[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#273087: kernel-source-2.6.8: CONFIG_DEVFS_MOUNT=y makes disk partitions inaccessible



Andres Salomon wrote:

On Thu, 23 Sep 2004 12:55:04 -0700, David Liontooth wrote:

Package: kernel-source-2.6.8
Version: 2.6.8-6
Severity: normal

When I use CONFIG_DEVFS_MOUNT=y on a dual xeon, the main /dev/hda drive still boots
fine to /dev/hda1, but swap on /dev/hda2 won't mount on boot or with "swapon -a", fdisk
sees nothing, and cfdisk says "FATAL ERROR: Cannot open disk drive". Booting with
"devfs=nomount" solves the problem.

I'm failing to see the bug, here.  Your system didn't mount swap because
you told the kernel to mount devfs on /dev; devfs uses a different naming
scheme for devices.  So, you must modify /etc/fstab to use the devfs
names, or you can install the devfsd package for backwards compatibility.

CONFIG_DEVFS_FS's help screen says, "This is work in progress. If you want to use this,
you *must* read the material in <file:Documentation/filesystems/devfs/>,
especially the file README there."

The README says things like, "IF YOU CONFIGURE TO MOUNT DEVFS AT BOOT, MAKE SURE YOU
INSTALL DEVFSD BEFORE YOU BOOT A DEVFS-ENABLED KERNEL!"  I don't know how
to make it more clear than that.



This is pretty alarming and it took me a while to pinpoint the problem,
so I hope at least a warning can be inserted in the help screen.


There's far too much devfs documentation to fit into the help screen,
that's why it says to read the README.


Fair enough. Please close the bug report. However, I suggest adding the warning you quote from the README into the
CONFIG_DEVFS_MOUNT help screen -- lightly edited, "If you say yes to this option, make sure you modify /etc/fstab accordingly or install devfsd."

Imagine all the Linux users with even less clue than me -- there's just so much to learn! The main constraint on Linux adoption by the great masses is simply knowledge, so making it slightly simpler to avoid major mistakes will be greatly appreciated.

Cheers,
David










Reply to: