Shenandoah and aarch64

[Andrew Haley]

On 7/15/19 12:19 PM, Roman Kennke wrote:
>>> For the Shenandoah part, would shenandoah/jdk11 work as staging repo? It
>>> seems to be the obvious choice because we already have it, and it is
>>> supposed to only contain the actual difference between upstream jdk11u
>>> and Shenandoah.
>>
>> Perfect. That could not be any better.
> 
> Great. Let's work with that then. We want to backport the LRB and
> related stuff first, and comb through and sort out any possibly unneeded
> changes.

Yeah. This may disappoint a few people, but we should perhaps let
Shenandoah bake in JDK 11 for a while before pushing it to 8. We'll
need to go through the code (again) with a fine-toothed comb to make
sure that the changes, particularly to C2, do not leak out if
Shenandoah is disabled.

> Well, pretty much same as jdk11. We want to first backport LRB and
> friends, which should clean up the shared code changes very
> considerably. And after that, we shall comb through:
> 
> https://builds.shipilev.net/patch-openjdk-shenandoah-jdk8-only-shared/hotspot/
> 
> and sort out whatever is no longer needed.
> 
> Right now it's mixed up with aarch64 changes, which is why I'd probably
> push aarch64 to jdk8u first, which should be whatever is:
> 
> http://hg.openjdk.java.net/aarch64-port/jdk8u/hotspot/
> 
> When this is done, we can take the complete patch and push it at once to
> jdk8u.
> 
> Do you agree?

That sounds sensible. AArch64 is much less intrusive to the 8u code
base than Shenandoah.

-- 
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671