2020-03-26 06:49:55

by Srinath Mannam

[permalink] [raw]
Subject: [PATCH v5 0/6] PAXB INTx support with proper model

This patch series adds PCIe legacy interrupt (INTx) support to the iProc
PCIe driver by modeling it with its own IRQ domain. All 4 interrupts INTA,
INTB, INTC, INTD share the same interrupt line connected to the GIC
in the system. This is now modeled by using its own IRQ domain.

Also update all relevant devicetree files to adapt to the new model.

This patch set is based on Linux-5.5-rc1.

Changes from v4:
- Addressed Lorenzo's comments
- Add iProc irq chip descriptor in the place of dummy irq chip to
provide mask/un-mask and enable/disable handlers to configure
individual INTx.

Changes from v3:
- Addressed Andrew Murray's comments
- Add change to dispose VIRQ when disabling INTx

Changes from v2:
- Addressed Lorenzo's comments
- Corrected INTx to PIN mapping.

Changes from v1:
- Addressed Rob, Lorenzo, Arnd's comments
- Used child node for interrupt controller.
- Addressed Andy Shevchenko's comments
- Replaced while loop with do-while.

Ray Jui (6):
dt-bindings: pci: Update iProc PCI binding for INTx support
PCI: iproc: Add INTx support with better modeling
arm: dts: Change PCIe INTx mapping for Cygnus
arm: dts: Change PCIe INTx mapping for NSP
arm: dts: Change PCIe INTx mapping for HR2
arm64: dts: Change PCIe INTx mapping for NS2

.../devicetree/bindings/pci/brcm,iproc-pcie.txt | 48 ++++++-
arch/arm/boot/dts/bcm-cygnus.dtsi | 30 ++++-
arch/arm/boot/dts/bcm-hr2.dtsi | 30 ++++-
arch/arm/boot/dts/bcm-nsp.dtsi | 45 ++++++-
arch/arm64/boot/dts/broadcom/northstar2/ns2.dtsi | 28 +++-
drivers/pci/controller/pcie-iproc.c | 147 ++++++++++++++++++++-
drivers/pci/controller/pcie-iproc.h | 8 ++
7 files changed, 309 insertions(+), 27 deletions(-)

--
2.7.4


2020-03-26 06:50:06

by Srinath Mannam

[permalink] [raw]
Subject: [PATCH v5 1/6] dt-bindings: pci: Update iProc PCI binding for INTx support

From: Ray Jui <[email protected]>

Update the iProc PCIe binding document for better modeling of the legacy
interrupt (INTx) support.

Signed-off-by: Ray Jui <[email protected]>
Signed-off-by: Srinath Mannam <[email protected]>
Reviewed-by: Rob Herring <[email protected]>
---
.../devicetree/bindings/pci/brcm,iproc-pcie.txt | 48 ++++++++++++++++++----
1 file changed, 41 insertions(+), 7 deletions(-)

diff --git a/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt b/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
index df065aa..d3f833a 100644
--- a/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
+++ b/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
@@ -13,9 +13,6 @@ controller, used in Stingray
PAXB-based root complex is used for external endpoint devices. PAXC-based
root complex is connected to emulated endpoint devices internal to the ASIC
- reg: base address and length of the PCIe controller I/O register space
-- #interrupt-cells: set to <1>
-- interrupt-map-mask and interrupt-map, standard PCI properties to define the
- mapping of the PCIe interface to interrupt numbers
- linux,pci-domain: PCI domain ID. Should be unique for each host controller
- bus-range: PCI bus numbers covered
- #address-cells: set to <3>
@@ -41,6 +38,21 @@ Required:
- brcm,pcie-ob-axi-offset: The offset from the AXI address to the internal
address used by the iProc PCIe core (not the PCIe address)

+Legacy interrupt (INTx) support (optional):
+
+Note INTx is for PAXB only.
+- interrupt-map-mask and interrupt-map, standard PCI properties to define
+the mapping of the PCIe interface to interrupt numbers
+
+In addition, a sub-node that describes the legacy interrupt controller built
+into the PCIe controller.
+This sub-node must have the following properties:
+ - compatible: must be "brcm,iproc-intc"
+ - interrupt-controller: claims itself as an interrupt controller for INTx
+ - #interrupt-cells: set to <1>
+ - interrupts: interrupt line wired to the generic GIC for INTx support
+ - interrupt-parent: Phandle to the parent interrupt controller
+
MSI support (optional):

For older platforms without MSI integrated in the GIC, iProc PCIe core provides
@@ -77,8 +89,11 @@ Example:
reg = <0x18012000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 100 IRQ_TYPE_NONE>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie0_intc 0>,
+ <0 0 0 2 &pcie0_intc 1>,
+ <0 0 0 3 &pcie0_intc 2>,
+ <0 0 0 4 &pcie0_intc 3>;

linux,pci-domain = <0>;

@@ -98,6 +113,14 @@ Example:

msi-parent = <&msi0>;

+ pcie0_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 100 IRQ_TYPE_NONE>;
+ };
+
/* iProc event queue based MSI */
msi0: msi@18012000 {
compatible = "brcm,iproc-msi";
@@ -115,8 +138,11 @@ Example:
reg = <0x18013000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 106 IRQ_TYPE_NONE>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie1_intc 0>,
+ <0 0 0 2 &pcie1_intc 1>,
+ <0 0 0 3 &pcie1_intc 2>,
+ <0 0 0 4 &pcie1_intc 3>;

linux,pci-domain = <1>;

@@ -130,4 +156,12 @@ Example:

phys = <&phy 1 6>;
phy-names = "pcie-phy";
+
+ pcie1_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 106 IRQ_TYPE_NONE>;
+ };
};
--
2.7.4

2020-03-26 06:50:21

by Srinath Mannam

[permalink] [raw]
Subject: [PATCH v5 3/6] arm: dts: Change PCIe INTx mapping for Cygnus

From: Ray Jui <[email protected]>

Change the PCIe INTx mapping to model the 4 INTx interrupts in the
IRQ domain of the iProc PCIe controller itself.

Signed-off-by: Ray Jui <[email protected]>
Signed-off-by: Srinath Mannam <[email protected]>
---
arch/arm/boot/dts/bcm-cygnus.dtsi | 30 ++++++++++++++++++++++++++----
1 file changed, 26 insertions(+), 4 deletions(-)

diff --git a/arch/arm/boot/dts/bcm-cygnus.dtsi b/arch/arm/boot/dts/bcm-cygnus.dtsi
index 2dac3ef..ca23e82 100644
--- a/arch/arm/boot/dts/bcm-cygnus.dtsi
+++ b/arch/arm/boot/dts/bcm-cygnus.dtsi
@@ -264,8 +264,11 @@
reg = <0x18012000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 100 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie0_intc 0>,
+ <0 0 0 2 &pcie0_intc 1>,
+ <0 0 0 3 &pcie0_intc 2>,
+ <0 0 0 4 &pcie0_intc 3>;

linux,pci-domain = <0>;

@@ -292,6 +295,14 @@
<GIC_SPI 98 IRQ_TYPE_LEVEL_HIGH>,
<GIC_SPI 99 IRQ_TYPE_LEVEL_HIGH>;
};
+
+ pcie0_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 100 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

pcie1: pcie@18013000 {
@@ -299,8 +310,11 @@
reg = <0x18013000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 106 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie1_intc 0>,
+ <0 0 0 2 &pcie1_intc 1>,
+ <0 0 0 3 &pcie1_intc 2>,
+ <0 0 0 4 &pcie1_intc 3>;

linux,pci-domain = <1>;

@@ -327,6 +341,14 @@
<GIC_SPI 104 IRQ_TYPE_LEVEL_HIGH>,
<GIC_SPI 105 IRQ_TYPE_LEVEL_HIGH>;
};
+
+ pcie1_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 106 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

dma0: dma@18018000 {
--
2.7.4

2020-03-26 06:50:32

by Srinath Mannam

[permalink] [raw]
Subject: [PATCH v5 2/6] PCI: iproc: Add INTx support with better modeling

From: Ray Jui <[email protected]>

Add PCIe legacy interrupt INTx support to the iProc PCIe driver by
modeling it with its own IRQ domain. All 4 interrupts INTA, INTB, INTC,
INTD share the same interrupt line connected to the GIC in the system,
while the status of each INTx can be obtained through the INTX CSR
register.

Signed-off-by: Ray Jui <[email protected]>
Signed-off-by: Srinath Mannam <[email protected]>
Reviewed-by: Andrew Murray <[email protected]>
---
drivers/pci/controller/pcie-iproc.c | 147 +++++++++++++++++++++++++++++++++++-
drivers/pci/controller/pcie-iproc.h | 8 ++
2 files changed, 153 insertions(+), 2 deletions(-)

diff --git a/drivers/pci/controller/pcie-iproc.c b/drivers/pci/controller/pcie-iproc.c
index 0a468c7..62d8f43 100644
--- a/drivers/pci/controller/pcie-iproc.c
+++ b/drivers/pci/controller/pcie-iproc.c
@@ -14,6 +14,7 @@
#include <linux/delay.h>
#include <linux/interrupt.h>
#include <linux/irqchip/arm-gic-v3.h>
+#include <linux/irqchip/chained_irq.h>
#include <linux/platform_device.h>
#include <linux/of_address.h>
#include <linux/of_pci.h>
@@ -270,6 +271,7 @@ enum iproc_pcie_reg {

/* enable INTx */
IPROC_PCIE_INTX_EN,
+ IPROC_PCIE_INTX_CSR,

/* outbound address mapping */
IPROC_PCIE_OARR0,
@@ -314,6 +316,7 @@ static const u16 iproc_pcie_reg_paxb_bcma[] = {
[IPROC_PCIE_CFG_ADDR] = 0x1f8,
[IPROC_PCIE_CFG_DATA] = 0x1fc,
[IPROC_PCIE_INTX_EN] = 0x330,
+ [IPROC_PCIE_INTX_CSR] = 0x334,
[IPROC_PCIE_LINK_STATUS] = 0xf0c,
};

@@ -325,6 +328,7 @@ static const u16 iproc_pcie_reg_paxb[] = {
[IPROC_PCIE_CFG_ADDR] = 0x1f8,
[IPROC_PCIE_CFG_DATA] = 0x1fc,
[IPROC_PCIE_INTX_EN] = 0x330,
+ [IPROC_PCIE_INTX_CSR] = 0x334,
[IPROC_PCIE_OARR0] = 0xd20,
[IPROC_PCIE_OMAP0] = 0xd40,
[IPROC_PCIE_OARR1] = 0xd28,
@@ -341,6 +345,7 @@ static const u16 iproc_pcie_reg_paxb_v2[] = {
[IPROC_PCIE_CFG_ADDR] = 0x1f8,
[IPROC_PCIE_CFG_DATA] = 0x1fc,
[IPROC_PCIE_INTX_EN] = 0x330,
+ [IPROC_PCIE_INTX_CSR] = 0x334,
[IPROC_PCIE_OARR0] = 0xd20,
[IPROC_PCIE_OMAP0] = 0xd40,
[IPROC_PCIE_OARR1] = 0xd28,
@@ -846,9 +851,142 @@ static int iproc_pcie_check_link(struct iproc_pcie *pcie)
return link_is_active ? 0 : -ENODEV;
}

