Linux system-based serial RapidIo data transmission method

Abstract

The invention discloses a Linux system-based serial RapidIo data transmission method. By utilizing a few quantity of DMA (direct memory access) controller resources, adopting a DMA linked list mode and mapping a register space of a DMA controller to an FPGA (field programmable gate array), the [CC]bit of the FPGA is realized by writing a DMA mode register MRn of a CPU (central processing unit), so as to start the DMA linked list transmission and initialize the data transmission of a rapidio link; during the whole data transmission process, the CPU does not need to join no matter in writing the DMA resister, transferring DMA data or the last rapidio link data transmission, so that the CPU efficiency is greatly improved; and the data transmission is realized in accordance with time sequence demands strictly, so that the dilemma of performance reduction caused by the fact that the time sequence is realized through interruption and the interruption is responded constantly; the mapping of an internal-core space address to a user space address is realized through a uio technique, so that the internal-core space address can be directly operated at the user space, and the data copy at the user space and the internal-core space is avoided; and moreover DMA linked list nodes can be designed so as to realize the flexible control in data transmission of a rapidio link output window.

Description

technical field [0001] The invention relates to the technical field of wireless communication, in particular to a serial RapidIo data transmission method based on a Linux system. Background technique [0002] In the embedded system, the traditional timing control is notified by the external interrupt to the CPU, and then the CPU completes the work that the timing should achieve. Generally speaking, due to the preemptiveness and exclusiveness of the external interrupt, it is determined that if When the interrupt arrives very close, the phenomenon of lost interrupt will occur. If the interrupt is threaded to reduce the interrupt processing time, thereby alleviating the problem of lost interrupt, the real-time performance of the system will be reduced. [0003] The data exchange between user space and kernel space will inevitably result in the cost of context switching and data copying, while using uio technology (the technology of user space for IO access) is to directly map t...

AI Technical Summary

Problems solved by technology

When only one cell is managed, the data is copied 3 times, and the interrupt response is 2 times. When managing multiple cells, these operations will be doubled. At this time, the performance of the system according to the ordinary design will not meet the actual needs.

Images