Howdy,
Could someone on the list with access to the complete dladdr1() source
code take a look to see how the value of info is set? When I compile and
run the following code:
$ cat dladdr.c
#include <stdio.h>
#include <dlfcn.h>
#include <sys/elf.h>
static void static_func() { }
void global_func() { }
int main(int argc, char *argv[]) {
Dl_info dlip;
Elf32_Sym *info;
if (dladdr1(global_func, &dlip, (void **)&info, RTLD_DL_SYMENT))
printf("Global: info points to symbol table entry at 0x%0x\n"
" st_value = %d, st_size = %d\n", info,
info->st_value, info->st_size);
if (dladdr1(static_func, &dlip, (void **)&info, RTLD_DL_SYMENT))
printf("Static: info points to symbol table entry at 0x%0x\n"
" st_value = %d, st_size = %d\n", info,
info->st_value, info->st_size);
return(0);
}
I am having trouble correlating the values from dladdr1() with the
values provided by nm ( for the static function only ):
$ ./dladdr
Global: info points to symbol table entry at 0x1027c
st_value = 67264, st_size = 12
Static: info points to symbol table entry at 0x102ec
st_value = 66800, st_size = 116
$ nm dladdr | grep func
[84] | 67264| 12|FUNC |GLOB |0 |9 |global_func
[58] | 67252| 12|FUNC |LOCL |0 |9 |static_func
Does anyone happen to have any insight into why this may be occuring?
Secondly, in display_stack_info(), which is called from printstack() in
libc, no attempt is made to verify that sym is not a NULL pointer
prior to deferencing it:
482 if (dladdr1((void *) pc, &info, (void**) &sym,
RTLD_DL_SYMENT) == 0) {
483 /* no info at all */
484 if (signo == 0)
485 async_filenoprintf(filenum, "0x%x\n", pc);
486 else
487 async_filenoprintf(filenum,
488 "0x%x [ Signal %d (%s)]\n", pc,
489 (ulong_t)signo, sigbuf);
490
491 } else if ((pc - (unsigned long)info.dli_saddr) <
492 sym->st_size) {
This caused my segfault handler to SEGFAULT (cute isn't it), and
I would like to recommend adding a check similar to the following:
491 } else if (sym && ((pc - (unsigned long)info.dli_saddr) <
492 sym->st_size)) {
Thanks,
- Ryan
--
UNIX Administrator
http://daemons.net/~matty
_______________________________________________
opensolaris-code mailing list
[email protected]
https://opensolaris.org:444/mailman/listinfo/opensolaris-code
