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

Re: Bug#752742: udev: fails to boot if CONFIG_UEVENT_HELPER is disabled



On Thu, 2014-06-26 at 15:05 +0200, Michael Biebl wrote:
> Am 26.06.2014 08:43, schrieb Arto Jantunen:
> > Package: udev
> > Version: 204-12
> > Severity: normal
> > Tags: patch
> > 
> > The udev initramfs script is executed with set -e, and attempts to clear
> > out the contents of /sys/kernel/uevent_helper at boot. If
> > CONFIG_UEVENT_HELPER is not set this file doesn't exist, setting it
> > fails and the boot stops. The script should check that the file exists
> > and is writable before trying to write to it.
> > 
> > A trivial patch to do that is attached.
> 
> Makes sense.
> 
> > Additionally udev.init and udev.postinst have checks for the existence
> > of that file which would stop udev from being started under sysvinit if
> > that compatibility option is not set. I'm not entirely sure what should
> > be done about them.
> 
> Seems reasonable to drop the
> 
>   if [ ! -e /sys/kernel/uevent_helper ]; then
>     echo "udev requires hotplug support, not started."
>     return 1
>   fi
> 
> checks from postinst and the udev SysV init script.
> udev has been switched over to use the netlink based interface for
> communicating with the kernel a long time ago.
> 
> Marco, at least I don't see a good reason to keep them. Do you agree?
> 
> > The kernel team should possibly be notified about keeping this option
> > enabled for the jessie release to avoid breaking partial upgrades.
> 
> I don't think they have any plans to change that. But CCed the kernel
> team just in case.

I wasn't aware it could be disabled... indeed it still can't be in
released kernel versions, but I see it can be in Linux 3.16.

To avoid upgrade problems, I'm happy to keep it enabled for jessie if we
use 3.16, but would like to disable it for jessie+1.

Ben.

> Just curious, was this triggered by [1]. Usually CONFIG_UEVENT_HELPER is
> set (at least by distros for compat reasons) and contains "".
> 
> Cheers,
> Michael
> 
> [1] http://www.spinics.net/linux/lists/kernel/msg1770793.html

-- 
Ben Hutchings
For every complex problem
there is a solution that is simple, neat, and wrong.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: