possible bug with starvation

[James Hoffa]

Hello--
1. Thank you for all of your hard work on the FreeBSD
java project.  It  is really great work.
2.  I think I may have found a bug, but I am not sure.
 please see below.  It may just be a consequence of
the 4.8 threading model and is hopefully fixed in 5.X

Details:  
My codebase uses Apache Ant to build.  This works fine
on WinXP, Win2K, Solaris, and Linux.  Just check out
of cvs,  modify build.properties to reflect the
environment and go. One of the Ant tasks consists of 
signing a jar. 

I compiled the 1.4 jvm for FreeBSD and tried to build
my codebase.   When I run the equivalent of  make
world (dist-all) from a bash prompt,  it takes a long
time to build everything on this K62-350, but finishes
successfully.  

I set up a cron job to build the code a few times a
day.  Somehow, everything stops in the jarsigner task
and the cpu goes idle.   It is not a password/keystore
prompt issue -- the job finishes OK when run from the
command line. If I check back in a few hours, the same
cron job is still running, waiting for jarsigner to
finish.  As soon as I type a command as simple as "ls"
,  the cpu maxes out with jarsigner and the cron job
actually finishes.   It is as if I awoke the
sleeping/starving thread. 

below are some more details about my setup.  If you
need any additional information, I am happy to oblige.
 this bug is not critical to me, just trying to
contribute a bit to the project.  

 $ ant -version
Apache Ant version 1.5.2 compiled on February 28 2003

 $ java -version
java version "1.4.1-p3"
Java(TM) 2 Runtime Environment, Standard Edition
(build 1.4.1-p3-root_15_may_2003_02_10)

$ uname -a
FreeBSD althea.thinlizard.com 4.8-STABLE FreeBSD
4.8-STABLE #0: Sat May  3 17:39:26 PDT 2003    
root at althea.thinlizard.com:/usr/obj/usr/src/sys/ALTHEA
 i386