[11u] Backport of fix for CDS and JVM-TI agent crash ?

Sun Jun 23 20:33:37 UTC 2019

Hi Christoph,

Just an update on this effort:

https://bugs.openjdk.java.net/browse/JDK-8210926 has been backported to
11u. I've also labeled all dependencies (with google-interest) related to
JDK-8212200 and JDK-8213250, including additional bug fixes and
enhancements pulled in for the dependencies. There are total 33 bug fixes
and enhances. I'll go through them to backport to 11u. Please feel free to
take any of them if you can.

With this set of fixes and enhancements, it will bring in the startup
performance improvements using Java object archiving (enabled with the
default CDS) to OpenJDK 11 update. Users will see a nice speedup with the
JVM startup time.

Best regards,
Jiangli

On Wed, Jun 19, 2019 at 7:44 AM Jiangli Zhou <jianglizhou at google.com> wrote:

> My pleasure��.
>
> Best regards,
> Jiangli
>
> On Wed, Jun 19, 2019 at 12:43 AM Langer, Christoph <
> christoph.langer at sap.com> wrote:
>
>> Hi Jiangli,
>>
>>
>>
>> sounds good to me. Thanks for your effort with this ��
>>
>>
>>
>> Cheers
>>
>> Christoph
>>
>>
>>
>> *From:* Jiangli Zhou <jianglizhou at google.com>
>> *Sent:* Mittwoch, 19. Juni 2019 00:49
>> *To:* Langer, Christoph <christoph.langer at sap.com>
>> *Cc:* Michael Rasmussen <Michael.Rasmussen at roguewave.com>;
>> jdk-updates-dev at openjdk.java.net; Ioi Lam <ioi.lam at oracle.com>
>> *Subject:* Re: [11u] Backport of fix for CDS and JVM-TI agent crash ?
>>
>>
>>
>> https://bugs.openjdk.java.net/browse/JDK-8210926 can be applied cleanly
>> with no dependencies. I've added jdk11u-fix-request.
>>
>>
>>
>> Best regards,
>>
>> Jiangli
>>
>>
>>
>> On Tue, Jun 18, 2019 at 3:25 PM Jiangli Zhou <jianglizhou at google.com>
>> wrote:
>>
>> Hi Christoph and others,
>>
>>
>>
>> I did investigations for backporting the JDK-8212200 and JDK-8213250 changes.
>> There are some dependencies on Java object archiving work. The cleanest way
>> to handle the backports is to pull in all the dependencies one by one to
>> OpenJDK 11. It helps minimize the merge changes. It also helps JVM startup
>> time in JDK 11 by pulling in the additional optimizations. Thoughts from
>> anyone?
>>
>>
>>
>> Best regards,
>>
>> Jiangli
>>
>>
>>
>> On Tue, Jun 18, 2019 at 6:20 AM Langer, Christoph <
>> christoph.langer at sap.com> wrote:
>>
>> Hi Jiangli,
>>
>>
>>
>> thanks for taking this over. I tried to pick a few of the dependent fixes
>> but it got too much for me. ��
>>
>>
>>
>> Best regards
>>
>> Christoph
>>
>>
>>
>> *From:* Jiangli Zhou <jianglizhou at google.com>
>> *Sent:* Dienstag, 18. Juni 2019 01:14
>> *To:* Langer, Christoph <christoph.langer at sap.com>
>> *Cc:* Michael Rasmussen <Michael.Rasmussen at roguewave.com>;
>> jdk-updates-dev at openjdk.java.net; Ioi Lam <ioi.lam at oracle.com>
>> *Subject:* Re: [11u] Backport of fix for CDS and JVM-TI agent crash ?
>>
>>
>>
>> Hi Christoph,
>>
>>
>>
>> I'll help out (maybe slow). JDK-8213250 has quite a few dependencies that
>> needs to be backported first, but most can be done cleanly.
>>
>>
>>
>> There is also the https://bugs.openjdk.java.net/browse/JDK-8210926,
>> which I think should be backported to JDK 11 as well. The issue can surface
>> up when the default CDS is used with JVMTI.
>>
>>
>>
>> I'll start from the easy ones first.
>>
>>
>>
>> Best regards,
>>
>>
>>
>> Jiangli
>>
>>
>>
