1691194 - Inline more large self-hosted functions (substring-like)

Status: RESOLVED FIXED

(Core :: JavaScript Engine: JIT, task)

Comment 1

[Tom S. (please needinfo tschuster)]

Attachment: Bug 1691194 - Add more logging for non-inlinenable functions in TrialInlining. r?iain

Comment 2

[Tom S. (please needinfo tschuster)]

Attachment: Bug 1691194 - Inline substring-like functions. r?iain

They are just relatively slim wrappers around native SubstringKernel

Depends on D104307

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/1918677e9b5b
Add more logging for non-inlinenable functions in TrialInlining. r=iain
https://hg.mozilla.org/integration/autoland/rev/ff11b7b68ed0
Inline substring-like functions. r=iain

Comment 4

[Atila Butkovits]

Backed out for causing Spidermonkey failures.

Backout link: Backed out for causing Spidermonkey failures.

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&selectedTaskRun=PzFIYocgRZCKnnhdLtVCjw.0&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel%2Crunnable&revision=ff11b7b68ed04dc134191a9b6e6f2c19daa49ed0&searchStr=spidermonkey

Failure log: https://treeherder.mozilla.org/logviewer?job_id=329360655&repo=autoland&lineNumber=10902
https://treeherder.mozilla.org/logviewer?job_id=329363984&repo=autoland&lineNumber=11753

Comment 5

[Pulsebot]

Backout by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/fe2bff61b4ef
Backed out 2 changesets for causing Spidermonkey failures. CLOSED TREE

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/c4ee20b1c31b
Add more logging for non-inlinenable functions in TrialInlining. r=iain
https://hg.mozilla.org/integration/autoland/rev/33811f651b65
Inline substring-like functions. r=iain

Comment 7

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/c4ee20b1c31b
https://hg.mozilla.org/mozilla-central/rev/33811f651b65

Comment 8

[Tom S. (please needinfo tschuster)]

There are some more things we should probably inline in the future, but aren't necessarily useful or possible right now.

• ArrayIterator#next: I think we could inline this, but we can't currently eliminate the result object creation. We would need to improve scalar replacement for this.
• Other iterators: Probably similar
• Array#map, Array#forEach etc.: These methods currently use arguments[1] and need bug 1688033 with inlining support.

We should maybe create a checklist of all self-hosted builtins and decide case by case.