Received-SPF: pass (google.com: domain of linux-kernel+bounces-94513-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Date: Thu, 7 Mar 2024 04:54:21 +0900
From: Dominique Martinet <asmadeus@codewreck.org>
To: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Cc: Michael Grzeschik <m.grzeschik@pengutronix.de>,
	Eric Van Hensbergen <ericvh@kernel.org>,
	Latchesar Ionkov <lucho@ionkov.net>,
	Christian Schoenebeck <linux_oss@crudebyte.com>,
	Jonathan Corbet <corbet@lwn.net>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	v9fs@lists.linux.dev, linux-doc@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-usb@vger.kernel.org,
	kernel@pengutronix.de
Subject: Re: [PATCH v3 2/3] net/9p/usbg: Add new usb gadget function transport
Message-ID: <ZejJ7W_yUMBtqi70@codewreck.org>
References: <20240116-ml-topic-u9p-v3-0-c62a36eccda1@pengutronix.de>
 <20240116-ml-topic-u9p-v3-2-c62a36eccda1@pengutronix.de>
 <915890fa-e763-470a-a82b-eda97c47c0c8@collabora.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <915890fa-e763-470a-a82b-eda97c47c0c8@collabora.com>

Andrzej Pietrasiewicz wrote on Wed, Mar 06, 2024 at 04:18:54PM +0100:
> Reading the earlier discussions on v1 and v2 I was somehow under the impression
> that what you submitted here does not contain configfs support.
> 
> But once I started reading the code I realized it does.
> 
> It worked for me, both as a legacy gadget and when composed with configfs.
> 
> I noticed that when you stop the forwarder at the host side when 9pfs remains
> mounted at the gadget side, umount hangs at the gadget side until the forwarder
> is restarted. I know that once the host-side software dies, not much can be
> done, however, unmounting a dead filesystem seems a reasonable thing to do
> and the way usb9pfs worked for me it is not possible to unmount in this
> situation. Any ideas on improving this?

If the trans_usbg code can detect the host device is gone it should mark
client->status as Disconnected then wake up all pending requests
(something like p9_conn_cancel in trans_fd.c)

Sorry I haven't found time to test/review either; I'd still be
interested in running this in qemu for non-reg over using a real device
so need to spend a few minutes figuring out the exact command line
required.. I'll try to find some time this weekend

-- 
Dominique