EPICS Controls Argonne National Laboratory

Experimental Physics and
Industrial Control System

1994  1995  1996  1997  <19981999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024  Index 1994  1995  1996  1997  <19981999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
<== Date ==> <== Thread ==>

Subject: Channel Access with Java on Solaris
From: Ken Evans <[email protected]>
To: [email protected]
Cc: [email protected]
Date: Tue, 24 Mar 1998 11:19:21 -0600
     This note describes a bug in Java 1.1 that causes channel access
to not work properly with Java on Solaris.

     Channel access (CA) can be invoked as a native method in Java.
While this makes the code platform dependent and not "100% Java", it
is the only way to use EPICS with Java until someone writes a Java
version of CA.

     However, there is a bug, at least through JDK 1.1.5, that
prevents CA from working when using the default (green_threads)
implementation of Java on Solaris.

     The problem is caused by an improper implementation of the recv()
call.  This is the function that reads a socket.  By default, if there
is nothing to read on the socket, recv() waits until there is,
blocking the program.  TCP sockets in CA are set to be non-blocking,
in which case recv() returns with a status indicating that it "would
block".  CA typically reads the socket until it gets the "would block"
status and then goes on about its business.  The problem is that in
Java recv() blocks all the time.  This means that CA hangs as soon as
it has finished reading the first message.  It takes a certain amount
of programming time to determine that this is what is happening, as
well.  (Java is hard to debug with native methods since the program
that is running is java, not your own application.  Java, of course,
is not built with debugging information.)  Hopefully, this note may
help others avoid spending this time.

     The bug has been reported to Java but they do not respond to bug
reports, at least from ordinary people.  They just fix them as they
see fit.  Since they market the "100% Java" concept, one can assume
this particular, operating-system-dependent bug will not be a high
priority.

     The bug does not appear if one uses native_threads instead of
green_threads.  Neither is well documented, the documentation
apparently appearing in the .javawrapper shell script in the java
installation directory.

     My experience is that you can set the environment variable
THREADS_FLAG to native to use native threads.  CA needs to be built to
be thread safe.  Building shared libraries with the following options
seems to work:

-Kpic -G -D_REENTRANT.

However, building shared libraries is an exercise in itself.

	-Ken

Navigate by Date:
Prev: Channel Access and Windows 95 Ken Evans
Next: Re: April Collaboration Meeting - Hotels Bob Dalesio
Index: 1994  1995  1996  1997  <19981999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
Navigate by Thread:
Prev: Channel Access and Windows 95 Ken Evans
Next: Re: April Collaboration Meeting - Hotels Bob Dalesio
Index: 1994  1995  1996  1997  <19981999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
ANJ, 10 Aug 2010 Valid HTML 4.01! · Home · News · About · Base · Modules · Extensions · Distributions · Download ·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing ·