udev 103 freezes my machine

Se7enLC · Tux's lil' helper Joined: 22 Aug 2003 Posts: 88

udev 87-r1 works fine, but the newly-released 103 freezes my computer on startup. I forget the exact text of where in the boot process it freezes, but it's a complete hardware freeze, numlock doesn't even work, no useful debugging available.

I was able to mask udev-103 and revert back to 87-r1 without much problem (had to re-emerge coldplug and re-add /etc/init.d/coldplug to the boot runlevel).

Any idea what caused this crash? Is this a known bug or is there something I should check for?

max2k5 · n00b Joined: 15 May 2006 Posts: 35

same here.
with new udev machine freezes after udev started on boot.
but i think the reason is a new kernel, and particularly a new module, which is started by udev.
I have this bootup problem with kernel 2.6.18, and also with old udev and coldplug (which is replaced by the new udev).
With the older kernel 2.6.17 I have no problem at all. The config of the new kernel is derived from the old one with "make oldconfig", which should then really not differ much from the old one.

When I delete the modules, startup is fine.

but with the crash, my screen also turns black and logging is not started yet, so i have no idea which module is really the problem...

Help is appreciated!

max

Se7enLC · Tux's lil' helper Joined: 22 Aug 2003 Posts: 88

I too am running 2.6.18. I have a lot of hardware in this machine, so it really could be any piece causing the problem. I would suspect the TV tuner card (Hauppauge PVR-350) since it's less common a device for people to have. But it really could be anything.

max2k5 · n00b Joined: 15 May 2006 Posts: 35

i now found the problem:
i had support for intel framebuffer devices compiled into the kernel, but so as into the old kernel. But it never really worked. Anyway it was there, but not used. Now with the combination of the new kernel and the new udev system it was detected, and it tried to load it on startup. And finally this made the machine crash, with a black screen.
I figured it out with setting verbose logging in /etc/conf.d/rc and filming the output with my mobile phone. I needed this, because the output would blank to fast to see it, after loading this module.

Solution in my case:
Disable framebuffer support in kernel, and done!

I would guess in different cases you would need to figure out which module causes the mess.

Cheers,
max

Se7enLC · Tux's lil' helper Joined: 22 Aug 2003 Posts: 88

Henry78 · Posted: Thu Nov 30, 2006 9:55 am Post subject:

Same problem here. I use framebuffer a long time, an i am able to boot with an old kernel. If i use my most recent one, system freezes at start up.
As i got framebuffer in both kernels, i don't think that's the problem.

Downgrading udev didn't solve my problem. Seems like the etc-update messes something up...

I'll post again, with the latest news. Hopefully that'll be the solution ; -)

Se7enLC · Tux's lil' helper Joined: 22 Aug 2003 Posts: 88

Henry78 · Posted: Thu Nov 30, 2006 6:52 pm Post subject:

with my old (and differntly configured kernel), i am able to boot every fourth or fifth time. strange.

i'll ivestigate this in detail next week, 'cause i need the system this weekend running, so i don't dare to reboot. so no news for the next days...

nixnut · Posted: Thu Nov 30, 2006 7:49 pm Post subject:

Moved from Installing Gentoo to Other Things Gentoo.
Postinstallation problem, so moved here.
_________________
Please add [solved] to the initial post's subject line if you feel your problem is resolved. Help answer the unanswered

talk is cheap. supply exceeds demand

pillsnotbills · n00b Joined: 08 Oct 2006 Posts: 3 Location: Vancouver

I was able to boot, but the display became very dark just at "letting udev process events." Initially I thought it wasn't booting, but my serial terminal displayed the usual login prompt. Carefully examining the monitor I could see text displayed, but the colour was so dark that it almost appeared black. Video card is evga Nvidia 6800GT.

This seemed to occur after running etc-update, not the actual udev upgrade. I was able to "solve" the problem by downgrading udev (and conf files) back to 087-r1, from 103. For now I'm happy with that.

I should probably mention that I am running with kernel 2.6.18-mm3 (don't know if that's related). This is because of endless problems getting my stupid JMicron IDE ports to work (I have only PATA drives).

psychomunky · Guru Joined: 02 Nov 2004 Posts: 337 Location: Canada

I as well had issues after upgrading to udev-103.

I am not certain that it is a problem with udev per se, but that the config protection stuff doesn't properly fix up the /etc/udev directory structure. This seems to apply if you have other packages that use udev and install their own rule sets. Here are the steps I used to get udev back working again:

1. mask udev-103
2. re-emerge udev, so that you go back to udev-087-r1
3. reboot to verify that things are (temporarily) fixed.
4. copy the contents of /etc/udev/rules.d to somewhere safe
5. rm -r /etc/udev
6. unmask udev-103
7. re-emerge udev to get the basic 103 files (the rule files have changed quite a bit by the looks of it).
8. re-boot again to verify things are working

This got me to a system that was no longer spitting out errors upon boot and would actually boot in less than 60 seconds. I suspect the rest of the steps (which I am trying now) are the following:

1. look in the location you backed up rules.d to
2. for each rules file (except *-udev-*) run the following command to find out which packages should be re-emerged:

Henry78 · Posted: Mon Dec 04, 2006 11:09 am Post subject:

I tired it the psychomunkys way, but without luck.

I wanted to rebuild my kernel anyway, so i did it last night. That made it work again. My steps:

* reemerge udev < 103 && etc-update
* saved /etc/udev (espacially rules.d)
* rebooted some times, till my old kernel started

* genkernel latest gentoo-sources (2.6.18-r3)
* emerge udev-103
* module-rebuild
* rebuild (as psychomunky explained) all packages that put files into rules.d
* edit grub.conf

--
regards, Henry
Now it seems to work, though i only booted two times (really had to go to bed!). I'll report if anything breaks again

Henry78 · Posted: Wed Dec 06, 2006 10:11 pm Post subject:

And it broke again.

The boot process hangs during udev processing quite often. It seems to work better, if i keep pressing NumLock / Shift during the critical moments. But all together it's far from being OK.

So the next step would be analyzing what makes my brave computer hang itself. But I don't know how to get any useful debug info during boot. Can anybody help me with that?

--
regards

psychomunky · Guru Joined: 02 Nov 2004 Posts: 337 Location: Canada

From /etc/conf.d/rc:

pillsnotbills · n00b Joined: 08 Oct 2006 Posts: 3 Location: Vancouver

Gentree · Posted: Sat Dec 09, 2006 11:08 pm Post subject:

I've also been seeing this a lot since messing with 2.6.18* kernels.

My quick fix is to re-emerge udev, (I now have it as a package so I can emerge -k).

This seems ok for a while then it comes back.

sys-fs/udev-087-r1 , not tried 103.

:roll:

_________________
Linux, because I'd rather own a free OS than steal one that's not worth paying for.

KX7-333 , AthlonXP1800+ @2.3GHz
AthlonXP-M on A7N8X @ 2.6/2.4GHz (winter/summer)
2.6.32-hh1 : portage ~x86

Henry78 · Posted: Mon Feb 05, 2007 9:58 pm Post subject:

udev 104-r9 seems to fix it.

Henry78 · Posted: Wed Feb 21, 2007 10:46 pm Post subject:

jep. fixed it.

scawa · Posted: Sun Feb 25, 2007 11:47 am Post subject:

udev 104-r11 didn't
_________________
Boredom is a personal Defect
-- Lamar Stephens

Duncan Mac Leod · Posted: Sun Feb 25, 2007 6:02 pm Post subject:

jaa · n00b Joined: 01 Mar 2004 Posts: 4

Same here: upgraded from 087-r1 -> 104-r11, udev won't start at all, /dev is empty. Downgrade back to 087-r1 and everything is fine.

halfgaar · l33t Joined: 22 Feb 2004 Posts: 781 Location: Netherlands

I have udev 104-r11, and my computer sometimes hangs during boot. The stage where it crashes is the stage that has been added to udev to replace coldplug; the stage that tries to autoload modules. When it hangs, it hangs when auto-loading the usbstorage module. And the strange thing is, when my computer starts fine, the usbstorage module isn't loaded by this auto detect stage. So, it only sometimes tries to load this module, and when it does, it crashes.

My machine:
Athlon X2 3800+
2GB RAM
NVidia Nforce 4 Ultra chipset (on MSI KNeo4 mainboad)
Kernel: 2.6.18-gentoo-r6
_________________
Linux backups the right way.
Get surround sound working.

thepustule · Posted: Wed Mar 07, 2007 6:05 pm Post subject:

Oooooh - how I miss devfs...

halfgaar · l33t Joined: 22 Feb 2004 Posts: 781 Location: Netherlands

I upgraded to kernel 2.6.19-gentoo-r5, and I have seen usb-storage loaded in the auto detect phase several times now, so it would appear that fixed my problem. But, I'm not gonna declare victory just yet...
_________________
Linux backups the right way.
Get surround sound working.