Bug 168627 – Intrinsicify parseInt

• Component: JavaScriptCore
• Status: RESOLVED FIXED

[Saam Barati]

...

[Saam Barati]

box2D calls parseInt quite often with a single integer argument. We should be good at emitting good code here. We could also be good about removing various type checks for strings, and determining if the radix is a constant.

Implementing a dumbed down parseInt Intrinsic is a 5-6% progression on box2d.

[Saam Barati]

Created attachment 302289 [details]
WIP

Note: the patch currently has some register allocation bits in it from a different bug I'm working on.

[Saam Barati]

Created attachment 302565 [details]
WIP

Might be the patch. Still need to write tests and a changelog.

[WebKit Commit Bot]

Attachment 302565 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/runtime/ParseInt.h:161:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
ERROR: Source/JavaScriptCore/runtime/ParseInt.h:164:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
ERROR: Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h:2942:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
ERROR: Source/JavaScriptCore/dfg/DFGConstantFoldingPhase.cpp:619:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
Total errors found: 4 in 25 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Build Bot]

Comment on attachment 302565 [details]
WIP

Attachment 302565 [details] did not pass mac-debug-ews (mac):
Output: http://webkit-queues.webkit.org/results/3181060

New failing tests:
media/modern-media-controls/volume-down-support/volume-down-support.html

[Build Bot]

Created attachment 302579 [details]
Archive of layout-test-results from ews117 for mac-elcapitan

The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews117  Port: mac-elcapitan  Platform: Mac OS X 10.11.6

[Saam Barati]

Created attachment 302598 [details]
patch

[WebKit Commit Bot]

Attachment 302598 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/runtime/ParseInt.h:161:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
ERROR: Source/JavaScriptCore/runtime/ParseInt.h:164:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
ERROR: Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h:2941:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
ERROR: Source/JavaScriptCore/dfg/DFGConstantFoldingPhase.cpp:619:  Tests for true/false, null/non-null, and zero/non-zero should all be done without equality comparisons.  [readability/comparison_to_zero] [5]
Total errors found: 4 in 29 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[WebKit Commit Bot]

Comment on attachment 302598 [details]
patch

Clearing flags on attachment: 302598

Committed r212939: <http://trac.webkit.org/changeset/212939>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Darin Adler]

Comment on attachment 302598 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=302598&action=review

> Source/JavaScriptCore/dfg/DFGOperations.cpp:895
> +    // This version is as if radix was undefined. Hence, undefined.toNumber() === 0.
> +    return parseIntResult(parseInt(viewWithString.view, radix));

The comment here seems wrong.

[Saam Barati]

Comment on attachment 302598 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=302598&action=review

>> Source/JavaScriptCore/dfg/DFGOperations.cpp:895
>> +    return parseIntResult(parseInt(viewWithString.view, radix));
> 
> The comment here seems wrong.

Oops. Thanks for pointing this out.
(I actually spotted this locally after I uploaded the patch but forgot to upload a new patch before landing. Will fix when I'm back in front of my computer.)

[Saam Barati]

Removed comment in:
https://trac.webkit.org/changeset/212962