From thomas.schatzl at oracle.com Mon Mar 3 10:41:59 2025 From: thomas.schatzl at oracle.com (Thomas Schatzl) Date: Mon, 3 Mar 2025 11:41:59 +0100 Subject: Porting request/PR out: G1: Improve Application Throughput with a More Efficient Write-Barrier [Was: Re: Porting request for JEP Reduce Latency of G1 Write Barriers] In-Reply-To: <4503f576-8a54-44b2-9691-cdf089eee12e@oracle.com> References: <0e7ee9cf-87da-4706-982d-e08d2cfa732b@oracle.com> <4503f576-8a54-44b2-9691-cdf089eee12e@oracle.com> Message-ID: Hi all, > On 22.10.24 10:26, Thomas Schatzl wrote: >> Hi all, >> >> we in the GC team have been experimenting with new post-write >> barriers for G1 to reduce the throughput difference between G1 and >> Parallel GC. >> >> After quite a few attempts we think we found a balanced solution >> between overall complexity and impact on throughput and latency. >> >> [...] On 08.11.24 16:39, Thomas Schatzl wrote: > Hi all, > >[...] > > The implementation CR also already contains some more information about > the implemetation (https://bugs.openjdk.org/browse/JDK-8342382), which > will also be updated a little before asking for reviews. we in the Oracle gc team decided that due to the size of the change (+3k, -4.4k LOC) are not waiting any longer for the JEP moving to candidate before asking for reviews. There is a PR out for review for this change now [4]. I would like to ask you maintainers to have a look/test. M. Doerr from SAP already confirmed that the change is still working :)(thanks!). I am not expecting any or any serious issues for the other platforms since not much changed since asking for the ports. However in particular I hope for feedback from maintainers for the ARM (32 bit) and the RISCV port, as I have only been able to perform very limited testing using cross-compilers and emulators there. Another confirmation for the S390 port that things still work would be appreciated too. :) Thanks, Thomas [4] https://github.com/openjdk/jdk/pull/23739 From Amit.Kumar220 at ibm.com Mon Mar 3 14:26:02 2025 From: Amit.Kumar220 at ibm.com (Amit Kumar) Date: Mon, 3 Mar 2025 14:26:02 +0000 Subject: Porting request/PR out: G1: Improve Application Throughput with a More Efficient Write-Barrier [Was: Re: Porting request for JEP Reduce Latency of G1 Write Barriers] In-Reply-To: References: <0e7ee9cf-87da-4706-982d-e08d2cfa732b@oracle.com> <4503f576-8a54-44b2-9691-cdf089eee12e@oracle.com> Message-ID: Hi Thomas, > Another confirmation for the S390 port that things still work would be appreciated too. :) I have posted on PR as well. I ran tier1 tests and result still looks good on s390x. Thanks, /Amit > On 03/03/2025, at 4:11?PM, Thomas Schatzl wrote: > > Hi all, > > > On 22.10.24 10:26, Thomas Schatzl wrote: > >> Hi all, > >> > >> we in the GC team have been experimenting with new post-write > >> barriers for G1 to reduce the throughput difference between G1 and > >> Parallel GC. > >> > >> After quite a few attempts we think we found a balanced solution > >> between overall complexity and impact on throughput and latency. > >> > >> [...] > > On 08.11.24 16:39, Thomas Schatzl wrote: >> Hi all, > >[...] >> The implementation CR also already contains some more information about the implemetation (https://bugs.openjdk.org/browse/JDK-8342382 ), which will also be updated a little before asking for reviews. > > we in the Oracle gc team decided that due to the size of the change (+3k, -4.4k LOC) are not waiting any longer for the JEP moving to candidate before asking for reviews. > > There is a PR out for review for this change now [4]. I would like to ask you maintainers to have a look/test. > > M. Doerr from SAP already confirmed that the change is still working :)(thanks!). > > I am not expecting any or any serious issues for the other platforms since not much changed since asking for the ports. However in particular I hope for feedback from maintainers for the ARM (32 bit) and the RISCV port, as I have only been able to perform very limited testing using cross-compilers and emulators there. > > Another confirmation for the S390 port that things still work would be appreciated too. :) > > Thanks, > Thomas > > [4] https://github.com/openjdk/jdk/pull/23739 From harald+jdklists at volse.no Wed Mar 5 12:03:01 2025 From: harald+jdklists at volse.no (Harald Eilertsen) Date: Wed, 5 Mar 2025 13:03:01 +0100 Subject: Parameterize pthread and iconv flags Message-ID: <7k6keygfbgde67zitfkysfa7w4cnutw23w3mu4t7f5f5qjlnjh@czxl2jkm7eim> Hi, As suggested by @magicus, I have split out a couple of trivial patches to move the compiler and linker flags for pthread and iconv support to be pupulated by the configure scripts. As any PR needs to have an assigned bug ID, I have tried to submit enhancement requests for these through bugreport.java.com. Not sure how long time that usually takes before it get's published (or I hear anything back,) but if anyone could have a look, the internal ID's are 9078187 (pthread) and 9078196 (iconv). For now, both of these will probably seem entirely unnecessary, but they are needed for the BSD port, and it was suggested to split them out and merge them to prepare for the port. Take care! Harald From david.holmes at oracle.com Thu Mar 6 04:17:43 2025 From: david.holmes at oracle.com (David Holmes) Date: Thu, 6 Mar 2025 14:17:43 +1000 Subject: Parameterize pthread and iconv flags In-Reply-To: <7k6keygfbgde67zitfkysfa7w4cnutw23w3mu4t7f5f5qjlnjh@czxl2jkm7eim> References: <7k6keygfbgde67zitfkysfa7w4cnutw23w3mu4t7f5f5qjlnjh@czxl2jkm7eim> Message-ID: Hi Harald, On 5/03/2025 10:03 pm, Harald Eilertsen wrote: > Hi, > > As suggested by @magicus, I have split out a couple of trivial patches > to move the compiler and linker flags for pthread and iconv support to > be pupulated by the configure scripts. > > As any PR needs to have an assigned bug ID, I have tried to submit > enhancement requests for these through bugreport.java.com. Not sure how > long time that usually takes before it get's published (or I hear > anything back,) but if anyone could have a look, the internal ID's are > 9078187 (pthread) and 9078196 (iconv). I have requested these be transferred across to JBS. Cheers, David > For now, both of these will probably seem entirely unnecessary, but they > are needed for the BSD port, and it was suggested to split them out and > merge them to prepare for the port. > > Take care! > Harald From magnus.ihse.bursie at oracle.com Thu Mar 6 16:32:50 2025 From: magnus.ihse.bursie at oracle.com (Magnus Ihse Bursie) Date: Thu, 6 Mar 2025 17:32:50 +0100 Subject: Improving OpenJDK for FreeBSD In-Reply-To: References: <8bddd709-5747-4bd3-b6d5-dcd4b2e33886@oracle.com> Message-ID: <8bd44fbe-6efd-4e43-9683-e6be2c9458af@oracle.com> On 2025-02-21 04:03, David Holmes wrote: > On 21/02/2025 2:17 am, Magnus Ihse Bursie wrote: >> Hi Harald (and others), >> >> Great to see that you've made it this far! Sorry for the delayed >> responses, I've been ill for a while and are starting to work through >> my backlog now. >> >> On 2025-02-13 12:48, Harald Eilertsen wrote: >> >>> On Wed, Feb 05, 2025 at 11:25:51AM +1000, David Holmes wrote: >>>> Hi Harald, >>>> >>>> Can you create a PR against mainline so that we can see the scope >>>> of the >>>> changes and how they have been implemented? >>> Here you go: >>> https://github.com/openjdk/jdk/pull/23611 >>> >>> As mentioned in the notes for the PR, this is mainly ment to get your >>> feedback and as a draft to assess the scope, and to be a base for >>> discussing how (and if) to proceed with the port. >> >> This was a good move. When actually integrating any code, a different >> approach will be needed (for instance, the original writer/copyright >> holder must be clear for every line of code that should be integrated >> so we can check OCA acceptance). I do believe that a piecewise >> integration is much preferable to trying to add everything once. >> >> That will however require us to accept that we for a while are going >> to have a port in the mainline that might not even successfully >> build, even less work. But I think that is preferable to a big >> massive integration. > > I would not like to see it happen that way. If it is to happen then I > would prefer to see a project established and a project repo. This > phased approach can be applied to the project repo and supporting > changes upstreamed to mainline, then eventually the port itself will > be integrated to mainline. So are your suggested course of action to create a new bsd port repo, that starts out identical with mainline, and then integrate PRs into that, with usual review and JBS bug numbers, and when all PRs are in place, we can make a wholesale integration of that repo into mainline? I think that could work as well, but I guess the difficult part will be to get reviewers from the relevant areas. Making PRs to mainline will ensure they get