Armada 3720 PCIe fixes for 4.14

index a572b2fb7af8..fd605796b011 100644
@@ -932,7 +932,7 @@ static int advk_pcie_parse_request_of_pci_ranges(struct advk_pcie *pcie)
 					     0,	0xF8000000, 0,
 					     lower_32_bits(res->start),
 					     OB_PCIE_IO);
-			err = pci_remap_iospace(res, iobase);
+			err = devm_pci_remap_iospace(dev, res, iobase);
 			if (err) {
 				dev_warn(dev, "error %d: failed to map resource %pR\n",
 					 err, res);

index fd605796b011..79cd11b1c89a 100644
@@ -789,6 +789,7 @@ static int advk_pcie_init_irq_domain(struct advk_pcie *pcie)
 	struct device_node *node = dev->of_node;
 	struct device_node *pcie_intc_node;
 	struct irq_chip *irq_chip;
+	int ret = 0;
 
 	raw_spin_lock_init(&pcie->irq_lock);
 

@@ -803,8 +804,8 @@ static int advk_pcie_init_irq_domain(struct advk_pcie *pcie)
 	irq_chip->name = devm_kasprintf(dev, GFP_KERNEL, "%s-irq",
 					dev_name(dev));
 	if (!irq_chip->name) {
-		of_node_put(pcie_intc_node);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto out_put_node;
 	}
 
 	irq_chip->irq_mask = advk_pcie_irq_mask;

@@ -816,11 +817,13 @@ static int advk_pcie_init_irq_domain(struct advk_pcie *pcie)
 				      &advk_pcie_irq_domain_ops, pcie);
 	if (!pcie->irq_domain) {
 		dev_err(dev, "Failed to get a INTx IRQ domain\n");
-		of_node_put(pcie_intc_node);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto out_put_node;
 	}
 
-	return 0;
+out_put_node:
+	of_node_put(pcie_intc_node);
+	return ret;
 }
 
 static void advk_pcie_remove_irq_domain(struct advk_pcie *pcie)

@@ -789,6 +789,7 @@ static int advk_pcie_init_irq_domain(str
 	struct device_node *node = dev->of_node;
 	struct device_node *pcie_intc_node;
 	struct irq_chip *irq_chip;
+	int ret = 0;
 
 	raw_spin_lock_init(&pcie->irq_lock);
 

@@ -803,8 +804,8 @@ static int advk_pcie_init_irq_domain(str
 	irq_chip->name = devm_kasprintf(dev, GFP_KERNEL, "%s-irq",
 					dev_name(dev));
 	if (!irq_chip->name) {
-		of_node_put(pcie_intc_node);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto out_put_node;
 	}
 
 	irq_chip->irq_mask = advk_pcie_irq_mask;

@@ -816,11 +817,13 @@ static int advk_pcie_init_irq_domain(str
 				      &advk_pcie_irq_domain_ops, pcie);
 	if (!pcie->irq_domain) {
 		dev_err(dev, "Failed to get a INTx IRQ domain\n");
-		of_node_put(pcie_intc_node);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto out_put_node;
 	}
 
-	return 0;
+out_put_node:
+	of_node_put(pcie_intc_node);
+	return ret;
 }
 
 static void advk_pcie_remove_irq_domain(struct advk_pcie *pcie)

index 79cd11b1c89a..7ee5a91e5f7f 100644
@@ -362,6 +362,14 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	reg |= PIO_CTRL_ADDR_WIN_DISABLE;
 	advk_writel(pcie, reg, PIO_CTRL);
 
+	/*
+	 * PERST# signal could have been asserted by pinctrl subsystem before
+	 * probe() callback has been called, making the endpoint going into
+	 * fundamental reset. As required by PCI Express spec a delay for at
+	 * least 100ms after such a reset before link training is needed.
+	 */
+	msleep(PCI_PM_D3COLD_WAIT);
+
 	/* Start link training */
 	reg = advk_readl(pcie, PCIE_CORE_LINK_CTRL_STAT_REG);
 	reg |= PCIE_CORE_LINK_TRAINING;

index 7ee5a91e5f7f..8373f8cc4c52 100644
@@ -324,11 +324,6 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	reg |= LANE_COUNT_1;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
 
-	/* Enable link training */
-	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
-	reg |= LINK_TRAINING_EN;
-	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
-
 	/* Enable MSI */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL2_REG);
 	reg |= PCIE_CORE_CTRL2_MSI_ENABLE;

@@ -370,7 +365,15 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	 */
 	msleep(PCI_PM_D3COLD_WAIT);
 
-	/* Start link training */
+	/* Enable link training */
+	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
+	reg |= LINK_TRAINING_EN;
+	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
+
+	/*
+	 * Start link training immediately after enabling it.
+	 * This solves problems for some buggy cards.
+	 */
 	reg = advk_readl(pcie, PCIE_CORE_LINK_CTRL_STAT_REG);
 	reg |= PCIE_CORE_LINK_TRAINING;
 	advk_writel(pcie, reg, PCIE_CORE_LINK_CTRL_STAT_REG);

index 8373f8cc4c52..714d81462297 100644
@@ -36,6 +36,7 @@
 #define PCIE_CORE_LINK_CTRL_STAT_REG				0xd0
 #define     PCIE_CORE_LINK_L0S_ENTRY				BIT(0)
 #define     PCIE_CORE_LINK_TRAINING				BIT(5)
+#define     PCIE_CORE_LINK_SPEED_SHIFT				16
 #define     PCIE_CORE_LINK_WIDTH_SHIFT				20
 #define PCIE_CORE_ERR_CAPCTL_REG				0x118
 #define     PCIE_CORE_ERR_CAPCTL_ECRC_CHK_TX			BIT(5)

@@ -212,6 +213,7 @@ struct advk_pcie {
 	struct mutex msi_used_lock;
 	u16 msi_msg;
 	int root_bus_nr;
+	int link_gen;
 };
 
 static inline void advk_writel(struct advk_pcie *pcie, u32 val, u64 reg)

@@ -235,20 +237,16 @@ static int advk_pcie_link_up(struct advk_pcie *pcie)
 
 static int advk_pcie_wait_for_link(struct advk_pcie *pcie)
 {
-	struct device *dev = &pcie->pdev->dev;
 	int retries;
 
 	/* check if the link is up or not */
 	for (retries = 0; retries < LINK_WAIT_MAX_RETRIES; retries++) {
-		if (advk_pcie_link_up(pcie)) {
-			dev_info(dev, "link up\n");
+		if (advk_pcie_link_up(pcie))
 			return 0;
-		}
 
 		usleep_range(LINK_WAIT_USLEEP_MIN, LINK_WAIT_USLEEP_MAX);
 	}
 
-	dev_err(dev, "link never came up\n");
 	return -ETIMEDOUT;
 }
 

@@ -272,6 +270,85 @@ static void advk_pcie_set_ob_win(struct advk_pcie *pcie,
 	advk_writel(pcie, match_ls | BIT(0), OB_WIN_MATCH_LS(win_num));
 }
 
