we are trying to develop some sort of RTC virtualization,
that means several clients (userspace processes) which share the RTC timer.
Assume that we can use an rtc daemon (or a kernel module) (which actually opens
the RTC device and periodically wakes up the clients throught some sort of IPC
(socket/pipe etc) ).
Now comes the fun:
of course we want maximum efficiency, so it would be very useful to change the
RTC frequency when needed.
assume the following idiotic case:
client1 requests RTC at 256HZ (the daemon opens the RTC device at 256HZ and
periodically "wakes up" the client)
now comes client2:
client2 wants RTC at 128HZ
no problem for the server: just wakeup client2 every 2 RTC ticks.
now client3 wants RTC at 1024 HZ
in this case the server has to change the RTC frequency to 1024HZ and
wakeup client2 every 4ticks and client2 every 8 ticks.
now client1 exits, nothing happens just remove the client from the wakeup list.
later client3 exits, the server changes the RTC frequency to 128HZ and servers
client3.
My question: are there major delays when changing RTC frequency on the fly ?
If yes, what is the order of magnitude of the delay ?
I ask this because it would just be a waste to run an RTC at 8192HZ only to
serve potential 8192HZ clients (8192HZ causes nice IRQ overhead).
comments ?
PS: please CC me because I'm not on the list.
regards,
Benno.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/