[ANNOUNCE] 4.4.229-rt200

[ANNOUNCE] 4.4.229-rt200

[Daniel Wagner]

Hello RT-list!

I'm pleased to announce the 4.4.229-rt200 stable release.

This release is just an update to the new stable 4.4.215 version.

Note: This update required dealing with a conflict in net/core/dev.c, where
devnet_rename_seq was moved from a seqence count to an RWSEM and renamed
to devnet_rename_sem. If you encounter runtime issues that show tracebacks
in net/core/dev.c, please let me know.

Patch "net: Add a mutex around devnet_rename_seq" dropped in favour of
602c47fbf46b ("net: core: device_rename: Use rwsem instead of a seqcount")

Known issues:

  sigwaittest with hackbench as workload is able to trigger a crash on x86_64,
  the same as reported for the v4.4.220-rt196 release. The crash seems to
  be triggered by a BPF program. IIRC, BPF was never supported in v4.4-rt
  and I just forgot to disable in my configuration.

You can get this release via the git tree at:

  git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-stable-rt.git

  branch: v4.4-rt
  Head SHA1: dd9655ee04bfb54f781974e3074a0caeb2423623

Or to build 4.4.229-rt200 directly, the following patches should be applied:

  https://www.kernel.org/pub/linux/kernel/v4.x/linux-4.4.tar.xz

  https://www.kernel.org/pub/linux/kernel/v4.x/patch-4.4.229.xz

  https://www.kernel.org/pub/linux/kernel/projects/rt/4.4/patch-4.4.229-rt200.patch.xz

Enjoy!
Daniel

Re: [ANNOUNCE] 4.4.229-rt200

[Sebastian Andrzej Siewior]

On 2020-07-08 18:44:02 [-0000], Daniel Wagner wrote:
> I'm pleased to announce the 4.4.229-rt200 stable release.
> 
> This release is just an update to the new stable 4.4.215 version.

Your template is off here.

> Note: This update required dealing with a conflict in net/core/dev.c, where
> devnet_rename_seq was moved from a seqence count to an RWSEM and renamed
> to devnet_rename_sem. If you encounter runtime issues that show tracebacks
> in net/core/dev.c, please let me know.
>
> Patch "net: Add a mutex around devnet_rename_seq" dropped in favour of
> 602c47fbf46b ("net: core: device_rename: Use rwsem instead of a seqcount")

Yes, this was the intention of the exercise :)
 
> Known issues:
> 
>   sigwaittest with hackbench as workload is able to trigger a crash on x86_64,
>   the same as reported for the v4.4.220-rt196 release. The crash seems to
>   be triggered by a BPF program. IIRC, BPF was never supported in v4.4-rt
>   and I just forgot to disable in my configuration.

In v5.4.3-rt1 there is
    fe18e9e08e25d ("BPF: Disable on PREEMPT_RT")

so maybe backport this across the stable kernels. Fixing it is a way
larger queue. If the intention is to keep it disabled it could be
enforced via Kconfig so it is not enabled by accident (or by using
oldconfig on a distro config which has it enabled).

> Enjoy!
> Daniel

Sebastian