additional woes with 1.4.2-p7 on 5.4: Runtime.exec()?

[Nick Johnson]

Ran into some more problems today.  If I configure Resin to use jikes as 
its compiler, the first time it tries to compile something, the JVM goes 
unresponsive and begins consuming a large amount of CPU.  truss -p shows 
the JVM in a tight kse_release loop.

Ran gcore on it, and got this for a stack trace:

#0  0x48097353 in pthread_testcancel () from /usr/lib/libpthread.so.1
#1  0x48098293 in __error () from /usr/lib/libpthread.so.1
#2  0x4808493a in pthread_yield () from /usr/lib/libpthread.so.1
#3  0x48083a11 in pthread_create () from /usr/lib/libpthread.so.1
#4  0x4808348d in pthread_create () from /usr/lib/libpthread.so.1
#5  0x484609fd in os::create_thread () from /usr/local/jdk1.4.2/jre/lib/i386/client/libjvm.so
#6  0x484aa430 in JavaThread::JavaThread () from /usr/local/jdk1.4.2/jre/lib/i386/client/libjvm.so
#7  0x483f3f68 in JVM_StartThread () from /usr/local/jdk1.4.2/jre/lib/i386/client/libjvm.so

After frame 7, there's just a load of garbage like this:

#8  0x4a3d9200 in ?? ()
#9  0x082a7698 in ?? ()
#10 0xbf073be4 in ?? ()
#11 0x00000001 in ?? ()
#12 0x4c7d1b10 in ?? ()
#13 0xbf073bb0 in ?? ()
#14 0x00000000 in ?? ()
#15 0xbf073be4 in ?? ()

Which I assume is probably hotspot stuff executing?  Or maybe just garbage 
in the stack... Eventually it comes to this:

#66 0x485749c4 in __JCR_LIST__ () from 
/usr/local/jdk1.4.2/jre/lib/i386/client/libjvm.so
#67 0x4a3d00d7 in ?? ()
#68 0xbf073d00 in ?? ()
#69 0xbf073d48 in ?? ()
#70 0x4838b057 in JavaCalls::call_helper () from /usr/local/jdk1.4.2/jre/lib/i386/client/libjvm.so
Previous frame inner to this frame (corrupt stack?)

Again, it looks like there's some trouble at pthread_testcancel, though 
that could of course be coincidental.

If I have Resin use the internal compiler, then there's no problem; I 
suspect that the trouble may be related to making a Runtime.exec call to 
start jikes.

    Nick

-- 
"The aptly-named morons.org is an obscenity-laced screed..."
 	-- Robert P. Lockwood, Catholic League director of research
Nick Johnson, version 2.1                             http://web.morons.org/