Firefly Open Source Community

Title: Core-3308Y MAC configuration [Print This Page]
Author: ivan.p    Time: 8/22/2023 06:39
Title: Core-3308Y MAC configuration
Last edited by ivan.p In 8/22/2023 06:44 Editor

Hi everyone!

We experience an issue with MAC configuration using a custom Device Tree. The circuit is without external clock generation (MAC_CLK is expected to be provided by RK3308B).
An entry which is used in dts:
&mac {
        phy-supply = <&vcc_phy>;
        assigned-clocks = <&cru SCLK_MAC>;
        assigned-clock-parents = <&mac_clkin>;
        clock_in_out = "output";
        pinctrl-names = "default";
        pinctrl-0 = <&rmii_pins &mac_refclk_12ma>;
        //snps,reset-gpio = <&gpio0 RK_PB2 GPIO_ACTIVE_LOW>;
        //snps,reset-active-low;
        //snps,reset-delays-us = <0 50000 50000>;
        status = "disable";
};




Parameter "status" is set to "okay" in an upper-level dts.
I suppose that at least "assigned-clock-parents" parameter must have another value, but I could not find any appropriate variants in rk3308.dtsi.

The picture that is observed on a scope: MAC_CLK signal appears for a second or two, then is turned off. The signal appears even when the parameter "clock_in_out" is set to "input". Nothing can be seen on TXD0, TXD1.


We have tried three cases.
1. Clean module on a custom baseboard, default device tree (rmii-m1 signals are used). No PHY is detected according to the log:
[    1.430124] rk_gmac-dwmac ff4e0000.ethernet: clock input or output? (input).
[    1.430168] rk_gmac-dwmac ff4e0000.ethernet: Can not read property: tx_delay.
[    1.430190] rk_gmac-dwmac ff4e0000.ethernet: set tx_delay to 0x30
[    1.430205] rk_gmac-dwmac ff4e0000.ethernet: Can not read property: rx_delay.
[    1.430223] rk_gmac-dwmac ff4e0000.ethernet: set rx_delay to 0x10
[    1.430248] rk_gmac-dwmac ff4e0000.ethernet: integrated PHY? (no).
[    1.435583] rk_gmac-dwmac ff4e0000.ethernet: init for RMII
[    1.435739] stmmac - user ID: 0x10, Synopsys ID: 0x35
[    1.435750]  Ring mode enabled
[    1.435762]  DMA HW capability register supported
[    1.435778]  Normal descriptors
[    1.435791]  RX Checksum Offload Engine supported (type 2)
[    1.435801]  TX Checksum insertion supported
[    1.435811]  Wake-Up On Lan supported
[    1.435876]  Enable RX Mitigation via HW Watchdog Timer
[    1.438739] libphy: stmmac: probed
[    1.438758] eth%d: No PHY found


2. Module on a custom baseboard, with a custom device tree (&mac entry shown above, rmii signals are used) - the same picture.
3. Module without any baseboard, just 5V power and terminal connected, default device tree. It manages to find some PHYs! I suppose that this is due to the noise on unconnected pins.
[    0.198362] rk_gmac-dwmac ff4e0000.ethernet: clock input or output? (input).
[    0.198413] rk_gmac-dwmac ff4e0000.ethernet: Can not read property: tx_delay.
[    0.198431] rk_gmac-dwmac ff4e0000.ethernet: set tx_delay to 0x30
[    0.198455] rk_gmac-dwmac ff4e0000.ethernet: Can not read property: rx_delay.
[    0.198472] rk_gmac-dwmac ff4e0000.ethernet: set rx_delay to 0x10
[    0.198497] rk_gmac-dwmac ff4e0000.ethernet: integrated PHY? (no).
[    0.198802] rk_gmac-dwmac ff4e0000.ethernet: clock input from PHY
[    0.203840] rk_gmac-dwmac ff4e0000.ethernet: init for RMII
[    0.203997] stmmac - user ID: 0x10, Synopsys ID: 0x35
[    0.204014]  Ring mode enabled
[    0.204025]  DMA HW capability register supported
[    0.204034]  Normal descriptors
[    0.204046]  RX Checksum Offload Engine supported (type 2)
[    0.204056]  TX Checksum insertion supported
[    0.204065]  Wake-Up On Lan supported
[    0.204127]  Enable RX Mitigation via HW Watchdog Timer
[    0.215581] libphy: stmmac: probed
[    0.215625] eth%d: PHY ID 0000ffff at 0 IRQ POLL (stmmac-0:00) active
[    0.215642] eth%d: PHY ID 0000ffff at 1 IRQ POLL (stmmac-0:01)
[    0.215656] eth%d: PHY ID 0000ffff at 2 IRQ POLL (stmmac-0:02)
[    0.215669] eth%d: PHY ID 0000ffff at 3 IRQ POLL (stmmac-0:03)
[    0.215683] eth%d: PHY ID 0000ffff at 4 IRQ POLL (stmmac-0:04)
[    0.215696] eth%d: PHY ID 0000ffff at 5 IRQ POLL (stmmac-0:05)
[    0.215709] eth%d: PHY ID 0000ffff at 6 IRQ POLL (stmmac-0:06)
[    0.215765] eth%d: PHY ID 0000ffff at 7 IRQ POLL (stmmac-0:07)
[    0.215779] eth%d: PHY ID 0000ffff at 8 IRQ POLL (stmmac-0:08)
[    0.215792] eth%d: PHY ID 0000ffff at 9 IRQ POLL (stmmac-0:09)
[    0.215805] eth%d: PHY ID 0000ffff at 10 IRQ POLL (stmmac-0:0a)
[    0.215818] eth%d: PHY ID 0000ffff at 11 IRQ POLL (stmmac-0:0b)
[    0.215831] eth%d: PHY ID 0000ffff at 12 IRQ POLL (stmmac-0:0c)
[    0.215851] eth%d: PHY ID 0000ffff at 13 IRQ POLL (stmmac-0:0d)
[    0.215864] eth%d: PHY ID 0000ffff at 14 IRQ POLL (stmmac-0:0e)
[    0.215877] eth%d: PHY ID 0000ffff at 15 IRQ POLL (stmmac-0:0f)
[    0.215891] eth%d: PHY ID 0000ffff at 16 IRQ POLL (stmmac-0:10)
[    0.215903] eth%d: PHY ID 0000ffff at 17 IRQ POLL (stmmac-0:11)
[    0.215917] eth%d: PHY ID 0000ffff at 18 IRQ POLL (stmmac-0:12)
[    0.215930] eth%d: PHY ID 0000ffff at 19 IRQ POLL (stmmac-0:13)
[    0.215943] eth%d: PHY ID 0000ffff at 20 IRQ POLL (stmmac-0:14)
[    0.215956] eth%d: PHY ID 0000ffff at 21 IRQ POLL (stmmac-0:15)
[    0.215969] eth%d: PHY ID 0000ffff at 22 IRQ POLL (stmmac-0:16)
[    0.215982] eth%d: PHY ID 0000ffff at 23 IRQ POLL (stmmac-0:17)
[    0.215995] eth%d: PHY ID 0000ffff at 24 IRQ POLL (stmmac-0:18)
[    0.216008] eth%d: PHY ID 0000ffff at 25 IRQ POLL (stmmac-0:19)
[    0.216021] eth%d: PHY ID 0000ffff at 26 IRQ POLL (stmmac-0:1a)
[    0.216034] eth%d: PHY ID 0000ffff at 27 IRQ POLL (stmmac-0:1b)
[    0.216047] eth%d: PHY ID 0000ffff at 28 IRQ POLL (stmmac-0:1c)
[    0.216062] eth%d: PHY ID 0000ffff at 29 IRQ POLL (stmmac-0:1d)
[    0.216080] eth%d: PHY ID 0000ffff at 30 IRQ POLL (stmmac-0:1e)
[    0.216092] eth%d: PHY ID 0000ffff at 31 IRQ POLL (stmmac-0:1f)


In the last case (just a bare Core-3308Y)
cat /sys/kernel/debug/pinctrl/pinctrl/pinmux-pins
shows that the pins are configured properly:
pin 128 (gpio4-0): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 129 (gpio4-1): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 130 (gpio4-2): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 131 (gpio4-3): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 132 (gpio4-4): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 133 (gpio4-5): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 134 (gpio4-6): (MUX UNCLAIMED) (GPIO UNCLAIMED)
pin 135 (gpio4-7): (MUX UNCLAIMED) (GPIO UNCLAIMED)
pin 136 (gpio4-8): fiq-debugger (GPIO UNCLAIMED) function uart4 group uart4-xfer
pin 137 (gpio4-9): fiq-debugger (GPIO UNCLAIMED) function uart4 group uart4-xfer
pin 138 (gpio4-10): (MUX UNCLAIMED) gpio4:138
pin 139 (gpio4-11): (MUX UNCLAIMED) (GPIO UNCLAIMED)
pin 140 (gpio4-12): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group macm1-refclk-12ma
pin 141 (gpio4-13): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 142 (gpio4-14): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins
pin 143 (gpio4-15): ff4e0000.ethernet (GPIO UNCLAIMED) function gmac-m1 group rmiim1-pins


And in the first two cases (where no PHY is found) these pins are shown as unconfigured.

I can provide the whole DT source tree if it helps.

Thank you!



Author: 时间的蝴蝶    Time: 8/24/2023 19:33
1. It is recommended to use an external clock.

2. Internal clock reference:

&mac {
phy-supply = <&vcc_phy>;
  assigned-clocks = <&cru SCLK_MAC>;
  //assigned-clock-parents = <&mac_clkin>;
  clock_in_out = "output";
  pinctrl-names = "default";
  pinctrl-0 = <&rmiim_pins &macm_refclk>;
  snps,reset-gpio = <&gpio0 RK_PB2 GPIO_ACTIVE_LOW>;
  snps,reset-active-low;
  snps,reset-delays-us = <0 50000 50000>;
  status = "okay";
};
Author: ivan.p    Time: 8/25/2023 06:21
时间的蝴蝶 Posted at 8/24/2023 19:33
1. It is recommended to use an external clock.

2. Internal clock reference:

Thank you for the solution!

After my studying multiple reference schematics, it became clear that an external clock is the only  widespread variant.
Therefore we will modify our board schematic to use the generator integrated into PHY.
Author: hueljannie    Time: 1/3/2024 17:40
After my studying multiple reference schematics, it came clear that an external timepiece is the only widevariant.Therefore we will modify our board schematic to use the creator integrated into PHY.




Welcome Firefly Open Source Community (https://bbs.t-firefly.com/) Powered by Discuz! X3.1