[GIT PULL] Misc driver fix for 5.15-rc1

[GIT PULL] Misc driver fix for 5.15-rc1

[Greg KH]

The following changes since commit bf9f243f23e6623f310ba03fbb14e10ec3a61290:

  Merge tag '5.15-rc-ksmbd-part2' of git://git.samba.org/ksmbd (2021-09-09 16:17:14 -0700)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc.git tags/char-misc-5.15-rc1-lkdtm

for you to fetch changes up to 3a3a11e6e5a2bc0595c7e36ae33c861c9e8c75b1:

  lkdtm: Use init_uts_ns.name instead of macros (2021-09-10 17:09:40 +0200)

----------------------------------------------------------------
Misc driver fix for 5.15-rc1

Here is a single patch for 5.15-rc1, for the lkdtm misc driver.

It resolves a build issue that many people were hitting with your
current tree, and Kees and others felt would be good to get merged
before -rc1 comes out, to prevent them from having to constantly hit it
as many development trees restart on -rc1, not older -rc releases.

It has NOT been in linux-next, but has passed 0-day testing and looks
"obviously correct" when reviewing it locally :)

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

----------------------------------------------------------------
Kees Cook (1):
      lkdtm: Use init_uts_ns.name instead of macros

 drivers/misc/lkdtm/core.c  | 10 ++++++++++
 drivers/misc/lkdtm/lkdtm.h | 28 +++++++++++++++++-----------
 2 files changed, 27 insertions(+), 11 deletions(-)

Re: [GIT PULL] Misc driver fix for 5.15-rc1

[Linus Torvalds]

On Sun, Sep 12, 2021 at 7:03 AM Greg KH <gregkh@linuxfoundation.org> wrote:
>
> It resolves a build issue that many people were hitting with your
> current tree [...]

I was like "Oh, serious build problems"..

But as far as I can tell, the "build issue" is that a couple of files
get recompiled every time you add a commit. Ok, that added possibly a
few seconds to the build.

So fixing that is good, but why is that lkdtm driver insisting on
writing the release and machine information AT ALL?

IOW, the fix seems kind of silly. The kernel release information
should just have been removed entirely, because the kernel prints that
out as the very first thing anyway, so 'dmesg' will have something
like

    Linux version 5.14.0-11181-gd8e988b62f94 (torvalds@ryzen) (clang
version 12.0.1 ...

at the very top.

So I think that whole 'lkdtm_kernel_info' thing is all kinds of pointless.

Could we please just stop pointlessly printing out kernel version data
that  is already available other ways? Just do 'dmesg', or 'uname -r'
or whatever instead of insisting on printing out redundant
information?

           Linus

Re: [GIT PULL] Misc driver fix for 5.15-rc1

[Kees Cook]

On Sun, Sep 12, 2021 at 12:03:35PM -0700, Linus Torvalds wrote:
> Could we please just stop pointlessly printing out kernel version data
> that  is already available other ways? Just do 'dmesg', or 'uname -r'
> or whatever instead of insisting on printing out redundant
> information?

This was done to deal with the various mangling of reports (i.e.
"summaries") I've been getting from CI systems that run LKDTM. It has been
difficult to reliably extract actionable details, so instead I included
it with the test output. And the extra details go the other way too:
frequently folks configuring the CI have no context for why tests fail,
and LKDTM usually has a reasonable amount of information about what
Kconfig settings have gone missing, etc, so those hints are reported as
well.

-- 
Kees Cook

Re: [GIT PULL] Misc driver fix for 5.15-rc1

[Linus Torvalds]

On Sun, Sep 12, 2021 at 12:17 PM Kees Cook <keescook@chromium.org> wrote:
>
> This was done to deal with the various mangling of reports (i.e.
> "summaries") I've been getting from CI systems that run LKDTM.

.. and what makes LKDTM so special?

IOW, what about ALL THE OTHER REPORTS?

And no, my argument is most definitely not "ok, everything should do this".

It's the reverse. The CI systems should be the ones that are fixed,
not random messages from random places in the kernel have version
information added.

A CI report that doesn't report the version of the kernel is kind of
fundamentally broken, wouldn't you agree?

The fix is not to add the kernel version string into every single
message you print out.

             Linus

Re: [GIT PULL] Misc driver fix for 5.15-rc1

[Kees Cook]

On Sun, Sep 12, 2021 at 12:22:39PM -0700, Linus Torvalds wrote:
> On Sun, Sep 12, 2021 at 12:17 PM Kees Cook <keescook@chromium.org> wrote:
> >
> > This was done to deal with the various mangling of reports (i.e.
> > "summaries") I've been getting from CI systems that run LKDTM.
> 
> .. and what makes LKDTM so special?
> 
> IOW, what about ALL THE OTHER REPORTS?

I'm not sure I consider it "special", but since it reports hints about the
build/test configuration combinations (i.e. "this test failed probably
because CONFIG_FOO is missing"), it seemed trivial to also include the
specifics of the version and arch.

> And no, my argument is most definitely not "ok, everything should do this".

Right, I completely understand that. If you really want it gone, I
will rip it out; it'll just make it more time consuming to analyze some
CI reports.

> It's the reverse. The CI systems should be the ones that are fixed,
> not random messages from random places in the kernel have version
> information added.

Completely agreed, and I've been _also_ been spending my time sending
patches[1] to CI tooling too, trying to solve this from both sides. But
not all CIs have the source for their machinery open for patching. :(

-Kees

[1] https://github.com/Linaro/test-definitions/commit/8bd338bbcfa5a03efcf1d12e25b5d341d5a29cbc

-- 
Kees Cook

Re: [GIT PULL] Misc driver fix for 5.15-rc1

[Linus Torvalds]

On Sun, Sep 12, 2021 at 12:44 PM Kees Cook <keescook@chromium.org> wrote:
>
> Right, I completely understand that. If you really want it gone, I
> will rip it out; it'll just make it more time consuming to analyze some
> CI reports.

Well, I've pulled it, so it is what it is.

I would ask you to try to get the CI cases fixed, and then remove the
odd printouts from LKDTM, but it's certainly ok by me if that doesn't
happen immediately.

           Linus

Re: [GIT PULL] Misc driver fix for 5.15-rc1

[Kees Cook]

On Sun, Sep 12, 2021 at 12:55:54PM -0700, Linus Torvalds wrote:
> On Sun, Sep 12, 2021 at 12:44 PM Kees Cook <keescook@chromium.org> wrote:
> >
> > Right, I completely understand that. If you really want it gone, I
> > will rip it out; it'll just make it more time consuming to analyze some
> > CI reports.
> 
> Well, I've pulled it, so it is what it is.
> 
> I would ask you to try to get the CI cases fixed, and then remove the
> odd printouts from LKDTM, but it's certainly ok by me if that doesn't
> happen immediately.

Okay, sounds good. Thanks!

-- 
Kees Cook

Re: [GIT PULL] Misc driver fix for 5.15-rc1

[pr-tracker-bot]

The pull request you sent on Sun, 12 Sep 2021 16:03:40 +0200:

> git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc.git tags/char-misc-5.15-rc1-lkdtm

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/d8e988b62f948d47dd86ec655c89d54053df1754

Thank you!

-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html