Avoid division-by-zero when visual viewport width is 0.

Avoid division-by-zero when visual viewport width is 0.

The extension overlay/drop-down or what it's called seems to have
styles calculated with visual viewport size set to (0, 0) which caused
a division-by-zero error calculating the base value for vh units.

R=bokan@chromium.org
BUG=667712

[rune, 2016-12-09 15:24:31]

The CQ bit was checked by rune@opera.com to run a CQ dry run

[commit-bot: I haz the power, 2016-12-09 15:24:50]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[rune, 2016-12-09 15:25:43]

I don't know if this is the correct fix, or if the visual viewport width should
have been different from 0.

[rune, 2016-12-09 15:27:50]

https://codereview.chromium.org/2563783003/diff/1/third_party/WebKit/Source/c...
File third_party/WebKit/Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/2563783003/diff/1/third_party/WebKit/Source/c...
third_party/WebKit/Source/core/frame/FrameView.cpp:1357:
m_frame->host()->visualViewport().size().width()) {
I should probably check the pageScaleAtLayoutWidth for 0 below instead ...

[bokan, 2016-12-09 16:48:56]

https://codereview.chromium.org/2563783003/diff/1/third_party/WebKit/Source/c...
File third_party/WebKit/Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/2563783003/diff/1/third_party/WebKit/Source/c...
third_party/WebKit/Source/core/frame/FrameView.cpp:1357:
m_frame->host()->visualViewport().size().width()) {
On 2016/12/09 15:27:50, rune wrote:
> I should probably check the pageScaleAtLayoutWidth for 0 below instead ...

Visual Viewport size should never be 0, do you know why that happens? Sigh...I
battled many size regressions related to extensions and their autosizing
behavior but I thought it was over...

[commit-bot: I haz the power, 2016-12-09 18:00:53]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-12-09 18:00:54]

Dry run: This issue passed the CQ dry run.

[rune, 2016-12-09 18:47:26]

On 2016/12/09 16:48:56, bokan wrote:

> Visual Viewport size should never be 0, do you know why that happens? Sigh...I
> battled many size regressions related to extensions and their autosizing
> behavior but I thought it was over...

No, I never actually debugged, I used printf and didn't check further since I
didn't know if that should ever happen.

[bokan, 2016-12-09 18:48:16]

On 2016/12/09 18:47:26, rune wrote:
> On 2016/12/09 16:48:56, bokan wrote:
> 
> > Visual Viewport size should never be 0, do you know why that happens?
Sigh...I
> > battled many size regressions related to extensions and their autosizing
> > behavior but I thought it was over...
> 
> No, I never actually debugged, I used printf and didn't check further since I
> didn't know if that should ever happen.

Ok, I wrote all this so unless you're deep into it I can take over this bug for
you.

[rune, 2016-12-09 19:35:42]

On 2016/12/09 18:47:26, rune wrote:
> On 2016/12/09 16:48:56, bokan wrote:
> 
> > Visual Viewport size should never be 0, do you know why that happens?
Sigh...I
> > battled many size regressions related to extensions and their autosizing
> > behavior but I thought it was over...
> 
> No, I never actually debugged, I used printf and didn't check further since I
> didn't know if that should ever happen.

newSize is (0,0) here when I click the icon to open the extension drop-down.

#0  blink::WebViewImpl::resizeVisualViewport (this=0x4b3c3ae9e90, newSize=...)
at ../../third_party/WebKit/Source/web/WebViewImpl.cpp:1762
#1  0x00007fffe76c12ec in blink::WebViewFrameWidget::resizeVisualViewport
(this=0x280d50c75980, size=...) at
../../third_party/WebKit/Source/web/WebViewFrameWidget.cpp:49
#2  0x00007ffff210b3fe in content::RenderWidget::Resize (this=0x280d511e7a20,
params=...) at ../../content/renderer/render_widget.cc:1169
#3  0x00007ffff21086c2 in content::RenderWidget::OnResize (this=0x280d511e7a20,
params=...) at ../../content/renderer/render_widget.cc:724
#4  0x00007ffff20ede46 in content::RenderViewImpl::OnResize
(this=0x280d511e7a20, params=...) at
../../content/renderer/render_view_impl.cc:2335
#5  0x00007ffff20e2074 in
content::RenderViewImpl::Initialize(content::mojom::CreateViewParams const&,
base::Callback<void (content::RenderWidget*, blink::WebNavigationPolicy,
gfx::Rect const&), (base::internal::CopyMode)1, (base::internal::RepeatMode)1>
const&) (this=0x280d511e7a20, params=..., show_callback=...) at
../../content/renderer/render_view_impl.cc:708
#6  0x00007ffff20e4dc8 in
content::RenderViewImpl::Create(content::CompositorDependencies*,
content::mojom::CreateViewParams const&, base::Callback<void
(content::RenderWidget*, blink::WebNavigationPolicy, gfx::Rect const&),
(base::internal::CopyMode)1, (base::internal::RepeatMode)1> const&)
(compositor_deps=0x280d50118638, params=..., show_callback=...) at
../../content/renderer/render_view_impl.cc:1097
#7  0x00007ffff20bb9a9 in content::RenderThreadImpl::CreateView
(this=0x280d50118220, params=...) at
../../content/renderer/render_thread_impl.cc:2156
#8  0x00007ffff06aeb06 in content::mojom::RendererStubDispatch::Accept
(impl=0x280d50118630, context=0x280d50118a68, message=0x7fffffffb968) at
gen/content/common/renderer.mojom.cc:446
#9  0x00007ffff20c7b23 in
content::mojom::RendererStub<mojo::RawPtrImplRefTraits<content::mojom::Renderer>
>::Accept (this=0x280d50118a58, message=0x7fffffffb968) at
gen/content/common/renderer.mojom.h:273
#10 0x00007ffff7f3a1f9 in ?? () from
/chromium/dev/src/out/Debug/./libbindings.so

... and that's the only size I see passed into that method in this case.

[rune, 2016-12-09 19:39:05]

On 2016/12/09 18:48:16, bokan wrote:
> On 2016/12/09 18:47:26, rune wrote:
> > On 2016/12/09 16:48:56, bokan wrote:
> > 
> > > Visual Viewport size should never be 0, do you know why that happens?
> Sigh...I
> > > battled many size regressions related to extensions and their autosizing
> > > behavior but I thought it was over...
> > 
> > No, I never actually debugged, I used printf and didn't check further since
I
> > didn't know if that should ever happen.
> 
> Ok, I wrote all this so unless you're deep into it I can take over this bug
for
> you.

Please do.

[bokan, 2016-12-09 19:40:23]

On 2016/12/09 19:39:05, rune wrote:
> On 2016/12/09 18:48:16, bokan wrote:
> > On 2016/12/09 18:47:26, rune wrote:
> > > On 2016/12/09 16:48:56, bokan wrote:
> > > 
> > > > Visual Viewport size should never be 0, do you know why that happens?
> > Sigh...I
> > > > battled many size regressions related to extensions and their autosizing
> > > > behavior but I thought it was over...
> > > 
> > > No, I never actually debugged, I used printf and didn't check further
since
> I
> > > didn't know if that should ever happen.
> > 
> > Ok, I wrote all this so unless you're deep into it I can take over this bug
> for
> > you.
> 
> Please do.

Ok, thanks for digging into it.

[rune, 2016-12-09 19:57:03]

Description was changed from

==========
Avoid division-by-zero when visual viewport width is 0.

The extension overlay/drop-down or what it's called seems to have
styles calculated with visual viewport size set to (0, 0) which caused
a division-by-zero error calculating the base value for vh units.

R=bokan@chromium.org
BUG=667712
==========

to

==========
Avoid division-by-zero when visual viewport width is 0.

The extension overlay/drop-down or what it's called seems to have
styles calculated with visual viewport size set to (0, 0) which caused
a division-by-zero error calculating the base value for vh units.

R=bokan@chromium.org
BUG=667712
==========

[rune, 2016-12-09 19:57:03]

Description was changed from

==========
Avoid division-by-zero when visual viewport width is 0.

The extension overlay/drop-down or what it's called seems to have
styles calculated with visual viewport size set to (0, 0) which caused
a division-by-zero error calculating the base value for vh units.

R=bokan@chromium.org
BUG=667712
==========

to

==========
Avoid division-by-zero when visual viewport width is 0.

The extension overlay/drop-down or what it's called seems to have
styles calculated with visual viewport size set to (0, 0) which caused
a division-by-zero error calculating the base value for vh units.

R=bokan@chromium.org
BUG=667712
==========