From 13c9a48c26ba05678f34ead0f93e2873e00f93e8 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Fri, 5 Jul 2024 21:44:34 +0200
Subject: [PATCH 01/16] ui, minor menu fix

---
 platform/common/menu_pico.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/platform/common/menu_pico.c b/platform/common/menu_pico.c
index a4abd7a6..53e52b87 100644
--- a/platform/common/menu_pico.c
+++ b/platform/common/menu_pico.c
@@ -577,7 +577,7 @@ static int menu_loop_32x_options(int id, int keys)
 static const char *sms_hardwares[] = { "auto", "Game Gear", "Master System", "SG-1000", "SC-3000", NULL };
 static const char *gg_ghosting_opts[] = { "OFF", "weak", "normal", NULL };
 static const char *sms_mappers[] = { "auto", "Sega", "Codemasters", "Korea", "Korea MSX", "Korea X-in-1", "Korea 4-Pak", "Korea Janggun", "Korea Nemesis", "Taiwan 8K RAM", "Korea XOR", "Sega 32K RAM", NULL };
-static const char *sms_tmspalette[] = { "SMS", "SG-1000" };
+static const char *sms_tmspalette[] = { "SMS", "SG-1000", NULL };
 
 static const char h_smsfm[] = "FM sound is only supported by few games,\n"
 				"some games may crash with FM enabled";
-- 
2.39.2


From 4af69a1df0d4e5e16bb4648934e30441835fab93 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Fri, 5 Jul 2024 21:47:33 +0200
Subject: [PATCH 02/16] sms, improve pad handling (TR as output)

---
 pico/sms.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/pico/sms.c b/pico/sms.c
index 919b6ea6..0c2ee634 100644
--- a/pico/sms.c
+++ b/pico/sms.c
@@ -189,9 +189,11 @@ static unsigned char z80_sms_in(unsigned short a)
         break;
 
       case 0xc0: /* I/O port A and B */
-        if (! (PicoIn.AHW & PAHW_SC) || (Pico.ms.io_sg & 7) == 7)
+        if (! (PicoIn.AHW & PAHW_SC) || (Pico.ms.io_sg & 7) == 7) {
           d = ~((PicoIn.pad[0] & 0x3f) | (PicoIn.pad[1] << 6));
-        else
+          if (!(Pico.ms.io_ctl & 0x01)) // TR as output
+            d = (d & ~0x20) | ((Pico.ms.io_ctl << 1) & 0x20);
+        } else
           ; // read kbd 8 bits
         break;
 
@@ -199,6 +201,8 @@ static unsigned char z80_sms_in(unsigned short a)
         if (! (PicoIn.AHW & PAHW_SC) || (Pico.ms.io_sg & 7) == 7) {
           d = (Pico.ms.io_ctl & 0x80) | ((Pico.ms.io_ctl << 1) & 0x40) | 0x30;
           d |= ~(PicoIn.pad[1] >> 2) & 0x0f;
+          if (!(Pico.ms.io_ctl & 0x04)) // TR as output
+            d = (d & ~0x08) | ((Pico.ms.io_ctl >> 3) & 0x08);
           if (Pico.ms.io_ctl & 0x08) d |= 0x80; // TH as input is unconnected
           if (Pico.ms.io_ctl & 0x02) d |= 0x40;
         } else
@@ -309,6 +313,7 @@ static void write_bank_sega(unsigned short a, unsigned char d)
 
   elprintf(EL_Z80BNK, "bank sega %04x %02x @ %04x", a, d, z80_pc());
   Pico.ms.mapper = PMS_MAP_SEGA;
+  if (d == Pico.ms.carthw[a & 0x0f]) return;
   Pico.ms.carthw[a & 0x0f] = d;
 
   switch (a & 0x0f)
@@ -348,6 +353,7 @@ static void write_bank_codem(unsigned short a, unsigned char d)
   if (Pico.ms.mapper != PMS_MAP_CODEM && (Pico.ms.mapper || (a>>14) == d)) return;
   elprintf(EL_Z80BNK, "bank codem %04x %02x @ %04x", a, d, z80_pc());
   Pico.ms.mapper = PMS_MAP_CODEM;
+  if (Pico.ms.carthw[a>>14] == d) return;
   Pico.ms.carthw[a>>14] = d;
 
   d &= bank_mask;
@@ -690,8 +696,9 @@ void PicoResetMS(void)
 
   z80_reset();
   PsndReset(); // pal must be known here
+
+  Pico.ms.io_ctl = (PicoIn.AHW & (PAHW_SG|PAHW_SC)) ? 0xf5 : 0xff;
   Pico.ms.fm_ctl = 0xff;
-  Pico.m.dirtyPal = 1;
 
   // reset memory mapping
   PicoMemSetupMS();
@@ -708,6 +715,7 @@ void PicoResetMS(void)
   Pico.video.reg[8] = 0x00;
   Pico.video.reg[9] = 0x00;
   Pico.video.reg[10] = 0xff;
+  Pico.m.dirtyPal = 1;
 
   // BIOS, clear zram (unitialized on Mark-III, cf src/mame/drivers/sms.cpp)
   i = !(PicoIn.AHW & PAHW_GG) && (Pico.m.hardware & PMS_HW_JAP) ? 0xf0 : 0x00;
-- 
2.39.2


From 73411ac44791153f79f1f03d82103954619ad140 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Mon, 8 Jul 2024 22:41:01 +0200
Subject: [PATCH 03/16] sound, add panning for ym2612 dac

---
 pico/sound/sound.c | 31 ++++++++++++++++++++++---------
 1 file changed, 22 insertions(+), 9 deletions(-)

diff --git a/pico/sound/sound.c b/pico/sound/sound.c
index 650f7aca..74f9967b 100644
--- a/pico/sound/sound.c
+++ b/pico/sound/sound.c
@@ -16,6 +16,8 @@
 #include "resampler.h"
 #include "mix.h"
 
+#define YM2612_CH6PAN   0x1b6   // panning register for channel 6 (used for DAC)
+
 void (*PsndMix_32_to_16)(s16 *dest, s32 *src, int count) = mix_32_to_16_stereo;
 
 // master int buffer to mix to
@@ -282,9 +284,12 @@ PICO_INTERNAL void PsndDoDAC(int cyc_to)
   // 1 sample delay for correct IIR filtering over audio frame boundaries
   if (PicoIn.opt & POPT_EN_STEREO) {
     s16 *d = PicoIn.sndOut + pos*2;
-    // left channel only, mixed ro right channel in mixing phase
-    *d++ += Pico.snd.dac_val2, *d++ += Pico.snd.dac_val2;
-    while (--len) *d++ += Pico.snd.dac_val, *d++ += Pico.snd.dac_val;
+    int pan = ym2612.REGS[YM2612_CH6PAN];
+    int l = pan & 0x80 ? Pico.snd.dac_val : 0;
+    int r = pan & 0x40 ? Pico.snd.dac_val : 0;
+    *d++ += pan & 0x80 ? Pico.snd.dac_val2 : 0;
+    *d++ += pan & 0x40 ? Pico.snd.dac_val2 : 0;
+    while (--len) *d++ += l, *d++ += r;
   } else {
     s16 *d = PicoIn.sndOut + pos;
     *d++ += Pico.snd.dac_val2;
@@ -516,13 +521,21 @@ static int PsndRender(int offset, int length)
 
   // Fill up DAC output in case of missing samples (Q rounding errors)
   if (length-daclen > 0 && PicoIn.sndOut) {
-    s16 *dacbuf = PicoIn.sndOut + (daclen << stereo);
     Pico.snd.dac_pos += (length-daclen) << 20;
-    *dacbuf++ += Pico.snd.dac_val2;
-    if (stereo) *dacbuf++ += Pico.snd.dac_val2;
-    for (daclen++; length-daclen > 0; daclen++) {
-      *dacbuf++ += Pico.snd.dac_val;
-      if (stereo) *dacbuf++ += Pico.snd.dac_val;
+    if (PicoIn.opt & POPT_EN_STEREO) {
+      s16 *d = PicoIn.sndOut + daclen*2;
+      int pan = ym2612.REGS[YM2612_CH6PAN];
+      int l = pan & 0x80 ? Pico.snd.dac_val : 0;
+      int r = pan & 0x40 ? Pico.snd.dac_val : 0;
+      *d++ += pan & 0x80 ? Pico.snd.dac_val2 : 0;
+      *d++ += pan & 0x40 ? Pico.snd.dac_val2 : 0;
+      if (l|r) for (daclen++; length-daclen > 0; daclen++)
+          *d++ += l, *d++ += r;
+    } else {
+      s16 *d = PicoIn.sndOut + daclen;
+      *d++ += Pico.snd.dac_val2;
+      if (Pico.snd.dac_val) for (daclen++; length-daclen > 0; daclen++)
+        *d++ += Pico.snd.dac_val;
     }
     Pico.snd.dac_val2 = Pico.snd.dac_val;
   }
-- 
2.39.2


From 977e5db898cd5c1980db66953dbcbce8a1e1c7d0 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Mon, 8 Jul 2024 22:46:06 +0200
Subject: [PATCH 04/16] core, fix typo

---
 pico/pico_int.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pico/pico_int.h b/pico/pico_int.h
index 3d495739..4caa800d 100644
--- a/pico/pico_int.h
+++ b/pico/pico_int.h
@@ -203,7 +203,7 @@ extern struct DrZ80 drZ80;
 
 #define z80_resetCycles() { \
   Pico.t.z80c_cnt -= Pico.t.z80c_aim, Pico.t.z80c_aim = Pico.t.z80_scanline = 0; \
-  if (!Pico.m.z80Run | Pico.m.z80_reset) Pico.t.z80c_cnt = 0; \
+  if (!Pico.m.z80Run || Pico.m.z80_reset) Pico.t.z80c_cnt = 0; \
 }
 
 #define z80_cyclesDone() \
-- 
2.39.2


From 538982b062eaad7c1ec4eca046e5e7749dbe321a Mon Sep 17 00:00:00 2001
From: notaz <notasas@gmail.com>
Date: Tue, 9 Jul 2024 22:30:17 +0300
Subject: [PATCH 05/16] mcd: unbreak cue search

---
 pico/cd/cd_parse.c | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)

diff --git a/pico/cd/cd_parse.c b/pico/cd/cd_parse.c
index aebd7d12..eeb3e880 100644
--- a/pico/cd/cd_parse.c
+++ b/pico/cd/cd_parse.c
@@ -71,23 +71,23 @@ static int get_token(const char *buff, char *dest, int len)
 static int get_ext(const char *fname, char ext[4],
 	char *base, size_t base_size)
 {
-	int len, pos = 0;
+	size_t pos = 0;
+	char *p;
 	
-	len = strrchr(fname, '.') - fname;
-	if (len > 0)
-		pos = len;
+	ext[0] = 0;
+	if (!(p = strrchr(fname, '.')))
+		return -1;
+	pos = p - fname;
 
 	strncpy(ext, fname + pos + 1, 4/*sizeof(ext)*/-1);
 	ext[4/*sizeof(ext)*/-1] = '\0';
 
 	if (base != NULL && base_size > 0) {
-		if (pos + 1 < base_size)
+		if (pos >= base_size)
 			pos = base_size - 1;
 
-		len = (pos < len) ? pos : len;
-
-		memcpy(base, fname, len);
-		base[len] = 0;
+		memcpy(base, fname, pos);
+		base[pos] = 0;
 	}
 	return pos;
 }
@@ -219,17 +219,17 @@ cd_data_t *cue_parse(const char *fname)
 	if (fname == NULL || (fname_len = strlen(fname)) == 0)
 		return NULL;
 
-	ret = get_ext(fname, ext, cue_base, sizeof(cue_base));
+	ret = get_ext(fname, ext, cue_base, sizeof(cue_base) - 4);
 	if (strcasecmp(ext, "cue") == 0) {
 		f = fopen(fname, "r");
 	}
-	else {
+	else if (strcasecmp(ext, "chd") != 0) {
 		// not a .cue, try one with the same base name
-		if (ret + 3 < sizeof(cue_base)) {
-			strcpy(cue_base + ret, "cue");
+		if (0 < ret && ret < sizeof(cue_base)) {
+			strcpy(cue_base + ret, ".cue");
 			f = fopen(cue_base, "r");
 			if (f == NULL) {
-				strcpy(cue_base + ret, "CUE");
+				strcpy(cue_base + ret, ".CUE");
 				f = fopen(cue_base, "r");
 			}
 		}
-- 
2.39.2


From 34070ccc345572d73a59c1a27579dc7602272577 Mon Sep 17 00:00:00 2001
From: irixxxx <31696370+irixxxx@users.noreply.github.com>
Date: Wed, 10 Jul 2024 19:17:04 +0200
Subject: [PATCH 06/16] add some images to readme

---
 README.md | 29 +++++++++++++++++++++++++++--
 1 file changed, 27 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index dd5e3604..9e506c95 100644
--- a/README.md
+++ b/README.md
@@ -1,8 +1,6 @@
 This is yet another SEGA 8 bit and 16 bit console emulator for emulating most
 of the hardware SEGA has published up to and including the 32X.
 
-Emulated 16 bit systems: Mega Drive/Genesis, Sega/Mega CD, 32X, Pico.  
-Emulated 8 bit systems: SG-1000, SC-3000, Master System/Mark III, Game Gear.
 
 PicoDrive was originally written having ARM-based handheld devices
 in mind, but later had got various improvements for other architectures
@@ -16,6 +14,33 @@ Currently the main development happens in the
 [irixxxx's fork](https://github.com/irixxxx/picodrive),
 [notaz's repo](https://github.com/notaz/picodrive) isn't updated as much.
 
+### gallery
+
+Some images of demos and homebrew software:
+
+**16/32 bit systems**: Mega Drive/Genesis, Sega/Mega CD, 32X, Pico.
+
+![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/02a4295b-ac9d-4114-bcd1-b5dd6e5930d0)
+![Raycast Demo](https://github.com/irixxxx/picodrive/assets/31696370/6e9c0bfe-49a9-45aa-bad7-544de065e388)
+![OpenLara](https://github.com/irixxxx/picodrive/assets/31696370/8a00002a-5c10-4d1d-a948-739bf978282a)
+
+![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/2e263e81-51c8-4daa-ab16-0b2cd5554f84)
+![DMA David](https://github.com/irixxxx/picodrive/assets/31696370/fbbeac15-8665-4d3e-9729-d1f8c35e417a)
+![Doom Resurrection](https://github.com/irixxxx/picodrive/assets/31696370/db7b7153-b917-4850-8442-a748c2fbb968)
+
+$~$
+
+**8 bit systems**: SG-1000, SC-3000, Master System/Mark III, Game Gear.
+
+![Cheril Perils Classics](https://github.com/irixxxx/picodrive/assets/31696370/653914a4-9f90-45f8-bd91-56e784df7550)
+![Stygian Quest](https://github.com/irixxxx/picodrive/assets/31696370/8196801b-85c8-4d84-97e1-ae57ab3d577f)
+![Wing Warriors](https://github.com/irixxxx/picodrive/assets/31696370/3c4a8f40-dad6-4fa4-b188-46b428a4b8c6)
+
+![Nyan Cat](https://github.com/irixxxx/picodrive/assets/31696370/6fe0d38b-549d-4faa-9351-b260a89dc745)
+![Anguna the Prison Dungeon](https://github.com/irixxxx/picodrive/assets/31696370/3264b962-7da2-4257-9ff7-1b509bd50cdf)
+![Turrican](https://github.com/irixxxx/picodrive/assets/31696370/c4eb2f2c-806e-4f4b-ac94-5c2cda82e962)
+
+
 ### compiling
 
 For platforms for which release builds are supplied the most easy way is to 
-- 
2.39.2


From 29dcbc52b2f279970fd3b76cc320389356a9042a Mon Sep 17 00:00:00 2001
From: irixxxx <31696370+irixxxx@users.noreply.github.com>
Date: Thu, 11 Jul 2024 22:56:04 +0200
Subject: [PATCH 07/16] add links to images

---
 README.md | 29 ++++++++++++-----------------
 1 file changed, 12 insertions(+), 17 deletions(-)

diff --git a/README.md b/README.md
index 9e506c95..8da8c717 100644
--- a/README.md
+++ b/README.md
@@ -14,31 +14,26 @@ Currently the main development happens in the
 [irixxxx's fork](https://github.com/irixxxx/picodrive),
 [notaz's repo](https://github.com/notaz/picodrive) isn't updated as much.
 
+
 ### gallery
 
 Some images of demos and homebrew software:
 
-**16/32 bit systems**: Mega Drive/Genesis, Sega/Mega CD, 32X, Pico.
-
-![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/02a4295b-ac9d-4114-bcd1-b5dd6e5930d0)
-![Raycast Demo](https://github.com/irixxxx/picodrive/assets/31696370/6e9c0bfe-49a9-45aa-bad7-544de065e388)
-![OpenLara](https://github.com/irixxxx/picodrive/assets/31696370/8a00002a-5c10-4d1d-a948-739bf978282a)
+**16 bit systems**: Mega Drive/Genesis, Sega/Mega CD, 32X, Pico.
 
-![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/2e263e81-51c8-4daa-ab16-0b2cd5554f84)
-![DMA David](https://github.com/irixxxx/picodrive/assets/31696370/fbbeac15-8665-4d3e-9729-d1f8c35e417a)
-![Doom Resurrection](https://github.com/irixxxx/picodrive/assets/31696370/db7b7153-b917-4850-8442-a748c2fbb968)
-
-$~$
+| ![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/02a4295b-ac9d-4114-bcd1-b5dd6e5930d0) | ![Raycast Demo](https://github.com/irixxxx/picodrive/assets/31696370/6e9c0bfe-49a9-45aa-bad7-544de065e388) | ![OpenLara](https://github.com/irixxxx/picodrive/assets/31696370/8a00002a-5c10-4d1d-a948-739bf978282a) |
+| --- | --- | --- |
+| [_MegaDrive: Titan Overdrive 2_](https://demozoo.org/productions/170767/) | [_MegaCD: RaycastDemo_](https://github.com/matteusbeus/RaycastDemo) | [_32X: OpenLara_](https://github.com/XProger/OpenLara/releases) |
+|![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/2e263e81-51c8-4daa-ab16-0b2cd5554f84)|![DMA David](https://github.com/irixxxx/picodrive/assets/31696370/fbbeac15-8665-4d3e-9729-d1f8c35e417a)|![Doom Resurrection](https://github.com/irixxxx/picodrive/assets/31696370/db7b7153-b917-4850-8442-a748c2fbb968)|
+|  | [_MegaDrive: DMA David_](http://www.mode5.net/DMA_David.html) | [_32X: Doom Resurrection_](https://archive.org/details/doom-32x-all-versions) |
 
 **8 bit systems**: SG-1000, SC-3000, Master System/Mark III, Game Gear.
 
-![Cheril Perils Classics](https://github.com/irixxxx/picodrive/assets/31696370/653914a4-9f90-45f8-bd91-56e784df7550)
-![Stygian Quest](https://github.com/irixxxx/picodrive/assets/31696370/8196801b-85c8-4d84-97e1-ae57ab3d577f)
-![Wing Warriors](https://github.com/irixxxx/picodrive/assets/31696370/3c4a8f40-dad6-4fa4-b188-46b428a4b8c6)
-
-![Nyan Cat](https://github.com/irixxxx/picodrive/assets/31696370/6fe0d38b-549d-4faa-9351-b260a89dc745)
-![Anguna the Prison Dungeon](https://github.com/irixxxx/picodrive/assets/31696370/3264b962-7da2-4257-9ff7-1b509bd50cdf)
-![Turrican](https://github.com/irixxxx/picodrive/assets/31696370/c4eb2f2c-806e-4f4b-ac94-5c2cda82e962)
+| ![Cheril Perils Classics](https://github.com/irixxxx/picodrive/assets/31696370/653914a4-9f90-45f8-bd91-56e784df7550) | ![Stygian Quest](https://github.com/irixxxx/picodrive/assets/31696370/8196801b-85c8-4d84-97e1-ae57ab3d577f) | ![Sword of Stone](https://github.com/irixxxx/picodrive/assets/31696370/3c4a8f40-dad6-4fa4-b188-46b428a4b8c6) |
+| --- | --- | --- |
+| [_SG-1000: Cheril Perils Classic_](https://www.smspower.org/Homebrew/CherilPerilsClassic-SG) | [_MasterSystem: Stygian Quest_](https://www.smspower.org/Homebrew/StygianQuest-SMS) | [_GameGear: The Sword of Stone_](https://www.smspower.org/Homebrew/SwordOfStone-GG) |
+| ![Nyan Cat](https://github.com/irixxxx/picodrive/assets/31696370/6fe0d38b-549d-4faa-9351-b260a89dc745) | ![Anguna the Prison Dungeon](https://github.com/irixxxx/picodrive/assets/31696370/3264b962-7da2-4257-9ff7-1b509bd50cdf) | ![Turrican](https://github.com/irixxxx/picodrive/assets/31696370/c4eb2f2c-806e-4f4b-ac94-5c2cda82e962) |
+| [_SG-1000: Nyan Cat_](https://www.smspower.org/Homebrew/NyanCat-SG) | [_MS: Anguna the Prison Dungeon_](https://www.smspower.org/Homebrew/AngunaThePrisonDungeon-SMS) | [_GameGear: Turrican_](https://www.smspower.org/Homebrew/GGTurrican-GG) |
 
 
 ### compiling
-- 
2.39.2


From e3fa3ee270fc5905943719e966277b8ae10e1cd4 Mon Sep 17 00:00:00 2001
From: Apaczer <94932128+Apaczer@users.noreply.github.com>
Date: Thu, 11 Jul 2024 22:49:13 +0200
Subject: [PATCH 08/16] platform-miyoo: change to IPK release output

create `platform/miyoo/pkg.cfg`
---
 .github/workflows/ci.yml |  4 ++--
 .gitignore               |  1 +
 Makefile                 |  8 ++++---
 platform/miyoo/pkg.cfg   | 45 ++++++++++++++++++++++++++++++++++++++++
 4 files changed, 53 insertions(+), 5 deletions(-)
 create mode 100644 platform/miyoo/pkg.cfg

