Upgrade to google-perftools revision 77....

Upgrade to google-perftools revision 77.
This brings several bugfixes needed for turning on the heap leak checker inside
tcmalloc.
Unforked malloc_hook.cc for Linux

[dank, 2009-10-29 15:41:39]

win trybot failure?

c:\b\slave\win\build\src\third_party\tcmalloc\tcmalloc\src\debugallocation.cc(37)
: fatal error C1083: Cannot open include file: 'pthread.h': No such file or
directory

[willchan no longer on Chromium, 2009-10-29 16:56:59]

http://codereview.chromium.org/339065/diff/10/12
File third_party/tcmalloc/tcmalloc.gyp (right):

http://codereview.chromium.org/339065/diff/10/12#newcode102
Line 102: 'tcmalloc/src/debugallocation.cc',
We don't want to depend on debugallocation yet.  It's not ready for public
consumption.  It's just there in the package so people can hack on it.

http://codereview.chromium.org/339065/diff/10/12#newcode106
Line 106: 'tcmalloc/src/symbolize.cc',
This needs to be added to the sources! for windows, right?

[not_the_right_glider, 2009-10-29 18:58:07]

On 2009/10/29 16:56:59, willchan wrote:
> http://codereview.chromium.org/339065/diff/10/12
> File third_party/tcmalloc/tcmalloc.gyp (right):
> 
> http://codereview.chromium.org/339065/diff/10/12#newcode102
> Line 102: 'tcmalloc/src/debugallocation.cc',
> We don't want to depend on debugallocation yet.  It's not ready for public
> consumption.  It's just there in the package so people can hack on it.
> 
> http://codereview.chromium.org/339065/diff/10/12#newcode106
> Line 106: 'tcmalloc/src/symbolize.cc',
> This needs to be added to the sources! for windows, right?

Fixed both comments, thank you.

[Mike Belshe, 2009-10-29 19:05:30]

looks good to me.

My only question is has it been tested for performance regressions.  Check the
perf bots after checkin.

[not_the_right_glider, 2009-10-29 19:19:06]

Fixed.

The trybot is experiencing some problems with the repositories on Patch Set 6,
but you can refer to Patch Set 5, as the difference between them is already
committed as http://codereview.chromium.org/342039

On 2009/10/29 15:41:39, dank wrote:
> win trybot failure?
> 
>
c:\b\slave\win\build\src\third_party\tcmalloc\tcmalloc\src\debugallocation.cc(37)
> : fatal error C1083: Cannot open include file: 'pthread.h': No such file or
> directory

[not_the_right_glider, 2009-10-29 19:20:32]

On 2009/10/29 19:05:30, Mike Belshe wrote:
> looks good to me.
> 
> My only question is has it been tested for performance regressions.  Check the
> perf bots after checkin.

Do perf bots build the tests with tcmalloc? It is off on Linux at the moment.

[willchan no longer on Chromium, 2009-10-29 19:32:18]

On Thu, Oct 29, 2009 at 12:20 PM,  <ramosian.glider@gmail.com> wrote:
> On 2009/10/29 19:05:30, Mike Belshe wrote:
>>
>> looks good to me.
>
>> My only question is has it been tested for performance regressions. =A0C=
heck
>> the
>> perf bots after checkin.
>
> Do perf bots build the tests with tcmalloc? It is off on Linux at the
> moment.

It will be built and run for Windows.  It will not run for Linux, no.

>
> http://codereview.chromium.org/339065
>

[mbelshe, 2009-10-29 19:33:56]

They do for windows.  And your changes (rolling the google-perftools
version) will effect windows as well.  But moving forward is good.

Mike


On Thu, Oct 29, 2009 at 12:20 PM, <ramosian.glider@gmail.com> wrote:

> On 2009/10/29 19:05:30, Mike Belshe wrote:
>
>> looks good to me.
>>
>
>  My only question is has it been tested for performance regressions.  Check
>> the
>> perf bots after checkin.
>>
>
> Do perf bots build the tests with tcmalloc? It is off on Linux at the
> moment.
>
>
> http://codereview.chromium.org/339065
>

[willchan no longer on Chromium, 2009-10-29 19:36:39]

Mike, not all of windows (and linux too) will move forward, right?  We
still have those forked files which won't pick up the changes up to
r77.  We need to update all the forked files.

On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe <mbelshe@google.com> wrote:
> They do for windows. =A0And your changes (rolling the google-perftools
> version) will effect windows as well. =A0But moving forward is good.
> Mike
>
> On Thu, Oct 29, 2009 at 12:20 PM, <ramosian.glider@gmail.com> wrote:
>>
>> On 2009/10/29 19:05:30, Mike Belshe wrote:
>>>
>>> looks good to me.
>>
>>> My only question is has it been tested for performance regressions.
>>> =A0Check the
>>> perf bots after checkin.
>>
>> Do perf bots build the tests with tcmalloc? It is off on Linux at the
>> moment.
>>
>> http://codereview.chromium.org/339065
>
>

[mbelshe, 2009-10-29 19:41:24]

yes


On Thu, Oct 29, 2009 at 12:36 PM, William Chan (=E9=99=88=E6=99=BA=E6=98=8C=
)
<willchan@chromium.org>wrote:

> Mike, not all of windows (and linux too) will move forward, right?  We
> still have those forked files which won't pick up the changes up to
> r77.  We need to update all the forked files.
>
> On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe <mbelshe@google.com> wrote:
> > They do for windows.  And your changes (rolling the google-perftools
> > version) will effect windows as well.  But moving forward is good.
> > Mike
> >
> > On Thu, Oct 29, 2009 at 12:20 PM, <ramosian.glider@gmail.com> wrote:
> >>
> >> On 2009/10/29 19:05:30, Mike Belshe wrote:
> >>>
> >>> looks good to me.
> >>
> >>> My only question is has it been tested for performance regressions.
> >>>  Check the
> >>> perf bots after checkin.
> >>
> >> Do perf bots build the tests with tcmalloc? It is off on Linux at the
> >> moment.
> >>
> >> http://codereview.chromium.org/339065
> >
> >
>

[not_the_right_glider, 2009-10-30 09:57:03]

I've unforked third_party/tcmalloc/malloc_hook.cc -- seems that the problem is
fixed in r77. Are there any other local patches that aren't needed anymore?
On 2009/10/29 19:36:39, willchan wrote:
> Mike, not all of windows (and linux too) will move forward, right?  We
> still have those forked files which won't pick up the changes up to
> r77.  We need to update all the forked files.
> 
> On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe <mailto:mbelshe@google.com>
wrote:
> > They do for windows. =A0And your changes (rolling the google-perftools
> > version) will effect windows as well. =A0But moving forward is good.
> > Mike
> >
> > On Thu, Oct 29, 2009 at 12:20 PM, <mailto:ramosian.glider@gmail.com> wrote:
> >>
> >> On 2009/10/29 19:05:30, Mike Belshe wrote:
> >>>
> >>> looks good to me.
> >>
> >>> My only question is has it been tested for performance regressions.
> >>> =A0Check the
> >>> perf bots after checkin.
> >>
> >> Do perf bots build the tests with tcmalloc? It is off on Linux at the
> >> moment.
> >>
> >> http://codereview.chromium.org/339065
> >
> >

[mbelshe, 2009-10-30 17:05:49]

On Fri, Oct 30, 2009 at 2:57 AM, <ramosian.glider@gmail.com> wrote:

> I've unforked third_party/tcmalloc/malloc_hook.cc -- seems that the problem
> is
> fixed in r77. Are there any other local patches that aren't needed anymore?
>
>
Cool - thanks!

The other forks haven't been upstreamed yet (as far as I know)

Mike



> On 2009/10/29 19:36:39, willchan wrote:
>
>> Mike, not all of windows (and linux too) will move forward, right?  We
>> still have those forked files which won't pick up the changes up to
>> r77.  We need to update all the forked files.
>>
>
>  On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe <mailto:mbelshe@google.com>
>>
> wrote:
>
>> > They do for windows. =A0And your changes (rolling the google-perftools
>> > version) will effect windows as well. =A0But moving forward is good.
>> > Mike
>>
>> >
>> > On Thu, Oct 29, 2009 at 12:20 PM, <mailto:ramosian.glider@gmail.com>
>> wrote:
>> >>
>> >> On 2009/10/29 19:05:30, Mike Belshe wrote:
>> >>>
>> >>> looks good to me.
>> >>
>> >>> My only question is has it been tested for performance regressions.
>> >>> =A0Check the
>>
>> >>> perf bots after checkin.
>> >>
>> >> Do perf bots build the tests with tcmalloc? It is off on Linux at the
>> >> moment.
>> >>
>> >> http://codereview.chromium.org/339065
>> >
>> >
>>
>
>
>
> http://codereview.chromium.org/339065
>

[not_the_right_glider, 2009-10-30 20:08:15]

Should we submit and monitor the perf bots then?
On 2009/10/30 17:05:49, mbelshe wrote:
> On Fri, Oct 30, 2009 at 2:57 AM, <mailto:ramosian.glider@gmail.com> wrote:
> 
> > I've unforked third_party/tcmalloc/malloc_hook.cc -- seems that the problem
> > is
> > fixed in r77. Are there any other local patches that aren't needed anymore?
> >
> >
> Cool - thanks!
> 
> The other forks haven't been upstreamed yet (as far as I know)
> 
> Mike
> 
> 
> 
> > On 2009/10/29 19:36:39, willchan wrote:
> >
> >> Mike, not all of windows (and linux too) will move forward, right?  We
> >> still have those forked files which won't pick up the changes up to
> >> r77.  We need to update all the forked files.
> >>
> >
> >  On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe <mailto:mbelshe@google.com>
> >>
> > wrote:
> >
> >> > They do for windows. =A0And your changes (rolling the google-perftools
> >> > version) will effect windows as well. =A0But moving forward is good.
> >> > Mike
> >>
> >> >
> >> > On Thu, Oct 29, 2009 at 12:20 PM, <mailto:ramosian.glider@gmail.com>
> >> wrote:
> >> >>
> >> >> On 2009/10/29 19:05:30, Mike Belshe wrote:
> >> >>>
> >> >>> looks good to me.
> >> >>
> >> >>> My only question is has it been tested for performance regressions.
> >> >>> =A0Check the
> >>
> >> >>> perf bots after checkin.
> >> >>
> >> >> Do perf bots build the tests with tcmalloc? It is off on Linux at the
> >> >> moment.
> >> >>
> >> >> http://codereview.chromium.org/339065
> >> >
> >> >
> >>
> >
> >
> >
> > http://codereview.chromium.org/339065
> >
>

[not_the_right_glider, 2009-10-30 20:09:59]

Should we submit and monitor the perf bots then?
On 2009/10/30 17:05:49, mbelshe wrote:
> On Fri, Oct 30, 2009 at 2:57 AM, <mailto:ramosian.glider@gmail.com> wrote:
> 
> > I've unforked third_party/tcmalloc/malloc_hook.cc -- seems that the problem
> > is
> > fixed in r77. Are there any other local patches that aren't needed anymore?
> >
> >
> Cool - thanks!
> 
> The other forks haven't been upstreamed yet (as far as I know)
> 
> Mike
> 
> 
> 
> > On 2009/10/29 19:36:39, willchan wrote:
> >
> >> Mike, not all of windows (and linux too) will move forward, right?  We
> >> still have those forked files which won't pick up the changes up to
> >> r77.  We need to update all the forked files.
> >>
> >
> >  On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe <mailto:mbelshe@google.com>
> >>
> > wrote:
> >
> >> > They do for windows. =A0And your changes (rolling the google-perftools
> >> > version) will effect windows as well. =A0But moving forward is good.
> >> > Mike
> >>
> >> >
> >> > On Thu, Oct 29, 2009 at 12:20 PM, <mailto:ramosian.glider@gmail.com>
> >> wrote:
> >> >>
> >> >> On 2009/10/29 19:05:30, Mike Belshe wrote:
> >> >>>
> >> >>> looks good to me.
> >> >>
> >> >>> My only question is has it been tested for performance regressions.
> >> >>> =A0Check the
> >>
> >> >>> perf bots after checkin.
> >> >>
> >> >> Do perf bots build the tests with tcmalloc? It is off on Linux at the
> >> >> moment.
> >> >>
> >> >> http://codereview.chromium.org/339065
> >> >
> >> >
> >>
> >
> >
> >
> > http://codereview.chromium.org/339065
> >
>

[willchan no longer on Chromium, 2009-10-30 20:48:32]

I'm fine with that.  If you missed anything, I can verify later.

On Fri, Oct 30, 2009 at 1:09 PM,  <ramosian.glider@gmail.com> wrote:
> Should we submit and monitor the perf bots then?
> On 2009/10/30 17:05:49, mbelshe wrote:
>>
>> On Fri, Oct 30, 2009 at 2:57 AM, <mailto:ramosian.glider@gmail.com> wrot=
e:
>
>> > I've unforked third_party/tcmalloc/malloc_hook.cc -- seems that the
>> > problem
>> > is
>> > fixed in r77. Are there any other local patches that aren't needed
>> > anymore?
>> >
>> >
>> Cool - thanks!
>
>> The other forks haven't been upstreamed yet (as far as I know)
>
>> Mike
>
>
>
>> > On 2009/10/29 19:36:39, willchan wrote:
>> >
>> >> Mike, not all of windows (and linux too) will move forward, right? =
=A0We
>> >> still have those forked files which won't pick up the changes up to
>> >> r77. =A0We need to update all the forked files.
>> >>
>> >
>> > =A0On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe
>> > <mailto:mbelshe@google.com>
>> >>
>> > wrote:
>> >
>> >> > They do for windows. =3DA0And your changes (rolling the
>> >> > google-perftools
>> >> > version) will effect windows as well. =3DA0But moving forward is go=
od.
>> >> > Mike
>> >>
>> >> >
>> >> > On Thu, Oct 29, 2009 at 12:20 PM, <mailto:ramosian.glider@gmail.com=
>
>> >> wrote:
>> >> >>
>> >> >> On 2009/10/29 19:05:30, Mike Belshe wrote:
>> >> >>>
>> >> >>> looks good to me.
>> >> >>
>> >> >>> My only question is has it been tested for performance regression=
s.
>> >> >>> =3DA0Check the
>> >>
>> >> >>> perf bots after checkin.
>> >> >>
>> >> >> Do perf bots build the tests with tcmalloc? It is off on Linux at
>> >> >> the
>> >> >> moment.
>> >> >>
>> >> >> http://codereview.chromium.org/339065
>> >> >
>> >> >
>> >>
>> >
>> >
>> >
>> > http://codereview.chromium.org/339065
>> >
>
>
>
>
> http://codereview.chromium.org/339065
>

[dkegel, 2009-10-30 21:46:10]

+1

On Fri, Oct 30, 2009 at 1:48 PM, William Chan (=E9=99=88=E6=99=BA=E6=98=8C)
<willchan@chromium.org> wrote:
> I'm fine with that. =C2=A0If you missed anything, I can verify later.
>
> On Fri, Oct 30, 2009 at 1:09 PM, =C2=A0<ramosian.glider@gmail.com> wrote:
>> Should we submit and monitor the perf bots then?
>> On 2009/10/30 17:05:49, mbelshe wrote:
>>>
>>> On Fri, Oct 30, 2009 at 2:57 AM, <mailto:ramosian.glider@gmail.com> wro=
te:
>>
>>> > I've unforked third_party/tcmalloc/malloc_hook.cc -- seems that the
>>> > problem
>>> > is
>>> > fixed in r77. Are there any other local patches that aren't needed
>>> > anymore?
>>> >
>>> >
>>> Cool - thanks!
>>
>>> The other forks haven't been upstreamed yet (as far as I know)
>>
>>> Mike
>>
>>
>>
>>> > On 2009/10/29 19:36:39, willchan wrote:
>>> >
>>> >> Mike, not all of windows (and linux too) will move forward, right? =
=C2=A0We
>>> >> still have those forked files which won't pick up the changes up to
>>> >> r77. =C2=A0We need to update all the forked files.
>>> >>
>>> >
>>> > =C2=A0On Thu, Oct 29, 2009 at 12:33 PM, Mike Belshe
>>> > <mailto:mbelshe@google.com>
>>> >>
>>> > wrote:
>>> >
>>> >> > They do for windows. =3DA0And your changes (rolling the
>>> >> > google-perftools
>>> >> > version) will effect windows as well. =3DA0But moving forward is g=
ood.
>>> >> > Mike
>>> >>
>>> >> >
>>> >> > On Thu, Oct 29, 2009 at 12:20 PM, <mailto:ramosian.glider@gmail.co=
m>
>>> >> wrote:
>>> >> >>
>>> >> >> On 2009/10/29 19:05:30, Mike Belshe wrote:
>>> >> >>>
>>> >> >>> looks good to me.
>>> >> >>
>>> >> >>> My only question is has it been tested for performance regressio=
ns.
>>> >> >>> =3DA0Check the
>>> >>
>>> >> >>> perf bots after checkin.
>>> >> >>
>>> >> >> Do perf bots build the tests with tcmalloc? It is off on Linux at
>>> >> >> the
>>> >> >> moment.
>>> >> >>
>>> >> >> http://codereview.chromium.org/339065
>>> >> >
>>> >> >
>>> >>
>>> >
>>> >
>>> >
>>> > http://codereview.chromium.org/339065
>>> >
>>
>>
>>
>>
>> http://codereview.chromium.org/339065
>>
>

[not_the_right_glider, 2009-12-08 11:32:06]

Committed as http://codereview.chromium.org/340065