From: notaz <notasas@gmail.com>
Date: Fri, 28 Aug 2009 20:16:33 +0000 (+0000)
Subject: sms wip: random tweaks, refactoring (palette, sound stuff, etc)
X-Git-Tag: v1.85~285
X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=19954be1966a04304b16d9180b003ec8ca3bc532;p=picodrive.git

sms wip: random tweaks, refactoring (palette, sound stuff, etc)

git-svn-id: file:///home/notaz/opt/svn/PicoDrive@764 be3aeb3a-fb24-0410-a615-afba39da0efa
---

diff --git a/pico/draw.c b/pico/draw.c
index 747f4281..8647a36d 100644
--- a/pico/draw.c
+++ b/pico/draw.c
@@ -53,8 +53,8 @@ int  HighPreSpr[80*2+1]; // slightly preprocessed sprites
 #define SPRL_LO_ABOVE_HI 0x10 // low priority sprites may be on top of hi
 unsigned char HighLnSpr[240][3 + MAX_LINE_SPRITES]; // sprite_count, ^flags, tile_count, [spritep]...
 
-int rendstatus = 0;
-int DrawScanline = 0;
+int rendstatus, rendstatus_old;
+int DrawScanline;
 int PicoDrawMask = -1;
 
 static int skip_next_line=0;
@@ -79,7 +79,6 @@ void DrawTilesFromCache(int *hc, int sh, int rlim);
 void DrawSpritesSHi(unsigned char *sprited);
 void DrawLayer(int plane_sh, int *hcache, int cellskip, int maxcells);
 void FinalizeLineBGR444(int sh);
-void FinalizeLineRGB555(int sh);
 void *blockcpy(void *dst, const void *src, size_t n);
 void blockcpy_or(void *dst, void *src, size_t n, int pat);
 #else
@@ -1160,20 +1159,24 @@ unsigned short HighPal[0x100];
 #ifndef _ASM_DRAW_C
 void PicoDoHighPal555(int sh)
 {
+  unsigned int *spal, *dpal;
   unsigned short *pal=HighPal;
   int i, t;
 
   Pico.m.dirtyPal = 0;
 
-  {
-    unsigned int *spal=(void *)Pico.cram;
-    unsigned int *dpal=(void *)HighPal;
-    for (i = 0x3f/2; i >= 0; i--)
+  spal = (void *)Pico.cram;
+  dpal = (void *)HighPal;
+
+  for (i = 0; i < 0x40; i++) {
+    unsigned int t = spal[i];
 #ifdef USE_BGR555
-      dpal[i] = ((spal[i]&0x000f000f)<< 1)|((spal[i]&0x00f000f0)<<3)|((spal[i]&0x0f000f00)<<4);
+    t = ((t & 0x000e000e)<< 1) | ((t & 0x00e000e0)<<3) | ((t & 0x0e000e00)<<4);
 #else
-      dpal[i] = ((spal[i]&0x000f000f)<<12)|((spal[i]&0x00f000f0)<<3)|((spal[i]&0x0f000f00)>>7);
+    t = ((t & 0x000e000e)<<12) | ((t & 0x00e000e0)<<3) | ((t & 0x0e000e00)>>7);
 #endif
+    t |= (t >> 3) & 0x18e318e3;
+    dpal[i] = t;
   }
 
   if (sh)
@@ -1227,7 +1230,7 @@ static void FinalizeLineBGR444(int sh)
 }
 
 
-static void FinalizeLineRGB555(int sh)
+void FinalizeLineRGB555(int sh)
 {
   unsigned short *pd=DrawLineDest;
   unsigned char  *ps=HighCol+8;
@@ -1265,7 +1268,7 @@ static void FinalizeLineRGB555(int sh)
 
 static void FinalizeLine8bit(int sh)
 {
-  unsigned char *pd=DrawLineDest;
+  unsigned char *pd = DrawLineDest;
   int len, rs = rendstatus;
   static int dirty_count;
 
@@ -1287,7 +1290,8 @@ static void FinalizeLine8bit(int sh)
   if (Pico.video.reg[12]&1) {
     len = 320;
   } else {
-    if (!(PicoOpt&POPT_DIS_32C_BORDER)) pd+=32;
+    if (!(PicoOpt & POPT_DIS_32C_BORDER))
+      pd += 32;
     len = 256;
   }
 
@@ -1407,12 +1411,22 @@ PICO_INTERNAL void PicoFrameStart(void)
   if (Pico.video.reg[1] & 8)
     rendstatus |= PDRAW_240LINES;
 
+  DrawScanline = 0;
+  skip_next_line = 0;
+
+  if (rendstatus != rendstatus_old) {
+    rendstatus_old = rendstatus;
+    emu_video_mode_change((rendstatus & PDRAW_240LINES) ? 0 : 8,
+      (rendstatus & PDRAW_240LINES) ? 240 : 224,
+      (Pico.video.reg[12] & 1) ? 0 : 1);
+  }
+
+  if (PicoOpt & POPT_ALT_RENDERER)
+    return;
+
   if (Pico.m.dirtyPal)
     Pico.m.dirtyPal = 2; // reset dirty if needed
-
-  DrawScanline=0;
   PrepareSprites(1);
-  skip_next_line=0;
 }
 
 static void DrawBlankedLine(int line, int offs)
diff --git a/pico/mode4.c b/pico/mode4.c
index 095250f9..3beb709d 100644
--- a/pico/mode4.c
+++ b/pico/mode4.c
@@ -190,17 +190,26 @@ static void DrawDisplayM4(int scanline)
 
 void PicoFrameStartMode4(void)
 {
+  int lines = 192;
   skip_next_line = 0;
   screen_offset = 24;
-  rendstatus = PDRAW_192LINES;
+  rendstatus = 0;
+
   if ((Pico.video.reg[0] & 6) == 6 && (Pico.video.reg[1] & 0x18)) {
-    rendstatus &= ~PDRAW_192LINES;
     if (Pico.video.reg[1] & 0x08) {
-      screen_offset = 0;
       rendstatus |= PDRAW_240LINES;
+      screen_offset = 0;
+      lines = 240;
     }
-    else // it's 224 lines
+    else {
       screen_offset = 8;
+      lines = 224;
+    }
+  }
+
+  if (rendstatus != rendstatus_old) {
+    rendstatus_old = rendstatus;
+    emu_video_mode_change(screen_offset, lines, 1);
   }
 }
 
@@ -251,28 +260,22 @@ void PicoDoHighPal555M4(void)
 
 static void FinalizeLineRGB555M4(void)
 {
-  unsigned short *pd=DrawLineDest;
-  unsigned char  *ps=HighCol+8;
-  unsigned short *pal=HighPal;
-  int i;
-
   if (Pico.m.dirtyPal)
     PicoDoHighPal555M4();
 
-  if (!(PicoOpt & POPT_DIS_32C_BORDER))
-    pd += 32;
-
-  for (i = 256/4; i > 0; i--) {
-    *pd++ = pal[*ps++];
-    *pd++ = pal[*ps++];
-    *pd++ = pal[*ps++];
-    *pd++ = pal[*ps++];
-  }
+  // standard FinalizeLine can finish it for us,
+  // with features like scaling and such
+  FinalizeLineRGB555(0);
 }
 
 static void FinalizeLine8bitM4(void)
 {
-  memcpy32(DrawLineDest, (int *)(HighCol+8), 256/4);
+  unsigned char *pd = DrawLineDest;
+
+  if (!(PicoOpt & POPT_DIS_32C_BORDER))
+    pd += 32;
+
+  memcpy32((int *)pd, (int *)(HighCol+8), 256/4);
 }
 
 void PicoDrawSetColorFormatMode4(int which)
diff --git a/pico/pico.c b/pico/pico.c
index c919507e..9b390c70 100644
--- a/pico/pico.c
+++ b/pico/pico.c
@@ -293,20 +293,19 @@ void PicoFrame(void)
 {
   Pico.m.frame_count++;
 
-  if (PicoAHW & PAHW_MCD) {
-    PicoFrameMCD();
+  if (PicoAHW & PAHW_SMS) {
+    PicoFrameMS();
     return;
   }
-  else if (PicoAHW & PAHW_SMS) {
-    PicoFrameMS();
+
+  if (PicoAHW & PAHW_MCD) {
+    PicoFrameMCD();
     return;
   }
 
   //if(Pico.video.reg[12]&0x2) Pico.video.status ^= 0x10; // change odd bit in interlace mode
 
-  if (!(PicoOpt&POPT_ALT_RENDERER))
-    PicoFrameStart();
-
+  PicoFrameStart();
   PicoFrameHints();
 }
 
diff --git a/pico/pico.h b/pico/pico.h
index 7436a605..92ba9cff 100644
--- a/pico/pico.h
+++ b/pico/pico.h
@@ -4,7 +4,7 @@
 // Pico Library - Header File
 
 // (c) Copyright 2004 Dave, All rights reserved.
-// (c) Copyright 2006-2008 notaz, All rights reserved.
+// (c) Copyright 2006-2009 notaz, All rights reserved.
 // Free for non-commercial use.
 
 // For commercial use, separate licencing terms must be obtained.
@@ -31,6 +31,9 @@ extern void mp3_update(int *buffer, int length, int stereo);
 // used by SVP dynarec
 extern void cache_flush_d_inval_i(const void *start_addr, const void *end_addr);
 
+// this one should handle display mode changes
+extern void emu_video_mode_change(int start_line, int line_count, int is_32cols);
+
 // Pico.c
 #define POPT_EN_FM          (1<< 0) // 00 000x
 #define POPT_EN_PSG         (1<< 1)
@@ -172,8 +175,7 @@ extern int PicoDrawMask;
 #define PDRAW_PLANE_HI_PRIO (1<<6) // have layer with all hi prio tiles (mk3)
 #define PDRAW_SHHI_DONE     (1<<7) // layer sh/hi already processed
 #define PDRAW_240LINES      (1<<8) // 240 line display (224 if not set)
-#define PDRAW_192LINES      (1<<9) // 192 line display (for SMS games)
-extern int rendstatus;
+extern int rendstatus, rendstatus_old;
 extern unsigned short HighPal[0x100];
 
 // Draw2.c
diff --git a/pico/pico_int.h b/pico/pico_int.h
index 4a4418e4..b5f4a95c 100644
--- a/pico/pico_int.h
+++ b/pico/pico_int.h
@@ -186,6 +186,7 @@ extern struct DrZ80 drZ80;
 #define z80_int()          drZ80.Z80_IRQ = 1
 
 #define z80_cyclesLeft     drZ80.cycles
+#define z80_pc()           (drZ80.Z80PC - drZ80.Z80PC_BASE)
 
 #elif defined(_USE_CZ80)
 #include "../cpu/cz80/cz80.h"
@@ -195,6 +196,7 @@ extern struct DrZ80 drZ80;
 #define z80_int()          Cz80_Set_IRQ(&CZ80, 0, HOLD_LINE)
 
 #define z80_cyclesLeft     (CZ80.ICount - CZ80.ExtraCycles)
+#define z80_pc()           Cz80_Get_Reg(&CZ80, CZ80_PC)
 
 #else
 
@@ -416,6 +418,7 @@ int CM_compareRun(int cyc, int is_sub);
 PICO_INTERNAL void PicoFrameStart(void);
 void PicoDrawSync(int to, int blank_last_line);
 void BackFill(int reg7, int sh);
+void FinalizeLineRGB555(int sh);
 extern int DrawScanline;
 #define MAX_LINE_SPRITES 29
 extern unsigned char HighLnSpr[240][3 + MAX_LINE_SPRITES];
diff --git a/pico/sms.c b/pico/sms.c
index 6cc6f3ca..e483291a 100644
--- a/pico/sms.c
+++ b/pico/sms.c
@@ -235,6 +235,7 @@ void PicoFrameMS(void)
   int lines = is_pal ? 313 : 262;
   int cycles_line = is_pal ? 58020 : 58293; /* (226.6 : 227.7) * 256 */
   int cycles_done = 0, cycles_aim = 0;
+  int skip = PicoSkipFrame;
   int lines_vis = 192;
   int hint; // Hint counter
   int y;
@@ -246,7 +247,7 @@ void PicoFrameMS(void)
   {
     pv->v_counter = Pico.m.scanline = y;
 
-    if (y < lines_vis)
+    if (y < lines_vis && !skip)
       PicoLineMode4(y);
 
     if (y <= lines_vis)
diff --git a/pico/sound/sound.c b/pico/sound/sound.c
index 7f3fd8df..a7510db4 100644
--- a/pico/sound/sound.c
+++ b/pico/sound/sound.c
@@ -398,7 +398,6 @@ PICO_INTERNAL void PsndGetSamples(int y)
 
 PICO_INTERNAL void PsndGetSamplesMS(void)
 {
-  int *buf32 = PsndBuffer;
   int stereo = (PicoOpt & 8) >> 3;
   int length = PsndLen;
 
@@ -415,8 +414,12 @@ PICO_INTERNAL void PsndGetSamplesMS(void)
   if (PicoOpt & POPT_EN_PSG)
     SN76496Update(PsndOut, length, stereo);
 
-  // convert + limit to normal 16bit output
-  PsndMix_32_to_16l(PsndOut, buf32, length);
+  // upmix to "stereo" if needed
+  if (stereo) {
+    int i = length, *p = (void *)PsndOut;
+    while (i--)
+      *p |= *p << 16;
+  }
 
   if (PicoWriteSound != NULL)
     PicoWriteSound(length);
diff --git a/platform/common/arm_utils.h b/platform/common/arm_utils.h
index ae5ae24c..8cf34ce8 100644
--- a/platform/common/arm_utils.h
+++ b/platform/common/arm_utils.h
@@ -3,9 +3,9 @@
 
 // For commercial use, separate licencing terms must be obtained.
 
-void vidConvCpyRGB32  (void *to, void *from, int pixels);
-void vidConvCpyRGB32sh(void *to, void *from, int pixels);
-void vidConvCpyRGB32hi(void *to, void *from, int pixels);
+void bgr444_to_rgb32(void *to, void *from);
+void bgr444_to_rgb32_sh(void *to, void *from);
+
 void vidcpy_m2(void *dest, void *src, int m32col, int with_32c_border);
 void vidcpy_m2_rot(void *dest, void *src, int m32col, int with_32c_border);
 void spend_cycles(int c); // utility
diff --git a/platform/common/arm_utils.s b/platform/common/arm_utils.s
index b93b99f0..fd13cac6 100644
--- a/platform/common/arm_utils.s
+++ b/platform/common/arm_utils.s
@@ -13,7 +13,6 @@
 @ to      00000000 rrr00000 ggg00000 bbb00000 ...
 
 @ lr =  0x00e000e0, out: r3=lower_pix, r2=higher_pix; trashes rin
-@ if sh==2, r8=0x00404040 (sh!=0 destroys flags!)
 .macro convRGB32_2 rin sh=0
     and     r2,  lr, \rin, lsr #4 @ blue
     and     r3,  \rin, lr
@@ -59,16 +58,19 @@
 .endif
 
     orr     r2,  r2,   r2,  lsr #3
+.if \sh == 1
+    str     r2, [r0, #0x40*2*4]
+.endif
     str     r2, [r0], #4
 .endm
 
 
-.global vidConvCpyRGB32 @ void *to, void *from, int pixels
+.global bgr444_to_rgb32 @ void *to, void *from
 
-vidConvCpyRGB32:
+bgr444_to_rgb32:
     stmfd   sp!, {r4-r7,lr}
 
-    mov     r12, r2, lsr #3 @ repeats
+    mov     r12, #0x40>>3 @ repeats
     mov     lr, #0x00e00000
     orr     lr, lr, #0x00e0
 
@@ -80,19 +82,18 @@ vidConvCpyRGB32:
     convRGB32_2 r5
     convRGB32_2 r6
     convRGB32_2 r7
-
     bgt     .loopRGB32
 
-    ldmfd   sp!, {r4-r7,lr}
-    bx      lr
+    ldmfd   sp!, {r4-r7,pc}
 
 
-.global vidConvCpyRGB32sh @ void *to, void *from, int pixels
+.global bgr444_to_rgb32_sh @ void *to, void *from
 
-vidConvCpyRGB32sh:
+bgr444_to_rgb32_sh:
     stmfd   sp!, {r4-r7,lr}
 
-    mov     r12, r2, lsr #3 @ repeats
+    mov     r12, #0x40>>3 @ repeats
+    add     r0, r0, #0x40*4
     mov     lr, #0x00e00000
     orr     lr, lr, #0x00e0
 
@@ -104,21 +105,10 @@ vidConvCpyRGB32sh:
     convRGB32_2 r5, 1
     convRGB32_2 r6, 1
     convRGB32_2 r7, 1
-
     bgt     .loopRGB32sh
 
-    ldmfd   sp!, {r4-r7,lr}
-    bx      lr
-
-
-.global vidConvCpyRGB32hi @ void *to, void *from, int pixels
-
-vidConvCpyRGB32hi:
-    stmfd   sp!, {r4-r7,lr}
-
-    mov     r12, r2, lsr #3 @ repeats
-    mov     lr, #0x00e00000
-    orr     lr, lr, #0x00e0
+    mov     r12, #0x40>>3 @ repeats
+    sub     r1, r1, #0x40*2
 
 .loopRGB32hi:
      ldmia    r1!, {r4-r7}
diff --git a/platform/common/emu.c b/platform/common/emu.c
index aca54bef..4a95a3ff 100644
--- a/platform/common/emu.c
+++ b/platform/common/emu.c
@@ -1332,7 +1332,7 @@ void emu_loop(void)
 {
 	int pframes_done;		/* "period" frames, used for sync */
 	int frames_done, frames_shown;	/* actual frames for fps counter */
-	int oldmodes, target_fps, target_frametime;
+	int target_fps, target_frametime;
 	unsigned int timestamp_base = 0, timestamp_fps;
 	char *notice_msg = NULL;
 	char fpsbuff[24];
@@ -1341,8 +1341,8 @@ void emu_loop(void)
 	fpsbuff[0] = 0;
 
 	/* make sure we are in correct mode */
-	oldmodes = ((Pico.video.reg[12]&1)<<2) ^ 0xc;
 	Pico.m.dirtyPal = 1;
+	rendstatus_old = -1;
 
 	/* number of ticks per frame */
 	if (Pico.m.pal) {
@@ -1371,7 +1371,6 @@ void emu_loop(void)
 	{
 		unsigned int timestamp;
 		int diff, diff_lim;
-		int modes;
 
 		timestamp = get_ticks();
 		if (reset_timing) {
@@ -1398,13 +1397,6 @@ void emu_loop(void)
 			}
 		}
 
-		// check for mode changes
-		modes = ((Pico.video.reg[12]&1)<<2) | (Pico.video.reg[1]&8);
-		if (modes != oldmodes) {
-			oldmodes = modes;
-			pemu_video_mode_change(!(modes & 4), (modes & 8));
-		}
-
 		// second changed?
 		if (timestamp - timestamp_fps >= ms_to_ticks(1000))
 		{
diff --git a/platform/common/plat.h b/platform/common/plat.h
index a453431a..ca847c69 100644
--- a/platform/common/plat.h
+++ b/platform/common/plat.h
@@ -9,7 +9,6 @@ void pemu_prep_defconfig(void);
 void pemu_loop_prep(void);
 void pemu_loop_end(void);
 void pemu_forced_frame(int opts);
-void pemu_video_mode_change(int is_32col, int is_240_lines);
 void pemu_update_display(const char *fps, const char *notice_msg);
 
 void pemu_sound_start(void);
diff --git a/platform/gp2x/emu.c b/platform/gp2x/emu.c
index dd4b6867..77e33f28 100644
--- a/platform/gp2x/emu.c
+++ b/platform/gp2x/emu.c
@@ -36,7 +36,7 @@ extern int crashed_940;
 static short __attribute__((aligned(4))) sndBuffer[2*(44100+100)/50];
 static unsigned char PicoDraw2FB_[(8+320) * (8+240+8)];
 unsigned char *PicoDraw2FB = PicoDraw2FB_;
-static int osd_fps_x;
+static int osd_fps_x, osd_y;
 
 extern void *gp2x_screens[4];
 
@@ -265,21 +265,66 @@ static int EmuScanEnd8_rot(unsigned int num)
 	return 0;
 }
 
-int localPal[0x100];
-static void (*vidcpyM2)(void *dest, void *src, int m32col, int with_32c_border) = NULL;
+static int localPal[0x100];
+static void (*vidcpyM2)(void *dest, void *src, int m32col, int with_32c_border);
+static int (*make_local_pal)(int fast_mode);
+
+static int make_local_pal_md(int fast_mode)
+{
+	int pallen = 0xc0;
+
+	bgr444_to_rgb32(localPal, Pico.cram);
+	if (fast_mode)
+		return 0x40;
+
+	if (Pico.video.reg[0xC] & 8) { // shadow/hilight mode
+		bgr444_to_rgb32_sh(localPal, Pico.cram);
+		localPal[0xc0] = 0x0000c000;
+		localPal[0xd0] = 0x00c00000;
+		localPal[0xe0] = 0x00000000; // reserved pixels for OSD
+		localPal[0xf0] = 0x00ffffff;
+		pallen = 0x100;
+	}
+	else if (rendstatus & PDRAW_SONIC_MODE) { // mid-frame palette changes
+		bgr444_to_rgb32(localPal+0x40, HighPal);
+		bgr444_to_rgb32(localPal+0x80, HighPal+0x40);
+	}
+	else
+		memcpy32(localPal+0x80, localPal, 0x40); // for spr prio mess
+
+	return pallen;
+}
+
+static int make_local_pal_sms(int fast_mode)
+{
+	unsigned short *spal = Pico.cram;
+	unsigned int *dpal = (void *)localPal;
+	unsigned int i, t;
+
+	for (i = 0x40; i > 0; i--) {
+		t = *spal++;
+		t = ((t & 0x0003) << 22) | ((t & 0x000c) << 12) | ((t & 0x0030) << 2);
+		t |= t >> 2;
+		t |= t >> 4;
+		*dpal++ = t;
+	}
+
+	return 0x40;
+}
 
