On 2006-08-20 at 21:27:09 +0200, Noèl Köthe <[EMAIL PROTECTED]> wrote:
> forwarded 379657 http://tcpreplay.synfin.net/trac/ticket/81
> tags 379657 + upstream confirmed
> thanks
>
> Am Sonntag, den 20.08.2006, 02:12 +0200 schrieb Tobias Klauser:
> > This seems to be caused by code enclosed in #ifdef FORCE_ALIGN which,
> > according to configure.in is defined on ia64. Though this code doesn't
> > seem to be maintained. Upstream's lack of the architectures in question
> > maybe?
> >
> > FORCE_ALIGN is also defined on alpha, arm*, hp*, mips* and sparc*. Does
> > the error also show up on these architectures?
>
> Yes. The build daemons had these problems, too.
>
> > I'm trying to investigate this further, but the lack of an ia64 machine
> > makes
> > it a bit hard. :-/
>
> Thx for your feedback.
> I reported it upstream some days ago:
> http://tcpreplay.synfin.net/trac/ticket/81
The 'ipbuff issue' seems to have been fixed upstream but there still
exist another error if you look at the buildd log.
When looking a bit more deeply at the according section in the source
all seem to be ported from an old version of tcprelay, containing
variables which are not defined anymore.
Excerpt from src/tcpedit/tcpedit.c
#ifdef FORCE_ALIGN
/*
* put back the layer 3 and above back in the pkt.data buffer
* we can't edit the packet at layer 3 or above beyond this point
*/
memcpy(&newpkt[l2len], ip_hdr, pkthdr_ptr->caplen - l2len);
#endif
The variables newpkt and pkthdr_ptr are non-existent (grep -r across the
source showed nothing else). I'm really not sure but it might be
possible to just delete the section in question?
Cheers, Tobias
P.S. I might get access to a mips in the next week so I could
investigate this bug there.
