rose@galtee.cs.wisc.edu (Sullen and despondent) (05/29/91)
Coherent Digest Vol. 91.2, No. 11 Tue May 28 00:32:59 CDT 1991 Today's Topics: Digested MWC Tech Support Traffic Screen Blanker Screen Blanking RE: screen blanker Administrivia: The Coherent Digest is a forum for discussion of the MWC Coherent 3.x operating system. Send submissions to "coherent@cs.wisc.edu" and administrivia to "coherent-reqeust@cs.wisc.edu". Previous issues are archived for anonymous FTP access on piggy.ucsb.edu in the directory "pub/coherent/mail-list". ---------------------------------------------------------------------- Date: 21 May 91 16:11:14 From: mwcbbs!root@uunet.UU.NET (MWCBBS root (Bob Hemedinger)) To: dalek!mwc!cohmail@uunet.UU.NET Subject: Digested MWC Tech Support Traffic This is a digest of electronic support recently transacted at the Mark Williams Company for COHERENT users. Volume # 1 Start Date: 05/04/91 Today's topics: clearing inodes uucico NEVER calls crypt() function bug device driver info bootstrap usage cflow port jackpot! kermit vs. uucp ATI EGA Wonder board crontab ide info ramdisks Adaptec vs. fsck MCI codes in L.sys removing master bootstrap 1144a parameters accidental shutdown WD1007sv2 jumpers CoMb bug ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >>Date: 4 May 91 21:25:29 >>From: dalek!mwc!sevenup@netcom.COM >>To: mwcbbs!support >>Subject: I-nodes,uucp mail test >> This is a uucp mail test for the most part but I just got a new boot >>message and was hoping you could shed some light the correction procedure. The >>message says "I-node 1323 this is a multiply addressed I-node" or something to >>that affect. It is not one of the error messages listed in the Lexicon and when >>I try to run icheck I get a message that it can't open a /dev file?? Generally a corrupted tar archive can cause this problem when you attempt to untar it -or- untarring a file with our tar which was tarred with ustar will also cause this problem. This results in recursive directories that cannot be removed with an ordinary rm -r command. You will have to salvage the files from this directory and then use the clri command to clear the inode. After using clri, shut down and reboot the system. Fsck will ask if you want this inode removed, tell it yes. It will more than likely have to rebuild the free inode list and some of the superblock info and reboot itself again before the indoe is gone for good. >>More new and improved uucico bugs for you... >> >>L.sys says: >> >>mwcbbs Never ACU 2400 17085590412 ... >> >>and >> >>uucico -Smwcbbs >> >>does not place the call !!! It is my understanding that the -S (capital s) >>should force the call, even if L.sys says we're not in the call window. >>The old uucico did it correctly. What's happening ??? Under the old uucico, this was a *bug*. When specifying Never in L.sys, the site should *Never* be called. You will have to place a valid time in your L.sys to work around this, perhaps a ridiculous time when you know that your system will not be running so that you will not have to worry about a call being placed. >>Other questions: >> >>Where's the 386 version. Getting tired of small model C... The more the 386 version comes along, the less we'll actually be able to discuss it. The development of the 386 is 'on schedule' as far as we are concerned, but I can not guarantee any delivery time for it If we give you a date for the release, the first thing we'll wind up doing is breaking it. >>What about math coprocessor support (80387). I know C does not support >>it, but are coprocessor status and registers saved across task switches ? >> >>We need ANSI C real bad. Any plans ? ANSI C is currently planned for the 386 version. >>Message-Id: <9105072350.AA91.V2.20.HRAlcorn@hra.UUCP> >>Date: 7 May 91 23:50:54 >>From: HRAlcorn!herb (Herbert R. Alcorn) >>To: mwcbbs!admin >>/* TO: Doug Peterson, MWC Admin. >> * >> * FROM: Herbert R. Alcorn >> * 118 Andover >> * Victoria, Texas 77904 >> * (512) 578-5430 >> * >> * Ver: Coherent Ver 3.1.0, Serial No. 246414609 >> * >> * Re: Problem with updated crypt () function. >> * >> * Description of problem: The following program accepts a user-id number >> * form stdin, looks up the encrypted password (based on UID) >> * from /etc/passwd; then requests the user to enter the matching >> * password, which after crypting should match the entry from the >> * password file. I replaced the file /lib/libc.a with the bbs >> * file libc310c.a.Z (after decompression). The following program >> * does NOT work with the new version of crypt.o; however, I used >> * ar to extract a copy fo crypt.o from the version of libc.a that >> * was distributed with version 3.1.0, and it works just fine! >> * >> * Conclusion: Somebody or something broke the function crypt () between >> * distribution of version 3.1.0 and the update libc310c.a.Z! Try the >> * program below with both! that is correct, the crypt function IS broken in the libc.a posted to mwcbbs. This has been corrected and will post it as soon as I can get a go ahead on it. Watch the ~/mwcnews/news file for the announcement. Date: 9 May 91 12:31:21 From: norm@mwc.com (Norman Bartek) To: hal digest Subject: re: Device driver question The magic number of 138 is actually 10 + 128 which equals the number of "direct blocks" in the file's inode, plus the number of "single indirect" blocks which can be addressed. Due to size constraints on the bootstrap (lots of stuff in a 510 byte program ;-) the bootstrap can only handle direct and single indirect blocks. Double (and tripple) indirect would have taken more code than could fit, sad to say. We're looking at redoing the bootstrap and adding a bit of functionality to it as well. Support for double-indirect blocks is a must, especially with our 386 COHERENT product which will be too large to boot using our traditional bootstrap. >>Subject: Boot >>Bob, >> What is the best way to boot both or either Dos and Coherent from >>two drives? i.e. drive 1 is a DOS drive and drive #2 is Coherent. >>I want (need) to get more space so I am tring to get another drive (IDE) >>I want to put Coherent on one drive and Dos on the other. But I want to have the option of selecting Coherent or Dos. But, I will be going to another drive not a partition... The *best* way that I have found is to have no partition active so that the system stops and waits for a partition to be selected. I highly recommend that the 2nd drive be split into at least 2 partitions, it will aid in backing up and restoring files. >>Do you have any idae yet as the the space needed for Coherent 386???? No details for the 386 requirements yet, save that you'll need a 386(sx). >>what about the file I up to you is it ok to send the rest or did I break the >>Copywrite??? The file looked okay to me, and its posted in the downloads/toolbox directory. >>Date: 10 May 91 06:47:32 >>From: hornlo!loh (Larry Olin Horn) >>To: mwcbbs!support loh >>Subject: Jackpot, new uucp pkg >>1) the update to uucp now delivers mail properly; thanks! >>2) I grep'd fortunes and many other files -- no sign of "Jackpot" >> I'll certainly let you know if I ever see it again, but I'm hoping >> that the various updates have taken care of it Surprisingly enough, I was 'diffing' some files later that day and lo and behold, a whole series of 'jackpot' messages appeared on my console. My first thought was that it was a virus of some kind, luckily a programmer rounded the corner by my cubicle and basically said 'oh, that from the diff command'. Apparently some debugging statements are still lingering in the code for diff. RCS (at least a version I run) can also produce these messages. >>----- hornlo!loh ----- support info ------------------------------------------- >>Coherent version: V3.1.0U (plus downloaded udpates) / Serial Number: 142922804 >>equip: Zenith 386/20, 8MB memory, ESDI disks, VGA, no math coprocessor >>home: Larry Horn / 1718 Hazel Street / Jackson, MS 39202-1118 / (601) 355-7721 >>work: Millsaps College / 1701 N State St. / Jackson, MS 39202 / (601) 974-1142 >>From: duncan!david (David John Murdoch) >>To: mwcbbs!support >>Subject: new kermit problems etc >> >>The new version of kermit seems to work ok, however, >>after using it UUCP(old version) does not work. The chat >>sequence does not copy to the debug file (uucico ... -x7 2>debug) >>after the line >>... >>calling phone# 123-4567 >> >>and uulog shows dial failed ACU problem. >> >>After rebooting the system