Merge master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6:
  sh: Kill off I/O cruft for R7780RP.
  sh: Revert lazy dcache writeback changes.
  sh: Enable SM501 support for RTS7751R2D.
  sh: Use L1_CACHE_BYTES for .data.cacheline_aligned.
  sysctl: Support vdso_enabled sysctl on SH.
  sh: Fix kernel thread stack corruption with preempt.
  doc: Add SH to vdso and earlyprintk in kernel-parameters.txt
  sh: Fix sigmask trampling in signal delivery.
  sh: Clear UBC when not in use.

18 files changed, 353 insertions, 364 deletions

diff --git a/arch/sh/boards/renesas/r7780rp/Makefile b/arch/sh/boards/renesas/r7780rp/Makefile
index 3c93012..ed5f5a9 100644
--- a/arch/sh/boards/renesas/r7780rp/Makefile
+++ b/arch/sh/boards/renesas/r7780rp/Makefile
@@ -2,6 +2,6 @@
 # Makefile for the R7780RP-1 specific parts of the kernel
 #
 
-obj-y	 := setup.o io.o irq.o
+obj-y	 := setup.o irq.o
 
 obj-$(CONFIG_PUSH_SWITCH)	+= psw.o
diff --git a/arch/sh/boards/renesas/r7780rp/io.c b/arch/sh/boards/renesas/r7780rp/io.c
deleted file mode 100644
index f74d2ff..0000000
--- a/arch/sh/boards/renesas/r7780rp/io.c
+++ /dev/null
@@ -1,233 +0,0 @@
-/*
- * Copyright (C) 2001  Ian da Silva, Jeremy Siegel
- * Based largely on io_se.c.
- *
- * I/O routine for Renesas Solutions Highlander R7780RP-1
- *
- * Initial version only to support LAN access; some
- * placeholder code from io_r7780rp.c left in with the
- * expectation of later SuperIO and PCMCIA access.
- */
-#include <linux/pci.h>
-#include <linux/kernel.h>
-#include <linux/types.h>
-#include <linux/io.h>
-#include <asm/r7780rp.h>
-#include <asm/addrspace.h>
-
-static inline unsigned long port88796l(unsigned int port, int flag)
-{
-	unsigned long addr;
-
-	if (flag)
-		addr = PA_AX88796L + ((port - AX88796L_IO_BASE) << 1);
-	else
-		addr = PA_AX88796L + ((port - AX88796L_IO_BASE) << 1) + 0x1000;
-
-	return addr;
-}
-
-#if defined(CONFIG_NE2000) || defined(CONFIG_NE2000_MODULE)
-#define CHECK_AX88796L_PORT(port) \
-  ((port >= AX88796L_IO_BASE) && (port < (AX88796L_IO_BASE+0x20)))
-#else
-#define CHECK_AX88796L_PORT(port) (0)
-#endif
-
-/*
- * General outline: remap really low stuff [eventually] to SuperIO,
- * stuff in PCI IO space (at or above window at pci.h:PCIBIOS_MIN_IO)
- * is mapped through the PCI IO window.  Stuff with high bits (PXSEG)
- * should be way beyond the window, and is used  w/o translation for
- * compatibility.
- */
-u8 r7780rp_inb(unsigned long port)
-{
-	if (CHECK_AX88796L_PORT(port))
-		return ctrl_inw(port88796l(port, 0)) & 0xff;
-	else if (is_pci_ioaddr(port))
-		return ctrl_inb(pci_ioaddr(port));
-
-	return ctrl_inw(port) & 0xff;
-}
-
-u8 r7780rp_inb_p(unsigned long port)
-{
-	u8 v;
-
-	if (CHECK_AX88796L_PORT(port))
-		v = ctrl_inw(port88796l(port, 0)) & 0xff;
-	else if (is_pci_ioaddr(port))
-		v = ctrl_inb(pci_ioaddr(port));
-	else
-		v = ctrl_inw(port) & 0xff;
-
-	ctrl_delay();
-
-	return v;
-}
-
-u16 r7780rp_inw(unsigned long port)
-{
-	if (is_pci_ioaddr(port))
-		return ctrl_inw(pci_ioaddr(port));
-
-	return ctrl_inw(port);
-}
-
-u32 r7780rp_inl(unsigned long port)
-{
-	if (is_pci_ioaddr(port))
-		return ctrl_inl(pci_ioaddr(port));
-
-	return ctrl_inl(port);
-}
-
-void r7780rp_outb(u8 value, unsigned long port)
-{
-	if (CHECK_AX88796L_PORT(port))
-		ctrl_outw(value, port88796l(port, 0));
-	else if (is_pci_ioaddr(port))
-		ctrl_outb(value, pci_ioaddr(port));
-	else
-		ctrl_outb(value, port);
-}
-
-void r7780rp_outb_p(u8 value, unsigned long port)
-{
-	if (CHECK_AX88796L_PORT(port))
-		ctrl_outw(value, port88796l(port, 0));
-	else if (is_pci_ioaddr(port))
-		ctrl_outb(value, pci_ioaddr(port));
-	else
-		ctrl_outb(value, port);
-
-	ctrl_delay();
-}
-
-void r7780rp_outw(u16 value, unsigned long port)
-{
-	if (is_pci_ioaddr(port))
-		ctrl_outw(value, pci_ioaddr(port));
-	else
-		ctrl_outw(value, port);
-}
-
-void r7780rp_outl(u32 value, unsigned long port)
-{
-	if (is_pci_ioaddr(port))
-		ctrl_outl(value, pci_ioaddr(port));
-	else
-		ctrl_outl(value, port);
-}
-
-void r7780rp_insb(unsigned long port, void *dst, unsigned long count)
-{
-	volatile u16 *p;
-	u8 *buf = dst;
-
-	if (CHECK_AX88796L_PORT(port)) {
-		p = (volatile u16 *)port88796l(port, 0);
-		while (count--)
-			*buf++ = *p & 0xff;
-	} else if (is_pci_ioaddr(port)) {
-		volatile u8 *bp = (volatile u8 *)pci_ioaddr(port);
-
-		while (count--)
-			*buf++ = *bp;
-	} else {
-		p = (volatile u16 *)port;
-		while (count--)
-			*buf++ = *p & 0xff;
-	}
-}
-
-void r7780rp_insw(unsigned long port, void *dst, unsigned long count)
-{
-	volatile u16 *p;
-	u16 *buf = dst;
-
-	if (CHECK_AX88796L_PORT(port))
-		p = (volatile u16 *)port88796l(port, 1);
-	else if (is_pci_ioaddr(port))
-		p = (volatile u16 *)pci_ioaddr(port);
-	else
-		p = (volatile u16 *)port;
-
-	while (count--)
-		*buf++ = *p;
-
-	flush_dcache_all();
-}
-
-void r7780rp_insl(unsigned long port, void *dst, unsigned long count)
-{
-	if (is_pci_ioaddr(port)) {
-		volatile u32 *p = (volatile u32 *)pci_ioaddr(port);
-		u32 *buf = dst;
-
-		while (count--)
-			*buf++ = *p;
-	}
-}
-
-void r7780rp_outsb(unsigned long port, const void *src, unsigned long count)
-{
-	volatile u16 *p;
-	const u8 *buf = src;
-
-	if (CHECK_AX88796L_PORT(port)) {
-		p = (volatile u16 *)port88796l(port, 0);
-		while (count--)
-			*p = *buf++;
-	} else if (is_pci_ioaddr(port)) {
-		volatile u8 *bp = (volatile u8 *)pci_ioaddr(port);
-
-		while (count--)
-			*bp = *buf++;
-	} else
-		while (count--)
-			ctrl_outb(*buf++, port);
-}
-
-void r7780rp_outsw(unsigned long port, const void *src, unsigned long count)
-{
-	volatile u16 *p;
-	const u16 *buf = src;
-
-	if (CHECK_AX88796L_PORT(port))
-		p = (volatile u16 *)port88796l(port, 1);
-	else if (is_pci_ioaddr(port))
-		p = (volatile u16 *)pci_ioaddr(port);
-	else
-		p = (volatile u16 *)port;
-
-	while (count--)
-		*p = *buf++;
-
-	flush_dcache_all();
-}
-
-void r7780rp_outsl(unsigned long port, const void *src, unsigned long count)
-{
-	const u32 *buf = src;
-	u32 *p;
-
-	if (is_pci_ioaddr(port))
-		p = (u32 *)pci_ioaddr(port);
-	else
-		p = (u32 *)port;
-
-	while (count--)
-		ctrl_outl(*buf++, (unsigned long)p);
-}
-
-void __iomem *r7780rp_ioport_map(unsigned long port, unsigned int size)
-{
-	if (CHECK_AX88796L_PORT(port))
-		return (void __iomem *)port88796l(port, size > 1);
-	else if (is_pci_ioaddr(port))
-		return (void __iomem *)pci_ioaddr(port);
-
-	return (void __iomem *)port;
-}
diff --git a/arch/sh/boards/renesas/r7780rp/setup.c b/arch/sh/boards/renesas/r7780rp/setup.c
index 0d74db9..2faba66 100644
--- a/arch/sh/boards/renesas/r7780rp/setup.c
+++ b/arch/sh/boards/renesas/r7780rp/setup.c
@@ -187,31 +187,7 @@ static void __init r7780rp_setup(char **cmdline_p)
 struct sh_machine_vector mv_r7780rp __initmv = {
 	.mv_name		= "Highlander R7780RP-1",
 	.mv_setup		= r7780rp_setup,
-
 	.mv_nr_irqs		= 109,
-
-	.mv_inb			= r7780rp_inb,
-	.mv_inw			= r7780rp_inw,
-	.mv_inl			= r7780rp_inl,
-	.mv_outb		= r7780rp_outb,
-	.mv_outw		= r7780rp_outw,
-	.mv_outl		= r7780rp_outl,
-
-	.mv_inb_p		= r7780rp_inb_p,
-	.mv_inw_p		= r7780rp_inw,
-	.mv_inl_p		= r7780rp_inl,
-	.mv_outb_p		= r7780rp_outb_p,
-	.mv_outw_p		= r7780rp_outw,
-	.mv_outl_p		= r7780rp_outl,
-
-	.mv_insb		= r7780rp_insb,
-	.mv_insw		= r7780rp_insw,
-	.mv_insl		= r7780rp_insl,
-	.mv_outsb		= r7780rp_outsb,
-	.mv_outsw		= r7780rp_outsw,
-	.mv_outsl		= r7780rp_outsl,
-
-	.mv_ioport_map		= r7780rp_ioport_map,
 	.mv_init_irq		= init_r7780rp_IRQ,
 };
 ALIAS_MV(r7780rp)
diff --git a/arch/sh/boards/renesas/rts7751r2d/setup.c b/arch/sh/boards/renesas/rts7751r2d/setup.c
index 44b4208..593f26a 100644
--- a/arch/sh/boards/renesas/rts7751r2d/setup.c
+++ b/arch/sh/boards/renesas/rts7751r2d/setup.c
@@ -12,6 +12,7 @@
 #include <linux/platform_device.h>
 #include <linux/pata_platform.h>
 #include <linux/serial_8250.h>
+#include <linux/sm501.h>
 #include <linux/pm.h>
 #include <asm/machvec.h>
 #include <asm/rts7751r2d.h>
@@ -111,10 +112,35 @@ static struct platform_device heartbeat_device = {
 	.resource	= heartbeat_resources,
 };
 
