linux: add flash workarounds

linux: add flash workarounds

This adds a few workarounds for Flash bugs on linux, hooking gtk and X functions.

[Antoine Labour, 2009-07-15 05:42:40]

An attempt to work-around some of the bugs with flash on linux. Looking for
comments on what you guys think is the best way to do this.
Ideally we need to understand why flash is acting up with us and not with
firefox, and fix it (and/or get Adobe to fix flash) but in the mean time that
makes gmail usable.
In this version the gtk and X functions are always hooked, in all the processes.
It doesn't seem like a big deal (since we default to the real ones unless we are
the plugin process running flash), but if it is an issue, it should be possible
to extract them into a .so that we LD_PRELOAD on the flash plugin process.

[Evan Martin, 2009-07-15 16:49:53]

I will bring this attention to Adobe.  I think we should avoid it if at all
possible.

http://codereview.chromium.org/155558/diff/1/8
File webkit/glue/plugins/plugin_host.cc (right):

http://codereview.chromium.org/155558/diff/1/8#newcode760
Line 760: *static_cast<Display **>(value) = GDK_DISPLAY();
Does this help the NULL-display case?

[Antoine Labour, 2009-07-15 17:15:03]

On Wed, Jul 15, 2009 at 9:49 AM, <evan@chromium.org> wrote:

> I will bring this attention to Adobe.  I think we should avoid it if at
> all possible.


Agreed.


>
> http://codereview.chromium.org/155558/diff/1/8
> File webkit/glue/plugins/plugin_host.cc (right):
>
> http://codereview.chromium.org/155558/diff/1/8#newcode760
> Line 760: *static_cast<Display **>(value) = GDK_DISPLAY();
> Does this help the NULL-display case?


I thought it could, but it doesn't.
Still it's one of the things we are supposed to implement, and since I had
written the code, I thought we might as well put it in.


>
> http://codereview.chromium.org/155558
>

[Antoine Labour, 2009-07-15 17:15:36]

On Wed, Jul 15, 2009 at 9:49 AM, <evan@chromium.org> wrote:

> I will bring this attention to Adobe.  I think we should avoid it if at
> all possible.


Agreed.


>
> http://codereview.chromium.org/155558/diff/1/8
> File webkit/glue/plugins/plugin_host.cc (right):
>
> http://codereview.chromium.org/155558/diff/1/8#newcode760
> Line 760: *static_cast<Display **>(value) = GDK_DISPLAY();
> Does this help the NULL-display case?


I thought it could, but it doesn't.
Still it's one of the things we are supposed to implement, and since I had
written the code, I thought we might as well put it in.


>
> http://codereview.chromium.org/155558
>