From robbin.ehn at oracle.com Wed Aug 21 08:19:47 2019 From: robbin.ehn at oracle.com (Robbin Ehn) Date: Wed, 21 Aug 2019 10:19:47 +0200 Subject: RFR(xs): 8229967: Enable thread local handshakes on zero Message-ID: <5f5fa00b-aa51-7b57-c8e3-aa3664951176@oracle.com> Hi all, please consider. Zero can have handshakes enabled. This prepares the vm for removing global safepoints. Built x64 and sanity tested. (should be platform independent) x86 do not build for me due to workaround_expand_exec_shield_cs_limit. Thanks, Robbin diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globalDefinitions_zero.hpp --- a/src/hotspot/cpu/zero/globalDefinitions_zero.hpp Mon Aug 19 13:43:51 2019 +0200 +++ b/src/hotspot/cpu/zero/globalDefinitions_zero.hpp Wed Aug 21 10:02:44 2019 +0200 @@ -32,2 +32,4 @@ +#define THREAD_LOCAL_POLL + #include diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globals_zero.hpp --- a/src/hotspot/cpu/zero/globals_zero.hpp Mon Aug 19 13:43:51 2019 +0200 +++ b/src/hotspot/cpu/zero/globals_zero.hpp Wed Aug 21 10:02:44 2019 +0200 @@ -78,3 +78,3 @@ -define_pd_global(bool, ThreadLocalHandshakes, false); +define_pd_global(bool, ThreadLocalHandshakes, true); From sgehwolf at redhat.com Wed Aug 21 11:32:08 2019 From: sgehwolf at redhat.com (Severin Gehwolf) Date: Wed, 21 Aug 2019 13:32:08 +0200 Subject: RFR(xs): 8229967: Enable thread local handshakes on zero In-Reply-To: <5f5fa00b-aa51-7b57-c8e3-aa3664951176@oracle.com> References: <5f5fa00b-aa51-7b57-c8e3-aa3664951176@oracle.com> Message-ID: Hi Robin, On Wed, 2019-08-21 at 10:19 +0200, Robbin Ehn wrote: > Hi all, please consider. > > Zero can have handshakes enabled. > This prepares the vm for removing global safepoints. > > Built x64 and sanity tested. (should be platform independent) Thanks for doing this. Zero bootcycles fine with this for me on x86_64. So +1 from me. > x86 do not build for me due to workaround_expand_exec_shield_cs_limit. I've never tried 32 bit Zero on x86. Not sure if this was ever working. Thanks, Severin > Thanks, Robbin > > diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globalDefinitions_zero.hpp > --- a/src/hotspot/cpu/zero/globalDefinitions_zero.hpp Mon Aug 19 13:43:51 2019 +0200 > +++ b/src/hotspot/cpu/zero/globalDefinitions_zero.hpp Wed Aug 21 10:02:44 2019 +0200 > @@ -32,2 +32,4 @@ > > +#define THREAD_LOCAL_POLL > + > #include > diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globals_zero.hpp > --- a/src/hotspot/cpu/zero/globals_zero.hpp Mon Aug 19 13:43:51 2019 +0200 > +++ b/src/hotspot/cpu/zero/globals_zero.hpp Wed Aug 21 10:02:44 2019 +0200 > @@ -78,3 +78,3 @@ > > -define_pd_global(bool, ThreadLocalHandshakes, false); > +define_pd_global(bool, ThreadLocalHandshakes, true); From robbin.ehn at oracle.com Wed Aug 21 20:05:17 2019 From: robbin.ehn at oracle.com (Robbin Ehn) Date: Wed, 21 Aug 2019 22:05:17 +0200 Subject: RFR(xs): 8229967: Enable thread local handshakes on zero In-Reply-To: References: <5f5fa00b-aa51-7b57-c8e3-aa3664951176@oracle.com> Message-ID: <86154f6d-7941-1d3b-ab27-d0703c7f1836@oracle.com> Thanks! /Robbin On 2019-08-21 13:32, Severin Gehwolf wrote: > Hi Robin, > > On Wed, 2019-08-21 at 10:19 +0200, Robbin Ehn wrote: >> Hi all, please consider. >> >> Zero can have handshakes enabled. >> This prepares the vm for removing global safepoints. >> >> Built x64 and sanity tested. (should be platform independent) > > Thanks for doing this. Zero bootcycles fine with this for me on x86_64. > So +1 from me. > >> x86 do not build for me due to workaround_expand_exec_shield_cs_limit. > > I've never tried 32 bit Zero on x86. Not sure if this was ever working. > > Thanks, > Severin > >> Thanks, Robbin >> >> diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globalDefinitions_zero.hpp >> --- a/src/hotspot/cpu/zero/globalDefinitions_zero.hpp Mon Aug 19 13:43:51 2019 +0200 >> +++ b/src/hotspot/cpu/zero/globalDefinitions_zero.hpp Wed Aug 21 10:02:44 2019 +0200 >> @@ -32,2 +32,4 @@ >> >> +#define THREAD_LOCAL_POLL >> + >> #include >> diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globals_zero.hpp >> --- a/src/hotspot/cpu/zero/globals_zero.hpp Mon Aug 19 13:43:51 2019 +0200 >> +++ b/src/hotspot/cpu/zero/globals_zero.hpp Wed Aug 21 10:02:44 2019 +0200 >> @@ -78,3 +78,3 @@ >> >> -define_pd_global(bool, ThreadLocalHandshakes, false); >> +define_pd_global(bool, ThreadLocalHandshakes, true); > From robbin.ehn at oracle.com Wed Aug 21 20:05:52 2019 From: robbin.ehn at oracle.com (Robbin Ehn) Date: Wed, 21 Aug 2019 22:05:52 +0200 Subject: RFR(xs): 8229967: Enable thread local handshakes on zero In-Reply-To: <8ab8a5bd-d26b-80b2-ffb1-1d524ae22dcc@redhat.com> References: <5f5fa00b-aa51-7b57-c8e3-aa3664951176@oracle.com> <8ab8a5bd-d26b-80b2-ffb1-1d524ae22dcc@redhat.com> Message-ID: <56cffa29-d4e5-d02e-c7db-d5bcd5bb403a@oracle.com> Thanks! /Robbin On 2019-08-21 15:54, Aleksey Shipilev wrote: > On 8/21/19 10:19 AM, Robbin Ehn wrote: >> Hi all, please consider. >> >> Zero can have handshakes enabled. >> This prepares the vm for removing global safepoints. >> >> Built x64 and sanity tested. (should be platform independent) >> x86 do not build for me due to workaround_expand_exec_shield_cs_limit. >> >> Thanks, Robbin >> >> diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globalDefinitions_zero.hpp >> --- a/src/hotspot/cpu/zero/globalDefinitions_zero.hpp??? Mon Aug 19 13:43:51 2019 +0200 >> +++ b/src/hotspot/cpu/zero/globalDefinitions_zero.hpp??? Wed Aug 21 10:02:44 2019 +0200 >> @@ -32,2 +32,4 @@ >> >> +#define THREAD_LOCAL_POLL >> + >> ?#include >> diff -r 7c20dcb59a68 src/hotspot/cpu/zero/globals_zero.hpp >> --- a/src/hotspot/cpu/zero/globals_zero.hpp??? Mon Aug 19 13:43:51 2019 +0200 >> +++ b/src/hotspot/cpu/zero/globals_zero.hpp??? Wed Aug 21 10:02:44 2019 +0200 >> @@ -78,3 +78,3 @@ >> >> -define_pd_global(bool, ThreadLocalHandshakes, false); >> +define_pd_global(bool, ThreadLocalHandshakes, true); > > FWIW, linux-x86_64-zero-fastdebug bootcycles for me. Looks good! >