Sun-Spots-Request@RICE.EDU.UUCP (08/02/86)
SUN-SPOTS DIGEST Friday, 1 August 1986 Volume 4 : Issue 24 Today's Topics: Re: Strangeness with pty's and/or emacs on Sun-3's (3) Using suntools as a login shell (2) New .afm files for the LaserWriter + RCS on Sun 3.0 vt100 tool Setup - flames & comments TTY Hung on Sun 2? SUN PC-NFS support? Rdump problem with xylogics 472? STC 501 tape controller device driver wanted? Anti-glare/radiation screens for Suns? ------------------------------------------------------------------------ Date: Fri, 25 Jul 86 11:02:33 edt From: seismo!allegra!phri!roy@SALLY.UTEXAS.EDU (Roy Smith) Subject: Strangeness with pty's and/or emacs (1) Something very strange is going on with pty's on our Sun-3's running Sun 3.0 Unix. I don't know too much about pty's, so some of the assumptions I've made may be wrong. I'll try to stick to describing the symptoms. We've had some other troubles with emacs, so I'm not even 100% sure it's a generic pty problem, a problem with Sun windows or emacs's fault. Hence, this is going to sun, unix, and emacs groups. The first time it happened was a few weeks ago. Somebody called me to say a window had died (or something) on their workstation. When I did an rlogin to their Sun, instead of getting a shell, I got emacs coming at me, sort of. It was asking me "buffer changed since last write to file, save?" like it normally does when you try to exit emacs without saving your file (we run CCA version "164.03 #1 of 5/25/86"). For those of you not familiar with Emacs, if you type anything except y, n, ^G, or ^L in response to this question, it beeps at you, you get an error message, and the question repeats. This is what was happening, continuously. Today, somebody logged onto a different workstation and and when one of their default windows came up, there was the same sort of thing. An emacs left over from whoever used the workstation before was still running, doing the "do you want to save?" infinite loop business. When we did a ps, we found the following (I've edited out a lot of random junk, and some of the columns to make it fit on the screen): F UID PID PPID CP PRI NI STAT TT TIME COMMAND 1400200 212 5094 1 0 15 0 IW co 0:01 -csh (csh) 1008001 212 5397 5094 0 1 0 I co 0:03 suntools 1000200 156 5036 1 0 15 0 IW p2 0:01 -bin/csh (csh) 8001 156 5045 5036 45 36 0 R p2776:10 EMACS ddd 1000000 156 5046 5045 0 1 0 IW p2 0:00 EMACS ddd 408001 212 5402 5397 64 41 0 R p2 6:59 shelltool [lots of args...] 1000000 212 5403 5402 0 3 0 IW p2 0:00 -bin/csh (csh) 1000200 156 3796 1 0 15 0 IW p3 0:00 -bin/csh (csh) 1008000 156 3820 3796 0 4 0 IW p32731:11 EMACS dud 1000000 156 3821 3820 0 1 0 IW p3 0:00 EMACS dud 1008000 156 5035 1 0 4 0 IW p3 0:00 shelltool [lots of args] What you basically see is 2 emacs's left over from the last person who used the machine. From the huge amount of cpu time they have racked up, these are runaway processes (we've seen emacs do this before on our 4.2bsd vax). Note that ttyp2 is the controlling tty for both the runaway emacs and the new shell running in the window. Likewise for ttyp3. The situation with the rlogin I had a few weeks ago was essentially the same. So, what's going on? Is it emacs's fault or is there something flakey with pty's on the Suns? I've noticed a lot of strange things with emacs on the Suns. In all cases, I can't be sure if it's emacs messing up, or just tickling a bug in the Sun window system. We've tracked down one bug where if you let emacs sit idle for a while (5 minutes or more?), the display gets messed up in all sort of strange and wonderful ways (like it was sending the wrong escape sequences). CCA has been unresponsive to our bug reports, so I'm taking this to the net. ----------------------------- Date: Mon, 28 Jul 86 14:22:19 EDT From: Broadway's Streetsweeper <dupuy%amsterdam@columbia.edu> Subject: Re: Strangeness emacs on Sun-3's (2) In article <2401@phri.UUCP> roy@phri.UUCP (Roy Smith) also writes: > I've noticed a lot of strange things with emacs on the Suns. In >all cases, I can't be sure if it's emacs messing up, or just tickling a bug >in the Sun window system. We've tracked down one bug where if you let >emacs sit idle for a while (5 minutes or more?), the display gets messed up >in all sort of strange and wonderful ways (like it was sending the wrong >escape sequences). CCA has been unresponsive to our bug reports, so I'm >taking this to the net. >-- >Roy Smith, {allegra,philabs}!phri!roy >System Administrator, Public Health Research Institute >455 First Avenue, New York, NY 10016 There were some changes to the ttysubwindow escape code sequences between 1.x and 2.x (I have no idea about 3.x). These caused our ancient version of CCA Emacs to get rather confused. Perhaps there are other problems