Merge refs/heads/ieee80211-wifi from master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6 
diff --git a/arch/arm/common/gic.c b/arch/arm/common/gic.c
index 51dbf54..d749907 100644
--- a/arch/arm/common/gic.c
+++ b/arch/arm/common/gic.c
@@ -24,6 +24,7 @@
 #include <linux/kernel.h>
 #include <linux/list.h>
 #include <linux/smp.h>
+#include <linux/cpumask.h>
 
 #include <asm/irq.h>
 #include <asm/io.h>
diff --git a/arch/arm/kernel/time.c b/arch/arm/kernel/time.c
index 1b7fcd5..8880482 100644
--- a/arch/arm/kernel/time.c
+++ b/arch/arm/kernel/time.c
@@ -433,10 +433,12 @@
 {
 	struct dyn_tick_timer *dyn_tick = system_timer->dyn_tick;
 
-	write_seqlock(&xtime_lock);
-	if (dyn_tick->state & DYN_TICK_ENABLED)
-		dyn_tick->reprogram(next_timer_interrupt() - jiffies);
-	write_sequnlock(&xtime_lock);
+	if (dyn_tick) {
+		write_seqlock(&xtime_lock);
+		if (dyn_tick->state & DYN_TICK_ENABLED)
+			dyn_tick->reprogram(next_timer_interrupt() - jiffies);
+		write_sequnlock(&xtime_lock);
+	}
 }
 
 static ssize_t timer_show_dyn_tick(struct sys_device *dev, char *buf)
diff --git a/arch/arm/mach-ixp4xx/common.c b/arch/arm/mach-ixp4xx/common.c
index 04490a9..0422e90 100644
--- a/arch/arm/mach-ixp4xx/common.c
+++ b/arch/arm/mach-ixp4xx/common.c
@@ -38,90 +38,6 @@
 #include <asm/mach/irq.h>
 #include <asm/mach/time.h>
 
-enum ixp4xx_irq_type {
-	IXP4XX_IRQ_LEVEL, IXP4XX_IRQ_EDGE
-};
-static void ixp4xx_config_irq(unsigned irq, enum ixp4xx_irq_type type);
-
-/*************************************************************************
- * GPIO acces functions
- *************************************************************************/
-
-/*
- * Configure GPIO line for input, interrupt, or output operation
- *
- * TODO: Enable/disable the irq_desc based on interrupt or output mode.
- * TODO: Should these be named ixp4xx_gpio_?
- */
-void gpio_line_config(u8 line, u32 style)
-{
-	static const int gpio2irq[] = {
-		6, 7, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29
-	};
-	u32 enable;
-	volatile u32 *int_reg;
-	u32 int_style;
-	enum ixp4xx_irq_type irq_type;
-
-	enable = *IXP4XX_GPIO_GPOER;
-
-	if (style & IXP4XX_GPIO_OUT) {
-		enable &= ~((1) << line);
-	} else if (style & IXP4XX_GPIO_IN) {
-		enable |= ((1) << line);
-
-		switch (style & IXP4XX_GPIO_INTSTYLE_MASK)
-		{
-		case (IXP4XX_GPIO_ACTIVE_HIGH):
-			int_style = IXP4XX_GPIO_STYLE_ACTIVE_HIGH;
-			irq_type = IXP4XX_IRQ_LEVEL;
-			break;
-		case (IXP4XX_GPIO_ACTIVE_LOW):
-			int_style = IXP4XX_GPIO_STYLE_ACTIVE_LOW;
-			irq_type = IXP4XX_IRQ_LEVEL;
-			break;
-		case (IXP4XX_GPIO_RISING_EDGE):
-			int_style = IXP4XX_GPIO_STYLE_RISING_EDGE;
-			irq_type = IXP4XX_IRQ_EDGE;
-			break;
-		case (IXP4XX_GPIO_FALLING_EDGE):
-			int_style = IXP4XX_GPIO_STYLE_FALLING_EDGE;
-			irq_type = IXP4XX_IRQ_EDGE;
-			break;
-		case (IXP4XX_GPIO_TRANSITIONAL):
-			int_style = IXP4XX_GPIO_STYLE_TRANSITIONAL;
-			irq_type = IXP4XX_IRQ_EDGE;
-			break;
-		default:
-			int_style = IXP4XX_GPIO_STYLE_ACTIVE_HIGH;
-			irq_type = IXP4XX_IRQ_LEVEL;
-			break;
-		}
-
-		if (style & IXP4XX_GPIO_INTSTYLE_MASK)
-			ixp4xx_config_irq(gpio2irq[line], irq_type);
-
-		if (line >= 8) {	/* pins 8-15 */ 
-			line -= 8;
-			int_reg = IXP4XX_GPIO_GPIT2R;
-		}
-		else {			/* pins 0-7 */
-			int_reg = IXP4XX_GPIO_GPIT1R;
-		}
-
-		/* Clear the style for the appropriate pin */
-		*int_reg &= ~(IXP4XX_GPIO_STYLE_CLEAR << 
-		    		(line * IXP4XX_GPIO_STYLE_SIZE));
-
-		/* Set the new style */
-		*int_reg |= (int_style << (line * IXP4XX_GPIO_STYLE_SIZE));
-	}
-
-	*IXP4XX_GPIO_GPOER = enable;
-}
-
-EXPORT_SYMBOL(gpio_line_config);
-
 /*************************************************************************
  * IXP4xx chipset I/O mapping
  *************************************************************************/
@@ -165,6 +81,69 @@
  *       (be it PCI or something else) configures that GPIO line
  *       as an IRQ.
  **************************************************************************/
+enum ixp4xx_irq_type {
+	IXP4XX_IRQ_LEVEL, IXP4XX_IRQ_EDGE
+};
+
+static void ixp4xx_config_irq(unsigned irq, enum ixp4xx_irq_type type);
+
+/*
+ * IRQ -> GPIO mapping table
+ */
+static int irq2gpio[32] = {
+	-1, -1, -1, -1, -1, -1,  0,  1,
+	-1, -1, -1, -1, -1, -1, -1, -1,
+	-1, -1, -1,  2,  3,  4,  5,  6,
+	 7,  8,  9, 10, 11, 12, -1, -1,
+};
+
+static int ixp4xx_set_irq_type(unsigned int irq, unsigned int type)
+{
+	int line = irq2gpio[irq];
+	u32 int_style;
+	enum ixp4xx_irq_type irq_type;
+	volatile u32 *int_reg;
+
+	/*
+	 * Only for GPIO IRQs
+	 */
+	if (line < 0)
+		return -EINVAL;
+
+	if (type & IRQT_BOTHEDGE) {
+		int_style = IXP4XX_GPIO_STYLE_TRANSITIONAL;
+		irq_type = IXP4XX_IRQ_EDGE;
+	} else  if (type & IRQT_RISING) {
+		int_style = IXP4XX_GPIO_STYLE_RISING_EDGE;
+		irq_type = IXP4XX_IRQ_EDGE;
+	} else if (type & IRQT_FALLING) {
+		int_style = IXP4XX_GPIO_STYLE_FALLING_EDGE;
+		irq_type = IXP4XX_IRQ_EDGE;
+	} else if (type & IRQT_HIGH) {
+		int_style = IXP4XX_GPIO_STYLE_ACTIVE_HIGH;
+		irq_type = IXP4XX_IRQ_LEVEL;
+	} else if (type & IRQT_LOW) {
+		int_style = IXP4XX_GPIO_STYLE_ACTIVE_LOW;
+		irq_type = IXP4XX_IRQ_LEVEL;
+	}
+
+	ixp4xx_config_irq(irq, irq_type);
+
+	if (line >= 8) {	/* pins 8-15 */
+		line -= 8;
+		int_reg = IXP4XX_GPIO_GPIT2R;
+	} else {		/* pins 0-7 */
+		int_reg = IXP4XX_GPIO_GPIT1R;
+	}
+
+	/* Clear the style for the appropriate pin */
+	*int_reg &= ~(IXP4XX_GPIO_STYLE_CLEAR <<
+	    		(line * IXP4XX_GPIO_STYLE_SIZE));
+
+	/* Set the new style */
+	*int_reg |= (int_style << (line * IXP4XX_GPIO_STYLE_SIZE));
+}
+
 static void ixp4xx_irq_mask(unsigned int irq)
 {
 	if (cpu_is_ixp46x() && irq >= 32)
@@ -183,12 +162,6 @@
 
 static void ixp4xx_irq_ack(unsigned int irq)
 {
-	static int irq2gpio[32] = {
-		-1, -1, -1, -1, -1, -1,  0,  1,
-		-1, -1, -1, -1, -1, -1, -1, -1,
-		-1, -1, -1,  2,  3,  4,  5,  6,
-		 7,  8,  9, 10, 11, 12, -1, -1,
-	};
 	int line = (irq < 32) ? irq2gpio[irq] : -1;
 
 	if (line >= 0)
@@ -209,12 +182,14 @@
 	.ack	= ixp4xx_irq_mask,
 	.mask	= ixp4xx_irq_mask,
 	.unmask	= ixp4xx_irq_level_unmask,
+	.type	= ixp4xx_set_irq_type
 };
 
 static struct irqchip ixp4xx_irq_edge_chip = {
 	.ack	= ixp4xx_irq_ack,
 	.mask	= ixp4xx_irq_mask,
 	.unmask	= ixp4xx_irq_unmask,
+	.type	= ixp4xx_set_irq_type
 };
 
 static void ixp4xx_config_irq(unsigned irq, enum ixp4xx_irq_type type)
diff --git a/arch/arm/mach-ixp4xx/coyote-pci.c b/arch/arm/mach-ixp4xx/coyote-pci.c
index afafb42..60de8a9 100644
--- a/arch/arm/mach-ixp4xx/coyote-pci.c
+++ b/arch/arm/mach-ixp4xx/coyote-pci.c
@@ -30,11 +30,8 @@
 
 void __init coyote_pci_preinit(void)
 {
-	gpio_line_config(COYOTE_PCI_SLOT0_PIN,
-			IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
-
-	gpio_line_config(COYOTE_PCI_SLOT1_PIN,
-			IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
+	set_irq_type(IRQ_COYOTE_PCI_SLOT0, IRQT_LOW);
+	set_irq_type(IRQ_COYOTE_PCI_SLOT1, IRQT_LOW);
 
 	gpio_line_isr_clear(COYOTE_PCI_SLOT0_PIN);
 	gpio_line_isr_clear(COYOTE_PCI_SLOT1_PIN);
diff --git a/arch/arm/mach-ixp4xx/coyote-setup.c b/arch/arm/mach-ixp4xx/coyote-setup.c
index 411ea99..8b2f253 100644
--- a/arch/arm/mach-ixp4xx/coyote-setup.c
+++ b/arch/arm/mach-ixp4xx/coyote-setup.c
@@ -24,11 +24,6 @@
 #include <asm/mach/arch.h>
 #include <asm/mach/flash.h>
 
-void __init coyote_map_io(void)
-{
-	ixp4xx_map_io();
-}
-
 static struct flash_platform_data coyote_flash_data = {
 	.map_name	= "cfi_probe",
 	.width		= 2,
@@ -107,7 +102,7 @@
 	.phys_ram	= PHYS_OFFSET,
 	.phys_io	= IXP4XX_PERIPHERAL_BASE_PHYS,
 	.io_pg_offst	= ((IXP4XX_PERIPHERAL_BASE_VIRT) >> 18) & 0xfffc,
-	.map_io		= coyote_map_io,
+	.map_io		= ixp4xx_map_io,
 	.init_irq	= ixp4xx_init_irq,
 	.timer		= &ixp4xx_timer,
 	.boot_params	= 0x0100,
@@ -125,7 +120,7 @@
 	.phys_ram	= PHYS_OFFSET,
 	.phys_io	= IXP4XX_PERIPHERAL_BASE_PHYS,
 	.io_pg_offst	= ((IXP4XX_PERIPHERAL_BASE_VIRT) >> 18) & 0xfffc,
-	.map_io		= coyote_map_io,
+	.map_io		= ixp4xx_map_io,
 	.init_irq	= ixp4xx_init_irq,
 	.timer		= &ixp4xx_timer,
 	.boot_params	= 0x0100,
diff --git a/arch/arm/mach-ixp4xx/gtwx5715-pci.c b/arch/arm/mach-ixp4xx/gtwx5715-pci.c
index b180358..a66484b6 100644
--- a/arch/arm/mach-ixp4xx/gtwx5715-pci.c
+++ b/arch/arm/mach-ixp4xx/gtwx5715-pci.c
@@ -35,26 +35,20 @@
 extern int ixp4xx_setup(int nr, struct pci_sys_data *sys);
 extern struct pci_bus *ixp4xx_scan_bus(int nr, struct pci_sys_data *sys);
 
-        /*
-        * The exact GPIO pins and IRQs are defined in arch-ixp4xx/gtwx5715.h
-        * Slot 0 isn't actually populated with a card connector but
-        * we initialize it anyway in case a future version has the
-        * slot populated or someone with good soldering skills has
-        * some free time.
-        */
 
-
-static void gtwx5715_init_gpio(u8 pin, u32 style)
-{
-	gpio_line_config(pin, style | IXP4XX_GPIO_ACTIVE_LOW);
-
-	if (style & IXP4XX_GPIO_IN) gpio_line_isr_clear(pin);
-}
-
+/*
+ * The exact GPIO pins and IRQs are defined in arch-ixp4xx/gtwx5715.h
+ * Slot 0 isn't actually populated with a card connector but
+ * we initialize it anyway in case a future version has the
+ * slot populated or someone with good soldering skills has
+ * some free time.
+ */
 void __init gtwx5715_pci_preinit(void)
 {
-	gtwx5715_init_gpio(GTWX5715_PCI_SLOT0_INTA_GPIO,	IXP4XX_GPIO_IN);
-	gtwx5715_init_gpio(GTWX5715_PCI_SLOT1_INTA_GPIO,	IXP4XX_GPIO_IN);
+	set_irq_type(GTWX5715_PCI_SLOT0_INTA_IRQ, IRQT_LOW);
+	set_irq_type(GTWX5715_PCI_SLOT0_INTB_IRQ, IRQT_LOW);
+	set_irq_type(GTWX5715_PCI_SLOT1_INTA_IRQ, IRQT_LOW);
+	set_irq_type(GTWX5715_PCI_SLOT1_INTB_IRQ, IRQT_LOW);
 
 	ixp4xx_pci_preinit();
 }
diff --git a/arch/arm/mach-ixp4xx/gtwx5715-setup.c b/arch/arm/mach-ixp4xx/gtwx5715-setup.c
index 333459d..3fd92c5 100644
--- a/arch/arm/mach-ixp4xx/gtwx5715-setup.c
+++ b/arch/arm/mach-ixp4xx/gtwx5715-setup.c
@@ -101,12 +101,6 @@
 	.resource	= gtwx5715_uart_resources,
 };
 
-
-void __init gtwx5715_map_io(void)
-{
-	ixp4xx_map_io();
-}
-
 static struct flash_platform_data gtwx5715_flash_data = {
 	.map_name	= "cfi_probe",
 	.width		= 2,
@@ -144,7 +138,7 @@
 	.phys_ram	= PHYS_OFFSET,
 	.phys_io	= IXP4XX_UART2_BASE_PHYS,
 	.io_pg_offst	= ((IXP4XX_UART2_BASE_VIRT) >> 18) & 0xfffc,
-	.map_io		= gtwx5715_map_io,
+	.map_io		= ixp4xx_map_io,
 	.init_irq	= ixp4xx_init_irq,
 	.timer		= &ixp4xx_timer,
 	.boot_params	= 0x0100,
diff --git a/arch/arm/mach-ixp4xx/ixdp425-pci.c b/arch/arm/mach-ixp4xx/ixdp425-pci.c
index c2ab9eb..f9a1d3e 100644
--- a/arch/arm/mach-ixp4xx/ixdp425-pci.c
+++ b/arch/arm/mach-ixp4xx/ixdp425-pci.c
@@ -27,14 +27,10 @@
 
 void __init ixdp425_pci_preinit(void)
 {
-	gpio_line_config(IXDP425_PCI_INTA_PIN,
-				IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
-	gpio_line_config(IXDP425_PCI_INTB_PIN, 
-				IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
-	gpio_line_config(IXDP425_PCI_INTC_PIN, 
-				IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
-	gpio_line_config(IXDP425_PCI_INTD_PIN, 
-				IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
+	set_irq_type(IRQ_IXDP425_PCI_INTA, IRQT_LOW);
+	set_irq_type(IRQ_IXDP425_PCI_INTB, IRQT_LOW);
+	set_irq_type(IRQ_IXDP425_PCI_INTC, IRQT_LOW);
+	set_irq_type(IRQ_IXDP425_PCI_INTD, IRQT_LOW);
 
 	gpio_line_isr_clear(IXDP425_PCI_INTA_PIN);
 	gpio_line_isr_clear(IXDP425_PCI_INTB_PIN);
diff --git a/arch/arm/mach-ixp4xx/ixdp425-setup.c b/arch/arm/mach-ixp4xx/ixdp425-setup.c
index fa0646c..6c14ff3 100644
--- a/arch/arm/mach-ixp4xx/ixdp425-setup.c
+++ b/arch/arm/mach-ixp4xx/ixdp425-setup.c
@@ -24,11 +24,6 @@
 #include <asm/mach/arch.h>
 #include <asm/mach/flash.h>
 
-void __init ixdp425_map_io(void) 
-{
-	ixp4xx_map_io();
-}
-
 static struct flash_platform_data ixdp425_flash_data = {
 	.map_name	= "cfi_probe",
 	.width		= 2,
@@ -133,7 +128,7 @@
 	.phys_ram	= PHYS_OFFSET,
 	.phys_io	= IXP4XX_PERIPHERAL_BASE_PHYS,
 	.io_pg_offst	= ((IXP4XX_PERIPHERAL_BASE_VIRT) >> 18) & 0xfffc,
-	.map_io		= ixdp425_map_io,
+	.map_io		= ixp4xx_map_io,
 	.init_irq	= ixp4xx_init_irq,
 	.timer		= &ixp4xx_timer,
 	.boot_params	= 0x0100,
@@ -145,7 +140,7 @@
 	.phys_ram	= PHYS_OFFSET,
 	.phys_io	= IXP4XX_PERIPHERAL_BASE_PHYS,
 	.io_pg_offst	= ((IXP4XX_PERIPHERAL_BASE_VIRT) >> 18) & 0xfffc,
-	.map_io		= ixdp425_map_io,
+	.map_io		= ixp4xx_map_io,
 	.init_irq	= ixp4xx_init_irq,
 	.timer		= &ixp4xx_timer,
 	.boot_params	= 0x0100,
@@ -157,7 +152,7 @@
 	.phys_ram	= PHYS_OFFSET,
 	.phys_io	= IXP4XX_PERIPHERAL_BASE_PHYS,
 	.io_pg_offst	= ((IXP4XX_PERIPHERAL_BASE_VIRT) >> 18) & 0xfffc,
-	.map_io		= ixdp425_map_io,
+	.map_io		= ixp4xx_map_io,
 	.init_irq	= ixp4xx_init_irq,
 	.timer		= &ixp4xx_timer,
 	.boot_params	= 0x0100,
@@ -176,7 +171,7 @@
 	.phys_ram	= PHYS_OFFSET,
 	.phys_io	= IXP4XX_PERIPHERAL_BASE_PHYS,
 	.io_pg_offst	= ((IXP4XX_PERIPHERAL_BASE_VIRT) >> 18) & 0xfffc,
-	.map_io		= ixdp425_map_io,
+	.map_io		= ixp4xx_map_io,
 	.init_irq	= ixp4xx_init_irq,
 	.timer		= &ixp4xx_timer,
 	.boot_params	= 0x0100,
diff --git a/arch/arm/mach-ixp4xx/ixdpg425-pci.c b/arch/arm/mach-ixp4xx/ixdpg425-pci.c
index ce4563f..fe5e766 100644
--- a/arch/arm/mach-ixp4xx/ixdpg425-pci.c
+++ b/arch/arm/mach-ixp4xx/ixdpg425-pci.c
@@ -29,8 +29,8 @@
 
 void __init ixdpg425_pci_preinit(void)
 {
-	gpio_line_config(6, IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
-	gpio_line_config(7, IXP4XX_GPIO_IN | IXP4XX_GPIO_ACTIVE_LOW);
+	set_irq_type(IRQ_IXP4XX_GPIO6, IRQT_LOW);
+	set_irq_type(IRQ_IXP4XX_GPIO7, IRQT_LOW);
 
 	gpio_line_isr_clear(6);
 	gpio_line_isr_clear(7);
diff --git a/arch/arm/mach-s3c2410/clock.c b/arch/arm/mach-s3c2410/clock.c
index 9a66050..f596082 100644
--- a/arch/arm/mach-s3c2410/clock.c
+++ b/arch/arm/mach-s3c2410/clock.c
@@ -388,6 +388,7 @@
 				unsigned long hclk,
 				unsigned long pclk)
 {
+	unsigned long clkslow = __raw_readl(S3C2410_CLKSLOW);
 	struct clk *clkp = init_clocks;
 	int ptr;
 	int ret;
@@ -446,5 +447,13 @@
 		}
 	}
 
+	/* show the clock-slow value */
+
+	printk("CLOCK: Slow mode (%ld.%ld MHz), %s, MPLL %s, UPLL %s\n",
+	       print_mhz(xtal / ( 2 * S3C2410_CLKSLOW_GET_SLOWVAL(clkslow))),
+	       (clkslow & S3C2410_CLKSLOW_SLOW) ? "slow" : "fast",
+	       (clkslow & S3C2410_CLKSLOW_MPLL_OFF) ? "off" : "on",
+	       (clkslow & S3C2410_CLKSLOW_UCLK_OFF) ? "off" : "on");
+
 	return 0;
 }
diff --git a/arch/arm/mach-s3c2410/s3c2440-clock.c b/arch/arm/mach-s3c2410/s3c2440-clock.c
index b018a1f..c67e097 100644
--- a/arch/arm/mach-s3c2410/s3c2440-clock.c
+++ b/arch/arm/mach-s3c2410/s3c2440-clock.c
@@ -68,6 +68,7 @@
 static int s3c2440_clk_add(struct sys_device *sysdev)
 {
 	unsigned long upllcon = __raw_readl(S3C2410_UPLLCON);
+	unsigned long camdivn = __raw_readl(S3C2440_CAMDIVN);
 	struct clk *clk_h;
 	struct clk *clk_p;
 	struct clk *clk_xtal;
@@ -80,8 +81,9 @@
 
 	s3c2440_clk_upll.rate = s3c2410_get_pll(upllcon, clk_xtal->rate);
 
-	printk("S3C2440: Clock Support, UPLL %ld.%03ld MHz\n",
-	       print_mhz(s3c2440_clk_upll.rate));
+	printk("S3C2440: Clock Support, UPLL %ld.%03ld MHz, DVS %s\n",
+	       print_mhz(s3c2440_clk_upll.rate),
+	       (camdivn & S3C2440_CAMDIVN_DVSEN) ? "on" : "off");
 
 	clk_p = clk_get(NULL, "pclk");
 	clk_h = clk_get(NULL, "hclk");
diff --git a/arch/arm/mm/alignment.c b/arch/arm/mm/alignment.c
index 81f4a8a..4b39d86 100644
--- a/arch/arm/mm/alignment.c
+++ b/arch/arm/mm/alignment.c
@@ -45,7 +45,7 @@
 
 #define LDST_P_EQ_U(i)	((((i) ^ ((i) >> 1)) & (1 << 23)) == 0)
 
-#define LDSTH_I_BIT(i)	(i & (1 << 22))		/* half-word immed	*/
+#define LDSTHD_I_BIT(i)	(i & (1 << 22))		/* double/half-word immed */
 #define LDM_S_BIT(i)	(i & (1 << 22))		/* write CPSR from SPSR	*/
 
 #define RN_BITS(i)	((i >> 16) & 15)	/* Rn			*/
@@ -68,6 +68,7 @@
 static unsigned long ai_skipped;
 static unsigned long ai_half;
 static unsigned long ai_word;
+static unsigned long ai_dword;
 static unsigned long ai_multi;
 static int ai_usermode;
 
@@ -93,6 +94,8 @@
 	p += sprintf(p, "Skipped:\t%lu\n", ai_skipped);
 	p += sprintf(p, "Half:\t\t%lu\n", ai_half);
 	p += sprintf(p, "Word:\t\t%lu\n", ai_word);
+	if (cpu_architecture() >= CPU_ARCH_ARMv5TE)
+		p += sprintf(p, "DWord:\t\t%lu\n", ai_dword);
 	p += sprintf(p, "Multi:\t\t%lu\n", ai_multi);
 	p += sprintf(p, "User faults:\t%i (%s)\n", ai_usermode,
 			usermode_action[ai_usermode]);
@@ -283,12 +286,6 @@
 {
 	unsigned int rd = RD_BITS(instr);
 
-	if ((instr & 0x01f00ff0) == 0x01000090)
-		goto swp;
-
-	if ((instr & 0x90) != 0x90 || (instr & 0x60) == 0)
-		goto bad;
-
 	ai_half += 1;
 
 	if (user_mode(regs))
@@ -323,10 +320,47 @@
 
  	return TYPE_LDST;
 
- swp:
-	printk(KERN_ERR "Alignment trap: not handling swp instruction\n");
- bad:
-	return TYPE_ERROR;
+ fault:
+	return TYPE_FAULT;
+}
+
+static int
+do_alignment_ldrdstrd(unsigned long addr, unsigned long instr,
+		      struct pt_regs *regs)
+{
+	unsigned int rd = RD_BITS(instr);
+
+	ai_dword += 1;
+
+	if (user_mode(regs))
+		goto user;
+
+	if ((instr & 0xf0) == 0xd0) {
+		unsigned long val;
+		get32_unaligned_check(val, addr);
+		regs->uregs[rd] = val;
+		get32_unaligned_check(val, addr+4);
+		regs->uregs[rd+1] = val;
+	} else {
+		put32_unaligned_check(regs->uregs[rd], addr);
+		put32_unaligned_check(regs->uregs[rd+1], addr+4);
+	}
+
+	return TYPE_LDST;
+
+ user:
+	if ((instr & 0xf0) == 0xd0) {
+		unsigned long val;
+		get32t_unaligned_check(val, addr);
+		regs->uregs[rd] = val;
+		get32t_unaligned_check(val, addr+4);
+		regs->uregs[rd+1] = val;
+	} else {
+		put32t_unaligned_check(regs->uregs[rd], addr);
+		put32t_unaligned_check(regs->uregs[rd+1], addr+4);
+	}
+
+	return TYPE_LDST;
 
  fault:
 	return TYPE_FAULT;
@@ -617,12 +651,20 @@
 	regs->ARM_pc += thumb_mode(regs) ? 2 : 4;
 
 	switch (CODING_BITS(instr)) {
-	case 0x00000000:	/* ldrh or strh */
-		if (LDSTH_I_BIT(instr))
+	case 0x00000000:	/* 3.13.4 load/store instruction extensions */
+		if (LDSTHD_I_BIT(instr))
 			offset.un = (instr & 0xf00) >> 4 | (instr & 15);
 		else
 			offset.un = regs->uregs[RM_BITS(instr)];
-		handler = do_alignment_ldrhstrh;
+
+		if ((instr & 0x000000f0) == 0x000000b0 || /* LDRH, STRH */
+		    (instr & 0x001000f0) == 0x001000f0)   /* LDRSH */
+			handler = do_alignment_ldrhstrh;
+		else if ((instr & 0x001000f0) == 0x000000d0 || /* LDRD */
+			 (instr & 0x001000f0) == 0x000000f0)   /* STRD */
+			handler = do_alignment_ldrdstrd;
+		else
+			goto bad;
 		break;
 
 	case 0x04000000:	/* ldr or str immediate */
diff --git a/arch/arm/mm/mm-armv.c b/arch/arm/mm/mm-armv.c
index 3c655c5..4dae00b 100644
--- a/arch/arm/mm/mm-armv.c
+++ b/arch/arm/mm/mm-armv.c
@@ -275,11 +275,9 @@
 	int i;
 
 	for (i = 0; i < 16; i += 1) {
-		alloc_init_section(virt, phys & SUPERSECTION_MASK,
-				   prot | PMD_SECT_SUPER);
+		alloc_init_section(virt, phys, prot | PMD_SECT_SUPER);
 
 		virt += (PGDIR_SIZE / 2);
-		phys += (PGDIR_SIZE / 2);
 	}
 }
 
diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index 8098813..3deced6 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -383,6 +383,12 @@
 
 endif
 
+if PM
+
+source "arch/ia64/kernel/cpufreq/Kconfig"
+
+endif
+
 endmenu
 
 if !IA64_HP_SIM
diff --git a/arch/ia64/hp/sim/boot/fw-emu.c b/arch/ia64/hp/sim/boot/fw-emu.c
index 5c46928..30fdfb1 100644
--- a/arch/ia64/hp/sim/boot/fw-emu.c
+++ b/arch/ia64/hp/sim/boot/fw-emu.c
@@ -237,17 +237,6 @@
 	return ((struct sal_ret_values) {status, r9, r10, r11});
 }
 
-
-/*
- * This is here to work around a bug in egcs-1.1.1b that causes the
- * compiler to crash (seems like a bug in the new alias analysis code.
- */
-void *
-id (long addr)
-{
-	return (void *) addr;
-}
-
 struct ia64_boot_param *
 sys_fw_init (const char *args, int arglen)
 {
diff --git a/arch/ia64/ia32/ia32_signal.c b/arch/ia64/ia32/ia32_signal.c
index ebb89be..aa891c9 100644
--- a/arch/ia64/ia32/ia32_signal.c
+++ b/arch/ia64/ia32/ia32_signal.c
@@ -29,7 +29,6 @@
 #include <asm/uaccess.h>
 #include <asm/rse.h>
 #include <asm/sigcontext.h>
-#include <asm/segment.h>
 
 #include "ia32priv.h"
 
diff --git a/arch/ia64/kernel/Makefile b/arch/ia64/kernel/Makefile
index e1fb68d..b242594 100644
--- a/arch/ia64/kernel/Makefile
+++ b/arch/ia64/kernel/Makefile
@@ -20,6 +20,7 @@
 obj-$(CONFIG_NUMA)		+= numa.o
 obj-$(CONFIG_PERFMON)		+= perfmon_default_smpl.o
 obj-$(CONFIG_IA64_CYCLONE)	+= cyclone.o
+obj-$(CONFIG_CPU_FREQ)		+= cpufreq/
 obj-$(CONFIG_IA64_MCA_RECOVERY)	+= mca_recovery.o
 obj-$(CONFIG_KPROBES)		+= kprobes.o jprobes.o
 obj-$(CONFIG_IA64_UNCACHED_ALLOCATOR)	+= uncached.o
diff --git a/arch/ia64/kernel/cpufreq/Kconfig b/arch/ia64/kernel/cpufreq/Kconfig
new file mode 100644
index 0000000..2d9d527
--- /dev/null
+++ b/arch/ia64/kernel/cpufreq/Kconfig
@@ -0,0 +1,29 @@
+
+#
+# CPU Frequency scaling
+#
+
+menu "CPU Frequency scaling"
+
+source "drivers/cpufreq/Kconfig"
+
+if CPU_FREQ
+
+comment "CPUFreq processor drivers"
+
+config IA64_ACPI_CPUFREQ
+	tristate "ACPI Processor P-States driver"
+	select CPU_FREQ_TABLE
+	depends on ACPI_PROCESSOR
+	help
+	This driver adds a CPUFreq driver which utilizes the ACPI
+	Processor Performance States.
+
+	For details, take a look at <file:Documentation/cpu-freq/>.
+
+	If in doubt, say N.
+
+endif   # CPU_FREQ
+
+endmenu
+
diff --git a/arch/ia64/kernel/cpufreq/Makefile b/arch/ia64/kernel/cpufreq/Makefile
new file mode 100644
index 0000000..f748d34
--- /dev/null
+++ b/arch/ia64/kernel/cpufreq/Makefile
@@ -0,0 +1 @@
+obj-$(CONFIG_IA64_ACPI_CPUFREQ)		+= acpi-cpufreq.o
diff --git a/arch/ia64/kernel/cpufreq/acpi-cpufreq.c b/arch/ia64/kernel/cpufreq/acpi-cpufreq.c
new file mode 100644
index 0000000..da4d5cf
--- /dev/null
+++ b/arch/ia64/kernel/cpufreq/acpi-cpufreq.c
@@ -0,0 +1,499 @@
+/*
+ * arch/ia64/kernel/cpufreq/acpi-cpufreq.c
+ * This file provides the ACPI based P-state support. This
+ * module works with generic cpufreq infrastructure. Most of
+ * the code is based on i386 version
+ * (arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c)
+ *
+ * Copyright (C) 2005 Intel Corp
+ *      Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
+ */
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/cpufreq.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
+#include <asm/io.h>
+#include <asm/uaccess.h>
+#include <asm/pal.h>
+
+#include <linux/acpi.h>
+#include <acpi/processor.h>
+
+#define dprintk(msg...) cpufreq_debug_printk(CPUFREQ_DEBUG_DRIVER, "acpi-cpufreq", msg)
+
+MODULE_AUTHOR("Venkatesh Pallipadi");
+MODULE_DESCRIPTION("ACPI Processor P-States Driver");
+MODULE_LICENSE("GPL");
+
+
+struct cpufreq_acpi_io {
+	struct acpi_processor_performance	acpi_data;
+	struct cpufreq_frequency_table		*freq_table;
+	unsigned int				resume;
+};
+
+static struct cpufreq_acpi_io	*acpi_io_data[NR_CPUS];
+
+static struct cpufreq_driver acpi_cpufreq_driver;
+
+
+static int
+processor_set_pstate (
+	u32	value)
+{
+	s64 retval;
+
+	dprintk("processor_set_pstate\n");
+
+	retval = ia64_pal_set_pstate((u64)value);
+
+	if (retval) {
+		dprintk("Failed to set freq to 0x%x, with error 0x%x\n",
+		        value, retval);
+		return -ENODEV;
+	}
+	return (int)retval;
+}
+
+
+static int
+processor_get_pstate (
+	u32	*value)
+{
+	u64	pstate_index = 0;
+	s64 	retval;
+
+	dprintk("processor_get_pstate\n");
+
+	retval = ia64_pal_get_pstate(&pstate_index);
+	*value = (u32) pstate_index;
+
+	if (retval)
+		dprintk("Failed to get current freq with "
+		        "error 0x%x, idx 0x%x\n", retval, *value);
+
+	return (int)retval;
+}
+
+
+/* To be used only after data->acpi_data is initialized */
+static unsigned
+extract_clock (
+	struct cpufreq_acpi_io *data,
+	unsigned value,
+	unsigned int cpu)
+{
+	unsigned long i;
+
+	dprintk("extract_clock\n");
+
+	for (i = 0; i < data->acpi_data.state_count; i++) {
+		if (value >= data->acpi_data.states[i].control)
+			return data->acpi_data.states[i].core_frequency;
+	}
+	return data->acpi_data.states[i-1].core_frequency;
+}
+
+
+static unsigned int
+processor_get_freq (
+	struct cpufreq_acpi_io	*data,
+	unsigned int		cpu)
+{
+	int			ret = 0;
+	u32			value = 0;
+	cpumask_t		saved_mask;
+	unsigned long 		clock_freq;
+
+	dprintk("processor_get_freq\n");
+
+	saved_mask = current->cpus_allowed;
+	set_cpus_allowed(current, cpumask_of_cpu(cpu));
+	if (smp_processor_id() != cpu) {
+		ret = -EAGAIN;
+		goto migrate_end;
+	}
+
+	/*
+	 * processor_get_pstate gets the average frequency since the
+	 * last get. So, do two PAL_get_freq()...
+	 */
+	ret = processor_get_pstate(&value);
+	ret = processor_get_pstate(&value);
+
+	if (ret) {
+		set_cpus_allowed(current, saved_mask);
+		printk(KERN_WARNING "get performance failed with error %d\n",
+		       ret);
+		ret = -EAGAIN;
+		goto migrate_end;
+	}
+	clock_freq = extract_clock(data, value, cpu);
+	ret = (clock_freq*1000);
+
+migrate_end:
+	set_cpus_allowed(current, saved_mask);
+	return ret;
+}
+
+
+static int
+processor_set_freq (
+	struct cpufreq_acpi_io	*data,
+	unsigned int		cpu,
+	int			state)
+{
+	int			ret = 0;
+	u32			value = 0;
+	struct cpufreq_freqs    cpufreq_freqs;
+	cpumask_t		saved_mask;
+	int			retval;
+
+	dprintk("processor_set_freq\n");
+
+	saved_mask = current->cpus_allowed;
+	set_cpus_allowed(current, cpumask_of_cpu(cpu));
+	if (smp_processor_id() != cpu) {
+		retval = -EAGAIN;
+		goto migrate_end;
+	}
+
+	if (state == data->acpi_data.state) {
+		if (unlikely(data->resume)) {
+			dprintk("Called after resume, resetting to P%d\n", state);
+			data->resume = 0;
+		} else {
+			dprintk("Already at target state (P%d)\n", state);
+			retval = 0;
+			goto migrate_end;
+		}
+	}
+
+	dprintk("Transitioning from P%d to P%d\n",
+		data->acpi_data.state, state);
+
+	/* cpufreq frequency struct */
+	cpufreq_freqs.cpu = cpu;
+	cpufreq_freqs.old = data->freq_table[data->acpi_data.state].frequency;
+	cpufreq_freqs.new = data->freq_table[state].frequency;
+
+	/* notify cpufreq */
+	cpufreq_notify_transition(&cpufreq_freqs, CPUFREQ_PRECHANGE);
+
+	/*
+	 * First we write the target state's 'control' value to the
+	 * control_register.
+	 */
+
+	value = (u32) data->acpi_data.states[state].control;
+
+	dprintk("Transitioning to state: 0x%08x\n", value);
+
+	ret = processor_set_pstate(value);
+	if (ret) {
+		unsigned int tmp = cpufreq_freqs.new;
+		cpufreq_notify_transition(&cpufreq_freqs, CPUFREQ_POSTCHANGE);
+		cpufreq_freqs.new = cpufreq_freqs.old;
+		cpufreq_freqs.old = tmp;
+		cpufreq_notify_transition(&cpufreq_freqs, CPUFREQ_PRECHANGE);
+		cpufreq_notify_transition(&cpufreq_freqs, CPUFREQ_POSTCHANGE);
+		printk(KERN_WARNING "Transition failed with error %d\n", ret);
+		retval = -ENODEV;
+		goto migrate_end;
+	}
+
+	cpufreq_notify_transition(&cpufreq_freqs, CPUFREQ_POSTCHANGE);
+
+	data->acpi_data.state = state;
+
+	retval = 0;
+
+migrate_end:
+	set_cpus_allowed(current, saved_mask);
+	return (retval);
+}
+
+
+static unsigned int
+acpi_cpufreq_get (
+	unsigned int		cpu)
+{
+	struct cpufreq_acpi_io *data = acpi_io_data[cpu];
+
+	dprintk("acpi_cpufreq_get\n");
+
+	return processor_get_freq(data, cpu);
+}
+
+
+static int
+acpi_cpufreq_target (
+	struct cpufreq_policy   *policy,
+	unsigned int target_freq,
+	unsigned int relation)
+{
+	struct cpufreq_acpi_io *data = acpi_io_data[policy->cpu];
+	unsigned int next_state = 0;
+	unsigned int result = 0;
+
+	dprintk("acpi_cpufreq_setpolicy\n");
+
+	result = cpufreq_frequency_table_target(policy,
+			data->freq_table, target_freq, relation, &next_state);
+	if (result)
+		return (result);
+
+	result = processor_set_freq(data, policy->cpu, next_state);
+
+	return (result);
+}
+
+
+static int
+acpi_cpufreq_verify (
+	struct cpufreq_policy   *policy)
+{
+	unsigned int result = 0;
+	struct cpufreq_acpi_io *data = acpi_io_data[policy->cpu];
+
+	dprintk("acpi_cpufreq_verify\n");
+
+	result = cpufreq_frequency_table_verify(policy,
+			data->freq_table);
+
+	return (result);
+}
+
+
+/*
+ * processor_init_pdc - let BIOS know about the SMP capabilities
+ * of this driver
+ * @perf: processor-specific acpi_io_data struct
+ * @cpu: CPU being initialized
+ *
+ * To avoid issues with legacy OSes, some BIOSes require to be informed of
+ * the SMP capabilities of OS P-state driver. Here we set the bits in _PDC
+ * accordingly. Actual call to _PDC is done in driver/acpi/processor.c
+ */
+static void
+processor_init_pdc (
+		struct acpi_processor_performance *perf,
+		unsigned int cpu,
+		struct acpi_object_list *obj_list
+		)
+{
+	union acpi_object *obj;
+	u32 *buf;
+
+	dprintk("processor_init_pdc\n");
+
+	perf->pdc = NULL;
+	/* Initialize pdc. It will be used later. */
+	if (!obj_list)
+		return;
+
+	if (!(obj_list->count && obj_list->pointer))
+		return;
+
+	obj = obj_list->pointer;
+	if ((obj->buffer.length == 12) && obj->buffer.pointer) {
+		buf = (u32 *)obj->buffer.pointer;
+       		buf[0] = ACPI_PDC_REVISION_ID;
+       		buf[1] = 1;
+       		buf[2] = ACPI_PDC_EST_CAPABILITY_SMP;
+		perf->pdc = obj_list;
+	}
+	return;
+}
+
+
+static int
+acpi_cpufreq_cpu_init (
+	struct cpufreq_policy   *policy)
+{
+	unsigned int		i;
+	unsigned int		cpu = policy->cpu;
+	struct cpufreq_acpi_io	*data;
+	unsigned int		result = 0;
+
+	union acpi_object		arg0 = {ACPI_TYPE_BUFFER};
+	u32				arg0_buf[3];
+	struct acpi_object_list 	arg_list = {1, &arg0};
+
+	dprintk("acpi_cpufreq_cpu_init\n");
+	/* setup arg_list for _PDC settings */
+        arg0.buffer.length = 12;
+        arg0.buffer.pointer = (u8 *) arg0_buf;
+
+	data = kmalloc(sizeof(struct cpufreq_acpi_io), GFP_KERNEL);
+	if (!data)
+		return (-ENOMEM);
+
+	memset(data, 0, sizeof(struct cpufreq_acpi_io));
+
+	acpi_io_data[cpu] = data;
+
+	processor_init_pdc(&data->acpi_data, cpu, &arg_list);
+	result = acpi_processor_register_performance(&data->acpi_data, cpu);
+	data->acpi_data.pdc = NULL;
+
+	if (result)
+		goto err_free;
+
+	/* capability check */
+	if (data->acpi_data.state_count <= 1) {
+		dprintk("No P-States\n");
+		result = -ENODEV;
+		goto err_unreg;
+	}
+
+	if ((data->acpi_data.control_register.space_id !=
+					ACPI_ADR_SPACE_FIXED_HARDWARE) ||
+	    (data->acpi_data.status_register.space_id !=
+					ACPI_ADR_SPACE_FIXED_HARDWARE)) {
+		dprintk("Unsupported address space [%d, %d]\n",
+			(u32) (data->acpi_data.control_register.space_id),
+			(u32) (data->acpi_data.status_register.space_id));
+		result = -ENODEV;
+		goto err_unreg;
+	}
+
+	/* alloc freq_table */
+	data->freq_table = kmalloc(sizeof(struct cpufreq_frequency_table) *
+	                           (data->acpi_data.state_count + 1),
+	                           GFP_KERNEL);
+	if (!data->freq_table) {
+		result = -ENOMEM;
+		goto err_unreg;
+	}
+
+	/* detect transition latency */
+	policy->cpuinfo.transition_latency = 0;
+	for (i=0; i<data->acpi_data.state_count; i++) {
+		if ((data->acpi_data.states[i].transition_latency * 1000) >
+		    policy->cpuinfo.transition_latency) {
+			policy->cpuinfo.transition_latency =
+			    data->acpi_data.states[i].transition_latency * 1000;
+		}
+	}
+	policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
+
+	policy->cur = processor_get_freq(data, policy->cpu);
+
+	/* table init */
+	for (i = 0; i <= data->acpi_data.state_count; i++)
+	{
+		data->freq_table[i].index = i;
+		if (i < data->acpi_data.state_count) {
+			data->freq_table[i].frequency =
+			      data->acpi_data.states[i].core_frequency * 1000;
+		} else {
+			data->freq_table[i].frequency = CPUFREQ_TABLE_END;
+		}
+	}
+
+	result = cpufreq_frequency_table_cpuinfo(policy, data->freq_table);
+	if (result) {
+		goto err_freqfree;
+	}
+
+	/* notify BIOS that we exist */
+	acpi_processor_notify_smm(THIS_MODULE);
+
+	printk(KERN_INFO "acpi-cpufreq: CPU%u - ACPI performance management "
+	       "activated.\n", cpu);
+
+	for (i = 0; i < data->acpi_data.state_count; i++)
+		dprintk("     %cP%d: %d MHz, %d mW, %d uS, %d uS, 0x%x 0x%x\n",
+			(i == data->acpi_data.state?'*':' '), i,
+			(u32) data->acpi_data.states[i].core_frequency,
+			(u32) data->acpi_data.states[i].power,
+			(u32) data->acpi_data.states[i].transition_latency,
+			(u32) data->acpi_data.states[i].bus_master_latency,
+			(u32) data->acpi_data.states[i].status,
+			(u32) data->acpi_data.states[i].control);
+
+	cpufreq_frequency_table_get_attr(data->freq_table, policy->cpu);
+
+	/* the first call to ->target() should result in us actually
+	 * writing something to the appropriate registers. */
+	data->resume = 1;
+
+	return (result);
+
+ err_freqfree:
+	kfree(data->freq_table);
+ err_unreg:
+	acpi_processor_unregister_performance(&data->acpi_data, cpu);
+ err_free:
+	kfree(data);
+	acpi_io_data[cpu] = NULL;
+
+	return (result);
+}
+
+
+static int
+acpi_cpufreq_cpu_exit (
+	struct cpufreq_policy   *policy)
+{
+	struct cpufreq_acpi_io *data = acpi_io_data[policy->cpu];
+
+	dprintk("acpi_cpufreq_cpu_exit\n");
+
+	if (data) {
+		cpufreq_frequency_table_put_attr(policy->cpu);
+		acpi_io_data[policy->cpu] = NULL;
+		acpi_processor_unregister_performance(&data->acpi_data,
+		                                      policy->cpu);
+		kfree(data);
+	}
+
+	return (0);
+}
+
+
+static struct freq_attr* acpi_cpufreq_attr[] = {
+	&cpufreq_freq_attr_scaling_available_freqs,
+	NULL,
+};
+
+
+static struct cpufreq_driver acpi_cpufreq_driver = {
+	.verify 	= acpi_cpufreq_verify,
+	.target 	= acpi_cpufreq_target,
+	.get 		= acpi_cpufreq_get,
+	.init		= acpi_cpufreq_cpu_init,
+	.exit		= acpi_cpufreq_cpu_exit,
+	.name		= "acpi-cpufreq",
+	.owner		= THIS_MODULE,
+	.attr           = acpi_cpufreq_attr,
+};
+
+
+static int __init
+acpi_cpufreq_init (void)
+{
+	dprintk("acpi_cpufreq_init\n");
+
+ 	return cpufreq_register_driver(&acpi_cpufreq_driver);
+}
+
+
+static void __exit
+acpi_cpufreq_exit (void)
+{
+	dprintk("acpi_cpufreq_exit\n");
+
+	cpufreq_unregister_driver(&acpi_cpufreq_driver);
+	return;
+}
+
+
+late_initcall(acpi_cpufreq_init);
+module_exit(acpi_cpufreq_exit);
+
diff --git a/arch/ia64/kernel/sys_ia64.c b/arch/ia64/kernel/sys_ia64.c
index 770fab3..f2dbcd1 100644
--- a/arch/ia64/kernel/sys_ia64.c
+++ b/arch/ia64/kernel/sys_ia64.c
@@ -35,7 +35,7 @@
 		return -ENOMEM;
 
 #ifdef CONFIG_HUGETLB_PAGE
-	if (REGION_NUMBER(addr) == REGION_HPAGE)
+	if (REGION_NUMBER(addr) == RGN_HPAGE)
 		addr = 0;
 #endif
 	if (!addr)
diff --git a/arch/ia64/kernel/uncached.c b/arch/ia64/kernel/uncached.c
index 490dfc9..4e9d06c 100644
--- a/arch/ia64/kernel/uncached.c
+++ b/arch/ia64/kernel/uncached.c
@@ -184,7 +184,7 @@
 {
 	int node;
 
-	node = nasid_to_cnodeid(NASID_GET(maddr));
+	node = paddr_to_nid(maddr - __IA64_UNCACHED_OFFSET);
 
 	dprintk(KERN_DEBUG "uncached_free_page(%lx) on node %i\n", maddr, node);
 
@@ -217,7 +217,7 @@
 
 	memset((char *)vstart, 0, length);
 
-	node = nasid_to_cnodeid(NASID_GET(start));
+	node = paddr_to_nid(start);
 
 	for (; vstart < vend ; vstart += PAGE_SIZE) {
 		dprintk(KERN_INFO "sticking %lx into the pool!\n", vstart);
diff --git a/arch/ia64/lib/Makefile b/arch/ia64/lib/Makefile
index 1902c3c..799407e 100644
--- a/arch/ia64/lib/Makefile
+++ b/arch/ia64/lib/Makefile
@@ -6,7 +6,7 @@
 
 lib-y := __divsi3.o __udivsi3.o __modsi3.o __umodsi3.o			\
 	__divdi3.o __udivdi3.o __moddi3.o __umoddi3.o			\
-	bitop.o checksum.o clear_page.o csum_partial_copy.o copy_page.o	\
+	bitop.o checksum.o clear_page.o csum_partial_copy.o		\
 	clear_user.o strncpy_from_user.o strlen_user.o strnlen_user.o	\
 	flush.o ip_fast_csum.o do_csum.o				\
 	memset.o strlen.o swiotlb.o
diff --git a/arch/ia64/lib/swiotlb.c b/arch/ia64/lib/swiotlb.c
index ab7b3ad..dbc0b3e 100644
--- a/arch/ia64/lib/swiotlb.c
+++ b/arch/ia64/lib/swiotlb.c
@@ -93,8 +93,7 @@
 setup_io_tlb_npages(char *str)
 {
 	if (isdigit(*str)) {
-		io_tlb_nslabs = simple_strtoul(str, &str, 0) <<
-			(PAGE_SHIFT - IO_TLB_SHIFT);
+		io_tlb_nslabs = simple_strtoul(str, &str, 0);
 		/* avoid tail segment of size < IO_TLB_SEGSIZE */
 		io_tlb_nslabs = ALIGN(io_tlb_nslabs, IO_TLB_SEGSIZE);
 	}
@@ -117,7 +116,7 @@
 	unsigned long i;
 
 	if (!io_tlb_nslabs) {
-		io_tlb_nslabs = (default_size >> PAGE_SHIFT);
+		io_tlb_nslabs = (default_size >> IO_TLB_SHIFT);
 		io_tlb_nslabs = ALIGN(io_tlb_nslabs, IO_TLB_SEGSIZE);
 	}
 
diff --git a/arch/ia64/mm/hugetlbpage.c b/arch/ia64/mm/hugetlbpage.c
index e0a776a..2d13889 100644
--- a/arch/ia64/mm/hugetlbpage.c
+++ b/arch/ia64/mm/hugetlbpage.c
@@ -76,7 +76,7 @@
 		return -EINVAL;
 	if (addr & ~HPAGE_MASK)
 		return -EINVAL;
-	if (REGION_NUMBER(addr) != REGION_HPAGE)
+	if (REGION_NUMBER(addr) != RGN_HPAGE)
 		return -EINVAL;
 
 	return 0;
@@ -87,7 +87,7 @@
 	struct page *page;
 	pte_t *ptep;
 
-	if (REGION_NUMBER(addr) != REGION_HPAGE)
+	if (REGION_NUMBER(addr) != RGN_HPAGE)
 		return ERR_PTR(-EINVAL);
 
 	ptep = huge_pte_offset(mm, addr);
@@ -142,8 +142,8 @@
 		return -ENOMEM;
 	if (len & ~HPAGE_MASK)
 		return -EINVAL;
-	/* This code assumes that REGION_HPAGE != 0. */
-	if ((REGION_NUMBER(addr) != REGION_HPAGE) || (addr & (HPAGE_SIZE - 1)))
+	/* This code assumes that RGN_HPAGE != 0. */
+	if ((REGION_NUMBER(addr) != RGN_HPAGE) || (addr & (HPAGE_SIZE - 1)))
 		addr = HPAGE_REGION_BASE;
 	else
 		addr = ALIGN(addr, HPAGE_SIZE);
diff --git a/arch/ia64/pci/pci.c b/arch/ia64/pci/pci.c
index f9472c5..9977c12 100644
--- a/arch/ia64/pci/pci.c
+++ b/arch/ia64/pci/pci.c
@@ -24,7 +24,6 @@
 
 #include <asm/machvec.h>
 #include <asm/page.h>
-#include <asm/segment.h>
 #include <asm/system.h>
 #include <asm/io.h>
 #include <asm/sal.h>
diff --git a/arch/ia64/sn/include/tio.h b/arch/ia64/sn/include/tio.h
index 0139124..6b2e7b7 100644
--- a/arch/ia64/sn/include/tio.h
+++ b/arch/ia64/sn/include/tio.h
@@ -3,7 +3,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (C) 2000-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 2000-2005 Silicon Graphics, Inc. All rights reserved.
  */
 
 #ifndef _ASM_IA64_SN_TIO_H
@@ -26,6 +26,10 @@
 #define TIO_ITTE_VALID_MASK	0x1
 #define TIO_ITTE_VALID_SHIFT	16
 
+#define TIO_ITTE_WIDGET(itte) \
+	(((itte) >> TIO_ITTE_WIDGET_SHIFT) & TIO_ITTE_WIDGET_MASK)
+#define TIO_ITTE_VALID(itte) \
+	(((itte) >> TIO_ITTE_VALID_SHIFT) & TIO_ITTE_VALID_MASK)
 
 #define TIO_ITTE_PUT(nasid, bigwin, widget, addr, valid) \
         REMOTE_HUB_S((nasid), TIO_ITTE(bigwin), \
diff --git a/arch/ia64/sn/include/xtalk/hubdev.h b/arch/ia64/sn/include/xtalk/hubdev.h
index 580a1c0..71c2b27 100644
--- a/arch/ia64/sn/include/xtalk/hubdev.h
+++ b/arch/ia64/sn/include/xtalk/hubdev.h
@@ -3,7 +3,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (C) 1992 - 1997, 2000-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 1992 - 1997, 2000-2005 Silicon Graphics, Inc. All rights reserved.
  */
 #ifndef _ASM_IA64_SN_XTALK_HUBDEV_H
 #define _ASM_IA64_SN_XTALK_HUBDEV_H
@@ -16,6 +16,9 @@
 #define IIO_ITTE_WIDGET_MASK    ((1<<IIO_ITTE_WIDGET_BITS)-1)
 #define IIO_ITTE_WIDGET_SHIFT   8
 
+#define IIO_ITTE_WIDGET(itte)	\
+	(((itte) >> IIO_ITTE_WIDGET_SHIFT) & IIO_ITTE_WIDGET_MASK)
+
 /*
  * Use the top big window as a surrogate for the first small window
  */
@@ -34,7 +37,8 @@
 	unsigned long sfdl_force_int_addr;
 	unsigned long sfdl_flush_value;
 	volatile unsigned long *sfdl_flush_addr;
-	uint64_t sfdl_persistent_busnum;
+	uint32_t sfdl_persistent_busnum;
+	uint32_t sfdl_persistent_segment;
 	struct pcibus_info *sfdl_pcibus_info;
 	spinlock_t sfdl_flush_lock;
 };
@@ -58,7 +62,8 @@
 
 	void				*hdi_nodepda;
 	void				*hdi_node_vertex;
-	void				*hdi_xtalk_vertex;
+	uint32_t			max_segment_number;
+	uint32_t			max_pcibus_number;
 };
 
 extern void hubdev_init_node(nodepda_t *, cnodeid_t);
diff --git a/arch/ia64/sn/kernel/bte.c b/arch/ia64/sn/kernel/bte.c
index 647deae..45854c6 100644
--- a/arch/ia64/sn/kernel/bte.c
+++ b/arch/ia64/sn/kernel/bte.c
@@ -29,16 +29,30 @@
 
 /* two interfaces on two btes */
 #define MAX_INTERFACES_TO_TRY		4
+#define MAX_NODES_TO_TRY		2
 
 static struct bteinfo_s *bte_if_on_node(nasid_t nasid, int interface)
 {
 	nodepda_t *tmp_nodepda;
 
+	if (nasid_to_cnodeid(nasid) == -1)
+		return (struct bteinfo_s *)NULL;;
+
 	tmp_nodepda = NODEPDA(nasid_to_cnodeid(nasid));
 	return &tmp_nodepda->bte_if[interface];
 
 }
 
+static inline void bte_start_transfer(struct bteinfo_s *bte, u64 len, u64 mode)
+{
+	if (is_shub2()) {
+		BTE_CTRL_STORE(bte, (IBLS_BUSY | ((len) | (mode) << 24)));
+	} else {
+		BTE_LNSTAT_STORE(bte, len);
+		BTE_CTRL_STORE(bte, mode);
+	}
+}
+
 /************************************************************************
  * Block Transfer Engine copy related functions.
  *
@@ -67,13 +81,15 @@
 {
 	u64 transfer_size;
 	u64 transfer_stat;
+	u64 notif_phys_addr;
 	struct bteinfo_s *bte;
 	bte_result_t bte_status;
 	unsigned long irq_flags;
 	unsigned long itc_end = 0;
-	struct bteinfo_s *btes_to_try[MAX_INTERFACES_TO_TRY];
-	int bte_if_index;
-	int bte_pri, bte_sec;
+	int nasid_to_try[MAX_NODES_TO_TRY];
+	int my_nasid = get_nasid();
+	int bte_if_index, nasid_index;
+	int bte_first, btes_per_node = BTES_PER_NODE;
 
 	BTE_PRINTK(("bte_copy(0x%lx, 0x%lx, 0x%lx, 0x%lx, 0x%p)\n",
 		    src, dest, len, mode, notification));
@@ -86,36 +102,26 @@
 		 (src & L1_CACHE_MASK) || (dest & L1_CACHE_MASK));
 	BUG_ON(!(len < ((BTE_LEN_MASK + 1) << L1_CACHE_SHIFT)));
 
-	/* CPU 0 (per node) tries bte0 first, CPU 1 try bte1 first */
-	if (cpuid_to_subnode(smp_processor_id()) == 0) {
-		bte_pri = 0;
-		bte_sec = 1;
-	} else {
-		bte_pri = 1;
-		bte_sec = 0;
-	}
+	/*
+	 * Start with interface corresponding to cpu number
+	 */
+	bte_first = raw_smp_processor_id() % btes_per_node;
 
 	if (mode & BTE_USE_DEST) {
 		/* try remote then local */
-		btes_to_try[0] = bte_if_on_node(NASID_GET(dest), bte_pri);
-		btes_to_try[1] = bte_if_on_node(NASID_GET(dest), bte_sec);
+		nasid_to_try[0] = NASID_GET(dest);
 		if (mode & BTE_USE_ANY) {
-			btes_to_try[2] = bte_if_on_node(get_nasid(), bte_pri);
-			btes_to_try[3] = bte_if_on_node(get_nasid(), bte_sec);
+			nasid_to_try[1] = my_nasid;
 		} else {
-			btes_to_try[2] = NULL;
-			btes_to_try[3] = NULL;
+			nasid_to_try[1] = (int)NULL;
 		}
 	} else {
 		/* try local then remote */
-		btes_to_try[0] = bte_if_on_node(get_nasid(), bte_pri);
-		btes_to_try[1] = bte_if_on_node(get_nasid(), bte_sec);
+		nasid_to_try[0] = my_nasid;
 		if (mode & BTE_USE_ANY) {
-			btes_to_try[2] = bte_if_on_node(NASID_GET(dest), bte_pri);
-			btes_to_try[3] = bte_if_on_node(NASID_GET(dest), bte_sec);
+			nasid_to_try[1] = NASID_GET(dest);
 		} else {
-			btes_to_try[2] = NULL;
-			btes_to_try[3] = NULL;
+			nasid_to_try[1] = (int)NULL;
 		}
 	}
 
@@ -123,11 +129,12 @@
 	do {
 		local_irq_save(irq_flags);
 
-		bte_if_index = 0;
+		bte_if_index = bte_first;
+		nasid_index = 0;
 
 		/* Attempt to lock one of the BTE interfaces. */
-		while (bte_if_index < MAX_INTERFACES_TO_TRY) {
-			bte = btes_to_try[bte_if_index++];
+		while (nasid_index < MAX_NODES_TO_TRY) {
+			bte = bte_if_on_node(nasid_to_try[nasid_index],bte_if_index);
 
 			if (bte == NULL) {
 				continue;
@@ -143,6 +150,15 @@
 					break;
 				}
 			}
+
+			bte_if_index = (bte_if_index + 1) % btes_per_node; /* Next interface */
+			if (bte_if_index == bte_first) {
+				/*
+				 * We've tried all interfaces on this node
+				 */
+				nasid_index++;
+			}
+
 			bte = NULL;
 		}
 
@@ -169,7 +185,13 @@
 
 	/* Initialize the notification to a known value. */
 	*bte->most_rcnt_na = BTE_WORD_BUSY;
+	notif_phys_addr = TO_PHYS(ia64_tpa((unsigned long)bte->most_rcnt_na));
 
+	if (is_shub2()) {
+		src = SH2_TIO_PHYS_TO_DMA(src);
+		dest = SH2_TIO_PHYS_TO_DMA(dest);
+		notif_phys_addr = SH2_TIO_PHYS_TO_DMA(notif_phys_addr);
+	}
 	/* Set the source and destination registers */
 	BTE_PRINTKV(("IBSA = 0x%lx)\n", (TO_PHYS(src))));
 	BTE_SRC_STORE(bte, TO_PHYS(src));
@@ -177,14 +199,12 @@
 	BTE_DEST_STORE(bte, TO_PHYS(dest));
 
 	/* Set the notification register */
-	BTE_PRINTKV(("IBNA = 0x%lx)\n",
-		     TO_PHYS(ia64_tpa((unsigned long)bte->most_rcnt_na))));
-	BTE_NOTIF_STORE(bte,
-			TO_PHYS(ia64_tpa((unsigned long)bte->most_rcnt_na)));
+	BTE_PRINTKV(("IBNA = 0x%lx)\n", notif_phys_addr));
+	BTE_NOTIF_STORE(bte, notif_phys_addr);
 
 	/* Initiate the transfer */
 	BTE_PRINTK(("IBCT = 0x%lx)\n", BTE_VALID_MODE(mode)));
-	BTE_START_TRANSFER(bte, transfer_size, BTE_VALID_MODE(mode));
+	bte_start_transfer(bte, transfer_size, BTE_VALID_MODE(mode));
 
 	itc_end = ia64_get_itc() + (40000000 * local_cpu_data->cyc_per_usec);
 
@@ -195,6 +215,7 @@
 	}
 
 	while ((transfer_stat = *bte->most_rcnt_na) == BTE_WORD_BUSY) {
+		cpu_relax();
 		if (ia64_get_itc() > itc_end) {
 			BTE_PRINTK(("BTE timeout nasid 0x%x bte%d IBLS = 0x%lx na 0x%lx\n",
 				NASID_GET(bte->bte_base_addr), bte->bte_num,
diff --git a/arch/ia64/sn/kernel/huberror.c b/arch/ia64/sn/kernel/huberror.c
index 5c39b43..5c5eb01 100644
--- a/arch/ia64/sn/kernel/huberror.c
+++ b/arch/ia64/sn/kernel/huberror.c
@@ -76,7 +76,7 @@
 	 */
 	REMOTE_HUB_S(hubdev_info->hdi_nasid, IIO_ICDR, (IIO_ICDR_PND | crbnum));
 	while (REMOTE_HUB_L(hubdev_info->hdi_nasid, IIO_ICDR) & IIO_ICDR_PND)
-		udelay(1);
+		cpu_relax();
 
 }
 
diff --git a/arch/ia64/sn/kernel/io_init.c b/arch/ia64/sn/kernel/io_init.c
index 414cdf2..4564ed0 100644
--- a/arch/ia64/sn/kernel/io_init.c
+++ b/arch/ia64/sn/kernel/io_init.c
@@ -18,6 +18,7 @@
 #include <asm/sn/simulator.h>
 #include <asm/sn/sn_sal.h>
 #include <asm/sn/tioca_provider.h>
+#include <asm/sn/tioce_provider.h>
 #include "xtalk/hubdev.h"
 #include "xtalk/xwidgetdev.h"
 
@@ -44,6 +45,9 @@
 
 struct sn_pcibus_provider *sn_pci_provider[PCIIO_ASIC_MAX_TYPES];	/* indexed by asic type */
 
+static int max_segment_number = 0; /* Default highest segment number */
+static int max_pcibus_number = 255; /* Default highest pci bus number */
+
 /*
  * Hooks and struct for unsupported pci providers
  */
@@ -157,13 +161,28 @@
 	uint64_t nasid;
 	int i, widget;
 
+	/*
+	 * Get SGI Specific HUB chipset information.
+	 * Inform Prom that this kernel can support domain bus numbering.
+	 */
 	for (i = 0; i < numionodes; i++) {
 		hubdev = (struct hubdev_info *)(NODEPDA(i)->pdinfo);
 		nasid = cnodeid_to_nasid(i);
+		hubdev->max_segment_number = 0xffffffff;
+		hubdev->max_pcibus_number = 0xff;
 		status = sal_get_hubdev_info(nasid, (uint64_t) __pa(hubdev));
 		if (status)
 			continue;
 
+		/* Save the largest Domain and pcibus numbers found. */
+		if (hubdev->max_segment_number) {
+			/*
+			 * Dealing with a Prom that supports segments.
+			 */
+			max_segment_number = hubdev->max_segment_number;
+			max_pcibus_number = hubdev->max_pcibus_number;
+		}
+
 		/* Attach the error interrupt handlers */
 		if (nasid & 1)
 			ice_error_init(hubdev);
@@ -230,7 +249,7 @@
 void sn_pci_fixup_slot(struct pci_dev *dev)
 {
 	int idx;
-	int segment = 0;
+	int segment = pci_domain_nr(dev->bus);
 	int status = 0;
 	struct pcibus_bussoft *bs;
  	struct pci_bus *host_pci_bus;
@@ -283,9 +302,9 @@
  	 * PCI host_pci_dev struct and set up host bus linkages
  	 */
 
- 	bus_no = SN_PCIDEV_INFO(dev)->pdi_slot_host_handle >> 32;
+ 	bus_no = (SN_PCIDEV_INFO(dev)->pdi_slot_host_handle >> 32) & 0xff;
  	devfn = SN_PCIDEV_INFO(dev)->pdi_slot_host_handle & 0xffffffff;
- 	host_pci_bus = pci_find_bus(pci_domain_nr(dev->bus), bus_no);
+ 	host_pci_bus = pci_find_bus(segment, bus_no);
  	host_pci_dev = pci_get_slot(host_pci_bus, devfn);
 
 	SN_PCIDEV_INFO(dev)->host_pci_dev = host_pci_dev;
@@ -333,6 +352,7 @@
 	prom_bussoft_ptr = __va(prom_bussoft_ptr);
 
  	controller = kcalloc(1,sizeof(struct pci_controller), GFP_KERNEL);
+	controller->segment = segment;
  	if (!controller)
  		BUG();
 
@@ -390,7 +410,7 @@
 	if (controller->node >= num_online_nodes()) {
 		struct pcibus_bussoft *b = SN_PCIBUS_BUSSOFT(bus);
 
-		printk(KERN_WARNING "Device ASIC=%u XID=%u PBUSNUM=%lu"
+		printk(KERN_WARNING "Device ASIC=%u XID=%u PBUSNUM=%u"
 				    "L_IO=%lx L_MEM=%lx BASE=%lx\n",
 			b->bs_asic_type, b->bs_xid, b->bs_persist_busnum,
 			b->bs_legacy_io, b->bs_legacy_mem, b->bs_base);
@@ -445,6 +465,7 @@
 static int __init sn_pci_init(void)
 {
 	int i = 0;
+	int j = 0;
 	struct pci_dev *pci_dev = NULL;
 	extern void sn_init_cpei_timer(void);
 #ifdef CONFIG_PROC_FS
@@ -464,6 +485,7 @@
 
 	pcibr_init_provider();
 	tioca_init_provider();
+	tioce_init_provider();
 
 	/*
 	 * This is needed to avoid bounce limit checks in the blk layer
@@ -479,8 +501,9 @@
 #endif
 
 	/* busses are not known yet ... */
-	for (i = 0; i < PCI_BUSES_TO_SCAN; i++)
-		sn_pci_controller_fixup(0, i, NULL);
+	for (i = 0; i <= max_segment_number; i++)
+		for (j = 0; j <= max_pcibus_number; j++)
+			sn_pci_controller_fixup(i, j, NULL);
 
 	/*
 	 * Generic Linux PCI Layer has created the pci_bus and pci_dev 
diff --git a/arch/ia64/sn/kernel/irq.c b/arch/ia64/sn/kernel/irq.c
index 84d276a..9fc7463 100644
--- a/arch/ia64/sn/kernel/irq.c
+++ b/arch/ia64/sn/kernel/irq.c
@@ -5,7 +5,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (c) 2000-2004 Silicon Graphics, Inc.  All Rights Reserved.
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.  All Rights Reserved.
  */
 
 #include <linux/irq.h>
@@ -76,16 +76,14 @@
 
 static void sn_ack_irq(unsigned int irq)
 {
-	uint64_t event_occurred, mask = 0;
-	int nasid;
+	u64 event_occurred, mask = 0;
 
 	irq = irq & 0xff;
-	nasid = get_nasid();
 	event_occurred =
-	    HUB_L((uint64_t *) GLOBAL_MMR_ADDR(nasid, SH_EVENT_OCCURRED));
+	    HUB_L((u64*)LOCAL_MMR_ADDR(SH_EVENT_OCCURRED));
 	mask = event_occurred & SH_ALL_INT_MASK;
-	HUB_S((uint64_t *) GLOBAL_MMR_ADDR(nasid, SH_EVENT_OCCURRED_ALIAS),
-		 mask);
+	HUB_S((u64*)LOCAL_MMR_ADDR(SH_EVENT_OCCURRED_ALIAS),
+	      mask);
 	__set_bit(irq, (volatile void *)pda->sn_in_service_ivecs);
 
 	move_irq(irq);
@@ -93,15 +91,12 @@
 
 static void sn_end_irq(unsigned int irq)
 {
-	int nasid;
 	int ivec;
-	uint64_t event_occurred;
+	u64 event_occurred;
 
 	ivec = irq & 0xff;
 	if (ivec == SGI_UART_VECTOR) {
-		nasid = get_nasid();
-		event_occurred = HUB_L((uint64_t *) GLOBAL_MMR_ADDR
-				       (nasid, SH_EVENT_OCCURRED));
+		event_occurred = HUB_L((u64*)LOCAL_MMR_ADDR (SH_EVENT_OCCURRED));
 		/* If the UART bit is set here, we may have received an
 		 * interrupt from the UART that the driver missed.  To
 		 * make sure, we IPI ourselves to force us to look again.
@@ -132,6 +127,7 @@
 		int local_widget, status;
 		nasid_t local_nasid;
 		struct sn_irq_info *new_irq_info;
+		struct sn_pcibus_provider *pci_provider;
 
 		new_irq_info = kmalloc(sizeof(struct sn_irq_info), GFP_ATOMIC);
 		if (new_irq_info == NULL)
@@ -171,8 +167,9 @@
 		new_irq_info->irq_cpuid = cpuid;
 		register_intr_pda(new_irq_info);
 
-		if (IS_PCI_BRIDGE_ASIC(new_irq_info->irq_bridge_type))
-			pcibr_change_devices_irq(new_irq_info);
+		pci_provider = sn_pci_provider[new_irq_info->irq_bridge_type];
+		if (pci_provider && pci_provider->target_interrupt)
+			(pci_provider->target_interrupt)(new_irq_info);
 
 		spin_lock(&sn_irq_info_lock);
 		list_replace_rcu(&sn_irq_info->list, &new_irq_info->list);
@@ -317,6 +314,16 @@
 	pci_dev_put(pci_dev);
 }
 
+static inline void
+sn_call_force_intr_provider(struct sn_irq_info *sn_irq_info)
+{
+	struct sn_pcibus_provider *pci_provider;
+
+	pci_provider = sn_pci_provider[sn_irq_info->irq_bridge_type];
+	if (pci_provider && pci_provider->force_interrupt)
+		(*pci_provider->force_interrupt)(sn_irq_info);
+}
+
 static void force_interrupt(int irq)
 {
 	struct sn_irq_info *sn_irq_info;
@@ -325,11 +332,9 @@
 		return;
 
 	rcu_read_lock();
-	list_for_each_entry_rcu(sn_irq_info, sn_irq_lh[irq], list) {
-		if (IS_PCI_BRIDGE_ASIC(sn_irq_info->irq_bridge_type) &&
-		    (sn_irq_info->irq_bridge != NULL))
-			pcibr_force_interrupt(sn_irq_info);
-	}
+	list_for_each_entry_rcu(sn_irq_info, sn_irq_lh[irq], list)
+		sn_call_force_intr_provider(sn_irq_info);
+
 	rcu_read_unlock();
 }
 
@@ -351,6 +356,14 @@
 	struct pcidev_info *pcidev_info;
 	struct pcibus_info *pcibus_info;
 
+	/*
+	 * Bridge types attached to TIO (anything but PIC) do not need this WAR
+	 * since they do not target Shub II interrupt registers.  If that
+	 * ever changes, this check needs to accomodate.
+	 */
+	if (sn_irq_info->irq_bridge_type != PCIIO_ASIC_TYPE_PIC)
+		return;
+
 	pcidev_info = (struct pcidev_info *)sn_irq_info->irq_pciioinfo;
 	if (!pcidev_info)
 		return;
@@ -377,16 +390,12 @@
 		break;
 	}
 	if (!test_bit(irr_bit, &irr_reg)) {
-		if (!test_bit(irq, pda->sn_soft_irr)) {
-			if (!test_bit(irq, pda->sn_in_service_ivecs)) {
-				regval &= 0xff;
-				if (sn_irq_info->irq_int_bit & regval &
-				    sn_irq_info->irq_last_intr) {
-					regval &=
-					    ~(sn_irq_info->
-					      irq_int_bit & regval);
-					pcibr_force_interrupt(sn_irq_info);
-				}
+		if (!test_bit(irq, pda->sn_in_service_ivecs)) {
+			regval &= 0xff;
+			if (sn_irq_info->irq_int_bit & regval &
+			    sn_irq_info->irq_last_intr) {
+				regval &= ~(sn_irq_info->irq_int_bit & regval);
+				sn_call_force_intr_provider(sn_irq_info);
 			}
 		}
 	}
@@ -404,13 +413,7 @@
 	rcu_read_lock();
 	for (i = pda->sn_first_irq; i <= pda->sn_last_irq; i++) {
 		list_for_each_entry_rcu(sn_irq_info, sn_irq_lh[i], list) {
-			/*
-			 * Only call for PCI bridges that are fully
-			 * initialized.
-			 */
-			if (IS_PCI_BRIDGE_ASIC(sn_irq_info->irq_bridge_type) &&
-			    (sn_irq_info->irq_bridge != NULL))
-				sn_check_intr(i, sn_irq_info);
+			sn_check_intr(i, sn_irq_info);
 		}
 	}
 	rcu_read_unlock();
diff --git a/arch/ia64/sn/kernel/setup.c b/arch/ia64/sn/kernel/setup.c
index 7c7fe44..a594aca 100644
--- a/arch/ia64/sn/kernel/setup.c
+++ b/arch/ia64/sn/kernel/setup.c
@@ -80,8 +80,6 @@
 DEFINE_PER_CPU(struct nodepda_s *, __sn_nodepda);
 EXPORT_PER_CPU_SYMBOL(__sn_nodepda);
 
-partid_t sn_partid = -1;
-EXPORT_SYMBOL(sn_partid);
 char sn_system_serial_number_string[128];
 EXPORT_SYMBOL(sn_system_serial_number_string);
 u64 sn_partition_serial_number;
@@ -403,6 +401,7 @@
 		memset(nodepdaindr[cnode], 0, sizeof(nodepda_t));
 		memset(nodepdaindr[cnode]->phys_cpuid, -1,
 		    sizeof(nodepdaindr[cnode]->phys_cpuid));
+		spin_lock_init(&nodepdaindr[cnode]->ptc_lock);
 	}
 
 	/*
@@ -532,8 +531,8 @@
 	 */
 	{
 		u64 pio1[] = {SH1_PIO_WRITE_STATUS_0, 0, SH1_PIO_WRITE_STATUS_1, 0};
-		u64 pio2[] = {SH2_PIO_WRITE_STATUS_0, SH2_PIO_WRITE_STATUS_1,
-			SH2_PIO_WRITE_STATUS_2, SH2_PIO_WRITE_STATUS_3};
+		u64 pio2[] = {SH2_PIO_WRITE_STATUS_0, SH2_PIO_WRITE_STATUS_2,
+			SH2_PIO_WRITE_STATUS_1, SH2_PIO_WRITE_STATUS_3};
 		u64 *pio;
 		pio = is_shub1() ? pio1 : pio2;
 		pda->pio_write_status_addr = (volatile unsigned long *) LOCAL_MMR_ADDR(pio[slice]);
diff --git a/arch/ia64/sn/kernel/sn2/ptc_deadlock.S b/arch/ia64/sn/kernel/sn2/ptc_deadlock.S
index 96cb71d..3fa9506 100644
--- a/arch/ia64/sn/kernel/sn2/ptc_deadlock.S
+++ b/arch/ia64/sn/kernel/sn2/ptc_deadlock.S
@@ -3,7 +3,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (C) 2000-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 2000-2005 Silicon Graphics, Inc. All rights reserved.
  */
 
 #include <asm/types.h>
@@ -11,7 +11,7 @@
 
 #define DEADLOCKBIT	SH_PIO_WRITE_STATUS_WRITE_DEADLOCK_SHFT
 #define WRITECOUNTMASK	SH_PIO_WRITE_STATUS_PENDING_WRITE_COUNT_MASK
-#define ALIAS_OFFSET	(SH1_PIO_WRITE_STATUS_0_ALIAS-SH1_PIO_WRITE_STATUS_0)
+#define ALIAS_OFFSET	8
 
 
 	.global	sn2_ptc_deadlock_recovery_core
@@ -36,13 +36,15 @@
 	extr.u	piowcphy=piowc,0,61;;	// Convert piowc to uncached physical address
 	dep	piowcphy=-1,piowcphy,63,1
 	movl	mask=WRITECOUNTMASK
+	mov	r8=r0
 
 1:
 	add	scr2=ALIAS_OFFSET,piowc	// Address of WRITE_STATUS alias register 
-	mov	scr1=7;;		// Clear DEADLOCK, WRITE_ERROR, MULTI_WRITE_ERROR
-	st8.rel	[scr2]=scr1;;
+	;;
+	ld8.acq	scr1=[scr2];;
 
 5:	ld8.acq	scr1=[piowc];;		// Wait for PIOs to complete.
+	hint	@pause
 	and	scr2=scr1,mask;;	// mask of writecount bits
 	cmp.ne	p6,p0=zeroval,scr2
 (p6)	br.cond.sptk 5b
@@ -57,6 +59,7 @@
 	st8.rel [ptc0]=data0		// Write PTC0 & wait for completion.
 
 5:	ld8.acq	scr1=[piowcphy];;	// Wait for PIOs to complete.
+	hint	@pause
 	and	scr2=scr1,mask;;	// mask of writecount bits
 	cmp.ne	p6,p0=zeroval,scr2
 (p6)	br.cond.sptk 5b;;
@@ -67,6 +70,7 @@
 (p7)	st8.rel [ptc1]=data1;;		// Now write PTC1.
 
