FW: RFR(XXS): 8213172: CDS and JFR tests fail with assert(JdkJfrEvent::is(klass)) failed: invariant
Markus Gronlund
markus.gronlund at oracle.com
Wed Oct 31 08:05:14 UTC 2018
Thanks a lot David!
Markus
-----Original Message-----
From: David Holmes
Sent: den 31 oktober 2018 09:05
To: Markus Gronlund <markus.gronlund at oracle.com>; hotspot-runtime-dev at openjdk.java.net
Subject: Re: FW: RFR(XXS): 8213172: CDS and JFR tests fail with assert(JdkJfrEvent::is(klass)) failed: invariant
Looks good Markus - and trivial! Ship it!
Thanks,
David
On 31/10/2018 5:50 PM, Markus Gronlund wrote:
> Sending this to a larger set of people, in order to get a fix to the CI pipeline sooner.
>
> Thanks
> Markus
>
> -----Original Message-----
> From: Markus Gronlund
> Sent: den 31 oktober 2018 07:28
> To: hotspot-jfr-dev at openjdk.java.net; Jiangli Zhou <jiangli.zhou at oracle.com>; Ioi Lam <ioi.lam at oracle.com>; Calvin Cheung <calvin.cheung at oracle.com>
> Subject: RFR(XXS): 8213172: CDS and JFR tests fail with assert(JdkJfrEvent::is(klass)) failed: invariant
>
> Hello,
>
> JDK-8203629 changed the top level JFR event class hierarchy from jdk.jfr.Event to jdk.internal.event.Event.
>
> Unfortunately the CDS exclusion filter for the JFR event classes was not updated accordingly, so let's do that here instead:
>
> Bug: https://bugs.openjdk.java.net/browse/JDK-8213172
> Webrev: http://cr.openjdk.java.net/~mgronlun/8213172/webrev01/
>
> This change makes the adjustment to dictionary.cpp:
>
> diff -r 385609cec1f8 src/hotspot/share/classfile/dictionary.cpp
> --- a/src/hotspot/share/classfile/dictionary.cpp Wed Oct 31 05:26:07 2018 +0100
> +++ b/src/hotspot/share/classfile/dictionary.cpp Wed Oct 31 06:05:04 2018 +0100
> @@ -468,7 +468,7 @@
> #if INCLUDE_CDS
> static bool is_jfr_event_class(Klass *k) {
> while (k) {
> - if (k->name()->equals("jdk/jfr/Event")) {
> + if (k->name()->equals("jdk/internal/event/Event")) {
> return true;
> }
> k = k->super();
>
>
> Tests:
> open/test/hotspot/jtreg/:hotspot_appcds
> tier1, tier2, tier3
>
> Thanks
> Markus
>
More information about the hotspot-runtime-dev
mailing list