Initial commit; kernel source import

arch/x86/kernel/apic/Makefile

@@ -0,0 +1,27 @@
+#
+# Makefile for local APIC drivers and for the IO-APIC code
+#
+
+obj-$(CONFIG_X86_LOCAL_APIC)	+= apic.o apic_noop.o ipi.o
+obj-y				+= hw_nmi.o
+
+obj-$(CONFIG_X86_IO_APIC)	+= io_apic.o
+obj-$(CONFIG_SMP)		+= ipi.o
+
+ifeq ($(CONFIG_X86_64),y)
+# APIC probe will depend on the listing order here
+obj-$(CONFIG_X86_NUMACHIP)	+= apic_numachip.o
+obj-$(CONFIG_X86_UV)		+= x2apic_uv_x.o
+obj-$(CONFIG_X86_X2APIC)	+= x2apic_phys.o
+obj-$(CONFIG_X86_X2APIC)	+= x2apic_cluster.o
+obj-y				+= apic_flat_64.o
+endif
+
+# APIC probe will depend on the listing order here
+obj-$(CONFIG_X86_NUMAQ)		+= numaq_32.o
+obj-$(CONFIG_X86_SUMMIT)	+= summit_32.o
+obj-$(CONFIG_X86_BIGSMP)	+= bigsmp_32.o
+obj-$(CONFIG_X86_ES7000)	+= es7000_32.o
+
+# For 32bit, probe_32 need to be listed last
+obj-$(CONFIG_X86_LOCAL_APIC)	+= probe_$(BITS).o

arch/x86/kernel/apic/apic_flat_64.c

@@ -0,0 +1,336 @@
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Flat APIC subarch code.
+ *
+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
+ * James Cleverdon.
+ */
+#include <linux/errno.h>
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/hardirq.h>
+#include <linux/module.h>
+#include <asm/smp.h>
+#include <asm/apic.h>
+#include <asm/ipi.h>
+
+#ifdef CONFIG_ACPI
+#include <acpi/acpi_bus.h>
+#endif
+
+static struct apic apic_physflat;
+static struct apic apic_flat;
+
+struct apic __read_mostly *apic = &apic_flat;
+EXPORT_SYMBOL_GPL(apic);
+
+static int flat_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	return 1;
+}
+
+/*
+ * Set up the logical destination ID.
+ *
+ * Intel recommends to set DFR, LDR and TPR before enabling
+ * an APIC.  See e.g. "AP-388 82489DX User's Manual" (Intel
+ * document number 292116).  So here it goes...
+ */
+void flat_init_apic_ldr(void)
+{
+	unsigned long val;
+	unsigned long num, id;
+
+	num = smp_processor_id();
+	id = 1UL << num;
+	apic_write(APIC_DFR, APIC_DFR_FLAT);
+	val = apic_read(APIC_LDR) & ~APIC_LDR_MASK;
+	val |= SET_APIC_LOGICAL_ID(id);
+	apic_write(APIC_LDR, val);
+}
+
+static inline void _flat_send_IPI_mask(unsigned long mask, int vector)
+{
+	unsigned long flags;
+
+	local_irq_save(flags);
+	__default_send_IPI_dest_field(mask, vector, apic->dest_logical);
+	local_irq_restore(flags);
+}
+
+static void flat_send_IPI_mask(const struct cpumask *cpumask, int vector)
+{
+	unsigned long mask = cpumask_bits(cpumask)[0];
+
+	_flat_send_IPI_mask(mask, vector);
+}
+
+static void
+flat_send_IPI_mask_allbutself(const struct cpumask *cpumask, int vector)
+{
+	unsigned long mask = cpumask_bits(cpumask)[0];
+	int cpu = smp_processor_id();
+
+	if (cpu < BITS_PER_LONG)
+		clear_bit(cpu, &mask);
+
+	_flat_send_IPI_mask(mask, vector);
+}
+
+static void flat_send_IPI_allbutself(int vector)
+{
+	int cpu = smp_processor_id();
+#ifdef	CONFIG_HOTPLUG_CPU
+	int hotplug = 1;
+#else
+	int hotplug = 0;
+#endif
+	if (hotplug || vector == NMI_VECTOR) {
+		if (!cpumask_equal(cpu_online_mask, cpumask_of(cpu))) {
+			unsigned long mask = cpumask_bits(cpu_online_mask)[0];
+
+			if (cpu < BITS_PER_LONG)
+				clear_bit(cpu, &mask);
+
+			_flat_send_IPI_mask(mask, vector);
+		}
+	} else if (num_online_cpus() > 1) {
+		__default_send_IPI_shortcut(APIC_DEST_ALLBUT,
+					    vector, apic->dest_logical);
+	}
+}
+
+static void flat_send_IPI_all(int vector)
+{
+	if (vector == NMI_VECTOR) {
+		flat_send_IPI_mask(cpu_online_mask, vector);
+	} else {
+		__default_send_IPI_shortcut(APIC_DEST_ALLINC,
+					    vector, apic->dest_logical);
+	}
+}
+
+static unsigned int flat_get_apic_id(unsigned long x)
+{
+	unsigned int id;
+
+	id = (((x)>>24) & 0xFFu);
+
+	return id;
+}
+
+static unsigned long set_apic_id(unsigned int id)
+{
+	unsigned long x;
+
+	x = ((id & 0xFFu)<<24);
+	return x;
+}
+
+static unsigned int read_xapic_id(void)
+{
+	unsigned int id;
+
+	id = flat_get_apic_id(apic_read(APIC_ID));
+	return id;
+}
+
+static int flat_apic_id_registered(void)
+{
+	return physid_isset(read_xapic_id(), phys_cpu_present_map);
+}
+
+static int flat_phys_pkg_id(int initial_apic_id, int index_msb)
+{
+	return initial_apic_id >> index_msb;
+}
+
+static int flat_probe(void)
+{
+	return 1;
+}
+
+static struct apic apic_flat =  {
+	.name				= "flat",
+	.probe				= flat_probe,
+	.acpi_madt_oem_check		= flat_acpi_madt_oem_check,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= flat_apic_id_registered,
+
+	.irq_delivery_mode		= dest_LowestPrio,
+	.irq_dest_mode			= 1, /* logical */
+
+	.target_cpus			= online_target_cpus,
+	.disable_esr			= 0,
+	.dest_logical			= APIC_DEST_LOGICAL,
+	.check_apicid_used		= NULL,
+	.check_apicid_present		= NULL,
+
+	.vector_allocation_domain	= flat_vector_allocation_domain,
+	.init_apic_ldr			= flat_init_apic_ldr,
+
+	.ioapic_phys_id_map		= NULL,
+	.setup_apic_routing		= NULL,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= NULL,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= default_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= flat_phys_pkg_id,
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= flat_get_apic_id,
+	.set_apic_id			= set_apic_id,
+	.apic_id_mask			= 0xFFu << 24,
+
+	.cpu_mask_to_apicid_and		= flat_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= flat_send_IPI_mask,
+	.send_IPI_mask_allbutself	= flat_send_IPI_mask_allbutself,
+	.send_IPI_allbutself		= flat_send_IPI_allbutself,
+	.send_IPI_all			= flat_send_IPI_all,
+	.send_IPI_self			= apic_send_IPI_self,
+
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+	.wait_for_init_deassert		= NULL,
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= default_inquire_remote_apic,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+};
+
+/*
+ * Physflat mode is used when there are more than 8 CPUs on a system.
+ * We cannot use logical delivery in this case because the mask
+ * overflows, so use physical mode.
+ */
+static int physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+#ifdef CONFIG_ACPI
+	/*
+	 * Quirk: some x86_64 machines can only use physical APIC mode
+	 * regardless of how many processors are present (x86_64 ES7000
+	 * is an example).
+	 */
+	if (acpi_gbl_FADT.header.revision >= FADT2_REVISION_ID &&
+		(acpi_gbl_FADT.flags & ACPI_FADT_APIC_PHYSICAL)) {
+		printk(KERN_DEBUG "system APIC only can use physical flat");
+		return 1;
+	}
+
+	if (!strncmp(oem_id, "IBM", 3) && !strncmp(oem_table_id, "EXA", 3)) {
+		printk(KERN_DEBUG "IBM Summit detected, will use apic physical");
+		return 1;
+	}
+#endif
+
+	return 0;
+}
+
+static void physflat_send_IPI_mask(const struct cpumask *cpumask, int vector)
+{
+	default_send_IPI_mask_sequence_phys(cpumask, vector);
+}
+
+static void physflat_send_IPI_mask_allbutself(const struct cpumask *cpumask,
+					      int vector)
+{
+	default_send_IPI_mask_allbutself_phys(cpumask, vector);
+}
+
+static void physflat_send_IPI_allbutself(int vector)
+{
+	default_send_IPI_mask_allbutself_phys(cpu_online_mask, vector);
+}
+
+static void physflat_send_IPI_all(int vector)
+{
+	physflat_send_IPI_mask(cpu_online_mask, vector);
+}
+
+static int physflat_probe(void)
+{
+	if (apic == &apic_physflat || num_possible_cpus() > 8)
+		return 1;
+
+	return 0;
+}
+
+static struct apic apic_physflat =  {
+
+	.name				= "physical flat",
+	.probe				= physflat_probe,
+	.acpi_madt_oem_check		= physflat_acpi_madt_oem_check,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= flat_apic_id_registered,
+
+	.irq_delivery_mode		= dest_Fixed,
+	.irq_dest_mode			= 0, /* physical */
+
+	.target_cpus			= online_target_cpus,
+	.disable_esr			= 0,
+	.dest_logical			= 0,
+	.check_apicid_used		= NULL,
+	.check_apicid_present		= NULL,
+
+	.vector_allocation_domain	= default_vector_allocation_domain,
+	/* not needed, but shouldn't hurt: */
+	.init_apic_ldr			= flat_init_apic_ldr,
+
+	.ioapic_phys_id_map		= NULL,
+	.setup_apic_routing		= NULL,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= NULL,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= default_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= flat_phys_pkg_id,
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= flat_get_apic_id,
+	.set_apic_id			= set_apic_id,
+	.apic_id_mask			= 0xFFu << 24,
+
+	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= physflat_send_IPI_mask,
+	.send_IPI_mask_allbutself	= physflat_send_IPI_mask_allbutself,
+	.send_IPI_allbutself		= physflat_send_IPI_allbutself,
+	.send_IPI_all			= physflat_send_IPI_all,
+	.send_IPI_self			= apic_send_IPI_self,
+
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+	.wait_for_init_deassert		= NULL,
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= default_inquire_remote_apic,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+};
+
+/*
+ * We need to check for physflat first, so this order is important.
+ */
+apic_drivers(apic_physflat, apic_flat);

arch/x86/kernel/apic/apic_noop.c

@@ -0,0 +1,192 @@
+/*
+ * NOOP APIC driver.
+ *
+ * Does almost nothing and should be substituted by a real apic driver via
+ * probe routine.
+ *
+ * Though in case if apic is disabled (for some reason) we try
+ * to not uglify the caller's code and allow to call (some) apic routines
+ * like self-ipi, etc...
+ */
+
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <asm/fixmap.h>
+#include <asm/mpspec.h>
+#include <asm/apicdef.h>
+#include <asm/apic.h>
+#include <asm/setup.h>
+
+#include <linux/smp.h>
+#include <asm/ipi.h>
+
+#include <linux/interrupt.h>
+#include <asm/acpi.h>
+#include <asm/e820.h>
+
+static void noop_init_apic_ldr(void) { }
+static void noop_send_IPI_mask(const struct cpumask *cpumask, int vector) { }
+static void noop_send_IPI_mask_allbutself(const struct cpumask *cpumask, int vector) { }
+static void noop_send_IPI_allbutself(int vector) { }
+static void noop_send_IPI_all(int vector) { }
+static void noop_send_IPI_self(int vector) { }
+static void noop_apic_wait_icr_idle(void) { }
+static void noop_apic_icr_write(u32 low, u32 id) { }
+
+static int noop_wakeup_secondary_cpu(int apicid, unsigned long start_eip)
+{
+	return -1;
+}
+
+static u32 noop_safe_apic_wait_icr_idle(void)
+{
+	return 0;
+}
+
+static u64 noop_apic_icr_read(void)
+{
+	return 0;
+}
+
+static int noop_phys_pkg_id(int cpuid_apic, int index_msb)
+{
+	return 0;
+}
+
+static unsigned int noop_get_apic_id(unsigned long x)
+{
+	return 0;
+}
+
+static int noop_probe(void)
+{
+	/*
+	 * NOOP apic should not ever be
+	 * enabled via probe routine
+	 */
+	return 0;
+}
+
+static int noop_apic_id_registered(void)
+{
+	/*
+	 * if we would be really "pedantic"
+	 * we should pass read_apic_id() here
+	 * but since NOOP suppose APIC ID = 0
+	 * lets save a few cycles
+	 */
+	return physid_isset(0, phys_cpu_present_map);
+}
+
+static const struct cpumask *noop_target_cpus(void)
+{
+	/* only BSP here */
+	return cpumask_of(0);
+}
+
+static unsigned long noop_check_apicid_used(physid_mask_t *map, int apicid)
+{
+	return physid_isset(apicid, *map);
+}
+
+static unsigned long noop_check_apicid_present(int bit)
+{
+	return physid_isset(bit, phys_cpu_present_map);
+}
+
+static void noop_vector_allocation_domain(int cpu, struct cpumask *retmask,
+					  const struct cpumask *mask)
+{
+	if (cpu != 0)
+		pr_warning("APIC: Vector allocated for non-BSP cpu\n");
+	cpumask_copy(retmask, cpumask_of(cpu));
+}
+
+static u32 noop_apic_read(u32 reg)
+{
+	WARN_ON_ONCE((cpu_has_apic && !disable_apic));
+	return 0;
+}
+
+static void noop_apic_write(u32 reg, u32 v)
+{
+	WARN_ON_ONCE(cpu_has_apic && !disable_apic);
+}
+
+struct apic apic_noop = {
+	.name				= "noop",
+	.probe				= noop_probe,
+	.acpi_madt_oem_check		= NULL,
+
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= noop_apic_id_registered,
+
+	.irq_delivery_mode		= dest_LowestPrio,
+	/* logical delivery broadcast to all CPUs: */
+	.irq_dest_mode			= 1,
+
+	.target_cpus			= noop_target_cpus,
+	.disable_esr			= 0,
+	.dest_logical			= APIC_DEST_LOGICAL,
+	.check_apicid_used		= noop_check_apicid_used,
+	.check_apicid_present		= noop_check_apicid_present,
+
+	.vector_allocation_domain	= noop_vector_allocation_domain,
+	.init_apic_ldr			= noop_init_apic_ldr,
+
+	.ioapic_phys_id_map		= default_ioapic_phys_id_map,
+	.setup_apic_routing		= NULL,
+	.multi_timer_check		= NULL,
+
+	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
+
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= default_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+
+	.phys_pkg_id			= noop_phys_pkg_id,
+
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= noop_get_apic_id,
+	.set_apic_id			= NULL,
+	.apic_id_mask			= 0x0F << 24,
+
+	.cpu_mask_to_apicid_and		= flat_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= noop_send_IPI_mask,
+	.send_IPI_mask_allbutself	= noop_send_IPI_mask_allbutself,
+	.send_IPI_allbutself		= noop_send_IPI_allbutself,
+	.send_IPI_all			= noop_send_IPI_all,
+	.send_IPI_self			= noop_send_IPI_self,
+
+	.wakeup_secondary_cpu		= noop_wakeup_secondary_cpu,
+
+	/* should be safe */
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+
+	.wait_for_init_deassert		= NULL,
+
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= NULL,
+
+	.read				= noop_apic_read,
+	.write				= noop_apic_write,
+	.eoi_write			= noop_apic_write,
+	.icr_read			= noop_apic_icr_read,
+	.icr_write			= noop_apic_icr_write,
+	.wait_icr_idle			= noop_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= noop_safe_apic_wait_icr_idle,
+
+#ifdef CONFIG_X86_32
+	.x86_32_early_logical_apicid	= noop_x86_32_early_logical_apicid,
+#endif
+};

