BTW, be sure to read section 12, Compaq service(?)
. It details problems I've personally had with Compaq service.
I wouldn't not buy the machine because of these issues, but it is definitely
a consideration.
| 27-April-02: |
The machine is dead. |
| 14-April-01: |
Updated
LiViD DVD player information (it works now!). |
| 14-April-01: |
Added information on XFree86
hardware acceleration (DRI). |
| 24-March-01: |
Update to
working software hybernation. |
| 24-Feb-01: |
Added ancidotal evidence of a
working software hybernation. |
| 23-Feb-01 to 5-March-01: |
Added a section on my experience with
Compaq service. |
| 23-Feb-01: |
Added information about
XFree86 3.3.6 support for ATI 128 Mobility. |
I found the Storm Linux install to be quite straightforward. It autodetects
most of the hardware (except the display, of course). It stops just
short of helping you configure the modem, printer, etc. But a utility
is provided, called sas (Storm Administration System), which is a great help
on those items.
The laptop came with Windows Millenium Edition installed. Now, I'm
no fan of windows, but ME really takes the cake. I had the machine for
less than 12 hours and I'd already seen safe mode twice, and scandisk half
a dozen times. Definitely time to install Linux!
Based on my experience so far, I would highly recommend against Windows ME. If you want to dual boot your system as I do, get 98 or 2000, or even NT. Windows ME does.
I partitioned the disk as follows:
Although Storm Linux comes with a custom version of the 2.2.16 kernel (that's
good), it isn't very well built for this machine (that's bad). In the
next section we'll discuss what needs to be done to the kernel to make it
usable.
The reason APM support isn't included is because the default kernel supports multiple processors, which excludes most of the APM features.
I believe (somebody correct me on this) that most distributions such as RedHat and Suse are not so handicapped. From what I've read these distributions usually have a loptop-tuned kernel.
If you need to rebuild your kernel read the Linux Kernel HOWTO . The important areas that I had to look at were
There is one other thing that must be done to enable framebuffer support. A line must be added to your lilo.conf file. You must add a line that looks like:
vga=835I have not been able to find any real documentation on these numbers, except that on the Framebuffer HOWTO page (which only covers standard VESA codes, not ones for oddball display configurations like 1400x1050). I have found from research on the net, however, that 834 is the code (on ATI chips, at least) 1400x1050 resolution with 16 bit depth, while 835 is the same resolution with 24 bit depth. You must set this mode either in the lilo.conf file or as a kernel parameter at boot time.
Here is a copy of my lilo.conf file.
One thing I would recommend, while you are tinkering with the kernel, is enabling of Advanced RTC (Real Time Clock) support. This is most useful in multiprocessor systems (required, in fact) but there are apps that require it as well.
For those that are interested, here is a copy of my kernel configuration. You can load it using the menu or x configuration utilities in XFree86 (make menuconfig or make xconfig ).
23-Feb-01: Apparently my statement above about ATI Mobility 128 cards not
being supported under 3.3.6 is incorrect. Tom A. Cox sent in this email
correction:
Hi Ken,
I've attached my configuration file and the output from 'X -probeonly.'
Basically, the default Mandrake installation set things up to use the Mach64
server at 32bpp, 1024x768. This is a good resolution four me.
Although it set
up additional resolutions, trying to switch to one of them will cause the
screen to blank and the system to lock up. It appears that Mandrake
also set
up a frame buffer device, but it isn't used and I haven't experimented with
it.
I've run Mesa, OpenGL programs and the GIMP and everything seems to work
just
fine. I still need to work on the APM configuration. It's just
barely usable;
apm -s from the console works, but apm -s from inside X will never recover.
DPMS works at times, and doesn't at others.
On an interesting side note, I'm currently working with real-time systems.
I've run several variants of real-time Linux on this system without a problem.
.
.
..
Thanks,
Tom
Here is the configuration file and the X -probeonly file Tom sent. Thanks Tom!
14-April-01: In an attempt to get some of my Virtual Reality research
to run a little faster on this machine, I tried installing the DRI (Direct
Rendering Interface) support for
XFree86 4.0.2. See the DRI User Guide and the
DRI Compilation Guide
(BTW, as of this writing XFree86 is up to 4.0.3, so you might as well use
that). The installation proceedure is complicated, so if you aren't
comfortable with configuring kernels (it requires a kernel > 2.4) and
compiling applications that don't have much good documentation and don't
compile all by themselves (i.e., I had to go hunt down the pieces that didn't
compile from the main compile and compile them seperately), don't try
it. On the other hand, if you are reasonably careful there isn't
much harm you can do. I didn't replace my original X, so I can always
revert back to it if I have to. Just one more caution: follow
the directions. I thought I was smart enough to leave out things I
didn't think were important or that didn't apply. I was wrong.
