MemoryPressureListener: fixes ref-counting.

MemoryPressureListener: fixes ref-counting.

Following crrev.com/205796, the internal observer list is
ref-counted, the lazy instance must keep a ref to it.

BUG=246125
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=206968

[bulach, 2013-06-17 09:27:26]

sorry, there was a bug in my previous patch.
ptal and also, please let me know if you'd prefer the alternative suggested
below. thanks!

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
File base/memory/memory_pressure_listener.cc (right):

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
base/memory/memory_pressure_listener.cc:21: return new (instance)
scoped_refptr<ObserverListThreadSafe<
alternatively, the lazy instance could keep the ObserverListThreadSafe directly
(instead of the scoped_refptr) and this traits would just AddRef to it.. wdyt?

[joth, 2013-06-17 18:00:38]

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
File base/memory/memory_pressure_listener.cc (right):

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
base/memory/memory_pressure_listener.cc:14: struct LeakyLazyObserverListTraits {
if you derive this from LeakyLazyInstanceTraits<..> then you only have to define
New(), not the other things (consts, Delete)

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
base/memory/memory_pressure_listener.cc:21: return new (instance)
scoped_refptr<ObserverListThreadSafe<
On 2013/06/17 09:27:26, bulach wrote:
> alternatively, the lazy instance could keep the ObserverListThreadSafe
directly
> (instead of the scoped_refptr) and this traits would just AddRef to it.. wdyt?

Yeah that seems simpler.
I was about to suggest the same.

[jar (doing other things), 2013-06-17 18:03:12]

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
File base/memory/memory_pressure_listener.cc (right):

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
base/memory/memory_pressure_listener.cc:21: return new (instance)
scoped_refptr<ObserverListThreadSafe<
+1

On 2013/06/17 18:00:38, joth wrote:
> On 2013/06/17 09:27:26, bulach wrote:
> > alternatively, the lazy instance could keep the ObserverListThreadSafe
> directly
> > (instead of the scoped_refptr) and this traits would just AddRef to it..
wdyt?
> 
> Yeah that seems simpler.
> I was about to suggest the same.

[bulach, 2013-06-17 18:11:36]

thanks both! comments addressed, another look please?

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
File base/memory/memory_pressure_listener.cc (right):

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
base/memory/memory_pressure_listener.cc:14: struct LeakyLazyObserverListTraits {
On 2013/06/17 18:00:38, joth wrote:
> if you derive this from LeakyLazyInstanceTraits<..> then you only have to
define
> New(), not the other things (consts, Delete)

Done.

https://codereview.chromium.org/17242002/diff/1/base/memory/memory_pressure_l...
base/memory/memory_pressure_listener.cc:21: return new (instance)
scoped_refptr<ObserverListThreadSafe<
On 2013/06/17 18:00:38, joth wrote:
> On 2013/06/17 09:27:26, bulach wrote:
> > alternatively, the lazy instance could keep the ObserverListThreadSafe
> directly
> > (instead of the scoped_refptr) and this traits would just AddRef to it..
wdyt?
> 
> Yeah that seems simpler.
> I was about to suggest the same.

Done.

[joth, 2013-06-17 18:20:08]

lgtm

[jar (doing other things), 2013-06-17 18:26:15]

lgtm

[commit-bot: I haz the power, 2013-06-18 08:31:57]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bulach@chromium.org/17242002/9001

[commit-bot: I haz the power, 2013-06-18 13:31:57]

Change committed as 206968