diff mbox series

[5/5] arm64: dts: meson: sei510: add network support

Message ID 20190510164940.13496-6-jbrunet@baylibre.com (mailing list archive)
State Superseded
Headers show
Series arm64: dts: meson: g12a: add network support | expand

Commit Message

Jerome Brunet May 10, 2019, 4:49 p.m. UTC
Enable the network interface of the SEI510 which use the internal PHY.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
---
 arch/arm64/boot/dts/amlogic/meson-g12a-sei510.dts | 7 +++++++
 1 file changed, 7 insertions(+)

Comments

Kevin Hilman May 10, 2019, 10:45 p.m. UTC | #1
Jerome Brunet <jbrunet@baylibre.com> writes:

> Enable the network interface of the SEI510 which use the internal PHY.
>
> Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>

I tried testing this series on SEI510, but I must still be missing some
defconfig options, as the default defconfig doesn't lead to a working
interface.


I tried adding this kconfig fragment[1], and the dwmac probes/inits but
I must still be missing something, as the dwmac is still failing to find
a PHY.  Boot log: https://termbin.com/ivf3

I have the same result testing on the u200.

Kevin

[1] amlogic network kconfig fragment
CONFIG_STMMAC_ETH=y

# following are needed, but automatically enabled if above is set
#CONFIG_STMMAC_PLATFORM=m
#CONFIG_DWMAC_MESON=m

CONFIG_PHYLIB=y
CONFIG_MICREL_PHY=y
CONFIG_REALTEK_PHY=y

CONFIG_MDIO_BUS_MUX_MESON_G12A=y
CONFIG_MESON_GXL_PHY=y
Jerome Brunet May 11, 2019, 3:46 p.m. UTC | #2
On Fri, 2019-05-10 at 15:45 -0700, Kevin Hilman wrote:
> Jerome Brunet <jbrunet@baylibre.com> writes:
> 
> > Enable the network interface of the SEI510 which use the internal PHY.
> > 
> > Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
> 
> I tried testing this series on SEI510, but I must still be missing some
> defconfig options, as the default defconfig doesn't lead to a working
> interface.

That's weird. AFAICT, the net part has hit Linus's tree.
You should have everything needed by default, the mdio mux has

> default m if ARCH_MESON

> 
> 
> I tried adding this kconfig fragment[1], and the dwmac probes/inits but
> I must still be missing something, as the dwmac is still failing to find
> a PHY.  Boot log: https://termbin.com/ivf3
> 
> I have the same result testing on the u200.

I don't any other patch pending for the network of the g12a.
Maybe I've done something wrong while rebasing. I'll check on monday.

> 
> Kevin
> 
> [1] amlogic network kconfig fragment
> CONFIG_STMMAC_ETH=y
> 
> # following are needed, but automatically enabled if above is set
> #CONFIG_STMMAC_PLATFORM=m
> #CONFIG_DWMAC_MESON=m
> 
> CONFIG_PHYLIB=y
> CONFIG_MICREL_PHY=y
> CONFIG_REALTEK_PHY=y
> 
> CONFIG_MDIO_BUS_MUX_MESON_G12A=y
> CONFIG_MESON_GXL_PHY=y
Martin Blumenstingl May 11, 2019, 5:16 p.m. UTC | #3
Hi Kevin,

On Sat, May 11, 2019 at 12:45 AM Kevin Hilman <khilman@baylibre.com> wrote:
>
> Jerome Brunet <jbrunet@baylibre.com> writes:
>
> > Enable the network interface of the SEI510 which use the internal PHY.
> >
> > Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
>
> I tried testing this series on SEI510, but I must still be missing some
> defconfig options, as the default defconfig doesn't lead to a working
> interface.
>
>
> I tried adding this kconfig fragment[1], and the dwmac probes/inits but
> I must still be missing something, as the dwmac is still failing to find
> a PHY.  Boot log: https://termbin.com/ivf3
>
> I have the same result testing on the u200.
I wonder if we're simply missing the pinctrl definitions in the ethmac node:
  pinctrl-0 = <&eth_rmii_pins>;
  pinctrl-names = "default";

I don't know how the SoC works internally but I am assuming that the
MDIO pins are routed to the "internal PHY" (within the chip).
also we need the eth_rmii_pins anyways for the RXD/TXD pins which are
connected to the physical Ethernet port on the board.
bonus question: while writing this email I'm surprised to see that on
GXL we don't use the rmii pins anywhere, why is Ethernet working fine
there?


