glibc-2.2.2 breaks java 1.3?

[Michael R. Jinks]

Greetings; forgive me if this is a known issue, but I couldn't find
anything searching the glibc bug database.

I'm a sysadmin for a bunch of Java developers.  Lately we've been
rolling out new machines under Red Hat 7, and found that a known bug in
glibc-2.2 causes javac to miss error codes generated by its
subprocesses; the recommended fix from the online JDK knowledge base is
an upgrade to Red Hat's packaged release of glibc-2.2.2.

Well, we applied that to our machines, but now we have a worse problem. 
javac runs and gets its error codes just fine, but java itself hangs,
endlessly calling nanosleep(2).  From the output of strace, it seems as
though the glibc upgrade induced java to attempt to use native threads
instead of green threads, something goes wrong with that, java gets
confused, eternal loop.

Our workaround for now is to stay with glibc-2.2 and always pass the
-classic argument to java and javac, which works, but causes a
significant performance hit.

Can anyone here advise?  Is this a known issue?  User error?  Or (my
personal preference) is this just further proof that Java is a tool of
the Evil One? (I know about C#, I meant that other Evil One.)

I'm happy to provide any forensic information that might be helpful,
left it out of this posting because I'm not sure what would be useful. 
We're using the Sun JDK, version 1.3.0_02, in case that matters/helps.

Thanks a lot,
-m

-- 
Michael Jinks, IB // Technical Entity // Saecos Corporation
Opinions expressed above are my own, and not those of my employer.