RE: [PATCH 6/6] unicore32: using generic-y format for one line asm-generic files

["Guan Xuetao" <gxt@mprc.pku.edu.cn>]

> -----Original Message-----
> From: Arnd Bergmann [mailto:arnd@arndb.de]
> Sent: Thursday, May 26, 2011 8:45 PM
> To: GuanXuetao
> Cc: linux-kernel@vger.kernel.org; linux-arch@vger.kernel.org; greg@kroah.com; Sam Ravnborg
> Subject: Re: [PATCH 6/6] unicore32: using generic-y format for one line asm-generic files
> 
> On Thursday 26 May 2011, GuanXuetao wrote:
> > From: Guan Xuetao <gxt@mprc.pku.edu.cn>
> >
> > see commit from Sam Ravnborg <d8ecc5cd8e227bc318513b5306ae88a474b8886d>
> >   kbuild: asm-generic support
> >
> > Signed-off-by: Guan Xuetao <gxt@mprc.pku.edu.cn>
> > ---
> >  arch/unicore32/Makefile           |   35 ----------------------
> >  arch/unicore32/include/asm/Kbuild |   59 +++++++++++++++++++++++++++++++++++++
> >  2 files changed, 59 insertions(+), 35 deletions(-)
> 
> Acked-by: Arnd Bergmann <arnd@arndb.de>
> 
> This reminds me that I should do the same for ARM and maybe some others.

There is a potential problem.
When I add stat.h with generic-y method, a file is generated in arch/*/include/generated/asm/ dir.
And then,  a patch for old-abi is applied, which will restore an old stat.h in arch/*/include/asm/ dir.
Without removing the line in arch/*/include/asm/Kbuild, two files exist in the meantime.
It works, depending on the order in LINUXINCLUDE, and (perhaps) the handling order in gcc.
So, a check and warning mechanism should be provided to clear up the confusion.

Guan Xuetao