On 12 Aug 2003, Keith Owens <kaos@xxxxxxx> wrote:
> On Tue, 12 Aug 2003 15:59:00 +1000,
> Martin Pool <mbp@xxxxxxxxx> wrote:
> >+++ linux-2.6.0test2-ia64-kdb/kdb/kdbmain.c 2003-08-12 15:30:52.000000000
> >+1000
> >@@ -2959,12 +2959,20 @@ kdb_ps(int argc, const char **argv, cons
> > mask = kdb_task_state_string(argc, argv, envp);
> > /* Run the active tasks first */
> > for (cpu = 0; cpu < NR_CPUS; ++cpu) {
>
> That line is wrong. In kdb v4.3 it reads
>
> for (cpu = 0; cpu < smp_num_cpus; ++cpu) {
>
> which, together with code in sched.c that ensures kdb_active_task[cpu]
> is always set, removes the need for the other tests. IOW,
> kdb_active_task[cpu] is always defined.
Thanks for explaining that. There doesn't seem to be any variable
smp_num_cpus on ia64 2.6.0test2. (I realize your patch is just for
2.4 but I was trying to port it forward.) I'll try to work out what
happened to it.
> >+ /* FIXME: Running tasks are printed out twice: once above for
> >+ * the CPU, and then again here. */
>
> Deliberately so. The first bit displays the tasks that own the cpus,
> the second bit displays all of the tasks. It is harder for the user if
> they have to hunt between the two sections to find processes, a little
> bit of duplication to assist the user is no big deal.
OK. I was going to make it consistent with bta, but I suppose the
information is so much larger there that we have to avoid the
duplicate.
--
Martin
|