Sun-Spots-Request@RICE.EDU.UUCP (11/12/87)
SUN-SPOTS DIGEST Wednesday, 11 November 1987 Volume 5 : Issue 59 Today's Topics: Editor's remarks: uudecode, named pipes Re: GNU Emacs on the Sun-4 ESDI disks on Suns Fix for Sun 3.4 TCP problems with Symbolics LISP machines Problem with vt100tool from the Net Mysterious filesystem demolition Another 3.4 networking problem Good way to "flip a coin"? Answered my own "flip a coin" question How to test the end of a 1/4" tape? Making 2400 baud modems work on SunOS 2.x? Serial Line IP on Sun III/160 using SunOS 3.4? Adding CDC SMD drives to Sun-3? ICON previewer yet another yet another news icon ---------------------------------------------------------------------- Date: Wed, 11 Nov 87 15:47:34 CST From: William LeFebvre <phil@rice.edu> Subject: Editor's remarks: uudecode, named pipes This message concerns two different remarks that I made in recent digests. Regarding uudecode and the permissions of the current directory: someone pointed out to me that world write permisssion is usually not sufficient. You almost certainly need group write permission. In my remark, I was trying to be succinct without being misleading. Oh well. These are the steps you really need to take: obtain and remember the permissions on the current directory (i.e. with "ls -ld ."), do "chmod go+w .", do the uudecode, then set the permissions of the directory back to whatever they were. Regarding named pipes: SunOS 3.2 does have System V FIFO files. They just aren't documented in any obvious manner. A FIFO file is basically a named pipe. It can be created by anyone with a call to "mknod": `mknod("foo", S_IFIFO | 0666, 0);'. See mknod(2) for more details. You can also use the program "/etc/mknod" to create the file (even if you are not the super-user---see mknod(8)). For fun, try the following series of commands: % /etc/mknod foo p % cat /etc/termcap >foo & % more foo William LeFebvre ------------------------------ Date: 2 Nov 87 20:49:02 GMT From: janssen@milano.UUCP Subject: Re: GNU Emacs on the Sun-4 I've built gnu-emacs (18.46) for our new Sun-4, using crt0.c code from eb@kent-state.UUCP (Eric Benson): asm(".seg \"text\""); asm(".global __start"); asm("__start:"); asm("mov 0, %fp"); asm("ld [%sp + 64], %o0"); asm("add %sp, 68, %o1"); asm("sll %o0, 2, %o2"); asm("add %o2, 4, %o2"); asm("add %o1, %o2, %o2"); asm("sethi %hi(_environ), %o3"); asm("st %o2, [%o3+%lo(_environ)]"); asm("andn %sp, 7, %sp"); asm("call _main"); asm("sub %sp, 24, %sp"); asm("call _exit"); asm("nop"); asm("call __exit"); asm("nop"); It builds, and runs, but when find-file is called, crashes with a segmentation violation. I find that egetenv("HOME") returns a 0, rather than a string, and the error occurs in strlen'ing the NULL. I suspect that _environ is not being set correctly. Can you help? Thanks for any tips. Bill Bill Janssen, MCC Software Technology Program ARPA: janssen@sw.mcc.com UUCP: {seismo,harvard,gatech,pyramid}!ut-sally!im4u!milano!janssen ------------------------------ Date: 1 Nov 87 05:21:29 GMT From: sundc!potomac!jtn@uunet.uu.net (John T. Nelson) Subject: ESDI disks on Suns > 2. I'd like to buy another 140Mb disk for my sun. Is Sun the only source > for this? Can I just buy another ESDI disk and stuff it in the box? Does > anyone else make SCSI disks for the sun (ps: I know that SCSI is not a > standard, assume I've written a SCSI driver or two). You can buy ESDI SCSI interface disks from Highland Digital in Mt View CA. They have 155 meg ESDI / SCSI Ren-3 CDC disks for about $2500 and 350 meg Ren-4 disks for about $5000. The Ren-4's are a problem because the Sun formater doesn't recognize them. Highland says that they will supply a program that will format these disks. We recently purchased a Sun III/60 with two Ren-3's daisy chained onto the 60. The Ren-3's performance is comperable to that of the Fujitsu Eagle drives. Highland is a VAR of Sun equipment. Hope this helps. ------------------------------ Date: 1 Nov 87 22:19:46 GMT From: sundc!potomac!jtn@uunet.uu.net (John T. Nelson) Subject: Fix for Sun 3.4 TCP problems with Symbolics LISP machines Thanks