Resguardo inicial

linux-7.2.3/arch/x86/platform/uv/Makefile

@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+obj-$(CONFIG_X86_UV)		+= bios_uv.o uv_irq.o uv_time.o uv_nmi.o

linux-7.2.3/arch/x86/platform/uv/bios_uv.c

@@ -0,0 +1,269 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * BIOS run time interface routines.
+ *
+ * (C) Copyright 2020 Hewlett Packard Enterprise Development LP
+ * Copyright (C) 2007-2017 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (c) Russ Anderson <rja@sgi.com>
+ */
+
+#include <linux/efi.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <asm/efi.h>
+#include <linux/io.h>
+#include <asm/pgalloc.h>
+#include <asm/uv/bios.h>
+#include <asm/uv/uv_hub.h>
+
+unsigned long uv_systab_phys __ro_after_init = EFI_INVALID_TABLE_ADDR;
+
+struct uv_systab *uv_systab;
+
+static s64 __uv_bios_call(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3,
+			u64 a4, u64 a5)
+{
+	struct uv_systab *tab = uv_systab;
+	s64 ret;
+
+	if (!tab || !tab->function)
+		/*
+		 * BIOS does not support UV systab
+		 */
+		return BIOS_STATUS_UNIMPLEMENTED;
+
+	ret = efi_call_virt_pointer(tab, function, (u64)which, a1, a2, a3, a4, a5);
+
+	return ret;
+}
+
+static s64 uv_bios_call(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, u64 a4,
+		u64 a5)
+{
+	s64 ret;
+
+	if (down_interruptible(&__efi_uv_runtime_lock))
+		return BIOS_STATUS_ABORT;
+
+	ret = __uv_bios_call(which, a1, a2, a3, a4, a5);
+	up(&__efi_uv_runtime_lock);
+
+	return ret;
+}
+
+static s64 uv_bios_call_irqsave(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3,
+		u64 a4, u64 a5)
+{
+	unsigned long bios_flags;
+	s64 ret;
+
+	if (down_interruptible(&__efi_uv_runtime_lock))
+		return BIOS_STATUS_ABORT;
+
+	local_irq_save(bios_flags);
+	ret = __uv_bios_call(which, a1, a2, a3, a4, a5);
+	local_irq_restore(bios_flags);
+
+	up(&__efi_uv_runtime_lock);
+
+	return ret;
+}
+
+long sn_partition_id;
+EXPORT_SYMBOL_GPL(sn_partition_id);
+long sn_coherency_id;
+EXPORT_SYMBOL_GPL(sn_coherency_id);
+long sn_region_size;
+EXPORT_SYMBOL_GPL(sn_region_size);
+long system_serial_number;
+int uv_type;
+
+s64 uv_bios_get_sn_info(int fc, int *uvtype, long *partid, long *coher,
+		long *region, long *ssn)
+{
+	s64 ret;
+	u64 v0, v1;
+	union partition_info_u part;
+
+	ret = uv_bios_call_irqsave(UV_BIOS_GET_SN_INFO, fc,
+				(u64)(&v0), (u64)(&v1), 0, 0);
+	if (ret != BIOS_STATUS_SUCCESS)
+		return ret;
+
+	part.val = v0;
+	if (uvtype)
+		*uvtype = part.hub_version;
+	if (partid)
+		*partid = part.partition_id;
+	if (coher)
+		*coher = part.coherence_id;
+	if (region)
+		*region = part.region_size;
+	if (ssn)
+		*ssn = v1;
+	return ret;
+}
+
+int
+uv_bios_mq_watchlist_alloc(unsigned long addr, unsigned int mq_size,
+			   unsigned long *intr_mmr_offset)
+{
+	u64 watchlist;
+	s64 ret;
+
+	/*
+	 * bios returns watchlist number or negative error number.
+	 */
+	ret = (int)uv_bios_call_irqsave(UV_BIOS_WATCHLIST_ALLOC, addr,
+			mq_size, (u64)intr_mmr_offset,
+			(u64)&watchlist, 0);
+	if (ret < BIOS_STATUS_SUCCESS)
+		return ret;
+
+	return watchlist;
+}
+EXPORT_SYMBOL_GPL(uv_bios_mq_watchlist_alloc);
+
+int
+uv_bios_mq_watchlist_free(int blade, int watchlist_num)
+{
+	return (int)uv_bios_call_irqsave(UV_BIOS_WATCHLIST_FREE,
+				blade, watchlist_num, 0, 0, 0);
+}
+EXPORT_SYMBOL_GPL(uv_bios_mq_watchlist_free);
+
+s64
+uv_bios_change_memprotect(u64 paddr, u64 len, enum uv_memprotect perms)
+{
+	return uv_bios_call_irqsave(UV_BIOS_MEMPROTECT, paddr, len,
+					perms, 0, 0);
+}
+EXPORT_SYMBOL_GPL(uv_bios_change_memprotect);
+
+s64
+uv_bios_reserved_page_pa(u64 buf, u64 *cookie, u64 *addr, u64 *len)
+{
+	return uv_bios_call_irqsave(UV_BIOS_GET_PARTITION_ADDR, (u64)cookie,
+				    (u64)addr, buf, (u64)len, 0);
+}
+EXPORT_SYMBOL_GPL(uv_bios_reserved_page_pa);
+
+s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second)
+{
+	return uv_bios_call(UV_BIOS_FREQ_BASE, clock_type,
+			   (u64)ticks_per_second, 0, 0, 0);
+}
+
+/*
+ * uv_bios_set_legacy_vga_target - Set Legacy VGA I/O Target
+ * @decode: true to enable target, false to disable target
+ * @domain: PCI domain number
+ * @bus: PCI bus number
+ *
+ * Returns:
+ *    0: Success
+ *    -EINVAL: Invalid domain or bus number
+ *    -ENOSYS: Capability not available
+ *    -EBUSY: Legacy VGA I/O cannot be retargeted at this time
+ */
+int uv_bios_set_legacy_vga_target(bool decode, int domain, int bus)
+{
+	return uv_bios_call(UV_BIOS_SET_LEGACY_VGA_TARGET,
+				(u64)decode, (u64)domain, (u64)bus, 0, 0);
+}
+
+extern s64 uv_bios_get_master_nasid(u64 size, u64 *master_nasid)
+{
+	return uv_bios_call(UV_BIOS_EXTRA, 0, UV_BIOS_EXTRA_MASTER_NASID, 0,
+				size, (u64)master_nasid);
+}
+EXPORT_SYMBOL_GPL(uv_bios_get_master_nasid);
+
+extern s64 uv_bios_get_heapsize(u64 nasid, u64 size, u64 *heap_size)
+{
+	return uv_bios_call(UV_BIOS_EXTRA, nasid, UV_BIOS_EXTRA_GET_HEAPSIZE,
+				0, size, (u64)heap_size);
+}
+EXPORT_SYMBOL_GPL(uv_bios_get_heapsize);
+
+extern s64 uv_bios_install_heap(u64 nasid, u64 heap_size, u64 *bios_heap)
+{
+	return uv_bios_call(UV_BIOS_EXTRA, nasid, UV_BIOS_EXTRA_INSTALL_HEAP,
+				0, heap_size, (u64)bios_heap);
+}
+EXPORT_SYMBOL_GPL(uv_bios_install_heap);
+
