unveil(2) /etc/gettytab.db in getty(8) to avoid possible violation.

OK deraadt@

gmtime(3) / locatime(3) can fail when timestamps are way off.

Add missing error checks to all calls under libexec/

Input & OK millert

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

show more ...

Explicitly lock unveil. Noticed when testing the recent ps(1) commit exposing
the unveil state of running processes.
ok deraadt@

Pull tname initialization up since it's used in an error path.
Pointed out by jsg, just fix it deraadt@

need to closefrom(0) before execve(), otherwise tty isn't properly
'hung up'; some testing by Theodore Wynnychenko

Theodore Wynnychenko discovered the gettytab "lo=path" feature didn't work
anymore with unveil wired to /usr/bin/login. So let's parse gettytab a bit
earlier to learn which login path to unveil. Lat

Theodore Wynnychenko discovered the gettytab "lo=path" feature didn't work
anymore with unveil wired to /usr/bin/login. So let's parse gettytab a bit
earlier to learn which login path to unveil. Later in the loop gettytab is
re-parsed, if the login changes re-exec getty to reach the unveil from the top.
ok millert, also discussed with mestre

show more ...

unveil(2) in getty. This has been in snaps for more than 2 months,
since I worry that a mistake in here will cause significant grief.

do not need dev[] wasting bss space

Do a better job of not printing sequences we cannot reverse in DELETE or
KILL. Therefore we can do a better job cleaning up.
testing by benno

Stop echoing ANSI sequences back to the console, by expanding ESC character
to literal ^[. This makes getty interaction appear more like login, and
stops a mid-ANSI sequence parsing problem that mes

Stop echoing ANSI sequences back to the console, by expanding ESC character
to literal ^[. This makes getty interaction appear more like login, and
stops a mid-ANSI sequence parsing problem that messed up login, making it
freeze the session until timeout.
ok tedu fcambus

show more ...

Stop supporting primordial # (erase) and @ (kill) sequences.
ok tedu

remove the getname() code trying to handle baud < 1200 terminals.
leave the special down under code until somebody unearths why it exists.

pledge a tiny bit later. This results in the specified tty being
opened before the first pledge call.
testing by tb and benno and others

More "(<blah> *)0" -> NULL, avoiding any stdarg functions.

Feedback millert@ kettenis@

Observe that FIOASYNC clearing for stdin is only done in the case where
getty receives the fd from init, so hoist it upwards. Since revoke(2)
is now allowed by pledge "rpath tty", the pledges can be

Observe that FIOASYNC clearing for stdin is only done in the case where
getty receives the fd from init, so hoist it upwards. Since revoke(2)
is now allowed by pledge "rpath tty", the pledges can be hoisted much
higher.
ok millert semarie tedu guenther

show more ...

further PPP reduction, hint from sthen

remove ppplogin support. ok sthen

Hoist clearing of FIOASYNC to much earlier, then getty can use
pledge "stdio rpath fattr proc exec tty".

remove ancient support for edited hostnames. ok deraadt (long ago)

Log a more useful error message if ttyname() fails. OK deraadt@

adjust to HOST_NAME_MAX+1 & LOGIN_NAME_MAX

Nuke yet more obvious #include duplications.

ok deraadt@

Nuke some obvious #include duplications.

ok espie@ deraadt@ millert@ tedu@

unsigned char for ctype
ok okan kettenis