> On Tue, Feb 27, 2001 at 09:15:29PM +0000, Richard Earnshaw wrote:
> > > - The file system is failing to correctly fill small holes left in files 
> > > on a seek() with zero.
> > 
> > Following up on this idea, I've written the following test program.  I 
> > believe the output of this should be a file consiting purely of 0's.  On 
> > my machine this is not the case and I get:
> 
> 	ah! Perhaps this is the source of the gzip problems I've been
> having.  Your script fails for me:
> NetBSD 1.5R (DIANA) #3: Wed Feb 14 23:13:01 CST 2001
> 
> However, you can simplify the test program: anything which seeks
> beyond the end of the file will cause this behavior.  I am
> quite reliably getting fragments of other files by writing
> one byte and seeking 4k beyond it.

I started off by always seeking to a 4k boundary and couldn't reproduce 
the problem.  Maybe it only occurs if the seek remains in the same block, 
or if the seek is to a non-block boundary.

R.