Re: [PATCH 43/98] include/uapi/linux/hdlc/ioctl.h: include linux/if.h

From: Mikko Rapeli
Date: Tue Jun 02 2015 - 15:34:59 EST


On Sat, May 30, 2015 at 10:22:53PM +0200, Frans Klaver wrote:
> On Sat, May 30, 2015 at 05:38:35PM +0200, Mikko Rapeli wrote:
> > Fixes userspace compiler error:
> >
> > error: âIFNAMSIZâ undeclared here (not in a function)
> >
> > Signed-off-by: Mikko Rapeli <mikko.rapeli@xxxxxx>
> > ---
> > include/uapi/linux/hdlc/ioctl.h | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/include/uapi/linux/hdlc/ioctl.h b/include/uapi/linux/hdlc/ioctl.h
> > index 04bc027..8bbee11 100644
> > --- a/include/uapi/linux/hdlc/ioctl.h
> > +++ b/include/uapi/linux/hdlc/ioctl.h
> > @@ -1,6 +1,7 @@
> > #ifndef __HDLC_IOCTL_H__
> > #define __HDLC_IOCTL_H__
> >
> > +#include <linux/if.h>
> >
> > #define GENERIC_HDLC_VERSION 4 /* For synchronization with sethdlc utility */
>
> What's the point of this one if you break this circular dependency in
> patch 49? Are those in between related?

They are indeed...

On Sat, May 30, 2015 at 09:52:23PM +0200, Frans Klaver wrote:
> On Sat, May 30, 2015 at 05:38:41PM +0200, Mikko Rapeli wrote:
> > This isn't pretty but does the job. I assume including plain linux/hdlc/ioctl.h
> > without linux/if.h is much more rare than the other way round, and that
> > these header file names are part of API, and that creating a new header file
> > just for IFNAMSIZ isn't a good idea either.
> >
> > Fixes userspace compiler error:
> >
> > error: âIFNAMSIZâ undeclared here (not in a function)
> >
> > Signed-off-by: Mikko Rapeli <mikko.rapeli@xxxxxx>
> > ---
> > include/uapi/linux/hdlc/ioctl.h | 5 ++++-
> > include/uapi/linux/if.h | 4 ++++
> > 2 files changed, 8 insertions(+), 1 deletion(-)
> >
> > diff --git a/include/uapi/linux/hdlc/ioctl.h b/include/uapi/linux/hdlc/ioctl.h
> > index 8bbee11..73982ec 100644
> > --- a/include/uapi/linux/hdlc/ioctl.h
> > +++ b/include/uapi/linux/hdlc/ioctl.h
> > @@ -1,7 +1,10 @@
> > #ifndef __HDLC_IOCTL_H__
> > #define __HDLC_IOCTL_H__
> >
> > -#include <linux/if.h>
> > +/* For breaking dependency loop between if.h and hdlc/ioctl.h */
> > +#ifndef IFNAMSIZ
> > +#define IFNAMSIZ 16
> > +#endif
> >
> > #define GENERIC_HDLC_VERSION 4 /* For synchronization with sethdlc utility */
> >
> > diff --git a/include/uapi/linux/if.h b/include/uapi/linux/if.h
> > index 9cf2394..e4b130a 100644
> > --- a/include/uapi/linux/if.h
> > +++ b/include/uapi/linux/if.h
> > @@ -23,6 +23,10 @@
> > #include <linux/socket.h> /* for "struct sockaddr" et al */
> > #include <linux/compiler.h> /* for "__user" et al */
> >
> > +/**
> > + * IFNAMSIZ is also defined in linux/hdlc/ioctl.h if it does not exists
> > + * to break dependency loop between linux/if.h and linux/hdlc/ioctl.h.
> > + */
> > #define IFNAMSIZ 16
> > #define IFALIASZ 256
> > #include <linux/hdlc/ioctl.h>
>
> This feels backwards to me. Wouldn't it be less funky to accept that
> IFNAMSIZE isn't defined in if.h, but rather in hdlc/ioctl.h? If now

That is much less funky, thanks! I'll squash these to commits into a one
which moves IFNAMSIZ to linux/hdlc/ioctl.h.

-Mikko
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/