Browser Plugin: Remove MapInstance from BrowserPluginHost + More cross-process navigation work

Browser Plugin: Remove MapInstance from BrowserPluginHost + More cross-process navigation work

This patch does two things:

It addresses the issues mentioned regarding BrowserPluginHost::OnMapInstance.
Instead this work is done over the GuestToEmbedderChannel.

This also introduces some plumbing on ViewMsg_Navigate that allows for navigating
back to swapped out RenderViews.

BUG=128976
TEST=manually


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=138861

[Fady Samuel, 2012-05-23 16:08:58]

creis@: cross-process stuff + MapInstance removal

jam@: OWNER

[Charlie Reis, 2012-05-23 16:57:53]

Cross-process stuff LGTM, pending the questions below.

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/browser/web...
File content/browser/web_contents/web_contents_impl.cc (right):

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/browser/web...
content/browser/web_contents/web_contents_impl.cc:1171: embedder_process_id);
I remember this was discussed in an earlier CL but I didn't see the resolution. 
Did we decide it's a safe technique?

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/renderer/re...
File content/renderer/render_view_impl.cc (right):

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/renderer/re...
content/renderer/render_view_impl.cc:781:
RenderViewImpl::GetGuestToEmbedderChannel() const {
Why rename these to CamelCase if they're still one-liners and not defined in the
public interface?

[Fady Samuel, 2012-05-23 17:32:14]

http://codereview.chromium.org/10412056/diff/5001/content/browser/web_content...
File content/browser/web_contents/web_contents_impl.cc (right):

http://codereview.chromium.org/10412056/diff/5001/content/browser/web_content...
content/browser/web_contents/web_contents_impl.cc:1171: embedder_process_id);
On 2012/05/23 16:57:54, creis wrote:
> I remember this was discussed in an earlier CL but I didn't see the
resolution. 
> Did we decide it's a safe technique?

Yes, please see here: https://chromiumcodereview.appspot.com/10378089



Justin Schuh commented, and I pointed out that I'm doing more or less what he
suggested, he agreed.

http://codereview.chromium.org/10412056/diff/5001/content/renderer/render_vie...
File content/renderer/render_view_impl.cc (right):

http://codereview.chromium.org/10412056/diff/5001/content/renderer/render_vie...
content/renderer/render_view_impl.cc:781:
RenderViewImpl::GetGuestToEmbedderChannel() const {
On 2012/05/23 16:57:54, creis wrote:
> Why rename these to CamelCase if they're still one-liners and not defined in
the
> public interface?

I believe I moved this because I needed SetGuestToEmbedderChannel, and I
realized that there's a lot going on there because guest_to_embedder_channel_ is
a refptr. I'm not sure if this is considered trivial or not?

[Charlie Reis, 2012-05-23 17:52:54]

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/browser/web...
File content/browser/web_contents/web_contents_impl.cc (right):

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/browser/web...
content/browser/web_contents/web_contents_impl.cc:1171: embedder_process_id);
On 2012/05/23 17:32:15, Fady Samuel wrote:
> On 2012/05/23 16:57:54, creis wrote:
> > I remember this was discussed in an earlier CL but I didn't see the
> resolution. 
> > Did we decide it's a safe technique?
> 
> Yes, please see here: https://chromiumcodereview.appspot.com/10378089
> 
> 
> 
> Justin Schuh commented, and I pointed out that I'm doing more or less what he
> suggested, he agreed.

Great.  Perhaps we can abstract this out so it's not duplicated between here and
render_view_host_impl.cc?  I could see it as a static GetChannelName helper on
BrowserPluginHost, for example.

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/renderer/re...
File content/renderer/render_view_impl.cc (right):

https://chromiumcodereview.appspot.com/10412056/diff/5001/content/renderer/re...
content/renderer/render_view_impl.cc:781:
RenderViewImpl::GetGuestToEmbedderChannel() const {
On 2012/05/23 17:32:15, Fady Samuel wrote:
> On 2012/05/23 16:57:54, creis wrote:
> > Why rename these to CamelCase if they're still one-liners and not defined in
> the
> > public interface?
> 
> I believe I moved this because I needed SetGuestToEmbedderChannel, and I
> realized that there's a lot going on there because guest_to_embedder_channel_
is
> a refptr. I'm not sure if this is considered trivial or not?

Still seems trivial to me.

[Fady Samuel, 2012-05-23 18:20:38]

On 2012/05/23 17:52:54, creis wrote:
>
https://chromiumcodereview.appspot.com/10412056/diff/5001/content/browser/web...
> File content/browser/web_contents/web_contents_impl.cc (right):
> 
>
https://chromiumcodereview.appspot.com/10412056/diff/5001/content/browser/web...
> content/browser/web_contents/web_contents_impl.cc:1171: embedder_process_id);
> On 2012/05/23 17:32:15, Fady Samuel wrote:
> > On 2012/05/23 16:57:54, creis wrote:
> > > I remember this was discussed in an earlier CL but I didn't see the
> > resolution. 
> > > Did we decide it's a safe technique?
> > 
> > Yes, please see here: https://chromiumcodereview.appspot.com/10378089
> > 
> > 
> > 
> > Justin Schuh commented, and I pointed out that I'm doing more or less what
he
> > suggested, he agreed.
> 
> Great.  Perhaps we can abstract this out so it's not duplicated between here
and
> render_view_host_impl.cc?  I could see it as a static GetChannelName helper on
> BrowserPluginHost, for example.
> 
>
https://chromiumcodereview.appspot.com/10412056/diff/5001/content/renderer/re...
> File content/renderer/render_view_impl.cc (right):
> 
>
https://chromiumcodereview.appspot.com/10412056/diff/5001/content/renderer/re...
> content/renderer/render_view_impl.cc:781:
> RenderViewImpl::GetGuestToEmbedderChannel() const {
> On 2012/05/23 17:32:15, Fady Samuel wrote:
> > On 2012/05/23 16:57:54, creis wrote:
> > > Why rename these to CamelCase if they're still one-liners and not defined
in
> > the
> > > public interface?
> > 
> > I believe I moved this because I needed SetGuestToEmbedderChannel, and I
> > realized that there's a lot going on there because
guest_to_embedder_channel_
> is
> > a refptr. I'm not sure if this is considered trivial or not?
> 
> Still seems trivial to me.

Restored unix_hacker style. definition in cc file to avoid cyclic includes.

[Fady Samuel, 2012-05-23 18:56:41]

-jam@ for now, until final look by creis@, then will readd.

[Charlie Reis, 2012-05-23 19:07:58]

Thanks, LGTM.

[Fady Samuel, 2012-05-23 19:30:54]

+jam again, ready for final OWNER approval. thanks.

[jam, 2012-05-23 22:11:10]

lgtm with nits

http://codereview.chromium.org/10412056/diff/16003/content/renderer/browser_p...
File content/renderer/browser_plugin/browser_plugin.cc (right):

http://codereview.chromium.org/10412056/diff/16003/content/renderer/browser_p...
content/renderer/browser_plugin/browser_plugin.cc:141: DCHECK(dispatcher);
nit: skip this dcheck, it's unnecessary

http://codereview.chromium.org/10412056/diff/16003/content/renderer/render_vi...
File content/renderer/render_view_impl.cc (right):

http://codereview.chromium.org/10412056/diff/16003/content/renderer/render_vi...
content/renderer/render_view_impl.cc:781:
RenderViewImpl::guest_to_embedder_channel() const {
nit: if the function definition is in the cc file, then you can't use
unix_hacker style. so either move this to the header or rename this using
CamelCase. same below.

[Fady Samuel, 2012-05-23 22:23:42]

Fixed nits. Committing. Thanks.

https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/b...
File content/renderer/browser_plugin/browser_plugin.cc (right):

https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/b...
content/renderer/browser_plugin/browser_plugin.cc:141: DCHECK(dispatcher);
On 2012/05/23 22:11:10, John Abd-El-Malek wrote:
> nit: skip this dcheck, it's unnecessary

Done.

https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/r...
File content/renderer/render_view_impl.cc (right):

https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/r...
content/renderer/render_view_impl.cc:781:
RenderViewImpl::guest_to_embedder_channel() const {
On 2012/05/23 22:11:10, John Abd-El-Malek wrote:
> nit: if the function definition is in the cc file, then you can't use
> unix_hacker style. so either move this to the header or rename this using
> CamelCase. same below.

I left it in the CC file to avoid including guest_to_embedder_channel.h in
render_view_impl.h. Made it CamelCase again. Thanks.

[commit-bot: I haz the power, 2012-05-23 22:24:14]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10412056/19001

[jam, 2012-05-23 22:30:20]

https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/r...
File content/renderer/render_view_impl.cc (right):

https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/r...
content/renderer/render_view_impl.cc:781:
RenderViewImpl::guest_to_embedder_channel() const {
On 2012/05/23 22:23:43, Fady Samuel wrote:
> On 2012/05/23 22:11:10, John Abd-El-Malek wrote:
> > nit: if the function definition is in the cc file, then you can't use
> > unix_hacker style. so either move this to the header or rename this using
> > CamelCase. same below.
> 
> I left it in the CC file to avoid including guest_to_embedder_channel.h in
> render_view_impl.h. Made it CamelCase again. Thanks.

I'm not following. Why do you need to include guest_to_embedder_channel.h  to
inline this?

[Fady Samuel, 2012-05-23 22:38:06]

On 2012/05/23 22:30:20, John Abd-El-Malek wrote:
>
https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/r...
> File content/renderer/render_view_impl.cc (right):
> 
>
https://chromiumcodereview.appspot.com/10412056/diff/16003/content/renderer/r...
> content/renderer/render_view_impl.cc:781:
> RenderViewImpl::guest_to_embedder_channel() const {
> On 2012/05/23 22:23:43, Fady Samuel wrote:
> > On 2012/05/23 22:11:10, John Abd-El-Malek wrote:
> > > nit: if the function definition is in the cc file, then you can't use
> > > unix_hacker style. so either move this to the header or rename this using
> > > CamelCase. same below.
> > 
> > I left it in the CC file to avoid including guest_to_embedder_channel.h in
> > render_view_impl.h. Made it CamelCase again. Thanks.
> 
> I'm not following. Why do you need to include guest_to_embedder_channel.h  to
> inline this?

IIRC, SetGuestToEmbedderChannel assigns a refptr which wants to reach into
GuestToEmbedderChannel to inc/dec the refcount and so it needs
guest_to_embedder_channel.h for its declaration. I moved
GetGuestToEmbedderChannel into render_view_impl as well for consistency with the
setter.

[commit-bot: I haz the power, 2012-05-23 22:49:25]

Try job failure for 10412056-19001 (retry) on mac_rel for step "compile"
(clobber build).
It's a second try, previously, step "compile" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...

[commit-bot: I haz the power, 2012-05-24 01:26:20]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10412056/25001

[commit-bot: I haz the power, 2012-05-24 01:49:37]

Try job failure for 10412056-25001 (retry) on mac_rel for step "compile"
(clobber build).
It's a second try, previously, step "compile" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...

[Fady Samuel, 2012-05-24 02:00:33]

+sky chrome/browser/visitedlink OWNER

[commit-bot: I haz the power, 2012-05-24 02:00:50]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10412056/19003

[commit-bot: I haz the power, 2012-05-24 02:01:07]

Presubmit check for 10412056-19003 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Missing LGTM from an OWNER for files in these directories:
    chrome/browser/visitedlink

Presubmit checks took 3.4s to calculate.

[sky, 2012-05-24 15:52:17]

LGTM

[commit-bot: I haz the power, 2012-05-24 15:52:52]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10412056/19003

[commit-bot: I haz the power, 2012-05-24 16:21:55]

Try job failure for 10412056-19003 (retry) on linux_chromeos for step "compile"
(clobber build).
It's a second try, previously, step "compile" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_chro...

[commit-bot: I haz the power, 2012-05-24 16:48:57]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10412056/25003

[commit-bot: I haz the power, 2012-05-24 19:31:28]

Change committed as 138861