>> On Mon, Jun 17, 2019 at 8:17 AM Langer, Christoph <
>> christoph.langer at sap.com> wrote:
>>
>> Hi Michael,
>>
>> I looked into backports for JDK-8212200 and JDK-8213250 but I'm not able
>> to do them. The code level between current JDK11u hotspot and the jdk/jdk
>> repository at the time the patches were pushed (JDK12) has diverged too
>> much. So, it's not a simple resolve and test thing. People with more
>> experience in that area will have to do it. I don't have the required
>> expertise and also no time to dig deeper into it.
>>
>> So, maybe somebody is willing to step up here - otherwise I'm afraid
>> these bugs need to remain unfixed in OpenJDK 11 for the time being.
>>
>> Best regards
>> Christoph
>>
>> > -----Original Message-----
>> > From: Langer, Christoph
>> > Sent: Mittwoch, 20. März 2019 16:13
>> > To: 'Ioi Lam' <ioi.lam at oracle.com>
>> > Cc: jdk-updates-dev at openjdk.java.net; Michael Rasmussen
>> > <Michael.Rasmussen at roguewave.com>
>> > Subject: RE: [11u] Backport of fix for CDS and JVM-TI agent crash ?
>> >
>> > Hi Ioi,
>> >
>> > thanks again for the update. I'll take care of these backports then.
>> >
>> > Thanks
>> > Christoph
>> >
>> > > -----Original Message-----
>> > > From: Ioi Lam <ioi.lam at oracle.com>
>> > > Sent: Dienstag, 19. März 2019 22:38
>> > > To: Langer, Christoph <christoph.langer at sap.com>
>> > > Cc: jdk-updates-dev at openjdk.java.net; Michael Rasmussen
>> > > <Michael.Rasmussen at roguewave.com>
>> > > Subject: Re: [11u] Backport of fix for CDS and JVM-TI agent crash ?
>> > >
>> > > The 3 issues that I wanted to backport were:
>> > >
>> > > Issue       BP-of   Synopsis
>> > > JDK-8212991 8212205 VM asserts after CDS archive has been unmapped
>> > > JDK-8213273 8213250 CDS archive creation aborts due to metaspace
>> object
>> > > allocation failure
>> > > JDK-8213164 8212200 assert(on_stack()) failed when shared
>> > > java.lang.object is redefined by JVMTI agent
>> > >
>> > > They are all backport issues related to CDS with (fixVersion =
>> 11-pool).
>> > > Only the last one is related to JVMTI, but the other 2 are kind of
>> > > serious as well. But due to reassignment, I wasn't able to do the
>> > > backport :-(
>> > >
>> > > Thanks
>> > > - Ioi
>> > >
>> > > On 3/18/19 2:08 PM, Langer, Christoph wrote:
>> > > > Hi Ioi,
>> > > >
>> > > > ok, thanks for that information.
>> > > >
>> > > > Can you let us know, what exactly these 3 bugs were, that should be
>> > > backported? I guess it's about JDK-8212200, JDK-8213182 which fixes a
>> build
>> > > failure after the former one. And what's the 3rd issue?
>> > > >
>> > > > Thanks
>> > > > Christoph
>> > > >
>> > > >> -----Original Message-----
>> > > >> From: Ioi Lam <ioi.lam at oracle.com>
>> > > >> Sent: Montag, 18. März 2019 19:27
>> > > >> To: Langer, Christoph <christoph.langer at sap.com>
>> > > >> Cc: jdk-updates-dev at openjdk.java.net; Michael Rasmussen
>> > > >> <Michael.Rasmussen at roguewave.com>
>> > > >> Subject: Re: [11u] Backport of fix for CDS and JVM-TI agent crash ?
>> > > >>
>> > > >> Hi Christoph, I have been reassigned (and I've removed myself as
>> the
>> > > >> assigned engineer for those bugs) so I think it's best for the
>> openjdk
>> > > >> community to pick up these backports.
>> > > >>
>> > > >> Thanks
>> > > >>
>> > > >> - Ioi
>> > > >>
>> > > >> On 3/17/19 11:46 PM, Langer, Christoph wrote:
>> > > >>> Hi Ioi,
>> > > >>>
>> > > >>> I'd like to ping you again on this one. What are your current
>> plans on
>> > > >> backporting https://bugs.openjdk.java.net/browse/JDK-8212200 and
>> > > >> whatever needs to go with it? Ideally you still want to do it, but
>> if not I'd
>> > > >> appreciate if you could let us know.
>> > > >>> Thanks a lot in advance
>> > > >>> Christoph
>> > > >>>
>> > > >>>
>> > > >>>> -----Original Message-----
>> > > >>>> From: jdk-updates-dev <jdk-updates-dev-
>> > > bounces at openjdk.java.net>
>> > > >> On
>> > > >>>> Behalf Of Ioi Lam
>> > > >>>> Sent: Dienstag, 29. Januar 2019 01:33
>> > > >>>> To: jdk-updates-dev at openjdk.java.net
>> > > >>>> Subject: Re: Backport of fix for CDS and JVM-TI agent crash ?
>> > > >>>>
>> > > >>>> There are CDS 3 bugs that I plan to backport to 11. I hope to
>> post the
>> > > >>>> requests this week.
>> > > >>>>
>> > > >>>> https://bugs.openjdk.java.net/issues/?filter=36209
>> > > >>>>
>> > > >>>> Thanks
>> > > >>>>
>> > > >>>> - Ioi
>> > > >>>>
>> > > >>>> On 1/28/19 8:35 AM, Lindenmaier, Goetz wrote:
>> > > >>>>> Hi,
>> > > >>>>>
>> > > >>>>> The fact a downport bug was opened does not matter,
>> > > >>>>> you could still flag jdk11u-fix-request and just push it
>> > > >>>>> if approved.  This bug will then be closed as it is flagged
>> 11-pool.
>> > > >>>>>
>> > > >>>>> More gentle it would be to ask Ioi what's going on ...
>> > > >>>>>
>> > > >>>>> Best regards,
>> > > >>>>>      Goetz.
>> > > >>>>>
>> > > >>>>>> -----Original Message-----
>> > > >>>>>> From: jdk-updates-dev <jdk-updates-dev-
>> > > >> bounces at openjdk.java.net>
>> > > >>>> On
>> > > >>>>>> Behalf Of Volker Simonis
>> > > >>>>>> Sent: Monday, January 28, 2019 5:02 PM
>> > > >>>>>> To: Michael Rasmussen <Michael.Rasmussen at roguewave.com>
>> > > >>>>>> Cc: jdk-updates-dev at openjdk.java.net
>> > > >>>>>> Subject: Re: Backport of fix for CDS and JVM-TI agent crash ?
>> > > >>>>>>
>> > > >>>>>> On Mon, Jan 28, 2019 at 4:17 PM Michael Rasmussen
>> > > >>>>>> <Michael.Rasmussen at roguewave.com> wrote:
>> > > >>>>>>> Hi,
>> > > >>>>>>>
>> > > >>>>>>> I was wondering if there are any plans to backport
>> > > >>>>>> https://bugs.openjdk.java.net/browse/JDK-8212200 to jdk11u.
>> > > >>>>>>> I can see https://bugs.openjdk.java.net/browse/JDK-8213164
>> has
>> > > >> been
>> > > >>>>>> created, but don't know if there are any active plans to
>> actually do
>> > it?
>> > > >>>>>> Hi Michael,
>> > > >>>>>>
>> > > >>>>>> I actually don't really understand what this means. Usually,
>> you do
>> > a
>> > > >>>>>> down-port by first requesting it on the original bug by
>> applying a
>> > > >>>>>> "jdk<release>u-fix-request" tag. Once the downport will be
>> > > approved
>> > > >>>>>> (by application of a "jdk<release>u-fix-yes" tag), the
>> > corresponding
>> > > >>>>>> fix can be pushed to the updates repository. This push
>> > automatically
>> > > >>>>>> triggers the creation of the corresponding backport issue in
>> JBS.
>> > > >>>>>>
>> > > >>>>>> I don't know what it means for the downport process [1] if
>> > > somebody
>> > > >>>>>> (apparently manually) creates a backport issue like
>> > > >>>>>> https://bugs.openjdk.java.net/browse/JDK-8213164 ?
>> > > >>>>>>
>> > > >>>>>> Regards,
>> > > >>>>>> Volker
>> > > >>>>>>
>> > > >>>>>> [1] http://openjdk.java.net/projects/jdk-updates/approval.html
>> > > >>>>>>
>> > > >>>>>>> Kind regards
>> > > >>>>>>> /Michael
>>
>>