Merge branch 'master' into sh/cachetlb

43 files changed, 1533 insertions, 211 deletions

diff --git a/arch/sh/kernel/Makefile_32 b/arch/sh/kernel/Makefile_32
index 94ed99b6800..f2245ebf0b3 100644
--- a/arch/sh/kernel/Makefile_32
+++ b/arch/sh/kernel/Makefile_32
@@ -9,10 +9,10 @@ ifdef CONFIG_FUNCTION_TRACER
 CFLAGS_REMOVE_ftrace.o = -pg
 endif
 
-obj-y	:= debugtraps.o idle.o io.o io_generic.o irq.o			\
+obj-y	:= debugtraps.o dumpstack.o idle.o io.o io_generic.o irq.o	\
 	   machvec.o process_32.o ptrace_32.o setup.o signal_32.o	\
-	   sys_sh.o sys_sh32.o syscalls_32.o time.o topology.o	\
-	   traps.o traps_32.o
+	   sys_sh.o sys_sh32.o syscalls_32.o time.o topology.o		\
+	   traps.o traps_32.o unwinder.o
 
 obj-y				+= cpu/
 obj-$(CONFIG_VSYSCALL)		+= vsyscall/
@@ -33,6 +33,7 @@ obj-$(CONFIG_FTRACE_SYSCALLS)	+= ftrace.o
 obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
 obj-$(CONFIG_DUMP_CODE)		+= disassemble.o
 obj-$(CONFIG_HIBERNATION)	+= swsusp.o
+obj-$(CONFIG_DWARF_UNWINDER)	+= dwarf.o
 
 obj-$(CONFIG_GENERIC_CLOCKEVENTS_BROADCAST)	+= localtimer.o
 
diff --git a/arch/sh/kernel/Makefile_64 b/arch/sh/kernel/Makefile_64
index 67b9f6c6326..639ee514266 100644
--- a/arch/sh/kernel/Makefile_64
+++ b/arch/sh/kernel/Makefile_64
@@ -2,7 +2,7 @@ extra-y	:= head_64.o init_task.o vmlinux.lds
 
 obj-y	:= debugtraps.o idle.o io.o io_generic.o irq.o machvec.o process_64.o \
 	   ptrace_64.o setup.o signal_64.o sys_sh.o sys_sh64.o \
-	   syscalls_64.o time.o topology.o traps.o traps_64.o
+	   syscalls_64.o time.o topology.o traps.o traps_64.o unwinder.o
 
 obj-y				+= cpu/
 obj-$(CONFIG_SMP)		+= smp.o
@@ -13,6 +13,7 @@ obj-$(CONFIG_CRASH_DUMP)	+= crash_dump.o
 obj-$(CONFIG_STACKTRACE)	+= stacktrace.o
 obj-$(CONFIG_IO_TRAPPED)	+= io_trapped.o
 obj-$(CONFIG_GENERIC_GPIO)	+= gpio.o
+obj-$(CONFIG_DWARF_UNWINDER)	+= dwarf.o
 
 obj-$(CONFIG_GENERIC_CLOCKEVENTS_BROADCAST)	+= localtimer.o
 
diff --git a/arch/sh/kernel/cpu/sh2/setup-sh7619.c b/arch/sh/kernel/cpu/sh2/setup-sh7619.c
index 13798733f2d..8555c05e866 100644
--- a/arch/sh/kernel/cpu/sh2/setup-sh7619.c
+++ b/arch/sh/kernel/cpu/sh2/setup-sh7619.c
@@ -187,7 +187,7 @@ static int __init sh7619_devices_setup(void)
 	return platform_add_devices(sh7619_devices,
 				    ARRAY_SIZE(sh7619_devices));
 }
-__initcall(sh7619_devices_setup);
+arch_initcall(sh7619_devices_setup);
 
 void __init plat_irq_setup(void)
 {
diff --git a/arch/sh/kernel/cpu/sh2a/setup-mxg.c b/arch/sh/kernel/cpu/sh2a/setup-mxg.c
index 869c2da4820..b6737644531 100644
--- a/arch/sh/kernel/cpu/sh2a/setup-mxg.c
+++ b/arch/sh/kernel/cpu/sh2a/setup-mxg.c
@@ -238,7 +238,7 @@ static int __init mxg_devices_setup(void)
 	return platform_add_devices(mxg_devices,
 				    ARRAY_SIZE(mxg_devices));
 }
-__initcall(mxg_devices_setup);
+arch_initcall(mxg_devices_setup);
 
 void __init plat_irq_setup(void)
 {
diff --git a/arch/sh/kernel/cpu/sh2a/setup-sh7201.c b/arch/sh/kernel/cpu/sh2a/setup-sh7201.c
index d8febe12806..fbde5b75deb 100644
--- a/arch/sh/kernel/cpu/sh2a/setup-sh7201.c
+++ b/arch/sh/kernel/cpu/sh2a/setup-sh7201.c
@@ -357,7 +357,7 @@ static int __init sh7201_devices_setup(void)
 	return platform_add_devices(sh7201_devices,
 				    ARRAY_SIZE(sh7201_devices));
 }
-__initcall(sh7201_devices_setup);
+arch_initcall(sh7201_devices_setup);
 
 void __init plat_irq_setup(void)
 {
diff --git a/arch/sh/kernel/cpu/sh2a/setup-sh7203.c b/arch/sh/kernel/cpu/sh2a/setup-sh7203.c
index 62e3039d239..d3fd536c9a8 100644
--- a/arch/sh/kernel/cpu/sh2a/setup-sh7203.c
+++ b/arch/sh/kernel/cpu/sh2a/setup-sh7203.c
@@ -367,7 +367,7 @@ static int __init sh7203_devices_setup(void)
 	return platform_add_devices(sh7203_devices,
 				    ARRAY_SIZE(sh7203_devices));
 }
-__initcall(sh7203_devices_setup);
+arch_initcall(sh7203_devices_setup);
 
 void __init plat_irq_setup(void)
 {
diff --git a/arch/sh/kernel/cpu/sh2a/setup-sh7206.c b/arch/sh/kernel/cpu/sh2a/setup-sh7206.c
index 3e6f3d7a58b..a9ccc5e8d9e 100644
--- a/arch/sh/kernel/cpu/sh2a/setup-sh7206.c
+++ b/arch/sh/kernel/cpu/sh2a/setup-sh7206.c
@@ -338,7 +338,7 @@ static int __init sh7206_devices_setup(void)
 	return platform_add_devices(sh7206_devices,
 				    ARRAY_SIZE(sh7206_devices));
 }
-__initcall(sh7206_devices_setup);
+arch_initcall(sh7206_devices_setup);
 
 void __init plat_irq_setup(void)
 {
diff --git a/arch/sh/kernel/cpu/sh3/entry.S b/arch/sh/kernel/cpu/sh3/entry.S
index a701fac8ed4..854921c6f45 100644
--- a/arch/sh/kernel/cpu/sh3/entry.S
+++ b/arch/sh/kernel/cpu/sh3/entry.S
@@ -137,6 +137,7 @@ ENTRY(tlb_protection_violation_store)
 	 mov	#1, r5
 
 call_handle_tlbmiss:
+	setup_frame_reg
 	mov.l	1f, r0
 	mov	r5, r8
 	mov.l	@r0, r6
diff --git a/arch/sh/kernel/cpu/sh3/setup-sh7705.c b/arch/sh/kernel/cpu/sh3/setup-sh7705.c
index 88f742fed9e..c2310598387 100644
--- a/arch/sh/kernel/cpu/sh3/setup-sh7705.c
+++ b/arch/sh/kernel/cpu/sh3/setup-sh7705.c
@@ -222,7 +222,7 @@ static int __init sh7705_devices_setup(void)
 	return platform_add_devices(sh7705_devices,
 				    ARRAY_SIZE(sh7705_devices));
 }
-__initcall(sh7705_devices_setup);
+arch_initcall(sh7705_devices_setup);
 
 static struct platform_device *sh7705_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh3/setup-sh770x.c b/arch/sh/kernel/cpu/sh3/setup-sh770x.c
index c5630679858..347ab35d069 100644
--- a/arch/sh/kernel/cpu/sh3/setup-sh770x.c
+++ b/arch/sh/kernel/cpu/sh3/setup-sh770x.c
@@ -250,7 +250,7 @@ static int __init sh770x_devices_setup(void)
 	return platform_add_devices(sh770x_devices,
 		ARRAY_SIZE(sh770x_devices));
 }
-__initcall(sh770x_devices_setup);
+arch_initcall(sh770x_devices_setup);
 
 static struct platform_device *sh770x_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh3/setup-sh7710.c b/arch/sh/kernel/cpu/sh3/setup-sh7710.c
index efa76c8148f..717e90ae109 100644
--- a/arch/sh/kernel/cpu/sh3/setup-sh7710.c
+++ b/arch/sh/kernel/cpu/sh3/setup-sh7710.c
@@ -226,7 +226,7 @@ static int __init sh7710_devices_setup(void)
 	return platform_add_devices(sh7710_devices,
 				    ARRAY_SIZE(sh7710_devices));
 }
-__initcall(sh7710_devices_setup);
+arch_initcall(sh7710_devices_setup);
 
 static struct platform_device *sh7710_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh3/setup-sh7720.c b/arch/sh/kernel/cpu/sh3/setup-sh7720.c
index 5b2107798ed..74d8baaf8e9 100644
--- a/arch/sh/kernel/cpu/sh3/setup-sh7720.c
+++ b/arch/sh/kernel/cpu/sh3/setup-sh7720.c
@@ -388,7 +388,7 @@ static int __init sh7720_devices_setup(void)
 	return platform_add_devices(sh7720_devices,
 				    ARRAY_SIZE(sh7720_devices));
 }
-__initcall(sh7720_devices_setup);
+arch_initcall(sh7720_devices_setup);
 
 static struct platform_device *sh7720_early_devices[] __initdata = {
 	&cmt0_device,
diff --git a/arch/sh/kernel/cpu/sh4/setup-sh4-202.c b/arch/sh/kernel/cpu/sh4/setup-sh4-202.c
index 6d088d12359..de4827df19a 100644
--- a/arch/sh/kernel/cpu/sh4/setup-sh4-202.c
+++ b/arch/sh/kernel/cpu/sh4/setup-sh4-202.c
@@ -138,7 +138,7 @@ static int __init sh4202_devices_setup(void)
 	return platform_add_devices(sh4202_devices,
 				    ARRAY_SIZE(sh4202_devices));
 }
-__initcall(sh4202_devices_setup);
+arch_initcall(sh4202_devices_setup);
 
 static struct platform_device *sh4202_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh4/setup-sh7750.c b/arch/sh/kernel/cpu/sh4/setup-sh7750.c
index 851672d15cf..1b8b122e8f3 100644
--- a/arch/sh/kernel/cpu/sh4/setup-sh7750.c
+++ b/arch/sh/kernel/cpu/sh4/setup-sh7750.c
@@ -239,7 +239,7 @@ static int __init sh7750_devices_setup(void)
 	return platform_add_devices(sh7750_devices,
 				    ARRAY_SIZE(sh7750_devices));
 }
-__initcall(sh7750_devices_setup);
+arch_initcall(sh7750_devices_setup);
 
 static struct platform_device *sh7750_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh4/setup-sh7760.c b/arch/sh/kernel/cpu/sh4/setup-sh7760.c
index 5b822519bd9..7fbb7be9284 100644
--- a/arch/sh/kernel/cpu/sh4/setup-sh7760.c
+++ b/arch/sh/kernel/cpu/sh4/setup-sh7760.c
@@ -265,7 +265,7 @@ static int __init sh7760_devices_setup(void)
 	return platform_add_devices(sh7760_devices,
 				    ARRAY_SIZE(sh7760_devices));
 }
-__initcall(sh7760_devices_setup);
+arch_initcall(sh7760_devices_setup);
 
 static struct platform_device *sh7760_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh4a/Makefile b/arch/sh/kernel/cpu/sh4a/Makefile
index 1d7ae38bc61..12cddf4c721 100644
--- a/arch/sh/kernel/cpu/sh4a/Makefile
+++ b/arch/sh/kernel/cpu/sh4a/Makefile
@@ -27,7 +27,7 @@ clock-$(CONFIG_CPU_SUBTYPE_SH7786)	:= clock-sh7786.o
 clock-$(CONFIG_CPU_SUBTYPE_SH7343)	:= clock-sh7343.o
 clock-$(CONFIG_CPU_SUBTYPE_SH7722)	:= clock-sh7722.o hwblk-sh7722.o
 clock-$(CONFIG_CPU_SUBTYPE_SH7723)	:= clock-sh7723.o hwblk-sh7723.o
-clock-$(CONFIG_CPU_SUBTYPE_SH7724)	:= clock-sh7724.o
+clock-$(CONFIG_CPU_SUBTYPE_SH7724)	:= clock-sh7724.o hwblk-sh7724.o
 clock-$(CONFIG_CPU_SUBTYPE_SH7366)	:= clock-sh7366.o
 clock-$(CONFIG_CPU_SUBTYPE_SHX3)	:= clock-shx3.o
 
diff --git a/arch/sh/kernel/cpu/sh4a/clock-sh7722.c b/arch/sh/kernel/cpu/sh4a/clock-sh7722.c
index 1fa9e1dd1cc..5b1bbbe63b1 100644
--- a/arch/sh/kernel/cpu/sh4a/clock-sh7722.c
+++ b/arch/sh/kernel/cpu/sh4a/clock-sh7722.c
@@ -32,9 +32,6 @@
 #define SCLKBCR		0xa415000c
 #define IRDACLKCR	0xa4150018
 #define PLLCR		0xa4150024
-#define MSTPCR0		0xa4150030
-#define MSTPCR1		0xa4150034
-#define MSTPCR2		0xa4150038
 #define DLLFRQ		0xa4150050
 
 /* Fixed 32 KHz root clock for RTC and Power Management purposes */
diff --git a/arch/sh/kernel/cpu/sh4a/clock-sh7723.c b/arch/sh/kernel/cpu/sh4a/clock-sh7723.c
index bf64c78eee3..e5c63911403 100644
--- a/arch/sh/kernel/cpu/sh4a/clock-sh7723.c
+++ b/arch/sh/kernel/cpu/sh4a/clock-sh7723.c
@@ -32,9 +32,6 @@
 #define SCLKBCR		0xa415000c
 #define IRDACLKCR	0xa4150018
 #define PLLCR		0xa4150024
-#define MSTPCR0		0xa4150030
-#define MSTPCR1		0xa4150034
-#define MSTPCR2		0xa4150038
 #define DLLFRQ		0xa4150050
 
 /* Fixed 32 KHz root clock for RTC and Power Management purposes */
diff --git a/arch/sh/kernel/cpu/sh4a/clock-sh7724.c b/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
index 5d5c9b95288..34611d97378 100644
--- a/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
+++ b/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
@@ -22,6 +22,8 @@
 #include <linux/kernel.h>
 #include <linux/io.h>
 #include <asm/clock.h>
+#include <asm/hwblk.h>
+#include <cpu/sh7724.h>
 
 /* SH7724 registers */
 #define FRQCRA		0xa4150000
@@ -31,9 +33,6 @@
 #define FCLKBCR		0xa415000c
 #define IRDACLKCR	0xa4150018
 #define PLLCR		0xa4150024
-#define MSTPCR0		0xa4150030
-#define MSTPCR1		0xa4150034
-#define MSTPCR2		0xa4150038
 #define SPUCLKCR	0xa415003c
 #define FLLFRQ		0xa4150050
 #define LSTATS		0xa4150060
@@ -156,64 +155,67 @@ struct clk div6_clks[] = {
 	SH_CLK_DIV6("spu_clk", &div3_clk, SPUCLKCR, 0),
 };
 
-#define MSTP(_str, _parent, _reg, _bit, _force_on, _need_cpg, _need_ram) \
-  SH_CLK_MSTP32(_str, -1, _parent, _reg, _bit, _force_on * CLK_ENABLE_ON_INIT)
+#define R_CLK (&r_clk)
+#define P_CLK (&div4_clks[DIV4_P])
+#define B_CLK (&div4_clks[DIV4_B])
+#define I_CLK (&div4_clks[DIV4_I])
+#define SH_CLK (&div4_clks[DIV4_SH])
 
 static struct clk mstp_clks[] = {
-	MSTP("tlb0", &div4_clks[DIV4_I], MSTPCR0, 31, 1, 1, 0),
-	MSTP("ic0", &div4_clks[DIV4_I], MSTPCR0, 30, 1, 1, 0),
-	MSTP("oc0", &div4_clks[DIV4_I], MSTPCR0, 29, 1, 1, 0),
-	MSTP("rs0", &div4_clks[DIV4_B], MSTPCR0, 28, 1, 1, 0),
-	MSTP("ilmem0", &div4_clks[DIV4_I], MSTPCR0, 27, 1, 1, 0),
-	MSTP("l2c0", &div4_clks[DIV4_SH], MSTPCR0, 26, 1, 1, 0),
-	MSTP("fpu0", &div4_clks[DIV4_I], MSTPCR0, 24, 1, 1, 0),
-	MSTP("intc0", &div4_clks[DIV4_P], MSTPCR0, 22, 1, 1, 0),
-	MSTP("dmac0", &div4_clks[DIV4_B], MSTPCR0, 21, 0, 1, 1),
-	MSTP("sh0", &div4_clks[DIV4_SH], MSTPCR0, 20, 0, 1, 0),
-	MSTP("hudi0", &div4_clks[DIV4_P], MSTPCR0, 19, 0, 1, 0),
-	MSTP("ubc0", &div4_clks[DIV4_I], MSTPCR0, 17, 0, 1, 0),
-	MSTP("tmu0", &div4_clks[DIV4_P], MSTPCR0, 15, 0, 1, 0),
-	MSTP("cmt0", &r_clk, MSTPCR0, 14, 0, 0, 0),
-	MSTP("rwdt0", &r_clk, MSTPCR0, 13, 0, 0, 0),
-	MSTP("dmac1", &div4_clks[DIV4_B], MSTPCR0, 12, 0, 1, 1),
-	MSTP("tmu1", &div4_clks[DIV4_P], MSTPCR0, 10, 0, 1, 0),
-	MSTP("scif0", &div4_clks[DIV4_P], MSTPCR0, 9, 0, 1, 0),
-	MSTP("scif1", &div4_clks[DIV4_P], MSTPCR0, 8, 0, 1, 0),
-	MSTP("scif2", &div4_clks[DIV4_P], MSTPCR0, 7, 0, 1, 0),
-	MSTP("scif3", &div4_clks[DIV4_B], MSTPCR0, 6, 0, 1, 0),
-	MSTP("scif4", &div4_clks[DIV4_B], MSTPCR0, 5, 0, 1, 0),
-	MSTP("scif5", &div4_clks[DIV4_B], MSTPCR0, 4, 0, 1, 0),
-	MSTP("msiof0", &div4_clks[DIV4_B], MSTPCR0, 2, 0, 1, 0),
-	MSTP("msiof1", &div4_clks[DIV4_B], MSTPCR0, 1, 0, 1, 0),
-
-	MSTP("keysc0", &r_clk, MSTPCR1, 12, 0, 0, 0),
-	MSTP("rtc0", &r_clk, MSTPCR1, 11, 0, 0, 0),
-	MSTP("i2c0", &div4_clks[DIV4_P], MSTPCR1, 9, 0, 1, 0),
-	MSTP("i2c1", &div4_clks[DIV4_P], MSTPCR1, 8, 0, 1, 0),
-
-	MSTP("mmc0", &div4_clks[DIV4_B], MSTPCR2, 29, 0, 1, 0),
-	MSTP("eth0", &div4_clks[DIV4_B], MSTPCR2, 28, 0, 1, 0),
-	MSTP("atapi0", &div4_clks[DIV4_B], MSTPCR2, 26, 0, 1, 0),
-	MSTP("tpu0", &div4_clks[DIV4_B], MSTPCR2, 25, 0, 1, 0),
-	MSTP("irda0", &div4_clks[DIV4_P], MSTPCR2, 24, 0, 1, 0),
-	MSTP("tsif0", &div4_clks[DIV4_B], MSTPCR2, 22, 0, 1, 0),
-	MSTP("usb1", &div4_clks[DIV4_B], MSTPCR2, 21, 0, 1, 1),
-	MSTP("usb0", &div4_clks[DIV4_B], MSTPCR2, 20, 0, 1, 1),
-	MSTP("2dg0", &div4_clks[DIV4_B], MSTPCR2, 19, 0, 1, 1),
-	MSTP("sdhi0", &div4_clks[DIV4_B], MSTPCR2, 18, 0, 1, 0),
-	MSTP("sdhi1", &div4_clks[DIV4_B], MSTPCR2, 17, 0, 1, 0),
-	MSTP("veu1", &div4_clks[DIV4_B], MSTPCR2, 15, 1, 1, 1),
-	MSTP("ceu1", &div4_clks[DIV4_B], MSTPCR2, 13, 0, 1, 1),
-	MSTP("beu1", &div4_clks[DIV4_B], MSTPCR2, 12, 0, 1, 1),
-	MSTP("2ddmac0", &div4_clks[DIV4_SH], MSTPCR2, 10, 0, 1, 1),
-	MSTP("spu0", &div4_clks[DIV4_B], MSTPCR2, 9, 0, 1, 0),
-	MSTP("jpu0", &div4_clks[DIV4_B], MSTPCR2, 6, 1, 1, 1),
-	MSTP("vou0", &div4_clks[DIV4_B], MSTPCR2, 5, 0, 1, 1),
-	MSTP("beu0", &div4_clks[DIV4_B], MSTPCR2, 4, 0, 1, 1),
-	MSTP("ceu0", &div4_clks[DIV4_B], MSTPCR2, 3, 0, 1, 1),
-	MSTP("veu0", &div4_clks[DIV4_B], MSTPCR2, 2, 1, 1, 1),
-	MSTP("vpu0", &div4_clks[DIV4_B], MSTPCR2, 1, 1, 1, 1),
-	MSTP("lcdc0", &div4_clks[DIV4_B], MSTPCR2, 0, 0, 1, 1),
+	SH_HWBLK_CLK("tlb0", -1, I_CLK, HWBLK_TLB, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("ic0", -1, I_CLK, HWBLK_IC, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("oc0", -1, I_CLK, HWBLK_OC, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("rs0", -1, B_CLK, HWBLK_RSMEM, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("ilmem0", -1, I_CLK, HWBLK_ILMEM, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("l2c0", -1, SH_CLK, HWBLK_L2C, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("fpu0", -1, I_CLK, HWBLK_FPU, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("intc0", -1, P_CLK, HWBLK_INTC, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("dmac0", -1, B_CLK, HWBLK_DMAC0, 0),
+	SH_HWBLK_CLK("sh0", -1, SH_CLK, HWBLK_SHYWAY, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("hudi0", -1, P_CLK, HWBLK_HUDI, 0),
+	SH_HWBLK_CLK("ubc0", -1, I_CLK, HWBLK_UBC, 0),
+	SH_HWBLK_CLK("tmu0", -1, P_CLK, HWBLK_TMU0, 0),
+	SH_HWBLK_CLK("cmt0", -1, R_CLK, HWBLK_CMT, 0),
+	SH_HWBLK_CLK("rwdt0", -1, R_CLK, HWBLK_RWDT, 0),
+	SH_HWBLK_CLK("dmac1", -1, B_CLK, HWBLK_DMAC1, 0),
+	SH_HWBLK_CLK("tmu1", -1, P_CLK, HWBLK_TMU1, 0),
+	SH_HWBLK_CLK("scif0", -1, P_CLK, HWBLK_SCIF0, 0),
+	SH_HWBLK_CLK("scif1", -1, P_CLK, HWBLK_SCIF1, 0),
+	SH_HWBLK_CLK("scif2", -1, P_CLK, HWBLK_SCIF2, 0),
+	SH_HWBLK_CLK("scif3", -1, B_CLK, HWBLK_SCIF3, 0),
+	SH_HWBLK_CLK("scif4", -1, B_CLK, HWBLK_SCIF4, 0),
+	SH_HWBLK_CLK("scif5", -1, B_CLK, HWBLK_SCIF5, 0),
+	SH_HWBLK_CLK("msiof0", -1, B_CLK, HWBLK_MSIOF0, 0),
+	SH_HWBLK_CLK("msiof1", -1, B_CLK, HWBLK_MSIOF1, 0),
+
+	SH_HWBLK_CLK("keysc0", -1, R_CLK, HWBLK_KEYSC, 0),
+	SH_HWBLK_CLK("rtc0", -1, R_CLK, HWBLK_RTC, 0),
+	SH_HWBLK_CLK("i2c0", -1, P_CLK, HWBLK_IIC0, 0),
+	SH_HWBLK_CLK("i2c1", -1, P_CLK, HWBLK_IIC1, 0),
+
+	SH_HWBLK_CLK("mmc0", -1, B_CLK, HWBLK_MMC, 0),
+	SH_HWBLK_CLK("eth0", -1, B_CLK, HWBLK_ETHER, 0),
+	SH_HWBLK_CLK("atapi0", -1, B_CLK, HWBLK_ATAPI, 0),
+	SH_HWBLK_CLK("tpu0", -1, B_CLK, HWBLK_TPU, 0),
+	SH_HWBLK_CLK("irda0", -1, P_CLK, HWBLK_IRDA, 0),
+	SH_HWBLK_CLK("tsif0", -1, B_CLK, HWBLK_TSIF, 0),
+	SH_HWBLK_CLK("usb1", -1, B_CLK, HWBLK_USB1, 0),
+	SH_HWBLK_CLK("usb0", -1, B_CLK, HWBLK_USB0, 0),
+	SH_HWBLK_CLK("2dg0", -1, B_CLK, HWBLK_2DG, 0),
+	SH_HWBLK_CLK("sdhi0", -1, B_CLK, HWBLK_SDHI0, 0),
+	SH_HWBLK_CLK("sdhi1", -1, B_CLK, HWBLK_SDHI1, 0),
+	SH_HWBLK_CLK("veu1", -1, B_CLK, HWBLK_VEU1, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("ceu1", -1, B_CLK, HWBLK_CEU1, 0),
+	SH_HWBLK_CLK("beu1", -1, B_CLK, HWBLK_BEU1, 0),
+	SH_HWBLK_CLK("2ddmac0", -1, SH_CLK, HWBLK_2DDMAC, 0),
+	SH_HWBLK_CLK("spu0", -1, B_CLK, HWBLK_SPU, 0),
+	SH_HWBLK_CLK("jpu0", -1, B_CLK, HWBLK_JPU, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("vou0", -1, B_CLK, HWBLK_VOU, 0),
+	SH_HWBLK_CLK("beu0", -1, B_CLK, HWBLK_BEU0, 0),
+	SH_HWBLK_CLK("ceu0", -1, B_CLK, HWBLK_CEU0, 0),
+	SH_HWBLK_CLK("veu0", -1, B_CLK, HWBLK_VEU0, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("vpu0", -1, B_CLK, HWBLK_VPU, CLK_ENABLE_ON_INIT),
+	SH_HWBLK_CLK("lcdc0", -1, B_CLK, HWBLK_LCDC, 0),
 };
 
 int __init arch_clk_init(void)
@@ -236,7 +238,7 @@ int __init arch_clk_init(void)
 		ret = sh_clk_div6_register(div6_clks, ARRAY_SIZE(div6_clks));
 
 	if (!ret)
-		ret = sh_clk_mstp32_register(mstp_clks, ARRAY_SIZE(mstp_clks));
+		ret = sh_hwblk_clk_register(mstp_clks, ARRAY_SIZE(mstp_clks));
 
 	return ret;
 }
diff --git a/arch/sh/kernel/cpu/sh4a/hwblk-sh7724.c b/arch/sh/kernel/cpu/sh4a/hwblk-sh7724.c
new file mode 100644
index 00000000000..1613ad6013c
--- /dev/null
+++ b/arch/sh/kernel/cpu/sh4a/hwblk-sh7724.c
@@ -0,0 +1,121 @@
+/*
+ * arch/sh/kernel/cpu/sh4a/hwblk-sh7724.c
+ *
+ * SH7724 hardware block support
+ *
+ * Copyright (C) 2009 Magnus Damm
+ *
+ * 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
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/io.h>
+#include <asm/suspend.h>
+#include <asm/hwblk.h>
+#include <cpu/sh7724.h>
+
+/* SH7724 registers */
+#define MSTPCR0		0xa4150030
+#define MSTPCR1		0xa4150034
+#define MSTPCR2		0xa4150038
+
+/* SH7724 Power Domains */
+enum { CORE_AREA, SUB_AREA, CORE_AREA_BM };
+static struct hwblk_area sh7724_hwblk_area[] = {
+	[CORE_AREA] = HWBLK_AREA(0, 0),
+	[CORE_AREA_BM] = HWBLK_AREA(HWBLK_AREA_FLAG_PARENT, CORE_AREA),
+	[SUB_AREA] = HWBLK_AREA(0, 0),
+};
+
+/* Table mapping HWBLK to Module Stop Bit and Power Domain */
+static struct hwblk sh7724_hwblk[HWBLK_NR] = {
+	[HWBLK_TLB] = HWBLK(MSTPCR0, 31, CORE_AREA),
+	[HWBLK_IC] = HWBLK(MSTPCR0, 30, CORE_AREA),
+	[HWBLK_OC] = HWBLK(MSTPCR0, 29, CORE_AREA),
+	[HWBLK_RSMEM] = HWBLK(MSTPCR0, 28, CORE_AREA),
+	[HWBLK_ILMEM] = HWBLK(MSTPCR0, 27, CORE_AREA),
+	[HWBLK_L2C] = HWBLK(MSTPCR0, 26, CORE_AREA),
+	[HWBLK_FPU] = HWBLK(MSTPCR0, 24, CORE_AREA),
+	[HWBLK_INTC] = HWBLK(MSTPCR0, 22, CORE_AREA),
+	[HWBLK_DMAC0] = HWBLK(MSTPCR0, 21, CORE_AREA_BM),
+	[HWBLK_SHYWAY] = HWBLK(MSTPCR0, 20, CORE_AREA),
+	[HWBLK_HUDI] = HWBLK(MSTPCR0, 19, CORE_AREA),
+	[HWBLK_DBG] = HWBLK(MSTPCR0, 18, CORE_AREA),
+	[HWBLK_UBC] = HWBLK(MSTPCR0, 17, CORE_AREA),
+	[HWBLK_TMU0] = HWBLK(MSTPCR0, 15, CORE_AREA),
+	[HWBLK_CMT] = HWBLK(MSTPCR0, 14, SUB_AREA),
+	[HWBLK_RWDT] = HWBLK(MSTPCR0, 13, SUB_AREA),
+	[HWBLK_DMAC1] = HWBLK(MSTPCR0, 12, CORE_AREA_BM),
+	[HWBLK_TMU1] = HWBLK(MSTPCR0, 10, CORE_AREA),
+	[HWBLK_SCIF0] = HWBLK(MSTPCR0, 9, CORE_AREA),
+	[HWBLK_SCIF1] = HWBLK(MSTPCR0, 8, CORE_AREA),
+	[HWBLK_SCIF2] = HWBLK(MSTPCR0, 7, CORE_AREA),
+	[HWBLK_SCIF3] = HWBLK(MSTPCR0, 6, CORE_AREA),
+	[HWBLK_SCIF4] = HWBLK(MSTPCR0, 5, CORE_AREA),
+	[HWBLK_SCIF5] = HWBLK(MSTPCR0, 4, CORE_AREA),
+	[HWBLK_MSIOF0] = HWBLK(MSTPCR0, 2, CORE_AREA),
+	[HWBLK_MSIOF1] = HWBLK(MSTPCR0, 1, CORE_AREA),
+
+	[HWBLK_KEYSC] = HWBLK(MSTPCR1, 12, SUB_AREA),
+	[HWBLK_RTC] = HWBLK(MSTPCR1, 11, SUB_AREA),
+	[HWBLK_IIC0] = HWBLK(MSTPCR1, 9, CORE_AREA),
+	[HWBLK_IIC1] = HWBLK(MSTPCR1, 8, CORE_AREA),
+
+	[HWBLK_MMC] = HWBLK(MSTPCR2, 29, CORE_AREA),
+	[HWBLK_ETHER] = HWBLK(MSTPCR2, 28, CORE_AREA_BM),
+	[HWBLK_ATAPI] = HWBLK(MSTPCR2, 26, CORE_AREA_BM),
+	[HWBLK_TPU] = HWBLK(MSTPCR2, 25, CORE_AREA),
+	[HWBLK_IRDA] = HWBLK(MSTPCR2, 24, CORE_AREA),
+	[HWBLK_TSIF] = HWBLK(MSTPCR2, 22, CORE_AREA),
+	[HWBLK_USB1] = HWBLK(MSTPCR2, 21, CORE_AREA),
+	[HWBLK_USB0] = HWBLK(MSTPCR2, 20, CORE_AREA),
+	[HWBLK_2DG] = HWBLK(MSTPCR2, 19, CORE_AREA_BM),
+	[HWBLK_SDHI0] = HWBLK(MSTPCR2, 18, CORE_AREA),
+	[HWBLK_SDHI1] = HWBLK(MSTPCR2, 17, CORE_AREA),
+	[HWBLK_VEU1] = HWBLK(MSTPCR2, 15, CORE_AREA_BM),
+	[HWBLK_CEU1] = HWBLK(MSTPCR2, 13, CORE_AREA_BM),
+	[HWBLK_BEU1] = HWBLK(MSTPCR2, 12, CORE_AREA_BM),
+	[HWBLK_2DDMAC] = HWBLK(MSTPCR2, 10, CORE_AREA_BM),
+	[HWBLK_SPU] = HWBLK(MSTPCR2, 9, CORE_AREA_BM),
+	[HWBLK_JPU] = HWBLK(MSTPCR2, 6, CORE_AREA_BM),
+	[HWBLK_VOU] = HWBLK(MSTPCR2, 5, CORE_AREA_BM),
+	[HWBLK_BEU0] = HWBLK(MSTPCR2, 4, CORE_AREA_BM),
+	[HWBLK_CEU0] = HWBLK(MSTPCR2, 3, CORE_AREA_BM),
+	[HWBLK_VEU0] = HWBLK(MSTPCR2, 2, CORE_AREA_BM),
+	[HWBLK_VPU] = HWBLK(MSTPCR2, 1, CORE_AREA_BM),
+	[HWBLK_LCDC] = HWBLK(MSTPCR2, 0, CORE_AREA_BM),
+};
+
+static struct hwblk_info sh7724_hwblk_info = {
+	.areas = sh7724_hwblk_area,
+	.nr_areas = ARRAY_SIZE(sh7724_hwblk_area),
+	.hwblks = sh7724_hwblk,
+	.nr_hwblks = ARRAY_SIZE(sh7724_hwblk),
+};
+
+int arch_hwblk_sleep_mode(void)
+{
+	if (!sh7724_hwblk_area[CORE_AREA].cnt[HWBLK_CNT_USAGE])
+		return SUSP_SH_STANDBY | SUSP_SH_SF;
+
+	if (!sh7724_hwblk_area[CORE_AREA_BM].cnt[HWBLK_CNT_USAGE])
+		return SUSP_SH_SLEEP | SUSP_SH_SF;
+
+	return SUSP_SH_SLEEP;
+}
+
+int __init arch_hwblk_init(void)
+{
+	return hwblk_register(&sh7724_hwblk_info);
+}
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7343.c b/arch/sh/kernel/cpu/sh4a/setup-sh7343.c
index 6307e087c86..ac4d5672ec1 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7343.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7343.c
@@ -325,7 +325,7 @@ static int __init sh7343_devices_setup(void)
 	return platform_add_devices(sh7343_devices,
 				    ARRAY_SIZE(sh7343_devices));
 }
-__initcall(sh7343_devices_setup);
+arch_initcall(sh7343_devices_setup);
 
 static struct platform_device *sh7343_early_devices[] __initdata = {
 	&cmt_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7366.c b/arch/sh/kernel/cpu/sh4a/setup-sh7366.c
index f6d20881356..4a9010bf4fd 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7366.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7366.c
@@ -318,7 +318,7 @@ static int __init sh7366_devices_setup(void)
 	return platform_add_devices(sh7366_devices,
 				    ARRAY_SIZE(sh7366_devices));
 }
-__initcall(sh7366_devices_setup);
+arch_initcall(sh7366_devices_setup);
 
 static struct platform_device *sh7366_early_devices[] __initdata = {
 	&cmt_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7722.c b/arch/sh/kernel/cpu/sh4a/setup-sh7722.c
index ea524a2da3e..67b0d87fcb2 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7722.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7722.c
@@ -13,6 +13,7 @@
 #include <linux/serial_sci.h>
 #include <linux/mm.h>
 #include <linux/uio_driver.h>
+#include <linux/usb/m66592.h>
 #include <linux/sh_timer.h>
 #include <asm/clock.h>
 #include <asm/mmzone.h>
@@ -47,9 +48,13 @@ static struct platform_device rtc_device = {
 	.resource	= rtc_resources,
 };
 
