Plugin Power Saver: Implement size-based heuristic for peripheral content.

Plugin Power Saver: Implement size-based heuristic for peripheral content.

Restricts Plugin Power Saver to small content that's cross-origin.

Also handles the temporary whitelist for cross-origin as described by
thestig@ in the design doc.

Design doc: http://goo.gl/iDix3p

BUG=403800
Committed: https://crrev.com/eb25b2a88b6bb848a3ebfe07d3737511ebf8685f
Cr-Commit-Position: refs/heads/master@{#302466}

[tommycli, 2014-10-27 23:35:52]

[email protected] changed reviewers:
+ [email protected], [email protected]

[tommycli, 2014-10-27 23:35:53]

groby, thestig: Let me know what you think

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
File content/renderer/pepper/pepper_plugin_instance_impl.cc (right):

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:2064: bool handled =
false;
I set (handled = false) to make sure every instance of this class processes this
message. This seems a bit hacky...

Alternatively, I could create a 'hub' class to receive and then forward the
message to all these instances, but it would be more SLOC.

[groby-ooo-7-16, 2014-10-28 00:39:12]

Implementation detail feedback inline, but: It bugs me that we round-trip to the
browser process for each peripheral plugin. 

What if each RenderFrame had a copy of the whitelist? This allows us to restrict
round-trips to updates to the whitelist. (With the downside that whitelist
updates need to be broadcast to all frames - but hopefully, there are less
whitelisted items than peripheral ones)

It would also simplify the code - you just check against the whitelist, and you
recheck if the whitelist changes. (And you could just add the mainframe's URL to
each render frame's whitelist)

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:2910:
plugin_content_origin_whitelist_[render_frame_message_source_].insert(
Why is this a map keyed by renderframe? AFAICT, we never look up by frame and
always traverse the whole thing?

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:2916: if
(peripheral_plugin_origins.second.erase(content_origin) > 0) {
Hm. Is that necessary? The message is broadcast anyways, can we just send it to
all renderframes? Or is that cost-prohibitive?

If it is: It feels like the set of plugin origins should be stored per frame,
not on WebContents. We only need WebContents for the document-wide whitelist,
right?

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:2931: if
(whitelisted_plugin_origins.second.find(content_origin) !=
Since you're just checking for set membership: 

if (whitelisted_plugin_origins.count(content_origin)) {
 ...
}

does the same, but is shorter.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
File content/renderer/pepper/pepper_plugin_instance_impl.cc (right):

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:2064: bool handled =
false;
On 2014/10/27 23:35:53, tommycli wrote:
> I set (handled = false) to make sure every instance of this class processes
this
> message. This seems a bit hacky...
> 
> Alternatively, I could create a 'hub' class to receive and then forward the
> message to all these instances, but it would be more SLOC.

You don't need this - as is, the code will _always_ return false, so you can
skip handled completely.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3373:
SetPluginThrottled(false);
Why not just SetPluginThrottled(false)?

[Lei Zhang, 2014-10-28 02:16:23]

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:2910:
plugin_content_origin_whitelist_[render_frame_message_source_].insert(
On 2014/10/28 00:39:12, groby wrote:
> Why is this a map keyed by renderframe? AFAICT, we never look up by frame and
> always traverse the whole thing?

I think it's just being used to limit the number of RenderFrames we send IPCs
to. I guess this all depends on how costly it is to broadcast to all frames. jam
or nasko can probably tell you? If we do end up broadcasting to all frames, we
probably don't need FrameHostMsg_PluginContentMarkedPeripheral.

https://codereview.chromium.org/680193002/diff/20001/content/common/frame_mes...
File content/common/frame_messages.h (right):

https://codereview.chromium.org/680193002/diff/20001/content/common/frame_mes...
content/common/frame_messages.h:655: // A renderer sends this to the brower
process when it wants to temporarily
typo: browser

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
File content/renderer/pepper/pepper_plugin_instance_impl.cc (right):

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:229: // Maximum
dimensions plug-in content may have while still being considered
Mention these match Safari's sizes.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:2062: // We set "handled"
to false here, because we want this broadcast message to
Do you want to coordinate this at the PluginModule level instead? It can
consolidate the results from all the PepperPluginInstances and send one bulk
IPC.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3334: GURL main_frame_url
= main_frame->document().url();
nit: replace |main_frame_url| with |main_frame_origin| ?

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3341: bool
content_is_large = bounds.width < kPeripheralContentMaxWidth ||
Shouldn't the variable be named "content_is_small" ? As such, the logic below is
reversed.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3348: Send(new
FrameHostMsg_PluginContentOriginAllowed(
It's weird that an IsFoo() method has side effects.

[tommycli, 2014-10-28 22:18:08]

thestig / groby: addressed your comments below.

I've revamped the system per Rachel's suggestions of having a broadcasted /
replicated whitelist stored on the renderer side.

The major benefit is that when there is no temporarily whitelisted origins,
there should be zero IPC traffic.

The con is that it's a bit more complicated.

I also pulled out the logic into a PluginPowerSaverHelper class. That's probably
a good thing to centralize the power saver code.

Let me know what you guys think.

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:2916: if
(peripheral_plugin_origins.second.erase(content_origin) > 0) {
On 2014/10/28 00:39:11, groby wrote:
> Hm. Is that necessary? The message is broadcast anyways, can we just send it
to
> all renderframes? Or is that cost-prohibitive?
> 
> If it is: It feels like the set of plugin origins should be stored per frame,
> not on WebContents. We only need WebContents for the document-wide whitelist,
> right?

The WebContents needs to aggregate all the per-frame whitelists to send to all
the RenderFrames.

https://codereview.chromium.org/680193002/diff/20001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:2931: if
(whitelisted_plugin_origins.second.find(content_origin) !=
On 2014/10/28 00:39:11, groby wrote:
> Since you're just checking for set membership: 
> 
> if (whitelisted_plugin_origins.count(content_origin)) {
>  ...
> }
> 
> does the same, but is shorter.

Done.

https://codereview.chromium.org/680193002/diff/20001/content/common/frame_mes...
File content/common/frame_messages.h (right):

https://codereview.chromium.org/680193002/diff/20001/content/common/frame_mes...
content/common/frame_messages.h:655: // A renderer sends this to the brower
process when it wants to temporarily
On 2014/10/28 02:16:23, Lei Zhang wrote:
> typo: browser

Done.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
File content/renderer/pepper/pepper_plugin_instance_impl.cc (right):

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:229: // Maximum
dimensions plug-in content may have while still being considered
On 2014/10/28 02:16:23, Lei Zhang wrote:
> Mention these match Safari's sizes.

Done.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:2062: // We set "handled"
to false here, because we want this broadcast message to
On 2014/10/28 02:16:23, Lei Zhang wrote:
> Do you want to coordinate this at the PluginModule level instead? It can
> consolidate the results from all the PepperPluginInstances and send one bulk
> IPC.

Done. I made a PluginPowerSaverHelper class to consolidate.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3334: GURL main_frame_url
= main_frame->document().url();
On 2014/10/28 02:16:23, Lei Zhang wrote:
> nit: replace |main_frame_url| with |main_frame_origin| ?

Done.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3341: bool
content_is_large = bounds.width < kPeripheralContentMaxWidth ||
On 2014/10/28 02:16:23, Lei Zhang wrote:
> Shouldn't the variable be named "content_is_small" ? As such, the logic below
is
> reversed.

Done.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3348: Send(new
FrameHostMsg_PluginContentOriginAllowed(
On 2014/10/28 02:16:23, Lei Zhang wrote:
> It's weird that an IsFoo() method has side effects.

Done.

https://codereview.chromium.org/680193002/diff/20001/content/renderer/pepper/...
content/renderer/pepper/pepper_plugin_instance_impl.cc:3373:
SetPluginThrottled(false);
On 2014/10/28 00:39:12, groby wrote:
> Why not just SetPluginThrottled(false)?

power_saver_enabled_ needs to be set to false, so it knows to ignore throttle
callbacks that are already in the message loop (from simple timer in our case)

[Lei Zhang, 2014-10-29 09:26:48]

https://codereview.chromium.org/680193002/diff/80001/content/browser/web_cont...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:3357:
ClearPowerSaveBlockers(render_frame_host);
This was fine before. Leave it alone?

https://codereview.chromium.org/680193002/diff/80001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:3361: SendToAllFrames(new
FrameMsg_UpdatePluginContentOriginWhitelist(
Do we have to do this? Imagine a web page with small flash embeds from origin
foo. The webpage also contains an iframe with a large flash embed from origin
foo. The page loads and all the Flash instances play. Then the user interacts
with the JS and somehow deletes the iframe. Now suddenly all the remaining flash
embeds would get throttled due to this. Wouldn't that be a weird user
experience?

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
File content/renderer/pepper/plugin_power_saver_helper.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:20: }
nit: blank line before and // namespace

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:77:
peripheral_plugins_.push_back(PeripheralPlugin(plugin_origin, unthrottle));
ShouldThrottleContent() should be const. You can call this or
origin_whitelist_.insert() separately depending on the return value.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:97: } else {
nit: you don't need the else, but if you make this method const, then you can
just return |content_is_small|.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:99: Send(new
FrameHostMsg_PluginContentOriginAllowed(
You only need to send the message if you actually inserted something new.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
File content/renderer/pepper/plugin_power_saver_helper.h (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.h:11: #include
"base/callback.h"
You can use callback_forward.h

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.h:47: ~PeripheralPlugin();
nit: blank line afterwards

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.h:49: base::Closure
unthrottle;
unthrottle_callback ?

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:84: #include
"content/renderer/pepper/plugin_power_saver_helper.h"
move to line 137

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:550: plugin_power_saver_helper_(NULL),
ifdef

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:819: plugin_power_saver_helper_ = new
PluginPowerSaverHelper(this);
Who owns this / frees it?

[tommycli, 2014-10-29 19:59:23]

thestig: Addressed your comments. I think this handles the same-origin
navigation correctly now.

https://codereview.chromium.org/680193002/diff/80001/content/browser/web_cont...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:3357:
ClearPowerSaveBlockers(render_frame_host);
On 2014/10/29 09:26:47, Lei Zhang wrote:
> This was fine before. Leave it alone?

Done.

https://codereview.chromium.org/680193002/diff/80001/content/browser/web_cont...
content/browser/web_contents/web_contents_impl.cc:3361: SendToAllFrames(new
FrameMsg_UpdatePluginContentOriginWhitelist(
On 2014/10/29 09:26:47, Lei Zhang wrote:
> Do we have to do this? Imagine a web page with small flash embeds from origin
> foo. The webpage also contains an iframe with a large flash embed from origin
> foo. The page loads and all the Flash instances play. Then the user interacts
> with the JS and somehow deletes the iframe. Now suddenly all the remaining
flash
> embeds would get throttled due to this. Wouldn't that be a weird user
> experience?

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
File content/renderer/pepper/plugin_power_saver_helper.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:20: }
On 2014/10/29 09:26:47, Lei Zhang wrote:
> nit: blank line before and // namespace

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:77:
peripheral_plugins_.push_back(PeripheralPlugin(plugin_origin, unthrottle));
On 2014/10/29 09:26:47, Lei Zhang wrote:
> ShouldThrottleContent() should be const. You can call this or
> origin_whitelist_.insert() separately depending on the return value.

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:97: } else {
On 2014/10/29 09:26:47, Lei Zhang wrote:
> nit: you don't need the else, but if you make this method const, then you can
> just return |content_is_small|.

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.cc:99: Send(new
FrameHostMsg_PluginContentOriginAllowed(
On 2014/10/29 09:26:47, Lei Zhang wrote:
> You only need to send the message if you actually inserted something new.

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
File content/renderer/pepper/plugin_power_saver_helper.h (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.h:11: #include
"base/callback.h"
On 2014/10/29 09:26:47, Lei Zhang wrote:
> You can use callback_forward.h

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.h:47: ~PeripheralPlugin();
On 2014/10/29 09:26:47, Lei Zhang wrote:
> nit: blank line afterwards

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/pepper/...
content/renderer/pepper/plugin_power_saver_helper.h:49: base::Closure
unthrottle;
On 2014/10/29 09:26:47, Lei Zhang wrote:
> unthrottle_callback ?

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:84: #include
"content/renderer/pepper/plugin_power_saver_helper.h"
On 2014/10/29 09:26:47, Lei Zhang wrote:
> move to line 137

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:550: plugin_power_saver_helper_(NULL),
On 2014/10/29 09:26:47, Lei Zhang wrote:
> ifdef

Done.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:819: plugin_power_saver_helper_ = new
PluginPowerSaverHelper(this);
On 2014/10/29 09:26:47, Lei Zhang wrote:
> Who owns this / frees it?

As a RenderFrameObserver, it's deleted when the RenderFrame gets deleted. Should
I add a comment to this effect?

[tommycli, 2014-10-29 20:53:10]

[email protected] changed reviewers:
+ [email protected]

[tommycli, 2014-10-29 20:53:11]

creis: Mind taking a look? Esp. at the RenderFrame and WebContents parts.

The tricky part is: Having a temporary whitelist of allowed plugin origins. The
whitelist is per-tab and cleared by top level navigation.

Each frame can whitelist origins, and the list needs to be replicated to
subframes.

[Lei Zhang, 2014-10-29 22:31:23]

Looking good. You need to find a content reviewer though. I would add some tests
to exercise the new code.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:819: plugin_power_saver_helper_ = new
PluginPowerSaverHelper(this);
On 2014/10/29 19:59:23, tommycli wrote:
> On 2014/10/29 09:26:47, Lei Zhang wrote:
> > Who owns this / frees it?
> 
> As a RenderFrameObserver, it's deleted when the RenderFrame gets deleted.
Should
> I add a comment to this effect?

