Re: Correct types for mod_devicetable.h (was: Re: m68k build failure)

From: Adrian Bunk <bunk@kernel.org>
To: Pierre Ossman <drzeus@drzeus.cx>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>,
	Al Viro <viro@ftp.linux.org.uk>,
	Andrew Morton <akpm@linux-foundation.org>,
	Sam Ravnborg <sam@ravnborg.org>,
	Marcel Holtmann <marcel@holtmann.org>,
	Linux Kernel Development <linux-kernel@vger.kernel.org>,
	Rusty Russell <rusty@rustcorp.com.au>,
	linux-arch@vger.kernel.org
Subject: Re: Correct types for mod_devicetable.h (was: Re: m68k build failure)
Date: Mon, 10 Dec 2007 19:11:12 +0100	[thread overview]
Message-ID: <20071210181112.GM20441@stusta.de> (raw)
In-Reply-To: <20071209180818.55bf1443@poseidon.drzeus.cx>

On Sun, Dec 09, 2007 at 06:08:18PM +0100, Pierre Ossman wrote:
> On Sun, 2 Dec 2007 12:22:31 +0100 (CET)
> Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> 
> > 
> > I gave it a try:
> >   - Remove existing alignment attributes from some device_id types
> >   - Introduce kernel_* types with proper size and alignment for
> >     cross-compilation (sample <asm/kerneltypes.h> for m68k included)
> >   - Introduce BITS_PER_KERNEL_LONG, to make it clearer it applies to the target
> > 
> > Apart from a cross-compile session for m68k, it's untested.
> > 
> 
> This still requires a bit of maintenance to set up a kerneltypes.h for every arch.

Better doing this work once than fixing similar issues again and again.

> It also means we have to be very careful that gcc's internal alignment settings matched the ones in our header.

There's nothing "gcc internal" about struct alignment - remember that 
any change in struct alignment would change our userspace ABI.

> Rgds
> Pierre

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed