Bug 170655 - [JSC] Enable JSRunLoopTimer for JSCOnly and Windows
Summary: [JSC] Enable JSRunLoopTimer for JSCOnly and Windows
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Yusuke Suzuki
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-04-09 08:54 PDT by Yusuke Suzuki
Modified: 2017-04-11 02:34 PDT (History)
12 users (show)

See Also:

Attachments
Patch (16.11 KB, patch)
2017-04-09 08:57 PDT, Yusuke Suzuki 		no flags Details | Formatted Diff | Diff
Patch (15.33 KB, patch)
2017-04-10 00:50 PDT, Yusuke Suzuki 		cgarcia: review+
Details | Formatted Diff | Diff
Show Obsolete (1) 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 2017-04-09 08:54:26 PDT 
[JSC] Enable JSRunLoopTimer for JSCOnly and Windows
Comment 1 Yusuke Suzuki 2017-04-09 08:57:21 PDT 
Created attachment 306622 [details]
Patch
Comment 2 Carlos Garcia Campos 2017-04-09 23:29:04 PDT 
Comment on attachment 306622 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=306622&action=review

> Source/JavaScriptCore/jsc.cpp:3891
>      WTF::initializeMainThread();
>      JSC::initializeThreading();
> +    RunLoop::initializeMainRunLoop();

Do we really need this? JS timers use RunLoop::current() not main(). If we need it, then we can remove WTF::initializeMainThread(); because initializeMainRunLoop already initializes the main thread.

> Source/JavaScriptCore/runtime/JSRunLoopTimer.cpp:190
> +    m_timer.stop();
> +    m_timer.startOneShot(s_decade);

RunLoop::Timer::start() already calls stop().

> Source/JavaScriptCore/runtime/JSRunLoopTimer.h:85
> +    static const Seconds s_decade;
> +    RunLoop::Timer<JSRunLoopTimer> m_timer;

Yes, I think we could adopt this implementation at least for non cocoa based ports. I think when we added support for heap timer, RunLoop was still in WebCore.
Comment 3 Yusuke Suzuki 2017-04-10 00:46:07 PDT 
Comment on attachment 306622 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=306622&action=review

Thanks, I'll update the patch.

>> Source/JavaScriptCore/jsc.cpp:3891
>> +    RunLoop::initializeMainRunLoop();
> 
> Do we really need this? JS timers use RunLoop::current() not main(). If we need it, then we can remove WTF::initializeMainThread(); because initializeMainRunLoop already initializes the main thread.

Right, it is not necessary. Dropped.

>> Source/JavaScriptCore/runtime/JSRunLoopTimer.cpp:190
>> +    m_timer.startOneShot(s_decade);
> 
> RunLoop::Timer::start() already calls stop().

Nice, dropped.

>> Source/JavaScriptCore/runtime/JSRunLoopTimer.h:85
>> +    RunLoop::Timer<JSRunLoopTimer> m_timer;
> 
> Yes, I think we could adopt this implementation at least for non cocoa based ports. I think when we added support for heap timer, RunLoop was still in WebCore.

Yeah, non cocoa based ports, I'll change to use this RunLoop::Timer in the subsequent patch.
This patch paves the way for that.
Comment 4 Yusuke Suzuki 2017-04-10 00:50:47 PDT 
Created attachment 306667 [details]
Patch
Comment 5 Yusuke Suzuki 2017-04-11 02:34:44 PDT 
Committed r215223: <http://trac.webkit.org/changeset/215223>