Add an option for custom vchan buffer size, to override default 64k (for
each direction). This is especially useful when the other side of
connection is MirageOS based, because of limited memory and default
grant table size (128 entries).
Previously the script was called through shell as:
execl(shell, "-sh", "-c", "/usr/lib/qubes/qubes-rpc-multiplexer
...", 0);
This tells the shell to load login scripts, including /etc/profile.
Since 5512e4eada this is no longer the
case and the script is called directly. Since most services do expect
proper user session initialized (/etc/profile loaded etc), adjust the
script's shebang to behave like a login shell and load those startup
scripts.
FixesQubesOS/qubes-issues#3615
This avoids duplicating service call parsing in multiple places.
Further improvements to that code (like avoid using shell) can be
implemented in one place.
If fork server is used, proper environment is inherited from the
session. But in other case (like non-default user), it needs to be
created by qrexec-agent itself. PAM provide some variables, but not the
most basic: HOME, SHELL, USER, LOGNAME. Also process should be started
in user home directory (if available).
FixesQubesOS/qubes-issues#3416
Some services require GUI access. Make qrexec-agent handling this, based
on per-service configuration, instead of forcing every caller to call
qubes.WaitForSession service first. This is especially important for
Disposable VMs, because those are destroyed after a single service call.
This needs to be done in qrexec-agent (instead of service script, or
qubes-rpc-multiplexer), because agent will behave differently depending
on GUI session being available or not. Namely, will use
qrexec-fork-server (so the process will be a child of session leader),
or will open new session.
Service configuration lives in /etc/qubes/rpc-config/SERVICE_NAME, can
can contain 'key=value' entries (no space around '=' allowed). Currently
the only settings supported is 'wait-for-session', with value either '0'
or '1'.
QubesOS/qubes-issues#2974
Prepend "-" to shell name, to instruct it being a login shell. This way
shell will initialize environment, load /etc/profile etc.
FixesQubesOS/qubes-issues#2903
Instead of calling 'su' to switch the user, use own implementation of
this. Thanks to PAM it's pretty simple. The main reason is to have
control over process waiting for session termination (to call
pam_close_sesion/pam_end). Especially we don't want it to keep std* fds
open, which would prevent qrexec-agent from receiving EOF when one of
them will be closed.
Also, this will preserve QREXEC_AGENT_PID environment variable.
FixesQubesOS/qubes-issues#2851
When qrexec-client-vm is started with socket on its stdout and no local
process requested, it will try to shutdown(SHUT_WR) this socket when
remote process exists. This is wrong, because this socket may be still
needed by other processes (for example shell from where qrexec-client-vm
was called).
In such a case, simple close() should be used.
1. If local process is started, report its exit code, instaed of remote
one. To get remote exit code, simply start qrexec-client-vm without
third argument (and connect its stdin/stdout with the other process some
other way).
2. Report process terminated by signal.
Don't pretend that process terminated by signal finished successfuly.
Copy shell behaviour of reporting it as 128+signum.
3. Do not wait() for any child process, just the one we expect. In case
of qrexec-client-vm the child process is started differently and
wait()ing on it inside main loop would break its exit code reporting.
FixesQubesOS/qubes-issues#2861
Exit code 1 is very common in all kind of programs, including qrexec
services, so it is hard to distinguish remote failure from service call
refusal. Use something from top of the range here (but not 127, as it is
commonly used to report "Command not found")
QubesOS/qubes-issues#2861
Always set QREXEC_AGENT_PID variable, setup SIGUSR1 handler. And do that
before starting child process to avoid race conditions.
Required for QubesOS/qubes-issues#
FixesQubesOS/qubes-issues#1863
In case of some services it makes much sense for caller to receive also
stderr in addition to stdout. For example:
- qubes.VMShell (stderr required for salt-ssh over qrexec)
- qubes.OpenInVM - especially when called to DispVM - otherwise
diagnosing errors can be hard
And generally all sort of error reporting (the purpose of stderr). It
would ease debugging - instead of message "error occurred, check here and
there for more details", it could be "error occurred: the reason".
FixesQubesOS/qubes-issues#1808
Implement one of TODOs left in the code. Without this buffering, it may
happen that qrexec-agent will hang waiting on write(2) to the child
process, while that child will do the same (try to write something to
the qrexec-agent), without reading its stdin. This would end up in a
deadlock.
FixesQubesOS/qubes-issues#1347
This will ensure that the child process will receive info that the
connection is closed. Otherwise it could hang on write() or in some
cases read() - on its stdin/stdout.
Thanks @adrelanos for help with debugging.
It should be remote process exit code, not the local one.
Also do not 'return' from the middle of the look, just use 'break' to
execute common cleanup code (which will be introduced in next commit).
Simply forget about that connection, instead of waiting for further
messages. If that connection is no longer available, select would return
EBADF, which would cause qrexec-agent termination.
/usr/local resides in private.img, so it is possible to define per-appvm RPC
Also, with the upcoming 3.0 release support for old (R1) paths is
removed.
In case of remote process exit even when some messages are still
waiting, vchan connection can be already closed. If we try to send some
data in this case (for example stdout of local process), there will be
an error, which will terminate qrexec-client-vm/qrexec-agent child. So
first check vchan data (where could be MSG_EXIT_CODE queued) , then
local process.
There is still some race condition in this code - remote process could
exit just after we check vchan, but before we send some data. But this
is much less probable and in the worst case we only loose remote process
exit code.
Child process can request to use single socket for both stdin and
stdout by sending SIGUSR1 signal. If it does so twice or more, previous
code broke the connection by closing the socket.
This doesn't cover all the cases, because local process could want to
receive that value (currently it cant), but I can't think of any simple,
*compatible* way to pass it there.
