * size;
 
 		if(os)			//Are we writing or calculating the size of this block?
@@ -668,16 +688,6 @@ static bool savestate_save(std::ostream* outstream, int compressionLevel)
 	writechunks(os);
 	ms.flush();
 
-	for(int i=0x2000;i<0x1000000;i++)
-		if(ARM9Mem.ARM9_REG[i] != 0) {
-			MessageBox(0,"Debug check fail: ARM9Mem.ARM9_REG",0,0);
-		}
-
-	for(int i=0;i<0x20000;i++)
-		if(ARM9Mem.blank_memory[i] != 0) {
-			MessageBox(0,"Debug check fail: ARM9Mem.blank_memory",0,0);
-		}
-
 	//save the length of the file
 	u32 len = ms.size();
 
@@ -737,8 +747,6 @@ bool savestate_save (const char *file_name)
 	} else return false;
 }
 
-//u8 GPU_screen[4*256*192];
-
 static void writechunks(std::ostream* os) {
 	savestate_WriteChunk(os,1,SF_ARM9);
 	savestate_WriteChunk(os,2,SF_ARM7);
@@ -812,6 +820,7 @@ static void loadstate()
 
 static bool savestate_load(std::istream* is)
 {
+	SAV_silent_fail_flag = false;
 	char header[16];
 	is->read(header,16);
 	if(is->fail() || memcmp(header,magic,16))
@@ -867,10 +876,11 @@ static bool savestate_load(std::istream* is)
 	memorystream mstemp(&buf);
 	bool x = ReadStateChunks(&mstemp,(s32)len);
 
-	if(!x)
+	if(!x && !SAV_silent_fail_flag)
 	{
 		printf("Error loading savestate. It failed halfway through;\nSince there is no savestate backup system, your current game session is wrecked");
 #ifdef _MSC_VER
+		//HACK! we really need a better way to handle this kind of feedback
 		MessageBox(0,"Error loading savestate. It failed halfway through;\nSince there is no savestate backup system, your current game session is wrecked",0,0);
 #endif
 		return false;
diff --git a/src/windows/main.cpp b/src/windows/main.cpp
index cf4852234..709b4e927 100644
--- a/src/windows/main.cpp
+++ b/src/windows/main.cpp
@@ -1464,6 +1464,11 @@ class WinDriver : public Driver
 	{
 		return ::AVI_IsRecording();
 	}
+
+	virtual void USR_InfoMessage(const char *message)
+	{
+		SetMessageToDisplay(message);
+	}
 };
 
 int WINAPI WinMain (HINSTANCE hThisInstance,
diff --git a/src/windows/ram_search.cpp b/src/windows/ram_search.cpp
index 16daf0c25..ad40237c6 100644
--- a/src/windows/ram_search.cpp
+++ b/src/windows/ram_search.cpp
@@ -71,7 +71,7 @@ static BOOL s_itemIndicesInvalid = true; // if true, the link from listbox items
 static BOOL s_prevValuesNeedUpdate = true; // if true, the "prev" values should be updated using the "cur" values on the next frame update signaled
 static unsigned int s_maxItemIndex = 0; // max currently valid item index, the listbox sometimes tries to update things past the end of the list so we need to know this to ignore those attempts
 
-static const MemoryRegion s_prgRegion    = {  0x02000000, 0x400000, (unsigned char*)ARM9Mem.MAIN_MEM,     true};
+static const MemoryRegion s_prgRegion    = {  0x02000000, 0x400000, (unsigned char*)ARM9Mem.MAIN_MEM,     false};
 
 /*
 static const MemoryRegion s_prgRegion    = {  0x020000, SEGACD_RAM_PRG_SIZE, (unsigned char*)Ram_Prg,     true};
diff --git a/src/windows/targetver.h b/src/windows/targetver.h
deleted file mode 100644
index f583181df..000000000
--- a/src/windows/targetver.h
+++ /dev/null
@@ -1,24 +0,0 @@
-#pragma once
-
-// The following macros define the minimum required platform.  The minimum required platform
-// is the earliest version of Windows, Internet Explorer etc. that has the necessary features to run 
-// your application.  The macros work by enabling all features available on platform versions up to and 
-// including the version specified.
-
-// Modify the following defines if you have to target a platform prior to the ones specified below.
-// Refer to MSDN for the latest info on corresponding values for different platforms.
-#ifndef WINVER                          // Specifies that the minimum required platform is Windows Vista.
-#define WINVER 0x0600           // Change this to the appropriate value to target other versions of Windows.
-#endif
-
-#ifndef _WIN32_WINNT            // Specifies that the minimum required platform is Windows Vista.
-#define _WIN32_WINNT 0x0600     // Change this to the appropriate value to target other versions of Windows.
-#endif
-
-#ifndef _WIN32_WINDOWS          // Specifies that the minimum required platform is Windows 98.
-#define _WIN32_WINDOWS 0x0410 // Change this to the appropriate value to target Windows Me or later.
-#endif
-
-#ifndef _WIN32_IE                       // Specifies that the minimum required platform is Internet Explorer 7.0.
-#define _WIN32_IE 0x0700        // Change this to the appropriate value to target other versions of IE.
-#endif