Visit certain unvisited ROMClass header slots

[cjjdespres]

The allSlotsInROMClassDo() function now visits every slot in a J9ROMClass header.

[cjjdespres]

Attn @mpirvu. While I was implementing the ROM class hashing changes, I found that the J9SRP callSiteData field of J9ROMClass didn't seem to be walked by allSlotsInROMClassDo. There's an allSlotsInCallSiteDataDo:

openj9/runtime/util/romclasswalk.c

Line 1419 in 051b08e

static void allSlotsInCallSiteDataDo (J9ROMClass* romClass, J9ROMClassWalkCallbacks* callbacks, void* userData)

but it doesn't appear to walk the actual SRP to the call site data itself. Since I was already fixing the one field, I added in slot callbacks for the other fields of J9ROMClass that didn't seem to have them. Link to J9ROMClass for comparison:

openj9/runtime/oti/j9nonbuilder.h

Line 3501 in 9ab4edb

typedef struct J9ROMClass {

It didn't seem likely that they were omitted for any particular reason, but it's possible they were. I'm unsure if other walking methods need to be similarly fixed.

[cjjdespres]

Ideally there would be a test that all of the slots are visited (or at least all of the SRPs). I could see a modified Cursor being created for the ROM class writer that would record the address of every SRP written to a ROM class. After writing, you could then run allSlotsInROMClassDo to collect the names and addresses of every SRP it touches, and then check that these two lists are equal.

[mpirvu]

LGTM

[mpirvu]

@TobiAjila Could you please suggest a reviewer for this VM related item? Thanks

[babsingh]

jenkins test sanity.functional,sanity.openjdk zlinux jdk21

[babsingh]

@cjjdespres Are you tracking any failures which will be resolved by this PR?

[cjjdespres]

@cjjdespres Are you tracking any failures which will be resolved by this PR?

Sort of. I came across it while testing some code I'll be adding to my PR #18301, which will involve visiting all the SRPs in a ROM class. This PR is a prerequisite for that one, in that sense. I'm not aware of any existing failures caused by not visiting these slots.

[cjjdespres]

The test failure in jdk_lang_1, java/lang/Thread/GenerifyStackTraces.java, on s390x looks like #18521:

22:53:40  STDERR:
22:53:40  java.lang.RuntimeException: Expected B in frame 3 but got waitForDump
22:53:40  	at GenerifyStackTraces.checkStack(GenerifyStackTraces.java:173)
22:53:40  	at GenerifyStackTraces.dumpStacks(GenerifyStackTraces.java:146)
22:53:40  	at GenerifyStackTraces$DumpThread.run(GenerifyStackTraces.java:75)
22:53:40  java.lang.RuntimeException: Test Failed.
22:53:40  	at GenerifyStackTraces.main(GenerifyStackTraces.java:60)
22:53:40  	at java.base/jdk.internal.reflect.DirectMethodHandleAccessor.invoke(DirectMethodHandleAccessor.java:103)
22:53:40  	at java.base/java.lang.reflect.Method.invoke(Method.java:586)
22:53:40  	at com.sun.javatest.regtest.agent.MainActionHelper$AgentVMRunnable.run(MainActionHelper.java:333)
22:53:40  	at java.base/java.lang.Thread.run(Thread.java:1595)

[mpirvu]

@babsingh The test was successful (excepting the known issue). If you agree, I could merge this item.

[babsingh]

• Only a known and unrelated failure is seen: #18521