@ assembly "optimized" version of some funtions from draw.c\r
@ this is highly specialized, be careful if changing related C code!\r
\r
-@ (c) Copyright 2006, notaz\r
+@ (c) Copyright 2007, Grazvydas "notaz" Ignotas\r
@ All Rights Reserved\r
\r
+.include "port_config.s"\r
\r
.extern Pico\r
.extern PicoOpt\r
.extern HighSprZ\r
.extern rendstatus\r
.extern DrawLineDest\r
-.extern DrawStripVSRam\r
.extern DrawStripInterlace\r
\r
\r
streqb r4, [r1,#\offs]\r
.endm\r
\r
-@ TileNorm (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: register with helper pattern 0xf, touches r3 high bits\r
+@ TileNormShHP (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: register with helper pattern 0xf, touches r3 high bits\r
.macro TileNormShHP\r
TilePixelShHP 12, 0 @ #0x0000f000\r
TilePixelShHP 8, 1 @ #0x00000f00\r
TilePixelShHP 16, 7 @ #0x000f0000\r
.endm\r
\r
-@ TileFlip (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf\r
+@ TileFlipShHP (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf\r
.macro TileFlipShHP\r
TilePixelShHP 16, 0 @ #0x000f0000\r
TilePixelShHP 20, 1 @ #0x00f00000\r
@ int cells; // 0x14\r
@ };\r
\r
-@ int DrawLayer(int plane, int *hcache, int maxcells, int sh)\r
+@ void DrawLayer(int plane_sh, int *hcache, int cellskip, int maxcells);\r
\r
-.global DrawLayer @ int plane, int *hcache, int maxcells, int sh\r
+.global DrawLayer\r
\r
DrawLayer:\r
stmfd sp!, {r4-r11,lr}\r
ldr r11, =(Pico+0x22228) @ Pico.video\r
mov r8, #1\r
\r
- ldrb r7, [r11, #16] @ ??hh??ww\r
+ ldrb r7, [r11, #16] @ ??vv??hh\r
\r
mov r6, r1 @ hcache\r
- orr r9, r2, r3, lsl #31 @ r9=maxcells|(sh<<31)\r
+ orr r9, r3, r0, lsl #30\r
+ orr r9, r9, r2, lsl #8 @ r9=sh[31]|cellskip[15:8]|maxcells[7:0] (tmp)\r
\r
mov r1, r7, lsl #4\r
orr r1, r1, #0x00ff\r
sub r5, r5, #1 @ r5=xmask\r
\r
@ Find name table:\r
- tst r0, r0\r
+ ands r0, r0, #1\r
ldreqb r12, [r11, #2]\r
ldrneb r12, [r11, #4]\r
\r
add r12, r12, r4, lsl r10 @ nametab+=(ts.line>>3)<<shift[width];\r
\r
@ ldmia r0, {r1,r2,r3,r5,r6,r9} @ r2=line, r3=ts->hscroll, r5=ts->xmask, r6=ts->hc, r9=ts->cells\r
-@ mov r12,r1, lsl #1 @ r12=(ts->nametab<<1) (halfword compliant)\r
\r
and r10,r2, #7\r
mov r10,r10, lsl #1 @ r10=ty=(ts->line&7)<<1;\r
\r
tst r9, #1<<31\r
mov r3, #0\r
- orrne r10,r10, #1<<23 @ r10=(cells<<24|sh<<23|hi_not_empty<<22|ty)\r
+ orrne r10,r10, #1<<23 @ r10=(cells<<24|sh<<23|hi_not_empty<<22|had_output<<21|ty)\r
movne r3, #0x40 @ default to shadowed pal on sh mode\r
\r
- mvn r9, #0 @ r9=prevcode=-1\r
-\r
cmp r7, #8\r
addne r10,r10, #0x01000000 @ we will loop cells+1 times if there is scroll\r
\r
+ and r9, r9, #0xff00\r
+ add r8, r8, r9, lsr #8 @ tilex+=cellskip\r
+ add r7, r7, r9, lsr #5 @ dx+=cellskip<<3;\r
+ sub r10,r10,r9, lsl #16 @ cells-=cellskip\r
+\r
@ cache some stuff to avoid mem access\r
+.if OVERRIDE_HIGHCOL\r
+ ldr r11,=HighCol\r
+ mov r0, #0xf\r
+ ldr r11,[r11]\r
+.else\r
ldr r11,=HighCol\r
mov r0, #0xf\r
+.endif\r
+\r
+ mvn r9, #0 @ r9=prevcode=-1\r
add r1, r11, r7 @ r1=pdest\r
\r
\r
beq .DrawStrip_samecode @ we know stuff about this tile already\r
\r
mov r9, r7 @ remember code\r
+ orr r10, r10, #1<<21 @ seen non hi-prio tile\r
\r
movs r2, r9, lsl #20 @ if (code&0x1000)\r
mov r2, r2, lsl #1\r
beq .DrawStrip_SingleColor @ tileline singlecolor \r
\r
tst r9, #0x0800\r
- beq .DrawStrip_TileNorm\r
+ bne .DrawStrip_TileFlip\r
\r
@ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern\r
- TileFlip r0\r
- b .dsloop\r
-\r
.DrawStrip_TileNorm:\r
TileNorm r0\r
b .dsloop\r
\r
+.DrawStrip_TileFlip:\r
+ TileFlip r0\r
+ b .dsloop\r
+\r
.DrawStrip_SingleColor:\r
and r4, r2, #0xf\r
orr r4, r3, r4\r
strneb r4, [r1], #1 @ have a remaining unaligned pixel?\r
b .dsloop_subr1\r
\r
+.DrawStrip_hiprio_maybempt:\r
+ cmp r7, r9\r
+ beq .dsloop @ must've been empty, otherwise we wouldn't get here\r
+ movs r2, r7, lsl #20 @ if (code&0x1000)\r
+ mov r2, r2, lsl #1\r
+ add r2, r2, r10, lsl #17\r
+ mov r2, r2, lsr #17\r
+ eorcs r2, r2, #0x0e @ if (code&0x1000) addr^=0xe;\r
+ ldr r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels\r
+ mov r9, r7 @ remember code\r
+ tst r2, r2\r
+ beq .dsloop\r
+ orr r10, r10, #1<<22\r
+\r
.DrawStrip_hiprio:\r
tst r10, #0x00c00000\r
beq .DrawStrip_hiprio_maybempt\r
mov r0, #0xf\r
b .dsloop\r
\r
-.DrawStrip_hiprio_maybempt:\r
+.dsloop_exit:\r
+ tst r10, #1<<21 @ seen non hi-prio tile\r
+ ldreq r1, =rendstatus\r
+ mov r0, #0\r
+ ldreq r2, [r1]\r
+ str r0, [r6] @ terminate the cache list\r
+ orreq r2, r2, #0x40 @ had a layer with all hi-prio tiles\r
+ streq r2, [r1]\r
+\r
+ ldmfd sp!, {r4-r11,lr}\r
+ bx lr\r
+\r
+@ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@\r
+\r
+.DrawStrip_vsscroll:\r
+ rsb r8, r3, #0\r
+ mov r8, r8, lsr #3 @ r8=tilex=(-ts->hscroll)>>3\r
+ bic r8, r8, #0xff000000\r
+ orr r8, r8, r5, lsl #25 @ r8=(xmask[31:25]|had_output[24]|tilex[23:0])\r
+\r
+ ldr r4, =Scanline\r
+ orr r5, r1, r10, lsl #24\r
+ ldr r4, [r4]\r
+ sub r1, r3, #1\r
+ orr r5, r5, r4, lsl #16 @ r5=(shift_width[31:24]|scanline[23:16]|ymask[15:0])\r
+ and r1, r1, #7\r
+ add r7, r1, #1 @ r7=dx=((ts->hscroll-1)&7)+1\r
+\r
+ mov r10,r9, lsl #16\r
+ tst r0, #1\r
+ orrne r10,r10, #0x8000\r
+ tst r9, #1<<31\r
+ mov r3, #0\r
+ orr r10,r10, #0xff000000 @ will be adjusted on entering loop\r
+ orrne r10,r10, #1<<23 @ r10=(cell[31:24]|sh[23]|hi_not_empty[22]|cells_max[21:16]|plane[15]|ty[14:0])\r
+ movne r3, #0x40 @ default to shadowed pal on sh mode\r
+\r
+ cmp r7, #8\r
+ subne r10,r10, #0x01000000 @ have hscroll, start with negative cell\r
+\r
+ and r9, r9, #0xff00\r
+ add r8, r8, r9, lsr #8 @ tilex+=cellskip\r
+ add r7, r7, r9, lsr #5 @ dx+=cellskip<<3;\r
+ add r10,r10,r9, lsl #16 @ cell+=cellskip\r
+\r
+ @ cache some stuff to avoid mem access\r
+.if OVERRIDE_HIGHCOL\r
+ ldr r11,=HighCol\r
+ mov r0, #0xf\r
+ ldr r11,[r11]\r
+.else\r
+ ldr r11,=HighCol\r
+ mov r0, #0xf\r
+.endif\r
+\r
+ mvn r9, #0 @ r9=prevcode=-1\r
+ add r1, r11, r7 @ r1=pdest\r
+\r
+ @ r4 & r7 are scratch in this loop\r
+.dsloop_vs_subr1:\r
+ sub r1, r1, #8\r
+.dsloop_vs: @ 40-41 times\r
+ add r10,r10, #0x01000000\r
+ and r4, r10, #0x003f0000\r
+ cmp r4, r10, asr #8\r
+ ble .dsloop_vs_exit\r
+\r
+ @ calc offset and read tileline code to r7, also calc ty\r
+ add r7, lr, #0x012000\r
+ add r7, r7, #0x000180 @ r7=Pico.vsram (Pico+0x22180)\r
+ add r7, r7, r10,asr #23 @ vsram + ((cell&~1)<<1)\r
+ bic r7, r7, #3\r
+ tst r10,#0x8000 @ plane1?\r
+ addne r7, r7, #2\r
+ ldrh r7, [r7] @ r7=vscroll\r
+\r
+ bic r10,r10,#0xff @ clear old ty\r
+ and r4, r5, #0xff0000\r
+ add r4, r4, r7, lsl #16\r
+ and r4, r4, r5, lsl #16 @ r4=line<<16\r
+ and r7, r4, #0x70000\r
+ orr r10,r10,r7, lsr #15 @ new ty\r
+\r
+ mov r4, r4, lsr #19\r
+ mov r7, r5, lsr #24\r
+ mov r4, r4, lsl r7 @ nametabadd\r
+\r
+ and r7, r8, r8, lsr #25\r
+ add r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords)\r
+ add r7, r7, r4, lsl #1\r
+ ldrh r7, [r7, r12] @ r7=code (int, but from unsigned, no sign extend)\r
+\r
+ add r1, r1, #8\r
+ add r8, r8, #1\r
+\r
+ tst r7, #0x8000\r
+ bne .DrawStrip_vs_hiprio\r
+\r
cmp r7, r9\r
- beq .dsloop @ must've been empty, otherwise we wouldn't get here\r
+ beq .DrawStrip_vs_samecode @ we know stuff about this tile already\r
+\r
+ mov r9, r7 @ remember code\r
+ orr r8, r8, #1<<24 @ seen non hi-prio tile\r
+\r
+ movs r2, r9, lsl #20 @ if (code&0x1000)\r
+ mov r2, r2, lsl #1\r
+ add r2, r2, r10, lsl #17\r
+ mov r2, r2, lsr #17\r
+ eorcs r2, r2, #0x0e @ if (code&0x1000) addr^=0xe;\r
+\r
+ ldr r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels\r
+\r
+ bic r7, r3, #0x3f\r
+ and r3, r9, #0x6000\r
+ add r3, r7, r3, lsr #9 @ r3=pal=((code&0x6000)>>9);\r
+\r
+.DrawStrip_vs_samecode:\r
+ tst r2, r2\r
+ beq .dsloop_vs @ tileline blank\r
+\r
+ cmp r2, r2, ror #4\r
+ beq .DrawStrip_vs_SingleColor @ tileline singlecolor \r
+\r
+ tst r9, #0x0800\r
+ bne .DrawStrip_vs_TileFlip\r
+\r
+ @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern\r
+.DrawStrip_vs_TileNorm:\r
+ TileNorm r0\r
+ b .dsloop_vs\r
+\r
+.DrawStrip_vs_TileFlip:\r
+ TileFlip r0\r
+ b .dsloop_vs\r
+\r
+.DrawStrip_vs_SingleColor:\r
+ and r4, r2, #0xf\r
+ orr r4, r3, r4\r
+ orr r4, r4, r4, lsl #8\r
+ tst r1, #1 @ not aligned?\r
