Sun-Spots-Request@RICE.EDU (William LeFebvre) (03/18/88)
SUN-SPOTS DIGEST Thursday, 17 March 1988 Volume 6 : Issue 31 Today's Topics: Re: Capacity of CDC Wren IV, diag/format problem Re: screendump from SUN 3/110 to laserprinter; greyscale 'rexd' is 'on' something Program to determine console type Sun's (lack of) support for INTERPHASE disk controllers Problems with VME bus timeouts: Ironics board on a Sun-3 Can't make a xylogics 450 and interphase controller live together Problems configuring cua0 on a 3/160 zs0 Questions about Sun Shoeboxes Grant Accounting on the SUN? Send contributions to: sun-spots@rice.edu Send subscription add/delete requests to: sun-spots-request@rice.edu Bitnet readers can subscribe directly with the CMS command: TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name Recent backissues are stored on "titan.rice.edu". For volume X, issue Y, "get sun-spots/vXnY". They are also accessible through the archive server: mail the word "help" to "archive-server@rice.edu". ---------------------------------------------------------------------- Date: Thu, 3 Mar 88 13:43:10 EST From: Root Boy Jim <rbj@icst-cmr.arpa> Subject: Re: Capacity of CDC Wren IV, diag/format problem Reference: v6n19 > From: Hans van Staveren <mcvax!cs.vu.nl!sater@uunet.uu.net> > We recently acquired a CDC Wren IV diskdrive and put it to work.... > Does anyone know the nitty gritty details about this? No, but I'll take a guess. Most likely the 307 refers to the `unformatted' capacity. Formatted capacity is often quite a bit (~20%) less. You seem to have gotten 285/307 or 92% out of it, which is higher than normal. There are other ways to lose space as well, even if the 307 truly represents the formatted capacity. Our Eagles on a VAX running 4.3 BSD have a geometry of 842 x 20 x 48. On a SUN 3 and a Sequent they only use 46 sectors, losing 5%. On another SUN 3 with sector slipping, they actually have 47 sectors, but you still only get 46. Creating file systems also eats space for inodes, cylinder groups, super blocks, etc. on the order of 7%. Further, the last 10% of real usable data blocks are not doled out to anyone but the superuser. Unless you can be more specific I wouldn't worry about it. Just think of it as another kind of tax :-) (Root Boy) Jim Cottrell <rbj@icst-cmr.arpa> National Bureau of Standards Flamer's Hotline: (301) 975-5688 ------------------------------ Date: Thu, 3 Mar 88 16:28:35 EST From: Root Boy Jim <rbj@icst-cmr.arpa> Subject: Re: screendump from SUN 3/110 to laserprinter; greyscale Reference: v6n19 > Are there laserprinters which can do greyscale? The Apple Laserwriter does greyscale. The is a Postscript command called `setgray' which takes a floating point number between zero and one, and interprets it in the obvious way. This is the SUN's standard laser printer, so I would imagine there is a way of producing screen dumps on it. Good luck. (Root Boy) Jim Cottrell <rbj@icst-cmr.arpa> National Bureau of Standards Flamer's Hotline: (301) 975-5688 YOW!!! I am having fun!!! ------------------------------ Date: Thu, 3 Mar 88 15:21:40 cst From: grunwald@m.cs.uiuc.edu (Dirk Grunwald) Subject: 'rexd' is 'on' something This is a summary of a problem which hit us early in February. We have 2 3/50's (Oboe & Melodian) and a 3/280 (Guitar) which is both a workstation & a server. We have modified the Sun NFS to allow root access across the workstations, primarily to allow us to read mail on any machine. One fateful day, I was sitting around compiling & said ``I'd like to use the 'on' command to use Oboe & Melodian''. Would that I had never done this. I set up the entry in services & everything worked fine -- I could do compiles on the 3/50's and life was happy. Until, of course, the day when I found 80% of our file system missing. It turns out that 'rexd', which is used by 'on', temporarily mounts file systems in /tmp. E.g. if I was in /foo/bar on Guitar, and said: % on oboe make then /foo is mounted at /tmp/rexdXXXXXX on Oboe, and the commands are executed from that mount point (i.e. /tmp/rexdXXXXX/bar, in this case). This occurs even if /foo is already mounted on Oboe. Now, this isn't a symlink, it's a mount point, which appears to be very different. The most important difference is that ``find'' traverses mount points. The same ``find'' which is used in /usr/lib/crontab to delete files older than 3 days. Now, in our case, 'rexd' did not remove those directories, when when the cleanup of /tmp occurred at 3:30am, our (largest) partition was also cleared. However, even in the case of rexd working correctly, if you were so foolish as to be using the 'on' command at 3:30, you'd be opening yourself to having your files wiped out. The same problem occurs, although to a lesser extent, if you don't have NFS set up for common-root. In that case, you'd just delete files with 'other' access instead of all files. However, it's still a problem. Two solutions to this are: 1) Don't use 'on' or 'rexd' 2) Have 'rexd' mount partitions somewhere else /rexd/rexdXXXX? We're doing the former. Dirk Grunwald Univ. of Illinois grunwald@m.cs.uiuc.edu ------------------------------ Date: Wed, 2 Mar 88 14:50:26 PST From: Doug Moran <moran@ai.sri.com> Subject: Program to determine console type There have been several requests in this digest for programs to determine the type of the console. Below is a program that I wrote to produce an identifying string (I start suntools in my .login file and use this pgm to determine which arguments to use). Caveat: my cluster has only a few of these monitor types, so the pgm has not been fully tested. Note on coding style: the function wu_fbid is actually located in a local library, accounting for what otherwise might appear to be a strange coding style. __________ main() { char *wu_fbid(); char *consoleid = wu_fbid("/dev/fb"); if ( consoleid == 0 ) write(1, "tty", 3); else write(1, consoleid, 3); write(1, "\n", 1); } #include <sys/ioctl.h> #include <sys/file.h> #include <sun/fbio.h> /* decoding as of Release 3.4 : fbio.h 1.3 87/01/09 SMI */ /* the convention for entries in this table is to translate the * macros for frame buffer codes (in <sun/fbio.h>) to short names * thus: * FBTYPE_SUNxBW becomes bwx * FBTYPE_SUNxCOLOR becomes cgx * FBTYPE_SUNxGP becomes gpx * FBTYPE_NOTSUN[1-9] becomes ns[A-J] */ static char *decode_fb[FBTYPE_LASTPLUSONE] = { "bw1", "cg1", "bw2", "cg2", "gp2", "bw3", "cg3", "bw4", "cg4", "nsA", "nsB", "nsC" /* Not Sun */ }; char * wu_fbid(fbname) char *fbname; { struct fbgattr fbattr; int fd, ioctl_ret; if ( (fd = open(fbname, O_RDWR, 0)) == -1 ) return(0); /* FBIOGATTR fails for early frame buffer types */ if (ioctl_ret = ioctl(fd,FBIOGATTR,&fbattr)) { /*success=>0(false)*/ ioctl_ret = ioctl(fd, FBIOGTYPE, &fbattr.fbtype); } close(fd); if ( ioctl_ret == -1 ) return(0); /* The binary is obsolete and needs to be re-compiled: * the ioctl returned a value beyond what was possible * when the program was compiled */ if (fbattr.fbtype.fb_type>=FBTYPE_LASTPLUSONE) return("unk"); /* The source is obsolete. The table "decode_fb" does not * have entries for some of the values returned by the ioctl. * Compare <sun/fbio.h> to the entries in "decode_fb" */ if ( decode_fb[fbattr.fbtype.fb_type] == 0 ) /* decode_fb is obs */ return("unk"); return(decode_fb[fbattr.fbtype.fb_type]); } ------------------------------ Date: Wed, 2 Mar 88 09:42:33 PST From: mordor!tolerant!vsi1!tim@sally.utexas.edu (Tim Richardson) Subject: Sun's (lack of) support for INTERPHASE disk controllers Is SUN going to support the INTERPHASE model 4200 disk controller. In our opinion, the 4200 is the best disk controller out there. What I don't care for is the apparent willingness of SUN to support one vendors equipment and a complete unwillingness to support another vendors equipment. This is inconsistent with the stated philosphy of having an "open architecture" and third party support, which is really hyped by the SUN sales dept. The next item is the INTERPHASE model 4201 with SCSI port. This controller is an ESDI (4 ported) controller which has an option (extra cost) as the SCSI port. Since you need the ability to boot from the 1/4inch tape, this board will eliminate the need to take up two backplane slots (1 for the disk controller and another for the SCSIbus Host Adaptor.) The problem right now is that SUN won't allow third party vendors to link their own boot device code to the SUN3/xxx CPU's boot EPROM. Can we get a resolution to this problem. Anyone who wants to integrate their own hardware, or be an OEM of SUN's, has these kind of problems. Tim Richardson VICOM Systems Inc. sun!pyramid----\ 2520 Junction Ave. uunet!ubvax-----!vsi1!tim San Jose, CA 95134 ucbvax!tolerant/ +1-408-432-8660 ------------------------------ Date: 3 Mar 88 19:10:21 GMT From: wiley!bob@uunet.uu.net (Bob Amstadt) Subject: Problems with VME bus timeouts: Ironics board on a Sun-3 I've been trying to use an Ironics IV-3272 Data Transporter board in a Sun 3/180. This board is a DMA engine that moves data from one place to another. We want to use it to move data between two custom VME boards. It's advantage is that it moves data faster than the Sun CPU. There is one problem though. The Ironics board does not always release the bus as quickly as the Sun CPU expects. Occasionally when doing transfers using the Ironics board, Sun UNIX crashes with the reason, "BUS TIMEOUT." Has anyone run into a similar problem? Can this timeout be changed or disabled either through software or hardware? ---Bob Amstadt bob@wiley.uucp {csvax.caltech.edu,uunet.uu.net,trwrb.uucp}!wiley!bob ------------------------------ Date: Thu, 3 Mar 88 14:13:05 CST From: aslam@p.cs.uiuc.edu (Sohail Aslam) Subject: Can't make a xylogics 450 and interphase controller live together I have a xylogics 450 controller managing a fujitsu eagle M2351 in a SUN 2/170 server. I got hold of an old fujitsu M2284 disk (128Mb) along with an interphase controller. Xylogics lives at address 0xee40. Following the configuration scheme I put the interphase controller at 0x40 and built a new kernel. Upon rebooting, the boot procedure tried to boot off of the interphase controller which is not what I wanted. I tried moving the xylogics to xyc1 at mbio ? csr 0xee48, the system correctly booted off of xy1(1,0,0) but then it complained that there is an unknown controller at 0x40. Is there a problem with my hardware or has someone else run into problems putting a xylogics 450 and an interphase controller in the same box? I am running SunOS 3.5. Adva-thanks-nce Sohail Aslam Department of Computer Science University of Illinois arpa aslam@a.cs.uiuc.edu csnet aslam@uiuc.csnet usenet {ihnp4,seismo}!uiucdcs!aslam ------------------------------ Date: Thu 3 Mar 88 11:59:02-EST From: Michael van Biema <MICHAEL@cs.columbia.edu> Subject: Problems configuring cua0 on a 3/160 zs0 I've been having trouble setting up /dev/cua0 and /dev/ttyd0 on a Sun 3/160 zs0 port 1. I've done it before under nearly identical circumstances, and given how simple it was last time, and given the step-by-step instructions in the documentation, I'm somewhat embarrassed by having to ask the net for help. However... As far as *I* can tell: /etc/ttys is correct (13ttyd0 and 02ttya -- don't need cua0 right?). /etc/gettytab is correct (standard entries for 2 and 3). kernel flags for zs0 are correct (0x2 since I'm using port A). major and minor numbers on cua0 and ttyd0 are correct (12 128 & 12 0). dip switches on the Hayes Smartmodem 1200 are correct (as per Sun doc). cable is correct (in fact all 25 pins, straight through). and of course ttya has been banished from dev. But I've obviously missed something because: I can dial out (with kermit or tip) as many times as I