Hi guys, On 4/12/06, MHD.Tayseer Alquoatli <idoit.ief@xxxxxxxxx> wrote:>>>> On 4/11/06, Ashok kumar <ashokmudalister@xxxxxxxxx> wrote:> >> >> > Hi MHD.Tayseer,> >> > this is the link.....> >> http://www.faqfarm.com/Q/What_is_the_difference_between_RTOS_and_OS>>> i've checked this article .. there is only one missing idea in our> discussion, that any algorithm used in any RTOS must be independent of the> load/mem/number of process .. i.e. the same algorithm will take the same> time to execute no matter what was the number of processes running on the> system while executing> example: O(1) scheduler comes with Linux kernel 2.6 .. which means that> scheduling algorithm is independent of the number of processes running (O(n)> is not acceptable for RTOS scheduler)> I couldn't understand the last point in the article (the hook).> hope this helps>> MHD.Tayseer> I had worked o RTLinux about a year ago.. but let me try to explain. By "hooks" the author refers to software interrupts.What happens is the following :When a process is running in RT context ( i.e when a real-timeprocess is running, the RT-core kernel sitting on top of the GPOSkernel, is running the process), all the hardware interrupts for theGPOS kernel are disabled with " cli" instructions.Now, when a hardware interrupt FOR THE GPOS comes, it is marked andstored by the RT-core.When the RT-core finished executing the real-time task andrelinquishes the CPU, the GPOS kernel runs as an idle task.At this time, the GPOS hardware interrupts are enabled with "sti" andthe RT-core sends software interrupts for the previously markedhardware interrupts and the GPOS finally services those interrupts. Hope this clears some dobts --Raseel.��������!���W��v������ޗ��{��f���� ��ޖw�n'�������Y�����
