THE UNIX NEWSLETTER VOLUME 2 NUMBER 7 AUGUST 1977 Page 1 NOTICE This document may contain information covered by one or more licenses, copy- rights, and non-disclosure agreements. Circulation of this document is restricted to holders of a license for the UNIX or Mini-UNIX software system from Western Electric. Such license holders may reproduce this document for uses in conformity with the Unix license. All other circulation or reproduction is prohibited. MEETINGS As was indicated at Urbana, the Stanford Research Institute is the host for a local West Coast Meeting in September. The details appear on the following pages. As you will note on reading the minutes of the Urbana meeting, the group indicated a desire to have two meet- ings per year. We are trying to arrange for meetings in January and May or June 1978. Details will be published soon. SOFTWARE DISTRIBUTION The listing of the contents of the third distribution in last month’s issue contained a large number of files in the directory "3iug". These were all of the files on line at the Urbana Users’ Group meeting. Apparently some of these files were not intended for general release. Accordingly, we have removed directory "3/ug" from the dis- tribution. Those of you who placed material on the Urbana system with the intent of releasing the software are asked to notify us and we will include it in the fourth distribution. The order form neglected to indicate how checks were to be made out if they were included. Any checks should be to the order of "Unix News" or the “Brooklyn College Assn.”. The TU16 driver from Harvard which appears in Distribution three seems to have one (and only one?) bug in it. If you have a TU16 drive, use the driver in Distribution two. The Harvard TU10 driver works correctly, at least on Digi-Data pseudo-TU10s. We will print the fix to the TU16 driver as soon as someone sends it to us. NEW TORONTO RELEASE We have been told that a new version of the Toronto Software Package was mailed to New York in mid-July. As of August 18 it has not arrived. Some tapes were shipped with the old version of the software, but as of this date we are delaying the preparation of tapes tequesting the Toronto package until the new version can. be.shipped. GUINNESS BOOK OF RECORDS A tape mailed from Portland, Oregon by first class mail took five weeks to reach New York. WORDS OF ONE SYLLABLE DEPARTMENT From the "PDP11/60 Processor Handbook", page 11-11: The design and packaging of the PDP- 11/60 has placed great emphasis on RAMP. This means reduced mean time between failures (MTBF) and reduced mean time to repair (MTTR). Address editorial material and software submission to Melvin Ferentz tla CUNY/UCC 555 West 57 Street New York, N.Y. 10019 Subscription requests, payments and address changes should be addressed to Armand Gazes Physics Department Brooklyn College Brooklyn, N.Y, 11210 Gx) STANFORO RESEARCH INSTITUTE MLNLO PARK, CALIFOANIA 94025 [4180 326 W200 July 18, 1977 Professor Melvin Ferentz Physica Department Brooklyn College of CUNY Brooklyn, New York 11210 Dear Professor Ferentz: I am pleased to announce, ae per your discussions with John Basa, that SRI International will hoat the UNIX West Coast Veers Group Meeting on Monday end Tuesday, September 12 and 13, 1977. The Conference will be held in SRI's main building (Building 1), Conference Rooma A and 8B, A map is enclosed for your information, T am alao enclosing @ liat of local, available motels. It is suggested that San Francisco International Airport would be the Most convenient airport with available limousine service and car rental facilities, A meeting agenda will follow in August. For more information, please contact our Chairman, Dr. Oliver Whicby (ext. 2791) or Hr. John Bass (ext. 3819). Sincerely, David H. Brandia Director Information Science Laboratory aie Enclosures CABLE STANAES. MENLO PARK / TWX 910-373-1246 Local Hotels/Motels Holiday Tan-Stanford 625 KE. camino ‘Real Palo Alto, California Mermaid Inn Motel 727 El Camino Real Henlo Park, California Red Cottage Motel 1704 El Camino, Real Henlo Pack, Calffornia Limousine Service Airport Limousine Service “ (415) 328-2800 (415) 323-9481 (415) 326-9010 (415) 961-8800 Single Ratea: $26.00 Single Ratea: $16,00 Single Ratee: $17.00 $10.00 (SFO-SRI) EL LAMIND HLA STAMFORD RESEARCH Ma IEXUTE San Francisco d So, . 4 D i sentra hon mi aq: | on Irs; | ES seat Da pena Inlernetiona: Aus por SS avensmoon AVENUE STANFORD RESEARCH INSTITUTE CONFERENCE PARKING SRI HEADQUARTERS MAP [Cl] omen amonas 0 termine MEMO MEUGLAL CLINIC Cee | inst stacey” eee rei aee ly STARE ORL uravemsi ty aa yea “afc 2 Sms wt eee ens be 4 i ait Rate ies ees to micoe tree vacuity ie me ing Parki Conference ht WAP OW REVERSE SIDE Gn UNIVERSITY OF MINNES@ Chronobiology Lisioratatier ‘ Department of Laboratory Medicine TWIN CITIES Peete, Medical School 380 Lyon Laboratories Minneapoils, Minnesota 55455 July 18, 1977 (612) 373-2916 or 373-2920 Profeesor Helvin Ferentz Physics Department Brooklyn College of CUNY Brooklyn, N.Y. 11210 Dear Professor Ferentz: We have just received a PDP 117/34 from DEC, We are having it installed this week and are in the process of having the UNIX license apreement with Bell modified to include us. Enclosed are copies of relevant communtcation with them should you need the information. We were not sble to obtain the serial number until now, so ve were unable to get the license modified before delivery. We should like to be added to the user's group as: Dewayne IHillman, Chranchiolory Laboratories, 380 Lyon Laboratories, University of Hinnesota, Minneapolis,!