RFR: 8341862: PPC64: C1 unwind_handler fails to unlock synchronized methods with LM_MONITOR

[Richard Reingruber]

On Mon, 14 Oct 2024 13:51:21 GMT, Richard Reingruber <rrich at openjdk.org> wrote:

> Make sure `LIR_Assembler::emit_unwind_handler()` jumps to the slow path directly for unlocking a synchronized method if `LM_MONITOR` is used.
> On the fast paths assertions are added that the mode is actually handled.
> 
> The change passed our CI testing:
> Tier 1-4 of hotspot and jdk. All of Langtools and jaxp. Renaissance Suite and SAP specific tests.
> Testing was done on the main platforms and also on Linux/PPC64le and AIX.

Thanks for the review!

> Do you think src/hotspot/cpu/s390/c1_LIRAssembler_s390.cpp
> 
> `LIR_Assembler::emit_unwind_handler() `
> 
> should be adjusted in a similar way (separate issue/change however).

You are right indeed. @offamitkumar as Matthias noticed it looks like s390 is affected by the issue too. You might want to check.

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

PR Comment: https://git.openjdk.org/jdk/pull/21497#issuecomment-2416971346