Final ideation before finish updates lua metaprogram before hello-cdrom

code/duffle/dsl.atom.h

@@ -160,7 +160,7 @@
  * ----------------------------------------------------------------------------*/
 #define atom_bind(binds_struct)  /* atom_bind(binds_struct) */
 
-#define Binds_(type) (tmpl(Binds,type) // TODO(Ed): Do we want to use this?
+#define Binds_(type) (tmpl(Binds,type)) // TODO(Ed): Do we want to use this?
 
 /* ============================================================================
  * atom_label / atom_offset — branch target machinery
@@ -179,9 +179,3 @@
 #define atom_offset(F, T) atom_offset_ ## F ## _ ## T
 // atom_label is a pure annotation for the metaprogram's offset calculations.
 #define atom_label(name) /* atom_label anchor: name */
-
-// WIP: Atoms Assocated closely with each other to form a tape procedure. (Maybe also a phase in a procedure/pipeline?)
-
-#define AtomBundle_(name)              Struct_(tmpl(AtomBundle,name))
-#define AtomBundle_Len(name)           S_(tmpl(AtomBundle,name))/S_(MipsAtom*)
-#define AtomBundleEntry_(bundle,entry) tmpl(bundle,entry)

code/duffle/gen/macs.h

@@ -228,21 +228,21 @@ WORD_COUNT(mac_gte_op_cross_v3s4, 13)
 WORD_COUNT(mac_gte_store_f3, 3)
 
 /* atom_dbg_skip */
-#define mac_gte_load_tri_verts(r_vert_base, r_v0, r_v1, r_v2) \
+#define mac_gte_load_tri_verts(vbase, v0, v1, v2) \
-	shift_lleft(R_AT, r_v0, v3s2_byteoff) \
+	shift_lleft(R_AT, v0, v3s2_byteoff) \
-,	add_u_self(R_AT, r_vert_base) \
+,	add_u_self(R_AT, vbase) \
 ,	load_word(R_V0, R_AT, O_(V3_S2,x)) \
 ,	load_word(R_V1, R_AT, O_(V3_S2,z)) \
 ,	LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY0) \
 ,	gte_mv_to_data_r(R_V1, C2_VZ0) \
-,	shift_lleft(R_AT, r_v1, v3s2_byteoff) \
+,	shift_lleft(R_AT, v1, v3s2_byteoff) \
-,	add_u_self(R_AT, r_vert_base) \
+,	add_u_self(R_AT, vbase) \
 ,	load_word(R_V0, R_AT, O_(V3_S2,x)) \
 ,	load_word(R_V1, R_AT, O_(V3_S2,z)) \
 ,	LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY1) \
 ,	gte_mv_to_data_r(R_V1, C2_VZ1) \
-,	shift_lleft(R_AT, r_v2, v3s2_byteoff) \
+,	shift_lleft(R_AT, v2, v3s2_byteoff) \
-,	add_u_self(R_AT, r_vert_base) \
+,	add_u_self(R_AT, vbase) \
 ,	load_word(R_V0, R_AT, O_(V3_S2,x)) \
 ,	load_word(R_V1, R_AT, O_(V3_S2,z)) \
 ,	LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY2) \

code/duffle/gte.atom.c

@@ -63,10 +63,10 @@ FI_ Slice_MipsCode ac_gte_store_f3(AtomBuilder_R ab, U4 r_primitive_cursor) atom
 })
 
