Are ut_user and ut_host in struct utmp NUL-terminated?

Are ut_user and ut_host in struct utmp NUL-terminated?

[Florian Weimer]

Can ut_user be exactly UT_NAMESIZE bytes long?  Or is it necessary to 
set aside a byte with the NUL terminator?

I'm not sure what POSIX says about struct utmpx.  It never says “string” 
in the context of ut_user, so I don't think it clarifies this matter.

Thanks,
Florian

Re: Are ut_user and ut_host in struct utmp NUL-terminated?

[Andreas Schwab]

On Aug 13 2018, Florian Weimer <fweimer@redhat.com> wrote:

> Can ut_user be exactly UT_NAMESIZE bytes long?  Or is it necessary to set
> aside a byte with the NUL terminator?

login from util-linux doesn't null terminate.

Andreas.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."

Re: Are ut_user and ut_host in struct utmp NUL-terminated?

[Michael Kerrisk (man-pages)]

On 08/13/2018 02:10 PM, Florian Weimer wrote:
> Can ut_user be exactly UT_NAMESIZE bytes long?  Or is it necessary to 
> set aside a byte with the NUL terminator?
> 
> I'm not sure what POSIX says about struct utmpx.  It never says “string” 
> in the context of ut_user, so I don't think it clarifies this matter.

Each of the char fields in utmp/utmpx can be completely filled with
characters. (I.e., there may not be a terminating null byte.)
As you note, POSIX does not say much here, but perhaps the fact that
the term "string" is not mentioned is a clue.

Thanks,

Michael


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

Re: Are ut_user and ut_host in struct utmp NUL-terminated?

[Florian Weimer]

On 08/21/2018 10:47 AM, Michael Kerrisk (man-pages) wrote:
> On 08/13/2018 02:10 PM, Florian Weimer wrote:
>> Can ut_user be exactly UT_NAMESIZE bytes long?  Or is it necessary to
>> set aside a byte with the NUL terminator?
>>
>> I'm not sure what POSIX says about struct utmpx.  It never says “string”
>> in the context of ut_user, so I don't think it clarifies this matter.
> 
> Each of the char fields in utmp/utmpx can be completely filled with
> characters. (I.e., there may not be a terminating null byte.)
> As you note, POSIX does not say much here, but perhaps the fact that
> the term "string" is not mentioned is a clue.

And indeed we already use __attribute__ ((__nonstring__)) for those.  I 
should have checked the current sources.

Thanks,
Florian