From rob.mckenna at oracle.com Wed Sep 12 19:57:49 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Wed, 12 Sep 2018 20:57:49 +0100 Subject: [11u Communication] jdk11u forest Message-ID: <20180912195749.GD6842@vimes> I'd like to announce the availibility of the JDK 11 Updates forest at: http://hg.openjdk.java.net/jdk-updates/jdk11u/ As with JDK9u & JDK10u, this forest has been opened under the jdk-updates project. As with jdk10u, once the fix has been approved [1] for inclusion in an update release and has subsequently been pushed to the open repo, a backport record will be created for a specific release. (e.g. 11.0.1) -Rob [1] http://openjdk.java.net/projects/jdk-updates/approval.html From li.jiang at oracle.com Thu Sep 13 08:04:02 2018 From: li.jiang at oracle.com (li.jiang at oracle.com) Date: Thu, 13 Sep 2018 16:04:02 +0800 Subject: [11u] RFA: JDK-8208746, JDK-8209775 and JDK-8210153 update currency data to ISO 4217 amendment 169 Message-ID: <3b324f68-8a8a-11e3-0172-9b59972c7e9f@oracle.com> Hi, We have updated the currency data to ISO 4217 amendment 169 in jdk/jdk mainline, so now I'd like to request the approval to backport the following two changes to 11u open repo. Bug: ISO 4217 Amendment #168 Update https://bugs.openjdk.java.net/browse/JDK-8208746 ISO 4217 Amendment #169 Update https://bugs.openjdk.java.net/browse/JDK-8209775 Review thread http://mail.openjdk.java.net/pipermail/i18n-dev/2018-August/002620.html Patch: http://hg.openjdk.java.net/jdk/jdk/raw-rev/67092c6be8e1 Bug: localized currency symbol of VES https://bugs.openjdk.java.net/browse/JDK-8210153 Review thread: http://mail.openjdk.java.net/pipermail/i18n-dev/2018-August/002624.html Patch: http://hg.openjdk.java.net/jdk/jdk/raw-rev/18ca918b4ed9 Note that JDK-8208746 and JDK-8209775 requires JDK-8210153 as a following-up fix to correct a missing currency symbol update. Both changes apply cleanly to 11u repo. Thank you. Best regards, Leo From sean.coffey at oracle.com Thu Sep 13 08:13:06 2018 From: sean.coffey at oracle.com (=?UTF-8?Q?Se=c3=a1n_Coffey?=) Date: Thu, 13 Sep 2018 09:13:06 +0100 Subject: [11u] RFA: JDK-8208746, JDK-8209775 and JDK-8210153 update currency data to ISO 4217 amendment 169 In-Reply-To: <3b324f68-8a8a-11e3-0172-9b59972c7e9f@oracle.com> References: <3b324f68-8a8a-11e3-0172-9b59972c7e9f@oracle.com> Message-ID: No requirement to send such an email Leo. The new process will be worked via bug database updates. I see that you've already added the necessary labels and justification statements. http://openjdk.java.net/projects/jdk-updates/approval.html regards, Sean. On 13/09/2018 09:04, li.jiang at oracle.com wrote: > Hi, > > We have updated the currency data to ISO 4217 amendment 169 in jdk/jdk > mainline, so now I'd like to request the approval to backport the > following two changes to 11u open repo. > > Bug: > ISO 4217 Amendment #168 Update > https://bugs.openjdk.java.net/browse/JDK-8208746 > ISO 4217 Amendment #169 Update > https://bugs.openjdk.java.net/browse/JDK-8209775 > > Review thread > http://mail.openjdk.java.net/pipermail/i18n-dev/2018-August/002620.html > > Patch: > http://hg.openjdk.java.net/jdk/jdk/raw-rev/67092c6be8e1 > > Bug: > localized currency symbol of VES > https://bugs.openjdk.java.net/browse/JDK-8210153 > > Review thread: > http://mail.openjdk.java.net/pipermail/i18n-dev/2018-August/002624.html > > Patch: > http://hg.openjdk.java.net/jdk/jdk/raw-rev/18ca918b4ed9 > > Note that JDK-8208746 and JDK-8209775 requires JDK-8210153 as a > following-up fix to correct a missing currency symbol update. > > Both changes apply cleanly to 11u repo. > > Thank you. > > Best regards, > Leo > From martinrb at google.com Wed Sep 19 23:15:30 2018 From: martinrb at google.com (Martin Buchholz) Date: Wed, 19 Sep 2018 16:15:30 -0700 Subject: [11u Communication] jdk11u forest In-Reply-To: <20180912195749.GD6842@vimes> References: <20180912195749.GD6842@vimes> Message-ID: Thanks for creating jdk11u. I'm happy it got created before jdk 11.0.0 GA, so backports from jdk head already have a place to soak. Things I'd like to see improved: - Rule 3 is too restrictive. P1 is too high a bar. It appears that jdk11 will have long-term broad ecosystem support, and any backports that make sense for a long lived stable release should be accepted. (But how to decide whether the risk is worth the benefit?! Humans will have to make tough decisions) - While it is understandable that the contents of security patches must be kept private, the security patches will likely be released along with garden variety bug fixes in jdk11u, and so the schedule for the 10.0.1 and 10.0.2 mini-trains should be public (e.g. I don't know what is the deadline to get fixes into 10.0.1). - The process documented at http://openjdk.java.net/projects/jdk-updates/approval.html is still not as clear as I'd like. Can backports be handled entirely without needing to send emails to jdk-updates-dev? On Wed, Sep 12, 2018 at 12:57 PM, Rob McKenna wrote: > I'd like to announce the availibility of the JDK 11 Updates forest at: > > http://hg.openjdk.java.net/jdk-updates/jdk11u/ > > As with JDK9u & JDK10u, this forest has been opened under the jdk-updates > project. > > As with jdk10u, once the fix has been approved [1] for inclusion in an > update release and has subsequently been pushed to the open repo, a backport > record will be created for a specific release. (e.g. 11.0.1) > > -Rob > > [1] http://openjdk.java.net/projects/jdk-updates/approval.html > From goetz.lindenmaier at sap.com Thu Sep 20 05:58:41 2018 From: goetz.lindenmaier at sap.com (Lindenmaier, Goetz) Date: Thu, 20 Sep 2018 05:58:41 +0000 Subject: [11u Communication] jdk11u forest In-Reply-To: References: <20180912195749.GD6842@vimes> Message-ID: <252adfa587234d72ba7d31fda74b8701@sap.com> Hi, I share the concerns about rule 3 in a LTE release. E.g., we would like to fix the s390 port to support JFR. This can not be considered a P1 bug. Another candidate is change "8207343: Automate vtable/itable stub size calculation" which fixes issues in the jdk11 dbg build. To jdk8, a row of intrinsics were downported that brought significant performance gains, similar issues will come up with jdk11. Best regards, Goetz. > -----Original Message----- > From: jdk-updates-dev On > Behalf Of Martin Buchholz > Sent: Thursday, September 20, 2018 1:16 AM > To: Rob McKenna > Cc: jdk-updates-dev at openjdk.java.net > Subject: Re: [11u Communication] jdk11u forest > > Thanks for creating jdk11u. I'm happy it got created before jdk > 11.0.0 GA, so backports from jdk head already have a place to soak. > > Things I'd like to see improved: > > - Rule 3 is too restrictive. P1 is too high a bar. It appears that > jdk11 will have long-term broad ecosystem support, and any backports > that make sense for a long lived stable release should be accepted. > (But how to decide whether the risk is worth the benefit?! Humans > will have to make tough decisions) > > - While it is understandable that the contents of security patches > must be kept private, the security patches will likely be released > along with garden variety bug fixes in jdk11u, and so the schedule for > the 10.0.1 and 10.0.2 mini-trains should be public (e.g. I don't know > what is the deadline to get fixes into 10.0.1). > > - The process documented at > http://openjdk.java.net/projects/jdk-updates/approval.html is still > not as clear as I'd like. Can backports be handled entirely without > needing to send emails to jdk-updates-dev? > > On Wed, Sep 12, 2018 at 12:57 PM, Rob McKenna > wrote: > > I'd like to announce the availibility of the JDK 11 Updates forest at: > > > > http://hg.openjdk.java.net/jdk-updates/jdk11u/ > > > > As with JDK9u & JDK10u, this forest has been opened under the jdk- > updates > > project. > > > > As with jdk10u, once the fix has been approved [1] for inclusion in an > > update release and has subsequently been pushed to the open repo, a > backport > > record will be created for a specific release. (e.g. 11.0.1) > > > > -Rob > > > > [1] http://openjdk.java.net/projects/jdk-updates/approval.html > > From volker.simonis at gmail.com Thu Sep 20 06:39:54 2018 From: volker.simonis at gmail.com (Volker Simonis) Date: Thu, 20 Sep 2018 08:39:54 +0200 Subject: [11u Communication] jdk11u forest In-Reply-To: <252adfa587234d72ba7d31fda74b8701@sap.com> References: <20180912195749.GD6842@vimes> <252adfa587234d72ba7d31fda74b8701@sap.com> Message-ID: While I completely share Goetz's and Martin's concerns, I think these are "just" Oracle's rules which only apply for the first two update releases (i.e. 11.0.1 and 11.0.2). From my understanding anybody who takes over the project lead after 11.0.2 (which might be already in about 4 month or so) can define whatever new rules he likes for the jdk11u project afterwards. Regards, Volker On Thu, Sep 20, 2018 at 7:59 AM Lindenmaier, Goetz wrote: > > Hi, > > I share the concerns about rule 3 in a LTE release. > E.g., we would like to fix the s390 port to support > JFR. This can not be considered a P1 bug. Another > candidate is change "8207343: Automate vtable/itable > stub size calculation" which fixes issues in the jdk11 > dbg build. > > To jdk8, a row of intrinsics were downported that > brought significant performance gains, similar issues > will come up with jdk11. > > Best regards, > Goetz. > > > > > > -----Original Message----- > > From: jdk-updates-dev On > > Behalf Of Martin Buchholz > > Sent: Thursday, September 20, 2018 1:16 AM > > To: Rob McKenna > > Cc: jdk-updates-dev at openjdk.java.net > > Subject: Re: [11u Communication] jdk11u forest > > > > Thanks for creating jdk11u. I'm happy it got created before jdk > > 11.0.0 GA, so backports from jdk head already have a place to soak. > > > > Things I'd like to see improved: > > > > - Rule 3 is too restrictive. P1 is too high a bar. It appears that > > jdk11 will have long-term broad ecosystem support, and any backports > > that make sense for a long lived stable release should be accepted. > > (But how to decide whether the risk is worth the benefit?! Humans > > will have to make tough decisions) > > > > - While it is understandable that the contents of security patches > > must be kept private, the security patches will likely be released > > along with garden variety bug fixes in jdk11u, and so the schedule for > > the 10.0.1 and 10.0.2 mini-trains should be public (e.g. I don't know > > what is the deadline to get fixes into 10.0.1). > > > > - The process documented at > > http://openjdk.java.net/projects/jdk-updates/approval.html is still > > not as clear as I'd like. Can backports be handled entirely without > > needing to send emails to jdk-updates-dev? > > > > On Wed, Sep 12, 2018 at 12:57 PM, Rob McKenna > > wrote: > > > I'd like to announce the availibility of the JDK 11 Updates forest at: > > > > > > http://hg.openjdk.java.net/jdk-updates/jdk11u/ > > > > > > As with JDK9u & JDK10u, this forest has been opened under the jdk- > > updates > > > project. > > > > > > As with jdk10u, once the fix has been approved [1] for inclusion in an > > > update release and has subsequently been pushed to the open repo, a > > backport > > > record will be created for a specific release. (e.g. 11.0.1) > > > > > > -Rob > > > > > > [1] http://openjdk.java.net/projects/jdk-updates/approval.html > > > From aph at redhat.com Thu Sep 20 08:41:52 2018 From: aph at redhat.com (Andrew Haley) Date: Thu, 20 Sep 2018 09:41:52 +0100 Subject: [11u Communication] jdk11u forest In-Reply-To: References: <20180912195749.GD6842@vimes> <252adfa587234d72ba7d31fda74b8701@sap.com> Message-ID: On 09/20/2018 07:39 AM, Volker Simonis wrote: > While I completely share Goetz's and Martin's concerns, I think these > are "just" Oracle's rules which only apply for the first two update > releases (i.e. 11.0.1 and 11.0.2). Yeah. P1-only seems much too strict for me. > From my understanding anybody who > takes over the project lead after 11.0.2 (which might be already in > about 4 month or so) can define whatever new rules he likes for the > jdk11u project afterwards. Yes, but we don't want it to be a free for all. I'm sure that we as a project can make sensible decisions about back porting. -- Andrew Haley Java Platform Lead Engineer Red Hat UK Ltd. EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671 From rob.mckenna at oracle.com Thu Sep 20 20:44:18 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Thu, 20 Sep 2018 21:44:18 +0100 Subject: [11u Communication] jdk11u forest In-Reply-To: References: <20180912195749.GD6842@vimes> <252adfa587234d72ba7d31fda74b8701@sap.com> Message-ID: <20180920204418.GA3760@tecra> Hi folks, Thanks for all the feedback. The maintainers are looking at the "high bar" issue and we'll have an update shortly. -Rob On 20/09/18 09:41, Andrew Haley wrote: > On 09/20/2018 07:39 AM, Volker Simonis wrote: > > While I completely share Goetz's and Martin's concerns, I think these > > are "just" Oracle's rules which only apply for the first two update > > releases (i.e. 11.0.1 and 11.0.2). > > Yeah. P1-only seems much too strict for me. > > > From my understanding anybody who > > takes over the project lead after 11.0.2 (which might be already in > > about 4 month or so) can define whatever new rules he likes for the > > jdk11u project afterwards. > > Yes, but we don't want it to be a free for all. I'm sure that we as a > project can make sensible decisions about back porting. > > -- > Andrew Haley > Java Platform Lead Engineer > Red Hat UK Ltd. > EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671 From rob.mckenna at oracle.com Thu Sep 20 20:46:13 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Thu, 20 Sep 2018 21:46:13 +0100 Subject: [11u Communication] jdk11u forest In-Reply-To: References: <20180912195749.GD6842@vimes> Message-ID: <20180920204613.GB3760@tecra> On 19/09/18 16:15, Martin Buchholz wrote: > Thanks for creating jdk11u. I'm happy it got created before jdk > 11.0.0 GA, so backports from jdk head already have a place to soak. > You're welcome, in fact it should have been created earlier and should be in future. > Things I'd like to see improved: > > - Rule 3 is too restrictive. P1 is too high a bar. It appears that > jdk11 will have long-term broad ecosystem support, and any backports > that make sense for a long lived stable release should be accepted. > (But how to decide whether the risk is worth the benefit?! Humans > will have to make tough decisions) We hear you. (see my other mail) > > - While it is understandable that the contents of security patches > must be kept private, the security patches will likely be released > along with garden variety bug fixes in jdk11u, and so the schedule for > the 10.0.1 and 10.0.2 mini-trains should be public (e.g. I don't know > what is the deadline to get fixes into 10.0.1). > Ah, I'll look at adding some language around this to the project page. Thanks. > - The process documented at > http://openjdk.java.net/projects/jdk-updates/approval.html is still > not as clear as I'd like. Can backports be handled entirely without > needing to send emails to jdk-updates-dev? > The project page is in need of a bit of TLC - I'll clear this up too. Thanks a lot Martin, -Rob > On Wed, Sep 12, 2018 at 12:57 PM, Rob McKenna wrote: > > I'd like to announce the availibility of the JDK 11 Updates forest at: > > > > http://hg.openjdk.java.net/jdk-updates/jdk11u/ > > > > As with JDK9u & JDK10u, this forest has been opened under the jdk-updates > > project. > > > > As with jdk10u, once the fix has been approved [1] for inclusion in an > > update release and has subsequently been pushed to the open repo, a backport > > record will be created for a specific release. (e.g. 11.0.1) > > > > -Rob > > > > [1] http://openjdk.java.net/projects/jdk-updates/approval.html > > From adinn at redhat.com Tue Sep 25 08:49:54 2018 From: adinn at redhat.com (Andrew Dinn) Date: Tue, 25 Sep 2018 09:49:54 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u Message-ID: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> Hi Rob, There are two critical AArch64 bugs in progress in jdk/jdk which are also present in jdk11u (JDK-8210858 and JDK-8210461). The generated machine code for maths intrinsics has been shown to contain bugs. The first bug has been fixed in jdk12 and we are pretty confident that the code has now been properly tested and is correct. the second has a proposed patch which is still being reviewed. We resolved this issue in the jdk11u-aarch64 releases built from our downstream repo with the following patches: http://pkgs.devel.redhat.com/cgit/rpms/java-11-openjdk/commit/?h=rhel-7.6&id=2d3a3f88c83016929e8624bc02685393e5d470f2 http://pkgs.devel.redhat.com/cgit/rpms/java-11-openjdk/commit/?h=rhel-7.6&id=038780bd09a09a3eb0aca70a0fbb341120b5f88f i.e. we disabled publishing of the generated assembler intrinsics. Rather than wait for a backport of the upstream fixes would it be possible to apply this patch to jdk11u now? That would allow other Linux distros to cherry-pick this patch for their own jdk11u releases. This is quite urgent as we want to be sure those distros get a chance to push an updated release before the next official jdk11u version is ready. I am not sure where this request sits as regards the current guidelines [1] for approving fixes nor what would be needed by way of JIRA to to cover this patch so your advice as to how we can proceed here would be very welcome. [1 http://openjdk.java.net/projects/jdk-updates/approval.html regards, Andrew Dinn ----------- Senior Principal Software Engineer Red Hat UK Ltd Registered in England and Wales under Company Registration No. 03798903 Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander From adinn at redhat.com Tue Sep 25 09:10:20 2018 From: adinn at redhat.com (Andrew Dinn) Date: Tue, 25 Sep 2018 10:10:20 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u In-Reply-To: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> References: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> Message-ID: <58c06db0-941a-e580-c314-e14db214a7f7@redhat.com> Oops. Apologies. Those were internal links! Corrections inline. On 25/09/18 09:49, Andrew Dinn wrote: > Hi Rob, > > There are two critical AArch64 bugs in progress in jdk/jdk which are > also present in jdk11u (JDK-8210858 and JDK-8210461). The generated > machine code for maths intrinsics has been shown to contain bugs. > > The first bug has been fixed in jdk12 and we are pretty confident that > the code has now been properly tested and is correct. the second has a > proposed patch which is still being reviewed. > > We resolved this issue in the jdk11u-aarch64 releases built from our > downstream repo with the following patches: https://src.fedoraproject.org/rpms/java-11-openjdk/blob/master/f/RHBZ-1628612-JDK-8210461-workaround-disable-aarch64-intrinsic.patch https://src.fedoraproject.org/rpms/java-11-openjdk/blob/master/f/RHBZ-1630996-JDK-8210858-workaround-disable-aarch64-intrinsic-log.patch > i.e. we disabled publishing of the generated assembler intrinsics. > > Rather than wait for a backport of the upstream fixes would it be > possible to apply this patch to jdk11u now? That would allow other Linux > distros to cherry-pick this patch for their own jdk11u releases. This is > quite urgent as we want to be sure those distros get a chance to push an > updated release before the next official jdk11u version is ready. > > I am not sure where this request sits as regards the current guidelines > [1] for approving fixes nor what would be needed by way of JIRA to to > cover this patch so your advice as to how we can proceed here would be > very welcome. > > [1 http://openjdk.java.net/projects/jdk-updates/approval.html regards, Andrew Dinn ----------- Senior Principal Software Engineer Red Hat UK Ltd Registered in England and Wales under Company Registration No. 03798903 Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander From aph at redhat.com Tue Sep 25 09:29:51 2018 From: aph at redhat.com (Andrew Haley) Date: Tue, 25 Sep 2018 10:29:51 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u In-Reply-To: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> References: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> Message-ID: <26c83f75-e4e2-37fd-1092-c4e5be4a9ed7@redhat.com> On 09/25/2018 09:49 AM, Andrew Dinn wrote: > I am not sure where this request sits as regards the current guidelines > [1] for approving fixes nor what would be needed by way of JIRA to to > cover this patch so your advice as to how we can proceed here would be > very welcome. The trick it to tag a P1 bug with "jdk11-fix-request". But I think we hit the last possible day on Friday. -- Andrew Haley Java Platform Lead Engineer Red Hat UK Ltd. EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671 From goetz.lindenmaier at sap.com Tue Sep 25 12:12:02 2018 From: goetz.lindenmaier at sap.com (Lindenmaier, Goetz) Date: Tue, 25 Sep 2018 12:12:02 +0000 Subject: How can I observe how my fix request is processed? Message-ID: <58ef737eb67040fc951ca0af62e49d43@sap.com> Hi, http://openjdk.java.net/projects/jdk-updates/approval.html does not say how I can observe what is going on with my fix request. E.g., I would like to know why https://bugs.openjdk.java.net/browse/JDK-8209786 which I tagged as jdk11u-fix-request is not handled. (Maybe I need to increase the Priority? I thought that this qualifies as "Fixes specific to downstream builders' projects".) Or is the procedure to just ask on this list? Best regards, Goetz. From rob.mckenna at oracle.com Tue Sep 25 13:38:02 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Tue, 25 Sep 2018 14:38:02 +0100 Subject: How can I observe how my fix request is processed? In-Reply-To: <58ef737eb67040fc951ca0af62e49d43@sap.com> References: <58ef737eb67040fc951ca0af62e49d43@sap.com> Message-ID: <20180925133802.GA2488@vimes> Hi Goetz! W.r.t. priority generally, the maintainers are discussing the push approval rules for jdk-updates this week and I'll get back to the lists with the results as soon as possible. In the mean time, this fix is somewhat awkward as it does have shared code so it isn't quite specific to a downstream builders project. I'll take a closer look however, as it seems pretty innocuous. -Rob On 25/09/18 12:12, Lindenmaier, Goetz wrote: > Hi, > > http://openjdk.java.net/projects/jdk-updates/approval.html does not say > how I can observe what is going on with my fix request. > > E.g., I would like to know why > https://bugs.openjdk.java.net/browse/JDK-8209786 > which I tagged as jdk11u-fix-request is not handled. > (Maybe I need to increase the Priority? I thought that > this qualifies as "Fixes specific to downstream builders' projects".) > > Or is the procedure to just ask on this list? > > Best regards, > Goetz. > From rob.mckenna at oracle.com Tue Sep 25 13:40:50 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Tue, 25 Sep 2018 14:40:50 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u In-Reply-To: <26c83f75-e4e2-37fd-1092-c4e5be4a9ed7@redhat.com> References: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> <26c83f75-e4e2-37fd-1092-c4e5be4a9ed7@redhat.com> Message-ID: <20180925134050.GB2488@vimes> Hi