Data transmission method and system

Abstract

The invention discloses a data transmission method and system applied to an I / O device to transmit data to a mainframe through a DMA. The method includes the steps that the index number S / P is calculated according to preset cache size S and average data packet length P, then the size I = (S / P) * U of an index memory block is calculated according to the size U of an index unit, and the index memory block is applied according to the size; according to the preset cache size S and the size of each data memory block M, the number N = S / M of the data memory blocks required to be allocated is calculated, and N data memory blocks are applied according to the preset data block size M; the starting physical address and length of the index memory block, starting physical addresses of all the data memory blocks and the size of the data memory blocks are written into the I / O device; the I / O device is controlled to transmit uplink data through the DMA; the mainframe reads index information and obtains uplink data packet contents according to records of the index unit. By adopting the data transmission method and system, the memory resource application success rate of a mainframe system can be improved, and smaller continuous memory blocks can be effectively utilized to make the I / O device provide more mainframe memory resources.

Description

technical field [0001] The invention relates to the field of data transmission, in particular to data transmission control for I / O devices such as PCIE, and mainly provides a method and system for memory resource allocation management and data transmission for DMA transmission of I / O devices such as PCIE. Background technique [0002] When the I / O device transmits data, it generally uses a Direct Memory Access (DMA) technology to transfer data on the I / O device, thereby realizing background batch data transmission independent of the CPU. In the DMA operation, the host computer allocates DMA memory address space for the DMA operation, and the I / O device directly reads and writes data in the DMA memory address space, thereby completing the DMA mode data transmission. The DMA operation is realized through the DMA message. The I / O device sends the DMA message to the host. The DMA message contains the DMA address. The DMA address is located in the DMA memory address space alloc...

AI Technical Summary

Problems solved by technology

In the past, the host generally applied for one or two larger continuous memories for data caching when allocating the uplink DMA cache. However, as the required cache becomes larger and larger, the disadvantages of the traditional method become more and more obvious:

[0004] Disadvantage 1: The probability of successful application for large continuous memory is low. Once the system cannot provide the expected continuous memory resources, it can only reduce the size and continue to apply until the application is successful, which often fails to obtain the expected cache size;

[0005] Disadvantage 2: The memory resources of the host system are not fully utilized, because a larger block of continuous memory needs to be applied for, so the smaller continuous memory blocks in the system cannot be used, which may cause the system to fail even if there are sufficient memory resources. Only part of it can be used by the corresponding PCIE device

Images