Unix is the Hole Hawg of operating systems, and Unix hackers, like Doug Barnes and the guy in the Dilbert cartoon and many of the other people who populate Silicon Valley, are like contractor's sons who grew up using only Hole Hawgs. They might use Apple/Microsoft OSes to write letters, play video games, or balance their checkbooks, but they cannot really bring themselves to take these operating systems seriously."
I've been moving from Solaris to Linux for some time now. Not entirely a pain-free experience. This page is a list of notes on obscure bits and pieces for those who are finding odd problems.
The hardware on my system, janus, is:
I'm using the RedHat 7.1 distribution.
janus is not a dual-boot system; I have another system with Windows 95
on it elsewhere.
Unfortunately, hardware manufacturers tend to distribute their firmware
upgrades in formats which require Windows 95 to install.
Thankfully, you only need a DOS-style bootable floppy disk.
When trying to install RedHat for the first time, it would go OK for a while and I would then get an Oops at some point. The trick here seems to be that the RedHat installation system doesn't like "large" amounts of memory. At the initial linux: prompt, boot with linux mem=32M and things work out.
On further reflection, this is probably another manifestation of the Sig 11 Problem.
If you are upgrading to 7.0, pay very careful attention to
the options you are being given!
The default disk partition option (I think) means that your
file system will be wiped.
Backups are your friends.
It's a good idea to upgrade the SCSI controller firmware to the latest version. You can get the latest firmware from here. To install, you will need to make up a Windows 95/98 boot disk, with the downloaded software on it. You then boot with the floppy disk, run the downloader and then reboot.
The Adaptec documentation, the Linux SCSI HOWTO and the aic7xxx documentation
all refer to SCAM support.
SCAM is no longer supported on these cards, as nobody liked it anyway.
I no longer have a WangDAT tape drive. But this information is still useful.
The SCSI controller driver is the aic7xxx module. Initially, on booting, the tape drive would be recognized as another disk drive, leading to all kinds of problems.
After a lot of fiddling about, checking terminations, cable integrity, etc. it turned out that the tape drive firmware and the aic7xxx driver were not entirely compatible. What was happening was that no data was being returned on the initial SCSI scan. The driver does not clear out its buffers between queries, so the old data from the disk drive was still there, leading to the confusion. The solution was to upgrade the firmware on the tape drive.
You can get firmware updates from here. My tape drive has an ID of 132XXXX-255A, rather than the -251 or -051 listed; for this model, version 3.0M is appropriate. To install the firmware, you need to make up a Windows 95/98 boot disk, install the DOS driver for your SCSI controller (and add it to config.sys) and the flash update software. You can then boot using this disk and install the software.
It would appear that
no longer has the firmware for 3200s listed on their web site and
I am one of the few people on the wide Internet who mention this.
As a result, I get occasional requests for the firmware.
So, here is a copy of 30MFLASH.EXE.
Please get this straight: You use this at your own risk.
This firmware only works for specific versions of the WangDAT 3200.
Other versions may be damaged or cease working.
If in doubt, bother the poor sods at Tecmar, who may be induced
to re-post the appropriate firmware.
I'm using the supplied XFree86 X-Windows system. The Viper 770 display card is not supported by older versions of XFree86, so you have a choice:
The monitor that I have is not part of the monitor database, so it needs to be added as a custom monitor to /etc/X11/XF86Config. The relevant parameters are:
Section "Monitor" Identifier "PL70i" VendorName "Panasonic" ModelName "TX-D7S55" HorizSync 30.0-97.0 VertRefresh 50.0-180.0 Option "DPMS" EndSection
The sig11 problem.
This usually shows up when you are recompiling the kernel. gcc will suddenly stop with a "signal 11". When you type in make again, the compilation will go on for a little longer and then fall over again.
Alternately, it will appear that a file has got a dud character in it. Or you will get just general randomized crashes in perfectly stable software.
Unfortunately, you have probably got a hardware problem: one of your memory chips has a fault in it. Just to make things difficult, however, there are other possible reasons for this problem. The Sig11 FAQ may help here, although it makes for depressing reading.
Another possibility is the APM power management is acting up.
You can disable APM in your BIOS while booting.
The Creative Vibra 128 is not officially supported by RedHat. However, sndconfig will, correctly, recognize this as equivalent to an Ensoniq 1371.
You will need to turn on sound in the GNOME settings, as well, to get sound effects.
The Ensoniq 1371 does not support the older ALaw/uLaw formats expected by OSS/Free or have a MIDI synthesiser. See the kernel documentation in /usr/src/linux/Documentation/sound/es1371 for more information on this. From kernel 2.4 on, it looks like support for OSS is being phased out (maybe to make way for ALSA). You will get a lot of complaints about missing sound-service-0- in the log. sound-service-0-0 is the mixer, sound-service-0-3 is the digital signal processor. I have no idea what 6 and 8 are.
You could install OSS, if you want to. This will solve most of your problems in exchange for USD20 or so.
You might want to put lines like
alias sound-service-0-0 off in your
I don't think this has any harmful effects.
I originally had a Ricoh MP7040A CD-RW CD writer. This is an IDE CD writer and the experience has not been a very happy one. CD writing in Linux requires the use of the ide-scsi module, to make the IDE drive look like a SCSI drive. This has always been unreliable, occasionally filling the logs with vaguely threatening messages. It also seemed to be affecting the ordinary CD-ROM drive as well, even if they were on separate IDE buses. (Opinion on the CLUG mailing list appears to be that the 7040 is not a good drive to use -- other Ricoh drives appear to be OK.)
The Yamaha drive is a SCSI drive and everything has started working
I'm using an old Texas Insruments MicroLaser Turbo as my Postscript printer. This works pretty well, provided that the printer is turned on and on-line before anything is sent to it. If there is something queued for printing when the printer comes up, something seems to get sent to the printer during the Initializing phase of the printers boot and the printer stays initializing forever. A quick check with tunelp shows that the system seems to think that the printer is on-line when it's still initializing.
Setting the printer device to check that the printer is on-line and generally happy before printing seems to fix this. Use the command
tunelp /dev/lp0 -o on
The Happy Hacking Keyboard
The Happy Hacking Keyboard is not particularly well supported by X11. The basic thing that you need to do (unless you are Japanese) is to set the keyboard switches to S2=on, SW3=off, SW4=off; these settings make the diamond keys Windows keys.
For a more complete set-up, including geometry files, download hhk.tar.gz. This package contains a Makefile to patch the XKB configuration files.
15-Feb-2003 Doug Palmer