 5:	ld8.acq	scr1=[piowcphy];;	// Wait for PIOs to complete.
+	hint	@pause
 	and	scr2=scr1,mask;;	// mask of writecount bits
 	cmp.ne	p6,p0=zeroval,scr2
 (p6)	br.cond.sptk 5b
@@ -77,6 +81,7 @@
 	srlz.i;;
 	////////////// END   PHYSICAL MODE ////////////////////
 
+(p8)	add	r8=1,r8
 (p8)	br.cond.spnt 1b;;		// Repeat if DEADLOCK occurred.
 
 	br.ret.sptk	rp
diff --git a/arch/ia64/sn/kernel/sn2/sn2_smp.c b/arch/ia64/sn/kernel/sn2/sn2_smp.c
index 7af05a7..0a4ee50 100644
--- a/arch/ia64/sn/kernel/sn2/sn2_smp.c
+++ b/arch/ia64/sn/kernel/sn2/sn2_smp.c
@@ -5,7 +5,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (C) 2000-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 2000-2005 Silicon Graphics, Inc. All rights reserved.
  */
 
 #include <linux/init.h>
@@ -20,6 +20,8 @@
 #include <linux/module.h>
 #include <linux/bitops.h>
 #include <linux/nodemask.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
 
 #include <asm/processor.h>
 #include <asm/irq.h>
@@ -39,12 +41,120 @@
 #include <asm/sn/nodepda.h>
 #include <asm/sn/rw_mmr.h>
 
-void sn2_ptc_deadlock_recovery(volatile unsigned long *, unsigned long data0, 
-	volatile unsigned long *, unsigned long data1);
+DEFINE_PER_CPU(struct ptc_stats, ptcstats);
+DECLARE_PER_CPU(struct ptc_stats, ptcstats);
 
 static  __cacheline_aligned DEFINE_SPINLOCK(sn2_global_ptc_lock);
 
-static unsigned long sn2_ptc_deadlock_count;
+void sn2_ptc_deadlock_recovery(short *, short, int, volatile unsigned long *, unsigned long data0,
+	volatile unsigned long *, unsigned long data1);
+
+#ifdef DEBUG_PTC
+/*
+ * ptctest:
+ *
+ * 	xyz - 3 digit hex number:
+ * 		x - Force PTC purges to use shub:
+ * 			0 - no force
+ * 			1 - force
+ * 		y - interupt enable
+ * 			0 - disable interrupts
+ * 			1 - leave interuupts enabled
+ * 		z - type of lock:
+ * 			0 - global lock
+ * 			1 - node local lock
+ * 			2 - no lock
+ *
+ *   	Note: on shub1, only ptctest == 0 is supported. Don't try other values!
+ */
+
+static unsigned int sn2_ptctest = 0;
+
+static int __init ptc_test(char *str)
+{
+	get_option(&str, &sn2_ptctest);
+	return 1;
+}
+__setup("ptctest=", ptc_test);
+
+static inline int ptc_lock(unsigned long *flagp)
+{
+	unsigned long opt = sn2_ptctest & 255;
+
+	switch (opt) {
+	case 0x00:
+		spin_lock_irqsave(&sn2_global_ptc_lock, *flagp);
+		break;
+	case 0x01:
+		spin_lock_irqsave(&sn_nodepda->ptc_lock, *flagp);
+		break;
+	case 0x02:
+		local_irq_save(*flagp);
+		break;
+	case 0x10:
+		spin_lock(&sn2_global_ptc_lock);
+		break;
+	case 0x11:
+		spin_lock(&sn_nodepda->ptc_lock);
+		break;
+	case 0x12:
+		break;
+	default:
+		BUG();
+	}
+	return opt;
+}
+
+static inline void ptc_unlock(unsigned long flags, int opt)
+{
+	switch (opt) {
+	case 0x00:
+		spin_unlock_irqrestore(&sn2_global_ptc_lock, flags);
+		break;
+	case 0x01:
+		spin_unlock_irqrestore(&sn_nodepda->ptc_lock, flags);
+		break;
+	case 0x02:
+		local_irq_restore(flags);
+		break;
+	case 0x10:
+		spin_unlock(&sn2_global_ptc_lock);
+		break;
+	case 0x11:
+		spin_unlock(&sn_nodepda->ptc_lock);
+		break;
+	case 0x12:
+		break;
+	default:
+		BUG();
+	}
+}
+#else
+
+#define sn2_ptctest	0
+
+static inline int ptc_lock(unsigned long *flagp)
+{
+	spin_lock_irqsave(&sn2_global_ptc_lock, *flagp);
+	return 0;
+}
+
+static inline void ptc_unlock(unsigned long flags, int opt)
+{
+	spin_unlock_irqrestore(&sn2_global_ptc_lock, flags);
+}
+#endif
+
+struct ptc_stats {
+	unsigned long ptc_l;
+	unsigned long change_rid;
+	unsigned long shub_ptc_flushes;
+	unsigned long nodes_flushed;
+	unsigned long deadlocks;
+	unsigned long lock_itc_clocks;
+	unsigned long shub_itc_clocks;
+	unsigned long shub_itc_clocks_max;
+};
 
 static inline unsigned long wait_piowc(void)
 {
@@ -89,9 +199,9 @@
 sn2_global_tlb_purge(unsigned long start, unsigned long end,
 		     unsigned long nbits)
 {
-	int i, shub1, cnode, mynasid, cpu, lcpu = 0, nasid, flushed = 0;
+	int i, opt, shub1, cnode, mynasid, cpu, lcpu = 0, nasid, flushed = 0;
 	volatile unsigned long *ptc0, *ptc1;
-	unsigned long flags = 0, data0 = 0, data1 = 0;
+	unsigned long itc, itc2, flags, data0 = 0, data1 = 0;
 	struct mm_struct *mm = current->active_mm;
 	short nasids[MAX_NUMNODES], nix;
 	nodemask_t nodes_flushed;
@@ -114,16 +224,19 @@
 			start += (1UL << nbits);
 		} while (start < end);
 		ia64_srlz_i();
+		__get_cpu_var(ptcstats).ptc_l++;
 		preempt_enable();
 		return;
 	}
 
 	if (atomic_read(&mm->mm_users) == 1) {
 		flush_tlb_mm(mm);
+		__get_cpu_var(ptcstats).change_rid++;
 		preempt_enable();
 		return;
 	}
 
+	itc = ia64_get_itc();
 	nix = 0;
 	for_each_node_mask(cnode, nodes_flushed)
 		nasids[nix++] = cnodeid_to_nasid(cnode);
@@ -148,7 +261,12 @@
 
 	mynasid = get_nasid();
 
-	spin_lock_irqsave(&sn2_global_ptc_lock, flags);
+	itc = ia64_get_itc();
+	opt = ptc_lock(&flags);
+	itc2 = ia64_get_itc();
+	__get_cpu_var(ptcstats).lock_itc_clocks += itc2 - itc;
+	__get_cpu_var(ptcstats).shub_ptc_flushes++;
+	__get_cpu_var(ptcstats).nodes_flushed += nix;
 
 	do {
 		if (shub1)
@@ -157,7 +275,7 @@
 			data0 = (data0 & ~SH2_PTC_ADDR_MASK) | (start & SH2_PTC_ADDR_MASK);
 		for (i = 0; i < nix; i++) {
 			nasid = nasids[i];
-			if (unlikely(nasid == mynasid)) {
+			if ((!(sn2_ptctest & 3)) && unlikely(nasid == mynasid)) {
 				ia64_ptcga(start, nbits << 2);
 				ia64_srlz_i();
 			} else {
@@ -169,18 +287,22 @@
 				flushed = 1;
 			}
 		}
-
 		if (flushed
 		    && (wait_piowc() &
-			SH_PIO_WRITE_STATUS_WRITE_DEADLOCK_MASK)) {
-			sn2_ptc_deadlock_recovery(ptc0, data0, ptc1, data1);
+				(SH_PIO_WRITE_STATUS_WRITE_DEADLOCK_MASK))) {
+			sn2_ptc_deadlock_recovery(nasids, nix, mynasid, ptc0, data0, ptc1, data1);
 		}
 
 		start += (1UL << nbits);
 
 	} while (start < end);
 
-	spin_unlock_irqrestore(&sn2_global_ptc_lock, flags);
+	itc2 = ia64_get_itc() - itc2;
+	__get_cpu_var(ptcstats).shub_itc_clocks += itc2;
+	if (itc2 > __get_cpu_var(ptcstats).shub_itc_clocks_max)
+		__get_cpu_var(ptcstats).shub_itc_clocks_max = itc2;
+
+	ptc_unlock(flags, opt);
 
 	preempt_enable();
 }
@@ -192,31 +314,29 @@
  * TLB flush transaction.  The recovery sequence is somewhat tricky & is
  * coded in assembly language.
  */
-void sn2_ptc_deadlock_recovery(volatile unsigned long *ptc0, unsigned long data0,
+void sn2_ptc_deadlock_recovery(short *nasids, short nix, int mynasid, volatile unsigned long *ptc0, unsigned long data0,
 	volatile unsigned long *ptc1, unsigned long data1)
 {
 	extern void sn2_ptc_deadlock_recovery_core(volatile unsigned long *, unsigned long,
 	        volatile unsigned long *, unsigned long, volatile unsigned long *, unsigned long);
-	int cnode, mycnode, nasid;
-	volatile unsigned long *piows;
-	volatile unsigned long zeroval;
+	short nasid, i;
+	unsigned long *piows, zeroval;
 
-	sn2_ptc_deadlock_count++;
+	__get_cpu_var(ptcstats).deadlocks++;
 
-	piows = pda->pio_write_status_addr;
+	piows = (unsigned long *) pda->pio_write_status_addr;
 	zeroval = pda->pio_write_status_val;
 
-	mycnode = numa_node_id();
-
-	for_each_online_node(cnode) {
-		if (is_headless_node(cnode) || cnode == mycnode)
+	for (i=0; i < nix; i++) {
+		nasid = nasids[i];
+		if (!(sn2_ptctest & 3) && nasid == mynasid)
 			continue;
-		nasid = cnodeid_to_nasid(cnode);
 		ptc0 = CHANGE_NASID(nasid, ptc0);
 		if (ptc1)
 			ptc1 = CHANGE_NASID(nasid, ptc1);
 		sn2_ptc_deadlock_recovery_core(ptc0, data0, ptc1, data1, piows, zeroval);
 	}
+
 }
 
 /**
@@ -293,3 +413,93 @@
 
 	sn_send_IPI_phys(nasid, physid, vector, delivery_mode);
 }
+
+#ifdef CONFIG_PROC_FS
+
+#define PTC_BASENAME	"sgi_sn/ptc_statistics"
+
+static void *sn2_ptc_seq_start(struct seq_file *file, loff_t * offset)
+{
+	if (*offset < NR_CPUS)
+		return offset;
+	return NULL;
+}
+
+static void *sn2_ptc_seq_next(struct seq_file *file, void *data, loff_t * offset)
+{
+	(*offset)++;
+	if (*offset < NR_CPUS)
+		return offset;
+	return NULL;
+}
+
+static void sn2_ptc_seq_stop(struct seq_file *file, void *data)
+{
+}
+
+static int sn2_ptc_seq_show(struct seq_file *file, void *data)
+{
+	struct ptc_stats *stat;
+	int cpu;
+
+	cpu = *(loff_t *) data;
+
+	if (!cpu) {
+		seq_printf(file, "# ptc_l change_rid shub_ptc_flushes shub_nodes_flushed deadlocks lock_nsec shub_nsec shub_nsec_max\n");
+		seq_printf(file, "# ptctest %d\n", sn2_ptctest);
+	}
+
+	if (cpu < NR_CPUS && cpu_online(cpu)) {
+		stat = &per_cpu(ptcstats, cpu);
+		seq_printf(file, "cpu %d %ld %ld %ld %ld %ld %ld %ld %ld\n", cpu, stat->ptc_l,
+				stat->change_rid, stat->shub_ptc_flushes, stat->nodes_flushed,
+				stat->deadlocks,
+				1000 * stat->lock_itc_clocks / per_cpu(cpu_info, cpu).cyc_per_usec,
+				1000 * stat->shub_itc_clocks / per_cpu(cpu_info, cpu).cyc_per_usec,
+				1000 * stat->shub_itc_clocks_max / per_cpu(cpu_info, cpu).cyc_per_usec);
+	}
+
+	return 0;
+}
+
+static struct seq_operations sn2_ptc_seq_ops = {
+	.start = sn2_ptc_seq_start,
+	.next = sn2_ptc_seq_next,
+	.stop = sn2_ptc_seq_stop,
+	.show = sn2_ptc_seq_show
+};
+
+int sn2_ptc_proc_open(struct inode *inode, struct file *file)
+{
+	return seq_open(file, &sn2_ptc_seq_ops);
+}
+
+static struct file_operations proc_sn2_ptc_operations = {
+	.open = sn2_ptc_proc_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = seq_release,
+};
+
+static struct proc_dir_entry *proc_sn2_ptc;
+
+static int __init sn2_ptc_init(void)
+{
+	if (!(proc_sn2_ptc = create_proc_entry(PTC_BASENAME, 0444, NULL))) {
+		printk(KERN_ERR "unable to create %s proc entry", PTC_BASENAME);
+		return -EINVAL;
+	}
+	proc_sn2_ptc->proc_fops = &proc_sn2_ptc_operations;
+	spin_lock_init(&sn2_global_ptc_lock);
+	return 0;
+}
+
+static void __exit sn2_ptc_exit(void)
+{
+	remove_proc_entry(PTC_BASENAME, NULL);
+}
+
+module_init(sn2_ptc_init);
+module_exit(sn2_ptc_exit);
+#endif /* CONFIG_PROC_FS */
+
diff --git a/arch/ia64/sn/kernel/sn2/sn_hwperf.c b/arch/ia64/sn/kernel/sn2/sn_hwperf.c
index 833e700..0513aac 100644
--- a/arch/ia64/sn/kernel/sn2/sn_hwperf.c
+++ b/arch/ia64/sn/kernel/sn2/sn_hwperf.c
@@ -36,7 +36,6 @@
 #include <asm/topology.h>
 #include <asm/smp.h>
 #include <asm/semaphore.h>
-#include <asm/segment.h>
 #include <asm/uaccess.h>
 #include <asm/sal.h>
 #include <asm/sn/io.h>
@@ -59,7 +58,7 @@
 	struct sn_hwperf_object_info *objbuf = NULL;
 
 	if ((e = sn_hwperf_init()) < 0) {
-		printk("sn_hwperf_init failed: err %d\n", e);
+		printk(KERN_ERR "sn_hwperf_init failed: err %d\n", e);
 		goto out;
 	}
 
@@ -111,7 +110,7 @@
 	if (sn_hwperf_location_to_bpos(location, &rack, &bay, &slot, &slab))
 		return -1;
 
-	for (cnode = 0; cnode < numionodes; cnode++) {
+	for_each_node(cnode) {
 		geoid = cnodeid_get_geoid(cnode);
 		module_id = geo_module(geoid);
 		this_rack = MODULE_GET_RACK(module_id);
@@ -124,11 +123,13 @@
 		}
 	}
 
-	return cnode < numionodes ? cnode : -1;
+	return node_possible(cnode) ? cnode : -1;
 }
 
 static int sn_hwperf_obj_to_cnode(struct sn_hwperf_object_info * obj)
 {
+	if (!SN_HWPERF_IS_NODE(obj) && !SN_HWPERF_IS_IONODE(obj))
+		BUG();
 	if (!obj->sn_hwp_this_part)
 		return -1;
 	return sn_hwperf_geoid_to_cnode(obj->location);
@@ -174,31 +175,199 @@
 	return slabname;
 }
 
-static void print_pci_topology(struct seq_file *s,
-	struct sn_hwperf_object_info *obj, int *ordinal,
-	u64 rack, u64 bay, u64 slot, u64 slab)
+static void print_pci_topology(struct seq_file *s)
 {
-	char *p1;
-	char *p2;
-	char *pg;
+	char *p;
+	size_t sz;
+	int e;
 
-	if (!(pg = (char *)get_zeroed_page(GFP_KERNEL)))
-		return; /* ignore */
-	if (ia64_sn_ioif_get_pci_topology(rack, bay, slot, slab,
-		__pa(pg), PAGE_SIZE) == SN_HWPERF_OP_OK) {
-		for (p1=pg; *p1 && p1 < pg + PAGE_SIZE;) {
-			if (!(p2 = strchr(p1, '\n')))
-				break;
-			*p2 = '\0';
-			seq_printf(s, "pcibus %d %s-%s\n",
-				*ordinal, obj->location, p1);
-			(*ordinal)++;
-			p1 = p2 + 1;
+	for (sz = PAGE_SIZE; sz < 16 * PAGE_SIZE; sz += PAGE_SIZE) {
+		if (!(p = (char *)kmalloc(sz, GFP_KERNEL)))
+			break;
+		e = ia64_sn_ioif_get_pci_topology(__pa(p), sz);
+		if (e == SALRET_OK)
+			seq_puts(s, p);
+		kfree(p);
+		if (e == SALRET_OK || e == SALRET_NOT_IMPLEMENTED)
+			break;
+	}
+}
+
+static inline int sn_hwperf_has_cpus(cnodeid_t node)
+{
+	return node_online(node) && nr_cpus_node(node);
+}
+
+static inline int sn_hwperf_has_mem(cnodeid_t node)
+{
+	return node_online(node) && NODE_DATA(node)->node_present_pages;
+}
+
+static struct sn_hwperf_object_info *
+sn_hwperf_findobj_id(struct sn_hwperf_object_info *objbuf,
+	int nobj, int id)
+{
+	int i;
+	struct sn_hwperf_object_info *p = objbuf;
+
+	for (i=0; i < nobj; i++, p++) {
+		if (p->id == id)
+			return p;
+	}
+
+	return NULL;
+
+}
+
+static int sn_hwperf_get_nearest_node_objdata(struct sn_hwperf_object_info *objbuf,
+	int nobj, cnodeid_t node, cnodeid_t *near_mem_node, cnodeid_t *near_cpu_node)
+{
+	int e;
+	struct sn_hwperf_object_info *nodeobj = NULL;
+	struct sn_hwperf_object_info *op;
+	struct sn_hwperf_object_info *dest;
+	struct sn_hwperf_object_info *router;
+	struct sn_hwperf_port_info ptdata[16];
+	int sz, i, j;
+	cnodeid_t c;
+	int found_mem = 0;
+	int found_cpu = 0;
+
+	if (!node_possible(node))
+		return -EINVAL;
+
+	if (sn_hwperf_has_cpus(node)) {
+		if (near_cpu_node)
+			*near_cpu_node = node;
+		found_cpu++;
+	}
+
+	if (sn_hwperf_has_mem(node)) {
+		if (near_mem_node)
+			*near_mem_node = node;
+		found_mem++;
+	}
+
+	if (found_cpu && found_mem)
+		return 0; /* trivially successful */
+
+	/* find the argument node object */
+	for (i=0, op=objbuf; i < nobj; i++, op++) {
+		if (!SN_HWPERF_IS_NODE(op) && !SN_HWPERF_IS_IONODE(op))
+			continue;
+		if (node == sn_hwperf_obj_to_cnode(op)) {
+			nodeobj = op;
+			break;
 		}
 	}
-	free_page((unsigned long)pg);
+	if (!nodeobj) {
+		e = -ENOENT;
+		goto err;
+	}
+
+	/* get it's interconnect topology */
+	sz = op->ports * sizeof(struct sn_hwperf_port_info);
+	if (sz > sizeof(ptdata))
+		BUG();
+	e = ia64_sn_hwperf_op(sn_hwperf_master_nasid,
+			      SN_HWPERF_ENUM_PORTS, nodeobj->id, sz,
+			      (u64)&ptdata, 0, 0, NULL);
+	if (e != SN_HWPERF_OP_OK) {
+		e = -EINVAL;
+		goto err;
+	}
+
+	/* find nearest node with cpus and nearest memory */
+	for (router=NULL, j=0; j < op->ports; j++) {
+		dest = sn_hwperf_findobj_id(objbuf, nobj, ptdata[j].conn_id);
+		if (!dest || SN_HWPERF_FOREIGN(dest) ||
+		    !SN_HWPERF_IS_NODE(dest) || SN_HWPERF_IS_IONODE(dest)) {
+			continue;
+		}
+		c = sn_hwperf_obj_to_cnode(dest);
+		if (!found_cpu && sn_hwperf_has_cpus(c)) {
+			if (near_cpu_node)
+				*near_cpu_node = c;
+			found_cpu++;
+		}
+		if (!found_mem && sn_hwperf_has_mem(c)) {
+			if (near_mem_node)
+				*near_mem_node = c;
+			found_mem++;
+		}
+		if (SN_HWPERF_IS_ROUTER(dest))
+			router = dest;
+	}
+
+	if (router && (!found_cpu || !found_mem)) {
+		/* search for a node connected to the same router */
+		sz = router->ports * sizeof(struct sn_hwperf_port_info);
+		if (sz > sizeof(ptdata))
+			BUG();
+		e = ia64_sn_hwperf_op(sn_hwperf_master_nasid,
+				      SN_HWPERF_ENUM_PORTS, router->id, sz,
+				      (u64)&ptdata, 0, 0, NULL);
+		if (e != SN_HWPERF_OP_OK) {
+			e = -EINVAL;
+			goto err;
+		}
+		for (j=0; j < router->ports; j++) {
+			dest = sn_hwperf_findobj_id(objbuf, nobj,
+				ptdata[j].conn_id);
+			if (!dest || dest->id == node ||
+			    SN_HWPERF_FOREIGN(dest) ||
+			    !SN_HWPERF_IS_NODE(dest) ||
+			    SN_HWPERF_IS_IONODE(dest)) {
+				continue;
+			}
+			c = sn_hwperf_obj_to_cnode(dest);
+			if (!found_cpu && sn_hwperf_has_cpus(c)) {
+				if (near_cpu_node)
+					*near_cpu_node = c;
+				found_cpu++;
+			}
+			if (!found_mem && sn_hwperf_has_mem(c)) {
+				if (near_mem_node)
+					*near_mem_node = c;
+				found_mem++;
+			}
+			if (found_cpu && found_mem)
+				break;
+		}
+	}
+
+	if (!found_cpu || !found_mem) {
+		/* resort to _any_ node with CPUs and memory */
+		for (i=0, op=objbuf; i < nobj; i++, op++) {
+			if (SN_HWPERF_FOREIGN(op) ||
+			    SN_HWPERF_IS_IONODE(op) ||
+			    !SN_HWPERF_IS_NODE(op)) {
+				continue;
+			}
+			c = sn_hwperf_obj_to_cnode(op);
+			if (!found_cpu && sn_hwperf_has_cpus(c)) {
+				if (near_cpu_node)
+					*near_cpu_node = c;
+				found_cpu++;
+			}
+			if (!found_mem && sn_hwperf_has_mem(c)) {
+				if (near_mem_node)
+					*near_mem_node = c;
+				found_mem++;
+			}
+			if (found_cpu && found_mem)
+				break;
+		}
+	}
+
+	if (!found_cpu || !found_mem)
+		e = -ENODATA;
+
+err:
+	return e;
 }
 
+
 static int sn_topology_show(struct seq_file *s, void *d)
 {
 	int sz;
@@ -215,7 +384,6 @@
 	struct sn_hwperf_object_info *p;
 	struct sn_hwperf_object_info *obj = d;	/* this object */
 	struct sn_hwperf_object_info *objs = s->private; /* all objects */
-	int rack, bay, slot, slab;
 	u8 shubtype;
 	u8 system_size;
 	u8 sharing_size;
@@ -225,7 +393,6 @@
 	u8 region_size;
 	u16 nasid_mask;
 	int nasid_msb;
-	int pci_bus_ordinal = 0;
 
 	if (obj == objs) {
 		seq_printf(s, "# sn_topology version 2\n");
@@ -253,6 +420,8 @@
 			shubtype ? "shub2" : "shub1", 
 			(u64)nasid_mask << nasid_shift, nasid_msb, nasid_shift,
 			system_size, sharing_size, coher, region_size);
+
+		print_pci_topology(s);
 	}
 
 	if (SN_HWPERF_FOREIGN(obj)) {
@@ -272,11 +441,24 @@
 	if (!SN_HWPERF_IS_NODE(obj) && !SN_HWPERF_IS_IONODE(obj))
 		seq_putc(s, '\n');
 	else {
+		cnodeid_t near_mem = -1;
+		cnodeid_t near_cpu = -1;
+
 		seq_printf(s, ", nasid 0x%x", cnodeid_to_nasid(ordinal));
-		for (i=0; i < numionodes; i++) {
-			seq_printf(s, i ? ":%d" : ", dist %d",
-				node_distance(ordinal, i));
+
+		if (sn_hwperf_get_nearest_node_objdata(objs, sn_hwperf_obj_cnt,
+			ordinal, &near_mem, &near_cpu) == 0) {
+			seq_printf(s, ", near_mem_nodeid %d, near_cpu_nodeid %d",
+				near_mem, near_cpu);
 		}
+
+		if (!SN_HWPERF_IS_IONODE(obj)) {
+			for_each_online_node(i) {
+				seq_printf(s, i ? ":%d" : ", dist %d",
+					node_distance(ordinal, i));
+			}
+		}
+
 		seq_putc(s, '\n');
 
 		/*
@@ -300,17 +482,6 @@
 				seq_putc(s, '\n');
 			}
 		}
-
-		/*
-		 * PCI busses attached to this node, if any
-		 */
-		if (sn_hwperf_location_to_bpos(obj->location,
-			&rack, &bay, &slot, &slab)) {
-			/* export pci bus info */
-			print_pci_topology(s, obj, &pci_bus_ordinal,
-				rack, bay, slot, slab);
-
-		}
 	}
 
 	if (obj->ports) {
@@ -572,6 +743,8 @@
 		if ((r = sn_hwperf_enum_objects(&nobj, &objs)) == 0) {
 			memset(p, 0, a.sz);
 			for (i = 0; i < nobj; i++) {
+				if (!SN_HWPERF_IS_NODE(objs + i))
+					continue;
 				node = sn_hwperf_obj_to_cnode(objs + i);
 				for_each_online_cpu(j) {
 					if (node != cpu_to_node(j))
@@ -598,7 +771,7 @@
 
 	case SN_HWPERF_GET_NODE_NASID:
 		if (a.sz != sizeof(u64) ||
-		   (node = a.arg) < 0 || node >= numionodes) {
+		   (node = a.arg) < 0 || !node_possible(node)) {
 			r = -EINVAL;
 			goto error;
 		}
@@ -627,6 +800,14 @@
 				vfree(objs);
 				goto error;
 			}
+
+			if (!SN_HWPERF_IS_NODE(objs + i) &&
+			    !SN_HWPERF_IS_IONODE(objs + i)) {
+			    	r = -ENOENT;
+				vfree(objs);
+				goto error;
+			}
+
 			*(u64 *)p = (u64)sn_hwperf_obj_to_cnode(objs + i);
 			vfree(objs);
 		}
@@ -692,6 +873,7 @@
 
 	/* single threaded, once-only initialization */
 	down(&sn_hwperf_init_mutex);
+
 	if (sn_hwperf_salheap) {
 		up(&sn_hwperf_init_mutex);
 		return e;
@@ -742,19 +924,6 @@
 		sn_hwperf_salheap = NULL;
 		sn_hwperf_obj_cnt = 0;
 	}
-
-	if (!e) {
-		/*
-		 * Register a dynamic misc device for ioctl. Platforms
-		 * supporting hotplug will create /dev/sn_hwperf, else
-		 * user can to look up the minor number in /proc/misc.
-		 */
-		if ((e = misc_register(&sn_hwperf_dev)) != 0) {
-			printk(KERN_ERR "sn_hwperf_init: misc register "
-			       "for \"sn_hwperf\" failed, err %d\n", e);
-		}
-	}
-
 	up(&sn_hwperf_init_mutex);
 	return e;
 }
@@ -782,3 +951,41 @@
 	vfree(seq->private);
 	return seq_release(inode, file);
 }
+
+int sn_hwperf_get_nearest_node(cnodeid_t node,
+	cnodeid_t *near_mem_node, cnodeid_t *near_cpu_node)
+{
+	int e;
+	int nobj;
+	struct sn_hwperf_object_info *objbuf;
+
+	if ((e = sn_hwperf_enum_objects(&nobj, &objbuf)) == 0) {
+		e = sn_hwperf_get_nearest_node_objdata(objbuf, nobj,
+			node, near_mem_node, near_cpu_node);
+		vfree(objbuf);
+	}
+
+	return e;
+}
+
+static int __devinit sn_hwperf_misc_register_init(void)
+{
+	int e;
+
+	sn_hwperf_init();
+
+	/*
+	 * Register a dynamic misc device for hwperf ioctls. Platforms
+	 * supporting hotplug will create /dev/sn_hwperf, else user
+	 * can to look up the minor number in /proc/misc.
+	 */
+	if ((e = misc_register(&sn_hwperf_dev)) != 0) {
+		printk(KERN_ERR "sn_hwperf_misc_register_init: failed to "
+		"register misc device for \"%s\"\n", sn_hwperf_dev.name);
+	}
+
+	return e;
+}
+
+device_initcall(sn_hwperf_misc_register_init); /* after misc_init() */
+EXPORT_SYMBOL(sn_hwperf_get_nearest_node);
diff --git a/arch/ia64/sn/kernel/sn2/sn_proc_fs.c b/arch/ia64/sn/kernel/sn2/sn_proc_fs.c
index 6a80fca..51bf827 100644
--- a/arch/ia64/sn/kernel/sn2/sn_proc_fs.c
+++ b/arch/ia64/sn/kernel/sn2/sn_proc_fs.c
@@ -3,7 +3,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (C) 2000-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 2000-2005 Silicon Graphics, Inc. All rights reserved.
  */
 #include <linux/config.h>
 #include <asm/uaccess.h>
@@ -15,7 +15,7 @@
 
 static int partition_id_show(struct seq_file *s, void *p)
 {
-	seq_printf(s, "%d\n", sn_local_partid());
+	seq_printf(s, "%d\n", sn_partition_id);
 	return 0;
 }
 
diff --git a/arch/ia64/sn/kernel/sn2/timer_interrupt.c b/arch/ia64/sn/kernel/sn2/timer_interrupt.c
index cde7375..adf5db2 100644
--- a/arch/ia64/sn/kernel/sn2/timer_interrupt.c
+++ b/arch/ia64/sn/kernel/sn2/timer_interrupt.c
@@ -1,7 +1,7 @@
 /*
  *
  *
- * Copyright (c) 2003 Silicon Graphics, Inc.  All Rights Reserved.
+ * Copyright (c) 2005 Silicon Graphics, Inc.  All Rights Reserved.
  * 
  * This program is free software; you can redistribute it and/or modify it 
  * under the terms of version 2 of the GNU General Public License 
@@ -50,14 +50,16 @@
 			     LED_CPU_HEARTBEAT, LED_CPU_HEARTBEAT);
 	}
 
-	if (enable_shub_wars_1_1()) {
-		/* Bugfix code for SHUB 1.1 */
-		if (pda->pio_shub_war_cam_addr)
-			*pda->pio_shub_war_cam_addr = 0x8000000000000010UL;
+	if (is_shub1()) {
+		if (enable_shub_wars_1_1()) {
+			/* Bugfix code for SHUB 1.1 */
+			if (pda->pio_shub_war_cam_addr)
+				*pda->pio_shub_war_cam_addr = 0x8000000000000010UL;
+		}
+		if (pda->sn_lb_int_war_ticks == 0)
+			sn_lb_int_war_check();
+		pda->sn_lb_int_war_ticks++;
+		if (pda->sn_lb_int_war_ticks >= SN_LB_INT_WAR_INTERVAL)
+			pda->sn_lb_int_war_ticks = 0;
 	}
-	if (pda->sn_lb_int_war_ticks == 0)
-		sn_lb_int_war_check();
-	pda->sn_lb_int_war_ticks++;
-	if (pda->sn_lb_int_war_ticks >= SN_LB_INT_WAR_INTERVAL)
-		pda->sn_lb_int_war_ticks = 0;
 }
diff --git a/arch/ia64/sn/pci/Makefile b/arch/ia64/sn/pci/Makefile
index 2f915bc..321576b 100644
--- a/arch/ia64/sn/pci/Makefile
+++ b/arch/ia64/sn/pci/Makefile
@@ -7,4 +7,4 @@
 #
 # Makefile for the sn pci general routines.
 
-obj-y := pci_dma.o tioca_provider.o pcibr/ 
+obj-y := pci_dma.o tioca_provider.o tioce_provider.o pcibr/
diff --git a/arch/ia64/sn/pci/pcibr/pcibr_dma.c b/arch/ia64/sn/pci/pcibr/pcibr_dma.c
index b058dc2..3409347 100644
--- a/arch/ia64/sn/pci/pcibr/pcibr_dma.c
+++ b/arch/ia64/sn/pci/pcibr/pcibr_dma.c
@@ -3,7 +3,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (C) 2001-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 2001-2005 Silicon Graphics, Inc. All rights reserved.
  */
 
 #include <linux/types.h>
@@ -215,8 +215,8 @@
 	int is_tio;
 	int wid_num;
 	int i, j;
-	int bwin;
 	uint64_t flags;
+	uint64_t itte;
 	struct hubdev_info *hubinfo;
 	volatile struct sn_flush_device_list *p;
 	struct sn_flush_nasid_entry *flush_nasid_list;
@@ -233,31 +233,36 @@
 	if (!hubinfo) {
 		BUG();
 	}
-	is_tio = (nasid & 1);
-	if (is_tio) {
-		wid_num = TIO_SWIN_WIDGETNUM(addr);
-		bwin = TIO_BWIN_WINDOWNUM(addr);
-	} else {
-		wid_num = SWIN_WIDGETNUM(addr);
-		bwin = BWIN_WINDOWNUM(addr);
-	}
 
 	flush_nasid_list = &hubinfo->hdi_flush_nasid_list;
 	if (flush_nasid_list->widget_p == NULL)
 		return;
-	if (bwin > 0) {
-		uint64_t itte = flush_nasid_list->iio_itte[bwin];
 
-		if (is_tio) {
-			wid_num = (itte >> TIO_ITTE_WIDGET_SHIFT) &
-			    TIO_ITTE_WIDGET_MASK;
-		} else {
-			wid_num = (itte >> IIO_ITTE_WIDGET_SHIFT) &
-			    IIO_ITTE_WIDGET_MASK;
-		}
+	is_tio = (nasid & 1);
+	if (is_tio) {
+		int itte_index;
+
+		if (TIO_HWIN(addr))
+			itte_index = 0;
+		else if (TIO_BWIN_WINDOWNUM(addr))
+			itte_index = TIO_BWIN_WINDOWNUM(addr);
+		else
+			itte_index = -1;
+
+		if (itte_index >= 0) {
+			itte = flush_nasid_list->iio_itte[itte_index];
+			if (! TIO_ITTE_VALID(itte))
+				return;
+			wid_num = TIO_ITTE_WIDGET(itte);
+		} else
+			wid_num = TIO_SWIN_WIDGETNUM(addr);
+	} else {
+		if (BWIN_WINDOWNUM(addr)) {
+			itte = flush_nasid_list->iio_itte[BWIN_WINDOWNUM(addr)];
+			wid_num = IIO_ITTE_WIDGET(itte);
+		} else
+			wid_num = SWIN_WIDGETNUM(addr);
 	}
-	if (flush_nasid_list->widget_p == NULL)
-		return;
 	if (flush_nasid_list->widget_p[wid_num] == NULL)
 		return;
 	p = &flush_nasid_list->widget_p[wid_num][0];
@@ -283,10 +288,16 @@
 	/*
 	 * For TIOCP use the Device(x) Write Request Buffer Flush Bridge
 	 * register since it ensures the data has entered the coherence
-	 * domain, unlike PIC
+	 * domain, unlike PIC.
 	 */
 	if (is_tio) {
-		uint32_t tio_id = REMOTE_HUB_L(nasid, TIO_NODE_ID);
+		/*
+	 	 * Note:  devices behind TIOCE should never be matched in the
+		 * above code, and so the following code is PIC/CP centric.
+		 * If CE ever needs the sn_dma_flush mechanism, we will have
+		 * to account for that here and in tioce_bus_fixup().
+	 	 */
+		uint32_t tio_id = HUB_L(TIO_IOSPACE_ADDR(nasid, TIO_NODE_ID));
 		uint32_t revnum = XWIDGET_PART_REV_NUM(tio_id);
 
 		/* TIOCP BRINGUP WAR (PV907516): Don't write buffer flush reg */
@@ -306,7 +317,8 @@
 		*(volatile uint32_t *)(p->sfdl_force_int_addr) = 1;
 
 		/* wait for the interrupt to come back. */
-		while (*(p->sfdl_flush_addr) != 0x10f) ;
+		while (*(p->sfdl_flush_addr) != 0x10f)
+			cpu_relax();
 
 		/* okay, everything is synched up. */
 		spin_unlock_irqrestore((spinlock_t *)&p->sfdl_flush_lock, flags);
diff --git a/arch/ia64/sn/pci/pcibr/pcibr_provider.c b/arch/ia64/sn/pci/pcibr/pcibr_provider.c
index b95e928..7b03b80 100644
--- a/arch/ia64/sn/pci/pcibr/pcibr_provider.c
+++ b/arch/ia64/sn/pci/pcibr/pcibr_provider.c
@@ -15,6 +15,7 @@
 #include <asm/sn/pcibus_provider_defs.h>
 #include <asm/sn/pcidev.h>
 #include <asm/sn/sn_sal.h>
+#include <asm/sn/sn2/sn_hwperf.h>
 #include "xtalk/xwidgetdev.h"
 #include "xtalk/hubdev.h"
 
@@ -60,7 +61,7 @@
 	ret_stuff.status = 0;
 	ret_stuff.v0 = 0;
 
-	segment = 0;
+	segment = soft->pbi_buscommon.bs_persist_segment;
 	busnum = soft->pbi_buscommon.bs_persist_busnum;
 	SAL_CALL_NOLOCK(ret_stuff,
 			(u64) SN_SAL_IOIF_ERROR_INTERRUPT,
@@ -88,6 +89,7 @@
 pcibr_bus_fixup(struct pcibus_bussoft *prom_bussoft, struct pci_controller *controller)
 {
 	int nasid, cnode, j;
+	cnodeid_t near_cnode;
 	struct hubdev_info *hubdev_info;
 	struct pcibus_info *soft;
 	struct sn_flush_device_list *sn_flush_device_list;
@@ -115,7 +117,7 @@
 	/*
 	 * register the bridge's error interrupt handler
 	 */
-	if (request_irq(SGI_PCIBR_ERROR, (void *)pcibr_error_intr_handler,
+	if (request_irq(SGI_PCIASIC_ERROR, (void *)pcibr_error_intr_handler,
 			SA_SHIRQ, "PCIBR error", (void *)(soft))) {
 		printk(KERN_WARNING
 		       "pcibr cannot allocate interrupt for error handler\n");
@@ -142,9 +144,12 @@
 			     j++, sn_flush_device_list++) {
 				if (sn_flush_device_list->sfdl_slot == -1)
 					continue;
-				if (sn_flush_device_list->
-				    sfdl_persistent_busnum ==
-				    soft->pbi_buscommon.bs_persist_busnum)
+				if ((sn_flush_device_list->
+				     sfdl_persistent_segment ==
+				     soft->pbi_buscommon.bs_persist_segment) &&
+				     (sn_flush_device_list->
+				     sfdl_persistent_busnum ==
+				     soft->pbi_buscommon.bs_persist_busnum))
 					sn_flush_device_list->sfdl_pcibus_info =
 					    soft;
 			}
@@ -158,12 +163,18 @@
 	memset(soft->pbi_int_ate_resource.ate, 0,
  	       (soft->pbi_int_ate_size * sizeof(uint64_t)));
 
-	if (prom_bussoft->bs_asic_type == PCIIO_ASIC_TYPE_TIOCP)
-		/*
-		 * TIO PCI Bridge with no closest node information.
-		 * FIXME: Find another way to determine the closest node
-		 */
-		controller->node = -1;
+	if (prom_bussoft->bs_asic_type == PCIIO_ASIC_TYPE_TIOCP) {
+		/* TIO PCI Bridge: find nearest node with CPUs */
+		int e = sn_hwperf_get_nearest_node(cnode, NULL, &near_cnode);
+
+		if (e < 0) {
+			near_cnode = (cnodeid_t)-1; /* use any node */
+			printk(KERN_WARNING "pcibr_bus_fixup: failed to find "
+				"near node with CPUs to TIO node %d, err=%d\n",
+				cnode, e);
+		}
+		controller->node = near_cnode;
+	}
 	else
 		controller->node = cnode;
 	return soft;
@@ -175,6 +186,9 @@
 	struct pcibus_info *pcibus_info;
 	int bit = sn_irq_info->irq_int_bit;
 
+	if (! sn_irq_info->irq_bridge)
+		return;
+
 	pcidev_info = (struct pcidev_info *)sn_irq_info->irq_pciioinfo;
 	if (pcidev_info) {
 		pcibus_info =
@@ -184,7 +198,7 @@
 	}
 }
 
-void pcibr_change_devices_irq(struct sn_irq_info *sn_irq_info)
+void pcibr_target_interrupt(struct sn_irq_info *sn_irq_info)
 {
 	struct pcidev_info *pcidev_info;
 	struct pcibus_info *pcibus_info;
@@ -219,6 +233,8 @@
 	.dma_map_consistent = pcibr_dma_map_consistent,
 	.dma_unmap = pcibr_dma_unmap,
 	.bus_fixup = pcibr_bus_fixup,
+	.force_interrupt = pcibr_force_interrupt,
+	.target_interrupt = pcibr_target_interrupt
 };
 
 int
diff --git a/arch/ia64/sn/pci/tioca_provider.c b/arch/ia64/sn/pci/tioca_provider.c
index 5d76a75..ea09c12f 100644
--- a/arch/ia64/sn/pci/tioca_provider.c
+++ b/arch/ia64/sn/pci/tioca_provider.c
@@ -559,7 +559,7 @@
 	ret_stuff.status = 0;
 	ret_stuff.v0 = 0;
 
-	segment = 0;
+	segment = soft->ca_common.bs_persist_segment;
 	busnum = soft->ca_common.bs_persist_busnum;
 
 	SAL_CALL_NOLOCK(ret_stuff,
@@ -622,7 +622,8 @@
 	    nasid_to_cnodeid(tioca_common->ca_closest_nasid);
 	tioca_common->ca_kernel_private = (uint64_t) tioca_kern;
 
-	bus = pci_find_bus(0, tioca_common->ca_common.bs_persist_busnum);
+	bus = pci_find_bus(tioca_common->ca_common.bs_persist_segment,
+		tioca_common->ca_common.bs_persist_busnum);
 	BUG_ON(!bus);
 	tioca_kern->ca_devices = &bus->devices;
 
@@ -656,6 +657,8 @@
 	.dma_map_consistent = tioca_dma_map,
 	.dma_unmap = tioca_dma_unmap,
 	.bus_fixup = tioca_bus_fixup,
+	.force_interrupt = NULL,
+	.target_interrupt = NULL
 };
 
 /**
diff --git a/arch/ia64/sn/pci/tioce_provider.c b/arch/ia64/sn/pci/tioce_provider.c
new file mode 100644
index 0000000..8e75db2
--- /dev/null
+++ b/arch/ia64/sn/pci/tioce_provider.c
@@ -0,0 +1,771 @@
+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2003-2005 Silicon Graphics, Inc.  All Rights Reserved.
+ */
+
+#include <linux/types.h>
+#include <linux/interrupt.h>
+#include <linux/pci.h>
+#include <asm/sn/sn_sal.h>
+#include <asm/sn/addrs.h>
+#include <asm/sn/pcidev.h>
+#include <asm/sn/pcibus_provider_defs.h>
+#include <asm/sn/tioce_provider.h>
+
+/**
+ * Bus address ranges for the 5 flavors of TIOCE DMA
+ */
+
+#define TIOCE_D64_MIN	0x8000000000000000UL
+#define TIOCE_D64_MAX	0xffffffffffffffffUL
+#define TIOCE_D64_ADDR(a)	((a) >= TIOCE_D64_MIN)
+
+#define TIOCE_D32_MIN	0x0000000080000000UL
+#define TIOCE_D32_MAX	0x00000000ffffffffUL
+#define TIOCE_D32_ADDR(a)	((a) >= TIOCE_D32_MIN && (a) <= TIOCE_D32_MAX)
+
+#define TIOCE_M32_MIN	0x0000000000000000UL
+#define TIOCE_M32_MAX	0x000000007fffffffUL
+#define TIOCE_M32_ADDR(a)	((a) >= TIOCE_M32_MIN && (a) <= TIOCE_M32_MAX)
+
+#define TIOCE_M40_MIN	0x0000004000000000UL
+#define TIOCE_M40_MAX	0x0000007fffffffffUL
+#define TIOCE_M40_ADDR(a)	((a) >= TIOCE_M40_MIN && (a) <= TIOCE_M40_MAX)
+
+#define TIOCE_M40S_MIN	0x0000008000000000UL
+#define TIOCE_M40S_MAX	0x000000ffffffffffUL
+#define TIOCE_M40S_ADDR(a)	((a) >= TIOCE_M40S_MIN && (a) <= TIOCE_M40S_MAX)
+
+/*
+ * ATE manipulation macros.
+ */
+
+#define ATE_PAGESHIFT(ps)	(__ffs(ps))
+#define ATE_PAGEMASK(ps)	((ps)-1)
+
+#define ATE_PAGE(x, ps) ((x) >> ATE_PAGESHIFT(ps))
+#define ATE_NPAGES(start, len, pagesize) \
+	(ATE_PAGE((start)+(len)-1, pagesize) - ATE_PAGE(start, pagesize) + 1)
+
+#define ATE_VALID(ate)	((ate) & (1UL << 63))
+#define ATE_MAKE(addr, ps) (((addr) & ~ATE_PAGEMASK(ps)) | (1UL << 63))
+
+/*
+ * Flavors of ate-based mapping supported by tioce_alloc_map()
+ */
+
+#define TIOCE_ATE_M32	1
+#define TIOCE_ATE_M40	2
+#define TIOCE_ATE_M40S	3
+
+#define KB(x)	((x) << 10)
+#define MB(x)	((x) << 20)
+#define GB(x)	((x) << 30)
+
+/**
+ * tioce_dma_d64 - create a DMA mapping using 64-bit direct mode
+ * @ct_addr: system coretalk address
+ *
+ * Map @ct_addr into 64-bit CE bus space.  No device context is necessary
+ * and no CE mapping are consumed.
+ *
+ * Bits 53:0 come from the coretalk address.  The remaining bits are set as
+ * follows:
+ *
+ * 63    - must be 1 to indicate d64 mode to CE hardware
+ * 62    - barrier bit ... controlled with tioce_dma_barrier()
+ * 61    - 0 since this is not an MSI transaction
+ * 60:54 - reserved, MBZ
+ */
+static uint64_t
+tioce_dma_d64(unsigned long ct_addr)
+{
+	uint64_t bus_addr;
+
+	bus_addr = ct_addr | (1UL << 63);
+
+	return bus_addr;
+}
+
+/**
+ * pcidev_to_tioce - return misc ce related pointers given a pci_dev
+ * @pci_dev: pci device context
+ * @base: ptr to store struct tioce_mmr * for the CE holding this device
+ * @kernel: ptr to store struct tioce_kernel * for the CE holding this device
+ * @port: ptr to store the CE port number that this device is on
+ *
+ * Return pointers to various CE-related structures for the CE upstream of
+ * @pci_dev.
+ */
+static inline void
+pcidev_to_tioce(struct pci_dev *pdev, struct tioce **base,
+		struct tioce_kernel **kernel, int *port)
+{
+	struct pcidev_info *pcidev_info;
+	struct tioce_common *ce_common;
+	struct tioce_kernel *ce_kernel;
+
+	pcidev_info = SN_PCIDEV_INFO(pdev);
+	ce_common = (struct tioce_common *)pcidev_info->pdi_pcibus_info;
+	ce_kernel = (struct tioce_kernel *)ce_common->ce_kernel_private;
+
+	if (base)
+		*base = (struct tioce *)ce_common->ce_pcibus.bs_base;
+	if (kernel)
+		*kernel = ce_kernel;
+
+	/*
+	 * we use port as a zero-based value internally, even though the
+	 * documentation is 1-based.
+	 */
+	if (port)
+		*port =
+		    (pdev->bus->number < ce_kernel->ce_port1_secondary) ? 0 : 1;
+}
+
+/**
+ * tioce_alloc_map - Given a coretalk address, map it to pcie bus address
+ * space using one of the various ATE-based address modes.
+ * @ce_kern: tioce context
+ * @type: map mode to use
+ * @port: 0-based port that the requesting device is downstream of
+ * @ct_addr: the coretalk address to map
+ * @len: number of bytes to map
+ *
+ * Given the addressing type, set up various paramaters that define the
+ * ATE pool to use.  Search for a contiguous block of entries to cover the
+ * length, and if enough resources exist, fill in the ATE's and construct a
+ * tioce_dmamap struct to track the mapping.
+ */
+static uint64_t
+tioce_alloc_map(struct tioce_kernel *ce_kern, int type, int port,
+		uint64_t ct_addr, int len)
+{
+	int i;
+	int j;
+	int first;
+	int last;
+	int entries;
+	int nates;
+	int pagesize;
+	uint64_t *ate_shadow;
+	uint64_t *ate_reg;
+	uint64_t addr;
+	struct tioce *ce_mmr;
+	uint64_t bus_base;
+	struct tioce_dmamap *map;
+
+	ce_mmr = (struct tioce *)ce_kern->ce_common->ce_pcibus.bs_base;
+
+	switch (type) {
+	case TIOCE_ATE_M32:
+		/*
+		 * The first 64 entries of the ate3240 pool are dedicated to
+		 * super-page (TIOCE_ATE_M40S) mode.
+		 */
+		first = 64;
+		entries = TIOCE_NUM_M3240_ATES - 64;
+		ate_shadow = ce_kern->ce_ate3240_shadow;
+		ate_reg = ce_mmr->ce_ure_ate3240;
+		pagesize = ce_kern->ce_ate3240_pagesize;
+		bus_base = TIOCE_M32_MIN;
+		break;
+	case TIOCE_ATE_M40:
+		first = 0;
+		entries = TIOCE_NUM_M40_ATES;
+		ate_shadow = ce_kern->ce_ate40_shadow;
+		ate_reg = ce_mmr->ce_ure_ate40;
+		pagesize = MB(64);
+		bus_base = TIOCE_M40_MIN;
+		break;
+	case TIOCE_ATE_M40S:
+		/*
+		 * ate3240 entries 0-31 are dedicated to port1 super-page
+		 * mappings.  ate3240 entries 32-63 are dedicated to port2.
+		 */
+		first = port * 32;
+		entries = 32;
+		ate_shadow = ce_kern->ce_ate3240_shadow;
+		ate_reg = ce_mmr->ce_ure_ate3240;
+		pagesize = GB(16);
+		bus_base = TIOCE_M40S_MIN;
+		break;
+	default:
+		return 0;
+	}
+
+	nates = ATE_NPAGES(ct_addr, len, pagesize);
+	if (nates > entries)
+		return 0;
+
+	last = first + entries - nates;
+	for (i = first; i <= last; i++) {
+		if (ATE_VALID(ate_shadow[i]))
+			continue;
+
+		for (j = i; j < i + nates; j++)
+			if (ATE_VALID(ate_shadow[j]))
+				break;
+
+		if (j >= i + nates)
+			break;
+	}
+
+	if (i > last)
+		return 0;
+
+	map = kcalloc(1, sizeof(struct tioce_dmamap), GFP_ATOMIC);
+	if (!map)
+		return 0;
+
+	addr = ct_addr;
+	for (j = 0; j < nates; j++) {
+		uint64_t ate;
+
+		ate = ATE_MAKE(addr, pagesize);
+		ate_shadow[i + j] = ate;
+		ate_reg[i + j] = ate;
+		addr += pagesize;
+	}
+
+	map->refcnt = 1;
+	map->nbytes = nates * pagesize;
+	map->ct_start = ct_addr & ~ATE_PAGEMASK(pagesize);
+	map->pci_start = bus_base + (i * pagesize);
+	map->ate_hw = &ate_reg[i];
+	map->ate_shadow = &ate_shadow[i];
+	map->ate_count = nates;
+
+	list_add(&map->ce_dmamap_list, &ce_kern->ce_dmamap_list);
+
+	return (map->pci_start + (ct_addr - map->ct_start));
+}
+
+/**
+ * tioce_dma_d32 - create a DMA mapping using 32-bit direct mode
+ * @pdev: linux pci_dev representing the function
+ * @paddr: system physical address
+ *
+ * Map @paddr into 32-bit bus space of the CE associated with @pcidev_info.
+ */
+static uint64_t
+tioce_dma_d32(struct pci_dev *pdev, uint64_t ct_addr)
+{
+	int dma_ok;
+	int port;
+	struct tioce *ce_mmr;
+	struct tioce_kernel *ce_kern;
+	uint64_t ct_upper;
+	uint64_t ct_lower;
+	dma_addr_t bus_addr;
+
+	ct_upper = ct_addr & ~0x3fffffffUL;
+	ct_lower = ct_addr & 0x3fffffffUL;
+
+	pcidev_to_tioce(pdev, &ce_mmr, &ce_kern, &port);
+
+	if (ce_kern->ce_port[port].dirmap_refcnt == 0) {
+		volatile uint64_t tmp;
+
+		ce_kern->ce_port[port].dirmap_shadow = ct_upper;
+		ce_mmr->ce_ure_dir_map[port] = ct_upper;
+		tmp = ce_mmr->ce_ure_dir_map[port];
+		dma_ok = 1;
+	} else
+		dma_ok = (ce_kern->ce_port[port].dirmap_shadow == ct_upper);
+
+	if (dma_ok) {
+		ce_kern->ce_port[port].dirmap_refcnt++;
+		bus_addr = TIOCE_D32_MIN + ct_lower;
+	} else
+		bus_addr = 0;
+
+	return bus_addr;
+}
+
+/**
+ * tioce_dma_barrier - swizzle a TIOCE bus address to include or exclude
+ * the barrier bit.
+ * @bus_addr:  bus address to swizzle
+ *
+ * Given a TIOCE bus address, set the appropriate bit to indicate barrier
+ * attributes.
+ */
+static uint64_t
+tioce_dma_barrier(uint64_t bus_addr, int on)
+{
+	uint64_t barrier_bit;
+
+	/* barrier not supported in M40/M40S mode */
+	if (TIOCE_M40_ADDR(bus_addr) || TIOCE_M40S_ADDR(bus_addr))
+		return bus_addr;
+
+	if (TIOCE_D64_ADDR(bus_addr))
+		barrier_bit = (1UL << 62);
+	else			/* must be m32 or d32 */
+		barrier_bit = (1UL << 30);
+
+	return (on) ? (bus_addr | barrier_bit) : (bus_addr & ~barrier_bit);
+}
+
+/**
+ * tioce_dma_unmap - release CE mapping resources
+ * @pdev: linux pci_dev representing the function
+ * @bus_addr: bus address returned by an earlier tioce_dma_map
+ * @dir: mapping direction (unused)
+ *
+ * Locate mapping resources associated with @bus_addr and release them.
+ * For mappings created using the direct modes there are no resources
+ * to release.
+ */
+void
+tioce_dma_unmap(struct pci_dev *pdev, dma_addr_t bus_addr, int dir)
+{
+	int i;
+	int port;
+	struct tioce_kernel *ce_kern;
+	struct tioce *ce_mmr;
+	unsigned long flags;
+
+	bus_addr = tioce_dma_barrier(bus_addr, 0);
+	pcidev_to_tioce(pdev, &ce_mmr, &ce_kern, &port);
+
+	/* nothing to do for D64 */
+
+	if (TIOCE_D64_ADDR(bus_addr))
+		return;
+
+	spin_lock_irqsave(&ce_kern->ce_lock, flags);
+
+	if (TIOCE_D32_ADDR(bus_addr)) {
+		if (--ce_kern->ce_port[port].dirmap_refcnt == 0) {
+			ce_kern->ce_port[port].dirmap_shadow = 0;
+			ce_mmr->ce_ure_dir_map[port] = 0;
+		}
+	} else {
+		struct tioce_dmamap *map;
+
+		list_for_each_entry(map, &ce_kern->ce_dmamap_list,
+				    ce_dmamap_list) {
+			uint64_t last;
+
+			last = map->pci_start + map->nbytes - 1;
+			if (bus_addr >= map->pci_start && bus_addr <= last)
+				break;
+		}
+
+		if (&map->ce_dmamap_list == &ce_kern->ce_dmamap_list) {
+			printk(KERN_WARNING
+			       "%s:  %s - no map found for bus_addr 0x%lx\n",
+			       __FUNCTION__, pci_name(pdev), bus_addr);
+		} else if (--map->refcnt == 0) {
+			for (i = 0; i < map->ate_count; i++) {
+				map->ate_shadow[i] = 0;
+				map->ate_hw[i] = 0;
+			}
+
+			list_del(&map->ce_dmamap_list);
+			kfree(map);
+		}
+	}
+
+	spin_unlock_irqrestore(&ce_kern->ce_lock, flags);
+}
+
+/**
+ * tioce_do_dma_map - map pages for PCI DMA
+ * @pdev: linux pci_dev representing the function
+ * @paddr: host physical address to map
+ * @byte_count: bytes to map
+ *
+ * This is the main wrapper for mapping host physical pages to CE PCI space.
+ * The mapping mode used is based on the device's dma_mask.
+ */
+static uint64_t
+tioce_do_dma_map(struct pci_dev *pdev, uint64_t paddr, size_t byte_count,
+		 int barrier)
+{
+	unsigned long flags;
+	uint64_t ct_addr;
+	uint64_t mapaddr = 0;
+	struct tioce_kernel *ce_kern;
+	struct tioce_dmamap *map;
+	int port;
+	uint64_t dma_mask;
+
+	dma_mask = (barrier) ? pdev->dev.coherent_dma_mask : pdev->dma_mask;
+
+	/* cards must be able to address at least 31 bits */
+	if (dma_mask < 0x7fffffffUL)
+		return 0;
+
+	ct_addr = PHYS_TO_TIODMA(paddr);
+
+	/*
+	 * If the device can generate 64 bit addresses, create a D64 map.
+	 * Since this should never fail, bypass the rest of the checks.
+	 */
+	if (dma_mask == ~0UL) {
+		mapaddr = tioce_dma_d64(ct_addr);
+		goto dma_map_done;
+	}
+
+	pcidev_to_tioce(pdev, NULL, &ce_kern, &port);
+
+	spin_lock_irqsave(&ce_kern->ce_lock, flags);
+
+	/*
+	 * D64 didn't work ... See if we have an existing map that covers
+	 * this address range.  Must account for devices dma_mask here since
+	 * an existing map might have been done in a mode using more pci
+	 * address bits than this device can support.
+	 */
+	list_for_each_entry(map, &ce_kern->ce_dmamap_list, ce_dmamap_list) {
+		uint64_t last;
+
+		last = map->ct_start + map->nbytes - 1;
+		if (ct_addr >= map->ct_start &&
+		    ct_addr + byte_count - 1 <= last &&
+		    map->pci_start <= dma_mask) {
+			map->refcnt++;
+			mapaddr = map->pci_start + (ct_addr - map->ct_start);
+			break;
+		}
+	}
+
+	/*
+	 * If we don't have a map yet, and the card can generate 40
+	 * bit addresses, try the M40/M40S modes.  Note these modes do not
+	 * support a barrier bit, so if we need a consistent map these
+	 * won't work.
+	 */
+	if (!mapaddr && !barrier && dma_mask >= 0xffffffffffUL) {
+		/*
+		 * We have two options for 40-bit mappings:  16GB "super" ATE's
+		 * and 64MB "regular" ATE's.  We'll try both if needed for a
+		 * given mapping but which one we try first depends on the
+		 * size.  For requests >64MB, prefer to use a super page with
+		 * regular as the fallback. Otherwise, try in the reverse order.
+		 */
+
+		if (byte_count > MB(64)) {
+			mapaddr = tioce_alloc_map(ce_kern, TIOCE_ATE_M40S,
+						  port, ct_addr, byte_count);
+			if (!mapaddr)
+				mapaddr =
+				    tioce_alloc_map(ce_kern, TIOCE_ATE_M40, -1,
+						    ct_addr, byte_count);
+		} else {
+			mapaddr = tioce_alloc_map(ce_kern, TIOCE_ATE_M40, -1,
+						  ct_addr, byte_count);
+			if (!mapaddr)
+				mapaddr =
+				    tioce_alloc_map(ce_kern, TIOCE_ATE_M40S,
+						    port, ct_addr, byte_count);
+		}
+	}
+
+	/*
+	 * 32-bit direct is the next mode to try
+	 */
+	if (!mapaddr && dma_mask >= 0xffffffffUL)
+		mapaddr = tioce_dma_d32(pdev, ct_addr);
+
+	/*
+	 * Last resort, try 32-bit ATE-based map.
+	 */
+	if (!mapaddr)
+		mapaddr =
+		    tioce_alloc_map(ce_kern, TIOCE_ATE_M32, -1, ct_addr,
+				    byte_count);
+
+	spin_unlock_irqrestore(&ce_kern->ce_lock, flags);
+
+dma_map_done:
+	if (mapaddr & barrier)
+		mapaddr = tioce_dma_barrier(mapaddr, 1);
+
+	return mapaddr;
+}
+
+/**
+ * tioce_dma - standard pci dma map interface
+ * @pdev: pci device requesting the map
+ * @paddr: system physical address to map into pci space
+ * @byte_count: # bytes to map
+ *
+ * Simply call tioce_do_dma_map() to create a map with the barrier bit clear
+ * in the address.
+ */
+static uint64_t
+tioce_dma(struct pci_dev *pdev, uint64_t paddr, size_t byte_count)
+{
+	return tioce_do_dma_map(pdev, paddr, byte_count, 0);
+}
+
+/**
+ * tioce_dma_consistent - consistent pci dma map interface
+ * @pdev: pci device requesting the map
+ * @paddr: system physical address to map into pci space
+ * @byte_count: # bytes to map
+ *
+ * Simply call tioce_do_dma_map() to create a map with the barrier bit set
+ * in the address.
+ */ static uint64_t
+tioce_dma_consistent(struct pci_dev *pdev, uint64_t paddr, size_t byte_count)
+{
+	return tioce_do_dma_map(pdev, paddr, byte_count, 1);
+}
+
+/**
+ * tioce_error_intr_handler - SGI TIO CE error interrupt handler
+ * @irq: unused
+ * @arg: pointer to tioce_common struct for the given CE
+ * @pt: unused
+ *
+ * Handle a CE error interrupt.  Simply a wrapper around a SAL call which
+ * defers processing to the SGI prom.
+ */ static irqreturn_t
+tioce_error_intr_handler(int irq, void *arg, struct pt_regs *pt)
+{
+	struct tioce_common *soft = arg;
+	struct ia64_sal_retval ret_stuff;
+	ret_stuff.status = 0;
+	ret_stuff.v0 = 0;
+
+	SAL_CALL_NOLOCK(ret_stuff, (u64) SN_SAL_IOIF_ERROR_INTERRUPT,
+			soft->ce_pcibus.bs_persist_segment,
+			soft->ce_pcibus.bs_persist_busnum, 0, 0, 0, 0, 0);
+
+	return IRQ_HANDLED;
+}
+
+/**
+ * tioce_kern_init - init kernel structures related to a given TIOCE
+ * @tioce_common: ptr to a cached tioce_common struct that originated in prom
+ */ static struct tioce_kernel *
+tioce_kern_init(struct tioce_common *tioce_common)
+{
+	int i;
+	uint32_t tmp;
+	struct tioce *tioce_mmr;
+	struct tioce_kernel *tioce_kern;
+
+	tioce_kern = kcalloc(1, sizeof(struct tioce_kernel), GFP_KERNEL);
+	if (!tioce_kern) {
+		return NULL;
+	}
+
+	tioce_kern->ce_common = tioce_common;
+	spin_lock_init(&tioce_kern->ce_lock);
+	INIT_LIST_HEAD(&tioce_kern->ce_dmamap_list);
+	tioce_common->ce_kernel_private = (uint64_t) tioce_kern;
+
+	/*
+	 * Determine the secondary bus number of the port2 logical PPB.
+	 * This is used to decide whether a given pci device resides on
+	 * port1 or port2.  Note:  We don't have enough plumbing set up
+	 * here to use pci_read_config_xxx() so use the raw_pci_ops vector.
+	 */
+
+	raw_pci_ops->read(tioce_common->ce_pcibus.bs_persist_segment,
+			  tioce_common->ce_pcibus.bs_persist_busnum,
+			  PCI_DEVFN(2, 0), PCI_SECONDARY_BUS, 1, &tmp);
+	tioce_kern->ce_port1_secondary = (uint8_t) tmp;
+
+	/*
+	 * Set PMU pagesize to the largest size available, and zero out
+	 * the ate's.
+	 */
+
+	tioce_mmr = (struct tioce *)tioce_common->ce_pcibus.bs_base;
+	tioce_mmr->ce_ure_page_map &= ~CE_URE_PAGESIZE_MASK;
+	tioce_mmr->ce_ure_page_map |= CE_URE_256K_PAGESIZE;
+	tioce_kern->ce_ate3240_pagesize = KB(256);
+
+	for (i = 0; i < TIOCE_NUM_M40_ATES; i++) {
+		tioce_kern->ce_ate40_shadow[i] = 0;
+		tioce_mmr->ce_ure_ate40[i] = 0;
+	}
+
+	for (i = 0; i < TIOCE_NUM_M3240_ATES; i++) {
+		tioce_kern->ce_ate3240_shadow[i] = 0;
+		tioce_mmr->ce_ure_ate3240[i] = 0;
+	}
+
+	return tioce_kern;
+}
+
+/**
+ * tioce_force_interrupt - implement altix force_interrupt() backend for CE
+ * @sn_irq_info: sn asic irq that we need an interrupt generated for
+ *
+ * Given an sn_irq_info struct, set the proper bit in ce_adm_force_int to
+ * force a secondary interrupt to be generated.  This is to work around an
+ * asic issue where there is a small window of opportunity for a legacy device
+ * interrupt to be lost.
+ */
+static void
+tioce_force_interrupt(struct sn_irq_info *sn_irq_info)
+{
+	struct pcidev_info *pcidev_info;
+	struct tioce_common *ce_common;
+	struct tioce *ce_mmr;
+	uint64_t force_int_val;
+
+	if (!sn_irq_info->irq_bridge)
+		return;
+
+	if (sn_irq_info->irq_bridge_type != PCIIO_ASIC_TYPE_TIOCE)
+		return;
+
+	pcidev_info = (struct pcidev_info *)sn_irq_info->irq_pciioinfo;
+	if (!pcidev_info)
+		return;
+
+	ce_common = (struct tioce_common *)pcidev_info->pdi_pcibus_info;
+	ce_mmr = (struct tioce *)ce_common->ce_pcibus.bs_base;
+
+	/*
+	 * irq_int_bit is originally set up by prom, and holds the interrupt
+	 * bit shift (not mask) as defined by the bit definitions in the
+	 * ce_adm_int mmr.  These shifts are not the same for the
+	 * ce_adm_force_int register, so do an explicit mapping here to make
+	 * things clearer.
+	 */
+
+	switch (sn_irq_info->irq_int_bit) {
+	case CE_ADM_INT_PCIE_PORT1_DEV_A_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT1_DEV_A_SHFT;
+		break;
+	case CE_ADM_INT_PCIE_PORT1_DEV_B_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT1_DEV_B_SHFT;
+		break;
+	case CE_ADM_INT_PCIE_PORT1_DEV_C_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT1_DEV_C_SHFT;
+		break;
+	case CE_ADM_INT_PCIE_PORT1_DEV_D_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT1_DEV_D_SHFT;
+		break;
+	case CE_ADM_INT_PCIE_PORT2_DEV_A_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT2_DEV_A_SHFT;
+		break;
+	case CE_ADM_INT_PCIE_PORT2_DEV_B_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT2_DEV_B_SHFT;
+		break;
+	case CE_ADM_INT_PCIE_PORT2_DEV_C_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT2_DEV_C_SHFT;
+		break;
+	case CE_ADM_INT_PCIE_PORT2_DEV_D_SHFT:
+		force_int_val = 1UL << CE_ADM_FORCE_INT_PCIE_PORT2_DEV_D_SHFT;
+		break;
+	default:
+		return;
+	}
+	ce_mmr->ce_adm_force_int = force_int_val;
+}
+
+/**
+ * tioce_target_interrupt - implement set_irq_affinity for tioce resident
+ * functions.  Note:  only applies to line interrupts, not MSI's.
+ *
+ * @sn_irq_info: SN IRQ context
+ *
+ * Given an sn_irq_info, set the associated CE device's interrupt destination
+ * register.  Since the interrupt destination registers are on a per-ce-slot
+ * basis, this will retarget line interrupts for all functions downstream of
+ * the slot.
+ */
+static void
+tioce_target_interrupt(struct sn_irq_info *sn_irq_info)
+{
+	struct pcidev_info *pcidev_info;
+	struct tioce_common *ce_common;
+	struct tioce *ce_mmr;
+	int bit;
+
+	pcidev_info = (struct pcidev_info *)sn_irq_info->irq_pciioinfo;
+	if (!pcidev_info)
+		return;
+
+	ce_common = (struct tioce_common *)pcidev_info->pdi_pcibus_info;
+	ce_mmr = (struct tioce *)ce_common->ce_pcibus.bs_base;
+
+	bit = sn_irq_info->irq_int_bit;
+
+	ce_mmr->ce_adm_int_mask |= (1UL << bit);
+	ce_mmr->ce_adm_int_dest[bit] =
+		((uint64_t)sn_irq_info->irq_irq << INTR_VECTOR_SHFT) |
+			   sn_irq_info->irq_xtalkaddr;
+	ce_mmr->ce_adm_int_mask &= ~(1UL << bit);
+
+	tioce_force_interrupt(sn_irq_info);
+}
+
+/**
+ * tioce_bus_fixup - perform final PCI fixup for a TIO CE bus
+ * @prom_bussoft: Common prom/kernel struct representing the bus
+ *
+ * Replicates the tioce_common pointed to by @prom_bussoft in kernel
+ * space.  Allocates and initializes a kernel-only area for a given CE,
+ * and sets up an irq for handling CE error interrupts.
+ *
+ * On successful setup, returns the kernel version of tioce_common back to
+ * the caller.
+ */
+static void *
+tioce_bus_fixup(struct pcibus_bussoft *prom_bussoft, struct pci_controller *controller)
+{
+	struct tioce_common *tioce_common;
+
+	/*
+	 * Allocate kernel bus soft and copy from prom.
+	 */
+
+	tioce_common = kcalloc(1, sizeof(struct tioce_common), GFP_KERNEL);
+	if (!tioce_common)
+		return NULL;
+
+	memcpy(tioce_common, prom_bussoft, sizeof(struct tioce_common));
+	tioce_common->ce_pcibus.bs_base |= __IA64_UNCACHED_OFFSET;
+
+	if (tioce_kern_init(tioce_common) == NULL) {
+		kfree(tioce_common);
+		return NULL;
+	}
+
+	if (request_irq(SGI_PCIASIC_ERROR,
+			tioce_error_intr_handler,
+			SA_SHIRQ, "TIOCE error", (void *)tioce_common))
+		printk(KERN_WARNING
+		       "%s:  Unable to get irq %d.  "
+		       "Error interrupts won't be routed for "
+		       "TIOCE bus %04x:%02x\n",
+		       __FUNCTION__, SGI_PCIASIC_ERROR,
+		       tioce_common->ce_pcibus.bs_persist_segment,
+		       tioce_common->ce_pcibus.bs_persist_busnum);
+
+	return tioce_common;
+}
+
+static struct sn_pcibus_provider tioce_pci_interfaces = {
+	.dma_map = tioce_dma,
+	.dma_map_consistent = tioce_dma_consistent,
+	.dma_unmap = tioce_dma_unmap,
+	.bus_fixup = tioce_bus_fixup,
+	.force_interrupt = tioce_force_interrupt,
+	.target_interrupt = tioce_target_interrupt
+};
+
+/**
+ * tioce_init_provider - init SN PCI provider ops for TIO CE
+ */
+int
+tioce_init_provider(void)
+{
+	sn_pci_provider[PCIIO_ASIC_TYPE_TIOCE] = &tioce_pci_interfaces;
+	return 0;
+}
diff --git a/arch/ppc/kernel/cpu_setup_6xx.S b/arch/ppc/kernel/cpu_setup_6xx.S
index 468721d..3fb1fb6 100644
--- a/arch/ppc/kernel/cpu_setup_6xx.S
+++ b/arch/ppc/kernel/cpu_setup_6xx.S
@@ -249,8 +249,11 @@
 	sync
 	isync
 
-	/* Enable L2 HW prefetch
+	/* Enable L2 HW prefetch, if L2 is enabled
 	 */
+	mfspr	r3,SPRN_L2CR
+	andis.	r3,r3,L2CR_L2E@h
+	beqlr
 	mfspr	r3,SPRN_MSSCR0
 	ori	r3,r3,3
 	sync
diff --git a/arch/ppc/kernel/l2cr.S b/arch/ppc/kernel/l2cr.S
index c394410..8611152 100644
--- a/arch/ppc/kernel/l2cr.S
+++ b/arch/ppc/kernel/l2cr.S
@@ -156,6 +156,26 @@
 	       The bit moved on the 7450.....
 	  ****/
 
+BEGIN_FTR_SECTION
+	/* Disable L2 prefetch on some 745x and try to ensure
+	 * L2 prefetch engines are idle. As explained by errata
+	 * text, we can't be sure they are, we just hope very hard
+	 * that well be enough (sic !). At least I noticed Apple
+	 * doesn't even bother doing the dcbf's here...
+	 */
+	mfspr	r4,SPRN_MSSCR0
+	rlwinm	r4,r4,0,0,29
+	sync
+	mtspr	SPRN_MSSCR0,r4
+	sync
+	isync
+	lis	r4,KERNELBASE@h
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+END_FTR_SECTION_IFSET(CPU_FTR_SPEC7450)
+
 	/* TODO: use HW flush assist when available */
 
 	lis	r4,0x0002
@@ -230,7 +250,16 @@
 	oris	r3,r3,0x8000
 	mtspr	SPRN_L2CR,r3
 	sync
-
+	
+	/* Enable L2 HW prefetch on 744x/745x */
+BEGIN_FTR_SECTION
+	mfspr	r3,SPRN_MSSCR0
+	ori	r3,r3,3
+	sync
+	mtspr	SPRN_MSSCR0,r3
+	sync
+	isync
+END_FTR_SECTION_IFSET(CPU_FTR_SPEC7450)
 4:
 
 	/* Restore HID0[DPM] to whatever it was before */
diff --git a/arch/ppc64/mm/slb_low.S b/arch/ppc64/mm/slb_low.S
index bab2558..698d6b9 100644
--- a/arch/ppc64/mm/slb_low.S
+++ b/arch/ppc64/mm/slb_low.S
@@ -97,25 +97,21 @@
 	lhz	r9,PACAHIGHHTLBAREAS(r13)
 	srdi	r11,r3,(HTLB_AREA_SHIFT-SID_SHIFT)
 	srd	r9,r9,r11
-	andi.	r9,r9,1
-	bne	5f
-
-	li	r11,SLB_VSID_USER
-
-	cmpldi	r3,16
-	bge	6f
-
-	lhz	r9,PACALOWHTLBAREAS(r13)
-	srd	r9,r9,r3
-	andi.	r9,r9,1
-
-	beq	6f
-
-5:	li	r11,SLB_VSID_USER|SLB_VSID_L
+	lhz	r11,PACALOWHTLBAREAS(r13)
+	srd	r11,r11,r3
+	or	r9,r9,r11
 END_FTR_SECTION_IFSET(CPU_FTR_16M_PAGE)
 #endif /* CONFIG_HUGETLB_PAGE */
 
-6:	ld	r9,PACACONTEXTID(r13)
+	li	r11,SLB_VSID_USER
+
+#ifdef CONFIG_HUGETLB_PAGE
+BEGIN_FTR_SECTION
+	rldimi	r11,r9,8,55		/* shift masked bit into SLB_VSID_L */
+END_FTR_SECTION_IFSET(CPU_FTR_16M_PAGE)
+#endif /* CONFIG_HUGETLB_PAGE */
+
+	ld	r9,PACACONTEXTID(r13)
 	rldimi	r3,r9,USER_ESID_BITS,0
 
 9:	/* r3 = protovsid, r11 = flags, r10 = esid_data, cr7 = <>KERNELBASE */
diff --git a/drivers/char/snsc_event.c b/drivers/char/snsc_event.c
index d692af5..baaa365 100644
--- a/drivers/char/snsc_event.c
+++ b/drivers/char/snsc_event.c
@@ -19,6 +19,7 @@
 #include <linux/sched.h>
 #include <linux/byteorder/generic.h>
 #include <asm/sn/sn_sal.h>
+#include <asm/unaligned.h>
 #include "snsc.h"
 
 static struct subch_data_s *event_sd;
@@ -62,13 +63,16 @@
 scdrv_parse_event(char *event, int *src, int *code, int *esp_code, char *desc)
 {
 	char *desc_end;
+	__be32 from_buf;
 
 	/* record event source address */
-	*src = be32_to_cpup((__be32 *)event);
+	from_buf = get_unaligned((__be32 *)event);
+	*src = be32_to_cpup(&from_buf);
 	event += 4; 			/* move on to event code */
 
 	/* record the system controller's event code */
-	*code = be32_to_cpup((__be32 *)event);
+	from_buf = get_unaligned((__be32 *)event);
+	*code = be32_to_cpup(&from_buf);
 	event += 4;			/* move on to event arguments */
 
 	/* how many arguments are in the packet? */
@@ -82,7 +86,8 @@
 		/* not an integer argument, so give up */
 		return -1;
 	}
-	*esp_code = be32_to_cpup((__be32 *)event);
+	from_buf = get_unaligned((__be32 *)event);
+	*esp_code = be32_to_cpup(&from_buf);
 	event += 4;
 
 	/* parse out the event description */
diff --git a/drivers/media/dvb/ttpci/Kconfig b/drivers/media/dvb/ttpci/Kconfig
index bf3c011..d8bf658 100644
--- a/drivers/media/dvb/ttpci/Kconfig
+++ b/drivers/media/dvb/ttpci/Kconfig
@@ -102,6 +102,9 @@
 	select VIDEO_DEV
 	select VIDEO_SAA7146_VV
 	select DVB_STV0299
+	select DVB_TDA1004X
+	select DVB_TDA10021
+	select FW_LOADER
 	help
 	  Support for simple SAA7146 based DVB cards
 	  (so called Budget- or Nova-PCI cards) without onboard
diff --git a/drivers/net/iseries_veth.c b/drivers/net/iseries_veth.c
index 183ba97..dc5d089 100644
--- a/drivers/net/iseries_veth.c
+++ b/drivers/net/iseries_veth.c
@@ -79,12 +79,55 @@
 #include <asm/iommu.h>
 #include <asm/vio.h>
 
-#include "iseries_veth.h"
+#undef DEBUG
 
 MODULE_AUTHOR("Kyle Lucke <klucke@us.ibm.com>");
 MODULE_DESCRIPTION("iSeries Virtual ethernet driver");
 MODULE_LICENSE("GPL");
 
+#define VETH_EVENT_CAP	(0)
+#define VETH_EVENT_FRAMES	(1)
+#define VETH_EVENT_MONITOR	(2)
+#define VETH_EVENT_FRAMES_ACK	(3)
+
+#define VETH_MAX_ACKS_PER_MSG	(20)
+#define VETH_MAX_FRAMES_PER_MSG	(6)
+
+struct veth_frames_data {
+	u32 addr[VETH_MAX_FRAMES_PER_MSG];
+	u16 len[VETH_MAX_FRAMES_PER_MSG];
+	u32 eofmask;
+};
+#define VETH_EOF_SHIFT		(32-VETH_MAX_FRAMES_PER_MSG)
+
+struct veth_frames_ack_data {
+	u16 token[VETH_MAX_ACKS_PER_MSG];
+};
+
+struct veth_cap_data {
+	u8 caps_version;
+	u8 rsvd1;
+	u16 num_buffers;
+	u16 ack_threshold;
+	u16 rsvd2;
+	u32 ack_timeout;
+	u32 rsvd3;
+	u64 rsvd4[3];
+};
+
+struct veth_lpevent {
+	struct HvLpEvent base_event;
+	union {
+		struct veth_cap_data caps_data;
+		struct veth_frames_data frames_data;
+		struct veth_frames_ack_data frames_ack_data;
+	} u;
+
+};
+
+#define DRV_NAME	"iseries_veth"
+#define DRV_VERSION	"2.0"
+
 #define VETH_NUMBUFFERS		(120)
 #define VETH_ACKTIMEOUT 	(1000000) /* microseconds */
 #define VETH_MAX_MCAST		(12)
@@ -113,9 +156,9 @@
 
 struct veth_msg {
 	struct veth_msg *next;
-	struct VethFramesData data;
+	struct veth_frames_data data;
 	int token;
-	unsigned long in_use;
+	int in_use;
 	struct sk_buff *skb;
 	struct device *dev;
 };
@@ -125,23 +168,28 @@
 	struct work_struct statemachine_wq;
 	struct veth_msg *msgs;
 	int num_events;
-	struct VethCapData local_caps;
+	struct veth_cap_data local_caps;
 
+	struct kobject kobject;
 	struct timer_list ack_timer;
 
+	struct timer_list reset_timer;
+	unsigned int reset_timeout;
+	unsigned long last_contact;
+	int outstanding_tx;
+
 	spinlock_t lock;
 	unsigned long state;
 	HvLpInstanceId src_inst;
 	HvLpInstanceId dst_inst;
-	struct VethLpEvent cap_event, cap_ack_event;
+	struct veth_lpevent cap_event, cap_ack_event;
 	u16 pending_acks[VETH_MAX_ACKS_PER_MSG];
 	u32 num_pending_acks;
 
 	int num_ack_events;
-	struct VethCapData remote_caps;
+	struct veth_cap_data remote_caps;
 	u32 ack_timeout;
 
-	spinlock_t msg_stack_lock;
 	struct veth_msg *msg_stack_head;
 };
 
@@ -151,15 +199,17 @@
 	u64 mac_addr;
 	HvLpIndexMap lpar_map;
 
-	spinlock_t pending_gate;
-	struct sk_buff *pending_skb;
-	HvLpIndexMap pending_lpmask;
+	/* queue_lock protects the stopped_map and dev's queue. */
+	spinlock_t queue_lock;
+	HvLpIndexMap stopped_map;
 
+	/* mcast_gate protects promiscuous, num_mcast & mcast_addr. */
 	rwlock_t mcast_gate;
 	int promiscuous;
-	int all_mcast;
 	int num_mcast;
 	u64 mcast_addr[VETH_MAX_MCAST];
+
+	struct kobject kobject;
 };
 
 static HvLpIndex this_lp;
@@ -168,44 +218,56 @@
 
 static int veth_start_xmit(struct sk_buff *skb, struct net_device *dev);
 static void veth_recycle_msg(struct veth_lpar_connection *, struct veth_msg *);
-static void veth_flush_pending(struct veth_lpar_connection *cnx);
-static void veth_receive(struct veth_lpar_connection *, struct VethLpEvent *);
-static void veth_timed_ack(unsigned long connectionPtr);
+static void veth_wake_queues(struct veth_lpar_connection *cnx);
+static void veth_stop_queues(struct veth_lpar_connection *cnx);
+static void veth_receive(struct veth_lpar_connection *, struct veth_lpevent *);
+static void veth_release_connection(struct kobject *kobject);
+static void veth_timed_ack(unsigned long ptr);
+static void veth_timed_reset(unsigned long ptr);
 
 /*
  * Utility functions
  */
 
-#define veth_printk(prio, fmt, args...) \
-	printk(prio "%s: " fmt, __FILE__, ## args)
+#define veth_info(fmt, args...) \
+	printk(KERN_INFO DRV_NAME ": " fmt, ## args)
 
 #define veth_error(fmt, args...) \
-	printk(KERN_ERR "(%s:%3.3d) ERROR: " fmt, __FILE__, __LINE__ , ## args)
+	printk(KERN_ERR DRV_NAME ": Error: " fmt, ## args)
 
+#ifdef DEBUG
+#define veth_debug(fmt, args...) \
+	printk(KERN_DEBUG DRV_NAME ": " fmt, ## args)
+#else
+#define veth_debug(fmt, args...) do {} while (0)
+#endif
+
+/* You must hold the connection's lock when you call this function. */
 static inline void veth_stack_push(struct veth_lpar_connection *cnx,
 				   struct veth_msg *msg)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(&cnx->msg_stack_lock, flags);
 	msg->next = cnx->msg_stack_head;
 	cnx->msg_stack_head = msg;
-	spin_unlock_irqrestore(&cnx->msg_stack_lock, flags);
 }
 
+/* You must hold the connection's lock when you call this function. */
 static inline struct veth_msg *veth_stack_pop(struct veth_lpar_connection *cnx)
 {
-	unsigned long flags;
 	struct veth_msg *msg;
 
-	spin_lock_irqsave(&cnx->msg_stack_lock, flags);
 	msg = cnx->msg_stack_head;
 	if (msg)
 		cnx->msg_stack_head = cnx->msg_stack_head->next;
-	spin_unlock_irqrestore(&cnx->msg_stack_lock, flags);
+
 	return msg;
 }
 
+/* You must hold the connection's lock when you call this function. */
+static inline int veth_stack_is_empty(struct veth_lpar_connection *cnx)
+{
+	return cnx->msg_stack_head == NULL;
+}
+
 static inline HvLpEvent_Rc
 veth_signalevent(struct veth_lpar_connection *cnx, u16 subtype,
 		 HvLpEvent_AckInd ackind, HvLpEvent_AckType acktype,
@@ -249,7 +311,7 @@
 	struct veth_allocation vc = { COMPLETION_INITIALIZER(vc.c), 0 };
 
 	mf_allocate_lp_events(rlp, HvLpEvent_Type_VirtualLan,
-			    sizeof(struct VethLpEvent), number,
+			    sizeof(struct veth_lpevent), number,
 			    &veth_complete_allocation, &vc);
 	wait_for_completion(&vc.c);
 
@@ -257,6 +319,137 @@
 }
 
 /*
+ * sysfs support
+ */
+
+struct veth_cnx_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct veth_lpar_connection *, char *buf);
+	ssize_t (*store)(struct veth_lpar_connection *, const char *buf);
+};
+
+static ssize_t veth_cnx_attribute_show(struct kobject *kobj,
+		struct attribute *attr, char *buf)
+{
+	struct veth_cnx_attribute *cnx_attr;
+	struct veth_lpar_connection *cnx;
+
+	cnx_attr = container_of(attr, struct veth_cnx_attribute, attr);
+	cnx = container_of(kobj, struct veth_lpar_connection, kobject);
+
+	if (!cnx_attr->show)
+		return -EIO;
+
+	return cnx_attr->show(cnx, buf);
+}
+
+#define CUSTOM_CNX_ATTR(_name, _format, _expression)			\
+static ssize_t _name##_show(struct veth_lpar_connection *cnx, char *buf)\
+{									\
+	return sprintf(buf, _format, _expression);			\
+}									\
+struct veth_cnx_attribute veth_cnx_attr_##_name = __ATTR_RO(_name)
+
+#define SIMPLE_CNX_ATTR(_name)	\
+	CUSTOM_CNX_ATTR(_name, "%lu\n", (unsigned long)cnx->_name)
+
+SIMPLE_CNX_ATTR(outstanding_tx);
+SIMPLE_CNX_ATTR(remote_lp);
+SIMPLE_CNX_ATTR(num_events);
+SIMPLE_CNX_ATTR(src_inst);
+SIMPLE_CNX_ATTR(dst_inst);
+SIMPLE_CNX_ATTR(num_pending_acks);
+SIMPLE_CNX_ATTR(num_ack_events);
+CUSTOM_CNX_ATTR(ack_timeout, "%d\n", jiffies_to_msecs(cnx->ack_timeout));
+CUSTOM_CNX_ATTR(reset_timeout, "%d\n", jiffies_to_msecs(cnx->reset_timeout));
+CUSTOM_CNX_ATTR(state, "0x%.4lX\n", cnx->state);
+CUSTOM_CNX_ATTR(last_contact, "%d\n", cnx->last_contact ?
+		jiffies_to_msecs(jiffies - cnx->last_contact) : 0);
+
+#define GET_CNX_ATTR(_name)	(&veth_cnx_attr_##_name.attr)
+
+static struct attribute *veth_cnx_default_attrs[] = {
+	GET_CNX_ATTR(outstanding_tx),
+	GET_CNX_ATTR(remote_lp),
+	GET_CNX_ATTR(num_events),
+	GET_CNX_ATTR(reset_timeout),
+	GET_CNX_ATTR(last_contact),
+	GET_CNX_ATTR(state),
+	GET_CNX_ATTR(src_inst),
+	GET_CNX_ATTR(dst_inst),
+	GET_CNX_ATTR(num_pending_acks),
+	GET_CNX_ATTR(num_ack_events),
+	GET_CNX_ATTR(ack_timeout),
+	NULL
+};
+
+static struct sysfs_ops veth_cnx_sysfs_ops = {
+		.show = veth_cnx_attribute_show
+};
+
+static struct kobj_type veth_lpar_connection_ktype = {
+	.release	= veth_release_connection,
+	.sysfs_ops	= &veth_cnx_sysfs_ops,
+	.default_attrs	= veth_cnx_default_attrs
+};
+
+struct veth_port_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct veth_port *, char *buf);
+	ssize_t (*store)(struct veth_port *, const char *buf);
+};
+
+static ssize_t veth_port_attribute_show(struct kobject *kobj,
+		struct attribute *attr, char *buf)
+{
+	struct veth_port_attribute *port_attr;
+	struct veth_port *port;
+
+	port_attr = container_of(attr, struct veth_port_attribute, attr);
+	port = container_of(kobj, struct veth_port, kobject);
+
+	if (!port_attr->show)
+		return -EIO;
+
+	return port_attr->show(port, buf);
+}
+
+#define CUSTOM_PORT_ATTR(_name, _format, _expression)			\
+static ssize_t _name##_show(struct veth_port *port, char *buf)		\
+{									\
+	return sprintf(buf, _format, _expression);			\
+}									\
+struct veth_port_attribute veth_port_attr_##_name = __ATTR_RO(_name)
+
+#define SIMPLE_PORT_ATTR(_name)	\
+	CUSTOM_PORT_ATTR(_name, "%lu\n", (unsigned long)port->_name)
+
+SIMPLE_PORT_ATTR(promiscuous);
+SIMPLE_PORT_ATTR(num_mcast);
+CUSTOM_PORT_ATTR(lpar_map, "0x%X\n", port->lpar_map);
+CUSTOM_PORT_ATTR(stopped_map, "0x%X\n", port->stopped_map);
+CUSTOM_PORT_ATTR(mac_addr, "0x%lX\n", port->mac_addr);
+
+#define GET_PORT_ATTR(_name)	(&veth_port_attr_##_name.attr)
+static struct attribute *veth_port_default_attrs[] = {
+	GET_PORT_ATTR(mac_addr),
+	GET_PORT_ATTR(lpar_map),
+	GET_PORT_ATTR(stopped_map),
+	GET_PORT_ATTR(promiscuous),
+	GET_PORT_ATTR(num_mcast),
+	NULL
+};
+
+static struct sysfs_ops veth_port_sysfs_ops = {
+	.show = veth_port_attribute_show
+};
+
+static struct kobj_type veth_port_ktype = {
+	.sysfs_ops	= &veth_port_sysfs_ops,
+	.default_attrs	= veth_port_default_attrs
+};
+
+/*
  * LPAR connection code
  */
 
@@ -266,7 +459,7 @@
 }
 
 static void veth_take_cap(struct veth_lpar_connection *cnx,
-			  struct VethLpEvent *event)
+			  struct veth_lpevent *event)
 {
 	unsigned long flags;
 
@@ -278,7 +471,7 @@
 						  HvLpEvent_Type_VirtualLan);
 
 	if (cnx->state & VETH_STATE_GOTCAPS) {
-		veth_error("Received a second capabilities from lpar %d\n",
+		veth_error("Received a second capabilities from LPAR %d.\n",
 			   cnx->remote_lp);
 		event->base_event.xRc = HvLpEvent_Rc_BufferNotAvailable;
 		HvCallEvent_ackLpEvent((struct HvLpEvent *) event);
@@ -291,13 +484,13 @@
 }
 
 static void veth_take_cap_ack(struct veth_lpar_connection *cnx,
-			      struct VethLpEvent *event)
+			      struct veth_lpevent *event)
 {
 	unsigned long flags;
 
 	spin_lock_irqsave(&cnx->lock, flags);
 	if (cnx->state & VETH_STATE_GOTCAPACK) {
-		veth_error("Received a second capabilities ack from lpar %d\n",
+		veth_error("Received a second capabilities ack from LPAR %d.\n",
 			   cnx->remote_lp);
 	} else {
 		memcpy(&cnx->cap_ack_event, event,
@@ -309,19 +502,24 @@
 }
 
 static void veth_take_monitor_ack(struct veth_lpar_connection *cnx,
-				  struct VethLpEvent *event)
+				  struct veth_lpevent *event)
 {
 	unsigned long flags;
 
 	spin_lock_irqsave(&cnx->lock, flags);
-	veth_printk(KERN_DEBUG, "Monitor ack returned for lpar %d\n",
-		    cnx->remote_lp);
-	cnx->state |= VETH_STATE_RESET;
-	veth_kick_statemachine(cnx);
+	veth_debug("cnx %d: lost connection.\n", cnx->remote_lp);
+
+	/* Avoid kicking the statemachine once we're shutdown.
+	 * It's unnecessary and it could break veth_stop_connection(). */
+
+	if (! (cnx->state & VETH_STATE_SHUTDOWN)) {
+		cnx->state |= VETH_STATE_RESET;
+		veth_kick_statemachine(cnx);
+	}
 	spin_unlock_irqrestore(&cnx->lock, flags);
 }
 
-static void veth_handle_ack(struct VethLpEvent *event)
+static void veth_handle_ack(struct veth_lpevent *event)
 {
 	HvLpIndex rlp = event->base_event.xTargetLp;
 	struct veth_lpar_connection *cnx = veth_cnx[rlp];
@@ -329,58 +527,67 @@
 	BUG_ON(! cnx);
 
 	switch (event->base_event.xSubtype) {
-	case VethEventTypeCap:
+	case VETH_EVENT_CAP:
 		veth_take_cap_ack(cnx, event);
 		break;
-	case VethEventTypeMonitor:
+	case VETH_EVENT_MONITOR:
 		veth_take_monitor_ack(cnx, event);
 		break;
 	default:
-		veth_error("Unknown ack type %d from lpar %d\n",
-			   event->base_event.xSubtype, rlp);
+		veth_error("Unknown ack type %d from LPAR %d.\n",
+				event->base_event.xSubtype, rlp);
 	};
 }
 
-static void veth_handle_int(struct VethLpEvent *event)
+static void veth_handle_int(struct veth_lpevent *event)
 {
 	HvLpIndex rlp = event->base_event.xSourceLp;
 	struct veth_lpar_connection *cnx = veth_cnx[rlp];
 	unsigned long flags;
-	int i;
+	int i, acked = 0;
 
 	BUG_ON(! cnx);
 
 	switch (event->base_event.xSubtype) {
-	case VethEventTypeCap:
+	case VETH_EVENT_CAP:
 		veth_take_cap(cnx, event);
 		break;
-	case VethEventTypeMonitor:
+	case VETH_EVENT_MONITOR:
 		/* do nothing... this'll hang out here til we're dead,
 		 * and the hypervisor will return it for us. */
 		break;
-	case VethEventTypeFramesAck:
+	case VETH_EVENT_FRAMES_ACK:
 		spin_lock_irqsave(&cnx->lock, flags);
+
 		for (i = 0; i < VETH_MAX_ACKS_PER_MSG; ++i) {
 			u16 msgnum = event->u.frames_ack_data.token[i];
 
-			if (msgnum < VETH_NUMBUFFERS)
+			if (msgnum < VETH_NUMBUFFERS) {
 				veth_recycle_msg(cnx, cnx->msgs + msgnum);
+				cnx->outstanding_tx--;
+				acked++;
+			}
 		}
+
+		if (acked > 0) {
+			cnx->last_contact = jiffies;
+			veth_wake_queues(cnx);
+		}
+
 		spin_unlock_irqrestore(&cnx->lock, flags);
-		veth_flush_pending(cnx);
 		break;
-	case VethEventTypeFrames:
+	case VETH_EVENT_FRAMES:
 		veth_receive(cnx, event);
 		break;
 	default:
-		veth_error("Unknown interrupt type %d from lpar %d\n",
-			   event->base_event.xSubtype, rlp);
+		veth_error("Unknown interrupt type %d from LPAR %d.\n",
+				event->base_event.xSubtype, rlp);
 	};
 }
 
 static void veth_handle_event(struct HvLpEvent *event, struct pt_regs *regs)
 {
-	struct VethLpEvent *veth_event = (struct VethLpEvent *)event;
+	struct veth_lpevent *veth_event = (struct veth_lpevent *)event;
 
 	if (event->xFlags.xFunction == HvLpEvent_Function_Ack)
 		veth_handle_ack(veth_event);
@@ -390,7 +597,7 @@
 
 static int veth_process_caps(struct veth_lpar_connection *cnx)
 {
-	struct VethCapData *remote_caps = &cnx->remote_caps;
+	struct veth_cap_data *remote_caps = &cnx->remote_caps;
 	int num_acks_needed;
 
 	/* Convert timer to jiffies */
@@ -400,8 +607,8 @@
 	     || (remote_caps->ack_threshold > VETH_MAX_ACKS_PER_MSG)
 	     || (remote_caps->ack_threshold == 0)
 	     || (cnx->ack_timeout == 0) ) {
-		veth_error("Received incompatible capabilities from lpar %d\n",
-			   cnx->remote_lp);
+		veth_error("Received incompatible capabilities from LPAR %d.\n",
+				cnx->remote_lp);
 		return HvLpEvent_Rc_InvalidSubtypeData;
 	}
 
@@ -418,8 +625,8 @@
 			cnx->num_ack_events += num;
 
 		if (cnx->num_ack_events < num_acks_needed) {
-			veth_error("Couldn't allocate enough ack events for lpar %d\n",
-				   cnx->remote_lp);
+			veth_error("Couldn't allocate enough ack events "
+					"for LPAR %d.\n", cnx->remote_lp);
 
 			return HvLpEvent_Rc_BufferNotAvailable;
 		}
@@ -440,15 +647,15 @@
 
  restart:
 	if (cnx->state & VETH_STATE_RESET) {
-		int i;
-
-		del_timer(&cnx->ack_timer);
-
 		if (cnx->state & VETH_STATE_OPEN)
 			HvCallEvent_closeLpEventPath(cnx->remote_lp,
 						     HvLpEvent_Type_VirtualLan);
 
-		/* reset ack data */
+		/*
+		 * Reset ack data. This prevents the ack_timer actually
+		 * doing anything, even if it runs one more time when
+		 * we drop the lock below.
+		 */
 		memset(&cnx->pending_acks, 0xff, sizeof (cnx->pending_acks));
 		cnx->num_pending_acks = 0;
 
@@ -458,14 +665,32 @@
 				| VETH_STATE_SENTCAPACK | VETH_STATE_READY);
 
 		/* Clean up any leftover messages */
-		if (cnx->msgs)
+		if (cnx->msgs) {
+			int i;
 			for (i = 0; i < VETH_NUMBUFFERS; ++i)
 				veth_recycle_msg(cnx, cnx->msgs + i);
+		}
+
+		cnx->outstanding_tx = 0;
+		veth_wake_queues(cnx);
+
+		/* Drop the lock so we can do stuff that might sleep or
+		 * take other locks. */
 		spin_unlock_irq(&cnx->lock);
-		veth_flush_pending(cnx);
+
+		del_timer_sync(&cnx->ack_timer);
+		del_timer_sync(&cnx->reset_timer);
+
 		spin_lock_irq(&cnx->lock);
+
 		if (cnx->state & VETH_STATE_RESET)
 			goto restart;
+
+		/* Hack, wait for the other end to reset itself. */
+		if (! (cnx->state & VETH_STATE_SHUTDOWN)) {
+			schedule_delayed_work(&cnx->statemachine_wq, 5 * HZ);
+			goto out;
+		}
 	}
 
 	if (cnx->state & VETH_STATE_SHUTDOWN)
@@ -488,7 +713,7 @@
 
 	if ( (cnx->state & VETH_STATE_OPEN)
 	     && !(cnx->state & VETH_STATE_SENTMON) ) {
-		rc = veth_signalevent(cnx, VethEventTypeMonitor,
+		rc = veth_signalevent(cnx, VETH_EVENT_MONITOR,
 				      HvLpEvent_AckInd_DoAck,
 				      HvLpEvent_AckType_DeferredAck,
 				      0, 0, 0, 0, 0, 0);
@@ -498,9 +723,8 @@
 		} else {
 			if ( (rc != HvLpEvent_Rc_PartitionDead)
 			     && (rc != HvLpEvent_Rc_PathClosed) )
-				veth_error("Error sending monitor to "
-					   "lpar %d, rc=%x\n",
-					   rlp, (int) rc);
+				veth_error("Error sending monitor to LPAR %d, "
+						"rc = %d\n", rlp, rc);
 
 			/* Oh well, hope we get a cap from the other
 			 * end and do better when that kicks us */
@@ -512,7 +736,7 @@
 	     && !(cnx->state & VETH_STATE_SENTCAPS)) {
 		u64 *rawcap = (u64 *)&cnx->local_caps;
 
-		rc = veth_signalevent(cnx, VethEventTypeCap,
+		rc = veth_signalevent(cnx, VETH_EVENT_CAP,
 				      HvLpEvent_AckInd_DoAck,
 				      HvLpEvent_AckType_ImmediateAck,
 				      0, rawcap[0], rawcap[1], rawcap[2],
@@ -523,9 +747,9 @@
 		} else {
 			if ( (rc != HvLpEvent_Rc_PartitionDead)
 			     && (rc != HvLpEvent_Rc_PathClosed) )
-				veth_error("Error sending caps to "
-					   "lpar %d, rc=%x\n",
-					   rlp, (int) rc);
+				veth_error("Error sending caps to LPAR %d, "
+						"rc = %d\n", rlp, rc);
+
 			/* Oh well, hope we get a cap from the other
 			 * end and do better when that kicks us */
 			goto out;
@@ -534,7 +758,7 @@
 
 	if ((cnx->state & VETH_STATE_GOTCAPS)
 	    && !(cnx->state & VETH_STATE_SENTCAPACK)) {
-		struct VethCapData *remote_caps = &cnx->remote_caps;
+		struct veth_cap_data *remote_caps = &cnx->remote_caps;
 
 		memcpy(remote_caps, &cnx->cap_event.u.caps_data,
 		       sizeof(*remote_caps));
@@ -565,10 +789,8 @@
 			add_timer(&cnx->ack_timer);
 			cnx->state |= VETH_STATE_READY;
 		} else {
-			veth_printk(KERN_ERR, "Caps rejected (rc=%d) by "
-				    "lpar %d\n",
-				    cnx->cap_ack_event.base_event.xRc,
-				    rlp);
+			veth_error("Caps rejected by LPAR %d, rc = %d\n",
+					rlp, cnx->cap_ack_event.base_event.xRc);
 			goto cant_cope;
 		}
 	}
@@ -581,8 +803,8 @@
 	/* FIXME: we get here if something happens we really can't
 	 * cope with.  The link will never work once we get here, and
 	 * all we can do is not lock the rest of the system up */
-	veth_error("Badness on connection to lpar %d (state=%04lx) "
-		   " - shutting down\n", rlp, cnx->state);
+	veth_error("Unrecoverable error on connection to LPAR %d, shutting down"
+			" (state = 0x%04lx)\n", rlp, cnx->state);
 	cnx->state |= VETH_STATE_SHUTDOWN;
 	spin_unlock_irq(&cnx->lock);
 }
@@ -591,7 +813,7 @@
 {
 	struct veth_lpar_connection *cnx;
 	struct veth_msg *msgs;
-	int i;
+	int i, rc;
 
 	if ( (rlp == this_lp)
 	     || ! HvLpConfig_doLpsCommunicateOnVirtualLan(this_lp, rlp) )
@@ -605,22 +827,36 @@
 	cnx->remote_lp = rlp;
 	spin_lock_init(&cnx->lock);
 	INIT_WORK(&cnx->statemachine_wq, veth_statemachine, cnx);
+
 	init_timer(&cnx->ack_timer);
 	cnx->ack_timer.function = veth_timed_ack;
 	cnx->ack_timer.data = (unsigned long) cnx;
+
+	init_timer(&cnx->reset_timer);
+	cnx->reset_timer.function = veth_timed_reset;
+	cnx->reset_timer.data = (unsigned long) cnx;
+	cnx->reset_timeout = 5 * HZ * (VETH_ACKTIMEOUT / 1000000);
+
 	memset(&cnx->pending_acks, 0xff, sizeof (cnx->pending_acks));
 
 	veth_cnx[rlp] = cnx;
 
+	/* This gets us 1 reference, which is held on behalf of the driver
+	 * infrastructure. It's released at module unload. */
+	kobject_init(&cnx->kobject);
+	cnx->kobject.ktype = &veth_lpar_connection_ktype;
+	rc = kobject_set_name(&cnx->kobject, "cnx%.2d", rlp);
+	if (rc != 0)
+		return rc;
+
 	msgs = kmalloc(VETH_NUMBUFFERS * sizeof(struct veth_msg), GFP_KERNEL);
 	if (! msgs) {
-		veth_error("Can't allocate buffers for lpar %d\n", rlp);
+		veth_error("Can't allocate buffers for LPAR %d.\n", rlp);
 		return -ENOMEM;
 	}
 
 	cnx->msgs = msgs;
 	memset(msgs, 0, VETH_NUMBUFFERS * sizeof(struct veth_msg));
-	spin_lock_init(&cnx->msg_stack_lock);
 
 	for (i = 0; i < VETH_NUMBUFFERS; i++) {
 		msgs[i].token = i;
@@ -630,8 +866,7 @@
 	cnx->num_events = veth_allocate_events(rlp, 2 + VETH_NUMBUFFERS);
 
 	if (cnx->num_events < (2 + VETH_NUMBUFFERS)) {
-		veth_error("Can't allocate events for lpar %d, only got %d\n",
-			   rlp, cnx->num_events);
+		veth_error("Can't allocate enough events for LPAR %d.\n", rlp);
 		return -ENOMEM;
 	}
 
@@ -642,11 +877,9 @@
 	return 0;
 }
 
-static void veth_stop_connection(u8 rlp)
+static void veth_stop_connection(struct veth_lpar_connection *cnx)
 {
-	struct veth_lpar_connection *cnx = veth_cnx[rlp];
-
-	if (! cnx)
+	if (!cnx)
 		return;
 
 	spin_lock_irq(&cnx->lock);
@@ -654,12 +887,23 @@
 	veth_kick_statemachine(cnx);
 	spin_unlock_irq(&cnx->lock);
 
-	flush_scheduled_work();
+	/* There's a slim chance the reset code has just queued the
+	 * statemachine to run in five seconds. If so we need to cancel
+	 * that and requeue the work to run now. */
+	if (cancel_delayed_work(&cnx->statemachine_wq)) {
+		spin_lock_irq(&cnx->lock);
+		veth_kick_statemachine(cnx);
+		spin_unlock_irq(&cnx->lock);
+	}
 
-	/* FIXME: not sure if this is necessary - will already have
-	 * been deleted by the state machine, just want to make sure
-	 * its not running any more */
-	del_timer_sync(&cnx->ack_timer);
+	/* Wait for the state machine to run. */
+	flush_scheduled_work();
+}
+
+static void veth_destroy_connection(struct veth_lpar_connection *cnx)
+{
+	if (!cnx)
+		return;
 
 	if (cnx->num_events > 0)
 		mf_deallocate_lp_events(cnx->remote_lp,
@@ -671,18 +915,18 @@
 				      HvLpEvent_Type_VirtualLan,
 				      cnx->num_ack_events,
 				      NULL, NULL);
-}
-
-static void veth_destroy_connection(u8 rlp)
-{
-	struct veth_lpar_connection *cnx = veth_cnx[rlp];
-
-	if (! cnx)
-		return;
 
 	kfree(cnx->msgs);
+	veth_cnx[cnx->remote_lp] = NULL;
 	kfree(cnx);
-	veth_cnx[rlp] = NULL;
+}
+
+static void veth_release_connection(struct kobject *kobj)
+{
+	struct veth_lpar_connection *cnx;
+	cnx = container_of(kobj, struct veth_lpar_connection, kobject);
+	veth_stop_connection(cnx);
+	veth_destroy_connection(cnx);
 }
 
 /*
@@ -726,17 +970,15 @@
 
 	write_lock_irqsave(&port->mcast_gate, flags);
 
-	if (dev->flags & IFF_PROMISC) {	/* set promiscuous mode */
-		printk(KERN_INFO "%s: Promiscuous mode enabled.\n",
-		       dev->name);
+	if ((dev->flags & IFF_PROMISC) || (dev->flags & IFF_ALLMULTI) ||
+			(dev->mc_count > VETH_MAX_MCAST)) {
 		port->promiscuous = 1;
-	} else if ( (dev->flags & IFF_ALLMULTI)
-		    || (dev->mc_count > VETH_MAX_MCAST) ) {
-		port->all_mcast = 1;
 	} else {
 		struct dev_mc_list *dmi = dev->mc_list;
 		int i;
 
+		port->promiscuous = 0;
+
 		/* Update table */
 		port->num_mcast = 0;
 
@@ -758,9 +1000,10 @@
 
 static void veth_get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *info)
 {
-	strncpy(info->driver, "veth", sizeof(info->driver) - 1);
+	strncpy(info->driver, DRV_NAME, sizeof(info->driver) - 1);
 	info->driver[sizeof(info->driver) - 1] = '\0';
-	strncpy(info->version, "1.0", sizeof(info->version) - 1);
+	strncpy(info->version, DRV_VERSION, sizeof(info->version) - 1);
+	info->version[sizeof(info->version) - 1] = '\0';
 }
 
 static int veth_get_settings(struct net_device *dev, struct ethtool_cmd *ecmd)
@@ -791,49 +1034,6 @@
 	.get_link = veth_get_link,
 };
 
-static void veth_tx_timeout(struct net_device *dev)
-{
-	struct veth_port *port = (struct veth_port *)dev->priv;
-	struct net_device_stats *stats = &port->stats;
-	unsigned long flags;
-	int i;
-
-	stats->tx_errors++;
-
-	spin_lock_irqsave(&port->pending_gate, flags);
-
-	if (!port->pending_lpmask) {
-		spin_unlock_irqrestore(&port->pending_gate, flags);
-		return;
-	}
-
-	printk(KERN_WARNING "%s: Tx timeout!  Resetting lp connections: %08x\n",
-	       dev->name, port->pending_lpmask);
-
-	for (i = 0; i < HVMAXARCHITECTEDLPS; i++) {
-		struct veth_lpar_connection *cnx = veth_cnx[i];
-
-		if (! (port->pending_lpmask & (1<<i)))
-			continue;
-
-		/* If we're pending on it, we must be connected to it,
-		 * so we should certainly have a structure for it. */
-		BUG_ON(! cnx);
-
-		/* Theoretically we could be kicking a connection
-		 * which doesn't deserve it, but in practice if we've
-		 * had a Tx timeout, the pending_lpmask will have
-		 * exactly one bit set - the connection causing the
-		 * problem. */
-		spin_lock(&cnx->lock);
-		cnx->state |= VETH_STATE_RESET;
-		veth_kick_statemachine(cnx);
-		spin_unlock(&cnx->lock);
-	}
-
-	spin_unlock_irqrestore(&port->pending_gate, flags);
-}
-
 static struct net_device * __init veth_probe_one(int vlan, struct device *vdev)
 {
 	struct net_device *dev;
@@ -848,8 +1048,9 @@
 
 	port = (struct veth_port *) dev->priv;
 
-	spin_lock_init(&port->pending_gate);
+	spin_lock_init(&port->queue_lock);
 	rwlock_init(&port->mcast_gate);
+	port->stopped_map = 0;
 
 	for (i = 0; i < HVMAXARCHITECTEDLPS; i++) {
 		HvLpVirtualLanIndexMap map;
@@ -882,22 +1083,24 @@
 	dev->set_multicast_list = veth_set_multicast_list;
 	SET_ETHTOOL_OPS(dev, &ops);
 
-	dev->watchdog_timeo = 2 * (VETH_ACKTIMEOUT * HZ / 1000000);
-	dev->tx_timeout = veth_tx_timeout;
-
 	SET_NETDEV_DEV(dev, vdev);
 
 	rc = register_netdev(dev);
 	if (rc != 0) {
-		veth_printk(KERN_ERR,
-			    "Failed to register ethernet device for vlan %d\n",
-			    vlan);
+		veth_error("Failed registering net device for vlan%d.\n", vlan);
 		free_netdev(dev);
 		return NULL;
 	}
 
-	veth_printk(KERN_DEBUG, "%s attached to iSeries vlan %d (lpar_map=0x%04x)\n",
-		    dev->name, vlan, port->lpar_map);
+	kobject_init(&port->kobject);
+	port->kobject.parent = &dev->class_dev.kobj;
+	port->kobject.ktype  = &veth_port_ktype;
+	kobject_set_name(&port->kobject, "veth_port");
+	if (0 != kobject_add(&port->kobject))
+		veth_error("Failed adding port for %s to sysfs.\n", dev->name);
+
+	veth_info("%s attached to iSeries vlan %d (LPAR map = 0x%.4X)\n",
+			dev->name, vlan, port->lpar_map);
 
 	return dev;
 }
@@ -912,98 +1115,95 @@
 	struct veth_lpar_connection *cnx = veth_cnx[rlp];
 	struct veth_port *port = (struct veth_port *) dev->priv;
 	HvLpEvent_Rc rc;
-	u32 dma_address, dma_length;
 	struct veth_msg *msg = NULL;
-	int err = 0;
 	unsigned long flags;
 
-	if (! cnx) {
-		port->stats.tx_errors++;
-		dev_kfree_skb(skb);
+	if (! cnx)
 		return 0;
-	}
 
 	spin_lock_irqsave(&cnx->lock, flags);
 
 	if (! (cnx->state & VETH_STATE_READY))
-		goto drop;
+		goto no_error;
 
-	if ((skb->len - 14) > VETH_MAX_MTU)
+	if ((skb->len - ETH_HLEN) > VETH_MAX_MTU)
 		goto drop;
 
 	msg = veth_stack_pop(cnx);
-
-	if (! msg) {
-		err = 1;
+	if (! msg)
 		goto drop;
-	}
 
-	dma_length = skb->len;
-	dma_address = dma_map_single(port->dev, skb->data,
-				     dma_length, DMA_TO_DEVICE);
+	msg->in_use = 1;
+	msg->skb = skb_get(skb);
 
-	if (dma_mapping_error(dma_address))
+	msg->data.addr[0] = dma_map_single(port->dev, skb->data,
+				skb->len, DMA_TO_DEVICE);
+
+	if (dma_mapping_error(msg->data.addr[0]))
 		goto recycle_and_drop;
 
-	/* Is it really necessary to check the length and address
-	 * fields of the first entry here? */
-	msg->skb = skb;
 	msg->dev = port->dev;
-	msg->data.addr[0] = dma_address;
-	msg->data.len[0] = dma_length;
+	msg->data.len[0] = skb->len;
 	msg->data.eofmask = 1 << VETH_EOF_SHIFT;
-	set_bit(0, &(msg->in_use));
-	rc = veth_signaldata(cnx, VethEventTypeFrames, msg->token, &msg->data);
+
+	rc = veth_signaldata(cnx, VETH_EVENT_FRAMES, msg->token, &msg->data);
 
 	if (rc != HvLpEvent_Rc_Good)
 		goto recycle_and_drop;
 
+	/* If the timer's not already running, start it now. */
+	if (0 == cnx->outstanding_tx)
+		mod_timer(&cnx->reset_timer, jiffies + cnx->reset_timeout);
+
+	cnx->last_contact = jiffies;
+	cnx->outstanding_tx++;
+
+	if (veth_stack_is_empty(cnx))
+		veth_stop_queues(cnx);
+
+ no_error:
 	spin_unlock_irqrestore(&cnx->lock, flags);
 	return 0;
 
  recycle_and_drop:
-	msg->skb = NULL;
-	/* need to set in use to make veth_recycle_msg in case this
-	 * was a mapping failure */
-	set_bit(0, &msg->in_use);
 	veth_recycle_msg(cnx, msg);
  drop:
-	port->stats.tx_errors++;
-	dev_kfree_skb(skb);
 	spin_unlock_irqrestore(&cnx->lock, flags);
-	return err;
+	return 1;
 }
 
-static HvLpIndexMap veth_transmit_to_many(struct sk_buff *skb,
+static void veth_transmit_to_many(struct sk_buff *skb,
 					  HvLpIndexMap lpmask,
 					  struct net_device *dev)
 {
 	struct veth_port *port = (struct veth_port *) dev->priv;
-	int i;
-	int rc;
+	int i, success, error;
+
+	success = error = 0;
 
 	for (i = 0; i < HVMAXARCHITECTEDLPS; i++) {
 		if ((lpmask & (1 << i)) == 0)
 			continue;
 
-		rc = veth_transmit_to_one(skb_get(skb), i, dev);
-		if (! rc)
-			lpmask &= ~(1<<i);
+		if (veth_transmit_to_one(skb, i, dev))
+			error = 1;
+		else
+			success = 1;
 	}
 
-	if (! lpmask) {
+	if (error)
+		port->stats.tx_errors++;
+
+	if (success) {
 		port->stats.tx_packets++;
 		port->stats.tx_bytes += skb->len;
 	}
-
-	return lpmask;
 }
 
 static int veth_start_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	unsigned char *frame = skb->data;
 	struct veth_port *port = (struct veth_port *) dev->priv;
-	unsigned long flags;
 	HvLpIndexMap lpmask;
 
 	if (! (frame[0] & 0x01)) {
@@ -1020,44 +1220,27 @@
 		lpmask = port->lpar_map;
 	}
 
-	spin_lock_irqsave(&port->pending_gate, flags);
+	veth_transmit_to_many(skb, lpmask, dev);
 
-	lpmask = veth_transmit_to_many(skb, lpmask, dev);
-
-	dev->trans_start = jiffies;
-
-	if (! lpmask) {
-		dev_kfree_skb(skb);
-	} else {
-		if (port->pending_skb) {
-			veth_error("%s: Tx while skb was pending!\n",
-				   dev->name);
-			dev_kfree_skb(skb);
-			spin_unlock_irqrestore(&port->pending_gate, flags);
-			return 1;
-		}
-
-		port->pending_skb = skb;
-		port->pending_lpmask = lpmask;
-		netif_stop_queue(dev);
-	}
-
-	spin_unlock_irqrestore(&port->pending_gate, flags);
+	dev_kfree_skb(skb);
 
 	return 0;
 }
 
+/* You must hold the connection's lock when you call this function. */
 static void veth_recycle_msg(struct veth_lpar_connection *cnx,
 			     struct veth_msg *msg)
 {
 	u32 dma_address, dma_length;
 
-	if (test_and_clear_bit(0, &msg->in_use)) {
+	if (msg->in_use) {
+		msg->in_use = 0;
 		dma_address = msg->data.addr[0];
 		dma_length = msg->data.len[0];
 
-		dma_unmap_single(msg->dev, dma_address, dma_length,
-				 DMA_TO_DEVICE);
+		if (!dma_mapping_error(dma_address))
+			dma_unmap_single(msg->dev, dma_address, dma_length,
+					DMA_TO_DEVICE);
 
 		if (msg->skb) {
 			dev_kfree_skb_any(msg->skb);
@@ -1066,15 +1249,16 @@
 
 		memset(&msg->data, 0, sizeof(msg->data));
 		veth_stack_push(cnx, msg);
-	} else
-		if (cnx->state & VETH_STATE_OPEN)
-			veth_error("Bogus frames ack from lpar %d (#%d)\n",
-				   cnx->remote_lp, msg->token);
+	} else if (cnx->state & VETH_STATE_OPEN) {
+		veth_error("Non-pending frame (# %d) acked by LPAR %d.\n",
+				cnx->remote_lp, msg->token);
+	}
 }
 
-static void veth_flush_pending(struct veth_lpar_connection *cnx)
+static void veth_wake_queues(struct veth_lpar_connection *cnx)
 {
 	int i;
+
 	for (i = 0; i < HVMAXARCHITECTEDVIRTUALLANS; i++) {
 		struct net_device *dev = veth_dev[i];
 		struct veth_port *port;
@@ -1088,22 +1272,79 @@
 		if (! (port->lpar_map & (1<<cnx->remote_lp)))
 			continue;
 
-		spin_lock_irqsave(&port->pending_gate, flags);
-		if (port->pending_skb) {
-			port->pending_lpmask =
-				veth_transmit_to_many(port->pending_skb,
-						      port->pending_lpmask,
-						      dev);
-			if (! port->pending_lpmask) {
-				dev_kfree_skb_any(port->pending_skb);
-				port->pending_skb = NULL;
-				netif_wake_queue(dev);
-			}
+		spin_lock_irqsave(&port->queue_lock, flags);
+
+		port->stopped_map &= ~(1 << cnx->remote_lp);
+
+		if (0 == port->stopped_map && netif_queue_stopped(dev)) {
+			veth_debug("cnx %d: woke queue for %s.\n",
+					cnx->remote_lp, dev->name);
+			netif_wake_queue(dev);
 		}
-		spin_unlock_irqrestore(&port->pending_gate, flags);
+		spin_unlock_irqrestore(&port->queue_lock, flags);
 	}
 }
 
+static void veth_stop_queues(struct veth_lpar_connection *cnx)
+{
+	int i;
+
+	for (i = 0; i < HVMAXARCHITECTEDVIRTUALLANS; i++) {
+		struct net_device *dev = veth_dev[i];
+		struct veth_port *port;
+
+		if (! dev)
+			continue;
+
+		port = (struct veth_port *)dev->priv;
+
+		/* If this cnx is not on the vlan for this port, continue */
+		if (! (port->lpar_map & (1 << cnx->remote_lp)))
+			continue;
+
+		spin_lock(&port->queue_lock);
+
+		netif_stop_queue(dev);
+		port->stopped_map |= (1 << cnx->remote_lp);
+
+		veth_debug("cnx %d: stopped queue for %s, map = 0x%x.\n",
+				cnx->remote_lp, dev->name, port->stopped_map);
+
+		spin_unlock(&port->queue_lock);
+	}
+}
+
+static void veth_timed_reset(unsigned long ptr)
+{
+	struct veth_lpar_connection *cnx = (struct veth_lpar_connection *)ptr;
+	unsigned long trigger_time, flags;
+
+	/* FIXME is it possible this fires after veth_stop_connection()?
+	 * That would reschedule the statemachine for 5 seconds and probably
+	 * execute it after the module's been unloaded. Hmm. */
+
+	spin_lock_irqsave(&cnx->lock, flags);
+
+	if (cnx->outstanding_tx > 0) {
+		trigger_time = cnx->last_contact + cnx->reset_timeout;
+
+		if (trigger_time < jiffies) {
+			cnx->state |= VETH_STATE_RESET;
+			veth_kick_statemachine(cnx);
+			veth_error("%d packets not acked by LPAR %d within %d "
+					"seconds, resetting.\n",
+					cnx->outstanding_tx, cnx->remote_lp,
+					cnx->reset_timeout / HZ);
+		} else {
+			/* Reschedule the timer */
+			trigger_time = jiffies + cnx->reset_timeout;
+			mod_timer(&cnx->reset_timer, trigger_time);
+		}
+	}
+
+	spin_unlock_irqrestore(&cnx->lock, flags);
+}
+
 /*
  * Rx path
  */
@@ -1117,12 +1358,9 @@
 	if ( (mac_addr == port->mac_addr) || (mac_addr == 0xffffffffffff0000) )
 		return 1;
 
-	if (! (((char *) &mac_addr)[0] & 0x01))
-		return 0;
-
 	read_lock_irqsave(&port->mcast_gate, flags);
 
-	if (port->promiscuous || port->all_mcast) {
+	if (port->promiscuous) {
 		wanted = 1;
 		goto out;
 	}
@@ -1175,21 +1413,21 @@
 {
 	HvLpEvent_Rc rc;
 
-	rc = veth_signaldata(cnx, VethEventTypeFramesAck,
+	rc = veth_signaldata(cnx, VETH_EVENT_FRAMES_ACK,
 			     0, &cnx->pending_acks);
 
 	if (rc != HvLpEvent_Rc_Good)
-		veth_error("Error 0x%x acking frames from lpar %d!\n",
-			   (unsigned)rc, cnx->remote_lp);
+		veth_error("Failed acking frames from LPAR %d, rc = %d\n",
+				cnx->remote_lp, (int)rc);
 
 	cnx->num_pending_acks = 0;
 	memset(&cnx->pending_acks, 0xff, sizeof(cnx->pending_acks));
 }
 
 static void veth_receive(struct veth_lpar_connection *cnx,
-			 struct VethLpEvent *event)
+			 struct veth_lpevent *event)
 {
-	struct VethFramesData *senddata = &event->u.frames_data;
+	struct veth_frames_data *senddata = &event->u.frames_data;
 	int startchunk = 0;
 	int nchunks;
 	unsigned long flags;
@@ -1216,9 +1454,10 @@
 		/* make sure that we have at least 1 EOF entry in the
 		 * remaining entries */
 		if (! (senddata->eofmask >> (startchunk + VETH_EOF_SHIFT))) {
-			veth_error("missing EOF frag in event "
-				   "eofmask=0x%x startchunk=%d\n",
-				   (unsigned) senddata->eofmask, startchunk);
+			veth_error("Missing EOF fragment in event "
+					"eofmask = 0x%x startchunk = %d\n",
+					(unsigned)senddata->eofmask,
+					startchunk);
 			break;
 		}
 
@@ -1237,8 +1476,9 @@
 		/* nchunks == # of chunks in this frame */
 
 		if ((length - ETH_HLEN) > VETH_MAX_MTU) {
-			veth_error("Received oversize frame from lpar %d "
-				   "(length=%d)\n", cnx->remote_lp, length);
+			veth_error("Received oversize frame from LPAR %d "
+					"(length = %d)\n",
+					cnx->remote_lp, length);
 			continue;
 		}
 
@@ -1331,15 +1571,33 @@
 
 static int veth_remove(struct vio_dev *vdev)
 {
-	int i = vdev->unit_address;
+	struct veth_lpar_connection *cnx;
 	struct net_device *dev;
+	struct veth_port *port;
+	int i;
 
-	dev = veth_dev[i];
-	if (dev != NULL) {
-		veth_dev[i] = NULL;
-		unregister_netdev(dev);
-		free_netdev(dev);
+	dev = veth_dev[vdev->unit_address];
+
+	if (! dev)
+		return 0;
+
+	port = netdev_priv(dev);
+
+	for (i = 0; i < HVMAXARCHITECTEDLPS; i++) {
+		cnx = veth_cnx[i];
+
+		if (cnx && (port->lpar_map & (1 << i))) {
+			/* Drop our reference to connections on our VLAN */
+			kobject_put(&cnx->kobject);
+		}
 	}
+
+	veth_dev[vdev->unit_address] = NULL;
+	kobject_del(&port->kobject);
+	kobject_put(&port->kobject);
+	unregister_netdev(dev);
+	free_netdev(dev);
+
 	return 0;
 }
 
@@ -1347,6 +1605,7 @@
 {
 	int i = vdev->unit_address;
 	struct net_device *dev;
+	struct veth_port *port;
 
 	dev = veth_probe_one(i, &vdev->dev);
 	if (dev == NULL) {
@@ -1355,11 +1614,23 @@
 	}
 	veth_dev[i] = dev;
 
-	/* Start the state machine on each connection, to commence
-	 * link negotiation */
-	for (i = 0; i < HVMAXARCHITECTEDLPS; i++)
-		if (veth_cnx[i])
-			veth_kick_statemachine(veth_cnx[i]);
+	port = (struct veth_port*)netdev_priv(dev);
+
+	/* Start the state machine on each connection on this vlan. If we're
+	 * the first dev to do so this will commence link negotiation */
+	for (i = 0; i < HVMAXARCHITECTEDLPS; i++) {
+		struct veth_lpar_connection *cnx;
+
+		if (! (port->lpar_map & (1 << i)))
+			continue;
+
+		cnx = veth_cnx[i];
+		if (!cnx)
+			continue;
+
+		kobject_get(&cnx->kobject);
+		veth_kick_statemachine(cnx);
+	}
 
 	return 0;
 }
@@ -1375,7 +1646,7 @@
 MODULE_DEVICE_TABLE(vio, veth_device_table);
 
 static struct vio_driver veth_driver = {
-	.name = "iseries_veth",
+	.name = DRV_NAME,
 	.id_table = veth_device_table,
 	.probe = veth_probe,
 	.remove = veth_remove
@@ -1388,29 +1659,29 @@
 void __exit veth_module_cleanup(void)
 {
 	int i;
+	struct veth_lpar_connection *cnx;
 
-	/* Stop the queues first to stop any new packets being sent. */
-	for (i = 0; i < HVMAXARCHITECTEDVIRTUALLANS; i++)
-		if (veth_dev[i])
-			netif_stop_queue(veth_dev[i]);
-
-	/* Stop the connections before we unregister the driver. This
-	 * ensures there's no skbs lying around holding the device open. */
-	for (i = 0; i < HVMAXARCHITECTEDLPS; ++i)
-		veth_stop_connection(i);
-
+	/* Disconnect our "irq" to stop events coming from the Hypervisor. */
 	HvLpEvent_unregisterHandler(HvLpEvent_Type_VirtualLan);
 
-	/* Hypervisor callbacks may have scheduled more work while we
-	 * were stoping connections. Now that we've disconnected from
-	 * the hypervisor make sure everything's finished. */
+	/* Make sure any work queued from Hypervisor callbacks is finished. */
 	flush_scheduled_work();
 
+	for (i = 0; i < HVMAXARCHITECTEDLPS; ++i) {
+		cnx = veth_cnx[i];
+
+		if (!cnx)
+			continue;
+
+		/* Remove the connection from sysfs */
+		kobject_del(&cnx->kobject);
+		/* Drop the driver's reference to the connection */
+		kobject_put(&cnx->kobject);
+	}
+
+	/* Unregister the driver, which will close all the netdevs and stop
+	 * the connections when they're no longer referenced. */
 	vio_unregister_driver(&veth_driver);
-
-	for (i = 0; i < HVMAXARCHITECTEDLPS; ++i)
-		veth_destroy_connection(i);
-
 }
 module_exit(veth_module_cleanup);
 
@@ -1423,15 +1694,37 @@
 
 	for (i = 0; i < HVMAXARCHITECTEDLPS; ++i) {
 		rc = veth_init_connection(i);
-		if (rc != 0) {
-			veth_module_cleanup();
-			return rc;
-		}
+		if (rc != 0)
+			goto error;
 	}
 
 	HvLpEvent_registerHandler(HvLpEvent_Type_VirtualLan,
 				  &veth_handle_event);
 
-	return vio_register_driver(&veth_driver);
+	rc = vio_register_driver(&veth_driver);
+	if (rc != 0)
+		goto error;
+
+	for (i = 0; i < HVMAXARCHITECTEDLPS; ++i) {
+		struct kobject *kobj;
+
+		if (!veth_cnx[i])
+			continue;
+
+		kobj = &veth_cnx[i]->kobject;
+		kobj->parent = &veth_driver.driver.kobj;
+		/* If the add failes, complain but otherwise continue */
+		if (0 != kobject_add(kobj))
+			veth_error("cnx %d: Failed adding to sysfs.\n", i);
+	}
+
+	return 0;
+
+error:
+	for (i = 0; i < HVMAXARCHITECTEDLPS; ++i) {
+		veth_destroy_connection(veth_cnx[i]);
+	}
+
+	return rc;
 }
 module_init(veth_module_init);
diff --git a/drivers/net/iseries_veth.h b/drivers/net/iseries_veth.h
deleted file mode 100644
index d9370f7..0000000
--- a/drivers/net/iseries_veth.h
+++ /dev/null
@@ -1,46 +0,0 @@
-/* File veth.h created by Kyle A. Lucke on Mon Aug  7 2000. */
-
-#ifndef _ISERIES_VETH_H
-#define _ISERIES_VETH_H
-
-#define VethEventTypeCap	(0)
-#define VethEventTypeFrames	(1)
-#define VethEventTypeMonitor	(2)
-#define VethEventTypeFramesAck	(3)
-
-#define VETH_MAX_ACKS_PER_MSG	(20)
-#define VETH_MAX_FRAMES_PER_MSG	(6)
-
-struct VethFramesData {
-	u32 addr[VETH_MAX_FRAMES_PER_MSG];
-	u16 len[VETH_MAX_FRAMES_PER_MSG];
-	u32 eofmask;
-};
-#define VETH_EOF_SHIFT		(32-VETH_MAX_FRAMES_PER_MSG)
-
-struct VethFramesAckData {
-	u16 token[VETH_MAX_ACKS_PER_MSG];
-};
-
-struct VethCapData {
-	u8 caps_version;
-	u8 rsvd1;
-	u16 num_buffers;
-	u16 ack_threshold;
-	u16 rsvd2;
-	u32 ack_timeout;
-	u32 rsvd3;
-	u64 rsvd4[3];
-};
-
-struct VethLpEvent {
-	struct HvLpEvent base_event;
-	union {
-		struct VethCapData caps_data;
-		struct VethFramesData frames_data;
-		struct VethFramesAckData frames_ack_data;
-	} u;
-
-};
-
-#endif	/* _ISERIES_VETH_H */
diff --git a/drivers/net/tulip/de2104x.c b/drivers/net/tulip/de2104x.c
index fc353e3..a22d001 100644
--- a/drivers/net/tulip/de2104x.c
+++ b/drivers/net/tulip/de2104x.c
@@ -1934,7 +1934,7 @@
 	struct de_private *de;
 	int rc;
 	void __iomem *regs;
-	long pciaddr;
+	unsigned long pciaddr;
 	static int board_idx = -1;
 
 	board_idx++;
diff --git a/drivers/net/tulip/tulip_core.c b/drivers/net/tulip/tulip_core.c
index 05da5be..6266a9a 100644
--- a/drivers/net/tulip/tulip_core.c
+++ b/drivers/net/tulip/tulip_core.c
@@ -238,6 +238,7 @@
 	{ 0x17B3, 0xAB08, PCI_ANY_ID, PCI_ANY_ID, 0, 0, COMET },
 	{ 0x10b7, 0x9300, PCI_ANY_ID, PCI_ANY_ID, 0, 0, COMET }, /* 3Com 3CSOHO100B-TX */
 	{ 0x14ea, 0xab08, PCI_ANY_ID, PCI_ANY_ID, 0, 0, COMET }, /* Planex FNW-3602-TX */
+	{ 0x1414, 0x0002, PCI_ANY_ID, PCI_ANY_ID, 0, 0, COMET },
 	{ } /* terminate list */
 };
 MODULE_DEVICE_TABLE(pci, tulip_pci_tbl);
diff --git a/drivers/pci/rom.c b/drivers/pci/rom.c
index 713c78f..49bd217 100644
--- a/drivers/pci/rom.c
+++ b/drivers/pci/rom.c
@@ -21,13 +21,21 @@
  * between the ROM and other resources, so enabling it may disable access
  * to MMIO registers or other card memory.
  */
-static void pci_enable_rom(struct pci_dev *pdev)
+static int pci_enable_rom(struct pci_dev *pdev)
 {
+	struct resource *res = pdev->resource + PCI_ROM_RESOURCE;
+	struct pci_bus_region region;
 	u32 rom_addr;
 
+	if (!res->flags)
+		return -1;
+
+	pcibios_resource_to_bus(pdev, &region, res);
 	pci_read_config_dword(pdev, pdev->rom_base_reg, &rom_addr);
-	rom_addr |= PCI_ROM_ADDRESS_ENABLE;
+	rom_addr &= ~PCI_ROM_ADDRESS_MASK;
+	rom_addr |= region.start | PCI_ROM_ADDRESS_ENABLE;
 	pci_write_config_dword(pdev, pdev->rom_base_reg, rom_addr);
+	return 0;
 }
 
 /**
@@ -71,19 +79,21 @@
 	} else {
 		if (res->flags & IORESOURCE_ROM_COPY) {
 			*size = pci_resource_len(pdev, PCI_ROM_RESOURCE);
-			return (void __iomem *)pci_resource_start(pdev, PCI_ROM_RESOURCE);
+			return (void __iomem *)pci_resource_start(pdev,
+							     PCI_ROM_RESOURCE);
 		} else {
 			/* assign the ROM an address if it doesn't have one */
-			if (res->parent == NULL)
-				pci_assign_resource(pdev, PCI_ROM_RESOURCE);
-
+			if (res->parent == NULL &&
+			    pci_assign_resource(pdev,PCI_ROM_RESOURCE))
+				return NULL;
 			start = pci_resource_start(pdev, PCI_ROM_RESOURCE);
 			*size = pci_resource_len(pdev, PCI_ROM_RESOURCE);
 			if (*size == 0)
 				return NULL;
 
 			/* Enable ROM space decodes */
-			pci_enable_rom(pdev);
+			if (pci_enable_rom(pdev))
+				return NULL;
 		}
 	}
 
diff --git a/include/asm-arm/arch-ixp4xx/io.h b/include/asm-arm/arch-ixp4xx/io.h
index 7495026..e350dcb 100644
--- a/include/asm-arm/arch-ixp4xx/io.h
+++ b/include/asm-arm/arch-ixp4xx/io.h
@@ -383,39 +383,45 @@
 		*vaddr++ = inl(io_addr);
 }
 
-#define	__is_io_address(p)	(((unsigned long)p >= 0x0) && \
-					((unsigned long)p <= 0x0000ffff))
+#define PIO_OFFSET      0x10000UL
+#define PIO_MASK        0x0ffffUL
+
+#define	__is_io_address(p)	(((unsigned long)p >= PIO_OFFSET) && \
+					((unsigned long)p <= (PIO_MASK + PIO_OFFSET)))
 static inline unsigned int
-__ixp4xx_ioread8(void __iomem *port)
+__ixp4xx_ioread8(void __iomem *addr)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		return	(unsigned int)__ixp4xx_inb((unsigned int)port);
+		return	(unsigned int)__ixp4xx_inb(port & PIO_MASK);
 	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
-		return (unsigned int)__raw_readb((u32)port);
+		return (unsigned int)__raw_readb(port);
 #else
-		return (unsigned int)__ixp4xx_readb((u32)port);
+		return (unsigned int)__ixp4xx_readb(port);
 #endif
 }
 
 static inline void
-__ixp4xx_ioread8_rep(u32 port, u8 *vaddr, u32 count)
+__ixp4xx_ioread8_rep(void __iomem *addr, void *vaddr, u32 count)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_insb(port, vaddr, count);
+		__ixp4xx_insb(port & PIO_MASK, vaddr, count);
 	else
 #ifndef	CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_readsb((void __iomem *)port, vaddr, count);
+		__raw_readsb(addr, vaddr, count);
 #else
 		__ixp4xx_readsb(port, vaddr, count);
 #endif
 }
 
 static inline unsigned int
-__ixp4xx_ioread16(void __iomem *port)
+__ixp4xx_ioread16(void __iomem *addr)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		return	(unsigned int)__ixp4xx_inw((unsigned int)port);
+		return	(unsigned int)__ixp4xx_inw(port & PIO_MASK);
 	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
 		return le16_to_cpu(__raw_readw((u32)port));
@@ -425,23 +431,25 @@
 }
 
 static inline void
-__ixp4xx_ioread16_rep(u32 port, u16 *vaddr, u32 count)
+__ixp4xx_ioread16_rep(void __iomem *addr, void *vaddr, u32 count)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_insw(port, vaddr, count);
+		__ixp4xx_insw(port & PIO_MASK, vaddr, count);
 	else
 #ifndef	CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_readsw((void __iomem *)port, vaddr, count);
+		__raw_readsw(addr, vaddr, count);
 #else
 		__ixp4xx_readsw(port, vaddr, count);
 #endif
 }
 
 static inline unsigned int
-__ixp4xx_ioread32(void __iomem *port)
+__ixp4xx_ioread32(void __iomem *addr)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		return	(unsigned int)__ixp4xx_inl((unsigned int)port);
+		return	(unsigned int)__ixp4xx_inl(port & PIO_MASK);
 	else {
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
 		return le32_to_cpu(__raw_readl((u32)port));
@@ -452,90 +460,100 @@
 }
 
 static inline void
-__ixp4xx_ioread32_rep(u32 port, u32 *vaddr, u32 count)
+__ixp4xx_ioread32_rep(void __iomem *addr, void *vaddr, u32 count)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_insl(port, vaddr, count);
+		__ixp4xx_insl(port & PIO_MASK, vaddr, count);
 	else
 #ifndef	CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_readsl((void __iomem *)port, vaddr, count);
+		__raw_readsl(addr, vaddr, count);
 #else
 		__ixp4xx_readsl(port, vaddr, count);
 #endif
 }
 
 static inline void
-__ixp4xx_iowrite8(u8 value, void __iomem *port)
+__ixp4xx_iowrite8(u8 value, void __iomem *addr)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_outb(value, (unsigned int)port);
+		__ixp4xx_outb(value, port & PIO_MASK);
 	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_writeb(value, (u32)port);
+		__raw_writeb(value, port);
 #else
-		__ixp4xx_writeb(value, (u32)port);
+		__ixp4xx_writeb(value, port);
 #endif
 }
 
 static inline void
-__ixp4xx_iowrite8_rep(u32 port, u8 *vaddr, u32 count)
+__ixp4xx_iowrite8_rep(void __iomem *addr, const void *vaddr, u32 count)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_outsb(port, vaddr, count);
+		__ixp4xx_outsb(port & PIO_MASK, vaddr, count);
+	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_writesb((void __iomem *)port, vaddr, count);
+		__raw_writesb(addr, vaddr, count);
 #else
 		__ixp4xx_writesb(port, vaddr, count);
 #endif
 }
 
 static inline void
-__ixp4xx_iowrite16(u16 value, void __iomem *port)
+__ixp4xx_iowrite16(u16 value, void __iomem *addr)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_outw(value, (unsigned int)port);
+		__ixp4xx_outw(value, port & PIO_MASK);
 	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_writew(cpu_to_le16(value), (u32)port);
+		__raw_writew(cpu_to_le16(value), addr);
 #else
-		__ixp4xx_writew(value, (u32)port);
+		__ixp4xx_writew(value, port);
 #endif
 }
 
 static inline void
-__ixp4xx_iowrite16_rep(u32 port, u16 *vaddr, u32 count)
+__ixp4xx_iowrite16_rep(void __iomem *addr, const void *vaddr, u32 count)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_outsw(port, vaddr, count);
+		__ixp4xx_outsw(port & PIO_MASK, vaddr, count);
+	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_readsw((void __iomem *)port, vaddr, count);
+		__raw_writesw(addr, vaddr, count);
 #else
 		__ixp4xx_writesw(port, vaddr, count);
 #endif
 }
 
 static inline void
-__ixp4xx_iowrite32(u32 value, void __iomem *port)
+__ixp4xx_iowrite32(u32 value, void __iomem *addr)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_outl(value, (unsigned int)port);
+		__ixp4xx_outl(value, port & PIO_MASK);
 	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_writel(cpu_to_le32(value), (u32)port);
+		__raw_writel(cpu_to_le32(value), port);
 #else
-		__ixp4xx_writel(value, (u32)port);
+		__ixp4xx_writel(value, port);
 #endif
 }
 
 static inline void
-__ixp4xx_iowrite32_rep(u32 port, u32 *vaddr, u32 count)
+__ixp4xx_iowrite32_rep(void __iomem *addr, const void *vaddr, u32 count)
 {
+	unsigned long port = (unsigned long __force)addr;
 	if (__is_io_address(port))
-		__ixp4xx_outsl(port, vaddr, count);
+		__ixp4xx_outsl(port & PIO_MASK, vaddr, count);
+	else
 #ifndef CONFIG_IXP4XX_INDIRECT_PCI
-		__raw_readsl((void __iomem *)port, vaddr, count);
+		__raw_writesl(addr, vaddr, count);
 #else
-		__ixp4xx_outsl(port, vaddr, count);
+		__ixp4xx_writesl(port, vaddr, count);
 #endif
 }
 
@@ -555,7 +573,7 @@
 #define	iowrite16_rep(p, v, c)		__ixp4xx_iowrite16_rep(p, v, c)
 #define	iowrite32_rep(p, v, c)		__ixp4xx_iowrite32_rep(p, v, c)
 
-#define	ioport_map(port, nr)		((void __iomem*)port)
+#define	ioport_map(port, nr)		((void __iomem*)(port + PIO_OFFSET))
 #define	ioport_unmap(addr)
 
 #endif	//  __ASM_ARM_ARCH_IO_H
diff --git a/include/asm-arm/arch-ixp4xx/platform.h b/include/asm-arm/arch-ixp4xx/platform.h
index 3a626c0..d13ee7f 100644
--- a/include/asm-arm/arch-ixp4xx/platform.h
+++ b/include/asm-arm/arch-ixp4xx/platform.h
@@ -83,17 +83,6 @@
 #define IXP4XX_GPIO_OUT 		0x1
 #define IXP4XX_GPIO_IN  		0x2
 
-#define IXP4XX_GPIO_INTSTYLE_MASK	0x7C  /* Bits [6:2] define interrupt style */
-
-/* 
- * GPIO interrupt types.
- */
-#define IXP4XX_GPIO_ACTIVE_HIGH		0x4 /* Default */
-#define IXP4XX_GPIO_ACTIVE_LOW		0x8
-#define IXP4XX_GPIO_RISING_EDGE		0x10
-#define IXP4XX_GPIO_FALLING_EDGE 	0x20
-#define IXP4XX_GPIO_TRANSITIONAL 	0x40
-
 /* GPIO signal types */
 #define IXP4XX_GPIO_LOW			0
 #define IXP4XX_GPIO_HIGH		1
@@ -102,7 +91,13 @@
 #define IXP4XX_GPIO_CLK_0		14
 #define IXP4XX_GPIO_CLK_1		15
 
-extern void gpio_line_config(u8 line, u32 style);
+static inline void gpio_line_config(u8 line, u32 direction)
+{
+	if (direction == IXP4XX_GPIO_OUT)
+		*IXP4XX_GPIO_GPOER |= (1 << line);
+	else
+		*IXP4XX_GPIO_GPOER &= ~(1 << line);
+}
 
 static inline void gpio_line_get(u8 line, int *value)
 {
diff --git a/include/asm-arm/arch-pxa/pxa-regs.h b/include/asm-arm/arch-pxa/pxa-regs.h
index 51f0fe0..939d9e5 100644
--- a/include/asm-arm/arch-pxa/pxa-regs.h
+++ b/include/asm-arm/arch-pxa/pxa-regs.h
@@ -818,6 +818,23 @@
 #define UDCOTGICR_IEIDF	(1 << 0)	/* OTG ID Change Falling Edge
 					   Interrupt Enable */
 
+#define UP2OCR		  __REG(0x40600020)  /* USB Port 2 Output Control register */
+
+#define UP2OCR_CPVEN	(1 << 0)	/* Charge Pump Vbus Enable */
+#define UP2OCR_CPVPE	(1 << 1)	/* Charge Pump Vbus Pulse Enable */
+#define UP2OCR_DPPDE	(1 << 2)	/* Host Port 2 Transceiver D+ Pull Down Enable */
+#define UP2OCR_DMPDE	(1 << 3)	/* Host Port 2 Transceiver D- Pull Down Enable */
+#define UP2OCR_DPPUE	(1 << 4)	/* Host Port 2 Transceiver D+ Pull Up Enable */
+#define UP2OCR_DMPUE	(1 << 5)	/* Host Port 2 Transceiver D- Pull Up Enable */
+#define UP2OCR_DPPUBE	(1 << 6)	/* Host Port 2 Transceiver D+ Pull Up Bypass Enable */
+#define UP2OCR_DMPUBE	(1 << 7)	/* Host Port 2 Transceiver D- Pull Up Bypass Enable */
+#define UP2OCR_EXSP		(1 << 8)	/* External Transceiver Speed Control */
+#define UP2OCR_EXSUS	(1 << 9)	/* External Transceiver Speed Enable */
+#define UP2OCR_IDON		(1 << 10)	/* OTG ID Read Enable */
+#define UP2OCR_HXS		(1 << 16)	/* Host Port 2 Transceiver Output Select */
+#define UP2OCR_HXOE		(1 << 17)	/* Host Port 2 Transceiver Output Enable */
+#define UP2OCR_SEOS		(1 << 24)	/* Single-Ended Output Select */
+
 #define UDCCSN(x)	__REG2(0x40600100, (x) << 2)
 #define UDCCSR0         __REG(0x40600100) /* UDC Control/Status register - Endpoint 0 */
 #define UDCCSR0_SA	(1 << 7)	/* Setup Active */
@@ -1423,6 +1440,7 @@
 #define GPIO84_NSSP_RX      	(84 | GPIO_ALT_FN_2_IN)
 #define GPIO85_nPCE_1_MD	(85 | GPIO_ALT_FN_1_OUT)
 #define GPIO92_MMCDAT0_MD	(92 | GPIO_ALT_FN_1_OUT)
+#define GPIO104_pSKTSEL_MD	(104 | GPIO_ALT_FN_1_OUT)
 #define GPIO109_MMCDAT1_MD	(109 | GPIO_ALT_FN_1_OUT)
 #define GPIO110_MMCDAT2_MD	(110 | GPIO_ALT_FN_1_OUT)
 #define GPIO110_MMCCS0_MD	(110 | GPIO_ALT_FN_1_OUT)
@@ -1510,6 +1528,8 @@
 #define PSSR_BFS	(1 << 1)	/* Battery Fault Status */
 #define PSSR_SSS	(1 << 0)	/* Software Sleep Status */
 
+#define PSLR_SL_ROD	(1 << 20)	/* Sleep-Mode/Depp-Sleep Mode nRESET_OUT Disable */
+
 #define PCFR_RO		(1 << 15)	/* RDH Override */
 #define PCFR_PO		(1 << 14)	/* PH Override */
 #define PCFR_GPROD	(1 << 12)	/* GPIO nRESET_OUT Disable */
@@ -1517,6 +1537,7 @@
 #define PCFR_FVC	(1 << 10)	/* Frequency/Voltage Change */
 #define PCFR_DC_EN	(1 << 7)	/* Sleep/deep-sleep DC-DC Converter Enable */
 #define PCFR_PI2CEN	(1 << 6)	/* Enable PI2C controller */
+#define PCFR_GPR_EN	(1 << 4)	/* nRESET_GPIO Pin Enable */
 #define PCFR_DS		(1 << 3)	/* Deep Sleep Mode */
 #define PCFR_FS		(1 << 2)	/* Float Static Chip Selects */
 #define PCFR_FP		(1 << 1)	/* Float PCMCIA controls */
@@ -1810,6 +1831,11 @@
 #define LCCR0_PDD_S	12
 #define LCCR0_BM	(1 << 20) 	/* Branch mask */
 #define LCCR0_OUM	(1 << 21)	/* Output FIFO underrun mask */
+#define LCCR0_LCDT      (1 << 22)       /* LCD panel type */
+#define LCCR0_RDSTM     (1 << 23)       /* Read status interrupt mask */
+#define LCCR0_CMDIM     (1 << 24)       /* Command interrupt mask */
+#define LCCR0_OUC       (1 << 25)       /* Overlay Underlay control bit */
+#define LCCR0_LDDALT    (1 << 26)       /* LDD alternate mapping control */
 
 #define LCCR1_PPL       Fld (10, 0)      /* Pixels Per Line - 1 */
 #define LCCR1_DisWdth(Pixel)            /* Display Width [1..800 pix.]  */ \
@@ -2062,7 +2088,10 @@
 #define UHCFMN		__REG(0x4C00003C) /* UHC Frame Number */
 #define UHCPERS		__REG(0x4C000040) /* UHC Periodic Start */
 #define UHCLS		__REG(0x4C000044) /* UHC Low Speed Threshold */
+
 #define UHCRHDA		__REG(0x4C000048) /* UHC Root Hub Descriptor A */
+#define UHCRHDA_NOCP	(1 << 12)	/* No over current protection */
+
 #define UHCRHDB		__REG(0x4C00004C) /* UHC Root Hub Descriptor B */
 #define UHCRHS		__REG(0x4C000050) /* UHC Root Hub Status */
 #define UHCRHPS1	__REG(0x4C000054) /* UHC Root Hub Port 1 Status */
diff --git a/include/asm-arm/arch-s3c2410/regs-clock.h b/include/asm-arm/arch-s3c2410/regs-clock.h
index e5e938b..16f4c3c 100644
--- a/include/asm-arm/arch-s3c2410/regs-clock.h
+++ b/include/asm-arm/arch-s3c2410/regs-clock.h
@@ -1,7 +1,7 @@
 /* linux/include/asm/arch-s3c2410/regs-clock.h
  *
- * Copyright (c) 2003,2004 Simtec Electronics <linux@simtec.co.uk>
- *		      http://www.simtec.co.uk/products/SWLINUX/
+ * Copyright (c) 2003,2004,2005 Simtec Electronics <linux@simtec.co.uk>
+ *		      http://armlinux.simtec.co.uk/
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 2 as
@@ -17,6 +17,7 @@
  *    29-Sep-2004 Ben Dooks	    Fixed usage for assembly inclusion
  *    10-Feb-2005 Ben Dooks	    Fixed CAMDIVN address (Guillaume Gourat)
  *    10-Mar-2005 Lucas Villa Real  Changed S3C2410_VA to S3C24XX_VA
+ *    27-Aug-2005 Ben Dooks	    Add clock-slow info
  */
 
 #ifndef __ASM_ARM_REGS_CLOCK
@@ -74,6 +75,12 @@
 #define S3C2410_CLKDIVN_PDIVN	     (1<<0)
 #define S3C2410_CLKDIVN_HDIVN	     (1<<1)
 
+#define S3C2410_CLKSLOW_UCLK_OFF	(1<<7)
+#define S3C2410_CLKSLOW_MPLL_OFF	(1<<5)
+#define S3C2410_CLKSLOW_SLOW		(1<<4)
+#define S3C2410_CLKSLOW_SLOWVAL(x)	(x)
+#define S3C2410_CLKSLOW_GET_SLOWVAL(x)	((x) & 7)
+
 #ifndef __ASSEMBLY__
 
 static inline unsigned int
diff --git a/include/asm-arm/unistd.h b/include/asm-arm/unistd.h
index abb36e5..94619cc 100644
--- a/include/asm-arm/unistd.h
+++ b/include/asm-arm/unistd.h
@@ -515,7 +515,6 @@
 #define __ARCH_WANT_SYS_TIME
 #define __ARCH_WANT_SYS_UTIME
 #define __ARCH_WANT_SYS_SOCKETCALL
-#define __ARCH_WANT_SYS_FADVISE64
 #define __ARCH_WANT_SYS_GETPGRP
 #define __ARCH_WANT_SYS_LLSEEK
 #define __ARCH_WANT_SYS_NICE
diff --git a/include/asm-ia64/acpi.h b/include/asm-ia64/acpi.h
index 4c06d45..3a544ff 100644
--- a/include/asm-ia64/acpi.h
+++ b/include/asm-ia64/acpi.h
@@ -116,6 +116,11 @@
 
 extern u16 ia64_acpiid_to_sapicid[];
 
+/*
+ * Refer Intel ACPI _PDC support document for bit definitions
+ */
+#define ACPI_PDC_EST_CAPABILITY_SMP     0x8
+
 #endif /*__KERNEL__*/
 
 #endif /*_ASM_ACPI_H*/
diff --git a/include/asm-ia64/fcntl.h b/include/asm-ia64/fcntl.h
index c9f8d83..cee16ea 100644
--- a/include/asm-ia64/fcntl.h
+++ b/include/asm-ia64/fcntl.h
@@ -81,6 +81,7 @@
 
 #define F_LINUX_SPECIFIC_BASE	1024
 
-#define force_o_largefile() ( ! (current->personality & PER_LINUX32) )
+#define force_o_largefile()	\
+		(personality(current->personality) != PER_LINUX32)
 
 #endif /* _ASM_IA64_FCNTL_H */
diff --git a/include/asm-ia64/io.h b/include/asm-ia64/io.h
index 54e7637..cf772a6 100644
--- a/include/asm-ia64/io.h
+++ b/include/asm-ia64/io.h
@@ -23,7 +23,7 @@
 #define __SLOW_DOWN_IO	do { } while (0)
 #define SLOW_DOWN_IO	do { } while (0)
 
-#define __IA64_UNCACHED_OFFSET	0xc000000000000000UL	/* region 6 */
+#define __IA64_UNCACHED_OFFSET	RGN_BASE(RGN_UNCACHED)
 
 /*
  * The legacy I/O space defined by the ia64 architecture supports only 65536 ports, but
@@ -41,7 +41,7 @@
 #define IO_SPACE_BASE(space)		((space) << IO_SPACE_BITS)
 #define IO_SPACE_PORT(port)		((port) & (IO_SPACE_SIZE - 1))
 
-#define IO_SPACE_SPARSE_ENCODING(p)	((((p) >> 2) << 12) | (p & 0xfff))
+#define IO_SPACE_SPARSE_ENCODING(p)	((((p) >> 2) << 12) | ((p) & 0xfff))
 
 struct io_space {
 	unsigned long mmio_base;	/* base in MMIO space */
diff --git a/include/asm-ia64/mmu.h b/include/asm-ia64/mmu.h
index ae15253..611432b 100644
--- a/include/asm-ia64/mmu.h
+++ b/include/asm-ia64/mmu.h
@@ -2,10 +2,12 @@
 #define __MMU_H
 
 /*
- * Type for a context number.  We declare it volatile to ensure proper ordering when it's
- * accessed outside of spinlock'd critical sections (e.g., as done in activate_mm() and
- * init_new_context()).
+ * Type for a context number.  We declare it volatile to ensure proper
+ * ordering when it's accessed outside of spinlock'd critical sections
+ * (e.g., as done in activate_mm() and init_new_context()).
  */
 typedef volatile unsigned long mm_context_t;
 
+typedef unsigned long nv_mm_context_t;
+
 #endif
diff --git a/include/asm-ia64/mmu_context.h b/include/asm-ia64/mmu_context.h
index e3e5fed..8d6e72f 100644
--- a/include/asm-ia64/mmu_context.h
+++ b/include/asm-ia64/mmu_context.h
@@ -19,6 +19,7 @@
 
 #define ia64_rid(ctx,addr)	(((ctx) << 3) | (addr >> 61))
 
+# include <asm/page.h>
 # ifndef __ASSEMBLY__
 
 #include <linux/compiler.h>
@@ -55,34 +56,46 @@
 delayed_tlb_flush (void)
 {
 	extern void local_flush_tlb_all (void);
+	unsigned long flags;
 
 	if (unlikely(__ia64_per_cpu_var(ia64_need_tlb_flush))) {
-		local_flush_tlb_all();
-		__ia64_per_cpu_var(ia64_need_tlb_flush) = 0;
+		spin_lock_irqsave(&ia64_ctx.lock, flags);
+		{
+			if (__ia64_per_cpu_var(ia64_need_tlb_flush)) {
+				local_flush_tlb_all();
+				__ia64_per_cpu_var(ia64_need_tlb_flush) = 0;
+			}
+		}
+		spin_unlock_irqrestore(&ia64_ctx.lock, flags);
 	}
 }
 
-static inline mm_context_t
+static inline nv_mm_context_t
 get_mmu_context (struct mm_struct *mm)
 {
 	unsigned long flags;
-	mm_context_t context = mm->context;
+	nv_mm_context_t context = mm->context;
 
-	if (context)
-		return context;
-
-	spin_lock_irqsave(&ia64_ctx.lock, flags);
-	{
-		/* re-check, now that we've got the lock: */
-		context = mm->context;
-		if (context == 0) {
-			cpus_clear(mm->cpu_vm_mask);
-			if (ia64_ctx.next >= ia64_ctx.limit)
-				wrap_mmu_context(mm);
-			mm->context = context = ia64_ctx.next++;
+	if (unlikely(!context)) {
+		spin_lock_irqsave(&ia64_ctx.lock, flags);
+		{
+			/* re-check, now that we've got the lock: */
+			context = mm->context;
+			if (context == 0) {
+				cpus_clear(mm->cpu_vm_mask);
+				if (ia64_ctx.next >= ia64_ctx.limit)
+					wrap_mmu_context(mm);
+				mm->context = context = ia64_ctx.next++;
+			}
 		}
+		spin_unlock_irqrestore(&ia64_ctx.lock, flags);
 	}
-	spin_unlock_irqrestore(&ia64_ctx.lock, flags);
+	/*
+	 * Ensure we're not starting to use "context" before any old
+	 * uses of it are gone from our TLB.
+	 */
+	delayed_tlb_flush();
+
 	return context;
 }
 
@@ -104,13 +117,13 @@
 }
 
 static inline void
-reload_context (mm_context_t context)
+reload_context (nv_mm_context_t context)
 {
 	unsigned long rid;
 	unsigned long rid_incr = 0;
 	unsigned long rr0, rr1, rr2, rr3, rr4, old_rr4;
 
-	old_rr4 = ia64_get_rr(0x8000000000000000UL);
+	old_rr4 = ia64_get_rr(RGN_BASE(RGN_HPAGE));
 	rid = context << 3;	/* make space for encoding the region number */
 	rid_incr = 1 << 8;
 
@@ -122,6 +135,10 @@
 	rr4 = rr0 + 4*rid_incr;
 #ifdef  CONFIG_HUGETLB_PAGE
 	rr4 = (rr4 & (~(0xfcUL))) | (old_rr4 & 0xfc);
+
+#  if RGN_HPAGE != 4
+#    error "reload_context assumes RGN_HPAGE is 4"
+#  endif
 #endif
 
 	ia64_set_rr(0x0000000000000000UL, rr0);
@@ -138,7 +155,7 @@
 static inline void
 activate_context (struct mm_struct *mm)
 {
-	mm_context_t context;
+	nv_mm_context_t context;
 
 	do {
 		context = get_mmu_context(mm);
@@ -157,8 +174,6 @@
 static inline void
 activate_mm (struct mm_struct *prev, struct mm_struct *next)
 {
-	delayed_tlb_flush();
-
 	/*
 	 * We may get interrupts here, but that's OK because interrupt handlers cannot
 	 * touch user-space.
diff --git a/include/asm-ia64/page.h b/include/asm-ia64/page.h
index 08894f7..9edffad 100644
--- a/include/asm-ia64/page.h
+++ b/include/asm-ia64/page.h
@@ -13,6 +13,19 @@
 #include <asm/types.h>
 
 /*
+ * The top three bits of an IA64 address are its Region Number.
+ * Different regions are assigned to different purposes.
+ */
+#define RGN_SHIFT	(61)
+#define RGN_BASE(r)	(__IA64_UL_CONST(r)<<RGN_SHIFT)
+#define RGN_BITS	(RGN_BASE(-1))
+
+#define RGN_KERNEL	7	/* Identity mapped region */
+#define RGN_UNCACHED    6	/* Identity mapped I/O region */
+#define RGN_GATE	5	/* Gate page, Kernel text, etc */
+#define RGN_HPAGE	4	/* For Huge TLB pages */
+
+/*
  * PAGE_SHIFT determines the actual kernel page size.
  */
 #if defined(CONFIG_IA64_PAGE_SIZE_4KB)
@@ -36,10 +49,9 @@
 
 #define RGN_MAP_LIMIT	((1UL << (4*PAGE_SHIFT - 12)) - PAGE_SIZE)	/* per region addr limit */
 
+
 #ifdef CONFIG_HUGETLB_PAGE
-# define REGION_HPAGE		(4UL)	/* note: this is hardcoded in reload_context()!*/
-# define REGION_SHIFT		61
-# define HPAGE_REGION_BASE	(REGION_HPAGE << REGION_SHIFT)
+# define HPAGE_REGION_BASE	RGN_BASE(RGN_HPAGE)
 # define HPAGE_SHIFT		hpage_shift
 # define HPAGE_SHIFT_DEFAULT	28	/* check ia64 SDM for architecture supported size */
 # define HPAGE_SIZE		(__IA64_UL_CONST(1) << HPAGE_SHIFT)
@@ -130,16 +142,13 @@
 #define REGION_NUMBER(x)	({ia64_va _v; _v.l = (long) (x); _v.f.reg;})
 #define REGION_OFFSET(x)	({ia64_va _v; _v.l = (long) (x); _v.f.off;})
 
-#define REGION_SIZE		REGION_NUMBER(1)
-#define REGION_KERNEL		7
-
 #ifdef CONFIG_HUGETLB_PAGE
 # define htlbpage_to_page(x)	(((unsigned long) REGION_NUMBER(x) << 61)			\
 				 | (REGION_OFFSET(x) >> (HPAGE_SHIFT-PAGE_SHIFT)))
 # define HUGETLB_PAGE_ORDER	(HPAGE_SHIFT - PAGE_SHIFT)
 # define is_hugepage_only_range(mm, addr, len)		\
-	 (REGION_NUMBER(addr) == REGION_HPAGE &&	\
-	  REGION_NUMBER((addr)+(len)-1) == REGION_HPAGE)
+	 (REGION_NUMBER(addr) == RGN_HPAGE &&	\
+	  REGION_NUMBER((addr)+(len)-1) == RGN_HPAGE)
 extern unsigned int hpage_shift;
 #endif
 
@@ -197,7 +206,7 @@
 # define __pgprot(x)	(x)
 #endif /* !STRICT_MM_TYPECHECKS */
 
-#define PAGE_OFFSET			__IA64_UL_CONST(0xe000000000000000)
+#define PAGE_OFFSET			RGN_BASE(RGN_KERNEL)
 
 #define VM_DATA_DEFAULT_FLAGS		(VM_READ | VM_WRITE |					\
 					 VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC |		\
diff --git a/include/asm-ia64/pal.h b/include/asm-ia64/pal.h
index 2303a10..e828377 100644
--- a/include/asm-ia64/pal.h
+++ b/include/asm-ia64/pal.h
@@ -75,6 +75,8 @@
 #define PAL_CACHE_READ		259	/* read tag & data of cacheline for diagnostic testing */
 #define PAL_CACHE_WRITE		260	/* write tag & data of cacheline for diagnostic testing */
 #define PAL_VM_TR_READ		261	/* read contents of translation register */
+#define PAL_GET_PSTATE		262	/* get the current P-state */
+#define PAL_SET_PSTATE		263	/* set the P-state */
 
 #ifndef __ASSEMBLY__
 
@@ -1111,6 +1113,25 @@
 	return iprv.status;
 }
 
+/* Get the current P-state information */
+static inline s64
+ia64_pal_get_pstate (u64 *pstate_index)
+{
+	struct ia64_pal_retval iprv;
+	PAL_CALL_STK(iprv, PAL_GET_PSTATE, 0, 0, 0);
+	*pstate_index = iprv.v0;
+	return iprv.status;
+}
+
+/* Set the P-state */
+static inline s64
+ia64_pal_set_pstate (u64 pstate_index)
+{
+	struct ia64_pal_retval iprv;
+	PAL_CALL_STK(iprv, PAL_SET_PSTATE, pstate_index, 0, 0);
+	return iprv.status;
+}
+
 /* Cause the processor to enter LIGHT HALT state, where prefetching and execution are
  * suspended, but cache and TLB coherency is maintained.
  */
diff --git a/include/asm-ia64/pgtable.h b/include/asm-ia64/pgtable.h
index 48586e0..2e34c06 100644
--- a/include/asm-ia64/pgtable.h
+++ b/include/asm-ia64/pgtable.h
@@ -204,21 +204,18 @@
 #define set_pte(ptep, pteval)	(*(ptep) = (pteval))
 #define set_pte_at(mm,addr,ptep,pteval) set_pte(ptep,pteval)
 
-#define RGN_SIZE	(1UL << 61)
-#define RGN_KERNEL	7
-
-#define VMALLOC_START		0xa000000200000000UL
+#define VMALLOC_START		(RGN_BASE(RGN_GATE) + 0x200000000UL)
 #ifdef CONFIG_VIRTUAL_MEM_MAP
-# define VMALLOC_END_INIT	(0xa000000000000000UL + (1UL << (4*PAGE_SHIFT - 9)))
+# define VMALLOC_END_INIT	(RGN_BASE(RGN_GATE) + (1UL << (4*PAGE_SHIFT - 9)))
 # define VMALLOC_END		vmalloc_end
   extern unsigned long vmalloc_end;
 #else
-# define VMALLOC_END		(0xa000000000000000UL + (1UL << (4*PAGE_SHIFT - 9)))
+# define VMALLOC_END		(RGN_BASE(RGN_GATE) + (1UL << (4*PAGE_SHIFT - 9)))
 #endif
 
 /* fs/proc/kcore.c */
-#define	kc_vaddr_to_offset(v) ((v) - 0xa000000000000000UL)
-#define	kc_offset_to_vaddr(o) ((o) + 0xa000000000000000UL)
+#define	kc_vaddr_to_offset(v) ((v) - RGN_BASE(RGN_GATE))
+#define	kc_offset_to_vaddr(o) ((o) + RGN_BASE(RGN_GATE))
 
 /*
  * Conversion functions: convert page frame number (pfn) and a protection value to a page
diff --git a/include/asm-ia64/rwsem.h b/include/asm-ia64/rwsem.h
index 6ece506..e18b5ab 100644
--- a/include/asm-ia64/rwsem.h
+++ b/include/asm-ia64/rwsem.h
@@ -3,6 +3,7 @@
  *
  * Copyright (C) 2003 Ken Chen <kenneth.w.chen@intel.com>
  * Copyright (C) 2003 Asit Mallick <asit.k.mallick@intel.com>
+ * Copyright (C) 2005 Christoph Lameter <clameter@sgi.com>
  *
  * Based on asm-i386/rwsem.h and other architecture implementation.
  *
@@ -11,9 +12,9 @@
  *
  * The lock count is initialized to 0 (no active and no waiting lockers).
  *
- * When a writer subtracts WRITE_BIAS, it'll get 0xffff0001 for the case
- * of an uncontended lock. Readers increment by 1 and see a positive value
- * when uncontended, negative if there are writers (and maybe) readers
+ * When a writer subtracts WRITE_BIAS, it'll get 0xffffffff00000001 for
+ * the case of an uncontended lock. Readers increment by 1 and see a positive
+ * value when uncontended, negative if there are writers (and maybe) readers
  * waiting (in which case it goes to sleep).
  */
 
@@ -29,7 +30,7 @@
  * the semaphore definition
  */
 struct rw_semaphore {
-	signed int		count;
+	signed long		count;
 	spinlock_t		wait_lock;
 	struct list_head	wait_list;
 #if RWSEM_DEBUG
@@ -37,10 +38,10 @@
 #endif
 };
 
-#define RWSEM_UNLOCKED_VALUE		0x00000000
-#define RWSEM_ACTIVE_BIAS		0x00000001
-#define RWSEM_ACTIVE_MASK		0x0000ffff
-#define RWSEM_WAITING_BIAS		(-0x00010000)
+#define RWSEM_UNLOCKED_VALUE		__IA64_UL_CONST(0x0000000000000000)
+#define RWSEM_ACTIVE_BIAS		__IA64_UL_CONST(0x0000000000000001)
+#define RWSEM_ACTIVE_MASK		__IA64_UL_CONST(0x00000000ffffffff)
+#define RWSEM_WAITING_BIAS		-__IA64_UL_CONST(0x0000000100000000)
 #define RWSEM_ACTIVE_READ_BIAS		RWSEM_ACTIVE_BIAS
 #define RWSEM_ACTIVE_WRITE_BIAS		(RWSEM_WAITING_BIAS + RWSEM_ACTIVE_BIAS)
 
@@ -83,7 +84,7 @@
 static inline void
 __down_read (struct rw_semaphore *sem)
 {
-	int result = ia64_fetchadd4_acq((unsigned int *)&sem->count, 1);
+	long result = ia64_fetchadd8_acq((unsigned long *)&sem->count, 1);
 
 	if (result < 0)
 		rwsem_down_read_failed(sem);
@@ -95,7 +96,7 @@
 static inline void
 __down_write (struct rw_semaphore *sem)
 {
-	int old, new;
+	long old, new;
 
 	do {
 		old = sem->count;
@@ -112,7 +113,7 @@
 static inline void
 __up_read (struct rw_semaphore *sem)
 {
-	int result = ia64_fetchadd4_rel((unsigned int *)&sem->count, -1);
+	long result = ia64_fetchadd8_rel((unsigned long *)&sem->count, -1);
 
 	if (result < 0 && (--result & RWSEM_ACTIVE_MASK) == 0)
 		rwsem_wake(sem);
@@ -124,7 +125,7 @@
 static inline void
 __up_write (struct rw_semaphore *sem)
 {
-	int old, new;
+	long old, new;
 
 	do {
 		old = sem->count;
@@ -141,7 +142,7 @@
 static inline int
 __down_read_trylock (struct rw_semaphore *sem)
 {
-	int tmp;
+	long tmp;
 	while ((tmp = sem->count) >= 0) {
 		if (tmp == cmpxchg_acq(&sem->count, tmp, tmp+1)) {
 			return 1;
@@ -156,7 +157,7 @@
 static inline int
 __down_write_trylock (struct rw_semaphore *sem)
 {
-	int tmp = cmpxchg_acq(&sem->count, RWSEM_UNLOCKED_VALUE,
+	long tmp = cmpxchg_acq(&sem->count, RWSEM_UNLOCKED_VALUE,
 			      RWSEM_ACTIVE_WRITE_BIAS);
 	return tmp == RWSEM_UNLOCKED_VALUE;
 }
@@ -167,7 +168,7 @@
 static inline void
 __downgrade_write (struct rw_semaphore *sem)
 {
-	int old, new;
+	long old, new;
 
 	do {
 		old = sem->count;
@@ -182,7 +183,7 @@
  * Implement atomic add functionality.  These used to be "inline" functions, but GCC v3.1
  * doesn't quite optimize this stuff right and ends up with bad calls to fetchandadd.
  */
-#define rwsem_atomic_add(delta, sem)	atomic_add(delta, (atomic_t *)(&(sem)->count))
-#define rwsem_atomic_update(delta, sem)	atomic_add_return(delta, (atomic_t *)(&(sem)->count))
+#define rwsem_atomic_add(delta, sem)	atomic64_add(delta, (atomic64_t *)(&(sem)->count))
+#define rwsem_atomic_update(delta, sem)	atomic64_add_return(delta, (atomic64_t *)(&(sem)->count))
 
 #endif /* _ASM_IA64_RWSEM_H */
diff --git a/include/asm-ia64/sn/addrs.h b/include/asm-ia64/sn/addrs.h
index 103d745..2c32e4b 100644
--- a/include/asm-ia64/sn/addrs.h
+++ b/include/asm-ia64/sn/addrs.h
@@ -3,7 +3,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (c) 1992-1999,2001-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (c) 1992-1999,2001-2005 Silicon Graphics, Inc. All rights reserved.
  */
 
 #ifndef _ASM_IA64_SN_ADDRS_H
@@ -65,7 +65,6 @@
 
 #define NASID_MASK              ((u64)NASID_BITMASK << NASID_SHIFT)
 #define AS_MASK			((u64)AS_BITMASK << AS_SHIFT)
-#define REGION_BITS		0xe000000000000000UL
 
 
 /*
@@ -79,38 +78,30 @@
 #define AS_CAC_SPACE		(AS_CAC_VAL << AS_SHIFT)
 
 
-/*
- * Base addresses for various address ranges.
- */
-#define CACHED			0xe000000000000000UL
-#define UNCACHED                0xc000000000000000UL
-#define UNCACHED_PHYS           0x8000000000000000UL
-
-
 /* 
  * Virtual Mode Local & Global MMR space.  
  */
 #define SH1_LOCAL_MMR_OFFSET	0x8000000000UL
 #define SH2_LOCAL_MMR_OFFSET	0x0200000000UL
 #define LOCAL_MMR_OFFSET	(is_shub2() ? SH2_LOCAL_MMR_OFFSET : SH1_LOCAL_MMR_OFFSET)
-#define LOCAL_MMR_SPACE		(UNCACHED | LOCAL_MMR_OFFSET)
-#define LOCAL_PHYS_MMR_SPACE	(UNCACHED_PHYS | LOCAL_MMR_OFFSET)
+#define LOCAL_MMR_SPACE		(__IA64_UNCACHED_OFFSET | LOCAL_MMR_OFFSET)
+#define LOCAL_PHYS_MMR_SPACE	(RGN_BASE(RGN_HPAGE) | LOCAL_MMR_OFFSET)
 
 #define SH1_GLOBAL_MMR_OFFSET	0x0800000000UL
 #define SH2_GLOBAL_MMR_OFFSET	0x0300000000UL
 #define GLOBAL_MMR_OFFSET	(is_shub2() ? SH2_GLOBAL_MMR_OFFSET : SH1_GLOBAL_MMR_OFFSET)
-#define GLOBAL_MMR_SPACE	(UNCACHED | GLOBAL_MMR_OFFSET)
+#define GLOBAL_MMR_SPACE	(__IA64_UNCACHED_OFFSET | GLOBAL_MMR_OFFSET)
 
 /*
  * Physical mode addresses
  */
-#define GLOBAL_PHYS_MMR_SPACE	(UNCACHED_PHYS | GLOBAL_MMR_OFFSET)
+#define GLOBAL_PHYS_MMR_SPACE	(RGN_BASE(RGN_HPAGE) | GLOBAL_MMR_OFFSET)
 
 
 /*
  * Clear region & AS bits.
  */
-#define TO_PHYS_MASK		(~(REGION_BITS | AS_MASK))
+#define TO_PHYS_MASK		(~(RGN_BITS | AS_MASK))
 
 
 /*
@@ -126,6 +117,7 @@
 #define GLOBAL_MMR_PHYS_ADDR(n,a) (GLOBAL_PHYS_MMR_SPACE | REMOTE_ADDR(n,a))
 #define GLOBAL_CAC_ADDR(n,a)	(CAC_BASE | REMOTE_ADDR(n,a))
 #define CHANGE_NASID(n,x)	((void *)(((u64)(x) & ~NASID_MASK) | NASID_SPACE(n)))
+#define IS_TIO_NASID(n)		((n) & 1)
 
 
 /* non-II mmr's start at top of big window space (4G) */
@@ -134,10 +126,10 @@
 /*
  * general address defines
  */
-#define CAC_BASE		(CACHED   | AS_CAC_SPACE)
-#define AMO_BASE		(UNCACHED | AS_AMO_SPACE)
-#define AMO_PHYS_BASE		(UNCACHED_PHYS | AS_AMO_SPACE)
-#define GET_BASE		(CACHED   | AS_GET_SPACE)
+#define CAC_BASE		(PAGE_OFFSET | AS_CAC_SPACE)
+#define AMO_BASE		(__IA64_UNCACHED_OFFSET | AS_AMO_SPACE)
+#define AMO_PHYS_BASE		(RGN_BASE(RGN_HPAGE) | AS_AMO_SPACE)
+#define GET_BASE		(PAGE_OFFSET | AS_GET_SPACE)
 
 /*
  * Convert Memory addresses between various addressing modes.
@@ -155,17 +147,35 @@
  *           the chiplet id is zero.  If we implement TIO-TIO dma, we might need
  *           to insert a chiplet id into this macro.  However, it is our belief
  *           right now that this chiplet id will be ICE, which is also zero.
- *           Nasid starts on bit 40.
  */
-#define PHYS_TO_TIODMA(x)	( (((u64)(NASID_GET(x))) << 40) | NODE_OFFSET(x))
-#define PHYS_TO_DMA(x)          ( (((u64)(x) & NASID_MASK) >> 2) | NODE_OFFSET(x))
+#define SH1_TIO_PHYS_TO_DMA(x) 						\
+	((((u64)(NASID_GET(x))) << 40) | NODE_OFFSET(x))
+
+#define SH2_NETWORK_BANK_OFFSET(x) 					\
+        ((u64)(x) & ((1UL << (sn_hub_info->nasid_shift - 4)) -1))
+
+#define SH2_NETWORK_BANK_SELECT(x) 					\
+        ((((u64)(x) & (0x3UL << (sn_hub_info->nasid_shift - 4)))	\
+        	>> (sn_hub_info->nasid_shift - 4)) << 36)
+
+#define SH2_NETWORK_ADDRESS(x) 						\
+	(SH2_NETWORK_BANK_OFFSET(x) | SH2_NETWORK_BANK_SELECT(x))
+
+#define SH2_TIO_PHYS_TO_DMA(x) 						\
+        (((u64)(NASID_GET(x)) << 40) | 	SH2_NETWORK_ADDRESS(x))
+
+#define PHYS_TO_TIODMA(x)						\
+	(is_shub1() ? SH1_TIO_PHYS_TO_DMA(x) : SH2_TIO_PHYS_TO_DMA(x))
+
+#define PHYS_TO_DMA(x)							\
+	((((u64)(x) & NASID_MASK) >> 2) | NODE_OFFSET(x))
 
 
 /*
  * Macros to test for address type.
  */
-#define IS_AMO_ADDRESS(x)	(((u64)(x) & (REGION_BITS | AS_MASK)) == AMO_BASE)
-#define IS_AMO_PHYS_ADDRESS(x)	(((u64)(x) & (REGION_BITS | AS_MASK)) == AMO_PHYS_BASE)
+#define IS_AMO_ADDRESS(x)	(((u64)(x) & (RGN_BITS | AS_MASK)) == AMO_BASE)
+#define IS_AMO_PHYS_ADDRESS(x)	(((u64)(x) & (RGN_BITS | AS_MASK)) == AMO_PHYS_BASE)
 
 
 /*
@@ -180,18 +190,20 @@
 #define TIO_SWIN_BASE(n, w) 		(TIO_IO_BASE(n) + \
 					    ((u64) (w) << TIO_SWIN_SIZE_BITS))
 #define NODE_IO_BASE(n)			(GLOBAL_MMR_SPACE | NASID_SPACE(n))
-#define TIO_IO_BASE(n)                  (UNCACHED | NASID_SPACE(n))
+#define TIO_IO_BASE(n)                  (__IA64_UNCACHED_OFFSET | NASID_SPACE(n))
 #define BWIN_SIZE			(1UL << BWIN_SIZE_BITS)
 #define NODE_BWIN_BASE0(n)		(NODE_IO_BASE(n) + BWIN_SIZE)
 #define NODE_BWIN_BASE(n, w)		(NODE_BWIN_BASE0(n) + ((u64) (w) << BWIN_SIZE_BITS))
 #define RAW_NODE_SWIN_BASE(n, w)	(NODE_IO_BASE(n) + ((u64) (w) << SWIN_SIZE_BITS))
 #define BWIN_WIDGET_MASK		0x7
 #define BWIN_WINDOWNUM(x)		(((x) >> BWIN_SIZE_BITS) & BWIN_WIDGET_MASK)
+#define SH1_IS_BIG_WINDOW_ADDR(x)	((x) & BWIN_TOP)
 
 #define TIO_BWIN_WINDOW_SELECT_MASK	0x7
 #define TIO_BWIN_WINDOWNUM(x)		(((x) >> TIO_BWIN_SIZE_BITS) & TIO_BWIN_WINDOW_SELECT_MASK)
 
-
+#define TIO_HWIN_SHIFT_BITS		33
+#define TIO_HWIN(x)			(NODE_OFFSET(x) >> TIO_HWIN_SHIFT_BITS)
 
 /*
  * The following definitions pertain to the IO special address
@@ -216,10 +228,6 @@
 #define TIO_SWIN_WIDGETNUM(x)		(((x)  >> TIO_SWIN_SIZE_BITS) & TIO_SWIN_WIDGET_MASK)
 
 
-#define TIO_IOSPACE_ADDR(n,x)					\
-	/* Move in the Chiplet ID for TIO Local Block MMR */	\
-	(REMOTE_ADDR(n,x) | 1UL << (NASID_SHIFT - 2))
-
 /*
  * The following macros produce the correct base virtual address for
  * the hub registers. The REMOTE_HUB_* macro produce
@@ -234,18 +242,40 @@
  *	Otherwise, the recommended approach is to use *_HUB_L() and *_HUB_S().
  *	They're always safe.
  */
+/* Shub1 TIO & MMR addressing macros */
+#define SH1_TIO_IOSPACE_ADDR(n,x)					\
+	GLOBAL_MMR_ADDR(n,x)
+
+#define SH1_REMOTE_BWIN_MMR(n,x)					\
+	GLOBAL_MMR_ADDR(n,x)
+
+#define SH1_REMOTE_SWIN_MMR(n,x)					\
+	(NODE_SWIN_BASE(n,1) + 0x800000UL + (x))
+
+#define SH1_REMOTE_MMR(n,x)						\
+	(SH1_IS_BIG_WINDOW_ADDR(x) ? SH1_REMOTE_BWIN_MMR(n,x) :		\
+	 	SH1_REMOTE_SWIN_MMR(n,x))
+
+/* Shub1 TIO & MMR addressing macros */
+#define SH2_TIO_IOSPACE_ADDR(n,x)					\
+	((__IA64_UNCACHED_OFFSET | REMOTE_ADDR(n,x) | 1UL << (NASID_SHIFT - 2)))
+
+#define SH2_REMOTE_MMR(n,x)						\
+	GLOBAL_MMR_ADDR(n,x)
+
+
+/* TIO & MMR addressing macros that work on both shub1 & shub2 */
+#define TIO_IOSPACE_ADDR(n,x)						\
+	((u64 *)(is_shub1() ? SH1_TIO_IOSPACE_ADDR(n,x) :		\
+		 SH2_TIO_IOSPACE_ADDR(n,x)))
+
+#define SH_REMOTE_MMR(n,x)						\
+	(is_shub1() ? SH1_REMOTE_MMR(n,x) : SH2_REMOTE_MMR(n,x))
+
 #define REMOTE_HUB_ADDR(n,x)						\
-	((n & 1) ?							\
-	/* TIO: */							\
-	(is_shub2() ?							\
-	/* TIO on Shub2 */						\
-	(volatile u64 *)(TIO_IOSPACE_ADDR(n,x))				\
-	: /* TIO on shub1 */						\
-	(volatile u64 *)(GLOBAL_MMR_ADDR(n,x)))				\
-									\
-	: /* SHUB1 and SHUB2 MMRs: */					\
-	(((x) & BWIN_TOP) ? ((volatile u64 *)(GLOBAL_MMR_ADDR(n,x)))	\
-	: ((volatile u64 *)(NODE_SWIN_BASE(n,1) + 0x800000 + (x)))))
+	(IS_TIO_NASID(n) ?  ((volatile u64*)TIO_IOSPACE_ADDR(n,x)) :	\
+	 ((volatile u64*)SH_REMOTE_MMR(n,x)))
+
 
 #define HUB_L(x)			(*((volatile typeof(*x) *)x))
 #define	HUB_S(x,d)			(*((volatile typeof(*x) *)x) = (d))
diff --git a/include/asm-ia64/sn/geo.h b/include/asm-ia64/sn/geo.h
index 84b2546..f083c94 100644
--- a/include/asm-ia64/sn/geo.h
+++ b/include/asm-ia64/sn/geo.h
@@ -3,7 +3,7 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  *
- * Copyright (C) 1992 - 1997, 2000-2004 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 1992 - 1997, 2000-2005 Silicon Graphics, Inc. All rights reserved.
  */
 
 #ifndef _ASM_IA64_SN_GEO_H
@@ -108,7 +108,6 @@
 #define INVALID_SLAB            (slabid_t)-1
 #define INVALID_SLOT            (slotid_t)-1
 #define INVALID_MODULE          ((moduleid_t)-1)
-#define INVALID_PARTID          ((partid_t)-1)
 
 static inline slabid_t geo_slab(geoid_t g)
 {
diff --git a/include/asm-ia64/sn/intr.h b/include/asm-ia64/sn/intr.h
index e190dd4..e35074f 100644
--- a/include/asm-ia64/sn/intr.h
+++ b/include/asm-ia64/sn/intr.h
@@ -12,13 +12,12 @@
 #include <linux/rcupdate.h>
 
 #define SGI_UART_VECTOR		(0xe9)
-#define SGI_PCIBR_ERROR		(0x33)
 
 /* Reserved IRQs : Note, not to exceed IA64_SN2_FIRST_DEVICE_VECTOR */
 #define SGI_XPC_ACTIVATE                (0x30)
 #define SGI_II_ERROR                    (0x31)
 #define SGI_XBOW_ERROR                  (0x32)
-#define SGI_PCIBR_ERROR                 (0x33)
+#define SGI_PCIASIC_ERROR               (0x33)
 #define SGI_ACPI_SCI_INT                (0x34)
 #define SGI_TIOCA_ERROR                 (0x35)
 #define SGI_TIO_ERROR                   (0x36)
diff --git a/include/asm-ia64/sn/nodepda.h b/include/asm-ia64/sn/nodepda.h
index 7138b1e..47bb810 100644
--- a/include/asm-ia64/sn/nodepda.h
+++ b/include/asm-ia64/sn/nodepda.h
@@ -37,7 +37,6 @@
 
 struct nodepda_s {
 	void 		*pdinfo;	/* Platform-dependent per-node info */
-	spinlock_t		bist_lock;
 
 	/*
 	 * The BTEs on this node are shared by the local cpus
@@ -55,6 +54,8 @@
 	 * Array of physical cpu identifiers. Indexed by cpuid.
 	 */
 	struct phys_cpuid	phys_cpuid[NR_CPUS];
+	spinlock_t		ptc_lock ____cacheline_aligned_in_smp;
+	spinlock_t		bist_lock;
 };
 
 typedef struct nodepda_s nodepda_t;
diff --git a/include/asm-ia64/sn/pcibus_provider_defs.h b/include/asm-ia64/sn/pcibus_provider_defs.h
index 976f5ef..ad0e8e8 100644
--- a/include/asm-ia64/sn/pcibus_provider_defs.h
+++ b/include/asm-ia64/sn/pcibus_provider_defs.h
@@ -18,8 +18,9 @@
 #define PCIIO_ASIC_TYPE_PIC	2
 #define PCIIO_ASIC_TYPE_TIOCP	3
 #define PCIIO_ASIC_TYPE_TIOCA	4
+#define PCIIO_ASIC_TYPE_TIOCE	5
 
-#define PCIIO_ASIC_MAX_TYPES	5
+#define PCIIO_ASIC_MAX_TYPES	6
 
 /*
  * Common pciio bus provider data.  There should be one of these as the
@@ -30,7 +31,8 @@
 struct pcibus_bussoft {
 	uint32_t		bs_asic_type;	/* chipset type */
 	uint32_t		bs_xid;		/* xwidget id */
-	uint64_t		bs_persist_busnum; /* Persistent Bus Number */
+	uint32_t		bs_persist_busnum; /* Persistent Bus Number */
+	uint32_t		bs_persist_segment; /* Segment Number */
 	uint64_t		bs_legacy_io;	/* legacy io pio addr */
 	uint64_t		bs_legacy_mem;	/* legacy mem pio addr */
 	uint64_t		bs_base;	/* widget base */
@@ -47,6 +49,8 @@
 	dma_addr_t	(*dma_map_consistent)(struct pci_dev *, unsigned long, size_t);
 	void		(*dma_unmap)(struct pci_dev *, dma_addr_t, int);
 	void *		(*bus_fixup)(struct pcibus_bussoft *, struct pci_controller *);
+ 	void		(*force_interrupt)(struct sn_irq_info *);
+ 	void		(*target_interrupt)(struct sn_irq_info *);
 };
 
 extern struct sn_pcibus_provider *sn_pci_provider[];
diff --git a/include/asm-ia64/sn/pda.h b/include/asm-ia64/sn/pda.h
index ea5590c..1c5108d 100644
--- a/include/asm-ia64/sn/pda.h
+++ b/include/asm-ia64/sn/pda.h
@@ -39,7 +39,6 @@
 	unsigned long pio_write_status_val;
 	volatile unsigned long *pio_shub_war_cam_addr;
 
-	unsigned long	sn_soft_irr[4];
 	unsigned long	sn_in_service_ivecs[4];
 	int		sn_lb_int_war_ticks;
 	int		sn_last_irq;
diff --git a/include/asm-ia64/sn/sn2/sn_hwperf.h b/include/asm-ia64/sn/sn2/sn_hwperf.h
index df75f4c..291ef3d 100644
--- a/include/asm-ia64/sn/sn2/sn_hwperf.h
+++ b/include/asm-ia64/sn/sn2/sn_hwperf.h
@@ -43,6 +43,7 @@
 
 /* macros for object classification */
 #define SN_HWPERF_IS_NODE(x)		((x) && strstr((x)->name, "SHub"))
+#define SN_HWPERF_IS_NODE_SHUB2(x)	((x) && strstr((x)->name, "SHub 2."))
 #define SN_HWPERF_IS_IONODE(x)		((x) && strstr((x)->name, "TIO"))
 #define SN_HWPERF_IS_ROUTER(x)		((x) && strstr((x)->name, "Router"))
 #define SN_HWPERF_IS_NL3ROUTER(x)	((x) && strstr((x)->name, "NL3Router"))
@@ -214,6 +215,15 @@
  */
 #define SN_HWPERF_GET_NODE_NASID	(102|SN_HWPERF_OP_MEM_COPYOUT)
 
+/*
+ * Given a node id, determine the id of the nearest node with CPUs
+ * and the id of the nearest node that has memory. The argument
+ * node would normally be a "headless" node, e.g. an "IO node".
+ * Return 0 on success.
+ */
+extern int sn_hwperf_get_nearest_node(cnodeid_t node,
+	cnodeid_t *near_mem, cnodeid_t *near_cpu);
+
 /* return codes */
 #define SN_HWPERF_OP_OK			0
 #define SN_HWPERF_OP_NOMEM		1
diff --git a/include/asm-ia64/sn/sn_sal.h b/include/asm-ia64/sn/sn_sal.h
index 27976d2..e67825a 100644
--- a/include/asm-ia64/sn/sn_sal.h
+++ b/include/asm-ia64/sn/sn_sal.h
@@ -55,7 +55,6 @@
 #define  SN_SAL_BUS_CONFIG		   	   0x02000037
 #define  SN_SAL_SYS_SERIAL_GET			   0x02000038
 #define  SN_SAL_PARTITION_SERIAL_GET		   0x02000039
-#define  SN_SAL_SYSCTL_PARTITION_GET		   0x0200003a
 #define  SN_SAL_SYSTEM_POWER_DOWN		   0x0200003b
 #define  SN_SAL_GET_MASTER_BASEIO_NASID		   0x0200003c
 #define  SN_SAL_COHERENCE                          0x0200003d
@@ -78,7 +77,8 @@
 
 #define SN_SAL_HUB_ERROR_INTERRUPT		   0x02000060
 #define SN_SAL_BTE_RECOVER			   0x02000061
-#define SN_SAL_IOIF_GET_PCI_TOPOLOGY	           0x02000062
+#define SN_SAL_RESERVED_DO_NOT_USE		   0x02000062
+#define SN_SAL_IOIF_GET_PCI_TOPOLOGY		   0x02000064
 
 /*
  * Service-specific constants
@@ -586,35 +586,6 @@
 }
 
 /*
- * Returns the partition id of the nasid passed in as an argument,
- * or INVALID_PARTID if the partition id cannot be retrieved.
- */
-static inline partid_t
-ia64_sn_sysctl_partition_get(nasid_t nasid)
-{
-	struct ia64_sal_retval ret_stuff;
-	ia64_sal_oemcall_nolock(&ret_stuff, SN_SAL_SYSCTL_PARTITION_GET, nasid,
-				0, 0, 0, 0, 0, 0);
-	if (ret_stuff.status != 0)
-	    return INVALID_PARTID;
-	return ((partid_t)ret_stuff.v0);
-}
-
-/*
- * Returns the partition id of the current processor.
- */
-
-extern partid_t sn_partid;
-
-static inline partid_t
-sn_local_partid(void) {
-	if (unlikely(sn_partid < 0)) {
-		sn_partid = ia64_sn_sysctl_partition_get(cpuid_to_nasid(smp_processor_id()));
-	}
-	return sn_partid;
-}
-
-/*
  * Returns the physical address of the partition's reserved page through
  * an iterative number of calls.
  *
@@ -749,7 +720,8 @@
 {
 	struct ia64_sal_retval ret_stuff;
 	SAL_CALL(ret_stuff, SN_SAL_SYSTEM_POWER_DOWN, 0, 0, 0, 0, 0, 0, 0);
-	while(1);
+	while(1)
+		cpu_relax();
 	/* never returns */
 }
 
@@ -1018,24 +990,6 @@
 	ret_stuff.v2 = 0;
 	SAL_CALL_NOLOCK(ret_stuff, SN_SAL_GET_SN_INFO, fc, 0, 0, 0, 0, 0, 0);
 
-/***** BEGIN HACK - temp til old proms no longer supported ********/
-	if (ret_stuff.status == SALRET_NOT_IMPLEMENTED) {
-		int nasid = get_sapicid() & 0xfff;;
-#define SH_SHUB_ID_NODES_PER_BIT_MASK 0x001f000000000000UL                                               
-#define SH_SHUB_ID_NODES_PER_BIT_SHFT 48                                                               
-		if (shubtype) *shubtype = 0;
-		if (nasid_bitmask) *nasid_bitmask = 0x7ff;
-		if (nasid_shift) *nasid_shift = 38;
-		if (systemsize) *systemsize = 11;
-		if (sharing_domain_size) *sharing_domain_size = 9;
-		if (partid) *partid = ia64_sn_sysctl_partition_get(nasid);
-		if (coher) *coher = nasid >> 9;
-		if (reg) *reg = (HUB_L((u64 *) LOCAL_MMR_ADDR(SH1_SHUB_ID)) & SH_SHUB_ID_NODES_PER_BIT_MASK) >>
-			SH_SHUB_ID_NODES_PER_BIT_SHFT;
-		return 0;
-	}
-/***** END HACK *******/
-
 	if (ret_stuff.status < 0)
 		return ret_stuff.status;
 
@@ -1068,12 +1022,10 @@
 }
 
 static inline int
-ia64_sn_ioif_get_pci_topology(u64 rack, u64 bay, u64 slot, u64 slab,
-			      u64 buf, u64 len)
+ia64_sn_ioif_get_pci_topology(u64 buf, u64 len)
 {
 	struct ia64_sal_retval rv;
-	SAL_CALL_NOLOCK(rv, SN_SAL_IOIF_GET_PCI_TOPOLOGY,
-		rack, bay, slot, slab, buf, len, 0);
+	SAL_CALL_NOLOCK(rv, SN_SAL_IOIF_GET_PCI_TOPOLOGY, buf, len, 0, 0, 0, 0, 0);
 	return (int) rv.status;
 }
 
diff --git a/include/asm-ia64/sn/tioce.h b/include/asm-ia64/sn/tioce.h
new file mode 100644
index 0000000..2287985
--- /dev/null
+++ b/include/asm-ia64/sn/tioce.h
@@ -0,0 +1,740 @@
+/**************************************************************************
+ *                                                                        *
+ *  Unpublished copyright (c) 2005, Silicon Graphics, Inc.                *
+ *  THIS IS UNPUBLISHED CONFIDENTIAL AND PROPRIETARY SOURCE CODE OF SGI.  *
+ *                                                                        *
+ *  The copyright notice above does  not evidence any actual or intended  *
+ *  publication  or  disclosure  of  this source  code,  which  includes  *
+ *  information that is confidential  and/or proprietary, and is a trade  *
+ *  secret, of  Silicon Graphics, Inc.   ANY REPRODUCTION, MODIFICATION,  *
+ *  DISTRIBUTION, PUBLIC  PERFORMANCE, OR  PUBLIC DISPLAY OF  OR THROUGH  *
+ *  USE  OF THIS  SOURCE CODE  WITHOUT  THE EXPRESS  WRITTEN CONSENT  OF  *
+ *  SILICON GRAPHICS, INC.  IS  STRICTLY PROHIBITED, AND IN VIOLATION OF  *
+ *  APPLICABLE  LAWS   AND  INTERNATIONAL  TREATIES.    THE  RECEIPT  OR  *
+ *  POSSESSION OF  THIS SOURCE CODE AND/OR RELATED  INFORMATION DOES NOT  *
+ *  CONVEY OR IMPLY ANY RIGHTS  TO REPRODUCE, DISCLOSE OR DISTRIBUTE ITS  *
+ *  CONTENTS,  OR TO  MANUFACTURE, USE,  OR  SELL ANYTHING  THAT IT  MAY  *
+ *  DESCRIBE, IN WHOLE OR IN PART.                                        *
+ *                                                                        *
+ **************************************************************************/
+
+#ifndef __ASM_IA64_SN_TIOCE_H__
+#define __ASM_IA64_SN_TIOCE_H__
+
+/* CE ASIC part & mfgr information  */
+#define TIOCE_PART_NUM			0xCE00
+#define TIOCE_MFGR_NUM			0x36
+#define TIOCE_REV_A			0x1
+
+/* CE Virtual PPB Vendor/Device IDs */
+#define CE_VIRT_PPB_VENDOR_ID		0x10a9
+#define CE_VIRT_PPB_DEVICE_ID		0x4002
+
+/* CE Host Bridge Vendor/Device IDs */
+#define CE_HOST_BRIDGE_VENDOR_ID	0x10a9
+#define CE_HOST_BRIDGE_DEVICE_ID	0x4003
+
+
+#define TIOCE_NUM_M40_ATES		4096
+#define TIOCE_NUM_M3240_ATES		2048
+#define TIOCE_NUM_PORTS			2
+
+/*
+ * Register layout for TIOCE.  MMR offsets are shown at the far right of the
+ * structure definition.
+ */
+typedef volatile struct tioce {
+	/*
+	 * ADMIN : Administration Registers
+	 */
+	uint64_t	ce_adm_id;				/* 0x000000 */
+	uint64_t	ce_pad_000008;				/* 0x000008 */
+	uint64_t	ce_adm_dyn_credit_status;		/* 0x000010 */
+	uint64_t	ce_adm_last_credit_status;		/* 0x000018 */
+	uint64_t	ce_adm_credit_limit;			/* 0x000020 */
+	uint64_t	ce_adm_force_credit;			/* 0x000028 */
+	uint64_t	ce_adm_control;				/* 0x000030 */
+	uint64_t	ce_adm_mmr_chn_timeout;			/* 0x000038 */
+	uint64_t	ce_adm_ssp_ure_timeout;			/* 0x000040 */
+	uint64_t	ce_adm_ssp_dre_timeout;			/* 0x000048 */
+	uint64_t	ce_adm_ssp_debug_sel;			/* 0x000050 */
+	uint64_t	ce_adm_int_status;			/* 0x000058 */
+	uint64_t	ce_adm_int_status_alias;		/* 0x000060 */
+	uint64_t	ce_adm_int_mask;			/* 0x000068 */
+	uint64_t	ce_adm_int_pending;			/* 0x000070 */
+	uint64_t	ce_adm_force_int;			/* 0x000078 */
+	uint64_t	ce_adm_ure_ups_buf_barrier_flush;	/* 0x000080 */
+	uint64_t	ce_adm_int_dest[15];	    /* 0x000088 -- 0x0000F8 */
+	uint64_t	ce_adm_error_summary;			/* 0x000100 */
+	uint64_t	ce_adm_error_summary_alias;		/* 0x000108 */
+	uint64_t	ce_adm_error_mask;			/* 0x000110 */
+	uint64_t	ce_adm_first_error;			/* 0x000118 */
+	uint64_t	ce_adm_error_overflow;			/* 0x000120 */
+	uint64_t	ce_adm_error_overflow_alias;		/* 0x000128 */
+	uint64_t	ce_pad_000130[2];	    /* 0x000130 -- 0x000138 */
+	uint64_t	ce_adm_tnum_error;			/* 0x000140 */
+	uint64_t	ce_adm_mmr_err_detail;			/* 0x000148 */
+	uint64_t	ce_adm_msg_sram_perr_detail;		/* 0x000150 */
+	uint64_t	ce_adm_bap_sram_perr_detail;		/* 0x000158 */
+	uint64_t	ce_adm_ce_sram_perr_detail;		/* 0x000160 */
+	uint64_t	ce_adm_ce_credit_oflow_detail;		/* 0x000168 */
+	uint64_t	ce_adm_tx_link_idle_max_timer;		/* 0x000170 */
+	uint64_t	ce_adm_pcie_debug_sel;			/* 0x000178 */
+	uint64_t	ce_pad_000180[16];	    /* 0x000180 -- 0x0001F8 */
+
+	uint64_t	ce_adm_pcie_debug_sel_top;		/* 0x000200 */
+	uint64_t	ce_adm_pcie_debug_lat_sel_lo_top;	/* 0x000208 */
+	uint64_t	ce_adm_pcie_debug_lat_sel_hi_top;	/* 0x000210 */
+	uint64_t	ce_adm_pcie_debug_trig_sel_top;		/* 0x000218 */
+	uint64_t	ce_adm_pcie_debug_trig_lat_sel_lo_top;	/* 0x000220 */
+	uint64_t	ce_adm_pcie_debug_trig_lat_sel_hi_top;	/* 0x000228 */
+	uint64_t	ce_adm_pcie_trig_compare_top;		/* 0x000230 */
+	uint64_t	ce_adm_pcie_trig_compare_en_top;	/* 0x000238 */
+	uint64_t	ce_adm_ssp_debug_sel_top;		/* 0x000240 */
+	uint64_t	ce_adm_ssp_debug_lat_sel_lo_top;	/* 0x000248 */
+	uint64_t	ce_adm_ssp_debug_lat_sel_hi_top;	/* 0x000250 */
+	uint64_t	ce_adm_ssp_debug_trig_sel_top;		/* 0x000258 */
+	uint64_t	ce_adm_ssp_debug_trig_lat_sel_lo_top;	/* 0x000260 */
+	uint64_t	ce_adm_ssp_debug_trig_lat_sel_hi_top;	/* 0x000268 */
+	uint64_t	ce_adm_ssp_trig_compare_top;		/* 0x000270 */
+	uint64_t	ce_adm_ssp_trig_compare_en_top;		/* 0x000278 */
+	uint64_t	ce_pad_000280[48];	    /* 0x000280 -- 0x0003F8 */
+
+	uint64_t	ce_adm_bap_ctrl;			/* 0x000400 */
+	uint64_t	ce_pad_000408[127];	    /* 0x000408 -- 0x0007F8 */
+
+	uint64_t	ce_msg_buf_data63_0[35];    /* 0x000800 -- 0x000918 */
+	uint64_t	ce_pad_000920[29];	    /* 0x000920 -- 0x0009F8 */
+
+	uint64_t	ce_msg_buf_data127_64[35];  /* 0x000A00 -- 0x000B18 */
+	uint64_t	ce_pad_000B20[29];	    /* 0x000B20 -- 0x000BF8 */
+
+	uint64_t	ce_msg_buf_parity[35];	    /* 0x000C00 -- 0x000D18 */
+	uint64_t	ce_pad_000D20[29];	    /* 0x000D20 -- 0x000DF8 */
+
+	uint64_t	ce_pad_000E00[576];	    /* 0x000E00 -- 0x001FF8 */
+
+	/*
+	 * LSI : LSI's PCI Express Link Registers (Link#1 and Link#2)
+	 * Link#1 MMRs at start at 0x002000, Link#2 MMRs at 0x003000
+	 * NOTE: the comment offsets at far right: let 'z' = {2 or 3}
+	 */
+	#define ce_lsi(link_num)	ce_lsi[link_num-1]
+	struct ce_lsi_reg {
+		uint64_t	ce_lsi_lpu_id;			/* 0x00z000 */
+		uint64_t	ce_lsi_rst;			/* 0x00z008 */
+		uint64_t	ce_lsi_dbg_stat;		/* 0x00z010 */
+		uint64_t	ce_lsi_dbg_cfg;			/* 0x00z018 */
+		uint64_t	ce_lsi_ltssm_ctrl;		/* 0x00z020 */
+		uint64_t	ce_lsi_lk_stat;			/* 0x00z028 */
+		uint64_t	ce_pad_00z030[2];   /* 0x00z030 -- 0x00z038 */
+		uint64_t	ce_lsi_int_and_stat;		/* 0x00z040 */
+		uint64_t	ce_lsi_int_mask;		/* 0x00z048 */
+		uint64_t	ce_pad_00z050[22];  /* 0x00z050 -- 0x00z0F8 */
+		uint64_t	ce_lsi_lk_perf_cnt_sel;		/* 0x00z100 */
+		uint64_t	ce_pad_00z108;			/* 0x00z108 */
+		uint64_t	ce_lsi_lk_perf_cnt_ctrl;	/* 0x00z110 */
+		uint64_t	ce_pad_00z118;			/* 0x00z118 */
+		uint64_t	ce_lsi_lk_perf_cnt1;		/* 0x00z120 */
+		uint64_t	ce_lsi_lk_perf_cnt1_test;	/* 0x00z128 */
+		uint64_t	ce_lsi_lk_perf_cnt2;		/* 0x00z130 */
+		uint64_t	ce_lsi_lk_perf_cnt2_test;	/* 0x00z138 */
+		uint64_t	ce_pad_00z140[24];  /* 0x00z140 -- 0x00z1F8 */
+		uint64_t	ce_lsi_lk_lyr_cfg;		/* 0x00z200 */
+		uint64_t	ce_lsi_lk_lyr_status;		/* 0x00z208 */
+		uint64_t	ce_lsi_lk_lyr_int_stat;		/* 0x00z210 */
+		uint64_t	ce_lsi_lk_ly_int_stat_test;	/* 0x00z218 */
+		uint64_t	ce_lsi_lk_ly_int_stat_mask;	/* 0x00z220 */
+		uint64_t	ce_pad_00z228[3];   /* 0x00z228 -- 0x00z238 */
+		uint64_t	ce_lsi_fc_upd_ctl;		/* 0x00z240 */
+		uint64_t	ce_pad_00z248[3];   /* 0x00z248 -- 0x00z258 */
+		uint64_t	ce_lsi_flw_ctl_upd_to_timer;	/* 0x00z260 */
+		uint64_t	ce_lsi_flw_ctl_upd_timer0;	/* 0x00z268 */
+		uint64_t	ce_lsi_flw_ctl_upd_timer1;	/* 0x00z270 */
+		uint64_t	ce_pad_00z278[49];  /* 0x00z278 -- 0x00z3F8 */
+		uint64_t	ce_lsi_freq_nak_lat_thrsh;	/* 0x00z400 */
+		uint64_t	ce_lsi_ack_nak_lat_tmr;		/* 0x00z408 */
+		uint64_t	ce_lsi_rply_tmr_thr;		/* 0x00z410 */
+		uint64_t	ce_lsi_rply_tmr;		/* 0x00z418 */
+		uint64_t	ce_lsi_rply_num_stat;		/* 0x00z420 */
+		uint64_t	ce_lsi_rty_buf_max_addr;	/* 0x00z428 */
+		uint64_t	ce_lsi_rty_fifo_ptr;		/* 0x00z430 */
+		uint64_t	ce_lsi_rty_fifo_rd_wr_ptr;	/* 0x00z438 */
+		uint64_t	ce_lsi_rty_fifo_cred;		/* 0x00z440 */
+		uint64_t	ce_lsi_seq_cnt;			/* 0x00z448 */
+		uint64_t	ce_lsi_ack_sent_seq_num;	/* 0x00z450 */
+		uint64_t	ce_lsi_seq_cnt_fifo_max_addr;	/* 0x00z458 */
+		uint64_t	ce_lsi_seq_cnt_fifo_ptr;	/* 0x00z460 */
+		uint64_t	ce_lsi_seq_cnt_rd_wr_ptr;	/* 0x00z468 */
+		uint64_t	ce_lsi_tx_lk_ts_ctl;		/* 0x00z470 */
+		uint64_t	ce_pad_00z478;			/* 0x00z478 */
+		uint64_t	ce_lsi_mem_addr_ctl;		/* 0x00z480 */
+		uint64_t	ce_lsi_mem_d_ld0;		/* 0x00z488 */
+		uint64_t	ce_lsi_mem_d_ld1;		/* 0x00z490 */
+		uint64_t	ce_lsi_mem_d_ld2;		/* 0x00z498 */
+		uint64_t	ce_lsi_mem_d_ld3;		/* 0x00z4A0 */
+		uint64_t	ce_lsi_mem_d_ld4;		/* 0x00z4A8 */
+		uint64_t	ce_pad_00z4B0[2];   /* 0x00z4B0 -- 0x00z4B8 */
+		uint64_t	ce_lsi_rty_d_cnt;		/* 0x00z4C0 */
+		uint64_t	ce_lsi_seq_buf_cnt;		/* 0x00z4C8 */
+		uint64_t	ce_lsi_seq_buf_bt_d;		/* 0x00z4D0 */
+		uint64_t	ce_pad_00z4D8;			/* 0x00z4D8 */
+		uint64_t	ce_lsi_ack_lat_thr;		/* 0x00z4E0 */
+		uint64_t	ce_pad_00z4E8[3];   /* 0x00z4E8 -- 0x00z4F8 */
+		uint64_t	ce_lsi_nxt_rcv_seq_1_cntr;	/* 0x00z500 */
+		uint64_t	ce_lsi_unsp_dllp_rcvd;		/* 0x00z508 */
+		uint64_t	ce_lsi_rcv_lk_ts_ctl;		/* 0x00z510 */
+		uint64_t	ce_pad_00z518[29];  /* 0x00z518 -- 0x00z5F8 */
+		uint64_t	ce_lsi_phy_lyr_cfg;		/* 0x00z600 */
+		uint64_t	ce_pad_00z608;			/* 0x00z608 */
+		uint64_t	ce_lsi_phy_lyr_int_stat;	/* 0x00z610 */
+		uint64_t	ce_lsi_phy_lyr_int_stat_test;	/* 0x00z618 */
+		uint64_t	ce_lsi_phy_lyr_int_mask;	/* 0x00z620 */
+		uint64_t	ce_pad_00z628[11];  /* 0x00z628 -- 0x00z678 */
+		uint64_t	ce_lsi_rcv_phy_cfg;		/* 0x00z680 */
+		uint64_t	ce_lsi_rcv_phy_stat1;		/* 0x00z688 */
+		uint64_t	ce_lsi_rcv_phy_stat2;		/* 0x00z690 */
+		uint64_t	ce_lsi_rcv_phy_stat3;		/* 0x00z698 */
+		uint64_t	ce_lsi_rcv_phy_int_stat;	/* 0x00z6A0 */
+		uint64_t	ce_lsi_rcv_phy_int_stat_test;	/* 0x00z6A8 */
+		uint64_t	ce_lsi_rcv_phy_int_mask;	/* 0x00z6B0 */
+		uint64_t	ce_pad_00z6B8[9];   /* 0x00z6B8 -- 0x00z6F8 */
+		uint64_t	ce_lsi_tx_phy_cfg;		/* 0x00z700 */
+		uint64_t	ce_lsi_tx_phy_stat;		/* 0x00z708 */
+		uint64_t	ce_lsi_tx_phy_int_stat;		/* 0x00z710 */
+		uint64_t	ce_lsi_tx_phy_int_stat_test;	/* 0x00z718 */
+		uint64_t	ce_lsi_tx_phy_int_mask;		/* 0x00z720 */
+		uint64_t	ce_lsi_tx_phy_stat2;		/* 0x00z728 */
+		uint64_t	ce_pad_00z730[10];  /* 0x00z730 -- 0x00z77F */
+		uint64_t	ce_lsi_ltssm_cfg1;		/* 0x00z780 */
+		uint64_t	ce_lsi_ltssm_cfg2;		/* 0x00z788 */
+		uint64_t	ce_lsi_ltssm_cfg3;		/* 0x00z790 */
+		uint64_t	ce_lsi_ltssm_cfg4;		/* 0x00z798 */
+		uint64_t	ce_lsi_ltssm_cfg5;		/* 0x00z7A0 */
+		uint64_t	ce_lsi_ltssm_stat1;		/* 0x00z7A8 */
+		uint64_t	ce_lsi_ltssm_stat2;		/* 0x00z7B0 */
+		uint64_t	ce_lsi_ltssm_int_stat;		/* 0x00z7B8 */
+		uint64_t	ce_lsi_ltssm_int_stat_test;	/* 0x00z7C0 */
+		uint64_t	ce_lsi_ltssm_int_mask;		/* 0x00z7C8 */
+		uint64_t	ce_lsi_ltssm_stat_wr_en;	/* 0x00z7D0 */
+		uint64_t	ce_pad_00z7D8[5];   /* 0x00z7D8 -- 0x00z7F8 */
+		uint64_t	ce_lsi_gb_cfg1;			/* 0x00z800 */
+		uint64_t	ce_lsi_gb_cfg2;			/* 0x00z808 */
+		uint64_t	ce_lsi_gb_cfg3;			/* 0x00z810 */
+		uint64_t	ce_lsi_gb_cfg4;			/* 0x00z818 */
+		uint64_t	ce_lsi_gb_stat;			/* 0x00z820 */
+		uint64_t	ce_lsi_gb_int_stat;		/* 0x00z828 */
+		uint64_t	ce_lsi_gb_int_stat_test;	/* 0x00z830 */
+		uint64_t	ce_lsi_gb_int_mask;		/* 0x00z838 */
+		uint64_t	ce_lsi_gb_pwr_dn1;		/* 0x00z840 */
+		uint64_t	ce_lsi_gb_pwr_dn2;		/* 0x00z848 */
+		uint64_t	ce_pad_00z850[246]; /* 0x00z850 -- 0x00zFF8 */
+	} ce_lsi[2];
+
+	uint64_t	ce_pad_004000[10];	    /* 0x004000 -- 0x004048 */
+
+	/*
+	 * CRM: Coretalk Receive Module Registers
+	 */
+	uint64_t	ce_crm_debug_mux;			/* 0x004050 */
+	uint64_t	ce_pad_004058;				/* 0x004058 */
+	uint64_t	ce_crm_ssp_err_cmd_wrd;			/* 0x004060 */
+	uint64_t	ce_crm_ssp_err_addr;			/* 0x004068 */
+	uint64_t	ce_crm_ssp_err_syn;			/* 0x004070 */
+
+	uint64_t	ce_pad_004078[499];	    /* 0x004078 -- 0x005008 */
+
+	/*
+         * CXM: Coretalk Xmit Module Registers
+         */
+	uint64_t	ce_cxm_dyn_credit_status;		/* 0x005010 */
+	uint64_t	ce_cxm_last_credit_status;		/* 0x005018 */
+	uint64_t	ce_cxm_credit_limit;			/* 0x005020 */
+	uint64_t	ce_cxm_force_credit;			/* 0x005028 */
+	uint64_t	ce_cxm_disable_bypass;			/* 0x005030 */
+	uint64_t	ce_pad_005038[3];	    /* 0x005038 -- 0x005048 */
+	uint64_t	ce_cxm_debug_mux;			/* 0x005050 */
+
+        uint64_t        ce_pad_005058[501];         /* 0x005058 -- 0x005FF8 */
+
+	/*
+	 * DTL: Downstream Transaction Layer Regs (Link#1 and Link#2)
+	 * DTL: Link#1 MMRs at start at 0x006000, Link#2 MMRs at 0x008000
+	 * DTL: the comment offsets at far right: let 'y' = {6 or 8}
+	 *
+	 * UTL: Downstream Transaction Layer Regs (Link#1 and Link#2)
+	 * UTL: Link#1 MMRs at start at 0x007000, Link#2 MMRs at 0x009000
+	 * UTL: the comment offsets at far right: let 'z' = {7 or 9}
+	 */
+	#define ce_dtl(link_num)	ce_dtl_utl[link_num-1]
+	#define ce_utl(link_num)	ce_dtl_utl[link_num-1]
+	struct ce_dtl_utl_reg {
+		/* DTL */
+		uint64_t	ce_dtl_dtdr_credit_limit;	/* 0x00y000 */
+		uint64_t	ce_dtl_dtdr_credit_force;	/* 0x00y008 */
+		uint64_t	ce_dtl_dyn_credit_status;	/* 0x00y010 */
+		uint64_t	ce_dtl_dtl_last_credit_stat;	/* 0x00y018 */
+		uint64_t	ce_dtl_dtl_ctrl;		/* 0x00y020 */
+		uint64_t	ce_pad_00y028[5];   /* 0x00y028 -- 0x00y048 */
+		uint64_t	ce_dtl_debug_sel;		/* 0x00y050 */
+		uint64_t	ce_pad_00y058[501]; /* 0x00y058 -- 0x00yFF8 */
+
+		/* UTL */
+		uint64_t	ce_utl_utl_ctrl;		/* 0x00z000 */
+		uint64_t	ce_utl_debug_sel;		/* 0x00z008 */
+		uint64_t	ce_pad_00z010[510]; /* 0x00z010 -- 0x00zFF8 */
+	} ce_dtl_utl[2];
+
+	uint64_t	ce_pad_00A000[514];	    /* 0x00A000 -- 0x00B008 */
+
+	/*
+	 * URE: Upstream Request Engine
+         */
+	uint64_t	ce_ure_dyn_credit_status;		/* 0x00B010 */
+	uint64_t	ce_ure_last_credit_status;		/* 0x00B018 */
+	uint64_t	ce_ure_credit_limit;			/* 0x00B020 */
+	uint64_t	ce_pad_00B028;				/* 0x00B028 */
+	uint64_t	ce_ure_control;				/* 0x00B030 */
+	uint64_t	ce_ure_status;				/* 0x00B038 */
+	uint64_t	ce_pad_00B040[2];	    /* 0x00B040 -- 0x00B048 */
+	uint64_t	ce_ure_debug_sel;			/* 0x00B050 */
+	uint64_t	ce_ure_pcie_debug_sel;			/* 0x00B058 */
+	uint64_t	ce_ure_ssp_err_cmd_wrd;			/* 0x00B060 */
+	uint64_t	ce_ure_ssp_err_addr;			/* 0x00B068 */
+	uint64_t	ce_ure_page_map;			/* 0x00B070 */
+	uint64_t	ce_ure_dir_map[TIOCE_NUM_PORTS];	/* 0x00B078 */
+	uint64_t	ce_ure_pipe_sel1;			/* 0x00B088 */
+	uint64_t	ce_ure_pipe_mask1;			/* 0x00B090 */
+	uint64_t	ce_ure_pipe_sel2;			/* 0x00B098 */
+	uint64_t	ce_ure_pipe_mask2;			/* 0x00B0A0 */
+	uint64_t	ce_ure_pcie1_credits_sent;		/* 0x00B0A8 */
+	uint64_t	ce_ure_pcie1_credits_used;		/* 0x00B0B0 */
+	uint64_t	ce_ure_pcie1_credit_limit;		/* 0x00B0B8 */
+	uint64_t	ce_ure_pcie2_credits_sent;		/* 0x00B0C0 */
+	uint64_t	ce_ure_pcie2_credits_used;		/* 0x00B0C8 */
+	uint64_t	ce_ure_pcie2_credit_limit;		/* 0x00B0D0 */
+	uint64_t	ce_ure_pcie_force_credit;		/* 0x00B0D8 */
+	uint64_t	ce_ure_rd_tnum_val;			/* 0x00B0E0 */
+	uint64_t	ce_ure_rd_tnum_rsp_rcvd;		/* 0x00B0E8 */
+	uint64_t	ce_ure_rd_tnum_esent_timer;		/* 0x00B0F0 */
+	uint64_t	ce_ure_rd_tnum_error;			/* 0x00B0F8 */
+	uint64_t	ce_ure_rd_tnum_first_cl;		/* 0x00B100 */
+	uint64_t	ce_ure_rd_tnum_link_buf;		/* 0x00B108 */
+	uint64_t	ce_ure_wr_tnum_val;			/* 0x00B110 */
+	uint64_t	ce_ure_sram_err_addr0;			/* 0x00B118 */
+	uint64_t	ce_ure_sram_err_addr1;			/* 0x00B120 */
+	uint64_t	ce_ure_sram_err_addr2;			/* 0x00B128 */
+	uint64_t	ce_ure_sram_rd_addr0;			/* 0x00B130 */
+	uint64_t	ce_ure_sram_rd_addr1;			/* 0x00B138 */
+	uint64_t	ce_ure_sram_rd_addr2;			/* 0x00B140 */
+	uint64_t	ce_ure_sram_wr_addr0;			/* 0x00B148 */
+	uint64_t	ce_ure_sram_wr_addr1;			/* 0x00B150 */
+	uint64_t	ce_ure_sram_wr_addr2;			/* 0x00B158 */
+	uint64_t	ce_ure_buf_flush10;			/* 0x00B160 */
+	uint64_t	ce_ure_buf_flush11;			/* 0x00B168 */
+	uint64_t	ce_ure_buf_flush12;			/* 0x00B170 */
+	uint64_t	ce_ure_buf_flush13;			/* 0x00B178 */
+	uint64_t	ce_ure_buf_flush20;			/* 0x00B180 */
+	uint64_t	ce_ure_buf_flush21;			/* 0x00B188 */
+	uint64_t	ce_ure_buf_flush22;			/* 0x00B190 */
+	uint64_t	ce_ure_buf_flush23;			/* 0x00B198 */
+	uint64_t	ce_ure_pcie_control1;			/* 0x00B1A0 */
+	uint64_t	ce_ure_pcie_control2;			/* 0x00B1A8 */
+
+	uint64_t	ce_pad_00B1B0[458];	    /* 0x00B1B0 -- 0x00BFF8 */
+
+	/* Upstream Data Buffer, Port1 */
+	struct ce_ure_maint_ups_dat1_data {
+		uint64_t	data63_0[512];	    /* 0x00C000 -- 0x00CFF8 */
+		uint64_t	data127_64[512];    /* 0x00D000 -- 0x00DFF8 */
+		uint64_t	parity[512];	    /* 0x00E000 -- 0x00EFF8 */
+	} ce_ure_maint_ups_dat1;
+
+	/* Upstream Header Buffer, Port1 */
+	struct ce_ure_maint_ups_hdr1_data {
+		uint64_t	data63_0[512];	    /* 0x00F000 -- 0x00FFF8 */
+		uint64_t	data127_64[512];    /* 0x010000 -- 0x010FF8 */
+		uint64_t	parity[512];	    /* 0x011000 -- 0x011FF8 */
+	} ce_ure_maint_ups_hdr1;
+
+	/* Upstream Data Buffer, Port2 */
+	struct ce_ure_maint_ups_dat2_data {
+		uint64_t	data63_0[512];	    /* 0x012000 -- 0x012FF8 */
+		uint64_t	data127_64[512];    /* 0x013000 -- 0x013FF8 */
+		uint64_t	parity[512];	    /* 0x014000 -- 0x014FF8 */
+	} ce_ure_maint_ups_dat2;
+
+	/* Upstream Header Buffer, Port2 */
+	struct ce_ure_maint_ups_hdr2_data {
+		uint64_t	data63_0[512];	    /* 0x015000 -- 0x015FF8 */
+		uint64_t	data127_64[512];    /* 0x016000 -- 0x016FF8 */
+		uint64_t	parity[512];	    /* 0x017000 -- 0x017FF8 */
+	} ce_ure_maint_ups_hdr2;
+
+	/* Downstream Data Buffer */
+	struct ce_ure_maint_dns_dat_data {
+		uint64_t	data63_0[512];	    /* 0x018000 -- 0x018FF8 */
+		uint64_t	data127_64[512];    /* 0x019000 -- 0x019FF8 */
+		uint64_t	parity[512];	    /* 0x01A000 -- 0x01AFF8 */
+	} ce_ure_maint_dns_dat;
+
+	/* Downstream Header Buffer */
+	struct	ce_ure_maint_dns_hdr_data {
+		uint64_t	data31_0[64];	    /* 0x01B000 -- 0x01B1F8 */
+		uint64_t	data95_32[64];	    /* 0x01B200 -- 0x01B3F8 */
+		uint64_t	parity[64];	    /* 0x01B400 -- 0x01B5F8 */
+	} ce_ure_maint_dns_hdr;
+
+	/* RCI Buffer Data */
+	struct	ce_ure_maint_rci_data {
+		uint64_t	data41_0[64];	    /* 0x01B600 -- 0x01B7F8 */
+		uint64_t	data69_42[64];	    /* 0x01B800 -- 0x01B9F8 */
+	} ce_ure_maint_rci;
+
+	/* Response Queue */
+	uint64_t	ce_ure_maint_rspq[64];	    /* 0x01BA00 -- 0x01BBF8 */
+
+	uint64_t	ce_pad_01C000[4224];	    /* 0x01BC00 -- 0x023FF8 */
+
+	/* Admin Build-a-Packet Buffer */
+	struct	ce_adm_maint_bap_buf_data {
+		uint64_t	data63_0[258];	    /* 0x024000 -- 0x024808 */
+		uint64_t	data127_64[258];    /* 0x024810 -- 0x025018 */
+		uint64_t	parity[258];	    /* 0x025020 -- 0x025828 */
+	} ce_adm_maint_bap_buf;
+
+	uint64_t	ce_pad_025830[5370];	    /* 0x025830 -- 0x02FFF8 */
+
+	/* URE: 40bit PMU ATE Buffer */		    /* 0x030000 -- 0x037FF8 */
+	uint64_t	ce_ure_ate40[TIOCE_NUM_M40_ATES];
+
+	/* URE: 32/40bit PMU ATE Buffer */	    /* 0x038000 -- 0x03BFF8 */
+	uint64_t	ce_ure_ate3240[TIOCE_NUM_M3240_ATES];
+
+	uint64_t	ce_pad_03C000[2050];	    /* 0x03C000 -- 0x040008 */
+
+	/*
+	 * DRE: Down Stream Request Engine
+         */
+	uint64_t	ce_dre_dyn_credit_status1;		/* 0x040010 */
+	uint64_t	ce_dre_dyn_credit_status2;		/* 0x040018 */
+	uint64_t	ce_dre_last_credit_status1;		/* 0x040020 */
+	uint64_t	ce_dre_last_credit_status2;		/* 0x040028 */
+	uint64_t	ce_dre_credit_limit1;			/* 0x040030 */
+	uint64_t	ce_dre_credit_limit2;			/* 0x040038 */
+	uint64_t	ce_dre_force_credit1;			/* 0x040040 */
+	uint64_t	ce_dre_force_credit2;			/* 0x040048 */
+	uint64_t	ce_dre_debug_mux1;			/* 0x040050 */
+	uint64_t	ce_dre_debug_mux2;			/* 0x040058 */
+	uint64_t	ce_dre_ssp_err_cmd_wrd;			/* 0x040060 */
+	uint64_t	ce_dre_ssp_err_addr;			/* 0x040068 */
+	uint64_t	ce_dre_comp_err_cmd_wrd;		/* 0x040070 */
+	uint64_t	ce_dre_comp_err_addr;			/* 0x040078 */
+	uint64_t	ce_dre_req_status;			/* 0x040080 */
+	uint64_t	ce_dre_config1;				/* 0x040088 */
+	uint64_t	ce_dre_config2;				/* 0x040090 */
+	uint64_t	ce_dre_config_req_status;		/* 0x040098 */
+	uint64_t	ce_pad_0400A0[12];	    /* 0x0400A0 -- 0x0400F8 */
+	uint64_t	ce_dre_dyn_fifo;			/* 0x040100 */
+	uint64_t	ce_pad_040108[3];	    /* 0x040108 -- 0x040118 */
+	uint64_t	ce_dre_last_fifo;			/* 0x040120 */
+
+	uint64_t	ce_pad_040128[27];	    /* 0x040128 -- 0x0401F8 */
+
+	/* DRE Downstream Head Queue */
+	struct	ce_dre_maint_ds_head_queue {
+		uint64_t	data63_0[32];	    /* 0x040200 -- 0x0402F8 */
+		uint64_t	data127_64[32];	    /* 0x040300 -- 0x0403F8 */
+		uint64_t	parity[32];	    /* 0x040400 -- 0x0404F8 */
+	} ce_dre_maint_ds_head_q;
+
+	uint64_t	ce_pad_040500[352];	    /* 0x040500 -- 0x040FF8 */
+
+	/* DRE Downstream Data Queue */
+	struct	ce_dre_maint_ds_data_queue {
+		uint64_t	data63_0[256];	    /* 0x041000 -- 0x0417F8 */
+		uint64_t	ce_pad_041800[256]; /* 0x041800 -- 0x041FF8 */
+		uint64_t	data127_64[256];    /* 0x042000 -- 0x0427F8 */
+		uint64_t	ce_pad_042800[256]; /* 0x042800 -- 0x042FF8 */
+		uint64_t	parity[256];	    /* 0x043000 -- 0x0437F8 */
+		uint64_t	ce_pad_043800[256]; /* 0x043800 -- 0x043FF8 */
+	} ce_dre_maint_ds_data_q;
+
+	/* DRE URE Upstream Response Queue */
+	struct	ce_dre_maint_ure_us_rsp_queue {
+		uint64_t	data63_0[8];	    /* 0x044000 -- 0x044038 */
+		uint64_t	ce_pad_044040[24];  /* 0x044040 -- 0x0440F8 */
+		uint64_t	data127_64[8];      /* 0x044100 -- 0x044138 */
+		uint64_t	ce_pad_044140[24];  /* 0x044140 -- 0x0441F8 */
+		uint64_t	parity[8];	    /* 0x044200 -- 0x044238 */
+		uint64_t	ce_pad_044240[24];  /* 0x044240 -- 0x0442F8 */
+	} ce_dre_maint_ure_us_rsp_q;
+
+	uint64_t 	ce_dre_maint_us_wrt_rsp[32];/* 0x044300 -- 0x0443F8 */
+
+	uint64_t	ce_end_of_struct;			/* 0x044400 */
+} tioce_t;
+
+
+/* ce_adm_int_mask/ce_adm_int_status register bit defines */
+#define CE_ADM_INT_CE_ERROR_SHFT		0
+#define CE_ADM_INT_LSI1_IP_ERROR_SHFT		1
+#define CE_ADM_INT_LSI2_IP_ERROR_SHFT		2
+#define CE_ADM_INT_PCIE_ERROR_SHFT		3
+#define CE_ADM_INT_PORT1_HOTPLUG_EVENT_SHFT	4
+#define CE_ADM_INT_PORT2_HOTPLUG_EVENT_SHFT	5
+#define CE_ADM_INT_PCIE_PORT1_DEV_A_SHFT	6
+#define CE_ADM_INT_PCIE_PORT1_DEV_B_SHFT	7
+#define CE_ADM_INT_PCIE_PORT1_DEV_C_SHFT	8
+#define CE_ADM_INT_PCIE_PORT1_DEV_D_SHFT	9
+#define CE_ADM_INT_PCIE_PORT2_DEV_A_SHFT	10
+#define CE_ADM_INT_PCIE_PORT2_DEV_B_SHFT	11
+#define CE_ADM_INT_PCIE_PORT2_DEV_C_SHFT	12
+#define CE_ADM_INT_PCIE_PORT2_DEV_D_SHFT	13
+#define CE_ADM_INT_PCIE_MSG_SHFT		14 /*see int_dest_14*/
+#define CE_ADM_INT_PCIE_MSG_SLOT_0_SHFT		14
+#define CE_ADM_INT_PCIE_MSG_SLOT_1_SHFT		15
+#define CE_ADM_INT_PCIE_MSG_SLOT_2_SHFT		16
+#define CE_ADM_INT_PCIE_MSG_SLOT_3_SHFT		17
+#define CE_ADM_INT_PORT1_PM_PME_MSG_SHFT	22
+#define CE_ADM_INT_PORT2_PM_PME_MSG_SHFT	23
+
+/* ce_adm_force_int register bit defines */
+#define CE_ADM_FORCE_INT_PCIE_PORT1_DEV_A_SHFT	0
+#define CE_ADM_FORCE_INT_PCIE_PORT1_DEV_B_SHFT	1
+#define CE_ADM_FORCE_INT_PCIE_PORT1_DEV_C_SHFT	2
+#define CE_ADM_FORCE_INT_PCIE_PORT1_DEV_D_SHFT	3
+#define CE_ADM_FORCE_INT_PCIE_PORT2_DEV_A_SHFT	4
+#define CE_ADM_FORCE_INT_PCIE_PORT2_DEV_B_SHFT	5
+#define CE_ADM_FORCE_INT_PCIE_PORT2_DEV_C_SHFT	6
+#define CE_ADM_FORCE_INT_PCIE_PORT2_DEV_D_SHFT	7
+#define CE_ADM_FORCE_INT_ALWAYS_SHFT		8
+
+/* ce_adm_int_dest register bit masks & shifts */
+#define INTR_VECTOR_SHFT			56
+
+/* ce_adm_error_mask and ce_adm_error_summary register bit masks */
+#define CE_ADM_ERR_CRM_SSP_REQ_INVALID			(0x1ULL <<  0)
+#define CE_ADM_ERR_SSP_REQ_HEADER			(0x1ULL <<  1)
+#define CE_ADM_ERR_SSP_RSP_HEADER			(0x1ULL <<  2)
+#define CE_ADM_ERR_SSP_PROTOCOL_ERROR			(0x1ULL <<  3)
+#define CE_ADM_ERR_SSP_SBE				(0x1ULL <<  4)
+#define CE_ADM_ERR_SSP_MBE				(0x1ULL <<  5)
+#define CE_ADM_ERR_CXM_CREDIT_OFLOW			(0x1ULL <<  6)
+#define CE_ADM_ERR_DRE_SSP_REQ_INVAL			(0x1ULL <<  7)
+#define CE_ADM_ERR_SSP_REQ_LONG				(0x1ULL <<  8)
+#define CE_ADM_ERR_SSP_REQ_OFLOW			(0x1ULL <<  9)
+#define CE_ADM_ERR_SSP_REQ_SHORT			(0x1ULL << 10)
+#define CE_ADM_ERR_SSP_REQ_SIDEBAND			(0x1ULL << 11)
+#define CE_ADM_ERR_SSP_REQ_ADDR_ERR			(0x1ULL << 12)
+#define CE_ADM_ERR_SSP_REQ_BAD_BE			(0x1ULL << 13)
+#define CE_ADM_ERR_PCIE_COMPL_TIMEOUT			(0x1ULL << 14)
+#define CE_ADM_ERR_PCIE_UNEXP_COMPL			(0x1ULL << 15)
+#define CE_ADM_ERR_PCIE_ERR_COMPL			(0x1ULL << 16)
+#define CE_ADM_ERR_DRE_CREDIT_OFLOW			(0x1ULL << 17)
+#define CE_ADM_ERR_DRE_SRAM_PE				(0x1ULL << 18)
+#define CE_ADM_ERR_SSP_RSP_INVALID			(0x1ULL << 19)
+#define CE_ADM_ERR_SSP_RSP_LONG				(0x1ULL << 20)
+#define CE_ADM_ERR_SSP_RSP_SHORT			(0x1ULL << 21)
+#define CE_ADM_ERR_SSP_RSP_SIDEBAND			(0x1ULL << 22)
+#define CE_ADM_ERR_URE_SSP_RSP_UNEXP			(0x1ULL << 23)
+#define CE_ADM_ERR_URE_SSP_WR_REQ_TIMEOUT		(0x1ULL << 24)
+#define CE_ADM_ERR_URE_SSP_RD_REQ_TIMEOUT		(0x1ULL << 25)
+#define CE_ADM_ERR_URE_ATE3240_PAGE_FAULT		(0x1ULL << 26)
+#define CE_ADM_ERR_URE_ATE40_PAGE_FAULT			(0x1ULL << 27)
+#define CE_ADM_ERR_URE_CREDIT_OFLOW			(0x1ULL << 28)
+#define CE_ADM_ERR_URE_SRAM_PE				(0x1ULL << 29)
+#define CE_ADM_ERR_ADM_SSP_RSP_UNEXP			(0x1ULL << 30)
+#define CE_ADM_ERR_ADM_SSP_REQ_TIMEOUT			(0x1ULL << 31)
+#define CE_ADM_ERR_MMR_ACCESS_ERROR			(0x1ULL << 32)
+#define CE_ADM_ERR_MMR_ADDR_ERROR			(0x1ULL << 33)
+#define CE_ADM_ERR_ADM_CREDIT_OFLOW			(0x1ULL << 34)
+#define CE_ADM_ERR_ADM_SRAM_PE				(0x1ULL << 35)
+#define CE_ADM_ERR_DTL1_MIN_PDATA_CREDIT_ERR		(0x1ULL << 36)
+#define CE_ADM_ERR_DTL1_INF_COMPL_CRED_UPDT_ERR		(0x1ULL << 37)
+#define CE_ADM_ERR_DTL1_INF_POSTED_CRED_UPDT_ERR	(0x1ULL << 38)
+#define CE_ADM_ERR_DTL1_INF_NPOSTED_CRED_UPDT_ERR	(0x1ULL << 39)
+#define CE_ADM_ERR_DTL1_COMP_HD_CRED_MAX_ERR		(0x1ULL << 40)
+#define CE_ADM_ERR_DTL1_COMP_D_CRED_MAX_ERR		(0x1ULL << 41)
+#define CE_ADM_ERR_DTL1_NPOSTED_HD_CRED_MAX_ERR		(0x1ULL << 42)
+#define CE_ADM_ERR_DTL1_NPOSTED_D_CRED_MAX_ERR		(0x1ULL << 43)
+#define CE_ADM_ERR_DTL1_POSTED_HD_CRED_MAX_ERR		(0x1ULL << 44)
+#define CE_ADM_ERR_DTL1_POSTED_D_CRED_MAX_ERR		(0x1ULL << 45)
+#define CE_ADM_ERR_DTL2_MIN_PDATA_CREDIT_ERR		(0x1ULL << 46)
+#define CE_ADM_ERR_DTL2_INF_COMPL_CRED_UPDT_ERR		(0x1ULL << 47)
+#define CE_ADM_ERR_DTL2_INF_POSTED_CRED_UPDT_ERR	(0x1ULL << 48)
+#define CE_ADM_ERR_DTL2_INF_NPOSTED_CRED_UPDT_ERR	(0x1ULL << 49)
+#define CE_ADM_ERR_DTL2_COMP_HD_CRED_MAX_ERR		(0x1ULL << 50)
+#define CE_ADM_ERR_DTL2_COMP_D_CRED_MAX_ERR		(0x1ULL << 51)
+#define CE_ADM_ERR_DTL2_NPOSTED_HD_CRED_MAX_ERR		(0x1ULL << 52)
+#define CE_ADM_ERR_DTL2_NPOSTED_D_CRED_MAX_ERR		(0x1ULL << 53)
+#define CE_ADM_ERR_DTL2_POSTED_HD_CRED_MAX_ERR		(0x1ULL << 54)
+#define CE_ADM_ERR_DTL2_POSTED_D_CRED_MAX_ERR		(0x1ULL << 55)
+#define CE_ADM_ERR_PORT1_PCIE_COR_ERR			(0x1ULL << 56)
+#define CE_ADM_ERR_PORT1_PCIE_NFAT_ERR			(0x1ULL << 57)
+#define CE_ADM_ERR_PORT1_PCIE_FAT_ERR			(0x1ULL << 58)
+#define CE_ADM_ERR_PORT2_PCIE_COR_ERR			(0x1ULL << 59)
+#define CE_ADM_ERR_PORT2_PCIE_NFAT_ERR			(0x1ULL << 60)
+#define CE_ADM_ERR_PORT2_PCIE_FAT_ERR			(0x1ULL << 61)
+
+/* ce_adm_ure_ups_buf_barrier_flush register bit masks and shifts */
+#define FLUSH_SEL_PORT1_PIPE0_SHFT	0
+#define FLUSH_SEL_PORT1_PIPE1_SHFT	4
+#define FLUSH_SEL_PORT1_PIPE2_SHFT	8
+#define FLUSH_SEL_PORT1_PIPE3_SHFT	12
+#define FLUSH_SEL_PORT2_PIPE0_SHFT	16
+#define FLUSH_SEL_PORT2_PIPE1_SHFT	20
+#define FLUSH_SEL_PORT2_PIPE2_SHFT	24
+#define FLUSH_SEL_PORT2_PIPE3_SHFT	28
+
+/* ce_dre_config1 register bit masks and shifts */
+#define CE_DRE_RO_ENABLE		(0x1ULL << 0)
+#define CE_DRE_DYN_RO_ENABLE		(0x1ULL << 1)
+#define CE_DRE_SUP_CONFIG_COMP_ERROR	(0x1ULL << 2)
+#define CE_DRE_SUP_IO_COMP_ERROR	(0x1ULL << 3)
+#define CE_DRE_ADDR_MODE_SHFT		4
+
+/* ce_dre_config_req_status register bit masks */
+#define CE_DRE_LAST_CONFIG_COMPLETION	(0x7ULL << 0)
+#define CE_DRE_DOWNSTREAM_CONFIG_ERROR	(0x1ULL << 3)
+#define CE_DRE_CONFIG_COMPLETION_VALID	(0x1ULL << 4)
+#define CE_DRE_CONFIG_REQUEST_ACTIVE	(0x1ULL << 5)
+
+/* ce_ure_control register bit masks & shifts */
+#define CE_URE_RD_MRG_ENABLE		(0x1ULL << 0)
+#define CE_URE_WRT_MRG_ENABLE1		(0x1ULL << 4)
+#define CE_URE_WRT_MRG_ENABLE2		(0x1ULL << 5)
+#define CE_URE_RSPQ_BYPASS_DISABLE	(0x1ULL << 24)
+#define CE_URE_UPS_DAT1_PAR_DISABLE	(0x1ULL << 32)
+#define CE_URE_UPS_HDR1_PAR_DISABLE	(0x1ULL << 33)
+#define CE_URE_UPS_DAT2_PAR_DISABLE	(0x1ULL << 34)
+#define CE_URE_UPS_HDR2_PAR_DISABLE	(0x1ULL << 35)
+#define CE_URE_ATE_PAR_DISABLE		(0x1ULL << 36)
+#define CE_URE_RCI_PAR_DISABLE		(0x1ULL << 37)
+#define CE_URE_RSPQ_PAR_DISABLE		(0x1ULL << 38)
+#define CE_URE_DNS_DAT_PAR_DISABLE	(0x1ULL << 39)
+#define CE_URE_DNS_HDR_PAR_DISABLE	(0x1ULL << 40)
+#define CE_URE_MALFORM_DISABLE		(0x1ULL << 44)
+#define CE_URE_UNSUP_DISABLE		(0x1ULL << 45)
+
+/* ce_ure_page_map register bit masks & shifts */
+#define CE_URE_ATE3240_ENABLE		(0x1ULL << 0)
+#define CE_URE_ATE40_ENABLE 		(0x1ULL << 1)
+#define CE_URE_PAGESIZE_SHFT		4
+#define CE_URE_PAGESIZE_MASK		(0x7ULL << CE_URE_PAGESIZE_SHFT)
+#define CE_URE_4K_PAGESIZE		(0x0ULL << CE_URE_PAGESIZE_SHFT)
+#define CE_URE_16K_PAGESIZE		(0x1ULL << CE_URE_PAGESIZE_SHFT)
+#define CE_URE_64K_PAGESIZE		(0x2ULL << CE_URE_PAGESIZE_SHFT)
+#define CE_URE_128K_PAGESIZE		(0x3ULL << CE_URE_PAGESIZE_SHFT)
+#define CE_URE_256K_PAGESIZE		(0x4ULL << CE_URE_PAGESIZE_SHFT)
+
+/* ce_ure_pipe_sel register bit masks & shifts */
+#define PKT_TRAFIC_SHRT			16
+#define BUS_SRC_ID_SHFT			8
+#define DEV_SRC_ID_SHFT			3
+#define FNC_SRC_ID_SHFT			0
+#define CE_URE_TC_MASK			(0x07ULL << PKT_TRAFIC_SHRT)
+#define CE_URE_BUS_MASK			(0xFFULL << BUS_SRC_ID_SHFT)
+#define CE_URE_DEV_MASK			(0x1FULL << DEV_SRC_ID_SHFT)
+#define CE_URE_FNC_MASK			(0x07ULL << FNC_SRC_ID_SHFT)
+#define CE_URE_PIPE_BUS(b)		(((uint64_t)(b) << BUS_SRC_ID_SHFT) & \
+					 CE_URE_BUS_MASK)
+#define CE_URE_PIPE_DEV(d)		(((uint64_t)(d) << DEV_SRC_ID_SHFT) & \
+					 CE_URE_DEV_MASK)
+#define CE_URE_PIPE_FNC(f)		(((uint64_t)(f) << FNC_SRC_ID_SHFT) & \
+					 CE_URE_FNC_MASK)
+
+#define CE_URE_SEL1_SHFT		0
+#define CE_URE_SEL2_SHFT		20
+#define CE_URE_SEL3_SHFT		40
+#define CE_URE_SEL1_MASK		(0x7FFFFULL << CE_URE_SEL1_SHFT)
+#define CE_URE_SEL2_MASK		(0x7FFFFULL << CE_URE_SEL2_SHFT)
+#define CE_URE_SEL3_MASK		(0x7FFFFULL << CE_URE_SEL3_SHFT)
+
+
+/* ce_ure_pipe_mask register bit masks & shifts */
+#define CE_URE_MASK1_SHFT		0
+#define CE_URE_MASK2_SHFT		20
+#define CE_URE_MASK3_SHFT		40
+#define CE_URE_MASK1_MASK		(0x7FFFFULL << CE_URE_MASK1_SHFT)
+#define CE_URE_MASK2_MASK		(0x7FFFFULL << CE_URE_MASK2_SHFT)
+#define CE_URE_MASK3_MASK		(0x7FFFFULL << CE_URE_MASK3_SHFT)
+
+
+/* ce_ure_pcie_control1 register bit masks & shifts */
+#define CE_URE_SI			(0x1ULL << 0)
+#define CE_URE_ELAL_SHFT		4
+#define CE_URE_ELAL_MASK		(0x7ULL << CE_URE_ELAL_SHFT)
+#define CE_URE_ELAL1_SHFT		8
+#define CE_URE_ELAL1_MASK		(0x7ULL << CE_URE_ELAL1_SHFT)
+#define CE_URE_SCC			(0x1ULL << 12)
+#define CE_URE_PN1_SHFT			16
+#define CE_URE_PN1_MASK			(0xFFULL << CE_URE_PN1_SHFT)
+#define CE_URE_PN2_SHFT			24
+#define CE_URE_PN2_MASK			(0xFFULL << CE_URE_PN2_SHFT)
+#define CE_URE_PN1_SET(n)		(((uint64_t)(n) << CE_URE_PN1_SHFT) & \
+					 CE_URE_PN1_MASK)
+#define CE_URE_PN2_SET(n)		(((uint64_t)(n) << CE_URE_PN2_SHFT) & \
+					 CE_URE_PN2_MASK)
+
+/* ce_ure_pcie_control2 register bit masks & shifts */
+#define CE_URE_ABP			(0x1ULL << 0)
+#define CE_URE_PCP			(0x1ULL << 1)
+#define CE_URE_MSP			(0x1ULL << 2)
+#define CE_URE_AIP			(0x1ULL << 3)
+#define CE_URE_PIP			(0x1ULL << 4)
+#define CE_URE_HPS			(0x1ULL << 5)
+#define CE_URE_HPC			(0x1ULL << 6)
+#define CE_URE_SPLV_SHFT		7
+#define CE_URE_SPLV_MASK		(0xFFULL << CE_URE_SPLV_SHFT)
+#define CE_URE_SPLS_SHFT		15
+#define CE_URE_SPLS_MASK		(0x3ULL << CE_URE_SPLS_SHFT)
+#define CE_URE_PSN1_SHFT		19
+#define CE_URE_PSN1_MASK		(0x1FFFULL << CE_URE_PSN1_SHFT)
+#define CE_URE_PSN2_SHFT		32
+#define CE_URE_PSN2_MASK		(0x1FFFULL << CE_URE_PSN2_SHFT)
+#define CE_URE_PSN1_SET(n)		(((uint64_t)(n) << CE_URE_PSN1_SHFT) & \
+					 CE_URE_PSN1_MASK)
+#define CE_URE_PSN2_SET(n)		(((uint64_t)(n) << CE_URE_PSN2_SHFT) & \
+					 CE_URE_PSN2_MASK)
+
+/*
+ * PIO address space ranges for CE
+ */
+
+/* Local CE Registers Space */
+#define CE_PIO_MMR			0x00000000
+#define CE_PIO_MMR_LEN			0x04000000
+
+/* PCI Compatible Config Space */
+#define CE_PIO_CONFIG_SPACE		0x04000000
+#define CE_PIO_CONFIG_SPACE_LEN		0x04000000
+
+/* PCI I/O Space Alias */
+#define CE_PIO_IO_SPACE_ALIAS		0x08000000
+#define CE_PIO_IO_SPACE_ALIAS_LEN	0x08000000
+
+/* PCI Enhanced Config Space */
+#define CE_PIO_E_CONFIG_SPACE		0x10000000
+#define CE_PIO_E_CONFIG_SPACE_LEN	0x10000000
+
+/* PCI I/O Space */
+#define CE_PIO_IO_SPACE			0x100000000
+#define CE_PIO_IO_SPACE_LEN		0x100000000
+
+/* PCI MEM Space */
+#define CE_PIO_MEM_SPACE		0x200000000
+#define CE_PIO_MEM_SPACE_LEN		TIO_HWIN_SIZE
+
+
+/*
+ * CE PCI Enhanced Config Space shifts & masks
+ */
+#define CE_E_CONFIG_BUS_SHFT		20
+#define CE_E_CONFIG_BUS_MASK		(0xFF << CE_E_CONFIG_BUS_SHFT)
+#define CE_E_CONFIG_DEVICE_SHFT		15
+#define CE_E_CONFIG_DEVICE_MASK		(0x1F << CE_E_CONFIG_DEVICE_SHFT)
+#define CE_E_CONFIG_FUNC_SHFT		12
+#define CE_E_CONFIG_FUNC_MASK		(0x7  << CE_E_CONFIG_FUNC_SHFT)
+
+#endif /* __ASM_IA64_SN_TIOCE_H__ */
diff --git a/include/asm-ia64/sn/tioce_provider.h b/include/asm-ia64/sn/tioce_provider.h
new file mode 100644
index 0000000..7f63dec
--- /dev/null
+++ b/include/asm-ia64/sn/tioce_provider.h
@@ -0,0 +1,66 @@
+/**************************************************************************
+ *             Copyright (C) 2005, Silicon Graphics, Inc.                 *
+ *									  *
+ *  These coded instructions, statements, and computer programs	 contain  *
+ *  unpublished	 proprietary  information of Silicon Graphics, Inc., and  *
+ *  are protected by Federal copyright law.  They  may	not be disclosed  *
+ *  to	third  parties	or copied or duplicated in any form, in whole or  *
+ *  in part, without the prior written consent of Silicon Graphics, Inc.  *
+ *									  *
+ **************************************************************************/
+
+#ifndef _ASM_IA64_SN_CE_PROVIDER_H
+#define _ASM_IA64_SN_CE_PROVIDER_H
+
+#include <asm/sn/pcibus_provider_defs.h>
+#include <asm/sn/tioce.h>
+
+/*
+ * Common TIOCE structure shared between the prom and kernel
+ *
+ * DO NOT CHANGE THIS STRUCT WITHOUT MAKING CORRESPONDING CHANGES TO THE
+ * PROM VERSION.
+ */
+struct tioce_common {
+	struct pcibus_bussoft	ce_pcibus;	/* common pciio header */
+
+	uint32_t		ce_rev;
+	uint64_t		ce_kernel_private;
+	uint64_t		ce_prom_private;
+};
+
+struct tioce_kernel {
+	struct tioce_common	*ce_common;
+	spinlock_t		ce_lock;
+	struct list_head	ce_dmamap_list;
+
+	uint64_t		ce_ate40_shadow[TIOCE_NUM_M40_ATES];
+	uint64_t		ce_ate3240_shadow[TIOCE_NUM_M3240_ATES];
+	uint32_t		ce_ate3240_pagesize;
+
+	uint8_t			ce_port1_secondary;
+
+	/* per-port resources */
+	struct {
+		int 		dirmap_refcnt;
+		uint64_t	dirmap_shadow;
+	} ce_port[TIOCE_NUM_PORTS];
+};
+
+struct tioce_dmamap {
+	struct list_head	ce_dmamap_list;	/* headed by tioce_kernel */
+	uint32_t		refcnt;
+
+	uint64_t		nbytes;		/* # bytes mapped */
+
+	uint64_t		ct_start;	/* coretalk start address */
+	uint64_t		pci_start;	/* bus start address */
+
+	uint64_t		*ate_hw;	/* hw ptr of first ate in map */
+	uint64_t		*ate_shadow;	/* shadow ptr of firat ate */
+	uint16_t		ate_count;	/* # ate's in the map */
+};
+
+extern int tioce_init_provider(void);
+
+#endif  /* __ASM_IA64_SN_CE_PROVIDER_H */
diff --git a/include/asm-ia64/spinlock.h b/include/asm-ia64/spinlock.h
index 909936f..d2430aa 100644
--- a/include/asm-ia64/spinlock.h
+++ b/include/asm-ia64/spinlock.h
@@ -93,7 +93,15 @@
 # endif /* CONFIG_MCKINLEY */
 #endif
 }
+
 #define _raw_spin_lock(lock) _raw_spin_lock_flags(lock, 0)
+
+/* Unlock by doing an ordered store and releasing the cacheline with nta */
+static inline void _raw_spin_unlock(spinlock_t *x) {
+	barrier();
+	asm volatile ("st4.rel.nta [%0] = r0\n\t" :: "r"(x));
+}
+
 #else /* !ASM_SUPPORTED */
 #define _raw_spin_lock_flags(lock, flags) _raw_spin_lock(lock)
 # define _raw_spin_lock(x)								\
@@ -109,16 +117,16 @@
 		} while (ia64_spinlock_val);						\
 	}										\
 } while (0)
+#define _raw_spin_unlock(x)	do { barrier(); ((spinlock_t *) x)->lock = 0; } while (0)
 #endif /* !ASM_SUPPORTED */
 
 #define spin_is_locked(x)	((x)->lock != 0)
-#define _raw_spin_unlock(x)	do { barrier(); ((spinlock_t *) x)->lock = 0; } while (0)
 #define _raw_spin_trylock(x)	(cmpxchg_acq(&(x)->lock, 0, 1) == 0)
 #define spin_unlock_wait(x)	do { barrier(); } while ((x)->lock)
 
 typedef struct {
-	volatile unsigned int read_counter	: 31;
-	volatile unsigned int write_lock	:  1;
+	volatile unsigned int read_counter	: 24;
+	volatile unsigned int write_lock	:  8;
 #ifdef CONFIG_PREEMPT
 	unsigned int break_lock;
 #endif
@@ -174,6 +182,13 @@
 	(result == 0);								\
 })
 
+static inline void _raw_write_unlock(rwlock_t *x)
+{
+	u8 *y = (u8 *)x;
+	barrier();
+	asm volatile ("st1.rel.nta [%0] = r0\n\t" :: "r"(y+3) : "memory" );
+}
+
 #else /* !ASM_SUPPORTED */
 
 #define _raw_write_lock(l)								\
@@ -195,14 +210,14 @@
 	(ia64_val == 0);						\
 })
 
+static inline void _raw_write_unlock(rwlock_t *x)
+{
+	barrier();
+	x->write_lock = 0;
+}
+
 #endif /* !ASM_SUPPORTED */
 
 #define _raw_read_trylock(lock) generic_raw_read_trylock(lock)
 
-#define _raw_write_unlock(x)								\
-({											\
-	smp_mb__before_clear_bit();	/* need barrier before releasing lock... */	\
-	clear_bit(31, (x));								\
-})
-
 #endif /*  _ASM_IA64_SPINLOCK_H */
diff --git a/include/asm-ia64/system.h b/include/asm-ia64/system.h
index cd2cf76..33256db 100644
--- a/include/asm-ia64/system.h
+++ b/include/asm-ia64/system.h
@@ -19,12 +19,13 @@
 #include <asm/pal.h>
 #include <asm/percpu.h>
 
-#define GATE_ADDR		__IA64_UL_CONST(0xa000000000000000)
+#define GATE_ADDR		RGN_BASE(RGN_GATE)
+
 /*
  * 0xa000000000000000+2*PERCPU_PAGE_SIZE
  * - 0xa000000000000000+3*PERCPU_PAGE_SIZE remain unmapped (guard page)
  */
-#define KERNEL_START		 __IA64_UL_CONST(0xa000000100000000)
+#define KERNEL_START		 (GATE_ADDR+0x100000000)
 #define PERCPU_ADDR		(-PERCPU_PAGE_SIZE)
 
 #ifndef __ASSEMBLY__