RFR: 8319778: Remove unreachable code in ObjectSynchronizer::exit

Mon Nov 13 07:24:57 UTC 2023

On Mon, 13 Nov 2023 00:59:00 GMT, David Holmes <dholmes at openjdk.org> wrote:

> Sorry I don't follow the rationale here. If the object is fast-locked to T1 and T2 tries to lock, it will inflate and have to install the anonymous-owner. WhenT1 exits it will see this state.

This is true. But the current post condition enforced by `ObjectSynchronizer::inflate` is that if the `current` JavaThread owns the monitor then it also fixes the the owner (, and in the recursive lightweight implementation, the recursions) on the `ObjectMonitor`. So in `ObjectSynchronizer::exit` which may only ever be called on an object owned by `current` this must be the case. 

If it would call exit on an object not owned by the current thread it may or may not say anonymous. If it were to say anonymous then the `current` JavaThread's lock stack would not contain the object, and it would not fix the owner. It would instead `assert(false, "Non-balanced monitor enter/exit!")` inside `ObjectMonitor::exit`.

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

PR Comment: https://git.openjdk.org/jdk/pull/16602#issuecomment-1807588853