[Bug 2160] IMFS pipe support broken

[bugzilla-daemon at rtems.org]

https://www.rtems.org/bugzilla/show_bug.cgi?id=2160

--- Comment #8 from Sebastian Huber <sebastian.huber at embedded-brains.de> 2013-12-11 02:13:15 CST ---
(In reply to comment #7)
> (In reply to comment #6)
> > I don't think it is good to break existing application configurations this way.
> 
> Hows does this break existing applications ?

With this change you get now FIFO support although the configuration said
otherwise.

> 
> > Why can't you use
> > 
> > #define CONFIGURE_FIFOS_ENABLED
> > 
> > #include <rtems/confdefs.h>
> > 
> > ?
> > 
> 
> I did not know this existed. I did not know various parts of POSIX where being
> controlled by confdefs.h.
> 
> As stated mkfifo does not fail however using it did 

Ok, this is a bug.

> and I spent a while
> thinking it was my code. Figuring out this related to the set up of the IMFS
> was not easy and not intuitive. If mkfifo failed with the errno for not
> supported that would be better but it does not.
> 
> > FIFO support is optional in IMFS and it should stay this way.  See also
> > 
> > http://git.rtems.org/rtems/tree/cpukit/libfs/src/imfs/fifoimfs_init.c
> > 
> > which is now identical your
> > 
> > http://git.rtems.org/rtems/tree/cpukit/libfs/src/imfs/imfs_init.c
> > 
> > This makes no sense at all.
> 
> Yes having both files does not make sense. If the open called for the FIFO
> failed then the change can be reverted. I wonder if the default handlers being
> used are wrong for the FIFO. This would also catch the mknod path to creating a
> FIFO.

I will fix it so that an error is returned in mkfifo().

-- 
Configure bugmail: https://www.rtems.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching all bug changes.