Installing from the external PCMCIA 4X CD-ROM Drive | |
Installing from a DOS-partition | |
Installing via NFS | |
Installing via FTP | |
Getting Things to Work | |
Open questions |
My Thinkpad 701 C came with an external PCMCIA 4X CD-ROM Drive (model type 1969-008/108), so I imagined installing Linux would be a breeze - just insert the CD and I'm all set. Wrong! Linux can not be installed from Windows 95 (the computer came with Win95). As a matter of fact it starts a small Linux-kernel early on in the setup process and does the entire installation under Linux. But: this small kernel does not come with support for the above CD-ROM drive, or I have been unable to do the correct things to enable the cdrom. It does however work once Linux is installed.
So I tried installing from DOS instead. This requires to copy a minimum set of files specified in the SuSE manual onto a DOS partition (about 90MB). Together with the approx. 100MB Windows 95 is currently installed on (so that I will be able to boot the computer if installing Linux fails) this leaves only 330MB of free disc-space on my 540MB HDD. Not much.
Well, it turned out that installing from HDD wasn't really an option, since Linux kept looking for lots of things deemed essential in other places.
Something else to watch out for: The (German) SuSE manual suggests to call the directory where all the files will be installed into /emil (this is supposedly funny) - however, YaST (the installation program) will look in /cdrom by default. Very funny indeed.
So next I tried installing over the net. I'm working at the University of Freiburg's computer science department, so I'm in the fortunate situation to have access to both an Ethernet and a PCMCIA Ethernet adapter. So I inserted the 3Com 3C574-TX Fast EtherLink 16-Bit PC Card and boot the computer using a Linux boot-disc - only to find that Linux doesn't support this particular Ethernet adapter. Well, a buggy driver exists, but for one thing it isn't part of the small installation kernel, and secondly it is, well, buggy.
By now I have borrowed a 3Com 3C589D-Combo EtherLink III LAN PC Card, and surprise, once I load the PCMCIA-module the card is immediately recognised and seems to work just fine. So all is set...
First I try to do an NFS install. All seems to work fine. I'm asked to provide the following information (note: this will obviously be different for you. However, if you have access to an Ethernet and an Ethernet-card, chances are good you also have access to someone who knows the correct values for you.):
IP-Address | 132.230.167.xxx |
Subnet-Mask | 255.255.255.0 |
Gateway | 132.230.167.254 |
DNS | 132.230.200.200 |
Server | 132.230.1.7 |
Directory | /ftp/pub/linux/suse/suse |
All seems to work well. I get YaST. I can format the free space on my disk and create a normal Linux partition and a swap partition (using the default sizes recommended by Linux). But then I get a message telling me that /mnt can not be mounted because only root can do this. What the... I never asked anybody to mount /mnt, and so far I didn't even have a chance to become root anywhere. Dread!
So I start over again, doing an ftp-install instead. And finally: success. At least until now (Installing package 118, 41 packages remaining) - it's slow, but it obviously does install something...
Success! Well, all the packages were installed at least. I've been selecting the stable 2.2.36 kernel with (E)IDE0 support.
This CD-ROM drive was sold by IBM probably from 6/96 till 7/97. The model number of my drive is 1969-108. On IBM's American website however only the model number 1969-008 can be found (maybe European and American model?). As you may have read above I could not install Linux using this drive. However, once Linux was installed successfully, everything went smoothly. I can now use the drive without any problem (it's even Plug'n'Play). The only thing worth mentioning is that the CD-ROM is not /dev/cdrom, but /dev/hdc.
Since I did not do anything in order to get it to work, I can not really tell you what to do it it doesn't work for you. However, I'll show you the output from lsmod to give you an idea what modules are loaded:
# lsmod ide_cs 1 0 3c589_cs 2 1 dummy0 1 1 (autoclean) ds 2 [ide_cs 3c589_cs] 5 i82365 6 2 pcmcia_core 9 [ide_cs 3c589_cs ds i82365] 0 memstat 1 0 nls_iso8859_1 1 1 (autoclean) Module Pages Used byif this doesn't help you, then have a look at the (relevant?) startup-messages:
Linux PCMCIA Card Services 3.0.7 kernel build: 2.0.36 #1 Mon Jan 18 19:58:46 /etc/localtime 1999 options: [pci] [cardbus] Intel PCIC probe: Cirrus PD672x ISA-to-PCMCIA at port 0x3e0 ofs 0x00, 2 sockets host opts [0]: [ring] [65/6/3] [1/15/3] host opts [1]: [ring] [65/6/3] [1/15/3] ISA irqs (default) = 3,4,5,7,9,10,11,12 polling interval = 1000 ms cs: IO port probe 0x1000-0x17ff: clean. cs: IO port probe 0x0100-0x04ff: excluding 0x220-0x22f 0x378-0x37f 0x388-0x38f 0x398-0x39f 0x3b8-0x3e7 cs: IO port probe 0x0a00-0x0aff: clean. cs: memory probe 0x0d0000-0x0dffff: clean. eth0: 3Com 3c589, port 0x300, irq 3, Auto port, hw_addr 00:10:4B:7E:39:BD hdc: CD-400 Series., ATAPI CDROM drive ide1 at 0x110-0x117,0x11e on irq 5 ide_cs: hdc: Vcc = 5.0, Vpp = 0.0 hdc: media changed hdc : tray open or drive not ready hdc : tray open or drive not ready hdc: media changedThe drive is mountable (and useable) despite the messages media changed and tray open or drive not ready.
I found that my thinkpad displayed a tendency to use irq 3 for the CDROM even when the MultiPort was used - in which case irq 3 is occupied by /dev/ttyS1 (that's com1 in DOS-speak). Luckily the problem is easily solved by editing the file /etc/pcmcia/config.opts and excluding irq 3. The CDROM now reliably uses irq 5 - which, I think, is also occupied by the second parallel-port. Well, I'll cross that bridge when I get access to a printer.
I now did, works as expected. Unfortunately I can not use the printer (my girl-friends) anyway, as it is a Win-printer. So here's a word of advice: do not get an inkjet from Lexmark, as virtually all are Windows-only!
I was somewhat afraid to use LILO. It supposedly doesn't work on Thinkpads (although I was able to use it from disk just fine), and I was afraid to loose my Win95 partition. Granted, not a great loss. But the Butterfly needs a FAT-partition for the hybernation file, so I thought I might as well keep Win95 around too. I'll probably have to change that once I get the 32MB RAM I'm looking for. So I made both a Linux rescue disk and a Windows startup-disk before installing LILO.
And right enough, once I had installed LILO I wasn't able to boot into Win95 anymore. I could press whatever key I liked, no menue appeared, and LILO would go straight into Linux every time. I was close to using the Windows startup-disk and nuke LILO using fdisk /mbr (undocumented!).
Luckily I read the instructions to LILO before doing so and found out that you need to press the SHIFT key in order to activate LILO, and the TAB key in order to get a list of possible selections. Hrumpf! What's wrong with providing a menue, I would like to know?
However, you can edit /etc/lilo.conf to display text from a file (/boot/message in my case) everytime you boot, so while this doesn't solve all problems - still no menu - a user now at least knows what options he's got.
Hybernation works out of the box, both in text-mode and under X. You press Fn-F4 to enter hybernation, and simply switch on the computer to get out of it. The screen will initially remain dark when doing so (and you might think hybernation didn't work after all), but pressing Fn and any of the funktion keys should remedy this.
The only problem with hybernation I have is that the clock completely looses track of the real time. But this might be due to the fact that my Thinkpad is somewhat defective and will not accept the battery (which is working).
Although, it would be nice to be able to enter hybernation from the command-line, but with the broken computer/battery I really didn't see a point in installing APM.
[22.2.1999] Ups. It seems hybernation doesn't work when the port adapter (multi-port II?) is used. Strange!
[26.3.1999] Not strange at all, it turns out. There's an option in the bios to disable hybernation while using the MultiPort. Once I enabled hybernation again, everything worked as expected.
[17.5.1999] It's not only the clock that won't wake up but also PCMCIA cardservices, agetty (if used) and (sometimes?) sound. I now run the folling script when returning from hybernation:
#!/usr/bin/tcsh -f # set clock /sbin/hwclock --hctosys # restart agetty /usr/bin/killall agetty # reset PCMCIA-cards /etc/rc.d/pcmcia stop /etc/rc.d/pcmcia start # Sound /usr/local/bin/soundon if ( -f /usr/share/sounds/au/hal9.au ) then cat /usr/share/sounds/au/hal9.au > /dev/audio endif
I'm using an external US-Robotics Sportster Vi 28.8 Faxmodem (however, it seems the original owner upgraded it to 33.6). This can only be connected through the MultiPort adapter (no serial link otherwise). Shame. The device corresponding to the port-adapters serial link is /dev/ttyS1. It seems like /dev/ttyS0 is the infrared-port.
I initially had some problem getting the modem to work. I now think this was due to the irq-conflict mentioned in the CDROM-section.
For my model, the 2630-TF4, sound worked out of the box using OSS 3.7.1z. Auto-detect reported an ess1688, and everything worked effortlessly - no kernel-build required, nothing.
This is a strange thing. The Thinkpad 701C reportedly uses MWave for sound, which currently doesn't work under Linux.