Re: [RFC PATCH 00/10] Netronix embedded controller driver for Kobo and Tolino ebook readers

[Andreas Kemnade]

On Sat, 4 Jul 2020 22:58:08 +0200
Jonathan NeuschÃfer <j.neuschaefer@xxxxxxx> wrote:

> On Tue, Jun 30, 2020 at 10:14:47PM +0200, Andreas Kemnade wrote:
> > On Tue, 30 Jun 2020 09:15:23 +0200
> > Jonathan NeuschÃfer <j.neuschaefer@xxxxxxx> wrote:
> >   
> > > On Tue, Jun 30, 2020 at 08:40:51AM +0200, Andreas Kemnade wrote:
> > > [...]  
> > > > got a chance to test it on a Tolino Shine 2 HD.
> > > > It uses the RTC from the RC5T619 but backlight seems to go via MSP430
> > > > EC.
> > > > 
> > > > I got this.
> > > > 
> > > > [    1.453603] ntxec 0-0043: Netronix embedded controller version f110 detected.
> > > > [   10.723638] ntxec-rtc 21a0000.i2c:embedded-controller@43:rtc: registered as rtc0
> > > > [   10.775276] ntxec-pwm: probe of 21a0000.i2c:embedded-controller@43:pwm failed with error -5    
> > > 
> > > Hmm, -EIO from the PWM driver.
> > >   
> > turing debugging on:  
> 
> (edited for compactness:)
> > [  330.332971] i2c i2c-0: write slave address: addr=0x86   ACK received
> > [  330.334420] i2c i2c-0: write byte: B0=0xA3              ACK received
> > [  330.334790] i2c i2c-0: write byte: B1=0x0               No ACK  
> 
> > [  330.352339] i2c i2c-0: write slave address: addr=0x86   ACK received
> > [  330.362208] i2c i2c-0: write byte: B0=0xA1              ACK received
> > [  330.362479] i2c i2c-0: write byte: B1=0xFF              No ACK  
> 
> > [  330.363112] i2c i2c-0: write slave address: addr=0x86   ACK received
> > [  330.363362] i2c i2c-0: write byte: B0=0xA2              ACK received
> > [  330.363608] i2c i2c-0: write byte: B1=0xFF              No ACK  
> 
> Hmm, it doesn't ack the writes to 0xA3, 0xA1 and 0xA2, which should
> disable the PWM output and then disable the auto-off timer (according to
> the vendor kernel).
> 
> And you said in your other mail that you can actually toggle the light
> with writes to 0xA3, so I suspect a bug in the EC firmware here (which
> may have gone unnoticed because the vendor kernel doesn't check if the
> i2c transfers succeed). :/
> 
That is a also my theory.

> IMHO we should get this driver merged first, and perhaps add a quirk to
> deal with the missing ACKs later (unless a better solution is found).
> 
Yes, that can be done separately (after the Tolino Shine 2 HD dtb is
in, I am polishing it right now).

Regards,
Andreas