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;
Date:   Tue, 7 Jul 2020 11:18:05 +0100
From:   Christoph Hellwig <hch@infradead.org>
To:     Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc:     Christoph Hellwig <hch@infradead.org>,
        Simon Arlott <simon@octiron.net>,
        "James E.J. Bottomley" <jejb@linux.ibm.com>,
        "Martin K. Petersen" <martin.petersen@oracle.com>,
        Jonathan Corbet <corbet@lwn.net>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        linux-scsi@vger.kernel.org, linux-doc@vger.kernel.org
Subject: Re: [PATCH] scsi: sd: stop SSD (non-rotational) disks before reboot
Message-ID: <20200707101805.GA11097@infradead.org>
References: <499138c8-b6d5-ef4a-2780-4f750ed337d3@0882a8b5-c6c3-11e9-b005-00805fc181fe>
 <20200618072138.GA11778@infradead.org>
 <9877e7de-d573-694b-2b75-95192756684b@0882a8b5-c6c3-11e9-b005-00805fc181fe>
 <20200618134904.GA26650@infradead.org>
 <20200705213125.GC8285@khazad-dum.debian.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200705213125.GC8285@khazad-dum.debian.net>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sun, Jul 05, 2020 at 06:31:25PM -0300, Henrique de Moraes Holschuh wrote:
> On Thu, 18 Jun 2020, Christoph Hellwig wrote:
> > > For SSDs, I don't think an extra stop should ever be an issue.
> > 
> > Extra shutdowns will usually cause additional P/E cycles.
> 
> I am not so sure.  We're talking about enforcing clean shutdowns here
> (from the SSD PoV).
> 
> A system reboot takes enough time that the SSD is likely to do about the
> same amount of P cycles commiting to FLASH any important data that it
> would trigger by a shutdown sequence, simply because it should not keep
> important data in RAM for too long.  By extension, it would not increase
> E cycles either.
> 
> OTOH, unclean shutdowns *always* cause extra P/E, and that's if you're
> lucky enough for it to not cause anything much worse.

The point is - with a normal system that doesn't required your odd
reboot method we'll normally not shut down the SSD at all, and that
won't require a P/E cycle.

But the whole thing is a moot point - if you quirk your system to
require a poweroff to reboot the kernel should trat it as a power off
as far as shutdown/remove callbacks are concerned and everything will
just work as intended.