Speedup ContextInstances::removeEach if there are null instances #38737
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
franz1981
force-pushed
the
main_fast_remove
branch
from
ec09241 to
82a333c
Compare
Ladicek
reviewed
Feb 12, 2024
...projects/arc/processor/src/main/java/io/quarkus/arc/processor/ContextInstancesGenerator.java
Outdated
Show resolved
Hide resolved
Ladicek
reviewed
Feb 12, 2024
...projects/arc/processor/src/main/java/io/quarkus/arc/processor/ContextInstancesGenerator.java
Outdated
Show resolved
Hide resolved
franz1981
force-pushed
the
main_fast_remove
branch
from
82a333c to
e2c22a3
Compare
franz1981
force-pushed
the
main_fast_remove
branch
from
e2c22a3 to
2cf5be0
Compare
|
@mkouba what surprise me (mostly because I'm very ignorant on the topic) is that an empty quarkus application (not even an hello world endpoint) still generate 2
I would like to understand what is deciding how many fields should be created, if possible |
This comment has been minimized.
This comment has been minimized.
franz1981
force-pushed
the
main_fast_remove
branch
from
2cf5be0 to
2ab88ae
Compare
Each field represents a bean of the given scope. Since your app is empty these beans are provided by the extensions. They are also unremovable; either by definition or marked by an extension. |
Status for workflow
|
|
@Ladicek @mkouba putting this on draft: this could be a better and more complete fix -> https://github.com/franz1981/quarkus/commits/main_lazy_ctx_instances/ |
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Outdated suggestions cannot be applied.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 change was due to:
which has a compute as
in short: is the case where there's a generated
ContextInstanceswith N > 1 instances fields but just a single one computed (ie not null).Before this PR we are going to perform 3 * N atomic volatile ops (using strong memory barriers), despite just 1 would be necessary.
The 3 atomic ops are:
ie:
nullon the instance field@mkouba @Ladicek
I have no idea which cutoff value make sense here and, as said in the previous comment at
this is another small problem, which could be solved by using
VarHandle::setRelease(..., null)on the volatile instance field, while holding the lock, but I would raise this in a separate PR.TBH this one is already decent, because it saves the 3 atomic ops to happen in one go; but is assuming that when a value is
nullis a terminal behavior and if a compute happen concurrently,removecan still observe it as uninitialized.My only concern ATM is that we have (regardless this PR) a lot of "undefined" states in the 2
ContextInstancesimpls, which are prone to both optimizations/concurrency misbehaviour, and I would like, in a separate issue, to state which ones are admissible/wrong/rare but still admissible based on how we use things.