Yes please.

https://codereview.chromium.org/680193002/diff/160001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper.cc (right):

https://codereview.chromium.org/680193002/diff/160001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper.cc:123:
origin_whitelist_.insert(content_origin);
nit: you can just always insert, and check the return value to see if the
insertion succeeded or not. Then you can omit the count() call.

[tommycli, 2014-10-29 22:53:23]

thestig: sounds good. I'll work on a browsertest.

Also: Seems like once you put one #if defined(ENABLE_PLUGINS) in, it spreads
like cancer.

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/680193002/diff/80001/content/renderer/render_...
content/renderer/render_frame_impl.cc:819: plugin_power_saver_helper_ = new
PluginPowerSaverHelper(this);
On 2014/10/29 22:31:22, Lei Zhang wrote:
> On 2014/10/29 19:59:23, tommycli wrote:
> > On 2014/10/29 09:26:47, Lei Zhang wrote:
> > > Who owns this / frees it?
> > 
> > As a RenderFrameObserver, it's deleted when the RenderFrame gets deleted.
> Should
> > I add a comment to this effect?
> 
> Yes please.

Done.

https://codereview.chromium.org/680193002/diff/160001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper.cc (right):

https://codereview.chromium.org/680193002/diff/160001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper.cc:123:
origin_whitelist_.insert(content_origin);
On 2014/10/29 22:31:23, Lei Zhang wrote:
> nit: you can just always insert, and check the return value to see if the
> insertion succeeded or not. Then you can omit the count() call.

Done.

[tommycli, 2014-10-30 21:14:24]

thestig: Added browsertest and spunoff the ENABLE_PLUGINS changes to a separate
CL.

[Lei Zhang, 2014-10-30 21:41:58]

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper_browsertest.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper_browsertest.cc:31: virtual
void SetUp() {
no virtual, override at the end.

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper_browsertest.cc:79:
EXPECT_EQ(1u, sink_->message_count());
ASSERT_EQ

[tommycli, 2014-10-30 21:46:22]

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper_browsertest.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper_browsertest.cc:31: virtual
void SetUp() {
On 2014/10/30 21:41:58, Lei Zhang wrote:
> no virtual, override at the end.

Done.

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper_browsertest.cc:79:
EXPECT_EQ(1u, sink_->message_count());
On 2014/10/30 21:41:58, Lei Zhang wrote:
> ASSERT_EQ

Done.

[Charlie Reis, 2014-10-30 22:13:16]

Nice-- I like the use of tab-level browser state, frame-level renderer state,
and SendToAllFrames.  That fits in well with out-of-process iframes.

A few thoughts below, but nothing major.

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:3364: if
(!plugin_content_origin_whitelist_.empty()) {
This feels like it belongs in a WebContentsObserver, rather than living here.

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
File content/browser/web_contents/web_contents_impl.h (right):

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.h:1003: // Plugin content origins
that have been whitelisted.
It's not clear from the comment what these have been whitelisted for.  Please
elaborate.

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.h:1005: PluginContentOriginSet
plugin_content_origin_whitelist_;
I'm wondering if this belongs on a WebContentsObserver rather than baked into
WebContents itself.  Would that make sense?

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper.cc:88: // is fixed. For now,
case 3 in the comment above doesn't work in
nit: comment above -> class level comment

https://codereview.chromium.org/680193002/diff/350001/content/renderer/render...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/renderer/render...
content/renderer/render_frame_impl.cc:587: // Everything below are
RenderFrameObservers are are automatically deleted
are -> and?

[tommycli, 2014-10-31 00:01:22]

creis: I split it off into a WebContentsObserver. Not sure if I put it in the
right place exactly.

Thanks

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:3364: if
(!plugin_content_origin_whitelist_.empty()) {
On 2014/10/30 22:13:16, Charlie Reis wrote:
> This feels like it belongs in a WebContentsObserver, rather than living here.

Done.

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
File content/browser/web_contents/web_contents_impl.h (right):

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.h:1003: // Plugin content origins
that have been whitelisted.
On 2014/10/30 22:13:16, Charlie Reis wrote:
> It's not clear from the comment what these have been whitelisted for.  Please
> elaborate.

Done.

https://codereview.chromium.org/680193002/diff/350001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.h:1005: PluginContentOriginSet
plugin_content_origin_whitelist_;
On 2014/10/30 22:13:16, Charlie Reis wrote:
> I'm wondering if this belongs on a WebContentsObserver rather than baked into
> WebContents itself.  Would that make sense?

