Hi, Thanks for the reply. But i still have some problem inspite of the explanation. When the interrupt is occured, it takes a maximum of 10ms for the tasklet to get scheduled. The possibility for the occurance of the interrupt is 10ns. So Assume on the occurance of the first interrupt, the tasklet is not yet scheduled. In the meanwhile, another interrupt has occured. Then the previous tasklet would not be scheduled anymore..Am i right. Plz let me know if my understanding is right. If it is right, can u plz pointer me a way by which all the tasklets be scheduled without the loss of any of them??? Is it possible at all???? Thanks in advance Regards, Vijay Ram.C ----- Original Message ----- From: "Mulyadi Santosa" <mulyadi.santosa@xxxxxxxxx> To: "Vijay Ram.C" <cvijay@xxxxxxxxxxxxxxxxxx>; "kernel newbies" <kernelnewbies@xxxxxxxxxxxx> Sent: Friday, September 30, 2005 8:30 AM Subject: Re: Doubt in bottom halves. > Hello... > > > I have a doubt regarding interrupt handling and bottom halves. > > I have registered an interrupt handler which is not a shared > > interrupt. When the interrupt occurs, the interrupt handler is called > > which schedules the tasklet and is proccessed During the proccessing > > of the tasklet, if the same interrupt comes again, the interrupt will > > the handler again. At that time, will the tasklet be scheduled again > > after the current executing tasklet ?? Can anybody please explain > > this scenario briefly ? > > Yes, you can still schedule the new tasklet so it is executed right > after the last same tasklet. I never tried it by myself, but according > to what I read on Linux kernel development, unless the current > tasklet's state is still in TASKLET_STATE_SCHED and/or disabled, you > are free to reschedule your tasklet (inside the interrupt handler) > > This is possible, thanks to the fact that tasklet_action() and > tasklet_hi_action() always check for pending tasklet continously. When > the number of pending tasklets are above certain limits, big chance it > is executed by ksoftirqd, not when do_irq() is about to return to > kernel/user context. But you don't have to worry about it. > > regards > > Mulyadi > > > -- > Kernelnewbies: Help each other learn about the Linux kernel. > Archive: http://mail.nl.linux.org/kernelnewbies/ > FAQ: http://kernelnewbies.org/faq/ -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
