/* * assembly optimized versions of most funtions from draw.c * (C) notaz, 2006-2010,2017 * * This work is licensed under the terms of MAME license. * See COPYING file in the top-level directory. * * this is highly specialized, be careful if changing related C code! */ #include "pico_int_o32.h" .extern DrawStripInterlace .equ PDRAW_SPRITES_MOVED, (1<<0) .equ PDRAW_WND_DIFF_PRIO, (1<<1) .equ PDRAW_ACC_SPRITES, (1<<2) .equ PDRAW_DIRTY_SPRITES, (1<<4) .equ PDRAW_PLANE_HI_PRIO, (1<<6) .equ PDRAW_SHHI_DONE, (1<<7) @ helper .macro TilePixel pat lsrr offs .if !\lsrr ands r4, \pat, r2 .else ands r4, \pat, r2, lsr #\lsrr .endif orrne r4, r3, r4 strneb r4, [r1,#\offs] .endm @ TileNorm (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf .macro TileNorm pat TilePixel \pat, 12, 0 @ #0x0000f000 TilePixel \pat, 8, 1 @ #0x00000f00 TilePixel \pat, 4, 2 @ #0x000000f0 TilePixel \pat, 0, 3 @ #0x0000000f TilePixel \pat, 28, 4 @ #0xf0000000 TilePixel \pat, 24, 5 @ #0x0f000000 TilePixel \pat, 20, 6 @ #0x00f00000 TilePixel \pat, 16, 7 @ #0x000f0000 .endm @ TileFlip (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf .macro TileFlip pat TilePixel \pat, 16, 0 @ #0x000f0000 TilePixel \pat, 20, 1 @ #0x00f00000 TilePixel \pat, 24, 2 @ #0x0f000000 TilePixel \pat, 28, 3 @ #0xf0000000 TilePixel \pat, 0, 4 @ #0x0000000f TilePixel \pat, 4, 5 @ #0x000000f0 TilePixel \pat, 8, 6 @ #0x00000f00 TilePixel \pat, 12, 7 @ #0x0000f000 .endm @ shadow/hilight mode @ this one is for hi priority layer .macro TilePixelShHP lsrr offs .if !\lsrr ands r4, r12, r2 .else ands r4, r12, r2, lsr #\lsrr .endif ldreqb r4, [r1,#\offs] orrne r4, r3, r4 andeq r4, r4, #0xbf strb r4, [r1,#\offs] .endm @ TileNormShHP (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: register with helper pattern 0xf, touches r3 high bits .macro TileNormShHP TilePixelShHP 12, 0 @ #0x0000f000 TilePixelShHP 8, 1 @ #0x00000f00 TilePixelShHP 4, 2 @ #0x000000f0 TilePixelShHP 0, 3 @ #0x0000000f TilePixelShHP 28, 4 @ #0xf0000000 TilePixelShHP 24, 5 @ #0x0f000000 TilePixelShHP 20, 6 @ #0x00f00000 TilePixelShHP 16, 7 @ #0x000f0000 .endm @ TileFlipShHP (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf .macro TileFlipShHP TilePixelShHP 16, 0 @ #0x000f0000 TilePixelShHP 20, 1 @ #0x00f00000 TilePixelShHP 24, 2 @ #0x0f000000 TilePixelShHP 28, 3 @ #0xf0000000 TilePixelShHP 0, 4 @ #0x0000000f TilePixelShHP 4, 5 @ #0x000000f0 TilePixelShHP 8, 6 @ #0x00000f00 TilePixelShHP 12, 7 @ #0x0000f000 .endm @ TileSingleSh (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx; r12: helper pattern 0xf .macro TileSingleSh tst r0, #1 @ not aligned? mov r7, #0x00c000 orr r7, r7, #0xc0 ldrneb r4, [r1] ldreqh r4, [r1] orr r4, r4, r7 strneb r4, [r1], #1 streqh r4, [r1], #2 ldrh r4, [r1] orr r4, r4, r7 strh r4, [r1], #2 ldrh r4, [r1] orr r4, r4, r7 strh r4, [r1], #2 ldrh r4, [r1] orr r4, r4, r7 strh r4, [r1], #2 ldrneb r4, [r1] orr r4, r4, r7 strneb r4, [r1], #1 .endm @ TileSingleHi (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx, r12: register with helper pattern 0xf .macro TileSingleHi tst r1, #1 @ not aligned? mov r7, #0x008000 orr r7, r7, #0x80 ldrneb r4, [r1], #1 ldreqh r4, [r1], #2 @ 1ci ldrh r12, [r1], #2 bic r4, r4, r7, lsr #1 orr r4, r4, r7 strneb r4, [r1, #-3] streqh r4, [r1, #-4] ldrh r4, [r1], #2 bic r12, r12, r7, lsr #1 orr r12, r12, r7 strh r12, [r1, #-4] ldrh r12, [r1], #2 bic r4, r4, r7, lsr #1 orr r4, r4, r7 strh r4, [r1, #-4] ldrneb r4, [r1] bic r12, r12, r7, lsr #1 orr r12, r12, r7 strh r12, [r1, #-2] bicne r4, r4, r7, lsr #1 orrne r4, r4, r7 strneb r4, [r1], #1 mov r12, #0xf .endm .macro TileDoShGenPixel shift ofs .if \shift ands r4, r12, r2, lsr #\shift .else ands r4, r12, r2 .endif beq 0f cmp r4, #0xe ldrgeb r7, [r1,#\ofs] orrlt r7, r3, r4 @ normal bicge r7, r7, #0xc0 orrge r7, r7, r4, lsl #6 strb r7, [r1,#\ofs] 0: .endm @ TileFlipSh (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx, r12: register with helper pattern 0xf .macro TileFlipSh TileDoShGenPixel 16, 0 @ #0x000f0000 TileDoShGenPixel 20, 1 @ #0x00f00000 TileDoShGenPixel 24, 2 @ #0x0f000000 TileDoShGenPixel 28, 3 @ #0xf0000000 TileDoShGenPixel 0, 4 @ #0x0000000f TileDoShGenPixel 4, 5 @ #0x000000f0 TileDoShGenPixel 8, 6 @ #0x00000f00 TileDoShGenPixel 12, 7 @ #0x0000f000 .endm @ TileNormSh (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx, r12: register with helper pattern 0xf .macro TileNormSh TileDoShGenPixel 12, 0 @ #0x0000f000 TileDoShGenPixel 8, 1 @ #0x00000f00 TileDoShGenPixel 4, 2 @ #0x000000f0 TileDoShGenPixel 0, 3 @ #0x0000000f TileDoShGenPixel 28, 4 @ #0xf0000000 TileDoShGenPixel 24, 5 @ #0x0f000000 TileDoShGenPixel 20, 6 @ #0x00f00000 TileDoShGenPixel 16, 7 @ #0x000f0000 .endm .macro TileDoShGenPixel_markop shift ofs .if \shift ands r4, r12, r2, lsr #\shift .else ands r4, r12, r2 .endif beq 0f cmp r4, #0xe ldrgeb r4, [r1,#\ofs] orrlt r4, r3, r4 orrge r4, r4, #0x80 strb r4, [r1,#\ofs] 0: .endm .macro TileFlipSh_markop TileDoShGenPixel_markop 16, 0 @ #0x000f0000 TileDoShGenPixel_markop 20, 1 @ #0x00f00000 TileDoShGenPixel_markop 24, 2 @ #0x0f000000 TileDoShGenPixel_markop 28, 3 @ #0xf0000000 TileDoShGenPixel_markop 0, 4 @ #0x0000000f TileDoShGenPixel_markop 4, 5 @ #0x000000f0 TileDoShGenPixel_markop 8, 6 @ #0x00000f00 TileDoShGenPixel_markop 12, 7 @ #0x0000f000 .endm .macro TileNormSh_markop TileDoShGenPixel_markop 12, 0 @ #0x0000f000 TileDoShGenPixel_markop 8, 1 @ #0x00000f00 TileDoShGenPixel_markop 4, 2 @ #0x000000f0 TileDoShGenPixel_markop 0, 3 @ #0x0000000f TileDoShGenPixel_markop 28, 4 @ #0xf0000000 TileDoShGenPixel_markop 24, 5 @ #0x0f000000 TileDoShGenPixel_markop 20, 6 @ #0x00f00000 TileDoShGenPixel_markop 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] cmp r7, #0xe blt 0f tst r4, #0xc0 bicne r4, r4, #0xc0 orrne r4, r4, r7, lsl #6 strneb 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 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ struct TileStrip @ { @ int nametab; // 0x00 @ int line; // 0x04 @ int hscroll; // 0x08 @ int xmask; // 0x0C @ int *hc; // 0x10 (pointer to cache buffer) @ int cells; // 0x14 @ }; @ void DrawLayer(int lflags, int *hcache, int cellskip, int maxcells, @ struct PicoEState *est) .global DrawLayer DrawLayer: ldr r12, [sp] @ est stmfd sp!, {r4-r11,lr} ldr r11, [r12, #OFS_Pico_video] mov r8, #1 ldrb r7, [r11, #16] @ ??vv??hh mov r6, r1 @ hcache orr r9, r3, r0, lsl #29 @ r9=force[31]|sh[30]|plane[29] orr r9, r9, r2, lsl #8 @ |cellskip[15:8]|maxcells[7:0] (tmp) mov r1, r7, lsl #4 orr r1, r1, #0x00ff and r10, r7, #3 cmp r10, #1 biclt r1, r1, #0xfc00 biceq r1, r1, #0xfe00 cmp r10, #2 moveq r1, #0x0007 movgt r1, #0x00ff @ r1=ymask=(height<<8)|0xff; ...; // Y Mask in pixels add r10, r10, #5 cmp r10, #7 subge r10, r10, #1 @ r10=shift[width] (5,6,6,7) ldr r2, [r12, #OFS_DrawScanline] ldr lr, [r12, #OFS_Pico_vram] @ Find name table: ands r0, r0, #1 ldreqb r12, [r11, #2] ldrneb r12, [r11, #4] @ calculate xmask: mov r5, r8, lsl r10 sub r5, r5, #1 @ r5=xmask moveq r12, r12, lsl #10 movne r12, r12, lsl #13 and r12, r12, #(7<<13) @ r12=(ts->nametab<<1) (halfword compliant) ldrh r8, [r11, #12] ldrb r7, [r11, #11] mov r4, r8, lsr #8 @ pvid->reg[13] mov r4, r4, lsl #10 @ htab=pvid->reg[13]<<9; (halfwords) tst r7, #2 addne r4, r4, r2, lsl #2 @ htab+=DrawScanline<<1; // Offset by line tst r7, #1 biceq r4, r4, #0x1f @ htab&=~0xf; // Offset by tile add r4, r4, r0, lsl #1 @ htab+=plane bic r4, r4, #0x00ff0000 @ just in case ldrh r3, [lr, r4] @ r3=hscroll tst r7, #4 bne .DrawStrip_vsscroll @ Get vertical scroll value: add r7, lr, #0x012000 add r7, r7, #0x000180 @ r7=Pico.vsram (Pico+0x22180) ldr r7, [r7] tst r8, #2 tstne r8, #4 bne .DrawStrip_interlace tst r0, r0 movne r7, r7, lsr #16 @ Find the line in the name table add r2, r2, r7 and r2, r2, r1 mov r4, r2, lsr #3 add r10, r10, #1 @ shift[width]++ add r12, r12, r4, lsl r10 @ nametab+=(ts.line>>3)<hscroll, r5=ts->xmask, r6=ts->hc, r9=ts->cells and r10,r2, #7 mov r10,r10, lsl #1 @ r10=ty=(ts->line&7)<<1; orr r10,r10, r9, lsl #24 rsb r8, r3, #0 mov r8, r8, lsr #3 @ r8=tilex=(-ts->hscroll)>>3 sub r1, r3, #1 and r1, r1, #7 add r7, r1, #1 @ r7=dx=((ts->hscroll-1)&7)+1 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] orrcs r10,r10, #1<<20 @ |had_output[21]|force[20]|ty[15:0] movmi r3, #0x40 @ 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 r9, r9, #0xff00 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 @ cache some stuff to avoid mem access ldr r11,[sp, #9*4] @ est mov r0, #0xf ldr r11,[r11, #OFS_HighCol] mvn r9, #0 @ r9=prevcode=-1 add r1, r11, r7 @ r1=pdest @ r4 & r7 are scratch in this loop .dsloop_subr1: sub r1, r1, #8 .dsloop: @ 40-41 times subs r10,r10, #0x01000000 bmi .dsloop_exit .dsloop_enter: and r7, r5, r8 add r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords) ldrh r7, [r7, r12] @ r7=code (int, but from unsigned, no sign extend) add r1, r1, #8 add r8, r8, #1 tst r7, #0x8000 tsteq r10, #1<<20 @ force? bne .DrawStrip_hiprio cmp r7, r9 beq .DrawStrip_samecode @ we know stuff about this tile already mov r9, r7 @ remember code orr r10, r10, #1<<21 @ seen non hi-prio tile 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 bic r7, r3, #0x3f and r3, r9, #0x6000 add r3, r7, r3, lsr #9 @ r3=pal=((code&0x6000)>>9); .DrawStrip_samecode: tst r2, r2 beq .dsloop @ tileline blank cmp r2, r2, ror #4 beq .DrawStrip_SingleColor @ tileline singlecolor tst r9, #0x0800 bne .DrawStrip_TileFlip @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern .DrawStrip_TileNorm: TileNorm r0 b .dsloop .DrawStrip_TileFlip: TileFlip r0 b .dsloop .DrawStrip_SingleColor: and r4, r2, #0xf orr r4, r3, r4 orr r4, r4, r4, lsl #8 tst r1, #1 @ not aligned? strneb r4, [r1], #1 streqh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strneb r4, [r1], #1 @ have a remaining unaligned pixel? b .dsloop_subr1 .DrawStrip_hiprio_maybempt: cmp r7, r9 beq .dsloop @ must've been empty, otherwise we wouldn't get here movs r2, r7, 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 mov r9, r7 @ remember code tst r2, r2 beq .dsloop orr r10, r10, #1<<22 .DrawStrip_hiprio: tst r10, #0x00d00000 @ sh[23]|hi_not_empty[22]|force[20] beq .DrawStrip_hiprio_maybempt sub r0, r1, r11 orr r7, r7, r0, lsl #16 orr r7, r7, r10, lsl #25 @ (ty<<25) tst r7, #0x1000 eorne r7, r7, #7<<26 @ if(code&0x1000) cval^=7<<26; str r7, [r6], #4 @ cache hi priority tile mov r0, #0xf b .dsloop .dsloop_exit: tst r10, #1<<21 @ seen non hi-prio tile ldr r1, [sp, #9*4] @ est mov r0, #0 ldreq r2, [r1, #OFS_rendstatus] str r0, [r6] @ terminate the cache list orreq r2, r2, #PDRAW_PLANE_HI_PRIO @ had a layer with all hi-prio tiles streq r2, [r1, #OFS_rendstatus] ldmfd sp!, {r4-r11,lr} bx lr @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ .DrawStrip_vsscroll: rsb r8, r3, #0 mov r8, r8, lsr #3 @ r8=tilex=(-ts->hscroll)>>3 bic r8, r8, #0x3fc00000 orr r8, r8, r5, lsl #25 @ r8=(xmask[31:25]|had_output[24]|tilex[21:0]) ldr r11, [sp, #9*4] @ est orr r5, r1, r10, lsl #24 ldr r4, [r11, #OFS_DrawScanline] sub r1, r3, #1 orr r5, r5, r4, lsl #16 @ r5=(shift_width[31:24]|scanline[23:16]|ymask[15:0]) and r1, r1, #7 add r7, r1, #1 @ r7=dx=((ts->hscroll-1)&7)+1 mov r10,r9, lsl #16 tst r0, #1 orrne r10,r10, #0x8000 tst r9, #1<<31 mov r3, #0 orr r10,r10, #0xff000000 @ will be adjusted on entering loop orrne r10,r10, #1<<23 @ r10=(cell[31:24]|sh[23]|hi_not_empty[22]|cells_max[21:16]|plane[15]|ty[14:0]) movne r3, #0x40 @ default to shadowed pal on sh mode cmp r7, #8 subne r10,r10, #0x01000000 @ have hscroll, start with negative cell and r9, r9, #0xff00 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 @ cache some stuff to avoid mem access ldr r11,[sp, #9*4] @ est mov r0, #0xf ldr r11,[r11, #OFS_HighCol] mvn r9, #0 @ r9=prevcode=-1 add r1, r11, r7 @ r1=pdest @ r4 & r7 are scratch in this loop .dsloop_vs_subr1: sub r1, r1, #8 .dsloop_vs: @ 40-41 times add r10,r10, #0x01000000 and r4, r10, #0x003f0000 cmp r4, r10, asr #8 ble .dsloop_vs_exit @ calc offset and read tileline code to r7, also calc ty add r7, lr, #0x012000 add r7, r7, #0x000180 @ r7=Pico.vsram (Pico+0x22180) add r7, r7, r10,asr #23 @ vsram + ((cell&~1)<<1) bic r7, r7, #3 tst r10,#0x8000 @ plane1? addne r7, r7, #2 ldrh r7, [r7] @ r7=vscroll bic r10,r10,#0xff @ clear old ty 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 mov r4, r4, lsr #19 mov r7, r5, lsr #24 mov r4, r4, lsl r7 @ nametabadd and r7, r8, r8, lsr #25 add r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords) add r7, r7, r4, lsl #1 ldrh r7, [r7, r12] @ r7=code (int, but from unsigned, no sign extend) add r1, r1, #8 add r8, r8, #1 tst r7, #0x8000 bne .DrawStrip_vs_hiprio cmp r7, r9 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 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 bic r7, r3, #0x3f and r3, r9, #0x6000 add r3, r7, r3, lsr #9 @ r3=pal=((code&0x6000)>>9); .DrawStrip_vs_samecode: tst r2, r2 beq .dsloop_vs @ tileline blank cmp r2, r2, ror #4 beq .DrawStrip_vs_SingleColor @ tileline singlecolor tst r9, #0x0800 bne .DrawStrip_vs_TileFlip @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern .DrawStrip_vs_TileNorm: TileNorm r0 b .dsloop_vs .DrawStrip_vs_TileFlip: TileFlip r0 b .dsloop_vs .DrawStrip_vs_SingleColor: and r4, r2, #0xf orr r4, r3, r4 orr r4, r4, r4, lsl #8 tst r1, #1 @ not aligned? strneb r4, [r1], #1 streqh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strneb r4, [r1], #1 @ have a remaining unaligned pixel? b .dsloop_vs_subr1 .DrawStrip_vs_hiprio: tst r10, #0x00c00000 beq .DrawStrip_vs_hiprio_maybempt sub r0, r1, r11 orr r7, r7, r0, lsl #16 orr r7, r7, r10, lsl #25 @ (ty<<25) tst r7, #0x1000 eorne r7, r7, #7<<26 @ if(code&0x1000) cval^=7<<26; str r7, [r6], #4 @ cache hi priority tile mov r0, #0xf b .dsloop_vs .DrawStrip_vs_hiprio_maybempt: cmp r7, r9 beq .dsloop_vs @ must've been empty, otherwise we wouldn't get here movs r2, r7, 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 mov r9, r7 @ remember code tst r2, r2 orrne r10, r10, #1<<22 bne .DrawStrip_vs_hiprio b .dsloop_vs .dsloop_vs_exit: tst r8, #(1<<24) @ seen non hi-prio tile ldr r1, [sp, #9*4] @ est mov r0, #0 ldreq r2, [r1, #OFS_rendstatus] str r0, [r6] @ terminate the cache list orreq r2, r2, #PDRAW_PLANE_HI_PRIO @ had a layer with all hi-prio tiles streq r2, [r1, #OFS_rendstatus] ldmfd sp!, {r4-r11,lr} bx lr @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ interlace mode 2? Sonic 2? .DrawStrip_interlace: tst r0, r0 moveq r7, r7, lsl #21 movne r7, r7, lsl #5 @ Find the line in the name table add r2, r7, r2, lsl #22 @ r2=(vscroll+(DrawScanline<<1))<<21 (11 bits); orr r1, r1, #0x80000000 and r2, r2, r1, ror #10 @ &((ymask<<1)|1)<<21; mov r2, r2, lsr #21 mov r4, r2, lsr #4 mov r12, r12, lsr #1 @ halfwords add r0, r12, r4, lsl r10 @ nametab+=(ts.line>>4)<>9); ldr r2, [lr, r2, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels .dtfc_samecode: rsbs r4, r4, r8, lsr #1 bmi .dtfc_cut_tile tst r8, #1 bne .dtfc_shadow tst r2, r2 beq .dtfc_loop cmp r2, r2, ror #4 beq .dtfc_SingleColor @ tileline singlecolor tst r5, #0x0800 bne .dtfc_TileFlip @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern .dtfc_TileNorm: TileNorm r12 b .dtfc_loop .dtfc_TileFlip: TileFlip r12 b .dtfc_loop .dtfc_SingleColor: and r4, r2, #0xf orr r4, r3, r4 orr r4, r4, r4, lsl #8 tst r1, #1 @ not aligned? strneb r4, [r1], #1 streqh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strneb r4, [r1], #1 @ have a remaining unaligned pixel? b .dtfc_loop .dtfc_shadow: tst r2, r2 beq .dtfc_shadow_blank cmp r2, r2, ror #4 beq .dtfc_SingleColor @ tileline singlecolor tst r5, #0x0800 bne .dtfc_TileFlipShHP @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern .dtfc_TileNormShHP: TileNormShHP b .dtfc_loop .dtfc_TileFlipShHP: TileFlipShHP b .dtfc_loop .dtfc_shadow_blank: tst r1, #1 ldrneb r4, [r1] mov r6, #0xbf and r4, r4, #0xbf 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: add r4, r4, #7 @ 0-6 mov r4, r4, lsl #2 mov r12,#0xf<<28 mov r12,r12,asr r4 mov r2, r2, ror #16 tst r5, #0x0800 @ flipped? mvnne r12,r12 and r2, r2, r12 mov r2, r2, ror #16 mov r12,#0xf tst r8, #1 bne .dtfc_shadow tst r2, r2 beq .dtfc_loop tst r5, #0x0800 beq .dtfc_TileNorm b .dtfc_TileFlip @ check if we have detected layer covered with hi-prio tiles: .dtfc_check_rendflags: ldr r2, [r9, #OFS_rendstatus] tst r2, #(PDRAW_PLANE_HI_PRIO|PDRAW_SHHI_DONE) beq .dtfc_loop bic r8, r8, #1 @ sh/hi mode off tst r2, #PDRAW_SHHI_DONE bne .dtfc_loop @ already processed orr r2, r2, #PDRAW_SHHI_DONE str r2, [r9, #OFS_rendstatus] add r1, r11,#8 mov r3, #320/4/4 mov r6, #0xbf 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 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 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ void DrawSpritesSHi(unsigned char *sprited, struct PicoEState *est) .global DrawSpritesSHi DrawSpritesSHi: ldr r3, [r0] mov r12,#0xff ands r3, r3, #0x7f bxeq lr stmfd sp!, {r1,r4-r11,lr} @ +est strb r12,[r0,#2] @ set end marker add r10,r0, #3 @ r10=HighLnSpr end add r10,r10,r3 @ r10=HighLnSpr end ldr r11,[r1, #OFS_HighCol] mov r12,#0xf ldr lr, [r1, #OFS_Pico_vram] DrawSpriteSHi: @ draw next sprite ldrb r0, [r10,#-1]! ldr r7, [sp] @ est ldr r1, [r7, #OFS_HighPreSpr] cmp r0, #0xff ldmeqfd sp!, {r1,r4-r11,pc} @ end of list and r0, r0, #0x7f add r0, r1, r0, lsl #3 ldr r9, [r0, #4] @ sprite[1] mov r2, r9, asr #16 @ r2=sx mov r9, r9, lsl #16 mov r3, r9, lsr #31 @ priority mov r9, r9, lsr #16 @ orr r9, r9, r8, lsl #31 @ r9=code|sh[31] @@ sh is always on here now and r4, r9, #0x6000 orr r9, r9, r4, lsl #16 orr r9, r9, #0x90000000 @ r9=scc1 ???? ... (s=shadow/hilight, cc=pal) cmp r12,r9, lsr #28 @ sh/hi with pal3? cmpne r3, #1 @ if not, is it hi prio? bne DrawSpriteSHi @ non-operator low sprite, already drawn ldr r3, [r0] @ sprite[0] mov r6, r3, lsr #28 sub r6, r6, #1 @ r6=width-1 (inc later) mov r5, r3, lsr #24 and r5, r5, #7 @ r5=height ldr r7, [r7, #OFS_DrawScanline] mov r0, r3, lsl #16 @ r4=sy<<16 (tmp) sub r7, r7, r0, asr #16 @ r7=row=DrawScanline-sy tst r9, #0x1000 movne r0, r5, lsl #3 subne r0, r0, #1 subne r7, r0, r7 @ if (code&0x1000) row=(height<<3)-1-row; // Flip Y 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 #21 mov r8, r8, lsr #17 and r7, r7, #7 add r8, r8, r7, lsl #1 @ tile+=(row&7)<<1; // Tile address mov r5, r5, lsl #4 @ delta<<=4; // Delta of address mov r3, r4, lsr #9 @ r3=pal=((code>>9)&0x30); add r6, r6, #1 @ inc now adds r0, r2, #0 @ mov sx to r0 and set ZV flags b .dsprShi_loop_enter .dsprShi_loop: subs r6, r6, #1 @ width-- beq DrawSpriteSHi adds r0, r0, #8 @ sx+=8 add r8, r8, r5 @ tile+=delta .dsprShi_loop_enter: ble .dsprShi_loop @ sx <= 0 cmp r0, #328 bge DrawSpriteSHi mov r8, r8, lsl #17 mov r8, r8, lsr #17 @ tile&=0x7fff; // Clip tile address ldr r2, [lr, r8, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels add r1, r11, r0 @ r1=pdest tst r2, r2 beq .dsprShi_loop cmp r12, r9, lsr #28 beq .dsprShi_shadow cmp r2, r2, ror #4 beq .dsprShi_SingleColor @ tileline singlecolor tst r9, #0x0800 bne .dsprShi_TileFlip @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern @ scratch: r4, r7 .dsprShi_TileNorm: TileNorm r12 b .dsprShi_loop .dsprShi_TileFlip: TileFlip r12 b .dsprShi_loop .dsprShi_SingleColor: and r4, r2, #0xf orr r4, r3, r4 orr r4, r4, r4, lsl #8 tst r0, #1 @ not aligned? strneb r4, [r1], #1 streqh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strneb r4, [r1], #1 b .dsprShi_loop .dsprShi_shadow: tst r9, #0x8000 beq .dsprShi_shadow_lowpri cmp r2, r2, ror #4 beq .dsprShi_singlec_sh tst r9, #0x0800 bne .dsprShi_TileFlip_sh @ (r1=pdest, r2=pixels8, r3=pal) r4, r7: scratch, r12: helper pattern .dsprShi_TileNorm_sh: TileNormSh b .dsprShi_loop .dsprShi_TileFlip_sh: TileFlipSh b .dsprShi_loop .dsprShi_singlec_sh: cmp r2, #0xe0000000 bcc .dsprShi_SingleColor @ normal singlecolor tileline (carry inverted in ARM) tst r2, #0x10000000 bne .dsprShi_sh_sh TileSingleHi b .dsprShi_loop .dsprShi_sh_sh: TileSingleSh b .dsprShi_loop .dsprShi_shadow_lowpri: tst r9, #0x800 bne .dsprShi_TileFlip_sh_lp .dsprShi_TileNorm_sh_lp: TileNormSh_onlyop_lp b .dsprShi_loop .dsprShi_TileFlip_sh_lp: TileFlipSh_onlyop_lp b .dsprShi_loop .pool @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ void DrawAllSprites(unsigned char *sprited, int prio, int sh, @ struct PicoEState *est) .global DrawAllSprites DrawAllSprites: orr r1, r2, r1, lsl #1 ldr r2, [r0] ands r2, r2, #0x7f bxeq lr @ time to do some real work stmfd sp!, {r1,r3-r11,lr} @ +sh|prio<<1 +est mov r12,#0xff strb r12,[r0,#2] @ set end marker add r10,r0, #3 add r10,r10,r2 @ r10=HighLnSpr end ldr r11,[r3, #OFS_HighCol] mov r12,#0xf ldr lr, [r3, #OFS_Pico_vram] @ + 0 : hhhhvvvv ----hhvv yyyyyyyy yyyyyyyy // v, h: horiz. size @ + 4 : xxxxxxxx xxxxxxxx pccvhnnn nnnnnnnn // x: x coord + 8 DrawSprite: @ draw next sprite ldrb r0, [r10,#-1]! ldr r8, [sp] @ sh|prio<<1 ldr r7, [sp, #4] @ est mov r2, r0, lsr #7 cmp r0, #0xff ldmeqfd sp!, {r1,r3-r11,pc} @ end of list cmp r2, r8, lsr #1 bne DrawSprite @ wrong priority ldr r1, [r7, #OFS_HighPreSpr] and r0, r0, #0x7f add r0, r1, r0, lsl #3 ldr r3, [r0] @ sprite[0] ldr r7, [r7, #OFS_DrawScanline] mov r6, r3, lsr #28 sub r6, r6, #1 @ r6=width-1 (inc later) mov r5, r3, lsr #24 and r5, r5, #7 @ r5=height mov r4, r3, lsl #16 @ r4=sy<<16 (tmp) ldr r9, [r0, #4] sub r7, r7, r4, asr #16 @ r7=row=DrawScanline-sy mov r2, r9, asr #16 @ r2=sx mov r9, r9, lsl #16 mov r9, r9, lsr #16 orr r9, r9, r8, lsl #31 @ r9=code|sh[31] 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 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 #21 mov r8, r8, lsr #17 and r7, r7, #7 add r8, r8, r7, lsl #1 @ tile+=(row&7)<<1; // Tile address .dspr_continue: @ cache some stuff to avoid mem access mov r5, r5, lsl #4 @ delta<<=4; // Delta of address and r4, r9, #0x6000 orr r9, r9, r4, lsl #16 orrs r9, r9, #0x10000000 @ r9=scc1 ???? ... (s=shadow/hilight, cc=pal) mov r3, r4, lsr #9 @ r3=pal=((code>>9)&0x30); orrmi r3, r3, #0x40 @ for sh/hi add r6, r6, #1 @ inc now adds r0, r2, #0 @ mov sx to r0 and set ZV flags b .dspr_loop_enter .dspr_loop: subs r6, r6, #1 @ width-- beq DrawSprite adds r0, r0, #8 @ sx+=8 add r8, r8, r5 @ tile+=delta .dspr_loop_enter: ble .dspr_loop @ sx <= 0 cmp r0, #328 bge DrawSprite mov r8, r8, lsl #17 mov r8, r8, lsr #17 @ tile&=0x7fff; // Clip tile address ldr r2, [lr, r8, lsl #1] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels add r1, r11, r0 @ r1=pdest tst r2, r2 beq .dspr_loop cmp r12, r9, lsr #28 beq .dspr_shadow cmp r2, r2, ror #4 beq .dspr_SingleColor @ tileline singlecolor tst r9, #0x0800 bne .dspr_TileFlip @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern @ scratch: r4, r7 .dspr_TileNorm: TileNorm r12 b .dspr_loop .dspr_TileFlip: TileFlip r12 b .dspr_loop .dspr_singlec_sh: cmp r2, #0xe0000000 bcs .dspr_TileNorm_sh @ op. tileline, markop. XXX: maybe add a spec. handler? .dspr_SingleColor: and r4, r2, #0xf orr r4, r3, r4 orr r4, r4, r4, lsl #8 tst r0, #1 @ not aligned? strneb r4, [r1], #1 streqh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strh r4, [r1], #2 strneb r4, [r1], #1 b .dspr_loop .dspr_shadow: cmp r2, r2, ror #4 beq .dspr_singlec_sh tst r9, #0x0800 bne .dspr_TileFlip_sh @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern .dspr_TileNorm_sh: TileNormSh_markop b .dspr_loop .dspr_TileFlip_sh: TileFlipSh_markop b .dspr_loop @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ void DrawWindow(int tstart, int tend, int prio, int sh @ struct PicoEState *est) .global DrawWindow DrawWindow: ldr r12, [sp] @ est stmfd sp!, {r4-r11,lr} ldr r6, [r12, #OFS_Pico_video] ldr r10, [r12, #OFS_DrawScanline] mov r11, r12 @ est ldrb r12, [r6, #3] @ pvid->reg[3] ldr r4, [r6, #12] mov r5, r10, lsr #3 and r10, r10, #7 mov r10, r10, lsl #1 @ r10=ty mov r12, r12, lsl #10 tst r4, #1 @ 40 cell mode? andne r12, r12, #0xf000 @ 0x3c<<10 andeq r12, r12, #0xf800 addne r12, r12, r5, lsl #7 addeq r12, r12, r5, lsl #6 @ nametab add r12, r12, r0, lsl #2 @ +starttile ldr lr, [r11, #OFS_Pico_vram] ldr r6, [r11, #OFS_rendstatus] @ fetch the first code now ldrh r7, [lr, r12] ands r6, r6, #PDRAW_WND_DIFF_PRIO orr r6, r6, r2 eoreq r8, r2, r7, lsr #15 @ do prio bits differ? cmpeq r8, #1 ldmeqfd sp!, {r4-r11,pc} @ yes, assume that whole window uses same priority orr r6, r6, r3, lsl #8 @ shadow mode sub r8, r1, r0 @ cache some stuff to avoid mem access ldr r11, [r11, #OFS_HighCol] mov r8, r8, lsl #1 @ cells add r11,r11,#8 mvn r9, #0 @ r9=prevcode=-1 add r1, r11, r0, lsl #4 @ r1=pdest mov r0, #0xf b .dwloop_enter @ r4,r5 are scratch in this loop .dwloop: add r1, r1, #8 .dwloop_nor1: add r12, r12, #2 @ halfwords ldrh r7, [lr, r12] @ r7=code (int, but from unsigned, no sign extend) subs r8, r8, #1 beq .dwloop_end @ done eor r5, r6, r7, lsr #15 tst r5, #1 orrne r6, r6, #2 @ wrong pri bne .dwloop cmp r7, r9 beq .dw_samecode @ we know stuff about this tile already .dwloop_enter: 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; 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 .dw_shadow_done: tst r2, r2 beq .dwloop @ tileline blank cmp r2, r2, ror #4 beq .dw_SingleColor @ tileline singlecolor tst r9, #0x0800 bne .dw_TileFlip @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern .dw_TileNorm: TileNorm r0 b .dwloop .dw_TileFlip: TileFlip r0 b .dwloop .dw_SingleColor: and r4, r0, r2 @ #0x0000000f orr r4, r3, r4 orr r4, r4, r4, lsl #8 orr r4, r4, r4, lsl #16 mov r5, r4 stmia r1!, {r4,r5} b .dwloop_nor1 @ we incremeted r1 ourselves .dw_shadow: tst r6, #1 @ hi pri? orreq r3, r3, #0x40 beq .dw_shadow_done ldr r4, [r1] 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] and r4, r4, r5 str r4, [r1,#4] b .dw_shadow_done .dwloop_end: and r2, r6, #PDRAW_WND_DIFF_PRIO ldmfd sp!, {r4-r11,lr} ldr r0, [sp] ldr r1, [r0, #OFS_rendstatus] orr r1, r1, r2 str r1, [r0, #OFS_rendstatus] bx lr @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ hilights 2 pixels in RGB444/BGR444 format .macro TileDoShHi2Pixels444 reg mov \reg, \reg, ror #12 adds \reg, \reg, #0x40000000 orrcs \reg, \reg, #0xf0000000 mov \reg, \reg, ror #28 adds \reg, \reg, #0x40000000 orrcs \reg, \reg, #0xf0000000 mov \reg, \reg, ror #28 adds \reg, \reg, #0x40000000 orrcs \reg, \reg, #0xf0000000 mov \reg, \reg, ror #24 adds \reg, \reg, #0x40000000 orrcs \reg, \reg, #0xf0000000 mov \reg, \reg, ror #28 adds \reg, \reg, #0x40000000 orrcs \reg, \reg, #0xf0000000 mov \reg, \reg, ror #28 adds \reg, \reg, #0x40000000 orrcs \reg, \reg, #0xf0000000 mov \reg, \reg, ror #12 .endm @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ Convert 0000bbb0 ggg0rrr0 @ to rrrrrggg gggbbbbb @ r2,r3 - scratch, lr = 0x001c001c, r8 = 0x08610861 .macro convRGB565 reg and r2, lr, \reg,lsr #7 @ b and r3, lr, \reg,lsr #3 @ g and \reg, lr, \reg,lsl #1 @ r orr r2, r2, r3, lsl #6 orr \reg, r2, \reg,lsl #11 and r2, r8, \reg,lsr #4 orr \reg, \reg, r2 .endm @ trashes: r2-r8,r12,lr; r8 = 0x08610861; r0,r1 are advanced .macro vidConvCpyRGB565_local mov r12, r2, lsr #3 @ repeats mov lr, #0x001c0000 orr lr, lr, #0x01c @ lr == pattern 0x001c001c 0: ldmia r1!, {r4-r7} subs r12, r12, #1 convRGB565 r4 str r4, [r0], #4 convRGB565 r5 str r5, [r0], #4 convRGB565 r6 str r6, [r0], #4 convRGB565 r7 str r7, [r0], #4 bgt 0b .endm .global vidConvCpyRGB565 vidConvCpyRGB565: @ void *to, void *from, int pixels stmfd sp!, {r4-r9,lr} mov r8, #0x0061 orr r8, r8, #0x0800 orr r8, r8, r8, lsl #16 vidConvCpyRGB565_local ldmfd sp!, {r4-r9,lr} bx lr @ void PicoDoHighPal555(int sh, int line, struct PicoEState *est) .global PicoDoHighPal555 PicoDoHighPal555: stmfd sp!, {r4-r10,lr} mov r10,r2 @ est mov r1, #0 ldr r8, [r10, #OFS_Pico_video] PicoDoHighPal555_nopush: orr r9, r1, r0, lsl #31 @ 0:called from FinalizeLine555, 31: s/h add r0, r10, #OFS_HighPal mov r1, #0 strb r1, [r8, #-0x1a] @ 0x2220e ~ dirtyPal sub r1, r8, #0x128 @ r1=Pico.cram mov r2, #0x40 mov r8, #0x0061 orr r8, r8, #0x0800 orr r8, r8, r8, lsl #16 vidConvCpyRGB565_local tst r9, #(1<<31) beq PicoDoHighPal555_end add r3, r10, #OFS_HighPal @ shadowed pixels: mov r12, #0x008e add r4, r3, #0x40*2 orr r12,r12,#0x7300 add r5, r3, #0xc0*2 orr r12,r12,r12,lsl #16 mov lr, #0x40/4 .fl_loopcpRGB555_sh: ldmia r3!, {r1,r6} subs lr, lr, #1 and r1, r12, r1, lsr #1 and r6, r12, r6, lsr #1 stmia r4!, {r1,r6} stmia