[PATCHv3 0/5] PCI: dwc: add prefetchable memory range support

[PATCHv3 0/5] PCI: dwc: add prefetchable memory range support

[Z.q. Hou]

From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

This patch set is to add prefetchable memory range support, patch 5/5.
Patch 1/5 and 2/5 are to initialize the number of viewport for
layerscape PCIe.
BTW, fix 2 bugs, see patch 3/5 and 4/5.

Hou Zhiqiang (5):
  ARM: dts: ls1021a: add num-viewport property for PCIe DT nodes
  arm64: dts: layerscape: add num-viewport property for PCIe DT nodes
  PCI: dwc: fix potential memory leak
  PCI: dwc: fix 4GiB outbound window size truncated to zero issue
  PCI: dwc: add prefetchable memory range support

 arch/arm/boot/dts/ls1021a.dtsi                |   2 +
 .../arm64/boot/dts/freescale/fsl-ls1012a.dtsi |   1 +
 .../arm64/boot/dts/freescale/fsl-ls1043a.dtsi |   3 +
 .../arm64/boot/dts/freescale/fsl-ls1046a.dtsi |   3 +
 .../arm64/boot/dts/freescale/fsl-ls1088a.dtsi |   3 +
 .../arm64/boot/dts/freescale/fsl-ls208xa.dtsi |   4 +
 .../pci/controller/dwc/pcie-designware-host.c | 135 +++++++++++++-----
 drivers/pci/controller/dwc/pcie-designware.c  |   4 +-
 drivers/pci/controller/dwc/pcie-designware.h  |  11 +-
 9 files changed, 128 insertions(+), 38 deletions(-)

-- 
2.17.1

[PATCHv3 1/5] ARM: dts: ls1021a: add num-viewport property for PCIe DT nodes

[Z.q. Hou]

From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

Add num-viewport property for PCIe DT nodes to specify how many
viewports are implemented.

Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
---
V3:
 - New patch 

 arch/arm/boot/dts/ls1021a.dtsi | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/arm/boot/dts/ls1021a.dtsi b/arch/arm/boot/dts/ls1021a.dtsi
index b769e0e40553..b455acc7afb4 100644
--- a/arch/arm/boot/dts/ls1021a.dtsi
+++ b/arch/arm/boot/dts/ls1021a.dtsi
@@ -726,6 +726,7 @@
 			#size-cells = <2>;
 			device_type = "pci";
 			num-lanes = <4>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x40 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x40 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
@@ -750,6 +751,7 @@
 			#size-cells = <2>;
 			device_type = "pci";
 			num-lanes = <4>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x48 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x48 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
-- 
2.17.1

Re: [PATCHv3 1/5] ARM: dts: ls1021a: add num-viewport property for PCIe DT nodes

[Shawn Guo]

On Tue, Dec 18, 2018 at 04:19:26AM +0000, Z.q. Hou wrote:
> From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
> 
> Add num-viewport property for PCIe DT nodes to specify how many
> viewports are implemented.
> 
> Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

Applied, thanks.

[PATCHv3 2/5] arm64: dts: layerscape: add num-viewport property for PCIe DT nodes

[Z.q. Hou]

From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

Add num-viewport property for PCIe DT nodes to specify how many
viewports are implemented.

Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
---
V3:
 - New patch 

 arch/arm64/boot/dts/freescale/fsl-ls1012a.dtsi | 1 +
 arch/arm64/boot/dts/freescale/fsl-ls1043a.dtsi | 3 +++
 arch/arm64/boot/dts/freescale/fsl-ls1046a.dtsi | 3 +++
 arch/arm64/boot/dts/freescale/fsl-ls1088a.dtsi | 3 +++
 arch/arm64/boot/dts/freescale/fsl-ls208xa.dtsi | 4 ++++
 5 files changed, 14 insertions(+)

diff --git a/arch/arm64/boot/dts/freescale/fsl-ls1012a.dtsi b/arch/arm64/boot/dts/freescale/fsl-ls1012a.dtsi
index 21f2b3ba6b58..7ce0aa52292c 100644
--- a/arch/arm64/boot/dts/freescale/fsl-ls1012a.dtsi
+++ b/arch/arm64/boot/dts/freescale/fsl-ls1012a.dtsi
@@ -486,6 +486,7 @@
 			#size-cells = <2>;
 			device_type = "pci";
 			num-lanes = <4>;
+			num-viewport = <2>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x40 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x40 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
diff --git a/arch/arm64/boot/dts/freescale/fsl-ls1043a.dtsi b/arch/arm64/boot/dts/freescale/fsl-ls1043a.dtsi
index 760d510d78de..f42368afeea9 100644
--- a/arch/arm64/boot/dts/freescale/fsl-ls1043a.dtsi
+++ b/arch/arm64/boot/dts/freescale/fsl-ls1043a.dtsi
@@ -673,6 +673,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <4>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x40 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x40 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
@@ -699,6 +700,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <2>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x48 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x48 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
@@ -725,6 +727,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <2>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x50 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x50 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
diff --git a/arch/arm64/boot/dts/freescale/fsl-ls1046a.dtsi b/arch/arm64/boot/dts/freescale/fsl-ls1046a.dtsi
index 64d334c6b0b4..10be89168793 100644
--- a/arch/arm64/boot/dts/freescale/fsl-ls1046a.dtsi
+++ b/arch/arm64/boot/dts/freescale/fsl-ls1046a.dtsi
@@ -642,6 +642,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <4>;
+			num-viewport = <8>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x40 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x40 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
@@ -668,6 +669,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <2>;
+			num-viewport = <8>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x48 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x48 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
@@ -694,6 +696,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <2>;
+			num-viewport = <8>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x50 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x50 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
diff --git a/arch/arm64/boot/dts/freescale/fsl-ls1088a.dtsi b/arch/arm64/boot/dts/freescale/fsl-ls1088a.dtsi
index 9deb9cb83046..7d303bd40047 100644
--- a/arch/arm64/boot/dts/freescale/fsl-ls1088a.dtsi
+++ b/arch/arm64/boot/dts/freescale/fsl-ls1088a.dtsi
@@ -523,6 +523,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <4>;
+			num-viewport = <256>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x20 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x20 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
@@ -548,6 +549,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <4>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x28 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x28 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
@@ -573,6 +575,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <8>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			ranges = <0x81000000 0x0 0x00000000 0x30 0x00010000 0x0 0x00010000   /* downstream I/O */
 				  0x82000000 0x0 0x40000000 0x30 0x40000000 0x0 0x40000000>; /* non-prefetchable memory */
diff --git a/arch/arm64/boot/dts/freescale/fsl-ls208xa.dtsi b/arch/arm64/boot/dts/freescale/fsl-ls208xa.dtsi
index 5732e3b48be7..4bb9aa2dd8b2 100644
--- a/arch/arm64/boot/dts/freescale/fsl-ls208xa.dtsi
+++ b/arch/arm64/boot/dts/freescale/fsl-ls208xa.dtsi
@@ -640,6 +640,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <4>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			msi-parent = <&its>;
 			#interrupt-cells = <1>;
@@ -662,6 +663,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <4>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			msi-parent = <&its>;
 			#interrupt-cells = <1>;
@@ -684,6 +686,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <8>;
+			num-viewport = <256>;
 			bus-range = <0x0 0xff>;
 			msi-parent = <&its>;
 			#interrupt-cells = <1>;
@@ -706,6 +709,7 @@
 			device_type = "pci";
 			dma-coherent;
 			num-lanes = <4>;
+			num-viewport = <6>;
 			bus-range = <0x0 0xff>;
 			msi-parent = <&its>;
 			#interrupt-cells = <1>;
-- 
2.17.1

Re: [PATCHv3 2/5] arm64: dts: layerscape: add num-viewport property for PCIe DT nodes

[Shawn Guo]

On Tue, Dec 18, 2018 at 04:19:32AM +0000, Z.q. Hou wrote:
> From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
> 
> Add num-viewport property for PCIe DT nodes to specify how many
> viewports are implemented.
> 
> Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

Applied, thanks.

[PATCHv3 3/5] PCI: dwc: fix potential memory leak

[Z.q. Hou]

From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

To avoid memory leak on error return of the adjacent function
devm_of_pci_get_host_bridge_resources(), change to use
devm_pci_alloc_host_bridge() to allocate host bridge structure,
then it will be managed automatically.

Fixes: 295aeb98a322 ("PCI: designware: Convert PCI scan API to
pci_scan_root_bus_bridge()")
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
---
V3:
 - Changed to use devm_* to allocate host bridge. 
 - Added Fixes info.

 .../pci/controller/dwc/pcie-designware-host.c | 28 ++++++++-----------
 1 file changed, 11 insertions(+), 17 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
index 29a05759a294..33b5a3815d24 100644
--- a/drivers/pci/controller/dwc/pcie-designware-host.c
+++ b/drivers/pci/controller/dwc/pcie-designware-host.c
@@ -346,7 +346,7 @@ int dw_pcie_host_init(struct pcie_port *pp)
 		dev_err(dev, "Missing *config* reg space\n");
 	}
 
-	bridge = pci_alloc_host_bridge(0);
+	bridge = devm_pci_alloc_host_bridge(dev, 0);
 	if (!bridge)
 		return -ENOMEM;
 
@@ -357,7 +357,7 @@ int dw_pcie_host_init(struct pcie_port *pp)
 
 	ret = devm_request_pci_bus_resources(dev, &bridge->windows);
 	if (ret)
-		goto error;
+		return ret;
 
 	/* Get the I/O and memory ranges from DT */
 	resource_list_for_each_entry_safe(win, tmp, &bridge->windows) {
@@ -401,8 +401,7 @@ int dw_pcie_host_init(struct pcie_port *pp)
 						resource_size(pp->cfg));
 		if (!pci->dbi_base) {
 			dev_err(dev, "Error with ioremap\n");
-			ret = -ENOMEM;
-			goto error;
+			return -ENOMEM;
 		}
 	}
 
@@ -413,8 +412,7 @@ int dw_pcie_host_init(struct pcie_port *pp)
 					pp->cfg0_base, pp->cfg0_size);
 		if (!pp->va_cfg0_base) {
 			dev_err(dev, "Error with ioremap in function\n");
-			ret = -ENOMEM;
-			goto error;
+			return -ENOMEM;
 		}
 	}
 
@@ -424,8 +422,7 @@ int dw_pcie_host_init(struct pcie_port *pp)
 						pp->cfg1_size);
 		if (!pp->va_cfg1_base) {
 			dev_err(dev, "Error with ioremap\n");
-			ret = -ENOMEM;
-			goto error;
+			return -ENOMEM;
 		}
 	}
 
@@ -448,14 +445,14 @@ int dw_pcie_host_init(struct pcie_port *pp)
 			    pp->num_vectors == 0) {
 				dev_err(dev,
 					"Invalid number of vectors\n");
-				goto error;
+				return -EINVAL;
 			}
 		}
 
 		if (!pp->ops->msi_host_init) {
 			ret = dw_pcie_allocate_domains(pp);
 			if (ret)
-				goto error;
+				return ret;
 
 			if (pp->msi_irq)
 				irq_set_chained_handler_and_data(pp->msi_irq,
@@ -464,14 +461,14 @@ int dw_pcie_host_init(struct pcie_port *pp)
 		} else {
 			ret = pp->ops->msi_host_init(pp);
 			if (ret < 0)
-				goto error;
+				return ret;
 		}
 	}
 
 	if (pp->ops->host_init) {
 		ret = pp->ops->host_init(pp);
 		if (ret)
-			goto error;
+			return ret;
 	}
 
 	pp->root_bus_nr = pp->busn->start;
@@ -485,7 +482,7 @@ int dw_pcie_host_init(struct pcie_port *pp)
 
 	ret = pci_scan_root_bus_bridge(bridge);
 	if (ret)
-		goto error;
+		return ret;
 
 	bus = bridge->bus;
 
@@ -499,11 +496,8 @@ int dw_pcie_host_init(struct pcie_port *pp)
 		pcie_bus_configure_settings(child);
 
 	pci_bus_add_devices(bus);
-	return 0;
 
-error:
-	pci_free_host_bridge(bridge);
-	return ret;
+	return 0;
 }
 
 static int dw_pcie_rd_other_conf(struct pcie_port *pp, struct pci_bus *bus,
-- 
2.17.1

[PATCHv3 4/5] PCI: dwc: fix 4GiB outbound window size truncated to zero issue

[Z.q. Hou]

From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

The current type of mem_size is 'u32', so when resource_size()
return 4G it will be truncated to zero. This patch fix it by
changing its type to 'u64'.

Fixes: 340cba6092c2 ("pci: Add PCIe driver for Samsung Exynos")
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Acked-by: Gustavo Pimentel <gustavo.pimentel@synopsys.com>
---
V3:
 - No change 

 drivers/pci/controller/dwc/pcie-designware.c | 4 ++--
 drivers/pci/controller/dwc/pcie-designware.h | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-designware.c b/drivers/pci/controller/dwc/pcie-designware.c
index 2153956a0b20..7ac5989c23ef 100644
--- a/drivers/pci/controller/dwc/pcie-designware.c
+++ b/drivers/pci/controller/dwc/pcie-designware.c
@@ -106,7 +106,7 @@ static void dw_pcie_writel_ob_unroll(struct dw_pcie *pci, u32 index, u32 reg,
 
 static void dw_pcie_prog_outbound_atu_unroll(struct dw_pcie *pci, int index,
 					     int type, u64 cpu_addr,
-					     u64 pci_addr, u32 size)
+					     u64 pci_addr, u64 size)
 {
 	u32 retries, val;
 
@@ -141,7 +141,7 @@ static void dw_pcie_prog_outbound_atu_unroll(struct dw_pcie *pci, int index,
 }
 
 void dw_pcie_prog_outbound_atu(struct dw_pcie *pci, int index, int type,
-			       u64 cpu_addr, u64 pci_addr, u32 size)
+			       u64 cpu_addr, u64 pci_addr, u64 size)
 {
 	u32 retries, val;
 
diff --git a/drivers/pci/controller/dwc/pcie-designware.h b/drivers/pci/controller/dwc/pcie-designware.h
index 0989d880ac46..25604387d13e 100644
--- a/drivers/pci/controller/dwc/pcie-designware.h
+++ b/drivers/pci/controller/dwc/pcie-designware.h
@@ -157,7 +157,7 @@ struct pcie_port {
 	u32			io_size;
 	u64			mem_base;
 	phys_addr_t		mem_bus_addr;
-	u32			mem_size;
+	u64			mem_size;
 	struct resource		*cfg;
 	struct resource		*io;
 	struct resource		*mem;
@@ -242,7 +242,7 @@ int dw_pcie_link_up(struct dw_pcie *pci);
 int dw_pcie_wait_for_link(struct dw_pcie *pci);
 void dw_pcie_prog_outbound_atu(struct dw_pcie *pci, int index,
 			       int type, u64 cpu_addr, u64 pci_addr,
-			       u32 size);
+			       u64 size);
 int dw_pcie_prog_inbound_atu(struct dw_pcie *pci, int index, int bar,
 			     u64 cpu_addr, enum dw_pcie_as_type as_type);
 void dw_pcie_disable_atu(struct dw_pcie *pci, int index,
-- 
2.17.1

[PATCHv3 5/5] PCI: dwc: add prefetchable memory range support

[Z.q. Hou]

From: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>

The current code only support non-prefetchable memory range,
as the non-prefetchable memory range must not be greater than
4GiB, one viewport can cover it, which supports upto 4GiB.

To support prefetchable memory range, which is upto 64-bit
memory space and can be greater than 4GiB, so we need multiple
viewports. And added separate vars to store prefetchable memory
range info to prevent overriding the non-prefetchable memory
range info.

And this patch explicitly assigned the last (if there are only
2 viewports) or last 2 viewports for CFG and I/O windows and the
rests for MEM windows.

Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
---
V3:
 - Changed back to get num-viewport from DTS.
 - Added print info upon non-pref memory truncated.
 - Corrected typo.

 .../pci/controller/dwc/pcie-designware-host.c | 107 +++++++++++++++---
 drivers/pci/controller/dwc/pcie-designware.h  |   7 ++
 2 files changed, 97 insertions(+), 17 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
index 33b5a3815d24..2d1dd3dba1ba 100644
--- a/drivers/pci/controller/dwc/pcie-designware-host.c
+++ b/drivers/pci/controller/dwc/pcie-designware-host.c
@@ -346,6 +346,28 @@ int dw_pcie_host_init(struct pcie_port *pp)
 		dev_err(dev, "Missing *config* reg space\n");
 	}
 
+	ret = of_property_read_u32(np, "num-viewport", &pci->num_viewport);
+	if (ret || pci->num_viewport < 2)
+		pci->num_viewport = 2;
+
+	/*
+	 * if there are only 2 viewports, assign the last viewport for
+	 * both CFG and IO window, otherwise assign the last 2 viewport
+	 * for CFG and IO window specific. And the rest viewports are
+	 * assigned to MEM windows.
+	 */
+	if (pci->num_viewport == 2) {
+		pp->cfg_idx = pp->io_idx = PCIE_ATU_REGION_INDEX1;
+		pp->mem_wins = 1;
+	} else {
+		pp->cfg_idx = pci->num_viewport - 1;
+		pp->io_idx = pci->num_viewport - 2;
+		pp->mem_wins = pci->num_viewport - 2;
+	}
+
+	dev_dbg(dev, "CFG win id: %d, I/O win id: %d, Total MEM win: %d\n",
+		pp->cfg_idx, pp->io_idx, pp->mem_wins);
+
 	bridge = devm_pci_alloc_host_bridge(dev, 0);
 	if (!bridge)
 		return -ENOMEM;
@@ -377,10 +399,20 @@ int dw_pcie_host_init(struct pcie_port *pp)
 			}
 			break;
 		case IORESOURCE_MEM:
-			pp->mem = win->res;
-			pp->mem->name = "MEM";
-			pp->mem_size = resource_size(pp->mem);
-			pp->mem_bus_addr = pp->mem->start - win->offset;
+			if (win->res->flags & IORESOURCE_PREFETCH) {
+				pp->mem_pref = win->res;
+				pp->mem_pref->name = "MEM pref";
+				pp->mem_pref_size = resource_size(pp->mem_pref);
+				pp->mem_pref_bus_addr = pp->mem_pref->start -
+							win->offset;
+				pp->mem_pref_base = pp->mem_pref->start;
+			} else {
+				pp->mem = win->res;
+				pp->mem->name = "MEM";
+				pp->mem_size = resource_size(pp->mem);
+				pp->mem_bus_addr = pp->mem->start - win->offset;
+				pp->mem_base = pp->mem->start;
+			}
 			break;
 		case 0:
 			pp->cfg = win->res;
@@ -405,8 +437,6 @@ int dw_pcie_host_init(struct pcie_port *pp)
 		}
 	}
 
-	pp->mem_base = pp->mem->start;
-
 	if (!pp->va_cfg0_base) {
 		pp->va_cfg0_base = devm_pci_remap_cfgspace(dev,
 					pp->cfg0_base, pp->cfg0_size);
@@ -527,12 +557,12 @@ static int dw_pcie_rd_other_conf(struct pcie_port *pp, struct pci_bus *bus,
 		va_cfg_base = pp->va_cfg1_base;
 	}
 
-	dw_pcie_prog_outbound_atu(pci, PCIE_ATU_REGION_INDEX1,
+	dw_pcie_prog_outbound_atu(pci, pp->cfg_idx,
 				  type, cpu_addr,
 				  busdev, cfg_size);
 	ret = dw_pcie_read(va_cfg_base + where, size, val);
-	if (pci->num_viewport <= 2)
-		dw_pcie_prog_outbound_atu(pci, PCIE_ATU_REGION_INDEX1,
+	if (pp->cfg_idx == pp->io_idx)
+		dw_pcie_prog_outbound_atu(pci, pp->io_idx,
 					  PCIE_ATU_TYPE_IO, pp->io_base,
 					  pp->io_bus_addr, pp->io_size);
 
@@ -566,12 +596,12 @@ static int dw_pcie_wr_other_conf(struct pcie_port *pp, struct pci_bus *bus,
 		va_cfg_base = pp->va_cfg1_base;
 	}
 
-	dw_pcie_prog_outbound_atu(pci, PCIE_ATU_REGION_INDEX1,
+	dw_pcie_prog_outbound_atu(pci, pp->cfg_idx,
 				  type, cpu_addr,
 				  busdev, cfg_size);
 	ret = dw_pcie_write(va_cfg_base + where, size, val);
-	if (pci->num_viewport <= 2)
-		dw_pcie_prog_outbound_atu(pci, PCIE_ATU_REGION_INDEX1,
+	if (pp->cfg_idx == pp->io_idx)
+		dw_pcie_prog_outbound_atu(pci, pp->io_idx,
 					  PCIE_ATU_TYPE_IO, pp->io_base,
 					  pp->io_bus_addr, pp->io_size);
 
@@ -645,6 +675,9 @@ static u8 dw_pcie_iatu_unroll_enabled(struct dw_pcie *pci)
 void dw_pcie_setup_rc(struct pcie_port *pp)
 {
 	u32 val, ctrl, num_ctrls;
+	u64 unmapped_size, base, win_size;
+	phys_addr_t bus_addr;
+	int i;
 	struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
 
 	dw_pcie_setup(pci);
@@ -693,13 +726,53 @@ void dw_pcie_setup_rc(struct pcie_port *pp)
 		dev_dbg(pci->dev, "iATU unroll: %s\n",
 			pci->iatu_unroll_enabled ? "enabled" : "disabled");
 
+		/*
+		 * The maximum region size is 4 GB, and a region
+		 * must not cross a 4 GB boundary.
+		 */
+		win_size = SZ_4G - (pp->mem_base & (SZ_4G - 1));
+		win_size = min(win_size, pp->mem_size);
+		if (win_size < pp->mem_size)
+			dev_info(pci->dev,
+				 "iATU: non-pref MEM size is truncated to 0x%llx\n",
+				 win_size);
+
 		dw_pcie_prog_outbound_atu(pci, PCIE_ATU_REGION_INDEX0,
 					  PCIE_ATU_TYPE_MEM, pp->mem_base,
-					  pp->mem_bus_addr, pp->mem_size);
-		if (pci->num_viewport > 2)
-			dw_pcie_prog_outbound_atu(pci, PCIE_ATU_REGION_INDEX2,
-						  PCIE_ATU_TYPE_IO, pp->io_base,
-						  pp->io_bus_addr, pp->io_size);
+					  pp->mem_bus_addr, win_size);
+		dev_dbg(pci->dev,
+			"iATU: non-pref MEM: win = %d: base = 0x%llx, bus_addr = %pa, size = 0x%llx\n",
+			0, pp->mem_base, &pp->mem_bus_addr, win_size);
+
+		/*
+		 * Prefetchable memory range can be 64bit space,
+		 * so may need multiple viewports.
+		 */
+		unmapped_size = pp->mem_pref_size;
+		base = pp->mem_pref_base;
+		bus_addr = pp->mem_pref_bus_addr;
+		for (i = PCIE_ATU_REGION_INDEX1;
+		     unmapped_size > 0 && i < pp->mem_wins; i++) {
+			win_size = SZ_4G - (base & (SZ_4G - 1));
+			win_size = min(win_size, unmapped_size);
+			dw_pcie_prog_outbound_atu(pci, i, PCIE_ATU_TYPE_MEM,
+						  base, bus_addr, win_size);
+			dev_dbg(pci->dev,
+				"iATU: pref MEM: win = %d: base = 0x%llx, bus_addr = %pa, size = 0x%llx\n",
+				i, base, &bus_addr, win_size);
+
+			base += win_size;
+			bus_addr += win_size;
+			unmapped_size -= win_size;
+		}
+
+		if (unmapped_size > 0)
+			dev_info(pci->dev,
+				 "iATU: can't cover pref memory range\n");
+
+		dw_pcie_prog_outbound_atu(pci, pp->io_idx, PCIE_ATU_TYPE_IO,
+					  pp->io_base, pp->io_bus_addr,
+					  pp->io_size);
 	}
 
 	dw_pcie_wr_own_conf(pp, PCI_BASE_ADDRESS_0, 4, 0);
diff --git a/drivers/pci/controller/dwc/pcie-designware.h b/drivers/pci/controller/dwc/pcie-designware.h
index 25604387d13e..1e87f18bc417 100644
--- a/drivers/pci/controller/dwc/pcie-designware.h
+++ b/drivers/pci/controller/dwc/pcie-designware.h
@@ -152,15 +152,22 @@ struct pcie_port {
 	u64			cfg1_base;
 	void __iomem		*va_cfg1_base;
 	u32			cfg1_size;
+	u32			cfg_idx;
 	resource_size_t		io_base;
 	phys_addr_t		io_bus_addr;
 	u32			io_size;
+	u32			io_idx;
 	u64			mem_base;
 	phys_addr_t		mem_bus_addr;
 	u64			mem_size;
+	phys_addr_t		mem_pref_base;
+	pci_bus_addr_t		mem_pref_bus_addr;
+	u64			mem_pref_size;
+	u32			mem_wins;
 	struct resource		*cfg;
 	struct resource		*io;
 	struct resource		*mem;
+	struct resource		*mem_pref;
 	struct resource		*busn;
 	int			irq;
 	const struct dw_pcie_host_ops *ops;
-- 
2.17.1