folks, Its not quite clear what you're aiming for here, but if (as I suspect) you want to get this into jdk11.0.X, please make sure to use the jdk11u-fix-request label. (note the 'u'!) As noted earlier, the maintainers are discussing relaxing the push approval criteria, but in any case these look like they qualify for the current high bar. (p1 / serious regression) -Rob On 25/09/18 10:29, Andrew Haley wrote: > On 09/25/2018 09:49 AM, Andrew Dinn wrote: > > I am not sure where this request sits as regards the current guidelines > > [1] for approving fixes nor what would be needed by way of JIRA to to > > cover this patch so your advice as to how we can proceed here would be > > very welcome. > > The trick it to tag a P1 bug with "jdk11-fix-request". But I think we hit the > last possible day on Friday. > > -- > Andrew Haley > Java Platform Lead Engineer > Red Hat UK Ltd. > EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671 From rob.mckenna at oracle.com Tue Sep 25 13:41:55 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Tue, 25 Sep 2018 14:41:55 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u In-Reply-To: <58c06db0-941a-e580-c314-e14db214a7f7@redhat.com> References: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> <58c06db0-941a-e580-c314-e14db214a7f7@redhat.com> Message-ID: <20180925134155.GC2488@vimes> Hi Andrew, Would you mind generating a webrev and posting to cr.openjdk.java.net or attaching the patches to email? (or the bug!) Thanks, -Rob On 25/09/18 10:10, Andrew Dinn wrote: > Oops. Apologies. Those were internal links! > > Corrections inline. > > On 25/09/18 09:49, Andrew Dinn wrote: > > Hi Rob, > > > > There are two critical AArch64 bugs in progress in jdk/jdk which are > > also present in jdk11u (JDK-8210858 and JDK-8210461). The generated > > machine code for maths intrinsics has been shown to contain bugs. > > > > The first bug has been fixed in jdk12 and we are pretty confident that > > the code has now been properly tested and is correct. the second has a > > proposed patch which is still being reviewed. > > > > We resolved this issue in the jdk11u-aarch64 releases built from our > > downstream repo with the following patches: > > > https://src.fedoraproject.org/rpms/java-11-openjdk/blob/master/f/RHBZ-1628612-JDK-8210461-workaround-disable-aarch64-intrinsic.patch > > > https://src.fedoraproject.org/rpms/java-11-openjdk/blob/master/f/RHBZ-1630996-JDK-8210858-workaround-disable-aarch64-intrinsic-log.patch > > > i.e. we disabled publishing of the generated assembler intrinsics. > > > > Rather than wait for a backport of the upstream fixes would it be > > possible to apply this patch to jdk11u now? That would allow other Linux > > distros to cherry-pick this patch for their own jdk11u releases. This is > > quite urgent as we want to be sure those distros get a chance to push an > > updated release before the next official jdk11u version is ready. > > > > I am not sure where this request sits as regards the current guidelines > > [1] for approving fixes nor what would be needed by way of JIRA to to > > cover this patch so your advice as to how we can proceed here would be > > very welcome. > > > > [1 http://openjdk.java.net/projects/jdk-updates/approval.html > > regards, > > > Andrew Dinn > ----------- > Senior Principal Software Engineer > Red Hat UK Ltd > Registered in England and Wales under Company Registration No. 03798903 > Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander From goetz.lindenmaier at sap.com Tue Sep 25 15:00:30 2018 From: goetz.lindenmaier at sap.com (Lindenmaier, Goetz) Date: Tue, 25 Sep 2018 15:00:30 +0000 Subject: How can I observe how my fix request is processed? In-Reply-To: <20180925133802.GA2488@vimes> References: <58ef737eb67040fc951ca0af62e49d43@sap.com> <20180925133802.GA2488@vimes> Message-ID: <562d2a13be214a3c8c7c2c83a1d344f6@sap.com> Hi Rob, thanks for the update! > In the mean time, this fix is somewhat awkward as it does have shared > code so it isn't quite specific to a downstream builders project. I'll take If " downstream builders project" excludes shared changes it's unclear to me what it is supposed to mean. I understood a "downstream builder" is anybody who builds the repo with a different setup than listed for Oracle among "JDK 11 build platforms supported by Oracle" here: https://wiki.openjdk.java.net/display/Build/Supported+Build+Platforms Maybe you need to update that wording, too? Best regards, Goetz. > -----Original Message----- > From: Rob McKenna > Sent: Dienstag, 25. September 2018 15:38 > To: Lindenmaier, Goetz > Cc: jdk-updates-dev at openjdk.java.net > Subject: Re: How can I observe how my fix request is processed? > > Hi Goetz! > > W.r.t. priority generally, the maintainers are discussing the push approval > rules for jdk-updates this week and I'll get back to the lists with the > results as soon as possible. > > In the mean time, this fix is somewhat awkward as it does have shared > code so it isn't quite specific to a downstream builders project. I'll take > a closer look however, as it seems pretty innocuous. > > -Rob > > On 25/09/18 12:12, Lindenmaier, Goetz wrote: > > Hi, > > > > http://openjdk.java.net/projects/jdk-updates/approval.html does not say > > how I can observe what is going on with my fix request. > > > > E.g., I would like to know why > > https://bugs.openjdk.java.net/browse/JDK-8209786 > > which I tagged as jdk11u-fix-request is not handled. > > (Maybe I need to increase the Priority? I thought that > > this qualifies as "Fixes specific to downstream builders' projects".) > > > > Or is the procedure to just ask on this list? > > > > Best regards, > > Goetz. > > From adinn at redhat.com Tue Sep 25 15:48:55 2018 From: adinn at redhat.com (Andrew Dinn) Date: Tue, 25 Sep 2018 16:48:55 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u In-Reply-To: <20180925134155.GC2488@vimes> References: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> <58c06db0-941a-e580-c314-e14db214a7f7@redhat.com> <20180925134155.GC2488@vimes> Message-ID: <4b17f4f8-b2b3-4539-7ad7-e22637d872c0@redhat.com> On 25/09/18 14:41, Rob McKenna wrote: > Would you mind generating a webrev and posting to cr.openjdk.java.net or > attaching the patches to email? (or the bug!) Thanks, No problem. diffs wrt the latest jdk11u are included below. The first diff indicates patch needed to stop the sin_cos intrinsic being published. The second diff inhibits publication of the log intrinsic. regards, Andrew Dinn ----------- Senior Principal Software Engineer Red Hat UK Ltd Registered in England and Wales under Company Registration No. 03798903 Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander ---- 8< ------- 8< ------- 8< ------- 8< ------- 8< ------- 8< --- diff -r 1ddf9a99e4ad src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp --- a/src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp? Wed Aug 22 21:50:12 2018 +0200 +++ b/src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp? Thu Sep 13 13:51:53 2018 +0100 @@ -5745,11 +5745,13 @@ } if (vmIntrinsics::is_intrinsic_available(vmIntrinsics::_dsin)) { - StubRoutines::_dsin = generate_dsin_dcos(/* isCos = */ false); + // disabled pending fix and retest of generated code + // StubRoutines::_dsin = generate_dsin_dcos(/* isCos = */ false); } if (vmIntrinsics::is_intrinsic_available(vmIntrinsics::_dcos)) { - StubRoutines::_dcos = generate_dsin_dcos(/* isCos = */ true); + // disabled pending fix and retest of generated code + // StubRoutines::_dcos = generate_dsin_dcos(/* isCos = */ true); } } ---- 8< ------- 8< ------- 8< ------- 8< ------- 8< ------- 8< --- diff --git a/src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp b/src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp --- a/src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp +++ b/src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp @@ -5741,7 +5741,8 @@ } if (vmIntrinsics::is_intrinsic_available(vmIntrinsics::_dlog)) { - StubRoutines::_dlog = generate_dlog(); + // disabled pending fix and retest of generated code + // StubRoutines::_dlog = generate_dlog(); } if (vmIntrinsics::is_intrinsic_available(vmIntrinsics::_dsin)) { ---- 8< ------- 8< ------- 8< ------- 8< ------- 8< ------- 8< --- From adinn at redhat.com Tue Sep 25 16:17:25 2018 From: adinn at redhat.com (Andrew Dinn) Date: Tue, 25 Sep 2018 17:17:25 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u In-Reply-To: <20180925134050.GB2488@vimes> References: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> <26c83f75-e4e2-37fd-1092-c4e5be4a9ed7@redhat.com> <20180925134050.GB2488@vimes> Message-ID: <99fda1e4-7e8e-88cf-64c7-ee2e4e8cbf48@redhat.com> Hi Rob, On 25/09/18 14:40, Rob McKenna wrote: > Its not quite clear what you're aiming for here, but if (as I suspect) > you want to get this into jdk11.0.X, please make sure to use the > jdk11u-fix-request label. (note the 'u'!) > > As noted earlier, the maintainers are discussing relaxing the push > approval criteria, but in any case these look like they qualify for > the current high bar. (p1 / serious regression) I have created the following JIRA and webrev: JIRA: https://bugs.openjdk.java.net/browse/JDK-8211105 webrev: http://cr.openjdk.java.net/~adinn/8211105/webrev.00/ n.b. the JIRA has the jdk11u-fix-request label set. If this is ok then I'll post an RFR to compiler-dev. If not please let me know what else I need to do. regards, Andrew Dinn ----------- Senior Principal Software Engineer Red Hat UK Ltd Registered in England and Wales under Company Registration No. 03798903 Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander From rob.mckenna at oracle.com Wed Sep 26 15:08:56 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Wed, 26 Sep 2018 16:08:56 +0100 Subject: AArch64: temporary fix for trig and log functions in jdk11u In-Reply-To: <99fda1e4-7e8e-88cf-64c7-ee2e4e8cbf48@redhat.com> References: <48af6d2d-b8dd-6c65-b077-4aad0b16b616@redhat.com> <26c83f75-e4e2-37fd-1092-c4e5be4a9ed7@redhat.com> <20180925134050.GB2488@vimes> <99fda1e4-7e8e-88cf-64c7-ee2e4e8cbf48@redhat.com> Message-ID: <20180926150856.GA3892@vimes> Looks good to me - if the code is different from 12 in any way please add a link to the review on openjdk to the bug. Thanks, -Rob On 25/09/18 17:17, Andrew Dinn wrote: > Hi Rob, > > On 25/09/18 14:40, Rob McKenna wrote: > > Its not quite clear what you're aiming for here, but if (as I suspect) > > you want to get this into jdk11.0.X, please make sure to use the > > jdk11u-fix-request label. (note the 'u'!) > > > > As noted earlier, the maintainers are discussing relaxing the push > > approval criteria, but in any case these look like they qualify for > > the current high bar. (p1 / serious regression) > I have created the following JIRA and webrev: > > JIRA: https://bugs.openjdk.java.net/browse/JDK-8211105 > webrev: http://cr.openjdk.java.net/~adinn/8211105/webrev.00/ > > n.b. the JIRA has the jdk11u-fix-request label set. > > If this is ok then I'll post an RFR to compiler-dev. If not please let > me know what else I need to do. > > regards, > > > Andrew Dinn > ----------- > Senior Principal Software Engineer > Red Hat UK Ltd > Registered in England and Wales under Company Registration No. 03798903 > Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander From rob.mckenna at oracle.com Wed Sep 26 15:13:32 2018 From: rob.mckenna at oracle.com (Rob McKenna) Date: Wed, 26 Sep 2018 16:13:32 +0100 Subject: How can I observe how my fix request is processed? In-Reply-To: <562d2a13be214a3c8c7c2c83a1d344f6@sap.com> References: <58ef737eb67040fc951ca0af62e49d43@sap.com> <20180925133802.GA2488@vimes> <562d2a13be214a3c8c7c2c83a1d344f6@sap.com> Message-ID: <20180926151332.GB3892@vimes> Examples of changes that are *specific* to downstream builders projects would be fixes that only affect a certain platform. (aarch64 for example) If the shared changes are minimal/harmless then a patch would usually qualify. -Rob On 25/09/18 15:00, Lindenmaier, Goetz wrote: > Hi Rob, > > thanks for the update! > > > In the mean time, this fix is somewhat awkward as it does have shared > > code so it isn't quite specific to a downstream builders project. I'll take > If " downstream builders project" excludes shared changes it's unclear to > me what it is supposed to mean. I understood a "downstream builder" > is anybody who builds the repo with a different setup than listed > for Oracle among "JDK 11 build platforms supported by Oracle" here: > https://wiki.openjdk.java.net/display/Build/Supported+Build+Platforms > Maybe you need to update that wording, too? > > Best regards, > Goetz. > > > > > > -----Original Message----- > > From: Rob McKenna > > Sent: Dienstag, 25. September 2018 15:38 > > To: Lindenmaier, Goetz > > Cc: jdk-updates-dev at openjdk.java.net > > Subject: Re: How can I observe how my fix request is processed? > > > > Hi Goetz! > > > > W.r.t. priority generally, the maintainers are discussing the push approval > > rules for jdk-updates this week and I'll get back to the lists with the > > results as soon as possible. > > > > In the mean time, this fix is somewhat awkward as it does have shared > > code so it isn't quite specific to a downstream builders project. I'll take > > a closer look however, as it seems pretty innocuous. > > > > -Rob > > > > On 25/09/18 12:12, Lindenmaier, Goetz wrote: > > > Hi, > > > > > > http://openjdk.java.net/projects/jdk-updates/approval.html does not say > > > how I can observe what is going on with my fix request. > > > > > > E.g., I would like to know why > > > https://bugs.openjdk.java.net/browse/JDK-8209786 > > > which I tagged as jdk11u-fix-request is not handled. > > > (Maybe I need to increase the Priority? I thought that > > > this qualifies as "Fixes specific to downstream builders' projects".) > > > > > > Or is the procedure to just ask on this list? > > > > > > Best regards, > > > Goetz. > > >