Re: [PATCH 6/9] btrfs: calculate end of bio offset properly

[Omar Sandoval <osandov@osandov.com>]

On Wed, Nov 23, 2016 at 12:32:26PM +0800, Anand Jain wrote:
> 
> 
> On 11/23/16 12:21, Omar Sandoval wrote:
> > On Wed, Nov 23, 2016 at 12:21:41PM +0800, Anand Jain wrote:
> > > 
> > > 
> > > > > Can anyone help me on how to get test coverage for the compression
> > > > > code?
> > > > 
> > > > I'm not surprised xfstests missed this one since it's just readahead.
> > > > You might be able to get better coverage with
> > > > 
> > > > export MOUNT_OPTS="-o compress-force"
> > > 
> > >  And where the data is /dev/urandom the btrfs compression will
> > >  bail out, so xfstest cases which uses /dev/urandom won't test
> > >  the compression code.
> > 
> > Isn't that just with "-o compress"? That's why I recommended "-o
> > compress-force".
> 
>  Nope. compress-force doesn't enforce compress even if the data
>  isn't compressible. I am not sure if its a bug, but its been
>  like that.
> 
>  The difference between compress and compress-force is that
>  compress-force will never give up and compress will give up
>  compress by setting nocompress flag if the first extent is
>  not compressible.
> 
> Thanks, Anand

Huh, you're right. The wording in man 5 btrfs could use some
clarification.

-- 
Omar