mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Grant Likely <grant.likely@linaro.org>
To: NeilBrown <neilb@suse.de>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Mark Rutland <mark.rutland@arm.com>, Jiri Slaby <jslaby@suse.cz>
Cc: devicetree@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/3] TTY: add support for "tty slave" devices.
Date: Fri, 12 Dec 2014 11:59:20 +0000	[thread overview]
Message-ID: <20141212115920.4277AC408F1@trevor.secretlab.ca> (raw)
In-Reply-To: <20141211215943.4127.24792.stgit@notabene.brown>

On Fri, 12 Dec 2014 08:59:44 +1100
, NeilBrown <neilb@suse.de>
 wrote:
> A "tty slave" is a device connected via UART.
> It may need a driver to, for example, power the device on
> when the tty is opened, and power it off when the tty
> is released.

Some nit picking below, mostly about avoiding nasty surprises on devices
that may already have child nodes below the tty node.

> 
> A "tty slave" is a platform device which is declared as a
> child of the uart in device-tree:

As far as the binding is concerned, talking about platform devices isn't
relevant. That's an implementation detail. They are just devices.

Using a platform device is somewhat problematic for the kernel because
the DT could specify pretty much *any* platform device compatible value
here and the kernel will happily bind a driver to it. I strongly
recommend a separate bus type so that the pool of drivers remains
separate. It would be a good time to look at platform_bus_type and see
if more of it can be generalized so that subsystems can have custom bus
types without very much code at all.

> 
> &uart1 {
> 	bluetooth {
> 		compatible = "tty,regulator";
> 		vdd-supply = <&vaux4>;
> 	};

The example here isn't great. The compatible value should specify the
specific device, not a generic "tty,regulator" type binding. The
expectation is the driver understands the attachment and understands how
to do PM on the device.

> };
> 
> runtime power management is used to power-up the device
> on tty_open() and  power-down on tty_release().

What about devices that need to stay powered up even if the tty is not
opened? The runtime PM model needs to be better described for tty slaves.

> 
> Signed-off-by: NeilBrown <neilb@suse.de>
> ---
>  .../devicetree/bindings/serial/of-serial.txt       |    4 ++++
>  drivers/tty/tty_io.c                               |   22 ++++++++++++++++++++
>  2 files changed, 26 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/serial/of-serial.txt b/Documentation/devicetree/bindings/serial/of-serial.txt
> index 8c4fd0332028..b59501ee2f21 100644
> --- a/Documentation/devicetree/bindings/serial/of-serial.txt
> +++ b/Documentation/devicetree/bindings/serial/of-serial.txt
> @@ -39,6 +39,10 @@ Optional properties:
>    driver is allowed to detect support for the capability even without this
>    property.
>  
> +Optional child node:
> +- a platform device listed as a child node will be probed and
> +  powered-on whenever the tty is in use (open).
> +

The biggest concern I have is what happens to nodes that already have
child devices that /don't/ match this use case? It is possible that some
UART nodes already have a child node used to store other data. There are
two ways to handle this; 1) add a new bool property that indicates the
child nodes are tty slave devices, or 2) Make each uart driver
explicitly enable the feature so that driver authors can check if it is
a problem for that device. I personally would suggest #1 because then it
can be enabled in generic code.

>  Example:
>  
>  	uart@80230000 {
> diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
> index 0508a1d8e4cd..7acdc6f093f4 100644
> --- a/drivers/tty/tty_io.c
> +++ b/drivers/tty/tty_io.c
> @@ -95,6 +95,8 @@
>  #include <linux/seq_file.h>
>  #include <linux/serial.h>
>  #include <linux/ratelimit.h>
> +#include <linux/pm_runtime.h>
> +#include <linux/of_platform.h>
>  
>  #include <linux/uaccess.h>
>  
> @@ -1683,6 +1685,17 @@ static int tty_release_checks(struct tty_struct *tty, struct tty_struct *o_tty,
>  	return 0;
>  }
>  
> +static int open_child(struct device *dev, void *data)
> +{
> +	pm_runtime_get_sync(dev);
> +	return 0;
> +}
> +static int release_child(struct device *dev, void *data)
> +{
> +	pm_runtime_put_autosuspend(dev);
> +	return 0;
> +}
> +
>  /**
>   *	tty_release		-	vfs callback for close
>   *	@inode: inode of tty
> @@ -1712,6 +1725,8 @@ int tty_release(struct inode *inode, struct file *filp)
>  	long	timeout = 0;
>  	int	once = 1;
>  
> +	if (tty->dev)
> +		device_for_each_child(tty->dev, NULL, release_child);
>  	if (tty_paranoia_check(tty, inode, __func__))
>  		return 0;
>  
> @@ -2118,6 +2133,8 @@ retry_open:
>  		__proc_set_tty(current, tty);
>  	spin_unlock_irq(&current->sighand->siglock);
>  	tty_unlock(tty);
> +	if (tty->dev)
> +		device_for_each_child(tty->dev, NULL, open_child);
>  	mutex_unlock(&tty_mutex);
>  	return 0;
>  err_unlock:
> @@ -3207,6 +3224,11 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
>  	retval = device_register(dev);
>  	if (retval)
>  		goto error;
> +	if (device && device->of_node)
> +		/* Children are platform devices and will be
> +		 * runtime_pm managed by this tty.
> +		 */
> +		of_platform_populate(device->of_node, NULL, NULL, dev);

Also need to remove all the tty slaves on driver unbind.

>  
>  	return dev;
>  
> 
> 


  parent reply	other threads:[~2014-12-12 17:46 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-12-11 21:59 [PATCH 0/3] Add support for 'tty-slaves' described by devicetree NeilBrown
2014-12-11 21:59 ` [PATCH 1/3] TTY: add support for "tty slave" devices NeilBrown
2014-12-11 22:41   ` Sebastian Reichel
2014-12-11 23:18   ` Peter Hurley
2014-12-12  5:23     ` NeilBrown
2014-12-12 13:02       ` Peter Hurley
2014-12-13 14:23         ` One Thousand Gnomes
2014-12-16 16:14           ` Peter Hurley
2014-12-13 14:12     ` One Thousand Gnomes
2014-12-12 11:59   ` Grant Likely [this message]
2014-12-13 17:46     ` Sebastian Reichel
2014-12-13 22:22       ` Grant Likely
2014-12-28 14:20   ` Pavel Machek
2015-01-02 21:33     ` NeilBrown
2015-01-04 10:18       ` Pavel Machek
2015-01-05  7:09         ` NeilBrown
2015-01-05 13:43           ` Pavel Machek
2015-01-05 15:41       ` One Thousand Gnomes
2015-01-05 16:28         ` Pavel Machek
2014-12-11 21:59 ` [PATCH 2/3] TTY: add slave driver to power-on device via a regulator NeilBrown
2014-12-11 22:58   ` Sebastian Reichel
2014-12-12  0:46     ` Marcel Holtmann
2014-12-12  1:31       ` Sebastian Reichel
2014-12-12  5:01     ` NeilBrown
2014-12-11 23:32   ` Peter Hurley
2014-12-12  5:27     ` NeilBrown
2014-12-12 11:59       ` Peter Hurley
2014-12-12 12:05   ` Grant Likely
2014-12-11 21:59 ` [PATCH 3/3] TTY/slave: add driver for w2sg0004 GPS NeilBrown
2014-12-11 23:04   ` Sebastian Reichel
2014-12-11 23:11   ` One Thousand Gnomes
2014-12-12  5:06     ` NeilBrown
2014-12-15 11:39       ` One Thousand Gnomes
2014-12-12 12:11   ` Grant Likely

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20141212115920.4277AC408F1@trevor.secretlab.ca \
    --to=grant.likely@linaro.org \
    --cc=devicetree@vger.kernel.org \
    --cc=gregkh@linuxfoundation.org \
    --cc=jslaby@suse.cz \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=neilb@suse.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®