+static int advk_pcie_train_at_gen(struct advk_pcie *pcie, int gen)
+{
+	int ret, neg_gen;
+	u32 reg;
+
+	/* Setup link speed */
+	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
+	reg &= ~PCIE_GEN_SEL_MSK;
+	if (gen == 3)
+		reg |= SPEED_GEN_3;
+	else if (gen == 2)
+		reg |= SPEED_GEN_2;
+	else
+		reg |= SPEED_GEN_1;
+	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
+
+	/*
+	 * Enable link training. This is not needed in every call to this
+	 * function, just once suffices, but it does not break anything either.
+	 */
+	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
+	reg |= LINK_TRAINING_EN;
+	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
+
+	/*
+	 * Start link training immediately after enabling it.
+	 * This solves problems for some buggy cards.
+	 */
+	reg = advk_readl(pcie, PCIE_CORE_LINK_CTRL_STAT_REG);
+	reg |= PCIE_CORE_LINK_TRAINING;
+	advk_writel(pcie, reg, PCIE_CORE_LINK_CTRL_STAT_REG);
+
+	ret = advk_pcie_wait_for_link(pcie);
+	if (ret)
+		return ret;
+
+	reg = advk_readl(pcie, PCIE_CORE_LINK_CTRL_STAT_REG);
+	neg_gen = (reg >> PCIE_CORE_LINK_SPEED_SHIFT) & 0xf;
+
+	return neg_gen;
+}
+
+static void advk_pcie_train_link(struct advk_pcie *pcie)
+{
+	struct device *dev = &pcie->pdev->dev;
+	int neg_gen = -1, gen;
+
+	/*
+	 * Try link training at link gen specified by device tree property
+	 * 'max-link-speed'. If this fails, iteratively train at lower gen.
+	 */
+	for (gen = pcie->link_gen; gen > 0; --gen) {
+		neg_gen = advk_pcie_train_at_gen(pcie, gen);
+		if (neg_gen > 0)
+			break;
+	}
+
+	if (neg_gen < 0)
+		goto err;
+
+	/*
+	 * After successful training if negotiated gen is lower than requested,
+	 * train again on negotiated gen. This solves some stability issues for
+	 * some buggy gen1 cards.
+	 */
+	if (neg_gen < gen) {
+		gen = neg_gen;
+		neg_gen = advk_pcie_train_at_gen(pcie, gen);
+	}
+
+	if (neg_gen == gen) {
+		dev_info(dev, "link up at gen %i\n", gen);
+		return;
+	}
+
+err:
+	dev_err(dev, "link never came up\n");
+}
+
 static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 {
 	u32 reg;

@@ -312,12 +389,6 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 		PCIE_CORE_CTRL2_TD_ENABLE;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL2_REG);
 
-	/* Set GEN2 */
-	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
-	reg &= ~PCIE_GEN_SEL_MSK;
-	reg |= SPEED_GEN_2;
-	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
-
 	/* Set lane X1 */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
 	reg &= ~LANE_CNT_MSK;

@@ -365,20 +436,7 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	 */
 	msleep(PCI_PM_D3COLD_WAIT);
 
-	/* Enable link training */
-	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
-	reg |= LINK_TRAINING_EN;
-	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
-
-	/*
-	 * Start link training immediately after enabling it.
-	 * This solves problems for some buggy cards.
-	 */
-	reg = advk_readl(pcie, PCIE_CORE_LINK_CTRL_STAT_REG);
-	reg |= PCIE_CORE_LINK_TRAINING;
-	advk_writel(pcie, reg, PCIE_CORE_LINK_CTRL_STAT_REG);
-
-	advk_pcie_wait_for_link(pcie);
+	advk_pcie_train_link(pcie);
 
 	reg = advk_readl(pcie, PCIE_CORE_CMD_STATUS_REG);
 	reg |= PCIE_CORE_CMD_MEM_ACCESS_EN |

@@ -1017,6 +1075,12 @@ static int advk_pcie_probe(struct platform_device *pdev)
 		return ret;
 	}
 
+	ret = of_pci_get_max_link_speed(dev->of_node);
+	if (ret <= 0 || ret > 3)
+		pcie->link_gen = 3;
+	else
+		pcie->link_gen = ret;
+
 	advk_pcie_setup_hw(pcie);
 
 	ret = advk_pcie_init_irq_domain(pcie);

index 714d81462297..0cfec5660c69 100644
@@ -12,6 +12,7 @@
  */
 
 #include <linux/delay.h>
+#include <linux/gpio.h>
 #include <linux/interrupt.h>
 #include <linux/irq.h>
 #include <linux/irqdomain.h>

@@ -20,6 +21,7 @@
 #include <linux/init.h>
 #include <linux/platform_device.h>
 #include <linux/of_address.h>
+#include <linux/of_gpio.h>
 #include <linux/of_pci.h>
 
 /* PCIe core registers */

@@ -214,6 +216,7 @@ struct advk_pcie {
 	u16 msi_msg;
 	int root_bus_nr;
 	int link_gen;
+	struct gpio_desc *reset_gpio;
 };
 
 static inline void advk_writel(struct advk_pcie *pcie, u32 val, u64 reg)

@@ -349,6 +352,25 @@ static void advk_pcie_train_link(struct advk_pcie *pcie)
 	dev_err(dev, "link never came up\n");
 }
 
+static void advk_pcie_issue_perst(struct advk_pcie *pcie)
+{
+	u32 reg;
+
+	if (!pcie->reset_gpio)
+		return;
+
+	/* PERST does not work for some cards when link training is enabled */
+	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
+	reg &= ~LINK_TRAINING_EN;
+	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
+
+	/* 10ms delay is needed for some cards */
+	dev_info(&pcie->pdev->dev, "issuing PERST via reset GPIO for 10ms\n");
+	gpiod_set_value_cansleep(pcie->reset_gpio, 1);
+	usleep_range(10000, 11000);
+	gpiod_set_value_cansleep(pcie->reset_gpio, 0);
+}
+
 static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 {
 	u32 reg;

@@ -358,6 +380,8 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	for (i = 0; i < 8; i++)
 		advk_pcie_set_ob_win(pcie, i, 0, 0, 0, 0, 0, 0, 0);
 
+	advk_pcie_issue_perst(pcie);
+
 	/* Set to Direct mode */
 	reg = advk_readl(pcie, CTRL_CONFIG_REG);
 	reg &= ~(CTRL_MODE_MASK << CTRL_MODE_SHIFT);

@@ -430,7 +454,8 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 
 	/*
 	 * PERST# signal could have been asserted by pinctrl subsystem before
-	 * probe() callback has been called, making the endpoint going into
+	 * probe() callback has been called or issued explicitly by reset gpio
+	 * function advk_pcie_issue_perst(), making the endpoint going into
 	 * fundamental reset. As required by PCI Express spec a delay for at
 	 * least 100ms after such a reset before link training is needed.
 	 */

@@ -1075,6 +1100,23 @@ static int advk_pcie_probe(struct platform_device *pdev)
 		return ret;
 	}
 
+	pcie->reset_gpio = devm_fwnode_get_index_gpiod_from_child(dev, "reset",
+								  0,
+								  dev_fwnode(dev),
+								  GPIOD_OUT_LOW,
+								  "pcie1-reset");
+	ret = PTR_ERR_OR_ZERO(pcie->reset_gpio);
+	if (ret) {
+		if (ret == -ENOENT) {
+			pcie->reset_gpio = NULL;
+		} else {
+			if (ret != -EPROBE_DEFER)
+				dev_err(dev, "Failed to get reset-gpio: %i\n",
+					ret);
+			return ret;
+		}
+	}
+
 	ret = of_pci_get_max_link_speed(dev->of_node);
 	if (ret <= 0 || ret > 3)
 		pcie->link_gen = 3;

index 0cfec5660c69..3ad6b6245f94 100644
@@ -29,17 +29,7 @@
 #define     PCIE_CORE_CMD_IO_ACCESS_EN				BIT(0)
 #define     PCIE_CORE_CMD_MEM_ACCESS_EN				BIT(1)
 #define     PCIE_CORE_CMD_MEM_IO_REQ_EN				BIT(2)
-#define PCIE_CORE_DEV_CTRL_STATS_REG				0xc8
-#define     PCIE_CORE_DEV_CTRL_STATS_RELAX_ORDER_DISABLE	(0 << 4)
-#define     PCIE_CORE_DEV_CTRL_STATS_MAX_PAYLOAD_SZ_SHIFT	5
-#define     PCIE_CORE_DEV_CTRL_STATS_SNOOP_DISABLE		(0 << 11)
-#define     PCIE_CORE_DEV_CTRL_STATS_MAX_RD_REQ_SIZE_SHIFT	12
-#define     PCIE_CORE_DEV_CTRL_STATS_MAX_RD_REQ_SZ		0x2
-#define PCIE_CORE_LINK_CTRL_STAT_REG				0xd0
-#define     PCIE_CORE_LINK_L0S_ENTRY				BIT(0)
-#define     PCIE_CORE_LINK_TRAINING				BIT(5)
-#define     PCIE_CORE_LINK_SPEED_SHIFT				16
-#define     PCIE_CORE_LINK_WIDTH_SHIFT				20
+#define PCIE_CORE_PCIEXP_CAP					0xc0
 #define PCIE_CORE_ERR_CAPCTL_REG				0x118
 #define     PCIE_CORE_ERR_CAPCTL_ECRC_CHK_TX			BIT(5)
 #define     PCIE_CORE_ERR_CAPCTL_ECRC_CHK_TX_EN			BIT(6)

