RFR: JDK-8212603: Need to step over GC barriers in Node::eqv_uncast()
rkennke at redhat.com
Wed Oct 17 20:33:19 UTC 2018
Thanks Vladimir for the review! I moved the method and will push this
.. as soon as jdk/submit says 'green' :-)
Am 17.10.18 um 20:49 schrieb Vladimir Kozlov:
> Good to me too. I would only suggest to move Node::eqv_uncast() body
> near Node::uncast() in node.cpp
> On 10/17/18 11:39 AM, Erik Österlund wrote:
>> Hi Roman,
>> Looks good.
>> On 2018-10-17 19:54, Roman Kennke wrote:
>>> Node::eqv_uncast() checks if two nodes are equal (equivalent) behind
>>> casts. It's used in many places concerning lock elimination. The trouble
>>> is if the actual nodes are behind GC barriers, and we get the same node
>>> behind two different GC barrier nodes, this would return false negative.
>>> We have seen a bad case of this with Shenandoah, where lock elimination
>>> was subtly thrown off by this, which led to eliminated locks not
>>> re-locked properly during deoptimization.
>>> I propose to also strip any possible GC barriers like this:
>>> Testing: hotspot/tier1, will push through jdk/submit in a bit
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 833 bytes
Desc: OpenPGP digital signature
More information about the hotspot-compiler-dev