Sun-Spots-Request@RICE.EDU (William LeFebvre) (07/15/88)
SUN-SPOTS DIGEST Wednesday, 13 July 1988 Volume 6 : Issue 139 Today's Topics: Re: Error messages not appearing in console window on client Re: Problems with Apple Laser-writer and Sun Re: 4.0 vs kill -HUP 1 Re: Sun losing track of console Re: Resolver based gethostby* in 4.0 (2) vxWorks and dbxtool 4.0 vs kill -HUP 1 SunOS 4.0 on Sun2's Problems with CDC EMD (368M) drives Anyone have driver for Xylogics 772? missed optimization? Using a BITNET host as a gateway? 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 available via anonymous FTP from "titan.rice.edu". For volume X, issue Y, "get sun-spots/vXnY". They are also accessible through the archive server: mail the request "send sun-spots vXnY" to "archive-server@rice.edu" or mail the word "help" to the same address for more information. ---------------------------------------------------------------------- Date: Wed, 6 Jul 88 07:45:22 EDT From: Chuck Musciano <chuck@trantor.harris-atd.com> Subject: Re: Error messages not appearing in console window on client > Under suntools ... error messages ... which should > appear in the console window simply do not. Instead they appear > splattered across the console as if no console window has been opened. > Any ideas what might be wrong here? Sounds like you might be creating a console window, then ANOTHER console window (which steals the console attribute from the first window), and then killing the second console window, leaving no console window. Although the first console window is still around, it is no longer attached to the console, and the messages just spill to the screen. You may want to try "contool" (available from the archive server) which does not prevent this problem, but does make it clearer that there is exactly one console window which is not used for any other functions. Chuck Musciano Advanced Technology Department Harris Corporation (407) 727-6131 ARPA: chuck@trantor.harris-atd.com ------------------------------ Date: Wed, 6 Jul 88 07:58:51 EDT From: Chuck Musciano <chuck@trantor.harris-atd.com> Subject: Re: Problems with Apple Laser-writer and Sun > The laser-writer randomly produces a test page, like the ones that are > spit out when you first turn the sucker on. I have never seen this happen. That page only comes out when we power cycle the printer. > We are also use Valid Logic Systems software, and I know that that can > really screw up the laser-writer, because they didn't implement their > postscript drivers properly. (Hear that, Valid?????) Aha! We use "real" software, the TranScript driver from Sun. I believe you have found the bug. No offense, but I am surprised by the number of people who buy a LaserWriter (usually from Apple) but decline to get the drivers from Sun, and then wonder why they can't get the printer to do everything it's supposed to. We have several groups here at Harris who have LaserWriters, and they have hacked up drivers to support "line printer" mode for their LW. Thus, they lose troff, TeX, Frame Maker, capability. Spend the money, get the right stuff, and use the device as it was intended. Penny-wise and pound-foolish never wins. Chuck Musciano Advanced Technology Department Harris Corporation (407) 727-6131 ARPA: chuck@trantor.harris-atd.com ------------------------------ Date: Wed, 6 Jul 88 08:22:46 EDT From: reg%lti.UUCP@bu-it.bu.edu (Rick Genter x18) Subject: Re: 4.0 vs kill -HUP 1 Reference: v6n124 Careful here; changing speeds on a line and doing kill -HUP 1 has caused a new getty to appear on that line for a long time (possibly dating back to 4.1 BSD). In fact, I don't think it is just changing speeds; if the line is in any "significantly different" mode other than how getty likes it, init will start a new getty, or at least it used to. I used to run into this problem all the time if I had a terminal running an application in CBREAK or RAW mode and someone did a kill -HUP 1. I believe the CBREAK/RAW problem was fixed; line speed was not, however. - reg ------------------------------ Date: Wed, 6 Jul 88 03:39:56 EDT From: budd@bu-it.bu.edu Subject: Re: Sun losing track of console Phase-Of-Moon: LQ-0D,22H,0M,57S strings of /etc/init and the sources only come up with /dev/console. ------------------------------ Date: Wed, 6 Jul 88 12:31:47 EDT From: Root Boy Jim <rbj@cmr.icst.nbs.gov> Subject: Re: Resolver based gethostby* in 4.0 (1) ? From: libes@cme-durer.arpa (Don Libes) ? I mentioned previously that upon request Sun will give you a new sendmail ? (with MX), new yp, and appropriate libraries all using the resolver. ? ... ? The only drawback of all this is that if you don't run yp, this isn't a ? solution for you. I agree with everything Don said except for the last sentence. The nameserver kit also includes `nslookup' a program that translates host names to internet addresses, among other things. One merely needs to use nslookup to do this translation and then telnet or ftp to the appropriate internet address directly. Unfortunately, this leaves out rlogin, rcp, and rsh, since they use the host tables. But it would seem possible to add any `invisible' hosts to ones localhosts file. ? Don Libes cme-durer.arpa ...!uunet!cme-durer!libes (Root Boy) Jim Cottrell <rbj@icst-cmr.arpa> National Bureau of Standards Flamer's Hotline: (301) 975-5688 ------------------------------ Date: Wed, 6 Jul 88 12:39:34 EDT From: Root Boy Jim <rbj@cmr.icst.nbs.gov> Subject: Re: Resolver based gethostby* in 4.0 (2) Reference: v6n108 ? From: franz!frisky!jkf@ucbarpa.berkeley.edu (John Foderaro) ? Beware, the 'struct hostent' (netdb.h) structure has changed form in 4.0. ? The last field is now points to a vector of addresses, not just one ? address.... Not only that, but the format of addresses that are returned is different. The old addresses were returned in printable form, i.e. `129.6.48.63', with a length of (in this case) 11, while the new addresses are returned as the bytes (hex) 81 06 30 3f, with a length of four. Actually, I feel this latter format is more useful, since it is easier to convert binary to ascii than verse vica. ? [[ ...As for recompiling ? software: it is always a good idea to recompile as much as you can after ? installing a major system upgrade (such as 3.x to 4.0). What other ? include files might have changed? --wnl ]] We live in interesting times, don't we? :-) (Root Boy) Jim Cottrell <rbj@icst-cmr.arpa> National Bureau of Standards Flamer's Hotline: (301) 975-5688 ------------------------------ Date: 5 Jul 88 15:27:52 CDT (Tue) From: texsun!killer!linimon@sun.com (Mark Linimon) Subject: vxWorks and dbxtool When you are running vxWorks on your real-time target, and granted that you compiled your vxWorks board support package for that CPU with the INCLUDE_DBX flag turned on, then you can use a special version of dbxtool (or dbx, really if you want), called dbxWorks. Actually you order it from Sun Consulting as dbxWorks, but it installs on top of your old dbxtool, of which it is a complete superset. two commands are added: vxdebug <vxhost> [ <function> [<kernel>]] which reloads and restarts dbx in vxWorks debugging mode; vxload <objfile> which loads the specified file into vxWorks, updated the on-board symbol table as it goes. Some of the other commands change a little bit. Since I do very little application work I don't use it much myself, but customers go slackjawed when they see demos. I guess that after seeing how solid vxWorks is I just expected it to work :-). BTW, you need either vxWorks 3.21 or 4.0 (the latter currently "just out"), the support in 3.20 was too rough. I claim to be totally biased both in favor of VxWorks and Mizar. So be it. Mark Linimon Mizar Digital Systems sun!convex!mizarvme!linimon killer!linimon ;guest account ------------------------------ Date: Wed, 6 Jul 88 12:47:08 EDT From: Root Boy Jim <rbj@cmr.icst.nbs.gov> Subject: 4.0 vs kill -HUP 1 ? From: Jean-Francois Lamy <lamy@ai.toronto.edu> ? ... ? Repeat-by: edit /etc/ttytab, change your terminal entry to some other ? speed and kill -HUP 1. You should get kicked out. That's right, you should. Your two statements are inconsistent. What happens if you change nothing? RTFM! Any change should kick you off. As to what is desirable, it is arguable that logged in users should never be kicked off, that init should merely remember the change for the next getty, but that's how init works, and always has. (Root Boy) Jim Cottrell <rbj@icst-cmr.arpa> National Bureau of Standards Flamer's Hotline: (301) 975-5688 ------------------------------ Date: Wed, 6 Jul 88 10:07:42 PDT From: stevo@jane.jpl.nasa.gov (Steve Groom) Subject: SunOS 4.0 on Sun2's We too have had great difficulty in getting SunOS 4.0 to run on 'elroy,' a Sun 2/170 with the old ('non-prime') CPU board. It seems that there is a bug in the *kernel* (not the hardware) that when reading some bus-related register neglects to mask out some unwanted bits. These bits, when seen as set, make the kernel very unhappy. The problem is identifiable as a bus error when trying to boot either MUNIX or the miniroot like the one given as an example by nosun!cvedc!opus!markh@sun.com (Mark Holm): dump on ns0b fstype spec : trap address 0x8,pid 1,pc=470e,sr=2004,stkfmt 8,context 0 Bus Error Reg98<VALID,PROTERR> access .... (ommitted for brevity) traceback ... (ditto) End traceback Panic: Bus error The problem is that the 701 jumper on the CPU board controls a bit in the register, and when the jumper is removed, the bit is set and the kernel gets upset. Installing the jumper turns the bit off, and the kernel is happy. My old Sun 2/170 manual (dated March 1984 - yes it's an oldie, but so is elroy) says that this jumper controls how the CPU handles the Multibus CBRQ/ line. If all devices on the Multibus support CBRQ/ correctly, the jumper does not need to be installed. If not, it must be installed so devices that don't support CBRQ/ can get the bus. When it is installed, the CPU gives up the Multibus after every bus cycle. This configuration also causes three additional wait states for every Multibus access. Go ahead, read those last two sentences again. I didn't believe them either, but that's what the manual says. We didn't have the jumper in before (we didn't need it), but we have to have it in now, or nothing will run. Elroy has been running fine for more than 4 1/2 years, with only minor problems with peripherals, but not with the CPU. It is slow compared to the Sun3's and 4's that we now have, but when I came to work here in 1986 it was our group's only system and supported 15 (or so) users. For the past year and a half it has bee devoted mostly to news/mail/networking, not interactive use, but performance was still OK. Elroy's performance under 4.0 is abysmal. It is running VEEERRRYYYY SLLLLOOOOOOWWWWLLLLYYYY since the upgrade. We have at present two theories about why. The first is the jumper problem, over which we have no control, and can only wait for a fix from Sun. The second possibility is that elroy just doesn't have enough memory to run 4.0 comfortably. Elroy has the 4MB that Sun says is a minimum, but we see *a lot* of paging and swapping activity that wasn't there under 3.4. We have plans to borrow some memory and maybe a CPU from a neighbor and try this out. However this could pose even more problems for us, because our older version of the CPU (or maybe it's the PROMs?) has fairly low upper limit on the amount of memory it supports, something like 6MB. Then again, the increased paging and swapping might not be so much of a problem if we didn't have to pay such a high (newly added) penalty for using the Multibus! COME ON SUN! FIX THE STUPID KERNEL! I DON'T WANT THAT JUMPER IN! -steve /* Steve Groom, MS 168-522, Jet Propulsion Laboratory, Pasadena, CA 91109 * Internet: stevo@elroy.jpl.nasa.gov UUCP: {ames,cit-vax}!elroy!stevo * Disclaimer: (thick German accent) "I know noothingg! Noothingg!" */ ------------------------------ Date: Wed, 6 Jul 88 13:08:44 EDT From: Root Boy Jim <rbj@cmr.icst.nbs.gov> Subject: Problems with CDC EMD (368M) drives We have a bit different configuration(s). We have an early 3/160, with an XY450 and an Eagle, and it works fine. We are running with 46 sectors per track, using no slipping. We have a newer 3/180 with an XY451, also with an Eagle (both eagles are the small ones, 2351's), using 47 sectors per track, with slipping enabled. Some time ago, the latter system started flaking out, with random errors. We found that the 451 we had was not up to the current rev. I forget the exact numbers (and I'm not going to shut down the system and pull the board just for you :-), but you should call Xylogics and ask them exactly what you need. I will hazard a guess that our version says Rev F2, and another tag says something like 451009. The 451 is obvious, and the `9' represents the specific mod that Xylogics made so that the controller would handle less than 600 bytes per sector (I think it works out to 594 with 47 sectors). In short, I was a little miffed that Sun sold us a configuration that really wouldn't do the job. I felt we should have had the board replaced for free. Fortunately, the repair took two days to diagnose, so we got the seven day rate rather than the one day rate. I also talked the FE into not charging us for the kluge board, just the controller, so that the total bill was something like $1500. Your tax dollars at work. ? Andy Wilcox (ajw@beach.cis.ufl.edu): ? If anybody has this drive working, I'd love to hear from you. Good luck, and let us all know what happens. It is possible that if you run the drive without slipping, that the problems will go away. Let us all know how you make out. Both machines were running 3.4 at the time,, and now run 3.5 with no problems. (Root Boy) Jim Cottrell <rbj@icst-cmr.arpa> National Bureau of Standards Flamer's Hotline: (301) 975-5688 ------------------------------ Date: Wed, 06 Jul 88 09:47:21 -0400 From: erc <@vb.cc.cmu.edu,@GOLD.SCRIBE.COM:erc@astatine> Subject: Anyone have driver for Xylogics 772? Has anyone out there written a driver for the Xylogics 772 tape controller? I am currently using SunOS 3.5 on a Sun 3. - Eric Crane Scribe Systems ------------------------------ Date: Wed, 06 Jul 88 05:24:12 -0400 From: Ken Yap <ken@cs.rochester.edu> Subject: missed optimization? On my Sun 3/50, running SunOS 3.4, I compiled a short C program to assembly, with and without -O. The program is not important. Without -O the generated prologue for a function was: link a6,#0 addl #-LF39,sp With -O, this became: link a6,#-24 (LF39 =24, the size of the local frame.) My question is, why didn't the compiler generate: link a6,#-LF39 right away, even without -O? Surely it doesn't take lookahead to collapse the link and the allocate instructions, since the assembler is taking care for the forward reference? Ken ------------------------------ Date: Wed, 6 Jul 88 17:59 U From: <VIJAY@ITIVAX.BITNET> Subject: Using a BITNET host as a gateway? I have had previous success in soliciting help from Sunspotters about e-mail problems so here goes another try... My Suns sit on a LAN on which is a Vax that is our Bitnet node. We have Wollongong's TCP/IP package, Gmail and Jnet software on the Vax. We are able to exchange mail between the Suns and the Vax via the TCP mailers on each side. Now for the interesting bit - how can I arrange things so that I can use the Vax as a gateway to the wide world? No one here has any idea how we might be able to do this. I am optimistic because I am sure that others have faced and licked this particular dragon :-) Please save me from having to log on to this yuk Vax every time I want to talk to the world. ADVdankeANCE. ------------------------------ End of SUN-Spots Digest ***********************