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 50847c34..40b77daf 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -196,12 +196,12 @@ jobs:
         ver=$(cut -d'"' -f2 platform/common/version.h)-$(git rev-parse --short HEAD)
         ./configure --platform=miyoo
         make -j$(getconf _NPROCESSORS_ONLN)
-        mv PicoDrive-miyoo.zip PicoDrive-miyoo-$ver.zip
+        mv PicoDrive-miyoo.ipk PicoDrive-miyoo-$ver.ipk
     - name: artifacts
       uses: actions/upload-artifact@v4
       with:
         name: Miyoo
-        path: PicoDrive-miyoo*.zip
+        path: PicoDrive-miyoo*.ipk
 
   build-retrofw:
     runs-on: ubuntu-latest
diff --git a/.gitignore b/.gitignore
index 79cc126b..742cba8e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -30,6 +30,7 @@ PicoDrive
 picodrive
 picodrive_libretro*
 PicoDrive*.opk
+PicoDrive*.ipk
 PicoDrive*.zip
 pico_int_offs.h
 amalgamate
diff --git a/Makefile b/Makefile
index f6ae8a5a..48fa6a4d 100644
--- a/Makefile
+++ b/Makefile
@@ -141,10 +141,12 @@ all: PicoDrive-dge.zip
 CFLAGS += -DSDL_SURFACE_SW # some legacy dinguces had bugs in HWSURFACE
 else
 ifneq (,$(filter %__MIYOO__, $(CFLAGS)))
-PicoDrive-miyoo.zip: .od_data
+PicoDrive-miyoo.ipk: .od_data
 	rm -f .od_data/default.*.desktop .od_data/PicoDrive.dge
-	cd .od_data && zip -9 -r ../$@ *
-all: PicoDrive-miyoo.zip
+	gm2xpkg -i platform/miyoo/pkg.cfg
+	mv picodrive.ipk $@
+	@gm2xpkg -c platform/miyoo/pkg.cfg >/dev/null 2>&1
+all: PicoDrive-miyoo.ipk
 else
 PicoDrive.opk: .od_data
 	rm -f .od_data/PicoDrive.dge
diff --git a/platform/miyoo/pkg.cfg b/platform/miyoo/pkg.cfg
new file mode 100644
index 00000000..dd5fa0d8
--- /dev/null
+++ b/platform/miyoo/pkg.cfg
@@ -0,0 +1,45 @@
+# NOTES:
+## All variable values should enclosed within double quotes: "<value>"
+## CONFIGURATION FILE for `gm2xpkg` script version:
+PKGVER="0.4"
+
+# EXEC commands (set to "1" anyone for desired outcome), you can instead use [OPTIONS] of `gm2xpkg`:
+PACKAGE=""
+ZIP=""
+IPK=""
+CLEAN=""
+
+# ENV VAR.
+## Specific (mandatory to provide!)
+TARGET="picodrive"  # replace with binary name
+VERSION="v2.00"  # replace with correct release version if exist
+
+## Generic - common to all apps (better to not modify)
+HOMEPATH="/mnt"
+RELEASEDIR=""
+ASSETSDIR=".od_data"
+OPKG_ASSETSDIR=""
+LINK="" # full name of gm2x link, modify if exec binary name may be different from target name - place in CWD (warning: it may be removed with CLEAN=1)
+ALIASES="" # full name (with ext) of *.txt file with new names for selector e.g. old_title=new_title - place in CWD
+MANUAL="" # full name (with ext) of *.man.txt file with usage description of target app - place in CWD
+
+## Link entries (better modify if no <target_name>.lnk file provided)
+TITLE="PicoDrive"
+DESCRI="MegaDrive/MegaCD/32X emulator"
+SELDIR="/mnt/roms/SMD"
+DESTDIR="emus" # default=apps
+SECTION="emulators" # default=applications
+
+## Custom entries (if needed then modify)
+TARGET_DIR=""  # the directory /$HOMEPATH/$DESTDIR/TARGET_DIR of executable binary if not provided the TARGET_DIR=$TARGET
+DOCS=("COPYING" "AUTHORS") # array of extra text files e.g. =("LICENSE" "CHANGELOG" "CONTRIBUTORS") which will be copied & converted to *.txt files for ease of use by frontend 
+
+## IPK control entries (if needed then modify)
+PRIORITY=""
+MAINTAINER="irixxxx"
+CONFFILES=""
+ARCH="" # default=arm - do not modify for ARM chips
+# CONTROL= # automated output of *.ipk control config file
+DEPENDS="" # list of dependency packages e.g. ="sdl, libpng" or ="sdl (>= 2.9.2), sdl_mixer (= ed76d39cda0735d26c14a3e4f4da996e420f6478)" provide only for shared libs build, otherwise ignored
+SOURCE="https://github.com/irixxxx/picodrive"
+LICENSE="Custom"
-- 
2.39.2


From a9d1e99533de1fd378892c0da5caa564f2bb639b Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Fri, 12 Jul 2024 22:11:51 +0200
Subject: [PATCH 09/16] core, adjustment for background color DMA

---
 pico/videoport.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/pico/videoport.c b/pico/videoport.c
index 9663e2bf..9c35162d 100644
--- a/pico/videoport.c
+++ b/pico/videoport.c
@@ -626,7 +626,8 @@ static void DmaSlow(int len, u32 source)
         if (sl > VdpFIFO.fifo_hcounts[0]-5) // hint delay is 5 slots
           sl = (s8)sl;
         // TODO this is needed to cover timing inaccuracies
-        if (sl <= 12) sl = -2;
+        if (sl <= 12) sl = -3;
+        else if (sl <= 40) sl = 30;
         PicoDrawBgcDMA(base, source, mask, len, sl);
         // do last DMA cycle since it's all going to the same cram location
         source = source+len-1;
-- 
2.39.2


From c12b126f06818948dd89e7fd8524158582773e2a Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Fri, 12 Jul 2024 22:57:25 +0200
Subject: [PATCH 10/16] platform miyoo, switch menu key to RESET

---
 platform/common/inputmap_kbd.c |  3 ++-
 platform/common/plat_sdl.c     |  5 ++---
 platform/common/plat_sdl.h     |  2 +-
 platform/opendingux/inputmap.c | 13 ++++++++++++-
 4 files changed, 17 insertions(+), 6 deletions(-)

diff --git a/platform/common/inputmap_kbd.c b/platform/common/inputmap_kbd.c
index c41abbb3..8fbb29f8 100644
--- a/platform/common/inputmap_kbd.c
+++ b/platform/common/inputmap_kbd.c
@@ -7,7 +7,7 @@
 #include "../common/input_pico.h"
 #include "../common/plat_sdl.h"
 
-const struct in_default_bind in_sdl_defbinds[] = {
+const struct in_default_bind _in_sdl_defbinds[] = {
 	{ SDLK_UP,     IN_BINDTYPE_PLAYER12, GBTN_UP },
 	{ SDLK_DOWN,   IN_BINDTYPE_PLAYER12, GBTN_DOWN },
 	{ SDLK_LEFT,   IN_BINDTYPE_PLAYER12, GBTN_LEFT },
@@ -35,6 +35,7 @@ const struct in_default_bind in_sdl_defbinds[] = {
 	{ SDLK_BACKSPACE, IN_BINDTYPE_EMU, PEVB_FF },
 	{ 0, 0, 0 }
 };
+const struct in_default_bind *in_sdl_defbinds = _in_sdl_defbinds;
 
 const struct menu_keymap _in_sdl_key_map[] = {
 	{ SDLK_UP,	PBTN_UP },
diff --git a/platform/common/plat_sdl.c b/platform/common/plat_sdl.c
index 140480e8..d1e833f3 100644
--- a/platform/common/plat_sdl.c
+++ b/platform/common/plat_sdl.c
@@ -26,9 +26,7 @@ static void *shadow_fb;
 static int shadow_size;
 static struct area { int w, h; } area;
 
-static struct in_pdata in_sdl_platform_data = {
-	.defbinds = in_sdl_defbinds,
-};
+static struct in_pdata in_sdl_platform_data;
 
 static int sound_rates[] = { 8000, 11025, 16000, 22050, 32000, 44100, 53000, -1 };
 struct plat_target plat_target = { .sound_rates = sound_rates };
@@ -438,6 +436,7 @@ void plat_init(void)
 	g_screen_ptr = shadow_fb;
 
 	plat_target_setup_input();
+	in_sdl_platform_data.defbinds = in_sdl_defbinds,
 	in_sdl_platform_data.kmap_size = in_sdl_key_map_sz,
 	in_sdl_platform_data.key_map = in_sdl_key_map,
 	in_sdl_platform_data.jmap_size = in_sdl_joy_map_sz,
diff --git a/platform/common/plat_sdl.h b/platform/common/plat_sdl.h
index aefa8076..cecdb2b3 100644
--- a/platform/common/plat_sdl.h
+++ b/platform/common/plat_sdl.h
@@ -1,5 +1,5 @@
 
-extern const struct in_default_bind in_sdl_defbinds[];
+extern const struct in_default_bind *in_sdl_defbinds;
 extern const struct menu_keymap *in_sdl_key_map;
 extern const int in_sdl_key_map_sz;
 extern const struct menu_keymap *in_sdl_joy_map;
diff --git a/platform/opendingux/inputmap.c b/platform/opendingux/inputmap.c
index e8eace45..30c4639a 100644
--- a/platform/opendingux/inputmap.c
+++ b/platform/opendingux/inputmap.c
@@ -10,7 +10,7 @@
 #include "../common/plat_sdl.h"
 #include "../common/emu.h"
 
-const struct in_default_bind in_sdl_defbinds[] = {
+struct in_default_bind _in_sdl_defbinds[] = {
 	{ SDLK_UP,	IN_BINDTYPE_PLAYER12, GBTN_UP },
 	{ SDLK_DOWN,	IN_BINDTYPE_PLAYER12, GBTN_DOWN },
 	{ SDLK_LEFT,	IN_BINDTYPE_PLAYER12, GBTN_LEFT },
@@ -27,6 +27,7 @@ const struct in_default_bind in_sdl_defbinds[] = {
 	{ SDLK_SPACE,	IN_BINDTYPE_EMU, PEVB_FF },
 	{ 0, 0, 0 }
 };
+const struct in_default_bind *in_sdl_defbinds = _in_sdl_defbinds;
 
 struct menu_keymap _in_sdl_key_map[] = {
 	{ SDLK_UP,	PBTN_UP },
@@ -122,6 +123,15 @@ static void keyswap(int k1, int k2)
 	}
 }
 
+static void bindswap(int k1, int k2)
+{
+	int i;
+
+	for (i = 0; _in_sdl_defbinds[i].code; i++)
+		if (_in_sdl_defbinds[i].code == k1)
+			_in_sdl_defbinds[i].code = k2;
+}
+
 void plat_target_setup_input(void)
 {
 	if (strcmp(plat_device, "miyoo") == 0) {
@@ -129,6 +139,7 @@ void plat_target_setup_input(void)
 		keyswap(SDLK_LALT, SDLK_LCTRL);
 		nameswap(SDLK_LALT, SDLK_LCTRL);
 		nameswap(SDLK_SPACE, SDLK_LSHIFT);
+		bindswap(SDLK_ESCAPE, SDLK_RCTRL);
 	} else if (strcmp(plat_device, "gcw0") == 0) {
 		/* swapped X/Y keys, single L/R keys */
 		nameswap(SDLK_SPACE, SDLK_LSHIFT);
-- 
2.39.2


From b4244cbb6dfe5607911fbe78c16ad1b1f7cc11fc Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Fri, 12 Jul 2024 22:15:21 +0200
Subject: [PATCH 11/16] update README

---
 README.md | 130 +++++++++++++++++++++++++++++-------------------------
 1 file changed, 71 insertions(+), 59 deletions(-)

diff --git a/README.md b/README.md
index 8da8c717..527aa29b 100644
--- a/README.md
+++ b/README.md
@@ -1,33 +1,39 @@
-This is yet another SEGA 8 bit and 16 bit console emulator for emulating most
-of the hardware SEGA has published up to and including the 32X.
+This is yet another SEGA 8 bit and 16 bit console emulator.
 
