Merge branch 'for-rmk' of git://gitorious.org/linux-gemini/mainline into devel

Conflicts:
	arch/arm/mm/Kconfig

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

36 files changed, 2149 insertions, 15 deletions

diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index e62b37a..e02b893 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -276,6 +276,14 @@ config ARCH_EP93XX
 	help
 	  This enables support for the Cirrus EP93xx series of CPUs.
 
+config ARCH_GEMINI
+	bool "Cortina Systems Gemini"
+	select CPU_FA526
+	select GENERIC_GPIO
+	select ARCH_REQUIRE_GPIOLIB
+	help
+	  Support for the Cortina Systems Gemini family SoCs
+
 config ARCH_FOOTBRIDGE
 	bool "FootBridge"
 	select CPU_SA110
@@ -616,6 +624,8 @@ source "arch/arm/mach-ep93xx/Kconfig"
 
 source "arch/arm/mach-footbridge/Kconfig"
 
+source "arch/arm/mach-gemini/Kconfig"
+
 source "arch/arm/mach-integrator/Kconfig"
 
 source "arch/arm/mach-iop32x/Kconfig"
diff --git a/arch/arm/Makefile b/arch/arm/Makefile
index 95186ef..e84729b 100644
--- a/arch/arm/Makefile
+++ b/arch/arm/Makefile
@@ -72,6 +72,7 @@ tune-$(CONFIG_CPU_ARM920T)	:=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_ARM922T)	:=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_ARM925T)	:=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_ARM926T)	:=-mtune=arm9tdmi
+tune-$(CONFIG_CPU_FA526)	:=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_SA110)	:=-mtune=strongarm110
 tune-$(CONFIG_CPU_SA1100)	:=-mtune=strongarm1100
 tune-$(CONFIG_CPU_XSCALE)	:=$(call cc-option,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
@@ -113,6 +114,7 @@ endif
     plat-$(CONFIG_PLAT_PXA)	   := pxa
  machine-$(CONFIG_ARCH_L7200)	   := l7200
  machine-$(CONFIG_ARCH_INTEGRATOR) := integrator
+ machine-$(CONFIG_ARCH_GEMINI)     := gemini
  textofs-$(CONFIG_ARCH_CLPS711X)   := 0x00028000
  machine-$(CONFIG_ARCH_CLPS711X)   := clps711x
  machine-$(CONFIG_ARCH_IOP32X)	   := iop32x
diff --git a/arch/arm/boot/compressed/head.S b/arch/arm/boot/compressed/head.S
index d14b827..b371fba 100644
--- a/arch/arm/boot/compressed/head.S
+++ b/arch/arm/boot/compressed/head.S
@@ -465,6 +465,20 @@ __armv7_mmu_cache_on:
 		mcr	p15, 0, r0, c7, c5, 4	@ ISB
 		mov	pc, r12
 
+__fa526_cache_on:
+		mov	r12, lr
+		bl	__setup_mmu
+		mov	r0, #0
+		mcr	p15, 0, r0, c7, c7, 0	@ Invalidate whole cache
+		mcr	p15, 0, r0, c7, c10, 4	@ drain write buffer
+		mcr	p15, 0, r0, c8, c7, 0	@ flush UTLB
+		mrc	p15, 0, r0, c1, c0, 0	@ read control reg
+		orr	r0, r0, #0x1000		@ I-cache enable
+		bl	__common_mmu_cache_on
+		mov	r0, #0
+		mcr	p15, 0, r0, c8, c7, 0	@ flush UTLB
+		mov	pc, r12
+
 __arm6_mmu_cache_on:
 		mov	r12, lr
 		bl	__setup_mmu
@@ -654,6 +668,12 @@ proc_types:
 		b	__armv4_mmu_cache_off
 		b	__armv5tej_mmu_cache_flush
 
+		.word	0x66015261		@ FA526
+		.word	0xff01fff1
+		b	__fa526_cache_on
+		b	__armv4_mmu_cache_off
+		b	__fa526_cache_flush
+
 		@ These match on the architecture ID
 
 		.word	0x00020000		@ ARMv4T
@@ -793,6 +813,12 @@ __armv4_mpu_cache_flush:
 		mcr	p15, 0, ip, c7, c10, 4	@ drain WB
 		mov	pc, lr
 		
+__fa526_cache_flush:
+		mov	r1, #0
+		mcr	p15, 0, r1, c7, c14, 0	@ clean and invalidate D cache
+		mcr	p15, 0, r1, c7, c5, 0	@ flush I cache
+		mcr	p15, 0, r1, c7, c10, 4	@ drain WB
+		mov	pc, lr
 
 __armv6_mmu_cache_flush:
 		mov	r1, #0
diff --git a/arch/arm/include/asm/cacheflush.h b/arch/arm/include/asm/cacheflush.h
index bfb0cb9..bb7d695 100644
--- a/arch/arm/include/asm/cacheflush.h
+++ b/arch/arm/include/asm/cacheflush.h
@@ -46,6 +46,14 @@
 # define MULTI_CACHE 1
 #endif
 
+#if defined(CONFIG_CPU_FA526)
+# ifdef _CACHE
+#  define MULTI_CACHE 1
+# else
+#  define _CACHE fa
+# endif
+#endif
+
 #if defined(CONFIG_CPU_ARM926T)
 # ifdef _CACHE
 #  define MULTI_CACHE 1
diff --git a/arch/arm/include/asm/page.h b/arch/arm/include/asm/page.h
index f341c9d..e6eb8a6 100644
--- a/arch/arm/include/asm/page.h
+++ b/arch/arm/include/asm/page.h
@@ -76,6 +76,14 @@
 # endif
 #endif
 
+#ifdef CONFIG_CPU_COPY_FA
+# ifdef _USER
+#  define MULTI_USER 1
+# else
+#  define _USER fa
+# endif
+#endif
+
 #ifdef CONFIG_CPU_SA1100
 # ifdef _USER
 #  define MULTI_USER 1
diff --git a/arch/arm/include/asm/proc-fns.h b/arch/arm/include/asm/proc-fns.h
index c625031..3976412 100644
--- a/arch/arm/include/asm/proc-fns.h
+++ b/arch/arm/include/asm/proc-fns.h
@@ -89,6 +89,14 @@
 #   define CPU_NAME cpu_arm922
 #  endif
 # endif
+# ifdef CONFIG_CPU_FA526
+#  ifdef CPU_NAME
+#   undef  MULTI_CPU
+#   define MULTI_CPU
+#  else
+#   define CPU_NAME cpu_fa526
+#  endif
+# endif
 # ifdef CONFIG_CPU_ARM925T
 #  ifdef CPU_NAME
 #   undef  MULTI_CPU
diff --git a/arch/arm/include/asm/system.h b/arch/arm/include/asm/system.h
index 0a0d49a..bd4dc8e 100644
--- a/arch/arm/include/asm/system.h
+++ b/arch/arm/include/asm/system.h
@@ -125,6 +125,12 @@ extern unsigned int user_debug;
 				    : : "r" (0) : "memory")
 #define dmb() __asm__ __volatile__ ("mcr p15, 0, %0, c7, c10, 5" \
 				    : : "r" (0) : "memory")
+#elif defined(CONFIG_CPU_FA526)
+#define isb() __asm__ __volatile__ ("mcr p15, 0, %0, c7, c5, 4" \
+				    : : "r" (0) : "memory")
+#define dsb() __asm__ __volatile__ ("mcr p15, 0, %0, c7, c10, 4" \
+				    : : "r" (0) : "memory")
+#define dmb() __asm__ __volatile__ ("" : : : "memory")
 #else
 #define isb() __asm__ __volatile__ ("" : : : "memory")
 #define dsb() __asm__ __volatile__ ("mcr p15, 0, %0, c7, c10, 4" \
diff --git a/arch/arm/include/asm/tlbflush.h b/arch/arm/include/asm/tlbflush.h
index b543a05..a622180 100644
--- a/arch/arm/include/asm/tlbflush.h
+++ b/arch/arm/include/asm/tlbflush.h
@@ -39,6 +39,7 @@
 #define TLB_V6_D_ASID	(1 << 17)
 #define TLB_V6_I_ASID	(1 << 18)
 
+#define TLB_BTB		(1 << 28)
 #define TLB_L2CLEAN_FR	(1 << 29)		/* Feroceon */
 #define TLB_DCLEAN	(1 << 30)
 #define TLB_WB		(1 << 31)
@@ -53,6 +54,7 @@
  *	  v4wb  - ARMv4 with write buffer without I TLB flush entry instruction
  *	  v4wbi - ARMv4 with write buffer with I TLB flush entry instruction
  *	  fr    - Feroceon (v4wbi with non-outer-cacheable page table walks)
+ *	  fa    - Faraday (v4 with write buffer with UTLB and branch target buffer (BTB))
  *	  v6wbi - ARMv6 with write buffer with I TLB flush entry instruction
  *	  v7wbi - identical to v6wbi
  */
@@ -89,6 +91,22 @@
 # define v4_always_flags	(-1UL)
 #endif
 
+#define fa_tlb_flags	(TLB_WB | TLB_BTB | TLB_DCLEAN | \
+			 TLB_V4_U_FULL | TLB_V4_U_PAGE)
+
+#ifdef CONFIG_CPU_TLB_FA
+# define fa_possible_flags	fa_tlb_flags
+# define fa_always_flags	fa_tlb_flags
+# ifdef _TLB
+#  define MULTI_TLB 1
+# else
+#  define _TLB fa
+# endif
+#else
+# define fa_possible_flags	0
+# define fa_always_flags	(-1UL)
+#endif
+
 #define v4wbi_tlb_flags	(TLB_WB | TLB_DCLEAN | \
 			 TLB_V4_I_FULL | TLB_V4_D_FULL | \
 			 TLB_V4_I_PAGE | TLB_V4_D_PAGE)
@@ -140,7 +158,7 @@
 # define v4wb_always_flags	(-1UL)
 #endif
 
-#define v6wbi_tlb_flags (TLB_WB | TLB_DCLEAN | \
+#define v6wbi_tlb_flags (TLB_WB | TLB_DCLEAN | TLB_BTB | \
 			 TLB_V6_I_FULL | TLB_V6_D_FULL | \
 			 TLB_V6_I_PAGE | TLB_V6_D_PAGE | \
 			 TLB_V6_I_ASID | TLB_V6_D_ASID)
@@ -267,6 +285,7 @@ extern struct cpu_tlb_fns cpu_tlb;
 				 v4wbi_possible_flags | \
 				 fr_possible_flags | \
 				 v4wb_possible_flags | \
+				 fa_possible_flags | \
 				 v6wbi_possible_flags | \
 				 v7wbi_possible_flags)
 
@@ -275,6 +294,7 @@ extern struct cpu_tlb_fns cpu_tlb;
 				 v4wbi_always_flags & \
 				 fr_always_flags & \
 				 v4wb_always_flags & \
+				 fa_always_flags & \
 				 v6wbi_always_flags & \
 				 v7wbi_always_flags)
 
@@ -297,9 +317,7 @@ static inline void local_flush_tlb_all(void)
 	if (tlb_flag(TLB_V4_I_FULL | TLB_V6_I_FULL))
 		asm("mcr p15, 0, %0, c8, c5, 0" : : "r" (zero) : "cc");
 
-	if (tlb_flag(TLB_V6_I_FULL | TLB_V6_D_FULL |
-		     TLB_V6_I_PAGE | TLB_V6_D_PAGE |
-		     TLB_V6_I_ASID | TLB_V6_D_ASID)) {
+	if (tlb_flag(TLB_BTB)) {
 		/* flush the branch target cache */
 		asm("mcr p15, 0, %0, c7, c5, 6" : : "r" (zero) : "cc");
 		dsb();
@@ -334,9 +352,7 @@ static inline void local_flush_tlb_mm(struct mm_struct *mm)
 	if (tlb_flag(TLB_V6_I_ASID))
 		asm("mcr p15, 0, %0, c8, c5, 2" : : "r" (asid) : "cc");
 
-	if (tlb_flag(TLB_V6_I_FULL | TLB_V6_D_FULL |
-		     TLB_V6_I_PAGE | TLB_V6_D_PAGE |
-		     TLB_V6_I_ASID | TLB_V6_D_ASID)) {
+	if (tlb_flag(TLB_BTB)) {
 		/* flush the branch target cache */
 		asm("mcr p15, 0, %0, c7, c5, 6" : : "r" (zero) : "cc");
 		dsb();
@@ -374,9 +390,7 @@ local_flush_tlb_page(struct vm_area_struct *vma, unsigned long uaddr)
 	if (tlb_flag(TLB_V6_I_PAGE))
 		asm("mcr p15, 0, %0, c8, c5, 1" : : "r" (uaddr) : "cc");
 
-	if (tlb_flag(TLB_V6_I_FULL | TLB_V6_D_FULL |
-		     TLB_V6_I_PAGE | TLB_V6_D_PAGE |
-		     TLB_V6_I_ASID | TLB_V6_D_ASID)) {
+	if (tlb_flag(TLB_BTB)) {
 		/* flush the branch target cache */
 		asm("mcr p15, 0, %0, c7, c5, 6" : : "r" (zero) : "cc");
 		dsb();
@@ -411,9 +425,7 @@ static inline void local_flush_tlb_kernel_page(unsigned long kaddr)
 	if (tlb_flag(TLB_V6_I_PAGE))
 		asm("mcr p15, 0, %0, c8, c5, 1" : : "r" (kaddr) : "cc");
 
-	if (tlb_flag(TLB_V6_I_FULL | TLB_V6_D_FULL |
-		     TLB_V6_I_PAGE | TLB_V6_D_PAGE |
-		     TLB_V6_I_ASID | TLB_V6_D_ASID)) {
+	if (tlb_flag(TLB_BTB)) {
 		/* flush the branch target cache */
 		asm("mcr p15, 0, %0, c7, c5, 6" : : "r" (zero) : "cc");
 		dsb();
diff --git a/arch/arm/mach-gemini/Kconfig b/arch/arm/mach-gemini/Kconfig
new file mode 100644
index 0000000..515b75c
--- /dev/null
+++ b/arch/arm/mach-gemini/Kconfig
@@ -0,0 +1,19 @@
+if ARCH_GEMINI
+
+menu "Cortina Systems Gemini Implementations"
+
+config MACH_RUT100
+	bool "Teltonika RUT100"
+	select GEMINI_MEM_SWAP
+	help
+	  Say Y here if you intend to run this kernel on a
+	  Teltonika 3G Router RUT100.
+
+endmenu
+
+config GEMINI_MEM_SWAP
+	bool "Gemini memory is swapped"
+	help
+	  Say Y here if Gemini memory is swapped by bootloader.
+
+endif
diff --git a/arch/arm/mach-gemini/Makefile b/arch/arm/mach-gemini/Makefile
new file mode 100644
index 0000000..719505b
--- /dev/null
+++ b/arch/arm/mach-gemini/Makefile
@@ -0,0 +1,10 @@
+#
+# Makefile for the linux kernel.
+#
+
+# Object file lists.
+
+obj-y			:= irq.o mm.o time.o devices.o gpio.o
+
+# Board-specific support
+obj-$(CONFIG_MACH_RUT100)	+= board-rut1xx.o
diff --git a/arch/arm/mach-gemini/Makefile.boot b/arch/arm/mach-gemini/Makefile.boot
new file mode 100644
index 0000000..22a52c22
--- /dev/null
+++ b/arch/arm/mach-gemini/Makefile.boot
@@ -0,0 +1,9 @@
+ifeq ($(CONFIG_GEMINI_MEM_SWAP),y)
+   zreladdr-y	:= 0x00008000
+params_phys-y	:= 0x00000100
+initrd_phys-y	:= 0x00800000
+else
+   zreladdr-y	:= 0x10008000
+params_phys-y	:= 0x10000100
+initrd_phys-y	:= 0x10800000
+endif
diff --git a/arch/arm/mach-gemini/board-rut1xx.c b/arch/arm/mach-gemini/board-rut1xx.c
new file mode 100644
index 0000000..e0de968
--- /dev/null
+++ b/arch/arm/mach-gemini/board-rut1xx.c
@@ -0,0 +1,95 @@
+/*
+ *  Support for Teltonika RUT1xx
+ *
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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/init.h>
+#include <linux/platform_device.h>
+#include <linux/leds.h>
+#include <linux/input.h>
+#include <linux/gpio_keys.h>
+
+#include <asm/mach-types.h>
+#include <asm/mach/arch.h>
+#include <asm/mach/time.h>
+
+#include "common.h"
+
+static struct gpio_keys_button rut1xx_keys[] = {
+	{
+		.code		= KEY_SETUP,
+		.gpio		= 60,
+		.active_low	= 1,
+		.desc		= "Reset to defaults",
+		.type		= EV_KEY,
+	},
+};
+
+static struct gpio_keys_platform_data rut1xx_keys_data = {
+	.buttons	= rut1xx_keys,
+	.nbuttons	= ARRAY_SIZE(rut1xx_keys),
+};
+
+static struct platform_device rut1xx_keys_device = {
+	.name	= "gpio-keys",
+	.id	= -1,
+	.dev	= {
+		.platform_data = &rut1xx_keys_data,
+	},
+};
+
+static struct gpio_led rut100_leds[] = {
+	{
+		.name			= "Power",
+		.default_trigger	= "heartbeat",
+		.gpio			= 17,
+	},
+	{
+		.name			= "GSM",
+		.default_trigger	= "default-on",
+		.gpio			= 7,
+		.active_low		= 1,
+	},
+};
+
+static struct gpio_led_platform_data rut100_leds_data = {
+	.num_leds	= ARRAY_SIZE(rut100_leds),
+	.leds		= rut100_leds,
+};
+
+static struct platform_device rut1xx_leds = {
+	.name	= "leds-gpio",
+	.id	= -1,
+	.dev	= {
+		.platform_data = &rut100_leds_data,
+	},
+};
+
+static struct sys_timer rut1xx_timer = {
+	.init	= gemini_timer_init,
+};
+
+static void __init rut1xx_init(void)
+{
+	gemini_gpio_init();
+	platform_register_uart();
+	platform_register_pflash(SZ_8M, NULL, 0);
+	platform_device_register(&rut1xx_leds);
+	platform_device_register(&rut1xx_keys_device);
+}
+
+MACHINE_START(RUT100, "Teltonika RUT100")
+	.phys_io	= 0x7fffc000,
+	.io_pg_offst	= ((0xffffc000) >> 18) & 0xfffc,
+	.boot_params	= 0x100,
+	.map_io		= gemini_map_io,
+	.init_irq	= gemini_init_irq,
+	.timer		= &rut1xx_timer,
+	.init_machine	= rut1xx_init,
+MACHINE_END
diff --git a/arch/arm/mach-gemini/common.h b/arch/arm/mach-gemini/common.h
new file mode 100644
index 0000000..9392834
--- /dev/null
+++ b/arch/arm/mach-gemini/common.h
@@ -0,0 +1,28 @@
+/*
+ * Common Gemini architecture functions
+ *
+ * Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __GEMINI_COMMON_H__
+#define __GEMINI_COMMON_H__
+
+struct mtd_partition;
+
+extern void gemini_map_io(void);
+extern void gemini_init_irq(void);
+extern void gemini_timer_init(void);
+extern void gemini_gpio_init(void);
+
+/* Common platform devices registration functions */
+extern int platform_register_uart(void);
+extern int platform_register_pflash(unsigned int size,
+				    struct mtd_partition *parts,
+				    unsigned int nr_parts);
+
+#endif /* __GEMINI_COMMON_H__ */
diff --git a/arch/arm/mach-gemini/devices.c b/arch/arm/mach-gemini/devices.c
new file mode 100644
index 0000000..6b52525
--- /dev/null
+++ b/arch/arm/mach-gemini/devices.c
@@ -0,0 +1,92 @@
+/*
+ * Common devices definition for Gemini
+ *
+ * Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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/init.h>
+#include <linux/io.h>
+#include <linux/platform_device.h>
+#include <linux/serial_8250.h>
+#include <linux/mtd/physmap.h>
+
+#include <mach/irqs.h>
+#include <mach/hardware.h>
+#include <mach/global_reg.h>
+
+static struct plat_serial8250_port serial_platform_data[] = {
+	{
+		.membase	= (void *)IO_ADDRESS(GEMINI_UART_BASE),
+		.mapbase	= GEMINI_UART_BASE,
+		.irq		= IRQ_UART,
+		.uartclk	= UART_CLK,
+		.regshift	= 2,
+		.iotype		= UPIO_MEM,
+		.type		= PORT_16550A,
+		.flags		= UPF_BOOT_AUTOCONF | UPF_SKIP_TEST | UPF_FIXED_TYPE,
+	},
+	{},
+};
+
+static struct platform_device serial_device = {
+	.name	= "serial8250",
+	.id	= PLAT8250_DEV_PLATFORM,
+	.dev	= {
+		.platform_data = serial_platform_data,
+	},
+};
+
+int platform_register_uart(void)
+{
+	return platform_device_register(&serial_device);
+}
+
+static struct resource flash_resource = {
+	.start	= GEMINI_FLASH_BASE,
+	.flags	= IORESOURCE_MEM,
+};
+
+static struct physmap_flash_data pflash_platform_data = {};
+
+static struct platform_device pflash_device = {
+	.name	= "physmap-flash",
+	.id	= 0,
+	.dev 	= {
+		.platform_data = &pflash_platform_data,
+	},
+	.resource = &flash_resource,
+	.num_resources = 1,
+};
+
+int platform_register_pflash(unsigned int size, struct mtd_partition *parts,
+			     unsigned int nr_parts)
+{
+	unsigned int reg;
+
+	reg = __raw_readl(IO_ADDRESS(GEMINI_GLOBAL_BASE) + GLOBAL_STATUS);
+
+	if ((reg & FLASH_TYPE_MASK) != FLASH_TYPE_PARALLEL)
+		return -ENXIO;
+
+	if (reg & FLASH_WIDTH_16BIT)
+		pflash_platform_data.width = 2;
+	else
+		pflash_platform_data.width = 1;
+
+	/* enable parallel flash pins and disable others */
+	reg = __raw_readl(IO_ADDRESS(GEMINI_GLOBAL_BASE) + GLOBAL_MISC_CTRL);
+	reg &= ~PFLASH_PADS_DISABLE;
+	reg |= SFLASH_PADS_DISABLE | NAND_PADS_DISABLE;
+	__raw_writel(reg, IO_ADDRESS(GEMINI_GLOBAL_BASE) + GLOBAL_MISC_CTRL);
+
+	flash_resource.end = flash_resource.start + size - 1;
+
+	pflash_platform_data.parts = parts;
+	pflash_platform_data.nr_parts = nr_parts;
+
+	return platform_device_register(&pflash_device);
+}
diff --git a/arch/arm/mach-gemini/gpio.c b/arch/arm/mach-gemini/gpio.c
new file mode 100644
index 0000000..e726385
--- /dev/null
+++ b/arch/arm/mach-gemini/gpio.c
@@ -0,0 +1,232 @@
+/*
+ * Gemini gpiochip and interrupt routines
+ *
+ * Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * Based on plat-mxc/gpio.c:
+ *  MXC GPIO support. (c) 2008 Daniel Mack <daniel@caiaq.de>
+ *  Copyright 2008 Juergen Beisert, kernel@pengutronix.de
+ *
+ * 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/init.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/gpio.h>
+
+#include <mach/hardware.h>
+#include <mach/irqs.h>
+
+#define GPIO_BASE(x)		IO_ADDRESS(GEMINI_GPIO_BASE(x))
+
+/* GPIO registers definition */
+#define GPIO_DATA_OUT		0x0
+#define GPIO_DATA_IN		0x4
+#define GPIO_DIR		0x8
+#define GPIO_DATA_SET		0x10
+#define GPIO_DATA_CLR		0x14
+#define GPIO_PULL_EN		0x18
+#define GPIO_PULL_TYPE		0x1C
+#define GPIO_INT_EN		0x20
+#define GPIO_INT_STAT		0x24
+#define GPIO_INT_MASK		0x2C
+#define GPIO_INT_CLR		0x30
+#define GPIO_INT_TYPE		0x34
+#define GPIO_INT_BOTH_EDGE	0x38
+#define GPIO_INT_LEVEL		0x3C
+#define GPIO_DEBOUNCE_EN	0x40
+#define GPIO_DEBOUNCE_PRESCALE	0x44
+
+#define GPIO_PORT_NUM		3
+
+static void _set_gpio_irqenable(unsigned int base, unsigned int index,
+				int enable)
+{
+	unsigned int reg;
+
+	reg = __raw_readl(base + GPIO_INT_EN);
+	reg = (reg & (~(1 << index))) | (!!enable << index);
+	__raw_writel(reg, base + GPIO_INT_EN);
+}
+
+static void gpio_ack_irq(unsigned int irq)
+{
+	unsigned int gpio = irq_to_gpio(irq);
+	unsigned int base = GPIO_BASE(gpio / 32);
+
+	__raw_writel(1 << (gpio % 32), base + GPIO_INT_CLR);
+}
+
+static void gpio_mask_irq(unsigned int irq)
+{
+	unsigned int gpio = irq_to_gpio(irq);
+	unsigned int base = GPIO_BASE(gpio / 32);
+
+	_set_gpio_irqenable(base, gpio % 32, 0);
+}
+
+static void gpio_unmask_irq(unsigned int irq)
+{
+	unsigned int gpio = irq_to_gpio(irq);
+	unsigned int base = GPIO_BASE(gpio / 32);
+
+	_set_gpio_irqenable(base, gpio % 32, 1);
+}
+
+static int gpio_set_irq_type(unsigned int irq, unsigned int type)
+{
+	unsigned int gpio = irq_to_gpio(irq);
+	unsigned int gpio_mask = 1 << (gpio % 32);
+	unsigned int base = GPIO_BASE(gpio / 32);
+	unsigned int reg_both, reg_level, reg_type;
+
+	reg_type = __raw_readl(base + GPIO_INT_TYPE);
+	reg_level = __raw_readl(base + GPIO_INT_BOTH_EDGE);
+	reg_both = __raw_readl(base + GPIO_INT_BOTH_EDGE);
+
+	switch (type) {
+	case IRQ_TYPE_EDGE_BOTH:
+		reg_type &= ~gpio_mask;
+		reg_both |= gpio_mask;
+		break;
+	case IRQ_TYPE_EDGE_RISING:
+		reg_type &= ~gpio_mask;
+		reg_both &= ~gpio_mask;
+		reg_level &= ~gpio_mask;
+		break;
+	case IRQ_TYPE_EDGE_FALLING:
+		reg_type &= ~gpio_mask;
+		reg_both &= ~gpio_mask;
+		reg_level |= gpio_mask;
+		break;
+	case IRQ_TYPE_LEVEL_HIGH:
+		reg_type |= gpio_mask;
+		reg_level &= ~gpio_mask;
+		break;
+	case IRQ_TYPE_LEVEL_LOW:
+		reg_type |= gpio_mask;
