I can't understand the way that binary multiplication takes place in avr code. Here is code from avr freaks. What exactly does "ror" and "lsr" do? I thought that for multiplication in binary we need lsl instead.
.def mc8s =r16 ;multiplicand
.def mp8s =r17 ;multiplier
.def m8sL =r17 ;result Low byte
.def m8sH =r18 ;result High byte
.def mcnt8s =r19 ;loop counter
mpy8s: sub m8sH,m8sH ;clear result High byte and carry
ldi mcnt8s,8 ;init loop counter
m8s_1: brcc m8s_2 ;if carry (previous bit) set
add m8sH,mc8s ; add multiplicand to result High byte
m8s_2: sbrc mp8s,0 ;if current bit set
sub m8sH,mc8s ; subtract multiplicand from result High
asr m8sH ;shift right result High byte
ror m8sL ;shift right result L byte and multiplier
enter code here
dec mcnt8s ;decrement loop counter
enter code here
brne m8s_1 ;if not done, loop more
ret
As you know, multiplication can be performed using shifting and addition, e.g.
101
101
===
101
1 01
= ====
1 1001
Notice that although the numbers fit in 4 bits, I needed to use more bits for the addition. To avoid this, the code you quote shifts the result rather than the number to be added.
101
--- -
10 1
-- --
1 01
101
===
110 01
--- --
11 001
-- ----
1 0001
This way only 4 bits of addition are needed.