Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

api: Add DMAPool. #231

Merged
merged 1 commit into from
May 6, 2024
Merged

api: Add DMAPool. #231

merged 1 commit into from
May 6, 2024

Conversation

iabdalkader
Copy link
Contributor

@iabdalkader iabdalkader commented May 2, 2024

The DMAPool class allocates and manages a pool of buffers that are DMA and cache-friendly, designed for applications that require frequent buffer exchanges between the CPU and DMA (such as audio applications).
The DMAPool maintains a read and write queues of DMA buffers. DMA buffers' sizes are rounded up to a multiple of the pool's alignment (which defaults to cache line size on platforms with caches), and their memory is initialized from one contiguous memory block.
A typical usage of DMAPool involves allocating a DMA buffer from the write queue for writing by a producer, updating its contents, and then releasing it. When released, the DMA buffer returns to the pool, which in turn places it back into the read queue for later consumption by the consumer. For example:

// Writer/Producer side (For example, an IRQ handler).
DMABuffer<uint16_t> *buf = pool->alloc(DMA_BUFFER_WRITE);
for (size_t i=0; i<buf.size(); i++) {
    buf[i] = 0xFFFF;
}
buf->release();

// Reader/Consumer side (User/library).
DMABuffer<uint16_t> *buf = pool->alloc(DMA_BUFFER_READ);
for (size_t i=0; i<buf.size(); i++) {
    print(buf[i]);
}
buf->release();

Note that the DMAPool uses single-writer, single-reader lock-free queues to store buffers, and as such, it can only be used by a single reader and a single writer. Locks are avoided to allow the DMAPool to be used from an ISR producer/consumer, with only the main thread, and without disabling IRQs.

The DMAPool class allocates and manages a pool of buffers that are DMA and cache
friendly, designed for applications that require frequent buffer exchanges between
the CPU and DMA (such as audio applications).
The DMAPool maintains a read and write queues of DMA buffers. DMA buffers' sizes
are rounded up to a multiple of the pool's alignment (which defaults to cache line
size on platforms with caches), and their memory is initialized from one contiguous
memory block.
A typical usage of DMAPool involves allocating a DMA buffer from the write queue
for writing by a producer, updating its contents, and then releasing it. When released,
the DMA buffer returns to the pool, which in turn places it back into the read queue
for later consumption by the consumer. For example:

```C++
// Writer/Producer side (For example, an IRQ handler).
DMABuffer<uint16_t> *buf = pool->alloc(DMA_BUFFER_WRITE);
for (size_t i=0; i<buf.size(); i++) {
    buf[i] = 0xFFFF;
}
buf->release();

// Reader/Consumer side (User/library).
DMABuffer<uint16_t> *buf = pool->alloc(DMA_BUFFER_READ);
for (size_t i=0; i<buf.size(); i++) {
    print(buf[i]);
}
buf->release();
```

Note that the DMAPool uses single-writer, single-reader lock-free queues to store
buffers, and as such, it can only be used by a single reader and a single writer.
Locks are avoided to allow the DMAPool to be used from an ISR producer/consumer,
with only the main thread, and without disabling IRQs.

Signed-off-by: iabdalkader <i.abdalkader@gmail.com>
@facchinm facchinm merged commit 6ebbb86 into arduino:master May 6, 2024
2 of 3 checks passed
@iabdalkader iabdalkader deleted the dma_pool branch May 6, 2024 10:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants