On 4/29/20 10:00 AM, Sascha Hauer wrote:
> On Wed, Apr 29, 2020 at 08:40:40AM +0200, Ahmad Fatoum wrote:
>> On little endian platforms that don't define __arch_swab32p, be32_to_cpu
>> ultimately expands to an expression that evaluates its argument multiple
>> times. Because of the increment, this is wrong (and UB because of
>> unsequenced modification).
>> Hoist the side effect out of the macro argument.
>
> I'm not sure if the reasoning is correct. That would mean that
> of_read_number() doesn't work on ARM, but I can assure it actually does.
Typo, sorry. arch needs to define __arch_swab32 (no p at the end).
of_read_number -> be32_to_cpu -> __be32_to_cpu -> __swab32 -> __fswab32
then
static inline __attribute_const__ __u32 __fswab32(__u32 val)
{
#ifdef __arch_swab32
return __arch_swab32(val);
#else
return ___constant_swab32(val);
#endif
}
where ___constant_swab32 is the macro that evaluates the argument
multiple times.
grep shows '#define __arch_swab32' matches for mips, arm, ppc and x86.
> That said, the implementation looks way better with this patch.
Could you amend the commit message when applying?
Thanks
Ahmad
--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http://www.pengutronix.de/ |
31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |
_______________________________________________
barebox mailing list
barebox@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/barebox