everything returns to normal >>until, you guessed it, kermit is used again. >>I've tried to access the modem configuration with at&v >>in kermit, to see if any settings are being changed, >>but this command will not work although other >>at commands will work such as atl. I can dial out to >>other sites requiring manual login using atdt7654321 >>but no connection is made. The attention command AT >>the escape code +++ and ATE1 are ignored in kermit except >>just after rebooting. >>Do you know what is happening? >>I'm willing to experiment further but could you suggest >>some things I could try? The symptoms appear to reflect the presence of a LCK file in /usr/spool uucp. Check for this after using ckermit. Xcmalt will produce these files, so I suspect that ckermit may be doing the same thing. Some other things may be happening as well... modem register settings changing and/or the serial port latching up. Are you running an updated kernel? Also, use a POLLED port to see if the behavior changes. Resetting your computer would most certainly reset the serial ports which would be a logical explanation, at the same time, however, you modem would also do a reset, and depending on its register settings, may be defaulting to something other than what you expect. >>Also, I edited /etc/rc so that /usr/bin/modeminit is read >>(I thought I could access the modem that way) >>and the Going multiuser... hangs up Remove the references to the enable and disable commands in the modeminit script. A race condition is being created because of their presence, OR, put modeminit in the background from the rc file. Either way will clear up the problem. >>: /usr/bin/modeminit 9/7/90 >>: Initialize modem, typically called from /etc/rc. >>: This example initializes a Hayes-compatible modem for echo off, autoanswer. >>: The sleep is required to keep the desired baud rate for the cat command. >>DEV=/dev/modem >>BAUD=2400 >>sleep 2 >$DEV& >>stty $BAUD >$DEV >>cat >$DEV <<\! >>AT E0 S0=1 >>When I comment out /usr/bin/modeminit in /etc/rc Going multiuser... Date: 9 May 91 14:29:25 From: htrails!root (Superuser) To: digest Subject: ATI EGA Wonder Card If used with a monochrome monitor, the following dip switches must be ON, 3 & 6. All the rest are off. If this is not done, the cursor does not move. Date: 9 May 91 14:30:42 From: htrails!root (Superuser) To: digest Subject; To make crontab run a command every minute, * * * * * date >/dev/console Set permissions to 777 for /dev/console while running as root. If you log into /dev/console, init changes the permissions, so you may have to set the permissions as part of your .profile or by hand each time. The alternative if you don't have a root passwd is to enter the following, * * * * * su root /usr/bin/shellscriptname which would run the script as root and the script could do output to /dev/console even if the permissions didn't allow it. From: htrails!doug (Douglas Peterson) To: support cc: norm Subject: Problem IDE drive A customer with a Fujitsu 90mb IDE drive, was having installation trouble. She consistantly got AT0: T0 messages when trying to boot the install disk. She HAD a drive type 47 with April 90 AMI Bios. She changed her cmos drive type to 27 667 cyls, 4 heads, 33 sec per track. The type 47 that failed had 1334 cyl, 8 heads, 33 sec per track. The type 27 configuration works but you lose 20 mb of drive. However, using Disk Manager, she can format the drive to its full 90 mmb capacity. The model number of the drive in question is Fujitsu 2612ET. Date: 9 May 91 14:33:29 From: htrails!root (Superuser) To: digest Subject; Building a RAMdisk. THe default ram0 uses a default size of 512k. If you don't have the ram available, you'll have to build a smaller ram# using the mknod command. As the su, type... rm /dev/ram0 <RETURN> The next step is to remake the ram device as a smaller size device. As an example, we'll make one that is just 64k in size. Type this command... /etc/mknod /dev/ram0 b 8 1 <RETURN> Lets break down what you just typed; /etc/mknod - Is the mknod command used to make a special file like a block special file through which a device like a printer or ram is accessed. /dev/ram0 - The directory path and name of your ram device. b - Tells mknod to build a block special file. All devices such as a printer, floppy drive, com port, ram drives etc., are considered a 'block special file.' 8 - Is the major device number for a ram drive. All major device numbers are listed in the Lexicon entry for 'device drivers.' 1 - This is the minor device number of your new ram0. This shows that the ram0 you are building will be 64k in size. If the minor device number would have been '2', then the ram0 size would have been 2 * 64k = 128k in size. Each increment in the minor device number is equal to an additional 64k in size for the ram device. A minor device of 16 multiplied by 64k would equal a 1 megabyte size ram drive. Next you'll make a file system in ram0 by typing... /etc/mkfs /dev/ram0 128 <RETURN> The number 128 is exactly twice the memory size, in this case 64k. Whatever size memory you choose to allocate to a ram device, the block size you specify in the mkfs command will be double. A one megabyte ram device for example would have 2048 blocks. Your new RAMdisk is now ready to be mounted. >>Date: 9 May 91 07:20:51 >>From: algol!bruce (Bruce Mardle) >>To: mwcbbs!support >>Subject: Re: Adaptec AHA1542B >>>>From uucp Thu May 9 00:43:33 1991 BST >>>>From support Wed May 8 06:10:21 1991 CDT remote from mwcbbs >>>>Received: by algol (mail v 2.20) >> id AA298; 9 May 91 00:43:33 >>>>Message-Id: <9105080610.AA25997.V2.20.mwcbbs@> >>>>Date: 8 May 91 06:10:21 >>>>From: mwcbbs!support (MWC Coherent Support) >>>>To: algol!algol!bruce >>>>Organization: >>>>X-Mailer: CoMB 1.29 [Coherent Mailbox] >>>>Subject: Re: Adaptec AHA1542B >>>>>>From uucp Tue May 7 23:43:56 1991 CDT >>>>>>From bruce Wed May 8 07:41:39 1991 BST remote from algol >>>>>>Received: by mwcbbs (mail v 2.20) >>>>>> id AA10273; 7 May 91 23:43:56 >>>>>>Message-Id: <9105080741.AA134.V2.20.algol@algol.UUCP> >>>>>>Date: 8 May 91 07:41:39 >>>>>>From: algol!bruce (Bruce Mardle) >>>>>>To: mwcbbs!support >>>>>>Subject: Adaptec AHA1542B >>>>>>A while ago I changed the disc in my PC from a Rodime 3055 to a Maxtor >>>>>>LXT213 driven by an Adaptec AHA1542B. It worked perfectly under MS-DOS 3.3 >>>>>>at the factory DMA setting of 5Mbyte/s. I tried it at 8MB/s and it didn't >>>>>>work at all. At 6.7MB/s it worked for about 30 minutes under DOS then >>>>>>trashed the FAT. Since I've set it back to 5MB/s I've had no problems with >>>>>>DOS. >>>>>>Having got DOS working on the new disc I then installed Coherent on it. The >>>>>>file system gets corrupted about once every three days. I've tried using a >>>>>>42Mbyte partition, instead of a 84Mbyte one, somewhere else on the disc. >>>>>>I've done a successful low-level verify using the Adaptec BIOS routines. >>>>>>DOS continues to work perfectly on this machine. DOS 4 works perfectly on >>>>>>the PCs at work (Research Machines VX486's with Adaptec AHA1540B's and HP >>>>>>or Maxtor discs (no LXT213, though)). >>>>>>Any suggestions (apart from using fewer parentheses)? >>>>>> >>>>>>My Coherent serial number: 220916044 >>>>>> >>>>When the filesystem becomes corrupted, is it corrupted to the point where >>>>it is necessary to reinstall? Does fsck report the same errors (same inode, >>>>same type of corruption) when the corruption occurs? >> >>>>Is this a 486 you're running COHERENT on? >> >>>>I'll need you to be as detailed as you can get for this one. >> >>fsck has never reported to me an error in any file that looked vital to >>Coherent. The system always seem okay after fsck has got through with it. >>It's usual complaint is that blocks are missing from the free list. On the >>84Mbyte partition as often as not it complains that 147524 blocks are >>missing. That's most of the free space. >>Sometimes fsck complains about dup blocks in the free list. >>I'll have to keep records of it the next time fsck reports something else. >>I don't remember there being any specially cursed i-numbers. Sometimes the >>corruption has its root in something I've done. >>`newusr heptaparaparshinokh "himself" /usr` was definitely a bad move on my >>part. >>I never had 'unsolicited' faults in the file system when I was using the >>Rodime disc. >>The computer is a Philips P3302 (16MHz 386DX) with 1Mbyte of RAM. The >>Coherent is version 3.1.0. 