Hi. I cannot use my sound because my kernel is "older" than 2.6.21. I cannot install a kernel that recent because all the distros hang in ``starting udev''. My Googling shows this to have started in early 2006 and it makes it impossible for me to install any distro after that time. My platform is _NOT_ a laptop, but an HP Pavilion a1632x (64 bit dual core AMD 64). So, it'd be nice to know if any fixes are in the pipe line. This is very frustrating!!!! Thank you.

You didn't mention what distro you are using or the ones you've tried that have the udev problem. Have you tried building your own current kernel and using it with your existing distro?

Sometimes when a bootup hangs on "Starting udev" it's actually hal and dbus that's causing the hangup. On my Slackware machines I was able to disable hal and dbus by making rc.hald and rc.dbus non-executable.

I had trouble installing SLED because of the same problem. I got it to install by disconnecting a secondary master hard disk. After installation I was able to reconnect it and everything was fine.

This problem happens with Red Hat, SuSE, and Debian. It happens with SuSE 10.2 but not 10.0, for example. I've downloaded, burned and tried quite enough different versions now. This patch:
https://bugzilla.redhat.com/attachment.cgi?id=125261
was to have fixed a remarkably similar problem 1+1/2 years ago. I see references to this symptom dating back to May of 2005. And, no, I've not built my own kernel. Also, since I'm talking about booting off the install CD/DVD, it's a little hard to fiddle with executable bits.

So, either this is a known problem hanging out or it is something peculiar to my system. But, since my system is an HP Pavilion a1632x and since that ought to be a reasonably common platform, it seems odd that this would be just me. The problem is that I must use Skype to connect with my co-workers, Skype requires audio and audio (on this year old box) requires a 2.6.21 or newer kernel. Quite the pain.

But, I do have two disks. One is the original with the Windows XP installation. Since I successfully installed SuSE 10.0, I hadn't considered that later editions might not be able to cope with two drives. I'd rather find out that there is a fix, but maybe I'll try unplugging the drive, too. (I hope I remember which was which....)

Thank you.

What kernel do you have in the system now? I had a problem like after allowing my kernel to get too old before upgrading; I wound up with a problem where I couldn't upgrade glib because my kernel was too old, but I needed the upgraded glib in order to upgrade the kernel.

I wound up saying: "chuck this, farley" and abandoning my long-time habit of piecemeal updates, and did a complete upgrade of my distro. This upgrade wound up with some trouble because ld.conf got changed incorrectly but after identifying that and fixing it, everything worked. Since that time, I have done a better job of keeping up on the upgrades.

You could well be facing something similar, and your solution might be a complete upgrade.

I stubbed my toe on that some time ago, too. That's why I generally want to avoid rolling my own kernel upgrade, though I'd do it now if I knew that there were a fix for this issue. Anyway, with three examples failing to boot the install disk (Debian 4.0, SuSE 10.1 and whatever the current Red Hat is), it has got to be some kernel issue and a "complete upgrade" is what I'm trying to do.

Oh, current kernel is 2.6.13 and, based on Googling, the dividing line is 2.6.15 and following will fail.

You might try this:

# back up the file so you can undo changes
cp /etc/modprobe.conf /etc/modprobe.conf.orig

# delete everything that is not absolutely necessary.
vi /etc/modprobe.conf
# delete everything that is not absolutely necessary.

This file seems to accumulate modules from multiple updates and
erroneous modules drastically effect boot time

I'm currently dealing with an Intel sound card issue that hangs my Toshiba Laptop.
I'm using Mandriva 2008 cooker (appox. Beta 1)
I have to boot to single mode and delete the lines from modprobe.conf to get it to boot.

Good Luck,
Darryl Scroggins

Thank you Darryl. I am probably not being very clear. Probably if I were able to be clear, perhaps I'd know the answer myself. Anyway, the hang is _on_the_installation_cd_. I could try copying the CD to disk, editing the files you mention and re-burn the bootable CD. I do know how to do all of that. I was hoping for something simpler though. Thank you again.
Regards, Bruce

Now this sounds exactly like what I'd previously mentioned with my SLED install. The installer is getting stuck trying to detect some hardware. Have you tried any LiveCD distros? Does your computer have some of the latest-and-greatest hardware, particularly stuff that udev would control: USB storage, SATA vs PATA, optical drives, etc.

My SLED install was in a virtual machine so disconnecting the secondary IDE channel was just a mouse click. My suggestion would be to try one or more live distros to see how well they handle the hardware. Ubuntu Feisty is good, Knoppix is good, and I'm beginning to like Wolvix (cub is only ~250MB).

Your specs sound awfully modern to me. I still run a Athlon 1.2GHz with 1GB SDRAM on a desktop and a Pavilion ze4315 with an Athlon 1.5GHz and 1GB DDR RAM. Both are about 6 years old now.

Those kernel versions were probably around the time of or before your system. I don't know how much different SuSE 10.2 is from 10.0, but maybe 10.0 didn't have hal to detect hardware before udev. Just guessing.

If your only problem is audio have you considered replacing the card? If the current audio device is built into the motherboard you should be able to disable it in the BIOS, then add a new card.

I use Mandriva and I have had issues on some computers that caused the install to hang.

To get around the problem I have had to install with the option "noapic".

I don't know what it does but it is necessary on about 2 of the 12 computers that I administer.

It might be worth a try.

``noapic''? Maybe ``acpi=off''? I tried the latter. It turns out that it cannot see my SATA drives. Very inconvenient.

Sigh. Yes. My 12 month old hardware is, indeed, using this new-fangled SATA thing. I did take a class on SATA devices 7 years ago, but it is true that it is only now becoming universal. Maybe just one more major release of the distros and the latest will work? Debian 4.0 did just come out. I'll try Debian version Bleeding Edge and see if I have luck.

Thanks! Regards, Bruce

This is driving me nuts.

I just installed Fedora Core 7 on a system that has a HighPoint RAID controller, and a single SATA 250 GB drive. Everything went perfectly during setup. I used the linux dd during install, loaded the driver for the RAID, and installed the OS on the 250.

Upon reboot I got the dreaded hang on 'starting udev'.

I tried to boot from 2 different live CD's [Ubuntu 7.04 and Fedora 7] and got a sata_mv: PCI error that scrolled at an insane rate.

I can boot to a prompt in rescue mode and see all the files on my RAID, but besides that, I'm F-ed. I'm so pissed that I'm about to throw this system out in the middle of the street.

Hi folks

installing opensuse 11 on my very cheap averatec-notebook i had the same issue - installation stucked at "starting udev ..."

(i guess the SIS900 is the problem ...)

Choosing "no local apic" kernel option in the installation menue was the solution ...

Good luck!
Denis

I'll try to remember that trick the next time it freezes like that.
My solution was to be unable to communicate with my co-workers and
use SuSE 10.0 until SuSE 11.0 came out. It's fixed now.

I know this thread is quite old but I would like to leave a solution for this issue as I was also struggling to come up with this problem

This issue generally occurs on virtual machines

Solution:

while booting go to the kernel edit menu
edit the kernel and make an entry of
boot the kernel

Best of luck