Linux IIO development
 help / color / mirror / Atom feed
From: Jonathan Cameron <jic23@kernel.org>
To: "Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>
Cc: William Breathitt Gray <vilhelm.gray@gmail.com>,
	Lars-Peter Clausen <lars@metafoo.de>,
	kernel@pengutronix.de,
	Jonathan Cameron <Jonathan.Cameron@huawei.com>,
	linux-iio@vger.kernel.org,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 13/23] counter: Provide alternative counter registration functions
Date: Tue, 28 Dec 2021 18:00:17 +0000	[thread overview]
Message-ID: <20211228180017.37c2703d@jic23-huawei> (raw)
In-Reply-To: <20211227094526.698714-14-u.kleine-koenig@pengutronix.de>

On Mon, 27 Dec 2021 10:45:16 +0100
Uwe Kleine-König         <u.kleine-koenig@pengutronix.de> wrote:

> The current implementation gets device lifetime tracking wrong. The
> problem is that allocation of struct counter_device is controlled by the
> individual drivers but this structure contains a struct device that
> might have to live longer than a driver is bound. As a result a command
> sequence like:
> 
> 	{ sleep 5; echo bang; } > /dev/counter0 &
> 	sleep 1;
> 	echo 40000000.timer:counter > /sys/bus/platform/drivers/stm32-timer-counter/unbind
> 
> can keep a reference to the struct device and unbinding results in
> freeing the memory occupied by this device resulting in an oops.
> 
> This commit provides two new functions (plus some helpers):
>  - counter_alloc() to allocate a struct counter_device that is
>    automatically freed once the embedded struct device is released
>  - counter_add() to register such a device.
> 
> Note that this commit doesn't fix any issues, all drivers have to be
> converted to these new functions to correct the lifetime problems.
> 
> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
In general looks very nice to me.

Few minor things inline.
> ---
>  drivers/counter/counter-core.c | 149 ++++++++++++++++++++++++++++++++-
>  include/linux/counter.h        |  15 ++++
>  2 files changed, 163 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/counter/counter-core.c b/drivers/counter/counter-core.c
> index 00c41f28c101..8261567b6272 100644
> --- a/drivers/counter/counter-core.c
> +++ b/drivers/counter/counter-core.c
> @@ -15,6 +15,7 @@
>  #include <linux/kdev_t.h>
>  #include <linux/module.h>
>  #include <linux/mutex.h>
> +#include <linux/slab.h>
>  #include <linux/types.h>
>  #include <linux/wait.h>
>  
> @@ -24,6 +25,11 @@
>  /* Provides a unique ID for each counter device */
>  static DEFINE_IDA(counter_ida);
>  
> +struct counter_device_allochelper {
> +	struct counter_device counter;
> +	unsigned long privdata[];
Nice to keep this opaque. We danced around how to do this in IIO for
a while and never got to a solution we all liked.  Slight disadvantage
is this might matter in hot paths where the compiler doesn't have visibility
to know it can very cheaply access the privdata.

I guess that can be looked at if anyone ever measures it as an important
element (they probably never will!)

> +};
> +
>  static void counter_device_release(struct device *dev)
>  {
>  	struct counter_device *const counter =
> @@ -31,6 +37,9 @@ static void counter_device_release(struct device *dev)
>  
>  	counter_chrdev_remove(counter);
>  	ida_free(&counter_ida, dev->id);
> +
> +	if (!counter->legacy_device)
> +		kfree(container_of(counter, struct counter_device_allochelper, counter));
>  }
>  
>  static struct device_type counter_device_type = {
> @@ -53,7 +62,14 @@ static dev_t counter_devt;
>   */
>  void *counter_priv(const struct counter_device *const counter)
>  {
> -	return counter->priv;
> +	if (counter->legacy_device) {
> +		return counter->priv;
> +	} else {
> +		struct counter_device_allochelper *ch =
> +			container_of(counter, struct counter_device_allochelper, counter);
> +
> +		return &ch->privdata;
> +	}
>  }
>  EXPORT_SYMBOL_GPL(counter_priv);
>  
> @@ -74,6 +90,8 @@ int counter_register(struct counter_device *const counter)
>  	int id;
>  	int err;
>  
> +	counter->legacy_device = true;
> +
>  	/* Acquire unique ID */
>  	id = ida_alloc(&counter_ida, GFP_KERNEL);
>  	if (id < 0)
> @@ -114,6 +132,100 @@ int counter_register(struct counter_device *const counter)
>  }
>  EXPORT_SYMBOL_GPL(counter_register);
>  
> +/**
> + * counter_alloc - allocate a counter_device
> + * @sizeof_priv: size of the driver private data
> + *
> + * This is part one of counter registration. The structure is allocated
> + * dynamically to ensure the right lifetime for the embedded struct device.
> + *
> + * If this succeeds, call counter_put() to get rid of the counter_device again.
> + */
> +struct counter_device *counter_alloc(size_t sizeof_priv)
> +{
> +	struct counter_device_allochelper *ch;
> +	struct counter_device *counter;
> +	struct device *dev;
> +	int id, err;
> +
> +	ch = kzalloc(sizeof(*ch) + sizeof_priv, GFP_KERNEL);
> +	if (!ch) {
> +		err = -ENOMEM;
> +		goto err_alloc_ch;
> +	}
> +
> +	counter = &ch->counter;
> +	dev = &counter->dev;
> +
> +	/* Acquire unique ID */
> +	err = ida_alloc(&counter_ida, GFP_KERNEL);
> +	if (err < 0) {
> +		goto err_ida_alloc;
> +	}
> +	dev->id = err;
> +
> +	err = counter_chrdev_add(counter);

Should think about renaming  counter_chrdev_add() given it
does init and allocation stuff but no adding.

Personal inclination here would be to keep the elements in here
in the same order as in counter_register() where it doesn't matter
in the interests of slightly easier comparison of the code.

> +	if (err < 0)
> +		goto err_chrdev_add;
> +
> +	device_initialize(dev);
> +	/* Configure device structure for Counter */
> +	dev->type = &counter_device_type;
> +	dev->bus = &counter_bus_type;
> +	dev->devt = MKDEV(MAJOR(counter_devt), id);

As 0-day pointed out id not initialized.

> +
> +	mutex_init(&counter->ops_exist_lock);
> +
> +	return counter;
> +
> +err_chrdev_add:
> +
> +	ida_free(&counter_ida, dev->id);
> +err_ida_alloc:
> +
> +	kfree(ch);
> +err_alloc_ch:
> +
> +	return ERR_PTR(err);
> +}
> +EXPORT_SYMBOL_GPL(counter_alloc);
> +
> +void counter_put(struct counter_device *counter)
> +{
> +	put_device(&counter->dev);
> +}
> +
> +/**
> + * counter_add - complete registration of a counter
> + * @counter: the counter to add
> + *
> + * This is part two of counter registration.
> + *
> + * If this succeeds, call counter_unregister() to get rid of the counter_device again.
> + */
> +int counter_add(struct counter_device *counter)
> +{
> +	int err;
> +	struct device *dev = &counter->dev;
> +
> +	get_device(&counter->dev);
> +
> +	if (counter->parent) {
> +		dev->parent = counter->parent;
> +		dev->of_node = counter->parent->of_node;
> +	}
> +
> +	err = counter_sysfs_add(counter);
> +	if (err < 0)
> +		return err;
> +
> +	/* implies device_add(dev) */
> +	err = cdev_device_add(&counter->chrdev, dev);
> +
> +	return err;

return cdev_device_add(&counter->chrdev, dev);
Lars got some of these points as well (maybe all of them but
I'm too lazy to filter them for you ;)


> +}
> +EXPORT_SYMBOL_GPL(counter_add);
> +
>  /**
>   * counter_unregister - unregister Counter from the system
>   * @counter:	pointer to Counter to unregister
> @@ -168,6 +280,41 @@ int devm_counter_register(struct device *dev,
>  }
>  EXPORT_SYMBOL_GPL(devm_counter_register);
>  
> +static void devm_counter_put(void *counter)
> +{
> +	counter_put(counter);
> +}
> +
> +struct counter_device *devm_counter_alloc(struct device *dev, size_t sizeof_priv)
> +{
> +	struct counter_device *counter;
> +	int err;
> +
> +	counter = counter_alloc(sizeof_priv);
> +	if (IS_ERR(counter))
> +		return counter;
> +
> +	err = devm_add_action_or_reset(dev, devm_counter_put, counter);
> +	if (err < 0)
> +		return ERR_PTR(err);
> +
> +	return counter;
> +}
> +EXPORT_SYMBOL_GPL(devm_counter_alloc);
> +
> +int devm_counter_add(struct device *dev,
> +		     struct counter_device *const counter)
> +{
> +	int err;
> +
> +	err = counter_add(counter);
> +	if (err < 0)
> +		return err;
> +
> +	return devm_add_action_or_reset(dev, devm_counter_release, counter);
> +}
> +EXPORT_SYMBOL_GPL(devm_counter_add);
> +
>  #define COUNTER_DEV_MAX 256
>  
>  static int __init counter_init(void)
> diff --git a/include/linux/counter.h b/include/linux/counter.h
> index 8daaa38c71d8..f1350a43cd48 100644
> --- a/include/linux/counter.h
> +++ b/include/linux/counter.h
> @@ -327,14 +327,29 @@ struct counter_device {
>  	spinlock_t events_in_lock;
>  	struct mutex events_out_lock;
>  	struct mutex ops_exist_lock;
> +
> +	/*
> +	 * This can go away once all drivers are converted to
> +	 * counter_alloc()/counter_add().
> +	 */
> +	bool legacy_device;
>  };
>  
>  void *counter_priv(const struct counter_device *const counter);
>  
>  int counter_register(struct counter_device *const counter);
> +
> +struct counter_device *counter_alloc(size_t sizeof_priv);
> +void counter_put(struct counter_device *const counter);
> +int counter_add(struct counter_device *const counter);
> +
>  void counter_unregister(struct counter_device *const counter);
>  int devm_counter_register(struct device *dev,
>  			  struct counter_device *const counter);
> +struct counter_device *devm_counter_alloc(struct device *dev,
> +					  size_t sizeof_priv);
> +int devm_counter_add(struct device *dev,
> +		     struct counter_device *const counter);
>  void counter_push_event(struct counter_device *const counter, const u8 event,
>  			const u8 channel);
>  


  parent reply	other threads:[~2021-12-28 17:54 UTC|newest]

Thread overview: 100+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-27  9:45 [PATCH v2 00/23] counter: cleanups and device lifetime fixes Uwe Kleine-König
2021-12-27  9:45 ` [PATCH v2 01/23] counter: Use container_of instead of drvdata to track counter_device Uwe Kleine-König
2021-12-28 17:14   ` Jonathan Cameron
2021-12-29  6:38   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 02/23] counter: ftm-quaddec: Drop unused platform_set_drvdata() Uwe Kleine-König
2021-12-28 17:37   ` Jonathan Cameron
2021-12-29  6:42   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 03/23] counter: microchip-tcb-capture: " Uwe Kleine-König
2021-12-28 17:38   ` Jonathan Cameron
2021-12-29  6:44   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 04/23] counter: Provide a wrapper to access device private data Uwe Kleine-König
2021-12-27 11:02   ` Greg Kroah-Hartman
2021-12-27 11:34     ` Lars-Peter Clausen
2021-12-27 11:52       ` Greg Kroah-Hartman
2021-12-27 12:28         ` Lars-Peter Clausen
2021-12-28 18:01   ` Jonathan Cameron
2021-12-29  6:47   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 05/23] counter: 104-quad-8: Convert to counter_priv() wrapper Uwe Kleine-König
2021-12-28 18:02   ` Jonathan Cameron
2021-12-29 11:11     ` Uwe Kleine-König
2021-12-29  7:27   ` William Breathitt Gray
2021-12-29 11:15     ` Uwe Kleine-König
2021-12-29 22:40       ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 06/23] counter: interrupt-cnt: " Uwe Kleine-König
2021-12-28 18:04   ` Jonathan Cameron
2021-12-29  7:34   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 07/23] counter: microchip-tcb-capture: " Uwe Kleine-König
2021-12-28 18:05   ` Jonathan Cameron
2021-12-29  7:35   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 08/23] counter: intel-qep: " Uwe Kleine-König
2021-12-28 18:06   ` Jonathan Cameron
2021-12-29  7:35   ` William Breathitt Gray
2021-12-29 12:43   ` Jarkko Nikula
2021-12-27  9:45 ` [PATCH v2 09/23] counter: ftm-quaddec: " Uwe Kleine-König
2021-12-28 18:07   ` Jonathan Cameron
2021-12-29  7:36   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 10/23] counter: ti-eqep: " Uwe Kleine-König
2021-12-28 18:07   ` Jonathan Cameron
2021-12-29  7:45   ` William Breathitt Gray
2021-12-29 11:19     ` Uwe Kleine-König
2021-12-27  9:45 ` [PATCH v2 11/23] counter: stm32-lptimer-cnt: " Uwe Kleine-König
2021-12-28 18:08   ` Jonathan Cameron
2021-12-29  7:48   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 12/23] counter: stm32-timer-cnt: " Uwe Kleine-König
2021-12-28 18:08   ` Jonathan Cameron
2021-12-29  7:48   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 13/23] counter: Provide alternative counter registration functions Uwe Kleine-König
2021-12-27 12:16   ` Lars-Peter Clausen
2021-12-28 11:22     ` Uwe Kleine-König
2021-12-27 12:18   ` kernel test robot
2021-12-28 18:00   ` Jonathan Cameron [this message]
2021-12-29 11:27     ` Uwe Kleine-König
2021-12-29 12:07       ` Uwe Kleine-König
2021-12-29 13:49     ` Jarkko Nikula
2021-12-29 15:47       ` Uwe Kleine-König
2021-12-29  8:13   ` William Breathitt Gray
2022-01-06 10:53   ` Dan Carpenter
2021-12-27  9:45 ` [PATCH v2 14/23] counter: Update documentation for new " Uwe Kleine-König
2021-12-28 18:12   ` Jonathan Cameron
2021-12-29  8:19     ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 15/23] counter: 104-quad-8: Convert to new counter registration Uwe Kleine-König
2021-12-27 12:19   ` Lars-Peter Clausen
2021-12-28 11:06     ` Uwe Kleine-König
2021-12-28 18:17   ` Jonathan Cameron
2021-12-29  8:24   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 16/23] counter: interrupt-cnt: " Uwe Kleine-König
2021-12-27 10:59   ` Greg Kroah-Hartman
2021-12-27 11:21     ` Lars-Peter Clausen
2021-12-27 11:34       ` Greg Kroah-Hartman
2021-12-27 11:44         ` Lars-Peter Clausen
2021-12-27 21:09         ` Uwe Kleine-König
2021-12-28 18:18   ` Jonathan Cameron
2021-12-29  8:27   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 17/23] counter: intel-qep: " Uwe Kleine-König
2021-12-27 15:02   ` Jarkko Nikula
2021-12-28 10:56     ` Uwe Kleine-König
2021-12-29 12:42       ` Jarkko Nikula
2021-12-28 18:20   ` Jonathan Cameron
2021-12-29  8:29   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 18/23] counter: ftm-quaddec: " Uwe Kleine-König
2021-12-28 18:23   ` Jonathan Cameron
2021-12-29  8:33   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 19/23] counter: microchip-tcb-capture: " Uwe Kleine-König
2021-12-28 18:24   ` Jonathan Cameron
2021-12-29  8:36   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 20/23] counter: stm32-timer-cnt: " Uwe Kleine-König
2021-12-28 18:25   ` Jonathan Cameron
2021-12-29  8:37   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 21/23] counter: stm32-lptimer-cnt: " Uwe Kleine-König
2021-12-28 18:26   ` Jonathan Cameron
2021-12-29  8:39   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 22/23] counter: ti-eqep: " Uwe Kleine-König
2021-12-28 18:29   ` Jonathan Cameron
2021-12-29  8:41   ` William Breathitt Gray
2021-12-27  9:45 ` [PATCH v2 23/23] counter: remove old and now unused registration API Uwe Kleine-König
2021-12-28 18:31   ` Jonathan Cameron
2021-12-29  8:43   ` William Breathitt Gray
2021-12-27 12:25 ` [PATCH v2 00/23] counter: cleanups and device lifetime fixes Lars-Peter Clausen
2021-12-28 17:35   ` Jonathan Cameron
2021-12-29  8:49     ` William Breathitt Gray

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=20211228180017.37c2703d@jic23-huawei \
    --to=jic23@kernel.org \
    --cc=Jonathan.Cameron@huawei.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=kernel@pengutronix.de \
    --cc=lars@metafoo.de \
    --cc=linux-iio@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=u.kleine-koenig@pengutronix.de \
    --cc=vilhelm.gray@gmail.com \
    /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