[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <dbb9955b-b12f-6b4b-d380-1f95d3d9fed6@huawei.com>
Date: Mon, 5 Aug 2024 10:00:01 +0800
From: Jinjie Ruan <ruanjinjie@...wei.com>
To: "Russell King (Oracle)" <linux@...linux.org.uk>
CC: <arnd@...db.de>, <afd@...com>, <akpm@...ux-foundation.org>,
<linus.walleij@...aro.org>, <eric.devolder@...cle.com>, <robh@...nel.org>,
<vincent.whitchurch@...s.com>, <bhe@...hat.com>, <nico@...xnic.net>,
<ardb@...nel.org>, <linux-arm-kernel@...ts.infradead.org>,
<linux-kernel@...r.kernel.org>, <linux-arch@...r.kernel.org>
Subject: Re: [PATCH] ARM: support PREEMPT_DYNAMIC
On 2024/8/2 19:44, Russell King (Oracle) wrote:
> On Wed, Jul 31, 2024 at 10:07:53AM +0800, Jinjie Ruan wrote:
>> On 2024/6/20 17:00, Jinjie Ruan wrote:
>>> Enable support for PREEMPT_DYNAMIC on arm32, allowing the preemption model
>>> to be chosen at boot time.
>>>
>>> Similar to arm64, arm32 does not yet use the generic entry code, we must
>>> define our own `sk_dynamic_irqentry_exit_cond_resched`, which will be
>>> enabled/disabled by the common code in kernel/sched/core.c.
>>>
>>> And arm32 use generic preempt.h, so declare
>>> `sk_dynamic_irqentry_exit_cond_resched` if the arch do not use generic
>>> entry. Other architectures which use generic preempt.h but not use generic
>>> entry can benefit from it.
>>>
>>> Test ok with the below cmdline parameters on Qemu versatilepb board:
>>> `preempt=none`
>>> `preempt=voluntary`
>>> `preempt=full`
>>>
>>> Update preempt mode with debugfs interface on above Qemu board is also
>>> tested ok:
>>> # cd /sys/kernel/debug/sched
>>> # echo none > preempt
>>> # echo voluntary > preempt
>>> # echo full > preempt
>
> Do you have a use case for this feature?
Yes, many of our ARM32 products use different preemption models, and it
would be much more convenient if we could configure it at startup.
>
Powered by blists - more mailing lists