RPi B+: linux-next / mainline and usb & ethernet

[Stephen Warren]

On 11/04/2014 01:31 AM, Matthias Klein wrote:
>
> Am 04.11.2014 um 05:24 schrieb Stephen Warren:
>> Are you able to run "git bisect" to find out which commit introduced the
>> problem?
 >
> Until now I didn't used "git bisect", but I think I need a "good" and
> "bad" commit. Can "git bisect" be used on a rebased tree like linux-next?

Yes, it works fine, providing the "good commit" is part of the base 
kernel that linux-next builds on.

> I don't know if the USB enumeration on the B+ did work on any
> stable/mainline/next tree. I only know that it works on Piotr's tree
> which contains non upstream patches:
> https://github.com/pietrushnic/rpi-dt-linux/tree/rpi-3.17.y
> (But Piotr has only a B rev 2.0 where the USB enumeration worked for him
> with linux-next)
>
> Today evening I will try a vanilla 3.17.1 because it is the base of
> Piotr's tree. If that works I will try to "git bisect" to mainline to
> find the problem commit.

You might want to try 3.17 rather than 3.17.1. I don't recall any 
differences relevant to the RPI, although I haven't been following any 
DWC2 fixes closely. The commits in 3.17 will all be part of linux-next, 
but anything in 3.17...3.17.1 won't be. Still, "git bisect" will work 
out what to do correctly just fine though (it'll likely just make you 
test 3.17 anyway as part of the bisect process)

> If it does not work on 3.17.1 I try to to find the commit in Piotr's
> tree which makes it working on the B+.
>
> Should I got that way, or do you recommend an another way?
>
>>
>> Do you know it the problem affects any RPi, or only B+? I'll try to see
>> if it repros on any of my systems tomorrow evening.
 >
> I have only a B+ to test here. I think it affects only the B+.

OK, that makes sense. I suppose I should buy one (and a compute module) 
to complete my collection of all the different types:-)