Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <CAHmME9pe2BEJV4WiZNHmDmH_XK621Qqr1JCBdgTNZmr4JGBA4w@mail.gmail.com>
 <20220128223548.97807-1-Jason@zx2c4.com> <CAHmME9qtjZX2kVNSQqUsTrZv1cdR8y6n3yZS-RnpVCCzX9okcA@mail.gmail.com>
 <Yf0JlXf3ARsBpL9K@linutronix.de> <CAHmME9r0XxX3LqNLpVeqAjDQ_OVskPf15QOwxtZYy0tb_x_7HQ@mail.gmail.com>
 <Yf0xy4kZ2Mn65yp8@linutronix.de>
In-Reply-To: <Yf0xy4kZ2Mn65yp8@linutronix.de>
From:   "Jason A. Donenfeld" <Jason@zx2c4.com>
Date:   Fri, 4 Feb 2022 15:11:34 +0100
Message-ID: <CAHmME9oOMhRVybTgHXT+oOXhMkdx7FVY7oSc-rHr=6AvZCVo=w@mail.gmail.com>
Subject: Re: [PATCH v2] random: remove batched entropy locking
To:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc:     Andy Lutomirski <luto@amacapital.net>,
        Boqun Feng <boqun.feng@gmail.com>,
        Will Deacon <will@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>,
        Waiman Long <longman@redhat.com>,
        Sultan Alsawaf <sultan@kerneltoast.com>,
        "Theodore Ts'o" <tytso@mit.edu>, Andy Lutomirski <luto@kernel.org>,
        =?UTF-8?Q?Jonathan_Neusch=C3=A4fer?= <j.neuschaefer@gmx.net>,
        LKML <linux-kernel@vger.kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Sebastian,

On Fri, Feb 4, 2022 at 3:02 PM Sebastian Andrzej Siewior
<bigeasy@linutronix.de> wrote:
> The commit in tree you cited is b43db859a36cb553102c9c80431fc44618703bda.
> It does not mention anything regarding faster nor the performance
> improvement and conditions (hoth path, etc). It still has a stable tag.

It dropped the Cc: stable@. It still has the Fixes:. I can get rid of
the Fixes: too. I'll improve that message a bunch for a potential v3.

> > Maybe it'd be best to retain the spinlock_t, which will amount to
> > disabling interrupts on !PREEMPT_RT, since it'll never be contended,
> > but will turn into a mutex on PREEMPT_RT, where it'll do the right
> > thing from an exclusivity perspective. Would this be reasonable?
>
> what does retain the spinlock_t mean since we already have a spinlock_t?

The idea would be to keep using spinlock_t like we do now -- no change
there -- but move to using this atomic generation counter so that
there's never any contention. Actually, though, I worry that that
approach would throw out the gains we're getting by chucking the
spinlock in the first place.

What if we keep a spinlock_t there on PREEMPT_RT but stick with
disabling interrupts on !PREEMPT_RT? I wish there was a solution or an
API that amounted to the same thing so there wouldn't need to be an
#ifdef, but I don't know what that'd be.

Jason