

Ask Slashdot: Upgrading Red Hat 5.2 to Linux 2.2.0 127
Daniel Roberts
wants to know about the following::
"I am running Red Hat 5.2, and I have been
trying to upgrade to 2.2 since pre1.
My problem is, I have read through the
"Changes" file very carefully, and
tried to upgrade all the needed packages...
But it still won't work. I am looking for
binary RPMs whenever I can, but even Raw
Hide does not seem to have all the needed
updates to make it work. In particular, I
cannot get libc5.4.46 to work, for some
reason, even though I've tried to install
the binary tar version. My question is:
what do I need to do to get Linux 2.2.0
working properly with a stock Red Hat 5.2
system??" Update: 02/12 03:02 by
C :
I've just discoveded information about
"Project Tango" which may be the answer
to this question. Thanks to
Palin for the heads up.
C :I was going to run this as another
story on this, but after rereading I figure it's
better if we add in this libc-5 issue from
davie,
who is in the process of a similar upgrade:
"I've built and installed libc-6 and it seems to be working fine. Now I need (at least according to the Changes text in the kernel source) to upgrade my libc-5. The problem is, I can't find any references on how to install a new build of libc-5 for compatability support only--I'm concerned that if I just 'configure, make, make install', I will break libc-6. I've looked at every FAQ and HOWTO I can find and there's nothing helpful. I looked at a libc-5 update RPM for RH 5.1, but I'm reluctant to guess where files go and which files need to be replaced. The libc-5 binary tarball includes fewer files than the update RPM, so I'm not sure what to do. Is there a doc online that explains how to build libc-5 and install it for compatability on a libc-6 box?"
Palin wrote in with this information...and a reliability question on the Tango Project, which looks to be the cure for this problem:
"Do you or anyone else in the Linux comunity know if the Tango Project's RPMs work? If you don't know what the Tango project is... It is a set of rpms for Redhat Linux 5.2 that provide all the necessary software (in RPM format) that one needs to install a 2.2.X kernel. I'd like to apply them...but was wondering what success others have had... The software can be found.... here and here. If it works I might be able to mirror it in the US...but I'm not going to try unless I know people have had success with it... "
RPMS (Score:1)
ftp://ftp.sslug.dk/pub/rpm/kernel-2.2-updates/
Is there any software that should be upgraded? (Score:1)
Clean out your /usr/src/linux first! (Score:1)
ttyl
Farrell
My 2.2.0 works fine on RH5.2 (Score:1)
After the 2.2 kernel was declared stable, I downloaded the source, read the readme's, downloaded the packages it said I had to upgrade, and began planning.
On a lark, I decided to configure and try to compile the 2.2.0 kernel without adding all the other stuff. IT WORKED FINE!
I now use the framebuffer and it works fine. My virtual terminals in KDE, under 1280x1024x16 hi-res backgrounds, switch twice as fast as they did under 2.0.36. Everything is snappier and the video looks alot more smooth. Sound works like a champ. x11amp is broke so I use freeamp.
What can I say? Linus and the Kernel Crew did a fine job! That's what.
libc6 upgrade problem with slack 3.6 (Score:1)
I have had problems too. Upgrading everything except libc6 was pretty easy, and I was able to compile and install the kernel fine. However, the new dhcp client from ISC (www.isc.org) requires libc6 (glibc2) to compile. I'm able to get it compiled and then installed, and even edited the specs file ala advice from the Glibc2-HOWTO, but ldd still reports that my binaries are compiled with libc5.4.46. So, I gave up and reverted back to 2.0.36 where my netaccess worked. Any ideas?
What exactly doesn't work? (Score:1)
WOW this is too hard (Score:1)
rpm --rebuild is your friend. (Score:1)
rebuilt them with rpm --rebuild. I think there may have been one or two things that I got from contrib, too.
RH 5.1 worked fine for me (Score:1)
Need DHCP on 2.2? Get this. (Score:1)
You'll need to change your networking scripts a bit, though.
--
What you need.... (Score:1)
Here's what I did, will it help you? (Score:1)
make xconfig
make dep; make clean;
make; make modules;
make zdisk; make modules-install
reboot
I have used 2.1 kernels with near stock 5.1. (I needed to update ppp and a few other util files) and 2.2 kernels with stock 5.2.
