Re: [PATCH] Phoenix-RTOS: Add caddr_t definition to <sys/types.h>.

[Jakub Sejdak <jakub dot sejdak at phoesys dot com>]

As I said in the initial commit, we prefer to avoid duplication of
type definitions between kernel and newlib.
We also don't want to include newlib headers in kernel (it should
compile without newlib). This is our company policy and I cannot do
anything about that.

This unfortunately requires "hacking" standard newlib headers and
moving some definitions to kernel headers.
I personally would avoid maintenance of every header, but idea of
user-space adjusting to OS (not the other way) sounds reasonable to
me.

2016-06-28 12:52 GMT+02:00 Corinna Vinschen <vinschen@redhat.com>:
> On Jun 28 10:47, Jakub Sejdak wrote:
>> From: Kuba Sejdak <jakub.sejdak@phoesys.com>
>>
>> ---
>>  newlib/libc/sys/phoenix/sys/types.h | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/newlib/libc/sys/phoenix/sys/types.h b/newlib/libc/sys/phoenix/sys/types.h
>> index f491bd8..c0c9198 100644
>> --- a/newlib/libc/sys/phoenix/sys/types.h
>> +++ b/newlib/libc/sys/phoenix/sys/types.h
>> @@ -39,6 +39,7 @@ typedef __uint64_t          __off64_t;
>>  typedef __uint32_t           __key_t;
>>  typedef __uint32_t           __useconds_t;
>>  typedef __uint32_t           __daddr_t;
>> +typedef char *                       caddr_t;
>>  typedef __uint32_t           __nlink_t;
>>  typedef __uint8_t            __u_char;
>>  typedef unsigned short       __u_short;
>> --
>> 2.7.4
>
> Thanks, applied.
>
> However, why do you do everything by yourself instead of using the
> already existing header structure?  This doesn't really make sense.
>
>
> Corinna
>
> --
> Corinna Vinschen
> Cygwin Maintainer
> Red Hat



-- 
Jakub Sejdak
Software Engineer
Phoenix Systems (www.phoesys.com)
+48 608 050 163