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;
Date:   Wed, 9 Nov 2022 10:42:37 +0100
From:   Peter Zijlstra <peterz@infradead.org>
To:     Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc:     linux-kernel@vger.kernel.org, Thomas Gleixner <tglx@linutronix.de>,
        "Paul E . McKenney" <paulmck@kernel.org>,
        Boqun Feng <boqun.feng@gmail.com>,
        "H . Peter Anvin" <hpa@zytor.com>, Paul Turner <pjt@google.com>,
        linux-api@vger.kernel.org, Christian Brauner <brauner@kernel.org>,
        Florian Weimer <fw@deneb.enyo.de>, David.Laight@aculab.com,
        carlos@redhat.com, Peter Oskolkov <posk@posk.io>,
        Alexander Mikhalitsyn <alexander@mihalicyn.com>,
        Chris Kennelly <ckennelly@google.com>
Subject: Re: [PATCH v5 08/24] sched: Introduce per memory space current
 virtual cpu id
Message-ID: <Y2t2DcHHLVB3tDfw@hirez.programming.kicks-ass.net>
References: <20221103200359.328736-1-mathieu.desnoyers@efficios.com>
 <20221103200359.328736-9-mathieu.desnoyers@efficios.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20221103200359.328736-9-mathieu.desnoyers@efficios.com>
Precedence: bulk

On Thu, Nov 03, 2022 at 04:03:43PM -0400, Mathieu Desnoyers wrote:

> +void sched_vcpu_exit_signals(struct task_struct *t)
> +{
> +	struct mm_struct *mm = t->mm;
> +	unsigned long flags;
> +
> +	if (!mm)
> +		return;
> +	local_irq_save(flags);
> +	mm_vcpu_put(mm, t->mm_vcpu);
> +	t->mm_vcpu = -1;
> +	t->mm_vcpu_active = 0;
> +	local_irq_restore(flags);
> +}
> +
> +void sched_vcpu_before_execve(struct task_struct *t)
> +{
> +	struct mm_struct *mm = t->mm;
> +	unsigned long flags;
> +
> +	if (!mm)
> +		return;
> +	local_irq_save(flags);
> +	mm_vcpu_put(mm, t->mm_vcpu);
> +	t->mm_vcpu = -1;
> +	t->mm_vcpu_active = 0;
> +	local_irq_restore(flags);
> +}
> +
> +void sched_vcpu_after_execve(struct task_struct *t)
> +{
> +	struct mm_struct *mm = t->mm;
> +	unsigned long flags;
> +
> +	WARN_ON_ONCE((t->flags & PF_KTHREAD) || !t->mm);
> +
> +	local_irq_save(flags);
> +	t->mm_vcpu = mm_vcpu_get(mm);
> +	t->mm_vcpu_active = 1;
> +	local_irq_restore(flags);
> +	rseq_set_notify_resume(t);
> +}

> +static inline void mm_vcpu_put(struct mm_struct *mm, int vcpu)
> +{
> +	lockdep_assert_irqs_disabled();
> +	if (vcpu < 0)
> +		return;
> +	spin_lock(&mm->vcpu_lock);
> +	__cpumask_clear_cpu(vcpu, mm_vcpumask(mm));
> +	spin_unlock(&mm->vcpu_lock);
> +}
> +
> +static inline int mm_vcpu_get(struct mm_struct *mm)
> +{
> +	int ret;
> +
> +	lockdep_assert_irqs_disabled();
> +	spin_lock(&mm->vcpu_lock);
> +	ret = __mm_vcpu_get(mm);
> +	spin_unlock(&mm->vcpu_lock);
> +	return ret;
> +}


This:

	local_irq_disable()
	spin_lock()

thing is a PREEMPT_RT anti-pattern.

At the very very least this should then be raw_spin_lock(), not in the
least because you're calling this from under rq->lock, which itself is a
raw_spin_lock_t.