From mboxrd@z Thu Jan 1 00:00:00 1970 From: Russell King - ARM Linux Subject: Re: [PATCH v2 7/7] ARM: smp: Add runtime PM support for CPU hotplug Date: Fri, 4 Sep 2015 18:46:59 +0100 Message-ID: <20150904174659.GH21084@n2100.arm.linux.org.uk> References: <1441310314-8857-1-git-send-email-lina.iyer@linaro.org> <1441310314-8857-8-git-send-email-lina.iyer@linaro.org> <20150904091735.GC21084@n2100.arm.linux.org.uk> <20150904092716.GD21084@n2100.arm.linux.org.uk> <20150904151202.GA876@linaro.org> <20150904162325.GG21084@n2100.arm.linux.org.uk> <20150904170211.GF876@linaro.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from pandora.arm.linux.org.uk ([78.32.30.218]:50656 "EHLO pandora.arm.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753601AbbIDRrO (ORCPT ); Fri, 4 Sep 2015 13:47:14 -0400 Content-Disposition: inline In-Reply-To: <20150904170211.GF876@linaro.org> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Lina Iyer Cc: Mark Rutland , rjw@rjwysocki.net, ulf.hansson@linaro.org, Lorenzo Pieralisi , linux-pm@vger.kernel.org, Catalin Marinas , sboyd@codeaurora.org, k.kozlowski@samsung.com, msivasub@codeaurora.org, khilman@linaro.org, geert@linux-m68k.org, agross@codeaurora.org, linux-arm-kernel@lists.infradead.org On Fri, Sep 04, 2015 at 11:02:11AM -0600, Lina Iyer wrote: > On Fri, Sep 04 2015 at 10:23 -0600, Russell King - ARM Linux wrote: > >You're taking the behaviour of the hardware you have in front of you > >and claiming that it's true everywhere, and shoving that into generic > >code. > > > >I know, for example on OMAP, you have to power up the CPU first before > >you can "wake" it. > > > >I wouldn't be surprised if other SoCs are like that: where they require > >the CPU core to be powered and held in reset, before releasing the reset > >to then allow them to start executing the secondary core bringup. > > > It would still require that the CPU's domain be powered on in the hw, > before the CPU can run Linux. > > >Relying on hardware to do this sounds really fragile and bad design to > >me. > > > >If you want to persue your current design, don't make it generic code, > >because it's got no right to be generic with assumptions like that. > > > Hmm, okay. I can look at alternatives like hotplug notifiers. How about putting the pm resume in path of the _requesting_ CPU? IOW, in __cpu_up(). -- FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up according to speedtest.net.