These functions only apply to utmp(5). They cannot be kept intact when
moving towards utmpx. The login(3) function would break, because its
argument is an utmp structure. The logout(3) and logwtmp(3) functions
cannot be used, since they provide a functionality which partially
overlaps.
Increment SHLIB_MAJOR to 9 to indicate the removal.
I'm not increasing the shlib major version for this, because not a
single application outside the base system should have used these
functions in such a short timespan.
Rewrite ulog_login(3) and ulog_logout(3) to build on top of the utmpx
implementation in libc.
- Partially revert r184122 (sshd.c). Our ut_host is now big enough to
fit proper hostnames.
- Change config.h to match reality.
- defines.h requires UTMPX_FILE to be set by <utmpx.h> before it allows
the utmpx code to work. This makes no sense to me. I've already
mentioned this upstream.
- Add our own platform-specific handling of lastlog. The version I will
send to the OpenSSH folks will use proper autoconf generated
definitions instead of `#if 1'.
Just like rlogind, there is no need to change the ownership of the
terminal during shutdown anymore. Also don't call logwtmp, because the
login(1)/PAM is responsible for doing this. Also use SHUT_RDWR instead
of 2.
Unfortunately I have to partially wreck its functionality, though. ftpd
used to keep a file descriptor to the wtmp, which allowed it to work
from within a chroot. The current utmpx implementation doesn't offer a
way to do this. Maybe we can address this in the future, if it turns out
to be a real issue.
It seems comsat stats the utmpx database each 15 seconds to see whether
it has been changed. I am changing this behaviour to look at the utmpx
database upon processing. I don't want to allow direct interference with
the database files. I also wonder whether this optimization has any
measurable performance benefit nowadays.
- It shouldn't call logwtmp(). Applications like login(1) already make
sure both login and logout entries are written to the storage.
- There's no need to restore permissions on the pseudo-terminal, since
it should be garbage collected by the kernel.
While there, fix a bug I introduced previously. We must reopen the
database for each username passed on the command line. We must rewind
the database and search from the beginning.
Similar to last(1), it must compare ut_id's instead of TTYs to determine
whether a session has been terminated. It must also use ut_type to
determine the type of the login record instead figuring it out by
itself.
Let it print "-" when the TTY string is empty. In this case, it must
also make sure it doesn't match processes who also have no controlling
TTY. Otherwise it will print random kernel processes when trying to pick
the best matching process.
Eventually it should look at the value of ut_pid as well.
Because getutxent also matches the typical get*ent format of library
routines, I thought it would be a good idea to teach it how to read
utmpx databases. getent(1) just gives a raw dump, which is very useful
when debugging problems related to parsing/logging.
Basically there are three major things I changed about last(1):
- It should use ut_type instead of determining by hand what type of
record was given.
- It should now keep track of ut_id's instead of TTYs. This means the
ttylist has been renamed to the idlist, storing all the ut_id's it has
processed until the next reboot.
- I've removed the signal handler. Because our wtmp is rotated so often,
it makes little sense. Even on a simple piece of hardware it should be
capable of grinding through megabytes of logs in a second.
According to a comment, we cannot safely remove utmpx entries here
anymore. This is because the libc routines may block on file locking. In
an ideal world login(1) should just remove the entries, which is why I'm
disabling this code for now. If it turns out we get lots of stale
entries here, we should figure out a way to deal with that.
A nice thing about utmpx is that it makes it very easy to log sessions
that don't use TTYs. This is because the file is not indexed by TTY
slots anymore.
Silence from: brian
It was already ported to use libulog, which makes it simpler now. Be
sure to catch the error returned by setutxdb(). Otherwise it may perform
a lookup on the utx.active database.
csh allows you to monitor the utmp(x) file to monitor certain user
logins. Unfortunately it needs to directly stat() this file. I don't
want to break this module , but eventually it shouldn't do that. The
idea of the getutxent(3) API is to hide file access.
Approved by: mp
The utmpx interface is the standardized interface of the user accounting
database. The standard only defines a subset of the functions that were
present in System V-like systems.
I'd like to highlight some of the traits my implementation has:
- The standard allows the on-disk format to be different than the
in-memory representation (struct utmpx). Most operating systems don't
do this, but we do. This allows us to keep our ABI more stable, while
giving us the opportunity to modify the on-disk format. It also allows
us to use a common file format across different architectures (i.e.
byte ordering).
- Our implementation of pututxline() also updates wtmp and lastlog (now
called utx.log and utx.lastlogin). This means the databases are more
likely to be in sync.
- Care must be taken that our implementation discard any fields that are
not applicable. For example, our DEAD_PROCESS records do not hold a
TTY name. Just a time stamp, a record identifier and a process
identifier. It also guarantees that strings (ut_host, ut_line and
ut_user) are null terminated. ut_id is obviously not null terminated,
because it's not a string.
- The API and its behaviour should be conformant to POSIX, but there may
be things that slightly deviate from the standard. This implementation
uses separate file descriptors when writing to the log files. It also
doesn't use getutxid() to search for a field to overwrite. It uses an
allocation strategy similar to getutxid(), but prevents DEAD_PROCESS
records from accumulating.
Make sure libulog doesn't overwrite the manpages shipped with our C
library. Also keep the symbol list in Symbol.map sorted.
I'll bump __FreeBSD_version later this evening. I first want to convert
everything to <utmpx.h> and get rid of <utmp.h>.
from standard 3G wireless units by supplying a raw IP/IPv6 endpoint rather than
using PPP over serial. uhsoctl(1) is used to initiate and close the WAN
connection.
Obtained from: Fredrik Lindberg <fli@shapeshifter.se>
as different nearby ones, and was grabbed by MIXER_xxx() handler.
While there, replace '(cmd & MIXER_xxx(0)) == MIXER_xxx(0)' expressions
with more correct '(cmd & ~0xff) == MIXER_READ(0)'. Use of bit operations
to compare numeric fields doesn't looks sane.
MFC after: 1 month
kern.ngroups+1. kern.ngroups can range from NGROUPS_MAX=1023 to
INT_MAX-1. Given that the Windows group limit is 1024, this range
should be sufficient for most applications.
MFC after: 1 month
