trace_events: add internal binding for intrinsic trace

[jasnell]

Continuation (and completion) of the work started in #20618 ...

Backports the intrinsic trace/isTraceCategoryEnabled functions that have landed upstream in V8 (v8/v8@0dd3390). These will need to be floated until we get a version of V8 that includes the commit.

This currently does not add code that uses the internal functions. That will come as separate PRs.

Note: There is currently one place in the javascript where we emit trace events through the old binding (in lib/internal/trace_events_async_hooks.js) but switching would require a change in the actual output format, which will break some tooling. We'll want to take it a bit slow on that. Will make those changes in a separate PR.

/cc @nodejs/v8 @nodejs/v8-update @nodejs/trace-events

Checklist

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

[nodejs-github-bot]

@jasnell build started: https://ci.nodejs.org/blue/organizations/jenkins/node-test-pull-request-lite-pipeline/detail/node-test-pull-request-lite-pipeline/319/pipeline

[jasnell]

Note: the benchmark added by this can occasionally Abort due to known issues in the NodeTraceWriter implementation (/cc @addaleax@eugeneo). This segfault is caused by the benchmark emitting trace events too quickly. The segfault is caused by any flaws in this particular PR.

[jasnell]

CI: https://ci.nodejs.org/job/node-test-pull-request/15955/

[jasnell]

Benchmark comparison with the existing legacy emit and category checks:

(trace and isTraceCategoryEnabled are the new versions)

misc/trace.js method="trace" n=100000: 509,507.6314919208 misc/trace.js method="emit" n=100000: 477,461.6085892631 misc/trace.js method="isTraceCategoryEnabled" n=100000: 4,949,474.281690222 misc/trace.js method="categoryGroupEnabled" n=100000: 1,904,456.030985652 

[jasnell]

New CI: https://ci.nodejs.org/job/node-test-pull-request/15956/

[jasnell]

@nodejs/build ... the OSX build bots are failing unfortunately... see https://ci.nodejs.org/job/node-test-commit-osx/19929/nodes=osx1012/console

[Trott]

@nodejs/build ... the OSX build bots are failing unfortunately... see https://ci.nodejs.org/job/node-test-commit-osx/19929/nodes=osx1012/console

Resume Build: https://ci.nodejs.org/job/node-test-pull-request/15957/

@jasnell: It seems to have gotten farther on in osx1012 than last two times. Since the console you link to shows git timing out, I'd chalk it up to network problems between the host and GitHub and hope it doesn't recur.

[jasnell]

CI is green!

[ofrobots]

LGTM w/ nit.

[ofrobots]

This doest not need to bumped given the bump in common.gypi above.

[ofrobots]

/cc @nodejs/v8-update If this lands before the V8 6.8 upgrade, this patch will need to be re-floated.

[jasnell]

/cc @nodejs/v8-update If this lands before the V8 6.8 upgrade, this patch will need to be re-floated.

Also note, there are some minor variances between this and the commit that landed in v8 master due to some minor internal API changes that were made. It's nothing that impacts Node.js' use of the API but does have some variances inside builtins_trace.cc and v8's bootstrapper.cc. I'm not sure exactly which version of V8 those API changes landed in so when relanding this there may need to be some fixups.

[jasnell]

Landed in a706456 and f7454a5

[MylesBorins]

Should this be backported to v8.x-staging? If yes please follow the guide and raise a backport PR, if not let me know or add the dont-land-on label.

[jasnell]

Putting the don't land labels for v8 and v6. This can go into v10 but let's hold off on bringing it back to v8 for the time being.