linux-kernel - Re: [PATCH] thermal: replace spin_lock_irqsave by spin

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <877drg62he.fsf@nanos.tec.linutronix.de>
Date:   Fri, 23 Oct 2020 23:45:17 +0200
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Geert Uytterhoeven <geert@...ux-m68k.org>,
        Tian Tao <tiantao6@...ilicon.com>
Cc:     Niklas Söderlund <niklas.soderlund@...natech.se>,
        Zhang Rui <rui.zhang@...el.com>,
        Daniel Lezcano <daniel.lezcano@...aro.org>, amitk@...nel.org,
        Linux-Renesas <linux-renesas-soc@...r.kernel.org>,
        Linux PM list <linux-pm@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] thermal: replace spin_lock_irqsave by spin_lock in hard IRQ

On Thu, Oct 22 2020 at 14:51, Geert Uytterhoeven wrote:
> On Wed, Oct 21, 2020 at 2:15 PM Tian Tao <tiantao6@...ilicon.com> wrote:
>> The code has been in a irq-disabled context since it is hard IRQ. There
>> is no necessity to do it again.
>>
> Is this also true if CONFIG_PREEMPT_RT=y, and all irq handlers execute
> in the context of special tasks?

On RT or even on mainline with 'threadirqs' on the command line all
interrupts which are not explicitly requested with IRQF_NO_THREAD run
their handlers in thread context. The same applies to soft interrupts.

That means they are subject to the normal scheduler rules and no other
code is going to acquire that lock from hard interrupt context either,
so the irqsave() here is pointless in all cases.

Famous last words...

  ... unless the driver does magic things like having a hrtimer armed
  which expires in hard interrupt context and touches the very same
  lock, but that's not the case in this particular driver.

So the change itself is correct, but the change log could do with some
polishing. :)

Thanks,

        tglx