Re: [ptxdist] OpenSSH patches

2015-07-29 Thread Clemens Gruber
On Tue, Jul 14, 2015 at 12:03:47PM +0200, Michael Olbrich wrote:
 On Tue, Jul 07, 2015 at 10:52:52AM +0200, Clemens Gruber wrote:
  what do you think about my recent OpenSSH patches (not the version bumps) 
  but
  enabling the sandbox per default (to use seccomp if available) and the 
  switch
  from DSA to Ed25519. ArchLinux and current Debian both generate Ed25519 
  pubkeys
  by default and add them as HostKey to sshd_config.
  They keep DSA and ECDSA but as they fall apart completely if the random 
  numbers
  used are not good, I am not sure this is a good idea for embedded systems 
  where
  entropy is often very scarce. Ed25519 is not that sensitive to entropy 
  problems.
 
 I looked at what Debian is doing, and I liked it: Basically, the postinst
 script checks the sshd_config and generated the needed keys. I think we can
 do the same in the ssh rc-once script.
 Then we can choose more restrictive defaults. Then those that need other
 keys just need to overwrite sshd_config in the BSP.
 
 Michael

Hi Michael,

what do you think about my recent patch (openssh: improve rc.once.d script and
harden sshd_config) from July 17th? I tried to follow what Debian is doing for
key generation and I also hardened the default config.

Regards,
Clemens

-- 
ptxdist mailing list
ptxdist@pengutronix.de


Re: [ptxdist] OpenSSH patches

2015-07-14 Thread Michael Olbrich
On Tue, Jul 07, 2015 at 10:52:52AM +0200, Clemens Gruber wrote:
 what do you think about my recent OpenSSH patches (not the version bumps) but
 enabling the sandbox per default (to use seccomp if available) and the switch
 from DSA to Ed25519. ArchLinux and current Debian both generate Ed25519 
 pubkeys
 by default and add them as HostKey to sshd_config.
 They keep DSA and ECDSA but as they fall apart completely if the random 
 numbers
 used are not good, I am not sure this is a good idea for embedded systems 
 where
 entropy is often very scarce. Ed25519 is not that sensitive to entropy 
 problems.

I looked at what Debian is doing, and I liked it: Basically, the postinst
script checks the sshd_config and generated the needed keys. I think we can
do the same in the ssh rc-once script.
Then we can choose more restrictive defaults. Then those that need other
keys just need to overwrite sshd_config in the BSP.

Michael

-- 
Pengutronix e.K.   | |
Industrial Linux Solutions | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0|
Amtsgericht Hildesheim, HRA 2686   | Fax:   +49-5121-206917- |

-- 
ptxdist mailing list
ptxdist@pengutronix.de


Re: [ptxdist] OpenSSH patches

2015-07-07 Thread Bruno Thomsen
Hey Clemens,

 what do you think about my recent OpenSSH patches (not the version bumps) but 
 enabling the sandbox per default (to use seccomp if available) and the switch 
 from DSA to Ed25519. ArchLinux and current Debian both generate Ed25519 
 pubkeys by default and add them as HostKey to sshd_config.
 They keep DSA and ECDSA but as they fall apart completely if the random 
 numbers used are not good, I am not sure this is a good idea for embedded 
 systems where entropy is often very scarce. Ed25519 is not that sensitive to 
 entropy problems.

I agree, when possible we should use ed25519. Adam Langley blog entry[1].

/Bruno

[1] https://www.imperialviolet.org/2013/06/15/suddendeathentropy.html

-- 
ptxdist mailing list
ptxdist@pengutronix.de