DomainKey-Signature: a=rsa-sha1; c=nofws; d=sourceware.org; h=list-id
	:list-unsubscribe:list-subscribe:list-archive:list-post
	:list-help:sender:date:from:to:subject:message-id:mime-version
	:content-type; q=dns; s=default; b=rMAc1dWeozWacuE8okYdLOwvwWEX3
	sLcJFyo+YMe4r4DcjPRr5In53EZg4UC5rLLgKYAIr39zXE60V/qhLN8ucKMh2tSl
	tGubcIDnPlilJuvBcg8bhXD0sZBNcm5C9ENBE64PtJ2oZnX2ZJUi2XzLum5Y4MDY
	yMSpRRUn6YIZyw=
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Date: Sat, 26 Aug 2017 22:52:24 +0100
From: Adam Dinwoodie <adam AT dinwoodie DOT org>
To: cygwin AT cygwin DOT com
Subject: fflush(NULL) empties stdin
Message-ID: <20170826215224.GD10378@dinwoodie.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.5.24 (2015-08-30)

Hello,

I seem to have found a bug in Cygwin's fflush implementation, where
fflush(NULL) is unexpectedly flushing stdin.

Working through https://cygwin.com/snapshots/ with the STC below, this
behaviour does not appear in the 2017-03-08 snapshot, and 2017-03-10 is
the first snapshot in which this does appear, with thanks to Ramsay
Jones on the Git mailing list for initially pointing me at the change
coming in some time between 2.7.0 and 2.8.0.

With thanks to Jeff King on the Git mailing list, here's a simple test
case:

    $ cat a.c
    #include <stdio.h>
    
    int main(void)
    {
            char buf[256];
            while (fgets(buf, sizeof(buf), stdin)) {
                    fprintf(stdout, "got: %s", buf);
                    fflush(NULL);
            }
            return 0;
    }
    
    $ gcc a.c
    
    $ seq 10 | ./a.exe
    got: 1

Compare this to the expected output, which I see on my handy CentOS 6
box:

    $ seq 10 | ./a.out
    got: 1
    got: 2
    got: 3
    got: 4
    got: 5
    got: 6
    got: 7
    got: 8
    got: 9
    got: 10

By my reading of Cygwin's fflush(3p), the stdin stream should be
unaffected by a fflush(NULL), as it is neither an output stream, an
update stream, nor a file capable of seeking.

I originally noticed this behaviour due to it causing a failure in one
of the Git test scripts.

Cheers,

Adam

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple