Fix regression in key usage purpose encoding

[djc]

Key usage purpose encoding regressed in #287.

I don't (yet) really know what I'm doing here, just implementing @ctz's suggestion from the issue to keep this moving.

Anyone want to suggest some tests that would prevent this from regressing again? There are some tests in https://github.com/rustls/rcgen/pull/287/files#diff-1340717e703c8244b2ad18d09955bd825f1565daee83f900a26a5c956b57c125R380 but we should have something that catches this.

Fixes #368.

[ctz]

For a test, how about this?

diff --git a/rcgen/src/certificate.rs b/rcgen/src/certificate.rs
index 8f625b8..9438018 100644
--- a/rcgen/src/certificate.rs
+++ b/rcgen/src/certificate.rs
@@ -1146,6 +1150,10 @@ mod tests {

                for ext in cert.extensions() {
                        if key_usage_oid_str == ext.oid.to_id_string() {
+                               // should have the minimal number of octets, and no extra trailing zero bytes
+                               // ref. https://github.com/rustls/rcgen/issues/368
+                               assert_eq!(ext.value, vec![0x03, 0x02, 0x05, 0xe0]);
+
                                if let x509_parser::extensions::ParsedExtension::KeyUsage(usage) =
                                        ext.parsed_extension()

[djc]

Thanks! I made you the commit author. 👍

[djc]

• Published rcgen v0.14.3 at registry crates-io
• [new tag] v0.14.3 -> v0.14.3
• Release notes