[POWERPC] 4xx: PLB to PCI Express support

This adds to the previous 2 patches the support for the 4xx PCI Express
cells as found in the 440SPe revA, revB and 405EX.

Unfortunately, due to significant differences between these, and other
interesting "features" of those pieces of HW, the code isn't as simple
as it is for PCI and PCI-X and some of the functions differ significantly
between the 3 implementations. Thus, not only this code can only support
those 3 implementations for now and will refuse to operate on any other,
but there are added ifdef's to avoid the bloat of building a fairly large
amount of code on platforms that don't need it.

Also, this code currently only supports fully initializing root complex
nodes, not endpoint. Some more code will have to be lifted from the
arch/ppc implementation to add the endpoint support, though it's mostly
differences in memory mapping, and the question on how to represent
endpoint mode PCI in the device-tree is thus open.

Many thanks to Stefan Roese for testing & fixing up the 405EX bits !

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

1 files changed, 976 insertions, 18 deletions

diff --git a/arch/powerpc/sysdev/ppc4xx_pci.c b/arch/powerpc/sysdev/ppc4xx_pci.c
index b7d79880c9b9..b986eff09f6b 100644
--- a/arch/powerpc/sysdev/ppc4xx_pci.c
+++ b/arch/powerpc/sysdev/ppc4xx_pci.c
@@ -3,16 +3,31 @@
  *
  * Copyright 2007 Ben. Herrenschmidt <benh@kernel.crashing.org>, IBM Corp.
  *
+ * Most PCI Express code is coming from Stefan Roese implementation for
+ * arch/ppc in the Denx tree, slightly reworked by me.
+ *
+ * Copyright 2007 DENX Software Engineering, Stefan Roese <sr@denx.de>
+ *
+ * Some of that comes itself from a previous implementation for 440SPE only
+ * by Roland Dreier:
+ *
+ * Copyright (c) 2005 Cisco Systems.  All rights reserved.
+ * Roland Dreier <rolandd@cisco.com>
+ *
  */
 
 #include <linux/kernel.h>
 #include <linux/pci.h>
 #include <linux/init.h>
 #include <linux/of.h>
+#include <linux/bootmem.h>
+#include <linux/delay.h>
 
 #include <asm/io.h>
 #include <asm/pci-bridge.h>
 #include <asm/machdep.h>
+#include <asm/dcr.h>
+#include <asm/dcr-regs.h>
 
 #include "ppc4xx_pci.h"
 
@@ -21,6 +36,17 @@ static int dma_offset_set;
 /* Move that to a useable header */
 extern unsigned long total_memory;
 
+#define U64_TO_U32_LOW(val)	((u32)((val) & 0x00000000ffffffffULL))
+#define U64_TO_U32_HIGH(val)	((u32)((val) >> 32))
+
+#ifdef CONFIG_RESOURCES_64BIT
+#define RES_TO_U32_LOW(val)	U64_TO_U32_LOW(val)
+#define RES_TO_U32_HIGH(val)	U64_TO_U32_HIGH(val)
+#else
+#define RES_TO_U32_LOW(val)	(val)
+#define RES_TO_U32_HIGH(val)	(0)
+#endif
+
 static void fixup_ppc4xx_pci_bridge(struct pci_dev *dev)
 {
 	struct pci_controller *hose;
@@ -178,13 +204,8 @@ static void __init ppc4xx_configure_pci_PMMs(struct pci_controller *hose,
 
 		/* Calculate register values */
 		la = res->start;
-#ifdef CONFIG_RESOURCES_64BIT
-		pciha = (res->start - hose->pci_mem_offset) >> 32;
-		pcila = (res->start - hose->pci_mem_offset) & 0xffffffffu;
-#else
-		pciha = 0;
-		pcila = res->start - hose->pci_mem_offset;
-#endif
+		pciha = RES_TO_U32_HIGH(res->start - hose->pci_mem_offset);
+		pcila = RES_TO_U32_LOW(res->start - hose->pci_mem_offset);
 
 		ma = res->end + 1 - res->start;
 		if (!is_power_of_2(ma) || ma < 0x1000 || ma > 0xffffffffu) {
@@ -333,16 +354,10 @@ static void __init ppc4xx_configure_pcix_POMs(struct pci_controller *hose,
 		}
 
 		/* Calculate register values */
-#ifdef CONFIG_RESOURCES_64BIT
-		lah = res->start >> 32;
-		lal = res->start & 0xffffffffu;
-		pciah = (res->start - hose->pci_mem_offset) >> 32;
-		pcial = (res->start - hose->pci_mem_offset) & 0xffffffffu;
-#else
-		lah = pciah = 0;
-		lal = res->start;
-		pcial = res->start - hose->pci_mem_offset;
-#endif
+		lah = RES_TO_U32_HIGH(res->start);
+		lal = RES_TO_U32_LOW(res->start);
+		pciah = RES_TO_U32_HIGH(res->start - hose->pci_mem_offset);
+		pcial = RES_TO_U32_LOW(res->start - hose->pci_mem_offset);
 		sa = res->end + 1 - res->start;
 		if (!is_power_of_2(sa) || sa < 0x100000 ||
 		    sa > 0xffffffffu) {
@@ -492,20 +507,963 @@ static void __init ppc4xx_probe_pcix_bridge(struct device_node *np)
 		iounmap(reg);
 }
 
+#ifdef CONFIG_PPC4xx_PCI_EXPRESS
+
 /*
  * 4xx PCI-Express part
+ *
+ * We support 3 parts currently based on the compatible property:
+ *
+ * ibm,plb-pciex-440speA
+ * ibm,plb-pciex-440speB
+ * ibm,plb-pciex-405ex
+ *
+ * Anything else will be rejected for now as they are all subtly
+ * different unfortunately.
+ *
  */
+
+#define MAX_PCIE_BUS_MAPPED	0x10
+
+struct ppc4xx_pciex_port
+{
+	struct pci_controller	*hose;
+	struct device_node	*node;
+	unsigned int		index;
+	int			endpoint;
+	unsigned int		sdr_base;
+	dcr_host_t		dcrs;
+	struct resource		cfg_space;
+	struct resource		utl_regs;
+};
+
+static struct ppc4xx_pciex_port *ppc4xx_pciex_ports;
+static unsigned int ppc4xx_pciex_port_count;
+
+struct ppc4xx_pciex_hwops
+{
+	int (*core_init)(struct device_node *np);
+	int (*port_init_hw)(struct ppc4xx_pciex_port *port);
+	int (*setup_utl)(struct ppc4xx_pciex_port *port);
+};
+
+static struct ppc4xx_pciex_hwops *ppc4xx_pciex_hwops;
+
+#ifdef CONFIG_44x
+
+/* Check various reset bits of the 440SPe PCIe core */
+static int __init ppc440spe_pciex_check_reset(struct device_node *np)
+{
+	u32 valPE0, valPE1, valPE2;
+	int err = 0;
+
+	/* SDR0_PEGPLLLCT1 reset */
+	if (!(mfdcri(SDR0, PESDR0_PLLLCT1) & 0x01000000)) {
+		/*
+		 * the PCIe core was probably already initialised
+		 * by firmware - let's re-reset RCSSET regs
+		 *
+		 * -- Shouldn't we also re-reset the whole thing ? -- BenH
+		 */
+		pr_debug("PCIE: SDR0_PLLLCT1 already reset.\n");
+		mtdcri(SDR0, PESDR0_440SPE_RCSSET, 0x01010000);
+		mtdcri(SDR0, PESDR1_440SPE_RCSSET, 0x01010000);
+		mtdcri(SDR0, PESDR2_440SPE_RCSSET, 0x01010000);
+	}
+
+	valPE0 = mfdcri(SDR0, PESDR0_440SPE_RCSSET);
+	valPE1 = mfdcri(SDR0, PESDR1_440SPE_RCSSET);
+	valPE2 = mfdcri(SDR0, PESDR2_440SPE_RCSSET);
+
+	/* SDR0_PExRCSSET rstgu */
+	if (!(valPE0 & 0x01000000) ||
+	    !(valPE1 & 0x01000000) ||
+	    !(valPE2 & 0x01000000)) {
+		printk(KERN_INFO "PCIE: SDR0_PExRCSSET rstgu error\n");
+		err = -1;
+	}
+
+	/* SDR0_PExRCSSET rstdl */
+	if (!(valPE0 & 0x00010000) ||
+	    !(valPE1 & 0x00010000) ||
+	    !(valPE2 & 0x00010000)) {
+		printk(KERN_INFO "PCIE: SDR0_PExRCSSET rstdl error\n");
+		err = -1;
+	}
+
+	/* SDR0_PExRCSSET rstpyn */
+	if ((valPE0 & 0x00001000) ||
+	    (valPE1 & 0x00001000) ||
+	    (valPE2 & 0x00001000)) {
+		printk(KERN_INFO "PCIE: SDR0_PExRCSSET rstpyn error\n");
+		err = -1;
+	}
+
+	/* SDR0_PExRCSSET hldplb */
+	if ((valPE0 & 0x10000000) ||
+	    (valPE1 & 0x10000000) ||
+	    (valPE2 & 0x10000000)) {
+		printk(KERN_INFO "PCIE: SDR0_PExRCSSET hldplb error\n");
+		err = -1;
+	}
+
+	/* SDR0_PExRCSSET rdy */
+	if ((valPE0 & 0x00100000) ||
+	    (valPE1 & 0x00100000) ||
+	    (valPE2 & 0x00100000)) {
+		printk(KERN_INFO "PCIE: SDR0_PExRCSSET rdy error\n");
+		err = -1;
+	}
+
+	/* SDR0_PExRCSSET shutdown */
+	if ((valPE0 & 0x00000100) ||
+	    (valPE1 & 0x00000100) ||
+	    (valPE2 & 0x00000100)) {
+		printk(KERN_INFO "PCIE: SDR0_PExRCSSET shutdown error\n");
+		err = -1;
+	}
+
+	return err;
+}
+
+/* Global PCIe core initializations for 440SPe core */
+static int __init ppc440spe_pciex_core_init(struct device_node *np)
+{
+	int time_out = 20;
+
+	/* Set PLL clock receiver to LVPECL */
+	mtdcri(SDR0, PESDR0_PLLLCT1, mfdcri(SDR0, PESDR0_PLLLCT1) | 1 << 28);
+
+	/* Shouldn't we do all the calibration stuff etc... here ? */
+	if (ppc440spe_pciex_check_reset(np))
+		return -ENXIO;
+
+	if (!(mfdcri(SDR0, PESDR0_PLLLCT2) & 0x10000)) {
+		printk(KERN_INFO "PCIE: PESDR_PLLCT2 resistance calibration "
+		       "failed (0x%08x)\n",
+		       mfdcri(SDR0, PESDR0_PLLLCT2));
+		return -1;
+	}
+
+	/* De-assert reset of PCIe PLL, wait for lock */
+	mtdcri(SDR0, PESDR0_PLLLCT1,
+	       mfdcri(SDR0, PESDR0_PLLLCT1) & ~(1 << 24));
+	udelay(3);
+
+	while (time_out) {
+		if (!(mfdcri(SDR0, PESDR0_PLLLCT3) & 0x10000000)) {
+			time_out--;
+			udelay(1);
+		} else
+			break;
+	}
+	if (!time_out) {
+		printk(KERN_INFO "PCIE: VCO output not locked\n");
+		return -1;
+	}
+
+	pr_debug("PCIE initialization OK\n");
+
+	return 3;
+}
+
+static int ppc440spe_pciex_init_port_hw(struct ppc4xx_pciex_port *port)
+{
+	u32 val = 1 << 24;
+
+	if (port->endpoint)
+		val = PTYPE_LEGACY_ENDPOINT << 20;
+	else
+		val = PTYPE_ROOT_PORT << 20;
+
+	if (port->index == 0)
+		val |= LNKW_X8 << 12;
+	else
+		val |= LNKW_X4 << 12;
+
+	mtdcri(SDR0, port->sdr_base + PESDRn_DLPSET, val);
+	mtdcri(SDR0, port->sdr_base + PESDRn_UTLSET1, 0x20222222);
+	if (of_device_is_compatible(port->node, "ibm,plb-pciex-440speA"))
+		mtdcri(SDR0, port->sdr_base + PESDRn_UTLSET2, 0x11000000);
+	mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL0SET1, 0x35000000);
+	mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL1SET1, 0x35000000);
+	mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL2SET1, 0x35000000);
+	mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL3SET1, 0x35000000);
+	if (port->index == 0) {
+		mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL4SET1,
+		       0x35000000);
+		mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL5SET1,
+		       0x35000000);
+		mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL6SET1,
+		       0x35000000);
+		mtdcri(SDR0, port->sdr_base + PESDRn_440SPE_HSSL7SET1,
+		       0x35000000);
+	}
+	val = mfdcri(SDR0, port->sdr_base + PESDRn_RCSSET);
+	mtdcri(SDR0, port->sdr_base + PESDRn_RCSSET,
+	       (val & ~(1 << 24 | 1 << 16)) | 1 << 12);
+
+	return 0;
+}
+
+static int ppc440speA_pciex_init_utl(struct ppc4xx_pciex_port *port)
+{
+	void __iomem *utl_base;
+
+	/* XXX Check what that value means... I hate magic */
+	dcr_write(port->dcrs, DCRO_PEGPL_SPECIAL, 0x68782800);
+
+	utl_base = ioremap(port->utl_regs.start, 0x100);
+	BUG_ON(utl_base == NULL);
+
+	/*
+	 * Set buffer allocations and then assert VRB and TXE.
+	 */
+	out_be32(utl_base + PEUTL_OUTTR,   0x08000000);
+	out_be32(utl_base + PEUTL_INTR,    0x02000000);
+	out_be32(utl_base + PEUTL_OPDBSZ,  0x10000000);
+	out_be32(utl_base + PEUTL_PBBSZ,   0x53000000);
+	out_be32(utl_base + PEUTL_IPHBSZ,  0x08000000);
+	out_be32(utl_base + PEUTL_IPDBSZ,  0x10000000);
+	out_be32(utl_base + PEUTL_RCIRQEN, 0x00f00000);
+	out_be32(utl_base + PEUTL_PCTL,    0x80800066);
+
+	iounmap(utl_base);
+
+	return 0;
+}
+
+static struct ppc4xx_pciex_hwops ppc440speA_pcie_hwops __initdata =
+{
+	.core_init	= ppc440spe_pciex_core_init,
+	.port_init_hw	= ppc440spe_pciex_init_port_hw,
+	.setup_utl	= ppc440speA_pciex_init_utl,
+};
+
+static struct ppc4xx_pciex_hwops ppc440speB_pcie_hwops __initdata =
+{
+	.core_init	= ppc440spe_pciex_core_init,
+	.port_init_hw	= ppc440spe_pciex_init_port_hw,
+};
+
+
+#endif /* CONFIG_44x */
+
+#ifdef CONFIG_40x
+
+static int __init ppc405ex_pciex_core_init(struct device_node *np)
+{
+	/* Nothing to do, return 2 ports */
+	return 2;
+}
+
+static void ppc405ex_pcie_phy_reset(struct ppc4xx_pciex_port *port)
+{
+	/* Assert the PE0_PHY reset */
+	mtdcri(SDR0, port->sdr_base + PESDRn_RCSSET, 0x01010000);
+	msleep(1);
+
+	/* deassert the PE0_hotreset */
+	if (port->endpoint)
+		mtdcri(SDR0, port->sdr_base + PESDRn_RCSSET, 0x01111000);
+	else
+		mtdcri(SDR0, port->sdr_base + PESDRn_RCSSET, 0x01101000);
+
+	/* poll for phy !reset */
+	/* XXX FIXME add timeout */
+	while (!(mfdcri(SDR0, port->sdr_base + PESDRn_405EX_PHYSTA) & 0x00001000))
+		;
+
+	/* deassert the PE0_gpl_utl_reset */
+	mtdcri(SDR0, port->sdr_base + PESDRn_RCSSET, 0x00101000);
+}
+
+static int ppc405ex_pciex_init_port_hw(struct ppc4xx_pciex_port *port)
+{
+	u32 val;
+
+	if (port->endpoint)
+		val = PTYPE_LEGACY_ENDPOINT;
+	else
+		val = PTYPE_ROOT_PORT;
+
+	mtdcri(SDR0, port->sdr_base + PESDRn_DLPSET,
+	       1 << 24 | val << 20 | LNKW_X1 << 12);
+
+	mtdcri(SDR0, port->sdr_base + PESDRn_UTLSET1, 0x00000000);
+	mtdcri(SDR0, port->sdr_base + PESDRn_UTLSET2, 0x01010000);
+	mtdcri(SDR0, port->sdr_base + PESDRn_405EX_PHYSET1, 0x720F0000);
+	mtdcri(SDR0, port->sdr_base + PESDRn_405EX_PHYSET2, 0x70600003);
+
+	/*
+	 * Only reset the PHY when no link is currently established.
+	 * This is for the Atheros PCIe board which has problems to establish
+	 * the link (again) after this PHY reset. All other currently tested
+	 * PCIe boards don't show this problem.
+	 * This has to be re-tested and fixed in a later release!
+	 */
+#if 0 /* XXX FIXME: Not resetting the PHY will leave all resources
+       * configured as done previously by U-Boot. Then Linux will currently
+       * not reassign them. So the PHY reset is now done always. This will
+       * lead to problems with the Atheros PCIe board again.
+       */
+	val = mfdcri(SDR0, port->sdr_base + PESDRn_LOOP);
+	if (!(val & 0x00001000))
+		ppc405ex_pcie_phy_reset(port);
+#else
+	ppc405ex_pcie_phy_reset(port);
+#endif
+
+	dcr_write(port->dcrs, DCRO_PEGPL_CFG, 0x10000000);  /* guarded on */
+
+	return 0;
+}
+
+static int ppc405ex_pciex_init_utl(struct ppc4xx_pciex_port *port)
+{
+	void __iomem *utl_base;
+
+	dcr_write(port->dcrs, DCRO_PEGPL_SPECIAL, 0x0);
+
+	utl_base = ioremap(port->utl_regs.start, 0x100);
+	BUG_ON(utl_base == NULL);
+
+	/*
+	 * Set buffer allocations and then assert VRB and TXE.
+	 */
+	out_be32(utl_base + PEUTL_OUTTR,   0x02000000);
+	out_be32(utl_base + PEUTL_INTR,    0x02000000);
+	out_be32(utl_base + PEUTL_OPDBSZ,  0x04000000);
+	out_be32(utl_base + PEUTL_PBBSZ,   0x21000000);
+	out_be32(utl_base + PEUTL_IPHBSZ,  0x02000000);
+	out_be32(utl_base + PEUTL_IPDBSZ,  0x04000000);
+	out_be32(utl_base + PEUTL_RCIRQEN, 0x00f00000);
+	out_be32(utl_base + PEUTL_PCTL,    0x80800066);
+
+	out_be32(utl_base + PEUTL_PBCTL,   0x0800000c);
+	out_be32(utl_base + PEUTL_RCSTA,
+		 in_be32(utl_base + PEUTL_RCSTA) | 0x000040000);
+
+	iounmap(utl_base);
+
+	return 0;
+}
+
+static struct ppc4xx_pciex_hwops ppc405ex_pcie_hwops __initdata =
+{
+	.core_init	= ppc405ex_pciex_core_init,
+	.port_init_hw	= ppc405ex_pciex_init_port_hw,
+	.setup_utl	= ppc405ex_pciex_init_utl,
+};
+
+#endif /* CONFIG_40x */
+
+
+/* Check that the core has been initied and if not, do it */
+static int __init ppc4xx_pciex_check_core_init(struct device_node *np)
+{
+	static int core_init;
+	int count = -ENODEV;
+
+	if (core_init++)
+		return 0;
+
+#ifdef CONFIG_44x
+	if (of_device_is_compatible(np, "ibm,plb-pciex-440speA"))
+		ppc4xx_pciex_hwops = &ppc440speA_pcie_hwops;
+	else if (of_device_is_compatible(np, "ibm,plb-pciex-440speB"))
+		ppc4xx_pciex_hwops = &ppc440speB_pcie_hwops;
+#endif /* CONFIG_44x    */
+#ifdef CONFIG_40x
+	if (of_device_is_compatible(np, "ibm,plb-pciex-405ex"))
+		ppc4xx_pciex_hwops = &ppc405ex_pcie_hwops;
+#endif
+	if (ppc4xx_pciex_hwops == NULL) {
+		printk(KERN_WARNING "PCIE: unknown host type %s\n",
+		       np->full_name);
+		return -ENODEV;
+	}
+
+	count = ppc4xx_pciex_hwops->core_init(np);
+	if (count > 0) {
+		ppc4xx_pciex_ports =
+		       kzalloc(count * sizeof(struct ppc4xx_pciex_port),
+			       GFP_KERNEL);
+		if (ppc4xx_pciex_ports) {
+			ppc4xx_pciex_port_count = count;
+			return 0;
+		}
+		printk(KERN_WARNING "PCIE: failed to allocate ports array\n");
+		return -ENOMEM;
+	}
+	return -ENODEV;
+}
+
+static void __init ppc4xx_pciex_port_init_mapping(struct ppc4xx_pciex_port *port)
+{
+	/* We map PCI Express configuration based on the reg property */
+	dcr_write(port->dcrs, DCRO_PEGPL_CFGBAH,
+		  RES_TO_U32_HIGH(port->cfg_space.start));
+	dcr_write(port->dcrs, DCRO_PEGPL_CFGBAL,
+		  RES_TO_U32_LOW(port->cfg_space.start));
+
+	/* XXX FIXME: Use size from reg property. For now, map 512M */
+	dcr_write(port->dcrs, DCRO_PEGPL_CFGMSK, 0xe0000001);
+
+	/* We map UTL registers based on the reg property */
+	dcr_write(port->dcrs, DCRO_PEGPL_REGBAH,
+		  RES_TO_U32_HIGH(port->utl_regs.start));
+	dcr_write(port->dcrs, DCRO_PEGPL_REGBAL,
+		  RES_TO_U32_LOW(port->utl_regs.start));
+
+	/* XXX FIXME: Use size from reg property */
+	dcr_write(port->dcrs, DCRO_PEGPL_REGMSK, 0x00007001);
+
+	/* Disable all other outbound windows */
+	dcr_write(port->dcrs, DCRO_PEGPL_OMR1MSKL, 0);
+	dcr_write(port->dcrs, DCRO_PEGPL_OMR2MSKL, 0);
+	dcr_write(port->dcrs, DCRO_PEGPL_OMR3MSKL, 0);
+	dcr_write(port->dcrs, DCRO_PEGPL_MSGMSK, 0);
+}
+
+static int __init ppc4xx_pciex_port_init(struct ppc4xx_pciex_port *port)
+{
+	int attempts, rc = 0;
+	u32 val;
+
+	/* Check if it's endpoint or root complex
+	 *
+	 * XXX Do we want to use the device-tree instead ? --BenH.
+	 */
+	val = mfdcri(SDR0, port->sdr_base + PESDRn_DLPSET);
+	port->endpoint = (((val >> 20) & 0xf) != PTYPE_ROOT_PORT);
+
+	/* Init HW */
+	if (ppc4xx_pciex_hwops->port_init_hw)
+		rc = ppc4xx_pciex_hwops->port_init_hw(port);
+	if (rc != 0)
+		return rc;
+
+	/*
+	 * Notice: the following delay has critical impact on device
+	 * initialization - if too short (<50ms) the link doesn't get up.
+	 *
+	 * XXX FIXME: There are various issues with that link up thingy,
+	 * we could just wait for the link with a timeout but Stefan says
+	 * some cards need more time even after the link is up. I'll
+	 * investigate. For now, we keep a fixed 1s delay.
+	 *
+	 * Ultimately, it should be made asynchronous so all ports are
+	 * brought up simultaneously though.
+	 */
+	printk(KERN_INFO "PCIE%d: Waiting for link to go up...\n",
+	       port->index);
+	msleep(1000);
+
+	/*
+	 * Check that we exited the reset state properly
+	 */
+	val = mfdcri(SDR0, port->sdr_base + PESDRn_RCSSTS);
+	if (val & (1 << 20)) {
+		printk(KERN_WARNING "PCIE%d: PGRST failed %08x\n",
+		       port->index, val);
+		return -1;
+	}
+
+	/*
+	 * Verify link is up
+	 */
+	val = mfdcri(SDR0, port->sdr_base + PESDRn_LOOP);
+	if (!(val & 0x00001000)) {
+		printk(KERN_INFO "PCIE%d: link is not up !\n",
+		       port->index);
+		return -1;
+	}
+
+	printk(KERN_INFO "PCIE%d: link is up !\n",
+	       port->index);
+
+	/*
+	 * Initialize mapping: disable all regions and configure
+	 * CFG and REG regions based on resources in the device tree
+	 */
+	ppc4xx_pciex_port_init_mapping(port);
+
+	/*
+	 * Setup UTL registers - but only on revA!
+	 * We use default settings for revB chip.
+	 *
+	 * To be reworked. We may also be able to move that to
+	 * before the link wait
+	 * --BenH.
+	 */
+	if (ppc4xx_pciex_hwops->setup_utl)
+		ppc4xx_pciex_hwops->setup_utl(port);
+
+	/*
+	 * Check for VC0 active and assert RDY.
+	 */
+	attempts = 10;
+	while (!(mfdcri(SDR0, port->sdr_base + PESDRn_RCSSTS) & (1 << 16))) {
+		if (!(attempts--)) {
+			printk(KERN_INFO "PCIE%d: VC0 not active\n",
+			       port->index);
+			return -1;
+		}
+		msleep(1000);
+	}
+	mtdcri(SDR0, port->sdr_base + PESDRn_RCSSET,
+	       mfdcri(SDR0, port->sdr_base + PESDRn_RCSSET) | 1 << 20);
+	msleep(100);
+
+	return 0;
+}
+
+static int ppc4xx_pciex_validate_bdf(struct ppc4xx_pciex_port *port,
+				     struct pci_bus *bus,
+				     unsigned int devfn)
+{
+	static int message;
+
+	/* Endpoint can not generate upstream(remote) config cycles */
+	if (port->endpoint && bus->number != port->hose->first_busno)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	/* Check we are within the mapped range */
+	if (bus->number > port->hose->last_busno) {
+		if (!message) {
+			printk(KERN_WARNING "Warning! Probing bus %u"
+			       " out of range !\n", bus->number);
+			message++;
+		}
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	}
+
+	/* The root complex has only one device / function */
+	if (bus->number == port->hose->first_busno && devfn != 0)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	/* The other side of the RC has only one device as well */
+	if (bus->number == (port->hose->first_busno + 1) &&
+	    PCI_SLOT(devfn) != 0)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	return 0;
+}
+
+static void __iomem *ppc4xx_pciex_get_config_base(struct ppc4xx_pciex_port *port,
+						  struct pci_bus *bus,
+						  unsigned int devfn)
+{
+	int relbus;
+
+	/* Remove the casts when we finally remove the stupid volatile
+	 * in struct pci_controller
+	 */
+	if (bus->number == port->hose->first_busno)
+		return (void __iomem *)port->hose->cfg_addr;
+
+	relbus = bus->number - (port->hose->first_busno + 1);
+	return (void __iomem *)port->hose->cfg_data +
+		((relbus  << 20) | (devfn << 12));
+}
+
+static int ppc4xx_pciex_read_config(struct pci_bus *bus, unsigned int devfn,
+				    int offset, int len, u32 *val)
+{
+	struct pci_controller *hose = (struct pci_controller *) bus->sysdata;
+	struct ppc4xx_pciex_port *port =
+		&ppc4xx_pciex_ports[hose->indirect_type];
+	void __iomem *addr;
+	u32 gpl_cfg;
+
+	BUG_ON(hose != port->hose);
+
+	if (ppc4xx_pciex_validate_bdf(port, bus, devfn) != 0)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = ppc4xx_pciex_get_config_base(port, bus, devfn);
+
+	/*
+	 * Reading from configuration space of non-existing device can
+	 * generate transaction errors. For the read duration we suppress
+	 * assertion of machine check exceptions to avoid those.
+	 */
+	gpl_cfg = dcr_read(port->dcrs, DCRO_PEGPL_CFG);
+	dcr_write(port->dcrs, DCRO_PEGPL_CFG, gpl_cfg | GPL_DMER_MASK_DISA);
+
+	switch (len) {
+	case 1:
+		*val = in_8((u8 *)(addr + offset));
+		break;
+	case 2:
+		*val = in_le16((u16 *)(addr + offset));
+		break;
+	default:
+		*val = in_le32((u32 *)(addr + offset));
+		break;
+	}
+
+	pr_debug("pcie-config-read: bus=%3d [%3d..%3d] devfn=0x%04x"
+		 " offset=0x%04x len=%d, addr=0x%p val=0x%08x\n",
+		 bus->number, hose->first_busno, hose->last_busno,
+		 devfn, offset, len, addr + offset, *val);
+
+	dcr_write(port->dcrs, DCRO_PEGPL_CFG, gpl_cfg);
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static int ppc4xx_pciex_write_config(struct pci_bus *bus, unsigned int devfn,
+				     int offset, int len, u32 val)
+{
+	struct pci_controller *hose = (struct pci_controller *) bus->sysdata;
+	struct ppc4xx_pciex_port *port =
+		&ppc4xx_pciex_ports[hose->indirect_type];
+	void __iomem *addr;
+	u32 gpl_cfg;
+
+	if (ppc4xx_pciex_validate_bdf(port, bus, devfn) != 0)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = ppc4xx_pciex_get_config_base(port, bus, devfn);
+
+	/*
+	 * Reading from configuration space of non-existing device can
+	 * generate transaction errors. For the read duration we suppress
+	 * assertion of machine check exceptions to avoid those.
+	 */
+	gpl_cfg = dcr_read(port->dcrs, DCRO_PEGPL_CFG);
+	dcr_write(port->dcrs, DCRO_PEGPL_CFG, gpl_cfg | GPL_DMER_MASK_DISA);
+
+	pr_debug("pcie-config-write: bus=%3d [%3d..%3d] devfn=0x%04x"
+		 " offset=0x%04x len=%d, addr=0x%p val=0x%08x\n",
+		 bus->number, hose->first_busno, hose->last_busno,
+		 devfn, offset, len, addr + offset, val);
+
+	switch (len) {
+	case 1:
+		out_8((u8 *)(addr + offset), val);
+		break;
+	case 2:
+		out_le16((u16 *)(addr + offset), val);
+		break;
+	default:
+		out_le32((u32 *)(addr + offset), val);
+		break;
+	}
+
+	dcr_write(port->dcrs, DCRO_PEGPL_CFG, gpl_cfg);
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static struct pci_ops ppc4xx_pciex_pci_ops =
+{
+	.read  = ppc4xx_pciex_read_config,
+	.write = ppc4xx_pciex_write_config,
+};
+
+static void __init ppc4xx_configure_pciex_POMs(struct ppc4xx_pciex_port *port,
+					       struct pci_controller *hose,
+					       void __iomem *mbase)
+{
+	u32 lah, lal, pciah, pcial, sa;
+	int i, j;
+
+	/* Setup outbound memory windows */
+	for (i = j = 0; i < 3; i++) {
+		struct resource *res = &hose->mem_resources[i];
+
+		/* we only care about memory windows */
+		if (!(res->flags & IORESOURCE_MEM))
+			continue;
+		if (j > 1) {
+			printk(KERN_WARNING "%s: Too many ranges\n",
+			       port->node->full_name);
+			break;
+		}
+
+		/* Calculate register values */
+		lah = RES_TO_U32_HIGH(res->start);
+		lal = RES_TO_U32_LOW(res->start);
+		pciah = RES_TO_U32_HIGH(res->start - hose->pci_mem_offset);
+		pcial = RES_TO_U32_LOW(res->start - hose->pci_mem_offset);
+		sa = res->end + 1 - res->start;
+		if (!is_power_of_2(sa) || sa < 0x100000 ||
+		    sa > 0xffffffffu) {
+			printk(KERN_WARNING "%s: Resource out of range\n",
+			       port->node->full_name);
+			continue;
+		}
+		sa = (0xffffffffu << ilog2(sa)) | 0x1;
+
+		/* Program register values */
+		switch (j) {
+		case 0:
+			out_le32(mbase + PECFG_POM0LAH, pciah);
+			out_le32(mbase + PECFG_POM0LAL, pcial);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR1BAH, lah);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR1BAL, lal);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR1MSKH, 0x7fffffff);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR1MSKL, sa | 3);
+			break;
+		case 1:
+			out_le32(mbase + PECFG_POM1LAH, pciah);
+			out_le32(mbase + PECFG_POM1LAL, pcial);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR2BAH, lah);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR2BAL, lal);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR2MSKH, 0x7fffffff);
+			dcr_write(port->dcrs, DCRO_PEGPL_OMR2MSKL, sa | 3);
+			break;
+		}
+		j++;
+	}
+
+	/* Configure IO, always 64K starting at 0 */
+	if (hose->io_resource.flags & IORESOURCE_IO) {
+		lah = RES_TO_U32_HIGH(hose->io_base_phys);
+		lal = RES_TO_U32_LOW(hose->io_base_phys);
+		out_le32(mbase + PECFG_POM2LAH, 0);
+		out_le32(mbase + PECFG_POM2LAL, 0);
+		dcr_write(port->dcrs, DCRO_PEGPL_OMR3BAH, lah);
+		dcr_write(port->dcrs, DCRO_PEGPL_OMR3BAL, lal);
+		dcr_write(port->dcrs, DCRO_PEGPL_OMR3MSKH, 0x7fffffff);
+		dcr_write(port->dcrs, DCRO_PEGPL_OMR3MSKL, 0xffff0000 | 3);
+	}
+}
+
+static void __init ppc4xx_configure_pciex_PIMs(struct ppc4xx_pciex_port *port,
+					       struct pci_controller *hose,
+					       void __iomem *mbase,
+					       struct resource *res)
+{
+	resource_size_t size = res->end - res->start + 1;
+	u64 sa;
+
+	/* Calculate window size */
+	sa = (0xffffffffffffffffull << ilog2(size));;
+	if (res->flags & IORESOURCE_PREFETCH)
+		sa |= 0x8;
+
+	out_le32(mbase + PECFG_BAR0HMPA, RES_TO_U32_HIGH(sa));
+	out_le32(mbase + PECFG_BAR0LMPA, RES_TO_U32_LOW(sa));
+
+	/* The setup of the split looks weird to me ... let's see if it works */
+	out_le32(mbase + PECFG_PIM0LAL, 0x00000000);
+	out_le32(mbase + PECFG_PIM0LAH, 0x00000000);
+	out_le32(mbase + PECFG_PIM1LAL, 0x00000000);
+	out_le32(mbase + PECFG_PIM1LAH, 0x00000000);
+	out_le32(mbase + PECFG_PIM01SAH, 0xffff0000);
+	out_le32(mbase + PECFG_PIM01SAL, 0x00000000);
+
+	/* Enable inbound mapping */
+	out_le32(mbase + PECFG_PIMEN, 0x1);
+
+	out_le32(mbase + PCI_BASE_ADDRESS_0, RES_TO_U32_LOW(res->start));
+	out_le32(mbase + PCI_BASE_ADDRESS_1, RES_TO_U32_HIGH(res->start));
+
+	/* Enable I/O, Mem, and Busmaster cycles */
+	out_le16(mbase + PCI_COMMAND,
+		 in_le16(mbase + PCI_COMMAND) |
+		 PCI_COMMAND_IO | PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER);
+}
+
+static void __init ppc4xx_pciex_port_setup_hose(struct ppc4xx_pciex_port *port)
+{
+	struct resource dma_window;
+	struct pci_controller *hose = NULL;
+	const int *bus_range;
+	int primary = 0, busses;
+	void __iomem *mbase = NULL, *cfg_data = NULL;
+
+	/* XXX FIXME: Handle endpoint mode properly */
+	if (port->endpoint)
+		return;
+
+	/* Check if primary bridge */
+	if (of_get_property(port->node, "primary", NULL))
+		primary = 1;
+
+	/* Get bus range if any */
+	bus_range = of_get_property(port->node, "bus-range", NULL);
+
+	/* Allocate the host controller data structure */
+	hose = pcibios_alloc_controller(port->node);
+	if (!hose)
+		goto fail;
+
+	/* We stick the port number in "indirect_type" so the config space
+	 * ops can retrieve the port data structure easily
+	 */
+	hose->indirect_type = port->index;
+
+	/* Get bus range */
+	hose->first_busno = bus_range ? bus_range[0] : 0x0;
+	hose->last_busno = bus_range ? bus_range[1] : 0xff;
+
+	/* Because of how big mapping the config space is (1M per bus), we
+	 * limit how many busses we support. In the long run, we could replace
+	 * that with something akin to kmap_atomic instead. We set aside 1 bus
+	 * for the host itself too.
+	 */
+	busses = hose->last_busno - hose->first_busno; /* This is off by 1 */
+	if (busses > MAX_PCIE_BUS_MAPPED) {
+		busses = MAX_PCIE_BUS_MAPPED;
+		hose->last_busno = hose->first_busno + busses;
+	}
+
+	/* We map the external config space in cfg_data and the host config
+	 * space in cfg_addr. External space is 1M per bus, internal space
+	 * is 4K
+	 */
+	cfg_data = ioremap(port->cfg_space.start +
+				 (hose->first_busno + 1) * 0x100000,
+				 busses * 0x100000);
+	mbase = ioremap(port->cfg_space.start + 0x10000000, 0x1000);
+	if (cfg_data == NULL || mbase == NULL) {
+		printk(KERN_ERR "%s: Can't map config space !",
+		       port->node->full_name);
+		goto fail;
+	}
+
+	hose->cfg_data = cfg_data;
+	hose->cfg_addr = mbase;
+
+	pr_debug("PCIE %s, bus %d..%d\n", port->node->full_name,
+		 hose->first_busno, hose->last_busno);
+	pr_debug("     config space mapped at: root @0x%p, other @0x%p\n",
+		 hose->cfg_addr, hose->cfg_data);
+
+	/* Setup config space */
+	hose->ops = &ppc4xx_pciex_pci_ops;
+	port->hose = hose;
+	mbase = (void __iomem *)hose->cfg_addr;
+
+	/*
+	 * Set bus numbers on our root port
+	 */
+	out_8(mbase + PCI_PRIMARY_BUS, hose->first_busno);
+	out_8(mbase + PCI_SECONDARY_BUS, hose->first_busno + 1);
+	out_8(mbase + PCI_SUBORDINATE_BUS, hose->last_busno);
+
+	/*
+	 * OMRs are already reset, also disable PIMs
+	 */
+	out_le32(mbase + PECFG_PIMEN, 0);
+
+	/* Parse outbound mapping resources */
+	pci_process_bridge_OF_ranges(hose, port->node, primary);
+
+	/* Parse inbound mapping resources */
+	if (ppc4xx_parse_dma_ranges(hose, mbase, &dma_window) != 0)
+		goto fail;
+
+	/* Configure outbound ranges POMs */
+	ppc4xx_configure_pciex_POMs(port, hose, mbase);
+
+	/* Configure inbound ranges PIMs */
+	ppc4xx_configure_pciex_PIMs(port, hose, mbase, &dma_window);
+
+	/* The root complex doesn't show up if we don't set some vendor
+	 * and device IDs into it. Those are the same bogus one that the
+	 * initial code in arch/ppc add. We might want to change that.
+	 */
+	out_le16(mbase + 0x200, 0xaaa0 + port->index);
+	out_le16(mbase + 0x202, 0xbed0 + port->index);
+
+	/* Set Class Code to PCI-PCI bridge and Revision Id to 1 */
+	out_le32(mbase + 0x208, 0x06040001);
+
+	printk(KERN_INFO "PCIE%d: successfully set as root-complex\n",
+	       port->index);
+	return;
+ fail:
+	if (hose)
+		pcibios_free_controller(hose);
+	if (cfg_data)
+		iounmap(cfg_data);
+	if (mbase)
+		iounmap(mbase);
+}
+
 static void __init ppc4xx_probe_pciex_bridge(struct device_node *np)
 {
-	/* NYI */
+	struct ppc4xx_pciex_port *port;
+	const u32 *pval;
+	int portno;
+	unsigned int dcrs;
+
+	/* First, proceed to core initialization as we assume there's
+	 * only one PCIe core in the system
+	 */
+	if (ppc4xx_pciex_check_core_init(np))
+		return;
+
+	/* Get the port number from the device-tree */
+	pval = of_get_property(np, "port", NULL);
+	if (pval == NULL) {
+		printk(KERN_ERR "PCIE: Can't find port number for %s\n",
+		       np->full_name);
+		return;
+	}
+	portno = *pval;
+	if (portno >= ppc4xx_pciex_port_count) {
+		printk(KERN_ERR "PCIE: port number out of range for %s\n",
+		       np->full_name);
+		return;
+	}
+	port = &ppc4xx_pciex_ports[portno];
+	port->index = portno;
+	port->node = of_node_get(np);
+	pval = of_get_property(np, "sdr-base", NULL);
+	if (pval == NULL) {
+		printk(KERN_ERR "PCIE: missing sdr-base for %s\n",
+		       np->full_name);
+		return;
+	}
+	port->sdr_base = *pval;
+
+	/* Fetch config space registers address */
+	if (of_address_to_resource(np, 0, &port->cfg_space)) {
+		printk(KERN_ERR "%s: Can't get PCI-E config space !",
+		       np->full_name);
+		return;
+	}
+	/* Fetch host bridge internal registers address */
+	if (of_address_to_resource(np, 1, &port->utl_regs)) {
+		printk(KERN_ERR "%s: Can't get UTL register base !",
+		       np->full_name);
+		return;
+	}
+
+	/* Map DCRs */
+	dcrs = dcr_resource_start(np, 0);
+	if (dcrs == 0) {
+		printk(KERN_ERR "%s: Can't get DCR register base !",
+		       np->full_name);
+		return;
+	}
+	port->dcrs = dcr_map(np, dcrs, dcr_resource_len(np, 0));
+
+	/* Initialize the port specific registers */
+	if (ppc4xx_pciex_port_init(port))
+		return;
+
+	/* Setup the linux hose data structure */
+	ppc4xx_pciex_port_setup_hose(port);
 }
 
+#endif /* CONFIG_PPC4xx_PCI_EXPRESS */
+
 static int __init ppc4xx_pci_find_bridges(void)
 {
 	struct device_node *np;
 
+#ifdef CONFIG_PPC4xx_PCI_EXPRESS
 	for_each_compatible_node(np, NULL, "ibm,plb-pciex")
 		ppc4xx_probe_pciex_bridge(np);
+#endif
 	for_each_compatible_node(np, NULL, "ibm,plb-pcix")
 		ppc4xx_probe_pcix_bridge(np);
 	for_each_compatible_node(np, NULL, "ibm,plb-pci")