@@ -229,6 +219,11 @@ static inline u32 advk_readl(struct advk_pcie *pcie, u64 reg)
 	return readl(pcie->base + reg);
 }
 
+static inline u16 advk_read16(struct advk_pcie *pcie, u64 reg)
+{
+	return advk_readl(pcie, (reg & ~0x3)) >> ((reg & 0x3) * 8);
+}
+
 static int advk_pcie_link_up(struct advk_pcie *pcie)
 {
 	u32 val, ltssm_state;

@@ -301,16 +296,16 @@ static int advk_pcie_train_at_gen(struct advk_pcie *pcie, int gen)
 	 * Start link training immediately after enabling it.
 	 * This solves problems for some buggy cards.
 	 */
-	reg = advk_readl(pcie, PCIE_CORE_LINK_CTRL_STAT_REG);
-	reg |= PCIE_CORE_LINK_TRAINING;
-	advk_writel(pcie, reg, PCIE_CORE_LINK_CTRL_STAT_REG);
+	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL);
+	reg |= PCI_EXP_LNKCTL_RL;
+	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL);
 
 	ret = advk_pcie_wait_for_link(pcie);
 	if (ret)
 		return ret;
 
-	reg = advk_readl(pcie, PCIE_CORE_LINK_CTRL_STAT_REG);
-	neg_gen = (reg >> PCIE_CORE_LINK_SPEED_SHIFT) & 0xf;
+	reg = advk_read16(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKSTA);
+	neg_gen = reg & PCI_EXP_LNKSTA_CLS;
 
 	return neg_gen;
 }

@@ -400,13 +395,14 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 		PCIE_CORE_ERR_CAPCTL_ECRC_CHCK_RCV;
 	advk_writel(pcie, reg, PCIE_CORE_ERR_CAPCTL_REG);
 
-	/* Set PCIe Device Control and Status 1 PF0 register */
-	reg = PCIE_CORE_DEV_CTRL_STATS_RELAX_ORDER_DISABLE |
-		(7 << PCIE_CORE_DEV_CTRL_STATS_MAX_PAYLOAD_SZ_SHIFT) |
-		PCIE_CORE_DEV_CTRL_STATS_SNOOP_DISABLE |
-		(PCIE_CORE_DEV_CTRL_STATS_MAX_RD_REQ_SZ <<
-		 PCIE_CORE_DEV_CTRL_STATS_MAX_RD_REQ_SIZE_SHIFT);
-	advk_writel(pcie, reg, PCIE_CORE_DEV_CTRL_STATS_REG);
+	/* Set PCIe Device Control register */
+	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_DEVCTL);
+	reg &= ~PCI_EXP_DEVCTL_RELAX_EN;
+	reg &= ~PCI_EXP_DEVCTL_NOSNOOP_EN;
+	reg &= ~PCI_EXP_DEVCTL_READRQ;
+	reg |= PCI_EXP_DEVCTL_PAYLOAD; /* Set max payload size */
+	reg |= PCI_EXP_DEVCTL_READRQ_512B;
+	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_DEVCTL);
 
 	/* Program PCIe Control 2 to disable strict ordering */
 	reg = PCIE_CORE_CTRL2_RESERVED |

index 3ad6b6245f94..ae67e5c3fe70 100644
@@ -631,8 +631,10 @@ static int advk_pcie_rd_conf(struct pci_bus *bus, u32 devfn,
 	advk_writel(pcie, 1, PIO_START);
 
 	ret = advk_pcie_wait_pio(pcie);
-	if (ret < 0)
+	if (ret < 0) {
+		*val = 0xffffffff;
 		return PCIBIOS_SET_FAILED;
+	}
 
 	/* Check PIO status and get the read result */
 	ret = advk_pcie_check_pio_status(pcie, val);

index ae67e5c3fe70..9426715316b2 100644
@@ -592,6 +592,15 @@ static bool advk_pcie_pio_is_running(struct advk_pcie *pcie)
 	return false;
 }
 
+static bool advk_pcie_valid_device(struct advk_pcie *pcie, struct pci_bus *bus,
+				  int devfn)
+{
+	if ((bus->number == pcie->root_bus_nr) && PCI_SLOT(devfn) != 0)
+		return false;
+
+	return true;
+}
+
 static int advk_pcie_rd_conf(struct pci_bus *bus, u32 devfn,
 			     int where, int size, u32 *val)
 {

@@ -599,7 +608,7 @@ static int advk_pcie_rd_conf(struct pci_bus *bus, u32 devfn,
 	u32 reg;
 	int ret;
 
-	if ((bus->number == pcie->root_bus_nr) && PCI_SLOT(devfn) != 0) {
+	if (!advk_pcie_valid_device(pcie, bus, devfn)) {
 		*val = 0xffffffff;
 		return PCIBIOS_DEVICE_NOT_FOUND;
 	}

@@ -660,7 +669,7 @@ static int advk_pcie_wr_conf(struct pci_bus *bus, u32 devfn,
 	int offset;
 	int ret;
 
-	if ((bus->number == pcie->root_bus_nr) && PCI_SLOT(devfn) != 0)
+	if (!advk_pcie_valid_device(pcie, bus, devfn))
 		return PCIBIOS_DEVICE_NOT_FOUND;
 
 	if (where % size)

index 9426715316b2..acc04c4a37df 100644
@@ -598,6 +598,13 @@ static bool advk_pcie_valid_device(struct advk_pcie *pcie, struct pci_bus *bus,
 	if ((bus->number == pcie->root_bus_nr) && PCI_SLOT(devfn) != 0)
 		return false;
 
+	/*
+	 * If the link goes down after we check for link-up, nothing bad
+	 * happens but the config access times out.
+	 */
+	if (bus->number != pcie->root_bus_nr && !advk_pcie_link_up(pcie))
+		return false;
+
 	return true;
 }
 

index acc04c4a37df..b3bce506d995 100644
@@ -12,7 +12,7 @@
  */
 
 #include <linux/delay.h>
-#include <linux/gpio.h>
+#include <linux/gpio/consumer.h>
 #include <linux/interrupt.h>
 #include <linux/irq.h>
 #include <linux/irqdomain.h>

index b3bce506d995..68ab3aef0e12 100644
@@ -268,6 +268,25 @@ static void advk_pcie_set_ob_win(struct advk_pcie *pcie,
 	advk_writel(pcie, match_ls | BIT(0), OB_WIN_MATCH_LS(win_num));
 }
 
+static void advk_pcie_issue_perst(struct advk_pcie *pcie)
+{
+	u32 reg;
+
+	if (!pcie->reset_gpio)
+		return;
+
+	/* PERST does not work for some cards when link training is enabled */
+	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
+	reg &= ~LINK_TRAINING_EN;
+	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
+
+	/* 10ms delay is needed for some cards */
+	dev_info(&pcie->pdev->dev, "issuing PERST via reset GPIO for 10ms\n");
+	gpiod_set_value_cansleep(pcie->reset_gpio, 1);
+	usleep_range(10000, 11000);
+	gpiod_set_value_cansleep(pcie->reset_gpio, 0);
+}
+
 static int advk_pcie_train_at_gen(struct advk_pcie *pcie, int gen)
 {
 	int ret, neg_gen;

@@ -315,6 +334,21 @@ static void advk_pcie_train_link(struct advk_pcie *pcie)
 	struct device *dev = &pcie->pdev->dev;
 	int neg_gen = -1, gen;
 
+	/*
+	 * Reset PCIe card via PERST# signal. Some cards are not detected
+	 * during link training when they are in some non-initial state.
+	 */
+	advk_pcie_issue_perst(pcie);
+
+	/*
+	 * PERST# signal could have been asserted by pinctrl subsystem before
+	 * probe() callback has been called or issued explicitly by reset gpio
+	 * function advk_pcie_issue_perst(), making the endpoint going into
+	 * fundamental reset. As required by PCI Express spec a delay for at
+	 * least 100ms after such a reset before link training is needed.
+	 */
+	msleep(PCI_PM_D3COLD_WAIT);
+
 	/*
 	 * Try link training at link gen specified by device tree property
 	 * 'max-link-speed'. If this fails, iteratively train at lower gen.

@@ -347,25 +381,6 @@ static void advk_pcie_train_link(struct advk_pcie *pcie)
 	dev_err(dev, "link never came up\n");
 }
 
-static void advk_pcie_issue_perst(struct advk_pcie *pcie)
-{
-	u32 reg;
-
-	if (!pcie->reset_gpio)
-		return;
-
-	/* PERST does not work for some cards when link training is enabled */
-	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
-	reg &= ~LINK_TRAINING_EN;
-	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
-
-	/* 10ms delay is needed for some cards */
-	dev_info(&pcie->pdev->dev, "issuing PERST via reset GPIO for 10ms\n");
-	gpiod_set_value_cansleep(pcie->reset_gpio, 1);
-	usleep_range(10000, 11000);
-	gpiod_set_value_cansleep(pcie->reset_gpio, 0);
-}
-
 static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 {
 	u32 reg;

@@ -375,8 +390,6 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	for (i = 0; i < 8; i++)
 		advk_pcie_set_ob_win(pcie, i, 0, 0, 0, 0, 0, 0, 0);
 
-	advk_pcie_issue_perst(pcie);
-
 	/* Set to Direct mode */
 	reg = advk_readl(pcie, CTRL_CONFIG_REG);
 	reg &= ~(CTRL_MODE_MASK << CTRL_MODE_SHIFT);

@@ -448,15 +461,6 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	reg |= PIO_CTRL_ADDR_WIN_DISABLE;
 	advk_writel(pcie, reg, PIO_CTRL);
 
-	/*
-	 * PERST# signal could have been asserted by pinctrl subsystem before
-	 * probe() callback has been called or issued explicitly by reset gpio
-	 * function advk_pcie_issue_perst(), making the endpoint going into
-	 * fundamental reset. As required by PCI Express spec a delay for at
-	 * least 100ms after such a reset before link training is needed.
-	 */
-	msleep(PCI_PM_D3COLD_WAIT);
-
 	advk_pcie_train_link(pcie);
 
 	reg = advk_readl(pcie, PCIE_CORE_CMD_STATUS_REG);

index 68ab3aef0e12..66c4b9282420 100644
@@ -275,7 +275,14 @@ static void advk_pcie_issue_perst(struct advk_pcie *pcie)
 	if (!pcie->reset_gpio)
 		return;
 
-	/* PERST does not work for some cards when link training is enabled */
+	/*
+	 * As required by PCI Express spec (PCI Express Base Specification, REV.
+	 * 4.0 PCI Express, February 19 2014, 6.6.1 Conventional Reset) a delay
+	 * for at least 100ms after de-asserting PERST# signal is needed before
+	 * link training is enabled. So ensure that link training is disabled
+	 * prior de-asserting PERST# signal to fulfill that PCI Express spec
+	 * requirement.
+	 */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
 	reg &= ~LINK_TRAINING_EN;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);

index 66c4b9282420..4021c4d1c6e8 100644
@@ -107,24 +107,6 @@
 #define PCIE_MSI_PAYLOAD_REG			(CONTROL_BASE_ADDR + 0x9C)
 #define     PCIE_MSI_DATA_MASK			GENMASK(15, 0)
 
-/* PCIe window configuration */
-#define OB_WIN_BASE_ADDR			0x4c00
-#define OB_WIN_BLOCK_SIZE			0x20
-#define OB_WIN_REG_ADDR(win, offset)		(OB_WIN_BASE_ADDR + \
-						 OB_WIN_BLOCK_SIZE * (win) + \
-						 (offset))
-#define OB_WIN_MATCH_LS(win)			OB_WIN_REG_ADDR(win, 0x00)
-#define OB_WIN_MATCH_MS(win)			OB_WIN_REG_ADDR(win, 0x04)
-#define OB_WIN_REMAP_LS(win)			OB_WIN_REG_ADDR(win, 0x08)
-#define OB_WIN_REMAP_MS(win)			OB_WIN_REG_ADDR(win, 0x0c)
-#define OB_WIN_MASK_LS(win)			OB_WIN_REG_ADDR(win, 0x10)
-#define OB_WIN_MASK_MS(win)			OB_WIN_REG_ADDR(win, 0x14)
-#define OB_WIN_ACTIONS(win)			OB_WIN_REG_ADDR(win, 0x18)
-
-/* PCIe window types */
-#define OB_PCIE_MEM				0x0
-#define OB_PCIE_IO				0x4
-
 /* LMI registers base address and register offsets */
 #define LMI_BASE_ADDR				0x6000
 #define CFG_REG					(LMI_BASE_ADDR + 0x0)

@@ -248,26 +230,6 @@ static int advk_pcie_wait_for_link(struct advk_pcie *pcie)
 	return -ETIMEDOUT;
 }
 
-/*
- * Set PCIe address window register which could be used for memory
- * mapping.
- */
-static void advk_pcie_set_ob_win(struct advk_pcie *pcie,
-				 u32 win_num, u32 match_ms,
-				 u32 match_ls, u32 mask_ms,
-				 u32 mask_ls, u32 remap_ms,
-				 u32 remap_ls, u32 action)
-{
-	advk_writel(pcie, match_ls, OB_WIN_MATCH_LS(win_num));
-	advk_writel(pcie, match_ms, OB_WIN_MATCH_MS(win_num));
-	advk_writel(pcie, mask_ms, OB_WIN_MASK_MS(win_num));
-	advk_writel(pcie, mask_ls, OB_WIN_MASK_LS(win_num));
-	advk_writel(pcie, remap_ms, OB_WIN_REMAP_MS(win_num));
-	advk_writel(pcie, remap_ls, OB_WIN_REMAP_LS(win_num));
-	advk_writel(pcie, action, OB_WIN_ACTIONS(win_num));
-	advk_writel(pcie, match_ls | BIT(0), OB_WIN_MATCH_LS(win_num));
-}
-
 static void advk_pcie_issue_perst(struct advk_pcie *pcie)
 {
 	u32 reg;

@@ -391,11 +353,6 @@ static void advk_pcie_train_link(struct advk_pcie *pcie)
 static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 {
 	u32 reg;
-	int i;
-
-	/* Point PCIe unit MBUS decode windows to DRAM space */
-	for (i = 0; i < 8; i++)
-		advk_pcie_set_ob_win(pcie, i, 0, 0, 0, 0, 0, 0, 0);
 
 	/* Set to Direct mode */
 	reg = advk_readl(pcie, CTRL_CONFIG_REG);

@@ -1048,12 +1005,6 @@ static int advk_pcie_parse_request_of_pci_ranges(struct advk_pcie *pcie)
 
 		switch (resource_type(res)) {
 		case IORESOURCE_IO:
-			advk_pcie_set_ob_win(pcie, 1,
-					     upper_32_bits(res->start),
-					     lower_32_bits(res->start),
-					     0,	0xF8000000, 0,
-					     lower_32_bits(res->start),
-					     OB_PCIE_IO);
 			err = devm_pci_remap_iospace(dev, res, iobase);
 			if (err) {
 				dev_warn(dev, "error %d: failed to map resource %pR\n",

@@ -1062,12 +1013,6 @@ static int advk_pcie_parse_request_of_pci_ranges(struct advk_pcie *pcie)
 			}
 			break;
 		case IORESOURCE_MEM:
-			advk_pcie_set_ob_win(pcie, 0,
-					     upper_32_bits(res->start),
-					     lower_32_bits(res->start),
-					     0x0, 0xF8000000, 0,
-					     lower_32_bits(res->start),
-					     (2 << 20) | OB_PCIE_MEM);
 			res_valid |= !(res->flags & IORESOURCE_PREFETCH);
 			break;
 		case IORESOURCE_BUS:

index 4021c4d1c6e8..fd5155b10c1d 100644
@@ -107,6 +107,46 @@
 #define PCIE_MSI_PAYLOAD_REG			(CONTROL_BASE_ADDR + 0x9C)
 #define     PCIE_MSI_DATA_MASK			GENMASK(15, 0)
 
+/* PCIe window configuration */
+#define OB_WIN_BASE_ADDR			0x4c00
+#define OB_WIN_BLOCK_SIZE			0x20
+#define OB_WIN_COUNT				8
+#define OB_WIN_REG_ADDR(win, offset)		(OB_WIN_BASE_ADDR + \
+						 OB_WIN_BLOCK_SIZE * (win) + \
+						 (offset))
+#define OB_WIN_MATCH_LS(win)			OB_WIN_REG_ADDR(win, 0x00)
+#define     OB_WIN_ENABLE			BIT(0)
+#define OB_WIN_MATCH_MS(win)			OB_WIN_REG_ADDR(win, 0x04)
+#define OB_WIN_REMAP_LS(win)			OB_WIN_REG_ADDR(win, 0x08)
+#define OB_WIN_REMAP_MS(win)			OB_WIN_REG_ADDR(win, 0x0c)
+#define OB_WIN_MASK_LS(win)			OB_WIN_REG_ADDR(win, 0x10)
+#define OB_WIN_MASK_MS(win)			OB_WIN_REG_ADDR(win, 0x14)
+#define OB_WIN_ACTIONS(win)			OB_WIN_REG_ADDR(win, 0x18)
+#define OB_WIN_DEFAULT_ACTIONS			(OB_WIN_ACTIONS(OB_WIN_COUNT-1) + 0x4)
+#define     OB_WIN_FUNC_NUM_MASK		GENMASK(31, 24)
+#define     OB_WIN_FUNC_NUM_SHIFT		24
+#define     OB_WIN_FUNC_NUM_ENABLE		BIT(23)
+#define     OB_WIN_BUS_NUM_BITS_MASK		GENMASK(22, 20)
+#define     OB_WIN_BUS_NUM_BITS_SHIFT		20
+#define     OB_WIN_MSG_CODE_ENABLE		BIT(22)
+#define     OB_WIN_MSG_CODE_MASK		GENMASK(21, 14)
+#define     OB_WIN_MSG_CODE_SHIFT		14
+#define     OB_WIN_MSG_PAYLOAD_LEN		BIT(12)
+#define     OB_WIN_ATTR_ENABLE			BIT(11)
+#define     OB_WIN_ATTR_TC_MASK			GENMASK(10, 8)
+#define     OB_WIN_ATTR_TC_SHIFT		8
+#define     OB_WIN_ATTR_RELAXED			BIT(7)
+#define     OB_WIN_ATTR_NOSNOOP			BIT(6)
+#define     OB_WIN_ATTR_POISON			BIT(5)
+#define     OB_WIN_ATTR_IDO			BIT(4)
+#define     OB_WIN_TYPE_MASK			GENMASK(3, 0)
+#define     OB_WIN_TYPE_SHIFT			0
+#define     OB_WIN_TYPE_MEM			0x0
+#define     OB_WIN_TYPE_IO			0x4
+#define     OB_WIN_TYPE_CONFIG_TYPE0		0x8
+#define     OB_WIN_TYPE_CONFIG_TYPE1		0x9
+#define     OB_WIN_TYPE_MSG			0xc
+
 /* LMI registers base address and register offsets */
 #define LMI_BASE_ADDR				0x6000
 #define CFG_REG					(LMI_BASE_ADDR + 0x0)

@@ -175,6 +215,13 @@ struct advk_pcie {
 	struct platform_device *pdev;
 	void __iomem *base;
 	struct list_head resources;
+	struct {
+		phys_addr_t match;
+		phys_addr_t remap;
+		phys_addr_t mask;
+		u32 actions;
+	} wins[OB_WIN_COUNT];
+	u8 wins_count;
 	struct irq_domain *irq_domain;
 	struct irq_chip irq_chip;
 	raw_spinlock_t irq_lock;

@@ -350,9 +397,39 @@ static void advk_pcie_train_link(struct advk_pcie *pcie)
 	dev_err(dev, "link never came up\n");
 }
 
+/*
+ * Set PCIe address window register which could be used for memory
+ * mapping.
+ */
+static void advk_pcie_set_ob_win(struct advk_pcie *pcie, u8 win_num,
+				 phys_addr_t match, phys_addr_t remap,
+				 phys_addr_t mask, u32 actions)
+{
+	advk_writel(pcie, OB_WIN_ENABLE |
+			  lower_32_bits(match), OB_WIN_MATCH_LS(win_num));
+	advk_writel(pcie, upper_32_bits(match), OB_WIN_MATCH_MS(win_num));
+	advk_writel(pcie, lower_32_bits(remap), OB_WIN_REMAP_LS(win_num));
+	advk_writel(pcie, upper_32_bits(remap), OB_WIN_REMAP_MS(win_num));
+	advk_writel(pcie, lower_32_bits(mask), OB_WIN_MASK_LS(win_num));
+	advk_writel(pcie, upper_32_bits(mask), OB_WIN_MASK_MS(win_num));
+	advk_writel(pcie, actions, OB_WIN_ACTIONS(win_num));
+}
+
+static void advk_pcie_disable_ob_win(struct advk_pcie *pcie, u8 win_num)
+{
+	advk_writel(pcie, 0, OB_WIN_MATCH_LS(win_num));
+	advk_writel(pcie, 0, OB_WIN_MATCH_MS(win_num));
+	advk_writel(pcie, 0, OB_WIN_REMAP_LS(win_num));
+	advk_writel(pcie, 0, OB_WIN_REMAP_MS(win_num));
+	advk_writel(pcie, 0, OB_WIN_MASK_LS(win_num));
+	advk_writel(pcie, 0, OB_WIN_MASK_MS(win_num));
+	advk_writel(pcie, 0, OB_WIN_ACTIONS(win_num));
+}
+
 static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 {
 	u32 reg;
+	int i;
 
 	/* Set to Direct mode */
 	reg = advk_readl(pcie, CTRL_CONFIG_REG);

@@ -416,15 +493,51 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	reg = PCIE_IRQ_ALL_MASK & (~PCIE_IRQ_ENABLE_INTS_MASK);
 	advk_writel(pcie, reg, HOST_CTRL_INT_MASK_REG);
 
+	/*
+	 * Enable AXI address window location generation:
+	 * When it is enabled, the default outbound window
+	 * configurations (Default User Field: 0xD0074CFC)
+	 * are used to transparent address translation for
+	 * the outbound transactions. Thus, PCIe address
+	 * windows are not required for transparent memory
+	 * access when default outbound window configuration
+	 * is set for memory access.
+	 */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL2_REG);
 	reg |= PCIE_CORE_CTRL2_OB_WIN_ENABLE;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL2_REG);
 
-	/* Bypass the address window mapping for PIO */
+	/*
+	 * Set memory access in Default User Field so it
+	 * is not required to configure PCIe address for
+	 * transparent memory access.
+	 */
+	advk_writel(pcie, OB_WIN_TYPE_MEM, OB_WIN_DEFAULT_ACTIONS);
+
+	/*
+	 * Bypass the address window mapping for PIO:
+	 * Since PIO access already contains all required
+	 * info over AXI interface by PIO registers, the
+	 * address window is not required.
+	 */
 	reg = advk_readl(pcie, PIO_CTRL);
 	reg |= PIO_CTRL_ADDR_WIN_DISABLE;
 	advk_writel(pcie, reg, PIO_CTRL);
 
+	/*
+	 * Configure PCIe address windows for non-memory or
+	 * non-transparent access as by default PCIe uses
+	 * transparent memory access.
+	 */
+	for (i = 0; i < pcie->wins_count; i++)
+		advk_pcie_set_ob_win(pcie, i,
+				     pcie->wins[i].match, pcie->wins[i].remap,
+				     pcie->wins[i].mask, pcie->wins[i].actions);
+
+	/* Disable remaining PCIe outbound windows */
+	for (i = pcie->wins_count; i < OB_WIN_COUNT; i++)
+		advk_pcie_disable_ob_win(pcie, i);
+
 	advk_pcie_train_link(pcie);
 
 	reg = advk_readl(pcie, PCIE_CORE_CMD_STATUS_REG);

