Re: [PATCH] Set bounds on what /proc/self/make-it-fail accepts.

[Dave Jones]

On Wed, Feb 19, 2014 at 02:00:21PM -0800, Andrew Morton wrote:
 
 > > I toyed with the idea of changing task_struct.make_it_fail to unsigned too,
 > > but only realized I missed that after I'd sent out the diff.
 > 
 > If we're touching the task_struct we could make it a bool.
 > 
 > Or just a single bit(field).  task_struct already has a bunch of
 > bitfields in it (strangely, they aren't contiguous).

afaics, asides from brk_randomized, they're contiguous, and gcc dtrt..

        unsigned int               in_execve:1;          /*   768:31  4 */
        unsigned int               in_iowait:1;          /*   768:30  4 */
        unsigned int               no_new_privs:1;       /*   768:29  4 */
        unsigned int               sched_reset_on_fork:1; /*   768:28  4 */
        unsigned int               sched_contributes_to_load:1; /*   768:27  4 */

So we could move the COMPAT_BRK ifdef and save 4 bytes for all the people still using libc5.
(Or those who are for some reason averse to heap randomization).

It's not really worth doing unless you're moving a bunch of other stuff around
in task_struct though, because as it is now, that struct has a bunch of alignment padding
& holes, so you're not going to save anything.

The other tricky part with reorganizing that struct is that so much of it is configurable.

	Dave

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/