¥ 55455. Past copies of the newsletter would be appreciated as thia would alleviate our having to copy them. Please advise. Our installation will soon consist of: 1) a PDP 117/34 package having §5KB HOS parity memory, RKOS3, RKOSF, Decuriter, 2) an 8-channel DZ11 RS232 compatible multiplexor, 3) two 65KB boards of compatible memory from Plesney, 4) a Printronix printer-plotter with RS232 interface to connect directly to !MIX, $) a Tektronix 4996 plotting terminal, 6) a Chataworth Data card reader with RS232 interface for connection to the "i or dn conjunction with any S232 terminal through telephone lines, 7) an Anderson-Jacobson letter~perfect terminal for the secretary, 8) a Research, Inc. mini-flopny disk drive designed to be uaed in conjunction with a terminal, 9) a Kennedy 9900 9-track tape drive at 99 API NRZI and 1609 BPI PL on a Western Peripherals embedded controller, 10)a home made interface to two different older Varian 620A computers so that ve can, among other things, have access to older 7-track tapes, 11)a home made interface to our older Calcomp microfilm plotter, and possibly to an older 1000 cpm card reader which we hope to make look like a CRIL. ‘le are just now starting on the latter design, 12)one or more terminals to be added in the future, 13) an auto-answer modem to connect to the IX. My present concept of our current needs includes a DZ11 driver and a better Fortran as vell as any other available software concerned with the above described system, fModifications to UNIX including memory parity chacking are also of interest. Any help we can get to help us get started will be aporeciated including any special considerations for the 11/34. ‘ Sincerely, eretnaps © lla HEALTH screncepeveyme C. [il1men University of Illinois at Urbana-Champaign DEPARTMENT OF COMPUTER SCIENCE Urbana, lllineis 6180) [by 333 4478 July 26, 1977 Professor Melvin Ferentz Physics Department Brooklyn College of CUNY Brooklyn, KY 11210 Dear Mel: Just wanted to let people know that ve have created an interprocess communications facility ourselves involving non-blocking I/O and ports, We opted for the absolute minimal size increase, quite vital to ue with our 11/40. Ports are consequently quite pipe-like, although much nicer. With thie facility we have been able to create a networking demon that implements a rather ARPAiuh sort of michine-to-machine comsunications facility. We have also used ports for programe such as the line printer demon so that they may receive requests without having to muck around in a directory. Those interested in copy may write and I will pass their request on to the Center for Advanced Computation which distributes things for a nominal charge (I believe about $100). Sincerely, fe Alfred D, Whaley Senior Research Programmer ADW:baa VANDERBILT UNIVERSITY rea NASHVILLE, TENNESSEE 37235 Trieprowe (615) 322-7311 / Box 1894 Station B Elctrical and Biomedical Engineering « Direct phone 322-3521 June 22, 1977 Pref, Melvin Ferentz Shysics Dept. Srookiyn Coljege of CUNY Brooklyn, N.Y. 11215 Dear Professor Ferentz, We obtained recently the UNIX system and plan to run it on an 11/34 with an RKCS disk. As of now we have not heard of an existing handler for the RXOS and we would appreciete very A if you could heip us. Any information will be greatly anpreciated. Sincerely yours, Hach ites Baruch Hamel ESINSR July 25th, 1977. Prof. Nelvin Ferentz, Brooklyn College of CUNY, Brooklyn, N.Y. 11210 U.S.A. Dear Professor Ferentz: Earlier this month I received a phone;cali from a Stu Overland of System Industries, California. He was seiijing disk storage products. He said that he had my name on a UNIX mailing list and figured that I therefore would be in a position to buy their hardware. i asked him where he had obtained +t , and Fe said where". I told him that the list proprictery as if was pert of the Newsletter but he didn't seem to care and changed the subject. I by no means object. to the publishing of < newsletter, but I must make you aware thet t somewhere, and that at least one copy of cxifting around commercial circles be warmed about vendors soliciting £ should make a point of avoid. Yours truly, NA i Rick Macfarlane RM/s5 B-N Software Research Ine. 522 University Ave. Toronto. Ontario MSG tW7 1416] $94 -0196 Computer Grofix Lab Now York Insiltute of Technology P.O. Box 178 Oid Westbury, L.t.. NY 11568 August 8, 1977 Prof. Mel Ferentz Physics Deportmont Brookiyn College of CUNY Brooklyn, KY 11218 Deor Met: Ue hove discovered yel another bug in setexit() ond reset(). Using the most rocent version published In UNIX sous, (ne following progremme slowly grows (Is stock until fl memory faults. maingot selexitO: £8, Bd: y tla, bie resetQ: ? Here Is my entry in the 'Con you awrite a correct version of sslaxlt/resot' suoepstakes. 7 C llbrary -~ reset, sefoxit reset (val) wil) generote o “return” from the lost coi? to setexitO by restoring sp, rS and doing o return. “val’ [5 returned to the cnller of setex!t. setexit itsnif returns O. useful for going back fo the main loop efter a horrible error in o lowlevel routine, SVAN ANS NAW AS sglobi tsetexit «globl ¢reset +Qlobl csv, ret NX tsetexit: mov eS.sr5 mv sp.ssp mov (sp) spc cir re —_— eer Lota te Ge 2h ~~ emp (75)..9r5 beq i wo SY PS a bre tb \ ?% ponic -- r2-r4 lost br 2r I: mov -(r5).074 mov ~Or5),03 mov =Or5).r2 2: nov sr5.c5 mov -38p-3p na spc. (sp) rts pe -bss sr: spe: s5p: LT nope this will be {he lost episode in the continuing saga of solenitvresal. With fingers crossed, I remain Very truly yours, Tore Df Tom Dutt July 14,1977 dr Ferentz: nave never met, tut spote several tines last fall. as I explained shat time, 1am an undergraduate student at Coluabia and nave work: NIA ona nuseer of installations over ‘the Bas t three years. Last Rott.an ad an unusual opportunity to work as tallation, and asa sult were able to develop- nich extend i to allow fairly ne support without the Ue 3 omas Eg $30, and over the past 3 month X with the real-ti x we stem Junctions are teing usel on a Shin the epi als _wnich has ,purcha 2 process. inc’: A&A real-tice ntly I have recived several inguindes fr IX installations interested e work Jeff and i have don and have teen mailing out copies ntaion enclosed. Lou Hatz, howeverz felt that this documentation axe an appropriate (if somewhat long) contribution to the newsletter. er has also encour2g¢i me to send you a copy. wSletter, I of what we have e questions reaAting jietails of the Re to the degree azle to in the encisset that I could provide jetail to provi € an overail picture of wnat we nave done. in lease feel free tsa with quessi or on this satter, I gan be reached at the Hospital ate 43t Black 3 ithe computer room) 635-3501/4013/3 er gate of either or. would forgive spoile pernacs have teen rather chen fort. Real-Time changes ta UNIX Kenneth Birman Many computer aprlications: especially those invoalvind the collection and analysis of time series datar need both real-time data aquisition facilities and a hish desree of Prodramming support suited to the implementation of sorhis~ ticated analysis techniques. Traditionally these needs have been hard to reconciler since by and larse the systems which are considered to provide adaquate real-time surrort are not also able to provide the sort of non-realtime data rrocess ing support which would lend itself ta assressive analysis effarts. Over the past year the author has worked with Jeff Rottman (3 sSraduate student at Rerketew) to privately develop a set of modifications of standard Version 6 UNIX which allow the system to surrort the realtime asuisition of dats without halting or disruptins normal timeshared orera~ tions. These chandes have succeeded ta the poind that a program previously implemented under RTIL which reauires the aquisition of data st a sampling rate of at Jeast 15kit (DMA) with moderate data compressian has been success transfered to an 11/70 UNIX system at Columbia Universit The changes made to suprort this realtime activity implement previously unused hardware present im PRP 11/4%’s and 70%s+ and are not tied to any particular device or rrocessind en- vironment. A process capable of satisfying the needs of a realtime ar- plication must meet several broad criteria. It must he able to control an interrupt driven devices and must therefore be capable of running at least treo processor erioritiesr ane of which must be lover than that at which the devier inter rusts. Entry overhead must he minimalre since very hidh sam- pling rates often require the smallest rossible system over- head. Lastiys the process should be capable of recieving dats from a UNIX process and of passing erocessed data to UNIX processes or writing it directly inta a disc file. The realtime surrort we have developed under UNIX attempts to satisfy each of these questions individually. Recause our aprlications have been scientific (at Columbia we are sork— ing. on cardiovascular problems involving the detection of heatt rate and function abnormalities under severe time care straints), the current system is intended to work with an LPS-11 or rossibly an AR-11 (which lacks DMA). Roth a/d core: verter sustems have their own clocks, and the INA ability of the LPS-11 has been exploited at Columbia to achieve sam~ pling rates of up to 15kHz Creauired bu the application) with some processing of data. The main constraints wr have encountered have proved to be the machine rrocessind sreed and the limits of the available mass storade device (at Columbiay 51000 blocks on an RP-04 are dedicated to the col-- lection of data). With minimal processina of datar much higher sampling rates than those we have employed should be possible. The realtime chanses to UNIX center around the use of the pachine’s “surervisor® mode to obtain a set of general eur~ rose resisters and marring resisters which can be dedicated to a resident erocess controlled by UNIX but which has acu cess to certain device redisters. This arrroach is aPprarent— lu similar to that used by MERTy where a seecial kernal ero- gram monitors operating systems which run in surervisor space controlling processes running in user space. Althoush Lhe "surervisor’ in our system would have ta save end ren sture the floatind point redisters prior to usinsd themr we have am this way been able to obtain a set of machine redis— ters which need not be saved and restored excert because of the anternal lodic of the supervisor erocess. On the 11/70» uiused UNIBUS mar resisters are sel ue to mar into the su- bervisor area in memoryr a#llavind physical [70 urerations to be anitaated by the Surervisar usind 18 bit addresses in Which the two high order address bits (the "xmem* bits) are both set. ln order te allow the surervisur to have access Lo UNIX con- trolled activityr EMT’s have beer aim-lemerted to let the su- pervisor request a isc write and ta reavest that a lower rriortiis ‘delayed eroc sind? routine be entereds important bevuise of the high anterrupt eriortty levels common om dale wuuisition devices. AL either eriority, the surervisor is fot scheduled and cam use the CFU for a» lomy as it must. Currently there gre onlu Pive surervisar emt’s? the other three allowing the ourervisar to request that it be ter- minated, to sidnal Lhe commletion of a data transferr and to determing 1ts ehystes! address for use in BMA transfers on 45's. There 16 no intrinsic lamits howevers om the the de- uree of system sureort thet cauld be prrovided in this manne ts In handling the lower eriority erocessing: and in the imptementation of the surervisor disc write emtr it has been convierdent to make use of the PIRQ (erodrammable in terrurt reauest) device eresect on the 70 and the 45. The PIRQ ss also used ta handle sroblems resulting from the euunchronous nature of a floating rorot interrupt which must Under some circumstances be redirected at lower priority for hendlina later. Theee out of ¢ rossible total of seven PIKU's are currently in use. An ameortanl consaderatcon an desisinindg the surervisor sur~ Port was that ik he roosible to test surervisors and install new ones wrth minghal Likelihood af disrurtind normal acm tevaty on the 70» which is used erimarily for data rrocess~ ind and analysis. LL has eraved possible ta intercert most sueervisor faulls end errorar and to write @ supervisor core dume when errors accur. Modifications to the C debudder heve made the debudding of surerviser routines a fairly painless matters in marked constrest to the usual environment ain which realtime rrusrans are develored. A UNIX Process whith wishes to interact in some way with the sueervisor 16 Frovided with several artions. Firsts it can reauest that a surervisor be imstalleds which is rossible rrovaded thet no other supervisor is currently installed and sufficient memory is available. Such a supervisor is loaded at one end of memory to avoid frasmentation. It will be dormant until accessed by,a UNIX Frocess in one of two ways» both concerned with rassind information to the surervisar. Firsts 3 UNIX erocess may ‘spira* data to an installed su- rervisors resullind in a transfer of a Sindle ward to a dedicates low core area within the surervisor and the ac- tivelion af the surervisor, af a -eredefined “entry points This 15 commonly used to pass sindle parameters or ta re- auest that some action be initiated. If -larse Volumes of dala ure to be passed it is wossible for a UNIX Frocess tu do whet looks like a DIMA read or write ta the supervisors whereby one rade of supervisor memory is marred ta overlay the buffer beind read or written by the UNIX process (which as locked in core). This shared data is accessible by both Lhe surervisor and the UNIX process which sleers waiting for the transfer to comelele. In order to facilitate the effi- cient maintainance af such a raw link between the surervisor and the UNIX process, a sustem call has been added to lock a process runnin under UNIX inte core. The Surervisor con~ trols the awakening of the UNIX process after the transfer throush a ‘resduy" emt, Thus a surervisor mau be very tisht~ ly linked Lo sur-ert rrodrams om the UNIX sider even to the point of sharin’ Tarde (BK) chunks of data with it. A more conventional mens of passing data to UNIX »rograms has also been im-lenented in the form of a rind buffer which can be read under UNIX through a dummy driver and involves copying datar but does not reauire that special precautions be taken to handle the core lockind problem, This is useful when? for nemvlers a supervisor monitors a larste volume of incoming deter but eraduces a small volume of outruts The chandes to UNIX I have described in this rarer are currently renniod at Columbia University, with no Fnoun buss. Samrlind rates as hidh as S3kKHZz with data comeression have heen successfully tested, and an assembly londuase ist hes been written to simplify the coding of urervisor rrocesses in C, Althoudh loosely tied to the LPS=11, it\ would mot be estecially difficult to modify the code tu handle some other device. Current distribution rlans would invelve some sort of software adreement between ihe licensed UNIX installation to recieve the coder myself? and Jerf Kotiman. There will be a distribution charse. ALL future decision, to distritute will be made by myself and Jeff Kottman. Tueuiries should be directed to Kenneth Bir- man care of lir, Louw Katz? Heady Cancer Research Center Com- ruter Facilitus Columbia University, Collese of Physicians and Surgeons? 630 West 148th Street# New York Citur N.Y. 10042, is kes 22 Juty 1977 Professor Melvin Ferentz CUNY Computing Center 555 W. 57th Street New York, New York 10019 Dear Mel: Enclosed are the minutes of the UNIX Users Group meeting. If any NEWSLETTER subscribers would like full size copies, Interactive will be glad to provide them. Cordially, p- Vere J. Steven Zucker Diretor of Systems JSZ:jni Enclosure: "Minutes of the First National UNIX Users Group Meeting" 1926 CLOVERFIELD PLYO. SANTA £178 HOA. CA. GO.i04 TELEPHONE (213)829-7741 J) MURUTES OF THE FIEST RATIOWUAL UkIX USERS GROVP MEETING Steven Zucker Interactive Systems Coxporaction 1526 Clove eld Elwd Santa Nonica, California 90404 The First National Meeting of thw UNIX Users Group was held at the University of Illinois, Ur Campus, on Hay 19-21, 1977. Steve Noln: gren of the University's Center for Advanced Computation chaired the meeting. The enthusiasm of the more than 159 participants ‘sulted in a very of ideas. The and the informal tone of the sessicus stimulating atmosphere for the exchen, meeting was divided into eight sessions: o UNIX Site Activities Data Ease Management Systeis o UCLA Data Secure UNIX o Interprocess Communication o Graphics o Languages o Networking o ° Phototypesetting My hope js that these notes on the sessions will be useful in directing those wishing more details to people who can 5 to these whose con- provide them. I offer my apologi. tributions I have inadvertently itted and urge then to send their eontributions to this Newsletter. Many of the sessions were replete with announcements by speakers as well as members of the: audience of new and/or improved drivers for one or another device, with the TL-16 Magnetic Tape Unit receiving the most attentica. Rather than list all the drivers mentioned here, 1 voule like to suggest that a column in the UNIX NEWSLETIFK be ‘devoted to information of this kind with installatiens oF individuals willing to disseminate such code supplying information as to features and requirements. UNIX STITH ACTIVITIES This "What's happening where” session is summarized here in outline form by presenting only the speaker's mame, the site he represented, and the list ef the UNIX software developed or under development at the site, Fer further details or to determine whether and under what terms the software is available, get in touch with the speaker directly. k 1 : kee RE Stephen Tepper Information Sciences Department The Rand Corporation 1730 Main Street Santa Monica, California 90406 o NED: A two-dimensional CKT Text Editor © RITA: The Rule-Directed Interactive Transaction Agent, a production system language o MS: Message System o PORTS: A "named pipe’ facility for interprocess communication eo EC: An encryption progxyam © VIRTUAL TERNIKALS: An experimental split screen facility for making one keybuard and screen look like multiple terminals : o REMIND: Reminder and delayed (scheduled batch) execution KKK RY he Alan Stoug UCLA Security Group University of California at Los Angeles Los Angeles, California 90024 NEW (nonprivileged) MOUNT comnanc SYSUP and SYSDOWI procedures SEMAFHORES supported in the kernel o TENEX-LIKE TTY DRIVER with interrupt time editing, user setable break characters, and CRT paging oe PASCAL SUBSET to C translator o ENHANCEMENTS to the Illinois Network Control Program © SYSTEM CRASH DUMP ANALYZER ooo EREKE Greg Chesson Bell Telephone Laboratories Murray Hill, New Jersey o UNIX T-shirts available (I guess this is softwear, not software) o DRAW: An interactive graphics systems For producing wirelists, proms, and for doing semiautomatic layout kKkKKEEK 7-11 Tucker Taft Harvard Science Center Harvard University Cambridge, Massachusetts 02138 o Supports 1500 student accounts cn a 192 KW 11/70 ,with 25 1200-baud terminal o TEXEX-LIKE Try DRIVE o PROCESS GROV: eo PROCESS QUOTAS o SUBMIT CO: RD Fan DOLAYED E. o INTERACTIVE £ ICAL PACKAG Tom Ryan at Fenn State) v LOSGER FO. KERNEL PRINIFS so thst system isn’t delayed for nonfatal errors name space A TERMINAL s TXECUTION PRRs Peter Weiner ¥nteractive Systems Corporation 1526 Cleverfield Riva Santa Moaniea, Califurnia 934904 Peter announced the formation of Interactive Systens Corporaticn. Its goal is lo provide UNIX systems and services including system configuration, installation, staff training, contract programming, and UNIX-based systems. kKRAEE Heinz Lycklama Bell Telephone Laboratories 600 Mountain Avenue Murray Hill, New Jersey 07974 Heinz described the several variants of UNIX that have been or are being developed at Bell Labs. In addition to the standard UNIX system which Western Electric already licenses, there are three other systems in use at Bell Labs. ; LSI UNIX (LSX): LSX occupies 8K words of main memory leaving up to 20K words of user space for the single user that it supports. Minimum memory requirements for running LSX are 20K words of main memory, the extended instruction set and two floppy disks. LSX is written in C and will run the C compiler. It runs at most three processes and supports the notion of contiguous files but pipes are not supported. . LSX will run on the 11/10, 11/20, 11/34, or 11/40 as well as the LSI 11. ° MINI-UNIX: Mini-UNIX supports up to four users running up to 13 concurrent processes on an 11/40, 11/34, 11/20, or 11/10. It occupies 12K words of memory leaving up to 16K words for user programs. It uses no memory mapping and, therefore, provides no memory protection. It requires an RKO5S or larger disk. : MERT (Multi-Environment Real-Time System): This system runs only on an 11/70 or 11/45 as it requires the separation of kernel and supervisor spaces. MERT supports a real-time supervisor which can lock processes in memory, perform preemptive scheduling or time-out scheduling. The communication facilities (events, messages, shared memory, and process ports) were described. File system support for MERT is embodied in independent processes which communicate with other levels via messages. kkekk Ken Thompson fell Telephone Laboratories 600 Mountain Avenue Murray Hili, New Jersey 07974 An effort is under way at Bell Labs toa convert UNIX to run on the INTERDATA 8/32. The conversion is being treated primarily as a portability exercise, As part of the portability exercise a pseudo C-has been developed which enforces strict typing of variables, A significant number of file system changes arc being planned for Version 7 of UNIX. The chenges would extend the allowable number of blocks in a file system from the present 216 to ae blocks, thus, making it easier to use large disk files such as the RPO4. The i-node size will be extended from the present 16 words to 32 words which will include space for 10 direct block pointers, one indirect block pointer, one double indirect block pointer, and one triple indirect block pointer, allowing files.to be as long as 232 bytes. Users IDs will be extended to 16 bits and the STAT and the FSTAT system ealls will hide the physical addresses. A long SEEK system call will replace the present SEEK and a TELL sys- tem call (the inverse of SEEK will be added). The SWITCHES call will be thrown away and the SLEEP call will be re- placed by PAUSE and ALARM. Significant changes are also anticipated in the STTY and GTTY system calls. It is unlikely that the new system will be available before the beginning of 1978. ‘ KK KEK Mike O'Brien Departnent of Information Engineering University of Tili Chicago Circle P.O. Box 4348 Chicago, Illinois 60626 @ presented a copy of tha latest UNIX Disrriburion Center release tape and a UNIX T-shirt to Ken Thompson. There was also a ccrtificate for Thompson and Ritchie from the Users Group honoring their work on UNIX. : KKK EK 71713 UCLA'S DATA SECUAR UNIX Jerry Propet Computer Science Nepartien 3532. Beelter Hall University of Cali Los Angeles, Cali 7 Jerry Fopek cescribed work at UCLA in which a secure version of the UNIX orerecing systen is being devetoped. The systen srchitecture is based or a kernel architecture, with progiem verification methods heing szplied to that seftvere, The kernel is composed of an operating systen nucleus, smaller and simpler than the UNIX kernel, which is responsible for all operational security. It provides a "capability machine" vith a number of simple keruel calls. Each one provides a primitive operating systen function, such as process invocation, swapping, 1/0, etc. Above the kernel, runnizg in supervisor mode, is a "UNIX interface" module, that is part cf each usei's precess (2 process has two address spaces}. That module is fesponsible for providing an interfece to user code thst is idcntical to UNIX, end either perforrs the function, or prepares kernel requests te accomplish them if security relevant. The secure UNIX system Popek described is to capable of supporting leree numbers of processes, and running virtually all non-cuper-user code without any chanze. A prototype implemontaticn has been delivered to the goverument, aud they are in the precess of Ictting a contract t: build a production version of secure UNIN. Fopek also described the proj,rar: verification pro- cedures necessary te show that provection is enicrced by the system in an uncircumventable way. kkk KE “10~ R. M. Walden Western Electric P. 0. Box 20046 Greensboro, North Carolina 27420 Following the Data Secure UNIX presentation, Bob Walden announced that the Government and International Systems Division of Western Electric has established an organization to provide support for UNIX, initially to government users. The service will include con- sultation, installation and training, trouble shooting or problem solving assistance, improved documentation, and new feature development. RRR KK 7-14 Ph INTERPR Alan N Bolt Beranck & Newman 50 Moulton Street: Cambridge, Massachusetts 02138 Alan Neineth reported on a series of meetings held te discuss interprocess cownmnication in UNIX. « The immediate goals of the meetings was to standardize on one or more interprocess commumication mechanisms to be supported in UNIX systems run by the Department of. Defense. Two such mechanisms have been tentatively adopted: the port mechanism developed at The Rand Corporation .and events developed at the University cf Illinois, _ The Rand port mechanism described in Rand Keport, R-2064/2, Inter- process Communication Extensions for the UNIX Operating Syste, provides a mechanism very ruch like pipes. except that they can be named and opened by readers unrelated to the creator of the port. Ports also support message- oriented (as opposed to stream-oriented) 1/9. While ports are intended for transfers cf large amount of. data between unrelated processes, the Illinois event mechanism provides a more efficient path for small one or two word messages. Each process has associated with it one event queue. Processes, using primitives provided by the kernel, ean send “events" to other processes, read an event from its queue if one is there, or wait for an event to appear on its queuc. At present the development of a suitable signalling mechanism to augment ports and events and provide process synchronizatior is a subject for further study. =12- Following an open discussion of ports, events, and sychronization, there were two presentations of segment sharing mechanism that have been employed in UNLX systems. Neinz lycklama of Bell Telephone Laboratories cescribed the MERT interprocess communication facilities. MERT prevides messages which are very similar to the Illinois events exeept that the messages may be somewhat longer -- 10 words instead of 2. The messages are employed in MERT for communication between the file manager process and the MERT kernel. In MERT the user is given the capability of manipulating memory segments. ‘The user may have up to 32 segments -- 6 of which may be in his active address space. Following Heinz's presentation, Steve Holmgren de- scribed the Illinois segment sharing mechanism by which processes may send segments to or receive segments from other processes. KKK THIS ~J3= GRAPHICS A special meeting of those attendees interested in praphics under UNIX was held on the evening of May 19. Karl Kelly of the University of Illinois Center for Advanced Computation presided at the meeting. This was a very informal session at which each mat.ufacturer of graphics hardware took its knocks. The tain conclusion that could be drawn from this session was that there exists a very large and very active group doing graphics under UNIX on a tremendous variety of equipment. It is ‘probably safe to say that there is a UNIX driver for most of the common commercially available graphics devices. Six people volunteered to make brief presentations on work in progress at their installations, The following is a summary outline of those presentations. The often valuable and even more often humourous remarks from the flocr are omitted. RRREE Lynn Broek 3 United Statés Navy NPRDC Pt. Loma, California 92152 eo Activities - psychological experiments and computer-aided instruction - has Tektronix software running under UNIX FORTRAN - AG-II: a set of FORTRAN callable subroutines that plots arrays of data points with grid, tick marks, and floating points labels o Other Packages - TCS, CGS, IGP, PLOT/10 4s Yom Duff Computer Grafix Laboratory Rew York Institute of Technology P. O. Box 170 Qld Westbury, New York 10036 o Activities - Animation for Disney-style Films o Hardware - Six PDP/11's - Six Evans & Sutherland Frame Buffers ~ Evans & Sutherland Picture Systems 1 & 2 ~ Dicomed Film Recorder with 1500 x 1500 or 3000 x 3000 resolution with 24 color bits at each point © Techniques in Use - Interpolation on line drawing - Patch rendering (3D) “program using bicubic patches kkk KE Mike Selander Center for Advanced Computation University of Illinois Urbana, lllinois 61801 o Activities ~- Image Processing and Robotics o Hardware - An 11/40 with a one-of-a-kind display called the "Elephant," a combination raster scan, lLine- drawing, and character device with 4000 x 4060 resolution and 256 gray levels Mike described the baroque software system which uses the core image of a separate process as the refresh buffer for the Elephant. Bill Reeves Dynamics Graphics Froiect University of Teronte Toronto, Canada o Activities animation: educational movies, e.g., movies illustrating sorting techniques and planetary motion circuit layout and simlation hewspaper page layout graphical input fot music computer art cardiovascular radiclogy o Hardware The package developed at Toronto is aimed at device independence and supports the following devices: Tektronix storage tube refresh displays ~- the YT11 and the Graphic Wonder color-video frame buffer Versatec and microfilm hardcopy scan converter Summagraphics