[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <Z-6jqf6QqRkAnTz8@gmail.com>
Date: Thu, 3 Apr 2025 17:05:13 +0200
From: Ingo Molnar <mingo@...nel.org>
To: "H. Peter Anvin" <hpa@...or.com>
Cc: linux-kernel@...r.kernel.org, Juergen Gross <jgross@...e.com>,
Linus Torvalds <torvalds@...ux-foundation.org>,
Peter Zijlstra <peterz@...radead.org>,
Borislav Petkov <bp@...en8.de>,
Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH 40/41] x86/alternatives: Rename 'text_poke_sync()' to
'text_poke_sync_each_cpu()'
* H. Peter Anvin <hpa@...or.com> wrote:
> Is that the only use case we have for syncing all CPUs?
So there's:
- kernel/sched/membarrier.c's ipi_sync_core() which does sync_core(),
but it's embedded into a larger array of IPI handlers, called via
smp_call_function_many()/on_each_cpu_mask(), so I don't think
functionality can be shared there.
- there's arch/x86/kernel/cpu/mce/core.c's machine_check_poll() handler
and the kill_me_maybe() function, but these are single thread.
- then there's arch/x86/kernel/static_call.c's
__static_call_update_early(), used indirectly by
arch/x86/xen/enlighten.c and arch/x86/xen/enlighten_pv.c, but these
are single-threaded too due to being early boot code.
So not much I think - at least what I've managed to find.
Thanks,
Ingo
Powered by blists - more mailing lists