Closes#3567
This doesn't feel complete, but there isn't much more I can do without
changing lots of code and doing more research, and I don't want to hold
up the Windows beta for error handling that may not be used.
I'm only handling crashes, errors, and panics in the `Controller` task
since it's hard to simulate them elsewhere. Everything else like the
wintun worker thread will have to be best-effort or tested by modifying
the code temporarily.
Crashes go to stderr and and the log file, and we can get a line number
from the crash dump:
Errors go to stderr and the log file, but we don't get a line number,
because we aren't capturing the backtrace:
Panics go to stderr and the log file, but we don't even get the panic
message. I don't think I can do much about that. Tokio doesn't give us
the full `PanicInfo`, which contains the message and location (line
number). If we end up debugging a panic in prod we could bisect by
putting tighter bounds until we narrow it down.
The panic hook ends up not triggered because:
- If I don't catch the panic, the app won't close, because Tokio will
catch the panic
- If I do catch the panic, I can't call the panic hook manually, because
Tokio doesn't give me a `PanicInfo` struct. And I can't resume the
panic, because I'm still inside a Tokio task.
I don't think there's any point capturing a crash dump on panic, because
Tokio will have already stopped the panicked thread, and the backtrace
will only show the catch location.
Maybe the message would print if I changed the controller task to a
controller thread, so that panicking that thread will trigger
`tracing_panic` and then abort the process? But that's a bigger change.
---------
Signed-off-by: Reactor Scram <ReactorScram@users.noreply.github.com>
Co-authored-by: Gabi <gabrielalejandro7@gmail.com>
Automatically write the wintun.dll file on startup and then detect
whether we need to elevate to admin privileges.
I check for privileges by making a test tunnel, so I did #2758 as part
of this, which bundles the DLL inside the exe, and then the exe deploys
it.
---------
Signed-off-by: Reactor Scram <ReactorScram@users.noreply.github.com>
Co-authored-by: Jamil <jamilbk@users.noreply.github.com>
Stacked on PR #2888
This will fix#2878 by setting some Windows security permissions to let
non-admin clients connect to a named pipe server owned by an admin
process.
It should also give us a path to fix 2808 (If I link it, Github assumes
this PR closes that issue?), since it'll be our own code instead of
tauri-plugin-deep-link, and we can just choose not to panic. I'm using
Tokio's `named_pipe` module where that panic happens, and Tokio is good
about just returning Results instead of panicking.
---------
Signed-off-by: Reactor Scram <ReactorScram@users.noreply.github.com>
Co-authored-by: Jamil <jamilbk@users.noreply.github.com>
