Re: pthread_mutex and resuming/stopping_java

Hi Martin,

> Hi all, we're new to sablevm and try to contribute to nizar's work.
> We've lot of questions!

Just ask...  We'll do what we can to answer.

> Why do we need these two calls in JVMDI (libsablevm/debug_interface.c)?
>    _svmm_resuming_java (env);
>    ...
>    _svmm_stopping_java (env);
> or
>    _svmm_stopping_java (env);
>    ...
>    _svmm_resuming_java (env);

This is due to garbage collection.  SableVM's collector moves objects,
so one must be very careful not to access an object while it is being moved.

"stopping_java" means: We're going to run native code (very soon).  We
won't be using any direct pointer to an object anymore.  GC is allowed
to proceed if needed (on another thread).

"resuming java" means: We're back from native code.  We need to access
objects, so, we now disable GC (other than at "checkpoints").

> Why is sometimes the following used
>    pthread_mutex_lock (&vm->global_mutex);
>    ...
>    pthread_mutex_unlock (&vm->global_mutex);
> 
> instead of this?
>    _svmm_mutex_lock (jdwp->global_mutex);
>    ...
>    _svmm_mutex_unlock ();

The second form should always be used.  The only exception is when it's
impossible to match lock/unlock.  [There's an example of this in
gc_copying.c].

Etienne

-- 
Etienne M. Gagnon, Ph.D.            http://www.info2.uqam.ca/~egagnon/
SableVM:                                       http://www.sablevm.org/
SableCC:                                       http://www.sablecc.org/

_______________________________________________
SableVM-devel mailing list
[email protected]
http://sablevm.org/lists/control/listinfo/sablevm-devel

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDS1LtjyrJi4rH84gRAmHsAJ41xDAaYkyiKeN62CwlswsUhcKDrwCfbHKk
emeC4965FsVYgSt2oED6ids=
=zU7H
-----END PGP SIGNATURE-----