Hi,

On Fri, 2014-01-31 at 12:22 -0500, Tysen Moore wrote:
> I have limited experience with
> Gadget behavior so please bear with me.

Likewise...

> Based on our HW/testing I see the following behavior.  (Admittedly I am
> limited to one setup--not a ton of exposure)
> 
> For Cable/Wired Ethernet:
> Driver probed, no cable connected              Powered=TRUE,
> Connected=false, No Services
> Cable Connected (No IP Address)              Powered=TRUE, Connected=false,
> No Services
> Cable Connected (IP Address Assigned)   Powered=TRUE, Connected=TRUE,
> ethernet_MAC_cable
> 
> This differs slightly from what you mentioned.  Ethernet is powered *prior*
> to carrier detect.

But of course. I actually should have said "autoconnected". Oh well.

> For USB Gadget Ethernet:
> Driver probed, no cable connected              Powered=TRUE,
> Connected=false, No Services
> Cable Connected (No IP Address)              Powered=TRUE, Connected=false,
> No Services
> Cable Connected (IP Address Assigned)   Powered=TRUE, Connected=TRUE,
> gadget_MAC_usb

That was interesting and good news. I have seen a gadget connection
announcing its existence and tried to connect without a cable attached.
Something has been fixed, or then some gadget drivers work properly
where others don't.

What is the kernel version and driver you are using?

> As I previously mentioned, we do get a cable connect from rtnl,
> no different than Ethernet.

Then all is fine.

> Again, why should these be different?  Based on my testing the two work the
> same, are you suggesting they should not?  If so, why?

If they really detect the cable fine, all is well and it should behave
the same as ethernet.

> 3) Since we are new at this, what is the best work-flow for reworking and
> splitting up the patches based on your feedback. The only (tedious)
> approach I can see it starting from a fresh baseline and selectively
> applying parts of the original patches while doing commits to achieve your
> suggested grouping. Is there a better/preferred method? Any advice for a
> newbie?

You could do 'git reset HEAD~1', whereby the last commit vanishes and
you are left with a bunch of modified files. Run 'git add -i', 'p',
select modified files, select similarly looking parts from the
interactively shown snippets, end with 'q', get out of interactive patch
mode, check with 'git diff --cached' if everything necessary is added.
If not, add the missing pieces with another run of 'git add -i'. When
happy, 'git commit'. If the patches end up the wrong way, do 'git rebase
-i master', reorder and/or squash/fixup the individual patches.

See 'git add --help' for the real instructions on interactive add.


Cheers,

        Patrik


_______________________________________________
connman mailing list
[email protected]
https://lists.connman.net/mailman/listinfo/connman

Reply via email to