| Message ID | 20210304114013.18494-1-qiangqing.zhang@nxp.com |
|---|---|
| State | New |
| Headers | show |
| Series | [V1] arm64: dts: imx8mp: fix FEC can't work when attached to generic phy driver | expand |
Kindly pinging... Best Regards, Joakim Zhang > -----Original Message----- > From: Joakim Zhang <qiangqing.zhang@nxp.com> > Sent: 2021年3月4日 19:40 > To: robh+dt@kernel.org; shawnguo@kernel.org; s.hauer@pengutronix.de; > festevam@gmail.com > Cc: kernel@pengutronix.de; dl-linux-imx <linux-imx@nxp.com>; > devicetree@vger.kernel.org; linux-arm-kernel@lists.infradead.org; > linux-kernel@vger.kernel.org > Subject: [PATCH V1] arm64: dts: imx8mp: fix FEC can't work when attached to > generic phy driver > > Some users report that FEC can't work on i.MX8MP EVK board, it brings > inconvenience. The root cause should be FEC controller attached to generic phy > driver, as Realtek phy driver is built as module in the defconfig file > (CONFIG_REALTEK_PHY=m), so it is unavailable. If provide "reset-gpios" > property, it will reset phy when probed, and no way to re-config phy since we > use the generic phy dirver, which leads FEC can't work. > > There are two ways to let FEC work: > > 1. If you want to use generic phy dirver, please delete "reset-gpios" > property, keep power-on strapping pins configurations. > > 2. If you want to use Realtek phy driver, please buildin driver > (CONFIG_REALTEK_PHY=y), and had better add another two reset > properties: > reset-assert-us = <20000>; > reset-deassert-us = <150000>; > According to RTL8211 serials PHY datasheet, for a complete PHY reset, reset > pin must be asserted low for at least 10ms for internal regulator. > Wait for at least 72ms (for internal circuits settling time) before accessing the > PHY register. > > This patch selects method 1, since users may waste time to find out FEC failure, > in most cases, they just want to use networking to debug other modules. > > Fixs: commit 9e847693c6f34 ("arm64: dts: freescale: Add i.MX8MP EVK board > support") > Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com> > --- > arch/arm64/boot/dts/freescale/imx8mp-evk.dts | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > index 7db4273cc88b..4f5c2fb33eda 100644 > --- a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > +++ b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > @@ -97,7 +97,6 @@ > compatible = "ethernet-phy-ieee802.3-c22"; > reg = <1>; > eee-broken-1000t; > - reset-gpios = <&gpio4 2 GPIO_ACTIVE_LOW>; > }; > }; > }; > -- > 2.17.1
On Thu, Mar 04, 2021 at 07:40:13PM +0800, Joakim Zhang wrote: > Some users report that FEC can't work on i.MX8MP EVK board, it brings > inconvenience. The root cause should be FEC controller attached to > generic phy driver, as Realtek phy driver is built as module in the > defconfig file (CONFIG_REALTEK_PHY=m), so it is unavailable. If provide > "reset-gpios" property, it will reset phy when probed, and no way to > re-config phy since we use the generic phy dirver, which leads FEC can't > work. > > There are two ways to let FEC work: > > 1. If you want to use generic phy dirver, please delete "reset-gpios" > property, keep power-on strapping pins configurations. > > 2. If you want to use Realtek phy driver, please buildin driver > (CONFIG_REALTEK_PHY=y), and had better add another two reset > properties: > reset-assert-us = <20000>; > reset-deassert-us = <150000>; > According to RTL8211 serials PHY datasheet, for a complete PHY reset, > reset pin must be asserted low for at least 10ms for internal regulator. > Wait for at least 72ms (for internal circuits settling time) before > accessing the PHY register. > > This patch selects method 1, since users may waste time to find out FEC > failure, in most cases, they just want to use networking to debug other > modules. > > Fixs: commit 9e847693c6f34 ("arm64: dts: freescale: Add i.MX8MP EVK board support") > Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com> > --- > arch/arm64/boot/dts/freescale/imx8mp-evk.dts | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > index 7db4273cc88b..4f5c2fb33eda 100644 > --- a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > +++ b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > @@ -97,7 +97,6 @@ > compatible = "ethernet-phy-ieee802.3-c22"; > reg = <1>; > eee-broken-1000t; > - reset-gpios = <&gpio4 2 GPIO_ACTIVE_LOW>; Hmm, DT is describing hardware. If board schematic says there is a reset GPIO, we should have it. Shawn > }; > }; > }; > -- > 2.17.1 >