+static struct m66592_platdata usbf_platdata = {
+	.on_chip = 1,
+};
+
 static struct resource usbf_resources[] = {
 	[0] = {
-		.name	= "m66592_udc",
+		.name	= "USBF",
 		.start	= 0x04480000,
 		.end	= 0x044800FF,
 		.flags	= IORESOURCE_MEM,
@@ -67,6 +72,7 @@ static struct platform_device usbf_device = {
 	.dev = {
 		.dma_mask		= NULL,
 		.coherent_dma_mask	= 0xffffffff,
+		.platform_data		= &usbf_platdata,
 	},
 	.num_resources	= ARRAY_SIZE(usbf_resources),
 	.resource	= usbf_resources,
@@ -359,7 +365,7 @@ static int __init sh7722_devices_setup(void)
 	return platform_add_devices(sh7722_devices,
 				    ARRAY_SIZE(sh7722_devices));
 }
-__initcall(sh7722_devices_setup);
+arch_initcall(sh7722_devices_setup);
 
 static struct platform_device *sh7722_early_devices[] __initdata = {
 	&cmt_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7723.c b/arch/sh/kernel/cpu/sh4a/setup-sh7723.c
index 28516499a2c..26dc4d32325 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7723.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7723.c
@@ -473,7 +473,7 @@ static int __init sh7723_devices_setup(void)
 	return platform_add_devices(sh7723_devices,
 				    ARRAY_SIZE(sh7723_devices));
 }
-__initcall(sh7723_devices_setup);
+arch_initcall(sh7723_devices_setup);
 
 static struct platform_device *sh7723_early_devices[] __initdata = {
 	&cmt_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7724.c b/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
index e5ac9eb11c6..a04edaab9a2 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
@@ -508,7 +508,7 @@ static int __init sh7724_devices_setup(void)
 	return platform_add_devices(sh7724_devices,
 				    ARRAY_SIZE(sh7724_devices));
 }
-device_initcall(sh7724_devices_setup);
+arch_initcall(sh7724_devices_setup);
 
 static struct platform_device *sh7724_early_devices[] __initdata = {
 	&cmt_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7763.c b/arch/sh/kernel/cpu/sh4a/setup-sh7763.c
index f1e0c0d36da..4659fff6b84 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7763.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7763.c
@@ -314,7 +314,7 @@ static int __init sh7763_devices_setup(void)
 	return platform_add_devices(sh7763_devices,
 				    ARRAY_SIZE(sh7763_devices));
 }
-__initcall(sh7763_devices_setup);
+arch_initcall(sh7763_devices_setup);
 
 static struct platform_device *sh7763_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7770.c b/arch/sh/kernel/cpu/sh4a/setup-sh7770.c
index 1e86209db28..eead08d89d3 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7770.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7770.c
@@ -368,7 +368,7 @@ static int __init sh7770_devices_setup(void)
 	return platform_add_devices(sh7770_devices,
 				    ARRAY_SIZE(sh7770_devices));
 }
-__initcall(sh7770_devices_setup);
+arch_initcall(sh7770_devices_setup);
 
 static struct platform_device *sh7770_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7780.c b/arch/sh/kernel/cpu/sh4a/setup-sh7780.c
index 715e05b431e..2c901f44695 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7780.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7780.c
@@ -256,7 +256,7 @@ static int __init sh7780_devices_setup(void)
 	return platform_add_devices(sh7780_devices,
 				    ARRAY_SIZE(sh7780_devices));
 }
-__initcall(sh7780_devices_setup);
+arch_initcall(sh7780_devices_setup);
 
 static struct platform_device *sh7780_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7785.c b/arch/sh/kernel/cpu/sh4a/setup-sh7785.c
index af561402570..7f6c718b6c3 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7785.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7785.c
@@ -263,7 +263,7 @@ static int __init sh7785_devices_setup(void)
 	return platform_add_devices(sh7785_devices,
 				    ARRAY_SIZE(sh7785_devices));
 }
-__initcall(sh7785_devices_setup);
+arch_initcall(sh7785_devices_setup);
 
 static struct platform_device *sh7785_early_devices[] __initdata = {
 	&tmu0_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7786.c b/arch/sh/kernel/cpu/sh4a/setup-sh7786.c
index b70049470a0..0104a8ec536 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7786.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7786.c
@@ -547,7 +547,7 @@ static int __init sh7786_devices_setup(void)
 	return platform_add_devices(sh7786_devices,
 				    ARRAY_SIZE(sh7786_devices));
 }
-device_initcall(sh7786_devices_setup);
+arch_initcall(sh7786_devices_setup);
 
 void __init plat_early_device_setup(void)
 {
diff --git a/arch/sh/kernel/cpu/sh4a/setup-shx3.c b/arch/sh/kernel/cpu/sh4a/setup-shx3.c
index 53c65fd9cce..07f078961c7 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-shx3.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-shx3.c
@@ -256,7 +256,7 @@ static int __init shx3_devices_setup(void)
 	return platform_add_devices(shx3_devices,
 				    ARRAY_SIZE(shx3_devices));
 }
-__initcall(shx3_devices_setup);
+arch_initcall(shx3_devices_setup);
 
 void __init plat_early_device_setup(void)
 {
diff --git a/arch/sh/kernel/cpu/sh5/setup-sh5.c b/arch/sh/kernel/cpu/sh5/setup-sh5.c
index f5ff1ac57fc..6a0f82f7003 100644
--- a/arch/sh/kernel/cpu/sh5/setup-sh5.c
+++ b/arch/sh/kernel/cpu/sh5/setup-sh5.c
@@ -186,7 +186,7 @@ static int __init sh5_devices_setup(void)
 	return platform_add_devices(sh5_devices,
 				    ARRAY_SIZE(sh5_devices));
 }
-__initcall(sh5_devices_setup);
+arch_initcall(sh5_devices_setup);
 