I never did get the accelleration I was looking for. In fact, my VR environment ran much slower. What seemed to happen is the simple examples worked like greased lightning, but the more complex ones didn't. I haven't tracked down what exactly slowed it down, but there was an unexpected side effect: the LiViD DVD player started working much better.
Unfortunately, while the kind people at Lucent have made a driver, they were a bit unkind to those of us who don't use RedHat Linux. The install script that comes with the driver makes some assumptions about how things work, assumptions geared for a RedHat box. What I ended up doing is performing the easy stuff by hand, and then putting a driver control script in /etc/init.d .
The hand setup stuff is pretty straightforward. As root run the following commands from the directory in which you unpacked the linmodem driver.
rm -f /dev/ttyS14Then put this control file (linmodem ) in /etc/init.d . Then to enable the driver when you boot run, as root
mknod /dev/ttyS14 c 62 78
rm -f /dev/modem
ln -s /dev/ttyS14 /dev/modem
chgrp uucp /dev/ttyS14
chmod 666 /dev/ttyS14
cp ltmodem.o /lib/modules/`uname -r`/misc/
update-rc.d linmodem defaults 20Now it can be started by executing (as root)
/etc/init.d/linmodem startand stopped by executing
/etc/init.d/linmodem stop.
X seems to recover fine from sleep mode. I've been led to believe that this wasn't always the case, but the combination of a 2.2.16 kernel and XFree86 4.0.1 seems to have solved that problem. On the other hand, no hibernation yet, either.
24-Feb-01: I received this email the other day. It seems that someone
has gotten software hibernation to work. Following is the email:
I have a 12XL300 with the suspend-to-disk working fine. I set up
/dev/hda4 as
type A0 (IIRC; the laptop is at home and I'm at work) and sized it to the
next
complete cylinder after sizeof(RAM) + sizeof(video) + 2 MB. I formatted
it with
lphdisk (which looks for /dev/hda4 and then checks to see if it is type IBM
Thinkpad before it formats it). Now I can suspend to disk (or hibernate,
if you
prefer) with the power button, suspend to memory with Fn-F4 or shut it down
completely with "shutdown -h now". The
thing that makes it easy is that the laptop is exclusively Linux with Windows
running as a task in X (with the use of Win4Lin). I haven't tried setting
up a
hard drive with an extended partition at /dev/hda3, a primary partition at.
/dev/hda4 and the logical drives at /dev/hda5+ and in fact don't even know
if it would
work. (The limiting factor here, AIUI, is the insistence of the hibernation
BIOS on working exclusively with /dev/hda4.)
HTH
Thanks HTH! (whoever you are :)
24-March-01: I tried this today with the same result: nada. I'm
begining to suspect, based on some other laptops I've seen, that the new
BIOSes are handling this in a new and interesting way. Dell's latest
uses a hibernation partition at /dev/hda1, rather than /dev/hda4. Maybe
this machine is simular. Well, onward and upward.
14-April-01: OK, now I have it working. It took at least a week
of pounding my head against the wall and fiddling with the supplied development
libraries. It is my fault, really. I decided that using the Storm
Linux CD was too much trouble, so I tried to convert the system to straight
Debian. This caused all sorts of havoc with the Storm supplied libraries
for which I had no development versions.
But now the DVD player works! At first I had a real problem with it skipping and popping and making these horrible squealing noises. Heavy Metal wasn't too bad, but the Matrix was just unwatchable. Then, for an unrelated reason I built an accellerated X. Sometime later I discovered that, even without running a kernel the X accellerator seemed to need, the player worked much better (I'm still running the original kernel I configured for the machine, above ). Now all movies play without all the annoying sound effects. Not perfectly, mind you, but many are now acceptable. Most all of them have just a slight little jerkyness to them, but it is, for the most part, ignorable. Some movies, though, like Austin Powers, the Spy Who Shagged Me, are really attrocious.
The controls are run by a GTK interface, which is really slick, but not entirely functional. You can select where to start the movie, play, pause, and select the soundtrack. But that's all. Anything else is either unavailable or locks it up tighter than a drum when you try to use it.
So the good news is, a free DVD player for Linux is working, and it works on this machine. The less than good news is that it is a little bit buggy. But an admirable feat as it is anyway. I really like it.
As for building it, I followed the directions. The place to start
is the Oms How To. It wasn't that complicated, and basically, once all
the prerequisite stuff was loaded, it worked as advertised. The only
caviat is that you must install the accellerated X server or it won't be
worth a piddle.
Since this section got to be enormously long, and doesn't have a direct
bearing on the subject at hand, it
has been moved.
Last modified March 5, 2001