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

Tokio v1.0.0 support. #33

Closed
wants to merge 5 commits into from
Closed
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
16 changes: 7 additions & 9 deletions src/lib.rs
Expand Up @@ -222,15 +222,13 @@ impl AsyncRead for Serial {
cx: &mut Context<'_>,
buf: &mut ReadBuf<'_>,
) -> Poll<io::Result<()>> {
loop {
let mut guard = ready!(self.io.poll_read_ready(cx))?;
match guard.try_io(|_| {
let read = self.io.get_ref().read(buf.initialize_unfilled())?;
return Ok(buf.advance(read));
}) {
Ok(result) => return Poll::Ready(result),
Err(_would_block) => continue,
}
let mut guard = ready!(self.io.poll_read_ready(cx))?;
match guard.try_io(|_| {
let read = self.io.get_ref().read(buf.initialize_unfilled())?;
return Ok(buf.advance(read));
}) {
Ok(result) => return Poll::Ready(result),
Err(_would_block) => return Poll::Pending,

Choose a reason for hiding this comment

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

This breaks as the read() can return would block in which case you go pending without the async fd actually registered with the runtime; The original version with the loop is correct as read() returns would_block then the poll_read_ready() call should return Pending and everything should work nicely :).

fwiw i hit this problem in practise while playing with this branch and a fdti usb dongle in linux

Choose a reason for hiding this comment

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

I'm confused about this: if the read returns EWOULDBLOCK doesn't that imply a bug somewhere else? We can't arrive at line 227 without being signaled by tokio that there are in fact some bytes to read. I haven't been using this branch but I have some code that is morally equiv that I have been testing with ftdi dongles (snap!) and I haven't seen this condition triggered (yet).

Copy link
Contributor

Choose a reason for hiding this comment

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

It is possible to get a false positive from poll_read_ready and get the would block error, but the snippet is still incorrect — you have to loop around and call poll_read_ready again if that happens.

Copy link

Choose a reason for hiding this comment

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

I see what you are saying. Thanks for making that connection. I'd conflated the clearing of the tokio-side readiness state (which is handled by try_io) with the need to poll again to get registered. I also completely misread the sample code for AsyncFd. I guess I'd just gotten (un)lucky with my testing :)

Copy link

@mateuszkj mateuszkj Jun 13, 2021

Choose a reason for hiding this comment

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

@mnbbrown Please look at PR: mnbbrown#1

We got bug with "hanging" usart and adding loops fixed the problem.

Copy link

Choose a reason for hiding this comment

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

I can confirm I have been seeing this bug too (when working on tokio-modbus's update to tokio-1.0) and rfael's patch fixed it.

}
}
}
Expand Down