Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting it: could not find any active user of that.
See also discussion in project-trim@chromium.org @ goo.gl/p3GEUy .

After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
Committed: https://crrev.com/196fb1f13813c464e3f1effac15f5c557308015b
Cr-Commit-Position: refs/heads/master@{#371770}

[Primiano Tucci (use gerrit), 2016-01-26 19:32:14]

Description was changed from

==========
Allocator cleanup: remove stale dependenceis on tcmalloc

After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
==========

to

==========
Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting that: could not find any active user of that.
After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
==========

[Primiano Tucci (use gerrit), 2016-01-26 19:32:15]

primiano@chromium.org changed reviewers:
+ fmeawad@chromium.org, jar@chromium.org, rtenneti@chromium.org,
sque@chromium.org

[Primiano Tucci (use gerrit), 2016-01-26 19:33:08]

Before I reach out to OWNERS, is anybody aware of an actual use of
CHROME_PROFILER_TIME env var ?
Couldn't find any in the codebase, will drop if nobody speaks.

[Primiano Tucci (use gerrit), 2016-01-26 19:35:34]

primiano@chromium.org changed reviewers:
- fmeawad@chromium.org, jar@chromium.org, rtenneti@chromium.org,
sque@chromium.org

[Primiano Tucci (use gerrit), 2016-01-26 19:35:34]

+brucedawson +crisha just in case somebody is using CHROME_PROFILER_TIME in
Windows-land.

For context: not asking for a review here, just asking if anybody still uses
CHROME_PROFILER_TIME

[Primiano Tucci (use gerrit), 2016-01-26 19:58:56]

On 2016/01/26 19:35:34, Primiano Tucci wrote:
> +brucedawson +crisha just in case somebody is using CHROME_PROFILER_TIME in
> Windows-land.
> 
> For context: not asking for a review here, just asking if anybody still uses
> CHROME_PROFILER_TIME

Ignore my last message. No way you could be using this on Windows as this is
#ifdef TCMALLOC, which means only Linux/CrOs.

[Primiano Tucci (use gerrit), 2016-01-26 19:59:46]

primiano@chromium.org changed reviewers:
+ andrewhayden@chromium.org, sievers@chromium.org, thakis@chromium.org

[Primiano Tucci (use gerrit), 2016-01-26 19:59:47]

+andrewhayden@ for third_party/cld

[Primiano Tucci (use gerrit), 2016-01-26 20:00:37]

(hit enter too quickly, apologies for the double mail)
+thakis for ---lines in base/
+sievers for ---lines in content/

[Nico, 2016-01-26 21:14:41]

lgtm

Sounds like about:profiler doesn't use this (anymore? or maybe never did)

https://codereview.chromium.org/1632253002/diff/20001/base/profiler/alternate...
File base/profiler/alternate_timer.h (right):

https://codereview.chromium.org/1632253002/diff/20001/base/profiler/alternate...
base/profiler/alternate_timer.h:15: enum TimeSourceType {
TIME_SOURCE_TYPE_WALL_TIME };
Maybe add a TODO to remove this enum altogether

[no sievers, 2016-01-26 22:04:12]

lgtm

[Primiano Tucci (use gerrit), 2016-01-26 23:40:16]

Description was changed from

==========
Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting that: could not find any active user of that.
After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
==========

to

==========
Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting it: could not find any active user of that.
See also discussion in project-trim@chromium.org @ goo.gl/p3GEUy .

After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
==========

[Primiano Tucci (use gerrit), 2016-01-26 23:45:25]

> Sounds like about:profiler doesn't use this (anymore? or maybe never did)
Yup. See discussion in https://goo.gl/p3GEUy (updated the CL description also to
point at this)

>
https://codereview.chromium.org/1632253002/diff/20001/base/profiler/alternate...
> base/profiler/alternate_timer.h:15: enum TimeSourceType {
> TIME_SOURCE_TYPE_WALL_TIME };
> Maybe add a TODO to remove this enum altogether

Ehm, at this point I plan to remove alternate_timer.cc/h altogether next not
just the enum 
Right now TIME_SOURCE_TYPE_WALL_TIME is relied on by some code in
tracked_object.cc, but given this CL that can be simplified a lot removing all
the branches that look for kAllowAlternateTimeSourceHandling.
Will give it a stab tomorrow.

[Primiano Tucci (use gerrit), 2016-01-27 09:25:23]

The CQ bit was checked by primiano@chromium.org

[commit-bot: I haz the power, 2016-01-27 09:26:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1632253002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1632253002/20001

[commit-bot: I haz the power, 2016-01-27 10:41:35]

Description was changed from

==========
Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting it: could not find any active user of that.
See also discussion in project-trim@chromium.org @ goo.gl/p3GEUy .

After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
==========

to

==========
Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting it: could not find any active user of that.
See also discussion in project-trim@chromium.org @ goo.gl/p3GEUy .

After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
==========

[commit-bot: I haz the power, 2016-01-27 10:41:36]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-01-27 10:42:55]

Description was changed from

==========
Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting it: could not find any active user of that.
See also discussion in project-trim@chromium.org @ goo.gl/p3GEUy .

After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365
==========

to

==========
Allocator cleanup: remove CHROME_PROFILER_TIME and rotten tcmalloc deps

This CL removes the support for the CHROME_PROFILER_TIME env var,
originally introduced by crrev.com/9212025 more than 3 years ago, which
allows to change the time source for the profiler via the environment.
Reason for deleting it: could not find any active user of that.
See also discussion in project-trim@chromium.org @ goo.gl/p3GEUy .

After this CL (almost*) nothing outside of base depends on tcmalloc.
This is the place where we want to be, as base should be the only thing
knowing about allocator internals.

* There is only one dep left in content_child_helpers.cc, which is
relied on by old-style telemetry memory benchmarks. But that code is
literally a dead man walking and there is no point refactoring it.

BUG=564618, 581365

Committed: https://crrev.com/196fb1f13813c464e3f1effac15f5c557308015b
Cr-Commit-Position: refs/heads/master@{#371770}
==========

[commit-bot: I haz the power, 2016-01-27 10:42:57]

Patchset 2 (id:??) landed as
https://crrev.com/196fb1f13813c464e3f1effac15f5c557308015b
Cr-Commit-Position: refs/heads/master@{#371770}