[PATCH 1/2] Provide __intmax_t and __uintmax_t

[Sebastian Huber]

----- Am 17. Okt 2016 um 17:03 schrieb Corinna Vinschen vinschen@redhat.com:

> Hi Sebastian,
> 
> On Oct 10 08:43, Sebastian Huber wrote:
>> Provide __intmax_t and __uintmax_t via <machine/_default_types> for
>> FreeBSD compatibility.
>> 
>> Signed-off-by: Sebastian Huber <sebastian.huber@embedded-brains.de>
>> ---
>>  newlib/libc/include/machine/_default_types.h | 9 +++++++++
>>  1 file changed, 9 insertions(+)
>> 
>> diff --git a/newlib/libc/include/machine/_default_types.h
>> b/newlib/libc/include/machine/_default_types.h
>> index ffc646d..eaa6cec 100644
>> --- a/newlib/libc/include/machine/_default_types.h
>> +++ b/newlib/libc/include/machine/_default_types.h
>> @@ -211,6 +211,15 @@ typedef long __intptr_t;
>>  typedef unsigned long __uintptr_t;
>>  #endif
>>  
>> +#ifdef __INTMAX_TYPE__
>> +typedef __INTMAX_TYPE__ __intmax_t;
>> +#ifdef __UINTMAX_TYPE__
>> +typedef __UINTMAX_TYPE__ __uintmax_t;
>> +#else
>> +typedef unsigned __INTMAX_TYPE__ __uintmax_t;
>> +#endif
>> +#endif
>> +
> 
> This doesn't match the definition of the non-underscored intmax_t and
> uintmax_t from stdint.h.  Is that intentional?  If not, wouldn't it
> make sense to align the definitions?  Or better, let's define __intmax_t
> and __uintmax_t in machine/_default_types.h according to the current
> definition in stdint.h and then use __intmax_t/__uintmax_t to define
> intmax_t/uintmax_t.

Yes, good point. I will fix this and send a second version in a couple of days.