| View previous topic :: View next topic |
| Author |
Message |
shoulder
n00b
Joined: 06 Feb 2012
Posts: 5
|
 Posted: Mon Feb 06, 2012 10:29 pm Post subject: Udev failed and local filesystems failed to mount! |
 |
|
Im about to start ripping my hair!
I've done 4 fresh installations of gentoo now, 2 on Virtualbox and 2 on the "real" hdd.
One have been succesfull and works fine, the trick i did with it was using genkernel.
My goal is to NOT use genkernel.
but ..
On ALL of my other tries i get the same error when booting (i get passt grub and it starts to mount(or whatever it dose))
First errors i get:
| Code: | *Assuming udev failed somewhere, as /dev/zero dose not exist.
*Start-stop-daemon: no matching process found.
*ERROR: udev failed to start. |
abit further down when its trying to mount the devices i get:
| Code: | mount: special device /dev/sda(x) dose not exist (all the disks show up like this)
*Some local filesystem failed to mount
|
I can then login as root but nothing is mounted nor can i see them under /dev
Anyone know what im failing with? Ive tryed doing several diffrent kernels but i get the same error anyway.
I am only using ext4 filesystems and it is enabled in kernal.
Please gentoo people, you are my only hope  |
|
| Back to top |
|
 |
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54216
Location: 56N 3W
|
| Posted: Mon Feb 06, 2012 11:33 pm Post subject: |
|
|
shoulder,
Welcome to Gentoo.
You either have something extra in the kernel or something missing.
For a wild guess, I suggest you have | Code: | | [ ] Enable deprecated sysfs features to support old userspace tools |
set on.
This prevents udev starting, then its all down hill from there as you don't get any nodes in /dev
If thats not your issue, make friends with wgetpaste and pastebin your /usr/src/linux/.config file so we can look over your kernel config.
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
shoulder
n00b
Joined: 06 Feb 2012
Posts: 5
|
| Posted: Tue Feb 07, 2012 12:06 am Post subject: |
|
|
Thank you!
Here is the .config
http://paste.pocoo.org/show/547069/
Just checked the Enable deprecated sysfs and it is not enabled. |
|
| Back to top |
|
|
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54216
Location: 56N 3W
|
| Posted: Tue Feb 07, 2012 1:18 pm Post subject: |
|
|
shoulder,
Hmm. Where to start.
You have told the kernel you have an intel core2 CPU. Is this right?
If so, go through the kernel and turn off every option that mentions AMD as you cannot use that code.
It may be harmless but its just useless baggage.
You have lots of options with debug in their names set. Unless you are a kernel developer or a kernel developer has asked you to turn them on, they must be off.
Some debug options interfere with noral operation. Turn them all off too.
| Code: | | # CONFIG_CONFIGFS_FS is not set |
should be on.
should be off. Its only useful on diskless systems that mount the root filesystem from an NFS server. (PXE booting)
In the Kernel Hacking menu, turn off everything to do with tracing, debug and checking. Tracing and cheking are forms of debug.
Its possible that with all the debug/tracing/checking going on that normal operation of udev is prevented.
Only change the configuration with Its search function will be useful to you - press /
There is lots more wrong too but lets get rid of the items that interfere with normal operation and see what happens.
IF you want a half way house because the full do it yourself approach takes too long, follow kernel-seeds.org
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
Hu
Moderator
Joined: 06 Mar 2007
Posts: 21602
|
| Posted: Wed Feb 08, 2012 3:52 am Post subject: |
|
|
| I never enable CONFIGFS_FS and have not noticed any problems arising from that decision. The help text did not make it sound like I needed it, so I left it out. |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Wed Feb 08, 2012 4:01 am Post subject: |
|
|
| shoulder wrote: | Thank you!
Here is the .config
http://paste.pocoo.org/show/547069/
| Code: | ...
CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug"
CONFIG_DEVTMPFS=y
# CONFIG_DEVTMPFS_MOUNT is not set
... |
Just checked the Enable deprecated sysfs and it is not enabled. |
You want to make the top one an empty string and the bottom one enabled. /sbin/hotplug doesn't exist on modern systems and it slows down every device event at boot. |
|
| Back to top |
|
|
shoulder
n00b
Joined: 06 Feb 2012
Posts: 5
|
| Posted: Wed Feb 08, 2012 12:29 pm Post subject: |
|
|
Thanks for all the help, I managed (after a few tries) to get a non-panicing kernel.
I think the
CONFIG_CONFIGFS_FS
and
CONFIG_ROOT_NFS=y
did the trick.