+It can run games developed for most consumer hardware released
+by SEGA, up to and including the 32X:
+- **16 bit systems:** Mega Drive/Genesis, Sega/Mega CD, 32X, Pico
+- **8 bit systems**: SG-1000, SC-3000, Master System/Mark III, Game Gear
 
-PicoDrive was originally written having ARM-based handheld devices
-in mind, but later had got various improvements for other architectures
-too, like SH2 recompilers for MIPS (mips32r2), ARM64 (armv8), RISC-V (RV64IM)
+PicoDrive was originally created with ARM-based handheld devices
+in mind, but later received various cross-platform improvements
+such as SH2 recompilers for MIPS (mips32r2), ARM64 (armv8), RISC-V (RV64IM)
 and PowerPC (G4/2.03).
 
-PicoDrive was the first emulator ever to properly emulate Virtua Racing and
+PicoDrive was the first software to properly emulate Virtua Racing and
 its SVP chip.
 
-Currently the main development happens in the
-[irixxxx's fork](https://github.com/irixxxx/picodrive),
-[notaz's repo](https://github.com/notaz/picodrive) isn't updated as much.
+At present, most development activity occurs in
+[irixxxx's fork](https://github.com/irixxxx/picodrive);
+[notaz's repo](https://github.com/notaz/picodrive) is updated less frequently.
 
+### Sega Pico and Storyware Pages
 
-### gallery
+PicoDrive can use Storyware pages and pad overlays in png format in the same
+directory as the cartridge image. The selected page is displayed automatically
+if the pen is used on the storyware or pad. Details about how to correctly name
+the pages can be found in the *How to run Sega Pico games* section in
+`platform/base_readme.txt`.
 
-Some images of demos and homebrew software:
+### Gallery
 
-**16 bit systems**: Mega Drive/Genesis, Sega/Mega CD, 32X, Pico.
+Some images of demos and homebrew software:
 
 | ![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/02a4295b-ac9d-4114-bcd1-b5dd6e5930d0) | ![Raycast Demo](https://github.com/irixxxx/picodrive/assets/31696370/6e9c0bfe-49a9-45aa-bad7-544de065e388) | ![OpenLara](https://github.com/irixxxx/picodrive/assets/31696370/8a00002a-5c10-4d1d-a948-739bf978282a) |
 | --- | --- | --- |
 | [_MegaDrive: Titan Overdrive 2_](https://demozoo.org/productions/170767/) | [_MegaCD: RaycastDemo_](https://github.com/matteusbeus/RaycastDemo) | [_32X: OpenLara_](https://github.com/XProger/OpenLara/releases) |
 |![Titan Overdrive 2](https://github.com/irixxxx/picodrive/assets/31696370/2e263e81-51c8-4daa-ab16-0b2cd5554f84)|![DMA David](https://github.com/irixxxx/picodrive/assets/31696370/fbbeac15-8665-4d3e-9729-d1f8c35e417a)|![Doom Resurrection](https://github.com/irixxxx/picodrive/assets/31696370/db7b7153-b917-4850-8442-a748c2fbb968)|
-|  | [_MegaDrive: DMA David_](http://www.mode5.net/DMA_David.html) | [_32X: Doom Resurrection_](https://archive.org/details/doom-32x-all-versions) |
-
-**8 bit systems**: SG-1000, SC-3000, Master System/Mark III, Game Gear.
+| [_MegaDrive: Titan Overdrive 2_](https://www.pouet.net/prod.php?which=69648) | [_MegaDrive: DMA David_](http://www.mode5.net/DMA_David.html) | [_32X: Doom Resurrection_](https://archive.org/details/doom-32x-all-versions) |
 
 | ![Cheril Perils Classics](https://github.com/irixxxx/picodrive/assets/31696370/653914a4-9f90-45f8-bd91-56e784df7550) | ![Stygian Quest](https://github.com/irixxxx/picodrive/assets/31696370/8196801b-85c8-4d84-97e1-ae57ab3d577f) | ![Sword of Stone](https://github.com/irixxxx/picodrive/assets/31696370/3c4a8f40-dad6-4fa4-b188-46b428a4b8c6) |
 | --- | --- | --- |
@@ -35,62 +41,68 @@ Some images of demos and homebrew software:
 | ![Nyan Cat](https://github.com/irixxxx/picodrive/assets/31696370/6fe0d38b-549d-4faa-9351-b260a89dc745) | ![Anguna the Prison Dungeon](https://github.com/irixxxx/picodrive/assets/31696370/3264b962-7da2-4257-9ff7-1b509bd50cdf) | ![Turrican](https://github.com/irixxxx/picodrive/assets/31696370/c4eb2f2c-806e-4f4b-ac94-5c2cda82e962) |
 | [_SG-1000: Nyan Cat_](https://www.smspower.org/Homebrew/NyanCat-SG) | [_MS: Anguna the Prison Dungeon_](https://www.smspower.org/Homebrew/AngunaThePrisonDungeon-SMS) | [_GameGear: Turrican_](https://www.smspower.org/Homebrew/GGTurrican-GG) |
 
+### Compiling
 
-### compiling
-
-For platforms for which release builds are supplied the most easy way is to 
-use the release script in tools/release.sh. See the release script for details.
-To make a platform build use
+For platforms where release builds are provided, the simplest method is to
+use the release script `tools/release.sh`. See the script itself for details.
+To create platform builds run the command:
 
-> tools/release.sh [version] [platforms...]
+```
+tools/release.sh [version] [platforms...]
+```
 
-This will deliver a file for each platform in a the release-[version] directory.
-A list of platforms is in the release script.
+This will generate a file for each platform in the `release-[version]` directory.
+A list of supported platforms can be found in the release script.
 
-If you want to build an executable for a unixoid platform not listed in the
-platform list, try using
+These commands should create an executable for a unixoid platform not included in the list:
 
-> configure --platform=generic
+```
+configure --platform=generic
+make
+```
 
-If DRC is available for the platform, it should be enabled automatically.
-For gp2x, wiz, and caanoo you may need to compile libpng first.
+To compile PicoDrive as a libretro core, use this command:
 
-After configure, compile with
+```
+make -f Makefile.libretro
+```
 
-> make
-
-
-### helix MP3 decoder for ARM
+### Helix MP3 decoder for ARM
 
 For 32 bit ARM platforms, the optimized helix MP3 decoder can be used to play
-MP3 audio files with CD games. The helix source files are however not supplied
-due to licensing issues. If you have legally obtained the sources, put them in
-the platform/common/helix directory. 
-
-To compile the helix sources, set CROSS_COMPILE to your cross compiler prefix
-(e.g. arm-linux-gnueabi-) and LIBGCC to your cross compiler's libgcc.a
-(e.g. /usr/lib/gcc-cross/arm-linux-gnueabi/4.7/libgcc.a), and compile with
-
-> make -C platform/common/helix CROSS_COMPILE=$CROSS_COMPILE LIBGCC=$LIBGCC
-
-This will result in a shared library named ${CROSS_COMPILE}helix_mp3.so. Copy
-this as libhelix.so to where the PicoDrive binary is on the target device.
-
-
-Also, the support for helix must be enabled in PicoDrive by compiling with
-
-> make PLATFORM_MP3=1
-
-This switch is automatically enabled for Gamepark Holdings devices (gp2x,
-caanoo and wiz). Without installing libhelix.so those devices will not play
+MP3 audio files with CD games. Due to licensing issues, the helix source files
+cannot be provided here; if you have obtained the sources legally, place them in
+the `platform/common/helix` directory.
+
+To compile the helix sources:
+
+- Set the environment variable `CROSS_COMPILE` to your cross compiler prefix
+(e.g. `arm-linux-gnueabi-`)
+- Set the environment variable `LIBGCC` to your cross compiler's `libgcc.a`
+(e.g. `/usr/lib/gcc-cross/arm-linux-gnueabi/4.7/libgcc.a`)
+- Run the command:
+```
+make -C platform/common/helix CROSS_COMPILE=$CROSS_COMPILE LIBGCC=$LIBGCC
+```
+- Copy the resulting shared library named `${CROSS_COMPILE}helix_mp3.so` as
+`libhelix.so` to the directory containing the PicoDrive binary on the target device.
+
+In addition, helix support must be enabled in PicoDrive itself by compiling with:
+
+```
+make PLATFORM_MP3=1
+```
+
+This switch is enabled automatically for Gamepark Holdings devices (`gp2x`,
+`caanoo` and `wiz`). Without installing `libhelix.so`, these devices will not play
 MP3 audio.
 
+### Installing
 
-### installing
+The release script produces packages or zip archives which have to be installed
+manually on the target device. Usually this involves unpacking the archive or 
+copying the package to a directory on either the internal device storage or an
+SD card. Device-specific instructions can be found on the internet.
 
-The release scripts produces zip files which need to be installed on the
-target device manually. Usually that means unpacking the file to some
-directory on the device storage or on an SD card. See device specific
-descriptions on the net.
 
