Re: [Sablevm-developer] Fixed(?) problem with exec()

[Etienne G. <gag...@uq...>]

Hi Greg,

You and David now have write access to create 1.1.x tags:

/sablevm/tags/sablevm-1\.1\.[^/]+/ = @sablevm-staging(add,remove,properties)
(same for sablevm-test-suite, sablevm-classpath, and sablejit).

So, feel free to make development releases when you wish.

Etienne

Grzegorz B. Prokopski wrote:
> Hi all,
> 
> I need a confirmation, as I haven't touched this stuff yet and
> I might be just missing something.
> 
> exec() is broken in current staging and 1.1.0. If you tried to run
> such a simple program:
> 
> public class ExecTest {
>   public static void main(String []args) {
>     try {
>       Process child = Runtime.getRuntime().exec("/bin/sleep 1");
>       child.waitFor();
>     }
>     catch(Exception e) {
>       System.out.println("Execution exception catched");
>     }
>     System.out.println("Execution finished");
>   }
> }
> 
> $ sablevm -Y ExecTest
> Java_java_lang_ProcessImpl_nativeWaitFor: No child processes
> Child did not exit normally.
> 
> strace showed that waitpid() is called twice, second time after
> the child exited. So I scratched my head and after 3 minutes in
> src/libsablevm/java_lang_ProcessImpl.c in ..._nativeWaitFor
> i saw more less such thing:
> 
> waitpid(pid, &status, 0);
> 
> if (nonblocking) {
>   waitpid(pid, &status, WNOHANG);
> } else {
>   waitpid(pid, &status, 0);
> }
> 
> Existence of the first waitpid makes no sense to me. I commented
> it out in just-uploaded sablevm Debian packages as it seemed to
> solve the problem.
> 
> Am I missing anything or should it get to staging too?
> 
> It's quite "grave" thing, I'd say. If it wasn't me who'd have to
> make the next release, I'd push for 1.1.1 ASAP :-/
> 
> Cheers,
> 
> 				Grzegorz B. Prokopski
> 


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