Date:   Mon, 15 Nov 2021 08:45:42 +0100
From:   Lukas Wunner <lukas@wunner.de>
To:     "Michael S. Tsirkin" <mst@redhat.com>
Cc:     Gerd Hoffmann <kraxel@redhat.com>, linux-pci@vger.kernel.org,
        Bjorn Helgaas <bhelgaas@google.com>,
        open list <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] pciehp: fast unplug for virtual machines
Message-ID: <20211115074542.GA24942@wunner.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20211114173550-mutt-send-email-mst@kernel.org>
 <20211114122249-mutt-send-email-mst@kernel.org>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Sun, Nov 14, 2021 at 05:37:15PM -0500, Michael S. Tsirkin wrote:
> On Sun, Nov 14, 2021 at 07:06:04PM +0100, Lukas Wunner wrote:
> > Why don't you just trigger surprise-removal from outside the guest?
> 
> Because linux does not handle it well for all devices.  Fixing that
> requires fixing all drivers.

Please expose surprise removal in addition to the Attention Button
to avoid perpetuating this situation.  Teach users to use that for
devices whose drivers already support it well and fall back to the
Attention Button only for those that don't.

macOS and Windows support surprise-removal of most hardware well.
We've made a lot of progress in that area (AMD for one has been
working on surprise removal of GPUs).  It's an embarrassment if
contortions such as abusing the Attention Button are needed just
because surprise removal doesn't work for everything.  I'd like
to see engineering resources spent on fixing surprise removal for
drivers that are relevant for you, rather than implementing
workarounds such as this one.

Thanks,

Lukas