+ strneb r4, [r1], #1\r
+ streqh r4, [r1], #2\r
+ strh r4, [r1], #2\r
+ strh r4, [r1], #2\r
+ strh r4, [r1], #2\r
+ strneb r4, [r1], #1 @ have a remaining unaligned pixel?\r
+ b .dsloop_vs_subr1\r
+\r
+.DrawStrip_vs_hiprio:\r
+ tst r10, #0x00c00000\r
+ beq .DrawStrip_vs_hiprio_maybempt\r
+ sub r0, r1, r11\r
+ orr r7, r7, r0, lsl #16\r
+ orr r7, r7, r10, lsl #25 @ (ty<<25)\r
+ tst r7, #0x1000\r
+ eorne r7, r7, #7<<26 @ if(code&0x1000) cval^=7<<26;\r
+ str r7, [r6], #4 @ cache hi priority tile\r
+ mov r0, #0xf\r
+ b .dsloop_vs\r
+\r
+.DrawStrip_vs_hiprio_maybempt:\r
+ cmp r7, r9\r
+ beq .dsloop_vs @ must've been empty, otherwise we wouldn't get here\r
movs r2, r7, lsl #20 @ if (code&0x1000)\r
mov r2, r2, lsl #1\r
add r2, r2, r10, lsl #17\r
mov r9, r7 @ remember code\r
tst r2, r2\r
orrne r10, r10, #1<<22\r
- bne .DrawStrip_hiprio\r
- b .dsloop\r
+ bne .DrawStrip_vs_hiprio\r
+ b .dsloop_vs\r
\r
-.dsloop_exit:\r
+.dsloop_vs_exit:\r
+ tst r8, #1<<24 @ seen non hi-prio tile\r
+ ldreq r1, =rendstatus\r
mov r0, #0\r
+ ldreq r2, [r1]\r
str r0, [r6] @ terminate the cache list\r
+ orreq r2, r2, #0x40 @ had a layer with all hi-prio tiles\r
+ streq r2, [r1]\r
\r
ldmfd sp!, {r4-r11,lr}\r
bx lr\r
\r
\r
-.DrawStrip_vsscroll:\r
- @ shit, we have 2-cell column based vscroll\r
- @ let the c code handle this (for now)\r
-\r
- @ int nametab; // 0x00\r
- @ int line; // 0x04\r
- @ int hscroll; // 0x08\r
- @ int xmask; // 0x0C\r
- @ int *hc; // 0x10 (pointer to cache buffer)\r
- @ int cells; // 0x14\r
-\r
- sub sp, sp, #6*4\r
- orr r2, r1, r10, lsl #24 @ ts.line=ymask|(shift[width]<<24); // save some stuff instead of line\r
- mov r1, r0 @ plane\r
- mov r0, r12, lsr #1 @ halfwords\r
- and r9, r9, #0xff\r
- stmia sp, {r0,r2,r3,r5,r6,r9}\r
-\r
- mov r0, sp\r
- bl DrawStripVSRam @ struct TileStrip *ts, int plane\r
-\r
- add sp, sp, #6*4\r
- ldmfd sp!, {r4-r11,lr}\r
- bx lr\r
+@ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@\r
\r
@ interlace mode 2? Sonic 2?\r
.DrawStrip_interlace:\r
BackFill:\r
stmfd sp!, {r4-r9,lr}\r
\r
+.if OVERRIDE_HIGHCOL\r
+ ldr lr, =HighCol\r
+ mov r0, r0, lsl #26\r
+ ldr lr, [lr]\r
+ mov r0, r0, lsr #26\r
+ add lr, lr, #8\r
+.else\r
ldr lr, =(HighCol+8)\r
-\r
mov r0, r0, lsl #26\r
mov r0, r0, lsr #26\r
+.endif\r
+\r
orr r0, r0, r1, lsl #6\r
orr r0, r0, r0, lsl #8\r
orr r0, r0, r0, lsl #16\r
@ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@\r
\r
\r
-.global DrawTilesFromCache @ int *hc, int sh\r
+.global DrawTilesFromCache @ int *hc, int sh, int rlim\r
\r
DrawTilesFromCache:\r
stmfd sp!, {r4-r8,r11,lr}\r
\r
- mvn r5, #0 @ r5=prevcode=-1\r
- mov r8, r1\r
-\r
@ cache some stuff to avoid mem access\r
+.if OVERRIDE_HIGHCOL\r
ldr r11,=HighCol\r
- ldr lr, =(Pico+0x10000) @ lr=Pico.vram\r
