Advanced Search

I was able to cause it to crash out without going into Linux CNC if i load it up too much with commands like
sudo chrt 99 ping -i .001

It handled running realtime at 1ms delay with averages of .101ms  for 10000, 20000 and 30000 pings, but when I jumped to 50000 pings it crashed out at about 40k pings..

So there is something that someone in the "chain" doesn't like under high network traffic..

and again, unplugging and plugging back in (just the network) fixes it... which is... wierd... (to me)..

Here is my Servo Period:
SERVO_PERIOD = 1000000

and here is my Servo Timeout
setp hm2_7i76e.0.watchdog.timeout_ns 5000000

the whole config can be found here (its where I wrote it a while ago, but my old pc died so working on getting it up and running on this pi)

github.com/Skyfire-community/SVM2-LinuxCNC

jhandel wrote:

Got a strange one with a Pi5...   Network seems fine when I am just pinging my Mesa card (a 7i76e) but when I run Linux CNC, things kick on but then the network crashes and I get a bazillion (technical term) hm2/hm2_7i76e.0: error queuing read iter=xxxx) and everything turns back off.. 

Once it dies I can't even ping the Mesa card (10.10.10.10) until I unplug and plug back in the ethernet cord... Then its happy as a clam pinging.. but once I start up Linux CNC again same crash right as the UI inits. 

(Linux CNC 2.9.3, PI5 8Gb, NVMe Boot)
 

You may want to start reading from here:
forum.linuxcnc.org/27-driver-boards/4223...read?start=20#207308

Unfortunately I don't have any ethernet hardware to test with (only SPI), I have a 7i92 but that's in the mill enclosure and is a PITA to access. From reports the 6.12 kernel on the most recent RPi images have better latency, but I can't say whether that would solve your issue.

Got a strange one with a Pi5...   Network seems fine when I am just pinging my Mesa card (a 7i76e) but when I run Linux CNC, things kick on but then the network crashes and I get a bazillion (technical term) hm2/hm2_7i76e.0: error queuing read iter=xxxx) and everything turns back off.. 

Once it dies I can't even ping the Mesa card (10.10.10.10) until I unplug and plug back in the ethernet cord... Then its happy as a clam pinging.. but once I start up Linux CNC again same crash right as the UI inits. 

(Linux CNC 2.9.3, PI5 8Gb, NVMe Boot)

I've recently updated from 2.8 to 2.9.2  and I'm having issues with the homing of my dual y axis that uses a single limit switch (i'll be adding a second one for auto squaring in the future). Since migrating the Y axis will go to home and them give a switch triggered while in home state 10 error. I've tried adjusting a number of the homing options but none of them resolve the issue. So I'm wondering if there been any changes since 2.8 that would require this setup to be configured differently.

NOTE: I am using a 7i76E and I've attached my config and hal file for reference.   

That looks correct (and indicates a 7I76+7I78 configuration)

all of my test nets stay low whether or not I apply 5v to the pins. 

That is expected as the input must be >60% of the field voltage to be recognized.

Here is 7i76e firmware for a 7I78 on P1 and 7I85 on P2: