5. IOAT DMA Device Driver

The ioat dmadev driver provides a poll-mode driver (PMD) for Intel® QuickData Technology which is part of part of Intel® I/O Acceleration Technology (Intel I/OAT). This PMD, when used on supported hardware, allows data copies, for example, cloning packet data, to be accelerated by IOAT hardware rather than having to be done by software, freeing up CPU cycles for other tasks.

5.1. Hardware Requirements

The dpdk-devbind.py script, included with DPDK, can be used to show the presence of supported hardware. Running dpdk-devbind.py --status-dev dma will show all the DMA devices on the system, IOAT devices are included in this list. For Intel® IOAT devices, the hardware will often be listed as “Crystal Beach DMA”, or “CBDMA” or on some newer systems ‘0b00’ due to the absence of pci-id database entries for them at this point.

Note

Error handling is not supported by this driver on hardware prior to Intel Ice Lake. Unsupported systems include Broadwell, Skylake and Cascade Lake.

5.2. Compilation

For builds using meson and ninja, the driver will be built when the target platform is x86-based. No additional compilation steps are necessary.

5.3. Device Setup

Intel® IOAT devices will need to be bound to a suitable DPDK-supported user-space IO driver such as vfio-pci in order to be used by DPDK.

The dpdk-devbind.py script can be used to view the state of the devices using:

$ dpdk-devbind.py --status-dev dma

The dpdk-devbind.py script can also be used to bind devices to a suitable driver. For example:

$ dpdk-devbind.py -b vfio-pci 00:01.0 00:01.1

5.3.1. Device Probing and Initialization

For devices bound to a suitable DPDK-supported driver (vfio-pci), the HW devices will be found as part of the device scan done at application initialization time without the need to pass parameters to the application.

If the application does not require all the devices available an allowlist can be used in the same way that other DPDK devices use them.

For example:

$ dpdk-test -a <b:d:f>

Once probed successfully, the device will appear as a dmadev, that is a “DMA device type” inside DPDK, and can be accessed using APIs from the rte_dmadev library.

5.4. Using IOAT DMAdev Devices

To use IOAT devices from an application, the dmadev API can be used.

5.4.1. Device Configuration

IOAT configuration requirements:

  • ring_size must be a power of two, between 64 and 4096.

  • Only one vchan is supported per device.

  • Silent mode is not supported.

  • The transfer direction must be set to RTE_DMA_DIR_MEM_TO_MEM to copy from memory to memory.

Once configured, the device can then be made ready for use by calling the rte_dma_start() API.

5.4.2. Performing Data Copies

Refer to the Enqueue / Dequeue APIs section of the dmadev library documentation for details on operation enqueue, submission and completion API usage.

It is expected that, for efficiency reasons, a burst of operations will be enqueued to the device via multiple enqueue calls between calls to the rte_dma_submit() function.

When gathering completions, rte_dma_completed() should be used, up until the point an error occurs with an operation. If an error was encountered, rte_dma_completed_status() must be used to reset the device and continue processing operations. This function will also gather the status of each individual operation which is filled in to the status array provided as parameter by the application.

The status codes supported by IOAT are:

  • RTE_DMA_STATUS_SUCCESSFUL: The operation was successful.

  • RTE_DMA_STATUS_INVALID_SRC_ADDR: The operation failed due to an invalid source address.

  • RTE_DMA_STATUS_INVALID_DST_ADDR: The operation failed due to an invalid destination address.

  • RTE_DMA_STATUS_INVALID_LENGTH: The operation failed due to an invalid descriptor length.

  • RTE_DMA_STATUS_DESCRIPTOR_READ_ERROR: The device could not read the descriptor.

  • RTE_DMA_STATUS_ERROR_UNKNOWN: The operation failed due to an unspecified error.

The following code shows how to retrieve the number of successfully completed copies within a burst and then uses rte_dma_completed_status() to check which operation failed and reset the device to continue processing operations:

enum rte_dma_status_code status[COMP_BURST_SZ];
uint16_t count, idx, status_count;
bool error = 0;

count = rte_dma_completed(dev_id, vchan, COMP_BURST_SZ, &idx, &error);

if (error){
   status_count = rte_dma_completed_status(dev_id, vchan, COMP_BURST_SZ, &idx, status);
}