ping: [11u] RFR(M): 8213250: CDS archive creation aborts due to metaspace object allocation failure
Lindenmaier, Goetz
goetz.lindenmaier at sap.com
Mon Mar 23 10:36:05 UTC 2020
Thanks Jiangli!
Best regards,
Goetz.
> -----Original Message-----
> From: Jiangli Zhou <jianglizhou at google.com>
> Sent: Saturday, March 21, 2020 1:59 AM
> To: Lindenmaier, Goetz <goetz.lindenmaier at sap.com>
> Cc: jdk-updates-dev at openjdk.java.net
> Subject: Re: ping: [11u] RFR(M): 8213250: CDS archive creation aborts due to
> metaspace object allocation failure
>
> On Thu, Mar 19, 2020 at 12:25 PM Jiangli Zhou <jianglizhou at google.com>
> wrote:
> >
> > Hi Goetz,
> >
> > The backport looks good to me. I will also run some tests on my side
> > for the backport. Will report back to you.
>
> Testing from my side looks good. Thanks!
>
> Best regards,
> Jiangli
>
> >
> > Best,
> > Jiangli
> >
> > On Thu, Mar 19, 2020 at 6:51 AM Lindenmaier, Goetz
> > <goetz.lindenmaier at sap.com> wrote:
> > >
> > > Hi,
> > >
> > > Could I please get a review for this downport?
> > >
> > > Thanks!
> > > Goetz
> > >
> > > > -----Original Message-----
> > > > From: Lindenmaier, Goetz
> > > > Sent: Friday, March 13, 2020 3:21 PM
> > > > To: jdk-updates-dev at openjdk.java.net
> > > > Subject: [11u] RFR(M): 8213250: CDS archive creation aborts due to
> > > > metaspace object allocation failure
> > > >
> > > > Hi
> > > >
> > > > I would like to downport this [1] for parity with 11.0.8-oracle.
> > > >
> > > > I had to resolve because in one context the InstandeKlass/Klass type
> differs.
> > > >
> > > > Further the change depends on some previous changes in jdk/jdk.
> > > > I had to do the following changes in systemDictionaryShared.cpp:
> > > >
> > > > Because [2] "8209645: Split ClassLoaderData and ClassLoaderDataGraph
> into
> > > > separate files"
> > > > is not in 11, the include of ClassLoaderDataGraph.hpp is not needed.
> > > >
> > > > I had to undo the effects of [3] "8209301: JVM rename is_anonymous,
> > > > host_klass to
> > > > unsafe specific terminology ahead of Unsafe.defineAnonymousClass
> > > > deprecation":
> > > > is_unsafe_anonymous() --> is_anonymous()
> > > >
> > > > The new SystemDictionaryShared::finalize_verification_constraints()
> locks
> > > > the
> > > > ClassLoaderDataGraph_lock. This lock was introduced by [4] "8210155:
> Lock
> > > > ClassLoaderDataGraph Summary: In preparation for concurrent class
> > > > unloading." and
> > > > is not in 11. As I understand, 11 has no concurrent class loading, so this is
> not
> > > > needed.
> > > > Removed.
> > > >
> > > > This webrev shows the edits I did after applying the change:
> > > > http://cr.openjdk.java.net/~goetz/wr20/8213250-CDS_aborts-
> jdk11/01-
> > > > edits/
> > > >
> > > > This is the resulting webrev:
> > > > http://cr.openjdk.java.net/~goetz/wr20/8213250-CDS_aborts-
> jdk11/01/
> > > >
> > > > Please review.
> > > >
> > > > Best regards,
> > > > Goetz.
> > > >
> > > > [1]
> > > > Original bug:
> > > > https://bugs.openjdk.java.net/browse/JDK-8213250
> > > > http://hg.openjdk.java.net/jdk/jdk/rev/e0fd97beab7e
> > > > [2]
> > > > 8209645: Split ClassLoaderData and ClassLoaderDataGraph into separate
> files
> > > > https://bugs.openjdk.java.net/browse/JDK-8209645
> > > > http://hg.openjdk.java.net/jdk/jdk/rev/db0c3952de52
> > > > [3]
> > > > 8209301: JVM rename is_anonymous, host_klass to unsafe specific
> > > > terminology ahead
> > > > of Unsafe.defineAnonymousClass deprecation
> > > > https://bugs.openjdk.java.net/browse/JDK-8209301
> > > > http://hg.openjdk.java.net/jdk/jdk/rev/3e5d28e6de32
> > > > [4]
> > > > 8210155: Lock ClassLoaderDataGraph Summary: In preparation for
> concurrent
> > > > class unloading
> > > > https://bugs.openjdk.java.net/browse/JDK-8210155
> > > > http://hg.openjdk.java.net/jdk/jdk/rev/625a5bdde0c5
> > > >
> > > >
> > >
More information about the jdk-updates-dev
mailing list