neuoy
commented
8 years ago Hi goial,
Thanks for your bug report.
Since you say your code is still running, this means the arduino did not crash or enter an infinite loop. The only thing I can imagine is that it is stuck doing nothing, with no timer or pin interrupt registered. How this can happen is unclear to me though. Since it's difficult to trace the sequence of events due to timing constraints (logging to the serial port is too slow), and even harder since the bug seems difficult to reproduce, I don't really know what the next move can be.
I don't have a RaspberryPi at hand to try to reproduce the issue. But I do have two arduinos (or is it "two arduini" ?) so I could try to reproduce the issue by tweaking a master library to have a longer reset (if I understand correctly you are suspecting that's the difference that make it fail with a Pi). But I really can't make any promise as to when that would be (like, maybe tonight, maybe in two years...).
Also, after reading my code again, I think I should make some charts about the various interrupt and timer sequences because currently it's hard to follow... Maybe the bug would be easier to find by reviewing the code carefully, than by trying to reproduce it, I don't know.
FYI, I'm using a DS9490R master adapter here (almost the cost of a RaspberryPi! but I bought it before the Pi even existed). And I never observed my arduino stuck pulling low, even after almost one year of continuous service (it's managing my boiler installation for house heating). Though, now you mention it, I did have to unplug/replug my master adapter on two occasions, for unknown reasons, maybe that was the same issue, hard to say. If that happens again I'll check the bus state before resetting it all.
Oh and one last thought : can you modify the Pi master library to change the reset duration? That could work around the issue for you, and also confirm the bug is triggered by a particular reset duration...
Hello @neuoy , I made a OneWire Slave with your Code and added some code for measurements. It worked very well over a long time(i think endless) with a Arduino as OneWire Master.
In my Project is a RaspberryPi as OneWire Master and that works too (Why not?). BUT the Slave get stuck some time. With the original OneWire Slaves, that does not happen. Without my code added to your "OneWireArduinoSlave" the Problem is the same. Only with a RaspberryPi as Master!
I tried to trigger the error, but I did not succeed. When the RaspberryPi only scans the bus for slaves, it worked without Problems for days (I tested it for ~30 hours). When the raspberry sent the slave a request (measure, read, write), enters the bug spontaneously. Sometimes after one request and sometimes after 100 requests. When i restart the slave it's working again for some time.
The bug looks like a Presence fail, because the Master pulled down the bus and release them after 480µS - 850µS. After the rising edge and 30-35µS the Slave pulled down the Bus and hold them for 296µS low. Then the magic begin:
My first Idea was the long reset pulse from master, but with various reset pulses from an arduino it worked too. The ArduinoSlave does not freeze when the bug appear, because my extra code run continues without problems when the fail happened.
Write me if I can do something for you to find the Issue (measurements, code changes... I think i can do every measurements what you want)! I will add later a picture of the bug.
greeting goial