From: kub <derkub@gmail.com>
Date: Wed, 13 Oct 2021 19:30:54 +0000 (+0200)
Subject: sms, basic gamegear support
X-Git-Tag: v2.00~472
X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=466fa079531d0605c73a65c4f499ef01aa9fdc7a;p=picodrive.git

sms, basic gamegear support
---

diff --git a/pico/draw.c b/pico/draw.c
index 3ae05c54..2b349228 100644
--- a/pico/draw.c
+++ b/pico/draw.c
@@ -135,7 +135,7 @@ void blockcpy_or(void *dst, void *src, size_t n, int pat)
   for (; n; n--)
     *pd++ = (unsigned char) (*ps++ | pat);
 }
-#define blockcpy memcpy
+#define blockcpy memmove
 #endif
 
 #define TileNormMaker_(pix_func,ret)                         \
@@ -1642,21 +1642,28 @@ void FinalizeLine555(int sh, int line, struct PicoEState *est)
 
   PicoDrawUpdateHighPal();
 
-  if (Pico.video.reg[12]&1) {
-    len = 320;
-  } else {
-    len = 256;
-  }
-
-  if ((*est->PicoOpt & POPT_EN_SOFTSCALE) && len == 256) {
-    switch (PicoIn.filter) {
-    case 3: h_upscale_bl4_4_5(pd, 320, ps, 256, 256, f_pal); break;
-    case 2: h_upscale_bl2_4_5(pd, 320, ps, 256, 256, f_pal); break;
-    case 1: h_upscale_snn_4_5(pd, 320, ps, 256, 256, f_pal); break;
-    default: h_upscale_nn_4_5(pd, 320, ps, 256, 256, f_pal); break;
-    }
+  if ((PicoIn.AHW & PAHW_SMS) && (Pico.m.hardware & 0x3) == 0x3)
+                                        len = 160;
+  else if (Pico.video.reg[12]&1)        len = 320;
+  else                                  len = 256;
+
+  if ((*est->PicoOpt & POPT_EN_SOFTSCALE) && len < 320) {
+    if (len == 256)
+      switch (PicoIn.filter) {
+      case 3: h_upscale_bl4_4_5(pd, 320, ps, 256, len, f_pal); break;
+      case 2: h_upscale_bl2_4_5(pd, 320, ps, 256, len, f_pal); break;
+      case 1: h_upscale_snn_4_5(pd, 320, ps, 256, len, f_pal); break;
+      default: h_upscale_nn_4_5(pd, 320, ps, 256, len, f_pal); break;
+      }
+    else if (len == 160)
+      switch (PicoIn.filter) {
+      case 3:
+      case 2: h_upscale_bl2_1_2(pd, 320, ps, 160, len, f_pal); break;
+      default: h_upscale_nn_1_2(pd, 320, ps, 160, len, f_pal); break;
+      }
   } else {
-    if (!(*est->PicoOpt & POPT_DIS_32C_BORDER) && len == 256) pd += 32;
+    if (!(*est->PicoOpt & POPT_DIS_32C_BORDER) && len < 320)
+      pd += (320-len) / 2;
 #if 1
     h_copy(pd, 320, ps, 320, len, f_pal);
 #else
@@ -1670,9 +1677,10 @@ void FinalizeLine555(int sh, int line, struct PicoEState *est)
 }
 #endif
 
-static void FinalizeLine8bit(int sh, int line, struct PicoEState *est)
+void FinalizeLine8bit(int sh, int line, struct PicoEState *est)
 {
   unsigned char *pd = est->DrawLineDest;
+  unsigned char *ps = est->HighCol+8;
   int len;
   static int dirty_line;
 
@@ -1680,8 +1688,8 @@ static void FinalizeLine8bit(int sh, int line, struct PicoEState *est)
   {
     // a hack for mid-frame palette changes
     if (!(est->rendstatus & PDRAW_SONIC_MODE) | (line - dirty_line > 4)) {
-      // store a maximum of 2 additional palettes in SonicPal
-      if (est->SonicPalCount < 2)
+      // store a maximum of 3 additional palettes in SonicPal
+      if (est->SonicPalCount < 3)
         est->SonicPalCount ++;
       dirty_line = line;
       est->rendstatus |= PDRAW_SONIC_MODE;
@@ -1690,35 +1698,33 @@ static void FinalizeLine8bit(int sh, int line, struct PicoEState *est)
     Pico.m.dirtyPal = 2;
   }
 
-  if (Pico.video.reg[12]&1) {
-    len = 320;
-  } else {
-    len = 256;
-  }
+  if ((PicoIn.AHW & PAHW_SMS) && (Pico.m.hardware & 0x3) == 0x3)
+                                        len = 160;
+  else if (Pico.video.reg[12]&1)        len = 320;
+  else                                  len = 256;
+
+  if (DrawLineDestIncrement == 0)
+    pd = est->HighCol+8;
 
-  if ((PicoIn.opt & POPT_EN_SOFTSCALE) && len == 256) {
-    unsigned char *ps = est->HighCol+8;
+  if ((PicoIn.opt & POPT_EN_SOFTSCALE) && len < 320) {
     unsigned char pal = 0;
 
     if (!sh && (est->rendstatus & PDRAW_SONIC_MODE))
       pal = est->SonicPalCount*0x40;
-    if (DrawLineDestIncrement == 0)
-      pd = est->HighCol+8;
     // Smoothing can't be used with CLUT, hence it's always Nearest Neighbour.
-    // use reverse version since src and dest ptr may be the same.
-    rh_upscale_nn_4_5(pd, 320, ps, 256, len, f_or);
-  } else if (DrawLineDestIncrement == 0) {
-    if (!sh && (est->rendstatus & PDRAW_SONIC_MODE))
-      blockcpy_or(est->HighCol+8, est->HighCol+8, len, est->SonicPalCount*0x40);
+    if (len == 256)
+      // use reverse version since src and dest ptr may be the same.
+      rh_upscale_nn_4_5(pd, 320, ps, 256, len, f_or);
+    else
+      rh_upscale_nn_1_2(pd, 320, ps, 256, len, f_or);
   } else {
-    if (!(PicoIn.opt & POPT_DIS_32C_BORDER))
-      pd += 32;
-    if (!sh && (est->rendstatus & PDRAW_SONIC_MODE)) {
+    if (!(*est->PicoOpt & POPT_DIS_32C_BORDER) && len < 320)
+      pd += (320-len) / 2;
+    if (!sh && (est->rendstatus & PDRAW_SONIC_MODE))
       // select active backup palette
-      blockcpy_or(pd, est->HighCol+8, len, est->SonicPalCount*0x40);
-    } else {
-      blockcpy(pd, est->HighCol+8, len);
-    }
+      blockcpy_or(pd, ps, len, est->SonicPalCount*0x40);
+    else if (pd != ps)
+      blockcpy(pd, ps, len);
   }
 }
 
@@ -1893,8 +1899,15 @@ PICO_INTERNAL void PicoFrameStart(void)
 
 static void DrawBlankedLine(int line, int offs, int sh, int bgc)
 {
-  if (PicoScanBegin != NULL)
-    PicoScanBegin(line + offs);
+  int skip = skip_next_line;
+
+  if (PicoScanBegin != NULL && skip == 0)
+    skip = PicoScanBegin(line + offs);
+
+  if (skip) {
+    skip_next_line = skip - 1;
+    return;
+  }
 
   BackFill(bgc, sh, &Pico.est);
 
@@ -1902,7 +1915,7 @@ static void DrawBlankedLine(int line, int offs, int sh, int bgc)
     FinalizeLine(sh, line, &Pico.est);
 
   if (PicoScanEnd != NULL)
-    PicoScanEnd(line + offs);
+    skip_next_line = PicoScanEnd(line + offs);
 
   Pico.est.HighCol += HighColIncrement;
   Pico.est.DrawLineDest = (char *)Pico.est.DrawLineDest + DrawLineDestIncrement;
@@ -1910,15 +1923,10 @@ static void DrawBlankedLine(int line, int offs, int sh, int bgc)
 
 static void PicoLine(int line, int offs, int sh, int bgc)
 {
-  int skip = 0;
-
-  if (skip_next_line > 0) {
-    skip_next_line--;
-    return;
-  }
+  int skip = skip_next_line;
 
   Pico.est.DrawScanline = line;
-  if (PicoScanBegin != NULL)
+  if (PicoScanBegin != NULL && skip == 0)
     skip = PicoScanBegin(line + offs);
 
   if (skip) {
diff --git a/pico/draw_arm.S b/pico/draw_arm.S
index 38d831f6..6a3641bd 100644
--- a/pico/draw_arm.S
+++ b/pico/draw_arm.S
@@ -1642,27 +1642,42 @@ FinalizeLine555:
     add     r3, r10, #OFS_EST_HighPal
 
     mov     lr, #0xff
+    mov     lr, lr, lsl #1
 
+    ldr     r5, [r10, #OFS_EST_PicoOpt]
     ldr     r1, [r10, #OFS_EST_HighCol]
     ldr     r0, [r10, #OFS_EST_DrawLineDest]
+    ldr     r4, [r5]
+    ldr     r7, [r5, #OFS_PicoIn_AHW-OFS_PicoIn_opt]
+    ldrb    r12,[r8, #OFS_Pico_video_reg+12]
+    ldr     r2, [r8, #OFS_Pico_m_hardware]
     add     r1, r1, #8
 
-    ldrb    r12, [r8, #OFS_Pico_video_reg+12]
-    mov     lr, lr, lsl #1
+    tst     r7, #0x10
+    beq     .fl_no20colRGB555
+    and     r7, r2, #0x3
+    cmp     r7, #0x3		@ Game Gear, LCD?
+    bne     .fl_no20colRGB555
 
-    tst     r12, #1
-    movne   r2, #320/8           @ len
+    mov     r2, #160/8		@ len = 160
+    tst     r4, #0x4000		@ EN_SOFTSCALE?
+    bne     .fl_20scale_RGB555	@ scale 160->320
+    beq     .fl_checkborder
+
+.fl_no20colRGB555:
+    tst     r12, #1		@ h32?
+    movne   r2, #320/8          @ len = 320
     bne     .fl_no32colRGB555
-    ldr     r5, [r10, #OFS_EST_PicoOpt]
-    mov     r2, #256/8
-    ldr     r4, [r5]
-    tst     r4, #0x4000
-    bne     .fl_32scale_RGB555
-    tst     r4, #0x0100
-    addeq   r0, r0, #32*2
+    moveq   r2, #256/8		@ len = 256
+    tst     r4, #0x4000		@ EN_SOFTSCALE?
+    bne     .fl_32scale_RGB555	@ scale 256->320
 
-.fl_no32colRGB555:
+.fl_checkborder:
+    tst     r4, #0x0100		@ DIS_32C_BORDER?
+    rsbeq   r4, r2, #320/8	@ pd += (320-len)/2
+    addeq   r0, r0, r4, lsl #3
 
+.fl_no32colRGB555:
 #ifdef UNALIGNED_DRAWLINEDEST
     @ this is basically for Gizmondo, which has unaligned odd lines in the framebuffer
     tst     r0, #2
@@ -1715,9 +1730,9 @@ FinalizeLine555:
     bne     .fl_32scale_RGB555u
 #endif
 
-    ands    r5, r5, #0x3
-    addne   pc, pc, r5, lsl #2
-    b       .fl_32scale_nn
+    and     r5, r5, #0x3
+    add     pc, pc, r5, lsl #2
+    nop
     b       .fl_32scale_nn
     b       .fl_32scale_snn
     b       .fl_32scale_bl2
@@ -1866,6 +1881,9 @@ FinalizeLine555:
     ldmfd   sp!, {r4-r10,pc}
 
 .fl_32scale_bl4:
+    // TODO this should reflect the bl4 C algorithm, but it doesn't, it's bln.
+    and     r9, r9, r9, lsl #1        @ nuke 2 LSBs to avoid spilling for n/4
+.fl_32loop_bl4:
     ldr     r12, [r1], #4
     ldr     r7,  [r1], #4
 
@@ -1926,10 +1944,147 @@ FinalizeLine555:
     subs    r2, r2, #1
 
     stmia   r0!, {r4,r5,r6,r8,r10}
-    bne     .fl_32scale_bl4
+    bne     .fl_32loop_bl4
+
+    ldmfd   sp!, {r4-r10,pc}
+
+.fl_20scale_RGB555:
+    ldr     r5, [r5, #OFS_PicoIn_filter-OFS_PicoIn_opt]
+
+    mov     r9, #0xf700 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007
+    orr     r9, r9, #0x00de
+
+#ifdef UNALIGNED_DRAWLINEDEST
+    tst     r0, #2
+    bne     .fl_20scale_RGB555u
+#endif
+
+    and     r5, r5, #0x2
+    add     pc, pc, r5, lsl #1
+    nop
+    b       .fl_20scale_nn
+    b       .fl_20scale_bl2
+
+.fl_20scale_nn:
+    ldr     r12, [r1], #4
+    ldr     r7,  [r1], #4
+
+    and     r4, lr, r12, lsl #1
+    ldrh    r4, [r3, r4]
+    and     r5, lr, r12, lsr #7
+    ldrh    r5, [r3, r5]
+    and     r6, lr, r12, lsr #15
+    ldrh    r6, [r3, r6]
+    and     r8 ,lr, r12, lsr #23
+    ldrh    r8 ,[r3, r8 ]
+
+    orr     r4, r4, r4, lsl #16
+    orr     r5, r5, r5, lsl #16
+    orr     r6, r6, r6, lsl #16
+    orr     r8, r8, r8, lsl #16
+    stmia   r0!, {r4,r5,r6,r8}
+
+    and     r4, lr, r7, lsl #1
+    ldrh    r4, [r3, r4]
+    and     r5, lr, r7, lsr #7
+    ldrh    r5, [r3, r5]
+    and     r6 ,lr, r7, lsr #15
+    ldrh    r6 ,[r3, r6]
+    and     r8, lr, r7, lsr #23
+    ldrh    r8, [r3, r8]
+
+    orr     r4, r4, r4, lsl #16
+    orr     r5, r5, r5, lsl #16
+    orr     r6, r6, r6, lsl #16
+    orr     r8, r8, r8, lsl #16
+    stmia   r0!, {r4,r5,r6,r8}
+
+    subs    r2, r2, #1
+    bne     .fl_20scale_nn
+
+    ldmfd   sp!, {r4-r10,pc}
+
+
+.fl_20scale_bl2:
+    ldr     r8,  [r1]
+    and     r8, lr, r8, lsl #1
+    ldrh    r8, [r3, r8]
+    and     r8, r8, r9
+    lsl     r8, r8, #16
+
+.fl_20loop_bl2:
+    ldr     r12, [r1], #4
+    ldr     r7,  [r1], #4
+
+    and     r4, lr, r12, lsl #1
+    ldrh    r4, [r3, r4]
+    and     r5, lr, r12, lsr #7
+    ldrh    r5, [r3, r5]
+    and     r6, lr, r12, lsr #15
+    ldrh    r6, [r3, r6]
+
+    and     r4, r4, r9
+    add     r10,r4, r8, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r4, r10,r4, lsl #16		@ (px-1+px0)/2 | px0
+
+    and     r8 ,lr, r12, lsr #23
+    ldrh    r8 ,[r3, r8]
+
+    and     r5, r5, r9
+    add     r10,r5, r4, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r5, r10,r5, lsl #16		@ (px0 +px1)/2 | px1
+
+    and     r6, r6, r9
+    add     r10,r6, r5, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r6, r10,r6, lsl #16		@ (px1 +px2)/2 | px2
+
+    and     r8, r8, r9
+    add     r10,r8, r6, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r8, r10,r8, lsl #16		@ (px2 +px3)/2 | px3
+
+    stmia   r0!, {r4,r5,r6,r8}
+
+    and     r4, lr, r7, lsl #1
+    ldrh    r4, [r3, r4]
+    and     r5, lr, r7, lsr #7
+    ldrh    r5, [r3, r5]
+    and     r6, lr, r7, lsr #15
+    ldrh    r6, [r3, r6]
+
+    and     r4, r4, r9
+    add     r10,r4, r8, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r4, r10,r4, lsl #16		@ (px-1+px0)/2 | px0
+
+    and     r8 ,lr, r7, lsr #23
+    ldrh    r8 ,[r3, r8]
+
+    and     r5, r5, r9
+    add     r10,r5, r4, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r5, r10,r5, lsl #16		@ (px0 +px1)/2 | px1
+
+    and     r6, r6, r9
+    add     r10,r6, r5, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r6, r10,r6, lsl #16		@ (px1 +px2)/2 | px2
+
+    and     r8, r8, r9
+    add     r10,r8, r6, lsr #16
+    mov     r10,r10,lsr #1
+    orr     r8, r10,r8, lsl #16		@ (px2 +px3)/2 | px3
+
+    subs    r2, r2, #1
+    stmia   r0!, {r4,r5,r6,r8}
+    bne     .fl_20loop_bl2
 
     ldmfd   sp!, {r4-r10,pc}
 
+
 #ifdef UNALIGNED_DRAWLINEDEST
     @ unaligned versions of loops
     @ warning: starts drawing 2bytes before dst
diff --git a/pico/media.c b/pico/media.c
index f4ad8cde..d7e91a0f 100644
--- a/pico/media.c
+++ b/pico/media.c
@@ -248,7 +248,6 @@ enum media_type_e PicoLoadMedia(const char *filename,
     }
   }
   else if (media_type == PM_MARK3) {
-    lprintf("detected SMS ROM\n");
     PicoIn.AHW = PAHW_SMS;
   }
 
@@ -297,9 +296,18 @@ enum media_type_e PicoLoadMedia(const char *filename,
     goto out;
   }
   rom_data = NULL; // now belongs to PicoCart
-  Pico.m.ncart_in = 0;
+
+  // simple test for GG. Do this here since m.hardware is nulled in Insert
+  if (PicoIn.AHW & PAHW_SMS) {
+    if (strstr(filename,".gg")) {
+      Pico.m.hardware |= 0x1;
+      lprintf("detected GG ROM\n");
+    } else
+      lprintf("detected SMS ROM\n");
+  }
 
   // insert CD if it was detected
+  Pico.m.ncart_in = 0;
   if (cd_img_type != CT_UNKNOWN) {
     ret = cdd_load(filename, cd_img_type);
     if (ret != 0) {
diff --git a/pico/mode4.c b/pico/mode4.c
index 889a43e0..2406dd4a 100644
--- a/pico/mode4.c
+++ b/pico/mode4.c
@@ -141,6 +141,8 @@ static void DrawSpritesM4(int scanline)
   if (pv->reg[0] & 8)
     xoff = 0;
   xoff += line_offset;
+  if ((Pico.m.hardware & 0x3) == 0x3)
+    xoff -= 48; // GG LCD, adjust to center 160 px
 
   sat = (u8 *)PicoMem.vram + ((pv->reg[5] & 0x7e) << 7);
   if (pv->reg[1] & 2) {
@@ -164,10 +166,12 @@ static void DrawSpritesM4(int scanline)
       break;
     }
 
-    sprites_x[s] = xoff + sat[MEM_LE2(0x80 + i*2)];
-    sprites_addr[s] = sprite_base + ((sat[MEM_LE2(0x80 + i*2 + 1)] & addr_mask) << (5-1)) +
-      ((scanline - y) >> zoomed << (2-1));
-    s++;
+    if (xoff + sat[MEM_LE2(0x80 + i*2)] >= 0) {
+      sprites_x[s] = xoff + sat[MEM_LE2(0x80 + i*2)];
+      sprites_addr[s] = sprite_base + ((sat[MEM_LE2(0x80 + i*2 + 1)] & addr_mask) << (5-1)) +
+        ((scanline - y) >> zoomed << (2-1));
+      s++;
+    }
   }
 
   // really half-assed but better than nothing
@@ -296,9 +300,12 @@ static void DrawDisplayM4(int scanline)
 
   // low priority tiles
   if (!(pv->debug_p & PVD_KILL_B)) {
-    if (pv->reg[0] & 0x80) {
+    if ((Pico.m.hardware & 0x3) == 0x3) {
+      // on GG render only the center 160 px
+      DrawStripLowM4(nametab , dx | ((cells-12)<< 16),(tilex+6) | (ty  << 16));
+    } else if (pv->reg[0] & 0x80) {
       // vscroll disabled for rightmost 8 columns (e.g. Gauntlet)
-      int dx2 = dx + (cells-8)*8, tilex2 = tilex + (cells-8), ty2 = scanline & 7;
+      int dx2 = dx + (cells-8)*8, tilex2 = tilex + (cells-8), ty2 = scanline&7;
       DrawStripLowM4(nametab,  dx | ((cells-8) << 16), tilex  | (ty  << 16));
       DrawStripLowM4(nametab2, dx2 |       (8  << 16), tilex2 | (ty2 << 17));
     } else
@@ -311,8 +318,10 @@ static void DrawDisplayM4(int scanline)
 
   // high priority tiles (use virtual layer switch just for fun)
   if (!(pv->debug_p & PVD_KILL_A)) {
-    if (pv->reg[0] & 0x80) {
-      int dx2 = dx + (cells-8)*8, tilex2 = tilex + (cells-8), ty2 = scanline & 7;
+    if ((Pico.m.hardware & 0x3) == 0x3) {
+      DrawStripHighM4(nametab , dx | ((cells-12)<< 16),(tilex+6) | (ty  << 16));
+    } else if (pv->reg[0] & 0x80) {
+      int dx2 = dx + (cells-8)*8, tilex2 = tilex + (cells-8), ty2 = scanline&7;
       DrawStripHighM4(nametab,  dx | ((cells-8) << 16), tilex  | (ty  << 16));
       DrawStripHighM4(nametab2, dx2 |       (8  << 16), tilex2 | (ty2 << 17));
     } else
@@ -451,20 +460,21 @@ static void DrawSpritesM2(int scanline)
 
