Only scan for correct architecture plugin binaries

Only load platform specific architecture plugins from the correct registry hive.

This fixes a regression in r298758 where if a 64-bit plugin was loaded after
a 32-bit plugin, and the versions were the same, then the 32-bit plugin would
be removed from the list, the 64-bit plugin plugin would be removed and
neither plugin would successfully load.

Solution is to only load the correct architecture plugins from the registry.

BUG=428549
Committed: https://crrev.com/6c217552734bf7516fdd47495cdc37948e5d2a93
Cr-Commit-Position: refs/heads/master@{#302713}

[Will Harris, 2014-10-30 00:39:06]

wfh@chromium.org changed reviewers:
+ ananta@chromium.org

[Will Harris, 2014-10-31 04:11:19]

I know why the tests are failing and will upload a new patchset tomororw

[Will Harris, 2014-11-01 01:55:52]

ananta PTAL. thanks.

[Will Harris, 2014-11-03 20:05:47]

wfh@chromium.org changed reviewers:
+ bauerb@chromium.org

[Will Harris, 2014-11-03 20:06:11]

I think ananta is OOO today, bauerb can you take a look?

[ananta, 2014-11-03 20:28:04]

https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
File content/common/plugin_list_win.cc (right):

https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
content/common/plugin_list_win.cc:401: 
Can we bail if wrong_architecture is true? here?

[Will Harris, 2014-11-03 20:55:29]

On 2014/11/03 20:28:04, ananta wrote:
>
https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
> File content/common/plugin_list_win.cc (right):
> 
>
https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
> content/common/plugin_list_win.cc:401: 
> Can we bail if wrong_architecture is true? here?

The test expectations are quite subtle, see previous attempts where I did
exactly that...

[Bernhard Bauer, 2014-11-03 22:24:23]

https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
File content/common/plugin_list_win.cc (right):

https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
content/common/plugin_list_win.cc:401: 
On 2014/11/03 20:28:04, ananta wrote:
> Can we bail if wrong_architecture is true? here?

Alternatively:

  bool should_check_version;
  if (wrong_architecture) {
    should_check_version = false;
  } else {
    base::AutoLock lock(lock_);
    should_check_version = ...;
  }

That would at least get rid of one level of indentation.

[Will Harris, 2014-11-03 22:38:12]

On 2014/11/03 22:24:23, Bernhard Bauer wrote:
>
https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
> File content/common/plugin_list_win.cc (right):
> 
>
https://codereview.chromium.org/674073003/diff/40001/content/common/plugin_li...
> content/common/plugin_list_win.cc:401: 
> On 2014/11/03 20:28:04, ananta wrote:
> > Can we bail if wrong_architecture is true? here?
> 
> Alternatively:
> 
>   bool should_check_version;
>   if (wrong_architecture) {
>     should_check_version = false;
>   } else {
>     base::AutoLock lock(lock_);
>     should_check_version = ...;
>   }
> 
> That would at least get rid of one level of indentation.

after chatting to ananta I'm just going to change
PluginList::GetPluginPathsFromRegistry to only pull from correct bitness part of
HKLM, reverting the behavior back to before r298758

[Will Harris, 2014-11-03 23:09:35]

PTAL - this code is far simpler.

[ananta, 2014-11-04 01:06:03]

lgtm

https://codereview.chromium.org/674073003/diff/60001/content/common/plugin_li...
File content/common/plugin_list_win.cc (right):

https://codereview.chromium.org/674073003/diff/60001/content/common/plugin_li...
content/common/plugin_list_win.cc:371: 0,
Please add a comment here saying that this will only work for plugins
which install under the correct subkeys for 32 and 64 bit Windows.

[Will Harris, 2014-11-04 22:58:44]

The CQ bit was checked by wfh@chromium.org

[commit-bot: I haz the power, 2014-11-04 23:01:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/674073003/80001

[commit-bot: I haz the power, 2014-11-04 23:54:57]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2014-11-04 23:56:24]

Patchset 5 (id:??) landed as
https://crrev.com/6c217552734bf7516fdd47495cdc37948e5d2a93
Cr-Commit-Position: refs/heads/master@{#302713}