Add support clock mode for ESP32 ethernet module

[petrkr]

I've added optional parameter support for network.LAN clock_mode and implemented new enum string for network module class to support Clock mode for ESP32.

Can be used for some cases when especially LAN8720 modules uses clock source from ESP's pin GPIO17.

usage:

import machine
import network

lan = network.LAN(mdc = machine.Pin(23), mdio = machine.Pin(18), power=None, phy_type = network.PHY_LAN8720, phy_addr=1, clock_mode=network.ETH_CLOCK_GPIO17_OUT)
lan.active(1)

new enum at network module:

>>> for i in dir(network): print(i)
... 
ETH_CLOCK_GPIO0_IN
ETH_CLOCK_GPIO16_OUT
ETH_CLOCK_GPIO17_OUT
...

Also added catch events for ethernet module:

I (44868) ethernet: start
True
>>> I (48868) ethernet: LAN cable connected
I (49798) event: eth ip: 10.0.0.137, mask: 255.255.255.0, gw: 10.0.0.1
I (49798) ethernet: Got IP
I (64868) ethernet: LAN cable disconnected
I (72868) ethernet: LAN cable connected
I (73798) event: eth ip: 10.0.0.137, mask: 255.255.255.0, gw: 10.0.0.1
I (73798) ethernet: Got IP

should fix #4502

Later I think about do some events like wifi has. But for fix work with different clock modes should be this enough.

[petrkr]

Rebase to latest master

[dpgeorge]

Thanks, it looks really good!

One point: don't you think it's better to use the ESP-IDF default clock_mode rather than ETH_CLOCK_GPIO0_IN? In case the IDF changes it from GPIO0. So the default value of the clock_mode keyword arg would be "-1", which indicates to use the IDF's value. Then the code is something like:

if (args[ARG_clock_mode].u_int != -1) {
    // validate value
    config.clock_mode = args[ARG_clock_mode].u_int;
}

[petrkr]

Thanks, it looks really good!

One point: don't you think it's better to use the ESP-IDF default clock_mode rather than ETH_CLOCK_GPIO0_IN? In case the IDF changes it from GPIO0. So the default value of the clock_mode keyword arg would be "-1", which indicates to use the IDF's value. Then the code is something like:

if (args[ARG_clock_mode].u_int != -1) {
    // validate value
    config.clock_mode = args[ARG_clock_mode].u_int;
}

Hi,
yeah, you are right, I supposed default value from SDK was GPIO0_IN, so I used it, but yes, it's better to do not set it at all and let SDK to use default from phy_lan8720_default_ethernet_config call. I will commit change later today.

[petrkr]

One more question before I will test it on HW later... u_int does not stands for unsigned int so what about -1 value ?

[dpgeorge]

u_int does not stands for unsigned int so what about -1 value ?

The "u" stands for union, so it is a signed int. -1 is a good value to indicate default.

[petrkr]

u_int does not stands for unsigned int so what about -1 value ?

The "u" stands for union, so it is a signed int. -1 is a good value to indicate default.

I see. In that case, pushed change (with rebase to actual master)

[dpgeorge]

Thanks for updating. Merged in 7d8c71c through 5801a00 (the 3 commits related to adding clock_mode were squashed together).