On Wednesday 10 July 2002 01:40 am, Seth Arnold wrote: > On Wed, Jul 10, 2002 at 02:24:26PM +0900, Jonathan Khoo wrote: > > However, I also realize that by adding unnecessary printk statements > > will affect the system performance. How can I measure the "actual" > > timings? > > Heisenburg's uncertainty principal plays a part in kernel development :) > You will always be stuck with "bad" timings from measuring the timings. > There is no way around that. However, by just storing the timings in a > memory buffer to be retrieved later (say, through a character device, > ioctl, /proc file, sysctl, or your own filesystem) _AFTER_ you are done > doing whatever it is you care about, ought to help your timing immensely. > > printk is just so blasted slow. Is printk really that slow if you kill the lines in release_console_sem that try to wake up klogd, and in addition kill syslogd and klogd? Isn't printk just writing in a circular buffer at that point? Anton -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