84Mb partition? That *could* explain some, if not all, of this. Two things... shrink the partition. 70 Mb is the safe limit OR, if you have the memory, increase the size of /dev/ram1, as fsck uses ram1 to handle large partitions, but it does have its limits, which are dependent upon the size of the ramdisk. If you will send me a snail mail address, I will send you an updated kernel which addresses some changes to the Adaptec driver. Date: 9 May 91 14:32:28 From: htrails!root (Superuser) To: digest Subject: uucp with MCI access code The only magic in my L.sys is a six second pause (three commas) after dialing the local MCI node access number. My L.sys entry, slightly purged: mwc Any ACU 9600 9501022,,,0xxxxxxxxxxyyyyyyyyyyzzzz "" \r\d\r ... where: 9501022 is the local MCI long distance access number, ,,, is a six-second pause to wait for MCI node dial tone, xxxxxxxxxx is the ten-digit number of the system called (mwc), yyyyyyyyyy is my ten-digit phone number, and zzzz is my four-digit MCI credit card code. The user may want to experiment with more commas; with too long a pause, MCI will disconnect you before you dial the next digits, with too short a pause, your modem will dial before the MCI node puts up its dial tone. Try dialing by hand and listening a few times to see how long the pause should be. Date: 9 May 91 14:33:47 From: htrails!root (Superuser) To: digest Subject; Removing MWC Master Boot (This will allow you to restor your drive to be an all dos system.) To remove the Mark Williams master boot program, you need to overwrite the master boot block on hard drive 0 with another boot program. Usually this is the DOS master boot program. DOS fdisk will write a new master boot program if there is not a valid signature at the end of the current contents of the master boot block. So, all you need to do is modify the contents of the last two bytes of the master boot block, and run DOS fdisk. WARNING: see the note below about DOS fdisk - be sure your hard drive is backed up before you try this!!! One way to invalidate the signature is to copy any sort of garbage into the master boot block. You can (a) reformat cylinder 0 of your hard drive - for example using the "DIAGNOSTICS" menu of the AMI BIOS - or (b) use COHERENT to overwrite the block, e.g. via dd if=/coherent of=/dev/at0x count=1 The master boot block is the first physical sector of the hard drive, e.g. cylinder 0, head 0, sector 1 (don't ask why sector numbering starts at 1!). The MWC master boot program is part of the initial program load, and does not belong to any operating system since it runs before an operating system is loaded. We can make no promises or guarantees what one version or other of DOS fdisk may do. All versions of the program that we have tried do not recognize partitions allocated for other operating systems - such partitions can not be deleted, and are not even displayed, by DOS. Certain configurations of empty partitions cause DOS fdisk to hang. And, most unfortunately, don't expect any data on your hard drive to be available after DOS fdisk rewrites an invalid master boot block; our experience is that, even if previously existing DOS partitions are re-allocated with identical cylinder ranges as at the time of their initial creation, data in these partitions disappears. One other note - if you do NOT select the MWC master boot when you install COHERENT, you can save the DOS or other master boot program using COHERENT' dd command, e.g. dd if=/dev/at0x of=/conf/mboot.DOS count=1 You can then use COHERENT fdisk to write the MWC master boot to your drive from /conf/mboot and you will always have the option of restoring the old boot program later with COHERENT fdisk or dd. Date: 9 May 91 14:34:04 From: htrails!root (Superuser) To: digest Subject; Seagate 1144A Hard Drives Here's what some customers used in the setup table to define the drive. This information may or may not work and is meant only as suggestions not written in STONE! Model # Type & Size Defined As Cyl Heads Sec Per Track 1144A IDE 125mb 47 1001 15 17 Alternative setup 2; 47 1024 14 17 ..and a 1024 write precomp. Keep in mind, that even using these parameters, some customers still have not been able to install