Re: [PATCH 1/3] generic: workaround device where glibc is not installed

[Eryu Guan <guaneryu@gmail.com>]

On Thu, Nov 29, 2018 at 01:10:03PM +1100, Dave Chinner wrote:
> On Wed, Nov 28, 2018 at 11:30:39AM +0800, Eryu Guan wrote:
> > >  	export CONFIG_INCLUDED=true
> > > diff --git a/common/rc b/common/rc
> > > index be1ed68c..992cb3cc 100644
> > > --- a/common/rc
> > > +++ b/common/rc
> > > @@ -3686,12 +3686,6 @@ _get_block_size()
> > >  	stat -f -c %S $1
> > >  }
> > >  
> > > -get_page_size()
> > > -{
> > > -	echo $(getconf PAGE_SIZE)
> > > -}
> > > -
> > > -
> > 
> > But I think we could just use "$here/src/feature -s" to get page size in
> > get_page_size() and "$here/src/feature -w" to get bits per long. But we
> > need to add
> > 
> > _require_test_program "feature"
> 
> IMO, that is unnecessary. `feature` is test harness infrastructure,
> like common/rc and check.  If the feature binary does not build then
> the test harness is running on a broken platform. IOWs, we shouldn't
> be requiring functionality tests for core infrastructure - if the
> core infrastructure didn't build, then there's bigger problems that
> need to be fixed...

Makes sense. Then it'd be better to require it explicitly in
common/config as what we did to fsstress and xfs_io etc.

Thanks,
Eryu