patchset update for Xenomai/IPIPE's kernel 5.4 porting

Jan Kiszka jan.kiszka at siemens.com
Mon Sep 14 17:51:01 CEST 2020


On 14.09.20 17:40, Meng, Fino via Xenomai wrote:
> Hi all,
> 
> A updated patchset was updated to this branch:
> https://github.com/intel/linux-stable-xenomai/tree/review/5.4.59/stable/ipipe/xenomai-3.1
> this version is more clean than previous, only one more patch for IPIPE part,
> and one more patch for Xenomai 3.1 part.
> 
> ....
> 0150-x86-ipipe-Harden-__speculation_ctrl_update-for-share.patch
> 0151-x86-ipipe-Let-HAVE_ARCH_JUMP_LABEL-depend-on-IPIPE.patch
> 0152-drm-i915-gt-rename-setup_irq-to-avoid-compile-confli.patch
> 0153-xenomai-patch-xenomai-3.1-directly.patch
> 0154-WIP-xenomai-fix-xenomai-kernel-part-for-kernel-5.4.y.patch

Please make this one ready for upstream, possibly breaking it up into 
logical chunks. Will help to keep your queue cleanly focused on I-pipe.

BTW, commenting out fpu__initialize() is likely problematic. Better make 
that function available again with the I-pipe patch.

> 0155-test-kernel-config-for-xenomai.patch
> 
> On my test device, boot OK and Xenomai's latency can run, other smokey test may fail.
> 

Does the switchtest explode? That's the first one to expose FPU issues, 
see above.

> A quick guide:
> 
>      usage:
>      $cp kernel_config_xenomai .config
>      $make olddefconfig
>      $make bindeb-pkg -j16 2>&1 | tee build.log
>      if build ok, install it over a Debian or Ubuntu on test board,
>      #dpkg -i  linux-image-*.deb
>      #reboot
> 
>      to clean up:
>      $make distclean
> 
>      bug:
>      out of tree build will fail for kernel 5.4, need fix.
> 
> BR / Fino (孟祥夫)
> Intel – IOTG Developer Enabling
> 

In any case: great progress!

Jan

-- 
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux



More information about the Xenomai mailing list