Implement an IsAllowed function in the pepper PPB_Broker_Trusted API

Implement an IsAllowed function in the pepper PPB_Broker_Trusted API

Flash sometimes needs to synchronously know if it can launch the broker, otherwise it will try to launch the broker when it shouldn't, and end up popping an infobar. This adds an IsAllowed function to synchronously test whether the broker is allowed to launch without popping the infobar.

Note that the document URL of the plugin instance is needed in order to check the broker permissions in the browser process. This is only available in the renderer process. In order to avoid an extra hop to the renderer process just to get this URL, it is sent to the browser (with the render view ID) upon initialization of the instance when the instance is registered with the browser process.

BUG=163248
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=172104

[raymes, 2012-12-04 17:57:44]

There are 2 small plumbing related changes to allow the Document/Plugin URLs to
be accessed in the browser process.
1) A new struct PepperRendererInstanceData is introduced which groups together
data sent from the renderer to the browser when the instance is initialized.
2) The plugin url and webkit container which are associated with the instance
are passed into the PluginInstance on construction instead of when the
PluginInstance::Initialize() function is called so that they are available when
the instance is registered in the browser.

jschuh: (broker-related code)
pepper_broker_host.*
ppapi_browsertest.cc
test_broker.*
ppb_broker_trusted.idl
ppapi_messages.h

brettw: (plumbing for document/plugin url)
content/*
webkit/*

yzshen:
The rest.

[jschuh, 2012-12-05 00:14:10]

lgtm in terms of exposing the correct functionality to the broker.

[brettw, 2012-12-05 23:00:26]

lgtm

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
File chrome/browser/renderer_host/pepper/pepper_broker_host.cc (right):

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
chrome/browser/renderer_host/pepper/pepper_broker_host.cc:82: if
(!document_url_.is_valid())
Can you add a DCHECK on the thread you should be on here? In addition to being a
sanity check, it makes it more clear when reading it.

https://codereview.chromium.org/11316316/diff/7004/content/common/pepper_rend...
File content/common/pepper_renderer_instance_data.h (right):

https://codereview.chromium.org/11316316/diff/7004/content/common/pepper_rend...
content/common/pepper_renderer_instance_data.h:18:
PepperRendererInstanceData(int render_process,
Do we need a destructor here since there are non-trivial destructors (GURL)? I
think clang sometimes complains about this kind of thing. Probably a good idea
anyway to prevent inlining.

https://codereview.chromium.org/11316316/diff/7004/ppapi/proxy/ppapi_messages.h
File ppapi/proxy/ppapi_messages.h (right):

https://codereview.chromium.org/11316316/diff/7004/ppapi/proxy/ppapi_messages...
ppapi/proxy/ppapi_messages.h:1438:
IPC_MESSAGE_CONTROL0(PpapiHostMsg_Broker_IsAllowed)
Can you document this here? I.e. that it's a call and the reply's error value
contains the result of the query. I'd like to start using the format for the
Talk interface below (big section for interface name, then separate comments for
each message).

[yzshen1, 2012-12-06 19:59:46]

mostly nits. Thanks!

https://chromiumcodereview.appspot.com/11316316/diff/7004/chrome/browser/rend...
File chrome/browser/renderer_host/pepper/pepper_broker_host.cc (right):

https://chromiumcodereview.appspot.com/11316316/diff/7004/chrome/browser/rend...
chrome/browser/renderer_host/pepper/pepper_broker_host.cc:46:
~BrokerMessageFilter();
add 'virtual' please.
(I understand it doesn't have any semantic difference, but it is more clear this
way.)

https://chromiumcodereview.appspot.com/11316316/diff/7004/chrome/browser/rend...
chrome/browser/renderer_host/pepper/pepper_broker_host.cc:72:
ppapi::host::HostMessageContext* context) {
wrong indent.

https://chromiumcodereview.appspot.com/11316316/diff/7004/chrome/browser/rend...
File chrome/browser/renderer_host/pepper/pepper_broker_host.h (right):

https://chromiumcodereview.appspot.com/11316316/diff/7004/chrome/browser/rend...
chrome/browser/renderer_host/pepper/pepper_broker_host.h:9: #include
"ppapi/proxy/resource_message_params.h"
nit: you don't need it.

https://chromiumcodereview.appspot.com/11316316/diff/7004/content/browser/ren...
File content/browser/renderer_host/render_message_filter.h (right):

https://chromiumcodereview.appspot.com/11316316/diff/7004/content/browser/ren...
content/browser/renderer_host/render_message_filter.h:173:
PepperRendererInstanceData instance_data,
why it is not a const ref? (You could move the #include to the .cc as well.)

