分析了 Linux 的實時性，針對其在實時應(yīng)用中的技術(shù)障礙，在參考了與此相關(guān)研究基礎(chǔ)上，從三方面提出了改善Linux 實時性能的改進(jìn)措施。為提高嵌入式應(yīng)用響應(yīng)時間精度,提出兩種細(xì)化Linux 時鐘粒度方法；為增強(qiáng)系統(tǒng)內(nèi)核對實時任務(wù)的響應(yīng)能力,采用插入搶占點和修改內(nèi)核法增強(qiáng)Linux 內(nèi)核的可搶占性；為保證硬實時任務(wù)的時限要求，把原Linux 的單運(yùn)行隊列改為雙運(yùn)行隊列,硬實時任務(wù)單獨被放在一個隊列中,并采用MLF 調(diào)度算法代替原內(nèi)核的FIFO 調(diào)度算法。
關(guān)鍵詞：Linux；實時性；調(diào)度策略；搶占
Abstract: The real-time performance of Linux is analyzed, and its disadvantage is pointed out when applied to the embedded system. Inspired mostly by some real time improvements of Linux, this article gives rise to three aspects to improve the real-time capability of Linux. To improve the time precision for embedded applications, two methods for creating accurate timer are proposed. To enhance the response ability of real-time task, inserting preemption points and the preemption kernel methods are adopted to enhance the preemptability of Linux kernel. To ensure the requirements of the hard real-time tasks, a double run queue is used instead of the original single run queue, these tasks are organized in one run queue and MLF schedule strategy is used for them.
Key word: Linux；Real-time；Schedule strategy；Preemption