to Bill Nesheim from Thinking Machines for the fix to our TCP problems. Briefly, when trying to converse between our Suns running 3.4 and Symbolics LISP machines, one of the machines would open a TCP connection but leave it hanging open without transfering data. The fix: Symbolics LISP machines do not recognize trailers on ethernet. You must use ifconfig ie0 arp -trailers to get it to work. ------------------------------ Date: Mon, 2 Nov 87 12:09 EDT From: CROUCH%M_REA%sdr.slb.com@relay.cs.net Subject: Problem with vt100tool from the Net Hi, I have pulled the net source 'vt100tool' from the CSNET info-server and tried to install it on our Sun. I followed the installation instructions carefully but unfortuanately I got the following error when I tried a make ------ cc -g -o vt100tool vt100tool.c libdir/nlibsuntool.a -lsunwindow -lpixrect Undefined: _ttysw_setboldstyle _ttysel_clear _ttysw_handle_itimer _ttysw_pty_input _ttysw_consume_output _ttysw_pty_output *** Error code 1 Stop. ------ Has any one else copied this software and had problems or even had success. Failing that does any one know what I am doing wrong. All answers gratefully appreciated. Thanks & Regards Sean Crouch, Computer Operations, Fairchild Semiconductor Ltd., Reading, England. Email: crouch%m_rea@sdr.slb.com OR crouch%m_rea%sdr.slb.com@relay.cs.net SInet: M_REA::CROUCH ------------------------------ Date: Tue, 3 Nov 87 13:08:42 EST From: Tony Movshon <movshon@acf8.nyu.edu> Subject: Mysterious filesystem demolition Forgive the length of this message, but I'd like to be as complete as possible in describing my difficulty, and my machine is nonstandard ... I have a Sun 3/50 (SunOS 3.2) with a shoebox fitted with two disk drives and no tape. The disk controller is an Adaptec. Drive 0 is a Maxtor XT-2190 (150 Mbytes formatted), and drive 1 is a CDC 94155/86 (70 Mbytes formatted). Both drives were formatted and partitioned locally. The Maxtor is partitioned thus: sd0a 7+ Mbytes ("/") sd0b 20+ Mbytes (swap) sd0g 60+ Mbytes ("/usr") sd0h 60+ Mbytes ("/usr/vnl") The CDC is partitioned thus: sd1a 7+ Mbytes (not used) sd1b 16 Mbytes (swap) sd1g 43 Mbytes ("/usr2") The machine both mounts and exports several NFS filesystems, and is connected to a heterogeneous local network including 8 other Suns, 3 VAXes, and some AT&T 3B2s. I have an occasional problem in which the /dev/sd0g partition (/usr) gets completely trashed. The system crashes, and of course nothing helpful remains in /usr/adm/messages because /usr/adm has disappeared. The subsequent fsck on /dev/sd0g fails. Doing an fsck by hand, the first two super-blocks of the filesystem show up as unusable (fsck reports "bad magic number"). Using fsck -b 4144 (the third superblock) elicits the response "CG 0: BAD MAGIC NUMBER", then a whole host of errors from inode 1 on up, then finally the message "ROOT INODE UNALLOCATED, CANNOT CONTINUE". The whole pattern repeats very precisely whenever the problem occurs, and it's now happened four times. I have tried various things which have *not* made the problem go away, including: - Reformatting the Maxtor disk with interleaves of 1 and 2 - Repartitioning the Maxtor disk so that the first cylinder of sd0g is in a different location - Running without the CDC disk or without the extra swap partition on the CDC disk - Upgrading from release 3.2PILOT to release 3.2 I had thought that the upgrade had solved the problem, because the system ran flawlessly for nearly a month, but then -- poof -- it happened again. The problem does not seem to be provoked by especially heavy use -- in fact it typically occurs late at night when no one is logged in. Any help or advice would be appreciated. And before anyone asks, no, I haven't reported the problem to Sun because: - We have no software maintenance support - We have a nonstandard hardware configuration, including disks not purchased from Sun and an "unsupported" shoebox setup. Thanks in advance, Tony Movshon Internet: movshon@nyu.nyu.edu or tony@xp.psych.nyu.edu Usenet: {ihnp4|allegra}!cmcl2!xp!tony or movshon@cmcl2.uucp US Mail: Department of Psychology, NYU 6 Washington