Done.

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper.cc:88: // is fixed. For now,
case 3 in the comment above doesn't work in
On 2014/10/30 22:13:16, Charlie Reis wrote:
> nit: comment above -> class level comment

Done.

https://codereview.chromium.org/680193002/diff/350001/content/renderer/render...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/680193002/diff/350001/content/renderer/render...
content/renderer/render_frame_impl.cc:587: // Everything below are
RenderFrameObservers are are automatically deleted
On 2014/10/30 22:13:16, Charlie Reis wrote:
> are -> and?

Edited the comment. See if that makes more sense.

[Charlie Reis, 2014-10-31 16:33:40]

https://codereview.chromium.org/680193002/diff/430001/content/browser/browser...
File content/browser/browser_plugin/browser_plugin_content_origin_whitelist.h
(right):

https://codereview.chromium.org/680193002/diff/430001/content/browser/browser...
content/browser/browser_plugin/browser_plugin_content_origin_whitelist.h:5:
#ifndef
CONTENT_BROWSER_BROWSER_PLUGIN_BROWSER_PLUGIN_CONTENT_ORIGIN_WHITELIST_H_
BrowserPlugin isn't what you're looking for-- that's the framework for the
<webview> tag which loads a web page within an app.

Just put this in content/browser next to the other plugin_* files.

https://codereview.chromium.org/680193002/diff/430001/content/browser/browser...
content/browser/browser_plugin/browser_plugin_content_origin_whitelist.h:28:
class BrowserPluginContentOriginWhitelist : public WebContentsObserver {
PluginContentOriginWhitelist

https://codereview.chromium.org/680193002/diff/430001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper_browsertest.cc (right):

https://codereview.chromium.org/680193002/diff/430001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper_browsertest.cc:97: // Runs
until the unthrottle closure is run.
Is there an expectation we should be testing here?  How would this test fail?

[tommycli, 2014-10-31 18:32:28]

[email protected] changed reviewers:
+ [email protected]

[tommycli, 2014-10-31 18:32:29]

creis: Thanks. I moved the file.
tsepez: May i have a security review for frame_messages.h?

https://codereview.chromium.org/680193002/diff/430001/content/browser/browser...
File content/browser/browser_plugin/browser_plugin_content_origin_whitelist.h
(right):

https://codereview.chromium.org/680193002/diff/430001/content/browser/browser...
content/browser/browser_plugin/browser_plugin_content_origin_whitelist.h:5:
#ifndef
CONTENT_BROWSER_BROWSER_PLUGIN_BROWSER_PLUGIN_CONTENT_ORIGIN_WHITELIST_H_
On 2014/10/31 16:33:39, Charlie Reis wrote:
> BrowserPlugin isn't what you're looking for-- that's the framework for the
> <webview> tag which loads a web page within an app.
> 
> Just put this in content/browser next to the other plugin_* files.

Done.

https://codereview.chromium.org/680193002/diff/430001/content/browser/browser...
content/browser/browser_plugin/browser_plugin_content_origin_whitelist.h:28:
class BrowserPluginContentOriginWhitelist : public WebContentsObserver {
On 2014/10/31 16:33:39, Charlie Reis wrote:
> PluginContentOriginWhitelist

Done.

https://codereview.chromium.org/680193002/diff/430001/content/renderer/pepper...
File content/renderer/pepper/plugin_power_saver_helper_browsertest.cc (right):

https://codereview.chromium.org/680193002/diff/430001/content/renderer/pepper...
content/renderer/pepper/plugin_power_saver_helper_browsertest.cc:97: // Runs
until the unthrottle closure is run.
On 2014/10/31 16:33:40, Charlie Reis wrote:
> Is there an expectation we should be testing here?  How would this test fail?

The unthrottle closure is the loop.QuitClosure.

If it failed, the loop would never exit, and the test would hang.

[Charlie Reis, 2014-10-31 18:58:31]

Thanks, LGTM.

[Tom Sepez, 2014-11-03 19:02:36]

lgtm

[tommycli, 2014-11-03 19:09:20]

all: thank you!

[tommycli, 2014-11-03 19:09:26]

The CQ bit was checked by [email protected]

[commit-bot: I haz the power, 2014-11-03 19:10:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/680193002/490001

[commit-bot: I haz the power, 2014-11-03 19:45:22]

Committed patchset #27 (id:490001)

[commit-bot: I haz the power, 2014-11-03 19:46:06]

Patchset 27 (id:??) landed as
https://crrev.com/eb25b2a88b6bb848a3ebfe07d3737511ebf8685f
Cr-Commit-Position: refs/heads/master@{#302466}