+static struct resource sm501_resources[] = {
+	[0]	= {
+		.start	= 0x10000000,
+		.end	= 0x13e00000 - 1,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1]	= {
+		.start	= 0x13e00000,
+		.end	= 0x13ffffff,
+		.flags	= IORESOURCE_MEM,
+	},
+	[2]	= {
+		.start	= 32,
+		.flags	= IORESOURCE_IRQ,
+	},
+};
+
+static struct platform_device sm501_device = {
+	.name		= "sm501",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(sm501_resources),
+	.resource	= sm501_resources,
+};
+
 static struct platform_device *rts7751r2d_devices[] __initdata = {
 	&uart_device,
 	&heartbeat_device,
 	&cf_ide_device,
+	&sm501_device,
 };
 
 static int __init rts7751r2d_devices_setup(void)
diff --git a/arch/sh/configs/rts7751r2d_defconfig b/arch/sh/configs/rts7751r2d_defconfig
index db6a02d..a59bb78 100644
--- a/arch/sh/configs/rts7751r2d_defconfig
+++ b/arch/sh/configs/rts7751r2d_defconfig
@@ -1,14 +1,13 @@
 #
 # Automatically generated make config: don't edit
-# Linux kernel version: 2.6.20
-# Thu Feb 15 17:17:29 2007
+# Linux kernel version: 2.6.21-rc1
+# Thu Mar  1 16:42:40 2007
 #
 CONFIG_SUPERH=y
 CONFIG_RWSEM_GENERIC_SPINLOCK=y
 CONFIG_GENERIC_FIND_NEXT_BIT=y
 CONFIG_GENERIC_HWEIGHT=y
 CONFIG_GENERIC_HARDIRQS=y
-CONFIG_GENERIC_HARDIRQS_NO__DO_IRQ=y
 CONFIG_GENERIC_IRQ_PROBE=y
 CONFIG_GENERIC_CALIBRATE_DELAY=y
 # CONFIG_GENERIC_TIME is not set
@@ -33,6 +32,7 @@ CONFIG_LOCALVERSION_AUTO=y
 CONFIG_SWAP=y
 CONFIG_SYSVIPC=y
 # CONFIG_IPC_NS is not set
+CONFIG_SYSVIPC_SYSCTL=y
 # CONFIG_POSIX_MQUEUE is not set
 # CONFIG_BSD_PROCESS_ACCT is not set
 # CONFIG_TASKSTATS is not set
@@ -119,7 +119,6 @@ CONFIG_SH_RTS7751R2D=y
 # CONFIG_SH_SHMIN is not set
 # CONFIG_SH_7206_SOLUTION_ENGINE is not set
 # CONFIG_SH_7619_SOLUTION_ENGINE is not set
-# CONFIG_SH_ASDAP310 is not set
 # CONFIG_SH_UNKNOWN is not set
 
 #
@@ -281,7 +280,7 @@ CONFIG_ZERO_PAGE_OFFSET=0x00010000
 CONFIG_BOOT_LINK_OFFSET=0x00800000
 # CONFIG_UBC_WAKEUP is not set
 CONFIG_CMDLINE_BOOL=y
-CONFIG_CMDLINE="console=ttySC0,115200 root=/dev/sda1"
+CONFIG_CMDLINE="console=tty0 console=ttySC0,115200 root=/dev/sda1"
 
 #
 # Bus options
@@ -433,6 +432,7 @@ CONFIG_FW_LOADER=m
 #
 # Plug and Play support
 #
+# CONFIG_PNPACPI is not set
 
 #
 # Block devices
@@ -770,7 +770,26 @@ CONFIG_NET_WIRELESS=y
 #
 # Input device support
 #
-# CONFIG_INPUT is not set
+CONFIG_INPUT=y
+# CONFIG_INPUT_FF_MEMLESS is not set
+
+#
+# Userland interfaces
+#
+# CONFIG_INPUT_MOUSEDEV is not set
+# CONFIG_INPUT_JOYDEV is not set
+# CONFIG_INPUT_TSDEV is not set
+# CONFIG_INPUT_EVDEV is not set
+# CONFIG_INPUT_EVBUG is not set
+
+#
+# Input Device Drivers
+#
+# CONFIG_INPUT_KEYBOARD is not set
+# CONFIG_INPUT_MOUSE is not set
+# CONFIG_INPUT_JOYSTICK is not set
+# CONFIG_INPUT_TOUCHSCREEN is not set
+# CONFIG_INPUT_MISC is not set
 
 #
 # Hardware I/O ports
@@ -781,7 +800,10 @@ CONFIG_NET_WIRELESS=y
 #
 # Character devices
 #
-# CONFIG_VT is not set
+CONFIG_VT=y
+CONFIG_VT_CONSOLE=y
+CONFIG_HW_CONSOLE=y
+CONFIG_VT_HW_CONSOLE_BINDING=y
 # CONFIG_SERIAL_NONSTANDARD is not set
 
 #
@@ -857,6 +879,11 @@ CONFIG_HWMON=y
 # CONFIG_HWMON_DEBUG_CHIP is not set
 
 #
+# Multifunction device drivers
+#
+CONFIG_MFD_SM501=y
+
+#
 # Multimedia devices
 #
 # CONFIG_VIDEO_DEV is not set
@@ -869,9 +896,66 @@ CONFIG_HWMON=y
 #
 # Graphics support
 #
-CONFIG_FIRMWARE_EDID=y
-# CONFIG_FB is not set
 # CONFIG_BACKLIGHT_LCD_SUPPORT is not set
+CONFIG_FB=y
+# CONFIG_FIRMWARE_EDID is not set
+# CONFIG_FB_DDC is not set
+CONFIG_FB_CFB_FILLRECT=y
+CONFIG_FB_CFB_COPYAREA=y
+CONFIG_FB_CFB_IMAGEBLIT=y
+# CONFIG_FB_SVGALIB is not set
+# CONFIG_FB_MACMODES is not set
+# CONFIG_FB_BACKLIGHT is not set
+# CONFIG_FB_MODE_HELPERS is not set
+# CONFIG_FB_TILEBLITTING is not set
+
+#
+# Frambuffer hardware drivers
+#
+# CONFIG_FB_CIRRUS is not set
+# CONFIG_FB_PM2 is not set
+# CONFIG_FB_CYBER2000 is not set
+# CONFIG_FB_ASILIANT is not set
+# CONFIG_FB_IMSTT is not set
+# CONFIG_FB_EPSON1355 is not set
+# CONFIG_FB_S1D13XXX is not set
+# CONFIG_FB_NVIDIA is not set
+# CONFIG_FB_RIVA is not set
+# CONFIG_FB_MATROX is not set
+# CONFIG_FB_RADEON is not set
+# CONFIG_FB_ATY128 is not set
+# CONFIG_FB_ATY is not set
+# CONFIG_FB_S3 is not set
+# CONFIG_FB_SAVAGE is not set
+# CONFIG_FB_SIS is not set
+# CONFIG_FB_NEOMAGIC is not set
+# CONFIG_FB_KYRO is not set
+# CONFIG_FB_3DFX is not set
+# CONFIG_FB_VOODOO1 is not set
+# CONFIG_FB_TRIDENT is not set
+CONFIG_FB_SM501=y
+# CONFIG_FB_VIRTUAL is not set
+
+#
+# Console display driver support
+#
+CONFIG_DUMMY_CONSOLE=y
+CONFIG_FRAMEBUFFER_CONSOLE=y
+# CONFIG_FRAMEBUFFER_CONSOLE_ROTATION is not set
+# CONFIG_FONTS is not set
+CONFIG_FONT_8x8=y
+CONFIG_FONT_8x16=y
+
+#
+# Logo configuration
+#
+CONFIG_LOGO=y
+# CONFIG_LOGO_LINUX_MONO is not set
+# CONFIG_LOGO_LINUX_VGA16 is not set
+# CONFIG_LOGO_LINUX_CLUT224 is not set
+# CONFIG_LOGO_SUPERH_MONO is not set
+# CONFIG_LOGO_SUPERH_VGA16 is not set
+CONFIG_LOGO_SUPERH_CLUT224=y
 
 #
 # Sound
@@ -985,6 +1069,12 @@ CONFIG_SOUND_PRIME=m
 CONFIG_AC97_BUS=m
 
 #
+# HID Devices
+#
+CONFIG_HID=y
+# CONFIG_HID_DEBUG is not set
+
+#
 # USB support
 #
 CONFIG_USB_ARCH_HAS_HCD=y
@@ -1237,7 +1327,7 @@ CONFIG_LOG_BUF_SHIFT=14
 CONFIG_EARLY_SCIF_CONSOLE=y
 CONFIG_EARLY_SCIF_CONSOLE_PORT=0xffe80000
 CONFIG_EARLY_PRINTK=y
-# CONFIG_KGDB is not set
+# CONFIG_SH_KGDB is not set
 
 #
 # Security options
diff --git a/arch/sh/kernel/entry-common.S b/arch/sh/kernel/entry-common.S
index ab4ebb8..b467280 100644
--- a/arch/sh/kernel/entry-common.S
+++ b/arch/sh/kernel/entry-common.S
@@ -224,7 +224,7 @@ work_resched:
 syscall_exit_work:
 	! r0: current_thread_info->flags
 	! r8: current_thread_info
-	tst	#_TIF_SYSCALL_TRACE, r0
+	tst	#_TIF_SYSCALL_TRACE | _TIF_SINGLESTEP, r0
 	bt/s	work_pending
 	 tst	#_TIF_NEED_RESCHED, r0
 #ifdef CONFIG_TRACE_IRQFLAGS
diff --git a/arch/sh/kernel/io_generic.c b/arch/sh/kernel/io_generic.c
index 66626c0..771ea42 100644
--- a/arch/sh/kernel/io_generic.c
+++ b/arch/sh/kernel/io_generic.c
@@ -14,7 +14,6 @@
 #include <linux/module.h>
 #include <linux/io.h>
 #include <asm/machvec.h>
-#include <asm/cacheflush.h>
 
 #ifdef CONFIG_CPU_SH3
 /* SH3 has a PCMCIA bug that needs a dummy read from area 6 for a
@@ -96,7 +95,6 @@ void generic_insw(unsigned long port, void *dst, unsigned long count)
 	while (count--)
 		*buf++ = *port_addr;
 
-	flush_dcache_all();
 	dummy_read();
 }
 
@@ -171,7 +169,6 @@ void generic_outsw(unsigned long port, const void *src, unsigned long count)
 	while (count--)
 		*port_addr = *buf++;
 
