diff --git a/arch/sh/Kconfig b/arch/sh/Kconfig
index 80b4f9a743a1..2f5352c06a0e 100644
--- a/arch/sh/Kconfig
+++ b/arch/sh/Kconfig
@@ -767,12 +767,31 @@ config UBC_WAKEUP
 
 	  If unsure, say N.
 
-config CMDLINE_BOOL
-	bool "Default bootloader kernel arguments"
+choice
+	prompt "Kernel command line"
+	optional
+	default CMDLINE_OVERWRITE
+	help
+	  Setting this option allows the kernel command line arguments
+	  to be set.
+
+config CMDLINE_OVERWRITE
+	bool "Overwrite bootloader kernel arguments"
+	help
+	  Given string will overwrite any arguments passed in by
+	  a bootloader.
+
+config CMDLINE_EXTEND
+	bool "Extend bootloader kernel arguments"
+	help
+	  Given string will be concatenated with arguments passed in
+	  by a bootloader.
+
+endchoice
 
 config CMDLINE
-	string "Initial kernel command string"
-	depends on CMDLINE_BOOL
+	string "Kernel command line arguments string"
+	depends on CMDLINE_OVERWRITE || CMDLINE_EXTEND
 	default "console=ttySC1,115200"
 
 endmenu
diff --git a/arch/sh/boot/compressed/head_32.S b/arch/sh/boot/compressed/head_32.S
index 06ac31f3be88..02a30935f0b9 100644
--- a/arch/sh/boot/compressed/head_32.S
+++ b/arch/sh/boot/compressed/head_32.S
@@ -22,7 +22,7 @@ startup:
 	bt	clear_bss
 	sub	r0, r2
 	mov.l	bss_start_addr, r0
-	mov	#0xe0, r1
+	mov	#0xffffffe0, r1
 	and	r1, r0			! align cache line
 	mov.l	text_start_addr, r3
 	mov	r0, r1
diff --git a/arch/sh/drivers/pci/pci.c b/arch/sh/drivers/pci/pci.c
index 9a1c423ad167..c481df639022 100644
--- a/arch/sh/drivers/pci/pci.c
+++ b/arch/sh/drivers/pci/pci.c
@@ -295,6 +295,8 @@ int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
 			       vma->vm_page_prot);
 }
 
+#ifndef CONFIG_GENERIC_IOMAP
+
 static void __iomem *ioport_map_pci(struct pci_dev *dev,
 				    unsigned long port, unsigned int nr)
 {
@@ -346,6 +348,8 @@ void pci_iounmap(struct pci_dev *dev, void __iomem *addr)
 }
 EXPORT_SYMBOL(pci_iounmap);
 
+#endif /* CONFIG_GENERIC_IOMAP */
+
 #ifdef CONFIG_HOTPLUG
 EXPORT_SYMBOL(pcibios_resource_to_bus);
 EXPORT_SYMBOL(pcibios_bus_to_resource);
diff --git a/arch/sh/include/asm/Kbuild b/arch/sh/include/asm/Kbuild
index 43910cdf78a5..e121c30f797d 100644
--- a/arch/sh/include/asm/Kbuild
+++ b/arch/sh/include/asm/Kbuild
@@ -1,6 +1,6 @@
 include include/asm-generic/Kbuild.asm
 
-header-y += cpu-features.h
+header-y += cachectl.h cpu-features.h
 
 unifdef-y += unistd_32.h
 unifdef-y += unistd_64.h
diff --git a/arch/sh/include/asm/cachectl.h b/arch/sh/include/asm/cachectl.h
new file mode 100644
index 000000000000..6ffb4b7a212e
--- /dev/null
+++ b/arch/sh/include/asm/cachectl.h
@@ -0,0 +1,19 @@
+#ifndef _SH_CACHECTL_H
+#define _SH_CACHECTL_H
+
+/* Definitions for the cacheflush system call.  */
+
+#define CACHEFLUSH_D_INVAL	0x1	/* invalidate (without write back) */
+#define CACHEFLUSH_D_WB		0x2	/* write back (without invalidate) */
+#define CACHEFLUSH_D_PURGE	0x3	/* writeback and invalidate */
+
+#define CACHEFLUSH_I		0x4
+
+/*
+ * Options for cacheflush system call
+ */
+#define ICACHE	CACHEFLUSH_I		/* flush instruction cache */
+#define DCACHE	CACHEFLUSH_D_PURGE	/* writeback and flush data cache */
+#define BCACHE	(ICACHE|DCACHE)		/* flush both caches */
+
+#endif /* _SH_CACHECTL_H */
diff --git a/arch/sh/include/asm/entry-macros.S b/arch/sh/include/asm/entry-macros.S
index 64fd0de24daf..cc43a55e1fcf 100644
--- a/arch/sh/include/asm/entry-macros.S
+++ b/arch/sh/include/asm/entry-macros.S
@@ -7,7 +7,7 @@
 	.endm
 
 	.macro	sti
-	mov	#0xf0, r11
+	mov	#0xfffffff0, r11
 	extu.b	r11, r11
 	not	r11, r11
 	stc	sr, r10
diff --git a/arch/sh/include/asm/io.h b/arch/sh/include/asm/io.h
index 25348141674b..5be45ea4dfec 100644
--- a/arch/sh/include/asm/io.h
+++ b/arch/sh/include/asm/io.h
@@ -92,8 +92,12 @@
 
 static inline void ctrl_delay(void)
 {
-#ifdef P2SEG
+#ifdef CONFIG_CPU_SH4
+	__raw_readw(CCN_PVR);
+#elif defined(P2SEG)
 	__raw_readw(P2SEG);
+#else
+#error "Need a dummy address for delay"
 #endif
 }
 
@@ -146,6 +150,7 @@ __BUILD_MEMORY_STRING(q, u64)
 #define readl_relaxed(a)	readl(a)
 #define readq_relaxed(a)	readq(a)
 
+#ifndef CONFIG_GENERIC_IOMAP
 /* Simple MMIO */
 #define ioread8(a)		__raw_readb(a)
 #define ioread16(a)		__raw_readw(a)
@@ -166,6 +171,15 @@ __BUILD_MEMORY_STRING(q, u64)
 #define iowrite8_rep(a, s, c)	__raw_writesb((a), (s), (c))
 #define iowrite16_rep(a, s, c)	__raw_writesw((a), (s), (c))
 #define iowrite32_rep(a, s, c)	__raw_writesl((a), (s), (c))
+#endif
+
+#define mmio_insb(p,d,c)	__raw_readsb(p,d,c)
+#define mmio_insw(p,d,c)	__raw_readsw(p,d,c)
+#define mmio_insl(p,d,c)	__raw_readsl(p,d,c)
+
+#define mmio_outsb(p,s,c)	__raw_writesb(p,s,c)
+#define mmio_outsw(p,s,c)	__raw_writesw(p,s,c)
+#define mmio_outsl(p,s,c)	__raw_writesl(p,s,c)
 
 /* synco on SH-4A, otherwise a nop */
 #define mmiowb()		wmb()
diff --git a/arch/sh/include/asm/unistd_32.h b/arch/sh/include/asm/unistd_32.h
index 61d6ad93d786..925dd40d9d55 100644
--- a/arch/sh/include/asm/unistd_32.h
+++ b/arch/sh/include/asm/unistd_32.h
@@ -132,7 +132,7 @@
 #define __NR_clone		120
 #define __NR_setdomainname	121
 #define __NR_uname		122
-#define __NR_modify_ldt		123
+#define __NR_cacheflush		123
 #define __NR_adjtimex		124
 #define __NR_mprotect		125
 #define __NR_sigprocmask	126
diff --git a/arch/sh/include/asm/unistd_64.h b/arch/sh/include/asm/unistd_64.h
index a751699afda3..2b84bc916bc5 100644
--- a/arch/sh/include/asm/unistd_64.h
+++ b/arch/sh/include/asm/unistd_64.h
@@ -137,7 +137,7 @@
 #define __NR_clone		120
 #define __NR_setdomainname	121
 #define __NR_uname		122
-#define __NR_modify_ldt		123
+#define __NR_cacheflush		123
 #define __NR_adjtimex		124
 #define __NR_mprotect		125
 #define __NR_sigprocmask	126
diff --git a/arch/sh/kernel/cpu/irq/ipr.c b/arch/sh/kernel/cpu/irq/ipr.c
index 808d99a48efb..c1508a90fc6a 100644
--- a/arch/sh/kernel/cpu/irq/ipr.c
+++ b/arch/sh/kernel/cpu/irq/ipr.c
@@ -35,6 +35,7 @@ static void disable_ipr_irq(unsigned int irq)
 	unsigned long addr = get_ipr_desc(irq)->ipr_offsets[p->ipr_idx];
 	/* Set the priority in IPR to 0 */
 	__raw_writew(__raw_readw(addr) & (0xffff ^ (0xf << p->shift)), addr);
+	(void)__raw_readw(addr);	/* Read back to flush write posting */
 }
 
 static void enable_ipr_irq(unsigned int irq)
diff --git a/arch/sh/kernel/cpu/sh3/entry.S b/arch/sh/kernel/cpu/sh3/entry.S
index 8c19e21847d7..9421ec715fd2 100644
--- a/arch/sh/kernel/cpu/sh3/entry.S
+++ b/arch/sh/kernel/cpu/sh3/entry.S
@@ -257,7 +257,7 @@ restore_all:
 	!
 	! Calculate new SR value
 	mov	k3, k2			! original SR value
-	mov	#0xf0, k1
+	mov	#0xfffffff0, k1
 	extu.b	k1, k1
 	not	k1, k1
 	and	k1, k2			! Mask original SR value
diff --git a/arch/sh/kernel/entry-common.S b/arch/sh/kernel/entry-common.S
index 700477601c6f..68d9223b145e 100644
--- a/arch/sh/kernel/entry-common.S
+++ b/arch/sh/kernel/entry-common.S
@@ -98,8 +98,9 @@ need_resched:
 
 	mov	#OFF_SR, r0
 	mov.l	@(r0,r15), r0		! get status register
-	and	#0xf0, r0		! interrupts off (exception path)?
-	cmp/eq	#0xf0, r0
+	shlr	r0
+	and	#(0xf0>>1), r0		! interrupts off (exception path)?
+	cmp/eq	#(0xf0>>1), r0
 	bt	noresched
 	mov.l	3f, r0
 	jsr	@r0			! call preempt_schedule_irq
diff --git a/arch/sh/kernel/io.c b/arch/sh/kernel/io.c
index 4f85fffaa557..4770c241c679 100644
--- a/arch/sh/kernel/io.c
+++ b/arch/sh/kernel/io.c
@@ -1,12 +1,9 @@
 /*
- * linux/arch/sh/kernel/io.c
+ * arch/sh/kernel/io.c - Machine independent I/O functions.
  *
- * Copyright (C) 2000  Stuart Menefy
+ * Copyright (C) 2000 - 2009  Stuart Menefy
  * Copyright (C) 2005  Paul Mundt
  *
- * Provide real functions which expand to whatever the header file defined.
- * Also definitions of machine independent IO functions.
- *
  * This file is subject to the terms and conditions of the GNU General Public
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
@@ -18,33 +15,87 @@
 
 /*
  * Copy data from IO memory space to "real" memory space.
- * This needs to be optimized.
  */
 void memcpy_fromio(void *to, const volatile void __iomem *from, unsigned long count)
 {
-	unsigned char *p = to;
-        while (count) {
-                count--;
-                *p = readb(from);
-                p++;
-                from++;
-        }
+	/*
+	 * Would it be worthwhile doing byte and long transfers first
+	 * to try and get aligned?
+	 */
+#ifdef CONFIG_CPU_SH4
+	if ((count >= 0x20) &&
+	     (((u32)to & 0x1f) == 0) && (((u32)from & 0x3) == 0)) {
+		int tmp2, tmp3, tmp4, tmp5, tmp6;
+
+		__asm__ __volatile__(
+			"1:			\n\t"
+			"mov.l	@%7+, r0	\n\t"
+			"mov.l	@%7+, %2	\n\t"
+			"movca.l r0, @%0	\n\t"
+			"mov.l	@%7+, %3	\n\t"
+			"mov.l	@%7+, %4	\n\t"
+			"mov.l	@%7+, %5	\n\t"
+			"mov.l	@%7+, %6	\n\t"
+			"mov.l	@%7+, r7	\n\t"
+			"mov.l	@%7+, r0	\n\t"
+			"mov.l	%2, @(0x04,%0)	\n\t"
+			"mov	#0x20, %2	\n\t"
+			"mov.l	%3, @(0x08,%0)	\n\t"
+			"sub	%2, %1		\n\t"
+			"mov.l	%4, @(0x0c,%0)	\n\t"
+			"cmp/hi	%1, %2		! T if 32 > count	\n\t"
+			"mov.l	%5, @(0x10,%0)	\n\t"
+			"mov.l	%6, @(0x14,%0)	\n\t"
+			"mov.l	r7, @(0x18,%0)	\n\t"
+			"mov.l	r0, @(0x1c,%0)	\n\t"
+			"bf.s	1b		\n\t"
+			" add	#0x20, %0	\n\t"
+			: "=&r" (to), "=&r" (count),
+			  "=&r" (tmp2), "=&r" (tmp3), "=&r" (tmp4),
+			  "=&r" (tmp5), "=&r" (tmp6), "=&r" (from)
+			: "7"(from), "0" (to), "1" (count)
+			: "r0", "r7", "t", "memory");
+	}
+#endif
+
+	if ((((u32)to | (u32)from) & 0x3) == 0) {
+		for (; count > 3; count -= 4) {
+			*(u32 *)to = *(volatile u32 *)from;
+			to += 4;
+			from += 4;
+		}
+	}
+
+	for (; count > 0; count--) {
+		*(u8 *)to = *(volatile u8 *)from;
+		to++;
+		from++;
+	}
+
+	mb();
 }
 EXPORT_SYMBOL(memcpy_fromio);
 
 /*
  * Copy data from "real" memory space to IO memory space.
- * This needs to be optimized.
  */
 void memcpy_toio(volatile void __iomem *to, const void *from, unsigned long count)
 {
-	const unsigned char *p = from;
-        while (count) {
-                count--;
-                writeb(*p, to);
-                p++;
-                to++;
-        }
+	if ((((u32)to | (u32)from) & 0x3) == 0) {
+		for ( ; count > 3; count -= 4) {
+			*(volatile u32 *)to = *(u32 *)from;
+			to += 4;
+			from += 4;
+		}
+	}
+
+	for (; count > 0; count--) {
+		*(volatile u8 *)to = *(u8 *)from;
+		to++;
+		from++;
+	}
+
+	mb();
 }
 EXPORT_SYMBOL(memcpy_toio);
 
@@ -62,6 +113,8 @@ void memset_io(volatile void __iomem *dst, int c, unsigned long count)
 }
 EXPORT_SYMBOL(memset_io);
 
+#ifndef CONFIG_GENERIC_IOMAP
+
 void __iomem *ioport_map(unsigned long port, unsigned int nr)
 {
 	void __iomem *ret;
@@ -79,3 +132,5 @@ void ioport_unmap(void __iomem *addr)
 	sh_mv.mv_ioport_unmap(addr);
 }
 EXPORT_SYMBOL(ioport_unmap);
+
+#endif /* CONFIG_GENERIC_IOMAP */
diff --git a/arch/sh/kernel/io_generic.c b/arch/sh/kernel/io_generic.c
index 5a7f554d9ca1..4ff507239286 100644
--- a/arch/sh/kernel/io_generic.c
+++ b/arch/sh/kernel/io_generic.c
@@ -73,35 +73,19 @@ u32 generic_inl_p(unsigned long port)
 
 void generic_insb(unsigned long port, void *dst, unsigned long count)
 {
-	volatile u8 *port_addr;
-	u8 *buf = dst;
-
-	port_addr = (volatile u8 __force *)__ioport_map(port, 1);
-	while (count--)
-		*buf++ = *port_addr;
+	__raw_readsb(__ioport_map(port, 1), dst, count);
+	dummy_read();
 }
 
 void generic_insw(unsigned long port, void *dst, unsigned long count)
 {
-	volatile u16 *port_addr;
-	u16 *buf = dst;
-
-	port_addr = (volatile u16 __force *)__ioport_map(port, 2);
-	while (count--)
-		*buf++ = *port_addr;
-
+	__raw_readsw(__ioport_map(port, 2), dst, count);
 	dummy_read();
 }
 
 void generic_insl(unsigned long port, void *dst, unsigned long count)
 {
-	volatile u32 *port_addr;
-	u32 *buf = dst;
-
-	port_addr = (volatile u32 __force *)__ioport_map(port, 4);
-	while (count--)
-		*buf++ = *port_addr;
-
+	__raw_readsl(__ioport_map(port, 4), dst, count);
 	dummy_read();
 }
 