r5!, {r1,r6} bne .fl_loopcpRGB555_sh @ hilighted pixels: @ t = ((dpal[i] >> 1) & 0x738e738e) + 0x738e738e; @ t |= (t >> 4) & 0x08610861; @ r8=0x08610861 sub r3, r3, #0x40*2 mov lr, #0x40/4 .fl_loopcpRGB555_hi: ldmia r3!, {r1,r6} and r1, r12, r1, lsr #1 and r6, r12, r6, lsr #1 add r1, r12, r1 add r6, r12, r6 and r5, r8, r1, lsr #4 and r7, r8, r6, lsr #4 orr r1, r1, r5 orr r6, r6, r7 stmia r4!, {r1,r6} subs lr, lr, #1 bne .fl_loopcpRGB555_hi mov r0, #1 PicoDoHighPal555_end: tst r9, #1 ldmeqfd sp!, {r4-r10,pc} ldr r8, [r10, #OFS_Pico_video] b FinalizeLineRGB555_pal_done @ void FinalizeLine555(int sh, int line, struct PicoEState *est) .global FinalizeLine555 FinalizeLine555: stmfd sp!, {r4-r10,lr} mov r10,r2 @ est ldr r8, [r10, #OFS_Pico_video] ldrb r2, [r8, #-0x1a] @ 0x2220e ~ dirtyPal mov r1, #1 tst r2, r2 bne PicoDoHighPal555_nopush FinalizeLineRGB555_pal_done: add r3, r10, #OFS_HighPal ldr r12, [r10, #OFS_rendstatus] eors r0, r0, #1 @ sh is 0 mov lr, #0xff tstne r12,#PDRAW_ACC_SPRITES movne lr, #0x3f ldr r1, [r10, #OFS_HighCol] ldr r0, [r10, #OFS_DrawLineDest] add r1, r1, #8 ldrb r12, [r8, #12] mov lr, lr, lsl #1 tst r12, #1 movne r2, #320/8 @ len bne .fl_no32colRGB555 ldr r4, [r10, #OFS_PicoOpt] mov r2, #256/8 ldr r4, [r4] tst r4, #0x4000 bne .fl_32scale_RGB555 tst r4, #0x0100 addeq r0, r0, #32*2 .fl_no32colRGB555: #ifdef 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 and r4, lr, r12, lsl #1 ldrh r4, [r3, r4] and r5, lr, r12, lsr #7 ldrh r5, [r3, r5] and r6, lr, r12, lsr #15 ldrh r6, [r3, r6] orr r4, r4, r5, lsl #16 and r5, lr, r12, lsr #23 ldrh r5, [r3, r5] and r8, lr, r7, lsl #1 ldrh r8, [r3, r8] orr r5, r6, r5, lsl #16 and r6, lr, r7, lsr #7 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 subs r2, r2, #1 orr r12,r12, r7, lsl #16 stmia r0!, {r4,r5,r8,r12} bne .fl_loopRGB555 ldmfd sp!, {r4-r10,lr} bx lr .fl_32scale_RGB555: mov r9, #0x3900 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007 orr r9, r9, #0x00e7 #ifdef UNALIGNED_DRAWLINEDEST tst r0, #2 bne .fl_32scale_RGB555u #endif .fl_loop32scale_RGB555: ldr r12, [r1], #4 ldr r7, [r1], #4 and r4, lr, r12,lsl #1 ldrh r4, [r3, r4] and r5, lr, r12,lsr #7 ldrh r5, [r3, r5] and r4, r4, r9, lsl #2 orr r4, r4, r4, lsl #14 @ r4[31:16] = 1/4 pix_s 0 and r5, r5, r9, lsl #2 sub r6, r5, r5, lsr #2 @ r6 = 3/4 pix_s 1 add r4, r4, r6, lsl #16 @ pix_d 0, 1 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 mov r5, r5, lsr #1 sub r6, r6, r6, lsr #2 @ r6 = 3/4 pix_s 2 orr r5, r5, r6, lsl #16 and r6, lr, r7, lsl #1 ldrh r6, [r3, r6] and r12,r12,r9, lsl #2 add r5, r5, r12,lsl #14 @ pix_d 2, 3 and r6, r6, r9, lsl #2 orr r6, r12,r6, lsl #16 @ pix_d 4, 5 and r12,lr, r7, lsr #7 ldrh r12,[r3, r12] and r10,lr, r7, lsr #15 ldrh r10,[r3, r10] and r12,r12,r9, lsl #2 sub r8, r12,r12,lsr #2 @ r8 = 3/4 pix_s 1 add r8, r8, r6, lsr #18 and r7, lr, r7, lsr #23 ldrh r7, [r3, r7] and r10,r10,r9, lsl #2 orr r8, r8, r10,lsl #15 add r8, r8, r12,lsl #15 @ pix_d 6, 7 sub r10,r10,r10,lsr #2 @ r10= 3/4 pix_s 2 and r7, r7, r9, lsl #2 add r10,r10,r7, lsr #2 @ += 1/4 pix_s 3 orr r10,r10,r7, lsl #16 @ pix_d 8, 9 subs r2, r2, #1 stmia r0!, {r4,r5,r6,r8,r10} bne .fl_loop32scale_RGB555 ldmfd sp!, {r4-r10,lr} bx lr #ifdef 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-r10,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!, {r4-r10,lr} bx lr #endif /* UNALIGNED_DRAWLINEDEST */ @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ utility .global blockcpy @ void *dst, void *src, size_t n blockcpy: stmfd sp!, {r4,r5} mov r2, r2, lsr #4 blockcpy_loop: ldmia r1!, {r3-r5,r12} subs r2, r2, #1 stmia r0!, {r3-r5,r12} bne blockcpy_loop ldmfd sp!, {r4,r5} bx lr .global blockcpy_or @ void *dst, void *src, size_t n, int pat blockcpy_or: stmfd sp!, {r4-r6} orr r3, r3, r3, lsl #8 orr r3, r3, r3, lsl #16 mov r2, r2, lsr #4 blockcpy_loop_or: ldmia r1!, {r4-r6,r12} subs r2, r2, #1 orr r4, r4, r3 orr r5, r5, r3 orr r6, r6, r3 orr r12,r12,r3 stmia r0!, {r4-r6,r12} bne blockcpy_loop_or ldmfd sp!, {r4-r6} bx lr @ vim:filetype=armasm