linux/drivers/iio/buffer/industrialio-buffer-dma.c

685 lines
21 KiB
C
Raw Normal View History

iio: Add generic DMA buffer infrastructure The traditional approach used in IIO to implement buffered capture requires the generation of at least one interrupt per sample. In the interrupt handler the driver reads the sample from the device and copies it to a software buffer. This approach has a rather large per sample overhead associated with it. And while it works fine for samplerates in the range of up to 1000 samples per second it starts to consume a rather large share of the available CPU processing time once we go beyond that, this is especially true on an embedded system with limited processing power. The regular interrupt also causes increased power consumption by not allowing the hardware into deeper sleep states, which is something that becomes more and more important on mobile battery powered devices. And while the recently added watermark support mitigates some of the issues by allowing the device to generate interrupts at a rate lower than the data output rate, this still requires a storage buffer inside the device and even if it exists it is only a few 100 samples deep at most. DMA support on the other hand allows to capture multiple millions or even more samples without any CPU interaction. This allows the CPU to either go to sleep for longer periods or focus on other tasks which increases overall system performance and power consumption. In addition to that some devices might not even offer a way to read the data other than using DMA, which makes DMA mandatory to use for them. The tasks involved in implementing a DMA buffer can be divided into two categories. The first category is memory buffer management (allocation, mapping, etc.) and hooking this up the IIO buffer callbacks like read(), enable(), disable(), etc. The second category of tasks is to setup the DMA hardware and manage the DMA transfers. Tasks from the first category will be very similar for all IIO drivers supporting DMA buffers, while the tasks from the second category will be hardware specific. This patch implements a generic infrastructure that take care of the former tasks. It provides a set of functions that implement the standard IIO buffer iio_buffer_access_funcs callbacks. These can either be used as is or be overloaded and augmented with driver specific code where necessary. For the DMA buffer support infrastructure that is introduced in this series sample data is grouped by so called blocks. A block is the basic unit at which data is exchanged between the application and the hardware. The application is responsible for allocating the memory associated with the block and then passes the block to the hardware. When the hardware has captured the amount of samples equal to size of a block it will notify the application, which can then read the data from the block and process it. The block size can freely chosen (within the constraints of the hardware). This allows to make a trade-off between latency and management overhead. The larger the block size the lower the per sample overhead but the latency between when the data was captured and when the application will be able to access it increases, in a similar way smaller block sizes have a larger per sample management overhead but a lower latency. The ideal block size thus depends on system and application requirements. For the time being the infrastructure only implements a simple double buffered scheme which allocates two blocks each with half the size of the configured buffer size. This provides basic support for capturing continuous uninterrupted data over the existing file-IO ABI. Future extensions to the DMA buffer infrastructure will give applications a more fine grained control over how many blocks are allocated and the size of each block. But this requires userspace ABI additions which are intentionally not part of this patch and will be added separately. Tasks of the second category need to be implemented by a device specific driver. They can be hooked up into the generic infrastructure using two simple callbacks, submit() and abort(). The submit() callback is used to schedule DMA transfers for blocks. Once a DMA transfer has been completed it is expected that the buffer driver calls iio_dma_buffer_block_done() to notify. The abort() callback is used for stopping all pending and active DMA transfers when the buffer is disabled. Signed-off-by: Lars-Peter Clausen <lars@metafoo.de> Signed-off-by: Jonathan Cameron <jic23@kernel.org>
2015-10-14 00:10:28 +08:00
/*
* Copyright 2013-2015 Analog Devices Inc.
* Author: Lars-Peter Clausen <lars@metafoo.de>
*
* Licensed under the GPL-2.
*/
#include <linux/slab.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/device.h>
#include <linux/workqueue.h>
#include <linux/mutex.h>
#include <linux/sched.h>
#include <linux/poll.h>
#include <linux/iio/buffer.h>
#include <linux/iio/buffer_impl.h>
iio: Add generic DMA buffer infrastructure The traditional approach used in IIO to implement buffered capture requires the generation of at least one interrupt per sample. In the interrupt handler the driver reads the sample from the device and copies it to a software buffer. This approach has a rather large per sample overhead associated with it. And while it works fine for samplerates in the range of up to 1000 samples per second it starts to consume a rather large share of the available CPU processing time once we go beyond that, this is especially true on an embedded system with limited processing power. The regular interrupt also causes increased power consumption by not allowing the hardware into deeper sleep states, which is something that becomes more and more important on mobile battery powered devices. And while the recently added watermark support mitigates some of the issues by allowing the device to generate interrupts at a rate lower than the data output rate, this still requires a storage buffer inside the device and even if it exists it is only a few 100 samples deep at most. DMA support on the other hand allows to capture multiple millions or even more samples without any CPU interaction. This allows the CPU to either go to sleep for longer periods or focus on other tasks which increases overall system performance and power consumption. In addition to that some devices might not even offer a way to read the data other than using DMA, which makes DMA mandatory to use for them. The tasks involved in implementing a DMA buffer can be divided into two categories. The first category is memory buffer management (allocation, mapping, etc.) and hooking this up the IIO buffer callbacks like read(), enable(), disable(), etc. The second category of tasks is to setup the DMA hardware and manage the DMA transfers. Tasks from the first category will be very similar for all IIO drivers supporting DMA buffers, while the tasks from the second category will be hardware specific. This patch implements a generic infrastructure that take care of the former tasks. It provides a set of functions that implement the standard IIO buffer iio_buffer_access_funcs callbacks. These can either be used as is or be overloaded and augmented with driver specific code where necessary. For the DMA buffer support infrastructure that is introduced in this series sample data is grouped by so called blocks. A block is the basic unit at which data is exchanged between the application and the hardware. The application is responsible for allocating the memory associated with the block and then passes the block to the hardware. When the hardware has captured the amount of samples equal to size of a block it will notify the application, which can then read the data from the block and process it. The block size can freely chosen (within the constraints of the hardware). This allows to make a trade-off between latency and management overhead. The larger the block size the lower the per sample overhead but the latency between when the data was captured and when the application will be able to access it increases, in a similar way smaller block sizes have a larger per sample management overhead but a lower latency. The ideal block size thus depends on system and application requirements. For the time being the infrastructure only implements a simple double buffered scheme which allocates two blocks each with half the size of the configured buffer size. This provides basic support for capturing continuous uninterrupted data over the existing file-IO ABI. Future extensions to the DMA buffer infrastructure will give applications a more fine grained control over how many blocks are allocated and the size of each block. But this requires userspace ABI additions which are intentionally not part of this patch and will be added separately. Tasks of the second category need to be implemented by a device specific driver. They can be hooked up into the generic infrastructure using two simple callbacks, submit() and abort(). The submit() callback is used to schedule DMA transfers for blocks. Once a DMA transfer has been completed it is expected that the buffer driver calls iio_dma_buffer_block_done() to notify. The abort() callback is used for stopping all pending and active DMA transfers when the buffer is disabled. Signed-off-by: Lars-Peter Clausen <lars@metafoo.de> Signed-off-by: Jonathan Cameron <jic23@kernel.org>
2015-10-14 00:10:28 +08:00
#include <linux/iio/buffer-dma.h>
#include <linux/dma-mapping.h>
#include <linux/sizes.h>
/*
* For DMA buffers the storage is sub-divided into so called blocks. Each block
* has its own memory buffer. The size of the block is the granularity at which
* memory is exchanged between the hardware and the application. Increasing the
* basic unit of data exchange from one sample to one block decreases the
* management overhead that is associated with each sample. E.g. if we say the
* management overhead for one exchange is x and the unit of exchange is one
* sample the overhead will be x for each sample. Whereas when using a block
* which contains n samples the overhead per sample is reduced to x/n. This
* allows to achieve much higher samplerates than what can be sustained with
* the one sample approach.
*
* Blocks are exchanged between the DMA controller and the application via the
* means of two queues. The incoming queue and the outgoing queue. Blocks on the
* incoming queue are waiting for the DMA controller to pick them up and fill
* them with data. Block on the outgoing queue have been filled with data and
* are waiting for the application to dequeue them and read the data.
*
* A block can be in one of the following states:
* * Owned by the application. In this state the application can read data from
* the block.
* * On the incoming list: Blocks on the incoming list are queued up to be
* processed by the DMA controller.
* * Owned by the DMA controller: The DMA controller is processing the block
* and filling it with data.
* * On the outgoing list: Blocks on the outgoing list have been successfully
* processed by the DMA controller and contain data. They can be dequeued by
* the application.
* * Dead: A block that is dead has been marked as to be freed. It might still
* be owned by either the application or the DMA controller at the moment.
* But once they are done processing it instead of going to either the
* incoming or outgoing queue the block will be freed.
*
* In addition to this blocks are reference counted and the memory associated
* with both the block structure as well as the storage memory for the block
* will be freed when the last reference to the block is dropped. This means a
* block must not be accessed without holding a reference.
*
* The iio_dma_buffer implementation provides a generic infrastructure for
* managing the blocks.
*
* A driver for a specific piece of hardware that has DMA capabilities need to
* implement the submit() callback from the iio_dma_buffer_ops structure. This
* callback is supposed to initiate the DMA transfer copying data from the
* converter to the memory region of the block. Once the DMA transfer has been
* completed the driver must call iio_dma_buffer_block_done() for the completed
* block.
*
* Prior to this it must set the bytes_used field of the block contains
* the actual number of bytes in the buffer. Typically this will be equal to the
* size of the block, but if the DMA hardware has certain alignment requirements
* for the transfer length it might choose to use less than the full size. In
* either case it is expected that bytes_used is a multiple of the bytes per
* datum, i.e. the block must not contain partial samples.
*
* The driver must call iio_dma_buffer_block_done() for each block it has
* received through its submit_block() callback, even if it does not actually
* perform a DMA transfer for the block, e.g. because the buffer was disabled
* before the block transfer was started. In this case it should set bytes_used
* to 0.
*
* In addition it is recommended that a driver implements the abort() callback.
* It will be called when the buffer is disabled and can be used to cancel
* pending and stop active transfers.
*
* The specific driver implementation should use the default callback
* implementations provided by this module for the iio_buffer_access_funcs
* struct. It may overload some callbacks with custom variants if the hardware
* has special requirements that are not handled by the generic functions. If a
* driver chooses to overload a callback it has to ensure that the generic
* callback is called from within the custom callback.
*/
static void iio_buffer_block_release(struct kref *kref)
{
struct iio_dma_buffer_block *block = container_of(kref,
struct iio_dma_buffer_block, kref);
WARN_ON(block->state != IIO_BLOCK_STATE_DEAD);
dma_free_coherent(block->queue->dev, PAGE_ALIGN(block->size),
block->vaddr, block->phys_addr);
iio_buffer_put(&block->queue->buffer);
kfree(block);
}
static void iio_buffer_block_get(struct iio_dma_buffer_block *block)
{
kref_get(&block->kref);
}
static void iio_buffer_block_put(struct iio_dma_buffer_block *block)
{
kref_put(&block->kref, iio_buffer_block_release);
}
/*
* dma_free_coherent can sleep, hence we need to take some special care to be
* able to drop a reference from an atomic context.
*/
static LIST_HEAD(iio_dma_buffer_dead_blocks);
static DEFINE_SPINLOCK(iio_dma_buffer_dead_blocks_lock);
static void iio_dma_buffer_cleanup_worker(struct work_struct *work)
{
struct iio_dma_buffer_block *block, *_block;
LIST_HEAD(block_list);
spin_lock_irq(&iio_dma_buffer_dead_blocks_lock);
list_splice_tail_init(&iio_dma_buffer_dead_blocks, &block_list);
spin_unlock_irq(&iio_dma_buffer_dead_blocks_lock);
list_for_each_entry_safe(block, _block, &block_list, head)
iio_buffer_block_release(&block->kref);
}
static DECLARE_WORK(iio_dma_buffer_cleanup_work, iio_dma_buffer_cleanup_worker);
static void iio_buffer_block_release_atomic(struct kref *kref)
{
struct iio_dma_buffer_block *block;
unsigned long flags;
block = container_of(kref, struct iio_dma_buffer_block, kref);
spin_lock_irqsave(&iio_dma_buffer_dead_blocks_lock, flags);
list_add_tail(&block->head, &iio_dma_buffer_dead_blocks);
spin_unlock_irqrestore(&iio_dma_buffer_dead_blocks_lock, flags);
schedule_work(&iio_dma_buffer_cleanup_work);
}
/*
* Version of iio_buffer_block_put() that can be called from atomic context
*/
static void iio_buffer_block_put_atomic(struct iio_dma_buffer_block *block)
{
kref_put(&block->kref, iio_buffer_block_release_atomic);
}
static struct iio_dma_buffer_queue *iio_buffer_to_queue(struct iio_buffer *buf)
{
return container_of(buf, struct iio_dma_buffer_queue, buffer);
}
static struct iio_dma_buffer_block *iio_dma_buffer_alloc_block(
struct iio_dma_buffer_queue *queue, size_t size)
{
struct iio_dma_buffer_block *block;
block = kzalloc(sizeof(*block), GFP_KERNEL);
if (!block)
return NULL;
block->vaddr = dma_alloc_coherent(queue->dev, PAGE_ALIGN(size),
&block->phys_addr, GFP_KERNEL);
if (!block->vaddr) {
kfree(block);
return NULL;
}
block->size = size;
block->state = IIO_BLOCK_STATE_DEQUEUED;
block->queue = queue;
INIT_LIST_HEAD(&block->head);
kref_init(&block->kref);
iio_buffer_get(&queue->buffer);
return block;
}
static void _iio_dma_buffer_block_done(struct iio_dma_buffer_block *block)
{
struct iio_dma_buffer_queue *queue = block->queue;
/*
* The buffer has already been freed by the application, just drop the
* reference.
*/
if (block->state != IIO_BLOCK_STATE_DEAD) {
block->state = IIO_BLOCK_STATE_DONE;
list_add_tail(&block->head, &queue->outgoing);
}
}
/**
* iio_dma_buffer_block_done() - Indicate that a block has been completed
* @block: The completed block
*
* Should be called when the DMA controller has finished handling the block to
* pass back ownership of the block to the queue.
*/
void iio_dma_buffer_block_done(struct iio_dma_buffer_block *block)
{
struct iio_dma_buffer_queue *queue = block->queue;
unsigned long flags;
spin_lock_irqsave(&queue->list_lock, flags);
_iio_dma_buffer_block_done(block);
spin_unlock_irqrestore(&queue->list_lock, flags);
iio_buffer_block_put_atomic(block);
wake_up_interruptible_poll(&queue->buffer.pollq, EPOLLIN | EPOLLRDNORM);
iio: Add generic DMA buffer infrastructure The traditional approach used in IIO to implement buffered capture requires the generation of at least one interrupt per sample. In the interrupt handler the driver reads the sample from the device and copies it to a software buffer. This approach has a rather large per sample overhead associated with it. And while it works fine for samplerates in the range of up to 1000 samples per second it starts to consume a rather large share of the available CPU processing time once we go beyond that, this is especially true on an embedded system with limited processing power. The regular interrupt also causes increased power consumption by not allowing the hardware into deeper sleep states, which is something that becomes more and more important on mobile battery powered devices. And while the recently added watermark support mitigates some of the issues by allowing the device to generate interrupts at a rate lower than the data output rate, this still requires a storage buffer inside the device and even if it exists it is only a few 100 samples deep at most. DMA support on the other hand allows to capture multiple millions or even more samples without any CPU interaction. This allows the CPU to either go to sleep for longer periods or focus on other tasks which increases overall system performance and power consumption. In addition to that some devices might not even offer a way to read the data other than using DMA, which makes DMA mandatory to use for them. The tasks involved in implementing a DMA buffer can be divided into two categories. The first category is memory buffer management (allocation, mapping, etc.) and hooking this up the IIO buffer callbacks like read(), enable(), disable(), etc. The second category of tasks is to setup the DMA hardware and manage the DMA transfers. Tasks from the first category will be very similar for all IIO drivers supporting DMA buffers, while the tasks from the second category will be hardware specific. This patch implements a generic infrastructure that take care of the former tasks. It provides a set of functions that implement the standard IIO buffer iio_buffer_access_funcs callbacks. These can either be used as is or be overloaded and augmented with driver specific code where necessary. For the DMA buffer support infrastructure that is introduced in this series sample data is grouped by so called blocks. A block is the basic unit at which data is exchanged between the application and the hardware. The application is responsible for allocating the memory associated with the block and then passes the block to the hardware. When the hardware has captured the amount of samples equal to size of a block it will notify the application, which can then read the data from the block and process it. The block size can freely chosen (within the constraints of the hardware). This allows to make a trade-off between latency and management overhead. The larger the block size the lower the per sample overhead but the latency between when the data was captured and when the application will be able to access it increases, in a similar way smaller block sizes have a larger per sample management overhead but a lower latency. The ideal block size thus depends on system and application requirements. For the time being the infrastructure only implements a simple double buffered scheme which allocates two blocks each with half the size of the configured buffer size. This provides basic support for capturing continuous uninterrupted data over the existing file-IO ABI. Future extensions to the DMA buffer infrastructure will give applications a more fine grained control over how many blocks are allocated and the size of each block. But this requires userspace ABI additions which are intentionally not part of this patch and will be added separately. Tasks of the second category need to be implemented by a device specific driver. They can be hooked up into the generic infrastructure using two simple callbacks, submit() and abort(). The submit() callback is used to schedule DMA transfers for blocks. Once a DMA transfer has been completed it is expected that the buffer driver calls iio_dma_buffer_block_done() to notify. The abort() callback is used for stopping all pending and active DMA transfers when the buffer is disabled. Signed-off-by: Lars-Peter Clausen <lars@metafoo.de> Signed-off-by: Jonathan Cameron <jic23@kernel.org>
2015-10-14 00:10:28 +08:00
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_block_done);
/**
* iio_dma_buffer_block_list_abort() - Indicate that a list block has been
* aborted
* @queue: Queue for which to complete blocks.
