Enable rk356x PCIe controller
v9 cover Enable rk356x PCIe controller
This series enables the DesignWare based PCIe controller on the rk356x series of chips. We drop the fallback to the core driver due to compatibility issues. We reset the PCIe controller at driver probe to prevent issues in the future when firmware / kexec leaves the controller in an unknown state. We add support for legacy interrupts for cards that lack MSI support (which is partially broken currently). We then add the device tree nodes to enable PCIe on the Quartz64 Model A.
Patch 1 drops the snps,dw,pcie fallback from the dt-binding Patch 2 resets the PCIe controller to prevent configuration bugs Patch 3 adds legacy interrupt support to the driver Patch 4 adds the device tree binding to the rk356x.dtsi Patch 5 enables the PCIe controller on the Quartz64-A
Changelog:
v9:
- move reset_control_assert out of rockchip_pcie_resource_get
- fix various formatting mistakes
- fix a checkpatch warning
v8:
- add core reset patch
- simplify IRQ enable/disable functions
- drop spinlock
- only enable/disable IRQ requested
- only pass the IRQ register bits used to irq functions
v7:
- drop assigned-clocks
v6:
- fix a ranges issue
- point to GIC instead of ITS
v5:
- fix incorrect series (apologies for the v4 spam)
v4:
- drop the ITS modification, poor compatibility is better than completely broken
v3:
- drop select node from dt-binding
- convert to for_each_set_bit
- convert to generic_handle_domain_irq
- drop unncessary dev_err
- reorder irq_chip items
- change to level_irq
- install the handler after initializing the domain
v2:
- Define PCIE_CLIENT_INTR_STATUS_LEGACY
- Fix PCIE_LEGACY_INT_ENABLE to only enable the RC interrupts
- Add legacy interrupt enable/disable support
Peter Geis (5):
dt-bindings: PCI: Remove fallback from Rockchip DesignWare binding
PCI: rockchip-dwc: Reset core at driver probe
PCI: rockchip-dwc: Add legacy interrupt support
arm64: dts: rockchip: Add rk3568 PCIe2x1 controller
arm64: dts: rockchip: Enable PCIe controller on quartz64-a
.../bindings/pci/rockchip-dw-pcie.yaml | 12 +- .../boot/dts/rockchip/rk3566-quartz64-a.dts | 34 +++++ arch/arm64/boot/dts/rockchip/rk356x.dtsi | 52 ++++++++ drivers/pci/controller/dwc/pcie-dw-rockchip.c | 119 +++++++++++++++--- 4 files changed, 191 insertions(+), 26 deletions(-)
Signature · 1 line
2.25.1
PATCH v9 2/5 PCI: rockchip-dwc: Reset core at driver probe
The PCIe controller is in an unknown state at driver probe. This can lead to undesireable effects when the driver attempts to configure the controller.
Prevent issues in the future by resetting the core during probe.
--- drivers/pci/controller/dwc/pcie-dw-rockchip.c | 23 ++++++++----------- 1 file changed, 10 insertions(+), 13 deletions(-)
drivers/pci/controller/dwc/pcie-dw-rockchip.c +10 -13
@@ -152,6 +152,11 @@ static int rockchip_pcie_resource_get(struct platform_device *pdev,if (IS_ERR (rockchip->rst_gpio))return PTR_ERR (rockchip->rst_gpio); + rockchip->rst =devm_reset_control_array_get_exclusive (&pdev->dev); +if (IS_ERR (rockchip->rst)) +return dev_err_probe (&pdev->dev,PTR_ERR (rockchip->rst), +"failed to get reset lines\n" ); +return 0 ; }
@@ -182,18 +187,6 @@ static void rockchip_pcie_phy_deinit(struct rockchip_pcie *rockchip)phy_power_off (rockchip->phy); } -static int rockchip_pcie_reset_control_release (struct rockchip_pcie *rockchip) -{ -struct device *dev = rockchip->pci.dev; - - rockchip->rst =devm_reset_control_array_get_exclusive (dev); -if (IS_ERR (rockchip->rst)) -return dev_err_probe (dev,PTR_ERR (rockchip->rst), -"failed to get reset lines\n" ); - -return reset_control_deassert (rockchip->rst); -} -static const struct dw_pcie_ops dw_pcie_ops = { .link_up = rockchip_pcie_link_up, .start_link = rockchip_pcie_start_link,
@@ -222,6 +215,10 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret)return ret; + ret =reset_control_assert (rockchip->rst); +if (ret) +return ret; +/* DON'T MOVE ME: must be enable before PHY init */ rockchip->vpcie3v3 =devm_regulator_get_optional (dev,"vpcie3v3" );if (IS_ERR (rockchip->vpcie3v3)) {
@@ -241,7 +238,7 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret) goto disable_regulator; - ret =rockchip_pcie_reset_control_release (rockchip); + ret =reset_control_deassert (rockchip->rst);if (ret) goto deinit_phy;
Signature · 1 line
2.25.1
PATCH v9 1/5 dt-bindings: PCI: Remove fallback from Rockchip DesignWare binding
In reply to Peter GeisThe snps,dw-pcie binds to a standalone driver. It is not fully compatible with the Rockchip implementation and causes a hang if it binds to the device.
Remove this binding as a valid fallback.
--- .../devicetree/bindings/pci/rockchip-dw-pcie.yaml | 12 +----------- 1 file changed, 1 insertion(+), 11 deletions(-)
Documentation/devicetree/bindings/pci/rockchip-dw-pcie.yaml +1 -11
@@ -19,20 +19,10 @@ description: |+ allOf: - $ref: /schemas/pci/pci-bus.yaml# -# We need a select here so we don't match all nodes with 'snps,dw-pcie' -select: - properties: - compatible: - contains: - const: rockchip,rk3568-pcie - required: - - compatible - properties: compatible: items: - const: rockchip,rk3568-pcie - - const: snps,dw-pcie reg: items:
@@ -110,7 +100,7 @@ examples: #size-cells = <2>; pcie3x2: pcie@fe280000 { - compatible = "rockchip,rk3568-pcie", "snps,dw-pcie"; + compatible = "rockchip,rk3568-pcie"; reg = <0x3 0xc0800000 0x0 0x390000>, <0x0 0xfe280000 0x0 0x10000>, <0x3 0x80000000 0x0 0x100000>;
Signature · 1 line
2.25.1
PATCH v9 3/5 PCI: rockchip-dwc: Add legacy interrupt support
In reply to Peter GeisThe legacy interrupts on the rk356x PCIe controller are handled by a single muxed interrupt. Add IRQ domain support to the pcie-dw-rockchip driver to support the virtual domain.
--- drivers/pci/controller/dwc/pcie-dw-rockchip.c | 96 ++++++++++++++++++- 1 file changed, 94 insertions(+), 2 deletions(-)
drivers/pci/controller/dwc/pcie-dw-rockchip.c +94 -2
@@ -10,9 +10,12 @@#include <linux/clk.h> #include <linux/gpio/consumer.h> +#include <linux/irqchip/chained_irq.h> +#include <linux/irqdomain.h> #include <linux/mfd/syscon.h> #include <linux/module.h> #include <linux/of_device.h> +#include <linux/of_irq.h> #include <linux/phy/phy.h> #include <linux/platform_device.h> #include <linux/regmap.h>
@@ -26,6 +29,7 @@*/ #define HIWORD_UPDATE (mask, val) (((mask) << 16) | (val))#define HIWORD_UPDATE_BIT (val) HIWORD_UPDATE(val, val) +#define HIWORD_DISABLE_BIT (val) HIWORD_UPDATE(val, ~val)#define to_rockchip_pcie (x) dev_get_drvdata((x)->dev)
@@ -36,10 +40,12 @@#define PCIE_LINKUP (PCIE_SMLH_LINKUP | PCIE_RDLH_LINKUP)#define PCIE_L0S_ENTRY 0x11#define PCIE_CLIENT_GENERAL_CONTROL 0x0 +#define PCIE_CLIENT_INTR_STATUS_LEGACY 0x8 +#define PCIE_CLIENT_INTR_MASK_LEGACY 0x1c#define PCIE_CLIENT_GENERAL_DEBUG 0x104 -#define PCIE_CLIENT_HOT_RESET_CTRL 0x180 +#define PCIE_CLIENT_HOT_RESET_CTRL 0x180#define PCIE_CLIENT_LTSSM_STATUS 0x300 -#define PCIE_LTSSM_ENABLE_ENHANCE BIT(4) +#define PCIE_LTSSM_ENABLE_ENHANCE BIT(4)#define PCIE_LTSSM_STATUS_MASK GENMASK(5, 0)struct rockchip_pcie {
@@ -51,6 +57,7 @@ struct rockchip_pcie {struct reset_control *rst;struct gpio_desc *rst_gpio;struct regulator *vpcie3v3; +struct irq_domain *irq_domain; }; static int rockchip_pcie_readl_apb(struct rockchip_pcie *rockchip,
@@ -65,6 +72,78 @@ static void rockchip_pcie_writel_apb(struct rockchip_pcie *rockchip,writel_relaxed (val, rockchip->apb_base + reg); } +static void rockchip_pcie_legacy_int_handler (struct irq_desc *desc) +{ +struct irq_chip *chip =irq_desc_get_chip (desc); +struct rockchip_pcie *rockchip =irq_desc_get_handler_data (desc); +unsigned long reg, hwirq; + +chained_irq_enter (chip, desc); + + reg =rockchip_pcie_readl_apb (rockchip,PCIE_CLIENT_INTR_STATUS_LEGACY ); + +for_each_set_bit (hwirq, ®,4 ) +generic_handle_domain_irq (rockchip->irq_domain, hwirq); + +chained_irq_exit (chip, desc); +} + +static void rockchip_intx_mask (struct irq_data *data) +{ +rockchip_pcie_writel_apb (irq_data_get_irq_chip_data (data), +HIWORD_UPDATE_BIT (BIT (data->hwirq)), +PCIE_CLIENT_INTR_MASK_LEGACY ); +}; + +static void rockchip_intx_unmask (struct irq_data *data) +{ +rockchip_pcie_writel_apb (irq_data_get_irq_chip_data (data), +HIWORD_DISABLE_BIT (BIT (data->hwirq)), +PCIE_CLIENT_INTR_MASK_LEGACY ); +}; + +static struct irq_chip rockchip_intx_irq_chip = { + .name ="INTx" , + .irq_mask = rockchip_intx_mask, + .irq_unmask = rockchip_intx_unmask, + .flags =IRQCHIP_SKIP_SET_WAKE |IRQCHIP_MASK_ON_SUSPEND , +}; + +static int rockchip_pcie_intx_map (struct irq_domain *domain,unsigned int irq, +irq_hw_number_t hwirq) +{ +irq_set_chip_and_handler (irq, &rockchip_intx_irq_chip, handle_level_irq); +irq_set_chip_data (irq, domain->host_data); + +return 0 ; +} + +static const struct irq_domain_ops intx_domain_ops = { + .map = rockchip_pcie_intx_map, +}; + +static int rockchip_pcie_init_irq_domain (struct rockchip_pcie *rockchip) +{ +struct device *dev = rockchip->pci.dev; +struct device_node *intc; + + intc =of_get_child_by_name (dev->of_node,"legacy-interrupt-controller" ); +if (!intc) { +dev_err (dev,"missing child interrupt-controller node\n" ); +return -EINVAL ; + } + + rockchip->irq_domain =irq_domain_add_linear (intc,PCI_NUM_INTX , + &intx_domain_ops, rockchip); +of_node_put (intc); +if (!rockchip->irq_domain) { +dev_err (dev,"failed to get a INTx IRQ domain\n" ); +return -EINVAL ; + } + +return 0 ; +} +static void rockchip_pcie_enable_ltssm (struct rockchip_pcie *rockchip) { rockchip_pcie_writel_apb(rockchip, PCIE_CLIENT_ENABLE_LTSSM,
@@ -111,7 +190,20 @@ static int rockchip_pcie_host_init(struct pcie_port *pp) {struct dw_pcie *pci =to_dw_pcie_from_pp (pp);struct rockchip_pcie *rockchip =to_rockchip_pcie (pci); +struct device *dev = rockchip->pci.dev;u32 val =HIWORD_UPDATE_BIT (PCIE_LTSSM_ENABLE_ENHANCE ); +int irq, ret; + + irq =of_irq_get_byname (dev->of_node,"legacy" ); +if (irq <0 ) +return irq; + + ret =rockchip_pcie_init_irq_domain (rockchip); +if (ret <0 ) +dev_err (dev,"failed to init irq domain\n" ); + +irq_set_chained_handler_and_data (irq, rockchip_pcie_legacy_int_handler, + rockchip);/* LTSSM enable control mode */ rockchip_pcie_writel_apb (rockchip, val,PCIE_CLIENT_HOT_RESET_CTRL );
Signature · 1 line
2.25.1
PATCH v9 4/5 arm64: dts: rockchip: Add rk3568 PCIe2x1 controller
In reply to Peter GeisThe PCIe2x1 controller is common between the rk3568 and rk3566. It is a single lane PCIe2 compliant controller.
--- arch/arm64/boot/dts/rockchip/rk356x.dtsi | 52 ++++++++++++++++++++++++ 1 file changed, 52 insertions(+)
arch/arm64/boot/dts/rockchip/rk356x.dtsi +52 -0
@@ -689,6 +689,58 @@ qos_vop_m1: qos@fe1a8100 { reg = <0x0 0xfe1a8100 0x0 0x20>; }; + pcie2x1: pcie@fe260000 { + compatible = "rockchip,rk3568-pcie"; + #address-cells = <3>; + #size-cells = <2>; + bus-range = <0x0 0xf>; + clocks = <&cru ACLK_PCIE20_MST>, <&cru ACLK_PCIE20_SLV>, + <&cru ACLK_PCIE20_DBI>, <&cru PCLK_PCIE20>, + <&cru CLK_PCIE20_AUX_NDFT>; + clock-names = "aclk_mst", "aclk_slv", + "aclk_dbi", "pclk", "aux"; + device_type = "pci"; + interrupts = <GIC_SPI 75 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 74 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 73 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 72 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 71 IRQ_TYPE_LEVEL_HIGH>; + interrupt-names = "sys", "pmc", "msi", "legacy", "err"; + #interrupt-cells = <1>; + interrupt-map-mask = <0 0 0 7>; + interrupt-map = <0 0 0 1 &pcie_intc 0>, + <0 0 0 2 &pcie_intc 1>, + <0 0 0 3 &pcie_intc 2>, + <0 0 0 4 &pcie_intc 3>; + linux,pci-domain = <0>; + num-ib-windows = <6>; + num-ob-windows = <2>; + max-link-speed = <2>; + msi-map = <0x0 &gic 0x0 0x1000>; + num-lanes = <1>; + phys = <&combphy2 PHY_TYPE_PCIE>; + phy-names = "pcie-phy"; + power-domains = <&power RK3568_PD_PIPE>; + reg = <0x3 0xc0000000 0x0 0x00400000>, + <0x0 0xfe260000 0x0 0x00010000>, + <0x3 0x00000000 0x0 0x01000000>; + ranges = <0x01000000 0x0 0x01000000 0x3 0x01000000 0x0 0x00100000 + 0x02000000 0x0 0x02000000 0x3 0x01100000 0x0 0x3ef00000>; + reg-names = "dbi", "apb", "config"; + resets = <&cru SRST_PCIE20_POWERUP>; + reset-names = "pipe"; + status = "disabled"; + + pcie_intc: legacy-interrupt-controller { + #address-cells = <0>; + #interrupt-cells = <1>; + interrupt-controller; + interrupt-parent = <&gic>; + interrupts = <GIC_SPI 72 IRQ_TYPE_EDGE_RISING>; + }; + + }; + sdmmc0: mmc@fe2b0000 { compatible = "rockchip,rk3568-dw-mshc", "rockchip,rk3288-dw-mshc"; reg = <0x0 0xfe2b0000 0x0 0x4000>;
Signature · 1 line
2.25.1
PATCH v9 5/5 arm64: dts: rockchip: Enable PCIe controller on quartz64-a
In reply to Peter GeisAdd the nodes to enable the PCIe controller on the Quartz64 Model A board.
--- .../boot/dts/rockchip/rk3566-quartz64-a.dts | 34 +++++++++++++++++++ 1 file changed, 34 insertions(+)
arch/arm64/boot/dts/rockchip/rk3566-quartz64-a.dts +34 -0
@@ -125,6 +125,18 @@ vbus: vbus { vin-supply = <&vcc12v_dcin>; }; + vcc3v3_pcie_p: vcc3v3_pcie_p { + compatible = "regulator-fixed"; + enable-active-high; + gpio = <&gpio0 RK_PC6 GPIO_ACTIVE_HIGH>; + pinctrl-names = "default"; + pinctrl-0 = <&pcie_enable_h>; + regulator-name = "vcc3v3_pcie_p"; + regulator-min-microvolt = <3300000>; + regulator-max-microvolt = <3300000>; + vin-supply = <&vcc_3v3>; + }; + vcc5v0_usb: vcc5v0_usb { compatible = "regulator-fixed"; regulator-name = "vcc5v0_usb";
@@ -187,6 +199,10 @@ vcc_wl: vcc_wl { }; }; +&combphy2 { + status = "okay"; +}; + &cpu0 { cpu-supply = <&vdd_cpu>; };
@@ -495,6 +511,14 @@ rgmii_phy1: ethernet-phy@0 { }; }; +&pcie2x1 { + pinctrl-names = "default"; + pinctrl-0 = <&pcie_reset_h>; + reset-gpios = <&gpio1 RK_PB2 GPIO_ACTIVE_HIGH>; + vpcie3v3-supply = <&vcc3v3_pcie_p>; + status = "okay"; +}; + &pinctrl { bt { bt_enable_h: bt-enable-h {
@@ -520,6 +544,16 @@ diy_led_enable_h: diy-led-enable-h { }; }; + pcie { + pcie_enable_h: pcie-enable-h { + rockchip,pins = <0 RK_PC6 RK_FUNC_GPIO &pcfg_pull_none>; + }; + + pcie_reset_h: pcie-reset-h { + rockchip,pins = <1 RK_PB2 RK_FUNC_GPIO &pcfg_pull_none>; + }; + }; + pmic { pmic_int_l: pmic-int-l { rockchip,pins = <0 RK_PA3 RK_FUNC_GPIO &pcfg_pull_up>;
Signature · 1 line
2.25.1
Re: [PATCH v9 3/5] PCI: rockchip-dwc: Add legacy interrupt support
In reply to Peter GeisQuoted 6 lines · Peter Geis
The legacy interrupts on the rk356x PCIe controller are handled by a single muxed interrupt. Add IRQ domain support to the pcie-dw-rockchip driver to support the virtual domain.
Signed-off-by Peter Geis <[email protected]>
M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
Quoted 8 lines · Peter Geis
This series enables the DesignWare based PCIe controller on the rk356x series of chips. We drop the fallback to the core driver due to compatibility issues. We reset the PCIe controller at driver probe to prevent issues in the future when firmware / kexec leaves the controller in an unknown state. We add support for legacy interrupts for cards that lack MSI support (which is partially broken currently). We then add the device tree nodes to enable PCIe on the Quartz64 Model A.
Good Evening,
Just a gentle ping to see if there's anything outstanding here.
Very Respectfully,
Peter Geis
Quoted 65 lines
Patch 1 drops the snps,dw,pcie fallback from the dt-binding Patch 2 resets the PCIe controller to prevent configuration bugs Patch 3 adds legacy interrupt support to the driver Patch 4 adds the device tree binding to the
rk356x.dtsiPatch 5 enables the PCIe controller on the Quartz64-AChangelog:
v9:
- movereset_control_assertout ofrockchip_pcie_resource_get
- fix various formatting mistakes
- fix a checkpatch warningv8:
- add core reset patch
- simplify IRQ enable/disable functions
- drop spinlock
- only enable/disable IRQ requested
- only pass the IRQ register bits used to irq functionsv7:
- drop assigned-clocksv6:
- fix a ranges issue
- point to GIC instead of ITSv5:
- fix incorrect series (apologies for the v4 spam)v4:
- drop the ITS modification, poor compatibility is better than completely brokenv3:
- drop select node from dt-binding
- convert tofor_each_set_bit
- convert togeneric_handle_domain_irq
- drop unncessarydev_err
- reorderirq_chipitems
- change tolevel_irq
- install the handler after initializing the domainv2:
- DefinePCIE_CLIENT_INTR_STATUS_LEGACY
- FixPCIE_LEGACY_INT_ENABLEto only enable the RC interrupts
- Add legacy interrupt enable/disable supportPeter Geis (5):
dt-bindings: PCI: Remove fallback from Rockchip DesignWare binding
PCI: rockchip-dwc: Reset core at driver probe
PCI: rockchip-dwc: Add legacy interrupt support
arm64: dts: rockchip: Add rk3568 PCIe2x1 controller
arm64: dts: rockchip: Enable PCIe controller on quartz64-a.../bindings/pci/rockchip-dw-pcie.yaml | 12 +- .../boot/dts/rockchip/rk3566-quartz64-a.dts | 34 +++++ arch/arm64/boot/dts/rockchip/rk356x.dtsi | 52 ++++++++ drivers/pci/controller/dwc/pcie-dw-rockchip.c | 119 +++++++++++++++--- 4 files changed, 191 insertions(+), 26 deletions(-)--
2.25.1
Heiko Stübner
maintainerAm Dienstag, 10. Mai 2022, 23:11:18 CEST schrieb Peter Geis:
Quoted 10 lines · Peter Geis
Quoted 8 lines
This series enables the DesignWare based PCIe controller on the rk356x series of chips. We drop the fallback to the core driver due to compatibility issues. We reset the PCIe controller at driver probe to prevent issues in the future when firmware / kexec leaves the controller in an unknown state. We add support for legacy interrupts for cards that lack MSI support (which is partially broken currently). We then add the device tree nodes to enable PCIe on the Quartz64 Model A.
Quoted 3 lines
Good Evening,
Just a gentle ping to see if there's anything outstanding here.
From my side it looks good. I'll take patches 4+5 once the binding-change from patch1 has been applied to some tree.
Heiko
Quoted 66 lines
Quoted 65 lines
Patch 1 drops the snps,dw,pcie fallback from the dt-binding Patch 2 resets the PCIe controller to prevent configuration bugs Patch 3 adds legacy interrupt support to the driver Patch 4 adds the device tree binding to the
rk356x.dtsiPatch 5 enables the PCIe controller on the Quartz64-AChangelog:
v9:
- movereset_control_assertout ofrockchip_pcie_resource_get
- fix various formatting mistakes
- fix a checkpatch warningv8:
- add core reset patch
- simplify IRQ enable/disable functions
- drop spinlock
- only enable/disable IRQ requested
- only pass the IRQ register bits used to irq functionsv7:
- drop assigned-clocksv6:
- fix a ranges issue
- point to GIC instead of ITSv5:
- fix incorrect series (apologies for the v4 spam)v4:
- drop the ITS modification, poor compatibility is better than completely brokenv3:
- drop select node from dt-binding
- convert tofor_each_set_bit
- convert togeneric_handle_domain_irq
- drop unncessarydev_err
- reorderirq_chipitems
- change tolevel_irq
- install the handler after initializing the domainv2:
- DefinePCIE_CLIENT_INTR_STATUS_LEGACY
- FixPCIE_LEGACY_INT_ENABLEto only enable the RC interrupts
- Add legacy interrupt enable/disable supportPeter Geis (5):
dt-bindings: PCI: Remove fallback from Rockchip DesignWare binding
PCI: rockchip-dwc: Reset core at driver probe
PCI: rockchip-dwc: Add legacy interrupt support
arm64: dts: rockchip: Add rk3568 PCIe2x1 controller
arm64: dts: rockchip: Enable PCIe controller on quartz64-a.../bindings/pci/rockchip-dw-pcie.yaml | 12 +- .../boot/dts/rockchip/rk3566-quartz64-a.dts | 34 +++++ arch/arm64/boot/dts/rockchip/rk356x.dtsi | 52 ++++++++ drivers/pci/controller/dwc/pcie-dw-rockchip.c | 119 +++++++++++++++--- 4 files changed, 191 insertions(+), 26 deletions(-)--
2.25.1
Re: [PATCH v9 2/5] PCI: rockchip-dwc: Reset core at driver probe
In reply to Peter GeisQuoted 11 lines · Peter Geis
The PCIe controller is in an unknown state at driver probe. This can lead to undesireable effects when the driver attempts to configure the controller.
Prevent issues in the future by resetting the core during probe.
Signed-off-by Peter Geis <[email protected]>Tested-by Nicolas Frattaroli <[email protected]>--- drivers/pci/controller/dwc/pcie-dw-rockchip.c | 23 ++++++++----------- 1 file changed, 10 insertions(+), 13 deletions(-)
I fear that the controller reset behaviour is bootloader/firmware dependent.
Are we sure we are not triggering any regressions by resetting the controller in the middle of probe (aka is the driver implicitly relying on existing behaviour on systems that are not the ones you are testing on) ?
Just asking, the rockchip maintainers should be able to answer this question.
Thanks,
Lorenzo
Quoted 58 lines
drivers/pci/controller/dwc/pcie-dw-rockchip.c +10
-13@@ -152,6 +152,11 @@ static int rockchip_pcie_resource_get(struct platform_device *pdev,if (IS_ERR (rockchip->rst_gpio))return PTR_ERR (rockchip->rst_gpio); + rockchip->rst =devm_reset_control_array_get_exclusive (&pdev->dev); +if (IS_ERR (rockchip->rst)) +return dev_err_probe (&pdev->dev,PTR_ERR (rockchip->rst), +"failed to get reset lines\n" ); +return 0 ; }@@ -182,18 +187,6 @@ static void rockchip_pcie_phy_deinit(struct rockchip_pcie *rockchip)phy_power_off (rockchip->phy); } -static int rockchip_pcie_reset_control_release (struct rockchip_pcie *rockchip) -{ -struct device *dev = rockchip->pci.dev; - - rockchip->rst =devm_reset_control_array_get_exclusive (dev); -if (IS_ERR (rockchip->rst)) -return dev_err_probe (dev,PTR_ERR (rockchip->rst), -"failed to get reset lines\n" ); - -return reset_control_deassert (rockchip->rst); -} -static const struct dw_pcie_ops dw_pcie_ops = { .link_up = rockchip_pcie_link_up, .start_link = rockchip_pcie_start_link,@@ -222,6 +215,10 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret)return ret; + ret =reset_control_assert (rockchip->rst); +if (ret) +return ret; +/* DON'T MOVE ME: must be enable before PHY init */ rockchip->vpcie3v3 =devm_regulator_get_optional (dev,"vpcie3v3" );if (IS_ERR (rockchip->vpcie3v3)) {@@ -241,7 +238,7 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret) goto disable_regulator; - ret =rockchip_pcie_reset_control_release (rockchip); + ret =reset_control_deassert (rockchip->rst);if (ret) goto deinit_phy;Signature · 1 line
2.25.1
Re: [PATCH v9 2/5] PCI: rockchip-dwc: Reset core at driver probe
Quoted 17 lines · Lorenzo Pieralisi
Quoted 11 lines · Peter Geis
The PCIe controller is in an unknown state at driver probe. This can lead to undesireable effects when the driver attempts to configure the controller.
Prevent issues in the future by resetting the core during probe.
Signed-off-by Peter Geis <[email protected]>Tested-by Nicolas Frattaroli <[email protected]>--- drivers/pci/controller/dwc/pcie-dw-rockchip.c | 23 ++++++++----------- 1 file changed, 10 insertions(+), 13 deletions(-)I fear that the controller reset behaviour is bootloader/firmware dependent.
Quoted 7 lines
Are we sure we are not triggering any regressions by resetting the controller in the middle of probe (aka is the driver implicitly relying on existing behaviour on systems that are not the ones you are testing on) ?
Just asking, the rockchip maintainers should be able to answer this question.
This is a new driver with no current users, this series enables the first user. It does not support ACPI nor any sort of handoff at this time.
Quoted 62 lines
Thanks,
LorenzoQuoted 58 lines
drivers/pci/controller/dwc/pcie-dw-rockchip.c +10
-14@@ -152,6 +152,11 @@ static int rockchip_pcie_resource_get(struct platform_device *pdev,if (IS_ERR (rockchip->rst_gpio))return PTR_ERR (rockchip->rst_gpio); + rockchip->rst =devm_reset_control_array_get_exclusive (&pdev->dev); +if (IS_ERR (rockchip->rst)) +return dev_err_probe (&pdev->dev,PTR_ERR (rockchip->rst), +"failed to get reset lines\n" ); +return 0 ; }@@ -182,18 +187,6 @@ static void rockchip_pcie_phy_deinit(struct rockchip_pcie *rockchip)phy_power_off (rockchip->phy); } -static int rockchip_pcie_reset_control_release (struct rockchip_pcie *rockchip) -{ -struct device *dev = rockchip->pci.dev; - - rockchip->rst =devm_reset_control_array_get_exclusive (dev); -if (IS_ERR (rockchip->rst)) -return dev_err_probe (dev,PTR_ERR (rockchip->rst), -"failed to get reset lines\n" ); - -return reset_control_deassert (rockchip->rst); -} -static const struct dw_pcie_ops dw_pcie_ops = { .link_up = rockchip_pcie_link_up, .start_link = rockchip_pcie_start_link,@@ -222,6 +215,10 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret)return ret; + ret =reset_control_assert (rockchip->rst); +if (ret) +return ret; +/* DON'T MOVE ME: must be enable before PHY init */ rockchip->vpcie3v3 =devm_regulator_get_optional (dev,"vpcie3v3" );if (IS_ERR (rockchip->vpcie3v3)) {@@ -241,7 +238,7 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret) goto disable_regulator; - ret =rockchip_pcie_reset_control_release (rockchip); + ret =reset_control_deassert (rockchip->rst);if (ret) goto deinit_phy; --2.25.1
Re: [PATCH v9 2/5] PCI: rockchip-dwc: Reset core at driver probe
Quoted 24 lines · Peter Geis
Quoted 22 lines · Lorenzo Pieralisi
Quoted 11 lines · Peter Geis
The PCIe controller is in an unknown state at driver probe. This can lead to undesireable effects when the driver attempts to configure the controller.
Prevent issues in the future by resetting the core during probe.
Signed-off-by Peter Geis <[email protected]>Tested-by Nicolas Frattaroli <[email protected]>--- drivers/pci/controller/dwc/pcie-dw-rockchip.c | 23 ++++++++----------- 1 file changed, 10 insertions(+), 13 deletions(-)I fear that the controller reset behaviour is bootloader/firmware dependent.
Are we sure we are not triggering any regressions by resetting the controller in the middle of probe (aka is the driver implicitly relying on existing behaviour on systems that are not the ones you are testing on) ?
Quoted 6 lines
Quoted 2 lines
Just asking, the rockchip maintainers should be able to answer this question.
This is a new driver with no current users, this series enables the first user. It does not support ACPI nor any sort of handoff at this time.
Ok. I will take patches [1-3], dts changes will have to go via platform trees, I hope that's fine.
