On Sun, Oct 06, 2024 at 11:31:52PM +0530, Manivannan Sadhasivam wrote:
> On Sun, Oct 06, 2024 at 12:33:21AM +0530, Sibi Sankar wrote:
> >
> >
> > On 10/1/24 14:21, Manivannan Sadhasivam wrote:
> > > On Tue, Oct 01, 2024 at 09:56:30AM +0300, Dmitry Baryshkov wrote:
> > > > On October 1, 2024 5:42:35 AM GMT+03:00, Bjorn Andersson <andersson@xxxxxxxxxx> wrote:
> > > > > On Wed, Sep 11, 2024 at 10:55:05AM GMT, Dmitry Baryshkov wrote:
> > > > > > On Wed, Sep 11, 2024 at 01:03:37PM GMT, Sibi Sankar wrote:
> > > > > [..]
> > > > > > > diff --git a/arch/arm64/boot/dts/qcom/x1e001de-devkit.dts b/arch/arm64/boot/dts/qcom/x1e001de-devkit.dts
> > > > > [..]
> > > > > > > +
> > > > > > > +&pcie5 {
> > > > > > > + perst-gpios = <&tlmm 149 GPIO_ACTIVE_LOW>;
> > > > > > > + wake-gpios = <&tlmm 151 GPIO_ACTIVE_LOW>;
> > > > > > > +
> > > > > > > + vddpe-3v3-supply = <&vreg_wwan>;
> > > > > >
> > > > > > Please use pwrseq instead.
> > > > > >
> > > > >
> > > > > What benefit is there to wrap a single 3.3V regulator in pwrseq driver?
> > > >
> > > > First of all, is it really just a 3.3V? Second, is it actually powering up the host controller (as expressed in the device tree? Is it a power supply to the slot (in this case, I think, it should be expressed differently)? Or is it a power supply to the card itself?
> > > >
> > >
> > > Yeah, we should get into the details here. We were not paying attention till
> > > now, but with the advent of pwrseq, we should describe the power supply properly
> > > in DT.
> > >
> > > Here I believe the supply is to the PCIe Mini Card connector where a modem is
> > > connected. In that case, 3.3v supply should be connected to 3.3Vaux of the
> > > connector and we should have a generic pwrseq driver for the mini cards.
> > >
> >
> > Hey Mani, Dmitry,
> >
> > The schematics are identical to that of the X1E CRD with
> > the exception of the pcie daughter card having the rtl8125g
> > on it. Yes, the 3.3V supply is connected to the card as well.
> >
>
> Is this connected to the 3.3vaux of the card? Please specify the actual rail
> name as the 'PCI Express Mini Card Electromechanical Specification' specifies
> only 3.3Vaux and 1.5v supplies.
>
> > Doesn't this mean all other x1e boards out there needs to be
> > updated with pwrseq as well? Anway will get that addressed in
> > v3.
> >
>
> pwrseq is the kernel driver abstraction, nothing to do with DT. But for making
> use of pwrseq, the supplies need to be described in the proper place. In this
> case most likely under a separate node of PCIe bridge. Then you'd need a
> separate pwrseq driver in kernel to parse the supply and take care of it.
>
> I'm currently writing a pwrseq driver for standard slots (x8 for X1E) and should
> be able to post it early next week. So you or someone could use it as a
> reference to add a new driver for m-pcie cards.
>
> If no one picks it up, I may just do it.
>
Hi,
The slot driver is taking more time than anticipated due to the pwrctl rework.
So please go ahead with the current binding and we would switch to pwrseq
later once the driver is available.
- Mani
--
மணிவண்ணன் சதாசிவம்
