[master] RFR: ObjectMonitor Storage

[Roman Kennke]

On Tue, 29 Mar 2022 06:12:49 GMT, David Holmes <dholmes at openjdk.org> wrote:

> Effectively reviving the old TSM-style allocation strategy does make me cringe somewhat. :( And without seeing how exactly this is to be used (including if the index goes out-of-range) makes it hard to suggest alternatives - though index into a (per-thread?) table of locked monitors seems one.

The motivation for this is to be able to compress the OM pointer to 32bit (or less), so that it fits into the lower part of the object header (or, later, the 32-bit-header). In order for this to work, we need a known address range for OMs into which we can allocate (aligned) OMs.

Roman

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

PR: https://git.openjdk.java.net/lilliput/pull/39