From erik.osterlund at oracle.com Tue Aug 9 05:57:32 2022 From: erik.osterlund at oracle.com (Erik Osterlund) Date: Tue, 9 Aug 2022 05:57:32 +0000 Subject: About to integrate JDK-8290025 Message-ID: <35B5AC8E-1EC4-4E43-B30E-33F22EDFD176@oracle.com> Hi, This is just a heads-up that I am about to integrate JDK-8290025 soon ish. PR: https://github.com/openjdk/jdk/pull/9741/ It touches some platform specific code, so I would like to just check that nobody has objections, in case it does not work on some platform. Going forward, aging of ?old?/?cold? nmethods will only work if a platform has nmethod entry barriers. I do hope that can soon be a thing that we can rely on all platforms with JIT having. Thanks, /Erik -------------- next part -------------- An HTML attachment was scrubbed... URL: From thomas.stuefe at gmail.com Tue Aug 9 09:34:22 2022 From: thomas.stuefe at gmail.com (=?UTF-8?Q?Thomas_St=C3=BCfe?=) Date: Tue, 9 Aug 2022 11:34:22 +0200 Subject: About to integrate JDK-8290025 In-Reply-To: <35B5AC8E-1EC4-4E43-B30E-33F22EDFD176@oracle.com> References: <35B5AC8E-1EC4-4E43-B30E-33F22EDFD176@oracle.com> Message-ID: Hi Erik, give us a day or two more, please. I'll run the patch through our CI first to see if any of our platforms hiccup. Cheers, Thomas On Tue, Aug 9, 2022 at 11:25 AM Erik Osterlund wrote: > Hi, > > This is just a heads-up that I am about to integrate JDK-8290025 soon ish. > > PR: > https://github.com/openjdk/jdk/pull/9741/ > > It touches some platform specific code, so I would like to just check that > nobody has objections, > in case it does not work on some platform. > > Going forward, aging of ?old?/?cold? nmethods will only work if a platform > has nmethod entry barriers. > I do hope that can soon be a thing that we can rely on all platforms with > JIT having. > > Thanks, > /Erik > -------------- next part -------------- An HTML attachment was scrubbed... URL: From erik.osterlund at oracle.com Tue Aug 9 09:47:59 2022 From: erik.osterlund at oracle.com (Erik Osterlund) Date: Tue, 9 Aug 2022 09:47:59 +0000 Subject: [External] : Re: About to integrate JDK-8290025 In-Reply-To: References: <35B5AC8E-1EC4-4E43-B30E-33F22EDFD176@oracle.com> Message-ID: <985312DA-10D8-4EF7-BFE4-ABBBB73B16E1@oracle.com> Hi Thomas, Okay! Please let me know how it goes. Thanks, /Erik On 9 Aug 2022, at 11:34, Thomas St?fe > wrote: Hi Erik, give us a day or two more, please. I'll run the patch through our CI first to see if any of our platforms hiccup. Cheers, Thomas On Tue, Aug 9, 2022 at 11:25 AM Erik Osterlund > wrote: Hi, This is just a heads-up that I am about to integrate JDK-8290025 soon ish. PR: https://github.com/openjdk/jdk/pull/9741/ It touches some platform specific code, so I would like to just check that nobody has objections, in case it does not work on some platform. Going forward, aging of ?old?/?cold? nmethods will only work if a platform has nmethod entry barriers. I do hope that can soon be a thing that we can rely on all platforms with JIT having. Thanks, /Erik -------------- next part -------------- An HTML attachment was scrubbed... URL: From thomas.stuefe at gmail.com Thu Aug 11 06:24:21 2022 From: thomas.stuefe at gmail.com (=?UTF-8?Q?Thomas_St=C3=BCfe?=) Date: Thu, 11 Aug 2022 08:24:21 +0200 Subject: About to integrate JDK-8290025 In-Reply-To: References: <35B5AC8E-1EC4-4E43-B30E-33F22EDFD176@oracle.com> Message-ID: Hi Erik, as I wrote in the PR, we see regressions where we now get OOMs in tests that were deliberately started with small code cache sizes. Cheers, Thomas On Tue, Aug 9, 2022 at 11:34 AM Thomas St?fe wrote: > Hi Erik, > > give us a day or two more, please. I'll run the patch through our CI first > to see if any of our platforms hiccup. > > Cheers, Thomas > > On Tue, Aug 9, 2022 at 11:25 AM Erik Osterlund > wrote: > >> Hi, >> >> This is just a heads-up that I am about to integrate JDK-8290025 soon ish. >> >> PR: >> https://github.com/openjdk/jdk/pull/9741/ >> >> It touches some platform specific code, so I would like to just check >> that nobody has objections, >> in case it does not work on some platform. >> >> Going forward, aging of ?old?/?cold? nmethods will only work if a >> platform has nmethod entry barriers. >> I do hope that can soon be a thing that we can rely on all platforms with >> JIT having. >> >> Thanks, >> /Erik >> > -------------- next part -------------- An HTML attachment was scrubbed... URL: From ioi.lam at oracle.com Tue Aug 16 16:05:09 2022 From: ioi.lam at oracle.com (Ioi Lam) Date: Tue, 16 Aug 2022 09:05:09 -0700 Subject: JDK-8290981: Enable CDS for zero builds Message-ID: <7eb58409-c301-e0f6-ca06-540362f01b5e@oracle.com> I have prepared a PR for enabling CDS for zero (with help from Aleksey Shipil?v) ??? https://github.com/openjdk/jdk/pull/9869 If anyone is maintaining zero builds of the JDK, it would be great if you could try this patch and see if it introduce any regressions. Thanks - Ioi From shade at redhat.com Thu Aug 18 16:34:15 2022 From: shade at redhat.com (Aleksey Shipilev) Date: Thu, 18 Aug 2022 18:34:15 +0200 Subject: JDK-8290981: Enable CDS for zero builds In-Reply-To: <7eb58409-c301-e0f6-ca06-540362f01b5e@oracle.com> References: <7eb58409-c301-e0f6-ca06-540362f01b5e@oracle.com> Message-ID: On 8/16/22 18:05, Ioi Lam wrote: > I have prepared a PR for enabling CDS for zero (with help from Aleksey > Shipil?v) > > ??? https://github.com/openjdk/jdk/pull/9869 > > If anyone is maintaining zero builds of the JDK, it would be great if > you could try this patch and see if it introduce any regressions. Well, I am maintaining lots of Zero builds, and at least x86_64 Zero works. This should cover the majority of potential problems with Zero on other arches too. The rest of the problems, if any, would likely come from massive rebuilds of Java packages using the Zero VM in the system JDK, which would be hard to easily test. Adrian, tell us if you can/want to do that :) But I think it would be fine as it is. -- Thanks, -Aleksey From thomas.stuefe at gmail.com Fri Aug 19 05:21:55 2022 From: thomas.stuefe at gmail.com (=?UTF-8?Q?Thomas_St=C3=BCfe?=) Date: Fri, 19 Aug 2022 07:21:55 +0200 Subject: [External] : Re: About to integrate JDK-8290025 In-Reply-To: <985312DA-10D8-4EF7-BFE4-ABBBB73B16E1@oracle.com> References: <35B5AC8E-1EC4-4E43-B30E-33F22EDFD176@oracle.com> <985312DA-10D8-4EF7-BFE4-ABBBB73B16E1@oracle.com> Message-ID: Hi Erik, sorry, I could not find time this week. We still have crashes around methodEntryBarriers on ppc. Martin is back from vacation next week and will look at it. Can you hold off till then? Cheers, Thomas On Tue, Aug 9, 2022 at 11:48 AM Erik Osterlund wrote: > Hi Thomas, > > Okay! Please let me know how it goes. > > Thanks, > /Erik > > On 9 Aug 2022, at 11:34, Thomas St?fe wrote: > > Hi Erik, > > give us a day or two more, please. I'll run the patch through our CI first > to see if any of our platforms hiccup. > > Cheers, Thomas > > On Tue, Aug 9, 2022 at 11:25 AM Erik Osterlund > wrote: > >> Hi, >> >> This is just a heads-up that I am about to integrate JDK-8290025 soon ish. >> >> PR: >> https://github.com/openjdk/jdk/pull/9741/ >> >> >> It touches some platform specific code, so I would like to just check >> that nobody has objections, >> in case it does not work on some platform. >> >> Going forward, aging of ?old?/?cold? nmethods will only work if a >> platform has nmethod entry barriers. >> I do hope that can soon be a thing that we can rely on all platforms with >> JIT having. >> >> Thanks, >> /Erik >> > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From erik.osterlund at oracle.com Fri Aug 19 06:04:22 2022 From: erik.osterlund at oracle.com (Erik Osterlund) Date: Fri, 19 Aug 2022 06:04:22 +0000 Subject: [External] : Re: About to integrate JDK-8290025 In-Reply-To: References: <35B5AC8E-1EC4-4E43-B30E-33F22EDFD176@oracle.com> <985312DA-10D8-4EF7-BFE4-ABBBB73B16E1@oracle.com> Message-ID: <0DDD35CC-E8E0-49C7-8A5B-AC11EADA50EE@oracle.com> Hi Thomas, Yes, no worries. /Erik On 19 Aug 2022, at 07:22, Thomas St?fe wrote: ? Hi Erik, sorry, I could not find time this week. We still have crashes around methodEntryBarriers on ppc. Martin is back from vacation next week and will look at it. Can you hold off till then? Cheers, Thomas On Tue, Aug 9, 2022 at 11:48 AM Erik Osterlund > wrote: Hi Thomas, Okay! Please let me know how it goes. Thanks, /Erik On 9 Aug 2022, at 11:34, Thomas St?fe > wrote: Hi Erik, give us a day or two more, please. I'll run the patch through our CI first to see if any of our platforms hiccup. Cheers, Thomas On Tue, Aug 9, 2022 at 11:25 AM Erik Osterlund > wrote: Hi, This is just a heads-up that I am about to integrate JDK-8290025 soon ish. PR: https://github.com/openjdk/jdk/pull/9741/ It touches some platform specific code, so I would like to just check that nobody has objections, in case it does not work on some platform. Going forward, aging of ?old?/?cold? nmethods will only work if a platform has nmethod entry barriers. I do hope that can soon be a thing that we can rely on all platforms with JIT having. Thanks, /Erik -------------- next part -------------- An HTML attachment was scrubbed... URL: From ioi.lam at oracle.com Fri Aug 19 17:34:58 2022 From: ioi.lam at oracle.com (Ioi Lam) Date: Fri, 19 Aug 2022 10:34:58 -0700 Subject: [External] : Re: JDK-8290981: Enable CDS for zero builds In-Reply-To: References: <7eb58409-c301-e0f6-ca06-540362f01b5e@oracle.com> Message-ID: On 8/18/2022 9:34 AM, Aleksey Shipilev wrote: > On 8/16/22 18:05, Ioi Lam wrote: >> I have prepared a PR for enabling CDS for zero (with help from Aleksey >> Shipil?v) >> >> https://urldefense.com/v3/__https://github.com/openjdk/jdk/pull/9869__;!!ACWV5N9M2RV99hQ!L1VWYO1YSdu5JP8l9in8YrWDSe0jg5AqyT5h73A8stmwg_nEcC-81x9kQVua_xJhV-TC2ZOHLWk$ >> >> If anyone is maintaining zero builds of the JDK, it would be great if >> you could try this patch and see if it introduce any regressions. > > Well, I am maintaining lots of Zero builds, and at least x86_64 Zero > works. This should cover the majority of potential problems with Zero > on other arches too. The rest of the problems, if any, would likely > come from massive rebuilds of Java packages using the Zero VM in the > system JDK, which would be hard to easily test. > > Adrian, tell us if you can/want to do that :) But I think it would be > fine as it is. > I haven't heard back from Adrian but it looks like it's covered pretty well with Aleksey's tests. So I'll integrate soon. Thanks - Ioi