Fix early pipe shutdown with Windows ProcessMonitor #3726
Merged
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
SeanTAllen
added
the
changelog - fixed
|
@kulibali here's the PR for testing. |
SeanTAllen
commented
Feb 28, 2021
| @@ -172,14 +178,17 @@ class _Pipe | |||
| bytes_to_read, addressof bytes_read, USize(0)) | |||
| let winerr = @GetLastError[I32]() | |||
| if not ok then | |||
| if (winerr == _ERRORBROKENPIPE()) or (winerr == _ERRORNODATA()) then | |||
There was a problem hiding this comment.
_ERRORNODATA should be impossible to get here based on doing non blocking and us having previously peeked to know there is data to read.
SeanTAllen
force-pushed
the
3674
branch
5 times, most recently
from
3ee2d26 to
be44d0d
Compare
This commit fixes a fun bug that results from multiple bits coming together. First, the ProcessMonitor when running on Windows doesn't using ASIO. Instead of having events pushed to it, we poll periodically to see if there's anything to read or if we can write any pending writes. Because of the polling, it is possible that we will attempt to read when there's no data in the pipe. When using non-blocking IO with pipes, a read that would return 0 bytes returns an error and sets the error code to a no data error code. We were interpreting the no data code as it should be when using blocking io. When using non-blocking io, there error communicates "no data right now, try again later" which wasn't what we were doing. The same incorrect handling of the no data error code is also present in the write code for Windows and is fixed by this commit. Fixes #3674
chalcolith
approved these changes
Feb 28, 2021
There was a problem hiding this comment.
Thanks @SeanTAllen! I have run the corrall test suite a couple of dozen times in a row without failures. Previously they woud fail quite often.
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 commit fixes a fun bug that results from multiple bits coming together.
First, the ProcessMonitor when running on Windows doesn't using ASIO. Instead
of having events pushed to it, we poll periodically to see if there's anything
to read or if we can write any pending writes.
Because of the polling, it is possible that we will attempt to read when there's
no data in the pipe. When using non-blocking IO with pipes, a read that would return
0 bytes returns an error and sets the error code to a no data error code.
We were interpreting the no data code as it should be when using blocking io. When
using non-blocking io, there error communicates "no data right now, try again later"
which wasn't what we were doing.
The same incorrect handling of the "no data" error code was present in the write code
path as well. That has also been fixed.
Fixes #3674