[PATCH v2 2/2] ARM: dts: aspeed: Add Facebook Backpack-CMM BMC

Tao Ren taoren at fb.com
Thu Nov 8 11:46:34 AEDT 2018


On 11/7/18 4:24 PM, Andrew Jeffery wrote:

>> +&uart1 {
>> +	status = "okay";
>> +};
>> +
>> +&uart3 {
>> +	status = "okay";
>> +};
>> +
>> +&uart4 {
>> +	status = "okay";
>> +};
> 
> UARTs 1-4 should be explicitly pinmuxed. It's a bit fiddly as each of the UART lines has its own mux configuration, but it at least allows some flexibility (i.e. whether you just want Rx/Tx, or something more). See e.g. the uart1 node in arch/arm/boot/dts/aspeed-bmc-opp-romulus.dts.

Thank you for pointing it out and I will fix it.

> Speaking of pinctrl, as it's an AST2500 system you'll probably want to add the following to ensure you have full control of the pinmux configuration:
> 
> &pinctrl {
> 	aspeed,external-nodes = <&gfx &lhc>;
> };
> 
>> +
>> +&uart5 {
>> +	status = "okay";
>> +};
> 
> UART5 has fixed pins, so no pinctrl necessary here.

Got it. Thank you.


> You said this in the commit message:
> 
>> Note: I2C devices on Backpack Line Cards and Fabric Cards are not
>> listed in the device tree file because Line/Fabric Cards may be
>> unplugged.
> 
> It's not clear to me what this actually means, and I'm trying to reconcile it with enabling all of the i2c buses as I'm curious as to whether that's necessary.

Sorry for the confusion. In general, all the i2c buses are initialized in device tree, but i2c leaf devices are created from user space because these i2c devices may be unplugged at boot time (which causes driver_probe failures). Let me refine the commit message and will send out updated patch soon.

Thanks,
Tao Ren


More information about the openbmc mailing list