Re: Weirdness with stat()

[Steve Schlaifer <steve@kihakkt.jetcafe.org>]

On Tue, 4 Jul 2000, Jim Hull wrote:
> [...]
>
> struct stat *st;
> 
>    stat(CONFIG, st);
>    if ((st->st_uid != 0) || (st->st_mode != 33152)) {
>       fprintf(stderr, "Unsecure modes on configuration files\n");
>       exit(1);
>    }
> 
> This worked with absolutly no problems on my laptop. I went to run it on
> my desktop (2.4.0-test2 SMP) and it started segfaulting. It wasn't
> segfaulting at the stat() call, it was segfaulting in weird places like a
> syslog() or a execv().
> 
> [...]
>
> Would this be my poor implementation of stat, or would this actually be a
> kernel bug.

Neither, this is a bad use of stat.  In this case, st is an uninitialized
pointer and you are having stat treat that as a pointer to some random
location in memory which it dutifully fills with the stat information.
Change your code to

struct stat st;              /* note not *st, just st */

stat(CONFIG, &st);           /* note &st not st */

and things will get much better.

		--Steve

  "We have a government of law, and government officials must
  be held accountable under the law."

    -- Judge Royce Lamberth
       U.S. District Court for the District of Columbia


Kernel-audit:  discussion list for security and the linux kernel
Archive:       http://mail.nl.linux.org/kernel-audit/