From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Sat, 26 Jun 2010 23:43:59 +0200 From: Pavel Machek Subject: Re: [RFC] [PATCH] timer: Added usleep[_range][_interruptable] timer Message-ID: <20100626214359.GA1773@elf.ucw.cz> References: <4C225EED.5040600@codeaurora.org> <1277323537.15159.30.camel@c-dwalke-linux.qualcomm.com> <4C226CCA.20601@codeaurora.org> <1277326605.15159.70.camel@c-dwalke-linux.qualcomm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1277326605.15159.70.camel@c-dwalke-linux.qualcomm.com> To: Daniel Walker Cc: Patrick Pannuto , linux-kernel@vger.kernel.org, sboyd@codeaurora.org, tglx@linutronix.de, mingo@elte.hu, heiko.carstens@de.ibm.com, eranian@google.com, schwidefsky@de.ibm.com, linux-arm-msm@vger.kernel.org List-ID: Hi! > > Yes, this test was leftover from a different project that involved refactoring > > timers, so it was available and easy. My guess for the reduction in number of > > wakeups is that the processor is able to do other work during the 100us it was > > previously busy-waiting, and thus had to wake up less often. > > As I said in the prior email the udelay()'s don't preclude other types > of work since you can get preempted. Yes, you can get preempted, but you'll still spin in the tight loop counting... So it does not preclude other task, but then you'll spin unneccessarily. Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html