100 \s+2Interim Report on Operating Systems Research in the Berkeley UNIX Project\s-2 April 1, 1990 - June 30, 1990 Susan L. Graham Domenico Ferrari Marshall Kirk McKusick Michael J. Karels Keith Sklower

0 .pp Berkeley continued to participate in the POSIX Standardization effort that most recently involved attending the April POSIX meeting. At this meeting we made a submission to ANSI concerning the IS-IS routing protocol. The submission was a suggestion concerning a way to compress lists of host identifiers on Local Area Networks, thus making it more practical to run the protocol over moderate capacity lines. The suggestion will be forwarded to other ISO member bodies as an ``expert contribution''. .pp We concluded our initial development of X.25 support for CLNP and CONS. The implementation was done so that it can gracefully co-exist with IP over public data networks and the DDN, and should work over a variety of drivers, (including front end processors). We added support for the UBC socket interface to X.25 (for which there is already an ISODE driver for TP0). .pp We continued to test our recent revisions to the routing layer that permit link-level access for ISO IS-IS, and included revised code that does CLNP over IP encapsulation. We have successfully exchanged IP-encoded CLNP echo requests with the NSFnet Backbone, that were then routed to various nodes within the NSFnet Backbone. We have revised the ESIS code to make use of the kernel routing data base, instead of its own private cache, in part to ease the construction of a user-level IS-IS daemon. .pp We tested the ESIS code to make sure that it correctly generated and processed redirects. Some last minute changes included further enhancements to ESIS done in collaboration with the University of Wisconsin (required for construction of an IS-IS implementation). The changes provided a means of intercepting all transport connection requests destined for a given NSAP, to make it possible to construct a transport service bridge. During the process of installing the transport intercept, another couple of the COS failures were fixed. .pp We received a list of conformance test failures for the ISO TP4 implementation and corrected many of them. .pp We completed the first phase of our structural reorganization of the kernel that reduces the number of global variables (important for security considerations). We formalized several major interfaces to improve modularity. .pp The end of the quarter was spent gathering together a formal test release of Berkeley UNIX which we called 4.3BSD-Reno. This work included finishing up and documenting the projects that we have done and putting together a coherent test release. .pp There were some last minute fixes to the 4.3BSD-Reno release, contributed by the Ultrix engineering group after interoperability testing with their DECNET phase V routers, and FTAM implementation. .pp The 6.6 version of ISODE was also compiled almost without incident. We received the very latest version of ISODE, and made sure that it compiled and ran under 4.3BSD-Reno. Regrettably, this took a weeks time, and consequently had to be shipped separately. .pp The project has managed to secure some non-DARPA continuation funding. However, this funding is significantly below our previous funding and as a consequence our group will probably have to be sized down. In particular, our current funding groups are not very interested in ISO development. So, although we have managed to continue our ISO work in the short term, it is not clear how long this effort can be continued. If NIST is still interested in further ISO work, then they should take action to see that it gets done. We are interested in continuing the ISO work; should this be impossible, we recommend that further ISO work start with the 4.3BSD-Reno and ISODE version 6.6 distributions.