Refactor avoidsFloats() to reflect CSS2.1 spec more clearly

Refactor avoidsFloats() to reflect CSS2.1 spec more clearly

Introduce the 'createsBlockFormatting' concept to RenderBlock::avoidsFloats()
and remove a couple of now-redundant virtualizations of the function.

BUG=388757

[leviw_travelin_and_unemployed, 2014-05-06 18:16:14]

lgtm

[rhogan, 2014-05-07 18:09:56]

The CQ bit was checked by robhogan@gmail.com

[commit-bot: I haz the power, 2014-05-07 18:10:51]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/robhogan@gmail.com/268833003/1

[commit-bot: I haz the power, 2014-05-07 18:22:40]

Change committed as 173549

[rhogan, 2014-05-21 21:50:46]

A revert of this CL has been created in
https://codereview.chromium.org/298913002/ by robhogan@gmail.com.

The reason for reverting is: May have caused a clusterfuzz regression:
https://code.google.com/p/chromium/issues/detail?id=375954.

[rhogan, 2014-05-22 18:03:27]

A revert of this CL has been created in
https://codereview.chromium.org/296983005/ by robhogan@gmail.com.

The reason for reverting is: May have caused a clusterfuzz regression:
https://code.google.com/p/chromium/issues/detail?id=375954..

[rhogan, 2014-06-17 21:53:16]

https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
File Source/core/rendering/RenderBlockFlow.cpp (right):

https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
Source/core/rendering/RenderBlockFlow.cpp:1837: bool
canPropagateFloatIntoSibling = !avoidsOrIgnoresFloats();
This is why the previous version of this patch had to be reverted. I used
!createsBlockFormattingContext() instead of !avoidsOrIgnoresFloats().
Unfortunately I have not been able to reduce the clusterfuzz test that spotted
this to fail (at all) from a layout test so I have no test to add to this second
try.

[leviw_travelin_and_unemployed, 2014-06-17 22:08:59]

On 2014/06/17 at 21:53:16, robhogan wrote:
>
https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
> File Source/core/rendering/RenderBlockFlow.cpp (right):
> 
>
https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
> Source/core/rendering/RenderBlockFlow.cpp:1837: bool
canPropagateFloatIntoSibling = !avoidsOrIgnoresFloats();
> This is why the previous version of this patch had to be reverted. I used
!createsBlockFormattingContext() instead of !avoidsOrIgnoresFloats().
Unfortunately I have not been able to reduce the clusterfuzz test that spotted
this to fail (at all) from a layout test so I have no test to add to this second
try.

Are you actively trying to get it reduced, or hoping to land this without the
LayoutTest?

[rhogan, 2014-06-18 20:07:10]

On 2014/06/17 at 22:08:59, leviw wrote:
> On 2014/06/17 at 21:53:16, robhogan wrote:
> >
https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
> > File Source/core/rendering/RenderBlockFlow.cpp (right):
> > 
> >
https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
> > Source/core/rendering/RenderBlockFlow.cpp:1837: bool
canPropagateFloatIntoSibling = !avoidsOrIgnoresFloats();
> > This is why the previous version of this patch had to be reverted. I used
!createsBlockFormattingContext() instead of !avoidsOrIgnoresFloats().
Unfortunately I have not been able to reduce the clusterfuzz test that spotted
this to fail (at all) from a layout test so I have no test to add to this second
try.
> 
> Are you actively trying to get it reduced, or hoping to land this without the
LayoutTest?

I'm hoping to land it without the test. I can't get an Asan build going on my
modest 32-bit desktop and have not been able to get the test to fail on the asan
trybots or on a normal release build.

I'm very confident this was the source of the issue.

[leviw_travelin_and_unemployed, 2014-06-18 21:01:54]

On 2014/06/18 20:07:10, rhogan wrote:
> On 2014/06/17 at 22:08:59, leviw wrote:
> > On 2014/06/17 at 21:53:16, robhogan wrote:
> > >
>
https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
> > > File Source/core/rendering/RenderBlockFlow.cpp (right):
> > > 
> > >
>
https://codereview.chromium.org/268833003/diff/20001/Source/core/rendering/Re...
> > > Source/core/rendering/RenderBlockFlow.cpp:1837: bool
> canPropagateFloatIntoSibling = !avoidsOrIgnoresFloats();
> > > This is why the previous version of this patch had to be reverted. I used
> !createsBlockFormattingContext() instead of !avoidsOrIgnoresFloats().
> Unfortunately I have not been able to reduce the clusterfuzz test that spotted
> this to fail (at all) from a layout test so I have no test to add to this
second
> try.
> > 
> > Are you actively trying to get it reduced, or hoping to land this without
the
> LayoutTest?
> 
> I'm hoping to land it without the test. I can't get an Asan build going on my
> modest 32-bit desktop and have not been able to get the test to fail on the
asan
> trybots or on a normal release build.
> 
> I'm very confident this was the source of the issue.

Okay. LGTM.

[rhogan, 2014-06-19 21:28:46]

The CQ bit was checked by robhogan@gmail.com

[commit-bot: I haz the power, 2014-06-19 21:29:14]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/robhogan@gmail.com/268833003/20001

[commit-bot: I haz the power, 2014-06-19 22:29:16]

Change committed as 176557

[rhogan, 2014-06-25 17:20:41]

A revert of this CL has been created in
https://codereview.chromium.org/330913007/ by robhogan@gmail.com.

The reason for reverting is: Causing clusterfuzz crashes.

[rhogan, 2014-11-11 19:17:13]

https://codereview.chromium.org/268833003/diff/60001/Source/core/rendering/Re...
File Source/core/rendering/RenderBlockFlow.cpp (right):

https://codereview.chromium.org/268833003/diff/60001/Source/core/rendering/Re...
Source/core/rendering/RenderBlockFlow.cpp:1920: s_canPropagateFloatIntoSibling =
oldStyle ? !avoidsOrIgnoresFloats() : false;
I had !createsBlockFormattingContext() here; which is consistent with the logic
elsewhere. This seems like a good candidate for the cause of the crash in bug
388757 (which I can't recreate).