Return raw CSSValue pointer if not CSSImageValue

Return raw CSSValue pointer instead of cast to CSSImageValue

Avoid an assertion triggering detected when CSSValue pointer is either
void or not cast to CSSImageValue. As far as I've seen, this data()
function is used to compare objects in RenderStyle, so it makes sense to
return a raw pointer.

The assertion avoids successful run of the following layout tests:
fast/hidpi/image-set-out-of-order.html
fast/hidpi/image-set-without-specified-width.html
virtual/gpu/fast/hidpi/image-set-out-of-order.html
virtual/gpu/fast/hidpi/image-set-without-specified-width.html

BUG=269627
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=155874

[Julien - ping for review, 2013-08-09 16:02:49]

What is this change trying to achieve? Avoid an ASSERT triggering? (in which
case you should mention a test tripping on it)

https://codereview.chromium.org/22631003/diff/1/Source/core/rendering/style/S...
File Source/core/rendering/style/StylePendingImage.h (right):

https://codereview.chromium.org/22631003/diff/1/Source/core/rendering/style/S...
Source/core/rendering/style/StylePendingImage.h:46: virtual WrappedImagePtr
data() const { return m_value && m_value->isImageValue() ?
toCSSImageValue(m_value) : m_value; }
Why not just return m_value all the time?

[abinader, 2013-08-09 19:28:03]

On 2013/08/09 16:02:49, Julien Chaffraix wrote:
> What is this change trying to achieve? Avoid an ASSERT triggering? (in which
> case you should mention a test tripping on it)

Yes - avoid an assert triggering. The following layout tests are related with
this change:

fast/hidpi/image-set-out-of-order.html
fast/hidpi/image-set-without-specified-width.html
virtual/gpu/fast/hidpi/image-set-out-of-order.html
virtual/gpu/fast/hidpi/image-set-without-specified-width.html

I'll add that information to the issue description.

> 
>
https://codereview.chromium.org/22631003/diff/1/Source/core/rendering/style/S...
> File Source/core/rendering/style/StylePendingImage.h (right):
> 
>
https://codereview.chromium.org/22631003/diff/1/Source/core/rendering/style/S...
> Source/core/rendering/style/StylePendingImage.h:46: virtual WrappedImagePtr
> data() const { return m_value && m_value->isImageValue() ?
> toCSSImageValue(m_value) : m_value; }
> Why not just return m_value all the time?

My initial idea was that - though that was exactly what was changed in revision
100535:
https://src.chromium.org/viewvc/blink/trunk/Source/WebCore/rendering/style/St...

I'm not sure why Timothy changed it this way, but it sounds sane for me to just
send m_value as well. I'll update the patch now.

[Julien - ping for review, 2013-08-09 19:48:28]

It's weird that WrappedImagePtr is void* but until this is changed, there is no
need for any cast so lgtm.

[commit-bot: I haz the power, 2013-08-09 19:53:19]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bruno.d@partner.samsung.com/22631003/1...

[commit-bot: I haz the power, 2013-08-09 21:39:45]

Change committed as 155874