get rid of custom memcpy funcs
[picodrive.git] / platform / psp / emu.c
index edaf26a..8373a21 100644 (file)
@@ -1,3 +1,11 @@
+/*
+ * PicoDrive
+ * (C) notaz, 2007,2008
+ *
+ * This work is licensed under the terms of MAME license.
+ * See COPYING file in the top-level directory.
+ */
+
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <sys/syslimits.h> // PATH_MAX
 #include <pspdisplay.h>
 #include <psputils.h>
 #include <pspgu.h>
+#include <pspaudio.h>
 
 #include "psp.h"
 #include "menu.h"
 #include "emu.h"
+#include "mp3.h"
+#include "asm_utils.h"
 #include "../common/emu.h"
+#include "../common/config.h"
 #include "../common/lprintf.h"
-#include "../../Pico/PicoInt.h"
+#include <pico/pico_int.h>
+#include <pico/cd/cue.h>
 
-#ifdef BENCHMARK
-#define OSD_FPS_X 220
-#else
-#define OSD_FPS_X 260
-#endif
+#define OSD_FPS_X 432
 
-char romFileName[PATH_MAX];
-unsigned char *PicoDraw2FB = (unsigned char *)VRAM_CACHED_STUFF + 8; // +8 to be able to skip border with 1 quadword..
-int engineState;
+// additional pspaudio imports, credits to crazyc
+int sceAudio_38553111(unsigned short samples, unsigned short freq, char unknown);  // play with conversion?
+int sceAudio_5C37C0AE(void);                           // end play?
+int sceAudio_E0727056(int volume, void *buffer);       // blocking output
+int sceAudioOutput2GetRestSample();
 
-static int combo_keys = 0, combo_acts = 0; // keys and actions which need button combos
-static unsigned int noticeMsgTime = 0;
-int reset_timing = 0; // do we need this?
 
+//unsigned char *Draw2FB = (unsigned char *)VRAM_CACHED_STUFF + 8; // +8 to be able to skip border with 1 quadword..
+int engineStateSuspend;
 
-static void blit(const char *fps, const char *notice);
+#define PICO_PEN_ADJUST_X 4
+#define PICO_PEN_ADJUST_Y 2
+static int pico_pen_x = 320/2, pico_pen_y = 240/2;
+
+static void sound_init(void);
+static void sound_deinit(void);
+static void blit2(const char *fps, const char *notice, int lagging_behind);
 static void clearArea(int full);
 
-void emu_noticeMsgUpdated(void)
+int plat_get_root_dir(char *dst, int len)
 {
-       noticeMsgTime = sceKernelGetSystemTimeLow();
+       if (len > 0) *dst = 0;
+       return 0;
 }
 
-void emu_getMainDir(char *dst, int len)
+static void osd_text(int x, const char *text, int is_active, int clear_all)
 {
-       if (len > 0) *dst = 0;
+       unsigned short *screen = is_active ? psp_video_get_active_fb() : psp_screen;
+       int len = clear_all ? (480 / 2) : (strlen(text) * 8 / 2);
+       int *p, h;
+       void *tmp;
+       for (h = 0; h < 8; h++) {
+               p = (int *) (screen+x+512*(264+h));
+               p = (int *) ((int)p & ~3); // align
+               memset32_uncached(p, 0, len);
+       }
+       if (is_active) { tmp = psp_screen; psp_screen = screen; } // nasty pointer tricks
+       emu_text_out16(x, 264, text);
+       if (is_active) psp_screen = tmp;
 }
 
-static void emu_msg_cb(const char *msg)
+void emu_msg_cb(const char *msg)
 {
-       void *fb = psp_video_get_active_fb();
-
-       memset32((int *)((char *)fb + 512*264*2), 0, 512*8*2/4);
-       emu_textOut16(4, 264, msg);
+       osd_text(4, msg, 1, 1);
        noticeMsgTime = sceKernelGetSystemTimeLow() - 2000000;
 
        /* assumption: emu_msg_cb gets called only when something slow is about to happen */
        reset_timing = 1;
 }
 
-void emu_stateCb(const char *str)
-{
-       clearArea(0);
-       blit("", str);
-}
-
-static void emu_msg_tray_open(void)
-{
-       strcpy(noticeMsg, "CD tray opened");
-       noticeMsgTime = sceKernelGetSystemTimeLow();
-}
-
-
+/* FIXME: move to plat */
 void emu_Init(void)
 {
-       // make dirs for saves, cfgs, etc.
-       mkdir("mds", 0777);
-       mkdir("srm", 0777);
-       mkdir("brm", 0777);
-       mkdir("cfg", 0777);
-
-       PicoInit();
-       PicoMessage = emu_msg_cb;
-       PicoMCDopenTray = emu_msg_tray_open;
-       PicoMCDcloseTray = menu_loop_tray;
+       sound_init();
 }
 
 void emu_Deinit(void)
 {
-       // save SRAM
-       if ((currentConfig.EmuOpt & 1) && SRam.changed) {
-               emu_SaveLoadGame(0, 1);
-               SRam.changed = 0;
-       }
-
-       if (!(currentConfig.EmuOpt & 0x20)) {
-               FILE *f = fopen(PicoConfigFile, "r+b");
-               if (!f) emu_WriteConfig(0);
-               else {
-                       // if we already have config, reload it, except last ROM
-                       fseek(f, sizeof(currentConfig.lastRomFile), SEEK_SET);
-                       fread(&currentConfig.EmuOpt, 1, sizeof(currentConfig) - sizeof(currentConfig.lastRomFile), f);
-                       fseek(f, 0, SEEK_SET);
-                       fwrite(&currentConfig, 1, sizeof(currentConfig), f);
-                       fflush(f);
-                       fclose(f);
-               }
-       }
-
-       PicoExit();
+       sound_deinit();
 }
 
-void emu_setDefaultConfig(void)
+void pemu_prep_defconfig(void)
 {
-       memset(&currentConfig, 0, sizeof(currentConfig));
-       currentConfig.lastRomFile[0] = 0;
-       currentConfig.EmuOpt  = 0x1f | 0x680; // | confirm_save, cd_leds, 16bit rend
-       currentConfig.PicoOpt = 0x07 | 0xc00; // | cd_pcm, cd_cdda
-       currentConfig.PsndRate = 22050;
-       currentConfig.PicoRegion = 0; // auto
-       currentConfig.PicoAutoRgnOrder = 0x184; // US, EU, JP
-       currentConfig.Frameskip = -1; // auto
-       currentConfig.volume = 50;
-       currentConfig.CPUclock = 222;
-       currentConfig.KeyBinds[ 4] = 1<<0; // SACB RLDU
-       currentConfig.KeyBinds[ 6] = 1<<1;
-       currentConfig.KeyBinds[ 7] = 1<<2;
-       currentConfig.KeyBinds[ 5] = 1<<3;
-       currentConfig.KeyBinds[14] = 1<<4;
-       currentConfig.KeyBinds[13] = 1<<5;
-       currentConfig.KeyBinds[15] = 1<<6;
-       currentConfig.KeyBinds[ 3] = 1<<7;
-       currentConfig.KeyBinds[23] = 1<<26; // switch rend
-       currentConfig.KeyBinds[ 8] = 1<<27; // save state
-       currentConfig.KeyBinds[ 9] = 1<<28; // load state
-       currentConfig.PicoCDBuffers = 0;
-       currentConfig.scaling = 0;
+       defaultConfig.s_PsndRate = 22050;
+       defaultConfig.s_PicoCDBuffers = 64;
+       defaultConfig.CPUclock = 333;
+       defaultConfig.KeyBinds[ 4] = 1<<0; // SACB RLDU
+       defaultConfig.KeyBinds[ 6] = 1<<1;
+       defaultConfig.KeyBinds[ 7] = 1<<2;
+       defaultConfig.KeyBinds[ 5] = 1<<3;
+       defaultConfig.KeyBinds[14] = 1<<4;
+       defaultConfig.KeyBinds[13] = 1<<5;
+       defaultConfig.KeyBinds[15] = 1<<6;
+       defaultConfig.KeyBinds[ 3] = 1<<7;
+       defaultConfig.KeyBinds[12] = 1<<26; // switch rnd
+       defaultConfig.KeyBinds[ 8] = 1<<27; // save state
+       defaultConfig.KeyBinds[ 9] = 1<<28; // load state
+       defaultConfig.KeyBinds[28] = 1<<0; // num "buttons"
+       defaultConfig.KeyBinds[30] = 1<<1;
+       defaultConfig.KeyBinds[31] = 1<<2;
+       defaultConfig.KeyBinds[29] = 1<<3;
+       defaultConfig.scaling = 1;     // bilinear filtering for psp
+       defaultConfig.scale = 1.20;    // fullscreen
+       defaultConfig.hscale40 = 1.25;
+       defaultConfig.hscale32 = 1.56;
 }
 
 
-static unsigned short __attribute__((aligned(16))) localPal[0x100];
+extern void amips_clut(unsigned short *dst, unsigned char *src, unsigned short *pal, int count);
+extern void amips_clut_6bit(unsigned short *dst, unsigned char *src, unsigned short *pal, int count);
+
+static void (*amips_clut_f)(unsigned short *dst, unsigned char *src, unsigned short *pal, int count) = NULL;
 
 struct Vertex
 {
@@ -142,232 +128,325 @@ struct Vertex
        short x,y,z;
 };
 
-static void EmuScanPrepare(void)
+static struct Vertex __attribute__((aligned(4))) g_vertices[2];
+static unsigned short __attribute__((aligned(16))) localPal[0x100];
+static int dynamic_palette = 0, need_pal_upload = 0, blit_16bit_mode = 0;
+static int fbimg_offs = 0;
+
+static void set_scaling_params(void)
 {
-       HighCol = VRAM_STUFF;
+       int src_width, fbimg_width, fbimg_height, fbimg_xoffs, fbimg_yoffs, border_hack = 0;
+       g_vertices[0].x = g_vertices[0].y =
+       g_vertices[0].z = g_vertices[1].z = 0;
+
+       fbimg_height = (int)(240.0 * currentConfig.scale + 0.5);
+       if (Pico.video.reg[12] & 1) {
+               fbimg_width = (int)(320.0 * currentConfig.scale * currentConfig.hscale40 + 0.5);
+               src_width = 320;
+       } else {
+               fbimg_width = (int)(256.0 * currentConfig.scale * currentConfig.hscale32 + 0.5);
+               src_width = 256;
+       }
 
-#if 0
-       sceGuSync(0,0); // sync with prev
-       sceGuStart(GU_DIRECT, guCmdList);
-//     sceGuDispBuffer(480, 272, psp_screen == VRAM_FB0 ? VRAMOFFS_FB1 : VRAMOFFS_FB0, 512);
-       sceGuDrawBuffer(GU_PSM_5650, psp_screen == VRAM_FB0 ? VRAMOFFS_FB0 : VRAMOFFS_FB1, 512); // point to back fb?
-       sceGuFinish();
-#endif
+       if (fbimg_width  & 1) fbimg_width++;  // make even
+       if (fbimg_height & 1) fbimg_height++;
+
+       if (fbimg_width >= 480) {
+               g_vertices[0].u = (fbimg_width-480)/2;
+               g_vertices[1].u = src_width - (fbimg_width-480)/2 - 1;
+               fbimg_width = 480;
+               fbimg_xoffs = 0;
+       } else {
+               g_vertices[0].u = 0;
+               g_vertices[1].u = src_width;
+               fbimg_xoffs = 240 - fbimg_width/2;
+       }
+       if (fbimg_width > 320 && fbimg_width <= 480) border_hack = 1;
+
+       if (fbimg_height >= 272) {
+               g_vertices[0].v = (fbimg_height-272)/2;
+               g_vertices[1].v = 240 - (fbimg_height-272)/2;
+               fbimg_height = 272;
+               fbimg_yoffs = 0;
+       } else {
+               g_vertices[0].v = 0;
+               g_vertices[1].v = 240;
+               fbimg_yoffs = 136 - fbimg_height/2;
+       }
+
+       g_vertices[1].x = fbimg_width;
+       g_vertices[1].y = fbimg_height;
+       if (fbimg_xoffs < 0) fbimg_xoffs = 0;
+       if (fbimg_yoffs < 0) fbimg_yoffs = 0;
+       if (border_hack) {
+               g_vertices[0].u++;
+               g_vertices[0].x++;
+               g_vertices[1].u--;
+               g_vertices[1].x--;
+       }
+       fbimg_offs = (fbimg_yoffs*512 + fbimg_xoffs) * 2; // dst is always 16bit
+
+       /*
+       lprintf("set_scaling_params:\n");
+       lprintf("offs: %i, %i\n", fbimg_xoffs, fbimg_yoffs);
+       lprintf("xy0, xy1: %i, %i; %i, %i\n", g_vertices[0].x, g_vertices[0].y, g_vertices[1].x, g_vertices[1].y);
+       lprintf("uv0, uv1: %i, %i; %i, %i\n", g_vertices[0].u, g_vertices[0].v, g_vertices[1].u, g_vertices[1].v);
+       */
 }
 
-static int EmuScan16(unsigned int num, void *sdata)
+static void do_pal_update(int allow_sh, int allow_as)
 {
-//     struct Vertex* vertices;
+       unsigned int *dpal=(void *)localPal;
+       int i;
 
-       if (!(Pico.video.reg[1]&8)) num += 8;
-       //DrawLineDest = (unsigned short *) psp_screen + 512*(num+1);
-       HighCol = (unsigned char *)psp_screen + num*512;
+       //for (i = 0x3f/2; i >= 0; i--)
+       //      dpal[i] = ((spal[i]&0x000f000f)<< 1)|((spal[i]&0x00f000f0)<<3)|((spal[i]&0x0f000f00)<<4);
+       do_pal_convert(localPal, Pico.cram, currentConfig.gamma, currentConfig.gamma2);
 
-#if 0
-       sceGuSync(0,0); // sync with prev
-       sceGuStart(GU_DIRECT, guCmdList);
+       Pico.m.dirtyPal = 0;
+       need_pal_upload = 1;
 
-       if (Pico.m.dirtyPal) {
-               int i, *dpal = (void *)localPal, *spal = (int *)Pico.cram;
-               Pico.m.dirtyPal = 0;
+       if (allow_sh && (Pico.video.reg[0xC]&8)) // shadow/hilight?
+       {
+               // shadowed pixels
                for (i = 0x3f/2; i >= 0; i--)
-                       dpal[i] = ((spal[i]&0x000f000f)<< 1)|((spal[i]&0x00f000f0)<<3)|((spal[i]&0x0f000f00)<<4);
-
-               sceGuClutLoad((256/8), localPal); // upload 32*8 entries (256)
+                       dpal[0x20|i] = dpal[0x60|i] = (dpal[i]>>1)&0x7bcf7bcf;
+               // hilighted pixels
+               for (i = 0x3f; i >= 0; i--) {
+                       int t=localPal[i]&0xf79e;t+=0x4208;
+                       if (t&0x20) t|=0x1e;
+                       if (t&0x800) t|=0x780;
+                       if (t&0x10000) t|=0xf000;
+                       t&=0xf79e;
+                       localPal[0x80|i]=(unsigned short)t;
+               }
+               localPal[0xe0] = 0;
+               localPal[0xf0] = 0x001f;
+       }
+       else if (allow_as && (Pico.est.rendstatus & PDRAW_SPR_LO_ON_HI))
+       {
+               memcpy(dpal + 0x80/2, localPal, 0x40*2);
        }
+}
 
-       // setup CLUT texture
+static void do_slowmode_lines(int line_to)
+{
+       int line = 0, line_len = (Pico.video.reg[12]&1) ? 320 : 256;
+       unsigned short *dst = (unsigned short *)VRAM_STUFF + 512*240/2;
+       unsigned char  *src = (unsigned char  *)VRAM_CACHED_STUFF + 16;
+       if (!(Pico.video.reg[1]&8)) { line = 8; dst += 512*8; src += 512*8; }
 
-//     sceGuClutMode(GU_PSM_5650,0,0xff,0);
-//     sceGuClutLoad((256/8), localPal); // upload 32*8 entries (256)
-//     sceGuTexMode(GU_PSM_T8,0,0,0); // 8-bit image
-//     sceGuTexImage(0,512,1/*512*/,512,VRAM_STUFF);
-//     sceGuTexFunc(GU_TFX_REPLACE,GU_TCC_RGB);
-//     sceGuTexFilter(GU_LINEAR,GU_LINEAR);
-//     sceGuTexScale(1.0f,1.0f);
-//     sceGuTexOffset(0.0f,0.0f);
-//     sceGuAmbientColor(0xffffffff);
+       for (; line < line_to; line++, dst+=512, src+=512)
+               amips_clut_f(dst, src, localPal, line_len);
+}
 
-       // render sprite
+static void EmuScanPrepare(void)
+{
+       Pico.est.HighCol = (unsigned char *)VRAM_CACHED_STUFF + 8;
+       if (!(Pico.video.reg[1]&8)) Pico.est.HighCol += 8*512;
 
-//     sceGuColor(0xffffffff);
-       vertices = (struct Vertex*)sceGuGetMemory(2 * sizeof(struct Vertex));
-       vertices[0].u = 0; vertices[0].v = 0;
-       vertices[0].x = 0; vertices[0].y = num; vertices[0].z = 0;
-       vertices[1].u = 320; vertices[1].v = 512;
-       vertices[1].x = 320; vertices[1].y = num+1; vertices[1].z = 0;
-       //sceGuDrawArray(GU_SPRITES,GU_TEXTURE_16BIT|GU_VERTEX_16BIT|GU_TRANSFORM_2D,2,0,vertices);
+       if (dynamic_palette > 0)
+               dynamic_palette--;
 
-       sceGuFinish();
-#endif
-       return 0;
+       if (Pico.m.dirtyPal)
+               do_pal_update(1, 1);
+       if ((Pico.est.rendstatus & PDRAW_SPR_LO_ON_HI) && !(Pico.video.reg[0xC]&8))
+            amips_clut_f = amips_clut_6bit;
+       else amips_clut_f = amips_clut;
 }
 
+static int EmuScanSlowBegin(unsigned int num)
+{
+       if (!dynamic_palette)
+               Pico.est.HighCol = (unsigned char *)VRAM_CACHED_STUFF + num * 512 + 8;
 
-static void draw2_clut(void)
+       return 0;
+}
+
+static int EmuScanSlowEnd(unsigned int num)
 {
-       struct Vertex* vertices;
-       int x;
+       if (Pico.m.dirtyPal) {
+               if (!dynamic_palette) {
+                       do_slowmode_lines(num);
+                       dynamic_palette = 3; // last for 2 more frames
+               }
+               do_pal_update(1, 1);
+       }
 
-       sceKernelDcacheWritebackAll(); // for PicoDraw2FB
+       if (dynamic_palette) {
+               int line_len = (Pico.video.reg[12]&1) ? 320 : 256;
+               void *dst = (char *)VRAM_STUFF + 512*240 + 512*2*num;
+               amips_clut_f(dst, Pico.est.HighCol + 8, localPal, line_len);
+       }
+
+       return 0;
+}
+
+static void blitscreen_clut(void)
+{
+       int offs = fbimg_offs;
+       offs += (psp_screen == VRAM_FB0) ? VRAMOFFS_FB0 : VRAMOFFS_FB1;
 
        sceGuSync(0,0); // sync with prev
        sceGuStart(GU_DIRECT, guCmdList);
-//     sceGuDispBuffer(480, 272, psp_screen == VRAM_FB0 ? VRAMOFFS_FB1 : VRAMOFFS_FB0, 512);
-       sceGuDrawBuffer(GU_PSM_5650, psp_screen == VRAM_FB0 ? VRAMOFFS_FB0 : VRAMOFFS_FB1, 512); // point to back fb?
+       sceGuDrawBuffer(GU_PSM_5650, (void *)offs, 512); // point to back buffer
 
-       if (Pico.m.dirtyPal) {
-               int i, *dpal = (void *)localPal, *spal = (int *)Pico.cram;
-               Pico.m.dirtyPal = 0;
-               for (i = 0x3f/2; i >= 0; i--)
-                       dpal[i] = ((spal[i]&0x000f000f)<< 1)|((spal[i]&0x00f000f0)<<3)|((spal[i]&0x0f000f00)<<4);
+       if (dynamic_palette)
+       {
+               if (!blit_16bit_mode) { // the current mode is not 16bit
+                       sceGuTexMode(GU_PSM_5650, 0, 0, 0);
+                       sceGuTexImage(0,512,512,512,(char *)VRAM_STUFF + 512*240);
 
-               sceGuClutLoad((256/8), localPal); // upload 32*8 entries (256)
+                       blit_16bit_mode = 1;
+               }
        }
+       else
+       {
+               if (blit_16bit_mode) {
+                       sceGuClutMode(GU_PSM_5650,0,0xff,0);
+                       sceGuTexMode(GU_PSM_T8,0,0,0); // 8-bit image
+                       sceGuTexImage(0,512,512,512,(char *)VRAM_STUFF + 16);
+                       blit_16bit_mode = 0;
+               }
+
+               if ((PicoOpt&0x10) && Pico.m.dirtyPal)
+                       do_pal_update(0, 0);
+
+               sceKernelDcacheWritebackAll();
 
-       #define SLICE_WIDTH 32
+               if (need_pal_upload) {
+                       need_pal_upload = 0;
+                       sceGuClutLoad((256/8), localPal); // upload 32*8 entries (256)
+               }
+       }
 
-       for (x = 0; x < 320; x += SLICE_WIDTH)
+#if 1
+       if (g_vertices[0].u == 0 && g_vertices[1].u == g_vertices[1].x)
        {
-               // render sprite
-               vertices = (struct Vertex*)sceGuGetMemory(2 * sizeof(struct Vertex));
-               vertices[0].u = vertices[0].x = x;
-               vertices[0].v = vertices[0].y = 0;
-               vertices[0].z = 0;
-               vertices[1].u = vertices[1].x = x + SLICE_WIDTH;
-               vertices[1].v = vertices[1].y = 224;
-               vertices[1].z = 0;
-               sceGuDrawArray(GU_SPRITES,GU_TEXTURE_16BIT|GU_VERTEX_16BIT|GU_TRANSFORM_2D,2,0,vertices);
+               struct Vertex* vertices;
+               int x;
+
+               #define SLICE_WIDTH 32
+               for (x = 0; x < g_vertices[1].x; x += SLICE_WIDTH)
+               {
+                       // render sprite
+                       vertices = (struct Vertex*)sceGuGetMemory(2 * sizeof(struct Vertex));
+                       memcpy(vertices, g_vertices, 2 * sizeof(struct Vertex));
+                       vertices[0].u = vertices[0].x = x;
+                       vertices[1].u = vertices[1].x = x + SLICE_WIDTH;
+                       sceGuDrawArray(GU_SPRITES,GU_TEXTURE_16BIT|GU_VERTEX_16BIT|GU_TRANSFORM_2D,2,0,vertices);
+               }
+               // lprintf("listlen: %iB\n", sceGuCheckList()); // ~480 only
        }
+       else
+#endif
+               sceGuDrawArray(GU_SPRITES,GU_TEXTURE_16BIT|GU_VERTEX_16BIT|GU_TRANSFORM_2D,2,0,g_vertices);
 
        sceGuFinish();
 }
 
 
+static void cd_leds(void)
+{
+       unsigned int reg, col_g, col_r, *p;
+
+       reg = Pico_mcd->s68k_regs[0];
+
+       p = (unsigned int *)((short *)psp_screen + 512*2+4+2);
+       col_g = (reg & 2) ? 0x06000600 : 0;
+       col_r = (reg & 1) ? 0x00180018 : 0;
+       *p++ = col_g; *p++ = col_g; p+=2; *p++ = col_r; *p++ = col_r; p += 512/2 - 12/2;
+       *p++ = col_g; *p++ = col_g; p+=2; *p++ = col_r; *p++ = col_r; p += 512/2 - 12/2;
+       *p++ = col_g; *p++ = col_g; p+=2; *p++ = col_r; *p++ = col_r;
+}
 
-static int EmuScan8(unsigned int num, void *sdata)
+static void draw_pico_ptr(void)
 {
-       // draw like the fast renderer
-       // TODO?
-       //if (!(Pico.video.reg[1]&8)) num += 8;
-       //HighCol = gfx_buffer + 328*(num+1);
+       unsigned char *p = (unsigned char *)VRAM_STUFF + 16;
 
-       return 0;
+       // only if pen enabled and for 8bit mode
+       if (pico_inp_mode == 0 || blit_16bit_mode) return;
+
+       p += 512 * (pico_pen_y + PICO_PEN_ADJUST_Y);
+       p += pico_pen_x + PICO_PEN_ADJUST_X;
+       p[  -1] = 0xe0; p[   0] = 0xf0; p[   1] = 0xe0;
+       p[ 511] = 0xf0; p[ 512] = 0xf0; p[ 513] = 0xf0;
+       p[1023] = 0xe0; p[1024] = 0xf0; p[1025] = 0xe0;
 }
 
-static void osd_text(int x, const char *text)
+
+#if 0
+static void dbg_text(void)
 {
-       int len = strlen(text) * 8 / 2;
-       int *p, h;
+       int *p, h, len;
+       char text[128];
+
+       sprintf(text, "sl: %i, 16b: %i", g_vertices[0].u == 0 && g_vertices[1].u == g_vertices[1].x, blit_16bit_mode);
+       len = strlen(text) * 8 / 2;
        for (h = 0; h < 8; h++) {
-               p = (int *) ((unsigned short *) psp_screen+x+512*(264+h));
+               p = (int *) ((unsigned short *) psp_screen+2+512*(256+h));
                p = (int *) ((int)p & ~3); // align
-               memset32(p, 0, len);
+               memset32_uncached(p, 0, len);
        }
-       emu_textOut16(x, 264, text);
+       emu_text_out16(2, 256, text);
 }
+#endif
 
-
-static void cd_leds(void)
+/* called after rendering is done, but frame emulation is not finished */
+void blit1(void)
 {
-       static int old_reg = 0;
-       unsigned int col_g, col_r, *p;
+       if (PicoOpt&0x10)
+       {
+               int i;
+               unsigned char *pd;
+               // clear top and bottom trash
+               for (pd = Pico.est.Draw2FB+8, i = 8; i > 0; i--, pd += 512)
+                       memset32((int *)pd, 0xe0e0e0e0, 320/4);
+               for (pd = Pico.est.Draw2FB+512*232+8, i = 8; i > 0; i--, pd += 512)
+                       memset32((int *)pd, 0xe0e0e0e0, 320/4);
+       }
 
-       if (!((Pico_mcd->s68k_regs[0] ^ old_reg) & 3)) return; // no change
-       old_reg = Pico_mcd->s68k_regs[0];
+       if (PicoAHW & PAHW_PICO)
+               draw_pico_ptr();
 
-       p = (unsigned int *)((short *)psp_screen + 512*2+4+2);
-       col_g = (old_reg & 2) ? 0x06000600 : 0;
-       col_r = (old_reg & 1) ? 0xc000c000 : 0;
-       *p++ = col_g; *p++ = col_g; p+=2; *p++ = col_r; *p++ = col_r; p += 512/2 - 12/2;
-       *p++ = col_g; *p++ = col_g; p+=2; *p++ = col_r; *p++ = col_r; p += 512/2 - 12/2;
-       *p++ = col_g; *p++ = col_g; p+=2; *p++ = col_r; *p++ = col_r;
+       blitscreen_clut();
 }
 
 
-static void blit(const char *fps, const char *notice)
+static void blit2(const char *fps, const char *notice, int lagging_behind)
 {
-       int emu_opt = currentConfig.EmuOpt;
-
-       if (PicoOpt&0x10)
-       {
-#if 1
-               draw2_clut();
-#else
-               extern void amips_clut(unsigned short *dst, unsigned char *src, unsigned short *pal, int count);
-               int i; // , lines_flags = 224;
-               unsigned short *pd = psp_screen;
-               unsigned char  *ps = PicoDraw2FB+328*8+8;
-               // 8bit fast renderer
-               if (Pico.m.dirtyPal) {
-                       int *dpal = (void *)localPal;
-                       int *spal = (int *)Pico.cram;
-                       Pico.m.dirtyPal = 0;
-                       for (i = 0x3f/2; i >= 0; i--)
-                               dpal[i] = ((spal[i]&0x000f000f)<< 1)|((spal[i]&0x00f000f0)<<3)|((spal[i]&0x0f000f00)<<4);
-               }
-               // if (!(Pico.video.reg[12]&1)) lines_flags|=0x10000;
-               // if (currentConfig.EmuOpt&0x4000)
-               //      lines_flags|=0x40000; // (Pico.m.frame_count&1)?0x20000:0x40000;
-               //vidCpy8to16((unsigned short *)giz_screen+321*8, PicoDraw2FB+328*8, localPal, lines_flags);
-               for (i = 224; i > 0; i--, pd+=512, ps+=328)
-                       amips_clut(pd, ps, localPal, 320);
-#endif
-       }
-#if 0
-       else if (!(emu_opt&0x80))
-       {
-               int lines_flags;
-               // 8bit accurate renderer
-               if (Pico.m.dirtyPal) {
-                       Pico.m.dirtyPal = 0;
-                       vidConvCpyRGB565(localPal, Pico.cram, 0x40);
-                       if (Pico.video.reg[0xC]&8) { // shadow/hilight mode
-                               //vidConvCpyRGB32sh(localPal+0x40, Pico.cram, 0x40);
-                               //vidConvCpyRGB32hi(localPal+0x80, Pico.cram, 0x40); // TODO?
-                               blockcpy(localPal+0xc0, localPal+0x40, 0x40*2);
-                               localPal[0xc0] = 0x0600;
-                               localPal[0xd0] = 0xc000;
-                               localPal[0xe0] = 0x0000; // reserved pixels for OSD
-                               localPal[0xf0] = 0xffff;
-                       }
-                       /* no support
-                       else if (rendstatus & 0x20) { // mid-frame palette changes
-                               vidConvCpyRGB565(localPal+0x40, HighPal, 0x40);
-                               vidConvCpyRGB565(localPal+0x80, HighPal+0x40, 0x40);
-                       } */
-               }
-               lines_flags = (Pico.video.reg[1]&8) ? 240 : 224;
-               if (!(Pico.video.reg[12]&1)) lines_flags|=0x10000;
-               if (currentConfig.EmuOpt&0x4000)
-                       lines_flags|=0x40000; // (Pico.m.frame_count&1)?0x20000:0x40000;
-               vidCpy8to16((unsigned short *)giz_screen+321*8, PicoDraw2FB+328*8, localPal, lines_flags);
-       }
-#endif
+       int vsync = 0, emu_opt = currentConfig.EmuOpt;
 
        if (notice || (emu_opt & 2)) {
-               if (notice)      osd_text(4, notice);
-               if (emu_opt & 2) osd_text(OSD_FPS_X, fps);
+               if (notice)      osd_text(4, notice, 0, 0);
+               if (emu_opt & 2) osd_text(OSD_FPS_X, fps, 0, 0);
        }
 
-       if ((emu_opt & 0x400) && (PicoMCD & 1))
+       //dbg_text();
+
+       if ((emu_opt & 0x400) && (PicoAHW & PAHW_MCD))
                cd_leds();
 
-       sceGuSync(0,0);
-       psp_video_flip(0);
+       if (currentConfig.EmuOpt & 0x2000) { // want vsync
+               if (!(currentConfig.EmuOpt & 0x10000) || !lagging_behind) vsync = 1;
+       }
+
+       psp_video_flip(vsync);
 }
 
 // clears whole screen or just the notice area (in all buffers)
 static void clearArea(int full)
 {
        if (full) {
-               memset32(psp_screen, 0, 512*272*2/4);
+               memset32_uncached(psp_screen, 0, 512*272*2/4);
                psp_video_flip(0);
-               memset32(psp_screen, 0, 512*272*2/4);
+               memset32_uncached(psp_screen, 0, 512*272*2/4);
+               memset32(VRAM_CACHED_STUFF, 0xe0e0e0e0, 512*240/4);
+               memset32((int *)VRAM_CACHED_STUFF+512*240/4, 0, 512*240*2/4);
        } else {
                void *fb = psp_video_get_active_fb();
-               memset32((int *)((char *)psp_screen + 512*264*2), 0, 512*8*2/4);
-               memset32((int *)((char *)fb         + 512*264*2), 0, 512*8*2/4);
+               memset32_uncached((int *)((char *)psp_screen + 512*264*2), 0, 512*8*2/4);
+               memset32_uncached((int *)((char *)fb         + 512*264*2), 0, 512*8*2/4);
        }
 }
 
@@ -378,51 +457,210 @@ static void vidResetMode(void)
        sceGuStart(GU_DIRECT, guCmdList);
 
        sceGuClutMode(GU_PSM_5650,0,0xff,0);
-       //sceGuClutLoad((256/8), localPal); // upload 32*8 entries (256)
        sceGuTexMode(GU_PSM_T8,0,0,0); // 8-bit image
        sceGuTexFunc(GU_TFX_REPLACE,GU_TCC_RGB);
-       sceGuTexFilter(GU_LINEAR,GU_LINEAR);
+       if (currentConfig.scaling)
+            sceGuTexFilter(GU_LINEAR, GU_LINEAR);
+       else sceGuTexFilter(GU_NEAREST, GU_NEAREST);
        sceGuTexScale(1.0f,1.0f);
        sceGuTexOffset(0.0f,0.0f);
-       sceGuAmbientColor(0xffffffff);
-       sceGuColor(0xffffffff);
 
+       sceGuTexImage(0,512,512,512,(char *)VRAM_STUFF + 16);
+
+       // slow rend.
+       PicoDrawSetOutFormat(PDF_NONE, 0);
+       PicoDrawSetCallbacks(EmuScanSlowBegin, EmuScanSlowEnd);
+
+       localPal[0xe0] = 0;
+       localPal[0xf0] = 0x001f;
+       Pico.m.dirtyPal = 1;
+       blit_16bit_mode = dynamic_palette = 0;
+
+       sceGuFinish();
+       set_scaling_params();
+       sceGuSync(0,0);
+}
+
+void plat_debug_cat(char *str)
+{
+       strcat(str, blit_16bit_mode ? "soft clut\n" : "hard clut\n");
+}
+
+
+/* sound stuff */
+#define SOUND_BLOCK_SIZE_NTSC (1470*2) // 1024 // 1152
+#define SOUND_BLOCK_SIZE_PAL  (1764*2)
+#define SOUND_BLOCK_COUNT    8
+
+static short __attribute__((aligned(4))) sndBuffer[SOUND_BLOCK_SIZE_PAL*SOUND_BLOCK_COUNT + 44100/50*2];
+static short *snd_playptr = NULL, *sndBuffer_endptr = NULL;
+static int samples_made = 0, samples_done = 0, samples_block = 0;
+static int sound_thread_exit = 0;
+static SceUID sound_sem = -1;
+
+static void writeSound(int len);
+
+static int sound_thread(SceSize args, void *argp)
+{
+       int ret = 0;
+
+       lprintf("sthr: started, priority %i\n", sceKernelGetThreadCurrentPriority());
+
+       while (!sound_thread_exit)
+       {
+               if (samples_made - samples_done < samples_block) {
+                       // wait for data (use at least 2 blocks)
+                       //lprintf("sthr: wait... (%i)\n", samples_made - samples_done);
+                       while (samples_made - samples_done <= samples_block*2 && !sound_thread_exit)
+                               ret = sceKernelWaitSema(sound_sem, 1, 0);
+                       if (ret < 0) lprintf("sthr: sceKernelWaitSema: %i\n", ret);
+                       continue;
+               }
+
+               // lprintf("sthr: got data: %i\n", samples_made - samples_done);
+
+               ret = sceAudio_E0727056(PSP_AUDIO_VOLUME_MAX, snd_playptr);
+
+               samples_done += samples_block;
+               snd_playptr  += samples_block;
+               if (snd_playptr >= sndBuffer_endptr)
+                       snd_playptr = sndBuffer;
+               // 1.5 kernel returns 0, newer ones return # of samples queued
+               if (ret < 0)
+                       lprintf("sthr: sceAudio_E0727056: %08x; pos %i/%i\n", ret, samples_done, samples_made);
+
+               // shouln't happen, but just in case
+               if (samples_made - samples_done >= samples_block*3) {
+                       //lprintf("sthr: block skip (%i)\n", samples_made - samples_done);
+                       samples_done += samples_block; // skip
+                       snd_playptr  += samples_block;
+               }
+
+       }
+
+       lprintf("sthr: exit\n");
+       sceKernelExitDeleteThread(0);
+       return 0;
+}
+
+static void sound_init(void)
+{
+       SceUID thid;
+       int ret;
 
-       if (PicoOpt&0x10) {
-               sceGuTexImage(0,512,512,512,(char *)VRAM_STUFF + 8*512+16);
+       sound_sem = sceKernelCreateSema("sndsem", 0, 0, 1, NULL);
+       if (sound_sem < 0) lprintf("sceKernelCreateSema() failed: %i\n", sound_sem);
 
-       } else if (currentConfig.EmuOpt&0x80) {
-               PicoDrawSetColorFormat(/*1*/-1);
-               PicoScan = EmuScan16;
+       samples_made = samples_done = 0;
+       samples_block = SOUND_BLOCK_SIZE_NTSC; // make sure it goes to sema
+       sound_thread_exit = 0;
+       thid = sceKernelCreateThread("sndthread", sound_thread, 0x12, 0x10000, 0, NULL);
+       if (thid >= 0)
+       {
+               ret = sceKernelStartThread(thid, 0, 0);
+               if (ret < 0) lprintf("sound_init: sceKernelStartThread returned %08x\n", ret);
+       }
+       else
+               lprintf("sceKernelCreateThread failed: %i\n", thid);
+}
+
+void pemu_sound_start(void)
+{
+       static int PsndRate_old = 0, PicoOpt_old = 0, pal_old = 0;
+       int ret, stereo;
 
-               sceGuTexImage(0,512,1/*512*/,512,VRAM_STUFF);
+       samples_made = samples_done = 0;
 
+       if (PsndRate != PsndRate_old || (PicoOpt&0x0b) != (PicoOpt_old&0x0b) || Pico.m.pal != pal_old) {
+               PsndRerate(Pico.m.frame_count ? 1 : 0);
+       }
+       stereo=(PicoOpt&8)>>3;
+
+       samples_block = Pico.m.pal ? SOUND_BLOCK_SIZE_PAL : SOUND_BLOCK_SIZE_NTSC;
+       if (PsndRate <= 22050) samples_block /= 2;
+       sndBuffer_endptr = &sndBuffer[samples_block*SOUND_BLOCK_COUNT];
+
+       lprintf("starting audio: %i, len: %i, stereo: %i, pal: %i, block samples: %i\n",
+                       PsndRate, PsndLen, stereo, Pico.m.pal, samples_block);
+
+       // while (sceAudioOutput2GetRestSample() > 0) psp_msleep(100);
+       // sceAudio_5C37C0AE();
+       ret = sceAudio_38553111(samples_block/2, PsndRate, 2); // seems to not need that stupid 64byte alignment
+       if (ret < 0) {
+               lprintf("sceAudio_38553111() failed: %i\n", ret);
+               emu_status_msg("sound init failed (%i), snd disabled", ret);
+               currentConfig.EmuOpt &= ~EOPT_EN_SOUND;
        } else {
-               PicoDrawSetColorFormat(-1);
-               PicoScan = EmuScan8;
+               PicoWriteSound = writeSound;
+               memset32((int *)(void *)sndBuffer, 0, sizeof(sndBuffer)/4);
+               snd_playptr = sndBuffer_endptr - samples_block;
+               samples_made = samples_block; // send 1 empty block first..
+               PsndOut = sndBuffer;
+               PsndRate_old = PsndRate;
+               PicoOpt_old  = PicoOpt;
+               pal_old = Pico.m.pal;
        }
-       if ((PicoOpt&0x10) || !(currentConfig.EmuOpt&0x80)) {
-               // setup pal for 8-bit modes
-               localPal[0xc0] = 0x0600;
-               localPal[0xd0] = 0xc000;
-               localPal[0xe0] = 0x0000; // reserved pixels for OSD
-               localPal[0xf0] = 0xffff;
+}
+
+void pemu_sound_stop(void)
+{
+       int i;
+       if (samples_done == 0)
+       {
+               // if no data is written between sceAudio_38553111 and sceAudio_5C37C0AE calls,
+               // we get a deadlock on next sceAudio_38553111 call
+               // so this is yet another workaround:
+               memset32((int *)(void *)sndBuffer, 0, samples_block*4/4);
+               samples_made = samples_block * 3;
+               sceKernelSignalSema(sound_sem, 1);
        }
-       Pico.m.dirtyPal = 1;
+       sceKernelDelayThread(100*1000);
+       samples_made = samples_done = 0;
+       for (i = 0; sceAudioOutput2GetRestSample() > 0 && i < 16; i++)
+               psp_msleep(100);
+       sceAudio_5C37C0AE();
+}
 
-       sceGuFinish();
-       sceGuSync(0,0);
+/* wait until we can write more sound */
+void pemu_sound_wait(void)
+{
+       // TODO: test this
+       while (!sound_thread_exit && samples_made - samples_done > samples_block * 4)
+               psp_msleep(10);
+}
 
-       clearArea(1);
+static void sound_deinit(void)
+{
+       sound_thread_exit = 1;
+       sceKernelSignalSema(sound_sem, 1);
+       sceKernelDeleteSema(sound_sem);
+       sound_sem = -1;
 }
-/*
-static void updateSound(int len)
+
+static void writeSound(int len)
 {
-       if (PicoOpt&8) len<<=1;
+       int ret;
 
-       // TODO..
+       PsndOut += len / 2;
+       /*if (PsndOut > sndBuffer_endptr) {
+               memcpy32((int *)(void *)sndBuffer, (int *)endptr, (PsndOut - endptr + 1) / 2);
+               PsndOut = &sndBuffer[PsndOut - endptr];
+               lprintf("mov\n");
+       }
+       else*/
+       if (PsndOut > sndBuffer_endptr) lprintf("snd oflow %i!\n", PsndOut - sndBuffer_endptr);
+       if (PsndOut >= sndBuffer_endptr)
+               PsndOut = sndBuffer;
+
+       // signal the snd thread
+       samples_made += len / 2;
+       if (samples_made - samples_done > samples_block*2) {
+               // lprintf("signal, %i/%i\n", samples_done, samples_made);
+               ret = sceKernelSignalSema(sound_sem, 1);
+               //if (ret < 0) lprintf("snd signal ret %08x\n", ret);
+       }
 }
-*/
+
 
 static void SkipFrame(void)
 {
@@ -431,51 +669,83 @@ static void SkipFrame(void)
        PicoSkipFrame=0;
 }
 
-void emu_forcedFrame(void)
+void pemu_forced_frame(int no_scale, int do_emu)
 {
        int po_old = PicoOpt;
        int eo_old = currentConfig.EmuOpt;
 
-       PicoOpt &= ~0x0010;
-       PicoOpt |=  0x4080; // soft_scale | acc_sprites
+       PicoOpt &= ~POPT_ALT_RENDERER;
+       PicoOpt |= POPT_ACC_SPRITES;
+       if (!no_scale)
+               PicoOpt |= POPT_EN_SOFTSCALE;
        currentConfig.EmuOpt |= 0x80;
 
-       PicoDrawSetColorFormat(1);
-       PicoScan = EmuScan16;
+       vidResetMode();
+       memset32(VRAM_CACHED_STUFF, 0xe0e0e0e0, 512*8/4); // borders
+       memset32((int *)VRAM_CACHED_STUFF + 512*232/4, 0xe0e0e0e0, 512*8/4);
+       memset32_uncached((int *)psp_screen + 512*264*2/4, 0, 512*8*2/4);
+
+       PicoDrawSetOutFormat(PDF_NONE, 0);
+       PicoDrawSetCallbacks(EmuScanSlowBegin, EmuScanSlowEnd);
        EmuScanPrepare();
-       Pico.m.dirtyPal = 1;
        PicoFrameDrawOnly();
+       blit1();
+       sceGuSync(0,0);
 
        PicoOpt = po_old;
        currentConfig.EmuOpt = eo_old;
 }
 
 
+static void RunEventsPico(unsigned int events, unsigned int keys)
+{
+       emu_RunEventsPico(events);
+
+       if (pico_inp_mode != 0)
+       {
+               PicoPad[0] &= ~0x0f; // release UDLR
+               if (keys & PBTN_UP)   { pico_pen_y--; if (pico_pen_y < 8) pico_pen_y = 8; }
+               if (keys & PBTN_DOWN) { pico_pen_y++; if (pico_pen_y > 224-PICO_PEN_ADJUST_Y) pico_pen_y = 224-PICO_PEN_ADJUST_Y; }
+               if (keys & PBTN_LEFT) { pico_pen_x--; if (pico_pen_x < 0) pico_pen_x = 0; }
+               if (keys & PBTN_RIGHT) {
+                       int lim = (Pico.video.reg[12]&1) ? 319 : 255;
+                       pico_pen_x++;
+                       if (pico_pen_x > lim-PICO_PEN_ADJUST_X)
+                               pico_pen_x = lim-PICO_PEN_ADJUST_X;
+               }
+               PicoPicohw.pen_pos[0] = pico_pen_x;
+               if (!(Pico.video.reg[12]&1)) PicoPicohw.pen_pos[0] += pico_pen_x/4;
+               PicoPicohw.pen_pos[0] += 0x3c;
+               PicoPicohw.pen_pos[1] = pico_inp_mode == 1 ? (0x2f8 + pico_pen_y) : (0x1fc + pico_pen_y);
+       }
+}
+
 static void RunEvents(unsigned int which)
 {
        if (which & 0x1800) // save or load (but not both)
        {
                int do_it = 1;
 
-               if ( emu_checkSaveFile(state_slot) &&
+               if ( emu_check_save_file(state_slot) &&
                                (( (which & 0x1000) && (currentConfig.EmuOpt & 0x800)) || // load
                                 (!(which & 0x1000) && (currentConfig.EmuOpt & 0x200))) ) // save
                {
                        int keys;
-                       blit("", (which & 0x1000) ? "LOAD STATE? (X=yes, O=no)" : "OVERWRITE SAVE? (X=yes, O=no)");
-                       while( !((keys = psp_pad_read(1)) & (BTN_X|BTN_CIRCLE)) )
+                       sceGuSync(0,0);
+                       blit2("", (which & 0x1000) ? "LOAD STATE? (X=yes, O=no)" : "OVERWRITE SAVE? (X=yes, O=no)", 0);
+                       while( !((keys = psp_pad_read(1)) & (PBTN_X|PBTN_CIRCLE)) )
                                psp_msleep(50);
-                       if (keys & BTN_CIRCLE) do_it = 0;
-                       while(  ((keys = psp_pad_read(1)) & (BTN_X|BTN_CIRCLE)) ) // wait for release
+                       if (keys & PBTN_CIRCLE) do_it = 0;
+                       while(  ((keys = psp_pad_read(1)) & (PBTN_X|PBTN_CIRCLE)) ) // wait for release
                                psp_msleep(50);
                        clearArea(0);
                }
 
                if (do_it)
                {
-                       osd_text(4, (which & 0x1000) ? "LOADING GAME" : "SAVING GAME");
-                       PicoStateProgressCB = emu_stateCb;
-                       emu_SaveLoadGame((which & 0x1000) >> 12, 0);
+                       osd_text(4, (which & 0x1000) ? "LOADING GAME" : "SAVING GAME", 1, 0);
+                       PicoStateProgressCB = emu_msg_cb;
+                       emu_save_load_game((which & 0x1000) >> 12, 0);
                        PicoStateProgressCB = NULL;
                        psp_msleep(0);
                }
@@ -489,15 +759,10 @@ static void RunEvents(unsigned int which)
 
                vidResetMode();
 
