[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240802052206.GA4206@thinkpad>
Date: Fri, 2 Aug 2024 10:52:06 +0530
From: Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
To: Serge Semin <fancer.lancer@...il.com>
Cc: Prudhvi Yarlagadda <quic_pyarlaga@...cinc.com>,
Bjorn Helgaas <helgaas@...nel.org>, jingoohan1@...il.com,
lpieralisi@...nel.org, kw@...ux.com, robh@...nel.org,
bhelgaas@...gle.com, linux-pci@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-arm-msm@...r.kernel.org,
quic_mrana@...cinc.com
Subject: Re: [PATCH v3 1/2] PCI: dwc: Add dbi_phys_addr and atu_phys_addr to
struct dw_pcie
On Fri, Aug 02, 2024 at 12:59:57AM +0300, Serge Semin wrote:
> On Thu, Aug 01, 2024 at 02:29:49PM -0700, Prudhvi Yarlagadda wrote:
> > Hi Serge,
> >
> > Thanks for the review comment.
> >
> > On 8/1/2024 12:25 PM, Serge Semin wrote:
> > > On Tue, Jul 23, 2024 at 07:27:18PM -0700, Prudhvi Yarlagadda wrote:
> > >> Both DBI and ATU physical base addresses are needed by pcie_qcom.c
> > >> driver to program the location of DBI and ATU blocks in Qualcomm
> > >> PCIe Controller specific PARF hardware block.
> > >>
> > >> Signed-off-by: Prudhvi Yarlagadda <quic_pyarlaga@...cinc.com>
> > >> Reviewed-by: Mayank Rana <quic_mrana@...cinc.com>
> > >> ---
> > >> drivers/pci/controller/dwc/pcie-designware.c | 2 ++
> > >> drivers/pci/controller/dwc/pcie-designware.h | 2 ++
> > >> 2 files changed, 4 insertions(+)
> > >>
> > >> diff --git a/drivers/pci/controller/dwc/pcie-designware.c b/drivers/pci/controller/dwc/pcie-designware.c
> > >> index 1b5aba1f0c92..bc3a5d6b0177 100644
> > >> --- a/drivers/pci/controller/dwc/pcie-designware.c
> > >> +++ b/drivers/pci/controller/dwc/pcie-designware.c
> > >> @@ -112,6 +112,7 @@ int dw_pcie_get_resources(struct dw_pcie *pci)
> > >> pci->dbi_base = devm_pci_remap_cfg_resource(pci->dev, res);
> > >> if (IS_ERR(pci->dbi_base))
> > >> return PTR_ERR(pci->dbi_base);
> > >> + pci->dbi_phys_addr = res->start;
> > >> }
> > >>
> > >> /* DBI2 is mainly useful for the endpoint controller */
> > >> @@ -134,6 +135,7 @@ int dw_pcie_get_resources(struct dw_pcie *pci)
> > >> pci->atu_base = devm_ioremap_resource(pci->dev, res);
> > >> if (IS_ERR(pci->atu_base))
> > >> return PTR_ERR(pci->atu_base);
> > >> + pci->atu_phys_addr = res->start;
> > >> } else {
> > >> pci->atu_base = pci->dbi_base + DEFAULT_DBI_ATU_OFFSET;
> > >> }
> > >> diff --git a/drivers/pci/controller/dwc/pcie-designware.h b/drivers/pci/controller/dwc/pcie-designware.h
> > >> index 53c4c8f399c8..efc72989330c 100644
> > >> --- a/drivers/pci/controller/dwc/pcie-designware.h
> > >> +++ b/drivers/pci/controller/dwc/pcie-designware.h
> > >> @@ -407,8 +407,10 @@ struct dw_pcie_ops {
> > >> struct dw_pcie {
> > >> struct device *dev;
> > >> void __iomem *dbi_base;
> > >
> > >> + phys_addr_t dbi_phys_addr;
> > >> void __iomem *dbi_base2;
> > >> void __iomem *atu_base;
> > >> + phys_addr_t atu_phys_addr;
> > >
> > > What's the point in adding these fields to the generic DW PCIe private
> > > data if they are going to be used in the Qcom glue driver only?
> > >
> > > What about moving them to the qcom_pcie structure and initializing the
> > > fields in some place of the pcie-qcom.c driver?
> > >
> > > -Serge(y)
> > >
> >
>
> > These fields were in pcie-qcom.c driver in the v1 patch[1] and
> > Manivannan suggested to move these fields to 'struct dw_pcie' so that duplication
> > of resource fetching code 'platform_get_resource_byname()' can be avoided.
> >
> > [1] https://lore.kernel.org/linux-pci/a01404d2-2f4d-4fb8-af9d-3db66d39acf7@quicinc.com/T/#mf9843386d57e9003de983e24e17de4d54314ff73
>
> Em, polluting the core driver structure with data not being used by
> the core driver but by the glue-code doesn't seem like a better
> alternative to additional platform_get_resource_byname() call in the
> glue-driver. I would have got back v1 version so to keep the core
> driver simpler. Bjorn?
>
IDK how adding two fields which is very related to DWC code *pollutes* it. Since
there is already 'dbi_base', adding 'dbi_phys_addr' made sense to me even though
only glue drivers are using it. Otherwise, glue drivers have to duplicate the
platform_get_resource_byname() code which I find annoying.
- Mani
--
மணிவண்ணன் சதாசிவம்
Powered by blists - more mailing lists