* * * * ORR * * * * * * KK O* * x OR NOR Ok * * * Oe OX * * OR OR * * * KK * OK wR ROK * * TERK * * * * Volume 2 Number 5 May - Juno 1977 SRS ACCOR GR ISSR HESS a aS ASK a OK aK aK fs {ROK SE AGORA IR EVE ROOK RR AR IR aR OR OR aR OK * * * NOTICE * * * * This document mny contain information covered * +k by one or more ticenses, copyrights, ond * ok non-disclosure agreements, Clreulation of * * this document is restricted to holders of a * * license for the UNIX software system from 2k * Western Electric. Such license holders may * * reproduce this document for uses in conformi- * * ty with the Unix lleense,. * * oe * All other clreulation or reproduction is x ak prohibited. * * * SERA AICIOUGCAR ICIS ISIIORK IORI OR ICCC SOK SER SCSI ERK RE OK ORE THIRD SOFTWARE DISTRIBUTION The Third Software Distribution is now being prepared for relense. We expect to start mailing it out in late July, ‘The Softwnre Distribution Center has been moved from Chicago Clrele to the City University of New York. We all owe Mike O'Brien a debt of gra- titude for the work he has done in setting up the software distribution service. Mike is leaving Chicago for the West Const soon. He prepared the Third Distribution and tins passed on to me (Mel Ferentz) alt of the tapes people sent him as well as the entire correspondence file. The distributions will be preparcd on the City University's 970/168, which we view as a suitable back-end for a UNIX system. Complete details on the distribution will be con- tained in the next Unix News. Those of you who have already sent tapes to Chicago will receive your tapes malied from New York. No further tnupes should be matled to Uhicago. The CUNY Computer Center sells tapes over-the-counter and while we will continue to write onto your tape If you send one, the preferred medinm for us is to write your distribution on a virgin 2406 foot tape, An order form will be included with the next newsletter. URBANA MEETING The Urbana Meeting was attended by over 150 people and was a great success. The ato tendanee list will he published as soon ns we pet a tape from Steve Wolmgren to replace the one he sent us that was folded and spindled by our favorite postal service. We have been promised minutes of the meeting which will also nppear aS soon ag ren celved. CHILDREN'S MUSEUM INFORMATION SYSTEM The Children’s Museum hag announced the availability of ita "Information System - Verston 8", A four page product description was distributed at the Urbana Meeting. For a copy of the description, more details, or licensing information contact Bill Mayhew, The Children’s Museum, Jamaicaway, Boston, MA 02130, FUTURE SOFTWARE RELEASES At the Urbuna Meeting it was snid (announced ts too strong a verb) that Bell its preparing Progranmer’s Work Bench for release this summer with Version 7 of Unix soon thereafter. Mini-Unix has been relengsed nad LSI-Unix and Mert wlil probably follow along atowvome liter date. Prof. Melvin Ferentz Brooklyn, NY 11210 Amsterdamr 25-Nov-76 Prof. Melvin Ferentz Physics Derartment Brooklyri Collede of CUNY Brooklyn NeY. 11210) U.S.As Dear Professor Ferentz+ We are using UNIX on our FOP 11/45 for almost a year now and are verw enthusiastic shout it. Our system is somewhat overloaded but we hore that the disk drives we ordered will hele to solve this. Lately we found a ‘bus’ im the UNIX kernel. One of our users was having troubles with his Frrogram that was switching bsck snd forth between single and double precision Floatina Point nodes We discovered that the F.F. resisters are saved in ere mode the F.F. erocessar has at the moment the rrogram is storred. This means that the low order 32 bits of the users double Frreci- sion redisters were not saved whenever his Prodram was storred in single mode. By adding setd instructions in m45.s Jjust before the lines where the F.F. resisters are moved to and from ur we solved this rrotvlem. Consequently the F.P. resisters are always stored in double mod2. The rrograms db and cob will have to be chensed to reflect the neu situation. A few months ago somebody moticed that the times stated by the time comand were somewhat off,Time expects that the system commend times returns Frocess and system times im 60ths of seconds. ut since we have a 50 Hz Power supelyr times returned those times in SOths of seconds. He chansed time.s according to cur situstion. We had some problems with the pire mechanism. When several Processes were writing simultaneously on one rire their messades sot antermixed if the pire rointers reached the end of the rire buffer. . In case somebody is interested in a driver for the old IEC DMi1 multiplexers we would be glad to send a cory of our drivers Sincerelur ZC (eaer E.G.Keizer Vaksroes Informatika Vride Universiteit Tle Boelelaan 1105 kamer 4A-16 Austerdam The Netherlands bs FACULTEIT DER WISKUNDE EN NATUURWETENSCHAPPEN i Toemooiveld, Nimegea ' Telephone 050-55 68 33 Telex 4 $2 28 winat ol KATHOLIEKE UNIVERSITEST NIJMEGEN THE NETHERLANDS cae Computer Graphics Informatica April 4 1977 Prof. Melvin Ferentz Brooklyn College of CUNY Brooklyn, NY 11210 U.S.A. a Dear Professor Ferentz, ir no one else wrote you about the matter before, nere fix te the ttyn(II1) problem mentioned in the Fegruary of UNIX News. I found the bug about 3 months ago.\ « Tustale issue After the line that reads: wov buf+2,(sp) I inserted: cul, ri syS stat;dev;bur - bes er1 fi 1- Apet’- ae ‘ emp buf,ri ‘ne ert “tee A. similar change has to be made to nreff(I), file: §7/nroffi.s. This file contains a slightly different version of ityn. The Re commengs may be cons. sees can es Toy > pilation: Hh, goo, 2 login, mail, g, nx, we6. I reg¢ompiled only Ahe first four. I have already informed Mr. Lucas. Sincerely, George Rolf Copel Mode 4 3000 2.75 PURGUE UNIVERSITY SCHOOL OF ELECTRICAL ENGINEERING WEST LAFAYETTE, INDIANA 47507 April 9, 1977. Professor Melvin Ferentz. Fhysics Department, Brooklyn College of CUNY. Erooklyn, MY 11210. Froblers with creat system call on Unix version 6 Dear Frofessor Ferentz, We have discovered two problems with the "“creat™ system call. The Following sequence of commands will cause “orphaned” Files (files that are not in any directory) to be created: chéir /tmp mkdir a chéir a traic /trp/a ls -1 / >Dorphan eheir / The rmdir causes the link count for the /tmp/a inode to goto zero, however the inode is not dealloceted because it is the shell's current directory. At this point one can create files in the current directory. One (except Super user) cannot create directories in the current directory because mkdir does a stat on ".." which does not exist. Upon doing a chdir /, the reference count for the old current directory goes to 0, causing dealloca- tion of its inode and stranding the newly created files. The second problem occurs when the maknode call in creat() fails due to no inodes on the device. Namei leaves the last girectory inede in the pathame locked because a return is execut- e3 efter the maknode failure. The next process to reference the locked inode will go to sleep (and hang!) with PINOD (-90) prior- ity. The fix for the first problem consists of adding an error return if the current directory inode has a link count of zero. Eelow is a copy of the existing creat(}) in /usr/sys/ken/sys2.c ang the revised one. creat () Existing creat() in /ust/sys/ken/sys2.c¢ register *ip; extern uchar; ip = namei(suchar, 1); if(ip == NULL) [ if(u.u_error) return; ip = maknode (u.u_arg[1]&077778(~ISVTX)); if (ip==NULL) return; openl(ip, FWRITE, 2); } else creat () { register *ip; extern uchar; ip = namei(suchar, 1); if(ip == NULL) { openl(ip, FWRITE, 1); Kodified creat() in /usr/sys/ken/sys2.c welded A-See- 77 Vey if(u.u_error) ae 2 Feturny pao ae ~d£((u.u_cdir->i_nlink == 0) && (fubyte(ueu_arg[0]) 4= '/'}) { Sor u.u_error = ENOENT; f err: ae iput(u.u_pdir); /* namei left parent dir locked */ return; natbesp ine soone Bac | “ip = maknode(u.u_arg{1]&077778(~ISVTX}); if (ip==NULL) «goto err; )-— openl(ip, FWRITE, 2); } else openliip, FKRITE, 1); } Sincerely yours, George Goble May 26, 1977 Professor Melvin Ferentz UNIX User's Grou m College of CUNY My NEY I2210 Dear Professor Ferentz: We are using a UNIX operating system on our PDP 11/70. would like to know if there is a compiler available for CO5CL or RPG Il. a BNR INC. Stantore Industeat Park ‘gn74 Pores Dewe. 5~+ VRIJE UNIVERSITEIT SUBFACULTEIT WISKUNDE WISKUNDIG SEMINARIUM amsterdam 1011 eee toni Prof. Melvin Ferentz, postbus 7161 Br ge of CUNY, telefoon 020 - S48 24 10 Brooklyn, N.Y. 41210, UsSeAy uw kenmerk ow brief van ons kenmerk =P dam bi tagen onderwerp Dear Mel, I would like to announce the evellability of a Pascal Comiler for Uiix. Under separate cover i have sent two cestenes to Mike O'Brien for in- cerporation in the next distribution. Beth binar: cluded. The compiier is Neavily nedified cerivativ: fet which ds tte eec oT a Ses y working can be (oationaily) will previde for quicker itself, will not fit in the POP-14's acd Yours truly, Orcbs Andrew S. Tarenbeum. UNIVERSITY OF GLASGOW Computing Science Department, Tue University, GLascow, G12 8QQ. Te 29 3853 eet. 478/7358 10th June, 1977. Dear Professor Ferentz, i am writing to let you mow of the formation of a U.K. Unix Users The first meeting took the form of a Colloquium at Glasgow University dey 27th Nay, attended by e2bout 40 peopie. Short talks were presented sts of the kernel including the scheduler and the buffer cache system, ucture of CAC "Network Unix", the features of the Camegie Kellon INORES rience with the Toronto graphics ser soos we eae constituted. an executive committee. 2s agreed that an attenrt should be mde to constitute the group es terest Group under the umbrella of DECUS U.K. We are seeking approval ais rove both frca Sell and from the JECIS Executive Zoard. DECUS have to hendie distrioution of she U.K letter, and will undertake to t only to accredited Unix licence 28, so we don't foresee any problems General inforration adcut meetings etc., will be published in the Newsletter, tut all systen-specific reterial will he restricted to the Licetion, ne question of languages tne appearance in the U.K. of the Princeton isplenentation ves generally welcomed, at least by the "engineering" The availability (subject, of course, to having purchesed appropriate licences) of a good FORTRAN which can be configured for the full range of hard- ware is bound to erhanoe the appeal of Unix in non-ccmputer-science depertments. are Standards Concern was expressed cn several points in the area of system standards, particularly Rste ed software. Anong the points raised were the following:- (i) User group standard software: since it is increasingly difficult for U.K. users to attend personally any ¢f the U.S. meetings, it would ve mice if the views of users outside the U.S. could be sought defore a piece 22 sof teare orasystem nod.is adcpted es e standar: In the case of the Yale Shell, we are cil 4 future proposals could be more contro: (44) Assumed hardware: wherever possible should be configured for e "standard" systex, for nodifications required for o hs conventions about pathnames, etc., should ve made ex: (iii) Documentation: manual pages should te in 'nroff' form, using the stan:ard 'tmac.naz’ macro definitions, and 1 extension '1' or '6'. Cthez documentation should eny reouired nroff necro definitions. (iv) System cells: the adoption of s' systen call steniard was suggested. from % to 63 she reserved for locally added systea calls, and no dist software should make any assunstions about system calls in this range. Software Distritution The meeting cgrecd that Glasgow should enter negotiations with a view to > for the U.x. We heve three exchangeatle should have an800/1600 spi nesrstic tape drive. collection centre for software which U.X. users wish to contribute to tae U.S. distribution centre. : If any U.S. Unix addicts are visiting U.K. this summer, please drop in and see us. (i'm sure that goes for all of the U.K. Unix sites). Best wishes. flstay C- Kalgowt Alistair C. Kilgour. Au DEPARTMENT OF COMPUTER SCIENCE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN URBANA, ILLINO!S 61801, U.S.A. April 20, 1977 Richard Miller University of Wollongong New South Wales Australia 2500 Dear Mr. Miller, IT jest got the UNIX news today, and, was greatly interested to learn of ygur work with UNIX on the 732. It so happens that ve bought a 732 a year ago, running ST in background batch as a slave to an 11/35 UNIX, and have yet to obtain any real satisfae~ tion from it. Without getting into too many details, I will say that ovr problems were bad enough that Interdata is letting us have MI for no charge (although they are taking their time in sending it). I have been spending seme spare time in the last few weeks making the second pass cf the C compiler generate 732 code. Unfortunately, I haven't the time to give an appropriate level of effort to this project, and 1 have only gotten about half dene with dt, and some of that done a bit sloppily. I wonder if there is some possibility of obtaining a copy of your 732 € compiler? Perhaps you vould also be interested in corresponding or colla- borating on the project as well. I am quite familiar with the kernel of and I might be able to provide sone useful assis- tance in getting the full system up. I might also be able to help scrounge up souree for some of the programs ve need to change if they are te run on UNIX on the 732 (I would have to dig up the FORTRAN compiler for instance, as we use the 732 for num ber crunching). Please let me know how you feel about this as I am quite anxious to do something to improve the usefulness of our machine. If it 4s possible to obtain a copy of your C compiler, please let me know what I would have to send to effect the transfer. Sincerely, : Patrick J. Lucey Governor STATE OF WISCONSIN Rober: H. Dunn Secretary DEPARTMENT OF ADMINISTRATION One West Wilson Street © Madison, Wisconsin §3702 May 25, 1977 Professor M. Ferentz Physics Department Brooklyn College of CUNY Brooklyn, New York 11218 Dear Professor Ferentz, We have recently acquired UNIX and we would like to join the UNIX user's group. Our hardware configuration is: 11/45 with 124K of core Delta data termine1s Teletype (ASR33's) 30 Ipm line printer 890 bpi, 9 track ragtape drives RPQ2 disk drives (2¢ mb’s apiece) Dual dectape drive Compuscen OCR (Model 176) PRR me We have been running UNIX on this configuration for the last four renths. Our main interest is online editing and typesetting. We have an APS-4 phototypesetter. We would like to start getting the UNIX newsletter, etc. Sincerely, USD had Mahmecd Ahmed Management Information Specialist MAirwt 261522 Alfred D. Whaley Cu, go taf Fort,