 /* Words: 18; Translates indices to vertex addresses and pushes them to GTE  */
-I_ Slice_MipsCode ac_gte_load_tri_verts(AtomBuilder_R ab, U4 r_vert_base, U4 r_v0, U4 r_v1, U4 r_v2) atom_dbg_skip MipsAtomComp_Proc_(ab, {
+I_ Slice_MipsCode ac_gte_load_tri_verts(AtomBuilder_R ab, Reg vbase, Reg v0, Reg v1, Reg v2) atom_dbg_skip MipsAtomComp_Proc_(ab, {
-	shift_lleft(R_AT, r_v0, v3s2_byteoff), add_u_self(R_AT, r_vert_base), load_word(R_V0, R_AT, O_(V3_S2,x)), load_word(R_V1, R_AT, O_(V3_S2,z)), LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY0), gte_mv_to_data_r(R_V1, C2_VZ0),
+	shift_lleft(R_AT, v0, v3s2_byteoff), add_u_self(R_AT, vbase), load_word(R_V0, R_AT, O_(V3_S2,x)), load_word(R_V1, R_AT, O_(V3_S2,z)), LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY0), gte_mv_to_data_r(R_V1, C2_VZ0),
-	shift_lleft(R_AT, r_v1, v3s2_byteoff), add_u_self(R_AT, r_vert_base), load_word(R_V0, R_AT, O_(V3_S2,x)), load_word(R_V1, R_AT, O_(V3_S2,z)), LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY1), gte_mv_to_data_r(R_V1, C2_VZ1),
+	shift_lleft(R_AT, v1, v3s2_byteoff), add_u_self(R_AT, vbase), load_word(R_V0, R_AT, O_(V3_S2,x)), load_word(R_V1, R_AT, O_(V3_S2,z)), LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY1), gte_mv_to_data_r(R_V1, C2_VZ1),
-	shift_lleft(R_AT, r_v2, v3s2_byteoff), add_u_self(R_AT, r_vert_base), load_word(R_V0, R_AT, O_(V3_S2,x)), load_word(R_V1, R_AT, O_(V3_S2,z)), LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY2), gte_mv_to_data_r(R_V1, C2_VZ2),
+	shift_lleft(R_AT, v2, v3s2_byteoff), add_u_self(R_AT, vbase), load_word(R_V0, R_AT, O_(V3_S2,x)), load_word(R_V1, R_AT, O_(V3_S2,z)), LdSlot_ gte_mv_to_data_r(R_V0, C2_VXY2), gte_mv_to_data_r(R_V1, C2_VZ2),
 })
 
 /* Words: 3; Stores the 3 transformed (V2_S2 screen) vertices of the
@@ -267,11 +267,11 @@ internal S2 const gte_normalize_sqr_tbl[192] align_(2) = {
 	0x0820, 0x081c, 0x0818, 0x0814, 0x0810, 0x080c, 0x0808, 0x0804,
 };
 
-typedef Struct_(Binds_NormalizeV3S4) {
+typedef Struct_(Binds_normalize_v3s4) {
 	U2 src_offset;    /* offset of src V3_S4 within the BIOS scratchpad */
 	U2 dst_offset;    /* offset of dst V3_S4 within the BIOS scratchpad */
 };
-typedef Struct_(RegUse_build_normalize_v3s4) {
+typedef Struct_(RegUse_normalize_v3s4) {
 	union { Reg_(V3_S4) res, src; };
 	union { Reg r0, src_ptr, mac2; };
 	union { Reg r1, dst_ptr; };
@@ -282,13 +282,13 @@ typedef Struct_(RegUse_build_normalize_v3s4) {
 };
 /* ─── Full normalize (all 4 stages inline) ───
  * Generic 4-stage GTE normalize (SQR → sum+LZCR → align+sqrtbl → GPF+srav). */
-internal MipsAtom* build_normalize_v3s4(AtomArena_R aa, RegUse_build_normalize_v3s4 r)
+internal MipsAtom* normalize_v3s4(AtomArena_R aa, RegUse_normalize_v3s4 r)
 MipsAtom_Proc_(aa, {
-	load_half(r.src_offset, R_TapePtr, O_(Binds_NormalizeV3S4, src_offset)),
+	load_half(r.src_offset, R_TapePtr, O_(Binds_normalize_v3s4, src_offset)),
-	load_half(r.dst_offset, R_TapePtr, O_(Binds_NormalizeV3S4, dst_offset)),
+	load_half(r.dst_offset, R_TapePtr, O_(Binds_normalize_v3s4, dst_offset)),
 	LdSlot_ add_u(r.src_ptr, R_ScratchBase, r.src_offset),
 	LdSlot_ add_u(r.dst_ptr, R_ScratchBase, r.dst_offset),
-	LdSlot_ add_ui_self(R_TapePtr, S_(Binds_NormalizeV3S4)),
+	LdSlot_ add_ui_self(R_TapePtr, S_(Binds_normalize_v3s4)),
 
 	mac_load_v3s4(r.src, r.src_ptr, 0),
 

code/duffle/lottes_tape.h

@@ -163,6 +163,12 @@ typedef Slice_(MipsAtom);
 // Used for trivial mappings from one atom component proc to the command of a more baser (meant for type-mapping)
 #define MipsAtomComp_ProcMap_(ab, base_command) atom_dbg_skip MipsAtomComp_Proc_(ab, {base_command })
 
+// WIP: Atoms Assocated closely with each other to form a tape procedure. (Maybe also a phase in a procedure/pipeline?)
+
+#define AtomBundle_(name)              Struct_(tmpl(AtomBundle,name))
+#define AtomBundle_Len(name)           S_(tmpl(AtomBundle,name))/S_(MipsAtom*)
+#define AtomBundleEntry_(bundle,entry) tmpl(bundle,entry)
+
 /* Line-table anchor: gcc only adds a file to the .debug_line file table when the contains line-numbered content.
 	Files containing only atoms and atom components.
    Place `ATOM_FILE_LINE_MARKER();` once at file scope in any `.atom.c` that defines atoms.
@@ -247,6 +253,9 @@ FI_ void tb_data(TapeBuilder* tb, U4        data) { u4_r(tb->ptr)[tb->used] = u4
 FI_ void tb_bind(TapeBuilder* tb, Slice data) { mem_copy(tb->ptr + tb->used * S_(MipsCode), u4_(data.ptr), data.len); tb->used += data.len / S_(MipsCode); }
 #define tb_bind_(tb,type,...) tb_bind(tb, (Slice){ (B1*)(& (type){__VA_ARGS__}), S_(type) }); static_assert(S_(type) % S_(MipsCode) == 0)
 
+#define tb_emit_wbind_(tb,atom,...)       tb_emit(tb,atom); tb_bind_(tb,tmpl(Binds,atom),__VA_ARGS__)
+#define tb_emit_wbind2_(tb,atom,type,...) tb_emit(tb,atom); tb_bind_(tb,type,__VA_ARGS__)
+
 FI_ Tape tb_end  (TapeBuilder* tb) { tb_emit(tb,tape_exit); return (Tape){ C_(U4*,tb->ptr), tb->used }; }
 FI_ Tape tb_slice(TapeBuilder  tb) {                        return (Tape){ C_(U4*,tb.ptr),  tb.used }; }
 #define tb_scope(tb)     for(U4 tbs_once=0;tbs_once==0;++tbs_once,tb_emit(tb,tape_exit))

code/duffle/pad.h

@@ -4,10 +4,7 @@
 #	include "math.h"
 #endif
 
-/* PSX button bit positions — 1:1 with PSX-SPX docs at docs/psx-spx/docs/controllersandmemorycards.md:405-421.
+// PSX button bit positions: PSX-SPX docs/psx-spx/docs/controllersandmemorycards.md:405-421.
- * Wire is active-low (0 = pressed).
- * The decoder atom computes buttons = (~raw_buttons) & 0xFFFF; 
- * active-low-to-active-high inversion is applied bit-by-bit. */
 typedef Enum_(U2, PadBtns) {
 	Bit_(Pad_Select,    0),
 	Bit_(Pad_L3,        1),
@@ -62,11 +59,7 @@ typedef Enum_(U4, PadStatus) {
 	PadStatus_Invalid,
 };
 
-/* Distinct from the game-facing PadStatus enum: PadRawStatus_Ok and PadRawStatus_Timeout are raw BIOS values;
+// Distinct from the game-facing PadStatus enum: PadRawStatus_Ok and PadRawStatus_Timeout are raw BIOS values
- * PadStatus_* are game-facing post-decode states. PadUnknownId_Sentinel is written by the decoder 
- * when the controller id does not match any known controller type.
- * PadAxisCentered_Word: Four-byte 0x80 pattern used to clear / center 
- * four byte axes at PadState.left_x through PadState.right_y. */
 typedef Enum_(U1, PadRawStatus) {
 	PadRawStatus_Ok      = 0x00,
 	PadRawStatus_Timeout = 0xFF,

code/duffle/psyq.h

@@ -104,7 +104,7 @@ void gte_matrix_set_translation(MT3_S2S4* mat) asm("SetTransMatrix");
 
 // Einheit, Metrication to unit vector. "Normalization", not Orthogonal "Normal, Normalis". Directionalization.
 // RGA(Lengyel): Normalize the bulk of a zero-weight direction. This is not finite-point unitization (which forces w=1).
-S4 normalize_v3s4(V3_S4* v0, V3_S4* v1) asm("VectorNormal");
+S4 psy_normalize_v3s4(V3_S4* v0, V3_S4* v1) asm("VectorNormal");
 
 // RGA(Lengyel): Apply the matrix expansion of a rigid transformation.
 // Motor antiproduct is equivalent for unitized points; LA form is what GTE consumes.

code/hello_camera/hello_camera.atom.c

@@ -155,7 +155,7 @@ atom_info(atom_bind(Binds_ResolveLookAtSub)) MipsAtom_Proc_(aa, {
 })
 
 typedef Struct_(Binds_ResolveLookAt_PopulateMT3S4S2) {
-	U4 look_at;  /* MT3_S2S4* — destination matrix address */
+	MT3_S2S4* look_at;  /* MT3_S2S4* — destination matrix address */
 };
 typedef Struct_(RegUse_resolve_look_at_populate_mt3s4s2) {
 	Reg         look_at;

code/hello_camera/hello_camera.c

@@ -80,9 +80,6 @@ typedef Struct_(SMemory) {
 	PadBiosRaw pad_raw[2];
 	PadState   pad[2];
 
-	// TODO(Ed): We don't need this we can just cast at any point an address to a desired view of scratchpad, we have the address.
-	U4_V scratchpad; // d-cache
-
 	U1 ct_init_atom_mem[CT_InitAtomMem_Size];
 	MipsAtom* normalize_v3s4;
 	MipsAtom* gte_cross_v3s4;
@@ -114,7 +111,7 @@ I_ void resolve_look_at_c11(MT3_S2S4* look_at, P3_S4* eye, P3_S4* target, V3_S4*
 	V3_S4 pos, off;
 
 	forward = target[0]; sub_v3s4(& forward, eye[0]); // RGA(Lengyel): Affine point - point = zero-weight direction.
-	normalize_v3s4(& forward, & uz);                  // RGA(Lengyel): Normalize the direction bulk. Not finite-point unitization.
+	psy_normalize_v3s4(& forward, & uz);              // RGA(Lengyel): Normalize the direction bulk. Not finite-point unitization.
 
 	cross_v3s4(& uz,   up_in, & right); normalize_v3s4(& right, & ux); // RGA(Lengyel): Complement(Wedge(forward, up_in)) -> right axis.
 	cross_v3s4(& uz, & ux,    & up);    normalize_v3s4(& up,    & uy); // RGA(Lengyel): Complement(Wedge(forward, right)) -> up axis.
@@ -149,8 +146,8 @@ internal void compile_init_atoms(void) {
 		});
 	regfile_reset(& rf);
 
-	smem.normalize_v3s4 = build_normalize_v3s4(& ab,
+	smem.normalize_v3s4 = normalize_v3s4(& ab,
-		RegUse_(build_normalize_v3s4) {
+		RegUse_(normalize_v3s4) {
 			.res = ralloc_v3(),
 			.r0     = ralloc(),
 			.r1     = ralloc(),
@@ -208,37 +205,36 @@ I_ void resolve_look_at(TapeBuilder_R tb,	MT3_S2S4* look_at, P3_S4* eye, P3_S4*
 	/* Typed view of the scratchpad for field-address arithmetic. */
 	ResolveLookAtScratch*        sp     = C_scratch(ResolveLookAtScratch*);
 	AtomBundle_resolve_look_at_R bundle = C_(void*, smem.resolve_look_at_bundle);
-	tb_emit(tb, bundle->input_and_sub); {
+	tb_emit(tb, bundle->input_and_sub); tb_bind_(tb, Binds_ResolveLookAtSub, 
-		tb_data(tb, u4_(target));
+		.target = target,
-		tb_data(tb, u4_(eye));
+		.eye    = eye,
-		tb_data(tb, u4_(up_in));
+		.up_in  = up_in,
-	}
+	);
-	tb_emit(tb, bundle->normalize_fwd_uz); {
+	tb_emit(tb, bundle->normalize_fwd_uz); tb_bind_(tb, Binds_normalize_v3s4,
-		// tb_data(tb, u4_(O_(ResolveLookAtScratch, fwd) | (O_(ResolveLookAtScratch, uz) << 16)));
+		.src_offset = O_(ResolveLookAtScratch,fwd),
-		tb_bind_(tb, Binds_NormalizeV3S4,
+		.dst_offset = O_(ResolveLookAtScratch,uz),
-				.src_offset = O_(ResolveLookAtScratch,fwd),
+	);
-				.dst_offset = O_(ResolveLookAtScratch,uz),
+	tb_emit(tb, bundle->cross_to_right); tb_bind_(tb, Binds_gte_cross_v3s4,
-		);
+		.src_a = & sp->uz,
-	}
+		.src_b = & sp->up_in,
-	tb_emit(tb, bundle->cross_to_right); {
+		.out   = & sp->right,
-		tb_data(tb, u4_(& sp->uz));
+	);
-		tb_data(tb, u4_(& sp->up_in));
+	tb_emit(tb, bundle->normalize_right_ux); tb_bind_(tb, Binds_normalize_v3s4, 
-		tb_data(tb, u4_(& sp->right));
+		.src_offset = O_(ResolveLookAtScratch,right),
-	}
+		.dst_offset = O_(ResolveLookAtScratch,ux),
-	tb_emit(tb, bundle->normalize_right_ux); {
+	);
-		tb_data(tb, u4_(O_(ResolveLookAtScratch, right) | (O_(ResolveLookAtScratch, ux) << 16)));
+	tb_emit(tb, bundle->cross_to_up); tb_bind_(tb, Binds_gte_cross_v3s4, 
-	}
+		.src_a = & sp->uz,
-	tb_emit(tb, bundle->cross_to_up); {
+		.src_b = & sp->ux,
-		tb_data(tb, u4_(& sp->uz));
+		.out   = & sp->up,
-		tb_data(tb, u4_(& sp->ux));
+	);
-		tb_data(tb, u4_(& sp->up));
+	tb_emit(tb, bundle->normalize_up_uy); tb_bind_(tb, Binds_normalize_v3s4,
-	}
+		.src_offset = O_(ResolveLookAtScratch,up),
-	tb_emit(tb, bundle->normalize_up_uy); {
+		.dst_offset = O_(ResolveLookAtScratch,uy),
-		tb_data(tb, u4_(O_(ResolveLookAtScratch, up) | (O_(ResolveLookAtScratch, uy) << 16)));
+	);
-	}
+	tb_emit(tb, bundle->populate_mt3s4s2); tb_bind_(tb, Binds_ResolveLookAt_PopulateMT3S4S2, 
-	tb_emit(tb, bundle->populate_mt3s4s2); {
+		.look_at = look_at,
-		tb_data(tb, u4_(look_at));
+	);
-	}
 }
 FI_ void camera_look_at(TapeBuilder_R tb, Camera* c, P3_S4* target, V3_S4* up_in) { resolve_look_at(tb, & c->look_at, & c->pos, target, up_in); }
 
@@ -397,8 +393,6 @@ GCC_OPTIMIZATION_DISABLE
 int main(void)
 {
 	smem = (SMemory){0};
-	// TODO(Ed): remove this field we don't need it in smem.
-	smem.scratchpad = C_(U4_V, Scratchpad_Loc);
 	// smem.primitives.used = 0;
 	// smem.active_buf_id   = 0;
 	smem.cam.pos = v3s4(500, -1000, -1500);