Carnegie Mellon

Linking

1
 Carnegie Mellon

Example C Program
main.c swap.c
int buf[2] = {1, 2}; extern int buf[];

int main() int *bufp0 = &buf[0];

{ static int *bufp1;
swap();
return 0; void swap()
} {
int temp;

bufp1 = &buf[1];
temp = *bufp0;
*bufp0 = *bufp1;
*bufp1 = temp;
}

2
 Carnegie Mellon

Static Linking
 Programs are translated and linked using a compiler driver:
▪ unix> gcc -O2 -g -o p main.c swap.c
▪ unix> ./p

main.c swap.c Source files

Translators Translators
(cpp, cc1, as) (cpp, cc1, as)

main.o swap.o Separately compiled

relocatable object files

Linker (ld)

Fully linked executable object file

p (contains code and data for all functions
defined in main.c and swap.c)
3
 Carnegie Mellon

Why Linkers?
 Reason 1: Modularity

▪ Program can be written as a collection of smaller source files,

rather than one monolithic mass.

▪ Can build libraries of common functions (more on this later)

▪ e.g., Math library, standard C library

4
 Carnegie Mellon

Why Linkers? (cont)

 Reason 2: Efficiency

▪ Time: Separate compilation

▪ Change one source file, compile, and then relink.
▪ No need to recompile other source files.

▪ Space: Libraries
▪ Common functions can be aggregated into a single file...
▪ Yet executable files and running memory images contain only
code for the functions they actually use.

5
 Carnegie Mellon

What Do Linkers Do?

 Step 1. Symbol resolution

▪ Programs define and reference symbols (variables and functions):

▪ void swap() {…} /* define symbol swap */
▪ swap(); /* reference symbol a */
▪ int *xp = &x; /* define symbol xp, reference x */

▪ Symbol definitions are stored (by compiler) in symbol table.

▪ Symbol table is an array of structs
▪ Each entry includes name, size, and location of symbol.

▪ Linker associates each symbol reference with exactly one symbol definition.

6
 Carnegie Mellon

What Do Linkers Do? (cont)

 Step 2. Relocation

▪ Merges separate code and data sections into single sections

▪ Relocates symbols from their relative locations in the .o files to

their final absolute memory locations in the executable.

▪ Updates all references to these symbols to reflect their new

positions.

7
 Carnegie Mellon

Three Kinds of Object Files (Modules)

 Relocatable object file (.o file)
▪ Contains code and data in a form that can be combined with other
relocatable object files to form executable object file.
▪ Each .o file is produced from exactly one source (.c) file

 Executable object file (a.out file)

▪ Contains code and data in a form that can be copied directly into
memory and then executed.

 Shared object file (.so file)

▪ Special type of relocatable object file that can be loaded into
memory and linked dynamically, at either load time or run-time.
▪ Called Dynamic Link Libraries (DLLs) by Windows

8
 Carnegie Mellon

Executable and Linkable Format (ELF)

 Standard binary format for object files
 Originally proposed by AT&T System V Unix
▪ Later adopted by BSD Unix variants and Linux
 One unified format for
▪ Relocatable object files (.o),
▪ Executable object files (a.out)
▪ Shared object files (.so)

 Generic name: ELF binaries

9
 Carnegie Mellon

ELF Object File Format

 Elf header
▪ Word size, byte ordering, file type (.o, exec, 0
.so), machine type, etc. ELF header
 Segment header table Segment header table
▪ Page size, virtual addresses memory segments (required for executables)
(sections), segment sizes. .text section
 .text section .rodata section
▪ Code .data section
 .rodata section .bss section
▪ Read only data: jump tables, ... .symtab section
 .data section .rel.txt section
▪ Initialized global variables .rel.data section
 .bss section .debug section
▪ Uninitialized global variables
▪ “Block Started by Symbol” Section header table
▪ “Better Save Space”
▪ Has section header but occupies no space 10
 Carnegie Mellon

ELF Object File Format (cont.)

