RFR (XXS): 8152400: Enabling TASK_STATS_ONLY filters out just enabled messages anyway

Per Liden per.liden at oracle.com
Wed Mar 23 03:08:13 UTC 2016


> On 22 Mar 2016, at 21:05, Thomas Schatzl <thomas.schatzl at oracle.com> wrote:
> 
> Hi all,
> 
>  can I have reviews for the following one-liner?
> 
> If you enable TASK_QUEUE_STATS for a product build for analysis, you
> can't immediately use them. You also need to change another line of
> code from log_develop_is_enabled() to log_is_enabled().
> 
> This seems strange: if I want the task queue stats, I surely intend to
> use it, and since that latter condition only filters out the printing
> itself, the overhead has already been incurred anyway.
> 
> This change changes the log_develop_is_enabled() check to
> log_is_enabled() to avoid an additional round of changes after noticing
> that task queue stats are not printed.
> 
> There is no difference in performance for product builds, as the whole
> code is compiled out, in non-product builds there is an extra trivial
> condition to be checked per gc. I do not think this is noticeable at
> all.
> 
> CR:
> https://bugs.openjdk.java.net/browse/JDK-8152400
> Webrev:
> http://cr.openjdk.java.net/~tschatzl/8152400/webrev/ <http://cr.openjdk.java.net/~tschatzl/8152400/webrev/>

Looks good.

Per

> Testing:
> local testing, local building
> 
> Thanks,
>  Thomas

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20160323/e7ce376c/attachment.html>


More information about the hotspot-gc-dev mailing list