Re: "Boradcasting" MAC'd data
From: David Wagner (daw_at_taverner.cs.berkeley.edu)
Date: 04/19/05
- Next message: David Wagner: "Re: Crypto Mini-FAQ"
- Previous message: Tim Peters: "Re: Now you can see, my "crank" status"
- In reply to: Carlos Moreno: "Re: "Boradcasting" MAC'd data"
- Next in thread: Carlos Moreno: "Re: "Boradcasting" MAC'd data"
- Reply: Carlos Moreno: "Re: "Boradcasting" MAC'd data"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Tue, 19 Apr 2005 20:15:51 +0000 (UTC)
Carlos Moreno wrote:
>Huh? Didn't you say you're in a client/server scheme? Doesn't
>the server have a network card?
>
>If so, there is enough "random" activity for /dev/urandom to
>generate plenty of more-than-enough-quality random numbers;
A network card, including the time at which interrupts occur due
to network packets, is not a good source of cryptographic entropy.
The fundamental problem is that an eavesdropper (or worse, an active
attacker) will know a lot about those "random" values.
I don't recall whether the Linux kernel's entropy estimator treats
network card interrupts as providing any entropy. I was under the
impression that it does not.
- Next message: David Wagner: "Re: Crypto Mini-FAQ"
- Previous message: Tim Peters: "Re: Now you can see, my "crank" status"
- In reply to: Carlos Moreno: "Re: "Boradcasting" MAC'd data"
- Next in thread: Carlos Moreno: "Re: "Boradcasting" MAC'd data"
- Reply: Carlos Moreno: "Re: "Boradcasting" MAC'd data"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
