fixed K_AMF derivation

[koh-osug]

The K_AMF derivation for 5G is incorrect. According to ETSI TS 133 501 A.7 KAMF derivation function uses K_SEAF as input key.

[CLAassistant]

All committers have signed the CLA.

[koh-osug]

will this be added now after 3 years?

[andrepuschmann]

Hey @koh-osug - apologies for the delay. As you might have guessed there is very little activity on the 5G UE side in this repo. I'll cherry-pick the PR locally and run through our (private) CI. If all is good it'll be merged and pushed upstream with the next official push.

[andrepuschmann]

@koh-osug Just seeing this is the patch for the PCSC USIM code. Have you tried this with a USIM against a commercial 5GC and can confirm it's working?

[koh-osug]

I have used a smartcard from sysmocom in a lab set up with commercial gNBs. But the specification part seems to be quite clear, and it was not working otherwise:

A.7.0 Parameters for the input S to the KDF
When deriving a KAMF from KSEAF the following parameters shall be used to form the input S to the KDF.

• FC = 0x6D
• P0 = SUPI
• L0 = P0 length - number of octets in P0
• P1 = ABBA parameter
• L1 = P1 length - number of octets in P1
  The input key KEY shall be the 256-bit KSEAF.

[andrepuschmann]

Cherry-picked and put through internal CI. Thanks again