Bug 201513 - [JSC] Generalize Get/PutPromiseInternalField for InternalFieldObjectImpl
Summary: [JSC] Generalize Get/PutPromiseInternalField for InternalFieldObjectImpl
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: JavaScriptCore (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Yusuke Suzuki
URL:
Keywords: InRadar
Depends on:
Blocks:
 
Reported: 2019-09-05 13:01 PDT by Yusuke Suzuki
Modified: 2019-09-05 14:48 PDT (History)
2 users (show)

See Also:

Attachments
Patch (64.15 KB, patch)
2019-09-05 13:38 PDT, Yusuke Suzuki 		no flags Details | Formatted Diff | Diff
Patch (64.17 KB, patch)
2019-09-05 13:45 PDT, Yusuke Suzuki 		no flags Details | Formatted Diff | Diff
Patch (64.16 KB, patch)
2019-09-05 13:57 PDT, Yusuke Suzuki 		ross.kirsling: review+
Details | Formatted Diff | Diff
Show Obsolete (2) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Yusuke Suzuki 2019-09-05 13:01:02 PDT 
To reuse it for generators.
Comment 1 Radar WebKit Bug Importer 2019-09-05 13:09:56 PDT 
<rdar://problem/55080640>
Comment 2 Yusuke Suzuki 2019-09-05 13:13:13 PDT 
The goal of making Generator internal-field-object is avoiding op_get_by_id_direct for generators.
Each generator has different structure since [[Prototype]] of them are different (defined in the spec). As a result, Generator.prototype.next's internal field accesses are getting super generic get_by_id.
Comment 3 Yusuke Suzuki 2019-09-05 13:13:49 PDT 
(In reply to Yusuke Suzuki from comment #2)
> The goal of making Generator internal-field-object is avoiding
> op_get_by_id_direct for generators.
> Each generator has different structure since [[Prototype]] of them are
> different (defined in the spec). As a result, Generator.prototype.next's
> internal field accesses are getting super generic get_by_id.

If multiple generators exist in one app, we hits this slow-case. And JetStream2/Basic is the one I think.
Comment 4 Yusuke Suzuki 2019-09-05 13:38:45 PDT 
Created attachment 378113 [details]
Patch
Comment 5 Yusuke Suzuki 2019-09-05 13:45:50 PDT 
Created attachment 378114 [details]
Patch
Comment 6 Yusuke Suzuki 2019-09-05 13:57:19 PDT 
Created attachment 378116 [details]
Patch
Comment 7 Ross Kirsling 2019-09-05 14:01:52 PDT 
Comment on attachment 378116 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=378116&action=review

Looks nice and simple to me.

> Source/JavaScriptCore/llint/LowLevelInterpreter.asm:168
> +const JSInternalFieldObjectImpl_internalFields = JSInternalFieldObjectImpl::m_internalFields

Is this constant just a stylistic matter? Just curious since we weren't using one before.
Comment 8 Yusuke Suzuki 2019-09-05 14:39:09 PDT 
Comment on attachment 378116 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=378116&action=review

>> Source/JavaScriptCore/llint/LowLevelInterpreter.asm:168
>> +const JSInternalFieldObjectImpl_internalFields = JSInternalFieldObjectImpl::m_internalFields
> 
> Is this constant just a stylistic matter? Just curious since we weren't using one before.

I used it since `<>` syntax is not available in LLInt right now.
Comment 9 Yusuke Suzuki 2019-09-05 14:48:50 PDT 
Committed r249547: <https://trac.webkit.org/changeset/249547>