Making evented I/O work on macOS #5286
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
andrewrk
reviewed
May 7, 2020
| @@ -503,7 +503,7 @@ pub const Loop = struct { | |||
| } | |||
| } | |||
|
|
|||
| pub async fn bsdWaitKev(self: *Loop, ident: usize, filter: i16, fflags: u32) void { | |||
| pub async fn bsdWaitKev(self: *Loop, ident: usize, filter: i16, flags: u16) void { | |||
There was a problem hiding this comment.
No async needed here (this is leftover from before the async/await rewrite)
Suggested change
| pub async fn bsdWaitKev(self: *Loop, ident: usize, filter: i16, flags: u16) void { | |
| pub fn bsdWaitKev(self: *Loop, ident: usize, filter: i16, flags: u16) void { |
|
|
||
| defer { | ||
| // If the kevent was set to be ONESHOT, it doesn't need to be deleted manually. | ||
| if (flags & os.EV_ONESHOT != 0) { |
There was a problem hiding this comment.
Do we maybe want to assert that flags always contains EV_ONESHOT and document the requirement in doc comments? It's true in all the callsites of this function.
| } | ||
| } | ||
|
|
||
| /// resume_node must live longer than the anyframe that it holds a reference to. | ||
| pub fn bsdAddKev(self: *Loop, resume_node: *ResumeNode.Basic, ident: usize, filter: i16, fflags: u32) !void { | ||
| pub fn bsdAddKev(self: *Loop, resume_node: *ResumeNode.Basic, ident: usize, filter: i16, flags: u16) !void { |
There was a problem hiding this comment.
Wow nice catch on the filter flags, I can't believe I was passing EV_ONESHOT to the wrong parameter 😰
| .dragonfly, | ||
| => self.fs_thread.wait(), | ||
| else => {}, | ||
| if (!builtin.single_threaded) { |
|
For future reference, this is the moment when we made the code work https://clips.twitch.tv/AssiduousMotionlessScorpionDansGame |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR contains two reasonably solid fixes and one tentative fix that should be discussed first.
Fix 1
Simple fix on a crash caused by a missing
builtin.single_threadedcheck.Fix 2
Fixes a race condition crash based on
EV_ONESHOTbeing erroneusly passed as a filter_flag (thanks JX7P from IRC, we were running in circles :D)Fix 3
As of now, when calling
connectwe were checking for the socket to be either readable or writable. In kqueue this doesn't seem to be possible, which exposes the problem that if the other host is not sending anything, the connection never becomes readable. This makes me question whether we really need to check for both conditions before proceeding. For example this deadlocks my Redis client because it's the client that's supposed to speak first.Related issue: #5278.
Thanks @haze @FireFox317 @SuperAuguste, chaplchapl and @alexnask .