Hi Michael, I'm sorry but I hardly know anything about it, basically just what the mentioned link contains. loginuid is something that may or may not be compiled into the kernel. If available, it's initially -1, inherited on fork(), can be changed from -1 to any arbitrary value by anyone, but if it's not -1 then it can only be changed by root. Supposedly login programs (e.g. sshd, graphical display manager, etc., the one that changes from a system daemon to a user process) sets this to the user who logs in (probably via a pam module), for auditing (whatever it means) purposes. The value is supposed to be preserved across a su/sudo, that is, still contain the user who originally logged in, rather than the user su/sudo changes to. I came across this at https://bugzilla.gnome.org/show_bug.cgi?id=747046 which discusses the removal of utmp logging from gnome-terminal. Quite a few utilities rely on getlogin(3) which in turn looks at utmp and hence returns incorrect result there. On the other hand, as discussed there, utmp is unreliable and broken by design. I was surprised to learn about the existence of loginuid, as well as the fact that getlogin(3) actually first looks as loginuid, and only if it's unavailable or unset (-1) then it goes on to utmp. For the gnome-terminal utmp discussion I wish I had known this earlier. Sorry, I know this is nowhere close to the style and precision that a manpage requires. cheers, egmont On Wed, May 18, 2016 at 10:36 PM, Michael Kerrisk (man-pages) <mtk.manpages@xxxxxxxxx> wrote: > On 04/23/2016 11:00 PM, Egmont Koblinger wrote: >> Hi, >> >> I just got to know about /proc/self/loginuid, as well as getlogin(3) >> referring to this file prior to utmp, by browsing this bug: >> https://bugzilla.redhat.com/show_bug.cgi?id=1328947 (esp. comments >> 13-17 are relevant). >> >> getlogin(3)'s manual page does not mention this at all. (And, for the >> record, neither does proc(5), it has a FIXME comment about it). >> >> Could you please fill in this gap? > > Hello Egmont, > > Can you provide some information about this field? If not a patch > to the man page, then at least some descriptive text? > > Thanks, > > Michael > > > -- > Michael Kerrisk > Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ > Linux/UNIX System Programming Training: http://man7.org/training/ -- To unsubscribe from this list: send the line "unsubscribe linux-man" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
