[webkit-reviews] review denied: [Bug 74680] [Plugins GTK/QT] Signed to Unsigned Conversion Error : [Attachment 119550] recreate the patch with changelog

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Wed Dec 21 12:23:54 PST 2011


Eric Seidel <eric at webkit.org> has denied Kwangmin Bang
<justine.bang at samsung.com>'s request for review:
Bug 74680: [Plugins GTK/QT] Signed to Unsigned Conversion Error
https://bugs.webkit.org/show_bug.cgi?id=74680

Attachment 119550: recreate the patch with changelog
https://bugs.webkit.org/attachment.cgi?id=119550&action=review

------- Additional Comments from Eric Seidel <eric at webkit.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=119550&action=review


> Source/WebCore/ChangeLog:8
> +	   No new tests. (OOPS!)

This will cause the commit-queue to fail.  You shoul deithe rlist the tests
affected by this change, or explain why this change is impossible/impractical
to test.

We have a TestPlugin in DRT which can be helpful if you need to make a test
which involves NPAPI code.  Otherwise most tests use javascript to poke and
prod at things, or dump the painted pixels and compare them against a
reference.

> Source/WebCore/plugins/gtk/PluginViewGtk.cpp:527
> +	   // Clipping rectangle of the plug-in; the origin is the top left
corner of the drawable or window.
> +	   // Signed to Unsigned Conversion Error; if signed integer variable
has negative number,
> +	   // fill unsigned short variables to zero before type casting.
> +	   // If not, clipRect have an unexpected value cause of the first bit
for negative.
> +	   m_npWindow.clipRect.left = max(0, m_npWindow.x + m_clipRect.x());
> +	   m_npWindow.clipRect.top = max(0, m_npWindow.y + m_clipRect.y());
> +	   m_npWindow.clipRect.right = max(0, m_npWindow.x + m_clipRect.x() +
m_clipRect.width());
> +	   m_npWindow.clipRect.bottom = max(0, m_npWindow.y + m_clipRect.y() +
m_clipRect.height());

Seems we should be sharing this code on the root PluginView implementation with
a private helper method, no?


More information about the webkit-reviews mailing list