Enable the allocator shim on Linux Desktop / CrOS

Enable the allocator shim on Linux Desktop / CrOS

This CL enables the shim introduced in crrev.com/1675143004 by
default, only Linux desktop and CrOS.
This CL is functionally a no-op, in the sense that does not change
the default allocator (which remains tcmalloc on Linux/CrOS) and does
not introduce any new feature (other than the possibility, for future
CLs to install hooks in the allocator path).

See crrev.com/1675143004 for a longer description about the allocator
shim and the design doc bit.ly/allocator-shim.

Note for stability sheriffs
---------------------------
If you see suspicious crashes in tcmalloc (especially in free/tc_free)
there are pretty good chances that something unexpectedly went wrong
here and this CL is the culprit.

Note for perf sheriffs
----------------------
There is a possibility that this CL might cause a regression on the
perf waterfalls (only on Linux/CrOS bots) due to crbug.com/593344.
The telemetry tests I tried to run locally were all inconclusive.
Should a regression happen, a temporary workaround is possible as
discussed in https://codereview.chromium.org/1675143004/#msg38.

BUG=550886
TEST=base_unittests (AllocatorShimTest.*,OutOfMemoryDeathTest.*,TCMalloc*)

Committed: https://crrev.com/8712d70e87c419af545e2bad088126d0e19eb384
Cr-Commit-Position: refs/heads/master@{#380377}

[Primiano Tucci (use gerrit), 2016-03-09 16:42:34]

Description was changed from

==========
Enable allocator shim on Linux Desktop / CrOS

This CL make the change introduced in crrev.com/1675143004 on by
default on Linux desktop an CrOS.
This CL is functionally a no-op, in the sense that does not change
the default allocator (which remains tcmalloc on Linux/CrOS) and does
not introduce any new feature (other than the possibility, for future
CLs to install hooks in the allocator path).

See crrev.com/1675143004 for a longer description about the allocator
shim and the design doc bit.ly/allocator-shim.

Note for stability sheriffs
If you see suspicious crashes in tcmalloc (especially in free/tc_free)
there are pretty good chances that something unexpectedly went wrong
here and this CL is the culprit.

BUG=550886
TEST=base_unittests (AllocatorShimTest.*,OutOfMemoryDeathTest.*,TCMalloc*)
==========

to

==========
Enable the allocator shim on Linux Desktop / CrOS

This CL enables the shim introduced in crrev.com/1675143004 by
default, only Linux desktop and CrOS.
This CL is functionally a no-op, in the sense that does not change
the default allocator (which remains tcmalloc on Linux/CrOS) and does
not introduce any new feature (other than the possibility, for future
CLs to install hooks in the allocator path).

See crrev.com/1675143004 for a longer description about the allocator
shim and the design doc bit.ly/allocator-shim.

Note for stability sheriffs
---------------------------
If you see suspicious crashes in tcmalloc (especially in free/tc_free)
there are pretty good chances that something unexpectedly went wrong
here and this CL is the culprit.

Note for perf sheriffs
----------------------
There is a possibility that this CL might cause a regression on the
perf waterfalls (only on Linux/CrOS bots) due to crbug.com/593344.
The telemetry tests I tried to run locally were all inconclusive.
Should a regression happen, a temporary workaround is possible as
discussed in https://codereview.chromium.org/1675143004/#msg38.

BUG=550886
TEST=base_unittests (AllocatorShimTest.*,OutOfMemoryDeathTest.*,TCMalloc*)
==========

[Primiano Tucci (use gerrit), 2016-03-09 16:43:56]

primiano@chromium.org changed reviewers:
+ thakis@chromium.org

[Primiano Tucci (use gerrit), 2016-03-09 16:43:57]

Well, let's roll the dices :)

[Primiano Tucci (use gerrit), 2016-03-09 18:26:03]

Patchset #3 (id:40001) has been deleted

[Nico, 2016-03-09 18:41:12]

lgtm if you checked that this does in fact enable the shim in gyp (always hard
to tell if scoping is right in gyp from diffs :-/)

I kind of thought you'd implement this on all platforms and then switch, but 1
implement for one platform 2 turn on there 3 repeat makes more sense now that I
think about it.

[Primiano Tucci (use gerrit), 2016-03-09 18:42:57]

On 2016/03/09 18:41:12, Nico wrote:
> lgtm if you checked that this does in fact enable the shim in gyp (always hard
> to tell if scoping is right in gyp from diffs :-/)
> 
> I kind of thought you'd implement this on all platforms and then switch, but 1
> implement for one platform 2 turn on there 3 repeat makes more sense now that
I
> think about it.

Yup I'd be more for a progressive stickiness attempt.
I don't want to make all the chrome platform crash at the same time with one cl,
that spoils all the fun.

[Primiano Tucci (use gerrit), 2016-03-09 18:50:23]

On 2016/03/09 18:42:57, Primiano wrote:
> On 2016/03/09 18:41:12, Nico wrote:
> > lgtm if you checked that this does in fact enable the shim in gyp (always
hard
> > to tell if scoping is right in gyp from diffs :-/)

I was about to say "yes I tested it", and then realized that I screwed it up
last minute after adding the build_for_tool condition.
Fixed the check in the last patchset and, yes, PS4 tested again :)

[Primiano Tucci (use gerrit), 2016-03-10 08:24:26]

The CQ bit was checked by primiano@chromium.org

[Primiano Tucci (use gerrit), 2016-03-10 08:24:27]

The patchset sent to the CQ was uploaded after l-g-t-m from thakis@chromium.org
Link to the patchset: https://codereview.chromium.org/1781573002/#ps80001
(title: "Fix build_for_tool condition")

[commit-bot: I haz the power, 2016-03-10 08:24:40]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1781573002/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1781573002/80001

[commit-bot: I haz the power, 2016-03-10 09:17:13]

Description was changed from

==========
Enable the allocator shim on Linux Desktop / CrOS

This CL enables the shim introduced in crrev.com/1675143004 by
default, only Linux desktop and CrOS.
This CL is functionally a no-op, in the sense that does not change
the default allocator (which remains tcmalloc on Linux/CrOS) and does
not introduce any new feature (other than the possibility, for future
CLs to install hooks in the allocator path).

See crrev.com/1675143004 for a longer description about the allocator
shim and the design doc bit.ly/allocator-shim.

Note for stability sheriffs
---------------------------
If you see suspicious crashes in tcmalloc (especially in free/tc_free)
there are pretty good chances that something unexpectedly went wrong
here and this CL is the culprit.

Note for perf sheriffs
----------------------
There is a possibility that this CL might cause a regression on the
perf waterfalls (only on Linux/CrOS bots) due to crbug.com/593344.
The telemetry tests I tried to run locally were all inconclusive.
Should a regression happen, a temporary workaround is possible as
discussed in https://codereview.chromium.org/1675143004/#msg38.

BUG=550886
TEST=base_unittests (AllocatorShimTest.*,OutOfMemoryDeathTest.*,TCMalloc*)
==========

to

==========
Enable the allocator shim on Linux Desktop / CrOS

This CL enables the shim introduced in crrev.com/1675143004 by
default, only Linux desktop and CrOS.
This CL is functionally a no-op, in the sense that does not change
the default allocator (which remains tcmalloc on Linux/CrOS) and does
not introduce any new feature (other than the possibility, for future
CLs to install hooks in the allocator path).

See crrev.com/1675143004 for a longer description about the allocator
shim and the design doc bit.ly/allocator-shim.

Note for stability sheriffs
---------------------------
If you see suspicious crashes in tcmalloc (especially in free/tc_free)
there are pretty good chances that something unexpectedly went wrong
here and this CL is the culprit.

Note for perf sheriffs
----------------------
There is a possibility that this CL might cause a regression on the
perf waterfalls (only on Linux/CrOS bots) due to crbug.com/593344.
The telemetry tests I tried to run locally were all inconclusive.
Should a regression happen, a temporary workaround is possible as
discussed in https://codereview.chromium.org/1675143004/#msg38.

BUG=550886
TEST=base_unittests (AllocatorShimTest.*,OutOfMemoryDeathTest.*,TCMalloc*)
==========

[commit-bot: I haz the power, 2016-03-10 09:17:14]

Committed patchset #4 (id:80001)

[commit-bot: I haz the power, 2016-03-10 09:19:24]

Description was changed from

==========
Enable the allocator shim on Linux Desktop / CrOS

This CL enables the shim introduced in crrev.com/1675143004 by
default, only Linux desktop and CrOS.
This CL is functionally a no-op, in the sense that does not change
the default allocator (which remains tcmalloc on Linux/CrOS) and does
not introduce any new feature (other than the possibility, for future
CLs to install hooks in the allocator path).

See crrev.com/1675143004 for a longer description about the allocator
shim and the design doc bit.ly/allocator-shim.

Note for stability sheriffs
---------------------------
If you see suspicious crashes in tcmalloc (especially in free/tc_free)
there are pretty good chances that something unexpectedly went wrong
here and this CL is the culprit.

Note for perf sheriffs
----------------------
There is a possibility that this CL might cause a regression on the
perf waterfalls (only on Linux/CrOS bots) due to crbug.com/593344.
The telemetry tests I tried to run locally were all inconclusive.
Should a regression happen, a temporary workaround is possible as
discussed in https://codereview.chromium.org/1675143004/#msg38.

BUG=550886
TEST=base_unittests (AllocatorShimTest.*,OutOfMemoryDeathTest.*,TCMalloc*)
==========

to

==========
Enable the allocator shim on Linux Desktop / CrOS

This CL enables the shim introduced in crrev.com/1675143004 by
default, only Linux desktop and CrOS.
This CL is functionally a no-op, in the sense that does not change
the default allocator (which remains tcmalloc on Linux/CrOS) and does
not introduce any new feature (other than the possibility, for future
CLs to install hooks in the allocator path).

See crrev.com/1675143004 for a longer description about the allocator
shim and the design doc bit.ly/allocator-shim.

Note for stability sheriffs
---------------------------
If you see suspicious crashes in tcmalloc (especially in free/tc_free)
there are pretty good chances that something unexpectedly went wrong
here and this CL is the culprit.

Note for perf sheriffs
----------------------
There is a possibility that this CL might cause a regression on the
perf waterfalls (only on Linux/CrOS bots) due to crbug.com/593344.
The telemetry tests I tried to run locally were all inconclusive.
Should a regression happen, a temporary workaround is possible as
discussed in https://codereview.chromium.org/1675143004/#msg38.

BUG=550886
TEST=base_unittests (AllocatorShimTest.*,OutOfMemoryDeathTest.*,TCMalloc*)

Committed: https://crrev.com/8712d70e87c419af545e2bad088126d0e19eb384
Cr-Commit-Position: refs/heads/master@{#380377}
==========

[commit-bot: I haz the power, 2016-03-10 09:19:25]

Patchset 4 (id:??) landed as
https://crrev.com/8712d70e87c419af545e2bad088126d0e19eb384
Cr-Commit-Position: refs/heads/master@{#380377}