Fix some race conditions during create engine, also fix dpkg build error.

Fix some race conditions during create engine, also fix dpkg build error.


BUG=chromium-os:10750
TEST=on Linux desktop

[Yusuke Sato, 2011-01-21 02:16:07]

Thanks for the patch for 10750.

Could you explain (in the CL description) what kind of bad scenarios this patch
could handle?
It was not obvious to me...

http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c
File bus/engineproxy.c (right):

http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode660
bus/engineproxy.c:660: /* Handler of notify::factory should be remove. */
remove -> removed

http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode672
bus/engineproxy.c:672: }
What happens if data->factory is NULL for some reason?

My guess is that the timeout_cb is called eventually and the callback function
returns G_DBUS_ERROR. Is this correct?

[Peng, 2011-01-21 15:45:23]

I guess sometime, the timeout is triggered between notify::factory signal, and
bus_factory_proxy_create_engine finished. It may cause the problem.

http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c
File bus/engineproxy.c (right):

http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode660
bus/engineproxy.c:660: /* Handler of notify::factory should be remove. */
On 2011/01/21 02:16:07, Yusuke Sato wrote:
> remove -> removed

Done.

http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode672
bus/engineproxy.c:672: }
On 2011/01/21 02:16:07, Yusuke Sato wrote:
> What happens if data->factory is NULL for some reason?
> 
> My guess is that the timeout_cb is called eventually and the callback function
> returns G_DBUS_ERROR. Is this correct?

I think it is better to continue waiting for timeout or next notify::factory
signal.

http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode672
bus/engineproxy.c:672: }
On 2011/01/21 02:16:07, Yusuke Sato wrote:
> What happens if data->factory is NULL for some reason?
> 
> My guess is that the timeout_cb is called eventually and the callback function
> returns G_DBUS_ERROR. Is this correct?

Done.

[Yusuke Sato, 2011-01-21 15:50:01]

Thanks for adding comments, LGTM

On 2011/01/21 15:45:23, Shawn.P.Huang wrote:
> I guess sometime, the timeout is triggered between notify::factory signal, and
> bus_factory_proxy_create_engine finished. It may cause the problem.
> 
> http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c
> File bus/engineproxy.c (right):
> 
> http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode660
> bus/engineproxy.c:660: /* Handler of notify::factory should be remove. */
> On 2011/01/21 02:16:07, Yusuke Sato wrote:
> > remove -> removed
> 
> Done.
> 
> http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode672
> bus/engineproxy.c:672: }
> On 2011/01/21 02:16:07, Yusuke Sato wrote:
> > What happens if data->factory is NULL for some reason?
> > 
> > My guess is that the timeout_cb is called eventually and the callback
function
> > returns G_DBUS_ERROR. Is this correct?
> 
> I think it is better to continue waiting for timeout or next notify::factory
> signal.
> 
> http://codereview.appspot.com/3970044/diff/1/bus/engineproxy.c#newcode672
> bus/engineproxy.c:672: }
> On 2011/01/21 02:16:07, Yusuke Sato wrote:
> > What happens if data->factory is NULL for some reason?
> > 
> > My guess is that the timeout_cb is called eventually and the callback
function
> > returns G_DBUS_ERROR. Is this correct?
> 
> Done.