X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=Pico%2FDraw.s;h=f31025362dfd6b5e721d6e2ec2c035959acb48f6;hb=d95259bdaaf911218656d8a74b096ff7306034f6;hp=e31ba0baa6911ad14782bf7823d97d4b6051c6af;hpb=83c093a48ab58670ea82d0ec81658daa9f9b950a;p=picodrive.git diff --git a/Pico/Draw.s b/Pico/Draw.s index e31ba0b..f310253 100644 --- a/Pico/Draw.s +++ b/Pico/Draw.s @@ -1,11 +1,12 @@ @ vim:filetype=armasm -@ assembly "optimized" version of some funtions from draw.c +@ ARM assembly versions of some funtions from draw.c @ this is highly specialized, be careful if changing related C code! -@ (c) Copyright 2007, Grazvydas "notaz" Ignotas +@ (c) Copyright 2007-2008, Grazvydas "notaz" Ignotas @ All Rights Reserved +.include "port_config.s" .extern Pico .extern PicoOpt @@ -15,6 +16,7 @@ .extern rendstatus .extern DrawLineDest .extern DrawStripInterlace +.extern HighCacheS_ptr @ helper @@ -63,10 +65,8 @@ .endif ldreqb r4, [r1,#\offs] orrne r4, r3, r4 - strneb r4, [r1,#\offs] - tsteq r4, #0x80 andeq r4, r4, #0x3f - streqb r4, [r1,#\offs] + strb r4, [r1,#\offs] .endm @ TileNormShHP (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: register with helper pattern 0xf, touches r3 high bits @@ -154,24 +154,17 @@ .else ands r4, r12, r2 .endif - beq 3f + beq 0f cmp r4, #0xe - beq 2f - bgt 1f - orr r4, r3, r4 - strb r4, [r1,#\ofs] - b 3f -1: - ldrb r4, [r1,#\ofs] @ 2ci - orr r4, r4, #0xc0 - strb r4, [r1,#\ofs] - b 3f -2: - ldrb r4, [r1,#\ofs] @ 2ci - bic r4, r4, #0xc0 - orr r4, r4, #0x80 + ldrgeb r4, [r1,#\ofs] + orrlt r4, r3, r4 @ normal + + biceq r4, r4, #0xc0 @ hilight + orreq r4, r4, #0x80 + orrgt r4, r4, #0xc0 @ shadow + strb r4, [r1,#\ofs] -3: +0: .endm @ TileFlipSh (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx, r12: register with helper pattern 0xf @@ -198,6 +191,80 @@ TileDoShGenPixel 16, 7 @ #0x000f0000 .endm +.macro TileDoShGenPixel_noop shift ofs +.if \shift + and r4, r12, r2, lsr #\shift +.else + and r4, r12, r2 +.endif + sub r7, r4, #1 + cmp r7, #0xd + orrcc r4, r3, r4 @ 0-0xc (was 1-0xd) + strccb r4, [r1,#\ofs] +.endm + +.macro TileFlipSh_noop + TileDoShGenPixel_noop 16, 0 @ #0x000f0000 + TileDoShGenPixel_noop 20, 1 @ #0x00f00000 + TileDoShGenPixel_noop 24, 2 @ #0x0f000000 + TileDoShGenPixel_noop 28, 3 @ #0xf0000000 + TileDoShGenPixel_noop 0, 4 @ #0x0000000f + TileDoShGenPixel_noop 4, 5 @ #0x000000f0 + TileDoShGenPixel_noop 8, 6 @ #0x00000f00 + TileDoShGenPixel_noop 12, 7 @ #0x0000f000 +.endm + +.macro TileNormSh_noop + TileDoShGenPixel_noop 12, 0 @ #0x0000f000 + TileDoShGenPixel_noop 8, 1 @ #0x00000f00 + TileDoShGenPixel_noop 4, 2 @ #0x000000f0 + TileDoShGenPixel_noop 0, 3 @ #0x0000000f + TileDoShGenPixel_noop 28, 4 @ #0xf0000000 + TileDoShGenPixel_noop 24, 5 @ #0x0f000000 + TileDoShGenPixel_noop 20, 6 @ #0x00f00000 + TileDoShGenPixel_noop 16, 7 @ #0x000f0000 +.endm + +.macro TileDoShGenPixel_onlyop_lp shift ofs +.if \shift + ands r7, r12, r2, lsr #\shift +.else + ands r7, r12, r2 +.endif + ldrneb r4, [r1,#\ofs] + tstne r4, #0x40 + beq 0f + + cmp r7, #0xe + biceq r4, r4, #0xc0 @ hilight + orreq r4, r4, #0x80 + orrgt r4, r4, #0xc0 @ shadow + strgeb r4, [r1,#\ofs] +0: +.endm + +.macro TileFlipSh_onlyop_lp + TileDoShGenPixel_onlyop_lp 16, 0 @ #0x000f0000 + TileDoShGenPixel_onlyop_lp 20, 1 @ #0x00f00000 + TileDoShGenPixel_onlyop_lp 24, 2 @ #0x0f000000 + TileDoShGenPixel_onlyop_lp 28, 3 @ #0xf0000000 + TileDoShGenPixel_onlyop_lp 0, 4 @ #0x0000000f + TileDoShGenPixel_onlyop_lp 4, 5 @ #0x000000f0 + TileDoShGenPixel_onlyop_lp 8, 6 @ #0x00000f00 + TileDoShGenPixel_onlyop_lp 12, 7 @ #0x0000f000 +.endm + +.macro TileNormSh_onlyop_lp + TileDoShGenPixel_onlyop_lp 12, 0 @ #0x0000f000 + TileDoShGenPixel_onlyop_lp 8, 1 @ #0x00000f00 + TileDoShGenPixel_onlyop_lp 4, 2 @ #0x000000f0 + TileDoShGenPixel_onlyop_lp 0, 3 @ #0x0000000f + TileDoShGenPixel_onlyop_lp 28, 4 @ #0xf0000000 + TileDoShGenPixel_onlyop_lp 24, 5 @ #0x0f000000 + TileDoShGenPixel_onlyop_lp 20, 6 @ #0x00f00000 + TileDoShGenPixel_onlyop_lp 16, 7 @ #0x000f0000 +.endm + @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @@ -317,11 +384,18 @@ DrawLayer: add r8, r8, r9, lsr #8 @ tilex+=cellskip add r7, r7, r9, lsr #5 @ dx+=cellskip<<3; sub r10,r10,r9, lsl #16 @ cells-=cellskip - mvn r9, #0 @ r9=prevcode=-1 @ cache some stuff to avoid mem access +.if OVERRIDE_HIGHCOL + ldr r11,=HighCol + mov r0, #0xf + ldr r11,[r11] +.else ldr r11,=HighCol mov r0, #0xf +.endif + + mvn r9, #0 @ r9=prevcode=-1 add r1, r11, r7 @ r1=pdest @@ -436,8 +510,8 @@ DrawLayer: .DrawStrip_vsscroll: rsb r8, r3, #0 mov r8, r8, lsr #3 @ r8=tilex=(-ts->hscroll)>>3 - bic r8, r8, #0xff000000 - orr r8, r8, r5, lsl #25 @ r8=(xmask[31:25]|had_output[24]|tilex[23:0]) + bic r8, r8, #0x3fc00000 + orr r8, r8, r5, lsl #25 @ r8=(xmask[31:25]|had_output[24]|tilex[21:0]) ldr r4, =Scanline orr r5, r1, r10, lsl #24 @@ -463,13 +537,19 @@ DrawLayer: add r8, r8, r9, lsr #8 @ tilex+=cellskip add r7, r7, r9, lsr #5 @ dx+=cellskip<<3; add r10,r10,r9, lsl #16 @ cell+=cellskip - mvn r9, #0 @ r9=prevcode=-1 @ cache some stuff to avoid mem access +.if OVERRIDE_HIGHCOL ldr r11,=HighCol mov r0, #0xf - add r1, r11, r7 @ r1=pdest + ldr r11,[r11] +.else + ldr r11,=HighCol + mov r0, #0xf +.endif + mvn r9, #0 @ r9=prevcode=-1 + add r1, r11, r7 @ r1=pdest @ r4 & r7 are scratch in this loop .dsloop_vs_subr1: @@ -490,9 +570,9 @@ DrawLayer: ldrh r7, [r7] @ r7=vscroll bic r10,r10,#0xff @ clear old ty - and r4, r5, #0xff0000 - add r4, r4, r7, lsl #16 - and r4, r4, r5, lsl #16 @ r4=line<<16 + and r4, r5, #0xff0000 @ scanline + add r4, r4, r7, lsl #16 @ ... += vscroll + and r4, r4, r5, lsl #16 @ ... &= ymask and r7, r4, #0x70000 orr r10,r10,r7, lsr #15 @ new ty @@ -515,7 +595,7 @@ DrawLayer: beq .DrawStrip_vs_samecode @ we know stuff about this tile already mov r9, r7 @ remember code - orr r8, r8, #1<<24 @ seen non hi-prio tile + orr r8, r8, #(1<<24)@ seen non hi-prio tile movs r2, r9, lsl #20 @ if (code&0x1000) mov r2, r2, lsl #1 @@ -589,7 +669,7 @@ DrawLayer: b .dsloop_vs .dsloop_vs_exit: - tst r8, #1<<24 @ seen non hi-prio tile + tst r8, #(1<<24) @ seen non hi-prio tile ldreq r1, =rendstatus mov r0, #0 ldreq r2, [r1] @@ -639,10 +719,18 @@ DrawLayer: BackFill: stmfd sp!, {r4-r9,lr} +.if OVERRIDE_HIGHCOL + ldr lr, =HighCol + mov r0, r0, lsl #26 + ldr lr, [lr] + mov r0, r0, lsr #26 + add lr, lr, #8 +.else ldr lr, =(HighCol+8) - mov r0, r0, lsl #26 mov r0, r0, lsr #26 +.endif + orr r0, r0, r1, lsl #6 orr r0, r0, r0, lsl #8 orr r0, r0, r0, lsl #16 @@ -680,9 +768,15 @@ DrawTilesFromCache: stmfd sp!, {r4-r8,r11,lr} @ cache some stuff to avoid mem access +.if OVERRIDE_HIGHCOL ldr r11,=HighCol - ldr lr, =(Pico+0x10000) @ lr=Pico.vram mov r12,#0xf + ldr r11,[r11] +.else + ldr r11,=HighCol + mov r12,#0xf +.endif + ldr lr, =(Pico+0x10000) @ lr=Pico.vram mvn r5, #0 @ r5=prevcode=-1 ands r8, r1, #1 @@ -770,39 +864,25 @@ DrawTilesFromCache: b .dtfc_loop .dtfc_shadow_blank: - ldrb r4, [r1] @ 1ci - ldrb r12,[r1,#1] - tst r4, #0x80 - andeq r4, r4,#0x3f - streqb r4, [r1] - tst r12,#0x80 - ldrb r4, [r1,#2] - andeq r12,r12,#0x3f - streqb r12,[r1,#1] - tst r4, #0x80 - ldrb r12,[r1,#3] - andeq r4, r4,#0x3f - streqb r4, [r1,#2] - tst r12,#0x80 - ldrb r4, [r1,#4] - andeq r12,r12,#0x3f - streqb r12,[r1,#3] - tst r4, #0x80 - ldrb r12,[r1,#5] - andeq r4, r4,#0x3f - streqb r4, [r1,#4] - tst r12,#0x80 - ldrb r4, [r1,#6] - andeq r12,r12,#0x3f - streqb r12,[r1,#5] - tst r4, #0x80 - ldrb r12,[r1,#7] - andeq r4, r4,#0x3f - streqb r4, [r1,#6] - tst r12,#0x80 - andeq r12,r12,#0x3f - streqb r12,[r1,#7] - mov r12, #0xf + tst r1, #1 + ldrneb r4, [r1] + mov r6, #0x3f + and r4, r4, #0x3f + strneb r4, [r1], #1 + ldrh r4, [r1] + orr r6, r6, r6, lsl #8 + and r4, r4, r6 + strh r4, [r1], #2 + ldrh r4, [r1] + and r4, r4, r6 + strh r4, [r1], #2 + ldrh r4, [r1] + and r4, r4, r6 + strh r4, [r1], #2 + ldrh r4, [r1] + and r4, r4, r6 + streqh r4, [r1] + strneb r4, [r1] b .dtfc_loop .dtfc_cut_tile: @@ -837,48 +917,45 @@ DrawTilesFromCache: str r2, [r1] add r1, r11,#8 - mov r3, #320/4 - mov r7, #0x80 - orr r7, r7, r7, lsl #8 - orr r7, r7, r7, lsl #16 + mov r3, #320/4/4 mov r6, #0x3f orr r6, r6, r6, lsl #8 orr r6, r6, r6, lsl #16 .dtfc_loop_shprep: + ldmia r1, {r2,r4,r5,r7} subs r3, r3, #1 - bmi .dtfc_loop @ done - ldr r2, [r1] - tst r2, r7 - andeq r2, r2, r6 - streq r2, [r1], #4 - beq .dtfc_loop_shprep - tst r2, #0x80000000 - biceq r2, r2, #0xc0000000 - tst r2, #0x00800000 - biceq r2, r2, #0x00c00000 - tst r2, #0x00008000 - biceq r2, r2, #0x0000c000 - tst r2, #0x00000080 - biceq r2, r2, #0x000000c0 - str r2, [r1], #4 - b .dtfc_loop_shprep + and r2, r2, r6 + and r4, r4, r6 + and r5, r5, r6 + and r7, r7, r6 + stmia r1!,{r2,r4,r5,r7} + bne .dtfc_loop_shprep + + mvn r5, #0 @ r5=prevcode=-1 + b .dtfc_loop .pool @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ -.global DrawSpritesFromCache @ int *hc, int sh +.global DrawSpritesFromCache @ int *hc, int maxwidth, int prio, int sh DrawSpritesFromCache: stmfd sp!, {r4-r11,lr} @ cache some stuff to avoid mem access +.if OVERRIDE_HIGHCOL + ldr r11,=HighCol + mov r12,#0xf + ldr r11,[r11] +.else ldr r11,=HighCol + mov r12,#0xf +.endif ldr lr, =(Pico+0x10000) @ lr=Pico.vram - mov r6, r1, lsl #31 + mov r6, r3, lsl #31 orr r6, r6, #1<<30 - mov r12,#0xf mov r10, r0 @@ -964,6 +1041,9 @@ DrawSpritesFromCache: b .dsfc_inloop .dsfc_shadow: + tst r9, #0x80000000 + beq .dsfc_shadow_lowpri + cmp r2, r2, ror #4 beq .dsfc_singlec_sh @@ -991,6 +1071,18 @@ DrawSpritesFromCache: TileSingleSh b .dsfc_inloop +.dsfc_shadow_lowpri: + tst r9, #0x10000 + bne .dsfc_TileFlip_sh_lp + +.dsfc_TileNorm_sh_lp: + TileNormSh_onlyop_lp + b .dsfc_inloop + +.dsfc_TileFlip_sh_lp: + TileFlipSh_onlyop_lp + b .dsfc_inloop + .pool @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @@ -998,11 +1090,12 @@ DrawSpritesFromCache: @ + 0 : hhhhvvvv ab--hhvv yyyyyyyy yyyyyyyy // a: offscreen h, b: offs. v, h: horiz. size @ + 4 : xxxxxxxx xxxxxxxx pccvhnnn nnnnnnnn // x: x coord + 8 -.global DrawSprite @ unsigned int *sprite, int **hc, int sh +.global DrawSprite @ unsigned int *sprite, int sh, int acc_sprites DrawSprite: stmfd sp!, {r4-r9,r11,lr} + orr r8, r2, r1, lsl #4 ldr r3, [r0] @ sprite[0] ldr r7, =Scanline mov r6, r3, lsr #28 @@ -1016,44 +1109,53 @@ DrawSprite: ldr r9, [r0, #4] sub r7, r7, r4, asr #16 @ r7=row=Scanline-sy - tst r2, r2 mov r2, r9, asr #16 @ r2=sx - bic r9, r9, #0xfe000000 - orrne r9, r9, #1<<31 @ r9=code|(sh<<31) + mov r9, r9, lsl #16 + mov r9, r9, lsr #16 + orr r9, r9, r8, lsl #27 @ r9=code|sh[31]|as[27] tst r9, #0x1000 movne r4, r5, lsl #3 subne r4, r4, #1 subne r7, r4, r7 @ if (code&0x1000) row=(height<<3)-1-row; // Flip Y - mov r8, r9, lsl #21 - mov r8, r8, lsr #21 - add r8, r8, r7, lsr #3 @ tile+=row>>3; // Tile number increases going down - + add r8, r9, r7, lsr #3 @ tile+=row>>3; // Tile number increases going down tst r9, #0x0800 mlane r8, r5, r6, r8 @ if (code&0x0800) { tile+=delta*(width-1); rsbne r5, r5, #0 @ delta=-delta; } // r5=delta now - mov r8, r8, lsl #4 + mov r8, r8, lsl #21 + mov r8, r8, lsr #17 and r7, r7, #7 add r8, r8, r7, lsl #1 @ tile+=(row&7)<<1; // Tile address tst r9, #0x8000 - bne .dspr_cache @ if(code&0x8000) // high priority - cache it - + tsteq r9, #(1<<27) + bne .dspr_cache @ if(code&0x8000) || as + tst r6, #0x4000 + tstne r6, #0x2000 + tstne r9, #(1<<31) + bne .dspr_cache @ (sh && pal == 0x30) + +.dspr_continue: @ cache some stuff to avoid mem access +.if OVERRIDE_HIGHCOL ldr r11,=HighCol - ldr lr, =(Pico+0x10000) @ lr=Pico.vram mov r12,#0xf + ldr r11,[r11] +.else + ldr r11,=HighCol + mov r12,#0xf +.endif + ldr lr, =(Pico+0x10000) @ lr=Pico.vram mov r5, r5, lsl #4 @ delta<<=4; // Delta of address and r4, r9, #0x6000 orr r9, r9, r4, lsl #16 - orr r9, r9, #0x10000000 @ r9=scc1 ???? ... (s=shadow/hilight, cc=pal) + orrs r9, r9, #0x10000000 @ r9=scc1 a??? ... (s=shadow/hilight, cc=pal, a=acc_spr) - tst r9, #1<<31 mov r3, r4, lsr #9 @ r3=pal=((code>>9)&0x30); - orrne r3, r3, #0x40 @ shadow by default + orrmi r3, r3, #0x40 @ shadow by default add r6, r6, #1 @ inc now adds r0, r2, #0 @ mov sx to r0 and set ZV flags @@ -1097,6 +1199,10 @@ DrawSprite: TileFlip r12 b .dspr_loop +.dspr_singlec_sh: + cmp r2, #0xe0000000 + bcs .dspr_loop @ operator tileline, ignore + .dspr_SingleColor: and r4, r2, #0xf orr r4, r3, r4 @@ -1119,44 +1225,38 @@ DrawSprite: @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern .dspr_TileNorm_sh: - TileNormSh + TileNormSh_noop b .dspr_loop .dspr_TileFlip_sh: - TileFlipSh - b .dspr_loop - -.dspr_singlec_sh: - cmp r2, #0xe0000000 - bcc .dspr_SingleColor @ normal tileline - tst r2, #0x10000000 - bne .dspr_sh_sh - TileSingleHi - b .dspr_loop - -.dspr_sh_sh: - TileSingleSh + TileFlipSh_noop b .dspr_loop .dspr_cache: - @ *(*hc)++ = (tile<<16)|((code&0x0800)<<5)|((sx<<6)&0x0000ffc0)|((code>>9)&0x30)|((sprite[0]>>24)&0xf); + @ *HighCacheS_ptr++ = ((code&0x8000)<<16)|(tile<<16)|((code&0x0800)<<5)|((sx<<6)&0x0000ffc0)|pal|((sprite[0]>>16)&0xf); + ldr r1, =HighCacheS_ptr mov r4, r8, lsl #16 @ tile tst r9, #0x0800 orrne r4, r4, #0x10000 @ code&0x0800 - mov r2, r2, lsl #22 - orr r4, r4, r2, lsr #16 @ (sx<<6)&0x0000ffc0 - and r2, r9, #0x6000 - orr r4, r4, r2, lsr #9 @ (code>>9)&0x30 + mov r0, r2, lsl #22 + orr r4, r4, r0, lsr #16 @ (sx<<6)&0x0000ffc0 + and r0, r9, #0x6000 + orr r4, r4, r0, lsr #9 @ (code>>9)&0x30 mov r3, r3, lsl #12 - ldr r2, [r1] orr r4, r4, r3, lsr #28 @ (sprite[0]>>24)&0xf - str r4, [r2], #4 - str r2, [r1] + ldr r0, [r1] + tst r9, #0x8000 + orrne r4, r4, #0x80000000 @ prio - ldmfd sp!, {r4-r9,r11,lr} - bx lr + str r4, [r0], #4 + str r0, [r1] + + and r0, r9, #(1<<27) @ as + teqne r0, #(1<<27) @ (code&0x8000) && !as + ldmnefd sp!, {r4-r9,r11,pc} + b .dspr_continue @ draw anyway if accurate sprites enabled @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @@ -1186,33 +1286,38 @@ DrawWindow: ldr r6, =rendstatus ldr lr, =(Pico+0x10000) @ lr=Pico.vram - ldrb r6, [r6] + ldr r6, [r6] @ fetch the first code now ldrh r7, [lr, r12] ands r6, r6, #2 @ we care about bit 1 only orr r6, r6, r2 - bne .dw_no_sameprio - cmp r2, r7, lsr #15 - ldmnefd sp!, {r4-r11,pc} @ assume that whole window uses same priority + teqne r2, r7, lsr #15 @ do prio bits differ? + ldmnefd sp!, {r4-r11,pc} @ yes, assume that whole window uses same priority -.dw_no_sameprio: orr r6, r6, r3, lsl #8 @ shadow mode sub r8, r1, r0 - mov r8, r8, lsl #1 @ cells - - mvn r9, #0 @ r9=prevcode=-1 @ cache some stuff to avoid mem access +.if OVERRIDE_HIGHCOL + ldr r11,=HighCol + mov r8, r8, lsl #1 @ cells + ldr r11,[r11] + mvn r9, #0 @ r9=prevcode=-1 + add r11,r11,#8 +.else ldr r11,=(HighCol+8) - add r1, r11, r0, lsl #4 @ r1=pdest + mov r8, r8, lsl #1 @ cells + mvn r9, #0 @ r9=prevcode=-1 +.endif + add r1, r11, r0, lsl #4 @ r1=pdest mov r0, #0xf b .dwloop_enter - @ r4,r5 & r7 are scratch in this loop + @ r4,r5 are scratch in this loop .dwloop: add r1, r1, #8 .dwloop_nor1: @@ -1278,24 +1383,13 @@ DrawWindow: orreq r3, r3, #0x40 beq .dw_shadow_done ldr r4, [r1] - tst r4, #0x00000080 - biceq r4, r4, #0x000000c0 - tst r4, #0x00008000 - biceq r4, r4, #0x0000c000 - tst r4, #0x00800000 - biceq r4, r4, #0x00c00000 - tst r4, #0x80000000 - biceq r4, r4, #0xc0000000 + mov r5, #0x3f + orr r5, r5, r5, lsl #8 + orr r5, r5, r5, lsl #16 + and r4, r4, r5 str r4, [r1] ldr r4, [r1,#4] - tst r4, #0x00000080 - biceq r4, r4, #0x000000c0 - tst r4, #0x00008000 - biceq r4, r4, #0x0000c000 - tst r4, #0x00800000 - biceq r4, r4, #0x00c00000 - tst r4, #0x80000000 - biceq r4, r4, #0xc0000000 + and r4, r4, r5 str r4, [r1,#4] b .dw_shadow_done @@ -1405,15 +1499,29 @@ FinalizeLineBGR444: bne .fl_loopcpBGR444_hi sub r3, r4, #0x40*3*2 + mov r6, #1 .fl_noshBGR444: - ldr r1, =(HighCol+8) + ldr r12,=rendstatus + eors r6, r6, #1 @ sh is 0 + ldr r12,[r12] mov lr, #0xff + tstne r12,#(1<<2) @ and PDRAW_ACC_SPRITES + +.if OVERRIDE_HIGHCOL + ldr r1, =HighCol + movne lr, #0x3f + ldr r1, [r1] mov lr, lr, lsl #1 + add r1, r1, #8 +.else + ldr r1, =(HighCol+8) + movne lr, #0x3f + mov lr, lr, lsl #1 +.endif .fl_loopBGR444: - ldr r12, [r1], #4 subs r2, r2, #1 @@ -1423,11 +1531,10 @@ FinalizeLineBGR444: ldrh r5, [r3, r5] and r6, lr, r12, lsr #15 ldrh r6, [r3, r6] + and r12,lr, r12, lsr #23 + ldrh r12,[r3, r12] @ 1c.i. orr r4, r4, r5, lsl #16 - - and r5, lr, r12, lsr #23 - ldrh r5, [r3, r5] @ 2c.i. - orr r5, r6, r5, lsl #16 + orr r5, r6, r12,lsl #16 stmia r0!, {r4,r5} bne .fl_loopBGR444 @@ -1480,14 +1587,16 @@ FinalizeLineBGR444: orr \reg, \reg, r3 @ add blue back .endm +.global vidConvCpyRGB565 + vidConvCpyRGB565: @ void *to, void *from, int pixels stmfd sp!, {r4-r9,lr} - mov r12, r2, lsr #3 @ repeats + mov r12, r2, lsr #3 @ repeats mov lr, #0x001c0000 orr lr, lr, #0x01c @ lr == pattern 0x001c001c mov r8, #0x00030000 - orr r8, r8, #0x003 @ lr == pattern 0x001c001c + orr r8, r8, #0x003 .loopRGB565: ldmia r1!, {r4-r7} @@ -1560,14 +1669,29 @@ FinalizeLineRGB555: bne .fl_loopcpRGB555_hi sub r3, r3, #0x40*2 + mov r6, #1 .fl_noshRGB555: + ldr r12,=rendstatus + eors r6, r6, #1 @ sh is 0 + ldr r12,[r12] + mov lr, #0xff + tstne r12,#(1<<2) @ and PDRAW_ACC_SPRITES + movne lr, #0x3f + +.if OVERRIDE_HIGHCOL + ldr r1, =HighCol + ldr r0, =DrawLineDest + ldr r1, [r1] + ldr r0, [r0] + add r1, r1, #8 +.else ldr r0, =DrawLineDest ldr r1, =(HighCol+8) ldr r0, [r0] +.endif ldrb r12, [r8, #12] - mov lr, #0xff mov lr, lr, lsl #1 tst r12, #1 @@ -1582,8 +1706,14 @@ FinalizeLineRGB555: addeq r0, r0, #32*2 .fl_no32colRGB555: -.fl_loopRGB555: +.if UNALIGNED_DRAWLINEDEST + @ this is basically for Gizmondo, which has unaligned odd lines in the framebuffer + tst r0, #2 + bne .fl_RGB555u +.endif + +.fl_loopRGB555: ldr r12, [r1], #4 ldr r7, [r1], #4 @@ -1605,12 +1735,12 @@ FinalizeLineRGB555: ldrh r6, [r3, r6] and r12,lr, r7, lsr #15 ldrh r12,[r3, r12] + and r7, lr, r7, lsr #23 + ldrh r7, [r3, r7] orr r8, r8, r6, lsl #16 - and r6, lr, r7, lsr #23 - ldrh r6, [r3, r6] @ 1 cycle interlock here (r6) subs r2, r2, #1 - orr r12,r12, r6, lsl #16 + orr r12,r12, r7, lsl #16 stmia r0!, {r4,r5,r8,r12} bne .fl_loopRGB555 @@ -1624,6 +1754,11 @@ FinalizeLineRGB555: mov r9, #0x3900 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007 orr r9, r9, #0x00e7 +.if UNALIGNED_DRAWLINEDEST + tst r0, #2 + bne .fl_32scale_RGB555u +.endif + .fl_loop32scale_RGB555: ldr r12, [r1], #4 ldr r7, [r1], #4 @@ -1680,6 +1815,121 @@ FinalizeLineRGB555: ldmfd sp!, {r4-r8,lr} bx lr +.if UNALIGNED_DRAWLINEDEST + @ unaligned versions of loops + @ warning: starts drawing 2bytes before dst + +.fl_RGB555u: + sub r0, r0, #2 @ initial adjustment + mov r8, #0 + +.fl_loopRGB555u: + ldr r12, [r1], #4 + ldr r7, [r1], #4 + + and r6, lr, r12,lsl #1 + ldrh r6, [r3, r6] + and r5, lr, r12,lsr #7 + ldrh r5, [r3, r5] + orr r4, r8, r6, lsl #16 + + and r6, lr, r12,lsr #15 + ldrh r6, [r3, r6] + and r8, lr, r12,lsr #23 + ldrh r8, [r3, r8] + orr r5, r5, r6, lsl #16 + + and r6, lr, r7, lsl #1 + ldrh r6, [r3, r6] + and r12,lr, r7, lsr #7 + ldrh r12,[r3, r12] + orr r6, r8, r6, lsl #16 + + and r8, lr, r7, lsr #15 + ldrh r8, [r3, r8] + and r7, lr, r7, lsr #23 + + subs r2, r2, #1 + orr r12,r12,r8, lsl #16 + ldrh r8, [r3, r7] + + stmia r0!, {r4,r5,r6,r12} + bne .fl_loopRGB555u + + strh r8, [r0], #2 + + ldmfd sp!, {r4-r8,lr} + bx lr + + +.fl_32scale_RGB555u: + sub r0, r0, #2 @ initial adjustment + mov r4, #0 + + @ r9 f800 07e0 001f | e000 0780 001c | 3800 01e0 0007 +.fl_loop32scale_RGB555u: + ldr r12, [r1], #4 + ldr r7, [r1], #4 + + and r6, lr, r12,lsl #1 + ldrh r6, [r3, r6] + and r5, lr, r12,lsr #7 + ldrh r5, [r3, r5] + and r6, r6, r9, lsl #2 + orr r4, r4, r6, lsl #16 @ r4 = pix_d -1, 0 + + and r5, r5, r9, lsl #2 + sub r8, r5, r5, lsr #2 @ r8 = 3/4 pix_s 1 + add r6, r8, r6, lsr #2 @ r6 = (1/4 pix_s 0) + (3/4 pix_s 1) + orr r5, r6, r5, lsl #15 + + and r6, lr, r12,lsr #15 + ldrh r6, [r3, r6] + and r12,lr, r12,lsr #23 + ldrh r12,[r3, r12] + and r6, r6, r9, lsl #2 + add r5, r5, r6, lsl #15 @ r5 = pix_d 1, 2 + + and r8, lr, r7, lsl #1 + ldrh r8, [r3, r8] + and r10,lr, r7, lsr #7 + ldrh r10,[r3, r10] + and r12,r12,r9, lsl #2 + sub r6, r6, r6, lsr #2 @ r6 = 3/4 pix_s 2 + add r6, r6, r12,lsr #2 + orr r6, r6, r12,lsl #16 @ r6 = pix_d 3, 4 + + and r8, r8, r9, lsl #2 + and r10,r10,r9, lsl #2 + sub r12,r10,r10,lsr #2 @ r12 = 3/4 pix_s 5 + orr r8, r8, r8, lsl #14 + add r8, r8, r12,lsl #16 @ r8 = pix_d 5, 6 + and r12,lr, r7, lsr #15 + ldrh r12,[r3, r12] + and r7, lr, r7, lsr #23 + ldrh r7, [r3, r7] + and r12,r12,r9, lsl #2 + add r10,r10,r12 + mov r10,r10, lsr #1 + sub r12,r12,r12,lsr #2 @ r12 = 3/4 pix_s 6 + orr r10,r10,r12,lsl #16 + and r7, r7, r9, lsl #2 + add r10,r10,r7, lsl #14 @ r10 = pix_d 7, 8 + + subs r2, r2, #1 + + stmia r0!, {r4,r5,r6,r8,r10} + mov r4, r7 + bne .fl_loop32scale_RGB555u + + strh r4, [r0], #2 + + ldmfd sp!, {r9,r10} + ldmfd sp!, {r4-r8,lr} + bx lr + +.endif @ UNALIGNED_DRAWLINEDEST + @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@