[PATCH] Fix minor breakage on Cygwin: auth-passwd.c andsession.c

[Darren Tucker]

Corinna Vinschen wrote:
> On Mon, Aug 11, 2003 at 07:04:05PM +0200, Gert Doering wrote:
> > Well, it's not a "break" but a "BREAK" signal.
> > [...]
> 
> Ok, but which functionality is ioctl(TIOCSBRK/TIOCCBRK)?  Does
> it reflect tcflow(TCOOFF/TCOON), tcflow(TCIOFF/TCION) or tcsendbreak()?

tcsendbreak.

I had assumed (and this seems to be the case) that TIOCSBRK/TIOCCBRK are
"set break flag" and "clear break flag".  I'll bet UARTs have a
corresponding flag in a register somewhere.

So: ioctl(TIOCSBRK) and the TD line on the port goes to +12V and stays
there until ioctl(TIOCCBRK) is called.

> Either way, it would be helpful to substitute the ioctl() by a more
> portable call instead of just ignoring it entirely if the ioctl()
> code isn't available.

The catch is the tcsendbreak can't (generally? portably?) specify the
duration of the break, you get a hard-coded delay.

So, we could do:

#if defined(TIOCSBRK) && defined(TIOCCBRK)
    /* use ioctls */
#else
# if defined(HAVE_TCSENDBREAK)
    /* use tcsendbreak */
# else
    return 0;
# endif
#endif

If you forget about specifying duration and just use tcsendbreak, it'd be
cleaner.  You can build a compatibility function using the ioctls where
necessary.

Also, tcsendbreak doesn't seem to block.

-- 
Darren Tucker (dtucker at zip.com.au)
GPG key 8FF4FA69 / D9A3 86E9 7EEE AF4B B2D4  37C9 C982 80C7 8FF4 FA69
    Good judgement comes with experience. Unfortunately, the experience
usually comes from bad judgement.