> -----Original Message----- > From: Shawn Guo <shawnguo@kernel.org> > Sent: 2021年3月18日 18:48 > To: Joakim Zhang <qiangqing.zhang@nxp.com> > Cc: robh+dt@kernel.org; s.hauer@pengutronix.de; festevam@gmail.com; > kernel@pengutronix.de; dl-linux-imx <linux-imx@nxp.com>; > devicetree@vger.kernel.org; linux-arm-kernel@lists.infradead.org; > linux-kernel@vger.kernel.org > Subject: Re: [PATCH V1] arm64: dts: imx8mp: fix FEC can't work when attached > to generic phy driver > > On Thu, Mar 04, 2021 at 07:40:13PM +0800, Joakim Zhang wrote: > > Some users report that FEC can't work on i.MX8MP EVK board, it brings > > inconvenience. The root cause should be FEC controller attached to > > generic phy driver, as Realtek phy driver is built as module in the > > defconfig file (CONFIG_REALTEK_PHY=m), so it is unavailable. If > > provide "reset-gpios" property, it will reset phy when probed, and no > > way to re-config phy since we use the generic phy dirver, which leads > > FEC can't work. > > > > There are two ways to let FEC work: > > > > 1. If you want to use generic phy dirver, please delete "reset-gpios" > > property, keep power-on strapping pins configurations. > > > > 2. If you want to use Realtek phy driver, please buildin driver > > (CONFIG_REALTEK_PHY=y), and had better add another two reset > > properties: > > reset-assert-us = <20000>; > > reset-deassert-us = <150000>; > > According to RTL8211 serials PHY datasheet, for a complete PHY reset, > > reset pin must be asserted low for at least 10ms for internal regulator. > > Wait for at least 72ms (for internal circuits settling time) before > > accessing the PHY register. > > > > This patch selects method 1, since users may waste time to find out > > FEC failure, in most cases, they just want to use networking to debug > > other modules. > > > > Fixs: commit 9e847693c6f34 ("arm64: dts: freescale: Add i.MX8MP EVK > > board support") > > Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com> > > --- > > arch/arm64/boot/dts/freescale/imx8mp-evk.dts | 1 - > > 1 file changed, 1 deletion(-) > > > > diff --git a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > > b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > > index 7db4273cc88b..4f5c2fb33eda 100644 > > --- a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > > +++ b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts > > @@ -97,7 +97,6 @@ > > compatible = "ethernet-phy-ieee802.3-c22"; > > reg = <1>; > > eee-broken-1000t; > > - reset-gpios = <&gpio4 2 GPIO_ACTIVE_LOW>; > > Hmm, DT is describing hardware. If board schematic says there is a reset > GPIO, we should have it. Hi Shawn, Seems you prefer to 2, is it possible to buildin Realtek PHY(CONFIG_REALTEK_PHY=y)? If not, it is going to be tricky. Best Regards, Joakim Zhang > Shawn > > > }; > > }; > > }; > > -- > > 2.17.1 > >
Hi Joakim, On Thu, Mar 18, 2021 at 8:04 AM Joakim Zhang <qiangqing.zhang@nxp.com> wrote: > Hi Shawn, > > Seems you prefer to 2, is it possible to buildin Realtek PHY(CONFIG_REALTEK_PHY=y)? If not, it is going to be tricky. Yes, we should select the driver as built-in and describe the reset timing requirements as per the datasheet. I have just sent two patches to address the issue you reported. Please test them.
diff --git a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts index 7db4273cc88b..4f5c2fb33eda 100644 --- a/arch/arm64/boot/dts/freescale/imx8mp-evk.dts +++ b/arch/arm64/boot/dts/freescale/imx8mp-evk.dts @@ -97,7 +97,6 @@ compatible = "ethernet-phy-ieee802.3-c22"; reg = <1>; eee-broken-1000t; - reset-gpios = <&gpio4 2 GPIO_ACTIVE_LOW>; }; }; };
Some users report that FEC can't work on i.MX8MP EVK board, it brings inconvenience. The root cause should be FEC controller attached to generic phy driver, as Realtek phy driver is built as module in the defconfig file (CONFIG_REALTEK_PHY=m), so it is unavailable. If provide "reset-gpios" property, it will reset phy when probed, and no way to re-config phy since we use the generic phy dirver, which leads FEC can't work. There are two ways to let FEC work: 1. If you want to use generic phy dirver, please delete "reset-gpios" property, keep power-on strapping pins configurations. 2. If you want to use Realtek phy driver, please buildin driver (CONFIG_REALTEK_PHY=y), and had better add another two reset properties: reset-assert-us = <20000>; reset-deassert-us = <150000>; According to RTL8211 serials PHY datasheet, for a complete PHY reset, reset pin must be asserted low for at least 10ms for internal regulator. Wait for at least 72ms (for internal circuits settling time) before accessing the PHY register. This patch selects method 1, since users may waste time to find out FEC failure, in most cases, they just want to use networking to debug other modules. Fixs: commit 9e847693c6f34 ("arm64: dts: freescale: Add i.MX8MP EVK board support") Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com> --- arch/arm64/boot/dts/freescale/imx8mp-evk.dts | 1 - 1 file changed, 1 deletion(-)