svn commit: r294362 - in head/sys: dev/uart kern sys

[Marius Strobl]

On Tue, Jan 19, 2016 at 07:24:31PM -0700, Ian Lepore wrote:
> On Tue, 2016-01-19 at 23:34 +0000, Marius Strobl wrote:
> > Author: marius
> > Date: Tue Jan 19 23:34:27 2016
> > New Revision: 294362
> > URL: https://svnweb.freebsd.org/changeset/base/294362
> > 
> > Log:
> >   Fix tty_drain() and, thus, TIOCDRAIN of the current tty(4) incarnation
> >   to actually wait until the TX FIFOs of UARTs have be drained before
> >   returning. This is done by bringing the equivalent of the TS_BUSY flag
> >   found in the previous implementation back in an ABI-preserving way.
> >   Reported and tested by: Patrick Powell
> >   
> >   Most likely, drivers for USB-serial-adapters likewise incorporating
> >   TX FIFOs as well as other terminal devices that buffer output in some
> >   form should also provide implementations of tsw_busy.
> >   
> >   MFC after:> 	> 3 days
> > 
> > Modified:
> >   head/sys/dev/uart/uart_tty.c
> >   head/sys/kern/tty.c
> >   head/sys/sys/ttydevsw.h
> 
> This rocks.
> 
> It would rock even more if uart's sc_txbusy actually meant what its
> name implies, but for some hardware what it really indicates is "the
> hardware can/cannot accept more data" which is not the same as "all
> data has been transmitted."  Even for hardware that doesn't signal
> txidle until the fifo is completely empty (not just at a lowater mark),
> there may still be a last byte on the way out from a tx holding
> register.
> 

Yes, I'm aware that at least for uart_dev_ns8250.c, sc_txbusy
doesn't cover a character still being in the TX shift register
so far. I've started working on that but it turned out to be
non-trivial to get right.

Marius