[RFC PATCH 14/15] WIP: PCI: qcom: use pwrseq to power up bus devices

Dmitry Baryshkov dmitry.baryshkov at linaro.org
Fri Aug 20 01:50:04 PDT 2021


Hi,

On Fri, 20 Aug 2021 at 02:43, Bjorn Andersson
<bjorn.andersson at linaro.org> wrote:
>
> On Mon 16 Aug 17:55 PDT 2021, Dmitry Baryshkov wrote:
>
> > Use bus-pwrseq device tree node to power up the devices on the bus. This
> > is to be rewritten with the proper code parsing the device tree and
> > powering up individual devices.
> >
>
> How about describing the PCI device in DT and having the PCIe controller
> dig it up up from there? Although we won't have a struct device until
> later, so perhaps we need the of-based pwrseq_get() for that.

Yes, this is the plan. Currently I just wanted to have a way to power
up the PCIe part of the chip and be able to test that.
In the previous attempts to provide qca6390 support, Rob has clearly
stated that we'd have to have the following device tree snippet:

pcie0 {
    bridge at 0,0 {
        qca6390 at 1,0 {
            supplies-or-whatever = ....;
        };
    };
};

>
> Regards,
> Bjorn
>
> > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov at linaro.org>
> > ---
> >  drivers/pci/controller/dwc/pcie-qcom.c | 13 +++++++++++++
> >  drivers/power/pwrseq/pwrseq_qca.c      |  1 +
> >  2 files changed, 14 insertions(+)
> >
> > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
> > index 8a7a300163e5..a60d41fbcd6f 100644
> > --- a/drivers/pci/controller/dwc/pcie-qcom.c
> > +++ b/drivers/pci/controller/dwc/pcie-qcom.c
> > @@ -23,6 +23,7 @@
> >  #include <linux/pm_runtime.h>
> >  #include <linux/platform_device.h>
> >  #include <linux/phy/phy.h>
> > +#include <linux/pwrseq/consumer.h>
> >  #include <linux/regulator/consumer.h>
> >  #include <linux/reset.h>
> >  #include <linux/slab.h>
> > @@ -1467,6 +1468,7 @@ static int qcom_pcie_probe(struct platform_device *pdev)
> >       struct pcie_port *pp;
> >       struct dw_pcie *pci;
> >       struct qcom_pcie *pcie;
> > +     struct pwrseq *pwrseq;
> >       int ret;
> >
> >       pcie = devm_kzalloc(dev, sizeof(*pcie), GFP_KERNEL);
> > @@ -1520,6 +1522,17 @@ static int qcom_pcie_probe(struct platform_device *pdev)
> >
> >       pp->ops = &qcom_pcie_dw_ops;
> >
> > +     pwrseq = devm_pwrseq_get_optional(dev, "bus");
> > +     if (IS_ERR(pwrseq)) {
> > +             ret = PTR_ERR(pwrseq);
> > +             goto err_pm_runtime_put;
> > +     }
> > +     if (pwrseq) {
> > +             ret = pwrseq_full_power_on(pwrseq);
> > +             if (ret)
> > +                     goto err_pm_runtime_put;
> > +     }
> > +
> >       ret = phy_init(pcie->phy);
> >       if (ret) {
> >               pm_runtime_disable(&pdev->dev);
> > diff --git a/drivers/power/pwrseq/pwrseq_qca.c b/drivers/power/pwrseq/pwrseq_qca.c
> > index 3421a4821126..4107f0a9c05d 100644
> > --- a/drivers/power/pwrseq/pwrseq_qca.c
> > +++ b/drivers/power/pwrseq/pwrseq_qca.c
> > @@ -1,3 +1,4 @@
> > +#define DEBUG
> >  // SPDX-License-Identifier: GPL-2.0-only
> >  /*
> >   * Copyright (c) 2021, Linaro Ltd.
> > --
> > 2.30.2
> >



-- 
With best wishes
Dmitry



More information about the ath10k mailing list