change mode of dcan1_tx

[MartinVoelz]

Hello, the mode of P9_26_can_pin is not correct, it must not have receive-mode and it would be nice to add config-pin support to the pins P9_19 and P9_20 in the cape-universal for custom capes that do not use cape-eeprom. I made some posts to this on https://eewiki.net/display/linuxonarm/BeagleBone+Black+Comments?focusedCommentId=41418755#comment-41418755

diff --git a/cape-universal-00A0.dts b/cape-universal-00A0.dts index 2c3fca7..9fbcd44 100644 --- a/cape-universal-00A0.dts +++ b/cape-universal-00A0.dts @@ -552,7 +552,7 @@ P9_24_uart_pin: pinmux_P9_24_uart_pin { pinctrl-single,pins = <0x184 0x30>; }; /* Mode 0, Pull-Up, RxActive */ P9_24_can_pin: pinmux_P9_24_can_pin {

• pinctrl-single,pins = <0x184 0x32>; }; /* Mode 2, Pull-Up, RxActive */
• pinctrl-single,pins = <0x184 0x32>; }; /* Mode 2, dcan1rx, Pull-Up, RxActive / P9_24_i2c_pin: pinmux_P9_24_i2cpin { pinctrl-single,pins = <0x184 0x33>; }; / Mode 3, Pull-Up, RxActive _/ P9_24_pruin_pin: pinmux_P9_24_pruin_pin { @@ -586,7 +586,7 @@ P9_26_uart_pin: pinmux_P9_26_uartpin { pinctrl-single,pins = <0x180 0x30>; }; / Mode 0, Pull-Up, RxActive */ P9_26_can_pin: pinmux_P9_26_can_pin {
• pinctrl-single,pins = <0x180 0x32>; }; /* Mode 2, Pull-Up, RxActive */
• pinctrl-single,pins = <0x180 0x12>; }; /* Mode 2, dcan1tx, Pull-Up, TxActive / P9_26_i2c_pin: pinmux_P9_26_i2cpin { pinctrl-single,pins = <0x180 0x33>; }; / Mode 3, Pull-Up, RxActive */ P9_26_pruin_pin: pinmux_P9_26_pruin_pin {

[cdsteinkuehler]

Have you verified having the RxActive bit enabled causes problems with CAN? This bit does NOT control the direction of the I/O pin, it disables the pin's input buffer for minimal power consumption.

[MartinVoelz]

ok, you are right, it is also working with your setup 0x32 for both pins. Strange thing, I would have never come to the idea that RxActive means that Tx ist still possible.

[cdsteinkuehler]

I'm leaving things as-is for now, since the default for all pins in cape-universal is to leave the receiver active. This issue can be revisited if someone locates any sort of official recommendation from TI for pinmux settings for specific functions or if I ever test the theory that leaving the RxActive bit set lets the GPIO logic read the state of the pin when in special function modes.