When pid 1 crashes, the getty unit for the console will happily keep
running which means we end up with two shells competing for the same
tty. Let's call vhangup on /dev/console to kill every other process
attached to the console before we spawn the crash shell. The getty
units have Restart=always but lucky for us, pid 1 just crashed in fire
and flames so it isn't actually able to restart the getty unit.
Various tty types come up with cols/rows not initialized (i.e. set to
zero). Let's detect these cases, and return a better error than EIO,
simply to make things easier to debug.
/dev/console is sometimes a symlink in container managers. Let's handle
that correctly, and resolve the symlink, and not consider the data from
/sys/ in that case.
It doesn't really make sense to have that in dev-setup.c, which is
mostly about setting up /dev/, creating device nodes and stuff.
let's move it to the other stuff that deals with /dev/console's
peculiarities.
Numerous fixes:
1. use vtnr_from_tty() to parse out VT number from tty path
2. open tty for write only when we want to output just ansi sequences
3. open tty in asynchronous mode, and apply a timeout, just to be safe
4. propagate error from writing (most callers ignore it anyway, might as
well pass it along correctly)
This is a lot of stuff, and sometimes quite wild, let's turn this into
its own header.
All stuff color-related that just generates sequences is now in
ansi-color.h (no .c file!), and everything more complex that
probes/ineracts with terminals remains in termina-util.[ch]
Let's update the commentary a bit. Also, use a time-out of 100ms rather
than 50ms for this, simply to unify on the same value used in
vt_disallocate() in a similar case.
Let's put "terminal_" as prefix, like with the other reset calls, and
let's make clear that this only encapsulates the ioctl-based reset
logic, not the ANSI sequence based reset logic.
ESC c is a (vaguely defined) "reset to initial state" ANSI sequence.
Many terminals clear the screen in this case, but that's a bit drastic I
think for most resets.
ESC c was added to the reset logic in
00bc83a275 (i.e. very recently), and I
don't think the effect was clear at that time.
Let's keep the ESC c in place however when we actually want to clear the
screen. Hence move it from reset_terminal_fd() into vt_disallocate().
Fixes: #33689
When we are talking to a serial terminal quite commonly the dimensions
are not set properly, because the serial protocol has not handshake or
similar to transfer this information.
However, we can derive the dimensions via ANSI sequences too, which
should get us the right information, since ANSI sequences are
interpreted by the final terminal, rather than an intermediary local tty
driver (which is where TIOCGWINSZ is interpreted).
This adds a helper call that gets the dimensions this way.
Let's prefix these functions with the subsystem name, and clean them up
a bit. Specifically, drop the error logging, it's entirely duplicative,
since every single caller does it anyway.
Let's add an extra safety check: before issuing the ansi sequence to
query the bg color, let's make sure input and output fd actually
reference the same tty. because otherwise it's unlikely we'll be able to
read back the response from the tty driver.
This is mostly just paranoia.
If we only read partial information from the tty we ended up parsing it
again and again, confusing the state machine. hence, return how much
data we actually processed and drop it from the buffer.
We use this at various places, let's unify this in one global constant.
This changes flags in crash-handler.c in a tiny irrelevant way: we ask
syscalls to be continued on signal arrival, which we previously didn't.
But that shouldn't change anything, the only thing we'll do in the
relevant process is call raise(), and that's it, hence there definitely
are no syscalls to restart or not to restart.
- introduce or rename usual enum values _MAX and _INVALID,
- introduce and use string table lookup functions,
- split out implementation of get_color_mode() to _impl(),
- add tests for get_color_mode().
Doing this in reset_terminal_fd() is a bit too invasive, see
https://github.com/systemd/systemd/pull/32406#issuecomment-2070923583.
Let's only do this for /dev/console so that we work around weird firmwares
disabling line-wrapping, but avoid messing too much with other things.
While we're at it, let's handle more than just line wrapping, and do a
more general reset of stuff to get the terminal into a sane state.
The qemu seabios firmware disables serial console line wrapping. Let's
make sure we re-enable it again when we reset a terminal to some sane
defaults.
To avoid potentially blocking on writing to the terminal, we put it
in nonblocking mode and add a timeout of 50ms.
We shouldn't try to use any ANSI escape sequences if TERM=dumb.
Also, the "\r\n" we output can get interpreted as a double newline
(for example by Github Actions), so let's output just "\n" when
TERM=dumb to clean up the CI logs.
Currently scan_background_color_response only accepts BEL (\x07) to end
a response, however some terminals (namely kitty in my case) will reply
with the string terminator (ST - https://en.wikipedia.org/wiki/ANSI_escape_code).
This commit changes the behaviour to now accept either ending.
Currently the return value 0 is not checked for, this indicates a
timeout and should be handled to prevent doing a blocking read on a file
descriptor with no data ready.
Mike Yuan
Daan De Meyer
Lennart Poettering
Yu Watanabe
Zbigniew Jędrzejewski-Szmek
Sam Leonard