From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sudeep Holla Subject: Re: [PATCH] Revert "arm64: dts: marvell: add CPU Idle power state support on Armada 7K/8K" Date: Tue, 18 Dec 2018 17:11:44 +0000 Message-ID: <20181218171144.GA17290@e107155-lin> References: <87k1kowf7i.fsf@bootlin.com> <20181205101035.GA31059@e107155-lin> <877eg6hiv9.fsf@bootlin.com> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Return-path: Content-Disposition: inline In-Reply-To: <877eg6hiv9.fsf@bootlin.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=m.gmane.org@lists.infradead.org To: Gregory CLEMENT Cc: Andrew Lunn , Baruch Siach , Jason Cooper , Sergey Matyukevich , linux-pm@vger.kernel.org, orenbh , Daniel Lezcano , "Rafael J. Wysocki" , Russell King , Sudeep Holla , linux-arm-kernel@lists.infradead.org, Sebastian Hesselbarth List-Id: linux-pm@vger.kernel.org On Tue, Dec 18, 2018 at 05:58:34PM +0100, Gregory CLEMENT wrote: > Hi Sudeep, > > On mer., d=E9c. 05 2018, Sudeep Holla wrote: > > > On Wed, Dec 05, 2018 at 09:27:45AM +0100, Gregory CLEMENT wrote: > >> Hi Baruch, > >> > >> On mer., d=E9c. 05 2018, Baruch Siach wrote: > >> > >> > Hi Gregory, > >> > > >> > On Tue, Dec 04, 2018 at 06:48:43PM +0200, Baruch Siach wrote: > >> >> On Tue, Dec 04, 2018 at 05:21:25PM +0100, Gregory CLEMENT wrote: > >> >> > On mar., d=E9c. 04 2018, Baruch Siach wrote: > >> >> > > This reverts commit 8ed46368776b3bc93d74c1f8f2bfb9fd8a9ad805. > >> >> > > > >> >> > > This commit breaks boot on Armada 8K based systems. Reverting i= t makes > >> >> > > affected systems boot again. > >> >> > > > >> >> > > Reported-by: Sergey Matyukevich > >> >> > > Signed-off-by: Baruch Siach > >> >> > > >> >> > I would like to know more about it because on the McBin I use the= re is > >> >> > no issue with 4.20-rc1. > >> >> > > >> >> > Do you observe this issue with the arm64 defconfig or do you have= some > >> >> > specific configuration option? > >> >> > >> >> I tested the attached defconfig on Clearfog GT-8K and on another cu= stom Armada > >> >> 8K based board. > >> >> > >> >> Tomorrow I'll test Macchiatobin with this config and the arm64 defc= onfig. > >> > > >> > I reproduced same issue on a Macchiatobin board with both this confi= g and the > >> > arm64 defconfig. The latter produced this splat: > >> > >> OK thanks, for this, I also got the confirmation by my colleagues. The > >> only difference I can think of, is the firmware, but the surprising > >> thing, is that I have an old firmware: 17.06.2 and that with a recent > >> one, it doesn't work. > >> > >> So I will apply it in mvebu/fixes. > >> > > > > Sorry if it's too late. But I would rather disable the states that are > > broken on old firmware and see if it's possible for bootloader or boot > > firmware loading DT can modify the status to "enabled" if it can detect > > the firmware version. > > you mean adding status =3D "disable" on each idle-state subnode? > Yes indeed. Keep it disabled by default and can enable it in bootloader if you can detect it that it should work with particular version of firmware. Just a thought especially if certain set of platforms support it while another not. -- Regards, Sudeep