openafs: Support quantal’s kernel 3.5

[Anders Kaseorg]

As one might expect, quantal’s new kernel 3.5 breaks openafs again. The upstream deltas to fix it are

    - [cc63cbb] Linux 3.4: replace end_writeback with clear_inode
    - [407e7c9] Linux 3.5: encode_fh API change

[Impact]
Now that linux-image-generic-lts-quantal is in precise main and installed by default, the OpenAFS kernel module fails to build on precise.

[Test Case]
apt-get install openafs-modules-dkms
(This should be tested on both the 3.2 kernel and the 3.5 kernel.)

[Regression Potential]
This fix is already well-tested in quantal. (The only difference between precise’s 1.6.1-1 and quantal’s pre-change 1.6.1-2 is translation updates.) It’s a minimal change that has no effect on pre-3.5 kernels, via preprocessor defines. In the unlikely event of a regression, OpenAFS may fail to come up, but the local system will remain usable.

[Other Info]
Since debian/source/options contains ‘single-debian-patch’, you may need to check that the debdiff was really applied (see comments).

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Anders Kaseorg (andersk)]

Since a fix in Debian doesn’t seem to be coming soon, here’s a patch for Ubuntu.

[Launchpad Janitor (janitor)]

This bug was fixed in the package openafs - 1.6.1-2ubuntu1

---------------
openafs (1.6.1-2ubuntu1) quantal; urgency=low

  * Apply upstream deltas for Linux 3.5 (Closes: #685973)
    (LP: #1015925):
    - [2b33384] Linux 3.4: replace end_writeback with clear_inode
    - [5227148] Linux 3.5: encode_fh API change
 -- Anders Kaseorg <email address hidden> Tue, 11 Sep 2012 19:01:20 -0400

[Anders Kaseorg (andersk)]

Uh, Luke, the package you uploaded did not include the changes in my debdiff (except the changelog and the control file). It’s identical to 1.6.1-2.

http://launchpadlibrarian.net/115542497/openafs_1.6.1-2_1.6.1-2ubuntu1.diff.gz

Please try again.

[Luke Yelavich (themuso)]

Strange. I diply the debdff, thanks for the heads up, will fix it up.

[Luke Yelavich (themuso)]

You let out the series file in debian/patches, which is probably why the patch was not included, and I totally missed it.

[Luke Yelavich (themuso)]

...or not, the series file is already there, because the file is already there. Never mind my previous comment.

[Luke Yelavich (themuso)]

Ok, I see the problem, the changes get overridden when I run debbuild on the package. I'll have to move them to a new patch to be sure.

[Anders Kaseorg (andersk)]

You should note that debian/source/options contains ‘single-debian-patch’. (But I guess you’ll probably have figured that out by the time you read this.)

[Luke Yelavich (themuso)]

Yes I have, I had to make sure quilt was aware of the changes to the debian patch in the end.

[Launchpad Janitor (janitor)]

This bug was fixed in the package openafs - 1.6.1-2ubuntu2

---------------
openafs (1.6.1-2ubuntu2) quantal; urgency=low

  * Really fix for kernel 3.5. (LP: #1015925)
 -- Luke Yelavich <email address hidden> Thu, 13 Sep 2012 09:45:52 +1000

[Anders Kaseorg (andersk)]

Now that linux-image-generic-lts-quantal is in precise main, this fix needs to be SRU’d into precise.

[Anders Kaseorg (andersk)]

Here’s a debdiff for a proposed SRU in precise. (I’ve generated it in a way that should avoid the single-debian-patch application problems from earlier.)

[Scott Kitterman (kitterman)]

It's installed by default only for new installations. Does this change work
with the old kernel too?

[Anders Kaseorg (andersk)]

Yeah (as I noted in [Regression Potential]); the new API is enabled via autoconf tests.

[Scott Kitterman (kitterman)]

On Thursday, February 28, 2013 06:58:30 AM you wrote:
> Yeah (as I noted in [Regression Potential]); the new API is enabled via
> autoconf tests.

OK. Are we building openafs twice for 12.04 then? Both kernels still exist.
My question was a run time question, not a build time question.

[Anders Kaseorg (andersk)]

No no, the kernel module is compiled on the user’s system by DKMS, separately for each kernel. There are no binary openafs modules in the Ubuntu repository.

[Scott Kitterman (kitterman)]

OK. Makes sense. It should be tested with both kernels before it's marked
verification done. Please add that to the test case.

[Scott Kitterman (kitterman)]

There's an embedded copy of the debdiff in the source package. Rejecting. Please re-upload a fixed version (you can reuse the same version number).

[Luke Faraone (lfaraone)]

Fixed and reuploaded, apologies.

On 1 March 2013 19:23, Scott Kitterman <email address hidden> wrote:

> There's an embedded copy of the debdiff in the source package.
> Rejecting. Please re-upload a fixed version (you can reuse the same
> version number).
>
> --
> You received this bug notification because you are a bug assignee.
> https://bugs.launchpad.net/bugs/1015925
>
> Title:
> openafs: Support quantal’s kernel 3.5
>
> To manage notifications about this bug go to:
>
> https://bugs.launchpad.net/ubuntu/+source/openafs/+bug/1015925/+subscriptions
>

--

Luke Faraone;; Debian & Ubuntu Developer; Sugar Labs; GMU 2014
lfaraone on irc.[freenode,oftc].net -- https://luke.wf/ohhello
PGP fprint: 5189 2A7D 16D0 49BB 046B DC77 9732 5DD8 F9FD D506

[Scott Kitterman (kitterman)]

Hello Anders, or anyone else affected,

Accepted into quantal-proposed. The package will build now and be available in a few hours in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, and change the tag from verification-needed to verification-done. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed. In either case, details of your testing will help us make a better decision.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance!

[Scott Kitterman (kitterman)]

Correction. That was accepted for precise-proposed.

[Jonathan Reed (jdreed)]

Using openafs-modules-dkm 1.6.1-1+ubuntu0.1 from precise-proposed, the module now builds correctly for 3.5.0-23-generic (and also for -26-generic, which happens to be in -proposed).

openafs-modules-dkms:
  Installed: 1.6.1-1+ubuntu0.1
  Candidate: 1.6.1-1+ubuntu0.1
  Version table:
 *** 1.6.1-1+ubuntu0.1 0
        500 http://archive.ubuntu.com/ubuntu/ precise-proposed/universe amd64 Packages
        100 /var/lib/dpkg/status
     1.6.1-1 0
        500 http://us.archive.ubuntu.com/ubuntu/ precise/universe amd64 Packages

[Jonathan Reed (jdreed)]

I can also confirm there is no regression, as it works fine on 3.2.0-38-generic.

[Jonathan Reed (jdreed)]

Can this get moved to production any time soon? I understand there's a desire to get openafs 1.6.2 out for security updates, but these are not mutually exclusive. I have a large community of users who can't access their shared file storage if they installed from a CD after Feb 14.

[Luke Faraone (lfaraone)]

Packages need to be in -proposed for seven days. I'll check in with the SRU team on Monday to see if we can move the package over.

[Scott Kitterman (kitterman)]

--- Releasing openafs ---
Proposed: 1.6.1-1+ubuntu0.1
Release: 1.6.1-1
Copied to precise-updates

[Rich Wales (richw)]

Now that the 3.8.0 kernel in in Precise, openafs appears to be broken again.

My "uname -a" output is:
Linux liberation 3.8.0-31-generic #46~precise1-Ubuntu SMP Wed Sep 11 18:21:16 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux

My /var/lib/dkms/openafs/1.6.1/build/make.log file includes the following error:
/var/lib/dkms/openafs/1.6.1/build/src/afs/afs_osi.h:14:21: fatal error: h/param.h: No such file or directory

[Anders Kaseorg (andersk)]

Known (bug 1206387).