Re: [PATCH v4 1/6] iio: Add output buffer support

From: Jonathan Cameron
Date: Mon Aug 30 2021 - 12:02:50 EST


On Fri, 20 Aug 2021 16:59:22 +0000
Mihail Chindris <mihail.chindris@xxxxxxxxxx> wrote:

> From: Lars-Peter Clausen <lars@xxxxxxxxxx>
>
> Currently IIO only supports buffer mode for capture devices like ADCs. Add
> support for buffered mode for output devices like DACs.
>
> The output buffer implementation is analogous to the input buffer
> implementation. Instead of using read() to get data from the buffer write()
> is used to copy data into the buffer.
>
> poll() with POLLOUT will wakeup if there is space available for more or
> equal to the configured watermark of samples.
>
> Drivers can remove data from a buffer using iio_buffer_remove_sample(), the
> function can e.g. called from a trigger handler to write the data to
> hardware.
>
> A buffer can only be either a output buffer or an input, but not both. So,
> for a device that has an ADC and DAC path, this will mean 2 IIO buffers
> (one for each direction).
>
> The direction of the buffer is decided by the new direction field of the
> iio_buffer struct and should be set after allocating and before registering
> it.
>
> Signed-off-by: Lars-Peter Clausen <lars@xxxxxxxxxx>
> Signed-off-by: Alexandru Ardelean <alexandru.ardelean@xxxxxxxxxx>
> Signed-off-by: Mihail Chindris <mihail.chindris@xxxxxxxxxx>
Hi Mihial,

Welcome to IIO (I don't think I've seen you before?)

Given the somewhat odd sign off trail I'd add some comments to the description
(probably not saying that everyone who works on this ends up leaving Analog.
It's not cursed! Really it's not ;) Lars and I discussed this at least 7+ years
ago and he lasted ages at Analog after that *evil laugh*

I'm not really clear how the concept of a watermark applies here. It feels
like it's getting used for two unrelated things:
1) Space in buffer for polling form userspace. We let userspace know it can
write more data once the watermark worth of scans is empty.
2) Writing to the kfifo. If a large write is attempted we do smaller writes to
transfer some of the data into the kfifo which can then drain to the hardware.
I can sort of see this might be beneficial as it provides batching.
They are somewhat related but it's not totally clear to me they should be the
same parameter. Perhaps we need some more docs to explain how watermark is
used for output buffers?

As it stands there are some corner cases around this that look ominous to me...
See inline.

> ---
> drivers/iio/iio_core.h | 4 +
> drivers/iio/industrialio-buffer.c | 133 +++++++++++++++++++++++++++++-
> drivers/iio/industrialio-core.c | 1 +
> include/linux/iio/buffer.h | 7 ++
> include/linux/iio/buffer_impl.h | 11 +++
> 5 files changed, 154 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/iio/iio_core.h b/drivers/iio/iio_core.h
> index 8f4a9b264962..61e318431de9 100644
> --- a/drivers/iio/iio_core.h
> +++ b/drivers/iio/iio_core.h
> @@ -68,12 +68,15 @@ __poll_t iio_buffer_poll_wrapper(struct file *filp,
> struct poll_table_struct *wait);
> ssize_t iio_buffer_read_wrapper(struct file *filp, char __user *buf,
> size_t n, loff_t *f_ps);
> +ssize_t iio_buffer_write_wrapper(struct file *filp, const char __user *buf,
> + size_t n, loff_t *f_ps);
>
> int iio_buffers_alloc_sysfs_and_mask(struct iio_dev *indio_dev);
> void iio_buffers_free_sysfs_and_mask(struct iio_dev *indio_dev);
>
> #define iio_buffer_poll_addr (&iio_buffer_poll_wrapper)
> #define iio_buffer_read_outer_addr (&iio_buffer_read_wrapper)
> +#define iio_buffer_write_outer_addr (&iio_buffer_write_wrapper)
>
> void iio_disable_all_buffers(struct iio_dev *indio_dev);
> void iio_buffer_wakeup_poll(struct iio_dev *indio_dev);
> @@ -83,6 +86,7 @@ void iio_device_detach_buffers(struct iio_dev *indio_dev);
>
> #define iio_buffer_poll_addr NULL
> #define iio_buffer_read_outer_addr NULL
> +#define iio_buffer_write_outer_addr NULL
>
> static inline int iio_buffers_alloc_sysfs_and_mask(struct iio_dev *indio_dev)
> {
> diff --git a/drivers/iio/industrialio-buffer.c b/drivers/iio/industrialio-buffer.c
> index a95cc2da56be..73d4451a0572 100644
> --- a/drivers/iio/industrialio-buffer.c
> +++ b/drivers/iio/industrialio-buffer.c
> @@ -161,6 +161,69 @@ static ssize_t iio_buffer_read(struct file *filp, char __user *buf,
> return ret;
> }
>
> +static size_t iio_buffer_space_available(struct iio_buffer *buf)
> +{
> + if (buf->access->space_available)
> + return buf->access->space_available(buf);
> +
> + return SIZE_MAX;
> +}
> +
> +static ssize_t iio_buffer_write(struct file *filp, const char __user *buf,
> + size_t n, loff_t *f_ps)
> +{
> + struct iio_dev_buffer_pair *ib = filp->private_data;
> + struct iio_buffer *rb = ib->buffer;
> + struct iio_dev *indio_dev = ib->indio_dev;
> + DEFINE_WAIT_FUNC(wait, woken_wake_function);
> + size_t datum_size;
> + size_t to_wait;
> + int ret;
> +
> + if (!rb || !rb->access->write)
> + return -EINVAL;
> +
> + datum_size = rb->bytes_per_datum;
> +
> + /*
> + * If datum_size is 0 there will never be anything to read from the
> + * buffer, so signal end of file now.
> + */
> + if (!datum_size)
> + return 0;
> +
> + if (filp->f_flags & O_NONBLOCK)
> + to_wait = 0;
> + else
> + to_wait = min_t(size_t, n / datum_size, rb->watermark);

Why is the watermark relevant here? We need enough space for the data
as written whatever the watermark is set to.
Been a while since I looked at equivalent write path, but I think there
we are interested in ensuring a hwfifo flushes out. I'm don't think
the same concept exists in this direction.

> +
> + add_wait_queue(&rb->pollq, &wait);
> + do {
> + if (!indio_dev->info) {
> + ret = -ENODEV;
> + break;
> + }
> +
> + if (iio_buffer_space_available(rb) < to_wait) {
> + if (signal_pending(current)) {
> + ret = -ERESTARTSYS;
> + break;
> + }
> +
> + wait_woken(&wait, TASK_INTERRUPTIBLE,
> + MAX_SCHEDULE_TIMEOUT);
> + continue;
> + }
> +
> + ret = rb->access->write(rb, n, buf);
> + if (ret == 0 && (filp->f_flags & O_NONBLOCK))
> + ret = -EAGAIN;

Do we need to advance the buf pointer here and reduce n? We may have written
some but not all the data.

> + } while (ret == 0);
> + remove_wait_queue(&rb->pollq, &wait);
> +
> + return ret;
> +}
> +
> /**
> * iio_buffer_poll() - poll the buffer to find out if it has data
> * @filp: File structure pointer for device access
> @@ -181,8 +244,18 @@ static __poll_t iio_buffer_poll(struct file *filp,
> return 0;
>
> poll_wait(filp, &rb->pollq, wait);
> - if (iio_buffer_ready(indio_dev, rb, rb->watermark, 0))
> - return EPOLLIN | EPOLLRDNORM;
> +
> + switch (rb->direction) {
> + case IIO_BUFFER_DIRECTION_IN:
> + if (iio_buffer_ready(indio_dev, rb, rb->watermark, 0))
> + return EPOLLIN | EPOLLRDNORM;
> + break;
> + case IIO_BUFFER_DIRECTION_OUT:
> + if (iio_buffer_space_available(rb) >= rb->watermark)

That's interesting. We should probably make sure we update docs to make
it clear that it has a different meaning for output buffers. Guess that
might be later in this set though.

> + return EPOLLOUT | EPOLLWRNORM;
> + break;
> + }
> +
> return 0;
> }
>
> @@ -199,6 +272,19 @@ ssize_t iio_buffer_read_wrapper(struct file *filp, char __user *buf,
> return iio_buffer_read(filp, buf, n, f_ps);
> }
>
> +ssize_t iio_buffer_write_wrapper(struct file *filp, const char __user *buf,
> + size_t n, loff_t *f_ps)
> +{
> + struct iio_dev_buffer_pair *ib = filp->private_data;
> + struct iio_buffer *rb = ib->buffer;
> +
> + /* check if buffer was opened through new API */

This is new. We don't need to support the old API. If we can make sure
it never appears in the old API at all that would be great.

