| Message ID | 20220506152107.1527552-1-dmitry.baryshkov@linaro.org |
|---|---|
| Headers | show |
| Series | dt-bindings: YAMLify pci/qcom,pcie schema | expand |
On Fri, May 06, 2022 at 06:20:59PM +0300, Dmitry Baryshkov wrote: > Convert pci/qcom,pcie schema to YAML description. The first patch > introduces several warnings which are fixed by the other patches in the > series. > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > controller uses Synopsys PCIe IP core. However it is not just fused to > the address space. Accessing PCIe registers requires several clocks and > regulators to be powered up. Thus it can be assumed that the qcom,pcie > bindings are not fully compatible with the snps,dw-pcie schema. > > Changes since v5: > - s/stance/stanza (pointed out by Bjorn Helgaas) > > Changes since v4: > - Change subjects to follow convention (suggested by Bjorn Helgaas) > > Changes since v3: > - Rebase on linux-next to include sm8150 patches > > Changes since v2 (still kudos to Krzyshtof): > - Readded reg-names conversion patch > - Mention wake-gpio update in the commit message > - Remove extra quotes in the schema > > Changes since v1 (all kudos to Krzyshtof): > - Dropped the reg-names patch. It will be handled separately > - Squashed the snps,dw-pcie removal (from schema) into the first patch > - Replaced deprecated perst-gpio and wake-gpio with perst-gpios and > wake-gpios in the examples and in DT files > - Moved common clocks/clock-names, resets/reset-names and power-domains > properties to the top level of the schema, leaving only platform > specifics in the conditional branches > - Dropped iommu-map/iommu-map-mask for now > - Added (missed) interrupt-cells, clocks, clock-names, resets, > reset-names properties to the required list (resets/reset-names are > removed in the next patch, as they are not used on msm8996) > - Fixed IRQ flags in the examples > - Merged apq8064/ipq8064 into the single condition statement > - Added extra empty lines > > Dmitry Baryshkov (8): > dt-bindings: PCI: qcom: Convert to YAML > dt-bindings: PCI: qcom: Do not require resets on msm8996 platforms > dt-bindings: PCI: qcom: Specify reg-names explicitly > dt-bindings: PCI: qcom: Add schema for sc7280 chipset > arm64: dts: qcom: stop using snps,dw-pcie falback > arm: dts: qcom: stop using snps,dw-pcie falback > arm: dts: qcom-*: replace deprecated perst-gpio with perst-gpios > arm64: dts: qcom: replace deprecated perst-gpio with perst-gpios > > .../devicetree/bindings/pci/qcom,pcie.txt | 398 ---------- > .../devicetree/bindings/pci/qcom,pcie.yaml | 714 ++++++++++++++++++ What tree do these apply to because they don't apply to rc1. I'm assuming the PCI tree and Lorenzo should take them. Rob
On Tue, 10 May 2022 at 19:50, Rob Herring <robh@kernel.org> wrote: > > On Fri, May 06, 2022 at 06:20:59PM +0300, Dmitry Baryshkov wrote: > > Convert pci/qcom,pcie schema to YAML description. The first patch > > introduces several warnings which are fixed by the other patches in the > > series. > > > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > > controller uses Synopsys PCIe IP core. However it is not just fused to > > the address space. Accessing PCIe registers requires several clocks and > > regulators to be powered up. Thus it can be assumed that the qcom,pcie > > bindings are not fully compatible with the snps,dw-pcie schema. > > > > Changes since v5: > > - s/stance/stanza (pointed out by Bjorn Helgaas) > > > > Changes since v4: > > - Change subjects to follow convention (suggested by Bjorn Helgaas) > > > > Changes since v3: > > - Rebase on linux-next to include sm8150 patches > > > > Changes since v2 (still kudos to Krzyshtof): > > - Readded reg-names conversion patch > > - Mention wake-gpio update in the commit message > > - Remove extra quotes in the schema > > > > Changes since v1 (all kudos to Krzyshtof): > > - Dropped the reg-names patch. It will be handled separately > > - Squashed the snps,dw-pcie removal (from schema) into the first patch > > - Replaced deprecated perst-gpio and wake-gpio with perst-gpios and > > wake-gpios in the examples and in DT files > > - Moved common clocks/clock-names, resets/reset-names and power-domains > > properties to the top level of the schema, leaving only platform > > specifics in the conditional branches > > - Dropped iommu-map/iommu-map-mask for now > > - Added (missed) interrupt-cells, clocks, clock-names, resets, > > reset-names properties to the required list (resets/reset-names are > > removed in the next patch, as they are not used on msm8996) > > - Fixed IRQ flags in the examples > > - Merged apq8064/ipq8064 into the single condition statement > > - Added extra empty lines > > > > Dmitry Baryshkov (8): > > dt-bindings: PCI: qcom: Convert to YAML > > dt-bindings: PCI: qcom: Do not require resets on msm8996 platforms > > dt-bindings: PCI: qcom: Specify reg-names explicitly > > dt-bindings: PCI: qcom: Add schema for sc7280 chipset > > arm64: dts: qcom: stop using snps,dw-pcie falback > > arm: dts: qcom: stop using snps,dw-pcie falback > > arm: dts: qcom-*: replace deprecated perst-gpio with perst-gpios > > arm64: dts: qcom: replace deprecated perst-gpio with perst-gpios > > > > .../devicetree/bindings/pci/qcom,pcie.txt | 398 ---------- > > .../devicetree/bindings/pci/qcom,pcie.yaml | 714 ++++++++++++++++++ > > What tree do these apply to because they don't apply to rc1. I'm > assuming the PCI tree and Lorenzo should take them. The series depends on the patch in Lorenzo's tree (sm8150 bindings), so I'd assume it would be natural to merge these patches through his tree too.
On Wed, May 11, 2022 at 01:13:28PM +0300, Dmitry Baryshkov wrote: > On Tue, 10 May 2022 at 19:50, Rob Herring <robh@kernel.org> wrote: > > > > On Fri, May 06, 2022 at 06:20:59PM +0300, Dmitry Baryshkov wrote: > > > Convert pci/qcom,pcie schema to YAML description. The first patch > > > introduces several warnings which are fixed by the other patches in the > > > series. > > > > > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > > > controller uses Synopsys PCIe IP core. However it is not just fused to > > > the address space. Accessing PCIe registers requires several clocks and > > > regulators to be powered up. Thus it can be assumed that the qcom,pcie > > > bindings are not fully compatible with the snps,dw-pcie schema. > > > > > > Changes since v5: > > > - s/stance/stanza (pointed out by Bjorn Helgaas) > > > > > > Changes since v4: > > > - Change subjects to follow convention (suggested by Bjorn Helgaas) > > > > > > Changes since v3: > > > - Rebase on linux-next to include sm8150 patches > > > > > > Changes since v2 (still kudos to Krzyshtof): > > > - Readded reg-names conversion patch > > > - Mention wake-gpio update in the commit message > > > - Remove extra quotes in the schema > > > > > > Changes since v1 (all kudos to Krzyshtof): > > > - Dropped the reg-names patch. It will be handled separately > > > - Squashed the snps,dw-pcie removal (from schema) into the first patch > > > - Replaced deprecated perst-gpio and wake-gpio with perst-gpios and > > > wake-gpios in the examples and in DT files > > > - Moved common clocks/clock-names, resets/reset-names and power-domains > > > properties to the top level of the schema, leaving only platform > > > specifics in the conditional branches > > > - Dropped iommu-map/iommu-map-mask for now > > > - Added (missed) interrupt-cells, clocks, clock-names, resets, > > > reset-names