CS208 Introduction to Computer Systems
Friday, 2 February 2024

+ Office hours today
- Busy 10-11
- Available 9:35-10:00 and during convo

+ Today: key ideas for the new assignment

+ x86-64 syntax reminder
- Intel syntax
    - used by Windows compilers
    - used in nearly all online documentation
    - OP destination, source

- AT&T
    - used by Unix compilers (notably gcc)
    - OP source, destination

- Get a cheat sheet from somebody using gcc

+ Instructions of interest
- Resources
    - instruction references
    - cheat sheets

- Easy ones
    - mov -- it should have been named copy
    - add
    - sub
    - neg -- one operand, set to -1 * original

- Example
    mov $0x5, %rsi
    mov $0x9, %rax
    sub %rax, %rsi
        puts -4 = 0xFFFFFFFFFFFFFFFC into %rsi
    jl  .L8
        jumps to label .L8 because -4 < 0
        and "jl" means "jump if less than"
        Note: jl is looking at the SF

    dest = dest - source (see CMU cheat sheet)
    %rsi = %rsi - %rax
    modifies the EFLAGS register according to the final result
        look at SUB on the instruction ref (felixcloutier.com/x86/)
        look at FLAGS on Wikipedia
    SF bit gets set to 1 because the result was < 0
    ZF bit gets set to 0 because the result was != 0

- Other essentials
    - cmp a, b
        computes b - a, sets flags SF, ZF, etc.,
            then throws the result away
    
    - test a, b
        computes (a bitwise-AND b), sets flags SF, ZF,...
            then throws the result away
    - jcc "jump condition code", jump or not based on flags
        jl
        jge
        ...

    - lea "load effective address"
        sometimes used to compute addresses of things
            (pointer arithmetic)
        often used to just do arithmetic

    - push
    - pop
    - call
    - ret
- Others that show up in this assignment
    - set
    - or
    - movzb
    - movs
    - sal



+ Addressing modes
- Immediate
    $0x36 --> the number 0x36
- Register
    %rax --> the number in %rax
- Register Indirect
    (%rax) --> the thing in memory at address %rax
- Register Indirect with offset
    0x12(%rax) --> the thing in memory at address 0x12 + %rax
- Indexed
    (a, b, c) --> a + b*c
    (a, b) --> a + b*1 = a + b
- Indexed with offset
    d(a, b, c) --> a + b*c + d
    d(a, b) --> a + b*1 + d = a + b + d



=====
puzzle0.asm thinking

function1:
  movl %edi, %eax      --- %edi is the first parameter, no %esi
                            so this is certainly a 1-param function
                            retval (eax) = param
  testl %edi, %edi     --- bitwise AND of the param with itself
                            so SF and ZF get set based on the param
  js .L3               --- jump to .L3 iff the param < 0
.L2:
  ret
.L3:
  negl %eax            --- retval = retval * -1
  jmp .L2

----
int f(int edi) {  // absolute value ("abs")
    int eax = edi;
    if (edi < 0) {
        eax = eax * -1;
    } else {
    }
    
    return eax;
}