From: Max G. <gi...@li...> - 2000-11-14 21:06:28
|
Hi! I'm experiencing continuous JVM crashes because of GL4Java (GLFuncJauJNI is listed on top of program trace each time). The problem is that those crashes are semi-random ie. I change something insignificant (like variable initial value or even launch another application like currently Netscape) and problem disappears. I don't have any idea how this behavior can be reproduced but it's *very* tiresome and it occures *very* often (several times per hour - note that I have to wait for 130MB core to be dumped each time - it takes time). Again - I don't have to change *anything* in application (but it sometimes works too), just change system environment a little and it 'solves' problem. If anyone has ANY ideas what should I do, I would be VERY grateful. BTW, remember my posts from about two weeks ago? I believe it was one of such situations. I'm using GNU/Linux RedHat 7.0, GL4Java 2.4.0.0 and Sun JDK 1.3. Bye, Max -- ----------------------------------------------- Max Gilead gi...@li... XMage library http://xmage.sourceforge.net ----------------------------------------------- -- Sex on TV is bad. You might fall off... |
From: Sven G. <sgo...@ja...> - 2000-11-17 11:58:09
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Tue, 14 Nov 2000, Max Gilead wrote: > Hi! > > I'm experiencing continuous JVM crashes because of GL4Java (GLFuncJauJNI > is listed on top of program trace each time). The problem is that those > crashes are semi-random ie. I change something insignificant (like > variable initial value or even launch another application like currently > Netscape) and problem disappears. I don't have any idea how this > behavior can be reproduced but it's *very* tiresome and it occures > *very* often (several times per hour - note that I have to wait for > 130MB core to be dumped each time - it takes time). > > Again - I don't have to change *anything* in application (but it > sometimes works too), just change system environment a little and it > 'solves' problem. > > If anyone has ANY ideas what should I do, I would be VERY grateful. > Well - this looks like a stack overflow behavior ! I guess it is NOT GL4Java's native libraries, but I am not shure ! GL4Java gives you the following chances to create a stack overflow, or a mysterious memory crash: - allocating a too small array, passing the array to OpenGL, which needs a bigger array ! This semantics are not checked. I have just tried to create this kind of crash, but I was not succesfull :-( Demo "select.java": I gave the "vp"-array just a zero size. "vp" is passed to GL and GLU ! But it works ... may be because the JVM allocates enough memory, so we did not crossed the border to an illegal memory area .. ?! I suppose, that - if somebody has the time - to add semantic arrays checks for another debug-test library ! But this task would be _very_ time consuming, because the many functions we do have. This "user" mistake, or a bug within the GL driver (which one are you using ?), can create those mysterical results ! Try to make JVM's startup and average memory heaps bigger ! If no error will occure, I really would guess, it is about a wrong arrays size. Comments and Ideas about automatically array-size checking for Java<->JNI are welcome. > BTW, remember my posts from about two weeks ago? I believe it was one of > such situations. > I must check my email archive. > I'm using GNU/Linux RedHat 7.0, GL4Java 2.4.0.0 and Sun JDK 1.3. > Which OpenGL driver ? Mesa 3.X ? XFree86 4.0.X - DRI / XXX ? NVidia ? > Bye, > Max > It would be cool, to have a very array consuming Java/OpenGL example, which always tries to run out of any bounds, even the JVM pre-allocated memory bounds. Kind regards, Sven > -- > ----------------------------------------------- > Max Gilead gi...@li... > XMage library http://xmage.sourceforge.net > ----------------------------------------------- > -- Sex on TV is bad. You might fall off... - -- mailto:sgo...@ja... www : http://www.jausoft.com ; pgp: http://www.jausoft.com/gpg/ voice : +49-521-2399440, +49-170-2115963; fax: +49-521-2399442 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.4 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE6FR0DHdOA30NoFAARAgpNAJ9a7AvqGL+9BO9VB5sk+Jy+twggVQCgtEZM 8htlxooRaVG8TUjOGPty/tI= =xD7w -----END PGP SIGNATURE----- |
From: Max G. <gi...@li...> - 2000-11-18 23:14:07
|
Sven Goethel wrote: > Well - this looks like a stack overflow behavior ! > I guess it is NOT GL4Java's native libraries, but I am not shure ! This happens in various places but always when calling GL4J, is it still possible to be an overflow problem? If so, I'll investigate it deeply, thx for suggestion. BTW, would it be a big problem to check stack and throw an exception instead of segfaulting whole JVM? > Try to make JVM's startup and average memory heaps bigger ! > If no error will occure, I really would guess, it is about a wrong arrays > size. No difference, I tried. > Mesa 3.X ? 3.3 > XFree86 4.0.X - DRI / XXX ? 4.0.1 which came with RedHat > NVidia ? TNT, but not accelerated (didn't have the time to investigate how to turn it on) Max -- ----------------------------------------------- Max Gilead gi...@li... XMage library http://xmage.sourceforge.net ----------------------------------------------- -- Sex on TV is bad. You might fall off... |