-
Notifications
You must be signed in to change notification settings - Fork 29.8k
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
events: refactor to use primordials in lib/events #38117
Conversation
Benchmarks should be checked on this before landing |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Some significant perf regressions:
confidence improvement accuracy (*) (**) (***)
events/ee-add-remove.jsn=1000000 *** -27.64 % ±3.09% ±4.13% ±5.40%
events/ee-emit.jslisteners=10 argc=0 n=2000000 0.28 % ±1.22% ±1.63% ±2.12%
events/ee-emit.jslisteners=10 argc=10 n=2000000 0.44 % ±1.47% ±1.96% ±2.58%
events/ee-emit.jslisteners=10 argc=2 n=2000000 0.37 % ±1.27% ±1.69% ±2.20%
events/ee-emit.jslisteners=10 argc=4 n=2000000 0.04 % ±1.07% ±1.43% ±1.86%
events/ee-emit.jslisteners=1 argc=0 n=2000000 -2.06 % ±2.67% ±3.55% ±4.62%
events/ee-emit.jslisteners=1 argc=10 n=2000000 -1.66 % ±3.41% ±4.54% ±5.91%
events/ee-emit.jslisteners=1 argc=2 n=2000000 -4.38 % ±4.59% ±6.11% ±7.96%
events/ee-emit.jslisteners=1 argc=4 n=2000000 -1.97 % ±3.47% ±4.63% ±6.03%
events/ee-emit.jslisteners=5 argc=0 n=2000000 -0.08 % ±2.29% ±3.04% ±3.96%
events/ee-emit.jslisteners=5 argc=10 n=2000000 -0.77 % ±1.88% ±2.51% ±3.26%
events/ee-emit.jslisteners=5 argc=2 n=2000000 -0.76 % ±1.97% ±2.63% ±3.43%
events/ee-emit.jslisteners=5 argc=4 n=2000000 -0.21 % ±2.09% ±2.78% ±3.62%
events/ee-listener-count-on-prototype.jsn=50000000 * -5.06 % ±4.62% ±6.21% ±8.20%
events/ee-listeners.jsraw='false' listeners=50 n=5000000 1.91 % ±3.18% ±4.26% ±5.58%
events/ee-listeners.jsraw='false' listeners=5 n=5000000 -1.18 % ±3.36% ±4.50% ±5.91%
events/ee-listeners.jsraw='true' listeners=50 n=5000000 1.67 % ±2.85% ±3.80% ±4.98%
events/ee-listeners.jsraw='true' listeners=5 n=5000000 2.24 % ±3.54% ±4.71% ±6.14%
events/ee-once.jsargc=0 n=20000000 *** -6.95 % ±1.40% ±1.86% ±2.43%
events/ee-once.jsargc=1 n=20000000 *** -9.55 % ±1.17% ±1.56% ±2.03%
events/ee-once.jsargc=4 n=20000000 *** -9.36 % ±4.11% ±5.47% ±7.12%
events/ee-once.jsargc=5 n=20000000 *** -8.47 % ±2.74% ±3.64% ±4.74%
events/eventtarget.jslisteners=10 n=1000000 * 3.56 % ±3.01% ±4.02% ±5.28%
events/eventtarget.jslisteners=1 n=1000000 -3.34 % ±4.28% ±5.74% ±7.54%
events/eventtarget.jslisteners=5 n=1000000 0.94 % ±2.38% ±3.16% ±4.12%
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm -1 to land this with regressions.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These are the primordials causing the Regressions in my previous commit
ArrayPrototypeUnshift
and ArrayPrototypePush
in _addListener()
FunctionPrototypeCall
and ReflectApply
in onceWrapper()
FunctionPrototypeBind
in _onceWrap()
ArrayPrototypeShift
in removeListener()
ArrayPrototypeShift
in on()
I have removed these and benchmarked events again (locally on my computer).
The regressions are gone.
Kindly re-run benchmarks once again.
Sorry for leaving a review instead of comment. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There does not seem to be any significant performance regressions in the bechmark run.
14:30:14 confidence improvement accuracy (*) (**) (***)
14:30:15 events/ee-add-remove.js n=1000000 -1.98 % ±5.20% ±6.92% ±9.01%
14:30:15 events/ee-emit.js listeners=10 argc=0 n=2000000 0.07 % ±1.19% ±1.58% ±2.06%
14:30:15 events/ee-emit.js listeners=10 argc=10 n=2000000 1.95 % ±3.77% ±5.05% ±6.64%
14:30:15 events/ee-emit.js listeners=10 argc=2 n=2000000 0.08 % ±0.81% ±1.08% ±1.40%
14:30:15 events/ee-emit.js listeners=10 argc=4 n=2000000 0.40 % ±0.98% ±1.30% ±1.69%
14:30:15 events/ee-emit.js listeners=1 argc=0 n=2000000 1.63 % ±4.24% ±5.64% ±7.34%
14:30:15 events/ee-emit.js listeners=1 argc=10 n=2000000 -0.60 % ±3.49% ±4.64% ±6.04%
14:30:15 events/ee-emit.js listeners=1 argc=2 n=2000000 -1.54 % ±4.00% ±5.32% ±6.93%
14:30:15 events/ee-emit.js listeners=1 argc=4 n=2000000 -5.19 % ±7.62% ±10.23% ±13.53%
14:30:15 events/ee-emit.js listeners=5 argc=0 n=2000000 -1.80 % ±2.15% ±2.86% ±3.73%
14:30:15 events/ee-emit.js listeners=5 argc=10 n=2000000 -0.44 % ±2.13% ±2.83% ±3.69%
14:30:15 events/ee-emit.js listeners=5 argc=2 n=2000000 0.56 % ±2.15% ±2.86% ±3.73%
14:30:15 events/ee-emit.js listeners=5 argc=4 n=2000000 -1.72 % ±2.54% ±3.38% ±4.40%
14:30:15 events/ee-listener-count-on-prototype.js n=50000000 0.19 % ±5.10% ±6.79% ±8.85%
14:30:15 events/ee-listeners.js raw='false' listeners=50 n=5000000 0.57 % ±1.29% ±1.72% ±2.23%
14:30:15 events/ee-listeners.js raw='false' listeners=5 n=5000000 0.65 % ±1.37% ±1.82% ±2.37%
14:30:15 events/ee-listeners.js raw='true' listeners=50 n=5000000 -2.61 % ±3.80% ±5.10% ±6.73%
14:30:15 events/ee-listeners.js raw='true' listeners=5 n=5000000 -1.70 % ±5.65% ±7.58% ±9.99%
14:30:15 events/ee-once.js argc=0 n=20000000 1.25 % ±1.45% ±1.94% ±2.54%
14:30:15 events/ee-once.js argc=1 n=20000000 -1.49 % ±2.15% ±2.88% ±3.79%
14:30:15 events/ee-once.js argc=4 n=20000000 -0.29 % ±1.31% ±1.75% ±2.28%
14:30:15 events/ee-once.js argc=5 n=20000000 0.09 % ±1.34% ±1.79% ±2.32%
14:30:15 events/eventtarget.js listeners=10 n=1000000 0.94 % ±2.36% ±3.15% ±4.10%
14:30:15 events/eventtarget.js listeners=1 n=1000000 0.07 % ±1.75% ±2.34% ±3.05%
14:30:15 events/eventtarget.js listeners=5 n=1000000 * 7.80 % ±6.65% ±8.95% ±11.85%
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Benchmark results look promising! It's probably a good idea to wait for #38248 to land first as it's going to create conflicts and we want to include it in the v16.0.0 release.
Once this has been rebased on top of #38248 (after it has landed), we should also run HTTP benchmark to make sure we're not introducing hidden perf regressions. |
We need to validate that http did not regress. |
@marsonya if you want to rebase on top of master to fix the git conflict, we can spawn some benchmarks to see where this PR stands in term of perf regressions/improvment. |
Replace code that's vulnerable to Prototype Pollution with Primordials.
Benchmark CI (events): https://ci.nodejs.org/view/Node.js%20benchmark/job/benchmark-node-micro-benchmarks/1019/ Results
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Benchmark results look OK.
I would like to do some checks of my own before landing. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Temporary -1 until I can test this as well.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
lgtm on my end
Landed in 13ec317 |
Replace code that's vulnerable to Prototype Pollution with Primordials. PR-URL: #38117 Reviewed-By: Antoine du Hamel <[email protected]> Reviewed-By: Matteo Collina <[email protected]> Reviewed-By: Darshan Sen <[email protected]> Reviewed-By: James M Snell <[email protected]>
Replace code that's vulnerable to Prototype Pollution with Primordials. PR-URL: #38117 Reviewed-By: Antoine du Hamel <[email protected]> Reviewed-By: Matteo Collina <[email protected]> Reviewed-By: Darshan Sen <[email protected]> Reviewed-By: James M Snell <[email protected]>
I'm going to mark this (and any other "use primordials" I come across) as requiring a manual backport due to the recent discussions around use of primordials in current and the fact that v14.x has an older version of V8 so if we did land these we would probably want to check benchmarks. |
Replace code that's vulnerable to Prototype Pollution with Primordials.