This is actually quite common in some fields, for both conference and journal submission, particularly when you have coordinated submission from a number of different participants in a complex project.
The way that I have handled this when I've done so is to have the citations initially as "submitted" and either bundle the other submissions as supplementary information for reviewers (when possible), or with a note saying "available on request" when the venue doesn't support supplementary information. Another option is to make use of pre-print services like arXiv, though that can violate double-blind review.
Once the papers are accepted, you just change the camera-ready to the final information and the circular reference enters the literature without issue. It gets more awkward if one paper is accepted and the other is not, but this can be resolved with the help of a pre-print service.