[jdk8u-dev] RFR: 8202952: C2: Unexpected dead nodes after matching

[Andrew John Hughes]

On Fri, 24 Mar 2023 07:57:45 GMT, Dongbo He <dongbohe at openjdk.org> wrote:

> I would like to backport this patch to fix crash during C2 compilation. 
> 
> This patch doesn't apply cleanly due to context differences, and
> I selected fuzzer tests from Fei Yang because the tests in the original commit cannot be reproduced on JDK8u.
> 
> Performed hotspot jtreg test on linux x86-64 platforms with -Xcomp options, new test fails without product change and passes with it.

Bot, please keep this one open.

The code was split into aarch64, ppc64 and x86 specific handling with [JDK-8154826](https://bugs.openjdk.org/browse/JDK-8154826). I don't think we should try backporting such a change into 8u, especially as [Volker already investigated this](https://mail.openjdk.org/pipermail/jdk8u-dev/2021-May/013899.html) for an AArch64 issue.

Could we not just change the line to use `X86_ONLY` instead of `LP64_ONLY`? The change would then be equivalent to what is in 11u.

Actually you would need to make it `X86_ONLY(LP64_ONLY(x))`, I believe, in order to exclude i686.

-------------

PR Review: https://git.openjdk.org/jdk8u-dev/pull/291#pullrequestreview-1572960634
PR Comment: https://git.openjdk.org/jdk8u-dev/pull/291#issuecomment-1674173723