I have a kernel module that will do that by commenting out the printk
statement and then adding a printk statement in the module unload
routine. Its posted in here somewhere.
I didn't think about ttyam0 because nothing was connected to it. On
all my other boxes I direct kernel messages to a log file. It makes
sense now why it was slow. My application for the interrupt was 1
million per second, it took me a while to deal with the fact that its not
a microcontroller with no OS. ;)
At 01:24 PM 3/30/2005 -0600, you wrote:
> The cpu is already doing
100 timer interupts/sec non stop, and running the
> process scheduler each time, and this takes very very very little
time -
> if it didn't we wouldn't be running linux on the board becuase
there'd be
> no time left for user apps. Printk will be writing chars over
/dev/ttyAM0
> (the kernel console device) and from an interrupt routine!!! I bet
system
> clock went belly up too!
>
> Though I've not tested it, I'm sure a very simple IRQ service
routine just
> incrementing counters etc, which are accessible from some /proc
entry or
> similar will place negligable load on the CPU and board for upto
several
> hundred irqs/sec.
The trick will be getting the input pulses to generate
interrupt(s).
I've not recently dug into the chip docs for the GPIO pins, but
that will be a must-have.
--
Andy
Warner
Voice: (612) 801-8549 Fax: (208)
575-5634
Yahoo! Groups Sponsor
ADVERTISEMENT
Yahoo! Groups Links
| Yahoo! Groups Sponsor |
ADVERTISEMENT
![click here]() | |
![]() |
Yahoo! Groups Links
|