Judd, I saw this with my CUDA bindings but didn't know how to dig out a root cause. Fortunately for me, the problem never effected my calculations since it always happened after the last END block, and it went away after I upgraded my Ubuntu box to the version officially supported by nVidia.
My guess was that my underlying libs suffered some weird version skew problem. Have you tried this code on a clean install? David On Aug 12, 2011 2:27 PM, "Judd Taylor" <jud...@orbitalsystems.com> wrote: > I've got some code that used to be using Java::Import to talk to a java > server using RMI. > > The server code has had a major update, and is no longer compatible with > Java::Import, due to CNI limitations. > > So I've switched it over to Inline::Java. This has no inline Java code, > just some STUDY of the Java Naming and System libs. > > It seems to be working fine, but on some terminals, I get a segfault > when the perl process exits. It looks like a couple of times it happened > in a an sv_free section of libperl.so. > > On a different terminal (same user), it seems to working with no issues > at all. > > If I do a perl -w -c on the perl module, I get errors in DESTROY, saying > it can't find object 0 for the main thread for Naming and System (as I > don't store object, just make calls to static class methods). This > happens for both the working and non-working shell sessions. > > This is a pretty critical portion of software, and the last thing I need > is a JVM flaking out on me at random times. Anybody seen this before? > > Thanks, > Judd > > > > -- > ____________________________ > Judd Taylor > Software Engineer > > Orbital Systems, Ltd. > 3807 Carbon Rd. > Irving, TX 75038-3415 > > jud...@orbitalsystems.com > (972) 915-3669 x127
