/*
 * This's for exa composite operation in no mask picture case.
 * The simplest case is just sending what src picture has to dst picture.
 * XXX: This is still experimental, and should be fixed to support multiple texture
 * map, and conditional mul actions. 
 */

/* I think this should be same as in g4a program for texture video,
   as we also use 16-pixel dispatch. and SF scale in g3 is useful for us. */

/* The initial payload of the thread is always g0.
 * WM_URB (incoming URB entries) is g3
   As mask texture coeffient needs extra setup urb starting from g4, we should
   shift this location. 

 * X0_R is g4->g6
 * X1_R is g5->g7
 * Y0_R is g6->g8
 * Y1_R is g7->g9

     * X0: {ss0.x, ss0.x+1, ss0.x,   ss0.x+1, ss1.x, ss1.x+1, ss1.x,   ss1.x+y}
     * Y0: {ss0.y, ss0.y,   ss0.y+1, ss0.y+1, ss1.y, ss1.y,   ss1.y+1, ss1.y+1}
     * X1: {ss2.x, ss2.x+1, ss2.x,   ss2.x+1, ss3.x, ss3.x+1, ss3.x,   ss3.x+y}
     * Y1: {ss2.y, ss2.y,   ss2.y+1, ss2.y+1, ss3.y, ss3.y,   ss3.y+1, ss3.y+1}
 */

/* multitexture program with src and mask texture */
/* - load src texture */
/* - load mask texture */
/* - mul src.X with mask's alpha */
/* - write out src.X */

    /* Set up ss0.x coordinates*/
mov (1) g6<1>F g1.8<0,1,0>UW { align1 };
add (1) g6.4<1>F g1.8<0,1,0>UW 1UB { align1 };
mov (1) g6.8<1>F g1.8<0,1,0>UW { align1 };
add (1) g6.12<1>F g1.8<0,1,0>UW 1UB { align1 };
    /* Set up ss0.y coordinates */
mov (1) g8<1>F g1.10<0,1,0>UW { align1 };
mov (1) g8.4<1>F g1.10<0,1,0>UW { align1 };
add (1) g8.8<1>F g1.10<0,1,0>UW 1UB { align1 };
add (1) g8.12<1>F g1.10<0,1,0>UW 1UB { align1 };
    /* set up ss1.x coordinates */
mov (1) g6.16<1>F g1.12<0,1,0>UW { align1 };
add (1) g6.20<1>F g1.12<0,1,0>UW 1UB { align1 };
mov (1) g6.24<1>F g1.12<0,1,0>UW { align1 };
add (1) g6.28<1>F g1.12<0,1,0>UW 1UB { align1 };
    /* set up ss1.y coordinates */
mov (1) g8.16<1>F g1.14<0,1,0>UW { align1 };
mov (1) g8.20<1>F g1.14<0,1,0>UW { align1 };
add (1) g8.24<1>F g1.14<0,1,0>UW 1UB { align1 };
add (1) g8.28<1>F g1.14<0,1,0>UW 1UB { align1 };
    /* Set up ss2.x coordinates */
mov (1) g7<1>F g1.16<0,1,0>UW { align1 };
add (1) g7.4<1>F g1.16<0,1,0>UW 1UB { align1 };
mov (1) g7.8<1>F g1.16<0,1,0>UW { align1 };
add (1) g7.12<1>F g1.16<0,1,0>UW 1UB { align1 };
    /* Set up ss2.y coordinates */
mov (1) g9<1>F g1.18<0,1,0>UW { align1 };
mov (1) g9.4<1>F g1.18<0,1,0>UW { align1 };
add (1) g9.8<1>F g1.18<0,1,0>UW 1UB { align1 };
add (1) g9.12<1>F g1.18<0,1,0>UW 1UB { align1 };
    /* Set up ss3.x coordinates */
mov (1) g7.16<1>F g1.20<0,1,0>UW { align1 };
add (1) g7.20<1>F g1.20<0,1,0>UW 1UB { align1 };
mov (1) g7.24<1>F g1.20<0,1,0>UW { align1 };
add (1) g7.28<1>F g1.20<0,1,0>UW 1UB { align1 };
    /* Set up ss3.y coordinates */
mov (1) g9.16<1>F g1.22<0,1,0>UW { align1 };
mov (1) g9.20<1>F g1.22<0,1,0>UW { align1 };
add (1) g9.24<1>F g1.22<0,1,0>UW 1UB { align1 };
add (1) g9.28<1>F g1.22<0,1,0>UW 1UB { align1 };

    /* Now, map these screen space coordinates into texture coordinates. */
/* This is for src texture */
/* I don't want to change origin ssX coords, as it will be used later in mask */
/* so store tex coords in g10, g11, g12, g13 */

    /* subtract screen-space X origin of vertex 0. */
add (8) g10<1>F g6<8,8,1>F -g1<0,1,0>F { align1 };
add (8) g11<1>F g7<8,8,1>F -g1<0,1,0>F { align1 };
    /* scale by texture X increment */
/* Cx[0] */
mul (8) g10<1>F g10<8,8,1>F g3<0,1,0>F { align1 };
mul (8) g11<1>F g11<8,8,1>F g3<0,1,0>F { align1 };
    /* add in texture X offset */
/* Co[0] */
add (8) g10<1>F g10<8,8,1>F g3.12<0,1,0>F { align1 };
add (8) g11<1>F g11<8,8,1>F g3.12<0,1,0>F { align1 };
    /* subtract screen-space Y origin of vertex 0. */
add (8) g12<1>F g8<8,8,1>F -g1.4<0,1,0>F { align1 };
add (8) g13<1>F g9<8,8,1>F -g1.4<0,1,0>F { align1 };
    /* scale by texture Y increment */
/* Cy[0] */
mul (8) g12<1>F g12<8,8,1>F g3.4<0,1,0>F { align1 };
mul (8) g13<1>F g13<8,8,1>F g3.4<0,1,0>F { align1 };
    /* add in texture Y offset */
/* Co[1] */
add (8) g12<1>F g12<8,8,1>F g3.28<0,1,0>F { align1 };
add (8) g13<1>F g13<8,8,1>F g3.28<0,1,0>F { align1 };

/* prepare sampler read back gX register, which would be written back to output */

/* use simd16 sampler, param 0 is u, param 1 is v. */
/* 'payload' loading, assuming tex coord start from g4 */
mov (8) m1<1>F g10<8,8,1>F { align1 };
mov (8) m2<1>F g11<8,8,1>F { align1 }; /* param 0 u in m1, m2 */
mov (8) m3<1>F g12<8,8,1>F { align1 };
mov (8) m4<1>F g13<8,8,1>F { align1 }; /* param 1 v in m3, m4 */

/* m0 will be copied with g0, as it contains send desc */
/* emit sampler 'send' cmd */

/* src texture readback: g14-g21 */
send (16) 0 		/* msg reg index */
	g14<1>UW 	/* readback */
	g0<8,8,1>UW  	/* copy to msg start reg*/
	sampler (1,0,F)  /* sampler message description, 
				(binding_table,sampler_index,datatype). 
			    here(src->dst) we should use src_sampler and 
			    src_surface */
	mlen 5 rlen 8 { align1 };   /* required message len 5, readback len 8 */

mov (8) g21<1>UD g21<8,8,1>UD { align1 };  /* wait sampler return */

/* sampler mask texture, use g10, g11, g12, g13 */
    /* subtract screen-space X origin of vertex 0. */
add (8) g10<1>F g6<8,8,1>F -g1<0,1,0>F { align1 };
add (8) g11<1>F g7<8,8,1>F -g1<0,1,0>F { align1 };
    /* scale by texture X increment */
/* Cx[2] */
mul (8) g10<1>F g10<8,8,1>F g4<0,1,0>F { align1 };
mul (8) g11<1>F g11<8,8,1>F g4<0,1,0>F { align1 };
    /* add in texture X offset */
/* Co[2] */
add (8) g10<1>F g10<8,8,1>F g4.12<0,1,0>F { align1 };
add (8) g11<1>F g11<8,8,1>F g4.12<0,1,0>F { align1 };
    /* subtract screen-space Y origin of vertex 0. */
add (8) g12<1>F g8<8,8,1>F -g1.4<0,1,0>F { align1 };
add (8) g13<1>F g9<8,8,1>F -g1.4<0,1,0>F { align1 };
    /* scale by texture Y increment */
/* Cy[2] */
mul (8) g12<1>F g12<8,8,1>F g4.4<0,1,0>F { align1 };
mul (8) g13<1>F g13<8,8,1>F g4.4<0,1,0>F { align1 };
    /* add in texture Y offset */
/* Co[3] */
add (8) g12<1>F g12<8,8,1>F g4.28<0,1,0>F { align1 };
add (8) g13<1>F g13<8,8,1>F g4.28<0,1,0>F { align1 };

mov (8) m1<1>F g10<8,8,1>F { align1 };
mov (8) m2<1>F g11<8,8,1>F { align1 }; 
mov (8) m3<1>F g12<8,8,1>F { align1 };
mov (8) m4<1>F g13<8,8,1>F { align1 };

/* mask sampler g22-g29 */
/* binding_table (2), sampler (1) */
send (16) 0 g22<1>UW g0<8,8,1>UW sampler (2,1,F) mlen 5 rlen 8 { align1 };
mov (8) g29<1>UD g29<8,8,1>UD { align1 };  /* wait sampler return */

/* mul mask's channel to src, then write out src */
mul (8) g14<1>F g14<8,8,1>F g22<8,8,1>F { align1 };
mul (8) g15<1>F g15<8,8,1>F g23<8,8,1>F { align1 };
mul (8) g16<1>F g16<8,8,1>F g24<8,8,1>F { align1 };
mul (8) g17<1>F g17<8,8,1>F g25<8,8,1>F { align1 };
mul (8) g18<1>F g18<8,8,1>F g26<8,8,1>F { align1 };
mul (8) g19<1>F g19<8,8,1>F g27<8,8,1>F { align1 };
mul (8) g20<1>F g20<8,8,1>F g28<8,8,1>F { align1 };
mul (8) g21<1>F g21<8,8,1>F g29<8,8,1>F { align1 };

/* prepare data in m2-m5 for subspan(1,0), m6-m9 for subspan(3,2), then it's ready to write */
mov (8) m2<1>F g14<8,8,1>F { align1 };
mov (8) m3<1>F g16<8,8,1>F { align1 };
mov (8) m4<1>F g18<8,8,1>F { align1 };
mov (8) m5<1>F g20<8,8,1>F { align1 };
mov (8) m6<1>F g15<8,8,1>F { align1 };
mov (8) m7<1>F g17<8,8,1>F { align1 };
mov (8) m8<1>F g19<8,8,1>F { align1 };
mov (8) m9<1>F g21<8,8,1>F { align1 };

/* m0, m1 are all direct passed by PS thread payload */
mov (8) m1<1>UD g1<8,8,1>UD { align1 mask_disable };

/* write */
send (16) 0 acc0<1>UW g0<8,8,1>UW write (
	0,  /* binding_table */
	8,  /* pixel scordboard clear, msg type simd16 single source */
	4,  /* render target write */
	0   /* no write commit message */
	) 
	mlen 10
	rlen 0
	{ align1 EOT };

nop;
nop;
nop;
nop;
nop;
nop;
nop;
nop;
nop;