IssuingDistributionPoint has the wrong AssociatedOid (id-pe-subjectInfoAccess)

[eyusd]

x509-cert/src/ext/pkix/crl/dp.rs:

impl AssociatedOid for IssuingDistributionPoint {
    const OID: ObjectIdentifier = ID_PE_SUBJECT_INFO_ACCESS;   // 1.3.6.1.5.5.7.1.11
}

issuingDistributionPoint is 2.5.29.28 (RFC 5280 §5.2.5). 1.3.6.1.5.5.7.1.11 is id-pe-subjectInfoAccess, which ext/pkix/access.rs already binds to SubjectInfoAccessSyntax.

So a lookup by associated OID never finds an IDP. It searches for 1.3.6.1.5.5.7.1.11 while the CRL carries 2.5.29.28, and comes back with "no such extension" rather than an error. If you're reading indirect_crl to decide whether a CRL's entries apply to your own issuer's serial space, you'll treat an indirect CRL as a direct one.

impl_extension!(IssuingDistributionPoint, critical = true) uses the same constant, so the encode side writes the wrong OID too.

Present in 0.3.0 and on master at b1e1582.

Repro

use const_oid::{AssociatedOid, db::rfc5280::ID_CE_ISSUING_DISTRIBUTION_POINT};
use x509_cert::ext::pkix::IssuingDistributionPoint;

assert_eq!(IssuingDistributionPoint::OID, ID_CE_ISSUING_DISTRIBUTION_POINT);   // fails

Fix

-    const OID: ObjectIdentifier = ID_PE_SUBJECT_INFO_ACCESS;
+    const OID: ObjectIdentifier = ID_CE_ISSUING_DISTRIBUTION_POINT;

ID_CE_ISSUING_DISTRIBUTION_POINT is already in const_oid::db::rfc5280.

I have this locally with two regression tests in tests/crl.rs, one for the constant and one that puts a real IDP on the wire and finds it by associated OID. Both fail before, pass after, and the rest of the suite is unchanged (78 tests plus 10 doc-tests). The 5 failures in tests/builder.rs are zlint: command not found and happen on unmodified master too.

It's a behaviour change for anyone already working around it, so probably wants a CHANGELOG entry. Happy to send a PR.

[carl-wallace]

Your change looks right to me. I'm sure a PR would be welcomed.