Don't abort the MimeHandlerStreamManager stream if we receive a RenderProcessHostChanged notificati…

Don't abort the MimeHandlerStreamManager stream if we receive a RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class, this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 650985

BUG=650985
Committed: https://crrev.com/b8cfa3b172b2741635247a8541e3251688aaa04f
Cr-Commit-Position: refs/heads/master@{#425498}

[ananta, 2016-10-14 00:13:30]

Description was changed from

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this caues
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 654193

BUG=654193
==========

to

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 654193

BUG=654193
==========

[ananta, 2016-10-14 00:13:43]

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

[ananta, 2016-10-14 00:13:57]

The CQ bit was checked by [email protected] to run a CQ dry run

[commit-bot: I haz the power, 2016-10-14 00:14:27]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-14 00:59:55]

The CQ bit was unchecked by [email protected]

[commit-bot: I haz the power, 2016-10-14 00:59:56]

Dry run: This issue passed the CQ dry run.

[lazyboy, 2016-10-14 01:46:28]

https://codereview.chromium.org/2418963002/diff/1/extensions/browser/guest_vi...
File
extensions/browser/guest_view/mime_handler_view/mime_handler_stream_manager.cc
(right):

https://codereview.chromium.org/2418963002/diff/1/extensions/browser/guest_vi...
extensions/browser/guest_view/mime_handler_view/mime_handler_stream_manager.cc:235:
// If the old_host is null, then it means that a subframe is being created.
optional nit: "If |old_host| is null...

https://codereview.chromium.org/2418963002/diff/1/extensions/browser/guest_vi...
extensions/browser/guest_view/mime_handler_view/mime_handler_stream_manager.cc:238:
return;
In this case, |new_host| should correspond to render_frame_id_ and
render_process_id_? Can we DCHECK that those match too? But it seems that if
somehow an entirely different frame creation in Embedder WebContents happens,
that wouldn't work.

I think if old_host is not null, then we should check old_host->GetRoutingId()
equals render_frame_id_ before updating render_frame_id_ from new_host. Same
with render_process_id_. WDYT?

[ananta, 2016-10-14 19:54:47]

I also added a null FrameTreeNode check in WebContents::FromFrameTreeNodeId
function as I was hitting this on the pdfobject.com/static.html with PlzNavigate
enabled. From the initial looks it appears to be coming from FrameTreeNodes
getting destroyed in the prerender code path.

https://codereview.chromium.org/2418963002/diff/1/extensions/browser/guest_vi...
File
extensions/browser/guest_view/mime_handler_view/mime_handler_stream_manager.cc
(right):

https://codereview.chromium.org/2418963002/diff/1/extensions/browser/guest_vi...
extensions/browser/guest_view/mime_handler_view/mime_handler_stream_manager.cc:238:
return;
On 2016/10/14 01:46:28, lazyboy wrote:
> In this case, |new_host| should correspond to render_frame_id_ and
> render_process_id_? Can we DCHECK that those match too? But it seems that if
> somehow an entirely different frame creation in Embedder WebContents happens,
> that wouldn't work.
> 
> I think if old_host is not null, then we should check old_host->GetRoutingId()
> equals render_frame_id_ before updating render_frame_id_ from new_host. Same
> with render_process_id_. WDYT?

Thanks. Done

[ananta, 2016-10-14 19:55:04]

The CQ bit was checked by [email protected] to run a CQ dry run

[commit-bot: I haz the power, 2016-10-14 19:55:36]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[ananta, 2016-10-14 20:15:57]

Description was changed from

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 654193

BUG=654193
==========

to

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 654193

BUG=650985
==========

[lazyboy, 2016-10-14 20:22:26]

lgtm.

[commit-bot: I haz the power, 2016-10-14 20:54:56]

The CQ bit was unchecked by [email protected]

[commit-bot: I haz the power, 2016-10-14 20:54:57]

Dry run: This issue passed the CQ dry run.

[ananta, 2016-10-14 20:58:28]

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

[ananta, 2016-10-14 20:58:28]

+jam for content

[jam, 2016-10-14 22:34:20]

On 2016/10/14 20:58:28, ananta wrote:
> +jam for content

lgtm


the description has two different bug numbers, and one seems unrelated?

[ananta, 2016-10-14 22:36:39]

Description was changed from

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 654193

BUG=650985
==========

to

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 650985

BUG=650985
==========

[ananta, 2016-10-14 22:36:45]

The CQ bit was checked by [email protected]

[commit-bot: I haz the power, 2016-10-14 22:37:20]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-14 22:43:42]

Description was changed from

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 650985

BUG=650985
==========

to

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 650985

BUG=650985
==========

[commit-bot: I haz the power, 2016-10-14 22:43:43]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-10-14 22:45:38]

Description was changed from

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 650985

BUG=650985
==========

to

==========
Don't abort the MimeHandlerStreamManager stream if we receive a
RenderProcessHostChanged notification for a null old host.

This indicates that a subframe is being created. If we ignore  the old_host here
and set
the new host pointer in the MimeHandlerStreamManager::EmbedderObserver class,
this causes
the stream to be aborted prematurely leading to embedded PDFs not loading.

This should fix bug 650985

BUG=650985

Committed: https://crrev.com/b8cfa3b172b2741635247a8541e3251688aaa04f
Cr-Commit-Position: refs/heads/master@{#425498}
==========

[commit-bot: I haz the power, 2016-10-14 22:45:39]

Patchset 2 (id:??) landed as
https://crrev.com/b8cfa3b172b2741635247a8541e3251688aaa04f
Cr-Commit-Position: refs/heads/master@{#425498}