+		reg_level |= gpio_mask;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	__raw_writel(reg_type, base + GPIO_INT_TYPE);
+	__raw_writel(reg_level, base + GPIO_INT_BOTH_EDGE);
+	__raw_writel(reg_both, base + GPIO_INT_BOTH_EDGE);
+
+	gpio_ack_irq(irq);
+
+	return 0;
+}
+
+static void gpio_irq_handler(unsigned int irq, struct irq_desc *desc)
+{
+	unsigned int gpio_irq_no, irq_stat;
+	unsigned int port = (unsigned int)get_irq_data(irq);
+
+	irq_stat = __raw_readl(GPIO_BASE(port) + GPIO_INT_STAT);
+
+	gpio_irq_no = GPIO_IRQ_BASE + port * 32;
+	for (; irq_stat != 0; irq_stat >>= 1, gpio_irq_no++) {
+
+		if ((irq_stat & 1) == 0)
+			continue;
+
+		BUG_ON(!(irq_desc[gpio_irq_no].handle_irq));
+		irq_desc[gpio_irq_no].handle_irq(gpio_irq_no,
+				&irq_desc[gpio_irq_no]);
+	}
+}
+
+static struct irq_chip gpio_irq_chip = {
+	.name = "GPIO",
+	.ack = gpio_ack_irq,
+	.mask = gpio_mask_irq,
+	.unmask = gpio_unmask_irq,
+	.set_type = gpio_set_irq_type,
+};
+
+static void _set_gpio_direction(struct gpio_chip *chip, unsigned offset,
+				int dir)
+{
+	unsigned int base = GPIO_BASE(offset / 32);
+	unsigned int reg;
+
+	reg = __raw_readl(base + GPIO_DIR);
+	if (dir)
+		reg |= 1 << (offset % 32);
+	else
+		reg &= ~(1 << (offset % 32));
+	__raw_writel(reg, base + GPIO_DIR);
+}
+
+static void gemini_gpio_set(struct gpio_chip *chip, unsigned offset, int value)
+{
+	unsigned int base = GPIO_BASE(offset / 32);
+
+	if (value)
+		__raw_writel(1 << (offset % 32), base + GPIO_DATA_SET);
+	else
+		__raw_writel(1 << (offset % 32), base + GPIO_DATA_CLR);
+}
+
+static int gemini_gpio_get(struct gpio_chip *chip, unsigned offset)
+{
+	unsigned int base = GPIO_BASE(offset / 32);
+
+	return (__raw_readl(base + GPIO_DATA_IN) >> (offset % 32)) & 1;
+}
+
+static int gemini_gpio_direction_input(struct gpio_chip *chip, unsigned offset)
+{
+	_set_gpio_direction(chip, offset, 0);
+	return 0;
+}
+
+static int gemini_gpio_direction_output(struct gpio_chip *chip, unsigned offset,
+					int value)
+{
+	_set_gpio_direction(chip, offset, 1);
+	gemini_gpio_set(chip, offset, value);
+	return 0;
+}
+
+static struct gpio_chip gemini_gpio_chip = {
+	.label			= "Gemini",
+	.direction_input	= gemini_gpio_direction_input,
+	.get			= gemini_gpio_get,
+	.direction_output	= gemini_gpio_direction_output,
+	.set			= gemini_gpio_set,
+	.base			= 0,
+	.ngpio			= GPIO_PORT_NUM * 32,
+};
+
+void __init gemini_gpio_init(void)
+{
+	int i, j;
+
+	for (i = 0; i < GPIO_PORT_NUM; i++) {
+		/* disable, unmask and clear all interrupts */
+		__raw_writel(0x0, GPIO_BASE(i) + GPIO_INT_EN);
+		__raw_writel(0x0, GPIO_BASE(i) + GPIO_INT_MASK);
+		__raw_writel(~0x0, GPIO_BASE(i) + GPIO_INT_CLR);
+
+		for (j = GPIO_IRQ_BASE + i * 32;
+		     j < GPIO_IRQ_BASE + (i + 1) * 32; j++) {
+			set_irq_chip(j, &gpio_irq_chip);
+			set_irq_handler(j, handle_edge_irq);
+			set_irq_flags(j, IRQF_VALID);
+		}
+
+		set_irq_chained_handler(IRQ_GPIO(i), gpio_irq_handler);
+		set_irq_data(IRQ_GPIO(i), (void *)i);
+	}
+
+	BUG_ON(gpiochip_add(&gemini_gpio_chip));
+}
diff --git a/arch/arm/mach-gemini/include/mach/debug-macro.S b/arch/arm/mach-gemini/include/mach/debug-macro.S
new file mode 100644
index 0000000..d04a6ea
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/debug-macro.S
@@ -0,0 +1,23 @@
+/*
+ * Debugging macro include header
+ *
+ *  Copyright (C) 1994-1999 Russell King
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 <mach/hardware.h>
+
+	.macro	addruart,rx
+	mrc	p15, 0, \rx, c1, c0
+	tst	\rx, #1					@ MMU enabled?
+	ldreq	\rx, =GEMINI_UART_BASE			@ physical
+	ldrne	\rx, =IO_ADDRESS(GEMINI_UART_BASE)	@ virtual
+	.endm
+
+#define UART_SHIFT	2
+#define FLOW_CONTROL
+#include <asm/hardware/debug-8250.S>
diff --git a/arch/arm/mach-gemini/include/mach/entry-macro.S b/arch/arm/mach-gemini/include/mach/entry-macro.S
new file mode 100644
index 0000000..1624f91
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/entry-macro.S
@@ -0,0 +1,39 @@
+/*
+ * Low-level IRQ helper macros for Gemini platform.
+ *
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * This file is licensed under  the terms of the GNU General Public
+ * License version 2. This program is licensed "as is" without any
+ * warranty of any kind, whether express or implied.
+ */
+#include <mach/hardware.h>
+
+#define IRQ_STATUS	0x14
+
+	.macro  disable_fiq
+	.endm
+
+	.macro  get_irqnr_preamble, base, tmp
+	.endm
+
+	.macro  arch_ret_to_user, tmp1, tmp2
+	.endm
+
+	.macro  get_irqnr_and_base, irqnr, irqstat, base, tmp
+	ldr     \irqstat, =IO_ADDRESS(GEMINI_INTERRUPT_BASE + IRQ_STATUS)
+	ldr     \irqnr, [\irqstat]
+	cmp     \irqnr, #0
+	beq     2313f
+	mov     \tmp, \irqnr
+	mov     \irqnr, #0
+2312:
+	tst     \tmp, #1
+	bne     2313f
+	add     \irqnr, \irqnr, #1
+	mov     \tmp, \tmp, lsr #1
+	cmp     \irqnr, #31
+	bcc     2312b
+2313:
+	.endm
diff --git a/arch/arm/mach-gemini/include/mach/global_reg.h b/arch/arm/mach-gemini/include/mach/global_reg.h
new file mode 100644
index 0000000..de7ff7e
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/global_reg.h
@@ -0,0 +1,278 @@
+/*
+ *  This file contains the hardware definitions for Gemini.
+ *
+ *  Copyright (C) 2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __MACH_GLOBAL_REG_H
+#define __MACH_GLOBAL_REG_H
+
+/* Global Word ID Register*/
+#define GLOBAL_ID			0x00
+
+#define CHIP_ID(reg)			((reg) >> 8)
+#define CHIP_REVISION(reg)		((reg) & 0xFF)
+
+/* Global Status Register */
+#define GLOBAL_STATUS			0x04
+
+#define CPU_BIG_ENDIAN			(1 << 31)
+#define PLL_OSC_30M			(1 << 30)	/* else 60MHz */
+
+#define OPERATION_MODE_MASK		(0xF << 26)
+#define OPM_IDDQ			(0xF << 26)
+#define OPM_NAND			(0xE << 26)
+#define OPM_RING			(0xD << 26)
+#define OPM_DIRECT_BOOT			(0xC << 26)
+#define OPM_USB1_PHY_TEST		(0xB << 26)
+#define OPM_USB0_PHY_TEST		(0xA << 26)
+#define OPM_SATA1_PHY_TEST		(0x9 << 26)
+#define OPM_SATA0_PHY_TEST		(0x8 << 26)
+#define OPM_ICE_ARM			(0x7 << 26)
+#define OPM_ICE_FARADAY			(0x6 << 26)
+#define OPM_PLL_BYPASS			(0x5 << 26)
+#define OPM_DEBUG			(0x4 << 26)
+#define OPM_BURN_IN			(0x3 << 26)
+#define OPM_MBIST			(0x2 << 26)
+#define OPM_SCAN			(0x1 << 26)
+#define OPM_REAL			(0x0 << 26)
+
+#define FLASH_TYPE_MASK			(0x3 << 24)
+#define FLASH_TYPE_NAND_2K		(0x3 << 24)
+#define FLASH_TYPE_NAND_512		(0x2 << 24)
+#define FLASH_TYPE_PARALLEL		(0x1 << 24)
+#define FLASH_TYPE_SERIAL		(0x0 << 24)
+/* if parallel */
+#define FLASH_WIDTH_16BIT		(1 << 23)	/* else 8 bit */
+/* if serial */
+#define FLASH_ATMEL			(1 << 23)	/* else STM */
+
+#define FLASH_SIZE_MASK			(0x3 << 21)
+#define NAND_256M			(0x3 << 21)	/* and more */
+#define NAND_128M			(0x2 << 21)
+#define NAND_64M			(0x1 << 21)
+#define NAND_32M			(0x0 << 21)
+#define ATMEL_16M			(0x3 << 21)	/* and more */
+#define ATMEL_8M			(0x2 << 21)
+#define ATMEL_4M_2M			(0x1 << 21)
+#define ATMEL_1M			(0x0 << 21)	/* and less */
+#define STM_32M				(1 << 22)	/* and more */
+#define STM_16M				(0 << 22)	/* and less */
+
+#define FLASH_PARALLEL_HIGH_PIN_CNT	(1 << 20)	/* else low pin cnt */
+
+#define CPU_AHB_RATIO_MASK		(0x3 << 18)
+#define CPU_AHB_1_1			(0x0 << 18)
+#define CPU_AHB_3_2			(0x1 << 18)
+#define CPU_AHB_24_13			(0x2 << 18)
+#define CPU_AHB_2_1			(0x3 << 18)
+
+#define REG_TO_AHB_SPEED(reg)		((((reg) >> 15) & 0x7) * 10 + 130)
+#define AHB_SPEED_TO_REG(x)		((((x - 130)) / 10) << 15)
+
+/* it is posible to override some settings, use >> OVERRIDE_xxxx_SHIFT */
+#define OVERRIDE_FLASH_TYPE_SHIFT	16
+#define OVERRIDE_FLASH_WIDTH_SHIFT	16
+#define OVERRIDE_FLASH_SIZE_SHIFT	16
+#define OVERRIDE_CPU_AHB_RATIO_SHIFT	15
+#define OVERRIDE_AHB_SPEED_SHIFT	15
+
+/* Global PLL Control Register */
+#define GLOBAL_PLL_CTRL			0x08
+
+#define PLL_BYPASS			(1 << 31)
+#define PLL_POWER_DOWN			(1 << 8)
+#define PLL_CONTROL_Q			(0x1F << 0)
+
+/* Global Soft Reset Control Register */
+#define GLOBAL_RESET			0x0C
+
+#define RESET_GLOBAL			(1 << 31)
+#define RESET_CPU1			(1 << 30)
+#define RESET_TVE			(1 << 28)
+#define RESET_SATA1			(1 << 27)
+#define RESET_SATA0			(1 << 26)
+#define RESET_CIR			(1 << 25)
+#define RESET_EXT_DEV			(1 << 24)
+#define RESET_WD			(1 << 23)
+#define RESET_GPIO2			(1 << 22)
+#define RESET_GPIO1			(1 << 21)
+#define RESET_GPIO0			(1 << 20)
+#define RESET_SSP			(1 << 19)
+#define RESET_UART			(1 << 18)
+#define RESET_TIMER			(1 << 17)
+#define RESET_RTC			(1 << 16)
+#define RESET_INT1			(1 << 15)
+#define RESET_INT0			(1 << 14)
+#define RESET_LCD			(1 << 13)
+#define RESET_LPC			(1 << 12)
+#define RESET_APB			(1 << 11)
+#define RESET_DMA			(1 << 10)
+#define RESET_USB1			(1 << 9)
+#define RESET_USB0			(1 << 8)
+#define RESET_PCI			(1 << 7)
+#define RESET_GMAC1			(1 << 6)
+#define RESET_GMAC0			(1 << 5)
+#define RESET_SECURITY			(1 << 4)
+#define RESET_RAID			(1 << 3)
+#define RESET_IDE			(1 << 2)
+#define RESET_FLASH			(1 << 1)
+#define RESET_DRAM			(1 << 0)
+
+/* Global IO Pad Driving Capability Control Register */
+#define GLOBAL_IO_DRIVING_CTRL		0x10
+
+#define DRIVING_CURRENT_MASK		0x3
+
+/* here 00-4mA, 01-8mA, 10-12mA, 11-16mA */
+#define GPIO1_PADS_31_28_SHIFT		28
+#define GPIO0_PADS_31_16_SHIFT		26
+#define GPIO0_PADS_15_0_SHIFT		24
+#define PCI_AND_EXT_RESET_PADS_SHIFT	22
+#define IDE_PADS_SHIFT			20
+#define GMAC1_PADS_SHIFT		18
+#define GMAC0_PADS_SHIFT		16
+/* DRAM is not in mA and poorly documented */
+#define DRAM_CLOCK_PADS_SHIFT		8
+#define DRAM_DATA_PADS_SHIFT		4
+#define DRAM_CONTROL_PADS_SHIFT		0
+
+/* Global IO Pad Slew Rate Control Register */
+#define GLOBAL_IO_SLEW_RATE_CTRL	0x14
+
+#define GPIO1_PADS_31_28_SLOW		(1 << 10)
+#define GPIO0_PADS_31_16_SLOW		(1 << 9)
+#define GPIO0_PADS_15_0_SLOW		(1 << 8)
+#define PCI_PADS_SLOW			(1 << 7)
+#define IDE_PADS_SLOW			(1 << 6)
+#define GMAC1_PADS_SLOW			(1 << 5)
+#define GMAC0_PADS_SLOW			(1 << 4)
+#define DRAM_CLOCK_PADS_SLOW		(1 << 1)
+#define DRAM_IO_PADS_SLOW		(1 << 0)
+
+/*
+ * General skew control defines
+ * 16 steps, each step is around 0.2ns
+ */
