RFR: 8319897: Move StackWatermark handling out of LockStack::contains [v2]

[Stefan Karlsson]

On Mon, 13 Nov 2023 08:26:22 GMT, Stefan Karlsson <stefank at openjdk.org> wrote:

>> src/hotspot/share/runtime/lockStack.inline.hpp line 110:
>> 
>>> 108:   // Can't poke around in thread oops without having started stack watermark processing.
>>> 109:   StackWatermark* watermark = StackWatermarkSet::get(get_thread(), StackWatermarkKind::gc);
>>> 110:   assert(watermark == nullptr || watermark->processing_started(), "Processing must have started!");
>> 
>> Is line #109 only to support the assertion? If so it should be ifdef'd. Though for an assertion I would prefer to see some kind of `StackWatermark::is safe_for(Thread* t)` API.
>
> Yes, it is only for the assertions. I prefer to avoid adding ifdefs because I find them highly distracting. I was considering adding a function, but couldn't immediately come up with a great name. I'll send out a new proposal with a new function. Thanks.

We already had a `StackWatermarkSet::processing_started(JavaThread*)` function, so I hanged the code to use it.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/16609#discussion_r1390776299