Because you link the startup files to your program, which contains (usually) assembler code that calls your main. If main were static, that code wouldn't be able to call main.
external linkage
means that other so-called translation-units
can see your symbol declared extern in its own translation-unit. So, your main is extern, and it will have an entry in its translation-units symbol table that states its address. Other translation-units will then be able to jump to that address when they want to call main.
static linkage
means your symbol is strictly translation-unit local. This means other translation units
will not be able to see that symbol. Thus, symbols with static linkage can occur in different translation units multiple times, and they won't clash with each other because they are local.
Edit: Generally, files generated by the compiler from translation units are specific to that particular compiler. For gcc on linux, often the ELF object format is used. You can view its symbol table using readelf -sW <file>.o
(simple test-file below):
test.c
void bar(void);
static int foo(void) {
return 1;
}
int main(void) {
bar();
return foo();
}
Here is the output of readelf:
Symbol table '.symtab' contains 10 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 00000000 0 NOTYPE LOCAL DEFAULT UND
1: 00000000 0 FILE LOCAL DEFAULT ABS test.c
2: 00000000 0 SECTION LOCAL DEFAULT 1
3: 00000000 0 SECTION LOCAL DEFAULT 3
4: 00000000 0 SECTION LOCAL DEFAULT 4
5: 00000000 10 FUNC LOCAL DEFAULT 1 foo
6: 00000000 0 SECTION LOCAL DEFAULT 6
7: 00000000 0 SECTION LOCAL DEFAULT 5
8: 0000000a 36 FUNC GLOBAL DEFAULT 1 main
9: 00000000 0 NOTYPE GLOBAL DEFAULT UND bar
You see the main function, and a static foo function, called by main. Also there is a function called which is not defined in the file, but which is defined in another object file. As the object file wasn't finally linked yet, the functions don't have final addresses assigned yet. After the final link, these will be arranged into the executable and will have addresses assigned. The object file has entries for calls to not-yet defined functions, so that when the file is linked, those call instructions can have the final addresses stored (readelf -r <file>.o
):
Relocation section '.rel.text' at offset 0x308 contains 1 entries:
Offset Info Type Sym.Value Sym. Name
0000001c 00000902 R_386_PC32 00000000 bar