+#define SKEW_MASK			0xF
+
+/* Global IDE PAD Skew Control Register */
+#define GLOBAL_IDE_SKEW_CTRL		0x18
+
+#define IDE1_HOST_STROBE_DELAY_SHIFT	28
+#define IDE1_DEVICE_STROBE_DELAY_SHIFT	24
+#define IDE1_OUTPUT_IO_SKEW_SHIFT	20
+#define IDE1_INPUT_IO_SKEW_SHIFT	16
+#define IDE0_HOST_STROBE_DELAY_SHIFT	12
+#define IDE0_DEVICE_STROBE_DELAY_SHIFT	8
+#define IDE0_OUTPUT_IO_SKEW_SHIFT	4
+#define IDE0_INPUT_IO_SKEW_SHIFT	0
+
+/* Global GMAC Control Pad Skew Control Register */
+#define GLOBAL_GMAC_CTRL_SKEW_CTRL	0x1C
+
+#define GMAC1_TXC_SKEW_SHIFT		28
+#define GMAC1_TXEN_SKEW_SHIFT		24
+#define GMAC1_RXC_SKEW_SHIFT		20
+#define GMAC1_RXDV_SKEW_SHIFT		16
+#define GMAC0_TXC_SKEW_SHIFT		12
+#define GMAC0_TXEN_SKEW_SHIFT		8
+#define GMAC0_RXC_SKEW_SHIFT		4
+#define GMAC0_RXDV_SKEW_SHIFT		0
+
+/* Global GMAC0 Data PAD Skew Control Register */
+#define GLOBAL_GMAC0_DATA_SKEW_CTRL	0x20
+/* Global GMAC1 Data PAD Skew Control Register */
+#define GLOBAL_GMAC1_DATA_SKEW_CTRL	0x24
+
+#define GMAC_TXD_SKEW_SHIFT(x)		(((x) * 4) + 16)
+#define GMAC_RXD_SKEW_SHIFT(x)		((x) * 4)
+
+/* CPU has two AHB busses. */
+
+/* Global Arbitration0 Control Register */
+#define GLOBAL_ARBITRATION0_CTRL	0x28
+
+#define BOOT_CONTROLLER_HIGH_PRIO	(1 << 3)
+#define DMA_BUS1_HIGH_PRIO		(1 << 2)
+#define CPU0_HIGH_PRIO			(1 << 0)
+
+/* Global Arbitration1 Control Register */
+#define GLOBAL_ARBITRATION1_CTRL	0x2C
+
+#define TVE_HIGH_PRIO			(1 << 9)
+#define PCI_HIGH_PRIO			(1 << 8)
+#define USB1_HIGH_PRIO			(1 << 7)
+#define USB0_HIGH_PRIO			(1 << 6)
+#define GMAC1_HIGH_PRIO			(1 << 5)
+#define GMAC0_HIGH_PRIO			(1 << 4)
+#define SECURITY_HIGH_PRIO		(1 << 3)
+#define RAID_HIGH_PRIO			(1 << 2)
+#define IDE_HIGH_PRIO			(1 << 1)
+#define DMA_BUS2_HIGH_PRIO		(1 << 0)
+
+/* Common bits for both arbitration registers */
+#define BURST_LENGTH_SHIFT		16
+#define BURST_LENGTH_MASK		(0x3F << 16)
+
+/* Miscellaneous Control Register */
+#define GLOBAL_MISC_CTRL		0x30
+
+#define MEMORY_SPACE_SWAP		(1 << 31)
+#define USB1_PLUG_MINIB			(1 << 30) /* else plug is mini-A */
+#define USB0_PLUG_MINIB			(1 << 29)
+#define GMAC_GMII			(1 << 28)
+#define GMAC_1_ENABLE			(1 << 27)
+/* TODO: define ATA/SATA bits */
+#define USB1_VBUS_ON			(1 << 23)
+#define USB0_VBUS_ON			(1 << 22)
+#define APB_CLKOUT_ENABLE		(1 << 21)
+#define TVC_CLKOUT_ENABLE		(1 << 20)
+#define EXT_CLKIN_ENABLE		(1 << 19)
+#define PCI_66MHZ			(1 << 18) /* else 33 MHz */
+#define PCI_CLKOUT_ENABLE		(1 << 17)
+#define LPC_CLKOUT_ENABLE		(1 << 16)
+#define USB1_WAKEUP_ON			(1 << 15)
+#define USB0_WAKEUP_ON			(1 << 14)
+/* TODO: define PCI idle detect bits */
+#define TVC_PADS_ENABLE			(1 << 9)
+#define SSP_PADS_ENABLE			(1 << 8)
+#define LCD_PADS_ENABLE			(1 << 7)
+#define LPC_PADS_ENABLE			(1 << 6)
+#define PCI_PADS_ENABLE			(1 << 5)
+#define IDE_PADS_ENABLE			(1 << 4)
+#define DRAM_PADS_POWER_DOWN		(1 << 3)
+#define NAND_PADS_DISABLE		(1 << 2)
+#define PFLASH_PADS_DISABLE		(1 << 1)
+#define SFLASH_PADS_DISABLE		(1 << 0)
+
+/* Global Clock Control Register */
+#define GLOBAL_CLOCK_CTRL		0x34
+
+#define POWER_STATE_G0			(1 << 31)
+#define POWER_STATE_S1			(1 << 30) /* else it is S3/S4 state */
+#define SECURITY_APB_AHB		(1 << 29)
+/* else Security APB clk will be 0.75xAHB */
+/* TODO: TVC clock divider */
+#define PCI_CLKRUN_ENABLE		(1 << 16)
+#define BOOT_CLK_DISABLE		(1 << 13)
+#define TVC_CLK_DISABLE			(1 << 12)
+#define FLASH_CLK_DISABLE		(1 << 11)
+#define DDR_CLK_DISABLE			(1 << 10)
+#define PCI_CLK_DISABLE			(1 << 9)
+#define IDE_CLK_DISABLE			(1 << 8)
+#define USB1_CLK_DISABLE		(1 << 7)
+#define USB0_CLK_DISABLE		(1 << 6)
+#define SATA1_CLK_DISABLE		(1 << 5)
+#define SATA0_CLK_DISABLE		(1 << 4)
+#define GMAC1_CLK_DISABLE		(1 << 3)
+#define GMAC0_CLK_DISABLE		(1 << 2)
+#define SECURITY_CLK_DISABLE		(1 << 1)
+
+/* TODO: other registers definitions if needed */
+
+#endif /* __MACH_GLOBAL_REG_H */
diff --git a/arch/arm/mach-gemini/include/mach/gpio.h b/arch/arm/mach-gemini/include/mach/gpio.h
new file mode 100644
index 0000000..3bc2c70
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/gpio.h
@@ -0,0 +1,25 @@
+/*
+ * Gemini gpiolib specific defines
+ *
+ * Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __MACH_GPIO_H__
+#define __MACH_GPIO_H__
+
+#include <mach/irqs.h>
+#include <asm-generic/gpio.h>
+
+#define gpio_get_value	__gpio_get_value
+#define gpio_set_value	__gpio_set_value
+#define gpio_cansleep	__gpio_cansleep
+
+#define gpio_to_irq(x)	((x) + GPIO_IRQ_BASE)
+#define irq_to_gpio(x)	((x) - GPIO_IRQ_BASE)
+
+#endif /* __MACH_GPIO_H__ */
diff --git a/arch/arm/mach-gemini/include/mach/hardware.h b/arch/arm/mach-gemini/include/mach/hardware.h
new file mode 100644
index 0000000..de67526
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/hardware.h
@@ -0,0 +1,75 @@
+/*
+ *  This file contains the hardware definitions for Gemini.
+ *
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __MACH_HARDWARE_H
+#define __MACH_HARDWARE_H
+
+/*
+ * Memory Map definitions
+ */
+/* FIXME: Does it really swap SRAM like this? */
+#ifdef CONFIG_GEMINI_MEM_SWAP
+# define GEMINI_DRAM_BASE	0x00000000
+# define GEMINI_SRAM_BASE	0x20000000
+#else
+# define GEMINI_SRAM_BASE	0x00000000
+# define GEMINI_DRAM_BASE	0x10000000
+#endif
+#define GEMINI_FLASH_BASE	0x30000000
+#define GEMINI_GLOBAL_BASE	0x40000000
+#define GEMINI_WAQTCHDOG_BASE	0x41000000
+#define GEMINI_UART_BASE	0x42000000
+#define GEMINI_TIMER_BASE	0x43000000
+#define GEMINI_LCD_BASE		0x44000000
+#define GEMINI_RTC_BASE		0x45000000
+#define GEMINI_SATA_BASE	0x46000000
+#define GEMINI_LPC_HOST_BASE	0x47000000
+#define GEMINI_LPC_IO_BASE	0x47800000
+#define GEMINI_INTERRUPT_BASE	0x48000000
+/* TODO: Different interrupt controlers when SMP
+ * #define GEMINI_INTERRUPT0_BASE	0x48000000
+ * #define GEMINI_INTERRUPT1_BASE	0x49000000
+ */
+#define GEMINI_SSP_CTRL_BASE	0x4A000000
+#define GEMINI_POWER_CTRL_BASE	0x4B000000
+#define GEMINI_CIR_BASE		0x4C000000
+#define GEMINI_GPIO_BASE(x)	(0x4D000000 + (x) * 0x1000000)
+#define GEMINI_PCI_IO_BASE	0x50000000
+#define GEMINI_PCI_MEM_BASE	0x58000000
+#define GEMINI_TOE_BASE		0x60000000
+#define GEMINI_GMAC0_BASE	0x6000A000
+#define GEMINI_GMAC1_BASE	0x6000E000
+#define GEMINI_SECURITY_BASE	0x62000000
+#define GEMINI_IDE0_BASE	0x63000000
+#define GEMINI_IDE1_BASE	0x63400000
+#define GEMINI_RAID_BASE	0x64000000
+#define GEMINI_FLASH_CTRL_BASE	0x65000000
+#define GEMINI_DRAM_CTRL_BASE	0x66000000
+#define GEMINI_GENERAL_DMA_BASE	0x67000000
+#define GEMINI_USB0_BASE	0x68000000
+#define GEMINI_USB1_BASE	0x69000000
+#define GEMINI_BIG_ENDIAN_BASE	0x80000000
+
+#define GEMINI_TIMER1_BASE	GEMINI_TIMER_BASE
+#define GEMINI_TIMER2_BASE	(GEMINI_TIMER_BASE + 0x10)
+#define GEMINI_TIMER3_BASE	(GEMINI_TIMER_BASE + 0x20)
+
+/*
+ * UART Clock when System clk is 150MHz
+ */
+#define UART_CLK	48000000
+
+/*
+ * macro to get at IO space when running virtually
+ */
+#define IO_ADDRESS(x)	((((x) & 0xFFF00000) >> 4) | ((x) & 0x000FFFFF) | 0xF0000000)
+
+#endif
diff --git a/arch/arm/mach-gemini/include/mach/io.h b/arch/arm/mach-gemini/include/mach/io.h
new file mode 100644
index 0000000..c548056
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/io.h
@@ -0,0 +1,18 @@
+/*
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __MACH_IO_H
+#define __MACH_IO_H
+
+#define IO_SPACE_LIMIT	0xffffffff
+
+#define __io(a)		__typesafe_io(a)
+#define __mem_pci(a)	(a)
+
+#endif /* __MACH_IO_H */
diff --git a/arch/arm/mach-gemini/include/mach/irqs.h b/arch/arm/mach-gemini/include/mach/irqs.h
new file mode 100644
index 0000000..06bc47e
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/irqs.h
@@ -0,0 +1,53 @@
+/*
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __MACH_IRQS_H__
+#define __MACH_IRQS_H__
+
+#define IRQ_SERIRQ1	31
+#define IRQ_SERIRQ0	30
+#define IRQ_PCID	29
+#define IRQ_PCIC	28
+#define IRQ_PCIB	27
+#define IRQ_PWR		26
+#define IRQ_CIR		25
+#define IRQ_GPIO(x)	(22 + (x))
+#define IRQ_SSP		21
+#define IRQ_LPC		20
+#define IRQ_LCD		19
+#define IRQ_UART	18
+#define IRQ_RTC		17
+#define IRQ_TIMER3	16
+#define IRQ_TIMER2	15
+#define IRQ_TIMER1	14
+#define IRQ_FLASH	12
+#define IRQ_USB1	11
+#define IRQ_USB0	10
+#define IRQ_DMA		9
+#define IRQ_PCI		8
+#define IRQ_IPSEC	7
+#define IRQ_RAID	6
+#define IRQ_IDE1	5
+#define IRQ_IDE0	4
+#define IRQ_WATCHDOG	3
+#define IRQ_GMAC1	2
+#define IRQ_GMAC0	1
+#define IRQ_IPI		0
+
+#define NORMAL_IRQ_NUM	32
+
+#define GPIO_IRQ_BASE	NORMAL_IRQ_NUM
+#define GPIO_IRQ_NUM	(3 * 32)
+
+#define ARCH_TIMER_IRQ	IRQ_TIMER2
+
+#define NR_IRQS		(NORMAL_IRQ_NUM + GPIO_IRQ_NUM)
+
+#endif /* __MACH_IRQS_H__ */
diff --git a/arch/arm/mach-gemini/include/mach/memory.h b/arch/arm/mach-gemini/include/mach/memory.h
new file mode 100644
index 0000000..2d14d5b
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/memory.h
@@ -0,0 +1,19 @@
+/*
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __MACH_MEMORY_H
+#define __MACH_MEMORY_H
+
+#ifdef CONFIG_GEMINI_MEM_SWAP
+# define PHYS_OFFSET	UL(0x00000000)
+#else
+# define PHYS_OFFSET	UL(0x10000000)
+#endif
+
+#endif /* __MACH_MEMORY_H */
diff --git a/arch/arm/mach-gemini/include/mach/system.h b/arch/arm/mach-gemini/include/mach/system.h
new file mode 100644
index 0000000..bbbd727
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/system.h
@@ -0,0 +1,37 @@
+/*
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 __MACH_SYSTEM_H
+#define __MACH_SYSTEM_H
+
+#include <linux/io.h>
+#include <mach/hardware.h>
+#include <mach/global_reg.h>
+
+static inline void arch_idle(void)
+{
+	/*
+	 * Because of broken hardware we have to enable interrupts or the CPU
+	 * will never wakeup... Acctualy it is not very good to enable
+	 * interrupts here since scheduler can miss a tick, but there is
+	 * no other way around this. Platforms that needs it for power saving
+	 * should call enable_hlt() in init code, since by default it is
+	 * disabled.
+	 */
+	local_irq_enable();
+	cpu_do_idle();
+}
+
+static inline void arch_reset(char mode)
+{
+	__raw_writel(RESET_GLOBAL | RESET_CPU1,
+		     IO_ADDRESS(GEMINI_GLOBAL_BASE) + GLOBAL_RESET);
+}
+
+#endif /* __MACH_SYSTEM_H */
diff --git a/arch/arm/mach-gemini/include/mach/timex.h b/arch/arm/mach-gemini/include/mach/timex.h
new file mode 100644
index 0000000..dc5690b
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/timex.h
@@ -0,0 +1,13 @@
+/*
+ * Gemini timex specifications
+ *
+ * Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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.
+ */
+
+/* When AHB bus frequency is 150MHz */
+#define CLOCK_TICK_RATE	38000000
diff --git a/arch/arm/mach-gemini/include/mach/uncompress.h b/arch/arm/mach-gemini/include/mach/uncompress.h
new file mode 100644
index 0000000..59c5df7
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/uncompress.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * Based on mach-pxa/include/mach/uncompress.h:
+ * Copyright:	(C) 2001 MontaVista Software Inc.
+ *
+ * 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 __MACH_UNCOMPRESS_H
+#define __MACH_UNCOMPRESS_H
+
+#include <linux/serial_reg.h>
+#include <mach/hardware.h>
+
+static volatile unsigned long *UART = (unsigned long *)GEMINI_UART_BASE;
+
+/*
+ * The following code assumes the serial port has already been
+ * initialized by the bootloader.  If you didn't setup a port in
+ * your bootloader then nothing will appear (which might be desired).
+ */
+static inline void putc(char c)
+{
+	while (!(UART[UART_LSR] & UART_LSR_THRE))
+		barrier();
+	UART[UART_TX] = c;
+}
+
+#define flush() do { } while (0)
+
+/*
+ * nothing to do
+ */
+#define arch_decomp_setup()
+
+#define arch_decomp_wdog()
+
+#endif /* __MACH_UNCOMPRESS_H */
diff --git a/arch/arm/mach-gemini/include/mach/vmalloc.h b/arch/arm/mach-gemini/include/mach/vmalloc.h
new file mode 100644
index 0000000..83e536d
--- /dev/null
+++ b/arch/arm/mach-gemini/include/mach/vmalloc.h
@@ -0,0 +1,10 @@
+/*
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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 VMALLOC_END	0xF0000000
diff --git a/arch/arm/mach-gemini/irq.c b/arch/arm/mach-gemini/irq.c
new file mode 100644
index 0000000..9e613ca
--- /dev/null
+++ b/arch/arm/mach-gemini/irq.c
@@ -0,0 +1,102 @@
+/*
+ *  Interrupt routines for Gemini
+ *
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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/io.h>
+#include <linux/ioport.h>
+#include <linux/stddef.h>
+#include <linux/list.h>
+#include <linux/sched.h>
+#include <asm/irq.h>
+#include <asm/mach/irq.h>
+#include <mach/hardware.h>
+
+#define IRQ_SOURCE(base_addr)	(base_addr + 0x00)
+#define IRQ_MASK(base_addr)	(base_addr + 0x04)
+#define IRQ_CLEAR(base_addr)	(base_addr + 0x08)
+#define IRQ_TMODE(base_addr)	(base_addr + 0x0C)
+#define IRQ_TLEVEL(base_addr)	(base_addr + 0x10)
+#define IRQ_STATUS(base_addr)	(base_addr + 0x14)
+#define FIQ_SOURCE(base_addr)	(base_addr + 0x20)
+#define FIQ_MASK(base_addr)	(base_addr + 0x24)
+#define FIQ_CLEAR(base_addr)	(base_addr + 0x28)
+#define FIQ_TMODE(base_addr)	(base_addr + 0x2C)
+#define FIQ_LEVEL(base_addr)	(base_addr + 0x30)
+#define FIQ_STATUS(base_addr)	(base_addr + 0x34)
+
+static void gemini_ack_irq(unsigned int irq)
+{
+	__raw_writel(1 << irq, IRQ_CLEAR(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+}
+
+static void gemini_mask_irq(unsigned int irq)
+{
+	unsigned int mask;
+
+	mask = __raw_readl(IRQ_MASK(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+	mask &= ~(1 << irq);
+	__raw_writel(mask, IRQ_MASK(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+}
+
+static void gemini_unmask_irq(unsigned int irq)
+{
+	unsigned int mask;
+
+	mask = __raw_readl(IRQ_MASK(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+	mask |= (1 << irq);
+	__raw_writel(mask, IRQ_MASK(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+}
+
+static struct irq_chip gemini_irq_chip = {
+	.name	= "INTC",
+	.ack	= gemini_ack_irq,
+	.mask	= gemini_mask_irq,
+	.unmask	= gemini_unmask_irq,
+};
+
+static struct resource irq_resource = {
+	.name	= "irq_handler",
+	.start	= IO_ADDRESS(GEMINI_INTERRUPT_BASE),
+	.end	= IO_ADDRESS(FIQ_STATUS(GEMINI_INTERRUPT_BASE)) + 4,
+};
+
+void __init gemini_init_irq(void)
+{
+	unsigned int i, mode = 0, level = 0;
+
+	/*
+	 * Disable arch_idle() by default since it is buggy
+	 * For more info see arch/arm/mach-gemini/include/mach/system.h
+	 */
+	disable_hlt();
+
+	request_resource(&iomem_resource, &irq_resource);
+
+	for (i = 0; i < NR_IRQS; i++) {
+		set_irq_chip(i, &gemini_irq_chip);
+		if((i >= IRQ_TIMER1 && i <= IRQ_TIMER3) || (i >= IRQ_SERIRQ0 && i <= IRQ_SERIRQ1)) {
+			set_irq_handler(i, handle_edge_irq);
+			mode |= 1 << i;
+			level |= 1 << i;
+		} else {			
+			set_irq_handler(i, handle_level_irq);
+		}
+		set_irq_flags(i, IRQF_VALID | IRQF_PROBE);
+	}
+
+	/* Disable all interrupts */
+	__raw_writel(0, IRQ_MASK(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+	__raw_writel(0, FIQ_MASK(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+
+	/* Set interrupt mode */
+	__raw_writel(mode, IRQ_TMODE(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+	__raw_writel(level, IRQ_TLEVEL(IO_ADDRESS(GEMINI_INTERRUPT_BASE)));
+}
diff --git a/arch/arm/mach-gemini/mm.c b/arch/arm/mach-gemini/mm.c
new file mode 100644
index 0000000..5194824
--- /dev/null
+++ b/arch/arm/mach-gemini/mm.c
@@ -0,0 +1,82 @@
+/*
+ *  Static mappings for Gemini
+ *
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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/mm.h>
+#include <linux/init.h>
+
+#include <asm/mach/map.h>
+
+#include <mach/hardware.h>
+
+/* Page table mapping for I/O region */
+static struct map_desc gemini_io_desc[] __initdata = {
+	{
+		.virtual	= IO_ADDRESS(GEMINI_GLOBAL_BASE),
+		.pfn		=__phys_to_pfn(GEMINI_GLOBAL_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_UART_BASE),
+		.pfn		= __phys_to_pfn(GEMINI_UART_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_TIMER_BASE),
+		.pfn		= __phys_to_pfn(GEMINI_TIMER_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_INTERRUPT_BASE),
+		.pfn		= __phys_to_pfn(GEMINI_INTERRUPT_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_POWER_CTRL_BASE),
+		.pfn		= __phys_to_pfn(GEMINI_POWER_CTRL_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_GPIO_BASE(0)),
+		.pfn		= __phys_to_pfn(GEMINI_GPIO_BASE(0)),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_GPIO_BASE(1)),
+		.pfn		= __phys_to_pfn(GEMINI_GPIO_BASE(1)),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_GPIO_BASE(2)),
+		.pfn		= __phys_to_pfn(GEMINI_GPIO_BASE(2)),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_FLASH_CTRL_BASE),
+		.pfn		= __phys_to_pfn(GEMINI_FLASH_CTRL_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_DRAM_CTRL_BASE),
+		.pfn		= __phys_to_pfn(GEMINI_DRAM_CTRL_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	}, {
+		.virtual	= IO_ADDRESS(GEMINI_GENERAL_DMA_BASE),
+		.pfn		= __phys_to_pfn(GEMINI_GENERAL_DMA_BASE),
+		.length		= SZ_512K,
+		.type 		= MT_DEVICE,
+	},
+};
+
+void __init gemini_map_io(void)
+{
+	iotable_init(gemini_io_desc, ARRAY_SIZE(gemini_io_desc));
+}
diff --git a/arch/arm/mach-gemini/time.c b/arch/arm/mach-gemini/time.c
new file mode 100644
index 0000000..21dc5a8
--- /dev/null
+++ b/arch/arm/mach-gemini/time.c
@@ -0,0 +1,89 @@
+/*
+ *  Copyright (C) 2001-2006 Storlink, Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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/interrupt.h>
+#include <linux/irq.h>
+#include <linux/io.h>
+#include <mach/hardware.h>
+#include <mach/global_reg.h>
+#include <asm/mach/time.h>
+
+/*
+ * Register definitions for the timers
+ */
+#define TIMER_COUNT(BASE_ADDR)		(BASE_ADDR  + 0x00)
+#define TIMER_LOAD(BASE_ADDR)		(BASE_ADDR  + 0x04)
+#define TIMER_MATCH1(BASE_ADDR)		(BASE_ADDR  + 0x08)
+#define TIMER_MATCH2(BASE_ADDR)		(BASE_ADDR  + 0x0C)
+#define TIMER_CR(BASE_ADDR)		(BASE_ADDR  + 0x30)
+
+#define TIMER_1_CR_ENABLE		(1 << 0)
+#define TIMER_1_CR_CLOCK		(1 << 1)
+#define TIMER_1_CR_INT			(1 << 2)
+#define TIMER_2_CR_ENABLE		(1 << 3)
+#define TIMER_2_CR_CLOCK		(1 << 4)
+#define TIMER_2_CR_INT			(1 << 5)
+#define TIMER_3_CR_ENABLE		(1 << 6)
+#define TIMER_3_CR_CLOCK		(1 << 7)
+#define TIMER_3_CR_INT			(1 << 8)
+
+/*
+ * IRQ handler for the timer
+ */
+static irqreturn_t gemini_timer_interrupt(int irq, void *dev_id)
+{
+	timer_tick();
+
+	return IRQ_HANDLED;
+}
+
+static struct irqaction gemini_timer_irq = {
+	.name		= "Gemini Timer Tick",
+	.flags		= IRQF_DISABLED | IRQF_TIMER,
+	.handler	= gemini_timer_interrupt,
+};
+
+/*
+ * Set up timer interrupt, and return the current time in seconds.
+ */
+void __init gemini_timer_init(void)
+{
+	unsigned int tick_rate, reg_v;
+
+	reg_v = __raw_readl(IO_ADDRESS(GEMINI_GLOBAL_BASE + GLOBAL_STATUS));
+	tick_rate = REG_TO_AHB_SPEED(reg_v) * 1000000;
+
+	printk(KERN_INFO "Bus: %dMHz", tick_rate / 1000000);
+
+	tick_rate /= 6;		/* APB bus run AHB*(1/6) */
+
+	switch(reg_v & CPU_AHB_RATIO_MASK) {
+	case CPU_AHB_1_1:
+		printk(KERN_CONT "(1/1)\n");
+		break;
+	case CPU_AHB_3_2:
+		printk(KERN_CONT "(3/2)\n");
+		break;
+	case CPU_AHB_24_13:
+		printk(KERN_CONT "(24/13)\n");
+		break;
+	case CPU_AHB_2_1:
+		printk(KERN_CONT "(2/1)\n");
+		break;
+	}
+
+	/*
+	 * Make irqs happen for the system timer
+	 */
+	setup_irq(IRQ_TIMER2, &gemini_timer_irq);
+	/* Start the timer */
+	__raw_writel(tick_rate / HZ, TIMER_COUNT(IO_ADDRESS(GEMINI_TIMER2_BASE)));
+	__raw_writel(tick_rate / HZ, TIMER_LOAD(IO_ADDRESS(GEMINI_TIMER2_BASE)));
+	__raw_writel(TIMER_2_CR_ENABLE | TIMER_2_CR_INT, TIMER_CR(IO_ADDRESS(GEMINI_TIMER_BASE)));
+}
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index a6230f7..2097956 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -186,6 +186,24 @@ config CPU_ARM926T
 	  Say Y if you want support for the ARM926T processor.
 	  Otherwise, say N.
 
+# FA526
+config CPU_FA526
+	bool
+	select CPU_32v4
+	select CPU_ABRT_EV4
+	select CPU_PABRT_NOIFAR
+	select CPU_CACHE_VIVT
+	select CPU_CP15_MMU
+	select CPU_CACHE_FA
+	select CPU_COPY_FA if MMU
+	select CPU_TLB_FA if MMU
+	help
+	  The FA526 is a version of the ARMv4 compatible processor with
+	  Branch Target Buffer, Unified TLB and cache line size 16.
+
+	  Say Y if you want support for the FA526 processor.
+	  Otherwise, say N.
+
 # ARM940T
 config CPU_ARM940T
 	bool "Support ARM940T processor" if ARCH_INTEGRATOR
@@ -495,6 +513,9 @@ config CPU_CACHE_VIVT
 config CPU_CACHE_VIPT
 	bool
 
+config CPU_CACHE_FA
+	bool
+
 if MMU
 # The copy-page model
 config CPU_COPY_V3
@@ -509,6 +530,9 @@ config CPU_COPY_V4WB
 config CPU_COPY_FEROCEON
 	bool
 
+config CPU_COPY_FA
+	bool
+
 config CPU_COPY_V6
 	bool
 
@@ -539,6 +563,13 @@ config CPU_TLB_FEROCEON
 	help
 	  Feroceon TLB (v4wbi with non-outer-cachable page table walks).
 
+config CPU_TLB_FA
+	bool
+	help
+	  Faraday ARM FA526 architecture, unified TLB with writeback cache
+	  and invalidate instruction cache entry. Branch target buffer is
+	  also supported.
+
 config CPU_TLB_V6
 	bool
 
@@ -649,7 +680,7 @@ config CPU_DCACHE_SIZE
 
 config CPU_DCACHE_WRITETHROUGH
 	bool "Force write through D-cache"
-	depends on (CPU_ARM740T || CPU_ARM920T || CPU_ARM922T || CPU_ARM925T || CPU_ARM926T || CPU_ARM940T || CPU_ARM946E || CPU_ARM1020) && !CPU_DCACHE_DISABLE
+	depends on (CPU_ARM740T || CPU_ARM920T || CPU_ARM922T || CPU_ARM925T || CPU_ARM926T || CPU_ARM940T || CPU_ARM946E || CPU_ARM1020 || CPU_FA526) && !CPU_DCACHE_DISABLE
 	default y if CPU_ARM925T
 	help
 	  Say Y here to use the data cache in writethrough mode. Unless you
@@ -664,7 +695,7 @@ config CPU_CACHE_ROUND_ROBIN
 
 config CPU_BPREDICT_DISABLE
 	bool "Disable branch prediction"
-	depends on CPU_ARM1020 || CPU_V6 || CPU_MOHAWK || CPU_XSC3 || CPU_V7
+	depends on CPU_ARM1020 || CPU_V6 || CPU_MOHAWK || CPU_XSC3 || CPU_V7 || CPU_FA526
 	help
 	  Say Y here to disable branch prediction.  If unsure, say N.
 
diff --git a/arch/arm/mm/Makefile b/arch/arm/mm/Makefile
index c264683..63e3f6d 100644
--- a/arch/arm/mm/Makefile
+++ b/arch/arm/mm/Makefile
@@ -33,6 +33,7 @@ obj-$(CONFIG_CPU_CACHE_V4WT)	+= cache-v4wt.o
 obj-$(CONFIG_CPU_CACHE_V4WB)	+= cache-v4wb.o
 obj-$(CONFIG_CPU_CACHE_V6)	+= cache-v6.o
 obj-$(CONFIG_CPU_CACHE_V7)	+= cache-v7.o
+obj-$(CONFIG_CPU_CACHE_FA)	+= cache-fa.o
 
 obj-$(CONFIG_CPU_COPY_V3)	+= copypage-v3.o
 obj-$(CONFIG_CPU_COPY_V4WT)	+= copypage-v4wt.o
@@ -42,6 +43,7 @@ obj-$(CONFIG_CPU_COPY_V6)	+= copypage-v6.o context.o
 obj-$(CONFIG_CPU_SA1100)	+= copypage-v4mc.o
 obj-$(CONFIG_CPU_XSCALE)	+= copypage-xscale.o
 obj-$(CONFIG_CPU_XSC3)		+= copypage-xsc3.o
+obj-$(CONFIG_CPU_COPY_FA)	+= copypage-fa.o
 
 obj-$(CONFIG_CPU_TLB_V3)	+= tlb-v3.o
 obj-$(CONFIG_CPU_TLB_V4WT)	+= tlb-v4.o
@@ -50,6 +52,7 @@ obj-$(CONFIG_CPU_TLB_V4WBI)	+= tlb-v4wbi.o
 obj-$(CONFIG_CPU_TLB_FEROCEON)	+= tlb-v4wbi.o	# reuse v4wbi TLB functions
 obj-$(CONFIG_CPU_TLB_V6)	+= tlb-v6.o
 obj-$(CONFIG_CPU_TLB_V7)	+= tlb-v7.o
+obj-$(CONFIG_CPU_TLB_FA)	+= tlb-fa.o
 
 obj-$(CONFIG_CPU_ARM610)	+= proc-arm6_7.o
 obj-$(CONFIG_CPU_ARM710)	+= proc-arm6_7.o
@@ -63,6 +66,7 @@ obj-$(CONFIG_CPU_ARM925T)	+= proc-arm925.o
 obj-$(CONFIG_CPU_ARM926T)	+= proc-arm926.o
 obj-$(CONFIG_CPU_ARM940T)	+= proc-arm940.o
 obj-$(CONFIG_CPU_ARM946E)	+= proc-arm946.o
+obj-$(CONFIG_CPU_FA526)		+= proc-fa526.o
 obj-$(CONFIG_CPU_ARM1020)	+= proc-arm1020.o
 obj-$(CONFIG_CPU_ARM1020E)	+= proc-arm1020e.o
 obj-$(CONFIG_CPU_ARM1022)	+= proc-arm1022.o
diff --git a/arch/arm/mm/cache-fa.S b/arch/arm/mm/cache-fa.S
new file mode 100644
index 0000000..b63a8f7
--- /dev/null
+++ b/arch/arm/mm/cache-fa.S
@@ -0,0 +1,220 @@
+/*
+ *  linux/arch/arm/mm/cache-fa.S
+ *
+ *  Copyright (C) 2005 Faraday Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * Based on cache-v4wb.S:
+ *  Copyright (C) 1997-2002 Russell king
+ *
+ * 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.
+ *
+ *  Processors: FA520 FA526 FA626	
+ */
+#include <linux/linkage.h>
+#include <linux/init.h>
+#include <asm/memory.h>
+#include <asm/page.h>
+
+#include "proc-macros.S"
+
+/*
+ * The size of one data cache line.
+ */
+#define CACHE_DLINESIZE	16
+
+/*
+ * The total size of the data cache.
+ */
+#ifdef CONFIG_ARCH_GEMINI
+#define CACHE_DSIZE	8192
+#else
+#define CACHE_DSIZE	16384 
+#endif 
+
+/* FIXME: put optimal value here. Current one is just estimation */
+#define CACHE_DLIMIT	(CACHE_DSIZE * 2)
+
+/*
+ *	flush_user_cache_all()
+ *
+ *	Clean and invalidate all cache entries in a particular address
+ *	space.
+ */
+ENTRY(fa_flush_user_cache_all)
+	/* FALLTHROUGH */
+/*
+ *	flush_kern_cache_all()
+ *
+ *	Clean and invalidate the entire cache.
+ */
+ENTRY(fa_flush_kern_cache_all)
+	mov	ip, #0
+	mov	r2, #VM_EXEC
+__flush_whole_cache:
+	mcr	p15, 0, ip, c7, c14, 0		@ clean/invalidate D cache
+	tst	r2, #VM_EXEC
+	mcrne	p15, 0, ip, c7, c5, 0		@ invalidate I cache
+	mcrne	p15, 0, ip, c7, c5, 6		@ invalidate BTB
+	mcrne	p15, 0, ip, c7, c10, 4		@ drain write buffer
+	mcrne	p15, 0, ip, c7, c5, 4		@ prefetch flush
+	mov	pc, lr
+
+/*
+ *	flush_user_cache_range(start, end, flags)
+ *
+ *	Invalidate a range of cache entries in the specified
+ *	address space.
+ *
+ *	- start - start address (inclusive, page aligned)
+ *	- end	- end address (exclusive, page aligned)
+ *	- flags	- vma_area_struct flags describing address space
+ */
+ENTRY(fa_flush_user_cache_range)
+	mov	ip, #0
+	sub	r3, r1, r0			@ calculate total size
+	cmp	r3, #CACHE_DLIMIT		@ total size >= limit?
+	bhs	__flush_whole_cache		@ flush whole D cache
+
+1:	tst	r2, #VM_EXEC
+	mcrne	p15, 0, r0, c7, c5, 1		@ invalidate I line
+	mcr	p15, 0, r0, c7, c14, 1		@ clean and invalidate D entry
+	add	r0, r0, #CACHE_DLINESIZE
+	cmp	r0, r1
+	blo	1b
+	tst	r2, #VM_EXEC
+	mcrne	p15, 0, ip, c7, c5, 6		@ invalidate BTB
+	mcrne	p15, 0, ip, c7, c10, 4		@ data write barrier
+	mcrne	p15, 0, ip, c7, c5, 4		@ prefetch flush
+	mov	pc, lr
+
+/*
+ *	coherent_kern_range(start, end)
+ *
+ *	Ensure coherency between the Icache and the Dcache in the
+ *	region described by start.  If you have non-snooping
+ *	Harvard caches, you need to implement this function.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ */
+ENTRY(fa_coherent_kern_range)
+	/* fall through */
+
+/*
+ *	coherent_user_range(start, end)
+ *
+ *	Ensure coherency between the Icache and the Dcache in the
+ *	region described by start.  If you have non-snooping
+ *	Harvard caches, you need to implement this function.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ */
+ENTRY(fa_coherent_user_range)
+	bic	r0, r0, #CACHE_DLINESIZE - 1
+1:	mcr	p15, 0, r0, c7, c14, 1		@ clean and invalidate D entry
+	mcr	p15, 0, r0, c7, c5, 1		@ invalidate I entry
+	add	r0, r0, #CACHE_DLINESIZE
+	cmp	r0, r1
+	blo	1b
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c5, 6		@ invalidate BTB
+	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer
+	mcr	p15, 0, r0, c7, c5, 4		@ prefetch flush
+	mov	pc, lr
+
+/*
+ *	flush_kern_dcache_page(kaddr)
+ *
+ *	Ensure that the data held in the page kaddr is written back
+ *	to the page in question.
+ *
+ *	- kaddr   - kernel address (guaranteed to be page aligned)
+ */
+ENTRY(fa_flush_kern_dcache_page)
+	add	r1, r0, #PAGE_SZ
+1:	mcr	p15, 0, r0, c7, c14, 1		@ clean & invalidate D line
+	add	r0, r0, #CACHE_DLINESIZE
+	cmp	r0, r1
+	blo	1b
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c5, 0		@ invalidate I cache
+	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer
+	mov	pc, lr
+
+/*
+ *	dma_inv_range(start, end)
+ *
+ *	Invalidate (discard) the specified virtual address range.
+ *	May not write back any entries.  If 'start' or 'end'
+ *	are not cache line aligned, those lines must be written
+ *	back.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ */
+ENTRY(fa_dma_inv_range)
+	tst	r0, #CACHE_DLINESIZE - 1
+	bic	r0, r0, #CACHE_DLINESIZE - 1
+	mcrne	p15, 0, r0, c7, c14, 1		@ clean & invalidate D entry
+	tst	r1, #CACHE_DLINESIZE - 1
+	bic	r1, r1, #CACHE_DLINESIZE - 1
+	mcrne	p15, 0, r1, c7, c14, 1		@ clean & invalidate D entry
+1:	mcr	p15, 0, r0, c7, c6, 1		@ invalidate D entry
+	add	r0, r0, #CACHE_DLINESIZE
+	cmp	r0, r1
+	blo	1b
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer
+	mov	pc, lr
+
+/*
+ *	dma_clean_range(start, end)
+ *
+ *	Clean (write back) the specified virtual address range.
+ *
+ *	- start  - virtual start address
+ *	- end	 - virtual end address
+ */
+ENTRY(fa_dma_clean_range)
+	bic	r0, r0, #CACHE_DLINESIZE - 1
+1:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
+	add	r0, r0, #CACHE_DLINESIZE
+	cmp	r0, r1
+	blo	1b
+	mov	r0, #0	
+	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer
+	mov	pc, lr
+
+/*
+ *	dma_flush_range(start,end)
+ *	- start   - virtual start address of region
+ *	- end     - virtual end address of region
+ */
+ENTRY(fa_dma_flush_range)
+	bic	r0, r0, #CACHE_DLINESIZE - 1
+1:	mcr	p15, 0, r0, c7, c14, 1		@ clean & invalidate D entry
+	add	r0, r0, #CACHE_DLINESIZE
+	cmp	r0, r1
+	blo	1b
+	mov	r0, #0	
+	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer
+	mov	pc, lr
+
+	__INITDATA
+
+	.type	fa_cache_fns, #object
+ENTRY(fa_cache_fns)
+	.long	fa_flush_kern_cache_all
+	.long	fa_flush_user_cache_all
+	.long	fa_flush_user_cache_range
+	.long	fa_coherent_kern_range
+	.long	fa_coherent_user_range
+	.long	fa_flush_kern_dcache_page
+	.long	fa_dma_inv_range
+	.long	fa_dma_clean_range
+	.long	fa_dma_flush_range
+	.size	fa_cache_fns, . - fa_cache_fns
diff --git a/arch/arm/mm/copypage-fa.c b/arch/arm/mm/copypage-fa.c
new file mode 100644
index 0000000..b2a6008
--- /dev/null
+++ b/arch/arm/mm/copypage-fa.c
@@ -0,0 +1,86 @@
+/*
+ *  linux/arch/arm/lib/copypage-fa.S
+ *
+ *  Copyright (C) 2005 Faraday Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * Based on copypage-v4wb.S:
+ *  Copyright (C) 1995-1999 Russell King
+ *
+ * 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/init.h>
+#include <linux/highmem.h>
+
+/*
+ * Faraday optimised copy_user_page
+ */
+static void __naked
+fa_copy_user_page(void *kto, const void *kfrom)
+{
+	asm("\
+	stmfd	sp!, {r4, lr}			@ 2\n\
+	mov	r2, %0				@ 1\n\
+1:	ldmia	r1!, {r3, r4, ip, lr}		@ 4\n\
+	stmia	r0, {r3, r4, ip, lr}		@ 4\n\
+	mcr	p15, 0, r0, c7, c14, 1		@ 1   clean and invalidate D line\n\
+	add	r0, r0, #16			@ 1\n\
+	ldmia	r1!, {r3, r4, ip, lr}		@ 4\n\
+	stmia	r0, {r3, r4, ip, lr}		@ 4\n\
+	mcr	p15, 0, r0, c7, c14, 1		@ 1   clean and invalidate D line\n\
+	add	r0, r0, #16			@ 1\n\
+	subs	r2, r2, #1			@ 1\n\
+	bne	1b				@ 1\n\
+	mcr	p15, 0, r2, c7, c10, 4		@ 1   drain WB\n\
+	ldmfd	sp!, {r4, pc}			@ 3"
+	:
+	: "I" (PAGE_SIZE / 32));
+}
+
+void fa_copy_user_highpage(struct page *to, struct page *from,
+	unsigned long vaddr)
+{
+	void *kto, *kfrom;
+
+	kto = kmap_atomic(to, KM_USER0);
+	kfrom = kmap_atomic(from, KM_USER1);
+	fa_copy_user_page(kto, kfrom);
+	kunmap_atomic(kfrom, KM_USER1);
+	kunmap_atomic(kto, KM_USER0);
+}
+
+/*
+ * Faraday optimised clear_user_page
+ *
+ * Same story as above.
+ */
+void fa_clear_user_highpage(struct page *page, unsigned long vaddr)
+{
+	void *ptr, *kaddr = kmap_atomic(page, KM_USER0);
+	asm volatile("\
+	mov	r1, %2				@ 1\n\
+	mov	r2, #0				@ 1\n\
+	mov	r3, #0				@ 1\n\
+	mov	ip, #0				@ 1\n\
+	mov	lr, #0				@ 1\n\
+1:	stmia	%0, {r2, r3, ip, lr}		@ 4\n\
+	mcr	p15, 0, %0, c7, c14, 1		@ 1   clean and invalidate D line\n\
+	add	%0, %0, #16			@ 1\n\
+	stmia	%0, {r2, r3, ip, lr}		@ 4\n\
+	mcr	p15, 0, %0, c7, c14, 1		@ 1   clean and invalidate D line\n\
+	add	%0, %0, #16			@ 1\n\
+	subs	r1, r1, #1			@ 1\n\
+	bne	1b				@ 1\n\
+	mcr	p15, 0, r1, c7, c10, 4		@ 1   drain WB"
+	: "=r" (ptr)
+	: "0" (kaddr), "I" (PAGE_SIZE / 32)
+	: "r1", "r2", "r3", "ip", "lr");
+	kunmap_atomic(kaddr, KM_USER0);
+}
+
+struct cpu_user_fns fa_user_fns __initdata = {
+	.cpu_clear_user_highpage = fa_clear_user_highpage,
+	.cpu_copy_user_highpage	= fa_copy_user_highpage,
+};
diff --git a/arch/arm/mm/proc-fa526.S b/arch/arm/mm/proc-fa526.S
new file mode 100644
index 0000000..08b8a95
--- /dev/null
+++ b/arch/arm/mm/proc-fa526.S
@@ -0,0 +1,248 @@
+/*
+ *  linux/arch/arm/mm/proc-fa526.S: MMU functions for FA526
+ *
+ *  Written by : Luke Lee
+ *  Copyright (C) 2005 Faraday Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * 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.
+ *
+ *
+ * These are the low level assembler for performing cache and TLB
+ * functions on the fa526.
+ */
+#include <linux/linkage.h>
+#include <linux/init.h>
+#include <asm/assembler.h>
+#include <asm/hwcap.h>
+#include <asm/pgtable-hwdef.h>
+#include <asm/pgtable.h>
+#include <asm/page.h>
+#include <asm/ptrace.h>
+#include <asm/system.h>
+
+#include "proc-macros.S"
+
+#define CACHE_DLINESIZE	16
+
+	.text
+/*
+ * cpu_fa526_proc_init()
+ */
+ENTRY(cpu_fa526_proc_init)
+	mov	pc, lr
+
+/*
+ * cpu_fa526_proc_fin()
+ */
+ENTRY(cpu_fa526_proc_fin)
+	stmfd	sp!, {lr}
+	mov	ip, #PSR_F_BIT | PSR_I_BIT | SVC_MODE
+	msr	cpsr_c, ip
+	bl	fa_flush_kern_cache_all
+	mrc	p15, 0, r0, c1, c0, 0		@ ctrl register
+	bic	r0, r0, #0x1000			@ ...i............
+	bic	r0, r0, #0x000e			@ ............wca.
+	mcr	p15, 0, r0, c1, c0, 0		@ disable caches
+	nop
+	nop
+	ldmfd	sp!, {pc}
+
+/*
+ * cpu_fa526_reset(loc)
+ *
+ * Perform a soft reset of the system.  Put the CPU into the
+ * same state as it would be if it had been reset, and branch
+ * to what would be the reset vector.
+ *
+ * loc: location to jump to for soft reset
+ */
+	.align	4
+ENTRY(cpu_fa526_reset)
+/* TODO: Use CP8 if possible... */
+	mov	ip, #0
+	mcr	p15, 0, ip, c7, c7, 0		@ invalidate I,D caches
+	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
+#ifdef CONFIG_MMU
+	mcr	p15, 0, ip, c8, c7, 0		@ invalidate I & D TLBs
+#endif
+	mrc	p15, 0, ip, c1, c0, 0		@ ctrl register
+	bic	ip, ip, #0x000f			@ ............wcam
+	bic	ip, ip, #0x1100			@ ...i...s........
+	bic	ip, ip, #0x0800			@ BTB off
+	mcr	p15, 0, ip, c1, c0, 0		@ ctrl register
+	nop
+	nop
+	mov	pc, r0
+
+/*
+ * cpu_fa526_do_idle()
+ */
+	.align	4
+ENTRY(cpu_fa526_do_idle)
+	mcr	p15, 0, r0, c7, c0, 4		@ Wait for interrupt
+	mov	pc, lr
+
+
+ENTRY(cpu_fa526_dcache_clean_area)
+1:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
+	add	r0, r0, #CACHE_DLINESIZE
+	subs	r1, r1, #CACHE_DLINESIZE
+	bhi	1b
+	mcr	p15, 0, r0, c7, c10, 4		@ drain WB
+	mov	pc, lr
+
+/* =============================== PageTable ============================== */
+
+/*
+ * cpu_fa526_switch_mm(pgd)
+ *
+ * Set the translation base pointer to be as described by pgd.
+ *
+ * pgd: new page tables
+ */
+	.align	4
+ENTRY(cpu_fa526_switch_mm)
+#ifdef CONFIG_MMU
+	mov	ip, #0
+#ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
+	mcr	p15, 0, ip, c7, c6, 0		@ invalidate D cache
+#else
+	mcr	p15, 0, ip, c7, c14, 0		@ clean and invalidate whole D cache
+#endif
+	mcr	p15, 0, ip, c7, c5, 0		@ invalidate I cache
+	mcr	p15, 0, ip, c7, c5, 6		@ invalidate BTB since mm changed
+	mcr	p15, 0, ip, c7, c10, 4		@ data write barrier
+	mcr	p15, 0, ip, c7, c5, 4		@ prefetch flush
+	mcr	p15, 0, r0, c2, c0, 0		@ load page table pointer
+	mcr	p15, 0, ip, c8, c7, 0		@ invalidate UTLB
+#endif
+	mov	pc, lr
+
+/*
+ * cpu_fa526_set_pte_ext(ptep, pte, ext)
+ *
+ * Set a PTE and flush it out
+ */
+	.align	4
+ENTRY(cpu_fa526_set_pte_ext)
+#ifdef CONFIG_MMU
+	armv3_set_pte_ext
+	mov	r0, r0
+	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c10, 4		@ drain WB
+#endif
+	mov	pc, lr
+
+	__INIT
+
+	.type	__fa526_setup, #function
+__fa526_setup:
+	/* On return of this routine, r0 must carry correct flags for CFG register */
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c7		@ invalidate I,D caches on v4
+	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer on v4
+#ifdef CONFIG_MMU
+	mcr	p15, 0, r0, c8, c7		@ invalidate I,D TLBs on v4
+#endif
+	mcr	p15, 0, r0, c7, c5, 5		@ invalidate IScratchpad RAM
+
+	mov	r0, #1
+	mcr	p15, 0, r0, c1, c1, 0		@ turn-on ECR
+
+	mov	r0, #0
+	mcr	p15, 0, r0, c7, c5, 6		@ invalidate BTB All
+	mcr	p15, 0, r0, c7, c10, 4		@ data write barrier
+	mcr	p15, 0, r0, c7, c5, 4		@ prefetch flush
+
+	mov	r0, #0x1f			@ Domains 0, 1 = manager, 2 = client
+	mcr	p15, 0, r0, c3, c0		@ load domain access register
+
+	mrc	p15, 0, r0, c1, c0		@ get control register v4
+	ldr	r5, fa526_cr1_clear
+	bic	r0, r0, r5
+	ldr	r5, fa526_cr1_set
+	orr	r0, r0, r5
+	mov	pc, lr
+	.size	__fa526_setup, . - __fa526_setup
+
+	/*
+	 * .RVI ZFRS BLDP WCAM
+	 * ..11 1001 .111 1101
+	 *
+	 */
+	.type	fa526_cr1_clear, #object
+	.type	fa526_cr1_set, #object
+fa526_cr1_clear:
+	.word	0x3f3f
+fa526_cr1_set:
+	.word	0x397D
+
+	__INITDATA
+
+/*
+ * Purpose : Function pointers used to access above functions - all calls
+ *	     come through these
+ */
+	.type	fa526_processor_functions, #object
+fa526_processor_functions:
+	.word	v4_early_abort
+	.word	pabort_noifar
+	.word	cpu_fa526_proc_init
+	.word	cpu_fa526_proc_fin
+	.word	cpu_fa526_reset
+	.word   cpu_fa526_do_idle
+	.word	cpu_fa526_dcache_clean_area
+	.word	cpu_fa526_switch_mm
+	.word	cpu_fa526_set_pte_ext
+	.size	fa526_processor_functions, . - fa526_processor_functions
+
+	.section ".rodata"
+
+	.type	cpu_arch_name, #object
+cpu_arch_name:
+	.asciz	"armv4"
+	.size	cpu_arch_name, . - cpu_arch_name
+
+	.type	cpu_elf_name, #object
+cpu_elf_name:
+	.asciz	"v4"
+	.size	cpu_elf_name, . - cpu_elf_name
+
+	.type	cpu_fa526_name, #object
+cpu_fa526_name:
+	.asciz	"FA526"
+	.size	cpu_fa526_name, . - cpu_fa526_name
+
+	.align
+
+	.section ".proc.info.init", #alloc, #execinstr
+
+	.type	__fa526_proc_info,#object
+__fa526_proc_info:
+	.long	0x66015261
+	.long	0xff01fff1
+	.long   PMD_TYPE_SECT | \
+		PMD_SECT_BUFFERABLE | \
+		PMD_SECT_CACHEABLE | \
+		PMD_BIT4 | \
+		PMD_SECT_AP_WRITE | \
+		PMD_SECT_AP_READ
+	.long   PMD_TYPE_SECT | \
+		PMD_BIT4 | \
+		PMD_SECT_AP_WRITE | \
+		PMD_SECT_AP_READ
+	b	__fa526_setup
+	.long	cpu_arch_name
+	.long	cpu_elf_name
+	.long	HWCAP_SWP | HWCAP_HALF
+	.long	cpu_fa526_name
+	.long	fa526_processor_functions
+	.long	fa_tlb_fns
+	.long	fa_user_fns
+	.long	fa_cache_fns
+	.size	__fa526_proc_info, . - __fa526_proc_info
diff --git a/arch/arm/mm/tlb-fa.S b/arch/arm/mm/tlb-fa.S
new file mode 100644
index 0000000..9694f1f
--- /dev/null
+++ b/arch/arm/mm/tlb-fa.S
@@ -0,0 +1,75 @@
+/*
+ *  linux/arch/arm/mm/tlb-fa.S
+ *
+ *  Copyright (C) 2005 Faraday Corp.
+ *  Copyright (C) 2008-2009 Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
+ *
+ * Based on tlb-v4wbi.S:
+ *  Copyright (C) 1997-2002 Russell King
+ *
+ * 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.
+ *
+ *  ARM architecture version 4, Faraday variation.
+ *  This assume an unified TLBs, with a write buffer, and branch target buffer (BTB)
+ *
+ *  Processors: FA520 FA526 FA626
+ */
+#include <linux/linkage.h>
+#include <linux/init.h>
+#include <asm/asm-offsets.h>
+#include <asm/tlbflush.h>
+#include "proc-macros.S"
+
+
+/*
+ *	flush_user_tlb_range(start, end, mm)
+ *
+ *	Invalidate a range of TLB entries in the specified address space.
+ *
+ *	- start - range start address
+ *	- end   - range end address
+ *	- mm    - mm_struct describing address space
+ */
+	.align	4
+ENTRY(fa_flush_user_tlb_range)
+	vma_vm_mm ip, r2
+	act_mm	r3				@ get current->active_mm
+	eors	r3, ip, r3			@ == mm ?
+	movne	pc, lr				@ no, we dont do anything
+	mov	r3, #0
+	mcr	p15, 0, r3, c7, c10, 4		@ drain WB
+	bic	r0, r0, #0x0ff
+	bic	r0, r0, #0xf00
+1:	mcr	p15, 0, r0, c8, c7, 1		@ invalidate UTLB entry
+	add	r0, r0, #PAGE_SZ
+	cmp	r0, r1
+	blo	1b
+	mcr	p15, 0, r3, c7, c5, 6		@ invalidate BTB
+	mcr	p15, 0, r3, c7, c10, 4		@ data write barrier
+	mov	pc, lr
+
+
+ENTRY(fa_flush_kern_tlb_range)
+	mov	r3, #0
+	mcr	p15, 0, r3, c7, c10, 4		@ drain WB
+	bic	r0, r0, #0x0ff
+	bic	r0, r0, #0xf00
+1:	mcr	p15, 0, r0, c8, c7, 1		@ invalidate UTLB entry
+	add	r0, r0, #PAGE_SZ
+	cmp	r0, r1
+	blo	1b
+	mcr	p15, 0, r3, c7, c5, 6		@ invalidate BTB
+	mcr	p15, 0, r3, c7, c10, 4		@ data write barrier
+	mcr	p15, 0, r3, c7, c5, 4		@ prefetch flush
+	mov	pc, lr
+
+	__INITDATA
+
+	.type	fa_tlb_fns, #object
+ENTRY(fa_tlb_fns)
+	.long	fa_flush_user_tlb_range
+	.long	fa_flush_kern_tlb_range
+	.long	fa_tlb_flags
+	.size	fa_tlb_fns, . - fa_tlb_fns