proc iselect
	save
	pop [gp1]
	pop [gp0]
	
	adjust [gp1] 0 ; test for zero in-place
	cond zero jmp $retz
	push 1 ; i
	push 0 ; t
	
	label $loop
	over
	push [gp1]
	bitand
	drop
	cond zero jmp $else
	over ; i t i
	push [gp0] ; i t i a
	bitand ; i t (i&a)
	bitor ; i (t|(i&a))
	over ; i t i
	push [gp1] ; i t i b
	bitxor
	pop [gp1]
	swap ; t i
	push 1 ; t i 1
	shll ; t 2i
	swap ; i t
	jmp $while
	
	label $else
	push [gp0]
	push 1
	shlr
	pop [gp0]
	push [gp1]
	push 1
	shlr
	pop [gp1]
	
	label $while
	push [gp1]
	popbool
	cond jmp $loop
	swap
	drop
	jmp $ret
	
	label $retz
	push 0
	label $ret
	restore
	set [arg] 1
	return ; r
end

proc iinterleave
	save
	
	; extract lower halfwords
	explode 2 ; a b_h b_l
	pop [gp1] ; a b_h
	swap ; b_h a
	explode 2 ; b_h a_h a_l
	pop [gp0] ; b_h a_h
	bitor ; the entire result is in the lower halfwords
	movesh ; stash the overflow value
	set [counter] 0 ; used for holding the shift amount
	set [arg] 1 ; holds the mask
	
	push 0 ; result
	
	label $loop
	
	bitand [gp1] [arg] ; res Bbit
	shl %P [counter] ; res Bbit<<counter
	bitor ; res |= Bbit<<counter
	
	adjust [counter] 1 ; increment the counter
	
	bitand [gp0] [arg] ; res Abit
	shl %P [counter] ; res Abit<<counter
	bitor ; res |= Abit<<counter
	
	adjust [arg] [arg] ; double the mask
	
	cond ovfl skip ; if the mask overflows, then all bits have been set
	jmp $loop ; return to start of loop
	
	movehs ; get the overflow value
	restore ; restore also clears the flags
	; shift the overflow off the end
	shl %P 32 ; shl %P 64 doesn't work
	shl %P 32
	; now ovfl is set if the high halfwords contained any set bits, or
	; unset if they did not, matching the specification of the
	; interleave instruction.
	set [arg] 1 ; only a single item is being returned
	return ; result is still on stack
end

; brainfuck macros

; 5 words
macro pushptr
	bitand [gp0] 8
	shll %P 3
	shll [index] 20
	select [gp0] 0FFFF8H
	bitor
end

; 1 word
macro bf<
	set [gp0] [gp0#-1]
end

; 1 word
macro bf>
	set [gp0] [gp0]+1
end

; 8 words
macro bf+
	pushptr
	load_ua [PSV] [SV]
	wadd 1
	store_ua %P %P
end

; 8 words
macro bf-
	pushptr
	load_ua [PSV] [SV]
	wadd -1
	store_ua %P %P
end

; 6+? words
macro bf.
	pushptr
	load_ua %P %P
	; output
end

macro writeEOF
.if OnEOF == 0
	; 7 words
	push [zero]
	pushptr
	store_ua %P %P
.elseif OnEOF == -1
	; 7 words
	push 255
	pushptr
	store_ua %P %P
.else OnEOF == 1 ; unchanged - no writeback
	; 0 words
.endif
end

; 11+[0,7]+? words
macro bf,
	; input
	dup
	cmpne EOF
	cond jmp $$EOF
	pushptr
	store_ua %P %P
	jmp $$end
$$EOF:
	drop
	; 7 or 0 words
	writeEOF
$$end:
end

; 8 words before
macro bfLB (here) (dest)
	pushptr
	load_ua %P %P
	popbool
	cond 0Eh jmp (dest)
	label (here)
end