mov r12,#0xf\r
+ ldr r11,[r11]\r
+.else\r
+ ldr r11,=HighCol\r
+ mov r12,#0xf\r
+.endif\r
+ ldr lr, =(Pico+0x10000) @ lr=Pico.vram\r
+\r
+ mvn r5, #0 @ r5=prevcode=-1\r
+ ands r8, r1, #1\r
+ orr r8, r8, r2, lsl #1\r
+ bne .dtfc_check_rendflags\r
\r
@ scratch: r4, r7\r
.dtfc_loop:\r
ldr r6, [r0], #4 @ read code\r
movs r1, r6, lsr #16 @ r1=dx;\r
ldmeqfd sp!, {r4-r8,r11,pc} @ dx is never zero, this must be a terminator, return\r
- bic r1, r1, #0xfe00\r
- add r1, r11, r1 @ r1=pdest\r
+ bic r4, r1, #0xfe00\r
+ add r1, r11, r4 @ r1=pdest\r
\r
mov r7, r6, lsl #16\r
cmp r5, r7, lsr #16\r
ldr r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels\r
\r
.dtfc_samecode:\r
- tst r8, r8\r
+ rsbs r4, r4, r8, lsr #1\r
+ bmi .dtfc_cut_tile\r
+\r
+ tst r8, #1\r
bne .dtfc_shadow\r
\r
tst r2, r2\r
beq .dtfc_SingleColor @ tileline singlecolor \r
\r
tst r5, #0x0800\r
- beq .dtfc_TileNorm\r
+ bne .dtfc_TileFlip\r
\r
@ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern\r
- TileFlip r12\r
- b .dtfc_loop\r
-\r
.dtfc_TileNorm:\r
TileNorm r12\r
b .dtfc_loop\r
\r
+.dtfc_TileFlip:\r
+ TileFlip r12\r
+ b .dtfc_loop\r
+\r
.dtfc_SingleColor:\r
and r4, r2, #0xf\r
orr r4, r3, r4\r
beq .dtfc_SingleColor @ tileline singlecolor \r
\r
tst r5, #0x0800\r
- beq .dtfc_TileNormShHP\r
+ bne .dtfc_TileFlipShHP\r
\r
@ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern\r
- TileFlipShHP\r
- b .dtfc_loop\r
-\r
.dtfc_TileNormShHP:\r
TileNormShHP\r
b .dtfc_loop\r
\r
+.dtfc_TileFlipShHP:\r
+ TileFlipShHP\r
+ b .dtfc_loop\r
+\r
.dtfc_shadow_blank:\r
ldrb r4, [r1] @ 1ci\r
ldrb r12,[r1,#1]\r
mov r12, #0xf\r
b .dtfc_loop\r
\r
+.dtfc_cut_tile:\r
+ add r4, r4, #7 @ 0-6\r
+ mov r4, r4, lsl #2\r
+ mov r12,#0xf<<28\r
+ mov r12,r12,asr r4\r
+ mov r2, r2, ror #16\r
+ tst r5, #0x0800 @ flipped?\r
+ mvnne r12,r12\r
+ and r2, r2, r12\r
+ mov r2, r2, ror #16\r
+ mov r12,#0xf\r
+ tst r8, #1\r
+ bne .dtfc_shadow\r
+ tst r2, r2\r
+ beq .dtfc_loop\r
+ tst r5, #0x0800\r
+ beq .dtfc_TileNorm\r
+ b .dtfc_TileFlip\r
+\r
+@ check if we have detected layer covered with hi-prio tiles:\r
+.dtfc_check_rendflags:\r
+ ldr r1, =rendstatus\r
+ ldr r2, [r1]\r
+ tst r2, #0xc0\r
+ beq .dtfc_loop\r
+ bic r8, r8, #1 @ sh/hi mode off\r
+ tst r2, #0x80\r
+ bne .dtfc_loop @ already processed\r
+ orr r2, r2, #0x80\r
+ str r2, [r1]\r
+\r
+ add r1, r11,#8\r
+ mov r3, #320/4\r
+ mov r7, #0x80\r
+ orr r7, r7, r7, lsl #8\r
+ orr r7, r7, r7, lsl #16\r
+ mov r6, #0x3f\r
+ orr r6, r6, r6, lsl #8\r
+ orr r6, r6, r6, lsl #16\r
+.dtfc_loop_shprep:\r
+ subs r3, r3, #1\r
+ bmi .dtfc_loop @ done\r
+ ldr r2, [r1]\r
+ tst r2, r7\r
+ andeq r2, r2, r6\r
+ streq r2, [r1], #4\r
+ beq .dtfc_loop_shprep\r
+ tst r2, #0x80000000\r
+ biceq r2, r2, #0xc0000000\r
