jschopp at austin.ibm.com wrote:
> check_dr_state has a built it 5 second timeout so it doesn't keep retrying
> forever. rtas_get_sensor will keep trying forever. I am not opposed to
> using rtas_get_sensor (though we should change it to check 9900..9905
> instead of 9900..9909 for extended delay).
Ah, that is an important difference. Looks as if we should implement an
rtas_get_sensor_timeout() interface that allows the caller to specify a
timeout value. And I believe you're right about the range for extended
delay. I think we could move check_dr_state to rtas.c (along with your
fix) and tweak it for this purpose.
> + while (nr_threads--) {
> + if (0 == query_cpu_stopped(tid[nr_threads])) {
> + best = tid[nr_threads];
> + if (best == old_hwindex)
> goto out;
>
> This part is a little hard to read. I actually had to think about
> nr_threads being 1 for the while and 0 for the indexing tid. If you clean
> this bit up to be more readable I will love the patch.
Yeah I guess a traditional "for" loop would be nicer. I'll resubmit
once we're able to test that bit.
Nathan
** Sent via the linuxppc64-dev mail list. See http://lists.linuxppc.org/