-static void iproc_pcie_enable(struct iproc_pcie *pcie)
+static void iproc_pcie_mask_irq(struct irq_data *d)
{
+ struct iproc_pcie *pcie = irq_data_get_irq_chip_data(d);
+ u32 val;
+ unsigned long flags;
+
+ spin_lock_irqsave(&pcie->intx_lock, flags);
+ val = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_EN);
+ val &= ~(BIT(irqd_to_hwirq(d)));
+ iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, val);
+ spin_unlock_irqrestore(&pcie->intx_lock, flags);
+}
+
+static void iproc_pcie_unmask_irq(struct irq_data *d)
+{
+ struct iproc_pcie *pcie = irq_data_get_irq_chip_data(d);
+ u32 val;
+ unsigned long flags;
+
+ spin_lock_irqsave(&pcie->intx_lock, flags);
+ val = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_EN);
+ val |= (BIT(irqd_to_hwirq(d)));
+ iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, val);
+ spin_unlock_irqrestore(&pcie->intx_lock, flags);
+}
+
+static struct irq_chip iproc_pcie_irq_chip = {
+ .name = "pcie-iproc-intc",
+ .irq_enable = iproc_pcie_unmask_irq,
+ .irq_disable = iproc_pcie_mask_irq,
+ .irq_mask = iproc_pcie_mask_irq,
+ .irq_unmask = iproc_pcie_unmask_irq,
+};
+
+static int iproc_pcie_intx_map(struct irq_domain *domain, unsigned int irq,
+ irq_hw_number_t hwirq)
+{
+ irq_set_chip_and_handler(irq, &iproc_pcie_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 = iproc_pcie_intx_map,
+};
+
+static void iproc_pcie_isr(struct irq_desc *desc)
+{
+ struct irq_chip *chip = irq_desc_get_chip(desc);
+ struct iproc_pcie *pcie;
+ struct device *dev;
+ unsigned long status;
+ u32 bit, virq;
+
+ chained_irq_enter(chip, desc);
+ pcie = irq_desc_get_handler_data(desc);
+ dev = pcie->dev;
+
+ /* go through INTx A, B, C, D until all interrupts are handled */
+ do {
+ status = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_CSR);
+ for_each_set_bit(bit, &status, PCI_NUM_INTX) {
+ virq = irq_find_mapping(pcie->irq_domain, bit);
+ if (virq)
+ generic_handle_irq(virq);
+ else
+ dev_err(dev, "unexpected INTx%u\n", bit);
+ }
+ } while ((status & SYS_RC_INTX_MASK) != 0);
+
+ chained_irq_exit(chip, desc);
+}
+
+static int iproc_pcie_intx_enable(struct iproc_pcie *pcie)
+{
+ struct device *dev = pcie->dev;
+ struct device_node *node;
+ int ret;
+
+ /*
+ * BCMA devices do not map INTx the same way as platform devices. All
+ * BCMA needs below line to enable INTx
+ */
iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, SYS_RC_INTX_MASK);
+
+ node = of_get_compatible_child(dev->of_node, "brcm,iproc-intc");
+ if (node)
+ pcie->irq = of_irq_get(node, 0);
+
+ if (!node || pcie->irq <= 0)
+ return 0;
+
+ spin_lock_init(&pcie->intx_lock);
+
+ /* set IRQ handler */
+ irq_set_chained_handler_and_data(pcie->irq, iproc_pcie_isr, pcie);
+
+ /* add IRQ domain for INTx */
+ pcie->irq_domain = irq_domain_add_linear(node, PCI_NUM_INTX,
+ &intx_domain_ops, pcie);
+ if (!pcie->irq_domain) {
+ dev_err(dev, "failed to add INTx IRQ domain\n");
+ ret = -ENOMEM;
+ goto err_rm_handler_data;
+ }
+
+ return 0;
+
+err_rm_handler_data:
+ of_node_put(node);
+ irq_set_chained_handler_and_data(pcie->irq, NULL, NULL);
+
+ return ret;
+}
+
+static void iproc_pcie_intx_disable(struct iproc_pcie *pcie)
+{
+ uint32_t offset, virq;
+ unsigned long flags;
+
+ spin_lock_irqsave(&pcie->intx_lock, flags);
+ iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, 0x0);
+ spin_unlock_irqrestore(&pcie->intx_lock, flags);
+
+ if (pcie->irq <= 0)
+ return;
+
+ for (offset = 0; offset < PCI_NUM_INTX; offset++) {
+ virq = irq_find_mapping(pcie->irq_domain, offset);
+ if (virq)
+ irq_dispose_mapping(virq);
+ }
+
+ irq_domain_remove(pcie->irq_domain);
+ irq_set_chained_handler_and_data(pcie->irq, NULL, NULL);
}

static inline bool iproc_pcie_ob_is_valid(struct iproc_pcie *pcie,
@@ -1518,7 +1656,11 @@ int iproc_pcie_setup(struct iproc_pcie *pcie, struct list_head *res)
goto err_power_off_phy;
}

- iproc_pcie_enable(pcie);
+ ret = iproc_pcie_intx_enable(pcie);
+ if (ret) {
+ dev_err(dev, "failed to enable INTx\n");
+ goto err_power_off_phy;
+ }

if (IS_ENABLED(CONFIG_PCI_MSI))
if (iproc_pcie_msi_enable(pcie))
@@ -1562,6 +1704,7 @@ int iproc_pcie_remove(struct iproc_pcie *pcie)
pci_remove_root_bus(pcie->root_bus);

iproc_pcie_msi_disable(pcie);
+ iproc_pcie_intx_disable(pcie);

phy_power_off(pcie->phy);
phy_exit(pcie->phy);
diff --git a/drivers/pci/controller/pcie-iproc.h b/drivers/pci/controller/pcie-iproc.h
index 4f03ea5..787bfba 100644
--- a/drivers/pci/controller/pcie-iproc.h
+++ b/drivers/pci/controller/pcie-iproc.h
@@ -74,9 +74,13 @@ struct iproc_msi;
* @ib: inbound mapping related parameters
* @ib_map: outbound mapping region related parameters
*
+ * @irq: interrupt line wired to the generic GIC for INTx
+ * @irq_domain: IRQ domain for INTx
+ *
* @need_msi_steer: indicates additional configuration of the iProc PCIe
* controller is required to steer MSI writes to external interrupt controller
* @msi: MSI data
+ * @intx_lock: spinlock to protect access to INTx related registers
*/
struct iproc_pcie {
struct device *dev;
@@ -102,8 +106,12 @@ struct iproc_pcie {
struct iproc_pcie_ib ib;
const struct iproc_pcie_ib_map *ib_map;

+ int irq;
+ struct irq_domain *irq_domain;
+
bool need_msi_steer;
struct iproc_msi *msi;
+ spinlock_t intx_lock;
};

int iproc_pcie_setup(struct iproc_pcie *pcie, struct list_head *res);
--
2.7.4

2020-03-26 06:50:54

by Srinath Mannam

[permalink] [raw]
Subject: [PATCH v5 6/6] arm64: dts: Change PCIe INTx mapping for NS2

From: Ray Jui <[email protected]>

Change the PCIe INTx mapping to model the 4 INTx interrupts in the
IRQ domain of the iProc PCIe controller itself.

Signed-off-by: Ray Jui <[email protected]>
Signed-off-by: Srinath Mannam <[email protected]>
---
arch/arm64/boot/dts/broadcom/northstar2/ns2.dtsi | 28 ++++++++++++++++++++----
1 file changed, 24 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/boot/dts/broadcom/northstar2/ns2.dtsi b/arch/arm64/boot/dts/broadcom/northstar2/ns2.dtsi
index 15f7b0e..489bfd5 100644
--- a/arch/arm64/boot/dts/broadcom/northstar2/ns2.dtsi
+++ b/arch/arm64/boot/dts/broadcom/northstar2/ns2.dtsi
@@ -117,8 +117,11 @@
dma-coherent;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic 0 GIC_SPI 281 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie0_intc 0>,
+ <0 0 0 2 &pcie0_intc 1>,
+ <0 0 0 3 &pcie0_intc 2>,
+ <0 0 0 4 &pcie0_intc 3>;

linux,pci-domain = <0>;

@@ -140,6 +143,13 @@
phy-names = "pcie-phy";

msi-parent = <&v2m0>;
+ pcie0_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 281 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

pcie4: pcie@50020000 {
@@ -148,8 +158,11 @@
dma-coherent;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic 0 GIC_SPI 305 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie4_intc 0>,
+ <0 0 0 2 &pcie4_intc 1>,
+ <0 0 0 3 &pcie4_intc 2>,
+ <0 0 0 4 &pcie4_intc 3>;

linux,pci-domain = <4>;

@@ -171,6 +184,13 @@
phy-names = "pcie-phy";

msi-parent = <&v2m0>;
+ pcie4_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 305 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

pcie8: pcie@60c00000 {
--
2.7.4

2020-03-26 06:51:29

by Srinath Mannam

[permalink] [raw]
Subject: [PATCH v5 4/6] arm: dts: Change PCIe INTx mapping for NSP

From: Ray Jui <[email protected]>

Change the PCIe INTx mapping to model the 4 INTx interrupts in the
IRQ domain of the iProc PCIe controller itself.

Signed-off-by: Ray Jui <[email protected]>
Signed-off-by: Srinath Mannam <[email protected]>
---
arch/arm/boot/dts/bcm-nsp.dtsi | 45 ++++++++++++++++++++++++++++++++++++------
1 file changed, 39 insertions(+), 6 deletions(-)

diff --git a/arch/arm/boot/dts/bcm-nsp.dtsi b/arch/arm/boot/dts/bcm-nsp.dtsi
index da6d70f..6d73221 100644
--- a/arch/arm/boot/dts/bcm-nsp.dtsi
+++ b/arch/arm/boot/dts/bcm-nsp.dtsi
@@ -529,8 +529,11 @@
reg = <0x18012000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie0_intc 0>,
+ <0 0 0 2 &pcie0_intc 1>,
+ <0 0 0 3 &pcie0_intc 2>,
+ <0 0 0 4 &pcie0_intc 3>;

linux,pci-domain = <0>;

@@ -559,6 +562,14 @@
<GIC_SPI 130 IRQ_TYPE_LEVEL_HIGH>;
brcm,pcie-msi-inten;
};
+
+ pcie0_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

pcie1: pcie@18013000 {
@@ -566,8 +577,11 @@
reg = <0x18013000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 137 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie1_intc 0>,
+ <0 0 0 2 &pcie1_intc 1>,
+ <0 0 0 3 &pcie1_intc 2>,
+ <0 0 0 4 &pcie1_intc 3>;

linux,pci-domain = <1>;

@@ -596,6 +610,14 @@
<GIC_SPI 136 IRQ_TYPE_LEVEL_HIGH>;
brcm,pcie-msi-inten;
};
+
+ pcie1_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 137 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

pcie2: pcie@18014000 {
@@ -603,8 +625,11 @@
reg = <0x18014000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 143 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie2_intc 0>,
+ <0 0 0 2 &pcie2_intc 1>,
+ <0 0 0 3 &pcie2_intc 2>,
+ <0 0 0 4 &pcie2_intc 3>;

linux,pci-domain = <2>;

@@ -633,6 +658,14 @@
<GIC_SPI 142 IRQ_TYPE_LEVEL_HIGH>;
brcm,pcie-msi-inten;
};
+
+ pcie2_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 143 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

thermal-zones {
--
2.7.4

2020-03-26 06:51:32

by Srinath Mannam

[permalink] [raw]
Subject: [PATCH v5 5/6] arm: dts: Change PCIe INTx mapping for HR2

From: Ray Jui <[email protected]>

Change the PCIe INTx mapping to model the 4 INTx interrupts in the
IRQ domain of the iProc PCIe controller itself.

Signed-off-by: Ray Jui <[email protected]>
Signed-off-by: Srinath Mannam <[email protected]>
---
arch/arm/boot/dts/bcm-hr2.dtsi | 30 ++++++++++++++++++++++++++----
1 file changed, 26 insertions(+), 4 deletions(-)

diff --git a/arch/arm/boot/dts/bcm-hr2.dtsi b/arch/arm/boot/dts/bcm-hr2.dtsi
index 6142c67..80c3add 100644
--- a/arch/arm/boot/dts/bcm-hr2.dtsi
+++ b/arch/arm/boot/dts/bcm-hr2.dtsi
@@ -299,8 +299,11 @@
reg = <0x18012000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 186 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie0_intc 0>,
+ <0 0 0 2 &pcie0_intc 1>,
+ <0 0 0 3 &pcie0_intc 2>,
+ <0 0 0 4 &pcie0_intc 3>;

linux,pci-domain = <0>;

@@ -328,6 +331,14 @@
<GIC_SPI 185 IRQ_TYPE_LEVEL_HIGH>;
brcm,pcie-msi-inten;
};
+
+ pcie0_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 186 IRQ_TYPE_LEVEL_HIGH>;
+ };
};

pcie1: pcie@18013000 {
@@ -335,8 +346,11 @@
reg = <0x18013000 0x1000>;

#interrupt-cells = <1>;
- interrupt-map-mask = <0 0 0 0>;
- interrupt-map = <0 0 0 0 &gic GIC_SPI 192 IRQ_TYPE_LEVEL_HIGH>;
+ interrupt-map-mask = <0 0 0 7>;
+ interrupt-map = <0 0 0 1 &pcie1_intc 0>,
+ <0 0 0 2 &pcie1_intc 1>,
+ <0 0 0 3 &pcie1_intc 2>,
+ <0 0 0 4 &pcie1_intc 3>;

linux,pci-domain = <1>;

@@ -364,5 +378,13 @@
<GIC_SPI 191 IRQ_TYPE_LEVEL_HIGH>;
brcm,pcie-msi-inten;
};
+
+ pcie1_intc: interrupt-controller {
+ compatible = "brcm,iproc-intc";
+ interrupt-controller;
+ #interrupt-cells = <1>;
+ interrupt-parent = <&gic>;
+ interrupts = <GIC_SPI 192 IRQ_TYPE_LEVEL_HIGH>;
+ };
};
};
--
2.7.4

2020-03-26 12:35:43

by Arnd Bergmann

[permalink] [raw]
Subject: Re: [PATCH v5 2/6] PCI: iproc: Add INTx support with better modeling

On Thu, Mar 26, 2020 at 7:49 AM Srinath Mannam
<[email protected]> wrote:
> -static void iproc_pcie_enable(struct iproc_pcie *pcie)
> +static void iproc_pcie_mask_irq(struct irq_data *d)
> {
> + struct iproc_pcie *pcie = irq_data_get_irq_chip_data(d);
> + u32 val;
> + unsigned long flags;
> +
> + spin_lock_irqsave(&pcie->intx_lock, flags);
> + val = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_EN);
> + val &= ~(BIT(irqd_to_hwirq(d)));
> + iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, val);
> + spin_unlock_irqrestore(&pcie->intx_lock, flags);
> +}

I think these need to use a raw spinlock, or you get problems with
PREEMPT_RT

Arnd

2020-03-26 12:38:27

by Arnd Bergmann

[permalink] [raw]
Subject: Re: [PATCH v5 1/6] dt-bindings: pci: Update iProc PCI binding for INTx support

On Thu, Mar 26, 2020 at 7:49 AM Srinath Mannam
<[email protected]> wrote:
>
> From: Ray Jui <[email protected]>
>
> Update the iProc PCIe binding document for better modeling of the legacy
> interrupt (INTx) support.

Could you describe in the changelog what happens when you get an old dtb
with a new kernel or vice versa? If any combination won't work, please make
that very clear, and give a reason why you consider this not to be a problem.

Arnd

2020-03-26 14:23:27

by Bjorn Helgaas

[permalink] [raw]
Subject: Re: [PATCH v5 1/6] dt-bindings: pci: Update iProc PCI binding for INTx support

$ git log --oneline Documentation/devicetree/bindings/pci/

34129bb831cc ("dt-bindings: PCI: intel: Fix dt_binding_check compilation failure")
e1ac611f57c9 ("dt-bindings: PCI: Convert generic host binding to DT schema")
919ba6e739eb ("dt-bindings: PCI: Convert Arm Versatile binding to DT schema")
0956dcb853dc ("dt-bindings: PCI: Add bindings for brcmstb's PCIe device")
5d28bee7c91e ("dt-bindings: PCI: qcom: Add support for SDM845 PCIe")
e54ea45a4955 ("dt-bindings: PCI: intel: Add YAML schemas for the PCIe RC controller")
d8725e38dd9f ("dt-bindings: pci: layerscape-pci: add compatible strings "fsl, ls1028a-pcie"")

Capitalize yours match and put the useful information at the
beginning; maybe something like this:

dt-bindings: PCI: iproc: Improve INTx modeling

On Thu, Mar 26, 2020 at 12:18:41PM +0530, Srinath Mannam wrote:
> From: Ray Jui <[email protected]>
>
> Update the iProc PCIe binding document for better modeling of the legacy
> interrupt (INTx) support.

2020-03-26 14:38:08

by Andy Shevchenko

[permalink] [raw]
Subject: Re: [PATCH v5 2/6] PCI: iproc: Add INTx support with better modeling

On Thu, Mar 26, 2020 at 8:49 AM Srinath Mannam
<[email protected]> wrote:
>
> From: Ray Jui <[email protected]>
>
> Add PCIe legacy interrupt INTx support to the iProc PCIe driver by
> modeling it with its own IRQ domain. All 4 interrupts INTA, INTB, INTC,
> INTD share the same interrupt line connected to the GIC in the system,
> while the status of each INTx can be obtained through the INTX CSR
> register.

...
> + val &= ~(BIT(irqd_to_hwirq(d)));

Too many parentheses.

...

> + val |= (BIT(irqd_to_hwirq(d)));

Ditto.

...

> + /* go through INTx A, B, C, D until all interrupts are handled */
> + do {
> + status = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_CSR);
> + for_each_set_bit(bit, &status, PCI_NUM_INTX) {
> + virq = irq_find_mapping(pcie->irq_domain, bit);
> + if (virq)
> + generic_handle_irq(virq);

> + else
> + dev_err(dev, "unexpected INTx%u\n", bit);

Any guarantee it will be no storm of undesired messages here?

> + }

> + } while ((status & SYS_RC_INTX_MASK) != 0);

' != 0' part is not needed.

If there an interrupt storm the handler will never end, right?
Is it the idea by design?

...

> + node = of_get_compatible_child(dev->of_node, "brcm,iproc-intc");
> + if (node)
> + pcie->irq = of_irq_get(node, 0);
> +
> + if (!node || pcie->irq <= 0)
> + return 0;

Perhaps
node = of_get_compatible_child(dev->of_node, "brcm,iproc-intc");
if (!node)
return 0;

pcie->irq = of_irq_get(node, 0);
if (pcie->irq <= 0)
return 0;
?

--
With Best Regards,
Andy Shevchenko

2020-03-26 14:52:30

by Marc Zyngier

[permalink] [raw]
Subject: Re: [PATCH v5 1/6] dt-bindings: pci: Update iProc PCI binding for INTx support

Srinath,

Please note that Andrew's email address as changed (see the MAINTAINERS
file).

On 2020-03-26 06:48, Srinath Mannam wrote:
> From: Ray Jui <[email protected]>
>
> Update the iProc PCIe binding document for better modeling of the
> legacy
> interrupt (INTx) support.
>
> Signed-off-by: Ray Jui <[email protected]>
> Signed-off-by: Srinath Mannam <[email protected]>
> Reviewed-by: Rob Herring <[email protected]>
> ---
> .../devicetree/bindings/pci/brcm,iproc-pcie.txt | 48
> ++++++++++++++++++----
> 1 file changed, 41 insertions(+), 7 deletions(-)
>
> diff --git a/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
> b/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
> index df065aa..d3f833a 100644
> --- a/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
> +++ b/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
> @@ -13,9 +13,6 @@ controller, used in Stingray
> PAXB-based root complex is used for external endpoint devices.
> PAXC-based
> root complex is connected to emulated endpoint devices internal to the
> ASIC
> - reg: base address and length of the PCIe controller I/O register
> space
> -- #interrupt-cells: set to <1>
> -- interrupt-map-mask and interrupt-map, standard PCI properties to
> define the
> - mapping of the PCIe interface to interrupt numbers
> - linux,pci-domain: PCI domain ID. Should be unique for each host
> controller
> - bus-range: PCI bus numbers covered
> - #address-cells: set to <3>
> @@ -41,6 +38,21 @@ Required:
> - brcm,pcie-ob-axi-offset: The offset from the AXI address to the
> internal
> address used by the iProc PCIe core (not the PCIe address)
>
> +Legacy interrupt (INTx) support (optional):
> +
> +Note INTx is for PAXB only.
> +- interrupt-map-mask and interrupt-map, standard PCI properties to
> define
> +the mapping of the PCIe interface to interrupt numbers
> +
> +In addition, a sub-node that describes the legacy interrupt controller
> built
> +into the PCIe controller.
> +This sub-node must have the following properties:
> + - compatible: must be "brcm,iproc-intc"
> + - interrupt-controller: claims itself as an interrupt controller for
> INTx
> + - #interrupt-cells: set to <1>
> + - interrupts: interrupt line wired to the generic GIC for INTx
> support
> + - interrupt-parent: Phandle to the parent interrupt controller
> +
> MSI support (optional):
>
> For older platforms without MSI integrated in the GIC, iProc PCIe core
> provides
> @@ -77,8 +89,11 @@ Example:
> reg = <0x18012000 0x1000>;
>
> #interrupt-cells = <1>;
> - interrupt-map-mask = <0 0 0 0>;
> - interrupt-map = <0 0 0 0 &gic GIC_SPI 100 IRQ_TYPE_NONE>;
> + interrupt-map-mask = <0 0 0 7>;
> + interrupt-map = <0 0 0 1 &pcie0_intc 0>,
> + <0 0 0 2 &pcie0_intc 1>,
> + <0 0 0 3 &pcie0_intc 2>,
> + <0 0 0 4 &pcie0_intc 3>;
>
> linux,pci-domain = <0>;
>
> @@ -98,6 +113,14 @@ Example:
>
> msi-parent = <&msi0>;
>
> + pcie0_intc: interrupt-controller {
> + compatible = "brcm,iproc-intc";
> + interrupt-controller;
> + #interrupt-cells = <1>;
> + interrupt-parent = <&gic>;
> + interrupts = <GIC_SPI 100 IRQ_TYPE_NONE>;

There is no such thing as IRQ_TYPE_NONE in the GIC binding.
Please update this to the right trigger type (which better
be level high...)

M.
--
Jazz is not dead. It just smells funny...

2020-03-26 15:05:36

by Marc Zyngier

[permalink] [raw]
Subject: Re: [PATCH v5 2/6] PCI: iproc: Add INTx support with better modeling

On 2020-03-26 06:48, Srinath Mannam wrote:
> From: Ray Jui <[email protected]>
>
> Add PCIe legacy interrupt INTx support to the iProc PCIe driver by
> modeling it with its own IRQ domain. All 4 interrupts INTA, INTB, INTC,
> INTD share the same interrupt line connected to the GIC in the system,
> while the status of each INTx can be obtained through the INTX CSR
> register.
>
> Signed-off-by: Ray Jui <[email protected]>
> Signed-off-by: Srinath Mannam <[email protected]>
> Reviewed-by: Andrew Murray <[email protected]>
> ---
> drivers/pci/controller/pcie-iproc.c | 147
> +++++++++++++++++++++++++++++++++++-
> drivers/pci/controller/pcie-iproc.h | 8 ++
> 2 files changed, 153 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/pci/controller/pcie-iproc.c
> b/drivers/pci/controller/pcie-iproc.c
> index 0a468c7..62d8f43 100644
> --- a/drivers/pci/controller/pcie-iproc.c
> +++ b/drivers/pci/controller/pcie-iproc.c
> @@ -14,6 +14,7 @@
> #include <linux/delay.h>
> #include <linux/interrupt.h>
> #include <linux/irqchip/arm-gic-v3.h>
> +#include <linux/irqchip/chained_irq.h>
> #include <linux/platform_device.h>
> #include <linux/of_address.h>
> #include <linux/of_pci.h>
> @@ -270,6 +271,7 @@ enum iproc_pcie_reg {
>
> /* enable INTx */
> IPROC_PCIE_INTX_EN,
> + IPROC_PCIE_INTX_CSR,
>
> /* outbound address mapping */
> IPROC_PCIE_OARR0,
> @@ -314,6 +316,7 @@ static const u16 iproc_pcie_reg_paxb_bcma[] = {
> [IPROC_PCIE_CFG_ADDR] = 0x1f8,
> [IPROC_PCIE_CFG_DATA] = 0x1fc,
> [IPROC_PCIE_INTX_EN] = 0x330,
> + [IPROC_PCIE_INTX_CSR] = 0x334,
> [IPROC_PCIE_LINK_STATUS] = 0xf0c,
> };
>
> @@ -325,6 +328,7 @@ static const u16 iproc_pcie_reg_paxb[] = {
> [IPROC_PCIE_CFG_ADDR] = 0x1f8,
> [IPROC_PCIE_CFG_DATA] = 0x1fc,
> [IPROC_PCIE_INTX_EN] = 0x330,
> + [IPROC_PCIE_INTX_CSR] = 0x334,
> [IPROC_PCIE_OARR0] = 0xd20,
> [IPROC_PCIE_OMAP0] = 0xd40,
> [IPROC_PCIE_OARR1] = 0xd28,
> @@ -341,6 +345,7 @@ static const u16 iproc_pcie_reg_paxb_v2[] = {
> [IPROC_PCIE_CFG_ADDR] = 0x1f8,
> [IPROC_PCIE_CFG_DATA] = 0x1fc,
> [IPROC_PCIE_INTX_EN] = 0x330,
> + [IPROC_PCIE_INTX_CSR] = 0x334,
> [IPROC_PCIE_OARR0] = 0xd20,
> [IPROC_PCIE_OMAP0] = 0xd40,
> [IPROC_PCIE_OARR1] = 0xd28,
> @@ -846,9 +851,142 @@ static int iproc_pcie_check_link(struct
> iproc_pcie *pcie)
> return link_is_active ? 0 : -ENODEV;
> }
>
> -static void iproc_pcie_enable(struct iproc_pcie *pcie)
> +static void iproc_pcie_mask_irq(struct irq_data *d)
> {
> + struct iproc_pcie *pcie = irq_data_get_irq_chip_data(d);
> + u32 val;
> + unsigned long flags;
> +
> + spin_lock_irqsave(&pcie->intx_lock, flags);

As Arnd already noticed, this needs to be defined as a raw spinlock.

> + val = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_EN);
> + val &= ~(BIT(irqd_to_hwirq(d)));
> + iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, val);
> + spin_unlock_irqrestore(&pcie->intx_lock, flags);
> +}
> +
> +static void iproc_pcie_unmask_irq(struct irq_data *d)
> +{
> + struct iproc_pcie *pcie = irq_data_get_irq_chip_data(d);
> + u32 val;
> + unsigned long flags;
> +
> + spin_lock_irqsave(&pcie->intx_lock, flags);
> + val = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_EN);
> + val |= (BIT(irqd_to_hwirq(d)));
> + iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, val);
> + spin_unlock_irqrestore(&pcie->intx_lock, flags);
> +}
> +
> +static struct irq_chip iproc_pcie_irq_chip = {
> + .name = "pcie-iproc-intc",
> + .irq_enable = iproc_pcie_unmask_irq,
> + .irq_disable = iproc_pcie_mask_irq,

No. Either the enable/disable callbacks are different from unmask/mask,
or they don't exist.

- A disabled interrupt is free to loose pending interrupts
- A masked interrupt doesn't loose interrupts.

I'm pretty sure yours is the latter.

