> From: "Paul Wouters" <[email protected]> 
> Date: 04/14/15 11:17 

> Restriction of algorithms will be done post RHEl-7.1 (and is not strictly
> a requirement of FIPS, you can document that one should not use MD5
> without blocking MD5)

It's a few times now that I see this.  A device offers non-FIPS option but, the 
user guide says not to use them.  Seems all OK for validation purposes.  To 
extrapolate, I guess a device could offer SNMP v1, v2c and v3 witha  FIPS user 
guide that says 'please do not use SNMP v1 and v2c'.
 
> However, current libreswan git head (which will become 3.13) does have
> these restrictions enforced now. Which means, MD5, TWOFISH and SERPENT
> are not available for IKE or ESP.

OK ! Thanks !
 
> > So far I can say that putting the kernel through FIPS validation
> > is not something that was ever mentioned with the consultants.
> 
> It's very expensive. It might be much better to pick a kernel that has
> been FIPS certified when you can.

Hmmm...
 
> That's because the XFRM itself does not perform cryptographic
> operations. The kernel crypto API does that, and it is FIPS certified on
> its own:
 
> http://csrc.nist.gov/groups/STM/cmvp/documents/140-1/140sp/140sp1387.pdf

Ah.
 



_______________________________________________
Swan mailing list
[email protected]
https://lists.libreswan.org/mailman/listinfo/swan

Reply via email to