hello Vijay.... > I am sorry i was not able to present my requirement clearly..infact > of the two options that u mentioned, > the 1st one is wat i require..If i get 5 interrupts, then i want the > tasklet to get executed 5 times.Since this is the > requirement, the linked list implementation doesnt solve the purpose. OK....sounds like what you are trying to do is similar with how timer's softirq try to catch up the running clock. The popular term is clock drift. before I am going to explain a little idea, please try this simple test. Create a kernel module that declare and schedule tasklet. Try to schedule a simple tasklet 4-5 times in a row and inside the handler, do something that you plan to do on your real tasklet. From there, watch...is that tasklet executed 5 times or not? If it is....merry Christmast :) No need to sweat...if it isn't....let's continue... In summary, here is the idea you can use. WARNING: UNTESTED!! 1. declare a global counter. You can use "int", "long" or whatever you want. Declare it as volatile so it won't be cached on CPU register during certain interval. let's name it "actual_interrupt_counter". Declare another global counter and name it "current_tasklet_counter". Initialize both as 0 somewhere.... 2. Atomically increase the "actual_interrupt_counter" everytime your handler is called. 3. Schedule your tasklet as usual. 4. Your tasklet handler will be executed (sooner or later) . Now assume this is the first time your tasklet handler is called and also the first time your interupt handler is called. So "actual_interrupt_counter" is 1 and "current_tasklet_counter" is 0. Just before the end of handler's function, atomically increase the "current_tasklet_counter"...and then compare it with "actual_interrupt_counter"" If they are equal or "current_tasklet_counter" is greater (which is unlikely,but who knows? :) ), just exit the function. If "current_tasklet_counter" is less, re-schedule your tasklet. For initial test, set the maximum limit of "current_tasklet_counter", e.g 100. So whenever it hit 100, it won't be scheduled anymore. This is intended to prevent unlimited scheduling caused by coding mistake somewhere.. Hope it helps and let me us know if the idea works.... regards Mulyadi -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
