[CentOS] kernel eclipse segfault

David Sommerseth dazo at users.sourceforge.net
Tue Mar 1 14:23:15 UTC 2011


On 01/03/11 08:02, sync wrote:
> Hi, all :
> 
> My user called Tom, and his vnc desktop suddenly did not work just now.
> Then I went to see the server  log message , and found the following message:
> 
> kernel: Xvnc[2779]: segfault at 00000000008 rip 000000000077af44 rsp
> 00007fff5ee310c0 error 4
> kernel: eclipse[25300]: segfault at 00000000008 rip 000003210cfbc6c rsp
> 00007fff470c66a0 error 4
> gconfd( tom-2764): Gconf server is not in use , shutting down
> gconfd: Exitting ..
> 
> 
> By the way, my server os is CentOS 5.3 x86_64 and kernel version is :
> 2.6.18-128.el5
> 
> 
> It looks it is the eclipse tool  problem which cause the kernel crash,
> isn't it ?

Partly correct.  The kernel kills the program with SEGV, because the
program has tried to do something it shouldn't do.  That's why you see that
segfault message in dmesg/log files.  It is not an indication of kernel
failure.  It's an indication the kernel took action.

> Could someone can give me some suggestion which can avoid that message
> happend again?

You need to figure out which program causes this error.  You have both Xvnc
and eclipse failures.  It might be that when one of them segfaults, it
brings down the other one too, as a segfault.

It's impossible to say exactly what the error is, just based on this
information.  The only thing which is clear, it's a user space error and
not a kernel issue.


Btw.  you're running on a very old CentOS base (5.3).  I would first try to
upgrade to at 5.5, and see if the error still is persistent.  5.3 has not
received much love since 5.4 and 5.5 is released.  Now all the focus is on
getting 5.6 out the door too.


kind regards,

David Sommerseth




More information about the CentOS mailing list