Message ID | 1147be600733a1bcde194f7df61866019f34a531.1538504729.git.leonard.crestez@nxp.com |
---|---|
State | New |
Headers | show |
Series | ARM: dts: imx6sx-sdb: Fix enet phy regulator | expand |
On Tue, Oct 2, 2018 at 8:37 PM Leonard Crestez <leonard.crestez@nxp.com> wrote: > This issue was exposed by commit efdfeb079cc3 ("regulator: fixed: > Convert to use GPIO descriptor only") which causes the "gpios" property > to also be parsed. Before that commit the "gpios" property had no > effect and PHY reset was only handled in the bootloader. Wow that was a very interesting one. Turning over these stones sometimes expose the weirdest things. This is one of the reasons why we need formal validation of DTS files: things look OK to the human eye and get merged but they are just syntactically wrong. I guess I could hack to make "gpios" be ignored by the regulator GPIO quirks in gpiolib, but I take it you probably prefer to fix up the real issue like this. Anyways, good job! Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Yours, Linus Walleij
Hi Leonard, On Tue, Oct 2, 2018 at 3:42 PM Leonard Crestez <leonard.crestez@nxp.com> wrote: > @@ -371,10 +372,12 @@ > MX6SX_PAD_RGMII1_RD1__ENET1_RX_DATA_1 0x3081 > MX6SX_PAD_RGMII1_RD2__ENET1_RX_DATA_2 0x3081 > MX6SX_PAD_RGMII1_RD3__ENET1_RX_DATA_3 0x3081 > MX6SX_PAD_RGMII1_RX_CTL__ENET1_RX_EN 0x3081 > MX6SX_PAD_ENET2_RX_CLK__ENET2_REF_CLK_25M 0x91 > + /* phy reset */ > + MX6SX_PAD_ENET2_CRS__GPIO2_IO_7 0x80000000 We should avoid using 0x80000000 as this relies on the bootloader setting the IOMUX. Please do like this insted: MX6SX_PAD_ENET2_CRS__GPIO2_IO_7 0x10b0 Then you can add: Reviewed-by: Fabio Estevam <festevam@gmail.com>
On Tue, 2018-10-02 at 21:56 +0200, Linus Walleij wrote: > On Tue, Oct 2, 2018 at 8:37 PM Leonard Crestez < > leonard.crestez@nxp.com> wrote: > > > This issue was exposed by commit efdfeb079cc3 ("regulator: fixed: > > Convert to use GPIO descriptor only") which causes the "gpios" > > property > > to also be parsed. Before that commit the "gpios" property had no > > effect and PHY reset was only handled in the bootloader. > > Wow that was a very interesting one. > Turning over these stones sometimes expose the weirdest > things. > > This is one of the reasons why we need formal validation of > DTS files: things look OK to the human eye and get merged > but they are just syntactically wrong. > > I guess I could hack to make "gpios" be ignored by the > regulator GPIO quirks in gpiolib, but I take it you probably > prefer to fix up the real issue like this. Maybe you could document in fixed-regulator bindings that it now supports the standard "gpios" property as well? There still seems to be a restriction that the list must contain a single element. Anyway, what my patch fixes is a board-specific issue which exists even outside your patch: the enet phy regulator never worked as expected. -- Regards, Leonard
On Wed, Oct 3, 2018 at 1:49 PM Leonard Crestez <leonard.crestez@nxp.com> wrote: > On Tue, 2018-10-02 at 21:56 +0200, Linus Walleij wrote: > > I guess I could hack to make "gpios" be ignored by the > > regulator GPIO quirks in gpiolib, but I take it you probably > > prefer to fix up the real issue like this. > > Maybe you could document in fixed-regulator bindings that it now > supports the standard "gpios" property as well? There still seems to be > a restriction that the list must contain a single element. OK you're right, fixing the docs to be more standards compliant is indeed better. > Anyway, what my patch fixes is a board-specific issue which exists even > outside your patch: the enet phy regulator never worked as expected. Indeed. Yours, Linus Walleij
diff --git a/arch/arm/boot/dts/imx6sx-sdb.dtsi b/arch/arm/boot/dts/imx6sx-sdb.dtsi index 53b3408b5fab..1ed7a081c420 100644 --- a/arch/arm/boot/dts/imx6sx-sdb.dtsi +++ b/arch/arm/boot/dts/imx6sx-sdb.dtsi @@ -115,11 +115,11 @@ pinctrl-names = "default"; pinctrl-0 = <&pinctrl_enet_3v3>; regulator-name = "enet_3v3"; regulator-min-microvolt = <3300000>; regulator-max-microvolt = <3300000>; - gpios = <&gpio2 6 GPIO_ACTIVE_LOW>; + gpio = <&gpio2 6 GPIO_ACTIVE_LOW>; }; reg_pcie_gpio: regulator-pcie-gpio { compatible = "regulator-fixed"; pinctrl-names = "default"; @@ -178,10 +178,11 @@ pinctrl-names = "default"; pinctrl-0 = <&pinctrl_enet1>; phy-supply = <®_enet_3v3>; phy-mode = "rgmii"; phy-handle = <ðphy1>; + phy-reset-gpios = <&gpio2 7 GPIO_ACTIVE_LOW>; status = "okay"; mdio { #address-cells = <1>; #size-cells = <0>; @@ -371,10 +372,12 @@ MX6SX_PAD_RGMII1_RD1__ENET1_RX_DATA_1 0x3081 MX6SX_PAD_RGMII1_RD2__ENET1_RX_DATA_2 0x3081 MX6SX_PAD_RGMII1_RD3__ENET1_RX_DATA_3 0x3081 MX6SX_PAD_RGMII1_RX_CTL__ENET1_RX_EN 0x3081 MX6SX_PAD_ENET2_RX_CLK__ENET2_REF_CLK_25M 0x91 + /* phy reset */ + MX6SX_PAD_ENET2_CRS__GPIO2_IO_7 0x80000000 >; }; pinctrl_enet_3v3: enet3v3grp { fsl,pins = <
Bindings for "fixed-regulator" only explicitly support "gpio" property, not "gpios". Fix by correcting the property name. The enet PHYs on imx6sx-sdb needs to be explicitly reset after a power cycle, handle this by adding the phy-reset-gpios property. Both phys share a single reset, a scenario similar to imx7d-sdb. This issue was exposed by commit efdfeb079cc3 ("regulator: fixed: Convert to use GPIO descriptor only") which causes the "gpios" property to also be parsed. Before that commit the "gpios" property had no effect and PHY reset was only handled in the bootloader. This fixes linux-next netboot failures previously reported here: https://lore.kernel.org/patchwork/patch/982437/#1177900 https://lore.kernel.org/patchwork/patch/994091/#1178304 Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com> --- arch/arm/boot/dts/imx6sx-sdb.dtsi | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) I'm not entirely certain sharing the phy reset is OK, maybe somebody with more familiarity with the FEC driver can confirm? This design seems to be similar between multiple imx boards with dual ethernet ports. This took a surprising amount of digging.