RE: [ng-spice-devel] [ng-spice-bugs] Segmentation fault in realloc (fwd)
Paolo & all,
Regarding the segmentation fault, on Windows I get an error as well. On my
version, it happens in the following stack trace:
free(void * 0x003570a0) line 926 + 11 bytes
txfree(void * 0x003570a0) line 116 + 9 bytes
INPremTerm(char * 0x003570a0, sINPtables * 0x00352c30) line 243 + 11 bytes
IFdelUid(void * 0x00352840, char * 0x003570a0, int 16) line 100 + 19 bytes
CKTdltNNum(void * 0x00352840, int 11) line 55 + 23 bytes
BJTunsetup(sGENmodel * 0x00353a40, CKTcircuit * 0x00352840) line 283 + 16
bytes
CKTunsetup(CKTcircuit * 0x00352840) line 96 + 29 bytes
I'm guessing here, but perhaps someone decided to be aggressive about
cleaning up memory and added some delete calls.
From my previous experience, this is a risky endeavor. What happens is that
some other pointer in the program is pointing to that same object in memory,
and then all hell breaks loose. The worst part is that the error shows up
ages after the error really occurred, and it's damn near impossible to track
down.
Please let me and the others know if this is the case, and if so, where the
extra deletions are coming from. I have another idea to track it down.
There is an old memory bug tracking program called MemCheck from a company
called Stratosware. I don't know if they still exist, but I'm going to try
to get a copy and see if it helps. It might be able to identify the problem
at its source.
Jon Engelbert
President, Beige Bag Software
279 E. Liberty, Ann Arbor, MI 48105
jon@beigebag.com
-----Original Message-----
From: p.nenzi@ieee.org [mailto:pnenzi@ieee.ing.uniroma1.it]
Sent: Tuesday, June 26, 2001 10:25 AM
To: ng-spice-devel@ieee.ing.uniroma1.it
Subject: [ng-spice-devel] [ng-spice-bugs] Segmentation fault in realloc
(fwd)
---------- Forwarded message ----------
Date: Mon, 25 Jun 2001 07:51:59 +0200
From: Thomas.Budmiger@flowtec.endress.com
Reply-To: ng-spice-bugs@ieee.ing.uniroma1.it
To: ng-spice-bugs@ieee.ing.uniroma1.it
Subject: [ng-spice-bugs] Segmentation fault in realloc
Hello
Since I've updated my system to linux-2.4.4 with glibc2 and gcc-2.95.3,
ng-spice crashes as soon as
an analysis is started. Netlist reading seems to be ok, a segmentation
fault is the only thing which happens
after starting a transient (or ac) analysis. This occures with the
fourbitadder example as well as with other
circuits.
I tried both, with and without garbage collector 6.0alpha8 - there is no
difference. Also, it doesn't matter
whether I use ng-spice-rework13 or 14pre3 (april 22 release).
Any ideas to get it work again ? Thanks for helping.
Thomas
Partial thread listing: