Initiate webfont download right after style recalc

Initiate webfont download right after style recalc

Currently WebFont download is queued during layout, and actual download is
delayed until after layout. This patch makes CSSFontFace::willUseFontData()
enqueue font download, which is called in style calc time. StyleResolver
kicks off font downloads, along with other pending resources.

UMA metrics (WebFont.Resource.StyleRecalcToDownloadLatency) shows
this reduces webfont download latency 114ms at the 50 percentile.
This patch also removes FontResource::FontResourceHistograms as we
collected sufficient data.

BUG=246492
TEST=Covered by existing tests

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=162179

[dglazkov, 2013-11-12 17:58:55]

lgtm

https://codereview.chromium.org/63713002/diff/80001/Source/core/css/CSSFontFa...
File Source/core/css/CSSFontFace.cpp (right):

https://codereview.chromium.org/63713002/diff/80001/Source/core/css/CSSFontFa...
Source/core/css/CSSFontFace.cpp:160: m_activeSource = m_sources[i].get();
As an aside, I am struggling to understand the purpose of m_activeSource. Is
this just a boolean that's implemented as a pointer? We should probably turn it
into a bool then.

[Kunihiko Sakamoto, 2013-11-13 05:18:53]

Actually m_activeSource is more than just a boolean. Consider this example:

@font-face {
  font-family: AB;
  src: url(A.woff), url(B.woff);
}
@font-face {
  font-family: B
  src: url(B.woff);
}

...and assume AB is loading A.woff and B is loading B.woff.
m_activeSource prevents CSSFontFace for AB from reacting to fontLoaded() caused
by B.woff load.

[commit-bot: I haz the power, 2013-11-13 05:20:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ksakamoto@chromium.org/63713002/80001

[dglazkov, 2013-11-13 05:26:39]

yup, right here:
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...

[commit-bot: I haz the power, 2013-11-13 06:44:09]

Retried try job too often on linux_blink_rel for step(s) webkit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_blin...

[dglazkov, 2013-11-13 16:31:16]

I think these might be real failures...

[Kunihiko Sakamoto, 2013-11-14 00:55:05]

Yeah, investigating...

[Kunihiko Sakamoto, 2013-11-18 06:07:39]

https://codereview.chromium.org/63713002/diff/430001/Source/core/css/CSSFontF...
File Source/core/css/CSSFontFaceSource.cpp (right):

https://codereview.chromium.org/63713002/diff/430001/Source/core/css/CSSFontF...
Source/core/css/CSSFontFaceSource.cpp:270: if (m_face && m_font &&
m_font->stillNeedsLoad())
This condition was missing.

[commit-bot: I haz the power, 2013-11-18 06:08:26]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ksakamoto@chromium.org/63713002/430001

[commit-bot: I haz the power, 2013-11-18 06:41:50]

Change committed as 162179

[eseidel, 2013-11-18 20:16:25]

I wonder if this could have caused Acid3 flaky timeouts?
https://codereview.chromium.org/59813012/