[Ocfs2-test-devel] [PATCH] Test zeros in holes while writing on page boundaries

[Joel Becker]

On Mon, Feb 28, 2011 at 12:27:14PM -0600, Goldwyn Rodrigues wrote:
> On Mon, Feb 28, 2011 at 10:04:23AM -0800, Joel Becker wrote:
> > On Thu, Feb 24, 2011 at 10:55:17AM -0600, Goldwyn Rodrigues wrote:
> > > This checks for zeros while writing at page boundaries and lseeking. The
> > > precise conditions are:
> > > 
> > > 1. blocksize < clustersize
> > > 2.  the previous write ended in a page boundary, but not a cluster
> > > boundary. eg, clustersize=16K but pos is at the page boundary = 4k
> > > 3. you wrote something at the page boundary
> > > 4. there is a hole created in the page from 4k-8k in *between* say
> > > 4128-4640
> > > 5. i_size has moved past the second page, or past 8192
> > > 
> > > Sorry, my expected file is a bit too large because this is more
> > > probabilistic rather than deterministic.
> > 
> > 	Like I stated in my other response, this should be evidenced
> > directly by one test, not only sometimes, right?
> 
> This *is* evidenced directly by one test, and fails everytime I execute
> the test.

	I mean writing at one location, not one run of your current
write-at-64-locations test.  Each of your locations in the seq 0 64 is a
test in and of itself.

Joel

-- 

"Sometimes when reading Goethe I have the paralyzing suspicion
 that he is trying to be funny."
         - Guy Davenport

			http://www.jlbec.org/
			jlbec at evilplan.org