convert bitmask buffer to array indices

bitmap_decode_ctz.asm

; references:
;	https://branchfree.org/2018/05/22/bits-to-indexes-in-bmi2-and-avx-512/
;	https://lemire.me/blog/2018/03/08/iterating-over-set-bits-quickly-simd-edition/
;	http://0x80.pl/notesen/2019-01-05-avx512vbmi-remove-spaces.html
align 64
; RCX : number of quadword to convert
; RSI : source bit array
; RDI : destination index array
bitmap_decode_ctz:
	jrcxz .all_zero
	xor ebx,ebx
	shl rcx,3
	push rdi
.one_more:
	mov rdx,[rsi+rbx]
	add rbx,8
	test rdx,rdx
	jz .one_zero
.convert:
	tzcnt rax,rdx
	lea rax,[rbx*8+rax]
	stosd ; mov [rdi],eax | add rdi,4
	blsr rdx,rdx
	jnz .convert
.one_zero:
	cmp rbx,rcx
	jc .one_more

	sub rdi,[rsp]
	xchg rax,rdi
	pop rdi
	shr rcx,3 ; restore value
	shr rax,2
	retn
.all_zero:
	xor eax,eax
	retn
; RAX : popcount of bit array and length of indice array
; RCX,RSI,RDI : unchanged
; RBX : ECX * 8

kind of elegant (imho) how all the parameters are preserved 😊