Merge commit 'v2.6.35-rc1' into amd-iommu/2.6.35

74 files changed, 4662 insertions, 4062 deletions

diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index 4c58352209e..e77b2208372 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -47,8 +47,6 @@ obj-$(CONFIG_X86_TRAMPOLINE)	+= trampoline.o
 obj-y				+= process.o
 obj-y				+= i387.o xsave.o
 obj-y				+= ptrace.o
-obj-$(CONFIG_X86_DS)		+= ds.o
-obj-$(CONFIG_X86_DS_SELFTEST)		+= ds_selftest.o
 obj-$(CONFIG_X86_32)		+= tls.o
 obj-$(CONFIG_IA32_EMULATION)	+= tls.o
 obj-y				+= step.o
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index cd40aba6aa9..60cc4058ed5 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -63,7 +63,6 @@ EXPORT_SYMBOL(acpi_disabled);
 int acpi_noirq;				/* skip ACPI IRQ initialization */
 int acpi_pci_disabled;		/* skip ACPI PCI scan and IRQ initialization */
 EXPORT_SYMBOL(acpi_pci_disabled);
-int acpi_ht __initdata = 1;	/* enable HT */
 
 int acpi_lapic;
 int acpi_ioapic;
@@ -94,6 +93,53 @@ enum acpi_irq_model_id acpi_irq_model = ACPI_IRQ_MODEL_PIC;
 
 
 /*
+ * ISA irqs by default are the first 16 gsis but can be
+ * any gsi as specified by an interrupt source override.
+ */
+static u32 isa_irq_to_gsi[NR_IRQS_LEGACY] __read_mostly = {
+	0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
+};
+
+static unsigned int gsi_to_irq(unsigned int gsi)
+{
+	unsigned int irq = gsi + NR_IRQS_LEGACY;
+	unsigned int i;
+
+	for (i = 0; i < NR_IRQS_LEGACY; i++) {
+		if (isa_irq_to_gsi[i] == gsi) {
+			return i;
+		}
+	}
+
+	/* Provide an identity mapping of gsi == irq
+	 * except on truly weird platforms that have
+	 * non isa irqs in the first 16 gsis.
+	 */
+	if (gsi >= NR_IRQS_LEGACY)
+		irq = gsi;
+	else
+		irq = gsi_end + 1 + gsi;
+
+	return irq;
+}
+
+static u32 irq_to_gsi(int irq)
+{
+	unsigned int gsi;
+
+	if (irq < NR_IRQS_LEGACY)
+		gsi = isa_irq_to_gsi[irq];
+	else if (irq <= gsi_end)
+		gsi = irq;
+	else if (irq <= (gsi_end + NR_IRQS_LEGACY))
+		gsi = irq - gsi_end;
+	else
+		gsi = 0xffffffff;
+
+	return gsi;
+}
+
+/*
  * Temporarily use the virtual area starting from FIX_IO_APIC_BASE_END,
  * to map the target physical address. The problem is that set_fixmap()
  * provides a single page, and it is possible that the page is not
@@ -313,7 +359,7 @@ acpi_parse_ioapic(struct acpi_subtable_header * header, const unsigned long end)
 /*
  * Parse Interrupt Source Override for the ACPI SCI
  */
-static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
+static void __init acpi_sci_ioapic_setup(u8 bus_irq, u16 polarity, u16 trigger, u32 gsi)
 {
 	if (trigger == 0)	/* compatible SCI trigger is level */
 		trigger = 3;
@@ -333,7 +379,7 @@ static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
 	 * If GSI is < 16, this will update its flags,
 	 * else it will create a new mp_irqs[] entry.
 	 */
-	mp_override_legacy_irq(gsi, polarity, trigger, gsi);
+	mp_override_legacy_irq(bus_irq, polarity, trigger, gsi);
 
 	/*
 	 * stash over-ride to indicate we've been here
@@ -357,9 +403,10 @@ acpi_parse_int_src_ovr(struct acpi_subtable_header * header,
 	acpi_table_print_madt_entry(header);
 
 	if (intsrc->source_irq == acpi_gbl_FADT.sci_interrupt) {
-		acpi_sci_ioapic_setup(intsrc->global_irq,
+		acpi_sci_ioapic_setup(intsrc->source_irq,
 				      intsrc->inti_flags & ACPI_MADT_POLARITY_MASK,
-				      (intsrc->inti_flags & ACPI_MADT_TRIGGER_MASK) >> 2);
+				      (intsrc->inti_flags & ACPI_MADT_TRIGGER_MASK) >> 2,
+				      intsrc->global_irq);
 		return 0;
 	}
 
@@ -448,7 +495,7 @@ void __init acpi_pic_sci_set_trigger(unsigned int irq, u16 trigger)
 
 int acpi_gsi_to_irq(u32 gsi, unsigned int *irq)
 {
-	*irq = gsi;
+	*irq = gsi_to_irq(gsi);
 
 #ifdef CONFIG_X86_IO_APIC
 	if (acpi_irq_model == ACPI_IRQ_MODEL_IOAPIC)
@@ -458,6 +505,14 @@ int acpi_gsi_to_irq(u32 gsi, unsigned int *irq)
 	return 0;
 }
 
+int acpi_isa_irq_to_gsi(unsigned isa_irq, u32 *gsi)
+{
+	if (isa_irq >= 16)
+		return -1;
+	*gsi = irq_to_gsi(isa_irq);
+	return 0;
+}
+
 /*
  * success: return IRQ number (>=0)
  * failure: return < 0
@@ -482,7 +537,7 @@ int acpi_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity)
 		plat_gsi = mp_register_gsi(dev, gsi, trigger, polarity);
 	}
 #endif
-	irq = plat_gsi;
+	irq = gsi_to_irq(plat_gsi);
 
 	return irq;
 }
@@ -867,29 +922,6 @@ static int __init acpi_parse_madt_lapic_entries(void)
 extern int es7000_plat;
 #endif
 
-int __init acpi_probe_gsi(void)
-{
-	int idx;
-	int gsi;
-	int max_gsi = 0;
-
-	if (acpi_disabled)
-		return 0;
-
-	if (!acpi_ioapic)
-		return 0;
-
-	max_gsi = 0;
-	for (idx = 0; idx < nr_ioapics; idx++) {
-		gsi = mp_gsi_routing[idx].gsi_end;
-
-		if (gsi > max_gsi)
-			max_gsi = gsi;
-	}
-
-	return max_gsi + 1;
-}
-
 static void assign_to_mp_irq(struct mpc_intsrc *m,
 				    struct mpc_intsrc *mp_irq)
 {
@@ -947,13 +979,13 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi)
 	mp_irq.dstirq = pin;	/* INTIN# */
 
 	save_mp_irq(&mp_irq);
+
+	isa_irq_to_gsi[bus_irq] = gsi;
 }
 
 void __init mp_config_acpi_legacy_irqs(void)
 {
 	int i;
-	int ioapic;
-	unsigned int dstapic;
 	struct mpc_intsrc mp_irq;
 
 #if defined (CONFIG_MCA) || defined (CONFIG_EISA)
@@ -974,19 +1006,27 @@ void __init mp_config_acpi_legacy_irqs(void)
 #endif
 
 	/*
-	 * Locate the IOAPIC that manages the ISA IRQs (0-15).
-	 */
-	ioapic = mp_find_ioapic(0);
-	if (ioapic < 0)
-		return;
-	dstapic = mp_ioapics[ioapic].apicid;
-
-	/*
 	 * Use the default configuration for the IRQs 0-15.  Unless
 	 * overridden by (MADT) interrupt source override entries.
 	 */
 	for (i = 0; i < 16; i++) {
+		int ioapic, pin;
+		unsigned int dstapic;
 		int idx;
+		u32 gsi;
+
+		/* Locate the gsi that irq i maps to. */
+		if (acpi_isa_irq_to_gsi(i, &gsi))
+			continue;
+
+		/*
+		 * Locate the IOAPIC that manages the ISA IRQ.
+		 */
+		ioapic = mp_find_ioapic(gsi);
+		if (ioapic < 0)
+			continue;
+		pin = mp_find_ioapic_pin(ioapic, gsi);
+		dstapic = mp_ioapics[ioapic].apicid;
 
 		for (idx = 0; idx < mp_irq_entries; idx++) {
 			struct mpc_intsrc *irq = mp_irqs + idx;
@@ -996,7 +1036,7 @@ void __init mp_config_acpi_legacy_irqs(void)
 				break;
 
 			/* Do we already have a mapping for this IOAPIC pin */
-			if (irq->dstapic == dstapic && irq->dstirq == i)
+			if (irq->dstapic == dstapic && irq->dstirq == pin)
 				break;
 		}
 
@@ -1011,7 +1051,7 @@ void __init mp_config_acpi_legacy_irqs(void)
 		mp_irq.dstapic = dstapic;
 		mp_irq.irqtype = mp_INT;
 		mp_irq.srcbusirq = i; /* Identity mapped */
-		mp_irq.dstirq = i;
+		mp_irq.dstirq = pin;
 
 		save_mp_irq(&mp_irq);
 	}
@@ -1076,11 +1116,6 @@ int mp_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity)
 
 	ioapic_pin = mp_find_ioapic_pin(ioapic, gsi);
 
-#ifdef CONFIG_X86_32
-	if (ioapic_renumber_irq)
-		gsi = ioapic_renumber_irq(ioapic, gsi);
-#endif
-
 	if (ioapic_pin > MP_MAX_IOAPIC_PIN) {
 		printk(KERN_ERR "Invalid reference to IOAPIC pin "
 		       "%d-%d\n", mp_ioapics[ioapic].apicid,
@@ -1094,7 +1129,7 @@ int mp_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity)
 	set_io_apic_irq_attr(&irq_attr, ioapic, ioapic_pin,
 			     trigger == ACPI_EDGE_SENSITIVE ? 0 : 1,
 			     polarity == ACPI_ACTIVE_HIGH ? 0 : 1);
-	io_apic_set_pci_routing(dev, gsi, &irq_attr);
+	io_apic_set_pci_routing(dev, gsi_to_irq(gsi), &irq_attr);
 
 	return gsi;
 }
@@ -1154,7 +1189,8 @@ static int __init acpi_parse_madt_ioapic_entries(void)
 	 * pretend we got one so we can set the SCI flags.
 	 */
 	if (!acpi_sci_override_gsi)
-		acpi_sci_ioapic_setup(acpi_gbl_FADT.sci_interrupt, 0, 0);
+		acpi_sci_ioapic_setup(acpi_gbl_FADT.sci_interrupt, 0, 0,
+				      acpi_gbl_FADT.sci_interrupt);
 
 	/* Fill in identity legacy mappings where no override */
 	mp_config_acpi_legacy_irqs();
@@ -1464,9 +1500,8 @@ void __init acpi_boot_table_init(void)
 
 	/*
 	 * If acpi_disabled, bail out
-	 * One exception: acpi=ht continues far enough to enumerate LAPICs
 	 */
-	if (acpi_disabled && !acpi_ht)
+	if (acpi_disabled)
 		return; 
 
 	/*
@@ -1497,9 +1532,8 @@ int __init early_acpi_boot_init(void)
 {
 	/*
 	 * If acpi_disabled, bail out
-	 * One exception: acpi=ht continues far enough to enumerate LAPICs
 	 */
-	if (acpi_disabled && !acpi_ht)
+	if (acpi_disabled)
 		return 1;
 
 	/*
@@ -1517,9 +1551,8 @@ int __init acpi_boot_init(void)
 
 	/*
 	 * If acpi_disabled, bail out
-	 * One exception: acpi=ht continues far enough to enumerate LAPICs
 	 */
-	if (acpi_disabled && !acpi_ht)
+	if (acpi_disabled)
 		return 1;
 
 	acpi_table_parse(ACPI_SIG_BOOT, acpi_parse_sbf);
@@ -1554,21 +1587,12 @@ static int __init parse_acpi(char *arg)
 	/* acpi=force to over-ride black-list */
 	else if (strcmp(arg, "force") == 0) {
 		acpi_force = 1;
-		acpi_ht = 1;
 		acpi_disabled = 0;
 	}
 	/* acpi=strict disables out-of-spec workarounds */
 	else if (strcmp(arg, "strict") == 0) {
 		acpi_strict = 1;
 	}
-	/* Limit ACPI just to boot-time to enable HT */
-	else if (strcmp(arg, "ht") == 0) {
-		if (!acpi_force) {
-			printk(KERN_WARNING "acpi=ht will be removed in Linux-2.6.35\n");
-			disable_acpi();
-		}
-		acpi_ht = 1;
-	}
 	/* acpi=rsdt use RSDT instead of XSDT */
 	else if (strcmp(arg, "rsdt") == 0) {
 		acpi_rsdt_forced = 1;
@@ -1576,6 +1600,10 @@ static int __init parse_acpi(char *arg)
 	/* "acpi=noirq" disables ACPI interrupt routing */
 	else if (strcmp(arg, "noirq") == 0) {
 		acpi_noirq_set();
+	}
+	/* "acpi=copy_dsdt" copys DSDT */
+	else if (strcmp(arg, "copy_dsdt") == 0) {
+		acpi_gbl_copy_dsdt_locally = 1;
 	} else {
 		/* Core will printk when we return error. */
 		return -EINVAL;
diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c
index f9961034e55..82e508677b9 100644
--- a/arch/x86/kernel/acpi/sleep.c
+++ b/arch/x86/kernel/acpi/sleep.c
@@ -162,8 +162,6 @@ static int __init acpi_sleep_setup(char *str)
 #endif
 		if (strncmp(str, "old_ordering", 12) == 0)
 			acpi_old_suspend_ordering();
-		if (strncmp(str, "sci_force_enable", 16) == 0)
-			acpi_set_sci_en_on_resume();
 		str = strchr(str, ',');
 		if (str != NULL)
 			str += strspn(str, ", \t");
diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
index 1a160d5d44d..70237732a6c 100644
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -194,7 +194,7 @@ static void __init_or_module add_nops(void *insns, unsigned int len)
 }
 
 extern struct alt_instr __alt_instructions[], __alt_instructions_end[];
-extern u8 *__smp_locks[], *__smp_locks_end[];
+extern s32 __smp_locks[], __smp_locks_end[];
 static void *text_poke_early(void *addr, const void *opcode, size_t len);
 
 /* Replace instructions with better alternatives for this CPU type.
@@ -235,37 +235,41 @@ void __init_or_module apply_alternatives(struct alt_instr *start,
 
 #ifdef CONFIG_SMP
 
-static void alternatives_smp_lock(u8 **start, u8 **end, u8 *text, u8 *text_end)
+static void alternatives_smp_lock(const s32 *start, const s32 *end,
+				  u8 *text, u8 *text_end)
 {
-	u8 **ptr;
+	const s32 *poff;
 
 	mutex_lock(&text_mutex);
-	for (ptr = start; ptr < end; ptr++) {
-		if (*ptr < text)
-			continue;
-		if (*ptr > text_end)
+	for (poff = start; poff < end; poff++) {
+		u8 *ptr = (u8 *)poff + *poff;
+
+		if (!*poff || ptr < text || ptr >= text_end)
 			continue;
 		/* turn DS segment override prefix into lock prefix */
-		text_poke(*ptr, ((unsigned char []){0xf0}), 1);
+		if (*ptr == 0x3e)
+			text_poke(ptr, ((unsigned char []){0xf0}), 1);
 	};
 	mutex_unlock(&text_mutex);
 }
 
-static void alternatives_smp_unlock(u8 **start, u8 **end, u8 *text, u8 *text_end)
+static void alternatives_smp_unlock(const s32 *start, const s32 *end,
+				    u8 *text, u8 *text_end)
 {
-	u8 **ptr;
+	const s32 *poff;
 
 	if (noreplace_smp)
 		return;
 
 	mutex_lock(&text_mutex);
-	for (ptr = start; ptr < end; ptr++) {
-		if (*ptr < text)
-			continue;
-		if (*ptr > text_end)
+	for (poff = start; poff < end; poff++) {
+		u8 *ptr = (u8 *)poff + *poff;
+
+		if (!*poff || ptr < text || ptr >= text_end)
 			continue;
 		/* turn lock prefix into DS segment override prefix */
-		text_poke(*ptr, ((unsigned char []){0x3E}), 1);
+		if (*ptr == 0xf0)
+			text_poke(ptr, ((unsigned char []){0x3E}), 1);
 	};
 	mutex_unlock(&text_mutex);
 }
@@ -276,8 +280,8 @@ struct smp_alt_module {
 	char		*name;
 
 	/* ptrs to lock prefixes */
-	u8		**locks;
-	u8		**locks_end;
+	const s32	*locks;
+	const s32	*locks_end;
 
 	/* .text segment, needed to avoid patching init code ;) */
 	u8		*text;
@@ -398,16 +402,19 @@ void alternatives_smp_switch(int smp)
 int alternatives_text_reserved(void *start, void *end)
 {
 	struct smp_alt_module *mod;
-	u8 **ptr;
+	const s32 *poff;
 	u8 *text_start = start;
 	u8 *text_end = end;
 
 	list_for_each_entry(mod, &smp_alt_modules, next) {
 		if (mod->text > text_end || mod->text_end < text_start)
 			continue;
-		for (ptr = mod->locks; ptr < mod->locks_end; ptr++)
-			if (text_start <= *ptr && text_end >= *ptr)
+		for (poff = mod->locks; poff < mod->locks_end; poff++) {
+			const u8 *ptr = (const u8 *)poff + *poff;
+
+			if (text_start <= ptr && text_end > ptr)
 				return 1;
+		}
 	}
 
 	return 0;
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index e5a4a1e0161..c02cc692985 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -51,6 +51,7 @@
 #include <asm/smp.h>
 #include <asm/mce.h>
 #include <asm/kvm_para.h>
+#include <asm/tsc.h>
 
 unsigned int num_processors;
 
@@ -1151,8 +1152,13 @@ static void __cpuinit lapic_setup_esr(void)
  */
 void __cpuinit setup_local_APIC(void)
 {
-	unsigned int value;
-	int i, j;
+	unsigned int value, queued;
+	int i, j, acked = 0;
+	unsigned long long tsc = 0, ntsc;
+	long long max_loops = cpu_khz;
+
+	if (cpu_has_tsc)
+		rdtscll(tsc);
 
 	if (disable_apic) {
 		arch_disable_smp_support();
@@ -1204,13 +1210,32 @@ void __cpuinit setup_local_APIC(void)
 	 * the interrupt. Hence a vector might get locked. It was noticed
 	 * for timer irq (vector 0x31). Issue an extra EOI to clear ISR.
 	 */
-	for (i = APIC_ISR_NR - 1; i >= 0; i--) {
-		value = apic_read(APIC_ISR + i*0x10);
-		for (j = 31; j >= 0; j--) {
-			if (value & (1<<j))
-				ack_APIC_irq();
+	do {
+		queued = 0;
+		for (i = APIC_ISR_NR - 1; i >= 0; i--)
+			queued |= apic_read(APIC_IRR + i*0x10);
+
+		for (i = APIC_ISR_NR - 1; i >= 0; i--) {
+			value = apic_read(APIC_ISR + i*0x10);
+			for (j = 31; j >= 0; j--) {
+				if (value & (1<<j)) {
+					ack_APIC_irq();
+					acked++;
+				}
+			}
 		}
-	}
+		if (acked > 256) {
+			printk(KERN_ERR "LAPIC pending interrupts after %d EOI\n",
+			       acked);
+			break;
+		}
+		if (cpu_has_tsc) {
+			rdtscll(ntsc);
+			max_loops = (cpu_khz << 10) - (ntsc - tsc);
+		} else
+			max_loops--;
+	} while (queued && max_loops > 0);
+	WARN_ON(max_loops <= 0);
 
 	/*
 	 * Now that we are all set up, enable the APIC
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index 03ba1b895f5..425e53a87fe 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -131,24 +131,6 @@ int					es7000_plat;
 
 static unsigned int			base;
 
-static int
-es7000_rename_gsi(int ioapic, int gsi)
-{
-	if (es7000_plat == ES7000_ZORRO)
-		return gsi;
-
-	if (!base) {
-		int i;
-		for (i = 0; i < nr_ioapics; i++)
-			base += nr_ioapic_registers[i];
-	}
-
-	if (!ioapic && (gsi < 16))
-		gsi += base;
-
-	return gsi;
-}
-
 static int __cpuinit wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip)
 {
 	unsigned long vect = 0, psaival = 0;
@@ -190,7 +172,6 @@ static void setup_unisys(void)
 		es7000_plat = ES7000_ZORRO;
 	else
 		es7000_plat = ES7000_CLASSIC;
-	ioapic_renumber_irq = es7000_rename_gsi;
 }
 
 /*
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index eb2789c3f72..33f3563a2a5 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -89,6 +89,9 @@ int nr_ioapics;
 /* IO APIC gsi routing info */
 struct mp_ioapic_gsi  mp_gsi_routing[MAX_IO_APICS];
 
+/* The last gsi number used */
+u32 gsi_end;
+
 /* MP IRQ source entries */
 struct mpc_intsrc mp_irqs[MAX_IRQ_SOURCES];
 
@@ -1013,10 +1016,9 @@ static inline int irq_trigger(int idx)
 	return MPBIOS_trigger(idx);
 }
 
-int (*ioapic_renumber_irq)(int ioapic, int irq);
 static int pin_2_irq(int idx, int apic, int pin)
 {
-	int irq, i;
+	int irq;
 	int bus = mp_irqs[idx].srcbus;
 
 	/*
@@ -1028,18 +1030,12 @@ static int pin_2_irq(int idx, int apic, int pin)
 	if (test_bit(bus, mp_bus_not_pci)) {
 		irq = mp_irqs[idx].srcbusirq;
 	} else {
-		/*
-		 * PCI IRQs are mapped in order
-		 */
-		i = irq = 0;
-		while (i < apic)
-			irq += nr_ioapic_registers[i++];
-		irq += pin;
-		/*
-                 * For MPS mode, so far only needed by ES7000 platform
-                 */
-		if (ioapic_renumber_irq)
-			irq = ioapic_renumber_irq(apic, irq);
+		u32 gsi = mp_gsi_routing[apic].gsi_base + pin;
+
+		if (gsi >= NR_IRQS_LEGACY)
+			irq = gsi;
+		else
+			irq = gsi_end + 1 + gsi;
 	}
 
 #ifdef CONFIG_X86_32
@@ -1950,20 +1946,8 @@ static struct { int pin, apic; } ioapic_i8259 = { -1, -1 };
 
 void __init enable_IO_APIC(void)
 {
-	union IO_APIC_reg_01 reg_01;
 	int i8259_apic, i8259_pin;
 	int apic;
-	unsigned long flags;
-
-	/*
-	 * The number of IO-APIC IRQ registers (== #pins):
-	 */
-	for (apic = 0; apic < nr_ioapics; apic++) {
-		raw_spin_lock_irqsave(&ioapic_lock, flags);
-		reg_01.raw = io_apic_read(apic, 1);
-		raw_spin_unlock_irqrestore(&ioapic_lock, flags);
-		nr_ioapic_registers[apic] = reg_01.bits.entries+1;
-	}
 
 	if (!legacy_pic->nr_legacy_irqs)
 		return;
@@ -3858,27 +3842,20 @@ int __init io_apic_get_redir_entries (int ioapic)
 	reg_01.raw = io_apic_read(ioapic, 1);
 	raw_spin_unlock_irqrestore(&ioapic_lock, flags);
 
-	return reg_01.bits.entries;
+	/* The register returns the maximum index redir index
+	 * supported, which is one less than the total number of redir
+	 * entries.
+	 */
+	return reg_01.bits.entries + 1;
 }
 
 void __init probe_nr_irqs_gsi(void)
 {
-	int nr = 0;
+	int nr;
 
-	nr = acpi_probe_gsi();
-	if (nr > nr_irqs_gsi) {
+	nr = gsi_end + 1 + NR_IRQS_LEGACY;
+	if (nr > nr_irqs_gsi)
 		nr_irqs_gsi = nr;
-	} else {
-		/* for acpi=off or acpi is not compiled in */
-		int idx;
-
-		nr = 0;
-		for (idx = 0; idx < nr_ioapics; idx++)
-			nr += io_apic_get_redir_entries(idx) + 1;
-
-		if (nr > nr_irqs_gsi)
-			nr_irqs_gsi = nr;
-	}
 
 	printk(KERN_DEBUG "nr_irqs_gsi: %d\n", nr_irqs_gsi);
 }
@@ -4085,22 +4062,27 @@ int __init io_apic_get_version(int ioapic)
 	return reg_01.bits.version;
 }
 
-int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
+int acpi_get_override_irq(u32 gsi, int *trigger, int *polarity)
 {
-	int i;
+	int ioapic, pin, idx;
 
 	if (skip_ioapic_setup)
 		return -1;
 
-	for (i = 0; i < mp_irq_entries; i++)
-		if (mp_irqs[i].irqtype == mp_INT &&
-		    mp_irqs[i].srcbusirq == bus_irq)
-			break;
-	if (i >= mp_irq_entries)
+	ioapic = mp_find_ioapic(gsi);
+	if (ioapic < 0)
 		return -1;
 
-	*trigger = irq_trigger(i);
-	*polarity = irq_polarity(i);
+	pin = mp_find_ioapic_pin(ioapic, gsi);
+	if (pin < 0)
+		return -1;
+
+	idx = find_irq_entry(ioapic, pin, mp_INT);
+	if (idx < 0)
+		return -1;
+
+	*trigger = irq_trigger(idx);
+	*polarity = irq_polarity(idx);
 	return 0;
 }
 
@@ -4241,7 +4223,7 @@ void __init ioapic_insert_resources(void)
 	}
 }
 
-int mp_find_ioapic(int gsi)
+int mp_find_ioapic(u32 gsi)
 {
 	int i = 0;
 
@@ -4256,7 +4238,7 @@ int mp_find_ioapic(int gsi)
 	return -1;
 }
 
-int mp_find_ioapic_pin(int ioapic, int gsi)
+int mp_find_ioapic_pin(int ioapic, u32 gsi)
 {
 	if (WARN_ON(ioapic == -1))
 		return -1;
@@ -4284,6 +4266,7 @@ static int bad_ioapic(unsigned long address)
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 {
 	int idx = 0;
+	int entries;
 
 	if (bad_ioapic(address))
 		return;
@@ -4302,9 +4285,17 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 	 * Build basic GSI lookup table to facilitate gsi->io_apic lookups
 	 * and to prevent reprogramming of IOAPIC pins (PCI GSIs).
 	 */
+	entries = io_apic_get_redir_entries(idx);
 	mp_gsi_routing[idx].gsi_base = gsi_base;
-	mp_gsi_routing[idx].gsi_end = gsi_base +
-	    io_apic_get_redir_entries(idx);
+	mp_gsi_routing[idx].gsi_end = gsi_base + entries - 1;
+
+	/*
+	 * The number of IO-APIC IRQ registers (== #pins):
+	 */
+	nr_ioapic_registers[idx] = entries;
+
+	if (mp_gsi_routing[idx].gsi_end > gsi_end)
+		gsi_end = mp_gsi_routing[idx].gsi_end;
 
 	printk(KERN_INFO "IOAPIC[%d]: apic_id %d, version %d, address 0x%x, "
 	       "GSI %d-%d\n", idx, mp_ioapics[idx].apicid,
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index c085d52dbaf..e46f98f36e3 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -735,9 +735,6 @@ void __init uv_system_init(void)
 		uv_node_to_blade[nid] = blade;
 		uv_cpu_to_blade[cpu] = blade;
 		max_pnode = max(pnode, max_pnode);
-
-		printk(KERN_DEBUG "UV: cpu %d, apicid 0x%x, pnode %d, nid %d, lcpu %d, blade %d\n",
-			cpu, apicid, pnode, nid, lcpu, blade);
 	}
 
 	/* Add blade/pnode info for nodes without cpus */
diff --git a/arch/x86/kernel/apm_32.c b/arch/x86/kernel/apm_32.c
index 031aa887b0e..c4f9182ca3a 100644
--- a/arch/x86/kernel/apm_32.c
+++ b/arch/x86/kernel/apm_32.c
@@ -1224,7 +1224,7 @@ static void reinit_timer(void)
 #ifdef INIT_TIMER_AFTER_SUSPEND
 	unsigned long flags;
 
-	spin_lock_irqsave(&i8253_lock, flags);
+	raw_spin_lock_irqsave(&i8253_lock, flags);
 	/* set the clock to HZ */
 	outb_pit(0x34, PIT_MODE);		/* binary, mode 2, LSB/MSB, ch 0 */
 	udelay(10);
@@ -1232,7 +1232,7 @@ static void reinit_timer(void)
 	udelay(10);
 	outb_pit(LATCH >> 8, PIT_CH0);	/* MSB */
 	udelay(10);
-	spin_unlock_irqrestore(&i8253_lock, flags);
+	raw_spin_unlock_irqrestore(&i8253_lock, flags);
 #endif
 }
 
diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index c202b62f367..3a785da34b6 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -14,7 +14,7 @@ CFLAGS_common.o		:= $(nostackp)
 
 obj-y			:= intel_cacheinfo.o addon_cpuid_features.o
 obj-y			+= proc.o capflags.o powerflags.o common.o
-obj-y			+= vmware.o hypervisor.o sched.o
+obj-y			+= vmware.o hypervisor.o sched.o mshyperv.o
 
 obj-$(CONFIG_X86_32)	+= bugs.o cmpxchg.o
 obj-$(CONFIG_X86_64)	+= bugs_64.o
diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c
index 97ad79cdf68..10fa5684a66 100644
--- a/arch/x86/kernel/cpu/addon_cpuid_features.c
+++ b/arch/x86/kernel/cpu/addon_cpuid_features.c
@@ -30,12 +30,14 @@ void __cpuinit init_scattered_cpuid_features(struct cpuinfo_x86 *c)
 	const struct cpuid_bit *cb;
 
 	static const struct cpuid_bit __cpuinitconst cpuid_bits[] = {
-		{ X86_FEATURE_IDA, CR_EAX, 1, 0x00000006 },
-		{ X86_FEATURE_ARAT, CR_EAX, 2, 0x00000006 },
-		{ X86_FEATURE_NPT,   CR_EDX, 0, 0x8000000a },
-		{ X86_FEATURE_LBRV,  CR_EDX, 1, 0x8000000a },
-		{ X86_FEATURE_SVML,  CR_EDX, 2, 0x8000000a },
-		{ X86_FEATURE_NRIPS, CR_EDX, 3, 0x8000000a },
+		{ X86_FEATURE_IDA,   		CR_EAX, 1, 0x00000006 },
+		{ X86_FEATURE_ARAT,  		CR_EAX, 2, 0x00000006 },
+		{ X86_FEATURE_APERFMPERF,	CR_ECX, 0, 0x00000006 },
+		{ X86_FEATURE_CPB,   		CR_EDX, 9, 0x80000007 },
+		{ X86_FEATURE_NPT,   		CR_EDX, 0, 0x8000000a },
+		{ X86_FEATURE_LBRV,  		CR_EDX, 1, 0x8000000a },
+		{ X86_FEATURE_SVML,  		CR_EDX, 2, 0x8000000a },
+		{ X86_FEATURE_NRIPS, 		CR_EDX, 3, 0x8000000a },
 		{ 0, 0, 0, 0 }
 	};
 
diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
index 01a26521239..c39576cb301 100644
--- a/arch/x86/kernel/cpu/bugs.c
+++ b/arch/x86/kernel/cpu/bugs.c
@@ -86,7 +86,7 @@ static void __init check_fpu(void)
 
 static void __init check_hlt(void)
 {
-	if (paravirt_enabled())
+	if (boot_cpu_data.x86 >= 5 || paravirt_enabled())
 		return;
 
 	printk(KERN_INFO "Checking 'hlt' instruction... ");
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 4868e4a951e..68e4a6f2211 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -1084,6 +1084,20 @@ static void clear_all_debug_regs(void)
 	}
 }
 
+#ifdef CONFIG_KGDB
+/*
+ * Restore debug regs if using kgdbwait and you have a kernel debugger
+ * connection established.
+ */
+static void dbg_restore_debug_regs(void)
+{
+	if (unlikely(kgdb_connected && arch_kgdb_ops.correct_hw_break))
+		arch_kgdb_ops.correct_hw_break();
+}
+#else /* ! CONFIG_KGDB */
+#define dbg_restore_debug_regs()
+#endif /* ! CONFIG_KGDB */
+
 /*
  * cpu_init() initializes state that is per-CPU. Some data is already
  * initialized (naturally) in the bootstrap process, such as the GDT
@@ -1107,9 +1121,9 @@ void __cpuinit cpu_init(void)
 	oist = &per_cpu(orig_ist, cpu);
 
 #ifdef CONFIG_NUMA
-	if (cpu != 0 && percpu_read(node_number) == 0 &&
-	    cpu_to_node(cpu) != NUMA_NO_NODE)
-		percpu_write(node_number, cpu_to_node(cpu));
+	if (cpu != 0 && percpu_read(numa_node) == 0 &&
+	    early_cpu_to_node(cpu) != NUMA_NO_NODE)
+		set_numa_node(early_cpu_to_node(cpu));
 #endif
 
 	me = current;
@@ -1174,18 +1188,8 @@ void __cpuinit cpu_init(void)
 	load_TR_desc();
 	load_LDT(&init_mm.context);
 
-#ifdef CONFIG_KGDB
-	/*
-	 * If the kgdb is connected no debug regs should be altered.  This
-	 * is only applicable when KGDB and a KGDB I/O module are built
-	 * into the kernel and you are using early debugging with
-	 * kgdbwait. KGDB will control the kernel HW breakpoint registers.
-	 */
-	if (kgdb_connected && arch_kgdb_ops.correct_hw_break)
-		arch_kgdb_ops.correct_hw_break();
-	else
-#endif
-		clear_all_debug_regs();
+	clear_all_debug_regs();
+	dbg_restore_debug_regs();
 
 	fpu_init();
 
@@ -1239,14 +1243,12 @@ void __cpuinit cpu_init(void)
 #endif
 
 	clear_all_debug_regs();
+	dbg_restore_debug_regs();
 
 	/*
 	 * Force FPU initialization:
 	 */
-	if (cpu_has_xsave)
-		current_thread_info()->status = TS_XSAVE;
-	else
-		current_thread_info()->status = 0;
+	current_thread_info()->status = 0;
 	clear_used_math();
 	mxcsr_feature_mask_init();
 
diff --git a/arch/x86/kernel/cpu/cpufreq/Makefile b/arch/x86/kernel/cpu/cpufreq/Makefile
index 1840c0a5170..bd54bf67e6f 100644
--- a/arch/x86/kernel/cpu/cpufreq/Makefile
+++ b/arch/x86/kernel/cpu/cpufreq/Makefile
@@ -2,8 +2,8 @@
 # K8 systems. ACPI is preferred to all other hardware-specific drivers.
 # speedstep-* is preferred over p4-clockmod.
 
-obj-$(CONFIG_X86_POWERNOW_K8)		+= powernow-k8.o
-obj-$(CONFIG_X86_ACPI_CPUFREQ)		+= acpi-cpufreq.o
+obj-$(CONFIG_X86_POWERNOW_K8)		+= powernow-k8.o mperf.o
+obj-$(CONFIG_X86_ACPI_CPUFREQ)		+= acpi-cpufreq.o mperf.o
 obj-$(CONFIG_X86_PCC_CPUFREQ)		+= pcc-cpufreq.o
 obj-$(CONFIG_X86_POWERNOW_K6)		+= powernow-k6.o
 obj-$(CONFIG_X86_POWERNOW_K7)		+= powernow-k7.o
diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
index 459168083b7..1d3cddaa40e 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -46,6 +46,7 @@
 #include <asm/msr.h>
 #include <asm/processor.h>
 #include <asm/cpufeature.h>
+#include "mperf.h"
 
 #define dprintk(msg...) cpufreq_debug_printk(CPUFREQ_DEBUG_DRIVER, \
 		"acpi-cpufreq", msg)
@@ -71,8 +72,6 @@ struct acpi_cpufreq_data {
 
 static DEFINE_PER_CPU(struct acpi_cpufreq_data *, acfreq_data);
 
-static DEFINE_PER_CPU(struct aperfmperf, acfreq_old_perf);
-
 /* acpi_perf_data is a pointer to percpu data. */
 static struct acpi_processor_performance *acpi_perf_data;
 
@@ -240,45 +239,6 @@ static u32 get_cur_val(const struct cpumask *mask)
 	return cmd.val;
 }
 
-/* Called via smp_call_function_single(), on the target CPU */
-static void read_measured_perf_ctrs(void *_cur)
-{
-	struct aperfmperf *am = _cur;
-
-	get_aperfmperf(am);
-}
-
-/*
- * Return the measured active (C0) frequency on this CPU since last call
- * to this function.
- * Input: cpu number
- * Return: Average CPU frequency in terms of max frequency (zero on error)
- *
- * We use IA32_MPERF and IA32_APERF MSRs to get the measured performance
- * over a period of time, while CPU is in C0 state.
- * IA32_MPERF counts at the rate of max advertised frequency
- * IA32_APERF counts at the rate of actual CPU frequency
- * Only IA32_APERF/IA32_MPERF ratio is architecturally defined and
- * no meaning should be associated with absolute values of these MSRs.
- */
-static unsigned int get_measured_perf(struct cpufreq_policy *policy,
-				      unsigned int cpu)
-{
-	struct aperfmperf perf;
-	unsigned long ratio;
-	unsigned int retval;
-
-	if (smp_call_function_single(cpu, read_measured_perf_ctrs, &perf, 1))
-		return 0;
-
-	ratio = calc_aperfmperf_ratio(&per_cpu(acfreq_old_perf, cpu), &perf);
-	per_cpu(acfreq_old_perf, cpu) = perf;
-
-	retval = (policy->cpuinfo.max_freq * ratio) >> APERFMPERF_SHIFT;
-
-	return retval;
-}
-
 static unsigned int get_cur_freq_on_cpu(unsigned int cpu)
 {
 	struct acpi_cpufreq_data *data = per_cpu(acfreq_data, cpu);
@@ -702,7 +662,7 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
 
 	/* Check for APERF/MPERF support in hardware */
 	if (cpu_has(c, X86_FEATURE_APERFMPERF))
-		acpi_cpufreq_driver.getavg = get_measured_perf;
+		acpi_cpufreq_driver.getavg = cpufreq_get_measured_perf;
 
 	dprintk("CPU%u - ACPI performance management activated.\n", cpu);
 	for (i = 0; i < perf->state_count; i++)
diff --git a/arch/x86/kernel/cpu/cpufreq/mperf.c b/arch/x86/kernel/cpu/cpufreq/mperf.c
new file mode 100644
index 00000000000..911e193018a
--- /dev/null
+++ b/arch/x86/kernel/cpu/cpufreq/mperf.c
@@ -0,0 +1,51 @@
+#include <linux/kernel.h>
+#include <linux/smp.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/cpufreq.h>
+#include <linux/slab.h>
+
+#include "mperf.h"
+
+static DEFINE_PER_CPU(struct aperfmperf, acfreq_old_perf);
+
+/* Called via smp_call_function_single(), on the target CPU */
+static void read_measured_perf_ctrs(void *_cur)
+{
+	struct aperfmperf *am = _cur;
+
+	get_aperfmperf(am);
+}
+
+/*
+ * Return the measured active (C0) frequency on this CPU since last call
+ * to this function.
+ * Input: cpu number
+ * Return: Average CPU frequency in terms of max frequency (zero on error)
+ *
+ * We use IA32_MPERF and IA32_APERF MSRs to get the measured performance
+ * over a period of time, while CPU is in C0 state.
+ * IA32_MPERF counts at the rate of max advertised frequency
+ * IA32_APERF counts at the rate of actual CPU frequency
+ * Only IA32_APERF/IA32_MPERF ratio is architecturally defined and
+ * no meaning should be associated with absolute values of these MSRs.
+ */
+unsigned int cpufreq_get_measured_perf(struct cpufreq_policy *policy,
+					unsigned int cpu)
+{
+	struct aperfmperf perf;
+	unsigned long ratio;
+	unsigned int retval;
+
+	if (smp_call_function_single(cpu, read_measured_perf_ctrs, &perf, 1))
+		return 0;
+
+	ratio = calc_aperfmperf_ratio(&per_cpu(acfreq_old_perf, cpu), &perf);
+	per_cpu(acfreq_old_perf, cpu) = perf;
+
+	retval = (policy->cpuinfo.max_freq * ratio) >> APERFMPERF_SHIFT;
+
+	return retval;
+}
+EXPORT_SYMBOL_GPL(cpufreq_get_measured_perf);
+MODULE_LICENSE("GPL");
diff --git a/arch/x86/kernel/cpu/cpufreq/mperf.h b/arch/x86/kernel/cpu/cpufreq/mperf.h
new file mode 100644
index 00000000000..5dbf2950dc2
--- /dev/null
+++ b/arch/x86/kernel/cpu/cpufreq/mperf.h
@@ -0,0 +1,9 @@
+/*
+ *  (c) 2010 Advanced Micro Devices, Inc.
+ *  Your use of this code is subject to the terms and conditions of the
+ *  GNU general public license version 2. See "COPYING" or
+ *  http://www.gnu.org/licenses/gpl.html
+ */
+
+unsigned int cpufreq_get_measured_perf(struct cpufreq_policy *policy,
+					unsigned int cpu);
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
index b6215b9798e..7ec2123838e 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
@@ -1,6 +1,5 @@
-
 /*
- *   (c) 2003-2006 Advanced Micro Devices, Inc.
+ *   (c) 2003-2010 Advanced Micro Devices, Inc.
  *  Your use of this code is subject to the terms and conditions of the
  *  GNU general public license version 2. See "COPYING" or
  *  http://www.gnu.org/licenses/gpl.html
@@ -46,6 +45,7 @@
 #define PFX "powernow-k8: "
 #define VERSION "version 2.20.00"
 #include "powernow-k8.h"
+#include "mperf.h"
 
 /* serialize freq changes  */
 static DEFINE_MUTEX(fidvid_mutex);
@@ -54,6 +54,12 @@ static DEFINE_PER_CPU(struct powernow_k8_data *, powernow_data);
 
 static int cpu_family = CPU_OPTERON;
 
+/* core performance boost */
+static bool cpb_capable, cpb_enabled;
+static struct msr __percpu *msrs;
+
+static struct cpufreq_driver cpufreq_amd64_driver;
+
 #ifndef CONFIG_SMP
 static inline const struct cpumask *cpu_core_mask(int cpu)
 {
@@ -1249,6 +1255,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol)
 	struct powernow_k8_data *data;
 	struct init_on_cpu init_on_cpu;
 	int rc;
+	struct cpuinfo_x86 *c = &cpu_data(pol->cpu);
 
 	if (!cpu_online(pol->cpu))
 		return -ENODEV;
@@ -1323,6 +1330,10 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol)
 		return -EINVAL;
 	}
 
+	/* Check for APERF/MPERF support in hardware */
+	if (cpu_has(c, X86_FEATURE_APERFMPERF))
+		cpufreq_amd64_driver.getavg = cpufreq_get_measured_perf;
+
 	cpufreq_frequency_table_get_attr(data->powernow_table, pol->cpu);
 
 	if (cpu_family == CPU_HW_PSTATE)
@@ -1394,8 +1405,77 @@ out:
 	return khz;
 }
 
+static void _cpb_toggle_msrs(bool t)
+{
+	int cpu;
+
+	get_online_cpus();
+
+	rdmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs);
+
+	for_each_cpu(cpu, cpu_online_mask) {
+		struct msr *reg = per_cpu_ptr(msrs, cpu);
+		if (t)
+			reg->l &= ~BIT(25);
+		else
+			reg->l |= BIT(25);
+	}
+	wrmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs);
+
+	put_online_cpus();
+}
+
+/*
+ * Switch on/off core performance boosting.
+ *
+ * 0=disable
+ * 1=enable.
+ */
+static void cpb_toggle(bool t)
+{
+	if (!cpb_capable)
+		return;
+
+	if (t && !cpb_enabled) {
+		cpb_enabled = true;
+		_cpb_toggle_msrs(t);
+		printk(KERN_INFO PFX "Core Boosting enabled.\n");
+	} else if (!t && cpb_enabled) {
+		cpb_enabled = false;
+		_cpb_toggle_msrs(t);
+		printk(KERN_INFO PFX "Core Boosting disabled.\n");
+	}
+}
+
+static ssize_t store_cpb(struct cpufreq_policy *policy, const char *buf,
+				 size_t count)
+{
+	int ret = -EINVAL;
+	unsigned long val = 0;
+
+	ret = strict_strtoul(buf, 10, &val);
+	if (!ret && (val == 0 || val == 1) && cpb_capable)
+		cpb_toggle(val);
+	else
+		return -EINVAL;
+
+	return count;
+}
+
+static ssize_t show_cpb(struct cpufreq_policy *policy, char *buf)
+{
+	return sprintf(buf, "%u\n", cpb_enabled);
+}
+
+#define define_one_rw(_name) \
+static struct freq_attr _name = \
+__ATTR(_name, 0644, show_##_name, store_##_name)
+
+define_one_rw(cpb);
+
 static struct freq_attr *powernow_k8_attr[] = {
 	&cpufreq_freq_attr_scaling_available_freqs,
+	&cpb,
 	NULL,
 };
 
@@ -1411,10 +1491,51 @@ static struct cpufreq_driver cpufreq_amd64_driver = {
 	.attr		= powernow_k8_attr,
 };
 
+/*
+ * Clear the boost-disable flag on the CPU_DOWN path so that this cpu
+ * cannot block the remaining ones from boosting. On the CPU_UP path we
+ * simply keep the boost-disable flag in sync with the current global
+ * state.
+ */
+static int cpb_notify(struct notifier_block *nb, unsigned long action,
+		      void *hcpu)
+{
+	unsigned cpu = (long)hcpu;
+	u32 lo, hi;
+
+	switch (action) {
+	case CPU_UP_PREPARE:
+	case CPU_UP_PREPARE_FROZEN:
+
+		if (!cpb_enabled) {
+			rdmsr_on_cpu(cpu, MSR_K7_HWCR, &lo, &hi);
+			lo |= BIT(25);
+			wrmsr_on_cpu(cpu, MSR_K7_HWCR, lo, hi);
+		}
+		break;
+
+	case CPU_DOWN_PREPARE:
+	case CPU_DOWN_PREPARE_FROZEN:
+		rdmsr_on_cpu(cpu, MSR_K7_HWCR, &lo, &hi);
+		lo &= ~BIT(25);
+		wrmsr_on_cpu(cpu, MSR_K7_HWCR, lo, hi);
+		break;
+
+	default:
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block cpb_nb = {
+	.notifier_call		= cpb_notify,
+};
+
 /* driver entry point for init */
 static int __cpuinit powernowk8_init(void)
 {
-	unsigned int i, supported_cpus = 0;
+	unsigned int i, supported_cpus = 0, cpu;
 
 	for_each_online_cpu(i) {
 		int rc;
@@ -1423,15 +1544,36 @@ static int __cpuinit powernowk8_init(void)
 			supported_cpus++;
 	}
 
-	if (supported_cpus == num_online_cpus()) {
-		printk(KERN_INFO PFX "Found %d %s "
-			"processors (%d cpu cores) (" VERSION ")\n",
-			num_online_nodes(),
-			boot_cpu_data.x86_model_id, supported_cpus);
-		return cpufreq_register_driver(&cpufreq_amd64_driver);
+	if (supported_cpus != num_online_cpus())
+		return -ENODEV;
+
+	printk(KERN_INFO PFX "Found %d %s (%d cpu cores) (" VERSION ")\n",
+		num_online_nodes(), boot_cpu_data.x86_model_id, supported_cpus);
+
+	if (boot_cpu_has(X86_FEATURE_CPB)) {
+
+		cpb_capable = true;
+
+		register_cpu_notifier(&cpb_nb);
+
+		msrs = msrs_alloc();
+		if (!msrs) {
+			printk(KERN_ERR "%s: Error allocating msrs!\n", __func__);
+			return -ENOMEM;
+		}
+
+		rdmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs);
+
+		for_each_cpu(cpu, cpu_online_mask) {
+			struct msr *reg = per_cpu_ptr(msrs, cpu);
+			cpb_enabled |= !(!!(reg->l & BIT(25)));
+		}
+
+		printk(KERN_INFO PFX "Core Performance Boosting: %s.\n",
+			(cpb_enabled ? "on" : "off"));
 	}
 
-	return -ENODEV;
+	return cpufreq_register_driver(&cpufreq_amd64_driver);
 }
 
 /* driver entry point for term */
@@ -1439,6 +1581,13 @@ static void __exit powernowk8_exit(void)
 {
 	dprintk("exit\n");
 
+	if (boot_cpu_has(X86_FEATURE_CPB)) {
+		msrs_free(msrs);
+		msrs = NULL;
+
+		unregister_cpu_notifier(&cpb_nb);
+	}
+
 	cpufreq_unregister_driver(&cpufreq_amd64_driver);
 }
 
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
index 02ce824073c..df3529b1c02 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
@@ -5,7 +5,6 @@
  *  http://www.gnu.org/licenses/gpl.html
  */
 
-
 enum pstate {
 	HW_PSTATE_INVALID = 0xff,
 	HW_PSTATE_0 = 0,
@@ -55,7 +54,6 @@ struct powernow_k8_data {
 	struct cpumask *available_cores;
 };
 
-
 /* processor's cpuid instruction support */
 #define CPUID_PROCESSOR_SIGNATURE	1	/* function 1 */
 #define CPUID_XFAM			0x0ff00000	/* extended family */
diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c
index 08be922de33..dd531cc56a8 100644
--- a/arch/x86/kernel/cpu/hypervisor.c
+++ b/arch/x86/kernel/cpu/hypervisor.c
@@ -21,37 +21,55 @@
  *
  */
 
+#include <linux/module.h>
 #include <asm/processor.h>
-#include <asm/vmware.h>
 #include <asm/hypervisor.h>
 
-static inline void __cpuinit
-detect_hypervisor_vendor(struct cpuinfo_x86 *c)
+/*
+ * Hypervisor detect order.  This is specified explicitly here because
+ * some hypervisors might implement compatibility modes for other
+ * hypervisors and therefore need to be detected in specific sequence.
+ */
+static const __initconst struct hypervisor_x86 * const hypervisors[] =
 {
-	if (vmware_platform())
-		c->x86_hyper_vendor = X86_HYPER_VENDOR_VMWARE;
-	else
-		c->x86_hyper_vendor = X86_HYPER_VENDOR_NONE;
-}
+	&x86_hyper_vmware,
+	&x86_hyper_ms_hyperv,
+};
 
-static inline void __cpuinit
-hypervisor_set_feature_bits(struct cpuinfo_x86 *c)
+const struct hypervisor_x86 *x86_hyper;
+EXPORT_SYMBOL(x86_hyper);
+
+static inline void __init
+detect_hypervisor_vendor(void)
 {
-	if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE) {
-		vmware_set_feature_bits(c);
-		return;
+	const struct hypervisor_x86 *h, * const *p;
+
+	for (p = hypervisors; p < hypervisors + ARRAY_SIZE(hypervisors); p++) {
+		h = *p;
+		if (h->detect()) {
+			x86_hyper = h;
+			printk(KERN_INFO "Hypervisor detected: %s\n", h->name);
+			break;
+		}
 	}
 }
 
 void __cpuinit init_hypervisor(struct cpuinfo_x86 *c)
 {
-	detect_hypervisor_vendor(c);
-	hypervisor_set_feature_bits(c);
+	if (x86_hyper && x86_hyper->set_cpu_features)
+		x86_hyper->set_cpu_features(c);
 }
 
 void __init init_hypervisor_platform(void)
 {
+
+	detect_hypervisor_vendor();
+
+	if (!x86_hyper)
+		return;
+
 	init_hypervisor(&boot_cpu_data);
-	if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE)
-		vmware_platform_setup();
+
+	if (x86_hyper->init_platform)
+		x86_hyper->init_platform();
 }
diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c
index 1366c7cfd48..85f69cdeae1 100644
--- a/arch/x86/kernel/cpu/intel.c
+++ b/arch/x86/kernel/cpu/intel.c
@@ -12,7 +12,6 @@
 #include <asm/processor.h>
 #include <asm/pgtable.h>
 #include <asm/msr.h>
-#include <asm/ds.h>
 #include <asm/bugs.h>
 #include <asm/cpu.h>
 
@@ -373,12 +372,6 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c)
 			set_cpu_cap(c, X86_FEATURE_ARCH_PERFMON);
 	}
 
-	if (c->cpuid_level > 6) {
-		unsigned ecx = cpuid_ecx(6);
-		if (ecx & 0x01)
-			set_cpu_cap(c, X86_FEATURE_APERFMPERF);
-	}
-
 	if (cpu_has_xmm2)
 		set_cpu_cap(c, X86_FEATURE_LFENCE_RDTSC);
 	if (cpu_has_ds) {
@@ -388,7 +381,6 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c)
 			set_cpu_cap(c, X86_FEATURE_BTS);
 		if (!(l1 & (1<<12)))
 			set_cpu_cap(c, X86_FEATURE_PEBS);
-		ds_init_intel(c);
 	}
 
 	if (c->x86 == 6 && c->x86_model == 29 && cpu_has_clflush)
diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c
index b3eeb66c0a5..33eae2062cf 100644
--- a/arch/x86/kernel/cpu/intel_cacheinfo.c
+++ b/arch/x86/kernel/cpu/intel_cacheinfo.c
@@ -148,13 +148,19 @@ union _cpuid4_leaf_ecx {
 	u32 full;
 };
 
+struct amd_l3_cache {
+	struct	 pci_dev *dev;
+	bool	 can_disable;
+	unsigned indices;
+	u8	 subcaches[4];
+};
+
 struct _cpuid4_info {
 	union _cpuid4_leaf_eax eax;
 	union _cpuid4_leaf_ebx ebx;
 	union _cpuid4_leaf_ecx ecx;
 	unsigned long size;
-	bool can_disable;
-	unsigned int l3_indices;
+	struct amd_l3_cache *l3;
 	DECLARE_BITMAP(shared_cpu_map, NR_CPUS);
 };
 
@@ -164,8 +170,7 @@ struct _cpuid4_info_regs {
 	union _cpuid4_leaf_ebx ebx;
 	union _cpuid4_leaf_ecx ecx;
 	unsigned long size;
-	bool can_disable;
-	unsigned int l3_indices;
+	struct amd_l3_cache *l3;
 };
 
 unsigned short			num_cache_leaves;
@@ -302,87 +307,163 @@ struct _cache_attr {
 };
 
 #ifdef CONFIG_CPU_SUP_AMD
-static unsigned int __cpuinit amd_calc_l3_indices(void)
+
+/*
+ * L3 cache descriptors
+ */
+static struct amd_l3_cache **__cpuinitdata l3_caches;
+
+static void __cpuinit amd_calc_l3_indices(struct amd_l3_cache *l3)
 {
-	/*
-	 * We're called over smp_call_function_single() and therefore
-	 * are on the correct cpu.
-	 */
-	int cpu = smp_processor_id();
-	int node = cpu_to_node(cpu);
-	struct pci_dev *dev = node_to_k8_nb_misc(node);
 	unsigned int sc0, sc1, sc2, sc3;
 	u32 val = 0;
 
-	pci_read_config_dword(dev, 0x1C4, &val);
+	pci_read_config_dword(l3->dev, 0x1C4, &val);
 
 	/* calculate subcache sizes */
-	sc0 = !(val & BIT(0));
-	sc1 = !(val & BIT(4));
-	sc2 = !(val & BIT(8))  + !(val & BIT(9));
-	sc3 = !(val & BIT(12)) + !(val & BIT(13));
+	l3->subcaches[0] = sc0 = !(val & BIT(0));
+	l3->subcaches[1] = sc1 = !(val & BIT(4));
+	l3->subcaches[2] = sc2 = !(val & BIT(8))  + !(val & BIT(9));
+	l3->subcaches[3] = sc3 = !(val & BIT(12)) + !(val & BIT(13));
 
-	return (max(max(max(sc0, sc1), sc2), sc3) << 10) - 1;
+	l3->indices = (max(max(max(sc0, sc1), sc2), sc3) << 10) - 1;
+}
+
+static struct amd_l3_cache * __cpuinit amd_init_l3_cache(int node)
+{
+	struct amd_l3_cache *l3;
+	struct pci_dev *dev = node_to_k8_nb_misc(node);
+
+	l3 = kzalloc(sizeof(struct amd_l3_cache), GFP_ATOMIC);
+	if (!l3) {
+		printk(KERN_WARNING "Error allocating L3 struct\n");
+		return NULL;
+	}
+
+	l3->dev = dev;
+
+	amd_calc_l3_indices(l3);
+
+	return l3;
 }
 
 static void __cpuinit
 amd_check_l3_disable(int index, struct _cpuid4_info_regs *this_leaf)
 {
-	if (index < 3)
+	int node;
+
+	if (boot_cpu_data.x86 != 0x10)
 		return;
 
-	if (boot_cpu_data.x86 == 0x11)
+	if (index < 3)
 		return;
 
 	/* see errata #382 and #388 */
-	if ((boot_cpu_data.x86 == 0x10) &&
-	    ((boot_cpu_data.x86_model < 0x8) ||
-	     (boot_cpu_data.x86_mask  < 0x1)))
+	if (boot_cpu_data.x86_model < 0x8)
+		return;
+
+	if ((boot_cpu_data.x86_model == 0x8 ||
+	     boot_cpu_data.x86_model == 0x9)
+		&&
+	     boot_cpu_data.x86_mask < 0x1)
+			return;
+
+	/* not in virtualized environments */
+	if (num_k8_northbridges == 0)
 		return;
 
-	this_leaf->can_disable = true;
-	this_leaf->l3_indices  = amd_calc_l3_indices();
+	/*
+	 * Strictly speaking, the amount in @size below is leaked since it is
+	 * never freed but this is done only on shutdown so it doesn't matter.
+	 */
+	if (!l3_caches) {
+		int size = num_k8_northbridges * sizeof(struct amd_l3_cache *);
+
+		l3_caches = kzalloc(size, GFP_ATOMIC);
+		if (!l3_caches)
+			return;
+	}
+
+	node = amd_get_nb_id(smp_processor_id());
+
+	if (!l3_caches[node]) {
+		l3_caches[node] = amd_init_l3_cache(node);
+		l3_caches[node]->can_disable = true;
+	}
+
+	WARN_ON(!l3_caches[node]);
+
+	this_leaf->l3 = l3_caches[node];
 }
 
 static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf,
-				  unsigned int index)
+				  unsigned int slot)
 {
-	int cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map));
-	int node = amd_get_nb_id(cpu);
-	struct pci_dev *dev = node_to_k8_nb_misc(node);
+	struct pci_dev *dev = this_leaf->l3->dev;
 	unsigned int reg = 0;
 
-	if (!this_leaf->can_disable)
+	if (!this_leaf->l3 || !this_leaf->l3->can_disable)
 		return -EINVAL;
 
 	if (!dev)
 		return -EINVAL;
 
-	pci_read_config_dword(dev, 0x1BC + index * 4, &reg);
+	pci_read_config_dword(dev, 0x1BC + slot * 4, &reg);
 	return sprintf(buf, "0x%08x\n", reg);
 }
 
-#define SHOW_CACHE_DISABLE(index)					\
+#define SHOW_CACHE_DISABLE(slot)					\
 static ssize_t								\
-show_cache_disable_##index(struct _cpuid4_info *this_leaf, char *buf)	\
+show_cache_disable_##slot(struct _cpuid4_info *this_leaf, char *buf)	\
 {									\
-	return show_cache_disable(this_leaf, buf, index);		\
+	return show_cache_disable(this_leaf, buf, slot);		\
 }
 SHOW_CACHE_DISABLE(0)
 SHOW_CACHE_DISABLE(1)
 
+static void amd_l3_disable_index(struct amd_l3_cache *l3, int cpu,
+				 unsigned slot, unsigned long idx)
+{
+	int i;
+
+	idx |= BIT(30);
+
+	/*
+	 *  disable index in all 4 subcaches
+	 */
+	for (i = 0; i < 4; i++) {
+		u32 reg = idx | (i << 20);
+
+		if (!l3->subcaches[i])
+			continue;
+
+		pci_write_config_dword(l3->dev, 0x1BC + slot * 4, reg);
+
+		/*
+		 * We need to WBINVD on a core on the node containing the L3
+		 * cache which indices we disable therefore a simple wbinvd()
+		 * is not sufficient.
+		 */
+		wbinvd_on_cpu(cpu);
+
+		reg |= BIT(31);
+		pci_write_config_dword(l3->dev, 0x1BC + slot * 4, reg);
+	}
+}
+
+
 static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
-	const char *buf, size_t count, unsigned int index)
+				   const char *buf, size_t count,
+				   unsigned int slot)
 {
+	struct pci_dev *dev = this_leaf->l3->dev;
 	int cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map));
-	int node = amd_get_nb_id(cpu);
-	struct pci_dev *dev = node_to_k8_nb_misc(node);
 	unsigned long val = 0;
 
 #define SUBCACHE_MASK	(3UL << 20)
 #define SUBCACHE_INDEX	0xfff
 
-	if (!this_leaf->can_disable)
+	if (!this_leaf->l3 || !this_leaf->l3->can_disable)
 		return -EINVAL;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -396,26 +477,20 @@ static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
 
 	/* do not allow writes outside of allowed bits */
 	if ((val & ~(SUBCACHE_MASK | SUBCACHE_INDEX)) ||
-	    ((val & SUBCACHE_INDEX) > this_leaf->l3_indices))
+	    ((val & SUBCACHE_INDEX) > this_leaf->l3->indices))
 		return -EINVAL;
 
-	val |= BIT(30);
-	pci_write_config_dword(dev, 0x1BC + index * 4, val);
-	/*
-	 * We need to WBINVD on a core on the node containing the L3 cache which
-	 * indices we disable therefore a simple wbinvd() is not sufficient.
-	 */
-	wbinvd_on_cpu(cpu);
-	pci_write_config_dword(dev, 0x1BC + index * 4, val | BIT(31));
+	amd_l3_disable_index(this_leaf->l3, cpu, slot, val);
+
 	return count;
 }
 
-#define STORE_CACHE_DISABLE(index)					\
+#define STORE_CACHE_DISABLE(slot)					\
 static ssize_t								\
-store_cache_disable_##index(struct _cpuid4_info *this_leaf,		\
+store_cache_disable_##slot(struct _cpuid4_info *this_leaf,		\
 			    const char *buf, size_t count)		\
 {									\
-	return store_cache_disable(this_leaf, buf, count, index);	\
+	return store_cache_disable(this_leaf, buf, count, slot);	\
 }
 STORE_CACHE_DISABLE(0)
 STORE_CACHE_DISABLE(1)
@@ -443,8 +518,7 @@ __cpuinit cpuid4_cache_lookup_regs(int index,
 
 	if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD) {
 		amd_cpuid4(index, &eax, &ebx, &ecx);
-		if (boot_cpu_data.x86 >= 0x10)
-			amd_check_l3_disable(index, this_leaf);
+		amd_check_l3_disable(index, this_leaf);
 	} else {
 		cpuid_count(4, index, &eax.full, &ebx.full, &ecx.full, &edx);
 	}
@@ -701,6 +775,7 @@ static void __cpuinit free_cache_attributes(unsigned int cpu)
 	for (i = 0; i < num_cache_leaves; i++)
 		cache_remove_shared_cpu_map(cpu, i);
 
+	kfree(per_cpu(ici_cpuid4_info, cpu)->l3);
 	kfree(per_cpu(ici_cpuid4_info, cpu));
 	per_cpu(ici_cpuid4_info, cpu) = NULL;
 }
@@ -985,7 +1060,7 @@ static int __cpuinit cache_add_dev(struct sys_device * sys_dev)
 
 		this_leaf = CPUID4_INFO_IDX(cpu, i);
 
-		if (this_leaf->can_disable)
+		if (this_leaf->l3 && this_leaf->l3->can_disable)
 			ktype_cache.default_attrs = default_l3_attrs;
 		else
 			ktype_cache.default_attrs = default_attrs;
diff --git a/arch/x86/kernel/cpu/mcheck/Makefile b/arch/x86/kernel/cpu/mcheck/Makefile
index 4ac6d48fe11..bb34b03af25 100644
--- a/arch/x86/kernel/cpu/mcheck/Makefile
+++ b/arch/x86/kernel/cpu/mcheck/Makefile
@@ -7,3 +7,5 @@ obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o
 obj-$(CONFIG_X86_MCE_INJECT)	+= mce-inject.o
 
 obj-$(CONFIG_X86_THERMAL_VECTOR) += therm_throt.o
+
+obj-$(CONFIG_ACPI_APEI)		+= mce-apei.o
diff --git a/arch/x86/kernel/cpu/mcheck/mce-apei.c b/arch/x86/kernel/cpu/mcheck/mce-apei.c
new file mode 100644
index 00000000000..745b54f9be8
--- /dev/null
+++ b/arch/x86/kernel/cpu/mcheck/mce-apei.c
@@ -0,0 +1,138 @@
+/*
+ * Bridge between MCE and APEI
+ *
+ * On some machine, corrected memory errors are reported via APEI
+ * generic hardware error source (GHES) instead of corrected Machine
+ * Check. These corrected memory errors can be reported to user space
+ * through /dev/mcelog via faking a corrected Machine Check, so that
+ * the error memory page can be offlined by /sbin/mcelog if the error
+ * count for one page is beyond the threshold.
+ *
+ * For fatal MCE, save MCE record into persistent storage via ERST, so
+ * that the MCE record can be logged after reboot via ERST.
+ *
+ * Copyright 2010 Intel Corp.
+ *   Author: Huang Ying <ying.huang@intel.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version
+ * 2 as published by the Free Software Foundation.
+ *
+ * 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/kernel.h>
+#include <linux/acpi.h>
+#include <linux/cper.h>
+#include <acpi/apei.h>
+#include <asm/mce.h>
+
+#include "mce-internal.h"
+
+void apei_mce_report_mem_error(int corrected, struct cper_sec_mem_err *mem_err)
+{
+	struct mce m;
+
+	/* Only corrected MC is reported */
+	if (!corrected)
+		return;
+
+	mce_setup(&m);
+	m.bank = 1;
+	/* Fake a memory read corrected error with unknown channel */
+	m.status = MCI_STATUS_VAL | MCI_STATUS_EN | MCI_STATUS_ADDRV | 0x9f;
+	m.addr = mem_err->physical_addr;
+	mce_log(&m);
+	mce_notify_irq();
+}
+EXPORT_SYMBOL_GPL(apei_mce_report_mem_error);
+
+#define CPER_CREATOR_MCE						\
+	UUID_LE(0x75a574e3, 0x5052, 0x4b29, 0x8a, 0x8e, 0xbe, 0x2c,	\
+		0x64, 0x90, 0xb8, 0x9d)
+#define CPER_SECTION_TYPE_MCE						\
+	UUID_LE(0xfe08ffbe, 0x95e4, 0x4be7, 0xbc, 0x73, 0x40, 0x96,	\
+		0x04, 0x4a, 0x38, 0xfc)
+
+/*
+ * CPER specification (in UEFI specification 2.3 appendix N) requires
+ * byte-packed.
+ */
+struct cper_mce_record {
+	struct cper_record_header hdr;
+	struct cper_section_descriptor sec_hdr;
+	struct mce mce;
+} __packed;
+
+int apei_write_mce(struct mce *m)
+{
+	struct cper_mce_record rcd;
+
+	memset(&rcd, 0, sizeof(rcd));
+	memcpy(rcd.hdr.signature, CPER_SIG_RECORD, CPER_SIG_SIZE);
+	rcd.hdr.revision = CPER_RECORD_REV;
+	rcd.hdr.signature_end = CPER_SIG_END;
+	rcd.hdr.section_count = 1;
+	rcd.hdr.error_severity = CPER_SER_FATAL;
+	/* timestamp, platform_id, partition_id are all invalid */
+	rcd.hdr.validation_bits = 0;
+	rcd.hdr.record_length = sizeof(rcd);
+	rcd.hdr.creator_id = CPER_CREATOR_MCE;
+	rcd.hdr.notification_type = CPER_NOTIFY_MCE;
+	rcd.hdr.record_id = cper_next_record_id();
+	rcd.hdr.flags = CPER_HW_ERROR_FLAGS_PREVERR;
+
+	rcd.sec_hdr.section_offset = (void *)&rcd.mce - (void *)&rcd;
+	rcd.sec_hdr.section_length = sizeof(rcd.mce);
+	rcd.sec_hdr.revision = CPER_SEC_REV;
+	/* fru_id and fru_text is invalid */
+	rcd.sec_hdr.validation_bits = 0;
+	rcd.sec_hdr.flags = CPER_SEC_PRIMARY;
+	rcd.sec_hdr.section_type = CPER_SECTION_TYPE_MCE;
+	rcd.sec_hdr.section_severity = CPER_SER_FATAL;
+
+	memcpy(&rcd.mce, m, sizeof(*m));
+
+	return erst_write(&rcd.hdr);
+}
+
+ssize_t apei_read_mce(struct mce *m, u64 *record_id)
+{
+	struct cper_mce_record rcd;
+	ssize_t len;
+
+	len = erst_read_next(&rcd.hdr, sizeof(rcd));
+	if (len <= 0)
+		return len;
+	/* Can not skip other records in storage via ERST unless clear them */
+	else if (len != sizeof(rcd) ||
+		 uuid_le_cmp(rcd.hdr.creator_id, CPER_CREATOR_MCE)) {
+		if (printk_ratelimit())
+			pr_warning(
+			"MCE-APEI: Can not skip the unknown record in ERST");
+		return -EIO;
+	}
+
+	memcpy(m, &rcd.mce, sizeof(*m));
+	*record_id = rcd.hdr.record_id;
+
+	return sizeof(*m);
+}
+
+/* Check whether there is record in ERST */
+int apei_check_mce(void)
+{
+	return erst_get_record_count();
+}
+
+int apei_clear_mce(u64 record_id)
+{
+	return erst_clear(record_id);
+}
diff --git a/arch/x86/kernel/cpu/mcheck/mce-internal.h b/arch/x86/kernel/cpu/mcheck/mce-internal.h
index 32996f9fab6..fefcc69ee8b 100644
--- a/arch/x86/kernel/cpu/mcheck/mce-internal.h
+++ b/arch/x86/kernel/cpu/mcheck/mce-internal.h
@@ -28,3 +28,26 @@ extern int mce_ser;
 
 extern struct mce_bank *mce_banks;
 
+#ifdef CONFIG_ACPI_APEI
+int apei_write_mce(struct mce *m);
+ssize_t apei_read_mce(struct mce *m, u64 *record_id);
+int apei_check_mce(void);
+int apei_clear_mce(u64 record_id);
+#else
+static inline int apei_write_mce(struct mce *m)
+{
+	return -EINVAL;
+}
+static inline ssize_t apei_read_mce(struct mce *m, u64 *record_id)
+{
+	return 0;
+}
+static inline int apei_check_mce(void)
+{
+	return 0;
+}
+static inline int apei_clear_mce(u64 record_id)
+{
+	return -EINVAL;
+}
+#endif
diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
index 8a6f0afa767..707165dbc20 100644
--- a/arch/x86/kernel/cpu/mcheck/mce.c
+++ b/arch/x86/kernel/cpu/mcheck/mce.c
@@ -264,7 +264,7 @@ static void wait_for_panic(void)
 
 static void mce_panic(char *msg, struct mce *final, char *exp)
 {
-	int i;
+	int i, apei_err = 0;
 
 	if (!fake_panic) {
 		/*
@@ -287,8 +287,11 @@ static void mce_panic(char *msg, struct mce *final, char *exp)
 		struct mce *m = &mcelog.entry[i];
 		if (!(m->status & MCI_STATUS_VAL))
 			continue;
-		if (!(m->status & MCI_STATUS_UC))
+		if (!(m->status & MCI_STATUS_UC)) {
 			print_mce(m);
+			if (!apei_err)
+				apei_err = apei_write_mce(m);
+		}
 	}
 	/* Now print uncorrected but with the final one last */
 	for (i = 0; i < MCE_LOG_LEN; i++) {
@@ -297,11 +300,17 @@ static void mce_panic(char *msg, struct mce *final, char *exp)
 			continue;
 		if (!(m->status & MCI_STATUS_UC))
 			continue;
-		if (!final || memcmp(m, final, sizeof(struct mce)))
+		if (!final || memcmp(m, final, sizeof(struct mce))) {
 			print_mce(m);
+			if (!apei_err)
+				apei_err = apei_write_mce(m);
+		}
 	}
-	if (final)
+	if (final) {
 		print_mce(final);
+		if (!apei_err)
+			apei_err = apei_write_mce(final);
+	}
 	if (cpu_missing)
 		printk(KERN_EMERG "Some CPUs didn't answer in synchronization\n");
 	print_mce_tail();
@@ -539,7 +548,7 @@ void machine_check_poll(enum mcp_flags flags, mce_banks_t *b)
 	struct mce m;
 	int i;
 
-	__get_cpu_var(mce_poll_count)++;
+	percpu_inc(mce_poll_count);
 
 	mce_setup(&m);
 
@@ -934,7 +943,7 @@ void do_machine_check(struct pt_regs *regs, long error_code)
 
 	atomic_inc(&mce_entry);
 
-	__get_cpu_var(mce_exception_count)++;
+	percpu_inc(mce_exception_count);
 
 	if (notify_die(DIE_NMI, "machine check", regs, error_code,
 			   18, SIGKILL) == NOTIFY_STOP)
@@ -1493,6 +1502,43 @@ static void collect_tscs(void *data)
 	rdtscll(cpu_tsc[smp_processor_id()]);
 }
 
+static int mce_apei_read_done;
+
+/* Collect MCE record of previous boot in persistent storage via APEI ERST. */
+static int __mce_read_apei(char __user **ubuf, size_t usize)
+{
+	int rc;
+	u64 record_id;
+	struct mce m;
+
+	if (usize < sizeof(struct mce))
+		return -EINVAL;
+
+	rc = apei_read_mce(&m, &record_id);
+	/* Error or no more MCE record */
+	if (rc <= 0) {
+		mce_apei_read_done = 1;
+		return rc;
+	}
+	rc = -EFAULT;
+	if (copy_to_user(*ubuf, &m, sizeof(struct mce)))
+		return rc;
+	/*
+	 * In fact, we should have cleared the record after that has
+	 * been flushed to the disk or sent to network in
+	 * /sbin/mcelog, but we have no interface to support that now,
+	 * so just clear it to avoid duplication.
+	 */
+	rc = apei_clear_mce(record_id);
+	if (rc) {
+		mce_apei_read_done = 1;
+		return rc;
+	}
+	*ubuf += sizeof(struct mce);
+
+	return 0;
+}
+
 static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize,
 			loff_t *off)
 {
@@ -1506,15 +1552,19 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize,
 		return -ENOMEM;
 
 	mutex_lock(&mce_read_mutex);
+
+	if (!mce_apei_read_done) {
+		err = __mce_read_apei(&buf, usize);
+		if (err || buf != ubuf)
+			goto out;
+	}
+
 	next = rcu_dereference_check_mce(mcelog.next);
 
 	/* Only supports full reads right now */
-	if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) {
-		mutex_unlock(&mce_read_mutex);
-		kfree(cpu_tsc);
-
-		return -EINVAL;
-	}
+	err = -EINVAL;
+	if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce))
+		goto out;
 
 	err = 0;
 	prev = 0;
@@ -1562,10 +1612,15 @@ timeout:
 			memset(&mcelog.entry[i], 0, sizeof(struct mce));
 		}
 	}
+
+	if (err)
+		err = -EFAULT;
+
+out:
 	mutex_unlock(&mce_read_mutex);
 	kfree(cpu_tsc);
 
-	return err ? -EFAULT : buf - ubuf;
+	return err ? err : buf - ubuf;
 }
 
 static unsigned int mce_poll(struct file *file, poll_table *wait)
@@ -1573,6 +1628,8 @@ static unsigned int mce_poll(struct file *file, poll_table *wait)
 	poll_wait(file, &mce_wait, wait);
 	if (rcu_dereference_check_mce(mcelog.next))
 		return POLLIN | POLLRDNORM;
+	if (!mce_apei_read_done && apei_check_mce())
+		return POLLIN | POLLRDNORM;
 	return 0;
 }
 
diff --git a/arch/x86/kernel/cpu/mcheck/therm_throt.c b/arch/x86/kernel/cpu/mcheck/therm_throt.c
index 81c499eceb2..e1a0a3bf971 100644
--- a/arch/x86/kernel/cpu/mcheck/therm_throt.c
+++ b/arch/x86/kernel/cpu/mcheck/therm_throt.c
@@ -190,7 +190,7 @@ thermal_throttle_cpu_callback(struct notifier_block *nfb,
 		mutex_unlock(&therm_cpu_lock);
 		break;
 	}
-	return err ? NOTIFY_BAD : NOTIFY_OK;
+	return notifier_from_errno(err);
 }
 
 static struct notifier_block thermal_throttle_cpu_notifier __cpuinitdata =
diff --git a/arch/x86/kernel/cpu/mshyperv.c b/arch/x86/kernel/cpu/mshyperv.c
new file mode 100644
index 00000000000..16f41bbe46b
--- /dev/null
+++ b/arch/x86/kernel/cpu/mshyperv.c
@@ -0,0 +1,55 @@
+/*
+ * HyperV  Detection code.
+ *
+ * Copyright (C) 2010, Novell, Inc.
+ * Author : K. Y. Srinivasan <ksrinivasan@novell.com>
+ *
+ * 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; version 2 of the License.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/module.h>
+#include <asm/processor.h>
+#include <asm/hypervisor.h>
+#include <asm/hyperv.h>
+#include <asm/mshyperv.h>
+
+struct ms_hyperv_info ms_hyperv;
+
+static bool __init ms_hyperv_platform(void)
+{
+	u32 eax;
+	u32 hyp_signature[3];
+
+	if (!boot_cpu_has(X86_FEATURE_HYPERVISOR))
+		return false;
+
+	cpuid(HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS,
+	      &eax, &hyp_signature[0], &hyp_signature[1], &hyp_signature[2]);
+
+	return eax >= HYPERV_CPUID_MIN &&
+		eax <= HYPERV_CPUID_MAX &&
+		!memcmp("Microsoft Hv", hyp_signature, 12);
+}
+
+static void __init ms_hyperv_init_platform(void)
+{
+	/*
+	 * Extract the features and hints
+	 */
+	ms_hyperv.features = cpuid_eax(HYPERV_CPUID_FEATURES);
+	ms_hyperv.hints    = cpuid_eax(HYPERV_CPUID_ENLIGHTMENT_INFO);
+
+	printk(KERN_INFO "HyperV: features 0x%x, hints 0x%x\n",
+	       ms_hyperv.features, ms_hyperv.hints);
+}
+
+const __refconst struct hypervisor_x86 x86_hyper_ms_hyperv = {
+	.name			= "Microsoft HyperV",
+	.detect			= ms_hyperv_platform,
+	.init_platform		= ms_hyperv_init_platform,
+};
+EXPORT_SYMBOL(x86_hyper_ms_hyperv);
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index db5bdc8addf..c77586061bc 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -31,46 +31,51 @@
 #include <asm/nmi.h>
 #include <asm/compat.h>
 
-static u64 perf_event_mask __read_mostly;
+#if 0
+#undef wrmsrl
+#define wrmsrl(msr, val) 					\
+do {								\
+	trace_printk("wrmsrl(%lx, %lx)\n", (unsigned long)(msr),\
+			(unsigned long)(val));			\
+	native_write_msr((msr), (u32)((u64)(val)), 		\
+			(u32)((u64)(val) >> 32));		\
+} while (0)
+#endif
 
-/* The maximal number of PEBS events: */
-#define MAX_PEBS_EVENTS	4
+/*
+ * best effort, GUP based copy_from_user() that assumes IRQ or NMI context
+ */
+static unsigned long
+copy_from_user_nmi(void *to, const void __user *from, unsigned long n)
+{
+	unsigned long offset, addr = (unsigned long)from;
+	int type = in_nmi() ? KM_NMI : KM_IRQ0;
+	unsigned long size, len = 0;
+	struct page *page;
+	void *map;
+	int ret;
 
-/* The size of a BTS record in bytes: */
-#define BTS_RECORD_SIZE		24
+	do {
+		ret = __get_user_pages_fast(addr, 1, 0, &page);
+		if (!ret)
+			break;
 
-/* The size of a per-cpu BTS buffer in bytes: */
-#define BTS_BUFFER_SIZE		(BTS_RECORD_SIZE * 2048)
+		offset = addr & (PAGE_SIZE - 1);
+		size = min(PAGE_SIZE - offset, n - len);
 
-/* The BTS overflow threshold in bytes from the end of the buffer: */
-#define BTS_OVFL_TH		(BTS_RECORD_SIZE * 128)
+		map = kmap_atomic(page, type);
+		memcpy(to, map+offset, size);
+		kunmap_atomic(map, type);
+		put_page(page);
 
+		len  += size;
+		to   += size;
+		addr += size;
 
-/*
- * Bits in the debugctlmsr controlling branch tracing.
- */
-#define X86_DEBUGCTL_TR			(1 << 6)
-#define X86_DEBUGCTL_BTS		(1 << 7)
-#define X86_DEBUGCTL_BTINT		(1 << 8)
-#define X86_DEBUGCTL_BTS_OFF_OS		(1 << 9)
-#define X86_DEBUGCTL_BTS_OFF_USR	(1 << 10)
+	} while (len < n);
 
-/*
- * A debug store configuration.
- *
- * We only support architectures that use 64bit fields.
- */
-struct debug_store {
-	u64	bts_buffer_base;
-	u64	bts_index;
-	u64	bts_absolute_maximum;
-	u64	bts_interrupt_threshold;
-	u64	pebs_buffer_base;
-	u64	pebs_index;
-	u64	pebs_absolute_maximum;
-	u64	pebs_interrupt_threshold;
-	u64	pebs_event_reset[MAX_PEBS_EVENTS];
-};
+	return len;
+}
 
 struct event_constraint {
 	union {
@@ -89,18 +94,41 @@ struct amd_nb {
 	struct event_constraint event_constraints[X86_PMC_IDX_MAX];
 };
 
+#define MAX_LBR_ENTRIES		16
+
 struct cpu_hw_events {
+	/*
+	 * Generic x86 PMC bits
+	 */
 	struct perf_event	*events[X86_PMC_IDX_MAX]; /* in counter order */
 	unsigned long		active_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-	unsigned long		interrupts;
 	int			enabled;
-	struct debug_store	*ds;
 
 	int			n_events;
 	int			n_added;
 	int			assign[X86_PMC_IDX_MAX]; /* event to counter assignment */
 	u64			tags[X86_PMC_IDX_MAX];
 	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
+
+	unsigned int		group_flag;
+
+	/*
+	 * Intel DebugStore bits
+	 */
+	struct debug_store	*ds;
+	u64			pebs_enabled;
+
+	/*
+	 * Intel LBR bits
+	 */
+	int				lbr_users;
+	void				*lbr_context;
+	struct perf_branch_stack	lbr_stack;
+	struct perf_branch_entry	lbr_entries[MAX_LBR_ENTRIES];
+
+	/*
+	 * AMD specific bits
+	 */
 	struct amd_nb		*amd_nb;
 };
 
@@ -114,44 +142,75 @@ struct cpu_hw_events {
 #define EVENT_CONSTRAINT(c, n, m)	\
 	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n))
 
+/*
+ * Constraint on the Event code.
+ */
 #define INTEL_EVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVTSEL_MASK)
+	EVENT_CONSTRAINT(c, n, ARCH_PERFMON_EVENTSEL_EVENT)
 
+/*
+ * Constraint on the Event code + UMask + fixed-mask
+ *
+ * filter mask to validate fixed counter events.
+ * the following filters disqualify for fixed counters:
+ *  - inv
+ *  - edge
+ *  - cnt-mask
+ *  The other filters are supported by fixed counters.
+ *  The any-thread option is supported starting with v3.
+ */
 #define FIXED_EVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, (1ULL << (32+n)), INTEL_ARCH_FIXED_MASK)
+	EVENT_CONSTRAINT(c, (1ULL << (32+n)), X86_RAW_EVENT_MASK)
+
+/*
+ * Constraint on the Event code + UMask
+ */
+#define PEBS_EVENT_CONSTRAINT(c, n)	\
+	EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK)
 
 #define EVENT_CONSTRAINT_END		\
 	EVENT_CONSTRAINT(0, 0, 0)
 
 #define for_each_event_constraint(e, c)	\
-	for ((e) = (c); (e)->cmask; (e)++)
+	for ((e) = (c); (e)->weight; (e)++)
+
+union perf_capabilities {
+	struct {
+		u64	lbr_format    : 6;
+		u64	pebs_trap     : 1;
+		u64	pebs_arch_reg : 1;
+		u64	pebs_format   : 4;
+		u64	smm_freeze    : 1;
+	};
+	u64	capabilities;
+};
 
 /*
  * struct x86_pmu - generic x86 pmu
  */
 struct x86_pmu {
+	/*
+	 * Generic x86 PMC bits
+	 */
 	const char	*name;
 	int		version;
 	int		(*handle_irq)(struct pt_regs *);
 	void		(*disable_all)(void);
-	void		(*enable_all)(void);
+	void		(*enable_all)(int added);
 	void		(*enable)(struct perf_event *);
 	void		(*disable)(struct perf_event *);
+	int		(*hw_config)(struct perf_event *event);
+	int		(*schedule_events)(struct cpu_hw_events *cpuc, int n, int *assign);
 	unsigned	eventsel;
 	unsigned	perfctr;
 	u64		(*event_map)(int);
-	u64		(*raw_event)(u64);
 	int		max_events;
-	int		num_events;
-	int		num_events_fixed;
-	int		event_bits;
-	u64		event_mask;
+	int		num_counters;
+	int		num_counters_fixed;
+	int		cntval_bits;
+	u64		cntval_mask;
 	int		apic;
 	u64		max_period;
-	u64		intel_ctrl;
-	void		(*enable_bts)(u64 config);
-	void		(*disable_bts)(void);
-
 	struct event_constraint *
 			(*get_event_constraints)(struct cpu_hw_events *cpuc,
 						 struct perf_event *event);
@@ -159,11 +218,32 @@ struct x86_pmu {
 	void		(*put_event_constraints)(struct cpu_hw_events *cpuc,
 						 struct perf_event *event);
 	struct event_constraint *event_constraints;
+	void		(*quirks)(void);
 
 	int		(*cpu_prepare)(int cpu);
 	void		(*cpu_starting)(int cpu);
 	void		(*cpu_dying)(int cpu);
 	void		(*cpu_dead)(int cpu);
+
+	/*
+	 * Intel Arch Perfmon v2+
+	 */
+	u64			intel_ctrl;
+	union perf_capabilities intel_cap;
+
+	/*
+	 * Intel DebugStore bits
+	 */
+	int		bts, pebs;
+	int		pebs_record_size;
+	void		(*drain_pebs)(struct pt_regs *regs);
+	struct event_constraint *pebs_constraints;
+
+	/*
+	 * Intel LBR
+	 */
+	unsigned long	lbr_tos, lbr_from, lbr_to; /* MSR base regs       */
+	int		lbr_nr;			   /* hardware stack size */
 };
 
 static struct x86_pmu x86_pmu __read_mostly;
@@ -198,7 +278,7 @@ static u64
 x86_perf_event_update(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
-	int shift = 64 - x86_pmu.event_bits;
+	int shift = 64 - x86_pmu.cntval_bits;
 	u64 prev_raw_count, new_raw_count;
 	int idx = hwc->idx;
 	s64 delta;
@@ -241,33 +321,32 @@ again:
 static atomic_t active_events;
 static DEFINE_MUTEX(pmc_reserve_mutex);
 
+#ifdef CONFIG_X86_LOCAL_APIC
+
 static bool reserve_pmc_hardware(void)
 {
-#ifdef CONFIG_X86_LOCAL_APIC
 	int i;
 
 	if (nmi_watchdog == NMI_LOCAL_APIC)
 		disable_lapic_nmi_watchdog();
 
-	for (i = 0; i < x86_pmu.num_events; i++) {
+	for (i = 0; i < x86_pmu.num_counters; i++) {
 		if (!reserve_perfctr_nmi(x86_pmu.perfctr + i))
 			goto perfctr_fail;
 	}
 
-	for (i = 0; i < x86_pmu.num_events; i++) {
+	for (i = 0; i < x86_pmu.num_counters; i++) {
 		if (!reserve_evntsel_nmi(x86_pmu.eventsel + i))
 			goto eventsel_fail;
 	}
-#endif
 
 	return true;
 
-#ifdef CONFIG_X86_LOCAL_APIC
 eventsel_fail:
 	for (i--; i >= 0; i--)
 		release_evntsel_nmi(x86_pmu.eventsel + i);
 
-	i = x86_pmu.num_events;
+	i = x86_pmu.num_counters;
 
 perfctr_fail:
 	for (i--; i >= 0; i--)
@@ -277,128 +356,36 @@ perfctr_fail:
 		enable_lapic_nmi_watchdog();
 
 	return false;
-#endif
 }
 
 static void release_pmc_hardware(void)
 {
-#ifdef CONFIG_X86_LOCAL_APIC
 	int i;
 
-	for (i = 0; i < x86_pmu.num_events; i++) {
+	for (i = 0; i < x86_pmu.num_counters; i++) {
 		release_perfctr_nmi(x86_pmu.perfctr + i);
 		release_evntsel_nmi(x86_pmu.eventsel + i);
 	}
 
 	if (nmi_watchdog == NMI_LOCAL_APIC)
 		enable_lapic_nmi_watchdog();
-#endif
-}
-
-static inline bool bts_available(void)
-{
-	return x86_pmu.enable_bts != NULL;
-}
-
-static void init_debug_store_on_cpu(int cpu)
-{
-	struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
-
-	if (!ds)
-		return;
-
-	wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA,
-		     (u32)((u64)(unsigned long)ds),
-		     (u32)((u64)(unsigned long)ds >> 32));
-}
-
-static void fini_debug_store_on_cpu(int cpu)
-{
-	if (!per_cpu(cpu_hw_events, cpu).ds)
-		return;
-
-	wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, 0, 0);
-}
-
-static void release_bts_hardware(void)
-{
-	int cpu;
-
-	if (!bts_available())
-		return;
-
-	get_online_cpus();
-
-	for_each_online_cpu(cpu)
-		fini_debug_store_on_cpu(cpu);
-
-	for_each_possible_cpu(cpu) {
-		struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
-
-		if (!ds)
-			continue;
-
-		per_cpu(cpu_hw_events, cpu).ds = NULL;
-
-		kfree((void *)(unsigned long)ds->bts_buffer_base);
-		kfree(ds);
-	}
-
-	put_online_cpus();
 }
 
-static int reserve_bts_hardware(void)
-{
-	int cpu, err = 0;
-
-	if (!bts_available())
-		return 0;
-
-	get_online_cpus();
-
-	for_each_possible_cpu(cpu) {
-		struct debug_store *ds;
-		void *buffer;
-
-		err = -ENOMEM;
-		buffer = kzalloc(BTS_BUFFER_SIZE, GFP_KERNEL);
-		if (unlikely(!buffer))
-			break;
-
-		ds = kzalloc(sizeof(*ds), GFP_KERNEL);
-		if (unlikely(!ds)) {
-			kfree(buffer);
-			break;
-		}
-
-		ds->bts_buffer_base = (u64)(unsigned long)buffer;
-		ds->bts_index = ds->bts_buffer_base;
-		ds->bts_absolute_maximum =
-			ds->bts_buffer_base + BTS_BUFFER_SIZE;
-		ds->bts_interrupt_threshold =
-			ds->bts_absolute_maximum - BTS_OVFL_TH;
-
-		per_cpu(cpu_hw_events, cpu).ds = ds;
-		err = 0;
-	}
+#else
 
-	if (err)
-		release_bts_hardware();
-	else {
-		for_each_online_cpu(cpu)
-			init_debug_store_on_cpu(cpu);
-	}
+static bool reserve_pmc_hardware(void) { return true; }
+static void release_pmc_hardware(void) {}
 
-	put_online_cpus();
+#endif
 
-	return err;
-}
+static int reserve_ds_buffers(void);
+static void release_ds_buffers(void);
 
 static void hw_perf_event_destroy(struct perf_event *event)
 {
 	if (atomic_dec_and_mutex_lock(&active_events, &pmc_reserve_mutex)) {
 		release_pmc_hardware();
-		release_bts_hardware();
+		release_ds_buffers();
 		mutex_unlock(&pmc_reserve_mutex);
 	}
 }
@@ -441,54 +428,11 @@ set_ext_hw_attr(struct hw_perf_event *hwc, struct perf_event_attr *attr)
 	return 0;
 }
 
-/*
- * Setup the hardware configuration for a given attr_type
- */
-static int __hw_perf_event_init(struct perf_event *event)
+static int x86_setup_perfctr(struct perf_event *event)
 {
 	struct perf_event_attr *attr = &event->attr;
 	struct hw_perf_event *hwc = &event->hw;
 	u64 config;
-	int err;
-
-	if (!x86_pmu_initialized())
-		return -ENODEV;
-
-	err = 0;
-	if (!atomic_inc_not_zero(&active_events)) {
-		mutex_lock(&pmc_reserve_mutex);
-		if (atomic_read(&active_events) == 0) {
-			if (!reserve_pmc_hardware())
-				err = -EBUSY;
-			else
-				err = reserve_bts_hardware();
-		}
-		if (!err)
-			atomic_inc(&active_events);
-		mutex_unlock(&pmc_reserve_mutex);
-	}
-	if (err)
-		return err;
-
-	event->destroy = hw_perf_event_destroy;
-
-	/*
-	 * Generate PMC IRQs:
-	 * (keep 'enabled' bit clear for now)
-	 */
-	hwc->config = ARCH_PERFMON_EVENTSEL_INT;
-
-	hwc->idx = -1;
-	hwc->last_cpu = -1;
-	hwc->last_tag = ~0ULL;
-
-	/*
-	 * Count user and OS events unless requested not to.
-	 */
-	if (!attr->exclude_user)
-		hwc->config |= ARCH_PERFMON_EVENTSEL_USR;
-	if (!attr->exclude_kernel)
-		hwc->config |= ARCH_PERFMON_EVENTSEL_OS;
 
 	if (!hwc->sample_period) {
 		hwc->sample_period = x86_pmu.max_period;
@@ -505,16 +449,8 @@ static int __hw_perf_event_init(struct perf_event *event)
 			return -EOPNOTSUPP;
 	}
 
-	/*
-	 * Raw hw_event type provide the config in the hw_event structure
-	 */
-	if (attr->type == PERF_TYPE_RAW) {
-		hwc->config |= x86_pmu.raw_event(attr->config);
-		if ((hwc->config & ARCH_PERFMON_EVENTSEL_ANY) &&
-		    perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN))
-			return -EACCES;
+	if (attr->type == PERF_TYPE_RAW)
 		return 0;
-	}
 
 	if (attr->type == PERF_TYPE_HW_CACHE)
 		return set_ext_hw_attr(hwc, attr);
@@ -539,11 +475,11 @@ static int __hw_perf_event_init(struct perf_event *event)
 	if ((attr->config == PERF_COUNT_HW_BRANCH_INSTRUCTIONS) &&
 	    (hwc->sample_period == 1)) {
 		/* BTS is not supported by this architecture. */
-		if (!bts_available())
+		if (!x86_pmu.bts)
 			return -EOPNOTSUPP;
 
 		/* BTS is currently only allowed for user-mode. */
-		if (hwc->config & ARCH_PERFMON_EVENTSEL_OS)
+		if (!attr->exclude_kernel)
 			return -EOPNOTSUPP;
 	}
 
@@ -552,12 +488,87 @@ static int __hw_perf_event_init(struct perf_event *event)
 	return 0;
 }
 
+static int x86_pmu_hw_config(struct perf_event *event)
+{
+	if (event->attr.precise_ip) {
+		int precise = 0;
+
+		/* Support for constant skid */
+		if (x86_pmu.pebs)
+			precise++;
+
+		/* Support for IP fixup */
+		if (x86_pmu.lbr_nr)
+			precise++;
+
+		if (event->attr.precise_ip > precise)
+			return -EOPNOTSUPP;
+	}
+
+	/*
+	 * Generate PMC IRQs:
+	 * (keep 'enabled' bit clear for now)
+	 */
+	event->hw.config = ARCH_PERFMON_EVENTSEL_INT;
+
+	/*
+	 * Count user and OS events unless requested not to
+	 */
+	if (!event->attr.exclude_user)
+		event->hw.config |= ARCH_PERFMON_EVENTSEL_USR;
+	if (!event->attr.exclude_kernel)
+		event->hw.config |= ARCH_PERFMON_EVENTSEL_OS;
+
+	if (event->attr.type == PERF_TYPE_RAW)
+		event->hw.config |= event->attr.config & X86_RAW_EVENT_MASK;
+
+	return x86_setup_perfctr(event);
+}
+
+/*
+ * Setup the hardware configuration for a given attr_type
+ */
+static int __hw_perf_event_init(struct perf_event *event)
+{
+	int err;
+
+	if (!x86_pmu_initialized())
+		return -ENODEV;
+
+	err = 0;
+	if (!atomic_inc_not_zero(&active_events)) {
+		mutex_lock(&pmc_reserve_mutex);
+		if (atomic_read(&active_events) == 0) {
+			if (!reserve_pmc_hardware())
+				err = -EBUSY;
+			else {
+				err = reserve_ds_buffers();
+				if (err)
+					release_pmc_hardware();
+			}
+		}
+		if (!err)
+			atomic_inc(&active_events);
+		mutex_unlock(&pmc_reserve_mutex);
+	}
+	if (err)
+		return err;
+
+	event->destroy = hw_perf_event_destroy;
+
+	event->hw.idx = -1;
+	event->hw.last_cpu = -1;
+	event->hw.last_tag = ~0ULL;
+
+	return x86_pmu.hw_config(event);
+}
+
 static void x86_pmu_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	int idx;
 
-	for (idx = 0; idx < x86_pmu.num_events; idx++) {
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
 		u64 val;
 
 		if (!test_bit(idx, cpuc->active_mask))
@@ -587,12 +598,12 @@ void hw_perf_disable(void)
 	x86_pmu.disable_all();
 }
 
-static void x86_pmu_enable_all(void)
+static void x86_pmu_enable_all(int added)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	int idx;
 
-	for (idx = 0; idx < x86_pmu.num_events; idx++) {
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
 		struct perf_event *event = cpuc->events[idx];
 		u64 val;
 
@@ -667,14 +678,14 @@ static int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 	 * assign events to counters starting with most
 	 * constrained events.
 	 */
-	wmax = x86_pmu.num_events;
+	wmax = x86_pmu.num_counters;
 
 	/*
 	 * when fixed event counters are present,
 	 * wmax is incremented by 1 to account
 	 * for one more choice
 	 */
-	if (x86_pmu.num_events_fixed)
+	if (x86_pmu.num_counters_fixed)
 		wmax++;
 
 	for (w = 1, num = n; num && w <= wmax; w++) {
@@ -724,7 +735,7 @@ static int collect_events(struct cpu_hw_events *cpuc, struct perf_event *leader,
 	struct perf_event *event;
 	int n, max_count;
 
-	max_count = x86_pmu.num_events + x86_pmu.num_events_fixed;
+	max_count = x86_pmu.num_counters + x86_pmu.num_counters_fixed;
 
 	/* current number of events already accepted */
 	n = cpuc->n_events;
@@ -795,7 +806,7 @@ void hw_perf_enable(void)
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct perf_event *event;
 	struct hw_perf_event *hwc;
-	int i;
+	int i, added = cpuc->n_added;
 
 	if (!x86_pmu_initialized())
 		return;
@@ -847,19 +858,20 @@ void hw_perf_enable(void)
 	cpuc->enabled = 1;
 	barrier();
 
-	x86_pmu.enable_all();
+	x86_pmu.enable_all(added);
 }
 
-static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc)
+static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc,
+					  u64 enable_mask)
 {
-	(void)checking_wrmsrl(hwc->config_base + hwc->idx,
-			      hwc->config | ARCH_PERFMON_EVENTSEL_ENABLE);
+	wrmsrl(hwc->config_base + hwc->idx, hwc->config | enable_mask);
 }
 
 static inline void x86_pmu_disable_event(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
-	(void)checking_wrmsrl(hwc->config_base + hwc->idx, hwc->config);
+
+	wrmsrl(hwc->config_base + hwc->idx, hwc->config);
 }
 
 static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left);
@@ -874,7 +886,7 @@ x86_perf_event_set_period(struct perf_event *event)
 	struct hw_perf_event *hwc = &event->hw;
 	s64 left = atomic64_read(&hwc->period_left);
 	s64 period = hwc->sample_period;
-	int err, ret = 0, idx = hwc->idx;
+	int ret = 0, idx = hwc->idx;
 
 	if (idx == X86_PMC_IDX_FIXED_BTS)
 		return 0;
@@ -912,8 +924,8 @@ x86_perf_event_set_period(struct perf_event *event)
 	 */
 	atomic64_set(&hwc->prev_count, (u64)-left);
 
-	err = checking_wrmsrl(hwc->event_base + idx,
-			     (u64)(-left) & x86_pmu.event_mask);
+	wrmsrl(hwc->event_base + idx,
+			(u64)(-left) & x86_pmu.cntval_mask);
 
 	perf_event_update_userpage(event);
 
@@ -924,7 +936,8 @@ static void x86_pmu_enable_event(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	if (cpuc->enabled)
-		__x86_pmu_enable_event(&event->hw);
+		__x86_pmu_enable_event(&event->hw,
+				       ARCH_PERFMON_EVENTSEL_ENABLE);
 }
 
 /*
@@ -950,7 +963,15 @@ static int x86_pmu_enable(struct perf_event *event)
 	if (n < 0)
 		return n;
 
-	ret = x86_schedule_events(cpuc, n, assign);
+	/*
+	 * If group events scheduling transaction was started,
+	 * skip the schedulability test here, it will be peformed
+	 * at commit time(->commit_txn) as a whole
+	 */
+	if (cpuc->group_flag & PERF_EVENT_TXN_STARTED)
+		goto out;
+
+	ret = x86_pmu.schedule_events(cpuc, n, assign);
 	if (ret)
 		return ret;
 	/*
@@ -959,6 +980,7 @@ static int x86_pmu_enable(struct perf_event *event)
 	 */
 	memcpy(cpuc->assign, assign, n*sizeof(int));
 
+out:
 	cpuc->n_events = n;
 	cpuc->n_added += n - n0;
 
@@ -991,11 +1013,12 @@ static void x86_pmu_unthrottle(struct perf_event *event)
 void perf_event_print_debug(void)
 {
 	u64 ctrl, status, overflow, pmc_ctrl, pmc_count, prev_left, fixed;
+	u64 pebs;
 	struct cpu_hw_events *cpuc;
 	unsigned long flags;
 	int cpu, idx;
 
-	if (!x86_pmu.num_events)
+	if (!x86_pmu.num_counters)
 		return;
 
 	local_irq_save(flags);
@@ -1008,16 +1031,18 @@ void perf_event_print_debug(void)
 		rdmsrl(MSR_CORE_PERF_GLOBAL_STATUS, status);
 		rdmsrl(MSR_CORE_PERF_GLOBAL_OVF_CTRL, overflow);
 		rdmsrl(MSR_ARCH_PERFMON_FIXED_CTR_CTRL, fixed);
+		rdmsrl(MSR_IA32_PEBS_ENABLE, pebs);
 
 		pr_info("\n");
 		pr_info("CPU#%d: ctrl:       %016llx\n", cpu, ctrl);
 		pr_info("CPU#%d: status:     %016llx\n", cpu, status);
 		pr_info("CPU#%d: overflow:   %016llx\n", cpu, overflow);
 		pr_info("CPU#%d: fixed:      %016llx\n", cpu, fixed);
+		pr_info("CPU#%d: pebs:       %016llx\n", cpu, pebs);
 	}
-	pr_info("CPU#%d: active:       %016llx\n", cpu, *(u64 *)cpuc->active_mask);
+	pr_info("CPU#%d: active:     %016llx\n", cpu, *(u64 *)cpuc->active_mask);
 
-	for (idx = 0; idx < x86_pmu.num_events; idx++) {
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
 		rdmsrl(x86_pmu.eventsel + idx, pmc_ctrl);
 		rdmsrl(x86_pmu.perfctr  + idx, pmc_count);
 
@@ -1030,7 +1055,7 @@ void perf_event_print_debug(void)
 		pr_info("CPU#%d:   gen-PMC%d left:  %016llx\n",
 			cpu, idx, prev_left);
 	}
-	for (idx = 0; idx < x86_pmu.num_events_fixed; idx++) {
+	for (idx = 0; idx < x86_pmu.num_counters_fixed; idx++) {
 		rdmsrl(MSR_ARCH_PERFMON_FIXED_CTR0 + idx, pmc_count);
 
 		pr_info("CPU#%d: fixed-PMC%d count: %016llx\n",
@@ -1095,7 +1120,7 @@ static int x86_pmu_handle_irq(struct pt_regs *regs)
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
 
-	for (idx = 0; idx < x86_pmu.num_events; idx++) {
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
 		if (!test_bit(idx, cpuc->active_mask))
 			continue;
 
@@ -1103,7 +1128,7 @@ static int x86_pmu_handle_irq(struct pt_regs *regs)
 		hwc = &event->hw;
 
 		val = x86_perf_event_update(event);
-		if (val & (1ULL << (x86_pmu.event_bits - 1)))
+		if (val & (1ULL << (x86_pmu.cntval_bits - 1)))
 			continue;
 
 		/*
@@ -1146,7 +1171,6 @@ void set_perf_event_pending(void)
 
 void perf_events_lapic_init(void)
 {
-#ifdef CONFIG_X86_LOCAL_APIC
 	if (!x86_pmu.apic || !x86_pmu_initialized())
 		return;
 
@@ -1154,7 +1178,6 @@ void perf_events_lapic_init(void)
 	 * Always use NMI for PMU
 	 */
 	apic_write(APIC_LVTPC, APIC_DM_NMI);
-#endif
 }
 
 static int __kprobes
@@ -1178,9 +1201,7 @@ perf_event_nmi_handler(struct notifier_block *self,
 
 	regs = args->regs;
 
-#ifdef CONFIG_X86_LOCAL_APIC
 	apic_write(APIC_LVTPC, APIC_DM_NMI);
-#endif
 	/*
 	 * Can't rely on the handled return value to say it was our NMI, two
 	 * events could trigger 'simultaneously' raising two back-to-back NMIs.
@@ -1217,118 +1238,11 @@ x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 	return &unconstrained;
 }
 
-static int x86_event_sched_in(struct perf_event *event,
-			  struct perf_cpu_context *cpuctx)
-{
-	int ret = 0;
-
-	event->state = PERF_EVENT_STATE_ACTIVE;
-	event->oncpu = smp_processor_id();
-	event->tstamp_running += event->ctx->time - event->tstamp_stopped;
-
-	if (!is_x86_event(event))
-		ret = event->pmu->enable(event);
-
-	if (!ret && !is_software_event(event))
-		cpuctx->active_oncpu++;
-
-	if (!ret && event->attr.exclusive)
-		cpuctx->exclusive = 1;
-
-	return ret;
-}
-
-static void x86_event_sched_out(struct perf_event *event,
-			    struct perf_cpu_context *cpuctx)
-{
-	event->state = PERF_EVENT_STATE_INACTIVE;
-	event->oncpu = -1;
-
-	if (!is_x86_event(event))
-		event->pmu->disable(event);
-
-	event->tstamp_running -= event->ctx->time - event->tstamp_stopped;
-
-	if (!is_software_event(event))
-		cpuctx->active_oncpu--;
-
-	if (event->attr.exclusive || !cpuctx->active_oncpu)
-		cpuctx->exclusive = 0;
-}
-
-/*
- * Called to enable a whole group of events.
- * Returns 1 if the group was enabled, or -EAGAIN if it could not be.
- * Assumes the caller has disabled interrupts and has
- * frozen the PMU with hw_perf_save_disable.
- *
- * called with PMU disabled. If successful and return value 1,
- * then guaranteed to call perf_enable() and hw_perf_enable()
- */
-int hw_perf_group_sched_in(struct perf_event *leader,
-	       struct perf_cpu_context *cpuctx,
-	       struct perf_event_context *ctx)
-{
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
-	struct perf_event *sub;
-	int assign[X86_PMC_IDX_MAX];
-	int n0, n1, ret;
-
-	/* n0 = total number of events */
-	n0 = collect_events(cpuc, leader, true);
-	if (n0 < 0)
-		return n0;
-
-	ret = x86_schedule_events(cpuc, n0, assign);
-	if (ret)
-		return ret;
-
-	ret = x86_event_sched_in(leader, cpuctx);
-	if (ret)
-		return ret;
-
-	n1 = 1;
-	list_for_each_entry(sub, &leader->sibling_list, group_entry) {
-		if (sub->state > PERF_EVENT_STATE_OFF) {
-			ret = x86_event_sched_in(sub, cpuctx);
-			if (ret)
-				goto undo;
-			++n1;
-		}
-	}
-	/*
-	 * copy new assignment, now we know it is possible
-	 * will be used by hw_perf_enable()
-	 */
-	memcpy(cpuc->assign, assign, n0*sizeof(int));
-
-	cpuc->n_events  = n0;
-	cpuc->n_added  += n1;
-	ctx->nr_active += n1;
-
-	/*
-	 * 1 means successful and events are active
-	 * This is not quite true because we defer
-	 * actual activation until hw_perf_enable() but
-	 * this way we* ensure caller won't try to enable
-	 * individual events
-	 */
-	return 1;
-undo:
-	x86_event_sched_out(leader, cpuctx);
-	n0  = 1;
-	list_for_each_entry(sub, &leader->sibling_list, group_entry) {
-		if (sub->state == PERF_EVENT_STATE_ACTIVE) {
-			x86_event_sched_out(sub, cpuctx);
-			if (++n0 == n1)
-				break;
-		}
-	}
-	return ret;
-}
-
 #include "perf_event_amd.c"
 #include "perf_event_p6.c"
+#include "perf_event_p4.c"
+#include "perf_event_intel_lbr.c"
+#include "perf_event_intel_ds.c"
 #include "perf_event_intel.c"
 
 static int __cpuinit
@@ -1402,48 +1316,50 @@ void __init init_hw_perf_events(void)
 
 	pr_cont("%s PMU driver.\n", x86_pmu.name);
 
-	if (x86_pmu.num_events > X86_PMC_MAX_GENERIC) {
+	if (x86_pmu.quirks)
+		x86_pmu.quirks();
+
+	if (x86_pmu.num_counters > X86_PMC_MAX_GENERIC) {
 		WARN(1, KERN_ERR "hw perf events %d > max(%d), clipping!",
-		     x86_pmu.num_events, X86_PMC_MAX_GENERIC);
-		x86_pmu.num_events = X86_PMC_MAX_GENERIC;
+		     x86_pmu.num_counters, X86_PMC_MAX_GENERIC);
+		x86_pmu.num_counters = X86_PMC_MAX_GENERIC;
 	}
-	perf_event_mask = (1 << x86_pmu.num_events) - 1;
-	perf_max_events = x86_pmu.num_events;
+	x86_pmu.intel_ctrl = (1 << x86_pmu.num_counters) - 1;
+	perf_max_events = x86_pmu.num_counters;
 
-	if (x86_pmu.num_events_fixed > X86_PMC_MAX_FIXED) {
+	if (x86_pmu.num_counters_fixed > X86_PMC_MAX_FIXED) {
 		WARN(1, KERN_ERR "hw perf events fixed %d > max(%d), clipping!",
-		     x86_pmu.num_events_fixed, X86_PMC_MAX_FIXED);
-		x86_pmu.num_events_fixed = X86_PMC_MAX_FIXED;
+		     x86_pmu.num_counters_fixed, X86_PMC_MAX_FIXED);
+		x86_pmu.num_counters_fixed = X86_PMC_MAX_FIXED;
 	}
 
-	perf_event_mask |=
-		((1LL << x86_pmu.num_events_fixed)-1) << X86_PMC_IDX_FIXED;
-	x86_pmu.intel_ctrl = perf_event_mask;
+	x86_pmu.intel_ctrl |=
+		((1LL << x86_pmu.num_counters_fixed)-1) << X86_PMC_IDX_FIXED;
 
 	perf_events_lapic_init();
 	register_die_notifier(&perf_event_nmi_notifier);
 
 	unconstrained = (struct event_constraint)
-		__EVENT_CONSTRAINT(0, (1ULL << x86_pmu.num_events) - 1,
-				   0, x86_pmu.num_events);
+		__EVENT_CONSTRAINT(0, (1ULL << x86_pmu.num_counters) - 1,
+				   0, x86_pmu.num_counters);
 
 	if (x86_pmu.event_constraints) {
 		for_each_event_constraint(c, x86_pmu.event_constraints) {
-			if (c->cmask != INTEL_ARCH_FIXED_MASK)
+			if (c->cmask != X86_RAW_EVENT_MASK)
 				continue;
 
-			c->idxmsk64 |= (1ULL << x86_pmu.num_events) - 1;
-			c->weight += x86_pmu.num_events;
+			c->idxmsk64 |= (1ULL << x86_pmu.num_counters) - 1;
+			c->weight += x86_pmu.num_counters;
 		}
 	}
 
 	pr_info("... version:                %d\n",     x86_pmu.version);
-	pr_info("... bit width:              %d\n",     x86_pmu.event_bits);
-	pr_info("... generic registers:      %d\n",     x86_pmu.num_events);
-	pr_info("... value mask:             %016Lx\n", x86_pmu.event_mask);
+	pr_info("... bit width:              %d\n",     x86_pmu.cntval_bits);
+	pr_info("... generic registers:      %d\n",     x86_pmu.num_counters);
+	pr_info("... value mask:             %016Lx\n", x86_pmu.cntval_mask);
 	pr_info("... max period:             %016Lx\n", x86_pmu.max_period);
-	pr_info("... fixed-purpose events:   %d\n",     x86_pmu.num_events_fixed);
-	pr_info("... event mask:             %016Lx\n", perf_event_mask);
+	pr_info("... fixed-purpose events:   %d\n",     x86_pmu.num_counters_fixed);
+	pr_info("... event mask:             %016Lx\n", x86_pmu.intel_ctrl);
 
 	perf_cpu_notifier(x86_pmu_notifier);
 }
@@ -1453,6 +1369,59 @@ static inline void x86_pmu_read(struct perf_event *event)
 	x86_perf_event_update(event);
 }
 
+/*
+ * Start group events scheduling transaction
+ * Set the flag to make pmu::enable() not perform the
+ * schedulability test, it will be performed at commit time
+ */
+static void x86_pmu_start_txn(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	cpuc->group_flag |= PERF_EVENT_TXN_STARTED;
+}
+
+/*
+ * Stop group events scheduling transaction
+ * Clear the flag and pmu::enable() will perform the
+ * schedulability test.
+ */
+static void x86_pmu_cancel_txn(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	cpuc->group_flag &= ~PERF_EVENT_TXN_STARTED;
+}
+
+/*
+ * Commit group events scheduling transaction
+ * Perform the group schedulability test as a whole
+ * Return 0 if success
+ */
+static int x86_pmu_commit_txn(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	int assign[X86_PMC_IDX_MAX];
+	int n, ret;
+
+	n = cpuc->n_events;
+
+	if (!x86_pmu_initialized())
+		return -EAGAIN;
+
+	ret = x86_pmu.schedule_events(cpuc, n, assign);
+	if (ret)
+		return ret;
+
+	/*
+	 * copy new assignment, now we know it is possible
+	 * will be used by hw_perf_enable()
+	 */
+	memcpy(cpuc->assign, assign, n*sizeof(int));
+
+	return 0;
+}
+
 static const struct pmu pmu = {
 	.enable		= x86_pmu_enable,
 	.disable	= x86_pmu_disable,
@@ -1460,9 +1429,38 @@ static const struct pmu pmu = {
 	.stop		= x86_pmu_stop,
 	.read		= x86_pmu_read,
 	.unthrottle	= x86_pmu_unthrottle,
+	.start_txn	= x86_pmu_start_txn,
+	.cancel_txn	= x86_pmu_cancel_txn,
+	.commit_txn	= x86_pmu_commit_txn,
 };
 
 /*
+ * validate that we can schedule this event
+ */
+static int validate_event(struct perf_event *event)
+{
+	struct cpu_hw_events *fake_cpuc;
+	struct event_constraint *c;
+	int ret = 0;
+
+	fake_cpuc = kmalloc(sizeof(*fake_cpuc), GFP_KERNEL | __GFP_ZERO);
+	if (!fake_cpuc)
+		return -ENOMEM;
+
+	c = x86_pmu.get_event_constraints(fake_cpuc, event);
+
+	if (!c || !c->weight)
+		ret = -ENOSPC;
+
+	if (x86_pmu.put_event_constraints)
+		x86_pmu.put_event_constraints(fake_cpuc, event);
+
+	kfree(fake_cpuc);
+
+	return ret;
+}
+
+/*
  * validate a single event group
  *
  * validation include:
@@ -1502,7 +1500,7 @@ static int validate_group(struct perf_event *event)
 
 	fake_cpuc->n_events = n;
 
-	ret = x86_schedule_events(fake_cpuc, n, NULL);
+	ret = x86_pmu.schedule_events(fake_cpuc, n, NULL);
 
 out_free:
 	kfree(fake_cpuc);
@@ -1527,6 +1525,8 @@ const struct pmu *hw_perf_event_init(struct perf_event *event)
 
 		if (event->group_leader != event)
 			err = validate_group(event);
+		else
+			err = validate_event(event);
 
 		event->pmu = tmp;
 	}
@@ -1574,8 +1574,7 @@ static void backtrace_address(void *data, unsigned long addr, int reliable)
 {
 	struct perf_callchain_entry *entry = data;
 
-	if (reliable)
-		callchain_store(entry, addr);
+	callchain_store(entry, addr);
 }
 
 static const struct stacktrace_ops backtrace_ops = {
@@ -1597,41 +1596,6 @@ perf_callchain_kernel(struct pt_regs *regs, struct perf_callchain_entry *entry)
 	dump_trace(NULL, regs, NULL, regs->bp, &backtrace_ops, entry);
 }
 
-/*
- * best effort, GUP based copy_from_user() that assumes IRQ or NMI context
- */
-static unsigned long
-copy_from_user_nmi(void *to, const void __user *from, unsigned long n)
-{
-	unsigned long offset, addr = (unsigned long)from;
-	int type = in_nmi() ? KM_NMI : KM_IRQ0;
-	unsigned long size, len = 0;
-	struct page *page;
-	void *map;
-	int ret;
-
-	do {
-		ret = __get_user_pages_fast(addr, 1, 0, &page);
-		if (!ret)
-			break;
-
-		offset = addr & (PAGE_SIZE - 1);
-		size = min(PAGE_SIZE - offset, n - len);
-
-		map = kmap_atomic(page, type);
-		memcpy(to, map+offset, size);
-		kunmap_atomic(map, type);
-		put_page(page);
-
-		len  += size;
-		to   += size;
-		addr += size;
-
-	} while (len < n);
-
-	return len;
-}
-
 #ifdef CONFIG_COMPAT
 static inline int
 perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry)
@@ -1727,6 +1691,11 @@ struct perf_callchain_entry *perf_callchain(struct pt_regs *regs)
 {
 	struct perf_callchain_entry *entry;
 
+	if (perf_guest_cbs && perf_guest_cbs->is_in_guest()) {
+		/* TODO: We don't support guest os callchain now */
+		return NULL;
+	}
+
 	if (in_nmi())
 		entry = &__get_cpu_var(pmc_nmi_entry);
 	else
@@ -1748,5 +1717,43 @@ void perf_arch_fetch_caller_regs(struct pt_regs *regs, unsigned long ip, int ski
 	 */
 	regs->bp = rewind_frame_pointer(skip + 1);
 	regs->cs = __KERNEL_CS;
-	local_save_flags(regs->flags);
+	/*
+	 * We abuse bit 3 to pass exact information, see perf_misc_flags
+	 * and the comment with PERF_EFLAGS_EXACT.
+	 */
+	regs->flags = 0;
+}
+
+unsigned long perf_instruction_pointer(struct pt_regs *regs)
+{
+	unsigned long ip;
+
+	if (perf_guest_cbs && perf_guest_cbs->is_in_guest())
+		ip = perf_guest_cbs->get_guest_ip();
+	else
+		ip = instruction_pointer(regs);
+
+	return ip;
+}
+
+unsigned long perf_misc_flags(struct pt_regs *regs)
+{
+	int misc = 0;
+
+	if (perf_guest_cbs && perf_guest_cbs->is_in_guest()) {
+		if (perf_guest_cbs->is_user_mode())
+			misc |= PERF_RECORD_MISC_GUEST_USER;
+		else
+			misc |= PERF_RECORD_MISC_GUEST_KERNEL;
+	} else {
+		if (user_mode(regs))
+			misc |= PERF_RECORD_MISC_USER;
+		else
+			misc |= PERF_RECORD_MISC_KERNEL;
+	}
+
+	if (regs->flags & PERF_EFLAGS_EXACT)
+		misc |= PERF_RECORD_MISC_EXACT_IP;
+
+	return misc;
 }
diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c
index db6f7d4056e..611df11ba15 100644
--- a/arch/x86/kernel/cpu/perf_event_amd.c
+++ b/arch/x86/kernel/cpu/perf_event_amd.c
@@ -2,7 +2,7 @@
 
 static DEFINE_RAW_SPINLOCK(amd_nb_lock);
 
-static __initconst u64 amd_hw_cache_event_ids
+static __initconst const u64 amd_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX] =
@@ -111,22 +111,19 @@ static u64 amd_pmu_event_map(int hw_event)
 	return amd_perfmon_event_map[hw_event];
 }
 
-static u64 amd_pmu_raw_event(u64 hw_event)
+static int amd_pmu_hw_config(struct perf_event *event)
 {
-#define K7_EVNTSEL_EVENT_MASK	0xF000000FFULL
-#define K7_EVNTSEL_UNIT_MASK	0x00000FF00ULL
-#define K7_EVNTSEL_EDGE_MASK	0x000040000ULL
-#define K7_EVNTSEL_INV_MASK	0x000800000ULL
-#define K7_EVNTSEL_REG_MASK	0x0FF000000ULL
-
-#define K7_EVNTSEL_MASK			\
-	(K7_EVNTSEL_EVENT_MASK |	\
-	 K7_EVNTSEL_UNIT_MASK  |	\
-	 K7_EVNTSEL_EDGE_MASK  |	\
-	 K7_EVNTSEL_INV_MASK   |	\
-	 K7_EVNTSEL_REG_MASK)
-
-	return hw_event & K7_EVNTSEL_MASK;
+	int ret = x86_pmu_hw_config(event);
+
+	if (ret)
+		return ret;
+
+	if (event->attr.type != PERF_TYPE_RAW)
+		return 0;
+
+	event->hw.config |= event->attr.config & AMD64_RAW_EVENT_MASK;
+
+	return 0;
 }
 
 /*
@@ -165,7 +162,7 @@ static void amd_put_event_constraints(struct cpu_hw_events *cpuc,
 	 * be removed on one CPU at a time AND PMU is disabled
 	 * when we come here
 	 */
-	for (i = 0; i < x86_pmu.num_events; i++) {
+	for (i = 0; i < x86_pmu.num_counters; i++) {
 		if (nb->owners[i] == event) {
 			cmpxchg(nb->owners+i, event, NULL);
 			break;
@@ -215,7 +212,7 @@ amd_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 	struct hw_perf_event *hwc = &event->hw;
 	struct amd_nb *nb = cpuc->amd_nb;
 	struct perf_event *old = NULL;
-	int max = x86_pmu.num_events;
+	int max = x86_pmu.num_counters;
 	int i, j, k = -1;
 
 	/*
@@ -293,7 +290,7 @@ static struct amd_nb *amd_alloc_nb(int cpu, int nb_id)
 	/*
 	 * initialize all possible NB constraints
 	 */
-	for (i = 0; i < x86_pmu.num_events; i++) {
+	for (i = 0; i < x86_pmu.num_counters; i++) {
 		__set_bit(i, nb->event_constraints[i].idxmsk);
 		nb->event_constraints[i].weight = 1;
 	}
@@ -371,21 +368,22 @@ static void amd_pmu_cpu_dead(int cpu)
 	raw_spin_unlock(&amd_nb_lock);
 }
 
-static __initconst struct x86_pmu amd_pmu = {
+static __initconst const struct x86_pmu amd_pmu = {
 	.name			= "AMD",
 	.handle_irq		= x86_pmu_handle_irq,
 	.disable_all		= x86_pmu_disable_all,
 	.enable_all		= x86_pmu_enable_all,
 	.enable			= x86_pmu_enable_event,
 	.disable		= x86_pmu_disable_event,
+	.hw_config		= amd_pmu_hw_config,
+	.schedule_events	= x86_schedule_events,
 	.eventsel		= MSR_K7_EVNTSEL0,
 	.perfctr		= MSR_K7_PERFCTR0,
 	.event_map		= amd_pmu_event_map,
-	.raw_event		= amd_pmu_raw_event,
 	.max_events		= ARRAY_SIZE(amd_perfmon_event_map),
-	.num_events		= 4,
-	.event_bits		= 48,
-	.event_mask		= (1ULL << 48) - 1,
+	.num_counters		= 4,
+	.cntval_bits		= 48,
+	.cntval_mask		= (1ULL << 48) - 1,
 	.apic			= 1,
 	/* use highest bit to detect overflow */
 	.max_period		= (1ULL << 47) - 1,
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 9c794ac8783..fdbc652d3fe 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -88,7 +88,7 @@ static u64 intel_pmu_event_map(int hw_event)
 	return intel_perfmon_event_map[hw_event];
 }
 
-static __initconst u64 westmere_hw_cache_event_ids
+static __initconst const u64 westmere_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX] =
@@ -179,7 +179,7 @@ static __initconst u64 westmere_hw_cache_event_ids
  },
 };
 
-static __initconst u64 nehalem_hw_cache_event_ids
+static __initconst const u64 nehalem_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX] =
@@ -270,7 +270,7 @@ static __initconst u64 nehalem_hw_cache_event_ids
  },
 };
 
-static __initconst u64 core2_hw_cache_event_ids
+static __initconst const u64 core2_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX] =
@@ -361,7 +361,7 @@ static __initconst u64 core2_hw_cache_event_ids
  },
 };
 
-static __initconst u64 atom_hw_cache_event_ids
+static __initconst const u64 atom_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX] =
@@ -452,60 +452,6 @@ static __initconst u64 atom_hw_cache_event_ids
  },
 };
 
-static u64 intel_pmu_raw_event(u64 hw_event)
-{
-#define CORE_EVNTSEL_EVENT_MASK		0x000000FFULL
-#define CORE_EVNTSEL_UNIT_MASK		0x0000FF00ULL
-#define CORE_EVNTSEL_EDGE_MASK		0x00040000ULL
-#define CORE_EVNTSEL_INV_MASK		0x00800000ULL
-#define CORE_EVNTSEL_REG_MASK		0xFF000000ULL
-
-#define CORE_EVNTSEL_MASK		\
-	(INTEL_ARCH_EVTSEL_MASK |	\
-	 INTEL_ARCH_UNIT_MASK   |	\
-	 INTEL_ARCH_EDGE_MASK   |	\
-	 INTEL_ARCH_INV_MASK    |	\
-	 INTEL_ARCH_CNT_MASK)
-
-	return hw_event & CORE_EVNTSEL_MASK;
-}
-
-static void intel_pmu_enable_bts(u64 config)
-{
-	unsigned long debugctlmsr;
-
-	debugctlmsr = get_debugctlmsr();
-
-	debugctlmsr |= X86_DEBUGCTL_TR;
-	debugctlmsr |= X86_DEBUGCTL_BTS;
-	debugctlmsr |= X86_DEBUGCTL_BTINT;
-
-	if (!(config & ARCH_PERFMON_EVENTSEL_OS))
-		debugctlmsr |= X86_DEBUGCTL_BTS_OFF_OS;
-
-	if (!(config & ARCH_PERFMON_EVENTSEL_USR))
-		debugctlmsr |= X86_DEBUGCTL_BTS_OFF_USR;
-
-	update_debugctlmsr(debugctlmsr);
-}
-
-static void intel_pmu_disable_bts(void)
-{
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
-	unsigned long debugctlmsr;
-
-	if (!cpuc->ds)
-		return;
-
-	debugctlmsr = get_debugctlmsr();
-
-	debugctlmsr &=
-		~(X86_DEBUGCTL_TR | X86_DEBUGCTL_BTS | X86_DEBUGCTL_BTINT |
-		  X86_DEBUGCTL_BTS_OFF_OS | X86_DEBUGCTL_BTS_OFF_USR);
-
-	update_debugctlmsr(debugctlmsr);
-}
-
 static void intel_pmu_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
@@ -514,12 +460,17 @@ static void intel_pmu_disable_all(void)
 
 	if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask))
 		intel_pmu_disable_bts();
+
+	intel_pmu_pebs_disable_all();
+	intel_pmu_lbr_disable_all();
 }
 
-static void intel_pmu_enable_all(void)
+static void intel_pmu_enable_all(int added)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
+	intel_pmu_pebs_enable_all();
+	intel_pmu_lbr_enable_all();
 	wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, x86_pmu.intel_ctrl);
 
 	if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask)) {
@@ -533,6 +484,42 @@ static void intel_pmu_enable_all(void)
 	}
 }
 
+/*
+ * Workaround for:
+ *   Intel Errata AAK100 (model 26)
+ *   Intel Errata AAP53  (model 30)
+ *   Intel Errata BD53   (model 44)
+ *
+ * These chips need to be 'reset' when adding counters by programming
+ * the magic three (non counting) events 0x4300D2, 0x4300B1 and 0x4300B5
+ * either in sequence on the same PMC or on different PMCs.
+ */
+static void intel_pmu_nhm_enable_all(int added)
+{
+	if (added) {
+		struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+		int i;
+
+		wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 0, 0x4300D2);
+		wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 1, 0x4300B1);
+		wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 2, 0x4300B5);
+
+		wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0x3);
+		wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0x0);
+
+		for (i = 0; i < 3; i++) {
+			struct perf_event *event = cpuc->events[i];
+
+			if (!event)
+				continue;
+
+			__x86_pmu_enable_event(&event->hw,
+					       ARCH_PERFMON_EVENTSEL_ENABLE);
+		}
+	}
+	intel_pmu_enable_all(added);
+}
+
 static inline u64 intel_pmu_get_status(void)
 {
 	u64 status;
@@ -547,8 +534,7 @@ static inline void intel_pmu_ack_status(u64 ack)
 	wrmsrl(MSR_CORE_PERF_GLOBAL_OVF_CTRL, ack);
 }
 
-static inline void
-intel_pmu_disable_fixed(struct hw_perf_event *hwc)
+static void intel_pmu_disable_fixed(struct hw_perf_event *hwc)
 {
 	int idx = hwc->idx - X86_PMC_IDX_FIXED;
 	u64 ctrl_val, mask;
@@ -557,71 +543,10 @@ intel_pmu_disable_fixed(struct hw_perf_event *hwc)
 
 	rdmsrl(hwc->config_base, ctrl_val);
 	ctrl_val &= ~mask;
-	(void)checking_wrmsrl(hwc->config_base, ctrl_val);
-}
-
-static void intel_pmu_drain_bts_buffer(void)
-{
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
-	struct debug_store *ds = cpuc->ds;
-	struct bts_record {
-		u64	from;
-		u64	to;
-		u64	flags;
-	};
-	struct perf_event *event = cpuc->events[X86_PMC_IDX_FIXED_BTS];
-	struct bts_record *at, *top;
-	struct perf_output_handle handle;
-	struct perf_event_header header;
-	struct perf_sample_data data;
-	struct pt_regs regs;
-
-	if (!event)
-		return;
-
-	if (!ds)
-		return;
-
-	at  = (struct bts_record *)(unsigned long)ds->bts_buffer_base;
-	top = (struct bts_record *)(unsigned long)ds->bts_index;
-
-	if (top <= at)
-		return;
-
-	ds->bts_index = ds->bts_buffer_base;
-
-	perf_sample_data_init(&data, 0);
-
-	data.period	= event->hw.last_period;
-	regs.ip		= 0;
-
-	/*
-	 * Prepare a generic sample, i.e. fill in the invariant fields.
-	 * We will overwrite the from and to address before we output
-	 * the sample.
-	 */
-	perf_prepare_sample(&header, &data, event, &regs);
-
-	if (perf_output_begin(&handle, event,
-			      header.size * (top - at), 1, 1))
-		return;
-
-	for (; at < top; at++) {
-		data.ip		= at->from;
-		data.addr	= at->to;
-
-		perf_output_sample(&handle, &header, &data, event);
-	}
-
-	perf_output_end(&handle);
-
-	/* There's new data available. */
-	event->hw.interrupts++;
-	event->pending_kill = POLL_IN;
+	wrmsrl(hwc->config_base, ctrl_val);
 }
 
-static inline void
-intel_pmu_disable_event(struct perf_event *event)
+static void intel_pmu_disable_event(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 
@@ -637,14 +562,15 @@ intel_pmu_disable_event(struct perf_event *event)
 	}
 
 	x86_pmu_disable_event(event);
+
+	if (unlikely(event->attr.precise_ip))
+		intel_pmu_pebs_disable(event);
 }
 
-static inline void
-intel_pmu_enable_fixed(struct hw_perf_event *hwc)
+static void intel_pmu_enable_fixed(struct hw_perf_event *hwc)
 {
 	int idx = hwc->idx - X86_PMC_IDX_FIXED;
 	u64 ctrl_val, bits, mask;
-	int err;
 
 	/*
 	 * Enable IRQ generation (0x8),
@@ -669,7 +595,7 @@ intel_pmu_enable_fixed(struct hw_perf_event *hwc)
 	rdmsrl(hwc->config_base, ctrl_val);
 	ctrl_val &= ~mask;
 	ctrl_val |= bits;
-	err = checking_wrmsrl(hwc->config_base, ctrl_val);
+	wrmsrl(hwc->config_base, ctrl_val);
 }
 
 static void intel_pmu_enable_event(struct perf_event *event)
@@ -689,7 +615,10 @@ static void intel_pmu_enable_event(struct perf_event *event)
 		return;
 	}
 
-	__x86_pmu_enable_event(hwc);
+	if (unlikely(event->attr.precise_ip))
+		intel_pmu_pebs_enable(event);
+
+	__x86_pmu_enable_event(hwc, ARCH_PERFMON_EVENTSEL_ENABLE);
 }
 
 /*
@@ -708,20 +637,20 @@ static void intel_pmu_reset(void)
 	unsigned long flags;
 	int idx;
 
-	if (!x86_pmu.num_events)
+	if (!x86_pmu.num_counters)
 		return;
 
 	local_irq_save(flags);
 
 	printk("clearing PMU state on CPU#%d\n", smp_processor_id());
 
-	for (idx = 0; idx < x86_pmu.num_events; idx++) {
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
 		checking_wrmsrl(x86_pmu.eventsel + idx, 0ull);
 		checking_wrmsrl(x86_pmu.perfctr  + idx, 0ull);
 	}
-	for (idx = 0; idx < x86_pmu.num_events_fixed; idx++) {
+	for (idx = 0; idx < x86_pmu.num_counters_fixed; idx++)
 		checking_wrmsrl(MSR_ARCH_PERFMON_FIXED_CTR0 + idx, 0ull);
-	}
+
 	if (ds)
 		ds->bts_index = ds->bts_buffer_base;
 
@@ -747,7 +676,7 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
 	intel_pmu_drain_bts_buffer();
 	status = intel_pmu_get_status();
 	if (!status) {
-		intel_pmu_enable_all();
+		intel_pmu_enable_all(0);
 		return 0;
 	}
 
@@ -762,6 +691,15 @@ again:
 
 	inc_irq_stat(apic_perf_irqs);
 	ack = status;
+
+	intel_pmu_lbr_read();
+
+	/*
+	 * PEBS overflow sets bit 62 in the global status register
+	 */
+	if (__test_and_clear_bit(62, (unsigned long *)&status))
+		x86_pmu.drain_pebs(regs);
+
 	for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) {
 		struct perf_event *event = cpuc->events[bit];
 
@@ -787,26 +725,22 @@ again:
 		goto again;
 
 done:
-	intel_pmu_enable_all();
+	intel_pmu_enable_all(0);
 	return 1;
 }
 
-static struct event_constraint bts_constraint =
-	EVENT_CONSTRAINT(0, 1ULL << X86_PMC_IDX_FIXED_BTS, 0);
-
 static struct event_constraint *
-intel_special_constraints(struct perf_event *event)
+intel_bts_constraints(struct perf_event *event)
 {
-	unsigned int hw_event;
-
-	hw_event = event->hw.config & INTEL_ARCH_EVENT_MASK;
+	struct hw_perf_event *hwc = &event->hw;
+	unsigned int hw_event, bts_event;
 
-	if (unlikely((hw_event ==
-		      x86_pmu.event_map(PERF_COUNT_HW_BRANCH_INSTRUCTIONS)) &&
-		     (event->hw.sample_period == 1))) {
+	hw_event = hwc->config & INTEL_ARCH_EVENT_MASK;
+	bts_event = x86_pmu.event_map(PERF_COUNT_HW_BRANCH_INSTRUCTIONS);
 
+	if (unlikely(hw_event == bts_event && hwc->sample_period == 1))
 		return &bts_constraint;
-	}
+
 	return NULL;
 }
 
@@ -815,24 +749,53 @@ intel_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event
 {
 	struct event_constraint *c;
 
-	c = intel_special_constraints(event);
+	c = intel_bts_constraints(event);
+	if (c)
+		return c;
+
+	c = intel_pebs_constraints(event);
 	if (c)
 		return c;
 
 	return x86_get_event_constraints(cpuc, event);
 }
 
-static __initconst struct x86_pmu core_pmu = {
+static int intel_pmu_hw_config(struct perf_event *event)
+{
+	int ret = x86_pmu_hw_config(event);
+
+	if (ret)
+		return ret;
+
+	if (event->attr.type != PERF_TYPE_RAW)
+		return 0;
+
+	if (!(event->attr.config & ARCH_PERFMON_EVENTSEL_ANY))
+		return 0;
+
+	if (x86_pmu.version < 3)
+		return -EINVAL;
+
+	if (perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	event->hw.config |= ARCH_PERFMON_EVENTSEL_ANY;
+
+	return 0;
+}
+
+static __initconst const struct x86_pmu core_pmu = {
 	.name			= "core",
 	.handle_irq		= x86_pmu_handle_irq,
 	.disable_all		= x86_pmu_disable_all,
 	.enable_all		= x86_pmu_enable_all,
 	.enable			= x86_pmu_enable_event,
 	.disable		= x86_pmu_disable_event,
+	.hw_config		= x86_pmu_hw_config,
+	.schedule_events	= x86_schedule_events,
 	.eventsel		= MSR_ARCH_PERFMON_EVENTSEL0,
 	.perfctr		= MSR_ARCH_PERFMON_PERFCTR0,
 	.event_map		= intel_pmu_event_map,
-	.raw_event		= intel_pmu_raw_event,
 	.max_events		= ARRAY_SIZE(intel_perfmon_event_map),
 	.apic			= 1,
 	/*
@@ -845,17 +808,32 @@ static __initconst struct x86_pmu core_pmu = {
 	.event_constraints	= intel_core_event_constraints,
 };
 
-static __initconst struct x86_pmu intel_pmu = {
+static void intel_pmu_cpu_starting(int cpu)
+{
+	init_debug_store_on_cpu(cpu);
+	/*
+	 * Deal with CPUs that don't clear their LBRs on power-up.
+	 */
+	intel_pmu_lbr_reset();
+}
+
+static void intel_pmu_cpu_dying(int cpu)
+{
+	fini_debug_store_on_cpu(cpu);
+}
+
+static __initconst const struct x86_pmu intel_pmu = {
 	.name			= "Intel",
 	.handle_irq		= intel_pmu_handle_irq,
 	.disable_all		= intel_pmu_disable_all,
 	.enable_all		= intel_pmu_enable_all,
 	.enable			= intel_pmu_enable_event,
 	.disable		= intel_pmu_disable_event,
+	.hw_config		= intel_pmu_hw_config,
+	.schedule_events	= x86_schedule_events,
 	.eventsel		= MSR_ARCH_PERFMON_EVENTSEL0,
 	.perfctr		= MSR_ARCH_PERFMON_PERFCTR0,
 	.event_map		= intel_pmu_event_map,
-	.raw_event		= intel_pmu_raw_event,
 	.max_events		= ARRAY_SIZE(intel_perfmon_event_map),
 	.apic			= 1,
 	/*
@@ -864,14 +842,38 @@ static __initconst struct x86_pmu intel_pmu = {
 	 * the generic event period:
 	 */
 	.max_period		= (1ULL << 31) - 1,
-	.enable_bts		= intel_pmu_enable_bts,
-	.disable_bts		= intel_pmu_disable_bts,
 	.get_event_constraints	= intel_get_event_constraints,
 
-	.cpu_starting		= init_debug_store_on_cpu,
-	.cpu_dying		= fini_debug_store_on_cpu,
+	.cpu_starting		= intel_pmu_cpu_starting,
+	.cpu_dying		= intel_pmu_cpu_dying,
 };
 
+static void intel_clovertown_quirks(void)
+{
+	/*
+	 * PEBS is unreliable due to:
+	 *
+	 *   AJ67  - PEBS may experience CPL leaks
+	 *   AJ68  - PEBS PMI may be delayed by one event
+	 *   AJ69  - GLOBAL_STATUS[62] will only be set when DEBUGCTL[12]
+	 *   AJ106 - FREEZE_LBRS_ON_PMI doesn't work in combination with PEBS
+	 *
+	 * AJ67 could be worked around by restricting the OS/USR flags.
+	 * AJ69 could be worked around by setting PMU_FREEZE_ON_PMI.
+	 *
+	 * AJ106 could possibly be worked around by not allowing LBR
+	 *       usage from PEBS, including the fixup.
+	 * AJ68  could possibly be worked around by always programming
+	 * 	 a pebs_event_reset[0] value and coping with the lost events.
+	 *
+	 * But taken together it might just make sense to not enable PEBS on
+	 * these chips.
+	 */
+	printk(KERN_WARNING "PEBS disabled due to CPU errata.\n");
+	x86_pmu.pebs = 0;
+	x86_pmu.pebs_constraints = NULL;
+}
+
 static __init int intel_pmu_init(void)
 {
 	union cpuid10_edx edx;
@@ -881,12 +883,13 @@ static __init int intel_pmu_init(void)
 	int version;
 
 	if (!cpu_has(&boot_cpu_data, X86_FEATURE_ARCH_PERFMON)) {
-		/* check for P6 processor family */
-	   if (boot_cpu_data.x86 == 6) {
-		return p6_pmu_init();
-	   } else {
+		switch (boot_cpu_data.x86) {
+		case 0x6:
+			return p6_pmu_init();
+		case 0xf:
+			return p4_pmu_init();
+		}
 		return -ENODEV;
-	   }
 	}
 
 	/*
@@ -904,16 +907,28 @@ static __init int intel_pmu_init(void)
 		x86_pmu = intel_pmu;
 
 	x86_pmu.version			= version;
-	x86_pmu.num_events		= eax.split.num_events;
-	x86_pmu.event_bits		= eax.split.bit_width;
-	x86_pmu.event_mask		= (1ULL << eax.split.bit_width) - 1;
+	x86_pmu.num_counters		= eax.split.num_counters;
+	x86_pmu.cntval_bits		= eax.split.bit_width;
+	x86_pmu.cntval_mask		= (1ULL << eax.split.bit_width) - 1;
 
 	/*
 	 * Quirk: v2 perfmon does not report fixed-purpose events, so
 	 * assume at least 3 events:
 	 */
 	if (version > 1)
-		x86_pmu.num_events_fixed = max((int)edx.split.num_events_fixed, 3);
+		x86_pmu.num_counters_fixed = max((int)edx.split.num_counters_fixed, 3);
+
+	/*
+	 * v2 and above have a perf capabilities MSR
+	 */
+	if (version > 1) {
+		u64 capabilities;
+
+		rdmsrl(MSR_IA32_PERF_CAPABILITIES, capabilities);
+		x86_pmu.intel_cap.capabilities = capabilities;
+	}
+
+	intel_ds_init();
 
 	/*
 	 * Install the hw-cache-events table:
@@ -924,12 +939,15 @@ static __init int intel_pmu_init(void)
 		break;
 
 	case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */
+		x86_pmu.quirks = intel_clovertown_quirks;
 	case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */
 	case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */
 	case 29: /* six-core 45 nm xeon "Dunnington" */
 		memcpy(hw_cache_event_ids, core2_hw_cache_event_ids,
 		       sizeof(hw_cache_event_ids));
 
+		intel_pmu_lbr_init_core();
+
 		x86_pmu.event_constraints = intel_core2_event_constraints;
 		pr_cont("Core2 events, ");
 		break;
@@ -940,13 +958,19 @@ static __init int intel_pmu_init(void)
 		memcpy(hw_cache_event_ids, nehalem_hw_cache_event_ids,
 		       sizeof(hw_cache_event_ids));
 
+		intel_pmu_lbr_init_nhm();
+
 		x86_pmu.event_constraints = intel_nehalem_event_constraints;
-		pr_cont("Nehalem/Corei7 events, ");
+		x86_pmu.enable_all = intel_pmu_nhm_enable_all;
+		pr_cont("Nehalem events, ");
 		break;
+
 	case 28: /* Atom */
 		memcpy(hw_cache_event_ids, atom_hw_cache_event_ids,
 		       sizeof(hw_cache_event_ids));
 
+		intel_pmu_lbr_init_atom();
+
 		x86_pmu.event_constraints = intel_gen_event_constraints;
 		pr_cont("Atom events, ");
 		break;
@@ -956,7 +980,10 @@ static __init int intel_pmu_init(void)
 		memcpy(hw_cache_event_ids, westmere_hw_cache_event_ids,
 		       sizeof(hw_cache_event_ids));
 
+		intel_pmu_lbr_init_nhm();
+
 		x86_pmu.event_constraints = intel_westmere_event_constraints;
+		x86_pmu.enable_all = intel_pmu_nhm_enable_all;
 		pr_cont("Westmere events, ");
 		break;
 
diff --git a/arch/x86/kernel/cpu/perf_event_intel_ds.c b/arch/x86/kernel/cpu/perf_event_intel_ds.c
new file mode 100644
index 00000000000..18018d1311c
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event_intel_ds.c
@@ -0,0 +1,641 @@
+#ifdef CONFIG_CPU_SUP_INTEL
+
+/* The maximal number of PEBS events: */
+#define MAX_PEBS_EVENTS		4
+
+/* The size of a BTS record in bytes: */
+#define BTS_RECORD_SIZE		24
+
+#define BTS_BUFFER_SIZE		(PAGE_SIZE << 4)
+#define PEBS_BUFFER_SIZE	PAGE_SIZE
+
+/*
+ * pebs_record_32 for p4 and core not supported
+
+struct pebs_record_32 {
+	u32 flags, ip;
+	u32 ax, bc, cx, dx;
+	u32 si, di, bp, sp;
+};
+
+ */
+
+struct pebs_record_core {
+	u64 flags, ip;
+	u64 ax, bx, cx, dx;
+	u64 si, di, bp, sp;
+	u64 r8,  r9,  r10, r11;
+	u64 r12, r13, r14, r15;
+};
+
+struct pebs_record_nhm {
+	u64 flags, ip;
+	u64 ax, bx, cx, dx;
+	u64 si, di, bp, sp;
+	u64 r8,  r9,  r10, r11;
+	u64 r12, r13, r14, r15;
+	u64 status, dla, dse, lat;
+};
+
+/*
+ * A debug store configuration.
+ *
+ * We only support architectures that use 64bit fields.
+ */
+struct debug_store {
+	u64	bts_buffer_base;
+	u64	bts_index;
+	u64	bts_absolute_maximum;
+	u64	bts_interrupt_threshold;
+	u64	pebs_buffer_base;
+	u64	pebs_index;
+	u64	pebs_absolute_maximum;
+	u64	pebs_interrupt_threshold;
+	u64	pebs_event_reset[MAX_PEBS_EVENTS];
+};
+
+static void init_debug_store_on_cpu(int cpu)
+{
+	struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
+
+	if (!ds)
+		return;
+
+	wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA,
+		     (u32)((u64)(unsigned long)ds),
+		     (u32)((u64)(unsigned long)ds >> 32));
+}
+
+static void fini_debug_store_on_cpu(int cpu)
+{
+	if (!per_cpu(cpu_hw_events, cpu).ds)
+		return;
+
+	wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, 0, 0);
+}
+
+static void release_ds_buffers(void)
+{
+	int cpu;
+
+	if (!x86_pmu.bts && !x86_pmu.pebs)
+		return;
+
+	get_online_cpus();
+
+	for_each_online_cpu(cpu)
+		fini_debug_store_on_cpu(cpu);
+
+	for_each_possible_cpu(cpu) {
+		struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
+
+		if (!ds)
+			continue;
+
+		per_cpu(cpu_hw_events, cpu).ds = NULL;
+
+		kfree((void *)(unsigned long)ds->pebs_buffer_base);
+		kfree((void *)(unsigned long)ds->bts_buffer_base);
+		kfree(ds);
+	}
+
+	put_online_cpus();
+}
+
+static int reserve_ds_buffers(void)
+{
+	int cpu, err = 0;
+
+	if (!x86_pmu.bts && !x86_pmu.pebs)
+		return 0;
+
+	get_online_cpus();
+
+	for_each_possible_cpu(cpu) {
+		struct debug_store *ds;
+		void *buffer;
+		int max, thresh;
+
+		err = -ENOMEM;
+		ds = kzalloc(sizeof(*ds), GFP_KERNEL);
+		if (unlikely(!ds))
+			break;
+		per_cpu(cpu_hw_events, cpu).ds = ds;
+
+		if (x86_pmu.bts) {
+			buffer = kzalloc(BTS_BUFFER_SIZE, GFP_KERNEL);
+			if (unlikely(!buffer))
+				break;
+
+			max = BTS_BUFFER_SIZE / BTS_RECORD_SIZE;
+			thresh = max / 16;
+
+			ds->bts_buffer_base = (u64)(unsigned long)buffer;
+			ds->bts_index = ds->bts_buffer_base;
+			ds->bts_absolute_maximum = ds->bts_buffer_base +
+				max * BTS_RECORD_SIZE;
+			ds->bts_interrupt_threshold = ds->bts_absolute_maximum -
+				thresh * BTS_RECORD_SIZE;
+		}
+
+		if (x86_pmu.pebs) {
+			buffer = kzalloc(PEBS_BUFFER_SIZE, GFP_KERNEL);
+			if (unlikely(!buffer))
+				break;
+
+			max = PEBS_BUFFER_SIZE / x86_pmu.pebs_record_size;
+
+			ds->pebs_buffer_base = (u64)(unsigned long)buffer;
+			ds->pebs_index = ds->pebs_buffer_base;
+			ds->pebs_absolute_maximum = ds->pebs_buffer_base +
+				max * x86_pmu.pebs_record_size;
+			/*
+			 * Always use single record PEBS
+			 */
+			ds->pebs_interrupt_threshold = ds->pebs_buffer_base +
+				x86_pmu.pebs_record_size;
+		}
+
+		err = 0;
+	}
+
+	if (err)
+		release_ds_buffers();
+	else {
+		for_each_online_cpu(cpu)
+			init_debug_store_on_cpu(cpu);
+	}
+
+	put_online_cpus();
+
+	return err;
+}
+
+/*
+ * BTS
+ */
+
+static struct event_constraint bts_constraint =
+	EVENT_CONSTRAINT(0, 1ULL << X86_PMC_IDX_FIXED_BTS, 0);
+
+static void intel_pmu_enable_bts(u64 config)
+{
+	unsigned long debugctlmsr;
+
+	debugctlmsr = get_debugctlmsr();
+
+	debugctlmsr |= DEBUGCTLMSR_TR;
+	debugctlmsr |= DEBUGCTLMSR_BTS;
+	debugctlmsr |= DEBUGCTLMSR_BTINT;
+
+	if (!(config & ARCH_PERFMON_EVENTSEL_OS))
+		debugctlmsr |= DEBUGCTLMSR_BTS_OFF_OS;
+
+	if (!(config & ARCH_PERFMON_EVENTSEL_USR))
+		debugctlmsr |= DEBUGCTLMSR_BTS_OFF_USR;
+
+	update_debugctlmsr(debugctlmsr);
+}
+
+static void intel_pmu_disable_bts(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	unsigned long debugctlmsr;
+
+	if (!cpuc->ds)
+		return;
+
+	debugctlmsr = get_debugctlmsr();
+
+	debugctlmsr &=
+		~(DEBUGCTLMSR_TR | DEBUGCTLMSR_BTS | DEBUGCTLMSR_BTINT |
+		  DEBUGCTLMSR_BTS_OFF_OS | DEBUGCTLMSR_BTS_OFF_USR);
+
+	update_debugctlmsr(debugctlmsr);
+}
+
+static void intel_pmu_drain_bts_buffer(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct debug_store *ds = cpuc->ds;
+	struct bts_record {
+		u64	from;
+		u64	to;
+		u64	flags;
+	};
+	struct perf_event *event = cpuc->events[X86_PMC_IDX_FIXED_BTS];
+	struct bts_record *at, *top;
+	struct perf_output_handle handle;
+	struct perf_event_header header;
+	struct perf_sample_data data;
+	struct pt_regs regs;
+
+	if (!event)
+		return;
+
+	if (!ds)
+		return;
+
+	at  = (struct bts_record *)(unsigned long)ds->bts_buffer_base;
+	top = (struct bts_record *)(unsigned long)ds->bts_index;
+
+	if (top <= at)
+		return;
+
+	ds->bts_index = ds->bts_buffer_base;
+
+	perf_sample_data_init(&data, 0);
+	data.period = event->hw.last_period;
+	regs.ip     = 0;
+
+	/*
+	 * Prepare a generic sample, i.e. fill in the invariant fields.
+	 * We will overwrite the from and to address before we output
+	 * the sample.
+	 */
+	perf_prepare_sample(&header, &data, event, &regs);
+
+	if (perf_output_begin(&handle, event, header.size * (top - at), 1, 1))
+		return;
+
+	for (; at < top; at++) {
+		data.ip		= at->from;
+		data.addr	= at->to;
+
+		perf_output_sample(&handle, &header, &data, event);
+	}
+
+	perf_output_end(&handle);
+
+	/* There's new data available. */
+	event->hw.interrupts++;
+	event->pending_kill = POLL_IN;
+}
+
+/*
+ * PEBS
+ */
+
+static struct event_constraint intel_core_pebs_events[] = {
+	PEBS_EVENT_CONSTRAINT(0x00c0, 0x1), /* INSTR_RETIRED.ANY */
+	PEBS_EVENT_CONSTRAINT(0xfec1, 0x1), /* X87_OPS_RETIRED.ANY */
+	PEBS_EVENT_CONSTRAINT(0x00c5, 0x1), /* BR_INST_RETIRED.MISPRED */
+	PEBS_EVENT_CONSTRAINT(0x1fc7, 0x1), /* SIMD_INST_RETURED.ANY */
+	PEBS_EVENT_CONSTRAINT(0x01cb, 0x1), /* MEM_LOAD_RETIRED.L1D_MISS */
+	PEBS_EVENT_CONSTRAINT(0x02cb, 0x1), /* MEM_LOAD_RETIRED.L1D_LINE_MISS */
+	PEBS_EVENT_CONSTRAINT(0x04cb, 0x1), /* MEM_LOAD_RETIRED.L2_MISS */
+	PEBS_EVENT_CONSTRAINT(0x08cb, 0x1), /* MEM_LOAD_RETIRED.L2_LINE_MISS */
+	PEBS_EVENT_CONSTRAINT(0x10cb, 0x1), /* MEM_LOAD_RETIRED.DTLB_MISS */
+	EVENT_CONSTRAINT_END
+};
+
+static struct event_constraint intel_nehalem_pebs_events[] = {
+	PEBS_EVENT_CONSTRAINT(0x00c0, 0xf), /* INSTR_RETIRED.ANY */
+	PEBS_EVENT_CONSTRAINT(0xfec1, 0xf), /* X87_OPS_RETIRED.ANY */
+	PEBS_EVENT_CONSTRAINT(0x00c5, 0xf), /* BR_INST_RETIRED.MISPRED */
+	PEBS_EVENT_CONSTRAINT(0x1fc7, 0xf), /* SIMD_INST_RETURED.ANY */
+	PEBS_EVENT_CONSTRAINT(0x01cb, 0xf), /* MEM_LOAD_RETIRED.L1D_MISS */
+	PEBS_EVENT_CONSTRAINT(0x02cb, 0xf), /* MEM_LOAD_RETIRED.L1D_LINE_MISS */
+	PEBS_EVENT_CONSTRAINT(0x04cb, 0xf), /* MEM_LOAD_RETIRED.L2_MISS */
+	PEBS_EVENT_CONSTRAINT(0x08cb, 0xf), /* MEM_LOAD_RETIRED.L2_LINE_MISS */
+	PEBS_EVENT_CONSTRAINT(0x10cb, 0xf), /* MEM_LOAD_RETIRED.DTLB_MISS */
+	EVENT_CONSTRAINT_END
+};
+
+static struct event_constraint *
+intel_pebs_constraints(struct perf_event *event)
+{
+	struct event_constraint *c;
+
+	if (!event->attr.precise_ip)
+		return NULL;
+
+	if (x86_pmu.pebs_constraints) {
+		for_each_event_constraint(c, x86_pmu.pebs_constraints) {
+			if ((event->hw.config & c->cmask) == c->code)
+				return c;
+		}
+	}
+
+	return &emptyconstraint;
+}
+
+static void intel_pmu_pebs_enable(struct perf_event *event)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct hw_perf_event *hwc = &event->hw;
+
+	hwc->config &= ~ARCH_PERFMON_EVENTSEL_INT;
+
+	cpuc->pebs_enabled |= 1ULL << hwc->idx;
+	WARN_ON_ONCE(cpuc->enabled);
+
+	if (x86_pmu.intel_cap.pebs_trap && event->attr.precise_ip > 1)
+		intel_pmu_lbr_enable(event);
+}
+
+static void intel_pmu_pebs_disable(struct perf_event *event)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct hw_perf_event *hwc = &event->hw;
+
+	cpuc->pebs_enabled &= ~(1ULL << hwc->idx);
+	if (cpuc->enabled)
+		wrmsrl(MSR_IA32_PEBS_ENABLE, cpuc->pebs_enabled);
+
+	hwc->config |= ARCH_PERFMON_EVENTSEL_INT;
+
+	if (x86_pmu.intel_cap.pebs_trap && event->attr.precise_ip > 1)
+		intel_pmu_lbr_disable(event);
+}
+
+static void intel_pmu_pebs_enable_all(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	if (cpuc->pebs_enabled)
+		wrmsrl(MSR_IA32_PEBS_ENABLE, cpuc->pebs_enabled);
+}
+
+static void intel_pmu_pebs_disable_all(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	if (cpuc->pebs_enabled)
+		wrmsrl(MSR_IA32_PEBS_ENABLE, 0);
+}
+
+#include <asm/insn.h>
+
+static inline bool kernel_ip(unsigned long ip)
+{
+#ifdef CONFIG_X86_32
+	return ip > PAGE_OFFSET;
+#else
+	return (long)ip < 0;
+#endif
+}
+
+static int intel_pmu_pebs_fixup_ip(struct pt_regs *regs)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	unsigned long from = cpuc->lbr_entries[0].from;
+	unsigned long old_to, to = cpuc->lbr_entries[0].to;
+	unsigned long ip = regs->ip;
+
+	/*
+	 * We don't need to fixup if the PEBS assist is fault like
+	 */
+	if (!x86_pmu.intel_cap.pebs_trap)
+		return 1;
+
+	/*
+	 * No LBR entry, no basic block, no rewinding
+	 */
+	if (!cpuc->lbr_stack.nr || !from || !to)
+		return 0;
+
+	/*
+	 * Basic blocks should never cross user/kernel boundaries
+	 */
+	if (kernel_ip(ip) != kernel_ip(to))
+		return 0;
+
+	/*
+	 * unsigned math, either ip is before the start (impossible) or
+	 * the basic block is larger than 1 page (sanity)
+	 */
+	if ((ip - to) > PAGE_SIZE)
+		return 0;
+
+	/*
+	 * We sampled a branch insn, rewind using the LBR stack
+	 */
+	if (ip == to) {
+		regs->ip = from;
+		return 1;
+	}
+
+	do {
+		struct insn insn;
+		u8 buf[MAX_INSN_SIZE];
+		void *kaddr;
+
+		old_to = to;
+		if (!kernel_ip(ip)) {
+			int bytes, size = MAX_INSN_SIZE;
+
+			bytes = copy_from_user_nmi(buf, (void __user *)to, size);
+			if (bytes != size)
+				return 0;
+
+			kaddr = buf;
+		} else
+			kaddr = (void *)to;
+
+		kernel_insn_init(&insn, kaddr);
+		insn_get_length(&insn);
+		to += insn.length;
+	} while (to < ip);
+
+	if (to == ip) {
+		regs->ip = old_to;
+		return 1;
+	}
+
+	/*
+	 * Even though we decoded the basic block, the instruction stream
+	 * never matched the given IP, either the TO or the IP got corrupted.
+	 */
+	return 0;
+}
+
+static int intel_pmu_save_and_restart(struct perf_event *event);
+
+static void __intel_pmu_pebs_event(struct perf_event *event,
+				   struct pt_regs *iregs, void *__pebs)
+{
+	/*
+	 * We cast to pebs_record_core since that is a subset of
+	 * both formats and we don't use the other fields in this
+	 * routine.
+	 */
+	struct pebs_record_core *pebs = __pebs;
+	struct perf_sample_data data;
+	struct pt_regs regs;
+
+	if (!intel_pmu_save_and_restart(event))
+		return;
+
+	perf_sample_data_init(&data, 0);
+	data.period = event->hw.last_period;
+
+	/*
+	 * We use the interrupt regs as a base because the PEBS record
+	 * does not contain a full regs set, specifically it seems to
+	 * lack segment descriptors, which get used by things like
+	 * user_mode().
+	 *
+	 * In the simple case fix up only the IP and BP,SP regs, for
+	 * PERF_SAMPLE_IP and PERF_SAMPLE_CALLCHAIN to function properly.
+	 * A possible PERF_SAMPLE_REGS will have to transfer all regs.
+	 */
+	regs = *iregs;
+	regs.ip = pebs->ip;
+	regs.bp = pebs->bp;
+	regs.sp = pebs->sp;
+
+	if (event->attr.precise_ip > 1 && intel_pmu_pebs_fixup_ip(&regs))
+		regs.flags |= PERF_EFLAGS_EXACT;
+	else
+		regs.flags &= ~PERF_EFLAGS_EXACT;
+
+	if (perf_event_overflow(event, 1, &data, &regs))
+		x86_pmu_stop(event);
+}
+
+static void intel_pmu_drain_pebs_core(struct pt_regs *iregs)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct debug_store *ds = cpuc->ds;
+	struct perf_event *event = cpuc->events[0]; /* PMC0 only */
+	struct pebs_record_core *at, *top;
+	int n;
+
+	if (!ds || !x86_pmu.pebs)
+		return;
+
+	at  = (struct pebs_record_core *)(unsigned long)ds->pebs_buffer_base;
+	top = (struct pebs_record_core *)(unsigned long)ds->pebs_index;
+
+	/*
+	 * Whatever else happens, drain the thing
+	 */
+	ds->pebs_index = ds->pebs_buffer_base;
+
+	if (!test_bit(0, cpuc->active_mask))
+		return;
+
+	WARN_ON_ONCE(!event);
+
+	if (!event->attr.precise_ip)
+		return;
+
+	n = top - at;
+	if (n <= 0)
+		return;
+
+	/*
+	 * Should not happen, we program the threshold at 1 and do not
+	 * set a reset value.
+	 */
+	WARN_ON_ONCE(n > 1);
+	at += n - 1;
+
+	__intel_pmu_pebs_event(event, iregs, at);
+}
+
+static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct debug_store *ds = cpuc->ds;
+	struct pebs_record_nhm *at, *top;
+	struct perf_event *event = NULL;
+	u64 status = 0;
+	int bit, n;
+
+	if (!ds || !x86_pmu.pebs)
+		return;
+
+	at  = (struct pebs_record_nhm *)(unsigned long)ds->pebs_buffer_base;
+	top = (struct pebs_record_nhm *)(unsigned long)ds->pebs_index;
+
+	ds->pebs_index = ds->pebs_buffer_base;
+
+	n = top - at;
+	if (n <= 0)
+		return;
+
+	/*
+	 * Should not happen, we program the threshold at 1 and do not
+	 * set a reset value.
+	 */
+	WARN_ON_ONCE(n > MAX_PEBS_EVENTS);
+
+	for ( ; at < top; at++) {
+		for_each_set_bit(bit, (unsigned long *)&at->status, MAX_PEBS_EVENTS) {
+			event = cpuc->events[bit];
+			if (!test_bit(bit, cpuc->active_mask))
+				continue;
+
+			WARN_ON_ONCE(!event);
+
+			if (!event->attr.precise_ip)
+				continue;
+
+			if (__test_and_set_bit(bit, (unsigned long *)&status))
+				continue;
+
+			break;
+		}
+
+		if (!event || bit >= MAX_PEBS_EVENTS)
+			continue;
+
+		__intel_pmu_pebs_event(event, iregs, at);
+	}
+}
+
+/*
+ * BTS, PEBS probe and setup
+ */
+
+static void intel_ds_init(void)
+{
+	/*
+	 * No support for 32bit formats
+	 */
+	if (!boot_cpu_has(X86_FEATURE_DTES64))
+		return;
+
+	x86_pmu.bts  = boot_cpu_has(X86_FEATURE_BTS);
+	x86_pmu.pebs = boot_cpu_has(X86_FEATURE_PEBS);
+	if (x86_pmu.pebs) {
+		char pebs_type = x86_pmu.intel_cap.pebs_trap ?  '+' : '-';
+		int format = x86_pmu.intel_cap.pebs_format;
+
+		switch (format) {
+		case 0:
+			printk(KERN_CONT "PEBS fmt0%c, ", pebs_type);
+			x86_pmu.pebs_record_size = sizeof(struct pebs_record_core);
+			x86_pmu.drain_pebs = intel_pmu_drain_pebs_core;
+			x86_pmu.pebs_constraints = intel_core_pebs_events;
+			break;
+
+		case 1:
+			printk(KERN_CONT "PEBS fmt1%c, ", pebs_type);
+			x86_pmu.pebs_record_size = sizeof(struct pebs_record_nhm);
+			x86_pmu.drain_pebs = intel_pmu_drain_pebs_nhm;
+			x86_pmu.pebs_constraints = intel_nehalem_pebs_events;
+			break;
+
+		default:
+			printk(KERN_CONT "no PEBS fmt%d%c, ", format, pebs_type);
+			x86_pmu.pebs = 0;
+			break;
+		}
+	}
+}
+
+#else /* CONFIG_CPU_SUP_INTEL */
+
+static int reserve_ds_buffers(void)
+{
+	return 0;
+}
+
+static void release_ds_buffers(void)
+{
+}
+
+#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_intel_lbr.c b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
new file mode 100644
index 00000000000..d202c1bece1
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
@@ -0,0 +1,218 @@
+#ifdef CONFIG_CPU_SUP_INTEL
+
+enum {
+	LBR_FORMAT_32		= 0x00,
+	LBR_FORMAT_LIP		= 0x01,
+	LBR_FORMAT_EIP		= 0x02,
+	LBR_FORMAT_EIP_FLAGS	= 0x03,
+};
+
+/*
+ * We only support LBR implementations that have FREEZE_LBRS_ON_PMI
+ * otherwise it becomes near impossible to get a reliable stack.
+ */
+
+static void __intel_pmu_lbr_enable(void)
+{
+	u64 debugctl;
+
+	rdmsrl(MSR_IA32_DEBUGCTLMSR, debugctl);
+	debugctl |= (DEBUGCTLMSR_LBR | DEBUGCTLMSR_FREEZE_LBRS_ON_PMI);
+	wrmsrl(MSR_IA32_DEBUGCTLMSR, debugctl);
+}
+
+static void __intel_pmu_lbr_disable(void)
+{
+	u64 debugctl;
+
+	rdmsrl(MSR_IA32_DEBUGCTLMSR, debugctl);
+	debugctl &= ~(DEBUGCTLMSR_LBR | DEBUGCTLMSR_FREEZE_LBRS_ON_PMI);
+	wrmsrl(MSR_IA32_DEBUGCTLMSR, debugctl);
+}
+
+static void intel_pmu_lbr_reset_32(void)
+{
+	int i;
+
+	for (i = 0; i < x86_pmu.lbr_nr; i++)
+		wrmsrl(x86_pmu.lbr_from + i, 0);
+}
+
+static void intel_pmu_lbr_reset_64(void)
+{
+	int i;
+
+	for (i = 0; i < x86_pmu.lbr_nr; i++) {
+		wrmsrl(x86_pmu.lbr_from + i, 0);
+		wrmsrl(x86_pmu.lbr_to   + i, 0);
+	}
+}
+
+static void intel_pmu_lbr_reset(void)
+{
+	if (!x86_pmu.lbr_nr)
+		return;
+
+	if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_32)
+		intel_pmu_lbr_reset_32();
+	else
+		intel_pmu_lbr_reset_64();
+}
+
+static void intel_pmu_lbr_enable(struct perf_event *event)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	if (!x86_pmu.lbr_nr)
+		return;
+
+	WARN_ON_ONCE(cpuc->enabled);
+
+	/*
+	 * Reset the LBR stack if we changed task context to
+	 * avoid data leaks.
+	 */
+
+	if (event->ctx->task && cpuc->lbr_context != event->ctx) {
+		intel_pmu_lbr_reset();
+		cpuc->lbr_context = event->ctx;
+	}
+
+	cpuc->lbr_users++;
+}
+
+static void intel_pmu_lbr_disable(struct perf_event *event)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	if (!x86_pmu.lbr_nr)
+		return;
+
+	cpuc->lbr_users--;
+	WARN_ON_ONCE(cpuc->lbr_users < 0);
+
+	if (cpuc->enabled && !cpuc->lbr_users)
+		__intel_pmu_lbr_disable();
+}
+
+static void intel_pmu_lbr_enable_all(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	if (cpuc->lbr_users)
+		__intel_pmu_lbr_enable();
+}
+
+static void intel_pmu_lbr_disable_all(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	if (cpuc->lbr_users)
+		__intel_pmu_lbr_disable();
+}
+
+static inline u64 intel_pmu_lbr_tos(void)
+{
+	u64 tos;
+
+	rdmsrl(x86_pmu.lbr_tos, tos);
+
+	return tos;
+}
+
+static void intel_pmu_lbr_read_32(struct cpu_hw_events *cpuc)
+{
+	unsigned long mask = x86_pmu.lbr_nr - 1;
+	u64 tos = intel_pmu_lbr_tos();
+	int i;
+
+	for (i = 0; i < x86_pmu.lbr_nr; i++) {
+		unsigned long lbr_idx = (tos - i) & mask;
+		union {
+			struct {
+				u32 from;
+				u32 to;
+			};
+			u64     lbr;
+		} msr_lastbranch;
+
+		rdmsrl(x86_pmu.lbr_from + lbr_idx, msr_lastbranch.lbr);
+
+		cpuc->lbr_entries[i].from  = msr_lastbranch.from;
+		cpuc->lbr_entries[i].to    = msr_lastbranch.to;
+		cpuc->lbr_entries[i].flags = 0;
+	}
+	cpuc->lbr_stack.nr = i;
+}
+
+#define LBR_FROM_FLAG_MISPRED  (1ULL << 63)
+
+/*
+ * Due to lack of segmentation in Linux the effective address (offset)
+ * is the same as the linear address, allowing us to merge the LIP and EIP
+ * LBR formats.
+ */
+static void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc)
+{
+	unsigned long mask = x86_pmu.lbr_nr - 1;
+	int lbr_format = x86_pmu.intel_cap.lbr_format;
+	u64 tos = intel_pmu_lbr_tos();
+	int i;
+
+	for (i = 0; i < x86_pmu.lbr_nr; i++) {
+		unsigned long lbr_idx = (tos - i) & mask;
+		u64 from, to, flags = 0;
+
+		rdmsrl(x86_pmu.lbr_from + lbr_idx, from);
+		rdmsrl(x86_pmu.lbr_to   + lbr_idx, to);
+
+		if (lbr_format == LBR_FORMAT_EIP_FLAGS) {
+			flags = !!(from & LBR_FROM_FLAG_MISPRED);
+			from = (u64)((((s64)from) << 1) >> 1);
+		}
+
+		cpuc->lbr_entries[i].from  = from;
+		cpuc->lbr_entries[i].to    = to;
+		cpuc->lbr_entries[i].flags = flags;
+	}
+	cpuc->lbr_stack.nr = i;
+}
+
+static void intel_pmu_lbr_read(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	if (!cpuc->lbr_users)
+		return;
+
+	if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_32)
+		intel_pmu_lbr_read_32(cpuc);
+	else
+		intel_pmu_lbr_read_64(cpuc);
+}
+
+static void intel_pmu_lbr_init_core(void)
+{
+	x86_pmu.lbr_nr     = 4;
+	x86_pmu.lbr_tos    = 0x01c9;
+	x86_pmu.lbr_from   = 0x40;
+	x86_pmu.lbr_to     = 0x60;
+}
+
+static void intel_pmu_lbr_init_nhm(void)
+{
+	x86_pmu.lbr_nr     = 16;
+	x86_pmu.lbr_tos    = 0x01c9;
+	x86_pmu.lbr_from   = 0x680;
+	x86_pmu.lbr_to     = 0x6c0;
+}
+
+static void intel_pmu_lbr_init_atom(void)
+{
+	x86_pmu.lbr_nr	   = 8;
+	x86_pmu.lbr_tos    = 0x01c9;
+	x86_pmu.lbr_from   = 0x40;
+	x86_pmu.lbr_to     = 0x60;
+}
+
+#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_p4.c b/arch/x86/kernel/cpu/perf_event_p4.c
new file mode 100644
index 00000000000..ae85d69644d
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event_p4.c
@@ -0,0 +1,858 @@
+/*
+ * Netburst Perfomance Events (P4, old Xeon)
+ *
+ *  Copyright (C) 2010 Parallels, Inc., Cyrill Gorcunov <gorcunov@openvz.org>
+ *  Copyright (C) 2010 Intel Corporation, Lin Ming <ming.m.lin@intel.com>
+ *
+ *  For licencing details see kernel-base/COPYING
+ */
+
+#ifdef CONFIG_CPU_SUP_INTEL
+
+#include <asm/perf_event_p4.h>
+
+#define P4_CNTR_LIMIT 3
+/*
+ * array indices: 0,1 - HT threads, used with HT enabled cpu
+ */
+struct p4_event_bind {
+	unsigned int opcode;			/* Event code and ESCR selector */
+	unsigned int escr_msr[2];		/* ESCR MSR for this event */
+	char cntr[2][P4_CNTR_LIMIT];		/* counter index (offset), -1 on abscence */
+};
+
+struct p4_cache_event_bind {
+	unsigned int metric_pebs;
+	unsigned int metric_vert;
+};
+
+#define P4_GEN_CACHE_EVENT_BIND(name)		\
+	[P4_CACHE__##name] = {			\
+		.metric_pebs = P4_PEBS__##name,	\
+		.metric_vert = P4_VERT__##name,	\
+	}
+
+static struct p4_cache_event_bind p4_cache_event_bind_map[] = {
+	P4_GEN_CACHE_EVENT_BIND(1stl_cache_load_miss_retired),
+	P4_GEN_CACHE_EVENT_BIND(2ndl_cache_load_miss_retired),
+	P4_GEN_CACHE_EVENT_BIND(dtlb_load_miss_retired),
+	P4_GEN_CACHE_EVENT_BIND(dtlb_store_miss_retired),
+};
+
+/*
+ * Note that we don't use CCCR1 here, there is an
+ * exception for P4_BSQ_ALLOCATION but we just have
+ * no workaround
+ *
+ * consider this binding as resources which particular
+ * event may borrow, it doesn't contain EventMask,
+ * Tags and friends -- they are left to a caller
+ */
+static struct p4_event_bind p4_event_bind_map[] = {
+	[P4_EVENT_TC_DELIVER_MODE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_TC_DELIVER_MODE),
+		.escr_msr	= { MSR_P4_TC_ESCR0, MSR_P4_TC_ESCR1 },
+		.cntr		= { {4, 5, -1}, {6, 7, -1} },
+	},
+	[P4_EVENT_BPU_FETCH_REQUEST] = {
+		.opcode		= P4_OPCODE(P4_EVENT_BPU_FETCH_REQUEST),
+		.escr_msr	= { MSR_P4_BPU_ESCR0, MSR_P4_BPU_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_ITLB_REFERENCE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_ITLB_REFERENCE),
+		.escr_msr	= { MSR_P4_ITLB_ESCR0, MSR_P4_ITLB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_MEMORY_CANCEL] = {
+		.opcode		= P4_OPCODE(P4_EVENT_MEMORY_CANCEL),
+		.escr_msr	= { MSR_P4_DAC_ESCR0, MSR_P4_DAC_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_MEMORY_COMPLETE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_MEMORY_COMPLETE),
+		.escr_msr	= { MSR_P4_SAAT_ESCR0 , MSR_P4_SAAT_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_LOAD_PORT_REPLAY] = {
+		.opcode		= P4_OPCODE(P4_EVENT_LOAD_PORT_REPLAY),
+		.escr_msr	= { MSR_P4_SAAT_ESCR0, MSR_P4_SAAT_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_STORE_PORT_REPLAY] = {
+		.opcode		= P4_OPCODE(P4_EVENT_STORE_PORT_REPLAY),
+		.escr_msr	= { MSR_P4_SAAT_ESCR0 ,  MSR_P4_SAAT_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_MOB_LOAD_REPLAY] = {
+		.opcode		= P4_OPCODE(P4_EVENT_MOB_LOAD_REPLAY),
+		.escr_msr	= { MSR_P4_MOB_ESCR0, MSR_P4_MOB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_PAGE_WALK_TYPE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_PAGE_WALK_TYPE),
+		.escr_msr	= { MSR_P4_PMH_ESCR0, MSR_P4_PMH_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_BSQ_CACHE_REFERENCE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_BSQ_CACHE_REFERENCE),
+		.escr_msr	= { MSR_P4_BSU_ESCR0, MSR_P4_BSU_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_IOQ_ALLOCATION] = {
+		.opcode		= P4_OPCODE(P4_EVENT_IOQ_ALLOCATION),
+		.escr_msr	= { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_IOQ_ACTIVE_ENTRIES] = {	/* shared ESCR */
+		.opcode		= P4_OPCODE(P4_EVENT_IOQ_ACTIVE_ENTRIES),
+		.escr_msr	= { MSR_P4_FSB_ESCR1,  MSR_P4_FSB_ESCR1 },
+		.cntr		= { {2, -1, -1}, {3, -1, -1} },
+	},
+	[P4_EVENT_FSB_DATA_ACTIVITY] = {
+		.opcode		= P4_OPCODE(P4_EVENT_FSB_DATA_ACTIVITY),
+		.escr_msr	= { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_BSQ_ALLOCATION] = {		/* shared ESCR, broken CCCR1 */
+		.opcode		= P4_OPCODE(P4_EVENT_BSQ_ALLOCATION),
+		.escr_msr	= { MSR_P4_BSU_ESCR0, MSR_P4_BSU_ESCR0 },
+		.cntr		= { {0, -1, -1}, {1, -1, -1} },
+	},
+	[P4_EVENT_BSQ_ACTIVE_ENTRIES] = {	/* shared ESCR */
+		.opcode		= P4_OPCODE(P4_EVENT_BSQ_ACTIVE_ENTRIES),
+		.escr_msr	= { MSR_P4_BSU_ESCR1 , MSR_P4_BSU_ESCR1 },
+		.cntr		= { {2, -1, -1}, {3, -1, -1} },
+	},
+	[P4_EVENT_SSE_INPUT_ASSIST] = {
+		.opcode		= P4_OPCODE(P4_EVENT_SSE_INPUT_ASSIST),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_PACKED_SP_UOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_PACKED_SP_UOP),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_PACKED_DP_UOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_PACKED_DP_UOP),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_SCALAR_SP_UOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_SCALAR_SP_UOP),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_SCALAR_DP_UOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_SCALAR_DP_UOP),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_64BIT_MMX_UOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_64BIT_MMX_UOP),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_128BIT_MMX_UOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_128BIT_MMX_UOP),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_X87_FP_UOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_X87_FP_UOP),
+		.escr_msr	= { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_TC_MISC] = {
+		.opcode		= P4_OPCODE(P4_EVENT_TC_MISC),
+		.escr_msr	= { MSR_P4_TC_ESCR0, MSR_P4_TC_ESCR1 },
+		.cntr		= { {4, 5, -1}, {6, 7, -1} },
+	},
+	[P4_EVENT_GLOBAL_POWER_EVENTS] = {
+		.opcode		= P4_OPCODE(P4_EVENT_GLOBAL_POWER_EVENTS),
+		.escr_msr	= { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_TC_MS_XFER] = {
+		.opcode		= P4_OPCODE(P4_EVENT_TC_MS_XFER),
+		.escr_msr	= { MSR_P4_MS_ESCR0, MSR_P4_MS_ESCR1 },
+		.cntr		= { {4, 5, -1}, {6, 7, -1} },
+	},
+	[P4_EVENT_UOP_QUEUE_WRITES] = {
+		.opcode		= P4_OPCODE(P4_EVENT_UOP_QUEUE_WRITES),
+		.escr_msr	= { MSR_P4_MS_ESCR0, MSR_P4_MS_ESCR1 },
+		.cntr		= { {4, 5, -1}, {6, 7, -1} },
+	},
+	[P4_EVENT_RETIRED_MISPRED_BRANCH_TYPE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_RETIRED_MISPRED_BRANCH_TYPE),
+		.escr_msr	= { MSR_P4_TBPU_ESCR0 , MSR_P4_TBPU_ESCR0 },
+		.cntr		= { {4, 5, -1}, {6, 7, -1} },
+	},
+	[P4_EVENT_RETIRED_BRANCH_TYPE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_RETIRED_BRANCH_TYPE),
+		.escr_msr	= { MSR_P4_TBPU_ESCR0 , MSR_P4_TBPU_ESCR1 },
+		.cntr		= { {4, 5, -1}, {6, 7, -1} },
+	},
+	[P4_EVENT_RESOURCE_STALL] = {
+		.opcode		= P4_OPCODE(P4_EVENT_RESOURCE_STALL),
+		.escr_msr	= { MSR_P4_ALF_ESCR0, MSR_P4_ALF_ESCR1 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_WC_BUFFER] = {
+		.opcode		= P4_OPCODE(P4_EVENT_WC_BUFFER),
+		.escr_msr	= { MSR_P4_DAC_ESCR0, MSR_P4_DAC_ESCR1 },
+		.cntr		= { {8, 9, -1}, {10, 11, -1} },
+	},
+	[P4_EVENT_B2B_CYCLES] = {
+		.opcode		= P4_OPCODE(P4_EVENT_B2B_CYCLES),
+		.escr_msr	= { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_BNR] = {
+		.opcode		= P4_OPCODE(P4_EVENT_BNR),
+		.escr_msr	= { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_SNOOP] = {
+		.opcode		= P4_OPCODE(P4_EVENT_SNOOP),
+		.escr_msr	= { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_RESPONSE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_RESPONSE),
+		.escr_msr	= { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 },
+		.cntr		= { {0, -1, -1}, {2, -1, -1} },
+	},
+	[P4_EVENT_FRONT_END_EVENT] = {
+		.opcode		= P4_OPCODE(P4_EVENT_FRONT_END_EVENT),
+		.escr_msr	= { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_EXECUTION_EVENT] = {
+		.opcode		= P4_OPCODE(P4_EVENT_EXECUTION_EVENT),
+		.escr_msr	= { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_REPLAY_EVENT] = {
+		.opcode		= P4_OPCODE(P4_EVENT_REPLAY_EVENT),
+		.escr_msr	= { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_INSTR_RETIRED] = {
+		.opcode		= P4_OPCODE(P4_EVENT_INSTR_RETIRED),
+		.escr_msr	= { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_UOPS_RETIRED] = {
+		.opcode		= P4_OPCODE(P4_EVENT_UOPS_RETIRED),
+		.escr_msr	= { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_UOP_TYPE] = {
+		.opcode		= P4_OPCODE(P4_EVENT_UOP_TYPE),
+		.escr_msr	= { MSR_P4_RAT_ESCR0, MSR_P4_RAT_ESCR1 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_BRANCH_RETIRED] = {
+		.opcode		= P4_OPCODE(P4_EVENT_BRANCH_RETIRED),
+		.escr_msr	= { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_MISPRED_BRANCH_RETIRED] = {
+		.opcode		= P4_OPCODE(P4_EVENT_MISPRED_BRANCH_RETIRED),
+		.escr_msr	= { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_X87_ASSIST] = {
+		.opcode		= P4_OPCODE(P4_EVENT_X87_ASSIST),
+		.escr_msr	= { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_MACHINE_CLEAR] = {
+		.opcode		= P4_OPCODE(P4_EVENT_MACHINE_CLEAR),
+		.escr_msr	= { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+	[P4_EVENT_INSTR_COMPLETED] = {
+		.opcode		= P4_OPCODE(P4_EVENT_INSTR_COMPLETED),
+		.escr_msr	= { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 },
+		.cntr		= { {12, 13, 16}, {14, 15, 17} },
+	},
+};
+
+#define P4_GEN_CACHE_EVENT(event, bit, cache_event)			  \
+	p4_config_pack_escr(P4_ESCR_EVENT(event)			| \
+			    P4_ESCR_EMASK_BIT(event, bit))		| \
+	p4_config_pack_cccr(cache_event					| \
+			    P4_CCCR_ESEL(P4_OPCODE_ESEL(P4_OPCODE(event))))
+
+static __initconst const u64 p4_hw_cache_event_ids
+				[PERF_COUNT_HW_CACHE_MAX]
+				[PERF_COUNT_HW_CACHE_OP_MAX]
+				[PERF_COUNT_HW_CACHE_RESULT_MAX] =
+{
+ [ C(L1D ) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS,
+						P4_CACHE__1stl_cache_load_miss_retired),
+	},
+ },
+ [ C(LL  ) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS,
+						P4_CACHE__2ndl_cache_load_miss_retired),
+	},
+},
+ [ C(DTLB) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS,
+						P4_CACHE__dtlb_load_miss_retired),
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS,
+						P4_CACHE__dtlb_store_miss_retired),
+	},
+ },
+ [ C(ITLB) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = P4_GEN_CACHE_EVENT(P4_EVENT_ITLB_REFERENCE, HIT,
+						P4_CACHE__itlb_reference_hit),
+		[ C(RESULT_MISS)   ] = P4_GEN_CACHE_EVENT(P4_EVENT_ITLB_REFERENCE, MISS,
+						P4_CACHE__itlb_reference_miss),
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = -1,
+		[ C(RESULT_MISS)   ] = -1,
+	},
+	[ C(OP_PREFETCH) ] = {
+		[ C(RESULT_ACCESS) ] = -1,
+		[ C(RESULT_MISS)   ] = -1,
+	},
+ },
+};
+
+static u64 p4_general_events[PERF_COUNT_HW_MAX] = {
+  /* non-halted CPU clocks */
+  [PERF_COUNT_HW_CPU_CYCLES] =
+	p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_GLOBAL_POWER_EVENTS)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_GLOBAL_POWER_EVENTS, RUNNING)),
+
+  /*
+   * retired instructions
+   * in a sake of simplicity we don't use the FSB tagging
+   */
+  [PERF_COUNT_HW_INSTRUCTIONS] =
+	p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_INSTR_RETIRED)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_INSTR_RETIRED, NBOGUSNTAG)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_INSTR_RETIRED, BOGUSNTAG)),
+
+  /* cache hits */
+  [PERF_COUNT_HW_CACHE_REFERENCES] =
+	p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_BSQ_CACHE_REFERENCE)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_HITS)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_HITE)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_HITM)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_HITS)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_HITE)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_HITM)),
+
+  /* cache misses */
+  [PERF_COUNT_HW_CACHE_MISSES] =
+	p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_BSQ_CACHE_REFERENCE)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_MISS)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_MISS)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, WR_2ndL_MISS)),
+
+  /* branch instructions retired */
+  [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] =
+	p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_RETIRED_BRANCH_TYPE)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, CONDITIONAL)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, CALL)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, RETURN)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, INDIRECT)),
+
+  /* mispredicted branches retired */
+  [PERF_COUNT_HW_BRANCH_MISSES]	=
+	p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_MISPRED_BRANCH_RETIRED)	|
+		P4_ESCR_EMASK_BIT(P4_EVENT_MISPRED_BRANCH_RETIRED, NBOGUS)),
+
+  /* bus ready clocks (cpu is driving #DRDY_DRV\#DRDY_OWN):  */
+  [PERF_COUNT_HW_BUS_CYCLES] =
+	p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_FSB_DATA_ACTIVITY)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_FSB_DATA_ACTIVITY, DRDY_DRV)		|
+		P4_ESCR_EMASK_BIT(P4_EVENT_FSB_DATA_ACTIVITY, DRDY_OWN))	|
+	p4_config_pack_cccr(P4_CCCR_EDGE | P4_CCCR_COMPARE),
+};
+
+static struct p4_event_bind *p4_config_get_bind(u64 config)
+{
+	unsigned int evnt = p4_config_unpack_event(config);
+	struct p4_event_bind *bind = NULL;
+
+	if (evnt < ARRAY_SIZE(p4_event_bind_map))
+		bind = &p4_event_bind_map[evnt];
+
+	return bind;
+}
+
+static u64 p4_pmu_event_map(int hw_event)
+{
+	struct p4_event_bind *bind;
+	unsigned int esel;
+	u64 config;
+
+	config = p4_general_events[hw_event];
+	bind = p4_config_get_bind(config);
+	esel = P4_OPCODE_ESEL(bind->opcode);
+	config |= p4_config_pack_cccr(P4_CCCR_ESEL(esel));
+
+	return config;
+}
+
+static int p4_hw_config(struct perf_event *event)
+{
+	int cpu = get_cpu();
+	int rc = 0;
+	unsigned int evnt;
+	u32 escr, cccr;
+
+	/*
+	 * the reason we use cpu that early is that: if we get scheduled
+	 * first time on the same cpu -- we will not need swap thread
+	 * specific flags in config (and will save some cpu cycles)
+	 */
+
+	cccr = p4_default_cccr_conf(cpu);
+	escr = p4_default_escr_conf(cpu, event->attr.exclude_kernel,
+					 event->attr.exclude_user);
+	event->hw.config = p4_config_pack_escr(escr) |
+			   p4_config_pack_cccr(cccr);
+
+	if (p4_ht_active() && p4_ht_thread(cpu))
+		event->hw.config = p4_set_ht_bit(event->hw.config);
+
+	if (event->attr.type == PERF_TYPE_RAW) {
+
+		/* user data may have out-of-bound event index */
+		evnt = p4_config_unpack_event(event->attr.config);
+		if (evnt >= ARRAY_SIZE(p4_event_bind_map)) {
+			rc = -EINVAL;
+			goto out;
+		}
+
+		/*
+		 * We don't control raw events so it's up to the caller
+		 * to pass sane values (and we don't count the thread number
+		 * on HT machine but allow HT-compatible specifics to be
+		 * passed on)
+		 *
+		 * XXX: HT wide things should check perf_paranoid_cpu() &&
+		 *      CAP_SYS_ADMIN
+		 */
+		event->hw.config |= event->attr.config &
+			(p4_config_pack_escr(P4_ESCR_MASK_HT) |
+			 p4_config_pack_cccr(P4_CCCR_MASK_HT));
+	}
+
+	rc = x86_setup_perfctr(event);
+out:
+	put_cpu();
+	return rc;
+}
+
+static inline int p4_pmu_clear_cccr_ovf(struct hw_perf_event *hwc)
+{
+	int overflow = 0;
+	u32 low, high;
+
+	rdmsr(hwc->config_base + hwc->idx, low, high);
+
+	/* we need to check high bit for unflagged overflows */
+	if ((low & P4_CCCR_OVF) || !(high & (1 << 31))) {
+		overflow = 1;
+		(void)checking_wrmsrl(hwc->config_base + hwc->idx,
+			((u64)low) & ~P4_CCCR_OVF);
+	}
+
+	return overflow;
+}
+
+static inline void p4_pmu_disable_event(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+
+	/*
+	 * If event gets disabled while counter is in overflowed
+	 * state we need to clear P4_CCCR_OVF, otherwise interrupt get
+	 * asserted again and again
+	 */
+	(void)checking_wrmsrl(hwc->config_base + hwc->idx,
+		(u64)(p4_config_unpack_cccr(hwc->config)) &
+			~P4_CCCR_ENABLE & ~P4_CCCR_OVF & ~P4_CCCR_RESERVED);
+}
+
+static void p4_pmu_disable_all(void)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	int idx;
+
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
+		struct perf_event *event = cpuc->events[idx];
+		if (!test_bit(idx, cpuc->active_mask))
+			continue;
+		p4_pmu_disable_event(event);
+	}
+}
+
+static void p4_pmu_enable_event(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	int thread = p4_ht_config_thread(hwc->config);
+	u64 escr_conf = p4_config_unpack_escr(p4_clear_ht_bit(hwc->config));
+	unsigned int idx = p4_config_unpack_event(hwc->config);
+	unsigned int idx_cache = p4_config_unpack_cache_event(hwc->config);
+	struct p4_event_bind *bind;
+	struct p4_cache_event_bind *bind_cache;
+	u64 escr_addr, cccr;
+
+	bind = &p4_event_bind_map[idx];
+	escr_addr = (u64)bind->escr_msr[thread];
+
+	/*
+	 * - we dont support cascaded counters yet
+	 * - and counter 1 is broken (erratum)
+	 */
+	WARN_ON_ONCE(p4_is_event_cascaded(hwc->config));
+	WARN_ON_ONCE(hwc->idx == 1);
+
+	/* we need a real Event value */
+	escr_conf &= ~P4_ESCR_EVENT_MASK;
+	escr_conf |= P4_ESCR_EVENT(P4_OPCODE_EVNT(bind->opcode));
+
+	cccr = p4_config_unpack_cccr(hwc->config);
+
+	/*
+	 * it could be Cache event so that we need to
+	 * set metrics into additional MSRs
+	 */
+	BUILD_BUG_ON(P4_CACHE__MAX > P4_CCCR_CACHE_OPS_MASK);
+	if (idx_cache > P4_CACHE__NONE &&
+		idx_cache < ARRAY_SIZE(p4_cache_event_bind_map)) {
+		bind_cache = &p4_cache_event_bind_map[idx_cache];
+		(void)checking_wrmsrl(MSR_IA32_PEBS_ENABLE, (u64)bind_cache->metric_pebs);
+		(void)checking_wrmsrl(MSR_P4_PEBS_MATRIX_VERT, (u64)bind_cache->metric_vert);
+	}
+
+	(void)checking_wrmsrl(escr_addr, escr_conf);
+	(void)checking_wrmsrl(hwc->config_base + hwc->idx,
+				(cccr & ~P4_CCCR_RESERVED) | P4_CCCR_ENABLE);
+}
+
+static void p4_pmu_enable_all(int added)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	int idx;
+
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
+		struct perf_event *event = cpuc->events[idx];
+		if (!test_bit(idx, cpuc->active_mask))
+			continue;
+		p4_pmu_enable_event(event);
+	}
+}
+
+static int p4_pmu_handle_irq(struct pt_regs *regs)
+{
+	struct perf_sample_data data;
+	struct cpu_hw_events *cpuc;
+	struct perf_event *event;
+	struct hw_perf_event *hwc;
+	int idx, handled = 0;
+	u64 val;
+
+	data.addr = 0;
+	data.raw = NULL;
+
+	cpuc = &__get_cpu_var(cpu_hw_events);
+
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
+
+		if (!test_bit(idx, cpuc->active_mask))
+			continue;
+
+		event = cpuc->events[idx];
+		hwc = &event->hw;
+
+		WARN_ON_ONCE(hwc->idx != idx);
+
+		/* it might be unflagged overflow */
+		handled = p4_pmu_clear_cccr_ovf(hwc);
+
+		val = x86_perf_event_update(event);
+		if (!handled && (val & (1ULL << (x86_pmu.cntval_bits - 1))))
+			continue;
+
+		/* event overflow for sure */
+		data.period = event->hw.last_period;
+
+		if (!x86_perf_event_set_period(event))
+			continue;
+		if (perf_event_overflow(event, 1, &data, regs))
+			p4_pmu_disable_event(event);
+	}
+
+	if (handled) {
+		/* p4 quirk: unmask it again */
+		apic_write(APIC_LVTPC, apic_read(APIC_LVTPC) & ~APIC_LVT_MASKED);
+		inc_irq_stat(apic_perf_irqs);
+	}
+
+	return handled;
+}
+
+/*
+ * swap thread specific fields according to a thread
+ * we are going to run on
+ */
+static void p4_pmu_swap_config_ts(struct hw_perf_event *hwc, int cpu)
+{
+	u32 escr, cccr;
+
+	/*
+	 * we either lucky and continue on same cpu or no HT support
+	 */
+	if (!p4_should_swap_ts(hwc->config, cpu))
+		return;
+
+	/*
+	 * the event is migrated from an another logical
+	 * cpu, so we need to swap thread specific flags
+	 */
+
+	escr = p4_config_unpack_escr(hwc->config);
+	cccr = p4_config_unpack_cccr(hwc->config);
+
+	if (p4_ht_thread(cpu)) {
+		cccr &= ~P4_CCCR_OVF_PMI_T0;
+		cccr |= P4_CCCR_OVF_PMI_T1;
+		if (escr & P4_ESCR_T0_OS) {
+			escr &= ~P4_ESCR_T0_OS;
+			escr |= P4_ESCR_T1_OS;
+		}
+		if (escr & P4_ESCR_T0_USR) {
+			escr &= ~P4_ESCR_T0_USR;
+			escr |= P4_ESCR_T1_USR;
+		}
+		hwc->config  = p4_config_pack_escr(escr);
+		hwc->config |= p4_config_pack_cccr(cccr);
+		hwc->config |= P4_CONFIG_HT;
+	} else {
+		cccr &= ~P4_CCCR_OVF_PMI_T1;
+		cccr |= P4_CCCR_OVF_PMI_T0;
+		if (escr & P4_ESCR_T1_OS) {
+			escr &= ~P4_ESCR_T1_OS;
+			escr |= P4_ESCR_T0_OS;
+		}
+		if (escr & P4_ESCR_T1_USR) {
+			escr &= ~P4_ESCR_T1_USR;
+			escr |= P4_ESCR_T0_USR;
+		}
+		hwc->config  = p4_config_pack_escr(escr);
+		hwc->config |= p4_config_pack_cccr(cccr);
+		hwc->config &= ~P4_CONFIG_HT;
+	}
+}
+
+/*
+ * ESCR address hashing is tricky, ESCRs are not sequential
+ * in memory but all starts from MSR_P4_BSU_ESCR0 (0x03a0) and
+ * the metric between any ESCRs is laid in range [0xa0,0xe1]
+ *
+ * so we make ~70% filled hashtable
+ */
+
+#define P4_ESCR_MSR_BASE		0x000003a0
+#define P4_ESCR_MSR_MAX			0x000003e1
+#define P4_ESCR_MSR_TABLE_SIZE		(P4_ESCR_MSR_MAX - P4_ESCR_MSR_BASE + 1)
+#define P4_ESCR_MSR_IDX(msr)		(msr - P4_ESCR_MSR_BASE)
+#define P4_ESCR_MSR_TABLE_ENTRY(msr)	[P4_ESCR_MSR_IDX(msr)] = msr
+
+static const unsigned int p4_escr_table[P4_ESCR_MSR_TABLE_SIZE] = {
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ALF_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ALF_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BPU_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BPU_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BSU_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BSU_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR2),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR3),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR4),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR5),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_DAC_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_DAC_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FIRM_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FIRM_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FLAME_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FLAME_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FSB_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FSB_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IQ_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IQ_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IS_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IS_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ITLB_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ITLB_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IX_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IX_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MOB_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MOB_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MS_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MS_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_PMH_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_PMH_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_RAT_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_RAT_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SAAT_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SAAT_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SSU_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SSU_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TBPU_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TBPU_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TC_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TC_ESCR1),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_U2L_ESCR0),
+	P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_U2L_ESCR1),
+};
+
+static int p4_get_escr_idx(unsigned int addr)
+{
+	unsigned int idx = P4_ESCR_MSR_IDX(addr);
+
+	if (unlikely(idx >= P4_ESCR_MSR_TABLE_SIZE	||
+			!p4_escr_table[idx]		||
+			p4_escr_table[idx] != addr)) {
+		WARN_ONCE(1, "P4 PMU: Wrong address passed: %x\n", addr);
+		return -1;
+	}
+
+	return idx;
+}
+
+static int p4_next_cntr(int thread, unsigned long *used_mask,
+			struct p4_event_bind *bind)
+{
+	int i, j;
+
+	for (i = 0; i < P4_CNTR_LIMIT; i++) {
+		j = bind->cntr[thread][i];
+		if (j != -1 && !test_bit(j, used_mask))
+			return j;
+	}
+
+	return -1;
+}
+
+static int p4_pmu_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
+{
+	unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+	unsigned long escr_mask[BITS_TO_LONGS(P4_ESCR_MSR_TABLE_SIZE)];
+	int cpu = smp_processor_id();
+	struct hw_perf_event *hwc;
+	struct p4_event_bind *bind;
+	unsigned int i, thread, num;
+	int cntr_idx, escr_idx;
+
+	bitmap_zero(used_mask, X86_PMC_IDX_MAX);
+	bitmap_zero(escr_mask, P4_ESCR_MSR_TABLE_SIZE);
+
+	for (i = 0, num = n; i < n; i++, num--) {
+
+		hwc = &cpuc->event_list[i]->hw;
+		thread = p4_ht_thread(cpu);
+		bind = p4_config_get_bind(hwc->config);
+		escr_idx = p4_get_escr_idx(bind->escr_msr[thread]);
+		if (unlikely(escr_idx == -1))
+			goto done;
+
+		if (hwc->idx != -1 && !p4_should_swap_ts(hwc->config, cpu)) {
+			cntr_idx = hwc->idx;
+			if (assign)
+				assign[i] = hwc->idx;
+			goto reserve;
+		}
+
+		cntr_idx = p4_next_cntr(thread, used_mask, bind);
+		if (cntr_idx == -1 || test_bit(escr_idx, escr_mask))
+			goto done;
+
+		p4_pmu_swap_config_ts(hwc, cpu);
+		if (assign)
+			assign[i] = cntr_idx;
+reserve:
+		set_bit(cntr_idx, used_mask);
+		set_bit(escr_idx, escr_mask);
+	}
+
+done:
+	return num ? -ENOSPC : 0;
+}
+
+static __initconst const struct x86_pmu p4_pmu = {
+	.name			= "Netburst P4/Xeon",
+	.handle_irq		= p4_pmu_handle_irq,
+	.disable_all		= p4_pmu_disable_all,
+	.enable_all		= p4_pmu_enable_all,
+	.enable			= p4_pmu_enable_event,
+	.disable		= p4_pmu_disable_event,
+	.eventsel		= MSR_P4_BPU_CCCR0,
+	.perfctr		= MSR_P4_BPU_PERFCTR0,
+	.event_map		= p4_pmu_event_map,
+	.max_events		= ARRAY_SIZE(p4_general_events),
+	.get_event_constraints	= x86_get_event_constraints,
+	/*
+	 * IF HT disabled we may need to use all
+	 * ARCH_P4_MAX_CCCR counters simulaneously
+	 * though leave it restricted at moment assuming
+	 * HT is on
+	 */
+	.num_counters		= ARCH_P4_MAX_CCCR,
+	.apic			= 1,
+	.cntval_bits		= 40,
+	.cntval_mask		= (1ULL << 40) - 1,
+	.max_period		= (1ULL << 39) - 1,
+	.hw_config		= p4_hw_config,
+	.schedule_events	= p4_pmu_schedule_events,
+};
+
+static __init int p4_pmu_init(void)
+{
+	unsigned int low, high;
+
+	/* If we get stripped -- indexig fails */
+	BUILD_BUG_ON(ARCH_P4_MAX_CCCR > X86_PMC_MAX_GENERIC);
+
+	rdmsr(MSR_IA32_MISC_ENABLE, low, high);
+	if (!(low & (1 << 7))) {
+		pr_cont("unsupported Netburst CPU model %d ",
+			boot_cpu_data.x86_model);
+		return -ENODEV;
+	}
+
+	memcpy(hw_cache_event_ids, p4_hw_cache_event_ids,
+		sizeof(hw_cache_event_ids));
+
+	pr_cont("Netburst events, ");
+
+	x86_pmu = p4_pmu;
+
+	return 0;
+}
+
+#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_p6.c b/arch/x86/kernel/cpu/perf_event_p6.c
index a330485d14d..34ba07be2cd 100644
--- a/arch/x86/kernel/cpu/perf_event_p6.c
+++ b/arch/x86/kernel/cpu/perf_event_p6.c
@@ -27,24 +27,6 @@ static u64 p6_pmu_event_map(int hw_event)
  */
 #define P6_NOP_EVENT			0x0000002EULL
 
-static u64 p6_pmu_raw_event(u64 hw_event)
-{
-#define P6_EVNTSEL_EVENT_MASK		0x000000FFULL
-#define P6_EVNTSEL_UNIT_MASK		0x0000FF00ULL
-#define P6_EVNTSEL_EDGE_MASK		0x00040000ULL
-#define P6_EVNTSEL_INV_MASK		0x00800000ULL
-#define P6_EVNTSEL_REG_MASK		0xFF000000ULL
-
-#define P6_EVNTSEL_MASK			\
-	(P6_EVNTSEL_EVENT_MASK |	\
-	 P6_EVNTSEL_UNIT_MASK  |	\
-	 P6_EVNTSEL_EDGE_MASK  |	\
-	 P6_EVNTSEL_INV_MASK   |	\
-	 P6_EVNTSEL_REG_MASK)
-
-	return hw_event & P6_EVNTSEL_MASK;
-}
-
 static struct event_constraint p6_event_constraints[] =
 {
 	INTEL_EVENT_CONSTRAINT(0xc1, 0x1),	/* FLOPS */
@@ -66,7 +48,7 @@ static void p6_pmu_disable_all(void)
 	wrmsrl(MSR_P6_EVNTSEL0, val);
 }
 
-static void p6_pmu_enable_all(void)
+static void p6_pmu_enable_all(int added)
 {
 	unsigned long val;
 
@@ -102,22 +84,23 @@ static void p6_pmu_enable_event(struct perf_event *event)
 	(void)checking_wrmsrl(hwc->config_base + hwc->idx, val);
 }
 
-static __initconst struct x86_pmu p6_pmu = {
+static __initconst const struct x86_pmu p6_pmu = {
 	.name			= "p6",
 	.handle_irq		= x86_pmu_handle_irq,
 	.disable_all		= p6_pmu_disable_all,
 	.enable_all		= p6_pmu_enable_all,
 	.enable			= p6_pmu_enable_event,
 	.disable		= p6_pmu_disable_event,
+	.hw_config		= x86_pmu_hw_config,
+	.schedule_events	= x86_schedule_events,
 	.eventsel		= MSR_P6_EVNTSEL0,
 	.perfctr		= MSR_P6_PERFCTR0,
 	.event_map		= p6_pmu_event_map,
-	.raw_event		= p6_pmu_raw_event,
 	.max_events		= ARRAY_SIZE(p6_perfmon_event_map),
 	.apic			= 1,
 	.max_period		= (1ULL << 31) - 1,
 	.version		= 0,
-	.num_events		= 2,
+	.num_counters		= 2,
 	/*
 	 * Events have 40 bits implemented. However they are designed such
 	 * that bits [32-39] are sign extensions of bit 31. As such the
@@ -125,8 +108,8 @@ static __initconst struct x86_pmu p6_pmu = {
 	 *
 	 * See IA-32 Intel Architecture Software developer manual Vol 3B
 	 */
-	.event_bits		= 32,
-	.event_mask		= (1ULL << 32) - 1,
+	.cntval_bits		= 32,
+	.cntval_mask		= (1ULL << 32) - 1,
 	.get_event_constraints	= x86_get_event_constraints,
 	.event_constraints	= p6_event_constraints,
 };
diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c
index dfdb4dba232..b9d1ff58844 100644
--- a/arch/x86/kernel/cpu/vmware.c
+++ b/arch/x86/kernel/cpu/vmware.c
@@ -24,8 +24,8 @@
 #include <linux/dmi.h>
 #include <linux/module.h>
 #include <asm/div64.h>
-#include <asm/vmware.h>
 #include <asm/x86_init.h>
+#include <asm/hypervisor.h>
 
 #define CPUID_VMWARE_INFO_LEAF	0x40000000
 #define VMWARE_HYPERVISOR_MAGIC	0x564D5868
@@ -65,7 +65,7 @@ static unsigned long vmware_get_tsc_khz(void)
 	return tsc_hz;
 }
 
-void __init vmware_platform_setup(void)
+static void __init vmware_platform_setup(void)
 {
 	uint32_t eax, ebx, ecx, edx;
 
@@ -83,26 +83,22 @@ void __init vmware_platform_setup(void)
  * serial key should be enough, as this will always have a VMware
  * specific string when running under VMware hypervisor.
  */
-int vmware_platform(void)
+static bool __init vmware_platform(void)
 {
 	if (cpu_has_hypervisor) {
-		unsigned int eax, ebx, ecx, edx;
-		char hyper_vendor_id[13];
-
-		cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &ebx, &ecx, &edx);
-		memcpy(hyper_vendor_id + 0, &ebx, 4);
-		memcpy(hyper_vendor_id + 4, &ecx, 4);
-		memcpy(hyper_vendor_id + 8, &edx, 4);
-		hyper_vendor_id[12] = '\0';
-		if (!strcmp(hyper_vendor_id, "VMwareVMware"))
-			return 1;
+		unsigned int eax;
+		unsigned int hyper_vendor_id[3];
+
+		cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &hyper_vendor_id[0],
+		      &hyper_vendor_id[1], &hyper_vendor_id[2]);
+		if (!memcmp(hyper_vendor_id, "VMwareVMware", 12))
+			return true;
 	} else if (dmi_available && dmi_name_in_serial("VMware") &&
 		   __vmware_platform())
-		return 1;
+		return true;
 
-	return 0;
+	return false;
 }
-EXPORT_SYMBOL(vmware_platform);
 
 /*
  * VMware hypervisor takes care of exporting a reliable TSC to the guest.
@@ -116,8 +112,16 @@ EXPORT_SYMBOL(vmware_platform);
  * so that the kernel could just trust the hypervisor with providing a
  * reliable virtual TSC that is suitable for timekeeping.
  */
-void __cpuinit vmware_set_feature_bits(struct cpuinfo_x86 *c)
+static void __cpuinit vmware_set_cpu_features(struct cpuinfo_x86 *c)
 {
 	set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC);
 	set_cpu_cap(c, X86_FEATURE_TSC_RELIABLE);
 }
+
+const __refconst struct hypervisor_x86 x86_hyper_vmware = {
+	.name			= "VMware",
+	.detect			= vmware_platform,
+	.set_cpu_features	= vmware_set_cpu_features,
+	.init_platform		= vmware_platform_setup,
+};
+EXPORT_SYMBOL(x86_hyper_vmware);
diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c
index 8b862d5900f..1b7b31ab7d8 100644
--- a/arch/x86/kernel/cpuid.c
+++ b/arch/x86/kernel/cpuid.c
@@ -170,7 +170,7 @@ static int __cpuinit cpuid_class_cpu_callback(struct notifier_block *nfb,
 		cpuid_device_destroy(cpu);
 		break;
 	}
-	return err ? NOTIFY_BAD : NOTIFY_OK;
+	return notifier_from_errno(err);
 }
 
 static struct notifier_block __refdata cpuid_class_cpu_notifier =
diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c
deleted file mode 100644
index 1c47390dd0e..00000000000
--- a/arch/x86/kernel/ds.c
+++ /dev/null
@@ -1,1437 +0,0 @@
-/*
- * Debug Store support
- *
- * This provides a low-level interface to the hardware's Debug Store
- * feature that is used for branch trace store (BTS) and
- * precise-event based sampling (PEBS).
- *
- * It manages:
- * - DS and BTS hardware configuration
- * - buffer overflow handling (to be done)
- * - buffer access
- *
- * It does not do:
- * - security checking (is the caller allowed to trace the task)
- * - buffer allocation (memory accounting)
- *
- *
- * Copyright (C) 2007-2009 Intel Corporation.
- * Markus Metzger <markus.t.metzger@intel.com>, 2007-2009
- */
-
-#include <linux/kernel.h>
-#include <linux/string.h>
-#include <linux/errno.h>
-#include <linux/sched.h>
-#include <linux/slab.h>
-#include <linux/mm.h>
-#include <linux/trace_clock.h>
-
-#include <asm/ds.h>
-
-#include "ds_selftest.h"
-
-/*
- * The configuration for a particular DS hardware implementation:
- */
-struct ds_configuration {
-	/* The name of the configuration: */
-	const char		*name;
-
-	/* The size of pointer-typed fields in DS, BTS, and PEBS: */
-	unsigned char		sizeof_ptr_field;
-
-	/* The size of a BTS/PEBS record in bytes: */
-	unsigned char		sizeof_rec[2];
-
-	/* The number of pebs counter reset values in the DS structure. */
-	unsigned char		nr_counter_reset;
-
-	/* Control bit-masks indexed by enum ds_feature: */
-	unsigned long		ctl[dsf_ctl_max];
-};
-static struct ds_configuration ds_cfg __read_mostly;
-
-
-/* Maximal size of a DS configuration: */
-#define MAX_SIZEOF_DS		0x80
-
-/* Maximal size of a BTS record: */
-#define MAX_SIZEOF_BTS		(3 * 8)
-
-/* BTS and PEBS buffer alignment: */
-#define DS_ALIGNMENT		(1 << 3)
-
-/* Number of buffer pointers in DS: */
-#define NUM_DS_PTR_FIELDS	8
-
-/* Size of a pebs reset value in DS: */
-#define PEBS_RESET_FIELD_SIZE	8
-
-/* Mask of control bits in the DS MSR register: */
-#define BTS_CONTROL				  \
-	( ds_cfg.ctl[dsf_bts]			| \
-	  ds_cfg.ctl[dsf_bts_kernel]		| \
-	  ds_cfg.ctl[dsf_bts_user]		| \
-	  ds_cfg.ctl[dsf_bts_overflow] )
-
-/*
- * A BTS or PEBS tracer.
- *
- * This holds the configuration of the tracer and serves as a handle
- * to identify tracers.
- */
-struct ds_tracer {
-	/* The DS context (partially) owned by this tracer. */
-	struct ds_context	*context;
-	/* The buffer provided on ds_request() and its size in bytes. */
-	void			*buffer;
-	size_t			size;
-};
-
-struct bts_tracer {
-	/* The common DS part: */
-	struct ds_tracer	ds;
-
-	/* The trace including the DS configuration: */
-	struct bts_trace	trace;
-
-	/* Buffer overflow notification function: */
-	bts_ovfl_callback_t	ovfl;
-
-	/* Active flags affecting trace collection. */
-	unsigned int		flags;
-};
-
-struct pebs_tracer {
-	/* The common DS part: */
-	struct ds_tracer	ds;
-
-	/* The trace including the DS configuration: */
-	struct pebs_trace	trace;
-
-	/* Buffer overflow notification function: */
-	pebs_ovfl_callback_t	ovfl;
-};
-
-/*
- * Debug Store (DS) save area configuration (see Intel64 and IA32
- * Architectures Software Developer's Manual, section 18.5)
- *
- * The DS configuration consists of the following fields; different
- * architetures vary in the size of those fields.
- *
- * - double-word aligned base linear address of the BTS buffer
- * - write pointer into the BTS buffer
- * - end linear address of the BTS buffer (one byte beyond the end of
- *   the buffer)
- * - interrupt pointer into BTS buffer
- *   (interrupt occurs when write pointer passes interrupt pointer)
- * - double-word aligned base linear address of the PEBS buffer
- * - write pointer into the PEBS buffer
- * - end linear address of the PEBS buffer (one byte beyond the end of
- *   the buffer)
- * - interrupt pointer into PEBS buffer
- *   (interrupt occurs when write pointer passes interrupt pointer)
- * - value to which counter is reset following counter overflow
- *
- * Later architectures use 64bit pointers throughout, whereas earlier
- * architectures use 32bit pointers in 32bit mode.
- *
- *
- * We compute the base address for the first 8 fields based on:
- * - the field size stored in the DS configuration
- * - the relative field position
- * - an offset giving the start of the respective region
- *
- * This offset is further used to index various arrays holding
- * information for BTS and PEBS at the respective index.
- *
- * On later 32bit processors, we only access the lower 32bit of the
- * 64bit pointer fields. The upper halves will be zeroed out.
- */
-
-enum ds_field {
-	ds_buffer_base = 0,
-	ds_index,
-	ds_absolute_maximum,
-	ds_interrupt_threshold,
-};
-
-enum ds_qualifier {
-	ds_bts = 0,
-	ds_pebs
-};
-
-static inline unsigned long
-ds_get(const unsigned char *base, enum ds_qualifier qual, enum ds_field field)
-{
-	base += (ds_cfg.sizeof_ptr_field * (field + (4 * qual)));
-	return *(unsigned long *)base;
-}
-
-static inline void
-ds_set(unsigned char *base, enum ds_qualifier qual, enum ds_field field,
-       unsigned long value)
-{
-	base += (ds_cfg.sizeof_ptr_field * (field + (4 * qual)));
-	(*(unsigned long *)base) = value;
-}
-
-
-/*
- * Locking is done only for allocating BTS or PEBS resources.
- */
-static DEFINE_SPINLOCK(ds_lock);
-
-/*
- * We either support (system-wide) per-cpu or per-thread allocation.
- * We distinguish the two based on the task_struct pointer, where a
- * NULL pointer indicates per-cpu allocation for the current cpu.
- *
- * Allocations are use-counted. As soon as resources are allocated,
- * further allocations must be of the same type (per-cpu or
- * per-thread). We model this by counting allocations (i.e. the number
- * of tracers of a certain type) for one type negatively:
- *   =0  no tracers
- *   >0  number of per-thread tracers
- *   <0  number of per-cpu tracers
- *
- * Tracers essentially gives the number of ds contexts for a certain
- * type of allocation.
- */
-static atomic_t tracers = ATOMIC_INIT(0);
-
-static inline int get_tracer(struct task_struct *task)
-{
-	int error;
-
-	spin_lock_irq(&ds_lock);
-
-	if (task) {
-		error = -EPERM;
-		if (atomic_read(&tracers) < 0)
-			goto out;
-		atomic_inc(&tracers);
-	} else {
-		error = -EPERM;
-		if (atomic_read(&tracers) > 0)
-			goto out;
-		atomic_dec(&tracers);
-	}
-
-	error = 0;
-out:
-	spin_unlock_irq(&ds_lock);
-	return error;
-}
-
-static inline void put_tracer(struct task_struct *task)
-{
-	if (task)
-		atomic_dec(&tracers);
-	else
-		atomic_inc(&tracers);
-}
-
-/*
- * The DS context is either attached to a thread or to a cpu:
- * - in the former case, the thread_struct contains a pointer to the
- *   attached context.
- * - in the latter case, we use a static array of per-cpu context
- *   pointers.
- *
- * Contexts are use-counted. They are allocated on first access and
- * deallocated when the last user puts the context.
- */
-struct ds_context {
-	/* The DS configuration; goes into MSR_IA32_DS_AREA: */
-	unsigned char		ds[MAX_SIZEOF_DS];
-
-	/* The owner of the BTS and PEBS configuration, respectively: */
-	struct bts_tracer	*bts_master;
-	struct pebs_tracer	*pebs_master;
-
-	/* Use count: */
-	unsigned long		count;
-
-	/* Pointer to the context pointer field: */
-	struct ds_context	**this;
-
-	/* The traced task; NULL for cpu tracing: */
-	struct task_struct	*task;
-
-	/* The traced cpu; only valid if task is NULL: */
-	int			cpu;
-};
-
-static DEFINE_PER_CPU(struct ds_context *, cpu_ds_context);
-
-
-static struct ds_context *ds_get_context(struct task_struct *task, int cpu)
-{
-	struct ds_context **p_context =
-		(task ? &task->thread.ds_ctx : &per_cpu(cpu_ds_context, cpu));
-	struct ds_context *context = NULL;
-	struct ds_context *new_context = NULL;
-
-	/* Chances are small that we already have a context. */
-	new_context = kzalloc(sizeof(*new_context), GFP_KERNEL);
-	if (!new_context)
-		return NULL;
-
-	spin_lock_irq(&ds_lock);
-
-	context = *p_context;
-	if (likely(!context)) {
-		context = new_context;
-
-		context->this = p_context;
-		context->task = task;
-		context->cpu = cpu;
-		context->count = 0;
-
-		*p_context = context;
-	}
-
-	context->count++;
-
-	spin_unlock_irq(&ds_lock);
-
-	if (context != new_context)
-		kfree(new_context);
-
-	return context;
-}
-
-static void ds_put_context(struct ds_context *context)
-{
-	struct task_struct *task;
-	unsigned long irq;
-
-	if (!context)
-		return;
-
-	spin_lock_irqsave(&ds_lock, irq);
-
-	if (--context->count) {
-		spin_unlock_irqrestore(&ds_lock, irq);
-		return;
-	}
-
-	*(context->this) = NULL;
-
-	task = context->task;
-
-	if (task)
-		clear_tsk_thread_flag(task, TIF_DS_AREA_MSR);
-
-	/*
-	 * We leave the (now dangling) pointer to the DS configuration in
-	 * the DS_AREA msr. This is as good or as bad as replacing it with
-	 * NULL - the hardware would crash if we enabled tracing.
-	 *
-	 * This saves us some problems with having to write an msr on a
-	 * different cpu while preventing others from doing the same for the
-	 * next context for that same cpu.
-	 */
-
-	spin_unlock_irqrestore(&ds_lock, irq);
-
-	/* The context might still be in use for context switching. */
-	if (task && (task != current))
-		wait_task_context_switch(task);
-
-	kfree(context);
-}
-
-static void ds_install_ds_area(struct ds_context *context)
-{
-	unsigned long ds;
-
-	ds = (unsigned long)context->ds;
-
-	/*
-	 * There is a race between the bts master and the pebs master.
-	 *
-	 * The thread/cpu access is synchronized via get/put_cpu() for
-	 * task tracing and via wrmsr_on_cpu for cpu tracing.
-	 *
-	 * If bts and pebs are collected for the same task or same cpu,
-	 * the same confiuration is written twice.
-	 */
-	if (context->task) {
-		get_cpu();
-		if (context->task == current)
-			wrmsrl(MSR_IA32_DS_AREA, ds);
-		set_tsk_thread_flag(context->task, TIF_DS_AREA_MSR);
-		put_cpu();
-	} else
-		wrmsr_on_cpu(context->cpu, MSR_IA32_DS_AREA,
-			     (u32)((u64)ds), (u32)((u64)ds >> 32));
-}
-
-/*
- * Call the tracer's callback on a buffer overflow.
- *
- * context: the ds context
- * qual: the buffer type
- */
-static void ds_overflow(struct ds_context *context, enum ds_qualifier qual)
-{
-	switch (qual) {
-	case ds_bts:
-		if (context->bts_master &&
-		    context->bts_master->ovfl)
-			context->bts_master->ovfl(context->bts_master);
-		break;
-	case ds_pebs:
-		if (context->pebs_master &&
-		    context->pebs_master->ovfl)
-			context->pebs_master->ovfl(context->pebs_master);
-		break;
-	}
-}
-
-
-/*
- * Write raw data into the BTS or PEBS buffer.
- *
- * The remainder of any partially written record is zeroed out.
- *
- * context: the DS context
- * qual:    the buffer type
- * record:  the data to write
- * size:    the size of the data
- */
-static int ds_write(struct ds_context *context, enum ds_qualifier qual,
-		    const void *record, size_t size)
-{
-	int bytes_written = 0;
-
-	if (!record)
-		return -EINVAL;
-
-	while (size) {
-		unsigned long base, index, end, write_end, int_th;
-		unsigned long write_size, adj_write_size;
-
-		/*
-		 * Write as much as possible without producing an
-		 * overflow interrupt.
-		 *
-		 * Interrupt_threshold must either be
-		 * - bigger than absolute_maximum or
-		 * - point to a record between buffer_base and absolute_maximum
-		 *
-		 * Index points to a valid record.
-		 */
-		base   = ds_get(context->ds, qual, ds_buffer_base);
-		index  = ds_get(context->ds, qual, ds_index);
-		end    = ds_get(context->ds, qual, ds_absolute_maximum);
-		int_th = ds_get(context->ds, qual, ds_interrupt_threshold);
-
-		write_end = min(end, int_th);
-
-		/*
-		 * If we are already beyond the interrupt threshold,
-		 * we fill the entire buffer.
-		 */
-		if (write_end <= index)
-			write_end = end;
-
-		if (write_end <= index)
-			break;
-
-		write_size = min((unsigned long) size, write_end - index);
-		memcpy((void *)index, record, write_size);
-
-		record = (const char *)record + write_size;
-		size -= write_size;
-		bytes_written += write_size;
-
-		adj_write_size = write_size / ds_cfg.sizeof_rec[qual];
-		adj_write_size *= ds_cfg.sizeof_rec[qual];
-
-		/* Zero out trailing bytes. */
-		memset((char *)index + write_size, 0,
-		       adj_write_size - write_size);
-		index += adj_write_size;
-
-		if (index >= end)
-			index = base;
-		ds_set(context->ds, qual, ds_index, index);
-
-		if (index >= int_th)
-			ds_overflow(context, qual);
-	}
-
-	return bytes_written;
-}
-
-
-/*
- * Branch Trace Store (BTS) uses the following format. Different
- * architectures vary in the size of those fields.
- * - source linear address
- * - destination linear address
- * - flags
- *
- * Later architectures use 64bit pointers throughout, whereas earlier
- * architectures use 32bit pointers in 32bit mode.
- *
- * We compute the base address for the fields based on:
- * - the field size stored in the DS configuration
- * - the relative field position
- *
- * In order to store additional information in the BTS buffer, we use
- * a special source address to indicate that the record requires
- * special interpretation.
- *
- * Netburst indicated via a bit in the flags field whether the branch
- * was predicted; this is ignored.
- *
- * We use two levels of abstraction:
- * - the raw data level defined here
- * - an arch-independent level defined in ds.h
- */
-
-enum bts_field {
-	bts_from,
-	bts_to,
-	bts_flags,
-
-	bts_qual		= bts_from,
-	bts_clock		= bts_to,
-	bts_pid			= bts_flags,
-
-	bts_qual_mask		= (bts_qual_max - 1),
-	bts_escape		= ((unsigned long)-1 & ~bts_qual_mask)
-};
-
-static inline unsigned long bts_get(const char *base, unsigned long field)
-{
-	base += (ds_cfg.sizeof_ptr_field * field);
-	return *(unsigned long *)base;
-}
-
-static inline void bts_set(char *base, unsigned long field, unsigned long val)
-{
-	base += (ds_cfg.sizeof_ptr_field * field);
-	(*(unsigned long *)base) = val;
-}
-
-
-/*
- * The raw BTS data is architecture dependent.
- *
- * For higher-level users, we give an arch-independent view.
- * - ds.h defines struct bts_struct
- * - bts_read translates one raw bts record into a bts_struct
- * - bts_write translates one bts_struct into the raw format and
- *   writes it into the top of the parameter tracer's buffer.
- *
- * return: bytes read/written on success; -Eerrno, otherwise
- */
-static int
-bts_read(struct bts_tracer *tracer, const void *at, struct bts_struct *out)
-{
-	if (!tracer)
-		return -EINVAL;
-
-	if (at < tracer->trace.ds.begin)
-		return -EINVAL;
-
-	if (tracer->trace.ds.end < (at + tracer->trace.ds.size))
-		return -EINVAL;
-
-	memset(out, 0, sizeof(*out));
-	if ((bts_get(at, bts_qual) & ~bts_qual_mask) == bts_escape) {
-		out->qualifier = (bts_get(at, bts_qual) & bts_qual_mask);
-		out->variant.event.clock = bts_get(at, bts_clock);
-		out->variant.event.pid = bts_get(at, bts_pid);
-	} else {
-		out->qualifier = bts_branch;
-		out->variant.lbr.from = bts_get(at, bts_from);
-		out->variant.lbr.to   = bts_get(at, bts_to);
-
-		if (!out->variant.lbr.from && !out->variant.lbr.to)
-			out->qualifier = bts_invalid;
-	}
-
-	return ds_cfg.sizeof_rec[ds_bts];
-}
-
-static int bts_write(struct bts_tracer *tracer, const struct bts_struct *in)
-{
-	unsigned char raw[MAX_SIZEOF_BTS];
-
-	if (!tracer)
-		return -EINVAL;
-
-	if (MAX_SIZEOF_BTS < ds_cfg.sizeof_rec[ds_bts])
-		return -EOVERFLOW;
-
-	switch (in->qualifier) {
-	case bts_invalid:
-		bts_set(raw, bts_from, 0);
-		bts_set(raw, bts_to, 0);
-		bts_set(raw, bts_flags, 0);
-		break;
-	case bts_branch:
-		bts_set(raw, bts_from, in->variant.lbr.from);
-		bts_set(raw, bts_to,   in->variant.lbr.to);
-		bts_set(raw, bts_flags, 0);
-		break;
-	case bts_task_arrives:
-	case bts_task_departs:
-		bts_set(raw, bts_qual, (bts_escape | in->qualifier));
-		bts_set(raw, bts_clock, in->variant.event.clock);
-		bts_set(raw, bts_pid, in->variant.event.pid);
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	return ds_write(tracer->ds.context, ds_bts, raw,
-			ds_cfg.sizeof_rec[ds_bts]);
-}
-
-
-static void ds_write_config(struct ds_context *context,
-			    struct ds_trace *cfg, enum ds_qualifier qual)
-{
-	unsigned char *ds = context->ds;
-
-	ds_set(ds, qual, ds_buffer_base, (unsigned long)cfg->begin);
-	ds_set(ds, qual, ds_index, (unsigned long)cfg->top);
-	ds_set(ds, qual, ds_absolute_maximum, (unsigned long)cfg->end);
-	ds_set(ds, qual, ds_interrupt_threshold, (unsigned long)cfg->ith);
-}
-
-static void ds_read_config(struct ds_context *context,
-			   struct ds_trace *cfg, enum ds_qualifier qual)
-{
-	unsigned char *ds = context->ds;
-
-	cfg->begin = (void *)ds_get(ds, qual, ds_buffer_base);
-	cfg->top = (void *)ds_get(ds, qual, ds_index);
-	cfg->end = (void *)ds_get(ds, qual, ds_absolute_maximum);
-	cfg->ith = (void *)ds_get(ds, qual, ds_interrupt_threshold);
-}
-
-static void ds_init_ds_trace(struct ds_trace *trace, enum ds_qualifier qual,
-			     void *base, size_t size, size_t ith,
-			     unsigned int flags) {
-	unsigned long buffer, adj;
-
-	/*
-	 * Adjust the buffer address and size to meet alignment
-	 * constraints:
-	 * - buffer is double-word aligned
-	 * - size is multiple of record size
-	 *
-	 * We checked the size at the very beginning; we have enough
-	 * space to do the adjustment.
-	 */
-	buffer = (unsigned long)base;
-
-	adj = ALIGN(buffer, DS_ALIGNMENT) - buffer;
-	buffer += adj;
-	size   -= adj;
-
-	trace->n = size / ds_cfg.sizeof_rec[qual];
-	trace->size = ds_cfg.sizeof_rec[qual];
-
-	size = (trace->n * trace->size);
-
-	trace->begin = (void *)buffer;
-	trace->top = trace->begin;
-	trace->end = (void *)(buffer + size);
-	/*
-	 * The value for 'no threshold' is -1, which will set the
-	 * threshold outside of the buffer, just like we want it.
-	 */
-	ith *= ds_cfg.sizeof_rec[qual];
-	trace->ith = (void *)(buffer + size - ith);
-
-	trace->flags = flags;
-}
-
-
-static int ds_request(struct ds_tracer *tracer, struct ds_trace *trace,
-		      enum ds_qualifier qual, struct task_struct *task,
-		      int cpu, void *base, size_t size, size_t th)
-{
-	struct ds_context *context;
-	int error;
-	size_t req_size;
-
-	error = -EOPNOTSUPP;
-	if (!ds_cfg.sizeof_rec[qual])
-		goto out;
-
-	error = -EINVAL;
-	if (!base)
-		goto out;
-
-	req_size = ds_cfg.sizeof_rec[qual];
-	/* We might need space for alignment adjustments. */
-	if (!IS_ALIGNED((unsigned long)base, DS_ALIGNMENT))
-		req_size += DS_ALIGNMENT;
-
-	error = -EINVAL;
-	if (size < req_size)
-		goto out;
-
-	if (th != (size_t)-1) {
-		th *= ds_cfg.sizeof_rec[qual];
-
-		error = -EINVAL;
-		if (size <= th)
-			goto out;
-	}
-
-	tracer->buffer = base;
-	tracer->size = size;
-
-	error = -ENOMEM;
-	context = ds_get_context(task, cpu);
-	if (!context)
-		goto out;
-	tracer->context = context;
-
-	/*
-	 * Defer any tracer-specific initialization work for the context until
-	 * context ownership has been clarified.
-	 */
-
-	error = 0;
- out:
-	return error;
-}
-
-static struct bts_tracer *ds_request_bts(struct task_struct *task, int cpu,
-					 void *base, size_t size,
-					 bts_ovfl_callback_t ovfl, size_t th,
-					 unsigned int flags)
-{
-	struct bts_tracer *tracer;
-	int error;
-
-	/* Buffer overflow notification is not yet implemented. */
-	error = -EOPNOTSUPP;
-	if (ovfl)
-		goto out;
-
-	error = get_tracer(task);
-	if (error < 0)
-		goto out;
-
-	error = -ENOMEM;
-	tracer = kzalloc(sizeof(*tracer), GFP_KERNEL);
-	if (!tracer)
-		goto out_put_tracer;
-	tracer->ovfl = ovfl;
-
-	/* Do some more error checking and acquire a tracing context. */
-	error = ds_request(&tracer->ds, &tracer->trace.ds,
-			   ds_bts, task, cpu, base, size, th);
-	if (error < 0)
-		goto out_tracer;
-
-	/* Claim the bts part of the tracing context we acquired above. */
-	spin_lock_irq(&ds_lock);
-
-	error = -EPERM;
-	if (tracer->ds.context->bts_master)
-		goto out_unlock;
-	tracer->ds.context->bts_master = tracer;
-
-	spin_unlock_irq(&ds_lock);
-
-	/*
-	 * Now that we own the bts part of the context, let's complete the
-	 * initialization for that part.
-	 */
-	ds_init_ds_trace(&tracer->trace.ds, ds_bts, base, size, th, flags);
-	ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_bts);
-	ds_install_ds_area(tracer->ds.context);
-
-	tracer->trace.read  = bts_read;
-	tracer->trace.write = bts_write;
-
-	/* Start tracing. */
-	ds_resume_bts(tracer);
-
-	return tracer;
-
- out_unlock:
-	spin_unlock_irq(&ds_lock);
-	ds_put_context(tracer->ds.context);
- out_tracer:
-	kfree(tracer);
- out_put_tracer:
-	put_tracer(task);
- out:
-	return ERR_PTR(error);
-}
-
-struct bts_tracer *ds_request_bts_task(struct task_struct *task,
-				       void *base, size_t size,
-				       bts_ovfl_callback_t ovfl,
-				       size_t th, unsigned int flags)
-{
-	return ds_request_bts(task, 0, base, size, ovfl, th, flags);
-}
-
-struct bts_tracer *ds_request_bts_cpu(int cpu, void *base, size_t size,
-				      bts_ovfl_callback_t ovfl,
-				      size_t th, unsigned int flags)
-{
-	return ds_request_bts(NULL, cpu, base, size, ovfl, th, flags);
-}
-
-static struct pebs_tracer *ds_request_pebs(struct task_struct *task, int cpu,
-					   void *base, size_t size,
-					   pebs_ovfl_callback_t ovfl, size_t th,
-					   unsigned int flags)
-{
-	struct pebs_tracer *tracer;
-	int error;
-
-	/* Buffer overflow notification is not yet implemented. */
-	error = -EOPNOTSUPP;
-	if (ovfl)
-		goto out;
-
-	error = get_tracer(task);
-	if (error < 0)
-		goto out;
-
-	error = -ENOMEM;
-	tracer = kzalloc(sizeof(*tracer), GFP_KERNEL);
-	if (!tracer)
-		goto out_put_tracer;
-	tracer->ovfl = ovfl;
-
-	/* Do some more error checking and acquire a tracing context. */
-	error = ds_request(&tracer->ds, &tracer->trace.ds,
-			   ds_pebs, task, cpu, base, size, th);
-	if (error < 0)
-		goto out_tracer;
-
-	/* Claim the pebs part of the tracing context we acquired above. */
-	spin_lock_irq(&ds_lock);
-
-	error = -EPERM;
-	if (tracer->ds.context->pebs_master)
-		goto out_unlock;
-	tracer->ds.context->pebs_master = tracer;
-
-	spin_unlock_irq(&ds_lock);
-
-	/*
-	 * Now that we own the pebs part of the context, let's complete the
-	 * initialization for that part.
-	 */
-	ds_init_ds_trace(&tracer->trace.ds, ds_pebs, base, size, th, flags);
-	ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_pebs);
-	ds_install_ds_area(tracer->ds.context);
-
-	/* Start tracing. */
-	ds_resume_pebs(tracer);
-
-	return tracer;
-
- out_unlock:
-	spin_unlock_irq(&ds_lock);
-	ds_put_context(tracer->ds.context);
- out_tracer:
-	kfree(tracer);
- out_put_tracer:
-	put_tracer(task);
- out:
-	return ERR_PTR(error);
-}
-
-struct pebs_tracer *ds_request_pebs_task(struct task_struct *task,
-					 void *base, size_t size,
-					 pebs_ovfl_callback_t ovfl,
-					 size_t th, unsigned int flags)
-{
-	return ds_request_pebs(task, 0, base, size, ovfl, th, flags);
-}
-
-struct pebs_tracer *ds_request_pebs_cpu(int cpu, void *base, size_t size,
-					pebs_ovfl_callback_t ovfl,
-					size_t th, unsigned int flags)
-{
-	return ds_request_pebs(NULL, cpu, base, size, ovfl, th, flags);
-}
-
-static void ds_free_bts(struct bts_tracer *tracer)
-{
-	struct task_struct *task;
-
-	task = tracer->ds.context->task;
-
-	WARN_ON_ONCE(tracer->ds.context->bts_master != tracer);
-	tracer->ds.context->bts_master = NULL;
-
-	/* Make sure tracing stopped and the tracer is not in use. */
-	if (task && (task != current))
-		wait_task_context_switch(task);
-
-	ds_put_context(tracer->ds.context);
-	put_tracer(task);
-
-	kfree(tracer);
-}
-
-void ds_release_bts(struct bts_tracer *tracer)
-{
-	might_sleep();
-
-	if (!tracer)
-		return;
-
-	ds_suspend_bts(tracer);
-	ds_free_bts(tracer);
-}
-
-int ds_release_bts_noirq(struct bts_tracer *tracer)
-{
-	struct task_struct *task;
-	unsigned long irq;
-	int error;
-
-	if (!tracer)
-		return 0;
-
-	task = tracer->ds.context->task;
-
-	local_irq_save(irq);
-
-	error = -EPERM;
-	if (!task &&
-	    (tracer->ds.context->cpu != smp_processor_id()))
-		goto out;
-
-	error = -EPERM;
-	if (task && (task != current))
-		goto out;
-
-	ds_suspend_bts_noirq(tracer);
-	ds_free_bts(tracer);
-
-	error = 0;
- out:
-	local_irq_restore(irq);
-	return error;
-}
-
-static void update_task_debugctlmsr(struct task_struct *task,
-				    unsigned long debugctlmsr)
-{
-	task->thread.debugctlmsr = debugctlmsr;
-
-	get_cpu();
-	if (task == current)
-		update_debugctlmsr(debugctlmsr);
-	put_cpu();
-}
-
-void ds_suspend_bts(struct bts_tracer *tracer)
-{
-	struct task_struct *task;
-	unsigned long debugctlmsr;
-	int cpu;
-
-	if (!tracer)
-		return;
-
-	tracer->flags = 0;
-
-	task = tracer->ds.context->task;
-	cpu  = tracer->ds.context->cpu;
-
-	WARN_ON(!task && irqs_disabled());
-
-	debugctlmsr = (task ?
-		       task->thread.debugctlmsr :
-		       get_debugctlmsr_on_cpu(cpu));
-	debugctlmsr &= ~BTS_CONTROL;
-
-	if (task)
-		update_task_debugctlmsr(task, debugctlmsr);
-	else
-		update_debugctlmsr_on_cpu(cpu, debugctlmsr);
-}
-
-int ds_suspend_bts_noirq(struct bts_tracer *tracer)
-{
-	struct task_struct *task;
-	unsigned long debugctlmsr, irq;
-	int cpu, error = 0;
-
-	if (!tracer)
-		return 0;
-
-	tracer->flags = 0;
-
-	task = tracer->ds.context->task;
-	cpu  = tracer->ds.context->cpu;
-
-	local_irq_save(irq);
-
-	error = -EPERM;
-	if (!task && (cpu != smp_processor_id()))
-		goto out;
-
-	debugctlmsr = (task ?
-		       task->thread.debugctlmsr :
-		       get_debugctlmsr());
-	debugctlmsr &= ~BTS_CONTROL;
-
-	if (task)
-		update_task_debugctlmsr(task, debugctlmsr);
-	else
-		update_debugctlmsr(debugctlmsr);
-
-	error = 0;
- out:
-	local_irq_restore(irq);
-	return error;
-}
-
-static unsigned long ds_bts_control(struct bts_tracer *tracer)
-{
-	unsigned long control;
-
-	control = ds_cfg.ctl[dsf_bts];
-	if (!(tracer->trace.ds.flags & BTS_KERNEL))
-		control |= ds_cfg.ctl[dsf_bts_kernel];
-	if (!(tracer->trace.ds.flags & BTS_USER))
-		control |= ds_cfg.ctl[dsf_bts_user];
-
-	return control;
-}
-
-void ds_resume_bts(struct bts_tracer *tracer)
-{
-	struct task_struct *task;
-	unsigned long debugctlmsr;
-	int cpu;
-
-	if (!tracer)
-		return;
-
-	tracer->flags = tracer->trace.ds.flags;
-
-	task = tracer->ds.context->task;
-	cpu  = tracer->ds.context->cpu;
-
-	WARN_ON(!task && irqs_disabled());
-
-	debugctlmsr = (task ?
-		       task->thread.debugctlmsr :
-		       get_debugctlmsr_on_cpu(cpu));
-	debugctlmsr |= ds_bts_control(tracer);
-
-	if (task)
-		update_task_debugctlmsr(task, debugctlmsr);
-	else
-		update_debugctlmsr_on_cpu(cpu, debugctlmsr);
-}
-
-int ds_resume_bts_noirq(struct bts_tracer *tracer)
-{
-	struct task_struct *task;
-	unsigned long debugctlmsr, irq;
-	int cpu, error = 0;
-
-	if (!tracer)
-		return 0;
-
-	tracer->flags = tracer->trace.ds.flags;
-
-	task = tracer->ds.context->task;
-	cpu  = tracer->ds.context->cpu;
-
-	local_irq_save(irq);
-
-	error = -EPERM;
-	if (!task && (cpu != smp_processor_id()))
-		goto out;
-
-	debugctlmsr = (task ?
-		       task->thread.debugctlmsr :
-		       get_debugctlmsr());
-	debugctlmsr |= ds_bts_control(tracer);
-
-	if (task)
-		update_task_debugctlmsr(task, debugctlmsr);
-	else
-		update_debugctlmsr(debugctlmsr);
-
-	error = 0;
- out:
-	local_irq_restore(irq);
-	return error;
-}
-
-static void ds_free_pebs(struct pebs_tracer *tracer)
-{
-	struct task_struct *task;
-
-	task = tracer->ds.context->task;
-
-	WARN_ON_ONCE(tracer->ds.context->pebs_master != tracer);
-	tracer->ds.context->pebs_master = NULL;
-
-	ds_put_context(tracer->ds.context);
-	put_tracer(task);
-
-	kfree(tracer);
-}
-
-void ds_release_pebs(struct pebs_tracer *tracer)
-{
-	might_sleep();
-
-	if (!tracer)
-		return;
-
-	ds_suspend_pebs(tracer);
-	ds_free_pebs(tracer);
-}
-
-int ds_release_pebs_noirq(struct pebs_tracer *tracer)
-{
-	struct task_struct *task;
-	unsigned long irq;
-	int error;
-
-	if (!tracer)
-		return 0;
-
-	task = tracer->ds.context->task;
-
-	local_irq_save(irq);
-
-	error = -EPERM;
-	if (!task &&
-	    (tracer->ds.context->cpu != smp_processor_id()))
-		goto out;
-
-	error = -EPERM;
-	if (task && (task != current))
-		goto out;
-
-	ds_suspend_pebs_noirq(tracer);
-	ds_free_pebs(tracer);
-
-	error = 0;
- out:
-	local_irq_restore(irq);
-	return error;
-}
-
-void ds_suspend_pebs(struct pebs_tracer *tracer)
-{
-
-}
-
-int ds_suspend_pebs_noirq(struct pebs_tracer *tracer)
-{
-	return 0;
-}
-
-void ds_resume_pebs(struct pebs_tracer *tracer)
-{
-
-}
-
-int ds_resume_pebs_noirq(struct pebs_tracer *tracer)
-{
-	return 0;
-}
-
-const struct bts_trace *ds_read_bts(struct bts_tracer *tracer)
-{
-	if (!tracer)
-		return NULL;
-
-	ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_bts);
-	return &tracer->trace;
-}
-
-const struct pebs_trace *ds_read_pebs(struct pebs_tracer *tracer)
-{
-	if (!tracer)
-		return NULL;
-
-	ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_pebs);
-
-	tracer->trace.counters = ds_cfg.nr_counter_reset;
-	memcpy(tracer->trace.counter_reset,
-	       tracer->ds.context->ds +
-	       (NUM_DS_PTR_FIELDS * ds_cfg.sizeof_ptr_field),
-	       ds_cfg.nr_counter_reset * PEBS_RESET_FIELD_SIZE);
-
-	return &tracer->trace;
-}
-
-int ds_reset_bts(struct bts_tracer *tracer)
-{
-	if (!tracer)
-		return -EINVAL;
-
-	tracer->trace.ds.top = tracer->trace.ds.begin;
-
-	ds_set(tracer->ds.context->ds, ds_bts, ds_index,
-	       (unsigned long)tracer->trace.ds.top);
-
-	return 0;
-}
-
-int ds_reset_pebs(struct pebs_tracer *tracer)
-{
-	if (!tracer)
-		return -EINVAL;
-
-	tracer->trace.ds.top = tracer->trace.ds.begin;
-
-	ds_set(tracer->ds.context->ds, ds_pebs, ds_index,
-	       (unsigned long)tracer->trace.ds.top);
-
-	return 0;
-}
-
-int ds_set_pebs_reset(struct pebs_tracer *tracer,
-		      unsigned int counter, u64 value)
-{
-	if (!tracer)
-		return -EINVAL;
-
-	if (ds_cfg.nr_counter_reset < counter)
-		return -EINVAL;
-
-	*(u64 *)(tracer->ds.context->ds +
-		 (NUM_DS_PTR_FIELDS * ds_cfg.sizeof_ptr_field) +
-		 (counter * PEBS_RESET_FIELD_SIZE)) = value;
-
-	return 0;
-}
-
-static const struct ds_configuration ds_cfg_netburst = {
-	.name = "Netburst",
-	.ctl[dsf_bts]		= (1 << 2) | (1 << 3),
-	.ctl[dsf_bts_kernel]	= (1 << 5),
-	.ctl[dsf_bts_user]	= (1 << 6),
-	.nr_counter_reset	= 1,
-};
-static const struct ds_configuration ds_cfg_pentium_m = {
-	.name = "Pentium M",
-	.ctl[dsf_bts]		= (1 << 6) | (1 << 7),
-	.nr_counter_reset	= 1,
-};
-static const struct ds_configuration ds_cfg_core2_atom = {
-	.name = "Core 2/Atom",
-	.ctl[dsf_bts]		= (1 << 6) | (1 << 7),
-	.ctl[dsf_bts_kernel]	= (1 << 9),
-	.ctl[dsf_bts_user]	= (1 << 10),
-	.nr_counter_reset	= 1,
-};
-static const struct ds_configuration ds_cfg_core_i7 = {
-	.name = "Core i7",
-	.ctl[dsf_bts]		= (1 << 6) | (1 << 7),
-	.ctl[dsf_bts_kernel]	= (1 << 9),
-	.ctl[dsf_bts_user]	= (1 << 10),
-	.nr_counter_reset	= 4,
-};
-
-static void
-ds_configure(const struct ds_configuration *cfg,
-	     struct cpuinfo_x86 *cpu)
-{
-	unsigned long nr_pebs_fields = 0;
-
-	printk(KERN_INFO "[ds] using %s configuration\n", cfg->name);
-
-#ifdef __i386__
-	nr_pebs_fields = 10;
-#else
-	nr_pebs_fields = 18;
-#endif
-
-	/*
-	 * Starting with version 2, architectural performance
-	 * monitoring supports a format specifier.
-	 */
-	if ((cpuid_eax(0xa) & 0xff) > 1) {
-		unsigned long perf_capabilities, format;
-
-		rdmsrl(MSR_IA32_PERF_CAPABILITIES, perf_capabilities);
-
-		format = (perf_capabilities >> 8) & 0xf;
-
-		switch (format) {
-		case 0:
-			nr_pebs_fields = 18;
-			break;
-		case 1:
-			nr_pebs_fields = 22;
-			break;
-		default:
-			printk(KERN_INFO
-			       "[ds] unknown PEBS format: %lu\n", format);
-			nr_pebs_fields = 0;
-			break;
-		}
-	}
-
-	memset(&ds_cfg, 0, sizeof(ds_cfg));
-	ds_cfg = *cfg;
-
-	ds_cfg.sizeof_ptr_field =
-		(cpu_has(cpu, X86_FEATURE_DTES64) ? 8 : 4);
-
-	ds_cfg.sizeof_rec[ds_bts]  = ds_cfg.sizeof_ptr_field * 3;
-	ds_cfg.sizeof_rec[ds_pebs] = ds_cfg.sizeof_ptr_field * nr_pebs_fields;
-
-	if (!cpu_has(cpu, X86_FEATURE_BTS)) {
-		ds_cfg.sizeof_rec[ds_bts] = 0;
-		printk(KERN_INFO "[ds] bts not available\n");
-	}
-	if (!cpu_has(cpu, X86_FEATURE_PEBS)) {
-		ds_cfg.sizeof_rec[ds_pebs] = 0;
-		printk(KERN_INFO "[ds] pebs not available\n");
-	}
-
-	printk(KERN_INFO "[ds] sizes: address: %u bit, ",
-	       8 * ds_cfg.sizeof_ptr_field);
-	printk("bts/pebs record: %u/%u bytes\n",
-	       ds_cfg.sizeof_rec[ds_bts], ds_cfg.sizeof_rec[ds_pebs]);
-
-	WARN_ON_ONCE(MAX_PEBS_COUNTERS < ds_cfg.nr_counter_reset);
-}
-
-void __cpuinit ds_init_intel(struct cpuinfo_x86 *c)
-{
-	/* Only configure the first cpu. Others are identical. */
-	if (ds_cfg.name)
-		return;
-
-	switch (c->x86) {
-	case 0x6:
-		switch (c->x86_model) {
-		case 0x9:
-		case 0xd: /* Pentium M */
-			ds_configure(&ds_cfg_pentium_m, c);
-			break;
-		case 0xf:
-		case 0x17: /* Core2 */
-		case 0x1c: /* Atom */
-			ds_configure(&ds_cfg_core2_atom, c);
-			break;
-		case 0x1a: /* Core i7 */
-			ds_configure(&ds_cfg_core_i7, c);
-			break;
-		default:
-			/* Sorry, don't know about them. */
-			break;
-		}
-		break;
-	case 0xf:
-		switch (c->x86_model) {
-		case 0x0:
-		case 0x1:
-		case 0x2: /* Netburst */
-			ds_configure(&ds_cfg_netburst, c);
-			break;
-		default:
-			/* Sorry, don't know about them. */
-			break;
-		}
-		break;
-	default:
-		/* Sorry, don't know about them. */
-		break;
-	}
-}
-
-static inline void ds_take_timestamp(struct ds_context *context,
-				     enum bts_qualifier qualifier,
-				     struct task_struct *task)
-{
-	struct bts_tracer *tracer = context->bts_master;
-	struct bts_struct ts;
-
-	/* Prevent compilers from reading the tracer pointer twice. */
-	barrier();
-
-	if (!tracer || !(tracer->flags & BTS_TIMESTAMPS))
-		return;
-
-	memset(&ts, 0, sizeof(ts));
-	ts.qualifier		= qualifier;
-	ts.variant.event.clock	= trace_clock_global();
-	ts.variant.event.pid	= task->pid;
-
-	bts_write(tracer, &ts);
-}
-
-/*
- * Change the DS configuration from tracing prev to tracing next.
- */
-void ds_switch_to(struct task_struct *prev, struct task_struct *next)
-{
-	struct ds_context *prev_ctx	= prev->thread.ds_ctx;
-	struct ds_context *next_ctx	= next->thread.ds_ctx;
-	unsigned long debugctlmsr	= next->thread.debugctlmsr;
-
-	/* Make sure all data is read before we start. */
-	barrier();
-
-	if (prev_ctx) {
-		update_debugctlmsr(0);
-
-		ds_take_timestamp(prev_ctx, bts_task_departs, prev);
-	}
-
-	if (next_ctx) {
-		ds_take_timestamp(next_ctx, bts_task_arrives, next);
-
-		wrmsrl(MSR_IA32_DS_AREA, (unsigned long)next_ctx->ds);
-	}
-
-	update_debugctlmsr(debugctlmsr);
-}
-
-static __init int ds_selftest(void)
-{
-	if (ds_cfg.sizeof_rec[ds_bts]) {
-		int error;
-
-		error = ds_selftest_bts();
-		if (error) {
-			WARN(1, "[ds] selftest failed. disabling bts.\n");
-			ds_cfg.sizeof_rec[ds_bts] = 0;
-		}
-	}
-
-	if (ds_cfg.sizeof_rec[ds_pebs]) {
-		int error;
-
-		error = ds_selftest_pebs();
-		if (error) {
-			WARN(1, "[ds] selftest failed. disabling pebs.\n");
-			ds_cfg.sizeof_rec[ds_pebs] = 0;
-		}
-	}
-
-	return 0;
-}
-device_initcall(ds_selftest);
diff --git a/arch/x86/kernel/ds_selftest.c b/arch/x86/kernel/ds_selftest.c
deleted file mode 100644
index 6bc7c199ab9..00000000000
--- a/arch/x86/kernel/ds_selftest.c
+++ /dev/null
@@ -1,408 +0,0 @@
-/*
- * Debug Store support - selftest
- *
- *
- * Copyright (C) 2009 Intel Corporation.
- * Markus Metzger <markus.t.metzger@intel.com>, 2009
- */
-
-#include "ds_selftest.h"
-
-#include <linux/kernel.h>
-#include <linux/string.h>
-#include <linux/smp.h>
-#include <linux/cpu.h>
-
-#include <asm/ds.h>
-
-
-#define BUFFER_SIZE		521	/* Intentionally chose an odd size. */
-#define SMALL_BUFFER_SIZE	24	/* A single bts entry. */
-
-struct ds_selftest_bts_conf {
-	struct bts_tracer *tracer;
-	int error;
-	int (*suspend)(struct bts_tracer *);
-	int (*resume)(struct bts_tracer *);
-};
-
-static int ds_selftest_bts_consistency(const struct bts_trace *trace)
-{
-	int error = 0;
-
-	if (!trace) {
-		printk(KERN_CONT "failed to access trace...");
-		/* Bail out. Other tests are pointless. */
-		return -1;
-	}
-
-	if (!trace->read) {
-		printk(KERN_CONT "bts read not available...");
-		error = -1;
-	}
-
-	/* Do some sanity checks on the trace configuration. */
-	if (!trace->ds.n) {
-		printk(KERN_CONT "empty bts buffer...");
-		error = -1;
-	}
-	if (!trace->ds.size) {
-		printk(KERN_CONT "bad bts trace setup...");
-		error = -1;
-	}
-	if (trace->ds.end !=
-	    (char *)trace->ds.begin + (trace->ds.n * trace->ds.size)) {
-		printk(KERN_CONT "bad bts buffer setup...");
-		error = -1;
-	}
-	/*
-	 * We allow top in [begin; end], since its not clear when the
-	 * overflow adjustment happens: after the increment or before the
-	 * write.
-	 */
-	if ((trace->ds.top < trace->ds.begin) ||
-	    (trace->ds.end < trace->ds.top)) {
-		printk(KERN_CONT "bts top out of bounds...");
-		error = -1;
-	}
-
-	return error;
-}
-
-static int ds_selftest_bts_read(struct bts_tracer *tracer,
-				const struct bts_trace *trace,
-				const void *from, const void *to)
-{
-	const unsigned char *at;
-
-	/*
-	 * Check a few things which do not belong to this test.
-	 * They should be covered by other tests.
-	 */
-	if (!trace)
-		return -1;
-
-	if (!trace->read)
-		return -1;
-
-	if (to < from)
-		return -1;
-
-	if (from < trace->ds.begin)
-		return -1;
-
-	if (trace->ds.end < to)
-		return -1;
-
-	if (!trace->ds.size)
-		return -1;
-
-	/* Now to the test itself. */
-	for (at = from; (void *)at < to; at += trace->ds.size) {
-		struct bts_struct bts;
-		unsigned long index;
-		int error;
-
-		if (((void *)at - trace->ds.begin) % trace->ds.size) {
-			printk(KERN_CONT
-			       "read from non-integer index...");
-			return -1;
-		}
-		index = ((void *)at - trace->ds.begin) / trace->ds.size;
-
-		memset(&bts, 0, sizeof(bts));
-		error = trace->read(tracer, at, &bts);
-		if (error < 0) {
-			printk(KERN_CONT
-			       "error reading bts trace at [%lu] (0x%p)...",
-			       index, at);
-			return error;
-		}
-
-		switch (bts.qualifier) {
-		case BTS_BRANCH:
-			break;
-		default:
-			printk(KERN_CONT
-			       "unexpected bts entry %llu at [%lu] (0x%p)...",
-			       bts.qualifier, index, at);
-			return -1;
-		}
-	}
-
-	return 0;
-}
-
-static void ds_selftest_bts_cpu(void *arg)
-{
-	struct ds_selftest_bts_conf *conf = arg;
-	const struct bts_trace *trace;
-	void *top;
-
-	if (IS_ERR(conf->tracer)) {
-		conf->error = PTR_ERR(conf->tracer);
-		conf->tracer = NULL;
-
-		printk(KERN_CONT
-		       "initialization failed (err: %d)...", conf->error);
-		return;
-	}
-
-	/* We should meanwhile have enough trace. */
-	conf->error = conf->suspend(conf->tracer);
-	if (conf->error < 0)
-		return;
-
-	/* Let's see if we can access the trace. */
-	trace = ds_read_bts(conf->tracer);
-
-	conf->error = ds_selftest_bts_consistency(trace);
-	if (conf->error < 0)
-		return;
-
-	/* If everything went well, we should have a few trace entries. */
-	if (trace->ds.top == trace->ds.begin) {
-		/*
-		 * It is possible but highly unlikely that we got a
-		 * buffer overflow and end up at exactly the same
-		 * position we started from.
-		 * Let's issue a warning, but continue.
-		 */
-		printk(KERN_CONT "no trace/overflow...");
-	}
-
-	/* Let's try to read the trace we collected. */
-	conf->error =
-		ds_selftest_bts_read(conf->tracer, trace,
-				     trace->ds.begin, trace->ds.top);
-	if (conf->error < 0)
-		return;
-
-	/*
-	 * Let's read the trace again.
-	 * Since we suspended tracing, we should get the same result.
-	 */
-	top = trace->ds.top;
-
-	trace = ds_read_bts(conf->tracer);
-	conf->error = ds_selftest_bts_consistency(trace);
-	if (conf->error < 0)
-		return;
-
-	if (top != trace->ds.top) {
-		printk(KERN_CONT "suspend not working...");
-		conf->error = -1;
-		return;
-	}
-
-	/* Let's collect some more trace - see if resume is working. */
-	conf->error = conf->resume(conf->tracer);
-	if (conf->error < 0)
-		return;
-
-	conf->error = conf->suspend(conf->tracer);
-	if (conf->error < 0)
-		return;
-
-	trace = ds_read_bts(conf->tracer);
-
-	conf->error = ds_selftest_bts_consistency(trace);
-	if (conf->error < 0)
-		return;
-
-	if (trace->ds.top == top) {
-		/*
-		 * It is possible but highly unlikely that we got a
-		 * buffer overflow and end up at exactly the same
-		 * position we started from.
-		 * Let's issue a warning and check the full trace.
-		 */
-		printk(KERN_CONT
-		       "no resume progress/overflow...");
-
-		conf->error =
-			ds_selftest_bts_read(conf->tracer, trace,
-					     trace->ds.begin, trace->ds.end);
-	} else if (trace->ds.top < top) {
-		/*
-		 * We had a buffer overflow - the entire buffer should
-		 * contain trace records.
-		 */
-		conf->error =
-			ds_selftest_bts_read(conf->tracer, trace,
-					     trace->ds.begin, trace->ds.end);
-	} else {
-		/*
-		 * It is quite likely that the buffer did not overflow.
-		 * Let's just check the delta trace.
-		 */
-		conf->error =
-			ds_selftest_bts_read(conf->tracer, trace, top,
-					     trace->ds.top);
-	}
-	if (conf->error < 0)
-		return;
-
-	conf->error = 0;
-}
-
-static int ds_suspend_bts_wrap(struct bts_tracer *tracer)
-{
-	ds_suspend_bts(tracer);
-	return 0;
-}
-
-static int ds_resume_bts_wrap(struct bts_tracer *tracer)
-{
-	ds_resume_bts(tracer);
-	return 0;
-}
-
-static void ds_release_bts_noirq_wrap(void *tracer)
-{
-	(void)ds_release_bts_noirq(tracer);
-}
-
-static int ds_selftest_bts_bad_release_noirq(int cpu,
-					     struct bts_tracer *tracer)
-{
-	int error = -EPERM;
-
-	/* Try to release the tracer on the wrong cpu. */
-	get_cpu();
-	if (cpu != smp_processor_id()) {
-		error = ds_release_bts_noirq(tracer);
-		if (error != -EPERM)
-			printk(KERN_CONT "release on wrong cpu...");
-	}
-	put_cpu();
-
-	return error ? 0 : -1;
-}
-
-static int ds_selftest_bts_bad_request_cpu(int cpu, void *buffer)
-{
-	struct bts_tracer *tracer;
-	int error;
-
-	/* Try to request cpu tracing while task tracing is active. */
-	tracer = ds_request_bts_cpu(cpu, buffer, BUFFER_SIZE, NULL,
-				    (size_t)-1, BTS_KERNEL);
-	error = PTR_ERR(tracer);
-	if (!IS_ERR(tracer)) {
-		ds_release_bts(tracer);
-		error = 0;
-	}
-
-	if (error != -EPERM)
-		printk(KERN_CONT "cpu/task tracing overlap...");
-
-	return error ? 0 : -1;
-}
-
-static int ds_selftest_bts_bad_request_task(void *buffer)
-{
-	struct bts_tracer *tracer;
-	int error;
-
-	/* Try to request cpu tracing while task tracing is active. */
-	tracer = ds_request_bts_task(current, buffer, BUFFER_SIZE, NULL,
-				    (size_t)-1, BTS_KERNEL);
-	error = PTR_ERR(tracer);
-	if (!IS_ERR(tracer)) {
-		error = 0;
-		ds_release_bts(tracer);
-	}
-
-	if (error != -EPERM)
-		printk(KERN_CONT "task/cpu tracing overlap...");
-
-	return error ? 0 : -1;
-}
-
-int ds_selftest_bts(void)
-{
-	struct ds_selftest_bts_conf conf;
-	unsigned char buffer[BUFFER_SIZE], *small_buffer;
-	unsigned long irq;
-	int cpu;
-
-	printk(KERN_INFO "[ds] bts selftest...");
-	conf.error = 0;
-
-	small_buffer = (unsigned char *)ALIGN((unsigned long)buffer, 8) + 8;
-
-	get_online_cpus();
-	for_each_online_cpu(cpu) {
-		conf.suspend = ds_suspend_bts_wrap;
-		conf.resume = ds_resume_bts_wrap;
-		conf.tracer =
-			ds_request_bts_cpu(cpu, buffer, BUFFER_SIZE,
-					   NULL, (size_t)-1, BTS_KERNEL);
-		ds_selftest_bts_cpu(&conf);
-		if (conf.error >= 0)
-			conf.error = ds_selftest_bts_bad_request_task(buffer);
-		ds_release_bts(conf.tracer);
-		if (conf.error < 0)
-			goto out;
-
-		conf.suspend = ds_suspend_bts_noirq;
-		conf.resume = ds_resume_bts_noirq;
-		conf.tracer =
-			ds_request_bts_cpu(cpu, buffer, BUFFER_SIZE,
-					   NULL, (size_t)-1, BTS_KERNEL);
-		smp_call_function_single(cpu, ds_selftest_bts_cpu, &conf, 1);
-		if (conf.error >= 0) {
-			conf.error =
-				ds_selftest_bts_bad_release_noirq(cpu,
-								  conf.tracer);
-			/* We must not release the tracer twice. */
-			if (conf.error < 0)
-				conf.tracer = NULL;
-		}
-		if (conf.error >= 0)
-			conf.error = ds_selftest_bts_bad_request_task(buffer);
-		smp_call_function_single(cpu, ds_release_bts_noirq_wrap,
-					 conf.tracer, 1);
-		if (conf.error < 0)
-			goto out;
-	}
-
-	conf.suspend = ds_suspend_bts_wrap;
-	conf.resume = ds_resume_bts_wrap;
-	conf.tracer =
-		ds_request_bts_task(current, buffer, BUFFER_SIZE,
-				    NULL, (size_t)-1, BTS_KERNEL);
-	ds_selftest_bts_cpu(&conf);
-	if (conf.error >= 0)
-		conf.error = ds_selftest_bts_bad_request_cpu(0, buffer);
-	ds_release_bts(conf.tracer);
-	if (conf.error < 0)
-		goto out;
-
-	conf.suspend = ds_suspend_bts_noirq;
-	conf.resume = ds_resume_bts_noirq;
-	conf.tracer =
-		ds_request_bts_task(current, small_buffer, SMALL_BUFFER_SIZE,
-				   NULL, (size_t)-1, BTS_KERNEL);
-	local_irq_save(irq);
-	ds_selftest_bts_cpu(&conf);
-	if (conf.error >= 0)
-		conf.error = ds_selftest_bts_bad_request_cpu(0, buffer);
-	ds_release_bts_noirq(conf.tracer);
-	local_irq_restore(irq);
-	if (conf.error < 0)
-		goto out;
-
-	conf.error = 0;
- out:
-	put_online_cpus();
-	printk(KERN_CONT "%s.\n", (conf.error ? "failed" : "passed"));
-
-	return conf.error;
-}
-
-int ds_selftest_pebs(void)
-{
-	return 0;
-}
diff --git a/arch/x86/kernel/ds_selftest.h b/arch/x86/kernel/ds_selftest.h
deleted file mode 100644
index 2ba8745c666..00000000000
--- a/arch/x86/kernel/ds_selftest.h
+++ /dev/null
@@ -1,15 +0,0 @@
-/*
- * Debug Store support - selftest
- *
- *
- * Copyright (C) 2009 Intel Corporation.
- * Markus Metzger <markus.t.metzger@intel.com>, 2009
- */
-
-#ifdef CONFIG_X86_DS_SELFTEST
-extern int ds_selftest_bts(void);
-extern int ds_selftest_pebs(void);
-#else
-static inline int ds_selftest_bts(void) { return 0; }
-static inline int ds_selftest_pebs(void) { return 0; }
-#endif
diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c
index 6d817554780..c89a386930b 100644
--- a/arch/x86/kernel/dumpstack.c
+++ b/arch/x86/kernel/dumpstack.c
@@ -224,11 +224,6 @@ unsigned __kprobes long oops_begin(void)
 	int cpu;
 	unsigned long flags;
 
-	/* notify the hw-branch tracer so it may disable tracing and
-	   add the last trace to the trace buffer -
-	   the earlier this happens, the more useful the trace. */
-	trace_hw_branch_oops();
-
 	oops_enter();
 
 	/* racy, but better than risking deadlock. */
diff --git a/arch/x86/kernel/early_printk.c b/arch/x86/kernel/early_printk.c
index b9c830c12b4..fa99bae75ac 100644
--- a/arch/x86/kernel/early_printk.c
+++ b/arch/x86/kernel/early_printk.c
@@ -41,6 +41,14 @@ static void early_vga_write(struct console *con, const char *str, unsigned n)
 				writew(0x720, VGABASE + 2*(max_xpos*j + i));
 			current_ypos = max_ypos-1;
 		}
+#ifdef CONFIG_KGDB_KDB
+		if (c == '\b') {
+			if (current_xpos > 0)
+				current_xpos--;
+		} else if (c == '\r') {
+			current_xpos = 0;
+		} else
+#endif
 		if (c == '\n') {
 			current_xpos = 0;
 			current_ypos++;
diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S
index 44a8e0dc673..cd49141cf15 100644
--- a/arch/x86/kernel/entry_32.S
+++ b/arch/x86/kernel/entry_32.S
@@ -53,6 +53,7 @@
 #include <asm/processor-flags.h>
 #include <asm/ftrace.h>
 #include <asm/irq_vectors.h>
+#include <asm/cpufeature.h>
 
 /* Avoid __ASSEMBLER__'ifying <linux/audit.h> just for this.  */
 #include <linux/elf-em.h>
@@ -905,7 +906,25 @@ ENTRY(simd_coprocessor_error)
 	RING0_INT_FRAME
 	pushl $0
 	CFI_ADJUST_CFA_OFFSET 4
+#ifdef CONFIG_X86_INVD_BUG
+	/* AMD 486 bug: invd from userspace calls exception 19 instead of #GP */
+661:	pushl $do_general_protection
+662:
+.section .altinstructions,"a"
+	.balign 4
+	.long 661b
+	.long 663f
+	.byte X86_FEATURE_XMM
+	.byte 662b-661b
+	.byte 664f-663f
+.previous
+.section .altinstr_replacement,"ax"
+663:	pushl $do_simd_coprocessor_error
+664:
+.previous
+#else
 	pushl $do_simd_coprocessor_error
+#endif
 	CFI_ADJUST_CFA_OFFSET 4
 	jmp error_code
 	CFI_ENDPROC
diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index 23b4ecdffa9..a198b7c87a1 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -36,6 +36,7 @@
 unsigned long				hpet_address;
 u8					hpet_blockid; /* OS timer block num */
 u8					hpet_msi_disable;
+u8					hpet_readback_cmp;
 
 #ifdef CONFIG_PCI_MSI
 static unsigned long			hpet_num_timers;
@@ -395,19 +396,23 @@ static int hpet_next_event(unsigned long delta,
 	 * at that point and we would wait for the next hpet interrupt
 	 * forever. We found out that reading the CMP register back
 	 * forces the transfer so we can rely on the comparison with
-	 * the counter register below. If the read back from the
-	 * compare register does not match the value we programmed
-	 * then we might have a real hardware problem. We can not do
-	 * much about it here, but at least alert the user/admin with
-	 * a prominent warning.
-	 * An erratum on some chipsets (ICH9,..), results in comparator read
-	 * immediately following a write returning old value. Workaround
-	 * for this is to read this value second time, when first
-	 * read returns old value.
+	 * the counter register below.
+	 *
+	 * That works fine on those ATI chipsets, but on newer Intel
+	 * chipsets (ICH9...) this triggers due to an erratum: Reading
+	 * the comparator immediately following a write is returning
+	 * the old value.
+	 *
+	 * We restrict the read back to the affected ATI chipsets (set
+	 * by quirks) and also run it with hpet=verbose for debugging
+	 * purposes.
 	 */
-	if (unlikely((u32)hpet_readl(HPET_Tn_CMP(timer)) != cnt)) {
-		WARN_ONCE(hpet_readl(HPET_Tn_CMP(timer)) != cnt,
-		  KERN_WARNING "hpet: compare register read back failed.\n");
+	if (hpet_readback_cmp || hpet_verbose) {
+		u32 cmp = hpet_readl(HPET_Tn_CMP(timer));
+
+		if (cmp != cnt)
+			printk_once(KERN_WARNING
+			    "hpet: compare register read back failed.\n");
 	}
 
 	return (s32)(hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0;
diff --git a/arch/x86/kernel/hw_breakpoint.c b/arch/x86/kernel/hw_breakpoint.c
index d6cc065f519..a8f1b803d2f 100644
--- a/arch/x86/kernel/hw_breakpoint.c
+++ b/arch/x86/kernel/hw_breakpoint.c
@@ -189,25 +189,16 @@ static int get_hbp_len(u8 hbp_len)
 }
 
 /*
- * Check for virtual address in user space.
- */
-int arch_check_va_in_userspace(unsigned long va, u8 hbp_len)
-{
-	unsigned int len;
-
-	len = get_hbp_len(hbp_len);
-
-	return (va <= TASK_SIZE - len);
-}
-
-/*
  * Check for virtual address in kernel space.
  */
-static int arch_check_va_in_kernelspace(unsigned long va, u8 hbp_len)
+int arch_check_bp_in_kernelspace(struct perf_event *bp)
 {
 	unsigned int len;
+	unsigned long va;
+	struct arch_hw_breakpoint *info = counter_arch_bp(bp);
 
-	len = get_hbp_len(hbp_len);
+	va = info->address;
+	len = get_hbp_len(info->len);
 
 	return (va >= TASK_SIZE) && ((va + len - 1) >= TASK_SIZE);
 }
@@ -300,8 +291,7 @@ static int arch_build_bp_info(struct perf_event *bp)
 /*
  * Validate the arch-specific HW Breakpoint register settings
  */
-int arch_validate_hwbkpt_settings(struct perf_event *bp,
-				  struct task_struct *tsk)
+int arch_validate_hwbkpt_settings(struct perf_event *bp)
 {
 	struct arch_hw_breakpoint *info = counter_arch_bp(bp);
 	unsigned int align;
@@ -314,16 +304,6 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp,
 
 	ret = -EINVAL;
 
-	if (info->type == X86_BREAKPOINT_EXECUTE)
-		/*
-		 * Ptrace-refactoring code
-		 * For now, we'll allow instruction breakpoint only for user-space
-		 * addresses
-		 */
-		if ((!arch_check_va_in_userspace(info->address, info->len)) &&
-			info->len != X86_BREAKPOINT_EXECUTE)
-			return ret;
-
 	switch (info->len) {
 	case X86_BREAKPOINT_LEN_1:
 		align = 0;
@@ -350,15 +330,6 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp,
 	if (info->address & align)
 		return -EINVAL;
 
-	/* Check that the virtual address is in the proper range */
-	if (tsk) {
-		if (!arch_check_va_in_userspace(info->address, info->len))
-			return -EFAULT;
-	} else {
-		if (!arch_check_va_in_kernelspace(info->address, info->len))
-			return -EFAULT;
-	}
-
 	return 0;
 }
 
diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c
index 54c31c28548..86cef6b3225 100644
--- a/arch/x86/kernel/i387.c
+++ b/arch/x86/kernel/i387.c
@@ -102,65 +102,62 @@ void __cpuinit fpu_init(void)
 
 	mxcsr_feature_mask_init();
 	/* clean state in init */
-	if (cpu_has_xsave)
-		current_thread_info()->status = TS_XSAVE;
-	else
-		current_thread_info()->status = 0;
+	current_thread_info()->status = 0;
 	clear_used_math();
 }
 #endif	/* CONFIG_X86_64 */
 
-/*
- * The _current_ task is using the FPU for the first time
- * so initialize it and set the mxcsr to its default
- * value at reset if we support XMM instructions and then
- * remeber the current task has used the FPU.
- */
-int init_fpu(struct task_struct *tsk)
+static void fpu_finit(struct fpu *fpu)
 {
-	if (tsk_used_math(tsk)) {
-		if (HAVE_HWFP && tsk == current)
-			unlazy_fpu(tsk);
-		return 0;
-	}
-
-	/*
-	 * Memory allocation at the first usage of the FPU and other state.
-	 */
-	if (!tsk->thread.xstate) {
-		tsk->thread.xstate = kmem_cache_alloc(task_xstate_cachep,
-						      GFP_KERNEL);
-		if (!tsk->thread.xstate)
-			return -ENOMEM;
-	}
-
 #ifdef CONFIG_X86_32
 	if (!HAVE_HWFP) {
-		memset(tsk->thread.xstate, 0, xstate_size);
-		finit_task(tsk);
-		set_stopped_child_used_math(tsk);
-		return 0;
+		finit_soft_fpu(&fpu->state->soft);
+		return;
 	}
 #endif
 
 	if (cpu_has_fxsr) {
-		struct i387_fxsave_struct *fx = &tsk->thread.xstate->fxsave;
+		struct i387_fxsave_struct *fx = &fpu->state->fxsave;
 
 		memset(fx, 0, xstate_size);
 		fx->cwd = 0x37f;
 		if (cpu_has_xmm)
 			fx->mxcsr = MXCSR_DEFAULT;
 	} else {
-		struct i387_fsave_struct *fp = &tsk->thread.xstate->fsave;
+		struct i387_fsave_struct *fp = &fpu->state->fsave;
 		memset(fp, 0, xstate_size);
 		fp->cwd = 0xffff037fu;
 		fp->swd = 0xffff0000u;
 		fp->twd = 0xffffffffu;
 		fp->fos = 0xffff0000u;
 	}
+}
+
+/*
+ * The _current_ task is using the FPU for the first time
+ * so initialize it and set the mxcsr to its default
+ * value at reset if we support XMM instructions and then
+ * remeber the current task has used the FPU.
+ */
+int init_fpu(struct task_struct *tsk)
+{
+	int ret;
+
+	if (tsk_used_math(tsk)) {
+		if (HAVE_HWFP && tsk == current)
+			unlazy_fpu(tsk);
+		return 0;
+	}
+
 	/*
-	 * Only the device not available exception or ptrace can call init_fpu.
+	 * Memory allocation at the first usage of the FPU and other state.
 	 */
+	ret = fpu_alloc(&tsk->thread.fpu);
+	if (ret)
+		return ret;
+
+	fpu_finit(&tsk->thread.fpu);
+
 	set_stopped_child_used_math(tsk);
 	return 0;
 }
@@ -194,7 +191,7 @@ int xfpregs_get(struct task_struct *target, const struct user_regset *regset,
 		return ret;
 
 	return user_regset_copyout(&pos, &count, &kbuf, &ubuf,
-				   &target->thread.xstate->fxsave, 0, -1);
+				   &target->thread.fpu.state->fxsave, 0, -1);
 }
 
 int xfpregs_set(struct task_struct *target, const struct user_regset *regset,
@@ -211,19 +208,19 @@ int xfpregs_set(struct task_struct *target, const struct user_regset *regset,
 		return ret;
 
 	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf,
-				 &target->thread.xstate->fxsave, 0, -1);
+				 &target->thread.fpu.state->fxsave, 0, -1);
 
 	/*
 	 * mxcsr reserved bits must be masked to zero for security reasons.
 	 */
-	target->thread.xstate->fxsave.mxcsr &= mxcsr_feature_mask;
+	target->thread.fpu.state->fxsave.mxcsr &= mxcsr_feature_mask;
 
 	/*
 	 * update the header bits in the xsave header, indicating the
 	 * presence of FP and SSE state.
 	 */
 	if (cpu_has_xsave)
-		target->thread.xstate->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE;
+		target->thread.fpu.state->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE;
 
 	return ret;
 }
@@ -246,14 +243,14 @@ int xstateregs_get(struct task_struct *target, const struct user_regset *regset,
 	 * memory layout in the thread struct, so that we can copy the entire
 	 * xstateregs to the user using one user_regset_copyout().
 	 */
-	memcpy(&target->thread.xstate->fxsave.sw_reserved,
+	memcpy(&target->thread.fpu.state->fxsave.sw_reserved,
 	       xstate_fx_sw_bytes, sizeof(xstate_fx_sw_bytes));
 
 	/*
 	 * Copy the xstate memory layout.
 	 */
 	ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf,
-				  &target->thread.xstate->xsave, 0, -1);
+				  &target->thread.fpu.state->xsave, 0, -1);
 	return ret;
 }
 
@@ -272,14 +269,14 @@ int xstateregs_set(struct task_struct *target, const struct user_regset *regset,
 		return ret;
 
 	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf,
-				 &target->thread.xstate->xsave, 0, -1);
+				 &target->thread.fpu.state->xsave, 0, -1);
 
 	/*
 	 * mxcsr reserved bits must be masked to zero for security reasons.
 	 */
-	target->thread.xstate->fxsave.mxcsr &= mxcsr_feature_mask;
+	target->thread.fpu.state->fxsave.mxcsr &= mxcsr_feature_mask;
 
-	xsave_hdr = &target->thread.xstate->xsave.xsave_hdr;
+	xsave_hdr = &target->thread.fpu.state->xsave.xsave_hdr;
 
 	xsave_hdr->xstate_bv &= pcntxt_mask;
 	/*
@@ -365,7 +362,7 @@ static inline u32 twd_fxsr_to_i387(struct i387_fxsave_struct *fxsave)
 static void
 convert_from_fxsr(struct user_i387_ia32_struct *env, struct task_struct *tsk)
 {
-	struct i387_fxsave_struct *fxsave = &tsk->thread.xstate->fxsave;
+	struct i387_fxsave_struct *fxsave = &tsk->thread.fpu.state->fxsave;
 	struct _fpreg *to = (struct _fpreg *) &env->st_space[0];
 	struct _fpxreg *from = (struct _fpxreg *) &fxsave->st_space[0];
 	int i;
@@ -405,7 +402,7 @@ static void convert_to_fxsr(struct task_struct *tsk,
 			    const struct user_i387_ia32_struct *env)
 
 {
-	struct i387_fxsave_struct *fxsave = &tsk->thread.xstate->fxsave;
+	struct i387_fxsave_struct *fxsave = &tsk->thread.fpu.state->fxsave;
 	struct _fpreg *from = (struct _fpreg *) &env->st_space[0];
 	struct _fpxreg *to = (struct _fpxreg *) &fxsave->st_space[0];
 	int i;
@@ -445,7 +442,7 @@ int fpregs_get(struct task_struct *target, const struct user_regset *regset,
 
 	if (!cpu_has_fxsr) {
 		return user_regset_copyout(&pos, &count, &kbuf, &ubuf,
-					   &target->thread.xstate->fsave, 0,
+					   &target->thread.fpu.state->fsave, 0,
 					   -1);
 	}
 
@@ -475,7 +472,7 @@ int fpregs_set(struct task_struct *target, const struct user_regset *regset,
 
 	if (!cpu_has_fxsr) {
 		return user_regset_copyin(&pos, &count, &kbuf, &ubuf,
-					  &target->thread.xstate->fsave, 0, -1);
+					  &target->thread.fpu.state->fsave, 0, -1);
 	}
 
 	if (pos > 0 || count < sizeof(env))
@@ -490,7 +487,7 @@ int fpregs_set(struct task_struct *target, const struct user_regset *regset,
 	 * presence of FP.
 	 */
 	if (cpu_has_xsave)
-		target->thread.xstate->xsave.xsave_hdr.xstate_bv |= XSTATE_FP;
+		target->thread.fpu.state->xsave.xsave_hdr.xstate_bv |= XSTATE_FP;
 	return ret;
 }
 
@@ -501,7 +498,7 @@ int fpregs_set(struct task_struct *target, const struct user_regset *regset,
 static inline int save_i387_fsave(struct _fpstate_ia32 __user *buf)
 {
 	struct task_struct *tsk = current;
-	struct i387_fsave_struct *fp = &tsk->thread.xstate->fsave;
+	struct i387_fsave_struct *fp = &tsk->thread.fpu.state->fsave;
 
 	fp->status = fp->swd;
 	if (__copy_to_user(buf, fp, sizeof(struct i387_fsave_struct)))
@@ -512,7 +509,7 @@ static inline int save_i387_fsave(struct _fpstate_ia32 __user *buf)
 static int save_i387_fxsave(struct _fpstate_ia32 __user *buf)
 {
 	struct task_struct *tsk = current;
-	struct i387_fxsave_struct *fx = &tsk->thread.xstate->fxsave;
+	struct i387_fxsave_struct *fx = &tsk->thread.fpu.state->fxsave;
 	struct user_i387_ia32_struct env;
 	int err = 0;
 
@@ -547,7 +544,7 @@ static int save_i387_xsave(void __user *buf)
 	 * header as well as change any contents in the memory layout.
 	 * xrestore as part of sigreturn will capture all the changes.
 	 */
-	tsk->thread.xstate->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE;
+	tsk->thread.fpu.state->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE;
 
 	if (save_i387_fxsave(fx) < 0)
 		return -1;
@@ -599,7 +596,7 @@ static inline int restore_i387_fsave(struct _fpstate_ia32 __user *buf)
 {
 	struct task_struct *tsk = current;
 
-	return __copy_from_user(&tsk->thread.xstate->fsave, buf,
+	return __copy_from_user(&tsk->thread.fpu.state->fsave, buf,
 				sizeof(struct i387_fsave_struct));
 }
 
@@ -610,10 +607,10 @@ static int restore_i387_fxsave(struct _fpstate_ia32 __user *buf,
 	struct user_i387_ia32_struct env;
 	int err;
 
-	err = __copy_from_user(&tsk->thread.xstate->fxsave, &buf->_fxsr_env[0],
+	err = __copy_from_user(&tsk->thread.fpu.state->fxsave, &buf->_fxsr_env[0],
 			       size);
 	/* mxcsr reserved bits must be masked to zero for security reasons */
-	tsk->thread.xstate->fxsave.mxcsr &= mxcsr_feature_mask;
+	tsk->thread.fpu.state->fxsave.mxcsr &= mxcsr_feature_mask;
 	if (err || __copy_from_user(&env, buf, sizeof(env)))
 		return 1;
 	convert_to_fxsr(tsk, &env);
@@ -629,7 +626,7 @@ static int restore_i387_xsave(void __user *buf)
 	struct i387_fxsave_struct __user *fx =
 		(struct i387_fxsave_struct __user *) &fx_user->_fxsr_env[0];
 	struct xsave_hdr_struct *xsave_hdr =
-				&current->thread.xstate->xsave.xsave_hdr;
+				&current->thread.fpu.state->xsave.xsave_hdr;
 	u64 mask;
 	int err;
 
diff --git a/arch/x86/kernel/i8253.c b/arch/x86/kernel/i8253.c
index 23c167925a5..2dfd3159744 100644
--- a/arch/x86/kernel/i8253.c
+++ b/arch/x86/kernel/i8253.c
@@ -16,7 +16,7 @@
 #include <asm/hpet.h>
 #include <asm/smp.h>
 
-DEFINE_SPINLOCK(i8253_lock);
+DEFINE_RAW_SPINLOCK(i8253_lock);
 EXPORT_SYMBOL(i8253_lock);
 
 /*
@@ -33,7 +33,7 @@ struct clock_event_device *global_clock_event;
 static void init_pit_timer(enum clock_event_mode mode,
 			   struct clock_event_device *evt)
 {
-	spin_lock(&i8253_lock);
+	raw_spin_lock(&i8253_lock);
 
 	switch (mode) {
 	case CLOCK_EVT_MODE_PERIODIC:
@@ -62,7 +62,7 @@ static void init_pit_timer(enum clock_event_mode mode,
 		/* Nothing to do here */
 		break;
 	}
-	spin_unlock(&i8253_lock);
+	raw_spin_unlock(&i8253_lock);
 }
 
 /*
@@ -72,10 +72,10 @@ static void init_pit_timer(enum clock_event_mode mode,
  */
 static int pit_next_event(unsigned long delta, struct clock_event_device *evt)
 {
-	spin_lock(&i8253_lock);
+	raw_spin_lock(&i8253_lock);
 	outb_pit(delta & 0xff , PIT_CH0);	/* LSB */
 	outb_pit(delta >> 8 , PIT_CH0);		/* MSB */
-	spin_unlock(&i8253_lock);
+	raw_spin_unlock(&i8253_lock);
 
 	return 0;
 }
@@ -130,7 +130,7 @@ static cycle_t pit_read(struct clocksource *cs)
 	int count;
 	u32 jifs;
 
-	spin_lock_irqsave(&i8253_lock, flags);
+	raw_spin_lock_irqsave(&i8253_lock, flags);
 	/*
 	 * Although our caller may have the read side of xtime_lock,
 	 * this is now a seqlock, and we are cheating in this routine
@@ -176,7 +176,7 @@ static cycle_t pit_read(struct clocksource *cs)
 	old_count = count;
 	old_jifs = jifs;
 
-	spin_unlock_irqrestore(&i8253_lock, flags);
+	raw_spin_unlock_irqrestore(&i8253_lock, flags);
 
 	count = (LATCH - 1) - count;
 
diff --git a/arch/x86/kernel/irqinit.c b/arch/x86/kernel/irqinit.c
index 0ed2d300cd4..990ae7cfc57 100644
--- a/arch/x86/kernel/irqinit.c
+++ b/arch/x86/kernel/irqinit.c
@@ -60,7 +60,7 @@ static irqreturn_t math_error_irq(int cpl, void *dev_id)
 	outb(0, 0xF0);
 	if (ignore_fpu_irq || !boot_cpu_data.hard_math)
 		return IRQ_NONE;
-	math_error((void __user *)get_irq_regs()->ip);
+	math_error(get_irq_regs(), 0, 16);
 	return IRQ_HANDLED;
 }
 
diff --git a/arch/x86/kernel/kgdb.c b/arch/x86/kernel/kgdb.c
index b2258ca9100..4f4af75b948 100644
--- a/arch/x86/kernel/kgdb.c
+++ b/arch/x86/kernel/kgdb.c
@@ -47,20 +47,8 @@
 #include <asm/debugreg.h>
 #include <asm/apicdef.h>
 #include <asm/system.h>
-
 #include <asm/apic.h>
 
-/*
- * Put the error code here just in case the user cares:
- */
-static int gdb_x86errcode;
-
-/*
- * Likewise, the vector number here (since GDB only gets the signal
- * number through the usual means, and that's not very specific):
- */
-static int gdb_x86vector = -1;
-
 /**
  *	pt_regs_to_gdb_regs - Convert ptrace regs to GDB regs
  *	@gdb_regs: A pointer to hold the registers in the order GDB wants.
@@ -211,6 +199,8 @@ static struct hw_breakpoint {
 	struct perf_event	**pev;
 } breakinfo[4];
 
+static unsigned long early_dr7;
+
 static void kgdb_correct_hw_break(void)
 {
 	int breakno;
@@ -222,6 +212,14 @@ static void kgdb_correct_hw_break(void)
 		int cpu = raw_smp_processor_id();
 		if (!breakinfo[breakno].enabled)
 			continue;
+		if (dbg_is_early) {
+			set_debugreg(breakinfo[breakno].addr, breakno);
+			early_dr7 |= encode_dr7(breakno,
+						breakinfo[breakno].len,
+						breakinfo[breakno].type);
+			set_debugreg(early_dr7, 7);
+			continue;
+		}
 		bp = *per_cpu_ptr(breakinfo[breakno].pev, cpu);
 		info = counter_arch_bp(bp);
 		if (bp->attr.disabled != 1)
@@ -236,7 +234,8 @@ static void kgdb_correct_hw_break(void)
 		if (!val)
 			bp->attr.disabled = 0;
 	}
-	hw_breakpoint_restore();
+	if (!dbg_is_early)
+		hw_breakpoint_restore();
 }
 
 static int hw_break_reserve_slot(int breakno)
@@ -245,6 +244,9 @@ static int hw_break_reserve_slot(int breakno)
 	int cnt = 0;
 	struct perf_event **pevent;
 
+	if (dbg_is_early)
+		return 0;
+
 	for_each_online_cpu(cpu) {
 		cnt++;
 		pevent = per_cpu_ptr(breakinfo[breakno].pev, cpu);
@@ -270,6 +272,9 @@ static int hw_break_release_slot(int breakno)
 	struct perf_event **pevent;
 	int cpu;
 
+	if (dbg_is_early)
+		return 0;
+
 	for_each_online_cpu(cpu) {
 		pevent = per_cpu_ptr(breakinfo[breakno].pev, cpu);
 		if (dbg_release_bp_slot(*pevent))
@@ -314,7 +319,11 @@ static void kgdb_remove_all_hw_break(void)
 		bp = *per_cpu_ptr(breakinfo[i].pev, cpu);
 		if (bp->attr.disabled == 1)
 			continue;
-		arch_uninstall_hw_breakpoint(bp);
+		if (dbg_is_early)
+			early_dr7 &= ~encode_dr7(i, breakinfo[i].len,
+						 breakinfo[i].type);
+		else
+			arch_uninstall_hw_breakpoint(bp);
 		bp->attr.disabled = 1;
 	}
 }
@@ -391,6 +400,11 @@ void kgdb_disable_hw_debug(struct pt_regs *regs)
 	for (i = 0; i < 4; i++) {
 		if (!breakinfo[i].enabled)
 			continue;
+		if (dbg_is_early) {
+			early_dr7 &= ~encode_dr7(i, breakinfo[i].len,
+						 breakinfo[i].type);
+			continue;
+		}
 		bp = *per_cpu_ptr(breakinfo[i].pev, cpu);
 		if (bp->attr.disabled == 1)
 			continue;
@@ -399,23 +413,6 @@ void kgdb_disable_hw_debug(struct pt_regs *regs)
 	}
 }
 
-/**
- *	kgdb_post_primary_code - Save error vector/code numbers.
- *	@regs: Original pt_regs.
- *	@e_vector: Original error vector.
- *	@err_code: Original error code.
- *
- *	This is needed on architectures which support SMP and KGDB.
- *	This function is called after all the slave cpus have been put
- *	to a know spin state and the primary CPU has control over KGDB.
- */
-void kgdb_post_primary_code(struct pt_regs *regs, int e_vector, int err_code)
-{
-	/* primary processor is completely in the debugger */
-	gdb_x86vector = e_vector;
-	gdb_x86errcode = err_code;
-}
-
 #ifdef CONFIG_SMP
 /**
  *	kgdb_roundup_cpus - Get other CPUs into a holding pattern
@@ -567,7 +564,7 @@ static int __kgdb_notify(struct die_args *args, unsigned long cmd)
 			return NOTIFY_DONE;
 	}
 
-	if (kgdb_handle_exception(args->trapnr, args->signr, args->err, regs))
+	if (kgdb_handle_exception(args->trapnr, args->signr, cmd, regs))
 		return NOTIFY_DONE;
 
 	/* Must touch watchdog before return to normal operation */
@@ -575,6 +572,26 @@ static int __kgdb_notify(struct die_args *args, unsigned long cmd)
 	return NOTIFY_STOP;
 }
 
+#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP
+int kgdb_ll_trap(int cmd, const char *str,
+		 struct pt_regs *regs, long err, int trap, int sig)
+{
+	struct die_args args = {
+		.regs	= regs,
+		.str	= str,
+		.err	= err,
+		.trapnr	= trap,
+		.signr	= sig,
+
+	};
+
+	if (!kgdb_io_module_registered)
+		return NOTIFY_DONE;
+
+	return __kgdb_notify(&args, cmd);
+}
+#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */
+
 static int
 kgdb_notify(struct notifier_block *self, unsigned long cmd, void *ptr)
 {
@@ -605,14 +622,15 @@ static struct notifier_block kgdb_notifier = {
  */
 int kgdb_arch_init(void)
 {
+	return register_die_notifier(&kgdb_notifier);
+}
+
+void kgdb_arch_late(void)
+{
 	int i, cpu;
-	int ret;
 	struct perf_event_attr attr;
 	struct perf_event **pevent;
 
-	ret = register_die_notifier(&kgdb_notifier);
-	if (ret != 0)
-		return ret;
 	/*
 	 * Pre-allocate the hw breakpoint structions in the non-atomic
 	 * portion of kgdb because this operation requires mutexs to
@@ -624,12 +642,15 @@ int kgdb_arch_init(void)
 	attr.bp_type = HW_BREAKPOINT_W;
 	attr.disabled = 1;
 	for (i = 0; i < 4; i++) {
+		if (breakinfo[i].pev)
+			continue;
 		breakinfo[i].pev = register_wide_hw_breakpoint(&attr, NULL);
 		if (IS_ERR(breakinfo[i].pev)) {
-			printk(KERN_ERR "kgdb: Could not allocate hw breakpoints\n");
+			printk(KERN_ERR "kgdb: Could not allocate hw"
+			       "breakpoints\nDisabling the kernel debugger\n");
 			breakinfo[i].pev = NULL;
 			kgdb_arch_exit();
-			return -1;
+			return;
 		}
 		for_each_online_cpu(cpu) {
 			pevent = per_cpu_ptr(breakinfo[i].pev, cpu);
@@ -640,7 +661,6 @@ int kgdb_arch_init(void)
 			}
 		}
 	}
-	return ret;
 }
 
 /**
@@ -690,6 +710,11 @@ unsigned long kgdb_arch_pc(int exception, struct pt_regs *regs)
 	return instruction_pointer(regs);
 }
 
+void kgdb_arch_set_pc(struct pt_regs *regs, unsigned long ip)
+{
+	regs->ip = ip;
+}
+
 struct kgdb_arch arch_kgdb_ops = {
 	/* Breakpoint instruction: */
 	.gdb_bpt_instr		= { 0xcc },
diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
index b43bbaebe2c..345a4b1fe14 100644
--- a/arch/x86/kernel/kprobes.c
+++ b/arch/x86/kernel/kprobes.c
@@ -422,14 +422,22 @@ static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
 
 static void __kprobes clear_btf(void)
 {
-	if (test_thread_flag(TIF_DEBUGCTLMSR))
-		update_debugctlmsr(0);
+	if (test_thread_flag(TIF_BLOCKSTEP)) {
+		unsigned long debugctl = get_debugctlmsr();
+
+		debugctl &= ~DEBUGCTLMSR_BTF;
+		update_debugctlmsr(debugctl);
+	}
 }
 
 static void __kprobes restore_btf(void)
 {
-	if (test_thread_flag(TIF_DEBUGCTLMSR))
-		update_debugctlmsr(current->thread.debugctlmsr);
+	if (test_thread_flag(TIF_BLOCKSTEP)) {
+		unsigned long debugctl = get_debugctlmsr();
+
+		debugctl |= DEBUGCTLMSR_BTF;
+		update_debugctlmsr(debugctl);
+	}
 }
 
 void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
@@ -534,20 +542,6 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
 	struct kprobe_ctlblk *kcb;
 
 	addr = (kprobe_opcode_t *)(regs->ip - sizeof(kprobe_opcode_t));
-	if (*addr != BREAKPOINT_INSTRUCTION) {
-		/*
-		 * The breakpoint instruction was removed right
-		 * after we hit it.  Another cpu has removed
-		 * either a probepoint or a debugger breakpoint
-		 * at this address.  In either case, no further
-		 * handling of this interrupt is appropriate.
-		 * Back up over the (now missing) int3 and run
-		 * the original instruction.
-		 */
-		regs->ip = (unsigned long)addr;
-		return 1;
-	}
-
 	/*
 	 * We don't want to be preempted for the entire
 	 * duration of kprobe processing. We conditionally
@@ -579,6 +573,19 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
 				setup_singlestep(p, regs, kcb, 0);
 			return 1;
 		}
+	} else if (*addr != BREAKPOINT_INSTRUCTION) {
+		/*
+		 * The breakpoint instruction was removed right
+		 * after we hit it.  Another cpu has removed
+		 * either a probepoint or a debugger breakpoint
+		 * at this address.  In either case, no further
+		 * handling of this interrupt is appropriate.
+		 * Back up over the (now missing) int3 and run
+		 * the original instruction.
+		 */
+		regs->ip = (unsigned long)addr;
+		preempt_enable_no_resched();
+		return 1;
 	} else if (kprobe_running()) {
 		p = __get_cpu_var(current_kprobe);
 		if (p->break_handler && p->break_handler(p, regs)) {
diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c
index feaeb0d3aa4..eb9b76c716c 100644
--- a/arch/x86/kernel/kvmclock.c
+++ b/arch/x86/kernel/kvmclock.c
@@ -29,6 +29,8 @@
 #define KVM_SCALE 22
 
 static int kvmclock = 1;
+static int msr_kvm_system_time = MSR_KVM_SYSTEM_TIME;
+static int msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK;
 
 static int parse_no_kvmclock(char *arg)
 {
@@ -54,7 +56,8 @@ static unsigned long kvm_get_wallclock(void)
 
 	low = (int)__pa_symbol(&wall_clock);
 	high = ((u64)__pa_symbol(&wall_clock) >> 32);
-	native_write_msr(MSR_KVM_WALL_CLOCK, low, high);
+
+	native_write_msr(msr_kvm_wall_clock, low, high);
 
 	vcpu_time = &get_cpu_var(hv_clock);
 	pvclock_read_wallclock(&wall_clock, vcpu_time, &ts);
@@ -130,7 +133,8 @@ static int kvm_register_clock(char *txt)
 	high = ((u64)__pa(&per_cpu(hv_clock, cpu)) >> 32);
 	printk(KERN_INFO "kvm-clock: cpu %d, msr %x:%x, %s\n",
 	       cpu, high, low, txt);
-	return native_write_msr_safe(MSR_KVM_SYSTEM_TIME, low, high);
+
+	return native_write_msr_safe(msr_kvm_system_time, low, high);
 }
 
 #ifdef CONFIG_X86_LOCAL_APIC
@@ -165,14 +169,14 @@ static void __init kvm_smp_prepare_boot_cpu(void)
 #ifdef CONFIG_KEXEC
 static void kvm_crash_shutdown(struct pt_regs *regs)
 {
-	native_write_msr_safe(MSR_KVM_SYSTEM_TIME, 0, 0);
+	native_write_msr(msr_kvm_system_time, 0, 0);
 	native_machine_crash_shutdown(regs);
 }
 #endif
 
 static void kvm_shutdown(void)
 {
-	native_write_msr_safe(MSR_KVM_SYSTEM_TIME, 0, 0);
+	native_write_msr(msr_kvm_system_time, 0, 0);
 	native_machine_shutdown();
 }
 
@@ -181,27 +185,37 @@ void __init kvmclock_init(void)
 	if (!kvm_para_available())
 		return;
 
-	if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE)) {
-		if (kvm_register_clock("boot clock"))
-			return;
-		pv_time_ops.sched_clock = kvm_clock_read;
-		x86_platform.calibrate_tsc = kvm_get_tsc_khz;
-		x86_platform.get_wallclock = kvm_get_wallclock;
-		x86_platform.set_wallclock = kvm_set_wallclock;
+	if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE2)) {
+		msr_kvm_system_time = MSR_KVM_SYSTEM_TIME_NEW;
+		msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK_NEW;
+	} else if (!(kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE)))
+		return;
+
+	printk(KERN_INFO "kvm-clock: Using msrs %x and %x",
+		msr_kvm_system_time, msr_kvm_wall_clock);
+
+	if (kvm_register_clock("boot clock"))
+		return;
+	pv_time_ops.sched_clock = kvm_clock_read;
+	x86_platform.calibrate_tsc = kvm_get_tsc_khz;
+	x86_platform.get_wallclock = kvm_get_wallclock;
+	x86_platform.set_wallclock = kvm_set_wallclock;
 #ifdef CONFIG_X86_LOCAL_APIC
-		x86_cpuinit.setup_percpu_clockev =
-			kvm_setup_secondary_clock;
+	x86_cpuinit.setup_percpu_clockev =
+		kvm_setup_secondary_clock;
 #endif
 #ifdef CONFIG_SMP
-		smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu;
+	smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu;
 #endif
-		machine_ops.shutdown  = kvm_shutdown;
+	machine_ops.shutdown  = kvm_shutdown;
 #ifdef CONFIG_KEXEC
-		machine_ops.crash_shutdown  = kvm_crash_shutdown;
+	machine_ops.crash_shutdown  = kvm_crash_shutdown;
 #endif
-		kvm_get_preset_lpj();
-		clocksource_register(&kvm_clock);
-		pv_info.paravirt_enabled = 1;
-		pv_info.name = "KVM";
-	}
+	kvm_get_preset_lpj();
+	clocksource_register(&kvm_clock);
+	pv_info.paravirt_enabled = 1;
+	pv_info.name = "KVM";
+
+	if (kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE_STABLE_BIT))
+		pvclock_set_flags(PVCLOCK_TSC_STABLE_BIT);
 }
diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c
index cceb5bc3c3c..fa6551d36c1 100644
--- a/arch/x86/kernel/microcode_core.c
+++ b/arch/x86/kernel/microcode_core.c
@@ -201,9 +201,9 @@ static int do_microcode_update(const void __user *buf, size_t size)
 	return error;
 }
 
-static int microcode_open(struct inode *unused1, struct file *unused2)
+static int microcode_open(struct inode *inode, struct file *file)
 {
-	return capable(CAP_SYS_RAWIO) ? 0 : -EPERM;
+	return capable(CAP_SYS_RAWIO) ? nonseekable_open(inode, file) : -EPERM;
 }
 
 static ssize_t microcode_write(struct file *file, const char __user *buf,
@@ -260,6 +260,7 @@ static void microcode_dev_exit(void)
 }
 
 MODULE_ALIAS_MISCDEV(MICROCODE_MINOR);
+MODULE_ALIAS("devname:cpu/microcode");
 #else
 #define microcode_dev_init()	0
 #define microcode_dev_exit()	do { } while (0)
diff --git a/arch/x86/kernel/microcode_intel.c b/arch/x86/kernel/microcode_intel.c
index 85a343e2893..356170262a9 100644
--- a/arch/x86/kernel/microcode_intel.c
+++ b/arch/x86/kernel/microcode_intel.c
@@ -343,10 +343,11 @@ static enum ucode_state generic_load_microcode(int cpu, void *data, size_t size,
 				int (*get_ucode_data)(void *, const void *, size_t))
 {
 	struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
-	u8 *ucode_ptr = data, *new_mc = NULL, *mc;
+	u8 *ucode_ptr = data, *new_mc = NULL, *mc = NULL;
 	int new_rev = uci->cpu_sig.rev;
 	unsigned int leftover = size;
 	enum ucode_state state = UCODE_OK;
+	unsigned int curr_mc_size = 0;
 
 	while (leftover) {
 		struct microcode_header_intel mc_header;
@@ -361,9 +362,15 @@ static enum ucode_state generic_load_microcode(int cpu, void *data, size_t size,
 			break;
 		}
 
-		mc = vmalloc(mc_size);
-		if (!mc)
-			break;
+		/* For performance reasons, reuse mc area when possible */
+		if (!mc || mc_size > curr_mc_size) {
+			if (mc)
+				vfree(mc);
+			mc = vmalloc(mc_size);
+			if (!mc)
+				break;
+			curr_mc_size = mc_size;
+		}
 
 		if (get_ucode_data(mc, ucode_ptr, mc_size) ||
 		    microcode_sanity_check(mc) < 0) {
@@ -376,13 +383,16 @@ static enum ucode_state generic_load_microcode(int cpu, void *data, size_t size,
 				vfree(new_mc);
 			new_rev = mc_header.rev;
 			new_mc  = mc;
-		} else
-			vfree(mc);
+			mc = NULL;	/* trigger new vmalloc */
+		}
 
 		ucode_ptr += mc_size;
 		leftover  -= mc_size;
 	}
 
+	if (mc)
+		vfree(mc);
+
 	if (leftover) {
 		if (new_mc)
 			vfree(new_mc);
diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index e81030f71a8..5ae5d2426ed 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -115,21 +115,6 @@ static void __init MP_bus_info(struct mpc_bus *m)
 		printk(KERN_WARNING "Unknown bustype %s - ignoring\n", str);
 }
 
-static int bad_ioapic(unsigned long address)
-{
-	if (nr_ioapics >= MAX_IO_APICS) {
-		printk(KERN_ERR "ERROR: Max # of I/O APICs (%d) exceeded "
-		       "(found %d)\n", MAX_IO_APICS, nr_ioapics);
-		panic("Recompile kernel with bigger MAX_IO_APICS!\n");
-	}
-	if (!address) {
-		printk(KERN_ERR "WARNING: Bogus (zero) I/O APIC address"
-		       " found in table, skipping!\n");
-		return 1;
-	}
-	return 0;
-}
-
 static void __init MP_ioapic_info(struct mpc_ioapic *m)
 {
 	if (!(m->flags & MPC_APIC_USABLE))
@@ -138,15 +123,7 @@ static void __init MP_ioapic_info(struct mpc_ioapic *m)
 	printk(KERN_INFO "I/O APIC #%d Version %d at 0x%X.\n",
 	       m->apicid, m->apicver, m->apicaddr);
 
-	if (bad_ioapic(m->apicaddr))
-		return;
-
-	mp_ioapics[nr_ioapics].apicaddr = m->apicaddr;
-	mp_ioapics[nr_ioapics].apicid = m->apicid;
-	mp_ioapics[nr_ioapics].type = m->type;
-	mp_ioapics[nr_ioapics].apicver = m->apicver;
-	mp_ioapics[nr_ioapics].flags = m->flags;
-	nr_ioapics++;
+	mp_register_ioapic(m->apicid, m->apicaddr, gsi_end + 1);
 }
 
 static void print_MP_intsrc_info(struct mpc_intsrc *m)
diff --git a/arch/x86/kernel/mrst.c b/arch/x86/kernel/mrst.c
index 0aad8670858..e796448f0eb 100644
--- a/arch/x86/kernel/mrst.c
+++ b/arch/x86/kernel/mrst.c
@@ -237,4 +237,9 @@ void __init x86_mrst_early_setup(void)
 	x86_init.pci.fixup_irqs = x86_init_noop;
 
 	legacy_pic = &null_legacy_pic;
+
+	/* Avoid searching for BIOS MP tables */
+	x86_init.mpparse.find_smp_config = x86_init_noop;
+	x86_init.mpparse.get_smp_config = x86_init_uint_noop;
+
 }
diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c
index 4d4468e9f47..7bf2dc4c8f7 100644
--- a/arch/x86/kernel/msr.c
+++ b/arch/x86/kernel/msr.c
@@ -230,7 +230,7 @@ static int __cpuinit msr_class_cpu_callback(struct notifier_block *nfb,
 		msr_device_destroy(cpu);
 		break;
 	}
-	return err ? NOTIFY_BAD : NOTIFY_OK;
+	return notifier_from_errno(err);
 }
 
 static struct notifier_block __refdata msr_class_cpu_notifier = {
diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c
index 7d2829dde20..a5bc528d432 100644
--- a/arch/x86/kernel/pci-swiotlb.c
+++ b/arch/x86/kernel/pci-swiotlb.c
@@ -31,8 +31,6 @@ static struct dma_map_ops swiotlb_dma_ops = {
 	.free_coherent = swiotlb_free_coherent,
 	.sync_single_for_cpu = swiotlb_sync_single_for_cpu,
 	.sync_single_for_device = swiotlb_sync_single_for_device,
-	.sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
-	.sync_single_range_for_device = swiotlb_sync_single_range_for_device,
 	.sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
 	.sync_sg_for_device = swiotlb_sync_sg_for_device,
 	.map_sg = swiotlb_map_sg_attrs,
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index 28ad9f4d8b9..e7e35219b32 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -20,7 +20,6 @@
 #include <asm/idle.h>
 #include <asm/uaccess.h>
 #include <asm/i387.h>
-#include <asm/ds.h>
 #include <asm/debugreg.h>
 
 unsigned long idle_halt;
@@ -32,26 +31,22 @@ struct kmem_cache *task_xstate_cachep;
 
 int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
 {
+	int ret;
+
 	*dst = *src;
-	if (src->thread.xstate) {
-		dst->thread.xstate = kmem_cache_alloc(task_xstate_cachep,
-						      GFP_KERNEL);
-		if (!dst->thread.xstate)
-			return -ENOMEM;
-		WARN_ON((unsigned long)dst->thread.xstate & 15);
-		memcpy(dst->thread.xstate, src->thread.xstate, xstate_size);
+	if (fpu_allocated(&src->thread.fpu)) {
+		memset(&dst->thread.fpu, 0, sizeof(dst->thread.fpu));
+		ret = fpu_alloc(&dst->thread.fpu);
+		if (ret)
+			return ret;
+		fpu_copy(&dst->thread.fpu, &src->thread.fpu);
 	}
 	return 0;
 }
 
 void free_thread_xstate(struct task_struct *tsk)
 {
-	if (tsk->thread.xstate) {
-		kmem_cache_free(task_xstate_cachep, tsk->thread.xstate);
-		tsk->thread.xstate = NULL;
-	}
-
-	WARN(tsk->thread.ds_ctx, "leaking DS context\n");
+	fpu_free(&tsk->thread.fpu);
 }
 
 void free_thread_info(struct thread_info *ti)
@@ -198,11 +193,16 @@ void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p,
 	prev = &prev_p->thread;
 	next = &next_p->thread;
 
-	if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) ||
-	    test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR))
-		ds_switch_to(prev_p, next_p);
-	else if (next->debugctlmsr != prev->debugctlmsr)
-		update_debugctlmsr(next->debugctlmsr);
+	if (test_tsk_thread_flag(prev_p, TIF_BLOCKSTEP) ^
+	    test_tsk_thread_flag(next_p, TIF_BLOCKSTEP)) {
+		unsigned long debugctl = get_debugctlmsr();
+
+		debugctl &= ~DEBUGCTLMSR_BTF;
+		if (test_tsk_thread_flag(next_p, TIF_BLOCKSTEP))
+			debugctl |= DEBUGCTLMSR_BTF;
+
+		update_debugctlmsr(debugctl);
+	}
 
 	if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^
 	    test_tsk_thread_flag(next_p, TIF_NOTSC)) {
@@ -546,11 +546,13 @@ static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c)
 		 * check OSVW bit for CPUs that are not affected
 		 * by erratum #400
 		 */
-		rdmsrl(MSR_AMD64_OSVW_ID_LENGTH, val);
-		if (val >= 2) {
-			rdmsrl(MSR_AMD64_OSVW_STATUS, val);
-			if (!(val & BIT(1)))
-				goto no_c1e_idle;
+		if (cpu_has(c, X86_FEATURE_OSVW)) {
+			rdmsrl(MSR_AMD64_OSVW_ID_LENGTH, val);
+			if (val >= 2) {
+				rdmsrl(MSR_AMD64_OSVW_STATUS, val);
+				if (!(val & BIT(1)))
+					goto no_c1e_idle;
+			}
 		}
 		return 1;
 	}
diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c
index f6c62667e30..8d128783af4 100644
--- a/arch/x86/kernel/process_32.c
+++ b/arch/x86/kernel/process_32.c
@@ -55,7 +55,6 @@
 #include <asm/cpu.h>
 #include <asm/idle.h>
 #include <asm/syscalls.h>
-#include <asm/ds.h>
 #include <asm/debugreg.h>
 
 asmlinkage void ret_from_fork(void) __asm__("ret_from_fork");
@@ -238,13 +237,6 @@ int copy_thread(unsigned long clone_flags, unsigned long sp,
 		kfree(p->thread.io_bitmap_ptr);
 		p->thread.io_bitmap_max = 0;
 	}
-
-	clear_tsk_thread_flag(p, TIF_DS_AREA_MSR);
-	p->thread.ds_ctx = NULL;
-
-	clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR);
-	p->thread.debugctlmsr = 0;
-
 	return err;
 }
 
@@ -317,7 +309,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 
 	/* we're going to use this soon, after a few expensive things */
 	if (preload_fpu)
-		prefetch(next->xstate);
+		prefetch(next->fpu.state);
 
 	/*
 	 * Reload esp0.
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index 17cb3295cbf..3c2422a99f1 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -49,7 +49,6 @@
 #include <asm/ia32.h>
 #include <asm/idle.h>
 #include <asm/syscalls.h>
-#include <asm/ds.h>
 #include <asm/debugreg.h>
 
 asmlinkage extern void ret_from_fork(void);
@@ -313,13 +312,6 @@ int copy_thread(unsigned long clone_flags, unsigned long sp,
 		if (err)
 			goto out;
 	}
-
-	clear_tsk_thread_flag(p, TIF_DS_AREA_MSR);
-	p->thread.ds_ctx = NULL;
-
-	clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR);
-	p->thread.debugctlmsr = 0;
-
 	err = 0;
 out:
 	if (err && p->thread.io_bitmap_ptr) {
@@ -396,7 +388,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 
 	/* we're going to use this soon, after a few expensive things */
 	if (preload_fpu)
-		prefetch(next->xstate);
+		prefetch(next->fpu.state);
 
 	/*
 	 * Reload esp0, LDT and the page table pointer:
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index 2e9b55027b7..70c4872cd8a 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -2,9 +2,6 @@
 /*
  * Pentium III FXSR, SSE support
  *	Gareth Hughes <gareth@valinux.com>, May 2000
- *
- * BTS tracing
- *	Markus Metzger <markus.t.metzger@intel.com>, Dec 2007
  */
 
 #include <linux/kernel.h>
@@ -22,7 +19,6 @@
 #include <linux/audit.h>
 #include <linux/seccomp.h>
 #include <linux/signal.h>
-#include <linux/workqueue.h>
 #include <linux/perf_event.h>
 #include <linux/hw_breakpoint.h>
 
@@ -36,7 +32,6 @@
 #include <asm/desc.h>
 #include <asm/prctl.h>
 #include <asm/proto.h>
-#include <asm/ds.h>
 #include <asm/hw_breakpoint.h>
 
 #include "tls.h"
@@ -693,7 +688,7 @@ static int ptrace_set_breakpoint_addr(struct task_struct *tsk, int nr,
 	struct perf_event_attr attr;
 
 	if (!t->ptrace_bps[nr]) {
-		hw_breakpoint_init(&attr);
+		ptrace_breakpoint_init(&attr);
 		/*
 		 * Put stub len and type to register (reserve) an inactive but
 		 * correct bp
@@ -789,342 +784,6 @@ static int ioperm_get(struct task_struct *target,
 				   0, IO_BITMAP_BYTES);
 }
 
-#ifdef CONFIG_X86_PTRACE_BTS
-/*
- * A branch trace store context.
- *
- * Contexts may only be installed by ptrace_bts_config() and only for
- * ptraced tasks.
- *
- * Contexts are destroyed when the tracee is detached from the tracer.
- * The actual destruction work requires interrupts enabled, so the
- * work is deferred and will be scheduled during __ptrace_unlink().
- *
- * Contexts hold an additional task_struct reference on the traced
- * task, as well as a reference on the tracer's mm.
- *
- * Ptrace already holds a task_struct for the duration of ptrace operations,
- * but since destruction is deferred, it may be executed after both
- * tracer and tracee exited.
- */
-struct bts_context {
-	/* The branch trace handle. */
-	struct bts_tracer	*tracer;
-
-	/* The buffer used to store the branch trace and its size. */
-	void			*buffer;
-	unsigned int		size;
-
-	/* The mm that paid for the above buffer. */
-	struct mm_struct	*mm;
-
-	/* The task this context belongs to. */
-	struct task_struct	*task;
-
-	/* The signal to send on a bts buffer overflow. */
-	unsigned int		bts_ovfl_signal;
-
-	/* The work struct to destroy a context. */
-	struct work_struct	work;
-};
-
-static int alloc_bts_buffer(struct bts_context *context, unsigned int size)
-{
-	void *buffer = NULL;
-	int err = -ENOMEM;
-
-	err = account_locked_memory(current->mm, current->signal->rlim, size);
-	if (err < 0)
-		return err;
-
-	buffer = kzalloc(size, GFP_KERNEL);
-	if (!buffer)
-		goto out_refund;
-
-	context->buffer = buffer;
-	context->size = size;
-	context->mm = get_task_mm(current);
-
-	return 0;
-
- out_refund:
-	refund_locked_memory(current->mm, size);
-	return err;
-}
-
-static inline void free_bts_buffer(struct bts_context *context)
-{
-	if (!context->buffer)
-		return;
-
-	kfree(context->buffer);
-	context->buffer = NULL;
-
-	refund_locked_memory(context->mm, context->size);
-	context->size = 0;
-
-	mmput(context->mm);
-	context->mm = NULL;
-}
-
-static void free_bts_context_work(struct work_struct *w)
-{
-	struct bts_context *context;
-
-	context = container_of(w, struct bts_context, work);
-
-	ds_release_bts(context->tracer);
-	put_task_struct(context->task);
-	free_bts_buffer(context);
-	kfree(context);
-}
-
-static inline void free_bts_context(struct bts_context *context)
-{
-	INIT_WORK(&context->work, free_bts_context_work);
-	schedule_work(&context->work);
-}
-
-static inline struct bts_context *alloc_bts_context(struct task_struct *task)
-{
-	struct bts_context *context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (context) {
-		context->task = task;
-		task->bts = context;
-
-		get_task_struct(task);
-	}
-
-	return context;
-}
-
-static int ptrace_bts_read_record(struct task_struct *child, size_t index,
-				  struct bts_struct __user *out)
-{
-	struct bts_context *context;
-	const struct bts_trace *trace;
-	struct bts_struct bts;
-	const unsigned char *at;
-	int error;
-
-	context = child->bts;
-	if (!context)
-		return -ESRCH;
-
-	trace = ds_read_bts(context->tracer);
-	if (!trace)
-		return -ESRCH;
-
-	at = trace->ds.top - ((index + 1) * trace->ds.size);
-	if ((void *)at < trace->ds.begin)
-		at += (trace->ds.n * trace->ds.size);
-
-	if (!trace->read)
-		return -EOPNOTSUPP;
-
-	error = trace->read(context->tracer, at, &bts);
-	if (error < 0)
-		return error;
-
-	if (copy_to_user(out, &bts, sizeof(bts)))
-		return -EFAULT;
-
-	return sizeof(bts);
-}
-
-static int ptrace_bts_drain(struct task_struct *child,
-			    long size,
-			    struct bts_struct __user *out)
-{
-	struct bts_context *context;
-	const struct bts_trace *trace;
-	const unsigned char *at;
-	int error, drained = 0;
-
-	context = child->bts;
-	if (!context)
-		return -ESRCH;
-
-	trace = ds_read_bts(context->tracer);
-	if (!trace)
-		return -ESRCH;
-
-	if (!trace->read)
-		return -EOPNOTSUPP;
-
-	if (size < (trace->ds.top - trace->ds.begin))
-		return -EIO;
-
-	for (at = trace->ds.begin; (void *)at < trace->ds.top;
-	     out++, drained++, at += trace->ds.size) {
-		struct bts_struct bts;
-
-		error = trace->read(context->tracer, at, &bts);
-		if (error < 0)
-			return error;
-
-		if (copy_to_user(out, &bts, sizeof(bts)))
-			return -EFAULT;
-	}
-
-	memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size);
-
-	error = ds_reset_bts(context->tracer);
-	if (error < 0)
-		return error;
-
-	return drained;
-}
-
-static int ptrace_bts_config(struct task_struct *child,
-			     long cfg_size,
-			     const struct ptrace_bts_config __user *ucfg)
-{
-	struct bts_context *context;
-	struct ptrace_bts_config cfg;
-	unsigned int flags = 0;
-
-	if (cfg_size < sizeof(cfg))
-		return -EIO;
-
-	if (copy_from_user(&cfg, ucfg, sizeof(cfg)))
-		return -EFAULT;
-
-	context = child->bts;
-	if (!context)
-		context = alloc_bts_context(child);
-	if (!context)
-		return -ENOMEM;
-
-	if (cfg.flags & PTRACE_BTS_O_SIGNAL) {
-		if (!cfg.signal)
-			return -EINVAL;
-
-		return -EOPNOTSUPP;
-		context->bts_ovfl_signal = cfg.signal;
-	}
-
-	ds_release_bts(context->tracer);
-	context->tracer = NULL;
-
-	if ((cfg.flags & PTRACE_BTS_O_ALLOC) && (cfg.size != context->size)) {
-		int err;
-
-		free_bts_buffer(context);
-		if (!cfg.size)
-			return 0;
-
-		err = alloc_bts_buffer(context, cfg.size);
-		if (err < 0)
-			return err;
-	}
-
-	if (cfg.flags & PTRACE_BTS_O_TRACE)
-		flags |= BTS_USER;
-
-	if (cfg.flags & PTRACE_BTS_O_SCHED)
-		flags |= BTS_TIMESTAMPS;
-
-	context->tracer =
-		ds_request_bts_task(child, context->buffer, context->size,
-				    NULL, (size_t)-1, flags);
-	if (unlikely(IS_ERR(context->tracer))) {
-		int error = PTR_ERR(context->tracer);
-
-		free_bts_buffer(context);
-		context->tracer = NULL;
-		return error;
-	}
-
-	return sizeof(cfg);
-}
-
-static int ptrace_bts_status(struct task_struct *child,
-			     long cfg_size,
-			     struct ptrace_bts_config __user *ucfg)
-{
-	struct bts_context *context;
-	const struct bts_trace *trace;
-	struct ptrace_bts_config cfg;
-
-	context = child->bts;
-	if (!context)
-		return -ESRCH;
-
-	if (cfg_size < sizeof(cfg))
-		return -EIO;
-
-	trace = ds_read_bts(context->tracer);
-	if (!trace)
-		return -ESRCH;
-
-	memset(&cfg, 0, sizeof(cfg));
-	cfg.size	= trace->ds.end - trace->ds.begin;
-	cfg.signal	= context->bts_ovfl_signal;
-	cfg.bts_size	= sizeof(struct bts_struct);
-
-	if (cfg.signal)
-		cfg.flags |= PTRACE_BTS_O_SIGNAL;
-
-	if (trace->ds.flags & BTS_USER)
-		cfg.flags |= PTRACE_BTS_O_TRACE;
-
-	if (trace->ds.flags & BTS_TIMESTAMPS)
-		cfg.flags |= PTRACE_BTS_O_SCHED;
-
-	if (copy_to_user(ucfg, &cfg, sizeof(cfg)))
-		return -EFAULT;
-
-	return sizeof(cfg);
-}
-
-static int ptrace_bts_clear(struct task_struct *child)
-{
-	struct bts_context *context;
-	const struct bts_trace *trace;
-
-	context = child->bts;
-	if (!context)
-		return -ESRCH;
-
-	trace = ds_read_bts(context->tracer);
-	if (!trace)
-		return -ESRCH;
-
-	memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size);
-
-	return ds_reset_bts(context->tracer);
-}
-
-static int ptrace_bts_size(struct task_struct *child)
-{
-	struct bts_context *context;
-	const struct bts_trace *trace;
-
-	context = child->bts;
-	if (!context)
-		return -ESRCH;
-
-	trace = ds_read_bts(context->tracer);
-	if (!trace)
-		return -ESRCH;
-
-	return (trace->ds.top - trace->ds.begin) / trace->ds.size;
-}
-
-/*
- * Called from __ptrace_unlink() after the child has been moved back
- * to its original parent.
- */
-void ptrace_bts_untrace(struct task_struct *child)
-{
-	if (unlikely(child->bts)) {
-		free_bts_context(child->bts);
-		child->bts = NULL;
-	}
-}
-#endif /* CONFIG_X86_PTRACE_BTS */
-
 /*
  * Called by kernel/ptrace.c when detaching..
  *
@@ -1252,39 +911,6 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 		break;
 #endif
 
-	/*
-	 * These bits need more cooking - not enabled yet:
-	 */
-#ifdef CONFIG_X86_PTRACE_BTS
-	case PTRACE_BTS_CONFIG:
-		ret = ptrace_bts_config
-			(child, data, (struct ptrace_bts_config __user *)addr);
-		break;
-
-	case PTRACE_BTS_STATUS:
-		ret = ptrace_bts_status
-			(child, data, (struct ptrace_bts_config __user *)addr);
-		break;
-
-	case PTRACE_BTS_SIZE:
-		ret = ptrace_bts_size(child);
-		break;
-
-	case PTRACE_BTS_GET:
-		ret = ptrace_bts_read_record
-			(child, data, (struct bts_struct __user *) addr);
-		break;
-
-	case PTRACE_BTS_CLEAR:
-		ret = ptrace_bts_clear(child);
-		break;
-
-	case PTRACE_BTS_DRAIN:
-		ret = ptrace_bts_drain
-			(child, data, (struct bts_struct __user *) addr);
-		break;
-#endif /* CONFIG_X86_PTRACE_BTS */
-
 	default:
 		ret = ptrace_request(child, request, addr, data);
 		break;
@@ -1544,14 +1170,6 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request,
 
 	case PTRACE_GET_THREAD_AREA:
 	case PTRACE_SET_THREAD_AREA:
-#ifdef CONFIG_X86_PTRACE_BTS
-	case PTRACE_BTS_CONFIG:
-	case PTRACE_BTS_STATUS:
-	case PTRACE_BTS_SIZE:
-	case PTRACE_BTS_GET:
-	case PTRACE_BTS_CLEAR:
-	case PTRACE_BTS_DRAIN:
-#endif /* CONFIG_X86_PTRACE_BTS */
 		return arch_ptrace(child, request, addr, data);
 
 	default:
diff --git a/arch/x86/kernel/pvclock.c b/arch/x86/kernel/pvclock.c
index 03801f2f761..239427ca02a 100644
--- a/arch/x86/kernel/pvclock.c
+++ b/arch/x86/kernel/pvclock.c
@@ -31,8 +31,16 @@ struct pvclock_shadow_time {
 	u32 tsc_to_nsec_mul;
 	int tsc_shift;
 	u32 version;
+	u8  flags;
 };
 
+static u8 valid_flags __read_mostly = 0;
+
+void pvclock_set_flags(u8 flags)
+{
+	valid_flags = flags;
+}
+
 /*
  * Scale a 64-bit delta by scaling and multiplying by a 32-bit fraction,
  * yielding a 64-bit result.
@@ -91,6 +99,7 @@ static unsigned pvclock_get_time_values(struct pvclock_shadow_time *dst,
 		dst->system_timestamp  = src->system_time;
 		dst->tsc_to_nsec_mul   = src->tsc_to_system_mul;
 		dst->tsc_shift         = src->tsc_shift;
+		dst->flags             = src->flags;
 		rmb();		/* test version after fetching data */
 	} while ((src->version & 1) || (dst->version != src->version));
 
@@ -109,11 +118,14 @@ unsigned long pvclock_tsc_khz(struct pvclock_vcpu_time_info *src)
 	return pv_tsc_khz;
 }
 
+static atomic64_t last_value = ATOMIC64_INIT(0);
+
 cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
 {
 	struct pvclock_shadow_time shadow;
 	unsigned version;
 	cycle_t ret, offset;
+	u64 last;
 
 	do {
 		version = pvclock_get_time_values(&shadow, src);
@@ -123,6 +135,31 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
 		barrier();
 	} while (version != src->version);
 
+	if ((valid_flags & PVCLOCK_TSC_STABLE_BIT) &&
+		(shadow.flags & PVCLOCK_TSC_STABLE_BIT))
+		return ret;
+
+	/*
+	 * Assumption here is that last_value, a global accumulator, always goes
+	 * forward. If we are less than that, we should not be much smaller.
+	 * We assume there is an error marging we're inside, and then the correction
+	 * does not sacrifice accuracy.
+	 *
+	 * For reads: global may have changed between test and return,
+	 * but this means someone else updated poked the clock at a later time.
+	 * We just need to make sure we are not seeing a backwards event.
+	 *
+	 * For updates: last_value = ret is not enough, since two vcpus could be
+	 * updating at the same time, and one of them could be slightly behind,
+	 * making the assumption that last_value always go forward fail to hold.
+	 */
+	last = atomic64_read(&last_value);
+	do {
+		if (ret < last)
+			return last;
+		last = atomic64_cmpxchg(&last_value, last, ret);
+	} while (unlikely(last != ret));
+
 	return ret;
 }
 
diff --git a/arch/x86/kernel/quirks.c b/arch/x86/kernel/quirks.c
index 12e9feaa2f7..e72d3fc6547 100644
--- a/arch/x86/kernel/quirks.c
+++ b/arch/x86/kernel/quirks.c
@@ -495,10 +495,18 @@ void force_hpet_resume(void)
 /*
  * HPET MSI on some boards (ATI SB700/SB800) has side effect on
  * floppy DMA. Disable HPET MSI on such platforms.
+ * See erratum #27 (Misinterpreted MSI Requests May Result in
+ * Corrupted LPC DMA Data) in AMD Publication #46837,
+ * "SB700 Family Product Errata", Rev. 1.0, March 2010.
+ *
+ * Also force the read back of the CMP register in hpet_next_event()
+ * to work around the problem that the CMP register write seems to be
+ * delayed. See hpet_next_event() for details.
  */
 static void force_disable_hpet_msi(struct pci_dev *unused)
 {
 	hpet_msi_disable = 1;
+	hpet_readback_cmp = 1;
 }
 
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS,
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index c4851eff57b..b4ae4acbd03 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -676,6 +676,17 @@ static struct dmi_system_id __initdata bad_bios_dmi_table[] = {
 			DMI_MATCH(DMI_BOARD_NAME, "DG45FC"),
 		},
 	},
+	/*
+	 * The Dell Inspiron Mini 1012 has DMI_BIOS_VENDOR = "Dell Inc.", so
+	 * match on the product name.
+	 */
+	{
+		.callback = dmi_low_memory_corruption,
+		.ident = "Phoenix BIOS",
+		.matches = {
+			DMI_MATCH(DMI_PRODUCT_NAME, "Inspiron 1012"),
+		},
+	},
 #endif
 	{}
 };
@@ -725,6 +736,7 @@ void __init setup_arch(char **cmdline_p)
 	/* VMI may relocate the fixmap; do this before touching ioremap area */
 	vmi_init();
 
+	early_trap_init();
 	early_cpu_init();
 	early_ioremap_init();
 
diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c
index ef6370b00e7..a867940a6df 100644
--- a/arch/x86/kernel/setup_percpu.c
+++ b/arch/x86/kernel/setup_percpu.c
@@ -265,10 +265,10 @@ void __init setup_per_cpu_areas(void)
 
 #if defined(CONFIG_X86_64) && defined(CONFIG_NUMA)
 	/*
-	 * make sure boot cpu node_number is right, when boot cpu is on the
+	 * make sure boot cpu numa_node is right, when boot cpu is on the
 	 * node that doesn't have mem installed
 	 */
-	per_cpu(node_number, boot_cpu_id) = cpu_to_node(boot_cpu_id);
+	set_cpu_numa_node(boot_cpu_id, early_cpu_to_node(boot_cpu_id));
 #endif
 
 	/* Setup node to cpumask map */
diff --git a/arch/x86/kernel/sfi.c b/arch/x86/kernel/sfi.c
index 34e09938265..7ded57896c0 100644
--- a/arch/x86/kernel/sfi.c
+++ b/arch/x86/kernel/sfi.c
@@ -81,7 +81,6 @@ static int __init sfi_parse_cpus(struct sfi_table_header *table)
 #endif /* CONFIG_X86_LOCAL_APIC */
 
 #ifdef CONFIG_X86_IO_APIC
-static u32 gsi_base;
 
 static int __init sfi_parse_ioapic(struct sfi_table_header *table)
 {
@@ -94,8 +93,7 @@ static int __init sfi_parse_ioapic(struct sfi_table_header *table)
 	pentry = (struct sfi_apic_table_entry *)sb->pentry;
 
 	for (i = 0; i < num; i++) {
-		mp_register_ioapic(i, pentry->phys_addr, gsi_base);
-		gsi_base += io_apic_get_redir_entries(i);
+		mp_register_ioapic(i, pentry->phys_addr, gsi_end + 1);
 		pentry++;
 	}
 
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 763d815e27a..37462f1ddba 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -1215,9 +1215,17 @@ __init void prefill_possible_map(void)
 	if (!num_processors)
 		num_processors = 1;
 
-	if (setup_possible_cpus == -1)
-		possible = num_processors + disabled_cpus;
-	else
+	i = setup_max_cpus ?: 1;
+	if (setup_possible_cpus == -1) {
+		possible = num_processors;
+#ifdef CONFIG_HOTPLUG_CPU
+		if (setup_max_cpus)
+			possible += disabled_cpus;
+#else
+		if (possible > i)
+			possible = i;
+#endif
+	} else
 		possible = setup_possible_cpus;
 
 	total_cpus = max_t(int, possible, num_processors + disabled_cpus);
@@ -1230,11 +1238,23 @@ __init void prefill_possible_map(void)
 		possible = nr_cpu_ids;
 	}
 
+#ifdef CONFIG_HOTPLUG_CPU
+	if (!setup_max_cpus)
+#endif
+	if (possible > i) {
+		printk(KERN_WARNING
+			"%d Processors exceeds max_cpus limit of %u\n",
+			possible, setup_max_cpus);
+		possible = i;
+	}
+
 	printk(KERN_INFO "SMP: Allowing %d CPUs, %d hotplug CPUs\n",
 		possible, max_t(int, possible - num_processors, 0));
 
 	for (i = 0; i < possible; i++)
 		set_cpu_possible(i, true);
+	for (; i < NR_CPUS; i++)
+		set_cpu_possible(i, false);
 
 	nr_cpu_ids = possible;
 }
diff --git a/arch/x86/kernel/step.c b/arch/x86/kernel/step.c
index 3149032ff10..58de45ee08b 100644
--- a/arch/x86/kernel/step.c
+++ b/arch/x86/kernel/step.c
@@ -158,22 +158,6 @@ static int enable_single_step(struct task_struct *child)
 }
 
 /*
- * Install this value in MSR_IA32_DEBUGCTLMSR whenever child is running.
- */
-static void write_debugctlmsr(struct task_struct *child, unsigned long val)
-{
-	if (child->thread.debugctlmsr == val)
-		return;
-
-	child->thread.debugctlmsr = val;
-
-	if (child != current)
-		return;
-
-	update_debugctlmsr(val);
-}
-
-/*
  * Enable single or block step.
  */
 static void enable_step(struct task_struct *child, bool block)
@@ -186,15 +170,17 @@ static void enable_step(struct task_struct *child, bool block)
 	 * that uses user-mode single stepping itself.
 	 */
 	if (enable_single_step(child) && block) {
-		set_tsk_thread_flag(child, TIF_DEBUGCTLMSR);
-		write_debugctlmsr(child,
-				  child->thread.debugctlmsr | DEBUGCTLMSR_BTF);
-	} else {
-		write_debugctlmsr(child,
-				  child->thread.debugctlmsr & ~DEBUGCTLMSR_BTF);
-
-		if (!child->thread.debugctlmsr)
-			clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR);
+		unsigned long debugctl = get_debugctlmsr();
+
+		debugctl |= DEBUGCTLMSR_BTF;
+		update_debugctlmsr(debugctl);
+		set_tsk_thread_flag(child, TIF_BLOCKSTEP);
+	} else if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) {
+		unsigned long debugctl = get_debugctlmsr();
+
+		debugctl &= ~DEBUGCTLMSR_BTF;
+		update_debugctlmsr(debugctl);
+		clear_tsk_thread_flag(child, TIF_BLOCKSTEP);
 	}
 }
 
@@ -213,11 +199,13 @@ void user_disable_single_step(struct task_struct *child)
 	/*
 	 * Make sure block stepping (BTF) is disabled.
 	 */
-	write_debugctlmsr(child,
-			  child->thread.debugctlmsr & ~DEBUGCTLMSR_BTF);
+	if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) {
+		unsigned long debugctl = get_debugctlmsr();
 
-	if (!child->thread.debugctlmsr)
-		clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR);
+		debugctl &= ~DEBUGCTLMSR_BTF;
+		update_debugctlmsr(debugctl);
+		clear_tsk_thread_flag(child, TIF_BLOCKSTEP);
+	}
 
 	/* Always clear TIF_SINGLESTEP... */
 	clear_tsk_thread_flag(child, TIF_SINGLESTEP);
diff --git a/arch/x86/kernel/tboot.c b/arch/x86/kernel/tboot.c
index 86c9f91b48a..c2f1b26141e 100644
--- a/arch/x86/kernel/tboot.c
+++ b/arch/x86/kernel/tboot.c
@@ -46,6 +46,7 @@
 
 /* Global pointer to shared data; NULL means no measured launch. */
 struct tboot *tboot __read_mostly;
+EXPORT_SYMBOL(tboot);
 
 /* timeout for APs (in secs) to enter wait-for-SIPI state during shutdown */
 #define AP_WAIT_TIMEOUT		1
@@ -175,6 +176,9 @@ static void add_mac_region(phys_addr_t start, unsigned long size)
 	struct tboot_mac_region *mr;
 	phys_addr_t end = start + size;
 
+	if (tboot->num_mac_regions >= MAX_TB_MAC_REGIONS)
+		panic("tboot: Too many MAC regions\n");
+
 	if (start && size) {
 		mr = &tboot->mac_regions[tboot->num_mac_regions++];
 		mr->start = round_down(start, PAGE_SIZE);
@@ -184,18 +188,17 @@ static void add_mac_region(phys_addr_t start, unsigned long size)
 
 static int tboot_setup_sleep(void)
 {
+	int i;
+
 	tboot->num_mac_regions = 0;
 
-	/* S3 resume code */
-	add_mac_region(acpi_wakeup_address, WAKEUP_SIZE);
+	for (i = 0; i < e820.nr_map; i++) {
+		if ((e820.map[i].type != E820_RAM)
+		 && (e820.map[i].type != E820_RESERVED_KERN))
+			continue;
 
-#ifdef CONFIG_X86_TRAMPOLINE
-	/* AP trampoline code */
-	add_mac_region(virt_to_phys(trampoline_base), TRAMPOLINE_SIZE);
-#endif
-
-	/* kernel code + data + bss */
-	add_mac_region(virt_to_phys(_text), _end - _text);
+		add_mac_region(e820.map[i].addr, e820.map[i].size);
+	}
 
 	tboot->acpi_sinfo.kernel_s3_resume_vector = acpi_wakeup_address;
 
diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c
index 17b03dd3a6b..7fea555929e 100644
--- a/arch/x86/kernel/tlb_uv.c
+++ b/arch/x86/kernel/tlb_uv.c
@@ -1,7 +1,7 @@
 /*
  *	SGI UltraViolet TLB flush routines.
  *
- *	(c) 2008 Cliff Wickman <cpw@sgi.com>, SGI.
+ *	(c) 2008-2010 Cliff Wickman <cpw@sgi.com>, SGI.
  *
  *	This code is released under the GNU General Public License version 2 or
  *	later.
@@ -20,42 +20,67 @@
 #include <asm/idle.h>
 #include <asm/tsc.h>
 #include <asm/irq_vectors.h>
+#include <asm/timer.h>
 
-static struct bau_control	**uv_bau_table_bases __read_mostly;
-static int			uv_bau_retry_limit __read_mostly;
+struct msg_desc {
+	struct bau_payload_queue_entry *msg;
+	int msg_slot;
+	int sw_ack_slot;
+	struct bau_payload_queue_entry *va_queue_first;
+	struct bau_payload_queue_entry *va_queue_last;
+};
 
-/* base pnode in this partition */
-static int			uv_partition_base_pnode __read_mostly;
+#define UV_INTD_SOFT_ACK_TIMEOUT_PERIOD	0x000000000bUL
+
+static int uv_bau_max_concurrent __read_mostly;
+
+static int nobau;
+static int __init setup_nobau(char *arg)
+{
+	nobau = 1;
+	return 0;
+}
+early_param("nobau", setup_nobau);
 
-static unsigned long		uv_mmask __read_mostly;
+/* base pnode in this partition */
+static int uv_partition_base_pnode __read_mostly;
+/* position of pnode (which is nasid>>1): */
+static int uv_nshift __read_mostly;
+static unsigned long uv_mmask __read_mostly;
 
 static DEFINE_PER_CPU(struct ptc_stats, ptcstats);
 static DEFINE_PER_CPU(struct bau_control, bau_control);
+static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask);
+
+struct reset_args {
+	int sender;
+};
 
 /*
- * Determine the first node on a blade.
+ * Determine the first node on a uvhub. 'Nodes' are used for kernel
+ * memory allocation.
  */
-static int __init blade_to_first_node(int blade)
+static int __init uvhub_to_first_node(int uvhub)
 {
 	int node, b;
 
 	for_each_online_node(node) {
 		b = uv_node_to_blade_id(node);
-		if (blade == b)
+		if (uvhub == b)
 			return node;
 	}
-	return -1; /* shouldn't happen */
+	return -1;
 }
 
 /*
- * Determine the apicid of the first cpu on a blade.
+ * Determine the apicid of the first cpu on a uvhub.
  */
-static int __init blade_to_first_apicid(int blade)
+static int __init uvhub_to_first_apicid(int uvhub)
 {
 	int cpu;
 
 	for_each_present_cpu(cpu)
-		if (blade == uv_cpu_to_blade_id(cpu))
+		if (uvhub == uv_cpu_to_blade_id(cpu))
 			return per_cpu(x86_cpu_to_apicid, cpu);
 	return -1;
 }
@@ -68,195 +93,459 @@ static int __init blade_to_first_apicid(int blade)
  * clear of the Timeout bit (as well) will free the resource. No reply will
  * be sent (the hardware will only do one reply per message).
  */
-static void uv_reply_to_message(int resource,
-				struct bau_payload_queue_entry *msg,
-				struct bau_msg_status *msp)
+static inline void uv_reply_to_message(struct msg_desc *mdp,
+				       struct bau_control *bcp)
 {
 	unsigned long dw;
+	struct bau_payload_queue_entry *msg;
 
-	dw = (1 << (resource + UV_SW_ACK_NPENDING)) | (1 << resource);
+	msg = mdp->msg;
+	if (!msg->canceled) {
+		dw = (msg->sw_ack_vector << UV_SW_ACK_NPENDING) |
+						msg->sw_ack_vector;
+		uv_write_local_mmr(
+				UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS, dw);
+	}
 	msg->replied_to = 1;
 	msg->sw_ack_vector = 0;
-	if (msp)
-		msp->seen_by.bits = 0;
-	uv_write_local_mmr(UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS, dw);
 }
 
 /*
- * Do all the things a cpu should do for a TLB shootdown message.
- * Other cpu's may come here at the same time for this message.
+ * Process the receipt of a RETRY message
  */
-static void uv_bau_process_message(struct bau_payload_queue_entry *msg,
-				   int msg_slot, int sw_ack_slot)
+static inline void uv_bau_process_retry_msg(struct msg_desc *mdp,
+					    struct bau_control *bcp)
 {
-	unsigned long this_cpu_mask;
-	struct bau_msg_status *msp;
-	int cpu;
+	int i;
+	int cancel_count = 0;
+	int slot2;
+	unsigned long msg_res;
+	unsigned long mmr = 0;
+	struct bau_payload_queue_entry *msg;
+	struct bau_payload_queue_entry *msg2;
+	struct ptc_stats *stat;
 
-	msp = __get_cpu_var(bau_control).msg_statuses + msg_slot;
-	cpu = uv_blade_processor_id();
-	msg->number_of_cpus =
-		uv_blade_nr_online_cpus(uv_node_to_blade_id(numa_node_id()));
-	this_cpu_mask = 1UL << cpu;
-	if (msp->seen_by.bits & this_cpu_mask)
-		return;
-	atomic_or_long(&msp->seen_by.bits, this_cpu_mask);
+	msg = mdp->msg;
+	stat = &per_cpu(ptcstats, bcp->cpu);
+	stat->d_retries++;
+	/*
+	 * cancel any message from msg+1 to the retry itself
+	 */
+	for (msg2 = msg+1, i = 0; i < DEST_Q_SIZE; msg2++, i++) {
+		if (msg2 > mdp->va_queue_last)
+			msg2 = mdp->va_queue_first;
+		if (msg2 == msg)
+			break;
+
+		/* same conditions for cancellation as uv_do_reset */
+		if ((msg2->replied_to == 0) && (msg2->canceled == 0) &&
+		    (msg2->sw_ack_vector) && ((msg2->sw_ack_vector &
+			msg->sw_ack_vector) == 0) &&
+		    (msg2->sending_cpu == msg->sending_cpu) &&
+		    (msg2->msg_type != MSG_NOOP)) {
+			slot2 = msg2 - mdp->va_queue_first;
+			mmr = uv_read_local_mmr
+				(UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE);
+			msg_res = ((msg2->sw_ack_vector << 8) |
+				   msg2->sw_ack_vector);
+			/*
+			 * This is a message retry; clear the resources held
+			 * by the previous message only if they timed out.
+			 * If it has not timed out we have an unexpected
+			 * situation to report.
+			 */
+			if (mmr & (msg_res << 8)) {
+				/*
+				 * is the resource timed out?
+				 * make everyone ignore the cancelled message.
+				 */
+				msg2->canceled = 1;
+				stat->d_canceled++;
+				cancel_count++;
+				uv_write_local_mmr(
+				    UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS,
+					(msg_res << 8) | msg_res);
+			} else
+				printk(KERN_INFO "note bau retry: no effect\n");
+		}
+	}
+	if (!cancel_count)
+		stat->d_nocanceled++;
+}
 
-	if (msg->replied_to == 1)
-		return;
+/*
+ * Do all the things a cpu should do for a TLB shootdown message.
+ * Other cpu's may come here at the same time for this message.
+ */
+static void uv_bau_process_message(struct msg_desc *mdp,
+				   struct bau_control *bcp)
+{
+	int msg_ack_count;
+	short socket_ack_count = 0;
+	struct ptc_stats *stat;
+	struct bau_payload_queue_entry *msg;
+	struct bau_control *smaster = bcp->socket_master;
 
+	/*
+	 * This must be a normal message, or retry of a normal message
+	 */
+	msg = mdp->msg;
+	stat = &per_cpu(ptcstats, bcp->cpu);
 	if (msg->address == TLB_FLUSH_ALL) {
 		local_flush_tlb();
-		__get_cpu_var(ptcstats).alltlb++;
+		stat->d_alltlb++;
 	} else {
 		__flush_tlb_one(msg->address);
-		__get_cpu_var(ptcstats).onetlb++;
+		stat->d_onetlb++;
 	}
+	stat->d_requestee++;
+
+	/*
+	 * One cpu on each uvhub has the additional job on a RETRY
+	 * of releasing the resource held by the message that is
+	 * being retried.  That message is identified by sending
+	 * cpu number.
+	 */
+	if (msg->msg_type == MSG_RETRY && bcp == bcp->uvhub_master)
+		uv_bau_process_retry_msg(mdp, bcp);
 
-	__get_cpu_var(ptcstats).requestee++;
+	/*
+	 * This is a sw_ack message, so we have to reply to it.
+	 * Count each responding cpu on the socket. This avoids
+	 * pinging the count's cache line back and forth between
+	 * the sockets.
+	 */
+	socket_ack_count = atomic_add_short_return(1, (struct atomic_short *)
+			&smaster->socket_acknowledge_count[mdp->msg_slot]);
+	if (socket_ack_count == bcp->cpus_in_socket) {
+		/*
+		 * Both sockets dump their completed count total into
+		 * the message's count.
+		 */
+		smaster->socket_acknowledge_count[mdp->msg_slot] = 0;
+		msg_ack_count = atomic_add_short_return(socket_ack_count,
+				(struct atomic_short *)&msg->acknowledge_count);
+
+		if (msg_ack_count == bcp->cpus_in_uvhub) {
+			/*
+			 * All cpus in uvhub saw it; reply
+			 */
+			uv_reply_to_message(mdp, bcp);
+		}
+	}
 
-	atomic_inc_short(&msg->acknowledge_count);
-	if (msg->number_of_cpus == msg->acknowledge_count)
-		uv_reply_to_message(sw_ack_slot, msg, msp);
+	return;
 }
 
 /*
- * Examine the payload queue on one distribution node to see
- * which messages have not been seen, and which cpu(s) have not seen them.
+ * Determine the first cpu on a uvhub.
+ */
+static int uvhub_to_first_cpu(int uvhub)
+{
+	int cpu;
+	for_each_present_cpu(cpu)
+		if (uvhub == uv_cpu_to_blade_id(cpu))
+			return cpu;
+	return -1;
+}
+
+/*
+ * Last resort when we get a large number of destination timeouts is
+ * to clear resources held by a given cpu.
+ * Do this with IPI so that all messages in the BAU message queue
+ * can be identified by their nonzero sw_ack_vector field.
  *
- * Returns the number of cpu's that have not responded.
+ * This is entered for a single cpu on the uvhub.
+ * The sender want's this uvhub to free a specific message's
+ * sw_ack resources.
  */
-static int uv_examine_destination(struct bau_control *bau_tablesp, int sender)
+static void
+uv_do_reset(void *ptr)
 {
-	struct bau_payload_queue_entry *msg;
-	struct bau_msg_status *msp;
-	int count = 0;
 	int i;
-	int j;
+	int slot;
+	int count = 0;
+	unsigned long mmr;
+	unsigned long msg_res;
+	struct bau_control *bcp;
+	struct reset_args *rap;
+	struct bau_payload_queue_entry *msg;
+	struct ptc_stats *stat;
 
-	for (msg = bau_tablesp->va_queue_first, i = 0; i < DEST_Q_SIZE;
-	     msg++, i++) {
-		if ((msg->sending_cpu == sender) && (!msg->replied_to)) {
-			msp = bau_tablesp->msg_statuses + i;
-			printk(KERN_DEBUG
-			       "blade %d: address:%#lx %d of %d, not cpu(s): ",
-			       i, msg->address, msg->acknowledge_count,
-			       msg->number_of_cpus);
-			for (j = 0; j < msg->number_of_cpus; j++) {
-				if (!((1L << j) & msp->seen_by.bits)) {
-					count++;
-					printk("%d ", j);
-				}
+	bcp = &per_cpu(bau_control, smp_processor_id());
+	rap = (struct reset_args *)ptr;
+	stat = &per_cpu(ptcstats, bcp->cpu);
+	stat->d_resets++;
+
+	/*
+	 * We're looking for the given sender, and
+	 * will free its sw_ack resource.
+	 * If all cpu's finally responded after the timeout, its
+	 * message 'replied_to' was set.
+	 */
+	for (msg = bcp->va_queue_first, i = 0; i < DEST_Q_SIZE; msg++, i++) {
+		/* uv_do_reset: same conditions for cancellation as
+		   uv_bau_process_retry_msg() */
+		if ((msg->replied_to == 0) &&
+		    (msg->canceled == 0) &&
+		    (msg->sending_cpu == rap->sender) &&
+		    (msg->sw_ack_vector) &&
+		    (msg->msg_type != MSG_NOOP)) {
+			/*
+			 * make everyone else ignore this message
+			 */
+			msg->canceled = 1;
+			slot = msg - bcp->va_queue_first;
+			count++;
+			/*
+			 * only reset the resource if it is still pending
+			 */
+			mmr = uv_read_local_mmr
+					(UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE);
+			msg_res = ((msg->sw_ack_vector << 8) |
+						   msg->sw_ack_vector);
+			if (mmr & msg_res) {
+				stat->d_rcanceled++;
+				uv_write_local_mmr(
+				    UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS,
+							msg_res);
 			}
-			printk("\n");
 		}
 	}
-	return count;
+	return;
 }
 
 /*
- * Examine the payload queue on all the distribution nodes to see
- * which messages have not been seen, and which cpu(s) have not seen them.
- *
- * Returns the number of cpu's that have not responded.
+ * Use IPI to get all target uvhubs to release resources held by
+ * a given sending cpu number.
  */
-static int uv_examine_destinations(struct bau_target_nodemask *distribution)
+static void uv_reset_with_ipi(struct bau_target_uvhubmask *distribution,
+			      int sender)
 {
-	int sender;
-	int i;
-	int count = 0;
+	int uvhub;
+	int cpu;
+	cpumask_t mask;
+	struct reset_args reset_args;
+
+	reset_args.sender = sender;
 
-	sender = smp_processor_id();
-	for (i = 0; i < sizeof(struct bau_target_nodemask) * BITSPERBYTE; i++) {
-		if (!bau_node_isset(i, distribution))
+	cpus_clear(mask);
+	/* find a single cpu for each uvhub in this distribution mask */
+	for (uvhub = 0;
+		    uvhub < sizeof(struct bau_target_uvhubmask) * BITSPERBYTE;
+		    uvhub++) {
+		if (!bau_uvhub_isset(uvhub, distribution))
 			continue;
-		count += uv_examine_destination(uv_bau_table_bases[i], sender);
+		/* find a cpu for this uvhub */
+		cpu = uvhub_to_first_cpu(uvhub);
+		cpu_set(cpu, mask);
 	}
-	return count;
+	/* IPI all cpus; Preemption is already disabled */
+	smp_call_function_many(&mask, uv_do_reset, (void *)&reset_args, 1);
+	return;
+}
+
+static inline unsigned long
+cycles_2_us(unsigned long long cyc)
+{
+	unsigned long long ns;
+	unsigned long us;
+	ns =  (cyc * per_cpu(cyc2ns, smp_processor_id()))
+						>> CYC2NS_SCALE_FACTOR;
+	us = ns / 1000;
+	return us;
 }
 
 /*
- * wait for completion of a broadcast message
- *
- * return COMPLETE, RETRY or GIVEUP
+ * wait for all cpus on this hub to finish their sends and go quiet
+ * leaves uvhub_quiesce set so that no new broadcasts are started by
+ * bau_flush_send_and_wait()
+ */
+static inline void
+quiesce_local_uvhub(struct bau_control *hmaster)
+{
+	atomic_add_short_return(1, (struct atomic_short *)
+		 &hmaster->uvhub_quiesce);
+}
+
+/*
+ * mark this quiet-requestor as done
+ */
+static inline void
+end_uvhub_quiesce(struct bau_control *hmaster)
+{
+	atomic_add_short_return(-1, (struct atomic_short *)
+		&hmaster->uvhub_quiesce);
+}
+
+/*
+ * Wait for completion of a broadcast software ack message
+ * return COMPLETE, RETRY(PLUGGED or TIMEOUT) or GIVEUP
  */
 static int uv_wait_completion(struct bau_desc *bau_desc,
-			      unsigned long mmr_offset, int right_shift)
+	unsigned long mmr_offset, int right_shift, int this_cpu,
+	struct bau_control *bcp, struct bau_control *smaster, long try)
 {
-	int exams = 0;
-	long destination_timeouts = 0;
-	long source_timeouts = 0;
+	int relaxes = 0;
 	unsigned long descriptor_status;
+	unsigned long mmr;
+	unsigned long mask;
+	cycles_t ttime;
+	cycles_t timeout_time;
+	struct ptc_stats *stat = &per_cpu(ptcstats, this_cpu);
+	struct bau_control *hmaster;
+
+	hmaster = bcp->uvhub_master;
+	timeout_time = get_cycles() + bcp->timeout_interval;
 
+	/* spin on the status MMR, waiting for it to go idle */
 	while ((descriptor_status = (((unsigned long)
 		uv_read_local_mmr(mmr_offset) >>
 			right_shift) & UV_ACT_STATUS_MASK)) !=
 			DESC_STATUS_IDLE) {
-		if (descriptor_status == DESC_STATUS_SOURCE_TIMEOUT) {
-			source_timeouts++;
-			if (source_timeouts > SOURCE_TIMEOUT_LIMIT)
-				source_timeouts = 0;
-			__get_cpu_var(ptcstats).s_retry++;
-			return FLUSH_RETRY;
-		}
 		/*
-		 * spin here looking for progress at the destinations
+		 * Our software ack messages may be blocked because there are
+		 * no swack resources available.  As long as none of them
+		 * has timed out hardware will NACK our message and its
+		 * state will stay IDLE.
 		 */
-		if (descriptor_status == DESC_STATUS_DESTINATION_TIMEOUT) {
-			destination_timeouts++;
-			if (destination_timeouts > DESTINATION_TIMEOUT_LIMIT) {
-				/*
-				 * returns number of cpus not responding
-				 */
-				if (uv_examine_destinations
-				    (&bau_desc->distribution) == 0) {
-					__get_cpu_var(ptcstats).d_retry++;
-					return FLUSH_RETRY;
-				}
-				exams++;
-				if (exams >= uv_bau_retry_limit) {
-					printk(KERN_DEBUG
-					       "uv_flush_tlb_others");
-					printk("giving up on cpu %d\n",
-					       smp_processor_id());
+		if (descriptor_status == DESC_STATUS_SOURCE_TIMEOUT) {
+			stat->s_stimeout++;
+			return FLUSH_GIVEUP;
+		} else if (descriptor_status ==
+					DESC_STATUS_DESTINATION_TIMEOUT) {
+			stat->s_dtimeout++;
+			ttime = get_cycles();
+
+			/*
+			 * Our retries may be blocked by all destination
+			 * swack resources being consumed, and a timeout
+			 * pending.  In that case hardware returns the
+			 * ERROR that looks like a destination timeout.
+			 */
+			if (cycles_2_us(ttime - bcp->send_message) < BIOS_TO) {
+				bcp->conseccompletes = 0;
+				return FLUSH_RETRY_PLUGGED;
+			}
+
+			bcp->conseccompletes = 0;
+			return FLUSH_RETRY_TIMEOUT;
+		} else {
+			/*
+			 * descriptor_status is still BUSY
+			 */
+			cpu_relax();
+			relaxes++;
+			if (relaxes >= 10000) {
+				relaxes = 0;
+				if (get_cycles() > timeout_time) {
+					quiesce_local_uvhub(hmaster);
+
+					/* single-thread the register change */
+					spin_lock(&hmaster->masks_lock);
+					mmr = uv_read_local_mmr(mmr_offset);
+					mask = 0UL;
+					mask |= (3UL < right_shift);
+					mask = ~mask;
+					mmr &= mask;
+					uv_write_local_mmr(mmr_offset, mmr);
+					spin_unlock(&hmaster->masks_lock);
+					end_uvhub_quiesce(hmaster);
+					stat->s_busy++;
 					return FLUSH_GIVEUP;
 				}
-				/*
-				 * delays can hang the simulator
-				   udelay(1000);
-				 */
-				destination_timeouts = 0;
 			}
 		}
-		cpu_relax();
 	}
+	bcp->conseccompletes++;
 	return FLUSH_COMPLETE;
 }
 
+static inline cycles_t
+sec_2_cycles(unsigned long sec)
+{
+	unsigned long ns;
+	cycles_t cyc;
+
+	ns = sec * 1000000000;
+	cyc = (ns << CYC2NS_SCALE_FACTOR)/(per_cpu(cyc2ns, smp_processor_id()));
+	return cyc;
+}
+
+/*
+ * conditionally add 1 to *v, unless *v is >= u
+ * return 0 if we cannot add 1 to *v because it is >= u
+ * return 1 if we can add 1 to *v because it is < u
+ * the add is atomic
+ *
+ * This is close to atomic_add_unless(), but this allows the 'u' value
+ * to be lowered below the current 'v'.  atomic_add_unless can only stop
+ * on equal.
+ */
+static inline int atomic_inc_unless_ge(spinlock_t *lock, atomic_t *v, int u)
+{
+	spin_lock(lock);
+	if (atomic_read(v) >= u) {
+		spin_unlock(lock);
+		return 0;
+	}
+	atomic_inc(v);
+	spin_unlock(lock);
+	return 1;
+}
+
 /**
  * uv_flush_send_and_wait
  *
- * Send a broadcast and wait for a broadcast message to complete.
+ * Send a broadcast and wait for it to complete.
  *
- * The flush_mask contains the cpus the broadcast was sent to.
+ * The flush_mask contains the cpus the broadcast is to be sent to, plus
+ * cpus that are on the local uvhub.
  *
- * Returns NULL if all remote flushing was done. The mask is zeroed.
+ * Returns NULL if all flushing represented in the mask was done. The mask
+ * is zeroed.
  * Returns @flush_mask if some remote flushing remains to be done. The
- * mask will have some bits still set.
+ * mask will have some bits still set, representing any cpus on the local
+ * uvhub (not current cpu) and any on remote uvhubs if the broadcast failed.
  */
-const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode,
-					     struct bau_desc *bau_desc,
-					     struct cpumask *flush_mask)
+const struct cpumask *uv_flush_send_and_wait(struct bau_desc *bau_desc,
+					     struct cpumask *flush_mask,
+					     struct bau_control *bcp)
 {
-	int completion_status = 0;
 	int right_shift;
-	int tries = 0;
-	int pnode;
+	int uvhub;
 	int bit;
+	int completion_status = 0;
+	int seq_number = 0;
+	long try = 0;
+	int cpu = bcp->uvhub_cpu;
+	int this_cpu = bcp->cpu;
+	int this_uvhub = bcp->uvhub;
 	unsigned long mmr_offset;
 	unsigned long index;
 	cycles_t time1;
 	cycles_t time2;
+	struct ptc_stats *stat = &per_cpu(ptcstats, bcp->cpu);
+	struct bau_control *smaster = bcp->socket_master;
+	struct bau_control *hmaster = bcp->uvhub_master;
+
+	/*
+	 * Spin here while there are hmaster->max_concurrent or more active
+	 * descriptors. This is the per-uvhub 'throttle'.
+	 */
+	if (!atomic_inc_unless_ge(&hmaster->uvhub_lock,
+			&hmaster->active_descriptor_count,
+			hmaster->max_concurrent)) {
+		stat->s_throttles++;
+		do {
+			cpu_relax();
+		} while (!atomic_inc_unless_ge(&hmaster->uvhub_lock,
+			&hmaster->active_descriptor_count,
+			hmaster->max_concurrent));
+	}
+
+	while (hmaster->uvhub_quiesce)
+		cpu_relax();
 
 	if (cpu < UV_CPUS_PER_ACT_STATUS) {
 		mmr_offset = UVH_LB_BAU_SB_ACTIVATION_STATUS_0;
@@ -268,24 +557,108 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode,
 	}
 	time1 = get_cycles();
 	do {
-		tries++;
+		/*
+		 * Every message from any given cpu gets a unique message
+		 * sequence number. But retries use that same number.
+		 * Our message may have timed out at the destination because
+		 * all sw-ack resources are in use and there is a timeout
+		 * pending there.  In that case, our last send never got
+		 * placed into the queue and we need to persist until it
+		 * does.
+		 *
+		 * Make any retry a type MSG_RETRY so that the destination will
+		 * free any resource held by a previous message from this cpu.
+		 */
+		if (try == 0) {
+			/* use message type set by the caller the first time */
+			seq_number = bcp->message_number++;
+		} else {
+			/* use RETRY type on all the rest; same sequence */
+			bau_desc->header.msg_type = MSG_RETRY;
+			stat->s_retry_messages++;
+		}
+		bau_desc->header.sequence = seq_number;
 		index = (1UL << UVH_LB_BAU_SB_ACTIVATION_CONTROL_PUSH_SHFT) |
-			cpu;
+			bcp->uvhub_cpu;
+		bcp->send_message = get_cycles();
+
 		uv_write_local_mmr(UVH_LB_BAU_SB_ACTIVATION_CONTROL, index);
+
+		try++;
 		completion_status = uv_wait_completion(bau_desc, mmr_offset,
-					right_shift);
-	} while (completion_status == FLUSH_RETRY);
+			right_shift, this_cpu, bcp, smaster, try);
+
+		if (completion_status == FLUSH_RETRY_PLUGGED) {
+			/*
+			 * Our retries may be blocked by all destination swack
+			 * resources being consumed, and a timeout pending. In
+			 * that case hardware immediately returns the ERROR
+			 * that looks like a destination timeout.
+			 */
+			udelay(TIMEOUT_DELAY);
+			bcp->plugged_tries++;
+			if (bcp->plugged_tries >= PLUGSB4RESET) {
+				bcp->plugged_tries = 0;
+				quiesce_local_uvhub(hmaster);
+				spin_lock(&hmaster->queue_lock);
+				uv_reset_with_ipi(&bau_desc->distribution,
+							this_cpu);
+				spin_unlock(&hmaster->queue_lock);
+				end_uvhub_quiesce(hmaster);
+				bcp->ipi_attempts++;
+				stat->s_resets_plug++;
+			}
+		} else if (completion_status == FLUSH_RETRY_TIMEOUT) {
+			hmaster->max_concurrent = 1;
+			bcp->timeout_tries++;
+			udelay(TIMEOUT_DELAY);
+			if (bcp->timeout_tries >= TIMEOUTSB4RESET) {
+				bcp->timeout_tries = 0;
+				quiesce_local_uvhub(hmaster);
+				spin_lock(&hmaster->queue_lock);
+				uv_reset_with_ipi(&bau_desc->distribution,
+								this_cpu);
+				spin_unlock(&hmaster->queue_lock);
+				end_uvhub_quiesce(hmaster);
+				bcp->ipi_attempts++;
+				stat->s_resets_timeout++;
+			}
+		}
+		if (bcp->ipi_attempts >= 3) {
+			bcp->ipi_attempts = 0;
+			completion_status = FLUSH_GIVEUP;
+			break;
+		}
+		cpu_relax();
+	} while ((completion_status == FLUSH_RETRY_PLUGGED) ||
+		 (completion_status == FLUSH_RETRY_TIMEOUT));
 	time2 = get_cycles();
-	__get_cpu_var(ptcstats).sflush += (time2 - time1);
-	if (tries > 1)
-		__get_cpu_var(ptcstats).retriesok++;
 
-	if (completion_status == FLUSH_GIVEUP) {
+	if ((completion_status == FLUSH_COMPLETE) && (bcp->conseccompletes > 5)
+	    && (hmaster->max_concurrent < hmaster->max_concurrent_constant))
+			hmaster->max_concurrent++;
+
+	/*
+	 * hold any cpu not timing out here; no other cpu currently held by
+	 * the 'throttle' should enter the activation code
+	 */
+	while (hmaster->uvhub_quiesce)
+		cpu_relax();
+	atomic_dec(&hmaster->active_descriptor_count);
+
+	/* guard against cycles wrap */
+	if (time2 > time1)
+		stat->s_time += (time2 - time1);
+	else
+		stat->s_requestor--; /* don't count this one */
+	if (completion_status == FLUSH_COMPLETE && try > 1)
+		stat->s_retriesok++;
+	else if (completion_status == FLUSH_GIVEUP) {
 		/*
 		 * Cause the caller to do an IPI-style TLB shootdown on
-		 * the cpu's, all of which are still in the mask.
+		 * the target cpu's, all of which are still in the mask.
 		 */
-		__get_cpu_var(ptcstats).ptc_i++;
+		stat->s_giveup++;
 		return flush_mask;
 	}
 
@@ -294,18 +667,17 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode,
 	 * use the IPI method of shootdown on them.
 	 */
 	for_each_cpu(bit, flush_mask) {
-		pnode = uv_cpu_to_pnode(bit);
-		if (pnode == this_pnode)
+		uvhub = uv_cpu_to_blade_id(bit);
+		if (uvhub == this_uvhub)
 			continue;
 		cpumask_clear_cpu(bit, flush_mask);
 	}
 	if (!cpumask_empty(flush_mask))
 		return flush_mask;
+
 	return NULL;
 }
 
-static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask);
-
 /**
  * uv_flush_tlb_others - globally purge translation cache of a virtual
  * address or all TLB's
@@ -322,8 +694,8 @@ static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask);
  * The caller has derived the cpumask from the mm_struct.  This function
  * is called only if there are bits set in the mask. (e.g. flush_tlb_page())
  *
- * The cpumask is converted into a nodemask of the nodes containing
- * the cpus.
+ * The cpumask is converted into a uvhubmask of the uvhubs containing
+ * those cpus.
  *
  * Note that this function should be called with preemption disabled.
  *
@@ -335,52 +707,82 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
 					  struct mm_struct *mm,
 					  unsigned long va, unsigned int cpu)
 {
-	struct cpumask *flush_mask = __get_cpu_var(uv_flush_tlb_mask);
-	int i;
-	int bit;
-	int pnode;
-	int uv_cpu;
-	int this_pnode;
+	int remotes;
+	int tcpu;
+	int uvhub;
 	int locals = 0;
 	struct bau_desc *bau_desc;
+	struct cpumask *flush_mask;
+	struct ptc_stats *stat;
+	struct bau_control *bcp;
 
-	cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu));
+	if (nobau)
+		return cpumask;
 
-	uv_cpu = uv_blade_processor_id();
-	this_pnode = uv_hub_info->pnode;
-	bau_desc = __get_cpu_var(bau_control).descriptor_base;
-	bau_desc += UV_ITEMS_PER_DESCRIPTOR * uv_cpu;
+	bcp = &per_cpu(bau_control, cpu);
+	/*
+	 * Each sending cpu has a per-cpu mask which it fills from the caller's
+	 * cpu mask.  Only remote cpus are converted to uvhubs and copied.
+	 */
+	flush_mask = (struct cpumask *)per_cpu(uv_flush_tlb_mask, cpu);
+	/*
+	 * copy cpumask to flush_mask, removing current cpu
+	 * (current cpu should already have been flushed by the caller and
+	 *  should never be returned if we return flush_mask)
+	 */
+	cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu));
+	if (cpu_isset(cpu, *cpumask))
+		locals++;  /* current cpu was targeted */
 
-	bau_nodes_clear(&bau_desc->distribution, UV_DISTRIBUTION_SIZE);
+	bau_desc = bcp->descriptor_base;
+	bau_desc += UV_ITEMS_PER_DESCRIPTOR * bcp->uvhub_cpu;
 
-	i = 0;
-	for_each_cpu(bit, flush_mask) {
-		pnode = uv_cpu_to_pnode(bit);
-		BUG_ON(pnode > (UV_DISTRIBUTION_SIZE - 1));
-		if (pnode == this_pnode) {
+	bau_uvhubs_clear(&bau_desc->distribution, UV_DISTRIBUTION_SIZE);
+	remotes = 0;
+	for_each_cpu(tcpu, flush_mask) {
+		uvhub = uv_cpu_to_blade_id(tcpu);
+		if (uvhub == bcp->uvhub) {
 			locals++;
 			continue;
 		}
-		bau_node_set(pnode - uv_partition_base_pnode,
-				&bau_desc->distribution);
-		i++;
+		bau_uvhub_set(uvhub, &bau_desc->distribution);
+		remotes++;
 	}
-	if (i == 0) {
+	if (remotes == 0) {
 		/*
-		 * no off_node flushing; return status for local node
+		 * No off_hub flushing; return status for local hub.
+		 * Return the caller's mask if all were local (the current
+		 * cpu may be in that mask).
 		 */
 		if (locals)
-			return flush_mask;
+			return cpumask;
 		else
 			return NULL;
 	}
-	__get_cpu_var(ptcstats).requestor++;
-	__get_cpu_var(ptcstats).ntargeted += i;
+	stat = &per_cpu(ptcstats, cpu);
+	stat->s_requestor++;
+	stat->s_ntargcpu += remotes;
+	remotes = bau_uvhub_weight(&bau_desc->distribution);
+	stat->s_ntarguvhub += remotes;
+	if (remotes >= 16)
+		stat->s_ntarguvhub16++;
+	else if (remotes >= 8)
+		stat->s_ntarguvhub8++;
+	else if (remotes >= 4)
+		stat->s_ntarguvhub4++;
+	else if (remotes >= 2)
+		stat->s_ntarguvhub2++;
+	else
+		stat->s_ntarguvhub1++;
 
 	bau_desc->payload.address = va;
 	bau_desc->payload.sending_cpu = cpu;
 
-	return uv_flush_send_and_wait(uv_cpu, this_pnode, bau_desc, flush_mask);
+	/*
+	 * uv_flush_send_and_wait returns null if all cpu's were messaged, or
+	 * the adjusted flush_mask if any cpu's were not messaged.
+	 */
+	return uv_flush_send_and_wait(bau_desc, flush_mask, bcp);
 }
 
 /*
@@ -389,87 +791,70 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
  *
  * We received a broadcast assist message.
  *
- * Interrupts may have been disabled; this interrupt could represent
+ * Interrupts are disabled; this interrupt could represent
  * the receipt of several messages.
  *
- * All cores/threads on this node get this interrupt.
- * The last one to see it does the s/w ack.
+ * All cores/threads on this hub get this interrupt.
+ * The last one to see it does the software ack.
  * (the resource will not be freed until noninterruptable cpus see this
- *  interrupt; hardware will timeout the s/w ack and reply ERROR)
+ *  interrupt; hardware may timeout the s/w ack and reply ERROR)
  */
 void uv_bau_message_interrupt(struct pt_regs *regs)
 {
-	struct bau_payload_queue_entry *va_queue_first;
-	struct bau_payload_queue_entry *va_queue_last;
-	struct bau_payload_queue_entry *msg;
-	struct pt_regs *old_regs = set_irq_regs(regs);
-	cycles_t time1;
-	cycles_t time2;
-	int msg_slot;
-	int sw_ack_slot;
-	int fw;
 	int count = 0;
-	unsigned long local_pnode;
-
-	ack_APIC_irq();
-	exit_idle();
-	irq_enter();
-
-	time1 = get_cycles();
-
-	local_pnode = uv_blade_to_pnode(uv_numa_blade_id());
-
-	va_queue_first = __get_cpu_var(bau_control).va_queue_first;
-	va_queue_last = __get_cpu_var(bau_control).va_queue_last;
-
-	msg = __get_cpu_var(bau_control).bau_msg_head;
+	cycles_t time_start;
+	struct bau_payload_queue_entry *msg;
+	struct bau_control *bcp;
+	struct ptc_stats *stat;
+	struct msg_desc msgdesc;
+
+	time_start = get_cycles();
+	bcp = &per_cpu(bau_control, smp_processor_id());
+	stat = &per_cpu(ptcstats, smp_processor_id());
+	msgdesc.va_queue_first = bcp->va_queue_first;
+	msgdesc.va_queue_last = bcp->va_queue_last;
+	msg = bcp->bau_msg_head;
 	while (msg->sw_ack_vector) {
 		count++;
-		fw = msg->sw_ack_vector;
-		msg_slot = msg - va_queue_first;
-		sw_ack_slot = ffs(fw) - 1;
-
-		uv_bau_process_message(msg, msg_slot, sw_ack_slot);
-
+		msgdesc.msg_slot = msg - msgdesc.va_queue_first;
+		msgdesc.sw_ack_slot = ffs(msg->sw_ack_vector) - 1;
+		msgdesc.msg = msg;
+		uv_bau_process_message(&msgdesc, bcp);
 		msg++;
-		if (msg > va_queue_last)
-			msg = va_queue_first;
-		__get_cpu_var(bau_control).bau_msg_head = msg;
+		if (msg > msgdesc.va_queue_last)
+			msg = msgdesc.va_queue_first;
+		bcp->bau_msg_head = msg;
 	}
+	stat->d_time += (get_cycles() - time_start);
 	if (!count)
-		__get_cpu_var(ptcstats).nomsg++;
+		stat->d_nomsg++;
 	else if (count > 1)
-		__get_cpu_var(ptcstats).multmsg++;
-
-	time2 = get_cycles();
-	__get_cpu_var(ptcstats).dflush += (time2 - time1);
-
-	irq_exit();
-	set_irq_regs(old_regs);
+		stat->d_multmsg++;
+	ack_APIC_irq();
 }
 
 /*
  * uv_enable_timeouts
  *
- * Each target blade (i.e. blades that have cpu's) needs to have
+ * Each target uvhub (i.e. a uvhub that has no cpu's) needs to have
  * shootdown message timeouts enabled.  The timeout does not cause
  * an interrupt, but causes an error message to be returned to
  * the sender.
  */
 static void uv_enable_timeouts(void)
 {
-	int blade;
-	int nblades;
+	int uvhub;
+	int nuvhubs;
 	int pnode;
 	unsigned long mmr_image;
 
-	nblades = uv_num_possible_blades();
+	nuvhubs = uv_num_possible_blades();
 
-	for (blade = 0; blade < nblades; blade++) {
-		if (!uv_blade_nr_possible_cpus(blade))
+	for (uvhub = 0; uvhub < nuvhubs; uvhub++) {
+		if (!uv_blade_nr_possible_cpus(uvhub))
 			continue;
 
-		pnode = uv_blade_to_pnode(blade);
+		pnode = uv_blade_to_pnode(uvhub);
 		mmr_image =
 		    uv_read_global_mmr64(pnode, UVH_LB_BAU_MISC_CONTROL);
 		/*
@@ -479,16 +864,16 @@ static void uv_enable_timeouts(void)
 		 * To program the period, the SOFT_ACK_MODE must be off.
 		 */
 		mmr_image &= ~((unsigned long)1 <<
-			       UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT);
+		    UVH_LB_BAU_MISC_CONTROL_ENABLE_INTD_SOFT_ACK_MODE_SHFT);
 		uv_write_global_mmr64
 		    (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
 		/*
 		 * Set the 4-bit period.
 		 */
 		mmr_image &= ~((unsigned long)0xf <<
-			UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT);
+		     UVH_LB_BAU_MISC_CONTROL_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHFT);
 		mmr_image |= (UV_INTD_SOFT_ACK_TIMEOUT_PERIOD <<
-			     UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT);
+		     UVH_LB_BAU_MISC_CONTROL_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHFT);
 		uv_write_global_mmr64
 		    (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
 		/*
@@ -497,7 +882,7 @@ static void uv_enable_timeouts(void)
 		 * indicated in bits 2:0 (7 causes all of them to timeout).
 		 */
 		mmr_image |= ((unsigned long)1 <<
-			      UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT);
+		    UVH_LB_BAU_MISC_CONTROL_ENABLE_INTD_SOFT_ACK_MODE_SHFT);
 		uv_write_global_mmr64
 		    (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
 	}
@@ -522,9 +907,20 @@ static void uv_ptc_seq_stop(struct seq_file *file, void *data)
 {
 }
 
+static inline unsigned long long
+millisec_2_cycles(unsigned long millisec)
+{
+	unsigned long ns;
+	unsigned long long cyc;
+
+	ns = millisec * 1000;
+	cyc = (ns << CYC2NS_SCALE_FACTOR)/(per_cpu(cyc2ns, smp_processor_id()));
+	return cyc;
+}
+
 /*
- * Display the statistics thru /proc
- * data points to the cpu number
+ * Display the statistics thru /proc.
+ * 'data' points to the cpu number
  */
 static int uv_ptc_seq_show(struct seq_file *file, void *data)
 {
@@ -535,78 +931,155 @@ static int uv_ptc_seq_show(struct seq_file *file, void *data)
 
 	if (!cpu) {
 		seq_printf(file,
-		"# cpu requestor requestee one all sretry dretry ptc_i ");
+			"# cpu sent stime numuvhubs numuvhubs16 numuvhubs8 ");
 		seq_printf(file,
-		"sw_ack sflush dflush sok dnomsg dmult starget\n");
+			"numuvhubs4 numuvhubs2 numuvhubs1 numcpus dto ");
+		seq_printf(file,
+			"retries rok resetp resett giveup sto bz throt ");
+		seq_printf(file,
+			"sw_ack recv rtime all ");
+		seq_printf(file,
+			"one mult none retry canc nocan reset rcan\n");
 	}
 	if (cpu < num_possible_cpus() && cpu_online(cpu)) {
 		stat = &per_cpu(ptcstats, cpu);
-		seq_printf(file, "cpu %d %ld %ld %ld %ld %ld %ld %ld ",
-			   cpu, stat->requestor,
-			   stat->requestee, stat->onetlb, stat->alltlb,
-			   stat->s_retry, stat->d_retry, stat->ptc_i);
-		seq_printf(file, "%lx %ld %ld %ld %ld %ld %ld\n",
+		/* source side statistics */
+		seq_printf(file,
+			"cpu %d %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld ",
+			   cpu, stat->s_requestor, cycles_2_us(stat->s_time),
+			   stat->s_ntarguvhub, stat->s_ntarguvhub16,
+			   stat->s_ntarguvhub8, stat->s_ntarguvhub4,
+			   stat->s_ntarguvhub2, stat->s_ntarguvhub1,
+			   stat->s_ntargcpu, stat->s_dtimeout);
+		seq_printf(file, "%ld %ld %ld %ld %ld %ld %ld %ld ",
+			   stat->s_retry_messages, stat->s_retriesok,
+			   stat->s_resets_plug, stat->s_resets_timeout,
+			   stat->s_giveup, stat->s_stimeout,
+			   stat->s_busy, stat->s_throttles);
+		/* destination side statistics */
+		seq_printf(file,
+			   "%lx %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld\n",
 			   uv_read_global_mmr64(uv_cpu_to_pnode(cpu),
 					UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE),
-			   stat->sflush, stat->dflush,
-			   stat->retriesok, stat->nomsg,
-			   stat->multmsg, stat->ntargeted);
+			   stat->d_requestee, cycles_2_us(stat->d_time),
+			   stat->d_alltlb, stat->d_onetlb, stat->d_multmsg,
+			   stat->d_nomsg, stat->d_retries, stat->d_canceled,
+			   stat->d_nocanceled, stat->d_resets,
+			   stat->d_rcanceled);
 	}
 
 	return 0;
 }
 
 /*
+ * -1: resetf the statistics
  *  0: display meaning of the statistics
- * >0: retry limit
+ * >0: maximum concurrent active descriptors per uvhub (throttle)
  */
 static ssize_t uv_ptc_proc_write(struct file *file, const char __user *user,
 				 size_t count, loff_t *data)
 {
-	long newmode;
+	int cpu;
+	long input_arg;
 	char optstr[64];
+	struct ptc_stats *stat;
+	struct bau_control *bcp;
 
 	if (count == 0 || count > sizeof(optstr))
 		return -EINVAL;
 	if (copy_from_user(optstr, user, count))
 		return -EFAULT;
 	optstr[count - 1] = '\0';
-	if (strict_strtoul(optstr, 10, &newmode) < 0) {
+	if (strict_strtol(optstr, 10, &input_arg) < 0) {
 		printk(KERN_DEBUG "%s is invalid\n", optstr);
 		return -EINVAL;
 	}
 
-	if (newmode == 0) {
+	if (input_arg == 0) {
 		printk(KERN_DEBUG "# cpu:      cpu number\n");
+		printk(KERN_DEBUG "Sender statistics:\n");
+		printk(KERN_DEBUG
+		"sent:     number of shootdown messages sent\n");
+		printk(KERN_DEBUG
+		"stime:    time spent sending messages\n");
+		printk(KERN_DEBUG
+		"numuvhubs: number of hubs targeted with shootdown\n");
+		printk(KERN_DEBUG
+		"numuvhubs16: number times 16 or more hubs targeted\n");
+		printk(KERN_DEBUG
+		"numuvhubs8: number times 8 or more hubs targeted\n");
+		printk(KERN_DEBUG
+		"numuvhubs4: number times 4 or more hubs targeted\n");
+		printk(KERN_DEBUG
+		"numuvhubs2: number times 2 or more hubs targeted\n");
+		printk(KERN_DEBUG
+		"numuvhubs1: number times 1 hub targeted\n");
+		printk(KERN_DEBUG
+		"numcpus:  number of cpus targeted with shootdown\n");
+		printk(KERN_DEBUG
+		"dto:      number of destination timeouts\n");
+		printk(KERN_DEBUG
+		"retries:  destination timeout retries sent\n");
+		printk(KERN_DEBUG
+		"rok:   :  destination timeouts successfully retried\n");
+		printk(KERN_DEBUG
+		"resetp:   ipi-style resource resets for plugs\n");
+		printk(KERN_DEBUG
+		"resett:   ipi-style resource resets for timeouts\n");
+		printk(KERN_DEBUG
+		"giveup:   fall-backs to ipi-style shootdowns\n");
+		printk(KERN_DEBUG
+		"sto:      number of source timeouts\n");
+		printk(KERN_DEBUG
+		"bz:       number of stay-busy's\n");
+		printk(KERN_DEBUG
+		"throt:    number times spun in throttle\n");
+		printk(KERN_DEBUG "Destination side statistics:\n");
 		printk(KERN_DEBUG
-		"requestor:  times this cpu was the flush requestor\n");
+		"sw_ack:   image of UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE\n");
 		printk(KERN_DEBUG
-		"requestee:  times this cpu was requested to flush its TLBs\n");
+		"recv:     shootdown messages received\n");
 		printk(KERN_DEBUG
-		"one:        times requested to flush a single address\n");
+		"rtime:    time spent processing messages\n");
 		printk(KERN_DEBUG
-		"all:        times requested to flush all TLB's\n");
+		"all:      shootdown all-tlb messages\n");
 		printk(KERN_DEBUG
-		"sretry:     number of retries of source-side timeouts\n");
+		"one:      shootdown one-tlb messages\n");
 		printk(KERN_DEBUG
-		"dretry:     number of retries of destination-side timeouts\n");
+		"mult:     interrupts that found multiple messages\n");
 		printk(KERN_DEBUG
-		"ptc_i:      times UV fell through to IPI-style flushes\n");
+		"none:     interrupts that found no messages\n");
 		printk(KERN_DEBUG
-		"sw_ack:     image of UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE\n");
+		"retry:    number of retry messages processed\n");
 		printk(KERN_DEBUG
-		"sflush_us:  cycles spent in uv_flush_tlb_others()\n");
+		"canc:     number messages canceled by retries\n");
 		printk(KERN_DEBUG
-		"dflush_us:  cycles spent in handling flush requests\n");
-		printk(KERN_DEBUG "sok:        successes on retry\n");
-		printk(KERN_DEBUG "dnomsg:     interrupts with no message\n");
+		"nocan:    number retries that found nothing to cancel\n");
 		printk(KERN_DEBUG
-		"dmult:      interrupts with multiple messages\n");
-		printk(KERN_DEBUG "starget:    nodes targeted\n");
+		"reset:    number of ipi-style reset requests processed\n");
+		printk(KERN_DEBUG
+		"rcan:     number messages canceled by reset requests\n");
+	} else if (input_arg == -1) {
+		for_each_present_cpu(cpu) {
+			stat = &per_cpu(ptcstats, cpu);
+			memset(stat, 0, sizeof(struct ptc_stats));
+		}
 	} else {
-		uv_bau_retry_limit = newmode;
-		printk(KERN_DEBUG "timeout retry limit:%d\n",
-		       uv_bau_retry_limit);
+		uv_bau_max_concurrent = input_arg;
+		bcp = &per_cpu(bau_control, smp_processor_id());
+		if (uv_bau_max_concurrent < 1 ||
+		    uv_bau_max_concurrent > bcp->cpus_in_uvhub) {
+			printk(KERN_DEBUG
+				"Error: BAU max concurrent %d; %d is invalid\n",
+				bcp->max_concurrent, uv_bau_max_concurrent);
+			return -EINVAL;
+		}
+		printk(KERN_DEBUG "Set BAU max concurrent:%d\n",
+		       uv_bau_max_concurrent);
+		for_each_present_cpu(cpu) {
+			bcp = &per_cpu(bau_control, cpu);
+			bcp->max_concurrent = uv_bau_max_concurrent;
+		}
 	}
 
 	return count;
@@ -650,79 +1123,30 @@ static int __init uv_ptc_init(void)
 }
 
 /*
- * begin the initialization of the per-blade control structures
- */
-static struct bau_control * __init uv_table_bases_init(int blade, int node)
-{
-	int i;
-	struct bau_msg_status *msp;
-	struct bau_control *bau_tabp;
-
-	bau_tabp =
-	    kmalloc_node(sizeof(struct bau_control), GFP_KERNEL, node);
-	BUG_ON(!bau_tabp);
-
-	bau_tabp->msg_statuses =
-	    kmalloc_node(sizeof(struct bau_msg_status) *
-			 DEST_Q_SIZE, GFP_KERNEL, node);
-	BUG_ON(!bau_tabp->msg_statuses);
-
-	for (i = 0, msp = bau_tabp->msg_statuses; i < DEST_Q_SIZE; i++, msp++)
-		bau_cpubits_clear(&msp->seen_by, (int)
-				  uv_blade_nr_possible_cpus(blade));
-
-	uv_bau_table_bases[blade] = bau_tabp;
-
-	return bau_tabp;
-}
-
-/*
- * finish the initialization of the per-blade control structures
- */
-static void __init
-uv_table_bases_finish(int blade,
-		      struct bau_control *bau_tablesp,
-		      struct bau_desc *adp)
-{
-	struct bau_control *bcp;
-	int cpu;
-
-	for_each_present_cpu(cpu) {
-		if (blade != uv_cpu_to_blade_id(cpu))
-			continue;
-
-		bcp = (struct bau_control *)&per_cpu(bau_control, cpu);
-		bcp->bau_msg_head	= bau_tablesp->va_queue_first;
-		bcp->va_queue_first	= bau_tablesp->va_queue_first;
-		bcp->va_queue_last	= bau_tablesp->va_queue_last;
-		bcp->msg_statuses	= bau_tablesp->msg_statuses;
-		bcp->descriptor_base	= adp;
-	}
-}
-
-/*
  * initialize the sending side's sending buffers
  */
-static struct bau_desc * __init
+static void
 uv_activation_descriptor_init(int node, int pnode)
 {
 	int i;
+	int cpu;
 	unsigned long pa;
 	unsigned long m;
 	unsigned long n;
-	struct bau_desc *adp;
-	struct bau_desc *ad2;
+	struct bau_desc *bau_desc;
+	struct bau_desc *bd2;
+	struct bau_control *bcp;
 
 	/*
 	 * each bau_desc is 64 bytes; there are 8 (UV_ITEMS_PER_DESCRIPTOR)
-	 * per cpu; and up to 32 (UV_ADP_SIZE) cpu's per blade
+	 * per cpu; and up to 32 (UV_ADP_SIZE) cpu's per uvhub
 	 */
-	adp = (struct bau_desc *)kmalloc_node(sizeof(struct bau_desc)*
+	bau_desc = (struct bau_desc *)kmalloc_node(sizeof(struct bau_desc)*
 		UV_ADP_SIZE*UV_ITEMS_PER_DESCRIPTOR, GFP_KERNEL, node);
-	BUG_ON(!adp);
+	BUG_ON(!bau_desc);
 
-	pa = uv_gpa(adp); /* need the real nasid*/
-	n = uv_gpa_to_pnode(pa);
+	pa = uv_gpa(bau_desc); /* need the real nasid*/
+	n = pa >> uv_nshift;
 	m = pa & uv_mmask;
 
 	uv_write_global_mmr64(pnode, UVH_LB_BAU_SB_DESCRIPTOR_BASE,
@@ -731,96 +1155,188 @@ uv_activation_descriptor_init(int node, int pnode)
 	/*
 	 * initializing all 8 (UV_ITEMS_PER_DESCRIPTOR) descriptors for each
 	 * cpu even though we only use the first one; one descriptor can
-	 * describe a broadcast to 256 nodes.
+	 * describe a broadcast to 256 uv hubs.
 	 */
-	for (i = 0, ad2 = adp; i < (UV_ADP_SIZE*UV_ITEMS_PER_DESCRIPTOR);
-		i++, ad2++) {
-		memset(ad2, 0, sizeof(struct bau_desc));
-		ad2->header.sw_ack_flag = 1;
+	for (i = 0, bd2 = bau_desc; i < (UV_ADP_SIZE*UV_ITEMS_PER_DESCRIPTOR);
+		i++, bd2++) {
+		memset(bd2, 0, sizeof(struct bau_desc));
+		bd2->header.sw_ack_flag = 1;
 		/*
-		 * base_dest_nodeid is the first node in the partition, so
-		 * the bit map will indicate partition-relative node numbers.
-		 * note that base_dest_nodeid is actually a nasid.
+		 * base_dest_nodeid is the nasid (pnode<<1) of the first uvhub
+		 * in the partition. The bit map will indicate uvhub numbers,
+		 * which are 0-N in a partition. Pnodes are unique system-wide.
 		 */
-		ad2->header.base_dest_nodeid = uv_partition_base_pnode << 1;
-		ad2->header.dest_subnodeid = 0x10; /* the LB */
-		ad2->header.command = UV_NET_ENDPOINT_INTD;
-		ad2->header.int_both = 1;
+		bd2->header.base_dest_nodeid = uv_partition_base_pnode << 1;
+		bd2->header.dest_subnodeid = 0x10; /* the LB */
+		bd2->header.command = UV_NET_ENDPOINT_INTD;
+		bd2->header.int_both = 1;
 		/*
 		 * all others need to be set to zero:
 		 *   fairness chaining multilevel count replied_to
 		 */
 	}
-	return adp;
+	for_each_present_cpu(cpu) {
+		if (pnode != uv_blade_to_pnode(uv_cpu_to_blade_id(cpu)))
+			continue;
+		bcp = &per_cpu(bau_control, cpu);
+		bcp->descriptor_base = bau_desc;
+	}
 }
 
 /*
  * initialize the destination side's receiving buffers
+ * entered for each uvhub in the partition
+ * - node is first node (kernel memory notion) on the uvhub
+ * - pnode is the uvhub's physical identifier
  */
-static struct bau_payload_queue_entry * __init
-uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp)
+static void
+uv_payload_queue_init(int node, int pnode)
 {
-	struct bau_payload_queue_entry *pqp;
-	unsigned long pa;
 	int pn;
+	int cpu;
 	char *cp;
+	unsigned long pa;
+	struct bau_payload_queue_entry *pqp;
+	struct bau_payload_queue_entry *pqp_malloc;
+	struct bau_control *bcp;
 
 	pqp = (struct bau_payload_queue_entry *) kmalloc_node(
 		(DEST_Q_SIZE + 1) * sizeof(struct bau_payload_queue_entry),
 		GFP_KERNEL, node);
 	BUG_ON(!pqp);
+	pqp_malloc = pqp;
 
 	cp = (char *)pqp + 31;
 	pqp = (struct bau_payload_queue_entry *)(((unsigned long)cp >> 5) << 5);
-	bau_tablesp->va_queue_first = pqp;
+
+	for_each_present_cpu(cpu) {
+		if (pnode != uv_cpu_to_pnode(cpu))
+			continue;
+		/* for every cpu on this pnode: */
+		bcp = &per_cpu(bau_control, cpu);
+		bcp->va_queue_first = pqp;
+		bcp->bau_msg_head = pqp;
+		bcp->va_queue_last = pqp + (DEST_Q_SIZE - 1);
+	}
 	/*
 	 * need the pnode of where the memory was really allocated
 	 */
 	pa = uv_gpa(pqp);
-	pn = uv_gpa_to_pnode(pa);
+	pn = pa >> uv_nshift;
 	uv_write_global_mmr64(pnode,
 			      UVH_LB_BAU_INTD_PAYLOAD_QUEUE_FIRST,
 			      ((unsigned long)pn << UV_PAYLOADQ_PNODE_SHIFT) |
 			      uv_physnodeaddr(pqp));
 	uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_TAIL,
 			      uv_physnodeaddr(pqp));
-	bau_tablesp->va_queue_last = pqp + (DEST_Q_SIZE - 1);
 	uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_LAST,
 			      (unsigned long)
-			      uv_physnodeaddr(bau_tablesp->va_queue_last));
+			      uv_physnodeaddr(pqp + (DEST_Q_SIZE - 1)));
+	/* in effect, all msg_type's are set to MSG_NOOP */
 	memset(pqp, 0, sizeof(struct bau_payload_queue_entry) * DEST_Q_SIZE);
-
-	return pqp;
 }
 
 /*
- * Initialization of each UV blade's structures
+ * Initialization of each UV hub's structures
  */
-static int __init uv_init_blade(int blade)
+static void __init uv_init_uvhub(int uvhub, int vector)
 {
 	int node;
 	int pnode;
-	unsigned long pa;
 	unsigned long apicid;
-	struct bau_desc *adp;
-	struct bau_payload_queue_entry *pqp;
-	struct bau_control *bau_tablesp;
-
-	node = blade_to_first_node(blade);
-	bau_tablesp = uv_table_bases_init(blade, node);
-	pnode = uv_blade_to_pnode(blade);
-	adp = uv_activation_descriptor_init(node, pnode);
-	pqp = uv_payload_queue_init(node, pnode, bau_tablesp);
-	uv_table_bases_finish(blade, bau_tablesp, adp);
+
+	node = uvhub_to_first_node(uvhub);
+	pnode = uv_blade_to_pnode(uvhub);
+	uv_activation_descriptor_init(node, pnode);
+	uv_payload_queue_init(node, pnode);
 	/*
 	 * the below initialization can't be in firmware because the
 	 * messaging IRQ will be determined by the OS
 	 */
-	apicid = blade_to_first_apicid(blade);
-	pa = uv_read_global_mmr64(pnode, UVH_BAU_DATA_CONFIG);
+	apicid = uvhub_to_first_apicid(uvhub);
 	uv_write_global_mmr64(pnode, UVH_BAU_DATA_CONFIG,
-				      ((apicid << 32) | UV_BAU_MESSAGE));
-	return 0;
+				      ((apicid << 32) | vector));
+}
+
+/*
+ * initialize the bau_control structure for each cpu
+ */
+static void uv_init_per_cpu(int nuvhubs)
+{
+	int i, j, k;
+	int cpu;
+	int pnode;
+	int uvhub;
+	short socket = 0;
+	struct bau_control *bcp;
+	struct uvhub_desc *bdp;
+	struct socket_desc *sdp;
+	struct bau_control *hmaster = NULL;
+	struct bau_control *smaster = NULL;
+	struct socket_desc {
+		short num_cpus;
+		short cpu_number[16];
+	};
+	struct uvhub_desc {
+		short num_sockets;
+		short num_cpus;
+		short uvhub;
+		short pnode;
+		struct socket_desc socket[2];
+	};
+	struct uvhub_desc *uvhub_descs;
+
+	uvhub_descs = (struct uvhub_desc *)
+		kmalloc(nuvhubs * sizeof(struct uvhub_desc), GFP_KERNEL);
+	memset(uvhub_descs, 0, nuvhubs * sizeof(struct uvhub_desc));
+	for_each_present_cpu(cpu) {
+		bcp = &per_cpu(bau_control, cpu);
+		memset(bcp, 0, sizeof(struct bau_control));
+		spin_lock_init(&bcp->masks_lock);
+		bcp->max_concurrent = uv_bau_max_concurrent;
+		pnode = uv_cpu_hub_info(cpu)->pnode;
+		uvhub = uv_cpu_hub_info(cpu)->numa_blade_id;
+		bdp = &uvhub_descs[uvhub];
+		bdp->num_cpus++;
+		bdp->uvhub = uvhub;
+		bdp->pnode = pnode;
+		/* time interval to catch a hardware stay-busy bug */
+		bcp->timeout_interval = millisec_2_cycles(3);
+		/* kludge: assume uv_hub.h is constant */
+		socket = (cpu_physical_id(cpu)>>5)&1;
+		if (socket >= bdp->num_sockets)
+			bdp->num_sockets = socket+1;
+		sdp = &bdp->socket[socket];
+		sdp->cpu_number[sdp->num_cpus] = cpu;
+		sdp->num_cpus++;
+	}
+	socket = 0;
+	for_each_possible_blade(uvhub) {
+		bdp = &uvhub_descs[uvhub];
+		for (i = 0; i < bdp->num_sockets; i++) {
+			sdp = &bdp->socket[i];
+			for (j = 0; j < sdp->num_cpus; j++) {
+				cpu = sdp->cpu_number[j];
+				bcp = &per_cpu(bau_control, cpu);
+				bcp->cpu = cpu;
+				if (j == 0) {
+					smaster = bcp;
+					if (i == 0)
+						hmaster = bcp;
+				}
+				bcp->cpus_in_uvhub = bdp->num_cpus;
+				bcp->cpus_in_socket = sdp->num_cpus;
+				bcp->socket_master = smaster;
+				bcp->uvhub_master = hmaster;
+				for (k = 0; k < DEST_Q_SIZE; k++)
+					bcp->socket_acknowledge_count[k] = 0;
+				bcp->uvhub_cpu =
+				  uv_cpu_hub_info(cpu)->blade_processor_id;
+			}
+			socket++;
+		}
+	}
+	kfree(uvhub_descs);
 }
 
 /*
@@ -828,38 +1344,54 @@ static int __init uv_init_blade(int blade)
  */
 static int __init uv_bau_init(void)
 {
-	int blade;
-	int nblades;
+	int uvhub;
+	int pnode;
+	int nuvhubs;
 	int cur_cpu;
+	int vector;
+	unsigned long mmr;
 
 	if (!is_uv_system())
 		return 0;
 
+	if (nobau)
+		return 0;
+
 	for_each_possible_cpu(cur_cpu)
 		zalloc_cpumask_var_node(&per_cpu(uv_flush_tlb_mask, cur_cpu),
 				       GFP_KERNEL, cpu_to_node(cur_cpu));
 
-	uv_bau_retry_limit = 1;
+	uv_bau_max_concurrent = MAX_BAU_CONCURRENT;
+	uv_nshift = uv_hub_info->m_val;
 	uv_mmask = (1UL << uv_hub_info->m_val) - 1;
-	nblades = uv_num_possible_blades();
+	nuvhubs = uv_num_possible_blades();
 
-	uv_bau_table_bases = (struct bau_control **)
-	    kmalloc(nblades * sizeof(struct bau_control *), GFP_KERNEL);
-	BUG_ON(!uv_bau_table_bases);
+	uv_init_per_cpu(nuvhubs);
 
 	uv_partition_base_pnode = 0x7fffffff;
-	for (blade = 0; blade < nblades; blade++)
-		if (uv_blade_nr_possible_cpus(blade) &&
-			(uv_blade_to_pnode(blade) < uv_partition_base_pnode))
-			uv_partition_base_pnode = uv_blade_to_pnode(blade);
-	for (blade = 0; blade < nblades; blade++)
-		if (uv_blade_nr_possible_cpus(blade))
-			uv_init_blade(blade);
-
-	alloc_intr_gate(UV_BAU_MESSAGE, uv_bau_message_intr1);
+	for (uvhub = 0; uvhub < nuvhubs; uvhub++)
+		if (uv_blade_nr_possible_cpus(uvhub) &&
+			(uv_blade_to_pnode(uvhub) < uv_partition_base_pnode))
+			uv_partition_base_pnode = uv_blade_to_pnode(uvhub);
+
+	vector = UV_BAU_MESSAGE;
+	for_each_possible_blade(uvhub)
+		if (uv_blade_nr_possible_cpus(uvhub))
+			uv_init_uvhub(uvhub, vector);
+
 	uv_enable_timeouts();
+	alloc_intr_gate(vector, uv_bau_message_intr1);
+
+	for_each_possible_blade(uvhub) {
+		pnode = uv_blade_to_pnode(uvhub);
+		/* INIT the bau */
+		uv_write_global_mmr64(pnode, UVH_LB_BAU_SB_ACTIVATION_CONTROL,
+				      ((unsigned long)1 << 63));
+		mmr = 1; /* should be 1 to broadcast to both sockets */
+		uv_write_global_mmr64(pnode, UVH_BAU_DATA_BROADCAST, mmr);
+	}
 
 	return 0;
 }
-__initcall(uv_bau_init);
-__initcall(uv_ptc_init);
+core_initcall(uv_bau_init);
+core_initcall(uv_ptc_init);
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 1168e445418..142d70c74b0 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -15,6 +15,7 @@
 #include <linux/kprobes.h>
 #include <linux/uaccess.h>
 #include <linux/kdebug.h>
+#include <linux/kgdb.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
 #include <linux/ptrace.h>
@@ -108,15 +109,6 @@ static inline void preempt_conditional_cli(struct pt_regs *regs)
 	dec_preempt_count();
 }
 
-#ifdef CONFIG_X86_32
-static inline void
-die_if_kernel(const char *str, struct pt_regs *regs, long err)
-{
-	if (!user_mode_vm(regs))
-		die(str, regs, err);
-}
-#endif
-
 static void __kprobes
 do_trap(int trapnr, int signr, char *str, struct pt_regs *regs,
 	long error_code, siginfo_t *info)
@@ -460,6 +452,11 @@ void restart_nmi(void)
 /* May run on IST stack. */
 dotraplinkage void __kprobes do_int3(struct pt_regs *regs, long error_code)
 {
+#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP
+	if (kgdb_ll_trap(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP)
+			== NOTIFY_STOP)
+		return;
+#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */
 #ifdef CONFIG_KPROBES
 	if (notify_die(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP)
 			== NOTIFY_STOP)
@@ -543,11 +540,11 @@ dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code)
 
 	/* DR6 may or may not be cleared by the CPU */
 	set_debugreg(0, 6);
+
 	/*
 	 * The processor cleared BTF, so don't mark that we need it set.
 	 */
-	clear_tsk_thread_flag(tsk, TIF_DEBUGCTLMSR);
-	tsk->thread.debugctlmsr = 0;
+	clear_tsk_thread_flag(tsk, TIF_BLOCKSTEP);
 
 	/* Store the virtualized DR6 value */
 	tsk->thread.debugreg6 = dr6;
@@ -585,55 +582,67 @@ dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code)
 	return;
 }
 
-#ifdef CONFIG_X86_64
-static int kernel_math_error(struct pt_regs *regs, const char *str, int trapnr)
-{
-	if (fixup_exception(regs))
-		return 1;
-
-	notify_die(DIE_GPF, str, regs, 0, trapnr, SIGFPE);
-	/* Illegal floating point operation in the kernel */
-	current->thread.trap_no = trapnr;
-	die(str, regs, 0);
-	return 0;
-}
-#endif
-
 /*
  * Note that we play around with the 'TS' bit in an attempt to get
  * the correct behaviour even in the presence of the asynchronous
  * IRQ13 behaviour
  */
-void math_error(void __user *ip)
+void math_error(struct pt_regs *regs, int error_code, int trapnr)
 {
-	struct task_struct *task;
+	struct task_struct *task = current;
 	siginfo_t info;
-	unsigned short cwd, swd, err;
+	unsigned short err;
+	char *str = (trapnr == 16) ? "fpu exception" : "simd exception";
+
+	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, SIGFPE) == NOTIFY_STOP)
+		return;
+	conditional_sti(regs);
+
+	if (!user_mode_vm(regs))
+	{
+		if (!fixup_exception(regs)) {
+			task->thread.error_code = error_code;
+			task->thread.trap_no = trapnr;
+			die(str, regs, error_code);
+		}
+		return;
+	}
 
 	/*
 	 * Save the info for the exception handler and clear the error.
 	 */
-	task = current;
 	save_init_fpu(task);
-	task->thread.trap_no = 16;
-	task->thread.error_code = 0;
+	task->thread.trap_no = trapnr;
+	task->thread.error_code = error_code;
 	info.si_signo = SIGFPE;
 	info.si_errno = 0;
-	info.si_addr = ip;
-	/*
-	 * (~cwd & swd) will mask out exceptions that are not set to unmasked
-	 * status.  0x3f is the exception bits in these regs, 0x200 is the
-	 * C1 reg you need in case of a stack fault, 0x040 is the stack
-	 * fault bit.  We should only be taking one exception at a time,
-	 * so if this combination doesn't produce any single exception,
-	 * then we have a bad program that isn't synchronizing its FPU usage
-	 * and it will suffer the consequences since we won't be able to
-	 * fully reproduce the context of the exception
-	 */
-	cwd = get_fpu_cwd(task);
-	swd = get_fpu_swd(task);
+	info.si_addr = (void __user *)regs->ip;
+	if (trapnr == 16) {
+		unsigned short cwd, swd;
+		/*
+		 * (~cwd & swd) will mask out exceptions that are not set to unmasked
+		 * status.  0x3f is the exception bits in these regs, 0x200 is the
+		 * C1 reg you need in case of a stack fault, 0x040 is the stack
+		 * fault bit.  We should only be taking one exception at a time,
+		 * so if this combination doesn't produce any single exception,
+		 * then we have a bad program that isn't synchronizing its FPU usage
+		 * and it will suffer the consequences since we won't be able to
+		 * fully reproduce the context of the exception
+		 */
+		cwd = get_fpu_cwd(task);
+		swd = get_fpu_swd(task);
 
-	err = swd & ~cwd;
+		err = swd & ~cwd;
+	} else {
+		/*
+		 * The SIMD FPU exceptions are handled a little differently, as there
+		 * is only a single status/control register.  Thus, to determine which
+		 * unmasked exception was caught we must mask the exception mask bits
+		 * at 0x1f80, and then use these to mask the exception bits at 0x3f.
+		 */
+		unsigned short mxcsr = get_fpu_mxcsr(task);
+		err = ~(mxcsr >> 7) & mxcsr;
+	}
 
 	if (err & 0x001) {	/* Invalid op */
 		/*
@@ -662,97 +671,17 @@ void math_error(void __user *ip)
 
 dotraplinkage void do_coprocessor_error(struct pt_regs *regs, long error_code)
 {
-	conditional_sti(regs);
-
 #ifdef CONFIG_X86_32
 	ignore_fpu_irq = 1;
-#else
-	if (!user_mode(regs) &&
-	    kernel_math_error(regs, "kernel x87 math error", 16))
-		return;
 #endif
 
-	math_error((void __user *)regs->ip);
-}
-
-static void simd_math_error(void __user *ip)
-{
-	struct task_struct *task;
-	siginfo_t info;
-	unsigned short mxcsr;
-
-	/*
-	 * Save the info for the exception handler and clear the error.
-	 */
-	task = current;
-	save_init_fpu(task);
-	task->thread.trap_no = 19;
-	task->thread.error_code = 0;
-	info.si_signo = SIGFPE;
-	info.si_errno = 0;
-	info.si_code = __SI_FAULT;
-	info.si_addr = ip;
-	/*
-	 * The SIMD FPU exceptions are handled a little differently, as there
-	 * is only a single status/control register.  Thus, to determine which
-	 * unmasked exception was caught we must mask the exception mask bits
-	 * at 0x1f80, and then use these to mask the exception bits at 0x3f.
-	 */
-	mxcsr = get_fpu_mxcsr(task);
-	switch (~((mxcsr & 0x1f80) >> 7) & (mxcsr & 0x3f)) {
-	case 0x000:
-	default:
-		break;
-	case 0x001: /* Invalid Op */
-		info.si_code = FPE_FLTINV;
-		break;
-	case 0x002: /* Denormalize */
-	case 0x010: /* Underflow */
-		info.si_code = FPE_FLTUND;
-		break;
-	case 0x004: /* Zero Divide */
-		info.si_code = FPE_FLTDIV;
-		break;
-	case 0x008: /* Overflow */
-		info.si_code = FPE_FLTOVF;
-		break;
-	case 0x020: /* Precision */
-		info.si_code = FPE_FLTRES;
-		break;
-	}
-	force_sig_info(SIGFPE, &info, task);
+	math_error(regs, error_code, 16);
 }
 
 dotraplinkage void
 do_simd_coprocessor_error(struct pt_regs *regs, long error_code)
 {
-	conditional_sti(regs);
-
-#ifdef CONFIG_X86_32
-	if (cpu_has_xmm) {
-		/* Handle SIMD FPU exceptions on PIII+ processors. */
-		ignore_fpu_irq = 1;
-		simd_math_error((void __user *)regs->ip);
-		return;
-	}
-	/*
-	 * Handle strange cache flush from user space exception
-	 * in all other cases.  This is undocumented behaviour.
-	 */
-	if (regs->flags & X86_VM_MASK) {
-		handle_vm86_fault((struct kernel_vm86_regs *)regs, error_code);
-		return;
-	}
-	current->thread.trap_no = 19;
-	current->thread.error_code = error_code;
-	die_if_kernel("cache flush denied", regs, error_code);
-	force_sig(SIGSEGV, current);
-#else
-	if (!user_mode(regs) &&
-			kernel_math_error(regs, "kernel simd math error", 19))
-		return;
-	simd_math_error((void __user *)regs->ip);
-#endif
+	math_error(regs, error_code, 19);
 }
 
 dotraplinkage void
@@ -879,6 +808,16 @@ dotraplinkage void do_iret_error(struct pt_regs *regs, long error_code)
 }
 #endif
 
+/* Set of traps needed for early debugging. */
+void __init early_trap_init(void)
+{
+	set_intr_gate_ist(1, &debug, DEBUG_STACK);
+	/* int3 can be called from all */
+	set_system_intr_gate_ist(3, &int3, DEBUG_STACK);
+	set_intr_gate(14, &page_fault);
+	load_idt(&idt_descr);
+}
+
 void __init trap_init(void)
 {
 	int i;
@@ -892,10 +831,7 @@ void __init trap_init(void)
 #endif
 
 	set_intr_gate(0, &divide_error);
-	set_intr_gate_ist(1, &debug, DEBUG_STACK);
 	set_intr_gate_ist(2, &nmi, NMI_STACK);
-	/* int3 can be called from all */
-	set_system_intr_gate_ist(3, &int3, DEBUG_STACK);
 	/* int4 can be called from all */
 	set_system_intr_gate(4, &overflow);
 	set_intr_gate(5, &bounds);
@@ -911,7 +847,6 @@ void __init trap_init(void)
 	set_intr_gate(11, &segment_not_present);
 	set_intr_gate_ist(12, &stack_segment, STACKFAULT_STACK);
 	set_intr_gate(13, &general_protection);
-	set_intr_gate(14, &page_fault);
 	set_intr_gate(15, &spurious_interrupt_bug);
 	set_intr_gate(16, &coprocessor_error);
 	set_intr_gate(17, &alignment_check);
diff --git a/arch/x86/kernel/uv_irq.c b/arch/x86/kernel/uv_irq.c
index 1d40336b030..1132129db79 100644
--- a/arch/x86/kernel/uv_irq.c
+++ b/arch/x86/kernel/uv_irq.c
@@ -44,7 +44,7 @@ static void uv_ack_apic(unsigned int irq)
 	ack_APIC_irq();
 }
 
-struct irq_chip uv_irq_chip = {
+static struct irq_chip uv_irq_chip = {
 	.name		= "UV-CORE",
 	.startup	= uv_noop_ret,
 	.shutdown	= uv_noop,
@@ -141,7 +141,7 @@ int uv_irq_2_mmr_info(int irq, unsigned long *offset, int *pnode)
  */
 static int
 arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade,
-		       unsigned long mmr_offset, int restrict)
+		       unsigned long mmr_offset, int limit)
 {
 	const struct cpumask *eligible_cpu = cpumask_of(cpu);
 	struct irq_desc *desc = irq_to_desc(irq);
@@ -160,7 +160,7 @@ arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade,
 	if (err != 0)
 		return err;
 
-	if (restrict == UV_AFFINITY_CPU)
+	if (limit == UV_AFFINITY_CPU)
 		desc->status |= IRQ_NO_BALANCING;
 	else
 		desc->status |= IRQ_MOVE_PCNTXT;
@@ -214,7 +214,7 @@ static int uv_set_irq_affinity(unsigned int irq, const struct cpumask *mask)
 	unsigned long mmr_value;
 	struct uv_IO_APIC_route_entry *entry;
 	unsigned long mmr_offset;
-	unsigned mmr_pnode;
+	int mmr_pnode;
 
 	if (set_desc_affinity(desc, mask, &dest))
 		return -1;
@@ -248,7 +248,7 @@ static int uv_set_irq_affinity(unsigned int irq, const struct cpumask *mask)
  * interrupt is raised.
  */
 int uv_setup_irq(char *irq_name, int cpu, int mmr_blade,
-		 unsigned long mmr_offset, int restrict)
+		 unsigned long mmr_offset, int limit)
 {
 	int irq, ret;
 
@@ -258,7 +258,7 @@ int uv_setup_irq(char *irq_name, int cpu, int mmr_blade,
 		return -EBUSY;
 
 	ret = arch_enable_uv_irq(irq_name, irq, cpu, mmr_blade, mmr_offset,
-		restrict);
+		limit);
 	if (ret == irq)
 		uv_set_irq_2_mmr_info(irq, mmr_offset, mmr_blade);
 	else
diff --git a/arch/x86/kernel/x8664_ksyms_64.c b/arch/x86/kernel/x8664_ksyms_64.c
index 693920b2249..1b950d151e5 100644
--- a/arch/x86/kernel/x8664_ksyms_64.c
+++ b/arch/x86/kernel/x8664_ksyms_64.c
@@ -54,7 +54,6 @@ EXPORT_SYMBOL(memcpy);
 EXPORT_SYMBOL(__memcpy);
 
 EXPORT_SYMBOL(empty_zero_page);
-EXPORT_SYMBOL(init_level4_pgt);
 #ifndef CONFIG_PARAVIRT
 EXPORT_SYMBOL(native_load_gs_index);
 #endif
diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c
index 782c3a362ec..37e68fc5e24 100644
--- a/arch/x86/kernel/xsave.c
+++ b/arch/x86/kernel/xsave.c
@@ -99,7 +99,7 @@ int save_i387_xstate(void __user *buf)
 		if (err)
 			return err;
 
-		if (task_thread_info(tsk)->status & TS_XSAVE)
+		if (use_xsave())
 			err = xsave_user(buf);
 		else
 			err = fxsave_user(buf);
@@ -109,14 +109,14 @@ int save_i387_xstate(void __user *buf)
 		task_thread_info(tsk)->status &= ~TS_USEDFPU;
 		stts();
 	} else {
-		if (__copy_to_user(buf, &tsk->thread.xstate->fxsave,
+		if (__copy_to_user(buf, &tsk->thread.fpu.state->fxsave,
 				   xstate_size))
 			return -1;
 	}
 
 	clear_used_math(); /* trigger finit */
 
-	if (task_thread_info(tsk)->status & TS_XSAVE) {
+	if (use_xsave()) {
 		struct _fpstate __user *fx = buf;
 		struct _xstate __user *x = buf;
 		u64 xstate_bv;
@@ -225,7 +225,7 @@ int restore_i387_xstate(void __user *buf)
 		clts();
 		task_thread_info(current)->status |= TS_USEDFPU;
 	}
-	if (task_thread_info(tsk)->status & TS_XSAVE)
+	if (use_xsave())
 		err = restore_user_xstate(buf);
 	else
 		err = fxrstor_checking((__force struct i387_fxsave_struct *)