[11u] Respin of JDK 11.0.9 with JDK-8250861 and JDK-8255603 -> 11.0.9.1

[Severin Gehwolf]

Hi,

On Fri, 2020-10-30 at 10:23 +0000, Langer, Christoph wrote:
> Hi,
>  
> I hereby propose a re-spin of OpenJDK 11 update release 11.0.9 with the following two fixes to be added:
>  
> JDK-8250861: Crash in MinINode::Ideal(PhaseGVN*, bool) [0]
> JDK-8255603: Memory/Performance regression after JDK-8210985 [1]
>  
> The first one is a potential crash which comes as a regression to one of the October CPU patches. The second one fixes a memory regression that was introduced with 11.0.5 already but has just been analyzed and fixed within the last few days.
>  
> The version number of the update would be 11.0.9.1, as Gil correctly pointed out [2]
>  
> I will wait for feedback/objections until Tuesday 03/11/20 EOB. In case nobody vetoes, Goetz and me will go ahead on Wednesday 04/11/20 and push these patches into the jdk11u repository, along with the required version bumps and tags.

Here are my thoughts:

JDK-8250861 is a regression introduced in 11.0.9 and a serious crasher
bug in c2. If consensus is to fix it with an async update to 11.0.9,
then we should do the same for 8u272 (and possibly JDK 13). It looks
like for JDK 15 it'll end up in 15.0.2. Just some more data points :)

As to JDK-8255603, my preference would be to defer to 11.0.10. While
the fix seems benign, it appears nobody noticed this until very
recently. 11.0.5 got out in October 2019. There is a known work-around. 
At least it seems strange to rush this fix now, where it's been out in
the wild for about a year.

Thanks,
Severin

>  
> [0] https://bugs.openjdk.java.net/browse/JDK-8250861
> [1] https://bugs.openjdk.java.net/browse/JDK-8255603
> [2] https://mail.openjdk.java.net/pipermail/jdk-updates-dev/2020-October/004055.html
>