Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <1615949121-70877-1-git-send-email-jiapeng.chong@linux.alibaba.com>
 <20210317083947.GD3881262@gmail.com> <20210317084132.GA378908@gmail.com>
 <CAK8P3a2qqWV3RGvb4ooiz4LS5GAKL0OHEiVtdgnHAgtmiRDSNA@mail.gmail.com> <20210317200529.GB3830960@gmail.com>
In-Reply-To: <20210317200529.GB3830960@gmail.com>
From:   Arnd Bergmann <arnd@kernel.org>
Date:   Wed, 17 Mar 2021 22:38:55 +0100
Message-ID: <CAK8P3a2MQiqEkrfd+_gE_dbYHgGBb9JjuAW7RX=ySWUc084KGg@mail.gmail.com>
Subject: Re: [PATCH] sched: replace if (cond) BUG() with BUG_ON()
To:     Ingo Molnar <mingo@kernel.org>
Cc:     Jiapeng Chong <jiapeng.chong@linux.alibaba.com>,
        Ingo Molnar <mingo@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Juri Lelli <juri.lelli@redhat.com>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Dietmar Eggemann <dietmar.eggemann@arm.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Ben Segall <bsegall@google.com>, Mel Gorman <mgorman@suse.de>,
        Daniel Bristot de Oliveira <bristot@redhat.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Mar 17, 2021 at 9:05 PM Ingo Molnar <mingo@kernel.org> wrote:
> * Arnd Bergmann <arnd@kernel.org> wrote:
> > On Wed, Mar 17, 2021 at 9:45 AM Ingo Molnar <mingo@kernel.org> wrote:
> > >
> > > More importantly, we use this pattern when we don't want !CONFIG_BUG
> > > to remove the 'condition'.
> > >
> > > I.e. the "side effect" here is important scheduler logic. It must
> > > never be optimized out.
> >
> > This behavior for !CONFIG_BUG has changed a while ago, it is now safe
> > to rely on the side-effect of the BUG_ON() condition regardless of
> > CONFIG_BUG. When that option is disabled, running into the condition
> > just ends up in a "do {} while (1)" loop.
>
> Dunno, I still think it's not a particularly clean pattern to 'hide'
> significant side effects within a BUG_ON().

Fair enough. Readability really is the key here, and I agree the current
version is easier to understand. The only architectures that even define
an optimized BUG_ON() are mips and powerpc, and saving a few cycles
is barely worth it in a fast path, which this is clearly not.

       Arnd