Re: [PATCH v2] Saving agent-owned secrets for newly created connections
- From: Jirka Klimes <jklimes redhat com>
- To: Dan Williams <dcbw redhat com>
- Cc: networkmanager-list gnome org
- Subject: Re: [PATCH v2] Saving agent-owned secrets for newly created connections
- Date: Tue, 15 Nov 2011 13:46:54 +0100
On Tuesday 15 of November 2011 00:33:11 Dan Williams wrote:
> On Thu, 2011-11-10 at 13:09 +0100, Jirka Klimes wrote:
> > On Wednesday 09 of November 2011 19:12:34 Dan Williams wrote:
> > > On Wed, 2011-11-09 at 17:13 +0100, Jirka Klimes wrote:
> > > > Hello,
> > > >
> > > > attached is a patch for sending secrets to agents when a new
> > > > connection is created (AddConnection, AddAndActivateConnection).
> > > >
> > > > At present, when a new connection is created, the secrets are not
> > > > sent to agents and thus not saved. One has to edit the connection
> > > > again. It is easily reproducible by adding a VPN connection. When it
> > > > is opened again in the editor, the secrets are not there.
> > >
> > > Hmm, could we do the save from pk_add_cb() and make the patch smaller?
> > > That way nm_settings_connection_save_agent_secrets() could be made
> > > private to nm-settings.c. And we wouldn't have to pass the caller_uid
> > > to the nm_settings_add_connection() callback, which saves some churn.
> >
> > You are right. Done.
> > I wanted to save some (copy-paste) code, but due to passing caller_uid
> > didn't save much. I also didn't realize that having the stuff in
> > pk_add_cb() allows more contained code and not to pollute nm-manager.c.
>
> Looks good, thanks!
>
> Dan
>
> > > Ordering might also be interesting here; since this could trigger a
> > > SaveSecrets before the agent has been able to process the new
> > > connection, meaning that the agent may not actually have grabbed the
> > > connection yet (since that's another dbus call) when SaveSecrets comes
> > > in. I don't think that's a problem in practice for the nm-applet agent
> > > but it might be for others. Not sure if there's anything we can do
> > > about it though since the ordering would be correct.
> > >
> > > One thing we should probably do (later though) is not call SaveSecrets
> > > at all if there aren't any secrets left to send after filtering for
> > > AGENT_OWNED secrets. Shouldn't have any real effect, but would be more
> > > "correct".
> > >
> > > Dan
Thanks!
Pushed as 2106d94d08c27c38aa0b0a23d1631c18fd84d54f
Jirka
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]