Compilation stages — preprocessing, compilation, assembly, linking

5.3.1 · Coding › Build Systems & Toolchain

YEH pipeline exist KYUN karti hai?

Stage 1 — Preprocessing

YEH KYA karta hai:

#include <stdio.h> → us header file ka poora content physically paste kar deta hai.
#define PI 3.14 → har PI token ko 3.14 se replace kar deta hai.
#ifdef / #ifndef / #endif → code ke blocks ko conditionally rakhta ya delete karta hai.
Comments strip kar deta hai.

ISKO kaise dekhen:

gcc -E hello.c -o hello.i   # -E = stop after preprocessing

Worked example Macro expansion

#define SQ(x) ((x)*(x))
int y = SQ(3+1);

Preprocessing ke baad → int y = ((3+1)*(3+1)); = 16. Inner parentheses kyun? Kyunki macro pure text paste hai. Unke bina, SQ(3+1) ban jaata 3+1*3+1 = 7. Preprocessor math nahi jaanta — sirf text!

Stage 2 — Compilation

ANDAR kya hota hai:

Lexing → text ko tokens mein tod deta hai.
Parsing → ek Abstract Syntax Tree (AST) banata hai.
Semantic analysis → type checks (yahi se tumhari zyaadatar errors aati hain!).
Optimization → jaise constant folding 2*3 → 6.
Code generation → target CPU ke liye assembly emit karta hai (x86, ARM…).

ISKO kaise dekhen:

gcc -S hello.i -o hello.s   # -S = stop after compilation

Worked example

.s kaisa dikhta hai int main(){return 42;} compile ho sakta hai:

main:
    mov  eax, 42      ; put 42 in return register
    ret               ; return

eax kyun? x86 par, convention (the ABI) ke hisaab se function ka return value eax register mein pass hota hai.

Stage 3 — Assembly

ABHI runnable kyun nahi? Kyunki yeh aisi functions call kar sakta hai (jaise printf) jo doosri files mein rehti hain. Unke addresses abhi unknown placeholders hain.

ISKO kaise dekhen:

gcc -c hello.s -o hello.o   # -c = compile+assemble, stop before linking

Stage 4 — Linking

YEH kya resolve karta hai:

Tumhara hello.o kehta hai "mujhe printf chahiye". Linker C standard library (libc) ke andar printf dhundhta hai aur address patch kar deta hai.

Do prakar:

Static linking → library code executable mein copy ho jaata hai (bada file, self-contained).
Dynamic linking → sirf ek reference store hota hai; actual .so/.dll runtime par load hoti hai (chota file, shared).

KAISE:

gcc hello.o -o hello        # links, produces executable

Worked example Classic linker error

undefined reference to `foo'

Matlab: koi object file foo use kar raha hai, lekin koi bhi object file ya library usse define nahi karti. Yeh ek link-time error hai, compile error NAHI — har file ki compilation individually succeed hui thi!

Common Mistakes (Steel-manned)

Forecast-then-Verify

80/20 — The vital few

Recall Feynman: 12-saal ke bachhe ko samjhao

Socho tumne English mein ek recipe likhi. Pehle, ek helper saari sub-recipes jo tumne reference ki thi copy karke laata hai aur shorthand notes theek karta hai (preprocess). Phir ek translator use chef ki special cooking-shorthand mein rewrite karta hai (compile). Ek doosra insaan us shorthand ko exact robot-arm button presses mein badalta hai (assemble) — lekin recipe abhi bhi kehti hai "doosri kitchen se sauce lo". Aakhir mein, linker doosri kitchens se saari sauces collect karne dauda jaata hai aur sab kuch ek FINISHED cookbook mein glue kar deta hai jise robot start-to-finish run kar sake.

Connections

Build Systems & Toolchain
Static vs Dynamic Linking
Object files and symbol tables
Make and incremental builds
ABI and calling conventions
Preprocessor macros and include guards

Flashcards

4 compilation stages order mein kya hain?

Preprocessing → Compilation → Assembly → Linking

Preprocessing ke baad kaun sa gcc flag rukta hai?

-E (.i produce karta hai)

Compilation ke baad kaun sa gcc flag rukta hai (assembly emit karta hai)?

-S (.s produce karta hai)

Kaun sa gcc flag compile+assemble karta hai lekin link nahi karta?

-c (.o produce karta hai)

Preprocessor #include ke saath actually kya karta hai?

Poori header file ka text source mein physically paste kar deta hai.

Translation unit kya hai?

Preprocessor ka pure-C output (koi # directives nahi bache), jo compiler ka actual input hai.

SQ(x) ((x)*(x)) ko parenthesize kyun kiya gaya hai?

Macros text substitution hain; parens ke bina, operator precedence SQ(3+1) jaisi expressions tod deta hai.

Assembler kya produce karta hai aur woh runnable kyun nahi hota?

Ek object file (.o) machine code ke saath, lekin usmein unresolved symbols aur placeholder addresses hain.

Linker ka main kaam kya hai?

Object files + libraries combine karna, symbols resolve karna, aur addresses ko ek executable mein relocate karna.

undefined reference to 'x' kis stage par aata hai?

Linking par — ek symbol use kiya gaya lekin kabhi define nahi kiya gaya.

Static vs dynamic linking?

Static library code ko executable mein copy karta hai; dynamic ek reference store karta hai jo runtime par shared library se load hota hai.

Type-checking errors kis stage par hoti hain?

Compilation stage par (semantic analysis ke dauran).

Object file ke symbol table mein kya hota hai?

Woh symbols jo woh define karta hai aur woh symbols jo use chahiye (undefined references).

C program ke liye file extension chain?

.c → .i → .s → .o → executable