Browser Plugin: Fixed browser process crash on embedder reload.

Browser Plugin: Fixed browser process crash on embedder reload.

The browser process crashed because a CHECK that verifies that an existing browser_plugin_embedder_ does not yet exist fails.

This happens because CreateGuest should only be handled in WebContentsImpl on the first call to CreateGuest. On all subsequent calls, CreateGuest should be handled in BrowserPluginEmbedderHelper.

This patch makes this change.

I've also cleaned up some dead code.

I've also written a test that verifies that reloading the embedder
doesn't cause the browser plugin to misbehave.


BUG=153561
Test=BrowserPluginHostTest.ReloadEmbedder


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=159854

[Fady Samuel, 2012-10-02 18:27:04]

Hi Istiaque,

Could you please take a quick look before I get Charlie to OWNER review it?

Thanks,

Fady

[Fady Samuel, 2012-10-02 18:39:01]

PTAL. Made it behave the same as before as per our offline conversation.

[lazyboy, 2012-10-02 18:48:33]

lgtm

Thanks for fixing, some comments for the test.

https://codereview.chromium.org/11046002/diff/1001/content/browser/browser_pl...
File content/browser/browser_plugin/browser_plugin_host_browsertest.cc (right):

https://codereview.chromium.org/11046002/diff/1001/content/browser/browser_pl...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:836: // Change
the title of the page to 'ready'.
The word 'ready' is a bit misleading, I was thinking why do we need this to
check if embedder is ready or not... Change it to something different (maybe
'modified') and also note in the comments here that we change the title from
'embedder' to something else so we can track if the next reload completed.

https://codereview.chromium.org/11046002/diff/1001/content/browser/browser_pl...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:843:
ASCIIToUTF16("MarkAsReady();"));
SetTitle(str) is probably better than MarkAsReady()

https://codereview.chromium.org/11046002/diff/1001/content/browser/browser_pl...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:871: 
nit: remove whitespaces.

[Fady Samuel, 2012-10-02 19:39:46]

Fixed nits. +creis for OWNER review. Thanks.

https://codereview.chromium.org/11046002/diff/1001/content/browser/browser_pl...
File content/browser/browser_plugin/browser_plugin_host_browsertest.cc (right):

https://codereview.chromium.org/11046002/diff/1001/content/browser/browser_pl...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:836: // Change
the title of the page to 'ready'.
On 2012/10/02 18:48:33, lazyboy wrote:
> The word 'ready' is a bit misleading, I was thinking why do we need this to
> check if embedder is ready or not... Change it to something different (maybe
> 'modified') and also note in the comments here that we change the title from
> 'embedder' to something else so we can track if the next reload completed.

Done.

https://codereview.chromium.org/11046002/diff/1001/content/browser/browser_pl...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:843:
ASCIIToUTF16("MarkAsReady();"));
On 2012/10/02 18:48:33, lazyboy wrote:
> SetTitle(str) is probably better than MarkAsReady()

Done.

[Fady Samuel, 2012-10-02 20:00:06]

http://codereview.chromium.org/11046002/diff/1001/content/browser/browser_plu...
File content/browser/browser_plugin/browser_plugin_host_browsertest.cc (right):

http://codereview.chromium.org/11046002/diff/1001/content/browser/browser_plu...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:871: 
On 2012/10/02 18:48:33, lazyboy wrote:
> nit: remove whitespaces.

Done.

[Charlie Reis, 2012-10-02 23:22:18]

LGTM, with a few comments.

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
File content/browser/browser_plugin/browser_plugin_embedder_helper.cc (right):

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
content/browser/browser_plugin/browser_plugin_embedder_helper.cc:106:
embedder_->CreateGuest(render_view_host(),
Can you add a comment saying this is for all CreateGuest calls after the first
one, and that the first one is handled in
WebContentsImpl::OnBrowserPluginCreateGuest?

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
File content/browser/browser_plugin/browser_plugin_host_browsertest.cc (right):

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:852: // Then 
navigate the guest to verify that the browser process does not crash.
nit: extra space after "Then"

http://codereview.chromium.org/11046002/diff/5004/content/browser/web_content...
File content/browser/web_contents/web_contents_impl.cc (right):

http://codereview.chromium.org/11046002/diff/5004/content/browser/web_content...
content/browser/web_contents/web_contents_impl.cc:2380: // BrowserPluginGuest.
Similarly, we should update this to say that this is only for the first
CreateGuest call, and subsequent ones are handled in
BrowserPluginEmbedderHelper.

How does that happen, anyway?  Is it just that BrowserPluginEmbedderHelper
consumes the IPC message so it never makes it to WebContentsImpl after that? 
That's worth being a little more explicit about in this comment, since it's very
implicit in the code.

[Fady Samuel, 2012-10-03 01:48:16]

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
File content/browser/browser_plugin/browser_plugin_embedder_helper.cc (right):

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
content/browser/browser_plugin/browser_plugin_embedder_helper.cc:106:
embedder_->CreateGuest(render_view_host(),
On 2012/10/02 23:22:18, creis wrote:
> Can you add a comment saying this is for all CreateGuest calls after the first
> one, and that the first one is handled in
> WebContentsImpl::OnBrowserPluginCreateGuest?

Done.

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
File content/browser/browser_plugin/browser_plugin_host_browsertest.cc (right):

http://codereview.chromium.org/11046002/diff/5004/content/browser/browser_plu...
content/browser/browser_plugin/browser_plugin_host_browsertest.cc:852: // Then 
navigate the guest to verify that the browser process does not crash.
On 2012/10/02 23:22:18, creis wrote:
> nit: extra space after "Then"

Done.

http://codereview.chromium.org/11046002/diff/5004/content/browser/web_content...
File content/browser/web_contents/web_contents_impl.cc (right):

http://codereview.chromium.org/11046002/diff/5004/content/browser/web_content...
content/browser/web_contents/web_contents_impl.cc:2380: // BrowserPluginGuest.
On 2012/10/02 23:22:18, creis wrote:
> Similarly, we should update this to say that this is only for the first
> CreateGuest call, and subsequent ones are handled in
> BrowserPluginEmbedderHelper.
> 
> How does that happen, anyway?  Is it just that BrowserPluginEmbedderHelper
> consumes the IPC message so it never makes it to WebContentsImpl after that? 
> That's worth being a little more explicit about in this comment, since it's
very
> implicit in the code.

Done.

[commit-bot: I haz the power, 2012-10-03 01:48:29]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11046002/11001

[commit-bot: I haz the power, 2012-10-03 06:04:10]

Change committed as 159854