Re: [PATCH v2 2/2] ARM: dts: gpio-ranges property is now required

From: nicolas saenz julienne <nsaenz@kernel.org>
To: Florian Fainelli <f.fainelli@gmail.com>,
	Phil Elwell <phil@raspberrypi.com>,
	Marek Szyprowski <m.szyprowski@samsung.com>,
	"linus.walleij" <linus.walleij@linaro.org>
Cc: Rob Herring <robh+dt@kernel.org>,
	bcm-kernel-feedback-list@broadcom.com,
	Thierry Reding <treding@nvidia.com>,
	devicetree@vger.kernel.org, linux-rpi-kernel@lists.infradead.org,
	linux-gpio@vger.kernel.org
Subject: Re: [PATCH v2 2/2] ARM: dts: gpio-ranges property is now required
Date: Wed, 15 Dec 2021 10:02:54 +0100	[thread overview]
Message-ID: <83bda801c9be665086f94acce46d44da5d24ba9c.camel@kernel.org> (raw)
In-Reply-To: <dbba6665-9688-9a5e-091b-eaa8e92e6886@gmail.com>

Florian,

On Tue, 2021-12-14 at 09:12 -0800, Florian Fainelli wrote:
> On 12/14/21 6:32 AM, Phil Elwell wrote:
> > Hi Marek,
> > 
> > On 14/12/2021 14:21, Marek Szyprowski wrote:
> > > Hi Phil,
> > > 
> > > On 06.12.2021 10:22, Phil Elwell wrote:
> > > > Since [1], added in 5.7, the absence of a gpio-ranges property has
> > > > prevented GPIOs from being restored to inputs when released.
> > > > Add those properties for BCM283x and BCM2711 devices.
> > > > 
> > > > [1] commit 2ab73c6d8323 ("gpio: Support GPIO controllers without
> > > >       pin-ranges")
> > > > 
> > > > Fixes: 2ab73c6d8323 ("gpio: Support GPIO controllers without
> > > > pin-ranges")
> > > > Signed-off-by: Phil Elwell <phil@raspberrypi.com>
> > > 
> > > This patch breaks today's linux-next (next-20211214) on RPi3 and RPi4.
> > > Either there is something missing or wrong here. Booting stops after
> > > following messages (on RPi4):
> > > 
> > > [    3.186786] pinctrl-bcm2835 fe200000.gpio: could not add GPIO chip
> > > [    3.234513] pinctrl-bcm2835 fe200000.gpio: could not add GPIO chip
> > > [    3.276703] mmc0: SDHCI controller on fe340000.mmc [fe340000.mmc]
> > > using ADMA
> > > [    3.287191] pinctrl-bcm2835 fe200000.gpio
> > 
> > This patch is part of a two-patch set, the cover note for which says:
> > 
> >     2. Since [1], a "gpio-ranges" property is required in order for pins
> >     to be returned to inputs when freed. Note that without patch 1, the
> >     device never gets out of EPROBE_DEFER.
> > 
> > It looks as though patch 2 has been merged without/before patch 1
> > ("pinctrl: bcm2835: Change init order for gpio hogs").
> 
> Yes, the hope was that there would be no such breakage, I suppose we
> will have to work out a plan to address that and coordinate both changes
> landing in at the same time.
> 
> I will work with Arnd to back out the Device Tree changes, sorry about that.

This is linux-next, so I can back out the DT change myself. Sorry for the
breakage.

As for channeling the path, would it make sense for linusw to take it alonside
GPIO fix?

Regards,
Nicolas