-	flush_dcache_all();
 	dummy_read();
 }
 
diff --git a/arch/sh/kernel/process.c b/arch/sh/kernel/process.c
index 9d6a438..e760736 100644
--- a/arch/sh/kernel/process.c
+++ b/arch/sh/kernel/process.c
@@ -250,12 +250,11 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 		childregs->regs[15] = usp;
 		ti->addr_limit = USER_DS;
 	} else {
-		childregs->regs[15] = (unsigned long)task_stack_page(p) +
-							THREAD_SIZE;
+		childregs->regs[15] = (unsigned long)childregs;
 		ti->addr_limit = KERNEL_DS;
 	}
 
-        if (clone_flags & CLONE_SETTLS)
+	if (clone_flags & CLONE_SETTLS)
 		childregs->gbr = childregs->regs[0];
 
 	childregs->regs[0] = 0; /* Set return value for child */
diff --git a/arch/sh/kernel/ptrace.c b/arch/sh/kernel/ptrace.c
index 04ca13a0..855f724 100644
--- a/arch/sh/kernel/ptrace.c
+++ b/arch/sh/kernel/ptrace.c
@@ -8,7 +8,6 @@
  * SuperH version:   Copyright (C) 1999, 2000  Kaz Kojima & Niibe Yutaka
  *
  */
-
 #include <linux/kernel.h>
 #include <linux/sched.h>
 #include <linux/mm.h>
@@ -20,8 +19,7 @@
 #include <linux/slab.h>
 #include <linux/security.h>
 #include <linux/signal.h>
-
-#include <asm/io.h>
+#include <linux/io.h>
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
 #include <asm/system.h>
@@ -59,6 +57,23 @@ static inline int put_stack_long(struct task_struct *task, int offset,
 	return 0;
 }
 
+static void ptrace_disable_singlestep(struct task_struct *child)
+{
+	clear_tsk_thread_flag(child, TIF_SINGLESTEP);
+
+	/*
+	 * Ensure the UBC is not programmed at the next context switch.
+	 *
+	 * Normally this is not needed but there are sequences such as
+	 * singlestep, signal delivery, and continue that leave the
+	 * ubc_pc non-zero leading to spurious SIGTRAPs.
+	 */
+	if (child->thread.ubc_pc != 0) {
+		ubc_usercnt -= 1;
+		child->thread.ubc_pc = 0;
+	}
+}
+
 /*
  * Called by kernel/ptrace.c when detaching..
  *
@@ -66,7 +81,7 @@ static inline int put_stack_long(struct task_struct *task, int offset,
  */
 void ptrace_disable(struct task_struct *child)
 {
-	/* nothing to do.. */
+	ptrace_disable_singlestep(child);
 }
 
 long arch_ptrace(struct task_struct *child, long request, long addr, long data)
@@ -76,7 +91,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 
 	switch (request) {
 	/* when I and D space are separate, these will need to be fixed. */
-	case PTRACE_PEEKTEXT: /* read word at location addr. */ 
+	case PTRACE_PEEKTEXT: /* read word at location addr. */
 	case PTRACE_PEEKDATA: {
 		unsigned long tmp;
 		int copied;
@@ -94,7 +109,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 		unsigned long tmp;
 
 		ret = -EIO;
-		if ((addr & 3) || addr < 0 || 
+		if ((addr & 3) || addr < 0 ||
 		    addr > sizeof(struct user) - 3)
 			break;
 
@@ -129,7 +144,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 
 	case PTRACE_POKEUSR: /* write the word at location addr in the USER area */
 		ret = -EIO;
-		if ((addr & 3) || addr < 0 || 
+		if ((addr & 3) || addr < 0 ||
 		    addr > sizeof(struct user) - 3)
 			break;
 
@@ -156,6 +171,9 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 			set_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
 		else
 			clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
+
+		ptrace_disable_singlestep(child);
+
 		child->exit_code = data;
 		wake_up_process(child);
 		ret = 0;
@@ -163,14 +181,15 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 	}
 
 /*
- * make the child exit.  Best I can do is send it a sigkill. 
- * perhaps it should be put in the status that it wants to 
+ * make the child exit.  Best I can do is send it a sigkill.
+ * perhaps it should be put in the status that it wants to
  * exit.
  */
 	case PTRACE_KILL: {
 		ret = 0;
 		if (child->exit_state == EXIT_ZOMBIE)	/* already dead */
 			break;
+		ptrace_disable_singlestep(child);
 		child->exit_code = SIGKILL;
 		wake_up_process(child);
 		break;
@@ -196,6 +215,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 			ubc_usercnt += 1;
 		child->thread.ubc_pc = pc;
 
+		set_tsk_thread_flag(child, TIF_SINGLESTEP);
 		child->exit_code = data;
 		/* give it a chance to run. */
 		wake_up_process(child);
@@ -248,14 +268,15 @@ asmlinkage void do_syscall_trace(void)
 {
 	struct task_struct *tsk = current;
 
-	if (!test_thread_flag(TIF_SYSCALL_TRACE))
+	if (!test_thread_flag(TIF_SYSCALL_TRACE) &&
+	    !test_thread_flag(TIF_SINGLESTEP))
 		return;
 	if (!(tsk->ptrace & PT_PTRACED))
 		return;
 	/* the 0x80 provides a way for the tracing parent to distinguish
 	   between a syscall stop and SIGTRAP delivery */
-	ptrace_notify(SIGTRAP | ((current->ptrace & PT_TRACESYSGOOD)
-				 ? 0x80 : 0));
+	ptrace_notify(SIGTRAP | ((current->ptrace & PT_TRACESYSGOOD) &&
+				 !test_thread_flag(TIF_SINGLESTEP) ? 0x80 : 0));
 
 	/*
 	 * this isn't the same as continuing with a signal, but it will do
diff --git a/arch/sh/kernel/signal.c b/arch/sh/kernel/signal.c
index 32f10a0..9f39ef1 100644
--- a/arch/sh/kernel/signal.c
+++ b/arch/sh/kernel/signal.c
@@ -589,6 +589,8 @@ static void do_signal(struct pt_regs *regs, unsigned int save_r0)
 			if (test_thread_flag(TIF_RESTORE_SIGMASK))
 				clear_thread_flag(TIF_RESTORE_SIGMASK);
 		}
+
+		return;
 	}
 
  no_signal:
@@ -598,7 +600,7 @@ static void do_signal(struct pt_regs *regs, unsigned int save_r0)
 		if (regs->regs[0] == -ERESTARTNOHAND ||
 		    regs->regs[0] == -ERESTARTSYS ||
 		    regs->regs[0] == -ERESTARTNOINTR) {
-		    	regs->regs[0] = save_r0;
+			regs->regs[0] = save_r0;
 			regs->pc -= 2;
 		} else if (regs->regs[0] == -ERESTART_RESTARTBLOCK) {
 			regs->pc -= 2;
diff --git a/arch/sh/kernel/vmlinux.lds.S b/arch/sh/kernel/vmlinux.lds.S
index 75de165..78a6c09 100644
--- a/arch/sh/kernel/vmlinux.lds.S
+++ b/arch/sh/kernel/vmlinux.lds.S
@@ -3,6 +3,7 @@
  * Written by Niibe Yutaka
  */
 #include <asm/thread_info.h>
+#include <asm/cache.h>
 #include <asm-generic/vmlinux.lds.h>
 
 #ifdef CONFIG_CPU_LITTLE_ENDIAN
@@ -53,7 +54,7 @@ SECTIONS
   . = ALIGN(PAGE_SIZE);
   .data.page_aligned : { *(.data.page_aligned) }
 
-  . = ALIGN(32);
+  . = ALIGN(L1_CACHE_BYTES);
   __per_cpu_start = .;
   .data.percpu : { *(.data.percpu) }
   __per_cpu_end = .;
diff --git a/arch/sh/mm/cache-sh4.c b/arch/sh/mm/cache-sh4.c
index e0cd4b7..981b040 100644
--- a/arch/sh/mm/cache-sh4.c
+++ b/arch/sh/mm/cache-sh4.c
@@ -237,20 +237,10 @@ static inline void flush_cache_4096(unsigned long start,
 /*
  * Write back & invalidate the D-cache of the page.
  * (To avoid "alias" issues)
- *
- * This uses a lazy write-back on UP, which is explicitly
- * disabled on SMP.
  */
 void flush_dcache_page(struct page *page)
 {
-#ifndef CONFIG_SMP
-	struct address_space *mapping = page_mapping(page);
-
-	if (mapping && !mapping_mapped(mapping))
-		set_bit(PG_dcache_dirty, &page->flags);
-	else
-#endif
-	{
+	if (test_bit(PG_mapped, &page->flags)) {
 		unsigned long phys = PHYSADDR(page_address(page));
 		unsigned long addr = CACHE_OC_ADDRESS_ARRAY;
 		int i, n;
diff --git a/arch/sh/mm/cache-sh7705.c b/arch/sh/mm/cache-sh7705.c
index 31f8deb..4896d73 100644
--- a/arch/sh/mm/cache-sh7705.c
+++ b/arch/sh/mm/cache-sh7705.c
@@ -3,11 +3,11 @@
  *
  * Copyright (C) 1999, 2000  Niibe Yutaka
  * Copyright (C) 2004  Alex Song
- * Copyright (C) 2006  Paul Mundt
  *
  * 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.
+ *
  */
 #include <linux/init.h>
 #include <linux/mman.h>
@@ -51,6 +51,7 @@ static inline void cache_wback_all(void)
 
 			if ((data & v) == v)
 				ctrl_outl(data & ~v, addr);
+
 		}
 
 		addrstart += current_cpu_data.dcache.way_incr;
@@ -127,11 +128,7 @@ static void __flush_dcache_page(unsigned long phys)
  */
 void flush_dcache_page(struct page *page)
 {
-	struct address_space *mapping = page_mapping(page);
-
-	if (mapping && !mapping_mapped(mapping))
-		set_bit(PG_dcache_dirty, &page->flags);
-	else
+	if (test_bit(PG_mapped, &page->flags))
 		__flush_dcache_page(PHYSADDR(page_address(page)));
 }
 
diff --git a/arch/sh/mm/pg-sh4.c b/arch/sh/mm/pg-sh4.c
index 969efece..df69da9 100644
--- a/arch/sh/mm/pg-sh4.c
+++ b/arch/sh/mm/pg-sh4.c
@@ -23,6 +23,7 @@ extern struct mutex p3map_mutex[];
  */
 void clear_user_page(void *to, unsigned long address, struct page *page)
 {
+	__set_bit(PG_mapped, &page->flags);
 	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0)
 		clear_page(to);
 	else {
@@ -58,6 +59,7 @@ void clear_user_page(void *to, unsigned long address, struct page *page)
 void copy_user_page(void *to, void *from, unsigned long address,
 		    struct page *page)
 {
+	__set_bit(PG_mapped, &page->flags);
 	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0)
 		copy_page(to, from);
 	else {
@@ -82,3 +84,23 @@ void copy_user_page(void *to, void *from, unsigned long address,
 		mutex_unlock(&p3map_mutex[(address & CACHE_ALIAS)>>12]);
 	}
 }
+
+/*
+ * For SH-4, we have our own implementation for ptep_get_and_clear
+ */
+inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
+{
+	pte_t pte = *ptep;
+
+	pte_clear(mm, addr, ptep);
+	if (!pte_not_present(pte)) {
+		unsigned long pfn = pte_pfn(pte);
+		if (pfn_valid(pfn)) {
+			struct page *page = pfn_to_page(pfn);
+			struct address_space *mapping = page_mapping(page);
+			if (!mapping || !mapping_writably_mapped(mapping))
+				__clear_bit(PG_mapped, &page->flags);
+		}
+	}
+	return pte;
+}
diff --git a/arch/sh/mm/pg-sh7705.c b/arch/sh/mm/pg-sh7705.c
index 887ab9d..a4b015f 100644
--- a/arch/sh/mm/pg-sh7705.c
+++ b/arch/sh/mm/pg-sh7705.c
@@ -7,7 +7,9 @@
  * 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.
+ *
  */
+
 #include <linux/init.h>
 #include <linux/mman.h>
 #include <linux/mm.h>
@@ -74,6 +76,7 @@ void clear_user_page(void *to, unsigned long address, struct page *pg)
 {
 	struct page *page = virt_to_page(to);
 
+	__set_bit(PG_mapped, &page->flags);
 	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0) {
 		clear_page(to);
 		__flush_wback_region(to, PAGE_SIZE);
@@ -92,11 +95,12 @@ void clear_user_page(void *to, unsigned long address, struct page *pg)
  * @from: P1 address
  * @address: U0 address to be mapped
  */
-void copy_user_page(void *to, void *from, unsigned long address,
-		    struct page *pg)
+void copy_user_page(void *to, void *from, unsigned long address, struct page *pg)
 {
 	struct page *page = virt_to_page(to);
 
+
+	__set_bit(PG_mapped, &page->flags);
 	if (((address ^ (unsigned long)to) & CACHE_ALIAS) == 0) {
 		copy_page(to, from);
 		__flush_wback_region(to, PAGE_SIZE);
@@ -108,3 +112,26 @@ void copy_user_page(void *to, void *from, unsigned long address,
 		__flush_wback_region(to, PAGE_SIZE);
 	}
 }
+
+/*
+ * For SH7705, we have our own implementation for ptep_get_and_clear
+ * Copied from pg-sh4.c
+ */
+inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
+{
+	pte_t pte = *ptep;
+
+	pte_clear(mm, addr, ptep);
+	if (!pte_not_present(pte)) {
+		unsigned long pfn = pte_pfn(pte);
+		if (pfn_valid(pfn)) {
+			struct page *page = pfn_to_page(pfn);
+			struct address_space *mapping = page_mapping(page);
+			if (!mapping || !mapping_writably_mapped(mapping))
+				__clear_bit(PG_mapped, &page->flags);
+		}
+	}
+
+	return pte;
+}
+
diff --git a/arch/sh/mm/tlb-flush.c b/arch/sh/mm/tlb-flush.c
index d2f7b4a..6f45c1f 100644
--- a/arch/sh/mm/tlb-flush.c
+++ b/arch/sh/mm/tlb-flush.c
@@ -2,17 +2,15 @@
  * TLB flushing operations for SH with an MMU.
  *
  *  Copyright (C) 1999  Niibe Yutaka
- *  Copyright (C) 2003 - 2006  Paul Mundt
+ *  Copyright (C) 2003  Paul Mundt
  *
  * 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.
  */
 #include <linux/mm.h>
-#include <linux/io.h>
 #include <asm/mmu_context.h>
 #include <asm/tlbflush.h>
-#include <asm/cacheflush.h>
 
 void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
 {
@@ -140,54 +138,3 @@ void local_flush_tlb_all(void)
 	ctrl_barrier();
 	local_irq_restore(flags);
 }
-
-void update_mmu_cache(struct vm_area_struct *vma,
-		      unsigned long address, pte_t pte)
-{
-	unsigned long flags;
-	unsigned long pteval;
-	unsigned long vpn;
-	struct page *page;
-	unsigned long pfn = pte_pfn(pte);
-	struct address_space *mapping;
-
-	if (!pfn_valid(pfn))
-		return;
-
-	page = pfn_to_page(pfn);
-	mapping = page_mapping(page);
-	if (mapping) {
-		unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
-		int dirty = test_and_clear_bit(PG_dcache_dirty, &page->flags);
-
-		if (dirty)
-			__flush_wback_region((void *)P1SEGADDR(phys),
-					     PAGE_SIZE);
-	}
-
-	local_irq_save(flags);
-
-	/* Set PTEH register */
-	vpn = (address & MMU_VPN_MASK) | get_asid();
-	ctrl_outl(vpn, MMU_PTEH);
-
-	pteval = pte_val(pte);
-
-#ifdef CONFIG_CPU_HAS_PTEA
-	/* Set PTEA register */
-	/* TODO: make this look less hacky */
-	ctrl_outl(((pteval >> 28) & 0xe) | (pteval & 0x1), MMU_PTEA);
-#endif
-
-	/* Set PTEL register */
-	pteval &= _PAGE_FLAGS_HARDWARE_MASK; /* drop software flags */
-#if defined(CONFIG_SH_WRITETHROUGH) && defined(CONFIG_CPU_SH4)
-	pteval |= _PAGE_WT;
-#endif
-	/* conveniently, we want all the software flags to be 0 anyway */
-	ctrl_outl(pteval, MMU_PTEL);
-
-	/* Load the TLB */
-	asm volatile("ldtlb": /* no output */ : /* no input */ : "memory");
-	local_irq_restore(flags);
-}
diff --git a/arch/sh/mm/tlb-sh3.c b/arch/sh/mm/tlb-sh3.c
index e5e76eb..7fbfd5a 100644
--- a/arch/sh/mm/tlb-sh3.c
+++ b/arch/sh/mm/tlb-sh3.c
@@ -8,9 +8,69 @@
  *
  * Released under the terms of the GNU GPL v2.0.
  */
-#include <linux/io.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+
 #include <asm/system.h>
+#include <asm/io.h>
+#include <asm/uaccess.h>
+#include <asm/pgalloc.h>
 #include <asm/mmu_context.h>
+#include <asm/cacheflush.h>
+
+void update_mmu_cache(struct vm_area_struct * vma,
+		      unsigned long address, pte_t pte)
+{
+	unsigned long flags;
+	unsigned long pteval;
+	unsigned long vpn;
+
+	/* Ptrace may call this routine. */
+	if (vma && current->active_mm != vma->vm_mm)
+		return;
+
+#if defined(CONFIG_SH7705_CACHE_32KB)
+	{
+		struct page *page = pte_page(pte);
+		unsigned long pfn = pte_pfn(pte);
+
+		if (pfn_valid(pfn) && !test_bit(PG_mapped, &page->flags)) {
+			unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
+
+			__flush_wback_region((void *)P1SEGADDR(phys),
+					     PAGE_SIZE);
+			__set_bit(PG_mapped, &page->flags);
+		}
+	}
+#endif
+
+	local_irq_save(flags);
+
+	/* Set PTEH register */
+	vpn = (address & MMU_VPN_MASK) | get_asid();
+	ctrl_outl(vpn, MMU_PTEH);
+
+	pteval = pte_val(pte);
+
+	/* Set PTEL register */
+	pteval &= _PAGE_FLAGS_HARDWARE_MASK; /* drop software flags */
+	/* conveniently, we want all the software flags to be 0 anyway */
+	ctrl_outl(pteval, MMU_PTEL);
+
+	/* Load the TLB */
+	asm volatile("ldtlb": /* no output */ : /* no input */ : "memory");
+	local_irq_restore(flags);
+}
 
 void local_flush_tlb_one(unsigned long asid, unsigned long page)
 {
@@ -34,3 +94,4 @@ void local_flush_tlb_one(unsigned long asid, unsigned long page)
 	for (i = 0; i < ways; i++)
 		ctrl_outl(data, addr + (i << 8));
 }
+
diff --git a/arch/sh/mm/tlb-sh4.c b/arch/sh/mm/tlb-sh4.c
index 221e709..f74cf66 100644
--- a/arch/sh/mm/tlb-sh4.c
+++ b/arch/sh/mm/tlb-sh4.c
@@ -8,9 +8,74 @@
  *
  * Released under the terms of the GNU GPL v2.0.
  */
-#include <linux/io.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/interrupt.h>
+
 #include <asm/system.h>
+#include <asm/io.h>
+#include <asm/uaccess.h>
+#include <asm/pgalloc.h>
 #include <asm/mmu_context.h>
+#include <asm/cacheflush.h>
+
+void update_mmu_cache(struct vm_area_struct * vma,
+		      unsigned long address, pte_t pte)
+{
+	unsigned long flags;
+	unsigned long pteval;
+	unsigned long vpn;
+	struct page *page;
+	unsigned long pfn;
+
+	/* Ptrace may call this routine. */
+	if (vma && current->active_mm != vma->vm_mm)
+		return;
+
+	pfn = pte_pfn(pte);
+	if (pfn_valid(pfn)) {
+		page = pfn_to_page(pfn);
+		if (!test_bit(PG_mapped, &page->flags)) {
+			unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
+			__flush_wback_region((void *)P1SEGADDR(phys), PAGE_SIZE);
+			__set_bit(PG_mapped, &page->flags);
+		}
+	}
+
+	local_irq_save(flags);
+
+	/* Set PTEH register */
+	vpn = (address & MMU_VPN_MASK) | get_asid();
+	ctrl_outl(vpn, MMU_PTEH);
+
+	pteval = pte_val(pte);
+
+	/* Set PTEA register */
+	if (cpu_data->flags & CPU_HAS_PTEA)
+		/* TODO: make this look less hacky */
+		ctrl_outl(((pteval >> 28) & 0xe) | (pteval & 0x1), MMU_PTEA);
+
+	/* Set PTEL register */
+	pteval &= _PAGE_FLAGS_HARDWARE_MASK; /* drop software flags */
+#ifdef CONFIG_SH_WRITETHROUGH
+	pteval |= _PAGE_WT;
+#endif
+	/* conveniently, we want all the software flags to be 0 anyway */
+	ctrl_outl(pteval, MMU_PTEL);
+
+	/* Load the TLB */
+	asm volatile("ldtlb": /* no output */ : /* no input */ : "memory");
+	local_irq_restore(flags);
+}
 
 void local_flush_tlb_one(unsigned long asid, unsigned long page)
 {
@@ -28,3 +93,4 @@ void local_flush_tlb_one(unsigned long asid, unsigned long page)
 	ctrl_outl(data, addr);
 	back_to_P1();
 }
+