Re: [PATCH v2 1/5] pinctrl: qcom: Add ipq8074 pinctrl driver

[Bjorn Andersson]

On Mon 15 May 04:24 PDT 2017, Varadarajan Narayanan wrote:

> 
> 
> On 5/15/2017 2:35 PM, Varadarajan Narayanan wrote:
> > On 5/14/2017 9:53 AM, Bjorn Andersson wrote:
> > > On Thu 11 May 03:33 PDT 2017, Varadarajan Narayanan wrote:
> > > 
> > > > On 5/11/2017 4:13 AM, Bjorn Andersson wrote:
> > > > > On Thu 04 May 04:53 PDT 2017, Varadarajan Narayanan wrote:
> > > > > 
> > > [..]
> > > > > > +enum ipq8074_functions {
> > > > > 
> > > > > Please keep these sorted alphabetically.
> > > > 
> > > > Ok
> > > > 
> > > > > > +    msm_mux_gpio,
> > > > > > +    msm_mux_qpic_pad,
> > > > > > +    msm_mux_blsp5_i2c,
> > > > > > +    msm_mux_blsp5_spi,
> > > > > > +    msm_mux_wci20,
> > > > > 
> > > > > What does "20" mean here?
> > > > 
> > > > This is for Wireless Coex Interface. The same functionality can
> > > > be muxed on
> > > > to different GPIOs. WCI2, is the 2nd edition of the WCI standard
> > > > and 0, 1
> > > > are for the muxing to different GPIOs (alternate muxes).
> > > > 
> > > 
> > > In other Qualcomm platforms the alternative muxes are denoted by letters
> > > (a,b,c...). Would you mind picking up this naming scheme, or do you see
> > > any problems with that? (E.g. wci2a in this case)
> > 
> > Ok
> > 
> > > Btw, do you need any additional configuration for selecting alternative
> > > muxing or is that automagical these days?
> > 
> > No additional configuration is needed.
> > 
> > > > > > +    msm_mux_blsp3_spi3,
> > > > > > +    msm_mux_burn0,
> > > > > > +    msm_mux_pcm_zsi0,
> > > > > > +    msm_mux_blsp5_uart,
> > > > > > +    msm_mux_mac12,
> > > > > 
> > > > > What does "12" mean here?
> > > > 
> > > > The SoC has three MAC cores. Each core has two pins for the
> > > > smart antenna
> > > > feature. macXY indicates the function select for MAC no. X and
> > > > smart antenna
> > > > no. Y.
> > > > 
> > > 
> > > Ok
> > > 
> > > > > > +    msm_mux_blsp3_spi0,
> > > > > > +    msm_mux_burn1,
> > > > > > +    msm_mux_mac01,
> > > > > > +    msm_mux_qdss_cti_trig_out_b0,
> > > > > > +    msm_mux_qdss_cti_trig_in_b0,
> > > > > > +    msm_mux_qpic_pad4,
> > > > > 
> > > > > What are qpic_pad and qpic_pad0 through qpic_pad8? Different functions,
> > > > > alternative muxings...?
> > > > 
> > > > This is for the NAND and LCD display. The pins listed are the 9
> > > > data pins.
> > > > 
> > > 
> > > Then you can describe them all as "qpic_pad" (or simply "qpic"?). (It's
> > > possible to reference a partial group in the DTS, if that's necessary)
> > 
> > There are two sets of 9 pins, either of which can go to NAND or LCD.
> > Will rename qpic_pad as qpic_a and qpic_pad[0-8] as qpic_b.
> > Is that ok?
> > 
> > > > > > +    msm_mux_blsp4_uart0,
> > > > > > +    msm_mux_blsp4_i2c0,
> > > > > > +    msm_mux_blsp4_spi0,
> > > > > > +    msm_mux_mac21,
> > > > > > +    msm_mux_qdss_cti_trig_out_b1,
> > > > > > +    msm_mux_qpic_pad5,
> > > > > > +    msm_mux_qdss_cti_trig_in_b1,
> > > > > > +    msm_mux_qpic_pad6,
> > > > > > +    msm_mux_qpic_pad7,
> > > > > > +    msm_mux_cxc0,
> > > > > > +    msm_mux_mac13,
> > > > > > +    msm_mux_qdss_cti_trig_in_a1,
> > > > > > +    msm_mux_qdss_cti_trig_out_a1,
> > > > > > +    msm_mux_wci22,
> > > > > > +    msm_mux_qdss_cti_trig_in_a0,
> > > > > > +    msm_mux_qpic_pad1,
> > > > > > +    msm_mux_qdss_cti_trig_out_a0,
> > > > > > +    msm_mux_qpic_pad2,
> > > > > > +    msm_mux_qpic_pad3,
> > > > > > +    msm_mux_qdss_traceclk_b,
> > > > > > +    msm_mux_qpic_pad0,
> > > > > > +    msm_mux_qdss_tracectl_b,
> > > > > > +    msm_mux_qpic_pad8,
> > > > > > +    msm_mux_pcm_zsi1,
> > > > > > +    msm_mux_qdss_tracedata_b,
> > > > > > +    msm_mux_led0,
> > > > > > +    msm_mux_pwm04,
> > > > > 
> > > > > What does "04" mean here?
> > > > 
> > > > There are 4 Pulse Width Modulation channels, pwmXY is pwm
> > > > channel X and pin
> > > > Y.
> > > 
> > > So Y is alternative mux? Can we use letters for this as well?
> > 
> > Ok
> 
> Sorry, actually they are not alternative muxes. There are 4 different PWM
> instances. Each PWM instance can handle 'n' different GPIOs (for example,
> that can be connected to LED etc.).
> 

So we have 4 PWMs and the output from a single PWM can muxed to drive
the output of 4 pins?

Then skip the last digit and just go pwm0, ... pwm3 and include all four
pins that can be part of each instance. As the pingroup configuration is
per pin in the TLMM it's possible to configure subsets of the pins in a
group.

Regards,
Bjorn