on the potential reviewers radar. I'm afraid that posting PRs on a separate port will just have them end up with a bunch of "Anyone?" posted every week, and if someone do show up to review them, it might not be the right person for the job. But if you think we can solve that, I agree that it is probably a better solution. Also, I don't know if this has been said before, but this work requires a JEP. /Magnus > > Cheers, > David > >> That reasoning also indicates that maybe the first parts that should >> go in is the build system changes that are needed to even try to >> build the BSD port... >> >> /Magnus >> > From harald+jdklists at volse.no Thu Mar 6 18:38:24 2025 From: harald+jdklists at volse.no (Harald Eilertsen) Date: Thu, 6 Mar 2025 19:38:24 +0100 Subject: Improving OpenJDK for FreeBSD In-Reply-To: <8bd44fbe-6efd-4e43-9683-e6be2c9458af@oracle.com> References: <8bddd709-5747-4bd3-b6d5-dcd4b2e33886@oracle.com> <8bd44fbe-6efd-4e43-9683-e6be2c9458af@oracle.com> Message-ID: On Thu, Mar 06, 2025 at 05:32:50PM +0100, Magnus Ihse Bursie wrote: > On 2025-02-21 04:03, David Holmes wrote: > > I would not like to see it happen that way. If it is to happen then I > > would prefer to see a project established and a project repo. > > [...snip...] > > I think that could work as well, but I guess the difficult part will be to > get reviewers from the relevant areas. I'm happy to go whichever way the OpenJDK community thinks is the best approach. On advantage I see from our perspective with the separate project repo is that it may be easier to do a bit of experimentation and testing different approaches before a merge into the mainline. It will still be possible to do a phased approach to merging the changes into the mainline, so that should give any reviewers that didn't come around for the first merges into the project repo a second chance to voice their concerns. > Also, I don't know if this has been said before, but this work requires a > JEP. No, I haven't heard anything about that (I think). I'll read up on it, and get back if I need a hand to hold on to :) Take care! Harald From magnus.ihse.bursie at oracle.com Thu Mar 6 19:19:36 2025 From: magnus.ihse.bursie at oracle.com (Magnus Ihse Bursie) Date: Thu, 6 Mar 2025 20:19:36 +0100 Subject: Improving OpenJDK for FreeBSD In-Reply-To: References: <8bddd709-5747-4bd3-b6d5-dcd4b2e33886@oracle.com> <8bd44fbe-6efd-4e43-9683-e6be2c9458af@oracle.com> Message-ID: <251c2111-2923-4df0-a371-a61336230060@oracle.com> On 2025-03-06 19:38, Harald Eilertsen wrote: > On Thu, Mar 06, 2025 at 05:32:50PM +0100, Magnus Ihse Bursie wrote: >> On 2025-02-21 04:03, David Holmes wrote: >>> I would not like to see it happen that way. If it is to happen then I >>> would prefer to see a project established and a project repo. >> [...snip...] >> >> I think that could work as well, but I guess the difficult part will be to >> get reviewers from the relevant areas. > I'm happy to go whichever way the OpenJDK community thinks is the best > approach. On advantage I see from our perspective with the separate > project repo is that it may be easier to do a bit of experimentation and > testing different approaches before a merge into the mainline. > > It will still be possible to do a phased approach to merging the changes > into the mainline, so that should give any reviewers that didn't come > around for the first merges into the project repo a second chance to > voice their concerns. Formally, there is already a bsd-port project ("Port: BSD Project" is the formal name), and according to the Census, Greg Lewis is the Project Lead. That means he has the authority to request that a repo be setup for this project. I hope you have enough contact with him to ask him to send an email to ops at openjdk.org to request such a repo. Since the cogs of the OpenJDK administration moves slowly, I'd suggest trying get him to send such a request already; then the repo might be created in time for it to be actually needed. Formally, I believe it would be good if he also nominates you as member of the bsd-port project. In time, especially if your position at FreeBSD is turning out to be long-term, I think it would be good (and likely supported by Lewis) to have you moving into the Project Lead position of the BSD port project. There is also an old mailing list associated with the project; we might consider moving this discussion over there. (Otoh, the porters-dev list is