arl: DATA: thinkpad total dominator: index

T23 (2647-9KG)

Hardware

CPU (processor)

Intel Speedstep does not work correctly. When booting up the speed depends on is the machine power connected or not. When connected boots up 1133MHz when not ~700MHz. Ok, this is fine, but when you reconnect to power the speed stays because hardware does not do speed stepping - just software which was not there earlier.

Using ACPI you may set processor speed and other parameters with Linux.
/proc/acpi - but enabling acpi disables suspend functionality (apm_bios) [2.4.27].

CPU (socketed) itself is upgradeable. Even though only to 1,3GHz?

The original CPU is
7122A737 SL5K
RH80530 1133/512
FRU: XXXXXXX 11838L4564Z1J
OVR1800C 108
INTEL (m)(c) '01

Intel: Processor Spec Finder | Intel: Mobile Intel® Pentium® III Processor - M: Technical Documents

Ultrabay 2000

Hot-swap does not work - so it's not so hot swap.

IBM has a solution for you: reboot! .. [more: stinking pads]

But really - "hot swapping" may be done when the machine is suspended.

Hardware used for hot insertion contains
PI5C 32X384CB (5V 20-Bit Bus Switch)
PI5C 16862CB (5V 20-Bit Flow through Bus Switch)

DVD

Works fine when booted up with it. Hot-swap does not work. Device is /dev/hdc or /dev/hdd depending on the moon phase ;-). Mostly /dev/hdc.

Using ogle to view DVD movies.

Actually DVD problems can be divided into categories:

Thinkpad problems, i.e. problems with front processor: its poor coding, lousy support by IBM, and lacking documentation.
Linux problems: poorly coded drivers and fixed attitudes with IDE-based technology, visionless designs, and again: poorly coded drivers.

floppy disk

Works fine when booted up with it. Hot-swap does not work.

Ethernet

Works fine with Intel drivers (e100), no problems at all (not even 10/100 problems).

Hardware:
Intel DA82562ET Ethernet chip.

USB

Works fine, but it's only USB 1.1 (slow), so need to have CardBUS card for USB 2.0 use.
Some Linux versions contain "broken" USB support (2.4.20) - and hang when hotplugging USB devices.

T23 does not have bluetooth, but you may use bluetooth through USB dongle (using "ProDigi" adapter - works).

CardBUS / PCMCIA

Works fine, using USB 2.0 adapter.
Some Linux versions (2.4.20) contain "broken" hotplugging - hanging when hotplugging.

IrDA

Hardish to get it working, there are many different solutions and each one has its own way to do things. Is very slow when using GPRS with Siemens ME45 (/dev/ircomm0).

RS232 serial

Works fine. Using it for GPRS connection with PPP.

modem

Not tested, I have no use for common telephone system based modem i.e. POTS modem.

Removed the original MicroPCI modem card. Not used, so why should power be wasted..

sound

After wondering for a while which sound chip should be used with ALSA, got it working. No problems after that.

First tried to install CS46xx modules, because the chip itself informs to be CS. But it needed intel8x0 modules.

Notice, after starting module level is 0, so use mixer to set it up.

Also Notice, after suspend you need to set level, but use different level than original, because ALSA driver (?) remembers the original level.. aaaaaaargh .. and does not do I/O with sound chip(?)..

harddisk

After around 9 months 48 GB drive tilted. Nice to have on-site. Real soon another 9 months have passed by... .. it still works!
The second 48GB drive tilted 200412 totally. Upgraded - new 80GB drive which is actually much slimmer than the original..

keyboard

Although in Finland, and the keyboard has Finnish keys, I use US layout, because that's much easier for programming. Never used an external keyboad.

Specialized keys do not work (front processor takes over?). There's utility nowadays for Linux to control special keyboard keys.

mouse

Using "internal" mouse only. Some problems with X server, like when switching virtual consoles you might automatically paste your confidential mail into irc window .. waiting couple of seconds for X to settle seems to me too much time wasted ;-)

Internal mouse (the clit ;-) is an auxiliary device for the keyboard.

Mouse naturally also stucks after suspends (occasionally), so here's my program to fix the problem:

reset_stinkpad_mouse

After running it, you may need to suspend the machine again (by Fn+F4), because this might stuck the keyboard again.. Link it -static and put it into /sbin.

display

Only using TFT, in mode 1400x1050. No external monitor used, because TFT picture is stabile and does not cause as much headache as CRTs do, especially when used 16 hours or more daily.

TV out/S-Video

Using TV out for DVD and MPEG etc. (after 200410 rarely: bought DIVX capable DVD player). Took quite a while to get it working: adapter cable did not actually work and my TV needed special X11 configuration - could not find much help in Internet. S-Video to SCART needed still composite, so needed to connect S-Video pins 3 & 4 with a small value ceramic condensator (~470nF) together inside a SCART adapter. Without this end result will be B/W (black and white) - naturally this composite image quality is not good, but anyway TV sucks also, so...

X11 savage driver is not able to switch to TV mode, but another program s3switch does. So when TV out is used first do

s3switch tv pal

and start X server when using TV as monitor I use

xinit -- :1.0 -xf86config XF86Config.TV_PAL; s3switch lcd

X_to_tv script [here]. so the display will be back lcd after X server. Naturally I'm running at the same time "normal" X server on :0.0

Notice:

Do not close display (thinkpad cover), because there's some sort of a bug, which causes also TV out to be closed. I think a bad hardware design.
Set screen saver within boot/bios menus off, because it will also save your tv out screen middle of DVD movie .. this is also due a bad hardware design.
Naturally switching virtual console does not work correctly: you may need to have to run s3switch lcd again.
Some SCART adapters do have 3 RCA connectors (L+R audio and video) and 1 S-Video, do not connect video RCA connector (will cause no video at all).
First patched s3switch, because it seemed to put T23 graphics chip into wrong category, but the original code seems to work (have to check this).