Strange behavior from cu on armv7

[bob prohaska]

On Mon, Feb 26, 2018 at 11:31:21AM -0700, Ian Lepore wrote:
> On Mon, 2018-02-26 at 09:03 -0800, bob prohaska wrote:
> > Lately cu sessions (or something closely-related) have taken to
> > misbehaving on a Pi2 running -current.
> > 
> > The situation is:
> > 
> > ssh into armv7 host which has a pl2303 plugged into it
> > su to root
> > run cu -l cuaU0 -s 115200 to connect to the serial port of a Pi3??
> > 
> > Most of the time the connection behaves normally. The pl2303 is prone
> > to locking up over time, seemingly faster (an hour) when the armv7 host
> > is loaded. In that case, unplugging and replugging the pl2303 aborts
> > the cu connection, which can then be restarted and again behaves normally.
> > 
> > Lately, the unplug-replug cycle produces the normal recognition and??
> > allows the cu session to be restarted, but no data is tranferred.
> > A "connected" prompt comes back, but there's no echo and no data.??
> > Further unplug-replug attempts don't change anything. A reboot of
> > the armv7 host restores normal behavior.??
> > 
> > In a couple of cases unprintable characters displayed:
> > 
> > root at www:/home/bob # cu -l cuaU0 -s 11520oo}root at www:/home/bob # cu -l cuaU0 -s 115200
> > Connected
> > [hex characters didn't copy/paste]
> > 
> > FreeBSD/arm64 (www.zefox.org) (ttyu0)
> > 
> > login: ???
> > 
> > FreeBSD/arm64 (www.zefox.org) (ttyu0)
> > 
> > login:??
> > 
> > And all is well, for now.
> > 
> > Thanks for reading, and any ideas.
> > 
> > bob prohaska
> 
> Hmm. ??I've noticed for the past week or two (maybe longer) that if I
> connect to a wandboard serial console via an ftdi usb-serial and cu,
> and then I do "stty size 24 140" I get exactly the symptom you
> describe... no response to ^C or ^T, and neither un/replugging the ftdi
> nor closing and reopening the cu connection makes data flow in either
> direction until the arm board is rebooted. ??Ssh connections to the
> board continue to work fine and the system appears to be running fine,
> it's only the serial console that's dead. ??Killing the getty on the
> console doesn't help either, a new non-responsive getty starts up
> immediately.
> 
> So, I guess all in all I have nothing much to offer here except a too-
> wordy "me too".
> 
Well, at least I'm in good company.....8-)

Thanks for writing,

bob prohaska