On 12/17, Oleg Nesterov wrote:
>
> On 12/16, Richard Guy Briggs wrote:
> >
> > task->pid is only ever assigned once (well ok, twice). For system health
> > and
> > secure logging confidence, make it const to make it much more intentional
> > when
> > it is being changed.
>
> Hmm. I am a bit suprized you decided to constify task->pid, but OK.
>
> And we can do the same with task->signal, this can actually help to generate
> a better code, probably.
>
> > if (!thread_group_leader(tsk)) {
> > struct task_struct *leader = tsk->group_leader;
> > + /* tast_struct::pid is const pid_t, hence the ugly cast */
> > + pid_t *pid_p = (pid_t*)&(tsk->pid);
> >
> > sig->notify_count = -1; /* for exit_notify() */
> > for (;;) {
> > @@ -950,7 +952,7 @@ static int de_thread(struct task_struct *tsk)
> > * Note: The old leader also uses this pid until release_task
> > * is called. Odd but simple and correct.
> > */
> > - tsk->pid = leader->pid;
> > + *pid_p = leader->pid;
>
> Well, imho this (and de_thread) looks a bit ugly. Perhaps we should add
> something like
>
> #define ASSIGN_CONST(l, r) (*(typeof(r) *)&(l) = (r))
>
> into compiler.h ?
Or even
#define ASSIGN_CONST(l, r) \
({ BUILD_BUG_ON(sizeof(l) != sizeof(r)); *(typeof(r) *)&(l) =
(r); })
perhaps it will have more users.
Oleg.
--
Linux-audit mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-audit
