Re: [NFS] blocks of zeros (NULLs) in NFS files in kernels >= 2.6.20

[Aaron Straus <aaron@merfinllc.com>]

Hi,

On Sep 22 10:45 AM, Aaron Straus wrote:
> > diff --git a/fs/nfs/write.c b/fs/nfs/write.c
> > index 3229e21..eb6b211 100644
> > --- a/fs/nfs/write.c
> > +++ b/fs/nfs/write.c
> > @@ -1428,7 +1428,8 @@ static int nfs_write_mapping(struct address_space *mapping, int how)
> >  		.sync_mode = WB_SYNC_NONE,
> >  		.nr_to_write = LONG_MAX,
> >  		.for_writepages = 1,
> > -		.range_cyclic = 1,
> > +		.range_start = 0,
> > +		.range_end = LLONG_MAX,
> >  	};
> >  	int ret;
> >  
> 
> Also, I should test this, but I thought the VM was calling
> nfs_writepages directly i.e. not going through nfs_write_mapping.  Let
> me test with this patch.

Yes.  This patch doesn't seem to help.  The VM is calling nfs_writepages
directly.   I have debug statements in the nfs_write_mapping() call and it
hasn't been called when the hole appears.

It's the VM dirty page writeout that is happening out of order, *not*
nfs_wb_* stuff.   That's why we can see it so easily.  The dirty pages
are only scanned periodically.  In-between scans is when we see the
hole.

I think :).  I'm obviously not an expert here.

					Thanks,
					=a=


-- 
===================
Aaron Straus
aaron@merfinllc.com