-Send bug reports, fixes etc to <derkub@gmail.com>
+Send bug reports, fixes etc. to <derkub@gmail.com>
-- 
2.39.2


From 4a347fbe946e4ee4ae0687f4139a8f6f0aa179cd Mon Sep 17 00:00:00 2001
From: Apaczer <94932128+Apaczer@users.noreply.github.com>
Date: Mon, 15 Jul 2024 23:13:09 +0200
Subject: [PATCH 12/16] platform/miyoo: scripted versioning

update pkg.cfg
---
 platform/miyoo/pkg.cfg | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/platform/miyoo/pkg.cfg b/platform/miyoo/pkg.cfg
index dd5fa0d8..940cf277 100644
--- a/platform/miyoo/pkg.cfg
+++ b/platform/miyoo/pkg.cfg
@@ -12,8 +12,8 @@ CLEAN=""
 # ENV VAR.
 ## Specific (mandatory to provide!)
 TARGET="picodrive"  # replace with binary name
-VERSION="v2.00"  # replace with correct release version if exist
-
+VERSION="$(cut -d'"' -f2 platform/common/version.h)"  # replace with correct release version if exist
+test "$(git describe --tags --exact-match HEAD 2>/dev/null)" == "" && VERSION="${VERSION}-$(git rev-parse --short HEAD || echo unknown)"
 ## Generic - common to all apps (better to not modify)
 HOMEPATH="/mnt"
 RELEASEDIR=""
-- 
2.39.2


From 1fad746a1f0fa19c874be4130e328141bcefb974 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Wed, 17 Jul 2024 22:47:02 +0200
Subject: [PATCH 13/16] core vdp, fix layer/window borders for vertical window

---
 pico/draw.c     | 235 +++++++++++++++++++++++++++---------------------
 pico/draw_arm.S |  83 ++++++++++++++---
 2 files changed, 205 insertions(+), 113 deletions(-)

diff --git a/pico/draw.c b/pico/draw.c
index 5868268b..0f9b137a 100644
--- a/pico/draw.c
+++ b/pico/draw.c
@@ -304,13 +304,43 @@ TileFlipMakerAS(TileFlipSH_AS_and, pix_sh_as_and)
 // --------------------------------------------
 
 #ifndef _ASM_DRAW_C
+#define DrawTile(mask) { \
+  if (code!=oldcode) { \
+    oldcode = code; \
+ \
+    pack = 0; \
+    if (code != blank) { \
+      /* Get tile address/2: */\
+      u32 addr = ((code&0x7ff)<<4) + ty; \
+      if (code & 0x1000) addr ^= 0xe; /* Y-flip */ \
+ \
+      pal = ((code>>9)&0x30) | sh; /* shadow */ \
+ \
+      pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr)); \
+      if (!pack) \
+        blank = code; \
+    } \
+  } \
+ \
+  if (code & 0x8000) { /* (un-forced) high priority tile */ \
+    if (sh | (pack&mask)) { \
+      code |= (dx<<16) | (ty<<25); \
+      if (code & 0x1000) code ^= 0xe<<25; \
+      *hc++ = code, *hc++ = pack&mask; /* cache it */ \
+    } \
+  } else if (pack&mask) { \
+    if (code & 0x0800) TileFlip(pd + dx, pack&mask, pal); \
+    else               TileNorm(pd + dx, pack&mask, pal); \
+  } \
+}
+
 static void DrawStrip(struct TileStrip *ts, int lflags, int cellskip)
 {
   unsigned char *pd = Pico.est.HighCol;
   u32 *hc = ts->hc;
   int tilex, dx, ty, cells;
-  u32 pack = 0, oldcode = -1, blank = -1; // The tile we know is blank
-  unsigned int pal = 0, sh;
+  u32 code, oldcode = -1, blank = -1; // The tile we know is blank
+  unsigned int pal = 0, pack = 0, sh, mask = ~0;
 
   // Draw tiles across screen:
   sh = (lflags & LF_SH) << 6; // shadow
@@ -318,42 +348,39 @@ static void DrawStrip(struct TileStrip *ts, int lflags, int cellskip)
   ty=(ts->line&7)<<1; // Y-Offset into tile
   dx=((ts->hscroll-1)&7)+1;
   cells = ts->cells - cellskip;
-  if(dx != 8) cells++; // have hscroll, need to draw 1 cell more
   dx+=cellskip<<3;
 
+  if (dx & 7) {
+    code = PicoMem.vram[ts->nametab + (tilex & ts->xmask)];
+    mask = 0xffffffff<<((dx&7)*4);
+    if (code & 0x0800) mask = 0xffffffff>>((dx&7)*4);
+    mask = (~mask << 16) | (~mask >> 16);
+
+    DrawTile(mask);
+    dx += 8, tilex++, cells--;
+  }
+
 //  int force = (lflags&LF_FORCE) << 13;
   for (; cells > 0; dx+=8, tilex++, cells--)
   {
-    u32 code = PicoMem.vram[ts->nametab + (tilex & ts->xmask)];
+    code = PicoMem.vram[ts->nametab + (tilex & ts->xmask)];
 //    code &= ~force; // forced always draw everything
 
     if (code == blank && !((code & 0x8000) && sh))
       continue;
 
-    if (code!=oldcode) {
-      oldcode = code;
-
-      pack = 0;
-      if (code != blank) {
-        // Get tile address/2:
-        u32 addr = ((code&0x7ff)<<4) + ty;
-        if (code & 0x1000) addr ^= 0xe; // Y-flip
-
-        pal = ((code>>9)&0x30) | sh; // shadow
+    DrawTile(~0);
+  }
 
-        pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr));
-        if (!pack)
-          blank = code;
-      }
-    }
+  if (dx & 7) {
+    code = PicoMem.vram[ts->nametab + (tilex & ts->xmask)];
+//    code &= ~force; // forced always draw everything
+    if (!(code == blank && !((code & 0x8000) && sh))) {
+      mask = 0xffffffff<<((dx&7)*4);
+      if (code & 0x0800) mask = 0xffffffff>>((dx&7)*4);
+      mask = (mask << 16) | (mask >> 16);
 
-    if (code & 0x8000) { // (un-forced) high priority tile
-      code |= (dx<<16) | (ty<<25);
-      if (code & 0x1000) code ^= 0xe<<25;
-      *hc++ = code, *hc++ = pack; // cache it
-    } else if (code != blank) {
-      if (code & 0x0800) TileFlip(pd + dx, pack, pal);
-      else               TileNorm(pd + dx, pack, pal);
+      DrawTile(mask);
     }
   }
 
@@ -451,6 +478,36 @@ static void DrawStripVSRam(struct TileStrip *ts, int plane_sh, int cellskip)
 }
 #endif
 
+#define DrawTileInterlace(mask) { \
+  if (code!=oldcode) { \
+    oldcode = code; \
+ \
+    pack = 0; \
+    if (code != blank) { \
+      /* Get tile address/2: */ \
+      u32 addr = ((code&0x3ff)<<5) + ty; \
+      if (code & 0x1000) addr ^= 0x1e; /* Y-flip */ \
+ \
+      pal = ((code>>9)&0x30) | sh; /* shadow */ \
+ \
+      pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr)); \
+      if (!pack) \
+        blank = code; \
+    } \
+  } \
+ \
+  if (code & 0x8000) { /* high priority tile */ \
+    if (sh | (pack&mask)) { \
+      code = (code&0xfc00) | ((code&0x3ff)<<1) | (dx<<16) | (ty<<25); \
+      if (code & 0x1000) code ^= 0x1e<<25; \
+      *hc++ = code, *hc++ = pack&mask; /* cache it */ \
+    } \
+  } else if (pack&mask) { \
+    if (code & 0x0800) TileFlip(pd + dx, pack&mask, pal); \
+    else               TileNorm(pd + dx, pack&mask, pal); \
+  } \
+}
+
 #ifndef _ASM_DRAW_C
 static
 #endif
@@ -459,8 +516,8 @@ void DrawStripInterlace(struct TileStrip *ts, int plane_sh)
   unsigned char *pd = Pico.est.HighCol;
   u32 *hc = ts->hc;
   int tilex = 0, dx = 0, ty = 0, cells;
-  u32 oldcode = -1, blank = -1; // The tile we know is blank
-  unsigned int pal = 0, pack = 0, sh;
+  u32 code, oldcode = -1, blank = -1; // The tile we know is blank
+  unsigned int pal = 0, pack = 0, sh, mask = ~0;
 
   // Draw tiles across screen:
   sh = (plane_sh & LF_SH) << 6; // shadow
@@ -468,7 +525,16 @@ void DrawStripInterlace(struct TileStrip *ts, int plane_sh)
   ty=(ts->line&15)<<1; // Y-Offset into tile
   dx=((ts->hscroll-1)&7)+1;
   cells = ts->cells;
-  if(dx != 8) cells++; // have hscroll, need to draw 1 cell more
+
+  if (dx & 7) {
+    code = PicoMem.vram[ts->nametab + (tilex & ts->xmask)];
+    mask = 0xffffffff<<(dx*4);
+    if (code & 0x0800) mask = 0xffffffff>>(dx*4);
+    mask = (~mask << 16) | (~mask >> 16);
+
+    DrawTileInterlace(mask);
+    dx += 8, tilex++, cells--;
+  }
 
 //  int force = (plane_sh&LF_FORCE) << 13;
   for (; cells; dx+=8,tilex++,cells--)
@@ -479,33 +545,18 @@ void DrawStripInterlace(struct TileStrip *ts, int plane_sh)
     if (code == blank && !(code & 0x8000))
       continue;
 
-    if (code!=oldcode) {
-      oldcode = code;
-
-      pack = 0;
-      if (code != blank) {
-        // Get tile address/2:
-        u32 addr = ((code&0x3ff)<<5) + ty;
-        if (code & 0x1000) addr ^= 0x1e; // Y-flip
-
-        pal = ((code>>9)&0x30) | sh; // shadow
+    DrawTileInterlace(~0);
+  }
 
-        pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr));
-        if (!pack)
-          blank = code;
-      }
-    }
+  if (dx & 7) {
+    code = PicoMem.vram[ts->nametab + (tilex & ts->xmask)];
+//    code &= ~force; // forced always draw everything
+    if (!(code == blank && !((code & 0x8000) && sh))) {
+      mask = 0xffffffff<<((dx&7)*4);
+      if (code & 0x0800) mask = 0xffffffff>>((dx&7)*4);
+      mask = (mask << 16) | (mask >> 16);
 
-    if (code & 0x8000) { // high priority tile
-      if ((plane_sh&LF_SH) | (code!=blank)) {
-        code = (code&0xfc00) | ((code&0x3ff)<<1) | (dx<<16) | (ty<<25);
-        if (code & 0x1000) code ^= 0x1e<<25;
-        *hc++ = code, *hc++ = pack; // cache it
-      }
-      continue;
-    } else if (code != blank) {
-      if (code & 0x0800) TileFlip(pd + dx, pack, pal);
-      else               TileNorm(pd + dx, pack, pal);
+      DrawTileInterlace(mask);
     }
   }
 
@@ -698,19 +749,6 @@ static void DrawWindow(int tstart, int tend, int prio, int sh,
 
 // --------------------------------------------
 
-static void DrawTilesFromCacheShPrep(void)
-{
-  // as some layer has covered whole line with hi priority tiles,
-  // we can process whole line and then act as if sh/hi mode was off,
-  // but leave lo pri op sprite markers alone
-  int c = 320/4, *zb = (int *)(Pico.est.HighCol+8);
-  Pico.est.rendstatus |= PDRAW_SHHI_DONE;
-  while (c--)
-  {
-    *zb++ &= 0x7f7f7f7f;
-  }
-}
-
 static void DrawTilesFromCache(u32 *hc, int sh, int rlim, struct PicoEState *est)
 {
   unsigned char *pd = est->HighCol;
@@ -722,8 +760,18 @@ static void DrawTilesFromCache(u32 *hc, int sh, int rlim, struct PicoEState *est
 
   if (sh && (est->rendstatus & (PDRAW_SHHI_DONE|PDRAW_PLANE_HI_PRIO)))
   {
-    if (!(est->rendstatus & PDRAW_SHHI_DONE))
-      DrawTilesFromCacheShPrep();
+    if (!(est->rendstatus & PDRAW_SHHI_DONE)) {
+      // as some layer has covered whole line with hi priority tiles,
+      // we can process whole line and then act as if sh/hi mode was off,
+      // but leave lo pri op sprite markers alone
+      int *zb = (int *)(Pico.est.HighCol+8);
+      int c = rlim / 4;
+      while (c--)
+      {
+        *zb++ &= 0x7f7f7f7f;
+      }
+      Pico.est.rendstatus |= PDRAW_SHHI_DONE;
+    }
     sh = 0;
   }
 