properties to the required list (resets/reset-names are > > > removed in the next patch, as they are not used on msm8996) > > > - Fixed IRQ flags in the examples > > > - Merged apq8064/ipq8064 into the single condition statement > > > - Added extra empty lines > > > > > > Dmitry Baryshkov (8): > > > dt-bindings: PCI: qcom: Convert to YAML > > > dt-bindings: PCI: qcom: Do not require resets on msm8996 platforms > > > dt-bindings: PCI: qcom: Specify reg-names explicitly > > > dt-bindings: PCI: qcom: Add schema for sc7280 chipset > > > arm64: dts: qcom: stop using snps,dw-pcie falback > > > arm: dts: qcom: stop using snps,dw-pcie falback > > > arm: dts: qcom-*: replace deprecated perst-gpio with perst-gpios > > > arm64: dts: qcom: replace deprecated perst-gpio with perst-gpios > > > > > > .../devicetree/bindings/pci/qcom,pcie.txt | 398 ---------- > > > .../devicetree/bindings/pci/qcom,pcie.yaml | 714 ++++++++++++++++++ > > > > What tree do these apply to because they don't apply to rc1. I'm > > assuming the PCI tree and Lorenzo should take them. > > The series depends on the patch in Lorenzo's tree (sm8150 bindings), > so I'd assume it would be natural to merge these patches through his > tree too. I can take the DT bindings but the dts updates I'd prefer if they went via platform trees. Is that OK ? Thanks, Lorenzo
On Wed, 11 May 2022 at 16:10, Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote: > > On Wed, May 11, 2022 at 01:13:28PM +0300, Dmitry Baryshkov wrote: > > On Tue, 10 May 2022 at 19:50, Rob Herring <robh@kernel.org> wrote: > > > > > > On Fri, May 06, 2022 at 06:20:59PM +0300, Dmitry Baryshkov wrote: > > > > Convert pci/qcom,pcie schema to YAML description. The first patch > > > > introduces several warnings which are fixed by the other patches in the > > > > series. > > > > > > > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > > > > controller uses Synopsys PCIe IP core. However it is not just fused to > > > > the address space. Accessing PCIe registers requires several clocks and > > > > regulators to be powered up. Thus it can be assumed that the qcom,pcie > > > > bindings are not fully compatible with the snps,dw-pcie schema. > > > > > > > > Changes since v5: > > > > - s/stance/stanza (pointed out by Bjorn Helgaas) > > > > > > > > Changes since v4: > > > > - Change subjects to follow convention (suggested by Bjorn Helgaas) > > > > > > > > Changes since v3: > > > > - Rebase on linux-next to include sm8150 patches > > > > > > > > Changes since v2 (still kudos to Krzyshtof): > > > > - Readded reg-names conversion patch > > > > - Mention wake-gpio update in the commit message > > > > - Remove extra quotes in the schema > > > > > > > > Changes since v1 (all kudos to Krzyshtof): > > > > - Dropped the reg-names patch. It will be handled separately > > > > - Squashed the snps,dw-pcie removal (from schema) into the first patch > > > > - Replaced deprecated perst-gpio and wake-gpio with perst-gpios and > > > > wake-gpios in the examples and in DT files > > > > - Moved common clocks/clock-names, resets/reset-names and power-domains > > > > properties to the top level of the schema, leaving only platform > > > > specifics in the conditional branches > > > > - Dropped iommu-map/iommu-map-mask for now > > > > - Added (missed) interrupt-cells, clocks, clock-names, resets, > > > > reset-names properties to the required list (resets/reset-names are > > > > removed in the next patch, as they are not used on msm8996) > > > > - Fixed IRQ flags in the examples > > > > - Merged apq8064/ipq8064 into the single condition statement > > > > - Added extra empty lines > > > > > > > > Dmitry Baryshkov (8): > > > > dt-bindings: PCI: qcom: Convert to YAML > > > > dt-bindings: PCI: qcom: Do not require resets on msm8996 platforms > > > > dt-bindings: PCI: qcom: Specify reg-names explicitly > > > > dt-bindings: PCI: qcom: Add schema for sc7280 chipset > > > > arm64: dts: qcom: stop using snps,dw-pcie falback > > > > arm: dts: qcom: stop using snps,dw-pcie falback > > > > arm: dts: qcom-*: replace deprecated perst-gpio with perst-gpios > > > > arm64: dts: qcom: replace deprecated perst-gpio with perst-gpios > > > > > > > > .../devicetree/bindings/pci/qcom,pcie.txt | 398 ---------- > > > > .../devicetree/bindings/pci/qcom,pcie.yaml | 714 ++++++++++++++++++ > > > > > > What tree do these apply to because they don't apply to rc1. I'm > > > assuming the PCI tree and Lorenzo should take them. > > > > The series depends on the patch in Lorenzo's tree (sm8150 bindings), > > so I'd assume it would be natural to merge these patches through his > > tree too. > > I can take the DT bindings but the dts updates I'd prefer if they > went via platform trees. Is that OK ? Yes, that's fine with me. I think Bjorn has sent 5.19 pull request already, I'll track them getting merged into 5.19 or 5.20. > > Thanks, > Lorenzo
On Wed, May 11, 2022 at 04:53:43PM +0300, Dmitry Baryshkov wrote: > On Wed, 11 May 2022 at 16:10, Lorenzo Pieralisi > <lorenzo.pieralisi@arm.com> wrote: > > > > On Wed, May 11, 2022 at 01:13:28PM +0300, Dmitry Baryshkov wrote: > > > On Tue, 10 May 2022 at 19:50, Rob Herring <robh@kernel.org> wrote: > > > > > > > > On Fri, May 06, 2022 at 06:20:59PM +0300, Dmitry Baryshkov wrote: > > > > > Convert pci/qcom,pcie schema to YAML description. The first patch > > > > > introduces several warnings which are fixed by the other patches in the > > > > > series. > > > > > > > > > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > > > > > controller uses Synopsys PCIe IP core. However it is not just fused to > > > > > the address space. Accessing PCIe registers requires several clocks and > > > > > regulators to be powered up. Thus it can be assumed that the qcom,pcie > > > > > bindings are not fully compatible with the snps,dw-pcie schema. > > > > > > > > > > Changes since v5: > > > > > - s/stance/stanza (pointed out by Bjorn Helgaas) > > > > > > > > > > Changes since v4: > > > > > - Change subjects to follow convention (suggested by Bjorn Helgaas) > > > > > > > > > > Changes since v3: > > > > > - Rebase on linux-next to include sm8150 patches > > > > > > > > > > Changes since v2 (still kudos to Krzyshtof): > > > > > - Readded reg-names conversion patch > > > > > - Mention wake-gpio update in the commit message > > > > > - Remove extra quotes in the schema > > > > > > > > > > Changes since v1 (all kudos to Krzyshtof): > > > > > - Dropped the reg-names patch. It will be handled separately > > > > > - Squashed the snps,dw-pcie removal (from schema) into the first patch > > > > > - Replaced deprecated perst-gpio and wake-gpio with perst-gpios and > > > > > wake-gpios in the examples and in DT files > > > > > - Moved common clocks/clock-names, resets/reset-names and power-domains > > > > > properties to the top level of the schema, leaving only platform > > > > > specifics in the conditional branches > > > > > - Dropped iommu-map/iommu-map-mask for now > > > > > - Added (missed) interrupt-cells, clocks, clock-names, resets, > > > > > reset-names properties to the required list (resets/reset-names are > > > > > removed in the next patch, as they are not used on msm8996) > > > > > - Fixed IRQ flags in the examples > > > > > - Merged apq8064/ipq8064 into the single condition statement > > > > > - Added extra empty lines > > > > > > > > > > Dmitry Baryshkov (8): > > > > > dt-bindings: PCI: qcom: Convert to YAML > > > > > dt-bindings: PCI: qcom: Do not require resets on msm8996 platforms > > > > > dt-bindings: PCI: qcom: Specify reg-names explicitly > > > > > dt-bindings: PCI: qcom: Add schema for sc7280 chipset > > > > > arm64: dts: qcom: stop using snps,dw-pcie falback > > > > > arm: dts: qcom: stop using snps,dw-pcie falback > > > > > arm: dts: qcom-*: replace deprecated perst-gpio with perst-gpios > > > > > arm64: dts: qcom: replace deprecated perst-gpio with perst-gpios > > > > > > > > > > .../devicetree/bindings/pci/qcom,pcie.txt | 398 ---------- > > > > > .../devicetree/bindings/pci/qcom,pcie.yaml | 714 ++++++++++++++++++ > > > > > > > > What tree do these apply to because they don't apply to rc1. I'm > > > > assuming the PCI tree and Lorenzo should take them. > > > > > > The series depends on the patch in Lorenzo's tree (sm8150 bindings), > > > so I'd assume it would be natural to merge these patches through his > > > tree too. > > > > I can take the DT bindings but the dts updates I'd prefer if they > > went via platform trees. Is that OK ? > > Yes, that's fine with me. I think Bjorn has sent 5.19 pull request > already, I'll track them getting merged into 5.19 or 5.20. I assume you meant BjornA sent a PR with patches [5-8] in it, correct ? I will only pull the DT bindings, patches [1-4], please let me know if that's what you expect. Lorenzo
On Wed, 11 May 2022 at 17:00, Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> wrote: > > On Wed, May 11, 2022 at 04:53:43PM +0300, Dmitry Baryshkov wrote: > > On Wed, 11 May 2022 at 16:10, Lorenzo Pieralisi > > <lorenzo.pieralisi@arm.com> wrote: > > > > > > On Wed, May 11, 2022 at 01:13:28PM +0300, Dmitry Baryshkov wrote: > > > > On Tue, 10 May 2022 at 19:50, Rob Herring <robh@kernel.org> wrote: > > > > > > > > > > On Fri, May 06, 2022 at 06:20:59PM +0300, Dmitry Baryshkov wrote: > > > > > > Convert pci/qcom,pcie schema to YAML description. The first patch > > > > > > introduces several warnings which are fixed by the other patches in the > > > > > > series. > > > > > > > > > > > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > > > > > > controller uses Synopsys PCIe IP core. However it is not just fused to > > > > > > the address space. Accessing PCIe registers requires several clocks and > > > > > > regulators to be powered up. Thus it can be assumed that the qcom,pcie > > > > > > bindings are not fully compatible with the snps,dw-pcie schema. > > > > > > > > > > > > Changes since v5: > > > > > > - s/stance/stanza (pointed out by Bjorn Helgaas) > > > > > > > > > > > > Changes since v4: > > > > > > - Change subjects to follow convention (suggested by Bjorn Helgaas) > > > > > > > > > > > > Changes since v3: > > > > > > - Rebase on linux-next to include sm8150 patches > > > > > > > > > > > > Changes since v2 (still kudos to Krzyshtof): > > > > > > - Readded reg-names conversion patch > > > > > > - Mention wake-gpio update in the commit message > > > > > > - Remove extra quotes in the schema > > > > > > > > > > > > Changes since v1 (all kudos to Krzyshtof): > > > > > > - Dropped the reg-names patch. It will be handled separately > > > > > > - Squashed the snps,dw-pcie removal (from schema) into the first patch > > > > > > - Replaced deprecated perst-gpio and wake-gpio with perst-gpios and > > > > > > wake-gpios in the examples and in DT files > > > > > > - Moved common clocks/clock-names, resets/reset-names and power-domains > > > > > > properties to the top level of the schema, leaving only platform > > > > > > specifics in the conditional branches > > > > > > - Dropped iommu-map/iommu-map-mask for now > > > > > > - Added (missed) interrupt-cells, clocks, clock-names, resets, > > > > > > reset-names properties to the required list (resets/reset-names are > > > > > > removed in the next patch, as they are not used on msm8996) > > > > > > - Fixed IRQ flags in the examples > > > > > > - Merged apq8064/ipq8064 into the single condition statement > > > > > > - Added extra empty lines > > > > > > > > > > > > Dmitry Baryshkov (8): > > > > > > dt-bindings: PCI: qcom: Convert to YAML > > > > > > dt-bindings: PCI: qcom: Do not require resets on msm8996 platforms > > > > > > dt-bindings: PCI: qcom: Specify reg-names explicitly > > > > > > dt-bindings: PCI: qcom: Add schema for sc7280 chipset > > > > > > arm64: dts: qcom: stop using snps,dw-pcie falback > > > > > > arm: dts: qcom: stop using snps,dw-pcie falback > > > > > > arm: dts: qcom-*: replace deprecated perst-gpio with perst-gpios > > > > > > arm64: dts: qcom: replace deprecated perst-gpio with perst-gpios > > > > > > > > > > > > .../devicetree/bindings/pci/qcom,pcie.txt | 398 ---------- > > > > > > .../devicetree/bindings/pci/qcom,pcie.yaml | 714 ++++++++++++++++++ > > > > > > > > > > What tree do these apply to because they don't apply to rc1. I'm > > > > > assuming the PCI tree and Lorenzo should take them. > > > > > > > > The series depends on the patch in Lorenzo's tree (sm8150 bindings), > > > > so I'd assume it would be natural to merge these patches through his > > > > tree too. > > > > > > I can take the DT bindings but the dts updates I'd prefer if they > > > went via platform trees. Is that OK ? > > > > Yes, that's fine with me. I think Bjorn has sent 5.19 pull request > > already, I'll track them getting merged into 5.19 or 5.20. > > I assume you meant BjornA sent a PR with patches [5-8] in it, > correct ? > > I will only pull the DT bindings, patches [1-4], please let me > know if that's what you expect. Yes and Yes. > > Lorenzo
On Fri, 6 May 2022 18:20:59 +0300, Dmitry Baryshkov wrote: > Convert pci/qcom,pcie schema to YAML description. The first patch > introduces several warnings which are fixed by the other patches in the > series. > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > controller uses Synopsys PCIe IP core. However it is not just fused to > the address space. Accessing PCIe registers requires several clocks and > regulators to be powered up. Thus it can be assumed that the qcom,pcie > bindings are not fully compatible with the snps,dw-pcie schema. > > [...] Applied to pci/qcom, thanks! [1/8] dt-bindings: PCI: qcom: Convert to YAML https://git.kernel.org/lpieralisi/pci/c/5383d16f06 [2/8] dt-bindings: PCI: qcom: Do not require resets on msm8996 platforms https://git.kernel.org/lpieralisi/pci/c/81dab110d3 [3/8] dt-bindings: PCI: qcom: Specify reg-names explicitly https://git.kernel.org/lpieralisi/pci/c/a6f2d6b1b3 [4/8] dt-bindings: PCI: qcom: Add schema for sc7280 chipset https://git.kernel.org/lpieralisi/pci/c/bddedfeb13 Thanks, Lorenzo
On Fri, 6 May 2022 18:20:59 +0300, Dmitry Baryshkov wrote: > Convert pci/qcom,pcie schema to YAML description. The first patch > introduces several warnings which are fixed by the other patches in the > series. > > Note regarding the snps,dw-pcie compatibility. The Qualcomm PCIe > controller uses Synopsys PCIe IP core. However it is not just fused to > the address space. Accessing PCIe registers requires several clocks and > regulators to be powered up. Thus it can be assumed that the qcom,pcie > bindings are not fully compatible with the snps,dw-pcie schema. > > [...] Applied, thanks! [5/8] arm64: dts: qcom: stop using snps,dw-pcie falback commit: 3e4fec3bc8f8d1366cf2d8d8a054ed37e5a41cba [8/8] arm64: dts: qcom: replace deprecated perst-gpio with perst-gpios commit: f3f5fb3184ec0cd7f98267a8dc1c0538575fbb77 Best regards,