RFR: [8u] 2018-02-19: Bulk backports to sh/jdk8u
Roman Kennke
rkennke at redhat.com
Wed Feb 21 17:49:30 UTC 2018
Am 19.02.2018 um 19:04 schrieb Aleksey Shipilev:
> http://cr.openjdk.java.net/~shade/shenandoah/backports/jdk8u-20180219/webrev.01/
>
> This backports the bulk of improvements to sh/jdk8u:
>
> b5a09ff17768: [backport] LotsOfCycles test timeouts
> 8735773ec619: [backport] Single thread-local GC state flag for all barriers
> 544322604347: [backport] ShConcurrentThread races with set_gc_state_bit
> d55c6d5216d1: [backport] GC state testers (infra)
> ab9dd1d11dc1: [backport] Protect C2 matchers with UseShenandoahGC
> fd9724b26fdd: [backport] Refactor allocation failure and explicit GC handling
> e5398dce6e7b: [backport] Make concurrent precleaning log message optional again
> 26b9048c042a: [backport] Make degenerated update-refs use region-set cursor to hand over work
> 12654193e434: [backport] Demote warning message about OOM-during-evac to informational
> 67294a38c0c7: [backport] TestSelectiveBarrierFlags should accept multi-element flag selections
> ecb87af5e0d8: [backport] Implement flag to generate write-barriers without membars
> b8c39bdc0dac: [backport] Log message on ref processing, class unload, update refs for mark events
> bd01b07ba0d7: [backport] Add ShenandoahRootProcessor API to report threads while scanning roots
> 3a6457fecc72: [backport] Relax assert in SBS::is_safe()
> f8aa59bf8732: [backport] Conditionalize PerfDataMemorySize on enabled heap sampling
> 240af491d9b5: [backport] Make major GC phases exclusive from each other
> 869aee37dbf1: [backport] Single GCTimer shared by all operations
> dbdde8719baf: [backport] Don't treat allocation regions implicitely live during some GCs
> 9ea5ffefa388: [backport] Fix Mac OS build warnings
> 7c494d9db524: [backport] Record cycle start/end to avoid continuous periodic GC
> 437741c2bffd: [backport] Improve/more detailed timing stats for root queue work
> f753dccda9dd: [backport] Fix/improve CLD processing
> da727e3d353f: [backport] Use leftmost region in GC allocations
> eef04e0a0474: [backport] Cleanup and strengthen BrooksPointer verification
> 35b65240dc8c: [backport] Remove BS:is_safe in favor of logged BS::verify_safe_oop
> 2a9afc3843f8: [backport] Rich assertion failure logging
> 2a3a44377445: [backport] Rename *_oop_static/oop_ref to *_forwarded
> 9fd8fc2e8b47: [backport] Get easy on template instantiations in ShConcMark
> eada99b258fb: [backport] Improve assertion/verification messages a bit
> 00a3f7ca1c36: [backport] Demote ShenandoahAllocImplicitLive to diagnostic
>
> Apart from very recent work in sh/jdk10, the only significant thing missing is Degenerated GC, which
> gives me trouble in testing (I suspect another 8u-specific bug). I shall try and debug it for the
> next bulk backport.
>
> Testing: hotspot_gc_shenandoah {fastdebug|release}
>
> Thanks,
> -Aleksey
>
Looks good.
Roman
More information about the shenandoah-dev
mailing list