Thanks,
Lorenzo
Quoted 61 lines
Quoted 61 lines
Thanks,
LorenzoQuoted 58 lines
drivers/pci/controller/dwc/pcie-dw-rockchip.c +10
-14@@ -152,6 +152,11 @@ static int rockchip_pcie_resource_get(struct platform_device *pdev,if (IS_ERR (rockchip->rst_gpio))return PTR_ERR (rockchip->rst_gpio); + rockchip->rst =devm_reset_control_array_get_exclusive (&pdev->dev); +if (IS_ERR (rockchip->rst)) +return dev_err_probe (&pdev->dev,PTR_ERR (rockchip->rst), +"failed to get reset lines\n" ); +return 0 ; }@@ -182,18 +187,6 @@ static void rockchip_pcie_phy_deinit(struct rockchip_pcie *rockchip)phy_power_off (rockchip->phy); } -static int rockchip_pcie_reset_control_release (struct rockchip_pcie *rockchip) -{ -struct device *dev = rockchip->pci.dev; - - rockchip->rst =devm_reset_control_array_get_exclusive (dev); -if (IS_ERR (rockchip->rst)) -return dev_err_probe (dev,PTR_ERR (rockchip->rst), -"failed to get reset lines\n" ); - -return reset_control_deassert (rockchip->rst); -} -static const struct dw_pcie_ops dw_pcie_ops = { .link_up = rockchip_pcie_link_up, .start_link = rockchip_pcie_start_link,@@ -222,6 +215,10 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret)return ret; + ret =reset_control_assert (rockchip->rst); +if (ret) +return ret; +/* DON'T MOVE ME: must be enable before PHY init */ rockchip->vpcie3v3 =devm_regulator_get_optional (dev,"vpcie3v3" );if (IS_ERR (rockchip->vpcie3v3)) {@@ -241,7 +238,7 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret) goto disable_regulator; - ret =rockchip_pcie_reset_control_release (rockchip); + ret =reset_control_deassert (rockchip->rst);if (ret) goto deinit_phy; --2.25.1
Lorenzo Pieralisi
AppliedRe: (subset) [PATCH v9 0/5] Enable rk356x PCIe controller
In reply to Peter GeisQuoted 9 lines · Peter Geis
This series enables the DesignWare based PCIe controller on the rk356x series of chips. We drop the fallback to the core driver due to compatibility issues. We reset the PCIe controller at driver probe to prevent issues in the future when firmware / kexec leaves the controller in an unknown state. We add support for legacy interrupts for cards that lack MSI support (which is partially broken currently). We then add the device tree nodes to enable PCIe on the Quartz64 Model A.
[...]
Applied to pci/dwc, thanks!
[1/5] dt-bindings: PCI: Remove fallback from Rockchip DesignWare binding
https://git.kernel.org/lpieralisi/pci/c/931262e646 [2/5] PCI: rockchip-dwc: Reset core at driver probe https://git.kernel.org/lpieralisi/pci/c/431e7d2eec [3/5] PCI: rockchip-dwc: Add legacy interrupt support https://git.kernel.org/lpieralisi/pci/c/e8aae154df
Thanks,
Lorenzo
Heiko Stübner
maintainerRe: [PATCH v9 2/5] PCI: rockchip-dwc: Reset core at driver probe
In reply to Lorenzo PieralisiAm Mittwoch, 11. Mai 2022, 17:00:05 CEST schrieb Lorenzo Pieralisi:
Quoted 28 lines · Peter Geis
Quoted 27 lines
Quoted 24 lines · Lorenzo Pieralisi
Quoted 11 lines · Peter Geis
The PCIe controller is in an unknown state at driver probe. This can lead to undesireable effects when the driver attempts to configure the controller.
Prevent issues in the future by resetting the core during probe.
Signed-off-by Peter Geis <[email protected]>Tested-by Nicolas Frattaroli <[email protected]>--- drivers/pci/controller/dwc/pcie-dw-rockchip.c | 23 ++++++++----------- 1 file changed, 10 insertions(+), 13 deletions(-)I fear that the controller reset behaviour is bootloader/firmware dependent.
Are we sure we are not triggering any regressions by resetting the controller in the middle of probe (aka is the driver implicitly relying on existing behaviour on systems that are not the ones you are testing on) ?
Just asking, the rockchip maintainers should be able to answer this question.
Quoted 6 lines
Quoted 3 lines
This is a new driver with no current users, this series enables the first user. It does not support ACPI nor any sort of handoff at this time.
Ok. I will take patches [1-3], dts changes will have to go via platform trees, I hope that's fine.
yep, that sounds great and I'll pick the dts patches then :-)
Thanks
Heiko
Quoted 59 lines
Quoted 58 lines
Quoted 58 lines
Quoted 58 lines
drivers/pci/controller/dwc/pcie-dw-rockchip.c +10
-14@@ -152,6 +152,11 @@ static int rockchip_pcie_resource_get(struct platform_device *pdev,if (IS_ERR (rockchip->rst_gpio))return PTR_ERR (rockchip->rst_gpio); + rockchip->rst =devm_reset_control_array_get_exclusive (&pdev->dev); +if (IS_ERR (rockchip->rst)) +return dev_err_probe (&pdev->dev,PTR_ERR (rockchip->rst), +"failed to get reset lines\n" ); +return 0 ; }@@ -182,18 +187,6 @@ static void rockchip_pcie_phy_deinit(struct rockchip_pcie *rockchip)phy_power_off (rockchip->phy); } -static int rockchip_pcie_reset_control_release (struct rockchip_pcie *rockchip) -{ -struct device *dev = rockchip->pci.dev; - - rockchip->rst =devm_reset_control_array_get_exclusive (dev); -if (IS_ERR (rockchip->rst)) -return dev_err_probe (dev,PTR_ERR (rockchip->rst), -"failed to get reset lines\n" ); - -return reset_control_deassert (rockchip->rst); -} -static const struct dw_pcie_ops dw_pcie_ops = { .link_up = rockchip_pcie_link_up, .start_link = rockchip_pcie_start_link,@@ -222,6 +215,10 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret)return ret; + ret =reset_control_assert (rockchip->rst); +if (ret) +return ret; +/* DON'T MOVE ME: must be enable before PHY init */ rockchip->vpcie3v3 =devm_regulator_get_optional (dev,"vpcie3v3" );if (IS_ERR (rockchip->vpcie3v3)) {@@ -241,7 +238,7 @@ static int rockchip_pcie_probe(struct platform_device *pdev)if (ret) goto disable_regulator; - ret =rockchip_pcie_reset_control_release (rockchip); + ret =reset_control_deassert (rockchip->rst);if (ret) goto deinit_phy; --2.25.1
Heiko Stübner
maintainerRe: [PATCH v9 4/5] arm64: dts: rockchip: Add rk3568 PCIe2x1 controller
In reply to Peter GeisAm Freitag, 29. April 2022, 14:38:30 CEST schrieb Peter Geis:
Quoted 24 lines
The PCIe2x1 controller is common between the rk3568 and rk3566. It is a single lane PCIe2 compliant controller.
