Bug 233978 – Make FontCascade::setShouldUseSmoothing/shouldUseSmoothing thread safe

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

[Cameron McCormack (:heycam)]

Make FontCascade::setShouldUseSmoothing/shouldUseSmoothing thread safe

[Cameron McCormack (:heycam)]

Created attachment 446301 [details]
Patch

[Myles C. Maxfield]

Comment on attachment 446301 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=446301&action=review

> Source/WebCore/ChangeLog:11
> +        same time from the main thread. This setting is only changed from
> +        Internals and WebView and at WebProcess initilization, but should

WebKitLegacy and WebKit have SPI for this:

+ (void)_setShouldUseFontSmoothing:(BOOL)f
{
    WebCore::FontCascade::setShouldUseSmoothing(f);
}

void WKContextSetShouldUseFontSmoothing(WKContextRef contextRef, bool useFontSmoothing)
{
    WebKit::toImpl(contextRef)->setShouldUseFontSmoothing(useFontSmoothing);
}

> Source/WebCore/platform/graphics/FontCascade.cpp:442
> +static std::atomic<bool> shouldUseFontSmoothing = true;

Is it possible to make this no longer a global? I think that would be a better solution than this.

[Radar WebKit Bug Importer]

<rdar://problem/86506527>

[Matt Woodrow]

(In reply to Myles C. Maxfield from comment #2)
> 
> > Source/WebCore/platform/graphics/FontCascade.cpp:442
> > +static std::atomic<bool> shouldUseFontSmoothing = true;
> 
> Is it possible to make this no longer a global? I think that would be a
> better solution than this.

We could store this on the per-thread FontCache instances, but that then makes it harder to mutate with the SPI, since we'd have to dispatch it to all threads that have a FontCache initialised.

[Myles C. Maxfield]

(In reply to Matt Woodrow from comment #4)
> (In reply to Myles C. Maxfield from comment #2)
> > 
> > > Source/WebCore/platform/graphics/FontCascade.cpp:442
> > > +static std::atomic<bool> shouldUseFontSmoothing = true;
> > 
> > Is it possible to make this no longer a global? I think that would be a
> > better solution than this.
> 
> We could store this on the per-thread FontCache instances, but that then
> makes it harder to mutate with the SPI, since we'd have to dispatch it to
> all threads that have a FontCache initialised.

I guess I still think that's a better solution than a global :/

[Myles C. Maxfield]

Alternatively, we could try to find all callers of the SPI and see if we can remove it.

[Matt Woodrow]

Pull request: https://github.com/WebKit/WebKit/pull/4240

[EWS]

Committed 254500@main (eb547e04a187): <https://commits.webkit.org/254500@main>

Reviewed commits have been landed. Closing PR #4240 and removing active labels.