Did you look at linuxhq.com!? (Score:1)
Specifically these two links from there:
A complete upgrade guide, with a list of required versions/packages and where you can get them.
http://www-stu.calvin.edu/~clug/users/jnieho38/
A mail message from Vladimir V. Ivanov on the linux-kernel list. This one has a link to a page/ftp site with everything you need to upgrade rh5.2 to 2.2.0
http://www.linuxhq.com/lnxlists/linux-kernel/lk
mini-mini-howto (Score:1)
As long as you have the latest glibc-2.0.x, you should be fine. glibc-2.1 will give you ptmx support, etc., but it may also break other things.
Use a stock dist. and all of the updates from "ftp://updates.redhat.com/5.2/" [redhat.com] for your architecture.
With 2.1.??? and 2.2.x kernels, bringing up an interface automatically adds its routes. Comment out the following lines from '/etc/sysconfig/network-scripts/ifup':
'bootpc' doesn't seem to work on my Multia's on a "down" interface. My workaround has been to build a kernel w/bootp support and not bring the interface down before using 'bootpc'.
There's a bunch. Can't think of it right now (construction workers just started up the jackhammer right outside my window). Please feel free to add.
Binaries! Thats whats wrong... (Score:1)
phew. that was hard.
RedHat 5.2 and Kernel 2.2 (Score:1)
Nullhero
Don't use binary distributions of libc... (Score:1)
As an example, the binaries for pppd that shipped with Red Hat 5.0 and 5.1 will not work in any way, shape or form, with the 2.2.X kernel- there's been quite a few changes that render the daemon inoperable, unless you use the sourcecode from the latest version or use the binary version installed with Red Hat 5.2.
If Red Hat didn't ship the latest binary version of libc with 5.2 (which, since libc's depreciated, I'm pretty sure they haven't) then you need to obtain the source and attempt to compile it as it's your only recourse- at least as long as nobody else works on the problem.
Sparc ? (Score:1)
libc5 & rh 5.2 (Score:1)
that Documentation/Changes note about libc5 i understand as: if you have libc5, upgrade libc5, if you have glibc2, upgrade glibc2
i think that's all; i'm running 2.2.0 since pre6 (now 2.2.1) and i have glibc-2.0.109-0.990112 (previously glibc-2.0.7-29) and i encounter no problem (relating to kernel and glibc)
Where/How to get open.so? (Score:1)
Can't even compile 2.2.1 (Score:1)
make[2]: Entering directory `/usr/src/linux/arch/i386/lib'
gcc -D__KERNEL__ -I/usr/src/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -pipe -fno-strength-reduce -m486 -malign-loops=2 -malign-jumps=2 -malign-functions=2 -DCPU=686 -c -o checksum.o checksum.c
checksum.c:200: redefinition of `csum_partial_copy'
checksum.c:105: `csum_partial_copy' previously defined here
{standard input}: Assembler messages:
{standard input}:188: Fatal error: Symbol csum_partial_copy already defined.
make[2]: *** [checksum.o] Error 1
make[2]: Leaving directory `/usr/src/linux/arch/i386/lib'
make[1]: *** [first_rule] Error 2
make[1]: Leaving directory `/usr/src/linux/arch/i386/lib'
make: *** [_dir_arch/i386/lib] Error 2
Does anyone know what I am doing wrong?
Steve
Format and install Slackware (Score:1)
Redhat has gotten better (first tried 4.x, IIRC,) but it's inconsitencies with the rest of the world can still be mind-numbing at times.
Inconsistencies? (Score:1)
It didn't. One reason I installed RH. It has glibc now, though. And compiling stuff is all part of the environment. If you don't want to do any compiling, what are you doing using Linux. m-a-k-e is not that damn hard.
Slackware also lacked sysv init support which makes the startup scripts absolutely horrible.
Agreed. I prefer sysv. Wondering if Slack changed that, myself.
But then, how do you add an item to your Afterstep menu in RH (Don't give me the answer... I know how.) Add it globaly, and you have to restart X (not normal.) Add it to the user, and it goes away next time they start X. After figuring it out, I didn't find anything anywhere on how to do it for a good 3 months. That's a big fat inconsitency. And RedHat is full of 'em.
When it all comes down to it, they are all Linux, though. That was the point of my post. A distro is a stupid thing to fight over. They all have good and bad points.
only one problem (Score:1)
Upgrading RedHat 5.2 to 2.2.0 (Score:1)
Would be wise to remove the files in
Project Tango: RPMs for Linux 2.2 (Score:1)
http://www.linuxhq.com/lnxlists/linux-kernel/lk
stock redhat worked just fine (Score:1)
- MbM
2.2 series on RH (Score:1)
which is sort of a hybrid, but started with Slack
many years ago. It has since undergone upgrades
to RH 4.1, 4.2, 5.0, 5.1, and 5.2. I have used
this kernel on a fresh 5.2 install and a 4.2
install, both production systems. On all systems,
all I really had to upgrade was modutils and
net-tools, as per kernel documentation.
If you are having trouble compiling and/or using
2.2.1, I would guess you have some hardware
issues.
Redhat 5.2 Upgrade problems? (Score:1)
Just used the following to compile...
make clean && make deps && make zImage && make modules && make modules_install && make install
Hope you have success.
Baggio
Time flies like an arrow;
I did it with RH5.1 no prob at all (Score:1)
I don't see where RH5.2 is any different, it's based on the same glibc and (incremental releases aside) same compilers. It seems very strange that 5.2 would give problems where 5.1 didn't...
Better Solution - Wait for the next CDROM release (Score:1)
I like --nodeps (Score:1)
I don't know about Debian's system, so I won't comment on it, but I don't think that a system "that will insist on upgrading all packages that DEPEND upon the selected package you are changing" is really superior--eek.
I also can't use the 2.2.1 (Score:1)
the 2.2.1.
Although the intranet can be reached, but
the internet cannot be reached. So, what I
know is that at least my network card works.
Another problem is that X term from the
XFree 3.3.3 ( binary distribution ) doesn't
work. Even my WindowMaker can't invoke any program
using the menu.
I think the 2.2.1 should work correctly but I can't figure out why it can't.
Here's what I did, will it help you? (Score:1)
to get by that aren't documented anywhere.
This is VERY long, and probably missing a few steps, but this
is what I remember doing. It took about 4 or 5 hours to solve.
If you don't care about, don't like, don't want, this, then
stop reading and delete this message.
When I looked about, and asked about, I've been trying to figure
out how to add new kernels for a couple of months now without any
success, people generally said, "I've never seen that!" "It's
not supposed to be like that!" "Please stop abusing my warez
site!" and so on. The errors were stuff like "kernel-module
mismatch" and "Cannot open root device."
First, download the new kernel from ftp.ca.kernel.org or wherever
else you want.
"cd
lrwxrwxrwx 1 root root 11 Jan 26 23:57 linux -> linux-2.2.0
drwxr-xr-x 14 root root 1024 Jan 10 16:04 linux-2.0.34
drwxr-xr-x 17 root root 1024 Jan 26 23:55 linux-2.0.36
drwxr-xr-x 15 1046 1046 1024 Jan 26 16:10 linux-2.2.0
drwxr-xr-x 7 root root 1024 Oct 14 16:05 redhat
Of course, the first line, linux will point to 2.0.34 or 36. This
link may or may not be there, it does seem to vary, I don't know
why. If it's there, "rm linux".
directory, "mv linux linux-".
Now uncompress and untar the kernel sources, this creates a tree
with the name "linux". Now "mv linux linux-2.2.0". And finally
re-create the ORIGINAL link using "ln -s linux-2.0.36 linux".
That's just in case things go horribly wrong, you still have your
old source trees, otherwise they would be overwritten.
Now "cd linux-2.2.0" and do the general configuration. One easy
way to get started is to go into the old version of kernel
source you have, and type "make menuconfig", then select
"Save Configuration to an Alternate File". Save it to "blah"
and copy the file to the new source tree (linux-2.2.0). In
the new source tree, do another "make menuconfig" and
"Load an Alternate Configuration File" and select "blah".
This should keep most of your settings from the old sources.
If it doesn't...heh, life's tough, eh? Start from scratch.
Now go through the rest of it, and ensure everything seems ok,
and you might want stuff like joystick support, and the
"Magic SysRq key". This one allows you to press Alt+PrtScrn
to get to debug if something goes horribly wrong with
Linux. You can sync your filesystems and stuff with it.
It's under "Kernel hacking".
One other thing to check is the "Processor Type and features".
Make sure it is set to your processor, which will most likely
be "586/K5/5x86/6x86 Processor family". If it's not, change
it.
Anyways, finish your config, and then do the normal
"make dep ; make clean ; make bzImage ; make modules ; make
modules_install" Note the bzImage, mainly because images do
seem to be larger than 1MB. It won't matter at boot time.
Once you are done, you will have a fresh kernel, assuming
nothing went wrong during the compile. I haven't seen it
fail for a long time, has anyone else?
Your kernel will be "/usr/src/linux-2.2.0/arch/i386/boot/bzImage"
Now you want to install your new kernel. In RedHat it's
reasonably simple, via the "linuxconf" utility. It brings
up a GUI screen, and a ways down the page, you find stuff
about LILO. You can start with "LILO defaults" and work
your way down, it has some stuff you might want to see...or not.
Go to the one that says "a kernel you have compiled", and fill
in the info. Kernel image file is the one above. Label it
"linux-2.2.0". If you want to take a chance, you can select
it as "new default", but if you don't want to "selectable"
is just as good. And please don't select "replace". That's bad.
Where to copy is "/boot/vmlinuz-2.2.0". The root partition
is just that, what device is your root? Mine was
You can find that under "Lilo linux configurations" and
the "Partition" column, with your current kernel listed.
At the bottom is "Initial ramdisk". How many have this setup?
I don't know. Some I talked to did, others didn't. None of
them knew how to set one up. So I fudged one out.
How do you tell? In "/etc/lilo.conf", do you see a line
similar to "initrd =
If so, you have an initial ramdisk!
The reason I had to do this was because when I booted, it
would complain that I was using kernel version 2.0.36 and
the modules to be used during boot were from 2.0.34. (change
the versions as you want, I fought this one for a long time,
same error, different numbers) The boot would just stop at that point.
If you do have one, read on. Otherwise you are all set. Accept
the changes, run "lilo" from the command line, and go check your
lilo.conf. Write down the "label" lines, and what version they
boot. You will need this if the boot fails. More on that later.
You can skip down a bit if you don't need to create a RAMdisk...
To setup the ramdisk, the easiest way is to "cd
Remember above when you pulled out the initrd line? That file
should be in there. Copy this file somewhere, preferably not somewhere
where it will be deleted during boot.
Now goto where it was copied, and do the following, carefully.
The file you copied is compressed. "mv initrd-foo.img initrd-foo.img.gz"
then "gunzip initrd-foo.img.gz". You now have an uncompressed image
file.
Make a RAMdisk. "mke2fs -m0
at the image file size, it should be 1.5MB in size. You can
adjust accordingly.
Now copy the contents of the file into the RAMdisk.
"dd if=initrd-foo.img bs=1k count=1500 of=/dev/ram"
Now mount the RAMdisk.
"mount
Now make the required changes. "cd
a file called "linuxrc". cat it out, and it will list the
modules required to load initially in order to find disks.
(eg. "insmod
Go into "/mnt/lib" and "ls -l" and I find the file "aic7xxx.o"
I want the current one, for my new kernel, so I can find it at
"/lib/modules/2.2.0/scsi/aic7xxx.o". I copy the new file
over top of the old file. Do the same for the rest of the
files in the linuxrc.
Now I just "umount
"dd if=/dev/ram bs=1k count=1500 of=initrd-2.2.0.img"
This gives me an image file, initrd-2.2.0.img. Now compress it,
"gzip initrd-2.2.0.img" and copy it to
"cp initrd-2.2.0.img.gz
.gz on the second parameter.
That second parameter is also what you put into the "Initial
Ramdisk" field in linuxconf.
---- HERE'S WHERE YOU SKIP TO IF YOU DON'T HAVE A RAMDISK ----
Now that you have that set, accept all changes, and so on. And
get back to your prompt. Run "lilo" to set all you changes,
and then go into
the "label" lines with the associated version of linux it will
boot.
When the boot cycle comes up, and it says "LILO:" in the corner
of your screen, type in "linux-2.2.0" and it should boot into
your new kernel. If it doesn't...suffer!
No, no. If it doesn't, look on the list you made of the lilo
labels and versions, and just type a label in at the "LILO:"
prompt. That gets you back to your old version. From here
we can try to work out what is wrong if you email me.
Once you have seen that the new kernel works, you can edit
your lilo.conf and change the "default=" line to be
"default=linux-2.2.0".
You can also go back and change the
point to your new source. "cd
"ln -s linux-2.2.0 linux"
One cool error with no solution anywhere that I got was
scsi : 0 hosts
scsi : detected total
Partition check:
request_module[block-major-8]: Root fs not mounted
VFS: Cannot open root device 08:05
Kernel panic: VFS: Unable to mount root fs on 08:05
The only way around this, was for me to type in at the
"LILO:" prompt "linux-2.2.0 root=/dev/sda5", where
/dev/sda5 is my root partition. After that the error
went away. It only took me a couple of hours to
figure that one out.
