In Linux, the shared library for the GNU standard C library (libc.so) is not only a shared library, but can also be run as a standalone executable, which prints out version information:
[me@computer ~]$ /lib/libc.so.6
GNU C Library stable release version 2.12, by Roland McGrath et al.
Copyright (C) 2010 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.
There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE.
Compiled by GNU CC version 4.4.7 20120313 (Red Hat 4.4.7-11).
Compiled on a Linux 2.6.32 system on 2015-01-07.
Available extensions:
The C stubs add-on version 2.1.2.
crypt add-on version 2.1 by Michael Glad and others
GNU Libidn by Simon Josefsson
Native POSIX Threads Library by Ulrich Drepper et al
BIND-8.2.3-T5B
RT using linux kernel aio
libc ABIs: UNIQUE IFUNC
For bug reporting instructions, please see:
<http://www.gnu.org/software/libc/bugs.html>.
How did they do this? I tried creating a shared library that also happened to have a main()
function, but it segfaulted when I tried to run it.
What I tried:
/* myso.c */
#include <stdlib.h>
#include <stdio.h>
static void foo(void) {
printf("hello foo\n");
}
int main(int argc, char *argv[]) {
printf("hello world\n");
foo();
return 0;
}
Then:
$ gcc -Wall -fPIC -shared -o myso.so myso.c
$ ./myso.so
Segmentation fault (core dumped)
As a note: I don't really want to do this for any real purpose, I just want to know how the GNU guys (and gals) did it.
In file Makerules
of glibc
:
# Give libc.so an entry point and make it directly runnable itself.
LDFLAGS-c.so += -e __libc_main
And in csu/version.c
extern void __libc_print_version (void);
void
__libc_print_version (void)
{
__write (STDOUT_FILENO, banner, sizeof banner - 1);
}
This is called by __libc_main()
.
In file elf/interp.c
a program interpreter is added to the shared object.
This is working for me on 64 bit Linux: gcc -Wall -Wextra -fPIC -shared -o myso.so myso.c
//myso.c
#include <unistd.h>
char const __invoke_dynamic_linker__[] __attribute__ ((section (".interp")))
#ifdef __LP64__
= "/lib64/ld-linux-x86-64.so.2";
#else
= "/lib/ld-linux.so.2";
#endif
void _start(void)
{
static char const msg[] = "Hello world!\n";
write(STDOUT_FILENO, msg, sizeof msg - 1);
_exit(0);
}
The __invoke_dynamic_linker__[]
line is needed because without it dynamic linking of write
and _exit
can't be done. The name of this variable is not important. There is an option for ld
, -dynamic-linker
, which would do the same, but it is ignored with -shared
, but putting the line into the source code works.
Here is a variant without dynamic linking:
#include <unistd.h>
#include <sys/syscall.h>
__asm__(
#ifdef __LP64__
"syscall: mov %rdi,%rax;"
" mov %rsi,%rdi;"
" mov %rdx,%rsi;"
" mov %rcx,%rdx;"
" mov %r8,%r10;"
" mov %r9,%r8;"
" mov 0x8(%rsp),%r9;"
" syscall;"
" cmp $0xfffffffffffff001,%rax;"
" jae __syscall_error;"
" retq; "
"__syscall_error: neg %rax;"
" mov %eax,%fs:0xffffffffffffffd0;"
" or $0xffffffffffffffff,%rax;"
" retq;"
#else
"syscall: push %ebp;"
" push %edi;"
" push %esi;"
" push %ebx;"
" mov 0x2c(%esp),%ebp;"
" mov 0x28(%esp),%edi;"
" mov 0x24(%esp),%esi;"
" mov 0x20(%esp),%edx;"
" mov 0x1c(%esp),%ecx;"
" mov 0x18(%esp),%ebx;"
" mov 0x14(%esp),%eax;"
" int $0x80;"
" pop %ebx;"
" pop %esi;"
" pop %edi;"
" pop %ebp;"
" cmp $0xfffff001,%eax;"
" jae __syscall_error;"
" ret;"
"__syscall_error: neg %eax;"
" mov %gs:0x0,%ecx;"
" mov %eax,-0x18(%ecx);"
" mov $0xffffffff,%eax;"
" ret;"
#endif
);
void _start(void)
{
static char const msg[] = "Hello world!\n";
syscall(SYS_write, STDOUT_FILENO, msg, sizeof msg - 1);
syscall(SYS_exit, 0);
}
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With