Understanding assembly code for LED blink with ATmega328P

Question

I wrote the following program to turn an LED on and off at 1-second intervals:

#include <avr/io.h>
#include <util/delay.h>
int main(void)
{
 DDRB = 0b00100000;
 for (;;) {
 PORTB = 0b00100000;
 _delay_ms(1000);
 PORTB = 0b00000000;
 _delay_ms(1000);
 }
 return 0;
}

I tried disassembling the binary file of this code using the avr-objdump of which the output is:

00000080 <main>:
 80: 80 e2 ldi r24, 0x20 ; 32
 82: 84 b9 out 0x04, r24 ; 4
00000084 <.L2>:
 84: 85 b9 out 0x05, r24 ; 5
 86: 2f ef ldi r18, 0xFF ; 255
 88: 33 ed ldi r19, 0xD3 ; 211
 8a: 90 e3 ldi r25, 0x30 ; 48
0000008c <.L1^B1>:
 8c: 21 50 subi r18, 0x01 ; 1
 8e: 30 40 sbci r19, 0x00 ; 0
 90: 90 40 sbci r25, 0x00 ; 0
 92: e1 f7 brne .-8 ; 0x8c <.L1^B1>
 94: 00 c0 rjmp .+0 ; 0x96 <L0^A>
00000096 <L0^A>:
 96: 00 00 nop
 98: 15 b8 out 0x05, r1 ; 5
 9a: 2f ef ldi r18, 0xFF ; 255
 9c: 33 ed ldi r19, 0xD3 ; 211
 9e: 90 e3 ldi r25, 0x30 ; 48
000000a0 <.L1^B2>:
 a0: 21 50 subi r18, 0x01 ; 1
 a2: 30 40 sbci r19, 0x00 ; 0
 a4: 90 40 sbci r25, 0x00 ; 0
 a6: e1 f7 brne .-8 ; 0xa0 <.L1^B2>
 a8: 00 c0 rjmp .+0 ; 0xaa <L0^A>
000000aa <L0^A>:
 aa: 00 00 nop
 ac: eb cf rjmp .-42 ; 0x84 <.L2>
000000ae <_exit>:
 ae: f8 94 cli
000000b0 <__stop_program>:
 b0: ff cf rjmp .-2 ; 0xb0 <__stop_program>

My understanding of the above assembly code is:

• the two lines under 00000080 <main>: sets the data direction of PB5 to output.
• 00000084 <.L2>: section turns the LED on.
• 0000008c <.L1^B1>: waits for 1 second (first 4 lines) which then jumps to 00000096 <L0^A>:
• 00000096 <L0^A>: turns the LED off by writing the contents of r1 (which is currently 0s) to PB5
• then 000000a0 <.L1^B2>: waits for 1 second and jumps to 000000aa <L0^A>:
• 000000aa <L0^A>: spends one clock cycle and jumps to 00000084 <.L2>: which turns the LED on. So on and so forth.

Questions:

1. Is my understanding of the assembly code correct?
2. What is the purpose of a nop sometimes as the first instruction?

• Andy aka: Have you checked the data sheets to see if R24 is a data direction register?
• Mattman944: What is your goal? If you want to learn assembly, I don't recommend this method. Compilers often produce code that is harder to understand than assembly code generated by humans.
• kovac: @Andyaka I checked the datasheet and thought that 0x04 is the data direction register. I thought that I'm writing the contents of r24 to 0x04, I may be wrong...
• jonathanjo: @kovac you are right: r24 is often used as a temporary register. Data direction register B is at IO address 4, ie, when using OUT. (And also at memory address 0x24, for when using ST.) It's a pecularity of this CPU that some registers appear in both IO and memory address spaces, but with different addresses.
• G36: Take a look here godbolt.org/z/9jTd7sdnj

Answer 1

1. In general, yes, but disassembling the raw binary may not be very fruitful, as you could just look at the compiler output listing which includes the C source code and generated assembly opcodes.

2. You ask for 1000 millisecond delay, which is done by looping 3199999 times. As the loop will be a few clock cycles less than 1000 milliseconds, it adds extra useless RJMP and NOP opcodes to waste a few clock cycles to end up closer to 1000 milliseconds.

• kovac: For 1., how do I get this?
• jonathanjo: Most C compilers support -S for making the assmbler file filename.s from filename.c. If you're using the avr toolset, avr-gcc -S -Os -mmcu=atmega328 -c filename.c is likely to give you useful output. Also look at assembler listing output: avr-as -mmcu=atmega328 -o filename.o -a=filename.lst filename.s