; arguments inverted so that they are in the same order as the
; matching bfLB
; 8 words before
macro bfRB (dest) (here)
	pushptr
	load_ua %P %P
	popbool
	cond jmp (dest)
	label (here)
end

; For the BF code >>[-]<<[->>+<<]
; 64 words total
proc bfcode
	bf>
	bf>
	bfLB BL1 BR1
	bf-
	bfRB BL1 BR2
	bf<
	bf<
	bfLB BL2 BR2
	bf-
	bf>
	bf>
	bf+
	bf<
	bf<
	bfRB BL2 BR2
	ret
end


; brainfuck macros (64-bit words, with anding)

; 1 word
macro bf<
	set [gp0] [gp0#-1]
end

; 1 word
macro bf>
	set [gp0] [gp0]+1
end

; 4 words
macro bf+
	push *[gp0]
	wadd 1
	bitand 255
	pop *[gp0]
end

; 4 words
macro bf-
	push *[gp0]
	wsub 1
	bitand 255
	pop *[gp0]
end

; 1+? words
macro bf.
	push *[gp0]
	; output
end

macro writeEOF
.if OnEOF == 0
	; 1 word
	set *[gp0] [zero]
.elseif OnEOF == -1
	; 1 word
	set *[gp0] 255
.else OnEOF == 1 ; unchanged - no writeback
	; 0 words
.endif
end

; 6+[0,1]+? words
macro bf,
	; input
	dup
	cmpne EOF
	cond jmp $$EOF
	pop *[gp0]
	jmp $$end
$$EOF:
	drop
	; 1 or 0 words
	writeEOF
$$end:
end

; 3 words before
macro bfLB (here) (dest)
	push *[gp0]
	popbool
	cond 0Eh jmp (dest)
	label (here)
end

; arguments inverted so that they are in the same order as the
; matching bfLB
; 3 words before
macro bfRB (dest) (here)
	push *[gp0]
	popbool
	cond jmp (dest)
	label (here)
end

; For the BF code >>[-]<<[->>+<<]
; 32 words total
; could likely be hand-optimiaed after macro expansion
proc bfcode_compact
	bf>
	bf>
	bfLB BL1 BR1
	bf-
	bfRB BL1 BR2
	bf<
	bf<
	bfLB BL2 BR2
	bf-
	bf>
	bf>
	bf+
	bf<
	bf<
	bfRB BL2 BR2
	ret
end

; 32 words
proc bfcode_expanded
	set [gp0] [gp0]+1
	set [gp0] [gp0]+1
	push *[gp0]
	popbool
	cond 0Eh jmp BR1
BL1:
	push *[gp0]
	wsub 1
	bitand 255
	pop *[gp0]
	push *[gp0]
	popbool
	cond jmp BL1
BR2:
	set [gp0] [gp0#-1]
	set [gp0] [gp0#-1]

	push *[gp0]
	popbool
	cond 0Eh jmp BR2
BL2:
	push *[gp0]
	wsub 1
	bitand 255
	pop *[gp0]
	set [gp0] [gp0]+1
	set [gp0] [gp0]+1
	push *[gp0]
	wadd 1
	bitand 255
	pop *[gp0]
	set [gp0] [gp0#-1]
	set [gp0] [gp0#-1]
	push *[gp0]
	popbool
	cond jmp BL2
BR2:
	ret
end

; optimizations applied:
	; [-] -> set 0
	; >..N..>(...)<..N..< -> operate on [gp0]+N
	; pop *A then push *A -> peek *A (no change here)
; 15 words
proc bfcode_optimized1
	set *[gp0]+2 0
	push *[gp0]
	popbool
	cond 0Eh jmp BR2
BL2:
	push *[gp0]
	wsub 1
	bitand 255
	pop *[gp0]
	push *[gp0]+2
	wadd 1
	bitand 255
	pop *[gp0]+2
	push *[gp0]
	popbool
	cond 2 jmp BL2
BR2:
	ret
end