-               if (PicoOpt&0x10) {
-                       strcpy(noticeMsg, " 8bit fast renderer");
-               } else if (currentConfig.EmuOpt&0x80) {
-                       strcpy(noticeMsg, "16bit accurate renderer");
-               } else {
-                       strcpy(noticeMsg, " 8bit accurate renderer");
-               }
-
-               noticeMsgTime = sceKernelGetSystemTimeLow();
+               if (PicoOpt & POPT_ALT_RENDERER)
+                       emu_status_msg("fast renderer");
+               else if (currentConfig.EmuOpt&0x80)
+                       emu_status_msg("accurate renderer");
        }
        if (which & 0x0300)
        {
@@ -508,8 +773,8 @@ static void RunEvents(unsigned int which)
                        state_slot += 1;
                        if(state_slot > 9) state_slot = 0;
                }
-               sprintf(noticeMsg, "SAVE SLOT %i [%s]", state_slot, emu_checkSaveFile(state_slot) ? "USED" : "FREE");
-               noticeMsgTime = sceKernelGetSystemTimeLow();
+               emu_status_msg("SAVE SLOT %i [%s]", state_slot,
+                       emu_check_save_file(state_slot) ? "USED" : "FREE");
        }
 }
 
@@ -519,97 +784,39 @@ static void updateKeys(void)
        static unsigned int prevEvents = 0;
        int i;
 
+       /* FIXME: port to input fw, merge with emu.c:emu_update_input() */
        keys = psp_pad_read(0);
        if (keys & PSP_CTRL_HOME)
                sceDisplayWaitVblankStart();
 
-       if (keys & BTN_SELECT)
+       if (keys & PBTN_SELECT)
                engineState = PGS_Menu;
 
        keys &= CONFIGURABLE_KEYS;
 
-       for (i = 0; i < 32; i++)
-       {
-               if (keys & (1 << i))
-               {
-                       int pl, acts = currentConfig.KeyBinds[i];
-                       if (!acts) continue;
-                       pl = (acts >> 16) & 1;
-                       if (combo_keys & (1 << i))
-                       {
-                               int u = i+1, acts_c = acts & combo_acts;
-                               // let's try to find the other one
-                               if (acts_c)
-                                       for (; u < 32; u++)
-                                               if ( (currentConfig.KeyBinds[u] & acts_c) && (keys & (1 << u)) ) {
-                                                       allActions[pl] |= acts_c;
-                                                       keys &= ~((1 << i) | (1 << u));
-                                                       break;
-                                               }
-                               // add non-combo actions if combo ones were not found
-                               if (!acts_c || u == 32)
-                                       allActions[pl] |= acts & ~combo_acts;
-                       } else {
-                               allActions[pl] |= acts;
-                       }
-               }
-       }
+       PicoPad[0] = allActions[0] & 0xfff;
+       PicoPad[1] = allActions[1] & 0xfff;
 
-       PicoPad[0] = (unsigned short) allActions[0];
-       PicoPad[1] = (unsigned short) allActions[1];
+       if (allActions[0] & 0x7000) emu_DoTurbo(&PicoPad[0], allActions[0]);
+       if (allActions[1] & 0x7000) emu_DoTurbo(&PicoPad[1], allActions[1]);
 
        events = (allActions[0] | allActions[1]) >> 16;
 
-       // volume is treated in special way and triggered every frame
-       if ((events & 0x6000) && PsndOut != NULL)
-       {
-               int vol = currentConfig.volume;
-               if (events & 0x2000) {
-                       if (vol < 100) vol++;
-               } else {
-                       if (vol >   0) vol--;
-               }
-               // FrameworkAudio_SetVolume(vol, vol); // TODO
-               sprintf(noticeMsg, "VOL: %02i ", vol);
-               noticeMsgTime = sceKernelGetSystemTimeLow();
-               currentConfig.volume = vol;
+       if ((events ^ prevEvents) & 0x40) {
+               emu_set_fastforward(events & 0x40);
+               reset_timing = 1;
        }
 
        events &= ~prevEvents;
+
+       if (PicoAHW == PAHW_PICO)
+               RunEventsPico(events, keys);
        if (events) RunEvents(events);
        if (movie_data) emu_updateMovie();
 
        prevEvents = (allActions[0] | allActions[1]) >> 16;
 }
 
-static void find_combos(void)
-{
-       int act, u;
-
-       // find out which keys and actions are combos
-       combo_keys = combo_acts = 0;
-       for (act = 0; act < 32; act++)
-       {
-               int keyc = 0;
-               if (act == 16 || act == 17) continue; // player2 flag
-               for (u = 0; u < 32; u++)
-               {
-                       if (currentConfig.KeyBinds[u] & (1 << act)) keyc++;
-               }
-               if (keyc > 1)
-               {
-                       // loop again and mark those keys and actions as combo
-                       for (u = 0; u < 32; u++)
-                       {
-                               if (currentConfig.KeyBinds[u] & (1 << act)) {
-                                       combo_keys |= 1 << u;
-                                       combo_acts |= 1 << act;
-                               }
-                       }
-               }
-       }
-}
-
 
 static void simpleWait(unsigned int until)
 {
@@ -622,13 +829,14 @@ static void simpleWait(unsigned int until)
                sceKernelDelayThread(diff);
 }
 
-void emu_Loop(void)
+void pemu_loop(void)
 {
-       //static int PsndRate_old = 0, PicoOpt_old = 0, pal_old = 0;
+       static int mp3_init_done = 0;
        char fpsbuff[24]; // fps count c string
-       unsigned int tval, tval_prev = 0, tval_thissec = 0; // timing
-       int frames_done = 0, frames_shown = 0, oldmodes = 0;
-       int target_fps, target_frametime, lim_time, tval_diff, i;
+       unsigned int tval, tval_thissec = 0; // timing
+       int target_fps, target_frametime, lim_time, tval_diff, i, oldmodes = 0;
+       int pframes_done, pframes_shown; // "period" frames, used for sync
+       int  frames_done,  frames_shown, tval_fpsc = 0; // actual frames
        char *notice = NULL;
 
        lprintf("entered emu_Loop()\n");
@@ -644,50 +852,38 @@ void emu_Loop(void)
 
        // make sure we are in correct mode
        vidResetMode();
+       clearArea(1);
        Pico.m.dirtyPal = 1;
        oldmodes = ((Pico.video.reg[12]&1)<<2) ^ 0xc;
-       find_combos();
 
        // pal/ntsc might have changed, reset related stuff
        target_fps = Pico.m.pal ? 50 : 60;
        target_frametime = Pico.m.pal ? (1000000<<8)/50 : (1000000<<8)/60+1;
        reset_timing = 1;
 
-       // prepare CD buffer
-       if (PicoMCD & 1) PicoCDBufferInit();
+       if (PicoAHW & PAHW_MCD) {
+               // prepare CD buffer
+               PicoCDBufferInit();
+               // mp3...
+               if (!mp3_init_done) {
+                       i = mp3_init();
+                       mp3_init_done = 1;
+                       if (i) { engineState = PGS_Menu; return; }
+               }
+       }
 
        // prepare sound stuff
        PsndOut = NULL;
-#if 0 // TODO
-       if (currentConfig.EmuOpt & 4)
+       if (currentConfig.EmuOpt & EOPT_EN_SOUND)
        {
-               int ret, snd_excess_add, stereo;
-               if (PsndRate != PsndRate_old || (PicoOpt&0x0b) != (PicoOpt_old&0x0b) || Pico.m.pal != pal_old) {
-                       sound_rerate(Pico.m.frame_count ? 1 : 0);
-               }
-               stereo=(PicoOpt&8)>>3;
-               snd_excess_add = ((PsndRate - PsndLen*target_fps)<<16) / target_fps;
-               snd_cbuf_samples = (PsndRate<<stereo) * 16 / target_fps;
-               lprintf("starting audio: %i len: %i (ex: %04x) stereo: %i, pal: %i\n",
-                       PsndRate, PsndLen, snd_excess_add, stereo, Pico.m.pal);
-               ret = FrameworkAudio_Init(PsndRate, snd_cbuf_samples, stereo);
-               if (ret != 0) {
-                       lprintf("FrameworkAudio_Init() failed: %i\n", ret);
-                       sprintf(noticeMsg, "sound init failed (%i), snd disabled", ret);
-                       noticeMsgTime = sceKernelGetSystemTimeLow();
-                       currentConfig.EmuOpt &= ~4;
-               } else {
-                       FrameworkAudio_SetVolume(currentConfig.volume, currentConfig.volume);
-                       PicoWriteSound = updateSound;
-                       snd_cbuff = FrameworkAudio_56448Buffer();
-                       PsndOut = snd_cbuff + snd_cbuf_samples / 2; // start writing at the middle
-                       snd_all_samples = 0;
-                       PsndRate_old = PsndRate;
-                       PicoOpt_old  = PicoOpt;
-                       pal_old = Pico.m.pal;
-               }
+               pemu_sound_start();
        }
-#endif
+
+       sceDisplayWaitVblankStart();
+       pframes_shown = pframes_done =
+        frames_shown =  frames_done = 0;
+
+       tval_fpsc = sceKernelGetSystemTimeLow();
 
        // loop?
        while (engineState == PGS_Running)
@@ -695,11 +891,11 @@ void emu_Loop(void)
                int modes;
 
                tval = sceKernelGetSystemTimeLow();
-               if (reset_timing || tval < tval_prev) {
+               if (reset_timing || tval < tval_fpsc) {
                        //stdbg("timing reset");
                        reset_timing = 0;
                        tval_thissec = tval;
-                       frames_shown = frames_done = 0;
+                       pframes_shown = pframes_done = 0;
                }
 
                // show notice message?
@@ -721,47 +917,46 @@ void emu_Loop(void)
                if (modes != oldmodes) {
                        oldmodes = modes;
                        clearArea(1);
+                       set_scaling_params();
                }
 
                // second passed?
+               if (tval - tval_fpsc >= 1000000)
+               {
+                       if (currentConfig.EmuOpt & 2)
+                               sprintf(fpsbuff, "%02i/%02i  ", frames_shown, frames_done);
+                       frames_done = frames_shown = 0;
+                       tval_fpsc += 1000000;
+               }
+
                if (tval - tval_thissec >= 1000000)
                {
-#ifdef BENCHMARK
-                       static int bench = 0, bench_fps = 0, bench_fps_s = 0, bfp = 0, bf[4];
-                       if(++bench == 10) {
-                               bench = 0;
-                               bench_fps_s = bench_fps;
-                               bf[bfp++ & 3] = bench_fps;
-                               bench_fps = 0;
+                       // missing 1 frame?
+                       if (currentConfig.Frameskip < 0 && pframes_done < target_fps) {
+                               SkipFrame(); pframes_done++; frames_done++;
                        }
-                       bench_fps += frames_shown;
-                       sprintf(fpsbuff, "%02i/%02i/%02i", frames_shown, bench_fps_s, (bf[0]+bf[1]+bf[2]+bf[3])>>2);
-#else
-                       if(currentConfig.EmuOpt & 2)
-                               sprintf(fpsbuff, "%02i/%02i", frames_shown, frames_done);
-#endif
+
                        tval_thissec += 1000000;
 
                        if (currentConfig.Frameskip < 0) {
-                               frames_done  -= target_fps; if (frames_done  < 0) frames_done  = 0;
-                               frames_shown -= target_fps; if (frames_shown < 0) frames_shown = 0;
-                               if (frames_shown > frames_done) frames_shown = frames_done;
+                               pframes_done  -= target_fps; if (pframes_done  < 0) pframes_done  = 0;
+                               pframes_shown -= target_fps; if (pframes_shown < 0) pframes_shown = 0;
+                               if (pframes_shown > pframes_done) pframes_shown = pframes_done;
                        } else {
-                               frames_done = frames_shown = 0;
+                               pframes_done = pframes_shown = 0;
                        }
                }
 #ifdef PFRAMES
                sprintf(fpsbuff, "%i", Pico.m.frame_count);
 #endif
 
-               tval_prev = tval;
-               lim_time = (frames_done+1) * target_frametime;
+               lim_time = (pframes_done+1) * target_frametime;
                if (currentConfig.Frameskip >= 0) // frameskip enabled
                {
                        for (i = 0; i < currentConfig.Frameskip; i++) {
                                updateKeys();
-                               SkipFrame(); frames_done++;
-                               if (PsndOut) { // do framelimitting if sound is enabled
+                               SkipFrame(); pframes_done++; frames_done++;
+                               if (!(currentConfig.EmuOpt&0x40000)) { // do framelimitting if needed
                                        int tval_diff;
                                        tval = sceKernelGetSystemTimeLow();
                                        tval_diff = (int)(tval - tval_thissec) << 8;
@@ -776,17 +971,15 @@ void emu_Loop(void)
                        int tval_diff;
                        tval = sceKernelGetSystemTimeLow();
                        tval_diff = (int)(tval - tval_thissec) << 8;
-                       if (tval_diff > lim_time)
+                       if (tval_diff > lim_time && (pframes_done/16 < pframes_shown))
                        {
                                // no time left for this frame - skip
                                if (tval_diff - lim_time >= (300000<<8)) {
-                                       /* something caused a slowdown for us (disk access? cache flush?)
-                                        * try to recover by resetting timing... */
                                        reset_timing = 1;
                                        continue;
                                }
                                updateKeys();
-                               SkipFrame(); frames_done++;
+                               SkipFrame(); pframes_done++; frames_done++;
                                continue;
                        }
                }
@@ -798,15 +991,18 @@ void emu_Loop(void)
 
                PicoFrame();
 
-               blit(fpsbuff, notice);
+               sceGuSync(0,0);
 
                // check time
                tval = sceKernelGetSystemTimeLow();
                tval_diff = (int)(tval - tval_thissec) << 8;
 
-               if (currentConfig.Frameskip < 0 && tval_diff - lim_time >= (300000<<8)) // slowdown detection
+               blit2(fpsbuff, notice, tval_diff > lim_time);
+
+               if (currentConfig.Frameskip < 0 && tval_diff - lim_time >= (300000<<8)) { // slowdown detection
                        reset_timing = 1;
-               else if (PsndOut != NULL || currentConfig.Frameskip < 0)
+               }
+               else if (!(currentConfig.EmuOpt&0x40000) || currentConfig.Frameskip < 0)
                {
                        // sleep if we are still too fast
                        if (tval_diff < lim_time)
@@ -816,29 +1012,60 @@ void emu_Loop(void)
                        }
                }
 
-               frames_done++; frames_shown++;
+               pframes_done++; pframes_shown++;
+                frames_done++;  frames_shown++;
        }
 
 
-       if (PicoMCD & 1) PicoCDBufferFree();
-/*
+       emu_set_fastforward(0);
+
+       if (PicoAHW & PAHW_MCD) PicoCDBufferFree();
+
        if (PsndOut != NULL) {
-               PsndOut = snd_cbuff = NULL;
-               FrameworkAudio_Close();
+               pemu_sound_stop();
+               PsndOut = NULL;
        }
-*/
+
        // save SRAM
        if ((currentConfig.EmuOpt & 1) && SRam.changed) {
-               emu_stateCb("Writing SRAM/BRAM..");
-               emu_SaveLoadGame(0, 1);
+               emu_msg_cb("Writing SRAM/BRAM..");
+               emu_save_load_game(0, 1);
                SRam.changed = 0;
        }
-}
 
+       // clear fps counters and stuff
+       memset32_uncached((int *)psp_video_get_active_fb() + 512*264*2/4, 0, 512*8*2/4);
+}
 
-void emu_ResetGame(void)
+void emu_HandleResume(void)
 {
-       PicoReset(0);
-       reset_timing = 1;
+       if (!(PicoAHW & PAHW_MCD)) return;
+
+       // reopen first CD track
+       if (Pico_mcd->TOC.Tracks[0].F != NULL)
+       {
+               char *fname = rom_fname_reload;
+               int len = strlen(rom_fname_reload);
+               cue_data_t *cue_data = NULL;
+
+               if (len > 4 && strcasecmp(fname + len - 4,  ".cue") == 0)
+               {
+                       cue_data = cue_parse(rom_fname_reload);
+                       if (cue_data != NULL)
+                               fname = cue_data->tracks[1].fname;
+               }
+
+               lprintf("emu_HandleResume: reopen %s\n", fname);
+               pm_close(Pico_mcd->TOC.Tracks[0].F);
+               Pico_mcd->TOC.Tracks[0].F = pm_open(fname);
+               lprintf("reopen %s\n", Pico_mcd->TOC.Tracks[0].F != NULL ? "ok" : "failed");
+
+               if (cue_data != NULL) cue_destroy(cue_data);
+       }
+
+       mp3_reopen_file();
+
+       if (!(Pico_mcd->s68k_regs[0x36] & 1) && (Pico_mcd->scd.Status_CDC & 1))
+               cdda_start_play();
 }