[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: linux-kernel-headers 2.6.18-1



Hello all,

Thanks for the fix. As i have not seen this package at repositories yet (i'll have to wait a bit). There is an untested arm toolchain (gcc-4.0) for sid-i386 at:
http://www.emdebian.org/~zumbi/toolchain/
These are built with 2.6.17 linux-kernel-headers.

There are some other arches too, please if you use it, i would apreciate some comments and feedback.



2006/10/6, Wookey <wookey@aleph1.co.uk>:
On 2006-10-03 18:00 +0100, Wookey wrote:
> On 06-10-03 18:02 +0200, Hector Oron wrote:
> >    Hello,
> >
> >    Is it possible to build linux-kernel-headers - 2.6.18-1 - for ARM arch ?
>
> According to
> http://buildd.debian.org/~jeroen/status/package.php?p=linux-kernel-headers&a=arm
> they failed due to a syntax error:
> /build/buildd/linux-kernel-headers-2.6.18/testsuite/../debian/linux-kernel-headers/usr/include/asm/byteorder.h:21:
> error: expected '=', ',', ';', 'asm' or '__attribute__' before '__u32'
>
> Anyone had a look at that yet?

OK, I had a look. bug #391305.

This patch fixes it:
--- include/asm-arm/byteorder.h.orig    2006-10-05 22:44:38.679856441 +0100
+++ include/asm-arm/byteorder.h 2006-10-05 22:45: 16.206748589 +0100
@@ -18,7 +18,7 @@
#include <linux/compiler.h>
#include <asm/types.h>

-static inline __attribute_const__ __u32 ___arch__swab32(__u32 x)
+static __inline__ __attribute_const__ __u32 ___arch__swab32(__u32 x)
{
        __u32 t;

@@ -29,7 +29,7 @@
                 * right thing and not screw it up to different degrees
                 * depending on the gcc version.
                 */
-               asm ("eor\t%0, %1, %1, ror #16" : "=r" (t) : "r" (x));
+               __asm__ ("eor\t%0, %1, %1, ror #16" : "=r" (t) : "r" (x));
        } else
#endif
                t = x ^ ((x << 16) | (x >> 16)); /* eor r1,r0,r0,ror #16 */

I don't really understand the distinction here, but all the other
arches except sh64 define this function with __inline__ (and __asm__) so I assume
that's correct (especially as it builds).

It may well be that sh64 is broken without a similar patch too, I
don't know.

This may not be an optimium fix - I welcome feedback from people who
actually understand the ins and outs of it.

zumbi - you should be able to build arm cross-toolchains once this
filters through.

Wookey
--
Aleph One Ltd, Bottisham, CAMBRIDGE, CB5 9BA, UK  Tel +44 (0) 1223 811679
work: http://www.aleph1.co.uk/                 play: http://wookware.org/


--
To UNSUBSCRIBE, email to debian-arm-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org



Reply to: