Re: [RFT PATCH] Revert "power: sequencing: request the WLAN enable GPIO as-is"

[Manivannan Sadhasivam]

On Mon, Dec 16, 2024 at 02:36:24PM +0100, Bartosz Golaszewski wrote:

[...]

> > Your hack is making sure that the default state of the GPIO is not changed at
> > all after initializing the controller. So even if the pwrctrl driver probes
> > later, it will try to enable the module by doing,
> > 'gpiod_set_value_cansleep(ctx->wlan_gpio, 1)', which would do nothing to the
> > device state.
> >
> > So the issue is not with the pwrctrl driver but with the controller
> > implementation. Ideally, once the device is removed, the PCIe link should move
> > to Detect state and then to Polling state once the receiver is detected on the
> > lanes. But the DWC and Qcom glue has other logics that prevents the controller
> > from doing so.
> >
> > So until the link down handling is implemented in the controller driver, we need
> > to carry this hack that preserves the GPIO state.
> >
> 
> Thanks for the explanation Mani. Regarding this patch: I suggest we
> keep it for now but maybe I'll add a comment saying why it's still
> necessary?
> 

Yeah, that makes sense.

- Mani

-- 
மணிவண்ணன் சதாசிவம்