4.10-rc3 rpi issues

[Michael Zoran]

On Thu, 2017-01-19 at 08:22 +0100, Gerd Hoffmann wrote:
>   Hi,
> 
> > Anyway, if all you have is the edimax and a keyboard attached I
> > might
> > be able to reproduce the error locally on an RPI 2 myself and might
> > be
> > able to give the owners of dwc2 some debugging info. I have a USB
> > bus
> > analyzer that I bought a long time ago so I can probably send them
> > useful packet captures.
> > 
> > Maybe you can e-mail me your build .config file and I'll see if I
> > can
> > reproduce the issue.
> 
> There is a email with the .config earlier in this thread.
> 
> > I'm assuming nothing is special about the keyboard and the RPI 2,
> > correct?
> 
> The keyboard doesn't matter, it's running without kbd atm, I've
> connected a serial console instead now.  It has a touch screen
> display
> too, but otherwise it is a standard rpi2.
> 
> Another, possibly related data point:
> 
> On another rpi2, running openhab and connected via ethernet, I see
> errors like this now and then:
> 
> [26030.915862] dwc2 3f980000.usb: dwc2_hc_chhltd_intr_dma: Channel 4
> -
> ChHltd set, but reason is unknown
> [26030.916128] dwc2 3f980000.usb: hcint 0x00000002, intsts 0x04600029
> [52337.852814] dwc2 3f980000.usb: dwc2_hc_chhltd_intr_dma: Channel 3
> -
> ChHltd set, but reason is unknown
> [52337.853080] dwc2 3f980000.usb: hcint 0x00000002, intsts 0x04600029
> 
> That is the standard fedora distro kernel (4.9.3-200.fc25.armv7hl).
> 
> cheers,
>   Gerd
> 

Very interesting here...  This isn't related directly to your specific
issue but I'm finding that when I build this kernel with the exact
dot.config you sent out with the arm-linux-gnueabihf- compiler it boots
just fine on a RPI 2 but it completely fails to boot on a RPI 3.  In
both cases this is with the armhf version of debian sid.  Meaning I
just took the SD card out of the RPI 2 and put it in the RPI 3.

I thought the RPI 3 was 100% backward compatible with the RPI 2 from a
hardware point of view?

Anyway I have it booting on the RPI 2 now, so I can debug.  I have
noticed a very large number of experimental features being enabled in
the config file.  I'm wondering if that could be contributing to this
issue?

I'll continue debugging and see what I find...