Hi Bjorn,
> From: Bjorn Helgaas, Sent: Wednesday, March 27, 2024 5:21 AM
>
> Include the function pointer name in the subject so it's a little more
> specific.
I got it. I'll change the subject.
> On Tue, Mar 26, 2024 at 11:45:38AM +0900, Yoshihiro Shimoda wrote:
> > This driver can reuse other R-Car Gen4 SoC support. However, some
> > initializing settings differs between r8a779f0 and others. So, add
> > a new function pointer start_link_enable() to support other R-Car
> > Gen4 SoC in the future. No behavior changes.
>
> Make it clear here what the new SoC is. I think it's r8a779f0, but
> you have to read the patch and look for the new .compatible string to
> figure that out.
I got it. The new SoC is r8a779g0. So, I should add such description here.
> > Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@xxxxxxxxxxx>
> > ---
> > drivers/pci/controller/dwc/pcie-rcar-gen4.c | 57 +++++++++++++++++++--
> > 1 file changed, 52 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/pci/controller/dwc/pcie-rcar-gen4.c b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> > index 0be760ed420b..a37613dd9ff4 100644
> > --- a/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> > +++ b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> > @@ -53,9 +53,16 @@ struct rcar_gen4_pcie {
> > void __iomem *base;
> > struct platform_device *pdev;
> > enum dw_pcie_device_mode mode;
> > +
> > + int (*start_link_enable)(struct rcar_gen4_pcie *rcar);
> > };
> > #define to_rcar_gen4_pcie(_dw) container_of(_dw, struct rcar_gen4_pcie, dw)
> >
> > +struct rcar_gen4_pcie_platdata {
> > + enum dw_pcie_device_mode mode;
> > + int (*start_link_enable)(struct rcar_gen4_pcie *rcar);
>
> I think it's confusing to repeat "mode" and "start_link_enable" in
> both rcar_gen4_pcie and rcar_gen4_pcie_platdata. I know several other
> drivers use this pattern, but I think it is simpler overall to just
> save the pointer directly, e.g.,
>
> imx6_pcie_probe
> imx6_pcie->drvdata = of_device_get_match_data(dev);
>
> ls_pcie_probe
> pcie->drvdata = of_device_get_match_data(dev);
>
> tegra_pcie_dw_probe
> data = of_device_get_match_data(dev);
> pcie->of_data = (struct tegra_pcie_dw_of_data *)data;
>
> So I think the best thing would be to add struct
> rcar_gen4_pcie_platdata, *move* rcar_gen4_pcie.mode there, and save a
> pointer to the rcar_gen4_pcie_platdata in struct rcar_gen4_pcie.
I got it. I'll modify the patch.
> That could be its own separate patch, which is nice on its own because
> it gets rid of the (void *) casts in rcar_gen4_pcie_of_match[].
>
> Then add .start_link_enable() (or .ltssm_enable(), see below) and the
> r8a779f0 bits in another patch.
I got it. I'll make such a patch at first.
> > +};
> > +
> > /* Common */
> > static void rcar_gen4_pcie_ltssm_enable(struct rcar_gen4_pcie *rcar,
> > bool enable)
> > @@ -123,9 +130,13 @@ static int rcar_gen4_pcie_speed_change(struct dw_pcie *dw)
> > static int rcar_gen4_pcie_start_link(struct dw_pcie *dw)
> > {
> > struct rcar_gen4_pcie *rcar = to_rcar_gen4_pcie(dw);
> > - int i, changes;
> > + int i, changes, ret;
> >
> > - rcar_gen4_pcie_ltssm_enable(rcar, true);
> > + if (rcar->start_link_enable) {
> > + ret = rcar->start_link_enable(rcar);
>
> This looks basically like what qcom does:
>
> qcom_pcie_start_link
> if (pcie->cfg->ops->ltssm_enable)
> pcie->cfg->ops->ltssm_enable(pcie)
>
> Can you copy that and use the same name for the pointer and function
> name (.ltssm_enable, .*_ltssm_enable())?
Yes, I can. I'll rename the pointer and function.
> > + if (ret)
> > + return ret;
> > + }
> >
> > /*
> > * Require direct speed change with retrying here if the link_gen is
> > @@ -437,7 +448,10 @@ static void rcar_gen4_remove_dw_pcie_ep(struct rcar_gen4_pcie *rcar)
> > /* Common */
> > static int rcar_gen4_add_dw_pcie(struct rcar_gen4_pcie *rcar)
> > {
> > - rcar->mode = (uintptr_t)of_device_get_match_data(&rcar->pdev->dev);
> > + const struct rcar_gen4_pcie_platdata *pd = of_device_get_match_data(&rcar->pdev->dev);
> > +
> > + rcar->mode = pd->mode;
> > + rcar->start_link_enable = pd->start_link_enable;
> >
> > switch (rcar->mode) {
> > case DW_PCIE_RC_TYPE:
> > @@ -500,14 +514,47 @@ static void rcar_gen4_pcie_remove(struct platform_device *pdev)
> > rcar_gen4_pcie_unprepare(rcar);
> > }
> >
> > +static int r8a779f0_pcie_start_link_enable(struct rcar_gen4_pcie *rcar)
> > +{
> > + rcar_gen4_pcie_ltssm_enable(rcar, true);
>
> Previously we called rcar_gen4_pcie_ltssm_enable() for
> "renesas,rcar-gen4-pcie" and "renesas,rcar-gen4-pcie-ep". But after
> this patch, it looks like we only call it for "renesas,r8a779f0-pcie"
> and "renesas,r8a779f0-pcie-ep"?
Yes.
Best regards,
Yoshihiro Shimoda
> > +
> > + return 0;
> > +}
> > +
> > +static struct rcar_gen4_pcie_platdata platdata_r8a779f0_pcie = {
> > + .mode = DW_PCIE_RC_TYPE,
> > + .start_link_enable = r8a779f0_pcie_start_link_enable,
> > +};
> > +
> > +static struct rcar_gen4_pcie_platdata platdata_r8a779f0_pcie_ep = {
> > + .mode = DW_PCIE_EP_TYPE,
> > + .start_link_enable = r8a779f0_pcie_start_link_enable,
> > +};
> > +
> > +static struct rcar_gen4_pcie_platdata platdata_rcar_gen4_pcie = {
> > + .mode = DW_PCIE_RC_TYPE,
> > +};
> > +
> > +static struct rcar_gen4_pcie_platdata platdata_rcar_gen4_pcie_ep = {
> > + .mode = DW_PCIE_EP_TYPE,
> > +};
> > +
> > static const struct of_device_id rcar_gen4_pcie_of_match[] = {
> > + {
> > + .compatible = "renesas,r8a779f0-pcie",
> > + .data = &platdata_r8a779f0_pcie,
> > + },
> > + {
> > + .compatible = "renesas,r8a779f0-pcie-ep",
> > + .data = &platdata_r8a779f0_pcie_ep,
> > + },
> > {
> > .compatible = "renesas,rcar-gen4-pcie",
> > - .data = (void *)DW_PCIE_RC_TYPE,
> > + .data = &platdata_rcar_gen4_pcie,
> > },
> > {
> > .compatible = "renesas,rcar-gen4-pcie-ep",
> > - .data = (void *)DW_PCIE_EP_TYPE,
> > + .data = &platdata_rcar_gen4_pcie_ep,
> > },
> > {},
> > };
> > --
> > 2.25.1
> >