involved with the fact that CCA Emacs uses (Pavel Curtis' version of) terminfo rather than termcap? @alex ----------------------------- Date: Mon, 28 Jul 86 14:11:38 EDT From: Broadway's Streetsweeper <dupuy%amsterdam@columbia.edu> Subject: Re: Strangeness with pty's and/or emacs on Sun-3's (3) In article <2401@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: > > Something very strange is going on with pty's on our Sun-3's >running Sun 3.0 Unix. >... >I got emacs coming at me, sort of. It was asking me "buffer changed since >last write to file, save?" like it normally does when you try to exit emacs >without saving your file (we run CCA version "164.03 #1 of 5/25/86"). >... >When we did a ps, we found the following > F UID PID PPID CP PRI NI STAT TT TIME COMMAND >1008001 212 5397 5094 0 1 0 I co 0:03 suntools >1000200 156 5036 1 0 15 0 IW p2 0:01 -bin/csh (csh) > 8001 156 5045 5036 45 36 0 R p2776:10 EMACS ddd >1000000 156 5046 5045 0 1 0 IW p2 0:00 EMACS ddd > 408001 212 5402 5397 64 41 0 R p2 6:59 shelltool [lots of args...] >1000000 212 5403 5402 0 3 0 IW p2 0:00 -bin/csh (csh) >... >Roy Smith, {allegra,philabs}!phri!roy >System Administrator, Public Health Research Institute >455 First Avenue, New York, NY 10016 This is not a new problem--we had it here at Columbia over a year ago, when we still used CCA Emacs on ouro Suns. The problem is (almost) entirely CCA Emacs' fault, and since we switched over to Gnu Emacs we have not had any problems. Runaway EMACS processes are caused by exiting suntools with emacs running in a window, when the emacs has modified unsaved buffers. Exiting suntools either from the Exit root menu item, or through lockscreen -e, causes a call to the win_screendestroy routine. Win_screendestroy sends a SIGTERM signal to the owner of every window on the screen (e.g. a shelltool). When the shelltool dies, its side of the pty is closed. This causes a SIGHUP to be sent to any processes whose controlling terminal is the slave end of the pty (e.g. csh, emacs). CCA Emacs is not handling this SIGHUP in the proper way. You should be able to duplicate the problem by aborting an rlogin connection, without using Suntools. When a dialin terminal or pty is hung up, the status returned on read attempts is an EIO error. Hardwired terminals will not return this error. CCA Emacs is obviously not checking for this (or any other?) error in their SIGHUP cleanup processing. As a result, it loops indefinitely, trying to get a response when none will ever arrive. So now that we have diagnosed the problem, what is the solution? The best one, by far, is to toss CCA Emacs on the ashheap of history, and use GNU Emacs instead. If this is not practical, the next best solution is to hack an EMACS killer into suntools users' .logout files. I had written one, back in the days when we had this problem, but no longer have a copy. You can write one using ps and egrep; it's straightforward enough. You could use "skill -9 emacs", except that the two "EMACS" processes are really "emacs" and "csh", and killing one doesn't do in the other. Anyone want to hack skill to look at argv as well as u.u_comm? @alex ----------------------------- Date: Sat, 26 Jul 86 09:57:35 edt From: seismo!allegra!phri!roy@SALLY.UTEXAS.EDU (Roy Smith) Subject: Using suntools as a login shell (1) In Volume 4, Issue 23, David Chase <rbbb@titan.rice.edu> asked about some strangeness that happened when he killed his login shell out from under (over?) suntools. This isn't really on the original subject, but the past few days I've been toying with the idea of installing suntools as my login shell in /etc/passwd. This would speed up logins a bit, and for novices, it would make explaining how to log out easier ("What do you mean I still have to log out after picking "exit suntools" from the menu?"). I did a small experiment and tried doing "exec suntools" from my login shell. This didn't seem to have any ill effects so it doesn't look like suntools depends on having a shell running above it. If I just put /usr/bin/suntools in /etc/passwd, would that work? In vanilla 4.2, the login shell does a fair amount of work to set up your environment. Has this been moved into /bin/login where it really belongs in Sun 3.0? Another problem is rlogins -- since I'm not on the console then, suntools can't run. Is there any way to make suntools exec a normal login shell in that case (we have source on order, so easy source fixes might be feasible; I'm not up for a major re-write of suntools, however). Assuming you could get all that right, is there any magic reason for having a real login shell running? ----------------------------- Date: Fri, 1 Aug 86 15:31:27 cdt From: William LeFebvre <phil@Rice.edu> Subject: Re: Using suntools as a login shell (2) Although it is probably true that suntools doesn't need to have a shell running above it, I would not at all recommend changing your login shell to be suntools. Many programs (perhaps even suntools itself), look at either the login shell entry of /etc/passwd or at the environment variable SHELL to get the name of a program that can process shell-like commands. How do you suppose suntools finds out what process to start up in a window when you ask for a shelltool? Well, I just looked and it uses the environment variable "SHELL". If you change your login shell, this environment variable will reflect the change. Wouldn't that be interesting? If you really want to do this, I would recommend putting the line "exec /usr/bin/suntools" in your .login, along with appropriate checks to see if you are logging in on the console. And remember that everything in the .login after the "exec" will not get done (unless the "exec" fails). William LeFebvre <phil@Rice.edu> ----------------------------- Date: Tue, 29 Jul 86 9:31:26 BST From: William Roberts <liam%cs.qmc.ac.uk@Cs.Ucl.AC.UK> Subject: New .afm files for the LaserWriter + I sent a PostScript program to Info-PostScript which generates .afm files from the font descriptions in the LaserWriter +. These files can then be used with the TranScript software to produce new font files to allow these fonts to be used with Ditroff. I suspect that using them with troff will be about as much fun as pulling your own teeth out, but you are welcome to try. Anyone who wants a copy of this program but didn't catch it as it went past on Info-PostScript, mail me and I will send it to you. If there is sufficient interest I will send it to mod.sources as well. William Roberts ARPA: liam@cs.qmc.ac.uk (gw: cs.ucl.edu) Queen Mary College UUCP: liam@qmc-cs.UUCP LONDON, UK Tel: 01-980 4811 ext 3900 ----------------------------- Date: Tue, 29 Jul 86 17:11:45 PDT From: gerolima@Ford-wdl1.ARPA (Mark Gerolimatos) Subject: RCS on Sun 3.0 We here at Ford have a working copy of RCS (thanks, Vince!), which ALSO works on version 3.0 (!!). However, certain RCS functions core dump if supposedly [optional] flags are not provided. I think guy@sun explained the problem there (ie: assuming null pointers readable...shame shame shame...). As for the pedigree of our version, well, hum, gee, I got our copy from a friend at another company, so I know little about where it came from. But I can tell you that the copy you can get from ftp@perdue.edu does not work, at least on our Suns. "For almost a quarter of a century..." "...Change Baby, Don't Worry!... Mark Gerolimatos ...Welcome! Welcome!... ARPA: gerolima@ford-wdl1.arpa ...Change Baby, Don't Worry!... UUCP: {sun,fortune}!wdl1!gerolima ...Box! Box! Box! Box!... AT&T: (415) 852-4105 ...Now, We Say Good-Bye... Mail: c/o Ford Aerospace ...Welcome to the GALATT... 3939 Fabian Way ...G-A-L-A-T-T We're GALATT..." Palo Alto CA 94306 -English phrases from a Japanese song Mail Stop X20 ----------------------------- Date: Sat, 26 Jul 86 07:19:44 EDT From: Sid Stuart <sid%linus@mitre-bedford.ARPA> Subject: vt100tool We have just posted our vt100tool to mod.sources. This tool works under release 2 of the Sun operating system. We have already received some bug reports that will be helpful to those of you trying to install the tool. 1. `decode.sh' should cd back up a level before cd'ing to fontdir 2. uudecode leaves output files belonging to uucp We are working on a hack of the tool to make it run under release 3.0. As soon as we receive the sources to release 3, we will port the tool to it properly. In the meantime, Ed should have the hack going within the week and we will post a uuencoded version to mod.sources. sid@linus ----------------------------- Date: Thu, 31 Jul 86 18:02:06 -0300 From: Leonid Rosenboim <leonid%taurus.BITNET@WISCVM.ARPA> Subject: Setup - flames & comments I've experienced in the near past the installation of Relase 3.0 of SUN UNIX. I have a few comments & suggestions about that. Hope that someones takes it seriously. Why is the Setup program so bloody automatic ? Is it really intended for dumb people ? or was is ment only to impress ? It seems that once you pushed the SETUP-EXECUTE button, you have no choice but follow the instructions you get from it, or L1-A. If it works - ok, but when you have real problems with the tape drives, it's a disaster. There seems to be no way of skipping steps. I also noted that once you are in the SETUP-EXECUTE mode, there is nothing you can du, i.e. the scroll bar doesn't work nor does anything else, EXIT, REBOOT. There is no way to abort setup and sync the disks, so usually if you abort with L1-A, you must go back to standalone copy - and that takes time. The next problem occurrs when you mix SUN3 and SUN2 clients. Setup