As reported in PR253081, when a library is opened via fdlopen, it has a null pointer for its path and so _rtld_bind can crash as a result of passing the null pointer to basename (which passes it to strrchr, which doesn't do a null check).
I've been carrying a local patch for this for a while. It affects capsicum programs, possibly only ones that open libraries indirectly via LD_LIBRARY_PATH_FDS. In my case, I am opening a library via fdlopen, which depends on some libraries that are opened via directory descriptors inherited from the parent and passed in with LD_LIBRARY_PATH_FDS. It may not happen with just the single fdlopen call (it appears that it's the basename(defobject->path) call that's the problem, not the basename(obj->path) It may be that the correct fix is to initialise the path field, but nothing apart from this debug line appears to worry if it is null.