Some bugs - Githubissues

andyboeh commented 3 months ago

I saw that you created a PR for this device to get it into OpenWrt (great!), so I had a look at the improvements you've made to my daemon. In fact, I found a few bugs:

06f7bba doesn't make a difference: pollGpio is a boolean, the actual GPIO that is polled is defined in dns320l.h as GPIO_BUTTON_POWER https://github.com/wigyori/dns320l-daemon/blob/c74941880e0a8d2bc0344b3256b984397512e8c3/dns320l.h#L30C1-L30C33 - so it always was GPIO29.

I suppose that you don't get a proper GPIO reading for GPIO29 for one or all of the following:

The GPIO is read from sysfs. I don't know about the Marvell SoC but on other targets, the GPIO number in sysfs does not match the actual GPIO pin - so it's possibly not 29.
The Daemon does not export the GPIO. You have to take care yourself of this. The easiest way to achieve this by default would be to define a gpio-export node in the DTS.

And last but not least: Nowadays, I would remove or limit the socket server support. It doesn't have much benefits.

Great to see that this device is still round. Mine died a few years ago and I switched to a custom x86 system in the meantime.

wigyori commented 3 months ago

Hi Andy, great to see you here, thanks for the comments.

06f7bba - indeed, I probably didn't check the rest of the code and misunderstood this as "this is the GPIO to poll" - will fix this.
I think the GPIO number would match as this is on gpiochip0 (0-32), and gpiochip1 has only 16 GPIOs - thanks for the pointer that it needs to be exported "manually". I recall that I did try to export it and checked the value, but because I couldn't get any state changes on kicking the power button, I guess I ended up thinking the button is wired into the MCU instead.

root@OpenWrt:/sys/kernel/debug# cat gpio
gpiochip0: GPIOs 0-31, parent: platform/f1010100.gpio, f1010100.gpio:
 gpio-22  (dns320l:orange:r_hdd) out lo
 gpio-23  (dns320l:orange:l_hdd) out lo
 gpio-24  (regulators:regulator) out hi
 gpio-25  (dns320l:usb:blue    ) out lo
 gpio-26  (dns320l:usb:orange  ) out hi
 gpio-27  (USB unmount button  ) in  hi (act lo) - IRQ edge  (clear  )
 gpio-28  (Reset push button   ) in  hi (act lo) - IRQ edge  (clear  )

gpiochip1: GPIOs 32-49, parent: platform/f1010140.gpio, f1010140.gpio:

Yeah, this one is still alive, although the PSU is gone for good, and it was sitting for like 2 years with starting and re-starting working on it. It's also good to see that there is feedback on it. :)

andyboeh commented 1 month ago

Regarding the power GPIO, I found the following text on my website:

The power up / shutdown procedure of the device is as follows:

Upon startup, the OS sends DeviceReady to the MCU
the MCU stops blinking the power LED and does NOT immediately power down on power button press
Instead, long pressing the power button is registered again by the MCU and the Power LED starts blinking
A GPIO is pulled low that can be read by Linux and which triggers the shutdown process
Then again, DeviceShutdown is sent to the MCU to cut power

wigyori / dns320l-daemon

Some bugs #1