basically empty so I don't think we're overwhelming everyone by keeping the discussion here as well.) > >> Also, I don't know if this has been said before, but this work requires a >> JEP. > No, I haven't heard anything about that (I think). I'll read up on it, > and get back if I need a hand to hold on to :) Have a look at e.g. JEP 388 (https://openjdk.org/jeps/388) which introduced the Windows/aarch64 port. That was really about combining an existing OS and an existing CPU, but I'd say the amount of changes required is similar to the BSD port, so I guess aiming at a JEP of similar complexity level is fine. /Magnus > > Take care! > Harald From harald+jdklists at volse.no Fri Mar 7 13:36:33 2025 From: harald+jdklists at volse.no (Harald Eilertsen) Date: Fri, 7 Mar 2025 14:36:33 +0100 Subject: Improving OpenJDK for FreeBSD In-Reply-To: <251c2111-2923-4df0-a371-a61336230060@oracle.com> References: <8bddd709-5747-4bd3-b6d5-dcd4b2e33886@oracle.com> <8bd44fbe-6efd-4e43-9683-e6be2c9458af@oracle.com> <251c2111-2923-4df0-a371-a61336230060@oracle.com> Message-ID: On Thu, Mar 06, 2025 at 08:19:36PM +0100, Magnus Ihse Bursie wrote: > Formally, there is already a bsd-port project ("Port: BSD Project" is the > formal name), and according to the Census, Greg Lewis is the Project Lead. Yep that's right. I've mentioned to him that it would be a good idea to create a project repo, but did not get a response yet. I'll check with him again. > > > Also, I don't know if this has been said before, but this work requires a > > > JEP. > > No, I haven't heard anything about that (I think). I'll read up on it, > > and get back if I need a hand to hold on to :) > > Have a look at e.g. JEP 388 (https://openjdk.org/jeps/388) Thanks! Take care! Harald From thomas.stuefe at gmail.com Fri Mar 7 14:16:01 2025 From: thomas.stuefe at gmail.com (=?UTF-8?Q?Thomas_St=C3=BCfe?=) Date: Fri, 7 Mar 2025 15:16:01 +0100 Subject: Improving OpenJDK for FreeBSD In-Reply-To: <251c2111-2923-4df0-a371-a61336230060@oracle.com> References: <8bddd709-5747-4bd3-b6d5-dcd4b2e33886@oracle.com> <8bd44fbe-6efd-4e43-9683-e6be2c9458af@oracle.com> <251c2111-2923-4df0-a371-a61336230060@oracle.com> Message-ID: @Magnus Ihse Bursie Why would we need a JEP? That is a lot of unnecessary red tape. The BSD port already exists. It is technically not a new port. On Thu, Mar 6, 2025 at 8:19?PM Magnus Ihse Bursie < magnus.ihse.bursie at oracle.com> wrote: > On 2025-03-06 19:38, Harald Eilertsen wrote: > > > On Thu, Mar 06, 2025 at 05:32:50PM +0100, Magnus Ihse Bursie wrote: > >> On 2025-02-21 04:03, David Holmes wrote: > >>> I would not like to see it happen that way. If it is to happen then I > >>> would prefer to see a project established and a project repo. > >> [...snip...] > >> > >> I think that could work as well, but I guess the difficult part will be > to > >> get reviewers from the relevant areas. > > I'm happy to go whichever way the OpenJDK community thinks is the best > > approach. On advantage I see from our perspective with the separate > > project repo is that it may be easier to do a bit of experimentation and > > testing different approaches before a merge into the mainline. > > > > It will still be possible to do a phased approach to merging the changes > > into the mainline, so that should give any reviewers that didn't come > > around for the first merges into the project repo a second chance to > > voice their concerns. > > Formally, there is already a bsd-port project ("Port: BSD Project" is > the formal name), and according to the Census, Greg Lewis is the Project > Lead. That means he has the authority to request that a repo be setup > for this project. I hope you have enough contact with him to ask him to > send an email to ops at openjdk.org to request such a repo. Since the cogs > of the OpenJDK administration moves slowly, I'd suggest trying get him > to send such a request already; then the repo might be created in time > for it to be actually needed. > > Formally, I believe it would be good if he also nominates you as member > of the bsd-port project. In time, especially if your position at FreeBSD > is turning out to be long-term, I think it would be good (and likely > supported by Lewis) to have you moving into the Project Lead position of > the BSD port project. > > There is also an old mailing list associated with the project; we might > consider moving this discussion over there. (Otoh, the porters-dev