   // now draw all sprites backwards
   for (--s; s >= 0; s--) {
-    int x, w = (zoomed ? 16: 8);
+    int x, c, w = (zoomed ? 16: 8);
     i = sprites_x[s];
     x = sat[MEM_LE2(i+1)] + xoff;
     if (sat[MEM_LE2(i+3)] & 0x80)
       x -= 32;
+    c = sat[MEM_LE2(i+3)] & 0x0f;
     if (x > 0) {
       pack = PicoMem.vramb[MEM_LE2(sprites_addr[s])];
-      if (zoomed) TileDoubleSprM2(x, pack, sat[MEM_LE2(i+3)] & 0xf);
-      else        TileNormSprM2(x, pack, sat[MEM_LE2(i+3)] & 0xf);
+      if (zoomed) TileDoubleSprM2(x, pack, c);
+      else        TileNormSprM2(x, pack, c);
     }
     if((pv->reg[1] & 0x2) && (x+=w) > 0) {
       pack = PicoMem.vramb[MEM_LE2(sprites_addr[s]+0x10)];
-      if (zoomed) TileDoubleSprM2(x, pack, sat[MEM_LE2(i+3)] & 0xf);
-      else        TileNormSprM2(x, pack, sat[MEM_LE2(i+3)] & 0xf);
+      if (zoomed) TileDoubleSprM2(x, pack, c);
+      else        TileNormSprM2(x, pack, c);
     }
   }
 }
@@ -524,34 +534,45 @@ static void FinalizeLineRGB555SMS(int line);
 
 void PicoFrameStartSMS(void)
 {
-  int lines = 192, columns = 256, coffs;
+  int lines = 192, columns = 256, loffs, coffs;
   skip_next_line = 0;
-  screen_offset = 24;
+  loffs = screen_offset = 24; // 192 lines is really 224 with top/bottom bars
   Pico.est.rendstatus = PDRAW_32_COLS;
 
-  switch ((Pico.video.reg[0]&0x06) | (Pico.video.reg[1]&0x18)) {
+  // Copy LCD enable flag for easier handling
+  Pico.m.hardware &= ~0x2;
+  if (PicoIn.opt & POPT_EN_GG_LCD)
+    Pico.m.hardware |= 0x2;
+
+  if ((Pico.m.hardware & 0x3) == 0x3) {
+    // GG LCD always has 160x144 regardless of settings
+    screen_offset = 24; // nonetheless the vdp timing has 224 lines
+    loffs = 48;
+    lines = 144;
+    columns = 160;
+  } else switch ((Pico.video.reg[0]&0x06) | (Pico.video.reg[1]&0x18)) {
   // SMS2 only 224/240 line modes, e.g. Micro Machines
   case 0x06|0x08:
-      screen_offset = 0;
+      loffs = screen_offset = 0;
       lines = 240;
       break;
   case 0x06|0x10:
-      screen_offset = 8;
+      loffs = screen_offset = 8;
       lines = 224;
       break;
   }
   if (PicoIn.opt & POPT_EN_SOFTSCALE) {
-    line_offset = 0;
+    coffs = 0;
     columns = 320;
   } else
-    line_offset = PicoIn.opt & POPT_DIS_32C_BORDER ? 0 : 32;
+    coffs = PicoIn.opt & POPT_DIS_32C_BORDER ? 0:(320-columns)/2;
+  line_offset = (PicoIn.opt & POPT_ALT_RENDERER ? coffs : 0);
 
-  coffs = line_offset;
   if (FinalizeLineSMS == FinalizeLineRGB555SMS)
     line_offset = 0 /* done in FinalizeLine */;
 
   if (Pico.est.rendstatus != rendstatus_old || lines != rendlines) {
-    emu_video_mode_change(screen_offset, lines, coffs, columns);
+    emu_video_mode_change(loffs, lines, coffs, columns);
     rendstatus_old = Pico.est.rendstatus;
     rendlines = lines;
   }
@@ -562,14 +583,20 @@ void PicoFrameStartSMS(void)
 
 void PicoLineSMS(int line)
 {
-  if (skip_next_line > 0) {
-    skip_next_line--;
+  int skip = skip_next_line;
+
+  // GG LCD, render only visible part of screen
+  if ((Pico.m.hardware & 0x3) == 0x3 && (line < 24 || line >= 24+144))
+    goto norender;
+
+  if (PicoScanBegin != NULL && skip == 0)
+    skip = PicoScanBegin(line + screen_offset);
+
+  if (skip) {
+    skip_next_line = skip - 1;
     return;
   }
 
-  if (PicoScanBegin != NULL)
-    skip_next_line = PicoScanBegin(line + screen_offset);
-
   // Draw screen:
   BackFill(Pico.video.reg[7] & 0x0f, 0, &Pico.est);
   if (Pico.video.reg[1] & 0x40) {
@@ -583,19 +610,21 @@ void PicoLineSMS(int line)
   if (PicoScanEnd != NULL)
     skip_next_line = PicoScanEnd(line + screen_offset);
 
+norender:
   Pico.est.HighCol += HighColIncrement;
   Pico.est.DrawLineDest = (char *)Pico.est.DrawLineDest + DrawLineDestIncrement;
 }
 
 /* Fixed palette for TMS9918 modes */
 static u16 tmspal[32] = {
- 0x00,0x00,0x08,0x0c,0x10,0x30,0x01,0x3c,0x02,0x03,0x05,0x0f,0x04,0x33,0x15,0x3f
+  0x0000, 0x0000, 0x00a0, 0x00f0, 0x0500, 0x0f00, 0x0005, 0x0ff0,
+  0x000a, 0x000f, 0x0055, 0x00ff, 0x0050, 0x0f0f, 0x0555, 0x0fff,
 };
 
 void PicoDoHighPal555SMS(void)
 {
-  unsigned int *spal=(void *)PicoMem.cram;
-  unsigned int *dpal=(void *)Pico.est.HighPal;
+  u32 *spal=(void *)PicoMem.cram;
+  u32 *dpal=(void *)Pico.est.HighPal;
   unsigned int t;
   int i;
 
@@ -603,21 +632,9 @@ void PicoDoHighPal555SMS(void)
   if (!(Pico.video.reg[0] & 0x4))
     spal = (u32 *)tmspal;
 
-  /* cram is always stored as shorts, even though real hardware probably uses bytes */
-  if (PicoIn.AHW & PAHW_SMS) for (i = 0x20/2; i > 0; i--, spal++, dpal++) { 
-    t = *spal;
-#if defined(USE_BGR555)
-    t = ((t & 0x00030003)<< 3) | ((t & 0x000c000c)<<6) | ((t & 0x00300030)<<9);
-    t |= (t >> 2) | ((t >> 4) & 0x04210421);
-#elif defined(USE_BGR565)
-    t = ((t & 0x00030003)<< 3) | ((t & 0x000c000c)<<7) | ((t & 0x00300030)<<10);
-    t |= (t >> 2) | ((t >> 4) & 0x08610861);
-#else
-    t = ((t & 0x00030003)<<14) | ((t & 0x000c000c)<<7) | ((t & 0x00300030)>>1);
-    t |= (t >> 2) | ((t >> 4) & 0x08610861);
-#endif
-    *dpal = t;
-  } else for (i = 0x20/2; i > 0; i--, spal++, dpal++) { // GG palette 4 bit/col
+  /* SMS 6 bit cram data was already converted to MD/GG format by vdp write,
+   * hence GG/SMS/TMS can all be handled the same here */
+  for (i = 0x20/2; i > 0; i--, spal++, dpal++) { 
     t = *spal;
 #if defined(USE_BGR555)
     t = ((t & 0x000f000f)<< 1) | ((t & 0x00f000f0)<<2) | ((t & 0x0f000f00)<<3);
@@ -646,15 +663,7 @@ static void FinalizeLineRGB555SMS(int line)
 
 static void FinalizeLine8bitSMS(int line)
 {
-  u8 *pd = Pico.est.DrawLineDest + line_offset;
-  u8 *ps = Pico.est.HighCol + line_offset + 8;
-
-  if (DrawLineDestIncrement) {
-    if (PicoIn.opt & POPT_EN_SOFTSCALE)
-      rh_upscale_nn_4_5(pd, 320, ps, 256, 256, f_nop);
-    else if (pd != ps)
-      memcpy(pd, ps, 256);
-  }
+  FinalizeLine8bit(0, line, &Pico.est);
 }
 
 void PicoDrawSetOutputSMS(pdso_t which)
diff --git a/pico/pico.h b/pico/pico.h
index 76786d61..dc615d1e 100644
--- a/pico/pico.h
+++ b/pico/pico.h
@@ -64,7 +64,7 @@ extern void *p32x_bios_g, *p32x_bios_m, *p32x_bios_s;
 #define POPT_EN_MCD_PCM     (1<<10)
 #define POPT_EN_MCD_CDDA    (1<<11)
 #define POPT_EN_MCD_GFX     (1<<12) // 00 x000
-// unused                   (1<<13)
+#define POPT_EN_GG_LCD      (1<<13)
 #define POPT_EN_SOFTSCALE   (1<<14)
 #define POPT_EN_MCD_RAMCART (1<<15)
 #define POPT_DIS_VDP_FIFO   (1<<16) // 0x 0000
diff --git a/pico/pico_int.h b/pico/pico_int.h
index 9a66344c..e6019296 100644
--- a/pico/pico_int.h
+++ b/pico/pico_int.h
@@ -670,6 +670,7 @@ PICO_INTERNAL void PicoFrameStart(void);
 void PicoDrawSync(int to, int blank_last_line);
 void BackFill(int reg7, int sh, struct PicoEState *est);
 void FinalizeLine555(int sh, int line, struct PicoEState *est);
+void FinalizeLine8bit(int sh, int line, struct PicoEState *est);
 void PicoDrawSetOutBufMD(void *dest, int increment);
 extern int (*PicoScanBegin)(unsigned int num);
 extern int (*PicoScanEnd)(unsigned int num);
diff --git a/pico/sms.c b/pico/sms.c
index 5f175ca6..6c8ce358 100644
--- a/pico/sms.c
+++ b/pico/sms.c
@@ -52,8 +52,17 @@ static void vdp_data_write(unsigned char d)
   struct PicoVideo *pv = &Pico.video;
 
   if (pv->type == 3) {
-    if (PicoMem.cram[pv->addr & 0x1f] != d) Pico.m.dirtyPal = 1;
-    PicoMem.cram[pv->addr & 0x1f] = d;
+    if (Pico.m.hardware & 0x1) { // GG, same layout as MD
+      unsigned a = pv->addr & 0x3f;
+      if (a & 0x1) d &= 0x0f;
+      if (((u8 *)PicoMem.cram)[MEM_LE2(a)] != d) Pico.m.dirtyPal = 1;
+      ((u8 *)PicoMem.cram)[MEM_LE2(a)] = d;
+    } else { // SMS, convert to MD layout (00BbGgRr to 0000BbBbGgGgRrRr)
+      unsigned a = pv->addr & 0x1f;
+      u16 c = (d&0x30)*0x40 + (d&0x0c)*0x10 + (d&0x03)*0x04;
+      if (PicoMem.cram[a] != (c | (c>>2))) Pico.m.dirtyPal = 1;
+      PicoMem.cram[a] = c | (c>>2);
+    }
   } else {
     PicoMem.vramb[MEM_LE2(pv->addr)] = d;
   }
@@ -130,7 +139,7 @@ static unsigned char z80_sms_in(unsigned short a)
     {
       case 0x00:
       case 0x01:
-        d = 0xff;
+        d = 0xff & ~(PicoIn.pad[0] & 0x80);
         break;
 
       case 0x40: /* V counter */
diff --git a/platform/common/arm_utils.s b/platform/common/arm_utils.s
index 6696e5af..5f0c46c3 100644
--- a/platform/common/arm_utils.s
+++ b/platform/common/arm_utils.s
@@ -58,9 +58,6 @@
 .endif
 
     orr     r2,  r2,   r2,  lsr #3
-.if \sh == 1
-    str     r2, [r0, #0x40*2*4]
-.endif
     str     r2, [r0], #4
 .endm
 
@@ -101,10 +98,10 @@ bgr444_to_rgb32_sh:
     subs    r12, r12, #1
 
     ldmia    r1!, {r4-r7}
-    convRGB32_2 r4, 1
-    convRGB32_2 r5, 1
-    convRGB32_2 r6, 1
-    convRGB32_2 r7, 1
+    convRGB32_2 r4, 2
+    convRGB32_2 r5, 2
+    convRGB32_2 r6, 2
+    convRGB32_2 r7, 2
     bgt     .loopRGB32sh
 
     mov     r12, #0x40>>3 @ repeats
@@ -112,10 +109,10 @@ bgr444_to_rgb32_sh:
 
 .loopRGB32hi:
      ldmia    r1!, {r4-r7}
-    convRGB32_2 r4, 2
-    convRGB32_2 r5, 2
-    convRGB32_2 r6, 2
-    convRGB32_2 r7, 2
+    convRGB32_2 r4, 1
+    convRGB32_2 r5, 1
+    convRGB32_2 r6, 1
+    convRGB32_2 r7, 1
 
     subs    r12, r12, #1
     bgt     .loopRGB32hi
diff --git a/platform/common/emu.c b/platform/common/emu.c
index fd4a2519..1ae8cdb5 100644
--- a/platform/common/emu.c
+++ b/platform/common/emu.c
@@ -326,6 +326,8 @@ static void system_announce(void)
 
 	if (PicoIn.AHW & PAHW_SMS) {
 		sys_name = "Master System";
+		if (Pico.m.hardware & 0x1)
+			sys_name = "Game Gear";
 #ifdef NO_SMS
 		extra = " [no support]";
 #endif
@@ -587,7 +589,7 @@ void emu_prep_defconfig(void)
 	memset(&defaultConfig, 0, sizeof(defaultConfig));
 	defaultConfig.EmuOpt    = EOPT_EN_SRAM | EOPT_EN_SOUND | EOPT_16BPP |
 				  EOPT_EN_CD_LEDS | EOPT_GZIP_SAVES | 0x10/*?*/;
-	defaultConfig.s_PicoOpt = POPT_EN_SNDFILTER|POPT_EN_YM2413|
+	defaultConfig.s_PicoOpt = POPT_EN_SNDFILTER|POPT_EN_YM2413|POPT_EN_GG_LCD |
 				  POPT_EN_STEREO|POPT_EN_FM|POPT_EN_PSG|POPT_EN_Z80 |
 				  POPT_EN_MCD_PCM|POPT_EN_MCD_CDDA|POPT_EN_MCD_GFX |
 				  POPT_EN_DRC|POPT_ACC_SPRITES |
diff --git a/platform/common/menu_pico.c b/platform/common/menu_pico.c
index 6a6da396..c6162987 100644
--- a/platform/common/menu_pico.c
+++ b/platform/common/menu_pico.c
@@ -42,7 +42,7 @@ static const char *rom_exts[] = {
 	"bin", "smd", "gen", "md",
 	"iso", "cso", "cue", "chd",
 	"32x",
-	"sms",
+	"sms", "gg",
 	NULL
 };
 
