Re-land: base: Implement browser process support for discardable memory.

Re-land: base: Implement browser process support for discardable memory.

This implements support for in-process discardable
memory. Each in-process discardable memory instance is
associated with a discardable shared memory segment. This
is hopefully efficient enough with the limited use of
discardable memory in the browser process.
DiscardableSharedMemoryHeap can be used if this is not
enough.

This also moves the test implementation of the discardable
memory allocator into a TestDiscardableMemoryShmemAllocator
class and requires tests that use discardable memory to
explicitly use this test allocator. Browser tests uses the
real discardable memory allocator.

BUG=422953, 463250
Committed: https://crrev.com/2799f7fb3d3a2406b1608f0b9bc470ff76ad4ea0
Cr-Commit-Position: refs/heads/master@{#320764}

[reveman, 2015-03-13 19:43:11]

reveman@chromium.org changed reviewers:
+ avi@chromium.org

[Avi (use Gerrit), 2015-03-13 21:05:26]

LGTM with nits.

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
File base/test/test_discardable_memory_shmem_allocator.cc (right):

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
base/test/test_discardable_memory_shmem_allocator.cc:13: : memory_(new
uint8[size]) {}
As noted in basictypes.h, do not use these aliases. Include <stdint.h> and use
the standard uint8_t.

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
File base/test/test_discardable_memory_shmem_allocator.h (right):

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
base/test/test_discardable_memory_shmem_allocator.h:1: // Copyright (c) 2015 The
Chromium Authors. All rights reserved.
No (c)

[reveman, 2015-03-13 21:31:23]

reveman@chromium.org changed reviewers:
+ danakj@chromium.org, jcivelli@chromium.org, oshima@chromium.org,
stevenjb@chromium.org

[reveman, 2015-03-13 21:31:24]

+danakj for base/
+oshima for ash/test/
+jcivelli for chrome/test/
+stevenjb for ui/

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
File base/test/test_discardable_memory_shmem_allocator.cc (right):

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
base/test/test_discardable_memory_shmem_allocator.cc:13: : memory_(new
uint8[size]) {}
On 2015/03/13 at 21:05:26, Avi wrote:
> As noted in basictypes.h, do not use these aliases. Include <stdint.h> and use
the standard uint8_t.

Done

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
File base/test/test_discardable_memory_shmem_allocator.h (right):

https://codereview.chromium.org/1004043002/diff/60001/base/test/test_discarda...
base/test/test_discardable_memory_shmem_allocator.h:1: // Copyright (c) 2015 The
Chromium Authors. All rights reserved.
On 2015/03/13 at 21:05:26, Avi wrote:
> No (c)

Done

[stevenjb, 2015-03-13 21:42:41]

lgtm

[danakj, 2015-03-13 21:59:07]

https://codereview.chromium.org/1004043002/diff/80001/base/test/test_discarda...
File base/test/test_discardable_memory_shmem_allocator.cc (right):

https://codereview.chromium.org/1004043002/diff/80001/base/test/test_discarda...
base/test/test_discardable_memory_shmem_allocator.cc:18: bool Lock() override {
return false; }
lock returning false i found surprising, then things won't use this memory?

[oshima, 2015-03-13 22:03:32]

ash rs lgtm

[reveman, 2015-03-13 22:07:33]

https://codereview.chromium.org/1004043002/diff/80001/base/test/test_discarda...
File base/test/test_discardable_memory_shmem_allocator.cc (right):

https://codereview.chromium.org/1004043002/diff/80001/base/test/test_discarda...
base/test/test_discardable_memory_shmem_allocator.cc:18: bool Lock() override {
return false; }
On 2015/03/13 21:59:07, danakj wrote:
> lock returning false i found surprising, then things won't use this memory?

The memory is locked initially. It will always fail the second time the client
tries to use the memory and have to call Lock(). This is useful for testing
purposes as it makes sure code that is using discardable memory is handling the
case when memory has been purged properly.

[danakj, 2015-03-13 22:32:45]

LGTM

https://codereview.chromium.org/1004043002/diff/80001/base/test/test_discarda...
File base/test/test_discardable_memory_shmem_allocator.cc (right):

https://codereview.chromium.org/1004043002/diff/80001/base/test/test_discarda...
base/test/test_discardable_memory_shmem_allocator.cc:18: bool Lock() override {
return false; }
On 2015/03/13 22:07:33, reveman wrote:
> On 2015/03/13 21:59:07, danakj wrote:
> > lock returning false i found surprising, then things won't use this memory?
> 
> The memory is locked initially. It will always fail the second time the client
> tries to use the memory and have to call Lock(). This is useful for testing
> purposes as it makes sure code that is using discardable memory is handling
the
> case when memory has been purged properly.

OK.

[reveman, 2015-03-13 22:42:24]

reveman@chromium.org changed reviewers:
+ sky@chromium.org, thakis@chromium.org

[reveman, 2015-03-13 22:42:25]

+thakis,sky for chrome/test/

[sky, 2015-03-13 22:54:12]

https://codereview.chromium.org/1004043002/diff/80001/chrome/test/base/chrome...
File chrome/test/base/chrome_unit_test_suite.cc (right):

https://codereview.chromium.org/1004043002/diff/80001/chrome/test/base/chrome...
chrome/test/base/chrome_unit_test_suite.cc:114:
base::DiscardableMemoryShmemAllocator::SetInstance(
Do you need any cleanup code in Shutdown or the like?

[reveman, 2015-03-13 23:08:21]

https://codereview.chromium.org/1004043002/diff/80001/chrome/test/base/chrome...
File chrome/test/base/chrome_unit_test_suite.cc (right):

https://codereview.chromium.org/1004043002/diff/80001/chrome/test/base/chrome...
chrome/test/base/chrome_unit_test_suite.cc:114:
base::DiscardableMemoryShmemAllocator::SetInstance(
On 2015/03/13 22:54:12, sky wrote:
> Do you need any cleanup code in Shutdown or the like?

No, this is a set once at startup kind of thing and the test allocator is
stateless.

[Nico, 2015-03-16 14:21:07]

lgtm

[reveman, 2015-03-16 14:21:50]

The CQ bit was checked by reveman@chromium.org

[reveman, 2015-03-16 14:21:51]

The patchset sent to the CQ was uploaded after l-g-t-m from avi@chromium.org
Link to the patchset: https://codereview.chromium.org/1004043002/#ps80001
(title: "address review feedback")

[commit-bot: I haz the power, 2015-03-16 14:21:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1004043002/80001

[commit-bot: I haz the power, 2015-03-16 16:50:20]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-03-16 16:51:09]

Patchset 5 (id:??) landed as
https://crrev.com/ee5ad4b13cc36b7ee494bc2b04a356d73ad32e3a
Cr-Commit-Position: refs/heads/master@{#320737}

[reveman, 2015-03-16 18:02:46]

avi, fyi re-landing this with the change to
content/test/blink_test_environment.cc, which is needed for webkit_unit_tests

[reveman, 2015-03-16 18:07:11]

The CQ bit was checked by reveman@chromium.org

[reveman, 2015-03-16 18:07:12]

The patchset sent to the CQ was uploaded after l-g-t-m from avi@chromium.org,
oshima@chromium.org, stevenjb@chromium.org, danakj@chromium.org,
thakis@chromium.org
Link to the patchset: https://codereview.chromium.org/1004043002/#ps100001
(title: "set allocator instance for blink tests")

[commit-bot: I haz the power, 2015-03-16 18:07:39]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1004043002/100001

[commit-bot: I haz the power, 2015-03-16 19:06:42]

Committed patchset #6 (id:100001)

[commit-bot: I haz the power, 2015-03-16 19:07:12]

Patchset 6 (id:??) landed as
https://crrev.com/2799f7fb3d3a2406b1608f0b9bc470ff76ad4ea0
Cr-Commit-Position: refs/heads/master@{#320764}