(0.41) AArch64 macOS: Stop assigning x18

[knn-k]

This commit stops the JIT assigning the x18 register on AArch64 macOS.

Original PR in master: #18351

[knn-k]

Related issues:

• Local build failing with XCode 14.0.3 and above #17493
• Recent builds of packaged Eclipse fail to launch with Semeru but succeed with Adoptium on macOS ARM #18336

[0xdaryl]

I approve these changes, but will defer to @pshipton if it can be contained in 0.41.

[pshipton]

We can evaluate next week after going through some testing in the head stream.

[pshipton]

Running some 0.41 testing, which is still going. So far there are a couple of DAA hangs in -Xjit:count=0 modes with user generated core files.

https://hyc-runtimes-jenkins.swg-devops.com/job/Test_openjdk11_j9_special.system_aarch64_mac_testList_1/288/

• DaaLoadTest_all_special_5m_5

https://hyc-runtimes-jenkins.swg-devops.com/job/Test_openjdk11_j9_special.system_aarch64_mac_testList_4/65/

• DaaLoadTest_daa1_special_5m_7

[pshipton]

Also the following, but there are no diagnostic files.

https://hyc-runtimes-jenkins.swg-devops.com/job/Test_openjdk17_j9_special.system_aarch64_mac_testList_3/193/

• DaaLoadTest_daa1_special_5m_24

[knn-k]

I ran Grinder jobs for DaaLoadTest_all_special_5m_5 and DaaLoadTest_daa1_special_5m_7, 20x each, and they finished successfully.
job/Grinder/35708/
job/Grinder/35709/

[knn-k]

Those failures with DaaLoadTest do not seem to relate to PR #18351.
See the following job on October 24, for example.
https://hyc-runtimes-jenkins.swg-devops.com/job/Test_openjdk11_j9_special.system_aarch64_mac/298/

[pshipton]

@JasonFengJ9 do you see any concerns in the weekend internal testing of amac? Did we get testing on both jdk11 and 17?

[JasonFengJ9]

do you see any concerns in the weekend internal testing of amac? Did we get testing on both jdk11 and 17?

From JDK 11/17/21 aarch64_mac weekly run containing #18351, there were a few java interrupted! (timed out?) and network-related infra issues, I don't see an issue that could have been caused by the PR in question.

[knn-k]

See Issue #17090 for the timeouts with LoadTests on amac.

[pshipton]

As the 0.41 and head stream testing looks good I will merge.