From: Dmitry Osipenko Subject: Re: [PATCH v3 1/2] soc/tegra: pmc: Query PCLK clock rate at probe time Date: Fri, 2 Aug 2019 17:26:45 +0300 Message-ID: <218c3d60-1a56-7ee8-840b-7d0559cd8c8e@gmail.com> References: <20190730174020.15878-1-digetx@gmail.com> <20190802133846.GC3883@pdeschrijver-desktop.Nvidia.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit Return-path: In-Reply-To: <20190802133846.GC3883@pdeschrijver-desktop.Nvidia.com> Content-Language: en-US Sender: linux-kernel-owner@vger.kernel.org To: Peter De Schrijver Cc: Thierry Reding , Jonathan Hunter , linux-tegra@vger.kernel.org, linux-kernel@vger.kernel.org List-Id: linux-tegra@vger.kernel.org 02.08.2019 16:38, Peter De Schrijver пишет: > On Tue, Jul 30, 2019 at 08:40:19PM +0300, Dmitry Osipenko wrote: >> The PCLK clock is running off SCLK, which is a critical clock that is >> very unlikely to randomly change its rate. It is possible to get a >> lockup if kernel decides to enter LP2 cpuidle from a clk-notifier, which >> happens occasionally in a case of Tegra30 EMC driver that waits for the >> clk-change event in the clk-notify handler, because CCF's 'prepare' mutex >> in kept locked and thus clk_get_rate() wants to sleep with interrupts >> being disabled. >> > > I don't think this is the right solution. Eventually we will want to > scale sclk and pclk because the clock tree power of those is not > insignificant. Maybe register a notifier which updates the PMC timer > values when pclk changes? I also had a thought about the notifier for pclk, but wasn't sure if it's really worthwhile right now since there is no real use-case and it's not obvious when such case will materialize. So, I'd say that solution is correct but incomplete. I'll make a v4 with the notifier, since you're asking about it. Thanks for the review!