@idryomov I started looking at this but I can't find any tests that seem to exercise passing >1 spec to the *_load2 function. I did write a test that functionally replaces the _load version with the _load2 version with a single element array. In particular I'm not sure how to create a clone that has a different encryption from the parent. Any pointers? (C/C++ or python code examples are fine)

I started looking at this but I can't find any tests that seem to exercise passing >1 spec to the *_load2 function.

Search for encryption_load2 in src/test/librbd/test_librbd.cc.

I did write a test that functionally replaces the _load version with the _load2 version with a single element array.

If go-ceph continues to expose EncryptionLoad, I'd prefer for that to still map to rbd_encryption_load().

In particular I'm not sure how to create a clone that has a different encryption from the parent.

Simply clone as usual and call EncryptionFormat on the clone right after opening it. There are examples in src/test/librbd/test_librbd.cc, typically a few lines before the call to rbd_encryption_load2() (C API) or Image::encryption_load2() (C++ API).

I started looking at this but I can't find any tests that seem to exercise passing >1 spec to the *_load2 function.

Search for encryption_load2 in src/test/librbd/test_librbd.cc.

Thanks. Don't know how I missed that the first time. It certainly helped.

I did write a test that functionally replaces the _load version with the _load2 version with a single element array.

If go-ceph continues to expose EncryptionLoad, I'd prefer for that to still map to rbd_encryption_load().

There will be no change to EncryptionLoad. The new EncryptionLoad2 needed a test and that test was basically a copy of the test for EncryptionLoad but with the calls to EncryptionLoad swapped with EncryptionLoad2. I felt this was a borderline insufficient.

The PR #1061 has been updated with a new test now and that should have better coverage of the EncyptionLoad2 functionality. It is ready for review.