Changes between 1.1.1 and 3.0.0 [xx XXX xxxx]
+ *) X509 certificates signed using SHA1 are no longer allowed at security
+ level 1 and above.
+ In TLS/SSL the default security level is 1. It can be set either
+ using the cipher string with @SECLEVEL, or calling
+ SSL_CTX_set_security_level(). If the leaf certificate is signed with SHA-1,
+ a call to SSL_CTX_use_certificate() will fail if the security level is not
+ lowered first.
+ Outside TLS/SSL, the default security level is -1 (effectively 0). It can
+ be set using X509_VERIFY_PARAM_set_auth_level() or using the -auth_level
+ options of the apps.
+ [Kurt Roeckx]
+
*) Reworked the treatment of EC EVP_PKEYs with the SM2 curve to
automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC.
This means that applications don't have to look at the curve NID and
Major changes between OpenSSL 1.1.1 and OpenSSL 3.0.0 [under development]
+ o X509 certificates signed using SHA1 are no longer allowed at security
+ level 1 or higher. The default security level for TLS is 1, so
+ certificates signed using SHA1 are by default no longer trusted to
+ authenticate servers or clients.
o enable-crypto-mdebug and enable-crypto-mdebug-backtrace were mostly
disabled; the project uses address sanitize/leak-detect instead.
o Added OSSL_SERIALIZER, a generic serializer API.
uint32_t flags;
const EVP_MD *mgf1md = NULL, *md = NULL;
RSA_PSS_PARAMS *pss;
+ int secbits;
/* Sanity check: make sure it is PSS */
if (OBJ_obj2nid(sigalg->algorithm) != EVP_PKEY_RSA_PSS)
else
flags = 0;
/* Note: security bits half number of digest bits */
- X509_SIG_INFO_set(siginf, mdnid, EVP_PKEY_RSA_PSS, EVP_MD_size(md) * 4,
+ secbits = EVP_MD_size(md) * 4;
+ /*
+ * SHA1 and MD5 are known to be broken. Reduce security bits so that
+ * they're no longer accepted at security level 1. The real values don't
+ * really matter as long as they're lower than 80, which is our security
+ * level 1.
+ * https://eprint.iacr.org/2020/014 puts a chosen-prefix attack for SHA1 at
+ * 2^63.4
+ * https://documents.epfl.ch/users/l/le/lenstra/public/papers/lat.pdf
+ * puts a chosen-prefix attack for MD5 at 2^39.
+ */
+ if (mdnid == NID_sha1)
+ secbits = 64;
+ else if (mdnid == NID_md5_sha1)
+ secbits = 68;
+ else if (mdnid == NID_md5)
+ secbits = 39;
+ X509_SIG_INFO_set(siginf, mdnid, EVP_PKEY_RSA_PSS, secbits,
flags);
rv = 1;
err:
return;
/* Security bits: half number of bits in digest */
siginf->secbits = EVP_MD_size(md) * 4;
+ /*
+ * SHA1 and MD5 are known to be broken. Reduce security bits so that
+ * they're no longer accepted at security level 1. The real values don't
+ * really matter as long as they're lower than 80, which is our security
+ * level 1.
+ * https://eprint.iacr.org/2020/014 puts a chosen-prefix attack for SHA1 at
+ * 2^63.4
+ * https://documents.epfl.ch/users/l/le/lenstra/public/papers/lat.pdf
+ * puts a chosen-prefix attack for MD5 at 2^39.
+ */
+ if (mdnid == NID_sha1)
+ siginf->secbits = 63;
+ else if (mdnid == NID_md5)
+ siginf->secbits = 39;
switch (mdnid) {
case NID_sha1:
case NID_sha256:
ok(!verify("badalt10-cert", "sslserver", ["root-cert"], ["ncca1-cert", "ncca3-cert"], ),
"Name constraints nested DNS name excluded");
-ok(verify("ee-pss-sha1-cert", "sslserver", ["root-cert"], ["ca-cert"], ),
- "Certificate PSS signature using SHA1");
+ok(verify("ee-pss-sha1-cert", "sslserver", ["root-cert"], ["ca-cert"], "-auth_level", "0"),
+ "Accept PSS signature using SHA1 at auth level 0");
ok(verify("ee-pss-sha256-cert", "sslserver", ["root-cert"], ["ca-cert"], ),
"CA with PSS signature using SHA256");
-ok(!verify("ee-pss-sha1-cert", "sslserver", ["root-cert"], ["ca-cert"], "-auth_level", "2"),
- "Reject PSS signature using SHA1 and auth level 2");
+ok(!verify("ee-pss-sha1-cert", "sslserver", ["root-cert"], ["ca-cert"], "-auth_level", "1"),
+ "Reject PSS signature using SHA1 and auth level 1");
ok(verify("ee-pss-sha256-cert", "sslserver", ["root-cert"], ["ca-cert"], "-auth_level", "2"),
"PSS signature using SHA256 and auth level 2");