Hi Deepak,

if both sides use the same Linux kernel implementation then
interoperability with sha2_256_96 is of course possible.

Unfortunately there is no stronger alternative with true 128 bit
strength since AES-XCBC data integrity is also truncated to 96 bits
and AES-GCM is not available yet with 2.6.18.

Best regards

Andreas

Gupta, Deepak (Deepak) wrote:
> Andreas,
> 
> Many thanks for the info.
> 
> Please consider, if we do not have the luxury to use this kernel
> patch for esp+sha256, i.e., both ends of the tunnel, then can the 2
> ends still use esp128-sha2_256?  Or does this non-conforming 96bit
> truncation rule out this combination for phase2?  In the latter case,
> what are our options, are we then only limited to sha1?
> 
> -Deepak
> 
> 
> -----Original Message----- From: Andreas Steffen
> [mailto:[email protected]] Sent: Tuesday, November 10,
> 2009 3:22 PM To: Gupta, Deepak (Deepak) Cc: 'Martin Willi';
> '[email protected]' Subject: Re: [strongSwan] Strongswan
> support for RHEL5
> 
> Gupta, Deepak (Deepak) wrote:
>> Martin,
>> 
>> Many thanks for your reply! I appreciate it.
>> 
>> Your answers are very helpful.  To give you some context, we are in
>> an effort to support IPsec tunnels with RedHat RHEL 5.3 kernel 
>> (2.6.18-128) (I misstyped before, it is not 5.2).  And based on
>> your answers, it appears that strongswan meets our requirements
>> more than openswan (would you agree based on the requirements I had
>> mentioned?), however, is the kernel issue (sha256+esp) strongswan
>> specific?  I would tend to think that this would be an issue for
>> all Ipsec solutions.
>> 
> Yes, this is a general issue with the Linux 2.6 kernel. We offered a
> kernel patch more than a year ago which was unfortunately rejected by
> the kernel developers because they wanted backward compatibility with
> the old non-conforming 96-bit truncation.
> 
>> Regarding the bug for "ipsec statusall", I googled and saw an email
>>  from Johannes that said that RHEL version 5.4 contains the fix for
>>  this, however, it did not mention the RedHat bug number or the
>> kernel patch.  However, Johannes' email from this morning suggests
>> that this fix went into 5.3 (2.6.18-164 or earlier).  Is this issue
>> Ipsec issue or a strongswan issue?
>> 
> This is an IPsec issue with the Linux 2.6 kernel. After some haggling
> RedHat recognized this and fixed the bug.
> 
> BTW, if there is any info available on this
>> bug (redhat bug number etc.) or linux kernel patch, please let us
>> know so we can try to evaluate how best to apply it.  I have done
>> some limited searching for this on RedHat's web site but did not
>> find this bug for either the 5.3 (2.6.18-128) or 5.4 kernels.
>> 
>> Regards,
>> 
>> -Deepak
> 
> Best regards
> 
> Andreas
=====================================================================
Andreas Steffen                         [email protected]
strongSwan - the Linux VPN Solution!                www.strongswan.org

Institute for Internet Technologies and Applications
University of Applied Sciences Rapperswil
CH-8640 Rapperswil (Switzerland)
===========================================================[ITA-HSR]==

_______________________________________________
Users mailing list
[email protected]
https://lists.strongswan.org/mailman/listinfo/users

Reply via email to