> + .irq_mask = iproc_pcie_mask_irq,
> + .irq_unmask = iproc_pcie_unmask_irq,
> +};
> +
> +static int iproc_pcie_intx_map(struct irq_domain *domain, unsigned int
> irq,
> + irq_hw_number_t hwirq)
> +{
> + irq_set_chip_and_handler(irq, &iproc_pcie_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 = iproc_pcie_intx_map,
> +};
> +
> +static void iproc_pcie_isr(struct irq_desc *desc)
> +{
> + struct irq_chip *chip = irq_desc_get_chip(desc);
> + struct iproc_pcie *pcie;
> + struct device *dev;
> + unsigned long status;
> + u32 bit, virq;
> +
> + chained_irq_enter(chip, desc);
> + pcie = irq_desc_get_handler_data(desc);
> + dev = pcie->dev;
> +
> + /* go through INTx A, B, C, D until all interrupts are handled */
> + do {
> + status = iproc_pcie_read_reg(pcie, IPROC_PCIE_INTX_CSR);
> + for_each_set_bit(bit, &status, PCI_NUM_INTX) {
> + virq = irq_find_mapping(pcie->irq_domain, bit);
> + if (virq)
> + generic_handle_irq(virq);
> + else
> + dev_err(dev, "unexpected INTx%u\n", bit);

I'd rather you avoid this kind of unlimited console screaming in
interrupt
context. Either ratelimit it, or even better turn it into a debug
statement.

> + }
> + } while ((status & SYS_RC_INTX_MASK) != 0);
> +
> + chained_irq_exit(chip, desc);
> +}
> +
> +static int iproc_pcie_intx_enable(struct iproc_pcie *pcie)
> +{
> + struct device *dev = pcie->dev;
> + struct device_node *node;
> + int ret;
> +
> + /*
> + * BCMA devices do not map INTx the same way as platform devices. All
> + * BCMA needs below line to enable INTx
> + */
> iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, SYS_RC_INTX_MASK);
> +
> + node = of_get_compatible_child(dev->of_node, "brcm,iproc-intc");
> + if (node)
> + pcie->irq = of_irq_get(node, 0);
> +
> + if (!node || pcie->irq <= 0)
> + return 0;
> +
> + spin_lock_init(&pcie->intx_lock);
> +
> + /* set IRQ handler */
> + irq_set_chained_handler_and_data(pcie->irq, iproc_pcie_isr, pcie);

Please do this last, once the domain is allocated (you leave a window
where
a screaming interrupt can fire here).

> +
> + /* add IRQ domain for INTx */
> + pcie->irq_domain = irq_domain_add_linear(node, PCI_NUM_INTX,
> + &intx_domain_ops, pcie);
> + if (!pcie->irq_domain) {
> + dev_err(dev, "failed to add INTx IRQ domain\n");
> + ret = -ENOMEM;
> + goto err_rm_handler_data;
> + }
> +
> + return 0;
> +
> +err_rm_handler_data:
> + of_node_put(node);
> + irq_set_chained_handler_and_data(pcie->irq, NULL, NULL);
> +
> + return ret;
> +}
> +
> +static void iproc_pcie_intx_disable(struct iproc_pcie *pcie)
> +{
> + uint32_t offset, virq;
> + unsigned long flags;
> +
> + spin_lock_irqsave(&pcie->intx_lock, flags);
> + iproc_pcie_write_reg(pcie, IPROC_PCIE_INTX_EN, 0x0);
> + spin_unlock_irqrestore(&pcie->intx_lock, flags);

What's the reason for this lock/unlock? What are you racing against?

> +
> + if (pcie->irq <= 0)
> + return;
> +
> + for (offset = 0; offset < PCI_NUM_INTX; offset++) {
> + virq = irq_find_mapping(pcie->irq_domain, offset);
> + if (virq)
> + irq_dispose_mapping(virq);
> + }
> +
> + irq_domain_remove(pcie->irq_domain);
> + irq_set_chained_handler_and_data(pcie->irq, NULL, NULL);

The other way around: first you disconnect the interrupt, then you
free the data that the interrupt can make use of.

> }
>
> static inline bool iproc_pcie_ob_is_valid(struct iproc_pcie *pcie,
> @@ -1518,7 +1656,11 @@ int iproc_pcie_setup(struct iproc_pcie *pcie,
> struct list_head *res)
> goto err_power_off_phy;
> }
>
> - iproc_pcie_enable(pcie);
> + ret = iproc_pcie_intx_enable(pcie);
> + if (ret) {
> + dev_err(dev, "failed to enable INTx\n");
> + goto err_power_off_phy;
> + }
>
> if (IS_ENABLED(CONFIG_PCI_MSI))
> if (iproc_pcie_msi_enable(pcie))
> @@ -1562,6 +1704,7 @@ int iproc_pcie_remove(struct iproc_pcie *pcie)
> pci_remove_root_bus(pcie->root_bus);
>
> iproc_pcie_msi_disable(pcie);
> + iproc_pcie_intx_disable(pcie);
>
> phy_power_off(pcie->phy);
> phy_exit(pcie->phy);
> diff --git a/drivers/pci/controller/pcie-iproc.h
> b/drivers/pci/controller/pcie-iproc.h
> index 4f03ea5..787bfba 100644
> --- a/drivers/pci/controller/pcie-iproc.h
> +++ b/drivers/pci/controller/pcie-iproc.h
> @@ -74,9 +74,13 @@ struct iproc_msi;
> * @ib: inbound mapping related parameters
> * @ib_map: outbound mapping region related parameters
> *
> + * @irq: interrupt line wired to the generic GIC for INTx
> + * @irq_domain: IRQ domain for INTx
> + *
> * @need_msi_steer: indicates additional configuration of the iProc
> PCIe
> * controller is required to steer MSI writes to external interrupt
> controller
> * @msi: MSI data
> + * @intx_lock: spinlock to protect access to INTx related registers
> */
> struct iproc_pcie {
> struct device *dev;
> @@ -102,8 +106,12 @@ struct iproc_pcie {
> struct iproc_pcie_ib ib;
> const struct iproc_pcie_ib_map *ib_map;
>
> + int irq;
> + struct irq_domain *irq_domain;
> +
> bool need_msi_steer;
> struct iproc_msi *msi;
> + spinlock_t intx_lock;
> };
>
> int iproc_pcie_setup(struct iproc_pcie *pcie, struct list_head *res);

Thanks,

M.
--
Jazz is not dead. It just smells funny...