On 9/27/06, Gregory Shimansky <[EMAIL PROTECTED]> wrote:
On Tuesday 26 September 2006 17:25 Geir Magnusson Jr. wrote:
> Wanted to start a new thread with better subject line.
I want to take part in this thread too.
1. As Tim has mentioned in a separate thread the multiVM support requires that
we don't use global and static variables in native code. One global variable
used often in drlvm across all of the code is Global_Env::loader_env. I
didn't find it in the patch code so I want to ask you, did you move this
pointer into TLS?
Not exactly. You are right I tried to write the code which supports
multi VM. But that was not my main goal at this moment. So what I did
in that direction is
1) I added JNIEnv * into VM_thread structure. So you can always get a
Global_Env having JNIEnv. Probably it makes sense to use VM local
storage to keep Global_Env as well. I consider it as another
task....not InvocationAPI
2) Localized thread library instance for each VM.
3) Fixed those places where Global_Env can be extracted from JNI_Env
*. see jni_get_vm_env function.
2. Can we get avoid all of the CR-LF EOLs introduced in this patch?
I will do my best in the future. I promise :-)
Evgueni
--
Gregory Shimansky, Intel Middleware Products Division
---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
