[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200423114326.GQ18608@localhost>
Date: Thu, 23 Apr 2020 13:43:26 +0200
From: Johan Hovold <johan@...nel.org>
To: Tony Lindgren <tony@...mide.com>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
Rob Herring <robh@...nel.org>,
Alan Cox <gnomes@...rguk.ukuu.org.uk>,
Lee Jones <lee.jones@...aro.org>, Jiri Slaby <jslaby@...e.cz>,
Johan Hovold <johan@...nel.org>,
Merlijn Wajer <merlijn@...zup.org>,
Pavel Machek <pavel@....cz>,
Peter Hurley <peter@...leysoftware.com>,
Sebastian Reichel <sre@...nel.org>,
linux-serial@...r.kernel.org, devicetree@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-omap@...r.kernel.org
Subject: Re: [PATCHv6 0/4] n_gsm serdev support and protocol driver for
droid4 modem
Hi Tony,
On Tue, Apr 21, 2020 at 04:27:48PM -0700, Tony Lindgren wrote:
> Hi all,
>
> Here's v4 set of n_gsm serdev support patches, and the related protocol
> driver for the modem found on Motorola Mapphone phones and tablets
> like droid4.
>
> This series only adds basic character device support for the serdev
> driver. Other serdev consumer drivers for specific devices will be
> posted separately.
I'm still missing an architectural (design) overview here -- reviewer
time is a scarce resource.
I also suggested earlier that you include, at least as an RFC, one or
more of your child-device drivers so that we can see how this ends up
being used in the end (including an example devicetree).
Some high-level comments until then:
I'm not sure that a plain chardev interface for the mux channels is the
right interface. The n_gsm ldisc exposes tty devices and I think your
serdev adaptation should continue to do that.
On that note; you're not actually adding general TS 27.010 serdev
support, but rather some hooks and a custom driver and interface (mfd +
/dev/motmdmN) for one particular modem.
I'd rather see a generic implementation which can be used with other
modems and that continues to expose a /dev/gsmttyN interface to which we
could attach serdev clients instead (and not create a motmdm serdev
replica of sorts).
I know the location of this driver has been up for discussion already,
but drivers/tty/serdev/protocol still isn't right (e.g. we don't have an
drivers/i2c/protocol directory where we stuff random i2c client
drivers).
It's an mfd + custom chardev driver for a modem and related to n_gsm
(even more if you add generic serdev support). Currently, drivers/mfd or
drivers/misc appear to be better choices. Otherwise, n_gsm lives in
drivers/tty since it's a line discipline, but it could be moved to a new
drivers/modem if needed (cf. the bluetooth hci ldisc).
Last, it seems you've based the serdev-ngsm-motmdm.c chardev
implementation on a more or less verbatim copy of drivers/gnss/core.c.
I'd appreciate if you could mention that in the file header and
reproduce the copyright notice if you end up keeping that interface.
> Tony Lindgren (4):
> tty: n_gsm: Add support for serdev drivers
> serdev: ngsm-motmdm: Add Motorola TS 27.010 serdev modem driver for
> droid4
> dt-bindings: serdev: motmdm: Add binding for motorola-mdm
> ARM: dts: omap4-droid4: Enable basic modem support
>
> .../serdev/motorola,mapphone-mdm6600.yaml | 34 +
> .../boot/dts/motorola-mapphone-common.dtsi | 6 +
> drivers/tty/n_gsm.c | 372 +++++
> drivers/tty/serdev/Kconfig | 2 +
> drivers/tty/serdev/Makefile | 2 +
> drivers/tty/serdev/protocol/Kconfig | 14 +
> drivers/tty/serdev/protocol/Makefile | 3 +
> .../tty/serdev/protocol/serdev-ngsm-motmdm.c | 1191 +++++++++++++++++
> include/linux/serdev-gsm.h | 168 +++
> 9 files changed, 1792 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/serdev/motorola,mapphone-mdm6600.yaml
> create mode 100644 drivers/tty/serdev/protocol/Kconfig
> create mode 100644 drivers/tty/serdev/protocol/Makefile
> create mode 100644 drivers/tty/serdev/protocol/serdev-ngsm-motmdm.c
> create mode 100644 include/linux/serdev-gsm.h
Johan
Powered by blists - more mailing lists