lkml.org 
[lkml]   [1998]   [Jan]   [28]   [last100]   RSS Feed
Views: [more markup]  [less markup]  [headers]  [forward] 
 
Messages in this thread
/
DateWed, 28 Jan 1998 12:21:18 +0100 (MET)
From Andries.Brouwer@cwi ...
SubjectRe: losetup and dev_t
    From martin@mira.isdn.cs.tu-berlin.de Wed Jan 28 08:27:40 1998

    > For the time being this probably means that one has to do
    >     sed 's/dev_t/unsigned short/'
    > when using kernel include files that will be compiled with glibc.
    Why not use kdev_t?

Heresy!
kdev_t is a secret, hidden type.
Nobody outside the kernel ever gets a kdev_t.
Nobody inside the kernel but outside <linux/kdev_t.h> knows what it is.

(Two years ago or so I have been running a kernel where it was
a pointer; hope to do so soon again. Will submit some patches
polishing kernel parts where integers and kdev_t's are mixed up.
The worst sinners are ps2esdi.c and rocket.c.)

So what to do outside the kernel? Thickets of #ifdef's testing
kernel version and libc version and architecture?

<aj@arthur.rhein-neckar.de> asked: Why not use __kernel_dev_t?
And that is what I ended up doing:

	#include <asm/posix_types.h>
	#define dev_t __kernel_dev_t
	#include <linux/loop.h>

Not pretty, but as far as I can see it works on all architectures
and all libc's and all kernels since 1.3.79.

Andries

\
 
 \ /
  Last update: 2005-03-22 12:41    [W:0.107 / U:0.020 seconds]
©2003-2008 Jasper Spaans