"Hamid Ali Asgari" <[EMAIL PROTECTED]> wrote: > Think logically. I have compiled the FreeRadius on two different machines > with 100% diffrent hardware: > - The first one was an Intel 1.8G with 512MB PC2100 DDR, 60 GB HDD. > - The other one was an AMD Athlon XP 1400 with 256MB DDR, 40 GB HDD. > > The only thing that was the same was the IP address. Both machines were > tested with RedHat 7.3 nad also Redhat 8.0. Both were compiled with > default settings, just the prefix was changed and the --enable-developer > option was enabled. The source codes were not modified at all. > Even when no clients were configured on the Radius server, the server > exited with the "Segmentation Fault" error after about 20 seconds.
I agree that the server is dying. I'm not sure why you would say it's a security problem. Sure, having the server die isn't a good thing, but there's no security hole to be exploited. It sounds like it's a DoS attack. That's bad, but not necessarily a security exploit. > By removing the access-lists , the Radius core dumps. Setting the ACLs > again, evverything will be OK. > My conclusion, malformed packets might be the source. I have dumped the > packets as you will see bellow.(The IP addresses have been replaced by > A.B.C.D AND X.Y.Z.T, the IP addresses of the 2 Radius servers) Ok... you've included a lot of packets to ports other than 1812. The server isn't listening on those ports, so those packets can't affect it. The ones to 1812 appear to be garbage packets. They're nothing I recognize. Try the latest version from CVS. (I think you said you were running 0.7...). Version 0.7 had an issue where certain packets could cause the server to call memset() with a NULL pointer, which would cause it to crash. This problem has been fixed in the CVS head. Alan DeKok. - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