@@ -1041,6 +1154,7 @@ static int advk_pcie_probe(struct platform_device *pdev)
 	struct resource *res;
 	struct pci_bus *bus, *child;
 	struct pci_host_bridge *bridge;
+	struct resource_entry *entry;
 	int ret, irq;
 
 	bridge = devm_pci_alloc_host_bridge(dev, sizeof(struct advk_pcie));

@@ -1070,6 +1184,80 @@ static int advk_pcie_probe(struct platform_device *pdev)
 		return ret;
 	}
 
+	resource_list_for_each_entry(entry, &pcie->resources) {
+		resource_size_t start = entry->res->start;
+		resource_size_t size = resource_size(entry->res);
+		unsigned long type = resource_type(entry->res);
+		u64 win_size;
+
+		/*
+		 * Aardvark hardware allows to configure also PCIe window
+		 * for config type 0 and type 1 mapping, but driver uses
+		 * only PIO for issuing configuration transfers which does
+		 * not use PCIe window configuration.
+		 */
+		if (type != IORESOURCE_MEM && type != IORESOURCE_MEM_64 &&
+		    type != IORESOURCE_IO)
+			continue;
+
+		/*
+		 * Skip transparent memory resources. Default outbound access
+		 * configuration is set to transparent memory access so it
+		 * does not need window configuration.
+		 */
+		if ((type == IORESOURCE_MEM || type == IORESOURCE_MEM_64) &&
+		    entry->offset == 0)
+			continue;
+
+		/*
+		 * The n-th PCIe window is configured by tuple (match, remap, mask)
+		 * and an access to address A uses this window if A matches the
+		 * match with given mask.
+		 * So every PCIe window size must be a power of two and every start
+		 * address must be aligned to window size. Minimal size is 64 KiB
+		 * because lower 16 bits of mask must be zero. Remapped address
+		 * may have set only bits from the mask.
+		 */
+		while (pcie->wins_count < OB_WIN_COUNT && size > 0) {
+			/* Calculate the largest aligned window size */
+			win_size = (1ULL << (fls64(size)-1)) |
+				   (start ? (1ULL << __ffs64(start)) : 0);
+			win_size = 1ULL << __ffs64(win_size);
+			if (win_size < 0x10000)
+				break;
+
+			dev_dbg(dev,
+				"Configuring PCIe window %d: [0x%llx-0x%llx] as %lu\n",
+				pcie->wins_count, (unsigned long long)start,
+				(unsigned long long)start + win_size, type);
+
+			if (type == IORESOURCE_IO) {
+				pcie->wins[pcie->wins_count].actions = OB_WIN_TYPE_IO;
+				pcie->wins[pcie->wins_count].match = pci_pio_to_address(start);
+			} else {
+				pcie->wins[pcie->wins_count].actions = OB_WIN_TYPE_MEM;
+				pcie->wins[pcie->wins_count].match = start;
+			}
+			pcie->wins[pcie->wins_count].remap = start - entry->offset;
+			pcie->wins[pcie->wins_count].mask = ~(win_size - 1);
+
+			if (pcie->wins[pcie->wins_count].remap & (win_size - 1))
+				break;
+
+			start += win_size;
+			size -= win_size;
+			pcie->wins_count++;
+		}
+
+		if (size > 0) {
+			dev_err(&pcie->pdev->dev,
+				"Invalid PCIe region [0x%llx-0x%llx]\n",
+				(unsigned long long)entry->res->start,
+				(unsigned long long)entry->res->end + 1);
+			return -EINVAL;
+		}
+	}
+
 	pcie->reset_gpio = devm_fwnode_get_index_gpiod_from_child(dev, "reset",
 								  0,
 								  dev_fwnode(dev),

index fd5155b10c1d..fc198e6e7fa4 100644
@@ -453,8 +453,9 @@ static void advk_pcie_setup_hw(struct advk_pcie *pcie)
 	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_DEVCTL);
 	reg &= ~PCI_EXP_DEVCTL_RELAX_EN;
 	reg &= ~PCI_EXP_DEVCTL_NOSNOOP_EN;
+	reg &= ~PCI_EXP_DEVCTL_PAYLOAD;
 	reg &= ~PCI_EXP_DEVCTL_READRQ;
-	reg |= PCI_EXP_DEVCTL_PAYLOAD; /* Set max payload size */
+	reg |= PCI_EXP_DEVCTL_PAYLOAD_512B;
 	reg |= PCI_EXP_DEVCTL_READRQ_512B;
 	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_DEVCTL);
 

index c843929e0c4f..5db024aa00e9 100644
@@ -654,6 +654,11 @@
 #define  PCI_EXP_LNKCAP2_SLS_8_0GB	0x00000008 /* Supported Speed 8.0GT/s */
 #define  PCI_EXP_LNKCAP2_CROSSLINK	0x00000100 /* Crosslink supported */
 #define PCI_EXP_LNKCTL2		48	/* Link Control 2 */
+#define PCI_EXP_LNKCTL2_TLS		0x000f
+#define PCI_EXP_LNKCTL2_TLS_2_5GT	0x0001 /* Supported Speed 2.5GT/s */
+#define PCI_EXP_LNKCTL2_TLS_5_0GT	0x0002 /* Supported Speed 5GT/s */
+#define PCI_EXP_LNKCTL2_TLS_8_0GT	0x0003 /* Supported Speed 8GT/s */
+#define PCI_EXP_LNKCTL2_TLS_16_0GT	0x0004 /* Supported Speed 16GT/s */
 #define PCI_EXP_LNKSTA2		50	/* Link Status 2 */
 #define PCI_CAP_EXP_ENDPOINT_SIZEOF_V2	52	/* v2 endpoints with link end here */
 #define PCI_EXP_SLTCAP2		52	/* Slot Capabilities 2 */

index fc198e6e7fa4..7c3369726c87 100644
@@ -248,11 +248,6 @@ static inline u32 advk_readl(struct advk_pcie *pcie, u64 reg)
 	return readl(pcie->base + reg);
 }
 
-static inline u16 advk_read16(struct advk_pcie *pcie, u64 reg)
-{
-	return advk_readl(pcie, (reg & ~0x3)) >> ((reg & 0x3) * 8);
-}
-
 static int advk_pcie_link_up(struct advk_pcie *pcie)
 {
 	u32 val, ltssm_state;

@@ -279,23 +274,9 @@ static int advk_pcie_wait_for_link(struct advk_pcie *pcie)
 
 static void advk_pcie_issue_perst(struct advk_pcie *pcie)
 {
-	u32 reg;
-
 	if (!pcie->reset_gpio)
 		return;
 
-	/*
-	 * As required by PCI Express spec (PCI Express Base Specification, REV.
-	 * 4.0 PCI Express, February 19 2014, 6.6.1 Conventional Reset) a delay
-	 * for at least 100ms after de-asserting PERST# signal is needed before
-	 * link training is enabled. So ensure that link training is disabled
-	 * prior de-asserting PERST# signal to fulfill that PCI Express spec
-	 * requirement.
-	 */
-	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
-	reg &= ~LINK_TRAINING_EN;
-	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
-
 	/* 10ms delay is needed for some cards */
 	dev_info(&pcie->pdev->dev, "issuing PERST via reset GPIO for 10ms\n");
 	gpiod_set_value_cansleep(pcie->reset_gpio, 1);

@@ -303,53 +284,46 @@ static void advk_pcie_issue_perst(struct advk_pcie *pcie)
 	gpiod_set_value_cansleep(pcie->reset_gpio, 0);
 }
 
-static int advk_pcie_train_at_gen(struct advk_pcie *pcie, int gen)
+static void advk_pcie_train_link(struct advk_pcie *pcie)
 {
-	int ret, neg_gen;
+	struct device *dev = &pcie->pdev->dev;
 	u32 reg;
+	int ret;
 
-	/* Setup link speed */
+	/*
+	 * Setup PCIe rev / gen compliance based on device tree property
+	 * 'max-link-speed' which also forces maximal link speed.
+	 */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
 	reg &= ~PCIE_GEN_SEL_MSK;
-	if (gen == 3)
+	if (pcie->link_gen == 3)
 		reg |= SPEED_GEN_3;
-	else if (gen == 2)
+	else if (pcie->link_gen == 2)
 		reg |= SPEED_GEN_2;
 	else
 		reg |= SPEED_GEN_1;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
 
 	/*
-	 * Enable link training. This is not needed in every call to this
-	 * function, just once suffices, but it does not break anything either.
+	 * Set maximal link speed value also into PCIe Link Control 2 register.
+	 * Armada 3700 Functional Specification says that default value is based
+	 * on SPEED_GEN but tests showed that default value is always 8.0 GT/s.
 	 */
+	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL2);
+	reg &= ~PCI_EXP_LNKCTL2_TLS;
+	if (pcie->link_gen == 3)
+		reg |= PCI_EXP_LNKCTL2_TLS_8_0GT;
+	else if (pcie->link_gen == 2)
+		reg |= PCI_EXP_LNKCTL2_TLS_5_0GT;
+	else
+		reg |= PCI_EXP_LNKCTL2_TLS_2_5GT;
+	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL2);
+
+	/* Enable link training after selecting PCIe generation */
 	reg = advk_readl(pcie, PCIE_CORE_CTRL0_REG);
 	reg |= LINK_TRAINING_EN;
 	advk_writel(pcie, reg, PCIE_CORE_CTRL0_REG);
 
-	/*
-	 * Start link training immediately after enabling it.
-	 * This solves problems for some buggy cards.
-	 */
-	reg = advk_readl(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL);
-	reg |= PCI_EXP_LNKCTL_RL;
-	advk_writel(pcie, reg, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKCTL);
-
-	ret = advk_pcie_wait_for_link(pcie);
-	if (ret)
-		return ret;
-
-	reg = advk_read16(pcie, PCIE_CORE_PCIEXP_CAP + PCI_EXP_LNKSTA);
-	neg_gen = reg & PCI_EXP_LNKSTA_CLS;
-
-	return neg_gen;
-}
-
-static void advk_pcie_train_link(struct advk_pcie *pcie)
-{
-	struct device *dev = &pcie->pdev->dev;
-	int neg_gen = -1, gen;
-
 	/*
 	 * Reset PCIe card via PERST# signal. Some cards are not detected
 	 * during link training when they are in some non-initial state.

@@ -360,41 +334,18 @@ static void advk_pcie_train_link(struct advk_pcie *pcie)
 	 * PERST# signal could have been asserted by pinctrl subsystem before
 	 * probe() callback has been called or issued explicitly by reset gpio
 	 * function advk_pcie_issue_perst(), making the endpoint going into
-	 * fundamental reset. As required by PCI Express spec a delay for at
-	 * least 100ms after such a reset before link training is needed.
+	 * fundamental reset. As required by PCI Express spec (PCI Express
+	 * Base Specification, REV. 4.0 PCI Express, February 19 2014, 6.6.1
+	 * Conventional Reset) a delay for at least 100ms after such a reset
+	 * before sending a Configuration Request to the device is needed.
+	 * So wait until PCIe link is up. Function advk_pcie_wait_for_link()
+	 * waits for link at least 900ms.
 	 */
-	msleep(PCI_PM_D3COLD_WAIT);
-
-	/*
-	 * Try link training at link gen specified by device tree property
-	 * 'max-link-speed'. If this fails, iteratively train at lower gen.
-	 */
-	for (gen = pcie->link_gen; gen > 0; --gen) {
-		neg_gen = advk_pcie_train_at_gen(pcie, gen);
-		if (neg_gen > 0)
-			break;
-	}
-
-	if (neg_gen < 0)
-		goto err;
-
-	/*
-	 * After successful training if negotiated gen is lower than requested,
-	 * train again on negotiated gen. This solves some stability issues for
-	 * some buggy gen1 cards.
-	 */
-	if (neg_gen < gen) {
-		gen = neg_gen;
-		neg_gen = advk_pcie_train_at_gen(pcie, gen);
-	}
-
-	if (neg_gen == gen) {
-		dev_info(dev, "link up at gen %i\n", gen);
-		return;
-	}
-
-err:
-	dev_err(dev, "link never came up\n");
+	ret = advk_pcie_wait_for_link(pcie);
+	if (ret < 0)
+		dev_err(dev, "link never came up\n");
+	else
+		dev_info(dev, "link up\n");
 }
 
 /*

index 7c3369726c87..9ae544e113dc 100644
@@ -152,9 +152,50 @@
 #define CFG_REG					(LMI_BASE_ADDR + 0x0)
 #define     LTSSM_SHIFT				24
 #define     LTSSM_MASK				0x3f
-#define     LTSSM_L0				0x10
 #define     RC_BAR_CONFIG			0x300
 
+/* LTSSM values in CFG_REG */
+enum {
+	LTSSM_DETECT_QUIET			= 0x0,
+	LTSSM_DETECT_ACTIVE			= 0x1,
+	LTSSM_POLLING_ACTIVE			= 0x2,
+	LTSSM_POLLING_COMPLIANCE		= 0x3,
+	LTSSM_POLLING_CONFIGURATION		= 0x4,
+	LTSSM_CONFIG_LINKWIDTH_START		= 0x5,
+	LTSSM_CONFIG_LINKWIDTH_ACCEPT		= 0x6,
+	LTSSM_CONFIG_LANENUM_ACCEPT		= 0x7,
+	LTSSM_CONFIG_LANENUM_WAIT		= 0x8,
+	LTSSM_CONFIG_COMPLETE			= 0x9,
+	LTSSM_CONFIG_IDLE			= 0xa,
+	LTSSM_RECOVERY_RCVR_LOCK		= 0xb,
+	LTSSM_RECOVERY_SPEED			= 0xc,
+	LTSSM_RECOVERY_RCVR_CFG			= 0xd,
+	LTSSM_RECOVERY_IDLE			= 0xe,
+	LTSSM_L0				= 0x10,
+	LTSSM_RX_L0S_ENTRY			= 0x11,
+	LTSSM_RX_L0S_IDLE			= 0x12,
+	LTSSM_RX_L0S_FTS			= 0x13,
+	LTSSM_TX_L0S_ENTRY			= 0x14,
+	LTSSM_TX_L0S_IDLE			= 0x15,
+	LTSSM_TX_L0S_FTS			= 0x16,
+	LTSSM_L1_ENTRY				= 0x17,
+	LTSSM_L1_IDLE				= 0x18,
+	LTSSM_L2_IDLE				= 0x19,
+	LTSSM_L2_TRANSMIT_WAKE			= 0x1a,
+	LTSSM_DISABLED				= 0x20,
+	LTSSM_LOOPBACK_ENTRY_MASTER		= 0x21,
+	LTSSM_LOOPBACK_ACTIVE_MASTER		= 0x22,
+	LTSSM_LOOPBACK_EXIT_MASTER		= 0x23,
+	LTSSM_LOOPBACK_ENTRY_SLAVE		= 0x24,
+	LTSSM_LOOPBACK_ACTIVE_SLAVE		= 0x25,
+	LTSSM_LOOPBACK_EXIT_SLAVE		= 0x26,
+	LTSSM_HOT_RESET				= 0x27,
+	LTSSM_RECOVERY_EQUALIZATION_PHASE0	= 0x28,
+	LTSSM_RECOVERY_EQUALIZATION_PHASE1	= 0x29,
+	LTSSM_RECOVERY_EQUALIZATION_PHASE2	= 0x2a,
+	LTSSM_RECOVERY_EQUALIZATION_PHASE3	= 0x2b,
+};
+
 /* PCIe core controller registers */
 #define CTRL_CORE_BASE_ADDR			0x18000
 #define CTRL_CONFIG_REG				(CTRL_CORE_BASE_ADDR + 0x0)

@@ -248,13 +289,35 @@ static inline u32 advk_readl(struct advk_pcie *pcie, u64 reg)
 	return readl(pcie->base + reg);
 }
 
-static int advk_pcie_link_up(struct advk_pcie *pcie)
+static u8 advk_pcie_ltssm_state(struct advk_pcie *pcie)
 {
-	u32 val, ltssm_state;
+	u32 val;
+	u8 ltssm_state;
 
 	val = advk_readl(pcie, CFG_REG);
 	ltssm_state = (val >> LTSSM_SHIFT) & LTSSM_MASK;
-	return ltssm_state >= LTSSM_L0;
+	return ltssm_state;
+}
+
+static inline bool advk_pcie_link_up(struct advk_pcie *pcie)
+{
+	/* check if LTSSM is in normal operation - some L* state */
+	u8 ltssm_state = advk_pcie_ltssm_state(pcie);
+	return ltssm_state >= LTSSM_L0 && ltssm_state < LTSSM_DISABLED;
+}
+
+static inline bool advk_pcie_link_training(struct advk_pcie *pcie)
+{
+	/*
+	  * According to PCIe Base specification 3.0, Table 4-14: Link
+	  * Status Mapped to the LTSSM is Link Training mapped to LTSSM
+	  * Configuration and Recovery states.
+	  */
+	u8 ltssm_state = advk_pcie_ltssm_state(pcie);
+	return ((ltssm_state >= LTSSM_CONFIG_LINKWIDTH_START &&
+		  ltssm_state < LTSSM_L0) ||
+		(ltssm_state >= LTSSM_RECOVERY_EQUALIZATION_PHASE0 &&
+		  ltssm_state <= LTSSM_RECOVERY_EQUALIZATION_PHASE3));
 }
 
 static int advk_pcie_wait_for_link(struct advk_pcie *pcie)

index 262f591ad8a6..4b4fd4661268 100644
@@ -157,8 +157,8 @@ static struct armada_37xx_pin_group armada_37xx_nb_groups[] = {
 	PIN_GRP_GPIO("pwm1", 12, 1, BIT(4), "pwm"),
 	PIN_GRP_GPIO("pwm2", 13, 1, BIT(5), "pwm"),
 	PIN_GRP_GPIO("pwm3", 14, 1, BIT(6), "pwm"),
-	PIN_GRP_GPIO("pmic1", 17, 1, BIT(7), "pmic"),
-	PIN_GRP_GPIO("pmic0", 16, 1, BIT(8), "pmic"),
+	PIN_GRP_GPIO("pmic1", 7, 1, BIT(7), "pmic"),
+	PIN_GRP_GPIO("pmic0", 6, 1, BIT(8), "pmic"),
 	PIN_GRP_GPIO("i2c2", 2, 2, BIT(9), "i2c"),
 	PIN_GRP_GPIO("i2c1", 0, 2, BIT(10), "i2c"),
 	PIN_GRP_GPIO("spi_cs1", 17, 1, BIT(12), "spi"),

@@ -182,8 +182,10 @@ static struct armada_37xx_pin_group armada_37xx_sb_groups[] = {
 	PIN_GRP_GPIO("usb2_drvvbus1", 1, 1, BIT(1), "drvbus"),
 	PIN_GRP_GPIO("sdio_sb", 24, 6, BIT(2), "sdio"),
 	PIN_GRP_GPIO("rgmii", 6, 12, BIT(3), "mii"),
-	PIN_GRP_GPIO("pcie1", 3, 2, BIT(4), "pcie"),
-	PIN_GRP_GPIO("ptp", 20, 3, BIT(5), "ptp"),
+	PIN_GRP_GPIO("smi", 18, 2, BIT(4), "smi"),
+	PIN_GRP_GPIO("pcie1", 3, 1, BIT(5), "pcie"),
+	PIN_GRP_GPIO("pcie1_clkreq", 4, 1, BIT(9), "pcie"),
+	PIN_GRP_GPIO("ptp", 20, 3, BIT(11) | BIT(12) | BIT(13), "ptp"),
 	PIN_GRP("ptp_clk", 21, 1, BIT(6), "ptp", "mii"),
 	PIN_GRP("ptp_trig", 22, 1, BIT(7), "ptp", "mii"),
 	PIN_GRP_GPIO_3("mii_col", 23, 1, BIT(8) | BIT(14), 0, BIT(8), BIT(14),

index 4b4fd4661268..a58367909455 100644
@@ -185,6 +185,7 @@ static struct armada_37xx_pin_group armada_37xx_sb_groups[] = {
 	PIN_GRP_GPIO("smi", 18, 2, BIT(4), "smi"),
 	PIN_GRP_GPIO("pcie1", 3, 1, BIT(5), "pcie"),
 	PIN_GRP_GPIO("pcie1_clkreq", 4, 1, BIT(9), "pcie"),
+	PIN_GRP_GPIO("pcie1_wakeup", 5, 1, BIT(10), "pcie"),
 	PIN_GRP_GPIO("ptp", 20, 3, BIT(11) | BIT(12) | BIT(13), "ptp"),
 	PIN_GRP("ptp_clk", 21, 1, BIT(6), "ptp", "mii"),
 	PIN_GRP("ptp_trig", 22, 1, BIT(7), "ptp", "mii"),

index a58367909455..9f7d917458c7 100644
@@ -153,10 +153,14 @@ static struct armada_37xx_pin_group armada_37xx_nb_groups[] = {
 	PIN_GRP_GPIO("jtag", 20, 5, BIT(0), "jtag"),
 	PIN_GRP_GPIO("sdio0", 8, 3, BIT(1), "sdio"),
 	PIN_GRP_GPIO("emmc_nb", 27, 9, BIT(2), "emmc"),
-	PIN_GRP_GPIO("pwm0", 11, 1, BIT(3), "pwm"),
-	PIN_GRP_GPIO("pwm1", 12, 1, BIT(4), "pwm"),
-	PIN_GRP_GPIO("pwm2", 13, 1, BIT(5), "pwm"),
-	PIN_GRP_GPIO("pwm3", 14, 1, BIT(6), "pwm"),
+	PIN_GRP_GPIO_3("pwm0", 11, 1, BIT(3) | BIT(20), 0, BIT(20), BIT(3),
+		       "pwm", "led"),
+	PIN_GRP_GPIO_3("pwm1", 12, 1, BIT(4) | BIT(21), 0, BIT(21), BIT(4),
+		       "pwm", "led"),
+	PIN_GRP_GPIO_3("pwm2", 13, 1, BIT(5) | BIT(22), 0, BIT(22), BIT(5),
+		       "pwm", "led"),
+	PIN_GRP_GPIO_3("pwm3", 14, 1, BIT(6) | BIT(23), 0, BIT(23), BIT(6),
+		       "pwm", "led"),
 	PIN_GRP_GPIO("pmic1", 7, 1, BIT(7), "pmic"),
 	PIN_GRP_GPIO("pmic0", 6, 1, BIT(8), "pmic"),
 	PIN_GRP_GPIO("i2c2", 2, 2, BIT(9), "i2c"),

@@ -170,11 +174,6 @@ static struct armada_37xx_pin_group armada_37xx_nb_groups[] = {
 	PIN_GRP_EXTRA("uart2", 9, 2, BIT(1) | BIT(13) | BIT(14) | BIT(19),
 		      BIT(1) | BIT(13) | BIT(14), BIT(1) | BIT(19),
 		      18, 2, "gpio", "uart"),
-	PIN_GRP_GPIO_2("led0_od", 11, 1, BIT(20), BIT(20), 0, "led"),
-	PIN_GRP_GPIO_2("led1_od", 12, 1, BIT(21), BIT(21), 0, "led"),
-	PIN_GRP_GPIO_2("led2_od", 13, 1, BIT(22), BIT(22), 0, "led"),
-	PIN_GRP_GPIO_2("led3_od", 14, 1, BIT(23), BIT(23), 0, "led"),
-
 };
 
 static struct armada_37xx_pin_group armada_37xx_sb_groups[] = {