> + if (test_bit(IIO_BUSY_BIT_POS, &rb->flags))
> + return -EBUSY;
> +
> + return iio_buffer_write(filp, buf, n, f_ps);
> +}
> +
> __poll_t iio_buffer_poll_wrapper(struct file *filp,
> struct poll_table_struct *wait)
> {
> @@ -231,6 +317,15 @@ void iio_buffer_wakeup_poll(struct iio_dev *indio_dev)
> }
> }
>
> +int iio_buffer_remove_sample(struct iio_buffer *buffer, u8 *data)

sample or scan? Sample would be a single value for a single channel,
scan would be updates for all channels that are enabled.

> +{
> + if (!buffer || !buffer->access || buffer->access->remove_from)
> + return -EINVAL;
> +
> + return buffer->access->remove_from(buffer, data);
> +}
> +EXPORT_SYMBOL_GPL(iio_buffer_remove_sample);
> +
> void iio_buffer_init(struct iio_buffer *buffer)
> {
> INIT_LIST_HEAD(&buffer->demux_list);
> @@ -807,6 +902,8 @@ static int iio_verify_update(struct iio_dev *indio_dev,
> }
>
> if (insert_buffer) {
> + if (insert_buffer->direction == IIO_BUFFER_DIRECTION_OUT)
> + strict_scanmask = true;

As below, I'm surprised we can get to here..

> bitmap_or(compound_mask, compound_mask,
> insert_buffer->scan_mask, indio_dev->masklength);
> scan_timestamp |= insert_buffer->scan_timestamp;
> @@ -948,6 +1045,8 @@ static int iio_update_demux(struct iio_dev *indio_dev)
> int ret;
>
> list_for_each_entry(buffer, &iio_dev_opaque->buffer_list, buffer_list) {
> + if (buffer->direction == IIO_BUFFER_DIRECTION_OUT)
> + continue;

Given the below, how did it get into the list? I think that check should be
enough that we don't need to check it elsewhere.

> ret = iio_buffer_update_demux(indio_dev, buffer);
> if (ret < 0)
> goto error_clear_mux_table;
> @@ -1159,6 +1258,11 @@ int iio_update_buffers(struct iio_dev *indio_dev,
> mutex_lock(&iio_dev_opaque->info_exist_lock);
> mutex_lock(&indio_dev->mlock);
>
> + if (insert_buffer->direction == IIO_BUFFER_DIRECTION_OUT) {

Can you do this outside of the lock as a sanity check before this function really
gets going?

> + ret = -EINVAL;
> + goto out_unlock;
> + }
> +
> if (insert_buffer && iio_buffer_is_active(insert_buffer))
> insert_buffer = NULL;
>
> @@ -1277,6 +1381,22 @@ static ssize_t iio_dma_show_data_available(struct device *dev,
> return sysfs_emit(buf, "%zu\n", iio_buffer_data_available(buffer));
> }
>
> +static ssize_t direction_show(struct device *dev,
> + struct device_attribute *attr,
> + char *buf)
> +{
> + struct iio_buffer *buffer = to_iio_dev_attr(attr)->buffer;
> +
> + switch (buffer->direction) {
> + case IIO_BUFFER_DIRECTION_IN:
> + return sprintf(buf, "in\n");
> + case IIO_BUFFER_DIRECTION_OUT:
> + return sprintf(buf, "out\n");
> + default:
> + return -EINVAL;
> + }
> +}
> +
> static DEVICE_ATTR(length, S_IRUGO | S_IWUSR, iio_buffer_read_length,
> iio_buffer_write_length);
> static struct device_attribute dev_attr_length_ro = __ATTR(length,
> @@ -1289,12 +1409,20 @@ static struct device_attribute dev_attr_watermark_ro = __ATTR(watermark,
> S_IRUGO, iio_buffer_show_watermark, NULL);
> static DEVICE_ATTR(data_available, S_IRUGO,
> iio_dma_show_data_available, NULL);
> +static DEVICE_ATTR_RO(direction);
>
> +/**
> + * When adding new attributes here, put the at the end, at least until
> + * the code that handles the lengh/length_ro & watermark/watermark_ro
> + * assignments gets cleaned up. Otherwise these can create some weird
> + * duplicate attributes errors under some setups.
> + */
> static struct attribute *iio_buffer_attrs[] = {
> &dev_attr_length.attr,
> &dev_attr_enable.attr,
> &dev_attr_watermark.attr,
> &dev_attr_data_available.attr,
> + &dev_attr_direction.attr,
> };
>
> #define to_dev_attr(_attr) container_of(_attr, struct device_attribute, attr)
> @@ -1397,6 +1525,7 @@ static const struct file_operations iio_buffer_chrdev_fileops = {
> .owner = THIS_MODULE,
> .llseek = noop_llseek,
> .read = iio_buffer_read,
> + .write = iio_buffer_write,
> .poll = iio_buffer_poll,
> .release = iio_buffer_chrdev_release,
> };
> diff --git a/drivers/iio/industrialio-core.c b/drivers/iio/industrialio-core.c
> index 2dbb37e09b8c..537a08549a69 100644
> --- a/drivers/iio/industrialio-core.c
> +++ b/drivers/iio/industrialio-core.c
> @@ -1822,6 +1822,7 @@ static const struct file_operations iio_buffer_fileops = {
> .owner = THIS_MODULE,
> .llseek = noop_llseek,
> .read = iio_buffer_read_outer_addr,
> + .write = iio_buffer_write_outer_addr,
> .poll = iio_buffer_poll_addr,
> .unlocked_ioctl = iio_ioctl,
> .compat_ioctl = compat_ptr_ioctl,
> diff --git a/include/linux/iio/buffer.h b/include/linux/iio/buffer.h
> index b6928ac5c63d..e87b8773253d 100644
> --- a/include/linux/iio/buffer.h
> +++ b/include/linux/iio/buffer.h
> @@ -11,8 +11,15 @@
>
> struct iio_buffer;
>
> +enum iio_buffer_direction {
> + IIO_BUFFER_DIRECTION_IN,
> + IIO_BUFFER_DIRECTION_OUT,
> +};
> +
> int iio_push_to_buffers(struct iio_dev *indio_dev, const void *data);
>
> +int iio_buffer_remove_sample(struct iio_buffer *buffer, u8 *data);
> +
> /**
> * iio_push_to_buffers_with_timestamp() - push data and timestamp to buffers
> * @indio_dev: iio_dev structure for device.
> diff --git a/include/linux/iio/buffer_impl.h b/include/linux/iio/buffer_impl.h
> index 245b32918ae1..8a44c5321e19 100644
> --- a/include/linux/iio/buffer_impl.h
> +++ b/include/linux/iio/buffer_impl.h
> @@ -7,6 +7,7 @@
> #ifdef CONFIG_IIO_BUFFER
>
> #include <uapi/linux/iio/buffer.h>
> +#include <linux/iio/buffer.h>

>
> struct iio_dev;
> struct iio_buffer;
> @@ -23,6 +24,10 @@ struct iio_buffer;
> * @read: try to get a specified number of bytes (must exist)
> * @data_available: indicates how much data is available for reading from
> * the buffer.
> + * @remove_from: remove sample from buffer. Drivers should calls this to
> + * remove a sample from a buffer.
> + * @write: try to write a number of bytes
> + * @space_available: returns the amount of bytes available in a buffer
> * @request_update: if a parameter change has been marked, update underlying
> * storage.
> * @set_bytes_per_datum:set number of bytes per datum
> @@ -49,6 +54,9 @@ struct iio_buffer_access_funcs {
> int (*store_to)(struct iio_buffer *buffer, const void *data);
> int (*read)(struct iio_buffer *buffer, size_t n, char __user *buf);
> size_t (*data_available)(struct iio_buffer *buffer);
> + int (*remove_from)(struct iio_buffer *buffer, void *data);
> + int (*write)(struct iio_buffer *buffer, size_t n, const char __user *buf);
> + size_t (*space_available)(struct iio_buffer *buffer);
>
> int (*request_update)(struct iio_buffer *buffer);
>
> @@ -80,6 +88,9 @@ struct iio_buffer {
> /** @bytes_per_datum: Size of individual datum including timestamp. */
> size_t bytes_per_datum;
>
> + /* @direction: Direction of the data stream (in/out). */
> + enum iio_buffer_direction direction;
> +
> /**
> * @access: Buffer access functions associated with the
> * implementation.