[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openjdk-discuss
Subject:    Project proposal: Graal
From:       john.r.rose () oracle ! com (John Rose)
Date:       2012-02-22 6:09:43
Message-ID: 6885047A-5915-4A97-86B5-75CADEFF0913 () oracle ! com
[Download RAW message or body]

On Jan 30, 2012, at 4:32 AM, Douglas Simon wrote:

> > Copying may (or may not) result in a significant footprint increase
> > as well as require update synchronization.
> 
> In general, we will attempt to not duplicate VM data structures but
> access them directly from Java code. For example, Graal currently
> does not use the C++ CI interface code (used by c1 and c2).
> The current perm-gen removal project may have an adverse impact on
> this issue. However, we may also be able to directly access C++ data
> structures (we control the compiler ;-).

The copying in the Hotspot CI interface is not just buffering across handles.

There is a transactional aspect to the CI interface.  Some kinds of metadata are read \
under a lock.  A stable copy of the read is used optimistically, and rechecked at the \
end of the compile.  This may sometimes be faster and more correct than alternatives \
that keep only one copy of the metadata.

(FTR, the recheck is ciEnv::validate_compile_task_dependencies.)

? John


[prev in list] [next in list] [prev in thread] [next in thread]