💬 Building a Raspberry Pi Gateway

[Velo17]

Hello everyone!

I am trying to use a NRF24 module on the SPI1 of a Raspberry2 (so that I can also use a touch tft on SPI0).

I enabled it in the config.txt so that is lists under /dev/spidev1.0 - for conveniency I configured the CS0 to be at GPIO16=Pin36 (by using dtoverlay=spi1-1cs,cs0_pin=16).

I use the dev branch of MySensors.git as the master branch does not recognize my Pi. I used this configure line:

./configure --my-rf24-channel=124 --my-rf24-irq-pin=29 --my-rf24-ce-pin=31 --my-rf24-cs-pin=36 --spi-driver=SPIDEV --spi-spidev-device=/dev/spidev1.0

successful make
but when I run the gateway in the shell with sudo ./bin/mysgw -d, I get the following:

mysgw: Starting gateway...
mysgw: Protocol version - 2.2.0-beta
 __  __       ____
|  \/  |_   _/ ___|  ___ _ __  ___  ___  _ __ ___
| |\/| | | | \___ \ / _ \ `_ \/ __|/ _ \| `__/ __|
| |  | | |_| |___| |  __/ | | \__ \  _  | |  \__ \
|_|  |_|\__, |____/ \___|_| |_|___/\___/|_|  |___/
        |___/                      2.2.0-beta

mysgw: MCO:BGN:INIT CP=RNNG--Q-
mysgw: TSF:LRT:OK
mysgw: TSM:INIT
mysgw: TSF:WUR:MS=0
mysgw: pinMode: invalid pin: 29
mysgw: pinMode: invalid pin: 31
mysgw: pinMode: invalid pin: 36
mysgw: digitalWrite: invalid pin: 31
mysgw: digitalWrite: invalid pin: 36
mysgw: digitalPinToInterrupt: invalid pin: 29
mysgw: digitalPinToInterrupt: invalid pin: 29
mysgw: digitalWrite: invalid pin: 36

Any ideas what I can do to make it work?

[Velo17]

I think I might have found the root of the problem. The ./configure detects my Rpi2 as such (rpi2) but in the RPi.cpp in line 26 there is the pin to GPIO translation array and it is lacking translations for the high pin numbers - which is obviously wrong as on the RPi2 you have the extended GPIO connector with 40 pins...

const static int pin_to_gpio_rev1[41] = {-1, -1, -1, 0, -1, 1, -1, 4, 14, -1, 15, 17, 18, 21, -1, 22, 23, -1, 24, 10, -1, 9, 25, 11, 8, -1, 7, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 };
const static int pin_to_gpio_rev2[41] = {-1, -1, -1, 2, -1, 3, -1, 4, 14, -1, 15, 17, 18, 27, -1, 22, 23, -1, 24, 10, -1, 9, 25, 11, 8, -1, 7, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 };
const static int pin_to_gpio_rev3[41] = {-1, -1, -1, 2, -1, 3, -1, 4, 14, -1, 15, 17, 18, 27, -1, 22, 23, -1, 24, 10, -1, 9, 25, 11, 8, -1, 7, -1, -1, 5, -1, 6, 12, 13, -1, 19, 16, 26, 20, -1, 21 };

I copied the rev3 array to the rev2 array and compiled this and it seems to work in my case!

[wergeld]

@Velo17 Thank you! I had been pulling my hair out over this issue on installing a new RPi2 setup.

[mfalkvidd]

Adding link to the github issue (thanks for creating!) for future reference
https://github.com/mysensors/MySensors/issues/895

[Velo17]

I just noticed that mysgw is constantly consuming 95-100% CPU here. This even happens when I unplug the NRF24 module so that after starting mysgw basically nothing happens. Running development branch. Hmmmm. Any idea what to check?

[mfalkvidd]

@Velo17 strace might be able to help:

sudo strace ./bin/mysgw -d

Could you post your configure command? I'd like to try to reproduce the problem. I have a gateway with development from git as of about 2 weeks ago (a9418271) but I'm not using the nrf24. That gateway uses about 2% cpu on my Raspberry Pi 1. There have only been 6 commits since I built my gateway but I can update it and see if it starts acting up.

