[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Problem with Multicast classifier: Segmentation Fault
There was a bug in in MCastClassifier::clearHash(). Please apply the
following patch, or wait for a while and grab it off the cvs
tree. I'll commit the fix right now.
-Yuri
*** classifier-mcast.cc 1999/03/13 03:52:44 1.23
--- classifier-mcast.cc 1999/06/25 20:04:09
***************
*** 106,112 ****
void MCastClassifier::clearHash(hashnode* h[], int size)
{
! for (int i = 0; i < HASHSIZE; ++i) {
hashnode* p = h[i];
while (p != 0) {
hashnode* n = p->next;
--- 106,112 ----
void MCastClassifier::clearHash(hashnode* h[], int size)
{
! for (int i = 0; i < size; ++i) {
hashnode* p = h[i];
while (p != 0) {
hashnode* n = p->next;
***************
*** 114,120 ****
p = n;
}
}
! memset(h, 0, size * sizeof(hashnode));
}
void MCastClassifier::clearAll()
--- 114,120 ----
p = n;
}
}
! memset(h, 0, size * sizeof(hashnode*));
}
void MCastClassifier::clearAll()
Joerg Diederich <[email protected]> writes:
> Hi!
>
> I played around a (very) little bit with multicast support in ns and
> got the following problem in ns-2.1b5 and the current snapshot from
> June, 16th (using also both tclcl-1.0b8 and tclcl-1.0b8-current from
> June, 16th):
>
> % set mc [new Classifier/Multicast]
> _o4
> % delete $mc
> Segmentation fault
>
> The problem does not appear in ns-2.1b4.
>
> I could not really track down the problem, but the SegFault happens
> somewhere in delete-shadow at different (changing) locations.
>
> gdb output does not help, either:
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x4016c132 in __libc_free (mem=0x0)
> Current language: auto; currently c
> (gdb) bt
> #0 0x4016c132 in __libc_free (mem=0x0)
> #1 0x8383c40 in ?? ()
> #2 0x841 in ?? ()
> Cannot access memory at address 0x2d70a.
>
>
> I compiled ns with both gcc-2.7.2 and egcs-2.91.66, same results.
>
> This is not really a problem for me as I am not using multicast that
> much and as you can simply omit deleting objects, but it may be a hint
> for other strange errors in that field...
>
>
> With best regards,
>
> /J"org
> -----
> J"org Diederich
> Institute of Operating Systems and Computer Networks,
> Technical University Braunschweig, Germany
> e-mail: [email protected]