[Kst] [Bug 112689] crash after period of inactivity

[George Staikos]

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
         
http://bugs.kde.org/show_bug.cgi?id=112689         




------- Additional Comments From staikos kde org  2005-09-15 23:21 -------
On Thursday 15 September 2005 16:57, Matthew Truch wrote:

> I had a kst session open, looking at some static dirfile data, including
> displaying some psds of the data.  I was not doing anything with kst, and
> it spontaneously crashed after a few minutes of letting kst be idle.  It
> seems hard to reproduce, but I have seen a few 'quirks' like this in the
> recent past, although I didn't investigate.  Backtrace was invalid, but the
> following was spit out during the crash:
>
> $ kst
> kst: kstpsd.cpp:340: virtual KstObject::UpdateType KstPSD::update(int):
> Assertion `_last_n_new >= 0' failed. KCrash: Application 'kst' crashing...
> /proc/meminfo: Bad file descriptor
> $
>
> I don't know if anyone will be able to debug this, so feel free to close if
> you feel it isn't worth it.


   Excellent.  I've been waiting for this to come up.  It indicates a bug 
somewhere in Kst where either _last_n_new is overflowing, or NumNew is 
negative in KstVector.  I'll try reading the code to find more possible cases 
where this can happen, but it will be much easier if you can produce a 
testcase for me.  I need to know which types of objects are being used in 
particular, but being able to run it locally will really help.

   The backtrace will never be useful because it asserts in the worker thread 
which causes a nasty memory corruption.  It might be worth reporting this to 
trolltech as a QThread limitation/issue.