rtems_tarfs_load("/", ...) -> null_op_mknod()

[Nick Withers]

On Tue, 2013-10-08 at 20:22 -0500, Joel Sherrill wrote:
> Could this have been detected and reported better?

Sure - it could detect directories called (e.g.) "." and "/" and skip
through them...

But should it? It now does return an error in this event. If that'd been
the case when I'd knocked my code up earlier then it would have been
brought to my attention.

If you don't think it's just a case of "application developer did the
wrong thing and is now warned so it's their responsibility", I could
knock up a patch, though?
-- 
Nick Withers

Embedded Systems Programmer
Room 2.26, Building 57
Department of Nuclear Physics
Research School of Physics and Engineering
The Australian National University (CRICOS: 00120C)

eMail: nick.withers at anu.edu.au
Phone: +61 2 6125 2091
Mobile: +61 414 397 446

> Nick Withers <nick.withers at anu.edu.au> wrote:
> 
> 
> 'Twas my bad: I'd done my tar-ing strangely in a Makefile and had a "."
> directory in the top-level.
> 
> Thanks guys, sorry for the noise!
> --
> Nick Withers
> 
> Embedded Systems Programmer
> Room 2.26, Building 57
> Department of Nuclear Physics
> Research School of Physics and Engineering
> The Australian National University (CRICOS: 00120C)
> 
> eMail: nick.withers at anu.edu.au
> Phone: +61 2 6125 2091
> Mobile: +61 414 397 446
> 
> On Wed, 2013-10-02 at 13:46 -0400, Gedare Bloom wrote:
> > Joel, perhaps the patch I checked in propagated a bad return value
> > that was previously ignored?
> > -Gedare
> >
> > On Wed, Oct 2, 2013 at 11:22 AM, Joel Sherrill
> > <joel.sherrill at oarcorp.com> wrote:
> > > I assume this worked in the past but what root filesystem do you
> > > have configured?
> > >
> > >
> > > On 10/2/2013 5:52 AM, Gedare Bloom wrote:
> > >>
> > >> Nick,
> > >> It sounds like a bug somewhere to me. If you have a simple example
> > >> that reproduces this behavior, could you file a report on the RTEMS
> > >> bugzilla?
> > >>
> > >> -Gedare
> > >>
> > >> On Wed, Oct 2, 2013 at 2:46 AM, Nick Withers <nick.withers at anu.edu.au>
> > >> wrote:
> > >>>
> > >>> Hey guys,
> > >>>
> > >>>
> > >>> http://git.rtems.org/rtems/commit/?id=97ae1eeb38a6521cfd52171cf6b816bdebf59f6b
> > >>> seems to have broken rtems_tarfs_load("/", ...) for me.
> > >>>
> > >>> rtems_filesystem_mknod() ends up calling null_op_mknod(), which
> > >>> unconditionally returns -1 (which makes sense, given its name) and now
> > >>> the whole shebang fails.
> > >>>
> > >>> Am I doing something silly? I've really just mimicked some of the
> > >>> examples' usage, and it works a charm with the change backed out.
> > >>> --
> > >>> Nick Withers