Re: Linux JDK 1.3 and hotspot (native threads)
At Mon, 30 Apr 2001 11:59:38 +0200, Georg-W. Koltermann [EMAIL PROTECTED] wrote: On Fri, Apr 27, 2001 at 09:32:58PM -0400, Andrew Gallatin wrote: Georg-W. Koltermann writes: ... In order to get real performance I would like to run either the SUN JDK with -hotspot, or the IBM 1.3 JVM. Both of these use native linux threads. With a recent -current I can successfully execute small JAVA test programs, but when I start a real application (e.g. Together from togethersoft.com), it fails with a core dump. ... Also, are there any non commercial apps that demonstrate the problem? Or at least things that I don't have to sign my life away to get access to? Hi Andrew, I just tried Forte for Java 3.0 Early Access, and it shows the above problem quite easily during startup. I ran it with truss(1), and the last part of the output is: linux_kill(0x15bc,0x20) = 2 (0x2) linux_brk(0x8232000) = 1 (0x1) linux_rt_sigprocmask(0x2,0x0,0xbfbfc73c,0x8) = 4 (0x4) write(4,0xbfbfc728,148) = 3 (0x3) linux_rt_sigprocmask(0x2,0x0,0xbfbfc6a8,0x8) = 4 (0x4) SIGNAL 32 SIGNAL 32 SIGNAL 32 linux_rt_sigsuspend(0xbfbfc6a8,0x8) ERR#4 'Interrupted system call' linux_sigreturn(0xbfbfc3a0) ERR#4 'Interrupted system call' linux_sched_getscheduler(0x15c2) ERR#1 'Operation not permitted' linux_kill(0x15c2,0x20) = 2 (0x2) linux_kill(0x15c2,0x20) = 2 (0x2) getpid() = 0 (0x0) linux_mmap(0xbfbfc9bc) = 1 (0x1) mprotect(0xbfa0e000,0x2000,0x0) = 3 (0x3) linux_mmap(0xbfbfc9bc) = 1 (0x1) linux_sigaltstack(0xbfbfc9f8,0x0)= 2 (0x2) -- System info linux_rt_sigprocmask(0x1,0xbfbfc968,0x0,0x8) = 4 (0x4) linux_sched_getscheduler(0x15b9) ERR#1 'Operation not permitted' Product Version = Forte for Java, CE v. 3.0 (Build 010523) IDE Versioning= IDE/1 spec=1.2.1 impl=010523 Operating System = Linux version 2.2.12 running on i386 Java; VM; Vendor = 1.3.1; Java HotSpot(TM) Client VM 1.3.1-b24; Sun Microsystems Inc. Java Home = /opt/jdk1.3.1/jre System Locale = en_US (f4j_ce) Home Dir; Current Dir = /home/hunter/gwk; /home/hunter/gwk IDE Install; User Dir = /opt/forte30; /home/hunter/gwk/ffjuser30 CLASSPATH = /opt/forte30/lib/patches/openide-compat.jar:/opt/forte30/lib/ext/bsh-1_0-fj.jar:/opt/forte30/lib/ext/cmd.jar:/opt/forte30/lib/ext/cosnaming.jar:/opt/forte30/lib/ext/dd2beans.jar:/opt/forte30/lib/ext/ddl.jar:/opt/forte30/lib/ext/fjscript.jar:/opt/forte30/lib/ext/fscontext.jar:/opt/forte30/lib/ext/idlcompilers.jar:/opt/forte30/lib/ext/jaas.jar:/opt/forte30/lib/ext/jaxp.jar:/opt/forte30/lib/ext/jh.jar:/opt/forte30/lib/ext/jndi.jar:/opt/forte30/lib/ext/ldap.jar:/opt/forte30/lib/ext/ldapbp.jar:/opt/forte30/lib/ext/logger.jar:/opt/forte30/lib/ext/nis.jar:/opt/forte30/lib/ext/openorb-1.0.2.jar:/opt/forte30/lib/ext/oracle.jar:/opt/forte30/lib/ext/parser.jar:/opt/forte30/lib/ext/pbembeddedeval.jar:/opt/forte30/lib/ext/providerutil.jar:/opt/forte30/lib/ext/regexp.jar:/opt/forte30/lib/ext/rmi-ext.jar:/opt/forte30/lib/ext/rmiregistry.jar:/opt/forte30/lib/ext/sax2.jar:/opt/forte30/lib/ext/servlet.jar:/opt/forte30/lib/ext/xerces.jar:/opt/forte30/lib/ext/jdbc20x.zip:/opt/forte30/lib/locale/core_f4j.jar:/opt/forte30/lib/locale/core_f4j_ce.jar:/opt/forte30/lib/locale/openide_f4j.jar:/opt/forte30/lib/core.jar:/opt/forte30/lib/openide-fs.jar:/opt/forte30/lib/openide-nodes.jar:/opt/forte30/lib/openide-util.jar:/opt/forte30/lib/openide.jar:/opt/jdk1.3.1/lib/dt.jar:/opt/jdk1.3.1/lib/htmlconverter.jar:/opt/jdk1.3.1/lib/tools.jar --- linux_rt_sigprocmask(0x2,0x0,0xbfbfc9cc,0x8) = 4 (0x4) Unexpected Signal : 11 occurred at PC=0x806535a Function name=(N/A) Library=(N/A) NOTE: We are unable to locate the function name symbol for the error just occurred. Please refer to release documentation for possible reason and solutions. Current Java thread: Dynamic libraries: Can not get information for pid = 5570 Local Time = Fri Jun 8 23:14:38 2001 Elapsed Time = 2 # # HotSpot Virtual Machine Error : 11 # Error ID : 4F530E43505002CC # Please report this error at # http://java.sun.com/cgi-bin/bugreport.cgi # # Java VM: Java HotSpot(TM) Client VM (1.3.1-b24 mixed mode) # # An error report file has been saved as hs_err_pid5570.log. # Please refer to the file for
Re: Linux JDK 1.3 and hotspot (native threads)
Georg-W == Georg-W Koltermann [EMAIL PROTECTED] writes: Georg-W I tried the patch and found it makes no difference. The current SUN Georg-W JVM (1.3.0_02) gives the SIG11 as I indicated before. The latest IBM Georg-W JVM (IBM build cx130-20010329) hangs around and eats up CPU time. It Georg-W mostly eats system time (usage is about 8% user, 91% system). A Georg-W normal kill is ineffective, I have to send it a -9 in order to Georg-W terminate. Running SUn JVM 1.3.0_02 on Solaris 2.6 gives me exactly the same error once in a while, but only (I think) when using with -server, i.e. the hotspot server. This might well be a JVM problem, not a problem of the Linux emulation. -- Peter Mutsaers | Dübendorf| UNIX - Live free or die [EMAIL PROTECTED] | Switzerland | Sent via FreeBSD 4.3-stable To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: Linux JDK 1.3 and hotspot (native threads)
On Fri, Apr 27, 2001 at 09:32:58PM -0400, Andrew Gallatin wrote: Georg-W. Koltermann writes: ... In order to get real performance I would like to run either the SUN JDK with -hotspot, or the IBM 1.3 JVM. Both of these use native linux threads. With a recent -current I can successfully execute small JAVA test programs, but when I start a real application (e.g. Together from togethersoft.com), it fails with a core dump. ... Can you try the kernel patch in http://www.freebsd.org/cgi/query-pr.cgi?pr=26705 It sounds like this is going to get committed soon, but I'd like to know if it has any affect on your problem. Hi Drew, I tried the patch and found it makes no difference. The current SUN JVM (1.3.0_02) gives the SIG11 as I indicated before. The latest IBM JVM (IBM build cx130-20010329) hangs around and eats up CPU time. It mostly eats system time (usage is about 8% user, 91% system). A normal kill is ineffective, I have to send it a -9 in order to terminate. Also, are there any non commercial apps that demonstrate the problem? Or at least things that I don't have to sign my life away to get access to? Hmm, I don't use any. We are developing a large-scale JAVA servlet application, and that one provokes the SIG11, too. Moneydance, a shareware finance manager that I once evaluated (http://www.seanreilly.com), seems to run ok. If you really want to invest the time, you might go to togethersoft.com and download their product. They will send you an evaluation license as soon as you download the product, without any further request. But it's a big application, and you won't have any use for it unless you develop a large-scale JAVA application. -- Regards, Georg. -- Who in the world needs 2000 Windows? To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: Linux JDK 1.3 and hotspot (native threads)
Georg-W. Koltermann writes: ... In order to get real performance I would like to run either the SUN JDK with -hotspot, or the IBM 1.3 JVM. Both of these use native linux threads. With a recent -current I can successfully execute small JAVA test programs, but when I start a real application (e.g. Together from togethersoft.com), it fails with a core dump. ... Can you try the kernel patch in http://www.freebsd.org/cgi/query-pr.cgi?pr=26705 It sounds like this is going to get committed soon, but I'd like to know if it has any affect on your problem. Also, are there any non commercial apps that demonstrate the problem? Or at least things that I don't have to sign my life away to get access to? Thanks, Drew -- Andrew Gallatin, Sr Systems Programmer http://www.cs.duke.edu/~gallatin Duke University Email: [EMAIL PROTECTED] Department of Computer Science Phone: (919) 660-6590 To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Linux JDK 1.3 and hotspot (native threads)
I remember some reports of success running Linux JAVA with native threads on -current. I've tried several times myself, and can't get it to work. The Linux JDK 1.3 (SUN's version) runs fine on -current as well as 4.x as long as I use the classic VM. This version doesn't use native Linux threads. In order to get real performance I would like to run either the SUN JDK with -hotspot, or the IBM 1.3 JVM. Both of these use native linux threads. With a recent -current I can successfully execute small JAVA test programs, but when I start a real application (e.g. Together from togethersoft.com), it fails with a core dump. The SUN JDK 1.3 gives # # HotSpot Virtual Machine Error, Unexpected Signal 11 # Please report this error at # http://java.sun.com/cgi-bin/bugreport.cgi # # Error ID: 4F533F4C494E55580E43505005BC # # Problematic Thread: prio=671556272 tid=0x804f790 nid=0x2940 runnable # and the IBM JDK 1.3 dies with a similar message. Is anybody successfully using a Linux JVM with native threads on -current? Are there any pathes that you needed to apply? -- Regards, Georg. -- Who in the world needs 2000 Windows? To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message