Re: setauth segfaults when giving an inexistent username

2008-04-19 Thread Marc-Olivier Mercier
Samuel Thibault wrote: Hello, The patches seems to make sense. I guess we need copyright assignment for that, do we have it? Finally the copyright assignment process is complete. Marc. O.

Re: setauth segfaults when giving an inexistent username

2008-03-19 Thread Marc-Olivier Mercier
Samuel Thibault wrote: Hello, The patches seems to make sense. I guess we need copyright assignment for that, do we have it? Samuel No, you don't. How do I proceed? -- Marc O.

Re: setauth segfaults when giving an inexistent username

2008-03-19 Thread Marc-Olivier Mercier
Marc-Olivier Mercier wrote: Hi all, I tried to used setauth and it segfaults when I passed a wrong user name (or even a wrong group name). The problem is when the args are parsed. In libshouldbeinlibc/ugids-argp.c, the function getpwnam_r is used. It is assumed that getpwnam_r return non

setauth segfaults when giving an inexistent username

2008-03-13 Thread Marc-Olivier Mercier
Hi all, I tried to used setauth and it segfaults when I passed a wrong user name (or even a wrong group name). The problem is when the args are parsed. In libshouldbeinlibc/ugids-argp.c, the function getpwnam_r is used. It is assumed that getpwnam_r return non-zero if the user does not exist,

Re: Re : Concerns about login

2008-03-08 Thread Marc-Olivier Mercier
Samuel Thibault wrote: Mmm, actually maybe it would make even more sense to use if (id != (pid_t)-1) What's a pid_t ? Eeergl, sorry, that was meant to be read uid_t :) Ah! makes sense now... I was wondering if we should prevent the creation of users with uid -1? Marc O.

Re : Concerns about login

2008-03-07 Thread Marc-Olivier Mercier
Samuel Thibault wrote: Marc-Olivier Mercier, le Thu 06 Mar 2008 22:27:14 -0500, a écrit : @@ -258,7 +258,7 @@ verify_id (uid_t id, int is_group, int m /* VERIFY_FN should have been defaulted in idvec_verify if necessary. */ assert (verify_fn); - if (id >= 0) + if ((signed int

Re : Concerns about login

2008-03-06 Thread Marc-Olivier Mercier
Samuel Thibault wrote: Marc-Olivier Mercier, le Sun 24 Feb 2008 10:57:24 -0500, a écrit : There's many patches possible : 4): treat (signed long)(-1) as a special case, allowing value 0 to (signed long)(-2) to be used normally. I've checked in SUSv3: uid_t is allowed to b

Concerns about login

2008-02-24 Thread Marc-Olivier Mercier
I have installed Debian GNU/Hurd about 2 weeks ago. I have some concerns about the way login works. - GNU 0.3 (stracci) (tty5) Use `login USER' to login, or `help' for more information. login> login bob login: bob: Unknown user - 1) Shouldn't "paranoid" be the only behavior of login? C