Received-SPF: pass (google.com: domain of linux-kernel+bounces-184752-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Tue, 21 May 2024 12:06:58 +0200
From: Frederic Weisbecker <frederic@kernel.org>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: Ming Lei <ming.lei@redhat.com>, Christoph Hellwig <hch@lst.de>,
	Keith Busch <kbusch@kernel.org>, Keith Busch <kbusch@meta.com>,
	linux-nvme@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] nvme-pci: allow unmanaged interrupts
Message-ID: <ZkxyQmR0cpT7-zKY@localhost.localdomain>
References: <20240510141459.3207725-1-kbusch@meta.com>
 <20240510141459.3207725-2-kbusch@meta.com>
 <20240510151047.GA10486@lst.de>
 <Zj5JMqWRY187PqnD@kbusch-mbp.dhcp.thefacebook.com>
 <Zj6yvTxIpUnOXl7R@fedora>
 <Zj6-1sXvUNZWO1pB@kbusch-mbp.dhcp.thefacebook.com>
 <87r0e7mt9w.ffs@tglx>
 <20240520153742.GB921@lst.de>
 <ZkwHnw6HkGFFq7mR@fedora>
 <87a5kjimpq.ffs@tglx>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <87a5kjimpq.ffs@tglx>

Le Tue, May 21, 2024 at 10:38:25AM +0200, Thomas Gleixner a ?crit :
> On Tue, May 21 2024 at 10:31, Ming Lei wrote:
> 
> > On Mon, May 20, 2024 at 05:37:42PM +0200, Christoph Hellwig wrote:
> >> On Sun, May 12, 2024 at 08:35:55AM +0200, Thomas Gleixner wrote:
> >> > That's expected as you pin the I/O operation on the isolated CPUs which
> >> > in turn makes them use the per CPU queue.
> >> > 
> >> > The isolated CPUs are only excluded for device management interrupts,
> >> > but not for the affinity spread of the queues.
> >> 
> >> We'll probably need a version of isolcpus that also excludes the
> >> interrupt spread given that users are asking for it.  And I'd much
> >> prefer that over adding radom module options to every driver to disable
> >> managed interrupts.
> >
> > BTW, isolcpus has been marked as deprecated, and it can't be adjust
> > runtime.
> 
> Which is far from reality as cpusets do not allow to do what isolcpus
> does today.
> 
> Also runtime adjusting managed interrupts needs way more thoughts.

I'll remove that comment (unless someone beats me at it?). We used to think
that cpusets would indeed deprecate isolcpus but for several reasons this
will never be the case.

Thanks.