@@ -536,6 +536,7 @@ static menu_entry e_menu_adv_options[] =
 	mee_onoff     ("Disable YM2612 SSG-EG",    MA_OPT2_DISABLE_YM_SSG,PicoIn.opt, POPT_DIS_FM_SSGEG),
 	mee_onoff     ("Emulate SN76496 (PSG)",    MA_OPT2_ENABLE_SN76496,PicoIn.opt, POPT_EN_PSG),
 	mee_onoff     ("Emulate YM2413 (FM)",      MA_OPT2_ENABLE_YM2413 ,PicoIn.opt, POPT_EN_YM2413),
+	mee_onoff     ("Emulate Game Gear LCD",    MA_OPT2_ENABLE_GGLCD  ,PicoIn.opt, POPT_EN_GG_LCD),
 	mee_onoff     ("Disable idle loop patching",MA_OPT2_NO_IDLE_LOOPS,PicoIn.opt, POPT_DIS_IDLE_DET),
 	mee_onoff     ("Disable frame limiter",    MA_OPT2_NO_FRAME_LIMIT,currentConfig.EmuOpt, EOPT_NO_FRMLIMIT),
 	mee_onoff     ("Enable dynarecs",          MA_OPT2_DYNARECS,      PicoIn.opt, POPT_EN_DRC),
diff --git a/platform/common/menu_pico.h b/platform/common/menu_pico.h
index 2b699ef2..af365dbb 100644
--- a/platform/common/menu_pico.h
+++ b/platform/common/menu_pico.h
@@ -52,6 +52,7 @@ typedef enum
 	MA_OPT2_DISABLE_YM_SSG,
 	MA_OPT2_ENABLE_SN76496,
 	MA_OPT2_ENABLE_YM2413,
+	MA_OPT2_ENABLE_GGLCD,
 	MA_OPT2_NO_LAST_ROM,
 	MA_OPT2_RAMTIMINGS,	/* gp2x */
 	MA_OPT2_NO_FRAME_LIMIT,	/* psp */
diff --git a/platform/common/upscale.h b/platform/common/upscale.h
index 9719981e..1cf0fd9c 100644
--- a/platform/common/upscale.h
+++ b/platform/common/upscale.h
@@ -35,12 +35,20 @@
  */
 #include <pico/pico_types.h>
 
+
+/* LSB of all colors in a pixel */
+#if defined(USE_BGR555)
+#define PXLSB		0x0421
+#else
+#define PXLSB		0x0821
+#endif
+
 /* RGB565 pixel mixing, see https://www.compuphase.com/graphic/scale3.htm and
   			    http://blargg.8bitalley.com/info/rgb_mixing.html */
 /* 2-level mixing */
-//#define p_05(d,p1,p2)	d=(((p1)+(p2)  + ( ((p1)^(p2))&0x0821))>>1) // round up
-//#define p_05(d,p1,p2)	d=(((p1)+(p2)  - ( ((p1)^(p2))&0x0821))>>1) // round down
-#define p_05(d,p1,p2)	d=(((p1)&(p2)) + ((((p1)^(p2))&~0x0821)>>1))
+//#define p_05(d,p1,p2)	d=(((p1)+(p2)  + ( ((p1)^(p2))&PXLSB))>>1) // round up
+//#define p_05(d,p1,p2)	d=(((p1)+(p2)  - ( ((p1)^(p2))&PXLSB))>>1) // round down
+#define p_05(d,p1,p2)	d=(((p1)&(p2)) + ((((p1)^(p2))&~PXLSB)>>1))
 /* 4-level mixing, 2 times slower */
 // 1/4*p1 + 3/4*p2 = 1/2*(1/2*(p1+p2) + p2)
 #define p_025(d,p1,p2)	p_05(t, p1, p2); p_05( d, t, p2)
@@ -322,6 +330,21 @@ scalers h:
 	si += ss - w;					\
 } while (0)
 
+// reverse version for overlapping buffers
+#define rh_upscale_nn_1_2(di,ds,si,ss,w,f) do {	\
+	int i;						\
+	di += w*2;					\
+	si += w;					\
+	for (i = w/2; i > 0; i--, si -= 2, di -= 4) {	\
+		di[-1] = f(si[-1]);			\
+		di[-2] = f(si[-1]);			\
+		di[-3] = f(si[-2]);			\
+		di[-4] = f(si[-2]);			\
+	}						\
+	di += ds;					\
+	si += ss;					\
+} while (0)
+
 #define h_upscale_bl2_1_2(di,ds,si,ss,w,f) do {		\
 	int i; uint p = f(si[0]);			\
 	for (i = w/2; i > 0; i--, si += 2, di += 4) {	\
@@ -514,11 +537,12 @@ scalers v:
 	} else  {					\
 		int j;					\
 		l = 0;					\
-		di -= 4*ds;				\
+		di -= 3*ds;				\
 		for (j = 0; j < 2; j++) {		\
 			v_copy(&di[0], &di[-ds], w, f_nop); \
 			di += 2*ds;			\
 		}					\
+		di -= ds;			\
 	}						\
 } while (0)
 
@@ -528,11 +552,12 @@ scalers v:
 	} else  {					\
 		int j;					\
 		l = 0;					\
-		di -= 4*ds;				\
+		di -= 3*ds;				\
 		for (j = 0; j < 2; j++) {		\
 			v_mix(&di[0], &di[-ds], &di[ds], w, p_05, f_nop); \
 			di += 2*ds;			\
 		}					\
+		di -= ds;			\
 	}						\
 } while (0)
 
diff --git a/platform/gp2x/emu.c b/platform/gp2x/emu.c
index 0622691b..deb86581 100644
--- a/platform/gp2x/emu.c
+++ b/platform/gp2x/emu.c
@@ -105,7 +105,7 @@ static void change_renderer(int diff)
 }
 
 #define is_16bit_mode() \
-	(currentConfig.renderer == RT_16BIT || (PicoIn.AHW & PAHW_32X))
+	(currentConfig.renderer == RT_16BIT || (PicoIn.AHW & PAHW_32X) || doing_bg_frame)
 
 static void (*osd_text)(int x, int y, const char *text);
 
@@ -248,7 +248,7 @@ static int EmuScanEnd8_rot(unsigned int num)
 
 /* line doublers */
 static unsigned int ld_counter;
-static int ld_left, ld_lines;
+static int ld_left, ld_lines; // numbers in Q1 format
 
 static int EmuScanBegin16_ld(unsigned int num)
 {
@@ -271,9 +271,9 @@ static int EmuScanEnd16_ld(unsigned int num)
 		emu_scan_end(ld_counter);
 
 	ld_counter++;
-	ld_left--;
+	ld_left -= 2;
 	if (ld_left <= 0) {
-		ld_left = ld_lines;
+		ld_left += ld_lines;
 
 		EmuScanBegin16_ld(num);
 		memcpy(Pico.est.DrawLineDest, oldline, 320 * gp2x_current_bpp / 8);
@@ -313,6 +313,7 @@ static int make_local_pal_md(int fast_mode)
 	else if (Pico.video.reg[0xC] & 8) { // shadow/hilight mode
 		bgr444_to_rgb32(localPal, Pico.est.SonicPal);
 		bgr444_to_rgb32_sh(localPal, Pico.est.SonicPal);
+		memcpy(localPal+0xc0, localPal, 0x40*4); // for spr prio mess
 	}
 	else {
 		bgr444_to_rgb32(localPal, Pico.est.SonicPal);
@@ -331,18 +332,7 @@ static int make_local_pal_md(int fast_mode)
 
 static int make_local_pal_sms(int fast_mode)
 {
-	unsigned short *spal = PicoMem.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;
-	}
-
+	bgr444_to_rgb32(localPal, PicoMem.cram);
 	Pico.m.dirtyPal = 0;
 	return 0x40;
 }
@@ -486,6 +476,9 @@ static void vid_reset_mode(void)
 	int gp2x_mode = 16;
 	int renderer = get_renderer();
 
