Richard Hughes created an issue: https://gitlab.com/gnutls/gnutls/-/issues/1719
Hi all,
I'm building PQC into libjcat, using GnuTLS. I'm loading a ML-DSA87 PKCS#7
certificate (and private key) and then signing some data using
`gnutls_pkcs7_sign()`. This works wonderfully, until I add
`GNUTLS_PKCS7_INCLUDE_TIME` which causes the sign process to abort with
`GNUTLS_E_INVALID_REQUEST`.
The backtrace explains a little what's happening:
#1 0x00007ffff784c0ad in gnutls_hash_fast (algorithm=GNUTLS_DIG_UNKNOWN,
ptext=0x41fe68 <payload_str>, ptext_len=14, digest=digest@entry=0x7fffffffcdc0)
at ../../lib/crypto-api.c:968
#2 0x00007ffff78eb8b9 in write_attributes (root=0x7ffff79ba1c3
"signerInfos.?LAST.signedAttrs", c2=0x461740, data=0x7fffffffd010,
me=0x7ffff7a74eb0 <hash_algorithms+1456>, other_attrs=<optimized out>, flags=2)
at ../../../lib/x509/pkcs7.c:2273
#3 gnutls_pkcs7_sign (pkcs7=0x452d50, signer=0x45dfe0,
signer_key=0x458ed0, data=0x7fffffffd010, signed_attrs=<optimized out>,
unsigned_attrs=<optimized out>, dig=GNUTLS_DIG_SHAKE_256, flags=2) at
../../../lib/x509/pkcs7.c:2452
So, we're calling `gnutls_pkcs7_sign(dig:
gnutls_digest_algorithm_t=GNUTLS_DIG_SHAKE_256)`, `write_attributes(me:
mac_entry_st)` which when calls `gnutls_hash_fast(MAC_TO_DIG(me->id))` --
`MAC_TO_DIG()` being an alias to `_gnutls_mac_to_dig()`.
Looking at the source code for that, I see:
inline static gnutls_digest_algorithm_t
_gnutls_mac_to_dig(gnutls_mac_algorithm_t mac)
{
if (unlikely(mac >= GNUTLS_MAC_AEAD))
return GNUTLS_DIG_UNKNOWN;
return (gnutls_digest_algorithm_t)mac;
}
...and sure enough, `GNUTLS_MAC_SHAKE_256` is indeed larger than
`GNUTLS_MAC_AEAD` as defined in `gnutls_mac_algorithm_t` from
`lib/includes/gnutls/gnutls.h.in`
I'm happy to write a merge request to fix this, but I'm not sure what the
correct fix would be. Moving `GNUTLS_MAC_SHAKE_256` to be a lower number than
`GNUTLS_MAC_AEAD` would work -- i.e. just after `GNUTLS_MAC_STREEBOG_512` --
but that's probably an API break.
I personally think adding the two constants to `_gnutls_mac_to_dig()` would be
best. Thanks!
--
Reply to this email directly or view it on GitLab:
https://gitlab.com/gnutls/gnutls/-/issues/1719
You're receiving this email because of your account on gitlab.com.
_______________________________________________
Gnutls-devel mailing list
Gnutls-devel@lists.gnutls.org
http://lists.gnupg.org/mailman/listinfo/gnutls-devel
