http://bugzilla.novell.com/show_bug.cgi?id=619193 http://bugzilla.novell.com/show_bug.cgi?id=619193#c1 Karl Eichwalder changed: What |Removed |Added ---------------------------------------------------------------------------- Priority|P5 - None |P3 - Medium Status|NEW |ASSIGNED --- Comment #1 from Karl Eichwalder 2010-07-02 11:39:42 UTC --- Thanks for the detailed input--I hope I got it right. Here is my proposal: <!-- bnc#619193 --> <sect3 id="sec.113.ipsec"> <title>Incompatible IPsec and strongSwan Changes</title> <para>The "sha256"/"sha2_256" keywords now configure the kernel with 128-bit truncation, not the non-standard 96-bit truncation used by previous releases. If you depend on the 96-bit truncation scheme, use the new "sha256_96" keyword—this might be necessary, if you want to establish a connection with an old kernel (openSUSE 11.2 or earlier).</para> <para>In those case modify the connection settings to the old and non-standard 96-bit truncation in the <filename>ipsec.conf</filename> of the new system:</para> <screen>esp=aes128-sha256_96</screen> <para>Or on the peer using the old kernel, when it supports it:</para> <screen>"esp=aes128-sha256_128"</screen> <para>There is also an incompatible strongSwan change. IPComp in tunnel mode was fixed to strip out the duplicated outer header. This change makes IPcomp tunnel mode connections incompatible with previous releases. Disable compression on such tunnels.</para> </sect3> -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=619193 http://bugzilla.novell.com/show_bug.cgi?id=619193#c3 --- Comment #3 from Marius Tomaschewski 2010-07-02 13:21:27 UTC --- A "!" at the end forces to use the algorithms (not only propose them): <screen>esp=aes128-sha256_96!</screen> -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=619193 http://bugzilla.novell.com/show_bug.cgi?id=619193#c5 Marius Tomaschewski changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jbohac@novell.com --- Comment #5 from Marius Tomaschewski 2010-07-02 13:52:37 UTC --- The situation is confusing me a bit, so let's document it more exactly here :-) When I compare the kernel and the strongswan sources between 11.2 and 11.3, the changes are as follows: The kernel has new hmac(sha384) and hmac(sha512) algorithms in the xfrm interface (net/xfrm/xfrm_algo.c), but it is still using the non-standard 96 bit truncation for sha256 by default: [...] (SADB_X_AALG_SHA2_256HMAC) { .name = "hmac(sha256)", .compat = "sha256", .uinfo = { .auth = { .icv_truncbits = 96, .icv_fullbits = 256, } }, [...] (SADB_X_AALG_SHA2_384HMAC) { + .name = "hmac(sha384)", + + .uinfo = { + .auth = { + .icv_truncbits = 192, + .icv_fullbits = 384, + } + }, [...] (SADB_X_AALG_SHA2_512HMAC) +{ + .name = "hmac(sha512)", + + .uinfo = { + .auth = { + .icv_truncbits = 256, + .icv_fullbits = 512, + } + }, [...] There is a xfrm API extension (XFRMA_ALG_AUTH_TRUNC) allowing to set the truncation. Strongswan sets it for SADB_X_AALG_SHA2_256HMAC and registers SADB_X_AALG_SHA2_256_96HMAC additionally then. This means, we have to write: "[...] If you depend on the 96-bit truncation scheme, use the new "sha256_96" keyword—this might be necessary, if you want to establish a connection with an old kernel (openSUSE 11.2 or earlier) or an linux based ipsec implementation that does not set the standard truncation. [...]" The relevant kernel changes are: - xfrm: Fix truncation length of authentication algorithms installed via PF_KEY http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.33.y.git;a=commit... - xfrm: Add SHA384 and SHA512 HMAC authentication algorithms to XFRM http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.33.y.git;a=commit... - xfrm: Use the user specified truncation length in ESP and AH http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.33.y.git;a=commit... - xfrm: Store aalg in xfrm_state with a user specified truncation length http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.33.y.git;a=commit... - xfrm: Define new XFRM netlink auth attribute with specified truncation bits http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.33.y.git;a=commit... -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=619193 http://bugzilla.novell.com/show_bug.cgi?id=619193#c7 --- Comment #7 from Marius Tomaschewski 2010-07-02 14:05:58 UTC --- BTW: http://www.faqs.org/rfcs/rfc4868.html [...] 2.3. Truncation The HMAC-SHA-256+ algorithms each produce an nnn-bit value, where nnn corresponds to the output bit length of the algorithm, e.g., HMAC- SHA-nnn. For use as an authenticator, this nnn-bit value can be truncated as described in [HMAC]. When used as a data origin authentication and integrity verification algorithm in ESP, AH, IKE, or IKEv2, a truncated value using the first nnn/2 bits -- exactly half the algorithm output size -- MUST be supported. No other authenticator value lengths are supported by this specification. [...] -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=619193 http://bugzilla.novell.com/show_bug.cgi?id=619193#c9 --- Comment #9 from Karl Eichwalder 2010-07-08 12:02:13 UTC --- (In reply to comment #2) ..

But better you remove this statement again:

...

<para>Or on the peer using the old kernel, when it supports it:</para>

<screen>"esp=aes128-sha256_128"</screen>

It worked for me yesterday on SLE-11-SP1, but maybe I've made some mistake in my setup... I can't find any explicit _128 in the proposal_keywords table. ..

Ok, this is my version now: <!-- bnc#619193 --> <sect3 id="sec.113.ipsec"> <title>Incompatible IPsec and strongSwan Changes</title> <para>The "sha256"/"sha2_256" keywords now configure the kernel with 128-bit truncation, not the non-standard 96-bit truncation used by previous releases. If you depend on the 96-bit truncation scheme, use the new "sha256_96" keyword—this might be necessary, if you want to establish a connection with an old kernel (openSUSE 11.2 or earlier).</para> <para>In those case modify the connection settings to the old and non-standard 96-bit truncation in the <filename>ipsec.conf</filename> of the new system:</para> <screen>esp=aes128-sha256_96</screen> <para>There is also an incompatible strongSwan change. IPComp in tunnel mode was fixed to strip out the duplicated outer header. This change makes IPComp tunnel mode connections incompatible with previous releases. Disable compression on such tunnels.</para> </sect3> -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=619193 http://bugzilla.novell.com/show_bug.cgi?id=619193#c10 --- Comment #10 from Marius Tomaschewski 2010-07-20 06:59:13 UTC --- (In reply to comment #9)

Ok, this is my version now:

IMO it looks good. Thanks! -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=619193 http://bugzilla.novell.com/show_bug.cgi?id=619193#c12 Swamp Workflow Management changed: What |Removed |Added ---------------------------------------------------------------------------- Status Whiteboard| |maint:released:11.3:34647 --- Comment #12 from Swamp Workflow Management 2010-07-26 14:47:32 UTC --- Update released for: release-notes-openSUSE Products: openSUSE 11.3 (i586) -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.