insists on making separate copies for 68020 and 68010 binaries. But suppose there's not enough disk space for both ? I somehow managed to go around Setup - I gave it almost zero space for usr.MC68020, it failed to restore that system, and when I brought up the system I had to salvage the symbolic links manually. Is it that hard to have it as an select-option to have separate /usr.MC68020 and /usr.MC68010 or just a common /usr with 68010 binaries ????!!!! I guess you have noticed that I experienced installations with limited disk spaces, so here comes another suggestion: It is true that programs like troff, eqn, tbl and files such as ms/me macros, /usr/dict and others are UNIX standard utilities. But suppose there is a serious limit on the disk space, in which case it might be a good idea to skip such programs. I suggest to put the text processing stuff in a separate tape file and make it a separate software option. (by the way, I like the software options idea). Also I suggest to add to the panel that asks you to put in a tape, besides the "OK" answer an "STOP-EXECUTION" button that will enable you to return to setup without losing your configuration. Leonid ----------------------------- Date: Mon, 28 Jul 86 11:47:50 CDT From: wsmith@b.CS.UIUC.EDU (William Smith) Subject: TTY Hung on Sun 2? Is this a known problem in suntools?: If you type 253 characters into a shell tool window on the same line, all is lost. All further characters are ignored and bringing up a menu on the edge of the window also brings up behind it a menu saying "TTY Hung?" This behavior is not exhibited out-side of suntools, but it is within the program 'script'. This leads to the pseudo tty drivers as a source of suspicion. In suntools the same problem was exhibited on a SUN running 1.2 software. I hope this problem does not exist in Sun 3's. Bill Smith ihnp4!uiucdcs!wsmith wsmith@b.cs.uiuc.edu ----------------------------- Date: Wed, 30 Jul 86 08:20:26 EDT From: jqj@gvax.cs.cornell.edu (J Q Johnson) Subject: SUN PC-NFS support? Has anyone outside of SMI had any experience with the SUN NFS package for the IBM PC and clones? This product is either a software-only package that provides NFS (network file system) support for PCs based on the 3Com Ethernet board, or a bundled sw+3Com package. It appears to support not only NFS but also some of the more traditional tcp/ip services. My questions: 1/ how well does the NFS support work? How well integrated is it into MS-DOS? What is the user interface? 2/ what other networking services are supported for the PC? Telnet certainly. FTP? Rlogin? ARP? etc. How flexible is the programming interface (in case I want to add my own services); is there a library of routines for tcp/ip based networking? 3/ what minimal configuration is required on the SUN side in order to use a SUN effectively as an NFS server? Note that SMI sells a packaged system consisting of a SUN-3, 4MB memory, 71MB disk, and 6 PCNFS boards+sw. Is that configuration usable for anything (tiny disk, no tape backup)? ----------------------------- Date: Tue, 29 Jul 86 10:34:07 edt From: allegra!mp@seismo.CSS.GOV Subject: rdump problem with xylogics 472? We recently put a 1600/6250 tape system on a Sun-2/170 (Xylogics 472 tape controller and some type of CDC tape drive). The problem we're having is that we periodically get data late errors when doing rdumps. Since this is a hard error, it requires us to restart a particular dump tape after spending as much as 2 hours writing it. We checked with Sun, and all the jumpers on our (prime) cpu board and the 472 are correct. Here's the order of the boards in the card cage. The 472 is jumpered for priority 3. cpu-mem-mem---video-ie--472--450 We're running 3.0. The 170 currently does essentially nothing except act as a tape server - no nd activity, and little NFS activity. Any suggestions? Mark Plotnick allegra!mp ----------------------------- Date: Sat, 26 Jul 86 11:25:15 EDT From: dms@HERMES.AI.MIT.EDU (David M. Siegel) Subject: STC 501 tape controller device driver wanted? We would like to put an STC 501 compatible tape drive on a Sun. The Xylogics controller/device drive that Sun supports is Pertec compatible, and can't handle STC-type tape drives. I was wondering if anyone has hacked a device driver for an STC controller. Thanks, -Dave ----------------------------- Date: Fri, 25 Jul 86 13:06:51 edt From: seismo!allegra!phri!nomi@SALLY.UTEXAS.EDU (Nomi Voroba) Subject: Anti-glare/radiation screens for Suns Does anybody know a distributor who sells anti-glare/radiation screens for Sun Workstations? Nomi Voroba, {allegra,philabs}!phri!nomi Public Health Research Institute 455 First Avenue, New York, NY 10016 ----------------------------- End of SUN-Spots Digest ***********************