* @list: List of aborted blocks. All blocks in this list must be from @queue.
*
* Typically called from the abort() callback after the DMA controller has been
* stopped. This will set bytes_used to 0 for each block in the list and then
* hand the blocks back to the queue.
*/
void iio_dma_buffer_block_list_abort(struct iio_dma_buffer_queue *queue,
struct list_head *list)
{
struct iio_dma_buffer_block *block, *_block;
unsigned long flags;
spin_lock_irqsave(&queue->list_lock, flags);
list_for_each_entry_safe(block, _block, list, head) {
list_del(&block->head);
block->bytes_used = 0;
_iio_dma_buffer_block_done(block);
iio_buffer_block_put_atomic(block);
}
spin_unlock_irqrestore(&queue->list_lock, flags);
wake_up_interruptible_poll(&queue->buffer.pollq, EPOLLIN | EPOLLRDNORM);
iio: Add generic DMA buffer infrastructure The traditional approach used in IIO to implement buffered capture requires the generation of at least one interrupt per sample. In the interrupt handler the driver reads the sample from the device and copies it to a software buffer. This approach has a rather large per sample overhead associated with it. And while it works fine for samplerates in the range of up to 1000 samples per second it starts to consume a rather large share of the available CPU processing time once we go beyond that, this is especially true on an embedded system with limited processing power. The regular interrupt also causes increased power consumption by not allowing the hardware into deeper sleep states, which is something that becomes more and more important on mobile battery powered devices. And while the recently added watermark support mitigates some of the issues by allowing the device to generate interrupts at a rate lower than the data output rate, this still requires a storage buffer inside the device and even if it exists it is only a few 100 samples deep at most. DMA support on the other hand allows to capture multiple millions or even more samples without any CPU interaction. This allows the CPU to either go to sleep for longer periods or focus on other tasks which increases overall system performance and power consumption. In addition to that some devices might not even offer a way to read the data other than using DMA, which makes DMA mandatory to use for them. The tasks involved in implementing a DMA buffer can be divided into two categories. The first category is memory buffer management (allocation, mapping, etc.) and hooking this up the IIO buffer callbacks like read(), enable(), disable(), etc. The second category of tasks is to setup the DMA hardware and manage the DMA transfers. Tasks from the first category will be very similar for all IIO drivers supporting DMA buffers, while the tasks from the second category will be hardware specific. This patch implements a generic infrastructure that take care of the former tasks. It provides a set of functions that implement the standard IIO buffer iio_buffer_access_funcs callbacks. These can either be used as is or be overloaded and augmented with driver specific code where necessary. For the DMA buffer support infrastructure that is introduced in this series sample data is grouped by so called blocks. A block is the basic unit at which data is exchanged between the application and the hardware. The application is responsible for allocating the memory associated with the block and then passes the block to the hardware. When the hardware has captured the amount of samples equal to size of a block it will notify the application, which can then read the data from the block and process it. The block size can freely chosen (within the constraints of the hardware). This allows to make a trade-off between latency and management overhead. The larger the block size the lower the per sample overhead but the latency between when the data was captured and when the application will be able to access it increases, in a similar way smaller block sizes have a larger per sample management overhead but a lower latency. The ideal block size thus depends on system and application requirements. For the time being the infrastructure only implements a simple double buffered scheme which allocates two blocks each with half the size of the configured buffer size. This provides basic support for capturing continuous uninterrupted data over the existing file-IO ABI. Future extensions to the DMA buffer infrastructure will give applications a more fine grained control over how many blocks are allocated and the size of each block. But this requires userspace ABI additions which are intentionally not part of this patch and will be added separately. Tasks of the second category need to be implemented by a device specific driver. They can be hooked up into the generic infrastructure using two simple callbacks, submit() and abort(). The submit() callback is used to schedule DMA transfers for blocks. Once a DMA transfer has been completed it is expected that the buffer driver calls iio_dma_buffer_block_done() to notify. The abort() callback is used for stopping all pending and active DMA transfers when the buffer is disabled. Signed-off-by: Lars-Peter Clausen <lars@metafoo.de> Signed-off-by: Jonathan Cameron <jic23@kernel.org>
2015-10-14 00:10:28 +08:00
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_block_list_abort);
static bool iio_dma_block_reusable(struct iio_dma_buffer_block *block)
{
/*
* If the core owns the block it can be re-used. This should be the
* default case when enabling the buffer, unless the DMA controller does
* not support abort and has not given back the block yet.
