Merge branch 'linus' into x86/urgent

Merge reason: We need the x86/uv updates from upstream, to queue up
              dependent fix.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

13 files changed, 217 insertions, 131 deletions

diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 098ec84b8c0..f2870920f24 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -431,6 +431,12 @@ static void __cpuinit setup_APIC_timer(void)
 {
 	struct clock_event_device *levt = &__get_cpu_var(lapic_events);
 
+	if (cpu_has(&current_cpu_data, X86_FEATURE_ARAT)) {
+		lapic_clockevent.features &= ~CLOCK_EVT_FEAT_C3STOP;
+		/* Make LAPIC timer preferrable over percpu HPET */
+		lapic_clockevent.rating = 150;
+	}
+
 	memcpy(levt, &lapic_clockevent, sizeof(*levt));
 	levt->cpumask = cpumask_of(smp_processor_id());
 
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index 49d2f5c739b..d6712334ce4 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -563,7 +563,8 @@ void __init uv_system_init(void)
 	unsigned long gnode_upper, lowmem_redir_base, lowmem_redir_size;
 	int bytes, nid, cpu, lcpu, pnode, blade, i, j, m_val, n_val;
 	int max_pnode = 0;
-	unsigned long mmr_base, present;
+	unsigned long mmr_base, present, paddr;
+	unsigned short pnode_mask;
 
 	map_low_mmrs();
 
@@ -606,6 +607,7 @@ void __init uv_system_init(void)
 		}
 	}
 
+	pnode_mask = (1 << n_val) - 1;
 	node_id.v = uv_read_local_mmr(UVH_NODE_ID);
 	gnode_upper = (((unsigned long)node_id.s.node_id) &
 		       ~((1 << n_val) - 1)) << m_val;
@@ -629,7 +631,7 @@ void __init uv_system_init(void)
 		uv_cpu_hub_info(cpu)->numa_blade_id = blade;
 		uv_cpu_hub_info(cpu)->blade_processor_id = lcpu;
 		uv_cpu_hub_info(cpu)->pnode = pnode;
-		uv_cpu_hub_info(cpu)->pnode_mask = (1 << n_val) - 1;
+		uv_cpu_hub_info(cpu)->pnode_mask = pnode_mask;
 		uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1;
 		uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper;
 		uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base;
@@ -645,6 +647,16 @@ void __init uv_system_init(void)
 			lcpu, blade);
 	}
 
+	/* Add blade/pnode info for nodes without cpus */
+	for_each_online_node(nid) {
+		if (uv_node_to_blade[nid] >= 0)
+			continue;
+		paddr = node_start_pfn(nid) << PAGE_SHIFT;
+		pnode = (paddr >> m_val) & pnode_mask;
+		blade = boot_pnode_to_blade(pnode);
+		uv_node_to_blade[nid] = blade;
+	}
+
 	map_gru_high(max_pnode);
 	map_mmr_high(max_pnode);
 	map_config_high(max_pnode);
diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c
index f63882728d9..63a88e1f987 100644
--- a/arch/x86/kernel/bios_uv.c
+++ b/arch/x86/kernel/bios_uv.c
@@ -182,7 +182,8 @@ void uv_bios_init(void)
 	memcpy(&uv_systab, tab, sizeof(struct uv_systab));
 	iounmap(tab);
 
-	printk(KERN_INFO "EFI UV System Table Revision %d\n", tab->revision);
+	printk(KERN_INFO "EFI UV System Table Revision %d\n",
+					uv_systab.revision);
 }
 #else	/* !CONFIG_EFI */
 
diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c
index 8220ae69849..c965e521271 100644
--- a/arch/x86/kernel/cpu/addon_cpuid_features.c
+++ b/arch/x86/kernel/cpu/addon_cpuid_features.c
@@ -31,6 +31,7 @@ void __cpuinit init_scattered_cpuid_features(struct cpuinfo_x86 *c)
 
 	static const struct cpuid_bit __cpuinitconst cpuid_bits[] = {
 		{ X86_FEATURE_IDA, CR_EAX, 1, 0x00000006 },
+		{ X86_FEATURE_ARAT, CR_EAX, 2, 0x00000006 },
 		{ 0, 0, 0, 0 }
 	};
 
diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
index 19f6b9d27e8..ecdb682ab51 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -68,6 +68,7 @@ struct acpi_cpufreq_data {
 	unsigned int max_freq;
 	unsigned int resume;
 	unsigned int cpu_feature;
+	u64 saved_aperf, saved_mperf;
 };
 
 static DEFINE_PER_CPU(struct acpi_cpufreq_data *, drv_data);
@@ -152,7 +153,8 @@ struct drv_cmd {
 	u32 val;
 };
 
