1. (of 2)
Never forget that Linux is a UNIX-workalike.
That means, you can ask ANY REAL (!!)
UNIX freak / hacker / crack.
It also means, that you might want to get a copy of
"The UNIX Programming Environment" by Brian W. Kernighan and Rob Pike.
Data to get it: Prentice Hall Software Series, ISBN
0-13-937681-X.
When I ordered it - some years ago - I got a SHOCK.
I got that tip from "jax" at The Well, S.F., CA, and
re-confirmed he really meant THIS book.
I talked with one of the authors. Of course, again on The Well. (Visit The Well's Unix
Conference)
The reason: That book was published in 1978.
But since UNIX is stable, standard... no reason for major changes... IT APPLIES. Believe me,
it's good for you.
2. (of 2)
Index of Problems.
(Now you know what "IP" means...
;-^)
LILO doesn't work: LI 101010101010101010....
Manual Pages don't work
X doesn't work good enough
Unrecognized TERM type
Could not find termcap entry for linux.
Sorry - PPP is not available on this system
"Microwave oven? Whaddya mean, it's a microwave oven?
I've been watching Channel 4 on the thing for two weeks."
Before we start, this is an excrept from the RedHat Installation Booky:
"If you're not sure what option to choose, talk to your systems administrator.
If you are the systems administrator and you're not sure, select Manual,
and consider a career in the food-service industry..."
But we also found some help in RatHead by "using the force, reading the source":
From: Donnie Barnes
Xgettext-Options: --default-domain=Xconfigurator --add-comments
--keyword=_ --keyword=N_
Files: ../Xconfigurator.c
cantsell pick a card, any card clock probin failed clockchip
configurashun configurashun proceedin usin defaults. you wanna run 'X
-probeonly' now? don't probe i wrote da file. you mat wanna poke at it
before ya run 'startx'.
how much video memory you got? it's possible that hardware detectin can
crash da system and leave the screen blank. if'n dis happns, don't do
this step again next time. lemme choose moniter setup nawp we needs to set
the specificashuns of the moniter. dere's to important thangs...the
vertical refresh and the horizontal sync rate.
da valid range fer the horizontal sync and vertical refresh is in da
manual fer yer moniter. yep pick a card from da list (or pick "Unlisted
Card" at the bottom of da list if yer card ain't here): pick a server
probin probe fer clocks probin dun probin to begin RAMDAC configurashun
screen configurashun pick video modes gimme sum video modes ta use. 8 bit
is fer 256 cullers, 16 bit is fer 64 thousand a dem cullers, and 24 bits
gives ya true culler. ya lose some performance tha higher ya go. you
needs to pick one of the thangs below. skip it! dere was an error findin
the video ram on yer card. configurate it by hand. dis here program makes
a XF86Config file usin informashun that you give.
da XF86Config file usually hides in /usr/X11R6/lib/X11 or /etc/X11. you
gits a sample file from XFree86; it comes configurated fer a normal VGA
card and a 640x480 moniter.
ya can take the sample XF86Config and edit it yerself (we'uns don't
recommend that) or (and dis is da part we like) you can continue dis
program to build one fer ya.
you might need ta know the chipset and amount of memories your video board
has gots, but prolly not. can't probe - error runnin X
video memory warnin what kinda moniter you got? if'n you wanna give your
own sync frequency of yer moniter, pick "Custom" from the list. which
clockchip you got? which RAMDAC you got? which server you need?
Xconfigurator did sum probin and found yer card. da default video mode
will be:
color depth: %s bits per pixel
resolution : %s
you wanna keep dat, or pick em yerself? Xconfigurator needs to setup a
default restitution and culler amount. lots of PEE-SEE-EYE video cards
like bein probed and Xconfigurator will figger da best video mode fer you.
if'n you want ta give the required informashun instead of bein probed
then answer "don't probe" to the followin question. Xconfigurator will now
run the X server you picked to probe a bunch of stuff about yer card. it
is normal fer the screen to blank a bunch a times. yep ya gotta gimme da
horizontal sync range of yer moniter. ya can select one of da ranges
below or give a specific range.
it's VERY IMPORTANT dat you don't give a sync range higher than whatcher
moniter can really do. doin that could blow the thing up. pick something
low if you aren't sher. ya gotta give da vertical sync range of yer
moniter. ya can pick one of tha ranges below or give yer own. yer dun!
End Of The RatHead Stuff
DESCRIPTION:
LILO doesn't work. LI..101010101010101010101...
SYMPTOM:
See above.
SOLUTION:
>Can you help? I tried to install Red Hat 5.2 on my computer alongside
>Win95. I followed all of the instructions on the screen, and it all
>seemed to go well (but I've never installed it before),
So far, so good, Chris.
>but when the
>time came to restart the computer the booting up process stoped with
>'LI' written on the screen at the point where in Windows the autoexec
>message would have been displayed. I tried a few things but nothing
>worked. So I restarted the computer using a Linux boot disk that was
>made during the installation process. The computer loaded until the'LI'
>appeared, but this time instead of stopping it kept on printing the
>number 10 to the screen over and over again. Now the computer doesn't
>work in Win or Linux.
This is the "1023 cylinder problem", aka "512 MB problem.
It's discussed in detail in RedHat 5.2 Installation Book page 25ff.
You need to read that.
What I'm writing now is ADDITIONAL information to help you AND OTHERS.
EVERY PC has a BIOS chip.
BIOS means Basic I/O System.
I/O means Input Output.
The design of PCs, and especially of new PCs is to make a Microsoft Windows installation easy.
Therefore, it's a bit more difficult today because manuals aren't shipped anymore with PCs or
even with motherboard, where the problem is.
But things never changed.
The PC is nothing but some pieces of metal, plastic, silicon chips and wires and such...
If you connect the PC (I'm talking about a PC with a monitor and keyboard attached),
well if you connect the PC to power,
the BIOS chip gets working.
The BIOS chip has programs in it, that run at this time.
They do need memory, so a PC with all memory pieces removed will never work.
The BIOS gives birth to the PC as YOU know a PC.
Nothing more, no education, no advice.
What the BIOS chip tells the fresh awakened PC is what you see when you hit "DEL" during
booting and scroll down the various items.
Careful here !
Not everything you see will be told the PC during booting. A lot of stuff will be told later.
Sometimes, too late.
Today it seems to be very difficult for folks to accept, that a PC only sees the first two
drives, and only the first 1023 cylingers ("512 MB") of these two drives.
These are the drives connected to the "PRIMARY channel".
You don't need to understand this, but you MUST know and accept it.
RedHat gives a lot of information about this in their latest 5.2 distribution.
READ IT if you have it !
As of today, ANY operating sytem, wheather DOS, Windows, Linux, or whatever, has to be started
from this virtual "launching pad".
(Some new BIOSes allow more than these "ever since" scenario, but don't hold your breath.)
The PC's "runways" for EVERY operating system are:
The diskette in the first floppy drive.
The "Master Boot Record" ("MBR") on the first hard disk drive ("C:").
The "Master Boot Record" ("MBR") on the second (PHYSICAL) hard disk drive.
Optional: A CD-ROM attached to any "channel".
Optional: SCSI stuff, Network information, ......
This limit to 1023 cylinders or 512 MB (the later one is not really correct and exact) is BAD,
but of historical reasons we have to live with it,
if you install Linux or whatever second operating system onto an "old" computer (meaning a
computer that runs already Windows).
The BIOS can not see further than the first two drives on the first channel.
If you have 2 harddrives and a CD, you - hopefully - have a configuration like this:
Channel 1 Master: HDD
Channel 1 Slave : HDD
Channel 2 Slave : CD
Since this is a GENERAL HELP (unless someone asks for details),
the GENERAL ADVICE IS (BECAUSE THIS WILL NEVER FAIL):
Install Linux whereever you want to.
Install Linux in a way, that the kernel (the core of the operating sytem) is below cylinder
1023 on one of the first two physical hard drives.
Install LILO (the LInux LOader) on the Master Boot Record of the first physical hard drive
("C:").
You can install Linux on the fourth drive if you want to,
BUT
the Linux-kernel (the "nucleus", "core of the operating system") MUST BE VISIBILE AT BOOT-TIME
to the BIOS.
That means it MUST be on the first or second physical HDD completely below cylinder 1023.
TO INSTALL LINUX:
Start the PC, hit "DEL" during booty time, and go to "STANDARD CMOS SETUP".
Hit .
You'll see your hard drives listed:
>>> THIS IS AN EXAMPLE <<<<<
HARD DISKS TYPE SIZE CYLS HEAD PRECOMP LANDZ SECTOR MODE
Primary Master User 6449 784 255 0 13327 63 LBA
Primary Slave User 2564 5300 15 65535 5299 63 NORMAL
Secondary Master User 528 1024 16 65535 1023 63 NORMAL
Secondary Slave None
This is the table of Darkstar, my main Linux workstation.
The first drive (6,449 MB) is set to mode "LBA" ("Logical Block Adressing") for historical
reasons and because there's also a Windows system on the drive, that has of course the same
problems as Linux has. (If someone wants to know why you never see problems with drives under
Windows, ask me.)
The main Linux partion is on the first drive, but 3709 MB away from the start of the drive.
The BIOS-chip inside the PC would never allow Linux to start, since this chip can not "see"
that far.
To boot an operating system, it's not necessary for the hardware to read your diary, somewhere
on a compressed drive or whereever, but the system's "core" (no matter what system).
DOS/Windows manages this, by placing COMMAND.COM and Company close to the beginning of the
drive.
If you install Linux, you must do the same.
The "core" of Linux (the "kernel") is on the second hard drive, on the first partition.
As you can see above, that drive has a "geometry" of 5300 cylinders, 15 heads, 63 sectors.
At first sight, this would never work.
The trick is to split the drive into logical partions.
The Primary Slave - drive has three logical partions (created during Linux-setup):
hdb1 - /boot
hdb5 - swap
hdb6 - /pub
The "/boot" - partition is only 16 MB big, and so it is completely below cylinder 1023 (and
inside the first 512 MB).
When the system starts, the BIOS knows only about the Primary Master and the Primary Slave -
drives. (THIS system knows more, but that might not be true for your system.)
LILO - the fine thing to start both Windows and Linux as you like it, is installed inside the
"MBR", the master boot record, the only suitable place for a PROGRAM that starts automatically
(and for boot-sector-viruses of course).
"LILO" IS A PROGRAM.
It runs and prints on your screen.
It should print "LILO.....", and when it stops and displays only "LI" or/and followed by 1s
and 0s, the configuration is false.
The core of DOS/WINDOWS has to be close to the beginning of the drive.
The Windows-installation or the DOS-command "sys" takes care about that.
The core of Linux has to be close to the beginning of the drive.
More detailed: Has to be where the BIOS can see it.
As I explained, the BIOS can definetely see the first 1023 cylinders of the first two physical
drives.
Since the Linux core (the "kernel") is on the beginning of the second physical drive, there's
no problem.
The BIOS ("BASIC Input Output System") can see the "Master Boot Record" of the entire system,
which is usually on the first drive ("C:" in DOS-terminolgy, hda in Linux-terminology).
Every hard disk has a "boot record", but one has to be the PRIMARY one, the "MASTER", to avoid
internal discussions inside your PC.
Democracy doesn't really work for humans, and for computers it would mean pure chaos and no
computer would ever work.
In case the Master Boot Record (the beginning of your first drive) can be recognized as such
by the BIOS, the BIOS assumes that the MBR is able to take care of business now and hands over
control.
The program stored in the MBR ("Master Boot Record") now starts.
That might be a "boot sector virus".
That might be a set of commands like "load COMMAND.COM, execute CONFIG.SYS if present, execute
AUTOEXEC.BAT if present... and so on".
That might be LILO.
LILO
starts an operating system - that needn't be Linux. It's just a Boot Loader, like the one that
shipped with OS/2.
To do so, it must load the core of that operating system into memory, than give it a kick, and
exit. Now that operating-system-core is active and starts DOS/Windows or Linux or whatever.
Again, the core must be completely accessable.
At the time LILO starts, it has no help exept from the BIOS, and that is BASIC I/O.
Not full-featured, not able to read the 5300th cylinder of the fourth drive, ONLY BASIC things
like the first 1023 cylinders of the first two physical drives.
On my system the Linux kernel is stored in the first partition of the second drive.
With the help from the BIOS, LILO can completely "see" and read that "kernel", load it into
memory and give it a kick to run.
LILO can not see all the data on the third partition of the second drive ("/pub" over here).
LILO has not the slightest chance to see the Linux "root" partition ("/"), since this starts
3709 MB away from the drive's beginning.
It doesn't need to.
LILO starts the Linux kernel.
The kernel has I/O routines that allow to see the entire PCs hardware as well as drives on
other computers with different operating systems on the other side of the planet.
When starting the kernel, LILO gives it the information, where the "Linux root partition" is.
You can check that on your system with the command:
# cat /etc/lilo.conf
I hope this is clear, if not, read it again.
After the third time, email me your questions.
> The cover just happened to be off because I had
>installed an extra hard drive earlier in the day, so I swaped the hard
>drive with windows on it for the hard drive with Linux on it (at the
>PCI terminals on the mother board). When I restarted the computer it
These are not PCI connectors.
What you did is changing the CHANNELS, not just PRIMARY to SECONDARY.
>started printing the number
> 10s to the screen again (This time a lot faster as they were being
>read off the hard drive instead of a floppy). So I gave up trying to
>get the Linux to work as I am very inexperienced with it, and probably
>doing something wrong, and got Windows working again so I can get
>online and try to get some help with the installation of Linux as I
What did you changed to get Windows working again ???
>realy would like to get it working and see what makes it so good. Any
>help would be greatly appreciated.
Re-boot, including disconnecting the power, to make sure your Windows stuff keeps working.
Your new Linux drive should be the Primary Slave drive.
To make it that way, check the drive for jumpers that have printings nearby saying "Master",
"Slave", "CSEL".
If you find that on your dedicated Linux drive, set the jumpers to match the "Slave" position.
Insert the drive and connect it TO THE SAME CABLE THAT LEADS TO YOUR WINDOWS DRIVE.
This makes it the "Primary Slave" and will allow LILO to boot it.
Your /boot - partition, which contains the kernel, MUST be partition below cylinder 1023.
Refer to RedHat's 5.2 book pages 25ff.
If you need futher help,
hit "DEL" during booty time, and tell me what drive configuration you see.
I need to know as much details as you can give me,
but you should already (now) be able to get it running.
If you try to use your already installed Linux-stuff, you MUST go thru the installation
process again, so LILO has the new configuration (the linux drive as the primary slave).
Just don't format the linux-drives again, so you keep the stuff you already installed.
A complete new install however is recommended,
but ONLY if you are sure the drive is Primary Slave.
If you've any doubt,
feel free to ask again,
george./
Joerg F. "George" Wangerin - IT Professional since 1978
-=- jfw Computing Consulting / HelpDesk / Think Tank -=-
mailto:jfw@rp-plus.de
http://titan.glo.be/jfwcc/pull/index.html
phone: +32 (Belgium) 58-243036 /// fax: +32 (Belgium) 58-243730
>> System- & Application- Programming, Troubleshooting,
>> Connectivity, User- & System- Interfaces, Online-Systems
--- PCs: DOS based O/S, Linux (POSIX compliance)
--- GUIs: MS-Windows, X Window System, X
--- UIs: DOS, TSO/ESA, native UNIX
--- MAINFRAMEs: MVS/ESA (incl. JES3), UNIX (incl. AIX & PRIME), VM
--- CONNECTIVITY: RPC, NetView, Custom Solutions
+++ FATAL ERROR: Unable to locate cup~o~chino. Operator halted. +++
DESCRIPTION:
Manual Pages don't work.
SYMPTOM:
Under X, clicking to get some manual pages,
cause "xman" to crash and terminate,
with no messages on your X screen.
SOLUTION:
The compressed pages get copied to /tmp, uncompressed and displayed.
For whatever reason, sometimes "xman" crashes with a floating point exception.
(You don't see the "Floating Point Exception" message.)
Type: "find /* -depth -name [NAME_OF_THE_COMMAND_YOU_CLICKED_ON]* -print 2>>
/dev/null"
(Don't forget the "*" after, for instance, "Mail" or "xsetroot" or whatever you wanted to
see.)
Type: "cd /usr/man/preformat/cat1 ; gunzip *"
The above solution (to unzip these files) has of course a payoff on hard disk space,
but if you got the problem you gotta fix it. And you encrease the speed.
You'll know WHICH manual page you wanted to start, since "xman" crashed at that point.
So you can find the directory where the manual page is stored:
"cd / ; find -name THE-THING-YOU-CLICKED-ON* -print"
Don't forget the asterik (that "*") !!
The xconsole-man-page (an example) is actually stored as "xconsole.1.gz" in
"/usr/man/preformat/cat1", so the name of the file is longer than what you clicked on.
When *I* get this situation, I gunzip all files in that directory.
If you're tight on disk space, gunzip just your "problem file". Maybe others work.
You can safely reverse that with "gzip *". Links remain maintained - no damage.
DESCRIPTION:
Resolution / Color - Problems.
SYMPTOM:
Under X, the resolution or color-depth is just too low.
SOLUTION:
> Thank you for replying. I tried the way you had instructed in the
> mail you had written to me. I got all the Xservers installed. But when I
> configure it to the card which I have SiS 6C215 PCI(actually there is
> only SIS 6C215 in the list of cards in the Xconfigurator, no PCI
> extension, I dont know if there is a difference), the xwindows fails to
Hi Ashwin,
there's a BIG difference in cards, if even one letter or digit is different.
That's because vendors once made themself a name with a certain card, and wanna keep that
name. So the next card has the same name at the first sight, but you'll find a tiny difference
somewhere.
What's important is the chipset your card has.
I've no information on the chipset of your card, so get the package and/or manual, and check
it.
It should be ie. something like "S3 Virge", "P9000" or such.
That might be difficult, so if you don't find the word "chipset" or "chip " somewhere,
turn off your computer, get a screw driver and open the box.
Disconnect the monitor cable.
Then get out the screw holding the card the cable led to.
Get the card out. Even if you're a layman, there's nothing to be afraid for, you won't ruin it.
Now you should have the card in your hands.
Get a piece of paper and write down EVERYTHING that's printed on EVERY chip.
There might be a paper sticker on a chip. Write down what's on that "sticker", then remove the
sticker. Use your fingernails. Again, you won't ruin something, that chip has pins about one
mm thick, you won't break anything.
Write down the printing under the sticker.
Now you should have all the information you can get.
Plug in the card again.
Pull down first the side away from the computer's "wall" (where the monitor-cable goes in).
Gently assign the card so that it goes down about 1 cm without much force.
Next, pull down the other side of the card.
That might be "impossible" at first sight/try.
Hold the card with one hand, touch the metal bar with the cable-connector with the other hand.
Push the bottom of that metal bar towards the the other side (towards the direction "into" the
computer).
Now the card should easily go down on both sides.
If not, get the other side a bit up again, so you push down the card on both sides at once,
making sure the metal bar CAN go down because you push it towards the inner side of the
computer.
If the card is properly alligned, compared with the other cards, push it down again just to be
sure. Insert the screw on the metal bar, and fix it.
If you have to move the metal bar for about one milimeter to allign it with the hole in the
comp's chassis, that's okay. It's constructed this way to make sure the card is really
connected, and that there's a steady force to make sure it stays connected if you move the
whole comp.
Connect the monitor cable again, switch on the comp.
Everything should work as usual.
If so, put the comp's cover back on and fix it with the four to six screws if you want to.
My comps are always "open", because the cover itself fits into slots you don't see easily.
So don't get nervous when the cover doesn't seem to fit - it does, you just have to move it up
and down, left and right, and so on...
Best way is to let the cover "drop" onto the comp with two centimeters distance from the
comp's front (where the switches and floppy drive are).
Then pull and push it towards the front, holding the cover's sides close to the comp.
This way the cover "snaps" into the slots you won't see unless you've done this already a few
times.
Now ship over everything you wrote on that paper.
Would be helpful if you describe what was on what chips - there are a few chips with the same
or nearly the same reading.
Start Linux and start X with the command "startx 1>> ~/Xoutput 2>> ~/Xoutput".
This will create a file "Xoutput" in your home directory with all the output the current X
server generated.
Ship over this file too.
I case you dunno how:
If you have access to your windows-partitions from within Linux, you should know how there
names are. If unsure, type "cat /etc/fstab", what you'll see on the screen will assist you in
finding the name(s). Where the third row says "vfat" or "msdos" or "dos" the second row of
that line shows the name of your "DOS partition(s)".
You can copy the file, ie. "cp Xoutput /dosc/windows/temp/".
So in Windows, you can open Notepad and open the file C:\Windows\temp\xoutput".
Choose "all files" in notepad or rename the file in a DOS-box with "ren
c:\windows\temp\xoutput c:\windows\temp\xoutput.txt" so that it has the ".txt" extension that
tells Windows that the file is a text file and okay to open with notepad (or wordpad or
whatever).
You can also just open the file with your email program and include it in the body of your
reply.
> start and shows error like SIS()X000) unknown chipset then something
> like memory 64k no memory to run higher resolution etc.
>
> I have 2mb of memory and its shown during start up of the bios plus I
> also have windows on my system which does not give me any trouble on the
> resolution.
The X-server doesn't recognize the card completely.
The server doesn't know what chipset the card has, and can not read the video-RAM.
These 64k are not taken from the card, but from the Linux kernel, to ensure that you can at
least see somewhat so you can fix the problem.
> But when I set by using Xconfigurator to Generic VGA, I have no problem
> in getting xwindows started except the fact that the color is 16 colors.
That's because the video-RAM isn't recognized, so the kernel jumps in and provides 64ks.
That's of course not enough to allow a higher color-depth or a higher resolution, since
everything that you see on the screen is actually written to memory (normally the video-card's
memory).
The video-card then routes these data to the monitor.
This is how a monitor works (in principal, not in detail).
> I had read in a magazine to set some path or so which looks like
> X11-->SVGA etc (I am not sure it looks exactly like this but similar).
This means that the command "startx" calls the X-server "SVGA".
The "->" indicates that the file "X11" is a link to the file "SVGA".
A "link" is a bit like a "shortcut" in Windows.
> I hope you have some alternative for me.
Sure.
But before you buy another card, let's try to get the card working that you already have.
But you might need to buy a card that is reasonable supported by some X-server.
I had an ATI monster-video-card, that was listed as supported, but then it turns out, that
resolution and color-depth were so low, that I had to replace it with a much much cheaper
card, which provides much much better quality, because the X-server knows the chip in and out
and can make use of all it's features. ATI doesn't disclose the internals of there cards,
hence it's just to much work to try and guess to build a server for it (or to include the
support for that card into an existing server).
Ship over the infos, and I'll try to get that card working,
george./
Joerg F. "George" Wangerin - IT Professional since 1978
-=- jfw Computing Consulting / HelpDesk / Think Tank -=-
mailto:jfw@rp-plus.de
http://titan.glo.be/jfwcc/pull/index.html
phone: +32 (Belgium) 58-243036 /// fax: +32 (Belgium) 58-243730
>> System- & Application- Programming, Troubleshooting,
>> Connectivity, User- & System- Interfaces, Online-Systems
--- PCs: DOS based O/S, Linux (POSIX compliance)
--- GUIs: MS-Windows, X Window System, X
--- UIs: DOS, TSO/ESA, native UNIX
--- MAINFRAMEs: MVS/ESA (incl. JES3), UNIX (incl. AIX & PRIME), VM
--- CONNECTIVITY: RPC, NetView, Custom Solutions
+++ FATAL ERROR: Unable to locate cup~o~chino. Operator halted. +++
DESCRIPTION:
Error messages during system boot,
when starting vi; screen looks different.
SYMPTOM:
You can not use "vi",
the screen looks different when using "more",
error messages during boot... etc...
SOLUTION:
You upgraded to Linux 2.0.x, right ? Congratulations !!
For whatever reason, environment variable TERM is set to rubbish, to: "linux".
"linux" however is not a terminal.
Repair it:
Either boot with an old zImage (floppy-boot)
or
type: "TERM=console ; export TERM". Be sure you do this in a bash-shell.
Now your editor works again, so you can repair it:
"vi /etc/profile"
Start straight below the comment lines and insert two lines:
Change
"if [ "$TERM" = "" -o "$TERM" = "unknown" -o \"
to
"if [ "$TERM" = "" -o "$TERM" = "unknown" -o "$TERM" = "linux" -o \"
BTW, here you can also add:
"/bin/setterm -blank 30 # screen goes blank after 30 minutes"
"/bin/setterm -foreground cyan # cyan on black is better for my eyes and taste"
"# than white on black..."
"/bin/setterm -store # just storing these values..."
"MOZILLA_HOME=/usr/local/netscape"
"export MOZILLA_HOME"
... and whatever you like to set for everybody...
BACKWARDS COMPATIBLE TO LINUX 1.2.8 !!
Reboot and ENJOY!!
DESCRIPTION:
After upgrading to a 2.0.x kernel,
your PPP connection doesn't work anymore.
SYMPTOM:
You get nothing but the error message,
SOLUTION:
Your kernel's part of the PPP (Point to Point Protocol, but if you're a serious
reader, you know that...),
well your kernel doesn't support your ppp daemon "pppd" or vice-versa.
Upgrade your kernel by applying the latest patches.
At the time writing, I'd recommend going for Linux 2.0.35.
See /usr/src/linux/README for instructions - on YOUR SYSTEM of course.
