harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mikhail Fursov" <mike.fur...@gmail.com>
Subject Re: [drlvm] The first GC helper with fast-path implemented in Java: gc_alloc
Date Fri, 13 Oct 2006 07:39:57 GMT
On 10/13/06, Ivan Volosyuk <ivan.volosyuk@gmail.com> wrote:

AFAIU, Mikhail also works on something like fast-JNI - a way to call
> native function from managed code. To do this, we need some kind of
> representation of native function in java, a mapping of native call
> into java bytecode. It looks the class
> org.apache.harmony.vmhelper.native.Utils was created for this purpose.
> The memset is just a function we can start with if we really want the
> ability to call native functions from java.
>
> I would like to have this functionality, but if the implementation
> needs some resources it can be postponed until after the basic helper
> inlining functionality is ready.
> --
> Ivan
>
>
Ivan, I'm absolutely agree with you that support of different types of
native calls from Java is the same task as calling slow native methods from
a fast-path helper.
This task worth a separate discussion because it has multiple notable
solutions.

As for today I prefer to finish JIT framework for helpers inlining first and
report that it's ready with a minimum of new features. (I'm going  prototype
native call as Weldon proposed, so we refactor this part once we have an
agreement on final design)

+ Is it ok if I remove cleaning code from the helper to avoid memset? I
could replace it with a loop, but imo it's too much for a helper's fast
path.. ?


-- 
Mikhail Fursov

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message