RFR (XS): 8220388: Increase -inlinehint-threshold for Clang to avoid G1 pause time regression

Aleksey Shipilev shade at redhat.com
Sat Mar 9 08:25:39 UTC 2019


On 3/9/19 3:46 AM, Man Cao wrote:
> Webrev: https://cr.openjdk.java.net/~manc/8220388/webrev.00/
> RFE: https://bugs.openjdk.java.net/browse/JDK-8220388

I just briefly looked into the webrev. I think the "usual" place to do these adjustments is
JvmOverrideFiles.gmk, on per-file basis. We did it for Shenandoah, and there are also plenty others:
  http://hg.openjdk.java.net/jdk/jdk/file/0324b3756aa2/make/hotspot/lib/JvmOverrideFiles.gmk#l39

Also note it makes sense for "release" builds only? In Shenandoah, we used to do this for all build
types, only to find that extended inlining blew up fastdebug binary sizes and the
compilation/linkage times.

-Aleksey

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <https://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20190309/7482beff/signature.asc>


More information about the hotspot-gc-dev mailing list