+	if (doing_bg_frame)
+		renderer = RT_16BIT;
+
 	PicoIn.opt &= ~POPT_ALT_RENDERER;
 	emu_scan_begin = NULL;
 	emu_scan_end = NULL;
@@ -557,11 +550,13 @@ static void vid_reset_mode(void)
 
 	Pico.m.dirtyPal = 1;
 
-	PicoIn.opt &= ~POPT_EN_SOFTSCALE;
+	PicoIn.opt &= ~(POPT_DIS_32C_BORDER|POPT_EN_SOFTSCALE);
 	if (currentConfig.scaling == EOPT_SCALE_SW) {
 		PicoIn.opt |= POPT_EN_SOFTSCALE;
 		PicoIn.filter = EOPT_FILTER_BILINEAR2;
-	}
+	} else if (currentConfig.scaling == EOPT_SCALE_HW && is_16bit_mode())
+		// hw scaling, render without any padding
+		PicoIn.opt |= POPT_DIS_32C_BORDER;
 
 	// palette converters for 8bit modes
 	make_local_pal = (PicoIn.AHW & PAHW_SMS) ? make_local_pal_sms : make_local_pal_md;
@@ -572,6 +567,12 @@ void emu_video_mode_change(int start_line, int line_count, int start_col, int co
 	int scalex = 320, scaley = 240;
 	int ln_offs = 0;
 
+	/* line doubling for swscaling, also needed for bg frames */
+	if (currentConfig.vscaling == EOPT_SCALE_SW && line_count < 240) {
+		ld_lines = ld_left = 2*line_count / (240 - line_count);
+		PicoDrawSetCallbacks(EmuScanBegin16_ld, EmuScanEnd16_ld);
+	}
+
 	if (doing_bg_frame)
 		return;
 
@@ -579,10 +580,8 @@ void emu_video_mode_change(int start_line, int line_count, int start_col, int co
 	osd_y = 232;
 
 	/* set up hwscaling here */
-	PicoIn.opt &= ~POPT_DIS_32C_BORDER;
 	if (col_count < 320 && currentConfig.scaling == EOPT_SCALE_HW) {
 		scalex = col_count;
-		PicoIn.opt |= POPT_DIS_32C_BORDER;
 		osd_fps_x = col_count - (320-OSD_FPS_X);
 	}
 
@@ -594,11 +593,6 @@ void emu_video_mode_change(int start_line, int line_count, int start_col, int co
 
 	gp2x_video_RGB_setscaling(ln_offs, scalex, scaley);
 
-	/* line doubling */
-	if (currentConfig.vscaling == EOPT_SCALE_SW && line_count < 240) {
-		ld_lines = ld_left = line_count / (240 - line_count);
-		PicoDrawSetCallbacks(EmuScanBegin16_ld, EmuScanEnd16_ld);
-	}
 
 	// clear whole screen in all buffers
 	if (!is_16bit_mode())
@@ -731,6 +725,8 @@ void pemu_forced_frame(int no_scale, int do_emu)
 	doing_bg_frame = 1;
 	PicoDrawSetCallbacks(NULL, NULL);
 	Pico.m.dirtyPal = 1;
+	PicoIn.opt &= ~POPT_DIS_32C_BORDER;
+	gp2x_current_bpp = 16;
 
 	if (!no_scale)
 		no_scale = currentConfig.scaling == EOPT_SCALE_NONE;
@@ -749,6 +745,7 @@ void plat_video_loop_prepare(void)
 	// make sure we are in correct mode
 	change_renderer(0);
 	vid_reset_mode();
+	rendstatus_old = -1;
 }
 
 void pemu_loop_prep(void)
diff --git a/platform/libretro/libretro.c b/platform/libretro/libretro.c
index 7c63c97b..4a9abea8 100644
--- a/platform/libretro/libretro.c
+++ b/platform/libretro/libretro.c
@@ -89,8 +89,8 @@ static retro_audio_sample_batch_t audio_batch_cb;
 #define INITIAL_SND_RATE 44100
 
 static const float VOUT_PAR = 0.0;
-static const float VOUT_4_3 = (224.0f * (4.0f / 3.0f));
-static const float VOUT_CRT = (224.0f * 1.29911f);
+static const float VOUT_4_3 = (4.0f / 3.0f);
+static const float VOUT_CRT = (1.29911f);
 
 static bool show_overscan = false;
 static bool old_show_overscan = false;
@@ -746,7 +746,7 @@ void retro_get_system_info(struct retro_system_info *info)
 #define _GIT_VERSION "-" GIT_VERSION
 #endif
    info->library_version = VERSION _GIT_VERSION;
-   info->valid_extensions = "bin|gen|smd|md|32x|cue|iso|chd|sms";
+   info->valid_extensions = "bin|gen|smd|md|32x|cue|iso|chd|sms|gg";
    info->need_fullpath = true;
 }
 
@@ -899,7 +899,7 @@ typedef struct patch
 } patch;
 
 extern void decode(char *buff, patch *dest);
-extern uint16_t m68k_read16(uint32_t a);
+extern uint32_t m68k_read16(uint32_t a);
 extern void m68k_write16(uint32_t a, uint16_t d);
 
 void retro_cheat_reset(void)
@@ -1494,12 +1494,13 @@ static void update_variables(bool first_run)
    var.value = NULL;
    var.key = "picodrive_aspect";
    if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value) {
+      int height = vout_height >= 192 && vout_height <= 224 ? 224 : vout_height;
       if (strcmp(var.value, "4/3") == 0)
-         user_vout_width = VOUT_4_3;
+         user_vout_width = VOUT_4_3 * height;
       else if (strcmp(var.value, "CRT") == 0)
-         user_vout_width = VOUT_CRT;
+         user_vout_width = VOUT_CRT * height;
       else
-         user_vout_width = VOUT_PAR;
+         user_vout_width = VOUT_PAR * height;
    }
 
    if (user_vout_width != old_user_vout_width)
@@ -1801,7 +1802,8 @@ void retro_init(void)
    sceBlock = getVMBlock();
 #endif
 
-   PicoIn.opt = POPT_EN_STEREO|POPT_EN_FM|POPT_EN_PSG|POPT_EN_Z80|POPT_EN_YM2413
+   PicoIn.opt = POPT_EN_STEREO|POPT_EN_FM
+      | POPT_EN_PSG|POPT_EN_Z80|POPT_EN_YM2413|POPT_EN_GG_LCD
       | POPT_EN_MCD_PCM|POPT_EN_MCD_CDDA|POPT_EN_MCD_GFX
       | POPT_EN_32X|POPT_EN_PWM
       | POPT_ACC_SPRITES|POPT_DIS_32C_BORDER;
diff --git a/platform/linux/emu.c b/platform/linux/emu.c
index 36898e61..1b228e91 100644
--- a/platform/linux/emu.c
+++ b/platform/linux/emu.c
@@ -95,7 +95,7 @@ static void draw_cd_leds(void)
  * for display isn't always possible.
  */
 