@@ -145,37 +129,19 @@ void generic_outl_p(u32 b, unsigned long port)
  */
 void generic_outsb(unsigned long port, const void *src, unsigned long count)
 {
-	volatile u8 *port_addr;
-	const u8 *buf = src;
-
-	port_addr = (volatile u8 __force *)__ioport_map(port, 1);
-
-	while (count--)
-		*port_addr = *buf++;
+	__raw_writesb(__ioport_map(port, 1), src, count);
+	dummy_read();
 }
 
 void generic_outsw(unsigned long port, const void *src, unsigned long count)
 {
-	volatile u16 *port_addr;
-	const u16 *buf = src;
-
-	port_addr = (volatile u16 __force *)__ioport_map(port, 2);
-
-	while (count--)
-		*port_addr = *buf++;
-
+	__raw_writesw(__ioport_map(port, 2), src, count);
 	dummy_read();
 }
 
 void generic_outsl(unsigned long port, const void *src, unsigned long count)
 {
-	volatile u32 *port_addr;
-	const u32 *buf = src;
-
-	port_addr = (volatile u32 __force *)__ioport_map(port, 4);
-	while (count--)
-		*port_addr = *buf++;
-
+	__raw_writesl(__ioport_map(port, 4), src, count);
 	dummy_read();
 }
 
diff --git a/arch/sh/kernel/irq.c b/arch/sh/kernel/irq.c
index 278c68c60488..d1053392e287 100644
--- a/arch/sh/kernel/irq.c
+++ b/arch/sh/kernel/irq.c
@@ -114,7 +114,7 @@ asmlinkage int do_IRQ(unsigned int irq, struct pt_regs *regs)
 #endif
 
 	irq_enter();
-	irq = irq_demux(intc_evt2irq(irq));
+	irq = irq_demux(evt2irq(irq));
 
 #ifdef CONFIG_IRQSTACKS
 	curctx = (union irq_ctx *)current_thread_info();
diff --git a/arch/sh/kernel/kgdb.c b/arch/sh/kernel/kgdb.c
index 305aad742aec..d29de7864f32 100644
--- a/arch/sh/kernel/kgdb.c
+++ b/arch/sh/kernel/kgdb.c
@@ -195,8 +195,6 @@ void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs)
 	regs->gbr = gdb_regs[GDB_GBR];
 	regs->mach = gdb_regs[GDB_MACH];
 	regs->macl = gdb_regs[GDB_MACL];
-
-	__asm__ __volatile__ ("ldc %0, vbr" : : "r" (gdb_regs[GDB_VBR]));
 }
 
 void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *p)
diff --git a/arch/sh/kernel/process_32.c b/arch/sh/kernel/process_32.c
index 9fee977f176b..0673c4746be3 100644
--- a/arch/sh/kernel/process_32.c
+++ b/arch/sh/kernel/process_32.c
@@ -32,15 +32,35 @@
 #include <asm/ubc.h>
 #include <asm/fpu.h>
 #include <asm/syscalls.h>
+#include <asm/watchdog.h>
 
 int ubc_usercnt = 0;
 
+#ifdef CONFIG_32BIT
+static void watchdog_trigger_immediate(void)
+{
+	sh_wdt_write_cnt(0xFF);
+	sh_wdt_write_csr(0xC2);
+}
+
+void machine_restart(char * __unused)
+{
+	local_irq_disable();
+
+	/* Use watchdog timer to trigger reset */
+	watchdog_trigger_immediate();
+
+	while (1)
+		cpu_sleep();
+}
+#else
 void machine_restart(char * __unused)
 {
 	/* SR.BL=1 and invoke address error to let CPU reset (manual reset) */
 	asm volatile("ldc %0, sr\n\t"
 		     "mov.l @%1, %0" : : "r" (0x10000000), "r" (0x80000001));
 }
+#endif
 
 void machine_halt(void)
 {
diff --git a/arch/sh/kernel/setup.c b/arch/sh/kernel/setup.c
index 212e6bddaeb8..d13bbafb4e1b 100644
--- a/arch/sh/kernel/setup.c
+++ b/arch/sh/kernel/setup.c
@@ -404,10 +404,14 @@ void __init setup_arch(char **cmdline_p)
 	if (!memory_end)
 		memory_end = memory_start + __MEMORY_SIZE;
 
-#ifdef CONFIG_CMDLINE_BOOL
+#ifdef CONFIG_CMDLINE_OVERWRITE
 	strlcpy(command_line, CONFIG_CMDLINE, sizeof(command_line));
 #else
 	strlcpy(command_line, COMMAND_LINE, sizeof(command_line));
+#ifdef CONFIG_CMDLINE_EXTEND
+	strlcat(command_line, " ", sizeof(command_line));
+	strlcat(command_line, CONFIG_CMDLINE, sizeof(command_line));
+#endif
 #endif
 
 	/* Save unparsed command line copy for /proc/cmdline */
diff --git a/arch/sh/kernel/signal_32.c b/arch/sh/kernel/signal_32.c
index b5afbec1db59..6010750c90b4 100644
--- a/arch/sh/kernel/signal_32.c
+++ b/arch/sh/kernel/signal_32.c
@@ -40,6 +40,16 @@ struct fdpic_func_descriptor {
 	unsigned long	GOT;
 };
 
+/*
+ * The following define adds a 64 byte gap between the signal
+ * stack frame and previous contents of the stack.  This allows
+ * frame unwinding in a function epilogue but only if a frame
+ * pointer is used in the function.  This is necessary because
+ * current gcc compilers (<4.3) do not generate unwind info on
+ * SH for function epilogues.
+ */
+#define UNWINDGUARD 64
+
 /*
  * Atomically swap in the new signal mask, and wait for a signal.
  */
@@ -327,7 +337,7 @@ get_sigframe(struct k_sigaction *ka, unsigned long sp, size_t frame_size)
 			sp = current->sas_ss_sp + current->sas_ss_size;
 	}
 
-	return (void __user *)((sp - frame_size) & -8ul);
+	return (void __user *)((sp - (frame_size+UNWINDGUARD)) & -8ul);
 }
 
 /* These symbols are defined with the addresses in the vsyscall page.
diff --git a/arch/sh/kernel/sys_sh.c b/arch/sh/kernel/sys_sh.c
index 90d00e47264d..8aa5d1ceaf14 100644
--- a/arch/sh/kernel/sys_sh.c
+++ b/arch/sh/kernel/sys_sh.c
@@ -25,6 +25,8 @@
 #include <asm/syscalls.h>
 #include <asm/uaccess.h>
 #include <asm/unistd.h>
+#include <asm/cacheflush.h>
+#include <asm/cachectl.h>
 
 static inline long
 do_mmap2(unsigned long addr, unsigned long len, unsigned long prot,
@@ -179,6 +181,47 @@ asmlinkage int sys_ipc(uint call, int first, int second,
 	return -EINVAL;
 }
 