+ tst r2, #0x00800000\r
+ biceq r2, r2, #0x00c00000\r
+ tst r2, #0x00008000\r
+ biceq r2, r2, #0x0000c000\r
+ tst r2, #0x00000080\r
+ biceq r2, r2, #0x000000c0\r
+ str r2, [r1], #4\r
+ b .dtfc_loop_shprep\r
+\r
.pool\r
\r
@ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@\r
stmfd sp!, {r4-r11,lr}\r
\r
@ cache some stuff to avoid mem access\r
+.if OVERRIDE_HIGHCOL\r
ldr r11,=HighCol\r
+ mov r12,#0xf\r
+ ldr r11,[r11]\r
+.else\r
+ ldr r11,=HighCol\r
+ mov r12,#0xf\r
+.endif\r
ldr lr, =(Pico+0x10000) @ lr=Pico.vram\r
mov r6, r1, lsl #31\r
orr r6, r6, #1<<30\r
- mov r12,#0xf\r
\r
mov r10, r0\r
\r
beq .dsfc_SingleColor @ tileline singlecolor \r
\r
tst r9, #0x10000\r
- beq .dsfc_TileNorm\r
+ bne .dsfc_TileFlip\r
\r
@ TileFlip (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern\r
- TileFlip r12\r
- b .dsfc_inloop\r
-\r
.dsfc_TileNorm:\r
TileNorm r12\r
b .dsfc_inloop\r
\r
+.dsfc_TileFlip:\r
+ TileFlip r12\r
+ b .dsfc_inloop\r
+\r
.dsfc_SingleColor:\r
tst r0, #1 @ not aligned?\r
and r4, r2, #0xf\r
beq .dsfc_singlec_sh\r
\r
tst r9, #0x10000\r
- beq .dsfc_TileNorm_sh\r
+ bne .dsfc_TileFlip_sh\r
\r
@ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern\r
- TileFlipSh\r
- b .dsfc_inloop\r
-\r
.dsfc_TileNorm_sh:\r
TileNormSh\r
b .dsfc_inloop\r
\r
+.dsfc_TileFlip_sh:\r
+ TileFlipSh\r
+ b .dsfc_inloop\r
+\r
.dsfc_singlec_sh:\r
cmp r2, #0xe0000000\r
bcc .dsfc_SingleColor @ normal singlecolor tileline (carry inverted in ARM)\r
bne .dspr_cache @ if(code&0x8000) // high priority - cache it\r
\r
@ cache some stuff to avoid mem access\r
+.if OVERRIDE_HIGHCOL\r
+ ldr r11,=HighCol\r
+ mov r12,#0xf\r
+ ldr r11,[r11]\r
+.else\r
ldr r11,=HighCol\r
- ldr lr, =(Pico+0x10000) @ lr=Pico.vram\r
mov r12,#0xf\r
+.endif\r
+ ldr lr, =(Pico+0x10000) @ lr=Pico.vram\r
\r
mov r5, r5, lsl #4 @ delta<<=4; // Delta of address\r
and r4, r9, #0x6000\r
beq .dspr_SingleColor @ tileline singlecolor \r
\r
tst r9, #0x0800\r
- beq .dspr_TileNorm\r
+ bne .dspr_TileFlip\r
\r
@ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern\r
- TileFlip r12\r
- b .dspr_loop\r
-\r
@ scratch: r4, r7\r
.dspr_TileNorm:\r
TileNorm r12\r
b .dspr_loop\r
\r
+.dspr_TileFlip:\r
+ TileFlip r12\r
+ b .dspr_loop\r
+\r
.dspr_SingleColor:\r
and r4, r2, #0xf\r
orr r4, r3, r4\r
beq .dspr_singlec_sh\r
\r
tst r9, #0x0800\r
- beq .dspr_TileNorm_sh\r
+ bne .dspr_TileFlip_sh\r
\r
@ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern\r
- TileFlipSh\r
- b .dspr_loop\r
-\r
.dspr_TileNorm_sh:\r
TileNormSh\r
b .dspr_loop\r
\r
+.dspr_TileFlip_sh:\r
+ TileFlipSh\r
+ b .dspr_loop\r
+\r
.dspr_singlec_sh:\r
cmp r2, #0xe0000000\r
bcc .dspr_SingleColor @ normal tileline\r
orr r6, r6, r3, lsl #8 @ shadow mode\r
\r
sub r8, r1, r0\r