-static u16 *screen_buffer(u16 *buf)
+static inline u16 *screen_buffer(u16 *buf)
 {
 	return buf + screen_y * g_screen_ppitch + screen_x -
 			(out_y * g_screen_ppitch + out_x);
@@ -105,34 +105,46 @@ void screen_blit(u16 *pd, int pp, u8* ps, int ss, u16 *pal)
 {
 	typedef void (*upscale_t)
 			(u16 *di,int ds, u8 *si,int ss, int w,int h, u16 *pal);
-	upscale_t upscale_hv[] = {
+	static const upscale_t upscale_256_224_hv[] = {
 		upscale_rgb_nn_x_4_5_y_16_17,	upscale_rgb_snn_x_4_5_y_16_17,
 		upscale_rgb_bl2_x_4_5_y_16_17,	upscale_rgb_bl4_x_4_5_y_16_17,
 	};
-	upscale_t upscale_h[] = {
+	static const upscale_t upscale_256_224_h[] = {
 		upscale_rgb_nn_x_4_5,		upscale_rgb_snn_x_4_5,
 		upscale_rgb_bl2_x_4_5,		upscale_rgb_bl4_x_4_5,
 	};
-	upscale_t upscale_v[] = {
+	static const upscale_t upscale_256_224_v[] = {
 		upscale_rgb_nn_y_16_17,		upscale_rgb_snn_y_16_17,
 		upscale_rgb_bl2_y_16_17,	upscale_rgb_bl4_y_16_17,
 	};
-	upscale_t *upscale;
+	static const upscale_t upscale_160_144_hv[] = {
+		upscale_rgb_nn_x_1_2_y_3_5,	upscale_rgb_nn_x_1_2_y_3_5,
+		upscale_rgb_bl2_x_1_2_y_3_5,	upscale_rgb_bl4_x_1_2_y_3_5,
+	};
+	static const upscale_t upscale_160_144_h[] = {
+		upscale_rgb_nn_x_1_2,		upscale_rgb_nn_x_1_2,
+		upscale_rgb_bl2_x_1_2,		upscale_rgb_bl2_x_1_2,
+	};
+	static const upscale_t upscale_160_144_v[] = {
+		upscale_rgb_nn_y_3_5,		upscale_rgb_nn_y_3_5,
+		upscale_rgb_bl2_y_3_5,		upscale_rgb_bl4_y_3_5,
+	};
+	const upscale_t *upscale;
 	int y;
 
 	// handle software upscaling
 	upscale = NULL;
-	if (currentConfig.scaling == EOPT_SCALE_SW && out_w == 256) {
-		if (currentConfig.vscaling == EOPT_SCALE_SW && out_h <= 224)
-			// h+v scaling
-			upscale = upscale_hv;
-		else
-			// h scaling
-			upscale = upscale_h;
-	} else if (currentConfig.vscaling == EOPT_SCALE_SW && out_h <= 224) {
-			// v scaling
-			upscale = upscale_v;
-	} else {
+	if (currentConfig.scaling == EOPT_SCALE_SW) {
+	    if (currentConfig.vscaling == EOPT_SCALE_SW && out_h <= 224)
+		// h+v scaling
+		upscale = out_w == 256 ? upscale_256_224_hv: upscale_160_144_hv;
+	    else
+		// h scaling
+		upscale = out_w == 256 ? upscale_256_224_h : upscale_160_144_h;
+	} else if (currentConfig.vscaling == EOPT_SCALE_SW && out_h <= 224)
+		// v scaling
+		upscale = out_w == 256 ? upscale_256_224_v : upscale_160_144_v;
+	if (!upscale) {
 		// no scaling
 		for (y = 0; y < out_h; y++)
 			h_copy(pd, pp, ps, 328, out_w, f_pal);
@@ -280,16 +292,16 @@ static int vscale_state;
 
 static int cb_vscaling_begin(unsigned int line)
 {
-	static int prevline = 999;
-
 	// at start of new frame?
-	if (line < prevline) {
-		// set y frame offset (see emu_change_video_mode)
+	if (line <= out_y) {
+		// set y frame offset (see emu_video_mode_change)
 		Pico.est.DrawLineDest = screen_buffer(g_screen_ptr) +
-				(out_y * g_screen_ppitch + out_x);
+				(out_y * g_screen_ppitch /*+ out_x*/);
 		vscale_state = 0;
-	}
-	prevline = line;
+		return out_y - line;
+	} else if (line > out_y + out_h)
+		return 1;
+
 	return 0;
 }
 
@@ -301,16 +313,25 @@ static int cb_vscaling_nop(unsigned int line)
 static int cb_vscaling_end(unsigned int line)
 {
 	u16 *dest = Pico.est.DrawLineDest;
-	switch (currentConfig.filter) {
-	case 3: v_upscale_bl4_16_17(dest, g_screen_ppitch, 320, vscale_state);
-		break;
-	case 2: v_upscale_bl2_16_17(dest, g_screen_ppitch, 320, vscale_state);
-		break;
-	case 1: v_upscale_snn_16_17(dest, g_screen_ppitch, 320, vscale_state);
-		break;
-	default: v_upscale_nn_16_17(dest, g_screen_ppitch, 320, vscale_state);
-		break;
-	}
+
+	if (out_h == 144)
+	  switch (currentConfig.filter) {
+	  case 0: v_upscale_nn_3_5(dest, g_screen_ppitch, 320, vscale_state);
+		  break;
+	  default: v_upscale_snn_3_5(dest, g_screen_ppitch, 320, vscale_state);
+		  break;
+	  }
+	else
+	  switch (currentConfig.filter) {
+	  case 3: v_upscale_bl4_16_17(dest, g_screen_ppitch, 320, vscale_state);
+		  break;
+	  case 2: v_upscale_bl2_16_17(dest, g_screen_ppitch, 320, vscale_state);
+		  break;
+	  case 1: v_upscale_snn_16_17(dest, g_screen_ppitch, 320, vscale_state);
+		  break;
+	  default: v_upscale_nn_16_17(dest, g_screen_ppitch, 320, vscale_state);
+		  break;
+	  }
 	Pico.est.DrawLineDest = dest;
 	return 0;
 }
@@ -337,10 +358,10 @@ void emu_video_mode_change(int start_line, int line_count, int start_col, int co
 	}
 	switch (currentConfig.vscaling) {
 	case EOPT_SCALE_HW:
-		screen_h = (out_h < 224 ? 224 : out_h);
+		screen_h = (out_h < 224 && out_h > 144 ? 224 : out_h);
 		screen_y = 0;
 		// NTSC always has 224 visible lines, anything smaller has bars
-		if (out_h < 224)
+		if (out_h < 224 && out_h > 144)
 			screen_y += (224 - out_h)/2;
 		// handle vertical centering for 16 bit mode
 		if (is_16bit_mode())
@@ -349,7 +370,7 @@ void emu_video_mode_change(int start_line, int line_count, int start_col, int co
 	case EOPT_SCALE_SW:
 		screen_y = (screen_h - 240)/2;
 		// NTSC always has 224 visible lines, anything smaller has bars
-		if (out_h < 224)
+		if (out_h < 224 && out_h > 144)
 			screen_y += (224 - out_h)/2;
 		// in 16 bit mode sw scaling is divided between core and platform
 		if (is_16bit_mode() && out_h < 240)
diff --git a/tools/mkoffsets.sh b/tools/mkoffsets.sh
index 615e347f..d9a8db5c 100755
--- a/tools/mkoffsets.sh
+++ b/tools/mkoffsets.sh
@@ -136,6 +136,7 @@ get_define OFS_Pico_ Pico est			; echo "$line" >>$fn
 
 get_define OFS_PicoIn_ PicoInterface opt	; echo "$line" >>$fn
 get_define OFS_PicoIn_ PicoInterface filter	; echo "$line" >>$fn
+get_define OFS_PicoIn_ PicoInterface AHW	; echo "$line" >>$fn
 
 get_define OFS_EST_ PicoEState DrawScanline	; echo "$line" >>$fn
 get_define OFS_EST_ PicoEState rendstatus	; echo "$line" >>$fn