arch/x86/kernel/apic/apic_numachip.c

@@ -0,0 +1,264 @@
+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Numascale NumaConnect-Specific APIC Code
+ *
+ * Copyright (C) 2011 Numascale AS. All rights reserved.
+ *
+ * Send feedback to <support@numascale.com>
+ *
+ */
+
+#include <linux/errno.h>
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/hardirq.h>
+#include <linux/delay.h>
+
+#include <asm/numachip/numachip.h>
+#include <asm/numachip/numachip_csr.h>
+#include <asm/smp.h>
+#include <asm/apic.h>
+#include <asm/ipi.h>
+#include <asm/apic_flat_64.h>
+#include <asm/pgtable.h>
+
+static int numachip_system __read_mostly;
+
+static const struct apic apic_numachip __read_mostly;
+
+static unsigned int get_apic_id(unsigned long x)
+{
+	unsigned long value;
+	unsigned int id;
+
+	rdmsrl(MSR_FAM10H_NODE_ID, value);
+	id = ((x >> 24) & 0xffU) | ((value << 2) & 0x3f00U);
+
+	return id;
+}
+
+static unsigned long set_apic_id(unsigned int id)
+{
+	unsigned long x;
+
+	x = ((id & 0xffU) << 24);
+	return x;
+}
+
+static unsigned int read_xapic_id(void)
+{
+	return get_apic_id(apic_read(APIC_ID));
+}
+
+static int numachip_apic_id_valid(int apicid)
+{
+	/* Trust what bootloader passes in MADT */
+	return 1;
+}
+
+static int numachip_apic_id_registered(void)
+{
+	return physid_isset(read_xapic_id(), phys_cpu_present_map);
+}
+
+static int numachip_phys_pkg_id(int initial_apic_id, int index_msb)
+{
+	return initial_apic_id >> index_msb;
+}
+
+static int __cpuinit numachip_wakeup_secondary(int phys_apicid, unsigned long start_rip)
+{
+	union numachip_csr_g3_ext_irq_gen int_gen;
+
+	int_gen.s._destination_apic_id = phys_apicid;
+	int_gen.s._vector = 0;
+	int_gen.s._msgtype = APIC_DM_INIT >> 8;
+	int_gen.s._index = 0;
+
+	write_lcsr(CSR_G3_EXT_IRQ_GEN, int_gen.v);
+
+	int_gen.s._msgtype = APIC_DM_STARTUP >> 8;
+	int_gen.s._vector = start_rip >> 12;
+
+	write_lcsr(CSR_G3_EXT_IRQ_GEN, int_gen.v);
+
+	atomic_set(&init_deasserted, 1);
+	return 0;
+}
+
+static void numachip_send_IPI_one(int cpu, int vector)
+{
+	union numachip_csr_g3_ext_irq_gen int_gen;
+	int apicid = per_cpu(x86_cpu_to_apicid, cpu);
+
+	int_gen.s._destination_apic_id = apicid;
+	int_gen.s._vector = vector;
+	int_gen.s._msgtype = (vector == NMI_VECTOR ? APIC_DM_NMI : APIC_DM_FIXED) >> 8;
+	int_gen.s._index = 0;
+
+	write_lcsr(CSR_G3_EXT_IRQ_GEN, int_gen.v);
+}
+
+static void numachip_send_IPI_mask(const struct cpumask *mask, int vector)
+{
+	unsigned int cpu;
+
+	for_each_cpu(cpu, mask)
+		numachip_send_IPI_one(cpu, vector);
+}
+
+static void numachip_send_IPI_mask_allbutself(const struct cpumask *mask,
+						int vector)
+{
+	unsigned int this_cpu = smp_processor_id();
+	unsigned int cpu;
+
+	for_each_cpu(cpu, mask) {
+		if (cpu != this_cpu)
+			numachip_send_IPI_one(cpu, vector);
+	}
+}
+
+static void numachip_send_IPI_allbutself(int vector)
+{
+	unsigned int this_cpu = smp_processor_id();
+	unsigned int cpu;
+
+	for_each_online_cpu(cpu) {
+		if (cpu != this_cpu)
+			numachip_send_IPI_one(cpu, vector);
+	}
+}
+
+static void numachip_send_IPI_all(int vector)
+{
+	numachip_send_IPI_mask(cpu_online_mask, vector);
+}
+
+static void numachip_send_IPI_self(int vector)
+{
+	__default_send_IPI_shortcut(APIC_DEST_SELF, vector, APIC_DEST_PHYSICAL);
+}
+
+static int __init numachip_probe(void)
+{
+	return apic == &apic_numachip;
+}
+
+static void __init map_csrs(void)
+{
+	printk(KERN_INFO "NumaChip: Mapping local CSR space (%016llx - %016llx)\n",
+		NUMACHIP_LCSR_BASE, NUMACHIP_LCSR_BASE + NUMACHIP_LCSR_SIZE - 1);
+	init_extra_mapping_uc(NUMACHIP_LCSR_BASE, NUMACHIP_LCSR_SIZE);
+
+	printk(KERN_INFO "NumaChip: Mapping global CSR space (%016llx - %016llx)\n",
+		NUMACHIP_GCSR_BASE, NUMACHIP_GCSR_BASE + NUMACHIP_GCSR_SIZE - 1);
+	init_extra_mapping_uc(NUMACHIP_GCSR_BASE, NUMACHIP_GCSR_SIZE);
+}
+
+static void fixup_cpu_id(struct cpuinfo_x86 *c, int node)
+{
+
+	if (c->phys_proc_id != node) {
+		c->phys_proc_id = node;
+		per_cpu(cpu_llc_id, smp_processor_id()) = node;
+	}
+}
+
+static int __init numachip_system_init(void)
+{
+	unsigned int val;
+
+	if (!numachip_system)
+		return 0;
+
+	x86_cpuinit.fixup_cpu_id = fixup_cpu_id;
+	x86_init.pci.arch_init = pci_numachip_init;
+
+	map_csrs();
+
+	val = read_lcsr(CSR_G0_NODE_IDS);
+	printk(KERN_INFO "NumaChip: Local NodeID = %08x\n", val);
+
+	return 0;
+}
+early_initcall(numachip_system_init);
+
+static int numachip_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	if (!strncmp(oem_id, "NUMASC", 6)) {
+		numachip_system = 1;
+		return 1;
+	}
+
+	return 0;
+}
+
+static const struct apic apic_numachip __refconst = {
+
+	.name				= "NumaConnect system",
+	.probe				= numachip_probe,
+	.acpi_madt_oem_check		= numachip_acpi_madt_oem_check,
+	.apic_id_valid			= numachip_apic_id_valid,
+	.apic_id_registered		= numachip_apic_id_registered,
+
+	.irq_delivery_mode		= dest_Fixed,
+	.irq_dest_mode			= 0, /* physical */
+
+	.target_cpus			= online_target_cpus,
+	.disable_esr			= 0,
+	.dest_logical			= 0,
+	.check_apicid_used		= NULL,
+	.check_apicid_present		= NULL,
+
+	.vector_allocation_domain	= default_vector_allocation_domain,
+	.init_apic_ldr			= flat_init_apic_ldr,
+
+	.ioapic_phys_id_map		= NULL,
+	.setup_apic_routing		= NULL,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= NULL,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= default_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= numachip_phys_pkg_id,
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= get_apic_id,
+	.set_apic_id			= set_apic_id,
+	.apic_id_mask			= 0xffU << 24,
+
+	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= numachip_send_IPI_mask,
+	.send_IPI_mask_allbutself	= numachip_send_IPI_mask_allbutself,
+	.send_IPI_allbutself		= numachip_send_IPI_allbutself,
+	.send_IPI_all			= numachip_send_IPI_all,
+	.send_IPI_self			= numachip_send_IPI_self,
+
+	.wakeup_secondary_cpu		= numachip_wakeup_secondary,
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+	.wait_for_init_deassert		= NULL,
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= NULL, /* REMRD not supported */
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+};
+apic_driver(apic_numachip);
+

arch/x86/kernel/apic/bigsmp_32.c

@@ -0,0 +1,238 @@
+/*
+ * APIC driver for "bigsmp" xAPIC machines with more than 8 virtual CPUs.
+ *
+ * Drives the local APIC in "clustered mode".
+ */
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/dmi.h>
+#include <linux/smp.h>
+
+#include <asm/apicdef.h>
+#include <asm/fixmap.h>
+#include <asm/mpspec.h>
+#include <asm/apic.h>
+#include <asm/ipi.h>
+
+static unsigned bigsmp_get_apic_id(unsigned long x)
+{
+	return (x >> 24) & 0xFF;
+}
+
+static int bigsmp_apic_id_registered(void)
+{
+	return 1;
+}
+
+static unsigned long bigsmp_check_apicid_used(physid_mask_t *map, int apicid)
+{
+	return 0;
+}
+
+static unsigned long bigsmp_check_apicid_present(int bit)
+{
+	return 1;
+}
+
+static int bigsmp_early_logical_apicid(int cpu)
+{
+	/* on bigsmp, logical apicid is the same as physical */
+	return early_per_cpu(x86_cpu_to_apicid, cpu);
+}
+
+static inline unsigned long calculate_ldr(int cpu)
+{
+	unsigned long val, id;
+
+	val = apic_read(APIC_LDR) & ~APIC_LDR_MASK;
+	id = per_cpu(x86_bios_cpu_apicid, cpu);
+	val |= SET_APIC_LOGICAL_ID(id);
+
+	return val;
+}
+
+/*
+ * Set up the logical destination ID.
+ *
+ * Intel recommends to set DFR, LDR and TPR before enabling
+ * an APIC.  See e.g. "AP-388 82489DX User's Manual" (Intel
+ * document number 292116).  So here it goes...
+ */
+static void bigsmp_init_apic_ldr(void)
+{
+	unsigned long val;
+	int cpu = smp_processor_id();
+
+	apic_write(APIC_DFR, APIC_DFR_FLAT);
+	val = calculate_ldr(cpu);
+	apic_write(APIC_LDR, val);
+}
+
+static void bigsmp_setup_apic_routing(void)
+{
+	printk(KERN_INFO
+		"Enabling APIC mode:  Physflat.  Using %d I/O APICs\n",
+		nr_ioapics);
+}
+
+static int bigsmp_cpu_present_to_apicid(int mps_cpu)
+{
+	if (mps_cpu < nr_cpu_ids)
+		return (int) per_cpu(x86_bios_cpu_apicid, mps_cpu);
+
+	return BAD_APICID;
+}
+
+static void bigsmp_ioapic_phys_id_map(physid_mask_t *phys_map, physid_mask_t *retmap)
+{
+	/* For clustered we don't have a good way to do this yet - hack */
+	physids_promote(0xFFL, retmap);
+}
+
+static int bigsmp_check_phys_apicid_present(int phys_apicid)
+{
+	return 1;
+}
+
+static int bigsmp_phys_pkg_id(int cpuid_apic, int index_msb)
+{
+	return cpuid_apic >> index_msb;
+}
+
+static inline void bigsmp_send_IPI_mask(const struct cpumask *mask, int vector)
+{
+	default_send_IPI_mask_sequence_phys(mask, vector);
+}
+
+static void bigsmp_send_IPI_allbutself(int vector)
+{
+	default_send_IPI_mask_allbutself_phys(cpu_online_mask, vector);
+}
+
+static void bigsmp_send_IPI_all(int vector)
+{
+	bigsmp_send_IPI_mask(cpu_online_mask, vector);
+}
+
+static int dmi_bigsmp; /* can be set by dmi scanners */
+
+static int hp_ht_bigsmp(const struct dmi_system_id *d)
+{
+	printk(KERN_NOTICE "%s detected: force use of apic=bigsmp\n", d->ident);
+	dmi_bigsmp = 1;
+
+	return 0;
+}
+
+
+static const struct dmi_system_id bigsmp_dmi_table[] = {
+	{ hp_ht_bigsmp, "HP ProLiant DL760 G2",
+		{	DMI_MATCH(DMI_BIOS_VENDOR, "HP"),
+			DMI_MATCH(DMI_BIOS_VERSION, "P44-"),
+		}
+	},
+
+	{ hp_ht_bigsmp, "HP ProLiant DL740",
+		{	DMI_MATCH(DMI_BIOS_VENDOR, "HP"),
+			DMI_MATCH(DMI_BIOS_VERSION, "P47-"),
+		}
+	},
+	{ } /* NULL entry stops DMI scanning */
+};
+
+static int probe_bigsmp(void)
+{
+	if (def_to_bigsmp)
+		dmi_bigsmp = 1;
+	else
+		dmi_check_system(bigsmp_dmi_table);
+
+	return dmi_bigsmp;
+}
+
+static struct apic apic_bigsmp = {
+
+	.name				= "bigsmp",
+	.probe				= probe_bigsmp,
+	.acpi_madt_oem_check		= NULL,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= bigsmp_apic_id_registered,
+
+	.irq_delivery_mode		= dest_Fixed,
+	/* phys delivery to target CPU: */
+	.irq_dest_mode			= 0,
+
+	.target_cpus			= default_target_cpus,
+	.disable_esr			= 1,
+	.dest_logical			= 0,
+	.check_apicid_used		= bigsmp_check_apicid_used,
+	.check_apicid_present		= bigsmp_check_apicid_present,
+
+	.vector_allocation_domain	= default_vector_allocation_domain,
+	.init_apic_ldr			= bigsmp_init_apic_ldr,
+
+	.ioapic_phys_id_map		= bigsmp_ioapic_phys_id_map,
+	.setup_apic_routing		= bigsmp_setup_apic_routing,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= bigsmp_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= bigsmp_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= bigsmp_phys_pkg_id,
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= bigsmp_get_apic_id,
+	.set_apic_id			= NULL,
+	.apic_id_mask			= 0xFF << 24,
+
+	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= bigsmp_send_IPI_mask,
+	.send_IPI_mask_allbutself	= NULL,
+	.send_IPI_allbutself		= bigsmp_send_IPI_allbutself,
+	.send_IPI_all			= bigsmp_send_IPI_all,
+	.send_IPI_self			= default_send_IPI_self,
+
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+
+	.wait_for_init_deassert		= default_wait_for_init_deassert,
+
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= default_inquire_remote_apic,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+
+	.x86_32_early_logical_apicid	= bigsmp_early_logical_apicid,
+};
+
+void __init generic_bigsmp_probe(void)
+{
+	unsigned int cpu;
+
+	if (!probe_bigsmp())
+		return;
+
+	apic = &apic_bigsmp;
+
+	for_each_possible_cpu(cpu) {
+		if (early_per_cpu(x86_cpu_to_logical_apicid,
+				  cpu) == BAD_APICID)
+			continue;
+		early_per_cpu(x86_cpu_to_logical_apicid, cpu) =
+			bigsmp_early_logical_apicid(cpu);
+	}
+
+	pr_info("Overriding APIC driver with %s\n", apic_bigsmp.name);
+}
+
+apic_driver(apic_bigsmp);

arch/x86/kernel/apic/es7000_32.c

@@ -0,0 +1,746 @@
+/*
+ * Written by: Garry Forsgren, Unisys Corporation
+ *             Natalie Protasevich, Unisys Corporation
+ *
+ * This file contains the code to configure and interface
+ * with Unisys ES7000 series hardware system manager.
+ *
+ * Copyright (c) 2003 Unisys Corporation.
+ * Copyright (C) 2009, Red Hat, Inc., Ingo Molnar
+ *
+ *   All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write the Free Software Foundation, Inc., 59
+ * Temple Place - Suite 330, Boston MA 02111-1307, USA.
+ *
+ * Contact information: Unisys Corporation, Township Line & Union Meeting
+ * Roads-A, Unisys Way, Blue Bell, Pennsylvania, 19424, or:
+ *
+ * http://www.unisys.com
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/notifier.h>
+#include <linux/spinlock.h>
+#include <linux/cpumask.h>
+#include <linux/threads.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/reboot.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <linux/acpi.h>
+#include <linux/init.h>
+#include <linux/gfp.h>
+#include <linux/nmi.h>
+#include <linux/smp.h>
+#include <linux/io.h>
+
+#include <asm/apicdef.h>
+#include <linux/atomic.h>
+#include <asm/fixmap.h>
+#include <asm/mpspec.h>
+#include <asm/setup.h>
+#include <asm/apic.h>
+#include <asm/ipi.h>
+
+/*
+ * ES7000 chipsets
+ */
+
+#define NON_UNISYS			0
+#define ES7000_CLASSIC			1
+#define ES7000_ZORRO			2
+
+#define	MIP_REG				1
+#define	MIP_PSAI_REG			4
+
+#define	MIP_BUSY			1
+#define	MIP_SPIN			0xf0000
+#define	MIP_VALID			0x0100000000000000ULL
+#define	MIP_SW_APIC			0x1020b
+
+#define	MIP_PORT(val)			((val >> 32) & 0xffff)
+
+#define	MIP_RD_LO(val)			(val & 0xffffffff)
+
+struct mip_reg {
+	unsigned long long		off_0x00;
+	unsigned long long		off_0x08;
+	unsigned long long		off_0x10;
+	unsigned long long		off_0x18;
+	unsigned long long		off_0x20;
+	unsigned long long		off_0x28;
+	unsigned long long		off_0x30;
+	unsigned long long		off_0x38;
+};
+
+struct mip_reg_info {
+	unsigned long long		mip_info;
+	unsigned long long		delivery_info;
+	unsigned long long		host_reg;
+	unsigned long long		mip_reg;
+};
+
+struct psai {
+	unsigned long long		entry_type;
+	unsigned long long		addr;
+	unsigned long long		bep_addr;
+};
+
+#ifdef CONFIG_ACPI
+
+struct es7000_oem_table {
+	struct acpi_table_header	Header;
+	u32				OEMTableAddr;
+	u32				OEMTableSize;
+};
+
+static unsigned long			oem_addrX;
+static unsigned long			oem_size;
+
+#endif
+
+/*
+ * ES7000 Globals
+ */
+
+static volatile unsigned long		*psai;
+static struct mip_reg			*mip_reg;
+static struct mip_reg			*host_reg;
+static int 				mip_port;
+static unsigned long			mip_addr;
+static unsigned long			host_addr;
+
+int					es7000_plat;
+
+/*
+ * GSI override for ES7000 platforms.
+ */
+
+
+static int __cpuinit wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip)
+{
+	unsigned long vect = 0, psaival = 0;
+
+	if (psai == NULL)
+		return -1;
+
+	vect = ((unsigned long)__pa(eip)/0x1000) << 16;
+	psaival = (0x1000000 | vect | cpu);
+
+	while (*psai & 0x1000000)
+		;
+
+	*psai = psaival;
+
+	return 0;
+}
+
+static int es7000_apic_is_cluster(void)
+{
+	/* MPENTIUMIII */
+	if (boot_cpu_data.x86 == 6 &&
+	    (boot_cpu_data.x86_model >= 7 && boot_cpu_data.x86_model <= 11))
+		return 1;
+
+	return 0;
+}
+
+static void setup_unisys(void)
+{
+	/*
+	 * Determine the generation of the ES7000 currently running.
+	 *
+	 * es7000_plat = 1 if the machine is a 5xx ES7000 box
+	 * es7000_plat = 2 if the machine is a x86_64 ES7000 box
+	 *
+	 */
+	if (!(boot_cpu_data.x86 <= 15 && boot_cpu_data.x86_model <= 2))
+		es7000_plat = ES7000_ZORRO;
+	else
+		es7000_plat = ES7000_CLASSIC;
+}
+
+/*
+ * Parse the OEM Table:
+ */
+static int parse_unisys_oem(char *oemptr)
+{
+	int			i;
+	int 			success = 0;
+	unsigned char		type, size;
+	unsigned long		val;
+	char			*tp = NULL;
+	struct psai		*psaip = NULL;
+	struct mip_reg_info 	*mi;
+	struct mip_reg		*host, *mip;
+
+	tp = oemptr;
+
+	tp += 8;
+
+	for (i = 0; i <= 6; i++) {
+		type = *tp++;
+		size = *tp++;
+		tp -= 2;
+		switch (type) {
+		case MIP_REG:
+			mi = (struct mip_reg_info *)tp;
+			val = MIP_RD_LO(mi->host_reg);
+			host_addr = val;
+			host = (struct mip_reg *)val;
+			host_reg = __va(host);
+			val = MIP_RD_LO(mi->mip_reg);
+			mip_port = MIP_PORT(mi->mip_info);
+			mip_addr = val;
+			mip = (struct mip_reg *)val;
+			mip_reg = __va(mip);
+			pr_debug("host_reg = 0x%lx\n",
+				 (unsigned long)host_reg);
+			pr_debug("mip_reg = 0x%lx\n",
+				 (unsigned long)mip_reg);
+			success++;
+			break;
+		case MIP_PSAI_REG:
+			psaip = (struct psai *)tp;
+			if (tp != NULL) {
+				if (psaip->addr)
+					psai = __va(psaip->addr);
+				else
+					psai = NULL;
+				success++;
+			}
+			break;
+		default:
+			break;
+		}
+		tp += size;
+	}
+
+	if (success < 2)
+		es7000_plat = NON_UNISYS;
+	else
+		setup_unisys();
+
+	return es7000_plat;
+}
+
+#ifdef CONFIG_ACPI
+static int __init find_unisys_acpi_oem_table(unsigned long *oem_addr)
+{
+	struct acpi_table_header *header = NULL;
+	struct es7000_oem_table *table;
+	acpi_size tbl_size;
+	acpi_status ret;
+	int i = 0;
+
+	for (;;) {
+		ret = acpi_get_table_with_size("OEM1", i++, &header, &tbl_size);
+		if (!ACPI_SUCCESS(ret))
+			return -1;
+
+		if (!memcmp((char *) &header->oem_id, "UNISYS", 6))
+			break;
+
+		early_acpi_os_unmap_memory(header, tbl_size);
+	}
+
+	table = (void *)header;
+
+	oem_addrX	= table->OEMTableAddr;
+	oem_size	= table->OEMTableSize;
+
+	early_acpi_os_unmap_memory(header, tbl_size);
+
+	*oem_addr	= (unsigned long)__acpi_map_table(oem_addrX, oem_size);
+
+	return 0;
+}
+
+static void __init unmap_unisys_acpi_oem_table(unsigned long oem_addr)
+{
+	if (!oem_addr)
+		return;
+
+	__acpi_unmap_table((char *)oem_addr, oem_size);
+}
+
+static int es7000_check_dsdt(void)
+{
+	struct acpi_table_header header;
+
+	if (ACPI_SUCCESS(acpi_get_table_header(ACPI_SIG_DSDT, 0, &header)) &&
+	    !strncmp(header.oem_id, "UNISYS", 6))
+		return 1;
+	return 0;
+}
+
+static int es7000_acpi_ret;
+
+/* Hook from generic ACPI tables.c */
+static int __init es7000_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	unsigned long oem_addr = 0;
+	int check_dsdt;
+	int ret = 0;
+
+	/* check dsdt at first to avoid clear fix_map for oem_addr */
+	check_dsdt = es7000_check_dsdt();
+
+	if (!find_unisys_acpi_oem_table(&oem_addr)) {
+		if (check_dsdt) {
+			ret = parse_unisys_oem((char *)oem_addr);
+		} else {
+			setup_unisys();
+			ret = 1;
+		}
+		/*
+		 * we need to unmap it
+		 */
+		unmap_unisys_acpi_oem_table(oem_addr);
+	}
+
+	es7000_acpi_ret = ret;
+
+	return ret && !es7000_apic_is_cluster();
+}
+
+static int es7000_acpi_madt_oem_check_cluster(char *oem_id, char *oem_table_id)
+{
+	int ret = es7000_acpi_ret;
+
+	return ret && es7000_apic_is_cluster();
+}
+
+#else /* !CONFIG_ACPI: */
+static int es7000_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	return 0;
+}
+
+static int es7000_acpi_madt_oem_check_cluster(char *oem_id, char *oem_table_id)
+{
+	return 0;
+}
+#endif /* !CONFIG_ACPI */
+
+static void es7000_spin(int n)
+{
+	int i = 0;
+
+	while (i++ < n)
+		rep_nop();
+}
+
+static int es7000_mip_write(struct mip_reg *mip_reg)
+{
+	int status = 0;
+	int spin;
+
+	spin = MIP_SPIN;
+	while ((host_reg->off_0x38 & MIP_VALID) != 0) {
+		if (--spin <= 0) {
+			WARN(1,	"Timeout waiting for Host Valid Flag\n");
+			return -1;
+		}
+		es7000_spin(MIP_SPIN);
+	}
+
+	memcpy(host_reg, mip_reg, sizeof(struct mip_reg));
+	outb(1, mip_port);
+
+	spin = MIP_SPIN;
+
+	while ((mip_reg->off_0x38 & MIP_VALID) == 0) {
+		if (--spin <= 0) {
+			WARN(1,	"Timeout waiting for MIP Valid Flag\n");
+			return -1;
+		}
+		es7000_spin(MIP_SPIN);
+	}
+
+	status = (mip_reg->off_0x00 & 0xffff0000000000ULL) >> 48;
+	mip_reg->off_0x38 &= ~MIP_VALID;
+
+	return status;
+}
+
+static void es7000_enable_apic_mode(void)
+{
+	struct mip_reg es7000_mip_reg;
+	int mip_status;
+
+	if (!es7000_plat)
+		return;
+
+	pr_info("Enabling APIC mode.\n");
+	memset(&es7000_mip_reg, 0, sizeof(struct mip_reg));
+	es7000_mip_reg.off_0x00 = MIP_SW_APIC;
+	es7000_mip_reg.off_0x38 = MIP_VALID;
+
+	while ((mip_status = es7000_mip_write(&es7000_mip_reg)) != 0)
+		WARN(1, "Command failed, status = %x\n", mip_status);
+}
+
+static void es7000_wait_for_init_deassert(atomic_t *deassert)
+{
+	while (!atomic_read(deassert))
+		cpu_relax();
+}
+
+static unsigned int es7000_get_apic_id(unsigned long x)
+{
+	return (x >> 24) & 0xFF;
+}
+
+static void es7000_send_IPI_mask(const struct cpumask *mask, int vector)
+{
+	default_send_IPI_mask_sequence_phys(mask, vector);
+}
+
+static void es7000_send_IPI_allbutself(int vector)
+{
+	default_send_IPI_mask_allbutself_phys(cpu_online_mask, vector);
+}
+
+static void es7000_send_IPI_all(int vector)
+{
+	es7000_send_IPI_mask(cpu_online_mask, vector);
+}
+
+static int es7000_apic_id_registered(void)
+{
+	return 1;
+}
+
+static const struct cpumask *target_cpus_cluster(void)
+{
+	return cpu_all_mask;
+}
+
+static const struct cpumask *es7000_target_cpus(void)
+{
+	return cpumask_of(smp_processor_id());
+}
+
+static unsigned long es7000_check_apicid_used(physid_mask_t *map, int apicid)
+{
+	return 0;
+}
+
+static unsigned long es7000_check_apicid_present(int bit)
+{
+	return physid_isset(bit, phys_cpu_present_map);
+}
+
+static int es7000_early_logical_apicid(int cpu)
+{
+	/* on es7000, logical apicid is the same as physical */
+	return early_per_cpu(x86_bios_cpu_apicid, cpu);
+}
+
+static unsigned long calculate_ldr(int cpu)
+{
+	unsigned long id = per_cpu(x86_bios_cpu_apicid, cpu);
+
+	return SET_APIC_LOGICAL_ID(id);
+}
+
+/*
+ * Set up the logical destination ID.
+ *
+ * Intel recommends to set DFR, LdR and TPR before enabling
+ * an APIC.  See e.g. "AP-388 82489DX User's Manual" (Intel
+ * document number 292116).  So here it goes...
+ */
+static void es7000_init_apic_ldr_cluster(void)
+{
+	unsigned long val;
+	int cpu = smp_processor_id();
+
+	apic_write(APIC_DFR, APIC_DFR_CLUSTER);
+	val = calculate_ldr(cpu);
+	apic_write(APIC_LDR, val);
+}
+
+static void es7000_init_apic_ldr(void)
+{
+	unsigned long val;
+	int cpu = smp_processor_id();
+
+	apic_write(APIC_DFR, APIC_DFR_FLAT);
+	val = calculate_ldr(cpu);
+	apic_write(APIC_LDR, val);
+}
+
+static void es7000_setup_apic_routing(void)
+{
+	int apic = per_cpu(x86_bios_cpu_apicid, smp_processor_id());
+
+	pr_info("Enabling APIC mode:  %s. Using %d I/O APICs, target cpus %lx\n",
+		(apic_version[apic] == 0x14) ?
+			"Physical Cluster" : "Logical Cluster",
+		nr_ioapics, cpumask_bits(es7000_target_cpus())[0]);
+}
+
+static int es7000_cpu_present_to_apicid(int mps_cpu)
+{
+	if (!mps_cpu)
+		return boot_cpu_physical_apicid;
+	else if (mps_cpu < nr_cpu_ids)
+		return per_cpu(x86_bios_cpu_apicid, mps_cpu);
+	else
+		return BAD_APICID;
+}
+
+static int cpu_id;
+
+static void es7000_apicid_to_cpu_present(int phys_apicid, physid_mask_t *retmap)
+{
+	physid_set_mask_of_physid(cpu_id, retmap);
+	++cpu_id;
+}
+
+static void es7000_ioapic_phys_id_map(physid_mask_t *phys_map, physid_mask_t *retmap)
+{
+	/* For clustered we don't have a good way to do this yet - hack */
+	physids_promote(0xFFL, retmap);
+}
+
+static int es7000_check_phys_apicid_present(int cpu_physical_apicid)
+{
+	boot_cpu_physical_apicid = read_apic_id();
+	return 1;
+}
+
+static inline int
+es7000_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
+{
+	unsigned int round = 0;
+	unsigned int cpu, uninitialized_var(apicid);
+
+	/*
+	 * The cpus in the mask must all be on the apic cluster.
+	 */
+	for_each_cpu_and(cpu, cpumask, cpu_online_mask) {
+		int new_apicid = early_per_cpu(x86_cpu_to_logical_apicid, cpu);
+
+		if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
+			WARN(1, "Not a valid mask!");
+
+			return -EINVAL;
+		}
+		apicid |= new_apicid;
+		round++;
+	}
+	if (!round)
+		return -EINVAL;
+	*dest_id = apicid;
+	return 0;
+}
+
+static int
+es7000_cpu_mask_to_apicid_and(const struct cpumask *inmask,
+			      const struct cpumask *andmask,
+			      unsigned int *apicid)
+{
+	cpumask_var_t cpumask;
+	*apicid = early_per_cpu(x86_cpu_to_logical_apicid, 0);
+
+	if (!alloc_cpumask_var(&cpumask, GFP_ATOMIC))
+		return 0;
+
+	cpumask_and(cpumask, inmask, andmask);
+	es7000_cpu_mask_to_apicid(cpumask, apicid);
+
+	free_cpumask_var(cpumask);
+
+	return 0;
+}
+
+static int es7000_phys_pkg_id(int cpuid_apic, int index_msb)
+{
+	return cpuid_apic >> index_msb;
+}
+
+static int probe_es7000(void)
+{
+	/* probed later in mptable/ACPI hooks */
+	return 0;
+}
+
+static int es7000_mps_ret;
+static int es7000_mps_oem_check(struct mpc_table *mpc, char *oem,
+		char *productid)
+{
+	int ret = 0;
+
+	if (mpc->oemptr) {
+		struct mpc_oemtable *oem_table =
+			(struct mpc_oemtable *)mpc->oemptr;
+
+		if (!strncmp(oem, "UNISYS", 6))
+			ret = parse_unisys_oem((char *)oem_table);
+	}
+
+	es7000_mps_ret = ret;
+
+	return ret && !es7000_apic_is_cluster();
+}
+
+static int es7000_mps_oem_check_cluster(struct mpc_table *mpc, char *oem,
+		char *productid)
+{
+	int ret = es7000_mps_ret;
+
+	return ret && es7000_apic_is_cluster();
+}
+
+/* We've been warned by a false positive warning.Use __refdata to keep calm. */
+static struct apic __refdata apic_es7000_cluster = {
+
+	.name				= "es7000",
+	.probe				= probe_es7000,
+	.acpi_madt_oem_check		= es7000_acpi_madt_oem_check_cluster,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= es7000_apic_id_registered,
+
+	.irq_delivery_mode		= dest_LowestPrio,
+	/* logical delivery broadcast to all procs: */
+	.irq_dest_mode			= 1,
+
+	.target_cpus			= target_cpus_cluster,
+	.disable_esr			= 1,
+	.dest_logical			= 0,
+	.check_apicid_used		= es7000_check_apicid_used,
+	.check_apicid_present		= es7000_check_apicid_present,
+
+	.vector_allocation_domain	= flat_vector_allocation_domain,
+	.init_apic_ldr			= es7000_init_apic_ldr_cluster,
+
+	.ioapic_phys_id_map		= es7000_ioapic_phys_id_map,
+	.setup_apic_routing		= es7000_setup_apic_routing,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= es7000_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= es7000_apicid_to_cpu_present,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= es7000_check_phys_apicid_present,
+	.enable_apic_mode		= es7000_enable_apic_mode,
+	.phys_pkg_id			= es7000_phys_pkg_id,
+	.mps_oem_check			= es7000_mps_oem_check_cluster,
+
+	.get_apic_id			= es7000_get_apic_id,
+	.set_apic_id			= NULL,
+	.apic_id_mask			= 0xFF << 24,
+
+	.cpu_mask_to_apicid_and		= es7000_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= es7000_send_IPI_mask,
+	.send_IPI_mask_allbutself	= NULL,
+	.send_IPI_allbutself		= es7000_send_IPI_allbutself,
+	.send_IPI_all			= es7000_send_IPI_all,
+	.send_IPI_self			= default_send_IPI_self,
+
+	.wakeup_secondary_cpu		= wakeup_secondary_cpu_via_mip,
+
+	.trampoline_phys_low		= 0x467,
+	.trampoline_phys_high		= 0x469,
+
+	.wait_for_init_deassert		= NULL,
+
+	/* Nothing to do for most platforms, since cleared by the INIT cycle: */
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= default_inquire_remote_apic,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+
+	.x86_32_early_logical_apicid	= es7000_early_logical_apicid,
+};
+
+static struct apic __refdata apic_es7000 = {
+
+	.name				= "es7000",
+	.probe				= probe_es7000,
+	.acpi_madt_oem_check		= es7000_acpi_madt_oem_check,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= es7000_apic_id_registered,
+
+	.irq_delivery_mode		= dest_Fixed,
+	/* phys delivery to target CPUs: */
+	.irq_dest_mode			= 0,
+
+	.target_cpus			= es7000_target_cpus,
+	.disable_esr			= 1,
+	.dest_logical			= 0,
+	.check_apicid_used		= es7000_check_apicid_used,
+	.check_apicid_present		= es7000_check_apicid_present,
+
+	.vector_allocation_domain	= flat_vector_allocation_domain,
+	.init_apic_ldr			= es7000_init_apic_ldr,
+
+	.ioapic_phys_id_map		= es7000_ioapic_phys_id_map,
+	.setup_apic_routing		= es7000_setup_apic_routing,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= es7000_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= es7000_apicid_to_cpu_present,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= es7000_check_phys_apicid_present,
+	.enable_apic_mode		= es7000_enable_apic_mode,
+	.phys_pkg_id			= es7000_phys_pkg_id,
+	.mps_oem_check			= es7000_mps_oem_check,
+
+	.get_apic_id			= es7000_get_apic_id,
+	.set_apic_id			= NULL,
+	.apic_id_mask			= 0xFF << 24,
+
+	.cpu_mask_to_apicid_and		= es7000_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= es7000_send_IPI_mask,
+	.send_IPI_mask_allbutself	= NULL,
+	.send_IPI_allbutself		= es7000_send_IPI_allbutself,
+	.send_IPI_all			= es7000_send_IPI_all,
+	.send_IPI_self			= default_send_IPI_self,
+
+	.trampoline_phys_low		= 0x467,
+	.trampoline_phys_high		= 0x469,
+
+	.wait_for_init_deassert		= es7000_wait_for_init_deassert,
+
+	/* Nothing to do for most platforms, since cleared by the INIT cycle: */
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= default_inquire_remote_apic,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+
+	.x86_32_early_logical_apicid	= es7000_early_logical_apicid,
+};
+
+/*
+ * Need to check for es7000 followed by es7000_cluster, so this order
+ * in apic_drivers is important.
+ */
+apic_drivers(apic_es7000, apic_es7000_cluster);

arch/x86/kernel/apic/hw_nmi.c

@@ -0,0 +1,91 @@
+/*
+ *  HW NMI watchdog support
+ *
+ *  started by Don Zickus, Copyright (C) 2010 Red Hat, Inc.
+ *
+ *  Arch specific calls to support NMI watchdog
+ *
+ *  Bits copied from original nmi.c file
+ *
+ */
+#include <asm/apic.h>
+#include <asm/nmi.h>
+
+#include <linux/cpumask.h>
+#include <linux/kdebug.h>
+#include <linux/notifier.h>
+#include <linux/kprobes.h>
+#include <linux/nmi.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+
+#ifdef CONFIG_HARDLOCKUP_DETECTOR
+u64 hw_nmi_get_sample_period(int watchdog_thresh)
+{
+	return (u64)(cpu_khz) * 1000 * watchdog_thresh;
+}
+#endif
+
+#ifdef arch_trigger_all_cpu_backtrace
+/* For reliability, we're prepared to waste bits here. */
+static DECLARE_BITMAP(backtrace_mask, NR_CPUS) __read_mostly;
+
+/* "in progress" flag of arch_trigger_all_cpu_backtrace */
+static unsigned long backtrace_flag;
+
+void arch_trigger_all_cpu_backtrace(void)
+{
+	int i;
+
+	if (test_and_set_bit(0, &backtrace_flag))
+		/*
+		 * If there is already a trigger_all_cpu_backtrace() in progress
+		 * (backtrace_flag == 1), don't output double cpu dump infos.
+		 */
+		return;
+
+	cpumask_copy(to_cpumask(backtrace_mask), cpu_online_mask);
+
+	printk(KERN_INFO "sending NMI to all CPUs:\n");
+	apic->send_IPI_all(NMI_VECTOR);
+
+	/* Wait for up to 10 seconds for all CPUs to do the backtrace */
+	for (i = 0; i < 10 * 1000; i++) {
+		if (cpumask_empty(to_cpumask(backtrace_mask)))
+			break;
+		mdelay(1);
+	}
+
+	clear_bit(0, &backtrace_flag);
+	smp_mb__after_clear_bit();
+}
+
+static int __kprobes
+arch_trigger_all_cpu_backtrace_handler(unsigned int cmd, struct pt_regs *regs)
+{
+	int cpu;
+
+	cpu = smp_processor_id();
+
+	if (cpumask_test_cpu(cpu, to_cpumask(backtrace_mask))) {
+		static arch_spinlock_t lock = __ARCH_SPIN_LOCK_UNLOCKED;
+
+		arch_spin_lock(&lock);
+		printk(KERN_WARNING "NMI backtrace for cpu %d\n", cpu);
+		show_regs(regs);
+		arch_spin_unlock(&lock);
+		cpumask_clear_cpu(cpu, to_cpumask(backtrace_mask));
+		return NMI_HANDLED;
+	}
+
+	return NMI_DONE;
+}
+
+static int __init register_trigger_all_cpu_backtrace(void)
+{
+	register_nmi_handler(NMI_LOCAL, arch_trigger_all_cpu_backtrace_handler,
+				0, "arch_bt");
+	return 0;
+}
+early_initcall(register_trigger_all_cpu_backtrace);
+#endif

arch/x86/kernel/apic/ipi.c

@@ -0,0 +1,167 @@
+#include <linux/cpumask.h>
+#include <linux/interrupt.h>
+#include <linux/init.h>
+
+#include <linux/mm.h>
+#include <linux/delay.h>
+#include <linux/spinlock.h>
+#include <linux/kernel_stat.h>
+#include <linux/mc146818rtc.h>
+#include <linux/cache.h>
+#include <linux/cpu.h>
+#include <linux/module.h>
+
+#include <asm/smp.h>
+#include <asm/mtrr.h>
+#include <asm/tlbflush.h>
+#include <asm/mmu_context.h>
+#include <asm/apic.h>
+#include <asm/proto.h>
+#include <asm/ipi.h>
+
+void default_send_IPI_mask_sequence_phys(const struct cpumask *mask, int vector)
+{
+	unsigned long query_cpu;
+	unsigned long flags;
+
+	/*
+	 * Hack. The clustered APIC addressing mode doesn't allow us to send
+	 * to an arbitrary mask, so I do a unicast to each CPU instead.
+	 * - mbligh
+	 */
+	local_irq_save(flags);
+	for_each_cpu(query_cpu, mask) {
+		__default_send_IPI_dest_field(per_cpu(x86_cpu_to_apicid,
+				query_cpu), vector, APIC_DEST_PHYSICAL);
+	}
+	local_irq_restore(flags);
+}
+
+void default_send_IPI_mask_allbutself_phys(const struct cpumask *mask,
+						 int vector)
+{
+	unsigned int this_cpu = smp_processor_id();
+	unsigned int query_cpu;
+	unsigned long flags;
+
+	/* See Hack comment above */
+
+	local_irq_save(flags);
+	for_each_cpu(query_cpu, mask) {
+		if (query_cpu == this_cpu)
+			continue;
+		__default_send_IPI_dest_field(per_cpu(x86_cpu_to_apicid,
+				 query_cpu), vector, APIC_DEST_PHYSICAL);
+	}
+	local_irq_restore(flags);
+}
+
+#ifdef CONFIG_X86_32
+
+void default_send_IPI_mask_sequence_logical(const struct cpumask *mask,
+						 int vector)
+{
+	unsigned long flags;
+	unsigned int query_cpu;
+
+	/*
+	 * Hack. The clustered APIC addressing mode doesn't allow us to send
+	 * to an arbitrary mask, so I do a unicasts to each CPU instead. This
+	 * should be modified to do 1 message per cluster ID - mbligh
+	 */
+
+	local_irq_save(flags);
+	for_each_cpu(query_cpu, mask)
+		__default_send_IPI_dest_field(
+			early_per_cpu(x86_cpu_to_logical_apicid, query_cpu),
+			vector, apic->dest_logical);
+	local_irq_restore(flags);
+}
+
+void default_send_IPI_mask_allbutself_logical(const struct cpumask *mask,
+						 int vector)
+{
+	unsigned long flags;
+	unsigned int query_cpu;
+	unsigned int this_cpu = smp_processor_id();
+
+	/* See Hack comment above */
+
+	local_irq_save(flags);
+	for_each_cpu(query_cpu, mask) {
+		if (query_cpu == this_cpu)
+			continue;
+		__default_send_IPI_dest_field(
+			early_per_cpu(x86_cpu_to_logical_apicid, query_cpu),
+			vector, apic->dest_logical);
+		}
+	local_irq_restore(flags);
+}
+
+/*
+ * This is only used on smaller machines.
+ */
+void default_send_IPI_mask_logical(const struct cpumask *cpumask, int vector)
+{
+	unsigned long mask = cpumask_bits(cpumask)[0];
+	unsigned long flags;
+
+	if (!mask)
+		return;
+
+	local_irq_save(flags);
+	WARN_ON(mask & ~cpumask_bits(cpu_online_mask)[0]);
+	__default_send_IPI_dest_field(mask, vector, apic->dest_logical);
+	local_irq_restore(flags);
+}
+
+void default_send_IPI_allbutself(int vector)
+{
+	/*
+	 * if there are no other CPUs in the system then we get an APIC send
+	 * error if we try to broadcast, thus avoid sending IPIs in this case.
+	 */
+	if (!(num_online_cpus() > 1))
+		return;
+
+	__default_local_send_IPI_allbutself(vector);
+}
+
+void default_send_IPI_all(int vector)
+{
+	__default_local_send_IPI_all(vector);
+}
+
+void default_send_IPI_self(int vector)
+{
+	__default_send_IPI_shortcut(APIC_DEST_SELF, vector, apic->dest_logical);
+}
+
+/* must come after the send_IPI functions above for inlining */
+static int convert_apicid_to_cpu(int apic_id)
+{
+	int i;
+
+	for_each_possible_cpu(i) {
+		if (per_cpu(x86_cpu_to_apicid, i) == apic_id)
+			return i;
+	}
+	return -1;
+}
+
+int safe_smp_processor_id(void)
+{
+	int apicid, cpuid;
+
+	if (!cpu_has_apic)
+		return 0;
+
+	apicid = hard_smp_processor_id();
+	if (apicid == BAD_APICID)
+		return 0;
+
+	cpuid = convert_apicid_to_cpu(apicid);
+
+	return cpuid >= 0 ? cpuid : 0;
+}
+#endif

arch/x86/kernel/apic/numaq_32.c

@@ -0,0 +1,525 @@
+/*
+ * Written by: Patricia Gaughen, IBM Corporation
+ *
+ * Copyright (C) 2002, IBM Corp.
+ * Copyright (C) 2009, Red Hat, Inc., Ingo Molnar
+ *
+ * All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * 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, GOOD TITLE or
+ * NON INFRINGEMENT.  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., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ * Send feedback to <gone@us.ibm.com>
+ */
+#include <linux/nodemask.h>
+#include <linux/topology.h>
+#include <linux/bootmem.h>
+#include <linux/memblock.h>
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/kernel.h>
+#include <linux/mmzone.h>
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/numa.h>
+#include <linux/smp.h>
+#include <linux/io.h>
+#include <linux/mm.h>
+
+#include <asm/processor.h>
+#include <asm/fixmap.h>
+#include <asm/mpspec.h>
+#include <asm/numaq.h>
+#include <asm/setup.h>
+#include <asm/apic.h>
+#include <asm/e820.h>
+#include <asm/ipi.h>
+
+int found_numaq;
+
+/*
+ * Have to match translation table entries to main table entries by counter
+ * hence the mpc_record variable .... can't see a less disgusting way of
+ * doing this ....
+ */
+struct mpc_trans {
+	unsigned char			mpc_type;
+	unsigned char			trans_len;
+	unsigned char			trans_type;
+	unsigned char			trans_quad;
+	unsigned char			trans_global;
+	unsigned char			trans_local;
+	unsigned short			trans_reserved;
+};
+
+static int				mpc_record;
+
+static struct mpc_trans			*translation_table[MAX_MPC_ENTRY];
+
+int					mp_bus_id_to_node[MAX_MP_BUSSES];
+int					mp_bus_id_to_local[MAX_MP_BUSSES];
+int					quad_local_to_mp_bus_id[NR_CPUS/4][4];
+
+
+static inline void numaq_register_node(int node, struct sys_cfg_data *scd)
+{
+	struct eachquadmem *eq = scd->eq + node;
+	u64 start = (u64)(eq->hi_shrd_mem_start - eq->priv_mem_size) << 20;
+	u64 end = (u64)(eq->hi_shrd_mem_start + eq->hi_shrd_mem_size) << 20;
+	int ret;
+
+	node_set(node, numa_nodes_parsed);
+	ret = numa_add_memblk(node, start, end);
+	BUG_ON(ret < 0);
+}
+
+/*
+ * Function: smp_dump_qct()
+ *
+ * Description: gets memory layout from the quad config table.  This
+ * function also updates numa_nodes_parsed with the nodes (quads) present.
+ */
+static void __init smp_dump_qct(void)
+{
+	struct sys_cfg_data *scd;
+	int node;
+
+	scd = (void *)__va(SYS_CFG_DATA_PRIV_ADDR);
+
+	for_each_node(node) {
+		if (scd->quads_present31_0 & (1 << node))
+			numaq_register_node(node, scd);
+	}
+}
+
+void __cpuinit numaq_tsc_disable(void)
+{
+	if (!found_numaq)
+		return;
+
+	if (num_online_nodes() > 1) {
+		printk(KERN_DEBUG "NUMAQ: disabling TSC\n");
+		setup_clear_cpu_cap(X86_FEATURE_TSC);
+	}
+}
+
+static void __init numaq_tsc_init(void)
+{
+	numaq_tsc_disable();
+}
+
+static inline int generate_logical_apicid(int quad, int phys_apicid)
+{
+	return (quad << 4) + (phys_apicid ? phys_apicid << 1 : 1);
+}
+
+/* x86_quirks member */
+static int mpc_apic_id(struct mpc_cpu *m)
+{
+	int quad = translation_table[mpc_record]->trans_quad;
+	int logical_apicid = generate_logical_apicid(quad, m->apicid);
+
+	printk(KERN_DEBUG
+		"Processor #%d %u:%u APIC version %d (quad %d, apic %d)\n",
+		 m->apicid, (m->cpufeature & CPU_FAMILY_MASK) >> 8,
+		(m->cpufeature & CPU_MODEL_MASK) >> 4,
+		 m->apicver, quad, logical_apicid);
+
+	return logical_apicid;
+}
+
+/* x86_quirks member */
+static void mpc_oem_bus_info(struct mpc_bus *m, char *name)
+{
+	int quad = translation_table[mpc_record]->trans_quad;
+	int local = translation_table[mpc_record]->trans_local;
+
+	mp_bus_id_to_node[m->busid] = quad;
+	mp_bus_id_to_local[m->busid] = local;
+
+	printk(KERN_INFO "Bus #%d is %s (node %d)\n", m->busid, name, quad);
+}
+
+/* x86_quirks member */
+static void mpc_oem_pci_bus(struct mpc_bus *m)
+{
+	int quad = translation_table[mpc_record]->trans_quad;
+	int local = translation_table[mpc_record]->trans_local;
+
+	quad_local_to_mp_bus_id[quad][local] = m->busid;
+}
+
+/*
+ * Called from mpparse code.
+ * mode = 0: prescan
+ * mode = 1: one mpc entry scanned
+ */
+static void numaq_mpc_record(unsigned int mode)
+{
+	if (!mode)
+		mpc_record = 0;
+	else
+		mpc_record++;
+}
+
+static void __init MP_translation_info(struct mpc_trans *m)
+{
+	printk(KERN_INFO
+	    "Translation: record %d, type %d, quad %d, global %d, local %d\n",
+	       mpc_record, m->trans_type, m->trans_quad, m->trans_global,
+	       m->trans_local);
+
+	if (mpc_record >= MAX_MPC_ENTRY)
+		printk(KERN_ERR "MAX_MPC_ENTRY exceeded!\n");
+	else
+		translation_table[mpc_record] = m; /* stash this for later */
+
+	if (m->trans_quad < MAX_NUMNODES && !node_online(m->trans_quad))
+		node_set_online(m->trans_quad);
+}
+
+static int __init mpf_checksum(unsigned char *mp, int len)
+{
+	int sum = 0;
+
+	while (len--)
+		sum += *mp++;
+
+	return sum & 0xFF;
+}
+
+/*
+ * Read/parse the MPC oem tables
+ */
+static void __init smp_read_mpc_oem(struct mpc_table *mpc)
+{
+	struct mpc_oemtable *oemtable = (void *)(long)mpc->oemptr;
+	int count = sizeof(*oemtable);	/* the header size */
+	unsigned char *oemptr = ((unsigned char *)oemtable) + count;
+
+	mpc_record = 0;
+	printk(KERN_INFO
+		"Found an OEM MPC table at %8p - parsing it...\n", oemtable);
+
+	if (memcmp(oemtable->signature, MPC_OEM_SIGNATURE, 4)) {
+		printk(KERN_WARNING
+		       "SMP mpc oemtable: bad signature [%c%c%c%c]!\n",
+		       oemtable->signature[0], oemtable->signature[1],
+		       oemtable->signature[2], oemtable->signature[3]);
+		return;
+	}
+
+	if (mpf_checksum((unsigned char *)oemtable, oemtable->length)) {
+		printk(KERN_WARNING "SMP oem mptable: checksum error!\n");
+		return;
+	}
+
+	while (count < oemtable->length) {
+		switch (*oemptr) {
+		case MP_TRANSLATION:
+			{
+				struct mpc_trans *m = (void *)oemptr;
+
+				MP_translation_info(m);
+				oemptr += sizeof(*m);
+				count += sizeof(*m);
+				++mpc_record;
+				break;
+			}
+		default:
+			printk(KERN_WARNING
+			       "Unrecognised OEM table entry type! - %d\n",
+			       (int)*oemptr);
+			return;
+		}
+	}
+}
+
+static __init void early_check_numaq(void)
+{
+	/*
+	 * get boot-time SMP configuration:
+	 */
+	if (smp_found_config)
+		early_get_smp_config();
+
+	if (found_numaq) {
+		x86_init.mpparse.mpc_record = numaq_mpc_record;
+		x86_init.mpparse.setup_ioapic_ids = x86_init_noop;
+		x86_init.mpparse.mpc_apic_id = mpc_apic_id;
+		x86_init.mpparse.smp_read_mpc_oem = smp_read_mpc_oem;
+		x86_init.mpparse.mpc_oem_pci_bus = mpc_oem_pci_bus;
+		x86_init.mpparse.mpc_oem_bus_info = mpc_oem_bus_info;
+		x86_init.timers.tsc_pre_init = numaq_tsc_init;
+		x86_init.pci.init = pci_numaq_init;
+	}
+}
+
+int __init numaq_numa_init(void)
+{
+	early_check_numaq();
+	if (!found_numaq)
+		return -ENOENT;
+	smp_dump_qct();
+
+	return 0;
+}
+
+#define NUMAQ_APIC_DFR_VALUE	(APIC_DFR_CLUSTER)
+
+static inline unsigned int numaq_get_apic_id(unsigned long x)
+{
+	return (x >> 24) & 0x0F;
+}
+
+static inline void numaq_send_IPI_mask(const struct cpumask *mask, int vector)
+{
+	default_send_IPI_mask_sequence_logical(mask, vector);
+}
+
+static inline void numaq_send_IPI_allbutself(int vector)
+{
+	default_send_IPI_mask_allbutself_logical(cpu_online_mask, vector);
+}
+
+static inline void numaq_send_IPI_all(int vector)
+{
+	numaq_send_IPI_mask(cpu_online_mask, vector);
+}
+
+#define NUMAQ_TRAMPOLINE_PHYS_LOW	(0x8)
+#define NUMAQ_TRAMPOLINE_PHYS_HIGH	(0xa)
+
+/*
+ * Because we use NMIs rather than the INIT-STARTUP sequence to
+ * bootstrap the CPUs, the APIC may be in a weird state. Kick it:
+ */
+static inline void numaq_smp_callin_clear_local_apic(void)
+{
+	clear_local_APIC();
+}
+
+static inline const struct cpumask *numaq_target_cpus(void)
+{
+	return cpu_all_mask;
+}
+
+static unsigned long numaq_check_apicid_used(physid_mask_t *map, int apicid)
+{
+	return physid_isset(apicid, *map);
+}
+
+static inline unsigned long numaq_check_apicid_present(int bit)
+{
+	return physid_isset(bit, phys_cpu_present_map);
+}
+
+static inline int numaq_apic_id_registered(void)
+{
+	return 1;
+}
+
+static inline void numaq_init_apic_ldr(void)
+{
+	/* Already done in NUMA-Q firmware */
+}
+
+static inline void numaq_setup_apic_routing(void)
+{
+	printk(KERN_INFO
+		"Enabling APIC mode:  NUMA-Q.  Using %d I/O APICs\n",
+		nr_ioapics);
+}
+
+/*
+ * Skip adding the timer int on secondary nodes, which causes
+ * a small but painful rift in the time-space continuum.
+ */
+static inline int numaq_multi_timer_check(int apic, int irq)
+{
+	return apic != 0 && irq == 0;
+}
+
+static inline void numaq_ioapic_phys_id_map(physid_mask_t *phys_map, physid_mask_t *retmap)
+{
+	/* We don't have a good way to do this yet - hack */
+	return physids_promote(0xFUL, retmap);
+}
+
+/*
+ * Supporting over 60 cpus on NUMA-Q requires a locality-dependent
+ * cpu to APIC ID relation to properly interact with the intelligent
+ * mode of the cluster controller.
+ */
+static inline int numaq_cpu_present_to_apicid(int mps_cpu)
+{
+	if (mps_cpu < 60)
+		return ((mps_cpu >> 2) << 4) | (1 << (mps_cpu & 0x3));
+	else
+		return BAD_APICID;
+}
+
+static inline int numaq_apicid_to_node(int logical_apicid)
+{
+	return logical_apicid >> 4;
+}
+
+static int numaq_numa_cpu_node(int cpu)
+{
+	int logical_apicid = early_per_cpu(x86_cpu_to_logical_apicid, cpu);
+
+	if (logical_apicid != BAD_APICID)
+		return numaq_apicid_to_node(logical_apicid);
+	return NUMA_NO_NODE;
+}
+
+static void numaq_apicid_to_cpu_present(int logical_apicid, physid_mask_t *retmap)
+{
+	int node = numaq_apicid_to_node(logical_apicid);
+	int cpu = __ffs(logical_apicid & 0xf);
+
+	physid_set_mask_of_physid(cpu + 4*node, retmap);
+}
+
+/* Where the IO area was mapped on multiquad, always 0 otherwise */
+void *xquad_portio;
+
+static inline int numaq_check_phys_apicid_present(int phys_apicid)
+{
+	return 1;
+}
+
+/*
+ * We use physical apicids here, not logical, so just return the default
+ * physical broadcast to stop people from breaking us
+ */
+static int
+numaq_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
+			     const struct cpumask *andmask,
+			     unsigned int *apicid)
+{
+	*apicid = 0x0F;
+	return 0;
+}
+
+/* No NUMA-Q box has a HT CPU, but it can't hurt to use the default code. */
+static inline int numaq_phys_pkg_id(int cpuid_apic, int index_msb)
+{
+	return cpuid_apic >> index_msb;
+}
+
+static int
+numaq_mps_oem_check(struct mpc_table *mpc, char *oem, char *productid)
+{
+	if (strncmp(oem, "IBM NUMA", 8))
+		printk(KERN_ERR "Warning! Not a NUMA-Q system!\n");
+	else
+		found_numaq = 1;
+
+	return found_numaq;
+}
+
+static int probe_numaq(void)
+{
+	/* already know from get_memcfg_numaq() */
+	return found_numaq;
+}
+
+static void numaq_setup_portio_remap(void)
+{
+	int num_quads = num_online_nodes();
+
+	if (num_quads <= 1)
+		return;
+
+	printk(KERN_INFO
+		"Remapping cross-quad port I/O for %d quads\n", num_quads);
+
+	xquad_portio = ioremap(XQUAD_PORTIO_BASE, num_quads*XQUAD_PORTIO_QUAD);
+
+	printk(KERN_INFO
+		"xquad_portio vaddr 0x%08lx, len %08lx\n",
+		(u_long) xquad_portio, (u_long) num_quads*XQUAD_PORTIO_QUAD);
+}
+
+/* Use __refdata to keep false positive warning calm.  */
+static struct apic __refdata apic_numaq = {
+
+	.name				= "NUMAQ",
+	.probe				= probe_numaq,
+	.acpi_madt_oem_check		= NULL,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= numaq_apic_id_registered,
+
+	.irq_delivery_mode		= dest_LowestPrio,
+	/* physical delivery on LOCAL quad: */
+	.irq_dest_mode			= 0,
+
+	.target_cpus			= numaq_target_cpus,
+	.disable_esr			= 1,
+	.dest_logical			= APIC_DEST_LOGICAL,
+	.check_apicid_used		= numaq_check_apicid_used,
+	.check_apicid_present		= numaq_check_apicid_present,
+
+	.vector_allocation_domain	= flat_vector_allocation_domain,
+	.init_apic_ldr			= numaq_init_apic_ldr,
+
+	.ioapic_phys_id_map		= numaq_ioapic_phys_id_map,
+	.setup_apic_routing		= numaq_setup_apic_routing,
+	.multi_timer_check		= numaq_multi_timer_check,
+	.cpu_present_to_apicid		= numaq_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= numaq_apicid_to_cpu_present,
+	.setup_portio_remap		= numaq_setup_portio_remap,
+	.check_phys_apicid_present	= numaq_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= numaq_phys_pkg_id,
+	.mps_oem_check			= numaq_mps_oem_check,
+
+	.get_apic_id			= numaq_get_apic_id,
+	.set_apic_id			= NULL,
+	.apic_id_mask			= 0x0F << 24,
+
+	.cpu_mask_to_apicid_and		= numaq_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= numaq_send_IPI_mask,
+	.send_IPI_mask_allbutself	= NULL,
+	.send_IPI_allbutself		= numaq_send_IPI_allbutself,
+	.send_IPI_all			= numaq_send_IPI_all,
+	.send_IPI_self			= default_send_IPI_self,
+
+	.wakeup_secondary_cpu		= wakeup_secondary_cpu_via_nmi,
+	.trampoline_phys_low		= NUMAQ_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= NUMAQ_TRAMPOLINE_PHYS_HIGH,
+
+	/* We don't do anything here because we use NMI's to boot instead */
+	.wait_for_init_deassert		= NULL,
+
+	.smp_callin_clear_local_apic	= numaq_smp_callin_clear_local_apic,
+	.inquire_remote_apic		= NULL,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+
+	.x86_32_early_logical_apicid	= noop_x86_32_early_logical_apicid,
+	.x86_32_numa_cpu_node		= numaq_numa_cpu_node,
+};
+
+apic_driver(apic_numaq);

arch/x86/kernel/apic/probe_32.c

@@ -0,0 +1,259 @@
+/*
+ * Default generic APIC driver. This handles up to 8 CPUs.
+ *
+ * Copyright 2003 Andi Kleen, SuSE Labs.
+ * Subject to the GNU Public License, v.2
+ *
+ * Generic x86 APIC driver probe layer.
+ */
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <asm/fixmap.h>
+#include <asm/mpspec.h>
+#include <asm/apicdef.h>
+#include <asm/apic.h>
+#include <asm/setup.h>
+
+#include <linux/smp.h>
+#include <asm/ipi.h>
+
+#include <linux/interrupt.h>
+#include <asm/acpi.h>
+#include <asm/e820.h>
+
+#ifdef CONFIG_HOTPLUG_CPU
+#define DEFAULT_SEND_IPI	(1)
+#else
+#define DEFAULT_SEND_IPI	(0)
+#endif
+
+int no_broadcast = DEFAULT_SEND_IPI;
+
+static __init int no_ipi_broadcast(char *str)
+{
+	get_option(&str, &no_broadcast);
+	pr_info("Using %s mode\n",
+		no_broadcast ? "No IPI Broadcast" : "IPI Broadcast");
+	return 1;
+}
+__setup("no_ipi_broadcast=", no_ipi_broadcast);
+
+static int __init print_ipi_mode(void)
+{
+	pr_info("Using IPI %s mode\n",
+		no_broadcast ? "No-Shortcut" : "Shortcut");
+	return 0;
+}
+late_initcall(print_ipi_mode);
+
+static int default_x86_32_early_logical_apicid(int cpu)
+{
+	return 1 << cpu;
+}
+
+static void setup_apic_flat_routing(void)
+{
+#ifdef CONFIG_X86_IO_APIC
+	printk(KERN_INFO
+		"Enabling APIC mode:  Flat.  Using %d I/O APICs\n",
+		nr_ioapics);
+#endif
+}
+
+/* should be called last. */
+static int probe_default(void)
+{
+	return 1;
+}
+
+static struct apic apic_default = {
+
+	.name				= "default",
+	.probe				= probe_default,
+	.acpi_madt_oem_check		= NULL,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= default_apic_id_registered,
+
+	.irq_delivery_mode		= dest_LowestPrio,
+	/* logical delivery broadcast to all CPUs: */
+	.irq_dest_mode			= 1,
+
+	.target_cpus			= default_target_cpus,
+	.disable_esr			= 0,
+	.dest_logical			= APIC_DEST_LOGICAL,
+	.check_apicid_used		= default_check_apicid_used,
+	.check_apicid_present		= default_check_apicid_present,
+
+	.vector_allocation_domain	= flat_vector_allocation_domain,
+	.init_apic_ldr			= default_init_apic_ldr,
+
+	.ioapic_phys_id_map		= default_ioapic_phys_id_map,
+	.setup_apic_routing		= setup_apic_flat_routing,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= default_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= default_phys_pkg_id,
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= default_get_apic_id,
+	.set_apic_id			= NULL,
+	.apic_id_mask			= 0x0F << 24,
+
+	.cpu_mask_to_apicid_and		= flat_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= default_send_IPI_mask_logical,
+	.send_IPI_mask_allbutself	= default_send_IPI_mask_allbutself_logical,
+	.send_IPI_allbutself		= default_send_IPI_allbutself,
+	.send_IPI_all			= default_send_IPI_all,
+	.send_IPI_self			= default_send_IPI_self,
+
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+
+	.wait_for_init_deassert		= default_wait_for_init_deassert,
+
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= default_inquire_remote_apic,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+
+	.x86_32_early_logical_apicid	= default_x86_32_early_logical_apicid,
+};
+
+apic_driver(apic_default);
+
+struct apic *apic = &apic_default;
+EXPORT_SYMBOL_GPL(apic);
+
+static int cmdline_apic __initdata;
+static int __init parse_apic(char *arg)
+{
+	struct apic **drv;
+
+	if (!arg)
+		return -EINVAL;
+
+	for (drv = __apicdrivers; drv < __apicdrivers_end; drv++) {
+		if (!strcmp((*drv)->name, arg)) {
+			apic = *drv;
+			cmdline_apic = 1;
+			return 0;
+		}
+	}
+
+	/* Parsed again by __setup for debug/verbose */
+	return 0;
+}
+early_param("apic", parse_apic);
+
+void __init default_setup_apic_routing(void)
+{
+	int version = apic_version[boot_cpu_physical_apicid];
+
+	if (num_possible_cpus() > 8) {
+		switch (boot_cpu_data.x86_vendor) {
+		case X86_VENDOR_INTEL:
+			if (!APIC_XAPIC(version)) {
+				def_to_bigsmp = 0;
+				break;
+			}
+			/* If P4 and above fall through */
+		case X86_VENDOR_AMD:
+			def_to_bigsmp = 1;
+		}
+	}
+
+#ifdef CONFIG_X86_BIGSMP
+	/*
+	 * This is used to switch to bigsmp mode when
+	 * - There is no apic= option specified by the user
+	 * - generic_apic_probe() has chosen apic_default as the sub_arch
+	 * - we find more than 8 CPUs in acpi LAPIC listing with xAPIC support
+	 */
+
+	if (!cmdline_apic && apic == &apic_default)
+		generic_bigsmp_probe();
+#endif
+
+	if (apic->setup_apic_routing)
+		apic->setup_apic_routing();
+
+	if (x86_platform.apic_post_init)
+		x86_platform.apic_post_init();
+}
+
+void __init generic_apic_probe(void)
+{
+	if (!cmdline_apic) {
+		struct apic **drv;
+
+		for (drv = __apicdrivers; drv < __apicdrivers_end; drv++) {
+			if ((*drv)->probe()) {
+				apic = *drv;
+				break;
+			}
+		}
+		/* Not visible without early console */
+		if (drv == __apicdrivers_end)
+			panic("Didn't find an APIC driver");
+	}
+	printk(KERN_INFO "Using APIC driver %s\n", apic->name);
+}
+
+/* These functions can switch the APIC even after the initial ->probe() */
+
+int __init
+generic_mps_oem_check(struct mpc_table *mpc, char *oem, char *productid)
+{
+	struct apic **drv;
+
+	for (drv = __apicdrivers; drv < __apicdrivers_end; drv++) {
+		if (!((*drv)->mps_oem_check))
+			continue;
+		if (!(*drv)->mps_oem_check(mpc, oem, productid))
+			continue;
+
+		if (!cmdline_apic) {
+			apic = *drv;
+			printk(KERN_INFO "Switched to APIC driver `%s'.\n",
+			       apic->name);
+		}
+		return 1;
+	}
+	return 0;
+}
+
+int __init default_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	struct apic **drv;
+
+	for (drv = __apicdrivers; drv < __apicdrivers_end; drv++) {
+		if (!(*drv)->acpi_madt_oem_check)
+			continue;
+		if (!(*drv)->acpi_madt_oem_check(oem_id, oem_table_id))
+			continue;
+
+		if (!cmdline_apic) {
+			apic = *drv;
+			printk(KERN_INFO "Switched to APIC driver `%s'.\n",
+			       apic->name);
+		}
+		return 1;
+	}
+	return 0;
+}

arch/x86/kernel/apic/probe_64.c

@@ -0,0 +1,72 @@
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Generic APIC sub-arch probe layer.
+ *
+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
+ * James Cleverdon.
+ */
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/hardirq.h>
+#include <linux/dmar.h>
+
+#include <asm/smp.h>
+#include <asm/apic.h>
+#include <asm/ipi.h>
+#include <asm/setup.h>
+
+/*
+ * Check the APIC IDs in bios_cpu_apicid and choose the APIC mode.
+ */
+void __init default_setup_apic_routing(void)
+{
+	struct apic **drv;
+
+	enable_IR_x2apic();
+
+	for (drv = __apicdrivers; drv < __apicdrivers_end; drv++) {
+		if ((*drv)->probe && (*drv)->probe()) {
+			if (apic != *drv) {
+				apic = *drv;
+				pr_info("Switched APIC routing to %s.\n",
+					apic->name);
+			}
+			break;
+		}
+	}
+
+	if (x86_platform.apic_post_init)
+		x86_platform.apic_post_init();
+}
+
+/* Same for both flat and physical. */
+
+void apic_send_IPI_self(int vector)
+{
+	__default_send_IPI_shortcut(APIC_DEST_SELF, vector, APIC_DEST_PHYSICAL);
+}
+
+int __init default_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	struct apic **drv;
+
+	for (drv = __apicdrivers; drv < __apicdrivers_end; drv++) {
+		if ((*drv)->acpi_madt_oem_check(oem_id, oem_table_id)) {
+			if (apic != *drv) {
+				apic = *drv;
+				pr_info("Setting APIC routing to %s.\n",
+					apic->name);
+			}
+			return 1;
+		}
+	}
+	return 0;
+}

arch/x86/kernel/apic/summit_32.c

@@ -0,0 +1,552 @@
+/*
+ * IBM Summit-Specific Code
+ *
+ * Written By: Matthew Dobson, IBM Corporation
+ *
+ * Copyright (c) 2003 IBM Corp.
+ *
+ * All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or (at
+ * your option) any later version.
+ *
+ * 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, GOOD TITLE or
+ * NON INFRINGEMENT.  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., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ * Send feedback to <colpatch@us.ibm.com>
+ *
+ */
+
+#define pr_fmt(fmt) "summit: %s: " fmt, __func__
+
+#include <linux/mm.h>
+#include <linux/init.h>
+#include <asm/io.h>
+#include <asm/bios_ebda.h>
+
+/*
+ * APIC driver for the IBM "Summit" chipset.
+ */
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <asm/mpspec.h>
+#include <asm/apic.h>
+#include <asm/smp.h>
+#include <asm/fixmap.h>
+#include <asm/apicdef.h>
+#include <asm/ipi.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/gfp.h>
+#include <linux/smp.h>
+
+static unsigned summit_get_apic_id(unsigned long x)
+{
+	return (x >> 24) & 0xFF;
+}
+
+static inline void summit_send_IPI_mask(const struct cpumask *mask, int vector)
+{
+	default_send_IPI_mask_sequence_logical(mask, vector);
+}
+
+static void summit_send_IPI_allbutself(int vector)
+{
+	default_send_IPI_mask_allbutself_logical(cpu_online_mask, vector);
+}
+
+static void summit_send_IPI_all(int vector)
+{
+	summit_send_IPI_mask(cpu_online_mask, vector);
+}
+
+#include <asm/tsc.h>
+
+extern int use_cyclone;
+
+#ifdef CONFIG_X86_SUMMIT_NUMA
+static void setup_summit(void);
+#else
+static inline void setup_summit(void) {}
+#endif
+
+static int summit_mps_oem_check(struct mpc_table *mpc, char *oem,
+		char *productid)
+{
+	if (!strncmp(oem, "IBM ENSW", 8) &&
+			(!strncmp(productid, "VIGIL SMP", 9)
+			 || !strncmp(productid, "EXA", 3)
+			 || !strncmp(productid, "RUTHLESS SMP", 12))){
+		mark_tsc_unstable("Summit based system");
+		use_cyclone = 1; /*enable cyclone-timer*/
+		setup_summit();
+		return 1;
+	}
+	return 0;
+}
+
+/* Hook from generic ACPI tables.c */
+static int summit_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	if (!strncmp(oem_id, "IBM", 3) &&
+	    (!strncmp(oem_table_id, "SERVIGIL", 8)
+	     || !strncmp(oem_table_id, "EXA", 3))){
+		mark_tsc_unstable("Summit based system");
+		use_cyclone = 1; /*enable cyclone-timer*/
+		setup_summit();
+		return 1;
+	}
+	return 0;
+}
+
+struct rio_table_hdr {
+	unsigned char version;      /* Version number of this data structure           */
+	                            /* Version 3 adds chassis_num & WP_index           */
+	unsigned char num_scal_dev; /* # of Scalability devices (Twisters for Vigil)   */
+	unsigned char num_rio_dev;  /* # of RIO I/O devices (Cyclones and Winnipegs)   */
+} __attribute__((packed));
+
+struct scal_detail {
+	unsigned char node_id;      /* Scalability Node ID                             */
+	unsigned long CBAR;         /* Address of 1MB register space                   */
+	unsigned char port0node;    /* Node ID port connected to: 0xFF=None            */
+	unsigned char port0port;    /* Port num port connected to: 0,1,2, or 0xFF=None */
+	unsigned char port1node;    /* Node ID port connected to: 0xFF = None          */
+	unsigned char port1port;    /* Port num port connected to: 0,1,2, or 0xFF=None */
+	unsigned char port2node;    /* Node ID port connected to: 0xFF = None          */
+	unsigned char port2port;    /* Port num port connected to: 0,1,2, or 0xFF=None */
+	unsigned char chassis_num;  /* 1 based Chassis number (1 = boot node)          */
+} __attribute__((packed));
+
+struct rio_detail {
+	unsigned char node_id;      /* RIO Node ID                                     */
+	unsigned long BBAR;         /* Address of 1MB register space                   */
+	unsigned char type;         /* Type of device                                  */
+	unsigned char owner_id;     /* For WPEG: Node ID of Cyclone that owns this WPEG*/
+	                            /* For CYC:  Node ID of Twister that owns this CYC */
+	unsigned char port0node;    /* Node ID port connected to: 0xFF=None            */
+	unsigned char port0port;    /* Port num port connected to: 0,1,2, or 0xFF=None */
+	unsigned char port1node;    /* Node ID port connected to: 0xFF=None            */
+	unsigned char port1port;    /* Port num port connected to: 0,1,2, or 0xFF=None */
+	unsigned char first_slot;   /* For WPEG: Lowest slot number below this WPEG    */
+	                            /* For CYC:  0                                     */
+	unsigned char status;       /* For WPEG: Bit 0 = 1 : the XAPIC is used         */
+	                            /*                 = 0 : the XAPIC is not used, ie:*/
+	                            /*                     ints fwded to another XAPIC */
+	                            /*           Bits1:7 Reserved                      */
+	                            /* For CYC:  Bits0:7 Reserved                      */
+	unsigned char WP_index;     /* For WPEG: WPEG instance index - lower ones have */
+	                            /*           lower slot numbers/PCI bus numbers    */
+	                            /* For CYC:  No meaning                            */
+	unsigned char chassis_num;  /* 1 based Chassis number                          */
+	                            /* For LookOut WPEGs this field indicates the      */
+	                            /* Expansion Chassis #, enumerated from Boot       */
+	                            /* Node WPEG external port, then Boot Node CYC     */
+	                            /* external port, then Next Vigil chassis WPEG     */
+	                            /* external port, etc.                             */
+	                            /* Shared Lookouts have only 1 chassis number (the */
+	                            /* first one assigned)                             */
+} __attribute__((packed));
+
+
+typedef enum {
+	CompatTwister = 0,  /* Compatibility Twister               */
+	AltTwister    = 1,  /* Alternate Twister of internal 8-way */
+	CompatCyclone = 2,  /* Compatibility Cyclone               */
+	AltCyclone    = 3,  /* Alternate Cyclone of internal 8-way */
+	CompatWPEG    = 4,  /* Compatibility WPEG                  */
+	AltWPEG       = 5,  /* Second Planar WPEG                  */
+	LookOutAWPEG  = 6,  /* LookOut WPEG                        */
+	LookOutBWPEG  = 7,  /* LookOut WPEG                        */
+} node_type;
+
+static inline int is_WPEG(struct rio_detail *rio){
+	return (rio->type == CompatWPEG || rio->type == AltWPEG ||
+		rio->type == LookOutAWPEG || rio->type == LookOutBWPEG);
+}
+
+#define SUMMIT_APIC_DFR_VALUE	(APIC_DFR_CLUSTER)
+
+static const struct cpumask *summit_target_cpus(void)
+{
+	/* CPU_MASK_ALL (0xff) has undefined behaviour with
+	 * dest_LowestPrio mode logical clustered apic interrupt routing
+	 * Just start on cpu 0.  IRQ balancing will spread load
+	 */
+	return cpumask_of(0);
+}
+
+static unsigned long summit_check_apicid_used(physid_mask_t *map, int apicid)
+{
+	return 0;
+}
+
+/* we don't use the phys_cpu_present_map to indicate apicid presence */
+static unsigned long summit_check_apicid_present(int bit)
+{
+	return 1;
+}
+
+static int summit_early_logical_apicid(int cpu)
+{
+	int count = 0;
+	u8 my_id = early_per_cpu(x86_cpu_to_apicid, cpu);
+	u8 my_cluster = APIC_CLUSTER(my_id);
+#ifdef CONFIG_SMP
+	u8 lid;
+	int i;
+
+	/* Create logical APIC IDs by counting CPUs already in cluster. */
+	for (count = 0, i = nr_cpu_ids; --i >= 0; ) {
+		lid = early_per_cpu(x86_cpu_to_logical_apicid, i);
+		if (lid != BAD_APICID && APIC_CLUSTER(lid) == my_cluster)
+			++count;
+	}
+#endif
+	/* We only have a 4 wide bitmap in cluster mode.  If a deranged
+	 * BIOS puts 5 CPUs in one APIC cluster, we're hosed. */
+	BUG_ON(count >= XAPIC_DEST_CPUS_SHIFT);
+	return my_cluster | (1UL << count);
+}
+
+static void summit_init_apic_ldr(void)
+{
+	int cpu = smp_processor_id();
+	unsigned long id = early_per_cpu(x86_cpu_to_logical_apicid, cpu);
+	unsigned long val;
+
+	apic_write(APIC_DFR, SUMMIT_APIC_DFR_VALUE);
+	val = apic_read(APIC_LDR) & ~APIC_LDR_MASK;
+	val |= SET_APIC_LOGICAL_ID(id);
+	apic_write(APIC_LDR, val);
+}
+
+static int summit_apic_id_registered(void)
+{
+	return 1;
+}
+
+static void summit_setup_apic_routing(void)
+{
+	pr_info("Enabling APIC mode:  Summit.  Using %d I/O APICs\n",
+		nr_ioapics);
+}
+
+static int summit_cpu_present_to_apicid(int mps_cpu)
+{
+	if (mps_cpu < nr_cpu_ids)
+		return (int)per_cpu(x86_bios_cpu_apicid, mps_cpu);
+	else
+		return BAD_APICID;
+}
+
+static void summit_ioapic_phys_id_map(physid_mask_t *phys_id_map, physid_mask_t *retmap)
+{
+	/* For clustered we don't have a good way to do this yet - hack */
+	physids_promote(0x0FL, retmap);
+}
+
+static void summit_apicid_to_cpu_present(int apicid, physid_mask_t *retmap)
+{
+	physid_set_mask_of_physid(0, retmap);
+}
+
+static int summit_check_phys_apicid_present(int physical_apicid)
+{
+	return 1;
+}
+
+static inline int
+summit_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
+{
+	unsigned int round = 0;
+	unsigned int cpu, apicid = 0;
+
+	/*
+	 * The cpus in the mask must all be on the apic cluster.
+	 */
+	for_each_cpu_and(cpu, cpumask, cpu_online_mask) {
+		int new_apicid = early_per_cpu(x86_cpu_to_logical_apicid, cpu);
+
+		if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
+			pr_err("Not a valid mask!\n");
+			return -EINVAL;
+		}
+		apicid |= new_apicid;
+		round++;
+	}
+	if (!round)
+		return -EINVAL;
+	*dest_id = apicid;
+	return 0;
+}
+
+static int
+summit_cpu_mask_to_apicid_and(const struct cpumask *inmask,
+			      const struct cpumask *andmask,
+			      unsigned int *apicid)
+{
+	cpumask_var_t cpumask;
+	*apicid = early_per_cpu(x86_cpu_to_logical_apicid, 0);
+
+	if (!alloc_cpumask_var(&cpumask, GFP_ATOMIC))
+		return 0;
+
+	cpumask_and(cpumask, inmask, andmask);
+	summit_cpu_mask_to_apicid(cpumask, apicid);
+
+	free_cpumask_var(cpumask);
+
+	return 0;
+}
+
+/*
+ * cpuid returns the value latched in the HW at reset, not the APIC ID
+ * register's value.  For any box whose BIOS changes APIC IDs, like
+ * clustered APIC systems, we must use hard_smp_processor_id.
+ *
+ * See Intel's IA-32 SW Dev's Manual Vol2 under CPUID.
+ */
+static int summit_phys_pkg_id(int cpuid_apic, int index_msb)
+{
+	return hard_smp_processor_id() >> index_msb;
+}
+
+static int probe_summit(void)
+{
+	/* probed later in mptable/ACPI hooks */
+	return 0;
+}
+
+#ifdef CONFIG_X86_SUMMIT_NUMA
+static struct rio_table_hdr *rio_table_hdr;
+static struct scal_detail   *scal_devs[MAX_NUMNODES];
+static struct rio_detail    *rio_devs[MAX_NUMNODES*4];
+
+#ifndef CONFIG_X86_NUMAQ
+static int mp_bus_id_to_node[MAX_MP_BUSSES];
+#endif
+
+static int setup_pci_node_map_for_wpeg(int wpeg_num, int last_bus)
+{
+	int twister = 0, node = 0;
+	int i, bus, num_buses;
+
+	for (i = 0; i < rio_table_hdr->num_rio_dev; i++) {
+		if (rio_devs[i]->node_id == rio_devs[wpeg_num]->owner_id) {
+			twister = rio_devs[i]->owner_id;
+			break;
+		}
+	}
+	if (i == rio_table_hdr->num_rio_dev) {
+		pr_err("Couldn't find owner Cyclone for Winnipeg!\n");
+		return last_bus;
+	}
+
+	for (i = 0; i < rio_table_hdr->num_scal_dev; i++) {
+		if (scal_devs[i]->node_id == twister) {
+			node = scal_devs[i]->node_id;
+			break;
+		}
+	}
+	if (i == rio_table_hdr->num_scal_dev) {
+		pr_err("Couldn't find owner Twister for Cyclone!\n");
+		return last_bus;
+	}
+
+	switch (rio_devs[wpeg_num]->type) {
+	case CompatWPEG:
+		/*
+		 * The Compatibility Winnipeg controls the 2 legacy buses,
+		 * the 66MHz PCI bus [2 slots] and the 2 "extra" buses in case
+		 * a PCI-PCI bridge card is used in either slot: total 5 buses.
+		 */
+		num_buses = 5;
+		break;
+	case AltWPEG:
+		/*
+		 * The Alternate Winnipeg controls the 2 133MHz buses [1 slot
+		 * each], their 2 "extra" buses, the 100MHz bus [2 slots] and
+		 * the "extra" buses for each of those slots: total 7 buses.
+		 */
+		num_buses = 7;
+		break;
+	case LookOutAWPEG:
+	case LookOutBWPEG:
+		/*
+		 * A Lookout Winnipeg controls 3 100MHz buses [2 slots each]
+		 * & the "extra" buses for each of those slots: total 9 buses.
+		 */
+		num_buses = 9;
+		break;
+	default:
+		pr_info("Unsupported Winnipeg type!\n");
+		return last_bus;
+	}
+
+	for (bus = last_bus; bus < last_bus + num_buses; bus++)
+		mp_bus_id_to_node[bus] = node;
+	return bus;
+}
+
+static int build_detail_arrays(void)
+{
+	unsigned long ptr;
+	int i, scal_detail_size, rio_detail_size;
+
+	if (rio_table_hdr->num_scal_dev > MAX_NUMNODES) {
+		pr_warn("MAX_NUMNODES too low!  Defined as %d, but system has %d nodes\n",
+			MAX_NUMNODES, rio_table_hdr->num_scal_dev);
+		return 0;
+	}
+
+	switch (rio_table_hdr->version) {
+	default:
+		pr_warn("Invalid Rio Grande Table Version: %d\n",
+			rio_table_hdr->version);
+		return 0;
+	case 2:
+		scal_detail_size = 11;
+		rio_detail_size = 13;
+		break;
+	case 3:
+		scal_detail_size = 12;
+		rio_detail_size = 15;
+		break;
+	}
+
+	ptr = (unsigned long)rio_table_hdr + 3;
+	for (i = 0; i < rio_table_hdr->num_scal_dev; i++, ptr += scal_detail_size)
+		scal_devs[i] = (struct scal_detail *)ptr;
+
+	for (i = 0; i < rio_table_hdr->num_rio_dev; i++, ptr += rio_detail_size)
+		rio_devs[i] = (struct rio_detail *)ptr;
+
+	return 1;
+}
+
+void setup_summit(void)
+{
+	unsigned long		ptr;
+	unsigned short		offset;
+	int			i, next_wpeg, next_bus = 0;
+
+	/* The pointer to the EBDA is stored in the word @ phys 0x40E(40:0E) */
+	ptr = get_bios_ebda();
+	ptr = (unsigned long)phys_to_virt(ptr);
+
+	rio_table_hdr = NULL;
+	offset = 0x180;
+	while (offset) {
+		/* The block id is stored in the 2nd word */
+		if (*((unsigned short *)(ptr + offset + 2)) == 0x4752) {
+			/* set the pointer past the offset & block id */
+			rio_table_hdr = (struct rio_table_hdr *)(ptr + offset + 4);
+			break;
+		}
+		/* The next offset is stored in the 1st word.  0 means no more */
+		offset = *((unsigned short *)(ptr + offset));
+	}
+	if (!rio_table_hdr) {
+		pr_err("Unable to locate Rio Grande Table in EBDA - bailing!\n");
+		return;
+	}
+
+	if (!build_detail_arrays())
+		return;
+
+	/* The first Winnipeg we're looking for has an index of 0 */
+	next_wpeg = 0;
+	do {
+		for (i = 0; i < rio_table_hdr->num_rio_dev; i++) {
+			if (is_WPEG(rio_devs[i]) && rio_devs[i]->WP_index == next_wpeg) {
+				/* It's the Winnipeg we're looking for! */
+				next_bus = setup_pci_node_map_for_wpeg(i, next_bus);
+				next_wpeg++;
+				break;
+			}
+		}
+		/*
+		 * If we go through all Rio devices and don't find one with
+		 * the next index, it means we've found all the Winnipegs,
+		 * and thus all the PCI buses.
+		 */
+		if (i == rio_table_hdr->num_rio_dev)
+			next_wpeg = 0;
+	} while (next_wpeg != 0);
+}
+#endif
+
+static struct apic apic_summit = {
+
+	.name				= "summit",
+	.probe				= probe_summit,
+	.acpi_madt_oem_check		= summit_acpi_madt_oem_check,
+	.apic_id_valid			= default_apic_id_valid,
+	.apic_id_registered		= summit_apic_id_registered,
+
+	.irq_delivery_mode		= dest_LowestPrio,
+	/* logical delivery broadcast to all CPUs: */
+	.irq_dest_mode			= 1,
+
+	.target_cpus			= summit_target_cpus,
+	.disable_esr			= 1,
+	.dest_logical			= APIC_DEST_LOGICAL,
+	.check_apicid_used		= summit_check_apicid_used,
+	.check_apicid_present		= summit_check_apicid_present,
+
+	.vector_allocation_domain	= flat_vector_allocation_domain,
+	.init_apic_ldr			= summit_init_apic_ldr,
+
+	.ioapic_phys_id_map		= summit_ioapic_phys_id_map,
+	.setup_apic_routing		= summit_setup_apic_routing,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= summit_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= summit_apicid_to_cpu_present,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= summit_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= summit_phys_pkg_id,
+	.mps_oem_check			= summit_mps_oem_check,
+
+	.get_apic_id			= summit_get_apic_id,
+	.set_apic_id			= NULL,
+	.apic_id_mask			= 0xFF << 24,
+
+	.cpu_mask_to_apicid_and		= summit_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= summit_send_IPI_mask,
+	.send_IPI_mask_allbutself	= NULL,
+	.send_IPI_allbutself		= summit_send_IPI_allbutself,
+	.send_IPI_all			= summit_send_IPI_all,
+	.send_IPI_self			= default_send_IPI_self,
+
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+
+	.wait_for_init_deassert		= default_wait_for_init_deassert,
+
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= default_inquire_remote_apic,
+
+	.read				= native_apic_mem_read,
+	.write				= native_apic_mem_write,
+	.eoi_write			= native_apic_mem_write,
+	.icr_read			= native_apic_icr_read,
+	.icr_write			= native_apic_icr_write,
+	.wait_icr_idle			= native_apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_apic_wait_icr_idle,
+
+	.x86_32_early_logical_apicid	= summit_early_logical_apicid,
+};
+
+apic_driver(apic_summit);

arch/x86/kernel/apic/x2apic_cluster.c

@@ -0,0 +1,296 @@
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/dmar.h>
+#include <linux/cpu.h>
+
+#include <asm/smp.h>
+#include <asm/x2apic.h>
+
+static DEFINE_PER_CPU(u32, x86_cpu_to_logical_apicid);
+static DEFINE_PER_CPU(cpumask_var_t, cpus_in_cluster);
+static DEFINE_PER_CPU(cpumask_var_t, ipi_mask);
+
+static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	return x2apic_enabled();
+}
+
+static inline u32 x2apic_cluster(int cpu)
+{
+	return per_cpu(x86_cpu_to_logical_apicid, cpu) >> 16;
+}
+
+static void
+__x2apic_send_IPI_mask(const struct cpumask *mask, int vector, int apic_dest)
+{
+	struct cpumask *cpus_in_cluster_ptr;
+	struct cpumask *ipi_mask_ptr;
+	unsigned int cpu, this_cpu;
+	unsigned long flags;
+	u32 dest;
+
+	x2apic_wrmsr_fence();
+
+	local_irq_save(flags);
+
+	this_cpu = smp_processor_id();
+
+	/*
+	 * We are to modify mask, so we need an own copy
+	 * and be sure it's manipulated with irq off.
+	 */
+	ipi_mask_ptr = __raw_get_cpu_var(ipi_mask);
+	cpumask_copy(ipi_mask_ptr, mask);
+
+	/*
+	 * The idea is to send one IPI per cluster.
+	 */
+	for_each_cpu(cpu, ipi_mask_ptr) {
+		unsigned long i;
+
+		cpus_in_cluster_ptr = per_cpu(cpus_in_cluster, cpu);
+		dest = 0;
+
+		/* Collect cpus in cluster. */
+		for_each_cpu_and(i, ipi_mask_ptr, cpus_in_cluster_ptr) {
+			if (apic_dest == APIC_DEST_ALLINC || i != this_cpu)
+				dest |= per_cpu(x86_cpu_to_logical_apicid, i);
+		}
+
+		if (!dest)
+			continue;
+
+		__x2apic_send_IPI_dest(dest, vector, apic->dest_logical);
+		/*
+		 * Cluster sibling cpus should be discared now so
+		 * we would not send IPI them second time.
+		 */
+		cpumask_andnot(ipi_mask_ptr, ipi_mask_ptr, cpus_in_cluster_ptr);
+	}
+
+	local_irq_restore(flags);
+}
+
+static void x2apic_send_IPI_mask(const struct cpumask *mask, int vector)
+{
+	__x2apic_send_IPI_mask(mask, vector, APIC_DEST_ALLINC);
+}
+
+static void
+x2apic_send_IPI_mask_allbutself(const struct cpumask *mask, int vector)
+{
+	__x2apic_send_IPI_mask(mask, vector, APIC_DEST_ALLBUT);
+}
+
+static void x2apic_send_IPI_allbutself(int vector)
+{
+	__x2apic_send_IPI_mask(cpu_online_mask, vector, APIC_DEST_ALLBUT);
+}
+
+static void x2apic_send_IPI_all(int vector)
+{
+	__x2apic_send_IPI_mask(cpu_online_mask, vector, APIC_DEST_ALLINC);
+}
+
+static int
+x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
+			      const struct cpumask *andmask,
+			      unsigned int *apicid)
+{
+	u32 dest = 0;
+	u16 cluster;
+	int i;
+
+	for_each_cpu_and(i, cpumask, andmask) {
+		if (!cpumask_test_cpu(i, cpu_online_mask))
+			continue;
+		dest = per_cpu(x86_cpu_to_logical_apicid, i);
+		cluster = x2apic_cluster(i);
+		break;
+	}
+
+	if (!dest)
+		return -EINVAL;
+
+	for_each_cpu_and(i, cpumask, andmask) {
+		if (!cpumask_test_cpu(i, cpu_online_mask))
+			continue;
+		if (cluster != x2apic_cluster(i))
+			continue;
+		dest |= per_cpu(x86_cpu_to_logical_apicid, i);
+	}
+
+	*apicid = dest;
+
+	return 0;
+}
+
+static void init_x2apic_ldr(void)
+{
+	unsigned int this_cpu = smp_processor_id();
+	unsigned int cpu;
+
+	per_cpu(x86_cpu_to_logical_apicid, this_cpu) = apic_read(APIC_LDR);
+
+	__cpu_set(this_cpu, per_cpu(cpus_in_cluster, this_cpu));
+	for_each_online_cpu(cpu) {
+		if (x2apic_cluster(this_cpu) != x2apic_cluster(cpu))
+			continue;
+		__cpu_set(this_cpu, per_cpu(cpus_in_cluster, cpu));
+		__cpu_set(cpu, per_cpu(cpus_in_cluster, this_cpu));
+	}
+}
+
+ /*
+  * At CPU state changes, update the x2apic cluster sibling info.
+  */
+static int __cpuinit
+update_clusterinfo(struct notifier_block *nfb, unsigned long action, void *hcpu)
+{
+	unsigned int this_cpu = (unsigned long)hcpu;
+	unsigned int cpu;
+	int err = 0;
+
+	switch (action) {
+	case CPU_UP_PREPARE:
+		if (!zalloc_cpumask_var(&per_cpu(cpus_in_cluster, this_cpu),
+					GFP_KERNEL)) {
+			err = -ENOMEM;
+		} else if (!zalloc_cpumask_var(&per_cpu(ipi_mask, this_cpu),
+					       GFP_KERNEL)) {
+			free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
+			err = -ENOMEM;
+		}
+		break;
+	case CPU_UP_CANCELED:
+	case CPU_UP_CANCELED_FROZEN:
+	case CPU_DEAD:
+		for_each_online_cpu(cpu) {
+			if (x2apic_cluster(this_cpu) != x2apic_cluster(cpu))
+				continue;
+			__cpu_clear(this_cpu, per_cpu(cpus_in_cluster, cpu));
+			__cpu_clear(cpu, per_cpu(cpus_in_cluster, this_cpu));
+		}
+		free_cpumask_var(per_cpu(cpus_in_cluster, this_cpu));
+		free_cpumask_var(per_cpu(ipi_mask, this_cpu));
+		break;
+	}
+
+	return notifier_from_errno(err);
+}
+
+static struct notifier_block __refdata x2apic_cpu_notifier = {
+	.notifier_call = update_clusterinfo,
+};
+
+static int x2apic_init_cpu_notifier(void)
+{
+	int cpu = smp_processor_id();
+
+	zalloc_cpumask_var(&per_cpu(cpus_in_cluster, cpu), GFP_KERNEL);
+	zalloc_cpumask_var(&per_cpu(ipi_mask, cpu), GFP_KERNEL);
+
+	BUG_ON(!per_cpu(cpus_in_cluster, cpu) || !per_cpu(ipi_mask, cpu));
+
+	__cpu_set(cpu, per_cpu(cpus_in_cluster, cpu));
+	register_hotcpu_notifier(&x2apic_cpu_notifier);
+	return 1;
+}
+
+static int x2apic_cluster_probe(void)
+{
+	if (x2apic_mode)
+		return x2apic_init_cpu_notifier();
+	else
+		return 0;
+}
+
+static const struct cpumask *x2apic_cluster_target_cpus(void)
+{
+	return cpu_all_mask;
+}
+
+/*
+ * Each x2apic cluster is an allocation domain.
+ */
+static void cluster_vector_allocation_domain(int cpu, struct cpumask *retmask,
+					     const struct cpumask *mask)
+{
+	/*
+	 * To minimize vector pressure, default case of boot, device bringup
+	 * etc will use a single cpu for the interrupt destination.
+	 *
+	 * On explicit migration requests coming from irqbalance etc,
+	 * interrupts will be routed to the x2apic cluster (cluster-id
+	 * derived from the first cpu in the mask) members specified
+	 * in the mask.
+	 */
+	if (mask == x2apic_cluster_target_cpus())
+		cpumask_copy(retmask, cpumask_of(cpu));
+	else
+		cpumask_and(retmask, mask, per_cpu(cpus_in_cluster, cpu));
+}
+
+static struct apic apic_x2apic_cluster = {
+
+	.name				= "cluster x2apic",
+	.probe				= x2apic_cluster_probe,
+	.acpi_madt_oem_check		= x2apic_acpi_madt_oem_check,
+	.apic_id_valid			= x2apic_apic_id_valid,
+	.apic_id_registered		= x2apic_apic_id_registered,
+
+	.irq_delivery_mode		= dest_LowestPrio,
+	.irq_dest_mode			= 1, /* logical */
+
+	.target_cpus			= x2apic_cluster_target_cpus,
+	.disable_esr			= 0,
+	.dest_logical			= APIC_DEST_LOGICAL,
+	.check_apicid_used		= NULL,
+	.check_apicid_present		= NULL,
+
+	.vector_allocation_domain	= cluster_vector_allocation_domain,
+	.init_apic_ldr			= init_x2apic_ldr,
+
+	.ioapic_phys_id_map		= NULL,
+	.setup_apic_routing		= NULL,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= NULL,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= default_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= x2apic_phys_pkg_id,
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= x2apic_get_apic_id,
+	.set_apic_id			= x2apic_set_apic_id,
+	.apic_id_mask			= 0xFFFFFFFFu,
+
+	.cpu_mask_to_apicid_and		= x2apic_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= x2apic_send_IPI_mask,
+	.send_IPI_mask_allbutself	= x2apic_send_IPI_mask_allbutself,
+	.send_IPI_allbutself		= x2apic_send_IPI_allbutself,
+	.send_IPI_all			= x2apic_send_IPI_all,
+	.send_IPI_self			= x2apic_send_IPI_self,
+
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+	.wait_for_init_deassert		= NULL,
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= NULL,
+
+	.read				= native_apic_msr_read,
+	.write				= native_apic_msr_write,
+	.eoi_write			= native_apic_msr_eoi_write,
+	.icr_read			= native_x2apic_icr_read,
+	.icr_write			= native_x2apic_icr_write,
+	.wait_icr_idle			= native_x2apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_x2apic_wait_icr_idle,
+};
+
+apic_driver(apic_x2apic_cluster);

arch/x86/kernel/apic/x2apic_phys.c

@@ -0,0 +1,150 @@
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/dmar.h>
+
+#include <asm/smp.h>
+#include <asm/x2apic.h>
+
+int x2apic_phys;
+
+static struct apic apic_x2apic_phys;
+
+static int set_x2apic_phys_mode(char *arg)
+{
+	x2apic_phys = 1;
+	return 0;
+}
+early_param("x2apic_phys", set_x2apic_phys_mode);
+
+static bool x2apic_fadt_phys(void)
+{
+	if ((acpi_gbl_FADT.header.revision >= FADT2_REVISION_ID) &&
+		(acpi_gbl_FADT.flags & ACPI_FADT_APIC_PHYSICAL)) {
+		printk(KERN_DEBUG "System requires x2apic physical mode\n");
+		return true;
+	}
+	return false;
+}
+
+static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
+{
+	return x2apic_enabled() && (x2apic_phys || x2apic_fadt_phys());
+}
+
+static void
+__x2apic_send_IPI_mask(const struct cpumask *mask, int vector, int apic_dest)
+{
+	unsigned long query_cpu;
+	unsigned long this_cpu;
+	unsigned long flags;
+
+	x2apic_wrmsr_fence();
+
+	local_irq_save(flags);
+
+	this_cpu = smp_processor_id();
+	for_each_cpu(query_cpu, mask) {
+		if (apic_dest == APIC_DEST_ALLBUT && this_cpu == query_cpu)
+			continue;
+		__x2apic_send_IPI_dest(per_cpu(x86_cpu_to_apicid, query_cpu),
+				       vector, APIC_DEST_PHYSICAL);
+	}
+	local_irq_restore(flags);
+}
+
+static void x2apic_send_IPI_mask(const struct cpumask *mask, int vector)
+{
+	__x2apic_send_IPI_mask(mask, vector, APIC_DEST_ALLINC);
+}
+
+static void
+ x2apic_send_IPI_mask_allbutself(const struct cpumask *mask, int vector)
+{
+	__x2apic_send_IPI_mask(mask, vector, APIC_DEST_ALLBUT);
+}
+
+static void x2apic_send_IPI_allbutself(int vector)
+{
+	__x2apic_send_IPI_mask(cpu_online_mask, vector, APIC_DEST_ALLBUT);
+}
+
+static void x2apic_send_IPI_all(int vector)
+{
+	__x2apic_send_IPI_mask(cpu_online_mask, vector, APIC_DEST_ALLINC);
+}
+
+static void init_x2apic_ldr(void)
+{
+}
+
+static int x2apic_phys_probe(void)
+{
+	if (x2apic_mode && (x2apic_phys || x2apic_fadt_phys()))
+		return 1;
+
+	return apic == &apic_x2apic_phys;
+}
+
+static struct apic apic_x2apic_phys = {
+
+	.name				= "physical x2apic",
+	.probe				= x2apic_phys_probe,
+	.acpi_madt_oem_check		= x2apic_acpi_madt_oem_check,
+	.apic_id_valid			= x2apic_apic_id_valid,
+	.apic_id_registered		= x2apic_apic_id_registered,
+
+	.irq_delivery_mode		= dest_Fixed,
+	.irq_dest_mode			= 0, /* physical */
+
+	.target_cpus			= online_target_cpus,
+	.disable_esr			= 0,
+	.dest_logical			= 0,
+	.check_apicid_used		= NULL,
+	.check_apicid_present		= NULL,
+
+	.vector_allocation_domain	= default_vector_allocation_domain,
+	.init_apic_ldr			= init_x2apic_ldr,
+
+	.ioapic_phys_id_map		= NULL,
+	.setup_apic_routing		= NULL,
+	.multi_timer_check		= NULL,
+	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+	.apicid_to_cpu_present		= NULL,
+	.setup_portio_remap		= NULL,
+	.check_phys_apicid_present	= default_check_phys_apicid_present,
+	.enable_apic_mode		= NULL,
+	.phys_pkg_id			= x2apic_phys_pkg_id,
+	.mps_oem_check			= NULL,
+
+	.get_apic_id			= x2apic_get_apic_id,
+	.set_apic_id			= x2apic_set_apic_id,
+	.apic_id_mask			= 0xFFFFFFFFu,
+
+	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
+
+	.send_IPI_mask			= x2apic_send_IPI_mask,
+	.send_IPI_mask_allbutself	= x2apic_send_IPI_mask_allbutself,
+	.send_IPI_allbutself		= x2apic_send_IPI_allbutself,
+	.send_IPI_all			= x2apic_send_IPI_all,
+	.send_IPI_self			= x2apic_send_IPI_self,
+
+	.trampoline_phys_low		= DEFAULT_TRAMPOLINE_PHYS_LOW,
+	.trampoline_phys_high		= DEFAULT_TRAMPOLINE_PHYS_HIGH,
+	.wait_for_init_deassert		= NULL,
+	.smp_callin_clear_local_apic	= NULL,
+	.inquire_remote_apic		= NULL,
+
+	.read				= native_apic_msr_read,
+	.write				= native_apic_msr_write,
+	.eoi_write			= native_apic_msr_eoi_write,
+	.icr_read			= native_x2apic_icr_read,
+	.icr_write			= native_x2apic_icr_write,
+	.wait_icr_idle			= native_x2apic_wait_icr_idle,
+	.safe_wait_icr_idle		= native_safe_x2apic_wait_icr_idle,
+};
+
+apic_driver(apic_x2apic_phys);