Re: [GIT PULL] arm64 updates for 7.1

[Linus Torvalds]

On Tue, 14 Apr 2026 at 06:47, Catalin Marinas <catalin.marinas@xxxxxxx> wrote:
>
> The generic entry patches touch include/linux/*entry* and kernel/entry/*
> as they are based on a tag from tip. Not sure when Thomas will send them
> your way, so sending the arm64 pull request in the meantime.

Ok, so the conflict with the irq changes is something that I want
Thomas and PeterZ to take a look at, because it involved two calls to
hrtimer_rearm_deferred() that got added in the meantime - while in
this pull the call site then got split up and moved from
kernel/entry/common.c to include/linux/irq-entry-common.h.

In my initial resolution I placed the two hrtimer_rearm_deferred()
calls very similarly to how they were placed before (just in the moved
location).

So my initial resolution was the fairly direct "that's what the code
movement would do".

But then I double-checked what linux-next did, and it did something
quite different. The linux-next resolution isn't the naive direct code
movement, and seems to be a smarter semantic resolution with just one
call-site.

So after having stared at it for a long time, and convinced myself it
was the better resolution I went back and fixed up mine too.

It all makes me go "PeterZ and Thomas should double-check me that I
didn't miss something else".

I did the builds on both x86-64 and arm64, but I feel it should still
be checked by people involved with that hrtimer code.

               Linus