[change] ModemManager device option needs to update to ctl_device

[ghost]

Hey,

Just letting you know, we are changing this from device to ctl_device upstream in the OpenWrt ModemManager protocol handler because of compatibility issues with DSA. I will link the PR here when it is in.

openwrt-openwisp-monitoring/openwrt-openwisp-monitoring/files/lib/openwisp_monitoring/interfaces.lua

Line 21 in 261239e

local modem = uci_cursor.get('network', interface['interface'], 'device')

[ghost]

How's this look? It looks like the uci_cursor.get returns nil if that entry doesn't exist, right?

        local compat_device = uci_cursor.get('network', interface['interface'], 'device')
        local device = uci_cursor.get('network', interface['interface'], 'ctl_device')
        local modem = (compat_device or device)

        if (compat_device and device) then
            modem = device
        end

        local info = {}

        local general = io.popen('mmcli --output-json -m '..modem):read("*a")

[nemesifier]

@nickberry17 yes, it should, I haven't tested the following modified version yet but if it works is a bit shorter but still readable:

local device = uci_cursor.get('network', interface['interface'], 'ctl_device') or \
               uci_cursor.get('network', interface['interface'], 'device')
local info = {}
local general = io.popen('mmcli --output-json -m '..device):read("*a")

The second lineand the or won't be executed in the future and can be removed at some point in the future.

[ghost]

Is it an inclusive or? Do we need to check if both are set, so we don't get device set to /sys/devices/platform/soc/8af8800.usb3/8a00000.dwc3/xhci-hcd.0.auto/usb2/2-1/sys/devices/platform/soc/8af8800.usb3/8a00000.dwc3/xhci-hcd.0.auto/usb2/2-1 In a sysupgrade scenario where settings are kept, both device and ctl_device could be set.

[nemesifier]

Is it an inclusive or? Do we need to check if both are set, so we don't get device set to /sys/devices/platform/soc/8af8800.usb3/8a00000.dwc3/xhci-hcd.0.auto/usb2/2-1/sys/devices/platform/soc/8af8800.usb3/8a00000.dwc3/xhci-hcd.0.auto/usb2/2-1 In a sysupgrade scenario where settings are kept, both device and ctl_device could be set.

The one with precedence should come first, the or is executed only if the first value is nil.

[ghost]

@nickberry17 yes, it should, I haven't tested the following modified version yet but if it works is a bit shorter but still readable:

local device = uci_cursor.get('network', interface['interface'], 'ctl_device') or \
               uci_cursor.get('network', interface['interface'], 'device')
local info = {}
local general = io.popen('mmcli --output-json -m '..device):read("*a")

The second lineand the or won't be executed in the future and can be removed at some point in the future.

Didn't like the \

uci: Entry not found
lua: /usr/sbin/netjson-monitoring:326: unexpected symbol near '\'

Putting them on the same line:

uci: Entry not found
lua: /usr/sbin/netjson-monitoring:330: attempt to concatenate global 'modem' (a nil value)
stack traceback:
	/usr/sbin/netjson-monitoring:330: in function 'specialized_info'
	/usr/sbin/netjson-monitoring:392: in function 'get_interface_info'
	/usr/sbin/netjson-monitoring:578: in main chunk
	[C]: ?

The longhand way works though

[devkapilbansal]

Putting them on the same line:

uci: Entry not found
lua: /usr/sbin/netjson-monitoring:330: attempt to concatenate global 'modem' (a nil value)
stack traceback:
	/usr/sbin/netjson-monitoring:330: in function 'specialized_info'
	/usr/sbin/netjson-monitoring:392: in function 'get_interface_info'
	/usr/sbin/netjson-monitoring:578: in main chunk
	[C]: ?

What's the modem here? I think it is renamed to device in other prior instances expect one giving error

[shwetd19]

Hey @nemesifier Here's how I think we can solve it :

1. Locate the Lua script where the device option is being used.
2. Replace device with ctl_device and add backward compatibility:

   local device = uci_cursor.get('network', interface['interface'], 'ctl_device') or
                  uci_cursor.get('network', interface['interface'], 'device')

3. after testing the changes to ensure the ModemManager functionality works as expected.

[nemesifier]

@shwetd19 sounds good but we'll need to test this with OpenWrt 24.10 on a router which has modem-manager, It may take some time before we can test this effectively, but it seems like an easy one so you can open the PR and then we can seek feedback.