[aarch64-port-dev ] RFR: Critical Shenandoah patch for interpreter

[Andrew Hughes]

----- Original Message -----
> Am Donnerstag, den 14.07.2016, 17:36 +0200 schrieb Roman Kennke:
> > The aarch64-port/jdk8u-shenandoah forest already has the ztos
> > handling
> > code in the interpreter, that we don't yet have in Shenandoah's jdk8u
> > repository, and for that reason was lacking a write-barrier in the
> > bool
> > putfield handler.
> > 
> > http://cr.openjdk.java.net/~rkennke/interpr-boolean-wb/webrev.00/
> > 
> > You might need to do the same for aarch64's interpreter?
> 
> Ping?
> 

I think we need to clarify who should be reviewing patches to this
aarch64/jdk8u-shenandoah tree.  If we are going by who is eligible
from the aarch64 project [0], then only Andrew Haley can approve
such patches. There are no reviewers and this is not the first time
patches have stalled.

> Also, the following changesets need to be included into the integration
> repository:
> 
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/51d8c9b5de87
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/bf65b0423b2d
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/306ff307a68d
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/80821b48a837
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/b67217ccdc3c
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/01d7999614f6
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/1e2c94c741bb
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/9b261ed40799
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/fa985a203d39
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/a75014eb896d
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/7cc5667f2121
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/597d20bb52fb
> http://hg.openjdk.java.net/shenandoah/jdk8u/hotspot/rev/92d0c4de9788
> 
> Those fix:
> - Growing of the heap (i.e. when -Xms and -Xmx are different)
> - Add some missing barriers
> - Make CAS-obj a loop
> - Fix freeze in JNI critical region support code
> 
> All of those are straightforward backports of corresponding code in
> jdk9, where the patches have baked for a while. All of them fix some
> important issues (freezes, failures in concurrency code, etc). Can
> those be included into the integration repository, together with the
> original boolean putfield patch asap?

Are you planning to backport these yourself and post a webrev, or
do you want me to include them when importing aarch64-jdk8u101-b13?

> 
> Roman
> 

[0] http://openjdk.java.net/census#aarch64-port
-- 
Andrew :)

Senior Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

PGP Key: ed25519/35964222 (hkp://keys.gnupg.net)
Fingerprint = 5132 579D D154 0ED2 3E04  C5A0 CFDA 0F9B 3596 4222