[Letux-kernel] towards the the root of charging problems

[Andreas Kemnade]

On Fri, 29 Jul 2016 18:19:54 +0200
Andreas Kemnade <andreas at kemnade.info> wrote:

> Hi,
> 
> I am digging towards the root of charging problems and share some of
> my insights.
> phy_power_on is called with phy->power_count = -1
> so it does not call the corresponding power on function in phy-twl4030
> to set OTG_EN.
> 
> I do not know yet where and why that counter gets wrong.
> 

digging deeper:
musb_platform_enable()/musb_platform_disable()
is not balanced in musb_core.c
There is the following interesting piece of code
	/* be sure interrupts are disabled before connecting ISR */
        musb_platform_disable(musb);
        musb_generic_disable(musb);

in musb_init_controller() to switch from an unknown state to off state
which seems to be a good idea.
So there phy->power_count get -1. When the gadget driver gets loaded,
phy->power_count becomes 0 (not 1) so the phy is not powered.

I am developing a patch.

Regards,
Andreas