Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Cannot upgrade circular udev dependency

67 views
Skip to first unread message

David Liontooth

unread,
Feb 16, 2022, 10:40:06 AM2/16/22
to


Hi -- I have a machine, Linux ancient 2.6.36.2 #1 SMP Sun Dec 26 06:19:57 PST 2010 x86_64 GNU/Linux.

I can install a new kernel, but it won't boot into any of the new kernels:

root@ancient:~# update-grub
Generating grub.cfg ...
Found linux image: /boot/vmlinuz-4.9.0-16-amd64
Found initrd image: /boot/initrd.img-4.9.0-16-amd64
Found linux image: /boot/vmlinuz-3.16.0-6-amd64
Found initrd image: /boot/initrd.img-3.16.0-6-amd64
Found linux image: /boot/vmlinuz-2.6.36.2
Found linux image: /boot/vmlinuz-2.6.36.1


I also cannot upgrade udev:

root@ancient:~# apt --fix-broken install
Reading package lists... Done
Building dependency tree       
Reading state information... Done
Correcting dependencies... Done
...
The following additional packages will be installed:
  udev
The following packages will be upgraded:
  udev
1 upgraded, 0 newly installed, 0 to remove and 476 not upgraded.
Need to get 0 B/1,112 kB of archives.
After this operation, 6,381 kB of additional disk space will be used.
Do you want to continue? [Y/n]
Reading database ... 64907 files and directories currently installed.)
Preparing to unpack .../udev_232-25+deb9u13_amd64.deb ...
Since release 198, udev requires support for the following features in
the running kernel:

- inotify(2)            (CONFIG_INOTIFY_USER)
- signalfd(2)           (CONFIG_SIGNALFD)
- accept4(2)
- open_by_handle_at(2)  (CONFIG_FHANDLE)
- timerfd_create(2)     (CONFIG_TIMERFD)
- epoll_create(2)       (CONFIG_EPOLL)
dpkg: error processing archive /var/cache/apt/archives/udev_232-25+deb9u13_amd64.deb (--unpack):
 subprocess new pre-installation script returned error exit status 1
update-rc.d: warning: start and stop actions are no longer supported; falling back to defaults
update-rc.d: warning: start and stop actions are no longer supported; falling back to defaults
Errors were encountered while processing:
 /var/cache/apt/archives/udev_232-25+deb9u13_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)

I'm also unable to download or install anything else; it's all blocked by udev. I cannot for instance install any of these available versions:

root@ancient:~# just available udev
udev:
  Installed: 175-7.2
  Candidate: 232-25+deb9u13
  Version table:
     241-5~bpo9+1 100
        100 http://ftp.debian.org/debian stretch-backports/main amd64 Packages
     232-25+deb9u13 500
        500 http://security.debian.org stretch/updates/main amd64 Packages
     232-25+deb9u12 500
        500 http://mirrors.kernel.org/debian stretch/main amd64 Packages
        500 http://debian.lcs.mit.edu/debian stretch/main amd64 Packages
     215-17+deb8u13 500
        500 http://security.debian.org jessie/updates/main amd64 Packages
     215-17+deb8u7 500
        500 http://mirrors.kernel.org/debian jessie/main amd64 Packages
 *** 175-7.2 100
        100 /var/lib/dpkg/status

I'm not finding udev packages in pool. Where are they located in the repository tree? Is there an intermediate udev package that would allow me for instance to boot linux 3.16.0-6-amd64?

At one point there was an option to use "touch /etc/udev/kernel-upgrade" to overcome this exact problem, but it's not working to get me out of
2.6.36.2 and udev 175-7.2.

Finally. I tried using another machine to download udev 215-17+deb8u7 -- it lets me download libudev 215-17+deb8u7, but even just downloading the udev package is blocked by the dependency check.

    apt-get install --download-only udev=215-17+deb8u7 <== fails
    apt-get install --nodeps --download-only udev=215-17+deb8u7
E: Command line option --nodeps is not understood in combination with the other options

There are clearly lots of barriers put in to keep people from making mistakes, but it's also keeping me from upgrading. Is there a solution?

Cheers,
Dave


Greg Wooledge

unread,
Feb 16, 2022, 11:10:06 AM2/16/22
to
On Wed, Feb 16, 2022 at 07:19:34AM -0800, David Liontooth wrote:
>
> Hi -- I have a machine, Linux ancient 2.6.36.2 #1 SMP Sun Dec 26 06:19:57
> PST 2010 x86_64 GNU/Linux.

This is not a Debian kernel. You either built it yourself, or you got
it from some foreign Linux distribution.

> I can install a new kernel, but it won't boot into any of the new kernels:
>
> root@ancient:~# update-grub
> Generating grub.cfg ...
> Found linux image: /boot/vmlinuz-4.9.0-16-amd64
> Found initrd image: /boot/initrd.img-4.9.0-16-amd64
> Found linux image: /boot/vmlinuz-3.16.0-6-amd64
> Found initrd image: /boot/initrd.img-3.16.0-6-amd64
> Found linux image: /boot/vmlinuz-2.6.36.2
> Found linux image: /boot/vmlinuz-2.6.36.1

Why not? What happens when you try to boot one of them? Are they missing
whatever features you needed to build a custom kernel to support?
You could rebuild your custom kernel to include those features. Assuming
they existed in 2.6.36.

But... that says deb9. That's a *whole* lot newer than your running
system. Did you honestly expect a Debian 9 package to work on your
Debian 6 (or whatever it is) system?

> I'm not finding udev packages in pool. Where are they located in the
> repository tree? Is there an intermediate udev package that would allow me
> for instance to boot linux 3.16.0-6-amd64?

You're running a version that's so old that I wouldn't expect to find
its packages in the normal locations. You'd be better off using
snapshot.debian.org and downloading the obsolete versions that you require
by hand.

> Finally. I tried using another machine to download udev 215-17+deb8u7 -- it
> lets me download libudev 215-17+deb8u7, but even just downloading the udev
> package is blocked by the dependency check.
>
>     apt-get install --download-only udev=215-17+deb8u7 <== fails
>     apt-get install --nodeps --download-only udev=215-17+deb8u7
> E: Command line option --nodeps is not understood in combination with the
> other options

And this one says deb8.

Use dpkg -i to install individual Debian package files in these old
versions. "apt-get install ./filename" didn't exist back then. The
old way was to use dpkg -i, which led you to an incomplete state,
and then use "apt-get -f install" with no package names to let apt-get
try to fix the incomplete state.

Or... you could install a newer version of Debian from scratch. It might
be simpler than trying to salvage this Frankendebian installation. (You've
got mixed up versions all over the place. I don't think this is fixable
in any sane way.)

I'm still curious why you needed a custom kernel, though. You'll want to
solve that mystery. If there's something *unique* about this machine,
which would prevent a clean installation of Debian 11, you'll need to
address it, whatever it may be.

David Liontooth

unread,
Feb 16, 2022, 11:50:06 AM2/16/22
to


On 2/16/22 7:48 AM, Greg Wooledge wrote:
On Wed, Feb 16, 2022 at 07:19:34AM -0800, David Liontooth wrote:

Hi -- I have a machine, Linux ancient 2.6.36.2 #1 SMP Sun Dec 26 06:19:57
PST 2010 x86_64 GNU/Linux.

This is not a Debian kernel.  You either built it yourself, or you got
it from some foreign Linux distribution.
True -- I used to compile my own kernels, though in this case, there's no particular need to. I just want a vanilla system at this point.
I can install a new kernel, but it won't boot into any of the new kernels:

root@ancient:~# update-grub
Generating grub.cfg ...
Found linux image: /boot/vmlinuz-4.9.0-16-amd64
Found initrd image: /boot/initrd.img-4.9.0-16-amd64
Found linux image: /boot/vmlinuz-3.16.0-6-amd64
Found initrd image: /boot/initrd.img-3.16.0-6-amd64
Found linux image: /boot/vmlinuz-2.6.36.2
Found linux image: /boot/vmlinuz-2.6.36.1

Why not?  What happens when you try to boot one of them?  Are they missing
whatever features you needed to build a custom kernel to support?
I've only tried remotely, so I don't see what's happening. If I force it and the udev is wrong, won't the kernel fail?
I'm trying to upgrade from jesse to stretch. Maybe I should downgrade to jesse again and see if I can install a debian kernel first -- I'll try that!
I'm not finding udev packages in pool. Where are they located in the
repository tree? Is there an intermediate udev package that would allow me
for instance to boot linux 3.16.0-6-amd64?

You're running a version that's so old that I wouldn't expect to find
its packages in the normal locations.  You'd be better off using
snapshot.debian.org and downloading the obsolete versions that you require
by hand.

OK, very helpful, I'd forgotten about snapshots, that's great.

Still, where in pool is udev? 

Finally. I tried using another machine to download udev 215-17+deb8u7 -- it
lets me download libudev 215-17+deb8u7, but even just downloading the udev
package is blocked by the dependency check.

    apt-get install --download-only udev=215-17+deb8u7 <== fails
    apt-get install --nodeps --download-only udev=215-17+deb8u7
E: Command line option --nodeps is not understood in combination with the
other options

And this one says deb8.
I was just trying to find something that would install, so I went back to a jesse repository for this.
Use dpkg -i to install individual Debian package files in these old
versions.  "apt-get install ./filename" didn't exist back then.  The
old way was to use dpkg -i, which led you to an incomplete state,
and then use "apt-get -f install" with no package names to let apt-get
try to fix the incomplete state.
OK, cool, thanks!
Or... you could install a newer version of Debian from scratch.  It might
be simpler than trying to salvage this Frankendebian installation.  (You've
got mixed up versions all over the place.  I don't think this is fixable
in any sane way.)
You may be right, but the machine is old and I'm not entirely confident I can get it to boot from a USB stick.

I'm still curious why you needed a custom kernel, though.  You'll want to
solve that mystery.  If there's something *unique* about this machine,
which would prevent a clean installation of Debian 11, you'll need to
address it, whatever it may be.

I just used to compile my own kernels as a matter of course -- maybe from way back; I first installed potato. I don't think it means there's anything unusual about this Supermicro machine, I see nothing in my ancient notes that would suggest it needs anything special.

Cheers,
Dave

Andy Smith

unread,
Feb 16, 2022, 12:30:05 PM2/16/22
to
On Wed, Feb 16, 2022 at 07:19:34AM -0800, David Liontooth wrote:
> Hi -- I have a machine, Linux ancient 2.6.36.2 #1 SMP Sun Dec 26 06:19:57
> PST 2010 x86_64 GNU/Linux.

[…]

> Since release 198, udev requires support for the following features in
> the running kernel:
>
> - inotify(2)            (CONFIG_INOTIFY_USER)
> - signalfd(2)           (CONFIG_SIGNALFD)
> - accept4(2)
> - open_by_handle_at(2)  (CONFIG_FHANDLE)
> - timerfd_create(2)     (CONFIG_TIMERFD)
> - epoll_create(2)       (CONFIG_EPOLL)
> dpkg: error processing archive
> /var/cache/apt/archives/udev_232-25+deb9u13_amd64.deb (--unpack):
>  subprocess new pre-installation script returned error exit status 1
> update-rc.d: warning: start and stop actions are no longer supported;
> falling back to defaults
> update-rc.d: warning: start and stop actions are no longer supported;
> falling back to defaults
> Errors were encountered while processing:
>  /var/cache/apt/archives/udev_232-25+deb9u13_amd64.deb
> E: Sub-process /usr/bin/dpkg returned an error code (1)

[…]

> There are clearly lots of barriers put in to keep people from making
> mistakes, but it's also keeping me from upgrading. Is there a solution?

To have ended up trying to run Debian stretch software on a 2.6.32
kernel you have made some pretty extreme decisions far outside what
Debian supports.

You have got to get onto a more modern kernel so you can upgrade
udev. If you somehow force udev to get installed while still running
that 2.6.32 kernel you will be in a world of pain because it just
won't work and you very likely won't complete another boot.

You haven't elaborated as to why any of the newer kernels you have
installed won't boot.

Cheers,
Andy

--
https://bitfolk.com/ -- No-nonsense VPS hosting

Greg Wooledge

unread,
Feb 16, 2022, 12:50:05 PM2/16/22
to
On Wed, Feb 16, 2022 at 08:24:57AM -0800, David Liontooth wrote:
> Still, where in pool is udev?

unicorn:~$ apt-get --reinstall --print-uris install udev
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
0 upgraded, 0 newly installed, 1 reinstalled, 0 to remove and 0 not upgraded.
Need to get 1,464 kB of archives.
After this operation, 0 B of additional disk space will be used.
'http://ftp.us.debian.org/debian/pool/main/s/systemd/udev_247.3-6_amd64.deb' udev_247.3-6_amd64.deb 1463624 MD5Sum:2e6e11d7cfd3069201b3ea65853c5a61

Under s for systemd, apparently. Or at least, that's where it is *now*.
Who knows where it was in your version.

David Liontooth

unread,
Feb 17, 2022, 2:30:06 PM2/17/22
to
Very nice. I didn't know how to query the repository for location; thanks!

David Liontooth

unread,
Feb 17, 2022, 2:30:06 PM2/17/22
to

Thanks, Andy. I didn't have physical access to the machine, so I didn't know why the kernels didn't boot. But I revered to jesse in sources.list and was able to install a new kernel, which is now booting fine. So in spite of my extreme decisions, Debian handled the upgrade well.

Dave
0 new messages