Request for Review (xs) - 8159073: Error handling incomplete when creating GC threads lazily
jon.masamitsu at oracle.com
Thu Jul 21 13:48:26 UTC 2016
On 7/21/2016 1:20 AM, Thomas Schatzl wrote:
> Hi Jon,
> On Wed, 2016-07-20 at 20:43 -0700, Jon Masamitsu wrote:
>> It's ready again.
>> Delta: http://cr.openjdk.java.net/~jmasa/8159073/webrev_delta_01_02/
>> Full: http://cr.openjdk.java.net/~jmasa/8159073/webrev.02/
> this looks good, except for the two log messages. They seem to be
> superfluous because first the callers of run_task(, uint) already print
> mostly the same message, and the given code guarantee's that whatever
> is printed anyway. Further, a single invocation of run_task will now
> trigger three messages basically telling the same.
> So I would prefer if these messages were either moved to trace level,
> or just removed.
> Thanks a lot for your patience with me, :)
I deleted the log messages.
@@ -272,11 +272,9 @@
task->name(), num_workers, total_workers());
guarantee(num_workers > 0, "Trying to execute task %s with zero
uint old_num_workers = _active_workers;
- log_debug(gc)("run_task: updating active workers for %s from %u to
%u", task->name(), old_num_workers, num_workers);
guarantee(_active_workers == num_workers, "active workers %u
num_workers %u", _active_workers, num_workers);
- log_debug(gc)("run_task: restoring active workers from %u to %u",
More information about the hotspot-gc-dev