INLINE vs. NORMAL FUNCTIONS: What's Really Happening in Assembly?

#cpp #c #tutorial #programming

🔥 INLINE vs. NORMAL FUNCTIONS: What's Really Happening in Assembly?

All sources (code, etc.) - https://github.com/dima853/self_university/tree/main/network/c/fucntions/inline

Everyone talks about 'inline', but few have seen the difference in real assembly. See:

🎯 With INLINE — code is INSERTED directly:

test_with_inline:
movl (%rdi), %r8d # Byte comparison
cmpl (%rsi), %r8d # ← code INSERTED!
je .L18

movl (%rsi), %ecx
cmpl (%rdx), %ecx # ← INSERTED again!
je .L19

The gist: The compiler COPYS the function code to each call site.

🐌 WITHOUT INLINE — CALLS to the function:

test_without_inline:
call mac_equals_normal # ← JUMP into the function!
movzbl %al, %ecx

call mac_equals_normal # ← JUMP AGAIN!
movl %eax, %esi

The gist: Each call is a jump to a different memory location

🍔 SIMPLE ANALOGY:

Without inline = "Courier: you → courier → restaurant → courier → you"
With inline = "Microwave: you → microwave → done!"

⚡ RESULT:

With inline: ~4 instructions per comparison
Without inline: ~20+ instructions (call + return)

🎯 CONCLUSION:

inline is when the compiler stops "sending the courier" and starts "microwave" right there!

Command for testing:

gcc -S -O2 test.c -o test.s

programming #C #assembler #optimization #inline #compiler

DEV Community