Now i just have to battle the wifi!  |
|
| Back to top |
|
|
shoulder
n00b
Joined: 06 Feb 2012
Posts: 5
|
| Posted: Wed Feb 08, 2012 2:59 pm Post subject: |
|
|
Speaking of batteling the wifi,
When i use iwconfig i get no wireless extensions on lo, sith0 and eth0.
Whatever i do i cant seem to get gentoo to find my wireless.
If i do a lspci i get that the network controller is a Broadcom Corp device 4357 rev 01.
I have tryed installing firmware for it but i seem to be doing something wrong.
Reading from this http://en.gentoo-wiki.com/wiki/Broadcom_43xx but i dont even have all of the options to choose.
Could someone send me in the right path? Im lost.
http://paste.pocoo.org/show/547809/ Here is my new kernel btw. |
|
| Back to top |
|
|
Gusar
Advocate
Joined: 09 Apr 2005
Posts: 2665
Location: Slovenia
|
| Posted: Wed Feb 08, 2012 3:19 pm Post subject: |
|
|
I think it'd be much better to open a new thread for your wireless issues.
Anyway, Broadcom device 4357 is supported by *two* drivers - b43 and brcmsmac. I see your kernel is set up for b43. In this case, you need to install b43-fwcutter-015. That exact version, older versions won't work. Then run these commands: | Code: | export FIRMWARE_INSTALL_DIR="/lib/firmware"
wget http://www.lwfinger.com/b43-firmware/broadcom-wl-5.100.138.tar.bz2
tar xjf broadcom-wl-5.100.138.tar.bz2
sudo b43-fwcutter -w "$FIRMWARE_INSTALL_DIR" broadcom-wl-5.100.138/linux/wl_apsta.o |
They download the correct version of the closed driver and extract firmware from it.
Oh, I just noticed something - you have b43 built-in, that won't work. Compile it as a module (CONFIG_B43=m). Also, activate phy-N support (CONFIG_B43_PHY_N=y). |
|
| Back to top |
|
|
Hu
Moderator
Joined: 06 Mar 2007
Posts: 21602
|
| Posted: Thu Feb 09, 2012 3:02 am Post subject: |
|
|
| Gusar wrote: | | Oh, I just noticed something - you have b43 built-in, that won't work. Compile it as a module (CONFIG_B43=m). |
Theoretically, every component should work properly when built-in. I have no evidence that your statement is incorrect, but I am curious. Could you provide a citation for why this driver must not be built-in? |
|
| Back to top |
|
|
Gusar
Advocate
Joined: 09 Apr 2005
Posts: 2665
Location: Slovenia
|
| Posted: Thu Feb 09, 2012 9:49 am Post subject: |
|
|
| It's not a hard requirement, it's just easier, because the driver loads several firmware files. There are like more than 50 of them. Not all get loaded, it depends on which particular Broadcom card you have. But if you compile the driver as module, you don't need to care about that, the module will just take what it needs from the /lib/firmware directory. |
|
| Back to top |
|
|
iamben
Apprentice
Joined: 10 May 2004
Posts: 275
|
| Posted: Thu Feb 09, 2012 4:55 pm Post subject: |
|
|
| I can't speak for the b43 module specifically, but many of the wireless drivers that require firmware do not load it @ module load time, but later when the interface is actually brought up. So in those cases it is perfectly fine to build-in the driver but still leave firmware in /lib/firmware. |
|
| Back to top |
|
|
Gusar
Advocate
Joined: 09 Apr 2005
Posts: 2665
Location: Slovenia
|
| Posted: Thu Feb 09, 2012 5:38 pm Post subject: |
|
|
| iamben wrote: | | but many of the wireless drivers that require firmware do not load it @ module load time, but later when the interface is actually brought up. |
From my experience it's actually few drivers which do that. For example, since udev>=176 was unleashed on ArchLinux users, a lot of people have a 30s hang at boot and no working wireless afterwards. It's a deliberate change in udev that causes it. If drivers did what you say, there'd be no problem with the new udev. But there is. The linux-wireless mailing list now has patches for various drivers to switch to a new firmware loading method.
Among the drivers affected: ipw2100 and ipw2200 (this is a big ouch, as they're unmaintained), the rtlwifi family (rtl8192se and company), r8712u, b43 and b43legacy, brcmsmac, ath9k_htc, p54(usb|pci|spi). Pretty much the only one of the major drivers that is not affected is iwlwifi. And the only driver where I'm absolutely sure it does what you say is zd1211rw. |
|
| Back to top |
|
|
iamben
Apprentice
Joined: 10 May 2004
Posts: 275
|
| Posted: Thu Feb 09, 2012 6:12 pm Post subject: |
|
|
| Maybe it's not as many drivers as I thought =) I know all intel iwl* behave this way |
|
| Back to top |
|
|
|
Installing Gentoo
