Re: [PATCH v2 4/6] staging:iio:events: Use waitqueue lock to protect event queue

[Lars-Peter Clausen <lars@metafoo.de>]

On 12/19/2011 09:57 PM, Jonathan Cameron wrote:
> On 12/19/2011 10:29 AM, Lars-Peter Clausen wrote:
>> Use the waitqueue lock to protect the event queue instead of a custom mutex.
>> This has the advantage that we can call the waitqueue operations with the lock
>> held, which simplifies the code flow a bit.
>>
> Unusual ordering of this.  usually your sign off first then everyone
> elses whatever below.

This is mainly due to 'git commit --amend -s' adding another Signed-off-by if
your Signed-off-by is not the last line in the commit message and me getting
annoyed at it ;)

>> Acked-by: Jonathan Cameron <jic23@kernel.org>
>> Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
>> ---
>>  drivers/staging/iio/industrialio-event.c |   45 ++++++++++++-----------------
>>  1 files changed, 19 insertions(+), 26 deletions(-)
>>
>> diff --git a/drivers/staging/iio/industrialio-event.c b/drivers/staging/iio/industrialio-event.c
>> index d63aa0b..b5f7a2c 100644
>> --- a/drivers/staging/iio/industrialio-event.c
>> +++ b/drivers/staging/iio/industrialio-event.c
>> @@ -34,8 +34,7 @@
>>   * @group:		event interface sysfs attribute group
>>   */
>>  struct iio_event_interface {
>> -	wait_queue_head_t	wait;
>> -	struct mutex		event_list_lock;
>> +	wait_queue_head_t			wait;
>>  	DECLARE_KFIFO(det_events, struct iio_event_data, 16);
>>  
>>  	struct list_head	dev_attr_list;
>> @@ -50,19 +49,17 @@ int iio_push_event(struct iio_dev *indio_dev, u64 ev_code, s64 timestamp)
>>  	int copied;
>>  
>>  	/* Does anyone care? */
>> -	mutex_lock(&ev_int->event_list_lock);
>> +	spin_lock(&ev_int->wait.lock);
>>  	if (test_bit(IIO_BUSY_BIT_POS, &ev_int->flags)) {
>>  
>>  		ev.id = ev_code;
>>  		ev.timestamp = timestamp;
>>  
>>  		copied = kfifo_put(&ev_int->det_events, &ev);
>> -
>> -		mutex_unlock(&ev_int->event_list_lock);
>>  		if (copied != 0)
>> -			wake_up_interruptible(&ev_int->wait);
>> -	} else
>> -		mutex_unlock(&ev_int->event_list_lock);
>> +			wake_up_locked(&ev_int->wait);
>> +	}
>> +	spin_unlock(&ev_int->wait.lock);
>>  
>>  	return 0;
>>  }
>> @@ -80,28 +77,25 @@ static ssize_t iio_event_chrdev_read(struct file *filep,
>>  	if (count < sizeof(struct iio_event_data))
>>  		return -EINVAL;
>>  
>> -	mutex_lock(&ev_int->event_list_lock);
>> +	spin_lock(&ev_int->wait.lock);
>>  	if (kfifo_is_empty(&ev_int->det_events)) {
>>  		if (filep->f_flags & O_NONBLOCK) {
>>  			ret = -EAGAIN;
>> -			goto error_mutex_unlock;
>> +			goto error_unlock;
>>  		}
>> -		mutex_unlock(&ev_int->event_list_lock);
>>  		/* Blocking on device; waiting for something to be there */
>> -		ret = wait_event_interruptible(ev_int->wait,
>> +		ret = wait_event_interruptible_locked(ev_int->wait,
>>  					!kfifo_is_empty(&ev_int->det_events));
>>  		if (ret)
>> -			goto error_ret;
>> +			goto error_unlock;
>>  		/* Single access device so no one else can get the data */
>> -		mutex_lock(&ev_int->event_list_lock);
>>  	}
>>  
>> -	mutex_unlock(&ev_int->event_list_lock);
>>  	ret = kfifo_to_user(&ev_int->det_events, buf, count, &copied);
>>  
>> -error_mutex_unlock:
>> -	mutex_unlock(&ev_int->event_list_lock);
>> -error_ret:
>> +error_unlock:
>> +	spin_unlock(&ev_int->wait.lock);
>> +
>>  	return ret ? ret : copied;
>>  }
>>  
>> @@ -109,7 +103,7 @@ static int iio_event_chrdev_release(struct inode *inode, struct file *filep)
>>  {
>>  	struct iio_event_interface *ev_int = filep->private_data;
>>  
>> -	mutex_lock(&ev_int->event_list_lock);
>> +	spin_lock(&ev_int->wait.lock);
>>  	clear_bit(IIO_BUSY_BIT_POS, &ev_int->flags);
>>  	/*
>>  	 * In order to maintain a clean state for reopening,
>> @@ -117,7 +111,7 @@ static int iio_event_chrdev_release(struct inode *inode, struct file *filep)
>>  	 * any new __iio_push_event calls running.
>>  	 */
>>  	kfifo_reset_out(&ev_int->det_events);
>> -	mutex_unlock(&ev_int->event_list_lock);
>> +	spin_unlock(&ev_int->wait.lock);
>>  
>>  	return 0;
>>  }
>> @@ -137,18 +131,18 @@ int iio_event_getfd(struct iio_dev *indio_dev)
>>  	if (ev_int == NULL)
>>  		return -ENODEV;
>>  
>> -	mutex_lock(&ev_int->event_list_lock);
>> +	spin_lock(&ev_int->wait.lock);
>>  	if (test_and_set_bit(IIO_BUSY_BIT_POS, &ev_int->flags)) {
>> -		mutex_unlock(&ev_int->event_list_lock);
>> +		spin_unlock(&ev_int->wait.lock);
>>  		return -EBUSY;
>>  	}
>> -	mutex_unlock(&ev_int->event_list_lock);
>> +	spin_unlock(&ev_int->wait.lock);
>>  	fd = anon_inode_getfd("iio:event",
>>  				&iio_event_chrdev_fileops, ev_int, O_RDONLY);
>>  	if (fd < 0) {
>> -		mutex_lock(&ev_int->event_list_lock);
>> +		spin_lock(&ev_int->wait.lock);
>>  		clear_bit(IIO_BUSY_BIT_POS, &ev_int->flags);
>> -		mutex_unlock(&ev_int->event_list_lock);
>> +		spin_unlock(&ev_int->wait.lock);
>>  	}
>>  	return fd;
>>  }
>> @@ -357,7 +351,6 @@ static bool iio_check_for_dynamic_events(struct iio_dev *indio_dev)
>>  
>>  static void iio_setup_ev_int(struct iio_event_interface *ev_int)
>>  {
>> -	mutex_init(&ev_int->event_list_lock);
>>  	INIT_KFIFO(ev_int->det_events);
>>  	init_waitqueue_head(&ev_int->wait);
>>  }
>