udev [059-070] bug discussions

Message

macxs · Post by **macxs** » Fri Sep 09, 2005 3:32 pm

I got a predefined file from a friend. Thanks.

drphibes · Post by **drphibes** » Thu Sep 15, 2005 5:56 pm

thread renamed in case anyone is having difficulty with 069 or 070. i upgraded from 068 to 070 without incident.

stalynx · Post by **stalynx** » Fri Sep 16, 2005 1:08 pm

From 068 I have been unable to read from my cdrom as user... however I can mount as user.

relevant udev line

Code: Select all

ENV{ID_CDROM}=="?*",            SYMLINK+="cdrom%e", GROUP="cdrom"

user groups

Code: Select all

uid=1000(stalyn) gid=100(users) groups=10(wheel),18(audio),19(cdrom),27(video),35(games),80(cdrw),100(users),409(plugdev)

relevant permissions

Code: Select all

lrwxrwxrwx  1 root root 3 Sep 16 08:42 cdrom -> hdb

brw-rw----  1 root cdrom 3, 64 Sep 15 14:13 hdb

fstab entry

Code: Select all

/dev/hdb                /media/cdrecorder       auto    user,exec,noauto,manage$

versions

Code: Select all

[ebuild   R   ] sys-fs/udev-070
[ebuild   R   ] sys-apps/hal-0.5.4
[ebuild   R   ] sys-apps/pmount-0.9.3-r3

any help would be greatly appreciated.

wnelson · Post by **wnelson** » Fri Sep 16, 2005 1:57 pm

For fstab I have for my cdrw

/dev/hdd /mnt/cdrom iso9660 ro,user,noauto,unhide 0 0

Walt

linkan · Post by **linkan** » Thu Sep 22, 2005 11:17 pm

OK, just for info, going to try to replicate tomorrow and file a bug if it persists.

With udev 0.68-r1 my stable system wont boot, it complains about not being able to check /dev/hdb2, the normal press ctrl+D or root ruckus...

Code: Select all

/dev/hdb2 = boot (ext3)
/dev/hdb3 = swap
/dev/hdb4 = / (reiserfs)

First time after entering root pw I tried doing a manual fsck, but it could find the device. Did a ls -l /dev/hdb* and it listed hdb[5-20], none of the ones I need

same thing with hda and hdd. ctrl+d to logout and it rebooted like it was supposed to.

reboot (took a loooooong time at setting up udev), repeat, but after doing ctrl+d it continued bootup, X crashed of course though, but i could log in. / was mounted, but the device didn't show in /dev, set /etc/conf.d/rc to use udev explicitly and set RC_DEVICE_TARBALL=no.

reboot (same loooooong wait at udev). won't get past setting the time.

<browse forums on laptop>
live-CD, chroot, emerge udev-058. working system again, turn on music, write this.

sleeep.

PseudoRE · Post by **PseudoRE** » Fri Sep 23, 2005 12:45 pm

i have the same problem as you linkan. im on the live cd right now looking for a solution
i will edit this post if i find anything

jd5419 · Post by **jd5419** » Sun Oct 02, 2005 7:50 pm

I'm having problems too, but i have a feeling that my problems are worse then yours.. i cant emerge udev-058 so things are not looking good for that box. i miss devfsd.

bakaohki · Post by **bakaohki** » Mon Oct 10, 2005 3:31 pm

Some extra stuff to the /dev/cdrom vs /dev/cdroms/cdrom0 madness: the hdparm init script still uses the old format, so everyone should check if his or her cdrom drive is correctly set up for dma.

For the udev-68-r1 boot problems. It happened to me after a full system backup (missing partition problems), so here's what I did:
1. Live cd, chroot (etc-update, source profile) mount offline gentoo partitions.
2. delete everything from dev
3. mknod null and console
4. emerge system and/or bash udev sysfsutils baselayout sys-libs/readline sysvinit
5. etc-update and reboot

CodeHacker84 · Post by **CodeHacker84** » Wed Oct 19, 2005 4:23 am

Had some trouble updating to udev-070. udevstart ended up segfaulting. Have heard some whispers about this being related to the file /etc/nsswitch.conf, but mine is exactly the same as the ones that apparently work.

<complain>
I don't really want to mess with it too much, since udev-068-r1 works just fine. Think I'm going to keep a known-good version of udev packaged and ready to go on the root partition from now on. Doesn't it stink that an update to the device filesystem is now persistant across kernel builds...I remember when a problem like this was easily solved by downgrading to a backup kernel...didn't even require a LiveCD.
</complain>

BlueByte · Post by **BlueByte** » Wed Oct 19, 2005 11:43 am

udev-070 doesn't work my amd64 system! segfaults

Vandroiy · Post by **Vandroiy** » Wed Oct 19, 2005 1:16 pm

Please help. I have updated udev from v68 to v70 this morning and cannot boot my system now. I get a

Code: Select all

/sbin/rc: line 93: 997 Segmentation Fault  /sbin/udevstart

and after this it is unable to mount my root partition. It says it may have a bad superblock, but everything's fine when I mount it in Knoppix. I tried to emerge back to udev-68 but the emerge command doesn't work in the emergency console I get after this error.

Vandroiy.

nichocouk · Post by **nichocouk** » Wed Oct 19, 2005 1:25 pm

Vandroiy wrote:
Code: Select all
/sbin/rc: line 93: 997 Segmentation Fault  /sbin/udevstart

Same here after update to udev70. I'm going to try to downgrade to the previous version.
For that, Vandroiy, it should be possible to mount from the Live CD, chroot and mount and stuff, and redo the emerge. Well, I'm gonna try myself and let you know...

Vandroiy · Post by **Vandroiy** » Wed Oct 19, 2005 2:14 pm

Thanks a lot, nichocouk! After chrooting I was able to downgrade to 068 and everything's fine again

Vandroiy.

tomas.rollo · Post by **tomas.rollo** » Wed Oct 19, 2005 2:23 pm

I got the same problems with udev 70 too. Luckily I keep one of the older kernels without udev as a boot option so I could boot via it and downgrade to 68-r1.

BTW I got terrified at the first glance when I saw messages about system being unable to read /dev/hda3, but after a boot from Slax, I found the data safe and untouched. What a relief!

If anybody asks, there's already a bug about this in bugzilla: [bug=109789]http://bugs.gentoo.org/show_bug.cgi?id=109789[/bug]

nichocouk · Post by **nichocouk** » Wed Oct 19, 2005 7:57 pm

Yep, I managed to boot again after downgrading to 068-r1 from Live CD.
However I am not able any more to start X... Problem with nvidia it seems, and I've re-emerged nvidia-kernel without any success. Could it be related?

Anyway I'll try again tomorrow after a new emerge sync as there is a new udev ebuild (070-r1) which should be working.

wyv3rn · Post by **wyv3rn** » Wed Oct 19, 2005 9:30 pm

I'm also having problems with udev-70 segfaulting on boot. I believe it is trying to overstep resource limits and change the .core file size from 0 (core files disabled) to 4096 (I don't want core files all over the place) and the grsecurity kernel is denying the change. I don't know why udev doesn't handle this situation gracefully and just say ok, no core files then. Another machine (also using grsecurity kernel) was upgraded to udev-70 and has no problems. I suppose I will go back to udev-68-r1 on the problem machine.

Similar to this, except it does stop everything, udev doesn't run and doesn't populate /dev:
http://forums.gentoo.org/viewtopic-t-28 ... db07ab16ee

dmpogo · Post by **dmpogo** » Thu Oct 20, 2005 4:17 am

nichocouk wrote:Yep, I managed to boot again after downgrading to 068-r1 from Live CD.
However I am not able any more to start X... Problem with nvidia it seems, and I've re-emerged nvidia-kernel without any success. Could it be related?

Anyway I'll try again tomorrow after a new emerge sync as there is a new udev ebuild (070-r1) which should be working.

perhaps you have deleted nvidia devices /dev/nvidia0 and /dev/nvidiactl along the way. They are not created automatically now with 2.6.13 lernel, it seems. You have to do it by hand. There was a lot of discussion of this on the forum last week.

nichocouk · Post by **nichocouk** » Thu Oct 20, 2005 7:52 am

Thanks dmpogo, you're right. My upgrade to 070-r1 worked smoothly and reboot was no problem any more , but I'm still having the X problem and indeed, I've lost the 2 files you mention. I'll read these threads to see how I can solve that.

Cheers,

Bob P · Post by **Bob P** » Thu Oct 20, 2005 8:13 am

i got b0rked by udev-070 as well. add that ebuild to the long list of ebuilds that have had insufficient QA.

nichocouk · Post by **nichocouk** » Thu Oct 20, 2005 8:29 am

nichocouk wrote:Thanks dmpogo, you're right. My upgrade to 070-r1 worked smoothly and reboot was no problem any more , but I'm still having the X problem and indeed, I've lost the 2 files you mention. I'll read these threads to see how I can solve that.

Cheers,

Just to make it clear for others who would have the same problem: I ran

Code: Select all

/sbin/NVmakedevices.sh

and yipeee, all is back in place!

Bob P · Post by **Bob P** » Thu Oct 20, 2005 9:47 am

at this point, the ebuild is fixed and has been replaced by an updated revision. a simple fix is to just resync and emerge the newest package.

wyv3rn · Post by **wyv3rn** » Thu Oct 20, 2005 4:02 pm

I downgraded back to 68-r1 last night w/o issue. I tried 70-r1 today and the problem is gone, works great.

Merlin8000 · Post by **Merlin8000** » Tue Jun 13, 2006 12:48 am

I just got bitten by this bug. http://forums.gentoo.org/viewtopic-t-438226.html
I think I fixed my particular system, we'll find out in the morning when I can gain physical access to it.