On Jun  7 15:14, Sebastian Huber wrote:
> On 07/06/17 15:11, Corinna Vinschen wrote:
> 
> > On Jun  7 11:08, Sebastian Huber wrote:
> > > Using uint32_t for ioctl_command_t does not work well on 64-bit targets.
> > > 
> > > Signed-off-by: Sebastian Huber<sebastian.huber@embedded-brains.de>
> > > ---
> > >   newlib/libc/sys/rtems/include/sys/ioccom.h | 4 ++--
> > >   1 file changed, 2 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/newlib/libc/sys/rtems/include/sys/ioccom.h b/newlib/libc/sys/rtems/include/sys/ioccom.h
> > > index 9c5350d1e..2c0e9ee71 100644
> > > --- a/newlib/libc/sys/rtems/include/sys/ioccom.h
> > > +++ b/newlib/libc/sys/rtems/include/sys/ioccom.h
> > > @@ -36,7 +36,7 @@
> > >   #include <sys/types.h>
> > >   #include <stdint.h>
> > > -typedef uint32_t ioctl_command_t;
> > > +typedef unsigned long ioctl_command_t;
> > Why do you need a ioctl_command_t at all?  It's not in FreeBSD either.
> 
> Its used by the legacy network stack in RTEMS (e.g. various device drivers).
> I don't know who introduced this type and why. glibc has also simply
> "unsigned long".

Ok.  At this point I would have replied "please apply", but you
already did that.  It's write *after* approval ;)


Corinna

-- 
Corinna Vinschen
Cygwin Maintainer
Red Hat