WebKit: Add missing includes.

WebKit: Add missing includes.

Not having these includes present gives a linker error when building
Chrome and WebKit with -fprofile-generate.

BUG=253985
TEST=USE="pgo_generate" emerge-$board chromeos-chrome works.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=153163

[asharif1, 2013-06-25 17:24:18]

On 2013/06/25 17:24:09, asharif1 wrote:

PTAL.

[Nico, 2013-06-25 17:40:34]

https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp
File Source/core/css/CSSRuleList.cpp (right):

https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.c...
Source/core/css/CSSRuleList.cpp:56: MemoryClassInfo info(memoryObjectInfo, this,
WebCoreMemoryTypes::CSS);
Why does this compile without the include? If using this class requires
including MemoryInstrumentationVector.h, should the header declaring this class
include MIV.h?

[asharif1, 2013-06-25 17:51:30]

On 2013/06/25 17:40:34, Nico wrote:
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp
> File Source/core/css/CSSRuleList.cpp (right):
> 
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.c...
> Source/core/css/CSSRuleList.cpp:56: MemoryClassInfo info(memoryObjectInfo,
this,
> WebCoreMemoryTypes::CSS);
> Why does this compile without the include? If using this class requires

It actually fails to compile without the include with -fprofile-generate. I'm
guessing for the regular build it works because the function call is optimized
away for some reason.

See:
http://chromegw/i/chromeos/builders/lumpy%20pgo%20canary%20%28experimental%29...

It fails to link when built with USE="pgo_generate".

When built with -fprofile-generate, I ran:

readelf -sW CSSRuleList.o | grep
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE

   515: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE

 c++filt
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
void WTF::reportMemoryUsage<WTF::RefPtr<WebCore::CSSRule>,
0ul>(WTF::Vector<WTF::RefPtr<WebCore::CSSRule>, 0ul> const*,
WTF::MemoryObjectInfo*)


> including MemoryInstrumentationVector.h, should the header declaring this
class
> include MIV.h?

Done in the next patchset, though I don't completely understand when to use <>
vs "" for the includes. Some wtf includes use <> while others use "".

[Nico, 2013-06-25 18:12:22]

On 2013/06/25 17:51:30, asharif1 wrote:
> On 2013/06/25 17:40:34, Nico wrote:
> >
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp
> > File Source/core/css/CSSRuleList.cpp (right):
> > 
> >
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.c...
> > Source/core/css/CSSRuleList.cpp:56: MemoryClassInfo info(memoryObjectInfo,
> this,
> > WebCoreMemoryTypes::CSS);
> > Why does this compile without the include? If using this class requires
> 
> It actually fails to compile without the include with -fprofile-generate. I'm
> guessing for the regular build it works because the function call is optimized
> away for some reason.
> 
> See:
>
http://chromegw/i/chromeos/builders/lumpy%2520pgo%2520canary%2520%2528experim...
> 
> It fails to link when built with USE="pgo_generate".
> 
> When built with -fprofile-generate, I ran:
> 
> readelf -sW CSSRuleList.o | grep
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> 
>    515: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> 
>  c++filt
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> void WTF::reportMemoryUsage<WTF::RefPtr<WebCore::CSSRule>,
> 0ul>(WTF::Vector<WTF::RefPtr<WebCore::CSSRule>, 0ul> const*,
> WTF::MemoryObjectInfo*)
> 
> 
> > including MemoryInstrumentationVector.h, should the header declaring this
> class
> > include MIV.h?
> 
> Done in the next patchset, though I don't completely understand when to use <>
> vs "" for the includes. Some wtf includes use <> while others use "".

I meant move it to the header that requires including MIV.h (probably
MemoryInstrumentationVector.h?), not to the .h of the cc you touched.

Always use "", <> is deprecated for wtf includes.

[Nico, 2013-06-25 18:12:38]

+yurys for the memory instrumentation bits

[asharif1, 2013-06-25 18:41:09]

On 2013/06/25 18:12:22, Nico wrote:
> On 2013/06/25 17:51:30, asharif1 wrote:
> > On 2013/06/25 17:40:34, Nico wrote:
> > >
> >
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp
> > > File Source/core/css/CSSRuleList.cpp (right):
> > > 
> > >
> >
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.c...
> > > Source/core/css/CSSRuleList.cpp:56: MemoryClassInfo info(memoryObjectInfo,
> > this,
> > > WebCoreMemoryTypes::CSS);
> > > Why does this compile without the include? If using this class requires
> > 
> > It actually fails to compile without the include with -fprofile-generate.
I'm
> > guessing for the regular build it works because the function call is
optimized
> > away for some reason.
> > 
> > See:
> >
>
http://chromegw/i/chromeos/builders/lumpy%252520pgo%252520canary%252520%25252...
> > 
> > It fails to link when built with USE="pgo_generate".
> > 
> > When built with -fprofile-generate, I ran:
> > 
> > readelf -sW CSSRuleList.o | grep
> >
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> > 
> >    515: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND
> >
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> > 
> >  c++filt
> >
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> > void WTF::reportMemoryUsage<WTF::RefPtr<WebCore::CSSRule>,
> > 0ul>(WTF::Vector<WTF::RefPtr<WebCore::CSSRule>, 0ul> const*,
> > WTF::MemoryObjectInfo*)
> > 
> > 
> > > including MemoryInstrumentationVector.h, should the header declaring this
> > class
> > > include MIV.h?
> > 
> > Done in the next patchset, though I don't completely understand when to use
<>
> > vs "" for the includes. Some wtf includes use <> while others use "".
> 
> I meant move it to the header that requires including MIV.h (probably
> MemoryInstrumentationVector.h?), not to the .h of the cc you touched.

I still don't understand where I should move the includes to.

The include file that's missing is MemoryInstrumentationVector.h. Where should I
add this include?

I moved it to the cpp files for now.

> 
> Always use "", <> is deprecated for wtf includes.

[asharif1, 2013-06-26 19:43:21]

On 2013/06/25 18:41:09, asharif1 wrote:
> On 2013/06/25 18:12:22, Nico wrote:
> > On 2013/06/25 17:51:30, asharif1 wrote:
> > > On 2013/06/25 17:40:34, Nico wrote:
> > > >
> > >
> >
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp
> > > > File Source/core/css/CSSRuleList.cpp (right):
> > > > 
> > > >
> > >
> >
>
https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.c...
> > > > Source/core/css/CSSRuleList.cpp:56: MemoryClassInfo
info(memoryObjectInfo,
> > > this,
> > > > WebCoreMemoryTypes::CSS);
> > > > Why does this compile without the include? If using this class requires
> > > 
> > > It actually fails to compile without the include with -fprofile-generate.
> I'm
> > > guessing for the regular build it works because the function call is
> optimized
> > > away for some reason.
> > > 
> > > See:
> > >
> >
>
http://chromegw/i/chromeos/builders/lumpy%25252520pgo%25252520canary%25252520...
> > > 
> > > It fails to link when built with USE="pgo_generate".
> > > 
> > > When built with -fprofile-generate, I ran:
> > > 
> > > readelf -sW CSSRuleList.o | grep
> > >
> >
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> > > 
> > >    515: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND
> > >
> >
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> > > 
> > >  c++filt
> > >
> >
>
_ZN3WTF17reportMemoryUsageINS_6RefPtrIN7WebCore7CSSRuleEEELm0EEEvPKNS_6VectorIT_XT0_EEEPNS_16MemoryObjectInfoE
> > > void WTF::reportMemoryUsage<WTF::RefPtr<WebCore::CSSRule>,
> > > 0ul>(WTF::Vector<WTF::RefPtr<WebCore::CSSRule>, 0ul> const*,
> > > WTF::MemoryObjectInfo*)
> > > 
> > > 
> > > > including MemoryInstrumentationVector.h, should the header declaring
this
> > > class
> > > > include MIV.h?
> > > 
> > > Done in the next patchset, though I don't completely understand when to
use
> <>
> > > vs "" for the includes. Some wtf includes use <> while others use "".
> > 
> > I meant move it to the header that requires including MIV.h (probably
> > MemoryInstrumentationVector.h?), not to the .h of the cc you touched.
> 
> I still don't understand where I should move the includes to.
> 
> The include file that's missing is MemoryInstrumentationVector.h. Where should
I
> add this include?
> 
> I moved it to the cpp files for now.
> 
> > 
> > Always use "", <> is deprecated for wtf includes.

thakis/timloh, ping.

[Nico, 2013-06-26 19:44:45]

I was hoping yurys would take a look. +loislo for redundancy :-)


On Wed, Jun 26, 2013 at 12:43 PM, <asharif@chromium.org> wrote:

> On 2013/06/25 18:41:09, asharif1 wrote:
>
>> On 2013/06/25 18:12:22, Nico wrote:
>> > On 2013/06/25 17:51:30, asharif1 wrote:
>> > > On 2013/06/25 17:40:34, Nico wrote:
>> > > >
>> > >
>> >
>>
>
> https://codereview.chromium.**org/17707003/diff/1/Source/**
>
core/css/CSSRuleList.cpp<https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp>
>
>> > > > File Source/core/css/CSSRuleList.**cpp (right):
>> > > >
>> > > >
>> > >
>> >
>>
>
> https://codereview.chromium.**org/17707003/diff/1/Source/**
>
core/css/CSSRuleList.cpp#**newcode56<https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp#newcode56>
>
>> > > > Source/core/css/CSSRuleList.**cpp:56: MemoryClassInfo
>>
> info(memoryObjectInfo,
>
>> > > this,
>> > > > WebCoreMemoryTypes::CSS);
>> > > > Why does this compile without the include? If using this class
>> requires
>> > >
>> > > It actually fails to compile without the include with
>> -fprofile-generate.
>> I'm
>> > > guessing for the regular build it works because the function call is
>> optimized
>> > > away for some reason.
>> > >
>> > > See:
>> > >
>> >
>>
>
> http://chromegw/i/chromeos/**builders/lumpy%25252520pgo%**
> 25252520canary%25252520%**25252528experimental%25252529/**
> builds/210/steps/**BuildPackages%25252520%**2525255Bpgo_generate%2525255D%
>
**25252520%2525255Blumpy%**2525255D/logs/stdio<http://chromegw/i/chromeos/builders/lumpy%25252520pgo%25252520canary%25252520%25252528experimental%25252529/builds/210/steps/BuildPackages%25252520%2525255Bpgo_generate%2525255D%25252520%2525255Blumpy%2525255D/logs/stdio>
>
>  > >
>> > > It fails to link when built with USE="pgo_generate".
>> > >
>> > > When built with -fprofile-generate, I ran:
>> > >
>> > > readelf -sW CSSRuleList.o | grep
>> > >
>> >
>>
>
> _ZN3WTF17reportMemoryUsageINS_**6RefPtrIN7WebCore7CSSRuleEEELm**
> 0EEEvPKNS_6VectorIT_XT0_**EEEPNS_16MemoryObjectInfoE
>
>> > >
>> > >    515: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND
>> > >
>> >
>>
>
> _ZN3WTF17reportMemoryUsageINS_**6RefPtrIN7WebCore7CSSRuleEEELm**
> 0EEEvPKNS_6VectorIT_XT0_**EEEPNS_16MemoryObjectInfoE
>
>> > >
>> > >  c++filt
>> > >
>> >
>>
>
> _ZN3WTF17reportMemoryUsageINS_**6RefPtrIN7WebCore7CSSRuleEEELm**
> 0EEEvPKNS_6VectorIT_XT0_**EEEPNS_16MemoryObjectInfoE
>
>> > > void WTF::reportMemoryUsage<WTF::**RefPtr<WebCore::CSSRule>,
>> > > 0ul>(WTF::Vector<WTF::RefPtr<**WebCore::CSSRule>, 0ul> const*,
>> > > WTF::MemoryObjectInfo*)
>> > >
>> > >
>> > > > including MemoryInstrumentationVector.h, should the header declaring
>>
> this
>
>> > > class
>> > > > include MIV.h?
>> > >
>> > > Done in the next patchset, though I don't completely understand when
>> to
>>
> use
>
>> <>
>> > > vs "" for the includes. Some wtf includes use <> while others use "".
>> >
>> > I meant move it to the header that requires including MIV.h (probably
>> > MemoryInstrumentationVector.h?**), not to the .h of the cc you touched.
>>
>
>  I still don't understand where I should move the includes to.
>>
>
>  The include file that's missing is MemoryInstrumentationVector.h. Where
>> should
>>
> I
>
>> add this include?
>>
>
>  I moved it to the cpp files for now.
>>
>
>  >
>> > Always use "", <> is deprecated for wtf includes.
>>
>
> thakis/timloh, ping.
>
>
https://codereview.chromium.**org/17707003/<https://codereview.chromium.org/1...
>

[Timothy Loh, 2013-06-27 03:48:00]

On 2013/06/26 19:43:21, asharif1 wrote:
> thakis/timloh, ping.

This seems fine to me (I mistakenly removed the include in
one of my remove unused includes patches). There's a link
time guard which makes us fail to link if we forget the
include, but possibly we aren't using this particular
reportMemoryUsage function so it gets optimised away?
It looks like we should have these includes in here since
we're reporting usage on vectors.

Sorry for breaking your build. lgtm (I don't have owners).

[loislo, 2013-06-27 04:10:54]

lgtm

[Nico, 2013-06-27 04:20:43]

lgtm stamp

[yurys, 2013-06-27 07:41:48]

