UART high speed hang fixed - disabled wkup_m3!!

[Lee Armstrong]

Hi all,

I have been having an issue where when heavily using a serial UART at high speed I would get a lot of overrun errors on the port and eventually the BBB would hang/appear to power off.  It was very strange, we couldn't get kernel debugging at all, it did appear that is had powered down.

This morning I was making some changes to a device tree (attached) and managed to fix the problem, by disabling the wkup_m3 the problem appears to have disappeared.

Usually it would happen after around 2-3 hours and so far have uptime of 12+ hours on 2 test units I have on the bench so it is looking good.

Now....why is this the case.  UART 2 is the one heavily used and the baud rate is 3Mbps and we are pretty much saturating the link at times.  

• What exactly does the wkup_m3 do?
  
• Is it safe to keep it disabled?  (Older device trees didn't seem to use it)
  
• Where would this fault lie, hardware or software and who should I report it to?

I've been pulling my hair out for a week and thankfully appear to have a solution!!!

Happy Thanksgiving all.

Lee

[William Hermans]

http://free-electrons.com/kerneldoc/latest/devicetree/bindings/remoteproc/wkup_m3_rproc.txt

From the gist of things it seems to be part of the PRU power management code in relation to remoteproc.

--
For more options, visit http://beagleboard.org/discuss
---
You received this message because you are subscribed to the Google Groups "BeagleBoard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beagleboard...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Lee Armstrong]

Thanks William. If it is power related that would explain the end result potentially.

Now to try and work out why! I don't think leaving it disabled is a bad thing.

[William Hermans]

I was thinking that a possibility is that they're both peripherals, or device modules connected via the L4 interconnect. The PRU is for sure, and I would also think the UARTs sit on the L4 interconnect as well. Maybe I'm wrong.

On Sat, Nov 28, 2015 at 5:03 AM, Lee Armstrong <l...@pinkfroot.com> wrote:

Thanks William. If it is power related that would explain the end result potentially.

Now to try and work out why!  I don't think leaving it disabled is a bad thing.

[Lee Armstrong]

Yes that might make sense there, I've just looked at the docs myself.  

Weird behaviour though!