Merge branch 'master'

94 files changed, 27247 insertions, 0 deletions

diff --git a/arch/powerpc/platforms/4xx/Kconfig b/arch/powerpc/platforms/4xx/Kconfig
new file mode 100644
index 000000000000..ed39d6a3d22a
--- /dev/null
+++ b/arch/powerpc/platforms/4xx/Kconfig
@@ -0,0 +1,280 @@
+config 4xx
+	bool
+	depends on 40x || 44x
+	default y
+
+config WANT_EARLY_SERIAL
+	bool
+	select SERIAL_8250
+	default n
+
+menu "AMCC 4xx options"
+	depends on 4xx
+
+choice
+	prompt "Machine Type"
+	depends on 40x
+	default WALNUT
+
+config BUBINGA
+	bool "Bubinga"
+	select WANT_EARLY_SERIAL
+	help
+	  This option enables support for the IBM 405EP evaluation board.
+
+config CPCI405
+	bool "CPCI405"
+	help
+	  This option enables support for the CPCI405 board.
+
+config EP405
+	bool "EP405/EP405PC"
+	help
+	  This option enables support for the EP405/EP405PC boards.
+
+config REDWOOD_5
+	bool "Redwood-5"
+	help
+	  This option enables support for the IBM STB04 evaluation board.
+
+config REDWOOD_6
+	bool "Redwood-6"
+	help
+	  This option enables support for the IBM STBx25xx evaluation board.
+
+config SYCAMORE
+	bool "Sycamore"
+	help
+	  This option enables support for the IBM PPC405GPr evaluation board.
+
+config WALNUT
+	bool "Walnut"
+	help
+	  This option enables support for the IBM PPC405GP evaluation board.
+
+config XILINX_ML300
+	bool "Xilinx-ML300"
+	help
+	  This option enables support for the Xilinx ML300 evaluation board.
+
+endchoice
+
+choice
+	prompt "Machine Type"
+	depends on 44x
+	default EBONY
+
+config BAMBOO
+	bool "Bamboo"
+	select WANT_EARLY_SERIAL
+	help
+	  This option enables support for the IBM PPC440EP evaluation board.
+
+config EBONY
+	bool "Ebony"
+	select WANT_EARLY_SERIAL
+	help
+	  This option enables support for the IBM PPC440GP evaluation board.
+
+config LUAN
+	bool "Luan"
+	select WANT_EARLY_SERIAL
+	help
+	  This option enables support for the IBM PPC440SP evaluation board.
+
+config OCOTEA
+	bool "Ocotea"
+	select WANT_EARLY_SERIAL
+	help
+	  This option enables support for the IBM PPC440GX evaluation board.
+
+endchoice
+
+config EP405PC
+	bool "EP405PC Support"
+	depends on EP405
+
+
+# It's often necessary to know the specific 4xx processor type.
+# Fortunately, it is impled (so far) from the board type, so we
+# don't need to ask more redundant questions.
+config NP405H
+	bool
+	depends on ASH
+	default y
+
+config 440EP
+	bool
+	depends on BAMBOO
+	select PPC_FPU
+	default y
+
+config 440GP
+	bool
+	depends on EBONY
+	default y
+
+config 440GX
+	bool
+	depends on OCOTEA
+	default y
+
+config 440SP
+	bool
+	depends on LUAN
+	default y
+
+config 440
+	bool
+	depends on 440GP || 440SP || 440EP
+	default y
+
+config 440A
+	bool
+	depends on 440GX
+	default y
+
+config IBM440EP_ERR42
+	bool
+	depends on 440EP
+	default y
+
+# All 405-based cores up until the 405GPR and 405EP have this errata.
+config IBM405_ERR77
+	bool
+	depends on 40x && !403GCX && !405GPR && !405EP
+	default y
+
+# All 40x-based cores, up until the 405GPR and 405EP have this errata.
+config IBM405_ERR51
+	bool
+	depends on 40x && !405GPR && !405EP
+	default y
+
+config BOOKE
+	bool
+	depends on 44x
+	default y
+
+config IBM_OCP
+	bool
+	depends on ASH || BAMBOO || BUBINGA || CPCI405 || EBONY || EP405 || LUAN || OCOTEA || REDWOOD_5 || REDWOOD_6 || SYCAMORE || WALNUT
+	default y
+
+config XILINX_OCP
+	bool
+	depends on XILINX_ML300
+	default y
+
+config IBM_EMAC4
+	bool
+	depends on 440GX || 440SP
+	default y
+
+config BIOS_FIXUP
+	bool
+	depends on BUBINGA || EP405 || SYCAMORE || WALNUT
+	default y
+
+# OAK doesn't exist but wanted to keep this around for any future 403GCX boards
+config 403GCX
+	bool
+	depends OAK
+	default y
+
+config 405EP
+	bool
+	depends on BUBINGA
+	default y
+
+config 405GP
+	bool
+	depends on CPCI405 || EP405 || WALNUT
+	default y
+
+config 405GPR
+	bool
+	depends on SYCAMORE
+	default y
+
+config VIRTEX_II_PRO
+	bool
+	depends on XILINX_ML300
+	default y
+
+config STB03xxx
+	bool
+	depends on REDWOOD_5 || REDWOOD_6
+	default y
+
+config EMBEDDEDBOOT
+	bool
+	depends on EP405 || XILINX_ML300
+	default y
+
+config IBM_OPENBIOS
+	bool
+	depends on ASH || BUBINGA || REDWOOD_5 || REDWOOD_6 || SYCAMORE || WALNUT
+	default y
+
+config PPC4xx_DMA
+	bool "PPC4xx DMA controller support"
+	depends on 4xx
+
+config PPC4xx_EDMA
+	bool
+	depends on !STB03xxx && PPC4xx_DMA
+	default y
+
+config PPC_GEN550
+	bool
+	depends on 4xx
+	default y
+
+choice
+	prompt "TTYS0 device and default console"
+	depends on 40x
+	default UART0_TTYS0
+
+config UART0_TTYS0
+	bool "UART0"
+
+config UART0_TTYS1
+	bool "UART1"
+
+endchoice
+
+config SERIAL_SICC
+	bool "SICC Serial port support"
+	depends on STB03xxx
+
+config UART1_DFLT_CONSOLE
+	bool
+	depends on SERIAL_SICC && UART0_TTYS1
+	default y
+
+config SERIAL_SICC_CONSOLE
+	bool
+	depends on SERIAL_SICC && UART0_TTYS1
+	default y
+endmenu
+
+
+menu "IBM 40x options"
+	depends on 40x
+
+config SERIAL_SICC
+	bool "SICC Serial port"
+	depends on STB03xxx
+
+config UART1_DFLT_CONSOLE
+	bool
+	depends on SERIAL_SICC && UART0_TTYS1
+	default y
+
+config SERIAL_SICC_CONSOLE
+	bool
+	depends on SERIAL_SICC && UART0_TTYS1
+	default y
+
+endmenu
diff --git a/arch/powerpc/platforms/4xx/Makefile b/arch/powerpc/platforms/4xx/Makefile
new file mode 100644
index 000000000000..79ff6b1e887c
--- /dev/null
+++ b/arch/powerpc/platforms/4xx/Makefile
@@ -0,0 +1 @@
+# empty makefile so make clean works
\ No newline at end of file
diff --git a/arch/powerpc/platforms/85xx/Kconfig b/arch/powerpc/platforms/85xx/Kconfig
new file mode 100644
index 000000000000..c5bc2821d991
--- /dev/null
+++ b/arch/powerpc/platforms/85xx/Kconfig
@@ -0,0 +1,86 @@
+config 85xx
+	bool
+	depends on E500
+	default y
+
+config PPC_INDIRECT_PCI_BE
+	bool
+	depends on 85xx
+	default y
+
+menu "Freescale 85xx options"
+	depends on E500
+
+choice
+	prompt "Machine Type"
+	depends on 85xx
+	default MPC8540_ADS
+
+config MPC8540_ADS
+	bool "Freescale MPC8540 ADS"
+	help
+	  This option enables support for the MPC 8540 ADS evaluation board.
+
+config MPC8548_CDS
+	bool "Freescale MPC8548 CDS"
+	help
+	  This option enablese support for the MPC8548 CDS evaluation board.
+
+config MPC8555_CDS
+	bool "Freescale MPC8555 CDS"
+	help
+	  This option enablese support for the MPC8555 CDS evaluation board.
+
+config MPC8560_ADS
+	bool "Freescale MPC8560 ADS"
+	help
+	  This option enables support for the MPC 8560 ADS evaluation board.
+
+config SBC8560
+	bool "WindRiver PowerQUICC III SBC8560"
+	help
+	  This option enables support for the WindRiver PowerQUICC III 
+	  SBC8560 board.
+
+config STX_GP3
+	bool "Silicon Turnkey Express GP3"
+	help
+	  This option enables support for the Silicon Turnkey Express GP3
+	  board.
+
+endchoice
+
+# It's often necessary to know the specific 85xx processor type.
+# Fortunately, it is implied (so far) from the board type, so we
+# don't need to ask more redundant questions.
+config MPC8540
+	bool
+	depends on MPC8540_ADS
+	default y
+
+config MPC8548
+	bool
+	depends on MPC8548_CDS
+	default y
+
+config MPC8555
+	bool
+	depends on MPC8555_CDS
+	default y
+
+config MPC8560
+	bool
+	depends on SBC8560 || MPC8560_ADS || STX_GP3
+	default y
+
+config 85xx_PCI2
+	bool "Supprt for 2nd PCI host controller"
+	depends on MPC8555_CDS
+	default y
+
+config PPC_GEN550
+	bool
+	depends on MPC8540 || SBC8560 || MPC8555
+	default y
+
+endmenu
diff --git a/arch/powerpc/platforms/85xx/Makefile b/arch/powerpc/platforms/85xx/Makefile
new file mode 100644
index 000000000000..6407197ffd89
--- /dev/null
+++ b/arch/powerpc/platforms/85xx/Makefile
@@ -0,0 +1 @@
+# empty makefile so make clean works
diff --git a/arch/powerpc/platforms/8xx/Kconfig b/arch/powerpc/platforms/8xx/Kconfig
new file mode 100644
index 000000000000..c8c0ba3cf8e8
--- /dev/null
+++ b/arch/powerpc/platforms/8xx/Kconfig
@@ -0,0 +1,352 @@
+config FADS
+	bool
+
+choice
+	prompt "8xx Machine Type"
+	depends on 8xx
+	default RPXLITE
+
+config RPXLITE
+	bool "RPX-Lite"
+	---help---
+	  Single-board computers based around the PowerPC MPC8xx chips and
+	  intended for embedded applications.  The following types are
+	  supported:
+
+	  RPX-Lite:
+	  Embedded Planet RPX Lite. PC104 form-factor SBC based on the MPC823.
+
+	  RPX-Classic:
+	  Embedded Planet RPX Classic Low-fat. Credit-card-size SBC based on
+	  the MPC 860
+
+	  BSE-IP:
+	  Bright Star Engineering ip-Engine.
+
+	  TQM823L:
+	  TQM850L:
+	  TQM855L:
+	  TQM860L:
+	  MPC8xx based family of mini modules, half credit card size,
+	  up to 64 MB of RAM, 8 MB Flash, (Fast) Ethernet, 2 x serial ports,
+	  2 x CAN bus interface, ...
+	  Manufacturer: TQ Components, www.tq-group.de
+	  Date of Release: October (?) 1999
+	  End of Life: not yet :-)
+	  URL:
+	  - module: <http://www.denx.de/PDF/TQM8xxLHWM201.pdf>
+	  - starter kit: <http://www.denx.de/PDF/STK8xxLHWM201.pdf>
+	  - images: <http://www.denx.de/embedded-ppc-en.html>
+
+	  FPS850L:
+	  FingerPrint Sensor System (based on TQM850L)
+	  Manufacturer: IKENDI AG, <http://www.ikendi.com/>
+	  Date of Release: November 1999
+	  End of life: end 2000 ?
+	  URL: see TQM850L
+
+	  IVMS8:
+	  MPC860 based board used in the "Integrated Voice Mail System",
+	  Small Version (8 voice channels)
+	  Manufacturer: Speech Design, <http://www.speech-design.de/>
+	  Date of Release: December 2000 (?)
+	  End of life: -
+	  URL: <http://www.speech-design.de/>
+
+	  IVML24:
+	  MPC860 based board used in the "Integrated Voice Mail System",
+	  Large Version (24 voice channels)
+	  Manufacturer: Speech Design, <http://www.speech-design.de/>
+	  Date of Release: March 2001  (?)
+	  End of life: -
+	  URL: <http://www.speech-design.de/>
+
+	  HERMES:
+	  Hermes-Pro ISDN/LAN router with integrated 8 x hub
+	  Manufacturer: Multidata Gesellschaft fur Datentechnik und Informatik
+	  <http://www.multidata.de/>
+	  Date of Release: 2000 (?)
+	  End of life: -
+	  URL: <http://www.multidata.de/english/products/hpro.htm>
+
+	  IP860:
+	  VMEBus IP (Industry Pack) carrier board with MPC860
+	  Manufacturer: MicroSys GmbH, <http://www.microsys.de/>
+	  Date of Release: ?
+	  End of life: -
+	  URL: <http://www.microsys.de/html/ip860.html>
+
+	  PCU_E:
+	  PCU = Peripheral Controller Unit, Extended
+	  Manufacturer: Siemens AG, ICN (Information and Communication Networks)
+	  	<http://www.siemens.de/page/1,3771,224315-1-999_2_226207-0,00.html>
+	  Date of Release: April 2001
+	  End of life: August 2001
+	  URL: n. a.
+
+config RPXCLASSIC
+	bool "RPX-Classic"
+	help
+	  The RPX-Classic is a single-board computer based on the Motorola
+	  MPC860.  It features 16MB of DRAM and a variable amount of flash,
+	  I2C EEPROM, thermal monitoring, a PCMCIA slot, a DIP switch and two
+	  LEDs.  Variants with Ethernet ports exist.  Say Y here to support it
+	  directly.
+
+config BSEIP
+	bool "BSE-IP"
+	help
+	  Say Y here to support the Bright Star Engineering ipEngine SBC.
+	  This is a credit-card-sized device featuring a MPC823 processor,
+	  26MB DRAM, 4MB flash, Ethernet, a 16K-gate FPGA, USB, an LCD/video
+	  controller, and two RS232 ports.
+
+config MPC8XXFADS
+	bool "FADS"
+	select FADS
+
+config MPC86XADS
+	bool "MPC86XADS"
+	help
+	  MPC86x Application Development System by Freescale Semiconductor.
+	  The MPC86xADS is meant to serve as a platform for s/w and h/w
+	  development around the MPC86X processor families.
+	select FADS
+
+config MPC885ADS
+	bool "MPC885ADS"
+	help
+	  Freescale Semiconductor MPC885 Application Development System (ADS).
+	  Also known as DUET.
+	  The MPC885ADS is meant to serve as a platform for s/w and h/w
+	  development around the MPC885 processor family.
+
+config TQM823L
+	bool "TQM823L"
+	help
+	  Say Y here to support the TQM823L, one of an MPC8xx-based family of
+	  mini SBCs (half credit-card size) from TQ Components first released
+	  in late 1999.  Technical references are at
+	  <http://www.denx.de/PDF/TQM8xxLHWM201.pdf>, and
+	  <http://www.denx.de/PDF/STK8xxLHWM201.pdf>, and an image at
+	  <http://www.denx.de/embedded-ppc-en.html>.
+
+config TQM850L
+	bool "TQM850L"
+	help
+	  Say Y here to support the TQM850L, one of an MPC8xx-based family of
+	  mini SBCs (half credit-card size) from TQ Components first released
+	  in late 1999.  Technical references are at
+	  <http://www.denx.de/PDF/TQM8xxLHWM201.pdf>, and
+	  <http://www.denx.de/PDF/STK8xxLHWM201.pdf>, and an image at
+	  <http://www.denx.de/embedded-ppc-en.html>.
+
+config TQM855L
+	bool "TQM855L"
+	help
+	  Say Y here to support the TQM855L, one of an MPC8xx-based family of
+	  mini SBCs (half credit-card size) from TQ Components first released
+	  in late 1999.  Technical references are at
+	  <http://www.denx.de/PDF/TQM8xxLHWM201.pdf>, and
+	  <http://www.denx.de/PDF/STK8xxLHWM201.pdf>, and an image at
+	  <http://www.denx.de/embedded-ppc-en.html>.
+
+config TQM860L
+	bool "TQM860L"
+	help
+	  Say Y here to support the TQM860L, one of an MPC8xx-based family of
+	  mini SBCs (half credit-card size) from TQ Components first released
+	  in late 1999.  Technical references are at
+	  <http://www.denx.de/PDF/TQM8xxLHWM201.pdf>, and
+	  <http://www.denx.de/PDF/STK8xxLHWM201.pdf>, and an image at
+	  <http://www.denx.de/embedded-ppc-en.html>.
+
+config FPS850L
+	bool "FPS850L"
+
+config IVMS8
+	bool "IVMS8"
+	help
+	  Say Y here to support the Integrated Voice-Mail Small 8-channel SBC
+	  from Speech Design, released March 2001.  The manufacturer's website
+	  is at <http://www.speech-design.de/>.
+
+config IVML24
+	bool "IVML24"
+	help
+	  Say Y here to support the Integrated Voice-Mail Large 24-channel SBC
+	  from Speech Design, released March 2001.  The manufacturer's website
+	  is at <http://www.speech-design.de/>.
+
+config HERMES_PRO
+	bool "HERMES"
+
+config IP860
+	bool "IP860"
+
+config LWMON
+	bool "LWMON"
+
+config PCU_E
+	bool "PCU_E"
+
+config CCM
+	bool "CCM"
+
+config LANTEC
+	bool "LANTEC"
+
+config MBX
+	bool "MBX"
+	help
+	  MBX is a line of Motorola single-board computer based around the
+	  MPC821 and MPC860 processors, and intended for embedded-controller
+	  applications.  Say Y here to support these boards directly.
+
+config WINCEPT
+	bool "WinCept"
+	help
+	  The Wincept 100/110 is a Motorola single-board computer based on the
+	  MPC821 PowerPC, introduced in 1998 and designed to be used in
+	  thin-client machines.  Say Y to support it directly.
+
+endchoice
+
+#
+# MPC8xx Communication options
+#
+
+menu "MPC8xx CPM Options"
+	depends on 8xx
+
+config SCC_ENET
+	bool "CPM SCC Ethernet"
+	depends on NET_ETHERNET
+	help
+	  Enable Ethernet support via the Motorola MPC8xx serial
+	  communications controller.
+
+choice
+	prompt "SCC used for Ethernet"
+	depends on SCC_ENET
+	default SCC1_ENET
+
+config SCC1_ENET
+	bool "SCC1"
+	help
+	  Use MPC8xx serial communications controller 1 to drive Ethernet
+	  (default).
+
+config SCC2_ENET
+	bool "SCC2"
+	help
+	  Use MPC8xx serial communications controller 2 to drive Ethernet.
+
+config SCC3_ENET
+	bool "SCC3"
+	help
+	  Use MPC8xx serial communications controller 3 to drive Ethernet.
+
+endchoice
+
+config FEC_ENET
+	bool "860T FEC Ethernet"
+	depends on NET_ETHERNET
+	help
+	  Enable Ethernet support via the Fast Ethernet Controller (FCC) on
+	  the Motorola MPC8260.
+
+config USE_MDIO
+	bool "Use MDIO for PHY configuration"
+	depends on FEC_ENET
+	help
+	  On some boards the hardware configuration of the ethernet PHY can be
+	  used without any software interaction over the MDIO interface, so
+	  all MII code can be omitted. Say N here if unsure or if you don't
+	  need link status reports.
+
+config  FEC_AM79C874
+	bool "Support AMD79C874 PHY"
+	depends on USE_MDIO
+
+config FEC_LXT970
+	bool "Support LXT970 PHY"
+	depends on USE_MDIO
+
+config FEC_LXT971
+	bool "Support LXT971 PHY"
+	depends on USE_MDIO
+	
+config FEC_QS6612
+	bool "Support QS6612 PHY"
+	depends on USE_MDIO
+	
+config ENET_BIG_BUFFERS
+	bool "Use Big CPM Ethernet Buffers"
+	depends on SCC_ENET || FEC_ENET
+	help
+	  Allocate large buffers for MPC8xx Ethernet. Increases throughput
+	  and decreases the likelihood of dropped packets, but costs memory.
+
+config HTDMSOUND
+	bool "Embedded Planet HIOX Audio"
+	depends on SOUND=y
+
+# This doesn't really belong here, but it is convenient to ask
+# 8xx specific questions.
+comment "Generic MPC8xx Options"
+
+config 8xx_COPYBACK
+	bool "Copy-Back Data Cache (else Writethrough)"
+	help
+	  Saying Y here will cause the cache on an MPC8xx processor to be used
+	  in Copy-Back mode.  If you say N here, it is used in Writethrough
+	  mode.
+
+	  If in doubt, say Y here.
+
+config 8xx_CPU6
+	bool "CPU6 Silicon Errata (860 Pre Rev. C)"
+	help
+	  MPC860 CPUs, prior to Rev C have some bugs in the silicon, which
+	  require workarounds for Linux (and most other OSes to work).  If you
+	  get a BUG() very early in boot, this might fix the problem.  For
+	  more details read the document entitled "MPC860 Family Device Errata
+	  Reference" on Motorola's website.  This option also incurs a
+	  performance hit.
+
+	  If in doubt, say N here.
+
+choice
+	prompt "Microcode patch selection"
+	default NO_UCODE_PATCH
+	help
+	  Help not implemented yet, coming soon.
+
+config NO_UCODE_PATCH
+	bool "None"
+
+config USB_SOF_UCODE_PATCH
+	bool "USB SOF patch"
+	help
+	  Help not implemented yet, coming soon.
+
+config I2C_SPI_UCODE_PATCH
+	bool "I2C/SPI relocation patch"
+	help
+	  Help not implemented yet, coming soon.
+
+config I2C_SPI_SMC1_UCODE_PATCH
+	bool "I2C/SPI/SMC1 relocation patch"
+	help
+	  Help not implemented yet, coming soon.
+
+endchoice
+
+config UCODE_PATCH
+	bool
+	default y
+	depends on !NO_UCODE_PATCH
+
+endmenu
+
diff --git a/arch/powerpc/platforms/Makefile b/arch/powerpc/platforms/Makefile
new file mode 100644
index 000000000000..8836b3a00668
--- /dev/null
+++ b/arch/powerpc/platforms/Makefile
@@ -0,0 +1,14 @@
+ifeq ($(CONFIG_PPC_MERGE),y)
+obj-$(CONFIG_PPC_PMAC)		+= powermac/
+else
+ifeq ($(CONFIG_PPC64),y)
+obj-$(CONFIG_PPC_PMAC)		+= powermac/
+endif
+endif
+obj-$(CONFIG_PPC_CHRP)		+= chrp/
+obj-$(CONFIG_4xx)		+= 4xx/
+obj-$(CONFIG_85xx)		+= 85xx/
+obj-$(CONFIG_PPC_PSERIES)	+= pseries/
+obj-$(CONFIG_PPC_ISERIES)	+= iseries/
+obj-$(CONFIG_PPC_MAPLE)		+= maple/
+obj-$(CONFIG_PPC_CELL)		+= cell/
diff --git a/arch/powerpc/platforms/apus/Kconfig b/arch/powerpc/platforms/apus/Kconfig
new file mode 100644
index 000000000000..6bde3bffed86
--- /dev/null
+++ b/arch/powerpc/platforms/apus/Kconfig
@@ -0,0 +1,130 @@
+
+config AMIGA
+	bool
+	depends on APUS
+	default y
+	help
+	  This option enables support for the Amiga series of computers.
+
+config ZORRO
+	bool
+	depends on APUS
+	default y
+	help
+	  This enables support for the Zorro bus in the Amiga. If you have
+	  expansion cards in your Amiga that conform to the Amiga
+	  AutoConfig(tm) specification, say Y, otherwise N. Note that even
+	  expansion cards that do not fit in the Zorro slots but fit in e.g.
+	  the CPU slot may fall in this category, so you have to say Y to let
+	  Linux use these.
+
+config ABSTRACT_CONSOLE
+	bool
+	depends on APUS
+	default y
+
+config APUS_FAST_EXCEPT
+	bool
+	depends on APUS
+	default y
+
+config AMIGA_PCMCIA
+	bool "Amiga 1200/600 PCMCIA support"
+	depends on APUS && EXPERIMENTAL
+	help
+	  Include support in the kernel for pcmcia on Amiga 1200 and Amiga
+	  600. If you intend to use pcmcia cards say Y; otherwise say N.
+
+config AMIGA_BUILTIN_SERIAL
+	tristate "Amiga builtin serial support"
+	depends on APUS
+	help
+	  If you want to use your Amiga's built-in serial port in Linux,
+	  answer Y.
+
+	  To compile this driver as a module, choose M here.
+
+config GVPIOEXT
+	tristate "GVP IO-Extender support"
+	depends on APUS
+	help
+	  If you want to use a GVP IO-Extender serial card in Linux, say Y.
+	  Otherwise, say N.
+
+config GVPIOEXT_LP
+	tristate "GVP IO-Extender parallel printer support"
+	depends on GVPIOEXT
+	help
+	  Say Y to enable driving a printer from the parallel port on your
+	  GVP IO-Extender card, N otherwise.
+
+config GVPIOEXT_PLIP
+	tristate "GVP IO-Extender PLIP support"
+	depends on GVPIOEXT
+	help
+	  Say Y to enable doing IP over the parallel port on your GVP
+	  IO-Extender card, N otherwise.
+
+config MULTIFACE_III_TTY
+	tristate "Multiface Card III serial support"
+	depends on APUS
+	help
+	  If you want to use a Multiface III card's serial port in Linux,
+	  answer Y.
+
+	  To compile this driver as a module, choose M here.
+
+config A2232
+	tristate "Commodore A2232 serial support (EXPERIMENTAL)"
+	depends on EXPERIMENTAL && APUS
+	---help---
+	  This option supports the 2232 7-port serial card shipped with the
+	  Amiga 2000 and other Zorro-bus machines, dating from 1989.  At
+	  a max of 19,200 bps, the ports are served by a 6551 ACIA UART chip
+	  each, plus a 8520 CIA, and a master 6502 CPU and buffer as well. The
+	  ports were connected with 8 pin DIN connectors on the card bracket,
+	  for which 8 pin to DB25 adapters were supplied. The card also had
+	  jumpers internally to toggle various pinning configurations.
+
+	  This driver can be built as a module; but then "generic_serial"
+	  will also be built as a module. This has to be loaded before
+	  "ser_a2232". If you want to do this, answer M here.
+
+config WHIPPET_SERIAL
+	tristate "Hisoft Whippet PCMCIA serial support"
+	depends on AMIGA_PCMCIA
+	help
+	  HiSoft has a web page at <http://www.hisoft.co.uk/>, but there
+	  is no listing for the Whippet in their Amiga section.
+
+config APNE
+	tristate "PCMCIA NE2000 support"
+	depends on AMIGA_PCMCIA
+	help
+	  If you have a PCMCIA NE2000 compatible adapter, say Y.  Otherwise,
+	  say N.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called apne.
+
+config SERIAL_CONSOLE
+	bool "Support for serial port console"
+	depends on APUS && (AMIGA_BUILTIN_SERIAL=y || GVPIOEXT=y || MULTIFACE_III_TTY=y)
+
+config HEARTBEAT
+	bool "Use power LED as a heartbeat"
+	depends on APUS
+	help
+	  Use the power-on LED on your machine as a load meter.  The exact
+	  behavior is platform-dependent, but normally the flash frequency is
+	  a hyperbolic function of the 5-minute load average.
+
+config PROC_HARDWARE
+	bool "/proc/hardware support"
+	depends on APUS
+
+source "drivers/zorro/Kconfig"
+
+config PCI_PERMEDIA
+	bool "PCI for Permedia2"
+	depends on !4xx && !8xx && APUS
diff --git a/arch/powerpc/platforms/cell/Makefile b/arch/powerpc/platforms/cell/Makefile
new file mode 100644
index 000000000000..55e094b96bc0
--- /dev/null
+++ b/arch/powerpc/platforms/cell/Makefile
@@ -0,0 +1,2 @@
+obj-y			+= interrupt.o iommu.o setup.o spider-pic.o
+obj-$(CONFIG_SMP)	+= smp.o
diff --git a/arch/powerpc/platforms/cell/interrupt.c b/arch/powerpc/platforms/cell/interrupt.c
new file mode 100644
index 000000000000..7fbe78a9327d
--- /dev/null
+++ b/arch/powerpc/platforms/cell/interrupt.c
@@ -0,0 +1,284 @@
+/*
+ * Cell Internal Interrupt Controller
+ *
+ * (C) Copyright IBM Deutschland Entwicklung GmbH 2005
+ *
+ * Author: Arnd Bergmann <arndb@de.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/config.h>
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+#include <linux/percpu.h>
+#include <linux/types.h>
+
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/prom.h>
+#include <asm/ptrace.h>
+
+#include "interrupt.h"
+
+struct iic_pending_bits {
+	u32 data;
+	u8 flags;
+	u8 class;
+	u8 source;
+	u8 prio;
+};
+
+enum iic_pending_flags {
+	IIC_VALID = 0x80,
+	IIC_IPI   = 0x40,
+};
+
+struct iic_regs {
+	struct iic_pending_bits pending;
+	struct iic_pending_bits pending_destr;
+	u64 generate;
+	u64 prio;
+};
+
+struct iic {
+	struct iic_regs __iomem *regs;
+};
+
+static DEFINE_PER_CPU(struct iic, iic);
+
+void iic_local_enable(void)
+{
+	out_be64(&__get_cpu_var(iic).regs->prio, 0xff);
+}
+
+void iic_local_disable(void)
+{
+	out_be64(&__get_cpu_var(iic).regs->prio, 0x0);
+}
+
+static unsigned int iic_startup(unsigned int irq)
+{
+	return 0;
+}
+
+static void iic_enable(unsigned int irq)
+{
+	iic_local_enable();
+}
+
+static void iic_disable(unsigned int irq)
+{
+}
+
+static void iic_end(unsigned int irq)
+{
+	iic_local_enable();
+}
+
+static struct hw_interrupt_type iic_pic = {
+	.typename = " CELL-IIC ",
+	.startup = iic_startup,
+	.enable = iic_enable,
+	.disable = iic_disable,
+	.end = iic_end,
+};
+
+static int iic_external_get_irq(struct iic_pending_bits pending)
+{
+	int irq;
+	unsigned char node, unit;
+
+	node = pending.source >> 4;
+	unit = pending.source & 0xf;
+	irq = -1;
+
+	/*
+	 * This mapping is specific to the Cell Broadband
+	 * Engine. We might need to get the numbers
+	 * from the device tree to support future CPUs.
+	 */
+	switch (unit) {
+	case 0x00:
+	case 0x0b:
+		/*
+		 * One of these units can be connected
+		 * to an external interrupt controller.
+		 */
+		if (pending.prio > 0x3f ||
+		    pending.class != 2)
+			break;
+		irq = IIC_EXT_OFFSET
+			+ spider_get_irq(pending.prio + node * IIC_NODE_STRIDE)
+			+ node * IIC_NODE_STRIDE;
+		break;
+	case 0x01 ... 0x04:
+	case 0x07 ... 0x0a:
+		/*
+		 * These units are connected to the SPEs
+		 */
+		if (pending.class > 2)
+			break;
+		irq = IIC_SPE_OFFSET
+			+ pending.class * IIC_CLASS_STRIDE
+			+ node * IIC_NODE_STRIDE
+			+ unit;
+		break;
+	}
+	if (irq == -1)
+		printk(KERN_WARNING "Unexpected interrupt class %02x, "
+			"source %02x, prio %02x, cpu %02x\n", pending.class,
+			pending.source, pending.prio, smp_processor_id());
+	return irq;
+}
+
+/* Get an IRQ number from the pending state register of the IIC */
+int iic_get_irq(struct pt_regs *regs)
+{
+	struct iic *iic;
+	int irq;
+	struct iic_pending_bits pending;
+
+	iic = &__get_cpu_var(iic);
+	*(unsigned long *) &pending = 
+		in_be64((unsigned long __iomem *) &iic->regs->pending_destr);
+
+	irq = -1;
+	if (pending.flags & IIC_VALID) {
+		if (pending.flags & IIC_IPI) {
+			irq = IIC_IPI_OFFSET + (pending.prio >> 4);
+/*
+			if (irq > 0x80)
+				printk(KERN_WARNING "Unexpected IPI prio %02x"
+					"on CPU %02x\n", pending.prio,
+							smp_processor_id());
+*/
+		} else {
+			irq = iic_external_get_irq(pending);
+		}
+	}
+	return irq;
+}
+
+static struct iic_regs __iomem *find_iic(int cpu)
+{
+	struct device_node *np;
+	int nodeid = cpu / 2;
+	unsigned long regs;
+	struct iic_regs __iomem *iic_regs;
+
+	for (np = of_find_node_by_type(NULL, "cpu");
+	     np;
+	     np = of_find_node_by_type(np, "cpu")) {
+		if (nodeid == *(int *)get_property(np, "node-id", NULL))
+			break;
+	}
+
+	if (!np) {
+		printk(KERN_WARNING "IIC: CPU %d not found\n", cpu);
+		iic_regs = NULL;
+	} else {
+		regs = *(long *)get_property(np, "iic", NULL);
+
+		/* hack until we have decided on the devtree info */
+		regs += 0x400;
+		if (cpu & 1)
+			regs += 0x20;
+
+		printk(KERN_DEBUG "IIC for CPU %d at %lx\n", cpu, regs);
+		iic_regs = __ioremap(regs, sizeof(struct iic_regs),
+						 _PAGE_NO_CACHE);
+	}
+	return iic_regs;
+}
+
+#ifdef CONFIG_SMP
+
+/* Use the highest interrupt priorities for IPI */
+static inline int iic_ipi_to_irq(int ipi)
+{
+	return IIC_IPI_OFFSET + IIC_NUM_IPIS - 1 - ipi;
+}
+
+static inline int iic_irq_to_ipi(int irq)
+{
+	return IIC_NUM_IPIS - 1 - (irq - IIC_IPI_OFFSET);
+}
+
+void iic_setup_cpu(void)
+{
+	out_be64(&__get_cpu_var(iic).regs->prio, 0xff);
+}
+
+void iic_cause_IPI(int cpu, int mesg)
+{
+	out_be64(&per_cpu(iic, cpu).regs->generate, (IIC_NUM_IPIS - 1 - mesg) << 4);
+}
+
+static irqreturn_t iic_ipi_action(int irq, void *dev_id, struct pt_regs *regs)
+{
+	smp_message_recv(iic_irq_to_ipi(irq), regs);
+	return IRQ_HANDLED;
+}
+
+static void iic_request_ipi(int ipi, const char *name)
+{
+	int irq;
+
+	irq = iic_ipi_to_irq(ipi);
+	/* IPIs are marked SA_INTERRUPT as they must run with irqs
+	 * disabled */
+	get_irq_desc(irq)->handler = &iic_pic;
+	get_irq_desc(irq)->status |= IRQ_PER_CPU;
+	request_irq(irq, iic_ipi_action, SA_INTERRUPT, name, NULL);
+}
+
+void iic_request_IPIs(void)
+{
+	iic_request_ipi(PPC_MSG_CALL_FUNCTION, "IPI-call");
+	iic_request_ipi(PPC_MSG_RESCHEDULE, "IPI-resched");
+#ifdef CONFIG_DEBUGGER
+	iic_request_ipi(PPC_MSG_DEBUGGER_BREAK, "IPI-debug");
+#endif /* CONFIG_DEBUGGER */
+}
+#endif /* CONFIG_SMP */
+
+static void iic_setup_spe_handlers(void)
+{
+	int be, isrc;
+
+	/* Assume two threads per BE are present */
+	for (be=0; be < num_present_cpus() / 2; be++) {
+		for (isrc = 0; isrc < IIC_CLASS_STRIDE * 3; isrc++) {
+			int irq = IIC_NODE_STRIDE * be + IIC_SPE_OFFSET + isrc;
+			get_irq_desc(irq)->handler = &iic_pic;
+		}
+	}
+}
+
+void iic_init_IRQ(void)
+{
+	int cpu, irq_offset;
+	struct iic *iic;
+
+	irq_offset = 0;
+	for_each_cpu(cpu) {
+		iic = &per_cpu(iic, cpu);
+		iic->regs = find_iic(cpu);
+		if (iic->regs)
+			out_be64(&iic->regs->prio, 0xff);
+	}
+	iic_setup_spe_handlers();
+}
diff --git a/arch/powerpc/platforms/cell/interrupt.h b/arch/powerpc/platforms/cell/interrupt.h
new file mode 100644
index 000000000000..37d58e6fd0c6
--- /dev/null
+++ b/arch/powerpc/platforms/cell/interrupt.h
@@ -0,0 +1,62 @@
+#ifndef ASM_CELL_PIC_H
+#define ASM_CELL_PIC_H
+#ifdef __KERNEL__
+/*
+ * Mapping of IIC pending bits into per-node
+ * interrupt numbers.
+ *
+ * IRQ     FF CC SS PP   FF CC SS PP	Description
+ *
+ * 00-3f   80 02 +0 00 - 80 02 +0 3f	South Bridge
+ * 00-3f   80 02 +b 00 - 80 02 +b 3f	South Bridge
+ * 41-4a   80 00 +1 ** - 80 00 +a **	SPU Class 0
+ * 51-5a   80 01 +1 ** - 80 01 +a **	SPU Class 1
+ * 61-6a   80 02 +1 ** - 80 02 +a **	SPU Class 2
+ * 70-7f   C0 ** ** 00 - C0 ** ** 0f	IPI
+ *
+ *    F flags
+ *    C class
+ *    S source
+ *    P Priority
+ *    + node number
+ *    * don't care
+ *
+ * A node consists of a Cell Broadband Engine and an optional
+ * south bridge device providing a maximum of 64 IRQs.
+ * The south bridge may be connected to either IOIF0
+ * or IOIF1.
+ * Each SPE is represented as three IRQ lines, one per
+ * interrupt class.
+ * 16 IRQ numbers are reserved for inter processor
+ * interruptions, although these are only used in the
+ * range of the first node.
+ *
+ * This scheme needs 128 IRQ numbers per BIF node ID,
+ * which means that with the total of 512 lines
+ * available, we can have a maximum of four nodes.
+ */
+
+enum {
+	IIC_EXT_OFFSET   = 0x00, /* Start of south bridge IRQs */
+	IIC_NUM_EXT      = 0x40, /* Number of south bridge IRQs */
+	IIC_SPE_OFFSET   = 0x40, /* Start of SPE interrupts */
+	IIC_CLASS_STRIDE = 0x10, /* SPE IRQs per class    */
+	IIC_IPI_OFFSET   = 0x70, /* Start of IPI IRQs */
+	IIC_NUM_IPIS     = 0x10, /* IRQs reserved for IPI */
+	IIC_NODE_STRIDE  = 0x80, /* Total IRQs per node   */
+};
+
+extern void iic_init_IRQ(void);
+extern int  iic_get_irq(struct pt_regs *regs);
+extern void iic_cause_IPI(int cpu, int mesg);
+extern void iic_request_IPIs(void);
+extern void iic_setup_cpu(void);
+extern void iic_local_enable(void);
+extern void iic_local_disable(void);
+
+
+extern void spider_init_IRQ(void);
+extern int spider_get_irq(unsigned long int_pending);
+
+#endif
+#endif /* ASM_CELL_PIC_H */
diff --git a/arch/powerpc/platforms/cell/iommu.c b/arch/powerpc/platforms/cell/iommu.c
new file mode 100644
index 000000000000..74f999b4ac9e
--- /dev/null
+++ b/arch/powerpc/platforms/cell/iommu.c
@@ -0,0 +1,381 @@
+/*
+ * IOMMU implementation for Cell Broadband Processor Architecture
+ * We just establish a linear mapping at boot by setting all the
+ * IOPT cache entries in the CPU.
+ * The mapping functions should be identical to pci_direct_iommu, 
+ * except for the handling of the high order bit that is required
+ * by the Spider bridge. These should be split into a separate
+ * file at the point where we get a different bridge chip.
+ *
+ * Copyright (C) 2005 IBM Deutschland Entwicklung GmbH,
+ *			 Arnd Bergmann <arndb@de.ibm.com>
+ *
+ * Based on linear mapping
+ * Copyright (C) 2003 Benjamin Herrenschmidt (benh@kernel.crashing.org)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#undef DEBUG
+
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/bootmem.h>
+#include <linux/mm.h>
+#include <linux/dma-mapping.h>
+
+#include <asm/sections.h>
+#include <asm/iommu.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/pci-bridge.h>
+#include <asm/machdep.h>
+#include <asm/pmac_feature.h>
+#include <asm/abs_addr.h>
+#include <asm/system.h>
+#include <asm/ppc-pci.h>
+
+#include "iommu.h"
+
+static inline unsigned long 
+get_iopt_entry(unsigned long real_address, unsigned long ioid,
+			 unsigned long prot)
+{
+	return (prot & IOPT_PROT_MASK)
+	     | (IOPT_COHERENT)
+	     | (IOPT_ORDER_VC)
+	     | (real_address & IOPT_RPN_MASK)
+	     | (ioid & IOPT_IOID_MASK);
+}
+
+typedef struct {
+	unsigned long val;
+} ioste;
+
+static inline ioste
+mk_ioste(unsigned long val)
+{
+	ioste ioste = { .val = val, };
+	return ioste;
+}
+
+static inline ioste
+get_iost_entry(unsigned long iopt_base, unsigned long io_address, unsigned page_size)
+{
+	unsigned long ps;
+	unsigned long iostep;
+	unsigned long nnpt;
+	unsigned long shift;
+
+	switch (page_size) {
+	case 0x1000000:
+		ps = IOST_PS_16M;
+		nnpt = 0;  /* one page per segment */
+		shift = 5; /* segment has 16 iopt entries */
+		break;
+
+	case 0x100000:
+		ps = IOST_PS_1M;
+		nnpt = 0;  /* one page per segment */
+		shift = 1; /* segment has 256 iopt entries */
+		break;
+
+	case 0x10000:
+		ps = IOST_PS_64K;
+		nnpt = 0x07; /* 8 pages per io page table */
+		shift = 0;   /* all entries are used */
+		break;
+
+	case 0x1000:
+		ps = IOST_PS_4K;
+		nnpt = 0x7f; /* 128 pages per io page table */
+		shift = 0;   /* all entries are used */
+		break;
+
+	default: /* not a known compile time constant */
+		{
+			/* BUILD_BUG_ON() is not usable here */
+			extern void __get_iost_entry_bad_page_size(void);
+			__get_iost_entry_bad_page_size();
+		}
+		break;
+	}
+
+	iostep = iopt_base +
+			 /* need 8 bytes per iopte */
+			(((io_address / page_size * 8)
+			 /* align io page tables on 4k page boundaries */
+				 << shift) 
+			 /* nnpt+1 pages go into each iopt */
+				 & ~(nnpt << 12));
+
+	nnpt++; /* this seems to work, but the documentation is not clear
+		   about wether we put nnpt or nnpt-1 into the ioste bits.
+		   In theory, this can't work for 4k pages. */
+	return mk_ioste(IOST_VALID_MASK
+			| (iostep & IOST_PT_BASE_MASK)
+			| ((nnpt << 5) & IOST_NNPT_MASK)
+			| (ps & IOST_PS_MASK));
+}
+
+/* compute the address of an io pte */
+static inline unsigned long
+get_ioptep(ioste iost_entry, unsigned long io_address)
+{
+	unsigned long iopt_base;
+	unsigned long page_size;
+	unsigned long page_number;
+	unsigned long iopt_offset;
+
+	iopt_base = iost_entry.val & IOST_PT_BASE_MASK;
+	page_size = iost_entry.val & IOST_PS_MASK;
+
+	/* decode page size to compute page number */
+	page_number = (io_address & 0x0fffffff) >> (10 + 2 * page_size);
+	/* page number is an offset into the io page table */
+	iopt_offset = (page_number << 3) & 0x7fff8ul;
+	return iopt_base + iopt_offset;
+}
+
+/* compute the tag field of the iopt cache entry */
+static inline unsigned long
+get_ioc_tag(ioste iost_entry, unsigned long io_address)
+{
+	unsigned long iopte = get_ioptep(iost_entry, io_address);
+
+	return IOPT_VALID_MASK
+	     | ((iopte & 0x00000000000000ff8ul) >> 3)
+	     | ((iopte & 0x0000003fffffc0000ul) >> 9);
+}
+
+/* compute the hashed 6 bit index for the 4-way associative pte cache */
+static inline unsigned long
+get_ioc_hash(ioste iost_entry, unsigned long io_address)
+{
+	unsigned long iopte = get_ioptep(iost_entry, io_address);
+
+	return ((iopte & 0x000000000000001f8ul) >> 3)
+	     ^ ((iopte & 0x00000000000020000ul) >> 17)
+	     ^ ((iopte & 0x00000000000010000ul) >> 15)
+	     ^ ((iopte & 0x00000000000008000ul) >> 13)
+	     ^ ((iopte & 0x00000000000004000ul) >> 11)
+	     ^ ((iopte & 0x00000000000002000ul) >> 9)
+	     ^ ((iopte & 0x00000000000001000ul) >> 7);
+}
+
+/* same as above, but pretend that we have a simpler 1-way associative
+   pte cache with an 8 bit index */
+static inline unsigned long
+get_ioc_hash_1way(ioste iost_entry, unsigned long io_address)
+{
+	unsigned long iopte = get_ioptep(iost_entry, io_address);
+
+	return ((iopte & 0x000000000000001f8ul) >> 3)
+	     ^ ((iopte & 0x00000000000020000ul) >> 17)
+	     ^ ((iopte & 0x00000000000010000ul) >> 15)
+	     ^ ((iopte & 0x00000000000008000ul) >> 13)
+	     ^ ((iopte & 0x00000000000004000ul) >> 11)
+	     ^ ((iopte & 0x00000000000002000ul) >> 9)
+	     ^ ((iopte & 0x00000000000001000ul) >> 7)
+	     ^ ((iopte & 0x0000000000000c000ul) >> 8);
+}
+
+static inline ioste
+get_iost_cache(void __iomem *base, unsigned long index)
+{
+	unsigned long __iomem *p = (base + IOC_ST_CACHE_DIR);
+	return mk_ioste(in_be64(&p[index]));
+}
+
+static inline void
+set_iost_cache(void __iomem *base, unsigned long index, ioste ste)
+{
+	unsigned long __iomem *p = (base + IOC_ST_CACHE_DIR);
+	pr_debug("ioste %02lx was %016lx, store %016lx", index,
+			get_iost_cache(base, index).val, ste.val);
+	out_be64(&p[index], ste.val);
+	pr_debug(" now %016lx\n", get_iost_cache(base, index).val);
+}
+
+static inline unsigned long
+get_iopt_cache(void __iomem *base, unsigned long index, unsigned long *tag)
+{
+	unsigned long __iomem *tags = (void *)(base + IOC_PT_CACHE_DIR);
+	unsigned long __iomem *p = (void *)(base + IOC_PT_CACHE_REG);	
+
+	*tag = tags[index];
+	rmb();
+	return *p;
+}
+
+static inline void
+set_iopt_cache(void __iomem *base, unsigned long index,
+		 unsigned long tag, unsigned long val)
+{
+	unsigned long __iomem *tags = base + IOC_PT_CACHE_DIR;
+	unsigned long __iomem *p = base + IOC_PT_CACHE_REG;
+	pr_debug("iopt %02lx was v%016lx/t%016lx, store v%016lx/t%016lx\n",
+		index, get_iopt_cache(base, index, &oldtag), oldtag, val, tag);
+
+	out_be64(p, val);
+	out_be64(&tags[index], tag);
+}
+
+static inline void
+set_iost_origin(void __iomem *base)
+{
+	unsigned long __iomem *p = base + IOC_ST_ORIGIN;
+	unsigned long origin = IOSTO_ENABLE | IOSTO_SW;
+
+	pr_debug("iost_origin %016lx, now %016lx\n", in_be64(p), origin);
+	out_be64(p, origin);
+}
+
+static inline void
+set_iocmd_config(void __iomem *base)
+{
+	unsigned long __iomem *p = base + 0xc00;
+	unsigned long conf;
+
+	conf = in_be64(p);
+	pr_debug("iost_conf %016lx, now %016lx\n", conf, conf | IOCMD_CONF_TE);
+	out_be64(p, conf | IOCMD_CONF_TE);
+}
+
+/* FIXME: get these from the device tree */
+#define ioc_base	0x20000511000ull
+#define ioc_mmio_base	0x20000510000ull
+#define ioid		0x48a
+#define iopt_phys_offset (- 0x20000000) /* We have a 512MB offset from the SB */
+#define io_page_size	0x1000000
+
+static unsigned long map_iopt_entry(unsigned long address)
+{
+	switch (address >> 20) {
+	case 0x600:
+		address = 0x24020000000ull; /* spider i/o */
+		break;
+	default:
+		address += iopt_phys_offset;
+		break;
+	}
+
+	return get_iopt_entry(address, ioid, IOPT_PROT_RW);
+}
+
+static void iommu_bus_setup_null(struct pci_bus *b) { }
+static void iommu_dev_setup_null(struct pci_dev *d) { }
+
+/* initialize the iommu to support a simple linear mapping
+ * for each DMA window used by any device. For now, we
+ * happen to know that there is only one DMA window in use,
+ * starting at iopt_phys_offset. */
+static void cell_map_iommu(void)
+{
+	unsigned long address;
+	void __iomem *base;
+	ioste ioste;
+	unsigned long index;
+
+	base = __ioremap(ioc_base, 0x1000, _PAGE_NO_CACHE);
+	pr_debug("%lx mapped to %p\n", ioc_base, base);
+	set_iocmd_config(base);
+	iounmap(base);
+
+	base = __ioremap(ioc_mmio_base, 0x1000, _PAGE_NO_CACHE);
+	pr_debug("%lx mapped to %p\n", ioc_mmio_base, base);
+
+	set_iost_origin(base);
+
+	for (address = 0; address < 0x100000000ul; address += io_page_size) {
+		ioste = get_iost_entry(0x10000000000ul, address, io_page_size);
+		if ((address & 0xfffffff) == 0) /* segment start */
+			set_iost_cache(base, address >> 28, ioste);
+		index = get_ioc_hash_1way(ioste, address);
+		pr_debug("addr %08lx, index %02lx, ioste %016lx\n",
+					 address, index, ioste.val);
+		set_iopt_cache(base,
+			get_ioc_hash_1way(ioste, address),
+			get_ioc_tag(ioste, address),
+			map_iopt_entry(address));
+	}
+	iounmap(base);
+}
+
+
+static void *cell_alloc_coherent(struct device *hwdev, size_t size,
+			   dma_addr_t *dma_handle, gfp_t flag)
+{
+	void *ret;
+
+	ret = (void *)__get_free_pages(flag, get_order(size));
+	if (ret != NULL) {
+		memset(ret, 0, size);
+		*dma_handle = virt_to_abs(ret) | CELL_DMA_VALID;
+	}
+	return ret;
+}
+
+static void cell_free_coherent(struct device *hwdev, size_t size,
+				 void *vaddr, dma_addr_t dma_handle)
+{
+	free_pages((unsigned long)vaddr, get_order(size));
+}
+
+static dma_addr_t cell_map_single(struct device *hwdev, void *ptr,
+		size_t size, enum dma_data_direction direction)
+{
+	return virt_to_abs(ptr) | CELL_DMA_VALID;
+}
+
+static void cell_unmap_single(struct device *hwdev, dma_addr_t dma_addr,
+		size_t size, enum dma_data_direction direction)
+{
+}
+
+static int cell_map_sg(struct device *hwdev, struct scatterlist *sg,
+		int nents, enum dma_data_direction direction)
+{
+	int i;
+
+	for (i = 0; i < nents; i++, sg++) {
+		sg->dma_address = (page_to_phys(sg->page) + sg->offset)
+					| CELL_DMA_VALID;
+		sg->dma_length = sg->length;
+	}
+
+	return nents;
+}
+
+static void cell_unmap_sg(struct device *hwdev, struct scatterlist *sg,
+		int nents, enum dma_data_direction direction)
+{
+}
+
+static int cell_dma_supported(struct device *dev, u64 mask)
+{
+	return mask < 0x100000000ull;
+}
+
+void cell_init_iommu(void)
+{
+	cell_map_iommu();
+
+	/* Direct I/O, IOMMU off */
+	ppc_md.iommu_dev_setup = iommu_dev_setup_null;
+	ppc_md.iommu_bus_setup = iommu_bus_setup_null;
+
+	pci_dma_ops.alloc_coherent = cell_alloc_coherent;
+	pci_dma_ops.free_coherent = cell_free_coherent;
+	pci_dma_ops.map_single = cell_map_single;
+	pci_dma_ops.unmap_single = cell_unmap_single;
+	pci_dma_ops.map_sg = cell_map_sg;
+	pci_dma_ops.unmap_sg = cell_unmap_sg;
+	pci_dma_ops.dma_supported = cell_dma_supported;
+}
diff --git a/arch/powerpc/platforms/cell/iommu.h b/arch/powerpc/platforms/cell/iommu.h
new file mode 100644
index 000000000000..490d77abfe85
--- /dev/null
+++ b/arch/powerpc/platforms/cell/iommu.h
@@ -0,0 +1,65 @@
+#ifndef CELL_IOMMU_H
+#define CELL_IOMMU_H
+
+/* some constants */
+enum {
+	/* segment table entries */
+	IOST_VALID_MASK	  = 0x8000000000000000ul,
+	IOST_TAG_MASK     = 0x3000000000000000ul,
+	IOST_PT_BASE_MASK = 0x000003fffffff000ul,
+	IOST_NNPT_MASK	  = 0x0000000000000fe0ul,
+	IOST_PS_MASK	  = 0x000000000000000ful,
+
+	IOST_PS_4K	  = 0x1,
+	IOST_PS_64K	  = 0x3,
+	IOST_PS_1M	  = 0x5,
+	IOST_PS_16M	  = 0x7,
+
+	/* iopt tag register */
+	IOPT_VALID_MASK   = 0x0000000200000000ul,
+	IOPT_TAG_MASK	  = 0x00000001fffffffful,
+
+	/* iopt cache register */
+	IOPT_PROT_MASK	  = 0xc000000000000000ul,
+	IOPT_PROT_NONE	  = 0x0000000000000000ul,
+	IOPT_PROT_READ	  = 0x4000000000000000ul,
+	IOPT_PROT_WRITE	  = 0x8000000000000000ul,
+	IOPT_PROT_RW	  = 0xc000000000000000ul,
+	IOPT_COHERENT	  = 0x2000000000000000ul,
+	
+	IOPT_ORDER_MASK	  = 0x1800000000000000ul,
+	/* order access to same IOID/VC on same address */
+	IOPT_ORDER_ADDR	  = 0x0800000000000000ul,
+	/* similar, but only after a write access */
+	IOPT_ORDER_WRITES = 0x1000000000000000ul,
+	/* Order all accesses to same IOID/VC */
+	IOPT_ORDER_VC	  = 0x1800000000000000ul,
+	
+	IOPT_RPN_MASK	  = 0x000003fffffff000ul,
+	IOPT_HINT_MASK	  = 0x0000000000000800ul,
+	IOPT_IOID_MASK	  = 0x00000000000007fful,
+
+	IOSTO_ENABLE	  = 0x8000000000000000ul,
+	IOSTO_ORIGIN	  = 0x000003fffffff000ul,
+	IOSTO_HW	  = 0x0000000000000800ul,
+	IOSTO_SW	  = 0x0000000000000400ul,
+
+	IOCMD_CONF_TE	  = 0x0000800000000000ul,
+
+	/* memory mapped registers */
+	IOC_PT_CACHE_DIR  = 0x000,
+	IOC_ST_CACHE_DIR  = 0x800,
+	IOC_PT_CACHE_REG  = 0x910,
+	IOC_ST_ORIGIN     = 0x918,
+	IOC_CONF	  = 0x930,
+
+	/* The high bit needs to be set on every DMA address,
+	   only 2GB are addressable */
+	CELL_DMA_VALID	  = 0x80000000,
+	CELL_DMA_MASK	  = 0x7fffffff,
+};
+
+
+void cell_init_iommu(void);
+
+#endif
diff --git a/arch/powerpc/platforms/cell/setup.c b/arch/powerpc/platforms/cell/setup.c
new file mode 100644
index 000000000000..9a495634d0c2
--- /dev/null
+++ b/arch/powerpc/platforms/cell/setup.c
@@ -0,0 +1,141 @@
+/*
+ *  linux/arch/powerpc/platforms/cell/cell_setup.c
+ *
+ *  Copyright (C) 1995  Linus Torvalds
+ *  Adapted from 'alpha' version by Gary Thomas
+ *  Modified by Cort Dougan (cort@cs.nmt.edu)
+ *  Modified by PPC64 Team, IBM Corp
+ *  Modified by Cell Team, IBM Deutschland Entwicklung GmbH
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/slab.h>
+#include <linux/user.h>
+#include <linux/reboot.h>
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/irq.h>
+#include <linux/seq_file.h>
+#include <linux/root_dev.h>
+#include <linux/console.h>
+
+#include <asm/mmu.h>
+#include <asm/processor.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/prom.h>
+#include <asm/rtas.h>
+#include <asm/pci-bridge.h>
+#include <asm/iommu.h>
+#include <asm/dma.h>
+#include <asm/machdep.h>
+#include <asm/time.h>
+#include <asm/nvram.h>
+#include <asm/cputable.h>
+#include <asm/ppc-pci.h>
+#include <asm/irq.h>
+
+#include "interrupt.h"
+#include "iommu.h"
+
+#ifdef DEBUG
+#define DBG(fmt...) udbg_printf(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+void cell_show_cpuinfo(struct seq_file *m)
+{
+	struct device_node *root;
+	const char *model = "";
+
+	root = of_find_node_by_path("/");
+	if (root)
+		model = get_property(root, "model", NULL);
+	seq_printf(m, "machine\t\t: CHRP %s\n", model);
+	of_node_put(root);
+}
+
+static void cell_progress(char *s, unsigned short hex)
+{
+	printk("*** %04x : %s\n", hex, s ? s : "");
+}
+
+static void __init cell_setup_arch(void)
+{
+	ppc_md.init_IRQ       = iic_init_IRQ;
+	ppc_md.get_irq        = iic_get_irq;
+
+#ifdef CONFIG_SMP
+	smp_init_cell();
+#endif
+
+	/* init to some ~sane value until calibrate_delay() runs */
+	loops_per_jiffy = 50000000;
+
+	if (ROOT_DEV == 0) {
+		printk("No ramdisk, default root is /dev/hda2\n");
+		ROOT_DEV = Root_HDA2;
+	}
+
+	/* Find and initialize PCI host bridges */
+	init_pci_config_tokens();
+	find_and_init_phbs();
+	spider_init_IRQ();
+#ifdef CONFIG_DUMMY_CONSOLE
+	conswitchp = &dummy_con;
+#endif
+
+	mmio_nvram_init();
+}
+
+/*
+ * Early initialization.  Relocation is on but do not reference unbolted pages
+ */
+static void __init cell_init_early(void)
+{
+	DBG(" -> cell_init_early()\n");
+
+	hpte_init_native();
+
+	cell_init_iommu();
+
+	ppc64_interrupt_controller = IC_CELL_PIC;
+
+	DBG(" <- cell_init_early()\n");
+}
+
+
+static int __init cell_probe(int platform)
+{
+	if (platform != PLATFORM_CELL)
+		return 0;
+
+	return 1;
+}
+
+struct machdep_calls __initdata cell_md = {
+	.probe			= cell_probe,
+	.setup_arch		= cell_setup_arch,
+	.init_early		= cell_init_early,
+	.show_cpuinfo		= cell_show_cpuinfo,
+	.restart		= rtas_restart,
+	.power_off		= rtas_power_off,
+	.halt			= rtas_halt,
+	.get_boot_time		= rtas_get_boot_time,
+	.get_rtc_time		= rtas_get_rtc_time,
+	.set_rtc_time		= rtas_set_rtc_time,
+	.calibrate_decr		= generic_calibrate_decr,
+	.progress		= cell_progress,
+};
diff --git a/arch/powerpc/platforms/cell/smp.c b/arch/powerpc/platforms/cell/smp.c
new file mode 100644
index 000000000000..de96eadf419d
--- /dev/null
+++ b/arch/powerpc/platforms/cell/smp.c
@@ -0,0 +1,230 @@
+/*
+ * SMP support for BPA machines.
+ *
+ * Dave Engebretsen, Peter Bergner, and
+ * Mike Corrigan {engebret|bergner|mikec}@us.ibm.com
+ *
+ * Plus various changes from other IBM teams...
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/cache.h>
+#include <linux/err.h>
+#include <linux/sysdev.h>
+#include <linux/cpu.h>
+
+#include <asm/ptrace.h>
+#include <asm/atomic.h>
+#include <asm/irq.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/smp.h>
+#include <asm/paca.h>
+#include <asm/time.h>
+#include <asm/machdep.h>
+#include <asm/cputable.h>
+#include <asm/firmware.h>
+#include <asm/system.h>
+#include <asm/rtas.h>
+
+#include "interrupt.h"
+
+#ifdef DEBUG
+#define DBG(fmt...) udbg_printf(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+/*
+ * The primary thread of each non-boot processor is recorded here before
+ * smp init.
+ */
+static cpumask_t of_spin_map;
+
+extern void pSeries_secondary_smp_init(unsigned long);
+
+/**
+ * smp_startup_cpu() - start the given cpu
+ *
+ * At boot time, there is nothing to do for primary threads which were
+ * started from Open Firmware.  For anything else, call RTAS with the
+ * appropriate start location.
+ *
+ * Returns:
+ *	0	- failure
+ *	1	- success
+ */
+static inline int __devinit smp_startup_cpu(unsigned int lcpu)
+{
+	int status;
+	unsigned long start_here = __pa((u32)*((unsigned long *)
+					       pSeries_secondary_smp_init));
+	unsigned int pcpu;
+	int start_cpu;
+
+	if (cpu_isset(lcpu, of_spin_map))
+		/* Already started by OF and sitting in spin loop */
+		return 1;
+
+	pcpu = get_hard_smp_processor_id(lcpu);
+
+	/* Fixup atomic count: it exited inside IRQ handler. */
+	paca[lcpu].__current->thread_info->preempt_count	= 0;
+
+	/*
+	 * If the RTAS start-cpu token does not exist then presume the
+	 * cpu is already spinning.
+	 */
+	start_cpu = rtas_token("start-cpu");
+	if (start_cpu == RTAS_UNKNOWN_SERVICE)
+		return 1;
+
+	status = rtas_call(start_cpu, 3, 1, NULL, pcpu, start_here, lcpu);
+	if (status != 0) {
+		printk(KERN_ERR "start-cpu failed: %i\n", status);
+		return 0;
+	}
+
+	return 1;
+}
+
+static void smp_iic_message_pass(int target, int msg)
+{
+	unsigned int i;
+
+	if (target < NR_CPUS) {
+		iic_cause_IPI(target, msg);
+	} else {
+		for_each_online_cpu(i) {
+			if (target == MSG_ALL_BUT_SELF
+			    && i == smp_processor_id())
+				continue;
+			iic_cause_IPI(i, msg);
+		}
+	}
+}
+
+static int __init smp_iic_probe(void)
+{
+	iic_request_IPIs();
+
+	return cpus_weight(cpu_possible_map);
+}
+
+static void __devinit smp_iic_setup_cpu(int cpu)
+{
+	if (cpu != boot_cpuid)
+		iic_setup_cpu();
+}
+
+static DEFINE_SPINLOCK(timebase_lock);
+static unsigned long timebase = 0;
+
+static void __devinit cell_give_timebase(void)
+{
+	spin_lock(&timebase_lock);
+	rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL);
+	timebase = get_tb();
+	spin_unlock(&timebase_lock);
+
+	while (timebase)
+		barrier();
+	rtas_call(rtas_token("thaw-time-base"), 0, 1, NULL);
+}
+
+static void __devinit cell_take_timebase(void)
+{
+	while (!timebase)
+		barrier();
+	spin_lock(&timebase_lock);
+	set_tb(timebase >> 32, timebase & 0xffffffff);
+	timebase = 0;
+	spin_unlock(&timebase_lock);
+}
+
+static void __devinit smp_cell_kick_cpu(int nr)
+{
+	BUG_ON(nr < 0 || nr >= NR_CPUS);
+
+	if (!smp_startup_cpu(nr))
+		return;
+
+	/*
+	 * The processor is currently spinning, waiting for the
+	 * cpu_start field to become non-zero After we set cpu_start,
+	 * the processor will continue on to secondary_start
+	 */
+	paca[nr].cpu_start = 1;
+}
+
+static int smp_cell_cpu_bootable(unsigned int nr)
+{
+	/* Special case - we inhibit secondary thread startup
+	 * during boot if the user requests it.  Odd-numbered
+	 * cpus are assumed to be secondary threads.
+	 */
+	if (system_state < SYSTEM_RUNNING &&
+	    cpu_has_feature(CPU_FTR_SMT) &&
+	    !smt_enabled_at_boot && nr % 2 != 0)
+		return 0;
+
+	return 1;
+}
+static struct smp_ops_t bpa_iic_smp_ops = {
+	.message_pass	= smp_iic_message_pass,
+	.probe		= smp_iic_probe,
+	.kick_cpu	= smp_cell_kick_cpu,
+	.setup_cpu	= smp_iic_setup_cpu,
+	.cpu_bootable	= smp_cell_cpu_bootable,
+};
+
+/* This is called very early */
+void __init smp_init_cell(void)
+{
+	int i;
+
+	DBG(" -> smp_init_cell()\n");
+
+	smp_ops = &bpa_iic_smp_ops;
+
+	/* Mark threads which are still spinning in hold loops. */
+	if (cpu_has_feature(CPU_FTR_SMT)) {
+		for_each_present_cpu(i) {
+			if (i % 2 == 0)
+				/*
+				 * Even-numbered logical cpus correspond to
+				 * primary threads.
+				 */
+				cpu_set(i, of_spin_map);
+		}
+	} else {
+		of_spin_map = cpu_present_map;
+	}
+
+	cpu_clear(boot_cpuid, of_spin_map);
+
+	/* Non-lpar has additional take/give timebase */
+	if (rtas_token("freeze-time-base") != RTAS_UNKNOWN_SERVICE) {
+		smp_ops->give_timebase = cell_give_timebase;
+		smp_ops->take_timebase = cell_take_timebase;
+	}
+
+	DBG(" <- smp_init_cell()\n");
+}
diff --git a/arch/powerpc/platforms/cell/spider-pic.c b/arch/powerpc/platforms/cell/spider-pic.c
new file mode 100644
index 000000000000..e74132188bdf
--- /dev/null
+++ b/arch/powerpc/platforms/cell/spider-pic.c
@@ -0,0 +1,191 @@
+/*
+ * External Interrupt Controller on Spider South Bridge
+ *
+ * (C) Copyright IBM Deutschland Entwicklung GmbH 2005
+ *
+ * Author: Arnd Bergmann <arndb@de.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+
+#include <asm/pgtable.h>
+#include <asm/prom.h>
+#include <asm/io.h>
+
+#include "interrupt.h"
+
+/* register layout taken from Spider spec, table 7.4-4 */
+enum {
+	TIR_DEN		= 0x004, /* Detection Enable Register */
+	TIR_MSK		= 0x084, /* Mask Level Register */
+	TIR_EDC		= 0x0c0, /* Edge Detection Clear Register */
+	TIR_PNDA	= 0x100, /* Pending Register A */
+	TIR_PNDB	= 0x104, /* Pending Register B */
+	TIR_CS		= 0x144, /* Current Status Register */
+	TIR_LCSA	= 0x150, /* Level Current Status Register A */
+	TIR_LCSB	= 0x154, /* Level Current Status Register B */
+	TIR_LCSC	= 0x158, /* Level Current Status Register C */
+	TIR_LCSD	= 0x15c, /* Level Current Status Register D */
+	TIR_CFGA	= 0x200, /* Setting Register A0 */
+	TIR_CFGB	= 0x204, /* Setting Register B0 */
+			/* 0x208 ... 0x3ff Setting Register An/Bn */
+	TIR_PPNDA	= 0x400, /* Packet Pending Register A */
+	TIR_PPNDB	= 0x404, /* Packet Pending Register B */
+	TIR_PIERA	= 0x408, /* Packet Output Error Register A */
+	TIR_PIERB	= 0x40c, /* Packet Output Error Register B */
+	TIR_PIEN	= 0x444, /* Packet Output Enable Register */
+	TIR_PIPND	= 0x454, /* Packet Output Pending Register */
+	TIRDID		= 0x484, /* Spider Device ID Register */
+	REISTIM		= 0x500, /* Reissue Command Timeout Time Setting */
+	REISTIMEN	= 0x504, /* Reissue Command Timeout Setting */
+	REISWAITEN	= 0x508, /* Reissue Wait Control*/
+};
+
+static void __iomem *spider_pics[4];
+
+static void __iomem *spider_get_pic(int irq)
+{
+	int node = irq / IIC_NODE_STRIDE;
+	irq %= IIC_NODE_STRIDE;
+
+	if (irq >= IIC_EXT_OFFSET &&
+	    irq < IIC_EXT_OFFSET + IIC_NUM_EXT &&
+	    spider_pics)
+		return spider_pics[node];
+	return NULL;
+}
+
+static int spider_get_nr(unsigned int irq)
+{
+	return (irq % IIC_NODE_STRIDE) - IIC_EXT_OFFSET;
+}
+
+static void __iomem *spider_get_irq_config(int irq)
+{
+	void __iomem *pic;
+	pic = spider_get_pic(irq);
+	return pic + TIR_CFGA + 8 * spider_get_nr(irq);
+}
+
+static void spider_enable_irq(unsigned int irq)
+{
+	void __iomem *cfg = spider_get_irq_config(irq);
+	irq = spider_get_nr(irq);
+
+	out_be32(cfg, in_be32(cfg) | 0x3107000eu);
+	out_be32(cfg + 4, in_be32(cfg + 4) | 0x00020000u | irq);
+}
+
+static void spider_disable_irq(unsigned int irq)
+{
+	void __iomem *cfg = spider_get_irq_config(irq);
+	irq = spider_get_nr(irq);
+
+	out_be32(cfg, in_be32(cfg) & ~0x30000000u);
+}
+
+static unsigned int spider_startup_irq(unsigned int irq)
+{
+	spider_enable_irq(irq);
+	return 0;
+}
+
+static void spider_shutdown_irq(unsigned int irq)
+{
+	spider_disable_irq(irq);
+}
+
+static void spider_end_irq(unsigned int irq)
+{
+	spider_enable_irq(irq);
+}
+
+static void spider_ack_irq(unsigned int irq)
+{
+	spider_disable_irq(irq);
+	iic_local_enable();
+}
+
+static struct hw_interrupt_type spider_pic = {
+	.typename = " SPIDER   ",
+	.startup = spider_startup_irq,
+	.shutdown = spider_shutdown_irq,
+	.enable = spider_enable_irq,
+	.disable = spider_disable_irq,
+	.ack = spider_ack_irq,
+	.end = spider_end_irq,
+};
+
+
+int spider_get_irq(unsigned long int_pending)
+{
+	void __iomem *regs = spider_get_pic(int_pending);
+	unsigned long cs;
+	int irq;
+
+	cs = in_be32(regs + TIR_CS);
+
+	irq = cs >> 24;
+	if (irq != 63)
+		return irq;
+
+	return -1;
+}
+ 
+void spider_init_IRQ(void)
+{
+	int node;
+	struct device_node *dn;
+	unsigned int *property;
+	long spiderpic;
+	int n;
+
+/* FIXME: detect multiple PICs as soon as the device tree has them */
+	for (node = 0; node < 1; node++) {
+		dn = of_find_node_by_path("/");
+		n = prom_n_addr_cells(dn);
+		property = (unsigned int *) get_property(dn,
+				"platform-spider-pic", NULL);
+
+		if (!property)
+			continue;
+		for (spiderpic = 0; n > 0; --n)
+			spiderpic = (spiderpic << 32) + *property++;
+		printk(KERN_DEBUG "SPIDER addr: %lx\n", spiderpic);
+		spider_pics[node] = __ioremap(spiderpic, 0x800, _PAGE_NO_CACHE);
+		for (n = 0; n < IIC_NUM_EXT; n++) {
+			int irq = n + IIC_EXT_OFFSET + node * IIC_NODE_STRIDE;
+			get_irq_desc(irq)->handler = &spider_pic;
+
+ 		/* do not mask any interrupts because of level */
+ 		out_be32(spider_pics[node] + TIR_MSK, 0x0);
+ 		
+ 		/* disable edge detection clear */
+ 		/* out_be32(spider_pics[node] + TIR_EDC, 0x0); */
+ 		
+ 		/* enable interrupt packets to be output */
+ 		out_be32(spider_pics[node] + TIR_PIEN,
+			in_be32(spider_pics[node] + TIR_PIEN) | 0x1);
+ 		
+ 		/* Enable the interrupt detection enable bit. Do this last! */
+ 		out_be32(spider_pics[node] + TIR_DEN,
+			in_be32(spider_pics[node] +TIR_DEN) | 0x1);
+
+		}
+	}
+}
diff --git a/arch/powerpc/platforms/chrp/Makefile b/arch/powerpc/platforms/chrp/Makefile
new file mode 100644
index 000000000000..902feb1ac431
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/Makefile
@@ -0,0 +1,4 @@
+obj-y				+= setup.o time.o pegasos_eth.o
+obj-$(CONFIG_PCI)		+= pci.o
+obj-$(CONFIG_SMP)		+= smp.o
+obj-$(CONFIG_NVRAM)		+= nvram.o
diff --git a/arch/powerpc/platforms/chrp/chrp.h b/arch/powerpc/platforms/chrp/chrp.h
new file mode 100644
index 000000000000..3a2057fa314a
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/chrp.h
@@ -0,0 +1,12 @@
+/*
+ * Declarations of CHRP platform-specific things.
+ */
+
+extern void chrp_nvram_init(void);
+extern void chrp_get_rtc_time(struct rtc_time *);
+extern int chrp_set_rtc_time(struct rtc_time *);
+extern void chrp_calibrate_decr(void);
+extern long chrp_time_init(void);
+
+extern void chrp_find_bridges(void);
+extern void chrp_event_scan(void);
diff --git a/arch/powerpc/platforms/chrp/nvram.c b/arch/powerpc/platforms/chrp/nvram.c
new file mode 100644
index 000000000000..150f67d6f90c
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/nvram.c
@@ -0,0 +1,89 @@
+/*
+ *  c 2001 PPC 64 Team, IBM Corp
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ *
+ * /dev/nvram driver for PPC
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <asm/uaccess.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+#include <asm/rtas.h>
+#include "chrp.h"
+
+static unsigned int nvram_size;
+static unsigned char nvram_buf[4];
+static DEFINE_SPINLOCK(nvram_lock);
+
+static unsigned char chrp_nvram_read(int addr)
+{
+	unsigned int done;
+	unsigned long flags;
+	unsigned char ret;
+
+	if (addr >= nvram_size) {
+		printk(KERN_DEBUG "%s: read addr %d > nvram_size %u\n",
+		       current->comm, addr, nvram_size);
+		return 0xff;
+	}
+	spin_lock_irqsave(&nvram_lock, flags);
+	if ((rtas_call(rtas_token("nvram-fetch"), 3, 2, &done, addr,
+		       __pa(nvram_buf), 1) != 0) || 1 != done)
+		ret = 0xff;
+	else
+		ret = nvram_buf[0];
+	spin_unlock_irqrestore(&nvram_lock, flags);
+
+	return ret;
+}
+
+static void chrp_nvram_write(int addr, unsigned char val)
+{
+	unsigned int done;
+	unsigned long flags;
+
+	if (addr >= nvram_size) {
+		printk(KERN_DEBUG "%s: write addr %d > nvram_size %u\n",
+		       current->comm, addr, nvram_size);
+		return;
+	}
+	spin_lock_irqsave(&nvram_lock, flags);
+	nvram_buf[0] = val;
+	if ((rtas_call(rtas_token("nvram-store"), 3, 2, &done, addr,
+		       __pa(nvram_buf), 1) != 0) || 1 != done)
+		printk(KERN_DEBUG "rtas IO error storing 0x%02x at %d", val, addr);
+	spin_unlock_irqrestore(&nvram_lock, flags);
+}
+
+void __init chrp_nvram_init(void)
+{
+	struct device_node *nvram;
+	unsigned int *nbytes_p, proplen;
+
+	nvram = of_find_node_by_type(NULL, "nvram");
+	if (nvram == NULL)
+		return;
+
+	nbytes_p = (unsigned int *)get_property(nvram, "#bytes", &proplen);
+	if (nbytes_p == NULL || proplen != sizeof(unsigned int))
+		return;
+
+	nvram_size = *nbytes_p;
+
+	printk(KERN_INFO "CHRP nvram contains %u bytes\n", nvram_size);
+	of_node_put(nvram);
+
+	ppc_md.nvram_read_val = chrp_nvram_read;
+	ppc_md.nvram_write_val = chrp_nvram_write;
+
+	return;
+}
diff --git a/arch/powerpc/platforms/chrp/pci.c b/arch/powerpc/platforms/chrp/pci.c
new file mode 100644
index 000000000000..82c429d487f3
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/pci.c
@@ -0,0 +1,310 @@
+/*
+ * CHRP pci routines.
+ */
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/ide.h>
+
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/irq.h>
+#include <asm/hydra.h>
+#include <asm/prom.h>
+#include <asm/gg2.h>
+#include <asm/machdep.h>
+#include <asm/sections.h>
+#include <asm/pci-bridge.h>
+#include <asm/open_pic.h>
+#include <asm/grackle.h>
+#include <asm/rtas.h>
+
+/* LongTrail */
+void __iomem *gg2_pci_config_base;
+
+/*
+ * The VLSI Golden Gate II has only 512K of PCI configuration space, so we
+ * limit the bus number to 3 bits
+ */
+
+int gg2_read_config(struct pci_bus *bus, unsigned int devfn, int off,
+			   int len, u32 *val)
+{
+	volatile void __iomem *cfg_data;
+	struct pci_controller *hose = bus->sysdata;
+
+	if (bus->number > 7)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	/*
+	 * Note: the caller has already checked that off is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	cfg_data = hose->cfg_data + ((bus->number<<16) | (devfn<<8) | off);
+	switch (len) {
+	case 1:
+		*val =  in_8(cfg_data);
+		break;
+	case 2:
+		*val = in_le16(cfg_data);
+		break;
+	default:
+		*val = in_le32(cfg_data);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int gg2_write_config(struct pci_bus *bus, unsigned int devfn, int off,
+			    int len, u32 val)
+{
+	volatile void __iomem *cfg_data;
+	struct pci_controller *hose = bus->sysdata;
+
+	if (bus->number > 7)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	/*
+	 * Note: the caller has already checked that off is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	cfg_data = hose->cfg_data + ((bus->number<<16) | (devfn<<8) | off);
+	switch (len) {
+	case 1:
+		out_8(cfg_data, val);
+		break;
+	case 2:
+		out_le16(cfg_data, val);
+		break;
+	default:
+		out_le32(cfg_data, val);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static struct pci_ops gg2_pci_ops =
+{
+	gg2_read_config,
+	gg2_write_config
+};
+
+/*
+ * Access functions for PCI config space using RTAS calls.
+ */
+int rtas_read_config(struct pci_bus *bus, unsigned int devfn, int offset,
+		     int len, u32 *val)
+{
+	struct pci_controller *hose = bus->sysdata;
+	unsigned long addr = (offset & 0xff) | ((devfn & 0xff) << 8)
+		| (((bus->number - hose->first_busno) & 0xff) << 16)
+		| (hose->index << 24);
+        int ret = -1;
+	int rval;
+
+	rval = rtas_call(rtas_token("read-pci-config"), 2, 2, &ret, addr, len);
+	*val = ret;
+	return rval? PCIBIOS_DEVICE_NOT_FOUND: PCIBIOS_SUCCESSFUL;
+}
+
+int rtas_write_config(struct pci_bus *bus, unsigned int devfn, int offset,
+		      int len, u32 val)
+{
+	struct pci_controller *hose = bus->sysdata;
+	unsigned long addr = (offset & 0xff) | ((devfn & 0xff) << 8)
+		| (((bus->number - hose->first_busno) & 0xff) << 16)
+		| (hose->index << 24);
+	int rval;
+
+	rval = rtas_call(rtas_token("write-pci-config"), 3, 1, NULL,
+			 addr, len, val);
+	return rval? PCIBIOS_DEVICE_NOT_FOUND: PCIBIOS_SUCCESSFUL;
+}
+
+static struct pci_ops rtas_pci_ops =
+{
+	rtas_read_config,
+	rtas_write_config
+};
+
+volatile struct Hydra __iomem *Hydra = NULL;
+
+int __init
+hydra_init(void)
+{
+	struct device_node *np;
+
+	np = find_devices("mac-io");
+	if (np == NULL || np->n_addrs == 0)
+		return 0;
+	Hydra = ioremap(np->addrs[0].address, np->addrs[0].size);
+	printk("Hydra Mac I/O at %lx\n", np->addrs[0].address);
+	printk("Hydra Feature_Control was %x",
+	       in_le32(&Hydra->Feature_Control));
+	out_le32(&Hydra->Feature_Control, (HYDRA_FC_SCC_CELL_EN |
+					   HYDRA_FC_SCSI_CELL_EN |
+					   HYDRA_FC_SCCA_ENABLE |
+					   HYDRA_FC_SCCB_ENABLE |
+					   HYDRA_FC_ARB_BYPASS |
+					   HYDRA_FC_MPIC_ENABLE |
+					   HYDRA_FC_SLOW_SCC_PCLK |
+					   HYDRA_FC_MPIC_IS_MASTER));
+	printk(", now %x\n", in_le32(&Hydra->Feature_Control));
+	return 1;
+}
+
+void __init
+chrp_pcibios_fixup(void)
+{
+	struct pci_dev *dev = NULL;
+	struct device_node *np;
+
+	/* PCI interrupts are controlled by the OpenPIC */
+	for_each_pci_dev(dev) {
+		np = pci_device_to_OF_node(dev);
+		if ((np != 0) && (np->n_intrs > 0) && (np->intrs[0].line != 0))
+			dev->irq = np->intrs[0].line;
+		pci_write_config_byte(dev, PCI_INTERRUPT_LINE, dev->irq);
+	}
+}
+
+#define PRG_CL_RESET_VALID 0x00010000
+
+static void __init
+setup_python(struct pci_controller *hose, struct device_node *dev)
+{
+	u32 __iomem *reg;
+	u32 val;
+	unsigned long addr = dev->addrs[0].address;
+
+	setup_indirect_pci(hose, addr + 0xf8000, addr + 0xf8010);
+
+	/* Clear the magic go-slow bit */
+	reg = ioremap(dev->addrs[0].address + 0xf6000, 0x40);
+	val = in_be32(&reg[12]);
+	if (val & PRG_CL_RESET_VALID) {
+		out_be32(&reg[12], val & ~PRG_CL_RESET_VALID);
+		in_be32(&reg[12]);
+	}
+	iounmap(reg);
+}
+
+/* Marvell Discovery II based Pegasos 2 */
+static void __init setup_peg2(struct pci_controller *hose, struct device_node *dev)
+{
+	struct device_node *root = find_path_device("/");
+	struct device_node *rtas;
+
+	rtas = of_find_node_by_name (root, "rtas");
+	if (rtas) {
+		hose->ops = &rtas_pci_ops;
+	} else {
+		printk ("RTAS supporting Pegasos OF not found, please upgrade"
+			" your firmware\n");
+	}
+	pci_assign_all_buses = 1;
+}
+
+void __init
+chrp_find_bridges(void)
+{
+	struct device_node *dev;
+	int *bus_range;
+	int len, index = -1;
+	struct pci_controller *hose;
+	unsigned int *dma;
+	char *model, *machine;
+	int is_longtrail = 0, is_mot = 0, is_pegasos = 0;
+	struct device_node *root = find_path_device("/");
+
+	/*
+	 * The PCI host bridge nodes on some machines don't have
+	 * properties to adequately identify them, so we have to
+	 * look at what sort of machine this is as well.
+	 */
+	machine = get_property(root, "model", NULL);
+	if (machine != NULL) {
+		is_longtrail = strncmp(machine, "IBM,LongTrail", 13) == 0;
+		is_mot = strncmp(machine, "MOT", 3) == 0;
+		if (strncmp(machine, "Pegasos2", 8) == 0)
+			is_pegasos = 2;
+		else if (strncmp(machine, "Pegasos", 7) == 0)
+			is_pegasos = 1;
+	}
+	for (dev = root->child; dev != NULL; dev = dev->sibling) {
+		if (dev->type == NULL || strcmp(dev->type, "pci") != 0)
+			continue;
+		++index;
+		/* The GG2 bridge on the LongTrail doesn't have an address */
+		if (dev->n_addrs < 1 && !is_longtrail) {
+			printk(KERN_WARNING "Can't use %s: no address\n",
+			       dev->full_name);
+			continue;
+		}
+		bus_range = (int *) get_property(dev, "bus-range", &len);
+		if (bus_range == NULL || len < 2 * sizeof(int)) {
+			printk(KERN_WARNING "Can't get bus-range for %s\n",
+				dev->full_name);
+			continue;
+		}
+		if (bus_range[1] == bus_range[0])
+			printk(KERN_INFO "PCI bus %d", bus_range[0]);
+		else
+			printk(KERN_INFO "PCI buses %d..%d",
+			       bus_range[0], bus_range[1]);
+		printk(" controlled by %s", dev->type);
+		if (dev->n_addrs > 0)
+			printk(" at %lx", dev->addrs[0].address);
+		printk("\n");
+
+		hose = pcibios_alloc_controller();
+		if (!hose) {
+			printk("Can't allocate PCI controller structure for %s\n",
+				dev->full_name);
+			continue;
+		}
+		hose->arch_data = dev;
+		hose->first_busno = bus_range[0];
+		hose->last_busno = bus_range[1];
+
+		model = get_property(dev, "model", NULL);
+		if (model == NULL)
+			model = "<none>";
+		if (device_is_compatible(dev, "IBM,python")) {
+			setup_python(hose, dev);
+		} else if (is_mot
+			   || strncmp(model, "Motorola, Grackle", 17) == 0) {
+			setup_grackle(hose);
+		} else if (is_longtrail) {
+			void __iomem *p = ioremap(GG2_PCI_CONFIG_BASE, 0x80000);
+			hose->ops = &gg2_pci_ops;
+			hose->cfg_data = p;
+			gg2_pci_config_base = p;
+		} else if (is_pegasos == 1) {
+			setup_indirect_pci(hose, 0xfec00cf8, 0xfee00cfc);
+		} else if (is_pegasos == 2) {
+			setup_peg2(hose, dev);
+		} else {
+			printk("No methods for %s (model %s), using RTAS\n",
+			       dev->full_name, model);
+			hose->ops = &rtas_pci_ops;
+		}
+
+		pci_process_bridge_OF_ranges(hose, dev, index == 0);
+
+		/* check the first bridge for a property that we can
+		   use to set pci_dram_offset */
+		dma = (unsigned int *)
+			get_property(dev, "ibm,dma-ranges", &len);
+		if (index == 0 && dma != NULL && len >= 6 * sizeof(*dma)) {
+			pci_dram_offset = dma[2] - dma[3];
+			printk("pci_dram_offset = %lx\n", pci_dram_offset);
+		}
+	}
+
+	/* Do not fixup interrupts from OF tree on pegasos */
+	if (is_pegasos == 0)
+		ppc_md.pcibios_fixup = chrp_pcibios_fixup;
+}
diff --git a/arch/powerpc/platforms/chrp/pegasos_eth.c b/arch/powerpc/platforms/chrp/pegasos_eth.c
new file mode 100644
index 000000000000..29c86781c493
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/pegasos_eth.c
@@ -0,0 +1,214 @@
+/*
+ *  arch/ppc/platforms/chrp_pegasos_eth.c
+ *
+ *  Copyright (C) 2005 Sven Luther <sl@bplan-gmbh.de>
+ *  Thanks to :
+ *	Dale Farnsworth <dale@farnsworth.org>
+ *	Mark A. Greer <mgreer@mvista.com>
+ *	Nicolas DET <nd@bplan-gmbh.de>
+ *	Benjamin Herrenschmidt <benh@kernel.crashing.org>
+ *  And anyone else who helped me on this.
+ */
+
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/ioport.h>
+#include <linux/device.h>
+#include <linux/platform_device.h>
+#include <linux/mv643xx.h>
+#include <linux/pci.h>
+
+#define PEGASOS2_MARVELL_REGBASE 		(0xf1000000)
+#define PEGASOS2_MARVELL_REGSIZE 		(0x00004000)
+#define PEGASOS2_SRAM_BASE 			(0xf2000000)
+#define PEGASOS2_SRAM_SIZE			(256*1024)
+
+#define PEGASOS2_SRAM_BASE_ETH0			(PEGASOS2_SRAM_BASE)
+#define PEGASOS2_SRAM_BASE_ETH1			(PEGASOS2_SRAM_BASE_ETH0 + (PEGASOS2_SRAM_SIZE / 2) )
+
+
+#define PEGASOS2_SRAM_RXRING_SIZE		(PEGASOS2_SRAM_SIZE/4)
+#define PEGASOS2_SRAM_TXRING_SIZE		(PEGASOS2_SRAM_SIZE/4)
+
+#undef BE_VERBOSE
+
+static struct resource mv643xx_eth_shared_resources[] = {
+	[0] = {
+		.name	= "ethernet shared base",
+		.start	= 0xf1000000 + MV643XX_ETH_SHARED_REGS,
+		.end	= 0xf1000000 + MV643XX_ETH_SHARED_REGS +
+					MV643XX_ETH_SHARED_REGS_SIZE - 1,
+		.flags	= IORESOURCE_MEM,
+	},
+};
+
+static struct platform_device mv643xx_eth_shared_device = {
+	.name		= MV643XX_ETH_SHARED_NAME,
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(mv643xx_eth_shared_resources),
+	.resource	= mv643xx_eth_shared_resources,
+};
+
+static struct resource mv643xx_eth0_resources[] = {
+	[0] = {
+		.name	= "eth0 irq",
+		.start	= 9,
+		.end	= 9,
+		.flags	= IORESOURCE_IRQ,
+	},
+};
+
+
+static struct mv643xx_eth_platform_data eth0_pd = {
+	.tx_sram_addr = PEGASOS2_SRAM_BASE_ETH0,
+	.tx_sram_size = PEGASOS2_SRAM_TXRING_SIZE,
+	.tx_queue_size = PEGASOS2_SRAM_TXRING_SIZE/16,
+
+	.rx_sram_addr = PEGASOS2_SRAM_BASE_ETH0 + PEGASOS2_SRAM_TXRING_SIZE,
+	.rx_sram_size = PEGASOS2_SRAM_RXRING_SIZE,
+	.rx_queue_size = PEGASOS2_SRAM_RXRING_SIZE/16,
+};
+
+static struct platform_device eth0_device = {
+	.name		= MV643XX_ETH_NAME,
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(mv643xx_eth0_resources),
+	.resource	= mv643xx_eth0_resources,
+	.dev = {
+		.platform_data = &eth0_pd,
+	},
+};
+
+static struct resource mv643xx_eth1_resources[] = {
+	[0] = {
+		.name	= "eth1 irq",
+		.start	= 9,
+		.end	= 9,
+		.flags	= IORESOURCE_IRQ,
+	},
+};
+
+static struct mv643xx_eth_platform_data eth1_pd = {
+	.tx_sram_addr = PEGASOS2_SRAM_BASE_ETH1,
+	.tx_sram_size = PEGASOS2_SRAM_TXRING_SIZE,
+	.tx_queue_size = PEGASOS2_SRAM_TXRING_SIZE/16,
+
+	.rx_sram_addr = PEGASOS2_SRAM_BASE_ETH1 + PEGASOS2_SRAM_TXRING_SIZE,
+	.rx_sram_size = PEGASOS2_SRAM_RXRING_SIZE,
+	.rx_queue_size = PEGASOS2_SRAM_RXRING_SIZE/16,
+};
+
+static struct platform_device eth1_device = {
+	.name		= MV643XX_ETH_NAME,
+	.id		= 1,
+	.num_resources	= ARRAY_SIZE(mv643xx_eth1_resources),
+	.resource	= mv643xx_eth1_resources,
+	.dev = {
+		.platform_data = &eth1_pd,
+	},
+};
+
+static struct platform_device *mv643xx_eth_pd_devs[] __initdata = {
+	&mv643xx_eth_shared_device,
+	&eth0_device,
+	&eth1_device,
+};
+
+/***********/
+/***********/
+#define MV_READ(offset,val) 	{ val = readl(mv643xx_reg_base + offset); }
+#define MV_WRITE(offset,data) writel(data, mv643xx_reg_base + offset)
+
+static void __iomem *mv643xx_reg_base;
+
+static int Enable_SRAM(void)
+{
+	u32 ALong;
+
+	if (mv643xx_reg_base == NULL)
+		mv643xx_reg_base = ioremap(PEGASOS2_MARVELL_REGBASE,
+					PEGASOS2_MARVELL_REGSIZE);
+
+	if (mv643xx_reg_base == NULL)
+		return -ENOMEM;
+
+#ifdef BE_VERBOSE
+	printk("Pegasos II/Marvell MV64361: register remapped from %p to %p\n",
+		(void *)PEGASOS2_MARVELL_REGBASE, (void *)mv643xx_reg_base);
+#endif
+
+	MV_WRITE(MV64340_SRAM_CONFIG, 0);
+
+	MV_WRITE(MV64340_INTEGRATED_SRAM_BASE_ADDR, PEGASOS2_SRAM_BASE >> 16);
+
+	MV_READ(MV64340_BASE_ADDR_ENABLE, ALong);
+	ALong &= ~(1 << 19);
+	MV_WRITE(MV64340_BASE_ADDR_ENABLE, ALong);
+
+	ALong = 0x02;
+	ALong |= PEGASOS2_SRAM_BASE & 0xffff0000;
+	MV_WRITE(MV643XX_ETH_BAR_4, ALong);
+
+	MV_WRITE(MV643XX_ETH_SIZE_REG_4, (PEGASOS2_SRAM_SIZE-1) & 0xffff0000);
+
+	MV_READ(MV643XX_ETH_BASE_ADDR_ENABLE_REG, ALong);
+	ALong &= ~(1 << 4);
+	MV_WRITE(MV643XX_ETH_BASE_ADDR_ENABLE_REG, ALong);
+
+#ifdef BE_VERBOSE
+	printk("Pegasos II/Marvell MV64361: register unmapped\n");
+	printk("Pegasos II/Marvell MV64361: SRAM at %p, size=%x\n", (void*) PEGASOS2_SRAM_BASE, PEGASOS2_SRAM_SIZE);
+#endif
+
+	iounmap(mv643xx_reg_base);
+	mv643xx_reg_base = NULL;
+
+	return 1;
+}
+
+
+/***********/
+/***********/
+int mv643xx_eth_add_pds(void)
+{
+	int ret = 0;
+	static struct pci_device_id pci_marvell_mv64360[] = {
+		{ PCI_DEVICE(PCI_VENDOR_ID_MARVELL, PCI_DEVICE_ID_MARVELL_MV64360) },
+		{ }
+	};
+
+#ifdef BE_VERBOSE
+	printk("Pegasos II/Marvell MV64361: init\n");
+#endif
+
+	if (pci_dev_present(pci_marvell_mv64360)) {
+		ret = platform_add_devices(mv643xx_eth_pd_devs,
+				ARRAY_SIZE(mv643xx_eth_pd_devs));
+
+		if ( Enable_SRAM() < 0)
+		{
+			eth0_pd.tx_sram_addr = 0;
+			eth0_pd.tx_sram_size = 0;
+			eth0_pd.rx_sram_addr = 0;
+			eth0_pd.rx_sram_size = 0;
+
+			eth1_pd.tx_sram_addr = 0;
+			eth1_pd.tx_sram_size = 0;
+			eth1_pd.rx_sram_addr = 0;
+			eth1_pd.rx_sram_size = 0;
+
+#ifdef BE_VERBOSE
+			printk("Pegasos II/Marvell MV64361: Can't enable the "
+				"SRAM\n");
+#endif
+		}
+	}
+
+#ifdef BE_VERBOSE
+	printk("Pegasos II/Marvell MV64361: init is over\n");
+#endif
+
+	return ret;
+}
+
+device_initcall(mv643xx_eth_add_pds);
diff --git a/arch/powerpc/platforms/chrp/setup.c b/arch/powerpc/platforms/chrp/setup.c
new file mode 100644
index 000000000000..ecd32d5d85f4
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/setup.c
@@ -0,0 +1,522 @@
+/*
+ *  arch/ppc/platforms/setup.c
+ *
+ *  Copyright (C) 1995  Linus Torvalds
+ *  Adapted from 'alpha' version by Gary Thomas
+ *  Modified by Cort Dougan (cort@cs.nmt.edu)
+ */
+
+/*
+ * bootup setup stuff..
+ */
+
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/slab.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/major.h>
+#include <linux/interrupt.h>
+#include <linux/reboot.h>
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/version.h>
+#include <linux/adb.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+#include <linux/ide.h>
+#include <linux/console.h>
+#include <linux/seq_file.h>
+#include <linux/root_dev.h>
+#include <linux/initrd.h>
+#include <linux/module.h>
+
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/prom.h>
+#include <asm/gg2.h>
+#include <asm/pci-bridge.h>
+#include <asm/dma.h>
+#include <asm/machdep.h>
+#include <asm/irq.h>
+#include <asm/hydra.h>
+#include <asm/sections.h>
+#include <asm/time.h>
+#include <asm/btext.h>
+#include <asm/i8259.h>
+#include <asm/mpic.h>
+#include <asm/rtas.h>
+#include <asm/xmon.h>
+
+#include "chrp.h"
+
+void rtas_indicator_progress(char *, unsigned short);
+void btext_progress(char *, unsigned short);
+
+int _chrp_type;
+EXPORT_SYMBOL(_chrp_type);
+
+struct mpic *chrp_mpic;
+
+/*
+ * XXX this should be in xmon.h, but putting it there means xmon.h
+ * has to include <linux/interrupt.h> (to get irqreturn_t), which
+ * causes all sorts of problems.  -- paulus
+ */
+extern irqreturn_t xmon_irq(int, void *, struct pt_regs *);
+
+extern unsigned long loops_per_jiffy;
+
+#ifdef CONFIG_SMP
+extern struct smp_ops_t chrp_smp_ops;
+#endif
+
+static const char *gg2_memtypes[4] = {
+	"FPM", "SDRAM", "EDO", "BEDO"
+};
+static const char *gg2_cachesizes[4] = {
+	"256 KB", "512 KB", "1 MB", "Reserved"
+};
+static const char *gg2_cachetypes[4] = {
+	"Asynchronous", "Reserved", "Flow-Through Synchronous",
+	"Pipelined Synchronous"
+};
+static const char *gg2_cachemodes[4] = {
+	"Disabled", "Write-Through", "Copy-Back", "Transparent Mode"
+};
+
+void chrp_show_cpuinfo(struct seq_file *m)
+{
+	int i, sdramen;
+	unsigned int t;
+	struct device_node *root;
+	const char *model = "";
+
+	root = find_path_device("/");
+	if (root)
+		model = get_property(root, "model", NULL);
+	seq_printf(m, "machine\t\t: CHRP %s\n", model);
+
+	/* longtrail (goldengate) stuff */
+	if (!strncmp(model, "IBM,LongTrail", 13)) {
+		/* VLSI VAS96011/12 `Golden Gate 2' */
+		/* Memory banks */
+		sdramen = (in_le32(gg2_pci_config_base + GG2_PCI_DRAM_CTRL)
+			   >>31) & 1;
+		for (i = 0; i < (sdramen ? 4 : 6); i++) {
+			t = in_le32(gg2_pci_config_base+
+						 GG2_PCI_DRAM_BANK0+
+						 i*4);
+			if (!(t & 1))
+				continue;
+			switch ((t>>8) & 0x1f) {
+			case 0x1f:
+				model = "4 MB";
+				break;
+			case 0x1e:
+				model = "8 MB";
+				break;
+			case 0x1c:
+				model = "16 MB";
+				break;
+			case 0x18:
+				model = "32 MB";
+				break;
+			case 0x10:
+				model = "64 MB";
+				break;
+			case 0x00:
+				model = "128 MB";
+				break;
+			default:
+				model = "Reserved";
+				break;
+			}
+			seq_printf(m, "memory bank %d\t: %s %s\n", i, model,
+				   gg2_memtypes[sdramen ? 1 : ((t>>1) & 3)]);
+		}
+		/* L2 cache */
+		t = in_le32(gg2_pci_config_base+GG2_PCI_CC_CTRL);
+		seq_printf(m, "board l2\t: %s %s (%s)\n",
+			   gg2_cachesizes[(t>>7) & 3],
+			   gg2_cachetypes[(t>>2) & 3],
+			   gg2_cachemodes[t & 3]);
+	}
+}
+
+/*
+ *  Fixes for the National Semiconductor PC78308VUL SuperI/O
+ *
+ *  Some versions of Open Firmware incorrectly initialize the IRQ settings
+ *  for keyboard and mouse
+ */
+static inline void __init sio_write(u8 val, u8 index)
+{
+	outb(index, 0x15c);
+	outb(val, 0x15d);
+}
+
+static inline u8 __init sio_read(u8 index)
+{
+	outb(index, 0x15c);
+	return inb(0x15d);
+}
+
+static void __init sio_fixup_irq(const char *name, u8 device, u8 level,
+				     u8 type)
+{
+	u8 level0, type0, active;
+
+	/* select logical device */
+	sio_write(device, 0x07);
+	active = sio_read(0x30);
+	level0 = sio_read(0x70);
+	type0 = sio_read(0x71);
+	if (level0 != level || type0 != type || !active) {
+		printk(KERN_WARNING "sio: %s irq level %d, type %d, %sactive: "
+		       "remapping to level %d, type %d, active\n",
+		       name, level0, type0, !active ? "in" : "", level, type);
+		sio_write(0x01, 0x30);
+		sio_write(level, 0x70);
+		sio_write(type, 0x71);
+	}
+}
+
+static void __init sio_init(void)
+{
+	struct device_node *root;
+
+	if ((root = find_path_device("/")) &&
+	    !strncmp(get_property(root, "model", NULL), "IBM,LongTrail", 13)) {
+		/* logical device 0 (KBC/Keyboard) */
+		sio_fixup_irq("keyboard", 0, 1, 2);
+		/* select logical device 1 (KBC/Mouse) */
+		sio_fixup_irq("mouse", 1, 12, 2);
+	}
+}
+
+
+static void __init pegasos_set_l2cr(void)
+{
+	struct device_node *np;
+
+	/* On Pegasos, enable the l2 cache if needed, as the OF forgets it */
+	if (_chrp_type != _CHRP_Pegasos)
+		return;
+
+	/* Enable L2 cache if needed */
+	np = find_type_devices("cpu");
+	if (np != NULL) {
+		unsigned int *l2cr = (unsigned int *)
+			get_property (np, "l2cr", NULL);
+		if (l2cr == NULL) {
+			printk ("Pegasos l2cr : no cpu l2cr property found\n");
+			return;
+		}
+		if (!((*l2cr) & 0x80000000)) {
+			printk ("Pegasos l2cr : L2 cache was not active, "
+				"activating\n");
+			_set_L2CR(0);
+			_set_L2CR((*l2cr) | 0x80000000);
+		}
+	}
+}
+
+void __init chrp_setup_arch(void)
+{
+	struct device_node *root = find_path_device ("/");
+	char *machine = NULL;
+	struct device_node *device;
+	unsigned int *p = NULL;
+
+	/* init to some ~sane value until calibrate_delay() runs */
+	loops_per_jiffy = 50000000/HZ;
+
+	if (root)
+		machine = get_property(root, "model", NULL);
+	if (machine && strncmp(machine, "Pegasos", 7) == 0) {
+		_chrp_type = _CHRP_Pegasos;
+	} else if (machine && strncmp(machine, "IBM", 3) == 0) {
+		_chrp_type = _CHRP_IBM;
+	} else if (machine && strncmp(machine, "MOT", 3) == 0) {
+		_chrp_type = _CHRP_Motorola;
+	} else {
+		/* Let's assume it is an IBM chrp if all else fails */
+		_chrp_type = _CHRP_IBM;
+	}
+	printk("chrp type = %x\n", _chrp_type);
+
+	rtas_initialize();
+	if (rtas_token("display-character") >= 0)
+		ppc_md.progress = rtas_progress;
+
+#ifdef CONFIG_BOOTX_TEXT
+	if (ppc_md.progress == NULL && boot_text_mapped)
+		ppc_md.progress = btext_progress;
+#endif
+
+#ifdef CONFIG_BLK_DEV_INITRD
+	/* this is fine for chrp */
+	initrd_below_start_ok = 1;
+
+	if (initrd_start)
+		ROOT_DEV = Root_RAM0;
+	else
+#endif
+		ROOT_DEV = Root_SDA2; /* sda2 (sda1 is for the kernel) */
+
+	/* On pegasos, enable the L2 cache if not already done by OF */
+	pegasos_set_l2cr();
+
+	/* Lookup PCI host bridges */
+	chrp_find_bridges();
+
+	/*
+	 *  Temporary fixes for PCI devices.
+	 *  -- Geert
+	 */
+	hydra_init();		/* Mac I/O */
+
+	/*
+	 *  Fix the Super I/O configuration
+	 */
+	sio_init();
+
+	/* Get the event scan rate for the rtas so we know how
+	 * often it expects a heartbeat. -- Cort
+	 */
+	device = find_devices("rtas");
+	if (device)
+		p = (unsigned int *) get_property
+			(device, "rtas-event-scan-rate", NULL);
+	if (p && *p) {
+		ppc_md.heartbeat = chrp_event_scan;
+		ppc_md.heartbeat_reset = HZ / (*p * 30) - 1;
+		ppc_md.heartbeat_count = 1;
+		printk("RTAS Event Scan Rate: %u (%lu jiffies)\n",
+		       *p, ppc_md.heartbeat_reset);
+	}
+
+	pci_create_OF_bus_map();
+
+	/*
+	 * Print the banner, then scroll down so boot progress
+	 * can be printed.  -- Cort
+	 */
+	if (ppc_md.progress) ppc_md.progress("Linux/PPC "UTS_RELEASE"\n", 0x0);
+}
+
+void
+chrp_event_scan(void)
+{
+	unsigned char log[1024];
+	int ret = 0;
+
+	/* XXX: we should loop until the hardware says no more error logs -- Cort */
+	rtas_call(rtas_token("event-scan"), 4, 1, &ret, 0xffffffff, 0,
+		  __pa(log), 1024);
+	ppc_md.heartbeat_count = ppc_md.heartbeat_reset;
+}
+
+/*
+ * Finds the open-pic node and sets up the mpic driver.
+ */
+static void __init chrp_find_openpic(void)
+{
+	struct device_node *np, *root;
+	int len, i, j, irq_count;
+	int isu_size, idu_size;
+	unsigned int *iranges, *opprop = NULL;
+	int oplen = 0;
+	unsigned long opaddr;
+	int na = 1;
+	unsigned char init_senses[NR_IRQS - NUM_8259_INTERRUPTS];
+
+	np = find_type_devices("open-pic");
+	if (np == NULL)
+		return;
+	root = find_path_device("/");
+	if (root) {
+		opprop = (unsigned int *) get_property
+			(root, "platform-open-pic", &oplen);
+		na = prom_n_addr_cells(root);
+	}
+	if (opprop && oplen >= na * sizeof(unsigned int)) {
+		opaddr = opprop[na-1];	/* assume 32-bit */
+		oplen /= na * sizeof(unsigned int);
+	} else {
+		if (np->n_addrs == 0)
+			return;
+		opaddr = np->addrs[0].address;
+		oplen = 0;
+	}
+
+	printk(KERN_INFO "OpenPIC at %lx\n", opaddr);
+
+	irq_count = NR_IRQS - NUM_ISA_INTERRUPTS - 4; /* leave room for IPIs */
+	prom_get_irq_senses(init_senses, NUM_8259_INTERRUPTS, NR_IRQS - 4);
+
+	iranges = (unsigned int *) get_property(np, "interrupt-ranges", &len);
+	if (iranges == NULL)
+		len = 0;	/* non-distributed mpic */
+	else
+		len /= 2 * sizeof(unsigned int);
+
+	/*
+	 * The first pair of cells in interrupt-ranges refers to the
+	 * IDU; subsequent pairs refer to the ISUs.
+	 */
+	if (oplen < len) {
+		printk(KERN_ERR "Insufficient addresses for distributed"
+		       " OpenPIC (%d < %d)\n", np->n_addrs, len);
+		len = oplen;
+	}
+
+	isu_size = 0;
+	idu_size = 0;
+	if (len > 0 && iranges[1] != 0) {
+		printk(KERN_INFO "OpenPIC irqs %d..%d in IDU\n",
+		       iranges[0], iranges[0] + iranges[1] - 1);
+		idu_size = iranges[1];
+	}
+	if (len > 1)
+		isu_size = iranges[3];
+
+	chrp_mpic = mpic_alloc(opaddr, MPIC_PRIMARY,
+			       isu_size, NUM_ISA_INTERRUPTS, irq_count,
+			       NR_IRQS - 4, init_senses, irq_count,
+			       " MPIC    ");
+	if (chrp_mpic == NULL) {
+		printk(KERN_ERR "Failed to allocate MPIC structure\n");
+		return;
+	}
+
+	j = na - 1;
+	for (i = 1; i < len; ++i) {
+		iranges += 2;
+		j += na;
+		printk(KERN_INFO "OpenPIC irqs %d..%d in ISU at %x\n",
+		       iranges[0], iranges[0] + iranges[1] - 1,
+		       opprop[j]);
+		mpic_assign_isu(chrp_mpic, i - 1, opprop[j]);
+	}
+
+	mpic_init(chrp_mpic);
+	mpic_setup_cascade(NUM_ISA_INTERRUPTS, i8259_irq_cascade, NULL);
+}
+
+#if defined(CONFIG_VT) && defined(CONFIG_INPUT_ADBHID) && defined(XMON)
+static struct irqaction xmon_irqaction = {
+	.handler = xmon_irq,
+	.mask = CPU_MASK_NONE,
+	.name = "XMON break",
+};
+#endif
+
+void __init chrp_init_IRQ(void)
+{
+	struct device_node *np;
+	unsigned long chrp_int_ack = 0;
+#if defined(CONFIG_VT) && defined(CONFIG_INPUT_ADBHID) && defined(XMON)
+	struct device_node *kbd;
+#endif
+
+	for (np = find_devices("pci"); np != NULL; np = np->next) {
+		unsigned int *addrp = (unsigned int *)
+			get_property(np, "8259-interrupt-acknowledge", NULL);
+
+		if (addrp == NULL)
+			continue;
+		chrp_int_ack = addrp[prom_n_addr_cells(np)-1];
+		break;
+	}
+	if (np == NULL)
+		printk(KERN_ERR "Cannot find PCI interrupt acknowledge address\n");
+
+	chrp_find_openpic();
+
+	i8259_init(chrp_int_ack, 0);
+
+	if (_chrp_type == _CHRP_Pegasos)
+		ppc_md.get_irq        = i8259_irq;
+	else
+		ppc_md.get_irq        = mpic_get_irq;
+
+#if defined(CONFIG_VT) && defined(CONFIG_INPUT_ADBHID) && defined(XMON)
+	/* see if there is a keyboard in the device tree
+	   with a parent of type "adb" */
+	for (kbd = find_devices("keyboard"); kbd; kbd = kbd->next)
+		if (kbd->parent && kbd->parent->type
+		    && strcmp(kbd->parent->type, "adb") == 0)
+			break;
+	if (kbd)
+		setup_irq(HYDRA_INT_ADB_NMI, &xmon_irqaction);
+#endif
+}
+
+void __init
+chrp_init2(void)
+{
+#ifdef CONFIG_NVRAM
+	chrp_nvram_init();
+#endif
+
+	request_region(0x20,0x20,"pic1");
+	request_region(0xa0,0x20,"pic2");
+	request_region(0x00,0x20,"dma1");
+	request_region(0x40,0x20,"timer");
+	request_region(0x80,0x10,"dma page reg");
+	request_region(0xc0,0x20,"dma2");
+
+	if (ppc_md.progress)
+		ppc_md.progress("  Have fun!    ", 0x7777);
+}
+
+void __init chrp_init(void)
+{
+	ISA_DMA_THRESHOLD = ~0L;
+	DMA_MODE_READ = 0x44;
+	DMA_MODE_WRITE = 0x48;
+	isa_io_base = CHRP_ISA_IO_BASE;		/* default value */
+	ppc_do_canonicalize_irqs = 1;
+
+	/* Assume we have an 8259... */
+	__irq_offset_value = NUM_ISA_INTERRUPTS;
+
+	ppc_md.setup_arch     = chrp_setup_arch;
+	ppc_md.show_cpuinfo   = chrp_show_cpuinfo;
+
+	ppc_md.init_IRQ       = chrp_init_IRQ;
+	ppc_md.init           = chrp_init2;
+
+	ppc_md.phys_mem_access_prot = pci_phys_mem_access_prot;
+
+	ppc_md.restart        = rtas_restart;
+	ppc_md.power_off      = rtas_power_off;
+	ppc_md.halt           = rtas_halt;
+
+	ppc_md.time_init      = chrp_time_init;
+	ppc_md.set_rtc_time   = chrp_set_rtc_time;
+	ppc_md.get_rtc_time   = chrp_get_rtc_time;
+	ppc_md.calibrate_decr = chrp_calibrate_decr;
+
+#ifdef CONFIG_SMP
+	smp_ops = &chrp_smp_ops;
+#endif /* CONFIG_SMP */
+}
+
+#ifdef CONFIG_BOOTX_TEXT
+void
+btext_progress(char *s, unsigned short hex)
+{
+	btext_drawstring(s);
+	btext_drawstring("\n");
+}
+#endif /* CONFIG_BOOTX_TEXT */
diff --git a/arch/powerpc/platforms/chrp/smp.c b/arch/powerpc/platforms/chrp/smp.c
new file mode 100644
index 000000000000..bb2315997d45
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/smp.c
@@ -0,0 +1,85 @@
+/*
+ * Smp support for CHRP machines.
+ *
+ * Written by Cort Dougan (cort@cs.nmt.edu) borrowing a great
+ * deal of code from the sparc and intel versions.
+ *
+ * Copyright (C) 1999 Cort Dougan <cort@cs.nmt.edu>
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+#include <linux/kernel_stat.h>
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+
+#include <asm/ptrace.h>
+#include <asm/atomic.h>
+#include <asm/irq.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/sections.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/smp.h>
+#include <asm/residual.h>
+#include <asm/time.h>
+#include <asm/open_pic.h>
+#include <asm/machdep.h>
+#include <asm/smp.h>
+#include <asm/mpic.h>
+
+static void __devinit smp_chrp_kick_cpu(int nr)
+{
+	*(unsigned long *)KERNELBASE = nr;
+	asm volatile("dcbf 0,%0"::"r"(KERNELBASE):"memory");
+}
+
+static void __devinit smp_chrp_setup_cpu(int cpu_nr)
+{
+	mpic_setup_this_cpu();
+}
+
+static DEFINE_SPINLOCK(timebase_lock);
+static unsigned int timebase_upper = 0, timebase_lower = 0;
+
+void __devinit smp_chrp_give_timebase(void)
+{
+	spin_lock(&timebase_lock);
+	rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL);
+	timebase_upper = get_tbu();
+	timebase_lower = get_tbl();
+	spin_unlock(&timebase_lock);
+
+	while (timebase_upper || timebase_lower)
+		barrier();
+	rtas_call(rtas_token("thaw-time-base"), 0, 1, NULL);
+}
+
+void __devinit smp_chrp_take_timebase(void)
+{
+	while (!(timebase_upper || timebase_lower))
+		barrier();
+	spin_lock(&timebase_lock);
+	set_tb(timebase_upper, timebase_lower);
+	timebase_upper = 0;
+	timebase_lower = 0;
+	spin_unlock(&timebase_lock);
+	printk("CPU %i taken timebase\n", smp_processor_id());
+}
+
+/* CHRP with openpic */
+struct smp_ops_t chrp_smp_ops = {
+	.message_pass = smp_mpic_message_pass,
+	.probe = smp_mpic_probe,
+	.kick_cpu = smp_chrp_kick_cpu,
+	.setup_cpu = smp_chrp_setup_cpu,
+	.give_timebase = smp_chrp_give_timebase,
+	.take_timebase = smp_chrp_take_timebase,
+};
diff --git a/arch/powerpc/platforms/chrp/time.c b/arch/powerpc/platforms/chrp/time.c
new file mode 100644
index 000000000000..9e53535ddb82
--- /dev/null
+++ b/arch/powerpc/platforms/chrp/time.c
@@ -0,0 +1,188 @@
+/*
+ *  arch/ppc/platforms/chrp_time.c
+ *
+ *  Copyright (C) 1991, 1992, 1995  Linus Torvalds
+ *
+ * Adapted for PowerPC (PReP) by Gary Thomas
+ * Modified by Cort Dougan (cort@cs.nmt.edu).
+ * Copied and modified from arch/i386/kernel/time.c
+ *
+ */
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <linux/time.h>
+#include <linux/timex.h>
+#include <linux/kernel_stat.h>
+#include <linux/mc146818rtc.h>
+#include <linux/init.h>
+#include <linux/bcd.h>
+
+#include <asm/io.h>
+#include <asm/nvram.h>
+#include <asm/prom.h>
+#include <asm/sections.h>
+#include <asm/time.h>
+
+extern spinlock_t rtc_lock;
+
+static int nvram_as1 = NVRAM_AS1;
+static int nvram_as0 = NVRAM_AS0;
+static int nvram_data = NVRAM_DATA;
+
+long __init chrp_time_init(void)
+{
+	struct device_node *rtcs;
+	int base;
+
+	rtcs = find_compatible_devices("rtc", "pnpPNP,b00");
+	if (rtcs == NULL)
+		rtcs = find_compatible_devices("rtc", "ds1385-rtc");
+	if (rtcs == NULL || rtcs->addrs == NULL)
+		return 0;
+	base = rtcs->addrs[0].address;
+	nvram_as1 = 0;
+	nvram_as0 = base;
+	nvram_data = base + 1;
+
+	return 0;
+}
+
+int chrp_cmos_clock_read(int addr)
+{
+	if (nvram_as1 != 0)
+		outb(addr>>8, nvram_as1);
+	outb(addr, nvram_as0);
+	return (inb(nvram_data));
+}
+
+void chrp_cmos_clock_write(unsigned long val, int addr)
+{
+	if (nvram_as1 != 0)
+		outb(addr>>8, nvram_as1);
+	outb(addr, nvram_as0);
+	outb(val, nvram_data);
+	return;
+}
+
+/*
+ * Set the hardware clock. -- Cort
+ */
+int chrp_set_rtc_time(struct rtc_time *tmarg)
+{
+	unsigned char save_control, save_freq_select;
+	struct rtc_time tm = *tmarg;
+
+	spin_lock(&rtc_lock);
+
+	save_control = chrp_cmos_clock_read(RTC_CONTROL); /* tell the clock it's being set */
+
+	chrp_cmos_clock_write((save_control|RTC_SET), RTC_CONTROL);
+
+	save_freq_select = chrp_cmos_clock_read(RTC_FREQ_SELECT); /* stop and reset prescaler */
+
+	chrp_cmos_clock_write((save_freq_select|RTC_DIV_RESET2), RTC_FREQ_SELECT);
+
+        tm.tm_year -= 1900;
+	if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
+		BIN_TO_BCD(tm.tm_sec);
+		BIN_TO_BCD(tm.tm_min);
+		BIN_TO_BCD(tm.tm_hour);
+		BIN_TO_BCD(tm.tm_mon);
+		BIN_TO_BCD(tm.tm_mday);
+		BIN_TO_BCD(tm.tm_year);
+	}
+	chrp_cmos_clock_write(tm.tm_sec,RTC_SECONDS);
+	chrp_cmos_clock_write(tm.tm_min,RTC_MINUTES);
+	chrp_cmos_clock_write(tm.tm_hour,RTC_HOURS);
+	chrp_cmos_clock_write(tm.tm_mon,RTC_MONTH);
+	chrp_cmos_clock_write(tm.tm_mday,RTC_DAY_OF_MONTH);
+	chrp_cmos_clock_write(tm.tm_year,RTC_YEAR);
+
+	/* The following flags have to be released exactly in this order,
+	 * otherwise the DS12887 (popular MC146818A clone with integrated
+	 * battery and quartz) will not reset the oscillator and will not
+	 * update precisely 500 ms later. You won't find this mentioned in
+	 * the Dallas Semiconductor data sheets, but who believes data
+	 * sheets anyway ...                           -- Markus Kuhn
+	 */
+	chrp_cmos_clock_write(save_control, RTC_CONTROL);
+	chrp_cmos_clock_write(save_freq_select, RTC_FREQ_SELECT);
+
+	spin_unlock(&rtc_lock);
+	return 0;
+}
+
+void chrp_get_rtc_time(struct rtc_time *tm)
+{
+	unsigned int year, mon, day, hour, min, sec;
+	int uip, i;
+
+	/* The Linux interpretation of the CMOS clock register contents:
+	 * When the Update-In-Progress (UIP) flag goes from 1 to 0, the
+	 * RTC registers show the second which has precisely just started.
+	 * Let's hope other operating systems interpret the RTC the same way.
+	 */
+
+	/* Since the UIP flag is set for about 2.2 ms and the clock
+	 * is typically written with a precision of 1 jiffy, trying
+	 * to obtain a precision better than a few milliseconds is
+	 * an illusion. Only consistency is interesting, this also
+	 * allows to use the routine for /dev/rtc without a potential
+	 * 1 second kernel busy loop triggered by any reader of /dev/rtc.
+	 */
+
+	for ( i = 0; i<1000000; i++) {
+		uip = chrp_cmos_clock_read(RTC_FREQ_SELECT);
+		sec = chrp_cmos_clock_read(RTC_SECONDS);
+		min = chrp_cmos_clock_read(RTC_MINUTES);
+		hour = chrp_cmos_clock_read(RTC_HOURS);
+		day = chrp_cmos_clock_read(RTC_DAY_OF_MONTH);
+		mon = chrp_cmos_clock_read(RTC_MONTH);
+		year = chrp_cmos_clock_read(RTC_YEAR);
+		uip |= chrp_cmos_clock_read(RTC_FREQ_SELECT);
+		if ((uip & RTC_UIP)==0) break;
+	}
+
+	if (!(chrp_cmos_clock_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;
+	tm->tm_sec = sec;
+	tm->tm_min = min;
+	tm->tm_hour = hour;
+	tm->tm_mday = day;
+	tm->tm_mon = mon;
+	tm->tm_year = year;
+}
+
+
+void __init chrp_calibrate_decr(void)
+{
+	struct device_node *cpu;
+	unsigned int freq, *fp;
+
+	/*
+	 * The cpu node should have a timebase-frequency property
+	 * to tell us the rate at which the decrementer counts.
+	 */
+	freq = 16666000;		/* hardcoded default */
+	cpu = find_type_devices("cpu");
+	if (cpu != 0) {
+		fp = (unsigned int *)
+			get_property(cpu, "timebase-frequency", NULL);
+		if (fp != 0)
+			freq = *fp;
+	}
+	ppc_tb_freq = freq;
+}
diff --git a/arch/powerpc/platforms/embedded6xx/Kconfig b/arch/powerpc/platforms/embedded6xx/Kconfig
new file mode 100644
index 000000000000..81250090f98d
--- /dev/null
+++ b/arch/powerpc/platforms/embedded6xx/Kconfig
@@ -0,0 +1,318 @@
+choice
+	prompt "Machine Type"
+	depends on EMBEDDED6xx
+
+config KATANA
+	bool "Artesyn-Katana"
+	help
+	  Select KATANA if configuring an Artesyn KATANA 750i or 3750
+	  cPCI board.
+
+config WILLOW
+	bool "Cogent-Willow"
+
+config CPCI690
+	bool "Force-CPCI690"
+	help
+	  Select CPCI690 if configuring a Force CPCI690 cPCI board.
+
+config POWERPMC250
+	bool "Force-PowerPMC250"
+
+config CHESTNUT
+	bool "IBM 750FX Eval board or 750GX Eval board"
+	help
+	  Select CHESTNUT if configuring an IBM 750FX Eval Board or a
+	  IBM 750GX Eval board.
+
+config SPRUCE
+	bool "IBM-Spruce"
+	select PPC_INDIRECT_PCI
+
+config HDPU
+	bool "Sky-HDPU"
+	help
+	  Select HDPU if configuring a Sky Computers Compute Blade.
+
+config HDPU_FEATURES
+	depends HDPU
+	tristate "HDPU-Features"
+	help
+	  Select to enable HDPU enhanced features.
+
+config EV64260
+	bool "Marvell-EV64260BP"
+	help
+	  Select EV64260 if configuring a Marvell (formerly Galileo)
+	  EV64260BP Evaluation platform.
+
+config LOPEC
+	bool "Motorola-LoPEC"
+	select PPC_I8259
+
+config MVME5100
+	bool "Motorola-MVME5100"
+	select PPC_INDIRECT_PCI
+
+config PPLUS
+	bool "Motorola-PowerPlus"
+	select PPC_I8259
+	select PPC_INDIRECT_PCI
+
+config PRPMC750
+	bool "Motorola-PrPMC750"
+	select PPC_INDIRECT_PCI
+
+config PRPMC800
+	bool "Motorola-PrPMC800"
+	select PPC_INDIRECT_PCI
+
+config SANDPOINT
+	bool "Motorola-Sandpoint"
+	select PPC_I8259
+	help
+	  Select SANDPOINT if configuring for a Motorola Sandpoint X3
+	  (any flavor).
+
+config RADSTONE_PPC7D
+	bool "Radstone Technology PPC7D board"
+	select PPC_I8259
+
+config PAL4
+	bool "SBS-Palomar4"
+
+config GEMINI
+	bool "Synergy-Gemini"
+	select PPC_INDIRECT_PCI
+	depends on BROKEN
+	help
+	  Select Gemini if configuring for a Synergy Microsystems' Gemini
+	  series Single Board Computer.  More information is available at:
+	  <http://www.synergymicro.com/PressRel/97_10_15.html>.
+
+config EST8260
+	bool "EST8260"
+	---help---
+	  The EST8260 is a single-board computer manufactured by Wind River
+	  Systems, Inc. (formerly Embedded Support Tools Corp.) and based on
+	  the MPC8260.  Wind River Systems has a website at
+	  <http://www.windriver.com/>, but the EST8260 cannot be found on it
+	  and has probably been discontinued or rebadged.
+
+config SBC82xx
+	bool "SBC82xx"
+	---help---
+	  SBC PowerQUICC II, single-board computer with MPC82xx CPU
+	  Manufacturer: Wind River Systems, Inc.
+	  Date of Release: May 2003
+	  End of Life: -
+	  URL: <http://www.windriver.com/>
+
+config SBS8260
+	bool "SBS8260"
+
+config RPX8260
+	bool "RPXSUPER"
+
+config TQM8260
+	bool "TQM8260"
+	---help---
+	  MPC8260 based module, little larger than credit card,
+	  up to 128 MB global + 64 MB local RAM, 32 MB Flash,
+	  32 kB EEPROM, 256 kB L@ Cache, 10baseT + 100baseT Ethernet,
+	  2 x serial ports, ...
+	  Manufacturer: TQ Components, www.tq-group.de
+	  Date of Release: June 2001
+	  End of Life: not yet :-)
+	  URL: <http://www.denx.de/PDF/TQM82xx_SPEC_Rev005.pdf>
+
+config ADS8272
+	bool "ADS8272"
+
+config PQ2FADS
+	bool "Freescale-PQ2FADS"
+	help
+	  Select PQ2FADS if you wish to configure for a Freescale
+	  PQ2FADS board (-VR or -ZU).
+
+config LITE5200
+	bool "Freescale LITE5200 / (IceCube)"
+	select PPC_MPC52xx
+	help
+	  Support for the LITE5200 dev board for the MPC5200 from Freescale.
+	  This is for the LITE5200 version 2.0 board. Don't know if it changes
+	  much but it's only been tested on this board version. I think this
+	  board is also known as IceCube.
+
+config MPC834x_SYS
+	bool "Freescale MPC834x SYS"
+	help
+	  This option enables support for the MPC 834x SYS evaluation board.
+
+	  Be aware that PCI buses can only function when SYS board is plugged
+	  into the PIB (Platform IO Board) board from Freescale which provide
+	  3 PCI slots.  The PIBs PCI initialization is the bootloader's
+	  responsiblilty.
+
+config EV64360
+	bool "Marvell-EV64360BP"
+	help
+	  Select EV64360 if configuring a Marvell EV64360BP Evaluation
+	  platform.
+endchoice
+
+config PQ2ADS
+	bool
+	depends on ADS8272
+	default y
+
+config TQM8xxL
+	bool
+	depends on 8xx && (TQM823L || TQM850L || FPS850L || TQM855L || TQM860L)
+	default y
+
+config PPC_MPC52xx
+	bool
+
+config 8260
+	bool "CPM2 Support" if WILLOW
+	depends on 6xx
+	default y if TQM8260 || RPX8260 || EST8260 || SBS8260 || SBC82xx || PQ2FADS
+	help
+	  The MPC8260 is a typical embedded CPU made by Motorola.  Selecting
+	  this option means that you wish to build a kernel for a machine with
+	  an 8260 class CPU.
+
+config 8272
+	bool
+	depends on 6xx
+	default y if ADS8272
+	select 8260
+	help
+	  The MPC8272 CPM has a different internal dpram setup than other CPM2
+	  devices
+
+config 83xx
+	bool
+	default y if MPC834x_SYS
+
+config MPC834x
+	bool
+	default y if MPC834x_SYS
+
+config CPM2
+	bool
+	depends on 8260 || MPC8560 || MPC8555
+	default y
+	help
+	  The CPM2 (Communications Processor Module) is a coprocessor on
+	  embedded CPUs made by Motorola.  Selecting this option means that
+	  you wish to build a kernel for a machine with a CPM2 coprocessor
+	  on it (826x, 827x, 8560).
+
+config PPC_GEN550
+	bool
+	depends on SANDPOINT || SPRUCE || PPLUS || \
+		PRPMC750 || PRPMC800 || LOPEC || \
+		(EV64260 && !SERIAL_MPSC) || CHESTNUT || RADSTONE_PPC7D || \
+		83xx
+	default y
+
+config FORCE
+	bool
+	depends on 6xx && POWERPMC250
+	default y
+
+config GT64260
+	bool
+	depends on EV64260 || CPCI690
+	default y
+
+config MV64360		# Really MV64360 & MV64460
+	bool
+	depends on CHESTNUT || KATANA || RADSTONE_PPC7D || HDPU || EV64360
+	default y
+
+config MV64X60
+	bool
+	depends on (GT64260 || MV64360)
+	select PPC_INDIRECT_PCI
+	default y
+
+menu "Set bridge options"
+	depends on MV64X60
+
+config NOT_COHERENT_CACHE
+	bool "Turn off Cache Coherency"
+	default n
+	help
+	  Some 64x60 bridges lock up when trying to enforce cache coherency.
+	  When this option is selected, cache coherency will be turned off.
+	  Note that this can cause other problems (e.g., stale data being
+	  speculatively loaded via a cached mapping).  Use at your own risk.
+
+config MV64X60_BASE
+	hex "Set bridge base used by firmware"
+	default "0xf1000000"
+	help
+	  A firmware can leave the base address of the bridge's registers at
+	  a non-standard location.  If so, set this value to reflect the
+	  address of that non-standard location.
+
+config MV64X60_NEW_BASE
+	hex "Set bridge base used by kernel"
+	default "0xf1000000"
+	help
+	  If the current base address of the bridge's registers is not where
+	  you want it, set this value to the address that you want it moved to.
+
+endmenu
+
+config NONMONARCH_SUPPORT
+	bool "Enable Non-Monarch Support"
+	depends on PRPMC800
+
+config HARRIER
+	bool
+	depends on PRPMC800
+	default y
+
+config EPIC_SERIAL_MODE
+	bool
+	depends on 6xx && (LOPEC || SANDPOINT)
+	default y
+
+config MPC10X_BRIDGE
+	bool
+	depends on POWERPMC250 || LOPEC || SANDPOINT
+	select PPC_INDIRECT_PCI
+	default y
+
+config MPC10X_OPENPIC
+	bool
+	depends on POWERPMC250 || LOPEC || SANDPOINT
+	default y
+
+config MPC10X_STORE_GATHERING
+	bool "Enable MPC10x store gathering"
+	depends on MPC10X_BRIDGE
+
+config SANDPOINT_ENABLE_UART1
+	bool "Enable DUART mode on Sandpoint"
+	depends on SANDPOINT
+	help
+	  If this option is enabled then the MPC824x processor will run
+	  in DUART mode instead of UART mode.
+
+config HARRIER_STORE_GATHERING
+	bool "Enable Harrier store gathering"
+	depends on HARRIER
+
+config MVME5100_IPMC761_PRESENT
+	bool "MVME5100 configured with an IPMC761"
+	depends on MVME5100
+	select PPC_I8259
+
+config SPRUCE_BAUD_33M
+	bool "Spruce baud clock support"
+	depends on SPRUCE
diff --git a/arch/powerpc/platforms/iseries/Kconfig b/arch/powerpc/platforms/iseries/Kconfig
new file mode 100644
index 000000000000..3d957a30c8c2
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/Kconfig
@@ -0,0 +1,31 @@
+
+menu "iSeries device drivers"
+	depends on PPC_ISERIES
+
+config VIOCONS
+	tristate "iSeries Virtual Console Support"
+
+config VIODASD
+	tristate "iSeries Virtual I/O disk support"
+	help
+	  If you are running on an iSeries system and you want to use
+ 	  virtual disks created and managed by OS/400, say Y.
+
+config VIOCD
+	tristate "iSeries Virtual I/O CD support"
+	help
+	  If you are running Linux on an IBM iSeries system and you want to
+	  read a CD drive owned by OS/400, say Y here.
+
+config VIOTAPE
+	tristate "iSeries Virtual Tape Support"
+	help
+	  If you are running Linux on an iSeries system and you want Linux
+	  to read and/or write a tape drive owned by OS/400, say Y here.
+
+endmenu
+
+config VIOPATH
+	bool
+	depends on VIOCONS || VIODASD || VIOCD || VIOTAPE || VETH
+	default y
diff --git a/arch/powerpc/platforms/iseries/Makefile b/arch/powerpc/platforms/iseries/Makefile
new file mode 100644
index 000000000000..127b465308be
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/Makefile
@@ -0,0 +1,9 @@
+EXTRA_CFLAGS	+= -mno-minimal-toc
+
+obj-y += hvlog.o hvlpconfig.o lpardata.o setup.o mf.o lpevents.o \
+	hvcall.o proc.o htab.o iommu.o misc.o
+obj-$(CONFIG_PCI) += pci.o irq.o vpdinfo.o
+obj-$(CONFIG_IBMVIO) += vio.o
+obj-$(CONFIG_SMP) += smp.o
+obj-$(CONFIG_VIOPATH) += viopath.o
+obj-$(CONFIG_MODULES) += ksyms.o
diff --git a/arch/powerpc/platforms/iseries/call_hpt.h b/arch/powerpc/platforms/iseries/call_hpt.h
new file mode 100644
index 000000000000..a843b0f87b72
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/call_hpt.h
@@ -0,0 +1,101 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#ifndef _PLATFORMS_ISERIES_CALL_HPT_H
+#define _PLATFORMS_ISERIES_CALL_HPT_H
+
+/*
+ * This file contains the "hypervisor call" interface which is used to
+ * drive the hypervisor from the OS.
+ */
+
+#include <asm/iseries/hv_call_sc.h>
+#include <asm/iseries/hv_types.h>
+#include <asm/mmu.h>
+
+#define HvCallHptGetHptAddress		HvCallHpt +  0
+#define HvCallHptGetHptPages		HvCallHpt +  1
+#define HvCallHptSetPp			HvCallHpt +  5
+#define HvCallHptSetSwBits		HvCallHpt +  6
+#define HvCallHptUpdate			HvCallHpt +  7
+#define HvCallHptInvalidateNoSyncICache	HvCallHpt +  8
+#define HvCallHptGet			HvCallHpt + 11
+#define HvCallHptFindNextValid		HvCallHpt + 12
+#define HvCallHptFindValid		HvCallHpt + 13
+#define HvCallHptAddValidate		HvCallHpt + 16
+#define HvCallHptInvalidateSetSwBitsGet HvCallHpt + 18
+
+
+static inline u64 HvCallHpt_getHptAddress(void)
+{
+	return HvCall0(HvCallHptGetHptAddress);
+}
+
+static inline u64 HvCallHpt_getHptPages(void)
+{
+	return HvCall0(HvCallHptGetHptPages);
+}
+
+static inline void HvCallHpt_setPp(u32 hpteIndex, u8 value)
+{
+	HvCall2(HvCallHptSetPp, hpteIndex, value);
+}
+
+static inline void HvCallHpt_setSwBits(u32 hpteIndex, u8 bitson, u8 bitsoff)
+{
+	HvCall3(HvCallHptSetSwBits, hpteIndex, bitson, bitsoff);
+}
+
+static inline void HvCallHpt_invalidateNoSyncICache(u32 hpteIndex)
+{
+	HvCall1(HvCallHptInvalidateNoSyncICache, hpteIndex);
+}
+
+static inline u64 HvCallHpt_invalidateSetSwBitsGet(u32 hpteIndex, u8 bitson,
+		u8 bitsoff)
+{
+	u64 compressedStatus;
+
+	compressedStatus = HvCall4(HvCallHptInvalidateSetSwBitsGet,
+			hpteIndex, bitson, bitsoff, 1);
+	HvCall1(HvCallHptInvalidateNoSyncICache, hpteIndex);
+	return compressedStatus;
+}
+
+static inline u64 HvCallHpt_findValid(hpte_t *hpte, u64 vpn)
+{
+	return HvCall3Ret16(HvCallHptFindValid, hpte, vpn, 0, 0);
+}
+
+static inline u64 HvCallHpt_findNextValid(hpte_t *hpte, u32 hpteIndex,
+		u8 bitson, u8 bitsoff)
+{
+	return HvCall3Ret16(HvCallHptFindNextValid, hpte, hpteIndex,
+			bitson, bitsoff);
+}
+
+static inline void HvCallHpt_get(hpte_t *hpte, u32 hpteIndex)
+{
+	HvCall2Ret16(HvCallHptGet, hpte, hpteIndex, 0);
+}
+
+static inline void HvCallHpt_addValidate(u32 hpteIndex, u32 hBit, hpte_t *hpte)
+{
+	HvCall4(HvCallHptAddValidate, hpteIndex, hBit, hpte->v, hpte->r);
+}
+
+#endif /* _PLATFORMS_ISERIES_CALL_HPT_H */
diff --git a/arch/powerpc/platforms/iseries/call_pci.h b/arch/powerpc/platforms/iseries/call_pci.h
new file mode 100644
index 000000000000..59d4e0ad5cf3
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/call_pci.h
@@ -0,0 +1,290 @@
+/*
+ * Provides the Hypervisor PCI calls for iSeries Linux Parition.
+ * Copyright (C) 2001  <Wayne G Holm> <IBM Corporation>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the:
+ * Free Software Foundation, Inc.,
+ * 59 Temple Place, Suite 330,
+ * Boston, MA  02111-1307  USA
+ *
+ * Change Activity:
+ *   Created, Jan 9, 2001
+ */
+
+#ifndef _PLATFORMS_ISERIES_CALL_PCI_H
+#define _PLATFORMS_ISERIES_CALL_PCI_H
+
+#include <asm/iseries/hv_call_sc.h>
+#include <asm/iseries/hv_types.h>
+
+/*
+ * DSA == Direct Select Address
+ * this struct must be 64 bits in total
+ */
+struct HvCallPci_DsaAddr {
+	u16		busNumber;		/* PHB index? */
+	u8		subBusNumber;		/* PCI bus number? */
+	u8		deviceId;		/* device and function? */
+	u8		barNumber;
+	u8		reserved[3];
+};
+
+union HvDsaMap {
+	u64	DsaAddr;
+	struct HvCallPci_DsaAddr Dsa;
+};
+
+struct HvCallPci_LoadReturn {
+	u64		rc;
+	u64		value;
+};
+
+enum HvCallPci_DeviceType {
+	HvCallPci_NodeDevice	= 1,
+	HvCallPci_SpDevice	= 2,
+	HvCallPci_IopDevice     = 3,
+	HvCallPci_BridgeDevice	= 4,
+	HvCallPci_MultiFunctionDevice = 5,
+	HvCallPci_IoaDevice	= 6
+};
+
+
+struct HvCallPci_DeviceInfo {
+	u32	deviceType;		/* See DeviceType enum for values */
+};
+
+struct HvCallPci_BusUnitInfo {
+	u32	sizeReturned;		/* length of data returned */
+	u32	deviceType;		/* see DeviceType enum for values */
+};
+
+struct HvCallPci_BridgeInfo {
+	struct HvCallPci_BusUnitInfo busUnitInfo;  /* Generic bus unit info */
+	u8		subBusNumber;	/* Bus number of secondary bus */
+	u8		maxAgents;	/* Max idsels on secondary bus */
+        u8              maxSubBusNumber; /* Max Sub Bus */
+	u8		logicalSlotNumber; /* Logical Slot Number for IOA */
+};
+
+
+/*
+ * Maximum BusUnitInfo buffer size.  Provided for clients so
+ * they can allocate a buffer big enough for any type of bus
+ * unit.  Increase as needed.
+ */
+enum {HvCallPci_MaxBusUnitInfoSize = 128};
+
+struct HvCallPci_BarParms {
+	u64		vaddr;
+	u64		raddr;
+	u64		size;
+	u64		protectStart;
+	u64		protectEnd;
+	u64		relocationOffset;
+	u64		pciAddress;
+	u64		reserved[3];
+};
+
+enum HvCallPci_VpdType {
+	HvCallPci_BusVpd	= 1,
+	HvCallPci_BusAdapterVpd	= 2
+};
+
+#define HvCallPciConfigLoad8		HvCallPci + 0
+#define HvCallPciConfigLoad16		HvCallPci + 1
+#define HvCallPciConfigLoad32		HvCallPci + 2
+#define HvCallPciConfigStore8		HvCallPci + 3
+#define HvCallPciConfigStore16		HvCallPci + 4
+#define HvCallPciConfigStore32		HvCallPci + 5
+#define HvCallPciEoi			HvCallPci + 16
+#define HvCallPciGetBarParms		HvCallPci + 18
+#define HvCallPciMaskFisr		HvCallPci + 20
+#define HvCallPciUnmaskFisr		HvCallPci + 21
+#define HvCallPciSetSlotReset		HvCallPci + 25
+#define HvCallPciGetDeviceInfo		HvCallPci + 27
+#define HvCallPciGetCardVpd		HvCallPci + 28
+#define HvCallPciBarLoad8		HvCallPci + 40
+#define HvCallPciBarLoad16		HvCallPci + 41
+#define HvCallPciBarLoad32		HvCallPci + 42
+#define HvCallPciBarLoad64		HvCallPci + 43
+#define HvCallPciBarStore8		HvCallPci + 44
+#define HvCallPciBarStore16		HvCallPci + 45
+#define HvCallPciBarStore32		HvCallPci + 46
+#define HvCallPciBarStore64		HvCallPci + 47
+#define HvCallPciMaskInterrupts		HvCallPci + 48
+#define HvCallPciUnmaskInterrupts	HvCallPci + 49
+#define HvCallPciGetBusUnitInfo		HvCallPci + 50
+
+static inline u64 HvCallPci_configLoad16(u16 busNumber, u8 subBusNumber,
+		u8 deviceId, u32 offset, u16 *value)
+{
+	struct HvCallPci_DsaAddr dsa;
+	struct HvCallPci_LoadReturn retVal;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumber;
+	dsa.subBusNumber = subBusNumber;
+	dsa.deviceId = deviceId;
+
+	HvCall3Ret16(HvCallPciConfigLoad16, &retVal, *(u64 *)&dsa, offset, 0);
+
+	*value = retVal.value;
+
+	return retVal.rc;
+}
+
+static inline u64 HvCallPci_configStore8(u16 busNumber, u8 subBusNumber,
+		u8 deviceId, u32 offset, u8 value)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumber;
+	dsa.subBusNumber = subBusNumber;
+	dsa.deviceId = deviceId;
+
+	return HvCall4(HvCallPciConfigStore8, *(u64 *)&dsa, offset, value, 0);
+}
+
+static inline u64 HvCallPci_eoi(u16 busNumberParm, u8 subBusParm,
+		u8 deviceIdParm)
+{
+	struct HvCallPci_DsaAddr dsa;
+	struct HvCallPci_LoadReturn retVal;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceIdParm;
+
+	HvCall1Ret16(HvCallPciEoi, &retVal, *(u64*)&dsa);
+
+	return retVal.rc;
+}
+
+static inline u64 HvCallPci_getBarParms(u16 busNumberParm, u8 subBusParm,
+		u8 deviceIdParm, u8 barNumberParm, u64 parms, u32 sizeofParms)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceIdParm;
+	dsa.barNumber = barNumberParm;
+
+	return HvCall3(HvCallPciGetBarParms, *(u64*)&dsa, parms, sizeofParms);
+}
+
+static inline u64 HvCallPci_maskFisr(u16 busNumberParm, u8 subBusParm,
+		u8 deviceIdParm, u64 fisrMask)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceIdParm;
+
+	return HvCall2(HvCallPciMaskFisr, *(u64*)&dsa, fisrMask);
+}
+
+static inline u64 HvCallPci_unmaskFisr(u16 busNumberParm, u8 subBusParm,
+		u8 deviceIdParm, u64 fisrMask)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceIdParm;
+
+	return HvCall2(HvCallPciUnmaskFisr, *(u64*)&dsa, fisrMask);
+}
+
+static inline u64 HvCallPci_getDeviceInfo(u16 busNumberParm, u8 subBusParm,
+		u8 deviceNumberParm, u64 parms, u32 sizeofParms)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceNumberParm << 4;
+
+	return HvCall3(HvCallPciGetDeviceInfo, *(u64*)&dsa, parms, sizeofParms);
+}
+
+static inline u64 HvCallPci_maskInterrupts(u16 busNumberParm, u8 subBusParm,
+		u8 deviceIdParm, u64 interruptMask)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceIdParm;
+
+	return HvCall2(HvCallPciMaskInterrupts, *(u64*)&dsa, interruptMask);
+}
+
+static inline u64 HvCallPci_unmaskInterrupts(u16 busNumberParm, u8 subBusParm,
+		u8 deviceIdParm, u64 interruptMask)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceIdParm;
+
+	return HvCall2(HvCallPciUnmaskInterrupts, *(u64*)&dsa, interruptMask);
+}
+
+static inline u64 HvCallPci_getBusUnitInfo(u16 busNumberParm, u8 subBusParm,
+		u8 deviceIdParm, u64 parms, u32 sizeofParms)
+{
+	struct HvCallPci_DsaAddr dsa;
+
+	*((u64*)&dsa) = 0;
+
+	dsa.busNumber = busNumberParm;
+	dsa.subBusNumber = subBusParm;
+	dsa.deviceId = deviceIdParm;
+
+	return HvCall3(HvCallPciGetBusUnitInfo, *(u64*)&dsa, parms,
+			sizeofParms);
+}
+
+static inline int HvCallPci_getBusVpd(u16 busNumParm, u64 destParm,
+		u16 sizeParm)
+{
+	u64 xRc = HvCall4(HvCallPciGetCardVpd, busNumParm, destParm,
+			sizeParm, HvCallPci_BusVpd);
+	if (xRc == -1)
+		return -1;
+	else
+		return xRc & 0xFFFF;
+}
+
+#endif /* _PLATFORMS_ISERIES_CALL_PCI_H */
diff --git a/arch/powerpc/platforms/iseries/call_sm.h b/arch/powerpc/platforms/iseries/call_sm.h
new file mode 100644
index 000000000000..c7e251619f48
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/call_sm.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#ifndef _ISERIES_CALL_SM_H
+#define _ISERIES_CALL_SM_H
+
+/*
+ * This file contains the "hypervisor call" interface which is used to
+ * drive the hypervisor from the OS.
+ */
+
+#include <asm/iseries/hv_call_sc.h>
+#include <asm/iseries/hv_types.h>
+
+#define HvCallSmGet64BitsOfAccessMap	HvCallSm  + 11
+
+static inline u64 HvCallSm_get64BitsOfAccessMap(HvLpIndex lpIndex,
+		u64 indexIntoBitMap)
+{
+	return HvCall2(HvCallSmGet64BitsOfAccessMap, lpIndex, indexIntoBitMap);
+}
+
+#endif /* _ISERIES_CALL_SM_H */
diff --git a/arch/powerpc/platforms/iseries/htab.c b/arch/powerpc/platforms/iseries/htab.c
new file mode 100644
index 000000000000..b3c6c3374ca6
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/htab.c
@@ -0,0 +1,257 @@
+/*
+ * iSeries hashtable management.
+ *	Derived from pSeries_htab.c
+ *
+ * SMP scalability work:
+ *    Copyright (C) 2001 Anton Blanchard <anton@au.ibm.com>, IBM
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+#include <asm/machdep.h>
+#include <asm/pgtable.h>
+#include <asm/mmu.h>
+#include <asm/mmu_context.h>
+#include <asm/abs_addr.h>
+#include <linux/spinlock.h>
+
+#include "call_hpt.h"
+
+static spinlock_t iSeries_hlocks[64] __cacheline_aligned_in_smp =
+	{ [0 ... 63] = SPIN_LOCK_UNLOCKED};
+
+/*
+ * Very primitive algorithm for picking up a lock
+ */
+static inline void iSeries_hlock(unsigned long slot)
+{
+	if (slot & 0x8)
+		slot = ~slot;
+	spin_lock(&iSeries_hlocks[(slot >> 4) & 0x3f]);
+}
+
+static inline void iSeries_hunlock(unsigned long slot)
+{
+	if (slot & 0x8)
+		slot = ~slot;
+	spin_unlock(&iSeries_hlocks[(slot >> 4) & 0x3f]);
+}
+
+static long iSeries_hpte_insert(unsigned long hpte_group, unsigned long va,
+				unsigned long prpn, unsigned long vflags,
+				unsigned long rflags)
+{
+	unsigned long arpn;
+	long slot;
+	hpte_t lhpte;
+	int secondary = 0;
+
+	/*
+	 * The hypervisor tries both primary and secondary.
+	 * If we are being called to insert in the secondary,
+	 * it means we have already tried both primary and secondary,
+	 * so we return failure immediately.
+	 */
+	if (vflags & HPTE_V_SECONDARY)
+		return -1;
+
+	iSeries_hlock(hpte_group);
+
+	slot = HvCallHpt_findValid(&lhpte, va >> PAGE_SHIFT);
+	BUG_ON(lhpte.v & HPTE_V_VALID);
+
+	if (slot == -1)	{ /* No available entry found in either group */
+		iSeries_hunlock(hpte_group);
+		return -1;
+	}
+
+	if (slot < 0) {		/* MSB set means secondary group */
+		vflags |= HPTE_V_SECONDARY;
+		secondary = 1;
+		slot &= 0x7fffffffffffffff;
+	}
+
+	arpn = phys_to_abs(prpn << PAGE_SHIFT) >> PAGE_SHIFT;
+
+	lhpte.v = (va >> 23) << HPTE_V_AVPN_SHIFT | vflags | HPTE_V_VALID;
+	lhpte.r = (arpn << HPTE_R_RPN_SHIFT) | rflags;
+
+	/* Now fill in the actual HPTE */
+	HvCallHpt_addValidate(slot, secondary, &lhpte);
+
+	iSeries_hunlock(hpte_group);
+
+	return (secondary << 3) | (slot & 7);
+}
+
+long iSeries_hpte_bolt_or_insert(unsigned long hpte_group,
+		unsigned long va, unsigned long prpn, unsigned long vflags,
+		unsigned long rflags)
+{
+	long slot;
+	hpte_t lhpte;
+
+	slot = HvCallHpt_findValid(&lhpte, va >> PAGE_SHIFT);
+
+	if (lhpte.v & HPTE_V_VALID) {
+		/* Bolt the existing HPTE */
+		HvCallHpt_setSwBits(slot, 0x10, 0);
+		HvCallHpt_setPp(slot, PP_RWXX);
+		return 0;
+	}
+
+	return iSeries_hpte_insert(hpte_group, va, prpn, vflags, rflags);
+}
+
+static unsigned long iSeries_hpte_getword0(unsigned long slot)
+{
+	hpte_t hpte;
+
+	HvCallHpt_get(&hpte, slot);
+	return hpte.v;
+}
+
+static long iSeries_hpte_remove(unsigned long hpte_group)
+{
+	unsigned long slot_offset;
+	int i;
+	unsigned long hpte_v;
+
+	/* Pick a random slot to start at */
+	slot_offset = mftb() & 0x7;
+
+	iSeries_hlock(hpte_group);
+
+	for (i = 0; i < HPTES_PER_GROUP; i++) {
+		hpte_v = iSeries_hpte_getword0(hpte_group + slot_offset);
+
+		if (! (hpte_v & HPTE_V_BOLTED)) {
+			HvCallHpt_invalidateSetSwBitsGet(hpte_group +
+							 slot_offset, 0, 0);
+			iSeries_hunlock(hpte_group);
+			return i;
+		}
+
+		slot_offset++;
+		slot_offset &= 0x7;
+	}
+
+	iSeries_hunlock(hpte_group);
+
+	return -1;
+}
+
+/*
+ * The HyperVisor expects the "flags" argument in this form:
+ *	bits  0..59 : reserved
+ *	bit      60 : N
+ *	bits 61..63 : PP2,PP1,PP0
+ */
+static long iSeries_hpte_updatepp(unsigned long slot, unsigned long newpp,
+				  unsigned long va, int large, int local)
+{
+	hpte_t hpte;
+	unsigned long avpn = va >> 23;
+
+	iSeries_hlock(slot);
+
+	HvCallHpt_get(&hpte, slot);
+	if ((HPTE_V_AVPN_VAL(hpte.v) == avpn) && (hpte.v & HPTE_V_VALID)) {
+		/*
+		 * Hypervisor expects bits as NPPP, which is
+		 * different from how they are mapped in our PP.
+		 */
+		HvCallHpt_setPp(slot, (newpp & 0x3) | ((newpp & 0x4) << 1));
+		iSeries_hunlock(slot);
+		return 0;
+	}
+	iSeries_hunlock(slot);
+
+	return -1;
+}
+
+/*
+ * Functions used to find the PTE for a particular virtual address.
+ * Only used during boot when bolting pages.
+ *
+ * Input : vpn      : virtual page number
+ * Output: PTE index within the page table of the entry
+ *         -1 on failure
+ */
+static long iSeries_hpte_find(unsigned long vpn)
+{
+	hpte_t hpte;
+	long slot;
+
+	/*
+	 * The HvCallHpt_findValid interface is as follows:
+	 * 0xffffffffffffffff : No entry found.
+	 * 0x00000000xxxxxxxx : Entry found in primary group, slot x
+	 * 0x80000000xxxxxxxx : Entry found in secondary group, slot x
+	 */
+	slot = HvCallHpt_findValid(&hpte, vpn);
+	if (hpte.v & HPTE_V_VALID) {
+		if (slot < 0) {
+			slot &= 0x7fffffffffffffff;
+			slot = -slot;
+		}
+	} else
+		slot = -1;
+	return slot;
+}
+
+/*
+ * Update the page protection bits. Intended to be used to create
+ * guard pages for kernel data structures on pages which are bolted
+ * in the HPT. Assumes pages being operated on will not be stolen.
+ * Does not work on large pages.
+ *
+ * No need to lock here because we should be the only user.
+ */
+static void iSeries_hpte_updateboltedpp(unsigned long newpp, unsigned long ea)
+{
+	unsigned long vsid,va,vpn;
+	long slot;
+
+	vsid = get_kernel_vsid(ea);
+	va = (vsid << 28) | (ea & 0x0fffffff);
+	vpn = va >> PAGE_SHIFT;
+	slot = iSeries_hpte_find(vpn);
+	if (slot == -1)
+		panic("updateboltedpp: Could not find page to bolt\n");
+	HvCallHpt_setPp(slot, newpp);
+}
+
+static void iSeries_hpte_invalidate(unsigned long slot, unsigned long va,
+				    int large, int local)
+{
+	unsigned long hpte_v;
+	unsigned long avpn = va >> 23;
+	unsigned long flags;
+
+	local_irq_save(flags);
+
+	iSeries_hlock(slot);
+
+	hpte_v = iSeries_hpte_getword0(slot);
+
+	if ((HPTE_V_AVPN_VAL(hpte_v) == avpn) && (hpte_v & HPTE_V_VALID))
+		HvCallHpt_invalidateSetSwBitsGet(slot, 0, 0);
+
+	iSeries_hunlock(slot);
+
+	local_irq_restore(flags);
+}
+
+void hpte_init_iSeries(void)
+{
+	ppc_md.hpte_invalidate	= iSeries_hpte_invalidate;
+	ppc_md.hpte_updatepp	= iSeries_hpte_updatepp;
+	ppc_md.hpte_updateboltedpp = iSeries_hpte_updateboltedpp;
+	ppc_md.hpte_insert	= iSeries_hpte_insert;
+	ppc_md.hpte_remove	= iSeries_hpte_remove;
+
+	htab_finish_init();
+}
diff --git a/arch/powerpc/platforms/iseries/hvcall.S b/arch/powerpc/platforms/iseries/hvcall.S
new file mode 100644
index 000000000000..07ae6ad5f49f
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/hvcall.S
@@ -0,0 +1,94 @@
+/*
+ * This file contains the code to perform calls to the
+ * iSeries LPAR hypervisor
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <asm/ppc_asm.h>
+#include <asm/processor.h>
+#include <asm/ptrace.h>		/* XXX for STACK_FRAME_OVERHEAD */
+
+	.text
+
+/*
+ * Hypervisor call
+ *
+ * Invoke the iSeries hypervisor via the System Call instruction
+ * Parameters are passed to this routine in registers r3 - r10
+ *
+ * r3 contains the HV function to be called
+ * r4-r10 contain the operands to the hypervisor function
+ *
+ */
+
+_GLOBAL(HvCall)
+_GLOBAL(HvCall0)
+_GLOBAL(HvCall1)
+_GLOBAL(HvCall2)
+_GLOBAL(HvCall3)
+_GLOBAL(HvCall4)
+_GLOBAL(HvCall5)
+_GLOBAL(HvCall6)
+_GLOBAL(HvCall7)
+
+
+	mfcr	r0
+	std	r0,-8(r1)
+	stdu	r1,-(STACK_FRAME_OVERHEAD+16)(r1)
+
+	/* r0 = 0xffffffffffffffff indicates a hypervisor call */
+
+	li	r0,-1
+
+	/* Invoke the hypervisor */
+
+	sc
+
+	ld	r1,0(r1)
+	ld	r0,-8(r1)
+	mtcrf	0xff,r0
+
+	/*  return to caller, return value in r3 */
+
+	blr
+
+_GLOBAL(HvCall0Ret16)
+_GLOBAL(HvCall1Ret16)
+_GLOBAL(HvCall2Ret16)
+_GLOBAL(HvCall3Ret16)
+_GLOBAL(HvCall4Ret16)
+_GLOBAL(HvCall5Ret16)
+_GLOBAL(HvCall6Ret16)
+_GLOBAL(HvCall7Ret16)
+
+	mfcr	r0
+	std	r0,-8(r1)
+	std	r31,-16(r1)
+	stdu	r1,-(STACK_FRAME_OVERHEAD+32)(r1)
+
+	mr	r31,r4
+	li	r0,-1
+	mr	r4,r5
+	mr	r5,r6
+	mr	r6,r7
+	mr	r7,r8
+	mr	r8,r9
+	mr	r9,r10
+
+	sc
+
+	std	r3,0(r31)
+	std	r4,8(r31)
+
+	mr	r3,r5
+
+	ld	r1,0(r1)
+	ld	r0,-8(r1)
+	mtcrf	0xff,r0
+	ld	r31,-16(r1)
+
+	blr
diff --git a/arch/powerpc/platforms/iseries/hvlog.c b/arch/powerpc/platforms/iseries/hvlog.c
new file mode 100644
index 000000000000..62ec73479687
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/hvlog.c
@@ -0,0 +1,35 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ * 
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <asm/page.h>
+#include <asm/abs_addr.h>
+#include <asm/iseries/hv_call.h>
+#include <asm/iseries/hv_call_sc.h>
+#include <asm/iseries/hv_types.h>
+
+
+void HvCall_writeLogBuffer(const void *buffer, u64 len)
+{
+	struct HvLpBufferList hv_buf;
+	u64 left_this_page;
+	u64 cur = virt_to_abs(buffer);
+
+	while (len) {
+		hv_buf.addr = cur;
+		left_this_page = ((cur & PAGE_MASK) + PAGE_SIZE) - cur;
+		if (left_this_page > len)
+			left_this_page = len;
+		hv_buf.len = left_this_page;
+		len -= left_this_page;
+		HvCall2(HvCallBaseWriteLogBuffer,
+				virt_to_abs(&hv_buf),
+				left_this_page);
+		cur = (cur & PAGE_MASK) + PAGE_SIZE;
+	}
+}
diff --git a/arch/powerpc/platforms/iseries/hvlpconfig.c b/arch/powerpc/platforms/iseries/hvlpconfig.c
new file mode 100644
index 000000000000..663a1affb4bb
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/hvlpconfig.c
@@ -0,0 +1,26 @@
+/*
+ * Copyright (C) 2001  Kyle A. Lucke, IBM Corporation
+ * 
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ * 
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ * 
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/module.h>
+#include <asm/iseries/hv_lp_config.h>
+
+HvLpIndex HvLpConfig_getLpIndex_outline(void)
+{
+	return HvLpConfig_getLpIndex();
+}
+EXPORT_SYMBOL(HvLpConfig_getLpIndex_outline);
diff --git a/arch/powerpc/platforms/iseries/iommu.c b/arch/powerpc/platforms/iseries/iommu.c
new file mode 100644
index 000000000000..1a6845b5c5a4
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/iommu.c
@@ -0,0 +1,178 @@
+/*
+ * Copyright (C) 2001 Mike Corrigan & Dave Engebretsen, IBM Corporation
+ *
+ * Rewrite, cleanup:
+ *
+ * Copyright (C) 2004 Olof Johansson <olof@austin.ibm.com>, IBM Corporation
+ *
+ * Dynamic DMA mapping support, iSeries-specific parts.
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/types.h>
+#include <linux/dma-mapping.h>
+#include <linux/list.h>
+
+#include <asm/iommu.h>
+#include <asm/tce.h>
+#include <asm/machdep.h>
+#include <asm/abs_addr.h>
+#include <asm/pci-bridge.h>
+#include <asm/iseries/hv_call_xm.h>
+
+extern struct list_head iSeries_Global_Device_List;
+
+
+static void tce_build_iSeries(struct iommu_table *tbl, long index, long npages,
+		unsigned long uaddr, enum dma_data_direction direction)
+{
+	u64 rc;
+	union tce_entry tce;
+
+	while (npages--) {
+		tce.te_word = 0;
+		tce.te_bits.tb_rpn = virt_to_abs(uaddr) >> PAGE_SHIFT;
+
+		if (tbl->it_type == TCE_VB) {
+			/* Virtual Bus */
+			tce.te_bits.tb_valid = 1;
+			tce.te_bits.tb_allio = 1;
+			if (direction != DMA_TO_DEVICE)
+				tce.te_bits.tb_rdwr = 1;
+		} else {
+			/* PCI Bus */
+			tce.te_bits.tb_rdwr = 1; /* Read allowed */
+			if (direction != DMA_TO_DEVICE)
+				tce.te_bits.tb_pciwr = 1;
+		}
+
+		rc = HvCallXm_setTce((u64)tbl->it_index, (u64)index,
+				tce.te_word);
+		if (rc)
+			panic("PCI_DMA: HvCallXm_setTce failed, Rc: 0x%lx\n",
+					rc);
+		index++;
+		uaddr += PAGE_SIZE;
+	}
+}
+
+static void tce_free_iSeries(struct iommu_table *tbl, long index, long npages)
+{
+	u64 rc;
+
+	while (npages--) {
+		rc = HvCallXm_setTce((u64)tbl->it_index, (u64)index, 0);
+		if (rc)
+			panic("PCI_DMA: HvCallXm_setTce failed, Rc: 0x%lx\n",
+					rc);
+		index++;
+	}
+}
+
+#ifdef CONFIG_PCI
+/*
+ * This function compares the known tables to find an iommu_table
+ * that has already been built for hardware TCEs.
+ */
+static struct iommu_table *iommu_table_find(struct iommu_table * tbl)
+{
+	struct pci_dn *pdn;
+
+	list_for_each_entry(pdn, &iSeries_Global_Device_List, Device_List) {
+		struct iommu_table *it = pdn->iommu_table;
+		if ((it != NULL) &&
+		    (it->it_type == TCE_PCI) &&
+		    (it->it_offset == tbl->it_offset) &&
+		    (it->it_index == tbl->it_index) &&
+		    (it->it_size == tbl->it_size))
+			return it;
+	}
+	return NULL;
+}
+
+/*
+ * Call Hv with the architected data structure to get TCE table info.
+ * info. Put the returned data into the Linux representation of the
+ * TCE table data.
+ * The Hardware Tce table comes in three flavors.
+ * 1. TCE table shared between Buses.
+ * 2. TCE table per Bus.
+ * 3. TCE Table per IOA.
+ */
+static void iommu_table_getparms(struct pci_dn *pdn,
+				 struct iommu_table* tbl)
+{
+	struct iommu_table_cb *parms;
+
+	parms = kmalloc(sizeof(*parms), GFP_KERNEL);
+	if (parms == NULL)
+		panic("PCI_DMA: TCE Table Allocation failed.");
+
+	memset(parms, 0, sizeof(*parms));
+
+	parms->itc_busno = pdn->busno;
+	parms->itc_slotno = pdn->LogicalSlot;
+	parms->itc_virtbus = 0;
+
+	HvCallXm_getTceTableParms(iseries_hv_addr(parms));
+
+	if (parms->itc_size == 0)
+		panic("PCI_DMA: parms->size is zero, parms is 0x%p", parms);
+
+	/* itc_size is in pages worth of table, it_size is in # of entries */
+	tbl->it_size = (parms->itc_size * PAGE_SIZE) / sizeof(union tce_entry);
+	tbl->it_busno = parms->itc_busno;
+	tbl->it_offset = parms->itc_offset;
+	tbl->it_index = parms->itc_index;
+	tbl->it_blocksize = 1;
+	tbl->it_type = TCE_PCI;
+
+	kfree(parms);
+}
+
+
+void iommu_devnode_init_iSeries(struct device_node *dn)
+{
+	struct iommu_table *tbl;
+	struct pci_dn *pdn = PCI_DN(dn);
+
+	tbl = kmalloc(sizeof(struct iommu_table), GFP_KERNEL);
+
+	iommu_table_getparms(pdn, tbl);
+
+	/* Look for existing tce table */
+	pdn->iommu_table = iommu_table_find(tbl);
+	if (pdn->iommu_table == NULL)
+		pdn->iommu_table = iommu_init_table(tbl);
+	else
+		kfree(tbl);
+}
+#endif
+
+static void iommu_dev_setup_iSeries(struct pci_dev *dev) { }
+static void iommu_bus_setup_iSeries(struct pci_bus *bus) { }
+
+void iommu_init_early_iSeries(void)
+{
+	ppc_md.tce_build = tce_build_iSeries;
+	ppc_md.tce_free  = tce_free_iSeries;
+
+	ppc_md.iommu_dev_setup = iommu_dev_setup_iSeries;
+	ppc_md.iommu_bus_setup = iommu_bus_setup_iSeries;
+
+	pci_iommu_init();
+}
diff --git a/arch/powerpc/platforms/iseries/ipl_parms.h b/arch/powerpc/platforms/iseries/ipl_parms.h
new file mode 100644
index 000000000000..77c135ddbf1b
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/ipl_parms.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#ifndef _ISERIES_IPL_PARMS_H
+#define _ISERIES_IPL_PARMS_H
+
+/*
+ *	This struct maps the IPL Parameters DMA'd from the SP.
+ *
+ * Warning:
+ *	This data must map in exactly 64 bytes and match the architecture for
+ *	the IPL parms
+ */
+
+#include <asm/types.h>
+
+struct ItIplParmsReal {
+	u8	xFormat;		// Defines format of IplParms	x00-x00
+	u8	xRsvd01:6;		// Reserved			x01-x01
+	u8	xAlternateSearch:1;	// Alternate search indicator	...
+	u8	xUaSupplied:1;		// UA Supplied on programmed IPL...
+	u8	xLsUaFormat;		// Format byte for UA		x02-x02
+	u8	xRsvd02;		// Reserved			x03-x03
+	u32	xLsUa;			// LS UA			x04-x07
+	u32	xUnusedLsLid;		// First OS LID to load		x08-x0B
+	u16	xLsBusNumber;		// LS Bus Number		x0C-x0D
+	u8	xLsCardAdr;		// LS Card Address		x0E-x0E
+	u8	xLsBoardAdr;		// LS Board Address		x0F-x0F
+	u32	xRsvd03;		// Reserved			x10-x13
+	u8	xSpcnPresent:1;		// SPCN present			x14-x14
+	u8	xCpmPresent:1;		// CPM present			...
+	u8	xRsvd04:6;		// Reserved			...
+	u8	xRsvd05:4;		// Reserved			x15-x15
+	u8	xKeyLock:4;		// Keylock setting		...
+	u8	xRsvd06:6;		// Reserved			x16-x16
+	u8	xIplMode:2;		// Ipl mode (A|B|C|D)		...
+	u8	xHwIplType;		// Fast v slow v slow EC HW IPL	x17-x17
+	u16	xCpmEnabledIpl:1;	// CPM in effect when IPL initiatedx18-x19
+	u16	xPowerOnResetIpl:1;	// Indicate POR condition	...
+	u16	xMainStorePreserved:1;	// Main Storage is preserved	...
+	u16	xRsvd07:13;		// Reserved			...
+	u16	xIplSource:16;		// Ipl source			x1A-x1B
+	u8	xIplReason:8;		// Reason for this IPL		x1C-x1C
+	u8	xRsvd08;		// Reserved			x1D-x1D
+	u16	xRsvd09;		// Reserved			x1E-x1F
+	u16	xSysBoxType;		// System Box Type		x20-x21
+	u16	xSysProcType;		// System Processor Type	x22-x23
+	u32	xRsvd10;		// Reserved			x24-x27
+	u64	xRsvd11;		// Reserved			x28-x2F
+	u64	xRsvd12;		// Reserved			x30-x37
+	u64	xRsvd13;		// Reserved			x38-x3F
+};
+
+extern struct ItIplParmsReal	xItIplParmsReal;
+
+#endif /* _ISERIES_IPL_PARMS_H */
diff --git a/arch/powerpc/platforms/iseries/irq.c b/arch/powerpc/platforms/iseries/irq.c
new file mode 100644
index 000000000000..c1135912cc05
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/irq.c
@@ -0,0 +1,366 @@
+/*
+ * This module supports the iSeries PCI bus interrupt handling
+ * Copyright (C) 20yy  <Robert L Holtorf> <IBM Corp>
+ * Copyright (C) 2004-2005 IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the:
+ * Free Software Foundation, Inc.,
+ * 59 Temple Place, Suite 330,
+ * Boston, MA  02111-1307  USA
+ *
+ * Change Activity:
+ *   Created, December 13, 2000 by Wayne Holm
+ * End Change Activity
+ */
+#include <linux/config.h>
+#include <linux/pci.h>
+#include <linux/init.h>
+#include <linux/threads.h>
+#include <linux/smp.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/bootmem.h>
+#include <linux/ide.h>
+#include <linux/irq.h>
+#include <linux/spinlock.h>
+
+#include <asm/ppcdebug.h>
+#include <asm/iseries/hv_types.h>
+#include <asm/iseries/hv_lp_event.h>
+#include <asm/iseries/hv_call_xm.h>
+
+#include "irq.h"
+#include "call_pci.h"
+
+/* This maps virtual irq numbers to real irqs */
+unsigned int virt_irq_to_real_map[NR_IRQS];
+
+/* The next available virtual irq number */
+/* Note: the pcnet32 driver assumes irq numbers < 2 aren't valid. :( */
+static int next_virtual_irq = 2;
+
+static long Pci_Interrupt_Count;
+static long Pci_Event_Count;
+
+enum XmPciLpEvent_Subtype {
+	XmPciLpEvent_BusCreated		= 0,	// PHB has been created
+	XmPciLpEvent_BusError		= 1,	// PHB has failed
+	XmPciLpEvent_BusFailed		= 2,	// Msg to Secondary, Primary failed bus
+	XmPciLpEvent_NodeFailed		= 4,	// Multi-adapter bridge has failed
+	XmPciLpEvent_NodeRecovered	= 5,	// Multi-adapter bridge has recovered
+	XmPciLpEvent_BusRecovered	= 12,	// PHB has been recovered
+	XmPciLpEvent_UnQuiesceBus	= 18,	// Secondary bus unqiescing
+	XmPciLpEvent_BridgeError	= 21,	// Bridge Error
+	XmPciLpEvent_SlotInterrupt	= 22	// Slot interrupt
+};
+
+struct XmPciLpEvent_BusInterrupt {
+	HvBusNumber	busNumber;
+	HvSubBusNumber	subBusNumber;
+};
+
+struct XmPciLpEvent_NodeInterrupt {
+	HvBusNumber	busNumber;
+	HvSubBusNumber	subBusNumber;
+	HvAgentId	deviceId;
+};
+
+struct XmPciLpEvent {
+	struct HvLpEvent hvLpEvent;
+
+	union {
+		u64 alignData;			// Align on an 8-byte boundary
+
+		struct {
+			u32		fisr;
+			HvBusNumber	busNumber;
+			HvSubBusNumber	subBusNumber;
+			HvAgentId	deviceId;
+		} slotInterrupt;
+
+		struct XmPciLpEvent_BusInterrupt busFailed;
+		struct XmPciLpEvent_BusInterrupt busRecovered;
+		struct XmPciLpEvent_BusInterrupt busCreated;
+
+		struct XmPciLpEvent_NodeInterrupt nodeFailed;
+		struct XmPciLpEvent_NodeInterrupt nodeRecovered;
+
+	} eventData;
+
+};
+
+static void intReceived(struct XmPciLpEvent *eventParm,
+		struct pt_regs *regsParm)
+{
+	int irq;
+
+	++Pci_Interrupt_Count;
+
+	switch (eventParm->hvLpEvent.xSubtype) {
+	case XmPciLpEvent_SlotInterrupt:
+		irq = eventParm->hvLpEvent.xCorrelationToken;
+		/* Dispatch the interrupt handlers for this irq */
+		ppc_irq_dispatch_handler(regsParm, irq);
+		HvCallPci_eoi(eventParm->eventData.slotInterrupt.busNumber,
+			eventParm->eventData.slotInterrupt.subBusNumber,
+			eventParm->eventData.slotInterrupt.deviceId);
+		break;
+		/* Ignore error recovery events for now */
+	case XmPciLpEvent_BusCreated:
+		printk(KERN_INFO "intReceived: system bus %d created\n",
+			eventParm->eventData.busCreated.busNumber);
+		break;
+	case XmPciLpEvent_BusError:
+	case XmPciLpEvent_BusFailed:
+		printk(KERN_INFO "intReceived: system bus %d failed\n",
+			eventParm->eventData.busFailed.busNumber);
+		break;
+	case XmPciLpEvent_BusRecovered:
+	case XmPciLpEvent_UnQuiesceBus:
+		printk(KERN_INFO "intReceived: system bus %d recovered\n",
+			eventParm->eventData.busRecovered.busNumber);
+		break;
+	case XmPciLpEvent_NodeFailed:
+	case XmPciLpEvent_BridgeError:
+		printk(KERN_INFO
+			"intReceived: multi-adapter bridge %d/%d/%d failed\n",
+			eventParm->eventData.nodeFailed.busNumber,
+			eventParm->eventData.nodeFailed.subBusNumber,
+			eventParm->eventData.nodeFailed.deviceId);
+		break;
+	case XmPciLpEvent_NodeRecovered:
+		printk(KERN_INFO
+			"intReceived: multi-adapter bridge %d/%d/%d recovered\n",
+			eventParm->eventData.nodeRecovered.busNumber,
+			eventParm->eventData.nodeRecovered.subBusNumber,
+			eventParm->eventData.nodeRecovered.deviceId);
+		break;
+	default:
+		printk(KERN_ERR
+			"intReceived: unrecognized event subtype 0x%x\n",
+			eventParm->hvLpEvent.xSubtype);
+		break;
+	}
+}
+
+static void XmPciLpEvent_handler(struct HvLpEvent *eventParm,
+		struct pt_regs *regsParm)
+{
+#ifdef CONFIG_PCI
+	++Pci_Event_Count;
+
+	if (eventParm && (eventParm->xType == HvLpEvent_Type_PciIo)) {
+		switch (eventParm->xFlags.xFunction) {
+		case HvLpEvent_Function_Int:
+			intReceived((struct XmPciLpEvent *)eventParm, regsParm);
+			break;
+		case HvLpEvent_Function_Ack:
+			printk(KERN_ERR
+				"XmPciLpEvent_handler: unexpected ack received\n");
+			break;
+		default:
+			printk(KERN_ERR
+				"XmPciLpEvent_handler: unexpected event function %d\n",
+				(int)eventParm->xFlags.xFunction);
+			break;
+		}
+	} else if (eventParm)
+		printk(KERN_ERR
+			"XmPciLpEvent_handler: Unrecognized PCI event type 0x%x\n",
+			(int)eventParm->xType);
+	else
+		printk(KERN_ERR "XmPciLpEvent_handler: NULL event received\n");
+#endif
+}
+
+/*
+ * This is called by init_IRQ.  set in ppc_md.init_IRQ by iSeries_setup.c
+ * It must be called before the bus walk.
+ */
+void __init iSeries_init_IRQ(void)
+{
+	/* Register PCI event handler and open an event path */
+	int xRc;
+
+	xRc = HvLpEvent_registerHandler(HvLpEvent_Type_PciIo,
+			&XmPciLpEvent_handler);
+	if (xRc == 0) {
+		xRc = HvLpEvent_openPath(HvLpEvent_Type_PciIo, 0);
+		if (xRc != 0)
+			printk(KERN_ERR "iSeries_init_IRQ: open event path "
+					"failed with rc 0x%x\n", xRc);
+	} else
+		printk(KERN_ERR "iSeries_init_IRQ: register handler "
+				"failed with rc 0x%x\n", xRc);
+}
+
+#define REAL_IRQ_TO_BUS(irq)	((((irq) >> 6) & 0xff) + 1)
+#define REAL_IRQ_TO_IDSEL(irq)	((((irq) >> 3) & 7) + 1)
+#define REAL_IRQ_TO_FUNC(irq)	((irq) & 7)
+
+/*
+ * This will be called by device drivers (via enable_IRQ)
+ * to enable INTA in the bridge interrupt status register.
+ */
+static void iSeries_enable_IRQ(unsigned int irq)
+{
+	u32 bus, deviceId, function, mask;
+	const u32 subBus = 0;
+	unsigned int rirq = virt_irq_to_real_map[irq];
+
+	/* The IRQ has already been locked by the caller */
+	bus = REAL_IRQ_TO_BUS(rirq);
+	function = REAL_IRQ_TO_FUNC(rirq);
+	deviceId = (REAL_IRQ_TO_IDSEL(rirq) << 4) + function;
+
+	/* Unmask secondary INTA */
+	mask = 0x80000000;
+	HvCallPci_unmaskInterrupts(bus, subBus, deviceId, mask);
+	PPCDBG(PPCDBG_BUSWALK, "iSeries_enable_IRQ 0x%02X.%02X.%02X 0x%04X\n",
+			bus, subBus, deviceId, irq);
+}
+
+/* This is called by iSeries_activate_IRQs */
+static unsigned int iSeries_startup_IRQ(unsigned int irq)
+{
+	u32 bus, deviceId, function, mask;
+	const u32 subBus = 0;
+	unsigned int rirq = virt_irq_to_real_map[irq];
+
+	bus = REAL_IRQ_TO_BUS(rirq);
+	function = REAL_IRQ_TO_FUNC(rirq);
+	deviceId = (REAL_IRQ_TO_IDSEL(rirq) << 4) + function;
+
+	/* Link the IRQ number to the bridge */
+	HvCallXm_connectBusUnit(bus, subBus, deviceId, irq);
+
+	/* Unmask bridge interrupts in the FISR */
+	mask = 0x01010000 << function;
+	HvCallPci_unmaskFisr(bus, subBus, deviceId, mask);
+	iSeries_enable_IRQ(irq);
+	return 0;
+}
+
+/*
+ * This is called out of iSeries_fixup to activate interrupt
+ * generation for usable slots
+ */
+void __init iSeries_activate_IRQs()
+{
+	int irq;
+	unsigned long flags;
+
+	for_each_irq (irq) {
+		irq_desc_t *desc = get_irq_desc(irq);
+
+		if (desc && desc->handler && desc->handler->startup) {
+			spin_lock_irqsave(&desc->lock, flags);
+			desc->handler->startup(irq);
+			spin_unlock_irqrestore(&desc->lock, flags);
+		}
+	}
+}
+
+/*  this is not called anywhere currently */
+static void iSeries_shutdown_IRQ(unsigned int irq)
+{
+	u32 bus, deviceId, function, mask;
+	const u32 subBus = 0;
+	unsigned int rirq = virt_irq_to_real_map[irq];
+
+	/* irq should be locked by the caller */
+	bus = REAL_IRQ_TO_BUS(rirq);
+	function = REAL_IRQ_TO_FUNC(rirq);
+	deviceId = (REAL_IRQ_TO_IDSEL(rirq) << 4) + function;
+
+	/* Invalidate the IRQ number in the bridge */
+	HvCallXm_connectBusUnit(bus, subBus, deviceId, 0);
+
+	/* Mask bridge interrupts in the FISR */
+	mask = 0x01010000 << function;
+	HvCallPci_maskFisr(bus, subBus, deviceId, mask);
+}
+
+/*
+ * This will be called by device drivers (via disable_IRQ)
+ * to disable INTA in the bridge interrupt status register.
+ */
+static void iSeries_disable_IRQ(unsigned int irq)
+{
+	u32 bus, deviceId, function, mask;
+	const u32 subBus = 0;
+	unsigned int rirq = virt_irq_to_real_map[irq];
+
+	/* The IRQ has already been locked by the caller */
+	bus = REAL_IRQ_TO_BUS(rirq);
+	function = REAL_IRQ_TO_FUNC(rirq);
+	deviceId = (REAL_IRQ_TO_IDSEL(rirq) << 4) + function;
+
+	/* Mask secondary INTA   */
+	mask = 0x80000000;
+	HvCallPci_maskInterrupts(bus, subBus, deviceId, mask);
+	PPCDBG(PPCDBG_BUSWALK, "iSeries_disable_IRQ 0x%02X.%02X.%02X 0x%04X\n",
+			bus, subBus, deviceId, irq);
+}
+
+/*
+ * Need to define this so ppc_irq_dispatch_handler will NOT call
+ * enable_IRQ at the end of interrupt handling.  However, this does
+ * nothing because there is not enough information provided to do
+ * the EOI HvCall.  This is done by XmPciLpEvent.c
+ */
+static void iSeries_end_IRQ(unsigned int irq)
+{
+}
+
+static hw_irq_controller iSeries_IRQ_handler = {
+	.typename = "iSeries irq controller",
+	.startup = iSeries_startup_IRQ,
+	.shutdown = iSeries_shutdown_IRQ,
+	.enable = iSeries_enable_IRQ,
+	.disable = iSeries_disable_IRQ,
+	.end = iSeries_end_IRQ
+};
+
+/*
+ * This is called out of iSeries_scan_slot to allocate an IRQ for an EADS slot
+ * It calculates the irq value for the slot.
+ * Note that subBusNumber is always 0 (at the moment at least).
+ */
+int __init iSeries_allocate_IRQ(HvBusNumber busNumber,
+		HvSubBusNumber subBusNumber, HvAgentId deviceId)
+{
+	unsigned int realirq, virtirq;
+	u8 idsel = (deviceId >> 4);
+	u8 function = deviceId & 7;
+
+	virtirq = next_virtual_irq++;
+	realirq = ((busNumber - 1) << 6) + ((idsel - 1) << 3) + function;
+	virt_irq_to_real_map[virtirq] = realirq;
+
+	irq_desc[virtirq].handler = &iSeries_IRQ_handler;
+	return virtirq;
+}
+
+int virt_irq_create_mapping(unsigned int real_irq)
+{
+	BUG(); /* Don't call this on iSeries, yet */
+
+	return 0;
+}
+
+void virt_irq_init(void)
+{
+	return;
+}
diff --git a/arch/powerpc/platforms/iseries/irq.h b/arch/powerpc/platforms/iseries/irq.h
new file mode 100644
index 000000000000..5f643f16ecc0
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/irq.h
@@ -0,0 +1,8 @@
+#ifndef	_ISERIES_IRQ_H
+#define	_ISERIES_IRQ_H
+
+extern void iSeries_init_IRQ(void);
+extern int  iSeries_allocate_IRQ(HvBusNumber, HvSubBusNumber, HvAgentId);
+extern void iSeries_activate_IRQs(void);
+
+#endif /* _ISERIES_IRQ_H */
diff --git a/arch/powerpc/platforms/iseries/ksyms.c b/arch/powerpc/platforms/iseries/ksyms.c
new file mode 100644
index 000000000000..a2200842f4e5
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/ksyms.c
@@ -0,0 +1,27 @@
+/*
+ * (C) 2001-2005 PPC 64 Team, IBM Corp
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+#include <linux/module.h>
+
+#include <asm/hw_irq.h>
+#include <asm/iseries/hv_call_sc.h>
+
+EXPORT_SYMBOL(HvCall0);
+EXPORT_SYMBOL(HvCall1);
+EXPORT_SYMBOL(HvCall2);
+EXPORT_SYMBOL(HvCall3);
+EXPORT_SYMBOL(HvCall4);
+EXPORT_SYMBOL(HvCall5);
+EXPORT_SYMBOL(HvCall6);
+EXPORT_SYMBOL(HvCall7);
+
+#ifdef CONFIG_SMP
+EXPORT_SYMBOL(local_get_flags);
+EXPORT_SYMBOL(local_irq_disable);
+EXPORT_SYMBOL(local_irq_restore);
+#endif
diff --git a/arch/powerpc/platforms/iseries/lpardata.c b/arch/powerpc/platforms/iseries/lpardata.c
new file mode 100644
index 000000000000..bb8c91537f35
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/lpardata.c
@@ -0,0 +1,227 @@
+/*
+ * Copyright 2001 Mike Corrigan, IBM Corp
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+#include <linux/config.h>
+#include <linux/types.h>
+#include <linux/threads.h>
+#include <linux/module.h>
+#include <linux/bitops.h>
+#include <asm/processor.h>
+#include <asm/ptrace.h>
+#include <asm/abs_addr.h>
+#include <asm/iseries/it_lp_naca.h>
+#include <asm/lppaca.h>
+#include <asm/iseries/it_lp_reg_save.h>
+#include <asm/paca.h>
+#include <asm/iseries/lpar_map.h>
+#include <asm/iseries/it_exp_vpd_panel.h>
+#include <asm/iseries/it_lp_queue.h>
+
+#include "naca.h"
+#include "vpd_areas.h"
+#include "spcomm_area.h"
+#include "ipl_parms.h"
+#include "processor_vpd.h"
+#include "release_data.h"
+
+/* The HvReleaseData is the root of the information shared between
+ * the hypervisor and Linux.
+ */
+struct HvReleaseData hvReleaseData = {
+	.xDesc = 0xc8a5d9c4,	/* "HvRD" ebcdic */
+	.xSize = sizeof(struct HvReleaseData),
+	.xVpdAreasPtrOffset = offsetof(struct naca_struct, xItVpdAreas),
+	.xSlicNacaAddr = &naca,		/* 64-bit Naca address */
+	.xMsNucDataOffset = LPARMAP_PHYS,
+	.xFlags = HVREL_TAGSINACTIVE	/* tags inactive       */
+					/* 64 bit              */
+					/* shared processors   */
+					/* HMT allowed         */
+		  | 6,			/* TEMP: This allows non-GA driver */
+	.xVrmIndex = 4,			/* We are v5r2m0               */
+	.xMinSupportedPlicVrmIndex = 3,		/* v5r1m0 */
+	.xMinCompatablePlicVrmIndex = 3,	/* v5r1m0 */
+	.xVrmName = { 0xd3, 0x89, 0x95, 0xa4,	/* "Linux 2.4.64" ebcdic */
+		0xa7, 0x40, 0xf2, 0x4b,
+		0xf4, 0x4b, 0xf6, 0xf4 },
+};
+
+/*
+ * The NACA.  The first dword of the naca is required by the iSeries
+ * hypervisor to point to itVpdAreas.  The hypervisor finds the NACA
+ * through the pointer in hvReleaseData.
+ */
+struct naca_struct naca = {
+	.xItVpdAreas = &itVpdAreas,
+	.xRamDisk = 0,
+	.xRamDiskSize = 0,
+};
+
+extern void system_reset_iSeries(void);
+extern void machine_check_iSeries(void);
+extern void data_access_iSeries(void);
+extern void instruction_access_iSeries(void);
+extern void hardware_interrupt_iSeries(void);
+extern void alignment_iSeries(void);
+extern void program_check_iSeries(void);
+extern void fp_unavailable_iSeries(void);
+extern void decrementer_iSeries(void);
+extern void trap_0a_iSeries(void);
+extern void trap_0b_iSeries(void);
+extern void system_call_iSeries(void);
+extern void single_step_iSeries(void);
+extern void trap_0e_iSeries(void);
+extern void performance_monitor_iSeries(void);
+extern void data_access_slb_iSeries(void);
+extern void instruction_access_slb_iSeries(void);
+
+struct ItLpNaca itLpNaca = {
+	.xDesc = 0xd397d581,		/* "LpNa" ebcdic */
+	.xSize = 0x0400,		/* size of ItLpNaca */
+	.xIntHdlrOffset = 0x0300,	/* offset to int array */
+	.xMaxIntHdlrEntries = 19,	/* # ents */
+	.xPrimaryLpIndex = 0,		/* Part # of primary */
+	.xServiceLpIndex = 0,		/* Part # of serv */
+	.xLpIndex = 0,			/* Part # of me */
+	.xMaxLpQueues = 0,		/* # of LP queues */
+	.xLpQueueOffset = 0x100,	/* offset of start of LP queues */
+	.xPirEnvironMode = 0,		/* Piranha stuff */
+	.xPirConsoleMode = 0,
+	.xPirDasdMode = 0,
+	.xLparInstalled = 0,
+	.xSysPartitioned = 0,
+	.xHwSyncedTBs = 0,
+	.xIntProcUtilHmt = 0,
+	.xSpVpdFormat = 0,
+	.xIntProcRatio = 0,
+	.xPlicVrmIndex = 0,		/* VRM index of PLIC */
+	.xMinSupportedSlicVrmInd = 0,	/* min supported SLIC */
+	.xMinCompatableSlicVrmInd = 0,	/* min compat SLIC */
+	.xLoadAreaAddr = 0,		/* 64-bit addr of load area */
+	.xLoadAreaChunks = 0,		/* chunks for load area */
+	.xPaseSysCallCRMask = 0,	/* PASE mask */
+	.xSlicSegmentTablePtr = 0,	/* seg table */
+	.xOldLpQueue = { 0 },		/* Old LP Queue */
+	.xInterruptHdlr = {
+		(u64)system_reset_iSeries,	/* 0x100 System Reset */
+		(u64)machine_check_iSeries,	/* 0x200 Machine Check */
+		(u64)data_access_iSeries,	/* 0x300 Data Access */
+		(u64)instruction_access_iSeries, /* 0x400 Instruction Access */
+		(u64)hardware_interrupt_iSeries, /* 0x500 External */
+		(u64)alignment_iSeries,		/* 0x600 Alignment */
+		(u64)program_check_iSeries,	/* 0x700 Program Check */
+		(u64)fp_unavailable_iSeries,	/* 0x800 FP Unavailable */
+		(u64)decrementer_iSeries,	/* 0x900 Decrementer */
+		(u64)trap_0a_iSeries,		/* 0xa00 Trap 0A */
+		(u64)trap_0b_iSeries,		/* 0xb00 Trap 0B */
+		(u64)system_call_iSeries,	/* 0xc00 System Call */
+		(u64)single_step_iSeries,	/* 0xd00 Single Step */
+		(u64)trap_0e_iSeries,		/* 0xe00 Trap 0E */
+		(u64)performance_monitor_iSeries,/* 0xf00 Performance Monitor */
+		0,				/* int 0x1000 */
+		0,				/* int 0x1010 */
+		0,				/* int 0x1020 CPU ctls */
+		(u64)hardware_interrupt_iSeries, /* SC Ret Hdlr */
+		(u64)data_access_slb_iSeries,	/* 0x380 D-SLB */
+		(u64)instruction_access_slb_iSeries /* 0x480 I-SLB */
+	}
+};
+EXPORT_SYMBOL(itLpNaca);
+
+/* May be filled in by the hypervisor so cannot end up in the BSS */
+struct ItIplParmsReal xItIplParmsReal __attribute__((__section__(".data")));
+
+/* May be filled in by the hypervisor so cannot end up in the BSS */
+struct ItExtVpdPanel xItExtVpdPanel __attribute__((__section__(".data")));
+EXPORT_SYMBOL(xItExtVpdPanel);
+
+#define maxPhysicalProcessors 32
+
+struct IoHriProcessorVpd xIoHriProcessorVpd[maxPhysicalProcessors] = {
+	{
+		.xInstCacheOperandSize = 32,
+		.xDataCacheOperandSize = 32,
+		.xProcFreq     = 50000000,
+		.xTimeBaseFreq = 50000000,
+		.xPVR = 0x3600
+	}
+};
+
+/* Space for Main Store Vpd 27,200 bytes */
+/* May be filled in by the hypervisor so cannot end up in the BSS */
+u64    xMsVpd[3400] __attribute__((__section__(".data")));
+
+/* Space for Recovery Log Buffer */
+/* May be filled in by the hypervisor so cannot end up in the BSS */
+u64    xRecoveryLogBuffer[32] __attribute__((__section__(".data")));
+
+struct SpCommArea xSpCommArea = {
+	.xDesc = 0xE2D7C3C2,
+	.xFormat = 1,
+};
+
+/* The LparMap data is now located at offset 0x6000 in head.S
+ * It was put there so that the HvReleaseData could address it
+ * with a 32-bit offset as required by the iSeries hypervisor
+ *
+ * The Naca has a pointer to the ItVpdAreas.  The hypervisor finds
+ * the Naca via the HvReleaseData area.  The HvReleaseData has the
+ * offset into the Naca of the pointer to the ItVpdAreas.
+ */
+struct ItVpdAreas itVpdAreas = {
+	.xSlicDesc = 0xc9a3e5c1,		/* "ItVA" */
+	.xSlicSize = sizeof(struct ItVpdAreas),
+	.xSlicVpdEntries = ItVpdMaxEntries,	/* # VPD array entries */
+	.xSlicDmaEntries = ItDmaMaxEntries,	/* # DMA array entries */
+	.xSlicMaxLogicalProcs = NR_CPUS * 2,	/* Max logical procs */
+	.xSlicMaxPhysicalProcs = maxPhysicalProcessors,	/* Max physical procs */
+	.xSlicDmaToksOffset = offsetof(struct ItVpdAreas, xPlicDmaToks),
+	.xSlicVpdAdrsOffset = offsetof(struct ItVpdAreas, xSlicVpdAdrs),
+	.xSlicDmaLensOffset = offsetof(struct ItVpdAreas, xPlicDmaLens),
+	.xSlicVpdLensOffset = offsetof(struct ItVpdAreas, xSlicVpdLens),
+	.xSlicMaxSlotLabels = 0,		/* max slot labels */
+	.xSlicMaxLpQueues = 1,			/* max LP queues */
+	.xPlicDmaLens = { 0 },			/* DMA lengths */
+	.xPlicDmaToks = { 0 },			/* DMA tokens */
+	.xSlicVpdLens = {			/* VPD lengths */
+	        0,0,0,		        /*  0 - 2 */
+		sizeof(xItExtVpdPanel), /*       3 Extended VPD   */
+		sizeof(struct paca_struct),	/*       4 length of Paca  */
+		0,			/*       5 */
+		sizeof(struct ItIplParmsReal),/* 6 length of IPL parms */
+		26992,			/*	 7 length of MS VPD */
+		0,			/*       8 */
+		sizeof(struct ItLpNaca),/*       9 length of LP Naca */
+		0,			/*	10 */
+		256,			/*	11 length of Recovery Log Buf */
+		sizeof(struct SpCommArea), /*   12 length of SP Comm Area */
+		0,0,0,			/* 13 - 15 */
+		sizeof(struct IoHriProcessorVpd),/* 16 length of Proc Vpd */
+		0,0,0,0,0,0,		/* 17 - 22  */
+		sizeof(struct hvlpevent_queue),	/* 23 length of Lp Queue */
+		0,0			/* 24 - 25 */
+		},
+	.xSlicVpdAdrs = {			/* VPD addresses */
+		0,0,0,			/*	 0 -  2 */
+		&xItExtVpdPanel,        /*       3 Extended VPD */
+		&paca[0],		/*       4 first Paca */
+		0,			/*       5 */
+		&xItIplParmsReal,	/*	 6 IPL parms */
+		&xMsVpd,		/*	 7 MS Vpd */
+		0,			/*       8 */
+		&itLpNaca,		/*       9 LpNaca */
+		0,			/*	10 */
+		&xRecoveryLogBuffer,	/*	11 Recovery Log Buffer */
+		&xSpCommArea,		/*	12 SP Comm Area */
+		0,0,0,			/* 13 - 15 */
+		&xIoHriProcessorVpd,	/*      16 Proc Vpd */
+		0,0,0,0,0,0,		/* 17 - 22 */
+		&hvlpevent_queue,	/*      23 Lp Queue */
+		0,0
+	}
+};
diff --git a/arch/powerpc/platforms/iseries/lpevents.c b/arch/powerpc/platforms/iseries/lpevents.c
new file mode 100644
index 000000000000..e9fb98bf895f
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/lpevents.c
@@ -0,0 +1,325 @@
+/*
+ * Copyright (C) 2001 Mike Corrigan  IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/stddef.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/bootmem.h>
+#include <linux/seq_file.h>
+#include <linux/proc_fs.h>
+#include <linux/module.h>
+
+#include <asm/system.h>
+#include <asm/paca.h>
+#include <asm/iseries/it_lp_queue.h>
+#include <asm/iseries/hv_lp_event.h>
+#include <asm/iseries/hv_call_event.h>
+#include <asm/iseries/it_lp_naca.h>
+
+/*
+ * The LpQueue is used to pass event data from the hypervisor to
+ * the partition.  This is where I/O interrupt events are communicated.
+ *
+ * It is written to by the hypervisor so cannot end up in the BSS.
+ */
+struct hvlpevent_queue hvlpevent_queue __attribute__((__section__(".data")));
+
+DEFINE_PER_CPU(unsigned long[HvLpEvent_Type_NumTypes], hvlpevent_counts);
+
+static char *event_types[HvLpEvent_Type_NumTypes] = {
+	"Hypervisor",
+	"Machine Facilities",
+	"Session Manager",
+	"SPD I/O",
+	"Virtual Bus",
+	"PCI I/O",
+	"RIO I/O",
+	"Virtual Lan",
+	"Virtual I/O"
+};
+
+/* Array of LpEvent handler functions */
+static LpEventHandler lpEventHandler[HvLpEvent_Type_NumTypes];
+static unsigned lpEventHandlerPaths[HvLpEvent_Type_NumTypes];
+
+static struct HvLpEvent * get_next_hvlpevent(void)
+{
+	struct HvLpEvent * event;
+	event = (struct HvLpEvent *)hvlpevent_queue.xSlicCurEventPtr;
+
+	if (event->xFlags.xValid) {
+		/* rmb() needed only for weakly consistent machines (regatta) */
+		rmb();
+		/* Set pointer to next potential event */
+		hvlpevent_queue.xSlicCurEventPtr += ((event->xSizeMinus1 +
+				LpEventAlign) / LpEventAlign) * LpEventAlign;
+
+		/* Wrap to beginning if no room at end */
+		if (hvlpevent_queue.xSlicCurEventPtr >
+				hvlpevent_queue.xSlicLastValidEventPtr) {
+			hvlpevent_queue.xSlicCurEventPtr =
+				hvlpevent_queue.xSlicEventStackPtr;
+		}
+	} else {
+		event = NULL;
+	}
+
+	return event;
+}
+
+static unsigned long spread_lpevents = NR_CPUS;
+
+int hvlpevent_is_pending(void)
+{
+	struct HvLpEvent *next_event;
+
+	if (smp_processor_id() >= spread_lpevents)
+		return 0;
+
+	next_event = (struct HvLpEvent *)hvlpevent_queue.xSlicCurEventPtr;
+
+	return next_event->xFlags.xValid |
+		hvlpevent_queue.xPlicOverflowIntPending;
+}
+
+static void hvlpevent_clear_valid(struct HvLpEvent * event)
+{
+	/* Tell the Hypervisor that we're done with this event.
+	 * Also clear bits within this event that might look like valid bits.
+	 * ie. on 64-byte boundaries.
+	 */
+	struct HvLpEvent *tmp;
+	unsigned extra = ((event->xSizeMinus1 + LpEventAlign) /
+						 LpEventAlign) - 1;
+
+	switch (extra) {
+	case 3:
+		tmp = (struct HvLpEvent*)((char*)event + 3 * LpEventAlign);
+		tmp->xFlags.xValid = 0;
+	case 2:
+		tmp = (struct HvLpEvent*)((char*)event + 2 * LpEventAlign);
+		tmp->xFlags.xValid = 0;
+	case 1:
+		tmp = (struct HvLpEvent*)((char*)event + 1 * LpEventAlign);
+		tmp->xFlags.xValid = 0;
+	}
+
+	mb();
+
+	event->xFlags.xValid = 0;
+}
+
+void process_hvlpevents(struct pt_regs *regs)
+{
+	struct HvLpEvent * event;
+
+	/* If we have recursed, just return */
+	if (!spin_trylock(&hvlpevent_queue.lock))
+		return;
+
+	for (;;) {
+		event = get_next_hvlpevent();
+		if (event) {
+			/* Call appropriate handler here, passing
+			 * a pointer to the LpEvent.  The handler
+			 * must make a copy of the LpEvent if it
+			 * needs it in a bottom half. (perhaps for
+			 * an ACK)
+			 *
+			 *  Handlers are responsible for ACK processing
+			 *
+			 * The Hypervisor guarantees that LpEvents will
+			 * only be delivered with types that we have
+			 * registered for, so no type check is necessary
+			 * here!
+			 */
+			if (event->xType < HvLpEvent_Type_NumTypes)
+				__get_cpu_var(hvlpevent_counts)[event->xType]++;
+			if (event->xType < HvLpEvent_Type_NumTypes &&
+					lpEventHandler[event->xType])
+				lpEventHandler[event->xType](event, regs);
+			else
+				printk(KERN_INFO "Unexpected Lp Event type=%d\n", event->xType );
+
+			hvlpevent_clear_valid(event);
+		} else if (hvlpevent_queue.xPlicOverflowIntPending)
+			/*
+			 * No more valid events. If overflow events are
+			 * pending process them
+			 */
+			HvCallEvent_getOverflowLpEvents(hvlpevent_queue.xIndex);
+		else
+			break;
+	}
+
+	spin_unlock(&hvlpevent_queue.lock);
+}
+
+static int set_spread_lpevents(char *str)
+{
+	unsigned long val = simple_strtoul(str, NULL, 0);
+
+	/*
+	 * The parameter is the number of processors to share in processing
+	 * lp events.
+	 */
+	if (( val > 0) && (val <= NR_CPUS)) {
+		spread_lpevents = val;
+		printk("lpevent processing spread over %ld processors\n", val);
+	} else {
+		printk("invalid spread_lpevents %ld\n", val);
+	}
+
+	return 1;
+}
+__setup("spread_lpevents=", set_spread_lpevents);
+
+void setup_hvlpevent_queue(void)
+{
+	void *eventStack;
+
+	/* Allocate a page for the Event Stack. */
+	eventStack = alloc_bootmem_pages(LpEventStackSize);
+	memset(eventStack, 0, LpEventStackSize);
+
+	/* Invoke the hypervisor to initialize the event stack */
+	HvCallEvent_setLpEventStack(0, eventStack, LpEventStackSize);
+
+	hvlpevent_queue.xSlicEventStackPtr = (char *)eventStack;
+	hvlpevent_queue.xSlicCurEventPtr = (char *)eventStack;
+	hvlpevent_queue.xSlicLastValidEventPtr = (char *)eventStack +
+					(LpEventStackSize - LpEventMaxSize);
+	hvlpevent_queue.xIndex = 0;
+}
+
+/* Register a handler for an LpEvent type */
+int HvLpEvent_registerHandler(HvLpEvent_Type eventType, LpEventHandler handler)
+{
+	if (eventType < HvLpEvent_Type_NumTypes) {
+		lpEventHandler[eventType] = handler;
+		return 0;
+	}
+	return 1;
+}
+EXPORT_SYMBOL(HvLpEvent_registerHandler);
+
+int HvLpEvent_unregisterHandler(HvLpEvent_Type eventType)
+{
+	might_sleep();
+
+	if (eventType < HvLpEvent_Type_NumTypes) {
+		if (!lpEventHandlerPaths[eventType]) {
+			lpEventHandler[eventType] = NULL;
+			/*
+			 * We now sleep until all other CPUs have scheduled.
+			 * This ensures that the deletion is seen by all
+			 * other CPUs, and that the deleted handler isn't
+			 * still running on another CPU when we return.
+			 */
+			synchronize_rcu();
+			return 0;
+		}
+	}
+	return 1;
+}
+EXPORT_SYMBOL(HvLpEvent_unregisterHandler);
+
+/*
+ * lpIndex is the partition index of the target partition.
+ * needed only for VirtualIo, VirtualLan and SessionMgr.  Zero
+ * indicates to use our partition index - for the other types.
+ */
+int HvLpEvent_openPath(HvLpEvent_Type eventType, HvLpIndex lpIndex)
+{
+	if ((eventType < HvLpEvent_Type_NumTypes) &&
+			lpEventHandler[eventType]) {
+		if (lpIndex == 0)
+			lpIndex = itLpNaca.xLpIndex;
+		HvCallEvent_openLpEventPath(lpIndex, eventType);
+		++lpEventHandlerPaths[eventType];
+		return 0;
+	}
+	return 1;
+}
+
+int HvLpEvent_closePath(HvLpEvent_Type eventType, HvLpIndex lpIndex)
+{
+	if ((eventType < HvLpEvent_Type_NumTypes) &&
+			lpEventHandler[eventType] &&
+			lpEventHandlerPaths[eventType]) {
+		if (lpIndex == 0)
+			lpIndex = itLpNaca.xLpIndex;
+		HvCallEvent_closeLpEventPath(lpIndex, eventType);
+		--lpEventHandlerPaths[eventType];
+		return 0;
+	}
+	return 1;
+}
+
+static int proc_lpevents_show(struct seq_file *m, void *v)
+{
+	int cpu, i;
+	unsigned long sum;
+	static unsigned long cpu_totals[NR_CPUS];
+
+	/* FIXME: do we care that there's no locking here? */
+	sum = 0;
+	for_each_online_cpu(cpu) {
+		cpu_totals[cpu] = 0;
+		for (i = 0; i < HvLpEvent_Type_NumTypes; i++) {
+			cpu_totals[cpu] += per_cpu(hvlpevent_counts, cpu)[i];
+		}
+		sum += cpu_totals[cpu];
+	}
+
+	seq_printf(m, "LpEventQueue 0\n");
+	seq_printf(m, "  events processed:\t%lu\n", sum);
+
+	for (i = 0; i < HvLpEvent_Type_NumTypes; ++i) {
+		sum = 0;
+		for_each_online_cpu(cpu) {
+			sum += per_cpu(hvlpevent_counts, cpu)[i];
+		}
+
+		seq_printf(m, "    %-20s %10lu\n", event_types[i], sum);
+	}
+
+	seq_printf(m, "\n  events processed by processor:\n");
+
+	for_each_online_cpu(cpu) {
+		seq_printf(m, "    CPU%02d  %10lu\n", cpu, cpu_totals[cpu]);
+	}
+
+	return 0;
+}
+
+static int proc_lpevents_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, proc_lpevents_show, NULL);
+}
+
+static struct file_operations proc_lpevents_operations = {
+	.open		= proc_lpevents_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static int __init proc_lpevents_init(void)
+{
+	struct proc_dir_entry *e;
+
+	e = create_proc_entry("iSeries/lpevents", S_IFREG|S_IRUGO, NULL);
+	if (e)
+		e->proc_fops = &proc_lpevents_operations;
+
+	return 0;
+}
+__initcall(proc_lpevents_init);
+
diff --git a/arch/powerpc/platforms/iseries/main_store.h b/arch/powerpc/platforms/iseries/main_store.h
new file mode 100644
index 000000000000..74f6889f834f
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/main_store.h
@@ -0,0 +1,165 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#ifndef _ISERIES_MAIN_STORE_H
+#define _ISERIES_MAIN_STORE_H
+
+/* Main Store Vpd for Condor,iStar,sStar */
+struct IoHriMainStoreSegment4 {
+	u8	msArea0Exists:1;
+	u8	msArea1Exists:1;
+	u8	msArea2Exists:1;
+	u8	msArea3Exists:1;
+	u8	reserved1:4;
+	u8	reserved2;
+
+	u8	msArea0Functional:1;
+	u8	msArea1Functional:1;
+	u8	msArea2Functional:1;
+	u8	msArea3Functional:1;
+	u8	reserved3:4;
+	u8	reserved4;
+
+	u32	totalMainStore;
+
+	u64	msArea0Ptr;
+	u64	msArea1Ptr;
+	u64	msArea2Ptr;
+	u64	msArea3Ptr;
+
+	u32	cardProductionLevel;
+
+	u32	msAdrHole;
+
+	u8	msArea0HasRiserVpd:1;
+	u8	msArea1HasRiserVpd:1;
+	u8	msArea2HasRiserVpd:1;
+	u8	msArea3HasRiserVpd:1;
+	u8	reserved5:4;
+	u8	reserved6;
+	u16	reserved7;
+
+	u8	reserved8[28];
+
+	u64	nonInterleavedBlocksStartAdr;
+	u64	nonInterleavedBlocksEndAdr;
+};
+
+/* Main Store VPD for Power4 */
+struct IoHriMainStoreChipInfo1 {
+	u32	chipMfgID	__attribute((packed));
+	char	chipECLevel[4]	__attribute((packed));
+};
+
+struct IoHriMainStoreVpdIdData {
+	char	typeNumber[4];
+	char	modelNumber[4];
+	char	partNumber[12];
+	char	serialNumber[12];
+};
+
+struct IoHriMainStoreVpdFruData {
+	char	fruLabel[8]	__attribute((packed));
+	u8	numberOfSlots	__attribute((packed));
+	u8	pluggingType	__attribute((packed));
+	u16	slotMapIndex	__attribute((packed));
+};
+
+struct IoHriMainStoreAdrRangeBlock {
+	void	*blockStart      __attribute((packed));
+	void	*blockEnd        __attribute((packed));
+	u32	blockProcChipId __attribute((packed));
+};
+
+#define MaxAreaAdrRangeBlocks 4
+
+struct IoHriMainStoreArea4 {
+	u32	msVpdFormat			__attribute((packed));
+	u8	containedVpdType		__attribute((packed));
+	u8	reserved1			__attribute((packed));
+	u16	reserved2			__attribute((packed));
+
+	u64	msExists			__attribute((packed));
+	u64	msFunctional			__attribute((packed));
+
+	u32	memorySize			__attribute((packed));
+	u32	procNodeId			__attribute((packed));
+
+	u32	numAdrRangeBlocks		__attribute((packed));
+	struct IoHriMainStoreAdrRangeBlock xAdrRangeBlock[MaxAreaAdrRangeBlocks]	__attribute((packed));
+
+	struct IoHriMainStoreChipInfo1	chipInfo0	__attribute((packed));
+	struct IoHriMainStoreChipInfo1	chipInfo1	__attribute((packed));
+	struct IoHriMainStoreChipInfo1	chipInfo2	__attribute((packed));
+	struct IoHriMainStoreChipInfo1	chipInfo3	__attribute((packed));
+	struct IoHriMainStoreChipInfo1	chipInfo4	__attribute((packed));
+	struct IoHriMainStoreChipInfo1	chipInfo5	__attribute((packed));
+	struct IoHriMainStoreChipInfo1	chipInfo6	__attribute((packed));
+	struct IoHriMainStoreChipInfo1	chipInfo7	__attribute((packed));
+
+	void	*msRamAreaArray			__attribute((packed));
+	u32	msRamAreaArrayNumEntries	__attribute((packed));
+	u32	msRamAreaArrayEntrySize		__attribute((packed));
+
+	u32	numaDimmExists			__attribute((packed));
+	u32	numaDimmFunctional		__attribute((packed));
+	void	*numaDimmArray			__attribute((packed));
+	u32	numaDimmArrayNumEntries		__attribute((packed));
+	u32	numaDimmArrayEntrySize		__attribute((packed));
+
+	struct IoHriMainStoreVpdIdData idData	__attribute((packed));
+
+	u64	powerData			__attribute((packed));
+	u64	cardAssemblyPartNum		__attribute((packed));
+	u64	chipSerialNum			__attribute((packed));
+
+	u64	reserved3			__attribute((packed));
+	char	reserved4[16]			__attribute((packed));
+
+	struct IoHriMainStoreVpdFruData fruData	__attribute((packed));
+
+	u8	vpdPortNum			__attribute((packed));
+	u8	reserved5			__attribute((packed));
+	u8	frameId				__attribute((packed));
+	u8	rackUnit			__attribute((packed));
+	char	asciiKeywordVpd[256]		__attribute((packed));
+	u32	reserved6			__attribute((packed));
+};
+
+
+struct IoHriMainStoreSegment5 {
+	u16	reserved1;
+	u8	reserved2;
+	u8	msVpdFormat;
+
+	u32	totalMainStore;
+	u64	maxConfiguredMsAdr;
+
+	struct IoHriMainStoreArea4	*msAreaArray;
+	u32	msAreaArrayNumEntries;
+	u32	msAreaArrayEntrySize;
+
+	u32	msAreaExists;
+	u32	msAreaFunctional;
+
+	u64	reserved3;
+};
+
+extern u64	xMsVpd[];
+
+#endif	/* _ISERIES_MAIN_STORE_H */
diff --git a/arch/powerpc/platforms/iseries/mf.c b/arch/powerpc/platforms/iseries/mf.c
new file mode 100644
index 000000000000..49e7e4b85847
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/mf.c
@@ -0,0 +1,1321 @@
+/*
+ * Copyright (C) 2001 Troy D. Armstrong  IBM Corporation
+ * Copyright (C) 2004-2005 Stephen Rothwell  IBM Corporation
+ *
+ * This modules exists as an interface between a Linux secondary partition
+ * running on an iSeries and the primary partition's Virtual Service
+ * Processor (VSP) object.  The VSP has final authority over powering on/off
+ * all partitions in the iSeries.  It also provides miscellaneous low-level
+ * machine facility type operations.
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/completion.h>
+#include <linux/delay.h>
+#include <linux/dma-mapping.h>
+#include <linux/bcd.h>
+#include <linux/rtc.h>
+
+#include <asm/time.h>
+#include <asm/uaccess.h>
+#include <asm/paca.h>
+#include <asm/abs_addr.h>
+#include <asm/iseries/vio.h>
+#include <asm/iseries/mf.h>
+#include <asm/iseries/hv_lp_config.h>
+#include <asm/iseries/it_lp_queue.h>
+
+#include "setup.h"
+
+extern int piranha_simulator;
+
+/*
+ * This is the structure layout for the Machine Facilites LPAR event
+ * flows.
+ */
+struct vsp_cmd_data {
+	u64 token;
+	u16 cmd;
+	HvLpIndex lp_index;
+	u8 result_code;
+	u32 reserved;
+	union {
+		u64 state;	/* GetStateOut */
+		u64 ipl_type;	/* GetIplTypeOut, Function02SelectIplTypeIn */
+		u64 ipl_mode;	/* GetIplModeOut, Function02SelectIplModeIn */
+		u64 page[4];	/* GetSrcHistoryIn */
+		u64 flag;	/* GetAutoIplWhenPrimaryIplsOut,
+				   SetAutoIplWhenPrimaryIplsIn,
+				   WhiteButtonPowerOffIn,
+				   Function08FastPowerOffIn,
+				   IsSpcnRackPowerIncompleteOut */
+		struct {
+			u64 token;
+			u64 address_type;
+			u64 side;
+			u32 length;
+			u32 offset;
+		} kern;		/* SetKernelImageIn, GetKernelImageIn,
+				   SetKernelCmdLineIn, GetKernelCmdLineIn */
+		u32 length_out;	/* GetKernelImageOut, GetKernelCmdLineOut */
+		u8 reserved[80];
+	} sub_data;
+};
+
+struct vsp_rsp_data {
+	struct completion com;
+	struct vsp_cmd_data *response;
+};
+
+struct alloc_data {
+	u16 size;
+	u16 type;
+	u32 count;
+	u16 reserved1;
+	u8 reserved2;
+	HvLpIndex target_lp;
+};
+
+struct ce_msg_data;
+
+typedef void (*ce_msg_comp_hdlr)(void *token, struct ce_msg_data *vsp_cmd_rsp);
+
+struct ce_msg_comp_data {
+	ce_msg_comp_hdlr handler;
+	void *token;
+};
+
+struct ce_msg_data {
+	u8 ce_msg[12];
+	char reserved[4];
+	struct ce_msg_comp_data *completion;
+};
+
+struct io_mf_lp_event {
+	struct HvLpEvent hp_lp_event;
+	u16 subtype_result_code;
+	u16 reserved1;
+	u32 reserved2;
+	union {
+		struct alloc_data alloc;
+		struct ce_msg_data ce_msg;
+		struct vsp_cmd_data vsp_cmd;
+	} data;
+};
+
+#define subtype_data(a, b, c, d)	\
+		(((a) << 24) + ((b) << 16) + ((c) << 8) + (d))
+
+/*
+ * All outgoing event traffic is kept on a FIFO queue.  The first
+ * pointer points to the one that is outstanding, and all new
+ * requests get stuck on the end.  Also, we keep a certain number of
+ * preallocated pending events so that we can operate very early in
+ * the boot up sequence (before kmalloc is ready).
+ */
+struct pending_event {
+	struct pending_event *next;
+	struct io_mf_lp_event event;
+	MFCompleteHandler hdlr;
+	char dma_data[72];
+	unsigned dma_data_length;
+	unsigned remote_address;
+};
+static spinlock_t pending_event_spinlock;
+static struct pending_event *pending_event_head;
+static struct pending_event *pending_event_tail;
+static struct pending_event *pending_event_avail;
+static struct pending_event pending_event_prealloc[16];
+
+/*
+ * Put a pending event onto the available queue, so it can get reused.
+ * Attention! You must have the pending_event_spinlock before calling!
+ */
+static void free_pending_event(struct pending_event *ev)
+{
+	if (ev != NULL) {
+		ev->next = pending_event_avail;
+		pending_event_avail = ev;
+	}
+}
+
+/*
+ * Enqueue the outbound event onto the stack.  If the queue was
+ * empty to begin with, we must also issue it via the Hypervisor
+ * interface.  There is a section of code below that will touch
+ * the first stack pointer without the protection of the pending_event_spinlock.
+ * This is OK, because we know that nobody else will be modifying
+ * the first pointer when we do this.
+ */
+static int signal_event(struct pending_event *ev)
+{
+	int rc = 0;
+	unsigned long flags;
+	int go = 1;
+	struct pending_event *ev1;
+	HvLpEvent_Rc hv_rc;
+
+	/* enqueue the event */
+	if (ev != NULL) {
+		ev->next = NULL;
+		spin_lock_irqsave(&pending_event_spinlock, flags);
+		if (pending_event_head == NULL)
+			pending_event_head = ev;
+		else {
+			go = 0;
+			pending_event_tail->next = ev;
+		}
+		pending_event_tail = ev;
+		spin_unlock_irqrestore(&pending_event_spinlock, flags);
+	}
+
+	/* send the event */
+	while (go) {
+		go = 0;
+
+		/* any DMA data to send beforehand? */
+		if (pending_event_head->dma_data_length > 0)
+			HvCallEvent_dmaToSp(pending_event_head->dma_data,
+					pending_event_head->remote_address,
+					pending_event_head->dma_data_length,
+					HvLpDma_Direction_LocalToRemote);
+
+		hv_rc = HvCallEvent_signalLpEvent(
+				&pending_event_head->event.hp_lp_event);
+		if (hv_rc != HvLpEvent_Rc_Good) {
+			printk(KERN_ERR "mf.c: HvCallEvent_signalLpEvent() "
+					"failed with %d\n", (int)hv_rc);
+
+			spin_lock_irqsave(&pending_event_spinlock, flags);
+			ev1 = pending_event_head;
+			pending_event_head = pending_event_head->next;
+			if (pending_event_head != NULL)
+				go = 1;
+			spin_unlock_irqrestore(&pending_event_spinlock, flags);
+
+			if (ev1 == ev)
+				rc = -EIO;
+			else if (ev1->hdlr != NULL)
+				(*ev1->hdlr)((void *)ev1->event.hp_lp_event.xCorrelationToken, -EIO);
+
+			spin_lock_irqsave(&pending_event_spinlock, flags);
+			free_pending_event(ev1);
+			spin_unlock_irqrestore(&pending_event_spinlock, flags);
+		}
+	}
+
+	return rc;
+}
+
+/*
+ * Allocate a new pending_event structure, and initialize it.
+ */
+static struct pending_event *new_pending_event(void)
+{
+	struct pending_event *ev = NULL;
+	HvLpIndex primary_lp = HvLpConfig_getPrimaryLpIndex();
+	unsigned long flags;
+	struct HvLpEvent *hev;
+
+	spin_lock_irqsave(&pending_event_spinlock, flags);
+	if (pending_event_avail != NULL) {
+		ev = pending_event_avail;
+		pending_event_avail = pending_event_avail->next;
+	}
+	spin_unlock_irqrestore(&pending_event_spinlock, flags);
+	if (ev == NULL) {
+		ev = kmalloc(sizeof(struct pending_event), GFP_ATOMIC);
+		if (ev == NULL) {
+			printk(KERN_ERR "mf.c: unable to kmalloc %ld bytes\n",
+					sizeof(struct pending_event));
+			return NULL;
+		}
+	}
+	memset(ev, 0, sizeof(struct pending_event));
+	hev = &ev->event.hp_lp_event;
+	hev->xFlags.xValid = 1;
+	hev->xFlags.xAckType = HvLpEvent_AckType_ImmediateAck;
+	hev->xFlags.xAckInd = HvLpEvent_AckInd_DoAck;
+	hev->xFlags.xFunction = HvLpEvent_Function_Int;
+	hev->xType = HvLpEvent_Type_MachineFac;
+	hev->xSourceLp = HvLpConfig_getLpIndex();
+	hev->xTargetLp = primary_lp;
+	hev->xSizeMinus1 = sizeof(ev->event) - 1;
+	hev->xRc = HvLpEvent_Rc_Good;
+	hev->xSourceInstanceId = HvCallEvent_getSourceLpInstanceId(primary_lp,
+			HvLpEvent_Type_MachineFac);
+	hev->xTargetInstanceId = HvCallEvent_getTargetLpInstanceId(primary_lp,
+			HvLpEvent_Type_MachineFac);
+
+	return ev;
+}
+
+static int signal_vsp_instruction(struct vsp_cmd_data *vsp_cmd)
+{
+	struct pending_event *ev = new_pending_event();
+	int rc;
+	struct vsp_rsp_data response;
+
+	if (ev == NULL)
+		return -ENOMEM;
+
+	init_completion(&response.com);
+	response.response = vsp_cmd;
+	ev->event.hp_lp_event.xSubtype = 6;
+	ev->event.hp_lp_event.x.xSubtypeData =
+		subtype_data('M', 'F',  'V',  'I');
+	ev->event.data.vsp_cmd.token = (u64)&response;
+	ev->event.data.vsp_cmd.cmd = vsp_cmd->cmd;
+	ev->event.data.vsp_cmd.lp_index = HvLpConfig_getLpIndex();
+	ev->event.data.vsp_cmd.result_code = 0xFF;
+	ev->event.data.vsp_cmd.reserved = 0;
+	memcpy(&(ev->event.data.vsp_cmd.sub_data),
+			&(vsp_cmd->sub_data), sizeof(vsp_cmd->sub_data));
+	mb();
+
+	rc = signal_event(ev);
+	if (rc == 0)
+		wait_for_completion(&response.com);
+	return rc;
+}
+
+
+/*
+ * Send a 12-byte CE message to the primary partition VSP object
+ */
+static int signal_ce_msg(char *ce_msg, struct ce_msg_comp_data *completion)
+{
+	struct pending_event *ev = new_pending_event();
+
+	if (ev == NULL)
+		return -ENOMEM;
+
+	ev->event.hp_lp_event.xSubtype = 0;
+	ev->event.hp_lp_event.x.xSubtypeData =
+		subtype_data('M',  'F',  'C',  'E');
+	memcpy(ev->event.data.ce_msg.ce_msg, ce_msg, 12);
+	ev->event.data.ce_msg.completion = completion;
+	return signal_event(ev);
+}
+
+/*
+ * Send a 12-byte CE message (with no data) to the primary partition VSP object
+ */
+static int signal_ce_msg_simple(u8 ce_op, struct ce_msg_comp_data *completion)
+{
+	u8 ce_msg[12];
+
+	memset(ce_msg, 0, sizeof(ce_msg));
+	ce_msg[3] = ce_op;
+	return signal_ce_msg(ce_msg, completion);
+}
+
+/*
+ * Send a 12-byte CE message and DMA data to the primary partition VSP object
+ */
+static int dma_and_signal_ce_msg(char *ce_msg,
+		struct ce_msg_comp_data *completion, void *dma_data,
+		unsigned dma_data_length, unsigned remote_address)
+{
+	struct pending_event *ev = new_pending_event();
+
+	if (ev == NULL)
+		return -ENOMEM;
+
+	ev->event.hp_lp_event.xSubtype = 0;
+	ev->event.hp_lp_event.x.xSubtypeData =
+		subtype_data('M', 'F', 'C', 'E');
+	memcpy(ev->event.data.ce_msg.ce_msg, ce_msg, 12);
+	ev->event.data.ce_msg.completion = completion;
+	memcpy(ev->dma_data, dma_data, dma_data_length);
+	ev->dma_data_length = dma_data_length;
+	ev->remote_address = remote_address;
+	return signal_event(ev);
+}
+
+/*
+ * Initiate a nice (hopefully) shutdown of Linux.  We simply are
+ * going to try and send the init process a SIGINT signal.  If
+ * this fails (why?), we'll simply force it off in a not-so-nice
+ * manner.
+ */
+static int shutdown(void)
+{
+	int rc = kill_proc(1, SIGINT, 1);
+
+	if (rc) {
+		printk(KERN_ALERT "mf.c: SIGINT to init failed (%d), "
+				"hard shutdown commencing\n", rc);
+		mf_power_off();
+	} else
+		printk(KERN_INFO "mf.c: init has been successfully notified "
+				"to proceed with shutdown\n");
+	return rc;
+}
+
+/*
+ * The primary partition VSP object is sending us a new
+ * event flow.  Handle it...
+ */
+static void handle_int(struct io_mf_lp_event *event)
+{
+	struct ce_msg_data *ce_msg_data;
+	struct ce_msg_data *pce_msg_data;
+	unsigned long flags;
+	struct pending_event *pev;
+
+	/* ack the interrupt */
+	event->hp_lp_event.xRc = HvLpEvent_Rc_Good;
+	HvCallEvent_ackLpEvent(&event->hp_lp_event);
+
+	/* process interrupt */
+	switch (event->hp_lp_event.xSubtype) {
+	case 0:	/* CE message */
+		ce_msg_data = &event->data.ce_msg;
+		switch (ce_msg_data->ce_msg[3]) {
+		case 0x5B:	/* power control notification */
+			if ((ce_msg_data->ce_msg[5] & 0x20) != 0) {
+				printk(KERN_INFO "mf.c: Commencing partition shutdown\n");
+				if (shutdown() == 0)
+					signal_ce_msg_simple(0xDB, NULL);
+			}
+			break;
+		case 0xC0:	/* get time */
+			spin_lock_irqsave(&pending_event_spinlock, flags);
+			pev = pending_event_head;
+			if (pev != NULL)
+				pending_event_head = pending_event_head->next;
+			spin_unlock_irqrestore(&pending_event_spinlock, flags);
+			if (pev == NULL)
+				break;
+			pce_msg_data = &pev->event.data.ce_msg;
+			if (pce_msg_data->ce_msg[3] != 0x40)
+				break;
+			if (pce_msg_data->completion != NULL) {
+				ce_msg_comp_hdlr handler =
+					pce_msg_data->completion->handler;
+				void *token = pce_msg_data->completion->token;
+
+				if (handler != NULL)
+					(*handler)(token, ce_msg_data);
+			}
+			spin_lock_irqsave(&pending_event_spinlock, flags);
+			free_pending_event(pev);
+			spin_unlock_irqrestore(&pending_event_spinlock, flags);
+			/* send next waiting event */
+			if (pending_event_head != NULL)
+				signal_event(NULL);
+			break;
+		}
+		break;
+	case 1:	/* IT sys shutdown */
+		printk(KERN_INFO "mf.c: Commencing system shutdown\n");
+		shutdown();
+		break;
+	}
+}
+
+/*
+ * The primary partition VSP object is acknowledging the receipt
+ * of a flow we sent to them.  If there are other flows queued
+ * up, we must send another one now...
+ */
+static void handle_ack(struct io_mf_lp_event *event)
+{
+	unsigned long flags;
+	struct pending_event *two = NULL;
+	unsigned long free_it = 0;
+	struct ce_msg_data *ce_msg_data;
+	struct ce_msg_data *pce_msg_data;
+	struct vsp_rsp_data *rsp;
+
+	/* handle current event */
+	if (pending_event_head == NULL) {
+		printk(KERN_ERR "mf.c: stack empty for receiving ack\n");
+		return;
+	}
+
+	switch (event->hp_lp_event.xSubtype) {
+	case 0:     /* CE msg */
+		ce_msg_data = &event->data.ce_msg;
+		if (ce_msg_data->ce_msg[3] != 0x40) {
+			free_it = 1;
+			break;
+		}
+		if (ce_msg_data->ce_msg[2] == 0)
+			break;
+		free_it = 1;
+		pce_msg_data = &pending_event_head->event.data.ce_msg;
+		if (pce_msg_data->completion != NULL) {
+			ce_msg_comp_hdlr handler =
+				pce_msg_data->completion->handler;
+			void *token = pce_msg_data->completion->token;
+
+			if (handler != NULL)
+				(*handler)(token, ce_msg_data);
+		}
+		break;
+	case 4:	/* allocate */
+	case 5:	/* deallocate */
+		if (pending_event_head->hdlr != NULL)
+			(*pending_event_head->hdlr)((void *)event->hp_lp_event.xCorrelationToken, event->data.alloc.count);
+		free_it = 1;
+		break;
+	case 6:
+		free_it = 1;
+		rsp = (struct vsp_rsp_data *)event->data.vsp_cmd.token;
+		if (rsp == NULL) {
+			printk(KERN_ERR "mf.c: no rsp\n");
+			break;
+		}
+		if (rsp->response != NULL)
+			memcpy(rsp->response, &event->data.vsp_cmd,
+					sizeof(event->data.vsp_cmd));
+		complete(&rsp->com);
+		break;
+	}
+
+	/* remove from queue */
+	spin_lock_irqsave(&pending_event_spinlock, flags);
+	if ((pending_event_head != NULL) && (free_it == 1)) {
+		struct pending_event *oldHead = pending_event_head;
+
+		pending_event_head = pending_event_head->next;
+		two = pending_event_head;
+		free_pending_event(oldHead);
+	}
+	spin_unlock_irqrestore(&pending_event_spinlock, flags);
+
+	/* send next waiting event */
+	if (two != NULL)
+		signal_event(NULL);
+}
+
+/*
+ * This is the generic event handler we are registering with
+ * the Hypervisor.  Ensure the flows are for us, and then
+ * parse it enough to know if it is an interrupt or an
+ * acknowledge.
+ */
+static void hv_handler(struct HvLpEvent *event, struct pt_regs *regs)
+{
+	if ((event != NULL) && (event->xType == HvLpEvent_Type_MachineFac)) {
+		switch(event->xFlags.xFunction) {
+		case HvLpEvent_Function_Ack:
+			handle_ack((struct io_mf_lp_event *)event);
+			break;
+		case HvLpEvent_Function_Int:
+			handle_int((struct io_mf_lp_event *)event);
+			break;
+		default:
+			printk(KERN_ERR "mf.c: non ack/int event received\n");
+			break;
+		}
+	} else
+		printk(KERN_ERR "mf.c: alien event received\n");
+}
+
+/*
+ * Global kernel interface to allocate and seed events into the
+ * Hypervisor.
+ */
+void mf_allocate_lp_events(HvLpIndex target_lp, HvLpEvent_Type type,
+		unsigned size, unsigned count, MFCompleteHandler hdlr,
+		void *user_token)
+{
+	struct pending_event *ev = new_pending_event();
+	int rc;
+
+	if (ev == NULL) {
+		rc = -ENOMEM;
+	} else {
+		ev->event.hp_lp_event.xSubtype = 4;
+		ev->event.hp_lp_event.xCorrelationToken = (u64)user_token;
+		ev->event.hp_lp_event.x.xSubtypeData =
+			subtype_data('M', 'F', 'M', 'A');
+		ev->event.data.alloc.target_lp = target_lp;
+		ev->event.data.alloc.type = type;
+		ev->event.data.alloc.size = size;
+		ev->event.data.alloc.count = count;
+		ev->hdlr = hdlr;
+		rc = signal_event(ev);
+	}
+	if ((rc != 0) && (hdlr != NULL))
+		(*hdlr)(user_token, rc);
+}
+EXPORT_SYMBOL(mf_allocate_lp_events);
+
+/*
+ * Global kernel interface to unseed and deallocate events already in
+ * Hypervisor.
+ */
+void mf_deallocate_lp_events(HvLpIndex target_lp, HvLpEvent_Type type,
+		unsigned count, MFCompleteHandler hdlr, void *user_token)
+{
+	struct pending_event *ev = new_pending_event();
+	int rc;
+
+	if (ev == NULL)
+		rc = -ENOMEM;
+	else {
+		ev->event.hp_lp_event.xSubtype = 5;
+		ev->event.hp_lp_event.xCorrelationToken = (u64)user_token;
+		ev->event.hp_lp_event.x.xSubtypeData =
+			subtype_data('M', 'F', 'M', 'D');
+		ev->event.data.alloc.target_lp = target_lp;
+		ev->event.data.alloc.type = type;
+		ev->event.data.alloc.count = count;
+		ev->hdlr = hdlr;
+		rc = signal_event(ev);
+	}
+	if ((rc != 0) && (hdlr != NULL))
+		(*hdlr)(user_token, rc);
+}
+EXPORT_SYMBOL(mf_deallocate_lp_events);
+
+/*
+ * Global kernel interface to tell the VSP object in the primary
+ * partition to power this partition off.
+ */
+void mf_power_off(void)
+{
+	printk(KERN_INFO "mf.c: Down it goes...\n");
+	signal_ce_msg_simple(0x4d, NULL);
+	for (;;)
+		;
+}
+
+/*
+ * Global kernel interface to tell the VSP object in the primary
+ * partition to reboot this partition.
+ */
+void mf_reboot(void)
+{
+	printk(KERN_INFO "mf.c: Preparing to bounce...\n");
+	signal_ce_msg_simple(0x4e, NULL);
+	for (;;)
+		;
+}
+
+/*
+ * Display a single word SRC onto the VSP control panel.
+ */
+void mf_display_src(u32 word)
+{
+	u8 ce[12];
+
+	memset(ce, 0, sizeof(ce));
+	ce[3] = 0x4a;
+	ce[7] = 0x01;
+	ce[8] = word >> 24;
+	ce[9] = word >> 16;
+	ce[10] = word >> 8;
+	ce[11] = word;
+	signal_ce_msg(ce, NULL);
+}
+
+/*
+ * Display a single word SRC of the form "PROGXXXX" on the VSP control panel.
+ */
+void mf_display_progress(u16 value)
+{
+	u8 ce[12];
+	u8 src[72];
+
+	memcpy(ce, "\x00\x00\x04\x4A\x00\x00\x00\x48\x00\x00\x00\x00", 12);
+	memcpy(src, "\x01\x00\x00\x01\x00\x00\x00\x00\x00\x00\x00\x00"
+		"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"
+		"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"
+		"\x00\x00\x00\x00PROGxxxx                        ",
+		72);
+	src[6] = value >> 8;
+	src[7] = value & 255;
+	src[44] = "0123456789ABCDEF"[(value >> 12) & 15];
+	src[45] = "0123456789ABCDEF"[(value >> 8) & 15];
+	src[46] = "0123456789ABCDEF"[(value >> 4) & 15];
+	src[47] = "0123456789ABCDEF"[value & 15];
+	dma_and_signal_ce_msg(ce, NULL, src, sizeof(src), 9 * 64 * 1024);
+}
+
+/*
+ * Clear the VSP control panel.  Used to "erase" an SRC that was
+ * previously displayed.
+ */
+void mf_clear_src(void)
+{
+	signal_ce_msg_simple(0x4b, NULL);
+}
+
+/*
+ * Initialization code here.
+ */
+void mf_init(void)
+{
+	int i;
+
+	/* initialize */
+	spin_lock_init(&pending_event_spinlock);
+	for (i = 0;
+	     i < sizeof(pending_event_prealloc) / sizeof(*pending_event_prealloc);
+	     ++i)
+		free_pending_event(&pending_event_prealloc[i]);
+	HvLpEvent_registerHandler(HvLpEvent_Type_MachineFac, &hv_handler);
+
+	/* virtual continue ack */
+	signal_ce_msg_simple(0x57, NULL);
+
+	/* initialization complete */
+	printk(KERN_NOTICE "mf.c: iSeries Linux LPAR Machine Facilities "
+			"initialized\n");
+}
+
+struct rtc_time_data {
+	struct completion com;
+	struct ce_msg_data ce_msg;
+	int rc;
+};
+
+static void get_rtc_time_complete(void *token, struct ce_msg_data *ce_msg)
+{
+	struct rtc_time_data *rtc = token;
+
+	memcpy(&rtc->ce_msg, ce_msg, sizeof(rtc->ce_msg));
+	rtc->rc = 0;
+	complete(&rtc->com);
+}
+
+static int rtc_set_tm(int rc, u8 *ce_msg, struct rtc_time *tm)
+{
+	tm->tm_wday = 0;
+	tm->tm_yday = 0;
+	tm->tm_isdst = 0;
+	if (rc) {
+		tm->tm_sec = 0;
+		tm->tm_min = 0;
+		tm->tm_hour = 0;
+		tm->tm_mday = 15;
+		tm->tm_mon = 5;
+		tm->tm_year = 52;
+		return rc;
+	}
+
+	if ((ce_msg[2] == 0xa9) ||
+	    (ce_msg[2] == 0xaf)) {
+		/* TOD clock is not set */
+		tm->tm_sec = 1;
+		tm->tm_min = 1;
+		tm->tm_hour = 1;
+		tm->tm_mday = 10;
+		tm->tm_mon = 8;
+		tm->tm_year = 71;
+		mf_set_rtc(tm);
+	}
+	{
+		u8 year = ce_msg[5];
+		u8 sec = ce_msg[6];
+		u8 min = ce_msg[7];
+		u8 hour = ce_msg[8];
+		u8 day = ce_msg[10];
+		u8 mon = ce_msg[11];
+
+		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 <= 69)
+			year += 100;
+
+		tm->tm_sec = sec;
+		tm->tm_min = min;
+		tm->tm_hour = hour;
+		tm->tm_mday = day;
+		tm->tm_mon = mon;
+		tm->tm_year = year;
+	}
+
+	return 0;
+}
+
+int mf_get_rtc(struct rtc_time *tm)
+{
+	struct ce_msg_comp_data ce_complete;
+	struct rtc_time_data rtc_data;
+	int rc;
+
+	memset(&ce_complete, 0, sizeof(ce_complete));
+	memset(&rtc_data, 0, sizeof(rtc_data));
+	init_completion(&rtc_data.com);
+	ce_complete.handler = &get_rtc_time_complete;
+	ce_complete.token = &rtc_data;
+	rc = signal_ce_msg_simple(0x40, &ce_complete);
+	if (rc)
+		return rc;
+	wait_for_completion(&rtc_data.com);
+	return rtc_set_tm(rtc_data.rc, rtc_data.ce_msg.ce_msg, tm);
+}
+
+struct boot_rtc_time_data {
+	int busy;
+	struct ce_msg_data ce_msg;
+	int rc;
+};
+
+static void get_boot_rtc_time_complete(void *token, struct ce_msg_data *ce_msg)
+{
+	struct boot_rtc_time_data *rtc = token;
+
+	memcpy(&rtc->ce_msg, ce_msg, sizeof(rtc->ce_msg));
+	rtc->rc = 0;
+	rtc->busy = 0;
+}
+
+int mf_get_boot_rtc(struct rtc_time *tm)
+{
+	struct ce_msg_comp_data ce_complete;
+	struct boot_rtc_time_data rtc_data;
+	int rc;
+
+	memset(&ce_complete, 0, sizeof(ce_complete));
+	memset(&rtc_data, 0, sizeof(rtc_data));
+	rtc_data.busy = 1;
+	ce_complete.handler = &get_boot_rtc_time_complete;
+	ce_complete.token = &rtc_data;
+	rc = signal_ce_msg_simple(0x40, &ce_complete);
+	if (rc)
+		return rc;
+	/* We need to poll here as we are not yet taking interrupts */
+	while (rtc_data.busy) {
+		if (hvlpevent_is_pending())
+			process_hvlpevents(NULL);
+	}
+	return rtc_set_tm(rtc_data.rc, rtc_data.ce_msg.ce_msg, tm);
+}
+
+int mf_set_rtc(struct rtc_time *tm)
+{
+	char ce_time[12];
+	u8 day, mon, hour, min, sec, y1, y2;
+	unsigned year;
+
+	year = 1900 + tm->tm_year;
+	y1 = year / 100;
+	y2 = year % 100;
+
+	sec = tm->tm_sec;
+	min = tm->tm_min;
+	hour = tm->tm_hour;
+	day = tm->tm_mday;
+	mon = tm->tm_mon + 1;
+
+	BIN_TO_BCD(sec);
+	BIN_TO_BCD(min);
+	BIN_TO_BCD(hour);
+	BIN_TO_BCD(mon);
+	BIN_TO_BCD(day);
+	BIN_TO_BCD(y1);
+	BIN_TO_BCD(y2);
+
+	memset(ce_time, 0, sizeof(ce_time));
+	ce_time[3] = 0x41;
+	ce_time[4] = y1;
+	ce_time[5] = y2;
+	ce_time[6] = sec;
+	ce_time[7] = min;
+	ce_time[8] = hour;
+	ce_time[10] = day;
+	ce_time[11] = mon;
+
+	return signal_ce_msg(ce_time, NULL);
+}
+
+#ifdef CONFIG_PROC_FS
+
+static int proc_mf_dump_cmdline(char *page, char **start, off_t off,
+		int count, int *eof, void *data)
+{
+	int len;
+	char *p;
+	struct vsp_cmd_data vsp_cmd;
+	int rc;
+	dma_addr_t dma_addr;
+
+	/* The HV appears to return no more than 256 bytes of command line */
+	if (off >= 256)
+		return 0;
+	if ((off + count) > 256)
+		count = 256 - off;
+
+	dma_addr = dma_map_single(iSeries_vio_dev, page, off + count,
+			DMA_FROM_DEVICE);
+	if (dma_mapping_error(dma_addr))
+		return -ENOMEM;
+	memset(page, 0, off + count);
+	memset(&vsp_cmd, 0, sizeof(vsp_cmd));
+	vsp_cmd.cmd = 33;
+	vsp_cmd.sub_data.kern.token = dma_addr;
+	vsp_cmd.sub_data.kern.address_type = HvLpDma_AddressType_TceIndex;
+	vsp_cmd.sub_data.kern.side = (u64)data;
+	vsp_cmd.sub_data.kern.length = off + count;
+	mb();
+	rc = signal_vsp_instruction(&vsp_cmd);
+	dma_unmap_single(iSeries_vio_dev, dma_addr, off + count,
+			DMA_FROM_DEVICE);
+	if (rc)
+		return rc;
+	if (vsp_cmd.result_code != 0)
+		return -ENOMEM;
+	p = page;
+	len = 0;
+	while (len < (off + count)) {
+		if ((*p == '\0') || (*p == '\n')) {
+			if (*p == '\0')
+				*p = '\n';
+			p++;
+			len++;
+			*eof = 1;
+			break;
+		}
+		p++;
+		len++;
+	}
+
+	if (len < off) {
+		*eof = 1;
+		len = 0;
+	}
+	return len;
+}
+
+#if 0
+static int mf_getVmlinuxChunk(char *buffer, int *size, int offset, u64 side)
+{
+	struct vsp_cmd_data vsp_cmd;
+	int rc;
+	int len = *size;
+	dma_addr_t dma_addr;
+
+	dma_addr = dma_map_single(iSeries_vio_dev, buffer, len,
+			DMA_FROM_DEVICE);
+	memset(buffer, 0, len);
+	memset(&vsp_cmd, 0, sizeof(vsp_cmd));
+	vsp_cmd.cmd = 32;
+	vsp_cmd.sub_data.kern.token = dma_addr;
+	vsp_cmd.sub_data.kern.address_type = HvLpDma_AddressType_TceIndex;
+	vsp_cmd.sub_data.kern.side = side;
+	vsp_cmd.sub_data.kern.offset = offset;
+	vsp_cmd.sub_data.kern.length = len;
+	mb();
+	rc = signal_vsp_instruction(&vsp_cmd);
+	if (rc == 0) {
+		if (vsp_cmd.result_code == 0)
+			*size = vsp_cmd.sub_data.length_out;
+		else
+			rc = -ENOMEM;
+	}
+
+	dma_unmap_single(iSeries_vio_dev, dma_addr, len, DMA_FROM_DEVICE);
+
+	return rc;
+}
+
+static int proc_mf_dump_vmlinux(char *page, char **start, off_t off,
+		int count, int *eof, void *data)
+{
+	int sizeToGet = count;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	if (mf_getVmlinuxChunk(page, &sizeToGet, off, (u64)data) == 0) {
+		if (sizeToGet != 0) {
+			*start = page + off;
+			return sizeToGet;
+		}
+		*eof = 1;
+		return 0;
+	}
+	*eof = 1;
+	return 0;
+}
+#endif
+
+static int proc_mf_dump_side(char *page, char **start, off_t off,
+		int count, int *eof, void *data)
+{
+	int len;
+	char mf_current_side = ' ';
+	struct vsp_cmd_data vsp_cmd;
+
+	memset(&vsp_cmd, 0, sizeof(vsp_cmd));
+	vsp_cmd.cmd = 2;
+	vsp_cmd.sub_data.ipl_type = 0;
+	mb();
+
+	if (signal_vsp_instruction(&vsp_cmd) == 0) {
+		if (vsp_cmd.result_code == 0) {
+			switch (vsp_cmd.sub_data.ipl_type) {
+			case 0:	mf_current_side = 'A';
+				break;
+			case 1:	mf_current_side = 'B';
+				break;
+			case 2:	mf_current_side = 'C';
+				break;
+			default:	mf_current_side = 'D';
+				break;
+			}
+		}
+	}
+
+	len = sprintf(page, "%c\n", mf_current_side);
+
+	if (len <= (off + count))
+		*eof = 1;
+	*start = page + off;
+	len -= off;
+	if (len > count)
+		len = count;
+	if (len < 0)
+		len = 0;
+	return len;
+}
+
+static int proc_mf_change_side(struct file *file, const char __user *buffer,
+		unsigned long count, void *data)
+{
+	char side;
+	u64 newSide;
+	struct vsp_cmd_data vsp_cmd;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	if (count == 0)
+		return 0;
+
+	if (get_user(side, buffer))
+		return -EFAULT;
+
+	switch (side) {
+	case 'A':	newSide = 0;
+			break;
+	case 'B':	newSide = 1;
+			break;
+	case 'C':	newSide = 2;
+			break;
+	case 'D':	newSide = 3;
+			break;
+	default:
+		printk(KERN_ERR "mf_proc.c: proc_mf_change_side: invalid side\n");
+		return -EINVAL;
+	}
+
+	memset(&vsp_cmd, 0, sizeof(vsp_cmd));
+	vsp_cmd.sub_data.ipl_type = newSide;
+	vsp_cmd.cmd = 10;
+
+	(void)signal_vsp_instruction(&vsp_cmd);
+
+	return count;
+}
+
+#if 0
+static void mf_getSrcHistory(char *buffer, int size)
+{
+	struct IplTypeReturnStuff return_stuff;
+	struct pending_event *ev = new_pending_event();
+	int rc = 0;
+	char *pages[4];
+
+	pages[0] = kmalloc(4096, GFP_ATOMIC);
+	pages[1] = kmalloc(4096, GFP_ATOMIC);
+	pages[2] = kmalloc(4096, GFP_ATOMIC);
+	pages[3] = kmalloc(4096, GFP_ATOMIC);
+	if ((ev == NULL) || (pages[0] == NULL) || (pages[1] == NULL)
+			 || (pages[2] == NULL) || (pages[3] == NULL))
+		return -ENOMEM;
+
+	return_stuff.xType = 0;
+	return_stuff.xRc = 0;
+	return_stuff.xDone = 0;
+	ev->event.hp_lp_event.xSubtype = 6;
+	ev->event.hp_lp_event.x.xSubtypeData =
+		subtype_data('M', 'F', 'V', 'I');
+	ev->event.data.vsp_cmd.xEvent = &return_stuff;
+	ev->event.data.vsp_cmd.cmd = 4;
+	ev->event.data.vsp_cmd.lp_index = HvLpConfig_getLpIndex();
+	ev->event.data.vsp_cmd.result_code = 0xFF;
+	ev->event.data.vsp_cmd.reserved = 0;
+	ev->event.data.vsp_cmd.sub_data.page[0] = iseries_hv_addr(pages[0]);
+	ev->event.data.vsp_cmd.sub_data.page[1] = iseries_hv_addr(pages[1]);
+	ev->event.data.vsp_cmd.sub_data.page[2] = iseries_hv_addr(pages[2]);
+	ev->event.data.vsp_cmd.sub_data.page[3] = iseries_hv_addr(pages[3]);
+	mb();
+	if (signal_event(ev) != 0)
+		return;
+
+ 	while (return_stuff.xDone != 1)
+ 		udelay(10);
+ 	if (return_stuff.xRc == 0)
+ 		memcpy(buffer, pages[0], size);
+	kfree(pages[0]);
+	kfree(pages[1]);
+	kfree(pages[2]);
+	kfree(pages[3]);
+}
+#endif
+
+static int proc_mf_dump_src(char *page, char **start, off_t off,
+		int count, int *eof, void *data)
+{
+#if 0
+	int len;
+
+	mf_getSrcHistory(page, count);
+	len = count;
+	len -= off;
+	if (len < count) {
+		*eof = 1;
+		if (len <= 0)
+			return 0;
+	} else
+		len = count;
+	*start = page + off;
+	return len;
+#else
+	return 0;
+#endif
+}
+
+static int proc_mf_change_src(struct file *file, const char __user *buffer,
+		unsigned long count, void *data)
+{
+	char stkbuf[10];
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	if ((count < 4) && (count != 1)) {
+		printk(KERN_ERR "mf_proc: invalid src\n");
+		return -EINVAL;
+	}
+
+	if (count > (sizeof(stkbuf) - 1))
+		count = sizeof(stkbuf) - 1;
+	if (copy_from_user(stkbuf, buffer, count))
+		return -EFAULT;
+
+	if ((count == 1) && (*stkbuf == '\0'))
+		mf_clear_src();
+	else
+		mf_display_src(*(u32 *)stkbuf);
+
+	return count;
+}
+
+static int proc_mf_change_cmdline(struct file *file, const char __user *buffer,
+		unsigned long count, void *data)
+{
+	struct vsp_cmd_data vsp_cmd;
+	dma_addr_t dma_addr;
+	char *page;
+	int ret = -EACCES;
+
+	if (!capable(CAP_SYS_ADMIN))
+		goto out;
+
+	dma_addr = 0;
+	page = dma_alloc_coherent(iSeries_vio_dev, count, &dma_addr,
+			GFP_ATOMIC);
+	ret = -ENOMEM;
+	if (page == NULL)
+		goto out;
+
+	ret = -EFAULT;
+	if (copy_from_user(page, buffer, count))
+		goto out_free;
+
+	memset(&vsp_cmd, 0, sizeof(vsp_cmd));
+	vsp_cmd.cmd = 31;
+	vsp_cmd.sub_data.kern.token = dma_addr;
+	vsp_cmd.sub_data.kern.address_type = HvLpDma_AddressType_TceIndex;
+	vsp_cmd.sub_data.kern.side = (u64)data;
+	vsp_cmd.sub_data.kern.length = count;
+	mb();
+	(void)signal_vsp_instruction(&vsp_cmd);
+	ret = count;
+
+out_free:
+	dma_free_coherent(iSeries_vio_dev, count, page, dma_addr);
+out:
+	return ret;
+}
+
+static ssize_t proc_mf_change_vmlinux(struct file *file,
+				      const char __user *buf,
+				      size_t count, loff_t *ppos)
+{
+	struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
+	ssize_t rc;
+	dma_addr_t dma_addr;
+	char *page;
+	struct vsp_cmd_data vsp_cmd;
+
+	rc = -EACCES;
+	if (!capable(CAP_SYS_ADMIN))
+		goto out;
+
+	dma_addr = 0;
+	page = dma_alloc_coherent(iSeries_vio_dev, count, &dma_addr,
+			GFP_ATOMIC);
+	rc = -ENOMEM;
+	if (page == NULL) {
+		printk(KERN_ERR "mf.c: couldn't allocate memory to set vmlinux chunk\n");
+		goto out;
+	}
+	rc = -EFAULT;
+	if (copy_from_user(page, buf, count))
+		goto out_free;
+
+	memset(&vsp_cmd, 0, sizeof(vsp_cmd));
+	vsp_cmd.cmd = 30;
+	vsp_cmd.sub_data.kern.token = dma_addr;
+	vsp_cmd.sub_data.kern.address_type = HvLpDma_AddressType_TceIndex;
+	vsp_cmd.sub_data.kern.side = (u64)dp->data;
+	vsp_cmd.sub_data.kern.offset = *ppos;
+	vsp_cmd.sub_data.kern.length = count;
+	mb();
+	rc = signal_vsp_instruction(&vsp_cmd);
+	if (rc)
+		goto out_free;
+	rc = -ENOMEM;
+	if (vsp_cmd.result_code != 0)
+		goto out_free;
+
+	*ppos += count;
+	rc = count;
+out_free:
+	dma_free_coherent(iSeries_vio_dev, count, page, dma_addr);
+out:
+	return rc;
+}
+
+static struct file_operations proc_vmlinux_operations = {
+	.write		= proc_mf_change_vmlinux,
+};
+
+static int __init mf_proc_init(void)
+{
+	struct proc_dir_entry *mf_proc_root;
+	struct proc_dir_entry *ent;
+	struct proc_dir_entry *mf;
+	char name[2];
+	int i;
+
+	mf_proc_root = proc_mkdir("iSeries/mf", NULL);
+	if (!mf_proc_root)
+		return 1;
+
+	name[1] = '\0';
+	for (i = 0; i < 4; i++) {
+		name[0] = 'A' + i;
+		mf = proc_mkdir(name, mf_proc_root);
+		if (!mf)
+			return 1;
+
+		ent = create_proc_entry("cmdline", S_IFREG|S_IRUSR|S_IWUSR, mf);
+		if (!ent)
+			return 1;
+		ent->nlink = 1;
+		ent->data = (void *)(long)i;
+		ent->read_proc = proc_mf_dump_cmdline;
+		ent->write_proc = proc_mf_change_cmdline;
+
+		if (i == 3)	/* no vmlinux entry for 'D' */
+			continue;
+
+		ent = create_proc_entry("vmlinux", S_IFREG|S_IWUSR, mf);
+		if (!ent)
+			return 1;
+		ent->nlink = 1;
+		ent->data = (void *)(long)i;
+		ent->proc_fops = &proc_vmlinux_operations;
+	}
+
+	ent = create_proc_entry("side", S_IFREG|S_IRUSR|S_IWUSR, mf_proc_root);
+	if (!ent)
+		return 1;
+	ent->nlink = 1;
+	ent->data = (void *)0;
+	ent->read_proc = proc_mf_dump_side;
+	ent->write_proc = proc_mf_change_side;
+
+	ent = create_proc_entry("src", S_IFREG|S_IRUSR|S_IWUSR, mf_proc_root);
+	if (!ent)
+		return 1;
+	ent->nlink = 1;
+	ent->data = (void *)0;
+	ent->read_proc = proc_mf_dump_src;
+	ent->write_proc = proc_mf_change_src;
+
+	return 0;
+}
+
+__initcall(mf_proc_init);
+
+#endif /* CONFIG_PROC_FS */
+
+/*
+ * Get the RTC from the virtual service processor
+ * This requires flowing LpEvents to the primary partition
+ */
+void iSeries_get_rtc_time(struct rtc_time *rtc_tm)
+{
+	if (piranha_simulator)
+		return;
+
+	mf_get_rtc(rtc_tm);
+	rtc_tm->tm_mon--;
+}
+
+/*
+ * Set the RTC in the virtual service processor
+ * This requires flowing LpEvents to the primary partition
+ */
+int iSeries_set_rtc_time(struct rtc_time *tm)
+{
+	mf_set_rtc(tm);
+	return 0;
+}
+
+unsigned long iSeries_get_boot_time(void)
+{
+	struct rtc_time tm;
+
+	if (piranha_simulator)
+		return 0;
+
+	mf_get_boot_rtc(&tm);
+	return mktime(tm.tm_year + 1900, tm.tm_mon, tm.tm_mday,
+		      tm.tm_hour, tm.tm_min, tm.tm_sec);
+}
diff --git a/arch/powerpc/platforms/iseries/misc.S b/arch/powerpc/platforms/iseries/misc.S
new file mode 100644
index 000000000000..09f14522e176
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/misc.S
@@ -0,0 +1,55 @@
+/*
+ * This file contains miscellaneous low-level functions.
+ *    Copyright (C) 1995-2005 IBM Corp
+ *
+ * Largely rewritten by Cort Dougan (cort@cs.nmt.edu)
+ * and Paul Mackerras.
+ * Adapted for iSeries by Mike Corrigan (mikejc@us.ibm.com)
+ * PPC64 updates by Dave Engebretsen (engebret@us.ibm.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <asm/processor.h>
+#include <asm/asm-offsets.h>
+
+	.text
+
+/* unsigned long local_save_flags(void) */
+_GLOBAL(local_get_flags)
+	lbz	r3,PACAPROCENABLED(r13)
+	blr
+
+/* unsigned long local_irq_disable(void) */
+_GLOBAL(local_irq_disable)
+	lbz	r3,PACAPROCENABLED(r13)
+	li	r4,0
+	stb	r4,PACAPROCENABLED(r13)
+	blr			/* Done */
+
+/* void local_irq_restore(unsigned long flags) */
+_GLOBAL(local_irq_restore)
+	lbz	r5,PACAPROCENABLED(r13)
+	 /* Check if things are setup the way we want _already_. */
+	cmpw	0,r3,r5
+	beqlr
+	/* are we enabling interrupts? */
+	cmpdi	0,r3,0
+	stb	r3,PACAPROCENABLED(r13)
+	beqlr
+	/* Check pending interrupts */
+	/*   A decrementer, IPI or PMC interrupt may have occurred
+	 *   while we were in the hypervisor (which enables) */
+	ld	r4,PACALPPACA+LPPACAANYINT(r13)
+	cmpdi	r4,0
+	beqlr
+
+	/*
+	 * Handle pending interrupts in interrupt context
+	 */
+	li	r0,0x5555
+	sc
+	blr
diff --git a/arch/powerpc/platforms/iseries/naca.h b/arch/powerpc/platforms/iseries/naca.h
new file mode 100644
index 000000000000..ab2372eb8d2e
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/naca.h
@@ -0,0 +1,24 @@
+#ifndef _PLATFORMS_ISERIES_NACA_H
+#define _PLATFORMS_ISERIES_NACA_H
+
+/*
+ * c 2001 PPC 64 Team, IBM Corp
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <asm/types.h>
+
+struct naca_struct {
+	/* Kernel only data - undefined for user space */
+	void *xItVpdAreas;              /* VPD Data                  0x00 */
+	void *xRamDisk;                 /* iSeries ramdisk           0x08 */
+	u64   xRamDiskSize;		/* In pages                  0x10 */
+};
+
+extern struct naca_struct naca;
+
+#endif /* _PLATFORMS_ISERIES_NACA_H */
diff --git a/arch/powerpc/platforms/iseries/pci.c b/arch/powerpc/platforms/iseries/pci.c
new file mode 100644
index 000000000000..7d7d5884343f
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/pci.c
@@ -0,0 +1,906 @@
+/*
+ * Copyright (C) 2001 Allan Trautman, IBM Corporation
+ *
+ * iSeries specific routines for PCI.
+ *
+ * Based on code from pci.c and iSeries_pci.c 32bit
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/ide.h>
+#include <linux/pci.h>
+
+#include <asm/io.h>
+#include <asm/irq.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+#include <asm/pci-bridge.h>
+#include <asm/ppcdebug.h>
+#include <asm/iommu.h>
+#include <asm/abs_addr.h>
+
+#include <asm/iseries/hv_call_xm.h>
+#include <asm/iseries/mf.h>
+
+#include <asm/ppc-pci.h>
+
+#include "irq.h"
+#include "pci.h"
+#include "call_pci.h"
+
+extern unsigned long io_page_mask;
+
+/*
+ * Forward declares of prototypes.
+ */
+static struct device_node *find_Device_Node(int bus, int devfn);
+static void scan_PHB_slots(struct pci_controller *Phb);
+static void scan_EADS_bridge(HvBusNumber Bus, HvSubBusNumber SubBus, int IdSel);
+static int scan_bridge_slot(HvBusNumber Bus, struct HvCallPci_BridgeInfo *Info);
+
+LIST_HEAD(iSeries_Global_Device_List);
+
+static int DeviceCount;
+
+/* Counters and control flags. */
+static long Pci_Io_Read_Count;
+static long Pci_Io_Write_Count;
+#if 0
+static long Pci_Cfg_Read_Count;
+static long Pci_Cfg_Write_Count;
+#endif
+static long Pci_Error_Count;
+
+static int Pci_Retry_Max = 3;	/* Only retry 3 times  */
+static int Pci_Error_Flag = 1;	/* Set Retry Error on. */
+
+static struct pci_ops iSeries_pci_ops;
+
+/*
+ * Table defines
+ * Each Entry size is 4 MB * 1024 Entries = 4GB I/O address space.
+ */
+#define IOMM_TABLE_MAX_ENTRIES	1024
+#define IOMM_TABLE_ENTRY_SIZE	0x0000000000400000UL
+#define BASE_IO_MEMORY		0xE000000000000000UL
+
+static unsigned long max_io_memory = 0xE000000000000000UL;
+static long current_iomm_table_entry;
+
+/*
+ * Lookup Tables.
+ */
+static struct device_node **iomm_table;
+static u8 *iobar_table;
+
+/*
+ * Static and Global variables
+ */
+static char *pci_io_text = "iSeries PCI I/O";
+static DEFINE_SPINLOCK(iomm_table_lock);
+
+/*
+ * iomm_table_initialize
+ *
+ * Allocates and initalizes the Address Translation Table and Bar
+ * Tables to get them ready for use.  Must be called before any
+ * I/O space is handed out to the device BARs.
+ */
+static void iomm_table_initialize(void)
+{
+	spin_lock(&iomm_table_lock);
+	iomm_table = kmalloc(sizeof(*iomm_table) * IOMM_TABLE_MAX_ENTRIES,
+			GFP_KERNEL);
+	iobar_table = kmalloc(sizeof(*iobar_table) * IOMM_TABLE_MAX_ENTRIES,
+			GFP_KERNEL);
+	spin_unlock(&iomm_table_lock);
+	if ((iomm_table == NULL) || (iobar_table == NULL))
+		panic("PCI: I/O tables allocation failed.\n");
+}
+
+/*
+ * iomm_table_allocate_entry
+ *
+ * Adds pci_dev entry in address translation table
+ *
+ * - Allocates the number of entries required in table base on BAR
+ *   size.
+ * - Allocates starting at BASE_IO_MEMORY and increases.
+ * - The size is round up to be a multiple of entry size.
+ * - CurrentIndex is incremented to keep track of the last entry.
+ * - Builds the resource entry for allocated BARs.
+ */
+static void iomm_table_allocate_entry(struct pci_dev *dev, int bar_num)
+{
+	struct resource *bar_res = &dev->resource[bar_num];
+	long bar_size = pci_resource_len(dev, bar_num);
+
+	/*
+	 * No space to allocate, quick exit, skip Allocation.
+	 */
+	if (bar_size == 0)
+		return;
+	/*
+	 * Set Resource values.
+	 */
+	spin_lock(&iomm_table_lock);
+	bar_res->name = pci_io_text;
+	bar_res->start =
+		IOMM_TABLE_ENTRY_SIZE * current_iomm_table_entry;
+	bar_res->start += BASE_IO_MEMORY;
+	bar_res->end = bar_res->start + bar_size - 1;
+	/*
+	 * Allocate the number of table entries needed for BAR.
+	 */
+	while (bar_size > 0 ) {
+		iomm_table[current_iomm_table_entry] = dev->sysdata;
+		iobar_table[current_iomm_table_entry] = bar_num;
+		bar_size -= IOMM_TABLE_ENTRY_SIZE;
+		++current_iomm_table_entry;
+	}
+	max_io_memory = BASE_IO_MEMORY +
+		(IOMM_TABLE_ENTRY_SIZE * current_iomm_table_entry);
+	spin_unlock(&iomm_table_lock);
+}
+
+/*
+ * allocate_device_bars
+ *
+ * - Allocates ALL pci_dev BAR's and updates the resources with the
+ *   BAR value.  BARS with zero length will have the resources
+ *   The HvCallPci_getBarParms is used to get the size of the BAR
+ *   space.  It calls iomm_table_allocate_entry to allocate
+ *   each entry.
+ * - Loops through The Bar resources(0 - 5) including the ROM
+ *   is resource(6).
+ */
+static void allocate_device_bars(struct pci_dev *dev)
+{
+	struct resource *bar_res;
+	int bar_num;
+
+	for (bar_num = 0; bar_num <= PCI_ROM_RESOURCE; ++bar_num) {
+		bar_res = &dev->resource[bar_num];
+		iomm_table_allocate_entry(dev, bar_num);
+	}
+}
+
+/*
+ * Log error information to system console.
+ * Filter out the device not there errors.
+ * PCI: EADs Connect Failed 0x18.58.10 Rc: 0x00xx
+ * PCI: Read Vendor Failed 0x18.58.10 Rc: 0x00xx
+ * PCI: Connect Bus Unit Failed 0x18.58.10 Rc: 0x00xx
+ */
+static void pci_Log_Error(char *Error_Text, int Bus, int SubBus,
+		int AgentId, int HvRc)
+{
+	if (HvRc == 0x0302)
+		return;
+	printk(KERN_ERR "PCI: %s Failed: 0x%02X.%02X.%02X Rc: 0x%04X",
+	       Error_Text, Bus, SubBus, AgentId, HvRc);
+}
+
+/*
+ * build_device_node(u16 Bus, int SubBus, u8 DevFn)
+ */
+static struct device_node *build_device_node(HvBusNumber Bus,
+		HvSubBusNumber SubBus, int AgentId, int Function)
+{
+	struct device_node *node;
+	struct pci_dn *pdn;
+
+	PPCDBG(PPCDBG_BUSWALK,
+			"-build_device_node 0x%02X.%02X.%02X Function: %02X\n",
+			Bus, SubBus, AgentId, Function);
+
+	node = kmalloc(sizeof(struct device_node), GFP_KERNEL);
+	if (node == NULL)
+		return NULL;
+	memset(node, 0, sizeof(struct device_node));
+	pdn = kzalloc(sizeof(*pdn), GFP_KERNEL);
+	if (pdn == NULL) {
+		kfree(node);
+		return NULL;
+	}
+	node->data = pdn;
+	pdn->node = node;
+	list_add_tail(&pdn->Device_List, &iSeries_Global_Device_List);
+	pdn->busno = Bus;
+	pdn->bussubno = SubBus;
+	pdn->devfn = PCI_DEVFN(ISERIES_ENCODE_DEVICE(AgentId), Function);
+	return node;
+}
+
+/*
+ * unsigned long __init find_and_init_phbs(void)
+ *
+ * Description:
+ *   This function checks for all possible system PCI host bridges that connect
+ *   PCI buses.  The system hypervisor is queried as to the guest partition
+ *   ownership status.  A pci_controller is built for any bus which is partially
+ *   owned or fully owned by this guest partition.
+ */
+unsigned long __init find_and_init_phbs(void)
+{
+	struct pci_controller *phb;
+	HvBusNumber bus;
+
+	PPCDBG(PPCDBG_BUSWALK, "find_and_init_phbs Entry\n");
+
+	/* Check all possible buses. */
+	for (bus = 0; bus < 256; bus++) {
+		int ret = HvCallXm_testBus(bus);
+		if (ret == 0) {
+			printk("bus %d appears to exist\n", bus);
+
+			phb = (struct pci_controller *)kmalloc(sizeof(struct pci_controller), GFP_KERNEL);
+			if (phb == NULL)
+				return -ENOMEM;
+			pci_setup_pci_controller(phb);
+
+			phb->pci_mem_offset = phb->local_number = bus;
+			phb->first_busno = bus;
+			phb->last_busno = bus;
+			phb->ops = &iSeries_pci_ops;
+
+			PPCDBG(PPCDBG_BUSWALK, "PCI:Create iSeries pci_controller(%p), Bus: %04X\n",
+					phb, bus);
+
+			/* Find and connect the devices. */
+			scan_PHB_slots(phb);
+		}
+		/*
+		 * Check for Unexpected Return code, a clue that something
+		 * has gone wrong.
+		 */
+		else if (ret != 0x0301)
+			printk(KERN_ERR "Unexpected Return on Probe(0x%04X): 0x%04X",
+			       bus, ret);
+	}
+	return 0;
+}
+
+/*
+ * iSeries_pcibios_init
+ *
+ * Chance to initialize and structures or variable before PCI Bus walk.
+ */
+void iSeries_pcibios_init(void)
+{
+	PPCDBG(PPCDBG_BUSWALK, "iSeries_pcibios_init Entry.\n");
+	iomm_table_initialize();
+	find_and_init_phbs();
+	io_page_mask = -1;
+	PPCDBG(PPCDBG_BUSWALK, "iSeries_pcibios_init Exit.\n");
+}
+
+/*
+ * iSeries_pci_final_fixup(void)
+ */
+void __init iSeries_pci_final_fixup(void)
+{
+	struct pci_dev *pdev = NULL;
+	struct device_node *node;
+	int DeviceCount = 0;
+
+	PPCDBG(PPCDBG_BUSWALK, "iSeries_pcibios_fixup Entry.\n");
+
+	/* Fix up at the device node and pci_dev relationship */
+	mf_display_src(0xC9000100);
+
+	printk("pcibios_final_fixup\n");
+	for_each_pci_dev(pdev) {
+		node = find_Device_Node(pdev->bus->number, pdev->devfn);
+		printk("pci dev %p (%x.%x), node %p\n", pdev,
+		       pdev->bus->number, pdev->devfn, node);
+
+		if (node != NULL) {
+			++DeviceCount;
+			pdev->sysdata = (void *)node;
+			PCI_DN(node)->pcidev = pdev;
+			PPCDBG(PPCDBG_BUSWALK,
+					"pdev 0x%p <==> DevNode 0x%p\n",
+					pdev, node);
+			allocate_device_bars(pdev);
+			iSeries_Device_Information(pdev, DeviceCount);
+			iommu_devnode_init_iSeries(node);
+		} else
+			printk("PCI: Device Tree not found for 0x%016lX\n",
+					(unsigned long)pdev);
+		pdev->irq = PCI_DN(node)->Irq;
+	}
+	iSeries_activate_IRQs();
+	mf_display_src(0xC9000200);
+}
+
+void pcibios_fixup_bus(struct pci_bus *PciBus)
+{
+	PPCDBG(PPCDBG_BUSWALK, "iSeries_pcibios_fixup_bus(0x%04X) Entry.\n",
+			PciBus->number);
+}
+
+void pcibios_fixup_resources(struct pci_dev *pdev)
+{
+	PPCDBG(PPCDBG_BUSWALK, "fixup_resources pdev %p\n", pdev);
+}
+
+/*
+ * Loop through each node function to find usable EADs bridges.
+ */
+static void scan_PHB_slots(struct pci_controller *Phb)
+{
+	struct HvCallPci_DeviceInfo *DevInfo;
+	HvBusNumber bus = Phb->local_number;	/* System Bus */
+	const HvSubBusNumber SubBus = 0;	/* EADs is always 0. */
+	int HvRc = 0;
+	int IdSel;
+	const int MaxAgents = 8;
+
+	DevInfo = (struct HvCallPci_DeviceInfo*)
+		kmalloc(sizeof(struct HvCallPci_DeviceInfo), GFP_KERNEL);
+	if (DevInfo == NULL)
+		return;
+
+	/*
+	 * Probe for EADs Bridges
+	 */
+	for (IdSel = 1; IdSel < MaxAgents; ++IdSel) {
+		HvRc = HvCallPci_getDeviceInfo(bus, SubBus, IdSel,
+				iseries_hv_addr(DevInfo),
+				sizeof(struct HvCallPci_DeviceInfo));
+		if (HvRc == 0) {
+			if (DevInfo->deviceType == HvCallPci_NodeDevice)
+				scan_EADS_bridge(bus, SubBus, IdSel);
+			else
+				printk("PCI: Invalid System Configuration(0x%02X)"
+				       " for bus 0x%02x id 0x%02x.\n",
+				       DevInfo->deviceType, bus, IdSel);
+		}
+		else
+			pci_Log_Error("getDeviceInfo", bus, SubBus, IdSel, HvRc);
+	}
+	kfree(DevInfo);
+}
+
+static void scan_EADS_bridge(HvBusNumber bus, HvSubBusNumber SubBus,
+		int IdSel)
+{
+	struct HvCallPci_BridgeInfo *BridgeInfo;
+	HvAgentId AgentId;
+	int Function;
+	int HvRc;
+
+	BridgeInfo = (struct HvCallPci_BridgeInfo *)
+		kmalloc(sizeof(struct HvCallPci_BridgeInfo), GFP_KERNEL);
+	if (BridgeInfo == NULL)
+		return;
+
+	/* Note: hvSubBus and irq is always be 0 at this level! */
+	for (Function = 0; Function < 8; ++Function) {
+		AgentId = ISERIES_PCI_AGENTID(IdSel, Function);
+		HvRc = HvCallXm_connectBusUnit(bus, SubBus, AgentId, 0);
+		if (HvRc == 0) {
+			printk("found device at bus %d idsel %d func %d (AgentId %x)\n",
+			       bus, IdSel, Function, AgentId);
+			/*  Connect EADs: 0x18.00.12 = 0x00 */
+			PPCDBG(PPCDBG_BUSWALK,
+					"PCI:Connect EADs: 0x%02X.%02X.%02X\n",
+					bus, SubBus, AgentId);
+			HvRc = HvCallPci_getBusUnitInfo(bus, SubBus, AgentId,
+					iseries_hv_addr(BridgeInfo),
+					sizeof(struct HvCallPci_BridgeInfo));
+			if (HvRc == 0) {
+				printk("bridge info: type %x subbus %x maxAgents %x maxsubbus %x logslot %x\n",
+					BridgeInfo->busUnitInfo.deviceType,
+					BridgeInfo->subBusNumber,
+					BridgeInfo->maxAgents,
+					BridgeInfo->maxSubBusNumber,
+					BridgeInfo->logicalSlotNumber);
+				PPCDBG(PPCDBG_BUSWALK,
+					"PCI: BridgeInfo, Type:0x%02X, SubBus:0x%02X, MaxAgents:0x%02X, MaxSubBus: 0x%02X, LSlot: 0x%02X\n",
+					BridgeInfo->busUnitInfo.deviceType,
+					BridgeInfo->subBusNumber,
+					BridgeInfo->maxAgents,
+					BridgeInfo->maxSubBusNumber,
+					BridgeInfo->logicalSlotNumber);
+
+				if (BridgeInfo->busUnitInfo.deviceType ==
+						HvCallPci_BridgeDevice)  {
+					/* Scan_Bridge_Slot...: 0x18.00.12 */
+					scan_bridge_slot(bus, BridgeInfo);
+				} else
+					printk("PCI: Invalid Bridge Configuration(0x%02X)",
+						BridgeInfo->busUnitInfo.deviceType);
+			}
+		} else if (HvRc != 0x000B)
+			pci_Log_Error("EADs Connect",
+					bus, SubBus, AgentId, HvRc);
+	}
+	kfree(BridgeInfo);
+}
+
+/*
+ * This assumes that the node slot is always on the primary bus!
+ */
+static int scan_bridge_slot(HvBusNumber Bus,
+		struct HvCallPci_BridgeInfo *BridgeInfo)
+{
+	struct device_node *node;
+	HvSubBusNumber SubBus = BridgeInfo->subBusNumber;
+	u16 VendorId = 0;
+	int HvRc = 0;
+	u8 Irq = 0;
+	int IdSel = ISERIES_GET_DEVICE_FROM_SUBBUS(SubBus);
+	int Function = ISERIES_GET_FUNCTION_FROM_SUBBUS(SubBus);
+	HvAgentId EADsIdSel = ISERIES_PCI_AGENTID(IdSel, Function);
+
+	/* iSeries_allocate_IRQ.: 0x18.00.12(0xA3) */
+	Irq = iSeries_allocate_IRQ(Bus, 0, EADsIdSel);
+	PPCDBG(PPCDBG_BUSWALK,
+		"PCI:- allocate and assign IRQ 0x%02X.%02X.%02X = 0x%02X\n",
+		Bus, 0, EADsIdSel, Irq);
+
+	/*
+	 * Connect all functions of any device found.
+	 */
+	for (IdSel = 1; IdSel <= BridgeInfo->maxAgents; ++IdSel) {
+		for (Function = 0; Function < 8; ++Function) {
+			HvAgentId AgentId = ISERIES_PCI_AGENTID(IdSel, Function);
+			HvRc = HvCallXm_connectBusUnit(Bus, SubBus,
+					AgentId, Irq);
+			if (HvRc != 0) {
+				pci_Log_Error("Connect Bus Unit",
+					      Bus, SubBus, AgentId, HvRc);
+				continue;
+			}
+
+			HvRc = HvCallPci_configLoad16(Bus, SubBus, AgentId,
+						      PCI_VENDOR_ID, &VendorId);
+			if (HvRc != 0) {
+				pci_Log_Error("Read Vendor",
+					      Bus, SubBus, AgentId, HvRc);
+				continue;
+			}
+			printk("read vendor ID: %x\n", VendorId);
+
+			/* FoundDevice: 0x18.28.10 = 0x12AE */
+			PPCDBG(PPCDBG_BUSWALK,
+			       "PCI:- FoundDevice: 0x%02X.%02X.%02X = 0x%04X, irq %d\n",
+			       Bus, SubBus, AgentId, VendorId, Irq);
+			HvRc = HvCallPci_configStore8(Bus, SubBus, AgentId,
+						      PCI_INTERRUPT_LINE, Irq);
+			if (HvRc != 0)
+				pci_Log_Error("PciCfgStore Irq Failed!",
+					      Bus, SubBus, AgentId, HvRc);
+
+			++DeviceCount;
+			node = build_device_node(Bus, SubBus, EADsIdSel, Function);
+			PCI_DN(node)->Irq = Irq;
+			PCI_DN(node)->LogicalSlot = BridgeInfo->logicalSlotNumber;
+
+		} /* for (Function = 0; Function < 8; ++Function) */
+	} /* for (IdSel = 1; IdSel <= MaxAgents; ++IdSel) */
+	return HvRc;
+}
+
+/*
+ * I/0 Memory copy MUST use mmio commands on iSeries
+ * To do; For performance, include the hv call directly
+ */
+void iSeries_memset_io(volatile void __iomem *dest, char c, size_t Count)
+{
+	u8 ByteValue = c;
+	long NumberOfBytes = Count;
+
+	while (NumberOfBytes > 0) {
+		iSeries_Write_Byte(ByteValue, dest++);
+		-- NumberOfBytes;
+	}
+}
+EXPORT_SYMBOL(iSeries_memset_io);
+
+void iSeries_memcpy_toio(volatile void __iomem *dest, void *source, size_t count)
+{
+	char *src = source;
+	long NumberOfBytes = count;
+
+	while (NumberOfBytes > 0) {
+		iSeries_Write_Byte(*src++, dest++);
+		-- NumberOfBytes;
+	}
+}
+EXPORT_SYMBOL(iSeries_memcpy_toio);
+
+void iSeries_memcpy_fromio(void *dest, const volatile void __iomem *src, size_t count)
+{
+	char *dst = dest;
+	long NumberOfBytes = count;
+
+	while (NumberOfBytes > 0) {
+		*dst++ = iSeries_Read_Byte(src++);
+		-- NumberOfBytes;
+	}
+}
+EXPORT_SYMBOL(iSeries_memcpy_fromio);
+
+/*
+ * Look down the chain to find the matching Device Device
+ */
+static struct device_node *find_Device_Node(int bus, int devfn)
+{
+	struct pci_dn *pdn;
+
+	list_for_each_entry(pdn, &iSeries_Global_Device_List, Device_List) {
+		if ((bus == pdn->busno) && (devfn == pdn->devfn))
+			return pdn->node;
+	}
+	return NULL;
+}
+
+#if 0
+/*
+ * Returns the device node for the passed pci_dev
+ * Sanity Check Node PciDev to passed pci_dev
+ * If none is found, returns a NULL which the client must handle.
+ */
+static struct device_node *get_Device_Node(struct pci_dev *pdev)
+{
+	struct device_node *node;
+
+	node = pdev->sysdata;
+	if (node == NULL || PCI_DN(node)->pcidev != pdev)
+		node = find_Device_Node(pdev->bus->number, pdev->devfn);
+	return node;
+}
+#endif
+
+/*
+ * Config space read and write functions.
+ * For now at least, we look for the device node for the bus and devfn
+ * that we are asked to access.  It may be possible to translate the devfn
+ * to a subbus and deviceid more directly.
+ */
+static u64 hv_cfg_read_func[4]  = {
+	HvCallPciConfigLoad8, HvCallPciConfigLoad16,
+	HvCallPciConfigLoad32, HvCallPciConfigLoad32
+};
+
+static u64 hv_cfg_write_func[4] = {
+	HvCallPciConfigStore8, HvCallPciConfigStore16,
+	HvCallPciConfigStore32, HvCallPciConfigStore32
+};
+
+/*
+ * Read PCI config space
+ */
+static int iSeries_pci_read_config(struct pci_bus *bus, unsigned int devfn,
+		int offset, int size, u32 *val)
+{
+	struct device_node *node = find_Device_Node(bus->number, devfn);
+	u64 fn;
+	struct HvCallPci_LoadReturn ret;
+
+	if (node == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	if (offset > 255) {
+		*val = ~0;
+		return PCIBIOS_BAD_REGISTER_NUMBER;
+	}
+
+	fn = hv_cfg_read_func[(size - 1) & 3];
+	HvCall3Ret16(fn, &ret, iseries_ds_addr(node), offset, 0);
+
+	if (ret.rc != 0) {
+		*val = ~0;
+		return PCIBIOS_DEVICE_NOT_FOUND;	/* or something */
+	}
+
+	*val = ret.value;
+	return 0;
+}
+
+/*
+ * Write PCI config space
+ */
+
+static int iSeries_pci_write_config(struct pci_bus *bus, unsigned int devfn,
+		int offset, int size, u32 val)
+{
+	struct device_node *node = find_Device_Node(bus->number, devfn);
+	u64 fn;
+	u64 ret;
+
+	if (node == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	if (offset > 255)
+		return PCIBIOS_BAD_REGISTER_NUMBER;
+
+	fn = hv_cfg_write_func[(size - 1) & 3];
+	ret = HvCall4(fn, iseries_ds_addr(node), offset, val, 0);
+
+	if (ret != 0)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	return 0;
+}
+
+static struct pci_ops iSeries_pci_ops = {
+	.read = iSeries_pci_read_config,
+	.write = iSeries_pci_write_config
+};
+
+/*
+ * Check Return Code
+ * -> On Failure, print and log information.
+ *    Increment Retry Count, if exceeds max, panic partition.
+ *
+ * PCI: Device 23.90 ReadL I/O Error( 0): 0x1234
+ * PCI: Device 23.90 ReadL Retry( 1)
+ * PCI: Device 23.90 ReadL Retry Successful(1)
+ */
+static int CheckReturnCode(char *TextHdr, struct device_node *DevNode,
+		int *retry, u64 ret)
+{
+	if (ret != 0)  {
+		struct pci_dn *pdn = PCI_DN(DevNode);
+
+		++Pci_Error_Count;
+		(*retry)++;
+		printk("PCI: %s: Device 0x%04X:%02X  I/O Error(%2d): 0x%04X\n",
+				TextHdr, pdn->busno, pdn->devfn,
+				*retry, (int)ret);
+		/*
+		 * Bump the retry and check for retry count exceeded.
+		 * If, Exceeded, panic the system.
+		 */
+		if (((*retry) > Pci_Retry_Max) &&
+				(Pci_Error_Flag > 0)) {
+			mf_display_src(0xB6000103);
+			panic_timeout = 0;
+			panic("PCI: Hardware I/O Error, SRC B6000103, "
+					"Automatic Reboot Disabled.\n");
+		}
+		return -1;	/* Retry Try */
+	}
+	return 0;
+}
+
+/*
+ * Translate the I/O Address into a device node, bar, and bar offset.
+ * Note: Make sure the passed variable end up on the stack to avoid
+ * the exposure of being device global.
+ */
+static inline struct device_node *xlate_iomm_address(
+		const volatile void __iomem *IoAddress,
+		u64 *dsaptr, u64 *BarOffsetPtr)
+{
+	unsigned long OrigIoAddr;
+	unsigned long BaseIoAddr;
+	unsigned long TableIndex;
+	struct device_node *DevNode;
+
+	OrigIoAddr = (unsigned long __force)IoAddress;
+	if ((OrigIoAddr < BASE_IO_MEMORY) || (OrigIoAddr >= max_io_memory))
+		return NULL;
+	BaseIoAddr = OrigIoAddr - BASE_IO_MEMORY;
+	TableIndex = BaseIoAddr / IOMM_TABLE_ENTRY_SIZE;
+	DevNode = iomm_table[TableIndex];
+
+	if (DevNode != NULL) {
+		int barnum = iobar_table[TableIndex];
+		*dsaptr = iseries_ds_addr(DevNode) | (barnum << 24);
+		*BarOffsetPtr = BaseIoAddr % IOMM_TABLE_ENTRY_SIZE;
+	} else
+		panic("PCI: Invalid PCI IoAddress detected!\n");
+	return DevNode;
+}
+
+/*
+ * Read MM I/O Instructions for the iSeries
+ * On MM I/O error, all ones are returned and iSeries_pci_IoError is cal
+ * else, data is returned in big Endian format.
+ *
+ * iSeries_Read_Byte = Read Byte  ( 8 bit)
+ * iSeries_Read_Word = Read Word  (16 bit)
+ * iSeries_Read_Long = Read Long  (32 bit)
+ */
+u8 iSeries_Read_Byte(const volatile void __iomem *IoAddress)
+{
+	u64 BarOffset;
+	u64 dsa;
+	int retry = 0;
+	struct HvCallPci_LoadReturn ret;
+	struct device_node *DevNode =
+		xlate_iomm_address(IoAddress, &dsa, &BarOffset);
+
+	if (DevNode == NULL) {
+		static unsigned long last_jiffies;
+		static int num_printed;
+
+		if ((jiffies - last_jiffies) > 60 * HZ) {
+			last_jiffies = jiffies;
+			num_printed = 0;
+		}
+		if (num_printed++ < 10)
+			printk(KERN_ERR "iSeries_Read_Byte: invalid access at IO address %p\n", IoAddress);
+		return 0xff;
+	}
+	do {
+		++Pci_Io_Read_Count;
+		HvCall3Ret16(HvCallPciBarLoad8, &ret, dsa, BarOffset, 0);
+	} while (CheckReturnCode("RDB", DevNode, &retry, ret.rc) != 0);
+
+	return (u8)ret.value;
+}
+EXPORT_SYMBOL(iSeries_Read_Byte);
+
+u16 iSeries_Read_Word(const volatile void __iomem *IoAddress)
+{
+	u64 BarOffset;
+	u64 dsa;
+	int retry = 0;
+	struct HvCallPci_LoadReturn ret;
+	struct device_node *DevNode =
+		xlate_iomm_address(IoAddress, &dsa, &BarOffset);
+
+	if (DevNode == NULL) {
+		static unsigned long last_jiffies;
+		static int num_printed;
+
+		if ((jiffies - last_jiffies) > 60 * HZ) {
+			last_jiffies = jiffies;
+			num_printed = 0;
+		}
+		if (num_printed++ < 10)
+			printk(KERN_ERR "iSeries_Read_Word: invalid access at IO address %p\n", IoAddress);
+		return 0xffff;
+	}
+	do {
+		++Pci_Io_Read_Count;
+		HvCall3Ret16(HvCallPciBarLoad16, &ret, dsa,
+				BarOffset, 0);
+	} while (CheckReturnCode("RDW", DevNode, &retry, ret.rc) != 0);
+
+	return swab16((u16)ret.value);
+}
+EXPORT_SYMBOL(iSeries_Read_Word);
+
+u32 iSeries_Read_Long(const volatile void __iomem *IoAddress)
+{
+	u64 BarOffset;
+	u64 dsa;
+	int retry = 0;
+	struct HvCallPci_LoadReturn ret;
+	struct device_node *DevNode =
+		xlate_iomm_address(IoAddress, &dsa, &BarOffset);
+
+	if (DevNode == NULL) {
+		static unsigned long last_jiffies;
+		static int num_printed;
+
+		if ((jiffies - last_jiffies) > 60 * HZ) {
+			last_jiffies = jiffies;
+			num_printed = 0;
+		}
+		if (num_printed++ < 10)
+			printk(KERN_ERR "iSeries_Read_Long: invalid access at IO address %p\n", IoAddress);
+		return 0xffffffff;
+	}
+	do {
+		++Pci_Io_Read_Count;
+		HvCall3Ret16(HvCallPciBarLoad32, &ret, dsa,
+				BarOffset, 0);
+	} while (CheckReturnCode("RDL", DevNode, &retry, ret.rc) != 0);
+
+	return swab32((u32)ret.value);
+}
+EXPORT_SYMBOL(iSeries_Read_Long);
+
+/*
+ * Write MM I/O Instructions for the iSeries
+ *
+ * iSeries_Write_Byte = Write Byte (8 bit)
+ * iSeries_Write_Word = Write Word(16 bit)
+ * iSeries_Write_Long = Write Long(32 bit)
+ */
+void iSeries_Write_Byte(u8 data, volatile void __iomem *IoAddress)
+{
+	u64 BarOffset;
+	u64 dsa;
+	int retry = 0;
+	u64 rc;
+	struct device_node *DevNode =
+		xlate_iomm_address(IoAddress, &dsa, &BarOffset);
+
+	if (DevNode == NULL) {
+		static unsigned long last_jiffies;
+		static int num_printed;
+
+		if ((jiffies - last_jiffies) > 60 * HZ) {
+			last_jiffies = jiffies;
+			num_printed = 0;
+		}
+		if (num_printed++ < 10)
+			printk(KERN_ERR "iSeries_Write_Byte: invalid access at IO address %p\n", IoAddress);
+		return;
+	}
+	do {
+		++Pci_Io_Write_Count;
+		rc = HvCall4(HvCallPciBarStore8, dsa, BarOffset, data, 0);
+	} while (CheckReturnCode("WWB", DevNode, &retry, rc) != 0);
+}
+EXPORT_SYMBOL(iSeries_Write_Byte);
+
+void iSeries_Write_Word(u16 data, volatile void __iomem *IoAddress)
+{
+	u64 BarOffset;
+	u64 dsa;
+	int retry = 0;
+	u64 rc;
+	struct device_node *DevNode =
+		xlate_iomm_address(IoAddress, &dsa, &BarOffset);
+
+	if (DevNode == NULL) {
+		static unsigned long last_jiffies;
+		static int num_printed;
+
+		if ((jiffies - last_jiffies) > 60 * HZ) {
+			last_jiffies = jiffies;
+			num_printed = 0;
+		}
+		if (num_printed++ < 10)
+			printk(KERN_ERR "iSeries_Write_Word: invalid access at IO address %p\n", IoAddress);
+		return;
+	}
+	do {
+		++Pci_Io_Write_Count;
+		rc = HvCall4(HvCallPciBarStore16, dsa, BarOffset, swab16(data), 0);
+	} while (CheckReturnCode("WWW", DevNode, &retry, rc) != 0);
+}
+EXPORT_SYMBOL(iSeries_Write_Word);
+
+void iSeries_Write_Long(u32 data, volatile void __iomem *IoAddress)
+{
+	u64 BarOffset;
+	u64 dsa;
+	int retry = 0;
+	u64 rc;
+	struct device_node *DevNode =
+		xlate_iomm_address(IoAddress, &dsa, &BarOffset);
+
+	if (DevNode == NULL) {
+		static unsigned long last_jiffies;
+		static int num_printed;
+
+		if ((jiffies - last_jiffies) > 60 * HZ) {
+			last_jiffies = jiffies;
+			num_printed = 0;
+		}
+		if (num_printed++ < 10)
+			printk(KERN_ERR "iSeries_Write_Long: invalid access at IO address %p\n", IoAddress);
+		return;
+	}
+	do {
+		++Pci_Io_Write_Count;
+		rc = HvCall4(HvCallPciBarStore32, dsa, BarOffset, swab32(data), 0);
+	} while (CheckReturnCode("WWL", DevNode, &retry, rc) != 0);
+}
+EXPORT_SYMBOL(iSeries_Write_Long);
diff --git a/arch/powerpc/platforms/iseries/pci.h b/arch/powerpc/platforms/iseries/pci.h
new file mode 100644
index 000000000000..33a8489fde54
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/pci.h
@@ -0,0 +1,63 @@
+#ifndef _PLATFORMS_ISERIES_PCI_H
+#define _PLATFORMS_ISERIES_PCI_H
+
+/*
+ * Created by Allan Trautman on Tue Feb 20, 2001.
+ *
+ * Define some useful macros for the iSeries pci routines.
+ * Copyright (C) 2001  Allan H Trautman, IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the:
+ * Free Software Foundation, Inc.,
+ * 59 Temple Place, Suite 330,
+ * Boston, MA  02111-1307  USA
+ *
+ * Change Activity:
+ *   Created Feb 20, 2001
+ *   Added device reset, March 22, 2001
+ *   Ported to ppc64, May 25, 2001
+ * End Change Activity
+ */
+
+#include <asm/pci-bridge.h>
+
+struct pci_dev;				/* For Forward Reference */
+
+/*
+ * Decodes Linux DevFn to iSeries DevFn, bridge device, or function.
+ * For Linux, see PCI_SLOT and PCI_FUNC in include/linux/pci.h
+ */
+
+#define ISERIES_PCI_AGENTID(idsel, func)	\
+	(((idsel & 0x0F) << 4) | (func & 0x07))
+#define ISERIES_ENCODE_DEVICE(agentid)		\
+	((0x10) | ((agentid & 0x20) >> 2) | (agentid & 0x07))
+
+#define ISERIES_GET_DEVICE_FROM_SUBBUS(subbus)		((subbus >> 5) & 0x7)
+#define ISERIES_GET_FUNCTION_FROM_SUBBUS(subbus)	((subbus >> 2) & 0x7)
+
+/*
+ * Generate a Direct Select Address for the Hypervisor
+ */
+static inline u64 iseries_ds_addr(struct device_node *node)
+{
+	struct pci_dn *pdn = PCI_DN(node);
+
+	return ((u64)pdn->busno << 48) + ((u64)pdn->bussubno << 40)
+			+ ((u64)0x10 << 32);
+}
+
+extern void	iSeries_Device_Information(struct pci_dev*, int);
+
+#endif /* _PLATFORMS_ISERIES_PCI_H */
diff --git a/arch/powerpc/platforms/iseries/proc.c b/arch/powerpc/platforms/iseries/proc.c
new file mode 100644
index 000000000000..e68b6b5fa89f
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/proc.c
@@ -0,0 +1,116 @@
+/*
+ * Copyright (C) 2001  Kyle A. Lucke IBM Corporation
+ * Copyright (C) 2001 Mike Corrigan & Dave Engebretsen IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#include <linux/init.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
+#include <linux/param.h>		/* for HZ */
+#include <asm/paca.h>
+#include <asm/processor.h>
+#include <asm/time.h>
+#include <asm/lppaca.h>
+#include <asm/iseries/it_lp_queue.h>
+#include <asm/iseries/hv_call_xm.h>
+
+#include "processor_vpd.h"
+#include "main_store.h"
+
+static int __init iseries_proc_create(void)
+{
+	struct proc_dir_entry *e = proc_mkdir("iSeries", 0);
+	if (!e)
+		return 1;
+
+	return 0;
+}
+core_initcall(iseries_proc_create);
+
+static unsigned long startTitan = 0;
+static unsigned long startTb = 0;
+
+static int proc_titantod_show(struct seq_file *m, void *v)
+{
+	unsigned long tb0, titan_tod;
+
+	tb0 = get_tb();
+	titan_tod = HvCallXm_loadTod();
+
+	seq_printf(m, "Titan\n" );
+	seq_printf(m, "  time base =          %016lx\n", tb0);
+	seq_printf(m, "  titan tod =          %016lx\n", titan_tod);
+	seq_printf(m, "  xProcFreq =          %016x\n",
+		   xIoHriProcessorVpd[0].xProcFreq);
+	seq_printf(m, "  xTimeBaseFreq =      %016x\n",
+		   xIoHriProcessorVpd[0].xTimeBaseFreq);
+	seq_printf(m, "  tb_ticks_per_jiffy = %lu\n", tb_ticks_per_jiffy);
+	seq_printf(m, "  tb_ticks_per_usec  = %lu\n", tb_ticks_per_usec);
+
+	if (!startTitan) {
+		startTitan = titan_tod;
+		startTb = tb0;
+	} else {
+		unsigned long titan_usec = (titan_tod - startTitan) >> 12;
+		unsigned long tb_ticks = (tb0 - startTb);
+		unsigned long titan_jiffies = titan_usec / (1000000/HZ);
+		unsigned long titan_jiff_usec = titan_jiffies * (1000000/HZ);
+		unsigned long titan_jiff_rem_usec =
+			titan_usec - titan_jiff_usec;
+		unsigned long tb_jiffies = tb_ticks / tb_ticks_per_jiffy;
+		unsigned long tb_jiff_ticks = tb_jiffies * tb_ticks_per_jiffy;
+		unsigned long tb_jiff_rem_ticks = tb_ticks - tb_jiff_ticks;
+		unsigned long tb_jiff_rem_usec =
+			tb_jiff_rem_ticks / tb_ticks_per_usec;
+		unsigned long new_tb_ticks_per_jiffy =
+			(tb_ticks * (1000000/HZ))/titan_usec;
+
+		seq_printf(m, "  titan elapsed = %lu uSec\n", titan_usec);
+		seq_printf(m, "  tb elapsed    = %lu ticks\n", tb_ticks);
+		seq_printf(m, "  titan jiffies = %lu.%04lu \n", titan_jiffies,
+			   titan_jiff_rem_usec);
+		seq_printf(m, "  tb jiffies    = %lu.%04lu\n", tb_jiffies,
+			   tb_jiff_rem_usec);
+		seq_printf(m, "  new tb_ticks_per_jiffy = %lu\n",
+			   new_tb_ticks_per_jiffy);
+	}
+
+	return 0;
+}
+
+static int proc_titantod_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, proc_titantod_show, NULL);
+}
+
+static struct file_operations proc_titantod_operations = {
+	.open		= proc_titantod_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static int __init iseries_proc_init(void)
+{
+	struct proc_dir_entry *e;
+
+	e = create_proc_entry("iSeries/titanTod", S_IFREG|S_IRUGO, NULL);
+	if (e)
+		e->proc_fops = &proc_titantod_operations;
+
+	return 0;
+}
+__initcall(iseries_proc_init);
diff --git a/arch/powerpc/platforms/iseries/processor_vpd.h b/arch/powerpc/platforms/iseries/processor_vpd.h
new file mode 100644
index 000000000000..7ac5d0d0dbfa
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/processor_vpd.h
@@ -0,0 +1,85 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#ifndef _ISERIES_PROCESSOR_VPD_H
+#define _ISERIES_PROCESSOR_VPD_H
+
+#include <asm/types.h>
+
+/*
+ * This struct maps Processor Vpd that is DMAd to SLIC by CSP
+ */
+struct IoHriProcessorVpd {
+	u8	xFormat;		// VPD format indicator		x00-x00
+	u8	xProcStatus:8;		// Processor State		x01-x01
+	u8	xSecondaryThreadCount;	// Secondary thread cnt		x02-x02
+	u8	xSrcType:1;		// Src Type			x03-x03
+	u8	xSrcSoft:1;		// Src stay soft		...
+	u8	xSrcParable:1;		// Src parable			...
+	u8	xRsvd1:5;		// Reserved			...
+	u16	xHvPhysicalProcIndex;	// Hypervisor physical proc index04-x05
+	u16	xRsvd2;			// Reserved			x06-x07
+	u32	xHwNodeId;		// Hardware node id		x08-x0B
+	u32	xHwProcId;		// Hardware processor id	x0C-x0F
+
+	u32	xTypeNum;		// Card Type/CCIN number	x10-x13
+	u32	xModelNum;		// Model/Feature number		x14-x17
+	u64	xSerialNum;		// Serial number		x18-x1F
+	char	xPartNum[12];		// Book Part or FPU number	x20-x2B
+	char	xMfgID[4];		// Manufacturing ID		x2C-x2F
+
+	u32	xProcFreq;		// Processor Frequency		x30-x33
+	u32	xTimeBaseFreq;		// Time Base Frequency		x34-x37
+
+	u32	xChipEcLevel;		// Chip EC Levels		x38-x3B
+	u32	xProcIdReg;		// PIR SPR value		x3C-x3F
+	u32	xPVR;			// PVR value			x40-x43
+	u8	xRsvd3[12];		// Reserved			x44-x4F
+
+	u32	xInstCacheSize;		// Instruction cache size in KB	x50-x53
+	u32	xInstBlockSize;		// Instruction cache block size	x54-x57
+	u32	xDataCacheOperandSize;	// Data cache operand size	x58-x5B
+	u32	xInstCacheOperandSize;	// Inst cache operand size	x5C-x5F
+
+	u32	xDataL1CacheSizeKB;	// L1 data cache size in KB	x60-x63
+	u32	xDataL1CacheLineSize;	// L1 data cache block size	x64-x67
+	u64	xRsvd4;			// Reserved			x68-x6F
+
+	u32	xDataL2CacheSizeKB;	// L2 data cache size in KB	x70-x73
+	u32	xDataL2CacheLineSize;	// L2 data cache block size	x74-x77
+	u64	xRsvd5;			// Reserved			x78-x7F
+
+	u32	xDataL3CacheSizeKB;	// L3 data cache size in KB	x80-x83
+	u32	xDataL3CacheLineSize;	// L3 data cache block size	x84-x87
+	u64	xRsvd6;			// Reserved			x88-x8F
+
+	u64	xFruLabel;		// Card Location Label		x90-x97
+	u8	xSlotsOnCard;		// Slots on card (0=no slots)	x98-x98
+	u8	xPartLocFlag;		// Location flag (0-pluggable 1-imbedded) x99-x99
+	u16	xSlotMapIndex;		// Index in slot map table	x9A-x9B
+	u8	xSmartCardPortNo;	// Smart card port number	x9C-x9C
+	u8	xRsvd7;			// Reserved			x9D-x9D
+	u16	xFrameIdAndRackUnit;	// Frame ID and rack unit adr	x9E-x9F
+
+	u8	xRsvd8[24];		// Reserved			xA0-xB7
+
+	char	xProcSrc[72];		// CSP format SRC		xB8-xFF
+};
+
+extern struct IoHriProcessorVpd	xIoHriProcessorVpd[];
+
+#endif /* _ISERIES_PROCESSOR_VPD_H */
diff --git a/arch/powerpc/platforms/iseries/release_data.h b/arch/powerpc/platforms/iseries/release_data.h
new file mode 100644
index 000000000000..66189fd2e32d
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/release_data.h
@@ -0,0 +1,63 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#ifndef _ISERIES_RELEASE_DATA_H
+#define _ISERIES_RELEASE_DATA_H
+
+/*
+ * This control block contains the critical information about the
+ * release so that it can be changed in the future (ie, the virtual
+ * address of the OS's NACA).
+ */
+#include <asm/types.h>
+#include "naca.h"
+
+/*
+ * When we IPL a secondary partition, we will check if if the
+ * secondary xMinPlicVrmIndex > the primary xVrmIndex.
+ * If it is then this tells PLIC that this secondary is not
+ * supported running on this "old" of a level of PLIC.
+ *
+ * Likewise, we will compare the primary xMinSlicVrmIndex to
+ * the secondary xVrmIndex.
+ * If the primary xMinSlicVrmDelta > secondary xVrmDelta then we
+ * know that this PLIC does not support running an OS "that old".
+ */
+
+#define	HVREL_TAGSINACTIVE	0x8000
+#define HVREL_32BIT		0x4000
+#define HVREL_NOSHAREDPROCS	0x2000
+#define HVREL_NOHMT		0x1000
+
+struct HvReleaseData {
+	u32	xDesc;		/* Descriptor "HvRD" ebcdic	x00-x03 */
+	u16	xSize;		/* Size of this control block	x04-x05 */
+	u16	xVpdAreasPtrOffset; /* Offset in NACA of ItVpdAreas x06-x07 */
+	struct  naca_struct	*xSlicNacaAddr; /* Virt addr of SLIC NACA x08-x0F */
+	u32	xMsNucDataOffset; /* Offset of Linux Mapping Data x10-x13 */
+	u32	xRsvd1;		/* Reserved			x14-x17 */
+	u16	xFlags;
+	u16	xVrmIndex;	/* VRM Index of OS image	x1A-x1B */
+	u16	xMinSupportedPlicVrmIndex; /* Min PLIC level  (soft) x1C-x1D */
+	u16	xMinCompatablePlicVrmIndex; /* Min PLIC levelP (hard) x1E-x1F */
+	char	xVrmName[12];	/* Displayable name		x20-x2B */
+	char	xRsvd3[20];	/* Reserved			x2C-x3F */
+};
+
+extern struct HvReleaseData	hvReleaseData;
+
+#endif /* _ISERIES_RELEASE_DATA_H */
diff --git a/arch/powerpc/platforms/iseries/setup.c b/arch/powerpc/platforms/iseries/setup.c
new file mode 100644
index 000000000000..fda712b42168
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/setup.c
@@ -0,0 +1,1004 @@
+/*
+ *    Copyright (c) 2000 Mike Corrigan <mikejc@us.ibm.com>
+ *    Copyright (c) 1999-2000 Grant Erickson <grant@lcse.umn.edu>
+ *
+ *    Description:
+ *      Architecture- / platform-specific boot-time initialization code for
+ *      the IBM iSeries LPAR.  Adapted from original code by Grant Erickson and
+ *      code by Gary Thomas, Cort Dougan <cort@fsmlabs.com>, and Dan Malek
+ *      <dan@net4x.com>.
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/init.h>
+#include <linux/threads.h>
+#include <linux/smp.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/initrd.h>
+#include <linux/seq_file.h>
+#include <linux/kdev_t.h>
+#include <linux/major.h>
+#include <linux/root_dev.h>
+#include <linux/kernel.h>
+
+#include <asm/processor.h>
+#include <asm/machdep.h>
+#include <asm/page.h>
+#include <asm/mmu.h>
+#include <asm/pgtable.h>
+#include <asm/mmu_context.h>
+#include <asm/cputable.h>
+#include <asm/sections.h>
+#include <asm/iommu.h>
+#include <asm/firmware.h>
+
+#include <asm/time.h>
+#include <asm/paca.h>
+#include <asm/cache.h>
+#include <asm/sections.h>
+#include <asm/abs_addr.h>
+#include <asm/iseries/hv_lp_config.h>
+#include <asm/iseries/hv_call_event.h>
+#include <asm/iseries/hv_call_xm.h>
+#include <asm/iseries/it_lp_queue.h>
+#include <asm/iseries/mf.h>
+#include <asm/iseries/hv_lp_event.h>
+#include <asm/iseries/lpar_map.h>
+
+#include "naca.h"
+#include "setup.h"
+#include "irq.h"
+#include "vpd_areas.h"
+#include "processor_vpd.h"
+#include "main_store.h"
+#include "call_sm.h"
+#include "call_hpt.h"
+
+extern void hvlog(char *fmt, ...);
+
+#ifdef DEBUG
+#define DBG(fmt...) hvlog(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+/* Function Prototypes */
+extern void ppcdbg_initialize(void);
+
+static void build_iSeries_Memory_Map(void);
+static void iseries_shared_idle(void);
+static void iseries_dedicated_idle(void);
+#ifdef CONFIG_PCI
+extern void iSeries_pci_final_fixup(void);
+#else
+static void iSeries_pci_final_fixup(void) { }
+#endif
+
+/* Global Variables */
+int piranha_simulator;
+
+extern int rd_size;		/* Defined in drivers/block/rd.c */
+extern unsigned long klimit;
+extern unsigned long embedded_sysmap_start;
+extern unsigned long embedded_sysmap_end;
+
+extern unsigned long iSeries_recal_tb;
+extern unsigned long iSeries_recal_titan;
+
+static int mf_initialized;
+
+static unsigned long cmd_mem_limit;
+
+struct MemoryBlock {
+	unsigned long absStart;
+	unsigned long absEnd;
+	unsigned long logicalStart;
+	unsigned long logicalEnd;
+};
+
+/*
+ * Process the main store vpd to determine where the holes in memory are
+ * and return the number of physical blocks and fill in the array of
+ * block data.
+ */
+static unsigned long iSeries_process_Condor_mainstore_vpd(
+		struct MemoryBlock *mb_array, unsigned long max_entries)
+{
+	unsigned long holeFirstChunk, holeSizeChunks;
+	unsigned long numMemoryBlocks = 1;
+	struct IoHriMainStoreSegment4 *msVpd =
+		(struct IoHriMainStoreSegment4 *)xMsVpd;
+	unsigned long holeStart = msVpd->nonInterleavedBlocksStartAdr;
+	unsigned long holeEnd = msVpd->nonInterleavedBlocksEndAdr;
+	unsigned long holeSize = holeEnd - holeStart;
+
+	printk("Mainstore_VPD: Condor\n");
+	/*
+	 * Determine if absolute memory has any
+	 * holes so that we can interpret the
+	 * access map we get back from the hypervisor
+	 * correctly.
+	 */
+	mb_array[0].logicalStart = 0;
+	mb_array[0].logicalEnd = 0x100000000;
+	mb_array[0].absStart = 0;
+	mb_array[0].absEnd = 0x100000000;
+
+	if (holeSize) {
+		numMemoryBlocks = 2;
+		holeStart = holeStart & 0x000fffffffffffff;
+		holeStart = addr_to_chunk(holeStart);
+		holeFirstChunk = holeStart;
+		holeSize = addr_to_chunk(holeSize);
+		holeSizeChunks = holeSize;
+		printk( "Main store hole: start chunk = %0lx, size = %0lx chunks\n",
+				holeFirstChunk, holeSizeChunks );
+		mb_array[0].logicalEnd = holeFirstChunk;
+		mb_array[0].absEnd = holeFirstChunk;
+		mb_array[1].logicalStart = holeFirstChunk;
+		mb_array[1].logicalEnd = 0x100000000 - holeSizeChunks;
+		mb_array[1].absStart = holeFirstChunk + holeSizeChunks;
+		mb_array[1].absEnd = 0x100000000;
+	}
+	return numMemoryBlocks;
+}
+
+#define MaxSegmentAreas			32
+#define MaxSegmentAdrRangeBlocks	128
+#define MaxAreaRangeBlocks		4
+
+static unsigned long iSeries_process_Regatta_mainstore_vpd(
+		struct MemoryBlock *mb_array, unsigned long max_entries)
+{
+	struct IoHriMainStoreSegment5 *msVpdP =
+		(struct IoHriMainStoreSegment5 *)xMsVpd;
+	unsigned long numSegmentBlocks = 0;
+	u32 existsBits = msVpdP->msAreaExists;
+	unsigned long area_num;
+
+	printk("Mainstore_VPD: Regatta\n");
+
+	for (area_num = 0; area_num < MaxSegmentAreas; ++area_num ) {
+		unsigned long numAreaBlocks;
+		struct IoHriMainStoreArea4 *currentArea;
+
+		if (existsBits & 0x80000000) {
+			unsigned long block_num;
+
+			currentArea = &msVpdP->msAreaArray[area_num];
+			numAreaBlocks = currentArea->numAdrRangeBlocks;
+			printk("ms_vpd: processing area %2ld  blocks=%ld",
+					area_num, numAreaBlocks);
+			for (block_num = 0; block_num < numAreaBlocks;
+					++block_num ) {
+				/* Process an address range block */
+				struct MemoryBlock tempBlock;
+				unsigned long i;
+
+				tempBlock.absStart =
+					(unsigned long)currentArea->xAdrRangeBlock[block_num].blockStart;
+				tempBlock.absEnd =
+					(unsigned long)currentArea->xAdrRangeBlock[block_num].blockEnd;
+				tempBlock.logicalStart = 0;
+				tempBlock.logicalEnd   = 0;
+				printk("\n          block %ld absStart=%016lx absEnd=%016lx",
+						block_num, tempBlock.absStart,
+						tempBlock.absEnd);
+
+				for (i = 0; i < numSegmentBlocks; ++i) {
+					if (mb_array[i].absStart ==
+							tempBlock.absStart)
+						break;
+				}
+				if (i == numSegmentBlocks) {
+					if (numSegmentBlocks == max_entries)
+						panic("iSeries_process_mainstore_vpd: too many memory blocks");
+					mb_array[numSegmentBlocks] = tempBlock;
+					++numSegmentBlocks;
+				} else
+					printk(" (duplicate)");
+			}
+			printk("\n");
+		}
+		existsBits <<= 1;
+	}
+	/* Now sort the blocks found into ascending sequence */
+	if (numSegmentBlocks > 1) {
+		unsigned long m, n;
+
+		for (m = 0; m < numSegmentBlocks - 1; ++m) {
+			for (n = numSegmentBlocks - 1; m < n; --n) {
+				if (mb_array[n].absStart <
+						mb_array[n-1].absStart) {
+					struct MemoryBlock tempBlock;
+
+					tempBlock = mb_array[n];
+					mb_array[n] = mb_array[n-1];
+					mb_array[n-1] = tempBlock;
+				}
+			}
+		}
+	}
+	/*
+	 * Assign "logical" addresses to each block.  These
+	 * addresses correspond to the hypervisor "bitmap" space.
+	 * Convert all addresses into units of 256K chunks.
+	 */
+	{
+	unsigned long i, nextBitmapAddress;
+
+	printk("ms_vpd: %ld sorted memory blocks\n", numSegmentBlocks);
+	nextBitmapAddress = 0;
+	for (i = 0; i < numSegmentBlocks; ++i) {
+		unsigned long length = mb_array[i].absEnd -
+			mb_array[i].absStart;
+
+		mb_array[i].logicalStart = nextBitmapAddress;
+		mb_array[i].logicalEnd = nextBitmapAddress + length;
+		nextBitmapAddress += length;
+		printk("          Bitmap range: %016lx - %016lx\n"
+				"        Absolute range: %016lx - %016lx\n",
+				mb_array[i].logicalStart,
+				mb_array[i].logicalEnd,
+				mb_array[i].absStart, mb_array[i].absEnd);
+		mb_array[i].absStart = addr_to_chunk(mb_array[i].absStart &
+				0x000fffffffffffff);
+		mb_array[i].absEnd = addr_to_chunk(mb_array[i].absEnd &
+				0x000fffffffffffff);
+		mb_array[i].logicalStart =
+			addr_to_chunk(mb_array[i].logicalStart);
+		mb_array[i].logicalEnd = addr_to_chunk(mb_array[i].logicalEnd);
+	}
+	}
+
+	return numSegmentBlocks;
+}
+
+static unsigned long iSeries_process_mainstore_vpd(struct MemoryBlock *mb_array,
+		unsigned long max_entries)
+{
+	unsigned long i;
+	unsigned long mem_blocks = 0;
+
+	if (cpu_has_feature(CPU_FTR_SLB))
+		mem_blocks = iSeries_process_Regatta_mainstore_vpd(mb_array,
+				max_entries);
+	else
+		mem_blocks = iSeries_process_Condor_mainstore_vpd(mb_array,
+				max_entries);
+
+	printk("Mainstore_VPD: numMemoryBlocks = %ld \n", mem_blocks);
+	for (i = 0; i < mem_blocks; ++i) {
+		printk("Mainstore_VPD: block %3ld logical chunks %016lx - %016lx\n"
+		       "                             abs chunks %016lx - %016lx\n",
+			i, mb_array[i].logicalStart, mb_array[i].logicalEnd,
+			mb_array[i].absStart, mb_array[i].absEnd);
+	}
+	return mem_blocks;
+}
+
+static void __init iSeries_get_cmdline(void)
+{
+	char *p, *q;
+
+	/* copy the command line parameter from the primary VSP  */
+	HvCallEvent_dmaToSp(cmd_line, 2 * 64* 1024, 256,
+			HvLpDma_Direction_RemoteToLocal);
+
+	p = cmd_line;
+	q = cmd_line + 255;
+	while(p < q) {
+		if (!*p || *p == '\n')
+			break;
+		++p;
+	}
+	*p = 0;
+}
+
+static void __init iSeries_init_early(void)
+{
+	DBG(" -> iSeries_init_early()\n");
+
+	ppc64_firmware_features = FW_FEATURE_ISERIES;
+
+	ppcdbg_initialize();
+
+	ppc64_interrupt_controller = IC_ISERIES;
+
+#if defined(CONFIG_BLK_DEV_INITRD)
+	/*
+	 * If the init RAM disk has been configured and there is
+	 * a non-zero starting address for it, set it up
+	 */
+	if (naca.xRamDisk) {
+		initrd_start = (unsigned long)__va(naca.xRamDisk);
+		initrd_end = initrd_start + naca.xRamDiskSize * PAGE_SIZE;
+		initrd_below_start_ok = 1;	// ramdisk in kernel space
+		ROOT_DEV = Root_RAM0;
+		if (((rd_size * 1024) / PAGE_SIZE) < naca.xRamDiskSize)
+			rd_size = (naca.xRamDiskSize * PAGE_SIZE) / 1024;
+	} else
+#endif /* CONFIG_BLK_DEV_INITRD */
+	{
+	    /* ROOT_DEV = MKDEV(VIODASD_MAJOR, 1); */
+	}
+
+	iSeries_recal_tb = get_tb();
+	iSeries_recal_titan = HvCallXm_loadTod();
+
+	/*
+	 * Initialize the hash table management pointers
+	 */
+	hpte_init_iSeries();
+
+	/*
+	 * Initialize the DMA/TCE management
+	 */
+	iommu_init_early_iSeries();
+
+	/* Initialize machine-dependency vectors */
+#ifdef CONFIG_SMP
+	smp_init_iSeries();
+#endif
+	if (itLpNaca.xPirEnvironMode == 0)
+		piranha_simulator = 1;
+
+	/* Associate Lp Event Queue 0 with processor 0 */
+	HvCallEvent_setLpEventQueueInterruptProc(0, 0);
+
+	mf_init();
+	mf_initialized = 1;
+	mb();
+
+	/* If we were passed an initrd, set the ROOT_DEV properly if the values
+	 * look sensible. If not, clear initrd reference.
+	 */
+#ifdef CONFIG_BLK_DEV_INITRD
+	if (initrd_start >= KERNELBASE && initrd_end >= KERNELBASE &&
+	    initrd_end > initrd_start)
+		ROOT_DEV = Root_RAM0;
+	else
+		initrd_start = initrd_end = 0;
+#endif /* CONFIG_BLK_DEV_INITRD */
+
+	DBG(" <- iSeries_init_early()\n");
+}
+
+struct mschunks_map mschunks_map = {
+	/* XXX We don't use these, but Piranha might need them. */
+	.chunk_size  = MSCHUNKS_CHUNK_SIZE,
+	.chunk_shift = MSCHUNKS_CHUNK_SHIFT,
+	.chunk_mask  = MSCHUNKS_OFFSET_MASK,
+};
+EXPORT_SYMBOL(mschunks_map);
+
+void mschunks_alloc(unsigned long num_chunks)
+{
+	klimit = _ALIGN(klimit, sizeof(u32));
+	mschunks_map.mapping = (u32 *)klimit;
+	klimit += num_chunks * sizeof(u32);
+	mschunks_map.num_chunks = num_chunks;
+}
+
+/*
+ * The iSeries may have very large memories ( > 128 GB ) and a partition
+ * may get memory in "chunks" that may be anywhere in the 2**52 real
+ * address space.  The chunks are 256K in size.  To map this to the
+ * memory model Linux expects, the AS/400 specific code builds a
+ * translation table to translate what Linux thinks are "physical"
+ * addresses to the actual real addresses.  This allows us to make
+ * it appear to Linux that we have contiguous memory starting at
+ * physical address zero while in fact this could be far from the truth.
+ * To avoid confusion, I'll let the words physical and/or real address
+ * apply to the Linux addresses while I'll use "absolute address" to
+ * refer to the actual hardware real address.
+ *
+ * build_iSeries_Memory_Map gets information from the Hypervisor and
+ * looks at the Main Store VPD to determine the absolute addresses
+ * of the memory that has been assigned to our partition and builds
+ * a table used to translate Linux's physical addresses to these
+ * absolute addresses.  Absolute addresses are needed when
+ * communicating with the hypervisor (e.g. to build HPT entries)
+ */
+
+static void __init build_iSeries_Memory_Map(void)
+{
+	u32 loadAreaFirstChunk, loadAreaLastChunk, loadAreaSize;
+	u32 nextPhysChunk;
+	u32 hptFirstChunk, hptLastChunk, hptSizeChunks, hptSizePages;
+	u32 totalChunks,moreChunks;
+	u32 currChunk, thisChunk, absChunk;
+	u32 currDword;
+	u32 chunkBit;
+	u64 map;
+	struct MemoryBlock mb[32];
+	unsigned long numMemoryBlocks, curBlock;
+
+	/* Chunk size on iSeries is 256K bytes */
+	totalChunks = (u32)HvLpConfig_getMsChunks();
+	mschunks_alloc(totalChunks);
+
+	/*
+	 * Get absolute address of our load area
+	 * and map it to physical address 0
+	 * This guarantees that the loadarea ends up at physical 0
+	 * otherwise, it might not be returned by PLIC as the first
+	 * chunks
+	 */
+
+	loadAreaFirstChunk = (u32)addr_to_chunk(itLpNaca.xLoadAreaAddr);
+	loadAreaSize =  itLpNaca.xLoadAreaChunks;
+
+	/*
+	 * Only add the pages already mapped here.
+	 * Otherwise we might add the hpt pages
+	 * The rest of the pages of the load area
+	 * aren't in the HPT yet and can still
+	 * be assigned an arbitrary physical address
+	 */
+	if ((loadAreaSize * 64) > HvPagesToMap)
+		loadAreaSize = HvPagesToMap / 64;
+
+	loadAreaLastChunk = loadAreaFirstChunk + loadAreaSize - 1;
+
+	/*
+	 * TODO Do we need to do something if the HPT is in the 64MB load area?
+	 * This would be required if the itLpNaca.xLoadAreaChunks includes
+	 * the HPT size
+	 */
+
+	printk("Mapping load area - physical addr = 0000000000000000\n"
+		"                    absolute addr = %016lx\n",
+		chunk_to_addr(loadAreaFirstChunk));
+	printk("Load area size %dK\n", loadAreaSize * 256);
+
+	for (nextPhysChunk = 0; nextPhysChunk < loadAreaSize; ++nextPhysChunk)
+		mschunks_map.mapping[nextPhysChunk] =
+			loadAreaFirstChunk + nextPhysChunk;
+
+	/*
+	 * Get absolute address of our HPT and remember it so
+	 * we won't map it to any physical address
+	 */
+	hptFirstChunk = (u32)addr_to_chunk(HvCallHpt_getHptAddress());
+	hptSizePages = (u32)HvCallHpt_getHptPages();
+	hptSizeChunks = hptSizePages >> (MSCHUNKS_CHUNK_SHIFT - PAGE_SHIFT);
+	hptLastChunk = hptFirstChunk + hptSizeChunks - 1;
+
+	printk("HPT absolute addr = %016lx, size = %dK\n",
+			chunk_to_addr(hptFirstChunk), hptSizeChunks * 256);
+
+	ppc64_pft_size = __ilog2(hptSizePages * PAGE_SIZE);
+
+	/*
+	 * The actual hashed page table is in the hypervisor,
+	 * we have no direct access
+	 */
+	htab_address = NULL;
+
+	/*
+	 * Determine if absolute memory has any
+	 * holes so that we can interpret the
+	 * access map we get back from the hypervisor
+	 * correctly.
+	 */
+	numMemoryBlocks = iSeries_process_mainstore_vpd(mb, 32);
+
+	/*
+	 * Process the main store access map from the hypervisor
+	 * to build up our physical -> absolute translation table
+	 */
+	curBlock = 0;
+	currChunk = 0;
+	currDword = 0;
+	moreChunks = totalChunks;
+
+	while (moreChunks) {
+		map = HvCallSm_get64BitsOfAccessMap(itLpNaca.xLpIndex,
+				currDword);
+		thisChunk = currChunk;
+		while (map) {
+			chunkBit = map >> 63;
+			map <<= 1;
+			if (chunkBit) {
+				--moreChunks;
+				while (thisChunk >= mb[curBlock].logicalEnd) {
+					++curBlock;
+					if (curBlock >= numMemoryBlocks)
+						panic("out of memory blocks");
+				}
+				if (thisChunk < mb[curBlock].logicalStart)
+					panic("memory block error");
+
+				absChunk = mb[curBlock].absStart +
+					(thisChunk - mb[curBlock].logicalStart);
+				if (((absChunk < hptFirstChunk) ||
+				     (absChunk > hptLastChunk)) &&
+				    ((absChunk < loadAreaFirstChunk) ||
+				     (absChunk > loadAreaLastChunk))) {
+					mschunks_map.mapping[nextPhysChunk] =
+						absChunk;
+					++nextPhysChunk;
+				}
+			}
+			++thisChunk;
+		}
+		++currDword;
+		currChunk += 64;
+	}
+
+	/*
+	 * main store size (in chunks) is
+	 *   totalChunks - hptSizeChunks
+	 * which should be equal to
+	 *   nextPhysChunk
+	 */
+	systemcfg->physicalMemorySize = chunk_to_addr(nextPhysChunk);
+}
+
+/*
+ * Document me.
+ */
+static void __init iSeries_setup_arch(void)
+{
+	unsigned procIx = get_paca()->lppaca.dyn_hv_phys_proc_index;
+
+	if (get_paca()->lppaca.shared_proc) {
+		ppc_md.idle_loop = iseries_shared_idle;
+		printk(KERN_INFO "Using shared processor idle loop\n");
+	} else {
+		ppc_md.idle_loop = iseries_dedicated_idle;
+		printk(KERN_INFO "Using dedicated idle loop\n");
+	}
+
+	/* Setup the Lp Event Queue */
+	setup_hvlpevent_queue();
+
+	printk("Max  logical processors = %d\n",
+			itVpdAreas.xSlicMaxLogicalProcs);
+	printk("Max physical processors = %d\n",
+			itVpdAreas.xSlicMaxPhysicalProcs);
+
+	systemcfg->processor = xIoHriProcessorVpd[procIx].xPVR;
+	printk("Processor version = %x\n", systemcfg->processor);
+}
+
+static void iSeries_show_cpuinfo(struct seq_file *m)
+{
+	seq_printf(m, "machine\t\t: 64-bit iSeries Logical Partition\n");
+}
+
+/*
+ * Document me.
+ * and Implement me.
+ */
+static int iSeries_get_irq(struct pt_regs *regs)
+{
+	/* -2 means ignore this interrupt */
+	return -2;
+}
+
+/*
+ * Document me.
+ */
+static void iSeries_restart(char *cmd)
+{
+	mf_reboot();
+}
+
+/*
+ * Document me.
+ */
+static void iSeries_power_off(void)
+{
+	mf_power_off();
+}
+
+/*
+ * Document me.
+ */
+static void iSeries_halt(void)
+{
+	mf_power_off();
+}
+
+static void __init iSeries_progress(char * st, unsigned short code)
+{
+	printk("Progress: [%04x] - %s\n", (unsigned)code, st);
+	if (!piranha_simulator && mf_initialized) {
+		if (code != 0xffff)
+			mf_display_progress(code);
+		else
+			mf_clear_src();
+	}
+}
+
+static void __init iSeries_fixup_klimit(void)
+{
+	/*
+	 * Change klimit to take into account any ram disk
+	 * that may be included
+	 */
+	if (naca.xRamDisk)
+		klimit = KERNELBASE + (u64)naca.xRamDisk +
+			(naca.xRamDiskSize * PAGE_SIZE);
+	else {
+		/*
+		 * No ram disk was included - check and see if there
+		 * was an embedded system map.  Change klimit to take
+		 * into account any embedded system map
+		 */
+		if (embedded_sysmap_end)
+			klimit = KERNELBASE + ((embedded_sysmap_end + 4095) &
+					0xfffffffffffff000);
+	}
+}
+
+static int __init iSeries_src_init(void)
+{
+        /* clear the progress line */
+        ppc_md.progress(" ", 0xffff);
+        return 0;
+}
+
+late_initcall(iSeries_src_init);
+
+static inline void process_iSeries_events(void)
+{
+	asm volatile ("li 0,0x5555; sc" : : : "r0", "r3");
+}
+
+static void yield_shared_processor(void)
+{
+	unsigned long tb;
+
+	HvCall_setEnabledInterrupts(HvCall_MaskIPI |
+				    HvCall_MaskLpEvent |
+				    HvCall_MaskLpProd |
+				    HvCall_MaskTimeout);
+
+	tb = get_tb();
+	/* Compute future tb value when yield should expire */
+	HvCall_yieldProcessor(HvCall_YieldTimed, tb+tb_ticks_per_jiffy);
+
+	/*
+	 * The decrementer stops during the yield.  Force a fake decrementer
+	 * here and let the timer_interrupt code sort out the actual time.
+	 */
+	get_paca()->lppaca.int_dword.fields.decr_int = 1;
+	process_iSeries_events();
+}
+
+static void iseries_shared_idle(void)
+{
+	while (1) {
+		while (!need_resched() && !hvlpevent_is_pending()) {
+			local_irq_disable();
+			ppc64_runlatch_off();
+
+			/* Recheck with irqs off */
+			if (!need_resched() && !hvlpevent_is_pending())
+				yield_shared_processor();
+
+			HMT_medium();
+			local_irq_enable();
+		}
+
+		ppc64_runlatch_on();
+
+		if (hvlpevent_is_pending())
+			process_iSeries_events();
+
+		schedule();
+	}
+}
+
+static void iseries_dedicated_idle(void)
+{
+	long oldval;
+
+	while (1) {
+		oldval = test_and_clear_thread_flag(TIF_NEED_RESCHED);
+
+		if (!oldval) {
+			set_thread_flag(TIF_POLLING_NRFLAG);
+
+			while (!need_resched()) {
+				ppc64_runlatch_off();
+				HMT_low();
+
+				if (hvlpevent_is_pending()) {
+					HMT_medium();
+					ppc64_runlatch_on();
+					process_iSeries_events();
+				}
+			}
+
+			HMT_medium();
+			clear_thread_flag(TIF_POLLING_NRFLAG);
+		} else {
+			set_need_resched();
+		}
+
+		ppc64_runlatch_on();
+		schedule();
+	}
+}
+
+#ifndef CONFIG_PCI
+void __init iSeries_init_IRQ(void) { }
+#endif
+
+static int __init iseries_probe(int platform)
+{
+	return PLATFORM_ISERIES_LPAR == platform;
+}
+
+struct machdep_calls __initdata iseries_md = {
+	.setup_arch	= iSeries_setup_arch,
+	.show_cpuinfo	= iSeries_show_cpuinfo,
+	.init_IRQ	= iSeries_init_IRQ,
+	.get_irq	= iSeries_get_irq,
+	.init_early	= iSeries_init_early,
+	.pcibios_fixup	= iSeries_pci_final_fixup,
+	.restart	= iSeries_restart,
+	.power_off	= iSeries_power_off,
+	.halt		= iSeries_halt,
+	.get_boot_time	= iSeries_get_boot_time,
+	.set_rtc_time	= iSeries_set_rtc_time,
+	.get_rtc_time	= iSeries_get_rtc_time,
+	.calibrate_decr	= generic_calibrate_decr,
+	.progress	= iSeries_progress,
+	.probe		= iseries_probe,
+	/* XXX Implement enable_pmcs for iSeries */
+};
+
+struct blob {
+	unsigned char data[PAGE_SIZE];
+	unsigned long next;
+};
+
+struct iseries_flat_dt {
+	struct boot_param_header header;
+	u64 reserve_map[2];
+	struct blob dt;
+	struct blob strings;
+};
+
+struct iseries_flat_dt iseries_dt;
+
+void dt_init(struct iseries_flat_dt *dt)
+{
+	dt->header.off_mem_rsvmap =
+		offsetof(struct iseries_flat_dt, reserve_map);
+	dt->header.off_dt_struct = offsetof(struct iseries_flat_dt, dt);
+	dt->header.off_dt_strings = offsetof(struct iseries_flat_dt, strings);
+	dt->header.totalsize = sizeof(struct iseries_flat_dt);
+	dt->header.dt_strings_size = sizeof(struct blob);
+
+	/* There is no notion of hardware cpu id on iSeries */
+	dt->header.boot_cpuid_phys = smp_processor_id();
+
+	dt->dt.next = (unsigned long)&dt->dt.data;
+	dt->strings.next = (unsigned long)&dt->strings.data;
+
+	dt->header.magic = OF_DT_HEADER;
+	dt->header.version = 0x10;
+	dt->header.last_comp_version = 0x10;
+
+	dt->reserve_map[0] = 0;
+	dt->reserve_map[1] = 0;
+}
+
+void dt_check_blob(struct blob *b)
+{
+	if (b->next >= (unsigned long)&b->next) {
+		DBG("Ran out of space in flat device tree blob!\n");
+		BUG();
+	}
+}
+
+void dt_push_u32(struct iseries_flat_dt *dt, u32 value)
+{
+	*((u32*)dt->dt.next) = value;
+	dt->dt.next += sizeof(u32);
+
+	dt_check_blob(&dt->dt);
+}
+
+void dt_push_u64(struct iseries_flat_dt *dt, u64 value)
+{
+	*((u64*)dt->dt.next) = value;
+	dt->dt.next += sizeof(u64);
+
+	dt_check_blob(&dt->dt);
+}
+
+unsigned long dt_push_bytes(struct blob *blob, char *data, int len)
+{
+	unsigned long start = blob->next - (unsigned long)blob->data;
+
+	memcpy((char *)blob->next, data, len);
+	blob->next = _ALIGN(blob->next + len, 4);
+
+	dt_check_blob(blob);
+
+	return start;
+}
+
+void dt_start_node(struct iseries_flat_dt *dt, char *name)
+{
+	dt_push_u32(dt, OF_DT_BEGIN_NODE);
+	dt_push_bytes(&dt->dt, name, strlen(name) + 1);
+}
+
+#define dt_end_node(dt) dt_push_u32(dt, OF_DT_END_NODE)
+
+void dt_prop(struct iseries_flat_dt *dt, char *name, char *data, int len)
+{
+	unsigned long offset;
+
+	dt_push_u32(dt, OF_DT_PROP);
+
+	/* Length of the data */
+	dt_push_u32(dt, len);
+
+	/* Put the property name in the string blob. */
+	offset = dt_push_bytes(&dt->strings, name, strlen(name) + 1);
+
+	/* The offset of the properties name in the string blob. */
+	dt_push_u32(dt, (u32)offset);
+
+	/* The actual data. */
+	dt_push_bytes(&dt->dt, data, len);
+}
+
+void dt_prop_str(struct iseries_flat_dt *dt, char *name, char *data)
+{
+	dt_prop(dt, name, data, strlen(data) + 1); /* + 1 for NULL */
+}
+
+void dt_prop_u32(struct iseries_flat_dt *dt, char *name, u32 data)
+{
+	dt_prop(dt, name, (char *)&data, sizeof(u32));
+}
+
+void dt_prop_u64(struct iseries_flat_dt *dt, char *name, u64 data)
+{
+	dt_prop(dt, name, (char *)&data, sizeof(u64));
+}
+
+void dt_prop_u64_list(struct iseries_flat_dt *dt, char *name, u64 *data, int n)
+{
+	dt_prop(dt, name, (char *)data, sizeof(u64) * n);
+}
+
+void dt_prop_empty(struct iseries_flat_dt *dt, char *name)
+{
+	dt_prop(dt, name, NULL, 0);
+}
+
+void dt_cpus(struct iseries_flat_dt *dt)
+{
+	unsigned char buf[32];
+	unsigned char *p;
+	unsigned int i, index;
+	struct IoHriProcessorVpd *d;
+
+	/* yuck */
+	snprintf(buf, 32, "PowerPC,%s", cur_cpu_spec->cpu_name);
+	p = strchr(buf, ' ');
+	if (!p) p = buf + strlen(buf);
+
+	dt_start_node(dt, "cpus");
+	dt_prop_u32(dt, "#address-cells", 1);
+	dt_prop_u32(dt, "#size-cells", 0);
+
+	for (i = 0; i < NR_CPUS; i++) {
+		if (paca[i].lppaca.dyn_proc_status >= 2)
+			continue;
+
+		snprintf(p, 32 - (p - buf), "@%d", i);
+		dt_start_node(dt, buf);
+
+		dt_prop_str(dt, "device_type", "cpu");
+
+		index = paca[i].lppaca.dyn_hv_phys_proc_index;
+		d = &xIoHriProcessorVpd[index];
+
+		dt_prop_u32(dt, "i-cache-size", d->xInstCacheSize * 1024);
+		dt_prop_u32(dt, "i-cache-line-size", d->xInstCacheOperandSize);
+
+		dt_prop_u32(dt, "d-cache-size", d->xDataL1CacheSizeKB * 1024);
+		dt_prop_u32(dt, "d-cache-line-size", d->xDataCacheOperandSize);
+
+		/* magic conversions to Hz copied from old code */
+		dt_prop_u32(dt, "clock-frequency",
+			((1UL << 34) * 1000000) / d->xProcFreq);
+		dt_prop_u32(dt, "timebase-frequency",
+			((1UL << 32) * 1000000) / d->xTimeBaseFreq);
+
+		dt_prop_u32(dt, "reg", i);
+
+		dt_end_node(dt);
+	}
+
+	dt_end_node(dt);
+}
+
+void build_flat_dt(struct iseries_flat_dt *dt)
+{
+	u64 tmp[2];
+
+	dt_init(dt);
+
+	dt_start_node(dt, "");
+
+	dt_prop_u32(dt, "#address-cells", 2);
+	dt_prop_u32(dt, "#size-cells", 2);
+
+	/* /memory */
+	dt_start_node(dt, "memory@0");
+	dt_prop_str(dt, "name", "memory");
+	dt_prop_str(dt, "device_type", "memory");
+	tmp[0] = 0;
+	tmp[1] = systemcfg->physicalMemorySize;
+	dt_prop_u64_list(dt, "reg", tmp, 2);
+	dt_end_node(dt);
+
+	/* /chosen */
+	dt_start_node(dt, "chosen");
+	dt_prop_u32(dt, "linux,platform", PLATFORM_ISERIES_LPAR);
+	if (cmd_mem_limit)
+		dt_prop_u64(dt, "linux,memory-limit", cmd_mem_limit);
+	dt_end_node(dt);
+
+	dt_cpus(dt);
+
+	dt_end_node(dt);
+
+	dt_push_u32(dt, OF_DT_END);
+}
+
+void * __init iSeries_early_setup(void)
+{
+	iSeries_fixup_klimit();
+
+	/*
+	 * Initialize the table which translate Linux physical addresses to
+	 * AS/400 absolute addresses
+	 */
+	build_iSeries_Memory_Map();
+
+	iSeries_get_cmdline();
+
+	/* Save unparsed command line copy for /proc/cmdline */
+	strlcpy(saved_command_line, cmd_line, COMMAND_LINE_SIZE);
+
+	/* Parse early parameters, in particular mem=x */
+	parse_early_param();
+
+	build_flat_dt(&iseries_dt);
+
+	return (void *) __pa(&iseries_dt);
+}
+
+/*
+ * On iSeries we just parse the mem=X option from the command line.
+ * On pSeries it's a bit more complicated, see prom_init_mem()
+ */
+static int __init early_parsemem(char *p)
+{
+	if (p)
+		cmd_mem_limit = ALIGN(memparse(p, &p), PAGE_SIZE);
+	return 0;
+}
+early_param("mem", early_parsemem);
diff --git a/arch/powerpc/platforms/iseries/setup.h b/arch/powerpc/platforms/iseries/setup.h
new file mode 100644
index 000000000000..5213044ec411
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/setup.h
@@ -0,0 +1,24 @@
+/*
+ *    Copyright (c) 2000 Mike Corrigan <mikejc@us.ibm.com>
+ *    Copyright (c) 1999-2000 Grant Erickson <grant@lcse.umn.edu>
+ *
+ *    Description:
+ *      Architecture- / platform-specific boot-time initialization code for
+ *      the IBM AS/400 LPAR. Adapted from original code by Grant Erickson and
+ *      code by Gary Thomas, Cort Dougan <cort@cs.nmt.edu>, and Dan Malek
+ *      <dan@netx4.com>.
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+
+#ifndef	__ISERIES_SETUP_H__
+#define	__ISERIES_SETUP_H__
+
+extern unsigned long iSeries_get_boot_time(void);
+extern int iSeries_set_rtc_time(struct rtc_time *tm);
+extern void iSeries_get_rtc_time(struct rtc_time *tm);
+
+#endif /* __ISERIES_SETUP_H__ */
diff --git a/arch/powerpc/platforms/iseries/smp.c b/arch/powerpc/platforms/iseries/smp.c
new file mode 100644
index 000000000000..3336bad67724
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/smp.c
@@ -0,0 +1,121 @@
+/*
+ * SMP support for iSeries machines.
+ *
+ * Dave Engebretsen, Peter Bergner, and
+ * Mike Corrigan {engebret|bergner|mikec}@us.ibm.com
+ *
+ * Plus various changes from other IBM teams...
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+#include <linux/kernel_stat.h>
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/cache.h>
+#include <linux/err.h>
+#include <linux/sysdev.h>
+#include <linux/cpu.h>
+
+#include <asm/ptrace.h>
+#include <asm/atomic.h>
+#include <asm/irq.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/io.h>
+#include <asm/smp.h>
+#include <asm/paca.h>
+#include <asm/iseries/hv_call.h>
+#include <asm/time.h>
+#include <asm/ppcdebug.h>
+#include <asm/machdep.h>
+#include <asm/cputable.h>
+#include <asm/system.h>
+
+static unsigned long iSeries_smp_message[NR_CPUS];
+
+void iSeries_smp_message_recv(struct pt_regs *regs)
+{
+	int cpu = smp_processor_id();
+	int msg;
+
+	if (num_online_cpus() < 2)
+		return;
+
+	for (msg = 0; msg < 4; msg++)
+		if (test_and_clear_bit(msg, &iSeries_smp_message[cpu]))
+			smp_message_recv(msg, regs);
+}
+
+static inline void smp_iSeries_do_message(int cpu, int msg)
+{
+	set_bit(msg, &iSeries_smp_message[cpu]);
+	HvCall_sendIPI(&(paca[cpu]));
+}
+
+static void smp_iSeries_message_pass(int target, int msg)
+{
+	int i;
+
+	if (target < NR_CPUS)
+		smp_iSeries_do_message(target, msg);
+	else {
+		for_each_online_cpu(i) {
+			if ((target == MSG_ALL_BUT_SELF) &&
+					(i == smp_processor_id()))
+				continue;
+			smp_iSeries_do_message(i, msg);
+		}
+	}
+}
+
+static int smp_iSeries_probe(void)
+{
+	return cpus_weight(cpu_possible_map);
+}
+
+static void smp_iSeries_kick_cpu(int nr)
+{
+	BUG_ON((nr < 0) || (nr >= NR_CPUS));
+
+	/* Verify that our partition has a processor nr */
+	if (paca[nr].lppaca.dyn_proc_status >= 2)
+		return;
+
+	/* The processor is currently spinning, waiting
+	 * for the cpu_start field to become non-zero
+	 * After we set cpu_start, the processor will
+	 * continue on to secondary_start in iSeries_head.S
+	 */
+	paca[nr].cpu_start = 1;
+}
+
+static void __devinit smp_iSeries_setup_cpu(int nr)
+{
+}
+
+static struct smp_ops_t iSeries_smp_ops = {
+	.message_pass = smp_iSeries_message_pass,
+	.probe        = smp_iSeries_probe,
+	.kick_cpu     = smp_iSeries_kick_cpu,
+	.setup_cpu    = smp_iSeries_setup_cpu,
+};
+
+/* This is called very early. */
+void __init smp_init_iSeries(void)
+{
+	smp_ops = &iSeries_smp_ops;
+}
diff --git a/arch/powerpc/platforms/iseries/spcomm_area.h b/arch/powerpc/platforms/iseries/spcomm_area.h
new file mode 100644
index 000000000000..6e3b685115c9
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/spcomm_area.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#ifndef _ISERIES_SPCOMM_AREA_H
+#define _ISERIES_SPCOMM_AREA_H
+
+
+struct SpCommArea {
+	u32	xDesc;			// Descriptor (only in new formats)	000-003
+	u8	xFormat;		// Format (only in new formats)		004-004
+	u8	xRsvd1[11];		// Reserved				005-00F
+	u64	xRawTbAtIplStart;	// Raw HW TB value when IPL is started	010-017
+	u64	xRawTodAtIplStart;	// Raw HW TOD value when IPL is started	018-01F
+	u64	xBcdTimeAtIplStart;	// BCD time when IPL is started		020-027
+	u64	xBcdTimeAtOsStart;	// BCD time when OS passed control	028-02F
+	u8	xRsvd2[80];		// Reserved				030-07F
+};
+
+extern struct SpCommArea xSpCommArea;
+
+#endif /* _ISERIES_SPCOMM_AREA_H */
diff --git a/arch/powerpc/platforms/iseries/vio.c b/arch/powerpc/platforms/iseries/vio.c
new file mode 100644
index 000000000000..c27a66876c2c
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/vio.c
@@ -0,0 +1,156 @@
+/*
+ * IBM PowerPC iSeries Virtual I/O Infrastructure Support.
+ *
+ *    Copyright (c) 2005 Stephen Rothwell, IBM Corp.
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+#include <linux/types.h>
+#include <linux/device.h>
+#include <linux/init.h>
+
+#include <asm/vio.h>
+#include <asm/iommu.h>
+#include <asm/tce.h>
+#include <asm/abs_addr.h>
+#include <asm/page.h>
+#include <asm/iseries/vio.h>
+#include <asm/iseries/hv_types.h>
+#include <asm/iseries/hv_lp_config.h>
+#include <asm/iseries/hv_call_xm.h>
+
+struct device *iSeries_vio_dev = &vio_bus_device.dev;
+EXPORT_SYMBOL(iSeries_vio_dev);
+
+static struct iommu_table veth_iommu_table;
+static struct iommu_table vio_iommu_table;
+
+static void __init iommu_vio_init(void)
+{
+	struct iommu_table *t;
+	struct iommu_table_cb cb;
+	unsigned long cbp;
+	unsigned long itc_entries;
+
+	cb.itc_busno = 255;    /* Bus 255 is the virtual bus */
+	cb.itc_virtbus = 0xff; /* Ask for virtual bus */
+
+	cbp = virt_to_abs(&cb);
+	HvCallXm_getTceTableParms(cbp);
+
+	itc_entries = cb.itc_size * PAGE_SIZE / sizeof(union tce_entry);
+	veth_iommu_table.it_size        = itc_entries / 2;
+	veth_iommu_table.it_busno       = cb.itc_busno;
+	veth_iommu_table.it_offset      = cb.itc_offset;
+	veth_iommu_table.it_index       = cb.itc_index;
+	veth_iommu_table.it_type        = TCE_VB;
+	veth_iommu_table.it_blocksize	= 1;
+
+	t = iommu_init_table(&veth_iommu_table);
+
+	if (!t)
+		printk("Virtual Bus VETH TCE table failed.\n");
+
+	vio_iommu_table.it_size         = itc_entries - veth_iommu_table.it_size;
+	vio_iommu_table.it_busno        = cb.itc_busno;
+	vio_iommu_table.it_offset       = cb.itc_offset +
+					  veth_iommu_table.it_size;
+	vio_iommu_table.it_index        = cb.itc_index;
+	vio_iommu_table.it_type         = TCE_VB;
+	vio_iommu_table.it_blocksize	= 1;
+
+	t = iommu_init_table(&vio_iommu_table);
+
+	if (!t)
+		printk("Virtual Bus VIO TCE table failed.\n");
+}
+
+/**
+ * vio_register_device_iseries: - Register a new iSeries vio device.
+ * @voidev:	The device to register.
+ */
+static struct vio_dev *__init vio_register_device_iseries(char *type,
+		uint32_t unit_num)
+{
+	struct vio_dev *viodev;
+
+	/* allocate a vio_dev for this device */
+	viodev = kmalloc(sizeof(struct vio_dev), GFP_KERNEL);
+	if (!viodev)
+		return NULL;
+	memset(viodev, 0, sizeof(struct vio_dev));
+
+	snprintf(viodev->dev.bus_id, BUS_ID_SIZE, "%s%d", type, unit_num);
+
+	viodev->name = viodev->dev.bus_id;
+	viodev->type = type;
+	viodev->unit_address = unit_num;
+	viodev->iommu_table = &vio_iommu_table;
+	if (vio_register_device(viodev) == NULL) {
+		kfree(viodev);
+		return NULL;
+	}
+	return viodev;
+}
+
+void __init probe_bus_iseries(void)
+{
+	HvLpIndexMap vlan_map;
+	struct vio_dev *viodev;
+	int i;
+
+	/* there is only one of each of these */
+	vio_register_device_iseries("viocons", 0);
+	vio_register_device_iseries("vscsi", 0);
+
+	vlan_map = HvLpConfig_getVirtualLanIndexMap();
+	for (i = 0; i < HVMAXARCHITECTEDVIRTUALLANS; i++) {
+		if ((vlan_map & (0x8000 >> i)) == 0)
+			continue;
+		viodev = vio_register_device_iseries("vlan", i);
+		/* veth is special and has it own iommu_table */
+		viodev->iommu_table = &veth_iommu_table;
+	}
+	for (i = 0; i < HVMAXARCHITECTEDVIRTUALDISKS; i++)
+		vio_register_device_iseries("viodasd", i);
+	for (i = 0; i < HVMAXARCHITECTEDVIRTUALCDROMS; i++)
+		vio_register_device_iseries("viocd", i);
+	for (i = 0; i < HVMAXARCHITECTEDVIRTUALTAPES; i++)
+		vio_register_device_iseries("viotape", i);
+}
+
+/**
+ * vio_match_device_iseries: - Tell if a iSeries VIO device matches a
+ *	vio_device_id
+ */
+static int vio_match_device_iseries(const struct vio_device_id *id,
+		const struct vio_dev *dev)
+{
+	return strncmp(dev->type, id->type, strlen(id->type)) == 0;
+}
+
+static struct vio_bus_ops vio_bus_ops_iseries = {
+	.match = vio_match_device_iseries,
+};
+
+/**
+ * vio_bus_init_iseries: - Initialize the iSeries virtual IO bus
+ */
+static int __init vio_bus_init_iseries(void)
+{
+	int err;
+
+	err = vio_bus_init(&vio_bus_ops_iseries);
+	if (err == 0) {
+		iommu_vio_init();
+		vio_bus_device.iommu_table = &vio_iommu_table;
+		iSeries_vio_dev = &vio_bus_device.dev;
+		probe_bus_iseries();
+	}
+	return err;
+}
+
+__initcall(vio_bus_init_iseries);
diff --git a/arch/powerpc/platforms/iseries/viopath.c b/arch/powerpc/platforms/iseries/viopath.c
new file mode 100644
index 000000000000..fe97bfbf7463
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/viopath.c
@@ -0,0 +1,672 @@
+/* -*- linux-c -*-
+ *
+ *  iSeries Virtual I/O Message Path code
+ *
+ *  Authors: Dave Boutcher <boutcher@us.ibm.com>
+ *           Ryan Arnold <ryanarn@us.ibm.com>
+ *           Colin Devilbiss <devilbis@us.ibm.com>
+ *
+ * (C) Copyright 2000-2005 IBM Corporation
+ *
+ * This code is used by the iSeries virtual disk, cd,
+ * tape, and console to communicate with OS/400 in another
+ * partition.
+ *
+ * This program is free software;  you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of the
+ * License, or (at your option) anyu later version.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ *
+ */
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/vmalloc.h>
+#include <linux/string.h>
+#include <linux/proc_fs.h>
+#include <linux/dma-mapping.h>
+#include <linux/wait.h>
+#include <linux/seq_file.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+
+#include <asm/system.h>
+#include <asm/uaccess.h>
+#include <asm/iseries/hv_types.h>
+#include <asm/iseries/it_exp_vpd_panel.h>
+#include <asm/iseries/hv_lp_event.h>
+#include <asm/iseries/hv_lp_config.h>
+#include <asm/iseries/mf.h>
+#include <asm/iseries/vio.h>
+
+/* Status of the path to each other partition in the system.
+ * This is overkill, since we will only ever establish connections
+ * to our hosting partition and the primary partition on the system.
+ * But this allows for other support in the future.
+ */
+static struct viopathStatus {
+	int isOpen;		/* Did we open the path?            */
+	int isActive;		/* Do we have a mon msg outstanding */
+	int users[VIO_MAX_SUBTYPES];
+	HvLpInstanceId mSourceInst;
+	HvLpInstanceId mTargetInst;
+	int numberAllocated;
+} viopathStatus[HVMAXARCHITECTEDLPS];
+
+static DEFINE_SPINLOCK(statuslock);
+
+/*
+ * For each kind of event we allocate a buffer that is
+ * guaranteed not to cross a page boundary
+ */
+static unsigned char event_buffer[VIO_MAX_SUBTYPES * 256] __page_aligned;
+static atomic_t event_buffer_available[VIO_MAX_SUBTYPES];
+static int event_buffer_initialised;
+
+static void handleMonitorEvent(struct HvLpEvent *event);
+
+/*
+ * We use this structure to handle asynchronous responses.  The caller
+ * blocks on the semaphore and the handler posts the semaphore.  However,
+ * if system_state is not SYSTEM_RUNNING, then wait_atomic is used ...
+ */
+struct alloc_parms {
+	struct semaphore sem;
+	int number;
+	atomic_t wait_atomic;
+	int used_wait_atomic;
+};
+
+/* Put a sequence number in each mon msg.  The value is not
+ * important.  Start at something other than 0 just for
+ * readability.  wrapping this is ok.
+ */
+static u8 viomonseq = 22;
+
+/* Our hosting logical partition.  We get this at startup
+ * time, and different modules access this variable directly.
+ */
+HvLpIndex viopath_hostLp = HvLpIndexInvalid;
+EXPORT_SYMBOL(viopath_hostLp);
+HvLpIndex viopath_ourLp = HvLpIndexInvalid;
+EXPORT_SYMBOL(viopath_ourLp);
+
+/* For each kind of incoming event we set a pointer to a
+ * routine to call.
+ */
+static vio_event_handler_t *vio_handler[VIO_MAX_SUBTYPES];
+
+#define VIOPATH_KERN_WARN	KERN_WARNING "viopath: "
+#define VIOPATH_KERN_INFO	KERN_INFO "viopath: "
+
+static int proc_viopath_show(struct seq_file *m, void *v)
+{
+	char *buf;
+	u16 vlanMap;
+	dma_addr_t handle;
+	HvLpEvent_Rc hvrc;
+	DECLARE_MUTEX_LOCKED(Semaphore);
+
+	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!buf)
+		return 0;
+	memset(buf, 0, PAGE_SIZE);
+
+	handle = dma_map_single(iSeries_vio_dev, buf, PAGE_SIZE,
+				DMA_FROM_DEVICE);
+
+	hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp,
+			HvLpEvent_Type_VirtualIo,
+			viomajorsubtype_config | vioconfigget,
+			HvLpEvent_AckInd_DoAck, HvLpEvent_AckType_ImmediateAck,
+			viopath_sourceinst(viopath_hostLp),
+			viopath_targetinst(viopath_hostLp),
+			(u64)(unsigned long)&Semaphore, VIOVERSION << 16,
+			((u64)handle) << 32, PAGE_SIZE, 0, 0);
+
+	if (hvrc != HvLpEvent_Rc_Good)
+		printk(VIOPATH_KERN_WARN "hv error on op %d\n", (int)hvrc);
+
+	down(&Semaphore);
+
+	vlanMap = HvLpConfig_getVirtualLanIndexMap();
+
+	buf[PAGE_SIZE-1] = '\0';
+	seq_printf(m, "%s", buf);
+	seq_printf(m, "AVAILABLE_VETH=%x\n", vlanMap);
+	seq_printf(m, "SRLNBR=%c%c%c%c%c%c%c\n",
+		   e2a(xItExtVpdPanel.mfgID[2]),
+		   e2a(xItExtVpdPanel.mfgID[3]),
+		   e2a(xItExtVpdPanel.systemSerial[1]),
+		   e2a(xItExtVpdPanel.systemSerial[2]),
+		   e2a(xItExtVpdPanel.systemSerial[3]),
+		   e2a(xItExtVpdPanel.systemSerial[4]),
+		   e2a(xItExtVpdPanel.systemSerial[5]));
+
+	dma_unmap_single(iSeries_vio_dev, handle, PAGE_SIZE, DMA_FROM_DEVICE);
+	kfree(buf);
+
+	return 0;
+}
+
+static int proc_viopath_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, proc_viopath_show, NULL);
+}
+
+static struct file_operations proc_viopath_operations = {
+	.open		= proc_viopath_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static int __init vio_proc_init(void)
+{
+	struct proc_dir_entry *e;
+
+	e = create_proc_entry("iSeries/config", 0, NULL);
+	if (e)
+		e->proc_fops = &proc_viopath_operations;
+
+        return 0;
+}
+__initcall(vio_proc_init);
+
+/* See if a given LP is active.  Allow for invalid lps to be passed in
+ * and just return invalid
+ */
+int viopath_isactive(HvLpIndex lp)
+{
+	if (lp == HvLpIndexInvalid)
+		return 0;
+	if (lp < HVMAXARCHITECTEDLPS)
+		return viopathStatus[lp].isActive;
+	else
+		return 0;
+}
+EXPORT_SYMBOL(viopath_isactive);
+
+/*
+ * We cache the source and target instance ids for each
+ * partition.
+ */
+HvLpInstanceId viopath_sourceinst(HvLpIndex lp)
+{
+	return viopathStatus[lp].mSourceInst;
+}
+EXPORT_SYMBOL(viopath_sourceinst);
+
+HvLpInstanceId viopath_targetinst(HvLpIndex lp)
+{
+	return viopathStatus[lp].mTargetInst;
+}
+EXPORT_SYMBOL(viopath_targetinst);
+
+/*
+ * Send a monitor message.  This is a message with the acknowledge
+ * bit on that the other side will NOT explicitly acknowledge.  When
+ * the other side goes down, the hypervisor will acknowledge any
+ * outstanding messages....so we will know when the other side dies.
+ */
+static void sendMonMsg(HvLpIndex remoteLp)
+{
+	HvLpEvent_Rc hvrc;
+
+	viopathStatus[remoteLp].mSourceInst =
+		HvCallEvent_getSourceLpInstanceId(remoteLp,
+				HvLpEvent_Type_VirtualIo);
+	viopathStatus[remoteLp].mTargetInst =
+		HvCallEvent_getTargetLpInstanceId(remoteLp,
+				HvLpEvent_Type_VirtualIo);
+
+	/*
+	 * Deliberately ignore the return code here.  if we call this
+	 * more than once, we don't care.
+	 */
+	vio_setHandler(viomajorsubtype_monitor, handleMonitorEvent);
+
+	hvrc = HvCallEvent_signalLpEventFast(remoteLp, HvLpEvent_Type_VirtualIo,
+			viomajorsubtype_monitor, HvLpEvent_AckInd_DoAck,
+			HvLpEvent_AckType_DeferredAck,
+			viopathStatus[remoteLp].mSourceInst,
+			viopathStatus[remoteLp].mTargetInst,
+			viomonseq++, 0, 0, 0, 0, 0);
+
+	if (hvrc == HvLpEvent_Rc_Good)
+		viopathStatus[remoteLp].isActive = 1;
+	else {
+		printk(VIOPATH_KERN_WARN "could not connect to partition %d\n",
+				remoteLp);
+		viopathStatus[remoteLp].isActive = 0;
+	}
+}
+
+static void handleMonitorEvent(struct HvLpEvent *event)
+{
+	HvLpIndex remoteLp;
+	int i;
+
+	/*
+	 * This handler is _also_ called as part of the loop
+	 * at the end of this routine, so it must be able to
+	 * ignore NULL events...
+	 */
+	if (!event)
+		return;
+
+	/*
+	 * First see if this is just a normal monitor message from the
+	 * other partition
+	 */
+	if (event->xFlags.xFunction == HvLpEvent_Function_Int) {
+		remoteLp = event->xSourceLp;
+		if (!viopathStatus[remoteLp].isActive)
+			sendMonMsg(remoteLp);
+		return;
+	}
+
+	/*
+	 * This path is for an acknowledgement; the other partition
+	 * died
+	 */
+	remoteLp = event->xTargetLp;
+	if ((event->xSourceInstanceId != viopathStatus[remoteLp].mSourceInst) ||
+	    (event->xTargetInstanceId != viopathStatus[remoteLp].mTargetInst)) {
+		printk(VIOPATH_KERN_WARN "ignoring ack....mismatched instances\n");
+		return;
+	}
+
+	printk(VIOPATH_KERN_WARN "partition %d ended\n", remoteLp);
+
+	viopathStatus[remoteLp].isActive = 0;
+
+	/*
+	 * For each active handler, pass them a NULL
+	 * message to indicate that the other partition
+	 * died
+	 */
+	for (i = 0; i < VIO_MAX_SUBTYPES; i++) {
+		if (vio_handler[i] != NULL)
+			(*vio_handler[i])(NULL);
+	}
+}
+
+int vio_setHandler(int subtype, vio_event_handler_t *beh)
+{
+	subtype = subtype >> VIOMAJOR_SUBTYPE_SHIFT;
+	if ((subtype < 0) || (subtype >= VIO_MAX_SUBTYPES))
+		return -EINVAL;
+	if (vio_handler[subtype] != NULL)
+		return -EBUSY;
+	vio_handler[subtype] = beh;
+	return 0;
+}
+EXPORT_SYMBOL(vio_setHandler);
+
+int vio_clearHandler(int subtype)
+{
+	subtype = subtype >> VIOMAJOR_SUBTYPE_SHIFT;
+	if ((subtype < 0) || (subtype >= VIO_MAX_SUBTYPES))
+		return -EINVAL;
+	if (vio_handler[subtype] == NULL)
+		return -EAGAIN;
+	vio_handler[subtype] = NULL;
+	return 0;
+}
+EXPORT_SYMBOL(vio_clearHandler);
+
+static void handleConfig(struct HvLpEvent *event)
+{
+	if (!event)
+		return;
+	if (event->xFlags.xFunction == HvLpEvent_Function_Int) {
+		printk(VIOPATH_KERN_WARN
+		       "unexpected config request from partition %d",
+		       event->xSourceLp);
+
+		if ((event->xFlags.xFunction == HvLpEvent_Function_Int) &&
+		    (event->xFlags.xAckInd == HvLpEvent_AckInd_DoAck)) {
+			event->xRc = HvLpEvent_Rc_InvalidSubtype;
+			HvCallEvent_ackLpEvent(event);
+		}
+		return;
+	}
+
+	up((struct semaphore *)event->xCorrelationToken);
+}
+
+/*
+ * Initialization of the hosting partition
+ */
+void vio_set_hostlp(void)
+{
+	/*
+	 * If this has already been set then we DON'T want to either change
+	 * it or re-register the proc file system
+	 */
+	if (viopath_hostLp != HvLpIndexInvalid)
+		return;
+
+	/*
+	 * Figure out our hosting partition.  This isn't allowed to change
+	 * while we're active
+	 */
+	viopath_ourLp = HvLpConfig_getLpIndex();
+	viopath_hostLp = HvLpConfig_getHostingLpIndex(viopath_ourLp);
+
+	if (viopath_hostLp != HvLpIndexInvalid)
+		vio_setHandler(viomajorsubtype_config, handleConfig);
+}
+EXPORT_SYMBOL(vio_set_hostlp);
+
+static void vio_handleEvent(struct HvLpEvent *event, struct pt_regs *regs)
+{
+	HvLpIndex remoteLp;
+	int subtype = (event->xSubtype & VIOMAJOR_SUBTYPE_MASK)
+		>> VIOMAJOR_SUBTYPE_SHIFT;
+
+	if (event->xFlags.xFunction == HvLpEvent_Function_Int) {
+		remoteLp = event->xSourceLp;
+		/*
+		 * The isActive is checked because if the hosting partition
+		 * went down and came back up it would not be active but it
+		 * would have different source and target instances, in which
+		 * case we'd want to reset them.  This case really protects
+		 * against an unauthorized active partition sending interrupts
+		 * or acks to this linux partition.
+		 */
+		if (viopathStatus[remoteLp].isActive
+		    && (event->xSourceInstanceId !=
+			viopathStatus[remoteLp].mTargetInst)) {
+			printk(VIOPATH_KERN_WARN
+			       "message from invalid partition. "
+			       "int msg rcvd, source inst (%d) doesnt match (%d)\n",
+			       viopathStatus[remoteLp].mTargetInst,
+			       event->xSourceInstanceId);
+			return;
+		}
+
+		if (viopathStatus[remoteLp].isActive
+		    && (event->xTargetInstanceId !=
+			viopathStatus[remoteLp].mSourceInst)) {
+			printk(VIOPATH_KERN_WARN
+			       "message from invalid partition. "
+			       "int msg rcvd, target inst (%d) doesnt match (%d)\n",
+			       viopathStatus[remoteLp].mSourceInst,
+			       event->xTargetInstanceId);
+			return;
+		}
+	} else {
+		remoteLp = event->xTargetLp;
+		if (event->xSourceInstanceId !=
+		    viopathStatus[remoteLp].mSourceInst) {
+			printk(VIOPATH_KERN_WARN
+			       "message from invalid partition. "
+			       "ack msg rcvd, source inst (%d) doesnt match (%d)\n",
+			       viopathStatus[remoteLp].mSourceInst,
+			       event->xSourceInstanceId);
+			return;
+		}
+
+		if (event->xTargetInstanceId !=
+		    viopathStatus[remoteLp].mTargetInst) {
+			printk(VIOPATH_KERN_WARN
+			       "message from invalid partition. "
+			       "viopath: ack msg rcvd, target inst (%d) doesnt match (%d)\n",
+			       viopathStatus[remoteLp].mTargetInst,
+			       event->xTargetInstanceId);
+			return;
+		}
+	}
+
+	if (vio_handler[subtype] == NULL) {
+		printk(VIOPATH_KERN_WARN
+		       "unexpected virtual io event subtype %d from partition %d\n",
+		       event->xSubtype, remoteLp);
+		/* No handler.  Ack if necessary */
+		if ((event->xFlags.xFunction == HvLpEvent_Function_Int) &&
+		    (event->xFlags.xAckInd == HvLpEvent_AckInd_DoAck)) {
+			event->xRc = HvLpEvent_Rc_InvalidSubtype;
+			HvCallEvent_ackLpEvent(event);
+		}
+		return;
+	}
+
+	/* This innocuous little line is where all the real work happens */
+	(*vio_handler[subtype])(event);
+}
+
+static void viopath_donealloc(void *parm, int number)
+{
+	struct alloc_parms *parmsp = parm;
+
+	parmsp->number = number;
+	if (parmsp->used_wait_atomic)
+		atomic_set(&parmsp->wait_atomic, 0);
+	else
+		up(&parmsp->sem);
+}
+
+static int allocateEvents(HvLpIndex remoteLp, int numEvents)
+{
+	struct alloc_parms parms;
+
+	if (system_state != SYSTEM_RUNNING) {
+		parms.used_wait_atomic = 1;
+		atomic_set(&parms.wait_atomic, 1);
+	} else {
+		parms.used_wait_atomic = 0;
+		init_MUTEX_LOCKED(&parms.sem);
+	}
+	mf_allocate_lp_events(remoteLp, HvLpEvent_Type_VirtualIo, 250,	/* It would be nice to put a real number here! */
+			    numEvents, &viopath_donealloc, &parms);
+	if (system_state != SYSTEM_RUNNING) {
+		while (atomic_read(&parms.wait_atomic))
+			mb();
+	} else
+		down(&parms.sem);
+	return parms.number;
+}
+
+int viopath_open(HvLpIndex remoteLp, int subtype, int numReq)
+{
+	int i;
+	unsigned long flags;
+	int tempNumAllocated;
+
+	if ((remoteLp >= HVMAXARCHITECTEDLPS) || (remoteLp == HvLpIndexInvalid))
+		return -EINVAL;
+
+	subtype = subtype >> VIOMAJOR_SUBTYPE_SHIFT;
+	if ((subtype < 0) || (subtype >= VIO_MAX_SUBTYPES))
+		return -EINVAL;
+
+	spin_lock_irqsave(&statuslock, flags);
+
+	if (!event_buffer_initialised) {
+		for (i = 0; i < VIO_MAX_SUBTYPES; i++)
+			atomic_set(&event_buffer_available[i], 1);
+		event_buffer_initialised = 1;
+	}
+
+	viopathStatus[remoteLp].users[subtype]++;
+
+	if (!viopathStatus[remoteLp].isOpen) {
+		viopathStatus[remoteLp].isOpen = 1;
+		HvCallEvent_openLpEventPath(remoteLp, HvLpEvent_Type_VirtualIo);
+
+		/*
+		 * Don't hold the spinlock during an operation that
+		 * can sleep.
+		 */
+		spin_unlock_irqrestore(&statuslock, flags);
+		tempNumAllocated = allocateEvents(remoteLp, 1);
+		spin_lock_irqsave(&statuslock, flags);
+
+		viopathStatus[remoteLp].numberAllocated += tempNumAllocated;
+
+		if (viopathStatus[remoteLp].numberAllocated == 0) {
+			HvCallEvent_closeLpEventPath(remoteLp,
+					HvLpEvent_Type_VirtualIo);
+
+			spin_unlock_irqrestore(&statuslock, flags);
+			return -ENOMEM;
+		}
+
+		viopathStatus[remoteLp].mSourceInst =
+			HvCallEvent_getSourceLpInstanceId(remoteLp,
+					HvLpEvent_Type_VirtualIo);
+		viopathStatus[remoteLp].mTargetInst =
+			HvCallEvent_getTargetLpInstanceId(remoteLp,
+					HvLpEvent_Type_VirtualIo);
+		HvLpEvent_registerHandler(HvLpEvent_Type_VirtualIo,
+					  &vio_handleEvent);
+		sendMonMsg(remoteLp);
+		printk(VIOPATH_KERN_INFO "opening connection to partition %d, "
+				"setting sinst %d, tinst %d\n",
+				remoteLp, viopathStatus[remoteLp].mSourceInst,
+				viopathStatus[remoteLp].mTargetInst);
+	}
+
+	spin_unlock_irqrestore(&statuslock, flags);
+	tempNumAllocated = allocateEvents(remoteLp, numReq);
+	spin_lock_irqsave(&statuslock, flags);
+	viopathStatus[remoteLp].numberAllocated += tempNumAllocated;
+	spin_unlock_irqrestore(&statuslock, flags);
+
+	return 0;
+}
+EXPORT_SYMBOL(viopath_open);
+
+int viopath_close(HvLpIndex remoteLp, int subtype, int numReq)
+{
+	unsigned long flags;
+	int i;
+	int numOpen;
+	struct alloc_parms parms;
+
+	if ((remoteLp >= HVMAXARCHITECTEDLPS) || (remoteLp == HvLpIndexInvalid))
+		return -EINVAL;
+
+	subtype = subtype >> VIOMAJOR_SUBTYPE_SHIFT;
+	if ((subtype < 0) || (subtype >= VIO_MAX_SUBTYPES))
+		return -EINVAL;
+
+	spin_lock_irqsave(&statuslock, flags);
+	/*
+	 * If the viopath_close somehow gets called before a
+	 * viopath_open it could decrement to -1 which is a non
+	 * recoverable state so we'll prevent this from
+	 * happening.
+	 */
+	if (viopathStatus[remoteLp].users[subtype] > 0)
+		viopathStatus[remoteLp].users[subtype]--;
+
+	spin_unlock_irqrestore(&statuslock, flags);
+
+	parms.used_wait_atomic = 0;
+	init_MUTEX_LOCKED(&parms.sem);
+	mf_deallocate_lp_events(remoteLp, HvLpEvent_Type_VirtualIo,
+			      numReq, &viopath_donealloc, &parms);
+	down(&parms.sem);
+
+	spin_lock_irqsave(&statuslock, flags);
+	for (i = 0, numOpen = 0; i < VIO_MAX_SUBTYPES; i++)
+		numOpen += viopathStatus[remoteLp].users[i];
+
+	if ((viopathStatus[remoteLp].isOpen) && (numOpen == 0)) {
+		printk(VIOPATH_KERN_INFO "closing connection to partition %d",
+				remoteLp);
+
+		HvCallEvent_closeLpEventPath(remoteLp,
+					     HvLpEvent_Type_VirtualIo);
+		viopathStatus[remoteLp].isOpen = 0;
+		viopathStatus[remoteLp].isActive = 0;
+
+		for (i = 0; i < VIO_MAX_SUBTYPES; i++)
+			atomic_set(&event_buffer_available[i], 0);
+		event_buffer_initialised = 0;
+	}
+	spin_unlock_irqrestore(&statuslock, flags);
+	return 0;
+}
+EXPORT_SYMBOL(viopath_close);
+
+void *vio_get_event_buffer(int subtype)
+{
+	subtype = subtype >> VIOMAJOR_SUBTYPE_SHIFT;
+	if ((subtype < 0) || (subtype >= VIO_MAX_SUBTYPES))
+		return NULL;
+
+	if (atomic_dec_if_positive(&event_buffer_available[subtype]) == 0)
+		return &event_buffer[subtype * 256];
+	else
+		return NULL;
+}
+EXPORT_SYMBOL(vio_get_event_buffer);
+
+void vio_free_event_buffer(int subtype, void *buffer)
+{
+	subtype = subtype >> VIOMAJOR_SUBTYPE_SHIFT;
+	if ((subtype < 0) || (subtype >= VIO_MAX_SUBTYPES)) {
+		printk(VIOPATH_KERN_WARN
+		       "unexpected subtype %d freeing event buffer\n", subtype);
+		return;
+	}
+
+	if (atomic_read(&event_buffer_available[subtype]) != 0) {
+		printk(VIOPATH_KERN_WARN
+		       "freeing unallocated event buffer, subtype %d\n",
+		       subtype);
+		return;
+	}
+
+	if (buffer != &event_buffer[subtype * 256]) {
+		printk(VIOPATH_KERN_WARN
+		       "freeing invalid event buffer, subtype %d\n", subtype);
+	}
+
+	atomic_set(&event_buffer_available[subtype], 1);
+}
+EXPORT_SYMBOL(vio_free_event_buffer);
+
+static const struct vio_error_entry vio_no_error =
+    { 0, 0, "Non-VIO Error" };
+static const struct vio_error_entry vio_unknown_error =
+    { 0, EIO, "Unknown Error" };
+
+static const struct vio_error_entry vio_default_errors[] = {
+	{0x0001, EIO, "No Connection"},
+	{0x0002, EIO, "No Receiver"},
+	{0x0003, EIO, "No Buffer Available"},
+	{0x0004, EBADRQC, "Invalid Message Type"},
+	{0x0000, 0, NULL},
+};
+
+const struct vio_error_entry *vio_lookup_rc(
+		const struct vio_error_entry *local_table, u16 rc)
+{
+	const struct vio_error_entry *cur;
+
+	if (!rc)
+		return &vio_no_error;
+	if (local_table)
+		for (cur = local_table; cur->rc; ++cur)
+			if (cur->rc == rc)
+				return cur;
+	for (cur = vio_default_errors; cur->rc; ++cur)
+		if (cur->rc == rc)
+			return cur;
+	return &vio_unknown_error;
+}
+EXPORT_SYMBOL(vio_lookup_rc);
diff --git a/arch/powerpc/platforms/iseries/vpd_areas.h b/arch/powerpc/platforms/iseries/vpd_areas.h
new file mode 100644
index 000000000000..601e6dd860ed
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/vpd_areas.h
@@ -0,0 +1,88 @@
+/*
+ * Copyright (C) 2001  Mike Corrigan IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+#ifndef _ISERIES_VPD_AREAS_H
+#define _ISERIES_VPD_AREAS_H
+
+/*
+ * This file defines the address and length of all of the VPD area passed to
+ * the OS from PLIC (most of which start from the SP).
+ */
+
+#include <asm/types.h>
+
+/* VPD Entry index is carved in stone - cannot be changed (easily). */
+#define ItVpdCecVpd				0
+#define ItVpdDynamicSpace			1
+#define ItVpdExtVpd				2
+#define ItVpdExtVpdOnPanel			3
+#define ItVpdFirstPaca				4
+#define ItVpdIoVpd				5
+#define ItVpdIplParms				6
+#define ItVpdMsVpd				7
+#define ItVpdPanelVpd				8
+#define ItVpdLpNaca				9
+#define ItVpdBackplaneAndMaybeClockCardVpd	10
+#define ItVpdRecoveryLogBuffer			11
+#define ItVpdSpCommArea				12
+#define ItVpdSpLogBuffer			13
+#define ItVpdSpLogBufferSave			14
+#define ItVpdSpCardVpd				15
+#define ItVpdFirstProcVpd			16
+#define ItVpdApModelVpd				17
+#define ItVpdClockCardVpd			18
+#define ItVpdBusExtCardVpd			19
+#define ItVpdProcCapacityVpd			20
+#define ItVpdInteractiveCapacityVpd		21
+#define ItVpdFirstSlotLabel			22
+#define ItVpdFirstLpQueue			23
+#define ItVpdFirstL3CacheVpd			24
+#define ItVpdFirstProcFruVpd			25
+
+#define ItVpdMaxEntries				26
+
+#define ItDmaMaxEntries				10
+
+#define ItVpdAreasMaxSlotLabels			192
+
+
+struct ItVpdAreas {
+	u32	xSlicDesc;		// Descriptor			000-003
+	u16	xSlicSize;		// Size of this control block	004-005
+	u16	xPlicAdjustVpdLens:1;	// Flag to indicate new interface006-007
+	u16	xRsvd1:15;		// Reserved bits		...
+	u16	xSlicVpdEntries;	// Number of VPD entries	008-009
+	u16	xSlicDmaEntries;	// Number of DMA entries	00A-00B
+	u16	xSlicMaxLogicalProcs;	// Maximum logical processors	00C-00D
+	u16	xSlicMaxPhysicalProcs;	// Maximum physical processors	00E-00F
+	u16	xSlicDmaToksOffset;	// Offset into this of array	010-011
+	u16	xSlicVpdAdrsOffset;	// Offset into this of array	012-013
+	u16	xSlicDmaLensOffset;	// Offset into this of array	014-015
+	u16	xSlicVpdLensOffset;	// Offset into this of array	016-017
+	u16	xSlicMaxSlotLabels;	// Maximum number of slot labels018-019
+	u16	xSlicMaxLpQueues;	// Maximum number of LP Queues	01A-01B
+	u8	xRsvd2[4];		// Reserved			01C-01F
+	u64	xRsvd3[12];		// Reserved			020-07F
+	u32	xPlicDmaLens[ItDmaMaxEntries];// Array of DMA lengths	080-0A7
+	u32	xPlicDmaToks[ItDmaMaxEntries];// Array of DMA tokens	0A8-0CF
+	u32	xSlicVpdLens[ItVpdMaxEntries];// Array of VPD lengths	0D0-12F
+	void	*xSlicVpdAdrs[ItVpdMaxEntries];// Array of VPD buffers	130-1EF
+};
+
+extern struct ItVpdAreas	itVpdAreas;
+
+#endif /* _ISERIES_VPD_AREAS_H */
diff --git a/arch/powerpc/platforms/iseries/vpdinfo.c b/arch/powerpc/platforms/iseries/vpdinfo.c
new file mode 100644
index 000000000000..23a6d1e5b429
--- /dev/null
+++ b/arch/powerpc/platforms/iseries/vpdinfo.c
@@ -0,0 +1,271 @@
+/*
+ * This code gets the card location of the hardware
+ * Copyright (C) 2001  <Allan H Trautman> <IBM Corp>
+ * Copyright (C) 2005  Stephen Rothwel, IBM Corp
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the:
+ * Free Software Foundation, Inc.,
+ * 59 Temple Place, Suite 330,
+ * Boston, MA  02111-1307  USA
+ *
+ * Change Activity:
+ *   Created, Feb 2, 2001
+ *   Ported to ppc64, August 20, 2001
+ * End Change Activity
+ */
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include <asm/types.h>
+#include <asm/resource.h>
+#include <asm/abs_addr.h>
+#include <asm/pci-bridge.h>
+#include <asm/iseries/hv_types.h>
+
+#include "pci.h"
+#include "call_pci.h"
+
+/*
+ * Size of Bus VPD data
+ */
+#define BUS_VPDSIZE      1024
+
+/*
+ * Bus Vpd Tags
+ */
+#define  VpdEndOfAreaTag   0x79
+#define  VpdIdStringTag    0x82
+#define  VpdVendorAreaTag  0x84
+
+/*
+ * Mfg Area Tags
+ */
+#define  VpdFruFrameId    0x4649     // "FI"
+#define  VpdSlotMapFormat 0x4D46     // "MF"
+#define  VpdSlotMap       0x534D     // "SM"
+
+/*
+ * Structures of the areas
+ */
+struct MfgVpdAreaStruct {
+	u16 Tag;
+	u8  TagLength;
+	u8  AreaData1;
+	u8  AreaData2;
+};
+typedef struct MfgVpdAreaStruct MfgArea;
+#define MFG_ENTRY_SIZE   3
+
+struct SlotMapStruct {
+	u8   AgentId;
+	u8   SecondaryAgentId;
+	u8   PhbId;
+	char CardLocation[3];
+	char Parms[8];
+	char Reserved[2];
+};
+typedef struct SlotMapStruct SlotMap;
+#define SLOT_ENTRY_SIZE   16
+
+/*
+ * Parse the Slot Area
+ */
+static void __init iSeries_Parse_SlotArea(SlotMap *MapPtr, int MapLen,
+		HvAgentId agent, u8 *PhbId, char card[4])
+{
+	int SlotMapLen = MapLen;
+	SlotMap *SlotMapPtr = MapPtr;
+
+	/*
+	 * Parse Slot label until we find the one requested
+	 */
+	while (SlotMapLen > 0) {
+		if (SlotMapPtr->AgentId == agent) {
+			/*
+			 * If Phb wasn't found, grab the entry first one found.
+			 */
+			if (*PhbId == 0xff)
+				*PhbId = SlotMapPtr->PhbId;
+			/* Found it, extract the data. */
+			if (SlotMapPtr->PhbId == *PhbId) {
+				memcpy(card, &SlotMapPtr->CardLocation, 3);
+				card[3]  = 0;
+				break;
+			}
+		}
+		/* Point to the next Slot */
+		SlotMapPtr = (SlotMap *)((char *)SlotMapPtr + SLOT_ENTRY_SIZE);
+		SlotMapLen -= SLOT_ENTRY_SIZE;
+	}
+}
+
+/*
+ * Parse the Mfg Area
+ */
+static void __init iSeries_Parse_MfgArea(u8 *AreaData, int AreaLen,
+		HvAgentId agent, u8 *PhbId,
+		u8 *frame, char card[4])
+{
+	MfgArea *MfgAreaPtr = (MfgArea *)AreaData;
+	int MfgAreaLen = AreaLen;
+	u16 SlotMapFmt = 0;
+
+	/* Parse Mfg Data */
+	while (MfgAreaLen > 0) {
+		int MfgTagLen = MfgAreaPtr->TagLength;
+		/* Frame ID         (FI 4649020310 ) */
+		if (MfgAreaPtr->Tag == VpdFruFrameId)		/* FI  */
+			*frame = MfgAreaPtr->AreaData1;
+		/* Slot Map Format  (MF 4D46020004 ) */
+		else if (MfgAreaPtr->Tag == VpdSlotMapFormat)	/* MF  */
+			SlotMapFmt = (MfgAreaPtr->AreaData1 * 256)
+				+ MfgAreaPtr->AreaData2;
+		/* Slot Map         (SM 534D90 */
+		else if (MfgAreaPtr->Tag == VpdSlotMap)	{	/* SM  */
+			SlotMap *SlotMapPtr;
+
+			if (SlotMapFmt == 0x1004)
+				SlotMapPtr = (SlotMap *)((char *)MfgAreaPtr
+						+ MFG_ENTRY_SIZE + 1);
+			else
+				SlotMapPtr = (SlotMap *)((char *)MfgAreaPtr
+						+ MFG_ENTRY_SIZE);
+			iSeries_Parse_SlotArea(SlotMapPtr, MfgTagLen,
+					agent, PhbId, card);
+		}
+		/*
+		 * Point to the next Mfg Area
+		 * Use defined size, sizeof give wrong answer
+		 */
+		MfgAreaPtr = (MfgArea *)((char *)MfgAreaPtr + MfgTagLen
+				+ MFG_ENTRY_SIZE);
+		MfgAreaLen -= (MfgTagLen + MFG_ENTRY_SIZE);
+	}
+}
+
+/*
+ * Look for "BUS".. Data is not Null terminated.
+ * PHBID of 0xFF indicates PHB was not found in VPD Data.
+ */
+static int __init iSeries_Parse_PhbId(u8 *AreaPtr, int AreaLength)
+{
+	u8 *PhbPtr = AreaPtr;
+	int DataLen = AreaLength;
+	char PhbId = 0xFF;
+
+	while (DataLen > 0) {
+		if ((*PhbPtr == 'B') && (*(PhbPtr + 1) == 'U')
+				&& (*(PhbPtr + 2) == 'S')) {
+			PhbPtr += 3;
+			while (*PhbPtr == ' ')
+				++PhbPtr;
+			PhbId = (*PhbPtr & 0x0F);
+			break;
+		}
+		++PhbPtr;
+		--DataLen;
+	}
+	return PhbId;
+}
+
+/*
+ * Parse out the VPD Areas
+ */
+static void __init iSeries_Parse_Vpd(u8 *VpdData, int VpdDataLen,
+		HvAgentId agent, u8 *frame, char card[4])
+{
+	u8 *TagPtr = VpdData;
+	int DataLen = VpdDataLen - 3;
+	u8 PhbId;
+
+	while ((*TagPtr != VpdEndOfAreaTag) && (DataLen > 0)) {
+		int AreaLen = *(TagPtr + 1) + (*(TagPtr + 2) * 256);
+		u8 *AreaData  = TagPtr + 3;
+
+		if (*TagPtr == VpdIdStringTag)
+			PhbId = iSeries_Parse_PhbId(AreaData, AreaLen);
+		else if (*TagPtr == VpdVendorAreaTag)
+			iSeries_Parse_MfgArea(AreaData, AreaLen,
+					agent, &PhbId, frame, card);
+		/* Point to next Area. */
+		TagPtr  = AreaData + AreaLen;
+		DataLen -= AreaLen;
+	}
+}
+
+static void __init iSeries_Get_Location_Code(u16 bus, HvAgentId agent,
+		u8 *frame, char card[4])
+{
+	int BusVpdLen = 0;
+	u8 *BusVpdPtr = kmalloc(BUS_VPDSIZE, GFP_KERNEL);
+
+	if (BusVpdPtr == NULL) {
+		printk("PCI: Bus VPD Buffer allocation failure.\n");
+		return;
+	}
+	BusVpdLen = HvCallPci_getBusVpd(bus, iseries_hv_addr(BusVpdPtr),
+					BUS_VPDSIZE);
+	if (BusVpdLen == 0) {
+		printk("PCI: Bus VPD Buffer zero length.\n");
+		goto out_free;
+	}
+	/* printk("PCI: BusVpdPtr: %p, %d\n",BusVpdPtr, BusVpdLen); */
+	/* Make sure this is what I think it is */
+	if (*BusVpdPtr != VpdIdStringTag) {	/* 0x82 */
+		printk("PCI: Bus VPD Buffer missing starting tag.\n");
+		goto out_free;
+	}
+	iSeries_Parse_Vpd(BusVpdPtr, BusVpdLen, agent, frame, card);
+out_free:
+	kfree(BusVpdPtr);
+}
+
+/*
+ * Prints the device information.
+ * - Pass in pci_dev* pointer to the device.
+ * - Pass in the device count
+ *
+ * Format:
+ * PCI: Bus  0, Device 26, Vendor 0x12AE  Frame  1, Card  C10  Ethernet
+ * controller
+ */
+void __init iSeries_Device_Information(struct pci_dev *PciDev, int count)
+{
+	struct device_node *DevNode = PciDev->sysdata;
+	struct pci_dn *pdn;
+	u16 bus;
+	u8 frame;
+	char card[4];
+	HvSubBusNumber subbus;
+	HvAgentId agent;
+
+	if (DevNode == NULL) {
+		printk("%d. PCI: iSeries_Device_Information DevNode is NULL\n",
+				count);
+		return;
+	}
+
+	pdn = PCI_DN(DevNode);
+	bus = pdn->busno;
+	subbus = pdn->bussubno;
+	agent = ISERIES_PCI_AGENTID(ISERIES_GET_DEVICE_FROM_SUBBUS(subbus),
+			ISERIES_GET_FUNCTION_FROM_SUBBUS(subbus));
+	iSeries_Get_Location_Code(bus, agent, &frame, card);
+
+	printk("%d. PCI: Bus%3d, Device%3d, Vendor %04X Frame%3d, Card %4s  ",
+			count, bus, PCI_SLOT(PciDev->devfn), PciDev->vendor,
+			frame, card);
+	printk("0x%04X\n", (int)(PciDev->class >> 8));
+}
diff --git a/arch/powerpc/platforms/maple/Makefile b/arch/powerpc/platforms/maple/Makefile
new file mode 100644
index 000000000000..1be1a993c5f5
--- /dev/null
+++ b/arch/powerpc/platforms/maple/Makefile
@@ -0,0 +1 @@
+obj-y	+= setup.o pci.o time.o
diff --git a/arch/powerpc/platforms/maple/maple.h b/arch/powerpc/platforms/maple/maple.h
new file mode 100644
index 000000000000..0657c579b840
--- /dev/null
+++ b/arch/powerpc/platforms/maple/maple.h
@@ -0,0 +1,12 @@
+/*
+ * Declarations for maple-specific code.
+ *
+ * Maple is the name of a PPC970 evaluation board.
+ */
+extern int maple_set_rtc_time(struct rtc_time *tm);
+extern void maple_get_rtc_time(struct rtc_time *tm);
+extern unsigned long maple_get_boot_time(void);
+extern void maple_calibrate_decr(void);
+extern void maple_pci_init(void);
+extern void maple_pcibios_fixup(void);
+extern int maple_pci_get_legacy_ide_irq(struct pci_dev *dev, int channel);
diff --git a/arch/powerpc/platforms/maple/pci.c b/arch/powerpc/platforms/maple/pci.c
new file mode 100644
index 000000000000..340c21caeae2
--- /dev/null
+++ b/arch/powerpc/platforms/maple/pci.c
@@ -0,0 +1,522 @@
+/*
+ * Copyright (C) 2004 Benjamin Herrenschmuidt (benh@kernel.crashing.org),
+ *		      IBM Corp.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#define DEBUG
+
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/bootmem.h>
+
+#include <asm/sections.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/pci-bridge.h>
+#include <asm/machdep.h>
+#include <asm/iommu.h>
+#include <asm/ppc-pci.h>
+
+#include "maple.h"
+
+#ifdef DEBUG
+#define DBG(x...) printk(x)
+#else
+#define DBG(x...)
+#endif
+
+static struct pci_controller *u3_agp, *u3_ht;
+
+static int __init fixup_one_level_bus_range(struct device_node *node, int higher)
+{
+	for (; node != 0;node = node->sibling) {
+		int * bus_range;
+		unsigned int *class_code;
+		int len;
+
+		/* For PCI<->PCI bridges or CardBus bridges, we go down */
+		class_code = (unsigned int *) get_property(node, "class-code", NULL);
+		if (!class_code || ((*class_code >> 8) != PCI_CLASS_BRIDGE_PCI &&
+			(*class_code >> 8) != PCI_CLASS_BRIDGE_CARDBUS))
+			continue;
+		bus_range = (int *) get_property(node, "bus-range", &len);
+		if (bus_range != NULL && len > 2 * sizeof(int)) {
+			if (bus_range[1] > higher)
+				higher = bus_range[1];
+		}
+		higher = fixup_one_level_bus_range(node->child, higher);
+	}
+	return higher;
+}
+
+/* This routine fixes the "bus-range" property of all bridges in the
+ * system since they tend to have their "last" member wrong on macs
+ *
+ * Note that the bus numbers manipulated here are OF bus numbers, they
+ * are not Linux bus numbers.
+ */
+static void __init fixup_bus_range(struct device_node *bridge)
+{
+	int * bus_range;
+	int len;
+
+	/* Lookup the "bus-range" property for the hose */
+	bus_range = (int *) get_property(bridge, "bus-range", &len);
+	if (bus_range == NULL || len < 2 * sizeof(int)) {
+		printk(KERN_WARNING "Can't get bus-range for %s\n",
+			       bridge->full_name);
+		return;
+	}
+	bus_range[1] = fixup_one_level_bus_range(bridge->child, bus_range[1]);
+}
+
+
+#define U3_AGP_CFA0(devfn, off)	\
+	((1 << (unsigned long)PCI_SLOT(dev_fn)) \
+	| (((unsigned long)PCI_FUNC(dev_fn)) << 8) \
+	| (((unsigned long)(off)) & 0xFCUL))
+
+#define U3_AGP_CFA1(bus, devfn, off)	\
+	((((unsigned long)(bus)) << 16) \
+	|(((unsigned long)(devfn)) << 8) \
+	|(((unsigned long)(off)) & 0xFCUL) \
+	|1UL)
+
+static unsigned long u3_agp_cfg_access(struct pci_controller* hose,
+				       u8 bus, u8 dev_fn, u8 offset)
+{
+	unsigned int caddr;
+
+	if (bus == hose->first_busno) {
+		if (dev_fn < (11 << 3))
+			return 0;
+		caddr = U3_AGP_CFA0(dev_fn, offset);
+	} else
+		caddr = U3_AGP_CFA1(bus, dev_fn, offset);
+
+	/* Uninorth will return garbage if we don't read back the value ! */
+	do {
+		out_le32(hose->cfg_addr, caddr);
+	} while (in_le32(hose->cfg_addr) != caddr);
+
+	offset &= 0x07;
+	return ((unsigned long)hose->cfg_data) + offset;
+}
+
+static int u3_agp_read_config(struct pci_bus *bus, unsigned int devfn,
+			      int offset, int len, u32 *val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = u3_agp_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		*val = in_8((u8 *)addr);
+		break;
+	case 2:
+		*val = in_le16((u16 *)addr);
+		break;
+	default:
+		*val = in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static int u3_agp_write_config(struct pci_bus *bus, unsigned int devfn,
+			       int offset, int len, u32 val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = u3_agp_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		out_8((u8 *)addr, val);
+		(void) in_8((u8 *)addr);
+		break;
+	case 2:
+		out_le16((u16 *)addr, val);
+		(void) in_le16((u16 *)addr);
+		break;
+	default:
+		out_le32((u32 *)addr, val);
+		(void) in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static struct pci_ops u3_agp_pci_ops =
+{
+	u3_agp_read_config,
+	u3_agp_write_config
+};
+
+
+#define U3_HT_CFA0(devfn, off)		\
+		((((unsigned long)devfn) << 8) | offset)
+#define U3_HT_CFA1(bus, devfn, off)	\
+		(U3_HT_CFA0(devfn, off) \
+		+ (((unsigned long)bus) << 16) \
+		+ 0x01000000UL)
+
+static unsigned long u3_ht_cfg_access(struct pci_controller* hose,
+				      u8 bus, u8 devfn, u8 offset)
+{
+	if (bus == hose->first_busno) {
+		if (PCI_SLOT(devfn) == 0)
+			return 0;
+		return ((unsigned long)hose->cfg_data) + U3_HT_CFA0(devfn, offset);
+	} else
+		return ((unsigned long)hose->cfg_data) + U3_HT_CFA1(bus, devfn, offset);
+}
+
+static int u3_ht_read_config(struct pci_bus *bus, unsigned int devfn,
+			     int offset, int len, u32 *val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = u3_ht_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		*val = in_8((u8 *)addr);
+		break;
+	case 2:
+		*val = in_le16((u16 *)addr);
+		break;
+	default:
+		*val = in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static int u3_ht_write_config(struct pci_bus *bus, unsigned int devfn,
+			      int offset, int len, u32 val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = u3_ht_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		out_8((u8 *)addr, val);
+		(void) in_8((u8 *)addr);
+		break;
+	case 2:
+		out_le16((u16 *)addr, val);
+		(void) in_le16((u16 *)addr);
+		break;
+	default:
+		out_le32((u32 *)addr, val);
+		(void) in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static struct pci_ops u3_ht_pci_ops =
+{
+	u3_ht_read_config,
+	u3_ht_write_config
+};
+
+static void __init setup_u3_agp(struct pci_controller* hose)
+{
+	/* On G5, we move AGP up to high bus number so we don't need
+	 * to reassign bus numbers for HT. If we ever have P2P bridges
+	 * on AGP, we'll have to move pci_assign_all_buses to the
+	 * pci_controller structure so we enable it for AGP and not for
+	 * HT childs.
+	 * We hard code the address because of the different size of
+	 * the reg address cell, we shall fix that by killing struct
+	 * reg_property and using some accessor functions instead
+	 */
+	hose->first_busno = 0xf0;
+	hose->last_busno = 0xff;
+	hose->ops = &u3_agp_pci_ops;
+	hose->cfg_addr = ioremap(0xf0000000 + 0x800000, 0x1000);
+	hose->cfg_data = ioremap(0xf0000000 + 0xc00000, 0x1000);
+
+	u3_agp = hose;
+}
+
+static void __init setup_u3_ht(struct pci_controller* hose)
+{
+	hose->ops = &u3_ht_pci_ops;
+
+	/* We hard code the address because of the different size of
+	 * the reg address cell, we shall fix that by killing struct
+	 * reg_property and using some accessor functions instead
+	 */
+	hose->cfg_data = (volatile unsigned char *)ioremap(0xf2000000, 0x02000000);
+
+	hose->first_busno = 0;
+	hose->last_busno = 0xef;
+
+	u3_ht = hose;
+}
+
+static int __init add_bridge(struct device_node *dev)
+{
+	int len;
+	struct pci_controller *hose;
+	char* disp_name;
+	int *bus_range;
+	int primary = 1;
+	struct property *of_prop;
+
+	DBG("Adding PCI host bridge %s\n", dev->full_name);
+
+	bus_range = (int *) get_property(dev, "bus-range", &len);
+	if (bus_range == NULL || len < 2 * sizeof(int)) {
+		printk(KERN_WARNING "Can't get bus-range for %s, assume bus 0\n",
+		dev->full_name);
+	}
+
+	hose = alloc_bootmem(sizeof(struct pci_controller));
+	if (hose == NULL)
+		return -ENOMEM;
+	pci_setup_pci_controller(hose);
+
+	hose->arch_data = dev;
+	hose->first_busno = bus_range ? bus_range[0] : 0;
+	hose->last_busno = bus_range ? bus_range[1] : 0xff;
+
+	of_prop = alloc_bootmem(sizeof(struct property) +
+				sizeof(hose->global_number));
+	if (of_prop) {
+		memset(of_prop, 0, sizeof(struct property));
+		of_prop->name = "linux,pci-domain";
+		of_prop->length = sizeof(hose->global_number);
+		of_prop->value = (unsigned char *)&of_prop[1];
+		memcpy(of_prop->value, &hose->global_number, sizeof(hose->global_number));
+		prom_add_property(dev, of_prop);
+	}
+
+	disp_name = NULL;
+	if (device_is_compatible(dev, "u3-agp")) {
+		setup_u3_agp(hose);
+		disp_name = "U3-AGP";
+		primary = 0;
+	} else if (device_is_compatible(dev, "u3-ht")) {
+		setup_u3_ht(hose);
+		disp_name = "U3-HT";
+		primary = 1;
+	}
+	printk(KERN_INFO "Found %s PCI host bridge. Firmware bus number: %d->%d\n",
+		disp_name, hose->first_busno, hose->last_busno);
+
+	/* Interpret the "ranges" property */
+	/* This also maps the I/O region and sets isa_io/mem_base */
+	pci_process_bridge_OF_ranges(hose, dev, primary);
+	pci_setup_phb_io(hose, primary);
+
+	/* Fixup "bus-range" OF property */
+	fixup_bus_range(dev);
+
+	return 0;
+}
+
+
+void __init maple_pcibios_fixup(void)
+{
+	struct pci_dev *dev = NULL;
+
+	DBG(" -> maple_pcibios_fixup\n");
+
+	for_each_pci_dev(dev)
+		pci_read_irq_line(dev);
+
+	/* Do the mapping of the IO space */
+	phbs_remap_io();
+
+	DBG(" <- maple_pcibios_fixup\n");
+}
+
+static void __init maple_fixup_phb_resources(void)
+{
+	struct pci_controller *hose, *tmp;
+	
+	list_for_each_entry_safe(hose, tmp, &hose_list, list_node) {
+		unsigned long offset = (unsigned long)hose->io_base_virt - pci_io_base;
+		hose->io_resource.start += offset;
+		hose->io_resource.end += offset;
+		printk(KERN_INFO "PCI Host %d, io start: %lx; io end: %lx\n",
+		       hose->global_number,
+		       hose->io_resource.start, hose->io_resource.end);
+	}
+}
+
+void __init maple_pci_init(void)
+{
+	struct device_node *np, *root;
+	struct device_node *ht = NULL;
+
+	/* Probe root PCI hosts, that is on U3 the AGP host and the
+	 * HyperTransport host. That one is actually "kept" around
+	 * and actually added last as it's resource management relies
+	 * on the AGP resources to have been setup first
+	 */
+	root = of_find_node_by_path("/");
+	if (root == NULL) {
+		printk(KERN_CRIT "maple_find_bridges: can't find root of device tree\n");
+		return;
+	}
+	for (np = NULL; (np = of_get_next_child(root, np)) != NULL;) {
+		if (np->name == NULL)
+			continue;
+		if (strcmp(np->name, "pci") == 0) {
+			if (add_bridge(np) == 0)
+				of_node_get(np);
+		}
+		if (strcmp(np->name, "ht") == 0) {
+			of_node_get(np);
+			ht = np;
+		}
+	}
+	of_node_put(root);
+
+	/* Now setup the HyperTransport host if we found any
+	 */
+	if (ht && add_bridge(ht) != 0)
+		of_node_put(ht);
+
+	/* Fixup the IO resources on our host bridges as the common code
+	 * does it only for childs of the host bridges
+	 */
+	maple_fixup_phb_resources();
+
+	/* Setup the linkage between OF nodes and PHBs */ 
+	pci_devs_phb_init();
+
+	/* Fixup the PCI<->OF mapping for U3 AGP due to bus renumbering. We
+	 * assume there is no P2P bridge on the AGP bus, which should be a
+	 * safe assumptions hopefully.
+	 */
+	if (u3_agp) {
+		struct device_node *np = u3_agp->arch_data;
+		PCI_DN(np)->busno = 0xf0;
+		for (np = np->child; np; np = np->sibling)
+			PCI_DN(np)->busno = 0xf0;
+	}
+
+	/* Tell pci.c to use the common resource allocation mecanism */
+	pci_probe_only = 0;
+	
+	/* Allow all IO */
+	io_page_mask = -1;
+}
+
+int maple_pci_get_legacy_ide_irq(struct pci_dev *pdev, int channel)
+{
+	struct device_node *np;
+	int irq = channel ? 15 : 14;
+
+	if (pdev->vendor != PCI_VENDOR_ID_AMD ||
+	    pdev->device != PCI_DEVICE_ID_AMD_8111_IDE)
+		return irq;
+
+	np = pci_device_to_OF_node(pdev);
+	if (np == NULL)
+		return irq;
+	if (np->n_intrs < 2)
+		return irq;
+	return np->intrs[channel & 0x1].line;
+}
+
+/* XXX: To remove once all firmwares are ok */
+static void fixup_maple_ide(struct pci_dev* dev)
+{
+#if 0 /* Enable this to enable IDE port 0 */
+	{
+		u8 v;
+
+		pci_read_config_byte(dev, 0x40, &v);
+		v |= 2;
+		pci_write_config_byte(dev, 0x40, v);
+	}
+#endif
+#if 0 /* fix bus master base */
+	pci_write_config_dword(dev, 0x20, 0xcc01);
+	printk("old ide resource: %lx -> %lx \n",
+	       dev->resource[4].start, dev->resource[4].end);
+	dev->resource[4].start = 0xcc00;
+	dev->resource[4].end = 0xcc10;
+#endif
+#if 1 /* Enable this to fixup IDE sense/polarity of irqs in IO-APICs */
+	{
+		struct pci_dev *apicdev;
+		u32 v;
+
+		apicdev = pci_get_slot (dev->bus, PCI_DEVFN(5,0));
+		if (apicdev == NULL)
+			printk("IDE Fixup IRQ: Can't find IO-APIC !\n");
+		else {
+			pci_write_config_byte(apicdev, 0xf2, 0x10 + 2*14);
+			pci_read_config_dword(apicdev, 0xf4, &v);
+			v &= ~0x00000022;
+			pci_write_config_dword(apicdev, 0xf4, v);
+			pci_write_config_byte(apicdev, 0xf2, 0x10 + 2*15);
+			pci_read_config_dword(apicdev, 0xf4, &v);
+			v &= ~0x00000022;
+			pci_write_config_dword(apicdev, 0xf4, v);
+			pci_dev_put(apicdev);
+		}
+	}
+#endif
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_8111_IDE,
+			 fixup_maple_ide);
diff --git a/arch/powerpc/platforms/maple/setup.c b/arch/powerpc/platforms/maple/setup.c
new file mode 100644
index 000000000000..7ece8983a105
--- /dev/null
+++ b/arch/powerpc/platforms/maple/setup.c
@@ -0,0 +1,295 @@
+/*
+ *  Maple (970 eval board) setup code
+ *
+ *  (c) Copyright 2004 Benjamin Herrenschmidt (benh@kernel.crashing.org),
+ *                     IBM Corp. 
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ */
+
+#define DEBUG
+
+#include <linux/config.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/slab.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/string.h>
+#include <linux/delay.h>
+#include <linux/ioport.h>
+#include <linux/major.h>
+#include <linux/initrd.h>
+#include <linux/vt_kern.h>
+#include <linux/console.h>
+#include <linux/ide.h>
+#include <linux/pci.h>
+#include <linux/adb.h>
+#include <linux/cuda.h>
+#include <linux/pmu.h>
+#include <linux/irq.h>
+#include <linux/seq_file.h>
+#include <linux/root_dev.h>
+#include <linux/serial.h>
+#include <linux/smp.h>
+
+#include <asm/processor.h>
+#include <asm/sections.h>
+#include <asm/prom.h>
+#include <asm/system.h>
+#include <asm/pgtable.h>
+#include <asm/bitops.h>
+#include <asm/io.h>
+#include <asm/pci-bridge.h>
+#include <asm/iommu.h>
+#include <asm/machdep.h>
+#include <asm/dma.h>
+#include <asm/cputable.h>
+#include <asm/time.h>
+#include <asm/of_device.h>
+#include <asm/lmb.h>
+#include <asm/mpic.h>
+#include <asm/udbg.h>
+
+#include "maple.h"
+
+#ifdef DEBUG
+#define DBG(fmt...) udbg_printf(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+extern void generic_find_legacy_serial_ports(u64 *physport,
+		unsigned int *default_speed);
+
+static void maple_restart(char *cmd)
+{
+	unsigned int maple_nvram_base;
+	unsigned int maple_nvram_offset;
+	unsigned int maple_nvram_command;
+	struct device_node *rtcs;
+
+	/* find NVRAM device */
+	rtcs = find_compatible_devices("nvram", "AMD8111");
+	if (rtcs && rtcs->addrs) {
+		maple_nvram_base = rtcs->addrs[0].address;
+	} else {
+		printk(KERN_EMERG "Maple: Unable to find NVRAM\n");
+		printk(KERN_EMERG "Maple: Manual Restart Required\n");
+		return;
+	}
+
+	/* find service processor device */
+	rtcs = find_devices("service-processor");
+	if (!rtcs) {
+		printk(KERN_EMERG "Maple: Unable to find Service Processor\n");
+		printk(KERN_EMERG "Maple: Manual Restart Required\n");
+		return;
+	}
+	maple_nvram_offset = *(unsigned int*) get_property(rtcs,
+			"restart-addr", NULL);
+	maple_nvram_command = *(unsigned int*) get_property(rtcs,
+			"restart-value", NULL);
+
+	/* send command */
+	outb_p(maple_nvram_command, maple_nvram_base + maple_nvram_offset);
+	for (;;) ;
+}
+
+static void maple_power_off(void)
+{
+	unsigned int maple_nvram_base;
+	unsigned int maple_nvram_offset;
+	unsigned int maple_nvram_command;
+	struct device_node *rtcs;
+
+	/* find NVRAM device */
+	rtcs = find_compatible_devices("nvram", "AMD8111");
+	if (rtcs && rtcs->addrs) {
+		maple_nvram_base = rtcs->addrs[0].address;
+	} else {
+		printk(KERN_EMERG "Maple: Unable to find NVRAM\n");
+		printk(KERN_EMERG "Maple: Manual Power-Down Required\n");
+		return;
+	}
+
+	/* find service processor device */
+	rtcs = find_devices("service-processor");
+	if (!rtcs) {
+		printk(KERN_EMERG "Maple: Unable to find Service Processor\n");
+		printk(KERN_EMERG "Maple: Manual Power-Down Required\n");
+		return;
+	}
+	maple_nvram_offset = *(unsigned int*) get_property(rtcs,
+			"power-off-addr", NULL);
+	maple_nvram_command = *(unsigned int*) get_property(rtcs,
+			"power-off-value", NULL);
+
+	/* send command */
+	outb_p(maple_nvram_command, maple_nvram_base + maple_nvram_offset);
+	for (;;) ;
+}
+
+static void maple_halt(void)
+{
+	maple_power_off();
+}
+
+#ifdef CONFIG_SMP
+struct smp_ops_t maple_smp_ops = {
+	.probe		= smp_mpic_probe,
+	.message_pass	= smp_mpic_message_pass,
+	.kick_cpu	= smp_generic_kick_cpu,
+	.setup_cpu	= smp_mpic_setup_cpu,
+	.give_timebase	= smp_generic_give_timebase,
+	.take_timebase	= smp_generic_take_timebase,
+};
+#endif /* CONFIG_SMP */
+
+void __init maple_setup_arch(void)
+{
+	/* init to some ~sane value until calibrate_delay() runs */
+	loops_per_jiffy = 50000000;
+
+	/* Setup SMP callback */
+#ifdef CONFIG_SMP
+	smp_ops = &maple_smp_ops;
+#endif
+	/* Lookup PCI hosts */
+       	maple_pci_init();
+
+#ifdef CONFIG_DUMMY_CONSOLE
+	conswitchp = &dummy_con;
+#endif
+
+	printk(KERN_INFO "Using native/NAP idle loop\n");
+}
+
+/* 
+ * Early initialization.
+ */
+static void __init maple_init_early(void)
+{
+	unsigned int default_speed;
+	u64 physport;
+
+	DBG(" -> maple_init_early\n");
+
+	/* Initialize hash table, from now on, we can take hash faults
+	 * and call ioremap
+	 */
+	hpte_init_native();
+
+	/* Find the serial port */
+	generic_find_legacy_serial_ports(&physport, &default_speed);
+
+	DBG("phys port addr: %lx\n", (long)physport);
+
+	if (physport) {
+		void *comport;
+		/* Map the uart for udbg. */
+		comport = (void *)ioremap(physport, 16);
+		udbg_init_uart(comport, default_speed);
+
+		DBG("Hello World !\n");
+	}
+
+	/* Setup interrupt mapping options */
+	ppc64_interrupt_controller = IC_OPEN_PIC;
+
+	iommu_init_early_u3();
+
+	DBG(" <- maple_init_early\n");
+}
+
+
+static __init void maple_init_IRQ(void)
+{
+	struct device_node *root;
+	unsigned int *opprop;
+	unsigned long opic_addr;
+	struct mpic *mpic;
+	unsigned char senses[128];
+	int n;
+
+	DBG(" -> maple_init_IRQ\n");
+
+	/* XXX: Non standard, replace that with a proper openpic/mpic node
+	 * in the device-tree. Find the Open PIC if present */
+	root = of_find_node_by_path("/");
+	opprop = (unsigned int *) get_property(root,
+				"platform-open-pic", NULL);
+	if (opprop == 0)
+		panic("OpenPIC not found !\n");
+
+	n = prom_n_addr_cells(root);
+	for (opic_addr = 0; n > 0; --n)
+		opic_addr = (opic_addr << 32) + *opprop++;
+	of_node_put(root);
+
+	/* Obtain sense values from device-tree */
+	prom_get_irq_senses(senses, 0, 128);
+
+	mpic = mpic_alloc(opic_addr,
+			  MPIC_PRIMARY | MPIC_BIG_ENDIAN |
+			  MPIC_BROKEN_U3 | MPIC_WANTS_RESET,
+			  0, 0, 128, 128, senses, 128, "U3-MPIC");
+	BUG_ON(mpic == NULL);
+	mpic_init(mpic);
+
+	DBG(" <- maple_init_IRQ\n");
+}
+
+static void __init maple_progress(char *s, unsigned short hex)
+{
+	printk("*** %04x : %s\n", hex, s ? s : "");
+}
+
+
+/*
+ * Called very early, MMU is off, device-tree isn't unflattened
+ */
+static int __init maple_probe(int platform)
+{
+	if (platform != PLATFORM_MAPLE)
+		return 0;
+	/*
+	 * On U3, the DART (iommu) must be allocated now since it
+	 * has an impact on htab_initialize (due to the large page it
+	 * occupies having to be broken up so the DART itself is not
+	 * part of the cacheable linar mapping
+	 */
+	alloc_u3_dart_table();
+
+	return 1;
+}
+
+struct machdep_calls __initdata maple_md = {
+	.probe			= maple_probe,
+	.setup_arch		= maple_setup_arch,
+	.init_early		= maple_init_early,
+	.init_IRQ		= maple_init_IRQ,
+	.get_irq		= mpic_get_irq,
+	.pcibios_fixup		= maple_pcibios_fixup,
+	.pci_get_legacy_ide_irq	= maple_pci_get_legacy_ide_irq,
+	.restart		= maple_restart,
+	.power_off		= maple_power_off,
+	.halt			= maple_halt,
+       	.get_boot_time		= maple_get_boot_time,
+       	.set_rtc_time		= maple_set_rtc_time,
+       	.get_rtc_time		= maple_get_rtc_time,
+      	.calibrate_decr		= generic_calibrate_decr,
+	.progress		= maple_progress,
+	.idle_loop		= native_idle,
+};
diff --git a/arch/powerpc/platforms/maple/time.c b/arch/powerpc/platforms/maple/time.c
new file mode 100644
index 000000000000..40fc07a8e606
--- /dev/null
+++ b/arch/powerpc/platforms/maple/time.c
@@ -0,0 +1,180 @@
+/*
+ *  arch/ppc64/kernel/maple_time.c
+ *
+ *  (c) Copyright 2004 Benjamin Herrenschmidt (benh@kernel.crashing.org),
+ *                     IBM Corp. 
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ */
+
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/init.h>
+#include <linux/time.h>
+#include <linux/adb.h>
+#include <linux/pmu.h>
+#include <linux/interrupt.h>
+#include <linux/mc146818rtc.h>
+#include <linux/bcd.h>
+
+#include <asm/sections.h>
+#include <asm/prom.h>
+#include <asm/system.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/machdep.h>
+#include <asm/time.h>
+
+#include "maple.h"
+
+#ifdef DEBUG
+#define DBG(x...) printk(x)
+#else
+#define DBG(x...)
+#endif
+
+extern void GregorianDay(struct rtc_time * tm);
+
+static int maple_rtc_addr;
+
+static int maple_clock_read(int addr)
+{
+	outb_p(addr, maple_rtc_addr);
+	return inb_p(maple_rtc_addr+1);
+}
+
+static void maple_clock_write(unsigned long val, int addr)
+{
+	outb_p(addr, maple_rtc_addr);
+	outb_p(val, maple_rtc_addr+1);
+}
+
+void maple_get_rtc_time(struct rtc_time *tm)
+{
+	int uip, i;
+
+	/* The Linux interpretation of the CMOS clock register contents:
+	 * When the Update-In-Progress (UIP) flag goes from 1 to 0, the
+	 * RTC registers show the second which has precisely just started.
+	 * Let's hope other operating systems interpret the RTC the same way.
+	 */
+
+	/* Since the UIP flag is set for about 2.2 ms and the clock
+	 * is typically written with a precision of 1 jiffy, trying
+	 * to obtain a precision better than a few milliseconds is
+	 * an illusion. Only consistency is interesting, this also
+	 * allows to use the routine for /dev/rtc without a potential
+	 * 1 second kernel busy loop triggered by any reader of /dev/rtc.
+	 */
+
+	for (i = 0; i<1000000; i++) {
+		uip = maple_clock_read(RTC_FREQ_SELECT);
+		tm->tm_sec = maple_clock_read(RTC_SECONDS);
+		tm->tm_min = maple_clock_read(RTC_MINUTES);
+		tm->tm_hour = maple_clock_read(RTC_HOURS);
+		tm->tm_mday = maple_clock_read(RTC_DAY_OF_MONTH);
+		tm->tm_mon = maple_clock_read(RTC_MONTH);
+		tm->tm_year = maple_clock_read(RTC_YEAR);
+		uip |= maple_clock_read(RTC_FREQ_SELECT);
+		if ((uip & RTC_UIP)==0)
+			break;
+	}
+
+	if (!(maple_clock_read(RTC_CONTROL) & RTC_DM_BINARY)
+	    || RTC_ALWAYS_BCD) {
+		BCD_TO_BIN(tm->tm_sec);
+		BCD_TO_BIN(tm->tm_min);
+		BCD_TO_BIN(tm->tm_hour);
+		BCD_TO_BIN(tm->tm_mday);
+		BCD_TO_BIN(tm->tm_mon);
+		BCD_TO_BIN(tm->tm_year);
+	  }
+	if ((tm->tm_year + 1900) < 1970)
+		tm->tm_year += 100;
+
+	GregorianDay(tm);
+}
+
+int maple_set_rtc_time(struct rtc_time *tm)
+{
+	unsigned char save_control, save_freq_select;
+	int sec, min, hour, mon, mday, year;
+
+	spin_lock(&rtc_lock);
+
+	save_control = maple_clock_read(RTC_CONTROL); /* tell the clock it's being set */
+
+	maple_clock_write((save_control|RTC_SET), RTC_CONTROL);
+
+	save_freq_select = maple_clock_read(RTC_FREQ_SELECT); /* stop and reset prescaler */
+
+	maple_clock_write((save_freq_select|RTC_DIV_RESET2), RTC_FREQ_SELECT);
+
+	sec = tm->tm_sec;
+	min = tm->tm_min;
+	hour = tm->tm_hour;
+	mon = tm->tm_mon;
+	mday = tm->tm_mday;
+	year = tm->tm_year;
+
+	if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
+		BIN_TO_BCD(sec);
+		BIN_TO_BCD(min);
+		BIN_TO_BCD(hour);
+		BIN_TO_BCD(mon);
+		BIN_TO_BCD(mday);
+		BIN_TO_BCD(year);
+	}
+	maple_clock_write(sec, RTC_SECONDS);
+	maple_clock_write(min, RTC_MINUTES);
+	maple_clock_write(hour, RTC_HOURS);
+	maple_clock_write(mon, RTC_MONTH);
+	maple_clock_write(mday, RTC_DAY_OF_MONTH);
+	maple_clock_write(year, RTC_YEAR);
+
+	/* The following flags have to be released exactly in this order,
+	 * otherwise the DS12887 (popular MC146818A clone with integrated
+	 * battery and quartz) will not reset the oscillator and will not
+	 * update precisely 500 ms later. You won't find this mentioned in
+	 * the Dallas Semiconductor data sheets, but who believes data
+	 * sheets anyway ...                           -- Markus Kuhn
+	 */
+	maple_clock_write(save_control, RTC_CONTROL);
+	maple_clock_write(save_freq_select, RTC_FREQ_SELECT);
+
+	spin_unlock(&rtc_lock);
+
+	return 0;
+}
+
+unsigned long __init maple_get_boot_time(void)
+{
+	struct rtc_time tm;
+	struct device_node *rtcs;
+
+	rtcs = find_compatible_devices("rtc", "pnpPNP,b00");
+	if (rtcs && rtcs->addrs) {
+		maple_rtc_addr = rtcs->addrs[0].address;
+		printk(KERN_INFO "Maple: Found RTC at 0x%x\n", maple_rtc_addr);
+	} else {
+		maple_rtc_addr = RTC_PORT(0); /* legacy address */
+		printk(KERN_INFO "Maple: No device node for RTC, assuming "
+		       "legacy address (0x%x)\n", maple_rtc_addr);
+	}
+	
+	maple_get_rtc_time(&tm);
+	return mktime(tm.tm_year+1900, tm.tm_mon+1, tm.tm_mday,
+		      tm.tm_hour, tm.tm_min, tm.tm_sec);
+}
+
diff --git a/arch/powerpc/platforms/powermac/Makefile b/arch/powerpc/platforms/powermac/Makefile
new file mode 100644
index 000000000000..4369676f1d54
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/Makefile
@@ -0,0 +1,8 @@
+obj-y				+= pic.o setup.o time.o feature.o pci.o \
+				   sleep.o low_i2c.o cache.o
+obj-$(CONFIG_PMAC_BACKLIGHT)	+= backlight.o
+obj-$(CONFIG_CPU_FREQ_PMAC)	+= cpufreq.o
+obj-$(CONFIG_NVRAM)		+= nvram.o
+# ppc64 pmac doesn't define CONFIG_NVRAM but needs nvram stuff
+obj-$(CONFIG_PPC64)		+= nvram.o
+obj-$(CONFIG_SMP)		+= smp.o
diff --git a/arch/powerpc/platforms/powermac/backlight.c b/arch/powerpc/platforms/powermac/backlight.c
new file mode 100644
index 000000000000..8be2f7d071f0
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/backlight.c
@@ -0,0 +1,202 @@
+/*
+ * Miscellaneous procedures for dealing with the PowerMac hardware.
+ * Contains support for the backlight.
+ *
+ *   Copyright (C) 2000 Benjamin Herrenschmidt
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/stddef.h>
+#include <linux/reboot.h>
+#include <linux/nvram.h>
+#include <linux/console.h>
+#include <asm/sections.h>
+#include <asm/ptrace.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/system.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+#include <asm/nvram.h>
+#include <asm/backlight.h>
+
+#include <linux/adb.h>
+#include <linux/pmu.h>
+
+static struct backlight_controller *backlighter;
+static void* backlighter_data;
+static int backlight_autosave;
+static int backlight_level = BACKLIGHT_MAX;
+static int backlight_enabled = 1;
+static int backlight_req_level = -1;
+static int backlight_req_enable = -1;
+
+static void backlight_callback(void *);
+static DECLARE_WORK(backlight_work, backlight_callback, NULL);
+
+void register_backlight_controller(struct backlight_controller *ctrler,
+					  void *data, char *type)
+{
+	struct device_node* bk_node;
+	char *prop;
+	int valid = 0;
+
+	/* There's already a matching controller, bail out */
+	if (backlighter != NULL)
+		return;
+
+	bk_node = find_devices("backlight");
+
+#ifdef CONFIG_ADB_PMU
+	/* Special case for the old PowerBook since I can't test on it */
+	backlight_autosave = machine_is_compatible("AAPL,3400/2400")
+		|| machine_is_compatible("AAPL,3500");
+	if ((backlight_autosave
+	     || machine_is_compatible("AAPL,PowerBook1998")
+	     || machine_is_compatible("PowerBook1,1"))
+	    && !strcmp(type, "pmu"))
+		valid = 1;
+#endif
+	if (bk_node) {
+		prop = get_property(bk_node, "backlight-control", NULL);
+		if (prop && !strncmp(prop, type, strlen(type)))
+			valid = 1;
+	}
+	if (!valid)
+		return;
+	backlighter = ctrler;
+	backlighter_data = data;
+
+	if (bk_node && !backlight_autosave)
+		prop = get_property(bk_node, "bklt", NULL);
+	else
+		prop = NULL;
+	if (prop) {
+		backlight_level = ((*prop)+1) >> 1;
+		if (backlight_level > BACKLIGHT_MAX)
+			backlight_level = BACKLIGHT_MAX;
+	}
+
+#ifdef CONFIG_ADB_PMU
+	if (backlight_autosave) {
+		struct adb_request req;
+		pmu_request(&req, NULL, 2, 0xd9, 0);
+		while (!req.complete)
+			pmu_poll();
+		backlight_level = req.reply[0] >> 4;
+	}
+#endif
+	acquire_console_sem();
+	if (!backlighter->set_enable(1, backlight_level, data))
+		backlight_enabled = 1;
+	release_console_sem();
+
+	printk(KERN_INFO "Registered \"%s\" backlight controller,"
+	       "level: %d/15\n", type, backlight_level);
+}
+EXPORT_SYMBOL(register_backlight_controller);
+
+void unregister_backlight_controller(struct backlight_controller
+					    *ctrler, void *data)
+{
+	/* We keep the current backlight level (for now) */
+	if (ctrler == backlighter && data == backlighter_data)
+		backlighter = NULL;
+}
+EXPORT_SYMBOL(unregister_backlight_controller);
+
+static int __set_backlight_enable(int enable)
+{
+	int rc;
+
+	if (!backlighter)
+		return -ENODEV;
+	acquire_console_sem();
+	rc = backlighter->set_enable(enable, backlight_level,
+				     backlighter_data);
+	if (!rc)
+		backlight_enabled = enable;
+	release_console_sem();
+	return rc;
+}
+int set_backlight_enable(int enable)
+{
+	if (!backlighter)
+		return -ENODEV;
+	backlight_req_enable = enable;
+	schedule_work(&backlight_work);
+	return 0;
+}
+
+EXPORT_SYMBOL(set_backlight_enable);
+
+int get_backlight_enable(void)
+{
+	if (!backlighter)
+		return -ENODEV;
+	return backlight_enabled;
+}
+EXPORT_SYMBOL(get_backlight_enable);
+
+static int __set_backlight_level(int level)
+{
+	int rc = 0;
+
+	if (!backlighter)
+		return -ENODEV;
+	if (level < BACKLIGHT_MIN)
+		level = BACKLIGHT_OFF;
+	if (level > BACKLIGHT_MAX)
+		level = BACKLIGHT_MAX;
+	acquire_console_sem();
+	if (backlight_enabled)
+		rc = backlighter->set_level(level, backlighter_data);
+	if (!rc)
+		backlight_level = level;
+	release_console_sem();
+	if (!rc && !backlight_autosave) {
+		level <<=1;
+		if (level & 0x10)
+			level |= 0x01;
+		// -- todo: save to property "bklt"
+	}
+	return rc;
+}
+int set_backlight_level(int level)
+{
+	if (!backlighter)
+		return -ENODEV;
+	backlight_req_level = level;
+	schedule_work(&backlight_work);
+	return 0;
+}
+
+EXPORT_SYMBOL(set_backlight_level);
+
+int get_backlight_level(void)
+{
+	if (!backlighter)
+		return -ENODEV;
+	return backlight_level;
+}
+EXPORT_SYMBOL(get_backlight_level);
+
+static void backlight_callback(void *dummy)
+{
+	int level, enable;
+
+	do {
+		level = backlight_req_level;
+		enable = backlight_req_enable;
+		mb();
+
+		if (level >= 0)
+			__set_backlight_level(level);
+		if (enable >= 0)
+			__set_backlight_enable(enable);
+	} while(cmpxchg(&backlight_req_level, level, -1) != level ||
+		cmpxchg(&backlight_req_enable, enable, -1) != enable);
+}
diff --git a/arch/powerpc/platforms/powermac/cache.S b/arch/powerpc/platforms/powermac/cache.S
new file mode 100644
index 000000000000..fb977de6b704
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/cache.S
@@ -0,0 +1,359 @@
+/*
+ * This file contains low-level cache management functions
+ * used for sleep and CPU speed changes on Apple machines.
+ * (In fact the only thing that is Apple-specific is that we assume
+ * that we can read from ROM at physical address 0xfff00000.)
+ *
+ *    Copyright (C) 2004 Paul Mackerras (paulus@samba.org) and
+ *                       Benjamin Herrenschmidt (benh@kernel.crashing.org)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ *
+ */
+
+#include <linux/config.h>
+#include <asm/processor.h>
+#include <asm/ppc_asm.h>
+#include <asm/cputable.h>
+
+/*
+ * Flush and disable all data caches (dL1, L2, L3). This is used
+ * when going to sleep, when doing a PMU based cpufreq transition,
+ * or when "offlining" a CPU on SMP machines. This code is over
+ * paranoid, but I've had enough issues with various CPU revs and
+ * bugs that I decided it was worth beeing over cautious
+ */
+
+_GLOBAL(flush_disable_caches)
+#ifndef CONFIG_6xx
+	blr
+#else
+BEGIN_FTR_SECTION
+	b	flush_disable_745x
+END_FTR_SECTION_IFSET(CPU_FTR_SPEC7450)
+BEGIN_FTR_SECTION
+	b	flush_disable_75x
+END_FTR_SECTION_IFSET(CPU_FTR_L2CR)
+	b	__flush_disable_L1
+
+/* This is the code for G3 and 74[01]0 */
+flush_disable_75x:
+	mflr	r10
+
+	/* Turn off EE and DR in MSR */
+	mfmsr	r11
+	rlwinm	r0,r11,0,~MSR_EE
+	rlwinm	r0,r0,0,~MSR_DR
+	sync
+	mtmsr	r0
+	isync
+
+	/* Stop DST streams */
+BEGIN_FTR_SECTION
+	DSSALL
+	sync
+END_FTR_SECTION_IFSET(CPU_FTR_ALTIVEC)
+
+	/* Stop DPM */
+	mfspr	r8,SPRN_HID0		/* Save SPRN_HID0 in r8 */
+	rlwinm	r4,r8,0,12,10		/* Turn off HID0[DPM] */
+	sync
+	mtspr	SPRN_HID0,r4		/* Disable DPM */
+	sync
+
+	/* Disp-flush L1. We have a weird problem here that I never
+	 * totally figured out. On 750FX, using the ROM for the flush
+	 * results in a non-working flush. We use that workaround for
+	 * now until I finally understand what's going on. --BenH
+	 */
+
+	/* ROM base by default */
+	lis	r4,0xfff0
+	mfpvr	r3
+	srwi	r3,r3,16
+	cmplwi	cr0,r3,0x7000
+	bne+	1f
+	/* RAM base on 750FX */
+	li	r4,0
+1:	li	r4,0x4000
+	mtctr	r4
+1:	lwz	r0,0(r4)
+	addi	r4,r4,32
+	bdnz	1b
+	sync
+	isync
+
+	/* Disable / invalidate / enable L1 data */
+	mfspr	r3,SPRN_HID0
+	rlwinm	r3,r3,0,~(HID0_DCE | HID0_ICE)
+	mtspr	SPRN_HID0,r3
+	sync
+	isync
+	ori	r3,r3,(HID0_DCE|HID0_DCI|HID0_ICE|HID0_ICFI)
+	sync
+	isync
+	mtspr	SPRN_HID0,r3
+	xori	r3,r3,(HID0_DCI|HID0_ICFI)
+	mtspr	SPRN_HID0,r3
+	sync
+
+	/* Get the current enable bit of the L2CR into r4 */
+	mfspr	r5,SPRN_L2CR
+	/* Set to data-only (pre-745x bit) */
+	oris	r3,r5,L2CR_L2DO@h
+	b	2f
+	/* When disabling L2, code must be in L1 */
+	.balign 32
+1:	mtspr	SPRN_L2CR,r3
+3:	sync
+	isync
+	b	1f
+2:	b	3f
+3:	sync
+	isync
+	b	1b
+1:	/* disp-flush L2. The interesting thing here is that the L2 can be
+	 * up to 2Mb ... so using the ROM, we'll end up wrapping back to memory
+	 * but that is probbaly fine. We disp-flush over 4Mb to be safe
+	 */
+	lis	r4,2
+	mtctr	r4
+	lis	r4,0xfff0
+1:	lwz	r0,0(r4)
+	addi	r4,r4,32
+	bdnz	1b
+	sync
+	isync
+	lis	r4,2
+	mtctr	r4
+	lis	r4,0xfff0
+1:	dcbf	0,r4
+	addi	r4,r4,32
+	bdnz	1b
+	sync
+	isync
+
+	/* now disable L2 */
+	rlwinm	r5,r5,0,~L2CR_L2E
+	b	2f
+	/* When disabling L2, code must be in L1 */
+	.balign 32
+1:	mtspr	SPRN_L2CR,r5
+3:	sync
+	isync
+	b	1f
+2:	b	3f
+3:	sync
+	isync
+	b	1b
+1:	sync
+	isync
+	/* Invalidate L2. This is pre-745x, we clear the L2I bit ourselves */
+	oris	r4,r5,L2CR_L2I@h
+	mtspr	SPRN_L2CR,r4
+	sync
+	isync
+
+	/* Wait for the invalidation to complete */
+1:	mfspr	r3,SPRN_L2CR
+	rlwinm.	r0,r3,0,31,31
+	bne	1b
+
+	/* Clear L2I */
+	xoris	r4,r4,L2CR_L2I@h
+	sync
+	mtspr	SPRN_L2CR,r4
+	sync
+
+	/* now disable the L1 data cache */
+	mfspr	r0,SPRN_HID0
+	rlwinm	r0,r0,0,~(HID0_DCE|HID0_ICE)
+	mtspr	SPRN_HID0,r0
+	sync
+	isync
+
+	/* Restore HID0[DPM] to whatever it was before */
+	sync
+	mfspr	r0,SPRN_HID0
+	rlwimi	r0,r8,0,11,11		/* Turn back HID0[DPM] */
+	mtspr	SPRN_HID0,r0
+	sync
+
+	/* restore DR and EE */
+	sync
+	mtmsr	r11
+	isync
+
+	mtlr	r10
+	blr
+
+/* This code is for 745x processors */
+flush_disable_745x:
+	/* Turn off EE and DR in MSR */
+	mfmsr	r11
+	rlwinm	r0,r11,0,~MSR_EE
+	rlwinm	r0,r0,0,~MSR_DR
+	sync
+	mtmsr	r0
+	isync
+
+	/* Stop prefetch streams */
+	DSSALL
+	sync
+
+	/* Disable L2 prefetching */
+	mfspr	r0,SPRN_MSSCR0
+	rlwinm	r0,r0,0,0,29
+	mtspr	SPRN_MSSCR0,r0
+	sync
+	isync
+	lis	r4,0
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+	dcbf	0,r4
+
+	/* Due to a bug with the HW flush on some CPU revs, we occasionally
+	 * experience data corruption. I'm adding a displacement flush along
+	 * with a dcbf loop over a few Mb to "help". The problem isn't totally
+	 * fixed by this in theory, but at least, in practice, I couldn't reproduce
+	 * it even with a big hammer...
+	 */
+
+        lis     r4,0x0002
+        mtctr   r4
+ 	li      r4,0
+1:
+        lwz     r0,0(r4)
+        addi    r4,r4,32                /* Go to start of next cache line */
+        bdnz    1b
+        isync
+
+        /* Now, flush the first 4MB of memory */
+        lis     r4,0x0002
+        mtctr   r4
+	li      r4,0
+        sync
+1:
+        dcbf    0,r4
+        addi    r4,r4,32                /* Go to start of next cache line */
+        bdnz    1b
+
+	/* Flush and disable the L1 data cache */
+	mfspr	r6,SPRN_LDSTCR
+	lis	r3,0xfff0	/* read from ROM for displacement flush */
+	li	r4,0xfe		/* start with only way 0 unlocked */
+	li	r5,128		/* 128 lines in each way */
+1:	mtctr	r5
+	rlwimi	r6,r4,0,24,31
+	mtspr	SPRN_LDSTCR,r6
+	sync
+	isync
+2:	lwz	r0,0(r3)	/* touch each cache line */
+	addi	r3,r3,32
+	bdnz	2b
+	rlwinm	r4,r4,1,24,30	/* move on to the next way */
+	ori	r4,r4,1
+	cmpwi	r4,0xff		/* all done? */
+	bne	1b
+	/* now unlock the L1 data cache */
+	li	r4,0
+	rlwimi	r6,r4,0,24,31
+	sync
+	mtspr	SPRN_LDSTCR,r6
+	sync
+	isync
+
+	/* Flush the L2 cache using the hardware assist */
+	mfspr	r3,SPRN_L2CR
+	cmpwi	r3,0		/* check if it is enabled first */
+	bge	4f
+	oris	r0,r3,(L2CR_L2IO_745x|L2CR_L2DO_745x)@h
+	b	2f
+	/* When disabling/locking L2, code must be in L1 */
+	.balign 32
+1:	mtspr	SPRN_L2CR,r0	/* lock the L2 cache */
+3:	sync
+	isync
+	b	1f
+2:	b	3f
+3:	sync
+	isync
+	b	1b
+1:	sync
+	isync
+	ori	r0,r3,L2CR_L2HWF_745x
+	sync
+	mtspr	SPRN_L2CR,r0	/* set the hardware flush bit */
+3:	mfspr	r0,SPRN_L2CR	/* wait for it to go to 0 */
+	andi.	r0,r0,L2CR_L2HWF_745x
+	bne	3b
+	sync
+	rlwinm	r3,r3,0,~L2CR_L2E
+	b	2f
+	/* When disabling L2, code must be in L1 */
+	.balign 32
+1:	mtspr	SPRN_L2CR,r3	/* disable the L2 cache */
+3:	sync
+	isync
+	b	1f
+2:	b	3f
+3:	sync
+	isync
+	b	1b
+1:	sync
+	isync
+	oris	r4,r3,L2CR_L2I@h
+	mtspr	SPRN_L2CR,r4
+	sync
+	isync
+1:	mfspr	r4,SPRN_L2CR
+	andis.	r0,r4,L2CR_L2I@h
+	bne	1b
+	sync
+
+BEGIN_FTR_SECTION
+	/* Flush the L3 cache using the hardware assist */
+4:	mfspr	r3,SPRN_L3CR
+	cmpwi	r3,0		/* check if it is enabled */
+	bge	6f
+	oris	r0,r3,L3CR_L3IO@h
+	ori	r0,r0,L3CR_L3DO
+	sync
+	mtspr	SPRN_L3CR,r0	/* lock the L3 cache */
+	sync
+	isync
+	ori	r0,r0,L3CR_L3HWF
+	sync
+	mtspr	SPRN_L3CR,r0	/* set the hardware flush bit */
+5:	mfspr	r0,SPRN_L3CR	/* wait for it to go to zero */
+	andi.	r0,r0,L3CR_L3HWF
+	bne	5b
+	rlwinm	r3,r3,0,~L3CR_L3E
+	sync
+	mtspr	SPRN_L3CR,r3	/* disable the L3 cache */
+	sync
+	ori	r4,r3,L3CR_L3I
+	mtspr	SPRN_L3CR,r4
+1:	mfspr	r4,SPRN_L3CR
+	andi.	r0,r4,L3CR_L3I
+	bne	1b
+	sync
+END_FTR_SECTION_IFSET(CPU_FTR_L3CR)
+
+6:	mfspr	r0,SPRN_HID0	/* now disable the L1 data cache */
+	rlwinm	r0,r0,0,~HID0_DCE
+	mtspr	SPRN_HID0,r0
+	sync
+	isync
+	mtmsr	r11		/* restore DR and EE */
+	isync
+	blr
+#endif	/* CONFIG_6xx */
diff --git a/arch/powerpc/platforms/powermac/cpufreq.c b/arch/powerpc/platforms/powermac/cpufreq.c
new file mode 100644
index 000000000000..c47f8b69725c
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/cpufreq.c
@@ -0,0 +1,726 @@
+/*
+ *  arch/ppc/platforms/pmac_cpufreq.c
+ *
+ *  Copyright (C) 2002 - 2005 Benjamin Herrenschmidt <benh@kernel.crashing.org>
+ *  Copyright (C) 2004        John Steele Scott <toojays@toojays.net>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * TODO: Need a big cleanup here. Basically, we need to have different
+ * cpufreq_driver structures for the different type of HW instead of the
+ * current mess. We also need to better deal with the detection of the
+ * type of machine.
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/delay.h>
+#include <linux/sched.h>
+#include <linux/adb.h>
+#include <linux/pmu.h>
+#include <linux/slab.h>
+#include <linux/cpufreq.h>
+#include <linux/init.h>
+#include <linux/sysdev.h>
+#include <linux/i2c.h>
+#include <linux/hardirq.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+#include <asm/irq.h>
+#include <asm/pmac_feature.h>
+#include <asm/mmu_context.h>
+#include <asm/sections.h>
+#include <asm/cputable.h>
+#include <asm/time.h>
+#include <asm/system.h>
+#include <asm/mpic.h>
+#include <asm/keylargo.h>
+
+/* WARNING !!! This will cause calibrate_delay() to be called,
+ * but this is an __init function ! So you MUST go edit
+ * init/main.c to make it non-init before enabling DEBUG_FREQ
+ */
+#undef DEBUG_FREQ
+
+/*
+ * There is a problem with the core cpufreq code on SMP kernels,
+ * it won't recalculate the Bogomips properly
+ */
+#ifdef CONFIG_SMP
+#warning "WARNING, CPUFREQ not recommended on SMP kernels"
+#endif
+
+extern void low_choose_7447a_dfs(int dfs);
+extern void low_choose_750fx_pll(int pll);
+extern void low_sleep_handler(void);
+
+/*
+ * Currently, PowerMac cpufreq supports only high & low frequencies
+ * that are set by the firmware
+ */
+static unsigned int low_freq;
+static unsigned int hi_freq;
+static unsigned int cur_freq;
+static unsigned int sleep_freq;
+
+/*
+ * Different models uses different mecanisms to switch the frequency
+ */
+static int (*set_speed_proc)(int low_speed);
+static unsigned int (*get_speed_proc)(void);
+
+/*
+ * Some definitions used by the various speedprocs
+ */
+static u32 voltage_gpio;
+static u32 frequency_gpio;
+static u32 slew_done_gpio;
+static int no_schedule;
+static int has_cpu_l2lve;
+static int is_pmu_based;
+
+/* There are only two frequency states for each processor. Values
+ * are in kHz for the time being.
+ */
+#define CPUFREQ_HIGH                  0
+#define CPUFREQ_LOW                   1
+
+static struct cpufreq_frequency_table pmac_cpu_freqs[] = {
+	{CPUFREQ_HIGH, 		0},
+	{CPUFREQ_LOW,		0},
+	{0,			CPUFREQ_TABLE_END},
+};
+
+static struct freq_attr* pmac_cpu_freqs_attr[] = {
+	&cpufreq_freq_attr_scaling_available_freqs,
+	NULL,
+};
+
+static inline void local_delay(unsigned long ms)
+{
+	if (no_schedule)
+		mdelay(ms);
+	else
+		msleep(ms);
+}
+
+#ifdef DEBUG_FREQ
+static inline void debug_calc_bogomips(void)
+{
+	/* This will cause a recalc of bogomips and display the
+	 * result. We backup/restore the value to avoid affecting the
+	 * core cpufreq framework's own calculation.
+	 */
+	extern void calibrate_delay(void);
+
+	unsigned long save_lpj = loops_per_jiffy;
+	calibrate_delay();
+	loops_per_jiffy = save_lpj;
+}
+#endif /* DEBUG_FREQ */
+
+/* Switch CPU speed under 750FX CPU control
+ */
+static int cpu_750fx_cpu_speed(int low_speed)
+{
+	u32 hid2;
+
+	if (low_speed == 0) {
+		/* ramping up, set voltage first */
+		pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, voltage_gpio, 0x05);
+		/* Make sure we sleep for at least 1ms */
+		local_delay(10);
+
+		/* tweak L2 for high voltage */
+		if (has_cpu_l2lve) {
+			hid2 = mfspr(SPRN_HID2);
+			hid2 &= ~0x2000;
+			mtspr(SPRN_HID2, hid2);
+		}
+	}
+#ifdef CONFIG_6xx
+	low_choose_750fx_pll(low_speed);
+#endif
+	if (low_speed == 1) {
+		/* tweak L2 for low voltage */
+		if (has_cpu_l2lve) {
+			hid2 = mfspr(SPRN_HID2);
+			hid2 |= 0x2000;
+			mtspr(SPRN_HID2, hid2);
+		}
+
+		/* ramping down, set voltage last */
+		pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, voltage_gpio, 0x04);
+		local_delay(10);
+	}
+
+	return 0;
+}
+
+static unsigned int cpu_750fx_get_cpu_speed(void)
+{
+	if (mfspr(SPRN_HID1) & HID1_PS)
+		return low_freq;
+	else
+		return hi_freq;
+}
+
+/* Switch CPU speed using DFS */
+static int dfs_set_cpu_speed(int low_speed)
+{
+	if (low_speed == 0) {
+		/* ramping up, set voltage first */
+		pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, voltage_gpio, 0x05);
+		/* Make sure we sleep for at least 1ms */
+		local_delay(1);
+	}
+
+	/* set frequency */
+#ifdef CONFIG_6xx
+	low_choose_7447a_dfs(low_speed);
+#endif
+	udelay(100);
+
+	if (low_speed == 1) {
+		/* ramping down, set voltage last */
+		pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, voltage_gpio, 0x04);
+		local_delay(1);
+	}
+
+	return 0;
+}
+
+static unsigned int dfs_get_cpu_speed(void)
+{
+	if (mfspr(SPRN_HID1) & HID1_DFS)
+		return low_freq;
+	else
+		return hi_freq;
+}
+
+
+/* Switch CPU speed using slewing GPIOs
+ */
+static int gpios_set_cpu_speed(int low_speed)
+{
+	int gpio, timeout = 0;
+
+	/* If ramping up, set voltage first */
+	if (low_speed == 0) {
+		pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, voltage_gpio, 0x05);
+		/* Delay is way too big but it's ok, we schedule */
+		local_delay(10);
+	}
+
+	/* Set frequency */
+	gpio = 	pmac_call_feature(PMAC_FTR_READ_GPIO, NULL, frequency_gpio, 0);
+	if (low_speed == ((gpio & 0x01) == 0))
+		goto skip;
+
+	pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, frequency_gpio,
+			  low_speed ? 0x04 : 0x05);
+	udelay(200);
+	do {
+		if (++timeout > 100)
+			break;
+		local_delay(1);
+		gpio = pmac_call_feature(PMAC_FTR_READ_GPIO, NULL, slew_done_gpio, 0);
+	} while((gpio & 0x02) == 0);
+ skip:
+	/* If ramping down, set voltage last */
+	if (low_speed == 1) {
+		pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, voltage_gpio, 0x04);
+		/* Delay is way too big but it's ok, we schedule */
+		local_delay(10);
+	}
+
+#ifdef DEBUG_FREQ
+	debug_calc_bogomips();
+#endif
+
+	return 0;
+}
+
+/* Switch CPU speed under PMU control
+ */
+static int pmu_set_cpu_speed(int low_speed)
+{
+	struct adb_request req;
+	unsigned long save_l2cr;
+	unsigned long save_l3cr;
+	unsigned int pic_prio;
+	unsigned long flags;
+
+	preempt_disable();
+
+#ifdef DEBUG_FREQ
+	printk(KERN_DEBUG "HID1, before: %x\n", mfspr(SPRN_HID1));
+#endif
+	pmu_suspend();
+
+	/* Disable all interrupt sources on openpic */
+ 	pic_prio = mpic_cpu_get_priority();
+	mpic_cpu_set_priority(0xf);
+
+	/* Make sure the decrementer won't interrupt us */
+	asm volatile("mtdec %0" : : "r" (0x7fffffff));
+	/* Make sure any pending DEC interrupt occuring while we did
+	 * the above didn't re-enable the DEC */
+	mb();
+	asm volatile("mtdec %0" : : "r" (0x7fffffff));
+
+	/* We can now disable MSR_EE */
+	local_irq_save(flags);
+
+	/* Giveup the FPU & vec */
+	enable_kernel_fp();
+
+#ifdef CONFIG_ALTIVEC
+	if (cpu_has_feature(CPU_FTR_ALTIVEC))
+		enable_kernel_altivec();
+#endif /* CONFIG_ALTIVEC */
+
+	/* Save & disable L2 and L3 caches */
+	save_l3cr = _get_L3CR();	/* (returns -1 if not available) */
+	save_l2cr = _get_L2CR();	/* (returns -1 if not available) */
+
+	/* Send the new speed command. My assumption is that this command
+	 * will cause PLL_CFG[0..3] to be changed next time CPU goes to sleep
+	 */
+	pmu_request(&req, NULL, 6, PMU_CPU_SPEED, 'W', 'O', 'O', 'F', low_speed);
+	while (!req.complete)
+		pmu_poll();
+
+	/* Prepare the northbridge for the speed transition */
+	pmac_call_feature(PMAC_FTR_SLEEP_STATE,NULL,1,1);
+
+	/* Call low level code to backup CPU state and recover from
+	 * hardware reset
+	 */
+	low_sleep_handler();
+
+	/* Restore the northbridge */
+	pmac_call_feature(PMAC_FTR_SLEEP_STATE,NULL,1,0);
+
+	/* Restore L2 cache */
+	if (save_l2cr != 0xffffffff && (save_l2cr & L2CR_L2E) != 0)
+ 		_set_L2CR(save_l2cr);
+	/* Restore L3 cache */
+	if (save_l3cr != 0xffffffff && (save_l3cr & L3CR_L3E) != 0)
+ 		_set_L3CR(save_l3cr);
+
+	/* Restore userland MMU context */
+	set_context(current->active_mm->context, current->active_mm->pgd);
+
+#ifdef DEBUG_FREQ
+	printk(KERN_DEBUG "HID1, after: %x\n", mfspr(SPRN_HID1));
+#endif
+
+	/* Restore low level PMU operations */
+	pmu_unlock();
+
+	/* Restore decrementer */
+	wakeup_decrementer();
+
+	/* Restore interrupts */
+ 	mpic_cpu_set_priority(pic_prio);
+
+	/* Let interrupts flow again ... */
+	local_irq_restore(flags);
+
+#ifdef DEBUG_FREQ
+	debug_calc_bogomips();
+#endif
+
+	pmu_resume();
+
+	preempt_enable();
+
+	return 0;
+}
+
+static int do_set_cpu_speed(int speed_mode, int notify)
+{
+	struct cpufreq_freqs freqs;
+	unsigned long l3cr;
+	static unsigned long prev_l3cr;
+
+	freqs.old = cur_freq;
+	freqs.new = (speed_mode == CPUFREQ_HIGH) ? hi_freq : low_freq;
+	freqs.cpu = smp_processor_id();
+
+	if (freqs.old == freqs.new)
+		return 0;
+
+	if (notify)
+		cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
+	if (speed_mode == CPUFREQ_LOW &&
+	    cpu_has_feature(CPU_FTR_L3CR)) {
+		l3cr = _get_L3CR();
+		if (l3cr & L3CR_L3E) {
+			prev_l3cr = l3cr;
+			_set_L3CR(0);
+		}
+	}
+	set_speed_proc(speed_mode == CPUFREQ_LOW);
+	if (speed_mode == CPUFREQ_HIGH &&
+	    cpu_has_feature(CPU_FTR_L3CR)) {
+		l3cr = _get_L3CR();
+		if ((prev_l3cr & L3CR_L3E) && l3cr != prev_l3cr)
+			_set_L3CR(prev_l3cr);
+	}
+	if (notify)
+		cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
+	cur_freq = (speed_mode == CPUFREQ_HIGH) ? hi_freq : low_freq;
+
+	return 0;
+}
+
+static unsigned int pmac_cpufreq_get_speed(unsigned int cpu)
+{
+	return cur_freq;
+}
+
+static int pmac_cpufreq_verify(struct cpufreq_policy *policy)
+{
+	return cpufreq_frequency_table_verify(policy, pmac_cpu_freqs);
+}
+
+static int pmac_cpufreq_target(	struct cpufreq_policy *policy,
+					unsigned int target_freq,
+					unsigned int relation)
+{
+	unsigned int    newstate = 0;
+
+	if (cpufreq_frequency_table_target(policy, pmac_cpu_freqs,
+			target_freq, relation, &newstate))
+		return -EINVAL;
+
+	return do_set_cpu_speed(newstate, 1);
+}
+
+unsigned int pmac_get_one_cpufreq(int i)
+{
+	/* Supports only one CPU for now */
+	return (i == 0) ? cur_freq : 0;
+}
+
+static int pmac_cpufreq_cpu_init(struct cpufreq_policy *policy)
+{
+	if (policy->cpu != 0)
+		return -ENODEV;
+
+	policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
+	policy->cpuinfo.transition_latency	= CPUFREQ_ETERNAL;
+	policy->cur = cur_freq;
+
+	cpufreq_frequency_table_get_attr(pmac_cpu_freqs, policy->cpu);
+	return cpufreq_frequency_table_cpuinfo(policy, pmac_cpu_freqs);
+}
+
+static u32 read_gpio(struct device_node *np)
+{
+	u32 *reg = (u32 *)get_property(np, "reg", NULL);
+	u32 offset;
+
+	if (reg == NULL)
+		return 0;
+	/* That works for all keylargos but shall be fixed properly
+	 * some day... The problem is that it seems we can't rely
+	 * on the "reg" property of the GPIO nodes, they are either
+	 * relative to the base of KeyLargo or to the base of the
+	 * GPIO space, and the device-tree doesn't help.
+	 */
+	offset = *reg;
+	if (offset < KEYLARGO_GPIO_LEVELS0)
+		offset += KEYLARGO_GPIO_LEVELS0;
+	return offset;
+}
+
+static int pmac_cpufreq_suspend(struct cpufreq_policy *policy, pm_message_t pmsg)
+{
+	/* Ok, this could be made a bit smarter, but let's be robust for now. We
+	 * always force a speed change to high speed before sleep, to make sure
+	 * we have appropriate voltage and/or bus speed for the wakeup process,
+	 * and to make sure our loops_per_jiffies are "good enough", that is will
+	 * not cause too short delays if we sleep in low speed and wake in high
+	 * speed..
+	 */
+	no_schedule = 1;
+	sleep_freq = cur_freq;
+	if (cur_freq == low_freq && !is_pmu_based)
+		do_set_cpu_speed(CPUFREQ_HIGH, 0);
+	return 0;
+}
+
+static int pmac_cpufreq_resume(struct cpufreq_policy *policy)
+{
+	/* If we resume, first check if we have a get() function */
+	if (get_speed_proc)
+		cur_freq = get_speed_proc();
+	else
+		cur_freq = 0;
+
+	/* We don't, hrm... we don't really know our speed here, best
+	 * is that we force a switch to whatever it was, which is
+	 * probably high speed due to our suspend() routine
+	 */
+	do_set_cpu_speed(sleep_freq == low_freq ?
+			 CPUFREQ_LOW : CPUFREQ_HIGH, 0);
+
+	no_schedule = 0;
+	return 0;
+}
+
+static struct cpufreq_driver pmac_cpufreq_driver = {
+	.verify 	= pmac_cpufreq_verify,
+	.target 	= pmac_cpufreq_target,
+	.get		= pmac_cpufreq_get_speed,
+	.init		= pmac_cpufreq_cpu_init,
+	.suspend	= pmac_cpufreq_suspend,
+	.resume		= pmac_cpufreq_resume,
+	.flags		= CPUFREQ_PM_NO_WARN,
+	.attr		= pmac_cpu_freqs_attr,
+	.name		= "powermac",
+	.owner		= THIS_MODULE,
+};
+
+
+static int pmac_cpufreq_init_MacRISC3(struct device_node *cpunode)
+{
+	struct device_node *volt_gpio_np = of_find_node_by_name(NULL,
+								"voltage-gpio");
+	struct device_node *freq_gpio_np = of_find_node_by_name(NULL,
+								"frequency-gpio");
+	struct device_node *slew_done_gpio_np = of_find_node_by_name(NULL,
+								     "slewing-done");
+	u32 *value;
+
+	/*
+	 * Check to see if it's GPIO driven or PMU only
+	 *
+	 * The way we extract the GPIO address is slightly hackish, but it
+	 * works well enough for now. We need to abstract the whole GPIO
+	 * stuff sooner or later anyway
+	 */
+
+	if (volt_gpio_np)
+		voltage_gpio = read_gpio(volt_gpio_np);
+	if (freq_gpio_np)
+		frequency_gpio = read_gpio(freq_gpio_np);
+	if (slew_done_gpio_np)
+		slew_done_gpio = read_gpio(slew_done_gpio_np);
+
+	/* If we use the frequency GPIOs, calculate the min/max speeds based
+	 * on the bus frequencies
+	 */
+	if (frequency_gpio && slew_done_gpio) {
+		int lenp, rc;
+		u32 *freqs, *ratio;
+
+		freqs = (u32 *)get_property(cpunode, "bus-frequencies", &lenp);
+		lenp /= sizeof(u32);
+		if (freqs == NULL || lenp != 2) {
+			printk(KERN_ERR "cpufreq: bus-frequencies incorrect or missing\n");
+			return 1;
+		}
+		ratio = (u32 *)get_property(cpunode, "processor-to-bus-ratio*2", NULL);
+		if (ratio == NULL) {
+			printk(KERN_ERR "cpufreq: processor-to-bus-ratio*2 missing\n");
+			return 1;
+		}
+
+		/* Get the min/max bus frequencies */
+		low_freq = min(freqs[0], freqs[1]);
+		hi_freq = max(freqs[0], freqs[1]);
+
+		/* Grrrr.. It _seems_ that the device-tree is lying on the low bus
+		 * frequency, it claims it to be around 84Mhz on some models while
+		 * it appears to be approx. 101Mhz on all. Let's hack around here...
+		 * fortunately, we don't need to be too precise
+		 */
+		if (low_freq < 98000000)
+			low_freq = 101000000;
+			
+		/* Convert those to CPU core clocks */
+		low_freq = (low_freq * (*ratio)) / 2000;
+		hi_freq = (hi_freq * (*ratio)) / 2000;
+
+		/* Now we get the frequencies, we read the GPIO to see what is out current
+		 * speed
+		 */
+		rc = pmac_call_feature(PMAC_FTR_READ_GPIO, NULL, frequency_gpio, 0);
+		cur_freq = (rc & 0x01) ? hi_freq : low_freq;
+
+		set_speed_proc = gpios_set_cpu_speed;
+		return 1;
+	}
+
+	/* If we use the PMU, look for the min & max frequencies in the
+	 * device-tree
+	 */
+	value = (u32 *)get_property(cpunode, "min-clock-frequency", NULL);
+	if (!value)
+		return 1;
+	low_freq = (*value) / 1000;
+	/* The PowerBook G4 12" (PowerBook6,1) has an error in the device-tree
+	 * here */
+	if (low_freq < 100000)
+		low_freq *= 10;
+
+	value = (u32 *)get_property(cpunode, "max-clock-frequency", NULL);
+	if (!value)
+		return 1;
+	hi_freq = (*value) / 1000;
+	set_speed_proc = pmu_set_cpu_speed;
+	is_pmu_based = 1;
+
+	return 0;
+}
+
+static int pmac_cpufreq_init_7447A(struct device_node *cpunode)
+{
+	struct device_node *volt_gpio_np;
+
+	if (get_property(cpunode, "dynamic-power-step", NULL) == NULL)
+		return 1;
+
+	volt_gpio_np = of_find_node_by_name(NULL, "cpu-vcore-select");
+	if (volt_gpio_np)
+		voltage_gpio = read_gpio(volt_gpio_np);
+	if (!voltage_gpio){
+		printk(KERN_ERR "cpufreq: missing cpu-vcore-select gpio\n");
+		return 1;
+	}
+
+	/* OF only reports the high frequency */
+	hi_freq = cur_freq;
+	low_freq = cur_freq/2;
+
+	/* Read actual frequency from CPU */
+	cur_freq = dfs_get_cpu_speed();
+	set_speed_proc = dfs_set_cpu_speed;
+	get_speed_proc = dfs_get_cpu_speed;
+
+	return 0;
+}
+
+static int pmac_cpufreq_init_750FX(struct device_node *cpunode)
+{
+	struct device_node *volt_gpio_np;
+	u32 pvr, *value;
+
+	if (get_property(cpunode, "dynamic-power-step", NULL) == NULL)
+		return 1;
+
+	hi_freq = cur_freq;
+	value = (u32 *)get_property(cpunode, "reduced-clock-frequency", NULL);
+	if (!value)
+		return 1;
+	low_freq = (*value) / 1000;
+
+	volt_gpio_np = of_find_node_by_name(NULL, "cpu-vcore-select");
+	if (volt_gpio_np)
+		voltage_gpio = read_gpio(volt_gpio_np);
+
+	pvr = mfspr(SPRN_PVR);
+	has_cpu_l2lve = !((pvr & 0xf00) == 0x100);
+
+	set_speed_proc = cpu_750fx_cpu_speed;
+	get_speed_proc = cpu_750fx_get_cpu_speed;
+	cur_freq = cpu_750fx_get_cpu_speed();
+
+	return 0;
+}
+
+/* Currently, we support the following machines:
+ *
+ *  - Titanium PowerBook 1Ghz (PMU based, 667Mhz & 1Ghz)
+ *  - Titanium PowerBook 800 (PMU based, 667Mhz & 800Mhz)
+ *  - Titanium PowerBook 400 (PMU based, 300Mhz & 400Mhz)
+ *  - Titanium PowerBook 500 (PMU based, 300Mhz & 500Mhz)
+ *  - iBook2 500/600 (PMU based, 400Mhz & 500/600Mhz)
+ *  - iBook2 700 (CPU based, 400Mhz & 700Mhz, support low voltage)
+ *  - Recent MacRISC3 laptops
+ *  - All new machines with 7447A CPUs
+ */
+static int __init pmac_cpufreq_setup(void)
+{
+	struct device_node	*cpunode;
+	u32			*value;
+
+	if (strstr(cmd_line, "nocpufreq"))
+		return 0;
+
+	/* Assume only one CPU */
+	cpunode = find_type_devices("cpu");
+	if (!cpunode)
+		goto out;
+
+	/* Get current cpu clock freq */
+	value = (u32 *)get_property(cpunode, "clock-frequency", NULL);
+	if (!value)
+		goto out;
+	cur_freq = (*value) / 1000;
+
+	/*  Check for 7447A based MacRISC3 */
+	if (machine_is_compatible("MacRISC3") &&
+	    get_property(cpunode, "dynamic-power-step", NULL) &&
+	    PVR_VER(mfspr(SPRN_PVR)) == 0x8003) {
+		pmac_cpufreq_init_7447A(cpunode);
+	/* Check for other MacRISC3 machines */
+	} else if (machine_is_compatible("PowerBook3,4") ||
+		   machine_is_compatible("PowerBook3,5") ||
+		   machine_is_compatible("MacRISC3")) {
+		pmac_cpufreq_init_MacRISC3(cpunode);
+	/* Else check for iBook2 500/600 */
+	} else if (machine_is_compatible("PowerBook4,1")) {
+		hi_freq = cur_freq;
+		low_freq = 400000;
+		set_speed_proc = pmu_set_cpu_speed;
+		is_pmu_based = 1;
+	}
+	/* Else check for TiPb 550 */
+	else if (machine_is_compatible("PowerBook3,3") && cur_freq == 550000) {
+		hi_freq = cur_freq;
+		low_freq = 500000;
+		set_speed_proc = pmu_set_cpu_speed;
+		is_pmu_based = 1;
+	}
+	/* Else check for TiPb 400 & 500 */
+	else if (machine_is_compatible("PowerBook3,2")) {
+		/* We only know about the 400 MHz and the 500Mhz model
+		 * they both have 300 MHz as low frequency
+		 */
+		if (cur_freq < 350000 || cur_freq > 550000)
+			goto out;
+		hi_freq = cur_freq;
+		low_freq = 300000;
+		set_speed_proc = pmu_set_cpu_speed;
+		is_pmu_based = 1;
+	}
+	/* Else check for 750FX */
+	else if (PVR_VER(mfspr(SPRN_PVR)) == 0x7000)
+		pmac_cpufreq_init_750FX(cpunode);
+out:
+	if (set_speed_proc == NULL)
+		return -ENODEV;
+
+	pmac_cpu_freqs[CPUFREQ_LOW].frequency = low_freq;
+	pmac_cpu_freqs[CPUFREQ_HIGH].frequency = hi_freq;
+
+	printk(KERN_INFO "Registering PowerMac CPU frequency driver\n");
+	printk(KERN_INFO "Low: %d Mhz, High: %d Mhz, Boot: %d Mhz\n",
+	       low_freq/1000, hi_freq/1000, cur_freq/1000);
+
+	return cpufreq_register_driver(&pmac_cpufreq_driver);
+}
+
+module_init(pmac_cpufreq_setup);
+
diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
new file mode 100644
index 000000000000..10f1d942c661
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/feature.c
@@ -0,0 +1,3063 @@
+/*
+ *  arch/ppc/platforms/pmac_feature.c
+ *
+ *  Copyright (C) 1996-2001 Paul Mackerras (paulus@cs.anu.edu.au)
+ *                          Ben. Herrenschmidt (benh@kernel.crashing.org)
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ *  TODO:
+ *
+ *   - Replace mdelay with some schedule loop if possible
+ *   - Shorten some obfuscated delays on some routines (like modem
+ *     power)
+ *   - Refcount some clocks (see darwin)
+ *   - Split split split...
+ *
+ */
+#include <linux/config.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/spinlock.h>
+#include <linux/adb.h>
+#include <linux/pmu.h>
+#include <linux/ioport.h>
+#include <linux/pci.h>
+#include <asm/sections.h>
+#include <asm/errno.h>
+#include <asm/ohare.h>
+#include <asm/heathrow.h>
+#include <asm/keylargo.h>
+#include <asm/uninorth.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+#include <asm/pmac_feature.h>
+#include <asm/dbdma.h>
+#include <asm/pci-bridge.h>
+#include <asm/pmac_low_i2c.h>
+
+#undef DEBUG_FEATURE
+
+#ifdef DEBUG_FEATURE
+#define DBG(fmt...) printk(KERN_DEBUG fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+#ifdef CONFIG_6xx
+extern int powersave_lowspeed;
+#endif
+
+extern int powersave_nap;
+extern struct device_node *k2_skiplist[2];
+
+
+/*
+ * We use a single global lock to protect accesses. Each driver has
+ * to take care of its own locking
+ */
+static DEFINE_SPINLOCK(feature_lock);
+
+#define LOCK(flags)	spin_lock_irqsave(&feature_lock, flags);
+#define UNLOCK(flags)	spin_unlock_irqrestore(&feature_lock, flags);
+
+
+/*
+ * Instance of some macio stuffs
+ */
+struct macio_chip macio_chips[MAX_MACIO_CHIPS];
+
+struct macio_chip *macio_find(struct device_node *child, int type)
+{
+	while(child) {
+		int	i;
+
+		for (i=0; i < MAX_MACIO_CHIPS && macio_chips[i].of_node; i++)
+			if (child == macio_chips[i].of_node &&
+			    (!type || macio_chips[i].type == type))
+				return &macio_chips[i];
+		child = child->parent;
+	}
+	return NULL;
+}
+EXPORT_SYMBOL_GPL(macio_find);
+
+static const char *macio_names[] =
+{
+	"Unknown",
+	"Grand Central",
+	"OHare",
+	"OHareII",
+	"Heathrow",
+	"Gatwick",
+	"Paddington",
+	"Keylargo",
+	"Pangea",
+	"Intrepid",
+	"K2"
+};
+
+
+
+/*
+ * Uninorth reg. access. Note that Uni-N regs are big endian
+ */
+
+#define UN_REG(r)	(uninorth_base + ((r) >> 2))
+#define UN_IN(r)	(in_be32(UN_REG(r)))
+#define UN_OUT(r,v)	(out_be32(UN_REG(r), (v)))
+#define UN_BIS(r,v)	(UN_OUT((r), UN_IN(r) | (v)))
+#define UN_BIC(r,v)	(UN_OUT((r), UN_IN(r) & ~(v)))
+
+static struct device_node *uninorth_node;
+static u32 __iomem *uninorth_base;
+static u32 uninorth_rev;
+static int uninorth_u3;
+static void __iomem *u3_ht;
+
+/*
+ * For each motherboard family, we have a table of functions pointers
+ * that handle the various features.
+ */
+
+typedef long (*feature_call)(struct device_node *node, long param, long value);
+
+struct feature_table_entry {
+	unsigned int	selector;
+	feature_call	function;
+};
+
+struct pmac_mb_def
+{
+	const char*			model_string;
+	const char*			model_name;
+	int				model_id;
+	struct feature_table_entry*	features;
+	unsigned long			board_flags;
+};
+static struct pmac_mb_def pmac_mb;
+
+/*
+ * Here are the chip specific feature functions
+ */
+
+static inline int simple_feature_tweak(struct device_node *node, int type,
+				       int reg, u32 mask, int value)
+{
+	struct macio_chip*	macio;
+	unsigned long		flags;
+
+	macio = macio_find(node, type);
+	if (!macio)
+		return -ENODEV;
+	LOCK(flags);
+	if (value)
+		MACIO_BIS(reg, mask);
+	else
+		MACIO_BIC(reg, mask);
+	(void)MACIO_IN32(reg);
+	UNLOCK(flags);
+
+	return 0;
+}
+
+#ifndef CONFIG_POWER4
+
+static long ohare_htw_scc_enable(struct device_node *node, long param,
+				 long value)
+{
+	struct macio_chip*	macio;
+	unsigned long		chan_mask;
+	unsigned long		fcr;
+	unsigned long		flags;
+	int			htw, trans;
+	unsigned long		rmask;
+
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+	if (!strcmp(node->name, "ch-a"))
+		chan_mask = MACIO_FLAG_SCCA_ON;
+	else if (!strcmp(node->name, "ch-b"))
+		chan_mask = MACIO_FLAG_SCCB_ON;
+	else
+		return -ENODEV;
+
+	htw = (macio->type == macio_heathrow || macio->type == macio_paddington
+		|| macio->type == macio_gatwick);
+	/* On these machines, the HRW_SCC_TRANS_EN_N bit mustn't be touched */
+	trans = (pmac_mb.model_id != PMAC_TYPE_YOSEMITE &&
+		 pmac_mb.model_id != PMAC_TYPE_YIKES);
+	if (value) {
+#ifdef CONFIG_ADB_PMU
+		if ((param & 0xfff) == PMAC_SCC_IRDA)
+			pmu_enable_irled(1);
+#endif /* CONFIG_ADB_PMU */
+		LOCK(flags);
+		fcr = MACIO_IN32(OHARE_FCR);
+		/* Check if scc cell need enabling */
+		if (!(fcr & OH_SCC_ENABLE)) {
+			fcr |= OH_SCC_ENABLE;
+			if (htw) {
+				/* Side effect: this will also power up the
+				 * modem, but it's too messy to figure out on which
+				 * ports this controls the tranceiver and on which
+				 * it controls the modem
+				 */
+				if (trans)
+					fcr &= ~HRW_SCC_TRANS_EN_N;
+				MACIO_OUT32(OHARE_FCR, fcr);
+				fcr |= (rmask = HRW_RESET_SCC);
+				MACIO_OUT32(OHARE_FCR, fcr);
+			} else {
+				fcr |= (rmask = OH_SCC_RESET);
+				MACIO_OUT32(OHARE_FCR, fcr);
+			}
+			UNLOCK(flags);
+			(void)MACIO_IN32(OHARE_FCR);
+			mdelay(15);
+			LOCK(flags);
+			fcr &= ~rmask;
+			MACIO_OUT32(OHARE_FCR, fcr);
+		}
+		if (chan_mask & MACIO_FLAG_SCCA_ON)
+			fcr |= OH_SCCA_IO;
+		if (chan_mask & MACIO_FLAG_SCCB_ON)
+			fcr |= OH_SCCB_IO;
+		MACIO_OUT32(OHARE_FCR, fcr);
+		macio->flags |= chan_mask;
+		UNLOCK(flags);
+		if (param & PMAC_SCC_FLAG_XMON)
+			macio->flags |= MACIO_FLAG_SCC_LOCKED;
+	} else {
+		if (macio->flags & MACIO_FLAG_SCC_LOCKED)
+			return -EPERM;
+		LOCK(flags);
+		fcr = MACIO_IN32(OHARE_FCR);
+		if (chan_mask & MACIO_FLAG_SCCA_ON)
+			fcr &= ~OH_SCCA_IO;
+		if (chan_mask & MACIO_FLAG_SCCB_ON)
+			fcr &= ~OH_SCCB_IO;
+		MACIO_OUT32(OHARE_FCR, fcr);
+		if ((fcr & (OH_SCCA_IO | OH_SCCB_IO)) == 0) {
+			fcr &= ~OH_SCC_ENABLE;
+			if (htw && trans)
+				fcr |= HRW_SCC_TRANS_EN_N;
+			MACIO_OUT32(OHARE_FCR, fcr);
+		}
+		macio->flags &= ~(chan_mask);
+		UNLOCK(flags);
+		mdelay(10);
+#ifdef CONFIG_ADB_PMU
+		if ((param & 0xfff) == PMAC_SCC_IRDA)
+			pmu_enable_irled(0);
+#endif /* CONFIG_ADB_PMU */
+	}
+	return 0;
+}
+
+static long ohare_floppy_enable(struct device_node *node, long param,
+				long value)
+{
+	return simple_feature_tweak(node, macio_ohare,
+		OHARE_FCR, OH_FLOPPY_ENABLE, value);
+}
+
+static long ohare_mesh_enable(struct device_node *node, long param, long value)
+{
+	return simple_feature_tweak(node, macio_ohare,
+		OHARE_FCR, OH_MESH_ENABLE, value);
+}
+
+static long ohare_ide_enable(struct device_node *node, long param, long value)
+{
+	switch(param) {
+	case 0:
+		/* For some reason, setting the bit in set_initial_features()
+		 * doesn't stick. I'm still investigating... --BenH.
+		 */
+		if (value)
+			simple_feature_tweak(node, macio_ohare,
+				OHARE_FCR, OH_IOBUS_ENABLE, 1);
+		return simple_feature_tweak(node, macio_ohare,
+			OHARE_FCR, OH_IDE0_ENABLE, value);
+	case 1:
+		return simple_feature_tweak(node, macio_ohare,
+			OHARE_FCR, OH_BAY_IDE_ENABLE, value);
+	default:
+		return -ENODEV;
+	}
+}
+
+static long ohare_ide_reset(struct device_node *node, long param, long value)
+{
+	switch(param) {
+	case 0:
+		return simple_feature_tweak(node, macio_ohare,
+			OHARE_FCR, OH_IDE0_RESET_N, !value);
+	case 1:
+		return simple_feature_tweak(node, macio_ohare,
+			OHARE_FCR, OH_IDE1_RESET_N, !value);
+	default:
+		return -ENODEV;
+	}
+}
+
+static long ohare_sleep_state(struct device_node *node, long param, long value)
+{
+	struct macio_chip*	macio = &macio_chips[0];
+
+	if ((pmac_mb.board_flags & PMAC_MB_CAN_SLEEP) == 0)
+		return -EPERM;
+	if (value == 1) {
+		MACIO_BIC(OHARE_FCR, OH_IOBUS_ENABLE);
+	} else if (value == 0) {
+		MACIO_BIS(OHARE_FCR, OH_IOBUS_ENABLE);
+	}
+
+	return 0;
+}
+
+static long heathrow_modem_enable(struct device_node *node, long param,
+				  long value)
+{
+	struct macio_chip*	macio;
+	u8			gpio;
+	unsigned long		flags;
+
+	macio = macio_find(node, macio_unknown);
+	if (!macio)
+		return -ENODEV;
+	gpio = MACIO_IN8(HRW_GPIO_MODEM_RESET) & ~1;
+	if (!value) {
+		LOCK(flags);
+		MACIO_OUT8(HRW_GPIO_MODEM_RESET, gpio);
+		UNLOCK(flags);
+		(void)MACIO_IN8(HRW_GPIO_MODEM_RESET);
+		mdelay(250);
+	}
+	if (pmac_mb.model_id != PMAC_TYPE_YOSEMITE &&
+	    pmac_mb.model_id != PMAC_TYPE_YIKES) {
+		LOCK(flags);
+		if (value)
+			MACIO_BIC(HEATHROW_FCR, HRW_SCC_TRANS_EN_N);
+		else
+			MACIO_BIS(HEATHROW_FCR, HRW_SCC_TRANS_EN_N);
+		UNLOCK(flags);
+		(void)MACIO_IN32(HEATHROW_FCR);
+		mdelay(250);
+	}
+	if (value) {
+		LOCK(flags);
+		MACIO_OUT8(HRW_GPIO_MODEM_RESET, gpio | 1);
+		(void)MACIO_IN8(HRW_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250); LOCK(flags);
+		MACIO_OUT8(HRW_GPIO_MODEM_RESET, gpio);
+		(void)MACIO_IN8(HRW_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250); LOCK(flags);
+		MACIO_OUT8(HRW_GPIO_MODEM_RESET, gpio | 1);
+		(void)MACIO_IN8(HRW_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250);
+	}
+	return 0;
+}
+
+static long heathrow_floppy_enable(struct device_node *node, long param,
+				   long value)
+{
+	return simple_feature_tweak(node, macio_unknown,
+		HEATHROW_FCR,
+		HRW_SWIM_ENABLE|HRW_BAY_FLOPPY_ENABLE,
+		value);
+}
+
+static long heathrow_mesh_enable(struct device_node *node, long param,
+				 long value)
+{
+	struct macio_chip*	macio;
+	unsigned long		flags;
+
+	macio = macio_find(node, macio_unknown);
+	if (!macio)
+		return -ENODEV;
+	LOCK(flags);
+	/* Set clear mesh cell enable */
+	if (value)
+		MACIO_BIS(HEATHROW_FCR, HRW_MESH_ENABLE);
+	else
+		MACIO_BIC(HEATHROW_FCR, HRW_MESH_ENABLE);
+	(void)MACIO_IN32(HEATHROW_FCR);
+	udelay(10);
+	/* Set/Clear termination power */
+	if (value)
+		MACIO_BIC(HEATHROW_MBCR, 0x04000000);
+	else
+		MACIO_BIS(HEATHROW_MBCR, 0x04000000);
+	(void)MACIO_IN32(HEATHROW_MBCR);
+	udelay(10);
+	UNLOCK(flags);
+
+	return 0;
+}
+
+static long heathrow_ide_enable(struct device_node *node, long param,
+				long value)
+{
+	switch(param) {
+	case 0:
+		return simple_feature_tweak(node, macio_unknown,
+			HEATHROW_FCR, HRW_IDE0_ENABLE, value);
+	case 1:
+		return simple_feature_tweak(node, macio_unknown,
+			HEATHROW_FCR, HRW_BAY_IDE_ENABLE, value);
+	default:
+		return -ENODEV;
+	}
+}
+
+static long heathrow_ide_reset(struct device_node *node, long param,
+			       long value)
+{
+	switch(param) {
+	case 0:
+		return simple_feature_tweak(node, macio_unknown,
+			HEATHROW_FCR, HRW_IDE0_RESET_N, !value);
+	case 1:
+		return simple_feature_tweak(node, macio_unknown,
+			HEATHROW_FCR, HRW_IDE1_RESET_N, !value);
+	default:
+		return -ENODEV;
+	}
+}
+
+static long heathrow_bmac_enable(struct device_node *node, long param,
+				 long value)
+{
+	struct macio_chip*	macio;
+	unsigned long		flags;
+
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+	if (value) {
+		LOCK(flags);
+		MACIO_BIS(HEATHROW_FCR, HRW_BMAC_IO_ENABLE);
+		MACIO_BIS(HEATHROW_FCR, HRW_BMAC_RESET);
+		UNLOCK(flags);
+		(void)MACIO_IN32(HEATHROW_FCR);
+		mdelay(10);
+		LOCK(flags);
+		MACIO_BIC(HEATHROW_FCR, HRW_BMAC_RESET);
+		UNLOCK(flags);
+		(void)MACIO_IN32(HEATHROW_FCR);
+		mdelay(10);
+	} else {
+		LOCK(flags);
+		MACIO_BIC(HEATHROW_FCR, HRW_BMAC_IO_ENABLE);
+		UNLOCK(flags);
+	}
+	return 0;
+}
+
+static long heathrow_sound_enable(struct device_node *node, long param,
+				  long value)
+{
+	struct macio_chip*	macio;
+	unsigned long		flags;
+
+	/* B&W G3 and Yikes don't support that properly (the
+	 * sound appear to never come back after beeing shut down).
+	 */
+	if (pmac_mb.model_id == PMAC_TYPE_YOSEMITE ||
+	    pmac_mb.model_id == PMAC_TYPE_YIKES)
+		return 0;
+
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+	if (value) {
+		LOCK(flags);
+		MACIO_BIS(HEATHROW_FCR, HRW_SOUND_CLK_ENABLE);
+		MACIO_BIC(HEATHROW_FCR, HRW_SOUND_POWER_N);
+		UNLOCK(flags);
+		(void)MACIO_IN32(HEATHROW_FCR);
+	} else {
+		LOCK(flags);
+		MACIO_BIS(HEATHROW_FCR, HRW_SOUND_POWER_N);
+		MACIO_BIC(HEATHROW_FCR, HRW_SOUND_CLK_ENABLE);
+		UNLOCK(flags);
+	}
+	return 0;
+}
+
+static u32 save_fcr[6];
+static u32 save_mbcr;
+static u32 save_gpio_levels[2];
+static u8 save_gpio_extint[KEYLARGO_GPIO_EXTINT_CNT];
+static u8 save_gpio_normal[KEYLARGO_GPIO_CNT];
+static u32 save_unin_clock_ctl;
+static struct dbdma_regs save_dbdma[13];
+static struct dbdma_regs save_alt_dbdma[13];
+
+static void dbdma_save(struct macio_chip *macio, struct dbdma_regs *save)
+{
+	int i;
+
+	/* Save state & config of DBDMA channels */
+	for (i = 0; i < 13; i++) {
+		volatile struct dbdma_regs __iomem * chan = (void __iomem *)
+			(macio->base + ((0x8000+i*0x100)>>2));
+		save[i].cmdptr_hi = in_le32(&chan->cmdptr_hi);
+		save[i].cmdptr = in_le32(&chan->cmdptr);
+		save[i].intr_sel = in_le32(&chan->intr_sel);
+		save[i].br_sel = in_le32(&chan->br_sel);
+		save[i].wait_sel = in_le32(&chan->wait_sel);
+	}
+}
+
+static void dbdma_restore(struct macio_chip *macio, struct dbdma_regs *save)
+{
+	int i;
+
+	/* Save state & config of DBDMA channels */
+	for (i = 0; i < 13; i++) {
+		volatile struct dbdma_regs __iomem * chan = (void __iomem *)
+			(macio->base + ((0x8000+i*0x100)>>2));
+		out_le32(&chan->control, (ACTIVE|DEAD|WAKE|FLUSH|PAUSE|RUN)<<16);
+		while (in_le32(&chan->status) & ACTIVE)
+			mb();
+		out_le32(&chan->cmdptr_hi, save[i].cmdptr_hi);
+		out_le32(&chan->cmdptr, save[i].cmdptr);
+		out_le32(&chan->intr_sel, save[i].intr_sel);
+		out_le32(&chan->br_sel, save[i].br_sel);
+		out_le32(&chan->wait_sel, save[i].wait_sel);
+	}
+}
+
+static void heathrow_sleep(struct macio_chip *macio, int secondary)
+{
+	if (secondary) {
+		dbdma_save(macio, save_alt_dbdma);
+		save_fcr[2] = MACIO_IN32(0x38);
+		save_fcr[3] = MACIO_IN32(0x3c);
+	} else {
+		dbdma_save(macio, save_dbdma);
+		save_fcr[0] = MACIO_IN32(0x38);
+		save_fcr[1] = MACIO_IN32(0x3c);
+		save_mbcr = MACIO_IN32(0x34);
+		/* Make sure sound is shut down */
+		MACIO_BIS(HEATHROW_FCR, HRW_SOUND_POWER_N);
+		MACIO_BIC(HEATHROW_FCR, HRW_SOUND_CLK_ENABLE);
+		/* This seems to be necessary as well or the fan
+		 * keeps coming up and battery drains fast */
+		MACIO_BIC(HEATHROW_FCR, HRW_IOBUS_ENABLE);
+		MACIO_BIC(HEATHROW_FCR, HRW_IDE0_RESET_N);
+		/* Make sure eth is down even if module or sleep
+		 * won't work properly */
+		MACIO_BIC(HEATHROW_FCR, HRW_BMAC_IO_ENABLE | HRW_BMAC_RESET);
+	}
+	/* Make sure modem is shut down */
+	MACIO_OUT8(HRW_GPIO_MODEM_RESET,
+		MACIO_IN8(HRW_GPIO_MODEM_RESET) & ~1);
+	MACIO_BIS(HEATHROW_FCR, HRW_SCC_TRANS_EN_N);
+	MACIO_BIC(HEATHROW_FCR, OH_SCCA_IO|OH_SCCB_IO|HRW_SCC_ENABLE);
+
+	/* Let things settle */
+	(void)MACIO_IN32(HEATHROW_FCR);
+}
+
+static void heathrow_wakeup(struct macio_chip *macio, int secondary)
+{
+	if (secondary) {
+		MACIO_OUT32(0x38, save_fcr[2]);
+		(void)MACIO_IN32(0x38);
+		mdelay(1);
+		MACIO_OUT32(0x3c, save_fcr[3]);
+		(void)MACIO_IN32(0x38);
+		mdelay(10);
+		dbdma_restore(macio, save_alt_dbdma);
+	} else {
+		MACIO_OUT32(0x38, save_fcr[0] | HRW_IOBUS_ENABLE);
+		(void)MACIO_IN32(0x38);
+		mdelay(1);
+		MACIO_OUT32(0x3c, save_fcr[1]);
+		(void)MACIO_IN32(0x38);
+		mdelay(1);
+		MACIO_OUT32(0x34, save_mbcr);
+		(void)MACIO_IN32(0x38);
+		mdelay(10);
+		dbdma_restore(macio, save_dbdma);
+	}
+}
+
+static long heathrow_sleep_state(struct device_node *node, long param,
+				 long value)
+{
+	if ((pmac_mb.board_flags & PMAC_MB_CAN_SLEEP) == 0)
+		return -EPERM;
+	if (value == 1) {
+		if (macio_chips[1].type == macio_gatwick)
+			heathrow_sleep(&macio_chips[0], 1);
+		heathrow_sleep(&macio_chips[0], 0);
+	} else if (value == 0) {
+		heathrow_wakeup(&macio_chips[0], 0);
+		if (macio_chips[1].type == macio_gatwick)
+			heathrow_wakeup(&macio_chips[0], 1);
+	}
+	return 0;
+}
+
+static long core99_scc_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip*	macio;
+	unsigned long		flags;
+	unsigned long		chan_mask;
+	u32			fcr;
+
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+	if (!strcmp(node->name, "ch-a"))
+		chan_mask = MACIO_FLAG_SCCA_ON;
+	else if (!strcmp(node->name, "ch-b"))
+		chan_mask = MACIO_FLAG_SCCB_ON;
+	else
+		return -ENODEV;
+
+	if (value) {
+		int need_reset_scc = 0;
+		int need_reset_irda = 0;
+
+		LOCK(flags);
+		fcr = MACIO_IN32(KEYLARGO_FCR0);
+		/* Check if scc cell need enabling */
+		if (!(fcr & KL0_SCC_CELL_ENABLE)) {
+			fcr |= KL0_SCC_CELL_ENABLE;
+			need_reset_scc = 1;
+		}
+		if (chan_mask & MACIO_FLAG_SCCA_ON) {
+			fcr |= KL0_SCCA_ENABLE;
+			/* Don't enable line drivers for I2S modem */
+			if ((param & 0xfff) == PMAC_SCC_I2S1)
+				fcr &= ~KL0_SCC_A_INTF_ENABLE;
+			else
+				fcr |= KL0_SCC_A_INTF_ENABLE;
+		}
+		if (chan_mask & MACIO_FLAG_SCCB_ON) {
+			fcr |= KL0_SCCB_ENABLE;
+			/* Perform irda specific inits */
+			if ((param & 0xfff) == PMAC_SCC_IRDA) {
+				fcr &= ~KL0_SCC_B_INTF_ENABLE;
+				fcr |= KL0_IRDA_ENABLE;
+				fcr |= KL0_IRDA_CLK32_ENABLE | KL0_IRDA_CLK19_ENABLE;
+				fcr |= KL0_IRDA_SOURCE1_SEL;
+				fcr &= ~(KL0_IRDA_FAST_CONNECT|KL0_IRDA_DEFAULT1|KL0_IRDA_DEFAULT0);
+				fcr &= ~(KL0_IRDA_SOURCE2_SEL|KL0_IRDA_HIGH_BAND);
+				need_reset_irda = 1;
+			} else
+				fcr |= KL0_SCC_B_INTF_ENABLE;
+		}
+		MACIO_OUT32(KEYLARGO_FCR0, fcr);
+		macio->flags |= chan_mask;
+		if (need_reset_scc)  {
+			MACIO_BIS(KEYLARGO_FCR0, KL0_SCC_RESET);
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+			UNLOCK(flags);
+			mdelay(15);
+			LOCK(flags);
+			MACIO_BIC(KEYLARGO_FCR0, KL0_SCC_RESET);
+		}
+		if (need_reset_irda)  {
+			MACIO_BIS(KEYLARGO_FCR0, KL0_IRDA_RESET);
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+			UNLOCK(flags);
+			mdelay(15);
+			LOCK(flags);
+			MACIO_BIC(KEYLARGO_FCR0, KL0_IRDA_RESET);
+		}
+		UNLOCK(flags);
+		if (param & PMAC_SCC_FLAG_XMON)
+			macio->flags |= MACIO_FLAG_SCC_LOCKED;
+	} else {
+		if (macio->flags & MACIO_FLAG_SCC_LOCKED)
+			return -EPERM;
+		LOCK(flags);
+		fcr = MACIO_IN32(KEYLARGO_FCR0);
+		if (chan_mask & MACIO_FLAG_SCCA_ON)
+			fcr &= ~KL0_SCCA_ENABLE;
+		if (chan_mask & MACIO_FLAG_SCCB_ON) {
+			fcr &= ~KL0_SCCB_ENABLE;
+			/* Perform irda specific clears */
+			if ((param & 0xfff) == PMAC_SCC_IRDA) {
+				fcr &= ~KL0_IRDA_ENABLE;
+				fcr &= ~(KL0_IRDA_CLK32_ENABLE | KL0_IRDA_CLK19_ENABLE);
+				fcr &= ~(KL0_IRDA_FAST_CONNECT|KL0_IRDA_DEFAULT1|KL0_IRDA_DEFAULT0);
+				fcr &= ~(KL0_IRDA_SOURCE1_SEL|KL0_IRDA_SOURCE2_SEL|KL0_IRDA_HIGH_BAND);
+			}
+		}
+		MACIO_OUT32(KEYLARGO_FCR0, fcr);
+		if ((fcr & (KL0_SCCA_ENABLE | KL0_SCCB_ENABLE)) == 0) {
+			fcr &= ~KL0_SCC_CELL_ENABLE;
+			MACIO_OUT32(KEYLARGO_FCR0, fcr);
+		}
+		macio->flags &= ~(chan_mask);
+		UNLOCK(flags);
+		mdelay(10);
+	}
+	return 0;
+}
+
+static long
+core99_modem_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip*	macio;
+	u8			gpio;
+	unsigned long		flags;
+
+	/* Hack for internal USB modem */
+	if (node == NULL) {
+		if (macio_chips[0].type != macio_keylargo)
+			return -ENODEV;
+		node = macio_chips[0].of_node;
+	}
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+	gpio = MACIO_IN8(KL_GPIO_MODEM_RESET);
+	gpio |= KEYLARGO_GPIO_OUTPUT_ENABLE;
+	gpio &= ~KEYLARGO_GPIO_OUTOUT_DATA;
+
+	if (!value) {
+		LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio);
+		UNLOCK(flags);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		mdelay(250);
+	}
+	LOCK(flags);
+	if (value) {
+		MACIO_BIC(KEYLARGO_FCR2, KL2_ALT_DATA_OUT);
+		UNLOCK(flags);
+		(void)MACIO_IN32(KEYLARGO_FCR2);
+		mdelay(250);
+	} else {
+		MACIO_BIS(KEYLARGO_FCR2, KL2_ALT_DATA_OUT);
+		UNLOCK(flags);
+	}
+	if (value) {
+		LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio | KEYLARGO_GPIO_OUTOUT_DATA);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250); LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250); LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio | KEYLARGO_GPIO_OUTOUT_DATA);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250);
+	}
+	return 0;
+}
+
+static long
+pangea_modem_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip*	macio;
+	u8			gpio;
+	unsigned long		flags;
+
+	/* Hack for internal USB modem */
+	if (node == NULL) {
+		if (macio_chips[0].type != macio_pangea &&
+		    macio_chips[0].type != macio_intrepid)
+			return -ENODEV;
+		node = macio_chips[0].of_node;
+	}
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+	gpio = MACIO_IN8(KL_GPIO_MODEM_RESET);
+	gpio |= KEYLARGO_GPIO_OUTPUT_ENABLE;
+	gpio &= ~KEYLARGO_GPIO_OUTOUT_DATA;
+
+	if (!value) {
+		LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio);
+		UNLOCK(flags);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		mdelay(250);
+	}
+	LOCK(flags);
+	if (value) {
+		MACIO_OUT8(KL_GPIO_MODEM_POWER,
+			KEYLARGO_GPIO_OUTPUT_ENABLE);
+		UNLOCK(flags);
+		(void)MACIO_IN32(KEYLARGO_FCR2);
+		mdelay(250);
+	} else {
+		MACIO_OUT8(KL_GPIO_MODEM_POWER,
+			KEYLARGO_GPIO_OUTPUT_ENABLE | KEYLARGO_GPIO_OUTOUT_DATA);
+		UNLOCK(flags);
+	}
+	if (value) {
+		LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio | KEYLARGO_GPIO_OUTOUT_DATA);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250); LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250); LOCK(flags);
+		MACIO_OUT8(KL_GPIO_MODEM_RESET, gpio | KEYLARGO_GPIO_OUTOUT_DATA);
+		(void)MACIO_IN8(KL_GPIO_MODEM_RESET);
+		UNLOCK(flags); mdelay(250);
+	}
+	return 0;
+}
+
+static long
+core99_ata100_enable(struct device_node *node, long value)
+{
+	unsigned long flags;
+	struct pci_dev *pdev = NULL;
+	u8 pbus, pid;
+
+	if (uninorth_rev < 0x24)
+		return -ENODEV;
+
+	LOCK(flags);
+	if (value)
+		UN_BIS(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_ATA100);
+	else
+		UN_BIC(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_ATA100);
+	(void)UN_IN(UNI_N_CLOCK_CNTL);
+	UNLOCK(flags);
+	udelay(20);
+
+	if (value) {
+		if (pci_device_from_OF_node(node, &pbus, &pid) == 0)
+			pdev = pci_find_slot(pbus, pid);
+		if (pdev == NULL)
+			return 0;
+		pci_enable_device(pdev);
+		pci_set_master(pdev);
+	}
+	return 0;
+}
+
+static long
+core99_ide_enable(struct device_node *node, long param, long value)
+{
+	/* Bus ID 0 to 2 are KeyLargo based IDE, busID 3 is U2
+	 * based ata-100
+	 */
+	switch(param) {
+	    case 0:
+		return simple_feature_tweak(node, macio_unknown,
+			KEYLARGO_FCR1, KL1_EIDE0_ENABLE, value);
+	    case 1:
+		return simple_feature_tweak(node, macio_unknown,
+			KEYLARGO_FCR1, KL1_EIDE1_ENABLE, value);
+	    case 2:
+		return simple_feature_tweak(node, macio_unknown,
+			KEYLARGO_FCR1, KL1_UIDE_ENABLE, value);
+	    case 3:
+		return core99_ata100_enable(node, value);
+	    default:
+		return -ENODEV;
+	}
+}
+
+static long
+core99_ide_reset(struct device_node *node, long param, long value)
+{
+	switch(param) {
+	    case 0:
+		return simple_feature_tweak(node, macio_unknown,
+			KEYLARGO_FCR1, KL1_EIDE0_RESET_N, !value);
+	    case 1:
+		return simple_feature_tweak(node, macio_unknown,
+			KEYLARGO_FCR1, KL1_EIDE1_RESET_N, !value);
+	    case 2:
+		return simple_feature_tweak(node, macio_unknown,
+			KEYLARGO_FCR1, KL1_UIDE_RESET_N, !value);
+	    default:
+		return -ENODEV;
+	}
+}
+
+static long
+core99_gmac_enable(struct device_node *node, long param, long value)
+{
+	unsigned long flags;
+
+	LOCK(flags);
+	if (value)
+		UN_BIS(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_GMAC);
+	else
+		UN_BIC(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_GMAC);
+	(void)UN_IN(UNI_N_CLOCK_CNTL);
+	UNLOCK(flags);
+	udelay(20);
+
+	return 0;
+}
+
+static long
+core99_gmac_phy_reset(struct device_node *node, long param, long value)
+{
+	unsigned long flags;
+	struct macio_chip *macio;
+
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo && macio->type != macio_pangea &&
+	    macio->type != macio_intrepid)
+		return -ENODEV;
+
+	LOCK(flags);
+	MACIO_OUT8(KL_GPIO_ETH_PHY_RESET, KEYLARGO_GPIO_OUTPUT_ENABLE);
+	(void)MACIO_IN8(KL_GPIO_ETH_PHY_RESET);
+	UNLOCK(flags);
+	mdelay(10);
+	LOCK(flags);
+	MACIO_OUT8(KL_GPIO_ETH_PHY_RESET, /*KEYLARGO_GPIO_OUTPUT_ENABLE | */
+		KEYLARGO_GPIO_OUTOUT_DATA);
+	UNLOCK(flags);
+	mdelay(10);
+
+	return 0;
+}
+
+static long
+core99_sound_chip_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip*	macio;
+	unsigned long		flags;
+
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+
+	/* Do a better probe code, screamer G4 desktops &
+	 * iMacs can do that too, add a recalibrate  in
+	 * the driver as well
+	 */
+	if (pmac_mb.model_id == PMAC_TYPE_PISMO ||
+	    pmac_mb.model_id == PMAC_TYPE_TITANIUM) {
+		LOCK(flags);
+		if (value)
+			MACIO_OUT8(KL_GPIO_SOUND_POWER,
+				KEYLARGO_GPIO_OUTPUT_ENABLE |
+				KEYLARGO_GPIO_OUTOUT_DATA);
+		else
+			MACIO_OUT8(KL_GPIO_SOUND_POWER,
+				KEYLARGO_GPIO_OUTPUT_ENABLE);
+		(void)MACIO_IN8(KL_GPIO_SOUND_POWER);
+		UNLOCK(flags);
+	}
+	return 0;
+}
+
+static long
+core99_airport_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip*	macio;
+	unsigned long		flags;
+	int			state;
+
+	macio = macio_find(node, 0);
+	if (!macio)
+		return -ENODEV;
+
+	/* Hint: we allow passing of macio itself for the sake of the
+	 * sleep code
+	 */
+	if (node != macio->of_node &&
+	    (!node->parent || node->parent != macio->of_node))
+		return -ENODEV;
+	state = (macio->flags & MACIO_FLAG_AIRPORT_ON) != 0;
+	if (value == state)
+		return 0;
+	if (value) {
+		/* This code is a reproduction of OF enable-cardslot
+		 * and init-wireless methods, slightly hacked until
+		 * I got it working.
+		 */
+		LOCK(flags);
+		MACIO_OUT8(KEYLARGO_GPIO_0+0xf, 5);
+		(void)MACIO_IN8(KEYLARGO_GPIO_0+0xf);
+		UNLOCK(flags);
+		mdelay(10);
+		LOCK(flags);
+		MACIO_OUT8(KEYLARGO_GPIO_0+0xf, 4);
+		(void)MACIO_IN8(KEYLARGO_GPIO_0+0xf);
+		UNLOCK(flags);
+
+		mdelay(10);
+
+		LOCK(flags);
+		MACIO_BIC(KEYLARGO_FCR2, KL2_CARDSEL_16);
+		(void)MACIO_IN32(KEYLARGO_FCR2);
+		udelay(10);
+		MACIO_OUT8(KEYLARGO_GPIO_EXTINT_0+0xb, 0);
+		(void)MACIO_IN8(KEYLARGO_GPIO_EXTINT_0+0xb);
+		udelay(10);
+		MACIO_OUT8(KEYLARGO_GPIO_EXTINT_0+0xa, 0x28);
+		(void)MACIO_IN8(KEYLARGO_GPIO_EXTINT_0+0xa);
+		udelay(10);
+		MACIO_OUT8(KEYLARGO_GPIO_EXTINT_0+0xd, 0x28);
+		(void)MACIO_IN8(KEYLARGO_GPIO_EXTINT_0+0xd);
+		udelay(10);
+		MACIO_OUT8(KEYLARGO_GPIO_0+0xd, 0x28);
+		(void)MACIO_IN8(KEYLARGO_GPIO_0+0xd);
+		udelay(10);
+		MACIO_OUT8(KEYLARGO_GPIO_0+0xe, 0x28);
+		(void)MACIO_IN8(KEYLARGO_GPIO_0+0xe);
+		UNLOCK(flags);
+		udelay(10);
+		MACIO_OUT32(0x1c000, 0);
+		mdelay(1);
+		MACIO_OUT8(0x1a3e0, 0x41);
+		(void)MACIO_IN8(0x1a3e0);
+		udelay(10);
+		LOCK(flags);
+		MACIO_BIS(KEYLARGO_FCR2, KL2_CARDSEL_16);
+		(void)MACIO_IN32(KEYLARGO_FCR2);
+		UNLOCK(flags);
+		mdelay(100);
+
+		macio->flags |= MACIO_FLAG_AIRPORT_ON;
+	} else {
+		LOCK(flags);
+		MACIO_BIC(KEYLARGO_FCR2, KL2_CARDSEL_16);
+		(void)MACIO_IN32(KEYLARGO_FCR2);
+		MACIO_OUT8(KL_GPIO_AIRPORT_0, 0);
+		MACIO_OUT8(KL_GPIO_AIRPORT_1, 0);
+		MACIO_OUT8(KL_GPIO_AIRPORT_2, 0);
+		MACIO_OUT8(KL_GPIO_AIRPORT_3, 0);
+		MACIO_OUT8(KL_GPIO_AIRPORT_4, 0);
+		(void)MACIO_IN8(KL_GPIO_AIRPORT_4);
+		UNLOCK(flags);
+
+		macio->flags &= ~MACIO_FLAG_AIRPORT_ON;
+	}
+	return 0;
+}
+
+#ifdef CONFIG_SMP
+static long
+core99_reset_cpu(struct device_node *node, long param, long value)
+{
+	unsigned int reset_io = 0;
+	unsigned long flags;
+	struct macio_chip *macio;
+	struct device_node *np;
+	const int dflt_reset_lines[] = {	KL_GPIO_RESET_CPU0,
+						KL_GPIO_RESET_CPU1,
+						KL_GPIO_RESET_CPU2,
+						KL_GPIO_RESET_CPU3 };
+
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo)
+		return -ENODEV;
+
+	np = find_path_device("/cpus");
+	if (np == NULL)
+		return -ENODEV;
+	for (np = np->child; np != NULL; np = np->sibling) {
+		u32 *num = (u32 *)get_property(np, "reg", NULL);
+		u32 *rst = (u32 *)get_property(np, "soft-reset", NULL);
+		if (num == NULL || rst == NULL)
+			continue;
+		if (param == *num) {
+			reset_io = *rst;
+			break;
+		}
+	}
+	if (np == NULL || reset_io == 0)
+		reset_io = dflt_reset_lines[param];
+
+	LOCK(flags);
+	MACIO_OUT8(reset_io, KEYLARGO_GPIO_OUTPUT_ENABLE);
+	(void)MACIO_IN8(reset_io);
+	udelay(1);
+	MACIO_OUT8(reset_io, 0);
+	(void)MACIO_IN8(reset_io);
+	UNLOCK(flags);
+
+	return 0;
+}
+#endif /* CONFIG_SMP */
+
+static long
+core99_usb_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip *macio;
+	unsigned long flags;
+	char *prop;
+	int number;
+	u32 reg;
+
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo && macio->type != macio_pangea &&
+	    macio->type != macio_intrepid)
+		return -ENODEV;
+
+	prop = (char *)get_property(node, "AAPL,clock-id", NULL);
+	if (!prop)
+		return -ENODEV;
+	if (strncmp(prop, "usb0u048", 8) == 0)
+		number = 0;
+	else if (strncmp(prop, "usb1u148", 8) == 0)
+		number = 2;
+	else if (strncmp(prop, "usb2u248", 8) == 0)
+		number = 4;
+	else
+		return -ENODEV;
+
+	/* Sorry for the brute-force locking, but this is only used during
+	 * sleep and the timing seem to be critical
+	 */
+	LOCK(flags);
+	if (value) {
+		/* Turn ON */
+		if (number == 0) {
+			MACIO_BIC(KEYLARGO_FCR0, (KL0_USB0_PAD_SUSPEND0 | KL0_USB0_PAD_SUSPEND1));
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+			UNLOCK(flags);
+			mdelay(1);
+			LOCK(flags);
+			MACIO_BIS(KEYLARGO_FCR0, KL0_USB0_CELL_ENABLE);
+		} else if (number == 2) {
+			MACIO_BIC(KEYLARGO_FCR0, (KL0_USB1_PAD_SUSPEND0 | KL0_USB1_PAD_SUSPEND1));
+			UNLOCK(flags);
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+			mdelay(1);
+			LOCK(flags);
+			MACIO_BIS(KEYLARGO_FCR0, KL0_USB1_CELL_ENABLE);
+		} else if (number == 4) {
+			MACIO_BIC(KEYLARGO_FCR1, (KL1_USB2_PAD_SUSPEND0 | KL1_USB2_PAD_SUSPEND1));
+			UNLOCK(flags);
+			(void)MACIO_IN32(KEYLARGO_FCR1);
+			mdelay(1);
+			LOCK(flags);
+			MACIO_BIS(KEYLARGO_FCR1, KL1_USB2_CELL_ENABLE);
+		}
+		if (number < 4) {
+			reg = MACIO_IN32(KEYLARGO_FCR4);
+			reg &=	~(KL4_PORT_WAKEUP_ENABLE(number) | KL4_PORT_RESUME_WAKE_EN(number) |
+				KL4_PORT_CONNECT_WAKE_EN(number) | KL4_PORT_DISCONNECT_WAKE_EN(number));
+			reg &=	~(KL4_PORT_WAKEUP_ENABLE(number+1) | KL4_PORT_RESUME_WAKE_EN(number+1) |
+				KL4_PORT_CONNECT_WAKE_EN(number+1) | KL4_PORT_DISCONNECT_WAKE_EN(number+1));
+			MACIO_OUT32(KEYLARGO_FCR4, reg);
+			(void)MACIO_IN32(KEYLARGO_FCR4);
+			udelay(10);
+		} else {
+			reg = MACIO_IN32(KEYLARGO_FCR3);
+			reg &=	~(KL3_IT_PORT_WAKEUP_ENABLE(0) | KL3_IT_PORT_RESUME_WAKE_EN(0) |
+				KL3_IT_PORT_CONNECT_WAKE_EN(0) | KL3_IT_PORT_DISCONNECT_WAKE_EN(0));
+			reg &=	~(KL3_IT_PORT_WAKEUP_ENABLE(1) | KL3_IT_PORT_RESUME_WAKE_EN(1) |
+				KL3_IT_PORT_CONNECT_WAKE_EN(1) | KL3_IT_PORT_DISCONNECT_WAKE_EN(1));
+			MACIO_OUT32(KEYLARGO_FCR3, reg);
+			(void)MACIO_IN32(KEYLARGO_FCR3);
+			udelay(10);
+		}
+		if (macio->type == macio_intrepid) {
+			/* wait for clock stopped bits to clear */
+			u32 test0 = 0, test1 = 0;
+			u32 status0, status1;
+			int timeout = 1000;
+
+			UNLOCK(flags);
+			switch (number) {
+			case 0:
+				test0 = UNI_N_CLOCK_STOPPED_USB0;
+				test1 = UNI_N_CLOCK_STOPPED_USB0PCI;
+				break;
+			case 2:
+				test0 = UNI_N_CLOCK_STOPPED_USB1;
+				test1 = UNI_N_CLOCK_STOPPED_USB1PCI;
+				break;
+			case 4:
+				test0 = UNI_N_CLOCK_STOPPED_USB2;
+				test1 = UNI_N_CLOCK_STOPPED_USB2PCI;
+				break;
+			}
+			do {
+				if (--timeout <= 0) {
+					printk(KERN_ERR "core99_usb_enable: "
+					       "Timeout waiting for clocks\n");
+					break;
+				}
+				mdelay(1);
+				status0 = UN_IN(UNI_N_CLOCK_STOP_STATUS0);
+				status1 = UN_IN(UNI_N_CLOCK_STOP_STATUS1);
+			} while ((status0 & test0) | (status1 & test1));
+			LOCK(flags);
+		}
+	} else {
+		/* Turn OFF */
+		if (number < 4) {
+			reg = MACIO_IN32(KEYLARGO_FCR4);
+			reg |=	KL4_PORT_WAKEUP_ENABLE(number) | KL4_PORT_RESUME_WAKE_EN(number) |
+				KL4_PORT_CONNECT_WAKE_EN(number) | KL4_PORT_DISCONNECT_WAKE_EN(number);
+			reg |=	KL4_PORT_WAKEUP_ENABLE(number+1) | KL4_PORT_RESUME_WAKE_EN(number+1) |
+				KL4_PORT_CONNECT_WAKE_EN(number+1) | KL4_PORT_DISCONNECT_WAKE_EN(number+1);
+			MACIO_OUT32(KEYLARGO_FCR4, reg);
+			(void)MACIO_IN32(KEYLARGO_FCR4);
+			udelay(1);
+		} else {
+			reg = MACIO_IN32(KEYLARGO_FCR3);
+			reg |=	KL3_IT_PORT_WAKEUP_ENABLE(0) | KL3_IT_PORT_RESUME_WAKE_EN(0) |
+				KL3_IT_PORT_CONNECT_WAKE_EN(0) | KL3_IT_PORT_DISCONNECT_WAKE_EN(0);
+			reg |=	KL3_IT_PORT_WAKEUP_ENABLE(1) | KL3_IT_PORT_RESUME_WAKE_EN(1) |
+				KL3_IT_PORT_CONNECT_WAKE_EN(1) | KL3_IT_PORT_DISCONNECT_WAKE_EN(1);
+			MACIO_OUT32(KEYLARGO_FCR3, reg);
+			(void)MACIO_IN32(KEYLARGO_FCR3);
+			udelay(1);
+		}
+		if (number == 0) {
+			if (macio->type != macio_intrepid)
+				MACIO_BIC(KEYLARGO_FCR0, KL0_USB0_CELL_ENABLE);
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+			udelay(1);
+			MACIO_BIS(KEYLARGO_FCR0, (KL0_USB0_PAD_SUSPEND0 | KL0_USB0_PAD_SUSPEND1));
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+		} else if (number == 2) {
+			if (macio->type != macio_intrepid)
+				MACIO_BIC(KEYLARGO_FCR0, KL0_USB1_CELL_ENABLE);
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+			udelay(1);
+			MACIO_BIS(KEYLARGO_FCR0, (KL0_USB1_PAD_SUSPEND0 | KL0_USB1_PAD_SUSPEND1));
+			(void)MACIO_IN32(KEYLARGO_FCR0);
+		} else if (number == 4) {
+			udelay(1);
+			MACIO_BIS(KEYLARGO_FCR1, (KL1_USB2_PAD_SUSPEND0 | KL1_USB2_PAD_SUSPEND1));
+			(void)MACIO_IN32(KEYLARGO_FCR1);
+		}
+		udelay(1);
+	}
+	UNLOCK(flags);
+
+	return 0;
+}
+
+static long
+core99_firewire_enable(struct device_node *node, long param, long value)
+{
+	unsigned long flags;
+	struct macio_chip *macio;
+
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo && macio->type != macio_pangea &&
+	    macio->type != macio_intrepid)
+		return -ENODEV;
+	if (!(macio->flags & MACIO_FLAG_FW_SUPPORTED))
+		return -ENODEV;
+
+	LOCK(flags);
+	if (value) {
+		UN_BIS(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_FW);
+		(void)UN_IN(UNI_N_CLOCK_CNTL);
+	} else {
+		UN_BIC(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_FW);
+		(void)UN_IN(UNI_N_CLOCK_CNTL);
+	}
+	UNLOCK(flags);
+	mdelay(1);
+
+	return 0;
+}
+
+static long
+core99_firewire_cable_power(struct device_node *node, long param, long value)
+{
+	unsigned long flags;
+	struct macio_chip *macio;
+
+	/* Trick: we allow NULL node */
+	if ((pmac_mb.board_flags & PMAC_MB_HAS_FW_POWER) == 0)
+		return -ENODEV;
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo && macio->type != macio_pangea &&
+	    macio->type != macio_intrepid)
+		return -ENODEV;
+	if (!(macio->flags & MACIO_FLAG_FW_SUPPORTED))
+		return -ENODEV;
+
+	LOCK(flags);
+	if (value) {
+		MACIO_OUT8(KL_GPIO_FW_CABLE_POWER , 0);
+		MACIO_IN8(KL_GPIO_FW_CABLE_POWER);
+		udelay(10);
+	} else {
+		MACIO_OUT8(KL_GPIO_FW_CABLE_POWER , 4);
+		MACIO_IN8(KL_GPIO_FW_CABLE_POWER); udelay(10);
+	}
+	UNLOCK(flags);
+	mdelay(1);
+
+	return 0;
+}
+
+static long
+intrepid_aack_delay_enable(struct device_node *node, long param, long value)
+{
+	unsigned long flags;
+
+	if (uninorth_rev < 0xd2)
+		return -ENODEV;
+
+	LOCK(flags);
+	if (param)
+		UN_BIS(UNI_N_AACK_DELAY, UNI_N_AACK_DELAY_ENABLE);
+	else
+		UN_BIC(UNI_N_AACK_DELAY, UNI_N_AACK_DELAY_ENABLE);
+	UNLOCK(flags);
+
+	return 0;
+}
+
+
+#endif /* CONFIG_POWER4 */
+
+static long
+core99_read_gpio(struct device_node *node, long param, long value)
+{
+	struct macio_chip *macio = &macio_chips[0];
+
+	return MACIO_IN8(param);
+}
+
+
+static long
+core99_write_gpio(struct device_node *node, long param, long value)
+{
+	struct macio_chip *macio = &macio_chips[0];
+
+	MACIO_OUT8(param, (u8)(value & 0xff));
+	return 0;
+}
+
+#ifdef CONFIG_POWER4
+static long g5_gmac_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip *macio = &macio_chips[0];
+	unsigned long flags;
+
+	if (node == NULL)
+		return -ENODEV;
+
+	LOCK(flags);
+	if (value) {
+		MACIO_BIS(KEYLARGO_FCR1, K2_FCR1_GMAC_CLK_ENABLE);
+		mb();
+		k2_skiplist[0] = NULL;
+	} else {
+		k2_skiplist[0] = node;
+		mb();
+		MACIO_BIC(KEYLARGO_FCR1, K2_FCR1_GMAC_CLK_ENABLE);
+	}
+	
+	UNLOCK(flags);
+	mdelay(1);
+
+	return 0;
+}
+
+static long g5_fw_enable(struct device_node *node, long param, long value)
+{
+	struct macio_chip *macio = &macio_chips[0];
+	unsigned long flags;
+
+	if (node == NULL)
+		return -ENODEV;
+
+	LOCK(flags);
+	if (value) {
+		MACIO_BIS(KEYLARGO_FCR1, K2_FCR1_FW_CLK_ENABLE);
+		mb();
+		k2_skiplist[1] = NULL;
+	} else {
+		k2_skiplist[1] = node;
+		mb();
+		MACIO_BIC(KEYLARGO_FCR1, K2_FCR1_FW_CLK_ENABLE);
+	}
+	
+	UNLOCK(flags);
+	mdelay(1);
+
+	return 0;
+}
+
+static long g5_mpic_enable(struct device_node *node, long param, long value)
+{
+	unsigned long flags;
+
+	if (node->parent == NULL || strcmp(node->parent->name, "u3"))
+		return 0;
+
+	LOCK(flags);
+	UN_BIS(U3_TOGGLE_REG, U3_MPIC_RESET | U3_MPIC_OUTPUT_ENABLE);
+	UNLOCK(flags);
+
+	return 0;
+}
+
+static long g5_eth_phy_reset(struct device_node *node, long param, long value)
+{
+	struct macio_chip *macio = &macio_chips[0];
+	struct device_node *phy;
+	int need_reset;
+
+	/*
+	 * We must not reset the combo PHYs, only the BCM5221 found in
+	 * the iMac G5.
+	 */
+	phy = of_get_next_child(node, NULL);
+	if (!phy)
+		return -ENODEV;
+	need_reset = device_is_compatible(phy, "B5221");
+	of_node_put(phy);
+	if (!need_reset)
+		return 0;
+
+	/* PHY reset is GPIO 29, not in device-tree unfortunately */
+	MACIO_OUT8(K2_GPIO_EXTINT_0 + 29,
+		   KEYLARGO_GPIO_OUTPUT_ENABLE | KEYLARGO_GPIO_OUTOUT_DATA);
+	/* Thankfully, this is now always called at a time when we can
+	 * schedule by sungem.
+	 */
+	msleep(10);
+	MACIO_OUT8(K2_GPIO_EXTINT_0 + 29, 0);
+
+	return 0;
+}
+
+static long g5_i2s_enable(struct device_node *node, long param, long value)
+{
+	/* Very crude implementation for now */
+	struct macio_chip *macio = &macio_chips[0];
+	unsigned long flags;
+
+	if (value == 0)
+		return 0; /* don't disable yet */
+
+	LOCK(flags);
+	MACIO_BIS(KEYLARGO_FCR3, KL3_CLK45_ENABLE | KL3_CLK49_ENABLE |
+		  KL3_I2S0_CLK18_ENABLE);
+	udelay(10);
+	MACIO_BIS(KEYLARGO_FCR1, K2_FCR1_I2S0_CELL_ENABLE |
+		  K2_FCR1_I2S0_CLK_ENABLE_BIT | K2_FCR1_I2S0_ENABLE);
+	udelay(10);
+	MACIO_BIC(KEYLARGO_FCR1, K2_FCR1_I2S0_RESET);
+	UNLOCK(flags);
+	udelay(10);
+
+	return 0;
+}
+
+
+#ifdef CONFIG_SMP
+static long g5_reset_cpu(struct device_node *node, long param, long value)
+{
+	unsigned int reset_io = 0;
+	unsigned long flags;
+	struct macio_chip *macio;
+	struct device_node *np;
+
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo2)
+		return -ENODEV;
+
+	np = find_path_device("/cpus");
+	if (np == NULL)
+		return -ENODEV;
+	for (np = np->child; np != NULL; np = np->sibling) {
+		u32 *num = (u32 *)get_property(np, "reg", NULL);
+		u32 *rst = (u32 *)get_property(np, "soft-reset", NULL);
+		if (num == NULL || rst == NULL)
+			continue;
+		if (param == *num) {
+			reset_io = *rst;
+			break;
+		}
+	}
+	if (np == NULL || reset_io == 0)
+		return -ENODEV;
+
+	LOCK(flags);
+	MACIO_OUT8(reset_io, KEYLARGO_GPIO_OUTPUT_ENABLE);
+	(void)MACIO_IN8(reset_io);
+	udelay(1);
+	MACIO_OUT8(reset_io, 0);
+	(void)MACIO_IN8(reset_io);
+	UNLOCK(flags);
+
+	return 0;
+}
+#endif /* CONFIG_SMP */
+
+/*
+ * This can be called from pmac_smp so isn't static
+ *
+ * This takes the second CPU off the bus on dual CPU machines
+ * running UP
+ */
+void g5_phy_disable_cpu1(void)
+{
+	UN_OUT(U3_API_PHY_CONFIG_1, 0);
+}
+#endif /* CONFIG_POWER4 */
+
+#ifndef CONFIG_POWER4
+
+static void
+keylargo_shutdown(struct macio_chip *macio, int sleep_mode)
+{
+	u32 temp;
+
+	if (sleep_mode) {
+		mdelay(1);
+		MACIO_BIS(KEYLARGO_FCR0, KL0_USB_REF_SUSPEND);
+		(void)MACIO_IN32(KEYLARGO_FCR0);
+		mdelay(1);
+	}
+
+	MACIO_BIC(KEYLARGO_FCR0,KL0_SCCA_ENABLE | KL0_SCCB_ENABLE |
+				KL0_SCC_CELL_ENABLE |
+				KL0_IRDA_ENABLE | KL0_IRDA_CLK32_ENABLE |
+				KL0_IRDA_CLK19_ENABLE);
+
+	MACIO_BIC(KEYLARGO_MBCR, KL_MBCR_MB0_DEV_MASK);
+	MACIO_BIS(KEYLARGO_MBCR, KL_MBCR_MB0_IDE_ENABLE);
+
+	MACIO_BIC(KEYLARGO_FCR1,
+		KL1_AUDIO_SEL_22MCLK | KL1_AUDIO_CLK_ENABLE_BIT |
+		KL1_AUDIO_CLK_OUT_ENABLE | KL1_AUDIO_CELL_ENABLE |
+		KL1_I2S0_CELL_ENABLE | KL1_I2S0_CLK_ENABLE_BIT |
+		KL1_I2S0_ENABLE | KL1_I2S1_CELL_ENABLE |
+		KL1_I2S1_CLK_ENABLE_BIT | KL1_I2S1_ENABLE |
+		KL1_EIDE0_ENABLE | KL1_EIDE0_RESET_N |
+		KL1_EIDE1_ENABLE | KL1_EIDE1_RESET_N |
+		KL1_UIDE_ENABLE);
+
+	MACIO_BIS(KEYLARGO_FCR2, KL2_ALT_DATA_OUT);
+	MACIO_BIC(KEYLARGO_FCR2, KL2_IOBUS_ENABLE);
+
+	temp = MACIO_IN32(KEYLARGO_FCR3);
+	if (macio->rev >= 2) {
+		temp |= KL3_SHUTDOWN_PLL2X;
+		if (sleep_mode)
+			temp |= KL3_SHUTDOWN_PLL_TOTAL;
+	}
+
+	temp |= KL3_SHUTDOWN_PLLKW6 | KL3_SHUTDOWN_PLLKW4 |
+		KL3_SHUTDOWN_PLLKW35;
+	if (sleep_mode)
+		temp |= KL3_SHUTDOWN_PLLKW12;
+	temp &= ~(KL3_CLK66_ENABLE | KL3_CLK49_ENABLE | KL3_CLK45_ENABLE
+		| KL3_CLK31_ENABLE | KL3_I2S1_CLK18_ENABLE | KL3_I2S0_CLK18_ENABLE);
+	if (sleep_mode)
+		temp &= ~(KL3_TIMER_CLK18_ENABLE | KL3_VIA_CLK16_ENABLE);
+	MACIO_OUT32(KEYLARGO_FCR3, temp);
+
+	/* Flush posted writes & wait a bit */
+	(void)MACIO_IN32(KEYLARGO_FCR0); mdelay(1);
+}
+
+static void
+pangea_shutdown(struct macio_chip *macio, int sleep_mode)
+{
+	u32 temp;
+
+	MACIO_BIC(KEYLARGO_FCR0,KL0_SCCA_ENABLE | KL0_SCCB_ENABLE |
+				KL0_SCC_CELL_ENABLE |
+				KL0_USB0_CELL_ENABLE | KL0_USB1_CELL_ENABLE);
+
+	MACIO_BIC(KEYLARGO_FCR1,
+		KL1_AUDIO_SEL_22MCLK | KL1_AUDIO_CLK_ENABLE_BIT |
+		KL1_AUDIO_CLK_OUT_ENABLE | KL1_AUDIO_CELL_ENABLE |
+		KL1_I2S0_CELL_ENABLE | KL1_I2S0_CLK_ENABLE_BIT |
+		KL1_I2S0_ENABLE | KL1_I2S1_CELL_ENABLE |
+		KL1_I2S1_CLK_ENABLE_BIT | KL1_I2S1_ENABLE |
+		KL1_UIDE_ENABLE);
+	if (pmac_mb.board_flags & PMAC_MB_MOBILE)
+		MACIO_BIC(KEYLARGO_FCR1, KL1_UIDE_RESET_N);
+
+	MACIO_BIS(KEYLARGO_FCR2, KL2_ALT_DATA_OUT);
+
+	temp = MACIO_IN32(KEYLARGO_FCR3);
+	temp |= KL3_SHUTDOWN_PLLKW6 | KL3_SHUTDOWN_PLLKW4 |
+		KL3_SHUTDOWN_PLLKW35;
+	temp &= ~(KL3_CLK49_ENABLE | KL3_CLK45_ENABLE | KL3_CLK31_ENABLE
+		| KL3_I2S0_CLK18_ENABLE | KL3_I2S1_CLK18_ENABLE);
+	if (sleep_mode)
+		temp &= ~(KL3_VIA_CLK16_ENABLE | KL3_TIMER_CLK18_ENABLE);
+	MACIO_OUT32(KEYLARGO_FCR3, temp);
+
+	/* Flush posted writes & wait a bit */
+	(void)MACIO_IN32(KEYLARGO_FCR0); mdelay(1);
+}
+
+static void
+intrepid_shutdown(struct macio_chip *macio, int sleep_mode)
+{
+	u32 temp;
+
+	MACIO_BIC(KEYLARGO_FCR0,KL0_SCCA_ENABLE | KL0_SCCB_ENABLE |
+		  KL0_SCC_CELL_ENABLE);
+
+	MACIO_BIC(KEYLARGO_FCR1,
+		  /*KL1_USB2_CELL_ENABLE |*/
+		KL1_I2S0_CELL_ENABLE | KL1_I2S0_CLK_ENABLE_BIT |
+		KL1_I2S0_ENABLE | KL1_I2S1_CELL_ENABLE |
+		KL1_I2S1_CLK_ENABLE_BIT | KL1_I2S1_ENABLE);
+	if (pmac_mb.board_flags & PMAC_MB_MOBILE)
+		MACIO_BIC(KEYLARGO_FCR1, KL1_UIDE_RESET_N);
+
+	temp = MACIO_IN32(KEYLARGO_FCR3);
+	temp &= ~(KL3_CLK49_ENABLE | KL3_CLK45_ENABLE |
+		  KL3_I2S1_CLK18_ENABLE | KL3_I2S0_CLK18_ENABLE);
+	if (sleep_mode)
+		temp &= ~(KL3_TIMER_CLK18_ENABLE | KL3_IT_VIA_CLK32_ENABLE);
+	MACIO_OUT32(KEYLARGO_FCR3, temp);
+
+	/* Flush posted writes & wait a bit */
+	(void)MACIO_IN32(KEYLARGO_FCR0);
+	mdelay(10);
+}
+
+
+void pmac_tweak_clock_spreading(int enable)
+{
+	struct macio_chip *macio = &macio_chips[0];
+
+	/* Hack for doing clock spreading on some machines PowerBooks and
+	 * iBooks. This implements the "platform-do-clockspreading" OF
+	 * property as decoded manually on various models. For safety, we also
+	 * check the product ID in the device-tree in cases we'll whack the i2c
+	 * chip to make reasonably sure we won't set wrong values in there
+	 *
+	 * Of course, ultimately, we have to implement a real parser for
+	 * the platform-do-* stuff...
+	 */
+
+	if (macio->type == macio_intrepid) {
+		if (enable)
+			UN_OUT(UNI_N_CLOCK_SPREADING, 2);
+		else
+			UN_OUT(UNI_N_CLOCK_SPREADING, 0);
+		mdelay(40);
+	}
+
+	while (machine_is_compatible("PowerBook5,2") ||
+	       machine_is_compatible("PowerBook5,3") ||
+	       machine_is_compatible("PowerBook6,2") ||
+	       machine_is_compatible("PowerBook6,3")) {
+		struct device_node *ui2c = of_find_node_by_type(NULL, "i2c");
+		struct device_node *dt = of_find_node_by_name(NULL, "device-tree");
+		u8 buffer[9];
+		u32 *productID;
+		int i, rc, changed = 0;
+
+		if (dt == NULL)
+			break;
+		productID = (u32 *)get_property(dt, "pid#", NULL);
+		if (productID == NULL)
+			break;
+		while(ui2c) {
+			struct device_node *p = of_get_parent(ui2c);
+			if (p && !strcmp(p->name, "uni-n"))
+				break;
+			ui2c = of_find_node_by_type(ui2c, "i2c");
+		}
+		if (ui2c == NULL)
+			break;
+		DBG("Trying to bump clock speed for PID: %08x...\n", *productID);
+		rc = pmac_low_i2c_open(ui2c, 1);
+		if (rc != 0)
+			break;
+		pmac_low_i2c_setmode(ui2c, pmac_low_i2c_mode_combined);
+		rc = pmac_low_i2c_xfer(ui2c, 0xd2 | pmac_low_i2c_read, 0x80, buffer, 9);
+		DBG("read result: %d,", rc);
+		if (rc != 0) {
+			pmac_low_i2c_close(ui2c);
+			break;
+		}
+		for (i=0; i<9; i++)
+			DBG(" %02x", buffer[i]);
+		DBG("\n");
+
+		switch(*productID) {
+		case 0x1182:	/* AlBook 12" rev 2 */
+		case 0x1183:	/* iBook G4 12" */
+			buffer[0] = (buffer[0] & 0x8f) | 0x70;
+			buffer[2] = (buffer[2] & 0x7f) | 0x00;
+			buffer[5] = (buffer[5] & 0x80) | 0x31;
+			buffer[6] = (buffer[6] & 0x40) | 0xb0;
+			buffer[7] = (buffer[7] & 0x00) | (enable ? 0xc0 : 0xba);
+			buffer[8] = (buffer[8] & 0x00) | 0x30;
+			changed = 1;
+			break;
+		case 0x3142:	/* AlBook 15" (ATI M10) */
+		case 0x3143:	/* AlBook 17" (ATI M10) */
+			buffer[0] = (buffer[0] & 0xaf) | 0x50;
+			buffer[2] = (buffer[2] & 0x7f) | 0x00;
+			buffer[5] = (buffer[5] & 0x80) | 0x31;
+			buffer[6] = (buffer[6] & 0x40) | 0xb0;
+			buffer[7] = (buffer[7] & 0x00) | (enable ? 0xd0 : 0xc0);
+			buffer[8] = (buffer[8] & 0x00) | 0x30;
+			changed = 1;
+			break;
+		default:
+			DBG("i2c-hwclock: Machine model not handled\n");
+			break;
+		}
+		if (!changed) {
+			pmac_low_i2c_close(ui2c);
+			break;
+		}
+		pmac_low_i2c_setmode(ui2c, pmac_low_i2c_mode_stdsub);
+		rc = pmac_low_i2c_xfer(ui2c, 0xd2 | pmac_low_i2c_write, 0x80, buffer, 9);
+		DBG("write result: %d,", rc);
+		pmac_low_i2c_setmode(ui2c, pmac_low_i2c_mode_combined);
+		rc = pmac_low_i2c_xfer(ui2c, 0xd2 | pmac_low_i2c_read, 0x80, buffer, 9);
+		DBG("read result: %d,", rc);
+		if (rc != 0) {
+			pmac_low_i2c_close(ui2c);
+			break;
+		}
+		for (i=0; i<9; i++)
+			DBG(" %02x", buffer[i]);
+		pmac_low_i2c_close(ui2c);
+		break;
+	}
+}
+
+
+static int
+core99_sleep(void)
+{
+	struct macio_chip *macio;
+	int i;
+
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo && macio->type != macio_pangea &&
+	    macio->type != macio_intrepid)
+		return -ENODEV;
+
+	/* We power off the wireless slot in case it was not done
+	 * by the driver. We don't power it on automatically however
+	 */
+	if (macio->flags & MACIO_FLAG_AIRPORT_ON)
+		core99_airport_enable(macio->of_node, 0, 0);
+
+	/* We power off the FW cable. Should be done by the driver... */
+	if (macio->flags & MACIO_FLAG_FW_SUPPORTED) {
+		core99_firewire_enable(NULL, 0, 0);
+		core99_firewire_cable_power(NULL, 0, 0);
+	}
+
+	/* We make sure int. modem is off (in case driver lost it) */
+	if (macio->type == macio_keylargo)
+		core99_modem_enable(macio->of_node, 0, 0);
+	else
+		pangea_modem_enable(macio->of_node, 0, 0);
+
+	/* We make sure the sound is off as well */
+	core99_sound_chip_enable(macio->of_node, 0, 0);
+
+	/*
+	 * Save various bits of KeyLargo
+	 */
+
+	/* Save the state of the various GPIOs */
+	save_gpio_levels[0] = MACIO_IN32(KEYLARGO_GPIO_LEVELS0);
+	save_gpio_levels[1] = MACIO_IN32(KEYLARGO_GPIO_LEVELS1);
+	for (i=0; i<KEYLARGO_GPIO_EXTINT_CNT; i++)
+		save_gpio_extint[i] = MACIO_IN8(KEYLARGO_GPIO_EXTINT_0+i);
+	for (i=0; i<KEYLARGO_GPIO_CNT; i++)
+		save_gpio_normal[i] = MACIO_IN8(KEYLARGO_GPIO_0+i);
+
+	/* Save the FCRs */
+	if (macio->type == macio_keylargo)
+		save_mbcr = MACIO_IN32(KEYLARGO_MBCR);
+	save_fcr[0] = MACIO_IN32(KEYLARGO_FCR0);
+	save_fcr[1] = MACIO_IN32(KEYLARGO_FCR1);
+	save_fcr[2] = MACIO_IN32(KEYLARGO_FCR2);
+	save_fcr[3] = MACIO_IN32(KEYLARGO_FCR3);
+	save_fcr[4] = MACIO_IN32(KEYLARGO_FCR4);
+	if (macio->type == macio_pangea || macio->type == macio_intrepid)
+		save_fcr[5] = MACIO_IN32(KEYLARGO_FCR5);
+
+	/* Save state & config of DBDMA channels */
+	dbdma_save(macio, save_dbdma);
+
+	/*
+	 * Turn off as much as we can
+	 */
+	if (macio->type == macio_pangea)
+		pangea_shutdown(macio, 1);
+	else if (macio->type == macio_intrepid)
+		intrepid_shutdown(macio, 1);
+	else if (macio->type == macio_keylargo)
+		keylargo_shutdown(macio, 1);
+
+	/*
+	 * Put the host bridge to sleep
+	 */
+
+	save_unin_clock_ctl = UN_IN(UNI_N_CLOCK_CNTL);
+	/* Note: do not switch GMAC off, driver does it when necessary, WOL must keep it
+	 * enabled !
+	 */
+	UN_OUT(UNI_N_CLOCK_CNTL, save_unin_clock_ctl &
+	       ~(/*UNI_N_CLOCK_CNTL_GMAC|*/UNI_N_CLOCK_CNTL_FW/*|UNI_N_CLOCK_CNTL_PCI*/));
+	udelay(100);
+	UN_OUT(UNI_N_HWINIT_STATE, UNI_N_HWINIT_STATE_SLEEPING);
+	UN_OUT(UNI_N_POWER_MGT, UNI_N_POWER_MGT_SLEEP);
+	mdelay(10);
+
+	/*
+	 * FIXME: A bit of black magic with OpenPIC (don't ask me why)
+	 */
+	if (pmac_mb.model_id == PMAC_TYPE_SAWTOOTH) {
+		MACIO_BIS(0x506e0, 0x00400000);
+		MACIO_BIS(0x506e0, 0x80000000);
+	}
+	return 0;
+}
+
+static int
+core99_wake_up(void)
+{
+	struct macio_chip *macio;
+	int i;
+
+	macio = &macio_chips[0];
+	if (macio->type != macio_keylargo && macio->type != macio_pangea &&
+	    macio->type != macio_intrepid)
+		return -ENODEV;
+
+	/*
+	 * Wakeup the host bridge
+	 */
+	UN_OUT(UNI_N_POWER_MGT, UNI_N_POWER_MGT_NORMAL);
+	udelay(10);
+	UN_OUT(UNI_N_HWINIT_STATE, UNI_N_HWINIT_STATE_RUNNING);
+	udelay(10);
+
+	/*
+	 * Restore KeyLargo
+	 */
+
+	if (macio->type == macio_keylargo) {
+		MACIO_OUT32(KEYLARGO_MBCR, save_mbcr);
+		(void)MACIO_IN32(KEYLARGO_MBCR); udelay(10);
+	}
+	MACIO_OUT32(KEYLARGO_FCR0, save_fcr[0]);
+	(void)MACIO_IN32(KEYLARGO_FCR0); udelay(10);
+	MACIO_OUT32(KEYLARGO_FCR1, save_fcr[1]);
+	(void)MACIO_IN32(KEYLARGO_FCR1); udelay(10);
+	MACIO_OUT32(KEYLARGO_FCR2, save_fcr[2]);
+	(void)MACIO_IN32(KEYLARGO_FCR2); udelay(10);
+	MACIO_OUT32(KEYLARGO_FCR3, save_fcr[3]);
+	(void)MACIO_IN32(KEYLARGO_FCR3); udelay(10);
+	MACIO_OUT32(KEYLARGO_FCR4, save_fcr[4]);
+	(void)MACIO_IN32(KEYLARGO_FCR4); udelay(10);
+	if (macio->type == macio_pangea || macio->type == macio_intrepid) {
+		MACIO_OUT32(KEYLARGO_FCR5, save_fcr[5]);
+		(void)MACIO_IN32(KEYLARGO_FCR5); udelay(10);
+	}
+
+	dbdma_restore(macio, save_dbdma);
+
+	MACIO_OUT32(KEYLARGO_GPIO_LEVELS0, save_gpio_levels[0]);
+	MACIO_OUT32(KEYLARGO_GPIO_LEVELS1, save_gpio_levels[1]);
+	for (i=0; i<KEYLARGO_GPIO_EXTINT_CNT; i++)
+		MACIO_OUT8(KEYLARGO_GPIO_EXTINT_0+i, save_gpio_extint[i]);
+	for (i=0; i<KEYLARGO_GPIO_CNT; i++)
+		MACIO_OUT8(KEYLARGO_GPIO_0+i, save_gpio_normal[i]);
+
+	/* FIXME more black magic with OpenPIC ... */
+	if (pmac_mb.model_id == PMAC_TYPE_SAWTOOTH) {
+		MACIO_BIC(0x506e0, 0x00400000);
+		MACIO_BIC(0x506e0, 0x80000000);
+	}
+
+	UN_OUT(UNI_N_CLOCK_CNTL, save_unin_clock_ctl);
+	udelay(100);
+
+	return 0;
+}
+
+static long
+core99_sleep_state(struct device_node *node, long param, long value)
+{
+	/* Param == 1 means to enter the "fake sleep" mode that is
+	 * used for CPU speed switch
+	 */
+	if (param == 1) {
+		if (value == 1) {
+			UN_OUT(UNI_N_HWINIT_STATE, UNI_N_HWINIT_STATE_SLEEPING);
+			UN_OUT(UNI_N_POWER_MGT, UNI_N_POWER_MGT_IDLE2);
+		} else {
+			UN_OUT(UNI_N_POWER_MGT, UNI_N_POWER_MGT_NORMAL);
+			udelay(10);
+			UN_OUT(UNI_N_HWINIT_STATE, UNI_N_HWINIT_STATE_RUNNING);
+			udelay(10);
+		}
+		return 0;
+	}
+	if ((pmac_mb.board_flags & PMAC_MB_CAN_SLEEP) == 0)
+		return -EPERM;
+
+	if (value == 1)
+		return core99_sleep();
+	else if (value == 0)
+		return core99_wake_up();
+	return 0;
+}
+
+#endif /* CONFIG_POWER4 */
+
+static long
+generic_dev_can_wake(struct device_node *node, long param, long value)
+{
+	/* Todo: eventually check we are really dealing with on-board
+	 * video device ...
+	 */
+
+	if (pmac_mb.board_flags & PMAC_MB_MAY_SLEEP)
+		pmac_mb.board_flags |= PMAC_MB_CAN_SLEEP;
+	return 0;
+}
+
+static long generic_get_mb_info(struct device_node *node, long param, long value)
+{
+	switch(param) {
+		case PMAC_MB_INFO_MODEL:
+			return pmac_mb.model_id;
+		case PMAC_MB_INFO_FLAGS:
+			return pmac_mb.board_flags;
+		case PMAC_MB_INFO_NAME:
+			/* hack hack hack... but should work */
+			*((const char **)value) = pmac_mb.model_name;
+			return 0;
+	}
+	return -EINVAL;
+}
+
+
+/*
+ * Table definitions
+ */
+
+/* Used on any machine
+ */
+static struct feature_table_entry any_features[] = {
+	{ PMAC_FTR_GET_MB_INFO,		generic_get_mb_info },
+	{ PMAC_FTR_DEVICE_CAN_WAKE,	generic_dev_can_wake },
+	{ 0, NULL }
+};
+
+#ifndef CONFIG_POWER4
+
+/* OHare based motherboards. Currently, we only use these on the
+ * 2400,3400 and 3500 series powerbooks. Some older desktops seem
+ * to have issues with turning on/off those asic cells
+ */
+static struct feature_table_entry ohare_features[] = {
+	{ PMAC_FTR_SCC_ENABLE,		ohare_htw_scc_enable },
+	{ PMAC_FTR_SWIM3_ENABLE,	ohare_floppy_enable },
+	{ PMAC_FTR_MESH_ENABLE,		ohare_mesh_enable },
+	{ PMAC_FTR_IDE_ENABLE,		ohare_ide_enable},
+	{ PMAC_FTR_IDE_RESET,		ohare_ide_reset},
+	{ PMAC_FTR_SLEEP_STATE,		ohare_sleep_state },
+	{ 0, NULL }
+};
+
+/* Heathrow desktop machines (Beige G3).
+ * Separated as some features couldn't be properly tested
+ * and the serial port control bits appear to confuse it.
+ */
+static struct feature_table_entry heathrow_desktop_features[] = {
+	{ PMAC_FTR_SWIM3_ENABLE,	heathrow_floppy_enable },
+	{ PMAC_FTR_MESH_ENABLE,		heathrow_mesh_enable },
+	{ PMAC_FTR_IDE_ENABLE,		heathrow_ide_enable },
+	{ PMAC_FTR_IDE_RESET,		heathrow_ide_reset },
+	{ PMAC_FTR_BMAC_ENABLE,		heathrow_bmac_enable },
+	{ 0, NULL }
+};
+
+/* Heathrow based laptop, that is the Wallstreet and mainstreet
+ * powerbooks.
+ */
+static struct feature_table_entry heathrow_laptop_features[] = {
+	{ PMAC_FTR_SCC_ENABLE,		ohare_htw_scc_enable },
+	{ PMAC_FTR_MODEM_ENABLE,	heathrow_modem_enable },
+	{ PMAC_FTR_SWIM3_ENABLE,	heathrow_floppy_enable },
+	{ PMAC_FTR_MESH_ENABLE,		heathrow_mesh_enable },
+	{ PMAC_FTR_IDE_ENABLE,		heathrow_ide_enable },
+	{ PMAC_FTR_IDE_RESET,		heathrow_ide_reset },
+	{ PMAC_FTR_BMAC_ENABLE,		heathrow_bmac_enable },
+	{ PMAC_FTR_SOUND_CHIP_ENABLE,	heathrow_sound_enable },
+	{ PMAC_FTR_SLEEP_STATE,		heathrow_sleep_state },
+	{ 0, NULL }
+};
+
+/* Paddington based machines
+ * The lombard (101) powerbook, first iMac models, B&W G3 and Yikes G4.
+ */
+static struct feature_table_entry paddington_features[] = {
+	{ PMAC_FTR_SCC_ENABLE,		ohare_htw_scc_enable },
+	{ PMAC_FTR_MODEM_ENABLE,	heathrow_modem_enable },
+	{ PMAC_FTR_SWIM3_ENABLE,	heathrow_floppy_enable },
+	{ PMAC_FTR_MESH_ENABLE,		heathrow_mesh_enable },
+	{ PMAC_FTR_IDE_ENABLE,		heathrow_ide_enable },
+	{ PMAC_FTR_IDE_RESET,		heathrow_ide_reset },
+	{ PMAC_FTR_BMAC_ENABLE,		heathrow_bmac_enable },
+	{ PMAC_FTR_SOUND_CHIP_ENABLE,	heathrow_sound_enable },
+	{ PMAC_FTR_SLEEP_STATE,		heathrow_sleep_state },
+	{ 0, NULL }
+};
+
+/* Core99 & MacRISC 2 machines (all machines released since the
+ * iBook (included), that is all AGP machines, except pangea
+ * chipset. The pangea chipset is the "combo" UniNorth/KeyLargo
+ * used on iBook2 & iMac "flow power".
+ */
+static struct feature_table_entry core99_features[] = {
+	{ PMAC_FTR_SCC_ENABLE,		core99_scc_enable },
+	{ PMAC_FTR_MODEM_ENABLE,	core99_modem_enable },
+	{ PMAC_FTR_IDE_ENABLE,		core99_ide_enable },
+	{ PMAC_FTR_IDE_RESET,		core99_ide_reset },
+	{ PMAC_FTR_GMAC_ENABLE,		core99_gmac_enable },
+	{ PMAC_FTR_GMAC_PHY_RESET,	core99_gmac_phy_reset },
+	{ PMAC_FTR_SOUND_CHIP_ENABLE,	core99_sound_chip_enable },
+	{ PMAC_FTR_AIRPORT_ENABLE,	core99_airport_enable },
+	{ PMAC_FTR_USB_ENABLE,		core99_usb_enable },
+	{ PMAC_FTR_1394_ENABLE,		core99_firewire_enable },
+	{ PMAC_FTR_1394_CABLE_POWER,	core99_firewire_cable_power },
+	{ PMAC_FTR_SLEEP_STATE,		core99_sleep_state },
+#ifdef CONFIG_SMP
+	{ PMAC_FTR_RESET_CPU,		core99_reset_cpu },
+#endif /* CONFIG_SMP */
+	{ PMAC_FTR_READ_GPIO,		core99_read_gpio },
+	{ PMAC_FTR_WRITE_GPIO,		core99_write_gpio },
+	{ 0, NULL }
+};
+
+/* RackMac
+ */
+static struct feature_table_entry rackmac_features[] = {
+	{ PMAC_FTR_SCC_ENABLE,		core99_scc_enable },
+	{ PMAC_FTR_IDE_ENABLE,		core99_ide_enable },
+	{ PMAC_FTR_IDE_RESET,		core99_ide_reset },
+	{ PMAC_FTR_GMAC_ENABLE,		core99_gmac_enable },
+	{ PMAC_FTR_GMAC_PHY_RESET,	core99_gmac_phy_reset },
+	{ PMAC_FTR_USB_ENABLE,		core99_usb_enable },
+	{ PMAC_FTR_1394_ENABLE,		core99_firewire_enable },
+	{ PMAC_FTR_1394_CABLE_POWER,	core99_firewire_cable_power },
+	{ PMAC_FTR_SLEEP_STATE,		core99_sleep_state },
+#ifdef CONFIG_SMP
+	{ PMAC_FTR_RESET_CPU,		core99_reset_cpu },
+#endif /* CONFIG_SMP */
+	{ PMAC_FTR_READ_GPIO,		core99_read_gpio },
+	{ PMAC_FTR_WRITE_GPIO,		core99_write_gpio },
+	{ 0, NULL }
+};
+
+/* Pangea features
+ */
+static struct feature_table_entry pangea_features[] = {
+	{ PMAC_FTR_SCC_ENABLE,		core99_scc_enable },
+	{ PMAC_FTR_MODEM_ENABLE,	pangea_modem_enable },
+	{ PMAC_FTR_IDE_ENABLE,		core99_ide_enable },
+	{ PMAC_FTR_IDE_RESET,		core99_ide_reset },
+	{ PMAC_FTR_GMAC_ENABLE,		core99_gmac_enable },
+	{ PMAC_FTR_GMAC_PHY_RESET,	core99_gmac_phy_reset },
+	{ PMAC_FTR_SOUND_CHIP_ENABLE,	core99_sound_chip_enable },
+	{ PMAC_FTR_AIRPORT_ENABLE,	core99_airport_enable },
+	{ PMAC_FTR_USB_ENABLE,		core99_usb_enable },
+	{ PMAC_FTR_1394_ENABLE,		core99_firewire_enable },
+	{ PMAC_FTR_1394_CABLE_POWER,	core99_firewire_cable_power },
+	{ PMAC_FTR_SLEEP_STATE,		core99_sleep_state },
+	{ PMAC_FTR_READ_GPIO,		core99_read_gpio },
+	{ PMAC_FTR_WRITE_GPIO,		core99_write_gpio },
+	{ 0, NULL }
+};
+
+/* Intrepid features
+ */
+static struct feature_table_entry intrepid_features[] = {
+	{ PMAC_FTR_SCC_ENABLE,		core99_scc_enable },
+	{ PMAC_FTR_MODEM_ENABLE,	pangea_modem_enable },
+	{ PMAC_FTR_IDE_ENABLE,		core99_ide_enable },
+	{ PMAC_FTR_IDE_RESET,		core99_ide_reset },
+	{ PMAC_FTR_GMAC_ENABLE,		core99_gmac_enable },
+	{ PMAC_FTR_GMAC_PHY_RESET,	core99_gmac_phy_reset },
+	{ PMAC_FTR_SOUND_CHIP_ENABLE,	core99_sound_chip_enable },
+	{ PMAC_FTR_AIRPORT_ENABLE,	core99_airport_enable },
+	{ PMAC_FTR_USB_ENABLE,		core99_usb_enable },
+	{ PMAC_FTR_1394_ENABLE,		core99_firewire_enable },
+	{ PMAC_FTR_1394_CABLE_POWER,	core99_firewire_cable_power },
+	{ PMAC_FTR_SLEEP_STATE,		core99_sleep_state },
+	{ PMAC_FTR_READ_GPIO,		core99_read_gpio },
+	{ PMAC_FTR_WRITE_GPIO,		core99_write_gpio },
+	{ PMAC_FTR_AACK_DELAY_ENABLE,	intrepid_aack_delay_enable },
+	{ 0, NULL }
+};
+
+#else /* CONFIG_POWER4 */
+
+/* G5 features
+ */
+static struct feature_table_entry g5_features[] = {
+	{ PMAC_FTR_GMAC_ENABLE,		g5_gmac_enable },
+	{ PMAC_FTR_1394_ENABLE,		g5_fw_enable },
+	{ PMAC_FTR_ENABLE_MPIC,		g5_mpic_enable },
+	{ PMAC_FTR_GMAC_PHY_RESET,	g5_eth_phy_reset },
+	{ PMAC_FTR_SOUND_CHIP_ENABLE,	g5_i2s_enable },
+#ifdef CONFIG_SMP
+	{ PMAC_FTR_RESET_CPU,		g5_reset_cpu },
+#endif /* CONFIG_SMP */
+	{ PMAC_FTR_READ_GPIO,		core99_read_gpio },
+	{ PMAC_FTR_WRITE_GPIO,		core99_write_gpio },
+	{ 0, NULL }
+};
+
+#endif /* CONFIG_POWER4 */
+
+static struct pmac_mb_def pmac_mb_defs[] = {
+#ifndef CONFIG_POWER4
+	/*
+	 * Desktops
+	 */
+
+	{	"AAPL,8500",			"PowerMac 8500/8600",
+		PMAC_TYPE_PSURGE,		NULL,
+		0
+	},
+	{	"AAPL,9500",			"PowerMac 9500/9600",
+		PMAC_TYPE_PSURGE,		NULL,
+		0
+	},
+	{	"AAPL,7200",			"PowerMac 7200",
+		PMAC_TYPE_PSURGE,		NULL,
+		0
+	},
+	{	"AAPL,7300",			"PowerMac 7200/7300",
+		PMAC_TYPE_PSURGE,		NULL,
+		0
+	},
+	{	"AAPL,7500",			"PowerMac 7500",
+		PMAC_TYPE_PSURGE,		NULL,
+		0
+	},
+	{	"AAPL,ShinerESB",		"Apple Network Server",
+		PMAC_TYPE_ANS,			NULL,
+		0
+	},
+	{	"AAPL,e407",			"Alchemy",
+		PMAC_TYPE_ALCHEMY,		NULL,
+		0
+	},
+	{	"AAPL,e411",			"Gazelle",
+		PMAC_TYPE_GAZELLE,		NULL,
+		0
+	},
+	{	"AAPL,Gossamer",		"PowerMac G3 (Gossamer)",
+		PMAC_TYPE_GOSSAMER,		heathrow_desktop_features,
+		0
+	},
+	{	"AAPL,PowerMac G3",		"PowerMac G3 (Silk)",
+		PMAC_TYPE_SILK,			heathrow_desktop_features,
+		0
+	},
+	{	"PowerMac1,1",			"Blue&White G3",
+		PMAC_TYPE_YOSEMITE,		paddington_features,
+		0
+	},
+	{	"PowerMac1,2",			"PowerMac G4 PCI Graphics",
+		PMAC_TYPE_YIKES,		paddington_features,
+		0
+	},
+	{	"PowerMac2,1",			"iMac FireWire",
+		PMAC_TYPE_FW_IMAC,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_OLD_CORE99
+	},
+	{	"PowerMac2,2",			"iMac FireWire",
+		PMAC_TYPE_FW_IMAC,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_OLD_CORE99
+	},
+	{	"PowerMac3,1",			"PowerMac G4 AGP Graphics",
+		PMAC_TYPE_SAWTOOTH,		core99_features,
+		PMAC_MB_OLD_CORE99
+	},
+	{	"PowerMac3,2",			"PowerMac G4 AGP Graphics",
+		PMAC_TYPE_SAWTOOTH,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_OLD_CORE99
+	},
+	{	"PowerMac3,3",			"PowerMac G4 AGP Graphics",
+		PMAC_TYPE_SAWTOOTH,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_OLD_CORE99
+	},
+	{	"PowerMac3,4",			"PowerMac G4 Silver",
+		PMAC_TYPE_QUICKSILVER,		core99_features,
+		PMAC_MB_MAY_SLEEP
+	},
+	{	"PowerMac3,5",			"PowerMac G4 Silver",
+		PMAC_TYPE_QUICKSILVER,		core99_features,
+		PMAC_MB_MAY_SLEEP
+	},
+	{	"PowerMac3,6",			"PowerMac G4 Windtunnel",
+		PMAC_TYPE_WINDTUNNEL,		core99_features,
+		PMAC_MB_MAY_SLEEP,
+	},
+	{	"PowerMac4,1",			"iMac \"Flower Power\"",
+		PMAC_TYPE_PANGEA_IMAC,		pangea_features,
+		PMAC_MB_MAY_SLEEP
+	},
+	{	"PowerMac4,2",			"Flat panel iMac",
+		PMAC_TYPE_FLAT_PANEL_IMAC,	pangea_features,
+		PMAC_MB_CAN_SLEEP
+	},
+	{	"PowerMac4,4",			"eMac",
+		PMAC_TYPE_EMAC,			core99_features,
+		PMAC_MB_MAY_SLEEP
+	},
+	{	"PowerMac5,1",			"PowerMac G4 Cube",
+		PMAC_TYPE_CUBE,			core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_OLD_CORE99
+	},
+	{	"PowerMac6,1",			"Flat panel iMac",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP,
+	},
+	{	"PowerMac6,3",			"Flat panel iMac",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP,
+	},
+	{	"PowerMac6,4",			"eMac",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP,
+	},
+	{	"PowerMac10,1",			"Mac mini",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER,
+	},
+	{	"iMac,1",			"iMac (first generation)",
+		PMAC_TYPE_ORIG_IMAC,		paddington_features,
+		0
+	},
+
+	/*
+	 * Xserve's
+	 */
+
+	{	"RackMac1,1",			"XServe",
+		PMAC_TYPE_RACKMAC,		rackmac_features,
+		0,
+	},
+	{	"RackMac1,2",			"XServe rev. 2",
+		PMAC_TYPE_RACKMAC,		rackmac_features,
+		0,
+	},
+
+	/*
+	 * Laptops
+	 */
+
+	{	"AAPL,3400/2400",		"PowerBook 3400",
+		PMAC_TYPE_HOOPER,		ohare_features,
+		PMAC_MB_CAN_SLEEP | PMAC_MB_MOBILE
+	},
+	{	"AAPL,3500",			"PowerBook 3500",
+		PMAC_TYPE_KANGA,		ohare_features,
+		PMAC_MB_CAN_SLEEP | PMAC_MB_MOBILE
+	},
+	{	"AAPL,PowerBook1998",		"PowerBook Wallstreet",
+		PMAC_TYPE_WALLSTREET,		heathrow_laptop_features,
+		PMAC_MB_CAN_SLEEP | PMAC_MB_MOBILE
+	},
+	{	"PowerBook1,1",			"PowerBook 101 (Lombard)",
+		PMAC_TYPE_101_PBOOK,		paddington_features,
+		PMAC_MB_CAN_SLEEP | PMAC_MB_MOBILE
+	},
+	{	"PowerBook2,1",			"iBook (first generation)",
+		PMAC_TYPE_ORIG_IBOOK,		core99_features,
+		PMAC_MB_CAN_SLEEP | PMAC_MB_OLD_CORE99 | PMAC_MB_MOBILE
+	},
+	{	"PowerBook2,2",			"iBook FireWire",
+		PMAC_TYPE_FW_IBOOK,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER |
+		PMAC_MB_OLD_CORE99 | PMAC_MB_MOBILE
+	},
+	{	"PowerBook3,1",			"PowerBook Pismo",
+		PMAC_TYPE_PISMO,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER |
+		PMAC_MB_OLD_CORE99 | PMAC_MB_MOBILE
+	},
+	{	"PowerBook3,2",			"PowerBook Titanium",
+		PMAC_TYPE_TITANIUM,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE
+	},
+	{	"PowerBook3,3",			"PowerBook Titanium II",
+		PMAC_TYPE_TITANIUM2,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE
+	},
+	{	"PowerBook3,4",			"PowerBook Titanium III",
+		PMAC_TYPE_TITANIUM3,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE
+	},
+	{	"PowerBook3,5",			"PowerBook Titanium IV",
+		PMAC_TYPE_TITANIUM4,		core99_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE
+	},
+	{	"PowerBook4,1",			"iBook 2",
+		PMAC_TYPE_IBOOK2,		pangea_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE
+	},
+	{	"PowerBook4,2",			"iBook 2",
+		PMAC_TYPE_IBOOK2,		pangea_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE
+	},
+	{	"PowerBook4,3",			"iBook 2 rev. 2",
+		PMAC_TYPE_IBOOK2,		pangea_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE
+	},
+	{	"PowerBook5,1",			"PowerBook G4 17\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook5,2",			"PowerBook G4 15\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook5,3",			"PowerBook G4 17\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook5,4",			"PowerBook G4 15\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook5,5",			"PowerBook G4 17\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook5,6",			"PowerBook G4 15\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook5,7",			"PowerBook G4 17\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook6,1",			"PowerBook G4 12\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook6,2",			"PowerBook G4",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook6,3",			"iBook G4",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook6,4",			"PowerBook G4 12\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook6,5",			"iBook G4",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook6,7",			"iBook G4",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+	{	"PowerBook6,8",			"PowerBook G4 12\"",
+		PMAC_TYPE_UNKNOWN_INTREPID,	intrepid_features,
+		PMAC_MB_MAY_SLEEP | PMAC_MB_HAS_FW_POWER | PMAC_MB_MOBILE,
+	},
+#else /* CONFIG_POWER4 */
+	{	"PowerMac7,2",			"PowerMac G5",
+		PMAC_TYPE_POWERMAC_G5,		g5_features,
+		0,
+	},
+#ifdef CONFIG_PPC64
+	{	"PowerMac7,3",			"PowerMac G5",
+		PMAC_TYPE_POWERMAC_G5,		g5_features,
+		0,
+	},
+	{	"PowerMac8,1",			"iMac G5",
+		PMAC_TYPE_IMAC_G5,		g5_features,
+		0,
+	},
+	{	"PowerMac9,1",			"PowerMac G5",
+		PMAC_TYPE_POWERMAC_G5_U3L,	g5_features,
+		0,
+	},
+	{       "RackMac3,1",                   "XServe G5",
+		PMAC_TYPE_XSERVE_G5,		g5_features,
+		0,
+	},
+#endif /* CONFIG_PPC64 */
+#endif /* CONFIG_POWER4 */
+};
+
+/*
+ * The toplevel feature_call callback
+ */
+long pmac_do_feature_call(unsigned int selector, ...)
+{
+	struct device_node *node;
+	long param, value;
+	int i;
+	feature_call func = NULL;
+	va_list args;
+
+	if (pmac_mb.features)
+		for (i=0; pmac_mb.features[i].function; i++)
+			if (pmac_mb.features[i].selector == selector) {
+				func = pmac_mb.features[i].function;
+				break;
+			}
+	if (!func)
+		for (i=0; any_features[i].function; i++)
+			if (any_features[i].selector == selector) {
+				func = any_features[i].function;
+				break;
+			}
+	if (!func)
+		return -ENODEV;
+
+	va_start(args, selector);
+	node = (struct device_node*)va_arg(args, void*);
+	param = va_arg(args, long);
+	value = va_arg(args, long);
+	va_end(args);
+
+	return func(node, param, value);
+}
+
+static int __init probe_motherboard(void)
+{
+	int i;
+	struct macio_chip *macio = &macio_chips[0];
+	const char *model = NULL;
+	struct device_node *dt;
+
+	/* Lookup known motherboard type in device-tree. First try an
+	 * exact match on the "model" property, then try a "compatible"
+	 * match is none is found.
+	 */
+	dt = find_devices("device-tree");
+	if (dt != NULL)
+		model = (const char *) get_property(dt, "model", NULL);
+	for(i=0; model && i<(sizeof(pmac_mb_defs)/sizeof(struct pmac_mb_def)); i++) {
+	    if (strcmp(model, pmac_mb_defs[i].model_string) == 0) {
+		pmac_mb = pmac_mb_defs[i];
+		goto found;
+	    }
+	}
+	for(i=0; i<(sizeof(pmac_mb_defs)/sizeof(struct pmac_mb_def)); i++) {
+	    if (machine_is_compatible(pmac_mb_defs[i].model_string)) {
+		pmac_mb = pmac_mb_defs[i];
+		goto found;
+	    }
+	}
+
+	/* Fallback to selection depending on mac-io chip type */
+	switch(macio->type) {
+#ifndef CONFIG_POWER4
+	    case macio_grand_central:
+		pmac_mb.model_id = PMAC_TYPE_PSURGE;
+		pmac_mb.model_name = "Unknown PowerSurge";
+		break;
+	    case macio_ohare:
+		pmac_mb.model_id = PMAC_TYPE_UNKNOWN_OHARE;
+		pmac_mb.model_name = "Unknown OHare-based";
+		break;
+	    case macio_heathrow:
+		pmac_mb.model_id = PMAC_TYPE_UNKNOWN_HEATHROW;
+		pmac_mb.model_name = "Unknown Heathrow-based";
+		pmac_mb.features = heathrow_desktop_features;
+		break;
+	    case macio_paddington:
+		pmac_mb.model_id = PMAC_TYPE_UNKNOWN_PADDINGTON;
+		pmac_mb.model_name = "Unknown Paddington-based";
+		pmac_mb.features = paddington_features;
+		break;
+	    case macio_keylargo:
+		pmac_mb.model_id = PMAC_TYPE_UNKNOWN_CORE99;
+		pmac_mb.model_name = "Unknown Keylargo-based";
+		pmac_mb.features = core99_features;
+		break;
+	    case macio_pangea:
+		pmac_mb.model_id = PMAC_TYPE_UNKNOWN_PANGEA;
+		pmac_mb.model_name = "Unknown Pangea-based";
+		pmac_mb.features = pangea_features;
+		break;
+	    case macio_intrepid:
+		pmac_mb.model_id = PMAC_TYPE_UNKNOWN_INTREPID;
+		pmac_mb.model_name = "Unknown Intrepid-based";
+		pmac_mb.features = intrepid_features;
+		break;
+#else /* CONFIG_POWER4 */
+	case macio_keylargo2:
+		pmac_mb.model_id = PMAC_TYPE_UNKNOWN_K2;
+		pmac_mb.model_name = "Unknown K2-based";
+		pmac_mb.features = g5_features;
+		break;
+#endif /* CONFIG_POWER4 */
+	default:
+		return -ENODEV;
+	}
+found:
+#ifndef CONFIG_POWER4
+	/* Fixup Hooper vs. Comet */
+	if (pmac_mb.model_id == PMAC_TYPE_HOOPER) {
+		u32 __iomem * mach_id_ptr = ioremap(0xf3000034, 4);
+		if (!mach_id_ptr)
+			return -ENODEV;
+		/* Here, I used to disable the media-bay on comet. It
+		 * appears this is wrong, the floppy connector is actually
+		 * a kind of media-bay and works with the current driver.
+		 */
+		if (__raw_readl(mach_id_ptr) & 0x20000000UL)
+			pmac_mb.model_id = PMAC_TYPE_COMET;
+		iounmap(mach_id_ptr);
+	}
+#endif /* CONFIG_POWER4 */
+
+#ifdef CONFIG_6xx
+	/* Set default value of powersave_nap on machines that support it.
+	 * It appears that uninorth rev 3 has a problem with it, we don't
+	 * enable it on those. In theory, the flush-on-lock property is
+	 * supposed to be set when not supported, but I'm not very confident
+	 * that all Apple OF revs did it properly, I do it the paranoid way.
+	 */
+	while (uninorth_base && uninorth_rev > 3) {
+		struct device_node *np = find_path_device("/cpus");
+		if (!np || !np->child) {
+			printk(KERN_WARNING "Can't find CPU(s) in device tree !\n");
+			break;
+		}
+		np = np->child;
+		/* Nap mode not supported on SMP */
+		if (np->sibling)
+			break;
+		/* Nap mode not supported if flush-on-lock property is present */
+		if (get_property(np, "flush-on-lock", NULL))
+			break;
+		powersave_nap = 1;
+		printk(KERN_INFO "Processor NAP mode on idle enabled.\n");
+		break;
+	}
+
+	/* On CPUs that support it (750FX), lowspeed by default during
+	 * NAP mode
+	 */
+	powersave_lowspeed = 1;
+#endif /* CONFIG_6xx */
+#ifdef CONFIG_POWER4
+	powersave_nap = 1;
+#endif
+	/* Check for "mobile" machine */
+	if (model && (strncmp(model, "PowerBook", 9) == 0
+		   || strncmp(model, "iBook", 5) == 0))
+		pmac_mb.board_flags |= PMAC_MB_MOBILE;
+
+
+	printk(KERN_INFO "PowerMac motherboard: %s\n", pmac_mb.model_name);
+	return 0;
+}
+
+/* Initialize the Core99 UniNorth host bridge and memory controller
+ */
+static void __init probe_uninorth(void)
+{
+	unsigned long actrl;
+
+	/* Locate core99 Uni-N */
+	uninorth_node = of_find_node_by_name(NULL, "uni-n");
+	/* Locate G5 u3 */
+	if (uninorth_node == NULL) {
+		uninorth_node = of_find_node_by_name(NULL, "u3");
+		uninorth_u3 = 1;
+	}
+	if (uninorth_node && uninorth_node->n_addrs > 0) {
+		unsigned long address = uninorth_node->addrs[0].address;
+		uninorth_base = ioremap(address, 0x40000);
+		uninorth_rev = in_be32(UN_REG(UNI_N_VERSION));
+		if (uninorth_u3)
+			u3_ht = ioremap(address + U3_HT_CONFIG_BASE, 0x1000);
+	} else
+		uninorth_node = NULL;
+
+	if (!uninorth_node)
+		return;
+
+	printk(KERN_INFO "Found %s memory controller & host bridge, revision: %d\n",
+	       uninorth_u3 ? "U3" : "UniNorth", uninorth_rev);
+	printk(KERN_INFO "Mapped at 0x%08lx\n", (unsigned long)uninorth_base);
+
+	/* Set the arbitrer QAck delay according to what Apple does
+	 */
+	if (uninorth_rev < 0x11) {
+		actrl = UN_IN(UNI_N_ARB_CTRL) & ~UNI_N_ARB_CTRL_QACK_DELAY_MASK;
+		actrl |= ((uninorth_rev < 3) ? UNI_N_ARB_CTRL_QACK_DELAY105 :
+			UNI_N_ARB_CTRL_QACK_DELAY) << UNI_N_ARB_CTRL_QACK_DELAY_SHIFT;
+		UN_OUT(UNI_N_ARB_CTRL, actrl);
+	}
+
+	/* Some more magic as done by them in recent MacOS X on UniNorth
+	 * revs 1.5 to 2.O and Pangea. Seem to toggle the UniN Maxbus/PCI
+	 * memory timeout
+	 */
+	if ((uninorth_rev >= 0x11 && uninorth_rev <= 0x24) || uninorth_rev == 0xc0)
+		UN_OUT(0x2160, UN_IN(0x2160) & 0x00ffffff);
+}
+
+static void __init probe_one_macio(const char *name, const char *compat, int type)
+{
+	struct device_node*	node;
+	int			i;
+	volatile u32 __iomem *	base;
+	u32*			revp;
+
+	node = find_devices(name);
+	if (!node || !node->n_addrs)
+		return;
+	if (compat)
+		do {
+			if (device_is_compatible(node, compat))
+				break;
+			node = node->next;
+		} while (node);
+	if (!node)
+		return;
+	for(i=0; i<MAX_MACIO_CHIPS; i++) {
+		if (!macio_chips[i].of_node)
+			break;
+		if (macio_chips[i].of_node == node)
+			return;
+	}
+	if (i >= MAX_MACIO_CHIPS) {
+		printk(KERN_ERR "pmac_feature: Please increase MAX_MACIO_CHIPS !\n");
+		printk(KERN_ERR "pmac_feature: %s skipped\n", node->full_name);
+		return;
+	}
+	base = ioremap(node->addrs[0].address, node->addrs[0].size);
+	if (!base) {
+		printk(KERN_ERR "pmac_feature: Can't map mac-io chip !\n");
+		return;
+	}
+	if (type == macio_keylargo) {
+		u32 *did = (u32 *)get_property(node, "device-id", NULL);
+		if (*did == 0x00000025)
+			type = macio_pangea;
+		if (*did == 0x0000003e)
+			type = macio_intrepid;
+	}
+	macio_chips[i].of_node	= node;
+	macio_chips[i].type	= type;
+	macio_chips[i].base	= base;
+	macio_chips[i].flags	= MACIO_FLAG_SCCB_ON | MACIO_FLAG_SCCB_ON;
+	macio_chips[i].name	= macio_names[type];
+	revp = (u32 *)get_property(node, "revision-id", NULL);
+	if (revp)
+		macio_chips[i].rev = *revp;
+	printk(KERN_INFO "Found a %s mac-io controller, rev: %d, mapped at 0x%p\n",
+		macio_names[type], macio_chips[i].rev, macio_chips[i].base);
+}
+
+static int __init
+probe_macios(void)
+{
+	/* Warning, ordering is important */
+	probe_one_macio("gc", NULL, macio_grand_central);
+	probe_one_macio("ohare", NULL, macio_ohare);
+	probe_one_macio("pci106b,7", NULL, macio_ohareII);
+	probe_one_macio("mac-io", "keylargo", macio_keylargo);
+	probe_one_macio("mac-io", "paddington", macio_paddington);
+	probe_one_macio("mac-io", "gatwick", macio_gatwick);
+	probe_one_macio("mac-io", "heathrow", macio_heathrow);
+	probe_one_macio("mac-io", "K2-Keylargo", macio_keylargo2);
+
+	/* Make sure the "main" macio chip appear first */
+	if (macio_chips[0].type == macio_gatwick
+	    && macio_chips[1].type == macio_heathrow) {
+		struct macio_chip temp = macio_chips[0];
+		macio_chips[0] = macio_chips[1];
+		macio_chips[1] = temp;
+	}
+	if (macio_chips[0].type == macio_ohareII
+	    && macio_chips[1].type == macio_ohare) {
+		struct macio_chip temp = macio_chips[0];
+		macio_chips[0] = macio_chips[1];
+		macio_chips[1] = temp;
+	}
+	macio_chips[0].lbus.index = 0;
+	macio_chips[1].lbus.index = 1;
+
+	return (macio_chips[0].of_node == NULL) ? -ENODEV : 0;
+}
+
+static void __init
+initial_serial_shutdown(struct device_node *np)
+{
+	int len;
+	struct slot_names_prop {
+		int	count;
+		char	name[1];
+	} *slots;
+	char *conn;
+	int port_type = PMAC_SCC_ASYNC;
+	int modem = 0;
+
+	slots = (struct slot_names_prop *)get_property(np, "slot-names", &len);
+	conn = get_property(np, "AAPL,connector", &len);
+	if (conn && (strcmp(conn, "infrared") == 0))
+		port_type = PMAC_SCC_IRDA;
+	else if (device_is_compatible(np, "cobalt"))
+		modem = 1;
+	else if (slots && slots->count > 0) {
+		if (strcmp(slots->name, "IrDA") == 0)
+			port_type = PMAC_SCC_IRDA;
+		else if (strcmp(slots->name, "Modem") == 0)
+			modem = 1;
+	}
+	if (modem)
+		pmac_call_feature(PMAC_FTR_MODEM_ENABLE, np, 0, 0);
+	pmac_call_feature(PMAC_FTR_SCC_ENABLE, np, port_type, 0);
+}
+
+static void __init
+set_initial_features(void)
+{
+	struct device_node *np;
+
+	/* That hack appears to be necessary for some StarMax motherboards
+	 * but I'm not too sure it was audited for side-effects on other
+	 * ohare based machines...
+	 * Since I still have difficulties figuring the right way to
+	 * differenciate them all and since that hack was there for a long
+	 * time, I'll keep it around
+	 */
+	if (macio_chips[0].type == macio_ohare && !find_devices("via-pmu")) {
+		struct macio_chip *macio = &macio_chips[0];
+		MACIO_OUT32(OHARE_FCR, STARMAX_FEATURES);
+	} else if (macio_chips[0].type == macio_ohare) {
+		struct macio_chip *macio = &macio_chips[0];
+		MACIO_BIS(OHARE_FCR, OH_IOBUS_ENABLE);
+	} else if (macio_chips[1].type == macio_ohare) {
+		struct macio_chip *macio = &macio_chips[1];
+		MACIO_BIS(OHARE_FCR, OH_IOBUS_ENABLE);
+	}
+
+#ifdef CONFIG_POWER4
+	if (macio_chips[0].type == macio_keylargo2) {
+#ifndef CONFIG_SMP
+		/* On SMP machines running UP, we have the second CPU eating
+		 * bus cycles. We need to take it off the bus. This is done
+		 * from pmac_smp for SMP kernels running on one CPU
+		 */
+		np = of_find_node_by_type(NULL, "cpu");
+		if (np != NULL)
+			np = of_find_node_by_type(np, "cpu");
+		if (np != NULL) {
+			g5_phy_disable_cpu1();
+			of_node_put(np);
+		}
+#endif /* CONFIG_SMP */
+		/* Enable GMAC for now for PCI probing. It will be disabled
+		 * later on after PCI probe
+		 */
+		np = of_find_node_by_name(NULL, "ethernet");
+		while(np) {
+			if (device_is_compatible(np, "K2-GMAC"))
+				g5_gmac_enable(np, 0, 1);
+			np = of_find_node_by_name(np, "ethernet");
+		}
+
+		/* Enable FW before PCI probe. Will be disabled later on
+		 * Note: We should have a batter way to check that we are
+		 * dealing with uninorth internal cell and not a PCI cell
+		 * on the external PCI. The code below works though.
+		 */
+		np = of_find_node_by_name(NULL, "firewire");
+		while(np) {
+			if (device_is_compatible(np, "pci106b,5811")) {
+				macio_chips[0].flags |= MACIO_FLAG_FW_SUPPORTED;
+				g5_fw_enable(np, 0, 1);
+			}
+			np = of_find_node_by_name(np, "firewire");
+		}
+	}
+#else /* CONFIG_POWER4 */
+
+	if (macio_chips[0].type == macio_keylargo ||
+	    macio_chips[0].type == macio_pangea ||
+	    macio_chips[0].type == macio_intrepid) {
+		/* Enable GMAC for now for PCI probing. It will be disabled
+		 * later on after PCI probe
+		 */
+		np = of_find_node_by_name(NULL, "ethernet");
+		while(np) {
+			if (np->parent
+			    && device_is_compatible(np->parent, "uni-north")
+			    && device_is_compatible(np, "gmac"))
+				core99_gmac_enable(np, 0, 1);
+			np = of_find_node_by_name(np, "ethernet");
+		}
+
+		/* Enable FW before PCI probe. Will be disabled later on
+		 * Note: We should have a batter way to check that we are
+		 * dealing with uninorth internal cell and not a PCI cell
+		 * on the external PCI. The code below works though.
+		 */
+		np = of_find_node_by_name(NULL, "firewire");
+		while(np) {
+			if (np->parent
+			    && device_is_compatible(np->parent, "uni-north")
+			    && (device_is_compatible(np, "pci106b,18") ||
+			        device_is_compatible(np, "pci106b,30") ||
+			        device_is_compatible(np, "pci11c1,5811"))) {
+				macio_chips[0].flags |= MACIO_FLAG_FW_SUPPORTED;
+				core99_firewire_enable(np, 0, 1);
+			}
+			np = of_find_node_by_name(np, "firewire");
+		}
+
+		/* Enable ATA-100 before PCI probe. */
+		np = of_find_node_by_name(NULL, "ata-6");
+		while(np) {
+			if (np->parent
+			    && device_is_compatible(np->parent, "uni-north")
+			    && device_is_compatible(np, "kauai-ata")) {
+				core99_ata100_enable(np, 1);
+			}
+			np = of_find_node_by_name(np, "ata-6");
+		}
+
+		/* Switch airport off */
+		np = find_devices("radio");
+		while(np) {
+			if (np && np->parent == macio_chips[0].of_node) {
+				macio_chips[0].flags |= MACIO_FLAG_AIRPORT_ON;
+				core99_airport_enable(np, 0, 0);
+			}
+			np = np->next;
+		}
+	}
+
+	/* On all machines that support sound PM, switch sound off */
+	if (macio_chips[0].of_node)
+		pmac_do_feature_call(PMAC_FTR_SOUND_CHIP_ENABLE,
+			macio_chips[0].of_node, 0, 0);
+
+	/* While on some desktop G3s, we turn it back on */
+	if (macio_chips[0].of_node && macio_chips[0].type == macio_heathrow
+		&& (pmac_mb.model_id == PMAC_TYPE_GOSSAMER ||
+		    pmac_mb.model_id == PMAC_TYPE_SILK)) {
+		struct macio_chip *macio = &macio_chips[0];
+		MACIO_BIS(HEATHROW_FCR, HRW_SOUND_CLK_ENABLE);
+		MACIO_BIC(HEATHROW_FCR, HRW_SOUND_POWER_N);
+	}
+
+	/* Some machine models need the clock chip to be properly setup for
+	 * clock spreading now. This should be a platform function but we
+	 * don't do these at the moment
+	 */
+	pmac_tweak_clock_spreading(1);
+
+#endif /* CONFIG_POWER4 */
+
+	/* On all machines, switch modem & serial ports off */
+	np = find_devices("ch-a");
+	while(np) {
+		initial_serial_shutdown(np);
+		np = np->next;
+	}
+	np = find_devices("ch-b");
+	while(np) {
+		initial_serial_shutdown(np);
+		np = np->next;
+	}
+}
+
+void __init
+pmac_feature_init(void)
+{
+	/* Detect the UniNorth memory controller */
+	probe_uninorth();
+
+	/* Probe mac-io controllers */
+	if (probe_macios()) {
+		printk(KERN_WARNING "No mac-io chip found\n");
+		return;
+	}
+
+	/* Setup low-level i2c stuffs */
+	pmac_init_low_i2c();
+
+	/* Probe machine type */
+	if (probe_motherboard())
+		printk(KERN_WARNING "Unknown PowerMac !\n");
+
+	/* Set some initial features (turn off some chips that will
+	 * be later turned on)
+	 */
+	set_initial_features();
+}
+
+int __init pmac_feature_late_init(void)
+{
+#if 0
+	struct device_node *np;
+
+	/* Request some resources late */
+	if (uninorth_node)
+		request_OF_resource(uninorth_node, 0, NULL);
+	np = find_devices("hammerhead");
+	if (np)
+		request_OF_resource(np, 0, NULL);
+	np = find_devices("interrupt-controller");
+	if (np)
+		request_OF_resource(np, 0, NULL);
+#endif
+	return 0;
+}
+
+device_initcall(pmac_feature_late_init);
+
+#if 0
+static void dump_HT_speeds(char *name, u32 cfg, u32 frq)
+{
+	int	freqs[16] = { 200,300,400,500,600,800,1000,0,0,0,0,0,0,0,0,0 };
+	int	bits[8] = { 8,16,0,32,2,4,0,0 };
+	int	freq = (frq >> 8) & 0xf;
+
+	if (freqs[freq] == 0)
+		printk("%s: Unknown HT link frequency %x\n", name, freq);
+	else
+		printk("%s: %d MHz on main link, (%d in / %d out) bits width\n",
+		       name, freqs[freq],
+		       bits[(cfg >> 28) & 0x7], bits[(cfg >> 24) & 0x7]);
+}
+
+void __init pmac_check_ht_link(void)
+{
+	u32	ufreq, freq, ucfg, cfg;
+	struct device_node *pcix_node;
+	u8	px_bus, px_devfn;
+	struct pci_controller *px_hose;
+
+	(void)in_be32(u3_ht + U3_HT_LINK_COMMAND);
+	ucfg = cfg = in_be32(u3_ht + U3_HT_LINK_CONFIG);
+	ufreq = freq = in_be32(u3_ht + U3_HT_LINK_FREQ);
+	dump_HT_speeds("U3 HyperTransport", cfg, freq);
+
+	pcix_node = of_find_compatible_node(NULL, "pci", "pci-x");
+	if (pcix_node == NULL) {
+		printk("No PCI-X bridge found\n");
+		return;
+	}
+	if (pci_device_from_OF_node(pcix_node, &px_bus, &px_devfn) != 0) {
+		printk("PCI-X bridge found but not matched to pci\n");
+		return;
+	}
+	px_hose = pci_find_hose_for_OF_device(pcix_node);
+	if (px_hose == NULL) {
+		printk("PCI-X bridge found but not matched to host\n");
+		return;
+	}	
+	early_read_config_dword(px_hose, px_bus, px_devfn, 0xc4, &cfg);
+	early_read_config_dword(px_hose, px_bus, px_devfn, 0xcc, &freq);
+	dump_HT_speeds("PCI-X HT Uplink", cfg, freq);
+	early_read_config_dword(px_hose, px_bus, px_devfn, 0xc8, &cfg);
+	early_read_config_dword(px_hose, px_bus, px_devfn, 0xd0, &freq);
+	dump_HT_speeds("PCI-X HT Downlink", cfg, freq);
+}
+#endif /* 0 */
+
+/*
+ * Early video resume hook
+ */
+
+static void (*pmac_early_vresume_proc)(void *data);
+static void *pmac_early_vresume_data;
+
+void pmac_set_early_video_resume(void (*proc)(void *data), void *data)
+{
+	if (_machine != _MACH_Pmac)
+		return;
+	preempt_disable();
+	pmac_early_vresume_proc = proc;
+	pmac_early_vresume_data = data;
+	preempt_enable();
+}
+EXPORT_SYMBOL(pmac_set_early_video_resume);
+
+void pmac_call_early_video_resume(void)
+{
+	if (pmac_early_vresume_proc)
+		pmac_early_vresume_proc(pmac_early_vresume_data);
+}
+
+/*
+ * AGP related suspend/resume code
+ */
+
+static struct pci_dev *pmac_agp_bridge;
+static int (*pmac_agp_suspend)(struct pci_dev *bridge);
+static int (*pmac_agp_resume)(struct pci_dev *bridge);
+
+void pmac_register_agp_pm(struct pci_dev *bridge,
+				 int (*suspend)(struct pci_dev *bridge),
+				 int (*resume)(struct pci_dev *bridge))
+{
+	if (suspend || resume) {
+		pmac_agp_bridge = bridge;
+		pmac_agp_suspend = suspend;
+		pmac_agp_resume = resume;
+		return;
+	}
+	if (bridge != pmac_agp_bridge)
+		return;
+	pmac_agp_suspend = pmac_agp_resume = NULL;
+	return;
+}
+EXPORT_SYMBOL(pmac_register_agp_pm);
+
+void pmac_suspend_agp_for_card(struct pci_dev *dev)
+{
+	if (pmac_agp_bridge == NULL || pmac_agp_suspend == NULL)
+		return;
+	if (pmac_agp_bridge->bus != dev->bus)
+		return;
+	pmac_agp_suspend(pmac_agp_bridge);
+}
+EXPORT_SYMBOL(pmac_suspend_agp_for_card);
+
+void pmac_resume_agp_for_card(struct pci_dev *dev)
+{
+	if (pmac_agp_bridge == NULL || pmac_agp_resume == NULL)
+		return;
+	if (pmac_agp_bridge->bus != dev->bus)
+		return;
+	pmac_agp_resume(pmac_agp_bridge);
+}
+EXPORT_SYMBOL(pmac_resume_agp_for_card);
diff --git a/arch/powerpc/platforms/powermac/low_i2c.c b/arch/powerpc/platforms/powermac/low_i2c.c
new file mode 100644
index 000000000000..f3f39e8e337a
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/low_i2c.c
@@ -0,0 +1,523 @@
+/*
+ *  arch/ppc/platforms/pmac_low_i2c.c
+ *
+ *  Copyright (C) 2003 Ben. Herrenschmidt (benh@kernel.crashing.org)
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ *  This file contains some low-level i2c access routines that
+ *  need to be used by various bits of the PowerMac platform code
+ *  at times where the real asynchronous & interrupt driven driver
+ *  cannot be used. The API borrows some semantics from the darwin
+ *  driver in order to ease the implementation of the platform
+ *  properties parser
+ */
+
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/types.h>
+#include <linux/sched.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/adb.h>
+#include <linux/pmu.h>
+#include <asm/keylargo.h>
+#include <asm/uninorth.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+#include <asm/pmac_low_i2c.h>
+
+#define MAX_LOW_I2C_HOST	4
+
+#ifdef DEBUG
+#define DBG(x...) do {\
+		printk(KERN_DEBUG "KW:" x);	\
+	} while(0)
+#else
+#define DBG(x...)
+#endif
+
+struct low_i2c_host;
+
+typedef int (*low_i2c_func_t)(struct low_i2c_host *host, u8 addr, u8 sub, u8 *data, int len);
+
+struct low_i2c_host
+{
+	struct device_node	*np;		/* OF device node */
+	struct semaphore	mutex;		/* Access mutex for use by i2c-keywest */
+	low_i2c_func_t		func;		/* Access function */
+	unsigned int		is_open : 1;	/* Poor man's access control */
+	int			mode;		/* Current mode */
+	int			channel;	/* Current channel */
+	int			num_channels;	/* Number of channels */
+	void __iomem		*base;		/* For keywest-i2c, base address */
+	int			bsteps;		/* And register stepping */
+	int			speed;		/* And speed */
+};
+
+static struct low_i2c_host	low_i2c_hosts[MAX_LOW_I2C_HOST];
+
+/* No locking is necessary on allocation, we are running way before
+ * anything can race with us
+ */
+static struct low_i2c_host *find_low_i2c_host(struct device_node *np)
+{
+	int i;
+
+	for (i = 0; i < MAX_LOW_I2C_HOST; i++)
+		if (low_i2c_hosts[i].np == np)
+			return &low_i2c_hosts[i];
+	return NULL;
+}
+
+/*
+ *
+ * i2c-keywest implementation (UniNorth, U2, U3, Keylargo's)
+ *
+ */
+
+/*
+ * Keywest i2c definitions borrowed from drivers/i2c/i2c-keywest.h,
+ * should be moved somewhere in include/asm-ppc/
+ */
+/* Register indices */
+typedef enum {
+	reg_mode = 0,
+	reg_control,
+	reg_status,
+	reg_isr,
+	reg_ier,
+	reg_addr,
+	reg_subaddr,
+	reg_data
+} reg_t;
+
+
+/* Mode register */
+#define KW_I2C_MODE_100KHZ	0x00
+#define KW_I2C_MODE_50KHZ	0x01
+#define KW_I2C_MODE_25KHZ	0x02
+#define KW_I2C_MODE_DUMB	0x00
+#define KW_I2C_MODE_STANDARD	0x04
+#define KW_I2C_MODE_STANDARDSUB	0x08
+#define KW_I2C_MODE_COMBINED	0x0C
+#define KW_I2C_MODE_MODE_MASK	0x0C
+#define KW_I2C_MODE_CHAN_MASK	0xF0
+
+/* Control register */
+#define KW_I2C_CTL_AAK		0x01
+#define KW_I2C_CTL_XADDR	0x02
+#define KW_I2C_CTL_STOP		0x04
+#define KW_I2C_CTL_START	0x08
+
+/* Status register */
+#define KW_I2C_STAT_BUSY	0x01
+#define KW_I2C_STAT_LAST_AAK	0x02
+#define KW_I2C_STAT_LAST_RW	0x04
+#define KW_I2C_STAT_SDA		0x08
+#define KW_I2C_STAT_SCL		0x10
+
+/* IER & ISR registers */
+#define KW_I2C_IRQ_DATA		0x01
+#define KW_I2C_IRQ_ADDR		0x02
+#define KW_I2C_IRQ_STOP		0x04
+#define KW_I2C_IRQ_START	0x08
+#define KW_I2C_IRQ_MASK		0x0F
+
+/* State machine states */
+enum {
+	state_idle,
+	state_addr,
+	state_read,
+	state_write,
+	state_stop,
+	state_dead
+};
+
+#define WRONG_STATE(name) do {\
+		printk(KERN_DEBUG "KW: wrong state. Got %s, state: %s (isr: %02x)\n", \
+		       name, __kw_state_names[state], isr); \
+	} while(0)
+
+static const char *__kw_state_names[] = {
+	"state_idle",
+	"state_addr",
+	"state_read",
+	"state_write",
+	"state_stop",
+	"state_dead"
+};
+
+static inline u8 __kw_read_reg(struct low_i2c_host *host, reg_t reg)
+{
+	return readb(host->base + (((unsigned int)reg) << host->bsteps));
+}
+
+static inline void __kw_write_reg(struct low_i2c_host *host, reg_t reg, u8 val)
+{
+	writeb(val, host->base + (((unsigned)reg) << host->bsteps));
+	(void)__kw_read_reg(host, reg_subaddr);
+}
+
+#define kw_write_reg(reg, val)	__kw_write_reg(host, reg, val) 
+#define kw_read_reg(reg)	__kw_read_reg(host, reg) 
+
+
+/* Don't schedule, the g5 fan controller is too
+ * timing sensitive
+ */
+static u8 kw_wait_interrupt(struct low_i2c_host* host)
+{
+	int i, j;
+	u8 isr;
+	
+	for (i = 0; i < 100000; i++) {
+		isr = kw_read_reg(reg_isr) & KW_I2C_IRQ_MASK;
+		if (isr != 0)
+			return isr;
+
+		/* This code is used with the timebase frozen, we cannot rely
+		 * on udelay ! For now, just use a bogus loop
+		 */
+		for (j = 1; j < 10000; j++)
+			mb();
+	}
+	return isr;
+}
+
+static int kw_handle_interrupt(struct low_i2c_host *host, int state, int rw, int *rc, u8 **data, int *len, u8 isr)
+{
+	u8 ack;
+
+	DBG("kw_handle_interrupt(%s, isr: %x)\n", __kw_state_names[state], isr);
+
+	if (isr == 0) {
+		if (state != state_stop) {
+			DBG("KW: Timeout !\n");
+			*rc = -EIO;
+			goto stop;
+		}
+		if (state == state_stop) {
+			ack = kw_read_reg(reg_status);
+			if (!(ack & KW_I2C_STAT_BUSY)) {
+				state = state_idle;
+				kw_write_reg(reg_ier, 0x00);
+			}
+		}
+		return state;
+	}
+
+	if (isr & KW_I2C_IRQ_ADDR) {
+		ack = kw_read_reg(reg_status);
+		if (state != state_addr) {
+			kw_write_reg(reg_isr, KW_I2C_IRQ_ADDR);
+			WRONG_STATE("KW_I2C_IRQ_ADDR"); 
+			*rc = -EIO;
+			goto stop;
+		}
+		if ((ack & KW_I2C_STAT_LAST_AAK) == 0) {			
+			*rc = -ENODEV;
+			DBG("KW: NAK on address\n");
+			return state_stop;		     
+		} else {
+			if (rw) {
+				state = state_read;
+				if (*len > 1)
+					kw_write_reg(reg_control, KW_I2C_CTL_AAK);
+			} else {
+				state = state_write;
+				kw_write_reg(reg_data, **data);
+				(*data)++; (*len)--;
+			}
+		}
+		kw_write_reg(reg_isr, KW_I2C_IRQ_ADDR);
+	}
+
+	if (isr & KW_I2C_IRQ_DATA) {
+		if (state == state_read) {
+			**data = kw_read_reg(reg_data);
+			(*data)++; (*len)--;
+			kw_write_reg(reg_isr, KW_I2C_IRQ_DATA);
+			if ((*len) == 0)
+				state = state_stop;
+			else if ((*len) == 1)
+				kw_write_reg(reg_control, 0);
+		} else if (state == state_write) {
+			ack = kw_read_reg(reg_status);
+			if ((ack & KW_I2C_STAT_LAST_AAK) == 0) {
+				DBG("KW: nack on data write\n");
+				*rc = -EIO;
+				goto stop;
+			} else if (*len) {
+				kw_write_reg(reg_data, **data);
+				(*data)++; (*len)--;
+			} else {
+				kw_write_reg(reg_control, KW_I2C_CTL_STOP);
+				state = state_stop;
+				*rc = 0;
+			}
+			kw_write_reg(reg_isr, KW_I2C_IRQ_DATA);
+		} else {
+			kw_write_reg(reg_isr, KW_I2C_IRQ_DATA);
+			WRONG_STATE("KW_I2C_IRQ_DATA"); 
+			if (state != state_stop) {
+				*rc = -EIO;
+				goto stop;
+			}
+		}
+	}
+
+	if (isr & KW_I2C_IRQ_STOP) {
+		kw_write_reg(reg_isr, KW_I2C_IRQ_STOP);
+		if (state != state_stop) {
+			WRONG_STATE("KW_I2C_IRQ_STOP");
+			*rc = -EIO;
+		}
+		return state_idle;
+	}
+
+	if (isr & KW_I2C_IRQ_START)
+		kw_write_reg(reg_isr, KW_I2C_IRQ_START);
+
+	return state;
+
+ stop:
+	kw_write_reg(reg_control, KW_I2C_CTL_STOP);	
+	return state_stop;
+}
+
+static int keywest_low_i2c_func(struct low_i2c_host *host, u8 addr, u8 subaddr, u8 *data, int len)
+{
+	u8 mode_reg = host->speed;
+	int state = state_addr;
+	int rc = 0;
+
+	/* Setup mode & subaddress if any */
+	switch(host->mode) {
+	case pmac_low_i2c_mode_dumb:
+		printk(KERN_ERR "low_i2c: Dumb mode not supported !\n");
+		return -EINVAL;
+	case pmac_low_i2c_mode_std:
+		mode_reg |= KW_I2C_MODE_STANDARD;
+		break;
+	case pmac_low_i2c_mode_stdsub:
+		mode_reg |= KW_I2C_MODE_STANDARDSUB;
+		break;
+	case pmac_low_i2c_mode_combined:
+		mode_reg |= KW_I2C_MODE_COMBINED;
+		break;
+	}
+
+	/* Setup channel & clear pending irqs */
+	kw_write_reg(reg_isr, kw_read_reg(reg_isr));
+	kw_write_reg(reg_mode, mode_reg | (host->channel << 4));
+	kw_write_reg(reg_status, 0);
+
+	/* Set up address and r/w bit */
+	kw_write_reg(reg_addr, addr);
+
+	/* Set up the sub address */
+	if ((mode_reg & KW_I2C_MODE_MODE_MASK) == KW_I2C_MODE_STANDARDSUB
+	    || (mode_reg & KW_I2C_MODE_MODE_MASK) == KW_I2C_MODE_COMBINED)
+		kw_write_reg(reg_subaddr, subaddr);
+
+	/* Start sending address & disable interrupt*/
+	kw_write_reg(reg_ier, 0 /*KW_I2C_IRQ_MASK*/);
+	kw_write_reg(reg_control, KW_I2C_CTL_XADDR);
+
+	/* State machine, to turn into an interrupt handler */
+	while(state != state_idle) {
+		u8 isr = kw_wait_interrupt(host);
+		state = kw_handle_interrupt(host, state, addr & 1, &rc, &data, &len, isr);
+	}
+
+	return rc;
+}
+
+static void keywest_low_i2c_add(struct device_node *np)
+{
+	struct low_i2c_host	*host = find_low_i2c_host(NULL);
+	u32			*psteps, *prate, steps, aoffset = 0;
+	struct device_node	*parent;
+
+	if (host == NULL) {
+		printk(KERN_ERR "low_i2c: Can't allocate host for %s\n",
+		       np->full_name);
+		return;
+	}
+	memset(host, 0, sizeof(*host));
+
+	init_MUTEX(&host->mutex);
+	host->np = of_node_get(np);	
+	psteps = (u32 *)get_property(np, "AAPL,address-step", NULL);
+	steps = psteps ? (*psteps) : 0x10;
+	for (host->bsteps = 0; (steps & 0x01) == 0; host->bsteps++)
+		steps >>= 1;
+	parent = of_get_parent(np);
+	host->num_channels = 1;
+	if (parent && parent->name[0] == 'u') {
+		host->num_channels = 2;
+		aoffset = 3;
+	}
+	/* Select interface rate */
+	host->speed = KW_I2C_MODE_100KHZ;
+	prate = (u32 *)get_property(np, "AAPL,i2c-rate", NULL);
+	if (prate) switch(*prate) {
+	case 100:
+		host->speed = KW_I2C_MODE_100KHZ;
+		break;
+	case 50:
+		host->speed = KW_I2C_MODE_50KHZ;
+		break;
+	case 25:
+		host->speed = KW_I2C_MODE_25KHZ;
+		break;
+	}	
+
+	host->mode = pmac_low_i2c_mode_std;
+	host->base = ioremap(np->addrs[0].address + aoffset,
+						np->addrs[0].size);
+	host->func = keywest_low_i2c_func;
+}
+
+/*
+ *
+ * PMU implementation
+ *
+ */
+
+
+#ifdef CONFIG_ADB_PMU
+
+static int pmu_low_i2c_func(struct low_i2c_host *host, u8 addr, u8 sub, u8 *data, int len)
+{
+	// TODO
+	return -ENODEV;
+}
+
+static void pmu_low_i2c_add(struct device_node *np)
+{
+	struct low_i2c_host	*host = find_low_i2c_host(NULL);
+
+	if (host == NULL) {
+		printk(KERN_ERR "low_i2c: Can't allocate host for %s\n",
+		       np->full_name);
+		return;
+	}
+	memset(host, 0, sizeof(*host));
+
+	init_MUTEX(&host->mutex);
+	host->np = of_node_get(np);	
+	host->num_channels = 3;
+	host->mode = pmac_low_i2c_mode_std;
+	host->func = pmu_low_i2c_func;
+}
+
+#endif /* CONFIG_ADB_PMU */
+
+void __init pmac_init_low_i2c(void)
+{
+	struct device_node *np;
+
+	/* Probe keywest-i2c busses */
+	np = of_find_compatible_node(NULL, "i2c", "keywest-i2c");
+	while(np) {
+		keywest_low_i2c_add(np);
+		np = of_find_compatible_node(np, "i2c", "keywest-i2c");
+	}
+
+#ifdef CONFIG_ADB_PMU
+	/* Probe PMU busses */
+	np = of_find_node_by_name(NULL, "via-pmu");
+	if (np)
+		pmu_low_i2c_add(np);
+#endif /* CONFIG_ADB_PMU */
+
+	/* TODO: Add CUDA support as well */
+}
+
+int pmac_low_i2c_lock(struct device_node *np)
+{
+	struct low_i2c_host *host = find_low_i2c_host(np);
+
+	if (!host)
+		return -ENODEV;
+	down(&host->mutex);
+	return 0;
+}
+EXPORT_SYMBOL(pmac_low_i2c_lock);
+
+int pmac_low_i2c_unlock(struct device_node *np)
+{
+	struct low_i2c_host *host = find_low_i2c_host(np);
+
+	if (!host)
+		return -ENODEV;
+	up(&host->mutex);
+	return 0;
+}
+EXPORT_SYMBOL(pmac_low_i2c_unlock);
+
+
+int pmac_low_i2c_open(struct device_node *np, int channel)
+{
+	struct low_i2c_host *host = find_low_i2c_host(np);
+
+	if (!host)
+		return -ENODEV;
+
+	if (channel >= host->num_channels)
+		return -EINVAL;
+
+	down(&host->mutex);
+	host->is_open = 1;
+	host->channel = channel;
+
+	return 0;
+}
+EXPORT_SYMBOL(pmac_low_i2c_open);
+
+int pmac_low_i2c_close(struct device_node *np)
+{
+	struct low_i2c_host *host = find_low_i2c_host(np);
+
+	if (!host)
+		return -ENODEV;
+
+	host->is_open = 0;
+	up(&host->mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL(pmac_low_i2c_close);
+
+int pmac_low_i2c_setmode(struct device_node *np, int mode)
+{
+	struct low_i2c_host *host = find_low_i2c_host(np);
+
+	if (!host)
+		return -ENODEV;
+	WARN_ON(!host->is_open);
+	host->mode = mode;
+
+	return 0;
+}
+EXPORT_SYMBOL(pmac_low_i2c_setmode);
+
+int pmac_low_i2c_xfer(struct device_node *np, u8 addrdir, u8 subaddr, u8 *data, int len)
+{
+	struct low_i2c_host *host = find_low_i2c_host(np);
+
+	if (!host)
+		return -ENODEV;
+	WARN_ON(!host->is_open);
+
+	return host->func(host, addrdir, subaddr, data, len);
+}
+EXPORT_SYMBOL(pmac_low_i2c_xfer);
+
diff --git a/arch/powerpc/platforms/powermac/nvram.c b/arch/powerpc/platforms/powermac/nvram.c
new file mode 100644
index 000000000000..4042e2f06ee0
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/nvram.c
@@ -0,0 +1,645 @@
+/*
+ *  arch/ppc/platforms/pmac_nvram.c
+ *
+ *  Copyright (C) 2002 Benjamin Herrenschmidt (benh@kernel.crashing.org)
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ *  Todo: - add support for the OF persistent properties
+ */
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/stddef.h>
+#include <linux/string.h>
+#include <linux/nvram.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/errno.h>
+#include <linux/adb.h>
+#include <linux/pmu.h>
+#include <linux/bootmem.h>
+#include <linux/completion.h>
+#include <linux/spinlock.h>
+#include <asm/sections.h>
+#include <asm/io.h>
+#include <asm/system.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+#include <asm/nvram.h>
+
+#define DEBUG
+
+#ifdef DEBUG
+#define DBG(x...) printk(x)
+#else
+#define DBG(x...)
+#endif
+
+#define NVRAM_SIZE		0x2000	/* 8kB of non-volatile RAM */
+
+#define CORE99_SIGNATURE	0x5a
+#define CORE99_ADLER_START	0x14
+
+/* On Core99, nvram is either a sharp, a micron or an AMD flash */
+#define SM_FLASH_STATUS_DONE	0x80
+#define SM_FLASH_STATUS_ERR	0x38
+
+#define SM_FLASH_CMD_ERASE_CONFIRM	0xd0
+#define SM_FLASH_CMD_ERASE_SETUP	0x20
+#define SM_FLASH_CMD_RESET		0xff
+#define SM_FLASH_CMD_WRITE_SETUP	0x40
+#define SM_FLASH_CMD_CLEAR_STATUS	0x50
+#define SM_FLASH_CMD_READ_STATUS	0x70
+
+/* CHRP NVRAM header */
+struct chrp_header {
+  u8		signature;
+  u8		cksum;
+  u16		len;
+  char          name[12];
+  u8		data[0];
+};
+
+struct core99_header {
+  struct chrp_header	hdr;
+  u32			adler;
+  u32			generation;
+  u32			reserved[2];
+};
+
+/*
+ * Read and write the non-volatile RAM on PowerMacs and CHRP machines.
+ */
+static int nvram_naddrs;
+static volatile unsigned char *nvram_data;
+static int is_core_99;
+static int core99_bank = 0;
+static int nvram_partitions[3];
+// XXX Turn that into a sem
+static DEFINE_SPINLOCK(nv_lock);
+
+extern int pmac_newworld;
+extern int system_running;
+
+static int (*core99_write_bank)(int bank, u8* datas);
+static int (*core99_erase_bank)(int bank);
+
+static char *nvram_image;
+
+
+static unsigned char core99_nvram_read_byte(int addr)
+{
+	if (nvram_image == NULL)
+		return 0xff;
+	return nvram_image[addr];
+}
+
+static void core99_nvram_write_byte(int addr, unsigned char val)
+{
+	if (nvram_image == NULL)
+		return;
+	nvram_image[addr] = val;
+}
+
+static ssize_t core99_nvram_read(char *buf, size_t count, loff_t *index)
+{
+	int i;
+
+	if (nvram_image == NULL)
+		return -ENODEV;
+	if (*index > NVRAM_SIZE)
+		return 0;
+
+	i = *index;
+	if (i + count > NVRAM_SIZE)
+		count = NVRAM_SIZE - i;
+
+	memcpy(buf, &nvram_image[i], count);
+	*index = i + count;
+	return count;
+}
+
+static ssize_t core99_nvram_write(char *buf, size_t count, loff_t *index)
+{
+	int i;
+
+	if (nvram_image == NULL)
+		return -ENODEV;
+	if (*index > NVRAM_SIZE)
+		return 0;
+
+	i = *index;
+	if (i + count > NVRAM_SIZE)
+		count = NVRAM_SIZE - i;
+
+	memcpy(&nvram_image[i], buf, count);
+	*index = i + count;
+	return count;
+}
+
+static ssize_t core99_nvram_size(void)
+{
+	if (nvram_image == NULL)
+		return -ENODEV;
+	return NVRAM_SIZE;
+}
+
+#ifdef CONFIG_PPC32
+static volatile unsigned char *nvram_addr;
+static int nvram_mult;
+
+static unsigned char direct_nvram_read_byte(int addr)
+{
+	return in_8(&nvram_data[(addr & (NVRAM_SIZE - 1)) * nvram_mult]);
+}
+
+static void direct_nvram_write_byte(int addr, unsigned char val)
+{
+	out_8(&nvram_data[(addr & (NVRAM_SIZE - 1)) * nvram_mult], val);
+}
+
+
+static unsigned char indirect_nvram_read_byte(int addr)
+{
+	unsigned char val;
+	unsigned long flags;
+
+	spin_lock_irqsave(&nv_lock, flags);
+	out_8(nvram_addr, addr >> 5);
+	val = in_8(&nvram_data[(addr & 0x1f) << 4]);
+	spin_unlock_irqrestore(&nv_lock, flags);
+
+	return val;
+}
+
+static void indirect_nvram_write_byte(int addr, unsigned char val)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&nv_lock, flags);
+	out_8(nvram_addr, addr >> 5);
+	out_8(&nvram_data[(addr & 0x1f) << 4], val);
+	spin_unlock_irqrestore(&nv_lock, flags);
+}
+
+
+#ifdef CONFIG_ADB_PMU
+
+static void pmu_nvram_complete(struct adb_request *req)
+{
+	if (req->arg)
+		complete((struct completion *)req->arg);
+}
+
+static unsigned char pmu_nvram_read_byte(int addr)
+{
+	struct adb_request req;
+	DECLARE_COMPLETION(req_complete); 
+	
+	req.arg = system_state == SYSTEM_RUNNING ? &req_complete : NULL;
+	if (pmu_request(&req, pmu_nvram_complete, 3, PMU_READ_NVRAM,
+			(addr >> 8) & 0xff, addr & 0xff))
+		return 0xff;
+	if (system_state == SYSTEM_RUNNING)
+		wait_for_completion(&req_complete);
+	while (!req.complete)
+		pmu_poll();
+	return req.reply[0];
+}
+
+static void pmu_nvram_write_byte(int addr, unsigned char val)
+{
+	struct adb_request req;
+	DECLARE_COMPLETION(req_complete); 
+	
+	req.arg = system_state == SYSTEM_RUNNING ? &req_complete : NULL;
+	if (pmu_request(&req, pmu_nvram_complete, 4, PMU_WRITE_NVRAM,
+			(addr >> 8) & 0xff, addr & 0xff, val))
+		return;
+	if (system_state == SYSTEM_RUNNING)
+		wait_for_completion(&req_complete);
+	while (!req.complete)
+		pmu_poll();
+}
+
+#endif /* CONFIG_ADB_PMU */
+#endif /* CONFIG_PPC32 */
+
+static u8 chrp_checksum(struct chrp_header* hdr)
+{
+	u8 *ptr;
+	u16 sum = hdr->signature;
+	for (ptr = (u8 *)&hdr->len; ptr < hdr->data; ptr++)
+		sum += *ptr;
+	while (sum > 0xFF)
+		sum = (sum & 0xFF) + (sum>>8);
+	return sum;
+}
+
+static u32 core99_calc_adler(u8 *buffer)
+{
+	int cnt;
+	u32 low, high;
+
+   	buffer += CORE99_ADLER_START;
+	low = 1;
+	high = 0;
+	for (cnt=0; cnt<(NVRAM_SIZE-CORE99_ADLER_START); cnt++) {
+		if ((cnt % 5000) == 0) {
+			high  %= 65521UL;
+			high %= 65521UL;
+		}
+		low += buffer[cnt];
+		high += low;
+	}
+	low  %= 65521UL;
+	high %= 65521UL;
+
+	return (high << 16) | low;
+}
+
+static u32 core99_check(u8* datas)
+{
+	struct core99_header* hdr99 = (struct core99_header*)datas;
+
+	if (hdr99->hdr.signature != CORE99_SIGNATURE) {
+		DBG("Invalid signature\n");
+		return 0;
+	}
+	if (hdr99->hdr.cksum != chrp_checksum(&hdr99->hdr)) {
+		DBG("Invalid checksum\n");
+		return 0;
+	}
+	if (hdr99->adler != core99_calc_adler(datas)) {
+		DBG("Invalid adler\n");
+		return 0;
+	}
+	return hdr99->generation;
+}
+
+static int sm_erase_bank(int bank)
+{
+	int stat, i;
+	unsigned long timeout;
+
+	u8* base = (u8 *)nvram_data + core99_bank*NVRAM_SIZE;
+
+       	DBG("nvram: Sharp/Micron Erasing bank %d...\n", bank);
+
+	out_8(base, SM_FLASH_CMD_ERASE_SETUP);
+	out_8(base, SM_FLASH_CMD_ERASE_CONFIRM);
+	timeout = 0;
+	do {
+		if (++timeout > 1000000) {
+			printk(KERN_ERR "nvram: Sharp/Micron flash erase timeout !\n");
+			break;
+		}
+		out_8(base, SM_FLASH_CMD_READ_STATUS);
+		stat = in_8(base);
+	} while (!(stat & SM_FLASH_STATUS_DONE));
+
+	out_8(base, SM_FLASH_CMD_CLEAR_STATUS);
+	out_8(base, SM_FLASH_CMD_RESET);
+
+	for (i=0; i<NVRAM_SIZE; i++)
+		if (base[i] != 0xff) {
+			printk(KERN_ERR "nvram: Sharp/Micron flash erase failed !\n");
+			return -ENXIO;
+		}
+	return 0;
+}
+
+static int sm_write_bank(int bank, u8* datas)
+{
+	int i, stat = 0;
+	unsigned long timeout;
+
+	u8* base = (u8 *)nvram_data + core99_bank*NVRAM_SIZE;
+
+       	DBG("nvram: Sharp/Micron Writing bank %d...\n", bank);
+
+	for (i=0; i<NVRAM_SIZE; i++) {
+		out_8(base+i, SM_FLASH_CMD_WRITE_SETUP);
+		udelay(1);
+		out_8(base+i, datas[i]);
+		timeout = 0;
+		do {
+			if (++timeout > 1000000) {
+				printk(KERN_ERR "nvram: Sharp/Micron flash write timeout !\n");
+				break;
+			}
+			out_8(base, SM_FLASH_CMD_READ_STATUS);
+			stat = in_8(base);
+		} while (!(stat & SM_FLASH_STATUS_DONE));
+		if (!(stat & SM_FLASH_STATUS_DONE))
+			break;
+	}
+	out_8(base, SM_FLASH_CMD_CLEAR_STATUS);
+	out_8(base, SM_FLASH_CMD_RESET);
+	for (i=0; i<NVRAM_SIZE; i++)
+		if (base[i] != datas[i]) {
+			printk(KERN_ERR "nvram: Sharp/Micron flash write failed !\n");
+			return -ENXIO;
+		}
+	return 0;
+}
+
+static int amd_erase_bank(int bank)
+{
+	int i, stat = 0;
+	unsigned long timeout;
+
+	u8* base = (u8 *)nvram_data + core99_bank*NVRAM_SIZE;
+
+       	DBG("nvram: AMD Erasing bank %d...\n", bank);
+
+	/* Unlock 1 */
+	out_8(base+0x555, 0xaa);
+	udelay(1);
+	/* Unlock 2 */
+	out_8(base+0x2aa, 0x55);
+	udelay(1);
+
+	/* Sector-Erase */
+	out_8(base+0x555, 0x80);
+	udelay(1);
+	out_8(base+0x555, 0xaa);
+	udelay(1);
+	out_8(base+0x2aa, 0x55);
+	udelay(1);
+	out_8(base, 0x30);
+	udelay(1);
+
+	timeout = 0;
+	do {
+		if (++timeout > 1000000) {
+			printk(KERN_ERR "nvram: AMD flash erase timeout !\n");
+			break;
+		}
+		stat = in_8(base) ^ in_8(base);
+	} while (stat != 0);
+	
+	/* Reset */
+	out_8(base, 0xf0);
+	udelay(1);
+	
+	for (i=0; i<NVRAM_SIZE; i++)
+		if (base[i] != 0xff) {
+			printk(KERN_ERR "nvram: AMD flash erase failed !\n");
+			return -ENXIO;
+		}
+	return 0;
+}
+
+static int amd_write_bank(int bank, u8* datas)
+{
+	int i, stat = 0;
+	unsigned long timeout;
+
+	u8* base = (u8 *)nvram_data + core99_bank*NVRAM_SIZE;
+
+       	DBG("nvram: AMD Writing bank %d...\n", bank);
+
+	for (i=0; i<NVRAM_SIZE; i++) {
+		/* Unlock 1 */
+		out_8(base+0x555, 0xaa);
+		udelay(1);
+		/* Unlock 2 */
+		out_8(base+0x2aa, 0x55);
+		udelay(1);
+
+		/* Write single word */
+		out_8(base+0x555, 0xa0);
+		udelay(1);
+		out_8(base+i, datas[i]);
+		
+		timeout = 0;
+		do {
+			if (++timeout > 1000000) {
+				printk(KERN_ERR "nvram: AMD flash write timeout !\n");
+				break;
+			}
+			stat = in_8(base) ^ in_8(base);
+		} while (stat != 0);
+		if (stat != 0)
+			break;
+	}
+
+	/* Reset */
+	out_8(base, 0xf0);
+	udelay(1);
+
+	for (i=0; i<NVRAM_SIZE; i++)
+		if (base[i] != datas[i]) {
+			printk(KERN_ERR "nvram: AMD flash write failed !\n");
+			return -ENXIO;
+		}
+	return 0;
+}
+
+static void __init lookup_partitions(void)
+{
+	u8 buffer[17];
+	int i, offset;
+	struct chrp_header* hdr;
+
+	if (pmac_newworld) {
+		nvram_partitions[pmac_nvram_OF] = -1;
+		nvram_partitions[pmac_nvram_XPRAM] = -1;
+		nvram_partitions[pmac_nvram_NR] = -1;
+		hdr = (struct chrp_header *)buffer;
+
+		offset = 0;
+		buffer[16] = 0;
+		do {
+			for (i=0;i<16;i++)
+				buffer[i] = ppc_md.nvram_read_val(offset+i);
+			if (!strcmp(hdr->name, "common"))
+				nvram_partitions[pmac_nvram_OF] = offset + 0x10;
+			if (!strcmp(hdr->name, "APL,MacOS75")) {
+				nvram_partitions[pmac_nvram_XPRAM] = offset + 0x10;
+				nvram_partitions[pmac_nvram_NR] = offset + 0x110;
+			}
+			offset += (hdr->len * 0x10);
+		} while(offset < NVRAM_SIZE);
+	} else {
+		nvram_partitions[pmac_nvram_OF] = 0x1800;
+		nvram_partitions[pmac_nvram_XPRAM] = 0x1300;
+		nvram_partitions[pmac_nvram_NR] = 0x1400;
+	}
+	DBG("nvram: OF partition at 0x%x\n", nvram_partitions[pmac_nvram_OF]);
+	DBG("nvram: XP partition at 0x%x\n", nvram_partitions[pmac_nvram_XPRAM]);
+	DBG("nvram: NR partition at 0x%x\n", nvram_partitions[pmac_nvram_NR]);
+}
+
+static void core99_nvram_sync(void)
+{
+	struct core99_header* hdr99;
+	unsigned long flags;
+
+	if (!is_core_99 || !nvram_data || !nvram_image)
+		return;
+
+	spin_lock_irqsave(&nv_lock, flags);
+	if (!memcmp(nvram_image, (u8*)nvram_data + core99_bank*NVRAM_SIZE,
+		NVRAM_SIZE))
+		goto bail;
+
+	DBG("Updating nvram...\n");
+
+	hdr99 = (struct core99_header*)nvram_image;
+	hdr99->generation++;
+	hdr99->hdr.signature = CORE99_SIGNATURE;
+	hdr99->hdr.cksum = chrp_checksum(&hdr99->hdr);
+	hdr99->adler = core99_calc_adler(nvram_image);
+	core99_bank = core99_bank ? 0 : 1;
+	if (core99_erase_bank)
+		if (core99_erase_bank(core99_bank)) {
+			printk("nvram: Error erasing bank %d\n", core99_bank);
+			goto bail;
+		}
+	if (core99_write_bank)
+		if (core99_write_bank(core99_bank, nvram_image))
+			printk("nvram: Error writing bank %d\n", core99_bank);
+ bail:
+	spin_unlock_irqrestore(&nv_lock, flags);
+
+#ifdef DEBUG
+       	mdelay(2000);
+#endif
+}
+
+static int __init core99_nvram_setup(struct device_node *dp)
+{
+	int i;
+	u32 gen_bank0, gen_bank1;
+
+	if (nvram_naddrs < 1) {
+		printk(KERN_ERR "nvram: no address\n");
+		return -EINVAL;
+	}
+	nvram_image = alloc_bootmem(NVRAM_SIZE);
+	if (nvram_image == NULL) {
+		printk(KERN_ERR "nvram: can't allocate ram image\n");
+		return -ENOMEM;
+	}
+	nvram_data = ioremap(dp->addrs[0].address, NVRAM_SIZE*2);
+	nvram_naddrs = 1; /* Make sure we get the correct case */
+
+	DBG("nvram: Checking bank 0...\n");
+
+	gen_bank0 = core99_check((u8 *)nvram_data);
+	gen_bank1 = core99_check((u8 *)nvram_data + NVRAM_SIZE);
+	core99_bank = (gen_bank0 < gen_bank1) ? 1 : 0;
+
+	DBG("nvram: gen0=%d, gen1=%d\n", gen_bank0, gen_bank1);
+	DBG("nvram: Active bank is: %d\n", core99_bank);
+
+	for (i=0; i<NVRAM_SIZE; i++)
+		nvram_image[i] = nvram_data[i + core99_bank*NVRAM_SIZE];
+
+	ppc_md.nvram_read_val	= core99_nvram_read_byte;
+	ppc_md.nvram_write_val	= core99_nvram_write_byte;
+	ppc_md.nvram_read	= core99_nvram_read;
+	ppc_md.nvram_write	= core99_nvram_write;
+	ppc_md.nvram_size	= core99_nvram_size;
+	ppc_md.nvram_sync	= core99_nvram_sync;
+	/* 
+	 * Maybe we could be smarter here though making an exclusive list
+	 * of known flash chips is a bit nasty as older OF didn't provide us
+	 * with a useful "compatible" entry. A solution would be to really
+	 * identify the chip using flash id commands and base ourselves on
+	 * a list of known chips IDs
+	 */
+	if (device_is_compatible(dp, "amd-0137")) {
+		core99_erase_bank = amd_erase_bank;
+		core99_write_bank = amd_write_bank;
+	} else {
+		core99_erase_bank = sm_erase_bank;
+		core99_write_bank = sm_write_bank;
+	}
+	return 0;
+}
+
+int __init pmac_nvram_init(void)
+{
+	struct device_node *dp;
+	int err = 0;
+
+	nvram_naddrs = 0;
+
+	dp = find_devices("nvram");
+	if (dp == NULL) {
+		printk(KERN_ERR "Can't find NVRAM device\n");
+		return -ENODEV;
+	}
+	nvram_naddrs = dp->n_addrs;
+	is_core_99 = device_is_compatible(dp, "nvram,flash");
+	if (is_core_99)
+		err = core99_nvram_setup(dp);
+#ifdef CONFIG_PPC32
+	else if (_machine == _MACH_chrp && nvram_naddrs == 1) {
+		nvram_data = ioremap(dp->addrs[0].address + isa_mem_base,
+				     dp->addrs[0].size);
+		nvram_mult = 1;
+		ppc_md.nvram_read_val	= direct_nvram_read_byte;
+		ppc_md.nvram_write_val	= direct_nvram_write_byte;
+	} else if (nvram_naddrs == 1) {
+		nvram_data = ioremap(dp->addrs[0].address, dp->addrs[0].size);
+		nvram_mult = (dp->addrs[0].size + NVRAM_SIZE - 1) / NVRAM_SIZE;
+		ppc_md.nvram_read_val	= direct_nvram_read_byte;
+		ppc_md.nvram_write_val	= direct_nvram_write_byte;
+	} else if (nvram_naddrs == 2) {
+		nvram_addr = ioremap(dp->addrs[0].address, dp->addrs[0].size);
+		nvram_data = ioremap(dp->addrs[1].address, dp->addrs[1].size);
+		ppc_md.nvram_read_val	= indirect_nvram_read_byte;
+		ppc_md.nvram_write_val	= indirect_nvram_write_byte;
+	} else if (nvram_naddrs == 0 && sys_ctrler == SYS_CTRLER_PMU) {
+#ifdef CONFIG_ADB_PMU
+		nvram_naddrs = -1;
+		ppc_md.nvram_read_val	= pmu_nvram_read_byte;
+		ppc_md.nvram_write_val	= pmu_nvram_write_byte;
+#endif /* CONFIG_ADB_PMU */
+	}
+#endif
+	else {
+		printk(KERN_ERR "Incompatible type of NVRAM\n");
+		return -ENXIO;
+	}
+	lookup_partitions();
+	return err;
+}
+
+int pmac_get_partition(int partition)
+{
+	return nvram_partitions[partition];
+}
+
+u8 pmac_xpram_read(int xpaddr)
+{
+	int offset = pmac_get_partition(pmac_nvram_XPRAM);
+
+	if (offset < 0 || xpaddr < 0 || xpaddr > 0x100)
+		return 0xff;
+
+	return ppc_md.nvram_read_val(xpaddr + offset);
+}
+
+void pmac_xpram_write(int xpaddr, u8 data)
+{
+	int offset = pmac_get_partition(pmac_nvram_XPRAM);
+
+	if (offset < 0 || xpaddr < 0 || xpaddr > 0x100)
+		return;
+
+	ppc_md.nvram_write_val(xpaddr + offset, data);
+}
+
+EXPORT_SYMBOL(pmac_get_partition);
+EXPORT_SYMBOL(pmac_xpram_read);
+EXPORT_SYMBOL(pmac_xpram_write);
diff --git a/arch/powerpc/platforms/powermac/pci.c b/arch/powerpc/platforms/powermac/pci.c
new file mode 100644
index 000000000000..8f818d092e2b
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/pci.c
@@ -0,0 +1,1170 @@
+/*
+ * Support for PCI bridges found on Power Macintoshes.
+ *
+ * Copyright (C) 2003 Benjamin Herrenschmuidt (benh@kernel.crashing.org)
+ * Copyright (C) 1997 Paul Mackerras (paulus@samba.org)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/bootmem.h>
+
+#include <asm/sections.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/pci-bridge.h>
+#include <asm/machdep.h>
+#include <asm/pmac_feature.h>
+#include <asm/grackle.h>
+#ifdef CONFIG_PPC64
+#include <asm/iommu.h>
+#include <asm/ppc-pci.h>
+#endif
+
+#undef DEBUG
+
+#ifdef DEBUG
+#define DBG(x...) printk(x)
+#else
+#define DBG(x...)
+#endif
+
+static int add_bridge(struct device_node *dev);
+
+/* XXX Could be per-controller, but I don't think we risk anything by
+ * assuming we won't have both UniNorth and Bandit */
+static int has_uninorth;
+#ifdef CONFIG_PPC64
+static struct pci_controller *u3_agp;
+static struct pci_controller *u3_ht;
+#endif /* CONFIG_PPC64 */
+
+extern u8 pci_cache_line_size;
+extern int pcibios_assign_bus_offset;
+
+struct device_node *k2_skiplist[2];
+
+/*
+ * Magic constants for enabling cache coherency in the bandit/PSX bridge.
+ */
+#define BANDIT_DEVID_2	8
+#define BANDIT_REVID	3
+
+#define BANDIT_DEVNUM	11
+#define BANDIT_MAGIC	0x50
+#define BANDIT_COHERENT	0x40
+
+static int __init fixup_one_level_bus_range(struct device_node *node, int higher)
+{
+	for (; node != 0;node = node->sibling) {
+		int * bus_range;
+		unsigned int *class_code;
+		int len;
+
+		/* For PCI<->PCI bridges or CardBus bridges, we go down */
+		class_code = (unsigned int *) get_property(node, "class-code", NULL);
+		if (!class_code || ((*class_code >> 8) != PCI_CLASS_BRIDGE_PCI &&
+			(*class_code >> 8) != PCI_CLASS_BRIDGE_CARDBUS))
+			continue;
+		bus_range = (int *) get_property(node, "bus-range", &len);
+		if (bus_range != NULL && len > 2 * sizeof(int)) {
+			if (bus_range[1] > higher)
+				higher = bus_range[1];
+		}
+		higher = fixup_one_level_bus_range(node->child, higher);
+	}
+	return higher;
+}
+
+/* This routine fixes the "bus-range" property of all bridges in the
+ * system since they tend to have their "last" member wrong on macs
+ *
+ * Note that the bus numbers manipulated here are OF bus numbers, they
+ * are not Linux bus numbers.
+ */
+static void __init fixup_bus_range(struct device_node *bridge)
+{
+	int * bus_range;
+	int len;
+
+	/* Lookup the "bus-range" property for the hose */
+	bus_range = (int *) get_property(bridge, "bus-range", &len);
+	if (bus_range == NULL || len < 2 * sizeof(int)) {
+		printk(KERN_WARNING "Can't get bus-range for %s\n",
+			       bridge->full_name);
+		return;
+	}
+	bus_range[1] = fixup_one_level_bus_range(bridge->child, bus_range[1]);
+}
+
+/*
+ * Apple MacRISC (U3, UniNorth, Bandit, Chaos) PCI controllers.
+ *
+ * The "Bandit" version is present in all early PCI PowerMacs,
+ * and up to the first ones using Grackle. Some machines may
+ * have 2 bandit controllers (2 PCI busses).
+ *
+ * "Chaos" is used in some "Bandit"-type machines as a bridge
+ * for the separate display bus. It is accessed the same
+ * way as bandit, but cannot be probed for devices. It therefore
+ * has its own config access functions.
+ *
+ * The "UniNorth" version is present in all Core99 machines
+ * (iBook, G4, new IMacs, and all the recent Apple machines).
+ * It contains 3 controllers in one ASIC.
+ *
+ * The U3 is the bridge used on G5 machines. It contains an
+ * AGP bus which is dealt with the old UniNorth access routines
+ * and a HyperTransport bus which uses its own set of access
+ * functions.
+ */
+
+#define MACRISC_CFA0(devfn, off)	\
+	((1 << (unsigned long)PCI_SLOT(dev_fn)) \
+	| (((unsigned long)PCI_FUNC(dev_fn)) << 8) \
+	| (((unsigned long)(off)) & 0xFCUL))
+
+#define MACRISC_CFA1(bus, devfn, off)	\
+	((((unsigned long)(bus)) << 16) \
+	|(((unsigned long)(devfn)) << 8) \
+	|(((unsigned long)(off)) & 0xFCUL) \
+	|1UL)
+
+static unsigned long macrisc_cfg_access(struct pci_controller* hose,
+					       u8 bus, u8 dev_fn, u8 offset)
+{
+	unsigned int caddr;
+
+	if (bus == hose->first_busno) {
+		if (dev_fn < (11 << 3))
+			return 0;
+		caddr = MACRISC_CFA0(dev_fn, offset);
+	} else
+		caddr = MACRISC_CFA1(bus, dev_fn, offset);
+
+	/* Uninorth will return garbage if we don't read back the value ! */
+	do {
+		out_le32(hose->cfg_addr, caddr);
+	} while (in_le32(hose->cfg_addr) != caddr);
+
+	offset &= has_uninorth ? 0x07 : 0x03;
+	return ((unsigned long)hose->cfg_data) + offset;
+}
+
+static int macrisc_read_config(struct pci_bus *bus, unsigned int devfn,
+				      int offset, int len, u32 *val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = macrisc_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		*val = in_8((u8 *)addr);
+		break;
+	case 2:
+		*val = in_le16((u16 *)addr);
+		break;
+	default:
+		*val = in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static int macrisc_write_config(struct pci_bus *bus, unsigned int devfn,
+				       int offset, int len, u32 val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = macrisc_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		out_8((u8 *)addr, val);
+		(void) in_8((u8 *)addr);
+		break;
+	case 2:
+		out_le16((u16 *)addr, val);
+		(void) in_le16((u16 *)addr);
+		break;
+	default:
+		out_le32((u32 *)addr, val);
+		(void) in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static struct pci_ops macrisc_pci_ops =
+{
+	macrisc_read_config,
+	macrisc_write_config
+};
+
+#ifdef CONFIG_PPC32
+/*
+ * Verify that a specific (bus, dev_fn) exists on chaos
+ */
+static int
+chaos_validate_dev(struct pci_bus *bus, int devfn, int offset)
+{
+	struct device_node *np;
+	u32 *vendor, *device;
+
+	np = pci_busdev_to_OF_node(bus, devfn);
+	if (np == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	vendor = (u32 *)get_property(np, "vendor-id", NULL);
+	device = (u32 *)get_property(np, "device-id", NULL);
+	if (vendor == NULL || device == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	if ((*vendor == 0x106b) && (*device == 3) && (offset >= 0x10)
+	    && (offset != 0x14) && (offset != 0x18) && (offset <= 0x24))
+		return PCIBIOS_BAD_REGISTER_NUMBER;
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static int
+chaos_read_config(struct pci_bus *bus, unsigned int devfn, int offset,
+		  int len, u32 *val)
+{
+	int result = chaos_validate_dev(bus, devfn, offset);
+	if (result == PCIBIOS_BAD_REGISTER_NUMBER)
+		*val = ~0U;
+	if (result != PCIBIOS_SUCCESSFUL)
+		return result;
+	return macrisc_read_config(bus, devfn, offset, len, val);
+}
+
+static int
+chaos_write_config(struct pci_bus *bus, unsigned int devfn, int offset,
+		   int len, u32 val)
+{
+	int result = chaos_validate_dev(bus, devfn, offset);
+	if (result != PCIBIOS_SUCCESSFUL)
+		return result;
+	return macrisc_write_config(bus, devfn, offset, len, val);
+}
+
+static struct pci_ops chaos_pci_ops =
+{
+	chaos_read_config,
+	chaos_write_config
+};
+
+static void __init setup_chaos(struct pci_controller *hose,
+			       struct reg_property *addr)
+{
+	/* assume a `chaos' bridge */
+	hose->ops = &chaos_pci_ops;
+	hose->cfg_addr = ioremap(addr->address + 0x800000, 0x1000);
+	hose->cfg_data = ioremap(addr->address + 0xc00000, 0x1000);
+}
+#else
+#define setup_chaos(hose, addr)
+#endif /* CONFIG_PPC32 */
+
+#ifdef CONFIG_PPC64
+/*
+ * These versions of U3 HyperTransport config space access ops do not
+ * implement self-view of the HT host yet
+ */
+
+/*
+ * This function deals with some "special cases" devices.
+ *
+ *  0 -> No special case
+ *  1 -> Skip the device but act as if the access was successfull
+ *       (return 0xff's on reads, eventually, cache config space
+ *       accesses in a later version)
+ * -1 -> Hide the device (unsuccessful acess)
+ */
+static int u3_ht_skip_device(struct pci_controller *hose,
+			     struct pci_bus *bus, unsigned int devfn)
+{
+	struct device_node *busdn, *dn;
+	int i;
+
+	/* We only allow config cycles to devices that are in OF device-tree
+	 * as we are apparently having some weird things going on with some
+	 * revs of K2 on recent G5s
+	 */
+	if (bus->self)
+		busdn = pci_device_to_OF_node(bus->self);
+	else
+		busdn = hose->arch_data;
+	for (dn = busdn->child; dn; dn = dn->sibling)
+		if (dn->data && PCI_DN(dn)->devfn == devfn)
+			break;
+	if (dn == NULL)
+		return -1;
+
+	/*
+	 * When a device in K2 is powered down, we die on config
+	 * cycle accesses. Fix that here.
+	 */
+	for (i=0; i<2; i++)
+		if (k2_skiplist[i] == dn)
+			return 1;
+
+	return 0;
+}
+
+#define U3_HT_CFA0(devfn, off)		\
+		((((unsigned long)devfn) << 8) | offset)
+#define U3_HT_CFA1(bus, devfn, off)	\
+		(U3_HT_CFA0(devfn, off) \
+		+ (((unsigned long)bus) << 16) \
+		+ 0x01000000UL)
+
+static unsigned long u3_ht_cfg_access(struct pci_controller* hose,
+					     u8 bus, u8 devfn, u8 offset)
+{
+	if (bus == hose->first_busno) {
+		/* For now, we don't self probe U3 HT bridge */
+		if (PCI_SLOT(devfn) == 0)
+			return 0;
+		return ((unsigned long)hose->cfg_data) + U3_HT_CFA0(devfn, offset);
+	} else
+		return ((unsigned long)hose->cfg_data) + U3_HT_CFA1(bus, devfn, offset);
+}
+
+static int u3_ht_read_config(struct pci_bus *bus, unsigned int devfn,
+				    int offset, int len, u32 *val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = u3_ht_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	switch (u3_ht_skip_device(hose, bus, devfn)) {
+	case 0:
+		break;
+	case 1:
+		switch (len) {
+		case 1:
+			*val = 0xff; break;
+		case 2:
+			*val = 0xffff; break;
+		default:
+			*val = 0xfffffffful; break;
+		}
+		return PCIBIOS_SUCCESSFUL;
+	default:
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	}
+
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		*val = in_8((u8 *)addr);
+		break;
+	case 2:
+		*val = in_le16((u16 *)addr);
+		break;
+	default:
+		*val = in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static int u3_ht_write_config(struct pci_bus *bus, unsigned int devfn,
+				     int offset, int len, u32 val)
+{
+	struct pci_controller *hose;
+	unsigned long addr;
+
+	hose = pci_bus_to_host(bus);
+	if (hose == NULL)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	addr = u3_ht_cfg_access(hose, bus->number, devfn, offset);
+	if (!addr)
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	switch (u3_ht_skip_device(hose, bus, devfn)) {
+	case 0:
+		break;
+	case 1:
+		return PCIBIOS_SUCCESSFUL;
+	default:
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	}
+
+	/*
+	 * Note: the caller has already checked that offset is
+	 * suitably aligned and that len is 1, 2 or 4.
+	 */
+	switch (len) {
+	case 1:
+		out_8((u8 *)addr, val);
+		(void) in_8((u8 *)addr);
+		break;
+	case 2:
+		out_le16((u16 *)addr, val);
+		(void) in_le16((u16 *)addr);
+		break;
+	default:
+		out_le32((u32 *)addr, val);
+		(void) in_le32((u32 *)addr);
+		break;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static struct pci_ops u3_ht_pci_ops =
+{
+	u3_ht_read_config,
+	u3_ht_write_config
+};
+#endif /* CONFIG_PPC64 */
+
+#ifdef CONFIG_PPC32
+/*
+ * For a bandit bridge, turn on cache coherency if necessary.
+ * N.B. we could clean this up using the hose ops directly.
+ */
+static void __init init_bandit(struct pci_controller *bp)
+{
+	unsigned int vendev, magic;
+	int rev;
+
+	/* read the word at offset 0 in config space for device 11 */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + PCI_VENDOR_ID);
+	udelay(2);
+	vendev = in_le32(bp->cfg_data);
+	if (vendev == (PCI_DEVICE_ID_APPLE_BANDIT << 16) +
+			PCI_VENDOR_ID_APPLE) {
+		/* read the revision id */
+		out_le32(bp->cfg_addr,
+			 (1UL << BANDIT_DEVNUM) + PCI_REVISION_ID);
+		udelay(2);
+		rev = in_8(bp->cfg_data);
+		if (rev != BANDIT_REVID)
+			printk(KERN_WARNING
+			       "Unknown revision %d for bandit\n", rev);
+	} else if (vendev != (BANDIT_DEVID_2 << 16) + PCI_VENDOR_ID_APPLE) {
+		printk(KERN_WARNING "bandit isn't? (%x)\n", vendev);
+		return;
+	}
+
+	/* read the word at offset 0x50 */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + BANDIT_MAGIC);
+	udelay(2);
+	magic = in_le32(bp->cfg_data);
+	if ((magic & BANDIT_COHERENT) != 0)
+		return;
+	magic |= BANDIT_COHERENT;
+	udelay(2);
+	out_le32(bp->cfg_data, magic);
+	printk(KERN_INFO "Cache coherency enabled for bandit/PSX\n");
+}
+
+/*
+ * Tweak the PCI-PCI bridge chip on the blue & white G3s.
+ */
+static void __init init_p2pbridge(void)
+{
+	struct device_node *p2pbridge;
+	struct pci_controller* hose;
+	u8 bus, devfn;
+	u16 val;
+
+	/* XXX it would be better here to identify the specific
+	   PCI-PCI bridge chip we have. */
+	if ((p2pbridge = find_devices("pci-bridge")) == 0
+	    || p2pbridge->parent == NULL
+	    || strcmp(p2pbridge->parent->name, "pci") != 0)
+		return;
+	if (pci_device_from_OF_node(p2pbridge, &bus, &devfn) < 0) {
+		DBG("Can't find PCI infos for PCI<->PCI bridge\n");
+		return;
+	}
+	/* Warning: At this point, we have not yet renumbered all busses.
+	 * So we must use OF walking to find out hose
+	 */
+	hose = pci_find_hose_for_OF_device(p2pbridge);
+	if (!hose) {
+		DBG("Can't find hose for PCI<->PCI bridge\n");
+		return;
+	}
+	if (early_read_config_word(hose, bus, devfn,
+				   PCI_BRIDGE_CONTROL, &val) < 0) {
+		printk(KERN_ERR "init_p2pbridge: couldn't read bridge control\n");
+		return;
+	}
+	val &= ~PCI_BRIDGE_CTL_MASTER_ABORT;
+	early_write_config_word(hose, bus, devfn, PCI_BRIDGE_CONTROL, val);
+}
+
+/*
+ * Some Apple desktop machines have a NEC PD720100A USB2 controller
+ * on the motherboard. Open Firmware, on these, will disable the
+ * EHCI part of it so it behaves like a pair of OHCI's. This fixup
+ * code re-enables it ;)
+ */
+static void __init fixup_nec_usb2(void)
+{
+	struct device_node *nec;
+
+	for (nec = NULL; (nec = of_find_node_by_name(nec, "usb")) != NULL;) {
+		struct pci_controller *hose;
+		u32 data, *prop;
+		u8 bus, devfn;
+
+		prop = (u32 *)get_property(nec, "vendor-id", NULL);
+		if (prop == NULL)
+			continue;
+		if (0x1033 != *prop)
+			continue;
+		prop = (u32 *)get_property(nec, "device-id", NULL);
+		if (prop == NULL)
+			continue;
+		if (0x0035 != *prop)
+			continue;
+		prop = (u32 *)get_property(nec, "reg", NULL);
+		if (prop == NULL)
+			continue;
+		devfn = (prop[0] >> 8) & 0xff;
+		bus = (prop[0] >> 16) & 0xff;
+		if (PCI_FUNC(devfn) != 0)
+			continue;
+		hose = pci_find_hose_for_OF_device(nec);
+		if (!hose)
+			continue;
+		early_read_config_dword(hose, bus, devfn, 0xe4, &data);
+		if (data & 1UL) {
+			printk("Found NEC PD720100A USB2 chip with disabled EHCI, fixing up...\n");
+			data &= ~1UL;
+			early_write_config_dword(hose, bus, devfn, 0xe4, data);
+			early_write_config_byte(hose, bus, devfn | 2, PCI_INTERRUPT_LINE,
+				nec->intrs[0].line);
+		}
+	}
+}
+
+static void __init setup_bandit(struct pci_controller *hose,
+				struct reg_property *addr)
+{
+	hose->ops = &macrisc_pci_ops;
+	hose->cfg_addr = ioremap(addr->address + 0x800000, 0x1000);
+	hose->cfg_data = ioremap(addr->address + 0xc00000, 0x1000);
+	init_bandit(hose);
+}
+
+static int __init setup_uninorth(struct pci_controller *hose,
+				 struct reg_property *addr)
+{
+	pci_assign_all_buses = 1;
+	has_uninorth = 1;
+	hose->ops = &macrisc_pci_ops;
+	hose->cfg_addr = ioremap(addr->address + 0x800000, 0x1000);
+	hose->cfg_data = ioremap(addr->address + 0xc00000, 0x1000);
+	/* We "know" that the bridge at f2000000 has the PCI slots. */
+	return addr->address == 0xf2000000;
+}
+#endif
+
+#ifdef CONFIG_PPC64
+static void __init setup_u3_agp(struct pci_controller* hose)
+{
+	/* On G5, we move AGP up to high bus number so we don't need
+	 * to reassign bus numbers for HT. If we ever have P2P bridges
+	 * on AGP, we'll have to move pci_assign_all_busses to the
+	 * pci_controller structure so we enable it for AGP and not for
+	 * HT childs.
+	 * We hard code the address because of the different size of
+	 * the reg address cell, we shall fix that by killing struct
+	 * reg_property and using some accessor functions instead
+	 */
+	hose->first_busno = 0xf0;
+	hose->last_busno = 0xff;
+	has_uninorth = 1;
+	hose->ops = &macrisc_pci_ops;
+	hose->cfg_addr = ioremap(0xf0000000 + 0x800000, 0x1000);
+	hose->cfg_data = ioremap(0xf0000000 + 0xc00000, 0x1000);
+
+	u3_agp = hose;
+}
+
+static void __init setup_u3_ht(struct pci_controller* hose)
+{
+	struct device_node *np = (struct device_node *)hose->arch_data;
+	int i, cur;
+
+	hose->ops = &u3_ht_pci_ops;
+
+	/* We hard code the address because of the different size of
+	 * the reg address cell, we shall fix that by killing struct
+	 * reg_property and using some accessor functions instead
+	 */
+	hose->cfg_data = (volatile unsigned char *)ioremap(0xf2000000, 0x02000000);
+
+	/*
+	 * /ht node doesn't expose a "ranges" property, so we "remove" regions that
+	 * have been allocated to AGP. So far, this version of the code doesn't assign
+	 * any of the 0xfxxxxxxx "fine" memory regions to /ht.
+	 * We need to fix that sooner or later by either parsing all child "ranges"
+	 * properties or figuring out the U3 address space decoding logic and
+	 * then read its configuration register (if any).
+	 */
+	hose->io_base_phys = 0xf4000000;
+	hose->pci_io_size = 0x00400000;
+	hose->io_resource.name = np->full_name;
+	hose->io_resource.start = 0;
+	hose->io_resource.end = 0x003fffff;
+	hose->io_resource.flags = IORESOURCE_IO;
+	hose->pci_mem_offset = 0;
+	hose->first_busno = 0;
+	hose->last_busno = 0xef;
+	hose->mem_resources[0].name = np->full_name;
+	hose->mem_resources[0].start = 0x80000000;
+	hose->mem_resources[0].end = 0xefffffff;
+	hose->mem_resources[0].flags = IORESOURCE_MEM;
+
+	u3_ht = hose;
+
+	if (u3_agp == NULL) {
+		DBG("U3 has no AGP, using full resource range\n");
+		return;
+	}
+
+	/* We "remove" the AGP resources from the resources allocated to HT, that
+	 * is we create "holes". However, that code does assumptions that so far
+	 * happen to be true (cross fingers...), typically that resources in the
+	 * AGP node are properly ordered
+	 */
+	cur = 0;
+	for (i=0; i<3; i++) {
+		struct resource *res = &u3_agp->mem_resources[i];
+		if (res->flags != IORESOURCE_MEM)
+			continue;
+		/* We don't care about "fine" resources */
+		if (res->start >= 0xf0000000)
+			continue;
+		/* Check if it's just a matter of "shrinking" us in one direction */
+		if (hose->mem_resources[cur].start == res->start) {
+			DBG("U3/HT: shrink start of %d, %08lx -> %08lx\n",
+			    cur, hose->mem_resources[cur].start, res->end + 1);
+			hose->mem_resources[cur].start = res->end + 1;
+			continue;
+		}
+		if (hose->mem_resources[cur].end == res->end) {
+			DBG("U3/HT: shrink end of %d, %08lx -> %08lx\n",
+			    cur, hose->mem_resources[cur].end, res->start - 1);
+			hose->mem_resources[cur].end = res->start - 1;
+			continue;
+		}
+		/* No, it's not the case, we need a hole */
+		if (cur == 2) {
+			/* not enough resources for a hole, we drop part of the range */
+			printk(KERN_WARNING "Running out of resources for /ht host !\n");
+			hose->mem_resources[cur].end = res->start - 1;
+			continue;
+		}
+		cur++;
+		DBG("U3/HT: hole, %d end at %08lx, %d start at %08lx\n",
+		    cur-1, res->start - 1, cur, res->end + 1);
+		hose->mem_resources[cur].name = np->full_name;
+		hose->mem_resources[cur].flags = IORESOURCE_MEM;
+		hose->mem_resources[cur].start = res->end + 1;
+		hose->mem_resources[cur].end = hose->mem_resources[cur-1].end;
+		hose->mem_resources[cur-1].end = res->start - 1;
+	}
+}
+
+/* XXX this needs to be converged between ppc32 and ppc64... */
+static struct pci_controller * __init pcibios_alloc_controller(void)
+{
+	struct pci_controller *hose;
+
+	hose = alloc_bootmem(sizeof(struct pci_controller));
+	if (hose)
+		pci_setup_pci_controller(hose);
+	return hose;
+}
+#endif
+
+/*
+ * We assume that if we have a G3 powermac, we have one bridge called
+ * "pci" (a MPC106) and no bandit or chaos bridges, and contrariwise,
+ * if we have one or more bandit or chaos bridges, we don't have a MPC106.
+ */
+static int __init add_bridge(struct device_node *dev)
+{
+	int len;
+	struct pci_controller *hose;
+#ifdef CONFIG_PPC32
+	struct reg_property *addr;
+#endif
+	char *disp_name;
+	int *bus_range;
+	int primary = 1;
+
+	DBG("Adding PCI host bridge %s\n", dev->full_name);
+
+#ifdef CONFIG_PPC32
+	/* XXX fix this */
+	addr = (struct reg_property *) get_property(dev, "reg", &len);
+	if (addr == NULL || len < sizeof(*addr)) {
+		printk(KERN_WARNING "Can't use %s: no address\n",
+		       dev->full_name);
+		return -ENODEV;
+	}
+#endif
+	bus_range = (int *) get_property(dev, "bus-range", &len);
+	if (bus_range == NULL || len < 2 * sizeof(int)) {
+		printk(KERN_WARNING "Can't get bus-range for %s, assume bus 0\n",
+			       dev->full_name);
+	}
+
+	hose = pcibios_alloc_controller();
+	if (!hose)
+		return -ENOMEM;
+	hose->arch_data = dev;
+	hose->first_busno = bus_range ? bus_range[0] : 0;
+	hose->last_busno = bus_range ? bus_range[1] : 0xff;
+
+	disp_name = NULL;
+#ifdef CONFIG_POWER4
+	if (device_is_compatible(dev, "u3-agp")) {
+		setup_u3_agp(hose);
+		disp_name = "U3-AGP";
+		primary = 0;
+	} else if (device_is_compatible(dev, "u3-ht")) {
+		setup_u3_ht(hose);
+		disp_name = "U3-HT";
+		primary = 1;
+	}
+	printk(KERN_INFO "Found %s PCI host bridge.  Firmware bus number: %d->%d\n",
+		disp_name, hose->first_busno, hose->last_busno);
+#else
+	if (device_is_compatible(dev, "uni-north")) {
+		primary = setup_uninorth(hose, addr);
+		disp_name = "UniNorth";
+	} else if (strcmp(dev->name, "pci") == 0) {
+		/* XXX assume this is a mpc106 (grackle) */
+		setup_grackle(hose);
+		disp_name = "Grackle (MPC106)";
+	} else if (strcmp(dev->name, "bandit") == 0) {
+		setup_bandit(hose, addr);
+		disp_name = "Bandit";
+	} else if (strcmp(dev->name, "chaos") == 0) {
+		setup_chaos(hose, addr);
+		disp_name = "Chaos";
+		primary = 0;
+	}
+	printk(KERN_INFO "Found %s PCI host bridge at 0x%08lx. Firmware bus number: %d->%d\n",
+		disp_name, addr->address, hose->first_busno, hose->last_busno);
+#endif
+	DBG(" ->Hose at 0x%p, cfg_addr=0x%p,cfg_data=0x%p\n",
+		hose, hose->cfg_addr, hose->cfg_data);
+
+	/* Interpret the "ranges" property */
+	/* This also maps the I/O region and sets isa_io/mem_base */
+	pci_process_bridge_OF_ranges(hose, dev, primary);
+
+	/* Fixup "bus-range" OF property */
+	fixup_bus_range(dev);
+
+	return 0;
+}
+
+static void __init
+pcibios_fixup_OF_interrupts(void)
+{
+	struct pci_dev* dev = NULL;
+
+	/*
+	 * Open Firmware often doesn't initialize the
+	 * PCI_INTERRUPT_LINE config register properly, so we
+	 * should find the device node and apply the interrupt
+	 * obtained from the OF device-tree
+	 */
+	for_each_pci_dev(dev) {
+		struct device_node *node;
+		node = pci_device_to_OF_node(dev);
+		/* this is the node, see if it has interrupts */
+		if (node && node->n_intrs > 0)
+			dev->irq = node->intrs[0].line;
+		pci_write_config_byte(dev, PCI_INTERRUPT_LINE, dev->irq);
+	}
+}
+
+void __init
+pmac_pcibios_fixup(void)
+{
+	/* Fixup interrupts according to OF tree */
+	pcibios_fixup_OF_interrupts();
+}
+
+#ifdef CONFIG_PPC64
+static void __init pmac_fixup_phb_resources(void)
+{
+	struct pci_controller *hose, *tmp;
+
+	list_for_each_entry_safe(hose, tmp, &hose_list, list_node) {
+		printk(KERN_INFO "PCI Host %d, io start: %lx; io end: %lx\n",
+		       hose->global_number,
+		       hose->io_resource.start, hose->io_resource.end);
+	}
+}
+#endif
+
+void __init pmac_pci_init(void)
+{
+	struct device_node *np, *root;
+	struct device_node *ht = NULL;
+
+	root = of_find_node_by_path("/");
+	if (root == NULL) {
+		printk(KERN_CRIT "pmac_pci_init: can't find root "
+		       "of device tree\n");
+		return;
+	}
+	for (np = NULL; (np = of_get_next_child(root, np)) != NULL;) {
+		if (np->name == NULL)
+			continue;
+		if (strcmp(np->name, "bandit") == 0
+		    || strcmp(np->name, "chaos") == 0
+		    || strcmp(np->name, "pci") == 0) {
+			if (add_bridge(np) == 0)
+				of_node_get(np);
+		}
+		if (strcmp(np->name, "ht") == 0) {
+			of_node_get(np);
+			ht = np;
+		}
+	}
+	of_node_put(root);
+
+#ifdef CONFIG_PPC64
+	/* Probe HT last as it relies on the agp resources to be already
+	 * setup
+	 */
+	if (ht && add_bridge(ht) != 0)
+		of_node_put(ht);
+
+	/*
+	 * We need to call pci_setup_phb_io for the HT bridge first
+	 * so it gets the I/O port numbers starting at 0, and we
+	 * need to call it for the AGP bridge after that so it gets
+	 * small positive I/O port numbers.
+	 */
+	if (u3_ht)
+		pci_setup_phb_io(u3_ht, 1);
+	if (u3_agp)
+		pci_setup_phb_io(u3_agp, 0);
+
+	/*
+	 * On ppc64, fixup the IO resources on our host bridges as
+	 * the common code does it only for children of the host bridges
+	 */
+	pmac_fixup_phb_resources();
+
+	/* Setup the linkage between OF nodes and PHBs */
+	pci_devs_phb_init();
+
+	/* Fixup the PCI<->OF mapping for U3 AGP due to bus renumbering. We
+	 * assume there is no P2P bridge on the AGP bus, which should be a
+	 * safe assumptions hopefully.
+	 */
+	if (u3_agp) {
+		struct device_node *np = u3_agp->arch_data;
+		PCI_DN(np)->busno = 0xf0;
+		for (np = np->child; np; np = np->sibling)
+			PCI_DN(np)->busno = 0xf0;
+	}
+
+	/* map in PCI I/O space */
+	phbs_remap_io();
+
+	/* pmac_check_ht_link(); */
+
+	/* Tell pci.c to not use the common resource allocation mechanism */
+	pci_probe_only = 1;
+
+	/* Allow all IO */
+	io_page_mask = -1;
+
+#else /* CONFIG_PPC64 */
+	init_p2pbridge();
+	fixup_nec_usb2();
+
+	/* We are still having some issues with the Xserve G4, enabling
+	 * some offset between bus number and domains for now when we
+	 * assign all busses should help for now
+	 */
+	if (pci_assign_all_buses)
+		pcibios_assign_bus_offset = 0x10;
+#endif
+}
+
+int
+pmac_pci_enable_device_hook(struct pci_dev *dev, int initial)
+{
+	struct device_node* node;
+	int updatecfg = 0;
+	int uninorth_child;
+
+	node = pci_device_to_OF_node(dev);
+
+	/* We don't want to enable USB controllers absent from the OF tree
+	 * (iBook second controller)
+	 */
+	if (dev->vendor == PCI_VENDOR_ID_APPLE
+	    && (dev->class == ((PCI_CLASS_SERIAL_USB << 8) | 0x10))
+	    && !node) {
+		printk(KERN_INFO "Apple USB OHCI %s disabled by firmware\n",
+		       pci_name(dev));
+		return -EINVAL;
+	}
+
+	if (!node)
+		return 0;
+
+	uninorth_child = node->parent &&
+		device_is_compatible(node->parent, "uni-north");
+
+	/* Firewire & GMAC were disabled after PCI probe, the driver is
+	 * claiming them, we must re-enable them now.
+	 */
+	if (uninorth_child && !strcmp(node->name, "firewire") &&
+	    (device_is_compatible(node, "pci106b,18") ||
+	     device_is_compatible(node, "pci106b,30") ||
+	     device_is_compatible(node, "pci11c1,5811"))) {
+		pmac_call_feature(PMAC_FTR_1394_CABLE_POWER, node, 0, 1);
+		pmac_call_feature(PMAC_FTR_1394_ENABLE, node, 0, 1);
+		updatecfg = 1;
+	}
+	if (uninorth_child && !strcmp(node->name, "ethernet") &&
+	    device_is_compatible(node, "gmac")) {
+		pmac_call_feature(PMAC_FTR_GMAC_ENABLE, node, 0, 1);
+		updatecfg = 1;
+	}
+
+	if (updatecfg) {
+		u16 cmd;
+
+		/*
+		 * Make sure PCI is correctly configured
+		 *
+		 * We use old pci_bios versions of the function since, by
+		 * default, gmac is not powered up, and so will be absent
+		 * from the kernel initial PCI lookup.
+		 *
+		 * Should be replaced by 2.4 new PCI mechanisms and really
+		 * register the device.
+		 */
+		pci_read_config_word(dev, PCI_COMMAND, &cmd);
+		cmd |= PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER
+			| PCI_COMMAND_INVALIDATE;
+		pci_write_config_word(dev, PCI_COMMAND, cmd);
+		pci_write_config_byte(dev, PCI_LATENCY_TIMER, 16);
+		pci_write_config_byte(dev, PCI_CACHE_LINE_SIZE,
+				      L1_CACHE_BYTES >> 2);
+	}
+
+	return 0;
+}
+
+/* We power down some devices after they have been probed. They'll
+ * be powered back on later on
+ */
+void __init pmac_pcibios_after_init(void)
+{
+	struct device_node* nd;
+
+#ifdef CONFIG_BLK_DEV_IDE
+	struct pci_dev *dev = NULL;
+
+	/* OF fails to initialize IDE controllers on macs
+	 * (and maybe other machines)
+	 *
+	 * Ideally, this should be moved to the IDE layer, but we need
+	 * to check specifically with Andre Hedrick how to do it cleanly
+	 * since the common IDE code seem to care about the fact that the
+	 * BIOS may have disabled a controller.
+	 *
+	 * -- BenH
+	 */
+	for_each_pci_dev(dev) {
+		if ((dev->class >> 16) == PCI_BASE_CLASS_STORAGE)
+			pci_enable_device(dev);
+	}
+#endif /* CONFIG_BLK_DEV_IDE */
+
+	nd = find_devices("firewire");
+	while (nd) {
+		if (nd->parent && (device_is_compatible(nd, "pci106b,18") ||
+				   device_is_compatible(nd, "pci106b,30") ||
+				   device_is_compatible(nd, "pci11c1,5811"))
+		    && device_is_compatible(nd->parent, "uni-north")) {
+			pmac_call_feature(PMAC_FTR_1394_ENABLE, nd, 0, 0);
+			pmac_call_feature(PMAC_FTR_1394_CABLE_POWER, nd, 0, 0);
+		}
+		nd = nd->next;
+	}
+	nd = find_devices("ethernet");
+	while (nd) {
+		if (nd->parent && device_is_compatible(nd, "gmac")
+		    && device_is_compatible(nd->parent, "uni-north"))
+			pmac_call_feature(PMAC_FTR_GMAC_ENABLE, nd, 0, 0);
+		nd = nd->next;
+	}
+}
+
+#ifdef CONFIG_PPC32
+void pmac_pci_fixup_cardbus(struct pci_dev* dev)
+{
+	if (_machine != _MACH_Pmac)
+		return;
+	/*
+	 * Fix the interrupt routing on the various cardbus bridges
+	 * used on powerbooks
+	 */
+	if (dev->vendor != PCI_VENDOR_ID_TI)
+		return;
+	if (dev->device == PCI_DEVICE_ID_TI_1130 ||
+	    dev->device == PCI_DEVICE_ID_TI_1131) {
+		u8 val;
+		/* Enable PCI interrupt */
+		if (pci_read_config_byte(dev, 0x91, &val) == 0)
+			pci_write_config_byte(dev, 0x91, val | 0x30);
+		/* Disable ISA interrupt mode */
+		if (pci_read_config_byte(dev, 0x92, &val) == 0)
+			pci_write_config_byte(dev, 0x92, val & ~0x06);
+	}
+	if (dev->device == PCI_DEVICE_ID_TI_1210 ||
+	    dev->device == PCI_DEVICE_ID_TI_1211 ||
+	    dev->device == PCI_DEVICE_ID_TI_1410 ||
+	    dev->device == PCI_DEVICE_ID_TI_1510) {
+		u8 val;
+		/* 0x8c == TI122X_IRQMUX, 2 says to route the INTA
+		   signal out the MFUNC0 pin */
+		if (pci_read_config_byte(dev, 0x8c, &val) == 0)
+			pci_write_config_byte(dev, 0x8c, (val & ~0x0f) | 2);
+		/* Disable ISA interrupt mode */
+		if (pci_read_config_byte(dev, 0x92, &val) == 0)
+			pci_write_config_byte(dev, 0x92, val & ~0x06);
+	}
+}
+
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_TI, PCI_ANY_ID, pmac_pci_fixup_cardbus);
+
+void pmac_pci_fixup_pciata(struct pci_dev* dev)
+{
+       u8 progif = 0;
+
+       /*
+        * On PowerMacs, we try to switch any PCI ATA controller to
+	* fully native mode
+        */
+	if (_machine != _MACH_Pmac)
+		return;
+	/* Some controllers don't have the class IDE */
+	if (dev->vendor == PCI_VENDOR_ID_PROMISE)
+		switch(dev->device) {
+		case PCI_DEVICE_ID_PROMISE_20246:
+		case PCI_DEVICE_ID_PROMISE_20262:
+		case PCI_DEVICE_ID_PROMISE_20263:
+		case PCI_DEVICE_ID_PROMISE_20265:
+		case PCI_DEVICE_ID_PROMISE_20267:
+		case PCI_DEVICE_ID_PROMISE_20268:
+		case PCI_DEVICE_ID_PROMISE_20269:
+		case PCI_DEVICE_ID_PROMISE_20270:
+		case PCI_DEVICE_ID_PROMISE_20271:
+		case PCI_DEVICE_ID_PROMISE_20275:
+		case PCI_DEVICE_ID_PROMISE_20276:
+		case PCI_DEVICE_ID_PROMISE_20277:
+			goto good;
+		}
+	/* Others, check PCI class */
+	if ((dev->class >> 8) != PCI_CLASS_STORAGE_IDE)
+		return;
+ good:
+	pci_read_config_byte(dev, PCI_CLASS_PROG, &progif);
+	if ((progif & 5) != 5) {
+		printk(KERN_INFO "Forcing PCI IDE into native mode: %s\n", pci_name(dev));
+		(void) pci_write_config_byte(dev, PCI_CLASS_PROG, progif|5);
+		if (pci_read_config_byte(dev, PCI_CLASS_PROG, &progif) ||
+		    (progif & 5) != 5)
+			printk(KERN_ERR "Rewrite of PROGIF failed !\n");
+	}
+}
+DECLARE_PCI_FIXUP_FINAL(PCI_ANY_ID, PCI_ANY_ID, pmac_pci_fixup_pciata);
+#endif
+
+/*
+ * Disable second function on K2-SATA, it's broken
+ * and disable IO BARs on first one
+ */
+static void fixup_k2_sata(struct pci_dev* dev)
+{
+	int i;
+	u16 cmd;
+
+	if (PCI_FUNC(dev->devfn) > 0) {
+		pci_read_config_word(dev, PCI_COMMAND, &cmd);
+		cmd &= ~(PCI_COMMAND_IO | PCI_COMMAND_MEMORY);
+		pci_write_config_word(dev, PCI_COMMAND, cmd);
+		for (i = 0; i < 6; i++) {
+			dev->resource[i].start = dev->resource[i].end = 0;
+			dev->resource[i].flags = 0;
+			pci_write_config_dword(dev, PCI_BASE_ADDRESS_0 + 4 * i, 0);
+		}
+	} else {
+		pci_read_config_word(dev, PCI_COMMAND, &cmd);
+		cmd &= ~PCI_COMMAND_IO;
+		pci_write_config_word(dev, PCI_COMMAND, cmd);
+		for (i = 0; i < 5; i++) {
+			dev->resource[i].start = dev->resource[i].end = 0;
+			dev->resource[i].flags = 0;
+			pci_write_config_dword(dev, PCI_BASE_ADDRESS_0 + 4 * i, 0);
+		}
+	}
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_SERVERWORKS, 0x0240, fixup_k2_sata);
+
diff --git a/arch/powerpc/platforms/powermac/pic.c b/arch/powerpc/platforms/powermac/pic.c
new file mode 100644
index 000000000000..83a49e80ac29
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/pic.c
@@ -0,0 +1,678 @@
+/*
+ *  Support for the interrupt controllers found on Power Macintosh,
+ *  currently Apple's "Grand Central" interrupt controller in all
+ *  it's incarnations. OpenPIC support used on newer machines is
+ *  in a separate file
+ *
+ *  Copyright (C) 1997 Paul Mackerras (paulus@samba.org)
+ *
+ *  Maintained by Benjamin Herrenschmidt (benh@kernel.crashing.org)
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/stddef.h>
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/signal.h>
+#include <linux/pci.h>
+#include <linux/interrupt.h>
+#include <linux/sysdev.h>
+#include <linux/adb.h>
+#include <linux/pmu.h>
+#include <linux/module.h>
+
+#include <asm/sections.h>
+#include <asm/io.h>
+#include <asm/smp.h>
+#include <asm/prom.h>
+#include <asm/pci-bridge.h>
+#include <asm/time.h>
+#include <asm/pmac_feature.h>
+#include <asm/mpic.h>
+
+#include "pmac.h"
+
+/*
+ * XXX this should be in xmon.h, but putting it there means xmon.h
+ * has to include <linux/interrupt.h> (to get irqreturn_t), which
+ * causes all sorts of problems.  -- paulus
+ */
+extern irqreturn_t xmon_irq(int, void *, struct pt_regs *);
+
+#ifdef CONFIG_PPC32
+struct pmac_irq_hw {
+        unsigned int    event;
+        unsigned int    enable;
+        unsigned int    ack;
+        unsigned int    level;
+};
+
+/* Default addresses */
+static volatile struct pmac_irq_hw *pmac_irq_hw[4] = {
+        (struct pmac_irq_hw *) 0xf3000020,
+        (struct pmac_irq_hw *) 0xf3000010,
+        (struct pmac_irq_hw *) 0xf4000020,
+        (struct pmac_irq_hw *) 0xf4000010,
+};
+
+#define GC_LEVEL_MASK		0x3ff00000
+#define OHARE_LEVEL_MASK	0x1ff00000
+#define HEATHROW_LEVEL_MASK	0x1ff00000
+
+static int max_irqs;
+static int max_real_irqs;
+static u32 level_mask[4];
+
+static DEFINE_SPINLOCK(pmac_pic_lock);
+
+#define GATWICK_IRQ_POOL_SIZE        10
+static struct interrupt_info gatwick_int_pool[GATWICK_IRQ_POOL_SIZE];
+
+/*
+ * Mark an irq as "lost".  This is only used on the pmac
+ * since it can lose interrupts (see pmac_set_irq_mask).
+ * -- Cort
+ */
+void
+__set_lost(unsigned long irq_nr, int nokick)
+{
+	if (!test_and_set_bit(irq_nr, ppc_lost_interrupts)) {
+		atomic_inc(&ppc_n_lost_interrupts);
+		if (!nokick)
+			set_dec(1);
+	}
+}
+
+static void
+pmac_mask_and_ack_irq(unsigned int irq_nr)
+{
+        unsigned long bit = 1UL << (irq_nr & 0x1f);
+        int i = irq_nr >> 5;
+        unsigned long flags;
+
+        if ((unsigned)irq_nr >= max_irqs)
+                return;
+
+        clear_bit(irq_nr, ppc_cached_irq_mask);
+        if (test_and_clear_bit(irq_nr, ppc_lost_interrupts))
+                atomic_dec(&ppc_n_lost_interrupts);
+	spin_lock_irqsave(&pmac_pic_lock, flags);
+        out_le32(&pmac_irq_hw[i]->enable, ppc_cached_irq_mask[i]);
+        out_le32(&pmac_irq_hw[i]->ack, bit);
+        do {
+                /* make sure ack gets to controller before we enable
+                   interrupts */
+                mb();
+        } while((in_le32(&pmac_irq_hw[i]->enable) & bit)
+                != (ppc_cached_irq_mask[i] & bit));
+	spin_unlock_irqrestore(&pmac_pic_lock, flags);
+}
+
+static void pmac_set_irq_mask(unsigned int irq_nr, int nokicklost)
+{
+        unsigned long bit = 1UL << (irq_nr & 0x1f);
+        int i = irq_nr >> 5;
+        unsigned long flags;
+
+        if ((unsigned)irq_nr >= max_irqs)
+                return;
+
+	spin_lock_irqsave(&pmac_pic_lock, flags);
+        /* enable unmasked interrupts */
+        out_le32(&pmac_irq_hw[i]->enable, ppc_cached_irq_mask[i]);
+
+        do {
+                /* make sure mask gets to controller before we
+                   return to user */
+                mb();
+        } while((in_le32(&pmac_irq_hw[i]->enable) & bit)
+                != (ppc_cached_irq_mask[i] & bit));
+
+        /*
+         * Unfortunately, setting the bit in the enable register
+         * when the device interrupt is already on *doesn't* set
+         * the bit in the flag register or request another interrupt.
+         */
+        if (bit & ppc_cached_irq_mask[i] & in_le32(&pmac_irq_hw[i]->level))
+		__set_lost((ulong)irq_nr, nokicklost);
+	spin_unlock_irqrestore(&pmac_pic_lock, flags);
+}
+
+/* When an irq gets requested for the first client, if it's an
+ * edge interrupt, we clear any previous one on the controller
+ */
+static unsigned int pmac_startup_irq(unsigned int irq_nr)
+{
+        unsigned long bit = 1UL << (irq_nr & 0x1f);
+        int i = irq_nr >> 5;
+
+	if ((irq_desc[irq_nr].status & IRQ_LEVEL) == 0)
+		out_le32(&pmac_irq_hw[i]->ack, bit);
+        set_bit(irq_nr, ppc_cached_irq_mask);
+        pmac_set_irq_mask(irq_nr, 0);
+
+	return 0;
+}
+
+static void pmac_mask_irq(unsigned int irq_nr)
+{
+        clear_bit(irq_nr, ppc_cached_irq_mask);
+        pmac_set_irq_mask(irq_nr, 0);
+        mb();
+}
+
+static void pmac_unmask_irq(unsigned int irq_nr)
+{
+        set_bit(irq_nr, ppc_cached_irq_mask);
+        pmac_set_irq_mask(irq_nr, 0);
+}
+
+static void pmac_end_irq(unsigned int irq_nr)
+{
+	if (!(irq_desc[irq_nr].status & (IRQ_DISABLED|IRQ_INPROGRESS))
+	    && irq_desc[irq_nr].action) {
+        	set_bit(irq_nr, ppc_cached_irq_mask);
+	        pmac_set_irq_mask(irq_nr, 1);
+	}
+}
+
+
+struct hw_interrupt_type pmac_pic = {
+	.typename	= " PMAC-PIC ",
+	.startup	= pmac_startup_irq,
+	.enable		= pmac_unmask_irq,
+	.disable	= pmac_mask_irq,
+	.ack		= pmac_mask_and_ack_irq,
+	.end		= pmac_end_irq,
+};
+
+struct hw_interrupt_type gatwick_pic = {
+	.typename	= " GATWICK  ",
+	.startup	= pmac_startup_irq,
+	.enable		= pmac_unmask_irq,
+	.disable	= pmac_mask_irq,
+	.ack		= pmac_mask_and_ack_irq,
+	.end		= pmac_end_irq,
+};
+
+static irqreturn_t gatwick_action(int cpl, void *dev_id, struct pt_regs *regs)
+{
+	int irq, bits;
+
+	for (irq = max_irqs; (irq -= 32) >= max_real_irqs; ) {
+		int i = irq >> 5;
+		bits = in_le32(&pmac_irq_hw[i]->event) | ppc_lost_interrupts[i];
+		/* We must read level interrupts from the level register */
+		bits |= (in_le32(&pmac_irq_hw[i]->level) & level_mask[i]);
+		bits &= ppc_cached_irq_mask[i];
+		if (bits == 0)
+			continue;
+		irq += __ilog2(bits);
+		__do_IRQ(irq, regs);
+		return IRQ_HANDLED;
+	}
+	printk("gatwick irq not from gatwick pic\n");
+	return IRQ_NONE;
+}
+
+int
+pmac_get_irq(struct pt_regs *regs)
+{
+	int irq;
+	unsigned long bits = 0;
+
+#ifdef CONFIG_SMP
+	void psurge_smp_message_recv(struct pt_regs *);
+
+       	/* IPI's are a hack on the powersurge -- Cort */
+       	if ( smp_processor_id() != 0 ) {
+		psurge_smp_message_recv(regs);
+		return -2;	/* ignore, already handled */
+        }
+#endif /* CONFIG_SMP */
+	for (irq = max_real_irqs; (irq -= 32) >= 0; ) {
+		int i = irq >> 5;
+		bits = in_le32(&pmac_irq_hw[i]->event) | ppc_lost_interrupts[i];
+		/* We must read level interrupts from the level register */
+		bits |= (in_le32(&pmac_irq_hw[i]->level) & level_mask[i]);
+		bits &= ppc_cached_irq_mask[i];
+		if (bits == 0)
+			continue;
+		irq += __ilog2(bits);
+		break;
+	}
+
+	return irq;
+}
+
+/* This routine will fix some missing interrupt values in the device tree
+ * on the gatwick mac-io controller used by some PowerBooks
+ */
+static void __init
+pmac_fix_gatwick_interrupts(struct device_node *gw, int irq_base)
+{
+	struct device_node *node;
+	int count;
+
+	memset(gatwick_int_pool, 0, sizeof(gatwick_int_pool));
+	node = gw->child;
+	count = 0;
+	while(node)
+	{
+		/* Fix SCC */
+		if (strcasecmp(node->name, "escc") == 0)
+			if (node->child) {
+				if (node->child->n_intrs < 3) {
+					node->child->intrs = &gatwick_int_pool[count];
+					count += 3;
+				}
+				node->child->n_intrs = 3;
+				node->child->intrs[0].line = 15+irq_base;
+				node->child->intrs[1].line =  4+irq_base;
+				node->child->intrs[2].line =  5+irq_base;
+				printk(KERN_INFO "irq: fixed SCC on second controller (%d,%d,%d)\n",
+					node->child->intrs[0].line,
+					node->child->intrs[1].line,
+					node->child->intrs[2].line);
+			}
+		/* Fix media-bay & left SWIM */
+		if (strcasecmp(node->name, "media-bay") == 0) {
+			struct device_node* ya_node;
+
+			if (node->n_intrs == 0)
+				node->intrs = &gatwick_int_pool[count++];
+			node->n_intrs = 1;
+			node->intrs[0].line = 29+irq_base;
+			printk(KERN_INFO "irq: fixed media-bay on second controller (%d)\n",
+					node->intrs[0].line);
+
+			ya_node = node->child;
+			while(ya_node)
+			{
+				if (strcasecmp(ya_node->name, "floppy") == 0) {
+					if (ya_node->n_intrs < 2) {
+						ya_node->intrs = &gatwick_int_pool[count];
+						count += 2;
+					}
+					ya_node->n_intrs = 2;
+					ya_node->intrs[0].line = 19+irq_base;
+					ya_node->intrs[1].line =  1+irq_base;
+					printk(KERN_INFO "irq: fixed floppy on second controller (%d,%d)\n",
+						ya_node->intrs[0].line, ya_node->intrs[1].line);
+				}
+				if (strcasecmp(ya_node->name, "ata4") == 0) {
+					if (ya_node->n_intrs < 2) {
+						ya_node->intrs = &gatwick_int_pool[count];
+						count += 2;
+					}
+					ya_node->n_intrs = 2;
+					ya_node->intrs[0].line = 14+irq_base;
+					ya_node->intrs[1].line =  3+irq_base;
+					printk(KERN_INFO "irq: fixed ide on second controller (%d,%d)\n",
+						ya_node->intrs[0].line, ya_node->intrs[1].line);
+				}
+				ya_node = ya_node->sibling;
+			}
+		}
+		node = node->sibling;
+	}
+	if (count > 10) {
+		printk("WARNING !! Gatwick interrupt pool overflow\n");
+		printk("  GATWICK_IRQ_POOL_SIZE = %d\n", GATWICK_IRQ_POOL_SIZE);
+		printk("              requested = %d\n", count);
+	}
+}
+
+/*
+ * The PowerBook 3400/2400/3500 can have a combo ethernet/modem
+ * card which includes an ohare chip that acts as a second interrupt
+ * controller.  If we find this second ohare, set it up and fix the
+ * interrupt value in the device tree for the ethernet chip.
+ */
+static int __init enable_second_ohare(void)
+{
+	unsigned char bus, devfn;
+	unsigned short cmd;
+        unsigned long addr;
+	struct device_node *irqctrler = find_devices("pci106b,7");
+	struct device_node *ether;
+
+	if (irqctrler == NULL || irqctrler->n_addrs <= 0)
+		return -1;
+	addr = (unsigned long) ioremap(irqctrler->addrs[0].address, 0x40);
+	pmac_irq_hw[1] = (volatile struct pmac_irq_hw *)(addr + 0x20);
+	max_irqs = 64;
+	if (pci_device_from_OF_node(irqctrler, &bus, &devfn) == 0) {
+		struct pci_controller* hose = pci_find_hose_for_OF_device(irqctrler);
+		if (!hose)
+		    printk(KERN_ERR "Can't find PCI hose for OHare2 !\n");
+		else {
+		    early_read_config_word(hose, bus, devfn, PCI_COMMAND, &cmd);
+		    cmd |= PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER;
+	  	    cmd &= ~PCI_COMMAND_IO;
+		    early_write_config_word(hose, bus, devfn, PCI_COMMAND, cmd);
+		}
+	}
+
+	/* Fix interrupt for the modem/ethernet combo controller. The number
+	   in the device tree (27) is bogus (correct for the ethernet-only
+	   board but not the combo ethernet/modem board).
+	   The real interrupt is 28 on the second controller -> 28+32 = 60.
+	*/
+	ether = find_devices("pci1011,14");
+	if (ether && ether->n_intrs > 0) {
+		ether->intrs[0].line = 60;
+		printk(KERN_INFO "irq: Fixed ethernet IRQ to %d\n",
+		       ether->intrs[0].line);
+	}
+
+	/* Return the interrupt number of the cascade */
+	return irqctrler->intrs[0].line;
+}
+
+#ifdef CONFIG_XMON
+static struct irqaction xmon_action = {
+	.handler	= xmon_irq,
+	.flags		= 0,
+	.mask		= CPU_MASK_NONE,
+	.name		= "NMI - XMON"
+};
+#endif
+
+static struct irqaction gatwick_cascade_action = {
+	.handler	= gatwick_action,
+	.flags		= SA_INTERRUPT,
+	.mask		= CPU_MASK_NONE,
+	.name		= "cascade",
+};
+#endif /* CONFIG_PPC32 */
+
+static int pmac_u3_cascade(struct pt_regs *regs, void *data)
+{
+	return mpic_get_one_irq((struct mpic *)data, regs);
+}
+
+void __init pmac_pic_init(void)
+{
+        struct device_node *irqctrler  = NULL;
+        struct device_node *irqctrler2 = NULL;
+	struct device_node *np;
+#ifdef CONFIG_PPC32
+        int i;
+        unsigned long addr;
+	int irq_cascade = -1;
+#endif
+	struct mpic *mpic1, *mpic2;
+
+	/* We first try to detect Apple's new Core99 chipset, since mac-io
+	 * is quite different on those machines and contains an IBM MPIC2.
+	 */
+	np = find_type_devices("open-pic");
+	while (np) {
+		if (np->parent && !strcmp(np->parent->name, "u3"))
+			irqctrler2 = np;
+		else
+			irqctrler = np;
+		np = np->next;
+	}
+	if (irqctrler != NULL && irqctrler->n_addrs > 0) {
+		unsigned char senses[128];
+
+		printk(KERN_INFO "PowerMac using OpenPIC irq controller at 0x%08x\n",
+		       (unsigned int)irqctrler->addrs[0].address);
+		pmac_call_feature(PMAC_FTR_ENABLE_MPIC, irqctrler, 0, 0);
+
+		prom_get_irq_senses(senses, 0, 128);
+		mpic1 = mpic_alloc(irqctrler->addrs[0].address,
+				   MPIC_PRIMARY | MPIC_WANTS_RESET,
+				   0, 0, 128, 252, senses, 128, " OpenPIC  ");
+		BUG_ON(mpic1 == NULL);
+		mpic_init(mpic1);		
+
+		if (irqctrler2 != NULL && irqctrler2->n_intrs > 0 &&
+		    irqctrler2->n_addrs > 0) {
+			printk(KERN_INFO "Slave OpenPIC at 0x%08x hooked on IRQ %d\n",
+			       (u32)irqctrler2->addrs[0].address,
+			       irqctrler2->intrs[0].line);
+
+			pmac_call_feature(PMAC_FTR_ENABLE_MPIC, irqctrler2, 0, 0);
+			prom_get_irq_senses(senses, 128, 128 + 124);
+
+			/* We don't need to set MPIC_BROKEN_U3 here since we don't have
+			 * hypertransport interrupts routed to it
+			 */
+			mpic2 = mpic_alloc(irqctrler2->addrs[0].address,
+					   MPIC_BIG_ENDIAN | MPIC_WANTS_RESET,
+					   0, 128, 124, 0, senses, 124,
+					   " U3-MPIC  ");
+			BUG_ON(mpic2 == NULL);
+			mpic_init(mpic2);
+			mpic_setup_cascade(irqctrler2->intrs[0].line,
+					   pmac_u3_cascade, mpic2);
+		}
+#if defined(CONFIG_XMON) && defined(CONFIG_PPC32)
+		{
+			struct device_node* pswitch;
+			int nmi_irq;
+
+			pswitch = find_devices("programmer-switch");
+			if (pswitch && pswitch->n_intrs) {
+				nmi_irq = pswitch->intrs[0].line;
+				mpic_irq_set_priority(nmi_irq, 9);
+				setup_irq(nmi_irq, &xmon_action);
+			}
+		}
+#endif	/* CONFIG_XMON */
+		return;
+	}
+	irqctrler = NULL;
+
+#ifdef CONFIG_PPC32
+	/* Get the level/edge settings, assume if it's not
+	 * a Grand Central nor an OHare, then it's an Heathrow
+	 * (or Paddington).
+	 */
+	ppc_md.get_irq = pmac_get_irq;
+	if (find_devices("gc"))
+		level_mask[0] = GC_LEVEL_MASK;
+	else if (find_devices("ohare")) {
+		level_mask[0] = OHARE_LEVEL_MASK;
+		/* We might have a second cascaded ohare */
+		level_mask[1] = OHARE_LEVEL_MASK;
+	} else {
+		level_mask[0] = HEATHROW_LEVEL_MASK;
+		level_mask[1] = 0;
+		/* We might have a second cascaded heathrow */
+		level_mask[2] = HEATHROW_LEVEL_MASK;
+		level_mask[3] = 0;
+	}
+
+	/*
+	 * G3 powermacs and 1999 G3 PowerBooks have 64 interrupts,
+	 * 1998 G3 Series PowerBooks have 128,
+	 * other powermacs have 32.
+	 * The combo ethernet/modem card for the Powerstar powerbooks
+	 * (2400/3400/3500, ohare based) has a second ohare chip
+	 * effectively making a total of 64.
+	 */
+	max_irqs = max_real_irqs = 32;
+	irqctrler = find_devices("mac-io");
+	if (irqctrler)
+	{
+		max_real_irqs = 64;
+		if (irqctrler->next)
+			max_irqs = 128;
+		else
+			max_irqs = 64;
+	}
+	for ( i = 0; i < max_real_irqs ; i++ )
+		irq_desc[i].handler = &pmac_pic;
+
+	/* get addresses of first controller */
+	if (irqctrler) {
+		if  (irqctrler->n_addrs > 0) {
+			addr = (unsigned long)
+				ioremap(irqctrler->addrs[0].address, 0x40);
+			for (i = 0; i < 2; ++i)
+				pmac_irq_hw[i] = (volatile struct pmac_irq_hw*)
+					(addr + (2 - i) * 0x10);
+		}
+
+		/* get addresses of second controller */
+		irqctrler = irqctrler->next;
+		if (irqctrler && irqctrler->n_addrs > 0) {
+			addr = (unsigned long)
+				ioremap(irqctrler->addrs[0].address, 0x40);
+			for (i = 2; i < 4; ++i)
+				pmac_irq_hw[i] = (volatile struct pmac_irq_hw*)
+					(addr + (4 - i) * 0x10);
+			irq_cascade = irqctrler->intrs[0].line;
+			if (device_is_compatible(irqctrler, "gatwick"))
+				pmac_fix_gatwick_interrupts(irqctrler, max_real_irqs);
+		}
+	} else {
+		/* older powermacs have a GC (grand central) or ohare at
+		   f3000000, with interrupt control registers at f3000020. */
+		addr = (unsigned long) ioremap(0xf3000000, 0x40);
+		pmac_irq_hw[0] = (volatile struct pmac_irq_hw *) (addr + 0x20);
+	}
+
+	/* PowerBooks 3400 and 3500 can have a second controller in a second
+	   ohare chip, on the combo ethernet/modem card */
+	if (machine_is_compatible("AAPL,3400/2400")
+	     || machine_is_compatible("AAPL,3500"))
+		irq_cascade = enable_second_ohare();
+
+	/* disable all interrupts in all controllers */
+	for (i = 0; i * 32 < max_irqs; ++i)
+		out_le32(&pmac_irq_hw[i]->enable, 0);
+	/* mark level interrupts */
+	for (i = 0; i < max_irqs; i++)
+		if (level_mask[i >> 5] & (1UL << (i & 0x1f)))
+			irq_desc[i].status = IRQ_LEVEL;
+
+	/* get interrupt line of secondary interrupt controller */
+	if (irq_cascade >= 0) {
+		printk(KERN_INFO "irq: secondary controller on irq %d\n",
+			(int)irq_cascade);
+		for ( i = max_real_irqs ; i < max_irqs ; i++ )
+			irq_desc[i].handler = &gatwick_pic;
+		setup_irq(irq_cascade, &gatwick_cascade_action);
+	}
+	printk("System has %d possible interrupts\n", max_irqs);
+	if (max_irqs != max_real_irqs)
+		printk(KERN_DEBUG "%d interrupts on main controller\n",
+			max_real_irqs);
+
+#ifdef CONFIG_XMON
+	setup_irq(20, &xmon_action);
+#endif	/* CONFIG_XMON */
+#endif	/* CONFIG_PPC32 */
+}
+
+#if defined(CONFIG_PM) && defined(CONFIG_PPC32)
+/*
+ * These procedures are used in implementing sleep on the powerbooks.
+ * sleep_save_intrs() saves the states of all interrupt enables
+ * and disables all interrupts except for the nominated one.
+ * sleep_restore_intrs() restores the states of all interrupt enables.
+ */
+unsigned long sleep_save_mask[2];
+
+/* This used to be passed by the PMU driver but that link got
+ * broken with the new driver model. We use this tweak for now...
+ */
+static int pmacpic_find_viaint(void)
+{
+	int viaint = -1;
+
+#ifdef CONFIG_ADB_PMU
+	struct device_node *np;
+
+	if (pmu_get_model() != PMU_OHARE_BASED)
+		goto not_found;
+	np = of_find_node_by_name(NULL, "via-pmu");
+	if (np == NULL)
+		goto not_found;
+	viaint = np->intrs[0].line;
+#endif /* CONFIG_ADB_PMU */
+
+not_found:
+	return viaint;
+}
+
+static int pmacpic_suspend(struct sys_device *sysdev, pm_message_t state)
+{
+	int viaint = pmacpic_find_viaint();
+
+	sleep_save_mask[0] = ppc_cached_irq_mask[0];
+	sleep_save_mask[1] = ppc_cached_irq_mask[1];
+	ppc_cached_irq_mask[0] = 0;
+	ppc_cached_irq_mask[1] = 0;
+	if (viaint > 0)
+		set_bit(viaint, ppc_cached_irq_mask);
+	out_le32(&pmac_irq_hw[0]->enable, ppc_cached_irq_mask[0]);
+	if (max_real_irqs > 32)
+		out_le32(&pmac_irq_hw[1]->enable, ppc_cached_irq_mask[1]);
+	(void)in_le32(&pmac_irq_hw[0]->event);
+	/* make sure mask gets to controller before we return to caller */
+	mb();
+        (void)in_le32(&pmac_irq_hw[0]->enable);
+
+        return 0;
+}
+
+static int pmacpic_resume(struct sys_device *sysdev)
+{
+	int i;
+
+	out_le32(&pmac_irq_hw[0]->enable, 0);
+	if (max_real_irqs > 32)
+		out_le32(&pmac_irq_hw[1]->enable, 0);
+	mb();
+	for (i = 0; i < max_real_irqs; ++i)
+		if (test_bit(i, sleep_save_mask))
+			pmac_unmask_irq(i);
+
+	return 0;
+}
+
+#endif /* CONFIG_PM && CONFIG_PPC32 */
+
+static struct sysdev_class pmacpic_sysclass = {
+	set_kset_name("pmac_pic"),
+};
+
+static struct sys_device device_pmacpic = {
+	.id		= 0,
+	.cls		= &pmacpic_sysclass,
+};
+
+static struct sysdev_driver driver_pmacpic = {
+#if defined(CONFIG_PM) && defined(CONFIG_PPC32)
+	.suspend	= &pmacpic_suspend,
+	.resume		= &pmacpic_resume,
+#endif /* CONFIG_PM && CONFIG_PPC32 */
+};
+
+static int __init init_pmacpic_sysfs(void)
+{
+#ifdef CONFIG_PPC32
+	if (max_irqs == 0)
+		return -ENODEV;
+#endif
+	printk(KERN_DEBUG "Registering pmac pic with sysfs...\n");
+	sysdev_class_register(&pmacpic_sysclass);
+	sysdev_register(&device_pmacpic);
+	sysdev_driver_register(&pmacpic_sysclass, &driver_pmacpic);
+	return 0;
+}
+
+subsys_initcall(init_pmacpic_sysfs);
+
diff --git a/arch/powerpc/platforms/powermac/pic.h b/arch/powerpc/platforms/powermac/pic.h
new file mode 100644
index 000000000000..664103dfeef9
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/pic.h
@@ -0,0 +1,11 @@
+#ifndef __PPC_PLATFORMS_PMAC_PIC_H
+#define __PPC_PLATFORMS_PMAC_PIC_H
+
+#include <linux/irq.h>
+
+extern struct hw_interrupt_type pmac_pic;
+
+void pmac_pic_init(void);
+int pmac_get_irq(struct pt_regs *regs);
+
+#endif /* __PPC_PLATFORMS_PMAC_PIC_H */
diff --git a/arch/powerpc/platforms/powermac/pmac.h b/arch/powerpc/platforms/powermac/pmac.h
new file mode 100644
index 000000000000..2ad25e13423e
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/pmac.h
@@ -0,0 +1,51 @@
+#ifndef __PMAC_H__
+#define __PMAC_H__
+
+#include <linux/pci.h>
+#include <linux/ide.h>
+#include <linux/irq.h>
+
+/*
+ * Declaration for the various functions exported by the
+ * pmac_* files. Mostly for use by pmac_setup
+ */
+
+struct rtc_time;
+
+extern long pmac_time_init(void);
+extern unsigned long pmac_get_boot_time(void);
+extern void pmac_get_rtc_time(struct rtc_time *);
+extern int pmac_set_rtc_time(struct rtc_time *);
+extern void pmac_read_rtc_time(void);
+extern void pmac_calibrate_decr(void);
+extern void pmac_pcibios_fixup(void);
+extern void pmac_pci_init(void);
+extern unsigned long pmac_ide_get_base(int index);
+extern void pmac_ide_init_hwif_ports(hw_regs_t *hw,
+	unsigned long data_port, unsigned long ctrl_port, int *irq);
+
+extern void pmac_nvram_update(void);
+extern unsigned char pmac_nvram_read_byte(int addr);
+extern void pmac_nvram_write_byte(int addr, unsigned char val);
+extern int pmac_pci_enable_device_hook(struct pci_dev *dev, int initial);
+extern void pmac_pcibios_after_init(void);
+extern int of_show_percpuinfo(struct seq_file *m, int i);
+
+extern void pmac_pci_init(void);
+extern void pmac_setup_pci_dma(void);
+extern void pmac_check_ht_link(void);
+
+extern void pmac_setup_smp(void);
+
+extern unsigned long pmac_ide_get_base(int index);
+extern void pmac_ide_init_hwif_ports(hw_regs_t *hw,
+	unsigned long data_port, unsigned long ctrl_port, int *irq);
+
+extern int pmac_nvram_init(void);
+
+extern struct hw_interrupt_type pmac_pic;
+
+void pmac_pic_init(void);
+int pmac_get_irq(struct pt_regs *regs);
+
+#endif /* __PMAC_H__ */
diff --git a/arch/powerpc/platforms/powermac/setup.c b/arch/powerpc/platforms/powermac/setup.c
new file mode 100644
index 000000000000..80b58c1ec412
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/setup.c
@@ -0,0 +1,795 @@
+/*
+ *  Powermac setup and early boot code plus other random bits.
+ *
+ *  PowerPC version
+ *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
+ *
+ *  Adapted for Power Macintosh by Paul Mackerras
+ *    Copyright (C) 1996 Paul Mackerras (paulus@samba.org)
+ *
+ *  Derived from "arch/alpha/kernel/setup.c"
+ *    Copyright (C) 1995 Linus Torvalds
+ *
+ *  Maintained by Benjamin Herrenschmidt (benh@kernel.crashing.org)
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ */
+
+/*
+ * bootup setup stuff..
+ */
+
+#include <linux/config.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/slab.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/string.h>
+#include <linux/delay.h>
+#include <linux/ioport.h>
+#include <linux/major.h>
+#include <linux/initrd.h>
+#include <linux/vt_kern.h>
+#include <linux/console.h>
+#include <linux/ide.h>
+#include <linux/pci.h>
+#include <linux/adb.h>
+#include <linux/cuda.h>
+#include <linux/pmu.h>
+#include <linux/irq.h>
+#include <linux/seq_file.h>
+#include <linux/root_dev.h>
+#include <linux/bitops.h>
+#include <linux/suspend.h>
+
+#include <asm/reg.h>
+#include <asm/sections.h>
+#include <asm/prom.h>
+#include <asm/system.h>
+#include <asm/pgtable.h>
+#include <asm/io.h>
+#include <asm/pci-bridge.h>
+#include <asm/ohare.h>
+#include <asm/mediabay.h>
+#include <asm/machdep.h>
+#include <asm/dma.h>
+#include <asm/cputable.h>
+#include <asm/btext.h>
+#include <asm/pmac_feature.h>
+#include <asm/time.h>
+#include <asm/of_device.h>
+#include <asm/mmu_context.h>
+#include <asm/iommu.h>
+#include <asm/smu.h>
+#include <asm/pmc.h>
+#include <asm/mpic.h>
+#include <asm/lmb.h>
+
+#include "pmac.h"
+
+#undef SHOW_GATWICK_IRQS
+
+unsigned char drive_info;
+
+int ppc_override_l2cr = 0;
+int ppc_override_l2cr_value;
+int has_l2cache = 0;
+
+int pmac_newworld = 1;
+
+static int current_root_goodness = -1;
+
+extern int pmac_newworld;
+extern struct machdep_calls pmac_md;
+
+#define DEFAULT_ROOT_DEVICE Root_SDA1	/* sda1 - slightly silly choice */
+
+#ifdef CONFIG_PPC64
+#include <asm/udbg.h>
+int sccdbg;
+#endif
+
+extern void zs_kgdb_hook(int tty_num);
+
+sys_ctrler_t sys_ctrler = SYS_CTRLER_UNKNOWN;
+EXPORT_SYMBOL(sys_ctrler);
+
+#ifdef CONFIG_PMAC_SMU
+unsigned long smu_cmdbuf_abs;
+EXPORT_SYMBOL(smu_cmdbuf_abs);
+#endif
+
+#ifdef CONFIG_SMP
+extern struct smp_ops_t psurge_smp_ops;
+extern struct smp_ops_t core99_smp_ops;
+#endif /* CONFIG_SMP */
+
+static void pmac_show_cpuinfo(struct seq_file *m)
+{
+	struct device_node *np;
+	char *pp;
+	int plen;
+	int mbmodel;
+	unsigned int mbflags;
+	char* mbname;
+
+	mbmodel = pmac_call_feature(PMAC_FTR_GET_MB_INFO, NULL,
+				    PMAC_MB_INFO_MODEL, 0);
+	mbflags = pmac_call_feature(PMAC_FTR_GET_MB_INFO, NULL,
+				    PMAC_MB_INFO_FLAGS, 0);
+	if (pmac_call_feature(PMAC_FTR_GET_MB_INFO, NULL, PMAC_MB_INFO_NAME,
+			      (long) &mbname) != 0)
+		mbname = "Unknown";
+
+	/* find motherboard type */
+	seq_printf(m, "machine\t\t: ");
+	np = of_find_node_by_path("/");
+	if (np != NULL) {
+		pp = (char *) get_property(np, "model", NULL);
+		if (pp != NULL)
+			seq_printf(m, "%s\n", pp);
+		else
+			seq_printf(m, "PowerMac\n");
+		pp = (char *) get_property(np, "compatible", &plen);
+		if (pp != NULL) {
+			seq_printf(m, "motherboard\t:");
+			while (plen > 0) {
+				int l = strlen(pp) + 1;
+				seq_printf(m, " %s", pp);
+				plen -= l;
+				pp += l;
+			}
+			seq_printf(m, "\n");
+		}
+		of_node_put(np);
+	} else
+		seq_printf(m, "PowerMac\n");
+
+	/* print parsed model */
+	seq_printf(m, "detected as\t: %d (%s)\n", mbmodel, mbname);
+	seq_printf(m, "pmac flags\t: %08x\n", mbflags);
+
+	/* find l2 cache info */
+	np = of_find_node_by_name(NULL, "l2-cache");
+	if (np == NULL)
+		np = of_find_node_by_type(NULL, "cache");
+	if (np != NULL) {
+		unsigned int *ic = (unsigned int *)
+			get_property(np, "i-cache-size", NULL);
+		unsigned int *dc = (unsigned int *)
+			get_property(np, "d-cache-size", NULL);
+		seq_printf(m, "L2 cache\t:");
+		has_l2cache = 1;
+		if (get_property(np, "cache-unified", NULL) != 0 && dc) {
+			seq_printf(m, " %dK unified", *dc / 1024);
+		} else {
+			if (ic)
+				seq_printf(m, " %dK instruction", *ic / 1024);
+			if (dc)
+				seq_printf(m, "%s %dK data",
+					   (ic? " +": ""), *dc / 1024);
+		}
+		pp = get_property(np, "ram-type", NULL);
+		if (pp)
+			seq_printf(m, " %s", pp);
+		seq_printf(m, "\n");
+		of_node_put(np);
+	}
+
+	/* Indicate newworld/oldworld */
+	seq_printf(m, "pmac-generation\t: %s\n",
+		   pmac_newworld ? "NewWorld" : "OldWorld");
+}
+
+static void pmac_show_percpuinfo(struct seq_file *m, int i)
+{
+#ifdef CONFIG_CPU_FREQ_PMAC
+	extern unsigned int pmac_get_one_cpufreq(int i);
+	unsigned int freq = pmac_get_one_cpufreq(i);
+	if (freq != 0) {
+		seq_printf(m, "clock\t\t: %dMHz\n", freq/1000);
+		return;
+	}
+#endif /* CONFIG_CPU_FREQ_PMAC */
+}
+
+#ifndef CONFIG_ADB_CUDA
+int find_via_cuda(void)
+{
+	if (!find_devices("via-cuda"))
+		return 0;
+	printk("WARNING ! Your machine is CUDA-based but your kernel\n");
+	printk("          wasn't compiled with CONFIG_ADB_CUDA option !\n");
+	return 0;
+}
+#endif
+
+#ifndef CONFIG_ADB_PMU
+int find_via_pmu(void)
+{
+	if (!find_devices("via-pmu"))
+		return 0;
+	printk("WARNING ! Your machine is PMU-based but your kernel\n");
+	printk("          wasn't compiled with CONFIG_ADB_PMU option !\n");
+	return 0;
+}
+#endif
+
+#ifndef CONFIG_PMAC_SMU
+int smu_init(void)
+{
+	/* should check and warn if SMU is present */
+	return 0;
+}
+#endif
+
+#ifdef CONFIG_PPC32
+static volatile u32 *sysctrl_regs;
+
+static void __init ohare_init(void)
+{
+	/* this area has the CPU identification register
+	   and some registers used by smp boards */
+	sysctrl_regs = (volatile u32 *) ioremap(0xf8000000, 0x1000);
+
+	/*
+	 * Turn on the L2 cache.
+	 * We assume that we have a PSX memory controller iff
+	 * we have an ohare I/O controller.
+	 */
+	if (find_devices("ohare") != NULL) {
+		if (((sysctrl_regs[2] >> 24) & 0xf) >= 3) {
+			if (sysctrl_regs[4] & 0x10)
+				sysctrl_regs[4] |= 0x04000020;
+			else
+				sysctrl_regs[4] |= 0x04000000;
+			if(has_l2cache)
+				printk(KERN_INFO "Level 2 cache enabled\n");
+		}
+	}
+}
+
+static void __init l2cr_init(void)
+{
+	/* Checks "l2cr-value" property in the registry */
+	if (cpu_has_feature(CPU_FTR_L2CR)) {
+		struct device_node *np = find_devices("cpus");
+		if (np == 0)
+			np = find_type_devices("cpu");
+		if (np != 0) {
+			unsigned int *l2cr = (unsigned int *)
+				get_property(np, "l2cr-value", NULL);
+			if (l2cr != 0) {
+				ppc_override_l2cr = 1;
+				ppc_override_l2cr_value = *l2cr;
+				_set_L2CR(0);
+				_set_L2CR(ppc_override_l2cr_value);
+			}
+		}
+	}
+
+	if (ppc_override_l2cr)
+		printk(KERN_INFO "L2CR overridden (0x%x), "
+		       "backside cache is %s\n",
+		       ppc_override_l2cr_value,
+		       (ppc_override_l2cr_value & 0x80000000)
+				? "enabled" : "disabled");
+}
+#endif
+
+void __init pmac_setup_arch(void)
+{
+	struct device_node *cpu, *ic;
+	int *fp;
+	unsigned long pvr;
+
+	pvr = PVR_VER(mfspr(SPRN_PVR));
+
+	/* Set loops_per_jiffy to a half-way reasonable value,
+	   for use until calibrate_delay gets called. */
+	loops_per_jiffy = 50000000 / HZ;
+	cpu = of_find_node_by_type(NULL, "cpu");
+	if (cpu != NULL) {
+		fp = (int *) get_property(cpu, "clock-frequency", NULL);
+		if (fp != NULL) {
+			if (pvr >= 0x30 && pvr < 0x80)
+				/* PPC970 etc. */
+				loops_per_jiffy = *fp / (3 * HZ);
+			else if (pvr == 4 || pvr >= 8)
+				/* 604, G3, G4 etc. */
+				loops_per_jiffy = *fp / HZ;
+			else
+				/* 601, 603, etc. */
+				loops_per_jiffy = *fp / (2 * HZ);
+		}
+		of_node_put(cpu);
+	}
+
+	/* See if newworld or oldworld */
+	for (ic = NULL; (ic = of_find_all_nodes(ic)) != NULL; )
+		if (get_property(ic, "interrupt-controller", NULL))
+			break;
+	pmac_newworld = (ic != NULL);
+	if (ic)
+		of_node_put(ic);
+
+	/* Lookup PCI hosts */
+	pmac_pci_init();
+
+#ifdef CONFIG_PPC32
+	ohare_init();
+	l2cr_init();
+#endif /* CONFIG_PPC32 */
+
+#ifdef CONFIG_PPC64
+	/* Probe motherboard chipset */
+	/* this is done earlier in setup_arch for 32-bit */
+	pmac_feature_init();
+
+	/* We can NAP */
+	powersave_nap = 1;
+	printk(KERN_INFO "Using native/NAP idle loop\n");
+#endif
+
+#ifdef CONFIG_KGDB
+	zs_kgdb_hook(0);
+#endif
+
+	find_via_cuda();
+	find_via_pmu();
+	smu_init();
+
+#if defined(CONFIG_NVRAM) || defined(CONFIG_PPC64)
+	pmac_nvram_init();
+#endif
+
+#ifdef CONFIG_PPC32
+#ifdef CONFIG_BLK_DEV_INITRD
+	if (initrd_start)
+		ROOT_DEV = Root_RAM0;
+	else
+#endif
+		ROOT_DEV = DEFAULT_ROOT_DEVICE;
+#endif
+
+#ifdef CONFIG_SMP
+	/* Check for Core99 */
+	if (find_devices("uni-n") || find_devices("u3"))
+		smp_ops = &core99_smp_ops;
+#ifdef CONFIG_PPC32
+	else
+		smp_ops = &psurge_smp_ops;
+#endif
+#endif /* CONFIG_SMP */
+}
+
+char *bootpath;
+char *bootdevice;
+void *boot_host;
+int boot_target;
+int boot_part;
+extern dev_t boot_dev;
+
+#ifdef CONFIG_SCSI
+void __init note_scsi_host(struct device_node *node, void *host)
+{
+	int l;
+	char *p;
+
+	l = strlen(node->full_name);
+	if (bootpath != NULL && bootdevice != NULL
+	    && strncmp(node->full_name, bootdevice, l) == 0
+	    && (bootdevice[l] == '/' || bootdevice[l] == 0)) {
+		boot_host = host;
+		/*
+		 * There's a bug in OF 1.0.5.  (Why am I not surprised.)
+		 * If you pass a path like scsi/sd@1:0 to canon, it returns
+		 * something like /bandit@F2000000/gc@10/53c94@10000/sd@0,0
+		 * That is, the scsi target number doesn't get preserved.
+		 * So we pick the target number out of bootpath and use that.
+		 */
+		p = strstr(bootpath, "/sd@");
+		if (p != NULL) {
+			p += 4;
+			boot_target = simple_strtoul(p, NULL, 10);
+			p = strchr(p, ':');
+			if (p != NULL)
+				boot_part = simple_strtoul(p + 1, NULL, 10);
+		}
+	}
+}
+EXPORT_SYMBOL(note_scsi_host);
+#endif
+
+#if defined(CONFIG_BLK_DEV_IDE) && defined(CONFIG_BLK_DEV_IDE_PMAC)
+static dev_t __init find_ide_boot(void)
+{
+	char *p;
+	int n;
+	dev_t __init pmac_find_ide_boot(char *bootdevice, int n);
+
+	if (bootdevice == NULL)
+		return 0;
+	p = strrchr(bootdevice, '/');
+	if (p == NULL)
+		return 0;
+	n = p - bootdevice;
+
+	return pmac_find_ide_boot(bootdevice, n);
+}
+#endif /* CONFIG_BLK_DEV_IDE && CONFIG_BLK_DEV_IDE_PMAC */
+
+static void __init find_boot_device(void)
+{
+#if defined(CONFIG_BLK_DEV_IDE) && defined(CONFIG_BLK_DEV_IDE_PMAC)
+	boot_dev = find_ide_boot();
+#endif
+}
+
+/* TODO: Merge the suspend-to-ram with the common code !!!
+ * currently, this is a stub implementation for suspend-to-disk
+ * only
+ */
+
+#ifdef CONFIG_SOFTWARE_SUSPEND
+
+static int pmac_pm_prepare(suspend_state_t state)
+{
+	printk(KERN_DEBUG "%s(%d)\n", __FUNCTION__, state);
+
+	return 0;
+}
+
+static int pmac_pm_enter(suspend_state_t state)
+{
+	printk(KERN_DEBUG "%s(%d)\n", __FUNCTION__, state);
+
+	/* Giveup the lazy FPU & vec so we don't have to back them
+	 * up from the low level code
+	 */
+	enable_kernel_fp();
+
+#ifdef CONFIG_ALTIVEC
+	if (cur_cpu_spec->cpu_features & CPU_FTR_ALTIVEC)
+		enable_kernel_altivec();
+#endif /* CONFIG_ALTIVEC */
+
+	return 0;
+}
+
+static int pmac_pm_finish(suspend_state_t state)
+{
+	printk(KERN_DEBUG "%s(%d)\n", __FUNCTION__, state);
+
+	/* Restore userland MMU context */
+	set_context(current->active_mm->context, current->active_mm->pgd);
+
+	return 0;
+}
+
+static struct pm_ops pmac_pm_ops = {
+	.pm_disk_mode	= PM_DISK_SHUTDOWN,
+	.prepare	= pmac_pm_prepare,
+	.enter		= pmac_pm_enter,
+	.finish		= pmac_pm_finish,
+};
+
+#endif /* CONFIG_SOFTWARE_SUSPEND */
+
+static int initializing = 1;
+
+static int pmac_late_init(void)
+{
+	initializing = 0;
+#ifdef CONFIG_SOFTWARE_SUSPEND
+	pm_set_ops(&pmac_pm_ops);
+#endif /* CONFIG_SOFTWARE_SUSPEND */
+	return 0;
+}
+
+late_initcall(pmac_late_init);
+
+/* can't be __init - can be called whenever a disk is first accessed */
+void note_bootable_part(dev_t dev, int part, int goodness)
+{
+	static int found_boot = 0;
+	char *p;
+
+	if (!initializing)
+		return;
+	if ((goodness <= current_root_goodness) &&
+	    ROOT_DEV != DEFAULT_ROOT_DEVICE)
+		return;
+	p = strstr(saved_command_line, "root=");
+	if (p != NULL && (p == saved_command_line || p[-1] == ' '))
+		return;
+
+	if (!found_boot) {
+		find_boot_device();
+		found_boot = 1;
+	}
+	if (!boot_dev || dev == boot_dev) {
+		ROOT_DEV = dev + part;
+		boot_dev = 0;
+		current_root_goodness = goodness;
+	}
+}
+
+#ifdef CONFIG_ADB_CUDA
+static void cuda_restart(void)
+{
+	struct adb_request req;
+
+	cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_RESET_SYSTEM);
+	for (;;)
+		cuda_poll();
+}
+
+static void cuda_shutdown(void)
+{
+	struct adb_request req;
+
+	cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_POWERDOWN);
+	for (;;)
+		cuda_poll();
+}
+
+#else
+#define cuda_restart()
+#define cuda_shutdown()
+#endif
+
+#ifndef CONFIG_ADB_PMU
+#define pmu_restart()
+#define pmu_shutdown()
+#endif
+
+#ifndef CONFIG_PMAC_SMU
+#define smu_restart()
+#define smu_shutdown()
+#endif
+
+static void pmac_restart(char *cmd)
+{
+	switch (sys_ctrler) {
+	case SYS_CTRLER_CUDA:
+		cuda_restart();
+		break;
+	case SYS_CTRLER_PMU:
+		pmu_restart();
+		break;
+	case SYS_CTRLER_SMU:
+		smu_restart();
+		break;
+	default: ;
+	}
+}
+
+static void pmac_power_off(void)
+{
+	switch (sys_ctrler) {
+	case SYS_CTRLER_CUDA:
+		cuda_shutdown();
+		break;
+	case SYS_CTRLER_PMU:
+		pmu_shutdown();
+		break;
+	case SYS_CTRLER_SMU:
+		smu_shutdown();
+		break;
+	default: ;
+	}
+}
+
+static void
+pmac_halt(void)
+{
+	pmac_power_off();
+}
+
+#ifdef CONFIG_PPC32
+void __init pmac_init(void)
+{
+	/* isa_io_base gets set in pmac_pci_init */
+	isa_mem_base = PMAC_ISA_MEM_BASE;
+	pci_dram_offset = PMAC_PCI_DRAM_OFFSET;
+	ISA_DMA_THRESHOLD = ~0L;
+	DMA_MODE_READ = 1;
+	DMA_MODE_WRITE = 2;
+
+	ppc_md = pmac_md;
+
+#if defined(CONFIG_BLK_DEV_IDE) || defined(CONFIG_BLK_DEV_IDE_MODULE)
+#ifdef CONFIG_BLK_DEV_IDE_PMAC
+        ppc_ide_md.ide_init_hwif	= pmac_ide_init_hwif_ports;
+        ppc_ide_md.default_io_base	= pmac_ide_get_base;
+#endif /* CONFIG_BLK_DEV_IDE_PMAC */
+#endif /* defined(CONFIG_BLK_DEV_IDE) || defined(CONFIG_BLK_DEV_IDE_MODULE) */
+
+	if (ppc_md.progress) ppc_md.progress("pmac_init(): exit", 0);
+
+}
+#endif
+
+/* 
+ * Early initialization.
+ */
+static void __init pmac_init_early(void)
+{
+#ifdef CONFIG_PPC64
+	/* Initialize hash table, from now on, we can take hash faults
+	 * and call ioremap
+	 */
+	hpte_init_native();
+
+	/* Init SCC */
+	if (strstr(cmd_line, "sccdbg")) {
+		sccdbg = 1;
+		udbg_init_scc(NULL);
+	}
+
+	/* Setup interrupt mapping options */
+	ppc64_interrupt_controller = IC_OPEN_PIC;
+
+	iommu_init_early_u3();
+#endif
+}
+
+static void __init pmac_progress(char *s, unsigned short hex)
+{
+#ifdef CONFIG_PPC64
+	if (sccdbg) {
+		udbg_puts(s);
+		udbg_puts("\n");
+		return;
+	}
+#endif
+#ifdef CONFIG_BOOTX_TEXT
+	if (boot_text_mapped) {
+		btext_drawstring(s);
+		btext_drawchar('\n');
+	}
+#endif /* CONFIG_BOOTX_TEXT */
+}
+
+/*
+ * pmac has no legacy IO, anything calling this function has to
+ * fail or bad things will happen
+ */
+static int pmac_check_legacy_ioport(unsigned int baseport)
+{
+	return -ENODEV;
+}
+
+static int __init pmac_declare_of_platform_devices(void)
+{
+	struct device_node *np, *npp;
+
+	np = find_devices("uni-n");
+	if (np) {
+		for (np = np->child; np != NULL; np = np->sibling)
+			if (strncmp(np->name, "i2c", 3) == 0) {
+				of_platform_device_create(np, "uni-n-i2c",
+							  NULL);
+				break;
+			}
+	}
+	np = find_devices("valkyrie");
+	if (np)
+		of_platform_device_create(np, "valkyrie", NULL);
+	np = find_devices("platinum");
+	if (np)
+		of_platform_device_create(np, "platinum", NULL);
+
+	npp = of_find_node_by_name(NULL, "u3");
+	if (npp) {
+		for (np = NULL; (np = of_get_next_child(npp, np)) != NULL;) {
+			if (strncmp(np->name, "i2c", 3) == 0) {
+				of_platform_device_create(np, "u3-i2c", NULL);
+				of_node_put(np);
+				break;
+			}
+		}
+		of_node_put(npp);
+	}
+        np = of_find_node_by_type(NULL, "smu");
+        if (np) {
+		of_platform_device_create(np, "smu", NULL);
+		of_node_put(np);
+	}
+
+	return 0;
+}
+
+device_initcall(pmac_declare_of_platform_devices);
+
+/*
+ * Called very early, MMU is off, device-tree isn't unflattened
+ */
+static int __init pmac_probe(int platform)
+{
+#ifdef CONFIG_PPC64
+	if (platform != PLATFORM_POWERMAC)
+		return 0;
+
+	/*
+	 * On U3, the DART (iommu) must be allocated now since it
+	 * has an impact on htab_initialize (due to the large page it
+	 * occupies having to be broken up so the DART itself is not
+	 * part of the cacheable linar mapping
+	 */
+	alloc_u3_dart_table();
+#endif
+
+#ifdef CONFIG_PMAC_SMU
+	/*
+	 * SMU based G5s need some memory below 2Gb, at least the current
+	 * driver needs that. We have to allocate it now. We allocate 4k
+	 * (1 small page) for now.
+	 */
+	smu_cmdbuf_abs = lmb_alloc_base(4096, 4096, 0x80000000UL);
+#endif /* CONFIG_PMAC_SMU */
+
+	return 1;
+}
+
+#ifdef CONFIG_PPC64
+static int pmac_probe_mode(struct pci_bus *bus)
+{
+	struct device_node *node = bus->sysdata;
+
+	/* We need to use normal PCI probing for the AGP bus,
+	   since the device for the AGP bridge isn't in the tree. */
+	if (bus->self == NULL && device_is_compatible(node, "u3-agp"))
+		return PCI_PROBE_NORMAL;
+
+	return PCI_PROBE_DEVTREE;
+}
+#endif
+
+struct machdep_calls __initdata pmac_md = {
+#if defined(CONFIG_HOTPLUG_CPU) && defined(CONFIG_PPC64)
+	.cpu_die		= generic_mach_cpu_die,
+#endif
+	.probe			= pmac_probe,
+	.setup_arch		= pmac_setup_arch,
+	.init_early		= pmac_init_early,
+	.show_cpuinfo		= pmac_show_cpuinfo,
+	.show_percpuinfo	= pmac_show_percpuinfo,
+	.init_IRQ		= pmac_pic_init,
+	.get_irq		= mpic_get_irq,	/* changed later */
+	.pcibios_fixup		= pmac_pcibios_fixup,
+	.restart		= pmac_restart,
+	.power_off		= pmac_power_off,
+	.halt			= pmac_halt,
+	.time_init		= pmac_time_init,
+	.get_boot_time		= pmac_get_boot_time,
+	.set_rtc_time		= pmac_set_rtc_time,
+	.get_rtc_time		= pmac_get_rtc_time,
+	.calibrate_decr		= pmac_calibrate_decr,
+	.feature_call		= pmac_do_feature_call,
+	.check_legacy_ioport	= pmac_check_legacy_ioport,
+	.progress		= pmac_progress,
+#ifdef CONFIG_PPC64
+	.pci_probe_mode		= pmac_probe_mode,
+	.idle_loop		= native_idle,
+	.enable_pmcs		= power4_enable_pmcs,
+#endif
+#ifdef CONFIG_PPC32
+	.pcibios_enable_device_hook = pmac_pci_enable_device_hook,
+	.pcibios_after_init	= pmac_pcibios_after_init,
+	.phys_mem_access_prot	= pci_phys_mem_access_prot,
+#endif
+};
diff --git a/arch/powerpc/platforms/powermac/sleep.S b/arch/powerpc/platforms/powermac/sleep.S
new file mode 100644
index 000000000000..22b113d19b24
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/sleep.S
@@ -0,0 +1,396 @@
+/*
+ * This file contains sleep low-level functions for PowerBook G3.
+ *    Copyright (C) 1999 Benjamin Herrenschmidt (benh@kernel.crashing.org)
+ *    and Paul Mackerras (paulus@samba.org).
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ *
+ */
+
+#include <linux/config.h>
+#include <asm/processor.h>
+#include <asm/page.h>
+#include <asm/ppc_asm.h>
+#include <asm/cputable.h>
+#include <asm/cache.h>
+#include <asm/thread_info.h>
+#include <asm/asm-offsets.h>
+
+#define MAGIC	0x4c617273	/* 'Lars' */
+
+/*
+ * Structure for storing CPU registers on the stack.
+ */
+#define SL_SP		0
+#define SL_PC		4
+#define SL_MSR		8
+#define SL_SDR1		0xc
+#define SL_SPRG0	0x10	/* 4 sprg's */
+#define SL_DBAT0	0x20
+#define SL_IBAT0	0x28
+#define SL_DBAT1	0x30
+#define SL_IBAT1	0x38
+#define SL_DBAT2	0x40
+#define SL_IBAT2	0x48
+#define SL_DBAT3	0x50
+#define SL_IBAT3	0x58
+#define SL_TB		0x60
+#define SL_R2		0x68
+#define SL_CR		0x6c
+#define SL_R12		0x70	/* r12 to r31 */
+#define SL_SIZE		(SL_R12 + 80)
+
+	.section .text
+	.align	5
+
+#if defined(CONFIG_PM) || defined(CONFIG_CPU_FREQ_PMAC)
+
+/* This gets called by via-pmu.c late during the sleep process.
+ * The PMU was already send the sleep command and will shut us down
+ * soon. We need to save all that is needed and setup the wakeup
+ * vector that will be called by the ROM on wakeup
+ */
+_GLOBAL(low_sleep_handler)
+#ifndef CONFIG_6xx
+	blr
+#else
+	mflr	r0
+	stw	r0,4(r1)
+	stwu	r1,-SL_SIZE(r1)
+	mfcr	r0
+	stw	r0,SL_CR(r1)
+	stw	r2,SL_R2(r1)
+	stmw	r12,SL_R12(r1)
+
+	/* Save MSR & SDR1 */
+	mfmsr	r4
+	stw	r4,SL_MSR(r1)
+	mfsdr1	r4
+	stw	r4,SL_SDR1(r1)
+
+	/* Get a stable timebase and save it */
+1:	mftbu	r4
+	stw	r4,SL_TB(r1)
+	mftb	r5
+	stw	r5,SL_TB+4(r1)
+	mftbu	r3
+	cmpw	r3,r4
+	bne	1b
+
+	/* Save SPRGs */
+	mfsprg	r4,0
+	stw	r4,SL_SPRG0(r1)
+	mfsprg	r4,1
+	stw	r4,SL_SPRG0+4(r1)
+	mfsprg	r4,2
+	stw	r4,SL_SPRG0+8(r1)
+	mfsprg	r4,3
+	stw	r4,SL_SPRG0+12(r1)
+
+	/* Save BATs */
+	mfdbatu	r4,0
+	stw	r4,SL_DBAT0(r1)
+	mfdbatl	r4,0
+	stw	r4,SL_DBAT0+4(r1)
+	mfdbatu	r4,1
+	stw	r4,SL_DBAT1(r1)
+	mfdbatl	r4,1
+	stw	r4,SL_DBAT1+4(r1)
+	mfdbatu	r4,2
+	stw	r4,SL_DBAT2(r1)
+	mfdbatl	r4,2
+	stw	r4,SL_DBAT2+4(r1)
+	mfdbatu	r4,3
+	stw	r4,SL_DBAT3(r1)
+	mfdbatl	r4,3
+	stw	r4,SL_DBAT3+4(r1)
+	mfibatu	r4,0
+	stw	r4,SL_IBAT0(r1)
+	mfibatl	r4,0
+	stw	r4,SL_IBAT0+4(r1)
+	mfibatu	r4,1
+	stw	r4,SL_IBAT1(r1)
+	mfibatl	r4,1
+	stw	r4,SL_IBAT1+4(r1)
+	mfibatu	r4,2
+	stw	r4,SL_IBAT2(r1)
+	mfibatl	r4,2
+	stw	r4,SL_IBAT2+4(r1)
+	mfibatu	r4,3
+	stw	r4,SL_IBAT3(r1)
+	mfibatl	r4,3
+	stw	r4,SL_IBAT3+4(r1)
+
+	/* Backup various CPU config stuffs */
+	bl	__save_cpu_setup
+
+	/* The ROM can wake us up via 2 different vectors:
+	 *  - On wallstreet & lombard, we must write a magic
+	 *    value 'Lars' at address 4 and a pointer to a
+	 *    memory location containing the PC to resume from
+	 *    at address 0.
+	 *  - On Core99, we must store the wakeup vector at
+	 *    address 0x80 and eventually it's parameters
+	 *    at address 0x84. I've have some trouble with those
+	 *    parameters however and I no longer use them.
+	 */
+	lis	r5,grackle_wake_up@ha
+	addi	r5,r5,grackle_wake_up@l
+	tophys(r5,r5)
+	stw	r5,SL_PC(r1)
+	lis	r4,KERNELBASE@h
+	tophys(r5,r1)
+	addi	r5,r5,SL_PC
+	lis	r6,MAGIC@ha
+	addi	r6,r6,MAGIC@l
+	stw	r5,0(r4)
+	stw	r6,4(r4)
+	/* Setup stuffs at 0x80-0x84 for Core99 */
+	lis	r3,core99_wake_up@ha
+	addi	r3,r3,core99_wake_up@l
+	tophys(r3,r3)
+	stw	r3,0x80(r4)
+	stw	r5,0x84(r4)
+	/* Store a pointer to our backup storage into
+	 * a kernel global
+	 */
+	lis r3,sleep_storage@ha
+	addi r3,r3,sleep_storage@l
+	stw r5,0(r3)
+
+	.globl	low_cpu_die
+low_cpu_die:
+	/* Flush & disable all caches */
+	bl	flush_disable_caches
+
+	/* Turn off data relocation. */
+	mfmsr	r3		/* Save MSR in r7 */
+	rlwinm	r3,r3,0,28,26	/* Turn off DR bit */
+	sync
+	mtmsr	r3
+	isync
+
+BEGIN_FTR_SECTION
+	/* Flush any pending L2 data prefetches to work around HW bug */
+	sync
+	lis	r3,0xfff0
+	lwz	r0,0(r3)	/* perform cache-inhibited load to ROM */
+	sync			/* (caches are disabled at this point) */
+END_FTR_SECTION_IFSET(CPU_FTR_SPEC7450)
+
+/*
+ * Set the HID0 and MSR for sleep.
+ */
+	mfspr	r2,SPRN_HID0
+	rlwinm	r2,r2,0,10,7	/* clear doze, nap */
+	oris	r2,r2,HID0_SLEEP@h
+	sync
+	isync
+	mtspr	SPRN_HID0,r2
+	sync
+
+/* This loop puts us back to sleep in case we have a spurrious
+ * wakeup so that the host bridge properly stays asleep. The
+ * CPU will be turned off, either after a known time (about 1
+ * second) on wallstreet & lombard, or as soon as the CPU enters
+ * SLEEP mode on core99
+ */
+	mfmsr	r2
+	oris	r2,r2,MSR_POW@h
+1:	sync
+	mtmsr	r2
+	isync
+	b	1b
+
+/*
+ * Here is the resume code.
+ */
+
+
+/*
+ * Core99 machines resume here
+ * r4 has the physical address of SL_PC(sp) (unused)
+ */
+_GLOBAL(core99_wake_up)
+	/* Make sure HID0 no longer contains any sleep bit and that data cache
+	 * is disabled
+	 */
+	mfspr	r3,SPRN_HID0
+	rlwinm	r3,r3,0,11,7		/* clear SLEEP, NAP, DOZE bits */
+	rlwinm	3,r3,0,18,15		/* clear DCE, ICE */
+	mtspr	SPRN_HID0,r3
+	sync
+	isync
+
+	/* sanitize MSR */
+	mfmsr	r3
+	ori	r3,r3,MSR_EE|MSR_IP
+	xori	r3,r3,MSR_EE|MSR_IP
+	sync
+	isync
+	mtmsr	r3
+	sync
+	isync
+
+	/* Recover sleep storage */
+	lis	r3,sleep_storage@ha
+	addi	r3,r3,sleep_storage@l
+	tophys(r3,r3)
+	lwz	r1,0(r3)
+
+	/* Pass thru to older resume code ... */
+/*
+ * Here is the resume code for older machines.
+ * r1 has the physical address of SL_PC(sp).
+ */
+
+grackle_wake_up:
+
+	/* Restore the kernel's segment registers before
+	 * we do any r1 memory access as we are not sure they
+	 * are in a sane state above the first 256Mb region
+	 */
+	li	r0,16		/* load up segment register values */
+	mtctr	r0		/* for context 0 */
+	lis	r3,0x2000	/* Ku = 1, VSID = 0 */
+	li	r4,0
+3:	mtsrin	r3,r4
+	addi	r3,r3,0x111	/* increment VSID */
+	addis	r4,r4,0x1000	/* address of next segment */
+	bdnz	3b
+	sync
+	isync
+
+	subi	r1,r1,SL_PC
+
+	/* Restore various CPU config stuffs */
+	bl	__restore_cpu_setup
+
+	/* Make sure all FPRs have been initialized */
+	bl	reloc_offset
+	bl	__init_fpu_registers
+
+	/* Invalidate & enable L1 cache, we don't care about
+	 * whatever the ROM may have tried to write to memory
+	 */
+	bl	__inval_enable_L1
+
+	/* Restore the BATs, and SDR1.  Then we can turn on the MMU. */
+	lwz	r4,SL_SDR1(r1)
+	mtsdr1	r4
+	lwz	r4,SL_SPRG0(r1)
+	mtsprg	0,r4
+	lwz	r4,SL_SPRG0+4(r1)
+	mtsprg	1,r4
+	lwz	r4,SL_SPRG0+8(r1)
+	mtsprg	2,r4
+	lwz	r4,SL_SPRG0+12(r1)
+	mtsprg	3,r4
+
+	lwz	r4,SL_DBAT0(r1)
+	mtdbatu	0,r4
+	lwz	r4,SL_DBAT0+4(r1)
+	mtdbatl	0,r4
+	lwz	r4,SL_DBAT1(r1)
+	mtdbatu	1,r4
+	lwz	r4,SL_DBAT1+4(r1)
+	mtdbatl	1,r4
+	lwz	r4,SL_DBAT2(r1)
+	mtdbatu	2,r4
+	lwz	r4,SL_DBAT2+4(r1)
+	mtdbatl	2,r4
+	lwz	r4,SL_DBAT3(r1)
+	mtdbatu	3,r4
+	lwz	r4,SL_DBAT3+4(r1)
+	mtdbatl	3,r4
+	lwz	r4,SL_IBAT0(r1)
+	mtibatu	0,r4
+	lwz	r4,SL_IBAT0+4(r1)
+	mtibatl	0,r4
+	lwz	r4,SL_IBAT1(r1)
+	mtibatu	1,r4
+	lwz	r4,SL_IBAT1+4(r1)
+	mtibatl	1,r4
+	lwz	r4,SL_IBAT2(r1)
+	mtibatu	2,r4
+	lwz	r4,SL_IBAT2+4(r1)
+	mtibatl	2,r4
+	lwz	r4,SL_IBAT3(r1)
+	mtibatu	3,r4
+	lwz	r4,SL_IBAT3+4(r1)
+	mtibatl	3,r4
+
+BEGIN_FTR_SECTION
+	li	r4,0
+	mtspr	SPRN_DBAT4U,r4
+	mtspr	SPRN_DBAT4L,r4
+	mtspr	SPRN_DBAT5U,r4
+	mtspr	SPRN_DBAT5L,r4
+	mtspr	SPRN_DBAT6U,r4
+	mtspr	SPRN_DBAT6L,r4
+	mtspr	SPRN_DBAT7U,r4
+	mtspr	SPRN_DBAT7L,r4
+	mtspr	SPRN_IBAT4U,r4
+	mtspr	SPRN_IBAT4L,r4
+	mtspr	SPRN_IBAT5U,r4
+	mtspr	SPRN_IBAT5L,r4
+	mtspr	SPRN_IBAT6U,r4
+	mtspr	SPRN_IBAT6L,r4
+	mtspr	SPRN_IBAT7U,r4
+	mtspr	SPRN_IBAT7L,r4
+END_FTR_SECTION_IFSET(CPU_FTR_HAS_HIGH_BATS)
+
+	/* Flush all TLBs */
+	lis	r4,0x1000
+1:	addic.	r4,r4,-0x1000
+	tlbie	r4
+	blt	1b
+	sync
+
+	/* restore the MSR and turn on the MMU */
+	lwz	r3,SL_MSR(r1)
+	bl	turn_on_mmu
+
+	/* get back the stack pointer */
+	tovirt(r1,r1)
+
+	/* Restore TB */
+	li	r3,0
+	mttbl	r3
+	lwz	r3,SL_TB(r1)
+	lwz	r4,SL_TB+4(r1)
+	mttbu	r3
+	mttbl	r4
+
+	/* Restore the callee-saved registers and return */
+	lwz	r0,SL_CR(r1)
+	mtcr	r0
+	lwz	r2,SL_R2(r1)
+	lmw	r12,SL_R12(r1)
+	addi	r1,r1,SL_SIZE
+	lwz	r0,4(r1)
+	mtlr	r0
+	blr
+
+turn_on_mmu:
+	mflr	r4
+	tovirt(r4,r4)
+	mtsrr0	r4
+	mtsrr1	r3
+	sync
+	isync
+	rfi
+
+#endif /* defined(CONFIG_PM) || defined(CONFIG_CPU_FREQ) */
+
+	.section .data
+	.balign	L1_CACHE_BYTES
+sleep_storage:
+	.long 0
+	.balign	L1_CACHE_BYTES, 0
+
+#endif /* CONFIG_6xx */
+	.section .text
diff --git a/arch/powerpc/platforms/powermac/smp.c b/arch/powerpc/platforms/powermac/smp.c
new file mode 100644
index 000000000000..e1f9443cc872
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/smp.c
@@ -0,0 +1,865 @@
+/*
+ * SMP support for power macintosh.
+ *
+ * We support both the old "powersurge" SMP architecture
+ * and the current Core99 (G4 PowerMac) machines.
+ *
+ * Note that we don't support the very first rev. of
+ * Apple/DayStar 2 CPUs board, the one with the funky
+ * watchdog. Hopefully, none of these should be there except
+ * maybe internally to Apple. I should probably still add some
+ * code to detect this card though and disable SMP. --BenH.
+ *
+ * Support Macintosh G4 SMP by Troy Benjegerdes (hozer@drgw.net)
+ * and Ben Herrenschmidt <benh@kernel.crashing.org>.
+ *
+ * Support for DayStar quad CPU cards
+ * Copyright (C) XLR8, Inc. 1994-2000
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ */
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+#include <linux/kernel_stat.h>
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/errno.h>
+#include <linux/hardirq.h>
+#include <linux/cpu.h>
+
+#include <asm/ptrace.h>
+#include <asm/atomic.h>
+#include <asm/irq.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/sections.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/smp.h>
+#include <asm/machdep.h>
+#include <asm/pmac_feature.h>
+#include <asm/time.h>
+#include <asm/mpic.h>
+#include <asm/cacheflush.h>
+#include <asm/keylargo.h>
+#include <asm/pmac_low_i2c.h>
+
+#undef DEBUG
+
+#ifdef DEBUG
+#define DBG(fmt...) udbg_printf(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+extern void __secondary_start_pmac_0(void);
+
+#ifdef CONFIG_PPC32
+
+/* Sync flag for HW tb sync */
+static volatile int sec_tb_reset = 0;
+
+/*
+ * Powersurge (old powermac SMP) support.
+ */
+
+/* Addresses for powersurge registers */
+#define HAMMERHEAD_BASE		0xf8000000
+#define HHEAD_CONFIG		0x90
+#define HHEAD_SEC_INTR		0xc0
+
+/* register for interrupting the primary processor on the powersurge */
+/* N.B. this is actually the ethernet ROM! */
+#define PSURGE_PRI_INTR		0xf3019000
+
+/* register for storing the start address for the secondary processor */
+/* N.B. this is the PCI config space address register for the 1st bridge */
+#define PSURGE_START		0xf2800000
+
+/* Daystar/XLR8 4-CPU card */
+#define PSURGE_QUAD_REG_ADDR	0xf8800000
+
+#define PSURGE_QUAD_IRQ_SET	0
+#define PSURGE_QUAD_IRQ_CLR	1
+#define PSURGE_QUAD_IRQ_PRIMARY	2
+#define PSURGE_QUAD_CKSTOP_CTL	3
+#define PSURGE_QUAD_PRIMARY_ARB	4
+#define PSURGE_QUAD_BOARD_ID	6
+#define PSURGE_QUAD_WHICH_CPU	7
+#define PSURGE_QUAD_CKSTOP_RDBK	8
+#define PSURGE_QUAD_RESET_CTL	11
+
+#define PSURGE_QUAD_OUT(r, v)	(out_8(quad_base + ((r) << 4) + 4, (v)))
+#define PSURGE_QUAD_IN(r)	(in_8(quad_base + ((r) << 4) + 4) & 0x0f)
+#define PSURGE_QUAD_BIS(r, v)	(PSURGE_QUAD_OUT((r), PSURGE_QUAD_IN(r) | (v)))
+#define PSURGE_QUAD_BIC(r, v)	(PSURGE_QUAD_OUT((r), PSURGE_QUAD_IN(r) & ~(v)))
+
+/* virtual addresses for the above */
+static volatile u8 __iomem *hhead_base;
+static volatile u8 __iomem *quad_base;
+static volatile u32 __iomem *psurge_pri_intr;
+static volatile u8 __iomem *psurge_sec_intr;
+static volatile u32 __iomem *psurge_start;
+
+/* values for psurge_type */
+#define PSURGE_NONE		-1
+#define PSURGE_DUAL		0
+#define PSURGE_QUAD_OKEE	1
+#define PSURGE_QUAD_COTTON	2
+#define PSURGE_QUAD_ICEGRASS	3
+
+/* what sort of powersurge board we have */
+static int psurge_type = PSURGE_NONE;
+
+/*
+ * Set and clear IPIs for powersurge.
+ */
+static inline void psurge_set_ipi(int cpu)
+{
+	if (psurge_type == PSURGE_NONE)
+		return;
+	if (cpu == 0)
+		in_be32(psurge_pri_intr);
+	else if (psurge_type == PSURGE_DUAL)
+		out_8(psurge_sec_intr, 0);
+	else
+		PSURGE_QUAD_OUT(PSURGE_QUAD_IRQ_SET, 1 << cpu);
+}
+
+static inline void psurge_clr_ipi(int cpu)
+{
+	if (cpu > 0) {
+		switch(psurge_type) {
+		case PSURGE_DUAL:
+			out_8(psurge_sec_intr, ~0);
+		case PSURGE_NONE:
+			break;
+		default:
+			PSURGE_QUAD_OUT(PSURGE_QUAD_IRQ_CLR, 1 << cpu);
+		}
+	}
+}
+
+/*
+ * On powersurge (old SMP powermac architecture) we don't have
+ * separate IPIs for separate messages like openpic does.  Instead
+ * we have a bitmap for each processor, where a 1 bit means that
+ * the corresponding message is pending for that processor.
+ * Ideally each cpu's entry would be in a different cache line.
+ *  -- paulus.
+ */
+static unsigned long psurge_smp_message[NR_CPUS];
+
+void psurge_smp_message_recv(struct pt_regs *regs)
+{
+	int cpu = smp_processor_id();
+	int msg;
+
+	/* clear interrupt */
+	psurge_clr_ipi(cpu);
+
+	if (num_online_cpus() < 2)
+		return;
+
+	/* make sure there is a message there */
+	for (msg = 0; msg < 4; msg++)
+		if (test_and_clear_bit(msg, &psurge_smp_message[cpu]))
+			smp_message_recv(msg, regs);
+}
+
+irqreturn_t psurge_primary_intr(int irq, void *d, struct pt_regs *regs)
+{
+	psurge_smp_message_recv(regs);
+	return IRQ_HANDLED;
+}
+
+static void smp_psurge_message_pass(int target, int msg)
+{
+	int i;
+
+	if (num_online_cpus() < 2)
+		return;
+
+	for (i = 0; i < NR_CPUS; i++) {
+		if (!cpu_online(i))
+			continue;
+		if (target == MSG_ALL
+		    || (target == MSG_ALL_BUT_SELF && i != smp_processor_id())
+		    || target == i) {
+			set_bit(msg, &psurge_smp_message[i]);
+			psurge_set_ipi(i);
+		}
+	}
+}
+
+/*
+ * Determine a quad card presence. We read the board ID register, we
+ * force the data bus to change to something else, and we read it again.
+ * It it's stable, then the register probably exist (ugh !)
+ */
+static int __init psurge_quad_probe(void)
+{
+	int type;
+	unsigned int i;
+
+	type = PSURGE_QUAD_IN(PSURGE_QUAD_BOARD_ID);
+	if (type < PSURGE_QUAD_OKEE || type > PSURGE_QUAD_ICEGRASS
+	    || type != PSURGE_QUAD_IN(PSURGE_QUAD_BOARD_ID))
+		return PSURGE_DUAL;
+
+	/* looks OK, try a slightly more rigorous test */
+	/* bogus is not necessarily cacheline-aligned,
+	   though I don't suppose that really matters.  -- paulus */
+	for (i = 0; i < 100; i++) {
+		volatile u32 bogus[8];
+		bogus[(0+i)%8] = 0x00000000;
+		bogus[(1+i)%8] = 0x55555555;
+		bogus[(2+i)%8] = 0xFFFFFFFF;
+		bogus[(3+i)%8] = 0xAAAAAAAA;
+		bogus[(4+i)%8] = 0x33333333;
+		bogus[(5+i)%8] = 0xCCCCCCCC;
+		bogus[(6+i)%8] = 0xCCCCCCCC;
+		bogus[(7+i)%8] = 0x33333333;
+		wmb();
+		asm volatile("dcbf 0,%0" : : "r" (bogus) : "memory");
+		mb();
+		if (type != PSURGE_QUAD_IN(PSURGE_QUAD_BOARD_ID))
+			return PSURGE_DUAL;
+	}
+	return type;
+}
+
+static void __init psurge_quad_init(void)
+{
+	int procbits;
+
+	if (ppc_md.progress) ppc_md.progress("psurge_quad_init", 0x351);
+	procbits = ~PSURGE_QUAD_IN(PSURGE_QUAD_WHICH_CPU);
+	if (psurge_type == PSURGE_QUAD_ICEGRASS)
+		PSURGE_QUAD_BIS(PSURGE_QUAD_RESET_CTL, procbits);
+	else
+		PSURGE_QUAD_BIC(PSURGE_QUAD_CKSTOP_CTL, procbits);
+	mdelay(33);
+	out_8(psurge_sec_intr, ~0);
+	PSURGE_QUAD_OUT(PSURGE_QUAD_IRQ_CLR, procbits);
+	PSURGE_QUAD_BIS(PSURGE_QUAD_RESET_CTL, procbits);
+	if (psurge_type != PSURGE_QUAD_ICEGRASS)
+		PSURGE_QUAD_BIS(PSURGE_QUAD_CKSTOP_CTL, procbits);
+	PSURGE_QUAD_BIC(PSURGE_QUAD_PRIMARY_ARB, procbits);
+	mdelay(33);
+	PSURGE_QUAD_BIC(PSURGE_QUAD_RESET_CTL, procbits);
+	mdelay(33);
+	PSURGE_QUAD_BIS(PSURGE_QUAD_PRIMARY_ARB, procbits);
+	mdelay(33);
+}
+
+static int __init smp_psurge_probe(void)
+{
+	int i, ncpus;
+
+	/* We don't do SMP on the PPC601 -- paulus */
+	if (PVR_VER(mfspr(SPRN_PVR)) == 1)
+		return 1;
+
+	/*
+	 * The powersurge cpu board can be used in the generation
+	 * of powermacs that have a socket for an upgradeable cpu card,
+	 * including the 7500, 8500, 9500, 9600.
+	 * The device tree doesn't tell you if you have 2 cpus because
+	 * OF doesn't know anything about the 2nd processor.
+	 * Instead we look for magic bits in magic registers,
+	 * in the hammerhead memory controller in the case of the
+	 * dual-cpu powersurge board.  -- paulus.
+	 */
+	if (find_devices("hammerhead") == NULL)
+		return 1;
+
+	hhead_base = ioremap(HAMMERHEAD_BASE, 0x800);
+	quad_base = ioremap(PSURGE_QUAD_REG_ADDR, 1024);
+	psurge_sec_intr = hhead_base + HHEAD_SEC_INTR;
+
+	psurge_type = psurge_quad_probe();
+	if (psurge_type != PSURGE_DUAL) {
+		psurge_quad_init();
+		/* All released cards using this HW design have 4 CPUs */
+		ncpus = 4;
+	} else {
+		iounmap(quad_base);
+		if ((in_8(hhead_base + HHEAD_CONFIG) & 0x02) == 0) {
+			/* not a dual-cpu card */
+			iounmap(hhead_base);
+			psurge_type = PSURGE_NONE;
+			return 1;
+		}
+		ncpus = 2;
+	}
+
+	psurge_start = ioremap(PSURGE_START, 4);
+	psurge_pri_intr = ioremap(PSURGE_PRI_INTR, 4);
+
+	/* this is not actually strictly necessary -- paulus. */
+	for (i = 1; i < ncpus; ++i)
+		smp_hw_index[i] = i;
+
+	if (ppc_md.progress) ppc_md.progress("smp_psurge_probe - done", 0x352);
+
+	return ncpus;
+}
+
+static void __init smp_psurge_kick_cpu(int nr)
+{
+	unsigned long start = __pa(__secondary_start_pmac_0) + nr * 8;
+	unsigned long a;
+
+	/* may need to flush here if secondary bats aren't setup */
+	for (a = KERNELBASE; a < KERNELBASE + 0x800000; a += 32)
+		asm volatile("dcbf 0,%0" : : "r" (a) : "memory");
+	asm volatile("sync");
+
+	if (ppc_md.progress) ppc_md.progress("smp_psurge_kick_cpu", 0x353);
+
+	out_be32(psurge_start, start);
+	mb();
+
+	psurge_set_ipi(nr);
+	udelay(10);
+	psurge_clr_ipi(nr);
+
+	if (ppc_md.progress) ppc_md.progress("smp_psurge_kick_cpu - done", 0x354);
+}
+
+/*
+ * With the dual-cpu powersurge board, the decrementers and timebases
+ * of both cpus are frozen after the secondary cpu is started up,
+ * until we give the secondary cpu another interrupt.  This routine
+ * uses this to get the timebases synchronized.
+ *  -- paulus.
+ */
+static void __init psurge_dual_sync_tb(int cpu_nr)
+{
+	int t;
+
+	set_dec(tb_ticks_per_jiffy);
+	set_tb(0, 0);
+	last_jiffy_stamp(cpu_nr) = 0;
+
+	if (cpu_nr > 0) {
+		mb();
+		sec_tb_reset = 1;
+		return;
+	}
+
+	/* wait for the secondary to have reset its TB before proceeding */
+	for (t = 10000000; t > 0 && !sec_tb_reset; --t)
+		;
+
+	/* now interrupt the secondary, starting both TBs */
+	psurge_set_ipi(1);
+
+	smp_tb_synchronized = 1;
+}
+
+static struct irqaction psurge_irqaction = {
+	.handler = psurge_primary_intr,
+	.flags = SA_INTERRUPT,
+	.mask = CPU_MASK_NONE,
+	.name = "primary IPI",
+};
+
+static void __init smp_psurge_setup_cpu(int cpu_nr)
+{
+
+	if (cpu_nr == 0) {
+		/* If we failed to start the second CPU, we should still
+		 * send it an IPI to start the timebase & DEC or we might
+		 * have them stuck.
+		 */
+		if (num_online_cpus() < 2) {
+			if (psurge_type == PSURGE_DUAL)
+				psurge_set_ipi(1);
+			return;
+		}
+		/* reset the entry point so if we get another intr we won't
+		 * try to startup again */
+		out_be32(psurge_start, 0x100);
+		if (setup_irq(30, &psurge_irqaction))
+			printk(KERN_ERR "Couldn't get primary IPI interrupt");
+	}
+
+	if (psurge_type == PSURGE_DUAL)
+		psurge_dual_sync_tb(cpu_nr);
+}
+
+void __init smp_psurge_take_timebase(void)
+{
+	/* Dummy implementation */
+}
+
+void __init smp_psurge_give_timebase(void)
+{
+	/* Dummy implementation */
+}
+
+/* PowerSurge-style Macs */
+struct smp_ops_t psurge_smp_ops = {
+	.message_pass	= smp_psurge_message_pass,
+	.probe		= smp_psurge_probe,
+	.kick_cpu	= smp_psurge_kick_cpu,
+	.setup_cpu	= smp_psurge_setup_cpu,
+	.give_timebase	= smp_psurge_give_timebase,
+	.take_timebase	= smp_psurge_take_timebase,
+};
+#endif /* CONFIG_PPC32 - actually powersurge support */
+
+#ifdef CONFIG_PPC64
+/*
+ * G5s enable/disable the timebase via an i2c-connected clock chip.
+ */
+static struct device_node *pmac_tb_clock_chip_host;
+static u8 pmac_tb_pulsar_addr;
+static void (*pmac_tb_freeze)(int freeze);
+static DEFINE_SPINLOCK(timebase_lock);
+static unsigned long timebase;
+
+static void smp_core99_cypress_tb_freeze(int freeze)
+{
+	u8 data;
+	int rc;
+
+	/* Strangely, the device-tree says address is 0xd2, but darwin
+	 * accesses 0xd0 ...
+	 */
+	pmac_low_i2c_setmode(pmac_tb_clock_chip_host, pmac_low_i2c_mode_combined);
+	rc = pmac_low_i2c_xfer(pmac_tb_clock_chip_host,
+			       0xd0 | pmac_low_i2c_read,
+			       0x81, &data, 1);
+	if (rc != 0)
+		goto bail;
+
+	data = (data & 0xf3) | (freeze ? 0x00 : 0x0c);
+
+       	pmac_low_i2c_setmode(pmac_tb_clock_chip_host, pmac_low_i2c_mode_stdsub);
+	rc = pmac_low_i2c_xfer(pmac_tb_clock_chip_host,
+			       0xd0 | pmac_low_i2c_write,
+			       0x81, &data, 1);
+
+ bail:
+	if (rc != 0) {
+		printk("Cypress Timebase %s rc: %d\n",
+		       freeze ? "freeze" : "unfreeze", rc);
+		panic("Timebase freeze failed !\n");
+	}
+}
+
+
+static void smp_core99_pulsar_tb_freeze(int freeze)
+{
+	u8 data;
+	int rc;
+
+	pmac_low_i2c_setmode(pmac_tb_clock_chip_host, pmac_low_i2c_mode_combined);
+	rc = pmac_low_i2c_xfer(pmac_tb_clock_chip_host,
+			       pmac_tb_pulsar_addr | pmac_low_i2c_read,
+			       0x2e, &data, 1);
+	if (rc != 0)
+		goto bail;
+
+	data = (data & 0x88) | (freeze ? 0x11 : 0x22);
+
+	pmac_low_i2c_setmode(pmac_tb_clock_chip_host, pmac_low_i2c_mode_stdsub);
+	rc = pmac_low_i2c_xfer(pmac_tb_clock_chip_host,
+			       pmac_tb_pulsar_addr | pmac_low_i2c_write,
+			       0x2e, &data, 1);
+ bail:
+	if (rc != 0) {
+		printk(KERN_ERR "Pulsar Timebase %s rc: %d\n",
+		       freeze ? "freeze" : "unfreeze", rc);
+		panic("Timebase freeze failed !\n");
+	}
+}
+
+
+static void smp_core99_give_timebase(void)
+{
+	/* Open i2c bus for synchronous access */
+	if (pmac_low_i2c_open(pmac_tb_clock_chip_host, 0))
+		panic("Can't open i2c for TB sync !\n");
+
+	spin_lock(&timebase_lock);
+	(*pmac_tb_freeze)(1);
+	mb();
+	timebase = get_tb();
+	spin_unlock(&timebase_lock);
+
+	while (timebase)
+		barrier();
+
+	spin_lock(&timebase_lock);
+	(*pmac_tb_freeze)(0);
+	spin_unlock(&timebase_lock);
+
+	/* Close i2c bus */
+	pmac_low_i2c_close(pmac_tb_clock_chip_host);
+}
+
+
+static void __devinit smp_core99_take_timebase(void)
+{
+	while (!timebase)
+		barrier();
+	spin_lock(&timebase_lock);
+	set_tb(timebase >> 32, timebase & 0xffffffff);
+	timebase = 0;
+	spin_unlock(&timebase_lock);
+}
+
+static void __init smp_core99_setup(int ncpus)
+{
+	struct device_node *cc = NULL;	
+	struct device_node *p;
+	u32 *reg;
+	int ok;
+
+	/* HW sync only on these platforms */
+	if (!machine_is_compatible("PowerMac7,2") &&
+	    !machine_is_compatible("PowerMac7,3") &&
+	    !machine_is_compatible("RackMac3,1"))
+		return;
+
+	/* Look for the clock chip */
+	while ((cc = of_find_node_by_name(cc, "i2c-hwclock")) != NULL) {
+		p = of_get_parent(cc);
+		ok = p && device_is_compatible(p, "uni-n-i2c");
+		of_node_put(p);
+		if (!ok)
+			continue;
+
+		reg = (u32 *)get_property(cc, "reg", NULL);
+		if (reg == NULL)
+			continue;
+
+		switch (*reg) {
+		case 0xd2:
+			if (device_is_compatible(cc, "pulsar-legacy-slewing")) {
+				pmac_tb_freeze = smp_core99_pulsar_tb_freeze;
+				pmac_tb_pulsar_addr = 0xd2;
+				printk(KERN_INFO "Timebase clock is Pulsar chip\n");
+			} else if (device_is_compatible(cc, "cy28508")) {
+				pmac_tb_freeze = smp_core99_cypress_tb_freeze;
+				printk(KERN_INFO "Timebase clock is Cypress chip\n");
+			}
+			break;
+		case 0xd4:
+			pmac_tb_freeze = smp_core99_pulsar_tb_freeze;
+			pmac_tb_pulsar_addr = 0xd4;
+			printk(KERN_INFO "Timebase clock is Pulsar chip\n");
+			break;
+		}
+		if (pmac_tb_freeze != NULL) {
+			pmac_tb_clock_chip_host = of_get_parent(cc);
+			of_node_put(cc);
+			break;
+		}
+	}
+	if (pmac_tb_freeze == NULL) {
+		smp_ops->give_timebase = smp_generic_give_timebase;
+		smp_ops->take_timebase = smp_generic_take_timebase;
+	}
+}
+
+/* nothing to do here, caches are already set up by service processor */
+static inline void __devinit core99_init_caches(int cpu)
+{
+}
+
+#else /* CONFIG_PPC64 */
+
+/*
+ * SMP G4 powermacs use a GPIO to enable/disable the timebase.
+ */
+
+static unsigned int core99_tb_gpio;	/* Timebase freeze GPIO */
+
+static unsigned int pri_tb_hi, pri_tb_lo;
+static unsigned int pri_tb_stamp;
+
+/* not __init, called in sleep/wakeup code */
+void smp_core99_give_timebase(void)
+{
+	unsigned long flags;
+	unsigned int t;
+
+	/* wait for the secondary to be in take_timebase */
+	for (t = 100000; t > 0 && !sec_tb_reset; --t)
+		udelay(10);
+	if (!sec_tb_reset) {
+		printk(KERN_WARNING "Timeout waiting sync on second CPU\n");
+		return;
+	}
+
+	/* freeze the timebase and read it */
+	/* disable interrupts so the timebase is disabled for the
+	   shortest possible time */
+	local_irq_save(flags);
+	pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, core99_tb_gpio, 4);
+	pmac_call_feature(PMAC_FTR_READ_GPIO, NULL, core99_tb_gpio, 0);
+	mb();
+	pri_tb_hi = get_tbu();
+	pri_tb_lo = get_tbl();
+	pri_tb_stamp = last_jiffy_stamp(smp_processor_id());
+	mb();
+
+	/* tell the secondary we're ready */
+	sec_tb_reset = 2;
+	mb();
+
+	/* wait for the secondary to have taken it */
+	for (t = 100000; t > 0 && sec_tb_reset; --t)
+		udelay(10);
+	if (sec_tb_reset)
+		printk(KERN_WARNING "Timeout waiting sync(2) on second CPU\n");
+	else
+		smp_tb_synchronized = 1;
+
+	/* Now, restart the timebase by leaving the GPIO to an open collector */
+       	pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, core99_tb_gpio, 0);
+        pmac_call_feature(PMAC_FTR_READ_GPIO, NULL, core99_tb_gpio, 0);
+	local_irq_restore(flags);
+}
+
+/* not __init, called in sleep/wakeup code */
+void smp_core99_take_timebase(void)
+{
+	unsigned long flags;
+
+	/* tell the primary we're here */
+	sec_tb_reset = 1;
+	mb();
+
+	/* wait for the primary to set pri_tb_hi/lo */
+	while (sec_tb_reset < 2)
+		mb();
+
+	/* set our stuff the same as the primary */
+	local_irq_save(flags);
+	set_dec(1);
+	set_tb(pri_tb_hi, pri_tb_lo);
+	last_jiffy_stamp(smp_processor_id()) = pri_tb_stamp;
+	mb();
+
+	/* tell the primary we're done */
+       	sec_tb_reset = 0;
+	mb();
+	local_irq_restore(flags);
+}
+
+/* L2 and L3 cache settings to pass from CPU0 to CPU1 on G4 cpus */
+volatile static long int core99_l2_cache;
+volatile static long int core99_l3_cache;
+
+static void __devinit core99_init_caches(int cpu)
+{
+	if (!cpu_has_feature(CPU_FTR_L2CR))
+		return;
+
+	if (cpu == 0) {
+		core99_l2_cache = _get_L2CR();
+		printk("CPU0: L2CR is %lx\n", core99_l2_cache);
+	} else {
+		printk("CPU%d: L2CR was %lx\n", cpu, _get_L2CR());
+		_set_L2CR(0);
+		_set_L2CR(core99_l2_cache);
+		printk("CPU%d: L2CR set to %lx\n", cpu, core99_l2_cache);
+	}
+
+	if (!cpu_has_feature(CPU_FTR_L3CR))
+		return;
+
+	if (cpu == 0){
+		core99_l3_cache = _get_L3CR();
+		printk("CPU0: L3CR is %lx\n", core99_l3_cache);
+	} else {
+		printk("CPU%d: L3CR was %lx\n", cpu, _get_L3CR());
+		_set_L3CR(0);
+		_set_L3CR(core99_l3_cache);
+		printk("CPU%d: L3CR set to %lx\n", cpu, core99_l3_cache);
+	}
+}
+
+static void __init smp_core99_setup(int ncpus)
+{
+	struct device_node *cpu;
+	u32 *tbprop = NULL;
+	int i;
+
+	core99_tb_gpio = KL_GPIO_TB_ENABLE;	/* default value */
+	cpu = of_find_node_by_type(NULL, "cpu");
+	if (cpu != NULL) {
+		tbprop = (u32 *)get_property(cpu, "timebase-enable", NULL);
+		if (tbprop)
+			core99_tb_gpio = *tbprop;
+		of_node_put(cpu);
+	}
+
+	/* XXX should get this from reg properties */
+	for (i = 1; i < ncpus; ++i)
+		smp_hw_index[i] = i;
+	powersave_nap = 0;
+}
+#endif
+
+static int __init smp_core99_probe(void)
+{
+	struct device_node *cpus;
+	int ncpus = 0;
+
+	if (ppc_md.progress) ppc_md.progress("smp_core99_probe", 0x345);
+
+	/* Count CPUs in the device-tree */
+       	for (cpus = NULL; (cpus = of_find_node_by_type(cpus, "cpu")) != NULL;)
+	       	++ncpus;
+
+	printk(KERN_INFO "PowerMac SMP probe found %d cpus\n", ncpus);
+
+	/* Nothing more to do if less than 2 of them */
+	if (ncpus <= 1)
+		return 1;
+
+	smp_core99_setup(ncpus);
+	mpic_request_ipis();
+	core99_init_caches(0);
+
+	return ncpus;
+}
+
+static void __devinit smp_core99_kick_cpu(int nr)
+{
+	unsigned int save_vector;
+	unsigned long new_vector;
+	unsigned long flags;
+	volatile unsigned int *vector
+		 = ((volatile unsigned int *)(KERNELBASE+0x100));
+
+	if (nr < 0 || nr > 3)
+		return;
+	if (ppc_md.progress) ppc_md.progress("smp_core99_kick_cpu", 0x346);
+
+	local_irq_save(flags);
+	local_irq_disable();
+
+	/* Save reset vector */
+	save_vector = *vector;
+
+	/* Setup fake reset vector that does	
+	 *   b __secondary_start_pmac_0 + nr*8 - KERNELBASE
+	 */
+	new_vector = (unsigned long) __secondary_start_pmac_0 + nr * 8;
+	*vector = 0x48000002 + new_vector - KERNELBASE;
+
+	/* flush data cache and inval instruction cache */
+	flush_icache_range((unsigned long) vector, (unsigned long) vector + 4);
+
+	/* Put some life in our friend */
+	pmac_call_feature(PMAC_FTR_RESET_CPU, NULL, nr, 0);
+
+	/* FIXME: We wait a bit for the CPU to take the exception, I should
+	 * instead wait for the entry code to set something for me. Well,
+	 * ideally, all that crap will be done in prom.c and the CPU left
+	 * in a RAM-based wait loop like CHRP.
+	 */
+	mdelay(1);
+
+	/* Restore our exception vector */
+	*vector = save_vector;
+	flush_icache_range((unsigned long) vector, (unsigned long) vector + 4);
+
+	local_irq_restore(flags);
+	if (ppc_md.progress) ppc_md.progress("smp_core99_kick_cpu done", 0x347);
+}
+
+static void __devinit smp_core99_setup_cpu(int cpu_nr)
+{
+	/* Setup L2/L3 */
+	if (cpu_nr != 0)
+		core99_init_caches(cpu_nr);
+
+	/* Setup openpic */
+	mpic_setup_this_cpu();
+
+	if (cpu_nr == 0) {
+#ifdef CONFIG_POWER4
+		extern void g5_phy_disable_cpu1(void);
+
+		/* If we didn't start the second CPU, we must take
+		 * it off the bus
+		 */
+		if (machine_is_compatible("MacRISC4") &&
+		    num_online_cpus() < 2)		
+			g5_phy_disable_cpu1();
+#endif /* CONFIG_POWER4 */
+		if (ppc_md.progress) ppc_md.progress("core99_setup_cpu 0 done", 0x349);
+	}
+}
+
+
+/* Core99 Macs (dual G4s and G5s) */
+struct smp_ops_t core99_smp_ops = {
+	.message_pass	= smp_mpic_message_pass,
+	.probe		= smp_core99_probe,
+	.kick_cpu	= smp_core99_kick_cpu,
+	.setup_cpu	= smp_core99_setup_cpu,
+	.give_timebase	= smp_core99_give_timebase,
+	.take_timebase	= smp_core99_take_timebase,
+};
+
+#if defined(CONFIG_HOTPLUG_CPU) && defined(CONFIG_PPC32)
+
+int __cpu_disable(void)
+{
+	cpu_clear(smp_processor_id(), cpu_online_map);
+
+	/* XXX reset cpu affinity here */
+	mpic_cpu_set_priority(0xf);
+	asm volatile("mtdec %0" : : "r" (0x7fffffff));
+	mb();
+	udelay(20);
+	asm volatile("mtdec %0" : : "r" (0x7fffffff));
+	return 0;
+}
+
+extern void low_cpu_die(void) __attribute__((noreturn)); /* in sleep.S */
+static int cpu_dead[NR_CPUS];
+
+void cpu_die(void)
+{
+	local_irq_disable();
+	cpu_dead[smp_processor_id()] = 1;
+	mb();
+	low_cpu_die();
+}
+
+void __cpu_die(unsigned int cpu)
+{
+	int timeout;
+
+	timeout = 1000;
+	while (!cpu_dead[cpu]) {
+		if (--timeout == 0) {
+			printk("CPU %u refused to die!\n", cpu);
+			break;
+		}
+		msleep(1);
+	}
+	cpu_callin_map[cpu] = 0;
+	cpu_dead[cpu] = 0;
+}
+
+#endif
diff --git a/arch/powerpc/platforms/powermac/time.c b/arch/powerpc/platforms/powermac/time.c
new file mode 100644
index 000000000000..5947b21a8588
--- /dev/null
+++ b/arch/powerpc/platforms/powermac/time.c
@@ -0,0 +1,360 @@
+/*
+ * Support for periodic interrupts (100 per second) and for getting
+ * the current time from the RTC on Power Macintoshes.
+ *
+ * We use the decrementer register for our periodic interrupts.
+ *
+ * Paul Mackerras	August 1996.
+ * Copyright (C) 1996 Paul Mackerras.
+ * Copyright (C) 2003-2005 Benjamin Herrenschmidt.
+ *
+ */
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/init.h>
+#include <linux/time.h>
+#include <linux/adb.h>
+#include <linux/cuda.h>
+#include <linux/pmu.h>
+#include <linux/interrupt.h>
+#include <linux/hardirq.h>
+#include <linux/rtc.h>
+
+#include <asm/sections.h>
+#include <asm/prom.h>
+#include <asm/system.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/machdep.h>
+#include <asm/time.h>
+#include <asm/nvram.h>
+#include <asm/smu.h>
+
+#undef DEBUG
+
+#ifdef DEBUG
+#define DBG(x...) printk(x)
+#else
+#define DBG(x...)
+#endif
+
+/* Apparently the RTC stores seconds since 1 Jan 1904 */
+#define RTC_OFFSET	2082844800
+
+/*
+ * Calibrate the decrementer frequency with the VIA timer 1.
+ */
+#define VIA_TIMER_FREQ_6	4700000	/* time 1 frequency * 6 */
+
+/* VIA registers */
+#define RS		0x200		/* skip between registers */
+#define T1CL		(4*RS)		/* Timer 1 ctr/latch (low 8 bits) */
+#define T1CH		(5*RS)		/* Timer 1 counter (high 8 bits) */
+#define T1LL		(6*RS)		/* Timer 1 latch (low 8 bits) */
+#define T1LH		(7*RS)		/* Timer 1 latch (high 8 bits) */
+#define ACR		(11*RS)		/* Auxiliary control register */
+#define IFR		(13*RS)		/* Interrupt flag register */
+
+/* Bits in ACR */
+#define T1MODE		0xc0		/* Timer 1 mode */
+#define T1MODE_CONT	0x40		/*  continuous interrupts */
+
+/* Bits in IFR and IER */
+#define T1_INT		0x40		/* Timer 1 interrupt */
+
+long __init pmac_time_init(void)
+{
+	s32 delta = 0;
+#ifdef CONFIG_NVRAM
+	int dst;
+	
+	delta = ((s32)pmac_xpram_read(PMAC_XPRAM_MACHINE_LOC + 0x9)) << 16;
+	delta |= ((s32)pmac_xpram_read(PMAC_XPRAM_MACHINE_LOC + 0xa)) << 8;
+	delta |= pmac_xpram_read(PMAC_XPRAM_MACHINE_LOC + 0xb);
+	if (delta & 0x00800000UL)
+		delta |= 0xFF000000UL;
+	dst = ((pmac_xpram_read(PMAC_XPRAM_MACHINE_LOC + 0x8) & 0x80) != 0);
+	printk("GMT Delta read from XPRAM: %d minutes, DST: %s\n", delta/60,
+		dst ? "on" : "off");
+#endif
+	return delta;
+}
+
+static void to_rtc_time(unsigned long now, struct rtc_time *tm)
+{
+	to_tm(now, tm);
+	tm->tm_year -= 1900;
+	tm->tm_mon -= 1;
+}
+
+static unsigned long from_rtc_time(struct rtc_time *tm)
+{
+	return mktime(tm->tm_year+1900, tm->tm_mon+1, tm->tm_mday,
+		      tm->tm_hour, tm->tm_min, tm->tm_sec);
+}
+
+#ifdef CONFIG_ADB_CUDA
+static unsigned long cuda_get_time(void)
+{
+	struct adb_request req;
+	unsigned long now;
+
+	if (cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_GET_TIME) < 0)
+		return 0;
+	while (!req.complete)
+		cuda_poll();
+	if (req.reply_len != 7)
+		printk(KERN_ERR "cuda_get_time: got %d byte reply\n",
+		       req.reply_len);
+	now = (req.reply[3] << 24) + (req.reply[4] << 16)
+		+ (req.reply[5] << 8) + req.reply[6];
+	return now - RTC_OFFSET;
+}
+
+#define cuda_get_rtc_time(tm)	to_rtc_time(cuda_get_time(), (tm))
+
+static int cuda_set_rtc_time(struct rtc_time *tm)
+{
+	unsigned int nowtime;
+	struct adb_request req;
+
+	nowtime = from_rtc_time(tm) + RTC_OFFSET;
+	if (cuda_request(&req, NULL, 6, CUDA_PACKET, CUDA_SET_TIME,
+			 nowtime >> 24, nowtime >> 16, nowtime >> 8,
+			 nowtime) < 0)
+		return -ENXIO;
+	while (!req.complete)
+		cuda_poll();
+	if ((req.reply_len != 3) && (req.reply_len != 7))
+		printk(KERN_ERR "cuda_set_rtc_time: got %d byte reply\n",
+		       req.reply_len);
+	return 0;
+}
+
+#else
+#define cuda_get_time()		0
+#define cuda_get_rtc_time(tm)
+#define cuda_set_rtc_time(tm)	0
+#endif
+
+#ifdef CONFIG_ADB_PMU
+static unsigned long pmu_get_time(void)
+{
+	struct adb_request req;
+	unsigned long now;
+
+	if (pmu_request(&req, NULL, 1, PMU_READ_RTC) < 0)
+		return 0;
+	pmu_wait_complete(&req);
+	if (req.reply_len != 4)
+		printk(KERN_ERR "pmu_get_time: got %d byte reply from PMU\n",
+		       req.reply_len);
+	now = (req.reply[0] << 24) + (req.reply[1] << 16)
+		+ (req.reply[2] << 8) + req.reply[3];
+	return now - RTC_OFFSET;
+}
+
+#define pmu_get_rtc_time(tm)	to_rtc_time(pmu_get_time(), (tm))
+
+static int pmu_set_rtc_time(struct rtc_time *tm)
+{
+	unsigned int nowtime;
+	struct adb_request req;
+
+	nowtime = from_rtc_time(tm) + RTC_OFFSET;
+	if (pmu_request(&req, NULL, 5, PMU_SET_RTC, nowtime >> 24,
+			nowtime >> 16, nowtime >> 8, nowtime) < 0)
+		return -ENXIO;
+	pmu_wait_complete(&req);
+	if (req.reply_len != 0)
+		printk(KERN_ERR "pmu_set_rtc_time: %d byte reply from PMU\n",
+		       req.reply_len);
+	return 0;
+}
+
+#else
+#define pmu_get_time()		0
+#define pmu_get_rtc_time(tm)
+#define pmu_set_rtc_time(tm)	0
+#endif
+
+#ifdef CONFIG_PMAC_SMU
+static unsigned long smu_get_time(void)
+{
+	struct rtc_time tm;
+
+	if (smu_get_rtc_time(&tm, 1))
+		return 0;
+	return from_rtc_time(&tm);
+}
+
+#else
+#define smu_get_time()			0
+#define smu_get_rtc_time(tm, spin)
+#define smu_set_rtc_time(tm, spin)	0
+#endif
+
+unsigned long pmac_get_boot_time(void)
+{
+	/* Get the time from the RTC, used only at boot time */
+	switch (sys_ctrler) {
+	case SYS_CTRLER_CUDA:
+		return cuda_get_time();
+	case SYS_CTRLER_PMU:
+		return pmu_get_time();
+	case SYS_CTRLER_SMU:
+		return smu_get_time();
+	default:
+		return 0;
+	}
+}
+
+void pmac_get_rtc_time(struct rtc_time *tm)
+{
+	/* Get the time from the RTC, used only at boot time */
+	switch (sys_ctrler) {
+	case SYS_CTRLER_CUDA:
+		cuda_get_rtc_time(tm);
+		break;
+	case SYS_CTRLER_PMU:
+		pmu_get_rtc_time(tm);
+		break;
+	case SYS_CTRLER_SMU:
+		smu_get_rtc_time(tm, 1);
+		break;
+	default:
+		;
+	}
+}
+
+int pmac_set_rtc_time(struct rtc_time *tm)
+{
+	switch (sys_ctrler) {
+	case SYS_CTRLER_CUDA:
+		return cuda_set_rtc_time(tm);
+	case SYS_CTRLER_PMU:
+		return pmu_set_rtc_time(tm);
+	case SYS_CTRLER_SMU:
+		return smu_set_rtc_time(tm, 1);
+	default:
+		return -ENODEV;
+	}
+}
+
+#ifdef CONFIG_PPC32
+/*
+ * Calibrate the decrementer register using VIA timer 1.
+ * This is used both on powermacs and CHRP machines.
+ */
+int __init via_calibrate_decr(void)
+{
+	struct device_node *vias;
+	volatile unsigned char __iomem *via;
+	int count = VIA_TIMER_FREQ_6 / 100;
+	unsigned int dstart, dend;
+
+	vias = find_devices("via-cuda");
+	if (vias == 0)
+		vias = find_devices("via-pmu");
+	if (vias == 0)
+		vias = find_devices("via");
+	if (vias == 0 || vias->n_addrs == 0)
+		return 0;
+	via = ioremap(vias->addrs[0].address, vias->addrs[0].size);
+
+	/* set timer 1 for continuous interrupts */
+	out_8(&via[ACR], (via[ACR] & ~T1MODE) | T1MODE_CONT);
+	/* set the counter to a small value */
+	out_8(&via[T1CH], 2);
+	/* set the latch to `count' */
+	out_8(&via[T1LL], count);
+	out_8(&via[T1LH], count >> 8);
+	/* wait until it hits 0 */
+	while ((in_8(&via[IFR]) & T1_INT) == 0)
+		;
+	dstart = get_dec();
+	/* clear the interrupt & wait until it hits 0 again */
+	in_8(&via[T1CL]);
+	while ((in_8(&via[IFR]) & T1_INT) == 0)
+		;
+	dend = get_dec();
+
+	ppc_tb_freq = (dstart - dend) * 100 / 6;
+
+	iounmap(via);
+	
+	return 1;
+}
+#endif
+
+#ifdef CONFIG_PM
+/*
+ * Reset the time after a sleep.
+ */
+static int
+time_sleep_notify(struct pmu_sleep_notifier *self, int when)
+{
+	static unsigned long time_diff;
+	unsigned long flags;
+	unsigned long seq;
+	struct timespec tv;
+
+	switch (when) {
+	case PBOOK_SLEEP_NOW:
+		do {
+			seq = read_seqbegin_irqsave(&xtime_lock, flags);
+			time_diff = xtime.tv_sec - pmac_get_boot_time();
+		} while (read_seqretry_irqrestore(&xtime_lock, seq, flags));
+		break;
+	case PBOOK_WAKE:
+		tv.tv_sec = pmac_get_boot_time() + time_diff;
+		tv.tv_nsec = 0;
+		do_settimeofday(&tv);
+		break;
+	}
+	return PBOOK_SLEEP_OK;
+}
+
+static struct pmu_sleep_notifier time_sleep_notifier = {
+	time_sleep_notify, SLEEP_LEVEL_MISC,
+};
+#endif /* CONFIG_PM */
+
+/*
+ * Query the OF and get the decr frequency.
+ */
+void __init pmac_calibrate_decr(void)
+{
+#ifdef CONFIG_PM
+	/* XXX why here? */
+	pmu_register_sleep_notifier(&time_sleep_notifier);
+#endif /* CONFIG_PM */
+
+	generic_calibrate_decr();
+
+#ifdef CONFIG_PPC32
+	/* We assume MacRISC2 machines have correct device-tree
+	 * calibration. That's better since the VIA itself seems
+	 * to be slightly off. --BenH
+	 */
+	if (!machine_is_compatible("MacRISC2") &&
+	    !machine_is_compatible("MacRISC3") &&
+	    !machine_is_compatible("MacRISC4"))
+		if (via_calibrate_decr())
+			return;
+
+	/* Special case: QuickSilver G4s seem to have a badly calibrated
+	 * timebase-frequency in OF, VIA is much better on these. We should
+	 * probably implement calibration based on the KL timer on these
+	 * machines anyway... -BenH
+	 */
+	if (machine_is_compatible("PowerMac3,5"))
+		if (via_calibrate_decr())
+			return;
+#endif
+}
diff --git a/arch/powerpc/platforms/prep/Kconfig b/arch/powerpc/platforms/prep/Kconfig
new file mode 100644
index 000000000000..673ac47a1626
--- /dev/null
+++ b/arch/powerpc/platforms/prep/Kconfig
@@ -0,0 +1,22 @@
+
+config PREP_RESIDUAL
+	bool "Support for PReP Residual Data"
+	depends on PPC_PREP
+	help
+	  Some PReP systems have residual data passed to the kernel by the
+	  firmware.  This allows detection of memory size, devices present and
+	  other useful pieces of information.  Sometimes this information is
+	  not present or incorrect, in which case it could lead to the machine 
+	  behaving incorrectly.  If this happens, either disable PREP_RESIDUAL
+	  or pass the 'noresidual' option to the kernel.
+
+	  If you are running a PReP system, say Y here, otherwise say N.
+
+config PROC_PREPRESIDUAL
+	bool "Support for reading of PReP Residual Data in /proc"
+	depends on PREP_RESIDUAL && PROC_FS
+	help
+	  Enabling this option will create a /proc/residual file which allows
+	  you to get at the residual data on PReP systems.  You will need a tool
+	  (lsresidual) to parse it.  If you aren't on a PReP system, you don't
+	  want this.
diff --git a/arch/powerpc/platforms/pseries/Kconfig b/arch/powerpc/platforms/pseries/Kconfig
new file mode 100644
index 000000000000..e3fc3407bb1f
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/Kconfig
@@ -0,0 +1,33 @@
+
+config PPC_SPLPAR
+	depends on PPC_PSERIES
+	bool "Support for shared-processor logical partitions"
+	default n
+	help
+	  Enabling this option will make the kernel run more efficiently
+	  on logically-partitioned pSeries systems which use shared
+	  processors, that is, which share physical processors between
+	  two or more partitions.
+
+config HMT
+	bool "Hardware multithreading"
+	depends on SMP && PPC_PSERIES && BROKEN
+	help
+	  This option enables hardware multithreading on RS64 cpus.
+	  pSeries systems p620 and p660 have such a cpu type.
+
+config EEH
+	bool "PCI Extended Error Handling (EEH)" if EMBEDDED
+	depends on PPC_PSERIES
+	default y if !EMBEDDED
+
+config SCANLOG
+	tristate "Scanlog dump interface"
+	depends on RTAS_PROC && PPC_PSERIES
+
+config LPARCFG
+	tristate "LPAR Configuration Data"
+	depends on PPC_PSERIES || PPC_ISERIES
+	help
+	Provide system capacity information via human readable
+	<key word>=<value> pairs through a /proc/ppc64/lparcfg interface.
diff --git a/arch/powerpc/platforms/pseries/Makefile b/arch/powerpc/platforms/pseries/Makefile
new file mode 100644
index 000000000000..b9938fece781
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/Makefile
@@ -0,0 +1,5 @@
+obj-y			:= pci.o lpar.o hvCall.o nvram.o reconfig.o \
+			   setup.o iommu.o ras.o rtasd.o
+obj-$(CONFIG_SMP)	+= smp.o
+obj-$(CONFIG_IBMVIO)	+= vio.o
+obj-$(CONFIG_XICS)	+= xics.o
diff --git a/arch/powerpc/platforms/pseries/hvCall.S b/arch/powerpc/platforms/pseries/hvCall.S
new file mode 100644
index 000000000000..176e8da76466
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/hvCall.S
@@ -0,0 +1,131 @@
+/*
+ * arch/ppc64/kernel/pSeries_hvCall.S
+ *
+ * This file contains the generic code to perform a call to the
+ * pSeries LPAR hypervisor.
+ * NOTE: this file will go away when we move to inline this work.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+#include <asm/hvcall.h>
+#include <asm/processor.h>
+#include <asm/ppc_asm.h>
+	
+#define STK_PARM(i)     (48 + ((i)-3)*8)
+
+	.text
+
+/* long plpar_hcall(unsigned long opcode,		R3
+			unsigned long arg1,		R4
+			unsigned long arg2,		R5
+			unsigned long arg3,		R6
+			unsigned long arg4,		R7
+			unsigned long *out1,		R8
+			unsigned long *out2,		R9
+			unsigned long *out3);		R10
+ */
+_GLOBAL(plpar_hcall)
+	HMT_MEDIUM
+
+	mfcr	r0
+
+	std	r8,STK_PARM(r8)(r1)	/* Save out ptrs */
+	std	r9,STK_PARM(r9)(r1)
+	std	r10,STK_PARM(r10)(r1)
+
+	stw	r0,8(r1)
+
+	HVSC				/* invoke the hypervisor */
+
+	lwz	r0,8(r1)
+
+	ld	r8,STK_PARM(r8)(r1)	/* Fetch r4-r6 ret args */
+	ld	r9,STK_PARM(r9)(r1)
+	ld	r10,STK_PARM(r10)(r1)
+	std	r4,0(r8)
+	std	r5,0(r9)
+	std	r6,0(r10)
+
+	mtcrf	0xff,r0
+	blr				/* return r3 = status */
+
+
+/* Simple interface with no output values (other than status) */
+_GLOBAL(plpar_hcall_norets)
+	HMT_MEDIUM
+
+	mfcr	r0
+	stw	r0,8(r1)
+
+	HVSC				/* invoke the hypervisor */
+
+	lwz	r0,8(r1)
+	mtcrf	0xff,r0
+	blr				/* return r3 = status */
+
+
+/* long plpar_hcall_8arg_2ret(unsigned long opcode,	R3
+			unsigned long arg1,		R4
+			unsigned long arg2,		R5
+			unsigned long arg3,		R6
+			unsigned long arg4,		R7
+			unsigned long arg5,		R8
+			unsigned long arg6,		R9
+			unsigned long arg7,		R10
+			unsigned long arg8,		112(R1)
+			unsigned long *out1);		120(R1)
+ */
+_GLOBAL(plpar_hcall_8arg_2ret)
+	HMT_MEDIUM
+
+	mfcr	r0
+	ld	r11,STK_PARM(r11)(r1)	/* put arg8 in R11 */
+	stw	r0,8(r1)
+
+	HVSC				/* invoke the hypervisor */
+
+	lwz	r0,8(r1)
+	ld	r10,STK_PARM(r12)(r1)	/* Fetch r4 ret arg */
+	std	r4,0(r10)
+	mtcrf	0xff,r0
+	blr				/* return r3 = status */
+
+
+/* long plpar_hcall_4out(unsigned long opcode,		R3
+		 	unsigned long arg1,		R4
+		 	unsigned long arg2,		R5
+		 	unsigned long arg3,		R6
+		 	unsigned long arg4,		R7
+		 	unsigned long *out1,		R8
+		 	unsigned long *out2,		R9
+		 	unsigned long *out3,		R10
+		 	unsigned long *out4);		112(R1)
+ */
+_GLOBAL(plpar_hcall_4out)
+	HMT_MEDIUM
+
+	mfcr	r0
+	stw	r0,8(r1)
+
+	std	r8,STK_PARM(r8)(r1)	/* Save out ptrs */
+	std	r9,STK_PARM(r9)(r1)
+	std	r10,STK_PARM(r10)(r1)
+
+	HVSC				/* invoke the hypervisor */
+
+	lwz	r0,8(r1)
+
+	ld	r8,STK_PARM(r8)(r1)	/* Fetch r4-r7 ret args */
+	ld	r9,STK_PARM(r9)(r1)
+	ld	r10,STK_PARM(r10)(r1)
+	ld	r11,STK_PARM(r11)(r1)
+	std	r4,0(r8)
+	std	r5,0(r9)
+	std	r6,0(r10)
+	std	r7,0(r11)
+
+	mtcrf	0xff,r0
+	blr				/* return r3 = status */
diff --git a/arch/powerpc/platforms/pseries/iommu.c b/arch/powerpc/platforms/pseries/iommu.c
new file mode 100644
index 000000000000..513e27231493
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/iommu.c
@@ -0,0 +1,607 @@
+/*
+ * arch/ppc64/kernel/pSeries_iommu.c
+ *
+ * Copyright (C) 2001 Mike Corrigan & Dave Engebretsen, IBM Corporation
+ *
+ * Rewrite, cleanup: 
+ *
+ * Copyright (C) 2004 Olof Johansson <olof@austin.ibm.com>, IBM Corporation
+ *
+ * Dynamic DMA mapping support, pSeries-specific parts, both SMP and LPAR.
+ *
+ * 
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ * 
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ * 
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/config.h>
+#include <linux/init.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/mm.h>
+#include <linux/spinlock.h>
+#include <linux/string.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/rtas.h>
+#include <asm/ppcdebug.h>
+#include <asm/iommu.h>
+#include <asm/pci-bridge.h>
+#include <asm/machdep.h>
+#include <asm/abs_addr.h>
+#include <asm/pSeries_reconfig.h>
+#include <asm/systemcfg.h>
+#include <asm/firmware.h>
+#include <asm/tce.h>
+#include <asm/ppc-pci.h>
+
+#include "plpar_wrappers.h"
+
+#define DBG(fmt...)
+
+extern int is_python(struct device_node *);
+
+static void tce_build_pSeries(struct iommu_table *tbl, long index, 
+			      long npages, unsigned long uaddr, 
+			      enum dma_data_direction direction)
+{
+	union tce_entry t;
+	union tce_entry *tp;
+
+	index <<= TCE_PAGE_FACTOR;
+	npages <<= TCE_PAGE_FACTOR;
+
+	t.te_word = 0;
+	t.te_rdwr = 1; // Read allowed 
+
+	if (direction != DMA_TO_DEVICE)
+		t.te_pciwr = 1;
+
+	tp = ((union tce_entry *)tbl->it_base) + index;
+
+	while (npages--) {
+		/* can't move this out since we might cross LMB boundary */
+		t.te_rpn = (virt_to_abs(uaddr)) >> TCE_SHIFT;
+	
+		tp->te_word = t.te_word;
+
+		uaddr += TCE_PAGE_SIZE;
+		tp++;
+	}
+}
+
+
+static void tce_free_pSeries(struct iommu_table *tbl, long index, long npages)
+{
+	union tce_entry t;
+	union tce_entry *tp;
+
+	npages <<= TCE_PAGE_FACTOR;
+	index <<= TCE_PAGE_FACTOR;
+
+	t.te_word = 0;
+	tp  = ((union tce_entry *)tbl->it_base) + index;
+		
+	while (npages--) {
+		tp->te_word = t.te_word;
+		
+		tp++;
+	}
+}
+
+
+static void tce_build_pSeriesLP(struct iommu_table *tbl, long tcenum,
+				long npages, unsigned long uaddr,
+				enum dma_data_direction direction)
+{
+	u64 rc;
+	union tce_entry tce;
+
+	tce.te_word = 0;
+	tce.te_rpn = (virt_to_abs(uaddr)) >> TCE_SHIFT;
+	tce.te_rdwr = 1;
+	if (direction != DMA_TO_DEVICE)
+		tce.te_pciwr = 1;
+
+	while (npages--) {
+		rc = plpar_tce_put((u64)tbl->it_index, 
+				   (u64)tcenum << 12, 
+				   tce.te_word );
+		
+		if (rc && printk_ratelimit()) {
+			printk("tce_build_pSeriesLP: plpar_tce_put failed. rc=%ld\n", rc);
+			printk("\tindex   = 0x%lx\n", (u64)tbl->it_index);
+			printk("\ttcenum  = 0x%lx\n", (u64)tcenum);
+			printk("\ttce val = 0x%lx\n", tce.te_word );
+			show_stack(current, (unsigned long *)__get_SP());
+		}
+			
+		tcenum++;
+		tce.te_rpn++;
+	}
+}
+
+static DEFINE_PER_CPU(void *, tce_page) = NULL;
+
+static void tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
+				     long npages, unsigned long uaddr,
+				     enum dma_data_direction direction)
+{
+	u64 rc;
+	union tce_entry tce, *tcep;
+	long l, limit;
+
+	tcenum <<= TCE_PAGE_FACTOR;
+	npages <<= TCE_PAGE_FACTOR;
+
+	if (npages == 1)
+		return tce_build_pSeriesLP(tbl, tcenum, npages, uaddr,
+					   direction);
+
+	tcep = __get_cpu_var(tce_page);
+
+	/* This is safe to do since interrupts are off when we're called
+	 * from iommu_alloc{,_sg}()
+	 */
+	if (!tcep) {
+		tcep = (void *)__get_free_page(GFP_ATOMIC);
+		/* If allocation fails, fall back to the loop implementation */
+		if (!tcep)
+			return tce_build_pSeriesLP(tbl, tcenum, npages,
+						   uaddr, direction);
+		__get_cpu_var(tce_page) = tcep;
+	}
+
+	tce.te_word = 0;
+	tce.te_rpn = (virt_to_abs(uaddr)) >> TCE_SHIFT;
+	tce.te_rdwr = 1;
+	if (direction != DMA_TO_DEVICE)
+		tce.te_pciwr = 1;
+
+	/* We can map max one pageful of TCEs at a time */
+	do {
+		/*
+		 * Set up the page with TCE data, looping through and setting
+		 * the values.
+		 */
+		limit = min_t(long, npages, 4096/sizeof(union tce_entry));
+
+		for (l = 0; l < limit; l++) {
+			tcep[l] = tce;
+			tce.te_rpn++;
+		}
+
+		rc = plpar_tce_put_indirect((u64)tbl->it_index,
+					    (u64)tcenum << 12,
+					    (u64)virt_to_abs(tcep),
+					    limit);
+
+		npages -= limit;
+		tcenum += limit;
+	} while (npages > 0 && !rc);
+
+	if (rc && printk_ratelimit()) {
+		printk("tce_buildmulti_pSeriesLP: plpar_tce_put failed. rc=%ld\n", rc);
+		printk("\tindex   = 0x%lx\n", (u64)tbl->it_index);
+		printk("\tnpages  = 0x%lx\n", (u64)npages);
+		printk("\ttce[0] val = 0x%lx\n", tcep[0].te_word);
+		show_stack(current, (unsigned long *)__get_SP());
+	}
+}
+
+static void tce_free_pSeriesLP(struct iommu_table *tbl, long tcenum, long npages)
+{
+	u64 rc;
+	union tce_entry tce;
+
+	tcenum <<= TCE_PAGE_FACTOR;
+	npages <<= TCE_PAGE_FACTOR;
+
+	tce.te_word = 0;
+
+	while (npages--) {
+		rc = plpar_tce_put((u64)tbl->it_index,
+				   (u64)tcenum << 12,
+				   tce.te_word);
+
+		if (rc && printk_ratelimit()) {
+			printk("tce_free_pSeriesLP: plpar_tce_put failed. rc=%ld\n", rc);
+			printk("\tindex   = 0x%lx\n", (u64)tbl->it_index);
+			printk("\ttcenum  = 0x%lx\n", (u64)tcenum);
+			printk("\ttce val = 0x%lx\n", tce.te_word );
+			show_stack(current, (unsigned long *)__get_SP());
+		}
+
+		tcenum++;
+	}
+}
+
+
+static void tce_freemulti_pSeriesLP(struct iommu_table *tbl, long tcenum, long npages)
+{
+	u64 rc;
+	union tce_entry tce;
+
+	tcenum <<= TCE_PAGE_FACTOR;
+	npages <<= TCE_PAGE_FACTOR;
+
+	tce.te_word = 0;
+
+	rc = plpar_tce_stuff((u64)tbl->it_index,
+			   (u64)tcenum << 12,
+			   tce.te_word,
+			   npages);
+
+	if (rc && printk_ratelimit()) {
+		printk("tce_freemulti_pSeriesLP: plpar_tce_stuff failed\n");
+		printk("\trc      = %ld\n", rc);
+		printk("\tindex   = 0x%lx\n", (u64)tbl->it_index);
+		printk("\tnpages  = 0x%lx\n", (u64)npages);
+		printk("\ttce val = 0x%lx\n", tce.te_word );
+		show_stack(current, (unsigned long *)__get_SP());
+	}
+}
+
+static void iommu_table_setparms(struct pci_controller *phb,
+				 struct device_node *dn,
+				 struct iommu_table *tbl) 
+{
+	struct device_node *node;
+	unsigned long *basep;
+	unsigned int *sizep;
+
+	node = (struct device_node *)phb->arch_data;
+
+	basep = (unsigned long *)get_property(node, "linux,tce-base", NULL);
+	sizep = (unsigned int *)get_property(node, "linux,tce-size", NULL);
+	if (basep == NULL || sizep == NULL) {
+		printk(KERN_ERR "PCI_DMA: iommu_table_setparms: %s has "
+				"missing tce entries !\n", dn->full_name);
+		return;
+	}
+
+	tbl->it_base = (unsigned long)__va(*basep);
+	memset((void *)tbl->it_base, 0, *sizep);
+
+	tbl->it_busno = phb->bus->number;
+	
+	/* Units of tce entries */
+	tbl->it_offset = phb->dma_window_base_cur >> PAGE_SHIFT;
+	
+	/* Test if we are going over 2GB of DMA space */
+	if (phb->dma_window_base_cur + phb->dma_window_size > 0x80000000ul) {
+		udbg_printf("PCI_DMA: Unexpected number of IOAs under this PHB.\n");
+		panic("PCI_DMA: Unexpected number of IOAs under this PHB.\n"); 
+	}
+	
+	phb->dma_window_base_cur += phb->dma_window_size;
+
+	/* Set the tce table size - measured in entries */
+	tbl->it_size = phb->dma_window_size >> PAGE_SHIFT;
+
+	tbl->it_index = 0;
+	tbl->it_blocksize = 16;
+	tbl->it_type = TCE_PCI;
+}
+
+/*
+ * iommu_table_setparms_lpar
+ *
+ * Function: On pSeries LPAR systems, return TCE table info, given a pci bus.
+ *
+ * ToDo: properly interpret the ibm,dma-window property.  The definition is:
+ *	logical-bus-number	(1 word)
+ *	phys-address		(#address-cells words)
+ *	size			(#cell-size words)
+ *
+ * Currently we hard code these sizes (more or less).
+ */
+static void iommu_table_setparms_lpar(struct pci_controller *phb,
+				      struct device_node *dn,
+				      struct iommu_table *tbl,
+				      unsigned int *dma_window)
+{
+	tbl->it_busno  = PCI_DN(dn)->bussubno;
+
+	/* TODO: Parse field size properties properly. */
+	tbl->it_size   = (((unsigned long)dma_window[4] << 32) |
+			   (unsigned long)dma_window[5]) >> PAGE_SHIFT;
+	tbl->it_offset = (((unsigned long)dma_window[2] << 32) |
+			   (unsigned long)dma_window[3]) >> PAGE_SHIFT;
+	tbl->it_base   = 0;
+	tbl->it_index  = dma_window[0];
+	tbl->it_blocksize  = 16;
+	tbl->it_type = TCE_PCI;
+}
+
+static void iommu_bus_setup_pSeries(struct pci_bus *bus)
+{
+	struct device_node *dn;
+	struct iommu_table *tbl;
+	struct device_node *isa_dn, *isa_dn_orig;
+	struct device_node *tmp;
+	struct pci_dn *pci;
+	int children;
+
+	DBG("iommu_bus_setup_pSeries, bus %p, bus->self %p\n", bus, bus->self);
+
+	dn = pci_bus_to_OF_node(bus);
+	pci = PCI_DN(dn);
+
+	if (bus->self) {
+		/* This is not a root bus, any setup will be done for the
+		 * device-side of the bridge in iommu_dev_setup_pSeries().
+		 */
+		return;
+	}
+
+	/* Check if the ISA bus on the system is under
+	 * this PHB.
+	 */
+	isa_dn = isa_dn_orig = of_find_node_by_type(NULL, "isa");
+
+	while (isa_dn && isa_dn != dn)
+		isa_dn = isa_dn->parent;
+
+	if (isa_dn_orig)
+		of_node_put(isa_dn_orig);
+
+	/* Count number of direct PCI children of the PHB.
+	 * All PCI device nodes have class-code property, so it's
+	 * an easy way to find them.
+	 */
+	for (children = 0, tmp = dn->child; tmp; tmp = tmp->sibling)
+		if (get_property(tmp, "class-code", NULL))
+			children++;
+
+	DBG("Children: %d\n", children);
+
+	/* Calculate amount of DMA window per slot. Each window must be
+	 * a power of two (due to pci_alloc_consistent requirements).
+	 *
+	 * Keep 256MB aside for PHBs with ISA.
+	 */
+
+	if (!isa_dn) {
+		/* No ISA/IDE - just set window size and return */
+		pci->phb->dma_window_size = 0x80000000ul; /* To be divided */
+
+		while (pci->phb->dma_window_size * children > 0x80000000ul)
+			pci->phb->dma_window_size >>= 1;
+		DBG("No ISA/IDE, window size is 0x%lx\n",
+			pci->phb->dma_window_size);
+		pci->phb->dma_window_base_cur = 0;
+
+		return;
+	}
+
+	/* If we have ISA, then we probably have an IDE
+	 * controller too. Allocate a 128MB table but
+	 * skip the first 128MB to avoid stepping on ISA
+	 * space.
+	 */
+	pci->phb->dma_window_size = 0x8000000ul;
+	pci->phb->dma_window_base_cur = 0x8000000ul;
+
+	tbl = kmalloc(sizeof(struct iommu_table), GFP_KERNEL);
+
+	iommu_table_setparms(pci->phb, dn, tbl);
+	pci->iommu_table = iommu_init_table(tbl);
+
+	/* Divide the rest (1.75GB) among the children */
+	pci->phb->dma_window_size = 0x80000000ul;
+	while (pci->phb->dma_window_size * children > 0x70000000ul)
+		pci->phb->dma_window_size >>= 1;
+
+	DBG("ISA/IDE, window size is 0x%lx\n", pci->phb->dma_window_size);
+
+}
+
+
+static void iommu_bus_setup_pSeriesLP(struct pci_bus *bus)
+{
+	struct iommu_table *tbl;
+	struct device_node *dn, *pdn;
+	struct pci_dn *ppci;
+	unsigned int *dma_window = NULL;
+
+	DBG("iommu_bus_setup_pSeriesLP, bus %p, bus->self %p\n", bus, bus->self);
+
+	dn = pci_bus_to_OF_node(bus);
+
+	/* Find nearest ibm,dma-window, walking up the device tree */
+	for (pdn = dn; pdn != NULL; pdn = pdn->parent) {
+		dma_window = (unsigned int *)get_property(pdn, "ibm,dma-window", NULL);
+		if (dma_window != NULL)
+			break;
+	}
+
+	if (dma_window == NULL) {
+		DBG("iommu_bus_setup_pSeriesLP: bus %s seems to have no ibm,dma-window property\n", dn->full_name);
+		return;
+	}
+
+	ppci = pdn->data;
+	if (!ppci->iommu_table) {
+		/* Bussubno hasn't been copied yet.
+		 * Do it now because iommu_table_setparms_lpar needs it.
+		 */
+
+		ppci->bussubno = bus->number;
+
+		tbl = (struct iommu_table *)kmalloc(sizeof(struct iommu_table),
+						    GFP_KERNEL);
+	
+		iommu_table_setparms_lpar(ppci->phb, pdn, tbl, dma_window);
+
+		ppci->iommu_table = iommu_init_table(tbl);
+	}
+
+	if (pdn != dn)
+		PCI_DN(dn)->iommu_table = ppci->iommu_table;
+}
+
+
+static void iommu_dev_setup_pSeries(struct pci_dev *dev)
+{
+	struct device_node *dn, *mydn;
+	struct iommu_table *tbl;
+
+	DBG("iommu_dev_setup_pSeries, dev %p (%s)\n", dev, pci_name(dev));
+
+	mydn = dn = pci_device_to_OF_node(dev);
+
+	/* If we're the direct child of a root bus, then we need to allocate
+	 * an iommu table ourselves. The bus setup code should have setup
+	 * the window sizes already.
+	 */
+	if (!dev->bus->self) {
+		DBG(" --> first child, no bridge. Allocating iommu table.\n");
+		tbl = kmalloc(sizeof(struct iommu_table), GFP_KERNEL);
+		iommu_table_setparms(PCI_DN(dn)->phb, dn, tbl);
+		PCI_DN(mydn)->iommu_table = iommu_init_table(tbl);
+
+		return;
+	}
+
+	/* If this device is further down the bus tree, search upwards until
+	 * an already allocated iommu table is found and use that.
+	 */
+
+	while (dn && dn->data && PCI_DN(dn)->iommu_table == NULL)
+		dn = dn->parent;
+
+	if (dn && dn->data) {
+		PCI_DN(mydn)->iommu_table = PCI_DN(dn)->iommu_table;
+	} else {
+		DBG("iommu_dev_setup_pSeries, dev %p (%s) has no iommu table\n", dev, pci_name(dev));
+	}
+}
+
+static int iommu_reconfig_notifier(struct notifier_block *nb, unsigned long action, void *node)
+{
+	int err = NOTIFY_OK;
+	struct device_node *np = node;
+	struct pci_dn *pci = np->data;
+
+	switch (action) {
+	case PSERIES_RECONFIG_REMOVE:
+		if (pci && pci->iommu_table &&
+		    get_property(np, "ibm,dma-window", NULL))
+			iommu_free_table(np);
+		break;
+	default:
+		err = NOTIFY_DONE;
+		break;
+	}
+	return err;
+}
+
+static struct notifier_block iommu_reconfig_nb = {
+	.notifier_call = iommu_reconfig_notifier,
+};
+
+static void iommu_dev_setup_pSeriesLP(struct pci_dev *dev)
+{
+	struct device_node *pdn, *dn;
+	struct iommu_table *tbl;
+	int *dma_window = NULL;
+	struct pci_dn *pci;
+
+	DBG("iommu_dev_setup_pSeriesLP, dev %p (%s)\n", dev, pci_name(dev));
+
+	/* dev setup for LPAR is a little tricky, since the device tree might
+	 * contain the dma-window properties per-device and not neccesarily
+	 * for the bus. So we need to search upwards in the tree until we
+	 * either hit a dma-window property, OR find a parent with a table
+	 * already allocated.
+	 */
+	dn = pci_device_to_OF_node(dev);
+
+	for (pdn = dn; pdn && pdn->data && !PCI_DN(pdn)->iommu_table;
+	     pdn = pdn->parent) {
+		dma_window = (unsigned int *)
+			get_property(pdn, "ibm,dma-window", NULL);
+		if (dma_window)
+			break;
+	}
+
+	/* Check for parent == NULL so we don't try to setup the empty EADS
+	 * slots on POWER4 machines.
+	 */
+	if (dma_window == NULL || pdn->parent == NULL) {
+		DBG("No dma window for device, linking to parent\n");
+		PCI_DN(dn)->iommu_table = PCI_DN(pdn)->iommu_table;
+		return;
+	} else {
+		DBG("Found DMA window, allocating table\n");
+	}
+
+	pci = pdn->data;
+	if (!pci->iommu_table) {
+		/* iommu_table_setparms_lpar needs bussubno. */
+		pci->bussubno = pci->phb->bus->number;
+
+		tbl = (struct iommu_table *)kmalloc(sizeof(struct iommu_table),
+						    GFP_KERNEL);
+
+		iommu_table_setparms_lpar(pci->phb, pdn, tbl, dma_window);
+
+		pci->iommu_table = iommu_init_table(tbl);
+	}
+
+	if (pdn != dn)
+		PCI_DN(dn)->iommu_table = pci->iommu_table;
+}
+
+static void iommu_bus_setup_null(struct pci_bus *b) { }
+static void iommu_dev_setup_null(struct pci_dev *d) { }
+
+/* These are called very early. */
+void iommu_init_early_pSeries(void)
+{
+	if (of_chosen && get_property(of_chosen, "linux,iommu-off", NULL)) {
+		/* Direct I/O, IOMMU off */
+		ppc_md.iommu_dev_setup = iommu_dev_setup_null;
+		ppc_md.iommu_bus_setup = iommu_bus_setup_null;
+		pci_direct_iommu_init();
+
+		return;
+	}
+
+	if (systemcfg->platform & PLATFORM_LPAR) {
+		if (firmware_has_feature(FW_FEATURE_MULTITCE)) {
+			ppc_md.tce_build = tce_buildmulti_pSeriesLP;
+			ppc_md.tce_free	 = tce_freemulti_pSeriesLP;
+		} else {
+			ppc_md.tce_build = tce_build_pSeriesLP;
+			ppc_md.tce_free	 = tce_free_pSeriesLP;
+		}
+		ppc_md.iommu_bus_setup = iommu_bus_setup_pSeriesLP;
+		ppc_md.iommu_dev_setup = iommu_dev_setup_pSeriesLP;
+	} else {
+		ppc_md.tce_build = tce_build_pSeries;
+		ppc_md.tce_free  = tce_free_pSeries;
+		ppc_md.iommu_bus_setup = iommu_bus_setup_pSeries;
+		ppc_md.iommu_dev_setup = iommu_dev_setup_pSeries;
+	}
+
+
+	pSeries_reconfig_notifier_register(&iommu_reconfig_nb);
+
+	pci_iommu_init();
+}
+
diff --git a/arch/powerpc/platforms/pseries/lpar.c b/arch/powerpc/platforms/pseries/lpar.c
new file mode 100644
index 000000000000..e384a5a91796
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/lpar.c
@@ -0,0 +1,514 @@
+/*
+ * pSeries_lpar.c
+ * Copyright (C) 2001 Todd Inglett, IBM Corporation
+ *
+ * pSeries LPAR support.
+ * 
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ * 
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ * 
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#define DEBUG
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/dma-mapping.h>
+#include <asm/processor.h>
+#include <asm/mmu.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/machdep.h>
+#include <asm/abs_addr.h>
+#include <asm/mmu_context.h>
+#include <asm/ppcdebug.h>
+#include <asm/iommu.h>
+#include <asm/tlbflush.h>
+#include <asm/tlb.h>
+#include <asm/prom.h>
+#include <asm/abs_addr.h>
+#include <asm/cputable.h>
+
+#include "plpar_wrappers.h"
+
+#ifdef DEBUG
+#define DBG(fmt...) udbg_printf(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+/* in pSeries_hvCall.S */
+EXPORT_SYMBOL(plpar_hcall);
+EXPORT_SYMBOL(plpar_hcall_4out);
+EXPORT_SYMBOL(plpar_hcall_norets);
+EXPORT_SYMBOL(plpar_hcall_8arg_2ret);
+
+extern void pSeries_find_serial_port(void);
+
+
+int vtermno;	/* virtual terminal# for udbg  */
+
+#define __ALIGNED__ __attribute__((__aligned__(sizeof(long))))
+static void udbg_hvsi_putc(unsigned char c)
+{
+	/* packet's seqno isn't used anyways */
+	uint8_t packet[] __ALIGNED__ = { 0xff, 5, 0, 0, c };
+	int rc;
+
+	if (c == '\n')
+		udbg_hvsi_putc('\r');
+
+	do {
+		rc = plpar_put_term_char(vtermno, sizeof(packet), packet);
+	} while (rc == H_Busy);
+}
+
+static long hvsi_udbg_buf_len;
+static uint8_t hvsi_udbg_buf[256];
+
+static int udbg_hvsi_getc_poll(void)
+{
+	unsigned char ch;
+	int rc, i;
+
+	if (hvsi_udbg_buf_len == 0) {
+		rc = plpar_get_term_char(vtermno, &hvsi_udbg_buf_len, hvsi_udbg_buf);
+		if (rc != H_Success || hvsi_udbg_buf[0] != 0xff) {
+			/* bad read or non-data packet */
+			hvsi_udbg_buf_len = 0;
+		} else {
+			/* remove the packet header */
+			for (i = 4; i < hvsi_udbg_buf_len; i++)
+				hvsi_udbg_buf[i-4] = hvsi_udbg_buf[i];
+			hvsi_udbg_buf_len -= 4;
+		}
+	}
+
+	if (hvsi_udbg_buf_len <= 0 || hvsi_udbg_buf_len > 256) {
+		/* no data ready */
+		hvsi_udbg_buf_len = 0;
+		return -1;
+	}
+
+	ch = hvsi_udbg_buf[0];
+	/* shift remaining data down */
+	for (i = 1; i < hvsi_udbg_buf_len; i++) {
+		hvsi_udbg_buf[i-1] = hvsi_udbg_buf[i];
+	}
+	hvsi_udbg_buf_len--;
+
+	return ch;
+}
+
+static unsigned char udbg_hvsi_getc(void)
+{
+	int ch;
+	for (;;) {
+		ch = udbg_hvsi_getc_poll();
+		if (ch == -1) {
+			/* This shouldn't be needed...but... */
+			volatile unsigned long delay;
+			for (delay=0; delay < 2000000; delay++)
+				;
+		} else {
+			return ch;
+		}
+	}
+}
+
+static void udbg_putcLP(unsigned char c)
+{
+	char buf[16];
+	unsigned long rc;
+
+	if (c == '\n')
+		udbg_putcLP('\r');
+
+	buf[0] = c;
+	do {
+		rc = plpar_put_term_char(vtermno, 1, buf);
+	} while(rc == H_Busy);
+}
+
+/* Buffered chars getc */
+static long inbuflen;
+static long inbuf[2];	/* must be 2 longs */
+
+static int udbg_getc_pollLP(void)
+{
+	/* The interface is tricky because it may return up to 16 chars.
+	 * We save them statically for future calls to udbg_getc().
+	 */
+	char ch, *buf = (char *)inbuf;
+	int i;
+	long rc;
+	if (inbuflen == 0) {
+		/* get some more chars. */
+		inbuflen = 0;
+		rc = plpar_get_term_char(vtermno, &inbuflen, buf);
+		if (rc != H_Success)
+			inbuflen = 0;	/* otherwise inbuflen is garbage */
+	}
+	if (inbuflen <= 0 || inbuflen > 16) {
+		/* Catch error case as well as other oddities (corruption) */
+		inbuflen = 0;
+		return -1;
+	}
+	ch = buf[0];
+	for (i = 1; i < inbuflen; i++)	/* shuffle them down. */
+		buf[i-1] = buf[i];
+	inbuflen--;
+	return ch;
+}
+
+static unsigned char udbg_getcLP(void)
+{
+	int ch;
+	for (;;) {
+		ch = udbg_getc_pollLP();
+		if (ch == -1) {
+			/* This shouldn't be needed...but... */
+			volatile unsigned long delay;
+			for (delay=0; delay < 2000000; delay++)
+				;
+		} else {
+			return ch;
+		}
+	}
+}
+
+/* call this from early_init() for a working debug console on
+ * vterm capable LPAR machines
+ */
+void udbg_init_debug_lpar(void)
+{
+	vtermno = 0;
+	udbg_putc = udbg_putcLP;
+	udbg_getc = udbg_getcLP;
+	udbg_getc_poll = udbg_getc_pollLP;
+}
+
+/* returns 0 if couldn't find or use /chosen/stdout as console */
+int find_udbg_vterm(void)
+{
+	struct device_node *stdout_node;
+	u32 *termno;
+	char *name;
+	int found = 0;
+
+	/* find the boot console from /chosen/stdout */
+	if (!of_chosen)
+		return 0;
+	name = (char *)get_property(of_chosen, "linux,stdout-path", NULL);
+	if (name == NULL)
+		return 0;
+	stdout_node = of_find_node_by_path(name);
+	if (!stdout_node)
+		return 0;
+
+	/* now we have the stdout node; figure out what type of device it is. */
+	name = (char *)get_property(stdout_node, "name", NULL);
+	if (!name) {
+		printk(KERN_WARNING "stdout node missing 'name' property!\n");
+		goto out;
+	}
+
+	if (strncmp(name, "vty", 3) == 0) {
+		if (device_is_compatible(stdout_node, "hvterm1")) {
+			termno = (u32 *)get_property(stdout_node, "reg", NULL);
+			if (termno) {
+				vtermno = termno[0];
+				udbg_putc = udbg_putcLP;
+				udbg_getc = udbg_getcLP;
+				udbg_getc_poll = udbg_getc_pollLP;
+				found = 1;
+			}
+		} else if (device_is_compatible(stdout_node, "hvterm-protocol")) {
+			termno = (u32 *)get_property(stdout_node, "reg", NULL);
+			if (termno) {
+				vtermno = termno[0];
+				udbg_putc = udbg_hvsi_putc;
+				udbg_getc = udbg_hvsi_getc;
+				udbg_getc_poll = udbg_hvsi_getc_poll;
+				found = 1;
+			}
+		}
+	} else if (strncmp(name, "serial", 6)) {
+		/* XXX fix ISA serial console */
+		printk(KERN_WARNING "serial stdout on LPAR ('%s')! "
+				"can't print udbg messages\n",
+		       stdout_node->full_name);
+	} else {
+		printk(KERN_WARNING "don't know how to print to stdout '%s'\n",
+		       stdout_node->full_name);
+	}
+
+out:
+	of_node_put(stdout_node);
+	return found;
+}
+
+void vpa_init(int cpu)
+{
+	int hwcpu = get_hard_smp_processor_id(cpu);
+	unsigned long vpa = __pa(&paca[cpu].lppaca);
+	long ret;
+
+	if (cpu_has_feature(CPU_FTR_ALTIVEC))
+		paca[cpu].lppaca.vmxregs_in_use = 1;
+
+	ret = register_vpa(hwcpu, vpa);
+
+	if (ret)
+		printk(KERN_ERR "WARNING: vpa_init: VPA registration for "
+				"cpu %d (hw %d) of area %lx returns %ld\n",
+				cpu, hwcpu, vpa, ret);
+}
+
+long pSeries_lpar_hpte_insert(unsigned long hpte_group,
+			      unsigned long va, unsigned long prpn,
+			      unsigned long vflags, unsigned long rflags)
+{
+	unsigned long lpar_rc;
+	unsigned long flags;
+	unsigned long slot;
+	unsigned long hpte_v, hpte_r;
+	unsigned long dummy0, dummy1;
+
+	hpte_v = ((va >> 23) << HPTE_V_AVPN_SHIFT) | vflags | HPTE_V_VALID;
+	if (vflags & HPTE_V_LARGE)
+		hpte_v &= ~(1UL << HPTE_V_AVPN_SHIFT);
+
+	hpte_r = (prpn << HPTE_R_RPN_SHIFT) | rflags;
+
+	/* Now fill in the actual HPTE */
+	/* Set CEC cookie to 0         */
+	/* Zero page = 0               */
+	/* I-cache Invalidate = 0      */
+	/* I-cache synchronize = 0     */
+	/* Exact = 0                   */
+	flags = 0;
+
+	/* XXX why is this here? - Anton */
+	if (rflags & (_PAGE_GUARDED|_PAGE_NO_CACHE))
+		hpte_r &= ~_PAGE_COHERENT;
+
+	lpar_rc = plpar_hcall(H_ENTER, flags, hpte_group, hpte_v,
+			      hpte_r, &slot, &dummy0, &dummy1);
+
+	if (unlikely(lpar_rc == H_PTEG_Full))
+		return -1;
+
+	/*
+	 * Since we try and ioremap PHBs we don't own, the pte insert
+	 * will fail. However we must catch the failure in hash_page
+	 * or we will loop forever, so return -2 in this case.
+	 */
+	if (unlikely(lpar_rc != H_Success))
+		return -2;
+
+	/* Because of iSeries, we have to pass down the secondary
+	 * bucket bit here as well
+	 */
+	return (slot & 7) | (!!(vflags & HPTE_V_SECONDARY) << 3);
+}
+
+static DEFINE_SPINLOCK(pSeries_lpar_tlbie_lock);
+
+static long pSeries_lpar_hpte_remove(unsigned long hpte_group)
+{
+	unsigned long slot_offset;
+	unsigned long lpar_rc;
+	int i;
+	unsigned long dummy1, dummy2;
+
+	/* pick a random slot to start at */
+	slot_offset = mftb() & 0x7;
+
+	for (i = 0; i < HPTES_PER_GROUP; i++) {
+
+		/* don't remove a bolted entry */
+		lpar_rc = plpar_pte_remove(H_ANDCOND, hpte_group + slot_offset,
+					   (0x1UL << 4), &dummy1, &dummy2);
+
+		if (lpar_rc == H_Success)
+			return i;
+
+		BUG_ON(lpar_rc != H_Not_Found);
+
+		slot_offset++;
+		slot_offset &= 0x7;
+	}
+
+	return -1;
+}
+
+static void pSeries_lpar_hptab_clear(void)
+{
+	unsigned long size_bytes = 1UL << ppc64_pft_size;
+	unsigned long hpte_count = size_bytes >> 4;
+	unsigned long dummy1, dummy2;
+	int i;
+
+	/* TODO: Use bulk call */
+	for (i = 0; i < hpte_count; i++)
+		plpar_pte_remove(0, i, 0, &dummy1, &dummy2);
+}
+
+/*
+ * NOTE: for updatepp ops we are fortunate that the linux "newpp" bits and
+ * the low 3 bits of flags happen to line up.  So no transform is needed.
+ * We can probably optimize here and assume the high bits of newpp are
+ * already zero.  For now I am paranoid.
+ */
+static long pSeries_lpar_hpte_updatepp(unsigned long slot, unsigned long newpp,
+				       unsigned long va, int large, int local)
+{
+	unsigned long lpar_rc;
+	unsigned long flags = (newpp & 7) | H_AVPN;
+	unsigned long avpn = va >> 23;
+
+	if (large)
+		avpn &= ~0x1UL;
+
+	lpar_rc = plpar_pte_protect(flags, slot, (avpn << 7));
+
+	if (lpar_rc == H_Not_Found)
+		return -1;
+
+	BUG_ON(lpar_rc != H_Success);
+
+	return 0;
+}
+
+static unsigned long pSeries_lpar_hpte_getword0(unsigned long slot)
+{
+	unsigned long dword0;
+	unsigned long lpar_rc;
+	unsigned long dummy_word1;
+	unsigned long flags;
+
+	/* Read 1 pte at a time                        */
+	/* Do not need RPN to logical page translation */
+	/* No cross CEC PFT access                     */
+	flags = 0;
+
+	lpar_rc = plpar_pte_read(flags, slot, &dword0, &dummy_word1);
+
+	BUG_ON(lpar_rc != H_Success);
+
+	return dword0;
+}
+
+static long pSeries_lpar_hpte_find(unsigned long vpn)
+{
+	unsigned long hash;
+	unsigned long i, j;
+	long slot;
+	unsigned long hpte_v;
+
+	hash = hpt_hash(vpn, 0);
+
+	for (j = 0; j < 2; j++) {
+		slot = (hash & htab_hash_mask) * HPTES_PER_GROUP;
+		for (i = 0; i < HPTES_PER_GROUP; i++) {
+			hpte_v = pSeries_lpar_hpte_getword0(slot);
+
+			if ((HPTE_V_AVPN_VAL(hpte_v) == (vpn >> 11))
+			    && (hpte_v & HPTE_V_VALID)
+			    && (!!(hpte_v & HPTE_V_SECONDARY) == j)) {
+				/* HPTE matches */
+				if (j)
+					slot = -slot;
+				return slot;
+			}
+			++slot;
+		}
+		hash = ~hash;
+	}
+
+	return -1;
+} 
+
+static void pSeries_lpar_hpte_updateboltedpp(unsigned long newpp,
+					     unsigned long ea)
+{
+	unsigned long lpar_rc;
+	unsigned long vsid, va, vpn, flags;
+	long slot;
+
+	vsid = get_kernel_vsid(ea);
+	va = (vsid << 28) | (ea & 0x0fffffff);
+	vpn = va >> PAGE_SHIFT;
+
+	slot = pSeries_lpar_hpte_find(vpn);
+	BUG_ON(slot == -1);
+
+	flags = newpp & 7;
+	lpar_rc = plpar_pte_protect(flags, slot, 0);
+
+	BUG_ON(lpar_rc != H_Success);
+}
+
+static void pSeries_lpar_hpte_invalidate(unsigned long slot, unsigned long va,
+					 int large, int local)
+{
+	unsigned long avpn = va >> 23;
+	unsigned long lpar_rc;
+	unsigned long dummy1, dummy2;
+
+	if (large)
+		avpn &= ~0x1UL;
+
+	lpar_rc = plpar_pte_remove(H_AVPN, slot, (avpn << 7), &dummy1,
+				   &dummy2);
+
+	if (lpar_rc == H_Not_Found)
+		return;
+
+	BUG_ON(lpar_rc != H_Success);
+}
+
+/*
+ * Take a spinlock around flushes to avoid bouncing the hypervisor tlbie
+ * lock.
+ */
+void pSeries_lpar_flush_hash_range(unsigned long number, int local)
+{
+	int i;
+	unsigned long flags = 0;
+	struct ppc64_tlb_batch *batch = &__get_cpu_var(ppc64_tlb_batch);
+	int lock_tlbie = !cpu_has_feature(CPU_FTR_LOCKLESS_TLBIE);
+
+	if (lock_tlbie)
+		spin_lock_irqsave(&pSeries_lpar_tlbie_lock, flags);
+
+	for (i = 0; i < number; i++)
+		flush_hash_page(batch->vaddr[i], batch->pte[i], local);
+
+	if (lock_tlbie)
+		spin_unlock_irqrestore(&pSeries_lpar_tlbie_lock, flags);
+}
+
+void hpte_init_lpar(void)
+{
+	ppc_md.hpte_invalidate	= pSeries_lpar_hpte_invalidate;
+	ppc_md.hpte_updatepp	= pSeries_lpar_hpte_updatepp;
+	ppc_md.hpte_updateboltedpp = pSeries_lpar_hpte_updateboltedpp;
+	ppc_md.hpte_insert	= pSeries_lpar_hpte_insert;
+	ppc_md.hpte_remove	= pSeries_lpar_hpte_remove;
+	ppc_md.flush_hash_range	= pSeries_lpar_flush_hash_range;
+	ppc_md.hpte_clear_all   = pSeries_lpar_hptab_clear;
+
+	htab_finish_init();
+}
diff --git a/arch/powerpc/platforms/pseries/nvram.c b/arch/powerpc/platforms/pseries/nvram.c
new file mode 100644
index 000000000000..18abfb1f4e24
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/nvram.c
@@ -0,0 +1,148 @@
+/*
+ *  c 2001 PPC 64 Team, IBM Corp
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ *
+ * /dev/nvram driver for PPC64
+ *
+ * This perhaps should live in drivers/char
+ */
+
+
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <asm/uaccess.h>
+#include <asm/nvram.h>
+#include <asm/rtas.h>
+#include <asm/prom.h>
+#include <asm/machdep.h>
+
+static unsigned int nvram_size;
+static int nvram_fetch, nvram_store;
+static char nvram_buf[NVRW_CNT];	/* assume this is in the first 4GB */
+static DEFINE_SPINLOCK(nvram_lock);
+
+
+static ssize_t pSeries_nvram_read(char *buf, size_t count, loff_t *index)
+{
+	unsigned int i;
+	unsigned long len;
+	int done;
+	unsigned long flags;
+	char *p = buf;
+
+
+	if (nvram_size == 0 || nvram_fetch == RTAS_UNKNOWN_SERVICE)
+		return -ENODEV;
+
+	if (*index >= nvram_size)
+		return 0;
+
+	i = *index;
+	if (i + count > nvram_size)
+		count = nvram_size - i;
+
+	spin_lock_irqsave(&nvram_lock, flags);
+
+	for (; count != 0; count -= len) {
+		len = count;
+		if (len > NVRW_CNT)
+			len = NVRW_CNT;
+		
+		if ((rtas_call(nvram_fetch, 3, 2, &done, i, __pa(nvram_buf),
+			       len) != 0) || len != done) {
+			spin_unlock_irqrestore(&nvram_lock, flags);
+			return -EIO;
+		}
+		
+		memcpy(p, nvram_buf, len);
+
+		p += len;
+		i += len;
+	}
+
+	spin_unlock_irqrestore(&nvram_lock, flags);
+	
+	*index = i;
+	return p - buf;
+}
+
+static ssize_t pSeries_nvram_write(char *buf, size_t count, loff_t *index)
+{
+	unsigned int i;
+	unsigned long len;
+	int done;
+	unsigned long flags;
+	const char *p = buf;
+
+	if (nvram_size == 0 || nvram_store == RTAS_UNKNOWN_SERVICE)
+		return -ENODEV;
+
+	if (*index >= nvram_size)
+		return 0;
+
+	i = *index;
+	if (i + count > nvram_size)
+		count = nvram_size - i;
+
+	spin_lock_irqsave(&nvram_lock, flags);
+
+	for (; count != 0; count -= len) {
+		len = count;
+		if (len > NVRW_CNT)
+			len = NVRW_CNT;
+
+		memcpy(nvram_buf, p, len);
+
+		if ((rtas_call(nvram_store, 3, 2, &done, i, __pa(nvram_buf),
+			       len) != 0) || len != done) {
+			spin_unlock_irqrestore(&nvram_lock, flags);
+			return -EIO;
+		}
+		
+		p += len;
+		i += len;
+	}
+	spin_unlock_irqrestore(&nvram_lock, flags);
+	
+	*index = i;
+	return p - buf;
+}
+
+static ssize_t pSeries_nvram_get_size(void)
+{
+	return nvram_size ? nvram_size : -ENODEV;
+}
+
+int __init pSeries_nvram_init(void)
+{
+	struct device_node *nvram;
+	unsigned int *nbytes_p, proplen;
+
+	nvram = of_find_node_by_type(NULL, "nvram");
+	if (nvram == NULL)
+		return -ENODEV;
+
+	nbytes_p = (unsigned int *)get_property(nvram, "#bytes", &proplen);
+	if (nbytes_p == NULL || proplen != sizeof(unsigned int))
+		return -EIO;
+
+	nvram_size = *nbytes_p;
+
+	nvram_fetch = rtas_token("nvram-fetch");
+	nvram_store = rtas_token("nvram-store");
+	printk(KERN_INFO "PPC64 nvram contains %d bytes\n", nvram_size);
+	of_node_put(nvram);
+
+	ppc_md.nvram_read	= pSeries_nvram_read;
+	ppc_md.nvram_write	= pSeries_nvram_write;
+	ppc_md.nvram_size	= pSeries_nvram_get_size;
+
+	return 0;
+}
diff --git a/arch/powerpc/platforms/pseries/pci.c b/arch/powerpc/platforms/pseries/pci.c
new file mode 100644
index 000000000000..c198656a3bb5
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/pci.c
@@ -0,0 +1,142 @@
+/*
+ * arch/ppc64/kernel/pSeries_pci.c
+ *
+ * Copyright (C) 2001 Dave Engebretsen, IBM Corporation
+ * Copyright (C) 2003 Anton Blanchard <anton@au.ibm.com>, IBM
+ *
+ * pSeries specific routines for PCI.
+ * 
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *    
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ * 
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/init.h>
+#include <linux/ioport.h>
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/string.h>
+
+#include <asm/pci-bridge.h>
+#include <asm/prom.h>
+#include <asm/ppc-pci.h>
+
+static int __devinitdata s7a_workaround = -1;
+
+#if 0
+void pcibios_name_device(struct pci_dev *dev)
+{
+	struct device_node *dn;
+
+	/*
+	 * Add IBM loc code (slot) as a prefix to the device names for service
+	 */
+	dn = pci_device_to_OF_node(dev);
+	if (dn) {
+		char *loc_code = get_property(dn, "ibm,loc-code", 0);
+		if (loc_code) {
+			int loc_len = strlen(loc_code);
+			if (loc_len < sizeof(dev->dev.name)) {
+				memmove(dev->dev.name+loc_len+1, dev->dev.name,
+					sizeof(dev->dev.name)-loc_len-1);
+				memcpy(dev->dev.name, loc_code, loc_len);
+				dev->dev.name[loc_len] = ' ';
+				dev->dev.name[sizeof(dev->dev.name)-1] = '\0';
+			}
+		}
+	}
+}   
+DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, pcibios_name_device);
+#endif
+
+static void __devinit check_s7a(void)
+{
+	struct device_node *root;
+	char *model;
+
+	s7a_workaround = 0;
+	root = of_find_node_by_path("/");
+	if (root) {
+		model = get_property(root, "model", NULL);
+		if (model && !strcmp(model, "IBM,7013-S7A"))
+			s7a_workaround = 1;
+		of_node_put(root);
+	}
+}
+
+void __devinit pSeries_irq_bus_setup(struct pci_bus *bus)
+{
+	struct pci_dev *dev;
+
+	if (s7a_workaround < 0)
+		check_s7a();
+	list_for_each_entry(dev, &bus->devices, bus_list) {
+		pci_read_irq_line(dev);
+		if (s7a_workaround) {
+			if (dev->irq > 16) {
+				dev->irq -= 3;
+				pci_write_config_byte(dev, PCI_INTERRUPT_LINE,
+					dev->irq);
+			}
+		}
+	}
+}
+
+static void __init pSeries_request_regions(void)
+{
+	if (!isa_io_base)
+		return;
+
+	request_region(0x20,0x20,"pic1");
+	request_region(0xa0,0x20,"pic2");
+	request_region(0x00,0x20,"dma1");
+	request_region(0x40,0x20,"timer");
+	request_region(0x80,0x10,"dma page reg");
+	request_region(0xc0,0x20,"dma2");
+}
+
+void __init pSeries_final_fixup(void)
+{
+	phbs_remap_io();
+	pSeries_request_regions();
+
+	pci_addr_cache_build();
+}
+
+/*
+ * Assume the winbond 82c105 is the IDE controller on a
+ * p610.  We should probably be more careful in case
+ * someone tries to plug in a similar adapter.
+ */
+static void fixup_winbond_82c105(struct pci_dev* dev)
+{
+	int i;
+	unsigned int reg;
+
+	if (!(systemcfg->platform & PLATFORM_PSERIES))
+		return;
+
+	printk("Using INTC for W82c105 IDE controller.\n");
+	pci_read_config_dword(dev, 0x40, &reg);
+	/* Enable LEGIRQ to use INTC instead of ISA interrupts */
+	pci_write_config_dword(dev, 0x40, reg | (1<<11));
+
+	for (i = 0; i < DEVICE_COUNT_RESOURCE; ++i) {
+		/* zap the 2nd function of the winbond chip */
+		if (dev->resource[i].flags & IORESOURCE_IO
+		    && dev->bus->number == 0 && dev->devfn == 0x81)
+			dev->resource[i].flags &= ~IORESOURCE_IO;
+	}
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_WINBOND, PCI_DEVICE_ID_WINBOND_82C105,
+			 fixup_winbond_82c105);
diff --git a/arch/powerpc/platforms/pseries/plpar_wrappers.h b/arch/powerpc/platforms/pseries/plpar_wrappers.h
new file mode 100644
index 000000000000..382f8c5b0e7c
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/plpar_wrappers.h
@@ -0,0 +1,120 @@
+#ifndef _PSERIES_PLPAR_WRAPPERS_H
+#define _PSERIES_PLPAR_WRAPPERS_H
+
+#include <asm/hvcall.h>
+
+static inline long poll_pending(void)
+{
+	unsigned long dummy;
+	return plpar_hcall(H_POLL_PENDING, 0, 0, 0, 0, &dummy, &dummy, &dummy);
+}
+
+static inline long prod_processor(void)
+{
+	plpar_hcall_norets(H_PROD);
+	return 0;
+}
+
+static inline long cede_processor(void)
+{
+	plpar_hcall_norets(H_CEDE);
+	return 0;
+}
+
+static inline long vpa_call(unsigned long flags, unsigned long cpu,
+		unsigned long vpa)
+{
+	/* flags are in bits 16-18 (counting from most significant bit) */
+	flags = flags << (63 - 18);
+
+	return plpar_hcall_norets(H_REGISTER_VPA, flags, cpu, vpa);
+}
+
+static inline long unregister_vpa(unsigned long cpu, unsigned long vpa)
+{
+	return vpa_call(0x5, cpu, vpa);
+}
+
+static inline long register_vpa(unsigned long cpu, unsigned long vpa)
+{
+	return vpa_call(0x1, cpu, vpa);
+}
+
+extern void vpa_init(int cpu);
+
+static inline long plpar_pte_remove(unsigned long flags, unsigned long ptex,
+		unsigned long avpn, unsigned long *old_pteh_ret,
+		unsigned long *old_ptel_ret)
+{
+	unsigned long dummy;
+	return plpar_hcall(H_REMOVE, flags, ptex, avpn, 0, old_pteh_ret,
+			old_ptel_ret, &dummy);
+}
+
+static inline long plpar_pte_read(unsigned long flags, unsigned long ptex,
+		unsigned long *old_pteh_ret, unsigned long *old_ptel_ret)
+{
+	unsigned long dummy;
+	return plpar_hcall(H_READ, flags, ptex, 0, 0, old_pteh_ret,
+			old_ptel_ret, &dummy);
+}
+
+static inline long plpar_pte_protect(unsigned long flags, unsigned long ptex,
+		unsigned long avpn)
+{
+	return plpar_hcall_norets(H_PROTECT, flags, ptex, avpn);
+}
+
+static inline long plpar_tce_get(unsigned long liobn, unsigned long ioba,
+		unsigned long *tce_ret)
+{
+	unsigned long dummy;
+	return plpar_hcall(H_GET_TCE, liobn, ioba, 0, 0, tce_ret, &dummy,
+			&dummy);
+}
+
+static inline long plpar_tce_put(unsigned long liobn, unsigned long ioba,
+		unsigned long tceval)
+{
+	return plpar_hcall_norets(H_PUT_TCE, liobn, ioba, tceval);
+}
+
+static inline long plpar_tce_put_indirect(unsigned long liobn,
+		unsigned long ioba, unsigned long page, unsigned long count)
+{
+	return plpar_hcall_norets(H_PUT_TCE_INDIRECT, liobn, ioba, page, count);
+}
+
+static inline long plpar_tce_stuff(unsigned long liobn, unsigned long ioba,
+		unsigned long tceval, unsigned long count)
+{
+	return plpar_hcall_norets(H_STUFF_TCE, liobn, ioba, tceval, count);
+}
+
+static inline long plpar_get_term_char(unsigned long termno,
+		unsigned long *len_ret, char *buf_ret)
+{
+	unsigned long *lbuf = (unsigned long *)buf_ret;	/* TODO: alignment? */
+	return plpar_hcall(H_GET_TERM_CHAR, termno, 0, 0, 0, len_ret,
+			lbuf + 0, lbuf + 1);
+}
+
+static inline long plpar_put_term_char(unsigned long termno, unsigned long len,
+		const char *buffer)
+{
+	unsigned long *lbuf = (unsigned long *)buffer;	/* TODO: alignment? */
+	return plpar_hcall_norets(H_PUT_TERM_CHAR, termno, len, lbuf[0],
+			lbuf[1]);
+}
+
+static inline long plpar_set_xdabr(unsigned long address, unsigned long flags)
+{
+	return plpar_hcall_norets(H_SET_XDABR, address, flags);
+}
+
+static inline long plpar_set_dabr(unsigned long val)
+{
+	return plpar_hcall_norets(H_SET_DABR, val);
+}
+
+#endif /* _PSERIES_PLPAR_WRAPPERS_H */
diff --git a/arch/powerpc/platforms/pseries/ras.c b/arch/powerpc/platforms/pseries/ras.c
new file mode 100644
index 000000000000..6562ff4b0a82
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/ras.c
@@ -0,0 +1,352 @@
+/*
+ * Copyright (C) 2001 Dave Engebretsen IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+/* Change Activity:
+ * 2001/09/21 : engebret : Created with minimal EPOW and HW exception support.
+ * End Change Activity
+ */
+
+#include <linux/errno.h>
+#include <linux/threads.h>
+#include <linux/kernel_stat.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/ioport.h>
+#include <linux/interrupt.h>
+#include <linux/timex.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/irq.h>
+#include <linux/random.h>
+#include <linux/sysrq.h>
+#include <linux/bitops.h>
+
+#include <asm/uaccess.h>
+#include <asm/system.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/irq.h>
+#include <asm/cache.h>
+#include <asm/prom.h>
+#include <asm/ptrace.h>
+#include <asm/machdep.h>
+#include <asm/rtas.h>
+#include <asm/ppcdebug.h>
+
+static unsigned char ras_log_buf[RTAS_ERROR_LOG_MAX];
+static DEFINE_SPINLOCK(ras_log_buf_lock);
+
+char mce_data_buf[RTAS_ERROR_LOG_MAX]
+;
+/* This is true if we are using the firmware NMI handler (typically LPAR) */
+extern int fwnmi_active;
+
+static int ras_get_sensor_state_token;
+static int ras_check_exception_token;
+
+#define EPOW_SENSOR_TOKEN	9
+#define EPOW_SENSOR_INDEX	0
+#define RAS_VECTOR_OFFSET	0x500
+
+static irqreturn_t ras_epow_interrupt(int irq, void *dev_id,
+					struct pt_regs * regs);
+static irqreturn_t ras_error_interrupt(int irq, void *dev_id,
+					struct pt_regs * regs);
+
+/* #define DEBUG */
+
+static void request_ras_irqs(struct device_node *np, char *propname,
+			irqreturn_t (*handler)(int, void *, struct pt_regs *),
+			const char *name)
+{
+	unsigned int *ireg, len, i;
+	int virq, n_intr;
+
+	ireg = (unsigned int *)get_property(np, propname, &len);
+	if (ireg == NULL)
+		return;
+	n_intr = prom_n_intr_cells(np);
+	len /= n_intr * sizeof(*ireg);
+
+	for (i = 0; i < len; i++) {
+		virq = virt_irq_create_mapping(*ireg);
+		if (virq == NO_IRQ) {
+			printk(KERN_ERR "Unable to allocate interrupt "
+			       "number for %s\n", np->full_name);
+			return;
+		}
+		if (request_irq(irq_offset_up(virq), handler, 0, name, NULL)) {
+			printk(KERN_ERR "Unable to request interrupt %d for "
+			       "%s\n", irq_offset_up(virq), np->full_name);
+			return;
+		}
+		ireg += n_intr;
+	}
+}
+
+/*
+ * Initialize handlers for the set of interrupts caused by hardware errors
+ * and power system events.
+ */
+static int __init init_ras_IRQ(void)
+{
+	struct device_node *np;
+
+	ras_get_sensor_state_token = rtas_token("get-sensor-state");
+	ras_check_exception_token = rtas_token("check-exception");
+
+	/* Internal Errors */
+	np = of_find_node_by_path("/event-sources/internal-errors");
+	if (np != NULL) {
+		request_ras_irqs(np, "open-pic-interrupt", ras_error_interrupt,
+				 "RAS_ERROR");
+		request_ras_irqs(np, "interrupts", ras_error_interrupt,
+				 "RAS_ERROR");
+		of_node_put(np);
+	}
+
+	/* EPOW Events */
+	np = of_find_node_by_path("/event-sources/epow-events");
+	if (np != NULL) {
+		request_ras_irqs(np, "open-pic-interrupt", ras_epow_interrupt,
+				 "RAS_EPOW");
+		request_ras_irqs(np, "interrupts", ras_epow_interrupt,
+				 "RAS_EPOW");
+		of_node_put(np);
+	}
+
+	return 1;
+}
+__initcall(init_ras_IRQ);
+
+/*
+ * Handle power subsystem events (EPOW).
+ *
+ * Presently we just log the event has occurred.  This should be fixed
+ * to examine the type of power failure and take appropriate action where
+ * the time horizon permits something useful to be done.
+ */
+static irqreturn_t
+ras_epow_interrupt(int irq, void *dev_id, struct pt_regs * regs)
+{
+	int status = 0xdeadbeef;
+	int state = 0;
+	int critical;
+
+	status = rtas_call(ras_get_sensor_state_token, 2, 2, &state,
+			   EPOW_SENSOR_TOKEN, EPOW_SENSOR_INDEX);
+
+	if (state > 3)
+		critical = 1;  /* Time Critical */
+	else
+		critical = 0;
+
+	spin_lock(&ras_log_buf_lock);
+
+	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
+			   RAS_VECTOR_OFFSET,
+			   virt_irq_to_real(irq_offset_down(irq)),
+			   RTAS_EPOW_WARNING | RTAS_POWERMGM_EVENTS,
+			   critical, __pa(&ras_log_buf),
+				rtas_get_error_log_max());
+
+	udbg_printf("EPOW <0x%lx 0x%x 0x%x>\n",
+		    *((unsigned long *)&ras_log_buf), status, state);
+	printk(KERN_WARNING "EPOW <0x%lx 0x%x 0x%x>\n",
+	       *((unsigned long *)&ras_log_buf), status, state);
+
+	/* format and print the extended information */
+	log_error(ras_log_buf, ERR_TYPE_RTAS_LOG, 0);
+
+	spin_unlock(&ras_log_buf_lock);
+	return IRQ_HANDLED;
+}
+
+/*
+ * Handle hardware error interrupts.
+ *
+ * RTAS check-exception is called to collect data on the exception.  If
+ * the error is deemed recoverable, we log a warning and return.
+ * For nonrecoverable errors, an error is logged and we stop all processing
+ * as quickly as possible in order to prevent propagation of the failure.
+ */
+static irqreturn_t
+ras_error_interrupt(int irq, void *dev_id, struct pt_regs * regs)
+{
+	struct rtas_error_log *rtas_elog;
+	int status = 0xdeadbeef;
+	int fatal;
+
+	spin_lock(&ras_log_buf_lock);
+
+	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
+			   RAS_VECTOR_OFFSET,
+			   virt_irq_to_real(irq_offset_down(irq)),
+			   RTAS_INTERNAL_ERROR, 1 /*Time Critical */,
+			   __pa(&ras_log_buf),
+				rtas_get_error_log_max());
+
+	rtas_elog = (struct rtas_error_log *)ras_log_buf;
+
+	if ((status == 0) && (rtas_elog->severity >= RTAS_SEVERITY_ERROR_SYNC))
+		fatal = 1;
+	else
+		fatal = 0;
+
+	/* format and print the extended information */
+	log_error(ras_log_buf, ERR_TYPE_RTAS_LOG, fatal);
+
+	if (fatal) {
+		udbg_printf("Fatal HW Error <0x%lx 0x%x>\n",
+			    *((unsigned long *)&ras_log_buf), status);
+		printk(KERN_EMERG "Error: Fatal hardware error <0x%lx 0x%x>\n",
+		       *((unsigned long *)&ras_log_buf), status);
+
+#ifndef DEBUG
+		/* Don't actually power off when debugging so we can test
+		 * without actually failing while injecting errors.
+		 * Error data will not be logged to syslog.
+		 */
+		ppc_md.power_off();
+#endif
+	} else {
+		udbg_printf("Recoverable HW Error <0x%lx 0x%x>\n",
+			    *((unsigned long *)&ras_log_buf), status);
+		printk(KERN_WARNING
+		       "Warning: Recoverable hardware error <0x%lx 0x%x>\n",
+		       *((unsigned long *)&ras_log_buf), status);
+	}
+
+	spin_unlock(&ras_log_buf_lock);
+	return IRQ_HANDLED;
+}
+
+/* Get the error information for errors coming through the
+ * FWNMI vectors.  The pt_regs' r3 will be updated to reflect
+ * the actual r3 if possible, and a ptr to the error log entry
+ * will be returned if found.
+ *
+ * The mce_data_buf does not have any locks or protection around it,
+ * if a second machine check comes in, or a system reset is done
+ * before we have logged the error, then we will get corruption in the
+ * error log.  This is preferable over holding off on calling
+ * ibm,nmi-interlock which would result in us checkstopping if a
+ * second machine check did come in.
+ */
+static struct rtas_error_log *fwnmi_get_errinfo(struct pt_regs *regs)
+{
+	unsigned long errdata = regs->gpr[3];
+	struct rtas_error_log *errhdr = NULL;
+	unsigned long *savep;
+
+	if ((errdata >= 0x7000 && errdata < 0x7fff0) ||
+	    (errdata >= rtas.base && errdata < rtas.base + rtas.size - 16)) {
+		savep = __va(errdata);
+		regs->gpr[3] = savep[0];	/* restore original r3 */
+		memset(mce_data_buf, 0, RTAS_ERROR_LOG_MAX);
+		memcpy(mce_data_buf, (char *)(savep + 1), RTAS_ERROR_LOG_MAX);
+		errhdr = (struct rtas_error_log *)mce_data_buf;
+	} else {
+		printk("FWNMI: corrupt r3\n");
+	}
+	return errhdr;
+}
+
+/* Call this when done with the data returned by FWNMI_get_errinfo.
+ * It will release the saved data area for other CPUs in the
+ * partition to receive FWNMI errors.
+ */
+static void fwnmi_release_errinfo(void)
+{
+	int ret = rtas_call(rtas_token("ibm,nmi-interlock"), 0, 1, NULL);
+	if (ret != 0)
+		printk("FWNMI: nmi-interlock failed: %d\n", ret);
+}
+
+void pSeries_system_reset_exception(struct pt_regs *regs)
+{
+	if (fwnmi_active) {
+		struct rtas_error_log *errhdr = fwnmi_get_errinfo(regs);
+		if (errhdr) {
+			/* XXX Should look at FWNMI information */
+		}
+		fwnmi_release_errinfo();
+	}
+}
+
+/*
+ * See if we can recover from a machine check exception.
+ * This is only called on power4 (or above) and only via
+ * the Firmware Non-Maskable Interrupts (fwnmi) handler
+ * which provides the error analysis for us.
+ *
+ * Return 1 if corrected (or delivered a signal).
+ * Return 0 if there is nothing we can do.
+ */
+static int recover_mce(struct pt_regs *regs, struct rtas_error_log * err)
+{
+	int nonfatal = 0;
+
+	if (err->disposition == RTAS_DISP_FULLY_RECOVERED) {
+		/* Platform corrected itself */
+		nonfatal = 1;
+	} else if ((regs->msr & MSR_RI) &&
+		   user_mode(regs) &&
+		   err->severity == RTAS_SEVERITY_ERROR_SYNC &&
+		   err->disposition == RTAS_DISP_NOT_RECOVERED &&
+		   err->target == RTAS_TARGET_MEMORY &&
+		   err->type == RTAS_TYPE_ECC_UNCORR &&
+		   !(current->pid == 0 || current->pid == 1)) {
+		/* Kill off a user process with an ECC error */
+		printk(KERN_ERR "MCE: uncorrectable ecc error for pid %d\n",
+		       current->pid);
+		/* XXX something better for ECC error? */
+		_exception(SIGBUS, regs, BUS_ADRERR, regs->nip);
+		nonfatal = 1;
+	}
+
+	log_error((char *)err, ERR_TYPE_RTAS_LOG, !nonfatal);
+
+	return nonfatal;
+}
+
+/*
+ * Handle a machine check.
+ *
+ * Note that on Power 4 and beyond Firmware Non-Maskable Interrupts (fwnmi)
+ * should be present.  If so the handler which called us tells us if the
+ * error was recovered (never true if RI=0).
+ *
+ * On hardware prior to Power 4 these exceptions were asynchronous which
+ * means we can't tell exactly where it occurred and so we can't recover.
+ */
+int pSeries_machine_check_exception(struct pt_regs *regs)
+{
+	struct rtas_error_log *errp;
+
+	if (fwnmi_active) {
+		errp = fwnmi_get_errinfo(regs);
+		fwnmi_release_errinfo();
+		if (errp && recover_mce(regs, errp))
+			return 1;
+	}
+
+	return 0;
+}
diff --git a/arch/powerpc/platforms/pseries/reconfig.c b/arch/powerpc/platforms/pseries/reconfig.c
new file mode 100644
index 000000000000..58c61219d08e
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/reconfig.c
@@ -0,0 +1,426 @@
+/*
+ * pSeries_reconfig.c - support for dynamic reconfiguration (including PCI
+ * Hotplug and Dynamic Logical Partitioning on RPA platforms).
+ *
+ * Copyright (C) 2005 Nathan Lynch
+ * Copyright (C) 2005 IBM Corporation
+ *
+ *
+ *	This program is free software; you can redistribute it and/or
+ *	modify it under the terms of the GNU General Public License version
+ *	2 as published by the Free Software Foundation.
+ */
+
+#include <linux/kernel.h>
+#include <linux/kref.h>
+#include <linux/notifier.h>
+#include <linux/proc_fs.h>
+
+#include <asm/prom.h>
+#include <asm/pSeries_reconfig.h>
+#include <asm/uaccess.h>
+
+
+
+/*
+ * Routines for "runtime" addition and removal of device tree nodes.
+ */
+#ifdef CONFIG_PROC_DEVICETREE
+/*
+ * Add a node to /proc/device-tree.
+ */
+static void add_node_proc_entries(struct device_node *np)
+{
+	struct proc_dir_entry *ent;
+
+	ent = proc_mkdir(strrchr(np->full_name, '/') + 1, np->parent->pde);
+	if (ent)
+		proc_device_tree_add_node(np, ent);
+}
+
+static void remove_node_proc_entries(struct device_node *np)
+{
+	struct property *pp = np->properties;
+	struct device_node *parent = np->parent;
+
+	while (pp) {
+		remove_proc_entry(pp->name, np->pde);
+		pp = pp->next;
+	}
+	if (np->pde)
+		remove_proc_entry(np->pde->name, parent->pde);
+}
+#else /* !CONFIG_PROC_DEVICETREE */
+static void add_node_proc_entries(struct device_node *np)
+{
+	return;
+}
+
+static void remove_node_proc_entries(struct device_node *np)
+{
+	return;
+}
+#endif /* CONFIG_PROC_DEVICETREE */
+
+/**
+ *	derive_parent - basically like dirname(1)
+ *	@path:  the full_name of a node to be added to the tree
+ *
+ *	Returns the node which should be the parent of the node
+ *	described by path.  E.g., for path = "/foo/bar", returns
+ *	the node with full_name = "/foo".
+ */
+static struct device_node *derive_parent(const char *path)
+{
+	struct device_node *parent = NULL;
+	char *parent_path = "/";
+	size_t parent_path_len = strrchr(path, '/') - path + 1;
+
+	/* reject if path is "/" */
+	if (!strcmp(path, "/"))
+		return ERR_PTR(-EINVAL);
+
+	if (strrchr(path, '/') != path) {
+		parent_path = kmalloc(parent_path_len, GFP_KERNEL);
+		if (!parent_path)
+			return ERR_PTR(-ENOMEM);
+		strlcpy(parent_path, path, parent_path_len);
+	}
+	parent = of_find_node_by_path(parent_path);
+	if (!parent)
+		return ERR_PTR(-EINVAL);
+	if (strcmp(parent_path, "/"))
+		kfree(parent_path);
+	return parent;
+}
+
+static struct notifier_block *pSeries_reconfig_chain;
+
+int pSeries_reconfig_notifier_register(struct notifier_block *nb)
+{
+	return notifier_chain_register(&pSeries_reconfig_chain, nb);
+}
+
+void pSeries_reconfig_notifier_unregister(struct notifier_block *nb)
+{
+	notifier_chain_unregister(&pSeries_reconfig_chain, nb);
+}
+
+static int pSeries_reconfig_add_node(const char *path, struct property *proplist)
+{
+	struct device_node *np;
+	int err = -ENOMEM;
+
+	np = kzalloc(sizeof(*np), GFP_KERNEL);
+	if (!np)
+		goto out_err;
+
+	np->full_name = kmalloc(strlen(path) + 1, GFP_KERNEL);
+	if (!np->full_name)
+		goto out_err;
+
+	strcpy(np->full_name, path);
+
+	np->properties = proplist;
+	OF_MARK_DYNAMIC(np);
+	kref_init(&np->kref);
+
+	np->parent = derive_parent(path);
+	if (IS_ERR(np->parent)) {
+		err = PTR_ERR(np->parent);
+		goto out_err;
+	}
+
+	err = notifier_call_chain(&pSeries_reconfig_chain,
+				  PSERIES_RECONFIG_ADD, np);
+	if (err == NOTIFY_BAD) {
+		printk(KERN_ERR "Failed to add device node %s\n", path);
+		err = -ENOMEM; /* For now, safe to assume kmalloc failure */
+		goto out_err;
+	}
+
+	of_attach_node(np);
+
+	add_node_proc_entries(np);
+
+	of_node_put(np->parent);
+
+	return 0;
+
+out_err:
+	if (np) {
+		of_node_put(np->parent);
+		kfree(np->full_name);
+		kfree(np);
+	}
+	return err;
+}
+
+static int pSeries_reconfig_remove_node(struct device_node *np)
+{
+	struct device_node *parent, *child;
+
+	parent = of_get_parent(np);
+	if (!parent)
+		return -EINVAL;
+
+	if ((child = of_get_next_child(np, NULL))) {
+		of_node_put(child);
+		return -EBUSY;
+	}
+
+	remove_node_proc_entries(np);
+
+	notifier_call_chain(&pSeries_reconfig_chain,
+			    PSERIES_RECONFIG_REMOVE, np);
+	of_detach_node(np);
+
+	of_node_put(parent);
+	of_node_put(np); /* Must decrement the refcount */
+	return 0;
+}
+
+/*
+ * /proc/ppc64/ofdt - yucky binary interface for adding and removing
+ * OF device nodes.  Should be deprecated as soon as we get an
+ * in-kernel wrapper for the RTAS ibm,configure-connector call.
+ */
+
+static void release_prop_list(const struct property *prop)
+{
+	struct property *next;
+	for (; prop; prop = next) {
+		next = prop->next;
+		kfree(prop->name);
+		kfree(prop->value);
+		kfree(prop);
+	}
+
+}
+
+/**
+ * parse_next_property - process the next property from raw input buffer
+ * @buf: input buffer, must be nul-terminated
+ * @end: end of the input buffer + 1, for validation
+ * @name: return value; set to property name in buf
+ * @length: return value; set to length of value
+ * @value: return value; set to the property value in buf
+ *
+ * Note that the caller must make copies of the name and value returned,
+ * this function does no allocation or copying of the data.  Return value
+ * is set to the next name in buf, or NULL on error.
+ */
+static char * parse_next_property(char *buf, char *end, char **name, int *length,
+				  unsigned char **value)
+{
+	char *tmp;
+
+	*name = buf;
+
+	tmp = strchr(buf, ' ');
+	if (!tmp) {
+		printk(KERN_ERR "property parse failed in %s at line %d\n",
+		       __FUNCTION__, __LINE__);
+		return NULL;
+	}
+	*tmp = '\0';
+
+	if (++tmp >= end) {
+		printk(KERN_ERR "property parse failed in %s at line %d\n",
+		       __FUNCTION__, __LINE__);
+		return NULL;
+	}
+
+	/* now we're on the length */
+	*length = -1;
+	*length = simple_strtoul(tmp, &tmp, 10);
+	if (*length == -1) {
+		printk(KERN_ERR "property parse failed in %s at line %d\n",
+		       __FUNCTION__, __LINE__);
+		return NULL;
+	}
+	if (*tmp != ' ' || ++tmp >= end) {
+		printk(KERN_ERR "property parse failed in %s at line %d\n",
+		       __FUNCTION__, __LINE__);
+		return NULL;
+	}
+
+	/* now we're on the value */
+	*value = tmp;
+	tmp += *length;
+	if (tmp > end) {
+		printk(KERN_ERR "property parse failed in %s at line %d\n",
+		       __FUNCTION__, __LINE__);
+		return NULL;
+	}
+	else if (tmp < end && *tmp != ' ' && *tmp != '\0') {
+		printk(KERN_ERR "property parse failed in %s at line %d\n",
+		       __FUNCTION__, __LINE__);
+		return NULL;
+	}
+	tmp++;
+
+	/* and now we should be on the next name, or the end */
+	return tmp;
+}
+
+static struct property *new_property(const char *name, const int length,
+				     const unsigned char *value, struct property *last)
+{
+	struct property *new = kmalloc(sizeof(*new), GFP_KERNEL);
+
+	if (!new)
+		return NULL;
+	memset(new, 0, sizeof(*new));
+
+	if (!(new->name = kmalloc(strlen(name) + 1, GFP_KERNEL)))
+		goto cleanup;
+	if (!(new->value = kmalloc(length + 1, GFP_KERNEL)))
+		goto cleanup;
+
+	strcpy(new->name, name);
+	memcpy(new->value, value, length);
+	*(((char *)new->value) + length) = 0;
+	new->length = length;
+	new->next = last;
+	return new;
+
+cleanup:
+	if (new->name)
+		kfree(new->name);
+	if (new->value)
+		kfree(new->value);
+	kfree(new);
+	return NULL;
+}
+
+static int do_add_node(char *buf, size_t bufsize)
+{
+	char *path, *end, *name;
+	struct device_node *np;
+	struct property *prop = NULL;
+	unsigned char* value;
+	int length, rv = 0;
+
+	end = buf + bufsize;
+	path = buf;
+	buf = strchr(buf, ' ');
+	if (!buf)
+		return -EINVAL;
+	*buf = '\0';
+	buf++;
+
+	if ((np = of_find_node_by_path(path))) {
+		of_node_put(np);
+		return -EINVAL;
+	}
+
+	/* rv = build_prop_list(tmp, bufsize - (tmp - buf), &proplist); */
+	while (buf < end &&
+	       (buf = parse_next_property(buf, end, &name, &length, &value))) {
+		struct property *last = prop;
+
+		prop = new_property(name, length, value, last);
+		if (!prop) {
+			rv = -ENOMEM;
+			prop = last;
+			goto out;
+		}
+	}
+	if (!buf) {
+		rv = -EINVAL;
+		goto out;
+	}
+
+	rv = pSeries_reconfig_add_node(path, prop);
+
+out:
+	if (rv)
+		release_prop_list(prop);
+	return rv;
+}
+
+static int do_remove_node(char *buf)
+{
+	struct device_node *node;
+	int rv = -ENODEV;
+
+	if ((node = of_find_node_by_path(buf)))
+		rv = pSeries_reconfig_remove_node(node);
+
+	of_node_put(node);
+	return rv;
+}
+
+/**
+ * ofdt_write - perform operations on the Open Firmware device tree
+ *
+ * @file: not used
+ * @buf: command and arguments
+ * @count: size of the command buffer
+ * @off: not used
+ *
+ * Operations supported at this time are addition and removal of
+ * whole nodes along with their properties.  Operations on individual
+ * properties are not implemented (yet).
+ */
+static ssize_t ofdt_write(struct file *file, const char __user *buf, size_t count,
+			  loff_t *off)
+{
+	int rv = 0;
+	char *kbuf;
+	char *tmp;
+
+	if (!(kbuf = kmalloc(count + 1, GFP_KERNEL))) {
+		rv = -ENOMEM;
+		goto out;
+	}
+	if (copy_from_user(kbuf, buf, count)) {
+		rv = -EFAULT;
+		goto out;
+	}
+
+	kbuf[count] = '\0';
+
+	tmp = strchr(kbuf, ' ');
+	if (!tmp) {
+		rv = -EINVAL;
+		goto out;
+	}
+	*tmp = '\0';
+	tmp++;
+
+	if (!strcmp(kbuf, "add_node"))
+		rv = do_add_node(tmp, count - (tmp - kbuf));
+	else if (!strcmp(kbuf, "remove_node"))
+		rv = do_remove_node(tmp);
+	else
+		rv = -EINVAL;
+out:
+	kfree(kbuf);
+	return rv ? rv : count;
+}
+
+static struct file_operations ofdt_fops = {
+	.write = ofdt_write
+};
+
+/* create /proc/ppc64/ofdt write-only by root */
+static int proc_ppc64_create_ofdt(void)
+{
+	struct proc_dir_entry *ent;
+
+	if (!(systemcfg->platform & PLATFORM_PSERIES))
+		return 0;
+
+	ent = create_proc_entry("ppc64/ofdt", S_IWUSR, NULL);
+	if (ent) {
+		ent->nlink = 1;
+		ent->data = NULL;
+		ent->size = 0;
+		ent->proc_fops = &ofdt_fops;
+	}
+
+	return 0;
+}
+__initcall(proc_ppc64_create_ofdt);
diff --git a/arch/powerpc/platforms/pseries/rtasd.c b/arch/powerpc/platforms/pseries/rtasd.c
new file mode 100644
index 000000000000..e26b0420b6dd
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/rtasd.c
@@ -0,0 +1,527 @@
+/*
+ * Copyright (C) 2001 Anton Blanchard <anton@au.ibm.com>, IBM
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ *
+ * Communication to userspace based on kernel/printk.c
+ */
+
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/poll.h>
+#include <linux/proc_fs.h>
+#include <linux/init.h>
+#include <linux/vmalloc.h>
+#include <linux/spinlock.h>
+#include <linux/cpu.h>
+#include <linux/delay.h>
+
+#include <asm/uaccess.h>
+#include <asm/io.h>
+#include <asm/rtas.h>
+#include <asm/prom.h>
+#include <asm/nvram.h>
+#include <asm/atomic.h>
+#include <asm/systemcfg.h>
+
+#if 0
+#define DEBUG(A...)	printk(KERN_ERR A)
+#else
+#define DEBUG(A...)
+#endif
+
+static DEFINE_SPINLOCK(rtasd_log_lock);
+
+DECLARE_WAIT_QUEUE_HEAD(rtas_log_wait);
+
+static char *rtas_log_buf;
+static unsigned long rtas_log_start;
+static unsigned long rtas_log_size;
+
+static int surveillance_timeout = -1;
+static unsigned int rtas_event_scan_rate;
+static unsigned int rtas_error_log_max;
+static unsigned int rtas_error_log_buffer_max;
+
+static int full_rtas_msgs = 0;
+
+extern int no_logging;
+
+volatile int error_log_cnt = 0;
+
+/*
+ * Since we use 32 bit RTAS, the physical address of this must be below
+ * 4G or else bad things happen. Allocate this in the kernel data and
+ * make it big enough.
+ */
+static unsigned char logdata[RTAS_ERROR_LOG_MAX];
+
+static int get_eventscan_parms(void);
+
+static char *rtas_type[] = {
+	"Unknown", "Retry", "TCE Error", "Internal Device Failure",
+	"Timeout", "Data Parity", "Address Parity", "Cache Parity",
+	"Address Invalid", "ECC Uncorrected", "ECC Corrupted",
+};
+
+static char *rtas_event_type(int type)
+{
+	if ((type > 0) && (type < 11))
+		return rtas_type[type];
+
+	switch (type) {
+		case RTAS_TYPE_EPOW:
+			return "EPOW";
+		case RTAS_TYPE_PLATFORM:
+			return "Platform Error";
+		case RTAS_TYPE_IO:
+			return "I/O Event";
+		case RTAS_TYPE_INFO:
+			return "Platform Information Event";
+		case RTAS_TYPE_DEALLOC:
+			return "Resource Deallocation Event";
+		case RTAS_TYPE_DUMP:
+			return "Dump Notification Event";
+	}
+
+	return rtas_type[0];
+}
+
+/* To see this info, grep RTAS /var/log/messages and each entry
+ * will be collected together with obvious begin/end.
+ * There will be a unique identifier on the begin and end lines.
+ * This will persist across reboots.
+ *
+ * format of error logs returned from RTAS:
+ * bytes	(size)	: contents
+ * --------------------------------------------------------
+ * 0-7		(8)	: rtas_error_log
+ * 8-47		(40)	: extended info
+ * 48-51	(4)	: vendor id
+ * 52-1023 (vendor specific) : location code and debug data
+ */
+static void printk_log_rtas(char *buf, int len)
+{
+
+	int i,j,n = 0;
+	int perline = 16;
+	char buffer[64];
+	char * str = "RTAS event";
+
+	if (full_rtas_msgs) {
+		printk(RTAS_DEBUG "%d -------- %s begin --------\n",
+		       error_log_cnt, str);
+
+		/*
+		 * Print perline bytes on each line, each line will start
+		 * with RTAS and a changing number, so syslogd will
+		 * print lines that are otherwise the same.  Separate every
+		 * 4 bytes with a space.
+		 */
+		for (i = 0; i < len; i++) {
+			j = i % perline;
+			if (j == 0) {
+				memset(buffer, 0, sizeof(buffer));
+				n = sprintf(buffer, "RTAS %d:", i/perline);
+			}
+
+			if ((i % 4) == 0)
+				n += sprintf(buffer+n, " ");
+
+			n += sprintf(buffer+n, "%02x", (unsigned char)buf[i]);
+
+			if (j == (perline-1))
+				printk(KERN_DEBUG "%s\n", buffer);
+		}
+		if ((i % perline) != 0)
+			printk(KERN_DEBUG "%s\n", buffer);
+
+		printk(RTAS_DEBUG "%d -------- %s end ----------\n",
+		       error_log_cnt, str);
+	} else {
+		struct rtas_error_log *errlog = (struct rtas_error_log *)buf;
+
+		printk(RTAS_DEBUG "event: %d, Type: %s, Severity: %d\n",
+		       error_log_cnt, rtas_event_type(errlog->type),
+		       errlog->severity);
+	}
+}
+
+static int log_rtas_len(char * buf)
+{
+	int len;
+	struct rtas_error_log *err;
+
+	/* rtas fixed header */
+	len = 8;
+	err = (struct rtas_error_log *)buf;
+	if (err->extended_log_length) {
+
+		/* extended header */
+		len += err->extended_log_length;
+	}
+
+	if (rtas_error_log_max == 0) {
+		get_eventscan_parms();
+	}
+	if (len > rtas_error_log_max)
+		len = rtas_error_log_max;
+
+	return len;
+}
+
+/*
+ * First write to nvram, if fatal error, that is the only
+ * place we log the info.  The error will be picked up
+ * on the next reboot by rtasd.  If not fatal, run the
+ * method for the type of error.  Currently, only RTAS
+ * errors have methods implemented, but in the future
+ * there might be a need to store data in nvram before a
+ * call to panic().
+ *
+ * XXX We write to nvram periodically, to indicate error has
+ * been written and sync'd, but there is a possibility
+ * that if we don't shutdown correctly, a duplicate error
+ * record will be created on next reboot.
+ */
+void pSeries_log_error(char *buf, unsigned int err_type, int fatal)
+{
+	unsigned long offset;
+	unsigned long s;
+	int len = 0;
+
+	DEBUG("logging event\n");
+	if (buf == NULL)
+		return;
+
+	spin_lock_irqsave(&rtasd_log_lock, s);
+
+	/* get length and increase count */
+	switch (err_type & ERR_TYPE_MASK) {
+	case ERR_TYPE_RTAS_LOG:
+		len = log_rtas_len(buf);
+		if (!(err_type & ERR_FLAG_BOOT))
+			error_log_cnt++;
+		break;
+	case ERR_TYPE_KERNEL_PANIC:
+	default:
+		spin_unlock_irqrestore(&rtasd_log_lock, s);
+		return;
+	}
+
+	/* Write error to NVRAM */
+	if (!no_logging && !(err_type & ERR_FLAG_BOOT))
+		nvram_write_error_log(buf, len, err_type);
+
+	/*
+	 * rtas errors can occur during boot, and we do want to capture
+	 * those somewhere, even if nvram isn't ready (why not?), and even
+	 * if rtasd isn't ready. Put them into the boot log, at least.
+	 */
+	if ((err_type & ERR_TYPE_MASK) == ERR_TYPE_RTAS_LOG)
+		printk_log_rtas(buf, len);
+
+	/* Check to see if we need to or have stopped logging */
+	if (fatal || no_logging) {
+		no_logging = 1;
+		spin_unlock_irqrestore(&rtasd_log_lock, s);
+		return;
+	}
+
+	/* call type specific method for error */
+	switch (err_type & ERR_TYPE_MASK) {
+	case ERR_TYPE_RTAS_LOG:
+		offset = rtas_error_log_buffer_max *
+			((rtas_log_start+rtas_log_size) & LOG_NUMBER_MASK);
+
+		/* First copy over sequence number */
+		memcpy(&rtas_log_buf[offset], (void *) &error_log_cnt, sizeof(int));
+
+		/* Second copy over error log data */
+		offset += sizeof(int);
+		memcpy(&rtas_log_buf[offset], buf, len);
+
+		if (rtas_log_size < LOG_NUMBER)
+			rtas_log_size += 1;
+		else
+			rtas_log_start += 1;
+
+		spin_unlock_irqrestore(&rtasd_log_lock, s);
+		wake_up_interruptible(&rtas_log_wait);
+		break;
+	case ERR_TYPE_KERNEL_PANIC:
+	default:
+		spin_unlock_irqrestore(&rtasd_log_lock, s);
+		return;
+	}
+
+}
+
+
+static int rtas_log_open(struct inode * inode, struct file * file)
+{
+	return 0;
+}
+
+static int rtas_log_release(struct inode * inode, struct file * file)
+{
+	return 0;
+}
+
+/* This will check if all events are logged, if they are then, we
+ * know that we can safely clear the events in NVRAM.
+ * Next we'll sit and wait for something else to log.
+ */
+static ssize_t rtas_log_read(struct file * file, char __user * buf,
+			 size_t count, loff_t *ppos)
+{
+	int error;
+	char *tmp;
+	unsigned long s;
+	unsigned long offset;
+
+	if (!buf || count < rtas_error_log_buffer_max)
+		return -EINVAL;
+
+	count = rtas_error_log_buffer_max;
+
+	if (!access_ok(VERIFY_WRITE, buf, count))
+		return -EFAULT;
+
+	tmp = kmalloc(count, GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+
+	spin_lock_irqsave(&rtasd_log_lock, s);
+	/* if it's 0, then we know we got the last one (the one in NVRAM) */
+	if (rtas_log_size == 0 && !no_logging)
+		nvram_clear_error_log();
+	spin_unlock_irqrestore(&rtasd_log_lock, s);
+
+
+	error = wait_event_interruptible(rtas_log_wait, rtas_log_size);
+	if (error)
+		goto out;
+
+	spin_lock_irqsave(&rtasd_log_lock, s);
+	offset = rtas_error_log_buffer_max * (rtas_log_start & LOG_NUMBER_MASK);
+	memcpy(tmp, &rtas_log_buf[offset], count);
+
+	rtas_log_start += 1;
+	rtas_log_size -= 1;
+	spin_unlock_irqrestore(&rtasd_log_lock, s);
+
+	error = copy_to_user(buf, tmp, count) ? -EFAULT : count;
+out:
+	kfree(tmp);
+	return error;
+}
+
+static unsigned int rtas_log_poll(struct file *file, poll_table * wait)
+{
+	poll_wait(file, &rtas_log_wait, wait);
+	if (rtas_log_size)
+		return POLLIN | POLLRDNORM;
+	return 0;
+}
+
+struct file_operations proc_rtas_log_operations = {
+	.read =		rtas_log_read,
+	.poll =		rtas_log_poll,
+	.open =		rtas_log_open,
+	.release =	rtas_log_release,
+};
+
+static int enable_surveillance(int timeout)
+{
+	int error;
+
+	error = rtas_set_indicator(SURVEILLANCE_TOKEN, 0, timeout);
+
+	if (error == 0)
+		return 0;
+
+	if (error == -EINVAL) {
+		printk(KERN_INFO "rtasd: surveillance not supported\n");
+		return 0;
+	}
+
+	printk(KERN_ERR "rtasd: could not update surveillance\n");
+	return -1;
+}
+
+static int get_eventscan_parms(void)
+{
+	struct device_node *node;
+	int *ip;
+
+	node = of_find_node_by_path("/rtas");
+
+	ip = (int *)get_property(node, "rtas-event-scan-rate", NULL);
+	if (ip == NULL) {
+		printk(KERN_ERR "rtasd: no rtas-event-scan-rate\n");
+		of_node_put(node);
+		return -1;
+	}
+	rtas_event_scan_rate = *ip;
+	DEBUG("rtas-event-scan-rate %d\n", rtas_event_scan_rate);
+
+	/* Make room for the sequence number */
+	rtas_error_log_max = rtas_get_error_log_max();
+	rtas_error_log_buffer_max = rtas_error_log_max + sizeof(int);
+
+	of_node_put(node);
+
+	return 0;
+}
+
+static void do_event_scan(int event_scan)
+{
+	int error;
+	do {
+		memset(logdata, 0, rtas_error_log_max);
+		error = rtas_call(event_scan, 4, 1, NULL,
+				  RTAS_EVENT_SCAN_ALL_EVENTS, 0,
+				  __pa(logdata), rtas_error_log_max);
+		if (error == -1) {
+			printk(KERN_ERR "event-scan failed\n");
+			break;
+		}
+
+		if (error == 0)
+			pSeries_log_error(logdata, ERR_TYPE_RTAS_LOG, 0);
+
+	} while(error == 0);
+}
+
+static void do_event_scan_all_cpus(long delay)
+{
+	int cpu;
+
+	lock_cpu_hotplug();
+	cpu = first_cpu(cpu_online_map);
+	for (;;) {
+		set_cpus_allowed(current, cpumask_of_cpu(cpu));
+		do_event_scan(rtas_token("event-scan"));
+		set_cpus_allowed(current, CPU_MASK_ALL);
+
+		/* Drop hotplug lock, and sleep for the specified delay */
+		unlock_cpu_hotplug();
+		msleep_interruptible(delay);
+		lock_cpu_hotplug();
+
+		cpu = next_cpu(cpu, cpu_online_map);
+		if (cpu == NR_CPUS)
+			break;
+	}
+	unlock_cpu_hotplug();
+}
+
+static int rtasd(void *unused)
+{
+	unsigned int err_type;
+	int event_scan = rtas_token("event-scan");
+	int rc;
+
+	daemonize("rtasd");
+
+	if (event_scan == RTAS_UNKNOWN_SERVICE || get_eventscan_parms() == -1)
+		goto error;
+
+	rtas_log_buf = vmalloc(rtas_error_log_buffer_max*LOG_NUMBER);
+	if (!rtas_log_buf) {
+		printk(KERN_ERR "rtasd: no memory\n");
+		goto error;
+	}
+
+	printk(KERN_INFO "RTAS daemon started\n");
+
+	DEBUG("will sleep for %d milliseconds\n", (30000/rtas_event_scan_rate));
+
+	/* See if we have any error stored in NVRAM */
+	memset(logdata, 0, rtas_error_log_max);
+
+	rc = nvram_read_error_log(logdata, rtas_error_log_max, &err_type);
+
+	/* We can use rtas_log_buf now */
+	no_logging = 0;
+
+	if (!rc) {
+		if (err_type != ERR_FLAG_ALREADY_LOGGED) {
+			pSeries_log_error(logdata, err_type | ERR_FLAG_BOOT, 0);
+		}
+	}
+
+	/* First pass. */
+	do_event_scan_all_cpus(1000);
+
+	if (surveillance_timeout != -1) {
+		DEBUG("enabling surveillance\n");
+		enable_surveillance(surveillance_timeout);
+		DEBUG("surveillance enabled\n");
+	}
+
+	/* Delay should be at least one second since some
+	 * machines have problems if we call event-scan too
+	 * quickly. */
+	for (;;)
+		do_event_scan_all_cpus(30000/rtas_event_scan_rate);
+
+error:
+	/* Should delete proc entries */
+	return -EINVAL;
+}
+
+static int __init rtas_init(void)
+{
+	struct proc_dir_entry *entry;
+
+	/* No RTAS, only warn if we are on a pSeries box  */
+	if (rtas_token("event-scan") == RTAS_UNKNOWN_SERVICE) {
+		if (systemcfg->platform & PLATFORM_PSERIES)
+			printk(KERN_INFO "rtasd: no event-scan on system\n");
+		return 1;
+	}
+
+	entry = create_proc_entry("ppc64/rtas/error_log", S_IRUSR, NULL);
+	if (entry)
+		entry->proc_fops = &proc_rtas_log_operations;
+	else
+		printk(KERN_ERR "Failed to create error_log proc entry\n");
+
+	if (kernel_thread(rtasd, NULL, CLONE_FS) < 0)
+		printk(KERN_ERR "Failed to start RTAS daemon\n");
+
+	return 0;
+}
+
+static int __init surveillance_setup(char *str)
+{
+	int i;
+
+	if (get_option(&str,&i)) {
+		if (i >= 0 && i <= 255)
+			surveillance_timeout = i;
+	}
+
+	return 1;
+}
+
+static int __init rtasmsgs_setup(char *str)
+{
+	if (strcmp(str, "on") == 0)
+		full_rtas_msgs = 1;
+	else if (strcmp(str, "off") == 0)
+		full_rtas_msgs = 0;
+
+	return 1;
+}
+__initcall(rtas_init);
+__setup("surveillance=", surveillance_setup);
+__setup("rtasmsgs=", rtasmsgs_setup);
diff --git a/arch/powerpc/platforms/pseries/setup.c b/arch/powerpc/platforms/pseries/setup.c
new file mode 100644
index 000000000000..65bee939eecc
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/setup.c
@@ -0,0 +1,618 @@
+/*
+ *  64-bit pSeries and RS/6000 setup code.
+ *
+ *  Copyright (C) 1995  Linus Torvalds
+ *  Adapted from 'alpha' version by Gary Thomas
+ *  Modified by Cort Dougan (cort@cs.nmt.edu)
+ *  Modified by PPC64 Team, IBM Corp
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+/*
+ * bootup setup stuff..
+ */
+
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/cpu.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/slab.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/major.h>
+#include <linux/interrupt.h>
+#include <linux/reboot.h>
+#include <linux/init.h>
+#include <linux/ioport.h>
+#include <linux/console.h>
+#include <linux/pci.h>
+#include <linux/utsname.h>
+#include <linux/adb.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+#include <linux/irq.h>
+#include <linux/seq_file.h>
+#include <linux/root_dev.h>
+
+#include <asm/mmu.h>
+#include <asm/processor.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/prom.h>
+#include <asm/rtas.h>
+#include <asm/pci-bridge.h>
+#include <asm/iommu.h>
+#include <asm/dma.h>
+#include <asm/machdep.h>
+#include <asm/irq.h>
+#include <asm/time.h>
+#include <asm/nvram.h>
+#include "xics.h"
+#include <asm/firmware.h>
+#include <asm/pmc.h>
+#include <asm/mpic.h>
+#include <asm/ppc-pci.h>
+#include <asm/i8259.h>
+#include <asm/udbg.h>
+
+#include "plpar_wrappers.h"
+
+#ifdef DEBUG
+#define DBG(fmt...) udbg_printf(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+extern void find_udbg_vterm(void);
+extern void system_reset_fwnmi(void);	/* from head.S */
+extern void machine_check_fwnmi(void);	/* from head.S */
+extern void generic_find_legacy_serial_ports(u64 *physport,
+		unsigned int *default_speed);
+
+int fwnmi_active;  /* TRUE if an FWNMI handler is present */
+
+extern void pSeries_system_reset_exception(struct pt_regs *regs);
+extern int pSeries_machine_check_exception(struct pt_regs *regs);
+
+static void pseries_shared_idle(void);
+static void pseries_dedicated_idle(void);
+
+struct mpic *pSeries_mpic;
+
+void pSeries_show_cpuinfo(struct seq_file *m)
+{
+	struct device_node *root;
+	const char *model = "";
+
+	root = of_find_node_by_path("/");
+	if (root)
+		model = get_property(root, "model", NULL);
+	seq_printf(m, "machine\t\t: CHRP %s\n", model);
+	of_node_put(root);
+}
+
+/* Initialize firmware assisted non-maskable interrupts if
+ * the firmware supports this feature.
+ *
+ */
+static void __init fwnmi_init(void)
+{
+	int ret;
+	int ibm_nmi_register = rtas_token("ibm,nmi-register");
+	if (ibm_nmi_register == RTAS_UNKNOWN_SERVICE)
+		return;
+	ret = rtas_call(ibm_nmi_register, 2, 1, NULL,
+			__pa((unsigned long)system_reset_fwnmi),
+			__pa((unsigned long)machine_check_fwnmi));
+	if (ret == 0)
+		fwnmi_active = 1;
+}
+
+static void __init pSeries_init_mpic(void)
+{
+        unsigned int *addrp;
+	struct device_node *np;
+	unsigned long intack = 0;
+
+	/* All ISUs are setup, complete initialization */
+	mpic_init(pSeries_mpic);
+
+	/* Check what kind of cascade ACK we have */
+        if (!(np = of_find_node_by_name(NULL, "pci"))
+            || !(addrp = (unsigned int *)
+                 get_property(np, "8259-interrupt-acknowledge", NULL)))
+                printk(KERN_ERR "Cannot find pci to get ack address\n");
+        else
+		intack = addrp[prom_n_addr_cells(np)-1];
+	of_node_put(np);
+
+	/* Setup the legacy interrupts & controller */
+	i8259_init(intack, 0);
+
+	/* Hook cascade to mpic */
+	mpic_setup_cascade(NUM_ISA_INTERRUPTS, i8259_irq_cascade, NULL);
+}
+
+static void __init pSeries_setup_mpic(void)
+{
+	unsigned int *opprop;
+	unsigned long openpic_addr = 0;
+        unsigned char senses[NR_IRQS - NUM_ISA_INTERRUPTS];
+        struct device_node *root;
+	int irq_count;
+
+	/* Find the Open PIC if present */
+	root = of_find_node_by_path("/");
+	opprop = (unsigned int *) get_property(root, "platform-open-pic", NULL);
+	if (opprop != 0) {
+		int n = prom_n_addr_cells(root);
+
+		for (openpic_addr = 0; n > 0; --n)
+			openpic_addr = (openpic_addr << 32) + *opprop++;
+		printk(KERN_DEBUG "OpenPIC addr: %lx\n", openpic_addr);
+	}
+	of_node_put(root);
+
+	BUG_ON(openpic_addr == 0);
+
+	/* Get the sense values from OF */
+	prom_get_irq_senses(senses, NUM_ISA_INTERRUPTS, NR_IRQS);
+	
+	/* Setup the openpic driver */
+	irq_count = NR_IRQS - NUM_ISA_INTERRUPTS - 4; /* leave room for IPIs */
+	pSeries_mpic = mpic_alloc(openpic_addr, MPIC_PRIMARY,
+				  16, 16, irq_count, /* isu size, irq offset, irq count */ 
+				  NR_IRQS - 4, /* ipi offset */
+				  senses, irq_count, /* sense & sense size */
+				  " MPIC     ");
+}
+
+static void pseries_lpar_enable_pmcs(void)
+{
+	unsigned long set, reset;
+
+	power4_enable_pmcs();
+
+	set = 1UL << 63;
+	reset = 0;
+	plpar_hcall_norets(H_PERFMON, set, reset);
+
+	/* instruct hypervisor to maintain PMCs */
+	if (firmware_has_feature(FW_FEATURE_SPLPAR))
+		get_paca()->lppaca.pmcregs_in_use = 1;
+}
+
+static void __init pSeries_setup_arch(void)
+{
+	/* Fixup ppc_md depending on the type of interrupt controller */
+	if (ppc64_interrupt_controller == IC_OPEN_PIC) {
+		ppc_md.init_IRQ       = pSeries_init_mpic;
+		ppc_md.get_irq        = mpic_get_irq;
+	 	ppc_md.cpu_irq_down   = mpic_teardown_this_cpu;
+		/* Allocate the mpic now, so that find_and_init_phbs() can
+		 * fill the ISUs */
+		pSeries_setup_mpic();
+	} else {
+		ppc_md.init_IRQ       = xics_init_IRQ;
+		ppc_md.get_irq        = xics_get_irq;
+		ppc_md.cpu_irq_down   = xics_teardown_cpu;
+	}
+
+#ifdef CONFIG_SMP
+	smp_init_pSeries();
+#endif
+	/* openpic global configuration register (64-bit format). */
+	/* openpic Interrupt Source Unit pointer (64-bit format). */
+	/* python0 facility area (mmio) (64-bit format) REAL address. */
+
+	/* init to some ~sane value until calibrate_delay() runs */
+	loops_per_jiffy = 50000000;
+
+	if (ROOT_DEV == 0) {
+		printk("No ramdisk, default root is /dev/sda2\n");
+		ROOT_DEV = Root_SDA2;
+	}
+
+	fwnmi_init();
+
+	/* Find and initialize PCI host bridges */
+	init_pci_config_tokens();
+	find_and_init_phbs();
+	eeh_init();
+
+	pSeries_nvram_init();
+
+	/* Choose an idle loop */
+	if (firmware_has_feature(FW_FEATURE_SPLPAR)) {
+		vpa_init(boot_cpuid);
+		if (get_paca()->lppaca.shared_proc) {
+			printk(KERN_INFO "Using shared processor idle loop\n");
+			ppc_md.idle_loop = pseries_shared_idle;
+		} else {
+			printk(KERN_INFO "Using dedicated idle loop\n");
+			ppc_md.idle_loop = pseries_dedicated_idle;
+		}
+	} else {
+		printk(KERN_INFO "Using default idle loop\n");
+		ppc_md.idle_loop = default_idle;
+	}
+
+	if (systemcfg->platform & PLATFORM_LPAR)
+		ppc_md.enable_pmcs = pseries_lpar_enable_pmcs;
+	else
+		ppc_md.enable_pmcs = power4_enable_pmcs;
+}
+
+static int __init pSeries_init_panel(void)
+{
+	/* Manually leave the kernel version on the panel. */
+	ppc_md.progress("Linux ppc64\n", 0);
+	ppc_md.progress(system_utsname.version, 0);
+
+	return 0;
+}
+arch_initcall(pSeries_init_panel);
+
+
+/* Build up the ppc64_firmware_features bitmask field
+ * using contents of device-tree/ibm,hypertas-functions.
+ * Ultimately this functionality may be moved into prom.c prom_init().
+ */
+static void __init fw_feature_init(void)
+{
+	struct device_node * dn;
+	char * hypertas;
+	unsigned int len;
+
+	DBG(" -> fw_feature_init()\n");
+
+	ppc64_firmware_features = 0;
+	dn = of_find_node_by_path("/rtas");
+	if (dn == NULL) {
+		printk(KERN_ERR "WARNING ! Cannot find RTAS in device-tree !\n");
+		goto no_rtas;
+	}
+
+	hypertas = get_property(dn, "ibm,hypertas-functions", &len);
+	if (hypertas) {
+		while (len > 0){
+			int i, hypertas_len;
+			/* check value against table of strings */
+			for(i=0; i < FIRMWARE_MAX_FEATURES ;i++) {
+				if ((firmware_features_table[i].name) &&
+				    (strcmp(firmware_features_table[i].name,hypertas))==0) {
+					/* we have a match */
+					ppc64_firmware_features |= 
+						(firmware_features_table[i].val);
+					break;
+				} 
+			}
+			hypertas_len = strlen(hypertas);
+			len -= hypertas_len +1;
+			hypertas+= hypertas_len +1;
+		}
+	}
+
+	of_node_put(dn);
+ no_rtas:
+	printk(KERN_INFO "firmware_features = 0x%lx\n", 
+	       ppc64_firmware_features);
+
+	DBG(" <- fw_feature_init()\n");
+}
+
+
+static  void __init pSeries_discover_pic(void)
+{
+	struct device_node *np;
+	char *typep;
+
+	/*
+	 * Setup interrupt mapping options that are needed for finish_device_tree
+	 * to properly parse the OF interrupt tree & do the virtual irq mapping
+	 */
+	__irq_offset_value = NUM_ISA_INTERRUPTS;
+	ppc64_interrupt_controller = IC_INVALID;
+	for (np = NULL; (np = of_find_node_by_name(np, "interrupt-controller"));) {
+		typep = (char *)get_property(np, "compatible", NULL);
+		if (strstr(typep, "open-pic"))
+			ppc64_interrupt_controller = IC_OPEN_PIC;
+		else if (strstr(typep, "ppc-xicp"))
+			ppc64_interrupt_controller = IC_PPC_XIC;
+		else
+			printk("pSeries_discover_pic: failed to recognize"
+			       " interrupt-controller\n");
+		break;
+	}
+}
+
+static void pSeries_mach_cpu_die(void)
+{
+	local_irq_disable();
+	idle_task_exit();
+	/* Some hardware requires clearing the CPPR, while other hardware does not
+	 * it is safe either way
+	 */
+	pSeriesLP_cppr_info(0, 0);
+	rtas_stop_self();
+	/* Should never get here... */
+	BUG();
+	for(;;);
+}
+
+static int pseries_set_dabr(unsigned long dabr)
+{
+	if (firmware_has_feature(FW_FEATURE_XDABR)) {
+		/* We want to catch accesses from kernel and userspace */
+		return plpar_set_xdabr(dabr, H_DABRX_KERNEL | H_DABRX_USER);
+	}
+
+	return plpar_set_dabr(dabr);
+}
+
+
+/*
+ * Early initialization.  Relocation is on but do not reference unbolted pages
+ */
+static void __init pSeries_init_early(void)
+{
+	void *comport;
+	int iommu_off = 0;
+	unsigned int default_speed;
+	u64 physport;
+
+	DBG(" -> pSeries_init_early()\n");
+
+	fw_feature_init();
+	
+	if (systemcfg->platform & PLATFORM_LPAR)
+		hpte_init_lpar();
+	else {
+		hpte_init_native();
+		iommu_off = (of_chosen &&
+			     get_property(of_chosen, "linux,iommu-off", NULL));
+	}
+
+	generic_find_legacy_serial_ports(&physport, &default_speed);
+
+	if (systemcfg->platform & PLATFORM_LPAR)
+		find_udbg_vterm();
+	else if (physport) {
+		/* Map the uart for udbg. */
+		comport = (void *)ioremap(physport, 16);
+		udbg_init_uart(comport, default_speed);
+
+		DBG("Hello World !\n");
+	}
+
+	if (firmware_has_feature(FW_FEATURE_XDABR | FW_FEATURE_DABR))
+		ppc_md.set_dabr = pseries_set_dabr;
+
+	iommu_init_early_pSeries();
+
+	pSeries_discover_pic();
+
+	DBG(" <- pSeries_init_early()\n");
+}
+
+
+static int pSeries_check_legacy_ioport(unsigned int baseport)
+{
+	struct device_node *np;
+
+#define I8042_DATA_REG	0x60
+#define FDC_BASE	0x3f0
+
+
+	switch(baseport) {
+	case I8042_DATA_REG:
+		np = of_find_node_by_type(NULL, "8042");
+		if (np == NULL)
+			return -ENODEV;
+		of_node_put(np);
+		break;
+	case FDC_BASE:
+		np = of_find_node_by_type(NULL, "fdc");
+		if (np == NULL)
+			return -ENODEV;
+		of_node_put(np);
+		break;
+	}
+	return 0;
+}
+
+/*
+ * Called very early, MMU is off, device-tree isn't unflattened
+ */
+extern struct machdep_calls pSeries_md;
+
+static int __init pSeries_probe(int platform)
+{
+	if (platform != PLATFORM_PSERIES &&
+	    platform != PLATFORM_PSERIES_LPAR)
+		return 0;
+
+	/* if we have some ppc_md fixups for LPAR to do, do
+	 * it here ...
+	 */
+
+	return 1;
+}
+
+DECLARE_PER_CPU(unsigned long, smt_snooze_delay);
+
+static inline void dedicated_idle_sleep(unsigned int cpu)
+{
+	struct paca_struct *ppaca = &paca[cpu ^ 1];
+
+	/* Only sleep if the other thread is not idle */
+	if (!(ppaca->lppaca.idle)) {
+		local_irq_disable();
+
+		/*
+		 * We are about to sleep the thread and so wont be polling any
+		 * more.
+		 */
+		clear_thread_flag(TIF_POLLING_NRFLAG);
+
+		/*
+		 * SMT dynamic mode. Cede will result in this thread going
+		 * dormant, if the partner thread is still doing work.  Thread
+		 * wakes up if partner goes idle, an interrupt is presented, or
+		 * a prod occurs.  Returning from the cede enables external
+		 * interrupts.
+		 */
+		if (!need_resched())
+			cede_processor();
+		else
+			local_irq_enable();
+	} else {
+		/*
+		 * Give the HV an opportunity at the processor, since we are
+		 * not doing any work.
+		 */
+		poll_pending();
+	}
+}
+
+static void pseries_dedicated_idle(void)
+{ 
+	long oldval;
+	struct paca_struct *lpaca = get_paca();
+	unsigned int cpu = smp_processor_id();
+	unsigned long start_snooze;
+	unsigned long *smt_snooze_delay = &__get_cpu_var(smt_snooze_delay);
+
+	while (1) {
+		/*
+		 * Indicate to the HV that we are idle. Now would be
+		 * a good time to find other work to dispatch.
+		 */
+		lpaca->lppaca.idle = 1;
+
+		oldval = test_and_clear_thread_flag(TIF_NEED_RESCHED);
+		if (!oldval) {
+			set_thread_flag(TIF_POLLING_NRFLAG);
+
+			start_snooze = __get_tb() +
+				*smt_snooze_delay * tb_ticks_per_usec;
+
+			while (!need_resched() && !cpu_is_offline(cpu)) {
+				ppc64_runlatch_off();
+
+				/*
+				 * Go into low thread priority and possibly
+				 * low power mode.
+				 */
+				HMT_low();
+				HMT_very_low();
+
+				if (*smt_snooze_delay != 0 &&
+				    __get_tb() > start_snooze) {
+					HMT_medium();
+					dedicated_idle_sleep(cpu);
+				}
+
+			}
+
+			HMT_medium();
+			clear_thread_flag(TIF_POLLING_NRFLAG);
+		} else {
+			set_need_resched();
+		}
+
+		lpaca->lppaca.idle = 0;
+		ppc64_runlatch_on();
+
+		schedule();
+
+		if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING)
+			cpu_die();
+	}
+}
+
+static void pseries_shared_idle(void)
+{
+	struct paca_struct *lpaca = get_paca();
+	unsigned int cpu = smp_processor_id();
+
+	while (1) {
+		/*
+		 * Indicate to the HV that we are idle. Now would be
+		 * a good time to find other work to dispatch.
+		 */
+		lpaca->lppaca.idle = 1;
+
+		while (!need_resched() && !cpu_is_offline(cpu)) {
+			local_irq_disable();
+			ppc64_runlatch_off();
+
+			/*
+			 * Yield the processor to the hypervisor.  We return if
+			 * an external interrupt occurs (which are driven prior
+			 * to returning here) or if a prod occurs from another
+			 * processor. When returning here, external interrupts
+			 * are enabled.
+			 *
+			 * Check need_resched() again with interrupts disabled
+			 * to avoid a race.
+			 */
+			if (!need_resched())
+				cede_processor();
+			else
+				local_irq_enable();
+
+			HMT_medium();
+		}
+
+		lpaca->lppaca.idle = 0;
+		ppc64_runlatch_on();
+
+		schedule();
+
+		if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING)
+			cpu_die();
+	}
+}
+
+static int pSeries_pci_probe_mode(struct pci_bus *bus)
+{
+	if (systemcfg->platform & PLATFORM_LPAR)
+		return PCI_PROBE_DEVTREE;
+	return PCI_PROBE_NORMAL;
+}
+
+struct machdep_calls __initdata pSeries_md = {
+	.probe			= pSeries_probe,
+	.setup_arch		= pSeries_setup_arch,
+	.init_early		= pSeries_init_early,
+	.show_cpuinfo		= pSeries_show_cpuinfo,
+	.log_error		= pSeries_log_error,
+	.pcibios_fixup		= pSeries_final_fixup,
+	.pci_probe_mode		= pSeries_pci_probe_mode,
+	.irq_bus_setup		= pSeries_irq_bus_setup,
+	.restart		= rtas_restart,
+	.power_off		= rtas_power_off,
+	.halt			= rtas_halt,
+	.panic			= rtas_os_term,
+	.cpu_die		= pSeries_mach_cpu_die,
+	.get_boot_time		= rtas_get_boot_time,
+	.get_rtc_time		= rtas_get_rtc_time,
+	.set_rtc_time		= rtas_set_rtc_time,
+	.calibrate_decr		= generic_calibrate_decr,
+	.progress		= rtas_progress,
+	.check_legacy_ioport	= pSeries_check_legacy_ioport,
+	.system_reset_exception = pSeries_system_reset_exception,
+	.machine_check_exception = pSeries_machine_check_exception,
+};
diff --git a/arch/powerpc/platforms/pseries/smp.c b/arch/powerpc/platforms/pseries/smp.c
new file mode 100644
index 000000000000..7a243e8ccd7e
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/smp.c
@@ -0,0 +1,472 @@
+/*
+ * SMP support for pSeries machines.
+ *
+ * Dave Engebretsen, Peter Bergner, and
+ * Mike Corrigan {engebret|bergner|mikec}@us.ibm.com
+ *
+ * Plus various changes from other IBM teams...
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+
+#undef DEBUG
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/cache.h>
+#include <linux/err.h>
+#include <linux/sysdev.h>
+#include <linux/cpu.h>
+
+#include <asm/ptrace.h>
+#include <asm/atomic.h>
+#include <asm/irq.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/smp.h>
+#include <asm/paca.h>
+#include <asm/time.h>
+#include <asm/machdep.h>
+#include "xics.h"
+#include <asm/cputable.h>
+#include <asm/firmware.h>
+#include <asm/system.h>
+#include <asm/rtas.h>
+#include <asm/pSeries_reconfig.h>
+#include <asm/mpic.h>
+
+#include "plpar_wrappers.h"
+
+#ifdef DEBUG
+#define DBG(fmt...) udbg_printf(fmt)
+#else
+#define DBG(fmt...)
+#endif
+
+/*
+ * The primary thread of each non-boot processor is recorded here before
+ * smp init.
+ */
+static cpumask_t of_spin_map;
+
+extern void pSeries_secondary_smp_init(unsigned long);
+
+#ifdef CONFIG_HOTPLUG_CPU
+
+/* Get state of physical CPU.
+ * Return codes:
+ *	0	- The processor is in the RTAS stopped state
+ *	1	- stop-self is in progress
+ *	2	- The processor is not in the RTAS stopped state
+ *	-1	- Hardware Error
+ *	-2	- Hardware Busy, Try again later.
+ */
+static int query_cpu_stopped(unsigned int pcpu)
+{
+	int cpu_status;
+	int status, qcss_tok;
+
+	qcss_tok = rtas_token("query-cpu-stopped-state");
+	if (qcss_tok == RTAS_UNKNOWN_SERVICE)
+		return -1;
+	status = rtas_call(qcss_tok, 1, 2, &cpu_status, pcpu);
+	if (status != 0) {
+		printk(KERN_ERR
+		       "RTAS query-cpu-stopped-state failed: %i\n", status);
+		return status;
+	}
+
+	return cpu_status;
+}
+
+int pSeries_cpu_disable(void)
+{
+	int cpu = smp_processor_id();
+
+	cpu_clear(cpu, cpu_online_map);
+	systemcfg->processorCount--;
+
+	/*fix boot_cpuid here*/
+	if (cpu == boot_cpuid)
+		boot_cpuid = any_online_cpu(cpu_online_map);
+
+	/* FIXME: abstract this to not be platform specific later on */
+	xics_migrate_irqs_away();
+	return 0;
+}
+
+void pSeries_cpu_die(unsigned int cpu)
+{
+	int tries;
+	int cpu_status;
+	unsigned int pcpu = get_hard_smp_processor_id(cpu);
+
+	for (tries = 0; tries < 25; tries++) {
+		cpu_status = query_cpu_stopped(pcpu);
+		if (cpu_status == 0 || cpu_status == -1)
+			break;
+		msleep(200);
+	}
+	if (cpu_status != 0) {
+		printk("Querying DEAD? cpu %i (%i) shows %i\n",
+		       cpu, pcpu, cpu_status);
+	}
+
+	/* Isolation and deallocation are definatly done by
+	 * drslot_chrp_cpu.  If they were not they would be
+	 * done here.  Change isolate state to Isolate and
+	 * change allocation-state to Unusable.
+	 */
+	paca[cpu].cpu_start = 0;
+}
+
+/*
+ * Update cpu_present_map and paca(s) for a new cpu node.  The wrinkle
+ * here is that a cpu device node may represent up to two logical cpus
+ * in the SMT case.  We must honor the assumption in other code that
+ * the logical ids for sibling SMT threads x and y are adjacent, such
+ * that x^1 == y and y^1 == x.
+ */
+static int pSeries_add_processor(struct device_node *np)
+{
+	unsigned int cpu;
+	cpumask_t candidate_map, tmp = CPU_MASK_NONE;
+	int err = -ENOSPC, len, nthreads, i;
+	u32 *intserv;
+
+	intserv = (u32 *)get_property(np, "ibm,ppc-interrupt-server#s", &len);
+	if (!intserv)
+		return 0;
+
+	nthreads = len / sizeof(u32);
+	for (i = 0; i < nthreads; i++)
+		cpu_set(i, tmp);
+
+	lock_cpu_hotplug();
+
+	BUG_ON(!cpus_subset(cpu_present_map, cpu_possible_map));
+
+	/* Get a bitmap of unoccupied slots. */
+	cpus_xor(candidate_map, cpu_possible_map, cpu_present_map);
+	if (cpus_empty(candidate_map)) {
+		/* If we get here, it most likely means that NR_CPUS is
+		 * less than the partition's max processors setting.
+		 */
+		printk(KERN_ERR "Cannot add cpu %s; this system configuration"
+		       " supports %d logical cpus.\n", np->full_name,
+		       cpus_weight(cpu_possible_map));
+		goto out_unlock;
+	}
+
+	while (!cpus_empty(tmp))
+		if (cpus_subset(tmp, candidate_map))
+			/* Found a range where we can insert the new cpu(s) */
+			break;
+		else
+			cpus_shift_left(tmp, tmp, nthreads);
+
+	if (cpus_empty(tmp)) {
+		printk(KERN_ERR "Unable to find space in cpu_present_map for"
+		       " processor %s with %d thread(s)\n", np->name,
+		       nthreads);
+		goto out_unlock;
+	}
+
+	for_each_cpu_mask(cpu, tmp) {
+		BUG_ON(cpu_isset(cpu, cpu_present_map));
+		cpu_set(cpu, cpu_present_map);
+		set_hard_smp_processor_id(cpu, *intserv++);
+	}
+	err = 0;
+out_unlock:
+	unlock_cpu_hotplug();
+	return err;
+}
+
+/*
+ * Update the present map for a cpu node which is going away, and set
+ * the hard id in the paca(s) to -1 to be consistent with boot time
+ * convention for non-present cpus.
+ */
+static void pSeries_remove_processor(struct device_node *np)
+{
+	unsigned int cpu;
+	int len, nthreads, i;
+	u32 *intserv;
+
+	intserv = (u32 *)get_property(np, "ibm,ppc-interrupt-server#s", &len);
+	if (!intserv)
+		return;
+
+	nthreads = len / sizeof(u32);
+
+	lock_cpu_hotplug();
+	for (i = 0; i < nthreads; i++) {
+		for_each_present_cpu(cpu) {
+			if (get_hard_smp_processor_id(cpu) != intserv[i])
+				continue;
+			BUG_ON(cpu_online(cpu));
+			cpu_clear(cpu, cpu_present_map);
+			set_hard_smp_processor_id(cpu, -1);
+			break;
+		}
+		if (cpu == NR_CPUS)
+			printk(KERN_WARNING "Could not find cpu to remove "
+			       "with physical id 0x%x\n", intserv[i]);
+	}
+	unlock_cpu_hotplug();
+}
+
+static int pSeries_smp_notifier(struct notifier_block *nb, unsigned long action, void *node)
+{
+	int err = NOTIFY_OK;
+
+	switch (action) {
+	case PSERIES_RECONFIG_ADD:
+		if (pSeries_add_processor(node))
+			err = NOTIFY_BAD;
+		break;
+	case PSERIES_RECONFIG_REMOVE:
+		pSeries_remove_processor(node);
+		break;
+	default:
+		err = NOTIFY_DONE;
+		break;
+	}
+	return err;
+}
+
+static struct notifier_block pSeries_smp_nb = {
+	.notifier_call = pSeries_smp_notifier,
+};
+
+#endif /* CONFIG_HOTPLUG_CPU */
+
+/**
+ * smp_startup_cpu() - start the given cpu
+ *
+ * At boot time, there is nothing to do for primary threads which were
+ * started from Open Firmware.  For anything else, call RTAS with the
+ * appropriate start location.
+ *
+ * Returns:
+ *	0	- failure
+ *	1	- success
+ */
+static inline int __devinit smp_startup_cpu(unsigned int lcpu)
+{
+	int status;
+	unsigned long start_here = __pa((u32)*((unsigned long *)
+					       pSeries_secondary_smp_init));
+	unsigned int pcpu;
+	int start_cpu;
+
+	if (cpu_isset(lcpu, of_spin_map))
+		/* Already started by OF and sitting in spin loop */
+		return 1;
+
+	pcpu = get_hard_smp_processor_id(lcpu);
+
+	/* Fixup atomic count: it exited inside IRQ handler. */
+	paca[lcpu].__current->thread_info->preempt_count	= 0;
+
+	/* 
+	 * If the RTAS start-cpu token does not exist then presume the
+	 * cpu is already spinning.
+	 */
+	start_cpu = rtas_token("start-cpu");
+	if (start_cpu == RTAS_UNKNOWN_SERVICE)
+		return 1;
+
+	status = rtas_call(start_cpu, 3, 1, NULL, pcpu, start_here, lcpu);
+	if (status != 0) {
+		printk(KERN_ERR "start-cpu failed: %i\n", status);
+		return 0;
+	}
+
+	return 1;
+}
+
+#ifdef CONFIG_XICS
+static inline void smp_xics_do_message(int cpu, int msg)
+{
+	set_bit(msg, &xics_ipi_message[cpu].value);
+	mb();
+	xics_cause_IPI(cpu);
+}
+
+static void smp_xics_message_pass(int target, int msg)
+{
+	unsigned int i;
+
+	if (target < NR_CPUS) {
+		smp_xics_do_message(target, msg);
+	} else {
+		for_each_online_cpu(i) {
+			if (target == MSG_ALL_BUT_SELF
+			    && i == smp_processor_id())
+				continue;
+			smp_xics_do_message(i, msg);
+		}
+	}
+}
+
+static int __init smp_xics_probe(void)
+{
+	xics_request_IPIs();
+
+	return cpus_weight(cpu_possible_map);
+}
+
+static void __devinit smp_xics_setup_cpu(int cpu)
+{
+	if (cpu != boot_cpuid)
+		xics_setup_cpu();
+
+	if (firmware_has_feature(FW_FEATURE_SPLPAR))
+		vpa_init(cpu);
+
+	cpu_clear(cpu, of_spin_map);
+
+}
+#endif /* CONFIG_XICS */
+
+static DEFINE_SPINLOCK(timebase_lock);
+static unsigned long timebase = 0;
+
+static void __devinit pSeries_give_timebase(void)
+{
+	spin_lock(&timebase_lock);
+	rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL);
+	timebase = get_tb();
+	spin_unlock(&timebase_lock);
+
+	while (timebase)
+		barrier();
+	rtas_call(rtas_token("thaw-time-base"), 0, 1, NULL);
+}
+
+static void __devinit pSeries_take_timebase(void)
+{
+	while (!timebase)
+		barrier();
+	spin_lock(&timebase_lock);
+	set_tb(timebase >> 32, timebase & 0xffffffff);
+	timebase = 0;
+	spin_unlock(&timebase_lock);
+}
+
+static void __devinit smp_pSeries_kick_cpu(int nr)
+{
+	BUG_ON(nr < 0 || nr >= NR_CPUS);
+
+	if (!smp_startup_cpu(nr))
+		return;
+
+	/*
+	 * The processor is currently spinning, waiting for the
+	 * cpu_start field to become non-zero After we set cpu_start,
+	 * the processor will continue on to secondary_start
+	 */
+	paca[nr].cpu_start = 1;
+}
+
+static int smp_pSeries_cpu_bootable(unsigned int nr)
+{
+	/* Special case - we inhibit secondary thread startup
+	 * during boot if the user requests it.  Odd-numbered
+	 * cpus are assumed to be secondary threads.
+	 */
+	if (system_state < SYSTEM_RUNNING &&
+	    cpu_has_feature(CPU_FTR_SMT) &&
+	    !smt_enabled_at_boot && nr % 2 != 0)
+		return 0;
+
+	return 1;
+}
+#ifdef CONFIG_MPIC
+static struct smp_ops_t pSeries_mpic_smp_ops = {
+	.message_pass	= smp_mpic_message_pass,
+	.probe		= smp_mpic_probe,
+	.kick_cpu	= smp_pSeries_kick_cpu,
+	.setup_cpu	= smp_mpic_setup_cpu,
+};
+#endif
+#ifdef CONFIG_XICS
+static struct smp_ops_t pSeries_xics_smp_ops = {
+	.message_pass	= smp_xics_message_pass,
+	.probe		= smp_xics_probe,
+	.kick_cpu	= smp_pSeries_kick_cpu,
+	.setup_cpu	= smp_xics_setup_cpu,
+	.cpu_bootable	= smp_pSeries_cpu_bootable,
+};
+#endif
+
+/* This is called very early */
+void __init smp_init_pSeries(void)
+{
+	int i;
+
+	DBG(" -> smp_init_pSeries()\n");
+
+	switch (ppc64_interrupt_controller) {
+#ifdef CONFIG_MPIC
+	case IC_OPEN_PIC:
+		smp_ops = &pSeries_mpic_smp_ops;
+		break;
+#endif
+#ifdef CONFIG_XICS
+	case IC_PPC_XIC:
+		smp_ops = &pSeries_xics_smp_ops;
+		break;
+#endif
+	default:
+		panic("Invalid interrupt controller");
+	}
+
+#ifdef CONFIG_HOTPLUG_CPU
+	smp_ops->cpu_disable = pSeries_cpu_disable;
+	smp_ops->cpu_die = pSeries_cpu_die;
+
+	/* Processors can be added/removed only on LPAR */
+	if (systemcfg->platform == PLATFORM_PSERIES_LPAR)
+		pSeries_reconfig_notifier_register(&pSeries_smp_nb);
+#endif
+
+	/* Mark threads which are still spinning in hold loops. */
+	if (cpu_has_feature(CPU_FTR_SMT)) {
+		for_each_present_cpu(i) { 
+			if (i % 2 == 0)
+				/*
+				 * Even-numbered logical cpus correspond to
+				 * primary threads.
+				 */
+				cpu_set(i, of_spin_map);
+		}
+	} else {
+		of_spin_map = cpu_present_map;
+	}
+
+	cpu_clear(boot_cpuid, of_spin_map);
+
+	/* Non-lpar has additional take/give timebase */
+	if (rtas_token("freeze-time-base") != RTAS_UNKNOWN_SERVICE) {
+		smp_ops->give_timebase = pSeries_give_timebase;
+		smp_ops->take_timebase = pSeries_take_timebase;
+	}
+
+	DBG(" <- smp_init_pSeries()\n");
+}
+
diff --git a/arch/powerpc/platforms/pseries/vio.c b/arch/powerpc/platforms/pseries/vio.c
new file mode 100644
index 000000000000..866379b80c09
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/vio.c
@@ -0,0 +1,274 @@
+/*
+ * IBM PowerPC pSeries Virtual I/O Infrastructure Support.
+ *
+ *    Copyright (c) 2003-2005 IBM Corp.
+ *     Dave Engebretsen engebret@us.ibm.com
+ *     Santiago Leon santil@us.ibm.com
+ *     Hollis Blanchard <hollisb@us.ibm.com>
+ *     Stephen Rothwell
+ *
+ *      This program is free software; you can redistribute it and/or
+ *      modify it under the terms of the GNU General Public License
+ *      as published by the Free Software Foundation; either version
+ *      2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/kobject.h>
+#include <asm/iommu.h>
+#include <asm/dma.h>
+#include <asm/prom.h>
+#include <asm/vio.h>
+#include <asm/hvcall.h>
+#include <asm/tce.h>
+
+extern struct subsystem devices_subsys; /* needed for vio_find_name() */
+
+static void probe_bus_pseries(void)
+{
+	struct device_node *node_vroot, *of_node;
+
+	node_vroot = find_devices("vdevice");
+	if ((node_vroot == NULL) || (node_vroot->child == NULL))
+		/* this machine doesn't do virtual IO, and that's ok */
+		return;
+
+	/*
+	 * Create struct vio_devices for each virtual device in the device tree.
+	 * Drivers will associate with them later.
+	 */
+	for (of_node = node_vroot->child; of_node != NULL;
+			of_node = of_node->sibling) {
+		printk(KERN_DEBUG "%s: processing %p\n", __FUNCTION__, of_node);
+		vio_register_device_node(of_node);
+	}
+}
+
+/**
+ * vio_match_device_pseries: - Tell if a pSeries VIO device matches a
+ *	vio_device_id
+ */
+static int vio_match_device_pseries(const struct vio_device_id *id,
+		const struct vio_dev *dev)
+{
+	return (strncmp(dev->type, id->type, strlen(id->type)) == 0) &&
+			device_is_compatible(dev->dev.platform_data, id->compat);
+}
+
+static void vio_release_device_pseries(struct device *dev)
+{
+	/* XXX free TCE table */
+	of_node_put(dev->platform_data);
+}
+
+static ssize_t viodev_show_devspec(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct device_node *of_node = dev->platform_data;
+
+	return sprintf(buf, "%s\n", of_node->full_name);
+}
+DEVICE_ATTR(devspec, S_IRUSR | S_IRGRP | S_IROTH, viodev_show_devspec, NULL);
+
+static void vio_unregister_device_pseries(struct vio_dev *viodev)
+{
+	device_remove_file(&viodev->dev, &dev_attr_devspec);
+}
+
+static struct vio_bus_ops vio_bus_ops_pseries = {
+	.match = vio_match_device_pseries,
+	.unregister_device = vio_unregister_device_pseries,
+	.release_device = vio_release_device_pseries,
+};
+
+/**
+ * vio_bus_init_pseries: - Initialize the pSeries virtual IO bus
+ */
+static int __init vio_bus_init_pseries(void)
+{
+	int err;
+
+	err = vio_bus_init(&vio_bus_ops_pseries);
+	if (err == 0)
+		probe_bus_pseries();
+	return err;
+}
+
+__initcall(vio_bus_init_pseries);
+
+/**
+ * vio_build_iommu_table: - gets the dma information from OF and
+ *	builds the TCE tree.
+ * @dev: the virtual device.
+ *
+ * Returns a pointer to the built tce tree, or NULL if it can't
+ * find property.
+*/
+static struct iommu_table *vio_build_iommu_table(struct vio_dev *dev)
+{
+	unsigned int *dma_window;
+	struct iommu_table *newTceTable;
+	unsigned long offset;
+	int dma_window_property_size;
+
+	dma_window = (unsigned int *) get_property(dev->dev.platform_data, "ibm,my-dma-window", &dma_window_property_size);
+	if(!dma_window) {
+		return NULL;
+	}
+
+	newTceTable = (struct iommu_table *) kmalloc(sizeof(struct iommu_table), GFP_KERNEL);
+
+	/*  There should be some code to extract the phys-encoded offset
+		using prom_n_addr_cells(). However, according to a comment
+		on earlier versions, it's always zero, so we don't bother */
+	offset = dma_window[1] >>  PAGE_SHIFT;
+
+	/* TCE table size - measured in tce entries */
+	newTceTable->it_size		= dma_window[4] >> PAGE_SHIFT;
+	/* offset for VIO should always be 0 */
+	newTceTable->it_offset		= offset;
+	newTceTable->it_busno		= 0;
+	newTceTable->it_index		= (unsigned long)dma_window[0];
+	newTceTable->it_type		= TCE_VB;
+
+	return iommu_init_table(newTceTable);
+}
+
+/**
+ * vio_register_device_node: - Register a new vio device.
+ * @of_node:	The OF node for this device.
+ *
+ * Creates and initializes a vio_dev structure from the data in
+ * of_node (dev.platform_data) and adds it to the list of virtual devices.
+ * Returns a pointer to the created vio_dev or NULL if node has
+ * NULL device_type or compatible fields.
+ */
+struct vio_dev * __devinit vio_register_device_node(struct device_node *of_node)
+{
+	struct vio_dev *viodev;
+	unsigned int *unit_address;
+	unsigned int *irq_p;
+
+	/* we need the 'device_type' property, in order to match with drivers */
+	if ((NULL == of_node->type)) {
+		printk(KERN_WARNING
+			"%s: node %s missing 'device_type'\n", __FUNCTION__,
+			of_node->name ? of_node->name : "<unknown>");
+		return NULL;
+	}
+
+	unit_address = (unsigned int *)get_property(of_node, "reg", NULL);
+	if (!unit_address) {
+		printk(KERN_WARNING "%s: node %s missing 'reg'\n", __FUNCTION__,
+			of_node->name ? of_node->name : "<unknown>");
+		return NULL;
+	}
+
+	/* allocate a vio_dev for this node */
+	viodev = kmalloc(sizeof(struct vio_dev), GFP_KERNEL);
+	if (!viodev) {
+		return NULL;
+	}
+	memset(viodev, 0, sizeof(struct vio_dev));
+
+	viodev->dev.platform_data = of_node_get(of_node);
+
+	viodev->irq = NO_IRQ;
+	irq_p = (unsigned int *)get_property(of_node, "interrupts", NULL);
+	if (irq_p) {
+		int virq = virt_irq_create_mapping(*irq_p);
+		if (virq == NO_IRQ) {
+			printk(KERN_ERR "Unable to allocate interrupt "
+			       "number for %s\n", of_node->full_name);
+		} else
+			viodev->irq = irq_offset_up(virq);
+	}
+
+	snprintf(viodev->dev.bus_id, BUS_ID_SIZE, "%x", *unit_address);
+	viodev->name = of_node->name;
+	viodev->type = of_node->type;
+	viodev->unit_address = *unit_address;
+	viodev->iommu_table = vio_build_iommu_table(viodev);
+
+	/* register with generic device framework */
+	if (vio_register_device(viodev) == NULL) {
+		/* XXX free TCE table */
+		kfree(viodev);
+		return NULL;
+	}
+	device_create_file(&viodev->dev, &dev_attr_devspec);
+
+	return viodev;
+}
+EXPORT_SYMBOL(vio_register_device_node);
+
+/**
+ * vio_get_attribute: - get attribute for virtual device
+ * @vdev:	The vio device to get property.
+ * @which:	The property/attribute to be extracted.
+ * @length:	Pointer to length of returned data size (unused if NULL).
+ *
+ * Calls prom.c's get_property() to return the value of the
+ * attribute specified by the preprocessor constant @which
+*/
+const void * vio_get_attribute(struct vio_dev *vdev, void* which, int* length)
+{
+	return get_property(vdev->dev.platform_data, (char*)which, length);
+}
+EXPORT_SYMBOL(vio_get_attribute);
+
+/* vio_find_name() - internal because only vio.c knows how we formatted the
+ * kobject name
+ * XXX once vio_bus_type.devices is actually used as a kset in
+ * drivers/base/bus.c, this function should be removed in favor of
+ * "device_find(kobj_name, &vio_bus_type)"
+ */
+static struct vio_dev *vio_find_name(const char *kobj_name)
+{
+	struct kobject *found;
+
+	found = kset_find_obj(&devices_subsys.kset, kobj_name);
+	if (!found)
+		return NULL;
+
+	return to_vio_dev(container_of(found, struct device, kobj));
+}
+
+/**
+ * vio_find_node - find an already-registered vio_dev
+ * @vnode: device_node of the virtual device we're looking for
+ */
+struct vio_dev *vio_find_node(struct device_node *vnode)
+{
+	uint32_t *unit_address;
+	char kobj_name[BUS_ID_SIZE];
+
+	/* construct the kobject name from the device node */
+	unit_address = (uint32_t *)get_property(vnode, "reg", NULL);
+	if (!unit_address)
+		return NULL;
+	snprintf(kobj_name, BUS_ID_SIZE, "%x", *unit_address);
+
+	return vio_find_name(kobj_name);
+}
+EXPORT_SYMBOL(vio_find_node);
+
+int vio_enable_interrupts(struct vio_dev *dev)
+{
+	int rc = h_vio_signal(dev->unit_address, VIO_IRQ_ENABLE);
+	if (rc != H_Success)
+		printk(KERN_ERR "vio: Error 0x%x enabling interrupts\n", rc);
+	return rc;
+}
+EXPORT_SYMBOL(vio_enable_interrupts);
+
+int vio_disable_interrupts(struct vio_dev *dev)
+{
+	int rc = h_vio_signal(dev->unit_address, VIO_IRQ_DISABLE);
+	if (rc != H_Success)
+		printk(KERN_ERR "vio: Error 0x%x disabling interrupts\n", rc);
+	return rc;
+}
+EXPORT_SYMBOL(vio_disable_interrupts);
diff --git a/arch/powerpc/platforms/pseries/xics.c b/arch/powerpc/platforms/pseries/xics.c
new file mode 100644
index 000000000000..c72c86f05cb6
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/xics.c
@@ -0,0 +1,747 @@
+/*
+ * arch/powerpc/platforms/pseries/xics.c
+ *
+ * Copyright 2000 IBM Corporation.
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ */
+#include <linux/config.h>
+#include <linux/types.h>
+#include <linux/threads.h>
+#include <linux/kernel.h>
+#include <linux/irq.h>
+#include <linux/smp.h>
+#include <linux/interrupt.h>
+#include <linux/signal.h>
+#include <linux/init.h>
+#include <linux/gfp.h>
+#include <linux/radix-tree.h>
+#include <linux/cpu.h>
+#include <asm/prom.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/smp.h>
+#include <asm/rtas.h>
+#include <asm/hvcall.h>
+#include <asm/machdep.h>
+#include <asm/i8259.h>
+
+#include "xics.h"
+
+static unsigned int xics_startup(unsigned int irq);
+static void xics_enable_irq(unsigned int irq);
+static void xics_disable_irq(unsigned int irq);
+static void xics_mask_and_ack_irq(unsigned int irq);
+static void xics_end_irq(unsigned int irq);
+static void xics_set_affinity(unsigned int irq_nr, cpumask_t cpumask);
+
+static struct hw_interrupt_type xics_pic = {
+	.typename = " XICS     ",
+	.startup = xics_startup,
+	.enable = xics_enable_irq,
+	.disable = xics_disable_irq,
+	.ack = xics_mask_and_ack_irq,
+	.end = xics_end_irq,
+	.set_affinity = xics_set_affinity
+};
+
+static struct hw_interrupt_type xics_8259_pic = {
+	.typename = " XICS/8259",
+	.ack = xics_mask_and_ack_irq,
+};
+
+/* This is used to map real irq numbers to virtual */
+static struct radix_tree_root irq_map = RADIX_TREE_INIT(GFP_ATOMIC);
+
+#define XICS_IPI		2
+#define XICS_IRQ_SPURIOUS	0
+
+/* Want a priority other than 0.  Various HW issues require this. */
+#define	DEFAULT_PRIORITY	5
+
+/*
+ * Mark IPIs as higher priority so we can take them inside interrupts that
+ * arent marked SA_INTERRUPT
+ */
+#define IPI_PRIORITY		4
+
+struct xics_ipl {
+	union {
+		u32 word;
+		u8 bytes[4];
+	} xirr_poll;
+	union {
+		u32 word;
+		u8 bytes[4];
+	} xirr;
+	u32 dummy;
+	union {
+		u32 word;
+		u8 bytes[4];
+	} qirr;
+};
+
+static struct xics_ipl __iomem *xics_per_cpu[NR_CPUS];
+
+static int xics_irq_8259_cascade = 0;
+static int xics_irq_8259_cascade_real = 0;
+static unsigned int default_server = 0xFF;
+static unsigned int default_distrib_server = 0;
+static unsigned int interrupt_server_size = 8;
+
+/*
+ * XICS only has a single IPI, so encode the messages per CPU
+ */
+struct xics_ipi_struct xics_ipi_message[NR_CPUS] __cacheline_aligned;
+
+/* RTAS service tokens */
+static int ibm_get_xive;
+static int ibm_set_xive;
+static int ibm_int_on;
+static int ibm_int_off;
+
+typedef struct {
+	int (*xirr_info_get)(int cpu);
+	void (*xirr_info_set)(int cpu, int val);
+	void (*cppr_info)(int cpu, u8 val);
+	void (*qirr_info)(int cpu, u8 val);
+} xics_ops;
+
+
+/* SMP */
+
+static int pSeries_xirr_info_get(int n_cpu)
+{
+	return in_be32(&xics_per_cpu[n_cpu]->xirr.word);
+}
+
+static void pSeries_xirr_info_set(int n_cpu, int value)
+{
+	out_be32(&xics_per_cpu[n_cpu]->xirr.word, value);
+}
+
+static void pSeries_cppr_info(int n_cpu, u8 value)
+{
+	out_8(&xics_per_cpu[n_cpu]->xirr.bytes[0], value);
+}
+
+static void pSeries_qirr_info(int n_cpu, u8 value)
+{
+	out_8(&xics_per_cpu[n_cpu]->qirr.bytes[0], value);
+}
+
+static xics_ops pSeries_ops = {
+	pSeries_xirr_info_get,
+	pSeries_xirr_info_set,
+	pSeries_cppr_info,
+	pSeries_qirr_info
+};
+
+static xics_ops *ops = &pSeries_ops;
+
+
+/* LPAR */
+
+static inline long plpar_eoi(unsigned long xirr)
+{
+	return plpar_hcall_norets(H_EOI, xirr);
+}
+
+static inline long plpar_cppr(unsigned long cppr)
+{
+	return plpar_hcall_norets(H_CPPR, cppr);
+}
+
+static inline long plpar_ipi(unsigned long servernum, unsigned long mfrr)
+{
+	return plpar_hcall_norets(H_IPI, servernum, mfrr);
+}
+
+static inline long plpar_xirr(unsigned long *xirr_ret)
+{
+	unsigned long dummy;
+	return plpar_hcall(H_XIRR, 0, 0, 0, 0, xirr_ret, &dummy, &dummy);
+}
+
+static int pSeriesLP_xirr_info_get(int n_cpu)
+{
+	unsigned long lpar_rc;
+	unsigned long return_value;
+
+	lpar_rc = plpar_xirr(&return_value);
+	if (lpar_rc != H_Success)
+		panic(" bad return code xirr - rc = %lx \n", lpar_rc);
+	return (int)return_value;
+}
+
+static void pSeriesLP_xirr_info_set(int n_cpu, int value)
+{
+	unsigned long lpar_rc;
+	unsigned long val64 = value & 0xffffffff;
+
+	lpar_rc = plpar_eoi(val64);
+	if (lpar_rc != H_Success)
+		panic("bad return code EOI - rc = %ld, value=%lx\n", lpar_rc,
+		      val64);
+}
+
+void pSeriesLP_cppr_info(int n_cpu, u8 value)
+{
+	unsigned long lpar_rc;
+
+	lpar_rc = plpar_cppr(value);
+	if (lpar_rc != H_Success)
+		panic("bad return code cppr - rc = %lx\n", lpar_rc);
+}
+
+static void pSeriesLP_qirr_info(int n_cpu , u8 value)
+{
+	unsigned long lpar_rc;
+
+	lpar_rc = plpar_ipi(get_hard_smp_processor_id(n_cpu), value);
+	if (lpar_rc != H_Success)
+		panic("bad return code qirr - rc = %lx\n", lpar_rc);
+}
+
+xics_ops pSeriesLP_ops = {
+	pSeriesLP_xirr_info_get,
+	pSeriesLP_xirr_info_set,
+	pSeriesLP_cppr_info,
+	pSeriesLP_qirr_info
+};
+
+static unsigned int xics_startup(unsigned int virq)
+{
+	unsigned int irq;
+
+	irq = irq_offset_down(virq);
+	if (radix_tree_insert(&irq_map, virt_irq_to_real(irq),
+			      &virt_irq_to_real_map[irq]) == -ENOMEM)
+		printk(KERN_CRIT "Out of memory creating real -> virtual"
+		       " IRQ mapping for irq %u (real 0x%x)\n",
+		       virq, virt_irq_to_real(irq));
+	xics_enable_irq(virq);
+	return 0;	/* return value is ignored */
+}
+
+static unsigned int real_irq_to_virt(unsigned int real_irq)
+{
+	unsigned int *ptr;
+
+	ptr = radix_tree_lookup(&irq_map, real_irq);
+	if (ptr == NULL)
+		return NO_IRQ;
+	return ptr - virt_irq_to_real_map;
+}
+
+#ifdef CONFIG_SMP
+static int get_irq_server(unsigned int irq)
+{
+	unsigned int server;
+	/* For the moment only implement delivery to all cpus or one cpu */
+	cpumask_t cpumask = irq_affinity[irq];
+	cpumask_t tmp = CPU_MASK_NONE;
+
+	if (!distribute_irqs)
+		return default_server;
+
+	if (cpus_equal(cpumask, CPU_MASK_ALL)) {
+		server = default_distrib_server;
+	} else {
+		cpus_and(tmp, cpu_online_map, cpumask);
+
+		if (cpus_empty(tmp))
+			server = default_distrib_server;
+		else
+			server = get_hard_smp_processor_id(first_cpu(tmp));
+	}
+
+	return server;
+
+}
+#else
+static int get_irq_server(unsigned int irq)
+{
+	return default_server;
+}
+#endif
+
+static void xics_enable_irq(unsigned int virq)
+{
+	unsigned int irq;
+	int call_status;
+	unsigned int server;
+
+	irq = virt_irq_to_real(irq_offset_down(virq));
+	if (irq == XICS_IPI)
+		return;
+
+	server = get_irq_server(virq);
+	call_status = rtas_call(ibm_set_xive, 3, 1, NULL, irq, server,
+				DEFAULT_PRIORITY);
+	if (call_status != 0) {
+		printk(KERN_ERR "xics_enable_irq: irq=%u: ibm_set_xive "
+		       "returned %d\n", irq, call_status);
+		printk("set_xive %x, server %x\n", ibm_set_xive, server);
+		return;
+	}
+
+	/* Now unmask the interrupt (often a no-op) */
+	call_status = rtas_call(ibm_int_on, 1, 1, NULL, irq);
+	if (call_status != 0) {
+		printk(KERN_ERR "xics_enable_irq: irq=%u: ibm_int_on "
+		       "returned %d\n", irq, call_status);
+		return;
+	}
+}
+
+static void xics_disable_real_irq(unsigned int irq)
+{
+	int call_status;
+	unsigned int server;
+
+	if (irq == XICS_IPI)
+		return;
+
+	call_status = rtas_call(ibm_int_off, 1, 1, NULL, irq);
+	if (call_status != 0) {
+		printk(KERN_ERR "xics_disable_real_irq: irq=%u: "
+		       "ibm_int_off returned %d\n", irq, call_status);
+		return;
+	}
+
+	server = get_irq_server(irq);
+	/* Have to set XIVE to 0xff to be able to remove a slot */
+	call_status = rtas_call(ibm_set_xive, 3, 1, NULL, irq, server, 0xff);
+	if (call_status != 0) {
+		printk(KERN_ERR "xics_disable_irq: irq=%u: ibm_set_xive(0xff)"
+		       " returned %d\n", irq, call_status);
+		return;
+	}
+}
+
+static void xics_disable_irq(unsigned int virq)
+{
+	unsigned int irq;
+
+	irq = virt_irq_to_real(irq_offset_down(virq));
+	xics_disable_real_irq(irq);
+}
+
+static void xics_end_irq(unsigned int irq)
+{
+	int cpu = smp_processor_id();
+
+	iosync();
+	ops->xirr_info_set(cpu, ((0xff << 24) |
+				 (virt_irq_to_real(irq_offset_down(irq)))));
+
+}
+
+static void xics_mask_and_ack_irq(unsigned int irq)
+{
+	int cpu = smp_processor_id();
+
+	if (irq < irq_offset_value()) {
+		i8259_pic.ack(irq);
+		iosync();
+		ops->xirr_info_set(cpu, ((0xff<<24) |
+					 xics_irq_8259_cascade_real));
+		iosync();
+	}
+}
+
+int xics_get_irq(struct pt_regs *regs)
+{
+	unsigned int cpu = smp_processor_id();
+	unsigned int vec;
+	int irq;
+
+	vec = ops->xirr_info_get(cpu);
+	/*  (vec >> 24) == old priority */
+	vec &= 0x00ffffff;
+
+	/* for sanity, this had better be < NR_IRQS - 16 */
+	if (vec == xics_irq_8259_cascade_real) {
+		irq = i8259_irq(regs);
+		if (irq == -1) {
+			/* Spurious cascaded interrupt.  Still must ack xics */
+			xics_end_irq(irq_offset_up(xics_irq_8259_cascade));
+
+			irq = -1;
+		}
+	} else if (vec == XICS_IRQ_SPURIOUS) {
+		irq = -1;
+	} else {
+		irq = real_irq_to_virt(vec);
+		if (irq == NO_IRQ)
+			irq = real_irq_to_virt_slowpath(vec);
+		if (irq == NO_IRQ) {
+			printk(KERN_ERR "Interrupt %u (real) is invalid,"
+			       " disabling it.\n", vec);
+			xics_disable_real_irq(vec);
+		} else
+			irq = irq_offset_up(irq);
+	}
+	return irq;
+}
+
+#ifdef CONFIG_SMP
+
+irqreturn_t xics_ipi_action(int irq, void *dev_id, struct pt_regs *regs)
+{
+	int cpu = smp_processor_id();
+
+	ops->qirr_info(cpu, 0xff);
+
+	WARN_ON(cpu_is_offline(cpu));
+
+	while (xics_ipi_message[cpu].value) {
+		if (test_and_clear_bit(PPC_MSG_CALL_FUNCTION,
+				       &xics_ipi_message[cpu].value)) {
+			mb();
+			smp_message_recv(PPC_MSG_CALL_FUNCTION, regs);
+		}
+		if (test_and_clear_bit(PPC_MSG_RESCHEDULE,
+				       &xics_ipi_message[cpu].value)) {
+			mb();
+			smp_message_recv(PPC_MSG_RESCHEDULE, regs);
+		}
+#if 0
+		if (test_and_clear_bit(PPC_MSG_MIGRATE_TASK,
+				       &xics_ipi_message[cpu].value)) {
+			mb();
+			smp_message_recv(PPC_MSG_MIGRATE_TASK, regs);
+		}
+#endif
+#ifdef CONFIG_DEBUGGER
+		if (test_and_clear_bit(PPC_MSG_DEBUGGER_BREAK,
+				       &xics_ipi_message[cpu].value)) {
+			mb();
+			smp_message_recv(PPC_MSG_DEBUGGER_BREAK, regs);
+		}
+#endif
+	}
+	return IRQ_HANDLED;
+}
+
+void xics_cause_IPI(int cpu)
+{
+	ops->qirr_info(cpu, IPI_PRIORITY);
+}
+#endif /* CONFIG_SMP */
+
+void xics_setup_cpu(void)
+{
+	int cpu = smp_processor_id();
+
+	ops->cppr_info(cpu, 0xff);
+	iosync();
+
+	/*
+	 * Put the calling processor into the GIQ.  This is really only
+	 * necessary from a secondary thread as the OF start-cpu interface
+	 * performs this function for us on primary threads.
+	 *
+	 * XXX: undo of teardown on kexec needs this too, as may hotplug
+	 */
+	rtas_set_indicator(GLOBAL_INTERRUPT_QUEUE,
+		(1UL << interrupt_server_size) - 1 - default_distrib_server, 1);
+}
+
+void xics_init_IRQ(void)
+{
+	int i;
+	unsigned long intr_size = 0;
+	struct device_node *np;
+	uint *ireg, ilen, indx = 0;
+	unsigned long intr_base = 0;
+	struct xics_interrupt_node {
+		unsigned long addr;
+		unsigned long size;
+	} intnodes[NR_CPUS];
+
+	ppc64_boot_msg(0x20, "XICS Init");
+
+	ibm_get_xive = rtas_token("ibm,get-xive");
+	ibm_set_xive = rtas_token("ibm,set-xive");
+	ibm_int_on  = rtas_token("ibm,int-on");
+	ibm_int_off = rtas_token("ibm,int-off");
+
+	np = of_find_node_by_type(NULL, "PowerPC-External-Interrupt-Presentation");
+	if (!np)
+		panic("xics_init_IRQ: can't find interrupt presentation");
+
+nextnode:
+	ireg = (uint *)get_property(np, "ibm,interrupt-server-ranges", NULL);
+	if (ireg) {
+		/*
+		 * set node starting index for this node
+		 */
+		indx = *ireg;
+	}
+
+	ireg = (uint *)get_property(np, "reg", &ilen);
+	if (!ireg)
+		panic("xics_init_IRQ: can't find interrupt reg property");
+
+	while (ilen) {
+		intnodes[indx].addr = (unsigned long)*ireg++ << 32;
+		ilen -= sizeof(uint);
+		intnodes[indx].addr |= *ireg++;
+		ilen -= sizeof(uint);
+		intnodes[indx].size = (unsigned long)*ireg++ << 32;
+		ilen -= sizeof(uint);
+		intnodes[indx].size |= *ireg++;
+		ilen -= sizeof(uint);
+		indx++;
+		if (indx >= NR_CPUS) break;
+	}
+
+	np = of_find_node_by_type(np, "PowerPC-External-Interrupt-Presentation");
+	if ((indx < NR_CPUS) && np) goto nextnode;
+
+	/* Find the server numbers for the boot cpu. */
+	for (np = of_find_node_by_type(NULL, "cpu");
+	     np;
+	     np = of_find_node_by_type(np, "cpu")) {
+		ireg = (uint *)get_property(np, "reg", &ilen);
+		if (ireg && ireg[0] == boot_cpuid_phys) {
+			ireg = (uint *)get_property(np, "ibm,ppc-interrupt-gserver#s",
+						    &ilen);
+			i = ilen / sizeof(int);
+			if (ireg && i > 0) {
+				default_server = ireg[0];
+				default_distrib_server = ireg[i-1]; /* take last element */
+			}
+			ireg = (uint *)get_property(np,
+					"ibm,interrupt-server#-size", NULL);
+			if (ireg)
+				interrupt_server_size = *ireg;
+			break;
+		}
+	}
+	of_node_put(np);
+
+	intr_base = intnodes[0].addr;
+	intr_size = intnodes[0].size;
+
+	np = of_find_node_by_type(NULL, "interrupt-controller");
+	if (!np) {
+		printk(KERN_WARNING "xics: no ISA interrupt controller\n");
+		xics_irq_8259_cascade_real = -1;
+		xics_irq_8259_cascade = -1;
+	} else {
+		ireg = (uint *) get_property(np, "interrupts", NULL);
+		if (!ireg)
+			panic("xics_init_IRQ: can't find ISA interrupts property");
+
+		xics_irq_8259_cascade_real = *ireg;
+		xics_irq_8259_cascade
+			= virt_irq_create_mapping(xics_irq_8259_cascade_real);
+		of_node_put(np);
+	}
+
+	if (systemcfg->platform == PLATFORM_PSERIES) {
+#ifdef CONFIG_SMP
+		for_each_cpu(i) {
+			int hard_id;
+
+			/* FIXME: Do this dynamically! --RR */
+			if (!cpu_present(i))
+				continue;
+
+			hard_id = get_hard_smp_processor_id(i);
+			xics_per_cpu[i] = ioremap(intnodes[hard_id].addr,
+						  intnodes[hard_id].size);
+		}
+#else
+		xics_per_cpu[0] = ioremap(intr_base, intr_size);
+#endif /* CONFIG_SMP */
+	} else if (systemcfg->platform == PLATFORM_PSERIES_LPAR) {
+		ops = &pSeriesLP_ops;
+	}
+
+	xics_8259_pic.enable = i8259_pic.enable;
+	xics_8259_pic.disable = i8259_pic.disable;
+	for (i = 0; i < 16; ++i)
+		get_irq_desc(i)->handler = &xics_8259_pic;
+	for (; i < NR_IRQS; ++i)
+		get_irq_desc(i)->handler = &xics_pic;
+
+	xics_setup_cpu();
+
+	ppc64_boot_msg(0x21, "XICS Done");
+}
+
+/*
+ * We cant do this in init_IRQ because we need the memory subsystem up for
+ * request_irq()
+ */
+static int __init xics_setup_i8259(void)
+{
+	if (ppc64_interrupt_controller == IC_PPC_XIC &&
+	    xics_irq_8259_cascade != -1) {
+		if (request_irq(irq_offset_up(xics_irq_8259_cascade),
+				no_action, 0, "8259 cascade", NULL))
+			printk(KERN_ERR "xics_setup_i8259: couldn't get 8259 "
+					"cascade\n");
+		i8259_init(0, 0);
+	}
+	return 0;
+}
+arch_initcall(xics_setup_i8259);
+
+#ifdef CONFIG_SMP
+void xics_request_IPIs(void)
+{
+	virt_irq_to_real_map[XICS_IPI] = XICS_IPI;
+
+	/* IPIs are marked SA_INTERRUPT as they must run with irqs disabled */
+	request_irq(irq_offset_up(XICS_IPI), xics_ipi_action, SA_INTERRUPT,
+		    "IPI", NULL);
+	get_irq_desc(irq_offset_up(XICS_IPI))->status |= IRQ_PER_CPU;
+}
+#endif
+
+static void xics_set_affinity(unsigned int virq, cpumask_t cpumask)
+{
+	unsigned int irq;
+	int status;
+	int xics_status[2];
+	unsigned long newmask;
+	cpumask_t tmp = CPU_MASK_NONE;
+
+	irq = virt_irq_to_real(irq_offset_down(virq));
+	if (irq == XICS_IPI || irq == NO_IRQ)
+		return;
+
+	status = rtas_call(ibm_get_xive, 1, 3, xics_status, irq);
+
+	if (status) {
+		printk(KERN_ERR "xics_set_affinity: irq=%u ibm,get-xive "
+		       "returns %d\n", irq, status);
+		return;
+	}
+
+	/* For the moment only implement delivery to all cpus or one cpu */
+	if (cpus_equal(cpumask, CPU_MASK_ALL)) {
+		newmask = default_distrib_server;
+	} else {
+		cpus_and(tmp, cpu_online_map, cpumask);
+		if (cpus_empty(tmp))
+			return;
+		newmask = get_hard_smp_processor_id(first_cpu(tmp));
+	}
+
+	status = rtas_call(ibm_set_xive, 3, 1, NULL,
+				irq, newmask, xics_status[1]);
+
+	if (status) {
+		printk(KERN_ERR "xics_set_affinity: irq=%u ibm,set-xive "
+		       "returns %d\n", irq, status);
+		return;
+	}
+}
+
+void xics_teardown_cpu(int secondary)
+{
+	int cpu = smp_processor_id();
+
+	ops->cppr_info(cpu, 0x00);
+	iosync();
+
+	/*
+	 * Some machines need to have at least one cpu in the GIQ,
+	 * so leave the master cpu in the group.
+	 */
+	if (secondary) {
+		/*
+		 * we need to EOI the IPI if we got here from kexec down IPI
+		 *
+		 * probably need to check all the other interrupts too
+		 * should we be flagging idle loop instead?
+		 * or creating some task to be scheduled?
+		 */
+		ops->xirr_info_set(cpu, XICS_IPI);
+		rtas_set_indicator(GLOBAL_INTERRUPT_QUEUE,
+			(1UL << interrupt_server_size) - 1 -
+			default_distrib_server, 0);
+	}
+}
+
+#ifdef CONFIG_HOTPLUG_CPU
+
+/* Interrupts are disabled. */
+void xics_migrate_irqs_away(void)
+{
+	int status;
+	unsigned int irq, virq, cpu = smp_processor_id();
+
+	/* Reject any interrupt that was queued to us... */
+	ops->cppr_info(cpu, 0);
+	iosync();
+
+	/* remove ourselves from the global interrupt queue */
+	status = rtas_set_indicator(GLOBAL_INTERRUPT_QUEUE,
+		(1UL << interrupt_server_size) - 1 - default_distrib_server, 0);
+	WARN_ON(status < 0);
+
+	/* Allow IPIs again... */
+	ops->cppr_info(cpu, DEFAULT_PRIORITY);
+	iosync();
+
+	for_each_irq(virq) {
+		irq_desc_t *desc;
+		int xics_status[2];
+		unsigned long flags;
+
+		/* We cant set affinity on ISA interrupts */
+		if (virq < irq_offset_value())
+			continue;
+
+		desc = get_irq_desc(virq);
+		irq = virt_irq_to_real(irq_offset_down(virq));
+
+		/* We need to get IPIs still. */
+		if (irq == XICS_IPI || irq == NO_IRQ)
+			continue;
+
+		/* We only need to migrate enabled IRQS */
+		if (desc == NULL || desc->handler == NULL
+		    || desc->action == NULL
+		    || desc->handler->set_affinity == NULL)
+			continue;
+
+		spin_lock_irqsave(&desc->lock, flags);
+
+		status = rtas_call(ibm_get_xive, 1, 3, xics_status, irq);
+		if (status) {
+			printk(KERN_ERR "migrate_irqs_away: irq=%u "
+					"ibm,get-xive returns %d\n",
+					virq, status);
+			goto unlock;
+		}
+
+		/*
+		 * We only support delivery to all cpus or to one cpu.
+		 * The irq has to be migrated only in the single cpu
+		 * case.
+		 */
+		if (xics_status[0] != get_hard_smp_processor_id(cpu))
+			goto unlock;
+
+		printk(KERN_WARNING "IRQ %u affinity broken off cpu %u\n",
+		       virq, cpu);
+
+		/* Reset affinity to all cpus */
+		desc->handler->set_affinity(virq, CPU_MASK_ALL);
+		irq_affinity[virq] = CPU_MASK_ALL;
+unlock:
+		spin_unlock_irqrestore(&desc->lock, flags);
+	}
+}
+#endif
diff --git a/arch/powerpc/platforms/pseries/xics.h b/arch/powerpc/platforms/pseries/xics.h
new file mode 100644
index 000000000000..e14c70868f1d
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/xics.h
@@ -0,0 +1,34 @@
+/*
+ * arch/powerpc/platforms/pseries/xics.h
+ *
+ * Copyright 2000 IBM Corporation.
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ */
+
+#ifndef _POWERPC_KERNEL_XICS_H
+#define _POWERPC_KERNEL_XICS_H
+
+#include <linux/cache.h>
+
+void xics_init_IRQ(void);
+int xics_get_irq(struct pt_regs *);
+void xics_setup_cpu(void);
+void xics_teardown_cpu(int secondary);
+void xics_cause_IPI(int cpu);
+void xics_request_IPIs(void);
+void xics_migrate_irqs_away(void);
+
+/* first argument is ignored for now*/
+void pSeriesLP_cppr_info(int n_cpu, u8 value);
+
+struct xics_ipi_struct {
+	volatile unsigned long value;
+} ____cacheline_aligned;
+
+extern struct xics_ipi_struct xics_ipi_message[NR_CPUS] __cacheline_aligned;
+
+#endif /* _POWERPC_KERNEL_XICS_H */