Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Ability to optionally log #1

Merged
merged 5 commits into from
Sep 26, 2023
Merged
Show file tree
Hide file tree
Changes from 4 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions etc/Dockerfile
Original file line number Diff line number Diff line change
@@ -1,5 +1,11 @@
FROM alpine

RUN apk update && apk add gcc musl-dev procps

COPY simple /simple

COPY zombie.c /zombie.c

RUN gcc /zombie.c

CMD ["/simple"]
24 changes: 24 additions & 0 deletions etc/zombie.c
Original file line number Diff line number Diff line change
@@ -0,0 +1,24 @@
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>

int main()
{
// fork() creates child process identical to parent
int pid = fork();

// if pid is greater than 0 than it is parent process
// if pid is 0 then it is child process
// if pid is -ve , it means fork() failed to create child process

// Parent process
if (pid > 0)
sleep(20);

// Child process
else {
exit(0);
}

return 0;
}
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sample program to test that it does reap the orphan process.

8 changes: 6 additions & 2 deletions examples/simple.rs
Original file line number Diff line number Diff line change
@@ -1,12 +1,16 @@
use pid1::Pid1Settings;

fn main() -> Result<(), Box<dyn std::error::Error>> {
pid1::relaunch_if_pid1()?;
pid1::relaunch_if_pid1(Pid1Settings { log: true })?;
let id = std::process::id();
println!("In the simple process, going to sleep. Process ID is {id}");
let args = std::env::args().collect::<Vec<_>>();
println!("Args: {args:?}");

for _ in 0..4 {
std::thread::sleep(std::time::Duration::from_secs(5));
std::process::Command::new("date").spawn().unwrap();
}
std::thread::sleep(std::time::Duration::from_secs(5));

Ok(())
}
16 changes: 10 additions & 6 deletions justfile
Original file line number Diff line number Diff line change
Expand Up @@ -2,10 +2,14 @@
default:
just --list --unsorted

# Basic test
# Test
test:
cargo build --target x86_64-unknown-linux-musl --example simple
cp target/x86_64-unknown-linux-musl/debug/examples/simple etc
docker build etc --tag pid1rstest
# Commenting this out to make the test pass
# docker run --rm -t pid1rstest
cargo build --target x86_64-unknown-linux-musl --example simple
cp target/x86_64-unknown-linux-musl/debug/examples/simple etc
docker build etc -f etc/Dockerfile --tag pid1rstest
docker rm pid1rs || exit 0
docker run --name pid1rs -t pid1rstest

# Exec into that docker container
exec-shell:
docker exec -it pid1rs sh
51 changes: 42 additions & 9 deletions src/lib.rs
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
use std::process::Child;

use nix::sys::wait::WaitStatus;
use nix::{sys::wait::WaitStatus, unistd::Pid};
use signal_hook::{
consts::{SIGCHLD, SIGINT, SIGTERM},
iterator::Signals,
Expand All @@ -12,15 +12,27 @@ pub enum Error {
SpawnChild(std::io::Error),
}

pub fn relaunch_if_pid1() -> Result<(), Error> {
if std::process::id() == 1 {
pub fn relaunch_if_pid1(option: Pid1Settings) -> Result<(), Error> {
let pid = std::process::id();
if pid == 1 {
let child = relaunch()?;
if option.log {
println!("pid1-rs: Process running as PID 1");
psibi marked this conversation as resolved.
Show resolved Hide resolved
}
pid1_handling(Some(child))
} else {
if option.log {
eprintln!("pid1-rs: Process not running as Pid 1: PID {pid}");
}
Ok(())
}
}

#[derive(Debug, Default, Copy, Clone)]
pub struct Pid1Settings {
pub log: bool,
}

fn relaunch() -> Result<Child, Error> {
let exe = std::env::current_exe().unwrap();
let args = std::env::args_os().skip(1).collect::<Vec<_>>();
Expand All @@ -31,17 +43,36 @@ fn relaunch() -> Result<Child, Error> {
}

fn pid1_handling(child: Option<Child>) -> ! {
let child = child.map(|x| x.id());
let mut signals = Signals::new([SIGTERM, SIGINT, SIGCHLD]).unwrap();

let child = child.map(|x| x.id());
struct ExitStatus {
pid: Pid,
exit_code: i32,
}

loop {
for signal in signals.pending() {
for signal in signals.forever() {
Copy link
Member Author

@psibi psibi Sep 26, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This fixes the busy loop issue. Earlier the above loop caused 100% cpu consumption.

This comment was marked as duplicate.

if signal == SIGTERM || signal == SIGINT {
// TODO: Should forward these signals to the
// application and then force kill the application
// after certain time.
graceful_shutdown();
}
if signal == SIGCHLD {
let pid = match nix::sys::wait::wait().unwrap() {
WaitStatus::Exited(pid, _) => Some(pid),
WaitStatus::Signaled(pid, _, _) => Some(pid),
WaitStatus::Exited(pid, exit_code) => {
let exit_status = ExitStatus { pid, exit_code };
Some(exit_status)
}
WaitStatus::Signaled(pid, signal, _) => {
let exit_status = ExitStatus {
pid,
// Translate signal to exit code
exit_code: signal as i32 + 128,

This comment was marked as duplicate.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The process can exit because of two reasons:

  • Normal exit
  • Exit via Unix signals

In case of Unix signals, we remap it to proper exit code.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I get that, but you seem to be conflating the process ID and the exit status.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you think renaming it from ExitStatus to ProcessStatus will be helpful ?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you look over the code one more time from scratch? I think you'll see what I'm talking about: in some branches you're returning a process ID, in some an exit code, and in some a signal, but assigning all of them to a variable called pid.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have gone ahead with some renaming which will hopefully help with the confusion.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you look over the code one more time from scratch? I think you'll see what I'm talking about: in some branches you're returning a process ID, in some an exit code, and in some a signal, but assigning all of them to a variable called pid.

Sure, I will let you know once it's ready for review. Thanks!

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Never mind, I see my confusion, you were completely correct. Sorry for the noise.

};
Some(exit_status)
}
WaitStatus::Stopped(_, _) => None,
WaitStatus::PtraceEvent(_, _, _) => None,
WaitStatus::PtraceSyscall(_) => None,
Expand All @@ -50,10 +81,12 @@ fn pid1_handling(child: Option<Child>) -> ! {
};
(|| {
let child = child?;
let pid = pid?;
let child_exit_status = pid?;

This comment was marked as duplicate.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sorry for the confusing name, I have renamed it. Can you see if that helps ?

let pid = child_exit_status.pid;
let pid = u32::try_from(pid.as_raw()).ok()?;
if pid == child {
graceful_shutdown()
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Calling the abort call previously resulted in the segmentation fault issue.

This comment was marked as duplicate.

// Propagate child exit status code
std::process::exit(child_exit_status.exit_code);
}
Some(())
})();
Expand Down
Loading