RFR (XS): 8186480: Remove warning when AssumeMP is true and ergonomics determines to use one worker thread

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

RFR (XS): 8186480: Remove warning when AssumeMP is true and ergonomics determines to use one worker thread

Thomas Schatzl
Hi all,

  can I have reviews for this change that removes some imho useless
warning that confuses some tests?

In particular, if AssumeMP is true, and ergonomics determines that it
only wants to use one thread (e.g. in a container/VM), it gives a
*warning * about that AssumeMP and the ergonomics decision seems to
contradict each other.

The problem is that, some time ago AssumeMP has been made true default,
and already deprecated in 10 (JDK-8185062) and the plan is to obsolete
it, so this message shows up in some tests if run on e.g. a single-
thread VM.

I suggest to remove this message because of this and it does not seem
to be terribly useful: to get only one thread by ergonomics you need to
explicitly run the VM in an environment that only provides a single
thread. It seems logical that in this case the user can expect (given
he relies on ergonomics) that the GC will only use one thread.

And alternative could be to only show this message if the user
explicitly sets AssumeMP to true, but given that AssumeMP is on its way
out and the above reasoning about the usefulness of the message I think
it is okay to just remove that message.

CR:
https://bugs.openjdk.java.net/browse/JDK-8186480
Webrev:
http://cr.openjdk.java.net/~tschatzl/8186480/webrev/
Testing:
compilation

Thanks,
  Thomas

Reply | Threaded
Open this post in threaded view
|

Re: RFR (XS): 8186480: Remove warning when AssumeMP is true and ergonomics determines to use one worker thread

Aleksey Shipilev-4
On 11/09/2017 01:17 PM, Thomas Schatzl wrote:

> The problem is that, some time ago AssumeMP has been made true default,
> and already deprecated in 10 (JDK-8185062) and the plan is to obsolete
> it, so this message shows up in some tests if run on e.g. a single-
> thread VM.
>
> I suggest to remove this message because of this and it does not seem
> to be terribly useful: to get only one thread by ergonomics you need to
> explicitly run the VM in an environment that only provides a single
> thread. It seems logical that in this case the user can expect (given
> he relies on ergonomics) that the GC will only use one thread.
>
> CR:
> https://bugs.openjdk.java.net/browse/JDK-8186480
> Webrev:
> http://cr.openjdk.java.net/~tschatzl/8186480/webrev/
That makes sense. Looks good to me.

-Aleksey



signature.asc (836 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: RFR (XS): 8186480: Remove warning when AssumeMP is true and ergonomics determines to use one worker thread

Thomas Schatzl
Hi Aleksey,

On Thu, 2017-11-09 at 14:01 +0100, Aleksey Shipilev wrote:

> On 11/09/2017 01:17 PM, Thomas Schatzl wrote:
> > The problem is that, some time ago AssumeMP has been made true
> > default,
> >
> > [...]
> > CR:
> > https://bugs.openjdk.java.net/browse/JDK-8186480
> > Webrev:
> > http://cr.openjdk.java.net/~tschatzl/8186480/webrev/
>
> That makes sense. Looks good to me.
>

 thanks for your review.

Thomas