@@ -731,13 +779,13 @@ static void DrawTilesFromCache(u32 *hc, int sh, int rlim, struct PicoEState *est
   {
     while ((code=*hc++)) {
       pack = *hc++;
+      if (rlim-dx < 0)
+        goto last_cut_tile;
       if (!pack)
         continue;
 
       dx = (code >> 16) & 0x1ff;
       pal = ((code >> 9) & 0x30);
-      if (rlim-dx < 0)
-        goto last_cut_tile;
 
       if (code & 0x0800) TileFlip(pd + dx, pack, pal);
       else               TileNorm(pd + dx, pack, pal);
@@ -754,12 +802,12 @@ static void DrawTilesFromCache(u32 *hc, int sh, int rlim, struct PicoEState *est
       *zb++ &= 0x7f; *zb++ &= 0x7f; *zb++ &= 0x7f; *zb++ &= 0x7f;
 
       pack = *hc++;
+      if (rlim - dx < 0)
+        goto last_cut_tile;
       if (!pack)
         continue;
 
       pal = ((code >> 9) & 0x30);
-      if (rlim - dx < 0)
-        goto last_cut_tile;
 
       if (code & 0x0800) TileFlip(pd + dx, pack, pal);
       else               TileNorm(pd + dx, pack, pal);
@@ -770,35 +818,18 @@ static void DrawTilesFromCache(u32 *hc, int sh, int rlim, struct PicoEState *est
 last_cut_tile:
   // for vertical window cutoff
   {
-    unsigned int t;
-
-    pd += dx;
-    if (code&0x0800)
-    {
-      switch (rlim-dx+8)
-      {
-        case 7: t=pack&0x00000f00; if (t) pd[6]=(unsigned char)(pal|(t>> 8)); // "break" is left out intentionally
-        case 6: t=pack&0x000000f0; if (t) pd[5]=(unsigned char)(pal|(t>> 4));
-        case 5: t=pack&0x0000000f; if (t) pd[4]=(unsigned char)(pal|(t    ));
-        case 4: t=pack&0xf0000000; if (t) pd[3]=(unsigned char)(pal|(t>>28));
-        case 3: t=pack&0x0f000000; if (t) pd[2]=(unsigned char)(pal|(t>>24));
-        case 2: t=pack&0x00f00000; if (t) pd[1]=(unsigned char)(pal|(t>>20));
-        case 1: t=pack&0x000f0000; if (t) pd[0]=(unsigned char)(pal|(t>>16));
-        default: break;
-      }
-    }
-    else
-    {
-      switch (rlim-dx+8)
-      {
-        case 7: t=pack&0x00f00000; if (t) pd[6]=(unsigned char)(pal|(t>>20));
-        case 6: t=pack&0x0f000000; if (t) pd[5]=(unsigned char)(pal|(t>>24));
-        case 5: t=pack&0xf0000000; if (t) pd[4]=(unsigned char)(pal|(t>>28));
-        case 4: t=pack&0x0000000f; if (t) pd[3]=(unsigned char)(pal|(t    ));
-        case 3: t=pack&0x000000f0; if (t) pd[2]=(unsigned char)(pal|(t>> 4));
-        case 2: t=pack&0x00000f00; if (t) pd[1]=(unsigned char)(pal|(t>> 8));
-        case 1: t=pack&0x0000f000; if (t) pd[0]=(unsigned char)(pal|(t>>12));
-        default: break;
+    unsigned int t, mask;
+
+    // rlim-dx + 8 px to draw -> mask shift 8-(rlim-dx + 8)
+    t = -(rlim - dx);
+    if (t < 8) {
+      mask = 0xffffffff<<(t*4);
+      if (code & 0x0800) mask = 0xffffffff>>(t*4);
+      mask = (mask << 16) | (mask >> 16);
+
+      if (pack&mask) {
+        if (code & 0x0800) TileFlip(pd + dx, pack&mask, pal);
+        else               TileNorm(pd + dx, pack&mask, pal);
       }
     }
   }
diff --git a/pico/draw_arm.S b/pico/draw_arm.S
index e3f6ce83..7a2e6f1d 100644
--- a/pico/draw_arm.S
+++ b/pico/draw_arm.S
@@ -433,11 +433,11 @@ DrawLayer:
     movs    r3, r9, lsl #1  @ (force[31]|sh[30]) << 1
     mov     r3, #0
     orrmi   r10,r10, #1<<23 @ r10=cells[31:24]|sh[23]|hi_not_empty[22]
-@    orrcc   r10,r10, #1<<20 @    |had_output[21]|!force[20]|ty[15:0]
+@    orrcc   r10,r10, #1<<20 @   |had_output[21]|!force[20]|hscroll[19:17]|ty[15:0]
     movmi   r3, #0x80       @ default to shadowed pal on sh mode
 
-    cmp     r7, #8
-    addne   r10,r10, #0x01000000 @ we will loop cells+1 times if there is scroll
+    and     r4, r7, #7
+    orr     r10,r10, r4, lsl #16 @ we will process cells+1 if there is scroll
 
     and     r9, r9, #0xff00
     add     r8, r8, r9, lsr #8   @ tilex+=cellskip
@@ -452,9 +452,40 @@ DrawLayer:
     mvn     r9, #0               @ r9=prevcode=-1
     add     r1, r11, r7          @ r1=pdest
 
-    @ r10=cells[31:24]|sh[23]|hi_not_empty[22]|had_output[21]|!force[20]|ty[15:0]
+    @ r10=cells[31:24]|sh[23]|hi_not_empty[22]|had_output[21]|!force[20]|hscroll[19:17]|ty[15:0]
     @ r1=pd+dx r2=pack r3=pal r5=xmask r6=hc r8=tilex r9=prevcode r11=HighCol r12=nametab lr=vram
     @ r4 & r7 are scratch in this loop
+
+    ands    r4, r10, #7<<16  @ hscroll?
+    beq     .dsloop_subr1
+    subs    r10,r10, #0x01000000
+    bmi     .dsloop_exit
+
+    and     r7, r5, r8         @ do first cut tile
+    add     r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords)
+    ldrh    r9, [r7, r12]      @ r7=code (int, but from unsigned, no sign extend)
+
+    add     r8, r8, #1
+
+    movs    r2, r9, lsl #20 @ if (code&0x1000)
+    mov     r2, r2, lsl #1
+    add     r2, r2, r10, lsl #17
+    mov     r2, r2, lsr #17
+    eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
+
+    ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
+
+    mvn     r7, #0
+    mov     r4, r4, lsr #16-2  @ (dx&7)*4
+    tst     r9, #0x0800
+    moveq   r7, r7, lsl r4     @ mask = ~0 [shift] (dx&7)*4
+    movne   r7, r7, lsr r4
+    mvn     r7, r7, ror #16
+    and     r2, r2, r7         @ pack&mask
+
+    orr     r9, r9, #0x80000000 @ invalidate oldcode since pack is masked
+    b       .DrawStrip_samecode
+
 .dsloop_subr1:
     sub     r1, r1, #8
 .dsloop: @ 40-41 times
@@ -480,7 +511,6 @@ DrawLayer:
     eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
 
     ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
-
 .DrawStrip_samecode:
     tst     r9, #0x8000
 @    tstne   r10, #1<<20     @ !force[20]
@@ -538,6 +568,35 @@ DrawLayer:
     b       .dsloop
 
 .dsloop_exit:
+    ands    r4,r10, #7<<16        @ hscroll?
+    beq     .DrawStrip_noscroll
+
+    and     r7, r5, r8         @ do one more cut tile
+    add     r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords)
+    ldrh    r9, [r7, r12]      @ r7=code (int, but from unsigned, no sign extend)
+
+    add     r1, r1, #8
+
+    movs    r2, r9, lsl #20 @ if (code&0x1000)
+    mov     r2, r2, lsl #1
+    add     r2, r2, r10, lsl #17
+    mov     r2, r2, lsr #17
+    eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
+
+    ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
+
+    mvn     r7, #0
+    mov     r4, r4, lsr #16-2  @ (dx&7)*4
+    tst     r9, #0x0800
+    moveq   r7, r7, lsl r4     @ mask = ~0 [shift] (dx&7)*4
+    movne   r7, r7, lsr r4
+    mov     r7, r7, ror #16
+    and     r2, r2, r7         @ pack&mask
+
+    bic     r10,r10, #7<<16
+    b       .DrawStrip_samecode @ one last time, with last tile now masked
+
+.DrawStrip_noscroll:
     tst     r10, #1<<21 @ seen non hi-prio tile
     ldr     r1, [sp, #9*4]  @ est
     mov     r0, #0
@@ -939,15 +998,17 @@ DrawTilesFromCache:
     b       .dtfc_loop
 
 .dtfc_cut_tile:
-    add     r4, r4, #7      @ 0-6
+    cmn     r4, #8
+    ble     .dtfc_loop      @ off limits
+
+    rsb     r4, r4, #0      @ 1-7
     mov     r4, r4, lsl #2
-    mov     r12,#0xf<<28
-    mov     r12,r12,asr r4
-    mov     r2, r2, ror #16
+    mvn     r12,#0
     tst     r6, #0x0800     @ flipped?
-    mvnne   r12,r12
+    moveq   r12,r12, lsl r4
+    movne   r12,r12, lsr r4
+    mov     r12,r12, ror #16
     and     r2, r2, r12
-    mov     r2, r2, ror #16
     mov     r12,#0xf
     tst     r8, #1
     bne     .dtfc_shadow
-- 
2.39.2


From 11a1966bf349ab74035df4673293bdeab41b1a96 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Thu, 18 Jul 2024 21:36:43 +0200
Subject: [PATCH 14/16] core vdp, arm rendering speed optimisation

---
 pico/draw.c     | 112 ++++++++++++++++++++++++------------------------
 pico/draw_arm.S |  69 +++++++++++++----------------
 2 files changed, 85 insertions(+), 96 deletions(-)

diff --git a/pico/draw.c b/pico/draw.c
index 0f9b137a..387f3a42 100644
--- a/pico/draw.c
+++ b/pico/draw.c
@@ -304,34 +304,34 @@ TileFlipMakerAS(TileFlipSH_AS_and, pix_sh_as_and)
 // --------------------------------------------
 
 #ifndef _ASM_DRAW_C
-#define DrawTile(mask) { \
-  if (code!=oldcode) { \
-    oldcode = code; \
- \
-    pack = 0; \
-    if (code != blank) { \
-      /* Get tile address/2: */\
-      u32 addr = ((code&0x7ff)<<4) + ty; \
-      if (code & 0x1000) addr ^= 0xe; /* Y-flip */ \
- \
-      pal = ((code>>9)&0x30) | sh; /* shadow */ \
- \
-      pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr)); \
-      if (!pack) \
-        blank = code; \
-    } \
-  } \
- \
-  if (code & 0x8000) { /* (un-forced) high priority tile */ \
-    if (sh | (pack&mask)) { \
-      code |= (dx<<16) | (ty<<25); \
-      if (code & 0x1000) code ^= 0xe<<25; \
-      *hc++ = code, *hc++ = pack&mask; /* cache it */ \
-    } \
-  } else if (pack&mask) { \
-    if (code & 0x0800) TileFlip(pd + dx, pack&mask, pal); \
-    else               TileNorm(pd + dx, pack&mask, pal); \
-  } \
+#define DrawTile(mask) {						\
+  if (code!=oldcode) {							\
+    oldcode = code;							\
+									\
+    pack = 0;								\
+    if (code != blank) {						\
+      /* Get tile address/2: */						\
+      u32 addr = ((code&0x7ff)<<4) + ty;				\
+      if (code & 0x1000) addr ^= 0xe; /* Y-flip */			\
+									\
+      pal = ((code>>9)&0x30) | sh; /* shadow */				\
+									\
+      pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr));			\
+      if (!pack)							\
+        blank = code;							\
+    }									\
+  }									\
+									\
+  if (code & 0x8000) { /* (un-forced) high priority tile */		\
+    if (sh | (pack&mask)) {						\
+      code |= (dx<<16) | (ty<<25);					\
+      if (code & 0x1000) code ^= 0xe<<25;				\
+      *hc++ = code, *hc++ = pack&mask; /* cache it */			\
+    }									\
+  } else if (pack&mask) {						\
+    if (code & 0x0800) TileFlip(pd + dx, pack&mask, pal);		\
+    else               TileNorm(pd + dx, pack&mask, pal);		\
+  }									\
 }
 
 static void DrawStrip(struct TileStrip *ts, int lflags, int cellskip)
@@ -478,34 +478,34 @@ static void DrawStripVSRam(struct TileStrip *ts, int plane_sh, int cellskip)
 }
 #endif
 
-#define DrawTileInterlace(mask) { \
-  if (code!=oldcode) { \
-    oldcode = code; \
- \
-    pack = 0; \
-    if (code != blank) { \
-      /* Get tile address/2: */ \
-      u32 addr = ((code&0x3ff)<<5) + ty; \
-      if (code & 0x1000) addr ^= 0x1e; /* Y-flip */ \
- \
-      pal = ((code>>9)&0x30) | sh; /* shadow */ \
- \
-      pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr)); \
-      if (!pack) \
-        blank = code; \
-    } \
-  } \
- \
-  if (code & 0x8000) { /* high priority tile */ \
-    if (sh | (pack&mask)) { \
-      code = (code&0xfc00) | ((code&0x3ff)<<1) | (dx<<16) | (ty<<25); \
-      if (code & 0x1000) code ^= 0x1e<<25; \
-      *hc++ = code, *hc++ = pack&mask; /* cache it */ \
-    } \
-  } else if (pack&mask) { \
-    if (code & 0x0800) TileFlip(pd + dx, pack&mask, pal); \
-    else               TileNorm(pd + dx, pack&mask, pal); \
-  } \
+#define DrawTileInterlace(mask) {					\
+  if (code!=oldcode) {							\
+    oldcode = code;							\
+									\
+    pack = 0;								\
+    if (code != blank) {						\
+      /* Get tile address/2: */						\
+      u32 addr = ((code&0x3ff)<<5) + ty;				\
+      if (code & 0x1000) addr ^= 0x1e; /* Y-flip */			\
+									\
+      pal = ((code>>9)&0x30) | sh; /* shadow */				\
+									\
+      pack = CPU_LE2(*(u32 *)(PicoMem.vram + addr));			\
+      if (!pack)							\
+        blank = code;							\
+    }									\
+  }									\
+									\
+  if (code & 0x8000) { /* high priority tile */				\
+    if (sh | (pack&mask)) {						\
+      code = (code&0xfc00) | ((code&0x3ff)<<1) | (dx<<16) | (ty<<25);	\
+      if (code & 0x1000) code ^= 0x1e<<25;				\
+      *hc++ = code, *hc++ = pack&mask; /* cache it */			\
+    }									\
+  } else if (pack&mask) {						\
+    if (code & 0x0800) TileFlip(pd + dx, pack&mask, pal);		\
+    else               TileNorm(pd + dx, pack&mask, pal);		\
+  }									\
 }
 
 #ifndef _ASM_DRAW_C
diff --git a/pico/draw_arm.S b/pico/draw_arm.S
index 7a2e6f1d..40d5c443 100644
--- a/pico/draw_arm.S
+++ b/pico/draw_arm.S
@@ -433,7 +433,7 @@ DrawLayer:
     movs    r3, r9, lsl #1  @ (force[31]|sh[30]) << 1
     mov     r3, #0
     orrmi   r10,r10, #1<<23 @ r10=cells[31:24]|sh[23]|hi_not_empty[22]
-@    orrcc   r10,r10, #1<<20 @   |had_output[21]|!force[20]|hscroll[19:17]|ty[15:0]
+@    orrcc   r10,r10, #1<<20 @   |had_output[21]|!force[20]|hscroll[18:16]|ty[15:0]
     movmi   r3, #0x80       @ default to shadowed pal on sh mode
 
     and     r4, r7, #7
@@ -452,7 +452,7 @@ DrawLayer:
     mvn     r9, #0               @ r9=prevcode=-1
     add     r1, r11, r7          @ r1=pdest
 
-    @ r10=cells[31:24]|sh[23]|hi_not_empty[22]|had_output[21]|!force[20]|hscroll[19:17]|ty[15:0]
+    @ r10=cells[31:24]|sh[23]|hi_not_empty[22]|had_output[21]|!force[20]|hscroll[18:16]|ty[15:0]
     @ r1=pd+dx r2=pack r3=pal r5=xmask r6=hc r8=tilex r9=prevcode r11=HighCol r12=nametab lr=vram
     @ r4 & r7 are scratch in this loop
 
@@ -467,21 +467,19 @@ DrawLayer:
 
     add     r8, r8, #1
 
-    movs    r2, r9, lsl #20 @ if (code&0x1000)
-    mov     r2, r2, lsl #1
+    tst     r9, #0x1000     @ if (code&0x1000)
+    mov     r2, r9, lsl #21
     add     r2, r2, r10, lsl #17
-    mov     r2, r2, lsr #17
-    eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
+    eorne   r2, r2, #0xe<<17 @ if (code&0x1000) addr^=0xe;
 
-    ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
+    ldr     r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
 
     mvn     r7, #0
     mov     r4, r4, lsr #16-2  @ (dx&7)*4
     tst     r9, #0x0800
     moveq   r7, r7, lsl r4     @ mask = ~0 [shift] (dx&7)*4
     movne   r7, r7, lsr r4
-    mvn     r7, r7, ror #16
-    and     r2, r2, r7         @ pack&mask
+    bic     r2, r2, r7, ror #16 @ pack&~mask
 
     orr     r9, r9, #0x80000000 @ invalidate oldcode since pack is masked
     b       .DrawStrip_samecode
@@ -504,13 +502,12 @@ DrawLayer:
 
     mov     r9, r7          @ remember code
 
-    movs    r2, r9, lsl #20 @ if (code&0x1000)
-    mov     r2, r2, lsl #1
+    tst     r9, #0x1000     @ if (code&0x1000)
+    mov     r2, r9, lsl #21
     add     r2, r2, r10, lsl #17
-    mov     r2, r2, lsr #17
-    eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
+    eorne   r2, r2, #0x0e<<17 @ if (code&0x1000) addr^=0xe;
 
-    ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
+    ldr     r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
 .DrawStrip_samecode:
     tst     r9, #0x8000
 @    tstne   r10, #1<<20     @ !force[20]
@@ -577,21 +574,19 @@ DrawLayer:
 
     add     r1, r1, #8
 
-    movs    r2, r9, lsl #20 @ if (code&0x1000)
-    mov     r2, r2, lsl #1
+    tst     r9, #0x1000     @ if (code&0x1000)
+    mov     r2, r9, lsl #21
     add     r2, r2, r10, lsl #17
-    mov     r2, r2, lsr #17
-    eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
+    eorne   r2, r2, #0x0e<<17 @ if (code&0x1000) addr^=0xe;
 
-    ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
+    ldr     r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
 
     mvn     r7, #0
     mov     r4, r4, lsr #16-2  @ (dx&7)*4
     tst     r9, #0x0800
     moveq   r7, r7, lsl r4     @ mask = ~0 [shift] (dx&7)*4
     movne   r7, r7, lsr r4
-    mov     r7, r7, ror #16
-    and     r2, r2, r7         @ pack&mask
+    and     r2, r2, r7, ror #16 @ pack&mask
 
     bic     r10,r10, #7<<16
     b       .DrawStrip_samecode @ one last time, with last tile now masked
@@ -743,13 +738,12 @@ DrawLayer:
 
     mov     r9, r7          @ remember code
 
-    movs    r2, r9, lsl #20 @ if (code&0x1000)
-    mov     r2, r2, lsl #1
+    tst     r9, #0x1000     @ if (code&0x1000)
+    mov     r2, r9, lsl #21
     add     r2, r2, r10, lsl #17
-    mov     r2, r2, lsr #17
-    eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
+    eorne   r2, r2, #0x0e<<17 @ if (code&0x1000) addr^=0xe;
 
-    ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(PicoMem.vram+addr); // Get 8 pixels
+    ldr     r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
 
 .DrawStrip_vs_samecode:
     tst     r9, #0x8000
@@ -1007,8 +1001,7 @@ DrawTilesFromCache:
     tst     r6, #0x0800     @ flipped?
     moveq   r12,r12, lsl r4
     movne   r12,r12, lsr r4
-    mov     r12,r12, ror #16
-    and     r2, r2, r12
+    and     r2, r2, r12, ror #16
     mov     r12,#0xf
     tst     r8, #1
     bne     .dtfc_shadow
@@ -1152,9 +1145,7 @@ DrawSpriteSHi:
     cmp     r0, #328
     bge     DrawSpriteSHi
 
-    mov     r8, r8, lsl #17
-    mov     r8, r8, lsr #17    @ tile&=0x7fff; // Clip tile address
-
+    bic     r8, r8, #0xf8000   @ tile&=0x7fff; // Clip tile address
     ldr     r2, [lr, r8, lsl #1] @ pack=*(unsigned int *)(PicoMem.vram+addr); // Get 8 pixels
     add     r1, r11, r0        @ r1=pdest
     tst     r2, r2
@@ -1346,9 +1337,7 @@ DrawSprite:
     cmp     r0, #328
     bge     DrawSprite
 
-    mov     r8, r8, lsl #17
-    mov     r8, r8, lsr #17    @ tile&=0x7fff; // Clip tile address
-
+    bic     r8, r8, #0xf8000   @ tile&=0x7fff; // Clip tile address
     ldr     r2, [lr, r8, lsl #1] @ pack=*(unsigned int *)(PicoMem.vram+addr); // Get 8 pixels
     add     r1, r11, r0        @ r1=pdest
     tst     r2, r2
@@ -1492,16 +1481,16 @@ DrawWindow:
 
     mov     r9, r7          @ remember code
 
-    movs    r2, r9, lsl #20 @ if (code&0x1000)
-    mov     r2, r2, lsl #1
-    add     r2, r10, r2, lsr #17 @ r2=addr=(code&0x7ff)<<4; addr+=ty
-    eorcs   r2, r2, #0x0e   @ if (code&0x1000) addr^=0xe;
+    tst     r9, #0x1000     @ if (code&0x1000)
+    mov     r2, r9, lsl #21
+    add     r2, r2, r10, lsl #17
+    eorne   r2, r2, #0xe<<17 @ if (code&0x1000) addr^=0xe;
+
+    ldr     r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
 
     and     r3, r9, #0x6000
     mov     r3, r3, lsr #9  @ r3=pal=((code&0x6000)>>9);
 
-    ldr     r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
-
 .dw_samecode:
     tst     r6, #0x100
     bne     .dw_shadow
-- 
2.39.2


From ecc3f2ecc052238d66d020839845c0e5c52e7fd2 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Fri, 19 Jul 2024 19:21:26 +0200
Subject: [PATCH 15/16] release 2.00

---
 platform/common/version.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/platform/common/version.h b/platform/common/version.h
index 1ab6c650..230f315a 100644
--- a/platform/common/version.h
+++ b/platform/common/version.h
@@ -1 +1 @@
-#define VERSION "pre-2.00-beta3" REVISION
+#define VERSION "2.00" REVISION
-- 
2.39.2


From 75db61b756baeb95547bb80775e07796de204810 Mon Sep 17 00:00:00 2001
From: kub <derkub@gmail.com>
Date: Sun, 21 Jul 2024 16:33:07 +0200
Subject: [PATCH 16/16] platform ps2, handle audio similar to psp

---
 platform/ps2/emu.c | 109 ++++++++++++++++++++++++++++++++++-----------
 platform/psp/emu.c |   2 +-
 2 files changed, 85 insertions(+), 26 deletions(-)

diff --git a/platform/ps2/emu.c b/platform/ps2/emu.c
index 32c3d6c4..09e13d25 100644
--- a/platform/ps2/emu.c
+++ b/platform/ps2/emu.c
@@ -87,9 +87,10 @@ static uint8_t vsync; /* 0 (Disabled), 1 (Enabled), 2 (Dynamic) */
 #define SOUND_BLOCK_COUNT    7
 #define SOUND_BUFFER_CHUNK   (2*54000/50) // max.rate/min.frames in stereo
 
+static short sndBuffer_emu[SOUND_BUFFER_CHUNK+4]; // 4 for sample rounding overhang
 static short __attribute__((aligned(4))) sndBuffer[SOUND_BUFFER_CHUNK*SOUND_BLOCK_COUNT];
 static short __attribute__((aligned(4))) nulBuffer[SOUND_BUFFER_CHUNK];
-static short *snd_playptr, *sndBuffer_endptr;
+static short *snd_playptr, *sndBuffer_ptr, *sndBuffer_endptr;
 static int samples_made, samples_done, samples_block;
 
 static int sound_thread_exit = 0, sound_stopped = 1;
@@ -122,24 +123,78 @@ static void writeSound(int len)
 
 	if (samples_made - samples_done < samples_block * (SOUND_BLOCK_COUNT-2) - 4) {
 		samples_made += len / 2;
-		PicoIn.sndOut += len / 2;
+		sndBuffer_ptr += len / 2;
+		l = sndBuffer_ptr - sndBuffer;
+
+		if (sndBuffer_ptr > sndBuffer_endptr)
+			sndBuffer_endptr = sndBuffer_ptr;
+		if (l > sizeof(sndBuffer)/2)
+			lprintf("snd ovrn %d %d\n", len, sndBuffer_ptr - sndBuffer);
+		if (l > samples_block * (SOUND_BLOCK_COUNT-2)) {
+			sndBuffer_endptr = sndBuffer_ptr;
+			sndBuffer_ptr = sndBuffer;
+		}
 	} else
 		lprintf("ovfl %d\n", samples_made - samples_done);
-	if (sndBuffer_endptr < PicoIn.sndOut)
-		sndBuffer_endptr = PicoIn.sndOut;
-
-	l = PicoIn.sndOut - sndBuffer;
-	if (l > sizeof(sndBuffer)/2)
-		lprintf("ovrn %d %d\n", len, PicoIn.sndOut - sndBuffer);
-	if (l > samples_block * (SOUND_BLOCK_COUNT-2)) {
-		sndBuffer_endptr = PicoIn.sndOut;
-		PicoIn.sndOut = sndBuffer;
-	}
 
 	// signal the snd thread
 	SignalSema(sound_sem);
 }
 
+static void writeSound_44100(int len)
+{
+	writeSound(len);
+	PicoIn.sndOut = sndBuffer_ptr;
+}
+
+static void writeSound_22050_stereo(int len)
+{
+	short *p = sndBuffer_ptr;
+	int i;
+
+	for (i = 0; i < len / 2; i+=2, p+=4) {
+		p[0] = p[2] = PicoIn.sndOut[i];
+		p[1] = p[3] = PicoIn.sndOut[i+1];
+	}
+	writeSound(2*len);
+}
+
+static void writeSound_22050_mono(int len)
+{
+	short *p = sndBuffer_ptr;
+	int i;
+
+	for (i = 0; i < len / 2; i++, p+=2) {
+		p[0] = p[1] = PicoIn.sndOut[i];
+	}
+	writeSound(2*len);
+}
+
+static void writeSound_11025_stereo(int len)
+{
+	short *p = sndBuffer_ptr;
+	int i;
+
+	for (i = 0; i < len / 2; i+=2, p+=8) {
+		p[0] = p[2] = p[4] = p[6] = PicoIn.sndOut[i];
+		p[1] = p[3] = p[5] = p[7] = PicoIn.sndOut[i+1];
+	}
+	writeSound(4*len);
+}
+
+static void writeSound_11025_mono(int len)
+{
+	short *p = sndBuffer_ptr;
+	int i;
+
+	for (i = 0; i < len / 2; i++, p+=4) {
+		p[0] = p[1] = p[2] = p[3] = PicoIn.sndOut[i];
+	}
+	writeSound(4*len);
+}
+
+#define PS2_RATE 44100 // PicoIn.sndRate
+
 static void resetSound()
 {
 	struct audsrv_fmt_t format;
@@ -147,7 +202,7 @@ static void resetSound()
 	int ret;
 
 	format.bits = 16;
-	format.freq = PicoIn.sndRate;
+	format.freq = PS2_RATE;
 	format.channels = stereo ? 2 : 1;
 	ret = audsrv_set_format(&format);
 	if (ret < 0) {
@@ -170,14 +225,10 @@ static int sound_thread(void *argp)
 		// if there aren't enough samples, queue silence
 		int queued = audsrv_queued()/2;
 		while (queued < 2*samples_block) {
-			short *sndOut = PicoIn.sndOut, *sndEnd = sndBuffer_endptr;
-
 			// compute sample chunk size
-			int buflen = sndEnd - snd_playptr;
-			if (sndOut >= snd_playptr)
-				buflen = sndOut - snd_playptr;
-			if (buflen > samples_made - samples_done)
-				buflen = samples_made - samples_done;
+			int buflen = samples_made - samples_done;
+			if (buflen > sndBuffer_endptr - snd_playptr)
+				buflen = sndBuffer_endptr - snd_playptr;
 			if (buflen > 3*samples_block - queued)
 				buflen = 3*samples_block - queued;
 
@@ -241,7 +292,7 @@ static void sound_init(void)
 	thread.initial_priority = 40;
 	thid = CreateThread(&thread);
 
-	samples_block = 22050/50; // needs to be initialized before thread start
+	samples_block = PS2_RATE/60; // needs to be initialized before thread start
 	if (thid >= 0) {
 		ret = StartThread(thid, NULL);
 		if (ret < 0) lprintf("sound_init: StartThread returned %08x\n", ret);
@@ -255,7 +306,7 @@ static void sound_init(void)
 void pemu_sound_start(void) {
 	static int PsndRate_old = 0, PicoOpt_old = 0, pal_old = 0;
 	static int mp3_init_done;
-	int ret, stereo;
+	int ret, stereo, factor;
 
 	samples_made = samples_done = 0;
 
@@ -279,14 +330,22 @@ void pemu_sound_start(void) {
 		PsndRerate(Pico.m.frame_count ? 1 : 0);
 	}
 	stereo = (PicoIn.opt&8)>>3;
-	samples_block = (PicoIn.sndRate / (Pico.m.pal ? 50 : 60)) * (stereo ? 2 : 1);
+
+	factor = PS2_RATE / PicoIn.sndRate;
+	samples_block = (PS2_RATE / (Pico.m.pal ? 50 : 60)) * (stereo ? 2 : 1);
 
 	lprintf("starting audio: %i, len: %i, stereo: %i, pal: %i, block samples: %i\n",
 			PicoIn.sndRate, Pico.snd.len, stereo, Pico.m.pal, samples_block);
 
 	resetSound();
-	PicoIn.writeSound = writeSound;
-	snd_playptr = PicoIn.sndOut = sndBuffer_endptr = sndBuffer;
+	switch (factor) {
+	case 1: PicoIn.writeSound = stereo ? writeSound_44100       :writeSound_44100     ; break;
+	case 2: PicoIn.writeSound = stereo ? writeSound_22050_stereo:writeSound_22050_mono; break;
+	case 4: PicoIn.writeSound = stereo ? writeSound_11025_stereo:writeSound_11025_mono; break;
+	}
+	sndBuffer_endptr = sndBuffer;
+	snd_playptr = sndBuffer_ptr = sndBuffer;
+	PicoIn.sndOut = (factor == 1 ? sndBuffer_ptr : sndBuffer_emu);
 
 	PsndRate_old = PicoIn.sndRate;
 	PicoOpt_old  = PicoIn.opt;
diff --git a/platform/psp/emu.c b/platform/psp/emu.c
index a3e22571..41deba8d 100644
--- a/platform/psp/emu.c
+++ b/platform/psp/emu.c
@@ -450,7 +450,7 @@ static void writeSound(int len)
 	if (samples_made - samples_done < samples_block * (SOUND_BLOCK_COUNT-2) - 4) {
 		sndBuffer_ptr += len / 2;
 		if (sndBuffer_ptr - sndBuffer > sizeof(sndBuffer)/2)
-			lprintf("snd ovrn %d %d\n", len, PicoIn.sndOut - sndBuffer);
+			lprintf("snd ovrn %d %d\n", len, sndBuffer_ptr - sndBuffer);
 		if (sndBuffer_ptr >= sndBuffer_endptr) {
 			int wrap = sndBuffer_ptr - sndBuffer_endptr;
 			if (wrap > 0)
-- 
2.39.2