 void __init plat_early_device_setup(void)
 {
diff --git a/arch/sh/kernel/dumpstack.c b/arch/sh/kernel/dumpstack.c
new file mode 100644
index 00000000000..6f5ad151340
--- /dev/null
+++ b/arch/sh/kernel/dumpstack.c
@@ -0,0 +1,123 @@
+/*
+ *  Copyright (C) 1991, 1992  Linus Torvalds
+ *  Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
+ *  Copyright (C) 2009  Matt Fleming
+ */
+#include <linux/kallsyms.h>
+#include <linux/ftrace.h>
+#include <linux/debug_locks.h>
+#include <asm/unwinder.h>
+#include <asm/stacktrace.h>
+
+void printk_address(unsigned long address, int reliable)
+{
+	printk(" [<%p>] %s%pS\n", (void *) address,
+			reliable ? "" : "? ", (void *) address);
+}
+
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+static void
+print_ftrace_graph_addr(unsigned long addr, void *data,
+			const struct stacktrace_ops *ops,
+			struct thread_info *tinfo, int *graph)
+{
+	struct task_struct *task = tinfo->task;
+	unsigned long ret_addr;
+	int index = task->curr_ret_stack;
+
+	if (addr != (unsigned long)return_to_handler)
+		return;
+
+	if (!task->ret_stack || index < *graph)
+		return;
+
+	index -= *graph;
+	ret_addr = task->ret_stack[index].ret;
+
+	ops->address(data, ret_addr, 1);
+
+	(*graph)++;
+}
+#else
+static inline void
+print_ftrace_graph_addr(unsigned long addr, void *data,
+			const struct stacktrace_ops *ops,
+			struct thread_info *tinfo, int *graph)
+{ }
+#endif
+
+void
+stack_reader_dump(struct task_struct *task, struct pt_regs *regs,
+		  unsigned long *sp, const struct stacktrace_ops *ops,
+		  void *data)
+{
+	struct thread_info *context;
+	int graph = 0;
+
+	context = (struct thread_info *)
+		((unsigned long)sp & (~(THREAD_SIZE - 1)));
+
+	while (!kstack_end(sp)) {
+		unsigned long addr = *sp++;
+
+		if (__kernel_text_address(addr)) {
+			ops->address(data, addr, 1);
+
+			print_ftrace_graph_addr(addr, data, ops,
+						context, &graph);
+		}
+	}
+}
+
+static void
+print_trace_warning_symbol(void *data, char *msg, unsigned long symbol)
+{
+	printk(data);
+	print_symbol(msg, symbol);
+	printk("\n");
+}
+
+static void print_trace_warning(void *data, char *msg)
+{
+	printk("%s%s\n", (char *)data, msg);
+}
+
+static int print_trace_stack(void *data, char *name)
+{
+	printk("%s <%s> ", (char *)data, name);
+	return 0;
+}
+
+/*
+ * Print one address/symbol entries per line.
+ */
+static void print_trace_address(void *data, unsigned long addr, int reliable)
+{
+	printk(data);
+	printk_address(addr, reliable);
+}
+
+static const struct stacktrace_ops print_trace_ops = {
+	.warning = print_trace_warning,
+	.warning_symbol = print_trace_warning_symbol,
+	.stack = print_trace_stack,
+	.address = print_trace_address,
+};
+
+void show_trace(struct task_struct *tsk, unsigned long *sp,
+		struct pt_regs *regs)
+{
+	if (regs && user_mode(regs))
+		return;
+
+	printk("\nCall trace:\n");
+
+	unwind_stack(tsk, regs, sp, &print_trace_ops, "");
+
+	printk("\n");
+
+	if (!tsk)
+		tsk = current;
+
+	debug_show_held_locks(tsk);
+}
diff --git a/arch/sh/kernel/dwarf.c b/arch/sh/kernel/dwarf.c
new file mode 100644
index 00000000000..db021361b16
--- /dev/null
+++ b/arch/sh/kernel/dwarf.c
@@ -0,0 +1,901 @@
+/*
+ * Copyright (C) 2009 Matt Fleming <matt@console-pimps.org>
+ *
+ * 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.
+ *
+ * This is an implementation of a DWARF unwinder. Its main purpose is
+ * for generating stacktrace information. Based on the DWARF 3
+ * specification from http://www.dwarfstd.org.
+ *
+ * TODO:
+ *	- DWARF64 doesn't work.
+ */
+
+/* #define DEBUG */
+#include <linux/kernel.h>
+#include <linux/io.h>
+#include <linux/list.h>
+#include <linux/mm.h>
+#include <asm/dwarf.h>
+#include <asm/unwinder.h>
+#include <asm/sections.h>
+#include <asm/unaligned.h>
+#include <asm/dwarf.h>
+#include <asm/stacktrace.h>
+
+static LIST_HEAD(dwarf_cie_list);
+DEFINE_SPINLOCK(dwarf_cie_lock);
+
+static LIST_HEAD(dwarf_fde_list);
+DEFINE_SPINLOCK(dwarf_fde_lock);
+
+static struct dwarf_cie *cached_cie;
+
+/*
+ * Figure out whether we need to allocate some dwarf registers. If dwarf
+ * registers have already been allocated then we may need to realloc
+ * them. "reg" is a register number that we need to be able to access
+ * after this call.
+ *
+ * Register numbers start at zero, therefore we need to allocate space
+ * for "reg" + 1 registers.
+ */
+static void dwarf_frame_alloc_regs(struct dwarf_frame *frame,
+				   unsigned int reg)
+{
+	struct dwarf_reg *regs;
+	unsigned int num_regs = reg + 1;
+	size_t new_size;
+	size_t old_size;
+
+	new_size = num_regs * sizeof(*regs);
+	old_size = frame->num_regs * sizeof(*regs);
+
+	/* Fast path: don't allocate any regs if we've already got enough. */
+	if (frame->num_regs >= num_regs)
+		return;
+
+	regs = kzalloc(new_size, GFP_ATOMIC);
+	if (!regs) {
+		printk(KERN_WARNING "Unable to allocate DWARF registers\n");
+		/*
+		 * Let's just bomb hard here, we have no way to
+		 * gracefully recover.
+		 */
+		BUG();
+	}
+
+	if (frame->regs) {
+		memcpy(regs, frame->regs, old_size);
+		kfree(frame->regs);
+	}
+
+	frame->regs = regs;
+	frame->num_regs = num_regs;
+}
+
+/**
+ *	dwarf_read_addr - read dwarf data
+ *	@src: source address of data
+ *	@dst: destination address to store the data to
+ *
+ *	Read 'n' bytes from @src, where 'n' is the size of an address on
+ *	the native machine. We return the number of bytes read, which
+ *	should always be 'n'. We also have to be careful when reading
+ *	from @src and writing to @dst, because they can be arbitrarily
+ *	aligned. Return 'n' - the number of bytes read.
+ */
+static inline int dwarf_read_addr(unsigned long *src, unsigned long *dst)
+{
+	u32 val = get_unaligned(src);
+	put_unaligned(val, dst);
+	return sizeof(unsigned long *);
+}
+
+/**
+ *	dwarf_read_uleb128 - read unsigned LEB128 data
+ *	@addr: the address where the ULEB128 data is stored
+ *	@ret: address to store the result
+ *
+ *	Decode an unsigned LEB128 encoded datum. The algorithm is taken
+ *	from Appendix C of the DWARF 3 spec. For information on the
+ *	encodings refer to section "7.6 - Variable Length Data". Return
+ *	the number of bytes read.
+ */
+static inline unsigned long dwarf_read_uleb128(char *addr, unsigned int *ret)
+{
+	unsigned int result;
+	unsigned char byte;
+	int shift, count;
+
+	result = 0;
+	shift = 0;
+	count = 0;
+
+	while (1) {
+		byte = __raw_readb(addr);
+		addr++;
+		count++;
+
+		result |= (byte & 0x7f) << shift;
+		shift += 7;
+
+		if (!(byte & 0x80))
+			break;
+	}
+
+	*ret = result;
+
+	return count;
+}
+
+/**
+ *	dwarf_read_leb128 - read signed LEB128 data
+ *	@addr: the address of the LEB128 encoded data
+ *	@ret: address to store the result
+ *
+ *	Decode signed LEB128 data. The algorithm is taken from Appendix
+ *	C of the DWARF 3 spec. Return the number of bytes read.
+ */
+static inline unsigned long dwarf_read_leb128(char *addr, int *ret)
+{
+	unsigned char byte;
+	int result, shift;
+	int num_bits;
+	int count;
+
+	result = 0;
+	shift = 0;
+	count = 0;
+
+	while (1) {
+		byte = __raw_readb(addr);
+		addr++;
+		result |= (byte & 0x7f) << shift;
+		shift += 7;
+		count++;
+
+		if (!(byte & 0x80))
+			break;
+	}
+
+	/* The number of bits in a signed integer. */
+	num_bits = 8 * sizeof(result);
+
+	if ((shift < num_bits) && (byte & 0x40))
+		result |= (-1 << shift);
+
+	*ret = result;
+
+	return count;
+}
+
+/**
+ *	dwarf_read_encoded_value - return the decoded value at @addr
+ *	@addr: the address of the encoded value
+ *	@val: where to write the decoded value
+ *	@encoding: the encoding with which we can decode @addr
+ *
+ *	GCC emits encoded address in the .eh_frame FDE entries. Decode
+ *	the value at @addr using @encoding. The decoded value is written
+ *	to @val and the number of bytes read is returned.
+ */
+static int dwarf_read_encoded_value(char *addr, unsigned long *val,
+				    char encoding)
+{
+	unsigned long decoded_addr = 0;
+	int count = 0;
+
+	switch (encoding & 0x70) {
+	case DW_EH_PE_absptr:
+		break;
+	case DW_EH_PE_pcrel:
+		decoded_addr = (unsigned long)addr;
+		break;
+	default:
+		pr_debug("encoding=0x%x\n", (encoding & 0x70));
+		BUG();
+	}
+
+	if ((encoding & 0x07) == 0x00)
+		encoding |= DW_EH_PE_udata4;
+
+	switch (encoding & 0x0f) {
+	case DW_EH_PE_sdata4:
+	case DW_EH_PE_udata4:
+		count += 4;
+		decoded_addr += get_unaligned((u32 *)addr);
+		__raw_writel(decoded_addr, val);
+		break;
+	default:
+		pr_debug("encoding=0x%x\n", encoding);
+		BUG();
+	}
+
+	return count;
+}
+
+/**
+ *	dwarf_entry_len - return the length of an FDE or CIE
+ *	@addr: the address of the entry
+ *	@len: the length of the entry
+ *
+ *	Read the initial_length field of the entry and store the size of
+ *	the entry in @len. We return the number of bytes read. Return a
+ *	count of 0 on error.
+ */
+static inline int dwarf_entry_len(char *addr, unsigned long *len)
+{
+	u32 initial_len;
+	int count;
+
+	initial_len = get_unaligned((u32 *)addr);
+	count = 4;
+
+	/*
+	 * An initial length field value in the range DW_LEN_EXT_LO -
+	 * DW_LEN_EXT_HI indicates an extension, and should not be
+	 * interpreted as a length. The only extension that we currently
+	 * understand is the use of DWARF64 addresses.
+	 */
+	if (initial_len >= DW_EXT_LO && initial_len <= DW_EXT_HI) {
+		/*
+		 * The 64-bit length field immediately follows the
+		 * compulsory 32-bit length field.
+		 */
+		if (initial_len == DW_EXT_DWARF64) {
+			*len = get_unaligned((u64 *)addr + 4);
+			count = 12;
+		} else {
+			printk(KERN_WARNING "Unknown DWARF extension\n");
+			count = 0;
+		}
+	} else
+		*len = initial_len;
+
+	return count;
+}
+
+/**
+ *	dwarf_lookup_cie - locate the cie
+ *	@cie_ptr: pointer to help with lookup
+ */
+static struct dwarf_cie *dwarf_lookup_cie(unsigned long cie_ptr)
+{
+	struct dwarf_cie *cie, *n;
+	unsigned long flags;
+
+	spin_lock_irqsave(&dwarf_cie_lock, flags);
+
+	/*
+	 * We've cached the last CIE we looked up because chances are
+	 * that the FDE wants this CIE.
+	 */
+	if (cached_cie && cached_cie->cie_pointer == cie_ptr) {
+		cie = cached_cie;
+		goto out;
+	}
+
+	list_for_each_entry_safe(cie, n, &dwarf_cie_list, link) {
+		if (cie->cie_pointer == cie_ptr) {
+			cached_cie = cie;
+			break;
+		}
+	}
+
+	/* Couldn't find the entry in the list. */
+	if (&cie->link == &dwarf_cie_list)
+		cie = NULL;
+out:
+	spin_unlock_irqrestore(&dwarf_cie_lock, flags);
+	return cie;
+}
+
+/**
+ *	dwarf_lookup_fde - locate the FDE that covers pc
+ *	@pc: the program counter
+ */
+struct dwarf_fde *dwarf_lookup_fde(unsigned long pc)
+{
+	unsigned long flags;
+	struct dwarf_fde *fde, *n;
+
+	spin_lock_irqsave(&dwarf_fde_lock, flags);
+	list_for_each_entry_safe(fde, n, &dwarf_fde_list, link) {
+		unsigned long start, end;
+
+		start = fde->initial_location;
+		end = fde->initial_location + fde->address_range;
+
+		if (pc >= start && pc < end)
+			break;
+	}
+
+	/* Couldn't find the entry in the list. */
+	if (&fde->link == &dwarf_fde_list)
+		fde = NULL;
+
+	spin_unlock_irqrestore(&dwarf_fde_lock, flags);
+
+	return fde;
+}
+
+/**
+ *	dwarf_cfa_execute_insns - execute instructions to calculate a CFA
+ *	@insn_start: address of the first instruction
+ *	@insn_end: address of the last instruction
+ *	@cie: the CIE for this function
+ *	@fde: the FDE for this function
+ *	@frame: the instructions calculate the CFA for this frame
+ *	@pc: the program counter of the address we're interested in
+ *	@define_ra: keep executing insns until the return addr reg is defined?
+ *
+ *	Execute the Call Frame instruction sequence starting at
+ *	@insn_start and ending at @insn_end. The instructions describe
+ *	how to calculate the Canonical Frame Address of a stackframe.
+ *	Store the results in @frame.
+ */
+static int dwarf_cfa_execute_insns(unsigned char *insn_start,
+				   unsigned char *insn_end,
+				   struct dwarf_cie *cie,
+				   struct dwarf_fde *fde,
+				   struct dwarf_frame *frame,
+				   unsigned long pc,
+				   bool define_ra)
+{
+	unsigned char insn;
+	unsigned char *current_insn;
+	unsigned int count, delta, reg, expr_len, offset;
+	bool seen_ra_reg;
+
+	current_insn = insn_start;
+
+	/*
+	 * If we're executing instructions for the dwarf_unwind_stack()
+	 * FDE we need to keep executing instructions until the value of
+	 * DWARF_ARCH_RA_REG is defined. See the comment in
+	 * dwarf_unwind_stack() for more details.
+	 */
+	if (define_ra)
+		seen_ra_reg = false;
+	else
+		seen_ra_reg = true;
+
+	while (current_insn < insn_end && (frame->pc <= pc || !seen_ra_reg) ) {
+		insn = __raw_readb(current_insn++);
+
+		if (!seen_ra_reg) {
+			if (frame->num_regs >= DWARF_ARCH_RA_REG &&
+			    frame->regs[DWARF_ARCH_RA_REG].flags)
+				seen_ra_reg = true;
+		}
+
+		/*
+		 * Firstly, handle the opcodes that embed their operands
+		 * in the instructions.
+		 */
+		switch (DW_CFA_opcode(insn)) {
+		case DW_CFA_advance_loc:
+			delta = DW_CFA_operand(insn);
+			delta *= cie->code_alignment_factor;
+			frame->pc += delta;
+			continue;
+			/* NOTREACHED */
+		case DW_CFA_offset:
+			reg = DW_CFA_operand(insn);
+			count = dwarf_read_uleb128(current_insn, &offset);
+			current_insn += count;
+			offset *= cie->data_alignment_factor;
+			dwarf_frame_alloc_regs(frame, reg);
+			frame->regs[reg].addr = offset;
+			frame->regs[reg].flags |= DWARF_REG_OFFSET;
+			continue;
+			/* NOTREACHED */
+		case DW_CFA_restore:
+			reg = DW_CFA_operand(insn);
+			continue;
+			/* NOTREACHED */
+		}
+
+		/*
+		 * Secondly, handle the opcodes that don't embed their
+		 * operands in the instruction.
+		 */
+		switch (insn) {
+		case DW_CFA_nop:
+			continue;
+		case DW_CFA_advance_loc1:
+			delta = *current_insn++;
+			frame->pc += delta * cie->code_alignment_factor;
+			break;
+		case DW_CFA_advance_loc2:
+			delta = get_unaligned((u16 *)current_insn);
+			current_insn += 2;
+			frame->pc += delta * cie->code_alignment_factor;
+			break;
+		case DW_CFA_advance_loc4:
+			delta = get_unaligned((u32 *)current_insn);
+			current_insn += 4;
+			frame->pc += delta * cie->code_alignment_factor;
+			break;
+		case DW_CFA_offset_extended:
+			count = dwarf_read_uleb128(current_insn, &reg);
+			current_insn += count;
+			count = dwarf_read_uleb128(current_insn, &offset);
+			current_insn += count;
+			offset *= cie->data_alignment_factor;
+			break;
+		case DW_CFA_restore_extended:
+			count = dwarf_read_uleb128(current_insn, &reg);
+			current_insn += count;
+			break;
+		case DW_CFA_undefined:
+			count = dwarf_read_uleb128(current_insn, &reg);
+			current_insn += count;
+			break;
+		case DW_CFA_def_cfa:
+			count = dwarf_read_uleb128(current_insn,
+						   &frame->cfa_register);
+			current_insn += count;
+			count = dwarf_read_uleb128(current_insn,
+						   &frame->cfa_offset);
+			current_insn += count;
+
+			frame->flags |= DWARF_FRAME_CFA_REG_OFFSET;
+			break;
+		case DW_CFA_def_cfa_register:
+			count = dwarf_read_uleb128(current_insn,
+						   &frame->cfa_register);
+			current_insn += count;
+			frame->flags |= DWARF_FRAME_CFA_REG_OFFSET;
+			break;
+		case DW_CFA_def_cfa_offset:
+			count = dwarf_read_uleb128(current_insn, &offset);
+			current_insn += count;
+			frame->cfa_offset = offset;
+			break;
+		case DW_CFA_def_cfa_expression:
+			count = dwarf_read_uleb128(current_insn, &expr_len);
+			current_insn += count;
+
+			frame->cfa_expr = current_insn;
+			frame->cfa_expr_len = expr_len;
+			current_insn += expr_len;
+
+			frame->flags |= DWARF_FRAME_CFA_REG_EXP;
+			break;
+		case DW_CFA_offset_extended_sf:
+			count = dwarf_read_uleb128(current_insn, &reg);
+			current_insn += count;
+			count = dwarf_read_leb128(current_insn, &offset);
+			current_insn += count;
+			offset *= cie->data_alignment_factor;
+			dwarf_frame_alloc_regs(frame, reg);
+			frame->regs[reg].flags |= DWARF_REG_OFFSET;
+			frame->regs[reg].addr = offset;
+			break;
+		case DW_CFA_val_offset:
+			count = dwarf_read_uleb128(current_insn, &reg);
+			current_insn += count;
+			count = dwarf_read_leb128(current_insn, &offset);
+			offset *= cie->data_alignment_factor;
+			frame->regs[reg].flags |= DWARF_REG_OFFSET;
+			frame->regs[reg].addr = offset;
+			break;
+		default:
+			pr_debug("unhandled DWARF instruction 0x%x\n", insn);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ *	dwarf_unwind_stack - recursively unwind the stack
+ *	@pc: address of the function to unwind
+ *	@prev: struct dwarf_frame of the previous stackframe on the callstack
+ *
+ *	Return a struct dwarf_frame representing the most recent frame
+ *	on the callstack. Each of the lower (older) stack frames are
+ *	linked via the "prev" member.
+ */
+struct dwarf_frame *dwarf_unwind_stack(unsigned long pc,
+				       struct dwarf_frame *prev)
+{
+	struct dwarf_frame *frame;
+	struct dwarf_cie *cie;
+	struct dwarf_fde *fde;
+	unsigned long addr;
+	int i, offset;
+	bool define_ra = false;
+
+	/*
+	 * If this is the first invocation of this recursive function we
+	 * need get the contents of a physical register to get the CFA
+	 * in order to begin the virtual unwinding of the stack.
+	 *
+	 * Setting "define_ra" to true indictates that we want
+	 * dwarf_cfa_execute_insns() to continue executing instructions
+	 * until we know how to calculate the value of DWARF_ARCH_RA_REG
+	 * (which we need in order to kick off the whole unwinding
+	 * process).
+	 *
+	 * NOTE: the return address is guaranteed to be setup by the
+	 * time this function makes its first function call.
+	 */
+	if (!pc && !prev) {
+		pc = (unsigned long)&dwarf_unwind_stack;
+		define_ra = true;
+	}
+
+	frame = kzalloc(sizeof(*frame), GFP_ATOMIC);
+	if (!frame)
+		return NULL;
+
+	frame->prev = prev;
+
+	fde = dwarf_lookup_fde(pc);
+	if (!fde) {
+		/*
+		 * This is our normal exit path - the one that stops the
+		 * recursion. There's two reasons why we might exit
+		 * here,
+		 *
+		 *	a) pc has no asscociated DWARF frame info and so
+		 *	we don't know how to unwind this frame. This is
+		 *	usually the case when we're trying to unwind a
+		 *	frame that was called from some assembly code
+		 *	that has no DWARF info, e.g. syscalls.
+		 *
+		 *	b) the DEBUG info for pc is bogus. There's
+		 *	really no way to distinguish this case from the
+		 *	case above, which sucks because we could print a
+		 *	warning here.
+		 */
+		return NULL;
+	}
+
+	cie = dwarf_lookup_cie(fde->cie_pointer);
+
+	frame->pc = fde->initial_location;
+
+	/* CIE initial instructions */
+	dwarf_cfa_execute_insns(cie->initial_instructions,
+				cie->instructions_end, cie, fde,
+				frame, pc, false);
+
+	/* FDE instructions */
+	dwarf_cfa_execute_insns(fde->instructions, fde->end, cie,
+				fde, frame, pc, define_ra);
+
+	/* Calculate the CFA */
+	switch (frame->flags) {
+	case DWARF_FRAME_CFA_REG_OFFSET:
+		if (prev) {
+			BUG_ON(!prev->regs[frame->cfa_register].flags);
+
+			addr = prev->cfa;
+			addr += prev->regs[frame->cfa_register].addr;
+			frame->cfa = __raw_readl(addr);
+
+		} else {
+			/*
+			 * Again, this is the first invocation of this
+			 * recurisve function. We need to physically
+			 * read the contents of a register in order to
+			 * get the Canonical Frame Address for this
+			 * function.
+			 */
+			frame->cfa = dwarf_read_arch_reg(frame->cfa_register);
+		}
+
+		frame->cfa += frame->cfa_offset;
+		break;
+	default:
+		BUG();
+	}
+
+	/* If we haven't seen the return address reg, we're screwed. */
+	BUG_ON(!frame->regs[DWARF_ARCH_RA_REG].flags);
+
+	for (i = 0; i <= frame->num_regs; i++) {
+		struct dwarf_reg *reg = &frame->regs[i];
+
+		if (!reg->flags)
+			continue;
+
+		offset = reg->addr;
+		offset += frame->cfa;
+	}
+
+	addr = frame->cfa + frame->regs[DWARF_ARCH_RA_REG].addr;
+	frame->return_addr = __raw_readl(addr);
+
+	frame->next = dwarf_unwind_stack(frame->return_addr, frame);
+	return frame;
+}
+
+static int dwarf_parse_cie(void *entry, void *p, unsigned long len,
+			   unsigned char *end)
+{
+	struct dwarf_cie *cie;
+	unsigned long flags;
+	int count;
+
+	cie = kzalloc(sizeof(*cie), GFP_KERNEL);
+	if (!cie)
+		return -ENOMEM;
+
+	cie->length = len;
+
+	/*
+	 * Record the offset into the .eh_frame section
+	 * for this CIE. It allows this CIE to be
+	 * quickly and easily looked up from the
+	 * corresponding FDE.
+	 */
+	cie->cie_pointer = (unsigned long)entry;
+
+	cie->version = *(char *)p++;
+	BUG_ON(cie->version != 1);
+
+	cie->augmentation = p;
+	p += strlen(cie->augmentation) + 1;
+
+	count = dwarf_read_uleb128(p, &cie->code_alignment_factor);
+	p += count;
+
+	count = dwarf_read_leb128(p, &cie->data_alignment_factor);
+	p += count;
+
+	/*
+	 * Which column in the rule table contains the
+	 * return address?
+	 */
+	if (cie->version == 1) {
+		cie->return_address_reg = __raw_readb(p);
+		p++;
+	} else {
+		count = dwarf_read_uleb128(p, &cie->return_address_reg);
+		p += count;
+	}
+
+	if (cie->augmentation[0] == 'z') {
+		unsigned int length, count;
+		cie->flags |= DWARF_CIE_Z_AUGMENTATION;
+
+		count = dwarf_read_uleb128(p, &length);
+		p += count;
+
+		BUG_ON((unsigned char *)p > end);
+
+		cie->initial_instructions = p + length;
+		cie->augmentation++;
+	}
+
+	while (*cie->augmentation) {
+		/*
+		 * "L" indicates a byte showing how the
+		 * LSDA pointer is encoded. Skip it.
+		 */
+		if (*cie->augmentation == 'L') {
+			p++;
+			cie->augmentation++;
+		} else if (*cie->augmentation == 'R') {
+			/*
+			 * "R" indicates a byte showing
+			 * how FDE addresses are
+			 * encoded.
+			 */
+			cie->encoding = *(char *)p++;
+			cie->augmentation++;
+		} else if (*cie->augmentation == 'P') {
+			/*
+			 * "R" indicates a personality
+			 * routine in the CIE
+			 * augmentation.
+			 */
+			BUG();
+		} else if (*cie->augmentation == 'S') {
+			BUG();
+		} else {
+			/*
+			 * Unknown augmentation. Assume
+			 * 'z' augmentation.
+			 */
+			p = cie->initial_instructions;
+			BUG_ON(!p);
+			break;
+		}
+	}
+
+	cie->initial_instructions = p;
+	cie->instructions_end = end;
+
+	/* Add to list */
+	spin_lock_irqsave(&dwarf_cie_lock, flags);
+	list_add_tail(&cie->link, &dwarf_cie_list);
+	spin_unlock_irqrestore(&dwarf_cie_lock, flags);
+
+	return 0;
+}
+
+static int dwarf_parse_fde(void *entry, u32 entry_type,
+			   void *start, unsigned long len)
+{
+	struct dwarf_fde *fde;
+	struct dwarf_cie *cie;
+	unsigned long flags;
+	int count;
+	void *p = start;
+
+	fde = kzalloc(sizeof(*fde), GFP_KERNEL);
+	if (!fde)
+		return -ENOMEM;
+
+	fde->length = len;
+
+	/*
+	 * In a .eh_frame section the CIE pointer is the
+	 * delta between the address within the FDE
+	 */
+	fde->cie_pointer = (unsigned long)(p - entry_type - 4);
+
+	cie = dwarf_lookup_cie(fde->cie_pointer);
+	fde->cie = cie;
+
+	if (cie->encoding)
+		count = dwarf_read_encoded_value(p, &fde->initial_location,
+						 cie->encoding);
+	else
+		count = dwarf_read_addr(p, &fde->initial_location);
+
+	p += count;
+
+	if (cie->encoding)
+		count = dwarf_read_encoded_value(p, &fde->address_range,
+						 cie->encoding & 0x0f);
+	else
+		count = dwarf_read_addr(p, &fde->address_range);
+
+	p += count;
+
+	if (fde->cie->flags & DWARF_CIE_Z_AUGMENTATION) {
+		unsigned int length;
+		count = dwarf_read_uleb128(p, &length);
+		p += count + length;
+	}
+
+	/* Call frame instructions. */
+	fde->instructions = p;
+	fde->end = start + len;
+
+	/* Add to list. */
+	spin_lock_irqsave(&dwarf_fde_lock, flags);
+	list_add_tail(&fde->link, &dwarf_fde_list);
+	spin_unlock_irqrestore(&dwarf_fde_lock, flags);
+
+	return 0;
+}
+
+static void dwarf_unwinder_dump(struct task_struct *task, struct pt_regs *regs,
+				unsigned long *sp,
+				const struct stacktrace_ops *ops, void *data)
+{
+	struct dwarf_frame *frame;
+
+	frame = dwarf_unwind_stack(0, NULL);
+
+	while (frame && frame->return_addr) {
+		ops->address(data, frame->return_addr, 1);
+		frame = frame->next;
+	}
+}
+
+static struct unwinder dwarf_unwinder = {
+	.name = "dwarf-unwinder",
+	.dump = dwarf_unwinder_dump,
+	.rating = 150,
+};
+
+static void dwarf_unwinder_cleanup(void)
+{
+	struct dwarf_cie *cie, *m;
+	struct dwarf_fde *fde, *n;
+	unsigned long flags;
+
+	/*
+	 * Deallocate all the memory allocated for the DWARF unwinder.
+	 * Traverse all the FDE/CIE lists and remove and free all the
+	 * memory associated with those data structures.
+	 */
+	spin_lock_irqsave(&dwarf_cie_lock, flags);
+	list_for_each_entry_safe(cie, m, &dwarf_cie_list, link)
+		kfree(cie);
+	spin_unlock_irqrestore(&dwarf_cie_lock, flags);
+
+	spin_lock_irqsave(&dwarf_fde_lock, flags);
+	list_for_each_entry_safe(fde, n, &dwarf_fde_list, link)
+		kfree(fde);
+	spin_unlock_irqrestore(&dwarf_fde_lock, flags);
+}
+
+/**
+ *	dwarf_unwinder_init - initialise the dwarf unwinder
+ *
+ *	Build the data structures describing the .dwarf_frame section to
+ *	make it easier to lookup CIE and FDE entries. Because the
+ *	.eh_frame section is packed as tightly as possible it is not
+ *	easy to lookup the FDE for a given PC, so we build a list of FDE
+ *	and CIE entries that make it easier.
+ */
+void dwarf_unwinder_init(void)
+{
+	u32 entry_type;
+	void *p, *entry;
+	int count, err;
+	unsigned long len;
+	unsigned int c_entries, f_entries;
+	unsigned char *end;
+	INIT_LIST_HEAD(&dwarf_cie_list);
+	INIT_LIST_HEAD(&dwarf_fde_list);
+
+	c_entries = 0;
+	f_entries = 0;
+	entry = &__start_eh_frame;
+
+	while ((char *)entry < __stop_eh_frame) {
+		p = entry;
+
+		count = dwarf_entry_len(p, &len);
+		if (count == 0) {
+			/*
+			 * We read a bogus length field value. There is
+			 * nothing we can do here apart from disabling
+			 * the DWARF unwinder. We can't even skip this
+			 * entry and move to the next one because 'len'
+			 * tells us where our next entry is.
+			 */
+			goto out;
+		} else
+			p += count;
+
+		/* initial length does not include itself */
+		end = p + len;
+
+		entry_type = get_unaligned((u32 *)p);
+		p += 4;
+
+		if (entry_type == DW_EH_FRAME_CIE) {
+			err = dwarf_parse_cie(entry, p, len, end);
+			if (err < 0)
+				goto out;
+			else
+				c_entries++;
+		} else {
+			err = dwarf_parse_fde(entry, entry_type, p, len);
+			if (err < 0)
+				goto out;
+			else
+				f_entries++;
+		}
+
+		entry = (char *)entry + len + 4;
+	}
+
+	printk(KERN_INFO "DWARF unwinder initialised: read %u CIEs, %u FDEs\n",
+	       c_entries, f_entries);
+
+	err = unwinder_register(&dwarf_unwinder);
+	if (err)
+		goto out;
+
+	return;
+
+out:
+	printk(KERN_ERR "Failed to initialise DWARF unwinder: %d\n", err);
+	dwarf_unwinder_cleanup();
+}
diff --git a/arch/sh/kernel/early_printk.c b/arch/sh/kernel/early_printk.c
index a952dcf9999..81a46145ffa 100644
--- a/arch/sh/kernel/early_printk.c
+++ b/arch/sh/kernel/early_printk.c
@@ -134,7 +134,7 @@ static void scif_sercon_init(char *s)
 	sci_out(&scif_port, SCFCR, 0x0030);	/* TTRG=b'11 */
 	sci_out(&scif_port, SCSCR, 0x0030);	/* TE, RE */
 }
-#elif defined(CONFIG_CPU_SH4)
+#elif defined(CONFIG_CPU_SH4) || defined(CONFIG_CPU_SH3)
 #define DEFAULT_BAUD 115200
 /*
  * Simple SCIF init, primarily aimed at SH7750 and other similar SH-4
@@ -220,8 +220,7 @@ static int __init setup_early_printk(char *buf)
 		early_console = &scif_console;
 
 #if !defined(CONFIG_SH_STANDARD_BIOS)
-#if defined(CONFIG_CPU_SH4) || defined(CONFIG_CPU_SUBTYPE_SH7720) || \
-    defined(CONFIG_CPU_SUBTYPE_SH7721)
+#if defined(CONFIG_CPU_SH4) || defined(CONFIG_CPU_SH3)
 		scif_sercon_init(buf + 6);
 #endif
 #endif
diff --git a/arch/sh/kernel/entry-common.S b/arch/sh/kernel/entry-common.S
index d62175650c5..e63178fefb9 100644
--- a/arch/sh/kernel/entry-common.S
+++ b/arch/sh/kernel/entry-common.S
@@ -43,9 +43,10 @@
  *	syscall #
  *
  */
+#include <asm/dwarf.h>
 
 #if defined(CONFIG_PREEMPT)
-#  define preempt_stop()	cli
+#  define preempt_stop()	cli ; TRACE_IRQS_OFF
 #else
 #  define preempt_stop()
 #  define resume_kernel		__restore_all
@@ -55,11 +56,7 @@
 	.align	2
 ENTRY(exception_error)
 	!
-#ifdef CONFIG_TRACE_IRQFLAGS
-	mov.l	2f, r0
-	jsr	@r0
-	 nop
-#endif
+	TRACE_IRQS_ON
 	sti
 	mov.l	1f, r0
 	jmp	@r0
@@ -67,22 +64,28 @@ ENTRY(exception_error)
 
 	.align	2
 1:	.long	do_exception_error
-#ifdef CONFIG_TRACE_IRQFLAGS
-2:	.long	trace_hardirqs_on
-#endif
 
 	.align	2
 ret_from_exception:
+	CFI_STARTPROC simple
+	CFI_DEF_CFA r14, 0
+	CFI_REL_OFFSET 17, 64
+	CFI_REL_OFFSET 15, 0
+	CFI_REL_OFFSET 14, 56
 	preempt_stop()
-#ifdef CONFIG_TRACE_IRQFLAGS
-	mov.l	4f, r0
-	jsr	@r0
-	 nop
-#endif
 ENTRY(ret_from_irq)
 	!
 	mov	#OFF_SR, r0
 	mov.l	@(r0,r15), r0	! get status register
+
+	shlr2	r0
+	and	#0x3c, r0
+	cmp/eq	#0x3c, r0
+	bt	9f
+	TRACE_IRQS_ON
+9:
+	mov	#OFF_SR, r0
+	mov.l	@(r0,r15), r0	! get status register
 	shll	r0
 	shll	r0		! kernel space?
 	get_current_thread_info r8, r0
@@ -125,11 +128,7 @@ noresched:
 ENTRY(resume_userspace)
 	! r8: current_thread_info
 	cli
-#ifdef CONFIG_TRACE_IRQFLAGS
-	mov.l	5f, r0
-	jsr	@r0
-	 nop
-#endif
+	TRACE_IRQS_OfF
 	mov.l	@(TI_FLAGS,r8), r0		! current_thread_info->flags
 	tst	#(_TIF_WORK_MASK & 0xff), r0
 	bt/s	__restore_all
@@ -156,11 +155,7 @@ work_resched:
 	jsr	@r1				! schedule
 	 nop
 	cli
-#ifdef CONFIG_TRACE_IRQFLAGS
-	mov.l	5f, r0
-	jsr	@r0
-	 nop
-#endif
+	TRACE_IRQS_OFF
 	!
 	mov.l	@(TI_FLAGS,r8), r0		! current_thread_info->flags
 	tst	#(_TIF_WORK_MASK & 0xff), r0
@@ -172,10 +167,6 @@ work_resched:
 1:	.long	schedule
 2:	.long	do_notify_resume
 3:	.long	resume_userspace
-#ifdef CONFIG_TRACE_IRQFLAGS
-4:	.long	trace_hardirqs_on
-5:	.long	trace_hardirqs_off
-#endif
 
 	.align	2
 syscall_exit_work:
@@ -184,11 +175,7 @@ syscall_exit_work:
 	tst	#(_TIF_WORK_SYSCALL_MASK & 0xff), r0
 	bt/s	work_pending
 	 tst	#_TIF_NEED_RESCHED, r0
-#ifdef CONFIG_TRACE_IRQFLAGS
-	mov.l	5f, r0
-	jsr	@r0
-	 nop
-#endif
+	TRACE_IRQS_ON
 	sti
 	mov	r15, r4
 	mov.l	8f, r0			! do_syscall_trace_leave
@@ -259,6 +246,7 @@ debug_trap:
 	 nop
 	bra	__restore_all
 	 nop
+	CFI_ENDPROC
 
 	.align	2
 1:	.long	debug_trap_table
@@ -304,6 +292,7 @@ ret_from_fork:
  * system calls and debug traps through their respective jump tables.
  */
 ENTRY(system_call)
+	setup_frame_reg
 #if !defined(CONFIG_CPU_SH2)
 	mov.l	1f, r9
 	mov.l	@r9, r8		! Read from TRA (Trap Address) Register
@@ -321,11 +310,7 @@ ENTRY(system_call)
 	bt/s	debug_trap		! it's a debug trap..
 	 nop
 
-#ifdef CONFIG_TRACE_IRQFLAGS
-	mov.l	5f, r10
-	jsr	@r10
-	 nop
-#endif
+	TRACE_IRQS_ON
 	sti
 
 	!
@@ -355,11 +340,7 @@ syscall_call:
 	!
 syscall_exit:
 	cli
-#ifdef CONFIG_TRACE_IRQFLAGS
-	mov.l	6f, r0
-	jsr	@r0
-	 nop
-#endif
+	TRACE_IRQS_OFF
 	!
 	get_current_thread_info r8, r0
 	mov.l	@(TI_FLAGS,r8), r0		! current_thread_info->flags
@@ -377,9 +358,5 @@ syscall_exit:
 #endif
 2:	.long	NR_syscalls
 3:	.long	sys_call_table
-#ifdef CONFIG_TRACE_IRQFLAGS
-5:	.long	trace_hardirqs_on
-6:	.long	trace_hardirqs_off
-#endif
 7:	.long	do_syscall_trace_enter
 8:	.long	do_syscall_trace_leave
diff --git a/arch/sh/kernel/io_trapped.c b/arch/sh/kernel/io_trapped.c
index 77dfecb6437..e27a19e1f46 100644
--- a/arch/sh/kernel/io_trapped.c
+++ b/arch/sh/kernel/io_trapped.c
@@ -112,14 +112,15 @@ void __iomem *match_trapped_io_handler(struct list_head *list,
 	struct trapped_io *tiop;
 	struct resource *res;
 	int k, len;
+	unsigned long flags;
 
-	spin_lock_irq(&trapped_lock);
+	spin_lock_irqsave(&trapped_lock, flags);
 	list_for_each_entry(tiop, list, list) {
 		voffs = 0;
 		for (k = 0; k < tiop->num_resources; k++) {
 			res = tiop->resource + k;
 			if (res->start == offset) {
-				spin_unlock_irq(&trapped_lock);
+				spin_unlock_irqrestore(&trapped_lock, flags);
 				return tiop->virt_base + voffs;
 			}
 
@@ -127,7 +128,7 @@ void __iomem *match_trapped_io_handler(struct list_head *list,
 			voffs += roundup(len, PAGE_SIZE);
 		}
 	}
-	spin_unlock_irq(&trapped_lock);
+	spin_unlock_irqrestore(&trapped_lock, flags);
 	return NULL;
 }
 EXPORT_SYMBOL_GPL(match_trapped_io_handler);
diff --git a/arch/sh/kernel/irq.c b/arch/sh/kernel/irq.c
index 278c68c6048..2bb43dc74f2 100644
--- a/arch/sh/kernel/irq.c
+++ b/arch/sh/kernel/irq.c
@@ -14,6 +14,7 @@
 #include <asm/processor.h>
 #include <asm/machvec.h>
 #include <asm/uaccess.h>
+#include <asm/dwarf.h>
 #include <asm/thread_info.h>
 #include <cpu/mmu_context.h>
 
@@ -261,6 +262,9 @@ void __init init_IRQ(void)
 		sh_mv.mv_init_irq();
 
 	irq_ctx_init(smp_processor_id());
+
+	/* This needs to be early, but not too early.. */
+	dwarf_unwinder_init();
 }
 
 #ifdef CONFIG_SPARSE_IRQ
diff --git a/arch/sh/kernel/stacktrace.c b/arch/sh/kernel/stacktrace.c
index 1a2a5eb76e4..c2e45c48409 100644
--- a/arch/sh/kernel/stacktrace.c
+++ b/arch/sh/kernel/stacktrace.c
@@ -13,47 +13,93 @@
 #include <linux/stacktrace.h>
 #include <linux/thread_info.h>
 #include <linux/module.h>
+#include <asm/unwinder.h>
 #include <asm/ptrace.h>
+#include <asm/stacktrace.h>
+
+static void save_stack_warning(void *data, char *msg)
+{
+}
+
+static void
+save_stack_warning_symbol(void *data, char *msg, unsigned long symbol)
+{
+}
+
+static int save_stack_stack(void *data, char *name)
+{
+	return 0;
+}
 
 /*
  * Save stack-backtrace addresses into a stack_trace buffer.
  */
+static void save_stack_address(void *data, unsigned long addr, int reliable)
+{
+	struct stack_trace *trace = data;
+
+	if (!reliable)
+		return;
+
+	if (trace->skip > 0) {
+		trace->skip--;
+		return;
+	}
+
+	if (trace->nr_entries < trace->max_entries)
+		trace->entries[trace->nr_entries++] = addr;
+}
+
+static const struct stacktrace_ops save_stack_ops = {
+	.warning = save_stack_warning,
+	.warning_symbol = save_stack_warning_symbol,
+	.stack = save_stack_stack,
+	.address = save_stack_address,
+};
+
 void save_stack_trace(struct stack_trace *trace)
 {
 	unsigned long *sp = (unsigned long *)current_stack_pointer;
 
-	while (!kstack_end(sp)) {
-		unsigned long addr = *sp++;
-
-		if (__kernel_text_address(addr)) {
-			if (trace->skip > 0)
-				trace->skip--;
-			else
-				trace->entries[trace->nr_entries++] = addr;
-			if (trace->nr_entries >= trace->max_entries)
-				break;
-		}
-	}
+	unwind_stack(current, NULL, sp,  &save_stack_ops, trace);
+	if (trace->nr_entries < trace->max_entries)
+		trace->entries[trace->nr_entries++] = ULONG_MAX;
 }
 EXPORT_SYMBOL_GPL(save_stack_trace);
 
+static void
+save_stack_address_nosched(void *data, unsigned long addr, int reliable)
+{
+	struct stack_trace *trace = (struct stack_trace *)data;
+
+	if (!reliable)
+		return;
+
+	if (in_sched_functions(addr))
+		return;
+
+	if (trace->skip > 0) {
+		trace->skip--;
+		return;
+	}
+
+	if (trace->nr_entries < trace->max_entries)
+		trace->entries[trace->nr_entries++] = addr;
+}
+
+static const struct stacktrace_ops save_stack_ops_nosched = {
+	.warning = save_stack_warning,
+	.warning_symbol = save_stack_warning_symbol,
+	.stack = save_stack_stack,
+	.address = save_stack_address_nosched,
+};
+
 void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
 {
 	unsigned long *sp = (unsigned long *)tsk->thread.sp;
 
-	while (!kstack_end(sp)) {
-		unsigned long addr = *sp++;
-
-		if (__kernel_text_address(addr)) {
-			if (in_sched_functions(addr))
-				break;
-			if (trace->skip > 0)
-				trace->skip--;
-			else
-				trace->entries[trace->nr_entries++] = addr;
-			if (trace->nr_entries >= trace->max_entries)
-				break;
-		}
-	}
+	unwind_stack(current, NULL, sp,  &save_stack_ops_nosched, trace);
+	if (trace->nr_entries < trace->max_entries)
+		trace->entries[trace->nr_entries++] = ULONG_MAX;
 }
 EXPORT_SYMBOL_GPL(save_stack_trace_tsk);
diff --git a/arch/sh/kernel/time.c b/arch/sh/kernel/time.c
index d2424b068b7..7f95f479060 100644
--- a/arch/sh/kernel/time.c
+++ b/arch/sh/kernel/time.c
@@ -92,6 +92,21 @@ module_init(rtc_generic_init);
 
 void (*board_time_init)(void);
 
+static void __init sh_late_time_init(void)
+{
+	/*
+	 * Make sure all compiled-in early timers register themselves.
+	 *
+	 * Run probe() for two "earlytimer" devices, these will be the
+	 * clockevents and clocksource devices respectively. In the event
+	 * that only a clockevents device is available, we -ENODEV on the
+	 * clocksource and the jiffies clocksource is used transparently
+	 * instead. No error handling is necessary here.
+	 */
+	early_platform_driver_register_all("earlytimer");
+	early_platform_driver_probe("earlytimer", 2, 0);
+}
+
 void __init time_init(void)
 {
 	if (board_time_init)
@@ -108,15 +123,5 @@ void __init time_init(void)
 	local_timer_setup(smp_processor_id());
 #endif
 
-	/*
-	 * Make sure all compiled-in early timers register themselves.
-	 *
-	 * Run probe() for two "earlytimer" devices, these will be the
-	 * clockevents and clocksource devices respectively. In the event
-	 * that only a clockevents device is available, we -ENODEV on the
-	 * clocksource and the jiffies clocksource is used transparently
-	 * instead. No error handling is necessary here.
-	 */
-	early_platform_driver_register_all("earlytimer");
-	early_platform_driver_probe("earlytimer", 2, 0);
+	late_time_init = sh_late_time_init;
 }
diff --git a/arch/sh/kernel/traps_32.c b/arch/sh/kernel/traps_32.c
index 2b772776fcd..563426487c6 100644
--- a/arch/sh/kernel/traps_32.c
+++ b/arch/sh/kernel/traps_32.c
@@ -858,30 +858,6 @@ void __init trap_init(void)
 	per_cpu_trap_init();
 }
 
-void show_trace(struct task_struct *tsk, unsigned long *sp,
-		struct pt_regs *regs)
-{
-	unsigned long addr;
-
-	if (regs && user_mode(regs))
-		return;
-
-	printk("\nCall trace:\n");
-
-	while (!kstack_end(sp)) {
-		addr = *sp++;
-		if (kernel_text_address(addr))
-			print_ip_sym(addr);
-	}
-
-	printk("\n");
-
-	if (!tsk)
-		tsk = current;
-
-	debug_show_held_locks(tsk);
-}
-
 void show_stack(struct task_struct *tsk, unsigned long *sp)
 {
 	unsigned long stack;
diff --git a/arch/sh/kernel/unwinder.c b/arch/sh/kernel/unwinder.c
new file mode 100644
index 00000000000..2b30fa28b44
--- /dev/null
+++ b/arch/sh/kernel/unwinder.c
@@ -0,0 +1,162 @@
+/*
+ * Copyright (C) 2009  Matt Fleming
+ *
+ * Based, in part, on kernel/time/clocksource.c.
+ *
+ * This file provides arbitration code for stack unwinders.
+ *
+ * Multiple stack unwinders can be available on a system, usually with
+ * the most accurate unwinder being the currently active one.
+ */
+#include <linux/errno.h>
+#include <linux/list.h>
+#include <linux/spinlock.h>
+#include <asm/unwinder.h>
+#include <asm/atomic.h>
+
+/*
+ * This is the most basic stack unwinder an architecture can
+ * provide. For architectures without reliable frame pointers, e.g.
+ * RISC CPUs, it can be implemented by looking through the stack for
+ * addresses that lie within the kernel text section.
+ *
+ * Other CPUs, e.g. x86, can use their frame pointer register to
+ * construct more accurate stack traces.
+ */
+static struct list_head unwinder_list;
+static struct unwinder stack_reader = {
+	.name = "stack-reader",
+	.dump = stack_reader_dump,
+	.rating = 50,
+	.list = {
+		.next = &unwinder_list,
+		.prev = &unwinder_list,
+	},
+};
+
+/*
+ * "curr_unwinder" points to the stack unwinder currently in use. This
+ * is the unwinder with the highest rating.
+ *
+ * "unwinder_list" is a linked-list of all available unwinders, sorted
+ * by rating.
+ *
+ * All modifications of "curr_unwinder" and "unwinder_list" must be
+ * performed whilst holding "unwinder_lock".
+ */
+static struct unwinder *curr_unwinder = &stack_reader;
+
+static struct list_head unwinder_list = {
+	.next = &stack_reader.list,
+	.prev = &stack_reader.list,
+};
+
+static DEFINE_SPINLOCK(unwinder_lock);
+
+static atomic_t unwinder_running = ATOMIC_INIT(0);
+
+/**
+ * select_unwinder - Select the best registered stack unwinder.
+ *
+ * Private function. Must hold unwinder_lock when called.
+ *
+ * Select the stack unwinder with the best rating. This is useful for
+ * setting up curr_unwinder.
+ */
+static struct unwinder *select_unwinder(void)
+{
+	struct unwinder *best;
+
+	if (list_empty(&unwinder_list))
+		return NULL;
+
+	best = list_entry(unwinder_list.next, struct unwinder, list);
+	if (best == curr_unwinder)
+		return NULL;
+
+	return best;
+}
+
+/*
+ * Enqueue the stack unwinder sorted by rating.
+ */
+static int unwinder_enqueue(struct unwinder *ops)
+{
+	struct list_head *tmp, *entry = &unwinder_list;
+
+	list_for_each(tmp, &unwinder_list) {
+		struct unwinder *o;
+
+		o = list_entry(tmp, struct unwinder, list);
+		if (o == ops)
+			return -EBUSY;
+		/* Keep track of the place, where to insert */
+		if (o->rating >= ops->rating)
+			entry = tmp;
+	}
+	list_add(&ops->list, entry);
+
+	return 0;
+}
+
+/**
+ * unwinder_register - Used to install new stack unwinder
+ * @u: unwinder to be registered
+ *
+ * Install the new stack unwinder on the unwinder list, which is sorted
+ * by rating.
+ *
+ * Returns -EBUSY if registration fails, zero otherwise.
+ */
+int unwinder_register(struct unwinder *u)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&unwinder_lock, flags);
+	ret = unwinder_enqueue(u);
+	if (!ret)
+		curr_unwinder = select_unwinder();
+	spin_unlock_irqrestore(&unwinder_lock, flags);
+
+	return ret;
+}
+
+/*
+ * Unwind the call stack and pass information to the stacktrace_ops
+ * functions. Also handle the case where we need to switch to a new
+ * stack dumper because the current one faulted unexpectedly.
+ */
+void unwind_stack(struct task_struct *task, struct pt_regs *regs,
+		  unsigned long *sp, const struct stacktrace_ops *ops,
+		  void *data)
+{
+	unsigned long flags;
+
+	/*
+	 * The problem with unwinders with high ratings is that they are
+	 * inherently more complicated than the simple ones with lower
+	 * ratings. We are therefore more likely to fault in the
+	 * complicated ones, e.g. hitting BUG()s. If we fault in the
+	 * code for the current stack unwinder we try to downgrade to
+	 * one with a lower rating.
+	 *
+	 * Hopefully this will give us a semi-reliable stacktrace so we
+	 * can diagnose why curr_unwinder->dump() faulted.
+	 */
+	if (atomic_inc_return(&unwinder_running) != 1) {
+		spin_lock_irqsave(&unwinder_lock, flags);
+
+		if (!list_is_singular(&unwinder_list)) {
+			list_del(&curr_unwinder->list);
+			curr_unwinder = select_unwinder();
+		}
+
+		spin_unlock_irqrestore(&unwinder_lock, flags);
+		atomic_dec(&unwinder_running);
+	}
+
+	curr_unwinder->dump(task, regs, sp, ops, data);
+
+	atomic_dec(&unwinder_running);
+}
diff --git a/arch/sh/kernel/vmlinux.lds.S b/arch/sh/kernel/vmlinux.lds.S
index 80dc9f8d941..1b7d9d541e0 100644
--- a/arch/sh/kernel/vmlinux.lds.S
+++ b/arch/sh/kernel/vmlinux.lds.S
@@ -12,7 +12,7 @@ OUTPUT_ARCH(sh)
 
 #include <asm/thread_info.h>
 #include <asm/cache.h>
-#include <asm-generic/vmlinux.lds.h>
+#include <asm/vmlinux.lds.h>
 
 ENTRY(_start)
 SECTIONS
@@ -70,6 +70,8 @@ SECTIONS
 
 	_edata = .;			/* End of data section */
 
+	DWARF_EH_FRAME
+
 	. = ALIGN(PAGE_SIZE);		/* Init code and data */
 	__init_begin = .;
 	INIT_TEXT_SECTION(PAGE_SIZE)