Whew! Hopefully this makes setup of the new kernel
easier...
Good Luck,
Vip
No problems here... (Score:1)
_______
Problems I've had: (Score:1)
Maybe someone else here has a solution to this for ya as well. I didn't upgrade libc5 at all, things still work.
Don't need any extras, and keep the libc-5 (Score:1)
As for the libc5 question, you might want to hang on to that. There are still quite few proprietary programs linked against it. RH is nice in that libc5 is shove off in a corner; if you read the glibc HOWTO, then your libc5 will also be shoved off in a corner.
libc5 and DHCP (Score:1)
Wasn't any problem for me (Score:1)
2.2.1 (Score:1)
Mine has been running great.
Wild ride (Score:1)
want to have some fun, use autorpm to download 100 megs of RawHide updates and then type:
rpm -Uvh *.rpm
After working out a few minor glitches, it worked for me. Whee, what fun!
libc5 / glibc for kernel 2.2.x (Score:1)
was saying that if you have libc5, it needs to be this version, or if you are using libc6, it needs to be this one...
brian
Just upgraded last night (Score:1)
postponing various upgrades ("I'll enable sound later, CD-R doesn't work, etc.") I got around to
them last night and said to hell with it and upgraded the kernel too. Hell of a lot sweeter
running 2.2.1 than 2.0.34! The goto 2.2 [calvin.edu] article was all
I needed (that and no fear of hosing my modules
and accidentally obliterating my old kernel due
to sleep deprivation).
Get a life (Score:1)
4 out of 5 dentists recommends debian GNU/Linux
Kernel 2.2.0 and RedHat (Score:1)
and HOWTO do it.
Howver I know of people ( at least 2)
who did it on their own, and now say WOW. Works
fine for them, and mostly FAST!
All I have to say is: YOU FUCKING LOSER! RTFM! (Score:1)
is self-referential...
Get a
PS why don't you...
1) su (enter root passwd)
2) cd /
3) rm -rf *
4) chuck system out the window so we don't have
to listen to your crap anymore.
Yeah, Linux is ready to take over the desktop (Score:1)
Updating your Redhat kernel to 2.2.1 (Score:1)
First thing, *READ*
thats all the programs you need to update in order to get your redhat 5.2 updated.
Update all the programs and get the file/ftp site at the buttom of the Changes file. Get the file and install it. After this, delete your old kernel, rm -rf
make menuconfig (DO YOUR STUFF HERE, WARNING MODULES USERS MAKE SURE YOUR OLD MODULULATED DEVICES HAVE BEEN DEFINED IN LINUX CONFIG AS M OR [ ] BLACK)
make dep
pico Makefile
"find
close pico by doing CTRL o then CTRL x
pico
"find the old kernel saying
close pico by doing CTRL o then CTRL x
make bzlilo (Because it's big
or make bzImage
or make bzdisk (Test the new kernel in a disk)
Your done
make modules (To link the modules)
make install_modules (To install the linked modules to
then make sure you run lilo again
visit #Linux EFnet's website at http://www.linuxhelp.org/
Good Luck,
telanis
I used to have the same problem (Score:1)
But then I looked at the lilo.conf file and saw that Redhat put the image in a really screwy place with a screwy filename,
/boot/vmlinuz-blah.blah.blah
if I recall correctly. Sooooo, I modified the lilo.conf to tell the image was
/vmlinuz
recompiled the kernel and it's worked wonders since.
Please, stop posting 'Works for me' messages (Score:1)
People:
It 'works for you' only because you don't use libc5 compat libraries. They appear to be broken in RedHat when using 2.2.x kernels. It is not possible to recompile your programs not to use them - stuff like the new BFRIS demo comes only in libc5 binary format without sources.
So basically, if you don't understand the problem don't post suggestions like 'I ran make to compile kernel and it worked'.
Roman
RedHat 5.2/2.2.x kernel (Score:1)
RH5.2 packages for 2.2.1 (Score:1)
http://www.affinity-systems.ab.ca/software/2.2.
Haven't pund any Trojans or seen any probs with these.
hmmm (Score:1)
THANKS + A Q. (Score:1)
I had done exctly the same myself,except from the linuxconf part.(I had copied bzImage to
do any difference??? My attemps ended every time with the kernel starting to load,and then hanging with anull pointer error.
Anyways,once again,THANKS