Re: + git-hdrcleanup-vs-git-klibc-on-ia64.patch added to -mm tree

Re: + git-hdrcleanup-vs-git-klibc-on-ia64.patch added to -mm tree

[David Woodhouse]

On Sun, 2006-05-14 at 10:54 -0700, akpm@osdl.org wrote:
> diff -puN include/asm/gcc_intrin.h~git-hdrcleanup-vs-git-klibc-on-ia64
> include/asm/gcc_intrin.h
> ---
> 25/include/asm/gcc_intrin.h~git-hdrcleanup-vs-git-klibc-on-ia64     Mon May 15 01:33:00 2006
> +++ 25-akpm/include/asm/gcc_intrin.h    Mon May 15 01:33:00 2006
> @@ -24,7 +24,9 @@
>  extern void ia64_bad_param_for_setreg (void);
>  extern void ia64_bad_param_for_getreg (void);
>  
> +#ifdef __KERNEL__
>  register unsigned long ia64_r13 asm ("r13") __attribute_used__;
> +#endif 

There's more than just that in asm-ia64/intrinsics.h that shouldn't be
visible to userspace. In fact I'm not entirely sure how much of
intrinsics.h _does_ make sense for userspace at all.

What does asm-ia64/byteorder.h include it for?

-- 
dwmw2

Re: + git-hdrcleanup-vs-git-klibc-on-ia64.patch added to -mm tree

[Christoph Hellwig]

On Sun, May 14, 2006 at 07:51:00PM +0100, David Woodhouse wrote:
> On Sun, 2006-05-14 at 10:54 -0700, akpm@osdl.org wrote:
> > diff -puN include/asm/gcc_intrin.h~git-hdrcleanup-vs-git-klibc-on-ia64
> > include/asm/gcc_intrin.h
> > ---
> > 25/include/asm/gcc_intrin.h~git-hdrcleanup-vs-git-klibc-on-ia64     Mon May 15 01:33:00 2006
> > +++ 25-akpm/include/asm/gcc_intrin.h    Mon May 15 01:33:00 2006
> > @@ -24,7 +24,9 @@
> >  extern void ia64_bad_param_for_setreg (void);
> >  extern void ia64_bad_param_for_getreg (void);
> >  
> > +#ifdef __KERNEL__
> >  register unsigned long ia64_r13 asm ("r13") __attribute_used__;
> > +#endif 
> 
> There's more than just that in asm-ia64/intrinsics.h that shouldn't be
> visible to userspace. In fact I'm not entirely sure how much of
> intrinsics.h _does_ make sense for userspace at all.
> 
> What does asm-ia64/byteorder.h include it for?

Using byte swizzling instrincts I supposed.  On the other hand asm/byteorder.h
really shouldn't be user-exposed to start with.

Re: + git-hdrcleanup-vs-git-klibc-on-ia64.patch added to -mm tree

[David Woodhouse]

On Sun, 2006-05-14 at 22:15 +0100, Christoph Hellwig wrote:
> Using byte swizzling instrincts I supposed.  On the other hand asm/byteorder.h
> really shouldn't be user-exposed to start with. 

A whole bunch of things include byteorder.h 

On Sun, 2006-05-14 at 22:17 +0100, Christoph Hellwig wrote:
> elf.h shouldn't be exported to userspace at all.  there's a separate incarnation
> of elf.h in userland, so anyone importing the kernel one is plain stupid.
> 
> similarly no one should be using page.h, there is nothing in there userspace
> should ever look at

ucontext.h includes elf.h for elf_gregset_t. 
elf.h includes page.h (probably for PAGE_SIZE).

Yes, they do want cleaning up -- but I want the first set of stuff in
the hdrcleanup git tree to be _entirely_ simple and uncontentious. We
can get on to the more interesting stuff later.

-- 
dwmw2

Re: + git-hdrcleanup-vs-git-klibc-on-ia64.patch added to -mm tree

[H. Peter Anvin]

Christoph Hellwig wrote:
> 
> Using byte swizzling instrincts I supposed.  On the other hand asm/byteorder.h
> really shouldn't be user-exposed to start with.

asm/byteorder.h is at least useful for klibc; it can, of course, be 
completely done in klibc itself as long as bi-endian platforms have 
appropriate intrinsic defines.

klibc (ab)uses the __le16_to_cpu() et al macros; the reason for that is 
just to cut down on the amount of architecture-specific code that has to 
be written for klibc itself.

	-hpa