with this model Seagate drive. Date: 9 May 91 14:34:41 From: htrails!root (Superuser) To: digest Subject; Shutdown If someone should run /etc/shutdown from a remote site and no one is logged into the console, the console terminal is left in an unfriendly state. Use this command to restore sanity to it. stty echo -nl -cbreak -raw^J <RETURN> Date: 9 May 91 14:36:05 From: htrails!root (Superuser) To: digest Subject: WD1007SV2 Controller Card A customer had install problems with a 160 Meg ESDI drive that he was able to solve by turning off jumper 56 on bank C1 of the WD controller card. >>From: landooz!creed (Creed A. Erickson) >>To: mwcbbs!admin >>X-Mailer: CoMB 0.90 [Coherent Mailbox] >>Subject: Bug fix update to CoMB >> >>This weekend I spent some time playing with CoMB and mail systems in general. >>The result is that I found and fixed one procedural and one cosmetic bug in >>the CoMB system. >> >>1) When used in conjunction with the less pager, CoMB would bomb out of >> reading messages with a "pipe broken" error. This appears to be a >> simple matter of improper closeing sequence in the readmsg.c module. >> >>2) I did not appreciate that comb left itself painted all over my screen >> and the cursor somewhere around line 19 when it exited. I caused the >> screen to clear upon exit. >> >>OK, so here's my questions: How do I upload these updates to mwcbbs? Should >>I wrap the whole thing back up in a .tar.Z package and ship it complete? Or >>should I simply wrap the changed modules and send only them? Third option >>would seem to be simply sending a diff file. So what makes your job easier >>without incuring tons of long distance modem time on my end (not that comb >>is all that large)? If you've changed just one modules, you can mail it to mwcbbs!admin or uucp it mwcbbs!/usr/spool/uucppublic/uploads. I will forward a copy of it to the author if you have not done this, since I talk to his system occassionally. >>NEXT QUESTION-- How do I get a reply to a query such as this? I guess I need >>to ask for a site login. May I have UUCP mail access? I am site 'landooz' >>and will receive mail there as 'creed' (i.e. ...landooz!creed) The file howto.start, recently updated, will give directions for mail accounts requests. >>Creed Erickson >>"Gee, Toto, I don't think we're in Kansas anymore!" ------------------------------ Date: Wed, 22 May 91 10:56:47 -0400 From: dan@ni.umd.edu (Dan Burns) To: coherent Subject: Screen Blanker I am looking for a screen blanker program that works with Coherent. Preferably, I would like one that activates after a number of seconds or minutes of inactivity. Does one come with the package that I can't find, or is one available via ftp from piggy? Thanks Dan Burns (dan@ni.umd.edu) ------------------------------ Date: Wed, 22 May 91 16:05:07 CDT From: "Grant B. Edwards" <rosevax.rosemount.com!grante%rutgers.UUCP@spool.cs.wisc.edu> To: coherent@spool.cs.wisc.edu Subject: Screen Blanking The console device driver does screen blanking. (At least it does on a Hercules video board). I don't know how to set the time-out period, but it is sufficiently long that it has never bothered me. The only decent way to do a screen blanker for the console would be in the device driver itself, so you probably won't see any screen blanking programs for Coherent. For serial terminals the screen blanking is usually handled by the terminal and not by the host computer. grante@hydro.rosemount.com Oh I used to be disgusted, And now I try to be amused. Rosemount Inc, CB7 But since their wings have gotten rusty, 12001 Technology Drive You know the angels want to wear my red shoes. Eden Prairie, MN 55344 - Elvis Costello ------------------------------ Date: 23 May 91 08:58:35 From: sdoucet@f1link.OCUnix.On.Ca (Stephane Doucet) To: coherent Subject: RE: screen blanker In Coherent 3.1.0 the screen on my console blanks out automatically. I fiddled around and found out that Coherent is the one doing it. If your version isn't then maybe it has something to do with your video card? I run and ATI Graphics Solution in my system and have it set to mono mode. - -Stephane ------------------------------ End of Coherent Digest Vol. 91.2, No. 11 ****************************************