[11u] RFR: JDK-8206895: aarch64: rework error-prone cmp instuction

[Andrew Haley]

On 11/06/2020 20:15, Lemmond, Dan wrote:
>> BTW, we already discussed backporting it to 11 here and it was rejected
>> as too risky:
>
> I haven’t been able to find any information about the risk
> associated with 8206895 (this patch), but the 8206265 thread you
> refer to reads to me that a smaller, easily reviewable fix was
> approved for 11.0.1 due to stability concerns right after GA, while
> a somewhat riskier but better fix went into 12.

Sure, but 12 wasn't a LTS release in 2018, it wasn't even released
yet. Over time these things firm up.

Here's what I wrote for 8u:

  All fixes that significantly improve stability, security or
  performance and do not change behavioural compatibility will be
  considered for jdk8u. To use the language of the JDK 6 project, by
  default such fixes are assumed to be applicable to jdk8u, especially
  if having "soaked" in later JDK releases for a time without
  incident. By "significant" I mean any bug that affects runtime
  behaviour in a way that either produces incorrect results, poor
  performance, or crashes the VM, especially TCK failures.

> If there were/are other reasons, I'd appreciate any insight you can provide.

So, to be considered a patch has to significantly improve stability,
security or performance.

>> Please explain why this should be back-ported.
>
> It looks like a nice stability and readability improvement to
> me. Given that it's been baking since 12, I see the backport risk as
> minor.

How does it make the platform more stable? I guess it might help when
back-porting new intrinsics, but I hope that's not going to be a huge
thing now.

-- 
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