Bug 233905 – TypedArray prototype set should go down the fast path when using non clamped integer types of the same byte size

• Component: JavaScriptCore
• Status: RESOLVED FIXED

[Saam Barati]

...

[Saam Barati]

Created attachment 446106 [details]
patch

[Keith Miller]

Comment on attachment 446106 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=446106&action=review

> Source/JavaScriptCore/runtime/JSGenericTypedArrayViewInlines.h:283
> +    if (isInt(Adaptor::typeValue) && isInt(ci->typedArrayStorageType) && !isClamped(Adaptor::typeValue) && JSC::elementSize(Adaptor::typeValue) == JSC::elementSize(ci->typedArrayStorageType))

Nit: Why do you need the isClamped check? Shouldn't that follow from the if above and the fact that the elementSizes are the same? It might still be worth having as an ASSERT in the body of this if though.

[Saam Barati]

Comment on attachment 446106 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=446106&action=review

>> Source/JavaScriptCore/runtime/JSGenericTypedArrayViewInlines.h:283
>> +    if (isInt(Adaptor::typeValue) && isInt(ci->typedArrayStorageType) && !isClamped(Adaptor::typeValue) && JSC::elementSize(Adaptor::typeValue) == JSC::elementSize(ci->typedArrayStorageType))
> 
> Nit: Why do you need the isClamped check? Shouldn't that follow from the if above and the fact that the elementSizes are the same? It might still be worth having as an ASSERT in the body of this if though.

No, because we could have something like, |this| is Uint8ClampedArray and the source is Int8Array. Above check doesn't hit, element sizes are the same, but we can't use this copying path.

[Saam Barati]

Created attachment 446120 [details]
patch for landing

[Saam Barati]

Comment on attachment 446120 [details]
patch for landing

View in context: https://bugs.webkit.org/attachment.cgi?id=446120&action=review

> Source/JavaScriptCore/runtime/JSGenericTypedArrayViewInlines.h:219
> +        || (elementSize == otherElementSize && (static_cast<void*>(typedVector() + offset) <= static_cast<void*>(other->typedVector() + otherOffset)))

this fixes a pre-existing bug where we sometimes do a forwards loop even though we should've been doing a backwards loop. I added a test for this.

[Keith Miller]

Comment on attachment 446120 [details]
patch for landing

r=me again.

[EWS]

Committed r286639 (244952@main): <https://commits.webkit.org/244952@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 446120 [details].

[Radar WebKit Bug Importer]

<rdar://problem/86189423>