PC110 and Linux --- Update
Vaughan Pratt
Tokyo's Electric City (Akhibara) two weeks ago.
I reported that I'd got Linux running, less Lilo and X-windows.
The problem with Lilo turned out to be in the install part: lilo was
treating the PCMCIA drive as the second disk and (so it seemed) getting
the geometry from another disk. I solved this by adding the disk geometry
to lilo.conf just before the image=/vmlinuz section, namely:
disk = /dev/hdc
bios = 0x80
sectors = 54
heads = 16
cylinders = 771
I'm *really* pleased at how well the APM works with Linux---just close
the lid (putting it in suspend) and everything powers down almost
completely, the battery loses no charge for days at a time. Open the
lid and everything picks up right where it left off. When the battery
drops to around 5%, the system suspends automatically. To recover, just
change batteries. You can do this with or without AC connected, without
panicking the kernel.
The battery mileage is good too---I was able to let
xv -wait 1 -wloop *
run in a directory full of jpeg's for two hours! The disk was running
the whole time.
X-windows (XFree86) was very easy to configure. I linked /var/X11R6/bin/X
to /usr/X11/bin/XF86_SVGA, and copied /usr/X11/lib/X11/XF86Config.eg to
/etc/XF86Config, with some minor changes: uncomment the four FontPath
entries and Emulate3Buttons, replace "Microsoft" and "/dev/mouse" with
"PS/2" and "/dev/psaux", and add
Chipset "ct65530"
VideoRam 512
to the generic svga device entry. (The actual chip is a 65535, which
XFree86 doesn't know about, anyone know what it adds to the 65530?)
Works like a charm.
Netscape runs, but for some reason it freezes the system for 40 seconds
when starting up. I'd guess this might be caused by some combination
of only 12MB RAM and no coprocessor.
Networking is not so hot. I tried running PLIP (parallel-port version
of SLIP) on the 110's parallel port, but for some reason the 110 kept
locking up after transferring a few kB. No idea what the problem is,
I've used PLIP successfully in the past. So I resorted to SLIP at 115Kb,
which is ok for telnet, but for file transfer you really notice the two
orders of magnitude slowdown compared with ethernet.
You can get an extender for a Type III drive from
http://users.aimnet.com/~avasales/adaptor.html for $49, that gives access
to the slot the Type III drive would normally block. (I'm using the
equivalent extender from Twin Industries.) However I haven't yet found
out how to access one port without knocking out the disk running in the
other---I need to talk to David Hinds about this.
Even nicer would be an extender for the 50-pin compactflash PCMCIA port
on the other side that allowed you to plug in a Type II PCMCIA Ethernet
card. But compactflash seems to assume ATA, so this would only work if
Hinds' code could take control and configure the compactflash port as
an ordinary PCMCIA port. Or, since the main drive is ATA (which has
fewer than 50 signals and is electrically identical to compact flash), I
suppose an extender (if such existed) would let me plug the drive into
the compact flash port freeing up *both* Type II slots.
As for modems, a PCMCIA modem will work iff a PCMCIA ethernet will work.
Of course there's always the option of connecting a modem to the serial
port and an pocket ethernet adaptor to the parallel port. And when
really desperate, there's always the 2400 baud date/9600 baud fax modem
to fall back on.
I tried using the infrared port, which has the advantage over the serial
and parallel ports of not needing the docking station (IBMese: port
replicator). IR worked ok at 115kb under Windows 3.1 with TranXit, a
package for copying files between PC's. (I used Adaptec's Airport for
the other end, which Tranxit knew about; it didn't of course know about
the PC110, but it worked fine when I described the 110 as "generic".)
On Linux however I couldn't run cu reliably at more than 19.2kb: it
lost every 20th character at 38kb and couldn't communicate at all at
57kb and 115kb, pure garbage both directions. I tried with no success
at all to run slip, which did not work at even 9600 baud. This may be
because IR demands half duplex---too much interference with full duplex.
I couldn't find any Linux IR code on the net.
Except for the communications headaches, the 110 is a great little
machine.
--
Vaughan Pratt http://boole.stanford.edu/nerdsheaven.html
Gary Hong
I have a compaq aero and I'm thinking about getting an IBM PC110.
It's a waste of money to have two subnotebooks and I was wondering, if I
decide I don't want it, anyone will to buy it off me when I come back to the
states? I live in Northern California (SF/Oakland/SantaCruz area).
Thanks,
Gary
--
ga...@sco.com
Amanda Walker
>run in a directory full of jpeg's for two hours! The disk was running
>the whole time.
Must be a lower power drive than the Maxtor I'm using :).
>Networking is not so hot. I tried running PLIP (parallel-port version
>of SLIP) on the 110's parallel port, but for some reason the 110 kept
>locking up after transferring a few kB.
Hmm. Well, the only think I can think of would be to check
the BIOS setup to make sure the port's configured to be bidirectional.
I think that's the default, though.
>Except for the communications headaches, the 110 is a great little
>machine.
Sounds like some of those are Linux-specific; I've had no trouble
using PCMCIA modems & SCSI cards while a drive is in the top slot with
an extender, using either the BIOS built-in slot mapping feature or the
DOS card & socket drivers.
IR is definitely half-duplex, though; IrDA includes a "com port emulation"
protocol, but it's a lot more complex than just talking to the serial
port that the IR transceiver is attached to.
Amanda Walker
--
"The nice thing about standards is that there are so many of them to
choose from." --Andrew S. Tanenbaum
Marko Schuster
In a previous article, ama...@intercon.com (Amanda Walker) says:
>>run in a directory full of jpeg's for two hours! The disk was running
>
>Must be a lower power drive than the Maxtor I'm using :).
Sounds interesting.. anyone has an overview of PCMCIA-III-Harddisks,
their capacity, speed and power consumption?
..bye..
--
ns...@fim.uni-erlangen.de - Real:Marko Schuster - Alias:Starli/Starliner/*---er
Tel&Fax:0911-3939031 - Data:0911-3939031t,#,#,# - Loc'n: D-90419 Nuernberg
Comp: A4000/040 (Amiga/Mac/BSD) - GRiD 80386 Penbook - Casio NX4000 -Sharp1403H
GU H+ s-:+ g+ p1(p+) !au a20 C++ L+ N++ M- t- tv++ !D u* h! f+@ !r@ n-- !x(x**)
Vaughan Pratt
>pr...@leland.Stanford.EDU (Vaughan Pratt) writes:
>>run in a directory full of jpeg's for two hours! The disk was running
>>the whole time.
>
I've now had time to measure battery lifetime more systematically.
My figure of two hours that Amanda is rightly questioning here had
been a hasty extrapolation assuming the APM's percentage readout meant
something. Now that I've tracked the readout more systematically,
I can see that it's far too nonlinear to extrapolate reliably from.
I ran several tests with two battery types, the 1200 mAh battery that
comes with the PC110 (I have two and tried both) and one 1350 mAh Sony
NP-520 used with Sony handycams. The OS was Linux, with XFree86 running.
In one test I had xv displaying a gif every few seconds. (This kept the CPU
busy, and the disk stayed spinning all the time though I don't know why
since the 17 images should have remained cached in RAM.) In this test
the IBM batteries took 90 minutes to discharge from 100% to 5% while the
NP-520 only took 60 minutes. (So either the latter is defective or these
mAh numbers are meaningless!).
In the other test I added the ethernet card, with a couple of xterms
opened, and played midi music on the built-in speaker (in the battery
cover) at the same time (still displaying the series of xv images).
I thought this would take a big chunk of power, but there was no
noticeable difference in battery lifetime from what I measured without
these two.
If I'd had the time and the data were of any consequence, I'd have
repeated these experiments. I didn't, so take them with the proverbial
grain of salt.
Incidentally you can plug the Sony battery into the PC110 but you can't
plug the IBM battery into the Handycam. Sony has a slot in their
battery and a plastic bar in the Handycam that fits into the slot. The
IBM battery doesn't have this slot and is blocked by the bar. Since
both devices can both charge and discharge their battery, I have *no*
guesses as to the significance of this difference. Anyone know if it
was IBM's idea or Sony's?
--
Vaughan Pratt http://boole.stanford.edu/nerdsheaven.html
My every word is copyright, especially
those not in the dictionary.
Robert Kawaratani
pr...@leland.Stanford.EDU (Vaughan Pratt) wrote:
>Netscape runs, but for some reason it freezes the system for 40 seconds
>when starting up. I'd guess this might be caused by some combination
>of only 12MB RAM and no coprocessor.
For the real crazies, there was a photograph of a 16 MB memory expansion
board
for the 110 in the May issue of Mobile PC which lists for 98,000 yen.
Anyway,
it's good to hear that the 110 users are having as much fun as the 200lx
users.
Ganbatte.
Amanda Walker
Kawaratani) writes:
>For the real crazies, there was a photograph of a 16 MB memory expansion
>board for the 110 in the May issue of Mobile PC which lists for 98,000 yen.
Cool; a 20MB PC110 would be fun. A little pricey, but fun :).
>Ganbatte.
That's what I told myself when I decided to get one...
Amanda Walker
--
"We are holding Elvis Presley's brain hostage on Planet Zort. Surrender Now."
--Bloom County