https://chromiumcodereview.appspot.com/11316316/diff/7004/content/common/view...
File content/common/view_messages.h (right):

https://chromiumcodereview.appspot.com/11316316/diff/7004/content/common/view...
content/common/view_messages.h:1969: // -The instance id of the instance being
created.
id -> ID to be consistent.

https://chromiumcodereview.appspot.com/11316316/diff/7004/content/public/brow...
File content/public/browser/browser_ppapi_host.h (right):

https://chromiumcodereview.appspot.com/11316316/diff/7004/content/public/brow...
content/public/browser/browser_ppapi_host.h:79: virtual GURL
GetPluginURLForInstance(PP_Instance instance) = 0;
I notice that the plugin url is not used. Is it useful for a future CL?

https://chromiumcodereview.appspot.com/11316316/diff/7004/ppapi/proxy/broker_...
File ppapi/proxy/broker_resource.cc (right):

https://chromiumcodereview.appspot.com/11316316/diff/7004/ppapi/proxy/broker_...
ppapi/proxy/broker_resource.cc:9: #include "ppapi/thunk/enter.h"
I don't think you need these two includes?

https://chromiumcodereview.appspot.com/11316316/diff/7004/ppapi/tests/test_br...
File ppapi/tests/test_broker.cc (right):

https://chromiumcodereview.appspot.com/11316316/diff/7004/ppapi/tests/test_br...
ppapi/tests/test_broker.cc:323: PP_Resource broker =
broker_interface_->CreateTrusted(
Is this leaked?

https://chromiumcodereview.appspot.com/11316316/diff/7004/ppapi/tests/test_br...
ppapi/tests/test_broker.cc:332: PP_Resource broker =
broker_interface_->CreateTrusted(
Is this leaked?

[raymes, 2012-12-07 21:32:19]

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
File chrome/browser/renderer_host/pepper/pepper_broker_host.cc (right):

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
chrome/browser/renderer_host/pepper/pepper_broker_host.cc:46:
~BrokerMessageFilter();
On 2012/12/06 19:59:46, yzshen1 wrote:
> add 'virtual' please.
> (I understand it doesn't have any semantic difference, but it is more clear
this
> way.)

Done.

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
chrome/browser/renderer_host/pepper/pepper_broker_host.cc:72:
ppapi::host::HostMessageContext* context) {
On 2012/12/06 19:59:46, yzshen1 wrote:
> wrong indent.

Done.

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
chrome/browser/renderer_host/pepper/pepper_broker_host.cc:82: if
(!document_url_.is_valid())
On 2012/12/05 23:00:26, brettw wrote:
> Can you add a DCHECK on the thread you should be on here? In addition to being
a
> sanity check, it makes it more clear when reading it.

Done.

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
File chrome/browser/renderer_host/pepper/pepper_broker_host.h (right):

https://codereview.chromium.org/11316316/diff/7004/chrome/browser/renderer_ho...
chrome/browser/renderer_host/pepper/pepper_broker_host.h:9: #include
"ppapi/proxy/resource_message_params.h"
On 2012/12/06 19:59:46, yzshen1 wrote:
> nit: you don't need it.

Done.

https://codereview.chromium.org/11316316/diff/7004/content/browser/renderer_h...
File content/browser/renderer_host/render_message_filter.h (right):

https://codereview.chromium.org/11316316/diff/7004/content/browser/renderer_h...
content/browser/renderer_host/render_message_filter.h:173:
PepperRendererInstanceData instance_data,
Mainly because I modify the struct in that function. Alternatively I can make it
a const ref and make a copy. Would you prefer that?

https://codereview.chromium.org/11316316/diff/7004/content/common/pepper_rend...
File content/common/pepper_renderer_instance_data.h (right):

https://codereview.chromium.org/11316316/diff/7004/content/common/pepper_rend...
content/common/pepper_renderer_instance_data.h:18:
PepperRendererInstanceData(int render_process,
On 2012/12/05 23:00:26, brettw wrote:
> Do we need a destructor here since there are non-trivial destructors (GURL)? I
> think clang sometimes complains about this kind of thing. Probably a good idea
> anyway to prevent inlining.

Done.

https://codereview.chromium.org/11316316/diff/7004/content/common/view_messag...
File content/common/view_messages.h (right):

https://codereview.chromium.org/11316316/diff/7004/content/common/view_messag...
content/common/view_messages.h:1969: // -The instance id of the instance being
created.
On 2012/12/06 19:59:46, yzshen1 wrote:
> id -> ID to be consistent.

Done.

https://codereview.chromium.org/11316316/diff/7004/content/public/browser/bro...
File content/public/browser/browser_ppapi_host.h (right):

https://codereview.chromium.org/11316316/diff/7004/content/public/browser/bro...
content/public/browser/browser_ppapi_host.h:79: virtual GURL
GetPluginURLForInstance(PP_Instance instance) = 0;
Yep this will be needed in the future for Flash GetSetting

https://codereview.chromium.org/11316316/diff/7004/ppapi/proxy/broker_resourc...
File ppapi/proxy/broker_resource.cc (right):

https://codereview.chromium.org/11316316/diff/7004/ppapi/proxy/broker_resourc...
ppapi/proxy/broker_resource.cc:9: #include "ppapi/thunk/enter.h"
On 2012/12/06 19:59:46, yzshen1 wrote:
> I don't think you need these two includes?

Done.

https://codereview.chromium.org/11316316/diff/7004/ppapi/proxy/ppapi_messages.h
File ppapi/proxy/ppapi_messages.h (right):

https://codereview.chromium.org/11316316/diff/7004/ppapi/proxy/ppapi_messages...
ppapi/proxy/ppapi_messages.h:1438:
IPC_MESSAGE_CONTROL0(PpapiHostMsg_Broker_IsAllowed)
On 2012/12/05 23:00:26, brettw wrote:
> Can you document this here? I.e. that it's a call and the reply's error value
> contains the result of the query. I'd like to start using the format for the
> Talk interface below (big section for interface name, then separate comments
for
> each message).

Done.

https://codereview.chromium.org/11316316/diff/7004/ppapi/tests/test_broker.cc
File ppapi/tests/test_broker.cc (right):

https://codereview.chromium.org/11316316/diff/7004/ppapi/tests/test_broker.cc...
ppapi/tests/test_broker.cc:323: PP_Resource broker =
broker_interface_->CreateTrusted(
Good catch. But this is the case with all of the tests in this file. Rather than
changing it in this CL I'd rather put a TODO to write a C++ wrapper which will
handle the cleanup for us. Let me know if that's ok with you.

https://codereview.chromium.org/11316316/diff/7004/ppapi/tests/test_broker.cc...
ppapi/tests/test_broker.cc:332: PP_Resource broker =
broker_interface_->CreateTrusted(
As above.

[yzshen1, 2012-12-07 21:54:34]

lgtm

https://codereview.chromium.org/11316316/diff/7004/content/browser/renderer_h...
File content/browser/renderer_host/render_message_filter.h (right):

https://codereview.chromium.org/11316316/diff/7004/content/browser/renderer_h...
content/browser/renderer_host/render_message_filter.h:173:
PepperRendererInstanceData instance_data,
I didn't notice that.
Then never mind. :)
On 2012/12/07 21:32:19, raymes wrote:
> Mainly because I modify the struct in that function. Alternatively I can make
it
> a const ref and make a copy. Would you prefer that?

https://codereview.chromium.org/11316316/diff/7004/ppapi/tests/test_broker.cc
File ppapi/tests/test_broker.cc (right):

https://codereview.chromium.org/11316316/diff/7004/ppapi/tests/test_broker.cc...
ppapi/tests/test_broker.cc:323: PP_Resource broker =
broker_interface_->CreateTrusted(
You can use the raw C interface PPB_Core to release this ref.

I don't think it is a lot of code. But if you prefer a later CL, that is fine.

On 2012/12/07 21:32:19, raymes wrote:
> Good catch. But this is the case with all of the tests in this file. Rather
than
> changing it in this CL I'd rather put a TODO to write a C++ wrapper which will
> handle the cleanup for us. Let me know if that's ok with you.

[commit-bot: I haz the power, 2012-12-10 17:12:01]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11316316/24018

[commit-bot: I haz the power, 2012-12-10 17:12:14]

Failed to apply patch for
content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc:
While running patch -p1 --forward --force --no-backup-if-mismatch;
  patching file content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc
  Hunk #1 FAILED at 3.
  Hunk #2 FAILED at 61.
  2 out of 2 hunks FAILED -- saving rejects to file
content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc.rej

Patch:       content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc
Index: content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc
diff --git a/content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc
b/content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc
index
3e074c41db1ab0ffb27500bc31a3d0d07cd4392d..25a28237ce68e223ad3be7ec9a2420187cd47273
100644
--- a/content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc
+++ b/content/browser/renderer_host/pepper/browser_ppapi_host_impl.cc
@@ -3,8 +3,9 @@
 // found in the LICENSE file.
 
 #include "content/browser/renderer_host/pepper/browser_ppapi_host_impl.h"
-#include "content/browser/renderer_host/pepper/pepper_message_filter.h"
 
+#include "content/browser/renderer_host/pepper/pepper_message_filter.h"
+#include "content/common/pepper_renderer_instance_data.h"
 #include "content/public/browser/browser_thread.h"
 #include "content/public/browser/render_view_host.h"
 #include "ipc/ipc_message_macros.h"
@@ -61,43 +62,53 @@ base::ProcessHandle
BrowserPpapiHostImpl::GetPluginProcessHandle() const {
 }
 
 bool BrowserPpapiHostImpl::IsValidInstance(PP_Instance instance) const {
-  return instance_to_view_.find(instance) != instance_to_view_.end();
+  return instance_map_.find(instance) != instance_map_.end();
 }
 
 bool BrowserPpapiHostImpl::GetRenderViewIDsForInstance(
     PP_Instance instance,
     int* render_process_id,
     int* render_view_id) const {
-  InstanceToViewMap::const_iterator found = instance_to_view_.find(instance);
-  if (found == instance_to_view_.end()) {
+  InstanceMap::const_iterator found = instance_map_.find(instance);
+  if (found == instance_map_.end()) {
     *render_process_id = 0;
     *render_view_id = 0;
     return false;
   }
 
-  *render_process_id = found->second.process_id;
-  *render_view_id = found->second.view_id;
+  *render_process_id = found->second.render_process_id;
+  *render_view_id = found->second.render_view_id;
   return true;
 }
 
-void BrowserPpapiHostImpl::AddInstanceForView(PP_Instance instance,
-                                              int render_process_id,
-                                              int render_view_id) {
-  DCHECK(instance_to_view_.find(instance) == instance_to_view_.end());
+GURL BrowserPpapiHostImpl::GetDocumentURLForInstance(PP_Instance instance) {
+  InstanceMap::const_iterator found = instance_map_.find(instance);
+  if (found == instance_map_.end())
+    return GURL();
+  return found->second.document_url;
+}
+
+GURL BrowserPpapiHostImpl::GetPluginURLForInstance(PP_Instance instance) {
+  InstanceMap::const_iterator found = instance_map_.find(instance);
+  if (found == instance_map_.end())
+    return GURL();
+  return found->second.plugin_url;
+}
 
-  RenderViewIDs ids;
-  ids.process_id = render_process_id;
-  ids.view_id = render_view_id;
-  instance_to_view_[instance] = ids;
+void BrowserPpapiHostImpl::AddInstance(
+    PP_Instance instance,
+    const PepperRendererInstanceData& instance_data) {
+  DCHECK(instance_map_.find(instance) == instance_map_.end());
+  instance_map_[instance] = instance_data;
 }
 
-void BrowserPpapiHostImpl::DeleteInstanceForView(PP_Instance instance) {
-  InstanceToViewMap::iterator found = instance_to_view_.find(instance);
-  if (found == instance_to_view_.end()) {
+void BrowserPpapiHostImpl::DeleteInstance(PP_Instance instance) {
+  InstanceMap::iterator found = instance_map_.find(instance);
+  if (found == instance_map_.end()) {
     NOTREACHED();
     return;
   }
-  instance_to_view_.erase(found);
+  instance_map_.erase(found);
 }
 
 bool BrowserPpapiHostImpl::HostMessageFilter::OnMessageReceived(

[commit-bot: I haz the power, 2012-12-10 17:23:11]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11316316/25052

[commit-bot: I haz the power, 2012-12-10 17:23:36]

Presubmit check for 11316316-25052 failed and returned exit status 1.


Running presubmit commit checks ...

** Presubmit Warnings **
Found line ending with white spaces in:

***************
content/public/browser/browser_ppapi_host.h, line 76
***************

Presubmit checks took 4.4s to calculate.

[commit-bot: I haz the power, 2012-12-10 17:36:56]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11316316/45001

[commit-bot: I haz the power, 2012-12-10 19:28:49]

Change committed as 172104