 .symtab section
0
▪ Symbol table ELF header
▪ Procedure and static variable names
Segment header table
▪ Section names and locations (required for executables)
 .rel.text section .text section
▪ Relocation info for .text section
.rodata section
▪ Addresses of instructions that will need to be
modified in the executable .data section
▪ Instructions for modifying. .bss section
 .rel.data section .symtab section
▪ Relocation info for .data section
▪ Addresses of pointer data that will need to be .rel.txt section
modified in the merged executable .rel.data section
 .debug section .debug section
▪ Info for symbolic debugging (gcc -g)
Section header table
 Section header table
▪ Offsets and sizes of each section
11
 Carnegie Mellon

Linker Symbols
 Global symbols
▪ Symbols defined by module m that can be referenced by other modules.
▪ E.g.: non-static C functions and non-static global variables.

 External symbols
▪ Global symbols that are referenced by module m but defined by some
other module.

 Local symbols
▪ Symbols that are defined and referenced exclusively by module m.
▪ E.g.: C functions and variables defined with the static attribute.
▪ Local linker symbols are not local program variables

12
 Carnegie Mellon

Resolving Symbols
Global External Local
Global

int buf[2] = {1, 2}; extern int buf[];

int main() int *bufp0 = &buf[0];

{ static int *bufp1;
swap();
return 0; void swap() Global
} main.c {
int temp;
External Linker knows bufp1 = &buf[1];
nothing of temp temp = *bufp0;
*bufp0 = *bufp1;
*bufp1 = temp;
} swap.c

13
 Carnegie Mellon

Relocating Code and Data

Relocatable Object Files Executable Object File

System code .text 0

Headers
System data .data
System code

main()
.text
main.o
swap()
main() .text
int buf[2]={1,2} .data More system code
System data
swap.o int buf[2]={1,2} .data
int *bufp0=&buf[0]
swap() .text int *bufp1 .bss
int *bufp0=&buf[0] .data .symtab
static int *bufp1 .bss .debug

Even though private to swap, requires allocation in .bss

14
 Carnegie Mellon

Relocation Info (main)

main.c main.o
int buf[2] = 0000000 <main>:
{1,2}; 0: 8d 4c 24 04 lea 0x4(%esp),%ecx
4: 83 e4 f0 and $0xfffffff0,%esp
7: ff 71 fc pushl 0xfffffffc(%ecx)
int main()
a: 55 push %ebp
{ b: 89 e5 mov %esp,%ebp
swap(); d: 51 push %ecx
return 0; e: 83 ec 04 sub $0x4,%esp
} 11: e8 fc ff ff ff call 12 <main+0x12>
12: R_386_PC32 swap
16: 83 c4 04 add $0x4,%esp
19: 31 c0 xor %eax,%eax
1b: 59 pop %ecx
1c: 5d pop %ebp
1d: 8d 61 fc lea 0xfffffffc(%ecx),%esp
20: c3 ret

Disassembly of section .data:

Source: objdump –r -d 00000000 <buf>:

0: 01 00 00 00 02 00 00 00
15
 Carnegie Mellon

Relocation Info (swap, .text)

swap.c swap.o
extern int buf[]; Disassembly of section .text:

int 00000000 <swap>:

0: 8b 15 00 00 00 00 mov 0x0,%edx
*bufp0 = &buf[0];
2: R_386_32 buf
6: a1 04 00 00 00 mov 0x4,%eax
static int *bufp1; 7: R_386_32 buf
b: 55 push %ebp
void swap() c: 89 e5 mov %esp,%ebp
{ e: c7 05 00 00 00 00 04 movl $0x4,0x0
int temp; 15: 00 00 00
10: R_386_32 .bss
14: R_386_32 buf
bufp1 = &buf[1];
18: 8b 08 mov (%eax),%ecx
temp = *bufp0; 1a: 89 10 mov %edx,(%eax)
*bufp0 = *bufp1; 1c: 5d pop %ebp
*bufp1 = temp; 1d: 89 0d 04 00 00 00 mov %ecx,0x4
} 1f: R_386_32 buf
23: c3 ret

16
 Carnegie Mellon

Relocation Info (swap, .data)

swap.c
extern int buf[]; Disassembly of section .data:

int *bufp0 = 00000000 <bufp0>:

0: 00 00 00 00
&buf[0];
static int *bufp1; 0: R_386_32 buf

void swap()
{
int temp;

bufp1 = &buf[1];
temp = *bufp0;
*bufp0 = *bufp1;
*bufp1 = temp;
}

17
 Carnegie Mellon

Executable Before/After Relocation (.text)

0000000 <main>:
. . . 0x8048396 + 0x1a
e: 83 ec 04 sub $0x4,%esp
11: e8 fc ff ff ff call 12 <main+0x12> = 0x80483b0
12: R_386_PC32 swap
16: 83 c4 04 add $0x4,%esp
. . .

08048380 <main>:
8048380: 8d 4c 24 04 lea 0x4(%esp),%ecx
8048384: 83 e4 f0 and $0xfffffff0,%esp
8048387: ff 71 fc pushl 0xfffffffc(%ecx)
804838a: 55 push %ebp
804838b: 89 e5 mov %esp,%ebp
804838d: 51 push %ecx
804838e: 83 ec 04 sub $0x4,%esp
8048391: e8 1a 00 00 00 call 80483b0 <swap>
8048396: 83 c4 04 add $0x4,%esp
8048399: 31 c0 xor %eax,%eax
804839b: 59 pop %ecx
804839c: 5d pop %ebp
804839d: 8d 61 fc lea 0xfffffffc(%ecx),%esp
80483a0: c3 ret

18
 Carnegie Mellon

0: 8b 15 00 00 00 00 mov 0x0,%edx
2: R_386_32 buf
6: a1 04 00 00 00 mov 0x4,%eax
7: R_386_32 buf
...
e: c7 05 00 00 00 00 04 movl $0x4,0x0
15: 00 00 00
10: R_386_32 .bss
14: R_386_32 buf
. . .
1d: 89 0d 04 00 00 00 mov %ecx,0x4
1f: R_386_32 buf
23: c3 ret

080483b0 <swap>:
80483b0: 8b 15 20 96 04 08 mov 0x8049620,%edx
80483b6: a1 24 96 04 08 mov 0x8049624,%eax
80483bb: 55 push %ebp
80483bc: 89 e5 mov %esp,%ebp
80483be: c7 05 30 96 04 08 24 movl $0x8049624,0x8049630
80483c5: 96 04 08
80483c8: 8b 08 mov (%eax),%ecx
80483ca: 89 10 mov %edx,(%eax)
80483cc: 5d pop %ebp
80483cd: 89 0d 24 96 04 08 mov %ecx,0x8049624
80483d3: c3 ret
19
 Carnegie Mellon

Executable After Relocation (.data)

Disassembly of section .data:

08049620 <buf>:
8049620: 01 00 00 00 02 00 00 00

08049628 <bufp0>:
8049628: 20 96 04 08

20
 Carnegie Mellon

Strong and Weak Symbols

 Program symbols are either strong or weak
▪ Strong: procedures and initialized globals
▪ Weak: uninitialized globals

p1.c p2.c
strong int foo=5; int foo; weak

strong p1() { p2() { strong

} }

21
 Carnegie Mellon

Linker’s Symbol Rules

 Rule 1: Multiple strong symbols are not allowed
▪ Each item can be defined only once
▪ Otherwise: Linker error

 Rule 2: Given a strong symbol and multiple weak symbol,

choose the strong symbol
▪ References to the weak symbol resolve to the strong symbol

 Rule 3: If there are multiple weak symbols, pick an arbitrary

one
▪ Can override this with gcc –fno-common

22
 Carnegie Mellon

Linker Puzzles
int x;
p1() {} p1() {} Link time error: two strong symbols (p1)

int x; int x; References to x will refer to the same

p1() {} p2() {} uninitialized int. Is this what you really want?

int x; double x;
int y; p2() {} Writes to x in p2 might overwrite y!
p1() {} Evil!

int x=7; double x; Writes to x in p2 will overwrite y!

int y=5; p2() {} Nasty!
p1() {}

int x=7; int x; References to x will refer to the same initialized

p1() {} p2() {} variable.

Nightmare scenario: two identical weak structs, compiled by different compilers

with different alignment rules.
23
 Carnegie Mellon

Role of .h Files
global.h
c1.c
#ifdef INITIALIZE
#include "global.h" int g = 23;
static int init = 1;
int f() { #else
return g+1; int g;
} static int init = 0;
#endif
c2.c
#include <stdio.h>
#include "global.h"

int main() {
if (!init)
g = 37;
int t = f();
printf("Calling f yields %d\n", t);
return 0;
}
24
 Carnegie Mellon

Running Preprocessor
c1.c global.h
#include "global.h" #ifdef INITIALIZE
int g = 23;
int f() { static int init = 1;
return g+1; #else
} int g;
static int init = 0;
#endif

-DINITIALIZE
no initialization

int g = 23; int g;

static int init = 1; static int init = 0;
int f() { int f() {
return g+1; return g+1;
} }

#include causes C preprocessor to insert file verbatim

25
 Carnegie Mellon

Role of .h Files (cont.)

global.h
c1.c
#ifdef INITIALIZE
#include "global.h" int g = 23;
static int init = 1;
int f() { #else
return g+1; int g;
} static int init = 0;
#endif
c2.c
#include <stdio.h> What happens:
#include "global.h"
gcc -o p c1.c c2.c
int main() { ??
if (!init) gcc -o p c1.c c2.c \
g = 37; -DINITIALIZE
int t = f(); ??
printf("Calling f yields %d\n", t);
return 0;
}
26
 Carnegie Mellon

Global Variables
 Avoid if you can

 Otherwise
▪ Use static if you can
▪ Initialize if you define a global variable
▪ Use extern if you use external global variable

27
 Carnegie Mellon

Packaging Commonly Used Functions

 How to package functions commonly used by programmers?
▪ Math, I/O, memory management, string manipulation, etc.

 Awkward, given the linker framework so far:

▪ Option 1: Put all functions into a single source file
▪Programmers link big object file into their programs
▪ Space and time inefficient
▪ Option 2: Put each function in a separate source file
▪ Programmers explicitly link appropriate binaries into their
programs
▪ More efficient, but burdensome on the programmer

28
 Carnegie Mellon

Solution: Static Libraries

 Static libraries (.a archive files)
▪ Concatenate related relocatable object files into a single file with an
index (called an archive).

▪ Enhance linker so that it tries to resolve unresolved external references

by looking for the symbols in one or more archives.

▪ If an archive member file resolves reference, link it into the executable.

29
 Carnegie Mellon

Creating Static Libraries

atoi.c printf.c random.c

Translator Translator ... Translator

atoi.o printf.o random.o

unix> ar rs libc.a \
Archiver (ar)
atoi.o printf.o … random.o

libc.a C standard library

 Archiver allows incremental updates

 Recompile function that changes and replace .o file in archive.

30
 Carnegie Mellon

Commonly Used Libraries

libc.a (the C standard library)
▪ 8 MB archive of 1392 object files.
▪ I/O, memory allocation, signal handling, string handling, data and time,
random numbers, integer math
libm.a (the C math library)
▪ 1 MB archive of 401 object files.
▪ floating point math (sin, cos, tan, log, exp, sqrt, …)

% ar -t /usr/lib/libc.a | sort % ar -t /usr/lib/libm.a | sort

… …
fork.o e_acos.o
… e_acosf.o
fprintf.o e_acosh.o
fpu_control.o e_acoshf.o
fputc.o e_acoshl.o
freopen.o e_acosl.o
fscanf.o e_asin.o
fseek.o e_asinf.o
fstab.o e_asinl.o
… …
31
 Carnegie Mellon

Linking with Static Libraries

addvec.o multvec.o

main2.c vector.h Archiver

(ar)
Translators
(cpp, cc1, as) libvector.a libc.a Static libraries

Relocatable main2.o addvec.o printf.o and any other

object files modules called by printf.o

Linker (ld)

Fully linked
p2
executable object file

32
 Carnegie Mellon

Using Static Libraries

 Linker’s algorithm for resolving external references:
▪ Scan .o files and .a files in the command line order.
▪ During the scan, keep a list of the current unresolved references.
▪ As each new .o or .a file, obj, is encountered, try to resolve each
unresolved reference in the list against the symbols defined in obj.
▪ If any entries in the unresolved list at end of scan, then error.

 Problem:
▪ Command line order matters!
▪ Moral: put libraries at the end of the command line.
unix> gcc -L. libtest.o -lmine
unix> gcc -L. -lmine libtest.o
libtest.o: In function `main':
libtest.o(.text+0x4): undefined reference to `libfun'

33
 Carnegie Mellon

Loading Executable Object Files Memory

outside 32-bit
Executable Object File Kernel virtual memory
0 address space
ELF header 0x100000000
User stack
Program header table (created at runtime)
%esp
(required for executables)
(stack
.init section pointer)
.text section Memory-mapped region for
.rodata section shared libraries
0xf7e9ddc0
.data section
.bss section brk
.symtab Run-time heap
(created by malloc)
.debug
Read/write segment Loaded
.line (.data, .bss) from
the
.strtab Read-only segment executable
Section header table (.init, .text, .rodata) file
0x08048000
(required for relocatables) Unused
0 34
 Carnegie Mellon

Shared Libraries
 Static libraries have the following disadvantages:
▪ Duplication in the stored executables (every function need std libc)
▪ Duplication in the running executables
▪ Minor bug fixes of system libraries require each application to explicitly
relink

 Modern solution: Shared Libraries

▪ Object files that contain code and data that are loaded and linked into
an application dynamically, at either load-time or run-time
▪ Also called: dynamic link libraries, DLLs, .so files

35
 Carnegie Mellon

Shared Libraries (cont.)

 Dynamic linking can occur when executable is first loaded
and run (load-time linking).
▪ Common case for Linux, handled automatically by the dynamic linker
(ld-linux.so).
▪ Standard C library (libc.so) usually dynamically linked.

 Dynamic linking can also occur after program has begun

(run-time linking).
▪ In Linux, this is done by calls to the dlopen() interface.
▪ Distributing software.
▪ High-performance web servers.
▪ Runtime library interpositioning.

 Shared library routines can be shared by multiple processes.

▪ More on this when we learn about virtual memory
36
 Carnegie Mellon

Dynamic Linking at Load-time

main2.c vector.h unix> gcc -shared -o libvector.so \
addvec.c multvec.c
Translators
(cpp, cc1, as) libc.so
libvector.so
Relocatable main2.o Relocation and symbol
object file table info

Linker (ld)

Partially linked p2
executable object file

Loader libc.so
(execve) libvector.so

Code and data

Fully linked
executable Dynamic linker (ld-linux.so)
in memory
37
 Carnegie Mellon

Dynamic Linking at Run-time

#include <stdio.h>
#include <dlfcn.h>

int x[2] = {1, 2};

int y[2] = {3, 4};
int z[2];

int main()
{
void *handle;
void (*addvec)(int *, int *, int *, int);
char *error;

/* dynamically load the shared lib that contains addvec() */

handle = dlopen("./libvector.so", RTLD_LAZY);
if (!handle) {
fprintf(stderr, "%s\n", dlerror());
exit(1);
}

38
 Carnegie Mellon

Dynamic Linking at Run-time

...

/* get a pointer to the addvec() function we just loaded */

addvec = dlsym(handle, "addvec");
if ((error = dlerror()) != NULL) {
fprintf(stderr, "%s\n", error);
exit(1);
}

/* Now we can call addvec() just like any other function */

addvec(x, y, z, 2);
printf("z = [%d %d]\n", z[0], z[1]);

/* unload the shared library */

if (dlclose(handle) < 0) {
fprintf(stderr, "%s\n", dlerror());
exit(1);
}
return 0;
}

39