Re: [PATCH v5 4/4] PCI: qcom: Switch pcie_1_pipe_clk_src after PHY init in SC7280

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Sep 09, 2021 at 11:21:22PM +0530, Prasad Malisetty wrote:
> On 2021-08-31 21:07, Bjorn Helgaas wrote:
> > On Tue, Aug 31, 2021 at 12:07:30PM +0530, Prasad Malisetty wrote:
> > > On 2021-08-26 18:07, Rob Herring wrote:
> > > > On Thu, Aug 26, 2021 at 2:22 AM Prasad Malisetty
> > > > <pmaliset@xxxxxxxxxxxxxx> wrote:
> > > > >
> > > > > On 2021-08-26 02:55, Bjorn Helgaas wrote:
> > > > > > [+cc linux-pci; patches to drivers/pci/ should always be cc'd there]
> > > > > >
> > > > > > On Wed, Aug 25, 2021 at 07:30:09PM +0000, Stephen Boyd wrote:
> > > > > >> Quoting Prasad Malisetty (2021-08-24 01:10:48)
> > > > > >> > On 2021-08-17 22:56, Prasad Malisetty wrote:
> > > > > >> > > On 2021-08-10 09:38, Prasad Malisetty wrote:
> > > > > >> > >> On the SC7280, By default the clock source for pcie_1_pipe is
> > > > > >> > >> TCXO for gdsc enable. But after the PHY is initialized, the clock
> > > > > >> > >> source must be switched to gcc_pcie_1_pipe_clk from TCXO.
> > > > > >> > >>
> > > > > >> > >> Signed-off-by: Prasad Malisetty <pmaliset@xxxxxxxxxxxxxx>
> > > > > >> > >> ---
> > > > > >> > >>  drivers/pci/controller/dwc/pcie-qcom.c | 18 ++++++++++++++++++
> > > > > >> > >>  1 file changed, 18 insertions(+)
> > > > > >> > >>
> > > > > >> > >> diff --git a/drivers/pci/controller/dwc/pcie-qcom.c
> > > > > >> > >> b/drivers/pci/controller/dwc/pcie-qcom.c
> > > > > >> > >> index 8a7a300..39e3b21 100644
> > > > > >> > >> --- a/drivers/pci/controller/dwc/pcie-qcom.c
> > > > > >> > >> +++ b/drivers/pci/controller/dwc/pcie-qcom.c
> > > > > >> > >> @@ -166,6 +166,8 @@ struct qcom_pcie_resources_2_7_0 {
> > > > > >> > >>      struct regulator_bulk_data supplies[2];
> > > > > >> > >>      struct reset_control *pci_reset;
> > > > > >> > >>      struct clk *pipe_clk;
> > > > > >> > >> +    struct clk *gcc_pcie_1_pipe_clk_src;
> > > > > >> > >> +    struct clk *phy_pipe_clk;
> > > > > >> > >>  };
> > > > > >> > >>
> > > > > >> > >>  union qcom_pcie_resources {
> > > > > >> > >> @@ -1167,6 +1169,16 @@ static int qcom_pcie_get_resources_2_7_0(struct
> > > > > >> > >> qcom_pcie *pcie)
> > > > > >> > >>      if (ret < 0)
> > > > > >> > >>              return ret;
> > > > > >> > >>
> > > > > >> > >> +    if (of_device_is_compatible(dev->of_node, "qcom,pcie-sc7280")) {
> > > > > >> > >> +            res->gcc_pcie_1_pipe_clk_src = devm_clk_get(dev, "pipe_mux");
> > > > > >> > >> +            if (IS_ERR(res->gcc_pcie_1_pipe_clk_src))
> > > > > >> > >> +                    return PTR_ERR(res->gcc_pcie_1_pipe_clk_src);
> > > > > >> > >> +
> > > > > >> > >> +            res->phy_pipe_clk = devm_clk_get(dev, "phy_pipe");
> > > > > >> > >> +            if (IS_ERR(res->phy_pipe_clk))
> > > > > >> > >> +                    return PTR_ERR(res->phy_pipe_clk);
> > > > > >> > >> +    }
> > > > > >> > >
> > > > > >> > > I would like to check is there any other better
> > > > > >> > > approach instead of compatible method here as well or
> > > > > >> > > is it fine to use compatible method.
> > > > > >>
> > > > > >> I'd prefer the compatible method. If nobody is responding
> > > > > >> then it's best to just resend the patches with the
> > > > > >> approach you prefer instead of waiting for someone to
> > > > > >> respond to a review comment.
> > > > > >
> > > > > > I'm missing some context here, so I'm not exactly sure
> > > > > > what your question is, Prasad, but IMO drivers generally
> > > > > > should not need to use of_device_is_compatible() if
> > > > > > they've already called of_device_get_match_data() (as
> > > > > > qcom_pcie_probe() has).
> > > > > >
> > > > > > of_device_is_compatible() does basically the same work of
> > > > > > looking for a match in qcom_pcie_match[] that
> > > > > > of_device_get_match_data() does, so it seems pointless to
> > > > > > repeat it.
> > > >
> > > > +1
> > > >
> > > > > > I am a little confused because while [1] adds "qcom,pcie-sc7280" to
> > > > > > qcom,pcie.txt, I don't see a patch that adds it to qcom_pcie_match[].
> > > >
> > > > Either that's missing or there's a fallback to 8250 that's not
> > > > documented.
> > > >
> > > > > I agree on your point, but the main reason is to use compatible in
> > > > > get_resources_2_7_0 is same hardware version. For SM8250 & SC7280
> > > > > platforms, the hw version is same. Since we can't have a separate ops
> > > > > for SC7280, we are using compatible method in get_resources_2_7_0 to
> > > > > differentiate SM8250 and SC7280.
> > > >
> > > > Then fix the match data to be not just ops, but ops and the flag you
> > > > need here.
> > > 
> > > This difference is not universal across all the platforms but
> > > instead this is specific to SC7280.  Hence it make sense to use
> > > compatible other than going for a flag.
> > 
> > There's no reason your qcom_pcie_match[].data pointers need to be
> > strictly based on the hardware version.
> > 
> > You can do something like what pcie-brcmstb.c does, e.g.,
> > 
> >   struct pcie_cfg_data {
> >     struct qcom_pcie_ops *ops;
> >     unsigned int pipe_mux:1;
> >   };
> > 
> >   static const struct pcie_cfg_data sm8250_cfg = {
> >     .ops = &ops_1_9_0,
> >   };
> > 
> >   static const struct pcie_cfg_data sc7280_cfg = {
> >     .ops = &ops_1_9_0,
> >     .pipe_mux = 1,
> >   };
> > 
> >   static const struct of_device_id qcom_pcie_match[] = {
> >     { .compatible = "qcom,pcie-sm8250", .data = &sm8250_cfg },
> >     { .compatible = "qcom,pcie-sc7280", .data = &sc7280_cfg },
> >   };
> 
> I have one quick query, If we use above approach, we should change platform
> data reading in PCIe probe to differentiate remaining platforms right.
> expect SM8250 and SC7280 all other platforms are using same qcom_pcie_ops
> structure pointer as data.

Yes.  of_device_get_match_data() must return the same type of pointer
(in the example above, "struct pcie_cfg_data *") for all platforms.
So you would have to add a struct for each of them, and each struct
would contain the ops pointer (&ops_1_0_0, &ops_2_1_0, etc).

Thanks for working on this!

Bjorn



[Index of Archives]     [DMA Engine]     [Linux Coverity]     [Linux USB]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Greybus]

  Powered by Linux