list > is basically empty so I don't think we're overwhelming everyone by > keeping the discussion here as well.) > > > > > >> Also, I don't know if this has been said before, but this work requires > a > >> JEP. > > No, I haven't heard anything about that (I think). I'll read up on it, > > and get back if I need a hand to hold on to :) > > Have a look at e.g. JEP 388 (https://openjdk.org/jeps/388) which > introduced the Windows/aarch64 port. That was really about combining an > existing OS and an existing CPU, but I'd say the amount of changes > required is similar to the BSD port, so I guess aiming at a JEP of > similar complexity level is fine. > > /Magnus > > > > > > Take care! > > Harald > -------------- next part -------------- An HTML attachment was scrubbed... URL: From magnus.ihse.bursie at oracle.com Fri Mar 7 14:30:36 2025 From: magnus.ihse.bursie at oracle.com (Magnus Ihse Bursie) Date: Fri, 7 Mar 2025 15:30:36 +0100 Subject: Improving OpenJDK for FreeBSD In-Reply-To: References: <8bddd709-5747-4bd3-b6d5-dcd4b2e33886@oracle.com> <8bd44fbe-6efd-4e43-9683-e6be2c9458af@oracle.com> <251c2111-2923-4df0-a371-a61336230060@oracle.com> Message-ID: On 2025-03-07 15:16, Thomas St?fe wrote: > @Magnus Ihse Bursie > > Why would we need a JEP? That is a lot of unnecessary red tape. The > BSD port already exists. It is technically not a new port. If we are to include it in the mainline, which is the goal here, a JEP is needed. We have only ever added or removed supported platforms in mainline with a JEP. /Magnus > > On Thu, Mar 6, 2025 at 8:19?PM Magnus Ihse Bursie > wrote: > > On 2025-03-06 19:38, Harald Eilertsen wrote: > > > On Thu, Mar 06, 2025 at 05:32:50PM +0100, Magnus Ihse Bursie wrote: > >> On 2025-02-21 04:03, David Holmes wrote: > >>> I would not like to see it happen that way. If it is to happen > then I > >>> would prefer to see a project established and a project repo. > >> [...snip...] > >> > >> I think that could work as well, but I guess the difficult part > will be to > >> get reviewers from the relevant areas. > > I'm happy to go whichever way the OpenJDK community thinks is > the best > > approach. On advantage I see from our perspective with the separate > > project repo is that it may be easier to do a bit of > experimentation and > > testing different approaches before a merge into the mainline. > > > > It will still be possible to do a phased approach to merging the > changes > > into the mainline, so that should give any reviewers that didn't > come > > around for the first merges into the project repo a second chance to > > voice their concerns. > > Formally, there is already a bsd-port project ("Port: BSD Project" is > the formal name), and according to the Census, Greg Lewis is the > Project > Lead. That means he has the authority to request that a repo be setup > for this project. I hope you have enough contact with him to ask > him to > send an email to ops at openjdk.org to request such a repo. Since the > cogs > of the OpenJDK administration moves slowly, I'd suggest trying get > him > to send such a request already; then the repo might be created in > time > for it to be actually needed. > > Formally, I believe it would be good if he also nominates you as > member > of the bsd-port project. In time, especially if your position at > FreeBSD > is turning out to be long-term, I think it would be good (and likely > supported by Lewis) to have you moving into the Project Lead > position of > the BSD port project. > > There is also an old mailing list associated with the project; we > might > consider moving this discussion over there. (Otoh, the porters-dev > list > is basically empty so I don't think we're overwhelming everyone by > keeping the discussion here as well.) > > > > > >> Also, I don't know if this has been said before, but this work > requires a > >> JEP. > > No, I haven't heard anything about that (I think). I'll read up > on it, > > and get back if I need a hand to hold on to :) > > Have a look at e.g. JEP 388 (https://openjdk.org/jeps/388) which > introduced the Windows/aarch64 port. That was really about > combining an > existing OS and an existing CPU, but I'd say the amount of changes > required is similar to the BSD port, so I guess aiming at a JEP of > similar complexity level is fine. > > /Magnus > > > > > > Take care! > > Harald > -------------- next part -------------- An HTML attachment was scrubbed... URL: