70 files changed, 524 insertions, 269 deletions

diff --git a/arch/arm/mach-omap1/board-ams-delta.c b/arch/arm/mach-omap1/board-ams-delta.c
index 9518bf5996dc..e255164ff087 100644
--- a/arch/arm/mach-omap1/board-ams-delta.c
+++ b/arch/arm/mach-omap1/board-ams-delta.c
@@ -444,16 +444,28 @@ static struct omap1_cam_platform_data ams_delta_camera_platform_data = {
         .lclk_khz_max   = 1334,         /* results in 5fps CIF, 10fps QCIF */
 };
 
+static struct platform_device ams_delta_audio_device = {
+        .name   = "ams-delta-audio",
+        .id     = -1,
+};
+
+static struct platform_device cx20442_codec_device = {
+        .name   = "cx20442-codec",
+        .id     = -1,
+};
+
 static struct platform_device *ams_delta_devices[] __initdata = {
         &latch1_gpio_device,
         &latch2_gpio_device,
         &ams_delta_kp_device,
         &ams_delta_camera_device,
+        &ams_delta_audio_device,
 };
 
 static struct platform_device *late_devices[] __initdata = {
         &ams_delta_nand_device,
         &ams_delta_lcd_device,
+        &cx20442_codec_device,
 };
 
 static void __init ams_delta_init(void)
diff --git a/arch/arm/mach-prima2/include/mach/gpio.h b/arch/arm/mach-prima2/include/mach/gpio.h
deleted file mode 100644
index 1904bb03876e..000000000000
--- a/arch/arm/mach-prima2/include/mach/gpio.h
+++ /dev/null
@@ -1,13 +0,0 @@
-#ifndef __MACH_GPIO_H
-#define __MACH_GPIO_H
-
-/* Pull up/down values */
-enum sirfsoc_gpio_pull {
-        SIRFSOC_GPIO_PULL_NONE,
-        SIRFSOC_GPIO_PULL_UP,
-        SIRFSOC_GPIO_PULL_DOWN,
-};
-
-void sirfsoc_gpio_set_pull(unsigned gpio, unsigned mode);
-
-#endif
diff --git a/arch/frv/kernel/entry.S b/arch/frv/kernel/entry.S
index 002732960315..ee0beb354e4d 100644
--- a/arch/frv/kernel/entry.S
+++ b/arch/frv/kernel/entry.S
@@ -867,8 +867,8 @@ ret_from_fork:
 ret_from_kernel_thread:
         lddi.p          @(gr28,#REG_GR(8)),gr20
         call            schedule_tail
-        or.p            gr20,gr20,gr8
-        calll           @(gr21,gr0)
+        calll.p         @(gr21,gr0)
+        or              gr20,gr20,gr8
         bra             sys_exit
 
         .globl          ret_from_kernel_execve
diff --git a/arch/frv/kernel/process.c b/arch/frv/kernel/process.c
index 655d90d20bb0..e1e3aa196aa4 100644
--- a/arch/frv/kernel/process.c
+++ b/arch/frv/kernel/process.c
@@ -194,7 +194,7 @@ int copy_thread(unsigned long clone_flags,
                 memset(childregs, 0, sizeof(struct pt_regs));
                 childregs->gr9 = usp; /* function */
                 childregs->gr8 = arg;
-                chilregs->psr = PSR_S;
+                childregs->psr = PSR_S;
                 p->thread.pc = (unsigned long) ret_from_kernel_thread;
                 save_user_regs(p->thread.user);
                 return 0;
diff --git a/arch/frv/kernel/setup.c b/arch/frv/kernel/setup.c
index 1f1e5efb3385..b8993c87d3de 100644
--- a/arch/frv/kernel/setup.c
+++ b/arch/frv/kernel/setup.c
@@ -112,9 +112,11 @@ char __initdata redboot_command_line[COMMAND_LINE_SIZE];
 #ifdef CONFIG_PM
 #define __pminit
 #define __pminitdata
+#define __pminitconst
 #else
 #define __pminit __init
 #define __pminitdata __initdata
+#define __pminitconst __initconst
 #endif
 
 struct clock_cmode {
diff --git a/arch/ia64/include/uapi/asm/Kbuild b/arch/ia64/include/uapi/asm/Kbuild
index 30cafac93703..1b3f5eb5fcdb 100644
--- a/arch/ia64/include/uapi/asm/Kbuild
+++ b/arch/ia64/include/uapi/asm/Kbuild
@@ -1,6 +1,8 @@
 # UAPI Header export list
 include include/uapi/asm-generic/Kbuild.asm
 
+generic-y += kvm_para.h
+
 header-y += auxvec.h
 header-y += bitsperlong.h
 header-y += break.h
diff --git a/arch/ia64/include/uapi/asm/kvm_para.h b/arch/ia64/include/uapi/asm/kvm_para.h
deleted file mode 100644
index e69de29bb2d1..000000000000
--- a/arch/ia64/include/uapi/asm/kvm_para.h
+++ /dev/null
diff --git a/arch/mips/include/asm/Kbuild b/arch/mips/include/asm/Kbuild
index e69de29bb2d1..533053d12ced 100644
--- a/arch/mips/include/asm/Kbuild
+++ b/arch/mips/include/asm/Kbuild
@@ -0,0 +1 @@
+# MIPS headers
diff --git a/arch/mips/include/asm/delay.h b/arch/mips/include/asm/delay.h
index e7cd78277c23..dc0a5f77a35c 100644
--- a/arch/mips/include/asm/delay.h
+++ b/arch/mips/include/asm/delay.h
@@ -13,9 +13,9 @@
 
 #include <linux/param.h>
 
-extern void __delay(unsigned int loops);
-extern void __ndelay(unsigned int ns);
-extern void __udelay(unsigned int us);
+extern void __delay(unsigned long loops);
+extern void __ndelay(unsigned long ns);
+extern void __udelay(unsigned long us);
 
 #define ndelay(ns) __ndelay(ns)
 #define udelay(us) __udelay(us)
diff --git a/arch/mips/include/asm/pgtable-64.h b/arch/mips/include/asm/pgtable-64.h
index c26e18250079..f5b521d5a67d 100644
--- a/arch/mips/include/asm/pgtable-64.h
+++ b/arch/mips/include/asm/pgtable-64.h
@@ -9,6 +9,7 @@
 #ifndef _ASM_PGTABLE_64_H
 #define _ASM_PGTABLE_64_H
 
+#include <linux/compiler.h>
 #include <linux/linkage.h>
 
 #include <asm/addrspace.h>
@@ -172,7 +173,19 @@ static inline int pmd_none(pmd_t pmd)
         return pmd_val(pmd) == (unsigned long) invalid_pte_table;
 }
 
-#define pmd_bad(pmd)            (pmd_val(pmd) & ~PAGE_MASK)
+static inline int pmd_bad(pmd_t pmd)
+{
+#ifdef CONFIG_HUGETLB_PAGE
+        /* pmd_huge(pmd) but inline */
+        if (unlikely(pmd_val(pmd) & _PAGE_HUGE))
+                return 0;
+#endif
+
+        if (unlikely(pmd_val(pmd) & ~PAGE_MASK))
+                return 1;
+
+        return 0;
+}
 
 static inline int pmd_present(pmd_t pmd)
 {
diff --git a/arch/mips/jz4740/serial.h b/arch/mips/jz4740/serial.h
index b9fe3ade0289..8eb715bb1ea8 100644
--- a/arch/mips/jz4740/serial.h
+++ b/arch/mips/jz4740/serial.h
@@ -14,6 +14,9 @@
  */
 
 #ifndef __MIPS_JZ4740_SERIAL_H__
+#define __MIPS_JZ4740_SERIAL_H__
+
+struct uart_port;
 
 void jz4740_serial_out(struct uart_port *p, int offset, int value);
 
diff --git a/arch/mips/kernel/smp-cmp.c b/arch/mips/kernel/smp-cmp.c
index afc379ca3753..06cd0c610f44 100644
--- a/arch/mips/kernel/smp-cmp.c
+++ b/arch/mips/kernel/smp-cmp.c
@@ -97,7 +97,7 @@ static void cmp_init_secondary(void)
 
         /* Enable per-cpu interrupts: platform specific */
 
-        c->core = (read_c0_ebase() >> 1) & 0xff;
+        c->core = (read_c0_ebase() >> 1) & 0x1ff;
 #if defined(CONFIG_MIPS_MT_SMP) || defined(CONFIG_MIPS_MT_SMTC)
         c->vpe_id = (read_c0_tcbind() >> TCBIND_CURVPE_SHIFT) & TCBIND_CURVPE;
 #endif
diff --git a/arch/mips/lib/delay.c b/arch/mips/lib/delay.c
index 5995969e8c42..dc81ca8dc0dd 100644
--- a/arch/mips/lib/delay.c
+++ b/arch/mips/lib/delay.c
@@ -15,13 +15,17 @@
 #include <asm/compiler.h>
 #include <asm/war.h>
 
-inline void __delay(unsigned int loops)
+void __delay(unsigned long loops)
 {
         __asm__ __volatile__ (
         "       .set    noreorder                               \n"
         "       .align  3                                       \n"
         "1:     bnez    %0, 1b                                  \n"
+#if __SIZEOF_LONG__ == 4
         "       subu    %0, 1                                   \n"
+#else
+        "       dsubu   %0, 1                                   \n"
+#endif
         "       .set    reorder                                 \n"
         : "=r" (loops)
         : "0" (loops));
diff --git a/arch/mips/lib/dump_tlb.c b/arch/mips/lib/dump_tlb.c
index 3f69725556af..a99c1d3fc567 100644
--- a/arch/mips/lib/dump_tlb.c
+++ b/arch/mips/lib/dump_tlb.c
@@ -50,8 +50,9 @@ static void dump_tlb(int first, int last)
 {
         unsigned long s_entryhi, entryhi, asid;
         unsigned long long entrylo0, entrylo1;
-        unsigned int s_index, pagemask, c0, c1, i;
+        unsigned int s_index, s_pagemask, pagemask, c0, c1, i;
 
+        s_pagemask = read_c0_pagemask();
         s_entryhi = read_c0_entryhi();
         s_index = read_c0_index();
         asid = s_entryhi & 0xff;
@@ -103,6 +104,7 @@ static void dump_tlb(int first, int last)
 
         write_c0_entryhi(s_entryhi);
         write_c0_index(s_index);
+        write_c0_pagemask(s_pagemask);
 }
 
 void dump_tlb_all(void)
diff --git a/arch/mips/mm/tlb-r4k.c b/arch/mips/mm/tlb-r4k.c
index 87b9cfcc30ff..4b9b935a070e 100644
--- a/arch/mips/mm/tlb-r4k.c
+++ b/arch/mips/mm/tlb-r4k.c
@@ -320,6 +320,7 @@ void __update_tlb(struct vm_area_struct * vma, unsigned long address, pte_t pte)
                         tlb_write_random();
                 else
                         tlb_write_indexed();
+                tlbw_use_hazard();
                 write_c0_pagemask(PM_DEFAULT_MASK);
         } else
 #endif
diff --git a/arch/mips/mm/tlbex.c b/arch/mips/mm/tlbex.c
index 658a520364ce..2833dcb67b5a 100644
--- a/arch/mips/mm/tlbex.c
+++ b/arch/mips/mm/tlbex.c
@@ -148,8 +148,8 @@ enum label_id {
         label_leave,
         label_vmalloc,
         label_vmalloc_done,
-        label_tlbw_hazard,
-        label_split,
+        label_tlbw_hazard_0,
+        label_split = label_tlbw_hazard_0 + 8,
         label_tlbl_goaround1,
         label_tlbl_goaround2,
         label_nopage_tlbl,
@@ -167,7 +167,7 @@ UASM_L_LA(_second_part)
 UASM_L_LA(_leave)
 UASM_L_LA(_vmalloc)
 UASM_L_LA(_vmalloc_done)
-UASM_L_LA(_tlbw_hazard)
+/* _tlbw_hazard_x is handled differently.  */
 UASM_L_LA(_split)
 UASM_L_LA(_tlbl_goaround1)
 UASM_L_LA(_tlbl_goaround2)
@@ -181,6 +181,30 @@ UASM_L_LA(_large_segbits_fault)
 UASM_L_LA(_tlb_huge_update)
 #endif
 
+static int __cpuinitdata hazard_instance;
+
+static void uasm_bgezl_hazard(u32 **p, struct uasm_reloc **r, int instance)
+{
+        switch (instance) {
+        case 0 ... 7:
+                uasm_il_bgezl(p, r, 0, label_tlbw_hazard_0 + instance);
+                return;
+        default:
+                BUG();
+        }
+}
+
+static void uasm_bgezl_label(struct uasm_label **l, u32 **p, int instance)
+{
+        switch (instance) {
+        case 0 ... 7:
+                uasm_build_label(l, *p, label_tlbw_hazard_0 + instance);
+                break;
+        default:
+                BUG();
+        }
+}
+
 /*
  * For debug purposes.
  */
@@ -478,21 +502,28 @@ static void __cpuinit build_tlb_write_entry(u32 **p, struct uasm_label **l,
                  * This branch uses up a mtc0 hazard nop slot and saves
                  * two nops after the tlbw instruction.
                  */
-                uasm_il_bgezl(p, r, 0, label_tlbw_hazard);
+                uasm_bgezl_hazard(p, r, hazard_instance);
                 tlbw(p);
-                uasm_l_tlbw_hazard(l, *p);
+                uasm_bgezl_label(l, p, hazard_instance);
+                hazard_instance++;
                 uasm_i_nop(p);
                 break;
 
         case CPU_R4600:
         case CPU_R4700:
-        case CPU_R5000:
-        case CPU_R5000A:
                 uasm_i_nop(p);
                 tlbw(p);
                 uasm_i_nop(p);
                 break;
 
+        case CPU_R5000:
+        case CPU_R5000A:
+        case CPU_NEVADA:
+                uasm_i_nop(p); /* QED specifies 2 nops hazard */
+                uasm_i_nop(p); /* QED specifies 2 nops hazard */
+                tlbw(p);
+                break;
+
         case CPU_R4300:
         case CPU_5KC:
         case CPU_TX49XX:
@@ -526,17 +557,6 @@ static void __cpuinit build_tlb_write_entry(u32 **p, struct uasm_label **l,
                 tlbw(p);
                 break;
 
-        case CPU_NEVADA:
-                uasm_i_nop(p); /* QED specifies 2 nops hazard */
-                /*
-                 * This branch uses up a mtc0 hazard nop slot and saves
-                 * a nop after the tlbw instruction.
-                 */
-                uasm_il_bgezl(p, r, 0, label_tlbw_hazard);
-                tlbw(p);
-                uasm_l_tlbw_hazard(l, *p);
-                break;
-
         case CPU_RM7000:
                 uasm_i_nop(p);
                 uasm_i_nop(p);
diff --git a/arch/mn10300/include/uapi/asm/setup.h b/arch/mn10300/include/uapi/asm/setup.h
index e69de29bb2d1..ae5704fa77ad 100644
--- a/arch/mn10300/include/uapi/asm/setup.h
+++ b/arch/mn10300/include/uapi/asm/setup.h
@@ -0,0 +1,4 @@
+/*
+ * There isn't anything here anymore, but the file must not be empty or patch
+ * will delete it.
+ */
diff --git a/arch/powerpc/include/asm/kvm_book3s_32.h b/arch/powerpc/include/asm/kvm_book3s_32.h
index 38040ff82063..ce0ef6ce8f86 100644
--- a/arch/powerpc/include/asm/kvm_book3s_32.h
+++ b/arch/powerpc/include/asm/kvm_book3s_32.h
@@ -42,5 +42,6 @@ static inline void svcpu_put(struct kvmppc_book3s_shadow_vcpu *svcpu)
 #define SID_SHIFT       28
 #define ESID_MASK       0xf0000000
 #define VSID_MASK       0x00fffffff0000000ULL
+#define VPN_SHIFT       12
 
 #endif /* __ASM_KVM_BOOK3S_32_H__ */
diff --git a/arch/powerpc/include/asm/processor.h b/arch/powerpc/include/asm/processor.h
index 8734b3855272..87502046c0dc 100644
--- a/arch/powerpc/include/asm/processor.h
+++ b/arch/powerpc/include/asm/processor.h
@@ -388,9 +388,9 @@ extern int powersave_nap;	/* set if nap mode can be used in idle loop */
 extern void power7_nap(void);
 
 #ifdef CONFIG_PSERIES_IDLE
-extern void update_smt_snooze_delay(int snooze);
+extern void update_smt_snooze_delay(int cpu, int residency);
 #else
-static inline void update_smt_snooze_delay(int snooze) {}
+static inline void update_smt_snooze_delay(int cpu, int residency) {}
 #endif
 
 extern void flush_instruction_cache(void);
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 8302af649219..cf357a059ddb 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -50,7 +50,7 @@ static ssize_t store_smt_snooze_delay(struct device *dev,
                 return -EINVAL;
 
         per_cpu(smt_snooze_delay, cpu->dev.id) = snooze;
-        update_smt_snooze_delay(snooze);
+        update_smt_snooze_delay(cpu->dev.id, snooze);
 
         return count;
 }
diff --git a/arch/powerpc/kvm/book3s_32_mmu_host.c b/arch/powerpc/kvm/book3s_32_mmu_host.c
index 00aa61268e0d..b0f625a33345 100644
--- a/arch/powerpc/kvm/book3s_32_mmu_host.c
+++ b/arch/powerpc/kvm/book3s_32_mmu_host.c
@@ -173,8 +173,8 @@ int kvmppc_mmu_map_page(struct kvm_vcpu *vcpu, struct kvmppc_pte *orig_pte)
         BUG_ON(!map);
 
         vsid = map->host_vsid;
-        vpn = (vsid << (SID_SHIFT - VPN_SHIFT)) | ((eaddr & ~ESID_MASK) >> VPN_SHIFT)
-
+        vpn = (vsid << (SID_SHIFT - VPN_SHIFT)) |
+                ((eaddr & ~ESID_MASK) >> VPN_SHIFT);
 next_pteg:
         if (rr == 16) {
                 primary = !primary;
diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 0db88f501f91..aa2465e21f1a 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c
@@ -1463,7 +1463,7 @@ static void perf_event_interrupt(struct pt_regs *regs)
                 if (!event->hw.idx || is_limited_pmc(event->hw.idx))
                         continue;
                 val = read_pmc(event->hw.idx);
-                if (pmc_overflow(val)) {
+                if ((int)val < 0) {
                         /* event has overflowed */
                         found = 1;
                         record_and_restart(event, val, regs);
diff --git a/arch/powerpc/platforms/pseries/processor_idle.c b/arch/powerpc/platforms/pseries/processor_idle.c
index 455760b1fe6e..45d00e5fe14d 100644
--- a/arch/powerpc/platforms/pseries/processor_idle.c
+++ b/arch/powerpc/platforms/pseries/processor_idle.c
@@ -33,13 +33,6 @@ static int max_idle_state = MAX_IDLE_STATE_COUNT - 1;
 static struct cpuidle_device __percpu *pseries_cpuidle_devices;
 static struct cpuidle_state *cpuidle_state_table;
 
-void update_smt_snooze_delay(int snooze)
-{
-        struct cpuidle_driver *drv = cpuidle_get_driver();
-        if (drv)
-                drv->states[0].target_residency = snooze;
-}
-
 static inline void idle_loop_prolog(unsigned long *in_purr, ktime_t *kt_before)
 {
 
@@ -66,32 +59,22 @@ static int snooze_loop(struct cpuidle_device *dev,
 {
         unsigned long in_purr;
         ktime_t kt_before;
-        unsigned long start_snooze;
-        long snooze = drv->states[0].target_residency;
+        int cpu = dev->cpu;
 
         idle_loop_prolog(&in_purr, &kt_before);
+        local_irq_enable();
+        set_thread_flag(TIF_POLLING_NRFLAG);
 
-        if (snooze) {
-                start_snooze = get_tb() + snooze * tb_ticks_per_usec;
-                local_irq_enable();
-                set_thread_flag(TIF_POLLING_NRFLAG);
-
-                while ((snooze < 0) || (get_tb() < start_snooze)) {
-                        if (need_resched() || cpu_is_offline(dev->cpu))
-                                goto out;
-                        ppc64_runlatch_off();
-                        HMT_low();
-                        HMT_very_low();
-                }
-
-                HMT_medium();
-                clear_thread_flag(TIF_POLLING_NRFLAG);
-                smp_mb();
-                local_irq_disable();
+        while ((!need_resched()) && cpu_online(cpu)) {
+                ppc64_runlatch_off();
+                HMT_low();
+                HMT_very_low();
         }
 
-out:
         HMT_medium();
+        clear_thread_flag(TIF_POLLING_NRFLAG);
+        smp_mb();
+
         dev->last_residency =
                 (int)idle_loop_epilog(in_purr, kt_before);
         return index;
@@ -172,8 +155,8 @@ static struct cpuidle_state dedicated_states[MAX_IDLE_STATE_COUNT] = {
                 .name = "CEDE",
                 .desc = "CEDE",
                 .flags = CPUIDLE_FLAG_TIME_VALID,
-                .exit_latency = 1,
-                .target_residency = 10,
+                .exit_latency = 10,
+                .target_residency = 100,
                 .enter = &dedicated_cede_loop },
 };
 
@@ -190,6 +173,23 @@ static struct cpuidle_state shared_states[MAX_IDLE_STATE_COUNT] = {
                 .enter = &shared_cede_loop },
 };
 
+void update_smt_snooze_delay(int cpu, int residency)
+{
+        struct cpuidle_driver *drv = cpuidle_get_driver();
+        struct cpuidle_device *dev = per_cpu(cpuidle_devices, cpu);
+
+        if (cpuidle_state_table != dedicated_states)
+                return;
+
+        if (residency < 0) {
+                /* Disable the Nap state on that cpu */
+                if (dev)
+                        dev->states_usage[1].disable = 1;
+        } else
+                if (drv)
+                        drv->states[1].target_residency = residency;
+}
+
 static int pseries_cpuidle_add_cpu_notifier(struct notifier_block *n,
                         unsigned long action, void *hcpu)
 {
@@ -246,10 +246,6 @@ static int pseries_cpuidle_driver_init(void)
                 drv->states[drv->state_count] = /* structure copy */
                         cpuidle_state_table[idle_state];
 
-                if (cpuidle_state_table == dedicated_states)
-                        drv->states[drv->state_count].target_residency =
-                                __get_cpu_var(smt_snooze_delay);
-
                 drv->state_count += 1;
         }
 
diff --git a/arch/s390/include/uapi/asm/Kbuild b/arch/s390/include/uapi/asm/Kbuild
index 7bf68fff7c5d..59b67ed423b4 100644
--- a/arch/s390/include/uapi/asm/Kbuild
+++ b/arch/s390/include/uapi/asm/Kbuild
@@ -1,6 +1,8 @@
 # UAPI Header export list
 include include/uapi/asm-generic/Kbuild.asm
 
+generic-y += kvm_para.h
+
 header-y += auxvec.h
 header-y += bitsperlong.h
 header-y += byteorder.h
diff --git a/arch/s390/include/uapi/asm/kvm_para.h b/arch/s390/include/uapi/asm/kvm_para.h
deleted file mode 100644
index e69de29bb2d1..000000000000
--- a/arch/s390/include/uapi/asm/kvm_para.h
+++ /dev/null
diff --git a/arch/sh/include/asm/Kbuild b/arch/sh/include/asm/Kbuild
index 86eadceff097..29f83beeef7a 100644
--- a/arch/sh/include/asm/Kbuild
+++ b/arch/sh/include/asm/Kbuild
@@ -1,4 +1,3 @@
-include include/asm-generic/Kbuild.asm
 
 generic-y += bitsperlong.h
 generic-y += cputime.h
@@ -34,13 +33,3 @@ generic-y += termbits.h
 generic-y += termios.h
 generic-y += ucontext.h
 generic-y += xor.h
-
-header-y += cachectl.h
-header-y += cpu-features.h
-header-y += hw_breakpoint.h
-header-y += posix_types_32.h
-header-y += posix_types_64.h
-header-y += ptrace_32.h
-header-y += ptrace_64.h
-header-y += unistd_32.h
-header-y += unistd_64.h
diff --git a/arch/sh/include/asm/hw_breakpoint.h b/arch/sh/include/asm/hw_breakpoint.h
index 89890f61a7b9..ec9ad593c3da 100644
--- a/arch/sh/include/asm/hw_breakpoint.h
+++ b/arch/sh/include/asm/hw_breakpoint.h
@@ -1,7 +1,8 @@
 #ifndef __ASM_SH_HW_BREAKPOINT_H
 #define __ASM_SH_HW_BREAKPOINT_H
 
-#ifdef __KERNEL__
+#include <uapi/asm/hw_breakpoint.h>
+
 #define __ARCH_HW_BREAKPOINT_H
 
 #include <linux/kdebug.h>
@@ -66,5 +67,4 @@ extern int register_sh_ubc(struct sh_ubc *);
 
 extern struct pmu perf_ops_bp;
 
-#endif /* __KERNEL__ */
 #endif /* __ASM_SH_HW_BREAKPOINT_H */
diff --git a/arch/sh/include/asm/posix_types.h b/arch/sh/include/asm/posix_types.h
index f08449bcbde7..1aa781079b1e 100644
--- a/arch/sh/include/asm/posix_types.h
+++ b/arch/sh/include/asm/posix_types.h
@@ -1,13 +1,5 @@
-#ifdef __KERNEL__
 # ifdef CONFIG_SUPERH32
 #  include <asm/posix_types_32.h>
 # else
 #  include <asm/posix_types_64.h>
 # endif
-#else
-# ifdef __SH5__
-#  include <asm/posix_types_64.h>
-# else
-#  include <asm/posix_types_32.h>
-# endif
-#endif /* __KERNEL__ */
diff --git a/arch/sh/include/asm/ptrace.h b/arch/sh/include/asm/ptrace.h
index a4a38dff997a..2506c7db76b7 100644
--- a/arch/sh/include/asm/ptrace.h
+++ b/arch/sh/include/asm/ptrace.h
@@ -1,42 +1,16 @@
-#ifndef __ASM_SH_PTRACE_H
-#define __ASM_SH_PTRACE_H
-
 /*
  * Copyright (C) 1999, 2000  Niibe Yutaka
  */
+#ifndef __ASM_SH_PTRACE_H
+#define __ASM_SH_PTRACE_H
 
-#define PTRACE_GETREGS          12      /* General registers */
-#define PTRACE_SETREGS          13
-
-#define PTRACE_GETFPREGS        14      /* FPU registers */
-#define PTRACE_SETFPREGS        15
-
-#define PTRACE_GETFDPIC         31      /* get the ELF fdpic loadmap address */
-
-#define PTRACE_GETFDPIC_EXEC    0       /* [addr] request the executable loadmap */
-#define PTRACE_GETFDPIC_INTERP  1       /* [addr] request the interpreter loadmap */
-
-#define PTRACE_GETDSPREGS       55      /* DSP registers */
-#define PTRACE_SETDSPREGS       56
-
-#define PT_TEXT_END_ADDR        240
-#define PT_TEXT_ADDR            244     /* &(struct user)->start_code */
-#define PT_DATA_ADDR            248     /* &(struct user)->start_data */
-#define PT_TEXT_LEN             252
-
-#if defined(__SH5__) || defined(CONFIG_CPU_SH5)
-#include <asm/ptrace_64.h>
-#else
-#include <asm/ptrace_32.h>
-#endif
-
-#ifdef __KERNEL__
 
 #include <linux/stringify.h>
 #include <linux/stddef.h>
 #include <linux/thread_info.h>
 #include <asm/addrspace.h>
 #include <asm/page.h>
+#include <uapi/asm/ptrace.h>
 
 #define user_mode(regs)                 (((regs)->sr & 0x40000000)==0)
 #define kernel_stack_pointer(_regs)     ((unsigned long)(_regs)->regs[15])
@@ -140,6 +114,4 @@ static inline unsigned long profile_pc(struct pt_regs *regs)
 #define profile_pc profile_pc
 
 #include <asm-generic/ptrace.h>
-#endif /* __KERNEL__ */
-
 #endif /* __ASM_SH_PTRACE_H */
diff --git a/arch/sh/include/asm/ptrace_32.h b/arch/sh/include/asm/ptrace_32.h
index 2d3e906aa722..1dd4480c5363 100644
--- a/arch/sh/include/asm/ptrace_32.h
+++ b/arch/sh/include/asm/ptrace_32.h
@@ -1,79 +1,8 @@
 #ifndef __ASM_SH_PTRACE_32_H
 #define __ASM_SH_PTRACE_32_H
 
-/*
- * GCC defines register number like this:
- * -----------------------------
- *       0 - 15 are integer registers
- *      17 - 22 are control/special registers
- *      24 - 39 fp registers
- *      40 - 47 xd registers
- *      48 -    fpscr register
- * -----------------------------
- *
- * We follows above, except:
- *      16 --- program counter (PC)
- *      22 --- syscall #
- *      23 --- floating point communication register
- */
-#define REG_REG0         0
-#define REG_REG15       15
+#include <uapi/asm/ptrace_32.h>
 
-#define REG_PC          16
-
-#define REG_PR          17
-#define REG_SR          18
-#define REG_GBR         19
-#define REG_MACH        20
-#define REG_MACL        21
-
-#define REG_SYSCALL     22
-
-#define REG_FPREG0      23
-#define REG_FPREG15     38
-#define REG_XFREG0      39
-#define REG_XFREG15     54
-
-#define REG_FPSCR       55
-#define REG_FPUL        56
-
-/*
- * This struct defines the way the registers are stored on the
- * kernel stack during a system call or other kernel entry.
- */
-struct pt_regs {
-        unsigned long regs[16];
-        unsigned long pc;
-        unsigned long pr;
-        unsigned long sr;
-        unsigned long gbr;
-        unsigned long mach;
-        unsigned long macl;
-        long tra;
-};
-
-/*
- * This struct defines the way the DSP registers are stored on the
- * kernel stack during a system call or other kernel entry.
- */
-struct pt_dspregs {
-        unsigned long   a1;
-        unsigned long   a0g;
-        unsigned long   a1g;
-        unsigned long   m0;
-        unsigned long   m1;
-        unsigned long   a0;
-        unsigned long   x0;
-        unsigned long   x1;
-        unsigned long   y0;
-        unsigned long   y1;
-        unsigned long   dsr;
-        unsigned long   rs;
-        unsigned long   re;
-        unsigned long   mod;
-};
-
-#ifdef __KERNEL__
 
 #define MAX_REG_OFFSET          offsetof(struct pt_regs, tra)
 static inline long regs_return_value(struct pt_regs *regs)
@@ -81,6 +10,4 @@ static inline long regs_return_value(struct pt_regs *regs)
         return regs->regs[0];
 }
 
-#endif /* __KERNEL__ */
-
 #endif /* __ASM_SH_PTRACE_32_H */
diff --git a/arch/sh/include/asm/ptrace_64.h b/arch/sh/include/asm/ptrace_64.h
index eb3fcceaf64b..97f4b5660f2c 100644
--- a/arch/sh/include/asm/ptrace_64.h
+++ b/arch/sh/include/asm/ptrace_64.h
@@ -1,16 +1,8 @@
 #ifndef __ASM_SH_PTRACE_64_H
 #define __ASM_SH_PTRACE_64_H
 
-struct pt_regs {
-        unsigned long long pc;
-        unsigned long long sr;
-        long long syscall_nr;
-        unsigned long long regs[63];
-        unsigned long long tregs[8];
-        unsigned long long pad[2];
-};
+#include <uapi/asm/ptrace_64.h>
 
-#ifdef __KERNEL__
 
 #define MAX_REG_OFFSET          offsetof(struct pt_regs, tregs[7])
 static inline long regs_return_value(struct pt_regs *regs)
@@ -18,6 +10,4 @@ static inline long regs_return_value(struct pt_regs *regs)
         return regs->regs[3];
 }
 
-#endif /* __KERNEL__ */
-
 #endif /* __ASM_SH_PTRACE_64_H */
diff --git a/arch/sh/include/asm/setup.h b/arch/sh/include/asm/setup.h
index 465a22df8fd0..99238108e7a5 100644
--- a/arch/sh/include/asm/setup.h
+++ b/arch/sh/include/asm/setup.h
@@ -1,9 +1,8 @@
 #ifndef _SH_SETUP_H
 #define _SH_SETUP_H
 
-#include <asm-generic/setup.h>
+#include <uapi/asm/setup.h>
 
-#ifdef __KERNEL__
 /*
  * This is set up by the setup-routine at boot-time
  */
@@ -22,6 +21,4 @@ void sh_mv_setup(void);
 void check_for_initrd(void);
 void per_cpu_trap_init(void);
 
-#endif /* __KERNEL__ */
-
 #endif /* _SH_SETUP_H */
diff --git a/arch/sh/include/asm/types.h b/arch/sh/include/asm/types.h
index f8421f7ad63a..6a31053fa5e3 100644
--- a/arch/sh/include/asm/types.h
+++ b/arch/sh/include/asm/types.h
@@ -1,12 +1,11 @@
 #ifndef __ASM_SH_TYPES_H
 #define __ASM_SH_TYPES_H
 
-#include <asm-generic/types.h>
+#include <uapi/asm/types.h>
 
 /*
  * These aren't exported outside the kernel to avoid name space clashes
  */
-#ifdef __KERNEL__
 #ifndef __ASSEMBLY__
 
 #ifdef CONFIG_SUPERH32
@@ -18,6 +17,4 @@ typedef u64 reg_size_t;
 #endif
 
 #endif /* __ASSEMBLY__ */
-#endif /* __KERNEL__ */
-
 #endif /* __ASM_SH_TYPES_H */
diff --git a/arch/sh/include/asm/unistd.h b/arch/sh/include/asm/unistd.h
index 307201a854f3..38956dfa76f7 100644
--- a/arch/sh/include/asm/unistd.h
+++ b/arch/sh/include/asm/unistd.h
@@ -1,4 +1,3 @@
-#ifdef __KERNEL__
 # ifdef CONFIG_SUPERH32
 #  include <asm/unistd_32.h>
 # else
@@ -38,10 +37,4 @@
  */
 # define cond_syscall(x) asm(".weak\t" #x "\n\t.set\t" #x ",sys_ni_syscall")
 
-#else
-# ifdef __SH5__
-#  include <asm/unistd_64.h>
-# else
-#  include <asm/unistd_32.h>
-# endif
-#endif
+#include <uapi/asm/unistd.h>
diff --git a/arch/sh/include/uapi/asm/Kbuild b/arch/sh/include/uapi/asm/Kbuild
index baebb3da1d44..60613ae78513 100644
--- a/arch/sh/include/uapi/asm/Kbuild
+++ b/arch/sh/include/uapi/asm/Kbuild
@@ -1,3 +1,25 @@
 # UAPI Header export list
 include include/uapi/asm-generic/Kbuild.asm
 
+header-y += auxvec.h
+header-y += byteorder.h
+header-y += cachectl.h
+header-y += cpu-features.h
+header-y += hw_breakpoint.h
+header-y += ioctls.h
+header-y += posix_types.h
+header-y += posix_types_32.h
+header-y += posix_types_64.h
+header-y += ptrace.h
+header-y += ptrace_32.h
+header-y += ptrace_64.h
+header-y += setup.h
+header-y += sigcontext.h
+header-y += signal.h
+header-y += sockios.h
+header-y += stat.h
+header-y += swab.h
+header-y += types.h
+header-y += unistd.h
+header-y += unistd_32.h
+header-y += unistd_64.h
diff --git a/arch/sh/include/asm/auxvec.h b/arch/sh/include/uapi/asm/auxvec.h
index 8bcc51af9367..8bcc51af9367 100644
--- a/arch/sh/include/asm/auxvec.h
+++ b/arch/sh/include/uapi/asm/auxvec.h
diff --git a/arch/sh/include/asm/byteorder.h b/arch/sh/include/uapi/asm/byteorder.h
index db2f5d7cb17d..db2f5d7cb17d 100644
--- a/arch/sh/include/asm/byteorder.h
+++ b/arch/sh/include/uapi/asm/byteorder.h
diff --git a/arch/sh/include/asm/cachectl.h b/arch/sh/include/uapi/asm/cachectl.h
index 6ffb4b7a212e..6ffb4b7a212e 100644
--- a/arch/sh/include/asm/cachectl.h
+++ b/arch/sh/include/uapi/asm/cachectl.h
diff --git a/arch/sh/include/asm/cpu-features.h b/arch/sh/include/uapi/asm/cpu-features.h
index 694abe490edb..694abe490edb 100644
--- a/arch/sh/include/asm/cpu-features.h
+++ b/arch/sh/include/uapi/asm/cpu-features.h
diff --git a/arch/sh/include/uapi/asm/hw_breakpoint.h b/arch/sh/include/uapi/asm/hw_breakpoint.h
new file mode 100644
index 000000000000..ae5704fa77ad
--- /dev/null
+++ b/arch/sh/include/uapi/asm/hw_breakpoint.h
@@ -0,0 +1,4 @@
+/*
+ * There isn't anything here anymore, but the file must not be empty or patch
+ * will delete it.
+ */
diff --git a/arch/sh/include/asm/ioctls.h b/arch/sh/include/uapi/asm/ioctls.h
index a6769f352bf6..a6769f352bf6 100644
--- a/arch/sh/include/asm/ioctls.h
+++ b/arch/sh/include/uapi/asm/ioctls.h
diff --git a/arch/sh/include/uapi/asm/posix_types.h b/arch/sh/include/uapi/asm/posix_types.h
new file mode 100644
index 000000000000..dc55e5adfe1e
--- /dev/null
+++ b/arch/sh/include/uapi/asm/posix_types.h
@@ -0,0 +1,7 @@
+#ifndef __KERNEL__
+# ifdef __SH5__
+#  include <asm/posix_types_64.h>
+# else
+#  include <asm/posix_types_32.h>
+# endif
+#endif /* __KERNEL__ */
diff --git a/arch/sh/include/asm/posix_types_32.h b/arch/sh/include/uapi/asm/posix_types_32.h
index ba0bdc423b07..ba0bdc423b07 100644
--- a/arch/sh/include/asm/posix_types_32.h
+++ b/arch/sh/include/uapi/asm/posix_types_32.h
diff --git a/arch/sh/include/asm/posix_types_64.h b/arch/sh/include/uapi/asm/posix_types_64.h
index 244f7e950e17..244f7e950e17 100644
--- a/arch/sh/include/asm/posix_types_64.h
+++ b/arch/sh/include/uapi/asm/posix_types_64.h
diff --git a/arch/sh/include/uapi/asm/ptrace.h b/arch/sh/include/uapi/asm/ptrace.h
new file mode 100644
index 000000000000..8b8c5aca9c28
--- /dev/null
+++ b/arch/sh/include/uapi/asm/ptrace.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright (C) 1999, 2000  Niibe Yutaka
+ */
+#ifndef _UAPI__ASM_SH_PTRACE_H
+#define _UAPI__ASM_SH_PTRACE_H
+
+
+#define PTRACE_GETREGS          12      /* General registers */
+#define PTRACE_SETREGS          13
+
+#define PTRACE_GETFPREGS        14      /* FPU registers */
+#define PTRACE_SETFPREGS        15
+
+#define PTRACE_GETFDPIC         31      /* get the ELF fdpic loadmap address */
+
+#define PTRACE_GETFDPIC_EXEC    0       /* [addr] request the executable loadmap */
+#define PTRACE_GETFDPIC_INTERP  1       /* [addr] request the interpreter loadmap */
+
+#define PTRACE_GETDSPREGS       55      /* DSP registers */
+#define PTRACE_SETDSPREGS       56
+
+#define PT_TEXT_END_ADDR        240
+#define PT_TEXT_ADDR            244     /* &(struct user)->start_code */
+#define PT_DATA_ADDR            248     /* &(struct user)->start_data */
+#define PT_TEXT_LEN             252
+
+#if defined(__SH5__) || defined(CONFIG_CPU_SH5)
+#include <asm/ptrace_64.h>
+#else
+#include <asm/ptrace_32.h>
+#endif
+
+
+#endif /* _UAPI__ASM_SH_PTRACE_H */
diff --git a/arch/sh/include/uapi/asm/ptrace_32.h b/arch/sh/include/uapi/asm/ptrace_32.h
new file mode 100644
index 000000000000..926e0cefc2bb
--- /dev/null
+++ b/arch/sh/include/uapi/asm/ptrace_32.h
@@ -0,0 +1,77 @@
+#ifndef _UAPI__ASM_SH_PTRACE_32_H
+#define _UAPI__ASM_SH_PTRACE_32_H
+
+/*
+ * GCC defines register number like this:
+ * -----------------------------
+ *       0 - 15 are integer registers
+ *      17 - 22 are control/special registers
+ *      24 - 39 fp registers
+ *      40 - 47 xd registers
+ *      48 -    fpscr register
+ * -----------------------------
+ *
+ * We follows above, except:
+ *      16 --- program counter (PC)
+ *      22 --- syscall #
+ *      23 --- floating point communication register
+ */
+#define REG_REG0         0
+#define REG_REG15       15
+
+#define REG_PC          16
+
+#define REG_PR          17
+#define REG_SR          18
+#define REG_GBR         19
+#define REG_MACH        20
+#define REG_MACL        21
+
+#define REG_SYSCALL     22
+
+#define REG_FPREG0      23
+#define REG_FPREG15     38
+#define REG_XFREG0      39
+#define REG_XFREG15     54
+
+#define REG_FPSCR       55
+#define REG_FPUL        56
+
+/*
+ * This struct defines the way the registers are stored on the
+ * kernel stack during a system call or other kernel entry.
+ */
+struct pt_regs {
+        unsigned long regs[16];
+        unsigned long pc;
+        unsigned long pr;
+        unsigned long sr;
+        unsigned long gbr;
+        unsigned long mach;
+        unsigned long macl;
+        long tra;
+};
+
+/*
+ * This struct defines the way the DSP registers are stored on the
+ * kernel stack during a system call or other kernel entry.
+ */
+struct pt_dspregs {
+        unsigned long   a1;
+        unsigned long   a0g;
+        unsigned long   a1g;
+        unsigned long   m0;
+        unsigned long   m1;
+        unsigned long   a0;
+        unsigned long   x0;
+        unsigned long   x1;
+        unsigned long   y0;
+        unsigned long   y1;
+        unsigned long   dsr;
+        unsigned long   rs;
+        unsigned long   re;
+        unsigned long   mod;
+};
+
+
+#endif /* _UAPI__ASM_SH_PTRACE_32_H */
diff --git a/arch/sh/include/uapi/asm/ptrace_64.h b/arch/sh/include/uapi/asm/ptrace_64.h
new file mode 100644
index 000000000000..0e52ee83e946
--- /dev/null
+++ b/arch/sh/include/uapi/asm/ptrace_64.h
@@ -0,0 +1,14 @@
+#ifndef _UAPI__ASM_SH_PTRACE_64_H
+#define _UAPI__ASM_SH_PTRACE_64_H
+
+struct pt_regs {
+        unsigned long long pc;
+        unsigned long long sr;
+        long long syscall_nr;
+        unsigned long long regs[63];
+        unsigned long long tregs[8];
+        unsigned long long pad[2];
+};
+
+
+#endif /* _UAPI__ASM_SH_PTRACE_64_H */
diff --git a/arch/sh/include/uapi/asm/setup.h b/arch/sh/include/uapi/asm/setup.h
new file mode 100644
index 000000000000..552df83f1a49
--- /dev/null
+++ b/arch/sh/include/uapi/asm/setup.h
@@ -0,0 +1 @@
+#include <asm-generic/setup.h>
diff --git a/arch/sh/include/asm/sigcontext.h b/arch/sh/include/uapi/asm/sigcontext.h
index 8ce1435bc0bf..8ce1435bc0bf 100644
--- a/arch/sh/include/asm/sigcontext.h
+++ b/arch/sh/include/uapi/asm/sigcontext.h
diff --git a/arch/sh/include/asm/signal.h b/arch/sh/include/uapi/asm/signal.h
index 9ac530a90bce..9ac530a90bce 100644
--- a/arch/sh/include/asm/signal.h
+++ b/arch/sh/include/uapi/asm/signal.h
diff --git a/arch/sh/include/asm/sockios.h b/arch/sh/include/uapi/asm/sockios.h
index cf8b96b1f9ab..cf8b96b1f9ab 100644
--- a/arch/sh/include/asm/sockios.h
+++ b/arch/sh/include/uapi/asm/sockios.h
diff --git a/arch/sh/include/asm/stat.h b/arch/sh/include/uapi/asm/stat.h
index e1810cc6e3da..e1810cc6e3da 100644
--- a/arch/sh/include/asm/stat.h
+++ b/arch/sh/include/uapi/asm/stat.h
diff --git a/arch/sh/include/asm/swab.h b/arch/sh/include/uapi/asm/swab.h
index 1cd09767a7a3..1cd09767a7a3 100644
--- a/arch/sh/include/asm/swab.h
+++ b/arch/sh/include/uapi/asm/swab.h
diff --git a/arch/sh/include/uapi/asm/types.h b/arch/sh/include/uapi/asm/types.h
new file mode 100644
index 000000000000..b9e79bc580dd
--- /dev/null
+++ b/arch/sh/include/uapi/asm/types.h
@@ -0,0 +1 @@
+#include <asm-generic/types.h>
diff --git a/arch/sh/include/uapi/asm/unistd.h b/arch/sh/include/uapi/asm/unistd.h
new file mode 100644
index 000000000000..eeef88dd53ce
--- /dev/null
+++ b/arch/sh/include/uapi/asm/unistd.h
@@ -0,0 +1,7 @@
+#ifndef __KERNEL__
+# ifdef __SH5__
+#  include <asm/unistd_64.h>
+# else
+#  include <asm/unistd_32.h>
+# endif
+#endif
diff --git a/arch/sh/include/asm/unistd_32.h b/arch/sh/include/uapi/asm/unistd_32.h
index 72fd1e061006..9e465f246dc1 100644
--- a/arch/sh/include/asm/unistd_32.h
+++ b/arch/sh/include/uapi/asm/unistd_32.h
@@ -378,7 +378,8 @@
 #define __NR_setns              364
 #define __NR_process_vm_readv   365
 #define __NR_process_vm_writev  366
+#define __NR_kcmp               367
 
-#define NR_syscalls 367
+#define NR_syscalls 368
 
 #endif /* __ASM_SH_UNISTD_32_H */
diff --git a/arch/sh/include/asm/unistd_64.h b/arch/sh/include/uapi/asm/unistd_64.h
index a28edc329692..8e3a2edd284e 100644
--- a/arch/sh/include/asm/unistd_64.h
+++ b/arch/sh/include/uapi/asm/unistd_64.h
@@ -398,7 +398,8 @@
 #define __NR_setns              375
 #define __NR_process_vm_readv   376
 #define __NR_process_vm_writev  377
+#define __NR_kcmp               378
 
-#define NR_syscalls 378
+#define NR_syscalls 379
 
 #endif /* __ASM_SH_UNISTD_64_H */
diff --git a/arch/sh/kernel/syscalls_32.S b/arch/sh/kernel/syscalls_32.S
index 4b68f0f79761..fe97ae5e56f1 100644
--- a/arch/sh/kernel/syscalls_32.S
+++ b/arch/sh/kernel/syscalls_32.S
@@ -384,3 +384,4 @@ ENTRY(sys_call_table)
         .long sys_setns
         .long sys_process_vm_readv      /* 365 */
         .long sys_process_vm_writev
+        .long sys_kcmp
diff --git a/arch/sh/kernel/syscalls_64.S b/arch/sh/kernel/syscalls_64.S
index 0956345b36ef..5c7b1c67bdc1 100644
--- a/arch/sh/kernel/syscalls_64.S
+++ b/arch/sh/kernel/syscalls_64.S
@@ -404,3 +404,4 @@ sys_call_table:
         .long sys_setns                 /* 375 */
         .long sys_process_vm_readv
         .long sys_process_vm_writev
+        .long sys_kcmp
diff --git a/arch/sparc/include/asm/ptrace.h b/arch/sparc/include/asm/ptrace.h
index 0c6f6b068289..da43bdc62294 100644
--- a/arch/sparc/include/asm/ptrace.h
+++ b/arch/sparc/include/asm/ptrace.h
@@ -42,7 +42,18 @@ struct global_reg_snapshot {
         struct thread_info      *thread;
         unsigned long           pad1;
 };
-extern struct global_reg_snapshot global_reg_snapshot[NR_CPUS];
+
+struct global_pmu_snapshot {
+        unsigned long           pcr[4];
+        unsigned long           pic[4];
+};
+
+union global_cpu_snapshot {
+        struct global_reg_snapshot      reg;
+        struct global_pmu_snapshot      pmu;
+};
+
+extern union global_cpu_snapshot global_cpu_snapshot[NR_CPUS];
 
 #define force_successful_syscall_return()           \
 do {    current_thread_info()->syscall_noerror = 1; \
diff --git a/arch/sparc/include/asm/smp_64.h b/arch/sparc/include/asm/smp_64.h
index 29862a9e9065..dd3bef4b9896 100644
--- a/arch/sparc/include/asm/smp_64.h
+++ b/arch/sparc/include/asm/smp_64.h
@@ -48,6 +48,7 @@ extern void smp_fill_in_sib_core_maps(void);
 extern void cpu_play_dead(void);
 
 extern void smp_fetch_global_regs(void);
+extern void smp_fetch_global_pmu(void);
 
 struct seq_file;
 void smp_bogo(struct seq_file *);
@@ -65,6 +66,7 @@ extern void __cpu_die(unsigned int cpu);
 #define hard_smp_processor_id()         0
 #define smp_fill_in_sib_core_maps() do { } while (0)
 #define smp_fetch_global_regs() do { } while (0)
+#define smp_fetch_global_pmu() do { } while (0)
 
 #endif /* !(CONFIG_SMP) */
 
diff --git a/arch/sparc/include/uapi/asm/sigcontext.h b/arch/sparc/include/uapi/asm/sigcontext.h
index e69de29bb2d1..ae5704fa77ad 100644
--- a/arch/sparc/include/uapi/asm/sigcontext.h
+++ b/arch/sparc/include/uapi/asm/sigcontext.h
@@ -0,0 +1,4 @@
+/*
+ * There isn't anything here anymore, but the file must not be empty or patch
+ * will delete it.
+ */
diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c
index e48651dace1b..885a8af74064 100644
--- a/arch/sparc/kernel/perf_event.c
+++ b/arch/sparc/kernel/perf_event.c
@@ -817,15 +817,17 @@ static u64 nop_for_index(int idx)
 
 static inline void sparc_pmu_enable_event(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc, int idx)
 {
-        u64 val, mask = mask_for_index(idx);
+        u64 enc, val, mask = mask_for_index(idx);
         int pcr_index = 0;
 
         if (sparc_pmu->num_pcrs > 1)
                 pcr_index = idx;
 
+        enc = perf_event_get_enc(cpuc->events[idx]);
+
         val = cpuc->pcr[pcr_index];
         val &= ~mask;
-        val |= hwc->config;
+        val |= event_encoding(enc, idx);
         cpuc->pcr[pcr_index] = val;
 
         pcr_ops->write_pcr(pcr_index, cpuc->pcr[pcr_index]);
@@ -1738,8 +1740,6 @@ static void perf_callchain_user_64(struct perf_callchain_entry *entry,
 {
         unsigned long ufp;
 
-        perf_callchain_store(entry, regs->tpc);
-
         ufp = regs->u_regs[UREG_I6] + STACK_BIAS;
         do {
                 struct sparc_stackf *usf, sf;
@@ -1760,8 +1760,6 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry,
 {
         unsigned long ufp;
 
-        perf_callchain_store(entry, regs->tpc);
-
         ufp = regs->u_regs[UREG_I6] & 0xffffffffUL;
         do {
                 struct sparc_stackf32 *usf, sf;
@@ -1780,6 +1778,11 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry,
 void
 perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs)
 {
+        perf_callchain_store(entry, regs->tpc);
+
+        if (!current->mm)
+                return;
+
         flushw_user();
         if (test_thread_flag(TIF_32BIT))
                 perf_callchain_user_32(entry, regs);
diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c
index fcaa59421126..d778248ef3f8 100644
--- a/arch/sparc/kernel/process_64.c
+++ b/arch/sparc/kernel/process_64.c
@@ -27,6 +27,7 @@
 #include <linux/tick.h>
 #include <linux/init.h>
 #include <linux/cpu.h>
+#include <linux/perf_event.h>
 #include <linux/elfcore.h>
 #include <linux/sysrq.h>
 #include <linux/nmi.h>
@@ -47,6 +48,7 @@
 #include <asm/syscalls.h>
 #include <asm/irq_regs.h>
 #include <asm/smp.h>
+#include <asm/pcr.h>
 
 #include "kstack.h"
 
@@ -204,18 +206,22 @@ void show_regs(struct pt_regs *regs)
         show_stack(current, (unsigned long *) regs->u_regs[UREG_FP]);
 }
 
-struct global_reg_snapshot global_reg_snapshot[NR_CPUS];
-static DEFINE_SPINLOCK(global_reg_snapshot_lock);
+union global_cpu_snapshot global_cpu_snapshot[NR_CPUS];
+static DEFINE_SPINLOCK(global_cpu_snapshot_lock);
 
 static void __global_reg_self(struct thread_info *tp, struct pt_regs *regs,
                               int this_cpu)
 {
+        struct global_reg_snapshot *rp;
+
         flushw_all();
 
-        global_reg_snapshot[this_cpu].tstate = regs->tstate;
-        global_reg_snapshot[this_cpu].tpc = regs->tpc;
-        global_reg_snapshot[this_cpu].tnpc = regs->tnpc;
-        global_reg_snapshot[this_cpu].o7 = regs->u_regs[UREG_I7];
+        rp = &global_cpu_snapshot[this_cpu].reg;
+
+        rp->tstate = regs->tstate;
+        rp->tpc = regs->tpc;
+        rp->tnpc = regs->tnpc;
+        rp->o7 = regs->u_regs[UREG_I7];
 
         if (regs->tstate & TSTATE_PRIV) {
                 struct reg_window *rw;
@@ -223,17 +229,17 @@ static void __global_reg_self(struct thread_info *tp, struct pt_regs *regs,
                 rw = (struct reg_window *)
                         (regs->u_regs[UREG_FP] + STACK_BIAS);
                 if (kstack_valid(tp, (unsigned long) rw)) {
-                        global_reg_snapshot[this_cpu].i7 = rw->ins[7];
+                        rp->i7 = rw->ins[7];
                         rw = (struct reg_window *)
                                 (rw->ins[6] + STACK_BIAS);
                         if (kstack_valid(tp, (unsigned long) rw))
-                                global_reg_snapshot[this_cpu].rpc = rw->ins[7];
+                                rp->rpc = rw->ins[7];
                 }
         } else {
-                global_reg_snapshot[this_cpu].i7 = 0;
-                global_reg_snapshot[this_cpu].rpc = 0;
+                rp->i7 = 0;
+                rp->rpc = 0;
         }
-        global_reg_snapshot[this_cpu].thread = tp;
+        rp->thread = tp;
 }
 
 /* In order to avoid hangs we do not try to synchronize with the
@@ -261,9 +267,9 @@ void arch_trigger_all_cpu_backtrace(void)
         if (!regs)
                 regs = tp->kregs;
 
-        spin_lock_irqsave(&global_reg_snapshot_lock, flags);
+        spin_lock_irqsave(&global_cpu_snapshot_lock, flags);
 
-        memset(global_reg_snapshot, 0, sizeof(global_reg_snapshot));
+        memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot));
 
         this_cpu = raw_smp_processor_id();
 
@@ -272,7 +278,7 @@ void arch_trigger_all_cpu_backtrace(void)
         smp_fetch_global_regs();
 
         for_each_online_cpu(cpu) {
-                struct global_reg_snapshot *gp = &global_reg_snapshot[cpu];
+                struct global_reg_snapshot *gp = &global_cpu_snapshot[cpu].reg;
 
                 __global_reg_poll(gp);
 
@@ -295,9 +301,9 @@ void arch_trigger_all_cpu_backtrace(void)
                 }
         }
 
-        memset(global_reg_snapshot, 0, sizeof(global_reg_snapshot));
+        memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot));
 
-        spin_unlock_irqrestore(&global_reg_snapshot_lock, flags);
+        spin_unlock_irqrestore(&global_cpu_snapshot_lock, flags);
 }
 
 #ifdef CONFIG_MAGIC_SYSRQ
@@ -309,16 +315,90 @@ static void sysrq_handle_globreg(int key)
 
 static struct sysrq_key_op sparc_globalreg_op = {
         .handler        = sysrq_handle_globreg,
-        .help_msg       = "Globalregs",
+        .help_msg       = "global-regs(Y)",
         .action_msg     = "Show Global CPU Regs",
 };
 
-static int __init sparc_globreg_init(void)
+static void __global_pmu_self(int this_cpu)
+{
+        struct global_pmu_snapshot *pp;
+        int i, num;
+
+        pp = &global_cpu_snapshot[this_cpu].pmu;
+
+        num = 1;
+        if (tlb_type == hypervisor &&
+            sun4v_chip_type >= SUN4V_CHIP_NIAGARA4)
+                num = 4;
+
+        for (i = 0; i < num; i++) {
+                pp->pcr[i] = pcr_ops->read_pcr(i);
+                pp->pic[i] = pcr_ops->read_pic(i);
+        }
+}
+
+static void __global_pmu_poll(struct global_pmu_snapshot *pp)
+{
+        int limit = 0;
+
+        while (!pp->pcr[0] && ++limit < 100) {
+                barrier();
+                udelay(1);
+        }
+}
+
+static void pmu_snapshot_all_cpus(void)
 {
-        return register_sysrq_key('y', &sparc_globalreg_op);
+        unsigned long flags;
+        int this_cpu, cpu;
+
+        spin_lock_irqsave(&global_cpu_snapshot_lock, flags);
+
+        memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot));
+
+        this_cpu = raw_smp_processor_id();
+
+        __global_pmu_self(this_cpu);
+
+        smp_fetch_global_pmu();
+
+        for_each_online_cpu(cpu) {
+                struct global_pmu_snapshot *pp = &global_cpu_snapshot[cpu].pmu;
+
+                __global_pmu_poll(pp);
+
+                printk("%c CPU[%3d]: PCR[%08lx:%08lx:%08lx:%08lx] PIC[%08lx:%08lx:%08lx:%08lx]\n",
+                       (cpu == this_cpu ? '*' : ' '), cpu,
+                       pp->pcr[0], pp->pcr[1], pp->pcr[2], pp->pcr[3],
+                       pp->pic[0], pp->pic[1], pp->pic[2], pp->pic[3]);
+        }
+
+        memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot));
+
+        spin_unlock_irqrestore(&global_cpu_snapshot_lock, flags);
+}
+
+static void sysrq_handle_globpmu(int key)
+{
+        pmu_snapshot_all_cpus();
+}
+
+static struct sysrq_key_op sparc_globalpmu_op = {
+        .handler        = sysrq_handle_globpmu,
+        .help_msg       = "global-pmu(X)",
+        .action_msg     = "Show Global PMU Regs",
+};
+
+static int __init sparc_sysrq_init(void)
+{
+        int ret = register_sysrq_key('y', &sparc_globalreg_op);
+
+        if (!ret)
+                ret = register_sysrq_key('x', &sparc_globalpmu_op);
+        return ret;
 }
 
-core_initcall(sparc_globreg_init);
+core_initcall(sparc_sysrq_init);
 
 #endif
 
diff --git a/arch/sparc/kernel/smp_64.c b/arch/sparc/kernel/smp_64.c
index 781bcb10b8bd..d94b878577b7 100644
--- a/arch/sparc/kernel/smp_64.c
+++ b/arch/sparc/kernel/smp_64.c
@@ -852,6 +852,8 @@ extern unsigned long xcall_flush_tlb_mm;
 extern unsigned long xcall_flush_tlb_pending;
 extern unsigned long xcall_flush_tlb_kernel_range;
 extern unsigned long xcall_fetch_glob_regs;
+extern unsigned long xcall_fetch_glob_pmu;
+extern unsigned long xcall_fetch_glob_pmu_n4;
 extern unsigned long xcall_receive_signal;
 extern unsigned long xcall_new_mmu_context_version;
 #ifdef CONFIG_KGDB
@@ -1000,6 +1002,15 @@ void smp_fetch_global_regs(void)
         smp_cross_call(&xcall_fetch_glob_regs, 0, 0, 0);
 }
 
+void smp_fetch_global_pmu(void)
+{
+        if (tlb_type == hypervisor &&
+            sun4v_chip_type >= SUN4V_CHIP_NIAGARA4)
+                smp_cross_call(&xcall_fetch_glob_pmu_n4, 0, 0, 0);
+        else
+                smp_cross_call(&xcall_fetch_glob_pmu, 0, 0, 0);
+}
+
 /* We know that the window frames of the user have been flushed
  * to the stack before we get here because all callers of us
  * are flush_tlb_*() routines, and these run after flush_cache_*()
diff --git a/arch/sparc/mm/ultra.S b/arch/sparc/mm/ultra.S
index 874162a11ceb..f8e13d421fcb 100644
--- a/arch/sparc/mm/ultra.S
+++ b/arch/sparc/mm/ultra.S
@@ -481,8 +481,8 @@ xcall_sync_tick:
 
         .globl          xcall_fetch_glob_regs
 xcall_fetch_glob_regs:
-        sethi           %hi(global_reg_snapshot), %g1
-        or              %g1, %lo(global_reg_snapshot), %g1
+        sethi           %hi(global_cpu_snapshot), %g1
+        or              %g1, %lo(global_cpu_snapshot), %g1
         __GET_CPUID(%g2)
         sllx            %g2, 6, %g3
         add             %g1, %g3, %g1
@@ -509,6 +509,66 @@ xcall_fetch_glob_regs:
         stx             %g3, [%g1 + GR_SNAP_THREAD]
         retry
 
+        .globl          xcall_fetch_glob_pmu
+xcall_fetch_glob_pmu:
+        sethi           %hi(global_cpu_snapshot), %g1
+        or              %g1, %lo(global_cpu_snapshot), %g1
+        __GET_CPUID(%g2)
+        sllx            %g2, 6, %g3
+        add             %g1, %g3, %g1
+        rd              %pic, %g7
+        stx             %g7, [%g1 + (4 * 8)]
+        rd              %pcr, %g7
+        stx             %g7, [%g1 + (0 * 8)]
+        retry
+
+        .globl          xcall_fetch_glob_pmu_n4
+xcall_fetch_glob_pmu_n4:
+        sethi           %hi(global_cpu_snapshot), %g1
+        or              %g1, %lo(global_cpu_snapshot), %g1
+        __GET_CPUID(%g2)
+        sllx            %g2, 6, %g3
+        add             %g1, %g3, %g1
+
+        ldxa            [%g0] ASI_PIC, %g7
+        stx             %g7, [%g1 + (4 * 8)]
+        mov             0x08, %g3
+        ldxa            [%g3] ASI_PIC, %g7
+        stx             %g7, [%g1 + (5 * 8)]
+        mov             0x10, %g3
+        ldxa            [%g3] ASI_PIC, %g7
+        stx             %g7, [%g1 + (6 * 8)]
+        mov             0x18, %g3
+        ldxa            [%g3] ASI_PIC, %g7
+        stx             %g7, [%g1 + (7 * 8)]
+
+        mov             %o0, %g2
+        mov             %o1, %g3
+        mov             %o5, %g7
+
+        mov             HV_FAST_VT_GET_PERFREG, %o5
+        mov             3, %o0
+        ta              HV_FAST_TRAP
+        stx             %o1, [%g1 + (3 * 8)]
+        mov             HV_FAST_VT_GET_PERFREG, %o5
+        mov             2, %o0
+        ta              HV_FAST_TRAP
+        stx             %o1, [%g1 + (2 * 8)]
+        mov             HV_FAST_VT_GET_PERFREG, %o5
+        mov             1, %o0
+        ta              HV_FAST_TRAP
+        stx             %o1, [%g1 + (1 * 8)]
+        mov             HV_FAST_VT_GET_PERFREG, %o5
+        mov             0, %o0
+        ta              HV_FAST_TRAP
+        stx             %o1, [%g1 + (0 * 8)]
+
+        mov             %g2, %o0
+        mov             %g3, %o1
+        mov             %g7, %o5
+
+        retry
+
 #ifdef DCACHE_ALIASING_POSSIBLE
         .align          32
         .globl          xcall_flush_dcache_page_cheetah
diff --git a/arch/tile/include/arch/Kbuild b/arch/tile/include/arch/Kbuild
index e69de29bb2d1..3751c9fabcf2 100644
--- a/arch/tile/include/arch/Kbuild
+++ b/arch/tile/include/arch/Kbuild
@@ -0,0 +1 @@
+# Tile arch headers
diff --git a/arch/tile/include/asm/Kbuild b/arch/tile/include/asm/Kbuild
index c68808a09da7..6948015e08a2 100644
--- a/arch/tile/include/asm/Kbuild
+++ b/arch/tile/include/asm/Kbuild
@@ -1,8 +1,6 @@
 
 header-y += ../arch/
 
-header-y += ucontext.h
-
 generic-y += bug.h
 generic-y += bugs.h
 generic-y += clkdev.h
@@ -37,5 +35,4 @@ generic-y += statfs.h
 generic-y += termbits.h
 generic-y += termios.h
 generic-y += types.h
-generic-y += ucontext.h
 generic-y += xor.h
diff --git a/arch/tile/include/uapi/asm/Kbuild b/arch/tile/include/uapi/asm/Kbuild
index 5c6915fd30b5..c20db8e428bf 100644
--- a/arch/tile/include/uapi/asm/Kbuild
+++ b/arch/tile/include/uapi/asm/Kbuild
@@ -15,4 +15,7 @@ header-y += siginfo.h
 header-y += signal.h
 header-y += stat.h
 header-y += swab.h
+header-y += ucontext.h
 header-y += unistd.h
+
+generic-y += ucontext.h
diff --git a/arch/x86/crypto/aesni-intel_glue.c b/arch/x86/crypto/aesni-intel_glue.c
index 7c04d0da709b..1b9c22bea8a7 100644
--- a/arch/x86/crypto/aesni-intel_glue.c
+++ b/arch/x86/crypto/aesni-intel_glue.c
@@ -515,6 +515,11 @@ static int xts_aesni_setkey(struct crypto_tfm *tfm, const u8 *key,
 }
 
 
+static void aesni_xts_tweak(void *ctx, u8 *out, const u8 *in)
+{
+        aesni_enc(ctx, out, in);
+}
+
 static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
                        struct scatterlist *src, unsigned int nbytes)
 {
@@ -525,7 +530,7 @@ static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
                 .tbuflen = sizeof(buf),
 
                 .tweak_ctx = aes_ctx(ctx->raw_tweak_ctx),
-                .tweak_fn = XTS_TWEAK_CAST(aesni_enc),
+                .tweak_fn = aesni_xts_tweak,
                 .crypt_ctx = aes_ctx(ctx->raw_crypt_ctx),
                 .crypt_fn = lrw_xts_encrypt_callback,
         };
@@ -550,7 +555,7 @@ static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
                 .tbuflen = sizeof(buf),
 
                 .tweak_ctx = aes_ctx(ctx->raw_tweak_ctx),
-                .tweak_fn = XTS_TWEAK_CAST(aesni_enc),
+                .tweak_fn = aesni_xts_tweak,
                 .crypt_ctx = aes_ctx(ctx->raw_crypt_ctx),
                 .crypt_fn = lrw_xts_decrypt_callback,
         };