[Letux-kernel] letux-4.7-rc7: charging & ethernet gadget

[Andreas Kemnade]

On Mon, 18 Jul 2016 18:08:38 +0200
"H. Nikolaus Schaller" <hns at goldelico.com> wrote:

> 
> > Am 18.07.2016 um 09:47 schrieb H. Nikolaus Schaller
> > <hns at goldelico.com>:
> > 
> > Next test:
> > 
> > letux-4.7-rc7 but the following files pulled from v4.3:
> > 
> > 	geändert:       drivers/hwmon/twl4030-madc-hwmon.c
> > 	geändert:       drivers/input/keyboard/twl4030_keypad.c
> > 	geändert:       drivers/input/misc/twl4030-pwrbutton.c
> > 	geändert:       drivers/input/misc/twl4030-vibra.c
> > 	geändert:       drivers/mfd/twl4030-audio.c
> > 	geändert:       drivers/mfd/twl4030-irq.c
> > 	geändert:       drivers/mfd/twl4030-power.c
> > 	geändert:       drivers/phy/phy-twl4030-usb.c
> > 	geändert:       drivers/power/twl4030_charger.c
> > 	geändert:       drivers/usb/musb/omap2430.c
> > 	geändert:       include/linux/platform_data/omap-twl4030.h
> > 	neue Datei:     include/linux/usb/musb-omap.h
> > 	geändert:       sound/soc/codecs/twl4030.c
> > 	geändert:       sound/soc/omap/omap-twl4030.c
> > 
> > This compiles and boots. But charging isn't working (nor gadget).
> > Not even in continuous mode.
> > 
> > There is one difference now:
> > when replugging (was plugged during boot) the USB cable I
> > get
> > 
> > [  202.650177] omap_musb_mailbox: musb core is not yet ready
> > 
> > Next I will try to backport the musb subsystem from v4.3.
> 
> Well, finally I gave up.
> 
> Initially it started fine. Just pull in musb from 4.3. And compile.
> But then there were dependencies. To usb-core. Ok, backport that
> as well. Next are dependencies on gadget.
> 
> And so on.
> 
> At the point where it started to require a different inode structure
> I gave up because it becomes unlikely that it will ever work before
> I have essentially done a git checkout 4.3...
> 
> Anyways this exercise gave me some hints that plugging events are
> related to the musb driver. And musb is the SoC circuits used for the
> OTG port. Maybe disabling musb makes charging work. But it might
> miss the power-plugin event.
> 

MUSB itself gets the plugging info via ULPI from twl4030...
And if VBus measurement somehow does not work, it also influences
musb. I just remembered the state we reached in the kernel workshop.
Linear charging was working. Vbus measurement was working. Gadget was
working, b_host.sh was working. That was 4.7rc1 + fixes done there. So
somewhere in between there and now caused something vbus measurement to
break.
As now b_host.sh, vbus and gadget do not work.
So maybe we should first concentrace on reaching the workshop state
again. So that is a bit like a git bisect between that 4.7rc1 and now.

> Or changes something in the madc/bciadc part.
> 
Any side effects from generic madc battery?

Regards,
Andreas