RFR: 8360707: Globally enumerate all blobs, stubs and entries

[Martin Doerr]

On Tue, 1 Jul 2025 08:34:32 GMT, Amit Kumar <amitkumar at openjdk.org> wrote:

>> @offamitkumar Thanks for following up.
>> 
>>> is this something which is expected:
>> 
>>     Stack: [0x000003ff82980000,0x000003ff82a80000],  sp=0x000003ff82a7e4c8,  free space=1017k
>>     Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
>>     J 21 c2 MachCodeFramesInErrorFile$Crasher.crashInJava3(J)V (27 bytes) @ 0x000003ff70657c8a [0x000003ff70657c40+0x000000000000004a]
>>     J 20 c2 MachCodeFramesInErrorFile$Crasher.crashInJava2(J)V (13 bytes) @ 0x000003ff706580c0 [0x000003ff70658040+0x0000000000000080]
>>     j  MachCodeFramesInErrorFile$Crasher.crashInJava1(J)V+9
>>     j  MachCodeFramesInErrorFile$Crasher.main([Ljava/lang/String;)V+71
>>     v  ~StubRoutines::Stub Generator call_stub_stub 0x000003ff70580edc
>>     
>>     [error occurred during error reporting (printing native stack (with source info)), id 0xe0000000, Internal Error (/home/amit/jdk/src/hotspot/cpu/s390/frame_s390.inline.hpp:40)]
>>     
>>     Retrying call stack printing without source information...
>> 
>> That does not look right. Was there any more hserr printout after that last line? -- in particular any [MachCode]...[/MachCode] sections? It would help if you could attach the full .jtr file.
>> 
>> The stack backtrace in the listing above is broken at the frame for a generated stub (call_stub_stub) so it looks like the stack walk is failing. It ought to continue with a C++ frame for JavaCalls::call_helper and so on right up to the frame for start_thread. I'll check to see if perhaps this might be to do with some change to the stub.
>
>> That does not look right. Was there any more hserr printout after that last line? -- in particular any [MachCode]...[/MachCode] sections? It would help if you could attach the full .jtr file.
> 
> There isn't much after this @adinn. It's the hs_err file generated during the test case execution. If you still want to take a look at jtr then I can attach it.

@offamitkumar: Is the crash you reported reproducible? Could it be that `save_return_pc()` is missing in `address generate_call_stub(address& return_address)`? PPC64 has it.

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

PR Comment: https://git.openjdk.org/jdk/pull/26004#issuecomment-3024189963