can't send G commands after FIRMWARE_RESTART

[rafaljot]

after FIRMWARE_RESTART I can't send G commands to printer from dwc2 web interface. I see nothing special in log, LCD works,

I tried to debug and I think the problem is with

self.reactor.register_callback(self.gcode_reactor_callback)

because this callback never triggers after FIRMWARE_RESTART

Do you have the same issue ?

[Stephan3]

just tested it her on my local machine and it works. So dont know what to say. Is your firmware_restart successfull in that case? The reactor callback works in any situation.

[rafaljot]

I put logging before this line and inside callback. And yes, the rest of Klipper seems to be all right after restart. So I will continue the investigation.

[rafaljot]

I think the problem in reactor.py here:

class ReactorCallback:
def __init__(self, reactor, callback, waketime):
        self.reactor = reactor
        self.timer = reactor.register_timer(self.invoke, waketime)
        self.callback = callback
    def invoke(self, eventtime):
        self.reactor.unregister_timer(self.timer)
        self.callback(eventtime)
        return self.reactor.NEVER

self.callback = callback is after register_timer. In fast multicores it can fire invoke before setting callback

[Stephan3]

Hmm i tried that meanwhile on my i7 6700. There it works. You can try to debug to put inside the callback

logging.error(“this is callback”)

So you can know that it was called. As I am not able to reproduce this, it’s hard to help here

[rafaljot]

I did it. It is not called after restart any more.

I have two environments

1. PyCharm on Ubuntu (vmWare) + trigorilla board
2. BananaPi Zero (4cores) + EinsyRambo

[rafaljot]

unfortunately in step-by-step debuggingi it works.

[Stephan3]

hm hm dafuq why does it work for me :/ Just tried a arch vm and Debian9 live in vm. Both did with Arduino Mega and smoothie.

That’s what’s here.

Do you get a reply (with logging)in reactor.py when this error occurred? May be I have an idea

[rafaljot]

I added some extra loggins and traps: web_dwc2_and_reactor.zip

First restart: INFO:root:DWC2:31850.92 - 127.0.0.1 - GET - /rr_gcode?gcode=RESTART ERROR:root:#1 rr_gcode ERROR:root:#5 ReactorCallback ERROR:root:#7 register_timer ERROR:root:#6 _check_timers ERROR:root:#6 ReactorCallback:invoke ERROR:root:#8 unregister_timer ERROR:root:#3 gcode_reactor_callback ERROR:root:#4 gcode_reactor_callback

Second Restart: INFO:root:DWC2:31939.93 - 127.0.0.1 - GET - /rr_gcode?gcode=RESTART ERROR:root:#1 rr_gcode ERROR:root:#5 ReactorCallback ERROR:root:#7 register_timer INFO:root:Stats 10148.8: gcodein=0 mcu: mcu_awake=0.006 mcu_task_avg=0.000078 mcu_task_stddev=0.000069 bytes_write=1828 bytes_read=8984 bytes_retransmit=0 bytes_invalid=4 send_seq=200 receive_seq=200 retransmit_seq=0 srtt=0.003 rttvar=0.001 rto=0.025 ready_bytes=0 stalled_bytes=0 freq=16003009 print_time=1.632 buffer_time=0.000 print_stall=0 extruder: target=0 temp=1208.0 pwm=0.000

[rafaljot]

It looks like I have two instances of Reactor and two web_dwc2 after RESTART So rr_gcode registers callbacks into old one

[Stephan3]

Okay looks promising. can you try to add theese:

        self.printer = config.get_printer()
        self.reactor = self.printer.get_reactor(

in line 81 in function handle ready for me? Then try to break it

[rafaljot]

nothing new. How to reproduce issue:

1. send command m117 #1 ;you will see #1 on LCD - if you have LCD
2. send RESTART ; #1 changes for "Ready"
3. send command m117 #2 ;you will see #2 on LCD - if you have LCD
4. send RESTART ; #2 changes for "Ready" etc.

step 1 and 2 always works.

3,4. etc sometime works sometimes not.
For sure I see two reactor.self ids in the same time

you can examine it by adding:

import threading logging.error('#threading.active_count:'+ str(threading.active_count()))

it increases by 1 with each RESTART

[rafaljot]

I changed

tornado.ioloop.IOLoop.instance().start()

to:

self.ioloop = tornado.ioloop.IOLoop.current()
self.ioloop.start()

and added:

self.ioloop.stop()

before:

self.http_server.stop()

It works :)

[Stephan3]

As i adopted this, i am going to close this. Thanks