core: memory limits & callbacks

[mraerino]

Fixes #6916

[CLAassistant]

All committers have signed the CLA.

[piscisaureus]

Maybe you forgot to set V8_FROM_SOURCE?

[mraerino]

@piscisaureus this needs a release in rusty_v8. were you going to make one soon?

[piscisaureus]

Yes this week.

[mraerino]

did this fail because of a flaky test?

[ry]

@mraerino Currently I don't think we have any flaky tests - but who knows it's always possible. Try pushing an empty commit to re-trigger CI.

My guess would be that it's related to your patch. I've pulled out the relevant failure logs from Github Actions:

2020-08-10T16:51:55.4750378Z test timeoutCancelSuccess ... test lock_check_ok2 ... ok
2020-08-10T16:51:55.7768303Z error: Uncaught AssertionError: Assertion failed.
2020-08-10T16:51:55.7769007Z ok (602ms)
2020-08-10T16:51:55.7779412Z     at assert (rt/06_util.js:33:13)
2020-08-10T16:51:55.7784881Z     at setGlobalTimeout (rt/11_timers.js:305:5)
2020-08-10T16:51:55.7820619Z     at setOrClearGlobalTimeout (rt/11_timers.js:349:7)
2020-08-10T16:51:55.8065898Z     at unschedule (rt/11_timers.js:398:9)
2020-08-10T16:51:55.8066362Z     at clearTimer (rt/11_timers.js:514:5)
2020-08-10T16:51:55.8067075Z     at clearTimeout (rt/11_timers.js:523:5)
2020-08-10T16:51:55.8067667Z     at timeoutCancelMultiple (file:///D:/a/deno/deno/cli/tests/unit/timers_test.ts:84:3)
2020-08-10T16:51:55.8068022Z     at asyncOpSanitizer (rt/40_testing.js:34:13)
2020-08-10T16:51:55.8068751Z     at Object.resourceSanitizer [as fn] (rt/40_testing.js:68:13)
2020-08-10T16:51:55.8070940Z     at TestRunner.[Symbol.asyncIterator] (rt/40_testing.js:240:24)
2020-08-10T16:51:55.8071934Z error: Uncaught ConnectionReset: An existing connection was forcibly closed by the remote host. (os error 10054)  
2020-08-10T16:51:55.8072673Z     at unwrapResponse (rt/10_dispatch_minimal.js:53:13)
2020-08-10T16:51:55.8090882Z     at sendAsync (rt/10_dispatch_minimal.js:96:12)
2020-08-10T16:51:55.8091311Z     at async read (rt/12_io.js:99:19)
2020-08-10T16:51:55.8091647Z     at async readDelim (file:///D:/a/deno/deno/std/io/bufio.ts:649:20)
2020-08-10T16:51:55.8091997Z     at async readStringDelim (file:///D:/a/deno/deno/std/io/bufio.ts:699:20)
2020-08-10T16:51:55.8092339Z     at async readLines (file:///D:/a/deno/deno/std/io/bufio.ts:709:3)
2020-08-10T16:51:55.8092694Z     at async runTestsForPermissionSet (file:///D:/a/deno/deno/cli/tests/unit/unit_test_runner.ts:142:22)
2020-08-10T16:51:55.8093047Z     at async masterRunnerMain (file:///D:/a/deno/deno/cli/tests/unit/unit_test_runner.ts:203:20)
2020-08-10T16:51:55.8285318Z test timeoutCancelMultiple ... test js_unit_tests ... FAILED

[mraerino]

ok, i can't explain the failures. between the two runs different things failed, so they seem non-deterministic.

my changes only add new interfaces to deno_core that are not being used in cli

[ry]

@mraerino Looks like it was indeed something flaky. Build is good. Reviewing now...

[ry]

Looks good - thank you @mraerino !

Just going to give you a chance to add AtomicUsize to the test - or not if you disagree.

[piscisaureus]

Curious question - what does a Pin around a RefCell (boxed or not) achieve?
I'd think that, since a RefCell allows interior mutability, this kinda negates the effect of the pin.

(In other words - why do you need to pin this?)

[mraerino]

good point. removed.

[piscisaureus]

This is good.

[piscisaureus]

LGTM

[mraerino]

Now that you say this:

Ah I see you've limited it to support only one callback.. That's ok.

I realized a small thing was missing: #7025