On my gateway, strace outputs loads and loads of these:

gettimeofday({1501098316, 668956}, NULL) = 0
ioctl(6, FIONREAD, [0])                 = 0
gettimeofday({1501098316, 671383}, NULL) = 0
gettimeofday({1501098316, 672447}, NULL) = 0
gettimeofday({1501098316, 673818}, NULL) = 0
nanosleep({0, 10000000}, NULL)          = 0
gettimeofday({1501098316, 685602}, NULL) = 0
getsockopt(6, SOL_TCP, TCP_INFO, "\1\0\0\0\0\7g\1P4\3\0@\234\0\0\250\5\0\0\30\2\0\0\0\0\0\0\0\0\0\0"..., [104]) = 0
getsockopt(6, SOL_TCP, TCP_INFO, "\1\0\0\0\0\7g\1P4\3\0@\234\0\0\250\5\0\0\30\2\0\0\0\0\0\0\0\0\0\0"..., [104]) = 0
gettimeofday({1501098316, 689179}, NULL) = 0
ioctl(6, FIONREAD, [0])                 = 0

Also, which git hash are you using? (git log -1)

Another thing to try could be to enable all the debug flags in MyConfig.h and see if they give you something interesting.

[mfalkvidd]

I updated my gateway to 10ef623 (latest from git). Still works ok.

[Velo17]

@mfalkvidd I make / make install a couple of times and maybe did a git pull inbetween as well - now everything is back to normal again. Thank you for giving me some dbug tips for the next issue :-)

My strace output looks similiar.

My configure line is

./configure --my-rf24-channel=124 --my-rf24-irq-pin=29 --my-rf24-ce-pin=31 --my-rf24-cs-pin=36 --spi-driver=SPIDEV --spi-spidev-device=/dev/spidev1.0 --my-debug=disable

[Sergio Rius]

Should be possible to connect two radio modules and run two instances at the same time?
It could be interesting to have a hat with nrf+rfm, all in one.

[mfalkvidd]

@Sergio-Rius it would be great if you could contribute to making that a reality.
An earlier discussion on the same topic: https://forum.mysensors.org/topic/4151/nrf24l01-and-rfm69-on-one-gw/

So far, people have found it easier to add a separate gateway for each transport.

[Sergio Rius]

@mfalkvidd thanks for guiding me on the starting point. Perhaps the raspberry can break the doble transport limitation.
I'll investigate in how to install two services at once. But it'll would be nice any advice on the device virtualization part.

[mfalkvidd]

Another thread: https://forum.mysensors.org/topic/2331/gateway-with-both-rfm69-and-nrf24l01/

[mfalkvidd]

Maybe it could be as simple as running two instances of mysgw, using different pins for the nrf and the rfm?

[Sergio Rius]

@mfalkvidd Maybe. I'm programmer, but not for Linux. My concern is the part of the service that creates the virtual device colliding between the instances.
I'll do a test in a virtual machine as soon as I have some time.
It should be something easy to test.

[mfalkvidd]

@Sergio-Rius by "virtual device", do you mean the com port? If that's the caee, it can be specified with --my-serial-pty=

Ot let one gateway be serial and the other ethernet.

[Sergio Rius]

@mfalkvidd yes. Should work. I admit I've not taken a look at the code nor I don't know how to install the service two times.
I'm used to having problems with programs that use static resources allocation. But I don't know if that's the case.

[gohan]

maybe you could also consider making 2 ethernet gateway one on port 5003 and the other on port 5004 so that you can use myscontroller to debug from remote.

[Matt]

Hi guys couple of questions, apologies if they have been covered elsewhere.
I am running latest dev branches of domoticz and MYS on a pi2.
At random times it seems the NRF24L01+ LNA/PA goes down, or MYS goes down and domotics stops receiving signals from all of the nodes at the same time. I have the interrupt feature enabled.
I have set the option under domoticz/hardware/data timeout to restart if no data received in five minutes but I dont think this is an ideal solution, and have yet to find out if this domoticz feature actually works in my scenario.
Any idea what causes this? I found an old thread over on the domoticz/mysensors forum but no answers there. I am using a new sd card class 10.
What I want to do is try out the different MYS builds on domoticz ie MQTT and SERIAL as well as ethernet which I am currently running. However I'm not sure how to stop the current MYS service and completely clean up the current build to start from scratch... I have tried just re entering the ./configure commands then 'make' but this does not seem to work.
Many thanks,
Matt

[mfalkvidd]

@Matt log files for the time where the problem occurred will be essential for troubleshooting.

make uninstall will stop the service and uninstall
make cleanconfig will remove previous ./configure settings
make clean will remove compiled code

[enlo]

@ccy that correction to the detect_machine function was necessary on my Rpi 3 too (I took the latest version from master). I got it working with that addition.
Soooo I guess that would be worth adding to the code / opening an issue on github, wouldn't it?