Re: Bash uses lseek while reading from serial device

[Corinna Vinschen <corinna-cygwin@cygwin.com>]

[-- Attachment #1: Type: text/plain, Size: 1818 bytes --]

On Aug 18 22:39, Ross Ridge wrote:
> Corinna Vinschen writes:
> > On Linux isatty on a descriptor connected to serial line returns 0,
> > on Cygwin it returned 1 so far.  I fixed both problems here, isatty
> > on a serial line returns 0 now, and lseek on serial (and, FWIW,
> > sockets) don't simply return 0 anymore, but rather -1 with errno set
> > to ESPIPE, as on Linux.
> 
> I'm not sure if Chet Ramey's suggestion that if isatty() returns 1
> then bash is allowed to assume reads are newline-delimited is correct.
> On Unix this would only be true if cannonical mode input processing was
> enabled (icanon), and Cygwin stty reports that this mode is disabled
> (-icanon) on serial devices.  Or at least it used to, with the snapshot
> DLL it now complains "/dev/ttyS0: Inappropriate ioctl for device".
> 
> For what its worth my tests on Linux shows that isatty() returns 1 on
> a serial device, namely /dev/ttyS0.  Which is what I would expect given
> that serial devices have traditionally been synonymous with ttys on Unix.

I reverted this part of the patch.  My test was flawed.  The problem is
that on Linux, /dev/ttyS[0-3] exists even if they are not backed by
hardware.  In this case, as is on my local machine, isatty returns 0.
However, if there's a device connected to /dev/ttySx, isatty returns 1.

So I just applied a patch which reverts the isatty status to /dev/ttySx
and thus, stty works as before.

> > Ross, please give it a try.
> 
> The snapshot DLL solves the bug and the script runs without any data
> being lost.  Thanks for looking into this.

Thanks for testing and discussing this problem.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 819 bytes --]