Re: [ROOT] CustomReAlloc2, 4.00.10, SuSE9.0

From: Rene Brun (brun@pcbrun.cern.ch)
Date: Sat Oct 16 2004 - 10:26:13 MEST


Hi Martin,

This looks like a mismatch between the new/delete operators.
When using shared memory, the liNew.so library should be loaded first.
This library redefines teh standard new/delete operators.

Could you check that this is true in your link sequence.

Rene Brun

On 
Sat, 16 Oct 
2004, schulte-wissermann wrote:

> hi all,
>  i use a TMapFile and a very qNd-programed class to
> display the online histos at our experiment. the general architecture of 
> this setup is from about 1999 and i know that it is everything 
> but 'a professional solution'. however, it displays our about 5000 histos
> and it used to work without problems since then. 
> yesterday, i upgraded the computer from 3.10.xx to 4.00.08 and from 
> suse7.3 (gcc2.95) to 9.0 (gcc3.3.1). 
> since then i get (about every 100 events) an error/warning (like):
> 	CustomReAlloc2: oldsize != newsize  
> everything works, so for the moment i consider this to be a 'warning'.
> the only thing that makes me nervous is that if this will
> change to an 'error' (e.g. seg-fault) when we will have full beam i would 
> not have a direct idea what to do. and since we do not have a backup solution 
> this would really be a problem.
> now my question:
>  did enybody encounter similar problems and/or does anybody have 
> a striking and simple solution? according to google it seems as if nowbody 
> had CustomReAlloc2-problems since root_v2.23.!?
>  thanx,
>    martin
> 
> ps: it is probably not possible to provide 'a simple application 
> 	that demonstrates the problem'. however, i could try to pin down the 
> 	problem (althoug we now have beamtime, so that even my 18h day is hardly 
> 	sufficent to get the work done).
> 



This archive was generated by hypermail 2b29 : Sun Jan 02 2005 - 05:50:10 MET