Properly handle autoread #35
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
Motivation: NIO's backpressure management systems rely on catching calls to read() in the ChannelPipeline. This only works if your Channel implementation actually makes those calls to read()! Sadly, SSHChildChannel did not: we instead just recursed directly into the Channel's read0(). While I'm here I also noticed that we produced a lot of channelReadComplete messages in the pipeline unnecessarily. We did this in the service of having a read "fast-path" where new frames would be automatically delivered into the ChannelPipeline without being buffered until channelReadComplete. This fast-path is probably inadvisable: I don't think it provided a meaningful performance benefit anyway, especially as the removal of that fast-path allowed us to suppress unnecessary calls to channelReadComplete, and therefore unnecessary calls to read() as well. Modifications: - Forced autoRead through the pipeline. - Removed read() fast-path. - Suppressed channelReadComplete if no frames were delivered. - Updated tests. Result: Clearer I/O path, better auto-read handling.
Lukasa
added
the
patch-version-bump-only
artemredkin
approved these changes
Aug 13, 2020
|
Sure is! |
artemredkin
pushed a commit
to artemredkin/swift-nio-ssh
that referenced
this pull request
Feb 19, 2021
Motivation: NIO's backpressure management systems rely on catching calls to read() in the ChannelPipeline. This only works if your Channel implementation actually makes those calls to read()! Sadly, SSHChildChannel did not: we instead just recursed directly into the Channel's read0(). While I'm here I also noticed that we produced a lot of channelReadComplete messages in the pipeline unnecessarily. We did this in the service of having a read "fast-path" where new frames would be automatically delivered into the ChannelPipeline without being buffered until channelReadComplete. This fast-path is probably inadvisable: I don't think it provided a meaningful performance benefit anyway, especially as the removal of that fast-path allowed us to suppress unnecessary calls to channelReadComplete, and therefore unnecessary calls to read() as well. Modifications: - Forced autoRead through the pipeline. - Removed read() fast-path. - Suppressed channelReadComplete if no frames were delivered. - Updated tests. Result: Clearer I/O path, better auto-read handling.
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.
Motivation:
NIO's backpressure management systems rely on catching calls to read()
in the ChannelPipeline. This only works if your Channel implementation
actually makes those calls to read()! Sadly, SSHChildChannel did not:
we instead just recursed directly into the Channel's read0().
While I'm here I also noticed that we produced a lot of
channelReadComplete messages in the pipeline unnecessarily. We did this
in the service of having a read "fast-path" where new frames would be
automatically delivered into the ChannelPipeline without being buffered
until channelReadComplete. This fast-path is probably inadvisable: I
don't think it provided a meaningful performance benefit anyway,
especially as the removal of that fast-path allowed us to suppress
unnecessary calls to channelReadComplete, and therefore unnecessary
calls to read() as well.
Modifications:
Result:
Clearer I/O path, better auto-read handling.