*/
switch (block->state) {
case IIO_BLOCK_STATE_DEQUEUED:
case IIO_BLOCK_STATE_QUEUED:
case IIO_BLOCK_STATE_DONE:
return true;
default:
return false;
}
}
/**
* iio_dma_buffer_request_update() - DMA buffer request_update callback
* @buffer: The buffer which to request an update
*
* Should be used as the iio_dma_buffer_request_update() callback for
* iio_buffer_access_ops struct for DMA buffers.
*/
int iio_dma_buffer_request_update(struct iio_buffer *buffer)
{
struct iio_dma_buffer_queue *queue = iio_buffer_to_queue(buffer);
struct iio_dma_buffer_block *block;
bool try_reuse = false;
size_t size;
int ret = 0;
int i;
/*
* Split the buffer into two even parts. This is used as a double
* buffering scheme with usually one block at a time being used by the
* DMA and the other one by the application.
*/
size = DIV_ROUND_UP(queue->buffer.bytes_per_datum *
queue->buffer.length, 2);
mutex_lock(&queue->lock);
/* Allocations are page aligned */
if (PAGE_ALIGN(queue->fileio.block_size) == PAGE_ALIGN(size))
try_reuse = true;
queue->fileio.block_size = size;
queue->fileio.active_block = NULL;
spin_lock_irq(&queue->list_lock);
for (i = 0; i < ARRAY_SIZE(queue->fileio.blocks); i++) {
iio: Add generic DMA buffer infrastructure The traditional approach used in IIO to implement buffered capture requires the generation of at least one interrupt per sample. In the interrupt handler the driver reads the sample from the device and copies it to a software buffer. This approach has a rather large per sample overhead associated with it. And while it works fine for samplerates in the range of up to 1000 samples per second it starts to consume a rather large share of the available CPU processing time once we go beyond that, this is especially true on an embedded system with limited processing power. The regular interrupt also causes increased power consumption by not allowing the hardware into deeper sleep states, which is something that becomes more and more important on mobile battery powered devices. And while the recently added watermark support mitigates some of the issues by allowing the device to generate interrupts at a rate lower than the data output rate, this still requires a storage buffer inside the device and even if it exists it is only a few 100 samples deep at most. DMA support on the other hand allows to capture multiple millions or even more samples without any CPU interaction. This allows the CPU to either go to sleep for longer periods or focus on other tasks which increases overall system performance and power consumption. In addition to that some devices might not even offer a way to read the data other than using DMA, which makes DMA mandatory to use for them. The tasks involved in implementing a DMA buffer can be divided into two categories. The first category is memory buffer management (allocation, mapping, etc.) and hooking this up the IIO buffer callbacks like read(), enable(), disable(), etc. The second category of tasks is to setup the DMA hardware and manage the DMA transfers. Tasks from the first category will be very similar for all IIO drivers supporting DMA buffers, while the tasks from the second category will be hardware specific. This patch implements a generic infrastructure that take care of the former tasks. It provides a set of functions that implement the standard IIO buffer iio_buffer_access_funcs callbacks. These can either be used as is or be overloaded and augmented with driver specific code where necessary. For the DMA buffer support infrastructure that is introduced in this series sample data is grouped by so called blocks. A block is the basic unit at which data is exchanged between the application and the hardware. The application is responsible for allocating the memory associated with the block and then passes the block to the hardware. When the hardware has captured the amount of samples equal to size of a block it will notify the application, which can then read the data from the block and process it. The block size can freely chosen (within the constraints of the hardware). This allows to make a trade-off between latency and management overhead. The larger the block size the lower the per sample overhead but the latency between when the data was captured and when the application will be able to access it increases, in a similar way smaller block sizes have a larger per sample management overhead but a lower latency. The ideal block size thus depends on system and application requirements. For the time being the infrastructure only implements a simple double buffered scheme which allocates two blocks each with half the size of the configured buffer size. This provides basic support for capturing continuous uninterrupted data over the existing file-IO ABI. Future extensions to the DMA buffer infrastructure will give applications a more fine grained control over how many blocks are allocated and the size of each block. But this requires userspace ABI additions which are intentionally not part of this patch and will be added separately. Tasks of the second category need to be implemented by a device specific driver. They can be hooked up into the generic infrastructure using two simple callbacks, submit() and abort(). The submit() callback is used to schedule DMA transfers for blocks. Once a DMA transfer has been completed it is expected that the buffer driver calls iio_dma_buffer_block_done() to notify. The abort() callback is used for stopping all pending and active DMA transfers when the buffer is disabled. Signed-off-by: Lars-Peter Clausen <lars@metafoo.de> Signed-off-by: Jonathan Cameron <jic23@kernel.org>
2015-10-14 00:10:28 +08:00
block = queue->fileio.blocks[i];
/* If we can't re-use it free it */
if (block && (!iio_dma_block_reusable(block) || !try_reuse))
block->state = IIO_BLOCK_STATE_DEAD;
}
/*
* At this point all blocks are either owned by the core or marked as
* dead. This means we can reset the lists without having to fear
* corrution.
*/
INIT_LIST_HEAD(&queue->outgoing);
spin_unlock_irq(&queue->list_lock);
INIT_LIST_HEAD(&queue->incoming);
for (i = 0; i < ARRAY_SIZE(queue->fileio.blocks); i++) {
iio: Add generic DMA buffer infrastructure The traditional approach used in IIO to implement buffered capture requires the generation of at least one interrupt per sample. In the interrupt handler the driver reads the sample from the device and copies it to a software buffer. This approach has a rather large per sample overhead associated with it. And while it works fine for samplerates in the range of up to 1000 samples per second it starts to consume a rather large share of the available CPU processing time once we go beyond that, this is especially true on an embedded system with limited processing power. The regular interrupt also causes increased power consumption by not allowing the hardware into deeper sleep states, which is something that becomes more and more important on mobile battery powered devices. And while the recently added watermark support mitigates some of the issues by allowing the device to generate interrupts at a rate lower than the data output rate, this still requires a storage buffer inside the device and even if it exists it is only a few 100 samples deep at most. DMA support on the other hand allows to capture multiple millions or even more samples without any CPU interaction. This allows the CPU to either go to sleep for longer periods or focus on other tasks which increases overall system performance and power consumption. In addition to that some devices might not even offer a way to read the data other than using DMA, which makes DMA mandatory to use for them. The tasks involved in implementing a DMA buffer can be divided into two categories. The first category is memory buffer management (allocation, mapping, etc.) and hooking this up the IIO buffer callbacks like read(), enable(), disable(), etc. The second category of tasks is to setup the DMA hardware and manage the DMA transfers. Tasks from the first category will be very similar for all IIO drivers supporting DMA buffers, while the tasks from the second category will be hardware specific. This patch implements a generic infrastructure that take care of the former tasks. It provides a set of functions that implement the standard IIO buffer iio_buffer_access_funcs callbacks. These can either be used as is or be overloaded and augmented with driver specific code where necessary. For the DMA buffer support infrastructure that is introduced in this series sample data is grouped by so called blocks. A block is the basic unit at which data is exchanged between the application and the hardware. The application is responsible for allocating the memory associated with the block and then passes the block to the hardware. When the hardware has captured the amount of samples equal to size of a block it will notify the application, which can then read the data from the block and process it. The block size can freely chosen (within the constraints of the hardware). This allows to make a trade-off between latency and management overhead. The larger the block size the lower the per sample overhead but the latency between when the data was captured and when the application will be able to access it increases, in a similar way smaller block sizes have a larger per sample management overhead but a lower latency. The ideal block size thus depends on system and application requirements. For the time being the infrastructure only implements a simple double buffered scheme which allocates two blocks each with half the size of the configured buffer size. This provides basic support for capturing continuous uninterrupted data over the existing file-IO ABI. Future extensions to the DMA buffer infrastructure will give applications a more fine grained control over how many blocks are allocated and the size of each block. But this requires userspace ABI additions which are intentionally not part of this patch and will be added separately. Tasks of the second category need to be implemented by a device specific driver. They can be hooked up into the generic infrastructure using two simple callbacks, submit() and abort(). The submit() callback is used to schedule DMA transfers for blocks. Once a DMA transfer has been completed it is expected that the buffer driver calls iio_dma_buffer_block_done() to notify. The abort() callback is used for stopping all pending and active DMA transfers when the buffer is disabled. Signed-off-by: Lars-Peter Clausen <lars@metafoo.de> Signed-off-by: Jonathan Cameron <jic23@kernel.org>
2015-10-14 00:10:28 +08:00
if (queue->fileio.blocks[i]) {
block = queue->fileio.blocks[i];
if (block->state == IIO_BLOCK_STATE_DEAD) {
/* Could not reuse it */
iio_buffer_block_put(block);
block = NULL;
} else {
block->size = size;
}
} else {
block = NULL;
}
if (!block) {
block = iio_dma_buffer_alloc_block(queue, size);
if (!block) {
ret = -ENOMEM;
goto out_unlock;
}
queue->fileio.blocks[i] = block;
}
block->state = IIO_BLOCK_STATE_QUEUED;
list_add_tail(&block->head, &queue->incoming);
}
out_unlock:
mutex_unlock(&queue->lock);
return ret;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_request_update);
static void iio_dma_buffer_submit_block(struct iio_dma_buffer_queue *queue,
struct iio_dma_buffer_block *block)
{
int ret;
/*
* If the hardware has already been removed we put the block into
* limbo. It will neither be on the incoming nor outgoing list, nor will
* it ever complete. It will just wait to be freed eventually.
*/
if (!queue->ops)
return;
block->state = IIO_BLOCK_STATE_ACTIVE;
iio_buffer_block_get(block);
ret = queue->ops->submit(queue, block);
if (ret) {
/*
* This is a bit of a problem and there is not much we can do
* other then wait for the buffer to be disabled and re-enabled
* and try again. But it should not really happen unless we run
* out of memory or something similar.
*
* TODO: Implement support in the IIO core to allow buffers to
* notify consumers that something went wrong and the buffer
* should be disabled.
*/
iio_buffer_block_put(block);
}
}
/**
* iio_dma_buffer_enable() - Enable DMA buffer
* @buffer: IIO buffer to enable
* @indio_dev: IIO device the buffer is attached to
*
* Needs to be called when the device that the buffer is attached to starts
* sampling. Typically should be the iio_buffer_access_ops enable callback.
*
* This will allocate the DMA buffers and start the DMA transfers.
*/
int iio_dma_buffer_enable(struct iio_buffer *buffer,
struct iio_dev *indio_dev)
{
struct iio_dma_buffer_queue *queue = iio_buffer_to_queue(buffer);
struct iio_dma_buffer_block *block, *_block;
mutex_lock(&queue->lock);
queue->active = true;
list_for_each_entry_safe(block, _block, &queue->incoming, head) {
list_del(&block->head);
iio_dma_buffer_submit_block(queue, block);
}
mutex_unlock(&queue->lock);
return 0;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_enable);
/**
* iio_dma_buffer_disable() - Disable DMA buffer
* @buffer: IIO DMA buffer to disable
* @indio_dev: IIO device the buffer is attached to
*
* Needs to be called when the device that the buffer is attached to stops
* sampling. Typically should be the iio_buffer_access_ops disable callback.
*/
int iio_dma_buffer_disable(struct iio_buffer *buffer,
struct iio_dev *indio_dev)
{
struct iio_dma_buffer_queue *queue = iio_buffer_to_queue(buffer);
mutex_lock(&queue->lock);
queue->active = false;
if (queue->ops && queue->ops->abort)
queue->ops->abort(queue);
mutex_unlock(&queue->lock);
return 0;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_disable);
static void iio_dma_buffer_enqueue(struct iio_dma_buffer_queue *queue,
struct iio_dma_buffer_block *block)
{
if (block->state == IIO_BLOCK_STATE_DEAD) {
iio_buffer_block_put(block);
} else if (queue->active) {
iio_dma_buffer_submit_block(queue, block);
} else {
block->state = IIO_BLOCK_STATE_QUEUED;
list_add_tail(&block->head, &queue->incoming);
}
}
static struct iio_dma_buffer_block *iio_dma_buffer_dequeue(
struct iio_dma_buffer_queue *queue)
{
struct iio_dma_buffer_block *block;
spin_lock_irq(&queue->list_lock);
block = list_first_entry_or_null(&queue->outgoing, struct
iio_dma_buffer_block, head);
if (block != NULL) {
list_del(&block->head);
block->state = IIO_BLOCK_STATE_DEQUEUED;
}
spin_unlock_irq(&queue->list_lock);
return block;
}
/**
* iio_dma_buffer_read() - DMA buffer read callback
* @buffer: Buffer to read form
* @n: Number of bytes to read
* @user_buffer: Userspace buffer to copy the data to
*
* Should be used as the read_first_n callback for iio_buffer_access_ops
* struct for DMA buffers.
*/
int iio_dma_buffer_read(struct iio_buffer *buffer, size_t n,
char __user *user_buffer)
{
struct iio_dma_buffer_queue *queue = iio_buffer_to_queue(buffer);
struct iio_dma_buffer_block *block;
int ret;
if (n < buffer->bytes_per_datum)
return -EINVAL;
mutex_lock(&queue->lock);
if (!queue->fileio.active_block) {
block = iio_dma_buffer_dequeue(queue);
if (block == NULL) {
ret = 0;
goto out_unlock;
}
queue->fileio.pos = 0;
queue->fileio.active_block = block;
} else {
block = queue->fileio.active_block;
}
n = rounddown(n, buffer->bytes_per_datum);
if (n > block->bytes_used - queue->fileio.pos)
n = block->bytes_used - queue->fileio.pos;
if (copy_to_user(user_buffer, block->vaddr + queue->fileio.pos, n)) {
ret = -EFAULT;
goto out_unlock;
}
queue->fileio.pos += n;
if (queue->fileio.pos == block->bytes_used) {
queue->fileio.active_block = NULL;
iio_dma_buffer_enqueue(queue, block);
}
ret = n;
out_unlock:
mutex_unlock(&queue->lock);
return ret;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_read);
/**
* iio_dma_buffer_data_available() - DMA buffer data_available callback
* @buf: Buffer to check for data availability
*
* Should be used as the data_available callback for iio_buffer_access_ops
* struct for DMA buffers.
*/
size_t iio_dma_buffer_data_available(struct iio_buffer *buf)
{
struct iio_dma_buffer_queue *queue = iio_buffer_to_queue(buf);
struct iio_dma_buffer_block *block;
size_t data_available = 0;
/*
* For counting the available bytes we'll use the size of the block not
* the number of actual bytes available in the block. Otherwise it is
* possible that we end up with a value that is lower than the watermark
* but won't increase since all blocks are in use.
*/
mutex_lock(&queue->lock);
if (queue->fileio.active_block)
data_available += queue->fileio.active_block->size;
spin_lock_irq(&queue->list_lock);
list_for_each_entry(block, &queue->outgoing, head)
data_available += block->size;
spin_unlock_irq(&queue->list_lock);
mutex_unlock(&queue->lock);
return data_available;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_data_available);
/**
* iio_dma_buffer_set_bytes_per_datum() - DMA buffer set_bytes_per_datum callback
* @buffer: Buffer to set the bytes-per-datum for
* @bpd: The new bytes-per-datum value
*
* Should be used as the set_bytes_per_datum callback for iio_buffer_access_ops
* struct for DMA buffers.
*/
int iio_dma_buffer_set_bytes_per_datum(struct iio_buffer *buffer, size_t bpd)
{
buffer->bytes_per_datum = bpd;
return 0;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_set_bytes_per_datum);
/**
* iio_dma_buffer_set_length - DMA buffer set_length callback
* @buffer: Buffer to set the length for
* @length: The new buffer length
*
* Should be used as the set_length callback for iio_buffer_access_ops
* struct for DMA buffers.
*/
int iio_dma_buffer_set_length(struct iio_buffer *buffer, unsigned int length)
iio: Add generic DMA buffer infrastructure The traditional approach used in IIO to implement buffered capture requires the generation of at least one interrupt per sample. In the interrupt handler the driver reads the sample from the device and copies it to a software buffer. This approach has a rather large per sample overhead associated with it. And while it works fine for samplerates in the range of up to 1000 samples per second it starts to consume a rather large share of the available CPU processing time once we go beyond that, this is especially true on an embedded system with limited processing power. The regular interrupt also causes increased power consumption by not allowing the hardware into deeper sleep states, which is something that becomes more and more important on mobile battery powered devices. And while the recently added watermark support mitigates some of the issues by allowing the device to generate interrupts at a rate lower than the data output rate, this still requires a storage buffer inside the device and even if it exists it is only a few 100 samples deep at most. DMA support on the other hand allows to capture multiple millions or even more samples without any CPU interaction. This allows the CPU to either go to sleep for longer periods or focus on other tasks which increases overall system performance and power consumption. In addition to that some devices might not even offer a way to read the data other than using DMA, which makes DMA mandatory to use for them. The tasks involved in implementing a DMA buffer can be divided into two categories. The first category is memory buffer management (allocation, mapping, etc.) and hooking this up the IIO buffer callbacks like read(), enable(), disable(), etc. The second category of tasks is to setup the DMA hardware and manage the DMA transfers. Tasks from the first category will be very similar for all IIO drivers supporting DMA buffers, while the tasks from the second category will be hardware specific. This patch implements a generic infrastructure that take care of the former tasks. It provides a set of functions that implement the standard IIO buffer iio_buffer_access_funcs callbacks. These can either be used as is or be overloaded and augmented with driver specific code where necessary. For the DMA buffer support infrastructure that is introduced in this series sample data is grouped by so called blocks. A block is the basic unit at which data is exchanged between the application and the hardware. The application is responsible for allocating the memory associated with the block and then passes the block to the hardware. When the hardware has captured the amount of samples equal to size of a block it will notify the application, which can then read the data from the block and process it. The block size can freely chosen (within the constraints of the hardware). This allows to make a trade-off between latency and management overhead. The larger the block size the lower the per sample overhead but the latency between when the data was captured and when the application will be able to access it increases, in a similar way smaller block sizes have a larger per sample management overhead but a lower latency. The ideal block size thus depends on system and application requirements. For the time being the infrastructure only implements a simple double buffered scheme which allocates two blocks each with half the size of the configured buffer size. This provides basic support for capturing continuous uninterrupted data over the existing file-IO ABI. Future extensions to the DMA buffer infrastructure will give applications a more fine grained control over how many blocks are allocated and the size of each block. But this requires userspace ABI additions which are intentionally not part of this patch and will be added separately. Tasks of the second category need to be implemented by a device specific driver. They can be hooked up into the generic infrastructure using two simple callbacks, submit() and abort(). The submit() callback is used to schedule DMA transfers for blocks. Once a DMA transfer has been completed it is expected that the buffer driver calls iio_dma_buffer_block_done() to notify. The abort() callback is used for stopping all pending and active DMA transfers when the buffer is disabled. Signed-off-by: Lars-Peter Clausen <lars@metafoo.de> Signed-off-by: Jonathan Cameron <jic23@kernel.org>
2015-10-14 00:10:28 +08:00
{
/* Avoid an invalid state */
if (length < 2)
length = 2;
buffer->length = length;
buffer->watermark = length / 2;
return 0;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_set_length);
/**
* iio_dma_buffer_init() - Initialize DMA buffer queue
* @queue: Buffer to initialize
* @dev: DMA device
* @ops: DMA buffer queue callback operations
*
* The DMA device will be used by the queue to do DMA memory allocations. So it
* should refer to the device that will perform the DMA to ensure that
* allocations are done from a memory region that can be accessed by the device.
*/
int iio_dma_buffer_init(struct iio_dma_buffer_queue *queue,
struct device *dev, const struct iio_dma_buffer_ops *ops)
{
iio_buffer_init(&queue->buffer);
queue->buffer.length = PAGE_SIZE;
queue->buffer.watermark = queue->buffer.length / 2;
queue->dev = dev;
queue->ops = ops;
INIT_LIST_HEAD(&queue->incoming);
INIT_LIST_HEAD(&queue->outgoing);
mutex_init(&queue->lock);
spin_lock_init(&queue->list_lock);
return 0;
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_init);
/**
* iio_dma_buffer_exit() - Cleanup DMA buffer queue
* @queue: Buffer to cleanup
*
* After this function has completed it is safe to free any resources that are
* associated with the buffer and are accessed inside the callback operations.
*/
void iio_dma_buffer_exit(struct iio_dma_buffer_queue *queue)
{
unsigned int i;
mutex_lock(&queue->lock);
spin_lock_irq(&queue->list_lock);
for (i = 0; i < ARRAY_SIZE(queue->fileio.blocks); i++) {
if (!queue->fileio.blocks[i])
continue;
queue->fileio.blocks[i]->state = IIO_BLOCK_STATE_DEAD;
}
INIT_LIST_HEAD(&queue->outgoing);
spin_unlock_irq(&queue->list_lock);
INIT_LIST_HEAD(&queue->incoming);
for (i = 0; i < ARRAY_SIZE(queue->fileio.blocks); i++) {
if (!queue->fileio.blocks[i])
continue;
iio_buffer_block_put(queue->fileio.blocks[i]);
queue->fileio.blocks[i] = NULL;
}
queue->fileio.active_block = NULL;
queue->ops = NULL;
mutex_unlock(&queue->lock);
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_exit);
/**
* iio_dma_buffer_release() - Release final buffer resources
* @queue: Buffer to release
*
* Frees resources that can't yet be freed in iio_dma_buffer_exit(). Should be
* called in the buffers release callback implementation right before freeing
* the memory associated with the buffer.
*/
void iio_dma_buffer_release(struct iio_dma_buffer_queue *queue)
{
mutex_destroy(&queue->lock);
}
EXPORT_SYMBOL_GPL(iio_dma_buffer_release);
MODULE_AUTHOR("Lars-Peter Clausen <lars@metafoo.de>");
MODULE_DESCRIPTION("DMA buffer for the IIO framework");
MODULE_LICENSE("GPL v2");