Martin
Jerome Brunet May 11, 2019, 5:25 p.m. UTC | #4
On Sat, 2019-05-11 at 19:16 +0200, Martin Blumenstingl wrote:
> Hi Kevin,
> 
> On Sat, May 11, 2019 at 12:45 AM Kevin Hilman <khilman@baylibre.com> wrote:
> > Jerome Brunet <jbrunet@baylibre.com> writes:
> > 
> > > Enable the network interface of the SEI510 which use the internal PHY.
> > > 
> > > Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
> > 
> > I tried testing this series on SEI510, but I must still be missing some
> > defconfig options, as the default defconfig doesn't lead to a working
> > interface.
> > 
> > 
> > I tried adding this kconfig fragment[1], and the dwmac probes/inits but
> > I must still be missing something, as the dwmac is still failing to find
> > a PHY.  Boot log: https://termbin.com/ivf3
> > 
> > I have the same result testing on the u200.
> I wonder if we're simply missing the pinctrl definitions in the ethmac node:
>   pinctrl-0 = <&eth_rmii_pins>;
>   pinctrl-names = "default";
> 
> I don't know how the SoC works internally but I am assuming that the
> MDIO pins are routed to the "internal PHY" (within the chip).
> also we need the eth_rmii_pins anyways for the RXD/TXD pins which are
> connected to the physical Ethernet port on the board.
> bonus question: while writing this email I'm surprised to see that on
> GXL we don't use the rmii pins anywhere, why is Ethernet working fine
> there?

AFAIK, the pinmux is for the external pad Martin
The internal phy does not use those pads.

> 
> 
> Martin
Kevin Hilman May 13, 2019, 4:31 p.m. UTC | #5
Jerome Brunet <jbrunet@baylibre.com> writes:

> On Fri, 2019-05-10 at 15:45 -0700, Kevin Hilman wrote:
>> Jerome Brunet <jbrunet@baylibre.com> writes:
>> 
>> > Enable the network interface of the SEI510 which use the internal PHY.
>> > 
>> > Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
>> 
>> I tried testing this series on SEI510, but I must still be missing some
>> defconfig options, as the default defconfig doesn't lead to a working
>> interface.
>
> That's weird. AFAICT, the net part has hit Linus's tree.

Ah, that's the key ingredient.  I was testing with a v5.1 baseline (not
linus/master) along with all my queued up changes.  I just pulled in
linus/master to test this series, and indeed, everything works, and I
don't need my config fragment either.

Thanks for clarifying.

If repost needed, feel free to add

Tested-by: Kevin Hilman <khilman@baylibre.com>

Kevin
diff mbox series

Patch

diff --git a/arch/arm64/boot/dts/amlogic/meson-g12a-sei510.dts b/arch/arm64/boot/dts/amlogic/meson-g12a-sei510.dts
index 61fb30047d7f..4a785b17c1af 100644
--- a/arch/arm64/boot/dts/amlogic/meson-g12a-sei510.dts
+++ b/arch/arm64/boot/dts/amlogic/meson-g12a-sei510.dts
@@ -29,6 +29,7 @@ 
 
 	aliases {
 		serial0 = &uart_AO;
+		ethernet0 = &ethmac;
 	};
 
 	chosen {
@@ -144,6 +145,12 @@ 
 	};
 };
 
+&ethmac {
+	status = "okay";
+	phy-handle = <&internal_ephy>;
+	phy-mode = "rmii";
+};
+
 &hdmi_tx {
 	status = "okay";
 	pinctrl-0 = <&hdmitx_hpd_pins>, <&hdmitx_ddc_pins>;