[Swan-dev] interop problem between pluto and strongswan with AES128-AES_XCBC-MODP2048

[Paul Wouters]

On Tue, 27 Mar 2018, Andrew Cagney wrote:

https://tools.ietf.org/html/rfc7296#section-2.14

    Ni and Nr are the nonces, stripped of any headers.  For
    historical backward-compatibility reasons, there are two PRFs that
    are treated specially in this calculation.  If the negotiated PRF is
    AES-XCBC-PRF-128 [AESXCBCPRF128] or AES-CMAC-PRF-128 [AESCMACPRF128],
    only the first 64 bits of Ni and the first 64 bits of Nr are used in
    calculating SKEYSEED, but all the bits are used for input to the prf+
    function.

> Date: Tue, 27 Mar 2018 00:17:54
> From: Andrew Cagney <andrew.cagney at gmail.com>
> To: Libreswan Development List <swan-dev at lists.libreswan.org>
> Subject: [Swan-dev] interop problem between pluto and strongswan with
>     AES128-AES_XCBC-MODP2048
> 
> Strongswan receives Ni:
>
> Mar 26 23:40:04 14[ENC] parsing NONCE payload, 100 bytes left
> ...
> Mar 26 23:40:04 14[ENC]   parsing rule 10 CHUNK_DATA
> Mar 26 23:40:04 14[ENC]    => 32 bytes @ 0x7fc800002330
> Mar 26 23:40:04 14[ENC]    0: 69 75 6C C9 B5 F5 68 1C 4D AC 18 68 D7
> 8D 75 61  iul...h.M..h..ua
> Mar 26 23:40:04 14[ENC]   16: 78 3F FB E6 B5 A4 BE F0 8A A3 2A AE 7C
> 5D 49 5D  x?........*.|]I]
> Mar 26 23:40:04 14[ENC] parsing NONCE payload finished
>
> Strongswan emits Nr:
>
> Mar 26 23:40:04 14[ENC] generating payload of type NONCE
> ///
> Mar 26 23:40:04 14[ENC]   generating rule 10 CHUNK_DATA
> Mar 26 23:40:04 14[ENC]    => 32 bytes @ 0x7fc800006520
> Mar 26 23:40:04 14[ENC]    0: 8D FC D7 85 AF 71 3C 46 1A CB CE D5 1A
> A1 2C 0D  .....q<F......,.
> Mar 26 23:40:04 14[ENC]   16: E6 AA 2A 14 2E 2E 19 3A 84 02 5C 0C C2
> 13 21 42  ..*....:..\...!B
> Mar 26 23:40:04 14[ENC] generating NONCE payload finished
>
> Pluto forms that into Ni|Nr:
>
> | ike sa SKEYSEED PRF aes_xcbc init Ni|Nr-chunk at 0x7fa9be9defb8 (length 64)
> |     ephemeral-key at 0x7fa9c00daf80, size: 16 bytes, type/mechanism: AES_KEY_GEN
> ...
> |     data-bytes at 0x7fa9be9defb8 (64 bytes)
> |       69 75 6c c9  b5 f5 68 1c  4d ac 18 68  d7 8d 75 61
> |       78 3f fb e6  b5 a4 be f0  8a a3 2a ae  7c 5d 49 5d
> |       8d fc d7 85  af 71 3c 46  1a cb ce d5  1a a1 2c 0d
> |       e6 aa 2a 14  2e 2e 19 3a  84 02 5c 0c  c2 13 21 42
>
> And feeds that into PRF(Ni|Nr, G^IR), since the key is too big it
> rehashes using K=0 before hashing the shared secret:
>
> | XCBC: Key 64<16 too big, rehashing to size
> ...
> | XCBC: data  69 75 6c c9  b5 f5 68 1c  4d ac 18 68  d7 8d 75 61
> | XCBC: data  78 3f fb e6  b5 a4 be f0  8a a3 2a ae  7c 5d 49 5d
> | XCBC: data  8d fc d7 85  af 71 3c 46  1a cb ce d5  1a a1 2c 0d
> | XCBC: data  e6 aa 2a 14  2e 2e 19 3a  84 02 5c 0c  c2 13 21 42
> ...
> | XCBC: K:  00 00 00 00  00 00 00 00  00 00 00 00  00 00 00 00
> | XCBC: K1  e1 4d 5d 0e  e2 77 15 df  08 b4 15 2b  a2 3d a8 e0
> ....
>
> On strongswan (strongswan-5.6.0?) if I'm reading the output from the
> attached patch correctly, seems to use the first 8 bytes of NI and Nr:
>
> Mar 26 23:40:04 14[LIB] XCBC: Setting key => 16 bytes @ 0x7fc800003450
> Mar 26 23:40:04 14[LIB]    0: 69 75 6C C9 B5 F5 68 1C 8D FC D7 85 AF
> 71 3C 46  iul...h......q<F
>
> Above is 69 75 6c c9  b5 f5 68 1c from Ni and 8d fc d7 85  af 71 3c 46
> from Nr leading to ...
>
> Mar 26 23:40:04 14[LIB] XCBC: Key is just right
> Mar 26 23:40:04 14[LIB] XCBC: K => 16 bytes @ 0x7fc800003450
> Mar 26 23:40:04 14[LIB]    0: 69 75 6C C9 B5 F5 68 1C 8D FC D7 85 AF
> 71 3C 46  iul...h......q<F
> Mar 26 23:40:04 14[LIB] XCBC: k1 => 16 bytes @ 0x7fc8238a0a90
> Mar 26 23:40:04 14[LIB]    0: 97 4E 06 B2 C1 CB 22 82 F6 95 69 EA 98
> B3 94 00  .N...."...i.....
> Mar 26 23:40:04 14[LIB] XCBC: k2 => 16 bytes @ 0x7fc800002810
> Mar 26 23:40:04 14[LIB]    0: 1A 56 1E 60 62 88 79 A3 67 C7 F8 F6 80
> F0 E3 04  .V.`b.y.g.......
> Mar 26 23:40:04 14[LIB] XCBC: k3 => 16 bytes @ 0x7fc800003270
> Mar 26 23:40:04 14[LIB]    0: F3 0E C0 35 BB 01 1D FC 3B A9 64 78 35
> 71 71 ED  ...5....;.dx5qq.
> Mar 26 23:40:04 14[LIB] XCBC: update => 256 bytes @ 0x7fc8000048d0
> Mar 26 23:40:04 14[LIB]    0: 6F 77 96 6A F6 A5 4D 06 9F 25 E2 96 D4
> C2 D6 E3  ow.j..M..%......
> (that's the start of the shared secret)
>
> stumped
>