From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Mon, 27 Jun 2016 11:35:31 +0300 From: Peter De Schrijver To: Thierry Reding CC: Rhyland Klein , Michael Turquette , Stephen Boyd , "Alexandre Courbot" , , , , Stephen Warren Subject: Re: [PATCH v2 01/11] clk: tegra: Switch to using critical clks Message-ID: <20160627083531.GA12863@tbergstrom-lnx.Nvidia.com> References: <1464381494-27096-1-git-send-email-rklein@nvidia.com> <1464381494-27096-2-git-send-email-rklein@nvidia.com> <20160622121630.GB26943@ulmo.ba.sec> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" In-Reply-To: <20160622121630.GB26943@ulmo.ba.sec> Return-Path: pdeschrijver@nvidia.com List-ID: On Wed, Jun 22, 2016 at 02:16:30PM +0200, Thierry Reding wrote: > * PGP Signed by an unknown key > > On Fri, May 27, 2016 at 04:38:04PM -0400, Rhyland Klein wrote: > > Mark some of the required-to-be-enabled clks as critical clks. These > > need to be kept on through the disabling of unused clks during init. > > They may not get any reference before or after init, but are required > > to be on, therefore let the core enable them. > > > > Signed-off-by: Rhyland Klein > > --- > > v2: > > - Remove mention of HAND_OFF clks as they are not supported yet. > > > > drivers/clk/tegra/clk-tegra-periph.c | 21 ++++++++++++++------- > > drivers/clk/tegra/clk-tegra-super-gen4.c | 12 +++++++----- > > 2 files changed, 21 insertions(+), 12 deletions(-) > > I have some difficulty to follow why some of these clocks are critical. > It might help if the commit message mentioned why each of these need to > remain enabled forever, even if never used. > > Also, it's fairly unlikely that pll_p for example would ever get > disabled because a bunch of others are derived from it. I'm also not > quite convinced yet that it really is critical. What does it drive which > isn't claimed by any drivers? > mselect and sclk are clocked from pll_p, however we don't have drivers for those. Most other peripherals clocked from pll_p can be clockgated by their driver which can lead to pll_p being turned off causing the system to hang. Cheers, Peter.