Possible segfault in nft utility

Possible segfault in nft utility

[Meyer Raffaele]

Dear Sir or Madam

I am currently using nftables under Ubuntu 15.10. Kernel version is 4.2.0-16-generic and nftables version is nftables v0.4 (Support Edward Snowden).
I was looking for a way to match packets based on the mss optional value in the tcp header. While using the describe command, I encountered a segfault:

user@machine:~$ nft describe tcp reserved
Segmentation fault (core dumped)

It is not really a bad bug since the reserved keyword is probably not used anyway but I thought to report it. At the same time I wanted to ask if it is possible to match packets on wether they have a MaximumSegmentSize set in the TCP header or not. I want this since most of the TCP-Syn flood tools seem to not set this option and most legitimate TCP stacks do set it.

Greetings
Raffaele Meyer

Re: Possible segfault in nft utility

[Pablo Neira Ayuso]

On Fri, Apr 08, 2016 at 11:10:47AM +0000, Meyer Raffaele wrote:
> Dear Sir or Madam
> 
> I am currently using nftables under Ubuntu 15.10. Kernel version is 4.2.0-16-generic and nftables version is nftables v0.4 (Support Edward Snowden).
> I was looking for a way to match packets based on the mss optional value in the tcp header. While using the describe command, I encountered a segfault:
> 
> user@machine:~$ nft describe tcp reserved
> Segmentation fault (core dumped)

This is already fixed in 0.5 as weel as in the upcoming 0.6 release.

# nft describe tcp reserved 
payload expression, datatype integer (integer), 4 bits

> It is not really a bad bug since the reserved keyword is probably
> not used anyway but I thought to report it. At the same time I
> wanted to ask if it is possible to match packets on wether they have
> a MaximumSegmentSize set in the TCP header or not.

Several people already asked for TCP options matching already, this is
coming soon.