- mov r8, r8, lsl #1 @ cells\r
-\r
- mvn r9, #0 @ r9=prevcode=-1\r
\r
@ cache some stuff to avoid mem access\r
+.if OVERRIDE_HIGHCOL\r
+ ldr r11,=HighCol\r
+ mov r8, r8, lsl #1 @ cells\r
+ ldr r11,[r11]\r
+ mvn r9, #0 @ r9=prevcode=-1\r
+ add r11,r11,#8\r
+.else\r
ldr r11,=(HighCol+8)\r
+ mov r8, r8, lsl #1 @ cells\r
+ mvn r9, #0 @ r9=prevcode=-1\r
+.endif\r
add r1, r11, r0, lsl #4 @ r1=pdest\r
mov r0, #0xf\r
b .dwloop_enter\r
beq .dw_SingleColor @ tileline singlecolor \r
\r
tst r9, #0x0800\r
- beq .dw_TileNorm\r
+ bne .dw_TileFlip\r
\r
@ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern\r
- TileFlip r0\r
- b .dwloop\r
-\r
.dw_TileNorm:\r
TileNorm r0\r
b .dwloop\r
\r
+.dw_TileFlip:\r
+ TileFlip r0\r
+ b .dwloop\r
+\r
.dw_SingleColor:\r
and r4, r0, r2 @ #0x0000000f\r
orr r4, r3, r4\r
\r
\r
.fl_noshBGR444:\r
+.if OVERRIDE_HIGHCOL\r
+ ldr r1, =HighCol\r
+ mov lr, #0xff\r
+ ldr r1, [r1]\r
+ mov lr, lr, lsl #1\r
+ add r1, r1, #8\r
+.else\r
ldr r1, =(HighCol+8)\r
mov lr, #0xff\r
mov lr, lr, lsl #1\r
+.endif\r
\r
.fl_loopBGR444:\r
\r
orr \reg, \reg, r3 @ add blue back\r
.endm\r
\r
+.global vidConvCpyRGB565\r
+\r
vidConvCpyRGB565: @ void *to, void *from, int pixels\r
stmfd sp!, {r4-r9,lr}\r
\r
- mov r12, r2, lsr #3 @ repeats\r
+ mov r12, r2, lsr #3 @ repeats\r
mov lr, #0x001c0000\r
orr lr, lr, #0x01c @ lr == pattern 0x001c001c\r
mov r8, #0x00030000\r
- orr r8, r8, #0x003 @ lr == pattern 0x001c001c\r
+ orr r8, r8, #0x003\r
\r
.loopRGB565:\r
ldmia r1!, {r4-r7}\r
sub r3, r3, #0x40*2\r
\r
.fl_noshRGB555:\r
+.if OVERRIDE_HIGHCOL\r
+ ldr r1, =HighCol\r
+ ldr r0, =DrawLineDest\r
+ ldr r1, [r1]\r
+ ldr r0, [r0]\r
+ add r1, r1, #8\r
+.else\r
ldr r0, =DrawLineDest\r
ldr r1, =(HighCol+8)\r
ldr r0, [r0]\r
+.endif\r
\r
ldrb r12, [r8, #12]\r
mov lr, #0xff\r
addeq r0, r0, #32*2\r
\r
.fl_no32colRGB555:\r
-.fl_loopRGB555:\r
\r
+.if UNALIGNED_DRAWLINEDEST\r
+ @ this is basically for Gizmondo, which has unaligned odd lines in the framebuffer\r
+ tst r0, #2\r
+ bne .fl_RGB555u\r
+.endif\r
+\r
+.fl_loopRGB555:\r
ldr r12, [r1], #4\r
ldr r7, [r1], #4\r
\r
ldrh r6, [r3, r6]\r
and r12,lr, r7, lsr #15\r
ldrh r12,[r3, r12]\r
+ and r7, lr, r7, lsr #23\r
+ ldrh r7, [r3, r7]\r
orr r8, r8, r6, lsl #16\r
\r
- and r6, lr, r7, lsr #23\r
- ldrh r6, [r3, r6] @ 1 cycle interlock here (r6)\r
subs r2, r2, #1\r
- orr r12,r12, r6, lsl #16\r
+ orr r12,r12, r7, lsl #16\r
\r
stmia r0!, {r4,r5,r8,r12}\r
bne .fl_loopRGB555\r
mov r9, #0x3900 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007\r
orr r9, r9, #0x00e7\r
\r
+.if UNALIGNED_DRAWLINEDEST\r
+ tst r0, #2\r
+ bne .fl_32scale_RGB555u\r
+.endif\r
+\r
.fl_loop32scale_RGB555:\r
ldr r12, [r1], #4\r
ldr r7, [r1], #4\r
ldmfd sp!, {r4-r8,lr}\r
bx lr\r
\r
+.if UNALIGNED_DRAWLINEDEST\r
+ @ unaligned versions of loops\r
+ @ warning: starts drawing 2bytes before dst\r
+\r
+.fl_RGB555u:\r
+ sub r0, r0, #2 @ initial adjustment\r
+ mov r8, #0\r
+\r
+.fl_loopRGB555u:\r
+ ldr r12, [r1], #4\r
+ ldr r7, [r1], #4\r
+\r
+ and r6, lr, r12,lsl #1\r
+ ldrh r6, [r3, r6]\r
+ and r5, lr, r12,lsr #7\r
+ ldrh r5, [r3, r5]\r
+ orr r4, r8, r6, lsl #16\r
+\r
+ and r6, lr, r12,lsr #15\r
+ ldrh r6, [r3, r6]\r
+ and r8, lr, r12,lsr #23\r
+ ldrh r8, [r3, r8]\r
+ orr r5, r5, r6, lsl #16\r
+\r
+ and r6, lr, r7, lsl #1\r
+ ldrh r6, [r3, r6]\r
+ and r12,lr, r7, lsr #7\r
+ ldrh r12,[r3, r12]\r
+ orr r6, r8, r6, lsl #16\r
+\r
+ and r8, lr, r7, lsr #15\r
+ ldrh r8, [r3, r8]\r
+ and r7, lr, r7, lsr #23\r
+\r
+ subs r2, r2, #1\r
+ orr r12,r12,r8, lsl #16\r
+ ldrh r8, [r3, r7]\r
+\r
+ stmia r0!, {r4,r5,r6,r12}\r
+ bne .fl_loopRGB555u\r
+\r
+ strh r8, [r0], #2\r
+\r
+ ldmfd sp!, {r4-r8,lr}\r
+ bx lr\r
+\r
+\r
+.fl_32scale_RGB555u:\r
+ sub r0, r0, #2 @ initial adjustment\r
+ mov r4, #0\r
+\r
+ @ r9 f800 07e0 001f | e000 0780 001c | 3800 01e0 0007\r
+.fl_loop32scale_RGB555u:\r
+ ldr r12, [r1], #4\r
+ ldr r7, [r1], #4\r
+\r
+ and r6, lr, r12,lsl #1\r
+ ldrh r6, [r3, r6]\r
+ and r5, lr, r12,lsr #7\r
+ ldrh r5, [r3, r5]\r
+ and r6, r6, r9, lsl #2\r
+ orr r4, r4, r6, lsl #16 @ r4 = pix_d -1, 0\r
+\r
+ and r5, r5, r9, lsl #2\r
+ sub r8, r5, r5, lsr #2 @ r8 = 3/4 pix_s 1\r
+ add r6, r8, r6, lsr #2 @ r6 = (1/4 pix_s 0) + (3/4 pix_s 1)\r
+ orr r5, r6, r5, lsl #15\r
+\r
+ and r6, lr, r12,lsr #15\r
+ ldrh r6, [r3, r6]\r
+ and r12,lr, r12,lsr #23\r
+ ldrh r12,[r3, r12]\r
+ and r6, r6, r9, lsl #2\r
+ add r5, r5, r6, lsl #15 @ r5 = pix_d 1, 2\r
+\r
+ and r8, lr, r7, lsl #1\r
+ ldrh r8, [r3, r8]\r
+ and r10,lr, r7, lsr #7\r
+ ldrh r10,[r3, r10]\r
+ and r12,r12,r9, lsl #2\r
+ sub r6, r6, r6, lsr #2 @ r6 = 3/4 pix_s 2\r
+ add r6, r6, r12,lsr #2\r
+ orr r6, r6, r12,lsl #16 @ r6 = pix_d 3, 4\r
+\r
+ and r8, r8, r9, lsl #2\r
+ and r10,r10,r9, lsl #2\r
+ sub r12,r10,r10,lsr #2 @ r12 = 3/4 pix_s 5\r
+ orr r8, r8, r8, lsl #14\r
+ add r8, r8, r12,lsl #16 @ r8 = pix_d 5, 6\r
+ and r12,lr, r7, lsr #15\r
+ ldrh r12,[r3, r12]\r
+ and r7, lr, r7, lsr #23\r
+ ldrh r7, [r3, r7]\r
+ and r12,r12,r9, lsl #2\r
+ add r10,r10,r12\r
+ mov r10,r10, lsr #1\r
+ sub r12,r12,r12,lsr #2 @ r12 = 3/4 pix_s 6\r
+ orr r10,r10,r12,lsl #16\r
+ and r7, r7, r9, lsl #2\r
+ add r10,r10,r7, lsl #14 @ r10 = pix_d 7, 8\r
+\r
+ subs r2, r2, #1\r
+\r
+ stmia r0!, {r4,r5,r6,r8,r10}\r
+ mov r4, r7\r
+ bne .fl_loop32scale_RGB555u\r
+\r
+ strh r4, [r0], #2\r
+\r
+ ldmfd sp!, {r9,r10}\r
+ ldmfd sp!, {r4-r8,lr}\r
+ bx lr\r
+\r
+.endif @ UNALIGNED_DRAWLINEDEST\r
+\r
\r
@ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@\r
\r