This is the mail archive of the
glibc-linux@ricardo.ecn.wfu.edu
mailing list for the glibc project.
who and utmp
- To: glibc-linux@ricardo.ecn.wfu.edu
- Subject: who and utmp
- From: Dale Harris <rodmur@maybe.org>
- Date: Sun, 15 Aug 1999 12:38:30 -0600
- Reply-To: glibc-linux@ricardo.ecn.wfu.edu
Ok, turns out that they are coming out with a new version of sh-utils and
it seems to pretty much work except that there may be a possible bug in
glibc, it doesn't seem to remove old entries from utmp, notice:
rodmur tty1 Aug 14 16:22
rodmur ttyp0 Aug 14 16:22
rodmur ttyp1 Aug 15 09:47
rodmur ttyp2 Aug 14 16:25
I'm logged into the ttyp2 at this time, w from procps-2.02 appears to handle
this correctly though:
12:35pm up 4 days, 22:01, 4 users, load average: 0.00, 0.01, 0.00
USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
rodmur tty1 - Sat 4pm 20:12m 12.43s 0.01s startx
rodmur ttyp0 :0.0 Sat 4pm 0.00s 0.94s 0.31s mutt
rodmur ttyp1 :0.0 9:47am 2:36m 0.26s 0.26s -bash
So it could be a bug in who or glibc.
Another thing that surprises me is the way utmp is accessed:
access("/var/run/utmpd.rw", F_OK) = 0
socket(PF_UNIX, SOCK_STREAM, 0) = 3
connect(3, {sun_family=AF_UNIX, sun_path="/var/run/utmpd.rw"}, 110) = -1 ECONNRE
FUSED (Connection refused)
close(3) = 0
socket(PF_UNIX, SOCK_STREAM, 0) = 3
connect(3, {sun_family=AF_UNIX, sun_path="/var/run/utmpd.ro"}, 110) = 0
That is from a trace of who (from sh-utils 1.16m), it doesn't have any
invocation of connect or any other socket types call in the source, it
in fact tries doing a open on /var/run/utmp. So I'm guessing that glibc
or utmpd somehow catches the open to /var/run/utmp and then directs it to
the /var/run/utmpd.* files? Why?
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dale Harris <rodmur@maybe.org> GPG key: 372FBD57 http://www.maybe.org/
M.A.Y.B.E. = Maybe is an Altruistic Yet Bohemian Enigma