-static long do_drv_read(void *_cmd)
+/* Called via smp_call_function_single(), on the target CPU */
+static void do_drv_read(void *_cmd)
 {
 	struct drv_cmd *cmd = _cmd;
 	u32 h;
@@ -169,10 +171,10 @@ static long do_drv_read(void *_cmd)
 	default:
 		break;
 	}
-	return 0;
 }
 
-static long do_drv_write(void *_cmd)
+/* Called via smp_call_function_many(), on the target CPUs */
+static void do_drv_write(void *_cmd)
 {
 	struct drv_cmd *cmd = _cmd;
 	u32 lo, hi;
@@ -191,23 +193,24 @@ static long do_drv_write(void *_cmd)
 	default:
 		break;
 	}
-	return 0;
 }
 
 static void drv_read(struct drv_cmd *cmd)
 {
 	cmd->val = 0;
 
-	work_on_cpu(cpumask_any(cmd->mask), do_drv_read, cmd);
+	smp_call_function_single(cpumask_any(cmd->mask), do_drv_read, cmd, 1);
 }
 
 static void drv_write(struct drv_cmd *cmd)
 {
-	unsigned int i;
+	int this_cpu;
 
-	for_each_cpu(i, cmd->mask) {
-		work_on_cpu(i, do_drv_write, cmd);
-	}
+	this_cpu = get_cpu();
+	if (cpumask_test_cpu(this_cpu, cmd->mask))
+		do_drv_write(cmd);
+	smp_call_function_many(cmd->mask, do_drv_write, cmd, 1);
+	put_cpu();
 }
 
 static u32 get_cur_val(const struct cpumask *mask)
@@ -241,28 +244,23 @@ static u32 get_cur_val(const struct cpumask *mask)
 	return cmd.val;
 }
 
-struct perf_cur {
+struct perf_pair {
 	union {
 		struct {
 			u32 lo;
 			u32 hi;
 		} split;
 		u64 whole;
-	} aperf_cur, mperf_cur;
+	} aperf, mperf;
 };
 
-
-static long read_measured_perf_ctrs(void *_cur)
+/* Called via smp_call_function_single(), on the target CPU */
+static void read_measured_perf_ctrs(void *_cur)
 {
-	struct perf_cur *cur = _cur;
+	struct perf_pair *cur = _cur;
 
-	rdmsr(MSR_IA32_APERF, cur->aperf_cur.split.lo, cur->aperf_cur.split.hi);
-	rdmsr(MSR_IA32_MPERF, cur->mperf_cur.split.lo, cur->mperf_cur.split.hi);
-
-	wrmsr(MSR_IA32_APERF, 0, 0);
-	wrmsr(MSR_IA32_MPERF, 0, 0);
-
-	return 0;
+	rdmsr(MSR_IA32_APERF, cur->aperf.split.lo, cur->aperf.split.hi);
+	rdmsr(MSR_IA32_MPERF, cur->mperf.split.lo, cur->mperf.split.hi);
 }
 
 /*
@@ -281,52 +279,57 @@ static long read_measured_perf_ctrs(void *_cur)
 static unsigned int get_measured_perf(struct cpufreq_policy *policy,
 				      unsigned int cpu)
 {
-	struct perf_cur cur;
+	struct perf_pair readin, cur;
 	unsigned int perf_percent;
 	unsigned int retval;
 
-	if (!work_on_cpu(cpu, read_measured_perf_ctrs, &cur))
+	if (smp_call_function_single(cpu, read_measured_perf_ctrs, &readin, 1))
 		return 0;
 
+	cur.aperf.whole = readin.aperf.whole -
+				per_cpu(drv_data, cpu)->saved_aperf;
+	cur.mperf.whole = readin.mperf.whole -
+				per_cpu(drv_data, cpu)->saved_mperf;
+	per_cpu(drv_data, cpu)->saved_aperf = readin.aperf.whole;
+	per_cpu(drv_data, cpu)->saved_mperf = readin.mperf.whole;
+
 #ifdef __i386__
 	/*
 	 * We dont want to do 64 bit divide with 32 bit kernel
 	 * Get an approximate value. Return failure in case we cannot get
 	 * an approximate value.
 	 */
-	if (unlikely(cur.aperf_cur.split.hi || cur.mperf_cur.split.hi)) {
+	if (unlikely(cur.aperf.split.hi || cur.mperf.split.hi)) {
 		int shift_count;
 		u32 h;
 
-		h = max_t(u32, cur.aperf_cur.split.hi, cur.mperf_cur.split.hi);
+		h = max_t(u32, cur.aperf.split.hi, cur.mperf.split.hi);
 		shift_count = fls(h);
 
-		cur.aperf_cur.whole >>= shift_count;
-		cur.mperf_cur.whole >>= shift_count;
+		cur.aperf.whole >>= shift_count;
+		cur.mperf.whole >>= shift_count;
 	}
 
-	if (((unsigned long)(-1) / 100) < cur.aperf_cur.split.lo) {
+	if (((unsigned long)(-1) / 100) < cur.aperf.split.lo) {
 		int shift_count = 7;
-		cur.aperf_cur.split.lo >>= shift_count;
-		cur.mperf_cur.split.lo >>= shift_count;
+		cur.aperf.split.lo >>= shift_count;
+		cur.mperf.split.lo >>= shift_count;
 	}
 
-	if (cur.aperf_cur.split.lo && cur.mperf_cur.split.lo)
-		perf_percent = (cur.aperf_cur.split.lo * 100) /
-				cur.mperf_cur.split.lo;
+	if (cur.aperf.split.lo && cur.mperf.split.lo)
+		perf_percent = (cur.aperf.split.lo * 100) / cur.mperf.split.lo;
 	else
 		perf_percent = 0;
 
 #else
-	if (unlikely(((unsigned long)(-1) / 100) < cur.aperf_cur.whole)) {
+	if (unlikely(((unsigned long)(-1) / 100) < cur.aperf.whole)) {
 		int shift_count = 7;
-		cur.aperf_cur.whole >>= shift_count;
-		cur.mperf_cur.whole >>= shift_count;
+		cur.aperf.whole >>= shift_count;
+		cur.mperf.whole >>= shift_count;
 	}
 
-	if (cur.aperf_cur.whole && cur.mperf_cur.whole)
-		perf_percent = (cur.aperf_cur.whole * 100) /
-				cur.mperf_cur.whole;
+	if (cur.aperf.whole && cur.mperf.whole)
+		perf_percent = (cur.aperf.whole * 100) / cur.mperf.whole;
 	else
 		perf_percent = 0;
 
diff --git a/arch/x86/kernel/cpu/cpufreq/longhaul.c b/arch/x86/kernel/cpu/cpufreq/longhaul.c
index 0bd48e65a0c..ce2ed3e4aad 100644
--- a/arch/x86/kernel/cpu/cpufreq/longhaul.c
+++ b/arch/x86/kernel/cpu/cpufreq/longhaul.c
@@ -33,7 +33,6 @@
 #include <linux/timex.h>
 #include <linux/io.h>
 #include <linux/acpi.h>
-#include <linux/kernel.h>
 
 #include <asm/msr.h>
 #include <acpi/processor.h>
diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c
index 70a10ca100f..18dfa30795c 100644
--- a/arch/x86/kernel/ftrace.c
+++ b/arch/x86/kernel/ftrace.c
@@ -18,6 +18,8 @@
 #include <linux/init.h>
 #include <linux/list.h>
 
+#include <trace/syscall.h>
+
 #include <asm/cacheflush.h>
 #include <asm/ftrace.h>
 #include <asm/nops.h>
diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c
index 4d420de9ac6..98c470c069d 100644
--- a/arch/x86/kernel/microcode_core.c
+++ b/arch/x86/kernel/microcode_core.c
@@ -108,40 +108,29 @@ struct ucode_cpu_info		ucode_cpu_info[NR_CPUS];
 EXPORT_SYMBOL_GPL(ucode_cpu_info);
 
 #ifdef CONFIG_MICROCODE_OLD_INTERFACE
-struct update_for_cpu {
-	const void __user	*buf;
-	size_t			size;
-};
-
-static long update_for_cpu(void *_ufc)
-{
-	struct update_for_cpu *ufc = _ufc;
-	int error;
-
-	error = microcode_ops->request_microcode_user(smp_processor_id(),
-						      ufc->buf, ufc->size);
-	if (error < 0)
-		return error;
-	if (!error)
-		microcode_ops->apply_microcode(smp_processor_id());
-	return error;
-}
-
 static int do_microcode_update(const void __user *buf, size_t size)
 {
+	cpumask_t old;
 	int error = 0;
 	int cpu;
-	struct update_for_cpu ufc = { .buf = buf, .size = size };
+
+	old = current->cpus_allowed;
 
 	for_each_online_cpu(cpu) {
 		struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
 
 		if (!uci->valid)
 			continue;
-		error = work_on_cpu(cpu, update_for_cpu, &ufc);
+
+		set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu));
+		error = microcode_ops->request_microcode_user(cpu, buf, size);
 		if (error < 0)
-			break;
+			goto out;
+		if (!error)
+			microcode_ops->apply_microcode(cpu);
 	}
+out:
+	set_cpus_allowed_ptr(current, &old);
 	return error;
 }
 
diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index dce99dca6cf..70fd7e414c1 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -679,7 +679,7 @@ void __init get_smp_config(void)
 	__get_smp_config(0);
 }
 
-static void smp_reserve_bootmem(struct mpf_intel *mpf)
+static void __init smp_reserve_bootmem(struct mpf_intel *mpf)
 {
 	unsigned long size = get_mpc_size(mpf->physptr);
 #ifdef CONFIG_X86_32
@@ -838,7 +838,7 @@ static int  __init get_MP_intsrc_index(struct mpc_intsrc *m)
 
 static struct mpc_intsrc __initdata *m_spare[SPARE_SLOT_NUM];
 
-static void check_irq_src(struct mpc_intsrc *m, int *nr_m_spare)
+static void __init check_irq_src(struct mpc_intsrc *m, int *nr_m_spare)
 {
 	int i;
 
@@ -866,7 +866,8 @@ static void check_irq_src(struct mpc_intsrc *m, int *nr_m_spare)
 	}
 }
 #else /* CONFIG_X86_IO_APIC */
-static inline void check_irq_src(struct mpc_intsrc *m, int *nr_m_spare) {}
+static
+inline void __init check_irq_src(struct mpc_intsrc *m, int *nr_m_spare) {}
 #endif /* CONFIG_X86_IO_APIC */
 
 static int check_slot(unsigned long mpc_new_phys, unsigned long mpc_new_length,
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index fe9345c967d..23b7c8f017e 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -21,7 +21,6 @@
 #include <linux/audit.h>
 #include <linux/seccomp.h>
 #include <linux/signal.h>
-#include <linux/ftrace.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
@@ -35,6 +34,8 @@
 #include <asm/proto.h>
 #include <asm/ds.h>
 
+#include <trace/syscall.h>
+
 #include "tls.h"
 
 enum x86_regset {
diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c
index deb5ebb32c3..ed0c33761e6 100644
--- a/arch/x86/kernel/tlb_uv.c
+++ b/arch/x86/kernel/tlb_uv.c
@@ -25,6 +25,8 @@ static int			uv_bau_retry_limit __read_mostly;
 
 /* position of pnode (which is nasid>>1): */
 static int			uv_nshift __read_mostly;
+/* base pnode in this partition */
+static int			uv_partition_base_pnode __read_mostly;
 
 static unsigned long		uv_mmask __read_mostly;
 
@@ -32,6 +34,34 @@ static DEFINE_PER_CPU(struct ptc_stats, ptcstats);
 static DEFINE_PER_CPU(struct bau_control, bau_control);
 
 /*
+ * Determine the first node on a blade.
+ */
+static int __init blade_to_first_node(int blade)
+{
+	int node, b;
+
+	for_each_online_node(node) {
+		b = uv_node_to_blade_id(node);
+		if (blade == b)
+			return node;
+	}
+	return -1; /* shouldn't happen */
+}
+
+/*
+ * Determine the apicid of the first cpu on a blade.
+ */
+static int __init blade_to_first_apicid(int blade)
+{
+	int cpu;
+
+	for_each_present_cpu(cpu)
+		if (blade == uv_cpu_to_blade_id(cpu))
+			return per_cpu(x86_cpu_to_apicid, cpu);
+	return -1;
+}
+
+/*
  * Free a software acknowledge hardware resource by clearing its Pending
  * bit. This will return a reply to the sender.
  * If the message has timed out, a reply has already been sent by the
@@ -67,7 +97,7 @@ static void uv_bau_process_message(struct bau_payload_queue_entry *msg,
 	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()));
+		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;
@@ -215,14 +245,14 @@ static int uv_wait_completion(struct bau_desc *bau_desc,
  * Returns @flush_mask if some remote flushing remains to be done. The
  * mask will have some bits still set.
  */
-const struct cpumask *uv_flush_send_and_wait(int cpu, int this_blade,
+const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode,
 					     struct bau_desc *bau_desc,
 					     struct cpumask *flush_mask)
 {
 	int completion_status = 0;
 	int right_shift;
 	int tries = 0;
-	int blade;
+	int pnode;
 	int bit;
 	unsigned long mmr_offset;
 	unsigned long index;
@@ -265,8 +295,8 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_blade,
 	 * use the IPI method of shootdown on them.
 	 */
 	for_each_cpu(bit, flush_mask) {
-		blade = uv_cpu_to_blade_id(bit);
-		if (blade == this_blade)
+		pnode = uv_cpu_to_pnode(bit);
+		if (pnode == this_pnode)
 			continue;
 		cpumask_clear_cpu(bit, flush_mask);
 	}
@@ -309,16 +339,16 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
 	struct cpumask *flush_mask = __get_cpu_var(uv_flush_tlb_mask);
 	int i;
 	int bit;
-	int blade;
+	int pnode;
 	int uv_cpu;
-	int this_blade;
+	int this_pnode;
 	int locals = 0;
 	struct bau_desc *bau_desc;
 
 	cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu));
 
 	uv_cpu = uv_blade_processor_id();
-	this_blade = uv_numa_blade_id();
+	this_pnode = uv_hub_info->pnode;
 	bau_desc = __get_cpu_var(bau_control).descriptor_base;
 	bau_desc += UV_ITEMS_PER_DESCRIPTOR * uv_cpu;
 
@@ -326,13 +356,14 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
 
 	i = 0;
 	for_each_cpu(bit, flush_mask) {
-		blade = uv_cpu_to_blade_id(bit);
-		BUG_ON(blade > (UV_DISTRIBUTION_SIZE - 1));
-		if (blade == this_blade) {
+		pnode = uv_cpu_to_pnode(bit);
+		BUG_ON(pnode > (UV_DISTRIBUTION_SIZE - 1));
+		if (pnode == this_pnode) {
 			locals++;
 			continue;
 		}
-		bau_node_set(blade, &bau_desc->distribution);
+		bau_node_set(pnode - uv_partition_base_pnode,
+				&bau_desc->distribution);
 		i++;
 	}
 	if (i == 0) {
@@ -350,7 +381,7 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
 	bau_desc->payload.address = va;
 	bau_desc->payload.sending_cpu = cpu;
 
-	return uv_flush_send_and_wait(uv_cpu, this_blade, bau_desc, flush_mask);
+	return uv_flush_send_and_wait(uv_cpu, this_pnode, bau_desc, flush_mask);
 }
 
 /*
@@ -418,24 +449,58 @@ void uv_bau_message_interrupt(struct pt_regs *regs)
 	set_irq_regs(old_regs);
 }
 
+/*
+ * uv_enable_timeouts
+ *
+ * Each target blade (i.e. blades that have 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 i;
 	int blade;
-	int last_blade;
+	int nblades;
 	int pnode;
-	int cur_cpu = 0;
-	unsigned long apicid;
+	unsigned long mmr_image;
 
-	last_blade = -1;
-	for_each_online_node(i) {
-		blade = uv_node_to_blade_id(i);
-		if (blade == last_blade)
+	nblades = uv_num_possible_blades();
+
+	for (blade = 0; blade < nblades; blade++) {
+		if (!uv_blade_nr_possible_cpus(blade))
 			continue;
-		last_blade = blade;
-		apicid = per_cpu(x86_cpu_to_apicid, cur_cpu);
+
 		pnode = uv_blade_to_pnode(blade);
-		cur_cpu += uv_blade_nr_possible_cpus(i);
+		mmr_image =
+		    uv_read_global_mmr64(pnode, UVH_LB_BAU_MISC_CONTROL);
+		/*
+		 * Set the timeout period and then lock it in, in three
+		 * steps; captures and locks in the period.
+		 *
+		 * To program the period, the SOFT_ACK_MODE must be off.
+		 */
+		mmr_image &= ~((unsigned long)1 <<
+			       UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT);
+		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);
+		mmr_image |= (UV_INTD_SOFT_ACK_TIMEOUT_PERIOD <<
+			     UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT);
+		uv_write_global_mmr64
+		    (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
+		/*
+		 * Subsequent reversals of the timebase bit (3) cause an
+		 * immediate timeout of one or all INTD resources as
+		 * indicated in bits 2:0 (7 causes all of them to timeout).
+		 */
+		mmr_image |= ((unsigned long)1 <<
+			      UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT);
+		uv_write_global_mmr64
+		    (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image);
 	}
 }
 
@@ -482,8 +547,7 @@ static int uv_ptc_seq_show(struct seq_file *file, void *data)
 			   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",
-			   uv_read_global_mmr64(uv_blade_to_pnode
-					(uv_cpu_to_blade_id(cpu)),
+			   uv_read_global_mmr64(uv_cpu_to_pnode(cpu),
 					UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE),
 			   stat->sflush, stat->dflush,
 			   stat->retriesok, stat->nomsg,
@@ -617,16 +681,18 @@ static struct bau_control * __init uv_table_bases_init(int blade, int node)
  * finish the initialization of the per-blade control structures
  */
 static void __init
-uv_table_bases_finish(int blade, int node, int cur_cpu,
+uv_table_bases_finish(int blade,
 		      struct bau_control *bau_tablesp,
 		      struct bau_desc *adp)
 {
 	struct bau_control *bcp;
-	int i;
+	int cpu;
 
-	for (i = cur_cpu; i < cur_cpu + uv_blade_nr_possible_cpus(blade); i++) {
-		bcp = (struct bau_control *)&per_cpu(bau_control, i);
+	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;
@@ -649,11 +715,10 @@ uv_activation_descriptor_init(int node, int pnode)
 	struct bau_desc *adp;
 	struct bau_desc *ad2;
 
-	adp = (struct bau_desc *)
-	    kmalloc_node(16384, GFP_KERNEL, node);
+	adp = (struct bau_desc *)kmalloc_node(16384, GFP_KERNEL, node);
 	BUG_ON(!adp);
 
-	pa = __pa((unsigned long)adp);
+	pa = uv_gpa(adp); /* need the real nasid*/
 	n = pa >> uv_nshift;
 	m = pa & uv_mmask;
 
@@ -667,8 +732,12 @@ uv_activation_descriptor_init(int node, int pnode)
 	for (i = 0, ad2 = adp; i < UV_ACTIVATION_DESCRIPTOR_SIZE; i++, ad2++) {
 		memset(ad2, 0, sizeof(struct bau_desc));
 		ad2->header.sw_ack_flag = 1;
-		ad2->header.base_dest_nodeid =
-		    uv_blade_to_pnode(uv_cpu_to_blade_id(0));
+		/*
+		 * 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.
+		 */
+		ad2->header.base_dest_nodeid = uv_partition_base_pnode << 1;
 		ad2->header.command = UV_NET_ENDPOINT_INTD;
 		ad2->header.int_both = 1;
 		/*
@@ -686,6 +755,8 @@ static struct bau_payload_queue_entry * __init
 uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp)
 {
 	struct bau_payload_queue_entry *pqp;
+	unsigned long pa;
+	int pn;
 	char *cp;
 
 	pqp = (struct bau_payload_queue_entry *) kmalloc_node(
@@ -696,10 +767,14 @@ uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp)
 	cp = (char *)pqp + 31;
 	pqp = (struct bau_payload_queue_entry *)(((unsigned long)cp >> 5) << 5);
 	bau_tablesp->va_queue_first = pqp;
+	/*
+	 * need the pnode of where the memory was really allocated
+	 */
+	pa = uv_gpa(pqp);
+	pn = pa >> uv_nshift;
 	uv_write_global_mmr64(pnode,
 			      UVH_LB_BAU_INTD_PAYLOAD_QUEUE_FIRST,
-			      ((unsigned long)pnode <<
-			       UV_PAYLOADQ_PNODE_SHIFT) |
+			      ((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));
@@ -715,8 +790,9 @@ uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp)
 /*
  * Initialization of each UV blade's structures
  */
-static int __init uv_init_blade(int blade, int node, int cur_cpu)
+static int __init uv_init_blade(int blade)
 {
+	int node;
 	int pnode;
 	unsigned long pa;
 	unsigned long apicid;
@@ -724,16 +800,17 @@ static int __init uv_init_blade(int blade, int node, int cur_cpu)
 	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, node, cur_cpu, bau_tablesp, adp);
+	uv_table_bases_finish(blade, bau_tablesp, adp);
 	/*
 	 * the below initialization can't be in firmware because the
 	 * messaging IRQ will be determined by the OS
 	 */
-	apicid = per_cpu(x86_cpu_to_apicid, cur_cpu);
+	apicid = blade_to_first_apicid(blade);
 	pa = uv_read_global_mmr64(pnode, UVH_BAU_DATA_CONFIG);
 	if ((pa & 0xff) != UV_BAU_MESSAGE) {
 		uv_write_global_mmr64(pnode, UVH_BAU_DATA_CONFIG,
@@ -748,9 +825,7 @@ static int __init uv_init_blade(int blade, int node, int cur_cpu)
 static int __init uv_bau_init(void)
 {
 	int blade;
-	int node;
 	int nblades;
-	int last_blade;
 	int cur_cpu;
 
 	if (!is_uv_system())
@@ -763,29 +838,21 @@ static int __init uv_bau_init(void)
 	uv_bau_retry_limit = 1;
 	uv_nshift = uv_hub_info->n_val;
 	uv_mmask = (1UL << uv_hub_info->n_val) - 1;
-	nblades = 0;
-	last_blade = -1;
-	cur_cpu = 0;
-	for_each_online_node(node) {
-		blade = uv_node_to_blade_id(node);
-		if (blade == last_blade)
-			continue;
-		last_blade = blade;
-		nblades++;
-	}
+	nblades = 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);
 
-	last_blade = -1;
-	for_each_online_node(node) {
-		blade = uv_node_to_blade_id(node);
-		if (blade == last_blade)
-			continue;
-		last_blade = blade;
-		uv_init_blade(blade, node, cur_cpu);
-		cur_cpu += uv_blade_nr_possible_cpus(blade);
-	}
+	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);
 	uv_enable_timeouts();
 
diff --git a/arch/x86/kernel/uv_sysfs.c b/arch/x86/kernel/uv_sysfs.c
index 67f9b9dbf80..36afb98675a 100644
--- a/arch/x86/kernel/uv_sysfs.c
+++ b/arch/x86/kernel/uv_sysfs.c
@@ -21,6 +21,7 @@
 
 #include <linux/sysdev.h>
 #include <asm/uv/bios.h>
+#include <asm/uv/uv.h>
 
 struct kobject *sgi_uv_kobj;
 
@@ -47,6 +48,9 @@ static int __init sgi_uv_sysfs_init(void)
 {
 	unsigned long ret;
 
+	if (!is_uv_system())
+		return -ENODEV;
+
 	if (!sgi_uv_kobj)
 		sgi_uv_kobj = kobject_create_and_add("sgi_uv", firmware_kobj);
 	if (!sgi_uv_kobj) {
diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c
index 2b54fe002e9..0a5b04aa98f 100644
--- a/arch/x86/kernel/xsave.c
+++ b/arch/x86/kernel/xsave.c
@@ -324,7 +324,7 @@ void __ref xsave_cntxt_init(void)
 	}
 
 	/*
-	 * for now OS knows only about FP/SSE
+	 * Support only the state known to OS.
 	 */
 	pcntxt_mask = pcntxt_mask & XCNTXT_MASK;
 	xsave_init();