[GSoC] : Interest in project

[Arjun S R]

I am final year Computer Science graduate student. I went through the
list of projects that NetBSD is offering through GSoC.
I found a few of them interesting to work on

1) Implement RFC 6056: "Recommendations for Transport-Protocol Port
Randomization"
2) Socket option to timestamp UDP packets in the kernel

Could someone throw light into the subject on what all files on the
source tree should I learn to work with?


``
Arjun S R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Chuck Swiger]

Hi, Arjun--

Post by Arjun S R
I am final year Computer Science graduate student. I went through the
list of projects that NetBSD is offering through GSoC.
I found a few of them interesting to work on
1) Implement RFC 6056: "Recommendations for Transport-Protocol Port
Randomization"
2) Socket option to timestamp UDP packets in the kernel
Could someone throw light into the subject on what all files on the
source tree should I learn to work with?

For #1, there's a patchset for FreeBSD, which ought to largely apply to NetBSD's sys/netinet/in_pcb.c as well:

http://people.freebsd.org/~bz/20110313-01-rfc6056.diff

For #2, I seem to see something called SO_TIMESTAMP & SO_OTIMESTAMP in TRUNK of sys/netinet/udp_usrreq.c. Perhaps someone already implemented it?

Regards,

[Alistair Crooks]

Post by Chuck Swiger
Hi, Arjun--

Post by Arjun S R
I am final year Computer Science graduate student. I went through the
list of projects that NetBSD is offering through GSoC.
I found a few of them interesting to work on
1) Implement RFC 6056: "Recommendations for Transport-Protocol Port
Randomization"
2) Socket option to timestamp UDP packets in the kernel
Could someone throw light into the subject on what all files on the
source tree should I learn to work with?

http://people.freebsd.org/~bz/20110313-01-rfc6056.diff

Well, RFC 6056 made me think that just cloning FreeBSD's
implementation would be, well, let's just say that implementing all of
the approaches in the RFC would be better.

And, for the minimalists amongst us, reed@ already did this:

http://mail-index.netbsd.org/tech-net/2008/07/11/msg000629.html

so there's a bit more to this project than just copying FreeBSD code.

Regards,
Alistair

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Chuck Swiger]

Post by Alistair Crooks

Post by Chuck Swiger
http://people.freebsd.org/~bz/20110313-01-rfc6056.diff

Well, RFC 6056 made me think that just cloning FreeBSD's
implementation would be, well, let's just say that implementing all of
the approaches in the RFC would be better.

I've read both the RFC and the patch, and, as far as I see, all five of the suggested algorithms are implemented in BZ's patch.

Post by Alistair Crooks
http://mail-index.netbsd.org/tech-net/2008/07/11/msg000629.html
so there's a bit more to this project than just copying FreeBSD code.

That appears to be implementing algorithm 1, much like FreeBSD, OpenBSD, and Solaris already do. From the RFC, NetBSD's current default behavior doesn't try to randomize port #'s at all:

"A.3. NetBSD

NetBSD 5.0.1 does not obfuscate its ephemeral port numbers. It
selects ephemeral port numbers from the range 49152-65535, starting
from port 65535, and decreasing the port number for each ephemeral
port number selected [NetBSD]."

Regards,

[Christos Zoulas]

Post by Chuck Swiger
For #2, I seem to see something called SO_TIMESTAMP & SO_OTIMESTAMP in
TRUNK of sys/netinet/udp_usrreq.c. Perhaps someone already implemented
it?

I have updated the project with the current status and more information.

http://wiki.netbsd.org/projects/gsoc_2011/kernel_udp_timestamping/

christos


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Alexander Nasonov]

Post by Christos Zoulas
I have updated the project with the current status and more information.
http://wiki.netbsd.org/projects/gsoc_2011/kernel_udp_timestamping/

It's worth mentioning there that Linux stuff shouldn't be blindly
copied. In my opition, it gives too many options to a user (e.g.
TX_SOFTWARE, TX_HARWARE, RAW_HARDWARE and SYS_HARDWARE) and returning
each full packet in ERRQUEUE or over IP_MULTICAST_LOOP is a bit too
expensive. Something that can truncate packets and aggregate them
would be a better fit. This sounds like bpf but not quite.

Alex

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Arjun S R]

Hi,
Went through the RFC 4086 "Randomness Requirements for Security".
Need to go through the code of linux random device code(or
documentation atleast ) to get some more idea to work on.

--
Arjun S R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Arjun S R]

Hi,
 Is there any research papers that I should read before i get to read
the kernel api to implement the required task ?

--
Arjun S R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Arjun S R]

My proposal

http://bit.ly/dMLYZx

Arjun S R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de