Fix crash under -Xaot:

[dsouzai]

When running with -Xjit: (i.e., not -Xjit but -Xjit: which has the colon followed by an empty string), the JVM prints out an error message and terminates. However, when running with -Xaot: the error message is printed out but then the JVM crashes. This happens because there is a call to codert_onload between initializing the compiler args for -Xjit and -Xaot.

This PR fixes this by invoking codert_onload after initializing both -Xjit and -Xaot. This is ok because the initization does not actually process the options. Rather, it sets the value of a pointer to point to the compiler args. Thus, calling codert_onload after initializing both is safe because the initialization does not depend on any initialization that codert_onload performs.

Fixes #20601

[dsouzai]

@hzongaro could you review/merge?

[hzongaro]

Looks good. Thanks!

[hzongaro]

I'm not sure what will happen if I run sanity.functional once with and once without aot. Giving it whirl. . . .

Jenkins test sanity.functional,sanity.functional+aot all jdk8,jdk17,jdk21

[hzongaro]

It looks like some testing was aborted. Restarting those runs:

Jenkins test sanity.functional+aot zlinux,win,win32 jdk8

[hzongaro]

Jenkins test sanity.functional+aot aix,plinux jdk17,jdk21

[hzongaro]

Jenkins test sanity.functional+aot zlinux jdk21

[dsouzai]

alinux and xmac failures are crashes in jfr tests, but they are run with -Xint, so the forceAOT option set in the TR_OptionsAOT env var does not take into effect.

Two of the aix failures are due to #17396

One aix failure is due to #21601

The last aix failure is

[2025-07-06T22:09:53.588Z] +++ j9vm.test.threadinterrupt.scenario7Test: +++
[2025-07-06T22:09:53.588Z] command: /home/jenkins/workspace/Test_openjdk21_j9_sanity.functional_ppc64_aix_Personal_testList_0/jdkbinary/j2sdk-image/bin/java -Xshareclasses:name=test_aot -Xscmx400M -Xscmaxaot256m  -Xjit -Xgcpolicy:gencon -Xnocompressedrefs  -Xdump -Xms64m -Xmx64m -classpath /home/jenkins/workspace/Test_openjdk21_j9_sanity.functional_ppc64_aix_Personal_testList_0/aqa-tests/TKG/../../jvmtest/functional/VM_Test/VM_Test.jar:/home/jenkins/workspace/Test_openjdk21_j9_sanity.functional_ppc64_aix_Personal_testList_0/../../testDependency/lib/asm-all.jar j9vm.test.threadinterrupt.scenario7Test 
[2025-07-06T22:09:53.588Z] 
[2025-07-06T22:09:53.846Z] main  waiting
[2025-07-06T22:09:58.021Z] inter syncing on a
[2025-07-06T22:09:58.021Z] inter synced on a
[2025-07-06T22:09:58.021Z] inter notifying main
[2025-07-06T22:09:58.021Z] inter done notifying a
[2025-07-06T22:09:58.021Z] inter interrupting main
[2025-07-06T22:09:58.021Z] inter done interrupting main
[2025-07-06T22:09:58.021Z] noti syncing on a
[2025-07-06T22:09:58.021Z] noti synced on a
[2025-07-06T22:09:58.021Z] noti waiting on a
[2025-07-06T22:09:58.585Z] main  done waiting
[2025-07-06T22:09:58.585Z] SUCCESS
[2025-07-06T22:09:59.150Z] Thread[#32,Thread-0,5,main] (interrupter) is done
[2025-07-07T06:59:27.547Z] Cancelling nested steps due to timeout
[2025-07-07T06:59:27.575Z] Sending interrupt signal to process
[2025-07-07T06:59:28.014Z] non-zero exit value: 143
[2025-07-07T06:59:28.014Z] *** Test FAILED *** (j9vm.test.threadinterrupt.scenario7Test)

I can't find an existing issue, but given that there is no -Xaot flag set on this test, I don't think this change is responsible for this.

[hzongaro]

Thanks for investigating the failures, @dsouzai! As the failures all appear to be unrelated to this change, and the change itself appears to be relatively safe, I will go ahead and merge.