[PATCH] libc: add issetugid()

[Anthony G. Basile]

On 07/27/14 05:36, Rich Felker wrote:
> On Sat, Jul 26, 2014 at 08:12:53AM -0400, Anthony G. Basile wrote:
>> On 07/24/14 16:41, Bernhard Reutner-Fischer wrote:
>>> On Wed, Jul 23, 2014 at 07:28:26AM -0400, Anthony G. Basile wrote:
>>>> I should add that this updated patch addresses Rich's points: 1) I've tested
>>>> this for both dynamic and static linking and I tested that libressl builds
>>>> and works correctly.  2) I added a link to the musl commit for the reasoning
>>>> behind this approach.
>>>>
>>>> On 07/22/14 13:27, basile at opensource.dyc.edu wrote:
>>>>> From: "Anthony G. Basile" <blueness at gentoo.org>
>>>>>
>>>>> issetugid() returns 1 if the process environment or memory address space
>>>>> is considered tainted, and returns 0 otherwise.  This happens, for example,
>>>>> when a process's privileges are elevated by the setuid or setgid flags on
>>>>> an executable belonging to root.  This function first appeard in OpenBSD 2.0
>>>>> and is needed for the LibreSSL.
>>>>>
>>>>> This patch follows the same logic as the equivalent musl commit.  For more
>>>>> information see the commit message at
>>>>>
>>>>> http://git.musl-libc.org/cgit/musl/commit/?id=ddddec106fd17c3aca3287005d21e92f742aa9d4
>>>>> ---
>>>>>   include/unistd.h                    |  8 ++++++++
>>>>>   libc/misc/file/issetugid.c          | 12 ++++++++++++
>>>>>   libc/misc/internals/__uClibc_main.c | 12 ++++++++++++
>>>>>   3 files changed, 32 insertions(+)
>>>>>   create mode 100644 libc/misc/file/issetugid.c
>>>>>
>>>>> diff --git a/include/unistd.h b/include/unistd.h
>>>>> index 540062a..6c2c8c2 100644
>>>>> --- a/include/unistd.h
>>>>> +++ b/include/unistd.h
>>>>> @@ -1168,6 +1168,14 @@ extern long int syscall (long int __sysno, ...) __THROW;
>>>>>
>>>>>   #endif	/* Use misc.  */
>>>>>
>>>>> +#ifdef __USE_MISC
>>>
>>> is MISC (or MISC alone) an appropriate guard?
>>
>> I had a hard time (and still have a hard time) deciding this even
>> after carefully reading include/features.h.  The function started in
>> openbsd and migrated to free and netbsd, but its not in 4.3BSD.
>> _USE_MISC is looser but does include SYS V.  I'm thinking now to
>> just remove the guard.  I did speak to Rich about what musl's doing
>> but it doesn't seem appropriate here.
>>
>> If there are no strong opinions, I'll just remove the guard and
>> resubmit in a few days.
>>
>> Your other comments below are good.
>
> I don't think removing the guard would be correct at all; that would
> expose it even in profiles where the namespace is supposed to conform
> to POSIX/XSI. If __USE_MISC is inappropriate, the solution would be to
> move it to a more-inclusive featureset, not a less-inclusive one.
>
> Rich
>

Yeah since unistd.h is POSIX/XSI and issetugid is not, I guess you 
really do need *some* guard.  As I see it, we have only two 
possibilities (from include/features.h):

    __USE_BSD            Define 4.3BSD things.
    __USE_MISC           Define things common to BSD and System V Unix.

The other choices _USE_POSIX* _USE_XOPEN* _USE_GNU etc are clearly wrong.

issetugid is not in 4.3BSD, and MISC is more inclusive, hence my 
original choice.

So now I'm leaning back using __USE_MISC.


-- 
Anthony G. Basile, Ph. D.
Chair of Information Technology
D'Youville College
Buffalo, NY 14201
(716) 829-8197