Re: [PATCH 0/5] sched: dynamic: Simplify PREEMPT_DYNAMIC

[Mete Durlu <meted@linux.ibm.com>]

On 03/07/2026 15:33, Mark Rutland wrote:
> All architectures which currently suppoort PREEMPT_DYNAMIC select
> ARCH_HAS_PREEMPT_LAZY.  On architectures which select
> ARCH_HAS_PREEMPT_LAZY, it has not been possible to select the NONE and
> VOLUNTARY preemption models since v7.0 due to commit:
> 
>    7dadeaa6e851 ("sched: Further restrict the preemption modes")
> 
> Hence in practice PREEMPT_DYNAMIC no longer supports the NONE or
> VOLUNTARY preemption models.
> 
> This series makes the de-facto situation official by making
> PREEMPT_DYNAMIC depend on ARCH_HAS_PREEMPT_LAZY, and removing all the
> redundant code.

Hi,

I got some results for s390.

> For x86_64 v7.2-rc1 defconfig built with GCC 10.2.1, this makes the
> bzImage 28K smaller, mostly due to removing NOPs that will never be
> patched to {cond,might}_resched() at runtime:
> 
>    [mark@lakrids:~/src/linux]% ls -al vmlinux-*
>    -rwxr-xr-x 1 mark mark 53844952 Jul  3 14:09 vmlinux-after
>    -rwxr-xr-x 1 mark mark 53842856 Jul  3 14:09 vmlinux-before
>    [mark@lakrids:~/src/linux]% ls -al bzImage-*
>    -rw-r--r-- 1 mark mark 14771200 Jul  3 14:09 bzImage-after
>    -rw-r--r-- 1 mark mark 14799872 Jul  3 14:08 bzImage-before
> 
I applied the series over e4de1b9cb3b5 ("powerpc/dt_cpu_ftrs: Set 
CPU_FTR_P11_PVR for Power11 and later processors")

I did git clean -xfd in between compiles and here is how it
looks like for s390;

vmlinux is almost 1M smaller

$ ls -la before/vmlinux
-rwxrwxr-x 1 meted meted 409306880 Jul 30 15:40 before/vmlinux
$ ls -la after/vmlinux
-rwxrwxr-x 1 meted meted 407987680 Jul 30 15:43 after/vmlinux

$ bloat-o-meter before/vmlinux after/vmlinux
add/remove: 67/50 grow/shrink: 731/2058 up/down: 54992/-161849 (-106857)

bzImage is ~32K smaller

$ ls -la before/bzImage
-rwxrwxr-x 1 meted meted 12914688 Jul 30 15:24 before/bzImage
$ ls -la after/bzImage
-rwxrwxr-x 1 meted meted 12881920 Jul 30 15:30 after/bzImage


I also did some sniff tests to see if it causes any
behavior changes, not that its expected, didn't see anything.