Signed-off-by Peter Geis <[email protected]>--- arch/arm64/boot/dts/rockchip/rk356x.dtsi | 52 ++++++++++++++++++++++++ 1 file changed, 52 insertions(+)arch/arm64/boot/dts/rockchip/rk356x.dtsi +8
-0@@ -689,6 +689,58 @@ qos_vop_m1: qos@fe1a8100 { reg = <0x0 0xfe1a8100 0x0 0x20>; }; + pcie2x1: pcie@fe260000 { + compatible = "rockchip,rk3568-pcie"; + #address-cells = <3>; + #size-cells = <2>; + bus-range = <0x0 0xf>; + clocks = <&cru ACLK_PCIE20_MST>, <&cru ACLK_PCIE20_SLV>, + <&cru ACLK_PCIE20_DBI>, <&cru PCLK_PCIE20>, + <&cru CLK_PCIE20_AUX_NDFT>;
arch/arm64/boot/dts/rockchip/rk356x.dtsi qos_vop_m1: qos@fe1a8100 { · line 709
Quoted 10 lines
arch/arm64/boot/dts/rockchip/rk356x.dtsi +10
-0+ clock-names = "aclk_mst", "aclk_slv", + "aclk_dbi", "pclk", "aux"; + device_type = "pci"; + interrupts = <GIC_SPI 75 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 74 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 73 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 72 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 71 IRQ_TYPE_LEVEL_HIGH>; + interrupt-names = "sys", "pmc", "msi", "legacy", "err"; + #interrupt-cells = <1>;
I guess #interrupt-cells shouldn't be necessary here, as that property is meant for interrupt-controller nodes - like the subnode here which already has its own #interrupt-cells, right?
Quoted 38 lines
arch/arm64/boot/dts/rockchip/rk356x.dtsi +34
-0+ interrupt-map-mask = <0 0 0 7>; + interrupt-map = <0 0 0 1 &pcie_intc 0>, + <0 0 0 2 &pcie_intc 1>, + <0 0 0 3 &pcie_intc 2>, + <0 0 0 4 &pcie_intc 3>; + linux,pci-domain = <0>; + num-ib-windows = <6>; + num-ob-windows = <2>; + max-link-speed = <2>; + msi-map = <0x0 &gic 0x0 0x1000>; + num-lanes = <1>; + phys = <&combphy2 PHY_TYPE_PCIE>; + phy-names = "pcie-phy"; + power-domains = <&power RK3568_PD_PIPE>; + reg = <0x3 0xc0000000 0x0 0x00400000>, + <0x0 0xfe260000 0x0 0x00010000>, + <0x3 0x00000000 0x0 0x01000000>; + ranges = <0x01000000 0x0 0x01000000 0x3 0x01000000 0x0 0x00100000 + 0x02000000 0x0 0x02000000 0x3 0x01100000 0x0 0x3ef00000>; + reg-names = "dbi", "apb", "config"; + resets = <&cru SRST_PCIE20_POWERUP>; + reset-names = "pipe"; + status = "disabled"; + + pcie_intc: legacy-interrupt-controller { + #address-cells = <0>; + #interrupt-cells = <1>; + interrupt-controller; + interrupt-parent = <&gic>; + interrupts = <GIC_SPI 72 IRQ_TYPE_EDGE_RISING>; + }; + + }; + sdmmc0: mmc@fe2b0000 { compatible = "rockchip,rk3568-dw-mshc", "rockchip,rk3288-dw-mshc"; reg = <0x0 0xfe2b0000 0x0 0x4000>;
Re: (subset) [PATCH v9 0/5] Enable rk356x PCIe controller
In reply to Peter GeisQuoted 9 lines · Peter Geis
This series enables the DesignWare based PCIe controller on the rk356x series of chips. We drop the fallback to the core driver due to compatibility issues. We reset the PCIe controller at driver probe to prevent issues in the future when firmware / kexec leaves the controller in an unknown state. We add support for legacy interrupts for cards that lack MSI support (which is partially broken currently). We then add the device tree nodes to enable PCIe on the Quartz64 Model A.
[...]
Applied, thanks!
[4/5] arm64: dts: rockchip: Add rk3568 PCIe2x1 controller
commit: c9168492af55bdbc811e05bfc55ae70880bf8ff3
[5/5] arm64: dts: rockchip: Enable PCIe controller on quartz64-a
commit: 4f4cbbb147b988daaa036dcf34628d93b2e22cd9Best regards,
Signature · 1 line
Heiko Stuebner <[email protected]>