like, with no problem. I can dial in as many times as I like, with no problem. On ttyd0: init turns to getty getty turns to login login turns to sh sh dies (on exit) and a new init (with a new pid) starts This is the way it's supposed to happen, right? However, once I dial in (log in and then exit), I can't dial out without first: Killing the new init process that started on ttyd0, and thus causing init (pid 1) to start another one on ttyd0 -or- momentarily disconnecting the RS232 cable -or- power-cycling the modem -or- momentarily driving DTR or DCD high (negative voltage, binary one). Note that the modem seems to do this itself as it disconnects. Why should I have to do it again manually? If I don't do one of the above, perror says "no such device or address" after trying to open /dev/cua0. open returns ok after I do one of the above, or if I haven't dialed in yet to begin with. I assume that among the things I've supposedly configured correctly, there is something that I've overlooked or just plain screwed up, but maybe I've just missed something completely. If anybody has any suggestions towards either end, I'd like to hear 'em. -Thanx Michael MICHAEL@CS.COLUMBIA.EDU (ARPANet) MICHAEL@CS.COLUMBIA.EDU@CSNET-RELAY (CSNet) MICHAEL%CS.COLUMBIA.EDU@CUNYVM (BITNet) ...rutgers!columbia!cs!michael (USENet) ------------------------------ Date: Thu, 03 Mar 88 12:16:24 EST From: williams%nrl-css.arpa@nrl-css.arpa Subject: Questions about Sun Shoeboxes We have a few of the Sun shoebox drives around here, and are planning to get more. In order to figure out what we really want, I need some information about the insides of these boxes. If anyone has dug into these things and can answer any of the questions below, I'd really appreciate it. First, I am working from the following assumptions/facts. Sun offers two types of shoebox: the Shoebox Subsystem (SS), which has a disk drive (ours are 141MByte) and an optional cartridge tape drive, and the Expansion Shoebox (ES), which is a disk only. (These may not be the exact names Sun uses, but that doesn't matter here.) The questions are: What SCSI addresses are used for the SS disk, the SS tape, and the ES disk? Where are the SCSI bus terminators, if any? It the SS? the ES? both?? Can the SCSI address(es) of an SS be changed easily, so that the drive in an SS can be the second SCSI drive (sd1[a-h])? One reason we want to do this is so we can make a "master SCSI disk", then use that to initialize a new (type SS) SCSI disk. By attaching the new disk to a Sun as sd1, with the master disk as sd0, we could simply dd the old disk onto the new one. Would modifying the SCSI addresses as suggested above void the warranty on the shoebox? Can the Expansion Shoebox be used as sd0, by modifying the SCSI address, and removing/adding a bus terminator? Much thanks in advance, Jim Williams James W. Williams williams@nrl-css.arpa Information Technology Division Naval Research Laboratory (202) 767-9035 Washington, DC ------------------------------ Date: Thu, 3 Mar 88 07:48 EST From: <STELLABO@CSHLAB.BITNET> Subject: Grant Accounting on the SUN? We are trying to bring our administrative functions on to the SUN computers to standardize all our computing. Our major stumbling point is an accounting package. We are a non-profit research laboratory, with funding from the NIH and private sources. We need a GRANTs management package that will run on a SUN-3. This package should be able to: 1) Process Purchase Orders with immediate Grant incumbrance. 2) Process Payroll grant incumbrances. 3) General ledger and sub legder accounting. 4) Grant information management. 5) and ... Inventory (not critical) Does anyone know of such a package.. ??? If so please contact me at: Fred J. Stellabotte Computer Systems Manager Cold Spring Harbor Laboratory stellabo@cshlab (516) 367-8420 ------------------------------ End of SUN-Spots Digest ***********************