 void pemu_update_display(const char *fps, const char *notice)
 {
 	int emu_opt = currentConfig.EmuOpt;
+	int ret;
 
 	if (PicoOpt & POPT_ALT_RENDERER)
 	{
 		// 8bit fast renderer
 		if (Pico.m.dirtyPal) {
 			Pico.m.dirtyPal = 0;
-			vidConvCpyRGB32(localPal, Pico.cram, 0x40);
+			ret = make_local_pal(1);
 			// feed new palette to our device
-			gp2x_video_setpalette(localPal, 0x40);
+			gp2x_video_setpalette(localPal, ret);
 		}
 		// a hack for VR
 		if (PicoRead16Hook == PicoSVPRead16)
@@ -293,43 +338,17 @@ void pemu_update_display(const char *fps, const char *notice)
 		// 8bit accurate renderer
 		if (Pico.m.dirtyPal)
 		{
-			int pallen = 0xc0;
 			Pico.m.dirtyPal = 0;
-			if (Pico.video.reg[0xC]&8) // shadow/hilight mode
-			{
-				vidConvCpyRGB32(localPal, Pico.cram, 0x40);
-				vidConvCpyRGB32sh(localPal+0x40, Pico.cram, 0x40);
-				vidConvCpyRGB32hi(localPal+0x80, Pico.cram, 0x40);
-				memcpy32(localPal+0xc0, localPal+0x40, 0x40);
-				pallen = 0x100;
-			}
-			else if (rendstatus & PDRAW_SONIC_MODE) { // mid-frame palette changes
-				vidConvCpyRGB32(localPal, Pico.cram, 0x40);
-				vidConvCpyRGB32(localPal+0x40, HighPal, 0x40);
-				vidConvCpyRGB32(localPal+0x80, HighPal+0x40, 0x40);
-			}
-			else {
-				vidConvCpyRGB32(localPal, Pico.cram, 0x40);
-				memcpy32(localPal+0x80, localPal, 0x40); // for spr prio mess
-			}
-			if (pallen > 0xc0) {
-				localPal[0xc0] = 0x0000c000;
-				localPal[0xd0] = 0x00c00000;
-				localPal[0xe0] = 0x00000000; // reserved pixels for OSD
-				localPal[0xf0] = 0x00ffffff;
-			}
-			gp2x_video_setpalette(localPal, pallen);
+			ret = make_local_pal(0);
+			gp2x_video_setpalette(localPal, ret);
 		}
 	}
 
 	if (notice || (emu_opt & 2)) {
-		int h = 232;
-		if (currentConfig.scaling == EOPT_SCALE_HW_HV && !(Pico.video.reg[1]&8))
-			h -= 8;
 		if (notice)
-			osd_text(4, h, notice);
+			osd_text(4, osd_y, notice);
 		if (emu_opt & 2)
-			osd_text(osd_fps_x, h, fps);
+			osd_text(osd_fps_x, osd_y, fps);
 	}
 	if ((emu_opt & 0x400) && (PicoAHW & PAHW_MCD))
 		draw_cd_leds();
@@ -474,6 +493,9 @@ static void vidResetMode(void)
 	if (currentConfig.scaling == EOPT_SCALE_HW_HV && !(Pico.video.reg[1]&8))
 	     gp2x_video_RGB_setscaling(8, (PicoOpt&0x100)&&!(Pico.video.reg[12]&1) ? 256 : 320, 224);
 	else gp2x_video_RGB_setscaling(0, (PicoOpt&0x100)&&!(Pico.video.reg[12]&1) ? 256 : 320, 240);
+
+	// palette converters for 8bit modes
+	make_local_pal = (PicoAHW & PAHW_SMS) ? make_local_pal_sms : make_local_pal_md;
 }
 
 void plat_video_toggle_renderer(int is_next, int is_menu)
@@ -498,6 +520,7 @@ void plat_video_toggle_renderer(int is_next, int is_menu)
 		return;
 
 	vidResetMode();
+	rendstatus_old = -1;
 
 	if (PicoOpt & POPT_ALT_RENDERER) {
 		emu_status_msg(" 8bit fast renderer");
@@ -694,18 +717,21 @@ void plat_debug_cat(char *str)
 {
 }
 
-void pemu_video_mode_change(int is_32col, int is_240_lines)
+void emu_video_mode_change(int start_line, int line_count, int is_32cols)
 {
 	int scalex = 320;
 	osd_fps_x = OSD_FPS_X;
-	if (is_32col && (PicoOpt & POPT_DIS_32C_BORDER)) {
+	osd_y = 232;
+	if (is_32cols && (PicoOpt & POPT_DIS_32C_BORDER)) {
 		scalex = 256;
 		osd_fps_x = OSD_FPS_X - 64;
 	}
+
 	/* want vertical scaling and game is not in 240 line mode */
-	if (currentConfig.scaling == EOPT_SCALE_HW_HV && !is_240_lines)
-		gp2x_video_RGB_setscaling(8, scalex, 224);
-	else
+	if (currentConfig.scaling == EOPT_SCALE_HW_HV) {
+		gp2x_video_RGB_setscaling(start_line, scalex, line_count);
+		osd_y = start_line + line_count - 8;
+	} else
 		gp2x_video_RGB_setscaling(0, scalex, 240);
 
 	// clear whole screen in all buffers
@@ -819,6 +845,7 @@ void pemu_loop_end(void)
 	int eo_old = currentConfig.EmuOpt;
 
 	pemu_sound_stop();
+	memset32(g_screen_ptr, 0, 320*240*2/4);
 
 	/* do one more frame for menu bg */
 	PicoOpt &= ~POPT_ALT_RENDERER;
diff --git a/platform/linux/blit.c b/platform/linux/blit.c
index 7f466237..fbaac03f 100644
--- a/platform/linux/blit.c
+++ b/platform/linux/blit.c
@@ -3,37 +3,36 @@
 // to      00000000 rrr00000 ggg00000 bbb00000 ...
 // TODO: rm when gp2x/emu.c is no longer used
 
-void vidConvCpyRGB32  (void *to, void *from, int pixels)
+void bgr444_to_rgb32(void *to, void *from)
 {
 	unsigned short *ps = from;
 	unsigned int   *pd = to;
+	int pixels;
 
-	for (; pixels; pixels--, ps++, pd++)
+	for (pixels = 0x40; pixels; pixels--, ps++, pd++)
 	{
 		*pd = ((*ps<<20)&0xe00000) | ((*ps<<8)&0xe000) | ((*ps>>4)&0xe0);
 		*pd |= *pd >> 3;
 	}
 }
 
-void vidConvCpyRGB32sh(void *to, void *from, int pixels)
+void bgr444_to_rgb32_sh(void *to, void *from)
 {
 	unsigned short *ps = from;
 	unsigned int   *pd = to;
+	int pixels;
 
-	for (; pixels; pixels--, ps++, pd++)
+	pd += 0x40;
+	for (pixels = 0x40; pixels; pixels--, ps++, pd++)
 	{
 		*pd = ((*ps<<20)&0xe00000) | ((*ps<<8)&0xe000) | ((*ps>>4)&0xe0);
 		*pd >>= 1;
 		*pd |= *pd >> 3;
+		pd[0x40*2] = *pd;
 	}
-}
-
-void vidConvCpyRGB32hi(void *to, void *from, int pixels)
-{
-	unsigned short *ps = from;
-	unsigned int   *pd = to;
 
-	for (; pixels; pixels--, ps++, pd++)
+	ps -= 0x40;
+	for (pixels = 0x40; pixels; pixels--, ps++, pd++)
 	{
 		*pd = ((*ps<<20)&0xe00000) | ((*ps<<8)&0xe000) | ((*ps>>4)&0xe0);
 		continue;