Re: [PATCH can-next v6] can: add tx/rx LED trigger support

[Fabio Baltieri <fabio.baltieri@gmail.com>]

On Fri, Sep 07, 2012 at 09:04:19AM +0200, Kurt Van Dijck wrote:
> On Thu, Sep 06, 2012 at 10:57:28PM +0200, Fabio Baltieri wrote:
> > On Thu, Sep 06, 2012 at 05:11:58PM +0200, Kurt Van Dijck wrote:
> > Can't argue with that... I'm trying to see how it comes but names like
> > "can-3-2:1.0-tx" doesn't looks that friendly to me...
> 
> "can-3-2:1.0" is an iface name?

No, I was trying some combination using device name instead of netdev
one (which remains can0), here 3-2:1.0 is the usb path, looks really bad
to be used for dev name if netdev name is still can0 - no way of
determine the link between the two.

> We must keep default iface names can%d. Only in the usecase that someone
> wants to name its ifaces according device names, he/she should manually
> adjust udev rules on his/her system ...
> 
> And if the iface is names "can-3-2:1.0", "can-3-2:1.0-tx" _is_
> a friendly trigger name.

Agreed.

> > > It looks even more inconvenient. It only works as expected when you don't
> > > change the trigger afterwards, but still it is possible (as should be),
> > > so the design of trigger is ... wrong.
> > > example:
> > > When you put another trigger to a led, and have a proper sequence of
> > > 'ip link set xxx up' and 'ip link set xxx down', you will end up
> > > with the default_trigger again.
> > > I realize I'm looking at unusual scenario's.
> > 
> > That's not correct, default trigger is going to be set only if there are
> > no other trigger assigned to the LED, and only on led probe and trigger
> > probe.
> > 
> > So, the LED framework is not going to change the trigger if you manually
> > changed it to something else, and in any case default_trigger assignment
> > only happens at probe/exit, not when interface is set up/down.
> 
> I think you're also arguing against a register/unregister during iface up/down.
> We agree.

Uh - yes! :-)

Fabio