Merge branch 'linus' into stackprotector

Conflicts:
	arch/x86/kernel/Makefile
	include/asm-x86/pda.h

69 files changed, 3930 insertions, 5277 deletions

diff --git a/arch/sparc64/Kconfig b/arch/sparc64/Kconfig
index eb36f3b746b..5446e2a499b 100644
--- a/arch/sparc64/Kconfig
+++ b/arch/sparc64/Kconfig
@@ -11,9 +11,20 @@ config SPARC
 config SPARC64
 	bool
 	default y
+	select HAVE_DYNAMIC_FTRACE
+	select HAVE_FTRACE
 	select HAVE_IDE
 	select HAVE_LMB
 	select HAVE_ARCH_KGDB
+	select USE_GENERIC_SMP_HELPERS if SMP
+	select HAVE_ARCH_TRACEHOOK
+	select ARCH_WANT_OPTIONAL_GPIOLIB
+	select RTC_CLASS
+	select RTC_DRV_M48T59
+	select RTC_DRV_CMOS
+	select RTC_DRV_BQ4802
+	select RTC_DRV_SUN4V
+	select RTC_DRV_STARFIRE
 
 config GENERIC_TIME
 	bool
@@ -27,6 +38,11 @@ config GENERIC_CLOCKEVENTS
 	bool
 	default y
 
+config GENERIC_GPIO
+	bool
+	help
+	  Generic GPIO API support
+
 config 64BIT
 	def_bool y
 
@@ -79,6 +95,10 @@ config GENERIC_HARDIRQS_NO__DO_IRQ
 	bool
 	def_bool y
 
+source "init/Kconfig"
+
+menu "Processor type and features"
+
 choice
 	prompt "Kernel page size"
 	default SPARC64_PAGE_SIZE_8KB
@@ -91,19 +111,11 @@ config SPARC64_PAGE_SIZE_8KB
 	  8KB and 64KB work quite well, since SPARC ELF sections
 	  provide for up to 64KB alignment.
 
-	  Therefore, 512KB and 4MB are for expert hackers only.
-
 	  If you don't know what to do, choose 8KB.
 
 config SPARC64_PAGE_SIZE_64KB
 	bool "64KB"
 
-config SPARC64_PAGE_SIZE_512KB
-	bool "512KB"
-
-config SPARC64_PAGE_SIZE_4MB
-	bool "4MB"
-
 endchoice
 
 config SECCOMP
@@ -134,14 +146,10 @@ config HOTPLUG_CPU
 	  can be controlled through /sys/devices/system/cpu/cpu#.
 	  Say N if you want to disable CPU hotplug.
 
-source "init/Kconfig"
-
 config GENERIC_HARDIRQS
 	bool
 	default y
 
-menu "General machine setup"
-
 source "kernel/time/Kconfig"
 
 config SMP
@@ -189,6 +197,17 @@ config US2E_FREQ
 
 	  If in doubt, say N.
 
+config US3_MC
+	tristate "UltraSPARC-III Memory Controller driver"
+	default y
+	help
+	  This adds a driver for the UltraSPARC-III memory controller.
+	  Loading this driver allows exact mnemonic strings to be
+	  printed in the event of a memory error, so that the faulty DIMM
+	  on the motherboard can be matched to the error.
+
+	  If in doubt, say Y, as this information can be very useful.
+
 # Global things across all Sun machines.
 config GENERIC_LOCKBREAK
 	bool
@@ -223,11 +242,10 @@ config HUGETLB_PAGE_SIZE_4MB
 	bool "4MB"
 
 config HUGETLB_PAGE_SIZE_512K
-	depends on !SPARC64_PAGE_SIZE_4MB && !SPARC64_PAGE_SIZE_512KB
 	bool "512K"
 
 config HUGETLB_PAGE_SIZE_64K
-	depends on !SPARC64_PAGE_SIZE_4MB && !SPARC64_PAGE_SIZE_512KB && !SPARC64_PAGE_SIZE_64KB
+	depends on !SPARC64_PAGE_SIZE_64KB
 	bool "64K"
 
 endchoice
@@ -236,6 +254,7 @@ endmenu
 
 config NUMA
 	bool "NUMA support"
+	depends on SMP
 
 config NODES_SHIFT
 	int
diff --git a/arch/sparc64/Kconfig.debug b/arch/sparc64/Kconfig.debug
index 6a4d28a4076..d6d32d178fc 100644
--- a/arch/sparc64/Kconfig.debug
+++ b/arch/sparc64/Kconfig.debug
@@ -33,7 +33,7 @@ config DEBUG_PAGEALLOC
 
 config MCOUNT
 	bool
-	depends on STACK_DEBUG
+	depends on STACK_DEBUG || FTRACE
 	default y
 
 config FRAME_POINTER
diff --git a/arch/sparc64/Makefile b/arch/sparc64/Makefile
index 4b8f2b084c2..c7214abc0d8 100644
--- a/arch/sparc64/Makefile
+++ b/arch/sparc64/Makefile
@@ -7,9 +7,11 @@
 # Copyright (C) 1998 Jakub Jelinek (jj@ultra.linux.cz)
 #
 
-CHECKFLAGS	+= -D__sparc__ -D__sparc_v9__ -m64
+CHECKFLAGS	+= -D__sparc__ -D__sparc_v9__ -D__arch64__ -m64
 
-CPPFLAGS_vmlinux.lds += -Usparc
+# Undefine sparc when processing vmlinux.lds - it is used
+# And teach CPP we are doing 64 bit builds (for this case)
+CPPFLAGS_vmlinux.lds += -m64 -Usparc
 
 LDFLAGS		:= -m elf64_sparc
 
diff --git a/arch/sparc64/defconfig b/arch/sparc64/defconfig
index 76eb832527f..82cab5cc807 100644
--- a/arch/sparc64/defconfig
+++ b/arch/sparc64/defconfig
@@ -1,7 +1,7 @@
 #
 # Automatically generated make config: don't edit
-# Linux kernel version: 2.6.26-rc2
-# Fri May 16 13:36:07 2008
+# Linux kernel version: 2.6.26
+# Fri Jul 18 00:47:07 2008
 #
 CONFIG_SPARC=y
 CONFIG_SPARC64=y
@@ -22,18 +22,6 @@ CONFIG_HAVE_SETUP_PER_CPU_AREA=y
 CONFIG_ARCH_NO_VIRT_TO_BUS=y
 CONFIG_OF=y
 CONFIG_GENERIC_HARDIRQS_NO__DO_IRQ=y
-CONFIG_SPARC64_PAGE_SIZE_8KB=y
-# CONFIG_SPARC64_PAGE_SIZE_64KB is not set
-# CONFIG_SPARC64_PAGE_SIZE_512KB is not set
-# CONFIG_SPARC64_PAGE_SIZE_4MB is not set
-CONFIG_SECCOMP=y
-CONFIG_HZ_100=y
-# CONFIG_HZ_250 is not set
-# CONFIG_HZ_300 is not set
-# CONFIG_HZ_1000 is not set
-CONFIG_HZ=100
-# CONFIG_SCHED_HRTICK is not set
-CONFIG_HOTPLUG_CPU=y
 CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
 
 #
@@ -105,6 +93,7 @@ CONFIG_KRETPROBES=y
 CONFIG_HAVE_KPROBES=y
 CONFIG_HAVE_KRETPROBES=y
 # CONFIG_HAVE_DMA_ATTRS is not set
+CONFIG_USE_GENERIC_SMP_HELPERS=y
 CONFIG_PROC_PAGE_MONITOR=y
 CONFIG_SLABINFO=y
 CONFIG_RT_MUTEXES=y
@@ -121,6 +110,7 @@ CONFIG_STOP_MACHINE=y
 CONFIG_BLOCK=y
 CONFIG_BLK_DEV_IO_TRACE=y
 CONFIG_BLK_DEV_BSG=y
+# CONFIG_BLK_DEV_INTEGRITY is not set
 CONFIG_BLOCK_COMPAT=y
 
 #
@@ -136,11 +126,21 @@ CONFIG_DEFAULT_AS=y
 # CONFIG_DEFAULT_NOOP is not set
 CONFIG_DEFAULT_IOSCHED="anticipatory"
 CONFIG_CLASSIC_RCU=y
-CONFIG_GENERIC_HARDIRQS=y
 
 #
-# General machine setup
+# Processor type and features
 #
+CONFIG_SPARC64_PAGE_SIZE_8KB=y
+# CONFIG_SPARC64_PAGE_SIZE_64KB is not set
+CONFIG_SECCOMP=y
+CONFIG_HZ_100=y
+# CONFIG_HZ_250 is not set
+# CONFIG_HZ_300 is not set
+# CONFIG_HZ_1000 is not set
+CONFIG_HZ=100
+# CONFIG_SCHED_HRTICK is not set
+CONFIG_HOTPLUG_CPU=y
+CONFIG_GENERIC_HARDIRQS=y
 CONFIG_TICK_ONESHOT=y
 CONFIG_NO_HZ=y
 CONFIG_HIGH_RES_TIMERS=y
@@ -342,6 +342,8 @@ CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug"
 CONFIG_STANDALONE=y
 # CONFIG_PREVENT_FIRMWARE_BUILD is not set
 CONFIG_FW_LOADER=y
+CONFIG_FIRMWARE_IN_KERNEL=y
+CONFIG_EXTRA_FIRMWARE=""
 # CONFIG_DEBUG_DRIVER is not set
 # CONFIG_DEBUG_DEVRES is not set
 # CONFIG_SYS_HYPERVISOR is not set
@@ -366,6 +368,7 @@ CONFIG_CDROM_PKTCDVD_BUFFERS=8
 CONFIG_CDROM_PKTCDVD_WCACHE=y
 CONFIG_ATA_OVER_ETH=m
 CONFIG_SUNVDC=m
+# CONFIG_BLK_DEV_HD is not set
 CONFIG_MISC_DEVICES=y
 # CONFIG_PHANTOM is not set
 # CONFIG_EEPROM_93CX6 is not set
@@ -379,6 +382,7 @@ CONFIG_BLK_DEV_IDE=y
 #
 # Please see Documentation/ide/ide.txt for help/info on IDE drives
 #
+CONFIG_IDE_TIMINGS=y
 # CONFIG_BLK_DEV_IDE_SATA is not set
 CONFIG_BLK_DEV_IDEDISK=y
 # CONFIG_IDEDISK_MULTI_MODE is not set
@@ -429,8 +433,6 @@ CONFIG_BLK_DEV_ALI15X3=y
 # CONFIG_BLK_DEV_VIA82CXXX is not set
 # CONFIG_BLK_DEV_TC86C001 is not set
 CONFIG_BLK_DEV_IDEDMA=y
-# CONFIG_BLK_DEV_HD_ONLY is not set
-# CONFIG_BLK_DEV_HD is not set
 
 #
 # SCSI device support
@@ -504,6 +506,7 @@ CONFIG_SCSI_LOWLEVEL=y
 # CONFIG_SCSI_DEBUG is not set
 # CONFIG_SCSI_SUNESP is not set
 # CONFIG_SCSI_SRP is not set
+# CONFIG_SCSI_DH is not set
 # CONFIG_ATA is not set
 CONFIG_MD=y
 CONFIG_BLK_DEV_MD=m
@@ -529,6 +532,10 @@ CONFIG_DM_ZERO=m
 #
 # IEEE 1394 (FireWire) support
 #
+
+#
+# Enable only one of the two stacks, unless you know what you are doing
+#
 # CONFIG_FIREWIRE is not set
 # CONFIG_IEEE1394 is not set
 # CONFIG_I2O is not set
@@ -745,7 +752,8 @@ CONFIG_SERIAL_CORE_CONSOLE=y
 CONFIG_UNIX98_PTYS=y
 # CONFIG_LEGACY_PTYS is not set
 # CONFIG_IPMI_HANDLER is not set
-# CONFIG_HW_RANDOM is not set
+CONFIG_HW_RANDOM=m
+CONFIG_HW_RANDOM_N2RNG=m
 # CONFIG_R3964 is not set
 # CONFIG_APPLICOM is not set
 # CONFIG_RAW_DRIVER is not set
@@ -759,38 +767,58 @@ CONFIG_I2C_ALGOBIT=y
 #
 # I2C Hardware Bus support
 #
+
+#
+# PC SMBus host controller drivers
+#
 # CONFIG_I2C_ALI1535 is not set
 # CONFIG_I2C_ALI1563 is not set
 # CONFIG_I2C_ALI15X3 is not set
 # CONFIG_I2C_AMD756 is not set
 # CONFIG_I2C_AMD8111 is not set
 # CONFIG_I2C_I801 is not set
-# CONFIG_I2C_I810 is not set
+# CONFIG_I2C_ISCH is not set
 # CONFIG_I2C_PIIX4 is not set
 # CONFIG_I2C_NFORCE2 is not set
-# CONFIG_I2C_OCORES is not set
-# CONFIG_I2C_PARPORT_LIGHT is not set
-# CONFIG_I2C_PROSAVAGE is not set
-# CONFIG_I2C_SAVAGE4 is not set
-# CONFIG_I2C_SIMTEC is not set
 # CONFIG_I2C_SIS5595 is not set
 # CONFIG_I2C_SIS630 is not set
 # CONFIG_I2C_SIS96X is not set
-# CONFIG_I2C_TAOS_EVM is not set
-# CONFIG_I2C_STUB is not set
-# CONFIG_I2C_TINY_USB is not set
 # CONFIG_I2C_VIA is not set
 # CONFIG_I2C_VIAPRO is not set
+
+#
+# I2C system bus drivers (mostly embedded / system-on-chip)
+#
+# CONFIG_I2C_OCORES is not set
+# CONFIG_I2C_SIMTEC is not set
+
+#
+# External I2C/SMBus adapter drivers
+#
+# CONFIG_I2C_PARPORT_LIGHT is not set
+# CONFIG_I2C_TAOS_EVM is not set
+# CONFIG_I2C_TINY_USB is not set
+
+#
+# Graphics adapter I2C/DDC channel drivers
+#
 # CONFIG_I2C_VOODOO3 is not set
+
+#
+# Other I2C/SMBus bus drivers
+#
 # CONFIG_I2C_PCA_PLATFORM is not set
+# CONFIG_I2C_STUB is not set
 
 #
 # Miscellaneous I2C Chip support
 #
 # CONFIG_DS1682 is not set
+# CONFIG_AT24 is not set
 # CONFIG_SENSORS_EEPROM is not set
 # CONFIG_SENSORS_PCF8574 is not set
 # CONFIG_PCF8575 is not set
+# CONFIG_SENSORS_PCA9539 is not set
 # CONFIG_SENSORS_PCF8591 is not set
 # CONFIG_SENSORS_MAX6875 is not set
 # CONFIG_SENSORS_TSL2550 is not set
@@ -856,6 +884,7 @@ CONFIG_HWMON=y
 # CONFIG_SENSORS_W83627EHF is not set
 # CONFIG_HWMON_DEBUG_CHIP is not set
 # CONFIG_THERMAL is not set
+# CONFIG_THERMAL_HWMON is not set
 # CONFIG_WATCHDOG is not set
 
 #
@@ -985,15 +1014,7 @@ CONFIG_LOGO=y
 # CONFIG_LOGO_LINUX_VGA16 is not set
 # CONFIG_LOGO_LINUX_CLUT224 is not set
 CONFIG_LOGO_SUN_CLUT224=y
-
-#
-# Sound
-#
 CONFIG_SOUND=m
-
-#
-# Advanced Linux Sound Architecture
-#
 CONFIG_SND=m
 CONFIG_SND_TIMER=m
 CONFIG_SND_PCM=m
@@ -1010,21 +1031,17 @@ CONFIG_SND_SUPPORT_OLD_API=y
 CONFIG_SND_VERBOSE_PROCFS=y
 # CONFIG_SND_VERBOSE_PRINTK is not set
 # CONFIG_SND_DEBUG is not set
-
-#
-# Generic devices
-#
+CONFIG_SND_VMASTER=y
 CONFIG_SND_MPU401_UART=m
 CONFIG_SND_AC97_CODEC=m
+CONFIG_SND_DRIVERS=y
 CONFIG_SND_DUMMY=m
 CONFIG_SND_VIRMIDI=m
 CONFIG_SND_MTPAV=m
 # CONFIG_SND_SERIAL_U16550 is not set
 # CONFIG_SND_MPU401 is not set
-
-#
-# PCI devices
-#
+# CONFIG_SND_AC97_POWER_SAVE is not set
+CONFIG_SND_PCI=y
 # CONFIG_SND_AD1889 is not set
 # CONFIG_SND_ALS300 is not set
 CONFIG_SND_ALI5451=m
@@ -1084,37 +1101,14 @@ CONFIG_SND_ALI5451=m
 # CONFIG_SND_VIRTUOSO is not set
 # CONFIG_SND_VX222 is not set
 # CONFIG_SND_YMFPCI is not set
-# CONFIG_SND_AC97_POWER_SAVE is not set
-
-#
-# USB devices
-#
+CONFIG_SND_USB=y
 # CONFIG_SND_USB_AUDIO is not set
 # CONFIG_SND_USB_CAIAQ is not set
-
-#
-# ALSA Sparc devices
-#
+CONFIG_SND_SPARC=y
 # CONFIG_SND_SUN_AMD7930 is not set
 CONFIG_SND_SUN_CS4231=m
 # CONFIG_SND_SUN_DBRI is not set
-
-#
-# System on Chip audio support
-#
 # CONFIG_SND_SOC is not set
-
-#
-# ALSA SoC audio for Freescale SOCs
-#
-
-#
-# SoC Audio for the Texas Instruments OMAP
-#
-
-#
-# Open Sound System
-#
 # CONFIG_SOUND_PRIME is not set
 CONFIG_AC97_BUS=m
 CONFIG_HID_SUPPORT=y
@@ -1167,6 +1161,7 @@ CONFIG_USB_UHCI_HCD=m
 #
 # CONFIG_USB_ACM is not set
 # CONFIG_USB_PRINTER is not set
+# CONFIG_USB_WDM is not set
 
 #
 # NOTE: USB_STORAGE enables SCSI, and 'SCSI disk support'
@@ -1226,6 +1221,7 @@ CONFIG_USB_STORAGE=m
 # CONFIG_USB_TRANCEVIBRATOR is not set
 # CONFIG_USB_IOWARRIOR is not set
 # CONFIG_USB_TEST is not set
+# CONFIG_USB_ISIGHTFW is not set
 # CONFIG_USB_GADGET is not set
 # CONFIG_MMC is not set
 # CONFIG_MEMSTICK is not set
@@ -1420,6 +1416,12 @@ CONFIG_DEBUG_BUGVERBOSE=y
 # CONFIG_BACKTRACE_SELF_TEST is not set
 # CONFIG_LKDTM is not set
 # CONFIG_FAULT_INJECTION is not set
+CONFIG_HAVE_FTRACE=y
+CONFIG_HAVE_DYNAMIC_FTRACE=y
+# CONFIG_FTRACE is not set
+# CONFIG_IRQSOFF_TRACER is not set
+# CONFIG_SCHED_TRACER is not set
+# CONFIG_CONTEXT_SWITCH_TRACER is not set
 # CONFIG_SAMPLES is not set
 CONFIG_HAVE_ARCH_KGDB=y
 # CONFIG_KGDB is not set
@@ -1486,6 +1488,10 @@ CONFIG_CRYPTO_CRC32C=m
 CONFIG_CRYPTO_MD4=y
 CONFIG_CRYPTO_MD5=y
 CONFIG_CRYPTO_MICHAEL_MIC=m
+# CONFIG_CRYPTO_RMD128 is not set
+# CONFIG_CRYPTO_RMD160 is not set
+# CONFIG_CRYPTO_RMD256 is not set
+# CONFIG_CRYPTO_RMD320 is not set
 CONFIG_CRYPTO_SHA1=y
 CONFIG_CRYPTO_SHA256=m
 CONFIG_CRYPTO_SHA512=m
@@ -1527,6 +1533,7 @@ CONFIG_BITREVERSE=y
 # CONFIG_GENERIC_FIND_FIRST_BIT is not set
 CONFIG_CRC_CCITT=m
 CONFIG_CRC16=m
+# CONFIG_CRC_T10DIF is not set
 # CONFIG_CRC_ITU_T is not set
 CONFIG_CRC32=y
 # CONFIG_CRC7 is not set
diff --git a/arch/sparc64/kernel/Makefile b/arch/sparc64/kernel/Makefile
index ec4f5ebb1ca..c0b8009ab19 100644
--- a/arch/sparc64/kernel/Makefile
+++ b/arch/sparc64/kernel/Makefile
@@ -7,15 +7,16 @@ EXTRA_CFLAGS := -Werror
 
 extra-y		:= head.o init_task.o vmlinux.lds
 
-obj-y		:= process.o setup.o cpu.o idprom.o \
+obj-y		:= process.o setup.o cpu.o idprom.o reboot.o \
 		   traps.o auxio.o una_asm.o sysfs.o iommu.o \
 		   irq.o ptrace.o time.o sys_sparc.o signal.o \
-		   unaligned.o central.o pci.o starfire.o \
-		   power.o sbus.o sparc64_ksyms.o chmc.o \
+		   unaligned.o central.o starfire.o \
+		   power.o sbus.o sparc64_ksyms.o ebus.o \
 		   visemul.o prom.o of_device.o hvapi.o sstate.o mdesc.o
 
+obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o
 obj-$(CONFIG_STACKTRACE) += stacktrace.o
-obj-$(CONFIG_PCI)	 += ebus.o pci_common.o \
+obj-$(CONFIG_PCI)	 += pci.o pci_common.o psycho_common.o \
 			    pci_psycho.o pci_sabre.o pci_schizo.o \
 			    pci_sun4v.o pci_sun4v_asm.o pci_fire.o
 obj-$(CONFIG_PCI_MSI)	+= pci_msi.o
@@ -24,6 +25,7 @@ obj-$(CONFIG_COMPAT) += sys32.o sys_sparc32.o signal32.o
 obj-$(CONFIG_MODULES) += module.o
 obj-$(CONFIG_US3_FREQ) += us3_cpufreq.o
 obj-$(CONFIG_US2E_FREQ) += us2e_cpufreq.o
+obj-$(CONFIG_US3_MC) += chmc.o
 obj-$(CONFIG_KPROBES) += kprobes.o
 obj-$(CONFIG_SUN_LDOMS) += ldc.o vio.o viohs.o ds.o
 obj-$(CONFIG_AUDIT) += audit.o
diff --git a/arch/sparc64/kernel/auxio.c b/arch/sparc64/kernel/auxio.c
index c55f0293eac..858beda8652 100644
--- a/arch/sparc64/kernel/auxio.c
+++ b/arch/sparc64/kernel/auxio.c
@@ -9,9 +9,9 @@
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/ioport.h>
+#include <linux/of_device.h>
 
 #include <asm/prom.h>
-#include <asm/of_device.h>
 #include <asm/io.h>
 #include <asm/auxio.h>
 
@@ -109,7 +109,7 @@ void auxio_set_lte(int on)
 	}
 }
 
-static struct of_device_id auxio_match[] = {
+static struct of_device_id __initdata auxio_match[] = {
 	{
 		.name = "auxio",
 	},
diff --git a/arch/sparc64/kernel/central.c b/arch/sparc64/kernel/central.c
index b61b8dfb09c..05f1c916db0 100644
--- a/arch/sparc64/kernel/central.c
+++ b/arch/sparc64/kernel/central.c
@@ -1,461 +1,268 @@
 /* central.c: Central FHC driver for Sunfire/Starfire/Wildfire.
  *
- * Copyright (C) 1997, 1999 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1997, 1999, 2008 David S. Miller (davem@davemloft.net)
  */
 
 #include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/string.h>
-#include <linux/timer.h>
-#include <linux/sched.h>
-#include <linux/delay.h>
 #include <linux/init.h>
-#include <linux/bootmem.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
 
-#include <asm/page.h>
 #include <asm/fhc.h>
-#include <asm/starfire.h>
-
-struct linux_central *central_bus = NULL;
-struct linux_fhc *fhc_list = NULL;
+#include <asm/upa.h>
+
+struct clock_board {
+	void __iomem		*clock_freq_regs;
+	void __iomem		*clock_regs;
+	void __iomem		*clock_ver_reg;
+	int			num_slots;
+	struct resource		leds_resource;
+	struct platform_device	leds_pdev;
+};
+
+struct fhc {
+	void __iomem		*pregs;
+	bool			central;
+	bool			jtag_master;
+	int			board_num;
+	struct resource		leds_resource;
+	struct platform_device	leds_pdev;
+};
+
+static int __devinit clock_board_calc_nslots(struct clock_board *p)
+{
+	u8 reg = upa_readb(p->clock_regs + CLOCK_STAT1) & 0xc0;
 
-#define IS_CENTRAL_FHC(__fhc)	((__fhc) == central_bus->child)
+	switch (reg) {
+	case 0x40:
+		return 16;
 
-static void central_probe_failure(int line)
-{
-	prom_printf("CENTRAL: Critical device probe failure at central.c:%d\n",
-		    line);
-	prom_halt();
-}
+	case 0xc0:
+		return 8;
 
-static void central_ranges_init(struct linux_central *central)
-{
-	struct device_node *dp = central->prom_node;
-	const void *pval;
-	int len;
-	
-	central->num_central_ranges = 0;
-	pval = of_get_property(dp, "ranges", &len);
-	if (pval) {
-		memcpy(central->central_ranges, pval, len);
-		central->num_central_ranges =
-			(len / sizeof(struct linux_prom_ranges));
+	case 0x80:
+		reg = 0;
+		if (p->clock_ver_reg)
+			reg = upa_readb(p->clock_ver_reg);
+		if (reg) {
+			if (reg & 0x80)
+				return 4;
+			else
+				return 5;
+		}
+		/* Fallthrough */
+	default:
+		return 4;
 	}
 }
 
-static void fhc_ranges_init(struct linux_fhc *fhc)
+static int __devinit clock_board_probe(struct of_device *op,
+				       const struct of_device_id *match)
 {
-	struct device_node *dp = fhc->prom_node;
-	const void *pval;
-	int len;
-	
-	fhc->num_fhc_ranges = 0;
-	pval = of_get_property(dp, "ranges", &len);
-	if (pval) {
-		memcpy(fhc->fhc_ranges, pval, len);
-		fhc->num_fhc_ranges =
-			(len / sizeof(struct linux_prom_ranges));
-	}
-}
+	struct clock_board *p = kzalloc(sizeof(*p), GFP_KERNEL);
+	int err = -ENOMEM;
 
-/* Range application routines are exported to various drivers,
- * so do not __init this.
- */
-static void adjust_regs(struct linux_prom_registers *regp, int nregs,
-			struct linux_prom_ranges *rangep, int nranges)
-{
-	int regc, rngc;
-
-	for (regc = 0; regc < nregs; regc++) {
-		for (rngc = 0; rngc < nranges; rngc++)
-			if (regp[regc].which_io == rangep[rngc].ot_child_space)
-				break; /* Fount it */
-		if (rngc == nranges) /* oops */
-			central_probe_failure(__LINE__);
-		regp[regc].which_io = rangep[rngc].ot_parent_space;
-		regp[regc].phys_addr -= rangep[rngc].ot_child_base;
-		regp[regc].phys_addr += rangep[rngc].ot_parent_base;
+	if (!p) {
+		printk(KERN_ERR "clock_board: Cannot allocate struct clock_board\n");
+		goto out;
 	}
-}
 
-/* Apply probed fhc ranges to registers passed, if no ranges return. */
-void apply_fhc_ranges(struct linux_fhc *fhc,
-		      struct linux_prom_registers *regs,
-		      int nregs)
-{
-	if (fhc->num_fhc_ranges)
-		adjust_regs(regs, nregs, fhc->fhc_ranges,
-			    fhc->num_fhc_ranges);
-}
+	p->clock_freq_regs = of_ioremap(&op->resource[0], 0,
+					resource_size(&op->resource[0]),
+					"clock_board_freq");
+	if (!p->clock_freq_regs) {
+		printk(KERN_ERR "clock_board: Cannot map clock_freq_regs\n");
+		goto out_free;
+	}
 
-/* Apply probed central ranges to registers passed, if no ranges return. */
-void apply_central_ranges(struct linux_central *central,
-			  struct linux_prom_registers *regs, int nregs)
-{
-	if (central->num_central_ranges)
-		adjust_regs(regs, nregs, central->central_ranges,
-			    central->num_central_ranges);
-}
+	p->clock_regs = of_ioremap(&op->resource[1], 0,
+				   resource_size(&op->resource[1]),
+				   "clock_board_regs");
+	if (!p->clock_regs) {
+		printk(KERN_ERR "clock_board: Cannot map clock_regs\n");
+		goto out_unmap_clock_freq_regs;
+	}
 
-static void * __init central_alloc_bootmem(unsigned long size)
-{
-	void *ret;
+	if (op->resource[2].flags) {
+		p->clock_ver_reg = of_ioremap(&op->resource[2], 0,
+					      resource_size(&op->resource[2]),
+					      "clock_ver_reg");
+		if (!p->clock_ver_reg) {
+			printk(KERN_ERR "clock_board: Cannot map clock_ver_reg\n");
+			goto out_unmap_clock_regs;
+		}
+	}
 
-	ret = __alloc_bootmem(size, SMP_CACHE_BYTES, 0UL);
-	if (ret != NULL)
-		memset(ret, 0, size);
+	p->num_slots = clock_board_calc_nslots(p);
 
-	return ret;
-}
+	p->leds_resource.start = (unsigned long)
+		(p->clock_regs + CLOCK_CTRL);
+	p->leds_resource.end = p->leds_resource.end;
+	p->leds_resource.name = "leds";
 
-static unsigned long prom_reg_to_paddr(struct linux_prom_registers *r)
-{
-	unsigned long ret = ((unsigned long) r->which_io) << 32;
+	p->leds_pdev.name = "sunfire-clockboard-leds";
+	p->leds_pdev.resource = &p->leds_resource;
+	p->leds_pdev.num_resources = 1;
+	p->leds_pdev.dev.parent = &op->dev;
 
-	return ret | (unsigned long) r->phys_addr;
-}
-
-static void __init probe_other_fhcs(void)
-{
-	struct device_node *dp;
-	const struct linux_prom64_registers *fpregs;
-
-	for_each_node_by_name(dp, "fhc") {
-		struct linux_fhc *fhc;
-		int board;
-		u32 tmp;
-
-		if (dp->parent &&
-		    dp->parent->parent != NULL)
-			continue;
-
-		fhc = (struct linux_fhc *)
-			central_alloc_bootmem(sizeof(struct linux_fhc));
-		if (fhc == NULL)
-			central_probe_failure(__LINE__);
-
-		/* Link it into the FHC chain. */
-		fhc->next = fhc_list;
-		fhc_list = fhc;
-
-		/* Toplevel FHCs have no parent. */
-		fhc->parent = NULL;
-		
-		fhc->prom_node = dp;
-		fhc_ranges_init(fhc);
-
-		/* Non-central FHC's have 64-bit OBP format registers. */
-		fpregs = of_get_property(dp, "reg", NULL);
-		if (!fpregs)
-			central_probe_failure(__LINE__);
-
-		/* Only central FHC needs special ranges applied. */
-		fhc->fhc_regs.pregs = fpregs[0].phys_addr;
-		fhc->fhc_regs.ireg = fpregs[1].phys_addr;
-		fhc->fhc_regs.ffregs = fpregs[2].phys_addr;
-		fhc->fhc_regs.sregs = fpregs[3].phys_addr;
-		fhc->fhc_regs.uregs = fpregs[4].phys_addr;
-		fhc->fhc_regs.tregs = fpregs[5].phys_addr;
-
-		board = of_getintprop_default(dp, "board#", -1);
-		fhc->board = board;
-
-		tmp = upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_JCTRL);
-		if ((tmp & FHC_JTAG_CTRL_MENAB) != 0)
-			fhc->jtag_master = 1;
-		else
-			fhc->jtag_master = 0;
-
-		tmp = upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_ID);
-		printk("FHC(board %d): Version[%x] PartID[%x] Manuf[%x] %s\n",
-		       board,
-		       (tmp & FHC_ID_VERS) >> 28,
-		       (tmp & FHC_ID_PARTID) >> 12,
-		       (tmp & FHC_ID_MANUF) >> 1,
-		       (fhc->jtag_master ? "(JTAG Master)" : ""));
-		
-		/* This bit must be set in all non-central FHC's in
-		 * the system.  When it is clear, this identifies
-		 * the central board.
-		 */
-		tmp = upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
-		tmp |= FHC_CONTROL_IXIST;
-		upa_writel(tmp, fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
+	err = platform_device_register(&p->leds_pdev);
+	if (err) {
+		printk(KERN_ERR "clock_board: Could not register LEDS "
+		       "platform device\n");
+		goto out_unmap_clock_ver_reg;
 	}
-}
 
-static void probe_clock_board(struct linux_central *central,
-			      struct linux_fhc *fhc,
-			      struct device_node *fp)
-{
-	struct device_node *dp;
-	struct linux_prom_registers cregs[3];
-	const struct linux_prom_registers *pr;
-	int nslots, tmp, nregs;
-
-	dp = fp->child;
-	while (dp) {
-		if (!strcmp(dp->name, "clock-board"))
-			break;
-		dp = dp->sibling;
-	}
-	if (!dp)
-		central_probe_failure(__LINE__);
+	printk(KERN_INFO "clock_board: Detected %d slot Enterprise system.\n",
+	       p->num_slots);
 
-	pr = of_get_property(dp, "reg", &nregs);
-	if (!pr)
-		central_probe_failure(__LINE__);
+	err = 0;
+out:
+	return err;
 
-	memcpy(cregs, pr, nregs);
-	nregs /= sizeof(struct linux_prom_registers);
+out_unmap_clock_ver_reg:
+	if (p->clock_ver_reg)
+		of_iounmap(&op->resource[2], p->clock_ver_reg,
+			   resource_size(&op->resource[2]));
 
-	apply_fhc_ranges(fhc, &cregs[0], nregs);
-	apply_central_ranges(central, &cregs[0], nregs);
-	central->cfreg = prom_reg_to_paddr(&cregs[0]);
-	central->clkregs = prom_reg_to_paddr(&cregs[1]);
+out_unmap_clock_regs:
+	of_iounmap(&op->resource[1], p->clock_regs,
+		   resource_size(&op->resource[1]));
 
-	if (nregs == 2)
-		central->clkver = 0UL;
-	else
-		central->clkver = prom_reg_to_paddr(&cregs[2]);
+out_unmap_clock_freq_regs:
+	of_iounmap(&op->resource[0], p->clock_freq_regs,
+		   resource_size(&op->resource[0]));
 
-	tmp = upa_readb(central->clkregs + CLOCK_STAT1);
-	tmp &= 0xc0;
-	switch(tmp) {
-	case 0x40:
-		nslots = 16;
-		break;
-	case 0xc0:
-		nslots = 8;
-		break;
-	case 0x80:
-		if (central->clkver != 0UL &&
-		   upa_readb(central->clkver) != 0) {
-			if ((upa_readb(central->clkver) & 0x80) != 0)
-				nslots = 4;
-			else
-				nslots = 5;
-			break;
-		}
-	default:
-		nslots = 4;
-		break;
-	};
-	central->slots = nslots;
-	printk("CENTRAL: Detected %d slot Enterprise system. cfreg[%02x] cver[%02x]\n",
-	       central->slots, upa_readb(central->cfreg),
-	       (central->clkver ? upa_readb(central->clkver) : 0x00));
+out_free:
+	kfree(p);
+	goto out;
 }
 
-static void ZAP(unsigned long iclr, unsigned long imap)
+static struct of_device_id __initdata clock_board_match[] = {
+	{
+		.name = "clock-board",
+	},
+	{},
+};
+
+static struct of_platform_driver clock_board_driver = {
+	.match_table	= clock_board_match,
+	.probe		= clock_board_probe,
+	.driver		= {
+		.name	= "clock_board",
+	},
+};
+
+static int __devinit fhc_probe(struct of_device *op,
+			       const struct of_device_id *match)
 {
-	u32 imap_tmp;
-
-	upa_writel(0, iclr);
-	upa_readl(iclr);
-	imap_tmp = upa_readl(imap);
-	imap_tmp &= ~(0x80000000);
-	upa_writel(imap_tmp, imap);
-	upa_readl(imap);
-}
+	struct fhc *p = kzalloc(sizeof(*p), GFP_KERNEL);
+	int err = -ENOMEM;
+	u32 reg;
 
-static void init_all_fhc_hw(void)
-{
-	struct linux_fhc *fhc;
-
-	for (fhc = fhc_list; fhc != NULL; fhc = fhc->next) {
-		u32 tmp;
-
-		/* Clear all of the interrupt mapping registers
-		 * just in case OBP left them in a foul state.
-		 */
-		ZAP(fhc->fhc_regs.ffregs + FHC_FFREGS_ICLR,
-		    fhc->fhc_regs.ffregs + FHC_FFREGS_IMAP);
-		ZAP(fhc->fhc_regs.sregs + FHC_SREGS_ICLR,
-		    fhc->fhc_regs.sregs + FHC_SREGS_IMAP);
-		ZAP(fhc->fhc_regs.uregs + FHC_UREGS_ICLR,
-		    fhc->fhc_regs.uregs + FHC_UREGS_IMAP);
-		ZAP(fhc->fhc_regs.tregs + FHC_TREGS_ICLR,
-		    fhc->fhc_regs.tregs + FHC_TREGS_IMAP);
-
-		/* Setup FHC control register. */
-		tmp = upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
-
-		/* All non-central boards have this bit set. */
-		if (! IS_CENTRAL_FHC(fhc))
-			tmp |= FHC_CONTROL_IXIST;
-
-		/* For all FHCs, clear the firmware synchronization
-		 * line and both low power mode enables.
-		 */
-		tmp &= ~(FHC_CONTROL_AOFF | FHC_CONTROL_BOFF |
-			 FHC_CONTROL_SLINE);
-
-		upa_writel(tmp, fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
-		upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
+	if (!p) {
+		printk(KERN_ERR "fhc: Cannot allocate struct fhc\n");
+		goto out;
 	}
 
-}
+	if (!strcmp(op->node->parent->name, "central"))
+		p->central = true;
 
-void __init central_probe(void)
-{
-	struct linux_prom_registers fpregs[6];
-	const struct linux_prom_registers *pr;
-	struct linux_fhc *fhc;
-	struct device_node *dp, *fp;
-	int err;
-
-	dp = of_find_node_by_name(NULL, "central");
-	if (!dp) {
-		if (this_is_starfire)
-			starfire_cpu_setup();
-		return;
+	p->pregs = of_ioremap(&op->resource[0], 0,
+			      resource_size(&op->resource[0]),
+			      "fhc_pregs");
+	if (!p->pregs) {
+		printk(KERN_ERR "fhc: Cannot map pregs\n");
+		goto out_free;
 	}
 
-	/* Ok we got one, grab some memory for software state. */
-	central_bus = (struct linux_central *)
-		central_alloc_bootmem(sizeof(struct linux_central));
-	if (central_bus == NULL)
-		central_probe_failure(__LINE__);
-
-	fhc = (struct linux_fhc *)
-		central_alloc_bootmem(sizeof(struct linux_fhc));
-	if (fhc == NULL)
-		central_probe_failure(__LINE__);
-
-	/* First init central. */
-	central_bus->child = fhc;
-	central_bus->prom_node = dp;
-	central_ranges_init(central_bus);
-
-	/* And then central's FHC. */
-	fhc->next = fhc_list;
-	fhc_list = fhc;
-
-	fhc->parent = central_bus;
-	fp = dp->child;
-	while (fp) {
-		if (!strcmp(fp->name, "fhc"))
-			break;
-		fp = fp->sibling;
+	if (p->central) {
+		reg = upa_readl(p->pregs + FHC_PREGS_BSR);
+		p->board_num = ((reg >> 16) & 1) | ((reg >> 12) & 0x0e);
+	} else {
+		p->board_num = of_getintprop_default(op->node, "board#", -1);
+		if (p->board_num == -1) {
+			printk(KERN_ERR "fhc: No board# property\n");
+			goto out_unmap_pregs;
+		}
+		if (upa_readl(p->pregs + FHC_PREGS_JCTRL) & FHC_JTAG_CTRL_MENAB)
+			p->jtag_master = true;
 	}
-	if (!fp)
-		central_probe_failure(__LINE__);
-
-	fhc->prom_node = fp;
-	fhc_ranges_init(fhc);
-
-	/* Now, map in FHC register set. */
-	pr = of_get_property(fp, "reg", NULL);
-	if (!pr)
-		central_probe_failure(__LINE__);
-	memcpy(fpregs, pr, sizeof(fpregs));
-
-	apply_central_ranges(central_bus, &fpregs[0], 6);
-	
-	fhc->fhc_regs.pregs = prom_reg_to_paddr(&fpregs[0]);
-	fhc->fhc_regs.ireg = prom_reg_to_paddr(&fpregs[1]);
-	fhc->fhc_regs.ffregs = prom_reg_to_paddr(&fpregs[2]);
-	fhc->fhc_regs.sregs = prom_reg_to_paddr(&fpregs[3]);
-	fhc->fhc_regs.uregs = prom_reg_to_paddr(&fpregs[4]);
-	fhc->fhc_regs.tregs = prom_reg_to_paddr(&fpregs[5]);
-
-	/* Obtain board number from board status register, Central's
-	 * FHC lacks "board#" property.
-	 */
-	err = upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_BSR);
-	fhc->board = (((err >> 16) & 0x01) |
-		      ((err >> 12) & 0x0e));
-
-	fhc->jtag_master = 0;
-
-	/* Attach the clock board registers for CENTRAL. */
-	probe_clock_board(central_bus, fhc, fp);
-
-	err = upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_ID);
-	printk("FHC(board %d): Version[%x] PartID[%x] Manuf[%x] (CENTRAL)\n",
-	       fhc->board,
-	       ((err & FHC_ID_VERS) >> 28),
-	       ((err & FHC_ID_PARTID) >> 12),
-	       ((err & FHC_ID_MANUF) >> 1));
-
-	probe_other_fhcs();
-
-	init_all_fhc_hw();
-}
 
-static inline void fhc_ledblink(struct linux_fhc *fhc, int on)
-{
-	u32 tmp;
+	if (!p->central) {
+		p->leds_resource.start = (unsigned long)
+			(p->pregs + FHC_PREGS_CTRL);
+		p->leds_resource.end = p->leds_resource.end;
+		p->leds_resource.name = "leds";
+
+		p->leds_pdev.name = "sunfire-fhc-leds";
+		p->leds_pdev.resource = &p->leds_resource;
+		p->leds_pdev.num_resources = 1;
+		p->leds_pdev.dev.parent = &op->dev;
+
+		err = platform_device_register(&p->leds_pdev);
+		if (err) {
+			printk(KERN_ERR "fhc: Could not register LEDS "
+			       "platform device\n");
+			goto out_unmap_pregs;
+		}
+	}
+	reg = upa_readl(p->pregs + FHC_PREGS_CTRL);
 
-	tmp = upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
+	if (!p->central)
+		reg |= FHC_CONTROL_IXIST;
 
-	/* NOTE: reverse logic on this bit */
-	if (on)
-		tmp &= ~(FHC_CONTROL_RLED);
-	else
-		tmp |= FHC_CONTROL_RLED;
-	tmp &= ~(FHC_CONTROL_AOFF | FHC_CONTROL_BOFF | FHC_CONTROL_SLINE);
+	reg &= ~(FHC_CONTROL_AOFF |
+		 FHC_CONTROL_BOFF |
+		 FHC_CONTROL_SLINE);
 
-	upa_writel(tmp, fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
-	upa_readl(fhc->fhc_regs.pregs + FHC_PREGS_CTRL);
-}
+	upa_writel(reg, p->pregs + FHC_PREGS_CTRL);
+	upa_readl(p->pregs + FHC_PREGS_CTRL);
 
-static inline void central_ledblink(struct linux_central *central, int on)
-{
-	u8 tmp;
-
-	tmp = upa_readb(central->clkregs + CLOCK_CTRL);
+	reg = upa_readl(p->pregs + FHC_PREGS_ID);
+	printk(KERN_INFO "fhc: Board #%d, Version[%x] PartID[%x] Manuf[%x] %s\n",
+	       p->board_num,
+	       (reg & FHC_ID_VERS) >> 28,
+	       (reg & FHC_ID_PARTID) >> 12,
+	       (reg & FHC_ID_MANUF) >> 1,
+	       (p->jtag_master ?
+		"(JTAG Master)" :
+		(p->central ? "(Central)" : "")));
 
-	/* NOTE: reverse logic on this bit */
-	if (on)
-		tmp &= ~(CLOCK_CTRL_RLED);
-	else
-		tmp |= CLOCK_CTRL_RLED;
+	err = 0;
 
-	upa_writeb(tmp, central->clkregs + CLOCK_CTRL);
-	upa_readb(central->clkregs + CLOCK_CTRL);
-}
+out:
+	return err;
 
-static struct timer_list sftimer;
-static int led_state;
+out_unmap_pregs:
+	of_iounmap(&op->resource[0], p->pregs, resource_size(&op->resource[0]));
 
-static void sunfire_timer(unsigned long __ignored)
-{
-	struct linux_fhc *fhc;
-
-	central_ledblink(central_bus, led_state);
-	for (fhc = fhc_list; fhc != NULL; fhc = fhc->next)
-		if (! IS_CENTRAL_FHC(fhc))
-			fhc_ledblink(fhc, led_state);
-	led_state = ! led_state;
-	sftimer.expires = jiffies + (HZ >> 1);
-	add_timer(&sftimer);
+out_free:
+	kfree(p);
+	goto out;
 }
 
-/* After PCI/SBUS busses have been probed, this is called to perform
- * final initialization of all FireHose Controllers in the system.
- */
-void firetruck_init(void)
+static struct of_device_id __initdata fhc_match[] = {
+	{
+		.name = "fhc",
+	},
+	{},
+};
+
+static struct of_platform_driver fhc_driver = {
+	.match_table	= fhc_match,
+	.probe		= fhc_probe,
+	.driver		= {
+		.name	= "fhc",
+	},
+};
+
+static int __init sunfire_init(void)
 {
-	struct linux_central *central = central_bus;
-	u8 ctrl;
-
-	/* No central bus, nothing to do. */
-	if (central == NULL)
-		return;
-
-	/* OBP leaves it on, turn it off so clock board timer LED
-	 * is in sync with FHC ones.
-	 */
-	ctrl = upa_readb(central->clkregs + CLOCK_CTRL);
-	ctrl &= ~(CLOCK_CTRL_RLED);
-	upa_writeb(ctrl, central->clkregs + CLOCK_CTRL);
-
-	led_state = 0;
-	init_timer(&sftimer);
-	sftimer.data = 0;
-	sftimer.function = &sunfire_timer;
-	sftimer.expires = jiffies + (HZ >> 1);
-	add_timer(&sftimer);
+	(void) of_register_driver(&fhc_driver, &of_platform_bus_type);
+	(void) of_register_driver(&clock_board_driver, &of_platform_bus_type);
+	return 0;
 }
+
+subsys_initcall(sunfire_init);
diff --git a/arch/sparc64/kernel/chmc.c b/arch/sparc64/kernel/chmc.c
index 6d4f02e8a4c..3b9f4d6e14a 100644
--- a/arch/sparc64/kernel/chmc.c
+++ b/arch/sparc64/kernel/chmc.c
@@ -1,6 +1,6 @@
-/* memctrlr.c: Driver for UltraSPARC-III memory controller.
+/* chmc.c: Driver for UltraSPARC-III memory controller.
  *
- * Copyright (C) 2001, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 2001, 2007, 2008 David S. Miller (davem@davemloft.net)
  */
 
 #include <linux/module.h>
@@ -13,45 +13,64 @@
 #include <linux/smp.h>
 #include <linux/errno.h>
 #include <linux/init.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
 #include <asm/spitfire.h>
 #include <asm/chmctrl.h>
 #include <asm/cpudata.h>
 #include <asm/oplib.h>
 #include <asm/prom.h>
+#include <asm/head.h>
 #include <asm/io.h>
+#include <asm/memctrl.h>
+
+#define DRV_MODULE_NAME		"chmc"
+#define PFX DRV_MODULE_NAME	": "
+#define DRV_MODULE_VERSION	"0.2"
+
+MODULE_AUTHOR("David S. Miller (davem@davemloft.net)");
+MODULE_DESCRIPTION("UltraSPARC-III memory controller driver");
+MODULE_LICENSE("GPL");
+MODULE_VERSION(DRV_MODULE_VERSION);
+
+static int mc_type;
+#define MC_TYPE_SAFARI		1
+#define MC_TYPE_JBUS		2
+
+static dimm_printer_t us3mc_dimm_printer;
 
 #define CHMCTRL_NDGRPS	2
 #define CHMCTRL_NDIMMS	4
 
-#define DIMMS_PER_MC	(CHMCTRL_NDGRPS * CHMCTRL_NDIMMS)
+#define CHMC_DIMMS_PER_MC	(CHMCTRL_NDGRPS * CHMCTRL_NDIMMS)
 
 /* OBP memory-layout property format. */
-struct obp_map {
+struct chmc_obp_map {
 	unsigned char	dimm_map[144];
 	unsigned char	pin_map[576];
 };
 
 #define DIMM_LABEL_SZ	8
 
-struct obp_mem_layout {
+struct chmc_obp_mem_layout {
 	/* One max 8-byte string label per DIMM.  Usually
 	 * this matches the label on the motherboard where
 	 * that DIMM resides.
 	 */
-	char		dimm_labels[DIMMS_PER_MC][DIMM_LABEL_SZ];
+	char			dimm_labels[CHMC_DIMMS_PER_MC][DIMM_LABEL_SZ];
 
 	/* If symmetric use map[0], else it is
 	 * asymmetric and map[1] should be used.
 	 */
-	char		symmetric;
+	char			symmetric;
 
-	struct obp_map	map[2];
+	struct chmc_obp_map	map[2];
 };
 
 #define CHMCTRL_NBANKS	4
 
-struct bank_info {
-	struct mctrl_info	*mp;
+struct chmc_bank_info {
+	struct chmc		*p;
 	int			bank_id;
 
 	u64			raw_reg;
@@ -65,28 +84,406 @@ struct bank_info {
 	unsigned long		size;
 };
 
-struct mctrl_info {
-	struct list_head	list;
-	int			portid;
+struct chmc {
+	struct list_head		list;
+	int				portid;
+
+	struct chmc_obp_mem_layout	layout_prop;
+	int				layout_size;
+
+	void __iomem			*regs;
 
-	struct obp_mem_layout	layout_prop;
-	int			layout_size;
+	u64				timing_control1;
+	u64				timing_control2;
+	u64				timing_control3;
+	u64				timing_control4;
+	u64				memaddr_control;
 
-	void __iomem		*regs;
+	struct chmc_bank_info		logical_banks[CHMCTRL_NBANKS];
+};
+
+#define JBUSMC_REGS_SIZE		8
+
+#define JB_MC_REG1_DIMM2_BANK3		0x8000000000000000UL
+#define JB_MC_REG1_DIMM1_BANK1		0x4000000000000000UL
+#define JB_MC_REG1_DIMM2_BANK2		0x2000000000000000UL
+#define JB_MC_REG1_DIMM1_BANK0		0x1000000000000000UL
+#define JB_MC_REG1_XOR			0x0000010000000000UL
+#define JB_MC_REG1_ADDR_GEN_2		0x000000e000000000UL
+#define JB_MC_REG1_ADDR_GEN_2_SHIFT	37
+#define JB_MC_REG1_ADDR_GEN_1		0x0000001c00000000UL
+#define JB_MC_REG1_ADDR_GEN_1_SHIFT	34
+#define JB_MC_REG1_INTERLEAVE		0x0000000001800000UL
+#define JB_MC_REG1_INTERLEAVE_SHIFT	23
+#define JB_MC_REG1_DIMM2_PTYPE		0x0000000000200000UL
+#define JB_MC_REG1_DIMM2_PTYPE_SHIFT	21
+#define JB_MC_REG1_DIMM1_PTYPE		0x0000000000100000UL
+#define JB_MC_REG1_DIMM1_PTYPE_SHIFT	20
+
+#define PART_TYPE_X8		0
+#define PART_TYPE_X4		1
+
+#define INTERLEAVE_NONE		0
+#define INTERLEAVE_SAME		1
+#define INTERLEAVE_INTERNAL	2
+#define INTERLEAVE_BOTH		3
+
+#define ADDR_GEN_128MB		0
+#define ADDR_GEN_256MB		1
+#define ADDR_GEN_512MB		2
+#define ADDR_GEN_1GB		3
+
+#define JB_NUM_DIMM_GROUPS	2
+#define JB_NUM_DIMMS_PER_GROUP	2
+#define JB_NUM_DIMMS		(JB_NUM_DIMM_GROUPS * JB_NUM_DIMMS_PER_GROUP)
+
+struct jbusmc_obp_map {
+	unsigned char	dimm_map[18];
+	unsigned char	pin_map[144];
+};
+
+struct jbusmc_obp_mem_layout {
+	/* One max 8-byte string label per DIMM.  Usually
+	 * this matches the label on the motherboard where
+	 * that DIMM resides.
+	 */
+	char		dimm_labels[JB_NUM_DIMMS][DIMM_LABEL_SZ];
+
+	/* If symmetric use map[0], else it is
+	 * asymmetric and map[1] should be used.
+	 */
+	char			symmetric;
+
+	struct jbusmc_obp_map	map;
+
+	char			_pad;
+};
 
-	u64			timing_control1;
-	u64			timing_control2;
-	u64			timing_control3;
-	u64			timing_control4;
-	u64			memaddr_control;
+struct jbusmc_dimm_group {
+	struct jbusmc			*controller;
+	int				index;
+	u64				base_addr;
+	u64				size;
+};
 
-	struct bank_info	logical_banks[CHMCTRL_NBANKS];
+struct jbusmc {
+	void __iomem			*regs;
+	u64				mc_reg_1;
+	u32				portid;
+	struct jbusmc_obp_mem_layout	layout;
+	int				layout_len;
+	int				num_dimm_groups;
+	struct jbusmc_dimm_group	dimm_groups[JB_NUM_DIMM_GROUPS];
+	struct list_head		list;
 };
 
+static DEFINE_SPINLOCK(mctrl_list_lock);
 static LIST_HEAD(mctrl_list);
 
+static void mc_list_add(struct list_head *list)
+{
+	spin_lock(&mctrl_list_lock);
+	list_add(list, &mctrl_list);
+	spin_unlock(&mctrl_list_lock);
+}
+
+static void mc_list_del(struct list_head *list)
+{
+	spin_lock(&mctrl_list_lock);
+	list_del_init(list);
+	spin_unlock(&mctrl_list_lock);
+}
+
+#define SYNDROME_MIN	-1
+#define SYNDROME_MAX	144
+
+/* Covert syndrome code into the way the bits are positioned
+ * on the bus.
+ */
+static int syndrome_to_qword_code(int syndrome_code)
+{
+	if (syndrome_code < 128)
+		syndrome_code += 16;
+	else if (syndrome_code < 128 + 9)
+		syndrome_code -= (128 - 7);
+	else if (syndrome_code < (128 + 9 + 3))
+		syndrome_code -= (128 + 9 - 4);
+	else
+		syndrome_code -= (128 + 9 + 3);
+	return syndrome_code;
+}
+
+/* All this magic has to do with how a cache line comes over the wire
+ * on Safari and JBUS.  A 64-bit line comes over in 1 or more quadword
+ * cycles, each of which transmit ECC/MTAG info as well as the actual
+ * data.
+ */
+#define L2_LINE_SIZE		64
+#define L2_LINE_ADDR_MSK	(L2_LINE_SIZE - 1)
+#define QW_PER_LINE		4
+#define QW_BYTES		(L2_LINE_SIZE / QW_PER_LINE)
+#define QW_BITS			144
+#define SAFARI_LAST_BIT		(576 - 1)
+#define JBUS_LAST_BIT		(144 - 1)
+
+static void get_pin_and_dimm_str(int syndrome_code, unsigned long paddr,
+				 int *pin_p, char **dimm_str_p, void *_prop,
+				 int base_dimm_offset)
+{
+	int qword_code = syndrome_to_qword_code(syndrome_code);
+	int cache_line_offset;
+	int offset_inverse;
+	int dimm_map_index;
+	int map_val;
+
+	if (mc_type == MC_TYPE_JBUS) {
+		struct jbusmc_obp_mem_layout *p = _prop;
+
+		/* JBUS */
+		cache_line_offset = qword_code;
+		offset_inverse = (JBUS_LAST_BIT - cache_line_offset);
+		dimm_map_index = offset_inverse / 8;
+		map_val = p->map.dimm_map[dimm_map_index];
+		map_val = ((map_val >> ((7 - (offset_inverse & 7)))) & 1);
+		*dimm_str_p = p->dimm_labels[base_dimm_offset + map_val];
+		*pin_p = p->map.pin_map[cache_line_offset];
+	} else {
+		struct chmc_obp_mem_layout *p = _prop;
+		struct chmc_obp_map *mp;
+		int qword;
+
+		/* Safari */
+		if (p->symmetric)
+			mp = &p->map[0];
+		else
+			mp = &p->map[1];
+
+		qword = (paddr & L2_LINE_ADDR_MSK) / QW_BYTES;
+		cache_line_offset = ((3 - qword) * QW_BITS) + qword_code;
+		offset_inverse = (SAFARI_LAST_BIT - cache_line_offset);
+		dimm_map_index = offset_inverse >> 2;
+		map_val = mp->dimm_map[dimm_map_index];
+		map_val = ((map_val >> ((3 - (offset_inverse & 3)) << 1)) & 0x3);
+		*dimm_str_p = p->dimm_labels[base_dimm_offset + map_val];
+		*pin_p = mp->pin_map[cache_line_offset];
+	}
+}
+
+static struct jbusmc_dimm_group *jbusmc_find_dimm_group(unsigned long phys_addr)
+{
+	struct jbusmc *p;
+
+	list_for_each_entry(p, &mctrl_list, list) {
+		int i;
+
+		for (i = 0; i < p->num_dimm_groups; i++) {
+			struct jbusmc_dimm_group *dp = &p->dimm_groups[i];
+
+			if (phys_addr < dp->base_addr ||
+			    (dp->base_addr + dp->size) <= phys_addr)
+				continue;
+
+			return dp;
+		}
+	}
+	return NULL;
+}
+
+static int jbusmc_print_dimm(int syndrome_code,
+			     unsigned long phys_addr,
+			     char *buf, int buflen)
+{
+	struct jbusmc_obp_mem_layout *prop;
+	struct jbusmc_dimm_group *dp;
+	struct jbusmc *p;
+	int first_dimm;
+
+	dp = jbusmc_find_dimm_group(phys_addr);
+	if (dp == NULL ||
+	    syndrome_code < SYNDROME_MIN ||
+	    syndrome_code > SYNDROME_MAX) {
+		buf[0] = '?';
+		buf[1] = '?';
+		buf[2] = '?';
+		buf[3] = '\0';
+	}
+	p = dp->controller;
+	prop = &p->layout;
+
+	first_dimm = dp->index * JB_NUM_DIMMS_PER_GROUP;
+
+	if (syndrome_code != SYNDROME_MIN) {
+		char *dimm_str;
+		int pin;
+
+		get_pin_and_dimm_str(syndrome_code, phys_addr, &pin,
+				     &dimm_str, prop, first_dimm);
+		sprintf(buf, "%s, pin %3d", dimm_str, pin);
+	} else {
+		int dimm;
+
+		/* Multi-bit error, we just dump out all the
+		 * dimm labels associated with this dimm group.
+		 */
+		for (dimm = 0; dimm < JB_NUM_DIMMS_PER_GROUP; dimm++) {
+			sprintf(buf, "%s ",
+				prop->dimm_labels[first_dimm + dimm]);
+			buf += strlen(buf);
+		}
+	}
+
+	return 0;
+}
+
+static u64 __devinit jbusmc_dimm_group_size(u64 base,
+					    const struct linux_prom64_registers *mem_regs,
+					    int num_mem_regs)
+{
+	u64 max = base + (8UL * 1024 * 1024 * 1024);
+	u64 max_seen = base;
+	int i;
+
+	for (i = 0; i < num_mem_regs; i++) {
+		const struct linux_prom64_registers *ent;
+		u64 this_base;
+		u64 this_end;
+
+		ent = &mem_regs[i];
+		this_base = ent->phys_addr;
+		this_end = this_base + ent->reg_size;
+		if (base < this_base || base >= this_end)
+			continue;
+		if (this_end > max)
+			this_end = max;
+		if (this_end > max_seen)
+			max_seen = this_end;
+	}
+
+	return max_seen - base;
+}
+
+static void __devinit jbusmc_construct_one_dimm_group(struct jbusmc *p,
+						      unsigned long index,
+						      const struct linux_prom64_registers *mem_regs,
+						      int num_mem_regs)
+{
+	struct jbusmc_dimm_group *dp = &p->dimm_groups[index];
+
+	dp->controller = p;
+	dp->index = index;
+
+	dp->base_addr  = (p->portid * (64UL * 1024 * 1024 * 1024));
+	dp->base_addr += (index * (8UL * 1024 * 1024 * 1024));
+	dp->size = jbusmc_dimm_group_size(dp->base_addr, mem_regs, num_mem_regs);
+}
+
+static void __devinit jbusmc_construct_dimm_groups(struct jbusmc *p,
+						   const struct linux_prom64_registers *mem_regs,
+						   int num_mem_regs)
+{
+	if (p->mc_reg_1 & JB_MC_REG1_DIMM1_BANK0) {
+		jbusmc_construct_one_dimm_group(p, 0, mem_regs, num_mem_regs);
+		p->num_dimm_groups++;
+	}
+	if (p->mc_reg_1 & JB_MC_REG1_DIMM2_BANK2) {
+		jbusmc_construct_one_dimm_group(p, 1, mem_regs, num_mem_regs);
+		p->num_dimm_groups++;
+	}
+}
+
+static int __devinit jbusmc_probe(struct of_device *op,
+				  const struct of_device_id *match)
+{
+	const struct linux_prom64_registers *mem_regs;
+	struct device_node *mem_node;
+	int err, len, num_mem_regs;
+	struct jbusmc *p;
+	const u32 *prop;
+	const void *ml;
+
+	err = -ENODEV;
+	mem_node = of_find_node_by_path("/memory");
+	if (!mem_node) {
+		printk(KERN_ERR PFX "Cannot find /memory node.\n");
+		goto out;
+	}
+	mem_regs = of_get_property(mem_node, "reg", &len);
+	if (!mem_regs) {
+		printk(KERN_ERR PFX "Cannot get reg property of /memory node.\n");
+		goto out;
+	}
+	num_mem_regs = len / sizeof(*mem_regs);
+
+	err = -ENOMEM;
+	p = kzalloc(sizeof(*p), GFP_KERNEL);
+	if (!p) {
+		printk(KERN_ERR PFX "Cannot allocate struct jbusmc.\n");
+		goto out;
+	}
+
+	INIT_LIST_HEAD(&p->list);
+
+	err = -ENODEV;
+	prop = of_get_property(op->node, "portid", &len);
+	if (!prop || len != 4) {
+		printk(KERN_ERR PFX "Cannot find portid.\n");
+		goto out_free;
+	}
+
+	p->portid = *prop;
+
+	prop = of_get_property(op->node, "memory-control-register-1", &len);
+	if (!prop || len != 8) {
+		printk(KERN_ERR PFX "Cannot get memory control register 1.\n");
+		goto out_free;
+	}
+
+	p->mc_reg_1 = ((u64)prop[0] << 32) | (u64) prop[1];
+
+	err = -ENOMEM;
+	p->regs = of_ioremap(&op->resource[0], 0, JBUSMC_REGS_SIZE, "jbusmc");
+	if (!p->regs) {
+		printk(KERN_ERR PFX "Cannot map jbusmc regs.\n");
+		goto out_free;
+	}
+
+	err = -ENODEV;
+	ml = of_get_property(op->node, "memory-layout", &p->layout_len);
+	if (!ml) {
+		printk(KERN_ERR PFX "Cannot get memory layout property.\n");
+		goto out_iounmap;
+	}
+	if (p->layout_len > sizeof(p->layout)) {
+		printk(KERN_ERR PFX "Unexpected memory-layout size %d\n",
+		       p->layout_len);
+		goto out_iounmap;
+	}
+	memcpy(&p->layout, ml, p->layout_len);
+
+	jbusmc_construct_dimm_groups(p, mem_regs, num_mem_regs);
+
+	mc_list_add(&p->list);
+
+	printk(KERN_INFO PFX "UltraSPARC-IIIi memory controller at %s\n",
+	       op->node->full_name);
+
+	dev_set_drvdata(&op->dev, p);
+
+	err = 0;
+
+out:
+	return err;
+
+out_iounmap:
+	of_iounmap(&op->resource[0], p->regs, JBUSMC_REGS_SIZE);
+
+out_free:
+	kfree(p);
+	goto out;
+}
+
 /* Does BANK decode PHYS_ADDR? */
-static int bank_match(struct bank_info *bp, unsigned long phys_addr)
+static int chmc_bank_match(struct chmc_bank_info *bp, unsigned long phys_addr)
 {
 	unsigned long upper_bits = (phys_addr & PA_UPPER_BITS) >> PA_UPPER_BITS_SHIFT;
 	unsigned long lower_bits = (phys_addr & PA_LOWER_BITS) >> PA_LOWER_BITS_SHIFT;
@@ -118,25 +515,18 @@ static int bank_match(struct bank_info *bp, unsigned long phys_addr)
 }
 
 /* Given PHYS_ADDR, search memory controller banks for a match. */
-static struct bank_info *find_bank(unsigned long phys_addr)
+static struct chmc_bank_info *chmc_find_bank(unsigned long phys_addr)
 {
-	struct list_head *mctrl_head = &mctrl_list;
-	struct list_head *mctrl_entry = mctrl_head->next;
+	struct chmc *p;
 
-	for (;;) {
-		struct mctrl_info *mp =
-			list_entry(mctrl_entry, struct mctrl_info, list);
+	list_for_each_entry(p, &mctrl_list, list) {
 		int bank_no;
 
-		if (mctrl_entry == mctrl_head)
-			break;
-		mctrl_entry = mctrl_entry->next;
-
 		for (bank_no = 0; bank_no < CHMCTRL_NBANKS; bank_no++) {
-			struct bank_info *bp;
+			struct chmc_bank_info *bp;
 
-			bp = &mp->logical_banks[bank_no];
-			if (bank_match(bp, phys_addr))
+			bp = &p->logical_banks[bank_no];
+			if (chmc_bank_match(bp, phys_addr))
 				return bp;
 		}
 	}
@@ -145,17 +535,15 @@ static struct bank_info *find_bank(unsigned long phys_addr)
 }
 
 /* This is the main purpose of this driver. */
-#define SYNDROME_MIN	-1
-#define SYNDROME_MAX	144
-int chmc_getunumber(int syndrome_code,
-		    unsigned long phys_addr,
-		    char *buf, int buflen)
+static int chmc_print_dimm(int syndrome_code,
+			   unsigned long phys_addr,
+			   char *buf, int buflen)
 {
-	struct bank_info *bp;
-	struct obp_mem_layout *prop;
+	struct chmc_bank_info *bp;
+	struct chmc_obp_mem_layout *prop;
 	int bank_in_controller, first_dimm;
 
-	bp = find_bank(phys_addr);
+	bp = chmc_find_bank(phys_addr);
 	if (bp == NULL ||
 	    syndrome_code < SYNDROME_MIN ||
 	    syndrome_code > SYNDROME_MAX) {
@@ -166,60 +554,18 @@ int chmc_getunumber(int syndrome_code,
 		return 0;
 	}
 
-	prop = &bp->mp->layout_prop;
+	prop = &bp->p->layout_prop;
 	bank_in_controller = bp->bank_id & (CHMCTRL_NBANKS - 1);
 	first_dimm  = (bank_in_controller & (CHMCTRL_NDGRPS - 1));
 	first_dimm *= CHMCTRL_NDIMMS;
 
 	if (syndrome_code != SYNDROME_MIN) {
-		struct obp_map *map;
-		int qword, where_in_line, where, map_index, map_offset;
-		unsigned int map_val;
+		char *dimm_str;
+		int pin;
 
-		/* Yaay, single bit error so we can figure out
-		 * the exact dimm.
-		 */
-		if (prop->symmetric)
-			map = &prop->map[0];
-		else
-			map = &prop->map[1];
-
-		/* Covert syndrome code into the way the bits are
-		 * positioned on the bus.
-		 */
-		if (syndrome_code < 144 - 16)
-			syndrome_code += 16;
-		else if (syndrome_code < 144)
-			syndrome_code -= (144 - 7);
-		else if (syndrome_code < (144 + 3))
-			syndrome_code -= (144 + 3 - 4);
-		else
-			syndrome_code -= 144 + 3;
-
-		/* All this magic has to do with how a cache line
-		 * comes over the wire on Safari.  A 64-bit line
-		 * comes over in 4 quadword cycles, each of which
-		 * transmit ECC/MTAG info as well as the actual
-		 * data.  144 bits per quadword, 576 total.
-		 */
-#define LINE_SIZE	64
-#define LINE_ADDR_MSK	(LINE_SIZE - 1)
-#define QW_PER_LINE	4
-#define QW_BYTES	(LINE_SIZE / QW_PER_LINE)
-#define QW_BITS		144
-#define LAST_BIT	(576 - 1)
-
-		qword = (phys_addr & LINE_ADDR_MSK) / QW_BYTES;
-		where_in_line = ((3 - qword) * QW_BITS) + syndrome_code;
-		where = (LAST_BIT - where_in_line);
-		map_index = where >> 2;
-		map_offset = where & 0x3;
-		map_val = map->dimm_map[map_index];
-		map_val = ((map_val >> ((3 - map_offset) << 1)) & (2 - 1));
-
-		sprintf(buf, "%s, pin %3d",
-			prop->dimm_labels[first_dimm + map_val],
-			map->pin_map[where_in_line]);
+		get_pin_and_dimm_str(syndrome_code, phys_addr, &pin,
+				     &dimm_str, prop, first_dimm);
+		sprintf(buf, "%s, pin %3d", dimm_str, pin);
 	} else {
 		int dimm;
 
@@ -240,7 +586,7 @@ int chmc_getunumber(int syndrome_code,
  * the code is executing, you must use special ASI load/store else
  * you go through the global mapping.
  */
-static u64 read_mcreg(struct mctrl_info *mp, unsigned long offset)
+static u64 chmc_read_mcreg(struct chmc *p, unsigned long offset)
 {
 	unsigned long ret, this_cpu;
 
@@ -248,14 +594,14 @@ static u64 read_mcreg(struct mctrl_info *mp, unsigned long offset)
 
 	this_cpu = real_hard_smp_processor_id();
 
-	if (mp->portid == this_cpu) {
+	if (p->portid == this_cpu) {
 		__asm__ __volatile__("ldxa	[%1] %2, %0"
 				     : "=r" (ret)
 				     : "r" (offset), "i" (ASI_MCU_CTRL_REG));
 	} else {
 		__asm__ __volatile__("ldxa	[%1] %2, %0"
 				     : "=r" (ret)
-				     : "r" (mp->regs + offset),
+				     : "r" (p->regs + offset),
 				       "i" (ASI_PHYS_BYPASS_EC_E));
 	}
 
@@ -265,178 +611,253 @@ static u64 read_mcreg(struct mctrl_info *mp, unsigned long offset)
 }
 
 #if 0 /* currently unused */
-static void write_mcreg(struct mctrl_info *mp, unsigned long offset, u64 val)
+static void chmc_write_mcreg(struct chmc *p, unsigned long offset, u64 val)
 {
-	if (mp->portid == smp_processor_id()) {
+	if (p->portid == smp_processor_id()) {
 		__asm__ __volatile__("stxa	%0, [%1] %2"
 				     : : "r" (val),
 				         "r" (offset), "i" (ASI_MCU_CTRL_REG));
 	} else {
 		__asm__ __volatile__("ldxa	%0, [%1] %2"
 				     : : "r" (val),
-				         "r" (mp->regs + offset),
+				         "r" (p->regs + offset),
 				         "i" (ASI_PHYS_BYPASS_EC_E));
 	}
 }
 #endif
 
-static void interpret_one_decode_reg(struct mctrl_info *mp, int which_bank, u64 val)
+static void chmc_interpret_one_decode_reg(struct chmc *p, int which_bank, u64 val)
 {
-	struct bank_info *p = &mp->logical_banks[which_bank];
-
-	p->mp = mp;
-	p->bank_id = (CHMCTRL_NBANKS * mp->portid) + which_bank;
-	p->raw_reg = val;
-	p->valid = (val & MEM_DECODE_VALID) >> MEM_DECODE_VALID_SHIFT;
-	p->uk = (val & MEM_DECODE_UK) >> MEM_DECODE_UK_SHIFT;
-	p->um = (val & MEM_DECODE_UM) >> MEM_DECODE_UM_SHIFT;
-	p->lk = (val & MEM_DECODE_LK) >> MEM_DECODE_LK_SHIFT;
-	p->lm = (val & MEM_DECODE_LM) >> MEM_DECODE_LM_SHIFT;
-
-	p->base  =  (p->um);
-	p->base &= ~(p->uk);
-	p->base <<= PA_UPPER_BITS_SHIFT;
-
-	switch(p->lk) {
+	struct chmc_bank_info *bp = &p->logical_banks[which_bank];
+
+	bp->p = p;
+	bp->bank_id = (CHMCTRL_NBANKS * p->portid) + which_bank;
+	bp->raw_reg = val;
+	bp->valid = (val & MEM_DECODE_VALID) >> MEM_DECODE_VALID_SHIFT;
+	bp->uk = (val & MEM_DECODE_UK) >> MEM_DECODE_UK_SHIFT;
+	bp->um = (val & MEM_DECODE_UM) >> MEM_DECODE_UM_SHIFT;
+	bp->lk = (val & MEM_DECODE_LK) >> MEM_DECODE_LK_SHIFT;
+	bp->lm = (val & MEM_DECODE_LM) >> MEM_DECODE_LM_SHIFT;
+
+	bp->base  =  (bp->um);
+	bp->base &= ~(bp->uk);
+	bp->base <<= PA_UPPER_BITS_SHIFT;
+
+	switch(bp->lk) {
 	case 0xf:
 	default:
-		p->interleave = 1;
+		bp->interleave = 1;
 		break;
 
 	case 0xe:
-		p->interleave = 2;
+		bp->interleave = 2;
 		break;
 
 	case 0xc:
-		p->interleave = 4;
+		bp->interleave = 4;
 		break;
 
 	case 0x8:
-		p->interleave = 8;
+		bp->interleave = 8;
 		break;
 
 	case 0x0:
-		p->interleave = 16;
+		bp->interleave = 16;
 		break;
 	};
 
 	/* UK[10] is reserved, and UK[11] is not set for the SDRAM
 	 * bank size definition.
 	 */
-	p->size = (((unsigned long)p->uk &
-		    ((1UL << 10UL) - 1UL)) + 1UL) << PA_UPPER_BITS_SHIFT;
-	p->size /= p->interleave;
+	bp->size = (((unsigned long)bp->uk &
+		     ((1UL << 10UL) - 1UL)) + 1UL) << PA_UPPER_BITS_SHIFT;
+	bp->size /= bp->interleave;
 }
 
-static void fetch_decode_regs(struct mctrl_info *mp)
+static void chmc_fetch_decode_regs(struct chmc *p)
 {
-	if (mp->layout_size == 0)
+	if (p->layout_size == 0)
 		return;
 
-	interpret_one_decode_reg(mp, 0,
-				 read_mcreg(mp, CHMCTRL_DECODE1));
-	interpret_one_decode_reg(mp, 1,
-				 read_mcreg(mp, CHMCTRL_DECODE2));
-	interpret_one_decode_reg(mp, 2,
-				 read_mcreg(mp, CHMCTRL_DECODE3));
-	interpret_one_decode_reg(mp, 3,
-				 read_mcreg(mp, CHMCTRL_DECODE4));
+	chmc_interpret_one_decode_reg(p, 0,
+				      chmc_read_mcreg(p, CHMCTRL_DECODE1));
+	chmc_interpret_one_decode_reg(p, 1,
+				      chmc_read_mcreg(p, CHMCTRL_DECODE2));
+	chmc_interpret_one_decode_reg(p, 2,
+				      chmc_read_mcreg(p, CHMCTRL_DECODE3));
+	chmc_interpret_one_decode_reg(p, 3,
+				      chmc_read_mcreg(p, CHMCTRL_DECODE4));
 }
 
-static int init_one_mctrl(struct device_node *dp)
+static int __devinit chmc_probe(struct of_device *op,
+				const struct of_device_id *match)
 {
-	struct mctrl_info *mp = kzalloc(sizeof(*mp), GFP_KERNEL);
-	int portid = of_getintprop_default(dp, "portid", -1);
-	const struct linux_prom64_registers *regs;
+	struct device_node *dp = op->node;
+	unsigned long ver;
 	const void *pval;
-	int len;
+	int len, portid;
+	struct chmc *p;
+	int err;
+
+	err = -ENODEV;
+	__asm__ ("rdpr %%ver, %0" : "=r" (ver));
+	if ((ver >> 32UL) == __JALAPENO_ID ||
+	    (ver >> 32UL) == __SERRANO_ID)
+		goto out;
 
-	if (!mp)
-		return -1;
+	portid = of_getintprop_default(dp, "portid", -1);
 	if (portid == -1)
-		goto fail;
+		goto out;
 
-	mp->portid = portid;
 	pval = of_get_property(dp, "memory-layout", &len);
-	mp->layout_size = len;
-	if (!pval)
-		mp->layout_size = 0;
-	else {
-		if (mp->layout_size > sizeof(mp->layout_prop))
-			goto fail;
-		memcpy(&mp->layout_prop, pval, len);
+	if (pval && len > sizeof(p->layout_prop)) {
+		printk(KERN_ERR PFX "Unexpected memory-layout property "
+		       "size %d.\n", len);
+		goto out;
 	}
 
-	regs = of_get_property(dp, "reg", NULL);
-	if (!regs || regs->reg_size != 0x48)
-		goto fail;
+	err = -ENOMEM;
+	p = kzalloc(sizeof(*p), GFP_KERNEL);
+	if (!p) {
+		printk(KERN_ERR PFX "Could not allocate struct chmc.\n");
+		goto out;
+	}
 
-	mp->regs = ioremap(regs->phys_addr, regs->reg_size);
-	if (mp->regs == NULL)
-		goto fail;
+	p->portid = portid;
+	p->layout_size = len;
+	if (!pval)
+		p->layout_size = 0;
+	else
+		memcpy(&p->layout_prop, pval, len);
+
+	p->regs = of_ioremap(&op->resource[0], 0, 0x48, "chmc");
+	if (!p->regs) {
+		printk(KERN_ERR PFX "Could not map registers.\n");
+		goto out_free;
+	}
 
-	if (mp->layout_size != 0UL) {
-		mp->timing_control1 = read_mcreg(mp, CHMCTRL_TCTRL1);
-		mp->timing_control2 = read_mcreg(mp, CHMCTRL_TCTRL2);
-		mp->timing_control3 = read_mcreg(mp, CHMCTRL_TCTRL3);
-		mp->timing_control4 = read_mcreg(mp, CHMCTRL_TCTRL4);
-		mp->memaddr_control = read_mcreg(mp, CHMCTRL_MACTRL);
+	if (p->layout_size != 0UL) {
+		p->timing_control1 = chmc_read_mcreg(p, CHMCTRL_TCTRL1);
+		p->timing_control2 = chmc_read_mcreg(p, CHMCTRL_TCTRL2);
+		p->timing_control3 = chmc_read_mcreg(p, CHMCTRL_TCTRL3);
+		p->timing_control4 = chmc_read_mcreg(p, CHMCTRL_TCTRL4);
+		p->memaddr_control = chmc_read_mcreg(p, CHMCTRL_MACTRL);
 	}
 
-	fetch_decode_regs(mp);
+	chmc_fetch_decode_regs(p);
 
-	list_add(&mp->list, &mctrl_list);
+	mc_list_add(&p->list);
 
-	/* Report the device. */
-	printk(KERN_INFO "%s: US3 memory controller at %p [%s]\n",
+	printk(KERN_INFO PFX "UltraSPARC-III memory controller at %s [%s]\n",
 	       dp->full_name,
-	       mp->regs, (mp->layout_size ? "ACTIVE" : "INACTIVE"));
+	       (p->layout_size ? "ACTIVE" : "INACTIVE"));
 
-	return 0;
+	dev_set_drvdata(&op->dev, p);
+
+	err = 0;
+
+out:
+	return err;
+
+out_free:
+	kfree(p);
+	goto out;
+}
 
-fail:
-	if (mp) {
-		if (mp->regs != NULL)
-			iounmap(mp->regs);
-		kfree(mp);
+static int __devinit us3mc_probe(struct of_device *op,
+				const struct of_device_id *match)
+{
+	if (mc_type == MC_TYPE_SAFARI)
+		return chmc_probe(op, match);
+	else if (mc_type == MC_TYPE_JBUS)
+		return jbusmc_probe(op, match);
+	return -ENODEV;
+}
+
+static void __devexit chmc_destroy(struct of_device *op, struct chmc *p)
+{
+	list_del(&p->list);
+	of_iounmap(&op->resource[0], p->regs, 0x48);
+	kfree(p);
+}
+
+static void __devexit jbusmc_destroy(struct of_device *op, struct jbusmc *p)
+{
+	mc_list_del(&p->list);
+	of_iounmap(&op->resource[0], p->regs, JBUSMC_REGS_SIZE);
+	kfree(p);
+}
+
+static int __devexit us3mc_remove(struct of_device *op)
+{
+	void *p = dev_get_drvdata(&op->dev);
+
+	if (p) {
+		if (mc_type == MC_TYPE_SAFARI)
+			chmc_destroy(op, p);
+		else if (mc_type == MC_TYPE_JBUS)
+			jbusmc_destroy(op, p);
 	}
-	return -1;
+	return 0;
+}
+
+static const struct of_device_id us3mc_match[] = {
+	{
+		.name = "memory-controller",
+	},
+	{},
+};
+MODULE_DEVICE_TABLE(of, us3mc_match);
+
+static struct of_platform_driver us3mc_driver = {
+	.name		= "us3mc",
+	.match_table	= us3mc_match,
+	.probe		= us3mc_probe,
+	.remove		= __devexit_p(us3mc_remove),
+};
+
+static inline bool us3mc_platform(void)
+{
+	if (tlb_type == cheetah || tlb_type == cheetah_plus)
+		return true;
+	return false;
 }
 
-static int __init chmc_init(void)
+static int __init us3mc_init(void)
 {
-	struct device_node *dp;
+	unsigned long ver;
+	int ret;
 
-	/* This driver is only for cheetah platforms. */
-	if (tlb_type != cheetah && tlb_type != cheetah_plus)
+	if (!us3mc_platform())
 		return -ENODEV;
 
-	for_each_node_by_name(dp, "memory-controller")
-		init_one_mctrl(dp);
+	__asm__ __volatile__("rdpr %%ver, %0" : "=r" (ver));
+	if ((ver >> 32UL) == __JALAPENO_ID ||
+	    (ver >> 32UL) == __SERRANO_ID) {
+		mc_type = MC_TYPE_JBUS;
+		us3mc_dimm_printer = jbusmc_print_dimm;
+	} else {
+		mc_type = MC_TYPE_SAFARI;
+		us3mc_dimm_printer = chmc_print_dimm;
+	}
 
-	for_each_node_by_name(dp, "mc-us3")
-		init_one_mctrl(dp);
+	ret = register_dimm_printer(us3mc_dimm_printer);
 
-	return 0;
+	if (!ret) {
+		ret = of_register_driver(&us3mc_driver, &of_bus_type);
+		if (ret)
+			unregister_dimm_printer(us3mc_dimm_printer);
+	}
+	return ret;
 }
 
-static void __exit chmc_cleanup(void)
+static void __exit us3mc_cleanup(void)
 {
-	struct list_head *head = &mctrl_list;
-	struct list_head *tmp = head->next;
-
-	for (;;) {
-		struct mctrl_info *p =
-			list_entry(tmp, struct mctrl_info, list);
-		if (tmp == head)
-			break;
-		tmp = tmp->next;
-
-		list_del(&p->list);
-		iounmap(p->regs);
-		kfree(p);
+	if (us3mc_platform()) {
+		unregister_dimm_printer(us3mc_dimm_printer);
+		of_unregister_driver(&us3mc_driver);
 	}
 }
 
-module_init(chmc_init);
-module_exit(chmc_cleanup);
+module_init(us3mc_init);
+module_exit(us3mc_cleanup);
diff --git a/arch/sparc64/kernel/compat_audit.c b/arch/sparc64/kernel/compat_audit.c
index c1979482aa9..c831b0a4e66 100644
--- a/arch/sparc64/kernel/compat_audit.c
+++ b/arch/sparc64/kernel/compat_audit.c
@@ -1,4 +1,4 @@
-#include <asm-sparc/unistd.h>
+#include <asm/unistd_32.h>
 
 unsigned sparc32_dir_class[] = {
 #include <asm-generic/audit_dir_write.h>
diff --git a/arch/sparc64/kernel/cpu.c b/arch/sparc64/kernel/cpu.c
index 0097c08dc60..0c9ac83ed0a 100644
--- a/arch/sparc64/kernel/cpu.c
+++ b/arch/sparc64/kernel/cpu.c
@@ -1,7 +1,7 @@
 /* cpu.c: Dinky routines to look for the kind of Sparc cpu
  *        we are on.
  *
- * Copyright (C) 1996, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1996, 2007, 2008 David S. Miller (davem@davemloft.net)
  */
 
 #include <linux/kernel.h>
@@ -19,53 +19,86 @@
 
 DEFINE_PER_CPU(cpuinfo_sparc, __cpu_data) = { 0 };
 
-struct cpu_iu_info {
-  short manuf;
-  short impl;
-  char* cpu_name;   /* should be enough I hope... */
+struct cpu_chip_info {
+	unsigned short	manuf;
+	unsigned short	impl;
+	const char	*cpu_name;
+	const char	*fp_name;
 };
 
-struct cpu_fp_info {
-  short manuf;
-  short impl;
-  char fpu_vers;
-  char* fp_name;
+static const struct cpu_chip_info cpu_chips[] = {
+	{
+		.manuf		= 0x17,
+		.impl		= 0x10,
+		.cpu_name	= "TI UltraSparc I   (SpitFire)",
+		.fp_name	= "UltraSparc I integrated FPU",
+	},
+	{
+		.manuf		= 0x22,
+		.impl		= 0x10,
+		.cpu_name	= "TI UltraSparc I   (SpitFire)",
+		.fp_name	= "UltraSparc I integrated FPU",
+	},
+	{
+		.manuf		= 0x17,
+		.impl		= 0x11,
+		.cpu_name	= "TI UltraSparc II  (BlackBird)",
+		.fp_name	= "UltraSparc II integrated FPU",
+	},
+	{
+		.manuf		= 0x17,
+		.impl		= 0x12,
+		.cpu_name	= "TI UltraSparc IIi (Sabre)",
+		.fp_name	= "UltraSparc IIi integrated FPU",
+	},
+	{
+		.manuf		= 0x17,
+		.impl		= 0x13,
+		.cpu_name	= "TI UltraSparc IIe (Hummingbird)",
+		.fp_name	= "UltraSparc IIe integrated FPU",
+	},
+	{
+		.manuf		= 0x3e,
+		.impl		= 0x14,
+		.cpu_name	= "TI UltraSparc III (Cheetah)",
+		.fp_name	= "UltraSparc III integrated FPU",
+	},
+	{
+		.manuf		= 0x3e,
+		.impl		= 0x15,
+		.cpu_name	= "TI UltraSparc III+ (Cheetah+)",
+		.fp_name	= "UltraSparc III+ integrated FPU",
+	},
+	{
+		.manuf		= 0x3e,
+		.impl		= 0x16,
+		.cpu_name	= "TI UltraSparc IIIi (Jalapeno)",
+		.fp_name	= "UltraSparc IIIi integrated FPU",
+	},
+	{
+		.manuf		= 0x3e,
+		.impl		= 0x18,
+		.cpu_name	= "TI UltraSparc IV (Jaguar)",
+		.fp_name	= "UltraSparc IV integrated FPU",
+	},
+	{
+		.manuf		= 0x3e,
+		.impl		= 0x19,
+		.cpu_name	= "TI UltraSparc IV+ (Panther)",
+		.fp_name	= "UltraSparc IV+ integrated FPU",
+	},
+	{
+		.manuf		= 0x3e,
+		.impl		= 0x22,
+		.cpu_name	= "TI UltraSparc IIIi+ (Serrano)",
+		.fp_name	= "UltraSparc IIIi+ integrated FPU",
+	},
 };
 
-static struct cpu_fp_info linux_sparc_fpu[] = {
-  { 0x17, 0x10, 0, "UltraSparc I integrated FPU"},
-  { 0x22, 0x10, 0, "UltraSparc I integrated FPU"},
-  { 0x17, 0x11, 0, "UltraSparc II integrated FPU"},
-  { 0x17, 0x12, 0, "UltraSparc IIi integrated FPU"},
-  { 0x17, 0x13, 0, "UltraSparc IIe integrated FPU"},
-  { 0x3e, 0x14, 0, "UltraSparc III integrated FPU"},
-  { 0x3e, 0x15, 0, "UltraSparc III+ integrated FPU"},
-  { 0x3e, 0x16, 0, "UltraSparc IIIi integrated FPU"},
-  { 0x3e, 0x18, 0, "UltraSparc IV integrated FPU"},
-  { 0x3e, 0x19, 0, "UltraSparc IV+ integrated FPU"},
-  { 0x3e, 0x22, 0, "UltraSparc IIIi+ integrated FPU"},
-};
-
-#define NSPARCFPU  ARRAY_SIZE(linux_sparc_fpu)
-
-static struct cpu_iu_info linux_sparc_chips[] = {
-  { 0x17, 0x10, "TI UltraSparc I   (SpitFire)"},
-  { 0x22, 0x10, "TI UltraSparc I   (SpitFire)"},
-  { 0x17, 0x11, "TI UltraSparc II  (BlackBird)"},
-  { 0x17, 0x12, "TI UltraSparc IIi (Sabre)"},
-  { 0x17, 0x13, "TI UltraSparc IIe (Hummingbird)"},
-  { 0x3e, 0x14, "TI UltraSparc III (Cheetah)"},
-  { 0x3e, 0x15, "TI UltraSparc III+ (Cheetah+)"},
-  { 0x3e, 0x16, "TI UltraSparc IIIi (Jalapeno)"},
-  { 0x3e, 0x18, "TI UltraSparc IV (Jaguar)"},
-  { 0x3e, 0x19, "TI UltraSparc IV+ (Panther)"},
-  { 0x3e, 0x22, "TI UltraSparc IIIi+ (Serrano)"},
-};
+#define NSPARCCHIPS ARRAY_SIZE(linux_sparc_chips)
 
-#define NSPARCCHIPS  ARRAY_SIZE(linux_sparc_chips)
-
-char *sparc_cpu_type;
-char *sparc_fpu_type;
+const char *sparc_cpu_type;
+const char *sparc_fpu_type;
 
 static void __init sun4v_cpu_probe(void)
 {
@@ -89,68 +122,45 @@ static void __init sun4v_cpu_probe(void)
 	}
 }
 
-void __init cpu_probe(void)
+static const struct cpu_chip_info * __init find_cpu_chip(unsigned short manuf,
+							 unsigned short impl)
 {
-	unsigned long ver, fpu_vers, manuf, impl, fprs;
 	int i;
-	
-	if (tlb_type == hypervisor) {
-		sun4v_cpu_probe();
-		return;
-	}
 
-	fprs = fprs_read();
-	fprs_write(FPRS_FEF);
-	__asm__ __volatile__ ("rdpr %%ver, %0; stx %%fsr, [%1]"
-			      : "=&r" (ver)
-			      : "r" (&fpu_vers));
-	fprs_write(fprs);
-	
-	manuf = ((ver >> 48) & 0xffff);
-	impl = ((ver >> 32) & 0xffff);
-
-	fpu_vers = ((fpu_vers >> 17) & 0x7);
-
-retry:
-	for (i = 0; i < NSPARCCHIPS; i++) {
-		if (linux_sparc_chips[i].manuf == manuf) {
-			if (linux_sparc_chips[i].impl == impl) {
-				sparc_cpu_type =
-					linux_sparc_chips[i].cpu_name;
-				break;
-			}
-		}
-	}
+	for (i = 0; i < ARRAY_SIZE(cpu_chips); i++) {
+		const struct cpu_chip_info *p = &cpu_chips[i];
 
-	if (i == NSPARCCHIPS) {
- 		/* Maybe it is a cheetah+ derivative, report it as cheetah+
- 		 * in that case until we learn the real names.
- 		 */
- 		if (manuf == 0x3e &&
- 		    impl > 0x15) {
- 			impl = 0x15;
- 			goto retry;
- 		} else {
- 			printk("DEBUG: manuf[%lx] impl[%lx]\n",
- 			       manuf, impl);
- 		}
-		sparc_cpu_type = "Unknown CPU";
+		if (p->manuf == manuf && p->impl == impl)
+			return p;
 	}
+	return NULL;
+}
 
-	for (i = 0; i < NSPARCFPU; i++) {
-		if (linux_sparc_fpu[i].manuf == manuf &&
-		    linux_sparc_fpu[i].impl == impl) {
-			if (linux_sparc_fpu[i].fpu_vers == fpu_vers) {
-				sparc_fpu_type =
-					linux_sparc_fpu[i].fp_name;
-				break;
-			}
+static int __init cpu_type_probe(void)
+{
+	if (tlb_type == hypervisor) {
+		sun4v_cpu_probe();
+	} else {
+		unsigned long ver, manuf, impl;
+		const struct cpu_chip_info *p;
+	
+		__asm__ __volatile__("rdpr %%ver, %0" : "=r" (ver));
+	
+		manuf = ((ver >> 48) & 0xffff);
+		impl = ((ver >> 32) & 0xffff);
+
+		p = find_cpu_chip(manuf, impl);
+		if (p) {
+			sparc_cpu_type = p->cpu_name;
+			sparc_fpu_type = p->fp_name;
+		} else {
+			printk(KERN_ERR "CPU: Unknown chip, manuf[%lx] impl[%lx]\n",
+			       manuf, impl);
+			sparc_cpu_type = "Unknown CPU";
+			sparc_fpu_type = "Unknown FPU";
 		}
 	}
-
-	if (i == NSPARCFPU) {
- 		printk("DEBUG: manuf[%lx] impl[%lx] fsr.vers[%lx]\n",
- 		       manuf, impl, fpu_vers);
-		sparc_fpu_type = "Unknown FPU";
-	}
+	return 0;
 }
+
+arch_initcall(cpu_type_probe);
diff --git a/arch/sparc64/kernel/ds.c b/arch/sparc64/kernel/ds.c
index edb74f5a118..f52e0534d91 100644
--- a/arch/sparc64/kernel/ds.c
+++ b/arch/sparc64/kernel/ds.c
@@ -1,6 +1,6 @@
 /* ds.c: Domain Services driver for Logical Domains
  *
- * Copyright (C) 2007 David S. Miller <davem@davemloft.net>
+ * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
  */
 
 #include <linux/kernel.h>
@@ -159,7 +159,7 @@ static void ds_var_data(struct ds_info *dp,
 			struct ds_cap_state *cp,
 			void *buf, int len);
 
-struct ds_cap_state ds_states_template[] = {
+static struct ds_cap_state ds_states_template[] = {
 	{
 		.service_id	= "md-update",
 		.data		= md_update_data,
@@ -1217,7 +1217,7 @@ static int ds_remove(struct vio_dev *vdev)
 	return 0;
 }
 
-static struct vio_device_id ds_match[] = {
+static struct vio_device_id __initdata ds_match[] = {
 	{
 		.type = "domain-services-port",
 	},
diff --git a/arch/sparc64/kernel/ebus.c b/arch/sparc64/kernel/ebus.c
index c49d0388b79..77dbf6d45fa 100644
--- a/arch/sparc64/kernel/ebus.c
+++ b/arch/sparc64/kernel/ebus.c
@@ -1,5 +1,4 @@
-/*
- * ebus.c: PCI to EBus bridge device.
+/* ebus.c: EBUS DMA library code.
  *
  * Copyright (C) 1997  Eddie C. Dost  (ecd@skynet.be)
  * Copyright (C) 1999  David S. Miller (davem@redhat.com)
@@ -9,23 +8,11 @@
 #include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/init.h>
-#include <linux/slab.h>
-#include <linux/string.h>
 #include <linux/interrupt.h>
 #include <linux/delay.h>
-#include <linux/pci.h>
-
-#include <asm/system.h>
-#include <asm/page.h>
-#include <asm/ebus.h>
-#include <asm/oplib.h>
-#include <asm/prom.h>
-#include <asm/of_device.h>
-#include <asm/bpp.h>
-#include <asm/irq.h>
-#include <asm/io.h>
 
-/* EBUS dma library. */
+#include <asm/ebus_dma.h>
+#include <asm/io.h>
 
 #define EBDMA_CSR	0x00UL	/* Control/Status */
 #define EBDMA_ADDR	0x04UL	/* DMA Address */
@@ -268,283 +255,3 @@ void ebus_dma_enable(struct ebus_dma_info *p, int on)
 	spin_unlock_irqrestore(&p->lock, flags);
 }
 EXPORT_SYMBOL(ebus_dma_enable);
-
-struct linux_ebus *ebus_chain = NULL;
-
-static inline void *ebus_alloc(size_t size)
-{
-	void *mem;
-
-	mem = kzalloc(size, GFP_ATOMIC);
-	if (!mem)
-		panic("ebus_alloc: out of memory");
-	return mem;
-}
-
-static void __init fill_ebus_child(struct device_node *dp,
-				   struct linux_ebus_child *dev,
-				   int non_standard_regs)
-{
-	struct of_device *op;
-	const int *regs;
-	int i, len;
-
-	dev->prom_node = dp;
-	printk(" (%s)", dp->name);
-
-	regs = of_get_property(dp, "reg", &len);
-	if (!regs)
-		dev->num_addrs = 0;
-	else
-		dev->num_addrs = len / sizeof(regs[0]);
-
-	if (non_standard_regs) {
-		/* This is to handle reg properties which are not
-		 * in the parent relative format.  One example are
-		 * children of the i2c device on CompactPCI systems.
-		 *
-		 * So, for such devices we just record the property
-		 * raw in the child resources.
-		 */
-		for (i = 0; i < dev->num_addrs; i++)
-			dev->resource[i].start = regs[i];
-	} else {
-		for (i = 0; i < dev->num_addrs; i++) {
-			int rnum = regs[i];
-			if (rnum >= dev->parent->num_addrs) {
-				prom_printf("UGH: property for %s was %d, need < %d\n",
-					    dp->name, len, dev->parent->num_addrs);
-				prom_halt();
-			}
-			dev->resource[i].start = dev->parent->resource[i].start;
-			dev->resource[i].end = dev->parent->resource[i].end;
-			dev->resource[i].flags = IORESOURCE_MEM;
-			dev->resource[i].name = dp->name;
-		}
-	}
-
-	op = of_find_device_by_node(dp);
-	if (!op) {
-		dev->num_irqs = 0;
-	} else {
-		dev->num_irqs = op->num_irqs;
-		for (i = 0; i < dev->num_irqs; i++)
-			dev->irqs[i] = op->irqs[i];
-	}
-
-	if (!dev->num_irqs) {
-		/*
-		 * Oh, well, some PROMs don't export interrupts
-		 * property to children of EBus devices...
-		 *
-		 * Be smart about PS/2 keyboard and mouse.
-		 */
-		if (!strcmp(dev->parent->prom_node->name, "8042")) {
-			if (!strcmp(dev->prom_node->name, "kb_ps2")) {
-				dev->num_irqs = 1;
-				dev->irqs[0] = dev->parent->irqs[0];
-			} else {
-				dev->num_irqs = 1;
-				dev->irqs[0] = dev->parent->irqs[1];
-			}
-		}
-	}
-}
-
-static int __init child_regs_nonstandard(struct linux_ebus_device *dev)
-{
-	if (!strcmp(dev->prom_node->name, "i2c") ||
-	    !strcmp(dev->prom_node->name, "SUNW,lombus"))
-		return 1;
-	return 0;
-}
-
-static void __init fill_ebus_device(struct device_node *dp, struct linux_ebus_device *dev)
-{
-	struct linux_ebus_child *child;
-	struct dev_archdata *sd;
-	struct of_device *op;
-	int i, len;
-
-	dev->prom_node = dp;
-
-	printk(" [%s", dp->name);
-
-	op = of_find_device_by_node(dp);
-	if (!op) {
-		dev->num_addrs = 0;
-		dev->num_irqs = 0;
-	} else {
-		const int *regs = of_get_property(dp, "reg", &len);
-
-		if (!regs)
-			len = 0;
-		dev->num_addrs = len / sizeof(struct linux_prom_registers);
-
-		for (i = 0; i < dev->num_addrs; i++)
-			memcpy(&dev->resource[i],
-			       &op->resource[i],
-			       sizeof(struct resource));
-
-		dev->num_irqs = op->num_irqs;
-		for (i = 0; i < dev->num_irqs; i++)
-			dev->irqs[i] = op->irqs[i];
-	}
-
-	sd = &dev->ofdev.dev.archdata;
-	sd->prom_node = dp;
-	sd->op = &dev->ofdev;
-	sd->iommu = dev->bus->ofdev.dev.parent->archdata.iommu;
-	sd->stc = dev->bus->ofdev.dev.parent->archdata.stc;
-	sd->numa_node = dev->bus->ofdev.dev.parent->archdata.numa_node;
-
-	dev->ofdev.node = dp;
-	dev->ofdev.dev.parent = &dev->bus->ofdev.dev;
-	dev->ofdev.dev.bus = &ebus_bus_type;
-	sprintf(dev->ofdev.dev.bus_id, "ebus[%08x]", dp->node);
-
-	/* Register with core */
-	if (of_device_register(&dev->ofdev) != 0)
-		printk(KERN_DEBUG "ebus: device registration error for %s!\n",
-		       dp->path_component_name);
-
-	dp = dp->child;
-	if (dp) {
-		printk(" ->");
-		dev->children = ebus_alloc(sizeof(struct linux_ebus_child));
-
-		child = dev->children;
-		child->next = NULL;
-		child->parent = dev;
-		child->bus = dev->bus;
-		fill_ebus_child(dp, child,
-				child_regs_nonstandard(dev));
-
-		while ((dp = dp->sibling) != NULL) {
-			child->next = ebus_alloc(sizeof(struct linux_ebus_child));
-
-			child = child->next;
-			child->next = NULL;
-			child->parent = dev;
-			child->bus = dev->bus;
-			fill_ebus_child(dp, child,
-					child_regs_nonstandard(dev));
-		}
-	}
-	printk("]");
-}
-
-static struct pci_dev *find_next_ebus(struct pci_dev *start, int *is_rio_p)
-{
-	struct pci_dev *pdev = start;
-
-	while ((pdev = pci_get_device(PCI_VENDOR_ID_SUN, PCI_ANY_ID, pdev)))
-		if (pdev->device == PCI_DEVICE_ID_SUN_EBUS ||
-			pdev->device == PCI_DEVICE_ID_SUN_RIO_EBUS)
-			break;
-
-	*is_rio_p = !!(pdev && (pdev->device == PCI_DEVICE_ID_SUN_RIO_EBUS));
-
-	return pdev;
-}
-
-void __init ebus_init(void)
-{
-	struct linux_ebus_device *dev;
-	struct linux_ebus *ebus;
-	struct pci_dev *pdev;
-	struct device_node *dp;
-	int is_rio;
-	int num_ebus = 0;
-
-	pdev = find_next_ebus(NULL, &is_rio);
-	if (!pdev) {
-		printk("ebus: No EBus's found.\n");
-		return;
-	}
-
-	dp = pci_device_to_OF_node(pdev);
-
-	ebus_chain = ebus = ebus_alloc(sizeof(struct linux_ebus));
-	ebus->next = NULL;
-	ebus->is_rio = is_rio;
-
-	while (dp) {
-		struct device_node *child;
-
-		/* SUNW,pci-qfe uses four empty ebuses on it.
-		   I think we should not consider them here,
-		   as they have half of the properties this
-		   code expects and once we do PCI hot-plug,
-		   we'd have to tweak with the ebus_chain
-		   in the runtime after initialization. -jj */
-		if (!dp->child) {
-			pdev = find_next_ebus(pdev, &is_rio);
-			if (!pdev) {
-				if (ebus == ebus_chain) {
-					ebus_chain = NULL;
-					printk("ebus: No EBus's found.\n");
-					return;
-				}
-				break;
-			}
-			ebus->is_rio = is_rio;
-			dp = pci_device_to_OF_node(pdev);
-			continue;
-		}
-		printk("ebus%d:", num_ebus);
-
-		ebus->index = num_ebus;
-		ebus->prom_node = dp;
-		ebus->self = pdev;
-
-		ebus->ofdev.node = dp;
-		ebus->ofdev.dev.parent = &pdev->dev;
-		ebus->ofdev.dev.bus = &ebus_bus_type;
-		sprintf(ebus->ofdev.dev.bus_id, "ebus%d", num_ebus);
-
-		/* Register with core */
-		if (of_device_register(&ebus->ofdev) != 0)
-			printk(KERN_DEBUG "ebus: device registration error for %s!\n",
-			       dp->path_component_name);
-
-
-		child = dp->child;
-		if (!child)
-			goto next_ebus;
-
-		ebus->devices = ebus_alloc(sizeof(struct linux_ebus_device));
-
-		dev = ebus->devices;
-		dev->next = NULL;
-		dev->children = NULL;
-		dev->bus = ebus;
-		fill_ebus_device(child, dev);
-
-		while ((child = child->sibling) != NULL) {
-			dev->next = ebus_alloc(sizeof(struct linux_ebus_device));
-
-			dev = dev->next;
-			dev->next = NULL;
-			dev->children = NULL;
-			dev->bus = ebus;
-			fill_ebus_device(child, dev);
-		}
-
-	next_ebus:
-		printk("\n");
-
-		pdev = find_next_ebus(pdev, &is_rio);
-		if (!pdev)
-			break;
-
-		dp = pci_device_to_OF_node(pdev);
-
-		ebus->next = ebus_alloc(sizeof(struct linux_ebus));
-		ebus = ebus->next;
-		ebus->next = NULL;
-		ebus->is_rio = is_rio;
-		++num_ebus;
-	}
-	pci_dev_put(pdev); /* XXX for the case, when ebusnd is 0, is it OK? */
-}
diff --git a/arch/sparc64/kernel/entry.h b/arch/sparc64/kernel/entry.h
index 32fbab62085..34d7ab5e10d 100644
--- a/arch/sparc64/kernel/entry.h
+++ b/arch/sparc64/kernel/entry.h
@@ -5,8 +5,8 @@
 #include <linux/types.h>
 #include <linux/init.h>
 
-extern char *sparc_cpu_type;
-extern char *sparc_fpu_type;
+extern const char *sparc_cpu_type;
+extern const char *sparc_fpu_type;
 
 extern void __init per_cpu_patch(void);
 extern void __init sun4v_patch(void);
@@ -22,8 +22,8 @@ extern void do_notify_resume(struct pt_regs *regs,
 			     unsigned long orig_i0,
 			     unsigned long thread_info_flags);
 
-extern asmlinkage void syscall_trace(struct pt_regs *regs,
-				     int syscall_exit_p);
+extern asmlinkage int syscall_trace_enter(struct pt_regs *regs);
+extern asmlinkage void syscall_trace_leave(struct pt_regs *regs);
 
 extern void bad_trap_tl1(struct pt_regs *regs, long lvl);
 
diff --git a/arch/sparc64/kernel/ftrace.c b/arch/sparc64/kernel/ftrace.c
new file mode 100644
index 00000000000..4298d0aee71
--- /dev/null
+++ b/arch/sparc64/kernel/ftrace.c
@@ -0,0 +1,94 @@
+#include <linux/spinlock.h>
+#include <linux/hardirq.h>
+#include <linux/ftrace.h>
+#include <linux/percpu.h>
+#include <linux/init.h>
+#include <linux/list.h>
+
+#include <asm/ftrace.h>
+
+static const u32 ftrace_nop = 0x01000000;
+
+notrace unsigned char *ftrace_nop_replace(void)
+{
+	return (char *)&ftrace_nop;
+}
+
+notrace unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
+{
+	static u32 call;
+	s32 off;
+
+	off = ((s32)addr - (s32)ip);
+	call = 0x40000000 | ((u32)off >> 2);
+
+	return (unsigned char *) &call;
+}
+
+notrace int
+ftrace_modify_code(unsigned long ip, unsigned char *old_code,
+		   unsigned char *new_code)
+{
+	u32 old = *(u32 *)old_code;
+	u32 new = *(u32 *)new_code;
+	u32 replaced;
+	int faulted;
+
+	__asm__ __volatile__(
+	"1:	cas	[%[ip]], %[old], %[new]\n"
+	"	flush	%[ip]\n"
+	"	mov	0, %[faulted]\n"
+	"2:\n"
+	"	.section .fixup,#alloc,#execinstr\n"
+	"	.align	4\n"
+	"3:	sethi	%%hi(2b), %[faulted]\n"
+	"	jmpl	%[faulted] + %%lo(2b), %%g0\n"
+	"	 mov	1, %[faulted]\n"
+	"	.previous\n"
+	"	.section __ex_table,\"a\"\n"
+	"	.align	4\n"
+	"	.word	1b, 3b\n"
+	"	.previous\n"
+	: "=r" (replaced), [faulted] "=r" (faulted)
+	: [new] "0" (new), [old] "r" (old), [ip] "r" (ip)
+	: "memory");
+
+	if (replaced != old && replaced != new)
+		faulted = 2;
+
+	return faulted;
+}
+
+notrace int ftrace_update_ftrace_func(ftrace_func_t func)
+{
+	unsigned long ip = (unsigned long)(&ftrace_call);
+	unsigned char old[MCOUNT_INSN_SIZE], *new;
+
+	memcpy(old, &ftrace_call, MCOUNT_INSN_SIZE);
+	new = ftrace_call_replace(ip, (unsigned long)func);
+	return ftrace_modify_code(ip, old, new);
+}
+
+notrace int ftrace_mcount_set(unsigned long *data)
+{
+	unsigned long ip = (long)(&mcount_call);
+	unsigned long *addr = data;
+	unsigned char old[MCOUNT_INSN_SIZE], *new;
+
+	/*
+	 * Replace the mcount stub with a pointer to the
+	 * ip recorder function.
+	 */
+	memcpy(old, &mcount_call, MCOUNT_INSN_SIZE);
+	new = ftrace_call_replace(ip, *addr);
+	*addr = ftrace_modify_code(ip, old, new);
+
+	return 0;
+}
+
+
+int __init ftrace_dyn_arch_init(void *data)
+{
+	ftrace_mcount_set(data);
+	return 0;
+}
diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S
index c9afef093d5..353226fa023 100644
--- a/arch/sparc64/kernel/head.S
+++ b/arch/sparc64/kernel/head.S
@@ -10,6 +10,7 @@
 #include <linux/errno.h>
 #include <linux/threads.h>
 #include <linux/init.h>
+#include <linux/linkage.h>
 #include <asm/thread_info.h>
 #include <asm/asi.h>
 #include <asm/pstate.h>
diff --git a/arch/sparc64/kernel/hvapi.c b/arch/sparc64/kernel/hvapi.c
index f34f5d6181e..1d272c3b574 100644
--- a/arch/sparc64/kernel/hvapi.c
+++ b/arch/sparc64/kernel/hvapi.c
@@ -9,7 +9,6 @@
 
 #include <asm/hypervisor.h>
 #include <asm/oplib.h>
-#include <asm/sstate.h>
 
 /* If the hypervisor indicates that the API setting
  * calls are unsupported, by returning HV_EBADTRAP or
@@ -34,8 +33,12 @@ static struct api_info api_table[] = {
 	{ .group = HV_GRP_LDOM,					},
 	{ .group = HV_GRP_SVC_CHAN,	.flags = FLAG_PRE_API	},
 	{ .group = HV_GRP_NCS,		.flags = FLAG_PRE_API	},
+	{ .group = HV_GRP_RNG,					},
 	{ .group = HV_GRP_NIAG_PERF,	.flags = FLAG_PRE_API	},
 	{ .group = HV_GRP_FIRE_PERF,				},
+	{ .group = HV_GRP_N2_CPU,				},
+	{ .group = HV_GRP_NIU,					},
+	{ .group = HV_GRP_VF_CPU,				},
 	{ .group = HV_GRP_DIAG,		.flags = FLAG_PRE_API	},
 };
 
@@ -180,8 +183,6 @@ void __init sun4v_hvapi_init(void)
 	if (sun4v_hvapi_register(group, major, &minor))
 		goto bad;
 
-	sun4v_sstate_init();
-
 	return;
 
 bad:
diff --git a/arch/sparc64/kernel/hvcalls.S b/arch/sparc64/kernel/hvcalls.S
index a2810f3ac70..e066269d159 100644
--- a/arch/sparc64/kernel/hvcalls.S
+++ b/arch/sparc64/kernel/hvcalls.S
@@ -3,89 +3,75 @@
 	 *
 	 * returns %o0: sysino
 	 */
-	.globl	sun4v_devino_to_sysino
-	.type	sun4v_devino_to_sysino,#function
-sun4v_devino_to_sysino:
+ENTRY(sun4v_devino_to_sysino)
 	mov	HV_FAST_INTR_DEVINO2SYSINO, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o1, %o0
-	.size	sun4v_devino_to_sysino, .-sun4v_devino_to_sysino
+ENDPROC(sun4v_devino_to_sysino)
 
 	/* %o0: sysino
 	 *
 	 * returns %o0: intr_enabled (HV_INTR_{DISABLED,ENABLED})
 	 */
-	.globl	sun4v_intr_getenabled
-	.type	sun4v_intr_getenabled,#function
-sun4v_intr_getenabled:
+ENTRY(sun4v_intr_getenabled)
 	mov	HV_FAST_INTR_GETENABLED, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o1, %o0
-	.size	sun4v_intr_getenabled, .-sun4v_intr_getenabled
+ENDPROC(sun4v_intr_getenabled)
 
 	/* %o0: sysino
 	 * %o1: intr_enabled (HV_INTR_{DISABLED,ENABLED})
 	 */
-	.globl	sun4v_intr_setenabled
-	.type	sun4v_intr_setenabled,#function
-sun4v_intr_setenabled:
+ENTRY(sun4v_intr_setenabled)
 	mov	HV_FAST_INTR_SETENABLED, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_intr_setenabled, .-sun4v_intr_setenabled
+ENDPROC(sun4v_intr_setenabled)
 
 	/* %o0: sysino
 	 *
 	 * returns %o0: intr_state (HV_INTR_STATE_*)
 	 */
-	.globl	sun4v_intr_getstate
-	.type	sun4v_intr_getstate,#function
-sun4v_intr_getstate:
+ENTRY(sun4v_intr_getstate)
 	mov	HV_FAST_INTR_GETSTATE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o1, %o0
-	.size	sun4v_intr_getstate, .-sun4v_intr_getstate
+ENDPROC(sun4v_intr_getstate)
 
 	/* %o0: sysino
 	 * %o1: intr_state (HV_INTR_STATE_*)
 	 */
-	.globl	sun4v_intr_setstate
-	.type	sun4v_intr_setstate,#function
-sun4v_intr_setstate:
+ENTRY(sun4v_intr_setstate)
 	mov	HV_FAST_INTR_SETSTATE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_intr_setstate, .-sun4v_intr_setstate
+ENDPROC(sun4v_intr_setstate)
 
 	/* %o0: sysino
 	 *
 	 * returns %o0: cpuid
 	 */
-	.globl	sun4v_intr_gettarget
-	.type	sun4v_intr_gettarget,#function
-sun4v_intr_gettarget:
+ENTRY(sun4v_intr_gettarget)
 	mov	HV_FAST_INTR_GETTARGET, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o1, %o0
-	.size	sun4v_intr_gettarget, .-sun4v_intr_gettarget
+ENDPROC(sun4v_intr_gettarget)
 
 	/* %o0: sysino
 	 * %o1: cpuid
 	 */
-	.globl	sun4v_intr_settarget
-	.type	sun4v_intr_settarget,#function
-sun4v_intr_settarget:
+ENTRY(sun4v_intr_settarget)
 	mov	HV_FAST_INTR_SETTARGET, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_intr_settarget, .-sun4v_intr_settarget
+ENDPROC(sun4v_intr_settarget)
 
 	/* %o0:	cpuid
 	 * %o1: pc
@@ -94,37 +80,31 @@ sun4v_intr_settarget:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_cpu_start
-	.type	sun4v_cpu_start,#function
-sun4v_cpu_start:
+ENTRY(sun4v_cpu_start)
 	mov	HV_FAST_CPU_START, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_cpu_start, .-sun4v_cpu_start
+ENDPROC(sun4v_cpu_start)
 
 	/* %o0:	cpuid
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_cpu_stop
-	.type	sun4v_cpu_stop,#function
-sun4v_cpu_stop:
+ENTRY(sun4v_cpu_stop)
 	mov	HV_FAST_CPU_STOP, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_cpu_stop, .-sun4v_cpu_stop
+ENDPROC(sun4v_cpu_stop)
 
 	/* returns %o0:	status  */
-	.globl	sun4v_cpu_yield
-	.type	sun4v_cpu_yield, #function
-sun4v_cpu_yield:
+ENTRY(sun4v_cpu_yield)
 	mov	HV_FAST_CPU_YIELD, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_cpu_yield, .-sun4v_cpu_yield
+ENDPROC(sun4v_cpu_yield)
 
 	/* %o0:	type
 	 * %o1:	queue paddr
@@ -132,14 +112,12 @@ sun4v_cpu_yield:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_cpu_qconf
-	.type	sun4v_cpu_qconf,#function
-sun4v_cpu_qconf:
+ENTRY(sun4v_cpu_qconf)
 	mov	HV_FAST_CPU_QCONF, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_cpu_qconf, .-sun4v_cpu_qconf
+ENDPROC(sun4v_cpu_qconf)
 
 	/* %o0:	num cpus in cpu list
 	 * %o1:	cpu list paddr
@@ -147,23 +125,19 @@ sun4v_cpu_qconf:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_cpu_mondo_send
-	.type	sun4v_cpu_mondo_send,#function
-sun4v_cpu_mondo_send:
+ENTRY(sun4v_cpu_mondo_send)
 	mov	HV_FAST_CPU_MONDO_SEND, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_cpu_mondo_send, .-sun4v_cpu_mondo_send
+ENDPROC(sun4v_cpu_mondo_send)
 
 	/* %o0:	CPU ID
 	 *
 	 * returns %o0:	-status if status non-zero, else
 	 *         %o0:	cpu state as HV_CPU_STATE_*
 	 */
-	.globl	sun4v_cpu_state
-	.type	sun4v_cpu_state,#function
-sun4v_cpu_state:
+ENTRY(sun4v_cpu_state)
 	mov	HV_FAST_CPU_STATE, %o5
 	ta	HV_FAST_TRAP
 	brnz,pn	%o0, 1f
@@ -171,7 +145,7 @@ sun4v_cpu_state:
 	mov	%o1, %o0
 1:	retl
 	 nop
-	.size	sun4v_cpu_state, .-sun4v_cpu_state
+ENDPROC(sun4v_cpu_state)
 
 	/* %o0: virtual address
 	 * %o1: must be zero
@@ -180,28 +154,24 @@ sun4v_cpu_state:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_mmu_map_perm_addr
-	.type	sun4v_mmu_map_perm_addr,#function
-sun4v_mmu_map_perm_addr:
+ENTRY(sun4v_mmu_map_perm_addr)
 	mov	HV_FAST_MMU_MAP_PERM_ADDR, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_mmu_map_perm_addr, .-sun4v_mmu_map_perm_addr
+ENDPROC(sun4v_mmu_map_perm_addr)
 
 	/* %o0: number of TSB descriptions
 	 * %o1: TSB descriptions real address
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_mmu_tsb_ctx0
-	.type	sun4v_mmu_tsb_ctx0,#function
-sun4v_mmu_tsb_ctx0:
+ENTRY(sun4v_mmu_tsb_ctx0)
 	mov	HV_FAST_MMU_TSB_CTX0, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_mmu_tsb_ctx0, .-sun4v_mmu_tsb_ctx0
+ENDPROC(sun4v_mmu_tsb_ctx0)
 
 	/* %o0:	API group number
 	 * %o1: pointer to unsigned long major number storage
@@ -209,9 +179,7 @@ sun4v_mmu_tsb_ctx0:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_get_version
-	.type	sun4v_get_version,#function
-sun4v_get_version:
+ENTRY(sun4v_get_version)
 	mov	HV_CORE_GET_VER, %o5
 	mov	%o1, %o3
 	mov	%o2, %o4
@@ -219,7 +187,7 @@ sun4v_get_version:
 	stx	%o1, [%o3]
 	retl
 	 stx	%o2, [%o4]
-	.size	sun4v_get_version, .-sun4v_get_version
+ENDPROC(sun4v_get_version)
 
 	/* %o0: API group number
 	 * %o1: desired major number
@@ -228,51 +196,43 @@ sun4v_get_version:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_set_version
-	.type	sun4v_set_version,#function
-sun4v_set_version:
+ENTRY(sun4v_set_version)
 	mov	HV_CORE_SET_VER, %o5
 	mov	%o3, %o4
 	ta	HV_CORE_TRAP
 	retl
 	 stx	%o1, [%o4]
-	.size	sun4v_set_version, .-sun4v_set_version
+ENDPROC(sun4v_set_version)
 
 	/* %o0: pointer to unsigned long time
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_tod_get
-	.type	sun4v_tod_get,#function
-sun4v_tod_get:
+ENTRY(sun4v_tod_get)
 	mov	%o0, %o4
 	mov	HV_FAST_TOD_GET, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_tod_get, .-sun4v_tod_get
+ENDPROC(sun4v_tod_get)
 
 	/* %o0: time
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_tod_set
-	.type	sun4v_tod_set,#function
-sun4v_tod_set:
+ENTRY(sun4v_tod_set)
 	mov	HV_FAST_TOD_SET, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_tod_set, .-sun4v_tod_set
+ENDPROC(sun4v_tod_set)
 
 	/* %o0: pointer to unsigned long status
 	 *
 	 * returns %o0: signed character
 	 */
-	.globl	sun4v_con_getchar
-	.type	sun4v_con_getchar,#function
-sun4v_con_getchar:
+ENTRY(sun4v_con_getchar)
 	mov	%o0, %o4
 	mov	HV_FAST_CONS_GETCHAR, %o5
 	clr	%o0
@@ -281,20 +241,18 @@ sun4v_con_getchar:
 	stx	%o0, [%o4]
 	retl
 	 sra	%o1, 0, %o0
-	.size	sun4v_con_getchar, .-sun4v_con_getchar
+ENDPROC(sun4v_con_getchar)
 
 	/* %o0: signed long character
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_con_putchar
-	.type	sun4v_con_putchar,#function
-sun4v_con_putchar:
+ENTRY(sun4v_con_putchar)
 	mov	HV_FAST_CONS_PUTCHAR, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 sra	%o0, 0, %o0
-	.size	sun4v_con_putchar, .-sun4v_con_putchar
+ENDPROC(sun4v_con_putchar)
 
 	/* %o0: buffer real address
 	 * %o1: buffer size
@@ -302,9 +260,7 @@ sun4v_con_putchar:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_con_read
-	.type	sun4v_con_read,#function
-sun4v_con_read:
+ENTRY(sun4v_con_read)
 	mov	%o2, %o4
 	mov	HV_FAST_CONS_READ, %o5
 	ta	HV_FAST_TRAP
@@ -318,7 +274,7 @@ sun4v_con_read:
 	stx	%o1, [%o4]
 1:	retl
 	 nop
-	.size	sun4v_con_read, .-sun4v_con_read
+ENDPROC(sun4v_con_read)
 
 	/* %o0: buffer real address
 	 * %o1: buffer size
@@ -326,43 +282,37 @@ sun4v_con_read:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_con_write
-	.type	sun4v_con_write,#function
-sun4v_con_write:
+ENTRY(sun4v_con_write)
 	mov	%o2, %o4
 	mov	HV_FAST_CONS_WRITE, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_con_write, .-sun4v_con_write
+ENDPROC(sun4v_con_write)
 
 	/* %o0:	soft state
 	 * %o1:	address of description string
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_mach_set_soft_state
-	.type	sun4v_mach_set_soft_state,#function
-sun4v_mach_set_soft_state:
+ENTRY(sun4v_mach_set_soft_state)
 	mov	HV_FAST_MACH_SET_SOFT_STATE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_mach_set_soft_state, .-sun4v_mach_set_soft_state
+ENDPROC(sun4v_mach_set_soft_state)
 
 	/* %o0: exit code
 	 *
 	 * Does not return.
 	 */
-	.globl	sun4v_mach_exit
-	.type	sun4v_mach_exit,#function
-sun4v_mach_exit:
+ENTRY(sun4v_mach_exit)
 	mov	HV_FAST_MACH_EXIT, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_mach_exit, .-sun4v_mach_exit
+ENDPROC(sun4v_mach_exit)
 
 	/* %o0: buffer real address
 	 * %o1: buffer length
@@ -370,44 +320,38 @@ sun4v_mach_exit:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_mach_desc
-	.type	sun4v_mach_desc,#function
-sun4v_mach_desc:
+ENTRY(sun4v_mach_desc)
 	mov	%o2, %o4
 	mov	HV_FAST_MACH_DESC, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_mach_desc, .-sun4v_mach_desc
+ENDPROC(sun4v_mach_desc)
 
 	/* %o0: new timeout in milliseconds
 	 * %o1: pointer to unsigned long orig_timeout
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_mach_set_watchdog
-	.type	sun4v_mach_set_watchdog,#function
-sun4v_mach_set_watchdog:
+ENTRY(sun4v_mach_set_watchdog)
 	mov	%o1, %o4
 	mov	HV_FAST_MACH_SET_WATCHDOG, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_mach_set_watchdog, .-sun4v_mach_set_watchdog
+ENDPROC(sun4v_mach_set_watchdog)
 
 	/* No inputs and does not return.  */
-	.globl	sun4v_mach_sir
-	.type	sun4v_mach_sir,#function
-sun4v_mach_sir:
+ENTRY(sun4v_mach_sir)
 	mov	%o1, %o4
 	mov	HV_FAST_MACH_SIR, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_mach_sir, .-sun4v_mach_sir
+ENDPROC(sun4v_mach_sir)
 
 	/* %o0: channel
 	 * %o1:	ra
@@ -415,14 +359,12 @@ sun4v_mach_sir:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_tx_qconf
-	.type	sun4v_ldc_tx_qconf,#function
-sun4v_ldc_tx_qconf:
+ENTRY(sun4v_ldc_tx_qconf)
 	mov	HV_FAST_LDC_TX_QCONF, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ldc_tx_qconf, .-sun4v_ldc_tx_qconf
+ENDPROC(sun4v_ldc_tx_qconf)
 
 	/* %o0: channel
 	 * %o1:	pointer to unsigned long ra
@@ -430,9 +372,7 @@ sun4v_ldc_tx_qconf:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_tx_qinfo
-	.type	sun4v_ldc_tx_qinfo,#function
-sun4v_ldc_tx_qinfo:
+ENTRY(sun4v_ldc_tx_qinfo)
 	mov	%o1, %g1
 	mov	%o2, %g2
 	mov	HV_FAST_LDC_TX_QINFO, %o5
@@ -441,7 +381,7 @@ sun4v_ldc_tx_qinfo:
 	stx	%o2, [%g2]
 	retl
 	 nop
-	.size	sun4v_ldc_tx_qinfo, .-sun4v_ldc_tx_qinfo
+ENDPROC(sun4v_ldc_tx_qinfo)
 
 	/* %o0: channel
 	 * %o1:	pointer to unsigned long head_off
@@ -450,9 +390,7 @@ sun4v_ldc_tx_qinfo:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_tx_get_state
-	.type	sun4v_ldc_tx_get_state,#function
-sun4v_ldc_tx_get_state:
+ENTRY(sun4v_ldc_tx_get_state)
 	mov	%o1, %g1
 	mov	%o2, %g2
 	mov	%o3, %g3
@@ -463,21 +401,19 @@ sun4v_ldc_tx_get_state:
 	stx	%o3, [%g3]
 	retl
 	 nop
-	.size	sun4v_ldc_tx_get_state, .-sun4v_ldc_tx_get_state
+ENDPROC(sun4v_ldc_tx_get_state)
 
 	/* %o0: channel
 	 * %o1:	tail_off
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_tx_set_qtail
-	.type	sun4v_ldc_tx_set_qtail,#function
-sun4v_ldc_tx_set_qtail:
+ENTRY(sun4v_ldc_tx_set_qtail)
 	mov	HV_FAST_LDC_TX_SET_QTAIL, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ldc_tx_set_qtail, .-sun4v_ldc_tx_set_qtail
+ENDPROC(sun4v_ldc_tx_set_qtail)
 
 	/* %o0: channel
 	 * %o1:	ra
@@ -485,14 +421,12 @@ sun4v_ldc_tx_set_qtail:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_rx_qconf
-	.type	sun4v_ldc_rx_qconf,#function
-sun4v_ldc_rx_qconf:
+ENTRY(sun4v_ldc_rx_qconf)
 	mov	HV_FAST_LDC_RX_QCONF, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ldc_rx_qconf, .-sun4v_ldc_rx_qconf
+ENDPROC(sun4v_ldc_rx_qconf)
 
 	/* %o0: channel
 	 * %o1:	pointer to unsigned long ra
@@ -500,9 +434,7 @@ sun4v_ldc_rx_qconf:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_rx_qinfo
-	.type	sun4v_ldc_rx_qinfo,#function
-sun4v_ldc_rx_qinfo:
+ENTRY(sun4v_ldc_rx_qinfo)
 	mov	%o1, %g1
 	mov	%o2, %g2
 	mov	HV_FAST_LDC_RX_QINFO, %o5
@@ -511,7 +443,7 @@ sun4v_ldc_rx_qinfo:
 	stx	%o2, [%g2]
 	retl
 	 nop
-	.size	sun4v_ldc_rx_qinfo, .-sun4v_ldc_rx_qinfo
+ENDPROC(sun4v_ldc_rx_qinfo)
 
 	/* %o0: channel
 	 * %o1:	pointer to unsigned long head_off
@@ -520,9 +452,7 @@ sun4v_ldc_rx_qinfo:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_rx_get_state
-	.type	sun4v_ldc_rx_get_state,#function
-sun4v_ldc_rx_get_state:
+ENTRY(sun4v_ldc_rx_get_state)
 	mov	%o1, %g1
 	mov	%o2, %g2
 	mov	%o3, %g3
@@ -533,21 +463,19 @@ sun4v_ldc_rx_get_state:
 	stx	%o3, [%g3]
 	retl
 	 nop
-	.size	sun4v_ldc_rx_get_state, .-sun4v_ldc_rx_get_state
+ENDPROC(sun4v_ldc_rx_get_state)
 
 	/* %o0: channel
 	 * %o1:	head_off
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_rx_set_qhead
-	.type	sun4v_ldc_rx_set_qhead,#function
-sun4v_ldc_rx_set_qhead:
+ENTRY(sun4v_ldc_rx_set_qhead)
 	mov	HV_FAST_LDC_RX_SET_QHEAD, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ldc_rx_set_qhead, .-sun4v_ldc_rx_set_qhead
+ENDPROC(sun4v_ldc_rx_set_qhead)
 
 	/* %o0: channel
 	 * %o1:	ra
@@ -555,14 +483,12 @@ sun4v_ldc_rx_set_qhead:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_ldc_set_map_table
-	.type	sun4v_ldc_set_map_table,#function
-sun4v_ldc_set_map_table:
+ENTRY(sun4v_ldc_set_map_table)
 	mov	HV_FAST_LDC_SET_MAP_TABLE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ldc_set_map_table, .-sun4v_ldc_set_map_table
+ENDPROC(sun4v_ldc_set_map_table)
 
 	/* %o0: channel
 	 * %o1:	pointer to unsigned long ra
@@ -570,9 +496,7 @@ sun4v_ldc_set_map_table:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_ldc_get_map_table
-	.type	sun4v_ldc_get_map_table,#function
-sun4v_ldc_get_map_table:
+ENTRY(sun4v_ldc_get_map_table)
 	mov	%o1, %g1
 	mov	%o2, %g2
 	mov	HV_FAST_LDC_GET_MAP_TABLE, %o5
@@ -581,7 +505,7 @@ sun4v_ldc_get_map_table:
 	stx	%o2, [%g2]
 	retl
 	 nop
-	.size	sun4v_ldc_get_map_table, .-sun4v_ldc_get_map_table
+ENDPROC(sun4v_ldc_get_map_table)
 
 	/* %o0:	channel
 	 * %o1:	dir_code
@@ -592,16 +516,14 @@ sun4v_ldc_get_map_table:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_copy
-	.type	sun4v_ldc_copy,#function
-sun4v_ldc_copy:
+ENTRY(sun4v_ldc_copy)
 	mov	%o5, %g1
 	mov	HV_FAST_LDC_COPY, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%g1]
 	retl
 	 nop
-	.size	sun4v_ldc_copy, .-sun4v_ldc_copy
+ENDPROC(sun4v_ldc_copy)
 
 	/* %o0:	channel
 	 * %o1:	cookie
@@ -610,9 +532,7 @@ sun4v_ldc_copy:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_mapin
-	.type	sun4v_ldc_mapin,#function
-sun4v_ldc_mapin:
+ENTRY(sun4v_ldc_mapin)
 	mov	%o2, %g1
 	mov	%o3, %g2
 	mov	HV_FAST_LDC_MAPIN, %o5
@@ -621,20 +541,18 @@ sun4v_ldc_mapin:
 	stx	%o2, [%g2]
 	retl
 	 nop
-	.size	sun4v_ldc_mapin, .-sun4v_ldc_mapin
+ENDPROC(sun4v_ldc_mapin)
 
 	/* %o0:	ra
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_unmap
-	.type	sun4v_ldc_unmap,#function
-sun4v_ldc_unmap:
+ENTRY(sun4v_ldc_unmap)
 	mov	HV_FAST_LDC_UNMAP, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ldc_unmap, .-sun4v_ldc_unmap
+ENDPROC(sun4v_ldc_unmap)
 
 	/* %o0: channel
 	 * %o1:	cookie
@@ -642,14 +560,12 @@ sun4v_ldc_unmap:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ldc_revoke
-	.type	sun4v_ldc_revoke,#function
-sun4v_ldc_revoke:
+ENTRY(sun4v_ldc_revoke)
 	mov	HV_FAST_LDC_REVOKE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ldc_revoke, .-sun4v_ldc_revoke
+ENDPROC(sun4v_ldc_revoke)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -657,16 +573,14 @@ sun4v_ldc_revoke:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_get_cookie
-	.type	sun4v_vintr_get_cookie,#function
-sun4v_vintr_get_cookie:
+ENTRY(sun4v_vintr_get_cookie)
 	mov	%o2, %g1
 	mov	HV_FAST_VINTR_GET_COOKIE, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%g1]
 	retl
 	 nop
-	.size	sun4v_vintr_get_cookie, .-sun4v_vintr_get_cookie
+ENDPROC(sun4v_vintr_get_cookie)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -674,14 +588,12 @@ sun4v_vintr_get_cookie:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_set_cookie
-	.type	sun4v_vintr_set_cookie,#function
-sun4v_vintr_set_cookie:
+ENTRY(sun4v_vintr_set_cookie)
 	mov	HV_FAST_VINTR_SET_COOKIE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_vintr_set_cookie, .-sun4v_vintr_set_cookie
+ENDPROC(sun4v_vintr_set_cookie)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -689,16 +601,14 @@ sun4v_vintr_set_cookie:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_get_valid
-	.type	sun4v_vintr_get_valid,#function
-sun4v_vintr_get_valid:
+ENTRY(sun4v_vintr_get_valid)
 	mov	%o2, %g1
 	mov	HV_FAST_VINTR_GET_VALID, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%g1]
 	retl
 	 nop
-	.size	sun4v_vintr_get_valid, .-sun4v_vintr_get_valid
+ENDPROC(sun4v_vintr_get_valid)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -706,14 +616,12 @@ sun4v_vintr_get_valid:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_set_valid
-	.type	sun4v_vintr_set_valid,#function
-sun4v_vintr_set_valid:
+ENTRY(sun4v_vintr_set_valid)
 	mov	HV_FAST_VINTR_SET_VALID, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_vintr_set_valid, .-sun4v_vintr_set_valid
+ENDPROC(sun4v_vintr_set_valid)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -721,16 +629,14 @@ sun4v_vintr_set_valid:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_get_state
-	.type	sun4v_vintr_get_state,#function
-sun4v_vintr_get_state:
+ENTRY(sun4v_vintr_get_state)
 	mov	%o2, %g1
 	mov	HV_FAST_VINTR_GET_STATE, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%g1]
 	retl
 	 nop
-	.size	sun4v_vintr_get_state, .-sun4v_vintr_get_state
+ENDPROC(sun4v_vintr_get_state)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -738,14 +644,12 @@ sun4v_vintr_get_state:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_set_state
-	.type	sun4v_vintr_set_state,#function
-sun4v_vintr_set_state:
+ENTRY(sun4v_vintr_set_state)
 	mov	HV_FAST_VINTR_SET_STATE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_vintr_set_state, .-sun4v_vintr_set_state
+ENDPROC(sun4v_vintr_set_state)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -753,16 +657,14 @@ sun4v_vintr_set_state:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_get_target
-	.type	sun4v_vintr_get_target,#function
-sun4v_vintr_get_target:
+ENTRY(sun4v_vintr_get_target)
 	mov	%o2, %g1
 	mov	HV_FAST_VINTR_GET_TARGET, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%g1]
 	retl
 	 nop
-	.size	sun4v_vintr_get_target, .-sun4v_vintr_get_target
+ENDPROC(sun4v_vintr_get_target)
 
 	/* %o0: device handle
 	 * %o1:	device INO
@@ -770,14 +672,12 @@ sun4v_vintr_get_target:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	sun4v_vintr_set_target
-	.type	sun4v_vintr_set_target,#function
-sun4v_vintr_set_target:
+ENTRY(sun4v_vintr_set_target)
 	mov	HV_FAST_VINTR_SET_TARGET, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_vintr_set_target, .-sun4v_vintr_set_target
+ENDPROC(sun4v_vintr_set_target)
 
 	/* %o0: NCS sub-function
 	 * %o1:	sub-function arg real-address
@@ -785,18 +685,14 @@ sun4v_vintr_set_target:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	sun4v_ncs_request
-	.type	sun4v_ncs_request,#function
-sun4v_ncs_request:
+ENTRY(sun4v_ncs_request)
 	mov	HV_FAST_NCS_REQUEST, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_ncs_request, .-sun4v_ncs_request
+ENDPROC(sun4v_ncs_request)
 
-	.globl	sun4v_svc_send
-	.type	sun4v_svc_send,#function
-sun4v_svc_send:
+ENTRY(sun4v_svc_send)
 	save	%sp, -192, %sp
 	mov	%i0, %o0
 	mov	%i1, %o1
@@ -806,11 +702,9 @@ sun4v_svc_send:
 	stx	%o1, [%i3]
 	ret
 	restore
-	.size	sun4v_svc_send, .-sun4v_svc_send
+ENDPROC(sun4v_svc_send)
 
-	.globl	sun4v_svc_recv
-	.type	sun4v_svc_recv,#function
-sun4v_svc_recv:
+ENTRY(sun4v_svc_recv)
 	save	%sp, -192, %sp
 	mov	%i0, %o0
 	mov	%i1, %o1
@@ -820,62 +714,50 @@ sun4v_svc_recv:
 	stx	%o1, [%i3]
 	ret
 	restore
-	.size	sun4v_svc_recv, .-sun4v_svc_recv
+ENDPROC(sun4v_svc_recv)
 
-	.globl	sun4v_svc_getstatus
-	.type	sun4v_svc_getstatus,#function
-sun4v_svc_getstatus:
+ENTRY(sun4v_svc_getstatus)
 	mov	HV_FAST_SVC_GETSTATUS, %o5
 	mov	%o1, %o4
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_svc_getstatus, .-sun4v_svc_getstatus
+ENDPROC(sun4v_svc_getstatus)
 
-	.globl	sun4v_svc_setstatus
-	.type	sun4v_svc_setstatus,#function
-sun4v_svc_setstatus:
+ENTRY(sun4v_svc_setstatus)
 	mov	HV_FAST_SVC_SETSTATUS, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_svc_setstatus, .-sun4v_svc_setstatus
+ENDPROC(sun4v_svc_setstatus)
 
-	.globl	sun4v_svc_clrstatus
-	.type	sun4v_svc_clrstatus,#function
-sun4v_svc_clrstatus:
+ENTRY(sun4v_svc_clrstatus)
 	mov	HV_FAST_SVC_CLRSTATUS, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_svc_clrstatus, .-sun4v_svc_clrstatus
+ENDPROC(sun4v_svc_clrstatus)
 
-	.globl	sun4v_mmustat_conf
-	.type	sun4v_mmustat_conf,#function
-sun4v_mmustat_conf:
+ENTRY(sun4v_mmustat_conf)
 	mov	%o1, %o4
 	mov	HV_FAST_MMUSTAT_CONF, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_mmustat_conf, .-sun4v_mmustat_conf
+ENDPROC(sun4v_mmustat_conf)
 
-	.globl	sun4v_mmustat_info
-	.type	sun4v_mmustat_info,#function
-sun4v_mmustat_info:
+ENTRY(sun4v_mmustat_info)
 	mov	%o0, %o4
 	mov	HV_FAST_MMUSTAT_INFO, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o4]
 	retl
 	 nop
-	.size	sun4v_mmustat_info, .-sun4v_mmustat_info
+ENDPROC(sun4v_mmustat_info)
 
-	.globl	sun4v_mmu_demap_all
-	.type	sun4v_mmu_demap_all,#function
-sun4v_mmu_demap_all:
+ENTRY(sun4v_mmu_demap_all)
 	clr	%o0
 	clr	%o1
 	mov	HV_MMU_ALL, %o2
@@ -883,4 +765,4 @@ sun4v_mmu_demap_all:
 	ta	HV_FAST_TRAP
 	retl
 	 nop
-	.size	sun4v_mmu_demap_all, .-sun4v_mmu_demap_all
+ENDPROC(sun4v_mmu_demap_all)
diff --git a/arch/sparc64/kernel/iommu_common.h b/arch/sparc64/kernel/iommu_common.h
index f3575a614fa..53b19c8231a 100644
--- a/arch/sparc64/kernel/iommu_common.h
+++ b/arch/sparc64/kernel/iommu_common.h
@@ -23,7 +23,7 @@
 #define IO_PAGE_SHIFT			13
 #define IO_PAGE_SIZE			(1UL << IO_PAGE_SHIFT)
 #define IO_PAGE_MASK			(~(IO_PAGE_SIZE-1))
-#define IO_PAGE_ALIGN(addr)		(((addr)+IO_PAGE_SIZE-1)&IO_PAGE_MASK)
+#define IO_PAGE_ALIGN(addr)		ALIGN(addr, IO_PAGE_SIZE)
 
 #define IO_TSB_ENTRIES			(128*1024)
 #define IO_TSB_SIZE			(IO_TSB_ENTRIES * 8)
diff --git a/arch/sparc64/kernel/irq.c b/arch/sparc64/kernel/irq.c
index b441a26b73b..52fc836f464 100644
--- a/arch/sparc64/kernel/irq.c
+++ b/arch/sparc64/kernel/irq.c
@@ -7,6 +7,7 @@
 
 #include <linux/module.h>
 #include <linux/sched.h>
+#include <linux/linkage.h>
 #include <linux/ptrace.h>
 #include <linux/errno.h>
 #include <linux/kernel_stat.h>
@@ -28,7 +29,6 @@
 #include <asm/system.h>
 #include <asm/irq.h>
 #include <asm/io.h>
-#include <asm/sbus.h>
 #include <asm/iommu.h>
 #include <asm/upa.h>
 #include <asm/oplib.h>
@@ -621,8 +621,9 @@ unsigned int sun4v_build_irq(u32 devhandle, unsigned int devino)
 unsigned int sun4v_build_virq(u32 devhandle, unsigned int devino)
 {
 	struct irq_handler_data *data;
-	struct ino_bucket *bucket;
 	unsigned long hv_err, cookie;
+	struct ino_bucket *bucket;
+	struct irq_desc *desc;
 	unsigned int virt_irq;
 
 	bucket = kzalloc(sizeof(struct ino_bucket), GFP_ATOMIC);
@@ -643,6 +644,13 @@ unsigned int sun4v_build_virq(u32 devhandle, unsigned int devino)
 	if (unlikely(!data))
 		return 0;
 
+	/* In order to make the LDC channel startup sequence easier,
+	 * especially wrt. locking, we do not let request_irq() enable
+	 * the interrupt.
+	 */
+	desc = irq_desc + virt_irq;
+	desc->status |= IRQ_NOAUTOEN;
+
 	set_irq_chip_data(virt_irq, data);
 
 	/* Catch accidental accesses to these things.  IMAP/ICLR handling
@@ -674,10 +682,32 @@ void ack_bad_irq(unsigned int virt_irq)
 	       ino, virt_irq);
 }
 
+void *hardirq_stack[NR_CPUS];
+void *softirq_stack[NR_CPUS];
+
+static __attribute__((always_inline)) void *set_hardirq_stack(void)
+{
+	void *orig_sp, *sp = hardirq_stack[smp_processor_id()];
+
+	__asm__ __volatile__("mov %%sp, %0" : "=r" (orig_sp));
+	if (orig_sp < sp ||
+	    orig_sp > (sp + THREAD_SIZE)) {
+		sp += THREAD_SIZE - 192 - STACK_BIAS;
+		__asm__ __volatile__("mov %0, %%sp" : : "r" (sp));
+	}
+
+	return orig_sp;
+}
+static __attribute__((always_inline)) void restore_hardirq_stack(void *orig_sp)
+{
+	__asm__ __volatile__("mov %0, %%sp" : : "r" (orig_sp));
+}
+
 void handler_irq(int irq, struct pt_regs *regs)
 {
 	unsigned long pstate, bucket_pa;
 	struct pt_regs *old_regs;
+	void *orig_sp;
 
 	clear_softint(1 << irq);
 
@@ -695,6 +725,8 @@ void handler_irq(int irq, struct pt_regs *regs)
 			       "i" (PSTATE_IE)
 			     : "memory");
 
+	orig_sp = set_hardirq_stack();
+
 	while (bucket_pa) {
 		struct irq_desc *desc;
 		unsigned long next_pa;
@@ -711,10 +743,38 @@ void handler_irq(int irq, struct pt_regs *regs)
 		bucket_pa = next_pa;
 	}
 
+	restore_hardirq_stack(orig_sp);
+
 	irq_exit();
 	set_irq_regs(old_regs);
 }
 
+void do_softirq(void)
+{
+	unsigned long flags;
+
+	if (in_interrupt())
+		return;
+
+	local_irq_save(flags);
+
+	if (local_softirq_pending()) {
+		void *orig_sp, *sp = softirq_stack[smp_processor_id()];
+
+		sp += THREAD_SIZE - 192 - STACK_BIAS;
+
+		__asm__ __volatile__("mov %%sp, %0\n\t"
+				     "mov %1, %%sp"
+				     : "=&r" (orig_sp)
+				     : "r" (sp));
+		__do_softirq();
+		__asm__ __volatile__("mov %0, %%sp"
+				     : : "r" (orig_sp));
+	}
+
+	local_irq_restore(flags);
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 void fixup_irqs(void)
 {
@@ -732,6 +792,8 @@ void fixup_irqs(void)
 		}
 		spin_unlock_irqrestore(&irq_desc[irq].lock, flags);
 	}
+
+	tick_ops->disable_irq();
 }
 #endif
 
@@ -804,7 +866,7 @@ static void kill_prom_timer(void)
 	: "g1", "g2");
 }
 
-void init_irqwork_curcpu(void)
+void notrace init_irqwork_curcpu(void)
 {
 	int cpu = hard_smp_processor_id();
 
@@ -835,7 +897,7 @@ static void __cpuinit register_one_mondo(unsigned long paddr, unsigned long type
 	}
 }
 
-void __cpuinit sun4v_register_mondo_queues(int this_cpu)
+void __cpuinit notrace sun4v_register_mondo_queues(int this_cpu)
 {
 	struct trap_per_cpu *tb = &trap_block[this_cpu];
 
@@ -907,12 +969,18 @@ static void __init sun4v_init_mondo_queues(void)
 		alloc_one_mondo(&tb->nonresum_mondo_pa, tb->nonresum_qmask);
 		alloc_one_kbuf(&tb->nonresum_kernel_buf_pa,
 			       tb->nonresum_qmask);
+	}
+}
+
+static void __init init_send_mondo_info(void)
+{
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		struct trap_per_cpu *tb = &trap_block[cpu];
 
 		init_cpu_send_mondo_info(tb);
 	}
-
-	/* Load up the boot cpu's entries.  */
-	sun4v_register_mondo_queues(hard_smp_processor_id());
 }
 
 static struct irqaction timer_irq_action = {
@@ -941,6 +1009,13 @@ void __init init_IRQ(void)
 	if (tlb_type == hypervisor)
 		sun4v_init_mondo_queues();
 
+	init_send_mondo_info();
+
+	if (tlb_type == hypervisor) {
+		/* Load up the boot cpu's entries.  */
+		sun4v_register_mondo_queues(hard_smp_processor_id());
+	}
+
 	/* We need to clear any IRQ's pending in the soft interrupt
 	 * registers, a spurious one could be left around from the
 	 * PROM timer which we just disabled.
diff --git a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c
index f43b5d75535..201a6e547e4 100644
--- a/arch/sparc64/kernel/kprobes.c
+++ b/arch/sparc64/kernel/kprobes.c
@@ -478,9 +478,9 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
 	return 0;
 }
 
-/* Called with kretprobe_lock held.  The value stored in the return
- * address register is actually 2 instructions before where the
- * callee will return to.  Sequences usually look something like this
+/* The value stored in the return address register is actually 2
+ * instructions before where the callee will return to.
+ * Sequences usually look something like this
  *
  *		call	some_function	<--- return register points here
  *		 nop			<--- call delay slot
@@ -512,8 +512,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
 	unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline;
 
 	INIT_HLIST_HEAD(&empty_rp);
-	spin_lock_irqsave(&kretprobe_lock, flags);
-	head = kretprobe_inst_table_head(current);
+	kretprobe_hash_lock(current, &head, &flags);
 
 	/*
 	 * It is possible to have multiple instances associated with a given
@@ -553,7 +552,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
 	regs->tnpc = orig_ret_address + 4;
 
 	reset_current_kprobe();
-	spin_unlock_irqrestore(&kretprobe_lock, flags);
+	kretprobe_hash_unlock(current, &flags);
 	preempt_enable_no_resched();
 
 	hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) {
diff --git a/arch/sparc64/kernel/kstack.h b/arch/sparc64/kernel/kstack.h
new file mode 100644
index 00000000000..4248d969272
--- /dev/null
+++ b/arch/sparc64/kernel/kstack.h
@@ -0,0 +1,60 @@
+#ifndef _KSTACK_H
+#define _KSTACK_H
+
+#include <linux/thread_info.h>
+#include <linux/sched.h>
+#include <asm/ptrace.h>
+#include <asm/irq.h>
+
+/* SP must be STACK_BIAS adjusted already.  */
+static inline bool kstack_valid(struct thread_info *tp, unsigned long sp)
+{
+	unsigned long base = (unsigned long) tp;
+
+	if (sp >= (base + sizeof(struct thread_info)) &&
+	    sp <= (base + THREAD_SIZE - sizeof(struct sparc_stackf)))
+		return true;
+
+	if (hardirq_stack[tp->cpu]) {
+		base = (unsigned long) hardirq_stack[tp->cpu];
+		if (sp >= base &&
+		    sp <= (base + THREAD_SIZE - sizeof(struct sparc_stackf)))
+			return true;
+		base = (unsigned long) softirq_stack[tp->cpu];
+		if (sp >= base &&
+		    sp <= (base + THREAD_SIZE - sizeof(struct sparc_stackf)))
+			return true;
+	}
+	return false;
+}
+
+/* Does "regs" point to a valid pt_regs trap frame?  */
+static inline bool kstack_is_trap_frame(struct thread_info *tp, struct pt_regs *regs)
+{
+	unsigned long base = (unsigned long) tp;
+	unsigned long addr = (unsigned long) regs;
+
+	if (addr >= base &&
+	    addr <= (base + THREAD_SIZE - sizeof(*regs)))
+		goto check_magic;
+
+	if (hardirq_stack[tp->cpu]) {
+		base = (unsigned long) hardirq_stack[tp->cpu];
+		if (addr >= base &&
+		    addr <= (base + THREAD_SIZE - sizeof(*regs)))
+			goto check_magic;
+		base = (unsigned long) softirq_stack[tp->cpu];
+		if (addr >= base &&
+		    addr <= (base + THREAD_SIZE - sizeof(*regs)))
+			goto check_magic;
+	}
+	return false;
+
+check_magic:
+	if ((regs->magic & ~0x1ff) == PT_REGS_MAGIC)
+		return true;
+	return false;
+
+}
+
+#endif /* _KSTACK_H */
diff --git a/arch/sparc64/kernel/ldc.c b/arch/sparc64/kernel/ldc.c
index 63969f61028..d68982330f6 100644
--- a/arch/sparc64/kernel/ldc.c
+++ b/arch/sparc64/kernel/ldc.c
@@ -1,6 +1,6 @@
 /* ldc.c: Logical Domain Channel link-layer protocol driver.
  *
- * Copyright (C) 2007 David S. Miller <davem@davemloft.net>
+ * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
  */
 
 #include <linux/kernel.h>
@@ -23,8 +23,8 @@
 
 #define DRV_MODULE_NAME		"ldc"
 #define PFX DRV_MODULE_NAME	": "
-#define DRV_MODULE_VERSION	"1.0"
-#define DRV_MODULE_RELDATE	"June 25, 2007"
+#define DRV_MODULE_VERSION	"1.1"
+#define DRV_MODULE_RELDATE	"July 22, 2008"
 
 static char version[] __devinitdata =
 	DRV_MODULE_NAME ".c:v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
@@ -1235,13 +1235,9 @@ int ldc_bind(struct ldc_channel *lp, const char *name)
 	unsigned long hv_err, flags;
 	int err = -EINVAL;
 
-	spin_lock_irqsave(&lp->lock, flags);
-
-	if (!name)
-		goto out_err;
-
-	if (lp->state != LDC_STATE_INIT)
-		goto out_err;
+	if (!name ||
+	    (lp->state != LDC_STATE_INIT))
+		return -EINVAL;
 
 	snprintf(lp->rx_irq_name, LDC_IRQ_NAME_MAX, "%s RX", name);
 	snprintf(lp->tx_irq_name, LDC_IRQ_NAME_MAX, "%s TX", name);
@@ -1250,25 +1246,32 @@ int ldc_bind(struct ldc_channel *lp, const char *name)
 			  IRQF_SAMPLE_RANDOM | IRQF_SHARED,
 			  lp->rx_irq_name, lp);
 	if (err)
-		goto out_err;
+		return err;
 
 	err = request_irq(lp->cfg.tx_irq, ldc_tx,
 			  IRQF_SAMPLE_RANDOM | IRQF_SHARED,
 			  lp->tx_irq_name, lp);
-	if (err)
-		goto out_free_rx_irq;
+	if (err) {
+		free_irq(lp->cfg.rx_irq, lp);
+		return err;
+	}
+
 
+	spin_lock_irqsave(&lp->lock, flags);
+
+	enable_irq(lp->cfg.rx_irq);
+	enable_irq(lp->cfg.tx_irq);
 
 	lp->flags |= LDC_FLAG_REGISTERED_IRQS;
 
 	err = -ENODEV;
 	hv_err = sun4v_ldc_tx_qconf(lp->id, 0, 0);
 	if (hv_err)
-		goto out_free_tx_irq;
+		goto out_free_irqs;
 
 	hv_err = sun4v_ldc_tx_qconf(lp->id, lp->tx_ra, lp->tx_num_entries);
 	if (hv_err)
-		goto out_free_tx_irq;
+		goto out_free_irqs;
 
 	hv_err = sun4v_ldc_rx_qconf(lp->id, 0, 0);
 	if (hv_err)
@@ -1304,14 +1307,11 @@ out_unmap_rx:
 out_unmap_tx:
 	sun4v_ldc_tx_qconf(lp->id, 0, 0);
 
-out_free_tx_irq:
+out_free_irqs:
 	lp->flags &= ~LDC_FLAG_REGISTERED_IRQS;
 	free_irq(lp->cfg.tx_irq, lp);
-
-out_free_rx_irq:
 	free_irq(lp->cfg.rx_irq, lp);
 
-out_err:
 	spin_unlock_irqrestore(&lp->lock, flags);
 
 	return err;
diff --git a/arch/sparc64/kernel/of_device.c b/arch/sparc64/kernel/of_device.c
index d569f60c24b..0f616ae3246 100644
--- a/arch/sparc64/kernel/of_device.c
+++ b/arch/sparc64/kernel/of_device.c
@@ -55,18 +55,38 @@ struct of_device *of_find_device_by_node(struct device_node *dp)
 }
 EXPORT_SYMBOL(of_find_device_by_node);
 
-#ifdef CONFIG_PCI
-struct bus_type isa_bus_type;
-EXPORT_SYMBOL(isa_bus_type);
+unsigned int irq_of_parse_and_map(struct device_node *node, int index)
+{
+	struct of_device *op = of_find_device_by_node(node);
+
+	if (!op || index >= op->num_irqs)
+		return 0;
+
+	return op->irqs[index];
+}
+EXPORT_SYMBOL(irq_of_parse_and_map);
 
-struct bus_type ebus_bus_type;
-EXPORT_SYMBOL(ebus_bus_type);
-#endif
+/* Take the archdata values for IOMMU, STC, and HOSTDATA found in
+ * BUS and propagate to all child of_device objects.
+ */
+void of_propagate_archdata(struct of_device *bus)
+{
+	struct dev_archdata *bus_sd = &bus->dev.archdata;
+	struct device_node *bus_dp = bus->node;
+	struct device_node *dp;
+
+	for (dp = bus_dp->child; dp; dp = dp->sibling) {
+		struct of_device *op = of_find_device_by_node(dp);
 
-#ifdef CONFIG_SBUS
-struct bus_type sbus_bus_type;
-EXPORT_SYMBOL(sbus_bus_type);
-#endif
+		op->dev.archdata.iommu = bus_sd->iommu;
+		op->dev.archdata.stc = bus_sd->stc;
+		op->dev.archdata.host_controller = bus_sd->host_controller;
+		op->dev.archdata.numa_node = bus_sd->numa_node;
+
+		if (dp->child)
+			of_propagate_archdata(op);
+	}
+}
 
 struct bus_type of_platform_bus_type;
 EXPORT_SYMBOL(of_platform_bus_type);
@@ -99,7 +119,7 @@ struct of_bus {
 				       int *addrc, int *sizec);
 	int		(*map)(u32 *addr, const u32 *range,
 			       int na, int ns, int pna);
-	unsigned int	(*get_flags)(const u32 *addr);
+	unsigned long	(*get_flags)(const u32 *addr, unsigned long);
 };
 
 /*
@@ -159,8 +179,10 @@ static int of_bus_default_map(u32 *addr, const u32 *range,
 	return 0;
 }
 
-static unsigned int of_bus_default_get_flags(const u32 *addr)
+static unsigned long of_bus_default_get_flags(const u32 *addr, unsigned long flags)
 {
+	if (flags)
+		return flags;
 	return IORESOURCE_MEM;
 }
 
@@ -170,7 +192,7 @@ static unsigned int of_bus_default_get_flags(const u32 *addr)
 
 static int of_bus_pci_match(struct device_node *np)
 {
-	if (!strcmp(np->type, "pci") || !strcmp(np->type, "pciex")) {
+	if (!strcmp(np->name, "pci")) {
 		const char *model = of_get_property(np, "model", NULL);
 
 		if (model && !strcmp(model, "SUNW,simba"))
@@ -201,7 +223,7 @@ static int of_bus_simba_match(struct device_node *np)
 	/* Treat PCI busses lacking ranges property just like
 	 * simba.
 	 */
-	if (!strcmp(np->type, "pci") || !strcmp(np->type, "pciex")) {
+	if (!strcmp(np->name, "pci")) {
 		if (!of_find_property(np, "ranges", NULL))
 			return 1;
 	}
@@ -252,17 +274,21 @@ static int of_bus_pci_map(u32 *addr, const u32 *range,
 	return 0;
 }
 
-static unsigned int of_bus_pci_get_flags(const u32 *addr)
+static unsigned long of_bus_pci_get_flags(const u32 *addr, unsigned long flags)
 {
-	unsigned int flags = 0;
 	u32 w = addr[0];
 
+	/* For PCI, we override whatever child busses may have used.  */
+	flags = 0;
 	switch((w >> 24) & 0x03) {
 	case 0x01:
 		flags |= IORESOURCE_IO;
+		break;
+
 	case 0x02: /* 32 bits */
 	case 0x03: /* 64 bits */
 		flags |= IORESOURCE_MEM;
+		break;
 	}
 	if (w & 0x40000000)
 		flags |= IORESOURCE_PREFETCH;
@@ -375,8 +401,7 @@ static int __init build_one_resource(struct device_node *parent,
 				     int na, int ns, int pna)
 {
 	const u32 *ranges;
-	unsigned int rlen;
-	int rone;
+	int rone, rlen;
 
 	ranges = of_get_property(parent, "ranges", &rlen);
 	if (ranges == NULL || rlen == 0) {
@@ -418,15 +443,24 @@ static int __init use_1to1_mapping(struct device_node *pp)
 
 	/* If the parent is the dma node of an ISA bus, pass
 	 * the translation up to the root.
+	 *
+	 * Some SBUS devices use intermediate nodes to express
+	 * hierarchy within the device itself.  These aren't
+	 * real bus nodes, and don't have a 'ranges' property.
+	 * But, we should still pass the translation work up
+	 * to the SBUS itself.
 	 */
-	if (!strcmp(pp->name, "dma"))
+	if (!strcmp(pp->name, "dma") ||
+	    !strcmp(pp->name, "espdma") ||
+	    !strcmp(pp->name, "ledma") ||
+	    !strcmp(pp->name, "lebuffer"))
 		return 0;
 
 	/* Similarly for all PCI bridges, if we get this far
 	 * it lacks a ranges property, and this will include
 	 * cases like Simba.
 	 */
-	if (!strcmp(pp->type, "pci") || !strcmp(pp->type, "pciex"))
+	if (!strcmp(pp->name, "pci"))
 		return 0;
 
 	return 1;
@@ -481,10 +515,10 @@ static void __init build_device_resources(struct of_device *op,
 		int pna, pns;
 
 		size = of_read_addr(reg + na, ns);
-		flags = bus->get_flags(reg);
-
 		memcpy(addr, reg, na * 4);
 
+		flags = bus->get_flags(addr, 0);
+
 		if (use_1to1_mapping(pp)) {
 			result = of_read_addr(addr, na);
 			goto build_res;
@@ -509,6 +543,8 @@ static void __init build_device_resources(struct of_device *op,
 					       dna, dns, pna))
 				break;
 
+			flags = pbus->get_flags(addr, flags);
+
 			dna = pna;
 			dns = pns;
 			dbus = pbus;
@@ -709,8 +745,7 @@ static unsigned int __init build_one_device_irq(struct of_device *op,
 				break;
 			}
 		} else {
-			if (!strcmp(pp->type, "pci") ||
-			    !strcmp(pp->type, "pciex")) {
+			if (!strcmp(pp->name, "pci")) {
 				unsigned int this_orig_irq = irq;
 
 				irq = pci_irq_swizzle(dp, pp, irq);
@@ -797,9 +832,9 @@ static struct of_device * __init scan_one_device(struct device_node *dp,
 	op->dev.parent = parent;
 	op->dev.bus = &of_platform_bus_type;
 	if (!parent)
-		strcpy(op->dev.bus_id, "root");
+		dev_set_name(&op->dev, "root");
 	else
-		sprintf(op->dev.bus_id, "%08x", dp->node);
+		dev_set_name(&op->dev, "%08x", dp->node);
 
 	if (of_device_register(op)) {
 		printk("%s: Could not register of device.\n",
@@ -840,17 +875,6 @@ static int __init of_bus_driver_init(void)
 	int err;
 
 	err = of_bus_type_init(&of_platform_bus_type, "of");
-#ifdef CONFIG_PCI
-	if (!err)
-		err = of_bus_type_init(&isa_bus_type, "isa");
-	if (!err)
-		err = of_bus_type_init(&ebus_bus_type, "ebus");
-#endif
-#ifdef CONFIG_SBUS
-	if (!err)
-		err = of_bus_type_init(&sbus_bus_type, "sbus");
-#endif
-
 	if (!err)
 		scan_of_devices();
 
diff --git a/arch/sparc64/kernel/pci.c b/arch/sparc64/kernel/pci.c
index 112b09f16f3..242ac1ccae7 100644
--- a/arch/sparc64/kernel/pci.c
+++ b/arch/sparc64/kernel/pci.c
@@ -18,32 +18,17 @@
 #include <linux/msi.h>
 #include <linux/irq.h>
 #include <linux/init.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
 #include <asm/irq.h>
-#include <asm/ebus.h>
 #include <asm/prom.h>
 #include <asm/apb.h>
 
 #include "pci_impl.h"
 
-#ifndef CONFIG_PCI
-/* A "nop" PCI implementation. */
-asmlinkage int sys_pciconfig_read(unsigned long bus, unsigned long dfn,
-				  unsigned long off, unsigned long len,
-				  unsigned char *buf)
-{
-	return 0;
-}
-asmlinkage int sys_pciconfig_write(unsigned long bus, unsigned long dfn,
-				   unsigned long off, unsigned long len,
-				   unsigned char *buf)
-{
-	return 0;
-}
-#else
-
 /* List of all PCI controllers found in the system. */
 struct pci_pbm_info *pci_pbm_root = NULL;
 
@@ -179,97 +164,6 @@ void pci_config_write32(u32 *addr, u32 val)
 	spin_unlock_irqrestore(&pci_poke_lock, flags);
 }
 
-/* Probe for all PCI controllers in the system. */
-extern void sabre_init(struct device_node *, const char *);
-extern void psycho_init(struct device_node *, const char *);
-extern void schizo_init(struct device_node *, const char *);
-extern void schizo_plus_init(struct device_node *, const char *);
-extern void tomatillo_init(struct device_node *, const char *);
-extern void sun4v_pci_init(struct device_node *, const char *);
-extern void fire_pci_init(struct device_node *, const char *);
-
-static struct {
-	char *model_name;
-	void (*init)(struct device_node *, const char *);
-} pci_controller_table[] __initdata = {
-	{ "SUNW,sabre", sabre_init },
-	{ "pci108e,a000", sabre_init },
-	{ "pci108e,a001", sabre_init },
-	{ "SUNW,psycho", psycho_init },
-	{ "pci108e,8000", psycho_init },
-	{ "SUNW,schizo", schizo_init },
-	{ "pci108e,8001", schizo_init },
-	{ "SUNW,schizo+", schizo_plus_init },
-	{ "pci108e,8002", schizo_plus_init },
-	{ "SUNW,tomatillo", tomatillo_init },
-	{ "pci108e,a801", tomatillo_init },
-	{ "SUNW,sun4v-pci", sun4v_pci_init },
-	{ "pciex108e,80f0", fire_pci_init },
-};
-#define PCI_NUM_CONTROLLER_TYPES	ARRAY_SIZE(pci_controller_table)
-
-static int __init pci_controller_init(const char *model_name, int namelen, struct device_node *dp)
-{
-	int i;
-
-	for (i = 0; i < PCI_NUM_CONTROLLER_TYPES; i++) {
-		if (!strncmp(model_name,
-			     pci_controller_table[i].model_name,
-			     namelen)) {
-			pci_controller_table[i].init(dp, model_name);
-			return 1;
-		}
-	}
-
-	return 0;
-}
-
-static int __init pci_controller_scan(int (*handler)(const char *, int, struct device_node *))
-{
-	struct device_node *dp;
-	int count = 0;
-
-	for_each_node_by_name(dp, "pci") {
-		struct property *prop;
-		int len;
-
-		prop = of_find_property(dp, "model", &len);
-		if (!prop)
-			prop = of_find_property(dp, "compatible", &len);
-
-		if (prop) {
-			const char *model = prop->value;
-			int item_len = 0;
-
-			/* Our value may be a multi-valued string in the
-			 * case of some compatible properties. For sanity,
-			 * only try the first one.
-			 */
-			while (model[item_len] && len) {
-				len--;
-				item_len++;
-			}
-
-			if (handler(model, item_len, dp))
-				count++;
-		}
-	}
-
-	return count;
-}
-
-/* Find each controller in the system, attach and initialize
- * software state structure for each and link into the
- * pci_pbm_root.  Setup the controller enough such
- * that bus scanning can be done.
- */
-static void __init pci_controller_probe(void)
-{
-	printk("PCI: Probing for controllers.\n");
-
-	pci_controller_scan(pci_controller_init);
-}
-
 static int ofpci_verbose;
 
 static int __init ofpci_debug(char *str)
@@ -348,11 +242,12 @@ static void pci_parse_of_addrs(struct of_device *op,
 	}
 }
 
-struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm,
-				  struct device_node *node,
-				  struct pci_bus *bus, int devfn)
+static struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm,
+					 struct device_node *node,
+					 struct pci_bus *bus, int devfn)
 {
 	struct dev_archdata *sd;
+	struct of_device *op;
 	struct pci_dev *dev;
 	const char *type;
 	u32 class;
@@ -366,14 +261,17 @@ struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm,
 	sd->stc = &pbm->stc;
 	sd->host_controller = pbm;
 	sd->prom_node = node;
-	sd->op = of_find_device_by_node(node);
+	sd->op = op = of_find_device_by_node(node);
 	sd->numa_node = pbm->numa_node;
 
-	sd = &sd->op->dev.archdata;
+	sd = &op->dev.archdata;
 	sd->iommu = pbm->iommu;
 	sd->stc = &pbm->stc;
 	sd->numa_node = pbm->numa_node;
 
+	if (!strcmp(node->name, "ebus"))
+		of_propagate_archdata(op);
+
 	type = of_get_property(node, "device_type", NULL);
 	if (type == NULL)
 		type = "";
@@ -408,7 +306,7 @@ struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm,
 	dev->class = class >> 8;
 	dev->revision = class & 0xff;
 
-	sprintf(pci_name(dev), "%04x:%02x:%02x.%d", pci_domain_nr(bus),
+	dev_set_name(&dev->dev, "%04x:%02x:%02x.%d", pci_domain_nr(bus),
 		dev->bus->number, PCI_SLOT(devfn), PCI_FUNC(devfn));
 
 	if (ofpci_verbose)
@@ -425,7 +323,7 @@ struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm,
 	dev->current_state = 4;		/* unknown power state */
 	dev->error_state = pci_channel_io_normal;
 
-	if (!strcmp(type, "pci") || !strcmp(type, "pciex")) {
+	if (!strcmp(node->name, "pci")) {
 		/* a PCI-PCI bridge */
 		dev->hdr_type = PCI_HEADER_TYPE_BRIDGE;
 		dev->rom_base_reg = PCI_ROM_ADDRESS1;
@@ -775,15 +673,15 @@ static void __devinit pci_bus_register_of_sysfs(struct pci_bus *bus)
 		pci_bus_register_of_sysfs(child_bus);
 }
 
-struct pci_bus * __devinit pci_scan_one_pbm(struct pci_pbm_info *pbm)
+struct pci_bus * __devinit pci_scan_one_pbm(struct pci_pbm_info *pbm,
+					    struct device *parent)
 {
-	struct device_node *node = pbm->prom_node;
+	struct device_node *node = pbm->op->node;
 	struct pci_bus *bus;
 
 	printk("PCI: Scanning PBM %s\n", node->full_name);
 
-	/* XXX parent device? XXX */
-	bus = pci_create_bus(NULL, pbm->pci_first_busno, pbm->pci_ops, pbm);
+	bus = pci_create_bus(parent, pbm->pci_first_busno, pbm->pci_ops, pbm);
 	if (!bus) {
 		printk(KERN_ERR "Failed to create bus for %s\n",
 		       node->full_name);
@@ -802,32 +700,6 @@ struct pci_bus * __devinit pci_scan_one_pbm(struct pci_pbm_info *pbm)
 	return bus;
 }
 
-static void __init pci_scan_each_controller_bus(void)
-{
-	struct pci_pbm_info *pbm;
-
-	for (pbm = pci_pbm_root; pbm; pbm = pbm->next)
-		pbm->scan_bus(pbm);
-}
-
-extern void power_init(void);
-
-static int __init pcibios_init(void)
-{
-	pci_controller_probe();
-	if (pci_pbm_root == NULL)
-		return 0;
-
-	pci_scan_each_controller_bus();
-
-	ebus_init();
-	power_init();
-
-	return 0;
-}
-
-subsys_initcall(pcibios_init);
-
 void __devinit pcibios_fixup_bus(struct pci_bus *pbus)
 {
 	struct pci_pbm_info *pbm = pbus->sysdata;
@@ -1105,14 +977,14 @@ int pcibus_to_node(struct pci_bus *pbus)
 EXPORT_SYMBOL(pcibus_to_node);
 #endif
 
-/* Return the domain nuber for this pci bus */
+/* Return the domain number for this pci bus */
 
 int pci_domain_nr(struct pci_bus *pbus)
 {
 	struct pci_pbm_info *pbm = pbus->sysdata;
 	int ret;
 
-	if (pbm == NULL || pbm->parent == NULL) {
+	if (!pbm) {
 		ret = -ENXIO;
 	} else {
 		ret = pbm->index;
@@ -1126,7 +998,7 @@ EXPORT_SYMBOL(pci_domain_nr);
 int arch_setup_msi_irq(struct pci_dev *pdev, struct msi_desc *desc)
 {
 	struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
-	int virt_irq;
+	unsigned int virt_irq;
 
 	if (!pbm->setup_msi_irq)
 		return -EINVAL;
@@ -1140,10 +1012,8 @@ void arch_teardown_msi_irq(unsigned int virt_irq)
 	struct pci_dev *pdev = entry->dev;
 	struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
 
-	if (!pbm->teardown_msi_irq)
-		return;
-
-	return pbm->teardown_msi_irq(virt_irq, pdev);
+	if (pbm->teardown_msi_irq)
+		pbm->teardown_msi_irq(virt_irq, pdev);
 }
 #endif /* !(CONFIG_PCI_MSI) */
 
@@ -1215,5 +1085,3 @@ void pci_resource_to_user(const struct pci_dev *pdev, int bar,
 	*start = rp->start - offset;
 	*end = rp->end - offset;
 }
-
-#endif /* !(CONFIG_PCI) */
diff --git a/arch/sparc64/kernel/pci_common.c b/arch/sparc64/kernel/pci_common.c
index 19fa621d6a6..23b88082d0b 100644
--- a/arch/sparc64/kernel/pci_common.c
+++ b/arch/sparc64/kernel/pci_common.c
@@ -8,9 +8,9 @@
 #include <linux/init.h>
 #include <linux/pci.h>
 #include <linux/device.h>
+#include <linux/of_device.h>
 
 #include <asm/prom.h>
-#include <asm/of_device.h>
 #include <asm/oplib.h>
 
 #include "pci_impl.h"
@@ -314,12 +314,12 @@ struct pci_ops sun4v_pci_ops = {
 
 void pci_get_pbm_props(struct pci_pbm_info *pbm)
 {
-	const u32 *val = of_get_property(pbm->prom_node, "bus-range", NULL);
+	const u32 *val = of_get_property(pbm->op->node, "bus-range", NULL);
 
 	pbm->pci_first_busno = val[0];
 	pbm->pci_last_busno = val[1];
 
-	val = of_get_property(pbm->prom_node, "ino-bitmap", NULL);
+	val = of_get_property(pbm->op->node, "ino-bitmap", NULL);
 	if (val) {
 		pbm->ino_bitmap = (((u64)val[1] << 32UL) |
 				   ((u64)val[0] <<  0UL));
@@ -365,7 +365,7 @@ static void pci_register_legacy_regions(struct resource *io_res,
 
 static void pci_register_iommu_region(struct pci_pbm_info *pbm)
 {
-	const u32 *vdma = of_get_property(pbm->prom_node, "virtual-dma", NULL);
+	const u32 *vdma = of_get_property(pbm->op->node, "virtual-dma", NULL);
 
 	if (vdma) {
 		struct resource *rp = kmalloc(sizeof(*rp), GFP_KERNEL);
@@ -389,7 +389,7 @@ void pci_determine_mem_io_space(struct pci_pbm_info *pbm)
 	int num_pbm_ranges;
 
 	saw_mem = saw_io = 0;
-	pbm_ranges = of_get_property(pbm->prom_node, "ranges", &i);
+	pbm_ranges = of_get_property(pbm->op->node, "ranges", &i);
 	if (!pbm_ranges) {
 		prom_printf("PCI: Fatal error, missing PBM ranges property "
 			    " for %s\n",
diff --git a/arch/sparc64/kernel/pci_fire.c b/arch/sparc64/kernel/pci_fire.c
index d23bb6f53cd..9462b68f489 100644
--- a/arch/sparc64/kernel/pci_fire.c
+++ b/arch/sparc64/kernel/pci_fire.c
@@ -8,34 +8,16 @@
 #include <linux/init.h>
 #include <linux/msi.h>
 #include <linux/irq.h>
+#include <linux/of_device.h>
 
-#include <asm/oplib.h>
 #include <asm/prom.h>
 #include <asm/irq.h>
+#include <asm/upa.h>
 
 #include "pci_impl.h"
 
-#define fire_read(__reg) \
-({	u64 __ret; \
-	__asm__ __volatile__("ldxa [%1] %2, %0" \
-			     : "=r" (__ret) \
-			     : "r" (__reg), "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory"); \
-	__ret; \
-})
-#define fire_write(__reg, __val) \
-	__asm__ __volatile__("stxa %0, [%1] %2" \
-			     : /* no outputs */ \
-			     : "r" (__val), "r" (__reg), \
-			       "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory")
-
-static void __init pci_fire_scan_bus(struct pci_pbm_info *pbm)
-{
-	pbm->pci_bus = pci_scan_one_pbm(pbm);
-
-	/* XXX register error interrupt handlers XXX */
-}
+#define DRIVER_NAME	"fire"
+#define PFX		DRIVER_NAME ": "
 
 #define FIRE_IOMMU_CONTROL	0x40000UL
 #define FIRE_IOMMU_TSBBASE	0x40008UL
@@ -69,21 +51,21 @@ static int pci_fire_pbm_iommu_init(struct pci_pbm_info *pbm)
 	/*
 	 * Invalidate TLB Entries.
 	 */
-	fire_write(iommu->iommu_flushinv, ~(u64)0);
+	upa_writeq(~(u64)0, iommu->iommu_flushinv);
 
 	err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask,
 			       pbm->numa_node);
 	if (err)
 		return err;
 
-	fire_write(iommu->iommu_tsbbase, __pa(iommu->page_table) | 0x7UL);
+	upa_writeq(__pa(iommu->page_table) | 0x7UL, iommu->iommu_tsbbase);
 
-	control = fire_read(iommu->iommu_control);
+	control = upa_readq(iommu->iommu_control);
 	control |= (0x00000400 /* TSB cache snoop enable */	|
 		    0x00000300 /* Cache mode */			|
 		    0x00000002 /* Bypass enable */		|
 		    0x00000001 /* Translation enable */);
-	fire_write(iommu->iommu_control, control);
+	upa_writeq(control, iommu->iommu_control);
 
 	return 0;
 }
@@ -165,7 +147,7 @@ struct pci_msiq_entry {
 static int pci_fire_get_head(struct pci_pbm_info *pbm, unsigned long msiqid,
 			     unsigned long *head)
 {
-	*head = fire_read(pbm->pbm_regs + EVENT_QUEUE_HEAD(msiqid));
+	*head = upa_readq(pbm->pbm_regs + EVENT_QUEUE_HEAD(msiqid));
 	return 0;
 }
 
@@ -191,8 +173,7 @@ static int pci_fire_dequeue_msi(struct pci_pbm_info *pbm, unsigned long msiqid,
 	*msi = msi_num = ((ep->word0 & MSIQ_WORD0_DATA0) >>
 			  MSIQ_WORD0_DATA0_SHIFT);
 
-	fire_write(pbm->pbm_regs + MSI_CLEAR(msi_num),
-		   MSI_CLEAR_EQWR_N);
+	upa_writeq(MSI_CLEAR_EQWR_N, pbm->pbm_regs + MSI_CLEAR(msi_num));
 
 	/* Clear the entry.  */
 	ep->word0 &= ~MSIQ_WORD0_FMT_TYPE;
@@ -208,7 +189,7 @@ static int pci_fire_dequeue_msi(struct pci_pbm_info *pbm, unsigned long msiqid,
 static int pci_fire_set_head(struct pci_pbm_info *pbm, unsigned long msiqid,
 			     unsigned long head)
 {
-	fire_write(pbm->pbm_regs + EVENT_QUEUE_HEAD(msiqid), head);
+	upa_writeq(head, pbm->pbm_regs + EVENT_QUEUE_HEAD(msiqid));
 	return 0;
 }
 
@@ -217,17 +198,16 @@ static int pci_fire_msi_setup(struct pci_pbm_info *pbm, unsigned long msiqid,
 {
 	u64 val;
 
-	val = fire_read(pbm->pbm_regs + MSI_MAP(msi));
+	val = upa_readq(pbm->pbm_regs + MSI_MAP(msi));
 	val &= ~(MSI_MAP_EQNUM);
 	val |= msiqid;
-	fire_write(pbm->pbm_regs + MSI_MAP(msi), val);
+	upa_writeq(val, pbm->pbm_regs + MSI_MAP(msi));
 
-	fire_write(pbm->pbm_regs + MSI_CLEAR(msi),
-		   MSI_CLEAR_EQWR_N);
+	upa_writeq(MSI_CLEAR_EQWR_N, pbm->pbm_regs + MSI_CLEAR(msi));
 
-	val = fire_read(pbm->pbm_regs + MSI_MAP(msi));
+	val = upa_readq(pbm->pbm_regs + MSI_MAP(msi));
 	val |= MSI_MAP_VALID;
-	fire_write(pbm->pbm_regs + MSI_MAP(msi), val);
+	upa_writeq(val, pbm->pbm_regs + MSI_MAP(msi));
 
 	return 0;
 }
@@ -237,12 +217,12 @@ static int pci_fire_msi_teardown(struct pci_pbm_info *pbm, unsigned long msi)
 	unsigned long msiqid;
 	u64 val;
 
-	val = fire_read(pbm->pbm_regs + MSI_MAP(msi));
+	val = upa_readq(pbm->pbm_regs + MSI_MAP(msi));
 	msiqid = (val & MSI_MAP_EQNUM);
 
 	val &= ~MSI_MAP_VALID;
 
-	fire_write(pbm->pbm_regs + MSI_MAP(msi), val);
+	upa_writeq(val, pbm->pbm_regs + MSI_MAP(msi));
 
 	return 0;
 }
@@ -261,22 +241,19 @@ static int pci_fire_msiq_alloc(struct pci_pbm_info *pbm)
 	memset((char *)pages, 0, PAGE_SIZE << order);
 	pbm->msi_queues = (void *) pages;
 
-	fire_write(pbm->pbm_regs + EVENT_QUEUE_BASE_ADDR_REG,
-		   (EVENT_QUEUE_BASE_ADDR_ALL_ONES |
-		    __pa(pbm->msi_queues)));
+	upa_writeq((EVENT_QUEUE_BASE_ADDR_ALL_ONES |
+		    __pa(pbm->msi_queues)),
+		   pbm->pbm_regs + EVENT_QUEUE_BASE_ADDR_REG);
 
-	fire_write(pbm->pbm_regs + IMONDO_DATA0,
-		   pbm->portid << 6);
-	fire_write(pbm->pbm_regs + IMONDO_DATA1, 0);
+	upa_writeq(pbm->portid << 6, pbm->pbm_regs + IMONDO_DATA0);
+	upa_writeq(0, pbm->pbm_regs + IMONDO_DATA1);
 
-	fire_write(pbm->pbm_regs + MSI_32BIT_ADDR,
-		   pbm->msi32_start);
-	fire_write(pbm->pbm_regs + MSI_64BIT_ADDR,
-		   pbm->msi64_start);
+	upa_writeq(pbm->msi32_start, pbm->pbm_regs + MSI_32BIT_ADDR);
+	upa_writeq(pbm->msi64_start, pbm->pbm_regs + MSI_64BIT_ADDR);
 
 	for (i = 0; i < pbm->msiq_num; i++) {
-		fire_write(pbm->pbm_regs + EVENT_QUEUE_HEAD(i), 0);
-		fire_write(pbm->pbm_regs + EVENT_QUEUE_TAIL(i), 0);
+		upa_writeq(0, pbm->pbm_regs + EVENT_QUEUE_HEAD(i));
+		upa_writeq(0, pbm->pbm_regs + EVENT_QUEUE_TAIL(i));
 	}
 
 	return 0;
@@ -310,9 +287,9 @@ static int pci_fire_msiq_build_irq(struct pci_pbm_info *pbm,
 	/* XXX iterate amongst the 4 IRQ controllers XXX */
 	int_ctrlr = (1UL << 6);
 
-	val = fire_read(imap_reg);
+	val = upa_readq(imap_reg);
 	val |= (1UL << 63) | int_ctrlr;
-	fire_write(imap_reg, val);
+	upa_writeq(val, imap_reg);
 
 	fixup = ((pbm->portid << 6) | devino) - int_ctrlr;
 
@@ -320,9 +297,8 @@ static int pci_fire_msiq_build_irq(struct pci_pbm_info *pbm,
 	if (!virt_irq)
 		return -ENOMEM;
 
-	fire_write(pbm->pbm_regs +
-		   EVENT_QUEUE_CONTROL_SET(msiqid),
-		   EVENT_QUEUE_CONTROL_SET_EN);
+	upa_writeq(EVENT_QUEUE_CONTROL_SET_EN,
+		   pbm->pbm_regs + EVENT_QUEUE_CONTROL_SET(msiqid));
 
 	return virt_irq;
 }
@@ -390,77 +366,65 @@ static void pci_fire_hw_init(struct pci_pbm_info *pbm)
 {
 	u64 val;
 
-	fire_write(pbm->controller_regs + FIRE_PARITY_CONTROL,
-		   FIRE_PARITY_ENAB);
+	upa_writeq(FIRE_PARITY_ENAB,
+		   pbm->controller_regs + FIRE_PARITY_CONTROL);
 
-	fire_write(pbm->controller_regs + FIRE_FATAL_RESET_CTL,
-		   (FIRE_FATAL_RESET_SPARE |
+	upa_writeq((FIRE_FATAL_RESET_SPARE |
 		    FIRE_FATAL_RESET_MB |
 		    FIRE_FATAL_RESET_CPE |
 		    FIRE_FATAL_RESET_APE |
 		    FIRE_FATAL_RESET_PIO |
 		    FIRE_FATAL_RESET_JW |
 		    FIRE_FATAL_RESET_JI |
-		    FIRE_FATAL_RESET_JR));
+		    FIRE_FATAL_RESET_JR),
+		   pbm->controller_regs + FIRE_FATAL_RESET_CTL);
 
-	fire_write(pbm->controller_regs + FIRE_CORE_INTR_ENABLE, ~(u64)0);
+	upa_writeq(~(u64)0, pbm->controller_regs + FIRE_CORE_INTR_ENABLE);
 
-	val = fire_read(pbm->pbm_regs + FIRE_TLU_CTRL);
+	val = upa_readq(pbm->pbm_regs + FIRE_TLU_CTRL);
 	val |= (FIRE_TLU_CTRL_TIM |
 		FIRE_TLU_CTRL_QDET |
 		FIRE_TLU_CTRL_CFG);
-	fire_write(pbm->pbm_regs + FIRE_TLU_CTRL, val);
-	fire_write(pbm->pbm_regs + FIRE_TLU_DEV_CTRL, 0);
-	fire_write(pbm->pbm_regs + FIRE_TLU_LINK_CTRL,
-		   FIRE_TLU_LINK_CTRL_CLK);
-
-	fire_write(pbm->pbm_regs + FIRE_LPU_RESET, 0);
-	fire_write(pbm->pbm_regs + FIRE_LPU_LLCFG,
-		   FIRE_LPU_LLCFG_VC0);
-	fire_write(pbm->pbm_regs + FIRE_LPU_FCTRL_UCTRL,
-		   (FIRE_LPU_FCTRL_UCTRL_N |
-		    FIRE_LPU_FCTRL_UCTRL_P));
-	fire_write(pbm->pbm_regs + FIRE_LPU_TXL_FIFOP,
-		   ((0xffff << 16) | (0x0000 << 0)));
-	fire_write(pbm->pbm_regs + FIRE_LPU_LTSSM_CFG2, 3000000);
-	fire_write(pbm->pbm_regs + FIRE_LPU_LTSSM_CFG3, 500000);
-	fire_write(pbm->pbm_regs + FIRE_LPU_LTSSM_CFG4,
-		   (2 << 16) | (140 << 8));
-	fire_write(pbm->pbm_regs + FIRE_LPU_LTSSM_CFG5, 0);
-
-	fire_write(pbm->pbm_regs + FIRE_DMC_IENAB, ~(u64)0);
-	fire_write(pbm->pbm_regs + FIRE_DMC_DBG_SEL_A, 0);
-	fire_write(pbm->pbm_regs + FIRE_DMC_DBG_SEL_B, 0);
-
-	fire_write(pbm->pbm_regs + FIRE_PEC_IENAB, ~(u64)0);
+	upa_writeq(val, pbm->pbm_regs + FIRE_TLU_CTRL);
+	upa_writeq(0, pbm->pbm_regs + FIRE_TLU_DEV_CTRL);
+	upa_writeq(FIRE_TLU_LINK_CTRL_CLK,
+		   pbm->pbm_regs + FIRE_TLU_LINK_CTRL);
+
+	upa_writeq(0, pbm->pbm_regs + FIRE_LPU_RESET);
+	upa_writeq(FIRE_LPU_LLCFG_VC0, pbm->pbm_regs + FIRE_LPU_LLCFG);
+	upa_writeq((FIRE_LPU_FCTRL_UCTRL_N | FIRE_LPU_FCTRL_UCTRL_P),
+		   pbm->pbm_regs + FIRE_LPU_FCTRL_UCTRL);
+	upa_writeq(((0xffff << 16) | (0x0000 << 0)),
+		   pbm->pbm_regs + FIRE_LPU_TXL_FIFOP);
+	upa_writeq(3000000, pbm->pbm_regs + FIRE_LPU_LTSSM_CFG2);
+	upa_writeq(500000, pbm->pbm_regs + FIRE_LPU_LTSSM_CFG3);
+	upa_writeq((2 << 16) | (140 << 8),
+		   pbm->pbm_regs + FIRE_LPU_LTSSM_CFG4);
+	upa_writeq(0, pbm->pbm_regs + FIRE_LPU_LTSSM_CFG5);
+
+	upa_writeq(~(u64)0, pbm->pbm_regs + FIRE_DMC_IENAB);
+	upa_writeq(0, pbm->pbm_regs + FIRE_DMC_DBG_SEL_A);
+	upa_writeq(0, pbm->pbm_regs + FIRE_DMC_DBG_SEL_B);
+
+	upa_writeq(~(u64)0, pbm->pbm_regs + FIRE_PEC_IENAB);
 }
 
-static int __init pci_fire_pbm_init(struct pci_controller_info *p,
-				    struct device_node *dp, u32 portid)
+static int __init pci_fire_pbm_init(struct pci_pbm_info *pbm,
+				    struct of_device *op, u32 portid)
 {
 	const struct linux_prom64_registers *regs;
-	struct pci_pbm_info *pbm;
+	struct device_node *dp = op->node;
 	int err;
 
-	if ((portid & 1) == 0)
-		pbm = &p->pbm_A;
-	else
-		pbm = &p->pbm_B;
-
-	pbm->next = pci_pbm_root;
-	pci_pbm_root = pbm;
-
 	pbm->numa_node = -1;
 
-	pbm->scan_bus = pci_fire_scan_bus;
 	pbm->pci_ops = &sun4u_pci_ops;
 	pbm->config_space_reg_bits = 12;
 
 	pbm->index = pci_num_pbms++;
 
 	pbm->portid = portid;
-	pbm->parent = p;
-	pbm->prom_node = dp;
+	pbm->op = op;
 	pbm->name = dp->full_name;
 
 	regs = of_get_property(dp, "reg", NULL);
@@ -481,53 +445,77 @@ static int __init pci_fire_pbm_init(struct pci_controller_info *p,
 
 	pci_fire_msi_init(pbm);
 
-	return 0;
-}
+	pbm->pci_bus = pci_scan_one_pbm(pbm, &op->dev);
+
+	/* XXX register error interrupt handlers XXX */
+
+	pbm->next = pci_pbm_root;
+	pci_pbm_root = pbm;
 
-static inline int portid_compare(u32 x, u32 y)
-{
-	if (x == (y ^ 1))
-		return 1;
 	return 0;
 }
 
-void __init fire_pci_init(struct device_node *dp, const char *model_name)
+static int __devinit fire_probe(struct of_device *op,
+				const struct of_device_id *match)
 {
-	struct pci_controller_info *p;
-	u32 portid = of_getintprop_default(dp, "portid", 0xff);
-	struct iommu *iommu;
+	struct device_node *dp = op->node;
 	struct pci_pbm_info *pbm;
+	struct iommu *iommu;
+	u32 portid;
+	int err;
 
-	for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
-		if (portid_compare(pbm->portid, portid)) {
-			if (pci_fire_pbm_init(pbm->parent, dp, portid))
-				goto fatal_memory_error;
-			return;
-		}
+	portid = of_getintprop_default(dp, "portid", 0xff);
+
+	err = -ENOMEM;
+	pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+	if (!pbm) {
+		printk(KERN_ERR PFX "Cannot allocate pci_pbminfo.\n");
+		goto out_err;
 	}
 
-	p = kzalloc(sizeof(struct pci_controller_info), GFP_ATOMIC);
-	if (!p)
-		goto fatal_memory_error;
+	iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+	if (!iommu) {
+		printk(KERN_ERR PFX "Cannot allocate PBM iommu.\n");
+		goto out_free_controller;
+	}
 
-	iommu = kzalloc(sizeof(struct iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
+	pbm->iommu = iommu;
 
-	p->pbm_A.iommu = iommu;
+	err = pci_fire_pbm_init(pbm, op, portid);
+	if (err)
+		goto out_free_iommu;
 
-	iommu = kzalloc(sizeof(struct iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
+	dev_set_drvdata(&op->dev, pbm);
 
-	p->pbm_B.iommu = iommu;
+	return 0;
 
-	if (pci_fire_pbm_init(p, dp, portid))
-		goto fatal_memory_error;
+out_free_iommu:
+	kfree(pbm->iommu);
+			
+out_free_controller:
+	kfree(pbm);
 
-	return;
+out_err:
+	return err;
+}
+
+static struct of_device_id __initdata fire_match[] = {
+	{
+		.name = "pci",
+		.compatible = "pciex108e,80f0",
+	},
+	{},
+};
 
-fatal_memory_error:
-	prom_printf("PCI_FIRE: Fatal memory allocation error.\n");
-	prom_halt();
+static struct of_platform_driver fire_driver = {
+	.name		= DRIVER_NAME,
+	.match_table	= fire_match,
+	.probe		= fire_probe,
+};
+
+static int __init fire_init(void)
+{
+	return of_register_driver(&fire_driver, &of_bus_type);
 }
+
+subsys_initcall(fire_init);
diff --git a/arch/sparc64/kernel/pci_impl.h b/arch/sparc64/kernel/pci_impl.h
index c385d126be1..03186824327 100644
--- a/arch/sparc64/kernel/pci_impl.h
+++ b/arch/sparc64/kernel/pci_impl.h
@@ -10,6 +10,7 @@
 #include <linux/spinlock.h>
 #include <linux/pci.h>
 #include <linux/msi.h>
+#include <linux/of_device.h>
 #include <asm/io.h>
 #include <asm/prom.h>
 #include <asm/iommu.h>
@@ -56,15 +57,11 @@ struct sparc64_msiq_cookie {
 };
 #endif
 
-struct pci_controller_info;
-
 struct pci_pbm_info {
 	struct pci_pbm_info		*next;
+	struct pci_pbm_info		*sibling;
 	int				index;
 
-	/* PCI controller we sit under. */
-	struct pci_controller_info	*parent;
-
 	/* Physical address base of controller registers. */
 	unsigned long			controller_regs;
 
@@ -94,7 +91,7 @@ struct pci_pbm_info {
 	char				*name;
 
 	/* OBP specific information. */
-	struct device_node		*prom_node;
+	struct of_device		*op;
 	u64				ino_bitmap;
 
 	/* PBM I/O and Memory space resources. */
@@ -107,6 +104,10 @@ struct pci_pbm_info {
 	/* This will be 12 on PCI-E controllers, 8 elsewhere.  */
 	unsigned long			config_space_reg_bits;
 
+	unsigned long			pci_afsr;
+	unsigned long			pci_afar;
+	unsigned long			pci_csr;
+
 	/* State of 66MHz capabilities on this PBM. */
 	int				is_66mhz_capable;
 	int				all_devs_66mhz;
@@ -146,25 +147,19 @@ struct pci_pbm_info {
 	unsigned int			pci_first_busno;
 	unsigned int			pci_last_busno;
 	struct pci_bus			*pci_bus;
-	void (*scan_bus)(struct pci_pbm_info *);
 	struct pci_ops			*pci_ops;
 
 	int				numa_node;
 };
 
-struct pci_controller_info {
-	/* The PCI bus modules controlled by us. */
-	struct pci_pbm_info		pbm_A;
-	struct pci_pbm_info		pbm_B;
-};
-
 extern struct pci_pbm_info *pci_pbm_root;
 
 extern int pci_num_pbms;
 
 /* PCI bus scanning and fixup support. */
 extern void pci_get_pbm_props(struct pci_pbm_info *pbm);
-extern struct pci_bus *pci_scan_one_pbm(struct pci_pbm_info *pbm);
+extern struct pci_bus *pci_scan_one_pbm(struct pci_pbm_info *pbm,
+					struct device *parent);
 extern void pci_determine_mem_io_space(struct pci_pbm_info *pbm);
 
 /* Error reporting support. */
@@ -183,4 +178,8 @@ extern void pci_config_write32(u32 *addr, u32 val);
 extern struct pci_ops sun4u_pci_ops;
 extern struct pci_ops sun4v_pci_ops;
 
+extern volatile int pci_poke_in_progress;
+extern volatile int pci_poke_cpu;
+extern volatile int pci_poke_faulted;
+
 #endif /* !(PCI_IMPL_H) */
diff --git a/arch/sparc64/kernel/pci_msi.c b/arch/sparc64/kernel/pci_msi.c
index db5e8fd8f67..2e680f34f72 100644
--- a/arch/sparc64/kernel/pci_msi.c
+++ b/arch/sparc64/kernel/pci_msi.c
@@ -120,9 +120,9 @@ static struct irq_chip msi_irq = {
 	/* XXX affinity XXX */
 };
 
-int sparc64_setup_msi_irq(unsigned int *virt_irq_p,
-			  struct pci_dev *pdev,
-			  struct msi_desc *entry)
+static int sparc64_setup_msi_irq(unsigned int *virt_irq_p,
+				 struct pci_dev *pdev,
+				 struct msi_desc *entry)
 {
 	struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
 	const struct sparc64_msiq_ops *ops = pbm->msi_ops;
@@ -179,8 +179,8 @@ out_err:
 	return err;
 }
 
-void sparc64_teardown_msi_irq(unsigned int virt_irq,
-			      struct pci_dev *pdev)
+static void sparc64_teardown_msi_irq(unsigned int virt_irq,
+				     struct pci_dev *pdev)
 {
 	struct pci_pbm_info *pbm = pdev->dev.archdata.host_controller;
 	const struct sparc64_msiq_ops *ops = pbm->msi_ops;
@@ -323,7 +323,7 @@ void sparc64_pbm_msi_init(struct pci_pbm_info *pbm,
 	const u32 *val;
 	int len;
 
-	val = of_get_property(pbm->prom_node, "#msi-eqs", &len);
+	val = of_get_property(pbm->op->node, "#msi-eqs", &len);
 	if (!val || len != 4)
 		goto no_msi;
 	pbm->msiq_num = *val;
@@ -346,16 +346,16 @@ void sparc64_pbm_msi_init(struct pci_pbm_info *pbm,
 			u32 msi64_len;
 		} *arng;
 
-		val = of_get_property(pbm->prom_node, "msi-eq-size", &len);
+		val = of_get_property(pbm->op->node, "msi-eq-size", &len);
 		if (!val || len != 4)
 			goto no_msi;
 
 		pbm->msiq_ent_count = *val;
 
-		mqp = of_get_property(pbm->prom_node,
+		mqp = of_get_property(pbm->op->node,
 				      "msi-eq-to-devino", &len);
 		if (!mqp)
-			mqp = of_get_property(pbm->prom_node,
+			mqp = of_get_property(pbm->op->node,
 					      "msi-eq-devino", &len);
 		if (!mqp || len != sizeof(struct msiq_prop))
 			goto no_msi;
@@ -363,27 +363,27 @@ void sparc64_pbm_msi_init(struct pci_pbm_info *pbm,
 		pbm->msiq_first = mqp->first_msiq;
 		pbm->msiq_first_devino = mqp->first_devino;
 
-		val = of_get_property(pbm->prom_node, "#msi", &len);
+		val = of_get_property(pbm->op->node, "#msi", &len);
 		if (!val || len != 4)
 			goto no_msi;
 		pbm->msi_num = *val;
 
-		mrng = of_get_property(pbm->prom_node, "msi-ranges", &len);
+		mrng = of_get_property(pbm->op->node, "msi-ranges", &len);
 		if (!mrng || len != sizeof(struct msi_range_prop))
 			goto no_msi;
 		pbm->msi_first = mrng->first_msi;
 
-		val = of_get_property(pbm->prom_node, "msi-data-mask", &len);
+		val = of_get_property(pbm->op->node, "msi-data-mask", &len);
 		if (!val || len != 4)
 			goto no_msi;
 		pbm->msi_data_mask = *val;
 
-		val = of_get_property(pbm->prom_node, "msix-data-width", &len);
+		val = of_get_property(pbm->op->node, "msix-data-width", &len);
 		if (!val || len != 4)
 			goto no_msi;
 		pbm->msix_data_width = *val;
 
-		arng = of_get_property(pbm->prom_node, "msi-address-ranges",
+		arng = of_get_property(pbm->op->node, "msi-address-ranges",
 				       &len);
 		if (!arng || len != sizeof(struct addr_range_prop))
 			goto no_msi;
diff --git a/arch/sparc64/kernel/pci_psycho.c b/arch/sparc64/kernel/pci_psycho.c
index 994dbe0603d..dfb3ec89298 100644
--- a/arch/sparc64/kernel/pci_psycho.c
+++ b/arch/sparc64/kernel/pci_psycho.c
@@ -11,35 +11,20 @@
 #include <linux/init.h>
 #include <linux/slab.h>
 #include <linux/interrupt.h>
+#include <linux/of_device.h>
 
 #include <asm/iommu.h>
 #include <asm/irq.h>
 #include <asm/starfire.h>
 #include <asm/prom.h>
-#include <asm/of_device.h>
-#include <asm/oplib.h>
+#include <asm/upa.h>
 
 #include "pci_impl.h"
 #include "iommu_common.h"
+#include "psycho_common.h"
 
-/* All PSYCHO registers are 64-bits.  The following accessor
- * routines are how they are accessed.  The REG parameter
- * is a physical address.
- */
-#define psycho_read(__reg) \
-({	u64 __ret; \
-	__asm__ __volatile__("ldxa [%1] %2, %0" \
-			     : "=r" (__ret) \
-			     : "r" (__reg), "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory"); \
-	__ret; \
-})
-#define psycho_write(__reg, __val) \
-	__asm__ __volatile__("stxa %0, [%1] %2" \
-			     : /* no outputs */ \
-			     : "r" (__val), "r" (__reg), \
-			       "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory")
+#define DRIVER_NAME	"psycho"
+#define PFX		DRIVER_NAME ": "
 
 /* Misc. PSYCHO PCI controller register offsets and definitions. */
 #define PSYCHO_CONTROL		0x0010UL
@@ -67,37 +52,7 @@
 #define  PSYCHO_PCICTRL_RESV4	 0x00000000000000c0UL /* Reserved                     */
 #define  PSYCHO_PCICTRL_AEN	 0x000000000000003fUL /* PCI DVMA Arbitration Enable  */
 
-/* U2P Programmer's Manual, page 13-55, configuration space
- * address format:
- * 
- *  32             24 23 16 15    11 10       8 7   2  1 0
- * ---------------------------------------------------------
- * |0 0 0 0 0 0 0 0 1| bus | device | function | reg | 0 0 |
- * ---------------------------------------------------------
- */
-#define PSYCHO_CONFIG_BASE(PBM)	\
-	((PBM)->config_space | (1UL << 24))
-#define PSYCHO_CONFIG_ENCODE(BUS, DEVFN, REG)	\
-	(((unsigned long)(BUS)   << 16) |	\
-	 ((unsigned long)(DEVFN) << 8)  |	\
-	 ((unsigned long)(REG)))
-
-static void *psycho_pci_config_mkaddr(struct pci_pbm_info *pbm,
-				      unsigned char bus,
-				      unsigned int devfn,
-				      int where)
-{
-	if (!pbm)
-		return NULL;
-	return (void *)
-		(PSYCHO_CONFIG_BASE(pbm) |
-		 PSYCHO_CONFIG_ENCODE(bus, devfn, where));
-}
-
 /* PSYCHO error handling support. */
-enum psycho_error_type {
-	UE_ERR, CE_ERR, PCI_ERR
-};
 
 /* Helper function of IOMMU error checking, which checks out
  * the state of the streaming buffers.  The IOMMU lock is
@@ -122,129 +77,10 @@ enum psycho_error_type {
 #define PSYCHO_STC_DATA_B	0xc000UL
 #define PSYCHO_STC_ERR_A	0xb400UL
 #define PSYCHO_STC_ERR_B	0xc400UL
-#define  PSYCHO_STCERR_WRITE	 0x0000000000000002UL	/* Write Error */
-#define  PSYCHO_STCERR_READ	 0x0000000000000001UL	/* Read Error */
 #define PSYCHO_STC_TAG_A	0xb800UL
 #define PSYCHO_STC_TAG_B	0xc800UL
-#define  PSYCHO_STCTAG_PPN	 0x0fffffff00000000UL	/* Physical Page Number */
-#define  PSYCHO_STCTAG_VPN	 0x00000000ffffe000UL	/* Virtual Page Number */
-#define  PSYCHO_STCTAG_VALID	 0x0000000000000002UL	/* Valid */
-#define  PSYCHO_STCTAG_WRITE	 0x0000000000000001UL	/* Writable */
 #define PSYCHO_STC_LINE_A	0xb900UL
 #define PSYCHO_STC_LINE_B	0xc900UL
-#define  PSYCHO_STCLINE_LINDX	 0x0000000001e00000UL	/* LRU Index */
-#define  PSYCHO_STCLINE_SPTR	 0x00000000001f8000UL	/* Dirty Data Start Pointer */
-#define  PSYCHO_STCLINE_LADDR	 0x0000000000007f00UL	/* Line Address */
-#define  PSYCHO_STCLINE_EPTR	 0x00000000000000fcUL	/* Dirty Data End Pointer */
-#define  PSYCHO_STCLINE_VALID	 0x0000000000000002UL	/* Valid */
-#define  PSYCHO_STCLINE_FOFN	 0x0000000000000001UL	/* Fetch Outstanding / Flush Necessary */
-
-static DEFINE_SPINLOCK(stc_buf_lock);
-static unsigned long stc_error_buf[128];
-static unsigned long stc_tag_buf[16];
-static unsigned long stc_line_buf[16];
-
-static void __psycho_check_one_stc(struct pci_pbm_info *pbm,
-				   int is_pbm_a)
-{
-	struct strbuf *strbuf = &pbm->stc;
-	unsigned long regbase = pbm->controller_regs;
-	unsigned long err_base, tag_base, line_base;
-	u64 control;
-	int i;
-
-	if (is_pbm_a) {
-		err_base = regbase + PSYCHO_STC_ERR_A;
-		tag_base = regbase + PSYCHO_STC_TAG_A;
-		line_base = regbase + PSYCHO_STC_LINE_A;
-	} else {
-		err_base = regbase + PSYCHO_STC_ERR_B;
-		tag_base = regbase + PSYCHO_STC_TAG_B;
-		line_base = regbase + PSYCHO_STC_LINE_B;
-	}
-
-	spin_lock(&stc_buf_lock);
-
-	/* This is __REALLY__ dangerous.  When we put the
-	 * streaming buffer into diagnostic mode to probe
-	 * it's tags and error status, we _must_ clear all
-	 * of the line tag valid bits before re-enabling
-	 * the streaming buffer.  If any dirty data lives
-	 * in the STC when we do this, we will end up
-	 * invalidating it before it has a chance to reach
-	 * main memory.
-	 */
-	control = psycho_read(strbuf->strbuf_control);
-	psycho_write(strbuf->strbuf_control,
-		     (control | PSYCHO_STRBUF_CTRL_DENAB));
-	for (i = 0; i < 128; i++) {
-		unsigned long val;
-
-		val = psycho_read(err_base + (i * 8UL));
-		psycho_write(err_base + (i * 8UL), 0UL);
-		stc_error_buf[i] = val;
-	}
-	for (i = 0; i < 16; i++) {
-		stc_tag_buf[i] = psycho_read(tag_base + (i * 8UL));
-		stc_line_buf[i] = psycho_read(line_base + (i * 8UL));
-		psycho_write(tag_base + (i * 8UL), 0UL);
-		psycho_write(line_base + (i * 8UL), 0UL);
-	}
-
-	/* OK, state is logged, exit diagnostic mode. */
-	psycho_write(strbuf->strbuf_control, control);
-
-	for (i = 0; i < 16; i++) {
-		int j, saw_error, first, last;
-
-		saw_error = 0;
-		first = i * 8;
-		last = first + 8;
-		for (j = first; j < last; j++) {
-			unsigned long errval = stc_error_buf[j];
-			if (errval != 0) {
-				saw_error++;
-				printk("%s: STC_ERR(%d)[wr(%d)rd(%d)]\n",
-				       pbm->name,
-				       j,
-				       (errval & PSYCHO_STCERR_WRITE) ? 1 : 0,
-				       (errval & PSYCHO_STCERR_READ) ? 1 : 0);
-			}
-		}
-		if (saw_error != 0) {
-			unsigned long tagval = stc_tag_buf[i];
-			unsigned long lineval = stc_line_buf[i];
-			printk("%s: STC_TAG(%d)[PA(%016lx)VA(%08lx)V(%d)W(%d)]\n",
-			       pbm->name,
-			       i,
-			       ((tagval & PSYCHO_STCTAG_PPN) >> 19UL),
-			       (tagval & PSYCHO_STCTAG_VPN),
-			       ((tagval & PSYCHO_STCTAG_VALID) ? 1 : 0),
-			       ((tagval & PSYCHO_STCTAG_WRITE) ? 1 : 0));
-			printk("%s: STC_LINE(%d)[LIDX(%lx)SP(%lx)LADDR(%lx)EP(%lx)"
-			       "V(%d)FOFN(%d)]\n",
-			       pbm->name,
-			       i,
-			       ((lineval & PSYCHO_STCLINE_LINDX) >> 21UL),
-			       ((lineval & PSYCHO_STCLINE_SPTR) >> 15UL),
-			       ((lineval & PSYCHO_STCLINE_LADDR) >> 8UL),
-			       ((lineval & PSYCHO_STCLINE_EPTR) >> 2UL),
-			       ((lineval & PSYCHO_STCLINE_VALID) ? 1 : 0),
-			       ((lineval & PSYCHO_STCLINE_FOFN) ? 1 : 0));
-		}
-	}
-
-	spin_unlock(&stc_buf_lock);
-}
-
-static void __psycho_check_stc_error(struct pci_pbm_info *pbm,
-				     unsigned long afsr,
-				     unsigned long afar,
-				     enum psycho_error_type type)
-{
-	__psycho_check_one_stc(pbm,
-			       (pbm == &pbm->parent->pbm_A));
-}
 
 /* When an Uncorrectable Error or a PCI Error happens, we
  * interrogate the IOMMU state to see if it is the cause.
@@ -271,122 +107,7 @@ static void __psycho_check_stc_error(struct pci_pbm_info *pbm,
 #define PSYCHO_IOMMU_TSBBASE	0x0208UL
 #define PSYCHO_IOMMU_FLUSH	0x0210UL
 #define PSYCHO_IOMMU_TAG	0xa580UL
-#define  PSYCHO_IOMMU_TAG_ERRSTS (0x3UL << 23UL)
-#define  PSYCHO_IOMMU_TAG_ERR	 (0x1UL << 22UL)
-#define  PSYCHO_IOMMU_TAG_WRITE	 (0x1UL << 21UL)
-#define  PSYCHO_IOMMU_TAG_STREAM (0x1UL << 20UL)
-#define  PSYCHO_IOMMU_TAG_SIZE	 (0x1UL << 19UL)
-#define  PSYCHO_IOMMU_TAG_VPAGE	 0x7ffffUL
 #define PSYCHO_IOMMU_DATA	0xa600UL
-#define  PSYCHO_IOMMU_DATA_VALID (1UL << 30UL)
-#define  PSYCHO_IOMMU_DATA_CACHE (1UL << 28UL)
-#define  PSYCHO_IOMMU_DATA_PPAGE 0xfffffffUL
-static void psycho_check_iommu_error(struct pci_pbm_info *pbm,
-				     unsigned long afsr,
-				     unsigned long afar,
-				     enum psycho_error_type type)
-{
-	struct iommu *iommu = pbm->iommu;
-	unsigned long iommu_tag[16];
-	unsigned long iommu_data[16];
-	unsigned long flags;
-	u64 control;
-	int i;
-
-	spin_lock_irqsave(&iommu->lock, flags);
-	control = psycho_read(iommu->iommu_control);
-	if (control & PSYCHO_IOMMU_CTRL_XLTEERR) {
-		char *type_string;
-
-		/* Clear the error encountered bit. */
-		control &= ~PSYCHO_IOMMU_CTRL_XLTEERR;
-		psycho_write(iommu->iommu_control, control);
-
-		switch((control & PSYCHO_IOMMU_CTRL_XLTESTAT) >> 25UL) {
-		case 0:
-			type_string = "Protection Error";
-			break;
-		case 1:
-			type_string = "Invalid Error";
-			break;
-		case 2:
-			type_string = "TimeOut Error";
-			break;
-		case 3:
-		default:
-			type_string = "ECC Error";
-			break;
-		};
-		printk("%s: IOMMU Error, type[%s]\n",
-		       pbm->name, type_string);
-
-		/* Put the IOMMU into diagnostic mode and probe
-		 * it's TLB for entries with error status.
-		 *
-		 * It is very possible for another DVMA to occur
-		 * while we do this probe, and corrupt the system
-		 * further.  But we are so screwed at this point
-		 * that we are likely to crash hard anyways, so
-		 * get as much diagnostic information to the
-		 * console as we can.
-		 */
-		psycho_write(iommu->iommu_control,
-			     control | PSYCHO_IOMMU_CTRL_DENAB);
-		for (i = 0; i < 16; i++) {
-			unsigned long base = pbm->controller_regs;
-
-			iommu_tag[i] =
-				psycho_read(base + PSYCHO_IOMMU_TAG + (i * 8UL));
-			iommu_data[i] =
-				psycho_read(base + PSYCHO_IOMMU_DATA + (i * 8UL));
-
-			/* Now clear out the entry. */
-			psycho_write(base + PSYCHO_IOMMU_TAG + (i * 8UL), 0);
-			psycho_write(base + PSYCHO_IOMMU_DATA + (i * 8UL), 0);
-		}
-
-		/* Leave diagnostic mode. */
-		psycho_write(iommu->iommu_control, control);
-
-		for (i = 0; i < 16; i++) {
-			unsigned long tag, data;
-
-			tag = iommu_tag[i];
-			if (!(tag & PSYCHO_IOMMU_TAG_ERR))
-				continue;
-
-			data = iommu_data[i];
-			switch((tag & PSYCHO_IOMMU_TAG_ERRSTS) >> 23UL) {
-			case 0:
-				type_string = "Protection Error";
-				break;
-			case 1:
-				type_string = "Invalid Error";
-				break;
-			case 2:
-				type_string = "TimeOut Error";
-				break;
-			case 3:
-			default:
-				type_string = "ECC Error";
-				break;
-			};
-			printk("%s: IOMMU TAG(%d)[error(%s) wr(%d) str(%d) sz(%dK) vpg(%08lx)]\n",
-			       pbm->name, i, type_string,
-			       ((tag & PSYCHO_IOMMU_TAG_WRITE) ? 1 : 0),
-			       ((tag & PSYCHO_IOMMU_TAG_STREAM) ? 1 : 0),
-			       ((tag & PSYCHO_IOMMU_TAG_SIZE) ? 64 : 8),
-			       (tag & PSYCHO_IOMMU_TAG_VPAGE) << IOMMU_PAGE_SHIFT);
-			printk("%s: IOMMU DATA(%d)[valid(%d) cache(%d) ppg(%016lx)]\n",
-			       pbm->name, i,
-			       ((data & PSYCHO_IOMMU_DATA_VALID) ? 1 : 0),
-			       ((data & PSYCHO_IOMMU_DATA_CACHE) ? 1 : 0),
-			       (data & PSYCHO_IOMMU_DATA_PPAGE) << IOMMU_PAGE_SHIFT);
-		}
-	}
-	__psycho_check_stc_error(pbm, afsr, afar, type);
-	spin_unlock_irqrestore(&iommu->lock, flags);
-}
 
 /* Uncorrectable Errors.  Cause of the error and the address are
  * recorded in the UE_AFSR and UE_AFAR of PSYCHO.  They are errors
@@ -410,15 +131,14 @@ static void psycho_check_iommu_error(struct pci_pbm_info *pbm,
 static irqreturn_t psycho_ue_intr(int irq, void *dev_id)
 {
 	struct pci_pbm_info *pbm = dev_id;
-	struct pci_controller_info *p = pbm->parent;
 	unsigned long afsr_reg = pbm->controller_regs + PSYCHO_UE_AFSR;
 	unsigned long afar_reg = pbm->controller_regs + PSYCHO_UE_AFAR;
 	unsigned long afsr, afar, error_bits;
 	int reported;
 
 	/* Latch uncorrectable error status. */
-	afar = psycho_read(afar_reg);
-	afsr = psycho_read(afsr_reg);
+	afar = upa_readq(afar_reg);
+	afsr = upa_readq(afsr_reg);
 
 	/* Clear the primary/secondary error status bits. */
 	error_bits = afsr &
@@ -426,7 +146,7 @@ static irqreturn_t psycho_ue_intr(int irq, void *dev_id)
 		 PSYCHO_UEAFSR_SPIO | PSYCHO_UEAFSR_SDRD | PSYCHO_UEAFSR_SDWR);
 	if (!error_bits)
 		return IRQ_NONE;
-	psycho_write(afsr_reg, error_bits);
+	upa_writeq(error_bits, afsr_reg);
 
 	/* Log the error. */
 	printk("%s: Uncorrectable Error, primary error type[%s]\n",
@@ -463,8 +183,9 @@ static irqreturn_t psycho_ue_intr(int irq, void *dev_id)
 	printk("]\n");
 
 	/* Interrogate both IOMMUs for error status. */
-	psycho_check_iommu_error(&p->pbm_A, afsr, afar, UE_ERR);
-	psycho_check_iommu_error(&p->pbm_B, afsr, afar, UE_ERR);
+	psycho_check_iommu_error(pbm, afsr, afar, UE_ERR);
+	if (pbm->sibling)
+		psycho_check_iommu_error(pbm->sibling, afsr, afar, UE_ERR);
 
 	return IRQ_HANDLED;
 }
@@ -495,8 +216,8 @@ static irqreturn_t psycho_ce_intr(int irq, void *dev_id)
 	int reported;
 
 	/* Latch error status. */
-	afar = psycho_read(afar_reg);
-	afsr = psycho_read(afsr_reg);
+	afar = upa_readq(afar_reg);
+	afsr = upa_readq(afsr_reg);
 
 	/* Clear primary/secondary error status bits. */
 	error_bits = afsr &
@@ -504,7 +225,7 @@ static irqreturn_t psycho_ce_intr(int irq, void *dev_id)
 		 PSYCHO_CEAFSR_SPIO | PSYCHO_CEAFSR_SDRD | PSYCHO_CEAFSR_SDWR);
 	if (!error_bits)
 		return IRQ_NONE;
-	psycho_write(afsr_reg, error_bits);
+	upa_writeq(error_bits, afsr_reg);
 
 	/* Log the error. */
 	printk("%s: Correctable Error, primary error type[%s]\n",
@@ -554,162 +275,9 @@ static irqreturn_t psycho_ce_intr(int irq, void *dev_id)
  */
 #define PSYCHO_PCI_AFSR_A	0x2010UL
 #define PSYCHO_PCI_AFSR_B	0x4010UL
-#define  PSYCHO_PCIAFSR_PMA	0x8000000000000000UL /* Primary Master Abort Error   */
-#define  PSYCHO_PCIAFSR_PTA	0x4000000000000000UL /* Primary Target Abort Error   */
-#define  PSYCHO_PCIAFSR_PRTRY	0x2000000000000000UL /* Primary Excessive Retries    */
-#define  PSYCHO_PCIAFSR_PPERR	0x1000000000000000UL /* Primary Parity Error         */
-#define  PSYCHO_PCIAFSR_SMA	0x0800000000000000UL /* Secondary Master Abort Error */
-#define  PSYCHO_PCIAFSR_STA	0x0400000000000000UL /* Secondary Target Abort Error */
-#define  PSYCHO_PCIAFSR_SRTRY	0x0200000000000000UL /* Secondary Excessive Retries  */
-#define  PSYCHO_PCIAFSR_SPERR	0x0100000000000000UL /* Secondary Parity Error       */
-#define  PSYCHO_PCIAFSR_RESV1	0x00ff000000000000UL /* Reserved                     */
-#define  PSYCHO_PCIAFSR_BMSK	0x0000ffff00000000UL /* Bytemask of failed transfer  */
-#define  PSYCHO_PCIAFSR_BLK	0x0000000080000000UL /* Trans was block operation    */
-#define  PSYCHO_PCIAFSR_RESV2	0x0000000040000000UL /* Reserved                     */
-#define  PSYCHO_PCIAFSR_MID	0x000000003e000000UL /* MID causing the error        */
-#define  PSYCHO_PCIAFSR_RESV3	0x0000000001ffffffUL /* Reserved                     */
 #define PSYCHO_PCI_AFAR_A	0x2018UL
 #define PSYCHO_PCI_AFAR_B	0x4018UL
 
-static irqreturn_t psycho_pcierr_intr_other(struct pci_pbm_info *pbm, int is_pbm_a)
-{
-	unsigned long csr_reg, csr, csr_error_bits;
-	irqreturn_t ret = IRQ_NONE;
-	u16 stat;
-
-	if (is_pbm_a) {
-		csr_reg = pbm->controller_regs + PSYCHO_PCIA_CTRL;
-	} else {
-		csr_reg = pbm->controller_regs + PSYCHO_PCIB_CTRL;
-	}
-	csr = psycho_read(csr_reg);
-	csr_error_bits =
-		csr & (PSYCHO_PCICTRL_SBH_ERR | PSYCHO_PCICTRL_SERR);
-	if (csr_error_bits) {
-		/* Clear the errors.  */
-		psycho_write(csr_reg, csr);
-
-		/* Log 'em.  */
-		if (csr_error_bits & PSYCHO_PCICTRL_SBH_ERR)
-			printk("%s: PCI streaming byte hole error asserted.\n",
-			       pbm->name);
-		if (csr_error_bits & PSYCHO_PCICTRL_SERR)
-			printk("%s: PCI SERR signal asserted.\n", pbm->name);
-		ret = IRQ_HANDLED;
-	}
-	pci_read_config_word(pbm->pci_bus->self, PCI_STATUS, &stat);
-	if (stat & (PCI_STATUS_PARITY |
-		    PCI_STATUS_SIG_TARGET_ABORT |
-		    PCI_STATUS_REC_TARGET_ABORT |
-		    PCI_STATUS_REC_MASTER_ABORT |
-		    PCI_STATUS_SIG_SYSTEM_ERROR)) {
-		printk("%s: PCI bus error, PCI_STATUS[%04x]\n",
-		       pbm->name, stat);
-		pci_write_config_word(pbm->pci_bus->self, PCI_STATUS, 0xffff);
-		ret = IRQ_HANDLED;
-	}
-	return ret;
-}
-
-static irqreturn_t psycho_pcierr_intr(int irq, void *dev_id)
-{
-	struct pci_pbm_info *pbm = dev_id;
-	struct pci_controller_info *p = pbm->parent;
-	unsigned long afsr_reg, afar_reg;
-	unsigned long afsr, afar, error_bits;
-	int is_pbm_a, reported;
-
-	is_pbm_a = (pbm == &pbm->parent->pbm_A);
-	if (is_pbm_a) {
-		afsr_reg = p->pbm_A.controller_regs + PSYCHO_PCI_AFSR_A;
-		afar_reg = p->pbm_A.controller_regs + PSYCHO_PCI_AFAR_A;
-	} else {
-		afsr_reg = p->pbm_A.controller_regs + PSYCHO_PCI_AFSR_B;
-		afar_reg = p->pbm_A.controller_regs + PSYCHO_PCI_AFAR_B;
-	}
-
-	/* Latch error status. */
-	afar = psycho_read(afar_reg);
-	afsr = psycho_read(afsr_reg);
-
-	/* Clear primary/secondary error status bits. */
-	error_bits = afsr &
-		(PSYCHO_PCIAFSR_PMA | PSYCHO_PCIAFSR_PTA |
-		 PSYCHO_PCIAFSR_PRTRY | PSYCHO_PCIAFSR_PPERR |
-		 PSYCHO_PCIAFSR_SMA | PSYCHO_PCIAFSR_STA |
-		 PSYCHO_PCIAFSR_SRTRY | PSYCHO_PCIAFSR_SPERR);
-	if (!error_bits)
-		return psycho_pcierr_intr_other(pbm, is_pbm_a);
-	psycho_write(afsr_reg, error_bits);
-
-	/* Log the error. */
-	printk("%s: PCI Error, primary error type[%s]\n",
-	       pbm->name,
-	       (((error_bits & PSYCHO_PCIAFSR_PMA) ?
-		 "Master Abort" :
-		 ((error_bits & PSYCHO_PCIAFSR_PTA) ?
-		  "Target Abort" :
-		  ((error_bits & PSYCHO_PCIAFSR_PRTRY) ?
-		   "Excessive Retries" :
-		   ((error_bits & PSYCHO_PCIAFSR_PPERR) ?
-		    "Parity Error" : "???"))))));
-	printk("%s: bytemask[%04lx] UPA_MID[%02lx] was_block(%d)\n",
-	       pbm->name,
-	       (afsr & PSYCHO_PCIAFSR_BMSK) >> 32UL,
-	       (afsr & PSYCHO_PCIAFSR_MID) >> 25UL,
-	       (afsr & PSYCHO_PCIAFSR_BLK) ? 1 : 0);
-	printk("%s: PCI AFAR [%016lx]\n", pbm->name, afar);
-	printk("%s: PCI Secondary errors [", pbm->name);
-	reported = 0;
-	if (afsr & PSYCHO_PCIAFSR_SMA) {
-		reported++;
-		printk("(Master Abort)");
-	}
-	if (afsr & PSYCHO_PCIAFSR_STA) {
-		reported++;
-		printk("(Target Abort)");
-	}
-	if (afsr & PSYCHO_PCIAFSR_SRTRY) {
-		reported++;
-		printk("(Excessive Retries)");
-	}
-	if (afsr & PSYCHO_PCIAFSR_SPERR) {
-		reported++;
-		printk("(Parity Error)");
-	}
-	if (!reported)
-		printk("(none)");
-	printk("]\n");
-
-	/* For the error types shown, scan PBM's PCI bus for devices
-	 * which have logged that error type.
-	 */
-
-	/* If we see a Target Abort, this could be the result of an
-	 * IOMMU translation error of some sort.  It is extremely
-	 * useful to log this information as usually it indicates
-	 * a bug in the IOMMU support code or a PCI device driver.
-	 */
-	if (error_bits & (PSYCHO_PCIAFSR_PTA | PSYCHO_PCIAFSR_STA)) {
-		psycho_check_iommu_error(pbm, afsr, afar, PCI_ERR);
-		pci_scan_for_target_abort(pbm, pbm->pci_bus);
-	}
-	if (error_bits & (PSYCHO_PCIAFSR_PMA | PSYCHO_PCIAFSR_SMA))
-		pci_scan_for_master_abort(pbm, pbm->pci_bus);
-
-	/* For excessive retries, PSYCHO/PBM will abort the device
-	 * and there is no way to specifically check for excessive
-	 * retries in the config space status registers.  So what
-	 * we hope is that we'll catch it via the master/target
-	 * abort events.
-	 */
-
-	if (error_bits & (PSYCHO_PCIAFSR_PPERR | PSYCHO_PCIAFSR_SPERR))
-		pci_scan_for_parity_error(pbm, pbm->pci_bus);
-
-	return IRQ_HANDLED;
-}
-
 /* XXX What about PowerFail/PowerManagement??? -DaveM */
 #define PSYCHO_ECC_CTRL		0x0020
 #define  PSYCHO_ECCCTRL_EE	 0x8000000000000000UL /* Enable ECC Checking */
@@ -717,7 +285,7 @@ static irqreturn_t psycho_pcierr_intr(int irq, void *dev_id)
 #define  PSYCHO_ECCCTRL_CE	 0x2000000000000000UL /* Enable CE INterrupts */
 static void psycho_register_error_handlers(struct pci_pbm_info *pbm)
 {
-	struct of_device *op = of_find_device_by_node(pbm->prom_node);
+	struct of_device *op = of_find_device_by_node(pbm->op->node);
 	unsigned long base = pbm->controller_regs;
 	u64 tmp;
 	int err;
@@ -744,43 +312,42 @@ static void psycho_register_error_handlers(struct pci_pbm_info *pbm)
 	 * the second will just error out since we do not pass in
 	 * IRQF_SHARED.
 	 */
-	err = request_irq(op->irqs[1], psycho_ue_intr, 0,
+	err = request_irq(op->irqs[1], psycho_ue_intr, IRQF_SHARED,
 			  "PSYCHO_UE", pbm);
-	err = request_irq(op->irqs[2], psycho_ce_intr, 0,
+	err = request_irq(op->irqs[2], psycho_ce_intr, IRQF_SHARED,
 			  "PSYCHO_CE", pbm);
 
 	/* This one, however, ought not to fail.  We can just warn
 	 * about it since the system can still operate properly even
 	 * if this fails.
 	 */
-	err = request_irq(op->irqs[0], psycho_pcierr_intr, 0,
+	err = request_irq(op->irqs[0], psycho_pcierr_intr, IRQF_SHARED,
 			  "PSYCHO_PCIERR", pbm);
 	if (err)
 		printk(KERN_WARNING "%s: Could not register PCIERR, "
 		       "err=%d\n", pbm->name, err);
 
 	/* Enable UE and CE interrupts for controller. */
-	psycho_write(base + PSYCHO_ECC_CTRL,
-		     (PSYCHO_ECCCTRL_EE |
-		      PSYCHO_ECCCTRL_UE |
-		      PSYCHO_ECCCTRL_CE));
+	upa_writeq((PSYCHO_ECCCTRL_EE |
+		    PSYCHO_ECCCTRL_UE |
+		    PSYCHO_ECCCTRL_CE), base + PSYCHO_ECC_CTRL);
 
 	/* Enable PCI Error interrupts and clear error
 	 * bits for each PBM.
 	 */
-	tmp = psycho_read(base + PSYCHO_PCIA_CTRL);
+	tmp = upa_readq(base + PSYCHO_PCIA_CTRL);
 	tmp |= (PSYCHO_PCICTRL_SERR |
 		PSYCHO_PCICTRL_SBH_ERR |
 		PSYCHO_PCICTRL_EEN);
 	tmp &= ~(PSYCHO_PCICTRL_SBH_INT);
-	psycho_write(base + PSYCHO_PCIA_CTRL, tmp);
+	upa_writeq(tmp, base + PSYCHO_PCIA_CTRL);
 		     
-	tmp = psycho_read(base + PSYCHO_PCIB_CTRL);
+	tmp = upa_readq(base + PSYCHO_PCIB_CTRL);
 	tmp |= (PSYCHO_PCICTRL_SERR |
 		PSYCHO_PCICTRL_SBH_ERR |
 		PSYCHO_PCICTRL_EEN);
 	tmp &= ~(PSYCHO_PCICTRL_SBH_INT);
-	psycho_write(base + PSYCHO_PCIB_CTRL, tmp);
+	upa_writeq(tmp, base + PSYCHO_PCIB_CTRL);
 }
 
 /* PSYCHO boot time probing and initialization. */
@@ -801,11 +368,12 @@ static void pbm_config_busmastering(struct pci_pbm_info *pbm)
 	pci_config_write8(addr, 64);
 }
 
-static void __init psycho_scan_bus(struct pci_pbm_info *pbm)
+static void __init psycho_scan_bus(struct pci_pbm_info *pbm,
+				   struct device *parent)
 {
 	pbm_config_busmastering(pbm);
 	pbm->is_66mhz_capable = 0;
-	pbm->pci_bus = pci_scan_one_pbm(pbm);
+	pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
 
 	/* After the PCI bus scan is complete, we can register
 	 * the error interrupt handlers.
@@ -813,61 +381,6 @@ static void __init psycho_scan_bus(struct pci_pbm_info *pbm)
 	psycho_register_error_handlers(pbm);
 }
 
-static int psycho_iommu_init(struct pci_pbm_info *pbm)
-{
-	struct iommu *iommu = pbm->iommu;
-	unsigned long i;
-	u64 control;
-	int err;
-
-	/* Register addresses. */
-	iommu->iommu_control  = pbm->controller_regs + PSYCHO_IOMMU_CONTROL;
-	iommu->iommu_tsbbase  = pbm->controller_regs + PSYCHO_IOMMU_TSBBASE;
-	iommu->iommu_flush    = pbm->controller_regs + PSYCHO_IOMMU_FLUSH;
-	iommu->iommu_tags     = iommu->iommu_flush + (0xa580UL - 0x0210UL);
-
-	/* PSYCHO's IOMMU lacks ctx flushing. */
-	iommu->iommu_ctxflush = 0;
-
-	/* We use the main control register of PSYCHO as the write
-	 * completion register.
-	 */
-	iommu->write_complete_reg = pbm->controller_regs + PSYCHO_CONTROL;
-
-	/*
-	 * Invalidate TLB Entries.
-	 */
-	control = psycho_read(pbm->controller_regs + PSYCHO_IOMMU_CONTROL);
-	control |= PSYCHO_IOMMU_CTRL_DENAB;
-	psycho_write(pbm->controller_regs + PSYCHO_IOMMU_CONTROL, control);
-	for(i = 0; i < 16; i++) {
-		psycho_write(pbm->controller_regs + PSYCHO_IOMMU_TAG + (i * 8UL), 0);
-		psycho_write(pbm->controller_regs + PSYCHO_IOMMU_DATA + (i * 8UL), 0);
-	}
-
-	/* Leave diag mode enabled for full-flushing done
-	 * in pci_iommu.c
-	 */
-	err = iommu_table_init(iommu, IO_TSB_SIZE, 0xc0000000, 0xffffffff,
-			       pbm->numa_node);
-	if (err)
-		return err;
-
-	psycho_write(pbm->controller_regs + PSYCHO_IOMMU_TSBBASE,
-		     __pa(iommu->page_table));
-
-	control = psycho_read(pbm->controller_regs + PSYCHO_IOMMU_CONTROL);
-	control &= ~(PSYCHO_IOMMU_CTRL_TSBSZ | PSYCHO_IOMMU_CTRL_TBWSZ);
-	control |= (PSYCHO_IOMMU_TSBSZ_128K | PSYCHO_IOMMU_CTRL_ENAB);
-	psycho_write(pbm->controller_regs + PSYCHO_IOMMU_CONTROL, control);
-
-	/* If necessary, hook us up for starfire IRQ translations. */
-	if (this_is_starfire)
-		starfire_hookup(pbm->portid);
-
-	return 0;
-}
-
 #define PSYCHO_IRQ_RETRY	0x1a00UL
 #define PSYCHO_PCIA_DIAG	0x2020UL
 #define PSYCHO_PCIB_DIAG	0x4020UL
@@ -884,28 +397,28 @@ static void psycho_controller_hwinit(struct pci_pbm_info *pbm)
 {
 	u64 tmp;
 
-	psycho_write(pbm->controller_regs + PSYCHO_IRQ_RETRY, 5);
+	upa_writeq(5, pbm->controller_regs + PSYCHO_IRQ_RETRY);
 
 	/* Enable arbiter for all PCI slots. */
-	tmp = psycho_read(pbm->controller_regs + PSYCHO_PCIA_CTRL);
+	tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIA_CTRL);
 	tmp |= PSYCHO_PCICTRL_AEN;
-	psycho_write(pbm->controller_regs + PSYCHO_PCIA_CTRL, tmp);
+	upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIA_CTRL);
 
-	tmp = psycho_read(pbm->controller_regs + PSYCHO_PCIB_CTRL);
+	tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIB_CTRL);
 	tmp |= PSYCHO_PCICTRL_AEN;
-	psycho_write(pbm->controller_regs + PSYCHO_PCIB_CTRL, tmp);
+	upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIB_CTRL);
 
 	/* Disable DMA write / PIO read synchronization on
 	 * both PCI bus segments.
 	 * [ U2P Erratum 1243770, STP2223BGA data sheet ]
 	 */
-	tmp = psycho_read(pbm->controller_regs + PSYCHO_PCIA_DIAG);
+	tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIA_DIAG);
 	tmp |= PSYCHO_PCIDIAG_DDWSYNC;
-	psycho_write(pbm->controller_regs + PSYCHO_PCIA_DIAG, tmp);
+	upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIA_DIAG);
 
-	tmp = psycho_read(pbm->controller_regs + PSYCHO_PCIB_DIAG);
+	tmp = upa_readq(pbm->controller_regs + PSYCHO_PCIB_DIAG);
 	tmp |= PSYCHO_PCIDIAG_DDWSYNC;
-	psycho_write(pbm->controller_regs + PSYCHO_PCIB_DIAG, tmp);
+	upa_writeq(tmp, pbm->controller_regs + PSYCHO_PCIB_DIAG);
 }
 
 static void psycho_pbm_strbuf_init(struct pci_pbm_info *pbm,
@@ -918,10 +431,16 @@ static void psycho_pbm_strbuf_init(struct pci_pbm_info *pbm,
 		pbm->stc.strbuf_control  = base + PSYCHO_STRBUF_CONTROL_A;
 		pbm->stc.strbuf_pflush   = base + PSYCHO_STRBUF_FLUSH_A;
 		pbm->stc.strbuf_fsync    = base + PSYCHO_STRBUF_FSYNC_A;
+		pbm->stc.strbuf_err_stat = base + PSYCHO_STC_ERR_A;
+		pbm->stc.strbuf_tag_diag = base + PSYCHO_STC_TAG_A;
+		pbm->stc.strbuf_line_diag= base + PSYCHO_STC_LINE_A;
 	} else {
 		pbm->stc.strbuf_control  = base + PSYCHO_STRBUF_CONTROL_B;
 		pbm->stc.strbuf_pflush   = base + PSYCHO_STRBUF_FLUSH_B;
 		pbm->stc.strbuf_fsync    = base + PSYCHO_STRBUF_FSYNC_B;
+		pbm->stc.strbuf_err_stat = base + PSYCHO_STC_ERR_B;
+		pbm->stc.strbuf_tag_diag = base + PSYCHO_STC_TAG_B;
+		pbm->stc.strbuf_line_diag= base + PSYCHO_STC_LINE_B;
 	}
 	/* PSYCHO's streaming buffer lacks ctx flushing. */
 	pbm->stc.strbuf_ctxflush      = 0;
@@ -944,7 +463,7 @@ static void psycho_pbm_strbuf_init(struct pci_pbm_info *pbm,
 	 */
 #undef PSYCHO_STRBUF_RERUN_ENABLE
 #undef PSYCHO_STRBUF_RERUN_DISABLE
-	control = psycho_read(pbm->stc.strbuf_control);
+	control = upa_readq(pbm->stc.strbuf_control);
 	control |= PSYCHO_STRBUF_CTRL_ENAB;
 	control &= ~(PSYCHO_STRBUF_CTRL_LENAB | PSYCHO_STRBUF_CTRL_LPTR);
 #ifdef PSYCHO_STRBUF_RERUN_ENABLE
@@ -954,7 +473,7 @@ static void psycho_pbm_strbuf_init(struct pci_pbm_info *pbm,
 	control |= PSYCHO_STRBUF_CTRL_RRDIS;
 #endif
 #endif
-	psycho_write(pbm->stc.strbuf_control, control);
+	upa_writeq(control, pbm->stc.strbuf_control);
 
 	pbm->stc.strbuf_enabled = 1;
 }
@@ -966,111 +485,134 @@ static void psycho_pbm_strbuf_init(struct pci_pbm_info *pbm,
 #define PSYCHO_MEMSPACE_B	0x180000000UL
 #define PSYCHO_MEMSPACE_SIZE	0x07fffffffUL
 
-static void __init psycho_pbm_init(struct pci_controller_info *p,
-			    struct device_node *dp, int is_pbm_a)
+static void __init psycho_pbm_init(struct pci_pbm_info *pbm,
+				   struct of_device *op, int is_pbm_a)
 {
-	struct property *prop;
-	struct pci_pbm_info *pbm;
-
-	if (is_pbm_a)
-		pbm = &p->pbm_A;
-	else
-		pbm = &p->pbm_B;
-
-	pbm->next = pci_pbm_root;
-	pci_pbm_root = pbm;
-
-	pbm->numa_node = -1;
-
-	pbm->scan_bus = psycho_scan_bus;
-	pbm->pci_ops = &sun4u_pci_ops;
-	pbm->config_space_reg_bits = 8;
-
-	pbm->index = pci_num_pbms++;
-
-	pbm->chip_type = PBM_CHIP_TYPE_PSYCHO;
-	pbm->chip_version = 0;
-	prop = of_find_property(dp, "version#", NULL);
-	if (prop)
-		pbm->chip_version = *(int *) prop->value;
-	pbm->chip_revision = 0;
-	prop = of_find_property(dp, "module-revision#", NULL);
-	if (prop)
-		pbm->chip_revision = *(int *) prop->value;
-
-	pbm->parent = p;
-	pbm->prom_node = dp;
-	pbm->name = dp->full_name;
-
-	printk("%s: PSYCHO PCI Bus Module ver[%x:%x]\n",
-	       pbm->name,
-	       pbm->chip_version, pbm->chip_revision);
-
-	pci_determine_mem_io_space(pbm);
+	psycho_pbm_init_common(pbm, op, "PSYCHO", PBM_CHIP_TYPE_PSYCHO);
+	psycho_pbm_strbuf_init(pbm, is_pbm_a);
+	psycho_scan_bus(pbm, &op->dev);
+}
 
-	pci_get_pbm_props(pbm);
+static struct pci_pbm_info * __devinit psycho_find_sibling(u32 upa_portid)
+{
+	struct pci_pbm_info *pbm;
 
-	psycho_pbm_strbuf_init(pbm, is_pbm_a);
+	for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
+		if (pbm->portid == upa_portid)
+			return pbm;
+	}
+	return NULL;
 }
 
 #define PSYCHO_CONFIGSPACE	0x001000000UL
 
-void __init psycho_init(struct device_node *dp, char *model_name)
+static int __devinit psycho_probe(struct of_device *op,
+				  const struct of_device_id *match)
 {
-	struct linux_prom64_registers *pr_regs;
-	struct pci_controller_info *p;
+	const struct linux_prom64_registers *pr_regs;
+	struct device_node *dp = op->node;
 	struct pci_pbm_info *pbm;
 	struct iommu *iommu;
-	struct property *prop;
+	int is_pbm_a, err;
 	u32 upa_portid;
-	int is_pbm_a;
 
-	upa_portid = 0xff;
-	prop = of_find_property(dp, "upa-portid", NULL);
-	if (prop)
-		upa_portid = *(u32 *) prop->value;
+	upa_portid = of_getintprop_default(dp, "upa-portid", 0xff);
 
-	for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
-		struct pci_controller_info *p = pbm->parent;
+	err = -ENOMEM;
+	pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+	if (!pbm) {
+		printk(KERN_ERR PFX "Cannot allocate pci_pbm_info.\n");
+		goto out_err;
+	}
 
-		if (p->pbm_A.portid == upa_portid) {
-			is_pbm_a = (p->pbm_A.prom_node == NULL);
-			psycho_pbm_init(p, dp, is_pbm_a);
-			return;
+	pbm->sibling = psycho_find_sibling(upa_portid);
+	if (pbm->sibling) {
+		iommu = pbm->sibling->iommu;
+	} else {
+		iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+		if (!iommu) {
+			printk(KERN_ERR PFX "Cannot allocate PBM iommu.\n");
+			goto out_free_controller;
 		}
 	}
 
-	p = kzalloc(sizeof(struct pci_controller_info), GFP_ATOMIC);
-	if (!p)
-		goto fatal_memory_error;
-	iommu = kzalloc(sizeof(struct iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
+	pbm->iommu = iommu;
+	pbm->portid = upa_portid;
 
-	p->pbm_A.iommu = p->pbm_B.iommu = iommu;
+	pr_regs = of_get_property(dp, "reg", NULL);
+	err = -ENODEV;
+	if (!pr_regs) {
+		printk(KERN_ERR PFX "No reg property.\n");
+		goto out_free_iommu;
+	}
 
-	p->pbm_A.portid = upa_portid;
-	p->pbm_B.portid = upa_portid;
+	is_pbm_a = ((pr_regs[0].phys_addr & 0x6000) == 0x2000);
 
-	prop = of_find_property(dp, "reg", NULL);
-	pr_regs = prop->value;
+	pbm->controller_regs = pr_regs[2].phys_addr;
+	pbm->config_space = (pr_regs[2].phys_addr + PSYCHO_CONFIGSPACE);
 
-	p->pbm_A.controller_regs = pr_regs[2].phys_addr;
-	p->pbm_B.controller_regs = pr_regs[2].phys_addr;
+	if (is_pbm_a) {
+		pbm->pci_afsr = pbm->controller_regs + PSYCHO_PCI_AFSR_A;
+		pbm->pci_afar = pbm->controller_regs + PSYCHO_PCI_AFAR_A;
+		pbm->pci_csr  = pbm->controller_regs + PSYCHO_PCIA_CTRL;
+	} else {
+		pbm->pci_afsr = pbm->controller_regs + PSYCHO_PCI_AFSR_B;
+		pbm->pci_afar = pbm->controller_regs + PSYCHO_PCI_AFAR_B;
+		pbm->pci_csr  = pbm->controller_regs + PSYCHO_PCIB_CTRL;
+	}
 
-	p->pbm_A.config_space = p->pbm_B.config_space =
-		(pr_regs[2].phys_addr + PSYCHO_CONFIGSPACE);
+	psycho_controller_hwinit(pbm);
+	if (!pbm->sibling) {
+		err = psycho_iommu_init(pbm, 128, 0xc0000000,
+					0xffffffff, PSYCHO_CONTROL);
+		if (err)
+			goto out_free_iommu;
 
-	psycho_controller_hwinit(&p->pbm_A);
+		/* If necessary, hook us up for starfire IRQ translations. */
+		if (this_is_starfire)
+			starfire_hookup(pbm->portid);
+	}
 
-	if (psycho_iommu_init(&p->pbm_A))
-		goto fatal_memory_error;
+	psycho_pbm_init(pbm, op, is_pbm_a);
 
-	is_pbm_a = ((pr_regs[0].phys_addr & 0x6000) == 0x2000);
-	psycho_pbm_init(p, dp, is_pbm_a);
-	return;
+	pbm->next = pci_pbm_root;
+	pci_pbm_root = pbm;
+
+	if (pbm->sibling)
+		pbm->sibling->sibling = pbm;
+
+	dev_set_drvdata(&op->dev, pbm);
+
+	return 0;
+
+out_free_iommu:
+	if (!pbm->sibling)
+		kfree(pbm->iommu);
 
-fatal_memory_error:
-	prom_printf("PSYCHO: Fatal memory allocation error.\n");
-	prom_halt();
+out_free_controller:
+	kfree(pbm);
+
+out_err:
+	return err;
 }
+
+static struct of_device_id __initdata psycho_match[] = {
+	{
+		.name = "pci",
+		.compatible = "pci108e,8000",
+	},
+	{},
+};
+
+static struct of_platform_driver psycho_driver = {
+	.name		= DRIVER_NAME,
+	.match_table	= psycho_match,
+	.probe		= psycho_probe,
+};
+
+static int __init psycho_init(void)
+{
+	return of_register_driver(&psycho_driver, &of_bus_type);
+}
+
+subsys_initcall(psycho_init);
diff --git a/arch/sparc64/kernel/pci_sabre.c b/arch/sparc64/kernel/pci_sabre.c
index 4c34195baf3..713257b6963 100644
--- a/arch/sparc64/kernel/pci_sabre.c
+++ b/arch/sparc64/kernel/pci_sabre.c
@@ -11,36 +11,20 @@
 #include <linux/init.h>
 #include <linux/slab.h>
 #include <linux/interrupt.h>
+#include <linux/of_device.h>
 
 #include <asm/apb.h>
 #include <asm/iommu.h>
 #include <asm/irq.h>
-#include <asm/smp.h>
-#include <asm/oplib.h>
 #include <asm/prom.h>
-#include <asm/of_device.h>
+#include <asm/upa.h>
 
 #include "pci_impl.h"
 #include "iommu_common.h"
+#include "psycho_common.h"
 
-/* All SABRE registers are 64-bits.  The following accessor
- * routines are how they are accessed.  The REG parameter
- * is a physical address.
- */
-#define sabre_read(__reg) \
-({	u64 __ret; \
-	__asm__ __volatile__("ldxa [%1] %2, %0" \
-			     : "=r" (__ret) \
-			     : "r" (__reg), "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory"); \
-	__ret; \
-})
-#define sabre_write(__reg, __val) \
-	__asm__ __volatile__("stxa %0, [%1] %2" \
-			     : /* no outputs */ \
-			     : "r" (__val), "r" (__reg), \
-			       "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory")
+#define DRIVER_NAME	"sabre"
+#define PFX		DRIVER_NAME ": "
 
 /* SABRE PCI controller register offsets and definitions. */
 #define SABRE_UE_AFSR		0x0030UL
@@ -208,95 +192,6 @@
 static int hummingbird_p;
 static struct pci_bus *sabre_root_bus;
 
-/* SABRE error handling support. */
-static void sabre_check_iommu_error(struct pci_pbm_info *pbm,
-				    unsigned long afsr,
-				    unsigned long afar)
-{
-	struct iommu *iommu = pbm->iommu;
-	unsigned long iommu_tag[16];
-	unsigned long iommu_data[16];
-	unsigned long flags;
-	u64 control;
-	int i;
-
-	spin_lock_irqsave(&iommu->lock, flags);
-	control = sabre_read(iommu->iommu_control);
-	if (control & SABRE_IOMMUCTRL_ERR) {
-		char *type_string;
-
-		/* Clear the error encountered bit.
-		 * NOTE: On Sabre this is write 1 to clear,
-		 *       which is different from Psycho.
-		 */
-		sabre_write(iommu->iommu_control, control);
-		switch((control & SABRE_IOMMUCTRL_ERRSTS) >> 25UL) {
-		case 1:
-			type_string = "Invalid Error";
-			break;
-		case 3:
-			type_string = "ECC Error";
-			break;
-		default:
-			type_string = "Unknown";
-			break;
-		};
-		printk("%s: IOMMU Error, type[%s]\n",
-		       pbm->name, type_string);
-
-		/* Enter diagnostic mode and probe for error'd
-		 * entries in the IOTLB.
-		 */
-		control &= ~(SABRE_IOMMUCTRL_ERRSTS | SABRE_IOMMUCTRL_ERR);
-		sabre_write(iommu->iommu_control,
-			    (control | SABRE_IOMMUCTRL_DENAB));
-		for (i = 0; i < 16; i++) {
-			unsigned long base = pbm->controller_regs;
-
-			iommu_tag[i] =
-				sabre_read(base + SABRE_IOMMU_TAG + (i * 8UL));
-			iommu_data[i] =
-				sabre_read(base + SABRE_IOMMU_DATA + (i * 8UL));
-			sabre_write(base + SABRE_IOMMU_TAG + (i * 8UL), 0);
-			sabre_write(base + SABRE_IOMMU_DATA + (i * 8UL), 0);
-		}
-		sabre_write(iommu->iommu_control, control);
-
-		for (i = 0; i < 16; i++) {
-			unsigned long tag, data;
-
-			tag = iommu_tag[i];
-			if (!(tag & SABRE_IOMMUTAG_ERR))
-				continue;
-
-			data = iommu_data[i];
-			switch((tag & SABRE_IOMMUTAG_ERRSTS) >> 23UL) {
-			case 1:
-				type_string = "Invalid Error";
-				break;
-			case 3:
-				type_string = "ECC Error";
-				break;
-			default:
-				type_string = "Unknown";
-				break;
-			};
-			printk("%s: IOMMU TAG(%d)[RAW(%016lx)error(%s)wr(%d)sz(%dK)vpg(%08lx)]\n",
-			       pbm->name, i, tag, type_string,
-			       ((tag & SABRE_IOMMUTAG_WRITE) ? 1 : 0),
-			       ((tag & SABRE_IOMMUTAG_SIZE) ? 64 : 8),
-			       ((tag & SABRE_IOMMUTAG_VPN) << IOMMU_PAGE_SHIFT));
-			printk("%s: IOMMU DATA(%d)[RAW(%016lx)valid(%d)used(%d)cache(%d)ppg(%016lx)\n",
-			       pbm->name, i, data,
-			       ((data & SABRE_IOMMUDATA_VALID) ? 1 : 0),
-			       ((data & SABRE_IOMMUDATA_USED) ? 1 : 0),
-			       ((data & SABRE_IOMMUDATA_CACHE) ? 1 : 0),
-			       ((data & SABRE_IOMMUDATA_PPN) << IOMMU_PAGE_SHIFT));
-		}
-	}
-	spin_unlock_irqrestore(&iommu->lock, flags);
-}
-
 static irqreturn_t sabre_ue_intr(int irq, void *dev_id)
 {
 	struct pci_pbm_info *pbm = dev_id;
@@ -306,8 +201,8 @@ static irqreturn_t sabre_ue_intr(int irq, void *dev_id)
 	int reported;
 
 	/* Latch uncorrectable error status. */
-	afar = sabre_read(afar_reg);
-	afsr = sabre_read(afsr_reg);
+	afar = upa_readq(afar_reg);
+	afsr = upa_readq(afsr_reg);
 
 	/* Clear the primary/secondary error status bits. */
 	error_bits = afsr &
@@ -316,7 +211,7 @@ static irqreturn_t sabre_ue_intr(int irq, void *dev_id)
 		 SABRE_UEAFSR_SDTE | SABRE_UEAFSR_PDTE);
 	if (!error_bits)
 		return IRQ_NONE;
-	sabre_write(afsr_reg, error_bits);
+	upa_writeq(error_bits, afsr_reg);
 
 	/* Log the error. */
 	printk("%s: Uncorrectable Error, primary error type[%s%s]\n",
@@ -352,7 +247,7 @@ static irqreturn_t sabre_ue_intr(int irq, void *dev_id)
 	printk("]\n");
 
 	/* Interrogate IOMMU for error status. */
-	sabre_check_iommu_error(pbm, afsr, afar);
+	psycho_check_iommu_error(pbm, afsr, afar, UE_ERR);
 
 	return IRQ_HANDLED;
 }
@@ -366,8 +261,8 @@ static irqreturn_t sabre_ce_intr(int irq, void *dev_id)
 	int reported;
 
 	/* Latch error status. */
-	afar = sabre_read(afar_reg);
-	afsr = sabre_read(afsr_reg);
+	afar = upa_readq(afar_reg);
+	afsr = upa_readq(afsr_reg);
 
 	/* Clear primary/secondary error status bits. */
 	error_bits = afsr &
@@ -375,7 +270,7 @@ static irqreturn_t sabre_ce_intr(int irq, void *dev_id)
 		 SABRE_CEAFSR_SDRD | SABRE_CEAFSR_SDWR);
 	if (!error_bits)
 		return IRQ_NONE;
-	sabre_write(afsr_reg, error_bits);
+	upa_writeq(error_bits, afsr_reg);
 
 	/* Log the error. */
 	printk("%s: Correctable Error, primary error type[%s]\n",
@@ -413,136 +308,9 @@ static irqreturn_t sabre_ce_intr(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static irqreturn_t sabre_pcierr_intr_other(struct pci_pbm_info *pbm)
-{
-	unsigned long csr_reg, csr, csr_error_bits;
-	irqreturn_t ret = IRQ_NONE;
-	u16 stat;
-
-	csr_reg = pbm->controller_regs + SABRE_PCICTRL;
-	csr = sabre_read(csr_reg);
-	csr_error_bits =
-		csr & SABRE_PCICTRL_SERR;
-	if (csr_error_bits) {
-		/* Clear the errors.  */
-		sabre_write(csr_reg, csr);
-
-		/* Log 'em.  */
-		if (csr_error_bits & SABRE_PCICTRL_SERR)
-			printk("%s: PCI SERR signal asserted.\n",
-			       pbm->name);
-		ret = IRQ_HANDLED;
-	}
-	pci_bus_read_config_word(sabre_root_bus, 0,
-				 PCI_STATUS, &stat);
-	if (stat & (PCI_STATUS_PARITY |
-		    PCI_STATUS_SIG_TARGET_ABORT |
-		    PCI_STATUS_REC_TARGET_ABORT |
-		    PCI_STATUS_REC_MASTER_ABORT |
-		    PCI_STATUS_SIG_SYSTEM_ERROR)) {
-		printk("%s: PCI bus error, PCI_STATUS[%04x]\n",
-		       pbm->name, stat);
-		pci_bus_write_config_word(sabre_root_bus, 0,
-					  PCI_STATUS, 0xffff);
-		ret = IRQ_HANDLED;
-	}
-	return ret;
-}
-
-static irqreturn_t sabre_pcierr_intr(int irq, void *dev_id)
-{
-	struct pci_pbm_info *pbm = dev_id;
-	unsigned long afsr_reg, afar_reg;
-	unsigned long afsr, afar, error_bits;
-	int reported;
-
-	afsr_reg = pbm->controller_regs + SABRE_PIOAFSR;
-	afar_reg = pbm->controller_regs + SABRE_PIOAFAR;
-
-	/* Latch error status. */
-	afar = sabre_read(afar_reg);
-	afsr = sabre_read(afsr_reg);
-
-	/* Clear primary/secondary error status bits. */
-	error_bits = afsr &
-		(SABRE_PIOAFSR_PMA | SABRE_PIOAFSR_PTA |
-		 SABRE_PIOAFSR_PRTRY | SABRE_PIOAFSR_PPERR |
-		 SABRE_PIOAFSR_SMA | SABRE_PIOAFSR_STA |
-		 SABRE_PIOAFSR_SRTRY | SABRE_PIOAFSR_SPERR);
-	if (!error_bits)
-		return sabre_pcierr_intr_other(pbm);
-	sabre_write(afsr_reg, error_bits);
-
-	/* Log the error. */
-	printk("%s: PCI Error, primary error type[%s]\n",
-	       pbm->name,
-	       (((error_bits & SABRE_PIOAFSR_PMA) ?
-		 "Master Abort" :
-		 ((error_bits & SABRE_PIOAFSR_PTA) ?
-		  "Target Abort" :
-		  ((error_bits & SABRE_PIOAFSR_PRTRY) ?
-		   "Excessive Retries" :
-		   ((error_bits & SABRE_PIOAFSR_PPERR) ?
-		    "Parity Error" : "???"))))));
-	printk("%s: bytemask[%04lx] was_block(%d)\n",
-	       pbm->name,
-	       (afsr & SABRE_PIOAFSR_BMSK) >> 32UL,
-	       (afsr & SABRE_PIOAFSR_BLK) ? 1 : 0);
-	printk("%s: PCI AFAR [%016lx]\n", pbm->name, afar);
-	printk("%s: PCI Secondary errors [", pbm->name);
-	reported = 0;
-	if (afsr & SABRE_PIOAFSR_SMA) {
-		reported++;
-		printk("(Master Abort)");
-	}
-	if (afsr & SABRE_PIOAFSR_STA) {
-		reported++;
-		printk("(Target Abort)");
-	}
-	if (afsr & SABRE_PIOAFSR_SRTRY) {
-		reported++;
-		printk("(Excessive Retries)");
-	}
-	if (afsr & SABRE_PIOAFSR_SPERR) {
-		reported++;
-		printk("(Parity Error)");
-	}
-	if (!reported)
-		printk("(none)");
-	printk("]\n");
-
-	/* For the error types shown, scan both PCI buses for devices
-	 * which have logged that error type.
-	 */
-
-	/* If we see a Target Abort, this could be the result of an
-	 * IOMMU translation error of some sort.  It is extremely
-	 * useful to log this information as usually it indicates
-	 * a bug in the IOMMU support code or a PCI device driver.
-	 */
-	if (error_bits & (SABRE_PIOAFSR_PTA | SABRE_PIOAFSR_STA)) {
-		sabre_check_iommu_error(pbm, afsr, afar);
-		pci_scan_for_target_abort(pbm, pbm->pci_bus);
-	}
-	if (error_bits & (SABRE_PIOAFSR_PMA | SABRE_PIOAFSR_SMA))
-		pci_scan_for_master_abort(pbm, pbm->pci_bus);
-
-	/* For excessive retries, SABRE/PBM will abort the device
-	 * and there is no way to specifically check for excessive
-	 * retries in the config space status registers.  So what
-	 * we hope is that we'll catch it via the master/target
-	 * abort events.
-	 */
-
-	if (error_bits & (SABRE_PIOAFSR_PPERR | SABRE_PIOAFSR_SPERR))
-		pci_scan_for_parity_error(pbm, pbm->pci_bus);
-
-	return IRQ_HANDLED;
-}
-
 static void sabre_register_error_handlers(struct pci_pbm_info *pbm)
 {
-	struct device_node *dp = pbm->prom_node;
+	struct device_node *dp = pbm->op->node;
 	struct of_device *op;
 	unsigned long base = pbm->controller_regs;
 	u64 tmp;
@@ -568,33 +336,34 @@ static void sabre_register_error_handlers(struct pci_pbm_info *pbm)
 	 * registering the handler so that we don't get spurious
 	 * interrupts.
 	 */
-	sabre_write(base + SABRE_UE_AFSR,
-		    (SABRE_UEAFSR_PDRD | SABRE_UEAFSR_PDWR |
-		     SABRE_UEAFSR_SDRD | SABRE_UEAFSR_SDWR |
-		     SABRE_UEAFSR_SDTE | SABRE_UEAFSR_PDTE));
+	upa_writeq((SABRE_UEAFSR_PDRD | SABRE_UEAFSR_PDWR |
+		    SABRE_UEAFSR_SDRD | SABRE_UEAFSR_SDWR |
+		    SABRE_UEAFSR_SDTE | SABRE_UEAFSR_PDTE),
+		   base + SABRE_UE_AFSR);
 
 	err = request_irq(op->irqs[1], sabre_ue_intr, 0, "SABRE_UE", pbm);
 	if (err)
 		printk(KERN_WARNING "%s: Couldn't register UE, err=%d.\n",
 		       pbm->name, err);
 
-	sabre_write(base + SABRE_CE_AFSR,
-		    (SABRE_CEAFSR_PDRD | SABRE_CEAFSR_PDWR |
-		     SABRE_CEAFSR_SDRD | SABRE_CEAFSR_SDWR));
+	upa_writeq((SABRE_CEAFSR_PDRD | SABRE_CEAFSR_PDWR |
+		    SABRE_CEAFSR_SDRD | SABRE_CEAFSR_SDWR),
+		   base + SABRE_CE_AFSR);
+
 
 	err = request_irq(op->irqs[2], sabre_ce_intr, 0, "SABRE_CE", pbm);
 	if (err)
 		printk(KERN_WARNING "%s: Couldn't register CE, err=%d.\n",
 		       pbm->name, err);
-	err = request_irq(op->irqs[0], sabre_pcierr_intr, 0,
+	err = request_irq(op->irqs[0], psycho_pcierr_intr, 0,
 			  "SABRE_PCIERR", pbm);
 	if (err)
 		printk(KERN_WARNING "%s: Couldn't register PCIERR, err=%d.\n",
 		       pbm->name, err);
 
-	tmp = sabre_read(base + SABRE_PCICTRL);
+	tmp = upa_readq(base + SABRE_PCICTRL);
 	tmp |= SABRE_PCICTRL_ERREN;
-	sabre_write(base + SABRE_PCICTRL, tmp);
+	upa_writeq(tmp, base + SABRE_PCICTRL);
 }
 
 static void apb_init(struct pci_bus *sabre_bus)
@@ -633,7 +402,8 @@ static void apb_init(struct pci_bus *sabre_bus)
 	}
 }
 
-static void __init sabre_scan_bus(struct pci_pbm_info *pbm)
+static void __init sabre_scan_bus(struct pci_pbm_info *pbm,
+				  struct device *parent)
 {
 	static int once;
 
@@ -656,12 +426,12 @@ static void __init sabre_scan_bus(struct pci_pbm_info *pbm)
 	 * to live at bus 0.
 	 */
 	if (once != 0) {
-		prom_printf("SABRE: Multiple controllers unsupported.\n");
-		prom_halt();
+		printk(KERN_ERR PFX "Multiple controllers unsupported.\n");
+		return;
 	}
 	once++;
 
-	pbm->pci_bus = pci_scan_one_pbm(pbm);
+	pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
 	if (!pbm->pci_bus)
 		return;
 
@@ -672,133 +442,58 @@ static void __init sabre_scan_bus(struct pci_pbm_info *pbm)
 	sabre_register_error_handlers(pbm);
 }
 
-static int sabre_iommu_init(struct pci_pbm_info *pbm,
-			    int tsbsize, unsigned long dvma_offset,
-			    u32 dma_mask)
-{
-	struct iommu *iommu = pbm->iommu;
-	unsigned long i;
-	u64 control;
-	int err;
-
-	/* Register addresses. */
-	iommu->iommu_control  = pbm->controller_regs + SABRE_IOMMU_CONTROL;
-	iommu->iommu_tsbbase  = pbm->controller_regs + SABRE_IOMMU_TSBBASE;
-	iommu->iommu_flush    = pbm->controller_regs + SABRE_IOMMU_FLUSH;
-	iommu->iommu_tags     = iommu->iommu_flush + (0xa580UL - 0x0210UL);
-	iommu->write_complete_reg = pbm->controller_regs + SABRE_WRSYNC;
-	/* Sabre's IOMMU lacks ctx flushing. */
-	iommu->iommu_ctxflush = 0;
-                                        
-	/* Invalidate TLB Entries. */
-	control = sabre_read(pbm->controller_regs + SABRE_IOMMU_CONTROL);
-	control |= SABRE_IOMMUCTRL_DENAB;
-	sabre_write(pbm->controller_regs + SABRE_IOMMU_CONTROL, control);
-
-	for(i = 0; i < 16; i++) {
-		sabre_write(pbm->controller_regs + SABRE_IOMMU_TAG + (i * 8UL), 0);
-		sabre_write(pbm->controller_regs + SABRE_IOMMU_DATA + (i * 8UL), 0);
-	}
-
-	/* Leave diag mode enabled for full-flushing done
-	 * in pci_iommu.c
-	 */
-	err = iommu_table_init(iommu, tsbsize * 1024 * 8,
-			       dvma_offset, dma_mask, pbm->numa_node);
-	if (err)
-		return err;
-
-	sabre_write(pbm->controller_regs + SABRE_IOMMU_TSBBASE,
-		    __pa(iommu->page_table));
-
-	control = sabre_read(pbm->controller_regs + SABRE_IOMMU_CONTROL);
-	control &= ~(SABRE_IOMMUCTRL_TSBSZ | SABRE_IOMMUCTRL_TBWSZ);
-	control |= SABRE_IOMMUCTRL_ENAB;
-	switch(tsbsize) {
-	case 64:
-		control |= SABRE_IOMMU_TSBSZ_64K;
-		break;
-	case 128:
-		control |= SABRE_IOMMU_TSBSZ_128K;
-		break;
-	default:
-		prom_printf("iommu_init: Illegal TSB size %d\n", tsbsize);
-		prom_halt();
-		break;
-	}
-	sabre_write(pbm->controller_regs + SABRE_IOMMU_CONTROL, control);
-
-	return 0;
-}
-
-static void __init sabre_pbm_init(struct pci_controller_info *p,
-				  struct pci_pbm_info *pbm, struct device_node *dp)
+static void __init sabre_pbm_init(struct pci_pbm_info *pbm,
+				  struct of_device *op)
 {
-	pbm->name = dp->full_name;
-	printk("%s: SABRE PCI Bus Module\n", pbm->name);
-
-	pbm->numa_node = -1;
-
-	pbm->scan_bus = sabre_scan_bus;
-	pbm->pci_ops = &sun4u_pci_ops;
-	pbm->config_space_reg_bits = 8;
-
-	pbm->index = pci_num_pbms++;
-
-	pbm->chip_type = PBM_CHIP_TYPE_SABRE;
-	pbm->parent = p;
-	pbm->prom_node = dp;
-	pci_get_pbm_props(pbm);
-
-	pci_determine_mem_io_space(pbm);
+	psycho_pbm_init_common(pbm, op, "SABRE", PBM_CHIP_TYPE_SABRE);
+	pbm->pci_afsr = pbm->controller_regs + SABRE_PIOAFSR;
+	pbm->pci_afar = pbm->controller_regs + SABRE_PIOAFAR;
+	pbm->pci_csr = pbm->controller_regs + SABRE_PCICTRL;
+	sabre_scan_bus(pbm, &op->dev);
 }
 
-void __init sabre_init(struct device_node *dp, char *model_name)
+static int __devinit sabre_probe(struct of_device *op,
+				 const struct of_device_id *match)
 {
 	const struct linux_prom64_registers *pr_regs;
-	struct pci_controller_info *p;
+	struct device_node *dp = op->node;
 	struct pci_pbm_info *pbm;
+	u32 upa_portid, dma_mask;
 	struct iommu *iommu;
-	int tsbsize;
+	int tsbsize, err;
 	const u32 *vdma;
-	u32 upa_portid, dma_mask;
 	u64 clear_irq;
 
-	hummingbird_p = 0;
-	if (!strcmp(model_name, "pci108e,a001"))
-		hummingbird_p = 1;
-	else if (!strcmp(model_name, "SUNW,sabre")) {
-		const char *compat = of_get_property(dp, "compatible", NULL);
-		if (compat && !strcmp(compat, "pci108e,a001"))
-			hummingbird_p = 1;
-		if (!hummingbird_p) {
-			struct device_node *dp;
-
-			/* Of course, Sun has to encode things a thousand
-			 * different ways, inconsistently.
-			 */
-			for_each_node_by_type(dp, "cpu") {
-				if (!strcmp(dp->name, "SUNW,UltraSPARC-IIe"))
-					hummingbird_p = 1;
-			}
+	hummingbird_p = (match->data != NULL);
+	if (!hummingbird_p) {
+		struct device_node *cpu_dp;
+
+		/* Of course, Sun has to encode things a thousand
+		 * different ways, inconsistently.
+		 */
+		for_each_node_by_type(cpu_dp, "cpu") {
+			if (!strcmp(cpu_dp->name, "SUNW,UltraSPARC-IIe"))
+				hummingbird_p = 1;
 		}
 	}
 
-	p = kzalloc(sizeof(*p), GFP_ATOMIC);
-	if (!p)
-		goto fatal_memory_error;
+	err = -ENOMEM;
+	pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+	if (!pbm) {
+		printk(KERN_ERR PFX "Cannot allocate pci_pbm_info.\n");
+		goto out_err;
+	}
+
+	iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
+	if (!iommu) {
+		printk(KERN_ERR PFX "Cannot allocate PBM iommu.\n");
+		goto out_free_controller;
+	}
 
-	iommu = kzalloc(sizeof(*iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
-	pbm = &p->pbm_A;
 	pbm->iommu = iommu;
 
 	upa_portid = of_getintprop_default(dp, "upa-portid", 0xff);
 
-	pbm->next = pci_pbm_root;
-	pci_pbm_root = pbm;
-
 	pbm->portid = upa_portid;
 
 	/*
@@ -806,6 +501,11 @@ void __init sabre_init(struct device_node *dp, char *model_name)
 	 */
 	
 	pr_regs = of_get_property(dp, "reg", NULL);
+	err = -ENODEV;
+	if (!pr_regs) {
+		printk(KERN_ERR PFX "No reg property\n");
+		goto out_free_iommu;
+	}
 
 	/*
 	 * First REG in property is base of entire SABRE register space.
@@ -816,22 +516,25 @@ void __init sabre_init(struct device_node *dp, char *model_name)
 
 	/* PCI first */
 	for (clear_irq = SABRE_ICLR_A_SLOT0; clear_irq < SABRE_ICLR_B_SLOT0 + 0x80; clear_irq += 8)
-		sabre_write(pbm->controller_regs + clear_irq, 0x0UL);
+		upa_writeq(0x0UL, pbm->controller_regs + clear_irq);
 
 	/* Then OBIO */
 	for (clear_irq = SABRE_ICLR_SCSI; clear_irq < SABRE_ICLR_SCSI + 0x80; clear_irq += 8)
-		sabre_write(pbm->controller_regs + clear_irq, 0x0UL);
+		upa_writeq(0x0UL, pbm->controller_regs + clear_irq);
 
 	/* Error interrupts are enabled later after the bus scan. */
-	sabre_write(pbm->controller_regs + SABRE_PCICTRL,
-		    (SABRE_PCICTRL_MRLEN   | SABRE_PCICTRL_SERR |
-		     SABRE_PCICTRL_ARBPARK | SABRE_PCICTRL_AEN));
+	upa_writeq((SABRE_PCICTRL_MRLEN   | SABRE_PCICTRL_SERR |
+		    SABRE_PCICTRL_ARBPARK | SABRE_PCICTRL_AEN),
+		   pbm->controller_regs + SABRE_PCICTRL);
 
 	/* Now map in PCI config space for entire SABRE. */
-	pbm->config_space =
-		(pbm->controller_regs + SABRE_CONFIGSPACE);
+	pbm->config_space = pbm->controller_regs + SABRE_CONFIGSPACE;
 
 	vdma = of_get_property(dp, "virtual-dma", NULL);
+	if (!vdma) {
+		printk(KERN_ERR PFX "No virtual-dma property\n");
+		goto out_free_iommu;
+	}
 
 	dma_mask = vdma[0];
 	switch(vdma[1]) {
@@ -849,20 +552,58 @@ void __init sabre_init(struct device_node *dp, char *model_name)
 			tsbsize = 128;
 			break;
 		default:
-			prom_printf("SABRE: strange virtual-dma size.\n");
-			prom_halt();
+			printk(KERN_ERR PFX "Strange virtual-dma size.\n");
+			goto out_free_iommu;
 	}
 
-	if (sabre_iommu_init(pbm, tsbsize, vdma[0], dma_mask))
-		goto fatal_memory_error;
+	err = psycho_iommu_init(pbm, tsbsize, vdma[0], dma_mask, SABRE_WRSYNC);
+	if (err)
+		goto out_free_iommu;
 
 	/*
 	 * Look for APB underneath.
 	 */
-	sabre_pbm_init(p, pbm, dp);
-	return;
+	sabre_pbm_init(pbm, op);
 
-fatal_memory_error:
-	prom_printf("SABRE: Fatal memory allocation error.\n");
-	prom_halt();
+	pbm->next = pci_pbm_root;
+	pci_pbm_root = pbm;
+
+	dev_set_drvdata(&op->dev, pbm);
+
+	return 0;
+
+out_free_iommu:
+	kfree(pbm->iommu);
+
+out_free_controller:
+	kfree(pbm);
+
+out_err:
+	return err;
+}
+
+static struct of_device_id __initdata sabre_match[] = {
+	{
+		.name = "pci",
+		.compatible = "pci108e,a001",
+		.data = (void *) 1,
+	},
+	{
+		.name = "pci",
+		.compatible = "pci108e,a000",
+	},
+	{},
+};
+
+static struct of_platform_driver sabre_driver = {
+	.name		= DRIVER_NAME,
+	.match_table	= sabre_match,
+	.probe		= sabre_probe,
+};
+
+static int __init sabre_init(void)
+{
+	return of_register_driver(&sabre_driver, &of_bus_type);
 }
+
+subsys_initcall(sabre_init);
diff --git a/arch/sparc64/kernel/pci_schizo.c b/arch/sparc64/kernel/pci_schizo.c
index 615edd9c8e2..45d9dba1ba1 100644
--- a/arch/sparc64/kernel/pci_schizo.c
+++ b/arch/sparc64/kernel/pci_schizo.c
@@ -1,6 +1,6 @@
 /* pci_schizo.c: SCHIZO/TOMATILLO specific PCI controller support.
  *
- * Copyright (C) 2001, 2002, 2003, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 2001, 2002, 2003, 2007, 2008 David S. Miller (davem@davemloft.net)
  */
 
 #include <linux/kernel.h>
@@ -9,36 +9,19 @@
 #include <linux/init.h>
 #include <linux/slab.h>
 #include <linux/interrupt.h>
+#include <linux/of_device.h>
 
 #include <asm/iommu.h>
 #include <asm/irq.h>
-#include <asm/upa.h>
 #include <asm/pstate.h>
 #include <asm/prom.h>
-#include <asm/of_device.h>
-#include <asm/oplib.h>
+#include <asm/upa.h>
 
 #include "pci_impl.h"
 #include "iommu_common.h"
 
-/* All SCHIZO registers are 64-bits.  The following accessor
- * routines are how they are accessed.  The REG parameter
- * is a physical address.
- */
-#define schizo_read(__reg) \
-({	u64 __ret; \
-	__asm__ __volatile__("ldxa [%1] %2, %0" \
-			     : "=r" (__ret) \
-			     : "r" (__reg), "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory"); \
-	__ret; \
-})
-#define schizo_write(__reg, __val) \
-	__asm__ __volatile__("stxa %0, [%1] %2" \
-			     : /* no outputs */ \
-			     : "r" (__val), "r" (__reg), \
-			       "i" (ASI_PHYS_BYPASS_EC_E) \
-			     : "memory")
+#define DRIVER_NAME	"schizo"
+#define PFX		DRIVER_NAME ": "
 
 /* This is a convention that at least Excalibur and Merlin
  * follow.  I suppose the SCHIZO used in Starcat and friends
@@ -163,25 +146,25 @@ static void __schizo_check_stc_error_pbm(struct pci_pbm_info *pbm,
 	 * invalidating it before it has a chance to reach
 	 * main memory.
 	 */
-	control = schizo_read(strbuf->strbuf_control);
-	schizo_write(strbuf->strbuf_control,
-		     (control | SCHIZO_STRBUF_CTRL_DENAB));
+	control = upa_readq(strbuf->strbuf_control);
+	upa_writeq((control | SCHIZO_STRBUF_CTRL_DENAB),
+		   strbuf->strbuf_control);
 	for (i = 0; i < 128; i++) {
 		unsigned long val;
 
-		val = schizo_read(err_base + (i * 8UL));
-		schizo_write(err_base + (i * 8UL), 0UL);
+		val = upa_readq(err_base + (i * 8UL));
+		upa_writeq(0UL, err_base + (i * 8UL));
 		stc_error_buf[i] = val;
 	}
 	for (i = 0; i < 16; i++) {
-		stc_tag_buf[i] = schizo_read(tag_base + (i * 8UL));
-		stc_line_buf[i] = schizo_read(line_base + (i * 8UL));
-		schizo_write(tag_base + (i * 8UL), 0UL);
-		schizo_write(line_base + (i * 8UL), 0UL);
+		stc_tag_buf[i] = upa_readq(tag_base + (i * 8UL));
+		stc_line_buf[i] = upa_readq(line_base + (i * 8UL));
+		upa_writeq(0UL, tag_base + (i * 8UL));
+		upa_writeq(0UL, line_base + (i * 8UL));
 	}
 
 	/* OK, state is logged, exit diagnostic mode. */
-	schizo_write(strbuf->strbuf_control, control);
+	upa_writeq(control, strbuf->strbuf_control);
 
 	for (i = 0; i < 16; i++) {
 		int j, saw_error, first, last;
@@ -258,14 +241,14 @@ static void schizo_check_iommu_error_pbm(struct pci_pbm_info *pbm,
 	int i;
 
 	spin_lock_irqsave(&iommu->lock, flags);
-	control = schizo_read(iommu->iommu_control);
+	control = upa_readq(iommu->iommu_control);
 	if (control & SCHIZO_IOMMU_CTRL_XLTEERR) {
 		unsigned long base;
 		char *type_string;
 
 		/* Clear the error encountered bit. */
 		control &= ~SCHIZO_IOMMU_CTRL_XLTEERR;
-		schizo_write(iommu->iommu_control, control);
+		upa_writeq(control, iommu->iommu_control);
 
 		switch((control & SCHIZO_IOMMU_CTRL_XLTESTAT) >> 25UL) {
 		case 0:
@@ -295,24 +278,24 @@ static void schizo_check_iommu_error_pbm(struct pci_pbm_info *pbm,
 		 * get as much diagnostic information to the
 		 * console as we can.
 		 */
-		schizo_write(iommu->iommu_control,
-			     control | SCHIZO_IOMMU_CTRL_DENAB);
+		upa_writeq(control | SCHIZO_IOMMU_CTRL_DENAB,
+			   iommu->iommu_control);
 
 		base = pbm->pbm_regs;
 
 		for (i = 0; i < 16; i++) {
 			iommu_tag[i] =
-				schizo_read(base + SCHIZO_IOMMU_TAG + (i * 8UL));
+				upa_readq(base + SCHIZO_IOMMU_TAG + (i * 8UL));
 			iommu_data[i] =
-				schizo_read(base + SCHIZO_IOMMU_DATA + (i * 8UL));
+				upa_readq(base + SCHIZO_IOMMU_DATA + (i * 8UL));
 
 			/* Now clear out the entry. */
-			schizo_write(base + SCHIZO_IOMMU_TAG + (i * 8UL), 0);
-			schizo_write(base + SCHIZO_IOMMU_DATA + (i * 8UL), 0);
+			upa_writeq(0, base + SCHIZO_IOMMU_TAG + (i * 8UL));
+			upa_writeq(0, base + SCHIZO_IOMMU_DATA + (i * 8UL));
 		}
 
 		/* Leave diagnostic mode. */
-		schizo_write(iommu->iommu_control, control);
+		upa_writeq(control, iommu->iommu_control);
 
 		for (i = 0; i < 16; i++) {
 			unsigned long tag, data;
@@ -357,11 +340,12 @@ static void schizo_check_iommu_error_pbm(struct pci_pbm_info *pbm,
 	spin_unlock_irqrestore(&iommu->lock, flags);
 }
 
-static void schizo_check_iommu_error(struct pci_controller_info *p,
+static void schizo_check_iommu_error(struct pci_pbm_info *pbm,
 				     enum schizo_error_type type)
 {
-	schizo_check_iommu_error_pbm(&p->pbm_A, type);
-	schizo_check_iommu_error_pbm(&p->pbm_B, type);
+	schizo_check_iommu_error_pbm(pbm, type);
+	if (pbm->sibling)
+		schizo_check_iommu_error_pbm(pbm->sibling, type);
 }
 
 /* Uncorrectable ECC error status gathering. */
@@ -386,14 +370,13 @@ static void schizo_check_iommu_error(struct pci_controller_info *p,
 static irqreturn_t schizo_ue_intr(int irq, void *dev_id)
 {
 	struct pci_pbm_info *pbm = dev_id;
-	struct pci_controller_info *p = pbm->parent;
 	unsigned long afsr_reg = pbm->controller_regs + SCHIZO_UE_AFSR;
 	unsigned long afar_reg = pbm->controller_regs + SCHIZO_UE_AFAR;
 	unsigned long afsr, afar, error_bits;
 	int reported, limit;
 
 	/* Latch uncorrectable error status. */
-	afar = schizo_read(afar_reg);
+	afar = upa_readq(afar_reg);
 
 	/* If either of the error pending bits are set in the
 	 * AFSR, the error status is being actively updated by
@@ -401,7 +384,7 @@ static irqreturn_t schizo_ue_intr(int irq, void *dev_id)
 	 */
 	limit = 1000;
 	do {
-		afsr = schizo_read(afsr_reg);
+		afsr = upa_readq(afsr_reg);
 	} while ((afsr & SCHIZO_UEAFSR_ERRPNDG) != 0 && --limit);
 
 	/* Clear the primary/secondary error status bits. */
@@ -410,7 +393,7 @@ static irqreturn_t schizo_ue_intr(int irq, void *dev_id)
 		 SCHIZO_UEAFSR_SPIO | SCHIZO_UEAFSR_SDMA);
 	if (!error_bits)
 		return IRQ_NONE;
-	schizo_write(afsr_reg, error_bits);
+	upa_writeq(error_bits, afsr_reg);
 
 	/* Log the error. */
 	printk("%s: Uncorrectable Error, primary error type[%s]\n",
@@ -449,7 +432,7 @@ static irqreturn_t schizo_ue_intr(int irq, void *dev_id)
 	printk("]\n");
 
 	/* Interrogate IOMMU for error status. */
-	schizo_check_iommu_error(p, UE_ERR);
+	schizo_check_iommu_error(pbm, UE_ERR);
 
 	return IRQ_HANDLED;
 }
@@ -481,7 +464,7 @@ static irqreturn_t schizo_ce_intr(int irq, void *dev_id)
 	int reported, limit;
 
 	/* Latch error status. */
-	afar = schizo_read(afar_reg);
+	afar = upa_readq(afar_reg);
 
 	/* If either of the error pending bits are set in the
 	 * AFSR, the error status is being actively updated by
@@ -489,7 +472,7 @@ static irqreturn_t schizo_ce_intr(int irq, void *dev_id)
 	 */
 	limit = 1000;
 	do {
-		afsr = schizo_read(afsr_reg);
+		afsr = upa_readq(afsr_reg);
 	} while ((afsr & SCHIZO_UEAFSR_ERRPNDG) != 0 && --limit);
 
 	/* Clear primary/secondary error status bits. */
@@ -498,7 +481,7 @@ static irqreturn_t schizo_ce_intr(int irq, void *dev_id)
 		 SCHIZO_CEAFSR_SPIO | SCHIZO_CEAFSR_SDMA);
 	if (!error_bits)
 		return IRQ_NONE;
-	schizo_write(afsr_reg, error_bits);
+	upa_writeq(error_bits, afsr_reg);
 
 	/* Log the error. */
 	printk("%s: Correctable Error, primary error type[%s]\n",
@@ -600,7 +583,7 @@ static irqreturn_t schizo_pcierr_intr_other(struct pci_pbm_info *pbm)
 	u16 stat;
 
 	csr_reg = pbm->pbm_regs + SCHIZO_PCI_CTRL;
-	csr = schizo_read(csr_reg);
+	csr = upa_readq(csr_reg);
 	csr_error_bits =
 		csr & (SCHIZO_PCICTRL_BUS_UNUS |
 		       SCHIZO_PCICTRL_TTO_ERR |
@@ -610,7 +593,7 @@ static irqreturn_t schizo_pcierr_intr_other(struct pci_pbm_info *pbm)
 		       SCHIZO_PCICTRL_SERR);
 	if (csr_error_bits) {
 		/* Clear the errors.  */
-		schizo_write(csr_reg, csr);
+		upa_writeq(csr, csr_reg);
 
 		/* Log 'em.  */
 		if (csr_error_bits & SCHIZO_PCICTRL_BUS_UNUS)
@@ -650,7 +633,6 @@ static irqreturn_t schizo_pcierr_intr_other(struct pci_pbm_info *pbm)
 static irqreturn_t schizo_pcierr_intr(int irq, void *dev_id)
 {
 	struct pci_pbm_info *pbm = dev_id;
-	struct pci_controller_info *p = pbm->parent;
 	unsigned long afsr_reg, afar_reg, base;
 	unsigned long afsr, afar, error_bits;
 	int reported;
@@ -661,8 +643,8 @@ static irqreturn_t schizo_pcierr_intr(int irq, void *dev_id)
 	afar_reg = base + SCHIZO_PCI_AFAR;
 
 	/* Latch error status. */
-	afar = schizo_read(afar_reg);
-	afsr = schizo_read(afsr_reg);
+	afar = upa_readq(afar_reg);
+	afsr = upa_readq(afsr_reg);
 
 	/* Clear primary/secondary error status bits. */
 	error_bits = afsr &
@@ -674,7 +656,7 @@ static irqreturn_t schizo_pcierr_intr(int irq, void *dev_id)
 		 SCHIZO_PCIAFSR_STTO | SCHIZO_PCIAFSR_SUNUS);
 	if (!error_bits)
 		return schizo_pcierr_intr_other(pbm);
-	schizo_write(afsr_reg, error_bits);
+	upa_writeq(error_bits, afsr_reg);
 
 	/* Log the error. */
 	printk("%s: PCI Error, primary error type[%s]\n",
@@ -744,7 +726,7 @@ static irqreturn_t schizo_pcierr_intr(int irq, void *dev_id)
 	 * a bug in the IOMMU support code or a PCI device driver.
 	 */
 	if (error_bits & (SCHIZO_PCIAFSR_PTA | SCHIZO_PCIAFSR_STA)) {
-		schizo_check_iommu_error(p, PCI_ERR);
+		schizo_check_iommu_error(pbm, PCI_ERR);
 		pci_scan_for_target_abort(pbm, pbm->pci_bus);
 	}
 	if (error_bits & (SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_SMA))
@@ -805,12 +787,11 @@ static irqreturn_t schizo_pcierr_intr(int irq, void *dev_id)
 static irqreturn_t schizo_safarierr_intr(int irq, void *dev_id)
 {
 	struct pci_pbm_info *pbm = dev_id;
-	struct pci_controller_info *p = pbm->parent;
 	u64 errlog;
 
-	errlog = schizo_read(pbm->controller_regs + SCHIZO_SAFARI_ERRLOG);
-	schizo_write(pbm->controller_regs + SCHIZO_SAFARI_ERRLOG,
-		     errlog & ~(SAFARI_ERRLOG_ERROUT));
+	errlog = upa_readq(pbm->controller_regs + SCHIZO_SAFARI_ERRLOG);
+	upa_writeq(errlog & ~(SAFARI_ERRLOG_ERROUT),
+		   pbm->controller_regs + SCHIZO_SAFARI_ERRLOG);
 
 	if (!(errlog & BUS_ERROR_UNMAP)) {
 		printk("%s: Unexpected Safari/JBUS error interrupt, errlog[%016lx]\n",
@@ -821,7 +802,7 @@ static irqreturn_t schizo_safarierr_intr(int irq, void *dev_id)
 
 	printk("%s: Safari/JBUS interrupt, UNMAPPED error, interrogating IOMMUs.\n",
 	       pbm->name);
-	schizo_check_iommu_error(p, SAFARI_ERR);
+	schizo_check_iommu_error(pbm, SAFARI_ERR);
 
 	return IRQ_HANDLED;
 }
@@ -863,7 +844,7 @@ static int pbm_routes_this_ino(struct pci_pbm_info *pbm, u32 ino)
  */
 static void tomatillo_register_error_handlers(struct pci_pbm_info *pbm)
 {
-	struct of_device *op = of_find_device_by_node(pbm->prom_node);
+	struct of_device *op = of_find_device_by_node(pbm->op->node);
 	u64 tmp, err_mask, err_no_mask;
 	int err;
 
@@ -910,10 +891,9 @@ static void tomatillo_register_error_handlers(struct pci_pbm_info *pbm)
 	}
 
 	/* Enable UE and CE interrupts for controller. */
-	schizo_write(pbm->controller_regs + SCHIZO_ECC_CTRL,
-		     (SCHIZO_ECCCTRL_EE |
-		      SCHIZO_ECCCTRL_UE |
-		      SCHIZO_ECCCTRL_CE));
+	upa_writeq((SCHIZO_ECCCTRL_EE |
+		    SCHIZO_ECCCTRL_UE |
+		    SCHIZO_ECCCTRL_CE), pbm->controller_regs + SCHIZO_ECC_CTRL);
 
 	/* Enable PCI Error interrupts and clear error
 	 * bits.
@@ -926,10 +906,10 @@ static void tomatillo_register_error_handlers(struct pci_pbm_info *pbm)
 
 	err_no_mask = SCHIZO_PCICTRL_DTO_ERR;
 
-	tmp = schizo_read(pbm->pbm_regs + SCHIZO_PCI_CTRL);
+	tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_CTRL);
 	tmp |= err_mask;
 	tmp &= ~err_no_mask;
-	schizo_write(pbm->pbm_regs + SCHIZO_PCI_CTRL, tmp);
+	upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_CTRL);
 
 	err_mask = (SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_PTA |
 		    SCHIZO_PCIAFSR_PRTRY | SCHIZO_PCIAFSR_PPERR |
@@ -938,7 +918,7 @@ static void tomatillo_register_error_handlers(struct pci_pbm_info *pbm)
 		    SCHIZO_PCIAFSR_SRTRY | SCHIZO_PCIAFSR_SPERR |
 		    SCHIZO_PCIAFSR_STTO);
 
-	schizo_write(pbm->pbm_regs + SCHIZO_PCI_AFSR, err_mask);
+	upa_writeq(err_mask, pbm->pbm_regs + SCHIZO_PCI_AFSR);
 
 	err_mask = (BUS_ERROR_BADCMD | BUS_ERROR_SNOOP_GR |
 		    BUS_ERROR_SNOOP_PCI | BUS_ERROR_SNOOP_RD |
@@ -950,16 +930,16 @@ static void tomatillo_register_error_handlers(struct pci_pbm_info *pbm)
 		    BUS_ERROR_APERR | BUS_ERROR_UNMAP |
 		    BUS_ERROR_BUSERR | BUS_ERROR_TIMEOUT);
 
-	schizo_write(pbm->controller_regs + SCHIZO_SAFARI_ERRCTRL,
-		     (SCHIZO_SAFERRCTRL_EN | err_mask));
+	upa_writeq((SCHIZO_SAFERRCTRL_EN | err_mask),
+		   pbm->controller_regs + SCHIZO_SAFARI_ERRCTRL);
 
-	schizo_write(pbm->controller_regs + SCHIZO_SAFARI_IRQCTRL,
-		     (SCHIZO_SAFIRQCTRL_EN | (BUS_ERROR_UNMAP)));
+	upa_writeq((SCHIZO_SAFIRQCTRL_EN | (BUS_ERROR_UNMAP)),
+		   pbm->controller_regs + SCHIZO_SAFARI_IRQCTRL);
 }
 
 static void schizo_register_error_handlers(struct pci_pbm_info *pbm)
 {
-	struct of_device *op = of_find_device_by_node(pbm->prom_node);
+	struct of_device *op = of_find_device_by_node(pbm->op->node);
 	u64 tmp, err_mask, err_no_mask;
 	int err;
 
@@ -1006,10 +986,9 @@ static void schizo_register_error_handlers(struct pci_pbm_info *pbm)
 	}
 
 	/* Enable UE and CE interrupts for controller. */
-	schizo_write(pbm->controller_regs + SCHIZO_ECC_CTRL,
-		     (SCHIZO_ECCCTRL_EE |
-		      SCHIZO_ECCCTRL_UE |
-		      SCHIZO_ECCCTRL_CE));
+	upa_writeq((SCHIZO_ECCCTRL_EE |
+		    SCHIZO_ECCCTRL_UE |
+		    SCHIZO_ECCCTRL_CE), pbm->controller_regs + SCHIZO_ECC_CTRL);
 
 	err_mask = (SCHIZO_PCICTRL_BUS_UNUS |
 		    SCHIZO_PCICTRL_ESLCK |
@@ -1025,18 +1004,18 @@ static void schizo_register_error_handlers(struct pci_pbm_info *pbm)
 	/* Enable PCI Error interrupts and clear error
 	 * bits for each PBM.
 	 */
-	tmp = schizo_read(pbm->pbm_regs + SCHIZO_PCI_CTRL);
+	tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_CTRL);
 	tmp |= err_mask;
 	tmp &= ~err_no_mask;
-	schizo_write(pbm->pbm_regs + SCHIZO_PCI_CTRL, tmp);
+	upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_CTRL);
 
-	schizo_write(pbm->pbm_regs + SCHIZO_PCI_AFSR,
-		     (SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_PTA |
-		      SCHIZO_PCIAFSR_PRTRY | SCHIZO_PCIAFSR_PPERR |
-		      SCHIZO_PCIAFSR_PTTO | SCHIZO_PCIAFSR_PUNUS |
-		      SCHIZO_PCIAFSR_SMA | SCHIZO_PCIAFSR_STA |
-		      SCHIZO_PCIAFSR_SRTRY | SCHIZO_PCIAFSR_SPERR |
-		      SCHIZO_PCIAFSR_STTO | SCHIZO_PCIAFSR_SUNUS));
+	upa_writeq((SCHIZO_PCIAFSR_PMA | SCHIZO_PCIAFSR_PTA |
+		    SCHIZO_PCIAFSR_PRTRY | SCHIZO_PCIAFSR_PPERR |
+		    SCHIZO_PCIAFSR_PTTO | SCHIZO_PCIAFSR_PUNUS |
+		    SCHIZO_PCIAFSR_SMA | SCHIZO_PCIAFSR_STA |
+		    SCHIZO_PCIAFSR_SRTRY | SCHIZO_PCIAFSR_SPERR |
+		    SCHIZO_PCIAFSR_STTO | SCHIZO_PCIAFSR_SUNUS),
+		   pbm->pbm_regs + SCHIZO_PCI_AFSR);
 
 	/* Make all Safari error conditions fatal except unmapped
 	 * errors which we make generate interrupts.
@@ -1063,8 +1042,8 @@ static void schizo_register_error_handlers(struct pci_pbm_info *pbm)
 		      BUS_ERROR_CPU0PS | BUS_ERROR_CPU0PB);
 #endif
 
-	schizo_write(pbm->controller_regs + SCHIZO_SAFARI_ERRCTRL,
-		     (SCHIZO_SAFERRCTRL_EN | err_mask));
+	upa_writeq((SCHIZO_SAFERRCTRL_EN | err_mask),
+		   pbm->controller_regs + SCHIZO_SAFARI_ERRCTRL);
 }
 
 static void pbm_config_busmastering(struct pci_pbm_info *pbm)
@@ -1084,14 +1063,15 @@ static void pbm_config_busmastering(struct pci_pbm_info *pbm)
 	pci_config_write8(addr, 64);
 }
 
-static void __init schizo_scan_bus(struct pci_pbm_info *pbm)
+static void __devinit schizo_scan_bus(struct pci_pbm_info *pbm,
+				      struct device *parent)
 {
 	pbm_config_busmastering(pbm);
 	pbm->is_66mhz_capable =
-		(of_find_property(pbm->prom_node, "66mhz-capable", NULL)
+		(of_find_property(pbm->op->node, "66mhz-capable", NULL)
 		 != NULL);
 
-	pbm->pci_bus = pci_scan_one_pbm(pbm);
+	pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
 
 	if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO)
 		tomatillo_register_error_handlers(pbm);
@@ -1133,12 +1113,12 @@ static void schizo_pbm_strbuf_init(struct pci_pbm_info *pbm)
 	 * streaming buffer and leave the rerun-disable
 	 * setting however OBP set it.
 	 */
-	control = schizo_read(pbm->stc.strbuf_control);
+	control = upa_readq(pbm->stc.strbuf_control);
 	control &= ~(SCHIZO_STRBUF_CTRL_LPTR |
 		     SCHIZO_STRBUF_CTRL_LENAB |
 		     SCHIZO_STRBUF_CTRL_DENAB);
 	control |= SCHIZO_STRBUF_CTRL_ENAB;
-	schizo_write(pbm->stc.strbuf_control, control);
+	upa_writeq(control, pbm->stc.strbuf_control);
 
 	pbm->stc.strbuf_enabled = 1;
 }
@@ -1150,24 +1130,17 @@ static void schizo_pbm_strbuf_init(struct pci_pbm_info *pbm)
 
 static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm)
 {
-	struct iommu *iommu = pbm->iommu;
+	static const u32 vdma_default[] = { 0xc0000000, 0x40000000 };
 	unsigned long i, tagbase, database;
-	struct property *prop;
-	u32 vdma[2], dma_mask;
+	struct iommu *iommu = pbm->iommu;
 	int tsbsize, err;
+	const u32 *vdma;
+	u32 dma_mask;
 	u64 control;
 
-	prop = of_find_property(pbm->prom_node, "virtual-dma", NULL);
-	if (prop) {
-		u32 *val = prop->value;
-
-		vdma[0] = val[0];
-		vdma[1] = val[1];
-	} else {
-		/* No property, use default values. */
-		vdma[0] = 0xc0000000;
-		vdma[1] = 0x40000000;
-	}
+	vdma = of_get_property(pbm->op->node, "virtual-dma", NULL);
+	if (!vdma)
+		vdma = vdma_default;
 
 	dma_mask = vdma[0];
 	switch (vdma[1]) {
@@ -1187,9 +1160,9 @@ static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm)
 			break;
 
 		default:
-			prom_printf("SCHIZO: strange virtual-dma size.\n");
-			prom_halt();
-	};
+			printk(KERN_ERR PFX "Strange virtual-dma size.\n");
+			return -EINVAL;
+	}
 
 	/* Register addresses, SCHIZO has iommu ctx flushing. */
 	iommu->iommu_control  = pbm->pbm_regs + SCHIZO_IOMMU_CONTROL;
@@ -1206,15 +1179,15 @@ static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm)
 	/*
 	 * Invalidate TLB Entries.
 	 */
-	control = schizo_read(iommu->iommu_control);
+	control = upa_readq(iommu->iommu_control);
 	control |= SCHIZO_IOMMU_CTRL_DENAB;
-	schizo_write(iommu->iommu_control, control);
+	upa_writeq(control, iommu->iommu_control);
 
 	tagbase = SCHIZO_IOMMU_TAG, database = SCHIZO_IOMMU_DATA;
 
-	for(i = 0; i < 16; i++) {
-		schizo_write(pbm->pbm_regs + tagbase + (i * 8UL), 0);
-		schizo_write(pbm->pbm_regs + database + (i * 8UL), 0);
+	for (i = 0; i < 16; i++) {
+		upa_writeq(0, pbm->pbm_regs + tagbase + (i * 8UL));
+		upa_writeq(0, pbm->pbm_regs + database + (i * 8UL));
 	}
 
 	/* Leave diag mode enabled for full-flushing done
@@ -1222,12 +1195,14 @@ static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm)
 	 */
 	err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask,
 			       pbm->numa_node);
-	if (err)
+	if (err) {
+		printk(KERN_ERR PFX "iommu_table_init() fails with %d\n", err);
 		return err;
+	}
 
-	schizo_write(iommu->iommu_tsbbase, __pa(iommu->page_table));
+	upa_writeq(__pa(iommu->page_table), iommu->iommu_tsbbase);
 
-	control = schizo_read(iommu->iommu_control);
+	control = upa_readq(iommu->iommu_control);
 	control &= ~(SCHIZO_IOMMU_CTRL_TSBSZ | SCHIZO_IOMMU_CTRL_TBWSZ);
 	switch (tsbsize) {
 	case 64:
@@ -1236,10 +1211,10 @@ static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm)
 	case 128:
 		control |= SCHIZO_IOMMU_TSBSZ_128K;
 		break;
-	};
+	}
 
 	control |= SCHIZO_IOMMU_CTRL_ENAB;
-	schizo_write(iommu->iommu_control, control);
+	upa_writeq(control, iommu->iommu_control);
 
 	return 0;
 }
@@ -1280,12 +1255,11 @@ static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm)
 
 static void schizo_pbm_hw_init(struct pci_pbm_info *pbm)
 {
-	struct property *prop;
 	u64 tmp;
 
-	schizo_write(pbm->pbm_regs + SCHIZO_PCI_IRQ_RETRY, 5);
+	upa_writeq(5, pbm->pbm_regs + SCHIZO_PCI_IRQ_RETRY);
 
-	tmp = schizo_read(pbm->pbm_regs + SCHIZO_PCI_CTRL);
+	tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_CTRL);
 
 	/* Enable arbiter for all PCI slots.  */
 	tmp |= 0xff;
@@ -1294,8 +1268,7 @@ static void schizo_pbm_hw_init(struct pci_pbm_info *pbm)
 	    pbm->chip_version >= 0x2)
 		tmp |= 0x3UL << SCHIZO_PCICTRL_PTO_SHIFT;
 
-	prop = of_find_property(pbm->prom_node, "no-bus-parking", NULL);
-	if (!prop)
+	if (!of_find_property(pbm->op->node, "no-bus-parking", NULL))
 		tmp |= SCHIZO_PCICTRL_PARK;
 	else
 		tmp &= ~SCHIZO_PCICTRL_PARK;
@@ -1311,13 +1284,13 @@ static void schizo_pbm_hw_init(struct pci_pbm_info *pbm)
 			SCHIZO_PCICTRL_RDO_PREF |
 			SCHIZO_PCICTRL_RDL_PREF);
 
-	schizo_write(pbm->pbm_regs + SCHIZO_PCI_CTRL, tmp);
+	upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_CTRL);
 
-	tmp = schizo_read(pbm->pbm_regs + SCHIZO_PCI_DIAG);
+	tmp = upa_readq(pbm->pbm_regs + SCHIZO_PCI_DIAG);
 	tmp &= ~(SCHIZO_PCIDIAG_D_RTRYARB |
 		 SCHIZO_PCIDIAG_D_RETRY |
 		 SCHIZO_PCIDIAG_D_INTSYNC);
-	schizo_write(pbm->pbm_regs + SCHIZO_PCI_DIAG, tmp);
+	upa_writeq(tmp, pbm->pbm_regs + SCHIZO_PCI_DIAG);
 
 	if (pbm->chip_type == PBM_CHIP_TYPE_TOMATILLO) {
 		/* Clear prefetch lengths to workaround a bug in
@@ -1329,17 +1302,16 @@ static void schizo_pbm_hw_init(struct pci_pbm_info *pbm)
 		       TOMATILLO_IOC_RDONE_CPENAB |
 		       TOMATILLO_IOC_RDLINE_CPENAB);
 
-		schizo_write(pbm->pbm_regs + TOMATILLO_PCI_IOC_CSR,
-			     tmp);
+		upa_writeq(tmp, pbm->pbm_regs + TOMATILLO_PCI_IOC_CSR);
 	}
 }
 
-static int __init schizo_pbm_init(struct pci_controller_info *p,
-				  struct device_node *dp, u32 portid,
-				  int chip_type)
+static int __devinit schizo_pbm_init(struct pci_pbm_info *pbm,
+				     struct of_device *op, u32 portid,
+				     int chip_type)
 {
 	const struct linux_prom64_registers *regs;
-	struct pci_pbm_info *pbm;
+	struct device_node *dp = op->node;
 	const char *chipset_name;
 	int is_pbm_a, err;
 
@@ -1372,25 +1344,19 @@ static int __init schizo_pbm_init(struct pci_controller_info *p,
 	regs = of_get_property(dp, "reg", NULL);
 
 	is_pbm_a = ((regs[0].phys_addr & 0x00700000) == 0x00600000);
-	if (is_pbm_a)
-		pbm = &p->pbm_A;
-	else
-		pbm = &p->pbm_B;
 
 	pbm->next = pci_pbm_root;
 	pci_pbm_root = pbm;
 
 	pbm->numa_node = -1;
 
-	pbm->scan_bus = schizo_scan_bus;
 	pbm->pci_ops = &sun4u_pci_ops;
 	pbm->config_space_reg_bits = 8;
 
 	pbm->index = pci_num_pbms++;
 
 	pbm->portid = portid;
-	pbm->parent = p;
-	pbm->prom_node = dp;
+	pbm->op = op;
 
 	pbm->chip_type = chip_type;
 	pbm->chip_version = of_getintprop_default(dp, "version#", 0);
@@ -1420,6 +1386,8 @@ static int __init schizo_pbm_init(struct pci_controller_info *p,
 
 	schizo_pbm_strbuf_init(pbm);
 
+	schizo_scan_bus(pbm, &op->dev);
+
 	return 0;
 }
 
@@ -1433,62 +1401,104 @@ static inline int portid_compare(u32 x, u32 y, int chip_type)
 	return (x == y);
 }
 
-static void __init __schizo_init(struct device_node *dp, char *model_name,
-				 int chip_type)
+static struct pci_pbm_info * __devinit schizo_find_sibling(u32 portid,
+							   int chip_type)
 {
-	struct pci_controller_info *p;
+	struct pci_pbm_info *pbm;
+
+	for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
+		if (portid_compare(pbm->portid, portid, chip_type))
+			return pbm;
+	}
+	return NULL;
+}
+
+static int __devinit __schizo_init(struct of_device *op, unsigned long chip_type)
+{
+	struct device_node *dp = op->node;
 	struct pci_pbm_info *pbm;
 	struct iommu *iommu;
 	u32 portid;
+	int err;
 
 	portid = of_getintprop_default(dp, "portid", 0xff);
 
-	for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
-		if (portid_compare(pbm->portid, portid, chip_type)) {
-			if (schizo_pbm_init(pbm->parent, dp,
-					    portid, chip_type))
-				goto fatal_memory_error;
-			return;
-		}
+	err = -ENOMEM;
+	pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+	if (!pbm) {
+		printk(KERN_ERR PFX "Cannot allocate pci_pbm_info.\n");
+		goto out_err;
+	}
+
+	pbm->sibling = schizo_find_sibling(portid, chip_type);
+
+	iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+	if (!iommu) {
+		printk(KERN_ERR PFX "Cannot allocate PBM A iommu.\n");
+		goto out_free_pbm;
 	}
 
-	p = kzalloc(sizeof(struct pci_controller_info), GFP_ATOMIC);
-	if (!p)
-		goto fatal_memory_error;
+	pbm->iommu = iommu;
 
-	iommu = kzalloc(sizeof(struct iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
+	if (schizo_pbm_init(pbm, op, portid, chip_type))
+		goto out_free_iommu;
 
-	p->pbm_A.iommu = iommu;
+	if (pbm->sibling)
+		pbm->sibling->sibling = pbm;
 
-	iommu = kzalloc(sizeof(struct iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
+	dev_set_drvdata(&op->dev, pbm);
 
-	p->pbm_B.iommu = iommu;
+	return 0;
 
-	if (schizo_pbm_init(p, dp, portid, chip_type))
-		goto fatal_memory_error;
+out_free_iommu:
+	kfree(pbm->iommu);
 
-	return;
+out_free_pbm:
+	kfree(pbm);
 
-fatal_memory_error:
-	prom_printf("SCHIZO: Fatal memory allocation error.\n");
-	prom_halt();
+out_err:
+	return err;
 }
 
-void __init schizo_init(struct device_node *dp, char *model_name)
+static int __devinit schizo_probe(struct of_device *op,
+				  const struct of_device_id *match)
 {
-	__schizo_init(dp, model_name, PBM_CHIP_TYPE_SCHIZO);
+	return __schizo_init(op, (unsigned long) match->data);
 }
 
-void __init schizo_plus_init(struct device_node *dp, char *model_name)
-{
-	__schizo_init(dp, model_name, PBM_CHIP_TYPE_SCHIZO_PLUS);
-}
+/* The ordering of this table is very important.  Some Tomatillo
+ * nodes announce that they are compatible with both pci108e,a801
+ * and pci108e,8001.  So list the chips in reverse chronological
+ * order.
+ */
+static struct of_device_id __initdata schizo_match[] = {
+	{
+		.name = "pci",
+		.compatible = "pci108e,a801",
+		.data = (void *) PBM_CHIP_TYPE_TOMATILLO,
+	},
+	{
+		.name = "pci",
+		.compatible = "pci108e,8002",
+		.data = (void *) PBM_CHIP_TYPE_SCHIZO_PLUS,
+	},
+	{
+		.name = "pci",
+		.compatible = "pci108e,8001",
+		.data = (void *) PBM_CHIP_TYPE_SCHIZO,
+	},
+	{},
+};
 
-void __init tomatillo_init(struct device_node *dp, char *model_name)
+static struct of_platform_driver schizo_driver = {
+	.name		= DRIVER_NAME,
+	.match_table	= schizo_match,
+	.probe		= schizo_probe,
+};
+
+static int __init schizo_init(void)
 {
-	__schizo_init(dp, model_name, PBM_CHIP_TYPE_TOMATILLO);
+	return of_register_driver(&schizo_driver, &of_bus_type);
 }
+
+subsys_initcall(schizo_init);
diff --git a/arch/sparc64/kernel/pci_sun4v.c b/arch/sparc64/kernel/pci_sun4v.c
index e2bb9790039..e86c73ec167 100644
--- a/arch/sparc64/kernel/pci_sun4v.c
+++ b/arch/sparc64/kernel/pci_sun4v.c
@@ -13,12 +13,10 @@
 #include <linux/irq.h>
 #include <linux/msi.h>
 #include <linux/log2.h>
+#include <linux/of_device.h>
 
 #include <asm/iommu.h>
 #include <asm/irq.h>
-#include <asm/upa.h>
-#include <asm/pstate.h>
-#include <asm/oplib.h>
 #include <asm/hypervisor.h>
 #include <asm/prom.h>
 
@@ -27,6 +25,9 @@
 
 #include "pci_sun4v.h"
 
+#define DRIVER_NAME	"pci_sun4v"
+#define PFX		DRIVER_NAME ": "
+
 static unsigned long vpci_major = 1;
 static unsigned long vpci_minor = 1;
 
@@ -41,6 +42,7 @@ struct iommu_batch {
 };
 
 static DEFINE_PER_CPU(struct iommu_batch, iommu_batch);
+static int iommu_batch_initialized;
 
 /* Interrupts must be disabled.  */
 static inline void iommu_batch_start(struct device *dev, unsigned long prot, unsigned long entry)
@@ -531,7 +533,7 @@ static void dma_4v_sync_sg_for_cpu(struct device *dev,
 	/* Nothing to do... */
 }
 
-const struct dma_ops sun4v_dma_ops = {
+static const struct dma_ops sun4v_dma_ops = {
 	.alloc_coherent			= dma_4v_alloc_coherent,
 	.free_coherent			= dma_4v_free_coherent,
 	.map_single			= dma_4v_map_single,
@@ -542,15 +544,16 @@ const struct dma_ops sun4v_dma_ops = {
 	.sync_sg_for_cpu		= dma_4v_sync_sg_for_cpu,
 };
 
-static void __init pci_sun4v_scan_bus(struct pci_pbm_info *pbm)
+static void __init pci_sun4v_scan_bus(struct pci_pbm_info *pbm,
+				      struct device *parent)
 {
 	struct property *prop;
 	struct device_node *dp;
 
-	dp = pbm->prom_node;
+	dp = pbm->op->node;
 	prop = of_find_property(dp, "66mhz-capable", NULL);
 	pbm->is_66mhz_capable = (prop != NULL);
-	pbm->pci_bus = pci_scan_one_pbm(pbm);
+	pbm->pci_bus = pci_scan_one_pbm(pbm, parent);
 
 	/* XXX register error interrupt handlers XXX */
 }
@@ -583,29 +586,22 @@ static unsigned long __init probe_existing_entries(struct pci_pbm_info *pbm,
 	return cnt;
 }
 
-static void __init pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
+static int __init pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
 {
+	static const u32 vdma_default[] = { 0x80000000, 0x80000000 };
 	struct iommu *iommu = pbm->iommu;
-	struct property *prop;
 	unsigned long num_tsb_entries, sz, tsbsize;
-	u32 vdma[2], dma_mask, dma_offset;
-
-	prop = of_find_property(pbm->prom_node, "virtual-dma", NULL);
-	if (prop) {
-		u32 *val = prop->value;
-
-		vdma[0] = val[0];
-		vdma[1] = val[1];
-	} else {
-		/* No property, use default values. */
-		vdma[0] = 0x80000000;
-		vdma[1] = 0x80000000;
-	}
+	u32 dma_mask, dma_offset;
+	const u32 *vdma;
+
+	vdma = of_get_property(pbm->op->node, "virtual-dma", NULL);
+	if (!vdma)
+		vdma = vdma_default;
 
 	if ((vdma[0] | vdma[1]) & ~IO_PAGE_MASK) {
-		prom_printf("PCI-SUN4V: strange virtual-dma[%08x:%08x].\n",
-			    vdma[0], vdma[1]);
-		prom_halt();
+		printk(KERN_ERR PFX "Strange virtual-dma[%08x:%08x].\n",
+		       vdma[0], vdma[1]);
+		return -EINVAL;
 	};
 
 	dma_mask = (roundup_pow_of_two(vdma[1]) - 1UL);
@@ -625,8 +621,8 @@ static void __init pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
 	sz = (sz + 7UL) & ~7UL;
 	iommu->arena.map = kzalloc(sz, GFP_KERNEL);
 	if (!iommu->arena.map) {
-		prom_printf("PCI_IOMMU: Error, kmalloc(arena.map) failed.\n");
-		prom_halt();
+		printk(KERN_ERR PFX "Error, kmalloc(arena.map) failed.\n");
+		return -ENOMEM;
 	}
 	iommu->arena.limit = num_tsb_entries;
 
@@ -634,6 +630,8 @@ static void __init pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
 	if (sz)
 		printk("%s: Imported %lu TSB entries from OBP\n",
 		       pbm->name, sz);
+
+	return 0;
 }
 
 #ifdef CONFIG_PCI_MSI
@@ -890,29 +888,20 @@ static void pci_sun4v_msi_init(struct pci_pbm_info *pbm)
 }
 #endif /* !(CONFIG_PCI_MSI) */
 
-static void __init pci_sun4v_pbm_init(struct pci_controller_info *p,
-				      struct device_node *dp, u32 devhandle)
+static int __init pci_sun4v_pbm_init(struct pci_pbm_info *pbm,
+				     struct of_device *op, u32 devhandle)
 {
-	struct pci_pbm_info *pbm;
-
-	if (devhandle & 0x40)
-		pbm = &p->pbm_B;
-	else
-		pbm = &p->pbm_A;
-
-	pbm->next = pci_pbm_root;
-	pci_pbm_root = pbm;
+	struct device_node *dp = op->node;
+	int err;
 
 	pbm->numa_node = of_node_to_nid(dp);
 
-	pbm->scan_bus = pci_sun4v_scan_bus;
 	pbm->pci_ops = &sun4v_pci_ops;
 	pbm->config_space_reg_bits = 12;
 
 	pbm->index = pci_num_pbms++;
 
-	pbm->parent = p;
-	pbm->prom_node = dp;
+	pbm->op = op;
 
 	pbm->devhandle = devhandle;
 
@@ -924,82 +913,120 @@ static void __init pci_sun4v_pbm_init(struct pci_controller_info *p,
 	pci_determine_mem_io_space(pbm);
 
 	pci_get_pbm_props(pbm);
-	pci_sun4v_iommu_init(pbm);
+
+	err = pci_sun4v_iommu_init(pbm);
+	if (err)
+		return err;
+
 	pci_sun4v_msi_init(pbm);
+
+	pci_sun4v_scan_bus(pbm, &op->dev);
+
+	pbm->next = pci_pbm_root;
+	pci_pbm_root = pbm;
+
+	return 0;
 }
 
-void __init sun4v_pci_init(struct device_node *dp, char *model_name)
+static int __devinit pci_sun4v_probe(struct of_device *op,
+				     const struct of_device_id *match)
 {
+	const struct linux_prom64_registers *regs;
 	static int hvapi_negotiated = 0;
-	struct pci_controller_info *p;
 	struct pci_pbm_info *pbm;
+	struct device_node *dp;
 	struct iommu *iommu;
-	struct property *prop;
-	struct linux_prom64_registers *regs;
 	u32 devhandle;
-	int i;
+	int i, err;
+
+	dp = op->node;
 
 	if (!hvapi_negotiated++) {
-		int err = sun4v_hvapi_register(HV_GRP_PCI,
-					       vpci_major,
-					       &vpci_minor);
+		err = sun4v_hvapi_register(HV_GRP_PCI,
+					   vpci_major,
+					   &vpci_minor);
 
 		if (err) {
-			prom_printf("SUN4V_PCI: Could not register hvapi, "
-				    "err=%d\n", err);
-			prom_halt();
+			printk(KERN_ERR PFX "Could not register hvapi, "
+			       "err=%d\n", err);
+			return err;
 		}
-		printk("SUN4V_PCI: Registered hvapi major[%lu] minor[%lu]\n",
+		printk(KERN_INFO PFX "Registered hvapi major[%lu] minor[%lu]\n",
 		       vpci_major, vpci_minor);
 
 		dma_ops = &sun4v_dma_ops;
 	}
 
-	prop = of_find_property(dp, "reg", NULL);
-	if (!prop) {
-		prom_printf("SUN4V_PCI: Could not find config registers\n");
-		prom_halt();
+	regs = of_get_property(dp, "reg", NULL);
+	err = -ENODEV;
+	if (!regs) {
+		printk(KERN_ERR PFX "Could not find config registers\n");
+		goto out_err;
 	}
-	regs = prop->value;
-
 	devhandle = (regs->phys_addr >> 32UL) & 0x0fffffff;
 
-	for (pbm = pci_pbm_root; pbm; pbm = pbm->next) {
-		if (pbm->devhandle == (devhandle ^ 0x40)) {
-			pci_sun4v_pbm_init(pbm->parent, dp, devhandle);
-			return;
+	err = -ENOMEM;
+	if (!iommu_batch_initialized) {
+		for_each_possible_cpu(i) {
+			unsigned long page = get_zeroed_page(GFP_KERNEL);
+
+			if (!page)
+				goto out_err;
+
+			per_cpu(iommu_batch, i).pglist = (u64 *) page;
 		}
+		iommu_batch_initialized = 1;
 	}
 
-	for_each_possible_cpu(i) {
-		unsigned long page = get_zeroed_page(GFP_ATOMIC);
-
-		if (!page)
-			goto fatal_memory_error;
+	pbm = kzalloc(sizeof(*pbm), GFP_KERNEL);
+	if (!pbm) {
+		printk(KERN_ERR PFX "Could not allocate pci_pbm_info\n");
+		goto out_err;
+	}
 
-		per_cpu(iommu_batch, i).pglist = (u64 *) page;
+	iommu = kzalloc(sizeof(struct iommu), GFP_KERNEL);
+	if (!iommu) {
+		printk(KERN_ERR PFX "Could not allocate pbm iommu\n");
+		goto out_free_controller;
 	}
 
-	p = kzalloc(sizeof(struct pci_controller_info), GFP_ATOMIC);
-	if (!p)
-		goto fatal_memory_error;
+	pbm->iommu = iommu;
 
-	iommu = kzalloc(sizeof(struct iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
+	err = pci_sun4v_pbm_init(pbm, op, devhandle);
+	if (err)
+		goto out_free_iommu;
+
+	dev_set_drvdata(&op->dev, pbm);
+
+	return 0;
+
+out_free_iommu:
+	kfree(pbm->iommu);
 
-	p->pbm_A.iommu = iommu;
+out_free_controller:
+	kfree(pbm);
 
-	iommu = kzalloc(sizeof(struct iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
+out_err:
+	return err;
+}
 
-	p->pbm_B.iommu = iommu;
+static struct of_device_id __initdata pci_sun4v_match[] = {
+	{
+		.name = "pci",
+		.compatible = "SUNW,sun4v-pci",
+	},
+	{},
+};
 
-	pci_sun4v_pbm_init(p, dp, devhandle);
-	return;
+static struct of_platform_driver pci_sun4v_driver = {
+	.name		= DRIVER_NAME,
+	.match_table	= pci_sun4v_match,
+	.probe		= pci_sun4v_probe,
+};
 
-fatal_memory_error:
-	prom_printf("SUN4V_PCI: Fatal memory allocation error.\n");
-	prom_halt();
+static int __init pci_sun4v_init(void)
+{
+	return of_register_driver(&pci_sun4v_driver, &of_bus_type);
 }
+
+subsys_initcall(pci_sun4v_init);
diff --git a/arch/sparc64/kernel/pci_sun4v_asm.S b/arch/sparc64/kernel/pci_sun4v_asm.S
index ecb81f389b0..e606d46c681 100644
--- a/arch/sparc64/kernel/pci_sun4v_asm.S
+++ b/arch/sparc64/kernel/pci_sun4v_asm.S
@@ -1,8 +1,9 @@
 /* pci_sun4v_asm: Hypervisor calls for PCI support.
  *
- * Copyright (C) 2006 David S. Miller <davem@davemloft.net>
+ * Copyright (C) 2006, 2008 David S. Miller <davem@davemloft.net>
  */
 
+#include <linux/linkage.h>
 #include <asm/hypervisor.h>
 
 	/* %o0: devhandle
@@ -14,8 +15,7 @@
 	 * returns %o0:	-status if status was non-zero, else
 	 *         %o0:	num pages mapped
 	 */
-	.globl	pci_sun4v_iommu_map
-pci_sun4v_iommu_map:
+ENTRY(pci_sun4v_iommu_map)
 	mov	%o5, %g1
 	mov	HV_FAST_PCI_IOMMU_MAP, %o5
 	ta	HV_FAST_TRAP
@@ -24,6 +24,7 @@ pci_sun4v_iommu_map:
 	mov	%o1, %o0
 1:	retl
 	 nop
+ENDPROC(pci_sun4v_iommu_map)
 
 	/* %o0: devhandle
 	 * %o1:	tsbid
@@ -31,12 +32,12 @@ pci_sun4v_iommu_map:
 	 *
 	 * returns %o0:	num ttes demapped
 	 */
-	.globl	pci_sun4v_iommu_demap
-pci_sun4v_iommu_demap:
+ENTRY(pci_sun4v_iommu_demap)
 	mov	HV_FAST_PCI_IOMMU_DEMAP, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o1, %o0
+ENDPROC(pci_sun4v_iommu_demap)
 
 	/* %o0: devhandle
 	 * %o1:	tsbid
@@ -45,8 +46,7 @@ pci_sun4v_iommu_demap:
 	 *
 	 * returns %o0:	status
 	 */
-	.globl	pci_sun4v_iommu_getmap
-pci_sun4v_iommu_getmap:
+ENTRY(pci_sun4v_iommu_getmap)
 	mov	%o2, %o4
 	mov	HV_FAST_PCI_IOMMU_GETMAP, %o5
 	ta	HV_FAST_TRAP
@@ -54,6 +54,7 @@ pci_sun4v_iommu_getmap:
 	stx	%o2, [%o3]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_iommu_getmap)
 
 	/* %o0: devhandle
 	 * %o1:	pci_device
@@ -65,14 +66,14 @@ pci_sun4v_iommu_getmap:
 	 * If there is an error, the data will be returned
 	 * as all 1's.
 	 */
-	.globl	pci_sun4v_config_get
-pci_sun4v_config_get:
+ENTRY(pci_sun4v_config_get)
 	mov	HV_FAST_PCI_CONFIG_GET, %o5
 	ta	HV_FAST_TRAP
 	brnz,a,pn %o1, 1f
 	 mov	-1, %o2
 1:	retl
 	 mov	%o2, %o0
+ENDPROC(pci_sun4v_config_get)
 
 	/* %o0: devhandle
 	 * %o1:	pci_device
@@ -85,14 +86,14 @@ pci_sun4v_config_get:
 	 * status will be zero if the operation completed
 	 * successfully, else -1 if not
 	 */
-	.globl	pci_sun4v_config_put
-pci_sun4v_config_put:
+ENTRY(pci_sun4v_config_put)
 	mov	HV_FAST_PCI_CONFIG_PUT, %o5
 	ta	HV_FAST_TRAP
 	brnz,a,pn %o1, 1f
 	 mov	-1, %o1
 1:	retl
 	 mov	%o1, %o0
+ENDPROC(pci_sun4v_config_put)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -104,12 +105,12 @@ pci_sun4v_config_put:
 	 * status will be zero if the operation completed
 	 * successfully, else -1 if not
 	 */
-	.globl	pci_sun4v_msiq_conf
-pci_sun4v_msiq_conf:
+ENTRY(pci_sun4v_msiq_conf)
 	mov	HV_FAST_PCI_MSIQ_CONF, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_conf)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -118,8 +119,7 @@ pci_sun4v_msiq_conf:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_info
-pci_sun4v_msiq_info:
+ENTRY(pci_sun4v_msiq_info)
 	mov	%o2, %o4
 	mov	HV_FAST_PCI_MSIQ_INFO, %o5
 	ta	HV_FAST_TRAP
@@ -127,6 +127,7 @@ pci_sun4v_msiq_info:
 	stx	%o2, [%o3]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_info)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -134,13 +135,13 @@ pci_sun4v_msiq_info:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_getvalid
-pci_sun4v_msiq_getvalid:
+ENTRY(pci_sun4v_msiq_getvalid)
 	mov	HV_FAST_PCI_MSIQ_GETVALID, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_getvalid)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -148,12 +149,12 @@ pci_sun4v_msiq_getvalid:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_setvalid
-pci_sun4v_msiq_setvalid:
+ENTRY(pci_sun4v_msiq_setvalid)
 	mov	HV_FAST_PCI_MSIQ_SETVALID, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_setvalid)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -161,13 +162,13 @@ pci_sun4v_msiq_setvalid:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_getstate
-pci_sun4v_msiq_getstate:
+ENTRY(pci_sun4v_msiq_getstate)
 	mov	HV_FAST_PCI_MSIQ_GETSTATE, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_getstate)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -175,12 +176,12 @@ pci_sun4v_msiq_getstate:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_setstate
-pci_sun4v_msiq_setstate:
+ENTRY(pci_sun4v_msiq_setstate)
 	mov	HV_FAST_PCI_MSIQ_SETSTATE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_setstate)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -188,13 +189,13 @@ pci_sun4v_msiq_setstate:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_gethead
-pci_sun4v_msiq_gethead:
+ENTRY(pci_sun4v_msiq_gethead)
 	mov	HV_FAST_PCI_MSIQ_GETHEAD, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_gethead)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -202,12 +203,12 @@ pci_sun4v_msiq_gethead:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_sethead
-pci_sun4v_msiq_sethead:
+ENTRY(pci_sun4v_msiq_sethead)
 	mov	HV_FAST_PCI_MSIQ_SETHEAD, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_sethead)
 
 	/* %o0: devhandle
 	 * %o1: msiqid
@@ -215,13 +216,13 @@ pci_sun4v_msiq_sethead:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msiq_gettail
-pci_sun4v_msiq_gettail:
+ENTRY(pci_sun4v_msiq_gettail)
 	mov	HV_FAST_PCI_MSIQ_GETTAIL, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msiq_gettail)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -229,13 +230,13 @@ pci_sun4v_msiq_gettail:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msi_getvalid
-pci_sun4v_msi_getvalid:
+ENTRY(pci_sun4v_msi_getvalid)
 	mov	HV_FAST_PCI_MSI_GETVALID, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msi_getvalid)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -243,12 +244,12 @@ pci_sun4v_msi_getvalid:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msi_setvalid
-pci_sun4v_msi_setvalid:
+ENTRY(pci_sun4v_msi_setvalid)
 	mov	HV_FAST_PCI_MSI_SETVALID, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msi_setvalid)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -256,13 +257,13 @@ pci_sun4v_msi_setvalid:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msi_getmsiq
-pci_sun4v_msi_getmsiq:
+ENTRY(pci_sun4v_msi_getmsiq)
 	mov	HV_FAST_PCI_MSI_GETMSIQ, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msi_getmsiq)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -271,12 +272,12 @@ pci_sun4v_msi_getmsiq:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msi_setmsiq
-pci_sun4v_msi_setmsiq:
+ENTRY(pci_sun4v_msi_setmsiq)
 	mov	HV_FAST_PCI_MSI_SETMSIQ, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msi_setmsiq)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -284,13 +285,13 @@ pci_sun4v_msi_setmsiq:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msi_getstate
-pci_sun4v_msi_getstate:
+ENTRY(pci_sun4v_msi_getstate)
 	mov	HV_FAST_PCI_MSI_GETSTATE, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msi_getstate)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -298,12 +299,12 @@ pci_sun4v_msi_getstate:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msi_setstate
-pci_sun4v_msi_setstate:
+ENTRY(pci_sun4v_msi_setstate)
 	mov	HV_FAST_PCI_MSI_SETSTATE, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msi_setstate)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -311,13 +312,13 @@ pci_sun4v_msi_setstate:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msg_getmsiq
-pci_sun4v_msg_getmsiq:
+ENTRY(pci_sun4v_msg_getmsiq)
 	mov	HV_FAST_PCI_MSG_GETMSIQ, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msg_getmsiq)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -325,12 +326,12 @@ pci_sun4v_msg_getmsiq:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msg_setmsiq
-pci_sun4v_msg_setmsiq:
+ENTRY(pci_sun4v_msg_setmsiq)
 	mov	HV_FAST_PCI_MSG_SETMSIQ, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msg_setmsiq)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -338,13 +339,13 @@ pci_sun4v_msg_setmsiq:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msg_getvalid
-pci_sun4v_msg_getvalid:
+ENTRY(pci_sun4v_msg_getvalid)
 	mov	HV_FAST_PCI_MSG_GETVALID, %o5
 	ta	HV_FAST_TRAP
 	stx	%o1, [%o2]
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msg_getvalid)
 
 	/* %o0: devhandle
 	 * %o1: msinum
@@ -352,10 +353,10 @@ pci_sun4v_msg_getvalid:
 	 *
 	 * returns %o0: status
 	 */
-	.globl	pci_sun4v_msg_setvalid
-pci_sun4v_msg_setvalid:
+ENTRY(pci_sun4v_msg_setvalid)
 	mov	HV_FAST_PCI_MSG_SETVALID, %o5
 	ta	HV_FAST_TRAP
 	retl
 	 mov	%o0, %o0
+ENDPROC(pci_sun4v_msg_setvalid)
 
diff --git a/arch/sparc64/kernel/power.c b/arch/sparc64/kernel/power.c
index eae8ca2a6ba..076cad7f975 100644
--- a/arch/sparc64/kernel/power.c
+++ b/arch/sparc64/kernel/power.c
@@ -1,34 +1,17 @@
 /* power.c: Power management driver.
  *
- * Copyright (C) 1999, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1999, 2007, 2008 David S. Miller (davem@davemloft.net)
  */
 
 #include <linux/kernel.h>
 #include <linux/module.h>
 #include <linux/init.h>
-#include <linux/sched.h>
-#include <linux/signal.h>
-#include <linux/delay.h>
 #include <linux/interrupt.h>
-#include <linux/pm.h>
-#include <linux/syscalls.h>
 #include <linux/reboot.h>
+#include <linux/of_device.h>
 
-#include <asm/system.h>
-#include <asm/auxio.h>
 #include <asm/prom.h>
-#include <asm/of_device.h>
 #include <asm/io.h>
-#include <asm/sstate.h>
-#include <asm/reboot.h>
-
-#include <linux/unistd.h>
-
-/*
- * sysctl - toggle power-off restriction for serial console 
- * systems in machine_power_off()
- */
-int scons_pwroff = 1; 
 
 static void __iomem *power_reg;
 
@@ -40,31 +23,6 @@ static irqreturn_t power_handler(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static void (*poweroff_method)(void) = machine_alt_power_off;
-
-void machine_power_off(void)
-{
-	sstate_poweroff();
-	if (strcmp(of_console_device->type, "serial") || scons_pwroff) {
-		if (power_reg) {
-			/* Both register bits seem to have the
-			 * same effect, so until I figure out
-			 * what the difference is...
-			 */
-			writel(AUXIO_PCIO_CPWR_OFF | AUXIO_PCIO_SPWR_OFF, power_reg);
-		} else {
-			if (poweroff_method != NULL) {
-				poweroff_method();
-				/* not reached */
-			}
-		}
-	}
-	machine_halt();
-}
-
-void (*pm_power_off)(void) = machine_power_off;
-EXPORT_SYMBOL(pm_power_off);
-
 static int __init has_button_interrupt(unsigned int irq, struct device_node *dp)
 {
 	if (irq == 0xffffffff)
@@ -85,8 +43,6 @@ static int __devinit power_probe(struct of_device *op, const struct of_device_id
 	printk(KERN_INFO "%s: Control reg at %lx\n",
 	       op->node->name, res->start);
 
-	poweroff_method = machine_halt;  /* able to use the standard halt */
-
 	if (has_button_interrupt(irq, op->node)) {
 		if (request_irq(irq,
 				power_handler, 0, "power", NULL) < 0)
@@ -96,7 +52,7 @@ static int __devinit power_probe(struct of_device *op, const struct of_device_id
 	return 0;
 }
 
-static struct of_device_id power_match[] = {
+static struct of_device_id __initdata power_match[] = {
 	{
 		.name = "power",
 	},
@@ -111,8 +67,9 @@ static struct of_platform_driver power_driver = {
 	},
 };
 
-void __init power_init(void)
+static int __init power_init(void)
 {
-	of_register_driver(&power_driver, &of_platform_bus_type);
-	return;
+	return of_register_driver(&power_driver, &of_platform_bus_type);
 }
+
+device_initcall(power_init);
diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c
index 2084f81a76e..d5e2acef987 100644
--- a/arch/sparc64/kernel/process.c
+++ b/arch/sparc64/kernel/process.c
@@ -15,7 +15,6 @@
 #include <linux/module.h>
 #include <linux/sched.h>
 #include <linux/kernel.h>
-#include <linux/kallsyms.h>
 #include <linux/mm.h>
 #include <linux/fs.h>
 #include <linux/smp.h>
@@ -23,7 +22,6 @@
 #include <linux/ptrace.h>
 #include <linux/slab.h>
 #include <linux/user.h>
-#include <linux/reboot.h>
 #include <linux/delay.h>
 #include <linux/compat.h>
 #include <linux/tick.h>
@@ -32,7 +30,6 @@
 #include <linux/elfcore.h>
 #include <linux/sysrq.h>
 
-#include <asm/oplib.h>
 #include <asm/uaccess.h>
 #include <asm/system.h>
 #include <asm/page.h>
@@ -47,13 +44,11 @@
 #include <asm/mmu_context.h>
 #include <asm/unistd.h>
 #include <asm/hypervisor.h>
-#include <asm/sstate.h>
-#include <asm/reboot.h>
 #include <asm/syscalls.h>
 #include <asm/irq_regs.h>
 #include <asm/smp.h>
 
-/* #define VERBOSE_SHOWREGS */
+#include "kstack.h"
 
 static void sparc64_yield(int cpu)
 {
@@ -97,7 +92,7 @@ void cpu_idle(void)
 	set_thread_flag(TIF_POLLING_NRFLAG);
 
 	while(1) {
-		tick_nohz_stop_sched_tick();
+		tick_nohz_stop_sched_tick(1);
 
 		while (!need_resched() && !cpu_is_offline(cpu))
 			sparc64_yield(cpu);
@@ -116,35 +111,6 @@ void cpu_idle(void)
 	}
 }
 
-void machine_halt(void)
-{
-	sstate_halt();
-	prom_halt();
-	panic("Halt failed!");
-}
-
-void machine_alt_power_off(void)
-{
-	sstate_poweroff();
-	prom_halt_power_off();
-	panic("Power-off failed!");
-}
-
-void machine_restart(char * cmd)
-{
-	char *p;
-	
-	sstate_reboot();
-	p = strchr (reboot_command, '\n');
-	if (p) *p = 0;
-	if (cmd)
-		prom_reboot(cmd);
-	if (*reboot_command)
-		prom_reboot(reboot_command);
-	prom_reboot("");
-	panic("Reboot failed!");
-}
-
 #ifdef CONFIG_COMPAT
 static void show_regwindow32(struct pt_regs *regs)
 {
@@ -211,28 +177,14 @@ static void show_regwindow(struct pt_regs *regs)
 	printk("i4: %016lx i5: %016lx i6: %016lx i7: %016lx\n",
 	       rwk->ins[4], rwk->ins[5], rwk->ins[6], rwk->ins[7]);
 	if (regs->tstate & TSTATE_PRIV)
-		print_symbol("I7: <%s>\n", rwk->ins[7]);
+		printk("I7: <%pS>\n", (void *) rwk->ins[7]);
 }
 
-#ifdef CONFIG_SMP
-static DEFINE_SPINLOCK(regdump_lock);
-#endif
-
-void __show_regs(struct pt_regs * regs)
+void show_regs(struct pt_regs *regs)
 {
-#ifdef CONFIG_SMP
-	unsigned long flags;
-
-	/* Protect against xcall ipis which might lead to livelock on the lock */
-	__asm__ __volatile__("rdpr      %%pstate, %0\n\t"
-			     "wrpr      %0, %1, %%pstate"
-			     : "=r" (flags)
-			     : "i" (PSTATE_IE));
-	spin_lock(&regdump_lock);
-#endif
 	printk("TSTATE: %016lx TPC: %016lx TNPC: %016lx Y: %08x    %s\n", regs->tstate,
 	       regs->tpc, regs->tnpc, regs->y, print_tainted());
-	print_symbol("TPC: <%s>\n", regs->tpc);
+	printk("TPC: <%pS>\n", (void *) regs->tpc);
 	printk("g0: %016lx g1: %016lx g2: %016lx g3: %016lx\n",
 	       regs->u_regs[0], regs->u_regs[1], regs->u_regs[2],
 	       regs->u_regs[3]);
@@ -245,63 +197,10 @@ void __show_regs(struct pt_regs * regs)
 	printk("o4: %016lx o5: %016lx sp: %016lx ret_pc: %016lx\n",
 	       regs->u_regs[12], regs->u_regs[13], regs->u_regs[14],
 	       regs->u_regs[15]);
-	print_symbol("RPC: <%s>\n", regs->u_regs[15]);
+	printk("RPC: <%pS>\n", (void *) regs->u_regs[15]);
 	show_regwindow(regs);
-#ifdef CONFIG_SMP
-	spin_unlock(&regdump_lock);
-	__asm__ __volatile__("wrpr	%0, 0, %%pstate"
-			     : : "r" (flags));
-#endif
-}
-
-#ifdef VERBOSE_SHOWREGS
-static void idump_from_user (unsigned int *pc)
-{
-	int i;
-	int code;
-	
-	if((((unsigned long) pc) & 3))
-		return;
-	
-	pc -= 3;
-	for(i = -3; i < 6; i++) {
-		get_user(code, pc);
-		printk("%c%08x%c",i?' ':'<',code,i?' ':'>');
-		pc++;
-	}
-	printk("\n");
-}
-#endif
-
-void show_regs(struct pt_regs *regs)
-{
-#ifdef VERBOSE_SHOWREGS
-	extern long etrap, etraptl1;
-#endif
-	__show_regs(regs);
-#if 0
-#ifdef CONFIG_SMP
-	{
-		extern void smp_report_regs(void);
-
-		smp_report_regs();
-	}
-#endif
-#endif
-
-#ifdef VERBOSE_SHOWREGS	
-	if (regs->tpc >= &etrap && regs->tpc < &etraptl1 &&
-	    regs->u_regs[14] >= (long)current - PAGE_SIZE &&
-	    regs->u_regs[14] < (long)current + 6 * PAGE_SIZE) {
-		printk ("*********parent**********\n");
-		__show_regs((struct pt_regs *)(regs->u_regs[14] + PTREGS_OFF));
-		idump_from_user(((struct pt_regs *)(regs->u_regs[14] + PTREGS_OFF))->tpc);
-		printk ("*********endpar**********\n");
-	}
-#endif
 }
 
-#ifdef CONFIG_MAGIC_SYSRQ
 struct global_reg_snapshot global_reg_snapshot[NR_CPUS];
 static DEFINE_SPINLOCK(global_reg_snapshot_lock);
 
@@ -320,10 +219,17 @@ static void __global_reg_self(struct thread_info *tp, struct pt_regs *regs,
 
 		rw = (struct reg_window *)
 			(regs->u_regs[UREG_FP] + STACK_BIAS);
-		global_reg_snapshot[this_cpu].i7 = rw->ins[6];
-	} else
+		if (kstack_valid(tp, (unsigned long) rw)) {
+			global_reg_snapshot[this_cpu].i7 = rw->ins[7];
+			rw = (struct reg_window *)
+				(rw->ins[6] + STACK_BIAS);
+			if (kstack_valid(tp, (unsigned long) rw))
+				global_reg_snapshot[this_cpu].rpc = rw->ins[7];
+		}
+	} else {
 		global_reg_snapshot[this_cpu].i7 = 0;
-
+		global_reg_snapshot[this_cpu].rpc = 0;
+	}
 	global_reg_snapshot[this_cpu].thread = tp;
 }
 
@@ -342,13 +248,10 @@ static void __global_reg_poll(struct global_reg_snapshot *gp)
 	}
 }
 
-static void sysrq_handle_globreg(int key, struct tty_struct *tty)
+void __trigger_all_cpu_backtrace(void)
 {
 	struct thread_info *tp = current_thread_info();
 	struct pt_regs *regs = get_irq_regs();
-#ifdef CONFIG_KALLSYMS
-	char buffer[KSYM_SYMBOL_LEN];
-#endif
 	unsigned long flags;
 	int this_cpu, cpu;
 
@@ -367,7 +270,6 @@ static void sysrq_handle_globreg(int key, struct tty_struct *tty)
 
 	for_each_online_cpu(cpu) {
 		struct global_reg_snapshot *gp = &global_reg_snapshot[cpu];
-		struct thread_info *tp;
 
 		__global_reg_poll(gp);
 
@@ -377,19 +279,16 @@ static void sysrq_handle_globreg(int key, struct tty_struct *tty)
 		       gp->tstate, gp->tpc, gp->tnpc,
 		       ((tp && tp->task) ? tp->task->comm : "NULL"),
 		       ((tp && tp->task) ? tp->task->pid : -1));
-#ifdef CONFIG_KALLSYMS
+
 		if (gp->tstate & TSTATE_PRIV) {
-			sprint_symbol(buffer, gp->tpc);
-			printk("             TPC[%s] ", buffer);
-			sprint_symbol(buffer, gp->o7);
-			printk("O7[%s] ", buffer);
-			sprint_symbol(buffer, gp->i7);
-			printk("I7[%s]\n", buffer);
-		} else
-#endif
-		{
-			printk("             TPC[%lx] O7[%lx] I7[%lx]\n",
-			       gp->tpc, gp->o7, gp->i7);
+			printk("             TPC[%pS] O7[%pS] I7[%pS] RPC[%pS]\n",
+			       (void *) gp->tpc,
+			       (void *) gp->o7,
+			       (void *) gp->i7,
+			       (void *) gp->rpc);
+		} else {
+			printk("             TPC[%lx] O7[%lx] I7[%lx] RPC[%lx]\n",
+			       gp->tpc, gp->o7, gp->i7, gp->rpc);
 		}
 	}
 
@@ -398,6 +297,13 @@ static void sysrq_handle_globreg(int key, struct tty_struct *tty)
 	spin_unlock_irqrestore(&global_reg_snapshot_lock, flags);
 }
 
+#ifdef CONFIG_MAGIC_SYSRQ
+
+static void sysrq_handle_globreg(int key, struct tty_struct *tty)
+{
+	__trigger_all_cpu_backtrace();
+}
+
 static struct sysrq_key_op sparc_globalreg_op = {
 	.handler	= sysrq_handle_globreg,
 	.help_msg	= "Globalregs",
@@ -691,9 +597,9 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
 		  ((unsigned long) child_sf) - STACK_BIAS;
 
 		/* Special case, if we are spawning a kernel thread from
-		 * a userspace task (via KMOD, NFS, or similar) we must
-		 * disable performance counters in the child because the
-		 * address space and protection realm are changing.
+		 * a userspace task (usermode helper, NFS or similar), we
+		 * must disable performance counters in the child because
+		 * the address space and protection realm are changing.
 		 */
 		if (t->flags & _TIF_PERFCTR) {
 			t->user_cntd0 = t->user_cntd1 = NULL;
@@ -876,7 +782,7 @@ out:
 unsigned long get_wchan(struct task_struct *task)
 {
 	unsigned long pc, fp, bias = 0;
-	unsigned long thread_info_base;
+	struct thread_info *tp;
 	struct reg_window *rw;
         unsigned long ret = 0;
 	int count = 0; 
@@ -885,14 +791,12 @@ unsigned long get_wchan(struct task_struct *task)
             task->state == TASK_RUNNING)
 		goto out;
 
-	thread_info_base = (unsigned long) task_stack_page(task);
+	tp = task_thread_info(task);
 	bias = STACK_BIAS;
 	fp = task_thread_info(task)->ksp + bias;
 
 	do {
-		/* Bogus frame pointer? */
-		if (fp < (thread_info_base + sizeof(struct thread_info)) ||
-		    fp >= (thread_info_base + THREAD_SIZE))
+		if (!kstack_valid(tp, fp))
 			break;
 		rw = (struct reg_window *) fp;
 		pc = rw->ins[7];
diff --git a/arch/sparc64/kernel/prom.c b/arch/sparc64/kernel/prom.c
index ed03a18d3b3..dbba82f9b14 100644
--- a/arch/sparc64/kernel/prom.c
+++ b/arch/sparc64/kernel/prom.c
@@ -21,9 +21,9 @@
 #include <linux/mm.h>
 #include <linux/module.h>
 #include <linux/lmb.h>
+#include <linux/of_device.h>
 
 #include <asm/prom.h>
-#include <asm/of_device.h>
 #include <asm/oplib.h>
 #include <asm/irq.h>
 #include <asm/asi.h>
@@ -38,7 +38,7 @@ struct device_node *of_find_node_by_phandle(phandle handle)
 {
 	struct device_node *np;
 
-	for (np = allnodes; np != 0; np = np->allnext)
+	for (np = allnodes; np; np = np->allnext)
 		if (np->node == handle)
 			break;
 
@@ -59,6 +59,9 @@ int of_getintprop_default(struct device_node *np, const char *name, int def)
 }
 EXPORT_SYMBOL(of_getintprop_default);
 
+DEFINE_MUTEX(of_set_property_mutex);
+EXPORT_SYMBOL(of_set_property_mutex);
+
 int of_set_property(struct device_node *dp, const char *name, void *val, int len)
 {
 	struct property **prevp;
@@ -82,7 +85,10 @@ int of_set_property(struct device_node *dp, const char *name, void *val, int len
 			void *old_val = prop->value;
 			int ret;
 
+			mutex_lock(&of_set_property_mutex);
 			ret = prom_setprop(dp->node, name, val, len);
+			mutex_unlock(&of_set_property_mutex);
+
 			err = -EINVAL;
 			if (ret >= 0) {
 				prop->value = new_val;
@@ -156,55 +162,11 @@ static unsigned long psycho_pcislot_imap_offset(unsigned long ino)
 		return PSYCHO_IMAP_B_SLOT0 + (slot * 8);
 }
 
-#define PSYCHO_IMAP_SCSI	0x1000UL
-#define PSYCHO_IMAP_ETH		0x1008UL
-#define PSYCHO_IMAP_BPP		0x1010UL
-#define PSYCHO_IMAP_AU_REC	0x1018UL
-#define PSYCHO_IMAP_AU_PLAY	0x1020UL
-#define PSYCHO_IMAP_PFAIL	0x1028UL
-#define PSYCHO_IMAP_KMS		0x1030UL
-#define PSYCHO_IMAP_FLPY	0x1038UL
-#define PSYCHO_IMAP_SHW		0x1040UL
-#define PSYCHO_IMAP_KBD		0x1048UL
-#define PSYCHO_IMAP_MS		0x1050UL
-#define PSYCHO_IMAP_SER		0x1058UL
-#define PSYCHO_IMAP_TIM0	0x1060UL
-#define PSYCHO_IMAP_TIM1	0x1068UL
-#define PSYCHO_IMAP_UE		0x1070UL
-#define PSYCHO_IMAP_CE		0x1078UL
-#define PSYCHO_IMAP_A_ERR	0x1080UL
-#define PSYCHO_IMAP_B_ERR	0x1088UL
-#define PSYCHO_IMAP_PMGMT	0x1090UL
-#define PSYCHO_IMAP_GFX		0x1098UL
-#define PSYCHO_IMAP_EUPA	0x10a0UL
-
-static unsigned long __psycho_onboard_imap_off[] = {
-/*0x20*/	PSYCHO_IMAP_SCSI,
-/*0x21*/	PSYCHO_IMAP_ETH,
-/*0x22*/	PSYCHO_IMAP_BPP,
-/*0x23*/	PSYCHO_IMAP_AU_REC,
-/*0x24*/	PSYCHO_IMAP_AU_PLAY,
-/*0x25*/	PSYCHO_IMAP_PFAIL,
-/*0x26*/	PSYCHO_IMAP_KMS,
-/*0x27*/	PSYCHO_IMAP_FLPY,
-/*0x28*/	PSYCHO_IMAP_SHW,
-/*0x29*/	PSYCHO_IMAP_KBD,
-/*0x2a*/	PSYCHO_IMAP_MS,
-/*0x2b*/	PSYCHO_IMAP_SER,
-/*0x2c*/	PSYCHO_IMAP_TIM0,
-/*0x2d*/	PSYCHO_IMAP_TIM1,
-/*0x2e*/	PSYCHO_IMAP_UE,
-/*0x2f*/	PSYCHO_IMAP_CE,
-/*0x30*/	PSYCHO_IMAP_A_ERR,
-/*0x31*/	PSYCHO_IMAP_B_ERR,
-/*0x32*/	PSYCHO_IMAP_PMGMT,
-/*0x33*/	PSYCHO_IMAP_GFX,
-/*0x34*/	PSYCHO_IMAP_EUPA,
-};
+#define PSYCHO_OBIO_IMAP_BASE	0x1000UL
+
 #define PSYCHO_ONBOARD_IRQ_BASE		0x20
-#define PSYCHO_ONBOARD_IRQ_LAST		0x34
 #define psycho_onboard_imap_offset(__ino) \
-	__psycho_onboard_imap_off[(__ino) - PSYCHO_ONBOARD_IRQ_BASE]
+	(PSYCHO_OBIO_IMAP_BASE + (((__ino) & 0x1f) << 3))
 
 #define PSYCHO_ICLR_A_SLOT0	0x1400UL
 #define PSYCHO_ICLR_SCSI	0x1800UL
@@ -228,10 +190,6 @@ static unsigned int psycho_irq_build(struct device_node *dp,
 		imap_off = psycho_pcislot_imap_offset(ino);
 	} else {
 		/* Onboard device */
-		if (ino > PSYCHO_ONBOARD_IRQ_LAST) {
-			prom_printf("psycho_irq_build: Wacky INO [%x]\n", ino);
-			prom_halt();
-		}
 		imap_off = psycho_onboard_imap_offset(ino);
 	}
 
@@ -318,23 +276,6 @@ static void sabre_wsync_handler(unsigned int ino, void *_arg1, void *_arg2)
 
 #define SABRE_IMAP_A_SLOT0	0x0c00UL
 #define SABRE_IMAP_B_SLOT0	0x0c20UL
-#define SABRE_IMAP_SCSI		0x1000UL
-#define SABRE_IMAP_ETH		0x1008UL
-#define SABRE_IMAP_BPP		0x1010UL
-#define SABRE_IMAP_AU_REC	0x1018UL
-#define SABRE_IMAP_AU_PLAY	0x1020UL
-#define SABRE_IMAP_PFAIL	0x1028UL
-#define SABRE_IMAP_KMS		0x1030UL
-#define SABRE_IMAP_FLPY		0x1038UL
-#define SABRE_IMAP_SHW		0x1040UL
-#define SABRE_IMAP_KBD		0x1048UL
-#define SABRE_IMAP_MS		0x1050UL
-#define SABRE_IMAP_SER		0x1058UL
-#define SABRE_IMAP_UE		0x1070UL
-#define SABRE_IMAP_CE		0x1078UL
-#define SABRE_IMAP_PCIERR	0x1080UL
-#define SABRE_IMAP_GFX		0x1098UL
-#define SABRE_IMAP_EUPA		0x10a0UL
 #define SABRE_ICLR_A_SLOT0	0x1400UL
 #define SABRE_ICLR_B_SLOT0	0x1480UL
 #define SABRE_ICLR_SCSI		0x1800UL
@@ -364,33 +305,10 @@ static unsigned long sabre_pcislot_imap_offset(unsigned long ino)
 		return SABRE_IMAP_B_SLOT0 + (slot * 8);
 }
 
-static unsigned long __sabre_onboard_imap_off[] = {
-/*0x20*/	SABRE_IMAP_SCSI,
-/*0x21*/	SABRE_IMAP_ETH,
-/*0x22*/	SABRE_IMAP_BPP,
-/*0x23*/	SABRE_IMAP_AU_REC,
-/*0x24*/	SABRE_IMAP_AU_PLAY,
-/*0x25*/	SABRE_IMAP_PFAIL,
-/*0x26*/	SABRE_IMAP_KMS,
-/*0x27*/	SABRE_IMAP_FLPY,
-/*0x28*/	SABRE_IMAP_SHW,
-/*0x29*/	SABRE_IMAP_KBD,
-/*0x2a*/	SABRE_IMAP_MS,
-/*0x2b*/	SABRE_IMAP_SER,
-/*0x2c*/	0 /* reserved */,
-/*0x2d*/	0 /* reserved */,
-/*0x2e*/	SABRE_IMAP_UE,
-/*0x2f*/	SABRE_IMAP_CE,
-/*0x30*/	SABRE_IMAP_PCIERR,
-/*0x31*/	0 /* reserved */,
-/*0x32*/	0 /* reserved */,
-/*0x33*/	SABRE_IMAP_GFX,
-/*0x34*/	SABRE_IMAP_EUPA,
-};
-#define SABRE_ONBOARD_IRQ_BASE		0x20
-#define SABRE_ONBOARD_IRQ_LAST		0x30
+#define SABRE_OBIO_IMAP_BASE	0x1000UL
+#define SABRE_ONBOARD_IRQ_BASE	0x20
 #define sabre_onboard_imap_offset(__ino) \
-	__sabre_onboard_imap_off[(__ino) - SABRE_ONBOARD_IRQ_BASE]
+	(SABRE_OBIO_IMAP_BASE + (((__ino) & 0x1f) << 3))
 
 #define sabre_iclr_offset(ino)					      \
 	((ino & 0x20) ? (SABRE_ICLR_SCSI + (((ino) & 0x1f) << 3)) :  \
@@ -453,10 +371,6 @@ static unsigned int sabre_irq_build(struct device_node *dp,
 		imap_off = sabre_pcislot_imap_offset(ino);
 	} else {
 		/* onboard device */
-		if (ino > SABRE_ONBOARD_IRQ_LAST) {
-			prom_printf("sabre_irq_build: Wacky INO [%x]\n", ino);
-			prom_halt();
-		}
 		imap_off = sabre_onboard_imap_offset(ino);
 	}
 
@@ -1037,22 +951,30 @@ static void __init irq_trans_init(struct device_node *dp)
 		for (i = 0; i < ARRAY_SIZE(pci_irq_trans_table); i++) {
 			struct irq_trans *t = &pci_irq_trans_table[i];
 
-			if (!strcmp(model, t->name))
-				return t->init(dp);
+			if (!strcmp(model, t->name)) {
+				t->init(dp);
+				return;
+			}
 		}
 	}
 #endif
 #ifdef CONFIG_SBUS
 	if (!strcmp(dp->name, "sbus") ||
-	    !strcmp(dp->name, "sbi"))
-		return sbus_irq_trans_init(dp);
+	    !strcmp(dp->name, "sbi")) {
+		sbus_irq_trans_init(dp);
+		return;
+	}
 #endif
 	if (!strcmp(dp->name, "fhc") &&
-	    !strcmp(dp->parent->name, "central"))
-		return central_irq_trans_init(dp);
+	    !strcmp(dp->parent->name, "central")) {
+		central_irq_trans_init(dp);
+		return;
+	}
 	if (!strcmp(dp->name, "virtual-devices") ||
-	    !strcmp(dp->name, "niu"))
-		return sun4v_vdev_irq_trans_init(dp);
+	    !strcmp(dp->name, "niu")) {
+		sun4v_vdev_irq_trans_init(dp);
+		return;
+	}
 }
 
 static int is_root_node(const struct device_node *dp)
@@ -1323,32 +1245,49 @@ static void __init __build_path_component(struct device_node *dp, char *tmp_buf)
 
 	if (parent != NULL) {
 		if (!strcmp(parent->type, "pci") ||
-		    !strcmp(parent->type, "pciex"))
-			return pci_path_component(dp, tmp_buf);
-		if (!strcmp(parent->type, "sbus"))
-			return sbus_path_component(dp, tmp_buf);
-		if (!strcmp(parent->type, "upa"))
-			return upa_path_component(dp, tmp_buf);
-		if (!strcmp(parent->type, "ebus"))
-			return ebus_path_component(dp, tmp_buf);
+		    !strcmp(parent->type, "pciex")) {
+			pci_path_component(dp, tmp_buf);
+			return;
+		}
+		if (!strcmp(parent->type, "sbus")) {
+			sbus_path_component(dp, tmp_buf);
+			return;
+		}
+		if (!strcmp(parent->type, "upa")) {
+			upa_path_component(dp, tmp_buf);
+			return;
+		}
+		if (!strcmp(parent->type, "ebus")) {
+			ebus_path_component(dp, tmp_buf);
+			return;
+		}
 		if (!strcmp(parent->name, "usb") ||
-		    !strcmp(parent->name, "hub"))
-			return usb_path_component(dp, tmp_buf);
-		if (!strcmp(parent->type, "i2c"))
-			return i2c_path_component(dp, tmp_buf);
-		if (!strcmp(parent->type, "firewire"))
-			return ieee1394_path_component(dp, tmp_buf);
-		if (!strcmp(parent->type, "virtual-devices"))
-			return vdev_path_component(dp, tmp_buf);
-
+		    !strcmp(parent->name, "hub")) {
+			usb_path_component(dp, tmp_buf);
+			return;
+		}
+		if (!strcmp(parent->type, "i2c")) {
+			i2c_path_component(dp, tmp_buf);
+			return;
+		}
+		if (!strcmp(parent->type, "firewire")) {
+			ieee1394_path_component(dp, tmp_buf);
+			return;
+		}
+		if (!strcmp(parent->type, "virtual-devices")) {
+			vdev_path_component(dp, tmp_buf);
+			return;
+		}
 		/* "isa" is handled with platform naming */
 	}
 
 	/* Use platform naming convention.  */
-	if (tlb_type == hypervisor)
-		return sun4v_path_component(dp, tmp_buf);
-	else
-		return sun4u_path_component(dp, tmp_buf);
+	if (tlb_type == hypervisor) {
+		sun4v_path_component(dp, tmp_buf);
+		return;
+	} else {
+		sun4u_path_component(dp, tmp_buf);
+	}
 }
 
 static char * __init build_path_component(struct device_node *dp)
diff --git a/arch/sparc64/kernel/psycho_common.c b/arch/sparc64/kernel/psycho_common.c
new file mode 100644
index 00000000000..790996428c1
--- /dev/null
+++ b/arch/sparc64/kernel/psycho_common.c
@@ -0,0 +1,470 @@
+/* psycho_common.c: Code common to PSYCHO and derivative PCI controllers.
+ *
+ * Copyright (C) 2008 David S. Miller <davem@davemloft.net>
+ */
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+
+#include <asm/upa.h>
+
+#include "pci_impl.h"
+#include "iommu_common.h"
+#include "psycho_common.h"
+
+#define  PSYCHO_STRBUF_CTRL_DENAB	0x0000000000000002UL
+#define  PSYCHO_STCERR_WRITE		0x0000000000000002UL
+#define  PSYCHO_STCERR_READ		0x0000000000000001UL
+#define  PSYCHO_STCTAG_PPN		0x0fffffff00000000UL
+#define  PSYCHO_STCTAG_VPN		0x00000000ffffe000UL
+#define  PSYCHO_STCTAG_VALID		0x0000000000000002UL
+#define  PSYCHO_STCTAG_WRITE		0x0000000000000001UL
+#define  PSYCHO_STCLINE_LINDX		0x0000000001e00000UL
+#define  PSYCHO_STCLINE_SPTR		0x00000000001f8000UL
+#define  PSYCHO_STCLINE_LADDR		0x0000000000007f00UL
+#define  PSYCHO_STCLINE_EPTR		0x00000000000000fcUL
+#define  PSYCHO_STCLINE_VALID		0x0000000000000002UL
+#define  PSYCHO_STCLINE_FOFN		0x0000000000000001UL
+
+static DEFINE_SPINLOCK(stc_buf_lock);
+static unsigned long stc_error_buf[128];
+static unsigned long stc_tag_buf[16];
+static unsigned long stc_line_buf[16];
+
+static void psycho_check_stc_error(struct pci_pbm_info *pbm)
+{
+	unsigned long err_base, tag_base, line_base;
+	struct strbuf *strbuf = &pbm->stc;
+	u64 control;
+	int i;
+
+	if (!strbuf->strbuf_control)
+		return;
+
+	err_base = strbuf->strbuf_err_stat;
+	tag_base = strbuf->strbuf_tag_diag;
+	line_base = strbuf->strbuf_line_diag;
+
+	spin_lock(&stc_buf_lock);
+
+	/* This is __REALLY__ dangerous.  When we put the streaming
+	 * buffer into diagnostic mode to probe it's tags and error
+	 * status, we _must_ clear all of the line tag valid bits
+	 * before re-enabling the streaming buffer.  If any dirty data
+	 * lives in the STC when we do this, we will end up
+	 * invalidating it before it has a chance to reach main
+	 * memory.
+	 */
+	control = upa_readq(strbuf->strbuf_control);
+	upa_writeq(control | PSYCHO_STRBUF_CTRL_DENAB, strbuf->strbuf_control);
+	for (i = 0; i < 128; i++) {
+		u64 val;
+
+		val = upa_readq(err_base + (i * 8UL));
+		upa_writeq(0UL, err_base + (i * 8UL));
+		stc_error_buf[i] = val;
+	}
+	for (i = 0; i < 16; i++) {
+		stc_tag_buf[i] = upa_readq(tag_base + (i * 8UL));
+		stc_line_buf[i] = upa_readq(line_base + (i * 8UL));
+		upa_writeq(0UL, tag_base + (i * 8UL));
+		upa_writeq(0UL, line_base + (i * 8UL));
+	}
+
+	/* OK, state is logged, exit diagnostic mode. */
+	upa_writeq(control, strbuf->strbuf_control);
+
+	for (i = 0; i < 16; i++) {
+		int j, saw_error, first, last;
+
+		saw_error = 0;
+		first = i * 8;
+		last = first + 8;
+		for (j = first; j < last; j++) {
+			u64 errval = stc_error_buf[j];
+			if (errval != 0) {
+				saw_error++;
+				printk(KERN_ERR "%s: STC_ERR(%d)[wr(%d)"
+				       "rd(%d)]\n",
+				       pbm->name,
+				       j,
+				       (errval & PSYCHO_STCERR_WRITE) ? 1 : 0,
+				       (errval & PSYCHO_STCERR_READ) ? 1 : 0);
+			}
+		}
+		if (saw_error != 0) {
+			u64 tagval = stc_tag_buf[i];
+			u64 lineval = stc_line_buf[i];
+			printk(KERN_ERR "%s: STC_TAG(%d)[PA(%016lx)VA(%08lx)"
+			       "V(%d)W(%d)]\n",
+			       pbm->name,
+			       i,
+			       ((tagval & PSYCHO_STCTAG_PPN) >> 19UL),
+			       (tagval & PSYCHO_STCTAG_VPN),
+			       ((tagval & PSYCHO_STCTAG_VALID) ? 1 : 0),
+			       ((tagval & PSYCHO_STCTAG_WRITE) ? 1 : 0));
+			printk(KERN_ERR "%s: STC_LINE(%d)[LIDX(%lx)SP(%lx)"
+			       "LADDR(%lx)EP(%lx)V(%d)FOFN(%d)]\n",
+			       pbm->name,
+			       i,
+			       ((lineval & PSYCHO_STCLINE_LINDX) >> 21UL),
+			       ((lineval & PSYCHO_STCLINE_SPTR) >> 15UL),
+			       ((lineval & PSYCHO_STCLINE_LADDR) >> 8UL),
+			       ((lineval & PSYCHO_STCLINE_EPTR) >> 2UL),
+			       ((lineval & PSYCHO_STCLINE_VALID) ? 1 : 0),
+			       ((lineval & PSYCHO_STCLINE_FOFN) ? 1 : 0));
+		}
+	}
+
+	spin_unlock(&stc_buf_lock);
+}
+
+#define PSYCHO_IOMMU_TAG		0xa580UL
+#define PSYCHO_IOMMU_DATA		0xa600UL
+
+static void psycho_record_iommu_tags_and_data(struct pci_pbm_info *pbm,
+					      u64 *tag, u64 *data)
+{
+	int i;
+
+	for (i = 0; i < 16; i++) {
+		unsigned long base = pbm->controller_regs;
+		unsigned long off = i * 8UL;
+
+		tag[i] = upa_readq(base + PSYCHO_IOMMU_TAG+off);
+		data[i] = upa_readq(base + PSYCHO_IOMMU_DATA+off);
+
+		/* Now clear out the entry. */
+		upa_writeq(0, base + PSYCHO_IOMMU_TAG + off);
+		upa_writeq(0, base + PSYCHO_IOMMU_DATA + off);
+	}
+}
+
+#define  PSYCHO_IOMMU_TAG_ERRSTS (0x3UL << 23UL)
+#define  PSYCHO_IOMMU_TAG_ERR	 (0x1UL << 22UL)
+#define  PSYCHO_IOMMU_TAG_WRITE	 (0x1UL << 21UL)
+#define  PSYCHO_IOMMU_TAG_STREAM (0x1UL << 20UL)
+#define  PSYCHO_IOMMU_TAG_SIZE	 (0x1UL << 19UL)
+#define  PSYCHO_IOMMU_TAG_VPAGE	 0x7ffffUL
+#define  PSYCHO_IOMMU_DATA_VALID (1UL << 30UL)
+#define  PSYCHO_IOMMU_DATA_CACHE (1UL << 28UL)
+#define  PSYCHO_IOMMU_DATA_PPAGE 0xfffffffUL
+
+static void psycho_dump_iommu_tags_and_data(struct pci_pbm_info *pbm,
+					    u64 *tag, u64 *data)
+{
+	int i;
+
+	for (i = 0; i < 16; i++) {
+		u64 tag_val, data_val;
+		const char *type_str;
+		tag_val = tag[i];
+		if (!(tag_val & PSYCHO_IOMMU_TAG_ERR))
+			continue;
+
+		data_val = data[i];
+		switch((tag_val & PSYCHO_IOMMU_TAG_ERRSTS) >> 23UL) {
+		case 0:
+			type_str = "Protection Error";
+			break;
+		case 1:
+			type_str = "Invalid Error";
+			break;
+		case 2:
+			type_str = "TimeOut Error";
+			break;
+		case 3:
+		default:
+			type_str = "ECC Error";
+			break;
+		}
+
+		printk(KERN_ERR "%s: IOMMU TAG(%d)[error(%s) wr(%d) "
+		       "str(%d) sz(%dK) vpg(%08lx)]\n",
+		       pbm->name, i, type_str,
+		       ((tag_val & PSYCHO_IOMMU_TAG_WRITE) ? 1 : 0),
+		       ((tag_val & PSYCHO_IOMMU_TAG_STREAM) ? 1 : 0),
+		       ((tag_val & PSYCHO_IOMMU_TAG_SIZE) ? 64 : 8),
+		       (tag_val & PSYCHO_IOMMU_TAG_VPAGE) << IOMMU_PAGE_SHIFT);
+		printk(KERN_ERR "%s: IOMMU DATA(%d)[valid(%d) cache(%d) "
+		       "ppg(%016lx)]\n",
+		       pbm->name, i,
+		       ((data_val & PSYCHO_IOMMU_DATA_VALID) ? 1 : 0),
+		       ((data_val & PSYCHO_IOMMU_DATA_CACHE) ? 1 : 0),
+		       (data_val & PSYCHO_IOMMU_DATA_PPAGE)<<IOMMU_PAGE_SHIFT);
+	}
+}
+
+#define  PSYCHO_IOMMU_CTRL_XLTESTAT	0x0000000006000000UL
+#define  PSYCHO_IOMMU_CTRL_XLTEERR	0x0000000001000000UL
+
+void psycho_check_iommu_error(struct pci_pbm_info *pbm,
+			      unsigned long afsr,
+			      unsigned long afar,
+			      enum psycho_error_type type)
+{
+	u64 control, iommu_tag[16], iommu_data[16];
+	struct iommu *iommu = pbm->iommu;
+	unsigned long flags;
+
+	spin_lock_irqsave(&iommu->lock, flags);
+	control = upa_readq(iommu->iommu_control);
+	if (control & PSYCHO_IOMMU_CTRL_XLTEERR) {
+		const char *type_str;
+
+		control &= ~PSYCHO_IOMMU_CTRL_XLTEERR;
+		upa_writeq(control, iommu->iommu_control);
+
+		switch ((control & PSYCHO_IOMMU_CTRL_XLTESTAT) >> 25UL) {
+		case 0:
+			type_str = "Protection Error";
+			break;
+		case 1:
+			type_str = "Invalid Error";
+			break;
+		case 2:
+			type_str = "TimeOut Error";
+			break;
+		case 3:
+		default:
+			type_str = "ECC Error";
+			break;
+		};
+		printk(KERN_ERR "%s: IOMMU Error, type[%s]\n",
+		       pbm->name, type_str);
+
+		/* It is very possible for another DVMA to occur while
+		 * we do this probe, and corrupt the system further.
+		 * But we are so screwed at this point that we are
+		 * likely to crash hard anyways, so get as much
+		 * diagnostic information to the console as we can.
+		 */
+		psycho_record_iommu_tags_and_data(pbm, iommu_tag, iommu_data);
+		psycho_dump_iommu_tags_and_data(pbm, iommu_tag, iommu_data);
+	}
+	psycho_check_stc_error(pbm);
+	spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+#define  PSYCHO_PCICTRL_SBH_ERR	 0x0000000800000000UL
+#define  PSYCHO_PCICTRL_SERR	 0x0000000400000000UL
+
+static irqreturn_t psycho_pcierr_intr_other(struct pci_pbm_info *pbm)
+{
+	irqreturn_t ret = IRQ_NONE;
+	u64 csr, csr_error_bits;
+	u16 stat, *addr;
+
+	csr = upa_readq(pbm->pci_csr);
+	csr_error_bits = csr & (PSYCHO_PCICTRL_SBH_ERR | PSYCHO_PCICTRL_SERR);
+	if (csr_error_bits) {
+		/* Clear the errors.  */
+		upa_writeq(csr, pbm->pci_csr);
+
+		/* Log 'em.  */
+		if (csr_error_bits & PSYCHO_PCICTRL_SBH_ERR)
+			printk(KERN_ERR "%s: PCI streaming byte hole "
+			       "error asserted.\n", pbm->name);
+		if (csr_error_bits & PSYCHO_PCICTRL_SERR)
+			printk(KERN_ERR "%s: PCI SERR signal asserted.\n",
+			       pbm->name);
+		ret = IRQ_HANDLED;
+	}
+	addr = psycho_pci_config_mkaddr(pbm, pbm->pci_first_busno,
+					0, PCI_STATUS);
+	pci_config_read16(addr, &stat);
+	if (stat & (PCI_STATUS_PARITY |
+		    PCI_STATUS_SIG_TARGET_ABORT |
+		    PCI_STATUS_REC_TARGET_ABORT |
+		    PCI_STATUS_REC_MASTER_ABORT |
+		    PCI_STATUS_SIG_SYSTEM_ERROR)) {
+		printk(KERN_ERR "%s: PCI bus error, PCI_STATUS[%04x]\n",
+		       pbm->name, stat);
+		pci_config_write16(addr, 0xffff);
+		ret = IRQ_HANDLED;
+	}
+	return ret;
+}
+
+#define  PSYCHO_PCIAFSR_PMA	0x8000000000000000UL
+#define  PSYCHO_PCIAFSR_PTA	0x4000000000000000UL
+#define  PSYCHO_PCIAFSR_PRTRY	0x2000000000000000UL
+#define  PSYCHO_PCIAFSR_PPERR	0x1000000000000000UL
+#define  PSYCHO_PCIAFSR_SMA	0x0800000000000000UL
+#define  PSYCHO_PCIAFSR_STA	0x0400000000000000UL
+#define  PSYCHO_PCIAFSR_SRTRY	0x0200000000000000UL
+#define  PSYCHO_PCIAFSR_SPERR	0x0100000000000000UL
+#define  PSYCHO_PCIAFSR_RESV1	0x00ff000000000000UL
+#define  PSYCHO_PCIAFSR_BMSK	0x0000ffff00000000UL
+#define  PSYCHO_PCIAFSR_BLK	0x0000000080000000UL
+#define  PSYCHO_PCIAFSR_RESV2	0x0000000040000000UL
+#define  PSYCHO_PCIAFSR_MID	0x000000003e000000UL
+#define  PSYCHO_PCIAFSR_RESV3	0x0000000001ffffffUL
+
+irqreturn_t psycho_pcierr_intr(int irq, void *dev_id)
+{
+	struct pci_pbm_info *pbm = dev_id;
+	u64 afsr, afar, error_bits;
+	int reported;
+
+	afsr = upa_readq(pbm->pci_afsr);
+	afar = upa_readq(pbm->pci_afar);
+	error_bits = afsr &
+		(PSYCHO_PCIAFSR_PMA | PSYCHO_PCIAFSR_PTA |
+		 PSYCHO_PCIAFSR_PRTRY | PSYCHO_PCIAFSR_PPERR |
+		 PSYCHO_PCIAFSR_SMA | PSYCHO_PCIAFSR_STA |
+		 PSYCHO_PCIAFSR_SRTRY | PSYCHO_PCIAFSR_SPERR);
+	if (!error_bits)
+		return psycho_pcierr_intr_other(pbm);
+	upa_writeq(error_bits, pbm->pci_afsr);
+	printk(KERN_ERR "%s: PCI Error, primary error type[%s]\n",
+	       pbm->name,
+	       (((error_bits & PSYCHO_PCIAFSR_PMA) ?
+		 "Master Abort" :
+		 ((error_bits & PSYCHO_PCIAFSR_PTA) ?
+		  "Target Abort" :
+		  ((error_bits & PSYCHO_PCIAFSR_PRTRY) ?
+		   "Excessive Retries" :
+		   ((error_bits & PSYCHO_PCIAFSR_PPERR) ?
+		    "Parity Error" : "???"))))));
+	printk(KERN_ERR "%s: bytemask[%04lx] UPA_MID[%02lx] was_block(%d)\n",
+	       pbm->name,
+	       (afsr & PSYCHO_PCIAFSR_BMSK) >> 32UL,
+	       (afsr & PSYCHO_PCIAFSR_MID) >> 25UL,
+	       (afsr & PSYCHO_PCIAFSR_BLK) ? 1 : 0);
+	printk(KERN_ERR "%s: PCI AFAR [%016lx]\n", pbm->name, afar);
+	printk(KERN_ERR "%s: PCI Secondary errors [", pbm->name);
+	reported = 0;
+	if (afsr & PSYCHO_PCIAFSR_SMA) {
+		reported++;
+		printk("(Master Abort)");
+	}
+	if (afsr & PSYCHO_PCIAFSR_STA) {
+		reported++;
+		printk("(Target Abort)");
+	}
+	if (afsr & PSYCHO_PCIAFSR_SRTRY) {
+		reported++;
+		printk("(Excessive Retries)");
+	}
+	if (afsr & PSYCHO_PCIAFSR_SPERR) {
+		reported++;
+		printk("(Parity Error)");
+	}
+	if (!reported)
+		printk("(none)");
+	printk("]\n");
+
+	if (error_bits & (PSYCHO_PCIAFSR_PTA | PSYCHO_PCIAFSR_STA)) {
+		psycho_check_iommu_error(pbm, afsr, afar, PCI_ERR);
+		pci_scan_for_target_abort(pbm, pbm->pci_bus);
+	}
+	if (error_bits & (PSYCHO_PCIAFSR_PMA | PSYCHO_PCIAFSR_SMA))
+		pci_scan_for_master_abort(pbm, pbm->pci_bus);
+
+	if (error_bits & (PSYCHO_PCIAFSR_PPERR | PSYCHO_PCIAFSR_SPERR))
+		pci_scan_for_parity_error(pbm, pbm->pci_bus);
+
+	return IRQ_HANDLED;
+}
+
+static void psycho_iommu_flush(struct pci_pbm_info *pbm)
+{
+	int i;
+
+	for (i = 0; i < 16; i++) {
+		unsigned long off = i * 8;
+
+		upa_writeq(0, pbm->controller_regs + PSYCHO_IOMMU_TAG + off);
+		upa_writeq(0, pbm->controller_regs + PSYCHO_IOMMU_DATA + off);
+	}
+}
+
+#define PSYCHO_IOMMU_CONTROL		0x0200UL
+#define  PSYCHO_IOMMU_CTRL_TSBSZ	0x0000000000070000UL
+#define  PSYCHO_IOMMU_TSBSZ_1K      	0x0000000000000000UL
+#define  PSYCHO_IOMMU_TSBSZ_2K      	0x0000000000010000UL
+#define  PSYCHO_IOMMU_TSBSZ_4K      	0x0000000000020000UL
+#define  PSYCHO_IOMMU_TSBSZ_8K      	0x0000000000030000UL
+#define  PSYCHO_IOMMU_TSBSZ_16K     	0x0000000000040000UL
+#define  PSYCHO_IOMMU_TSBSZ_32K     	0x0000000000050000UL
+#define  PSYCHO_IOMMU_TSBSZ_64K     	0x0000000000060000UL
+#define  PSYCHO_IOMMU_TSBSZ_128K    	0x0000000000070000UL
+#define  PSYCHO_IOMMU_CTRL_TBWSZ    	0x0000000000000004UL
+#define  PSYCHO_IOMMU_CTRL_DENAB    	0x0000000000000002UL
+#define  PSYCHO_IOMMU_CTRL_ENAB     	0x0000000000000001UL
+#define PSYCHO_IOMMU_FLUSH		0x0210UL
+#define PSYCHO_IOMMU_TSBBASE		0x0208UL
+
+int psycho_iommu_init(struct pci_pbm_info *pbm, int tsbsize,
+		      u32 dvma_offset, u32 dma_mask,
+		      unsigned long write_complete_offset)
+{
+	struct iommu *iommu = pbm->iommu;
+	u64 control;
+	int err;
+
+	iommu->iommu_control  = pbm->controller_regs + PSYCHO_IOMMU_CONTROL;
+	iommu->iommu_tsbbase  = pbm->controller_regs + PSYCHO_IOMMU_TSBBASE;
+	iommu->iommu_flush    = pbm->controller_regs + PSYCHO_IOMMU_FLUSH;
+	iommu->iommu_tags     = pbm->controller_regs + PSYCHO_IOMMU_TAG;
+	iommu->write_complete_reg = (pbm->controller_regs +
+				     write_complete_offset);
+
+	iommu->iommu_ctxflush = 0;
+
+	control = upa_readq(iommu->iommu_control);
+	control |= PSYCHO_IOMMU_CTRL_DENAB;
+	upa_writeq(control, iommu->iommu_control);
+
+	psycho_iommu_flush(pbm);
+
+	/* Leave diag mode enabled for full-flushing done in pci_iommu.c */
+	err = iommu_table_init(iommu, tsbsize * 1024 * 8,
+			       dvma_offset, dma_mask, pbm->numa_node);
+	if (err)
+		return err;
+
+	upa_writeq(__pa(iommu->page_table), iommu->iommu_tsbbase);
+
+	control = upa_readq(iommu->iommu_control);
+	control &= ~(PSYCHO_IOMMU_CTRL_TSBSZ | PSYCHO_IOMMU_CTRL_TBWSZ);
+	control |= PSYCHO_IOMMU_CTRL_ENAB;
+
+	switch (tsbsize) {
+	case 64:
+		control |= PSYCHO_IOMMU_TSBSZ_64K;
+		break;
+	case 128:
+		control |= PSYCHO_IOMMU_TSBSZ_128K;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	upa_writeq(control, iommu->iommu_control);
+
+	return 0;
+
+}
+
+void psycho_pbm_init_common(struct pci_pbm_info *pbm, struct of_device *op,
+			    const char *chip_name, int chip_type)
+{
+	struct device_node *dp = op->node;
+
+	pbm->name = dp->full_name;
+	pbm->numa_node = -1;
+	pbm->chip_type = chip_type;
+	pbm->chip_version = of_getintprop_default(dp, "version#", 0);
+	pbm->chip_revision = of_getintprop_default(dp, "module-revision#", 0);
+	pbm->op = op;
+	pbm->pci_ops = &sun4u_pci_ops;
+	pbm->config_space_reg_bits = 8;
+	pbm->index = pci_num_pbms++;
+	pci_get_pbm_props(pbm);
+	pci_determine_mem_io_space(pbm);
+
+	printk(KERN_INFO "%s: %s PCI Bus Module ver[%x:%x]\n",
+	       pbm->name, chip_name,
+	       pbm->chip_version, pbm->chip_revision);
+}
diff --git a/arch/sparc64/kernel/psycho_common.h b/arch/sparc64/kernel/psycho_common.h
new file mode 100644
index 00000000000..092c278ef28
--- /dev/null
+++ b/arch/sparc64/kernel/psycho_common.h
@@ -0,0 +1,48 @@
+#ifndef _PSYCHO_COMMON_H
+#define _PSYCHO_COMMON_H
+
+/* U2P Programmer's Manual, page 13-55, configuration space
+ * address format:
+ * 
+ *  32             24 23 16 15    11 10       8 7   2  1 0
+ * ---------------------------------------------------------
+ * |0 0 0 0 0 0 0 0 1| bus | device | function | reg | 0 0 |
+ * ---------------------------------------------------------
+ */
+#define PSYCHO_CONFIG_BASE(PBM)	\
+	((PBM)->config_space | (1UL << 24))
+#define PSYCHO_CONFIG_ENCODE(BUS, DEVFN, REG)	\
+	(((unsigned long)(BUS)   << 16) |	\
+	 ((unsigned long)(DEVFN) << 8)  |	\
+	 ((unsigned long)(REG)))
+
+static inline void *psycho_pci_config_mkaddr(struct pci_pbm_info *pbm,
+					     unsigned char bus,
+					     unsigned int devfn,
+					     int where)
+{
+	return (void *)
+		(PSYCHO_CONFIG_BASE(pbm) |
+		 PSYCHO_CONFIG_ENCODE(bus, devfn, where));
+}
+
+enum psycho_error_type {
+	UE_ERR, CE_ERR, PCI_ERR
+};
+
+extern void psycho_check_iommu_error(struct pci_pbm_info *pbm,
+				     unsigned long afsr,
+				     unsigned long afar,
+				     enum psycho_error_type type);
+
+extern irqreturn_t psycho_pcierr_intr(int irq, void *dev_id);
+
+extern int psycho_iommu_init(struct pci_pbm_info *pbm, int tsbsize,
+			     u32 dvma_offset, u32 dma_mask,
+			     unsigned long write_complete_offset);
+
+extern void psycho_pbm_init_common(struct pci_pbm_info *pbm,
+				   struct of_device *op,
+				   const char *chip_name, int chip_type);
+
+#endif /* _PSYCHO_COMMON_H */
diff --git a/arch/sparc64/kernel/ptrace.c b/arch/sparc64/kernel/ptrace.c
index f6c9fc92921..f43adbc773c 100644
--- a/arch/sparc64/kernel/ptrace.c
+++ b/arch/sparc64/kernel/ptrace.c
@@ -23,6 +23,7 @@
 #include <linux/audit.h>
 #include <linux/signal.h>
 #include <linux/regset.h>
+#include <linux/tracehook.h>
 #include <linux/compat.h>
 #include <linux/elf.h>
 
@@ -442,7 +443,7 @@ static const struct user_regset sparc64_regsets[] = {
 	 */
 	[REGSET_GENERAL] = {
 		.core_note_type = NT_PRSTATUS,
-		.n = 36 * sizeof(u64),
+		.n = 36,
 		.size = sizeof(u64), .align = sizeof(u64),
 		.get = genregs64_get, .set = genregs64_set
 	},
@@ -454,7 +455,7 @@ static const struct user_regset sparc64_regsets[] = {
 	 */
 	[REGSET_FP] = {
 		.core_note_type = NT_PRFPREG,
-		.n = 35 * sizeof(u64),
+		.n = 35,
 		.size = sizeof(u64), .align = sizeof(u64),
 		.get = fpregs64_get, .set = fpregs64_set
 	},
@@ -800,7 +801,7 @@ static const struct user_regset sparc32_regsets[] = {
 	 */
 	[REGSET_GENERAL] = {
 		.core_note_type = NT_PRSTATUS,
-		.n = 38 * sizeof(u32),
+		.n = 38,
 		.size = sizeof(u32), .align = sizeof(u32),
 		.get = genregs32_get, .set = genregs32_set
 	},
@@ -816,7 +817,7 @@ static const struct user_regset sparc32_regsets[] = {
 	 */
 	[REGSET_FP] = {
 		.core_note_type = NT_PRFPREG,
-		.n = 99 * sizeof(u32),
+		.n = 99,
 		.size = sizeof(u32), .align = sizeof(u32),
 		.get = fpregs32_get, .set = fpregs32_set
 	},
@@ -1049,42 +1050,17 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 	return ret;
 }
 
-asmlinkage void syscall_trace(struct pt_regs *regs, int syscall_exit_p)
+asmlinkage int syscall_trace_enter(struct pt_regs *regs)
 {
+	int ret = 0;
+
 	/* do the secure computing check first */
 	secure_computing(regs->u_regs[UREG_G1]);
 
-	if (unlikely(current->audit_context) && syscall_exit_p) {
-		unsigned long tstate = regs->tstate;
-		int result = AUDITSC_SUCCESS;
-
-		if (unlikely(tstate & (TSTATE_XCARRY | TSTATE_ICARRY)))
-			result = AUDITSC_FAILURE;
-
-		audit_syscall_exit(result, regs->u_regs[UREG_I0]);
-	}
-
-	if (!(current->ptrace & PT_PTRACED))
-		goto out;
-
-	if (!test_thread_flag(TIF_SYSCALL_TRACE))
-		goto out;
-
-	ptrace_notify(SIGTRAP | ((current->ptrace & PT_TRACESYSGOOD)
-				 ? 0x80 : 0));
-
-	/*
-	 * this isn't the same as continuing with a signal, but it will do
-	 * for normal use.  strace only continues with a signal if the
-	 * stopping signal is not SIGTRAP.  -brl
-	 */
-	if (current->exit_code) {
-		send_sig(current->exit_code, current, 1);
-		current->exit_code = 0;
-	}
+	if (test_thread_flag(TIF_SYSCALL_TRACE))
+		ret = tracehook_report_syscall_entry(regs);
 
-out:
-	if (unlikely(current->audit_context) && !syscall_exit_p)
+	if (unlikely(current->audit_context) && !ret)
 		audit_syscall_entry((test_thread_flag(TIF_32BIT) ?
 				     AUDIT_ARCH_SPARC :
 				     AUDIT_ARCH_SPARC64),
@@ -1093,4 +1069,22 @@ out:
 				    regs->u_regs[UREG_I1],
 				    regs->u_regs[UREG_I2],
 				    regs->u_regs[UREG_I3]);
+
+	return ret;
+}
+
+asmlinkage void syscall_trace_leave(struct pt_regs *regs)
+{
+	if (unlikely(current->audit_context)) {
+		unsigned long tstate = regs->tstate;
+		int result = AUDITSC_SUCCESS;
+
+		if (unlikely(tstate & (TSTATE_XCARRY | TSTATE_ICARRY)))
+			result = AUDITSC_FAILURE;
+
+		audit_syscall_exit(result, regs->u_regs[UREG_I0]);
+	}
+
+	if (test_thread_flag(TIF_SYSCALL_TRACE))
+		tracehook_report_syscall_exit(regs, 0);
 }
diff --git a/arch/sparc64/kernel/reboot.c b/arch/sparc64/kernel/reboot.c
new file mode 100644
index 00000000000..ef89d3d6974
--- /dev/null
+++ b/arch/sparc64/kernel/reboot.c
@@ -0,0 +1,53 @@
+/* reboot.c: reboot/shutdown/halt/poweroff handling
+ *
+ * Copyright (C) 2008 David S. Miller <davem@davemloft.net>
+ */
+#include <linux/kernel.h>
+#include <linux/reboot.h>
+#include <linux/module.h>
+#include <linux/pm.h>
+
+#include <asm/system.h>
+#include <asm/oplib.h>
+#include <asm/prom.h>
+
+/* sysctl - toggle power-off restriction for serial console
+ * systems in machine_power_off()
+ */
+int scons_pwroff = 1;
+
+/* This isn't actually used, it exists merely to satisfy the
+ * reference in kernel/sys.c
+ */
+void (*pm_power_off)(void) = machine_power_off;
+EXPORT_SYMBOL(pm_power_off);
+
+void machine_power_off(void)
+{
+	if (strcmp(of_console_device->type, "serial") || scons_pwroff)
+		prom_halt_power_off();
+
+	prom_halt();
+}
+
+void machine_halt(void)
+{
+	prom_halt();
+	panic("Halt failed!");
+}
+
+void machine_restart(char *cmd)
+{
+	char *p;
+
+	p = strchr(reboot_command, '\n');
+	if (p)
+		*p = 0;
+	if (cmd)
+		prom_reboot(cmd);
+	if (*reboot_command)
+		prom_reboot(reboot_command);
+	prom_reboot("");
+	panic("Reboot failed!");
+}
+
diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S
index c6fc695fe1f..97a993c1f7f 100644
--- a/arch/sparc64/kernel/rtrap.S
+++ b/arch/sparc64/kernel/rtrap.S
@@ -46,7 +46,7 @@ __handle_user_windows:
 		wrpr			%g0, RTRAP_PSTATE_IRQOFF, %pstate
 		ldx			[%g6 + TI_FLAGS], %l0
 
-1:		andcc			%l0, _TIF_SIGPENDING, %g0
+1:		andcc			%l0, _TIF_DO_NOTIFY_RESUME_MASK, %g0
 		be,pt			%xcc, __handle_user_windows_continue
 		 nop
 		mov			%l5, %o1
@@ -86,7 +86,7 @@ __handle_perfctrs:
 		 wrpr			%g0, RTRAP_PSTATE, %pstate
 		wrpr			%g0, RTRAP_PSTATE_IRQOFF, %pstate
 		ldx			[%g6 + TI_FLAGS], %l0
-1:		andcc			%l0, _TIF_SIGPENDING, %g0
+1:		andcc			%l0, _TIF_DO_NOTIFY_RESUME_MASK, %g0
 
 		be,pt			%xcc, __handle_perfctrs_continue
 		 sethi			%hi(TSTATE_PEF), %o0
@@ -195,7 +195,7 @@ __handle_preemption_continue:
 		 andcc			%l1, %o0, %g0
 		andcc			%l0, _TIF_NEED_RESCHED, %g0
 		bne,pn			%xcc, __handle_preemption
-		 andcc			%l0, _TIF_SIGPENDING, %g0
+		 andcc			%l0, _TIF_DO_NOTIFY_RESUME_MASK, %g0
 		bne,pn			%xcc, __handle_signal
 __handle_signal_continue:
 		 ldub			[%g6 + TI_WSAVED], %o2
diff --git a/arch/sparc64/kernel/sbus.c b/arch/sparc64/kernel/sbus.c
index e33a8a660e9..2ead310066d 100644
--- a/arch/sparc64/kernel/sbus.c
+++ b/arch/sparc64/kernel/sbus.c
@@ -11,15 +11,17 @@
 #include <linux/slab.h>
 #include <linux/init.h>
 #include <linux/interrupt.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
 
 #include <asm/page.h>
-#include <asm/sbus.h>
 #include <asm/io.h>
 #include <asm/upa.h>
 #include <asm/cache.h>
 #include <asm/dma.h>
 #include <asm/irq.h>
 #include <asm/prom.h>
+#include <asm/oplib.h>
 #include <asm/starfire.h>
 
 #include "iommu_common.h"
@@ -52,13 +54,23 @@
 #define STRBUF_TAG_VALID	0x02UL
 
 /* Enable 64-bit DVMA mode for the given device. */
-void sbus_set_sbus64(struct sbus_dev *sdev, int bursts)
+void sbus_set_sbus64(struct device *dev, int bursts)
 {
-	struct iommu *iommu = sdev->ofdev.dev.archdata.iommu;
-	int slot = sdev->slot;
+	struct iommu *iommu = dev->archdata.iommu;
+	struct of_device *op = to_of_device(dev);
+	const struct linux_prom_registers *regs;
 	unsigned long cfg_reg;
+	int slot;
 	u64 val;
 
+	regs = of_get_property(op->node, "reg", NULL);
+	if (!regs) {
+		printk(KERN_ERR "sbus_set_sbus64: Cannot find regs for %s\n",
+		       op->node->full_name);
+		return;
+	}
+	slot = regs->which_io;
+
 	cfg_reg = iommu->write_complete_reg;
 	switch (slot) {
 	case 0:
@@ -191,10 +203,9 @@ static unsigned long sysio_imap_to_iclr(unsigned long imap)
 	return imap + diff;
 }
 
-unsigned int sbus_build_irq(void *buscookie, unsigned int ino)
+static unsigned int sbus_build_irq(struct of_device *op, unsigned int ino)
 {
-	struct sbus_bus *sbus = (struct sbus_bus *)buscookie;
-	struct iommu *iommu = sbus->ofdev.dev.archdata.iommu;
+	struct iommu *iommu = op->dev.archdata.iommu;
 	unsigned long reg_base = iommu->write_complete_reg - 0x2000UL;
 	unsigned long imap, iclr;
 	int sbus_level = 0;
@@ -255,12 +266,12 @@ unsigned int sbus_build_irq(void *buscookie, unsigned int ino)
 #define  SYSIO_UEAFSR_RESV2 0x0000001fffffffffUL /* Reserved                  */
 static irqreturn_t sysio_ue_handler(int irq, void *dev_id)
 {
-	struct sbus_bus *sbus = dev_id;
-	struct iommu *iommu = sbus->ofdev.dev.archdata.iommu;
+	struct of_device *op = dev_id;
+	struct iommu *iommu = op->dev.archdata.iommu;
 	unsigned long reg_base = iommu->write_complete_reg - 0x2000UL;
 	unsigned long afsr_reg, afar_reg;
 	unsigned long afsr, afar, error_bits;
-	int reported;
+	int reported, portid;
 
 	afsr_reg = reg_base + SYSIO_UE_AFSR;
 	afar_reg = reg_base + SYSIO_UE_AFAR;
@@ -275,9 +286,11 @@ static irqreturn_t sysio_ue_handler(int irq, void *dev_id)
 		 SYSIO_UEAFSR_SPIO | SYSIO_UEAFSR_SDRD | SYSIO_UEAFSR_SDWR);
 	upa_writeq(error_bits, afsr_reg);
 
+	portid = of_getintprop_default(op->node, "portid", -1);
+
 	/* Log the error. */
 	printk("SYSIO[%x]: Uncorrectable ECC Error, primary error type[%s]\n",
-	       sbus->portid,
+	       portid,
 	       (((error_bits & SYSIO_UEAFSR_PPIO) ?
 		 "PIO" :
 		 ((error_bits & SYSIO_UEAFSR_PDRD) ?
@@ -285,12 +298,12 @@ static irqreturn_t sysio_ue_handler(int irq, void *dev_id)
 		  ((error_bits & SYSIO_UEAFSR_PDWR) ?
 		   "DVMA Write" : "???")))));
 	printk("SYSIO[%x]: DOFF[%lx] SIZE[%lx] MID[%lx]\n",
-	       sbus->portid,
+	       portid,
 	       (afsr & SYSIO_UEAFSR_DOFF) >> 45UL,
 	       (afsr & SYSIO_UEAFSR_SIZE) >> 42UL,
 	       (afsr & SYSIO_UEAFSR_MID) >> 37UL);
-	printk("SYSIO[%x]: AFAR[%016lx]\n", sbus->portid, afar);
-	printk("SYSIO[%x]: Secondary UE errors [", sbus->portid);
+	printk("SYSIO[%x]: AFAR[%016lx]\n", portid, afar);
+	printk("SYSIO[%x]: Secondary UE errors [", portid);
 	reported = 0;
 	if (afsr & SYSIO_UEAFSR_SPIO) {
 		reported++;
@@ -327,12 +340,12 @@ static irqreturn_t sysio_ue_handler(int irq, void *dev_id)
 #define  SYSIO_CEAFSR_RESV2 0x0000001fffffffffUL /* Reserved                  */
 static irqreturn_t sysio_ce_handler(int irq, void *dev_id)
 {
-	struct sbus_bus *sbus = dev_id;
-	struct iommu *iommu = sbus->ofdev.dev.archdata.iommu;
+	struct of_device *op = dev_id;
+	struct iommu *iommu = op->dev.archdata.iommu;
 	unsigned long reg_base = iommu->write_complete_reg - 0x2000UL;
 	unsigned long afsr_reg, afar_reg;
 	unsigned long afsr, afar, error_bits;
-	int reported;
+	int reported, portid;
 
 	afsr_reg = reg_base + SYSIO_CE_AFSR;
 	afar_reg = reg_base + SYSIO_CE_AFAR;
@@ -347,8 +360,10 @@ static irqreturn_t sysio_ce_handler(int irq, void *dev_id)
 		 SYSIO_CEAFSR_SPIO | SYSIO_CEAFSR_SDRD | SYSIO_CEAFSR_SDWR);
 	upa_writeq(error_bits, afsr_reg);
 
+	portid = of_getintprop_default(op->node, "portid", -1);
+
 	printk("SYSIO[%x]: Correctable ECC Error, primary error type[%s]\n",
-	       sbus->portid,
+	       portid,
 	       (((error_bits & SYSIO_CEAFSR_PPIO) ?
 		 "PIO" :
 		 ((error_bits & SYSIO_CEAFSR_PDRD) ?
@@ -360,14 +375,14 @@ static irqreturn_t sysio_ce_handler(int irq, void *dev_id)
 	 * XXX UDB CE trap handler does... -DaveM
 	 */
 	printk("SYSIO[%x]: DOFF[%lx] ECC Syndrome[%lx] Size[%lx] MID[%lx]\n",
-	       sbus->portid,
+	       portid,
 	       (afsr & SYSIO_CEAFSR_DOFF) >> 45UL,
 	       (afsr & SYSIO_CEAFSR_ESYND) >> 48UL,
 	       (afsr & SYSIO_CEAFSR_SIZE) >> 42UL,
 	       (afsr & SYSIO_CEAFSR_MID) >> 37UL);
-	printk("SYSIO[%x]: AFAR[%016lx]\n", sbus->portid, afar);
+	printk("SYSIO[%x]: AFAR[%016lx]\n", portid, afar);
 
-	printk("SYSIO[%x]: Secondary CE errors [", sbus->portid);
+	printk("SYSIO[%x]: Secondary CE errors [", portid);
 	reported = 0;
 	if (afsr & SYSIO_CEAFSR_SPIO) {
 		reported++;
@@ -404,11 +419,11 @@ static irqreturn_t sysio_ce_handler(int irq, void *dev_id)
 #define  SYSIO_SBAFSR_RESV3 0x0000001fffffffffUL /* Reserved                  */
 static irqreturn_t sysio_sbus_error_handler(int irq, void *dev_id)
 {
-	struct sbus_bus *sbus = dev_id;
-	struct iommu *iommu = sbus->ofdev.dev.archdata.iommu;
+	struct of_device *op = dev_id;
+	struct iommu *iommu = op->dev.archdata.iommu;
 	unsigned long afsr_reg, afar_reg, reg_base;
 	unsigned long afsr, afar, error_bits;
-	int reported;
+	int reported, portid;
 
 	reg_base = iommu->write_complete_reg - 0x2000UL;
 	afsr_reg = reg_base + SYSIO_SBUS_AFSR;
@@ -423,9 +438,11 @@ static irqreturn_t sysio_sbus_error_handler(int irq, void *dev_id)
 		 SYSIO_SBAFSR_SLE | SYSIO_SBAFSR_STO | SYSIO_SBAFSR_SBERR);
 	upa_writeq(error_bits, afsr_reg);
 
+	portid = of_getintprop_default(op->node, "portid", -1);
+
 	/* Log the error. */
 	printk("SYSIO[%x]: SBUS Error, primary error type[%s] read(%d)\n",
-	       sbus->portid,
+	       portid,
 	       (((error_bits & SYSIO_SBAFSR_PLE) ?
 		 "Late PIO Error" :
 		 ((error_bits & SYSIO_SBAFSR_PTO) ?
@@ -434,11 +451,11 @@ static irqreturn_t sysio_sbus_error_handler(int irq, void *dev_id)
 		   "Error Ack" : "???")))),
 	       (afsr & SYSIO_SBAFSR_RD) ? 1 : 0);
 	printk("SYSIO[%x]: size[%lx] MID[%lx]\n",
-	       sbus->portid,
+	       portid,
 	       (afsr & SYSIO_SBAFSR_SIZE) >> 42UL,
 	       (afsr & SYSIO_SBAFSR_MID) >> 37UL);
-	printk("SYSIO[%x]: AFAR[%016lx]\n", sbus->portid, afar);
-	printk("SYSIO[%x]: Secondary SBUS errors [", sbus->portid);
+	printk("SYSIO[%x]: AFAR[%016lx]\n", portid, afar);
+	printk("SYSIO[%x]: Secondary SBUS errors [", portid);
 	reported = 0;
 	if (afsr & SYSIO_SBAFSR_SLE) {
 		reported++;
@@ -470,34 +487,37 @@ static irqreturn_t sysio_sbus_error_handler(int irq, void *dev_id)
 #define SYSIO_CE_INO		0x35
 #define SYSIO_SBUSERR_INO	0x36
 
-static void __init sysio_register_error_handlers(struct sbus_bus *sbus)
+static void __init sysio_register_error_handlers(struct of_device *op)
 {
-	struct iommu *iommu = sbus->ofdev.dev.archdata.iommu;
+	struct iommu *iommu = op->dev.archdata.iommu;
 	unsigned long reg_base = iommu->write_complete_reg - 0x2000UL;
 	unsigned int irq;
 	u64 control;
+	int portid;
+
+	portid = of_getintprop_default(op->node, "portid", -1);
 
-	irq = sbus_build_irq(sbus, SYSIO_UE_INO);
+	irq = sbus_build_irq(op, SYSIO_UE_INO);
 	if (request_irq(irq, sysio_ue_handler, 0,
-			"SYSIO_UE", sbus) < 0) {
+			"SYSIO_UE", op) < 0) {
 		prom_printf("SYSIO[%x]: Cannot register UE interrupt.\n",
-			    sbus->portid);
+			    portid);
 		prom_halt();
 	}
 
-	irq = sbus_build_irq(sbus, SYSIO_CE_INO);
+	irq = sbus_build_irq(op, SYSIO_CE_INO);
 	if (request_irq(irq, sysio_ce_handler, 0,
-			"SYSIO_CE", sbus) < 0) {
+			"SYSIO_CE", op) < 0) {
 		prom_printf("SYSIO[%x]: Cannot register CE interrupt.\n",
-			    sbus->portid);
+			    portid);
 		prom_halt();
 	}
 
-	irq = sbus_build_irq(sbus, SYSIO_SBUSERR_INO);
+	irq = sbus_build_irq(op, SYSIO_SBUSERR_INO);
 	if (request_irq(irq, sysio_sbus_error_handler, 0,
-			"SYSIO_SBERR", sbus) < 0) {
+			"SYSIO_SBERR", op) < 0) {
 		prom_printf("SYSIO[%x]: Cannot register SBUS Error interrupt.\n",
-			    sbus->portid);
+			    portid);
 		prom_halt();
 	}
 
@@ -513,19 +533,15 @@ static void __init sysio_register_error_handlers(struct sbus_bus *sbus)
 }
 
 /* Boot time initialization. */
-static void __init sbus_iommu_init(int __node, struct sbus_bus *sbus)
+static void __init sbus_iommu_init(struct of_device *op)
 {
 	const struct linux_prom64_registers *pr;
-	struct device_node *dp;
+	struct device_node *dp = op->node;
 	struct iommu *iommu;
 	struct strbuf *strbuf;
 	unsigned long regs, reg_base;
+	int i, portid;
 	u64 control;
-	int i;
-
-	dp = of_find_node_by_phandle(__node);
-
-	sbus->portid = of_getintprop_default(dp, "upa-portid", -1);
 
 	pr = of_get_property(dp, "reg", NULL);
 	if (!pr) {
@@ -542,9 +558,9 @@ static void __init sbus_iommu_init(int __node, struct sbus_bus *sbus)
 	if (!strbuf)
 		goto fatal_memory_error;
 
-	sbus->ofdev.dev.archdata.iommu = iommu;
-	sbus->ofdev.dev.archdata.stc = strbuf;
-	sbus->ofdev.dev.archdata.numa_node = -1;
+	op->dev.archdata.iommu = iommu;
+	op->dev.archdata.stc = strbuf;
+	op->dev.archdata.numa_node = -1;
 
 	reg_base = regs + SYSIO_IOMMUREG_BASE;
 	iommu->iommu_control = reg_base + IOMMU_CONTROL;
@@ -572,8 +588,9 @@ static void __init sbus_iommu_init(int __node, struct sbus_bus *sbus)
 	 */
 	iommu->write_complete_reg = regs + 0x2000UL;
 
-	printk("SYSIO: UPA portID %x, at %016lx\n",
-	       sbus->portid, regs);
+	portid = of_getintprop_default(op->node, "portid", -1);
+	printk(KERN_INFO "SYSIO: UPA portID %x, at %016lx\n",
+	       portid, regs);
 
 	/* Setup for TSB_SIZE=7, TBW_SIZE=0, MMU_DE=1, MMU_EN=1 */
 	if (iommu_table_init(iommu, IO_TSB_SIZE, MAP_BASE, 0xffffffff, -1))
@@ -631,56 +648,27 @@ static void __init sbus_iommu_init(int __node, struct sbus_bus *sbus)
 
 	/* Now some Xfire specific grot... */
 	if (this_is_starfire)
-		starfire_hookup(sbus->portid);
+		starfire_hookup(portid);
 
-	sysio_register_error_handlers(sbus);
+	sysio_register_error_handlers(op);
 	return;
 
 fatal_memory_error:
 	prom_printf("sbus_iommu_init: Fatal memory allocation error.\n");
 }
 
-void sbus_fill_device_irq(struct sbus_dev *sdev)
+static int __init sbus_init(void)
 {
-	struct device_node *dp = of_find_node_by_phandle(sdev->prom_node);
-	const struct linux_prom_irqs *irqs;
-
-	irqs = of_get_property(dp, "interrupts", NULL);
-	if (!irqs) {
-		sdev->irqs[0] = 0;
-		sdev->num_irqs = 0;
-	} else {
-		unsigned int pri = irqs[0].pri;
+	struct device_node *dp;
 
-		sdev->num_irqs = 1;
-		if (pri < 0x20)
-			pri += sdev->slot * 8;
+	for_each_node_by_name(dp, "sbus") {
+		struct of_device *op = of_find_device_by_node(dp);
 
-		sdev->irqs[0] =	sbus_build_irq(sdev->bus, pri);
+		sbus_iommu_init(op);
+		of_propagate_archdata(op);
 	}
-}
 
-void __init sbus_arch_bus_ranges_init(struct device_node *pn, struct sbus_bus *sbus)
-{
-}
-
-void __init sbus_setup_iommu(struct sbus_bus *sbus, struct device_node *dp)
-{
-	sbus_iommu_init(dp->node, sbus);
-}
-
-void __init sbus_setup_arch_props(struct sbus_bus *sbus, struct device_node *dp)
-{
-}
-
-int __init sbus_arch_preinit(void)
-{
 	return 0;
 }
 
-void __init sbus_arch_postinit(void)
-{
-	extern void firetruck_init(void);
-
-	firetruck_init();
-}
+subsys_initcall(sbus_init);
diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c
index 9667e96fd51..ec82d76dc6f 100644
--- a/arch/sparc64/kernel/signal.c
+++ b/arch/sparc64/kernel/signal.c
@@ -2,7 +2,7 @@
  *  arch/sparc64/kernel/signal.c
  *
  *  Copyright (C) 1991, 1992  Linus Torvalds
- *  Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
+ *  Copyright (C) 1995, 2008 David S. Miller (davem@davemloft.net)
  *  Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx)
  *  Copyright (C) 1997 Eddie C. Dost   (ecd@skynet.be)
  *  Copyright (C) 1997,1998 Jakub Jelinek   (jj@sunsite.mff.cuni.cz)
@@ -17,6 +17,7 @@
 #include <linux/errno.h>
 #include <linux/wait.h>
 #include <linux/ptrace.h>
+#include <linux/tracehook.h>
 #include <linux/unistd.h>
 #include <linux/mm.h>
 #include <linux/tty.h>
@@ -89,7 +90,9 @@ asmlinkage void sparc64_set_context(struct pt_regs *regs)
 	err |= __get_user(regs->u_regs[UREG_G4], (&(*grp)[MC_G4]));
 	err |= __get_user(regs->u_regs[UREG_G5], (&(*grp)[MC_G5]));
 	err |= __get_user(regs->u_regs[UREG_G6], (&(*grp)[MC_G6]));
-	err |= __get_user(regs->u_regs[UREG_G7], (&(*grp)[MC_G7]));
+
+	/* Skip %g7 as that's the thread register in userspace.  */
+
 	err |= __get_user(regs->u_regs[UREG_I0], (&(*grp)[MC_O0]));
 	err |= __get_user(regs->u_regs[UREG_I1], (&(*grp)[MC_O1]));
 	err |= __get_user(regs->u_regs[UREG_I2], (&(*grp)[MC_O2]));
@@ -574,6 +577,8 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 		 * clear the TS_RESTORE_SIGMASK flag.
 		 */
 		current_thread_info()->status &= ~TS_RESTORE_SIGMASK;
+
+		tracehook_signal_handler(signr, &info, &ka, regs, 0);
 		return;
 	}
 	if (restart_syscall &&
@@ -605,4 +610,8 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 {
 	if (thread_info_flags & _TIF_SIGPENDING)
 		do_signal(regs, orig_i0);
+	if (thread_info_flags & _TIF_NOTIFY_RESUME) {
+		clear_thread_flag(TIF_NOTIFY_RESUME);
+		tracehook_notify_resume(regs);
+	}
 }
diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c
index 97cdd1bf4a1..ba5b09ad666 100644
--- a/arch/sparc64/kernel/signal32.c
+++ b/arch/sparc64/kernel/signal32.c
@@ -19,6 +19,7 @@
 #include <linux/binfmts.h>
 #include <linux/compat.h>
 #include <linux/bitops.h>
+#include <linux/tracehook.h>
 
 #include <asm/uaccess.h>
 #include <asm/ptrace.h>
@@ -794,6 +795,8 @@ void do_signal32(sigset_t *oldset, struct pt_regs * regs,
 		 * clear the TS_RESTORE_SIGMASK flag.
 		 */
 		current_thread_info()->status &= ~TS_RESTORE_SIGMASK;
+
+		tracehook_signal_handler(signr, &info, &ka, regs, 0);
 		return;
 	}
 	if (restart_syscall &&
diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c
index fa63c68a181..e5627118e61 100644
--- a/arch/sparc64/kernel/smp.c
+++ b/arch/sparc64/kernel/smp.c
@@ -21,6 +21,7 @@
 #include <linux/jiffies.h>
 #include <linux/profile.h>
 #include <linux/lmb.h>
+#include <linux/cpu.h>
 
 #include <asm/head.h>
 #include <asm/ptrace.h>
@@ -80,8 +81,6 @@ void smp_bogo(struct seq_file *m)
 			   i, cpu_data(i).clock_tick);
 }
 
-static __cacheline_aligned_in_smp DEFINE_SPINLOCK(call_lock);
-
 extern void setup_sparc64_timer(void);
 
 static volatile unsigned long callin_flag = 0;
@@ -117,12 +116,15 @@ void __cpuinit smp_callin(void)
 	atomic_inc(&init_mm.mm_count);
 	current->active_mm = &init_mm;
 
+	/* inform the notifiers about the new cpu */
+	notify_cpu_starting(cpuid);
+
 	while (!cpu_isset(cpuid, smp_commenced_mask))
 		rmb();
 
-	spin_lock(&call_lock);
+	ipi_call_lock();
 	cpu_set(cpuid, cpu_online_map);
-	spin_unlock(&call_lock);
+	ipi_call_unlock();
 
 	/* idle thread is expected to have preempt disabled */
 	preempt_disable();
@@ -459,27 +461,35 @@ again:
 	}
 }
 
-static inline void spitfire_xcall_deliver(u64 data0, u64 data1, u64 data2, cpumask_t mask)
+static void spitfire_xcall_deliver(struct trap_per_cpu *tb, int cnt)
 {
+	u64 *mondo, data0, data1, data2;
+	u16 *cpu_list;
 	u64 pstate;
 	int i;
 
 	__asm__ __volatile__("rdpr %%pstate, %0" : "=r" (pstate));
-	for_each_cpu_mask(i, mask)
-		spitfire_xcall_helper(data0, data1, data2, pstate, i);
+	cpu_list = __va(tb->cpu_list_pa);
+	mondo = __va(tb->cpu_mondo_block_pa);
+	data0 = mondo[0];
+	data1 = mondo[1];
+	data2 = mondo[2];
+	for (i = 0; i < cnt; i++)
+		spitfire_xcall_helper(data0, data1, data2, pstate, cpu_list[i]);
 }
 
 /* Cheetah now allows to send the whole 64-bytes of data in the interrupt
  * packet, but we have no use for that.  However we do take advantage of
  * the new pipelining feature (ie. dispatch to multiple cpus simultaneously).
  */
-static void cheetah_xcall_deliver(u64 data0, u64 data1, u64 data2, cpumask_t mask)
+static void cheetah_xcall_deliver(struct trap_per_cpu *tb, int cnt)
 {
-	u64 pstate, ver, busy_mask;
 	int nack_busy_id, is_jbus, need_more;
+	u64 *mondo, pstate, ver, busy_mask;
+	u16 *cpu_list;
 
-	if (cpus_empty(mask))
-		return;
+	cpu_list = __va(tb->cpu_list_pa);
+	mondo = __va(tb->cpu_mondo_block_pa);
 
 	/* Unfortunately, someone at Sun had the brilliant idea to make the
 	 * busy/nack fields hard-coded by ITID number for this Ultra-III
@@ -502,7 +512,7 @@ retry:
 			     "stxa	%2, [%5] %6\n\t"
 			     "membar	#Sync\n\t"
 			     : /* no outputs */
-			     : "r" (data0), "r" (data1), "r" (data2),
+			     : "r" (mondo[0]), "r" (mondo[1]), "r" (mondo[2]),
 			       "r" (0x40), "r" (0x50), "r" (0x60),
 			       "i" (ASI_INTR_W));
 
@@ -511,11 +521,16 @@ retry:
 	{
 		int i;
 
-		for_each_cpu_mask(i, mask) {
-			u64 target = (i << 14) | 0x70;
+		for (i = 0; i < cnt; i++) {
+			u64 target, nr;
+
+			nr = cpu_list[i];
+			if (nr == 0xffff)
+				continue;
 
+			target = (nr << 14) | 0x70;
 			if (is_jbus) {
-				busy_mask |= (0x1UL << (i * 2));
+				busy_mask |= (0x1UL << (nr * 2));
 			} else {
 				target |= (nack_busy_id << 24);
 				busy_mask |= (0x1UL <<
@@ -549,11 +564,13 @@ retry:
 				__asm__ __volatile__("wrpr %0, 0x0, %%pstate"
 						     : : "r" (pstate));
 				if (unlikely(need_more)) {
-					int i, cnt = 0;
-					for_each_cpu_mask(i, mask) {
-						cpu_clear(i, mask);
-						cnt++;
-						if (cnt == 32)
+					int i, this_cnt = 0;
+					for (i = 0; i < cnt; i++) {
+						if (cpu_list[i] == 0xffff)
+							continue;
+						cpu_list[i] = 0xffff;
+						this_cnt++;
+						if (this_cnt == 32)
 							break;
 					}
 					goto retry;
@@ -584,16 +601,20 @@ retry:
 			/* Clear out the mask bits for cpus which did not
 			 * NACK us.
 			 */
-			for_each_cpu_mask(i, mask) {
-				u64 check_mask;
+			for (i = 0; i < cnt; i++) {
+				u64 check_mask, nr;
+
+				nr = cpu_list[i];
+				if (nr == 0xffff)
+					continue;
 
 				if (is_jbus)
-					check_mask = (0x2UL << (2*i));
+					check_mask = (0x2UL << (2*nr));
 				else
 					check_mask = (0x2UL <<
 						      this_busy_nack);
 				if ((dispatch_stat & check_mask) == 0)
-					cpu_clear(i, mask);
+					cpu_list[i] = 0xffff;
 				this_busy_nack += 2;
 				if (this_busy_nack == 64)
 					break;
@@ -605,47 +626,17 @@ retry:
 }
 
 /* Multi-cpu list version.  */
-static void hypervisor_xcall_deliver(u64 data0, u64 data1, u64 data2, cpumask_t mask)
+static void hypervisor_xcall_deliver(struct trap_per_cpu *tb, int cnt)
 {
-	struct trap_per_cpu *tb;
+	int retries, this_cpu, prev_sent, i, saw_cpu_error;
+	unsigned long status;
 	u16 *cpu_list;
-	u64 *mondo;
-	cpumask_t error_mask;
-	unsigned long flags, status;
-	int cnt, retries, this_cpu, prev_sent, i;
-
-	if (cpus_empty(mask))
-		return;
-
-	/* We have to do this whole thing with interrupts fully disabled.
-	 * Otherwise if we send an xcall from interrupt context it will
-	 * corrupt both our mondo block and cpu list state.
-	 *
-	 * One consequence of this is that we cannot use timeout mechanisms
-	 * that depend upon interrupts being delivered locally.  So, for
-	 * example, we cannot sample jiffies and expect it to advance.
-	 *
-	 * Fortunately, udelay() uses %stick/%tick so we can use that.
-	 */
-	local_irq_save(flags);
 
 	this_cpu = smp_processor_id();
-	tb = &trap_block[this_cpu];
-
-	mondo = __va(tb->cpu_mondo_block_pa);
-	mondo[0] = data0;
-	mondo[1] = data1;
-	mondo[2] = data2;
-	wmb();
 
 	cpu_list = __va(tb->cpu_list_pa);
 
-	/* Setup the initial cpu list.  */
-	cnt = 0;
-	for_each_cpu_mask(i, mask)
-		cpu_list[cnt++] = i;
-
-	cpus_clear(error_mask);
+	saw_cpu_error = 0;
 	retries = 0;
 	prev_sent = 0;
 	do {
@@ -690,10 +681,9 @@ static void hypervisor_xcall_deliver(u64 data0, u64 data1, u64 data2, cpumask_t
 					continue;
 
 				err = sun4v_cpu_state(cpu);
-				if (err >= 0 &&
-				    err == HV_CPU_STATE_ERROR) {
+				if (err == HV_CPU_STATE_ERROR) {
+					saw_cpu_error = (cpu + 1);
 					cpu_list[i] = 0xffff;
-					cpu_set(cpu, error_mask);
 				}
 			}
 		} else if (unlikely(status != HV_EWOULDBLOCK))
@@ -717,32 +707,24 @@ static void hypervisor_xcall_deliver(u64 data0, u64 data1, u64 data2, cpumask_t
 		}
 	} while (1);
 
-	local_irq_restore(flags);
-
-	if (unlikely(!cpus_empty(error_mask)))
+	if (unlikely(saw_cpu_error))
 		goto fatal_mondo_cpu_error;
 
 	return;
 
 fatal_mondo_cpu_error:
 	printk(KERN_CRIT "CPU[%d]: SUN4V mondo cpu error, some target cpus "
-	       "were in error state\n",
-	       this_cpu);
-	printk(KERN_CRIT "CPU[%d]: Error mask [ ", this_cpu);
-	for_each_cpu_mask(i, error_mask)
-		printk("%d ", i);
-	printk("]\n");
+	       "(including %d) were in error state\n",
+	       this_cpu, saw_cpu_error - 1);
 	return;
 
 fatal_mondo_timeout:
-	local_irq_restore(flags);
 	printk(KERN_CRIT "CPU[%d]: SUN4V mondo timeout, no forward "
 	       " progress after %d retries.\n",
 	       this_cpu, retries);
 	goto dump_cpu_list_and_out;
 
 fatal_mondo_error:
-	local_irq_restore(flags);
 	printk(KERN_CRIT "CPU[%d]: Unexpected SUN4V mondo error %lu\n",
 	       this_cpu, status);
 	printk(KERN_CRIT "CPU[%d]: Args were cnt(%d) cpulist_pa(%lx) "
@@ -756,124 +738,103 @@ dump_cpu_list_and_out:
 	printk("]\n");
 }
 
-/* Send cross call to all processors mentioned in MASK
- * except self.
- */
-static void smp_cross_call_masked(unsigned long *func, u32 ctx, u64 data1, u64 data2, cpumask_t mask)
-{
-	u64 data0 = (((u64)ctx)<<32 | (((u64)func) & 0xffffffff));
-	int this_cpu = get_cpu();
-
-	cpus_and(mask, mask, cpu_online_map);
-	cpu_clear(this_cpu, mask);
-
-	if (tlb_type == spitfire)
-		spitfire_xcall_deliver(data0, data1, data2, mask);
-	else if (tlb_type == cheetah || tlb_type == cheetah_plus)
-		cheetah_xcall_deliver(data0, data1, data2, mask);
-	else
-		hypervisor_xcall_deliver(data0, data1, data2, mask);
-	/* NOTE: Caller runs local copy on master. */
+static void (*xcall_deliver_impl)(struct trap_per_cpu *, int);
 
-	put_cpu();
-}
+static void xcall_deliver(u64 data0, u64 data1, u64 data2, const cpumask_t *mask)
+{
+	struct trap_per_cpu *tb;
+	int this_cpu, i, cnt;
+	unsigned long flags;
+	u16 *cpu_list;
+	u64 *mondo;
 
-extern unsigned long xcall_sync_tick;
+	/* We have to do this whole thing with interrupts fully disabled.
+	 * Otherwise if we send an xcall from interrupt context it will
+	 * corrupt both our mondo block and cpu list state.
+	 *
+	 * One consequence of this is that we cannot use timeout mechanisms
+	 * that depend upon interrupts being delivered locally.  So, for
+	 * example, we cannot sample jiffies and expect it to advance.
+	 *
+	 * Fortunately, udelay() uses %stick/%tick so we can use that.
+	 */
+	local_irq_save(flags);
 
-static void smp_start_sync_tick_client(int cpu)
-{
-	cpumask_t mask = cpumask_of_cpu(cpu);
+	this_cpu = smp_processor_id();
+	tb = &trap_block[this_cpu];
 
-	smp_cross_call_masked(&xcall_sync_tick,
-			      0, 0, 0, mask);
-}
+	mondo = __va(tb->cpu_mondo_block_pa);
+	mondo[0] = data0;
+	mondo[1] = data1;
+	mondo[2] = data2;
+	wmb();
 
-/* Send cross call to all processors except self. */
-#define smp_cross_call(func, ctx, data1, data2) \
-	smp_cross_call_masked(func, ctx, data1, data2, cpu_online_map)
+	cpu_list = __va(tb->cpu_list_pa);
 
-struct call_data_struct {
-	void (*func) (void *info);
-	void *info;
-	atomic_t finished;
-	int wait;
-};
+	/* Setup the initial cpu list.  */
+	cnt = 0;
+	for_each_cpu_mask_nr(i, *mask) {
+		if (i == this_cpu || !cpu_online(i))
+			continue;
+		cpu_list[cnt++] = i;
+	}
 
-static struct call_data_struct *call_data;
+	if (cnt)
+		xcall_deliver_impl(tb, cnt);
 
-extern unsigned long xcall_call_function;
+	local_irq_restore(flags);
+}
 
-/**
- * smp_call_function(): Run a function on all other CPUs.
- * @func: The function to run. This must be fast and non-blocking.
- * @info: An arbitrary pointer to pass to the function.
- * @nonatomic: currently unused.
- * @wait: If true, wait (atomically) until function has completed on other CPUs.
- *
- * Returns 0 on success, else a negative status code. Does not return until
- * remote CPUs are nearly ready to execute <<func>> or are or have executed.
- *
- * You must not call this function with disabled interrupts or from a
- * hardware interrupt handler or from a bottom half handler.
+/* Send cross call to all processors mentioned in MASK_P
+ * except self.  Really, there are only two cases currently,
+ * "&cpu_online_map" and "&mm->cpu_vm_mask".
  */
-static int smp_call_function_mask(void (*func)(void *info), void *info,
-				  int nonatomic, int wait, cpumask_t mask)
+static void smp_cross_call_masked(unsigned long *func, u32 ctx, u64 data1, u64 data2, const cpumask_t *mask)
 {
-	struct call_data_struct data;
-	int cpus;
-
-	/* Can deadlock when called with interrupts disabled */
-	WARN_ON(irqs_disabled());
-
-	data.func = func;
-	data.info = info;
-	atomic_set(&data.finished, 0);
-	data.wait = wait;
-
-	spin_lock(&call_lock);
+	u64 data0 = (((u64)ctx)<<32 | (((u64)func) & 0xffffffff));
 
-	cpu_clear(smp_processor_id(), mask);
-	cpus = cpus_weight(mask);
-	if (!cpus)
-		goto out_unlock;
+	xcall_deliver(data0, data1, data2, mask);
+}
 
-	call_data = &data;
-	mb();
+/* Send cross call to all processors except self. */
+static void smp_cross_call(unsigned long *func, u32 ctx, u64 data1, u64 data2)
+{
+	smp_cross_call_masked(func, ctx, data1, data2, &cpu_online_map);
+}
 
-	smp_cross_call_masked(&xcall_call_function, 0, 0, 0, mask);
+extern unsigned long xcall_sync_tick;
 
-	/* Wait for response */
-	while (atomic_read(&data.finished) != cpus)
-		cpu_relax();
+static void smp_start_sync_tick_client(int cpu)
+{
+	xcall_deliver((u64) &xcall_sync_tick, 0, 0,
+		      &cpumask_of_cpu(cpu));
+}
 
-out_unlock:
-	spin_unlock(&call_lock);
+extern unsigned long xcall_call_function;
 
-	return 0;
+void arch_send_call_function_ipi(cpumask_t mask)
+{
+	xcall_deliver((u64) &xcall_call_function, 0, 0, &mask);
 }
 
-int smp_call_function(void (*func)(void *info), void *info,
-		      int nonatomic, int wait)
+extern unsigned long xcall_call_function_single;
+
+void arch_send_call_function_single_ipi(int cpu)
 {
-	return smp_call_function_mask(func, info, nonatomic, wait,
-				      cpu_online_map);
+	xcall_deliver((u64) &xcall_call_function_single, 0, 0,
+		      &cpumask_of_cpu(cpu));
 }
 
 void smp_call_function_client(int irq, struct pt_regs *regs)
 {
-	void (*func) (void *info) = call_data->func;
-	void *info = call_data->info;
+	clear_softint(1 << irq);
+	generic_smp_call_function_interrupt();
+}
 
+void smp_call_function_single_client(int irq, struct pt_regs *regs)
+{
 	clear_softint(1 << irq);
-	if (call_data->wait) {
-		/* let initiator proceed only after completion */
-		func(info);
-		atomic_inc(&call_data->finished);
-	} else {
-		/* let initiator proceed after getting data */
-		atomic_inc(&call_data->finished);
-		func(info);
-	}
+	generic_smp_call_function_single_interrupt();
 }
 
 static void tsb_sync(void *info)
@@ -893,16 +854,13 @@ static void tsb_sync(void *info)
 
 void smp_tsb_sync(struct mm_struct *mm)
 {
-	smp_call_function_mask(tsb_sync, mm, 0, 1, mm->cpu_vm_mask);
+	smp_call_function_mask(mm->cpu_vm_mask, tsb_sync, mm, 1);
 }
 
 extern unsigned long xcall_flush_tlb_mm;
 extern unsigned long xcall_flush_tlb_pending;
 extern unsigned long xcall_flush_tlb_kernel_range;
-extern unsigned long xcall_report_regs;
-#ifdef CONFIG_MAGIC_SYSRQ
 extern unsigned long xcall_fetch_glob_regs;
-#endif
 extern unsigned long xcall_receive_signal;
 extern unsigned long xcall_new_mmu_context_version;
 #ifdef CONFIG_KGDB
@@ -934,7 +892,6 @@ static inline void __local_flush_dcache_page(struct page *page)
 
 void smp_flush_dcache_page_impl(struct page *page, int cpu)
 {
-	cpumask_t mask = cpumask_of_cpu(cpu);
 	int this_cpu;
 
 	if (tlb_type == hypervisor)
@@ -950,29 +907,24 @@ void smp_flush_dcache_page_impl(struct page *page, int cpu)
 		__local_flush_dcache_page(page);
 	} else if (cpu_online(cpu)) {
 		void *pg_addr = page_address(page);
-		u64 data0;
+		u64 data0 = 0;
 
 		if (tlb_type == spitfire) {
-			data0 =
-				((u64)&xcall_flush_dcache_page_spitfire);
+			data0 = ((u64)&xcall_flush_dcache_page_spitfire);
 			if (page_mapping(page) != NULL)
 				data0 |= ((u64)1 << 32);
-			spitfire_xcall_deliver(data0,
-					       __pa(pg_addr),
-					       (u64) pg_addr,
-					       mask);
 		} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
 #ifdef DCACHE_ALIASING_POSSIBLE
-			data0 =
-				((u64)&xcall_flush_dcache_page_cheetah);
-			cheetah_xcall_deliver(data0,
-					      __pa(pg_addr),
-					      0, mask);
+			data0 =	((u64)&xcall_flush_dcache_page_cheetah);
 #endif
 		}
+		if (data0) {
+			xcall_deliver(data0, __pa(pg_addr),
+				      (u64) pg_addr, &cpumask_of_cpu(cpu));
 #ifdef CONFIG_DEBUG_DCFLUSH
-		atomic_inc(&dcpage_flushes_xcall);
+			atomic_inc(&dcpage_flushes_xcall);
 #endif
+		}
 	}
 
 	put_cpu();
@@ -980,66 +932,41 @@ void smp_flush_dcache_page_impl(struct page *page, int cpu)
 
 void flush_dcache_page_all(struct mm_struct *mm, struct page *page)
 {
-	void *pg_addr = page_address(page);
-	cpumask_t mask = cpu_online_map;
-	u64 data0;
+	void *pg_addr;
 	int this_cpu;
+	u64 data0;
 
 	if (tlb_type == hypervisor)
 		return;
 
 	this_cpu = get_cpu();
 
-	cpu_clear(this_cpu, mask);
-
 #ifdef CONFIG_DEBUG_DCFLUSH
 	atomic_inc(&dcpage_flushes);
 #endif
-	if (cpus_empty(mask))
-		goto flush_self;
+	data0 = 0;
+	pg_addr = page_address(page);
 	if (tlb_type == spitfire) {
 		data0 = ((u64)&xcall_flush_dcache_page_spitfire);
 		if (page_mapping(page) != NULL)
 			data0 |= ((u64)1 << 32);
-		spitfire_xcall_deliver(data0,
-				       __pa(pg_addr),
-				       (u64) pg_addr,
-				       mask);
 	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
 #ifdef DCACHE_ALIASING_POSSIBLE
 		data0 = ((u64)&xcall_flush_dcache_page_cheetah);
-		cheetah_xcall_deliver(data0,
-				      __pa(pg_addr),
-				      0, mask);
 #endif
 	}
+	if (data0) {
+		xcall_deliver(data0, __pa(pg_addr),
+			      (u64) pg_addr, &cpu_online_map);
 #ifdef CONFIG_DEBUG_DCFLUSH
-	atomic_inc(&dcpage_flushes_xcall);
+		atomic_inc(&dcpage_flushes_xcall);
 #endif
- flush_self:
+	}
 	__local_flush_dcache_page(page);
 
 	put_cpu();
 }
 
-static void __smp_receive_signal_mask(cpumask_t mask)
-{
-	smp_cross_call_masked(&xcall_receive_signal, 0, 0, 0, mask);
-}
-
-void smp_receive_signal(int cpu)
-{
-	cpumask_t mask = cpumask_of_cpu(cpu);
-
-	if (cpu_online(cpu))
-		__smp_receive_signal_mask(mask);
-}
-
-void smp_receive_signal_client(int irq, struct pt_regs *regs)
-{
-	clear_softint(1 << irq);
-}
-
 void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs)
 {
 	struct mm_struct *mm;
@@ -1078,17 +1005,10 @@ void kgdb_roundup_cpus(unsigned long flags)
 }
 #endif
 
-void smp_report_regs(void)
-{
-	smp_cross_call(&xcall_report_regs, 0, 0, 0);
-}
-
-#ifdef CONFIG_MAGIC_SYSRQ
 void smp_fetch_global_regs(void)
 {
 	smp_cross_call(&xcall_fetch_glob_regs, 0, 0, 0);
 }
-#endif
 
 /* We know that the window frames of the user have been flushed
  * to the stack before we get here because all callers of us
@@ -1145,7 +1065,7 @@ void smp_flush_tlb_mm(struct mm_struct *mm)
 
 	smp_cross_call_masked(&xcall_flush_tlb_mm,
 			      ctx, 0, 0,
-			      mm->cpu_vm_mask);
+			      &mm->cpu_vm_mask);
 
 local_flush_and_out:
 	__flush_tlb_mm(ctx, SECONDARY_CONTEXT);
@@ -1163,7 +1083,7 @@ void smp_flush_tlb_pending(struct mm_struct *mm, unsigned long nr, unsigned long
 	else
 		smp_cross_call_masked(&xcall_flush_tlb_pending,
 				      ctx, nr, (unsigned long) vaddrs,
-				      mm->cpu_vm_mask);
+				      &mm->cpu_vm_mask);
 
 	__flush_tlb_pending(ctx, nr, vaddrs);
 
@@ -1264,6 +1184,16 @@ void __devinit smp_prepare_boot_cpu(void)
 {
 }
 
+void __init smp_setup_processor_id(void)
+{
+	if (tlb_type == spitfire)
+		xcall_deliver_impl = spitfire_xcall_deliver;
+	else if (tlb_type == cheetah || tlb_type == cheetah_plus)
+		xcall_deliver_impl = cheetah_xcall_deliver;
+	else
+		xcall_deliver_impl = hypervisor_xcall_deliver;
+}
+
 void __devinit smp_fill_in_sib_core_maps(void)
 {
 	unsigned int i;
@@ -1377,10 +1307,6 @@ int __cpu_disable(void)
 	c->core_id = 0;
 	c->proc_id = -1;
 
-	spin_lock(&call_lock);
-	cpu_clear(cpu, cpu_online_map);
-	spin_unlock(&call_lock);
-
 	smp_wmb();
 
 	/* Make sure no interrupts point to this cpu.  */
@@ -1390,6 +1316,10 @@ int __cpu_disable(void)
 	mdelay(1);
 	local_irq_disable();
 
+	ipi_call_lock();
+	cpu_clear(cpu, cpu_online_map);
+	ipi_call_unlock();
+
 	return 0;
 }
 
@@ -1432,7 +1362,13 @@ void __init smp_cpus_done(unsigned int max_cpus)
 
 void smp_send_reschedule(int cpu)
 {
-	smp_receive_signal(cpu);
+	xcall_deliver((u64) &xcall_receive_signal, 0, 0,
+		      &cpumask_of_cpu(cpu));
+}
+
+void smp_receive_signal_client(int irq, struct pt_regs *regs)
+{
+	clear_softint(1 << irq);
 }
 
 /* This is a nop because we capture all other cpus
diff --git a/arch/sparc64/kernel/sparc64_ksyms.c b/arch/sparc64/kernel/sparc64_ksyms.c
index 8ac0b99f2c5..30bba8b0a3b 100644
--- a/arch/sparc64/kernel/sparc64_ksyms.c
+++ b/arch/sparc64/kernel/sparc64_ksyms.c
@@ -36,7 +36,6 @@
 #include <asm/elf.h>
 #include <asm/head.h>
 #include <asm/smp.h>
-#include <asm/mostek.h>
 #include <asm/ptrace.h>
 #include <asm/uaccess.h>
 #include <asm/checksum.h>
@@ -44,15 +43,12 @@
 #include <asm/pgalloc.h>
 #include <asm/cacheflush.h>
 #ifdef CONFIG_SBUS
-#include <asm/sbus.h>
 #include <asm/dma.h>
 #endif
-#ifdef CONFIG_PCI
-#include <asm/ebus.h>
-#endif
 #include <asm/ns87303.h>
 #include <asm/timer.h>
 #include <asm/cpudata.h>
+#include <asm/ftrace.h>
 
 struct poll {
 	int fd;
@@ -67,8 +63,6 @@ extern void *__memscan_zero(void *, size_t);
 extern void *__memscan_generic(void *, int, size_t);
 extern int __memcmp(const void *, const void *, __kernel_size_t);
 extern __kernel_size_t strlen(const char *);
-extern void show_regs(struct pt_regs *);
-extern void syscall_trace(struct pt_regs *, int);
 extern void sys_sigsuspend(void);
 extern int compat_sys_ioctl(unsigned int fd, unsigned int cmd, u32 arg);
 extern int (*handle_mathemu)(struct pt_regs *, struct fpustate *);
@@ -107,12 +101,9 @@ EXPORT_SYMBOL(__read_unlock);
 EXPORT_SYMBOL(__write_lock);
 EXPORT_SYMBOL(__write_unlock);
 EXPORT_SYMBOL(__write_trylock);
-
-EXPORT_SYMBOL(smp_call_function);
 #endif /* CONFIG_SMP */
 
-#if defined(CONFIG_MCOUNT)
-extern void _mcount(void);
+#ifdef CONFIG_MCOUNT
 EXPORT_SYMBOL(_mcount);
 #endif
 
@@ -157,26 +148,12 @@ EXPORT_SYMBOL(flush_dcache_page);
 EXPORT_SYMBOL(__flush_dcache_range);
 #endif
 
-EXPORT_SYMBOL(mostek_lock);
-EXPORT_SYMBOL(mstk48t02_regs);
 #ifdef CONFIG_SUN_AUXIO
 EXPORT_SYMBOL(auxio_set_led);
 EXPORT_SYMBOL(auxio_set_lte);
 #endif
 #ifdef CONFIG_SBUS
-EXPORT_SYMBOL(sbus_root);
-EXPORT_SYMBOL(dma_chain);
 EXPORT_SYMBOL(sbus_set_sbus64);
-EXPORT_SYMBOL(sbus_alloc_consistent);
-EXPORT_SYMBOL(sbus_free_consistent);
-EXPORT_SYMBOL(sbus_map_single);
-EXPORT_SYMBOL(sbus_unmap_single);
-EXPORT_SYMBOL(sbus_map_sg);
-EXPORT_SYMBOL(sbus_unmap_sg);
-EXPORT_SYMBOL(sbus_dma_sync_single_for_cpu);
-EXPORT_SYMBOL(sbus_dma_sync_single_for_device);
-EXPORT_SYMBOL(sbus_dma_sync_sg_for_cpu);
-EXPORT_SYMBOL(sbus_dma_sync_sg_for_device);
 #endif
 EXPORT_SYMBOL(outsb);
 EXPORT_SYMBOL(outsw);
@@ -185,7 +162,6 @@ EXPORT_SYMBOL(insb);
 EXPORT_SYMBOL(insw);
 EXPORT_SYMBOL(insl);
 #ifdef CONFIG_PCI
-EXPORT_SYMBOL(ebus_chain);
 EXPORT_SYMBOL(pci_alloc_consistent);
 EXPORT_SYMBOL(pci_free_consistent);
 EXPORT_SYMBOL(pci_map_single);
@@ -303,3 +279,5 @@ EXPORT_SYMBOL(xor_niagara_2);
 EXPORT_SYMBOL(xor_niagara_3);
 EXPORT_SYMBOL(xor_niagara_4);
 EXPORT_SYMBOL(xor_niagara_5);
+
+EXPORT_SYMBOL_GPL(real_hard_smp_processor_id);
diff --git a/arch/sparc64/kernel/sstate.c b/arch/sparc64/kernel/sstate.c
index 5b6e75b7f05..8cdbe5946b4 100644
--- a/arch/sparc64/kernel/sstate.c
+++ b/arch/sparc64/kernel/sstate.c
@@ -1,14 +1,15 @@
 /* sstate.c: System soft state support.
  *
- * Copyright (C) 2007 David S. Miller <davem@davemloft.net>
+ * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
  */
 
 #include <linux/kernel.h>
 #include <linux/notifier.h>
+#include <linux/reboot.h>
 #include <linux/init.h>
 
 #include <asm/hypervisor.h>
-#include <asm/sstate.h>
+#include <asm/spitfire.h>
 #include <asm/oplib.h>
 #include <asm/head.h>
 #include <asm/io.h>
@@ -50,31 +51,34 @@ static const char rebooting_msg[32] __attribute__((aligned(32))) =
 static const char panicing_msg[32] __attribute__((aligned(32))) =
 	"Linux panicing";
 
-void sstate_booting(void)
+static int sstate_reboot_call(struct notifier_block *np, unsigned long type, void *_unused)
 {
-	do_set_sstate(HV_SOFT_STATE_TRANSITION, booting_msg);
-}
+	const char *msg;
 
-void sstate_running(void)
-{
-	do_set_sstate(HV_SOFT_STATE_NORMAL, running_msg);
-}
+	switch (type) {
+	case SYS_DOWN:
+	default:
+		msg = rebooting_msg;
+		break;
 
-void sstate_halt(void)
-{
-	do_set_sstate(HV_SOFT_STATE_TRANSITION, halting_msg);
-}
+	case SYS_HALT:
+		msg = halting_msg;
+		break;
 
-void sstate_poweroff(void)
-{
-	do_set_sstate(HV_SOFT_STATE_TRANSITION, poweroff_msg);
-}
+	case SYS_POWER_OFF:
+		msg = poweroff_msg;
+		break;
+	}
 
-void sstate_reboot(void)
-{
-	do_set_sstate(HV_SOFT_STATE_TRANSITION, rebooting_msg);
+	do_set_sstate(HV_SOFT_STATE_TRANSITION, msg);
+
+	return NOTIFY_OK;
 }
 
+static struct notifier_block sstate_reboot_notifier = {
+	.notifier_call = sstate_reboot_call,
+};
+
 static int sstate_panic_event(struct notifier_block *n, unsigned long event, void *ptr)
 {
 	do_set_sstate(HV_SOFT_STATE_TRANSITION, panicing_msg);
@@ -87,18 +91,37 @@ static struct notifier_block sstate_panic_block = {
 	.priority	=	INT_MAX,
 };
 
-void __init sun4v_sstate_init(void)
+static int __init sstate_init(void)
 {
 	unsigned long major, minor;
 
+	if (tlb_type != hypervisor)
+		return 0;
+
 	major = 1;
 	minor = 0;
 	if (sun4v_hvapi_register(HV_GRP_SOFT_STATE, major, &minor))
-		return;
+		return 0;
 
 	hv_supports_soft_state = 1;
 
 	prom_sun4v_guest_soft_state();
+
+	do_set_sstate(HV_SOFT_STATE_TRANSITION, booting_msg);
+
 	atomic_notifier_chain_register(&panic_notifier_list,
 				       &sstate_panic_block);
+	register_reboot_notifier(&sstate_reboot_notifier);
+
+	return 0;
 }
+
+core_initcall(sstate_init);
+
+static int __init sstate_running(void)
+{
+	do_set_sstate(HV_SOFT_STATE_NORMAL, running_msg);
+	return 0;
+}
+
+late_initcall(sstate_running);
diff --git a/arch/sparc64/kernel/stacktrace.c b/arch/sparc64/kernel/stacktrace.c
index c73ce3f4197..4e21d4a57d3 100644
--- a/arch/sparc64/kernel/stacktrace.c
+++ b/arch/sparc64/kernel/stacktrace.c
@@ -1,13 +1,16 @@
 #include <linux/sched.h>
 #include <linux/stacktrace.h>
 #include <linux/thread_info.h>
+#include <linux/module.h>
 #include <asm/ptrace.h>
 #include <asm/stacktrace.h>
 
+#include "kstack.h"
+
 void save_stack_trace(struct stack_trace *trace)
 {
-	unsigned long ksp, fp, thread_base;
 	struct thread_info *tp = task_thread_info(current);
+	unsigned long ksp, fp;
 
 	stack_trace_flush();
 
@@ -17,21 +20,18 @@ void save_stack_trace(struct stack_trace *trace)
 	);
 
 	fp = ksp + STACK_BIAS;
-	thread_base = (unsigned long) tp;
 	do {
 		struct sparc_stackf *sf;
 		struct pt_regs *regs;
 		unsigned long pc;
 
-		/* Bogus frame pointer? */
-		if (fp < (thread_base + sizeof(struct thread_info)) ||
-		    fp >= (thread_base + THREAD_SIZE))
+		if (!kstack_valid(tp, fp))
 			break;
 
 		sf = (struct sparc_stackf *) fp;
 		regs = (struct pt_regs *) (sf + 1);
 
-		if ((regs->magic & ~0x1ff) == PT_REGS_MAGIC) {
+		if (kstack_is_trap_frame(tp, regs)) {
 			if (!(regs->tstate & TSTATE_PRIV))
 				break;
 			pc = regs->tpc;
@@ -47,3 +47,4 @@ void save_stack_trace(struct stack_trace *trace)
 			trace->entries[trace->nr_entries++] = pc;
 	} while (trace->nr_entries < trace->max_entries);
 }
+EXPORT_SYMBOL_GPL(save_stack_trace);
diff --git a/arch/sparc64/kernel/starfire.c b/arch/sparc64/kernel/starfire.c
index 7461581b3bb..060d0f3a615 100644
--- a/arch/sparc64/kernel/starfire.c
+++ b/arch/sparc64/kernel/starfire.c
@@ -28,11 +28,6 @@ void check_if_starfire(void)
 		this_is_starfire = 1;
 }
 
-void starfire_cpu_setup(void)
-{
-	/* Currently, nothing to do.  */
-}
-
 int starfire_hard_smp_processor_id(void)
 {
 	return upa_readl(0x1fff40000d0UL);
diff --git a/arch/sparc64/kernel/sys_sparc.c b/arch/sparc64/kernel/sys_sparc.c
index ac1bff58c1a..39749e32dc7 100644
--- a/arch/sparc64/kernel/sys_sparc.c
+++ b/arch/sparc64/kernel/sys_sparc.c
@@ -418,7 +418,7 @@ asmlinkage long sparc_pipe(struct pt_regs *regs)
 	int fd[2];
 	int error;
 
-	error = do_pipe(fd);
+	error = do_pipe_flags(fd, 0);
 	if (error)
 		goto out;
 	regs->u_regs[UREG_I1] = fd[1];
@@ -542,7 +542,7 @@ asmlinkage long sparc64_personality(unsigned long personality)
 	return ret;
 }
 
-int sparc64_mmap_check(unsigned long addr, unsigned long len)
+int sparc_mmap_check(unsigned long addr, unsigned long len)
 {
 	if (test_thread_flag(TIF_32BIT)) {
 		if (len >= STACK_TOP32)
@@ -614,9 +614,9 @@ asmlinkage unsigned long sys64_mremap(unsigned long addr,
 		goto out;
 	if (unlikely(new_len >= VA_EXCLUDE_START))
 		goto out;
-	if (unlikely(sparc64_mmap_check(addr, old_len)))
+	if (unlikely(sparc_mmap_check(addr, old_len)))
 		goto out;
-	if (unlikely(sparc64_mmap_check(new_addr, new_len)))
+	if (unlikely(sparc_mmap_check(new_addr, new_len)))
 		goto out;
 
 	down_write(&current->mm->mmap_sem);
diff --git a/arch/sparc64/kernel/sys_sparc32.c b/arch/sparc64/kernel/sys_sparc32.c
index ba5bd626b39..3320c9d0075 100644
--- a/arch/sparc64/kernel/sys_sparc32.c
+++ b/arch/sparc64/kernel/sys_sparc32.c
@@ -58,182 +58,6 @@
 #include <asm/mmu_context.h>
 #include <asm/compat_signal.h>
 
-asmlinkage long sys32_chown16(const char __user * filename, u16 user, u16 group)
-{
-	return sys_chown(filename, low2highuid(user), low2highgid(group));
-}
-
-asmlinkage long sys32_lchown16(const char __user * filename, u16 user, u16 group)
-{
-	return sys_lchown(filename, low2highuid(user), low2highgid(group));
-}
-
-asmlinkage long sys32_fchown16(unsigned int fd, u16 user, u16 group)
-{
-	return sys_fchown(fd, low2highuid(user), low2highgid(group));
-}
-
-asmlinkage long sys32_setregid16(u16 rgid, u16 egid)
-{
-	return sys_setregid(low2highgid(rgid), low2highgid(egid));
-}
-
-asmlinkage long sys32_setgid16(u16 gid)
-{
-	return sys_setgid((gid_t)gid);
-}
-
-asmlinkage long sys32_setreuid16(u16 ruid, u16 euid)
-{
-	return sys_setreuid(low2highuid(ruid), low2highuid(euid));
-}
-
-asmlinkage long sys32_setuid16(u16 uid)
-{
-	return sys_setuid((uid_t)uid);
-}
-
-asmlinkage long sys32_setresuid16(u16 ruid, u16 euid, u16 suid)
-{
-	return sys_setresuid(low2highuid(ruid), low2highuid(euid),
-		low2highuid(suid));
-}
-
-asmlinkage long sys32_getresuid16(u16 __user *ruid, u16 __user *euid, u16 __user *suid)
-{
-	int retval;
-
-	if (!(retval = put_user(high2lowuid(current->uid), ruid)) &&
-	    !(retval = put_user(high2lowuid(current->euid), euid)))
-		retval = put_user(high2lowuid(current->suid), suid);
-
-	return retval;
-}
-
-asmlinkage long sys32_setresgid16(u16 rgid, u16 egid, u16 sgid)
-{
-	return sys_setresgid(low2highgid(rgid), low2highgid(egid),
-		low2highgid(sgid));
-}
-
-asmlinkage long sys32_getresgid16(u16 __user *rgid, u16 __user *egid, u16 __user *sgid)
-{
-	int retval;
-
-	if (!(retval = put_user(high2lowgid(current->gid), rgid)) &&
-	    !(retval = put_user(high2lowgid(current->egid), egid)))
-		retval = put_user(high2lowgid(current->sgid), sgid);
-
-	return retval;
-}
-
-asmlinkage long sys32_setfsuid16(u16 uid)
-{
-	return sys_setfsuid((uid_t)uid);
-}
-
-asmlinkage long sys32_setfsgid16(u16 gid)
-{
-	return sys_setfsgid((gid_t)gid);
-}
-
-static int groups16_to_user(u16 __user *grouplist, struct group_info *group_info)
-{
-	int i;
-	u16 group;
-
-	for (i = 0; i < group_info->ngroups; i++) {
-		group = (u16)GROUP_AT(group_info, i);
-		if (put_user(group, grouplist+i))
-			return -EFAULT;
-	}
-
-	return 0;
-}
-
-static int groups16_from_user(struct group_info *group_info, u16 __user *grouplist)
-{
-	int i;
-	u16 group;
-
-	for (i = 0; i < group_info->ngroups; i++) {
-		if (get_user(group, grouplist+i))
-			return  -EFAULT;
-		GROUP_AT(group_info, i) = (gid_t)group;
-	}
-
-	return 0;
-}
-
-asmlinkage long sys32_getgroups16(int gidsetsize, u16 __user *grouplist)
-{
-	int i;
-
-	if (gidsetsize < 0)
-		return -EINVAL;
-
-	get_group_info(current->group_info);
-	i = current->group_info->ngroups;
-	if (gidsetsize) {
-		if (i > gidsetsize) {
-			i = -EINVAL;
-			goto out;
-		}
-		if (groups16_to_user(grouplist, current->group_info)) {
-			i = -EFAULT;
-			goto out;
-		}
-	}
-out:
-	put_group_info(current->group_info);
-	return i;
-}
-
-asmlinkage long sys32_setgroups16(int gidsetsize, u16 __user *grouplist)
-{
-	struct group_info *group_info;
-	int retval;
-
-	if (!capable(CAP_SETGID))
-		return -EPERM;
-	if ((unsigned)gidsetsize > NGROUPS_MAX)
-		return -EINVAL;
-
-	group_info = groups_alloc(gidsetsize);
-	if (!group_info)
-		return -ENOMEM;
-	retval = groups16_from_user(group_info, grouplist);
-	if (retval) {
-		put_group_info(group_info);
-		return retval;
-	}
-
-	retval = set_current_groups(group_info);
-	put_group_info(group_info);
-
-	return retval;
-}
-
-asmlinkage long sys32_getuid16(void)
-{
-	return high2lowuid(current->uid);
-}
-
-asmlinkage long sys32_geteuid16(void)
-{
-	return high2lowuid(current->euid);
-}
-
-asmlinkage long sys32_getgid16(void)
-{
-	return high2lowgid(current->gid);
-}
-
-asmlinkage long sys32_getegid16(void)
-{
-	return high2lowgid(current->egid);
-}
-
 /* 32-bit timeval and related flotsam.  */
 
 static inline long put_tv32(struct compat_timeval __user *o, struct timeval *i)
@@ -359,7 +183,8 @@ int cp_compat_stat(struct kstat *stat, struct compat_stat __user *statbuf)
 	return err;
 }
 
-int cp_compat_stat64(struct kstat *stat, struct compat_stat64 __user *statbuf)
+static int cp_compat_stat64(struct kstat *stat,
+			    struct compat_stat64 __user *statbuf)
 {
 	int err;
 
@@ -750,14 +575,6 @@ asmlinkage long sys32_settimeofday(struct compat_timeval __user *tv,
 	return do_sys_settimeofday(tv ? &kts : NULL, tz ? &ktz : NULL);
 }
 
-/* These are here just in case some old sparc32 binary calls it. */
-asmlinkage long sys32_pause(void)
-{
-	current->state = TASK_INTERRUPTIBLE;
-	schedule();
-	return -ERESTARTNOHAND;
-}
-
 asmlinkage compat_ssize_t sys32_pread64(unsigned int fd,
 					char __user *ubuf,
 					compat_size_t count,
@@ -870,9 +687,9 @@ asmlinkage unsigned long sys32_mremap(unsigned long addr,
 	unsigned long ret = -EINVAL;
 	unsigned long new_addr = __new_addr;
 
-	if (unlikely(sparc64_mmap_check(addr, old_len)))
+	if (unlikely(sparc_mmap_check(addr, old_len)))
 		goto out;
-	if (unlikely(sparc64_mmap_check(new_addr, new_len)))
+	if (unlikely(sparc_mmap_check(new_addr, new_len)))
 		goto out;
 	down_write(&current->mm->mmap_sem);
 	ret = do_mremap(addr, old_len, new_len, flags, new_addr);
diff --git a/arch/sparc64/kernel/syscalls.S b/arch/sparc64/kernel/syscalls.S
index db19ed67acf..7a6786a7136 100644
--- a/arch/sparc64/kernel/syscalls.S
+++ b/arch/sparc64/kernel/syscalls.S
@@ -65,9 +65,8 @@ sys32_rt_sigreturn:
 	andcc	%l5, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT), %g0
 	be,pt	%icc, rtrap
 	 nop
-	add	%sp, PTREGS_OFF, %o0
-	call	syscall_trace
-	 mov	1, %o1
+	call	syscall_trace_leave
+	 add	%sp, PTREGS_OFF, %o0
 	ba,pt	%xcc, rtrap
 	 nop
 
@@ -159,9 +158,10 @@ linux_sparc_ni_syscall:
 	 or	%l7, %lo(sys_ni_syscall), %l7
 
 linux_syscall_trace32:
-	add	%sp, PTREGS_OFF, %o0
-	call	syscall_trace
-	 clr	%o1
+	call	syscall_trace_enter
+	 add	%sp, PTREGS_OFF, %o0
+	brnz,pn	%o0, 3f
+	 mov	-ENOSYS, %o0
 	srl	%i0, 0, %o0
 	srl	%i4, 0, %o4
 	srl	%i1, 0, %o1
@@ -170,9 +170,10 @@ linux_syscall_trace32:
 	 srl	%i3, 0, %o3
 
 linux_syscall_trace:
-	add	%sp, PTREGS_OFF, %o0
-	call	syscall_trace
-	 clr	%o1
+	call	syscall_trace_enter
+	 add	%sp, PTREGS_OFF, %o0
+	brnz,pn	%o0, 3f
+	 mov	-ENOSYS, %o0
 	mov	%i0, %o0
 	mov	%i1, %o1
 	mov	%i2, %o2
@@ -271,9 +272,8 @@ ret_sys_call:
 	b,pt	%xcc, rtrap
 	 stx	%l2, [%sp + PTREGS_OFF + PT_V9_TNPC]
 linux_syscall_trace2:
-	add	%sp, PTREGS_OFF, %o0
-	call	syscall_trace
-	 mov	1, %o1
+	call	syscall_trace_leave
+	 add	%sp, PTREGS_OFF, %o0
 	stx	%l1, [%sp + PTREGS_OFF + PT_V9_TPC]
 	ba,pt	%xcc, rtrap
 	 stx	%l2, [%sp + PTREGS_OFF + PT_V9_TNPC]
diff --git a/arch/sparc64/kernel/sysfs.c b/arch/sparc64/kernel/sysfs.c
index e885034a6b7..84e5ce14671 100644
--- a/arch/sparc64/kernel/sysfs.c
+++ b/arch/sparc64/kernel/sysfs.c
@@ -14,7 +14,8 @@
 static DEFINE_PER_CPU(struct hv_mmu_statistics, mmu_stats) __attribute__((aligned(64)));
 
 #define SHOW_MMUSTAT_ULONG(NAME) \
-static ssize_t show_##NAME(struct sys_device *dev, char *buf) \
+static ssize_t show_##NAME(struct sys_device *dev, \
+			struct sysdev_attribute *attr, char *buf) \
 { \
 	struct hv_mmu_statistics *p = &per_cpu(mmu_stats, dev->id); \
 	return sprintf(buf, "%lu\n", p->NAME); \
@@ -135,13 +136,16 @@ static unsigned long write_mmustat_enable(unsigned long val)
 	return sun4v_mmustat_conf(ra, &orig_ra);
 }
 
-static ssize_t show_mmustat_enable(struct sys_device *s, char *buf)
+static ssize_t show_mmustat_enable(struct sys_device *s,
+				struct sysdev_attribute *attr, char *buf)
 {
 	unsigned long val = run_on_cpu(s->id, read_mmustat_enable, 0);
 	return sprintf(buf, "%lx\n", val);
 }
 
-static ssize_t store_mmustat_enable(struct sys_device *s, const char *buf, size_t count)
+static ssize_t store_mmustat_enable(struct sys_device *s,
+			struct sysdev_attribute *attr, const char *buf,
+			size_t count)
 {
 	unsigned long val, err;
 	int ret = sscanf(buf, "%ld", &val);
@@ -179,14 +183,16 @@ static void unregister_mmu_stats(struct sys_device *s)
 #endif
 
 #define SHOW_CPUDATA_ULONG_NAME(NAME, MEMBER) \
-static ssize_t show_##NAME(struct sys_device *dev, char *buf) \
+static ssize_t show_##NAME(struct sys_device *dev, \
+		struct sysdev_attribute *attr, char *buf) \
 { \
 	cpuinfo_sparc *c = &cpu_data(dev->id); \
 	return sprintf(buf, "%lu\n", c->MEMBER); \
 }
 
 #define SHOW_CPUDATA_UINT_NAME(NAME, MEMBER) \
-static ssize_t show_##NAME(struct sys_device *dev, char *buf) \
+static ssize_t show_##NAME(struct sys_device *dev, \
+		struct sysdev_attribute *attr, char *buf) \
 { \
 	cpuinfo_sparc *c = &cpu_data(dev->id); \
 	return sprintf(buf, "%u\n", c->MEMBER); \
diff --git a/arch/sparc64/kernel/systbls.S b/arch/sparc64/kernel/systbls.S
index 8b5282d433c..5daee4b04dd 100644
--- a/arch/sparc64/kernel/systbls.S
+++ b/arch/sparc64/kernel/systbls.S
@@ -20,21 +20,21 @@
 sys_call_table32:
 /*0*/	.word sys_restart_syscall, sys32_exit, sys_fork, sys_read, sys_write
 /*5*/	.word sys32_open, sys_close, sys32_wait4, sys32_creat, sys_link
-/*10*/  .word sys_unlink, sunos_execv, sys_chdir, sys32_chown16, sys32_mknod
-/*15*/	.word sys_chmod, sys32_lchown16, sparc_brk, sys32_perfctr, sys32_lseek
-/*20*/	.word sys_getpid, sys_capget, sys_capset, sys32_setuid16, sys32_getuid16
-/*25*/	.word sys32_vmsplice, compat_sys_ptrace, sys_alarm, sys32_sigaltstack, sys32_pause
+/*10*/  .word sys_unlink, sunos_execv, sys_chdir, sys_chown16, sys32_mknod
+/*15*/	.word sys_chmod, sys_lchown16, sparc_brk, sys32_perfctr, sys32_lseek
+/*20*/	.word sys_getpid, sys_capget, sys_capset, sys_setuid16, sys_getuid16
+/*25*/	.word sys32_vmsplice, compat_sys_ptrace, sys_alarm, sys32_sigaltstack, sys_pause
 /*30*/	.word compat_sys_utime, sys_lchown, sys_fchown, sys32_access, sys32_nice
 	.word sys_chown, sys_sync, sys32_kill, compat_sys_newstat, sys32_sendfile
 /*40*/	.word compat_sys_newlstat, sys_dup, sys_pipe, compat_sys_times, sys_getuid
-	.word sys32_umount, sys32_setgid16, sys32_getgid16, sys32_signal, sys32_geteuid16
-/*50*/	.word sys32_getegid16, sys_acct, sys_nis_syscall, sys_getgid, compat_sys_ioctl
+	.word sys32_umount, sys_setgid16, sys_getgid16, sys32_signal, sys_geteuid16
+/*50*/	.word sys_getegid16, sys_acct, sys_nis_syscall, sys_getgid, compat_sys_ioctl
 	.word sys32_reboot, sys32_mmap2, sys_symlink, sys32_readlink, sys32_execve
 /*60*/	.word sys32_umask, sys_chroot, compat_sys_newfstat, compat_sys_fstat64, sys_getpagesize
 	.word sys32_msync, sys_vfork, sys32_pread64, sys32_pwrite64, sys_geteuid
 /*70*/	.word sys_getegid, sys_mmap, sys_setreuid, sys_munmap, sys_mprotect
-	.word sys_madvise, sys_vhangup, sys32_truncate64, sys_mincore, sys32_getgroups16
-/*80*/	.word sys32_setgroups16, sys_getpgrp, sys32_setgroups, sys32_setitimer, sys32_ftruncate64
+	.word sys_madvise, sys_vhangup, sys32_truncate64, sys_mincore, sys_getgroups16
+/*80*/	.word sys_setgroups16, sys_getpgrp, sys32_setgroups, sys32_setitimer, sys32_ftruncate64
 	.word sys32_swapon, sys32_getitimer, sys_setuid, sys32_sethostname, sys_setgid
 /*90*/	.word sys_dup2, sys_setfsuid, compat_sys_fcntl, sys32_select, sys_setfsgid
 	.word sys_fsync, sys32_setpriority, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
@@ -42,8 +42,8 @@ sys_call_table32:
 	.word compat_sys_rt_sigtimedwait, sys32_rt_sigqueueinfo, compat_sys_rt_sigsuspend, sys_setresuid, sys_getresuid
 /*110*/	.word sys_setresgid, sys_getresgid, sys_setregid, sys_nis_syscall, sys_nis_syscall
 	.word sys32_getgroups, sys32_gettimeofday, sys32_getrusage, sys_nis_syscall, sys_getcwd
-/*120*/	.word compat_sys_readv, compat_sys_writev, sys32_settimeofday, sys32_fchown16, sys_fchmod
-	.word sys_nis_syscall, sys32_setreuid16, sys32_setregid16, sys_rename, sys_truncate
+/*120*/	.word compat_sys_readv, compat_sys_writev, sys32_settimeofday, sys_fchown16, sys_fchmod
+	.word sys_nis_syscall, sys_setreuid16, sys_setregid16, sys_rename, sys_truncate
 /*130*/	.word sys_ftruncate, sys_flock, compat_sys_lstat64, sys_nis_syscall, sys_nis_syscall
 	.word sys_nis_syscall, sys32_mkdir, sys_rmdir, compat_sys_utimes, compat_sys_stat64
 /*140*/	.word sys32_sendfile64, sys_nis_syscall, sys32_futex, sys_gettid, compat_sys_getrlimit
@@ -63,7 +63,7 @@ sys_call_table32:
 /*210*/	.word sys32_fadvise64_64, sys32_tgkill, sys32_waitpid, sys_swapoff, compat_sys_sysinfo
 	.word compat_sys_ipc, sys32_sigreturn, sys_clone, sys32_ioprio_get, compat_sys_adjtimex
 /*220*/	.word sys32_sigprocmask, sys_ni_syscall, sys32_delete_module, sys_ni_syscall, sys32_getpgid
-	.word sys32_bdflush, sys32_sysfs, sys_nis_syscall, sys32_setfsuid16, sys32_setfsgid16
+	.word sys32_bdflush, sys32_sysfs, sys_nis_syscall, sys_setfsuid16, sys_setfsgid16
 /*230*/	.word sys32_select, compat_sys_time, sys32_splice, compat_sys_stime, compat_sys_statfs64
 	.word compat_sys_fstatfs64, sys_llseek, sys_mlock, sys_munlock, sys32_mlockall
 /*240*/	.word sys_munlockall, sys32_sched_setparam, sys32_sched_getparam, sys32_sched_setscheduler, sys32_sched_getscheduler
@@ -81,7 +81,8 @@ sys_call_table32:
 /*300*/	.word compat_sys_set_robust_list, compat_sys_get_robust_list, compat_sys_migrate_pages, compat_sys_mbind, compat_sys_get_mempolicy
 	.word compat_sys_set_mempolicy, compat_sys_kexec_load, compat_sys_move_pages, sys_getcpu, compat_sys_epoll_pwait
 /*310*/	.word compat_sys_utimensat, compat_sys_signalfd, sys_timerfd_create, sys_eventfd, compat_sys_fallocate
-	.word compat_sys_timerfd_settime, compat_sys_timerfd_gettime
+	.word compat_sys_timerfd_settime, compat_sys_timerfd_gettime, compat_sys_signalfd4, sys_eventfd2, sys_epoll_create1
+/*320*/	.word sys_dup3, sys_pipe2, sys_inotify_init1
 
 #endif /* CONFIG_COMPAT */
 
@@ -154,4 +155,5 @@ sys_call_table:
 /*300*/	.word sys_set_robust_list, sys_get_robust_list, sys_migrate_pages, sys_mbind, sys_get_mempolicy
 	.word sys_set_mempolicy, sys_kexec_load, sys_move_pages, sys_getcpu, sys_epoll_pwait
 /*310*/	.word sys_utimensat, sys_signalfd, sys_timerfd_create, sys_eventfd, sys_fallocate
-	.word sys_timerfd_settime, sys_timerfd_gettime
+	.word sys_timerfd_settime, sys_timerfd_gettime, sys_signalfd4, sys_eventfd2, sys_epoll_create1
+/*320*/	.word sys_dup3, sys_pipe2, sys_inotify_init1
diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c
index e5d238970c7..80d71a5ce1e 100644
--- a/arch/sparc64/kernel/time.c
+++ b/arch/sparc64/kernel/time.c
@@ -11,6 +11,7 @@
 #include <linux/errno.h>
 #include <linux/module.h>
 #include <linux/sched.h>
+#include <linux/smp_lock.h>
 #include <linux/kernel.h>
 #include <linux/param.h>
 #include <linux/string.h>
@@ -29,17 +30,18 @@
 #include <linux/percpu.h>
 #include <linux/miscdevice.h>
 #include <linux/rtc.h>
+#include <linux/rtc/m48t59.h>
 #include <linux/kernel_stat.h>
 #include <linux/clockchips.h>
 #include <linux/clocksource.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
 
 #include <asm/oplib.h>
-#include <asm/mostek.h>
 #include <asm/timer.h>
 #include <asm/irq.h>
 #include <asm/io.h>
 #include <asm/prom.h>
-#include <asm/of_device.h>
 #include <asm/starfire.h>
 #include <asm/smp.h>
 #include <asm/sections.h>
@@ -49,18 +51,7 @@
 
 #include "entry.h"
 
-DEFINE_SPINLOCK(mostek_lock);
 DEFINE_SPINLOCK(rtc_lock);
-void __iomem *mstk48t02_regs = NULL;
-#ifdef CONFIG_PCI
-unsigned long ds1287_regs = 0UL;
-static void __iomem *bq4802_regs;
-#endif
-
-static void __iomem *mstk48t08_regs;
-static void __iomem *mstk48t59_regs;
-
-static int set_rtc_mmss(unsigned long);
 
 #define TICK_PRIV_BIT	(1UL << 63)
 #define TICKCMP_IRQ_BIT	(1UL << 63)
@@ -404,313 +395,167 @@ static unsigned long timer_ticks_per_nsec_quotient __read_mostly;
 
 int update_persistent_clock(struct timespec now)
 {
-	return set_rtc_mmss(now.tv_sec);
-}
+	struct rtc_device *rtc = rtc_class_open("rtc0");
+	int err = -1;
 
-/* Kick start a stopped clock (procedure from the Sun NVRAM/hostid FAQ). */
-static void __init kick_start_clock(void)
-{
-	void __iomem *regs = mstk48t02_regs;
-	u8 sec, tmp;
-	int i, count;
-
-	prom_printf("CLOCK: Clock was stopped. Kick start ");
-
-	spin_lock_irq(&mostek_lock);
-
-	/* Turn on the kick start bit to start the oscillator. */
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp |= MSTK_CREG_WRITE;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-	tmp = mostek_read(regs + MOSTEK_SEC);
-	tmp &= ~MSTK_STOP;
-	mostek_write(regs + MOSTEK_SEC, tmp);
-	tmp = mostek_read(regs + MOSTEK_HOUR);
-	tmp |= MSTK_KICK_START;
-	mostek_write(regs + MOSTEK_HOUR, tmp);
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp &= ~MSTK_CREG_WRITE;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-
-	spin_unlock_irq(&mostek_lock);
-
-	/* Delay to allow the clock oscillator to start. */
-	sec = MSTK_REG_SEC(regs);
-	for (i = 0; i < 3; i++) {
-		while (sec == MSTK_REG_SEC(regs))
-			for (count = 0; count < 100000; count++)
-				/* nothing */ ;
-		prom_printf(".");
-		sec = MSTK_REG_SEC(regs);
-	}
-	prom_printf("\n");
-
-	spin_lock_irq(&mostek_lock);
-
-	/* Turn off kick start and set a "valid" time and date. */
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp |= MSTK_CREG_WRITE;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-	tmp = mostek_read(regs + MOSTEK_HOUR);
-	tmp &= ~MSTK_KICK_START;
-	mostek_write(regs + MOSTEK_HOUR, tmp);
-	MSTK_SET_REG_SEC(regs,0);
-	MSTK_SET_REG_MIN(regs,0);
-	MSTK_SET_REG_HOUR(regs,0);
-	MSTK_SET_REG_DOW(regs,5);
-	MSTK_SET_REG_DOM(regs,1);
-	MSTK_SET_REG_MONTH(regs,8);
-	MSTK_SET_REG_YEAR(regs,1996 - MSTK_YEAR_ZERO);
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp &= ~MSTK_CREG_WRITE;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-
-	spin_unlock_irq(&mostek_lock);
-
-	/* Ensure the kick start bit is off. If it isn't, turn it off. */
-	while (mostek_read(regs + MOSTEK_HOUR) & MSTK_KICK_START) {
-		prom_printf("CLOCK: Kick start still on!\n");
-
-		spin_lock_irq(&mostek_lock);
-
-		tmp = mostek_read(regs + MOSTEK_CREG);
-		tmp |= MSTK_CREG_WRITE;
-		mostek_write(regs + MOSTEK_CREG, tmp);
-
-		tmp = mostek_read(regs + MOSTEK_HOUR);
-		tmp &= ~MSTK_KICK_START;
-		mostek_write(regs + MOSTEK_HOUR, tmp);
-
-		tmp = mostek_read(regs + MOSTEK_CREG);
-		tmp &= ~MSTK_CREG_WRITE;
-		mostek_write(regs + MOSTEK_CREG, tmp);
-
-		spin_unlock_irq(&mostek_lock);
+	if (rtc) {
+		err = rtc_set_mmss(rtc, now.tv_sec);
+		rtc_class_close(rtc);
 	}
 
-	prom_printf("CLOCK: Kick start procedure successful.\n");
+	return err;
 }
 
-/* Return nonzero if the clock chip battery is low. */
-static int __init has_low_battery(void)
-{
-	void __iomem *regs = mstk48t02_regs;
-	u8 data1, data2;
-
-	spin_lock_irq(&mostek_lock);
+unsigned long cmos_regs;
+EXPORT_SYMBOL(cmos_regs);
 
-	data1 = mostek_read(regs + MOSTEK_EEPROM);	/* Read some data. */
-	mostek_write(regs + MOSTEK_EEPROM, ~data1);	/* Write back the complement. */
-	data2 = mostek_read(regs + MOSTEK_EEPROM);	/* Read back the complement. */
-	mostek_write(regs + MOSTEK_EEPROM, data1);	/* Restore original value. */
+static struct resource rtc_cmos_resource;
 
-	spin_unlock_irq(&mostek_lock);
-
-	return (data1 == data2);	/* Was the write blocked? */
-}
+static struct platform_device rtc_cmos_device = {
+	.name		= "rtc_cmos",
+	.id		= -1,
+	.resource	= &rtc_cmos_resource,
+	.num_resources	= 1,
+};
 
-static void __init mostek_set_system_time(void __iomem *mregs)
+static int __devinit rtc_probe(struct of_device *op, const struct of_device_id *match)
 {
-	unsigned int year, mon, day, hour, min, sec;
-	u8 tmp;
-
-	spin_lock_irq(&mostek_lock);
-
-	/* Traditional Mostek chip. */
-	tmp = mostek_read(mregs + MOSTEK_CREG);
-	tmp |= MSTK_CREG_READ;
-	mostek_write(mregs + MOSTEK_CREG, tmp);
+	struct resource *r;
 
-	sec = MSTK_REG_SEC(mregs);
-	min = MSTK_REG_MIN(mregs);
-	hour = MSTK_REG_HOUR(mregs);
-	day = MSTK_REG_DOM(mregs);
-	mon = MSTK_REG_MONTH(mregs);
-	year = MSTK_CVT_YEAR( MSTK_REG_YEAR(mregs) );
+	printk(KERN_INFO "%s: RTC regs at 0x%lx\n",
+	       op->node->full_name, op->resource[0].start);
 
-	xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
-	xtime.tv_nsec = (INITIAL_JIFFIES % HZ) * (NSEC_PER_SEC / HZ);
-	set_normalized_timespec(&wall_to_monotonic,
- 	                        -xtime.tv_sec, -xtime.tv_nsec);
+	/* The CMOS RTC driver only accepts IORESOURCE_IO, so cons
+	 * up a fake resource so that the probe works for all cases.
+	 * When the RTC is behind an ISA bus it will have IORESOURCE_IO
+	 * already, whereas when it's behind EBUS is will be IORESOURCE_MEM.
+	 */
 
-	tmp = mostek_read(mregs + MOSTEK_CREG);
-	tmp &= ~MSTK_CREG_READ;
-	mostek_write(mregs + MOSTEK_CREG, tmp);
+	r = &rtc_cmos_resource;
+	r->flags = IORESOURCE_IO;
+	r->name = op->resource[0].name;
+	r->start = op->resource[0].start;
+	r->end = op->resource[0].end;
 
-	spin_unlock_irq(&mostek_lock);
+	cmos_regs = op->resource[0].start;
+	return platform_device_register(&rtc_cmos_device);
 }
 
-/* Probe for the real time clock chip. */
-static void __init set_system_time(void)
-{
-	unsigned int year, mon, day, hour, min, sec;
-	void __iomem *mregs = mstk48t02_regs;
-#ifdef CONFIG_PCI
-	unsigned long dregs = ds1287_regs;
-	void __iomem *bregs = bq4802_regs;
-#else
-	unsigned long dregs = 0UL;
-	void __iomem *bregs = 0UL;
-#endif
-
-	if (!mregs && !dregs && !bregs) {
-		prom_printf("Something wrong, clock regs not mapped yet.\n");
-		prom_halt();
-	}		
-
-	if (mregs) {
-		mostek_set_system_time(mregs);
-		return;
-	}
-
-	if (bregs) {
-		unsigned char val = readb(bregs + 0x0e);
-		unsigned int century;
+static struct of_device_id __initdata rtc_match[] = {
+	{
+		.name = "rtc",
+		.compatible = "m5819",
+	},
+	{
+		.name = "rtc",
+		.compatible = "isa-m5819p",
+	},
+	{
+		.name = "rtc",
+		.compatible = "isa-m5823p",
+	},
+	{
+		.name = "rtc",
+		.compatible = "ds1287",
+	},
+	{},
+};
 
-		/* BQ4802 RTC chip. */
+static struct of_platform_driver rtc_driver = {
+	.match_table	= rtc_match,
+	.probe		= rtc_probe,
+	.driver		= {
+		.name	= "rtc",
+	},
+};
 
-		writeb(val | 0x08, bregs + 0x0e);
+static struct platform_device rtc_bq4802_device = {
+	.name		= "rtc-bq4802",
+	.id		= -1,
+	.num_resources	= 1,
+};
 
-		sec  = readb(bregs + 0x00);
-		min  = readb(bregs + 0x02);
-		hour = readb(bregs + 0x04);
-		day  = readb(bregs + 0x06);
-		mon  = readb(bregs + 0x09);
-		year = readb(bregs + 0x0a);
-		century = readb(bregs + 0x0f);
+static int __devinit bq4802_probe(struct of_device *op, const struct of_device_id *match)
+{
 
-		writeb(val, bregs + 0x0e);
+	printk(KERN_INFO "%s: BQ4802 regs at 0x%lx\n",
+	       op->node->full_name, op->resource[0].start);
 
-		BCD_TO_BIN(sec);
-		BCD_TO_BIN(min);
-		BCD_TO_BIN(hour);
-		BCD_TO_BIN(day);
-		BCD_TO_BIN(mon);
-		BCD_TO_BIN(year);
-		BCD_TO_BIN(century);
+	rtc_bq4802_device.resource = &op->resource[0];
+	return platform_device_register(&rtc_bq4802_device);
+}
 
-		year += (century * 100);
-	} else {
-		/* Dallas 12887 RTC chip. */
-
-		do {
-			sec  = CMOS_READ(RTC_SECONDS);
-			min  = CMOS_READ(RTC_MINUTES);
-			hour = CMOS_READ(RTC_HOURS);
-			day  = CMOS_READ(RTC_DAY_OF_MONTH);
-			mon  = CMOS_READ(RTC_MONTH);
-			year = CMOS_READ(RTC_YEAR);
-		} while (sec != CMOS_READ(RTC_SECONDS));
-
-		if (!(CMOS_READ(RTC_CONTROL) & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
-			BCD_TO_BIN(sec);
-			BCD_TO_BIN(min);
-			BCD_TO_BIN(hour);
-			BCD_TO_BIN(day);
-			BCD_TO_BIN(mon);
-			BCD_TO_BIN(year);
-		}
-		if ((year += 1900) < 1970)
-			year += 100;
-	}
+static struct of_device_id __initdata bq4802_match[] = {
+	{
+		.name = "rtc",
+		.compatible = "bq4802",
+	},
+};
 
-	xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
-	xtime.tv_nsec = (INITIAL_JIFFIES % HZ) * (NSEC_PER_SEC / HZ);
-	set_normalized_timespec(&wall_to_monotonic,
- 	                        -xtime.tv_sec, -xtime.tv_nsec);
-}
+static struct of_platform_driver bq4802_driver = {
+	.match_table	= bq4802_match,
+	.probe		= bq4802_probe,
+	.driver		= {
+		.name	= "bq4802",
+	},
+};
 
-/* davem suggests we keep this within the 4M locked kernel image */
-static u32 starfire_get_time(void)
+static unsigned char mostek_read_byte(struct device *dev, u32 ofs)
 {
-	static char obp_gettod[32];
-	static u32 unix_tod;
+	struct platform_device *pdev = to_platform_device(dev);
+	struct m48t59_plat_data *pdata = pdev->dev.platform_data;
+	void __iomem *regs;
+	unsigned char val;
 
-	sprintf(obp_gettod, "h# %08x unix-gettod",
-		(unsigned int) (long) &unix_tod);
-	prom_feval(obp_gettod);
+	regs = (void __iomem *) pdev->resource[0].start;
+	val = readb(regs + ofs);
 
-	return unix_tod;
+	/* the year 0 is 1968 */
+	if (ofs == pdata->offset + M48T59_YEAR) {
+		val += 0x68;
+		if ((val & 0xf) > 9)
+			val += 6;
+	}
+	return val;
 }
 
-static int starfire_set_time(u32 val)
+static void mostek_write_byte(struct device *dev, u32 ofs, u8 val)
 {
-	/* Do nothing, time is set using the service processor
-	 * console on this platform.
-	 */
-	return 0;
-}
+	struct platform_device *pdev = to_platform_device(dev);
+	struct m48t59_plat_data *pdata = pdev->dev.platform_data;
+	void __iomem *regs;
 
-static u32 hypervisor_get_time(void)
-{
-	unsigned long ret, time;
-	int retries = 10000;
-
-retry:
-	ret = sun4v_tod_get(&time);
-	if (ret == HV_EOK)
-		return time;
-	if (ret == HV_EWOULDBLOCK) {
-		if (--retries > 0) {
-			udelay(100);
-			goto retry;
-		}
-		printk(KERN_WARNING "SUN4V: tod_get() timed out.\n");
-		return 0;
+	regs = (void __iomem *) pdev->resource[0].start;
+	if (ofs == pdata->offset + M48T59_YEAR) {
+		if (val < 0x68)
+			val += 0x32;
+		else
+			val -= 0x68;
+		if ((val & 0xf) > 9)
+			val += 6;
+		if ((val & 0xf0) > 0x9A)
+			val += 0x60;
 	}
-	printk(KERN_WARNING "SUN4V: tod_get() not supported.\n");
-	return 0;
+	writeb(val, regs + ofs);
 }
 
-static int hypervisor_set_time(u32 secs)
-{
-	unsigned long ret;
-	int retries = 10000;
-
-retry:
-	ret = sun4v_tod_set(secs);
-	if (ret == HV_EOK)
-		return 0;
-	if (ret == HV_EWOULDBLOCK) {
-		if (--retries > 0) {
-			udelay(100);
-			goto retry;
-		}
-		printk(KERN_WARNING "SUN4V: tod_set() timed out.\n");
-		return -EAGAIN;
-	}
-	printk(KERN_WARNING "SUN4V: tod_set() not supported.\n");
-	return -EOPNOTSUPP;
-}
+static struct m48t59_plat_data m48t59_data = {
+	.read_byte	= mostek_read_byte,
+	.write_byte	= mostek_write_byte,
+};
 
-static int __init clock_model_matches(const char *model)
-{
-	if (strcmp(model, "mk48t02") &&
-	    strcmp(model, "mk48t08") &&
-	    strcmp(model, "mk48t59") &&
-	    strcmp(model, "m5819") &&
-	    strcmp(model, "m5819p") &&
-	    strcmp(model, "m5823") &&
-	    strcmp(model, "ds1287") &&
-	    strcmp(model, "bq4802"))
-		return 0;
-
-	return 1;
-}
+static struct platform_device m48t59_rtc = {
+	.name		= "rtc-m48t59",
+	.id		= 0,
+	.num_resources	= 1,
+	.dev	= {
+		.platform_data = &m48t59_data,
+	},
+};
 
-static int __devinit clock_probe(struct of_device *op, const struct of_device_id *match)
+static int __devinit mostek_probe(struct of_device *op, const struct of_device_id *match)
 {
 	struct device_node *dp = op->node;
-	const char *model = of_get_property(dp, "model", NULL);
-	const char *compat = of_get_property(dp, "compatible", NULL);
-	unsigned long size, flags;
-	void __iomem *regs;
-
-	if (!model)
-		model = compat;
-
-	if (!model || !clock_model_matches(model))
-		return -ENODEV;
 
 	/* On an Enterprise system there can be multiple mostek clocks.
 	 * We should only match the one that is on the central FHC bus.
@@ -719,88 +564,51 @@ static int __devinit clock_probe(struct of_device *op, const struct of_device_id
 	    strcmp(dp->parent->parent->name, "central") != 0)
 		return -ENODEV;
 
-	size = (op->resource[0].end - op->resource[0].start) + 1;
-	regs = of_ioremap(&op->resource[0], 0, size, "clock");
-	if (!regs)
-		return -ENOMEM;
-
-#ifdef CONFIG_PCI
-	if (!strcmp(model, "ds1287") ||
-	    !strcmp(model, "m5819") ||
-	    !strcmp(model, "m5819p") ||
-	    !strcmp(model, "m5823")) {
-		ds1287_regs = (unsigned long) regs;
-	} else if (!strcmp(model, "bq4802")) {
-		bq4802_regs = regs;
-	} else
-#endif
-	if (model[5] == '0' && model[6] == '2') {
-		mstk48t02_regs = regs;
-	} else if(model[5] == '0' && model[6] == '8') {
-		mstk48t08_regs = regs;
-		mstk48t02_regs = mstk48t08_regs + MOSTEK_48T08_48T02;
-	} else {
-		mstk48t59_regs = regs;
-		mstk48t02_regs = mstk48t59_regs + MOSTEK_48T59_48T02;
-	}
-
-	printk(KERN_INFO "%s: Clock regs at %p\n", dp->full_name, regs);
-
-	local_irq_save(flags);
+	printk(KERN_INFO "%s: Mostek regs at 0x%lx\n",
+	       dp->full_name, op->resource[0].start);
 
-	if (mstk48t02_regs != NULL) {
-		/* Report a low battery voltage condition. */
-		if (has_low_battery())
-			prom_printf("NVRAM: Low battery voltage!\n");
-
-		/* Kick start the clock if it is completely stopped. */
-		if (mostek_read(mstk48t02_regs + MOSTEK_SEC) & MSTK_STOP)
-			kick_start_clock();
-	}
-
-	set_system_time();
-	
-	local_irq_restore(flags);
-
-	return 0;
+	m48t59_rtc.resource = &op->resource[0];
+	return platform_device_register(&m48t59_rtc);
 }
 
-static struct of_device_id clock_match[] = {
+static struct of_device_id __initdata mostek_match[] = {
 	{
 		.name = "eeprom",
 	},
-	{
-		.name = "rtc",
-	},
 	{},
 };
 
-static struct of_platform_driver clock_driver = {
-	.match_table	= clock_match,
-	.probe		= clock_probe,
+static struct of_platform_driver mostek_driver = {
+	.match_table	= mostek_match,
+	.probe		= mostek_probe,
 	.driver		= {
-		.name	= "clock",
+		.name	= "mostek",
 	},
 };
 
+static struct platform_device rtc_sun4v_device = {
+	.name		= "rtc-sun4v",
+	.id		= -1,
+};
+
+static struct platform_device rtc_starfire_device = {
+	.name		= "rtc-starfire",
+	.id		= -1,
+};
+
 static int __init clock_init(void)
 {
-	if (this_is_starfire) {
-		xtime.tv_sec = starfire_get_time();
-		xtime.tv_nsec = (INITIAL_JIFFIES % HZ) * (NSEC_PER_SEC / HZ);
-		set_normalized_timespec(&wall_to_monotonic,
-		                        -xtime.tv_sec, -xtime.tv_nsec);
-		return 0;
-	}
-	if (tlb_type == hypervisor) {
-		xtime.tv_sec = hypervisor_get_time();
-		xtime.tv_nsec = (INITIAL_JIFFIES % HZ) * (NSEC_PER_SEC / HZ);
-		set_normalized_timespec(&wall_to_monotonic,
-		                        -xtime.tv_sec, -xtime.tv_nsec);
-		return 0;
-	}
+	if (this_is_starfire)
+		return platform_device_register(&rtc_starfire_device);
 
-	return of_register_driver(&clock_driver, &of_platform_bus_type);
+	if (tlb_type == hypervisor)
+		return platform_device_register(&rtc_sun4v_device);
+
+	(void) of_register_driver(&rtc_driver, &of_platform_bus_type);
+	(void) of_register_driver(&mostek_driver, &of_platform_bus_type);
+	(void) of_register_driver(&bq4802_driver, &of_platform_bus_type);
+
+	return 0;
 }
 
 /* Must be after subsys_initcall() so that busses are probed.  Must
@@ -813,7 +621,7 @@ fs_initcall(clock_init);
 static unsigned long sparc64_init_timers(void)
 {
 	struct device_node *dp;
-	unsigned long clock;
+	unsigned long freq;
 
 	dp = of_find_node_by_path("/");
 	if (tlb_type == spitfire) {
@@ -826,17 +634,17 @@ static unsigned long sparc64_init_timers(void)
 		if (manuf == 0x17 && impl == 0x13) {
 			/* Hummingbird, aka Ultra-IIe */
 			tick_ops = &hbtick_operations;
-			clock = of_getintprop_default(dp, "stick-frequency", 0);
+			freq = of_getintprop_default(dp, "stick-frequency", 0);
 		} else {
 			tick_ops = &tick_operations;
-			clock = local_cpu_data().clock_tick;
+			freq = local_cpu_data().clock_tick;
 		}
 	} else {
 		tick_ops = &stick_operations;
-		clock = of_getintprop_default(dp, "stick-frequency", 0);
+		freq = of_getintprop_default(dp, "stick-frequency", 0);
 	}
 
-	return clock;
+	return freq;
 }
 
 struct freq_table {
@@ -883,6 +691,16 @@ static struct notifier_block sparc64_cpufreq_notifier_block = {
 	.notifier_call	= sparc64_cpufreq_notifier
 };
 
+static int __init register_sparc64_cpufreq_notifier(void)
+{
+
+	cpufreq_register_notifier(&sparc64_cpufreq_notifier_block,
+				  CPUFREQ_TRANSITION_NOTIFIER);
+	return 0;
+}
+
+core_initcall(register_sparc64_cpufreq_notifier);
+
 #endif /* CONFIG_CPU_FREQ */
 
 static int sparc64_next_event(unsigned long delta,
@@ -1018,16 +836,16 @@ EXPORT_SYMBOL(udelay);
 
 void __init time_init(void)
 {
-	unsigned long clock = sparc64_init_timers();
+	unsigned long freq = sparc64_init_timers();
 
-	tb_ticks_per_usec = clock / USEC_PER_SEC;
+	tb_ticks_per_usec = freq / USEC_PER_SEC;
 
 	timer_ticks_per_nsec_quotient =
-		clocksource_hz2mult(clock, SPARC64_NSEC_PER_CYC_SHIFT);
+		clocksource_hz2mult(freq, SPARC64_NSEC_PER_CYC_SHIFT);
 
 	clocksource_tick.name = tick_ops->name;
 	clocksource_tick.mult =
-		clocksource_hz2mult(clock,
+		clocksource_hz2mult(freq,
 				    clocksource_tick.shift);
 	clocksource_tick.read = tick_ops->get_tick;
 
@@ -1038,7 +856,7 @@ void __init time_init(void)
 
 	sparc64_clockevent.name = tick_ops->name;
 
-	setup_clockevent_multiplier(clock);
+	setup_clockevent_multiplier(freq);
 
 	sparc64_clockevent.max_delta_ns =
 		clockevent_delta2ns(0x7fffffffffffffffUL, &sparc64_clockevent);
@@ -1049,11 +867,6 @@ void __init time_init(void)
 	       sparc64_clockevent.mult, sparc64_clockevent.shift);
 
 	setup_sparc64_timer();
-
-#ifdef CONFIG_CPU_FREQ
-	cpufreq_register_notifier(&sparc64_cpufreq_notifier_block,
-				  CPUFREQ_TRANSITION_NOTIFIER);
-#endif
 }
 
 unsigned long long sched_clock(void)
@@ -1064,668 +877,8 @@ unsigned long long sched_clock(void)
 		>> SPARC64_NSEC_PER_CYC_SHIFT;
 }
 
-static int set_rtc_mmss(unsigned long nowtime)
-{
-	int real_seconds, real_minutes, chip_minutes;
-	void __iomem *mregs = mstk48t02_regs;
-#ifdef CONFIG_PCI
-	unsigned long dregs = ds1287_regs;
-	void __iomem *bregs = bq4802_regs;
-#else
-	unsigned long dregs = 0UL;
-	void __iomem *bregs = 0UL;
-#endif
-	unsigned long flags;
-	u8 tmp;
-
-	/* 
-	 * Not having a register set can lead to trouble.
-	 * Also starfire doesn't have a tod clock.
-	 */
-	if (!mregs && !dregs && !bregs)
-		return -1;
-
-	if (mregs) {
-		spin_lock_irqsave(&mostek_lock, flags);
-
-		/* Read the current RTC minutes. */
-		tmp = mostek_read(mregs + MOSTEK_CREG);
-		tmp |= MSTK_CREG_READ;
-		mostek_write(mregs + MOSTEK_CREG, tmp);
-
-		chip_minutes = MSTK_REG_MIN(mregs);
-
-		tmp = mostek_read(mregs + MOSTEK_CREG);
-		tmp &= ~MSTK_CREG_READ;
-		mostek_write(mregs + MOSTEK_CREG, tmp);
-
-		/*
-		 * since we're only adjusting minutes and seconds,
-		 * don't interfere with hour overflow. This avoids
-		 * messing with unknown time zones but requires your
-		 * RTC not to be off by more than 15 minutes
-		 */
-		real_seconds = nowtime % 60;
-		real_minutes = nowtime / 60;
-		if (((abs(real_minutes - chip_minutes) + 15)/30) & 1)
-			real_minutes += 30;	/* correct for half hour time zone */
-		real_minutes %= 60;
-
-		if (abs(real_minutes - chip_minutes) < 30) {
-			tmp = mostek_read(mregs + MOSTEK_CREG);
-			tmp |= MSTK_CREG_WRITE;
-			mostek_write(mregs + MOSTEK_CREG, tmp);
-
-			MSTK_SET_REG_SEC(mregs,real_seconds);
-			MSTK_SET_REG_MIN(mregs,real_minutes);
-
-			tmp = mostek_read(mregs + MOSTEK_CREG);
-			tmp &= ~MSTK_CREG_WRITE;
-			mostek_write(mregs + MOSTEK_CREG, tmp);
-
-			spin_unlock_irqrestore(&mostek_lock, flags);
-
-			return 0;
-		} else {
-			spin_unlock_irqrestore(&mostek_lock, flags);
-
-			return -1;
-		}
-	} else if (bregs) {
-		int retval = 0;
-		unsigned char val = readb(bregs + 0x0e);
-
-		/* BQ4802 RTC chip. */
-
-		writeb(val | 0x08, bregs + 0x0e);
-
-		chip_minutes = readb(bregs + 0x02);
-		BCD_TO_BIN(chip_minutes);
-		real_seconds = nowtime % 60;
-		real_minutes = nowtime / 60;
-		if (((abs(real_minutes - chip_minutes) + 15)/30) & 1)
-			real_minutes += 30;
-		real_minutes %= 60;
-
-		if (abs(real_minutes - chip_minutes) < 30) {
-			BIN_TO_BCD(real_seconds);
-			BIN_TO_BCD(real_minutes);
-			writeb(real_seconds, bregs + 0x00);
-			writeb(real_minutes, bregs + 0x02);
-		} else {
-			printk(KERN_WARNING
-			       "set_rtc_mmss: can't update from %d to %d\n",
-			       chip_minutes, real_minutes);
-			retval = -1;
-		}
-
-		writeb(val, bregs + 0x0e);
-
-		return retval;
-	} else {
-		int retval = 0;
-		unsigned char save_control, save_freq_select;
-
-		/* Stolen from arch/i386/kernel/time.c, see there for
-		 * credits and descriptive comments.
-		 */
-		spin_lock_irqsave(&rtc_lock, flags);
-		save_control = CMOS_READ(RTC_CONTROL); /* tell the clock it's being set */
-		CMOS_WRITE((save_control|RTC_SET), RTC_CONTROL);
-
-		save_freq_select = CMOS_READ(RTC_FREQ_SELECT); /* stop and reset prescaler */
-		CMOS_WRITE((save_freq_select|RTC_DIV_RESET2), RTC_FREQ_SELECT);
-
-		chip_minutes = CMOS_READ(RTC_MINUTES);
-		if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD)
-			BCD_TO_BIN(chip_minutes);
-		real_seconds = nowtime % 60;
-		real_minutes = nowtime / 60;
-		if (((abs(real_minutes - chip_minutes) + 15)/30) & 1)
-			real_minutes += 30;
-		real_minutes %= 60;
-
-		if (abs(real_minutes - chip_minutes) < 30) {
-			if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
-				BIN_TO_BCD(real_seconds);
-				BIN_TO_BCD(real_minutes);
-			}
-			CMOS_WRITE(real_seconds,RTC_SECONDS);
-			CMOS_WRITE(real_minutes,RTC_MINUTES);
-		} else {
-			printk(KERN_WARNING
-			       "set_rtc_mmss: can't update from %d to %d\n",
-			       chip_minutes, real_minutes);
-			retval = -1;
-		}
-
-		CMOS_WRITE(save_control, RTC_CONTROL);
-		CMOS_WRITE(save_freq_select, RTC_FREQ_SELECT);
-		spin_unlock_irqrestore(&rtc_lock, flags);
-
-		return retval;
-	}
-}
-
-#define RTC_IS_OPEN		0x01	/* means /dev/rtc is in use	*/
-static unsigned char mini_rtc_status;	/* bitmapped status byte.	*/
-
-#define FEBRUARY	2
-#define	STARTOFTIME	1970
-#define SECDAY		86400L
-#define SECYR		(SECDAY * 365)
-#define	leapyear(year)		((year) % 4 == 0 && \
-				 ((year) % 100 != 0 || (year) % 400 == 0))
-#define	days_in_year(a) 	(leapyear(a) ? 366 : 365)
-#define	days_in_month(a) 	(month_days[(a) - 1])
-
-static int month_days[12] = {
-	31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31
-};
-
-/*
- * This only works for the Gregorian calendar - i.e. after 1752 (in the UK)
- */
-static void GregorianDay(struct rtc_time * tm)
-{
-	int leapsToDate;
-	int lastYear;
-	int day;
-	int MonthOffset[] = { 0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334 };
-
-	lastYear = tm->tm_year - 1;
-
-	/*
-	 * Number of leap corrections to apply up to end of last year
-	 */
-	leapsToDate = lastYear / 4 - lastYear / 100 + lastYear / 400;
-
-	/*
-	 * This year is a leap year if it is divisible by 4 except when it is
-	 * divisible by 100 unless it is divisible by 400
-	 *
-	 * e.g. 1904 was a leap year, 1900 was not, 1996 is, and 2000 was
-	 */
-	day = tm->tm_mon > 2 && leapyear(tm->tm_year);
-
-	day += lastYear*365 + leapsToDate + MonthOffset[tm->tm_mon-1] +
-		   tm->tm_mday;
-
-	tm->tm_wday = day % 7;
-}
-
-static void to_tm(int tim, struct rtc_time *tm)
-{
-	register int    i;
-	register long   hms, day;
-
-	day = tim / SECDAY;
-	hms = tim % SECDAY;
-
-	/* Hours, minutes, seconds are easy */
-	tm->tm_hour = hms / 3600;
-	tm->tm_min = (hms % 3600) / 60;
-	tm->tm_sec = (hms % 3600) % 60;
-
-	/* Number of years in days */
-	for (i = STARTOFTIME; day >= days_in_year(i); i++)
-		day -= days_in_year(i);
-	tm->tm_year = i;
-
-	/* Number of months in days left */
-	if (leapyear(tm->tm_year))
-		days_in_month(FEBRUARY) = 29;
-	for (i = 1; day >= days_in_month(i); i++)
-		day -= days_in_month(i);
-	days_in_month(FEBRUARY) = 28;
-	tm->tm_mon = i;
-
-	/* Days are what is left over (+1) from all that. */
-	tm->tm_mday = day + 1;
-
-	/*
-	 * Determine the day of week
-	 */
-	GregorianDay(tm);
-}
-
-/* Both Starfire and SUN4V give us seconds since Jan 1st, 1970,
- * aka Unix time.  So we have to convert to/from rtc_time.
- */
-static void starfire_get_rtc_time(struct rtc_time *time)
-{
-	u32 seconds = starfire_get_time();
-
-	to_tm(seconds, time);
-	time->tm_year -= 1900;
-	time->tm_mon -= 1;
-}
-
-static int starfire_set_rtc_time(struct rtc_time *time)
-{
-	u32 seconds = mktime(time->tm_year + 1900, time->tm_mon + 1,
-			     time->tm_mday, time->tm_hour,
-			     time->tm_min, time->tm_sec);
-
-	return starfire_set_time(seconds);
-}
-
-static void hypervisor_get_rtc_time(struct rtc_time *time)
-{
-	u32 seconds = hypervisor_get_time();
-
-	to_tm(seconds, time);
-	time->tm_year -= 1900;
-	time->tm_mon -= 1;
-}
-
-static int hypervisor_set_rtc_time(struct rtc_time *time)
-{
-	u32 seconds = mktime(time->tm_year + 1900, time->tm_mon + 1,
-			     time->tm_mday, time->tm_hour,
-			     time->tm_min, time->tm_sec);
-
-	return hypervisor_set_time(seconds);
-}
-
-#ifdef CONFIG_PCI
-static void bq4802_get_rtc_time(struct rtc_time *time)
-{
-	unsigned char val = readb(bq4802_regs + 0x0e);
-	unsigned int century;
-
-	writeb(val | 0x08, bq4802_regs + 0x0e);
-
-	time->tm_sec = readb(bq4802_regs + 0x00);
-	time->tm_min = readb(bq4802_regs + 0x02);
-	time->tm_hour = readb(bq4802_regs + 0x04);
-	time->tm_mday = readb(bq4802_regs + 0x06);
-	time->tm_mon = readb(bq4802_regs + 0x09);
-	time->tm_year = readb(bq4802_regs + 0x0a);
-	time->tm_wday = readb(bq4802_regs + 0x08);
-	century = readb(bq4802_regs + 0x0f);
-
-	writeb(val, bq4802_regs + 0x0e);
-
-	BCD_TO_BIN(time->tm_sec);
-	BCD_TO_BIN(time->tm_min);
-	BCD_TO_BIN(time->tm_hour);
-	BCD_TO_BIN(time->tm_mday);
-	BCD_TO_BIN(time->tm_mon);
-	BCD_TO_BIN(time->tm_year);
-	BCD_TO_BIN(time->tm_wday);
-	BCD_TO_BIN(century);
-
-	time->tm_year += (century * 100);
-	time->tm_year -= 1900;
-
-	time->tm_mon--;
-}
-
-static int bq4802_set_rtc_time(struct rtc_time *time)
-{
-	unsigned char val = readb(bq4802_regs + 0x0e);
-	unsigned char sec, min, hrs, day, mon, yrs, century;
-	unsigned int year;
-
-	year = time->tm_year + 1900;
-	century = year / 100;
-	yrs = year % 100;
-
-	mon = time->tm_mon + 1;   /* tm_mon starts at zero */
-	day = time->tm_mday;
-	hrs = time->tm_hour;
-	min = time->tm_min;
-	sec = time->tm_sec;
-
-	BIN_TO_BCD(sec);
-	BIN_TO_BCD(min);
-	BIN_TO_BCD(hrs);
-	BIN_TO_BCD(day);
-	BIN_TO_BCD(mon);
-	BIN_TO_BCD(yrs);
-	BIN_TO_BCD(century);
-
-	writeb(val | 0x08, bq4802_regs + 0x0e);
-
-	writeb(sec, bq4802_regs + 0x00);
-	writeb(min, bq4802_regs + 0x02);
-	writeb(hrs, bq4802_regs + 0x04);
-	writeb(day, bq4802_regs + 0x06);
-	writeb(mon, bq4802_regs + 0x09);
-	writeb(yrs, bq4802_regs + 0x0a);
-	writeb(century, bq4802_regs + 0x0f);
-
-	writeb(val, bq4802_regs + 0x0e);
-
-	return 0;
-}
-
-static void cmos_get_rtc_time(struct rtc_time *rtc_tm)
-{
-	unsigned char ctrl;
-
-	rtc_tm->tm_sec = CMOS_READ(RTC_SECONDS);
-	rtc_tm->tm_min = CMOS_READ(RTC_MINUTES);
-	rtc_tm->tm_hour = CMOS_READ(RTC_HOURS);
-	rtc_tm->tm_mday = CMOS_READ(RTC_DAY_OF_MONTH);
-	rtc_tm->tm_mon = CMOS_READ(RTC_MONTH);
-	rtc_tm->tm_year = CMOS_READ(RTC_YEAR);
-	rtc_tm->tm_wday = CMOS_READ(RTC_DAY_OF_WEEK);
-
-	ctrl = CMOS_READ(RTC_CONTROL);
-	if (!(ctrl & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
-		BCD_TO_BIN(rtc_tm->tm_sec);
-		BCD_TO_BIN(rtc_tm->tm_min);
-		BCD_TO_BIN(rtc_tm->tm_hour);
-		BCD_TO_BIN(rtc_tm->tm_mday);
-		BCD_TO_BIN(rtc_tm->tm_mon);
-		BCD_TO_BIN(rtc_tm->tm_year);
-		BCD_TO_BIN(rtc_tm->tm_wday);
-	}
-
-	if (rtc_tm->tm_year <= 69)
-		rtc_tm->tm_year += 100;
-
-	rtc_tm->tm_mon--;
-}
-
-static int cmos_set_rtc_time(struct rtc_time *rtc_tm)
-{
-	unsigned char mon, day, hrs, min, sec;
-	unsigned char save_control, save_freq_select;
-	unsigned int yrs;
-
-	yrs = rtc_tm->tm_year;
-	mon = rtc_tm->tm_mon + 1;
-	day = rtc_tm->tm_mday;
-	hrs = rtc_tm->tm_hour;
-	min = rtc_tm->tm_min;
-	sec = rtc_tm->tm_sec;
-
-	if (yrs >= 100)
-		yrs -= 100;
-
-	if (!(CMOS_READ(RTC_CONTROL) & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
-		BIN_TO_BCD(sec);
-		BIN_TO_BCD(min);
-		BIN_TO_BCD(hrs);
-		BIN_TO_BCD(day);
-		BIN_TO_BCD(mon);
-		BIN_TO_BCD(yrs);
-	}
-
-	save_control = CMOS_READ(RTC_CONTROL);
-	CMOS_WRITE((save_control|RTC_SET), RTC_CONTROL);
-	save_freq_select = CMOS_READ(RTC_FREQ_SELECT);
-	CMOS_WRITE((save_freq_select|RTC_DIV_RESET2), RTC_FREQ_SELECT);
-
-	CMOS_WRITE(yrs, RTC_YEAR);
-	CMOS_WRITE(mon, RTC_MONTH);
-	CMOS_WRITE(day, RTC_DAY_OF_MONTH);
-	CMOS_WRITE(hrs, RTC_HOURS);
-	CMOS_WRITE(min, RTC_MINUTES);
-	CMOS_WRITE(sec, RTC_SECONDS);
-
-	CMOS_WRITE(save_control, RTC_CONTROL);
-	CMOS_WRITE(save_freq_select, RTC_FREQ_SELECT);
-
-	return 0;
-}
-#endif /* CONFIG_PCI */
-
-static void mostek_get_rtc_time(struct rtc_time *rtc_tm)
-{
-	void __iomem *regs = mstk48t02_regs;
-	u8 tmp;
-
-	spin_lock_irq(&mostek_lock);
-
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp |= MSTK_CREG_READ;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-
-	rtc_tm->tm_sec = MSTK_REG_SEC(regs);
-	rtc_tm->tm_min = MSTK_REG_MIN(regs);
-	rtc_tm->tm_hour = MSTK_REG_HOUR(regs);
-	rtc_tm->tm_mday = MSTK_REG_DOM(regs);
-	rtc_tm->tm_mon = MSTK_REG_MONTH(regs);
-	rtc_tm->tm_year = MSTK_CVT_YEAR( MSTK_REG_YEAR(regs) );
-	rtc_tm->tm_wday = MSTK_REG_DOW(regs);
-
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp &= ~MSTK_CREG_READ;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-
-	spin_unlock_irq(&mostek_lock);
-
-	rtc_tm->tm_mon--;
-	rtc_tm->tm_wday--;
-	rtc_tm->tm_year -= 1900;
-}
-
-static int mostek_set_rtc_time(struct rtc_time *rtc_tm)
-{
-	unsigned char mon, day, hrs, min, sec, wday;
-	void __iomem *regs = mstk48t02_regs;
-	unsigned int yrs;
-	u8 tmp;
-
-	yrs = rtc_tm->tm_year + 1900;
-	mon = rtc_tm->tm_mon + 1;
-	day = rtc_tm->tm_mday;
-	wday = rtc_tm->tm_wday + 1;
-	hrs = rtc_tm->tm_hour;
-	min = rtc_tm->tm_min;
-	sec = rtc_tm->tm_sec;
-
-	spin_lock_irq(&mostek_lock);
-
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp |= MSTK_CREG_WRITE;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-
-	MSTK_SET_REG_SEC(regs, sec);
-	MSTK_SET_REG_MIN(regs, min);
-	MSTK_SET_REG_HOUR(regs, hrs);
-	MSTK_SET_REG_DOW(regs, wday);
-	MSTK_SET_REG_DOM(regs, day);
-	MSTK_SET_REG_MONTH(regs, mon);
-	MSTK_SET_REG_YEAR(regs, yrs - MSTK_YEAR_ZERO);
-
-	tmp = mostek_read(regs + MOSTEK_CREG);
-	tmp &= ~MSTK_CREG_WRITE;
-	mostek_write(regs + MOSTEK_CREG, tmp);
-
-	spin_unlock_irq(&mostek_lock);
-
-	return 0;
-}
-
-struct mini_rtc_ops {
-	void (*get_rtc_time)(struct rtc_time *);
-	int (*set_rtc_time)(struct rtc_time *);
-};
-
-static struct mini_rtc_ops starfire_rtc_ops = {
-	.get_rtc_time = starfire_get_rtc_time,
-	.set_rtc_time = starfire_set_rtc_time,
-};
-
-static struct mini_rtc_ops hypervisor_rtc_ops = {
-	.get_rtc_time = hypervisor_get_rtc_time,
-	.set_rtc_time = hypervisor_set_rtc_time,
-};
-
-#ifdef CONFIG_PCI
-static struct mini_rtc_ops bq4802_rtc_ops = {
-	.get_rtc_time = bq4802_get_rtc_time,
-	.set_rtc_time = bq4802_set_rtc_time,
-};
-
-static struct mini_rtc_ops cmos_rtc_ops = {
-	.get_rtc_time = cmos_get_rtc_time,
-	.set_rtc_time = cmos_set_rtc_time,
-};
-#endif /* CONFIG_PCI */
-
-static struct mini_rtc_ops mostek_rtc_ops = {
-	.get_rtc_time = mostek_get_rtc_time,
-	.set_rtc_time = mostek_set_rtc_time,
-};
-
-static struct mini_rtc_ops *mini_rtc_ops;
-
-static inline void mini_get_rtc_time(struct rtc_time *time)
-{
-	unsigned long flags;
-
-	spin_lock_irqsave(&rtc_lock, flags);
-	mini_rtc_ops->get_rtc_time(time);
-	spin_unlock_irqrestore(&rtc_lock, flags);
-}
-
-static inline int mini_set_rtc_time(struct rtc_time *time)
-{
-	unsigned long flags;
-	int err;
-
-	spin_lock_irqsave(&rtc_lock, flags);
-	err = mini_rtc_ops->set_rtc_time(time);
-	spin_unlock_irqrestore(&rtc_lock, flags);
-
-	return err;
-}
-
-static int mini_rtc_ioctl(struct inode *inode, struct file *file,
-			  unsigned int cmd, unsigned long arg)
-{
-	struct rtc_time wtime;
-	void __user *argp = (void __user *)arg;
-
-	switch (cmd) {
-
-	case RTC_PLL_GET:
-		return -EINVAL;
-
-	case RTC_PLL_SET:
-		return -EINVAL;
-
-	case RTC_UIE_OFF:	/* disable ints from RTC updates.	*/
-		return 0;
-
-	case RTC_UIE_ON:	/* enable ints for RTC updates.	*/
-	        return -EINVAL;
-
-	case RTC_RD_TIME:	/* Read the time/date from RTC	*/
-		/* this doesn't get week-day, who cares */
-		memset(&wtime, 0, sizeof(wtime));
-		mini_get_rtc_time(&wtime);
-
-		return copy_to_user(argp, &wtime, sizeof(wtime)) ? -EFAULT : 0;
-
-	case RTC_SET_TIME:	/* Set the RTC */
-	    {
-		int year, days;
-
-		if (!capable(CAP_SYS_TIME))
-			return -EACCES;
-
-		if (copy_from_user(&wtime, argp, sizeof(wtime)))
-			return -EFAULT;
-
-		year = wtime.tm_year + 1900;
-		days = month_days[wtime.tm_mon] +
-		       ((wtime.tm_mon == 1) && leapyear(year));
-
-		if ((wtime.tm_mon < 0 || wtime.tm_mon > 11) ||
-		    (wtime.tm_mday < 1))
-			return -EINVAL;
-
-		if (wtime.tm_mday < 0 || wtime.tm_mday > days)
-			return -EINVAL;
-
-		if (wtime.tm_hour < 0 || wtime.tm_hour >= 24 ||
-		    wtime.tm_min < 0 || wtime.tm_min >= 60 ||
-		    wtime.tm_sec < 0 || wtime.tm_sec >= 60)
-			return -EINVAL;
-
-		return mini_set_rtc_time(&wtime);
-	    }
-	}
-
-	return -EINVAL;
-}
-
-static int mini_rtc_open(struct inode *inode, struct file *file)
-{
-	if (mini_rtc_status & RTC_IS_OPEN)
-		return -EBUSY;
-
-	mini_rtc_status |= RTC_IS_OPEN;
-
-	return 0;
-}
-
-static int mini_rtc_release(struct inode *inode, struct file *file)
-{
-	mini_rtc_status &= ~RTC_IS_OPEN;
-	return 0;
-}
-
-
-static const struct file_operations mini_rtc_fops = {
-	.owner		= THIS_MODULE,
-	.ioctl		= mini_rtc_ioctl,
-	.open		= mini_rtc_open,
-	.release	= mini_rtc_release,
-};
-
-static struct miscdevice rtc_mini_dev =
-{
-	.minor		= RTC_MINOR,
-	.name		= "rtc",
-	.fops		= &mini_rtc_fops,
-};
-
-static int __init rtc_mini_init(void)
-{
-	int retval;
-
-	if (tlb_type == hypervisor)
-		mini_rtc_ops = &hypervisor_rtc_ops;
-	else if (this_is_starfire)
-		mini_rtc_ops = &starfire_rtc_ops;
-#ifdef CONFIG_PCI
-	else if (bq4802_regs)
-		mini_rtc_ops = &bq4802_rtc_ops;
-	else if (ds1287_regs)
-		mini_rtc_ops = &cmos_rtc_ops;
-#endif /* CONFIG_PCI */
-	else if (mstk48t02_regs)
-		mini_rtc_ops = &mostek_rtc_ops;
-	else
-		return -ENODEV;
-
-	printk(KERN_INFO "Mini RTC Driver\n");
-
-	retval = misc_register(&rtc_mini_dev);
-	if (retval < 0)
-		return retval;
-
-	return 0;
-}
-
-static void __exit rtc_mini_exit(void)
-{
-	misc_deregister(&rtc_mini_dev);
-}
-
 int __devinit read_current_timer(unsigned long *timer_val)
 {
 	*timer_val = tick_ops->get_tick();
 	return 0;
 }
-
-module_init(rtc_mini_init);
-module_exit(rtc_mini_exit);
diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c
index 36974926265..81ccd22e78d 100644
--- a/arch/sparc64/kernel/traps.c
+++ b/arch/sparc64/kernel/traps.c
@@ -1,6 +1,6 @@
 /* arch/sparc64/kernel/traps.c
  *
- * Copyright (C) 1995,1997 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1995,1997,2008 David S. Miller (davem@davemloft.net)
  * Copyright (C) 1997,1999,2000 Jakub Jelinek (jakub@redhat.com)
  */
 
@@ -10,8 +10,8 @@
 
 #include <linux/module.h>
 #include <linux/sched.h>
+#include <linux/linkage.h>
 #include <linux/kernel.h>
-#include <linux/kallsyms.h>
 #include <linux/signal.h>
 #include <linux/smp.h>
 #include <linux/mm.h>
@@ -37,12 +37,11 @@
 #include <asm/processor.h>
 #include <asm/timer.h>
 #include <asm/head.h>
-#ifdef CONFIG_KMOD
-#include <linux/kmod.h>
-#endif
 #include <asm/prom.h>
+#include <asm/memctrl.h>
 
 #include "entry.h"
+#include "kstack.h"
 
 /* When an irrecoverable trap occurs at tl > 0, the trap entry
  * code logs the trap state registers at every level in the trap
@@ -74,7 +73,7 @@ static void dump_tl1_traplog(struct tl1_traplog *p)
 		       i + 1,
 		       p->trapstack[i].tstate, p->trapstack[i].tpc,
 		       p->trapstack[i].tnpc, p->trapstack[i].tt);
-		print_symbol("TRAPLOG: TPC<%s>\n", p->trapstack[i].tpc);
+		printk("TRAPLOG: TPC<%pS>\n", (void *) p->trapstack[i].tpc);
 	}
 }
 
@@ -131,6 +130,56 @@ void do_BUG(const char *file, int line)
 }
 #endif
 
+static DEFINE_SPINLOCK(dimm_handler_lock);
+static dimm_printer_t dimm_handler;
+
+static int sprintf_dimm(int synd_code, unsigned long paddr, char *buf, int buflen)
+{
+	unsigned long flags;
+	int ret = -ENODEV;
+
+	spin_lock_irqsave(&dimm_handler_lock, flags);
+	if (dimm_handler) {
+		ret = dimm_handler(synd_code, paddr, buf, buflen);
+	} else if (tlb_type == spitfire) {
+		if (prom_getunumber(synd_code, paddr, buf, buflen) == -1)
+			ret = -EINVAL;
+		else
+			ret = 0;
+	} else
+		ret = -ENODEV;
+	spin_unlock_irqrestore(&dimm_handler_lock, flags);
+
+	return ret;
+}
+
+int register_dimm_printer(dimm_printer_t func)
+{
+	unsigned long flags;
+	int ret = 0;
+
+	spin_lock_irqsave(&dimm_handler_lock, flags);
+	if (!dimm_handler)
+		dimm_handler = func;
+	else
+		ret = -EEXIST;
+	spin_unlock_irqrestore(&dimm_handler_lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(register_dimm_printer);
+
+void unregister_dimm_printer(dimm_printer_t func)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&dimm_handler_lock, flags);
+	if (dimm_handler == func)
+		dimm_handler = NULL;
+	spin_unlock_irqrestore(&dimm_handler_lock, flags);
+}
+EXPORT_SYMBOL_GPL(unregister_dimm_printer);
+
 void spitfire_insn_access_exception(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
 {
 	siginfo_t info;
@@ -293,10 +342,7 @@ void sun4v_data_access_exception_tl1(struct pt_regs *regs, unsigned long addr, u
 }
 
 #ifdef CONFIG_PCI
-/* This is really pathetic... */
-extern volatile int pci_poke_in_progress;
-extern volatile int pci_poke_cpu;
-extern volatile int pci_poke_faulted;
+#include "pci_impl.h"
 #endif
 
 /* When access exceptions happen, we must do this. */
@@ -378,8 +424,7 @@ static void spitfire_log_udb_syndrome(unsigned long afar, unsigned long udbh, un
 
 	if (udbl & bit) {
 		scode = ecc_syndrome_table[udbl & 0xff];
-		if (prom_getunumber(scode, afar,
-				    memmod_str, sizeof(memmod_str)) == -1)
+		if (sprintf_dimm(scode, afar, memmod_str, sizeof(memmod_str)) < 0)
 			p = syndrome_unknown;
 		else
 			p = memmod_str;
@@ -390,8 +435,7 @@ static void spitfire_log_udb_syndrome(unsigned long afar, unsigned long udbh, un
 
 	if (udbh & bit) {
 		scode = ecc_syndrome_table[udbh & 0xff];
-		if (prom_getunumber(scode, afar,
-				    memmod_str, sizeof(memmod_str)) == -1)
+		if (sprintf_dimm(scode, afar, memmod_str, sizeof(memmod_str)) < 0)
 			p = syndrome_unknown;
 		else
 			p = memmod_str;
@@ -1064,8 +1108,6 @@ static const char *cheetah_get_string(unsigned long bit)
 	return "???";
 }
 
-extern int chmc_getunumber(int, unsigned long, char *, int);
-
 static void cheetah_log_errors(struct pt_regs *regs, struct cheetah_err_info *info,
 			       unsigned long afsr, unsigned long afar, int recoverable)
 {
@@ -1081,7 +1123,7 @@ static void cheetah_log_errors(struct pt_regs *regs, struct cheetah_err_info *in
 	       regs->tpc, regs->tnpc, regs->u_regs[UREG_I7], regs->tstate);
 	printk("%s" "ERROR(%d): ",
 	       (recoverable ? KERN_WARNING : KERN_CRIT), smp_processor_id());
-	print_symbol("TPC<%s>\n", regs->tpc);
+	printk("TPC<%pS>\n", (void *) regs->tpc);
 	printk("%s" "ERROR(%d): M_SYND(%lx),  E_SYND(%lx)%s%s\n",
 	       (recoverable ? KERN_WARNING : KERN_CRIT), smp_processor_id(),
 	       (afsr & CHAFSR_M_SYNDROME) >> CHAFSR_M_SYNDROME_SHIFT,
@@ -1107,7 +1149,7 @@ static void cheetah_log_errors(struct pt_regs *regs, struct cheetah_err_info *in
 
 		syndrome = (afsr & CHAFSR_E_SYNDROME) >> CHAFSR_E_SYNDROME_SHIFT;
 		syndrome = cheetah_ecc_syntab[syndrome];
-		ret = chmc_getunumber(syndrome, afar, unum, sizeof(unum));
+		ret = sprintf_dimm(syndrome, afar, unum, sizeof(unum));
 		if (ret != -1)
 			printk("%s" "ERROR(%d): AFAR E-syndrome [%s]\n",
 			       (recoverable ? KERN_WARNING : KERN_CRIT),
@@ -1118,7 +1160,7 @@ static void cheetah_log_errors(struct pt_regs *regs, struct cheetah_err_info *in
 
 		syndrome = (afsr & CHAFSR_M_SYNDROME) >> CHAFSR_M_SYNDROME_SHIFT;
 		syndrome = cheetah_mtag_syntab[syndrome];
-		ret = chmc_getunumber(syndrome, afar, unum, sizeof(unum));
+		ret = sprintf_dimm(syndrome, afar, unum, sizeof(unum));
 		if (ret != -1)
 			printk("%s" "ERROR(%d): AFAR M-syndrome [%s]\n",
 			       (recoverable ? KERN_WARNING : KERN_CRIT),
@@ -1689,7 +1731,7 @@ void cheetah_plus_parity_error(int type, struct pt_regs *regs)
 		       smp_processor_id(),
 		       (type & 0x1) ? 'I' : 'D',
 		       regs->tpc);
-		print_symbol(KERN_EMERG "TPC<%s>\n", regs->tpc);
+		printk(KERN_EMERG "TPC<%pS>\n", (void *) regs->tpc);
 		panic("Irrecoverable Cheetah+ parity error.");
 	}
 
@@ -1697,7 +1739,7 @@ void cheetah_plus_parity_error(int type, struct pt_regs *regs)
 	       smp_processor_id(),
 	       (type & 0x1) ? 'I' : 'D',
 	       regs->tpc);
-	print_symbol(KERN_WARNING "TPC<%s>\n", regs->tpc);
+	printk(KERN_WARNING "TPC<%pS>\n", (void *) regs->tpc);
 }
 
 struct sun4v_error_entry {
@@ -1781,7 +1823,7 @@ static void sun4v_log_error(struct pt_regs *regs, struct sun4v_error_entry *ent,
 	       pfx,
 	       ent->err_raddr, ent->err_size, ent->err_cpu);
 
-	__show_regs(regs);
+	show_regs(regs);
 
 	if ((cnt = atomic_read(ocnt)) != 0) {
 		atomic_set(ocnt, 0);
@@ -1904,9 +1946,10 @@ void sun4v_itlb_error_report(struct pt_regs *regs, int tl)
 
 	printk(KERN_EMERG "SUN4V-ITLB: Error at TPC[%lx], tl %d\n",
 	       regs->tpc, tl);
-	print_symbol(KERN_EMERG "SUN4V-ITLB: TPC<%s>\n", regs->tpc);
+	printk(KERN_EMERG "SUN4V-ITLB: TPC<%pS>\n", (void *) regs->tpc);
 	printk(KERN_EMERG "SUN4V-ITLB: O7[%lx]\n", regs->u_regs[UREG_I7]);
-	print_symbol(KERN_EMERG "SUN4V-ITLB: O7<%s>\n", regs->u_regs[UREG_I7]);
+	printk(KERN_EMERG "SUN4V-ITLB: O7<%pS>\n",
+	       (void *) regs->u_regs[UREG_I7]);
 	printk(KERN_EMERG "SUN4V-ITLB: vaddr[%lx] ctx[%lx] "
 	       "pte[%lx] error[%lx]\n",
 	       sun4v_err_itlb_vaddr, sun4v_err_itlb_ctx,
@@ -1927,9 +1970,10 @@ void sun4v_dtlb_error_report(struct pt_regs *regs, int tl)
 
 	printk(KERN_EMERG "SUN4V-DTLB: Error at TPC[%lx], tl %d\n",
 	       regs->tpc, tl);
-	print_symbol(KERN_EMERG "SUN4V-DTLB: TPC<%s>\n", regs->tpc);
+	printk(KERN_EMERG "SUN4V-DTLB: TPC<%pS>\n", (void *) regs->tpc);
 	printk(KERN_EMERG "SUN4V-DTLB: O7[%lx]\n", regs->u_regs[UREG_I7]);
-	print_symbol(KERN_EMERG "SUN4V-DTLB: O7<%s>\n", regs->u_regs[UREG_I7]);
+	printk(KERN_EMERG "SUN4V-DTLB: O7<%pS>\n",
+	       (void *) regs->u_regs[UREG_I7]);
 	printk(KERN_EMERG "SUN4V-DTLB: vaddr[%lx] ctx[%lx] "
 	       "pte[%lx] error[%lx]\n",
 	       sun4v_err_dtlb_vaddr, sun4v_err_dtlb_ctx,
@@ -2111,23 +2155,18 @@ void show_stack(struct task_struct *tsk, unsigned long *_ksp)
 	fp = ksp + STACK_BIAS;
 	thread_base = (unsigned long) tp;
 
-	printk("Call Trace:");
-#ifdef CONFIG_KALLSYMS
-	printk("\n");
-#endif
+	printk("Call Trace:\n");
 	do {
 		struct sparc_stackf *sf;
 		struct pt_regs *regs;
 		unsigned long pc;
 
-		/* Bogus frame pointer? */
-		if (fp < (thread_base + sizeof(struct thread_info)) ||
-		    fp >= (thread_base + THREAD_SIZE))
+		if (!kstack_valid(tp, fp))
 			break;
 		sf = (struct sparc_stackf *) fp;
 		regs = (struct pt_regs *) (sf + 1);
 
-		if ((regs->magic & ~0x1ff) == PT_REGS_MAGIC) {
+		if (kstack_is_trap_frame(tp, regs)) {
 			if (!(regs->tstate & TSTATE_PRIV))
 				break;
 			pc = regs->tpc;
@@ -2137,12 +2176,8 @@ void show_stack(struct task_struct *tsk, unsigned long *_ksp)
 			fp = (unsigned long)sf->fp + STACK_BIAS;
 		}
 
-		printk(" [%016lx] ", pc);
-		print_symbol("%s\n", pc);
+		printk(" [%016lx] %pS\n", pc, (void *) pc);
 	} while (++count < 16);
-#ifndef CONFIG_KALLSYMS
-	printk("\n");
-#endif
 }
 
 void dump_stack(void)
@@ -2186,7 +2221,6 @@ static inline struct reg_window *kernel_stack_up(struct reg_window *rw)
 void die_if_kernel(char *str, struct pt_regs *regs)
 {
 	static int die_counter;
-	extern void smp_report_regs(void);
 	int count = 0;
 	
 	/* Amuse the user. */
@@ -2199,7 +2233,7 @@ void die_if_kernel(char *str, struct pt_regs *regs)
 	printk("%s(%d): %s [#%d]\n", current->comm, task_pid_nr(current), str, ++die_counter);
 	notify_die(DIE_OOPS, str, regs, 0, 255, SIGSEGV);
 	__asm__ __volatile__("flushw");
-	__show_regs(regs);
+	show_regs(regs);
 	add_taint(TAINT_DIE);
 	if (regs->tstate & TSTATE_PRIV) {
 		struct reg_window *rw = (struct reg_window *)
@@ -2211,9 +2245,8 @@ void die_if_kernel(char *str, struct pt_regs *regs)
 		while (rw &&
 		       count++ < 30&&
 		       is_kernel_stack(current, rw)) {
-			printk("Caller[%016lx]", rw->ins[7]);
-			print_symbol(": %s", rw->ins[7]);
-			printk("\n");
+			printk("Caller[%016lx]: %pS\n", rw->ins[7],
+			       (void *) rw->ins[7]);
 
 			rw = kernel_stack_up(rw);
 		}
@@ -2225,11 +2258,6 @@ void die_if_kernel(char *str, struct pt_regs *regs)
 		}
 		user_instruction_dump ((unsigned int __user *) regs->tpc);
 	}
-#if 0
-#ifdef CONFIG_SMP
-	smp_report_regs();
-#endif
-#endif                                                	
 	if (regs->tstate & TSTATE_PRIV)
 		do_exit(SIGKILL);
 	do_exit(SIGSEGV);
@@ -2240,7 +2268,6 @@ void die_if_kernel(char *str, struct pt_regs *regs)
 
 extern int handle_popc(u32 insn, struct pt_regs *regs);
 extern int handle_ldf_stq(u32 insn, struct pt_regs *regs);
-extern int vis_emul(struct pt_regs *, unsigned int);
 
 void do_illegal_instruction(struct pt_regs *regs)
 {
@@ -2470,7 +2497,7 @@ struct trap_per_cpu trap_block[NR_CPUS];
 /* This can get invoked before sched_init() so play it super safe
  * and use hard_smp_processor_id().
  */
-void init_cur_cpu_trap(struct thread_info *t)
+void notrace init_cur_cpu_trap(struct thread_info *t)
 {
 	int cpu = hard_smp_processor_id();
 	struct trap_per_cpu *p = &trap_block[cpu];
diff --git a/arch/sparc64/kernel/ttable.S b/arch/sparc64/kernel/ttable.S
index 450053af039..1ade3d6fb7f 100644
--- a/arch/sparc64/kernel/ttable.S
+++ b/arch/sparc64/kernel/ttable.S
@@ -58,7 +58,12 @@ tl0_irq3:	BTRAP(0x43)
 tl0_irq4:	BTRAP(0x44)
 #endif
 tl0_irq5:	TRAP_IRQ(handler_irq, 5)
-tl0_irq6:	BTRAP(0x46) BTRAP(0x47) BTRAP(0x48) BTRAP(0x49)
+#ifdef CONFIG_SMP
+tl0_irq6:	TRAP_IRQ(smp_call_function_single_client, 6)
+#else
+tl0_irq6:	BTRAP(0x46)
+#endif
+tl0_irq7:	BTRAP(0x47) BTRAP(0x48) BTRAP(0x49)
 tl0_irq10:	BTRAP(0x4a) BTRAP(0x4b) BTRAP(0x4c) BTRAP(0x4d)
 tl0_irq14:	TRAP_IRQ(timer_interrupt, 14)
 tl0_irq15:	TRAP_IRQ(handler_irq, 15)
diff --git a/arch/sparc64/kernel/unaligned.c b/arch/sparc64/kernel/unaligned.c
index afa7fc4f519..203ddfad9f2 100644
--- a/arch/sparc64/kernel/unaligned.c
+++ b/arch/sparc64/kernel/unaligned.c
@@ -2,7 +2,7 @@
  * unaligned.c: Unaligned load/store trap handling with special
  *              cases for the kernel to do them more quickly.
  *
- * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1996,2008 David S. Miller (davem@davemloft.net)
  * Copyright (C) 1996,1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
  */
 
@@ -20,7 +20,6 @@
 #include <asm/uaccess.h>
 #include <linux/smp.h>
 #include <linux/bitops.h>
-#include <linux/kallsyms.h>
 #include <asm/fpumacro.h>
 
 /* #define DEBUG_MNA */
@@ -289,8 +288,8 @@ static void log_unaligned(struct pt_regs *regs)
 	if (count < 5) {
 		last_time = jiffies;
 		count++;
-		printk("Kernel unaligned access at TPC[%lx] ", regs->tpc);
-		print_symbol("%s\n", regs->tpc);
+		printk("Kernel unaligned access at TPC[%lx] %pS\n",
+		       regs->tpc, (void *) regs->tpc);
 	}
 }
 
diff --git a/arch/sparc64/kernel/vio.c b/arch/sparc64/kernel/vio.c
index e78b3517940..92b1f8ec01d 100644
--- a/arch/sparc64/kernel/vio.c
+++ b/arch/sparc64/kernel/vio.c
@@ -152,7 +152,7 @@ show_pciobppath_attr(struct device *dev, struct device_attribute *attr,
 static DEVICE_ATTR(obppath, S_IRUSR | S_IRGRP | S_IROTH,
 		   show_pciobppath_attr, NULL);
 
-struct device_node *cdev_node;
+static struct device_node *cdev_node;
 
 static struct vio_dev *root_vdev;
 static u64 cdev_cfg_handle;
@@ -224,7 +224,7 @@ static struct vio_dev *vio_create_one(struct mdesc_handle *hp, u64 mp,
 	if (!strcmp(type, "domain-services-port"))
 		bus_id_name = "ds";
 
-	if (strlen(bus_id_name) >= KOBJ_NAME_LEN - 4) {
+	if (strlen(bus_id_name) >= BUS_ID_SIZE - 4) {
 		printk(KERN_ERR "VIO: bus_id_name [%s] is too long.\n",
 		       bus_id_name);
 		return NULL;
@@ -260,16 +260,14 @@ static struct vio_dev *vio_create_one(struct mdesc_handle *hp, u64 mp,
 	vio_fill_channel_info(hp, mp, vdev);
 
 	if (!id) {
-		snprintf(vdev->dev.bus_id, BUS_ID_SIZE, "%s",
-			 bus_id_name);
+		dev_set_name(&vdev->dev, "%s", bus_id_name);
 		vdev->dev_no = ~(u64)0;
 	} else if (!cfg_handle) {
-		snprintf(vdev->dev.bus_id, BUS_ID_SIZE, "%s-%lu",
-			 bus_id_name, *id);
+		dev_set_name(&vdev->dev, "%s-%lu", bus_id_name, *id);
 		vdev->dev_no = *id;
 	} else {
-		snprintf(vdev->dev.bus_id, BUS_ID_SIZE, "%s-%lu-%lu",
-			 bus_id_name, *cfg_handle, *id);
+		dev_set_name(&vdev->dev, "%s-%lu-%lu", bus_id_name,
+			     *cfg_handle, *id);
 		vdev->dev_no = *cfg_handle;
 	}
 
@@ -292,12 +290,12 @@ static struct vio_dev *vio_create_one(struct mdesc_handle *hp, u64 mp,
 	}
 	vdev->dp = dp;
 
-	printk(KERN_INFO "VIO: Adding device %s\n", vdev->dev.bus_id);
+	printk(KERN_INFO "VIO: Adding device %s\n", dev_name(&vdev->dev));
 
 	err = device_register(&vdev->dev);
 	if (err) {
 		printk(KERN_ERR "VIO: Could not register device %s, err=%d\n",
-		       vdev->dev.bus_id, err);
+		       dev_name(&vdev->dev), err);
 		kfree(vdev);
 		return NULL;
 	}
@@ -330,7 +328,7 @@ static void vio_remove(struct mdesc_handle *hp, u64 node)
 	dev = device_find_child(&root_vdev->dev, (void *) node,
 				vio_md_node_match);
 	if (dev) {
-		printk(KERN_INFO "VIO: Removing device %s\n", dev->bus_id);
+		printk(KERN_INFO "VIO: Removing device %s\n", dev_name(dev));
 
 		device_unregister(dev);
 	}
@@ -373,9 +371,9 @@ static struct mdesc_notifier_client vio_ds_notifier = {
 	.node_name	= "domain-services-port",
 };
 
-const char *channel_devices_node = "channel-devices";
-const char *channel_devices_compat = "SUNW,sun4v-channel-devices";
-const char *cfg_handle_prop = "cfg-handle";
+static const char *channel_devices_node = "channel-devices";
+static const char *channel_devices_compat = "SUNW,sun4v-channel-devices";
+static const char *cfg_handle_prop = "cfg-handle";
 
 static int __init vio_init(void)
 {
diff --git a/arch/sparc64/kernel/visemul.c b/arch/sparc64/kernel/visemul.c
index c3fd64706b5..9e05cb5cb85 100644
--- a/arch/sparc64/kernel/visemul.c
+++ b/arch/sparc64/kernel/visemul.c
@@ -243,7 +243,7 @@ static inline unsigned int *fps_regaddr(struct fpustate *f,
 struct edge_tab {
 	u16 left, right;
 };
-struct edge_tab edge8_tab[8] = {
+static struct edge_tab edge8_tab[8] = {
 	{ 0xff, 0x80 },
 	{ 0x7f, 0xc0 },
 	{ 0x3f, 0xe0 },
@@ -253,7 +253,7 @@ struct edge_tab edge8_tab[8] = {
 	{ 0x03, 0xfe },
 	{ 0x01, 0xff },
 };
-struct edge_tab edge8_tab_l[8] = {
+static struct edge_tab edge8_tab_l[8] = {
 	{ 0xff, 0x01 },
 	{ 0xfe, 0x03 },
 	{ 0xfc, 0x07 },
@@ -263,23 +263,23 @@ struct edge_tab edge8_tab_l[8] = {
 	{ 0xc0, 0x7f },
 	{ 0x80, 0xff },
 };
-struct edge_tab edge16_tab[4] = {
+static struct edge_tab edge16_tab[4] = {
 	{ 0xf, 0x8 },
 	{ 0x7, 0xc },
 	{ 0x3, 0xe },
 	{ 0x1, 0xf },
 };
-struct edge_tab edge16_tab_l[4] = {
+static struct edge_tab edge16_tab_l[4] = {
 	{ 0xf, 0x1 },
 	{ 0xe, 0x3 },
 	{ 0xc, 0x7 },
 	{ 0x8, 0xf },
 };
-struct edge_tab edge32_tab[2] = {
+static struct edge_tab edge32_tab[2] = {
 	{ 0x3, 0x2 },
 	{ 0x1, 0x3 },
 };
-struct edge_tab edge32_tab_l[2] = {
+static struct edge_tab edge32_tab_l[2] = {
 	{ 0x3, 0x1 },
 	{ 0x2, 0x3 },
 };
diff --git a/arch/sparc64/lib/copy_page.S b/arch/sparc64/lib/copy_page.S
index 37460666a5c..b243d3b606b 100644
--- a/arch/sparc64/lib/copy_page.S
+++ b/arch/sparc64/lib/copy_page.S
@@ -25,9 +25,9 @@
 
 #define	DCACHE_SIZE	(PAGE_SIZE * 2)
 
-#if (PAGE_SHIFT == 13) || (PAGE_SHIFT == 19)
+#if (PAGE_SHIFT == 13)
 #define PAGE_SIZE_REM	0x80
-#elif (PAGE_SHIFT == 16) || (PAGE_SHIFT == 22)
+#elif (PAGE_SHIFT == 16)
 #define PAGE_SIZE_REM	0x100
 #else
 #error Wrong PAGE_SHIFT specified
@@ -198,7 +198,7 @@ cheetah_copy_page_insn:
 	cmp		%o2, PAGE_SIZE_REM
 	bne,pt		%xcc, 1b
 	 add		%o0, 0x40, %o0
-#if (PAGE_SHIFT == 16) || (PAGE_SHIFT == 22)
+#if (PAGE_SHIFT == 16)
 	TOUCH(f0, f2, f4, f6, f8, f10, f12, f14)
 	ldda		[%o1] ASI_BLK_P, %f32
 	stda		%f48, [%o0] %asi
diff --git a/arch/sparc64/lib/mcount.S b/arch/sparc64/lib/mcount.S
index 9e4534b485c..fad90ddb3a2 100644
--- a/arch/sparc64/lib/mcount.S
+++ b/arch/sparc64/lib/mcount.S
@@ -28,10 +28,13 @@ ovstack:
 	.skip		OVSTACKSIZE
 #endif
 	.text
-	.align 32
-	.globl mcount, _mcount
-mcount:
+	.align		32
+	.globl		_mcount
+	.type		_mcount,#function
+	.globl		mcount
+	.type		mcount,#function
 _mcount:
+mcount:
 #ifdef CONFIG_STACK_DEBUG
 	/*
 	 * Check whether %sp is dangerously low.
@@ -45,16 +48,96 @@ _mcount:
 	sub		%g3, STACK_BIAS, %g3
 	cmp		%sp, %g3
 	bg,pt		%xcc, 1f
-	 sethi		%hi(panicstring), %g3
+	 nop
+	lduh		[%g6 + TI_CPU], %g1
+	sethi		%hi(hardirq_stack), %g3
+	or		%g3, %lo(hardirq_stack), %g3
+	sllx		%g1, 3, %g1
+	ldx		[%g3 + %g1], %g7
+	sub		%g7, STACK_BIAS, %g7
+	cmp		%sp, %g7
+	bleu,pt		%xcc, 2f
+	 sethi		%hi(THREAD_SIZE), %g3
+	add		%g7, %g3, %g7
+	cmp		%sp, %g7
+	blu,pn		%xcc, 1f
+2:	 sethi		%hi(softirq_stack), %g3
+	or		%g3, %lo(softirq_stack), %g3
+	ldx		[%g3 + %g1], %g7
+	cmp		%sp, %g7
+	bleu,pt		%xcc, 2f
+	 sethi		%hi(THREAD_SIZE), %g3
+	add		%g7, %g3, %g7
+	cmp		%sp, %g7
+	blu,pn		%xcc, 1f
+	 nop
+	/* If we are already on ovstack, don't hop onto it
+	 * again, we are already trying to output the stack overflow
+	 * message.
+	 */
 	sethi		%hi(ovstack), %g7		! cant move to panic stack fast enough
 	 or		%g7, %lo(ovstack), %g7
-	add		%g7, OVSTACKSIZE, %g7
+	add		%g7, OVSTACKSIZE, %g3
+	sub		%g3, STACK_BIAS + 192, %g3
 	sub		%g7, STACK_BIAS, %g7
-	mov		%g7, %sp
+	cmp		%sp, %g7
+	blu,pn		%xcc, 2f
+	 cmp		%sp, %g3
+	bleu,pn		%xcc, 1f
+	 nop
+2:	mov		%g3, %sp
+	sethi		%hi(panicstring), %g3
 	call		prom_printf
 	 or		%g3, %lo(panicstring), %o0
 	call		prom_halt
 	 nop
+1:
 #endif
-1:	retl
+#ifdef CONFIG_FTRACE
+#ifdef CONFIG_DYNAMIC_FTRACE
+	mov		%o7, %o0
+	.globl		mcount_call
+mcount_call:
+	call		ftrace_stub
+	 mov		%o0, %o7
+#else
+	sethi		%hi(ftrace_trace_function), %g1
+	sethi		%hi(ftrace_stub), %g2
+	ldx		[%g1 + %lo(ftrace_trace_function)], %g1
+	or		%g2, %lo(ftrace_stub), %g2
+	cmp		%g1, %g2
+	be,pn		%icc, 1f
+	 mov		%i7, %o1
+	jmpl		%g1, %g0
+	 mov		%o7, %o0
+	/* not reached */
+1:
+#endif
+#endif
+	retl
+	 nop
+	.size		_mcount,.-_mcount
+	.size		mcount,.-mcount
+
+#ifdef CONFIG_FTRACE
+	.globl		ftrace_stub
+	.type		ftrace_stub,#function
+ftrace_stub:
+	retl
 	 nop
+	.size		ftrace_stub,.-ftrace_stub
+#ifdef CONFIG_DYNAMIC_FTRACE
+	.globl		ftrace_caller
+	.type		ftrace_caller,#function
+ftrace_caller:
+	mov		%i7, %o1
+	mov		%o7, %o0
+	.globl		ftrace_call
+ftrace_call:
+	call		ftrace_stub
+	 mov		%o0, %o7
+	retl
+	 nop
+	.size		ftrace_caller,.-ftrace_caller
+#endif
+#endif
diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c
index 236f4d228d2..a9e474bf638 100644
--- a/arch/sparc64/mm/fault.c
+++ b/arch/sparc64/mm/fault.c
@@ -1,7 +1,7 @@
 /*
  * arch/sparc64/mm/fault.c: Page fault handlers for the 64-bit Sparc.
  *
- * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1996, 2008 David S. Miller (davem@davemloft.net)
  * Copyright (C) 1997, 1999 Jakub Jelinek (jj@ultra.linux.cz)
  */
 
@@ -18,7 +18,6 @@
 #include <linux/init.h>
 #include <linux/interrupt.h>
 #include <linux/kprobes.h>
-#include <linux/kallsyms.h>
 #include <linux/kdebug.h>
 
 #include <asm/page.h>
@@ -52,43 +51,6 @@ static inline int notify_page_fault(struct pt_regs *regs)
 }
 #endif
 
-/*
- * To debug kernel to catch accesses to certain virtual/physical addresses.
- * Mode = 0 selects physical watchpoints, mode = 1 selects virtual watchpoints.
- * flags = VM_READ watches memread accesses, flags = VM_WRITE watches memwrite accesses.
- * Caller passes in a 64bit aligned addr, with mask set to the bytes that need to be
- * watched. This is only useful on a single cpu machine for now. After the watchpoint
- * is detected, the process causing it will be killed, thus preventing an infinite loop.
- */
-void set_brkpt(unsigned long addr, unsigned char mask, int flags, int mode)
-{
-	unsigned long lsubits;
-
-	__asm__ __volatile__("ldxa [%%g0] %1, %0"
-			     : "=r" (lsubits)
-			     : "i" (ASI_LSU_CONTROL));
-	lsubits &= ~(LSU_CONTROL_PM | LSU_CONTROL_VM |
-		     LSU_CONTROL_PR | LSU_CONTROL_VR |
-		     LSU_CONTROL_PW | LSU_CONTROL_VW);
-
-	__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
-			     "membar	#Sync"
-			     : /* no outputs */
-			     : "r" (addr), "r" (mode ? VIRT_WATCHPOINT : PHYS_WATCHPOINT),
-			       "i" (ASI_DMMU));
-
-	lsubits |= ((unsigned long)mask << (mode ? 25 : 33));
-	if (flags & VM_READ)
-		lsubits |= (mode ? LSU_CONTROL_VR : LSU_CONTROL_PR);
-	if (flags & VM_WRITE)
-		lsubits |= (mode ? LSU_CONTROL_VW : LSU_CONTROL_PW);
-	__asm__ __volatile__("stxa %0, [%%g0] %1\n\t"
-			     "membar #Sync"
-			     : /* no outputs */
-			     : "r" (lsubits), "i" (ASI_LSU_CONTROL)
-			     : "memory");
-}
-
 static void __kprobes unhandled_fault(unsigned long address,
 				      struct task_struct *tsk,
 				      struct pt_regs *regs)
@@ -115,7 +77,7 @@ static void bad_kernel_pc(struct pt_regs *regs, unsigned long vaddr)
 	printk(KERN_CRIT "OOPS: Bogus kernel PC [%016lx] in fault handler\n",
 	       regs->tpc);
 	printk(KERN_CRIT "OOPS: RPC [%016lx]\n", regs->u_regs[15]);
-	print_symbol("RPC: <%s>\n", regs->u_regs[15]);
+	printk("OOPS: RPC <%pS>\n", (void *) regs->u_regs[15]);
 	printk(KERN_CRIT "OOPS: Fault was to vaddr[%lx]\n", vaddr);
 	dump_stack();
 	unhandled_fault(regs->tpc, current, regs);
diff --git a/arch/sparc64/mm/hugetlbpage.c b/arch/sparc64/mm/hugetlbpage.c
index 6cfab2e4d34..f27d10369e0 100644
--- a/arch/sparc64/mm/hugetlbpage.c
+++ b/arch/sparc64/mm/hugetlbpage.c
@@ -175,7 +175,7 @@ hugetlb_get_unmapped_area(struct file *file, unsigned long addr,
 		return -ENOMEM;
 
 	if (flags & MAP_FIXED) {
-		if (prepare_hugepage_range(addr, len))
+		if (prepare_hugepage_range(file, addr, len))
 			return -EINVAL;
 		return addr;
 	}
@@ -195,7 +195,8 @@ hugetlb_get_unmapped_area(struct file *file, unsigned long addr,
 				pgoff, flags);
 }
 
-pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr)
+pte_t *huge_pte_alloc(struct mm_struct *mm,
+			unsigned long addr, unsigned long sz)
 {
 	pgd_t *pgd;
 	pud_t *pud;
@@ -294,6 +295,11 @@ int pmd_huge(pmd_t pmd)
 	return 0;
 }
 
+int pud_huge(pud_t pud)
+{
+	return 0;
+}
+
 struct page *follow_huge_pmd(struct mm_struct *mm, unsigned long address,
 			     pmd_t *pmd, int write)
 {
@@ -344,7 +350,7 @@ void hugetlb_prefault_arch_hook(struct mm_struct *mm)
 			 * also executing in this address space.
 			 */
 			mm->context.sparc64_ctx_val = ctx;
-			on_each_cpu(context_reload, mm, 0, 0);
+			on_each_cpu(context_reload, mm, 0);
 		}
 		spin_unlock(&ctx_alloc_lock);
 	}
diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c
index 84898c44dd4..3c10daf8fc0 100644
--- a/arch/sparc64/mm/init.c
+++ b/arch/sparc64/mm/init.c
@@ -46,14 +46,11 @@
 #include <asm/tsb.h>
 #include <asm/hypervisor.h>
 #include <asm/prom.h>
-#include <asm/sstate.h>
 #include <asm/mdesc.h>
 #include <asm/cpudata.h>
+#include <asm/irq.h>
 
-#define MAX_PHYS_ADDRESS	(1UL << 42UL)
-#define KPTE_BITMAP_CHUNK_SZ	(256UL * 1024UL * 1024UL)
-#define KPTE_BITMAP_BYTES	\
-	((MAX_PHYS_ADDRESS / KPTE_BITMAP_CHUNK_SZ) / 8)
+#include "init.h"
 
 unsigned long kern_linear_pte_xor[2] __read_mostly;
 
@@ -392,51 +389,6 @@ void __kprobes flush_icache_range(unsigned long start, unsigned long end)
 	}
 }
 
-void show_mem(void)
-{
-	unsigned long total = 0, reserved = 0;
-	unsigned long shared = 0, cached = 0;
-	pg_data_t *pgdat;
-
-	printk(KERN_INFO "Mem-info:\n");
-	show_free_areas();
-	printk(KERN_INFO "Free swap:       %6ldkB\n",
-	       nr_swap_pages << (PAGE_SHIFT-10));
-	for_each_online_pgdat(pgdat) {
-		unsigned long i, flags;
-
-		pgdat_resize_lock(pgdat, &flags);
-		for (i = 0; i < pgdat->node_spanned_pages; i++) {
-			struct page *page = pgdat_page_nr(pgdat, i);
-			total++;
-			if (PageReserved(page))
-				reserved++;
-			else if (PageSwapCache(page))
-				cached++;
-			else if (page_count(page))
-				shared += page_count(page) - 1;
-		}
-		pgdat_resize_unlock(pgdat, &flags);
-	}
-
-	printk(KERN_INFO "%lu pages of RAM\n", total);
-	printk(KERN_INFO "%lu reserved pages\n", reserved);
-	printk(KERN_INFO "%lu pages shared\n", shared);
-	printk(KERN_INFO "%lu pages swap cached\n", cached);
-
-	printk(KERN_INFO "%lu pages dirty\n",
-	       global_page_state(NR_FILE_DIRTY));
-	printk(KERN_INFO "%lu pages writeback\n",
-	       global_page_state(NR_WRITEBACK));
-	printk(KERN_INFO "%lu pages mapped\n",
-	       global_page_state(NR_FILE_MAPPED));
-	printk(KERN_INFO "%lu pages slab\n",
-		global_page_state(NR_SLAB_RECLAIMABLE) +
-		global_page_state(NR_SLAB_UNRECLAIMABLE));
-	printk(KERN_INFO "%lu pages pagetables\n",
-	       global_page_state(NR_PAGETABLE));
-}
-
 void mmu_info(struct seq_file *m)
 {
 	if (tlb_type == cheetah)
@@ -460,17 +412,9 @@ void mmu_info(struct seq_file *m)
 #endif /* CONFIG_DEBUG_DCFLUSH */
 }
 
-struct linux_prom_translation {
-	unsigned long virt;
-	unsigned long size;
-	unsigned long data;
-};
-
-/* Exported for kernel TLB miss handling in ktlb.S */
 struct linux_prom_translation prom_trans[512] __read_mostly;
 unsigned int prom_trans_ents __read_mostly;
 
-/* Exported for SMP bootup purposes. */
 unsigned long kern_locked_tte_data;
 
 /* The obp translations are saved based on 8k pagesize, since obp can
@@ -788,7 +732,6 @@ int numa_cpu_lookup_table[NR_CPUS];
 cpumask_t numa_cpumask_lookup_table[MAX_NUMNODES];
 
 #ifdef CONFIG_NEED_MULTIPLE_NODES
-static bootmem_data_t plat_node_bdata[MAX_NUMNODES];
 
 struct mdesc_mblock {
 	u64	base;
@@ -841,6 +784,9 @@ static unsigned long nid_range(unsigned long start, unsigned long end,
 		start += PAGE_SIZE;
 	}
 
+	if (start > end)
+		start = end;
+
 	return start;
 }
 #else
@@ -871,7 +817,7 @@ static void __init allocate_node_data(int nid)
 	NODE_DATA(nid) = __va(paddr);
 	memset(NODE_DATA(nid), 0, sizeof(struct pglist_data));
 
-	NODE_DATA(nid)->bdata = &plat_node_bdata[nid];
+	NODE_DATA(nid)->bdata = &bootmem_node_data[nid];
 #endif
 
 	p = NODE_DATA(nid);
@@ -980,6 +926,10 @@ int of_node_to_nid(struct device_node *dp)
 	int count, nid;
 	u64 grp;
 
+	/* This is the right thing to do on currently supported
+	 * SUN4U NUMA platforms as well, as the PCI controller does
+	 * not sit behind any particular memory controller.
+	 */
 	if (!mlgroups)
 		return -1;
 
@@ -1248,8 +1198,44 @@ out:
 	return err;
 }
 
+static int __init numa_parse_jbus(void)
+{
+	unsigned long cpu, index;
+
+	/* NUMA node id is encoded in bits 36 and higher, and there is
+	 * a 1-to-1 mapping from CPU ID to NUMA node ID.
+	 */
+	index = 0;
+	for_each_present_cpu(cpu) {
+		numa_cpu_lookup_table[cpu] = index;
+		numa_cpumask_lookup_table[index] = cpumask_of_cpu(cpu);
+		node_masks[index].mask = ~((1UL << 36UL) - 1UL);
+		node_masks[index].val = cpu << 36UL;
+
+		index++;
+	}
+	num_node_masks = index;
+
+	add_node_ranges();
+
+	for (index = 0; index < num_node_masks; index++) {
+		allocate_node_data(index);
+		node_set_online(index);
+	}
+
+	return 0;
+}
+
 static int __init numa_parse_sun4u(void)
 {
+	if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		unsigned long ver;
+
+		__asm__ ("rdpr %%ver, %0" : "=r" (ver));
+		if ((ver >> 32UL) == __JALAPENO_ID ||
+		    (ver >> 32UL) == __SERRANO_ID)
+			return numa_parse_jbus();
+	}
 	return -1;
 }
 
@@ -1675,8 +1661,6 @@ void __cpuinit sun4v_ktsb_register(void)
 
 /* paging_init() sets up the page tables */
 
-extern void central_probe(void);
-
 static unsigned long last_valid_pfn;
 pgd_t swapper_pg_dir[2048];
 
@@ -1721,8 +1705,6 @@ void __init paging_init(void)
 	kern_base = (prom_boot_mapping_phys_low >> 22UL) << 22UL;
 	kern_size = (unsigned long)&_end - (unsigned long)KERNBASE;
 
-	sstate_booting();
-
 	/* Invalidate both kernel TSBs.  */
 	memset(swapper_tsb, 0x40, sizeof(swapper_tsb));
 #ifndef CONFIG_DEBUG_PAGEALLOC
@@ -1768,8 +1750,7 @@ void __init paging_init(void)
 
 	find_ramdisk(phys_base);
 
-	if (cmdline_memory_size)
-		lmb_enforce_memory_limit(phys_base + cmdline_memory_size);
+	lmb_enforce_memory_limit(cmdline_memory_size);
 
 	lmb_analyze();
 	lmb_dump_all();
@@ -1817,6 +1798,16 @@ void __init paging_init(void)
 	if (tlb_type == hypervisor)
 		sun4v_mdesc_init();
 
+	/* Once the OF device tree and MDESC have been setup, we know
+	 * the list of possible cpus.  Therefore we can allocate the
+	 * IRQ stacks.
+	 */
+	for_each_possible_cpu(i) {
+		/* XXX Use node local allocations... XXX */
+		softirq_stack[i] = __va(lmb_alloc(THREAD_SIZE, THREAD_SIZE));
+		hardirq_stack[i] = __va(lmb_alloc(THREAD_SIZE, THREAD_SIZE));
+	}
+
 	/* Setup bootmem... */
 	last_valid_pfn = end_pfn = bootmem_init(phys_base);
 
@@ -1836,9 +1827,6 @@ void __init paging_init(void)
 	}
 
 	printk("Booting Linux...\n");
-
-	central_probe();
-	cpu_probe();
 }
 
 int __init page_in_phys_avail(unsigned long paddr)
@@ -1876,7 +1864,7 @@ static int pavail_rescan_ents __initdata;
  * memory list again, and make sure it provides at least as much
  * memory as 'pavail' does.
  */
-static void setup_valid_addr_bitmap_from_pavail(void)
+static void __init setup_valid_addr_bitmap_from_pavail(void)
 {
 	int i;
 
@@ -1996,6 +1984,15 @@ void __init mem_init(void)
 void free_initmem(void)
 {
 	unsigned long addr, initend;
+	int do_free = 1;
+
+	/* If the physical memory maps were trimmed by kernel command
+	 * line options, don't even try freeing this initmem stuff up.
+	 * The kernel image could have been in the trimmed out region
+	 * and if so the freeing below will free invalid page structs.
+	 */
+	if (cmdline_memory_size)
+		do_free = 0;
 
 	/*
 	 * The init section is aligned to 8k in vmlinux.lds. Page align for >8k pagesizes.
@@ -2010,13 +2007,16 @@ void free_initmem(void)
 			((unsigned long) __va(kern_base)) -
 			((unsigned long) KERNBASE));
 		memset((void *)addr, POISON_FREE_INITMEM, PAGE_SIZE);
-		p = virt_to_page(page);
 
-		ClearPageReserved(p);
-		init_page_count(p);
-		__free_page(p);
-		num_physpages++;
-		totalram_pages++;
+		if (do_free) {
+			p = virt_to_page(page);
+
+			ClearPageReserved(p);
+			init_page_count(p);
+			__free_page(p);
+			num_physpages++;
+			totalram_pages++;
+		}
 	}
 }
 
@@ -2053,7 +2053,6 @@ pgprot_t PAGE_COPY __read_mostly;
 pgprot_t PAGE_SHARED __read_mostly;
 EXPORT_SYMBOL(PAGE_SHARED);
 
-pgprot_t PAGE_EXEC __read_mostly;
 unsigned long pg_iobits __read_mostly;
 
 unsigned long _PAGE_IE __read_mostly;
@@ -2066,14 +2065,6 @@ unsigned long _PAGE_CACHE __read_mostly;
 EXPORT_SYMBOL(_PAGE_CACHE);
 
 #ifdef CONFIG_SPARSEMEM_VMEMMAP
-
-#define VMEMMAP_CHUNK_SHIFT	22
-#define VMEMMAP_CHUNK		(1UL << VMEMMAP_CHUNK_SHIFT)
-#define VMEMMAP_CHUNK_MASK	~(VMEMMAP_CHUNK - 1UL)
-#define VMEMMAP_ALIGN(x)	(((x)+VMEMMAP_CHUNK-1UL)&VMEMMAP_CHUNK_MASK)
-
-#define VMEMMAP_SIZE	((((1UL << MAX_PHYSADDR_BITS) >> PAGE_SHIFT) * \
-			  sizeof(struct page *)) >> VMEMMAP_CHUNK_SHIFT)
 unsigned long vmemmap_table[VMEMMAP_SIZE];
 
 int __meminit vmemmap_populate(struct page *start, unsigned long nr, int node)
@@ -2157,7 +2148,6 @@ static void __init sun4u_pgprot_init(void)
 				       _PAGE_CACHE_4U | _PAGE_P_4U |
 				       __ACCESS_BITS_4U | __DIRTY_BITS_4U |
 				       _PAGE_EXEC_4U | _PAGE_L_4U);
-	PAGE_EXEC = __pgprot(_PAGE_EXEC_4U);
 
 	_PAGE_IE = _PAGE_IE_4U;
 	_PAGE_E = _PAGE_E_4U;
@@ -2168,10 +2158,10 @@ static void __init sun4u_pgprot_init(void)
 
 #ifdef CONFIG_DEBUG_PAGEALLOC
 	kern_linear_pte_xor[0] = (_PAGE_VALID | _PAGE_SZBITS_4U) ^
-		0xfffff80000000000;
+		0xfffff80000000000UL;
 #else
 	kern_linear_pte_xor[0] = (_PAGE_VALID | _PAGE_SZ4MB_4U) ^
-		0xfffff80000000000;
+		0xfffff80000000000UL;
 #endif
 	kern_linear_pte_xor[0] |= (_PAGE_CP_4U | _PAGE_CV_4U |
 				   _PAGE_P_4U | _PAGE_W_4U);
@@ -2209,7 +2199,6 @@ static void __init sun4v_pgprot_init(void)
 				__ACCESS_BITS_4V | __DIRTY_BITS_4V |
 				_PAGE_EXEC_4V);
 	PAGE_KERNEL_LOCKED = PAGE_KERNEL;
-	PAGE_EXEC = __pgprot(_PAGE_EXEC_4V);
 
 	_PAGE_IE = _PAGE_IE_4V;
 	_PAGE_E = _PAGE_E_4V;
@@ -2217,20 +2206,20 @@ static void __init sun4v_pgprot_init(void)
 
 #ifdef CONFIG_DEBUG_PAGEALLOC
 	kern_linear_pte_xor[0] = (_PAGE_VALID | _PAGE_SZBITS_4V) ^
-		0xfffff80000000000;
+		0xfffff80000000000UL;
 #else
 	kern_linear_pte_xor[0] = (_PAGE_VALID | _PAGE_SZ4MB_4V) ^
-		0xfffff80000000000;
+		0xfffff80000000000UL;
 #endif
 	kern_linear_pte_xor[0] |= (_PAGE_CP_4V | _PAGE_CV_4V |
 				   _PAGE_P_4V | _PAGE_W_4V);
 
 #ifdef CONFIG_DEBUG_PAGEALLOC
 	kern_linear_pte_xor[1] = (_PAGE_VALID | _PAGE_SZBITS_4V) ^
-		0xfffff80000000000;
+		0xfffff80000000000UL;
 #else
 	kern_linear_pte_xor[1] = (_PAGE_VALID | _PAGE_SZ256MB_4V) ^
-		0xfffff80000000000;
+		0xfffff80000000000UL;
 #endif
 	kern_linear_pte_xor[1] |= (_PAGE_CP_4V | _PAGE_CV_4V |
 				   _PAGE_P_4V | _PAGE_W_4V);
diff --git a/arch/sparc64/mm/init.h b/arch/sparc64/mm/init.h
new file mode 100644
index 00000000000..16063870a48
--- /dev/null
+++ b/arch/sparc64/mm/init.h
@@ -0,0 +1,49 @@
+#ifndef _SPARC64_MM_INIT_H
+#define _SPARC64_MM_INIT_H
+
+/* Most of the symbols in this file are defined in init.c and
+ * marked non-static so that assembler code can get at them.
+ */
+
+#define MAX_PHYS_ADDRESS	(1UL << 42UL)
+#define KPTE_BITMAP_CHUNK_SZ	(256UL * 1024UL * 1024UL)
+#define KPTE_BITMAP_BYTES	\
+	((MAX_PHYS_ADDRESS / KPTE_BITMAP_CHUNK_SZ) / 8)
+
+extern unsigned long kern_linear_pte_xor[2];
+extern unsigned long kpte_linear_bitmap[KPTE_BITMAP_BYTES / sizeof(unsigned long)];
+extern unsigned int sparc64_highest_unlocked_tlb_ent;
+extern unsigned long sparc64_kern_pri_context;
+extern unsigned long sparc64_kern_pri_nuc_bits;
+extern unsigned long sparc64_kern_sec_context;
+extern void mmu_info(struct seq_file *m);
+
+struct linux_prom_translation {
+	unsigned long virt;
+	unsigned long size;
+	unsigned long data;
+};
+
+/* Exported for kernel TLB miss handling in ktlb.S */
+extern struct linux_prom_translation prom_trans[512];
+extern unsigned int prom_trans_ents;
+
+/* Exported for SMP bootup purposes. */
+extern unsigned long kern_locked_tte_data;
+
+extern void prom_world(int enter);
+
+extern void free_initmem(void);
+
+#ifdef CONFIG_SPARSEMEM_VMEMMAP
+#define VMEMMAP_CHUNK_SHIFT	22
+#define VMEMMAP_CHUNK		(1UL << VMEMMAP_CHUNK_SHIFT)
+#define VMEMMAP_CHUNK_MASK	~(VMEMMAP_CHUNK - 1UL)
+#define VMEMMAP_ALIGN(x)	(((x)+VMEMMAP_CHUNK-1UL)&VMEMMAP_CHUNK_MASK)
+
+#define VMEMMAP_SIZE	((((1UL << MAX_PHYSADDR_BITS) >> PAGE_SHIFT) * \
+			  sizeof(struct page *)) >> VMEMMAP_CHUNK_SHIFT)
+extern unsigned long vmemmap_table[VMEMMAP_SIZE];
+#endif
+
+#endif /* _SPARC64_MM_INIT_H */
diff --git a/arch/sparc64/mm/tlb.c b/arch/sparc64/mm/tlb.c
index ae24919cba7..d8f21e24a82 100644
--- a/arch/sparc64/mm/tlb.c
+++ b/arch/sparc64/mm/tlb.c
@@ -19,7 +19,7 @@
 
 /* Heavily inspired by the ppc64 code.  */
 
-DEFINE_PER_CPU(struct mmu_gather, mmu_gathers) = { 0, };
+DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
 
 void flush_tlb_pending(void)
 {
diff --git a/arch/sparc64/mm/tsb.c b/arch/sparc64/mm/tsb.c
index fe70c8a557b..587f8efb2e0 100644
--- a/arch/sparc64/mm/tsb.c
+++ b/arch/sparc64/mm/tsb.c
@@ -1,9 +1,10 @@
 /* arch/sparc64/mm/tsb.c
  *
- * Copyright (C) 2006 David S. Miller <davem@davemloft.net>
+ * Copyright (C) 2006, 2008 David S. Miller <davem@davemloft.net>
  */
 
 #include <linux/kernel.h>
+#include <linux/preempt.h>
 #include <asm/system.h>
 #include <asm/page.h>
 #include <asm/tlbflush.h>
@@ -96,12 +97,6 @@ void flush_tsb_user(struct mmu_gather *mp)
 #elif defined(CONFIG_SPARC64_PAGE_SIZE_64KB)
 #define HV_PGSZ_IDX_BASE	HV_PGSZ_IDX_64K
 #define HV_PGSZ_MASK_BASE	HV_PGSZ_MASK_64K
-#elif defined(CONFIG_SPARC64_PAGE_SIZE_512KB)
-#define HV_PGSZ_IDX_BASE	HV_PGSZ_IDX_512K
-#define HV_PGSZ_MASK_BASE	HV_PGSZ_MASK_512K
-#elif defined(CONFIG_SPARC64_PAGE_SIZE_4MB)
-#define HV_PGSZ_IDX_BASE	HV_PGSZ_IDX_4MB
-#define HV_PGSZ_MASK_BASE	HV_PGSZ_MASK_4MB
 #else
 #error Broken base page size setting...
 #endif
@@ -421,7 +416,9 @@ retry_tsb_alloc:
 		tsb_context_switch(mm);
 
 		/* Now force other processors to do the same.  */
+		preempt_disable();
 		smp_tsb_sync(mm);
+		preempt_enable();
 
 		/* Now it is safe to free the old tsb.  */
 		kmem_cache_free(tsb_caches[old_cache_index], old_tsb);
diff --git a/arch/sparc64/mm/ultra.S b/arch/sparc64/mm/ultra.S
index 9bb2d90a9df..86773e89dc1 100644
--- a/arch/sparc64/mm/ultra.S
+++ b/arch/sparc64/mm/ultra.S
@@ -480,42 +480,6 @@ xcall_sync_tick:
 	b		rtrap_xcall
 	 ldx		[%sp + PTREGS_OFF + PT_V9_TSTATE], %l1
 
-	/* NOTE: This is SPECIAL!!  We do etrap/rtrap however
-	 *       we choose to deal with the "BH's run with
-	 *       %pil==15" problem (described in asm/pil.h)
-	 *       by just invoking rtrap directly past where
-	 *       BH's are checked for.
-	 *
-	 *       We do it like this because we do not want %pil==15
-	 *       lockups to prevent regs being reported.
-	 */
-	.globl		xcall_report_regs
-xcall_report_regs:
-
-661:	rdpr		%pstate, %g2
-	wrpr		%g2, PSTATE_IG | PSTATE_AG, %pstate
-	.section	.sun4v_2insn_patch, "ax"
-	.word		661b
-	nop
-	nop
-	.previous
-
-	rdpr		%pil, %g2
-	wrpr		%g0, 15, %pil
-	sethi		%hi(109f), %g7
-	b,pt		%xcc, etrap_irq
-109:	 or		%g7, %lo(109b), %g7
-#ifdef CONFIG_TRACE_IRQFLAGS
-	call		trace_hardirqs_off
-	 nop
-#endif
-	call		__show_regs
-	 add		%sp, PTREGS_OFF, %o0
-	/* Has to be a non-v9 branch due to the large distance. */
-	b		rtrap_xcall
-	 ldx		[%sp + PTREGS_OFF + PT_V9_TSTATE], %l1
-
-#ifdef CONFIG_MAGIC_SYSRQ
 	.globl		xcall_fetch_glob_regs
 xcall_fetch_glob_regs:
 	sethi		%hi(global_reg_snapshot), %g1
@@ -531,6 +495,13 @@ xcall_fetch_glob_regs:
 	stx		%g7, [%g1 + GR_SNAP_TNPC]
 	stx		%o7, [%g1 + GR_SNAP_O7]
 	stx		%i7, [%g1 + GR_SNAP_I7]
+	/* Don't try this at home kids... */
+	rdpr		%cwp, %g2
+	sub		%g2, 1, %g7
+	wrpr		%g7, %cwp
+	mov		%i7, %g7
+	wrpr		%g2, %cwp
+	stx		%g7, [%g1 + GR_SNAP_RPC]
 	sethi		%hi(trap_block), %g7
 	or		%g7, %lo(trap_block), %g7
 	sllx		%g2, TRAP_BLOCK_SZ_SHIFT, %g2
@@ -539,7 +510,6 @@ xcall_fetch_glob_regs:
 	membar		#StoreStore
 	stx		%g3, [%g1 + GR_SNAP_THREAD]
 	retry
-#endif /* CONFIG_MAGIC_SYSRQ */
 
 #ifdef DCACHE_ALIASING_POSSIBLE
 	.align		32
@@ -688,6 +658,11 @@ xcall_call_function:
 	wr		%g0, (1 << PIL_SMP_CALL_FUNC), %set_softint
 	retry
 
+	.globl		xcall_call_function_single
+xcall_call_function_single:
+	wr		%g0, (1 << PIL_SMP_CALL_FUNC_SNGL), %set_softint
+	retry
+
 	.globl		xcall_receive_signal
 xcall_receive_signal:
 	wr		%g0, (1 << PIL_SMP_RECEIVE_SIGNAL), %set_softint