Re: [PATCH v14 4/5] pkcs7, x509: Add ML-DSA support

["Christophe Leroy (CS GROUP)" <chleroy@kernel.org>]

Le 25/01/2026 à 15:42, Jarkko Sakkinen a écrit :
> On Wed, Jan 21, 2026 at 10:36:06PM +0000, David Howells wrote:
>> Add support for ML-DSA keys and signatures to the CMS/PKCS#7 and X.509
>> implementations.  ML-DSA-44, -65 and -87 are all supported.  For X.509
>> certificates, the TBSCertificate is required to be signed directly; for CMS,
>> direct signing of the data is preferred, though use of SHA512 (and only that)
>> as an intermediate hash of the content is permitted with signedAttrs.
>>
>> Signed-off-by: David Howells <dhowells@redhat.com>
>> cc: Lukas Wunner <lukas@wunner.de>
>> cc: Ignat Korchagin <ignat@cloudflare.com>
>> cc: Stephan Mueller <smueller@chronox.de>
>> cc: Eric Biggers <ebiggers@kernel.org>
>> cc: Herbert Xu <herbert@gondor.apana.org.au>
>> cc: keyrings@vger.kernel.org
>> cc: linux-crypto@vger.kernel.org
>> ---
>>   crypto/asymmetric_keys/pkcs7_parser.c     | 24 +++++++++++++++++++-
>>   crypto/asymmetric_keys/public_key.c       | 10 +++++++++
>>   crypto/asymmetric_keys/x509_cert_parser.c | 27 ++++++++++++++++++++++-
>>   include/linux/oid_registry.h              |  5 +++++
>>   4 files changed, 64 insertions(+), 2 deletions(-)
>>
>> diff --git a/crypto/asymmetric_keys/pkcs7_parser.c b/crypto/asymmetric_keys/pkcs7_parser.c
>> index 3cdbab3b9f50..594a8f1d9dfb 100644
>> --- a/crypto/asymmetric_keys/pkcs7_parser.c
>> +++ b/crypto/asymmetric_keys/pkcs7_parser.c
>> @@ -95,11 +95,18 @@ static int pkcs7_check_authattrs(struct pkcs7_message *msg)
>>   	if (sinfo->authattrs) {
>>   		want = true;
>>   		msg->have_authattrs = true;
>> +	} else if (sinfo->sig->algo_takes_data) {
>> +		sinfo->sig->hash_algo = "none";
>>   	}
>>   
>> -	for (sinfo = sinfo->next; sinfo; sinfo = sinfo->next)
>> +	for (sinfo = sinfo->next; sinfo; sinfo = sinfo->next) {
>>   		if (!!sinfo->authattrs != want)
>>   			goto inconsistent;
>> +
>> +		if (!sinfo->authattrs &&
>> +		    sinfo->sig->algo_takes_data)
>> +			sinfo->sig->hash_algo = "none";
> 
> Why don't we have a constant for "none"?
> 
> $ git grep "\"none\"" security/
> security/apparmor/audit.c:      "none",
> security/apparmor/lib.c:        { "none", DEBUG_NONE },
> security/security.c:    [LOCKDOWN_NONE] = "none",
> 
> $ git grep "\"none\"" crypto
> crypto/asymmetric_keys/public_key.c:                                    hash_algo = "none";
> crypto/asymmetric_keys/public_key.c:                            hash_algo = "none";
> crypto/testmgr.h: * PKCS#1 RSA test vectors for hash algorithm "none"
> 
> IMHO, this a bad practice.

What is a bad practice ?

$ git grep "\"sha256\"" security
security/apparmor/apparmorfs.c:         dent = 
aafs_create_file("sha256", S_IFREG | 0444, dir,
security/apparmor/apparmorfs.c: return rawdata_get_link_base(dentry, 
inode, done, "sha256");
security/apparmor/apparmorfs.c:         dent = create_profile_file(dir, 
"sha256", profile,
security/integrity/ima/Kconfig: default "sha256" if IMA_DEFAULT_HASH_SHA256
security/ipe/audit.c:#define IPE_AUDIT_HASH_ALG "sha256" /* keep in sync 
with audit_policy() */

$ git grep "\"sha256\"" crypto
crypto/asymmetric_keys/mscode_parser.c:         ctx->digest_algo = "sha256";
crypto/asymmetric_keys/pkcs7_parser.c: 
ctx->sinfo->sig->hash_algo = "sha256";
crypto/asymmetric_keys/public_key.c:                strcmp(hash_algo, 
"sha256") != 0 &&
crypto/asymmetric_keys/x509_cert_parser.c: 
ctx->cert->sig->hash_algo = "sha256";
crypto/asymmetric_keys/x509_cert_parser.c: 
ctx->cert->sig->hash_algo = "sha256";
crypto/drbg.c:          .cra_name = "sha256",
crypto/drbg.c:          .backend_cra_name = "sha256",
crypto/essiv.c: /* Synchronous hash, e.g., "sha256" */
crypto/krb5/rfc8009_aes2.c:     .hash_name      = "sha256",
crypto/sha256.c:                .base.cra_name          = "sha256",
crypto/sha256.c:MODULE_ALIAS_CRYPTO("sha256");
crypto/tcrypt.c:                ret = min(ret, tcrypt_test("sha256"));
crypto/tcrypt.c:                test_hash_speed("sha256", sec, 
generic_hash_speed_template);
crypto/tcrypt.c:                test_ahash_speed("sha256", sec, 
generic_hash_speed_template);
crypto/testmgr.c:               .alg = "sha256",

How is the handling of "none" different from other hash algorithms ?

Christophe