Place New York, NY 10003 Phone: (212) 998-7880 ------------------------------ Date: Tue, 27 Oct 87 23:57:42 EST From: Root <root%suny-sb.csnet@relay.cs.net> Subject: Another 3.4 networking problem I suppose I was a bit incredulous at all the reports of network bugs in SunOS 3.4, until I tripped over them myself. First, caveat emptor to Mt Xinu users - their 4.3 port suffers the same broken ICMP mask reply as does Berkeley 4.3 and Wollongong. The fix is simple, but frankly I'm feeling a bit burned by Mt Xinu - after all, the big bucks they get for their software would lead one to believe they actually track bug fixes.. My second problem is that I would like to use two different subnet masks, yet the SunOS 3.4 software doesn't seem to allow it. Here is what I want to do (4.3 allows this): ifconfig ec0 129.49.2.2 netmask 0xfffff800 ... and so on ifconfig ie0 129.49.89.65 netmask 0xffffff00 ... and so on The hitch is that the mask for BOTH controllers (actually all controllers) gets set to the netmask specified by whichever ifconfig runs last. This leads one to believe that for some reason, the network folks at Sun deviated from Berkeleys if design of one controller/one netmask, etc. So, following this hypothesis, I did the following: nm /vmunix | fgrep subnet and got: (among other things) 0f072380 D _subnet_mask Which when examined using adb on an active kernel, was set to the subnet mask. This seems to indicate that Sun (blast them) designed their low level stuff so that only ONE subnet mask is allowed per system. (Flame on) It seems to me that the Sun network group should try, as a whole, to be a little less free in deciding what parts of the 4.3 network code to drop into their kernel. I should like to see a complete port of the 4.3 network code, and not some half a**ed merge of new and old. (Flame off) Ahem. Now before I run off and remedy the aforementioned problem, has anyone else run into and fixed this? Does SunOS > 3.4 fix said bug? Has anyone dropped the entire Berkeley 4.3 network code into a 3.4 kernel? Thanks in advance, Rick Spanbauer SUNY/Stony Brook ------------------------------ Date: Mon, 2 Nov 87 17:10:04 PST From: ultra!wayne@ames.arpa (Wayne Hathaway) Subject: Good way to "flip a coin"? I have a driver for SunOS 3.3 in which I would like to "flip a coin." By this, I mean I would like a quick way to make a more-or-less evenly balanced two-valued random number test. One obvious way to do this would be to test some suitable bit in some rapidly-changing location, such as a real-time clock cell or something like an interrupt counter. Unfortunately, we do not have source. Could some kind soul please recommend a variable name and bit location I could look at? Hopefully something that is not going to go away in SunOS 4.0 or such ... Any assistance would as always be muchly appreciated! Thnx. Wayne Hathaway ultra!wayne@Ames.ARPA Ultra Network Technologies 2140 Bering drive with a domain server: San Jose, CA 95131 wayne@Ultra.COM 408-922-0100 ------------------------------ Date: Tue, 3 Nov 87 07:35:41 PST From: ultra!wayne@ames.arpa (Wayne Hathaway) Subject: Answered my own "flip a coin" question Regarding my search for a rapidly-changing bit which can be used to "flip a coin" in a device driver, I may have answered my own question about as well as it can be; please comment. Looking through header files, I settled on the kernel variable "time", in particular the long "time.tv_usec" looking at the bit which is the rightmost 1 bit in "tick". Presumably this bit toggles HZ times per second (50 on my Sun), so that's reasonably changing. Other ideas are of course welcome. Wayne Hathaway ultra!wayne@Ames.ARPA Ultra Network Technologies 2140 Bering drive with a domain server: San Jose, CA 95131 wayne@Ultra.COM 408-922-0100 ------------------------------ Date: Mon, 2 Nov 87 18:07:44 CST From: Dominique Petitpierre <petitp%cui.unige.ch@relay.cs.net> Subject: How to test the end of a 1/4" tape? How is it possible to test for the "end of tape" (ie after the last end of file mark) on a 1/4" tape? The only way I've found, is to try to write something (with, for example, 'mt -f /dev/nrst0 eof') at the current position, and check for an error message (which means that you are not at the end). It's not very nice since each time it is found, something is written at the end of the tape. And, furthermore, it is always possible to write something at the beginning of the tape, thus the test might erase information. So, since the driver knows when it has reached the last file, how can the user know without modifying the tape? I came to that problem, trying to write a small 'table of content' shell script that would give a rough idea on the information stored on a 1/4" tape (physical format (QIC 11,QIC 24), number and size of files, logical format (tar,cpio,etc)). If it matters, the workstation is a 3/50, with sun OS release 3.2 and we use the 'st' 1/4" tape driver (st(4), /dev/nrst[08]). Mr. Dominique Petitpierre ISSCO, University of Geneva 54 route des Acacias CH-1227 GENEVA (Switzerland) Tel: 0041/22/20 93 33 extension 2117 ean, BITNET, EARN, MHS, X.400: petitp@cui.unige.ch uucp: mcvax!cernvax!cui!petitp , petitp@cui.uucp JANET: petitp%cui.unige.ch@cs.ucl.ac.uk csnet, ARPA: petitp%cui.unige.ch@csnet-relay.csnet ------------------------------ Date: Mon, 2 Nov 87 11:03:32 PST From: <nsc!parns!harker@sun.com> Subject: Making 2400 baud modems work on SunOS 2.x? I have recently purchased a Desktalk II 2400 baud modem and was having problems with making it work under SunOS 2.4. It would work fine when connected directly to a terminal, but would fail when I attached it to my Sun. I had edited my /etc/remote file adding lines for a 2400 baud modem: UNIX-2400:\ :el=^D^U^C^S^Q^O@:du:at=at:ie=#$%:oe=^D:br#2400:tc=dialers: VMS-2400|TOPS20-2400:\ :el=^Z^U^C^S^Q^O:du:at=at:ie=$@:oe=^Z:br#2400:tc=dialers: adding a line to my /etc/gettytab file for a 2400 baud modem: 0|d2400|Dial-2400:nx=d1200:fd#1:sp#2400: and adding a line to my /etc/ttys file to set the tty line up for 2400 baud: 10ttyd0 But my new modem would not work. It would dial the number, I could hear the two modem tones, and then the modem would go quiet signaling that the modems has synchronized. But then tip would give me a "called failed" message. This had me very confused until I saw this message about problems with Suns and 2400 baud modems in the comp.unix.ultrix news group. > From: tim@okra.UUCP (Tim Wood) > Subject: Re: Problems with Tip on Ultrix 2.0 > > One bug I found with SunUNIX 'tip', possibly still existing, was that > it wouldn't communicate with Hayes-compatible 2400-baud (USR Courier). > The Courier had a default status code set that was different from > the Hayes 1200 one, so when it made a 2400-baud connect, it returned > a status code which tip couldn't recognize. Fix was/is to MANUALLY > enter the following whenever the modem is powered up (what you type > is preceded by a tab or prompt: > > blorp% tip cua0 > connected > ATX0 < this is not echoed > > 0 > ~. > disconnecting...[EOT] > blorp% > > ATX0 selects the ordinary 1200 status codes. Since I have done this I have been able to use my modem under SunOS 2.0. As a note this problem does not occur under SunOS 3.4. I am very happy with my Desktalk II modem. For a $200 modem it is quite nice, but ugly. I am having some problems with line noise, but this might only be a problem with my phone lines which I think go through our company phone switcher, although our phone department swears it is a AT&T phone line. Desktalk II 2400, $199, USA Flex (800) 872-3539 Features: Hayes clone, 5 year warranty, help screen, 11 LEDs, plastic case, 30 day money back guarantee. I hope this information is useful. Robert Harker, All around good guy. harker@parns.nsc.com {sun,decwrl,hplabs}!nsc!parns!harker ------------------------------ Date: 2 Nov 87 19:29:36 GMT From: Yorick Phoenix <mcvax!kbsc!yorick@uunet.uu.net> Subject: Serial Line IP on Sun III/160 using SunOS 3.4? I have in the past run the Serial Line Tcp-IP on my Sun III/160. This was added using patches to my Object Only Sun 3.0 and 3.2 Kernel from mark@mimsy. However, I have now upgraded to 3.4 [at last] and upon using the same set of patches I have discovered that an attempt to use slattach results in a great big hang. Is there anybody out there who has a set of patches to get Slip to work under 3.4 on an Object Only Kernel. Thanks, Yorick yorick@kbsc.UUCP ..mcvax!ukc!{idec,hrc63}!kbsc!yorick The Knowledge-Based Systems Center 58 Northside, Clapham Common LONDON SW4 9RZ England Voice: +44 1 350 1622 ------------------------------ Date: Tue, 3 Nov 87 08:34:16 PST From: calius@composite.stanford.edu (Emilio Calius) Subject: Adding CDC SMD drives to Sun-3? Help! If you have had some experience in hooking up CDC drives to a Sun, a bunch of novices would very much like to have your advice. CDC is now offering a new version of their 9720-515 (515 Mb un- formatted) 8 inch SMD drive at a very attractive price. ($4300 @ Kierulf Electronics in Sunnyvale, CA) We would like to add one to our Sun-3/160, presently configured with 1 Xylogics 450 (NOT 451) controller and 1 Fujitsu M2322 130 Mb drive. What do we have to watch out for when installing it & formatting with "diag" (assuming we get the heads/cylinders geometry from CDC) ? Will this drive work with our Xy-450 controller? What about the cabling (daisy-chain or radial) ? Any warnings/words of caution/great peals of laughter? Etc, etc.. Thank you. We really need to make a decision soon. Emilio P. Calius Structures & Composites Lab -- "We make it, you crash [in] it" Aeronautics & Astronautics Stanford University ------------------------------ Date: Wed, 14 Oct 87 16:23:18 edt From: decvax!genrad!jpn@ucbvax.berkeley.edu (John P. Nelson) Subject: ICON previewer The following is a handy little program to display all these ICONS being submitted to SUN-SPOTS. I am donating this to the public domain. /* dsplyicon.c 6-Jan-86 17:01 jpn Author * * Program to write an icon image on top of the current window * Useful for displaying icons without the overhead of starting * up an "icontool". Also displays icons with odd sizes ("icontool" * will only handle 64 x 64 * * Compile: cc -o dsplyicon dsplyicon.c -lsuntool -lsunwindow -lpixrect */ [[ The remainder of the source has been placed in the archives as "sun-source/dsplyicon.c". It can be retrieved via anonymous FTP from the host "titan.rice.edu" or via the archive server. For more information about the archive server, send a mail message containing the word "help" to the address "archive-server@rice.edu". -wnl ]] ------------------------------ Date: Tue 27 Oct 87 14:42:08-PST From: Alan R. Katz <KATZ@venera.isi.edu> Subject: yet another yet another news icon Here is another rn News icon (which I like better): /* Format_version=1, Width=64, Height=64, Depth=1, Valid_bits_per_item=16 */ 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x03FF,0xFFFF,0xFFFF,0xFF80,0x0000,0x0000,0x0000,0x0080, 0x0FFF,0xFFFF,0xFFFF,0xFE80,0x0000,0x0000,0x0000,0x0280, 0x3FFF,0xFFFF,0xFFFF,0xFA80,0x2000,0x0000,0x0000,0x0A80, 0x2000,0x0000,0x0000,0x0A80,0x2000,0x0000,0x0000,0x0A80, 0x2001,0x13E8,0x2700,0x0A80,0x2001,0x9208,0x2880,0x0A80, 0x2001,0x9208,0x2800,0x0A80,0x2001,0x5209,0x2400,0x0A80, 0x2001,0x53C5,0x4300,0x0A80,0x2001,0x3205,0x4080,0x0A80, 0x2001,0x3205,0x4880,0x0A80,0x2001,0x1202,0x8880,0x0A80, 0x2001,0x13E2,0x8700,0x0A80,0x2000,0x0000,0x0000,0x0A80, 0x2000,0x0000,0x0000,0x0A80,0x2222,0x21FF,0xF000,0x0A80, 0x2444,0x4000,0x01FF,0xEA80,0x2111,0x11FF,0xF000,0x0A80, 0x2088,0x8000,0x01FF,0xEA80,0x2222,0x21FF,0xF000,0x0A80, 0x2444,0x4000,0x01FF,0xEA80,0x2111,0x11FF,0xF000,0x0A80, 0x2088,0x8000,0x01FF,0xEA80,0x2222,0x21FF,0xF000,0x0A80, 0x2444,0x4000,0x01FF,0xEA80,0x2111,0x11FF,0xF000,0x0A80, 0x2000,0x0000,0x01FF,0xEA80,0x23FF,0xC1FF,0xF000,0x0A00, 0x2000,0x0000,0x01FF,0xEA00,0x2000,0x0000,0x0000,0x0800, 0x3FFF,0xFFFF,0xFFFF,0xF800,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0162,0xC000,0x0000, 0x0000,0x0193,0x2000,0x0000,0x0000,0x0102,0x2000,0x0000, 0x0000,0x0102,0x2000,0x0000,0x0000,0x0102,0x2000,0x0000, 0x0000,0x0102,0x2000,0x0000,0x0000,0x0102,0x2000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000 [[ In the archives as "sun-icons/ark-news.icon". --wnl ]] ------------------------------ End of SUN-Spots Digest ***********************