threads / discuss / 19486

[JGIT] Bus error (SIGBUS) on Solaris

Subject: [JGIT] Bus error (SIGBUS) on Solaris

## tl;dr

2 messages between May 24, 2009 and May 24, 2009.

replies: 1people: 2as markdown or json

John Bito· May 24, 2009, 18:13 UTC · lore

I updated with the integration build on the update site today, after restart, Eclipse failed with a bus error.

Show 5 quoted lines
> #  SIGBUS (0xa) at pc=0xf88162fc, pid=15191, tid=23
> #
> # Java VM: Java HotSpot(TM) Client VM (11.2-b01 mixed mode solaris-sparc)
> # Problematic frame:
> # j  org.spearce.jgit.lib.OffsetCache.getOrLoad(Lorg/spearce/jgit/lib/PackFile;J)Ljava/lang/Object;+17

I haven't updated my WinVista workstation, yet.  Shall I create a bug report with the full error log, or is this something that's already known?

The failure seems repeatable with my current workspace configuration.
Robin Rosenberg· May 24, 2009, 20:11 UTC · re: John Bito · lore

Re: [JGIT] Bus error (SIGBUS) on Solaris

söndag 24 maj 2009 20:13:24 skrev John Bito:
Show 14 quoted lines
> I updated with the integration build on the update site today, after
> restart, Eclipse failed with a bus error.
> 
> > #  SIGBUS (0xa) at pc=0xf88162fc, pid=15191, tid=23
> > #
> > # Java VM: Java HotSpot(TM) Client VM (11.2-b01 mixed mode solaris-sparc)
> > # Problematic frame:
> > # j  org.spearce.jgit.lib.OffsetCache.getOrLoad(Lorg/spearce/jgit/lib/PackFile;J)Ljava/lang/Object;+17
> 
> I haven't updated my WinVista workstation, yet.  Shall I create a bug
> report with the full error log, or is this something that's already
> known?
> 
> The failure seems repeatable with my current workspace configuration.

There is no way a Java bug can crash the VM by itself. You can hit either a real VM bug or a bug in native code used by the VM.

So you have one or more
1) A broken Windows installation (some might consider this an tautology).
2) A corrupt JRE installation. Reinstall and see if things works better
3) A corrupt Eclipse. Try a fresh Eclipse.
4) Bad hardward, memory, malware leading to broken files (1).

2 & 3 are easy to test. If reinstalling the JRE doesn't help, try t he newest JRE available from SUN and see, Sun JVM bug reports should go to SUN. We will do absolutely nothing with them, though it would be interesting to get the ticket id when it gets assigned one.

-- robin '

← back to recent threads