+/* sys_cacheflush -- flush (part of) the processor cache.  */
+asmlinkage int sys_cacheflush(unsigned long addr, unsigned long len, int op)
+{
+	struct vm_area_struct *vma;
+
+	if ((op <= 0) || (op > (CACHEFLUSH_D_PURGE|CACHEFLUSH_I)))
+		return -EINVAL;
+
+	/*
+	 * Verify that the specified address region actually belongs
+	 * to this process.
+	 */
+	if (addr + len < addr)
+		return -EFAULT;
+
+	down_read(&current->mm->mmap_sem);
+	vma = find_vma (current->mm, addr);
+	if (vma == NULL || addr < vma->vm_start || addr + len > vma->vm_end) {
+		up_read(&current->mm->mmap_sem);
+		return -EFAULT;
+	}
+
+	switch (op & CACHEFLUSH_D_PURGE) {
+		case CACHEFLUSH_D_INVAL:
+			__flush_invalidate_region((void *)addr, len);
+			break;
+		case CACHEFLUSH_D_WB:
+			__flush_wback_region((void *)addr, len);
+			break;
+		case CACHEFLUSH_D_PURGE:
+			__flush_purge_region((void *)addr, len);
+			break;
+	}
+
+	if (op & CACHEFLUSH_I)
+		flush_cache_all();
+
+	up_read(&current->mm->mmap_sem);
+	return 0;
+}
+
 asmlinkage int sys_uname(struct old_utsname __user *name)
 {
 	int err;
diff --git a/arch/sh/kernel/syscalls_32.S b/arch/sh/kernel/syscalls_32.S
index f9e21fa2f592..16ba225ede89 100644
--- a/arch/sh/kernel/syscalls_32.S
+++ b/arch/sh/kernel/syscalls_32.S
@@ -139,7 +139,7 @@ ENTRY(sys_call_table)
 	.long sys_clone		/* 120 */
 	.long sys_setdomainname
 	.long sys_newuname
-	.long sys_ni_syscall	/* sys_modify_ldt */
+	.long sys_cacheflush	/* x86: sys_modify_ldt */
 	.long sys_adjtimex
 	.long sys_mprotect		/* 125 */
 	.long sys_sigprocmask
diff --git a/arch/sh/kernel/syscalls_64.S b/arch/sh/kernel/syscalls_64.S
index bf420b616ae0..af6fb7410c21 100644
--- a/arch/sh/kernel/syscalls_64.S
+++ b/arch/sh/kernel/syscalls_64.S
@@ -143,7 +143,7 @@ sys_call_table:
 	.long sys_clone			/* 120 */
 	.long sys_setdomainname
 	.long sys_newuname
-	.long sys_ni_syscall	/* sys_modify_ldt */
+	.long sys_cacheflush	/* x86: sys_modify_ldt */
 	.long sys_adjtimex
 	.long sys_mprotect		/* 125 */
 	.long sys_sigprocmask
diff --git a/arch/sh/kernel/traps_32.c b/arch/sh/kernel/traps_32.c
index 05a04b6df844..c581dc31d92a 100644
--- a/arch/sh/kernel/traps_32.c
+++ b/arch/sh/kernel/traps_32.c
@@ -24,6 +24,7 @@
 #include <linux/kdebug.h>
 #include <linux/kexec.h>
 #include <linux/limits.h>
+#include <linux/proc_fs.h>
 #include <asm/system.h>
 #include <asm/uaccess.h>
 #include <asm/fpu.h>
@@ -44,6 +45,87 @@
 #define TRAP_ILLEGAL_SLOT_INST	13
 #endif
 
+static unsigned long se_user;
+static unsigned long se_sys;
+static unsigned long se_skipped;
+static unsigned long se_half;
+static unsigned long se_word;
+static unsigned long se_dword;
+static unsigned long se_multi;
+/* bitfield: 1: warn 2: fixup 4: signal -> combinations 2|4 && 1|2|4 are not
+   valid! */
+static int se_usermode = 3;
+/* 0: no warning 1: print a warning message */
+static int se_kernmode_warn = 1;
+
+#ifdef CONFIG_PROC_FS
+static const char *se_usermode_action[] = {
+	"ignored",
+	"warn",
+	"fixup",
+	"fixup+warn",
+	"signal",
+	"signal+warn"
+};
+
+static int
+proc_alignment_read(char *page, char **start, off_t off, int count, int *eof,
+		    void *data)
+{
+	char *p = page;
+	int len;
+
+	p += sprintf(p, "User:\t\t%lu\n", se_user);
+	p += sprintf(p, "System:\t\t%lu\n", se_sys);
+	p += sprintf(p, "Skipped:\t%lu\n", se_skipped);
+	p += sprintf(p, "Half:\t\t%lu\n", se_half);
+	p += sprintf(p, "Word:\t\t%lu\n", se_word);
+	p += sprintf(p, "DWord:\t\t%lu\n", se_dword);
+	p += sprintf(p, "Multi:\t\t%lu\n", se_multi);
+	p += sprintf(p, "User faults:\t%i (%s)\n", se_usermode,
+			se_usermode_action[se_usermode]);
+	p += sprintf(p, "Kernel faults:\t%i (fixup%s)\n", se_kernmode_warn,
+			se_kernmode_warn ? "+warn" : "");
+
+	len = (p - page) - off;
+	if (len < 0)
+		len = 0;
+
+	*eof = (len <= count) ? 1 : 0;
+	*start = page + off;
+
+	return len;
+}
+
+static int proc_alignment_write(struct file *file, const char __user *buffer,
+				unsigned long count, void *data)
+{
+	char mode;
+
+	if (count > 0) {
+		if (get_user(mode, buffer))
+			return -EFAULT;
+		if (mode >= '0' && mode <= '5')
+			se_usermode = mode - '0';
+	}
+	return count;
+}
+
+static int proc_alignment_kern_write(struct file *file, const char __user *buffer,
+				     unsigned long count, void *data)
+{
+	char mode;
+
+	if (count > 0) {
+		if (get_user(mode, buffer))
+			return -EFAULT;
+		if (mode >= '0' && mode <= '1')
+			se_kernmode_warn = mode - '0';
+	}
+	return count;
+}
+#endif
+
 static void dump_mem(const char *str, unsigned long bottom, unsigned long top)
 {
 	unsigned long p;
@@ -194,6 +276,13 @@ static int handle_unaligned_ins(insn_size_t instruction, struct pt_regs *regs,
 
 	count = 1<<(instruction&3);
 
+	switch (count) {
+	case 1: se_half  += 1; break;
+	case 2: se_word  += 1; break;
+	case 4: se_dword += 1; break;
+	case 8: se_multi += 1; break; /* ??? */
+	}
+
 	ret = -EFAULT;
 	switch (instruction>>12) {
 	case 0: /* mov.[bwl] to/from memory via r0+rn */
@@ -359,13 +448,6 @@ static inline int handle_delayslot(struct pt_regs *regs,
 #define SH_PC_8BIT_OFFSET(instr) ((((signed char)(instr))*2) + 4)
 #define SH_PC_12BIT_OFFSET(instr) ((((signed short)(instr<<4))>>3) + 4)
 
-/*
- * XXX: SH-2A needs this too, but it needs an overhaul thanks to mixed 32-bit
- * opcodes..
- */
-
-static int handle_unaligned_notify_count = 10;
-
 int handle_unaligned_access(insn_size_t instruction, struct pt_regs *regs,
 			    struct mem_access *ma)
 {
@@ -375,15 +457,13 @@ int handle_unaligned_access(insn_size_t instruction, struct pt_regs *regs,
 	index = (instruction>>8)&15;	/* 0x0F00 */
 	rm = regs->regs[index];
 
-	/* shout about the first ten userspace fixups */
-	if (user_mode(regs) && handle_unaligned_notify_count>0) {
-		handle_unaligned_notify_count--;
-
-		printk(KERN_NOTICE "Fixing up unaligned userspace access "
+	/* shout about fixups */
+	if (printk_ratelimit())
+		printk(KERN_NOTICE "Fixing up unaligned %s access "
 		       "in \"%s\" pid=%d pc=0x%p ins=0x%04hx\n",
+		       user_mode(regs) ? "userspace" : "kernel",
 		       current->comm, task_pid_nr(current),
 		       (void *)regs->pc, instruction);
-	}
 
 	ret = -EFAULT;
 	switch (instruction&0xF000) {
@@ -539,6 +619,36 @@ asmlinkage void do_address_error(struct pt_regs *regs,
 
 		local_irq_enable();
 
+		se_user += 1;
+
+#ifndef CONFIG_CPU_SH2A
+		set_fs(USER_DS);
+		if (copy_from_user(&instruction, (u16 *)(regs->pc & ~1), 2)) {
+			set_fs(oldfs);
+			goto uspace_segv;
+		}
+		set_fs(oldfs);
+
+		/* shout about userspace fixups */
+		if (se_usermode & 1)
+			printk(KERN_NOTICE "Unaligned userspace access "
+			       "in \"%s\" pid=%d pc=0x%p ins=0x%04hx\n",
+			       current->comm, current->pid, (void *)regs->pc,
+			       instruction);
+#endif
+
+		if (se_usermode & 2)
+			goto fixup;
+
+		if (se_usermode & 4)
+			goto uspace_segv;
+		else {
+			/* ignore */
+			regs->pc += instruction_size(instruction);
+			return;
+		}
+
+fixup:
 		/* bad PC is not something we can fix */
 		if (regs->pc & 1) {
 			si_code = BUS_ADRALN;
@@ -546,15 +656,6 @@ asmlinkage void do_address_error(struct pt_regs *regs,
 		}
 
 		set_fs(USER_DS);
-		if (copy_from_user(&instruction, (void __user *)(regs->pc),
-				   sizeof(instruction))) {
-			/* Argh. Fault on the instruction itself.
-			   This should never happen non-SMP
-			*/
-			set_fs(oldfs);
-			goto uspace_segv;
-		}
-
 		tmp = handle_unaligned_access(instruction, regs,
 					      &user_mem_access);
 		set_fs(oldfs);
@@ -572,6 +673,14 @@ uspace_segv:
 		info.si_addr = (void __user *)address;
 		force_sig_info(SIGBUS, &info, current);
 	} else {
+		se_sys += 1;
+
+		if (se_kernmode_warn)
+			printk(KERN_NOTICE "Unaligned kernel access "
+			       "on behalf of \"%s\" pid=%d pc=0x%p ins=0x%04hx\n",
+			       current->comm, current->pid, (void *)regs->pc,
+			       instruction);
+
 		if (regs->pc & 1)
 			die("unaligned program counter", regs, error_code);
 
@@ -881,3 +990,38 @@ void dump_stack(void)
 	show_stack(NULL, NULL);
 }
 EXPORT_SYMBOL(dump_stack);
+
+#ifdef CONFIG_PROC_FS
+/*
+ * This needs to be done after sysctl_init, otherwise sys/ will be
+ * overwritten.  Actually, this shouldn't be in sys/ at all since
+ * it isn't a sysctl, and it doesn't contain sysctl information.
+ * We now locate it in /proc/cpu/alignment instead.
+ */
+static int __init alignment_init(void)
+{
+	struct proc_dir_entry *dir, *res;
+
+	dir = proc_mkdir("cpu", NULL);
+	if (!dir)
+		return -ENOMEM;
+
+	res = create_proc_entry("alignment", S_IWUSR | S_IRUGO, dir);
+	if (!res)
+		return -ENOMEM;
+
+	res->read_proc = proc_alignment_read;
+	res->write_proc = proc_alignment_write;
+
+        res = create_proc_entry("kernel_alignment", S_IWUSR | S_IRUGO, dir);
+        if (!res)
+                return -ENOMEM;
+
+        res->read_proc = proc_alignment_read;
+        res->write_proc = proc_alignment_kern_write;
+
+	return 0;
+}
+
+fs_initcall(alignment_init);
+#endif
diff --git a/arch/sh/lib/clear_page.S b/arch/sh/lib/clear_page.S
index 8342bfbde64c..c92244d4ff9d 100644
--- a/arch/sh/lib/clear_page.S
+++ b/arch/sh/lib/clear_page.S
@@ -57,7 +57,7 @@ ENTRY(clear_page)
 ENTRY(__clear_user)
 	!
 	mov	#0, r0
-	mov	#0xe0, r1	! 0xffffffe0
+	mov	#0xffffffe0, r1
 	!
 	! r4..(r4+31)&~32 	   -------- not aligned	[ Area 0 ]
 	! (r4+31)&~32..(r4+r5)&~32 -------- aligned	[ Area 1 ]
diff --git a/arch/sh/lib/delay.c b/arch/sh/lib/delay.c
index f3ddd2133e6f..faa8f86c0db4 100644
--- a/arch/sh/lib/delay.c
+++ b/arch/sh/lib/delay.c
@@ -21,13 +21,14 @@ void __delay(unsigned long loops)
 
 inline void __const_udelay(unsigned long xloops)
 {
+	xloops *= 4;
 	__asm__("dmulu.l	%0, %2\n\t"
 		"sts	mach, %0"
 		: "=r" (xloops)
 		: "0" (xloops),
-		  "r" (HZ * cpu_data[raw_smp_processor_id()].loops_per_jiffy)
+		  "r" (cpu_data[raw_smp_processor_id()].loops_per_jiffy * (HZ/4))
 		: "macl", "mach");
-	__delay(xloops);
+	__delay(++xloops);
 }
 
 void __udelay(unsigned long usecs)
diff --git a/arch/sh/mm/cache-sh4.c b/arch/sh/mm/cache-sh4.c
index 5cfe08dbb59e..397c1030c7a6 100644
--- a/arch/sh/mm/cache-sh4.c
+++ b/arch/sh/mm/cache-sh4.c
@@ -581,6 +581,31 @@ static void __flush_cache_4096(unsigned long addr, unsigned long phys,
  * Break the 1, 2 and 4 way variants of this out into separate functions to
  * avoid nearly all the overhead of having the conditional stuff in the function
  * bodies (+ the 1 and 2 way cases avoid saving any registers too).
+ *
+ * We want to eliminate unnecessary bus transactions, so this code uses
+ * a non-obvious technique.
+ *
+ * Loop over a cache way sized block of, one cache line at a time. For each
+ * line, use movca.a to cause the current cache line contents to be written
+ * back, but without reading anything from main memory. However this has the
+ * side effect that the cache is now caching that memory location. So follow
+ * this with a cache invalidate to mark the cache line invalid. And do all
+ * this with interrupts disabled, to avoid the cache line being accidently
+ * evicted while it is holding garbage.
+ *
+ * This also breaks in a number of circumstances:
+ * - if there are modifications to the region of memory just above
+ *   empty_zero_page (for example because a breakpoint has been placed
+ *   there), then these can be lost.
+ *
+ *   This is because the the memory address which the cache temporarily
+ *   caches in the above description is empty_zero_page. So the
+ *   movca.l hits the cache (it is assumed that it misses, or at least
+ *   isn't dirty), modifies the line and then invalidates it, losing the
+ *   required change.
+ *
+ * - If caches are disabled or configured in write-through mode, then
+ *   the movca.l writes garbage directly into memory.
  */
 static void __flush_dcache_segment_1way(unsigned long start,
 					unsigned long extent_per_way)
@@ -630,6 +655,25 @@ static void __flush_dcache_segment_1way(unsigned long start,
 	} while (a0 < a0e);
 }
 
+#ifdef CONFIG_CACHE_WRITETHROUGH
+/* This method of cache flushing avoids the problems discussed
+ * in the comment above if writethrough caches are enabled. */
+static void __flush_dcache_segment_2way(unsigned long start,
+					unsigned long extent_per_way)
+{
+	unsigned long array_addr;
+
+	array_addr = CACHE_OC_ADDRESS_ARRAY |
+		(start & cpu_data->dcache.entry_mask);
+
+	while (extent_per_way) {
+		ctrl_outl(0, array_addr);
+		ctrl_outl(0, array_addr + cpu_data->dcache.way_incr);
+		array_addr += cpu_data->dcache.linesz;
+		extent_per_way -= cpu_data->dcache.linesz;
+	}
+}
+#else
 static void __flush_dcache_segment_2way(unsigned long start,
 					unsigned long extent_per_way)
 {
@@ -688,6 +732,7 @@ static void __flush_dcache_segment_2way(unsigned long start,
 		a1 += linesz;
 	} while (a0 < a0e);
 }
+#endif
 
 static void __flush_dcache_segment_4way(unsigned long start,
 					unsigned long extent_per_way)
diff --git a/arch/sh/mm/ioremap_32.c b/arch/sh/mm/ioremap_32.c
index da2f4186f2cd..c3250614e3ae 100644
--- a/arch/sh/mm/ioremap_32.c
+++ b/arch/sh/mm/ioremap_32.c
@@ -57,14 +57,6 @@ void __iomem *__ioremap(unsigned long phys_addr, unsigned long size,
 	if (is_pci_memory_fixed_range(phys_addr, size))
 		return (void __iomem *)phys_addr;
 
-#if !defined(CONFIG_PMB_FIXED)
-	/*
-	 * Don't allow anybody to remap normal RAM that we're using..
-	 */
-	if (phys_addr < virt_to_phys(high_memory))
-		return NULL;
-#endif
-
 	/*
 	 * Mappings have to be page-aligned
 	 */
diff --git a/drivers/sh/intc.c b/drivers/sh/intc.c
index 3dd231a643b5..a9174ec72853 100644
--- a/drivers/sh/intc.c
+++ b/drivers/sh/intc.c
@@ -77,7 +77,7 @@ static unsigned long ack_handle[NR_IRQS];
 static inline struct intc_desc_int *get_intc_desc(unsigned int irq)
 {
 	struct irq_chip *chip = get_irq_chip(irq);
-	return (void *)((char *)chip - offsetof(struct intc_desc_int, chip));
+	return container_of(chip, struct intc_desc_int, chip);
 }
 
 static inline unsigned int set_field(unsigned int value,
@@ -95,16 +95,19 @@ static inline unsigned int set_field(unsigned int value,
 static void write_8(unsigned long addr, unsigned long h, unsigned long data)
 {
 	__raw_writeb(set_field(0, data, h), addr);
+	(void)__raw_readb(addr);	/* Defeat write posting */
 }
 
 static void write_16(unsigned long addr, unsigned long h, unsigned long data)
 {
 	__raw_writew(set_field(0, data, h), addr);
+	(void)__raw_readw(addr);	/* Defeat write posting */
 }
 
 static void write_32(unsigned long addr, unsigned long h, unsigned long data)
 {
 	__raw_writel(set_field(0, data, h), addr);
+	(void)__raw_readl(addr);	/* Defeat write posting */
 }
 
 static void modify_8(unsigned long addr, unsigned long h, unsigned long data)
@@ -112,6 +115,7 @@ static void modify_8(unsigned long addr, unsigned long h, unsigned long data)
 	unsigned long flags;
 	local_irq_save(flags);
 	__raw_writeb(set_field(__raw_readb(addr), data, h), addr);
+	(void)__raw_readb(addr);	/* Defeat write posting */
 	local_irq_restore(flags);
 }
 
@@ -120,6 +124,7 @@ static void modify_16(unsigned long addr, unsigned long h, unsigned long data)
 	unsigned long flags;
 	local_irq_save(flags);
 	__raw_writew(set_field(__raw_readw(addr), data, h), addr);
+	(void)__raw_readw(addr);	/* Defeat write posting */
 	local_irq_restore(flags);
 }
 
@@ -128,6 +133,7 @@ static void modify_32(unsigned long addr, unsigned long h, unsigned long data)
 	unsigned long flags;
 	local_irq_save(flags);
 	__raw_writel(set_field(__raw_readl(addr), data, h), addr);
+	(void)__raw_readl(addr);	/* Defeat write posting */
 	local_irq_restore(flags);
 }
 
@@ -657,16 +663,9 @@ static unsigned int __init save_reg(struct intc_desc_int *d,
 	return 0;
 }
 
-static unsigned char *intc_evt2irq_table;
-
-unsigned int intc_evt2irq(unsigned int vector)
+static void intc_redirect_irq(unsigned int irq, struct irq_desc *desc)
 {
-	unsigned int irq = evt2irq(vector);
-
-	if (intc_evt2irq_table && intc_evt2irq_table[irq])
-		irq = intc_evt2irq_table[irq];
-
-	return irq;
+	generic_handle_irq((unsigned int)get_irq_data(irq));
 }
 
 void __init register_intc_controller(struct intc_desc *desc)
@@ -739,34 +738,6 @@ void __init register_intc_controller(struct intc_desc *desc)
 
 	BUG_ON(k > 256); /* _INTC_ADDR_E() and _INTC_ADDR_D() are 8 bits */
 
-	/* keep the first vector only if same enum is used multiple times */
-	for (i = 0; i < desc->nr_vectors; i++) {
-		struct intc_vect *vect = desc->vectors + i;
-		int first_irq = evt2irq(vect->vect);
-
-		if (!vect->enum_id)
-			continue;
-
-		for (k = i + 1; k < desc->nr_vectors; k++) {
-			struct intc_vect *vect2 = desc->vectors + k;
-
-			if (vect->enum_id != vect2->enum_id)
-				continue;
-
-			vect2->enum_id = 0;
-
-			if (!intc_evt2irq_table)
-				intc_evt2irq_table = kzalloc(NR_IRQS, GFP_NOWAIT);
-
-			if (!intc_evt2irq_table) {
-				pr_warning("intc: cannot allocate evt2irq!\n");
-				continue;
-			}
-
-			intc_evt2irq_table[evt2irq(vect2->vect)] = first_irq;
-		}
-	}
-
 	/* register the vectors one by one */
 	for (i = 0; i < desc->nr_vectors; i++) {
 		struct intc_vect *vect = desc->vectors + i;
@@ -783,6 +754,21 @@ void __init register_intc_controller(struct intc_desc *desc)
 		}
 
 		intc_register_irq(desc, d, vect->enum_id, irq);
+
+		for (k = i + 1; k < desc->nr_vectors; k++) {
+			struct intc_vect *vect2 = desc->vectors + k;
+			unsigned int irq2 = evt2irq(vect2->vect);
+
+			if (vect->enum_id != vect2->enum_id)
+				continue;
+
+			vect2->enum_id = 0;
+
+			/* redirect this interrupts to the first one */
+			set_irq_chip_and_handler_name(irq2, &d->chip,
+					intc_redirect_irq, "redirect");
+			set_irq_data(irq2, (void *)irq);
+		}
 	}
 }
 
diff --git a/include/linux/sh_intc.h b/include/linux/sh_intc.h
index eb1423a0078d..68e212ff9dde 100644
--- a/include/linux/sh_intc.h
+++ b/include/linux/sh_intc.h
@@ -85,7 +85,6 @@ struct intc_desc symbol __initdata = {					\
 }
 #endif
 
-unsigned int intc_evt2irq(unsigned int vector);
 void __init register_intc_controller(struct intc_desc *desc);
 int intc_set_priority(unsigned int irq, unsigned int prio);