Hi Thank you for your reply. You answered to most of all my questions. Additional RTFM is required for me :) > But you said, you want to communicate right? that means passing a data, > awaiting for certain kernel event, or something else? Yes, abs. correct. I would like to modify ns2 simulator to get an opportunity to emualate real-world IP packets. So, the ns2 application is a userspace application. Now my LLD is looked like this: the ns2 simulator has a timetable with events. This events should be executed at the certain time (as I already described above). There are two types of events: 1) Internal events of simulator 2) "Real events" - e.g. the new packet arriving to the interface. When real-world IP packet arrives to the interface, my driver buffers it and sends a signal to ns2 application. When ns2 application receives a signal, it asks driver for new arrived packets IDs (via ioctl function), generates event (internal) and schedule it for later execution in a simulator. Thus i can't use a block reading from /proc file because these "virtual events" should be executed from time to time. Also, this design distrubs me about perfomance. See: When virtual event starts execution, I block all singnals (using sigprocmask(SIG_BLOCK)) and release them after (using sigprocmask(SIG_SETMASK)). I would like to protect my timetable (scheduler) from multiply changes at the the same time (like Critical sections). In my test example simulator usually handles about 1800 events per second. So i'm really aware of perfomance results. So, the last question: Is there any possibility to make it faster? Can anybody advice me more efficient ways?Are there some examples of such kind communication between kernel and userspace? Thank you! Best regards, Sergey Semionov -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
