From mboxrd@z Thu Jan 1 00:00:00 1970 From: will.deacon@arm.com (Will Deacon) Date: Thu, 5 Jul 2012 15:15:43 +0100 Subject: [PATCH v2 2/2] ARM: delay: allow timer-based delay implementation to be selected In-Reply-To: <4FF59145.2020009@renesas.com> References: <1340991231-17682-1-git-send-email-will.deacon@arm.com> <1340991231-17682-3-git-send-email-will.deacon@arm.com> <4FF59145.2020009@renesas.com> Message-ID: <20120705141542.GH3399@mudshark.cambridge.arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Thu, Jul 05, 2012 at 02:06:13PM +0100, Shinya Kuribayashi wrote: > On 6/30/2012 2:33 AM, Will Deacon wrote: > > +extern void init_current_timer_delay(unsigned long freq); > > + > > This is needed for all users of the new timer-baed delay routine. > How about moving this extern declaration to more generic place, in > ? I mentioned this before in the original posting. Currently, only the architected timers are using this interface and I have some reservations about using memory-mapped timers instead (although there may be specific instances where they're suitable). In this case, I'd rather see a registration mechanism for the current_timer where we take a pointer to read_current_timer and register the delay at the same time. That can be added when somebody else decides they want to make use of the new delay routines. > Tested-off-by: Shinya Kuribayashi Thanks! Will