Split up trace_event.h into trace_event_impl.h and add webkitplatform support.

Split up trace_event.h into trace_event_impl.h and add webkitplatform support.

No new features here, this is just a code split and API glue.

(Good news! I copied over trace_event.h to WebKit and verified that this all works!)

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=118733

[jbates, 2012-01-13 22:19:15]

nduca: please review
darin: base/ and webkit/ owners

[darin (slow to review), 2012-01-13 22:37:12]

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
File webkit/glue/webkitplatformsupport_impl.cc (right):

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
webkit/glue/webkitplatformsupport_impl.cc:342: unsigned long long
WebKitPlatformSupportImpl::getInterProcessID(void* pointer) {
Is it necessary for the WebKit API to talk about "inter-process" stuff?  Can
this be more generic from the point of view of WebKit?

[nduca, 2012-01-13 22:40:41]

This should really be two cls, not one. One to wkplatformsupport and one to
base/.

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
File webkit/glue/webkitplatformsupport_impl.cc (right):

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
webkit/glue/webkitplatformsupport_impl.cc:342: unsigned long long
WebKitPlatformSupportImpl::getInterProcessID(void* pointer) {
manglePointerForTracing?

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
File webkit/glue/webkitplatformsupport_impl.h (right):

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
webkit/glue/webkitplatformsupport_impl.h:89: virtual void addCounterEvent(
Er, before this, is there any hope of folding counters into addTraceEvent? I
would have expected the arg_types refactoring stuff to have fixed the issue that
prevented me from using it originally [though I've forgotten what the issue was,
tbh]

[darin (slow to review), 2012-01-13 22:53:54]

On Fri, Jan 13, 2012 at 2:40 PM, <nduca@chromium.org> wrote:

> This should really be two cls, not one. One to wkplatformsupport and one to
> base/.
>
>
>
>
> http://codereview.chromium.**org/9213013/diff/1/webkit/**
>
glue/webkitplatformsupport_**impl.cc<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.cc>
> File webkit/glue/**webkitplatformsupport_impl.cc (right):
>
> http://codereview.chromium.**org/9213013/diff/1/webkit/**
>
glue/webkitplatformsupport_**impl.cc#newcode342<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.cc#newcode342>
> webkit/glue/**webkitplatformsupport_impl.cc:**342: unsigned long long
> WebKitPlatformSupportImpl::**getInterProcessID(void* pointer) {
> manglePointerForTracing?
>

What is the purpose of this function?  Is it to create a globally unique
identifier for the address?  Or, does it do something else?
-Darin


>
> http://codereview.chromium.**org/9213013/diff/1/webkit/**
>
glue/webkitplatformsupport_**impl.h<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.h>
> File webkit/glue/**webkitplatformsupport_impl.h (right):
>
> http://codereview.chromium.**org/9213013/diff/1/webkit/**
>
glue/webkitplatformsupport_**impl.h#newcode89<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.h#newcode89>
> webkit/glue/**webkitplatformsupport_impl.h:**89: virtual void
> addCounterEvent(
> Er, before this, is there any hope of folding counters into
> addTraceEvent? I would have expected the arg_types refactoring stuff to
> have fixed the issue that prevented me from using it originally [though
> I've forgotten what the issue was, tbh]
>
>
http://codereview.chromium.**org/9213013/<http://codereview.chromium.org/9213...
>

[jbates, 2012-01-13 22:57:22]

On 2012/01/13 22:53:54, darin wrote:
> On Fri, Jan 13, 2012 at 2:40 PM, <mailto:nduca@chromium.org> wrote:
> 
> > This should really be two cls, not one. One to wkplatformsupport and one to
> > base/.
> >
> >
> >
> >
> > http://codereview.chromium.**org/9213013/diff/1/webkit/**
> >
>
glue/webkitplatformsupport_**impl.cc<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.cc>
> > File webkit/glue/**webkitplatformsupport_impl.cc (right):
> >
> > http://codereview.chromium.**org/9213013/diff/1/webkit/**
> >
>
glue/webkitplatformsupport_**impl.cc#newcode342<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.cc#newcode342>
> > webkit/glue/**webkitplatformsupport_impl.cc:**342: unsigned long long
> > WebKitPlatformSupportImpl::**getInterProcessID(void* pointer) {
> > manglePointerForTracing?
> >
> 
> What is the purpose of this function?  Is it to create a globally unique
> identifier for the address? 

Yeah, really simple, it just casts to int and xor's with a hash of the process
ID. That way if you're tracing two renderer processes and the same pointer is
traced in both, we will know which is which. (There's code that needs to
associate begin/end events based on pointers.)

> Or, does it do something else?
> -Darin
> 
> 
> >
> > http://codereview.chromium.**org/9213013/diff/1/webkit/**
> >
>
glue/webkitplatformsupport_**impl.h<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.h>
> > File webkit/glue/**webkitplatformsupport_impl.h (right):
> >
> > http://codereview.chromium.**org/9213013/diff/1/webkit/**
> >
>
glue/webkitplatformsupport_**impl.h#newcode89<http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsupport_impl.h#newcode89>
> > webkit/glue/**webkitplatformsupport_impl.h:**89: virtual void
> > addCounterEvent(
> > Er, before this, is there any hope of folding counters into
> > addTraceEvent? I would have expected the arg_types refactoring stuff to
> > have fixed the issue that prevented me from using it originally [though
> > I've forgotten what the issue was, tbh]
> >
> >
>
http://codereview.chromium.**org/9213013/%3Chttp://codereview.chromium.org/92...>
> >

[jbates, 2012-01-13 23:56:03]

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
File webkit/glue/webkitplatformsupport_impl.cc (right):

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
webkit/glue/webkitplatformsupport_impl.cc:342: unsigned long long
WebKitPlatformSupportImpl::getInterProcessID(void* pointer) {
On 2012/01/13 22:40:41, nduca wrote:
> manglePointerForTracing?

Done.

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
File webkit/glue/webkitplatformsupport_impl.h (right):

http://codereview.chromium.org/9213013/diff/1/webkit/glue/webkitplatformsuppo...
webkit/glue/webkitplatformsupport_impl.h:89: virtual void addCounterEvent(
On 2012/01/13 22:40:41, nduca wrote:
> Er, before this, is there any hope of folding counters into addTraceEvent? I
> would have expected the arg_types refactoring stuff to have fixed the issue
that
> prevented me from using it originally [though I've forgotten what the issue
was,
> tbh]

Done.

[nduca, 2012-01-14 00:05:35]

LGTM

[darin (slow to review), 2012-01-14 00:13:53]

On Fri, Jan 13, 2012 at 2:57 PM, <jbates@chromium.org> wrote:
...

> What is the purpose of this function?  Is it to create a globally unique
>> identifier for the address?
>>
>
> Yeah, really simple, it just casts to int and xor's with a hash of the
> process
> ID. That way if you're tracing two renderer processes and the same pointer
> is
> traced in both, we will know which is which. (There's code that needs to
> associate begin/end events based on pointers.)


OK

http://codereview.chromium.**org/9213013/<http://codereview.chromium.org/9213...
>

[darin (slow to review), 2012-01-14 00:14:53]

LGTM

I'm curious what the changes to WebKit will look like.  Are you intending to
just copy the header into WebKit, or do you intend to refer to it in some other
way?

[jbates, 2012-01-23 18:51:34]

On 2012/01/14 00:14:53, darin wrote:
> LGTM
> 
> I'm curious what the changes to WebKit will look like.  Are you intending to
> just copy the header into WebKit, or do you intend to refer to it in some
other
> way?

Yeah, planning to copy the contents from base/debug/trace_event.h to
Source/WebCore/platform/chromium/TraceEvent.h. From WebKit, the macros are
changed to call through the WebKitPlatformSupport API.