Re-land: base: Enable browser-wide discardable memory on Linux, CrOS and Windows.

Re-land: base: Enable browser-wide discardable memory on Linux, CrOS and Windows.

This makes SHMEM implementation of discardable memory preferred
over EMULATED implementation. This effectively makes SHMEM the
implementation used by default on Linux, CrOS and Windows.

SHMEM implementation of discardable memory gives the browser
process control over the total amount of discardable memory used
and allows us to enforce a global limit of 512MB across all
renderers.

BUG=429415, 429416
Committed: https://crrev.com/76c0dbc0fa04aa03eb6c337b160fe25affbdce8f
Cr-Commit-Position: refs/heads/master@{#317520}

[reveman, 2014-12-28 09:46:21]

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

[Avi (use Gerrit), 2014-12-28 17:07:28]

lgtm

Good luck.

[reveman, 2014-12-28 20:12:33]

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

[reveman, 2014-12-28 20:12:34]

+danakj for base/

[danakj, 2015-01-06 16:30:35]

LGTM

[reveman, 2015-01-06 21:10:33]

The CQ bit was checked by reveman@chromium.org

[commit-bot: I haz the power, 2015-01-06 21:10:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/793443003/1

[commit-bot: I haz the power, 2015-01-06 21:58:26]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-01-06 21:59:13]

Patchset 1 (id:??) landed as
https://crrev.com/03ea50508953461c007316ef0bf2d5c68d097d4b
Cr-Commit-Position: refs/heads/master@{#310153}

[falken, 2015-01-07 03:22:53]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/812993007/ by falken@chromium.org.

The reason for reverting is: Likely causing many crashes in layout tests on
chromium.webkit's WebKit Win7 (dbg) builder as well as tryserver's
win_blink_rel.

The failures started here:
http://build.chromium.org/p/chromium.webkit/builders/WebKit%20Win7%20%28dbg%2...

Excerpt from log:
15:14:22.485 352 worker/3
virtual/deferred/inspector/tracing/timeline-style-recalc-with-invalidations.html
crashed, (stderr lines):
15:14:22.485 352  
[4300:4244:0106/151422:991308:FATAL:host_discardable_shared_memory_manager.cc(126)]
Check failed: bytes_allocated_ >= size (0 vs. 4194304)
15:14:22.485 352   Backtrace:
15:14:22.485 352   	base::debug::StackTrace::StackTrace [0x008CA731+33]
15:14:22.485 352   	logging::LogMessage::~LogMessage [0x0095B6FF+63]
15:14:22.485 352   	content::HostDiscardableSharedMemoryManager::ProcessRemoved
[0x117B47DB+283]
15:14:22.485 352   	content::RenderMessageFilter::~RenderMessageFilter
[0x10DD88B4+628]
15:14:22.485 352   	content::RenderMessageFilter::`vector deleting destructor'
[0x10DD9DC7+87]
15:14:22.485 352  
	content::BrowserThread::DeleteOnThread<6>::Destruct<content::RenderMessageFilter>
[0x10DC7A95+85]
15:14:22.485 352   	content::RenderMessageFilter::OnDestruct [0x10DDD437+23]
15:14:22.485 352   	content::BrowserMessageFilterTraits::Destruct
[0x102F99B3+19]
15:14:22.485 352  
	base::RefCountedThreadSafe<content::BrowserMessageFilter,content::BrowserMessageFilterTraits>::Release
[0x102FA602+82]
15:14:22.485 352   	scoped_refptr<content::BrowserMessageFilter>::Release
[0x102FA63E+14]
15:14:22.485 352  
	scoped_refptr<content::BrowserMessageFilter>::~scoped_refptr<content::BrowserMessageFilter>
[0x102F8D91+33]
15:14:22.485 352   	content::BrowserMessageFilter::Internal::~Internal
[0x102F8FA3+35]
15:14:22.485 352   	content::BrowserMessageFilter::Internal::`scalar deleting
destructor' [0x102F9536+22]
15:14:22.485 352  
	base::RefCountedThreadSafe<IPC::MessageFilter,base::DefaultRefCountedThreadSafeTraits<IPC::MessageFilter>
>::DeleteInternal [0x05C16FDF+63]
15:14:22.485 352  
	base::DefaultRefCountedThreadSafeTraits<IPC::MessageFilter>::Destruct
[0x05C1704C+12]
15:14:22.485 352  
	base::RefCountedThreadSafe<IPC::MessageFilter,base::DefaultRefCountedThreadSafeTraits<IPC::MessageFilter>
>::Release [0x05C18D92+82]
15:14:22.485 352   	scoped_refptr<IPC::MessageFilter>::Release [0x05C18DEE+14]
15:14:22.485 352  
	scoped_refptr<IPC::MessageFilter>::~scoped_refptr<IPC::MessageFilter>
[0x05C13CD1+33]
15:14:22.485 352   	scoped_refptr<IPC::MessageFilter>::`scalar deleting
destructor' [0x05C15E36+22]
15:14:22.485 352   	std::allocator<scoped_refptr<IPC::MessageFilter>
>::destroy<scoped_refptr<IPC::MessageFilter> > [0x05C11698+24]
15:14:22.485 352  
	std::allocator_traits<std::allocator<scoped_refptr<IPC::MessageFilter> >
>::destroy<scoped_refptr<IPC::MessageFilter> > [0x05C116CF+15]
15:14:22.485 352  
	std::_Wrap_alloc<std::allocator<scoped_refptr<IPC::MessageFilter> >
>::destroy<scoped_refptr<IPC::MessageFilter> > [0x05C1165B+27]
15:14:22.485 352  
	std::_Destroy_range<std::_Wrap_alloc<std::allocator<scoped_refptr<IPC::MessageFilter>
> > > [0x05C103C2+34]
15:14:22.485 352  
	std::_Destroy_range<std::_Wrap_alloc<std::allocator<scoped_refptr<IPC::MessageFilter>
> > > [0x05C10374+52]
15:14:22.485 352  
	std::vector<scoped_refptr<IPC::MessageFilter>,std::allocator<scoped_refptr<IPC::MessageFilter>
> >::_Destroy [0x05C1A867+55]
15:14:22.485 352  
	std::vector<scoped_refptr<IPC::MessageFilter>,std::allocator<scoped_refptr<IPC::MessageFilter>
> >::clear [0x05C1B647+55]
15:14:22.485 352   	IPC::ChannelProxy::Context::OnChannelClosed [0x05C17A0F+207]
.

[reveman, 2015-01-08 23:36:45]

The CQ bit was checked by reveman@chromium.org

[commit-bot: I haz the power, 2015-01-08 23:37:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/793443003/1

[commit-bot: I haz the power, 2015-01-08 23:39:47]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-01-08 23:41:52]

Patchset 1 (id:??) landed as
https://crrev.com/fdfea459b1e5b3ffae10b781a33f884b99a0f2a4
Cr-Commit-Position: refs/heads/master@{#310628}

[danakj, 2015-01-10 00:17:20]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/843153003/ by danakj@chromium.org.

The reason for reverting is: Seems to be causing crashes:
https://crash/browse?q=product.name%3D%27Chrome%27%20AND%20product.version%3D....

[reveman, 2015-01-17 20:55:35]

FYI, giving this another try now that the issue that was causing crashes has
been fixed.

[reveman, 2015-01-17 20:55:41]

The CQ bit was checked by reveman@chromium.org

[commit-bot: I haz the power, 2015-01-17 20:56:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/793443003/1

[commit-bot: I haz the power, 2015-01-17 21:14:23]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-01-17 21:15:07]

Patchset 1 (id:??) landed as
https://crrev.com/07e0fd329293cca145875e156b87afdf383dcdc0
Cr-Commit-Position: refs/heads/master@{#312045}

[reveman, 2015-01-20 17:12:19]

On 2015/01/17 21:15:07, I haz the power (commit-bot) wrote:
> Patchset 1 (id:??) landed as
> https://crrev.com/07e0fd329293cca145875e156b87afdf383dcdc0
> Cr-Commit-Position: refs/heads/master@{#312045}

Reverting this to solve the following issues before re-landing:
crbug.com/450096
crbug.com/450122
crbug.com/449907

[reveman, 2015-01-20 17:12:58]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/808743004/ by reveman@chromium.org.

The reason for reverting is: Need to fix these issues before reland:
crbug.com/450096
crbug.com/450122
crbug.com/449907.

[reveman, 2015-02-19 02:35:26]

The CQ bit was checked by reveman@chromium.org

[commit-bot: I haz the power, 2015-02-19 02:36:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/793443003/1

[commit-bot: I haz the power, 2015-02-19 02:37:24]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-02-19 02:38:01]

Patchset 1 (id:??) landed as
https://crrev.com/2f7a8e80b83aa581aa6de86117a692093d16c217
Cr-Commit-Position: refs/heads/master@{#316977}

[Ken Russell (switch to Gerrit), 2015-02-19 06:19:25]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/917023003/ by kbr@chromium.org.

The reason for reverting is: Caused random crashes on GPU bots with this
assertion failure:
WebglConformance.conformance_ogles_GL_log_log_001_to_008:
Error installing extension 'v8/gc'.
[3060:3996:0218/193241:FATAL:discardable_shared_memory_heap.cc(205)] Check
failed: spans_.find(span->start_ + span->length_ - 1) != spans_.end().
.

[Ken Russell (switch to Gerrit), 2015-02-19 06:42:46]

On 2015/02/19 06:19:25, Ken Russell wrote:
> A revert of this CL (patchset #1 id:1) has been created in
> https://codereview.chromium.org/917023003/ by mailto:kbr@chromium.org.
> 
> The reason for reverting is: Caused random crashes on GPU bots with this
> assertion failure:
> WebglConformance.conformance_ogles_GL_log_log_001_to_008:
> Error installing extension 'v8/gc'.
> [3060:3996:0218/193241:FATAL:discardable_shared_memory_heap.cc(205)] Check
> failed: spans_.find(span->start_ + span->length_ - 1) != spans_.end().
> .

A significant question is why this CL made it through the commit queue in the
first place. There are clearly failed win_chromium_rel_ng and
linux_chromium_rel_ng tryjobs that show the exact failure that later showed up
on the waterfall. Looking at
https://chromium-cq-status.appspot.com/patch-status/793443003/1 it claims that
all tryjobs passed, which isn't the case.

Please work with sergiyb@ to figure out why this CL made it through the CQ. I
think it is a mistake to repeatedly revert and re-land the same CL. It may
confuse the state of the commit queue.

[reveman, 2015-02-21 23:17:14]

On 2015/02/19 at 06:42:46, kbr wrote:
> On 2015/02/19 06:19:25, Ken Russell wrote:
> > A revert of this CL (patchset #1 id:1) has been created in
> > https://codereview.chromium.org/917023003/ by mailto:kbr@chromium.org.
> > 
> > The reason for reverting is: Caused random crashes on GPU bots with this
> > assertion failure:
> > WebglConformance.conformance_ogles_GL_log_log_001_to_008:
> > Error installing extension 'v8/gc'.
> > [3060:3996:0218/193241:FATAL:discardable_shared_memory_heap.cc(205)] Check
> > failed: spans_.find(span->start_ + span->length_ - 1) != spans_.end().
> > .
> 
> A significant question is why this CL made it through the commit queue in the
first place. There are clearly failed win_chromium_rel_ng and
linux_chromium_rel_ng tryjobs that show the exact failure that later showed up
on the waterfall. Looking at
https://chromium-cq-status.appspot.com/patch-status/793443003/1 it claims that
all tryjobs passed, which isn't the case.
> 
> Please work with sergiyb@ to figure out why this CL made it through the CQ. I
think it is a mistake to repeatedly revert and re-land the same CL. It may
confuse the state of the commit queue.

Created crbug.com/460743 for the issue with CQ landing the change even though
try jobs failed.

These try jobs are not failing since https://codereview.chromium.org/947743002
landed so I'm re-landing this change.

[reveman, 2015-02-21 23:17:21]

The CQ bit was checked by reveman@chromium.org

[commit-bot: I haz the power, 2015-02-21 23:17:31]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/793443003/1

[commit-bot: I haz the power, 2015-02-21 23:17:59]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-02-21 23:18:44]

Patchset 1 (id:??) landed as
https://crrev.com/76c0dbc0fa04aa03eb6c337b160fe25affbdce8f
Cr-Commit-Position: refs/heads/master@{#317520}