public inbox for development@lists.ipfire.org
 help / color / mirror / Atom feed
From: Michael Tremer <michael.tremer@ipfire.org>
To: development@lists.ipfire.org
Subject: Re: [PATCH 00/21] linux: Update to 5.15.85 and backport many IPFire 3.x changes
Date: Tue, 27 Dec 2022 11:36:12 +0100	[thread overview]
Message-ID: <00B6AF81-499B-42F2-B9D0-49DA35DF7FE4@ipfire.org> (raw)
In-Reply-To: <0e60a1de-6210-835e-54a4-ec5e3128e42e@ipfire.org>

[-- Attachment #1: Type: text/plain, Size: 6627 bytes --]

Hello Peter,

> On 26 Dec 2022, at 20:24, Peter Müller <peter.mueller(a)ipfire.org> wrote:
> 
> This patchset aims at updating the Linux kernel to 5.15.85, given that
> the last release we shipped dates back a while ago. However, its primary
> purpose is to backport some kernel changes recently made by Michael in
> IPFire 3.x, whenever bringing these to the IPFire 2.x userbase is sensible
> and/or feasible.

I am happy with updating the kernel.

> Patch descriptions are copy & past'ed from their IPFire 3.x counterparts,
> which are referred to by their commit IDs in ipfire-3.x. Due to different
> hardware situation as well as architecture maturity (this particularly
> affects ARM), not all changes could be backported 1:1 or to a near-complete
> extend.

As I said in our previous conversation about this, I am not too happy to see this patchset here, yet.

The current kernel in IPFire 3 is highly experimental. In order to try things out, I enabled lots of (let’s call them) risky features that are either not commonly enabled on off-the-shelf distributions, or are not tested by us.

That results in a kernel that currently does not even boot.

“Backporting” from a broken kernel that is so untested will only result in carrying over any problems from the testing environment into the production environment where they are so much more harmful.

We should test first, and then move on to the next step and figure out how we can roll out the successfully tested changes and how we can roll back those that don’t work well for us.

> Feedback is particularly appreciated regarding the last commit, which aims
> at aligning the ARM kernel configuration files to the x86_64 one. Since
> no real ARM hardware is at the author's disposal, this alignment has to be
> taken with a pinch of salt.

How is that supposed to be tested?

> As far as benchmarks are concerned, a 5.15.85 x86_64 kernel booted in an
> IPFire 2.x VM on the basis of Core Update 172 introduced the following changes
> in file size:
> 
> Location Before After
> -------------------------------------------
> /boot 48M 53M  (+ 5)
> /lib/modules 58M 71M  (+13)
> ISO 373M 394M (+21)

We cannot afford at all to make the kernel larger, since we still have plenty of installations out there is a small /boot partition and a / partition that is limited to 2GB. Not that another 13 MiB will break the camel’s back, but we should try to save space to keep those users up and running.

> Contrary to its documentation, enabling the GCC stackleak plugin (which
> is the current setting in IPFire 3.x as well) neither brought a notable
> compile time increase, nor does it seem to slow down runtime operations
> significantly. More thorough tests, especially on physical machines, are
> however, yet to come.

How many times did you rebuild the kernel with exactly the same configuration?

In IPFire 3 there is something that seems to limit the performance of ccache, which we cannot carry over into IPFire 2 under any circumstances. IPFire 2 is very sensitive towards compile time.

-Michael

> Peter Müller (21):
>  linux: Update to 5.15.85
>  linux: Disable the entire PCMCIA/CardBus subsystem
>  linux: Enable parallel crypto by default
>  linux: Disable syscalls that allows processes to r/w other processes'
>    memory
>  linux: Disable the latent entropy plugin
>  linux: Build all library routines as modules and disable self-tests
>  linux: Build all HWRNGs as modules
>  linux: Compile binfmt_misc as a module
>  linux: Wipe all memory when rebooting on EFI
>  linux: Disable the Distributed Lock Manager
>  linux: Disable some character devices that do not make sense
>  linux: Make graphics configruation sane
>  linux: Disable all sorts of useless Device Mapper targets
>  linux: Enable various modern ciphers/hashes/etc. and acceleration
>  linux: Compress the kernel, modules and firmware using Zstandard
>  linux: Disable ACPI configfs support
>  linux: Enable support for more USB host controllers as modules
>  linux: Poison kernel stack before returning from syscalls
>  linux: Enable Landlock support
>  linux: Update x86_64 rootfile
>  linux: Align ARM kernel configurations as much as possible
> 
> config/kernel/kernel.config.aarch64-ipfire    |  194 +-
> config/kernel/kernel.config.armv6l-ipfire     |  101 +-
> config/kernel/kernel.config.x86_64-ipfire     |  216 +-
> config/rootfiles/common/x86_64/linux          | 5954 ++++++++---------
> lfs/linux                                     |    9 +-
> .../linux-5.15-wifi-security-patches-1.patch  |   50 -
> .../linux-5.15-wifi-security-patches-10.patch |   98 -
> .../linux-5.15-wifi-security-patches-11.patch |   96 -
> .../linux-5.15-wifi-security-patches-12.patch | 1179 ----
> .../linux-5.15-wifi-security-patches-13.patch |  130 -
> .../linux-5.15-wifi-security-patches-14.patch |  107 -
> .../linux-5.15-wifi-security-patches-2.patch  |   59 -
> .../linux-5.15-wifi-security-patches-3.patch  |   49 -
> .../linux-5.15-wifi-security-patches-4.patch  |   96 -
> .../linux-5.15-wifi-security-patches-5.patch  |   56 -
> .../linux-5.15-wifi-security-patches-6.patch  |   39 -
> .../linux-5.15-wifi-security-patches-7.patch  |   60 -
> .../linux-5.15-wifi-security-patches-8.patch  |   94 -
> .../linux-5.15-wifi-security-patches-9.patch  |  126 -
> 19 files changed, 3183 insertions(+), 5530 deletions(-)
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-1.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-10.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-11.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-12.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-13.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-14.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-2.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-3.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-4.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-5.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-6.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-7.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-8.patch
> delete mode 100644 src/patches/linux/linux-5.15-wifi-security-patches-9.patch
> 
> -- 
> 2.35.3


      parent reply	other threads:[~2022-12-27 10:36 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-12-26 19:24 Peter Müller
2022-12-26 19:24 ` [PATCH 01/21] linux: Update to 5.15.85 Peter Müller
2022-12-27 10:37   ` Michael Tremer
2022-12-29 11:14     ` Peter Müller
2022-12-29 11:16       ` Michael Tremer
2022-12-26 19:24 ` [PATCH 02/21] linux: Disable the entire PCMCIA/CardBus subsystem Peter Müller
2022-12-27 10:39   ` Michael Tremer
2022-12-26 19:25 ` [PATCH 03/21] linux: Enable parallel crypto by default Peter Müller
2022-12-27 10:39   ` Michael Tremer
2022-12-26 19:25 ` [PATCH 04/21] linux: Disable syscalls that allows processes to r/w other processes' memory Peter Müller
2022-12-27 11:22   ` Michael Tremer
2022-12-26 19:26 ` [PATCH 05/21] linux: Disable the latent entropy plugin Peter Müller
2022-12-27 11:22   ` Michael Tremer
2022-12-26 19:26 ` [PATCH 06/21] linux: Build all library routines as modules and disable self-tests Peter Müller
2022-12-27 11:22   ` Michael Tremer
2022-12-26 19:26 ` [PATCH 07/21] linux: Build all HWRNGs as modules Peter Müller
2022-12-27 11:23   ` Michael Tremer
2022-12-26 19:27 ` [PATCH 08/21] linux: Compile binfmt_misc as a module Peter Müller
2022-12-27 11:23   ` Michael Tremer
2022-12-26 19:27 ` [PATCH 09/21] linux: Wipe all memory when rebooting on EFI Peter Müller
2022-12-27 11:23   ` Michael Tremer
2022-12-26 19:27 ` [PATCH 10/21] linux: Disable the Distributed Lock Manager Peter Müller
2022-12-27 11:24   ` Michael Tremer
2022-12-26 19:28 ` [PATCH 11/21] linux: Disable some character devices that do not make sense Peter Müller
2022-12-27 11:24   ` Michael Tremer
2022-12-26 19:28 ` [PATCH 12/21] linux: Make graphics configruation sane Peter Müller
2022-12-27 11:24   ` Michael Tremer
2022-12-26 19:28 ` [PATCH 13/21] linux: Disable all sorts of useless Device Mapper targets Peter Müller
2022-12-27 11:25   ` Michael Tremer
2022-12-26 19:29 ` [PATCH 14/21] linux: Enable various modern ciphers/hashes/etc. and acceleration Peter Müller
2022-12-27 11:25   ` Michael Tremer
2022-12-26 19:29 ` [PATCH 15/21] linux: Compress the kernel, modules and firmware using Zstandard Peter Müller
2022-12-27 11:26   ` Michael Tremer
2022-12-26 19:29 ` [PATCH 16/21] linux: Disable ACPI configfs support Peter Müller
2022-12-27 11:29   ` Michael Tremer
2022-12-26 19:30 ` [PATCH 17/21] linux: Enable support for more USB host controllers as modules Peter Müller
2022-12-27 11:33   ` Michael Tremer
2022-12-26 19:30 ` [PATCH 18/21] linux: Poison kernel stack before returning from syscalls Peter Müller
2022-12-27 11:35   ` Michael Tremer
2022-12-26 19:30 ` [PATCH 19/21] linux: Enable Landlock support Peter Müller
2022-12-27 11:36   ` Michael Tremer
2022-12-26 19:31 ` [PATCH 20/21] linux: Update x86_64 rootfile Peter Müller
2022-12-27 11:36   ` Michael Tremer
2022-12-26 19:31 ` [PATCH 21/21] linux: Align ARM kernel configurations as much as possible Peter Müller
2022-12-27 10:36 ` Michael Tremer [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=00B6AF81-499B-42F2-B9D0-49DA35DF7FE4@ipfire.org \
    --to=michael.tremer@ipfire.org \
    --cc=development@lists.ipfire.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox