Yes! If you have 8M of RAM, you must enable at least 8M of swap during the installation of Red Hat. If you have more RAM, you should still enable swap, though it isn't mandatory.
Red Hat 2.1 made it out of the office with a bug in the mkswap part of the
install. It only occurs when you use swap partitions of about 40M
or greater. The solution is to enable it by hand or use our updated
image. The updated image is available from any Red Hat mirror
in redhat-2.1/updates/images/ramdisk2-beta.img (Note: This image
currently only does a text install...if you want to do the X install you
must do the following ``hand'' fix with your original images).
If you want to fix it by hand, you can do the following:
mkswap /dev/blah
swapon /dev/blah
Some laptops will report that the ``device is in use'' when trying to use the mouse. This is most likely due to the fact that the IRQ is getting stolen by the PCMCIA card services package. To fix it, add the following line to /etc/sysconfig/pcmcia:
OPTS=irq_mask=0xefff
The Linux Library CD-ROM is a couple months old and is a.out.
You need to install the aout-libs package off your CD-ROM to get it to work.
You can use glint or rpm to do the install.
It does report that you ``do not have a Red Hat system''. This is normal. It is really checking for the existence of our old packaging system, RPP. Since we don't use that, it reports that error. You can disregard it and continue the install.
The Creative Labs kit comes with the IDE Interface set to the third IDE port by default (IDE2, I think). This won't work under Linux as it only recognizes two IDE ports. What you need to do is move a jumper on the card to configure it for IDE1 (the secondary IDE port). I'm not sure if you have to make changes under DOS for this configuration to work, but your manuals should help with this (it will still work, but you may have to edit a file or two).
You may also have to specify command line parameters like:
boot: linux hdN=cdrom
If you have one IDE drive, N is most likely ``b'' in this case (ie ``linux hdb=cdrom''). If you have two, then it will be ``c''.
Under 2.x, the LILO message when you boot the first floppy says:
Sony 31&33 CD-ROM REQUIRED cdu31a=base_addr,[irq|0][,PAS]
``REQUIRED'' means just that. There has been a driver change since whatever version you had that requires you to enter a parameter to get the drive recognized. The probe was breaking other things, so they took it out.
You need to do something like:
LILO boot: linux cdu31a=0x340
Also, at some point in the install it will ask if you need to enter additional hardware params for LILO. Answer yes, and then enter whatever it was you had to enter above to get the install to work.
Get the RedHat-HOWTO from the docs directory of any
Red Hat Mirror. It will tell you all you need to know.
It appears that some 1540's and 1542's are a bit hosed by the Buslogic probe. You will usually get an error referring to ``DMA'' in the Adaptec probe when this happens. To fix it, you need to enter command line parameters as follows:
boot: linux aha154x=0x340 buslogic=0x334
This is assuming, of course, that the base address of your Adaptec is 0x340. If not, put your address in there instead. This will force the buslogic to only probe an address where your Adaptec is not.
Reboot your machine using the three disks that you made for the installation. You will most likely need to completely reinstall your system.
If you just want to reconfigure networking or install more packages, try using the control-panel tools.
Try doing a `df' at the shell prompt. If you have ``/'' or ``/usr'' showing up as 100% used, chances are your partition was too small for the entire install. Unfortunately, this does not result in any type of error messages unless you are watching closely (which most people don't when the thing tells you it is going to take 2 hours to install :-). You will most likely need to reinstall and do one of two things: a) make your root or ``/usr'' partition larger or b) install fewer packages.
Also, if none of the partitions you made for Linux is full, you
are probably just missing some packages. You will need to
use glint or rpm to install those.
Not necessarily. You should be able to use the boot disk that you made for the install to boot to your root partition on the hard drive. Insert the disk, reboot the machine, and at the boot prompt, type:
linux ramdisk=0 root=/dev/xxxx
where xxxx is the hard drive and partition number. If you
installed to the first partition of your first IDE drive,
this would be hda1. If you installed to the third partition
on your second drive, this would be hdb3. The second partition
of your second SCSI drive would be sdb2.
If you don't remember, try what you think might work. The worst that will happen is a kernel panic or the machine will hang. In either case, reboot and try another partition.
Once you get it booted, you need to edit /etc/lilo.conf and then run 'lilo'. For more info on how to do that properly, you can then read the LILO-HOWTO in ``/usr/doc''.
An example lilo.conf to boot linux and DOS is below:
boot=/dev/sda
map=/boot/map
install=/boot/boot.b
prompt
timeout=50
image=/vmlinuz
label=linux
root=/dev/sdb2
read-only
other=/dev/sda1
label = dos
You may also need a line in the ``other'' section with ``table = /dev/hda'' or whatever drive your DOS partition is on. The only lines from above that should change if you boot linux and DOS are ``boot'', ``root'', and ``other''. Some of what is there is optional, like the append and what YOU actually label the dos partition (some people prefer msdos).
One caveat: Make sure the ``boot'' line points to your first IDE drive, OR your first SCSI drive IF AND ONLY IF you only have NO IDE drives. That is the only way the machine has to know where to find it. You need to also be sure that the partition in the ``boot'' line is marked ``Active'' by either the DOS fdisk, or the linux fdisk (or cfdisk). Here are the ``rules'' of where ``/'' must live:
These rules are non negotiable.
BUT I WANT DOS TO BOOT BY DEFAULT!!!
Easy. Just change the above /etc/lilo.conf to:
boot=/dev/sda
map=/boot/map
install=/boot/boot.b
prompt
timeout=50
other=/dev/sda1
label = dos
image=/vmlinuz
label=linux
root=/dev/sdb2
read-only
and then run `lilo' after saving the file.
If you installed from CD-ROM, you should be able to do:
mount /mnt/cdrom
If you get an error that says the mount point does not exist, do:
mkdir /mnt/cdrom
If you did an install via NFS:
You need to know what type of CD-ROM you have.
The best way is to watch the boot messages as you
reboot, or look at /var/adm/messages (more /var/adm/messages).
You most likely have an IDE CD-ROM that will show up as
/dev/hdc or /dev/hdd, or you have a Sound Blaster type
that is /dev/sbpcd, or you have a SCSI that is /dev/scd0.
In any case, you can do the following:
mount -t iso9660 /dev/xxxx /mnt/cdrom
and substitute xxxx for YOUR device. You can then:
cd /mnt/cdrom
and then:
ls
and you will be looking at the contents of the CD-ROM.
XFree86 3.1.2 currently doesn't support the advanced features of the 9400. The timing for the release of the 9400 was just bad for the release dates of 3.1.1, so they weren't able to add the support in. Now that the card is pretty popular and a new release is expected soon of XFree86, we expect to see direct support in the next version of XFree86. For now, you can run resolutions all the way to 1024x768 at 60 hz. If you have trouble using Xconfigurator to set it up, you will probably find that xf86config will do the job (it just doesn't look as nice doing it :-)
There are a couple things that could be wrong. On some
386's, you need to compile your kernel with
``Limit memory to 16M?'' enabled.
Some AMI BIOS motherboards have a memory option to split
memory into two segments. I'm not sure what the option
is called, but it has something to do with ``linear''.
You'll find it in your BIOS setup.
You may need to specify the amount of memory to the
kernel at boot time from lilo. To try this, type:
linux mem=32M
at the lilo prompt. If that works, you can add that line
to the append in your /etc/lilo.conf and rerun lilo.
The example from above would be:
append = "mem=32M"
You can't. None of the current Red Hat releases support UMSDOS installs. If you have one drive that has one big DOS partition with free space you want to use, you have two options. One is to backup all your DOS data somehow and then repartition the drive into smaller partitions and then restore the DOS data back to a DOS partition. You then have a free partition for linux. The second choice is to use fips. fips is a program that will let you change your partition size of a DOS partition without harming the data on it. It is STRONGLY suggested that you backup your data before using fips. Red Hat Software makes no warranties or claims as to the suitability of using fips for this purpose nor can we guarantee that data will remain unharmed. I haven't heard of anyone having a bad experience with fips, but I won't guarantee it either.
Okay, there are a lot of questions on this, so I'll go into it in better detail.
Line 1: Our name and the version you have.
Line 3-4: short description of what to do.
The rest of the message is for lilo parameters. These
should only be used if you have a Sony CDU31 or 33 -OR-
if you are having problems with the auto detection of other
hardware.
Be sure that for every parameter you enter that has commas that
you do NOT put a space after the comma. The only spaces
on the command line should be after 'linux' and between
multiple parameters (ie: mcd=0x340,11 bmouse=5).
Line 6: This line is almost NEVER necessary and should be used
only as a last resort. If your hard drives are not being
recognized or giving odd errors, you can try these parameters.
If your first hard drive acting up, you would enter:
linux hda=xxx,xxx,xxx
where the 'xxx' is your parameters from BIOS.
Line 7: Use this if your Mitsumi CD-ROM does not get recognized
properly. An example: linux mcd=0x340,11
Line 8: This line is required for the sony cdu31a or 33a.
Use "cdu31a" for BOTH the 31 and the 33 (ie. do NOT use
"cdu33a" as a parameter on the command line).
An example: linux cdu31a=0x340,0
Line 9: If you have an Aztech CD268, try:
linux aztcd=0x340
(or whatever your port is).
Line 10: If your ethernet card is not recognized, you can
try to use these parameters. For an ethernet card on
base address 0x340 and irq 10, enter:
linux ether=10,0x340
Line 11: If your bus mouse isn't recognized, try this parameter.
Line 12: For an old Seagate ST0x controller, use this parameter.
Line 13: For a Future Domain TMC-8xx controller, use this.
Line 14: For a trantor T128, use this line.
Line 15: For an NCR-5380, use this line. For a card at 0x330
and IRQ 11 and DMA 5, you would use:
linux ncr5380=0x330,11,5
Line 16: Use this line for an Adaptec 1505, 1510, 1515, 1520,
and 1522 (and Sound Blaster SCSI). The reconnect and parity
are optional. For a CD-ROM hooked to ID 1 an a controller
with base address of 0x330 and IRQ 11 you would use:
linux aha152x=0x330,11,1,reconnect,parity
(you can try it without the last two)
Line 17: Use this if you are having problems with an Adaptec
1542 (rare).
Line 18: Use this if you are having trouble with a Buslogic
controller (also rare).
Line 19: Use this for Pro Audio Spectrums and Pro Audio Studio
cards that aren't recognized.
Line 20: Use this if you have the "normal" Sound Blaster CD
interface and it doesn't work. This is the regular ole run
of the mill Sound Blaster type that controls the Panasonic
56x drives, Matsushta Kotobuki drives, TEAC CD55A, and a
few others. If you have a Sound Blaster with something
like the Panasonic connected using base address 0x240, you
would use:
linux sbpcd=0x240,SoundBlaster
Line 21: If you have an IDE (ATAPI compliant) CD-ROM that is
not getting recognized by the auto probe, you can use this
parameter to tell it where to look. Normally, if you have
IDE hard drives, they take up the first available drive
letters. If you had 2 IDE hard drives, your CD-ROM would be
hdc. If you had one IDE hard drive, it would be hdb. It's
usually okay to just try them all. You would use:
linux hdb=cdrom
if you had one IDE hard drive.
Line 22: A reminder that you must put a "0x" in front of
all Hex values (such as base addresses).
Line 23: Instructions on what to enter.
There are people who have linux installed by some means, but need their own special boot disk to install their system. This is possible in some cases. Do not try this with kernels greater than version 1.3.48. It will not work. See below for instructions for later kernels.
First of all, we recommend you only try this with the same kernel
source version that we ship (ie 1.2.13 for 2.0 and 2.1). You need to
build your kernel on your machine and make sure and say `y' to the
following options during the make config:
Then build the kernel with make dep, make clean, and
finally make zImage. Once done, your new kernel will be
/usr/src/linux/arch/i386/boot/zImage. You now need to put
it on an install floppy. Under linux, the steps to do this would
be something like:
dd of=/dev/fd0 if=/path/to/boot/images/boot0000.img
mkdir /mnt/floppy
mount -t ext2 /dev/fd0 /mnt/floppy
cp /usr/src/linux/arch/i386/boot/zImage /mnt/floppy/vmlinuz
lilo -r /mnt/floppy
umount /mnt/floppy
You should then be able to reboot with the new floppy and the new kernel.
If you must use a kernel version greater than 1.3.48, you can follow the following steps:
make config.make config.mknod dev/ram2 b 1 1There will be some weirdness and an error or two displayed as the second ramdisk loads, but it should work.
For 3.0.3, the boot disks are now UMSDOS filesystem. This means you
can use mcopy or the copy command from DOS to put a new kernel on your
disk. If you are using a kernel version less than 1.3.48, build according
to the directions for 2.x (above). Then just copy it to any boot floppy
as linux (not ``vmlinuz'' as your normally would). You can rawrite
or dd any boot image from the 1213 directory and just overwrite the kernel.
For versions later than 1.3.48, make sure and say 'y' to the ramdisk
question in make config. Then copy your kernel to a boot image (you
can write out any boot image) as above. Then, get the ramdisk.img
from the 1357 directory and write it to disk. You can then boot using
the boot disk and the parameters:
boot: linux ramdisk=256 prompt_ramdisk=1 load_ramdisk=1
Note that this will not work for PCMCIA installs unless you also build the PCMCIA Card Services package using 1.3.x kernel sources and also install the modules and binaries on your ramdisk. This is *not* trivial.
Please see the NYS-Tips. You can find it at
http://www.redhat.com/.
You need to boot Unixware off of a boot floppy and run 'fdisk /mbr'. That should allow you to get back into Unixware, but will not let you access Linux.
The Trantor T-130B is really an NCR-5380 in disguise. But, it will most likely not autodetect. You will need to use the NCR boot image along with a boot parameter. The registers are offset by 8 from a normal card. So, for a T-130B at base address 0x350 and IRQ 5, you would use:
ncr5380=0x358,5
If you experience problems with XFree86 and your Diamond video card, you may try putting:
"Option Diamond
In your /etc/X11/XF86Config (or in /etc/XF86Config if you use that).
With every release there seem to be some things that break or are overlooked. This things are getting fewer in number, but as of 3.0.3 some problems do still exist. To find solutions to known problems, please see:
http://www.redhat.com/
and look for the product page for your particular version. These sheets only exist for Red Hat Linux 3.0.3 and later.
To get the startx command to use Metro-X instead of XFree86 you
must make /etc/X11/X a symbolic link to /usr/X11R6/bin/Xmetro
This is summarized by:
rm /etc/X11/X<br>
ln -s /usr/X11R6/bin/Xmetro /etc/X11/X
Next Chapter, Previous Chapter
Table of contents of this chapter, General table of contents
Top of the document, Beginning of this Chapter