1

Summary: The Java VM running Eclipse on my system appears to be working, but internally it gets segfaults constantly.

Steps to reproduce:

  1. Download eclipse-cpp-kepler-SR2-linux-gtk-x86_64.tar.gz from http://www.eclipse.org/cdt/downloads.php.
  2. Extract it as /path/to/eclipse.
  3. Run gdb /path/to/eclipse/eclipse.
  4. Do set follow-fork child so that GDB will trace the actual Java process and not just the Eclipse launcher.
  5. Type run to start Eclipse. You might have to select a workspace directory.
  6. You should get a segmentation fault (SIGSEGV) pretty soon. Type cont in GDB and you will get another. Type cont again and you will get another. And so on ad nauseam.

Again, Eclipse appears to be working, presumably because it is catching the signal and recovering somehow.

My OS is 64-bit Red Hat Enterprise Linux 6.5 with all updates as of yesterday (2014-May-22). The Java RPM is java-1.7.0-openjdk-1.7.0.55-2.4.7.1.el6_5.x86_64, the current release (Version 7 Update 55) from http://java.com/ produces the same results.

I am curious to know whether this behavior happens for other people and on other flavors of Linux. More importantly, I am curious if anybody knows whether this is "normal".

(In case you are wondering... Although Eclipse appears to be working, I am concerned that some real problem is being hidden. I noticed this because I am getting occasional, not-entirely-reproducible segfaults in a plug-in that uses WebKit. I thought I would try to observe the segfault under GDB, but this turns out to be tricky because the JVM is segfaulting over and over right from the beginning, even on a fresh install without the offending plug-in.)

7
  • No; Its not normal. You will need to provide more information. For example what plugin?
    – Ramhound
    May 22, 2014 at 18:51
  • @Ramhound: It's a commercial plugin (IBM Rational Team Concert), but that is not relevant to my question... As I said, this "continuous segfault" behavior happens on a freshly-installed Eclipse with no plugins whatsoever (well, except for the integrated CDT). My "steps to reproduce" are complete; let me know if they are not.
    – Nemo
    May 22, 2014 at 18:56
  • I will decide if something is relevant I asked the question for a reason
    – Ramhound
    May 22, 2014 at 19:20
  • @Ramhound: And I answered your question, even though I do not see how a plug-in that was never installed could cause the problem.
    – Nemo
    May 22, 2014 at 21:16
  • 1
    See this it appears that SIGSEGV is expected
    – greg-449
    May 24, 2014 at 10:48

1 Answer 1

4

The JVM uses a number of signals during normal operation so the SIGSEGVs are expected. SIGSEGV is used when dealing with NullPointerException.

See this article for more detail.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .