test: increase coverage of async_hooks

[DavidCai1111]

• increase coverage of lib/async_hooks.js
• rename test/async_hooks/test-embedder.api.async-event.*.js to test/async_hooks/test-embedder.api.async-resource.*.js

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• tests and/or benchmarks are included
• commit message follows commit guidelines

Affected core subsystem(s)

async_hooks

[kunalspathak]

Unrelated to this PR, but I noticed that we async-hooks test don't run for windows. Any reason why? @trevnorris , @addaleax .

[addaleax]

I think this test can go into parallel (it doesn’t use any of the async hook test helpers here)

[DavidCai1111]

@addaleax Done, PTAL :=)

[AndreasMadsen]

Unrelated to this PR, but I noticed that we async-hooks test don't run for windows. Any reason why? @trevnorris , @addaleax .

Sigh, I guess it is because it doesn't appear in https://github.com/nodejs/node/blob/master/vcbuild.bat#L68L82

[AndreasMadsen]

A resource must be an object.

[DavidCai1111]

@AndreasMadsen Updated, PTAL :=)

[kunalspathak]

Sigh, I guess it is because it doesn't appear in https://github.com/nodejs/node/blob/master/vcbuild.bat#L68L82

Yep, that was my question 😄 . Why is it not present in vcbuild.bat?

[AndreasMadsen]

/cc @refack regarding vcbuild.bat

[refack]

/cc @refack regarding vcbuild.bat

AFAIK there's no reason except nobody bothered to add it.

[refack]

Cross-ref: #13378

[DavidCai1111]

CI: https://ci.nodejs.org/job/node-test-pull-request/8424/

[AndreasMadsen]

LGTM, I don't think we need to wait for the windows PR to be merged. These are mostly pure JS logic related tests.

[trevnorris]

Thanks for doing all this. Test coverage for async_hooks is important and happy to see more of it coming in. I have two suggestions and one of your test additions made me aware of a flaw in the AsyncResource() constructor logic that needs to be fixed.

[trevnorris]

There's the problem where if test_init_callback is never hit then the common.mustCall() will never run. Instead you'll need to do it like this:

constoninitMustCall=common.mustCall(()=>{thrownewError('test_init_callback');});switch(process.argv[2]){case'test_init_callback':
initHooks({oninit: oninitMustCall}).enable();

[DavidCai1111]

😕 But IMO the actual intention here is just to ensure that the common.mustCall() will run only when the test_init_callback case is hit? Every switch case will be hit in different process because of spawnSync()s below.

[trevnorris]

But IMO the actual intention here is just to ensure that the common.mustCall() will run only when the test_init_callback case is hit?

Whoops. Missed that. You're correct.

[trevnorris]

I screwed up the logic here. It shouldn't return early before the arguments are checked. Otherwise it'll cause an async hook stack corruption. Here's the minimal test case:

new(require('async_hooks').AsyncResource)().emitBefore();

Here's the diff for the fix. Feel free to add it to this PR along with the above test case:

diff --git a/lib/async_hooks.js b/lib/async_hooks.js
index d5d5407..aec73ed 100644
--- a/lib/async_hooks.js+++ b/lib/async_hooks.js@@ -205,15 +205,15 @@ class AsyncResource {
}
this[trigger_id_symbol] = triggerId;
- // Return immediately if there's nothing to do.- if (async_hook_fields[kInit] === 0)- return;-
if (typeof type !== 'string' || type.length <= 0)
throw new TypeError('type must be a string with length > 0');
if (!Number.isSafeInteger(triggerId) || triggerId < 0)
throw new RangeError('triggerId must be an unsigned integer');
+ // Return immediately if there's nothing to do.+ if (async_hook_fields[kInit] === 0)+ return;+
processing_hook = true;
for (var i = 0; i < active_hooks_array.length; i++) {
if (typeof active_hooks_array[i][init_symbol] === 'function') {

[DavidCai1111]

OK, i'll do it in this PR :=)

[DavidCai1111]

@trevnorris Done, added the diff to this PR, PTAL :=)

[trevnorris]

To make sure these run I' possibly do this:

constemitBeforeMustCall=common.mustCall(()=>async_hooks.emitBefore(-1));switch(process.argv[2]){case'test_invalid_async_id':
emitBeforeMustCall();

[refack]

Needs a rebase (just tried locally it goes smooth)

[DavidCai1111]

@refack I rebased it locally and the result shows that it goes very smooth 😃

[refack]

@refack I rebased it locally and the result shows that it goes very smooth 😃

Apparently GitHub is a whiny little ...

[DavidCai1111]

CI: https://ci.nodejs.org/job/node-test-pull-request/8503/

[DavidCai1111]

@refack Now no conflicts checking and CI both seems good ...

[DavidCai1111]

@trevnorris PTAL

[trevnorris]

@DavidCai1993 Much thanks for the fix.

[DavidCai1111]

Landed in 35353a4