Link 7. Static Linking

Transcription

1 Link 7. Static Linking Young W. Lim Fri Young W. Lim Link 7. Static Linking Fri 1 / 41

2 Outline 1 Linking - 7. Static Linking Based on Static Library Examples Linking with Static Libraries Resolving refereces with Static Libraries Static Linking and Symbol Relocation Example Young W. Lim Link 7. Static Linking Fri 2 / 41

3 Based on "Self-service Linux: Mastering the Art of Problem Determination", Mark Wilding "Computer Architecture: A Programmer s Perspective", Bryant & O Hallaron I, the copyright holder of this work, hereby publish it under the following licenses: GNU head Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the section entitled GNU Free Documentation License. CC BY SA This file is licensed under the Creative Commons Attribution ShareAlike 3.0 Unported License. In short: you are free to share and make derivative works of the file under the conditions that you appropriately attribute it, and that you distribute it only under a license compatible with this one. Young W. Lim Link 7. Static Linking Fri 3 / 41

4 Static Library Examples 1 addvec.c and mutvec.c 2 libvector.a 3 main.c 4 p Young W. Lim Link 7. Static Linking Fri 4 / 41

5 addvec.c and multvec.c! /*::::: addvec.c :::::::::::::::::::::::::*/ void addvec(int *x, int *y, int *z, int n) { int i; } for (i=0; i<n; i++) z[i] = x[i] + y[i]; /*::::: multvec.c :::::::::::::::::::::::::*/ void multvec(int *x, int *y, int *z, int n) { int i; } for (i=0; i<n; i++) z[i] = x[i] * y[i]; Young W. Lim Link 7. Static Linking Fri 5 / 41

6 libvector.a gcc -c addvec.c addvec.o gcc -c multvec.c multvec.o ar rcs libvector.a addvec.o multvec.o libvector.a Young W. Lim Link 7. Static Linking Fri 6 / 41

7 main.c /*::::: vector.h ::::::::::::::::::::::::::*/ void addvec(int *x, int *y, int *z, int n); void multvec(int *x, int *y, int *z, int n); /*::::: main.c ::::::::::::::::::::::::::::*/ #include <stdio.h> #include "vector.h" int x[2] = { 1, 2}; int y[2] = { 3, 4}; int z[2]; int main() { } addvec(x, y, z, 2); printf("z= [%d %d]\n", z[0], z[1]); Young W. Lim Link 7. Static Linking Fri 7 / 41

8 p gcc -O2 -c main.c main.o gcc -static -o p main.o./libvector.a p./p z= [4 6] Young W. Lim Link 7. Static Linking Fri 8 / 41

9 Linking with Static Libraries 1 Static Libraries 2 ANSI C libc.a 3 advantages of static libraries 4 Unix archive 5 Linking with static library examples Young W. Lim Link 7. Static Linking Fri 9 / 41

10 Static Libraries assumption the linker reads a colleciton of relocatable object files and links them together into an output file in practice a static library is a mechanism for packaging those related object modules into a single file this file is supplied as input to the linker the linker copies only those object modules in the library that are referenced by the application program Young W. Lim Link 7. Static Linking Fri 10 / 41

11 ANSI C libc.a libc.a library an extensive collection of standard I/O atoi, pritnf, scanf string manipulation strcpy integer math functions random libm.a library an extensive collection of floating-point math functions sin, cos, sqrt Young W. Lim Link 7. Static Linking Fri 11 / 41

12 advantages of static libraries (1) related functions can be compiled into separate object modules then packaged in a single static library file application program can then use any of the functions defined in the library by specifying the file name on the command line gcc main.c /usr/lib/libc.a /usr/lib/libm.a Young W. Lim Link 7. Static Linking Fri 12 / 41

13 advantages of static libraries (2) at link time, the linker will only copy the object modules that are referenced by the program which reduces the size of the executable on disk and in memory the application programmers only need to include the names of a few library files C compiler drivers always pass libc.a to the linker so the reference to libc.a is unnecessary Young W. Lim Link 7. Static Linking Fri 13 / 41

14 Unix archive an archive on Unix systems static libraries are stored on disk in a particular file format : archive a collection of concatenated relocatable object files a header describes the size and locaton of each member object file archive filenames are denoted with the.a suffix Young W. Lim Link 7. Static Linking Fri 14 / 41

15 Linking with static library examples (1) main2.c main2.o translators (cpp, ccl, as) libvector.a addvec.o libc.a printf.o any other modules called by printf.o also main2.o, addvec.o, printf.o, other object files p2 linker (ld) void addvec(int *x, int *y) Young W. Lim Link 7. Static Linking Fri 15 / 41

16 Linking with static library examples (2) source files : main2.c, vector.h static libraries : libvector.a, libc.a relocatable object files : main2.o, addvec.o, printf.o, any other modules called by printf.o fully linked executable object file : p2 Young W. Lim Link 7. Static Linking Fri 16 / 41

17 Resolving refereces with Static Libraries 1 Symbol resoltuion phase 2 for each input file f 3 for each member in the archive f 4 for undefined symbol set U is not empty 5 Link time error 6 Link time error example 7 Ordering libraries on the command line 8 Ordering libraries on the command line examples Young W. Lim Link 7. Static Linking Fri 17 / 41

18 Symbol resolution phase during the symbol resolution phase, the linker scans the relocatable object files and archives left to right ( ) in the same order that they appear on the command line the linker maintains a set E of relocatable object files that will be merged to form the executable a set U of unresolved symbols symbols referred to but not yet defined a set D of symbols that have been defined in the previous input files initially, all the set E, U, D are empty Young W. Lim Link 7. Static Linking Fri 18 / 41

19 for each input file f for each input file f on the command line, the linker determines if f is an object file or an archive for input file f, the linker adds f to E updates U and D to reflect the symbol definitions and references in f and proceeds to the next input file Young W. Lim Link 7. Static Linking Fri 19 / 41

20 for each input archive f the linker attempts to match the unresolved symbols in U against the symbols defined by the members of the archive any member object files which are not contained in E are discarded and the linker proceeds to the next input file Young W. Lim Link 7. Static Linking Fri 20 / 41

21 for each member in the archive f if some archive member m defines a symbol that resolves a reference in U then m is added to E then the linker updates U and D to reflect the symbol definitions and references in m this process iterates over the member of object files in f until a fixed point is reached where U and D no longer change Young W. Lim Link 7. Static Linking Fri 21 / 41

22 for undefined symbol set U is not empty if U is nonempty when the linker finishes scanning the input files on the command line, it prints an error and terminates otherwise, it merges and relocates the object files in E to build the output executable file Young W. Lim Link 7. Static Linking Fri 22 / 41

23 Link time error the ordering of libraries and object files on the command line is significant if the library that defines a symbol appear on the command line before the object file that references the symbol then the reference will not be resolved and the linking will fail void addvec(int *x, int *y) Young W. Lim Link 7. Static Linking Fri 23 / 41

24 Linke time error example when libvector.a is processed, U is empty therefore, no member object files from libvector.a is added to E the reference to addvec is never resolved error message gcc -static./libvector.a main2.c in function main : undefined reference to addvec gcc -static main2.c./libvector.a Young W. Lim Link 7. Static Linking Fri 24 / 41

25 Ordering libraries on the command line if the members of different libraries are independent (no member references a symbol defined by other member) then the libraries can be placed at the end of the command line in arbitrary order if they not independent, they must be ordered so that for each symbol s that is referenced externally by a member of an archive, at least one definition of s follows a reference to s Young W. Lim Link 7. Static Linking Fri 25 / 41

26 Ordering libraries on the command line example (1) foo.c calls functions in libx.a and libz.a which call functions in liby.a libx.a and libz.a must precedes liby.a on the command line gcc foo.c libx.a libz.a liby.a Young W. Lim Link 7. Static Linking Fri 26 / 41

27 Ordering libraries on the command line example (2) libraries can be repeated on the command line if necessary to satisfy the dependence requirements foo.c calls a function in libx.a which calls a function in liby.a which again calls a function in libx.a then libx.a must be repeated on the command line gcc foo.c libx.a liby.a libx.a alternatively, libx.a and liby.a can be combined into a single archive Young W. Lim Link 7. Static Linking Fri 27 / 41

28 n m ain.c // nothing.h static void donothingstatic(); void donothing(); void doalmostnothing(); // n_main.c #include "nothing.h" int main(int argc, const char *argv[]) { doalmostnothing(); return 0; } Young W. Lim Link 7. Static Linking Fri 28 / 41

29 donothingstatic, donothing, doalmostnothing // nothing.h static void donothingstatic(); void donothing(); void doalmostnothing(); // nothing.c #include "nothing.h" static void donothingstatic() { } void donothing() { } void doalmostnothing() { donothingstatic(); donothing(); } Young W. Lim Link 7. Static Linking Fri 29 / 41

30 disassemble (1) gcc -nostdlib -c -m32 nothing.c objdump -d nothing.o nothing.o: file format elf32-i386 Disassembly of section.text: <donothingstatic>: 0: 55 push %ebp 1: 89 e5 mov %esp,%ebp 3: e8 fc ff ff ff call 4 <donothingstatic+0x4> 8: add $0x1,%eax d: 90 nop e: 5d pop %ebp f: c3 ret Young W. Lim Link 7. Static Linking Fri 30 / 41

31 disassemble (2) <donothing>: 10: 55 push %ebp 11: 89 e5 mov %esp,%ebp 13: e8 fc ff ff ff call 14 <donothing+0x4> 18: add $0x1,%eax 1d: 90 nop 1e: 5d pop %ebp 1f: c3 ret <doalmostnothing>: 20: 55 push %ebp 21: 89 e5 mov %esp,%ebp 23: e8 fc ff ff ff call 24 <doalmostnothing+0x4> 28: add $0x1,%eax 2d: e8 ce ff ff ff call 0 <donothingstatic> 32: e8 fc ff ff ff call 33 <doalmostnothing+0x13> 37: 90 nop 38: 5d pop %ebp 39: c3 ret Young W. Lim Link 7. Static Linking Fri 31 / 41

32 disassemble (3) Disassembly of section.text. x86.get_pc_thunk.ax: < x86.get_pc_thunk.ax>: 0: 8b mov (%esp),%eax 3: c3 ret Young W. Lim Link 7. Static Linking Fri 32 / 41

33 relocation seciton readelf -r nothing.o Relocation section.rel.text at offset 0x288 contains 7 entries: Offset Info Type Sym.Value Sym. Name b02 R_386_PC x86.get_pc_thunk.ax c0a R_386_GOTPC _GLOBAL_OFFSET_TABLE_ b02 R_386_PC x86.get_pc_thunk.ax c0a R_386_GOTPC _GLOBAL_OFFSET_TABLE_ b02 R_386_PC x86.get_pc_thunk.ax c0a R_386_GOTPC _GLOBAL_OFFSET_TABLE_ d02 R_386_PC donothing Relocation section.rel.eh_frame at offset 0x2c0 contains 4 entries: Offset Info Type Sym.Value Sym. Name R_386_PC text R_386_PC text R_386_PC text R_386_PC text. x86.get_pc_thu Young W. Lim Link 7. Static Linking Fri 33 / 41

34 the final binary listing (1) objdump -D nothing.o nothing.o: file format elf32-i386 Disassembly of section.group: <.group>: 0: add %eax,(%eax) 2: add %al,(%eax) 4: 06 push %es 5: add %al,(%eax)... Disassembly of section.text: <donothingstatic>: 0: 55 push %ebp 1: 89 e5 mov %esp,%ebp 3: e8 fc ff ff ff call 4 <donothingstatic+0x4> 8: add $0x1,%eax d: 90 nop e: 5d pop %ebp f: c3 ret Young W. Lim Link 7. Static Linking Fri 34 / 41

35 the final binary listing (2) <donothing>: 10: 55 push %ebp 11: 89 e5 mov %esp,%ebp 13: e8 fc ff ff ff call 14 <donothing+0x4> 18: add $0x1,%eax 1d: 90 nop 1e: 5d pop %ebp 1f: c3 ret <doalmostnothing>: 20: 55 push %ebp 21: 89 e5 mov %esp,%ebp 23: e8 fc ff ff ff call 24 <doalmostnothing+0x4> 28: add $0x1,%eax 2d: e8 ce ff ff ff call 0 <donothingstatic> 32: e8 fc ff ff ff call 33 <doalmostnothing+0x13> 37: 90 nop 38: 5d pop %ebp 39: c3 ret Disassembly of section.text. x86.get_pc_thunk.ax: < x86.get_pc_thunk.ax>: 0: 8b mov (%esp),%eax 3: Young c3 W. Lim ret Link 7. Static Linking Fri 35 / 41

36 the final binary listing (3) Disassembly of section.comment: <.comment>: 0: add %al,0x43(%edi) 3: 43 inc %ebx 4: 3a 20 cmp (%eax),%ah 6: sub %dl,0x62(%ebp) 9: 75 6e jne 79 <doalmostnothing+0x59> b: je 82 <doalmostnothing+0x62> d: and %dh,(%edi) f: 2e 33 2e xor %cs:(%esi),%ebp 12: 30 2d xor %ch,0x : 75 6e jne 88 <doalmostnothing+0x68> 1a: je 91 <doalmostnothing+0x71> 1c: 31 7e 31 xor %edi,0x31(%esi) 1f: 38 2e cmp %ch,(%esi) 21: xor %dh,(%ecx,%ebp,1) 24: and %dh,(%edi) 26: 2e 33 2e xor %cs:(%esi),%ebp 29: xor %al,(%eax) Young W. Lim Link 7. Static Linking Fri 36 / 41

37 the final binary listing (4) Disassembly of section.eh_frame: <.eh_frame>: 0: adc $0x0,%al 2: add %al,(%eax) 4: add %al,(%eax) 6: add %al,(%eax) 8: 01 7a 52 add %edi,0x52(%edx) b: add %al,(%ecx) d: 7c 08 jl 17 <.eh_frame+0x17> f: 01 1b add %ebx,(%ebx) 11: 0c 04 or $0x4,%al 13: add $0x88,%al 15: add %eax,(%eax) 17: 00 1c 00 add %bl,(%eax,%eax,1) 1a: add %al,(%eax) 1c: 1c 00 sbb $0x0,%al 1e: add %al,(%eax) 20: add %al,(%eax) 22: add %al,(%eax) 24: adc %al,(%eax) 26: add %al,(%eax) 28: e add %al,0xe(%ecx) 2b: d 05 or %al,0x50d4202(%ebp) Young W. Lim Link 7. Static Linking Fri 37 / 41

38 the final binary listing (5) 31: 4c dec %esp 32: c5 0c 04 lds (%esp,%eax,1),%ecx 35: add $0x0,%al 37: 00 1c 00 add %bl,(%eax,%eax,1) 3a: add %al,(%eax) 3c: 3c 00 cmp $0x0,%al 3e: add %al,(%eax) 40: adc %al,(%eax) 42: add %al,(%eax) 44: adc %al,(%eax) 46: add %al,(%eax) 48: e add %al,0xe(%ecx) 4b: d 05 or %al,0x50d4202(%ebp) 51: 4c dec %esp 52: c5 0c 04 lds (%esp,%eax,1),%ecx 55: add $0x0,%al 57: 00 1c 00 add %bl,(%eax,%eax,1) 5a: add %al,(%eax) 5c: 5c pop %esp 5d: add %al,(%eax) 5f: add %ah,(%eax) Young W. Lim Link 7. Static Linking Fri 38 / 41

39 the final binary listing (6) 61: add %al,(%eax) 63: 00 1a add %bl,(%edx) 65: add %al,(%eax) 67: add %al,(%eax) 69: 41 inc %ecx 6a: 0e push %cs 6b: d 05 or %al,0x50d4202(%ebp) 71: 56 push %esi 72: c5 0c 04 lds (%esp,%eax,1),%ecx 75: add $0x0,%al 77: add %dl,(%eax) 79: add %al,(%eax) 7b: 00 7c add %bh,0x0(%eax,%eax,1) 7f: add %al,(%eax) 81: add %al,(%eax) 83: add %al,(%eax,%eax,1) 86: add %al,(%eax) 88: add %al,(%eax) Young W. Lim Link 7. Static Linking Fri 39 / 41

40 relocation information in nmain.o gcc -nostdlib -c -m32 nmain.c readelf -s --wide nmain.o grep doalmostnothing 13: NOTYPE GLOBAL DEFAULT UND doalmostnothing young@usys1:~$ readelf -r nmain.o Relocation section.rel.text at offset 0x224 contains 3 entries: Offset Info Type Sym.Value Sym. Name b02 R_386_PC x86.get_pc_thunk.ax c0a R_386_GOTPC _GLOBAL_OFFSET_TABLE_ c 00000d04 R_386_PLT doalmostnothing Relocation section.rel.eh_frame at offset 0x23c contains 2 entries: Offset Info Type Sym.Value Sym. Name R_386_PC text R_386_PC text. x86.get_pc_thu Young W. Lim Link 7. Static Linking Fri 40 / 41

41 relocation information in nmain.out gcc -m32 -o nmain.out nmain.o nothing.o readelf -r nmain.out Relocation section.rel.dyn at offset 0x310 contains 8 entries: Offset Info Type Sym.Value Sym. Name 00001edc R_386_RELATIVE 00001ee R_386_RELATIVE 00001ff R_386_RELATIVE R_386_RELATIVE 00001fec R_386_GLOB_DAT _ITM_deregisterTMClone 00001ff R_386_GLOB_DAT cxa_finalize@glibc_ ff R_386_GLOB_DAT gmon_start 00001ffc R_386_GLOB_DAT _ITM_registerTMCloneTa Relocation section.rel.plt at offset 0x350 contains 1 entry: Offset Info Type Sym.Value Sym. Name 00001fe R_386_JUMP_SLOT libc_start_main@glibc_2.0 Young W. Lim Link 7. Static Linking Fri 41 / 41