lgtm

On 2013/06/26 19:44:45, Nico wrote:
> I was hoping yurys would take a look. +loislo for redundancy :-)
> 
> 
> On Wed, Jun 26, 2013 at 12:43 PM, <mailto:asharif@chromium.org> wrote:
> 
> > On 2013/06/25 18:41:09, asharif1 wrote:
> >
> >> On 2013/06/25 18:12:22, Nico wrote:
> >> > On 2013/06/25 17:51:30, asharif1 wrote:
> >> > > On 2013/06/25 17:40:34, Nico wrote:
> >> > > >
> >> > >
> >> >
> >>
> >
> > https://codereview.chromium.**org/17707003/diff/1/Source/**
> >
>
core/css/CSSRuleList.cpp<https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp>
> >
> >> > > > File Source/core/css/CSSRuleList.**cpp (right):
> >> > > >
> >> > > >
> >> > >
> >> >
> >>
> >
> > https://codereview.chromium.**org/17707003/diff/1/Source/**
> >
>
core/css/CSSRuleList.cpp#**newcode56<https://codereview.chromium.org/17707003/diff/1/Source/core/css/CSSRuleList.cpp#newcode56>
> >
> >> > > > Source/core/css/CSSRuleList.**cpp:56: MemoryClassInfo
> >>
> > info(memoryObjectInfo,
> >
> >> > > this,
> >> > > > WebCoreMemoryTypes::CSS);
> >> > > > Why does this compile without the include? If using this class
> >> requires
> >> > >
> >> > > It actually fails to compile without the include with
> >> -fprofile-generate.
> >> I'm
> >> > > guessing for the regular build it works because the function call is
> >> optimized
> >> > > away for some reason.
> >> > >
> >> > > See:
> >> > >
> >> >
> >>
> >
> > http://chromegw/i/chromeos/**builders/lumpy%2525252520pgo%25**
> > 25252520canary%25252520%**25252528experimental%25252529/**
> > builds/210/steps/**BuildPackages%25252520%**2525255Bpgo_generate%2525255D%
> >
>
**25252520%2525255Blumpy%**2525255D/logs/stdio<http://chromegw/i/chromeos/builders/lumpy%25252520pgo%25252520canary%25252520%25252528experimental%25252529/builds/210/steps/BuildPackages%25252520%2525255Bpgo_generate%2525255D%25252520%2525255Blumpy%2525255D/logs/stdio>
> >
> >  > >
> >> > > It fails to link when built with USE="pgo_generate".
> >> > >
> >> > > When built with -fprofile-generate, I ran:
> >> > >
> >> > > readelf -sW CSSRuleList.o | grep
> >> > >
> >> >
> >>
> >
> > _ZN3WTF17reportMemoryUsageINS_**6RefPtrIN7WebCore7CSSRuleEEELm**
> > 0EEEvPKNS_6VectorIT_XT0_**EEEPNS_16MemoryObjectInfoE
> >
> >> > >
> >> > >    515: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND
> >> > >
> >> >
> >>
> >
> > _ZN3WTF17reportMemoryUsageINS_**6RefPtrIN7WebCore7CSSRuleEEELm**
> > 0EEEvPKNS_6VectorIT_XT0_**EEEPNS_16MemoryObjectInfoE
> >
> >> > >
> >> > >  c++filt
> >> > >
> >> >
> >>
> >
> > _ZN3WTF17reportMemoryUsageINS_**6RefPtrIN7WebCore7CSSRuleEEELm**
> > 0EEEvPKNS_6VectorIT_XT0_**EEEPNS_16MemoryObjectInfoE
> >
> >> > > void WTF::reportMemoryUsage<WTF::**RefPtr<WebCore::CSSRule>,
> >> > > 0ul>(WTF::Vector<WTF::RefPtr<**WebCore::CSSRule>, 0ul> const*,
> >> > > WTF::MemoryObjectInfo*)
> >> > >
> >> > >
> >> > > > including MemoryInstrumentationVector.h, should the header declaring
> >>
> > this
> >
> >> > > class
> >> > > > include MIV.h?
> >> > >
> >> > > Done in the next patchset, though I don't completely understand when
> >> to
> >>
> > use
> >
> >> <>
> >> > > vs "" for the includes. Some wtf includes use <> while others use "".
> >> >
> >> > I meant move it to the header that requires including MIV.h (probably
> >> > MemoryInstrumentationVector.h?**), not to the .h of the cc you touched.
> >>
> >
> >  I still don't understand where I should move the includes to.
> >>
> >
> >  The include file that's missing is MemoryInstrumentationVector.h. Where
> >> should
> >>
> > I
> >
> >> add this include?
> >>
> >
> >  I moved it to the cpp files for now.
> >>
> >
> >  >
> >> > Always use "", <> is deprecated for wtf includes.
> >>
> >
> > thakis/timloh, ping.
> >
> >
>
https://codereview.chromium.**org/17707003/%3Chttps://codereview.chromium.org...>
> >

[commit-bot: I haz the power, 2013-06-27 15:06:40]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/asharif@chromium.org/17707003/13001

[commit-bot: I haz the power, 2013-06-27 16:38:34]

Retried try job too often on win_layout_rel for step(s) webkit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_layout...

[commit-bot: I haz the power, 2013-06-27 16:44:55]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/asharif@chromium.org/17707003/13001

[commit-bot: I haz the power, 2013-06-27 18:26:25]

Change committed as 153163