It depends on what exactly you’re assuming. When both are injected as separate services hiding all logging and caching details from repo it’s OK. If they are part of repo implementation then it’s not OK.

I don’t see the problem. Why can’t we have logging and caching below the repo layer?

Don’t you think using such segregation you’re not correctly describe the purpose of the abstraction component? It’s too broad to say just “abstract” because it may be done by means of different approaches with different purposes. For example, a service class can abstract the way how data is construct to be used inside of the app layer, repository can abstract how data is retrieved, joined, access object can abstract the tech details of data retrieval. Although, for a small project where you have only web site and DB these segregation might be redundant.