+extern s64 uv_bios_obj_count(u64 nasid, u64 size, u64 *objcnt)
+{
+	return uv_bios_call(UV_BIOS_EXTRA, nasid, UV_BIOS_EXTRA_OBJECT_COUNT,
+				0, size, (u64)objcnt);
+}
+EXPORT_SYMBOL_GPL(uv_bios_obj_count);
+
+extern s64 uv_bios_enum_objs(u64 nasid, u64 size, u64 *objbuf)
+{
+	return uv_bios_call(UV_BIOS_EXTRA, nasid, UV_BIOS_EXTRA_ENUM_OBJECTS,
+				0, size, (u64)objbuf);
+}
+EXPORT_SYMBOL_GPL(uv_bios_enum_objs);
+
+extern s64 uv_bios_enum_ports(u64 nasid, u64 obj_id, u64 size, u64 *portbuf)
+{
+	return uv_bios_call(UV_BIOS_EXTRA, nasid, UV_BIOS_EXTRA_ENUM_PORTS,
+				obj_id, size, (u64)portbuf);
+}
+EXPORT_SYMBOL_GPL(uv_bios_enum_ports);
+
+extern s64 uv_bios_get_geoinfo(u64 nasid, u64 size, u64 *buf)
+{
+	return uv_bios_call(UV_BIOS_GET_GEOINFO, nasid, (u64)buf, size, 0, 0);
+}
+EXPORT_SYMBOL_GPL(uv_bios_get_geoinfo);
+
+extern s64 uv_bios_get_pci_topology(u64 size, u64 *buf)
+{
+	return uv_bios_call(UV_BIOS_GET_PCI_TOPOLOGY, (u64)buf, size, 0, 0, 0);
+}
+EXPORT_SYMBOL_GPL(uv_bios_get_pci_topology);
+
+unsigned long get_uv_systab_phys(bool msg)
+{
+	if ((uv_systab_phys == EFI_INVALID_TABLE_ADDR) ||
+	    !uv_systab_phys || efi_runtime_disabled()) {
+		if (msg)
+			pr_crit("UV: UVsystab: missing\n");
+		return 0;
+	}
+	return uv_systab_phys;
+}
+
+int uv_bios_init(void)
+{
+	unsigned long uv_systab_phys_addr;
+
+	uv_systab = NULL;
+	uv_systab_phys_addr = get_uv_systab_phys(1);
+	if (!uv_systab_phys_addr)
+		return -EEXIST;
+
+	uv_systab = ioremap(uv_systab_phys_addr, sizeof(struct uv_systab));
+	if (!uv_systab || strncmp(uv_systab->signature, UV_SYSTAB_SIG, 4)) {
+		pr_err("UV: UVsystab: bad signature!\n");
+		iounmap(uv_systab);
+		return -EINVAL;
+	}
+
+	/* Starting with UV4 the UV systab size is variable */
+	if (uv_systab->revision >= UV_SYSTAB_VERSION_UV4) {
+		int size = uv_systab->size;
+
+		iounmap(uv_systab);
+		uv_systab = ioremap(uv_systab_phys_addr, size);
+		if (!uv_systab) {
+			pr_err("UV: UVsystab: ioremap(%d) failed!\n", size);
+			return -EFAULT;
+		}
+	}
+	pr_info("UV: UVsystab: Revision:%x\n", uv_systab->revision);
+	return 0;
+}

linux-7.2.3/arch/x86/platform/uv/uv_irq.c

@@ -0,0 +1,213 @@
+/*
+ * 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.
+ *
+ * SGI UV IRQ functions
+ *
+ * Copyright (C) 2008 Silicon Graphics, Inc. All rights reserved.
+ */
+
+#include <linux/export.h>
+#include <linux/rbtree.h>
+#include <linux/slab.h>
+#include <linux/irq.h>
+
+#include <asm/irqdomain.h>
+#include <asm/apic.h>
+#include <asm/uv/uv_irq.h>
+#include <asm/uv/uv_hub.h>
+
+/* MMR offset and pnode of hub sourcing interrupts for a given irq */
+struct uv_irq_2_mmr_pnode {
+	unsigned long		offset;
+	int			pnode;
+};
+
+static void uv_program_mmr(struct irq_cfg *cfg, struct uv_irq_2_mmr_pnode *info)
+{
+	unsigned long mmr_value;
+	struct uv_IO_APIC_route_entry *entry;
+
+	BUILD_BUG_ON(sizeof(struct uv_IO_APIC_route_entry) !=
+		     sizeof(unsigned long));
+
+	mmr_value = 0;
+	entry = (struct uv_IO_APIC_route_entry *)&mmr_value;
+	entry->vector		= cfg->vector;
+	entry->delivery_mode	= APIC_DELIVERY_MODE_FIXED;
+	entry->dest_mode	= apic->dest_mode_logical;
+	entry->polarity		= 0;
+	entry->trigger		= 0;
+	entry->mask		= 0;
+	entry->dest		= cfg->dest_apicid;
+
+	uv_write_global_mmr64(info->pnode, info->offset, mmr_value);
+}
+
+static void uv_noop(struct irq_data *data) { }
+
+static int
+uv_set_irq_affinity(struct irq_data *data, const struct cpumask *mask,
+		    bool force)
+{
+	struct irq_data *parent = data->parent_data;
+	struct irq_cfg *cfg = irqd_cfg(data);
+	int ret;
+
+	ret = parent->chip->irq_set_affinity(parent, mask, force);
+	if (ret >= 0) {
+		uv_program_mmr(cfg, data->chip_data);
+		vector_schedule_cleanup(cfg);
+	}
+
+	return ret;
+}
+
+static struct irq_chip uv_irq_chip = {
+	.name			= "UV-CORE",
+	.irq_mask		= uv_noop,
+	.irq_unmask		= uv_noop,
+	.irq_eoi		= apic_ack_irq,
+	.irq_set_affinity	= uv_set_irq_affinity,
+};
+
+static int uv_domain_alloc(struct irq_domain *domain, unsigned int virq,
+			   unsigned int nr_irqs, void *arg)
+{
+	struct uv_irq_2_mmr_pnode *chip_data;
+	struct irq_alloc_info *info = arg;
+	struct irq_data *irq_data = irq_domain_get_irq_data(domain, virq);
+	int ret;
+
+	if (nr_irqs > 1 || !info || info->type != X86_IRQ_ALLOC_TYPE_UV)
+		return -EINVAL;
+
+	chip_data = kmalloc_node(sizeof(*chip_data), GFP_KERNEL,
+				 irq_data_get_node(irq_data));
+	if (!chip_data)
+		return -ENOMEM;
+
+	ret = irq_domain_alloc_irqs_parent(domain, virq, nr_irqs, arg);
+	if (ret >= 0) {
+		if (info->uv.limit == UV_AFFINITY_CPU)
+			irq_set_status_flags(virq, IRQ_NO_BALANCING);
+
+		chip_data->pnode = uv_blade_to_pnode(info->uv.blade);
+		chip_data->offset = info->uv.offset;
+		irq_domain_set_info(domain, virq, virq, &uv_irq_chip, chip_data,
+				    handle_percpu_irq, NULL, info->uv.name);
+	} else {
+		kfree(chip_data);
+	}
+
+	return ret;
+}
+
+static void uv_domain_free(struct irq_domain *domain, unsigned int virq,
+			   unsigned int nr_irqs)
+{
+	struct irq_data *irq_data = irq_domain_get_irq_data(domain, virq);
+
+	BUG_ON(nr_irqs != 1);
+	kfree(irq_data->chip_data);
+	irq_clear_status_flags(virq, IRQ_NO_BALANCING);
+	irq_domain_free_irqs_top(domain, virq, nr_irqs);
+}
+
+/*
+ * Re-target the irq to the specified CPU and enable the specified MMR located
+ * on the specified blade to allow the sending of MSIs to the specified CPU.
+ */
+static int uv_domain_activate(struct irq_domain *domain,
+			      struct irq_data *irq_data, bool reserve)
+{
+	uv_program_mmr(irqd_cfg(irq_data), irq_data->chip_data);
+	return 0;
+}
+
+/*
+ * Disable the specified MMR located on the specified blade so that MSIs are
+ * longer allowed to be sent.
+ */
+static void uv_domain_deactivate(struct irq_domain *domain,
+				 struct irq_data *irq_data)
+{
+	unsigned long mmr_value;
+	struct uv_IO_APIC_route_entry *entry;
+
+	mmr_value = 0;
+	entry = (struct uv_IO_APIC_route_entry *)&mmr_value;
+	entry->mask = 1;
+	uv_program_mmr(irqd_cfg(irq_data), irq_data->chip_data);
+}
+
+static const struct irq_domain_ops uv_domain_ops = {
+	.alloc		= uv_domain_alloc,
+	.free		= uv_domain_free,
+	.activate	= uv_domain_activate,
+	.deactivate	= uv_domain_deactivate,
+};
+
+static struct irq_domain *uv_get_irq_domain(void)
+{
+	static struct irq_domain *uv_domain;
+	static DEFINE_MUTEX(uv_lock);
+	struct fwnode_handle *fn;
+
+	mutex_lock(&uv_lock);
+	if (uv_domain)
+		goto out;
+
+	fn = irq_domain_alloc_named_fwnode("UV-CORE");
+	if (!fn)
+		goto out;
+
+	uv_domain = irq_domain_create_hierarchy(x86_vector_domain, 0, 0, fn,
+						&uv_domain_ops, NULL);
+	if (!uv_domain)
+		irq_domain_free_fwnode(fn);
+out:
+	mutex_unlock(&uv_lock);
+
+	return uv_domain;
+}
+
+/*
+ * Set up a mapping of an available irq and vector, and enable the specified
+ * MMR that defines the MSI that is to be sent to the specified CPU when an
+ * interrupt is raised.
+ */
+int uv_setup_irq(char *irq_name, int cpu, int mmr_blade,
+		 unsigned long mmr_offset, int limit)
+{
+	struct irq_alloc_info info;
+	struct irq_domain *domain = uv_get_irq_domain();
+
+	if (!domain)
+		return -ENOMEM;
+
+	init_irq_alloc_info(&info, cpumask_of(cpu));
+	info.type = X86_IRQ_ALLOC_TYPE_UV;
+	info.uv.limit = limit;
+	info.uv.blade = mmr_blade;
+	info.uv.offset = mmr_offset;
+	info.uv.name = irq_name;
+
+	return irq_domain_alloc_irqs(domain, 1,
+				     uv_blade_to_memory_nid(mmr_blade), &info);
+}
+EXPORT_SYMBOL_GPL(uv_setup_irq);
+
+/*
+ * Tear down a mapping of an irq and vector, and disable the specified MMR that
+ * defined the MSI that was to be sent to the specified CPU when an interrupt
+ * was raised.
+ *
+ * Set mmr_blade and mmr_offset to what was passed in on uv_setup_irq().
+ */
+void uv_teardown_irq(unsigned int irq)
+{
+	irq_domain_free_irqs(irq, 1);
+}
+EXPORT_SYMBOL_GPL(uv_teardown_irq);

linux-7.2.3/arch/x86/platform/uv/uv_time.c

@@ -0,0 +1,393 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * SGI RTC clock/timer routines.
+ *
+ *  (C) Copyright 2020 Hewlett Packard Enterprise Development LP
+ *  Copyright (c) 2009-2013 Silicon Graphics, Inc.  All Rights Reserved.
+ *  Copyright (c) Dimitri Sivanich
+ */
+#include <linux/clockchips.h>
+#include <linux/slab.h>
+
+#include <asm/uv/uv_mmrs.h>
+#include <asm/uv/uv_hub.h>
+#include <asm/uv/bios.h>
+#include <asm/uv/uv.h>
+#include <asm/apic.h>
+#include <asm/cpu.h>
+
+#define RTC_NAME		"sgi_rtc"
+
+static u64 uv_read_rtc(struct clocksource *cs);
+static int uv_rtc_next_event(unsigned long, struct clock_event_device *);
+static int uv_rtc_shutdown(struct clock_event_device *evt);
+
+static struct clocksource clocksource_uv = {
+	.name		= RTC_NAME,
+	.rating		= 299,
+	.read		= uv_read_rtc,
+	.mask		= (u64)UVH_RTC_REAL_TIME_CLOCK_MASK,
+	.flags		= CLOCK_SOURCE_IS_CONTINUOUS,
+};
+
+static struct clock_event_device clock_event_device_uv = {
+	.name			= RTC_NAME,
+	.features		= CLOCK_EVT_FEAT_ONESHOT,
+	.shift			= 20,
+	.rating			= 400,
+	.irq			= -1,
+	.set_next_event		= uv_rtc_next_event,
+	.set_state_shutdown	= uv_rtc_shutdown,
+	.event_handler		= NULL,
+};
+
+static DEFINE_PER_CPU(struct clock_event_device, cpu_ced);
+
+/* There is one of these allocated per node */
+struct uv_rtc_timer_head {
+	spinlock_t	lock;
+	/* next cpu waiting for timer, local node relative: */
+	int		next_cpu;
+	/* number of cpus on this node: */
+	int		ncpus;
+	struct {
+		int	lcpu;		/* systemwide logical cpu number */
+		u64	expires;	/* next timer expiration for this cpu */
+	} cpu[] __counted_by(ncpus);
+};
+
+/*
+ * Access to uv_rtc_timer_head via blade id.
+ */
+static struct uv_rtc_timer_head		**blade_info __read_mostly;
+
+static int				uv_rtc_evt_enable;
+
+/*
+ * Hardware interface routines
+ */
+
+/* Send IPIs to another node */
+static void uv_rtc_send_IPI(int cpu)
+{
+	unsigned long apicid, val;
+	int pnode;
+
+	apicid = cpu_physical_id(cpu);
+	pnode = uv_apicid_to_pnode(apicid);
+	val = (1UL << UVH_IPI_INT_SEND_SHFT) |
+	      (apicid << UVH_IPI_INT_APIC_ID_SHFT) |
+	      (X86_PLATFORM_IPI_VECTOR << UVH_IPI_INT_VECTOR_SHFT);
+
+	uv_write_global_mmr64(pnode, UVH_IPI_INT, val);
+}
+
+/* Check for an RTC interrupt pending */
+static int uv_intr_pending(int pnode)
+{
+	return uv_read_global_mmr64(pnode, UVH_EVENT_OCCURRED2) &
+		UVH_EVENT_OCCURRED2_RTC_1_MASK;
+}
+
+/* Setup interrupt and return non-zero if early expiration occurred. */
+static int uv_setup_intr(int cpu, u64 expires)
+{
+	u64 val;
+	unsigned long apicid = cpu_physical_id(cpu);
+	int pnode = uv_cpu_to_pnode(cpu);
+
+	uv_write_global_mmr64(pnode, UVH_RTC1_INT_CONFIG,
+		UVH_RTC1_INT_CONFIG_M_MASK);
+	uv_write_global_mmr64(pnode, UVH_INT_CMPB, -1L);
+
+	uv_write_global_mmr64(pnode, UVH_EVENT_OCCURRED2_ALIAS,
+			      UVH_EVENT_OCCURRED2_RTC_1_MASK);
+
+	val = (X86_PLATFORM_IPI_VECTOR << UVH_RTC1_INT_CONFIG_VECTOR_SHFT) |
+		((u64)apicid << UVH_RTC1_INT_CONFIG_APIC_ID_SHFT);
+
+	/* Set configuration */
+	uv_write_global_mmr64(pnode, UVH_RTC1_INT_CONFIG, val);
+	/* Initialize comparator value */
+	uv_write_global_mmr64(pnode, UVH_INT_CMPB, expires);
+
+	if (uv_read_rtc(NULL) <= expires)
+		return 0;
+
+	return !uv_intr_pending(pnode);
+}
+
+/*
+ * Per-cpu timer tracking routines
+ */
+
+static __init void uv_rtc_deallocate_timers(void)
+{
+	int bid;
+
+	for_each_possible_blade(bid) {
+		kfree(blade_info[bid]);
+	}
+	kfree(blade_info);
+}
+
+/* Allocate per-node list of cpu timer expiration times. */
+static __init int uv_rtc_allocate_timers(void)
+{
+	int cpu;
+
+	blade_info = kcalloc(uv_possible_blades, sizeof(void *), GFP_KERNEL);
+	if (!blade_info)
+		return -ENOMEM;
+
+	for_each_present_cpu(cpu) {
+		int nid = cpu_to_node(cpu);
+		int bid = uv_cpu_to_blade_id(cpu);
+		int bcpu = uv_cpu_blade_processor_id(cpu);
+		struct uv_rtc_timer_head *head = blade_info[bid];
+
+		if (!head) {
+			head = kmalloc_node(struct_size(head, cpu,
+				uv_blade_nr_possible_cpus(bid)),
+				GFP_KERNEL, nid);
+			if (!head) {
+				uv_rtc_deallocate_timers();
+				return -ENOMEM;
+			}
+			spin_lock_init(&head->lock);
+			head->ncpus = uv_blade_nr_possible_cpus(bid);
+			head->next_cpu = -1;
+			blade_info[bid] = head;
+		}
+
+		head->cpu[bcpu].lcpu = cpu;
+		head->cpu[bcpu].expires = ULLONG_MAX;
+	}
+
+	return 0;
+}
+
+/* Find and set the next expiring timer.  */
+static void uv_rtc_find_next_timer(struct uv_rtc_timer_head *head, int pnode)
+{
+	u64 lowest = ULLONG_MAX;
+	int c, bcpu = -1;
+
+	head->next_cpu = -1;
+	for (c = 0; c < head->ncpus; c++) {
+		u64 exp = head->cpu[c].expires;
+		if (exp < lowest) {
+			bcpu = c;
+			lowest = exp;
+		}
+	}
+	if (bcpu >= 0) {
+		head->next_cpu = bcpu;
+		c = head->cpu[bcpu].lcpu;
+		if (uv_setup_intr(c, lowest))
+			/* If we didn't set it up in time, trigger */
+			uv_rtc_send_IPI(c);
+	} else {
+		uv_write_global_mmr64(pnode, UVH_RTC1_INT_CONFIG,
+			UVH_RTC1_INT_CONFIG_M_MASK);
+	}
+}
+
+/*
+ * Set expiration time for current cpu.
+ *
+ * Returns 1 if we missed the expiration time.
+ */
+static int uv_rtc_set_timer(int cpu, u64 expires)
+{
+	int pnode = uv_cpu_to_pnode(cpu);
+	int bid = uv_cpu_to_blade_id(cpu);
+	struct uv_rtc_timer_head *head = blade_info[bid];
+	int bcpu = uv_cpu_blade_processor_id(cpu);
+	u64 *t = &head->cpu[bcpu].expires;
+	unsigned long flags;
+	int next_cpu;
+
+	spin_lock_irqsave(&head->lock, flags);
+
+	next_cpu = head->next_cpu;
+	*t = expires;
+
+	/* Will this one be next to go off? */
+	if (next_cpu < 0 || bcpu == next_cpu ||
+			expires < head->cpu[next_cpu].expires) {
+		head->next_cpu = bcpu;
+		if (uv_setup_intr(cpu, expires)) {
+			*t = ULLONG_MAX;
+			uv_rtc_find_next_timer(head, pnode);
+			spin_unlock_irqrestore(&head->lock, flags);
+			return -ETIME;
+		}
+	}
+
+	spin_unlock_irqrestore(&head->lock, flags);
+	return 0;
+}
+
+/*
+ * Unset expiration time for current cpu.
+ *
+ * Returns 1 if this timer was pending.
+ */
+static int uv_rtc_unset_timer(int cpu, int force)
+{
+	int pnode = uv_cpu_to_pnode(cpu);
+	int bid = uv_cpu_to_blade_id(cpu);
+	struct uv_rtc_timer_head *head = blade_info[bid];
+	int bcpu = uv_cpu_blade_processor_id(cpu);
+	u64 *t = &head->cpu[bcpu].expires;
+	unsigned long flags;
+	int rc = 0;
+
+	spin_lock_irqsave(&head->lock, flags);
+
+	if ((head->next_cpu == bcpu && uv_read_rtc(NULL) >= *t) || force)
+		rc = 1;
+
+	if (rc) {
+		*t = ULLONG_MAX;
+		/* Was the hardware setup for this timer? */
+		if (head->next_cpu == bcpu)
+			uv_rtc_find_next_timer(head, pnode);
+	}
+
+	spin_unlock_irqrestore(&head->lock, flags);
+
+	return rc;
+}
+
+
+/*
+ * Kernel interface routines.
+ */
+
+/*
+ * Read the RTC.
+ *
+ * Starting with HUB rev 2.0, the UV RTC register is replicated across all
+ * cachelines of its own page.  This allows faster simultaneous reads
+ * from a given socket.
+ */
+static u64 uv_read_rtc(struct clocksource *cs)
+{
+	unsigned long offset;
+
+	if (uv_get_min_hub_revision_id() == 1)
+		offset = 0;
+	else
+		offset = (uv_blade_processor_id() * L1_CACHE_BYTES) % PAGE_SIZE;
+
+	return (u64)uv_read_local_mmr(UVH_RTC | offset);
+}
+
+/*
+ * Program the next event, relative to now
+ */
+static int uv_rtc_next_event(unsigned long delta,
+			     struct clock_event_device *ced)
+{
+	int ced_cpu = cpumask_first(ced->cpumask);
+
+	return uv_rtc_set_timer(ced_cpu, delta + uv_read_rtc(NULL));
+}
+
+/*
+ * Shutdown the RTC timer
+ */
+static int uv_rtc_shutdown(struct clock_event_device *evt)
+{
+	int ced_cpu = cpumask_first(evt->cpumask);
+
+	uv_rtc_unset_timer(ced_cpu, 1);
+	return 0;
+}
+
+static void uv_rtc_interrupt(void)
+{
+	int cpu = smp_processor_id();
+	struct clock_event_device *ced = &per_cpu(cpu_ced, cpu);
+
+	if (!ced || !ced->event_handler)
+		return;
+
+	if (uv_rtc_unset_timer(cpu, 0) != 1)
+		return;
+
+	ced->event_handler(ced);
+}
+
+static int __init uv_enable_evt_rtc(char *str)
+{
+	uv_rtc_evt_enable = 1;
+
+	return 1;
+}
+__setup("uvrtcevt", uv_enable_evt_rtc);
+
+static __init void uv_rtc_register_clockevents(struct work_struct *dummy)
+{
+	struct clock_event_device *ced = this_cpu_ptr(&cpu_ced);
+
+	*ced = clock_event_device_uv;
+	ced->cpumask = cpumask_of(smp_processor_id());
+	clockevents_register_device(ced);
+}
+
+static __init int uv_rtc_setup_clock(void)
+{
+	int rc;
+
+	if (!is_uv_system())
+		return -ENODEV;
+
+	rc = clocksource_register_hz(&clocksource_uv, sn_rtc_cycles_per_second);
+	if (rc)
+		printk(KERN_INFO "UV RTC clocksource failed rc %d\n", rc);
+	else
+		printk(KERN_INFO "UV RTC clocksource registered freq %lu MHz\n",
+			sn_rtc_cycles_per_second/(unsigned long)1E6);
+
+	if (rc || !uv_rtc_evt_enable || x86_platform_ipi_callback)
+		return rc;
+
+	/* Setup and register clockevents */
+	rc = uv_rtc_allocate_timers();
+	if (rc)
+		goto error;
+
+	x86_platform_ipi_callback = uv_rtc_interrupt;
+
+	clock_event_device_uv.mult = div_sc(sn_rtc_cycles_per_second,
+				NSEC_PER_SEC, clock_event_device_uv.shift);
+
+	clock_event_device_uv.min_delta_ns = NSEC_PER_SEC /
+						sn_rtc_cycles_per_second;
+	clock_event_device_uv.min_delta_ticks = 1;
+
+	clock_event_device_uv.max_delta_ns = clocksource_uv.mask *
+				(NSEC_PER_SEC / sn_rtc_cycles_per_second);
+	clock_event_device_uv.max_delta_ticks = clocksource_uv.mask;
+
+	rc = schedule_on_each_cpu(uv_rtc_register_clockevents);
+	if (rc) {
+		x86_platform_ipi_callback = NULL;
+		uv_rtc_deallocate_timers();
+		goto error;
+	}
+
+	printk(KERN_INFO "UV RTC clockevents registered\n");
+
+	return 0;
+
+error:
+	clocksource_unregister(&clocksource_uv);
+	printk(KERN_INFO "UV RTC clockevents failed rc %d\n", rc);
+
+	return rc;
+}
+arch_initcall(uv_rtc_setup_clock);