[Chromium] Uninitialized access in SimpleFontDataSkia::platformInit()

• Component: Platform
• Status: RESOLVED FIXED

[Kenichi Ishibashi]

From http://crbug.com/76378

Conditional jump or move depends on uninitialised value(s)                                                                                                                                          
   at 0x16AEFEF: WebCore::SimpleFontData::platformInit() (third_party/WebKit/Source/WebCore/platform/graphics/SimpleFontData.h:312)                                                                 
   by 0x169AC95: WebCore::SimpleFontData::SimpleFontData(WebCore::FontPlatformData const&, bool, bool, bool) (third_party/WebKit/Source/WebCore/platform/graphics/SimpleFontData.cpp:61)

This implies that platformInit() calls widthForGlyph(), in which we compared the argument with m_zeroWidthSpaceGlyph, which isn't initialized yet.

[Kenichi Ishibashi]

Created attachment 134194 [details]
Patch

[Kenichi Ishibashi]

Kent-san, could you take a look?

[Kent Tamura]

Comment on attachment 134194 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=134194&action=review

> Source/WebCore/ChangeLog:12
> +        [Chromium] Uninitialized access in SimpleFontDataSkia::platformInit
> +        https://bugs.webkit.org/show_bug.cgi?id=82411
> +
> +        SimpleFontDataSkia::platformInit() could call widthForGlyph(), which
> +        accesses m_zeroWidthSpaceGlyph. This causes a valgrind memcheck error
> +        because m_zeroWidthSpaceGlyph isn't initialized at this point.
> +        Initialize m_zeroWidthSpaceGlyph with zero so that widthForGlyph() can
> +        return appropriate value(The value zero here means "unknown glyph").
> +
> +        Reviewed by NOBODY (OOPS!).

Usual order is:

  <summary>
  <bug URL>

  Reviewd by ...

  <Detail>

[Kenichi Ishibashi]

Created attachment 134200 [details]
Patch for landing

[Kenichi Ishibashi]

Comment on attachment 134194 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=134194&action=review

Thanks!

>> Source/WebCore/ChangeLog:12
>> +        Reviewed by NOBODY (OOPS!).
> 
> Usual order is:
> 
>   <summary>
>   <bug URL>
> 
>   Reviewd by ...
> 
>   <Detail>

Done.

[WebKit Review Bot]

Comment on attachment 134200 [details]
Patch for landing

Clearing flags on attachment: 134200

Committed r112357: <http://trac.webkit.org/changeset/112357>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.