cp command - problem with sparse [sparse file support under NTFS (Win2k)]

[Corinna Vinschen]

On Feb  2 17:18, Eric Blake wrote:
> > I am not sure whether this is a 128 k problem.  I tested with a file that
> > had 1 MB of data at the beginning and another 1 MB of data at the end.  In
> > between there were some 18 MB of zeroes (checked this with a hex-editor).
> 
> I did the following test on Win2000 SP4, using cygwin 1.5.12 and coreutils 
> 5.3.0-2 (actually, I strace'd the following calls to more closely track what 
> syscalls were happening).
> $ dd bs=1 seek=128K of=t < /dev/null 2> /dev/null
> $ cp --sparse=always t u
> $ cp u v
> 
> Since fsutil is not present except on win2003 or XP, I used an alternative 
> method to check whether a file is sparse: opening up Windows explorer, and 

ls -s works, too.  The size given is the actual size on disk in KB.  For
sparse files you'll see it's lower than the filesize.

> However, it points out a related bug in cygwin.  cp uses lseek() followed by 
> write(), so cygwin made u sparse because the write was at 128k beyond the end 
> of the current file size.  But dd uses lseek() followed by ftruncate(), and 
> cygwin just moved the end of the file without checking whether the file
> should 
> be made sparse.  The logic in write() needs to be copied to truncate() and 
> ftruncate() to allow dd to create sparse files in the first place.

Thanks for the hint.  I'll fix it at one point.

Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          mailto:cygwin@cygwin.com
Red Hat, Inc.

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