According to RFC5280, a subject/issuer Name is a sequence of RelativeDistinguishedName, which in turn in a set of AttributeTypeAndValue (see https://tools.ietf.org/html/rfc5280#section-18.104.22.168).
There is no doubt that (RFC2253 notation) these Names are valid:
Single value RDNs, repeating Type OID: CN=A,CN=B
Multivalue RDNs with separate type OIDs: CN=A+OU=B
But is my interpretation correct, that (RFC2253 notation) this Names would also be valid:
Multivalue RDNs with repeated type OID: CN=A+CN=B
But this one would not be:
Multivalue RDNs with repeated (type OID and value): CN=A+CN=A
My initial assumtion would have been that within any RDN every type OID may occur at most once (giving it a dictionary syntax), and I remember having that read somewhere, but am unable to find the reference again. From the pure syntax, it appears that "CN=A+CN=B" seems permissible.