tablet for input o Real-time Response o High-level Interface through a set of C-callable functions supporting character output, line deseriptions with color filling of areas, segmented display files, and windowing ak © A standard format is used for cemmuniceting picture descriptions between modules and filters are used to pro- duce the final pictures ~16- © UNIX kernel modifications -~ grabcore -- reserves a buffer in low core for various displays ~ weptouser -- enables the user. to set up segmen- tation registers to refer to buffers that have Leen grabbed RK KEK Steven Zucker Interactive System, Corporation 1526 Cloverfield Bivd Santa Monica, California 90404 Steve described the Rand Virtual Terminal concept and its implementation on a Genisco bit-map display system. Suppor! for multiple windows on each of several screens, with cvlor and vector capability as well as characters in each window, is provided in the Genisco. A new system eall, "split", creates a “virtual terminal” from a rectangular region of an existing terminal (real or virtual), and ‘gives the user another input/output port to/from UNIX, keKR KR Gary kaetz Navel Postgraduate School Code 52 Monterey, California 93940 Gary described a very powerful dual 12/50 system in which one of the 11/50s is used for ordinary UNIX Li ATE timeshariny, while the other is devoted to driving graphics devices. The two 11/59's communicate via shared files on three 80 mepabytc disks and via direct connections through tro DRI1L-Cs. o Hardware Supported | - Ramtek 645 x 240 x 4 byte wap display - CONUSRAPRICS device - Vector General tablet - Vector General refresh display with 3D hardware winduwing, rosation, and translation ~ A third processor,:an 11/44, shares 32K of mewory with the graphics 11/50. The 11/34 supports two Vector Gunera) displays and is downline leaded *via two DRlls back to back. o Software - Plot/10 converted to € - A distributable driver fur the DRL1-B kRKKSK aig Mike O'Brien Department of Intormation Lugineerin; University of Illinois Chicago Circle P. 0, Box 4348 Chicago, Illinois 60626 Mike spoke very bricfly about a new C cempiler which supports long variables with initialization, structures, containing variables with byte ficlds, conditional com- pilation, structure initialization, and a new printf program. eR RE Steve Bunch University of Illinois Centcr for Advanced Computation 5 Buena Vista Urbana, Illinois 61801 Steve spoke about a C compiler for the Honeywell level 6 which is to be in the publie domain. At this point in the meeting there were a number of announcements made from the floor of varicus languages available under UNIX from various srcuces. 1n particuler Commercial Unicn Leasing Corporatien, Now York City, sor as well as FURTRAN apparently has a © to FORTRAN pro IV PLUS running under UNIX. the Commercial Union FORTRAN TV PLUS is vastly better Reporis have indicated that 7-18 alge than UNIX I'GRTRAN and it js available for a license fee from Cow 2xeial Union. ee Ee eS, Tucker Taft Harvard Science Center Haryerd University Cans igpze, M. achusetts 02138 Tucker described ECL, an extensible language that is being run at Harvard, Documentation is available from the Harvard Science Center. : RRR EE Arthur Olson Department of Chemistry University of California San Diego/LaJolla, California , B-014 92093 Arthur Olson announced that San Diego was running the 11/40 floating point unit under UNIX. KEREE Evelyn Walton UCLA Security Group University of California at Los Angeles Westwood, California 90624 Evelyn Walton made a brief presention of the Pascal 26." to C translator being used by the UCLA security kernel project. The purpose of the translator was to enable the preductien of code in Pascal for which an automatic verifier oxists. No attempt was made to translate all of Pesecal to C. Thus the translator does not support sets or nested procedures. ; a Peter Weiner Interactive Systems Coxporatioa 1526 Cloverfield Blvd Santa Monica, Californie 90404 At this point in the meeting, Peter Weiner of Interactive Systems Corporation solicited suggestions from the floor en areas of UNIX that needed improvement or extenrion. . Several such suggestions were forthcoming especially in the networking area. Ree K Mel Ferentz City University of New York Brooklyn College Brooklyn, New York 11210 Mel announced that the UNIX Users Group will be incorporating as a nonprofit educational organization in ‘order to obtain such berefits as favorable postage rates on its mailings. He also announced that the software “distribution center will be moving from Chicago Circle to New York where the availability of greater machine resources will make it possible to speed the delivery of new distributions. The scheduic for the next Users Group meeting was:discussed; it will be published in the NEWSLETTER when fixed. kee KE <29— NETWORKING Jody Kravitz - Center for Advanced Computation University of Illinois Urbana, Illinois 61801 Steve Abraham JCLA Security Group University of California at Los Angeles Westwood, California 90024 The first networking presentation was made jointly by Jody Kravitz and Steve Abraham. They announced that there will be an official release of the UNIX ARPANET NCP (Network Control Program) combining changes made at the University of Illinois, at UCLA, and at Rand. The new release will be available in mid-summer From ILL-NTS. Re KE Ken Thompson Bell Laboratories 600 Mountin Avenue Murray Hill, New Jersey 07974 Ken described an experimental UNIX networking facility he is working on at Bell Laboratories. feature of the network is a protocol which provides a “directory assistance" facility. host accepts calls for directory assistance and provides An interesting A demon proccss on each routing information based on the part of the. network that it knows about. The call initiation protocol establishes : 3 : ; =2§< a path between the nodes on the network from source to destination and the messages transmitted from the source to the destination 211 follow the same path. Present plans call for the use of a new DEC device, the KiC-11, a small microprocessor, to provide support for the network. EER brian Lucas National Bureau of Standards Room A264 TECH Washington, D.C. 20234 Brian Lucas discussed the ETHERNET. The ETHERNET provides a very high bandwidth yet low cost means of connecting machines that are within a single building or cluster of buildings. The network utilizes coaxial cables which support a one to two megabaud signalling rate. Adding a new host is as simple as connecting to the cable with a high impedance tap. Microprocessors between the - host and the cable perform the actual signalling and detect and resolve conditions in which more than one host tries to signal simultanevusly. Ke EK ~24- DATA BASE MANAGEMENT SYSTENS John Hoskins Office of Insticutionsl Research 340 Edwards Street New Haven, Connecticut 06520 John Hoskins described the Yale University Registration System. The system manages ten years of Yale undergraduate student records, Each student record is a separate file and the system holds approximately 15,000 records of two to three thousand bytes each. The system has been very well received in the Registrar's office where personnel are trained in only 4 hours and become expert in the use of the system in only 2 weeks, The primary components of the system are the Text Editor, a program called the “fence (which makes available an editable copy of a student's record and prevents simultaneous update) and a number of shell files for producing grade reports, class schedules, and other reports as required. Those involved with it -- both developers and users -- speak very highly of the convenience and economy of using UNIX -~- even when compared with other larger and much more expensive operating systems and machines, tk eKK Dan Giclan New York Telephone 140 West Street, No. 550 Kew York, New York 10007 Dan Giclan reported om the development of an enhanced, 225 production version of the INGRES system developed at Berkeley. ‘fhe improved system is oriented towards pro- duction use rather than theoretical completeness. By vesting ownership of the data bases in the user rather than the system and by placing responsibility for avoiding the rare but potentially dangerous problem of concurrency (simu? taneous update) the system is able to run ten times faster than the original Berkeley system. KEK EK , Bill Mayhew The Children's Museum Jamaicaway Boston, Massachusetts 02130 Bill Mayhew described "The Information System," avail- able for license at the above address. The Information System is distributed as a collection of routines that perform the standard database operations: add item, delete item, add descriptor to item, remove descriptor from item, locate descriptor, retrieve next item in inverted list, delete descriptor from dictionary, plus the AND, OK, and AND NOT hitlist boolean operators. Also auppliecd is a user interface implementing a simple query language and providing facilities for entering, up- dating, and retrieving textual data items. The Information System can be applied to a wide range of information management problems, It has been success- fully used to develop interactive maintenance systems for mailing lists, membership and contribution records, and 2762 group visit and educational program reservations, and is about to be used as the foundation for a service to match the educational resources of cultural organizations with the needs nf teachers throughout the Commonwealth of Massachusetts. keEKKE 7-22 =2]= PHOTOTYPESETTING Joe Gssana Bell Laboretories 600 Mountain Avenue Murray Hill, Mew Jersey 07974 The principal speaker at the Phototypesetting session was Joe Ossana. He announced a new phototypesetting package, typesetter V7, which is or soon will be available from Western Electric for a $3300 license fee. The new package combines NROFF and TROFF and is written in-C. This resuits in TROFF being 50% larger and 20-30% slower than the earlier version. There have been a number of significant improvements in the package however. First TROFF font control and width calculations are now taken from files so it is relatively casy to use the package to drive other phototypesetters. Second, one can now specify artificia} belding which is performed ky over- striking characters with a small offset. EQN the mathe- matics typesetting program now works with NROIF. Also: o Bell Labs is looking into other typesetters (an APS4 or APS5S typesetters, which sells for about $100,000). TROFF will easily drive it, although makina use of the advanced features such as more fonts or sizes may be difficult. : co Bell Labs has a Tektronics 4014 TROFF simulator which, though slow, can show what a typeset page will look like. ge © Measurements have indicated that NKOFF hypenation is correct approximately 97-1/2% of the time. o There is a new columnar cable builder. kREKKE Two other announcements were made at the Phototypesetting session, Larry Smith Texas Student Publications Pb, 0, Box D University of Texas Austin, Texas 78703 Larry announced that the University of Texas at Austia Student Publications are running two PHOTON typesetters under UNIX. ke EK Gerry Barksdale Naval Postgraduate Schoel Computer Science Department (Code 52) Monterey, California 93940 Gerry announced the availability of fonts which can be printed on a VERSAYEK printer. Re KKK REE KEK kekKEK During a break: in the Pholetypesetting session, two avavds were presented. ‘The fixst was presented by Gieg Chesson to Steve Holmgren for his work in organizing ring the conference. Steve was pleased ta receive and « a stuffed pheasant for his mautlepiece. The second award was presented by Ken Thompsca to Deunis Mumaugh of the Department of Defense for huving the largest collection igstence. of UNIX users software in ex. Anyone wishing to purchase an angry-looking rubber chicken should get in touch with Dennis. KR REE keke EK Keka