[deliverable/linux.git] / arch / powerpc / platforms / pseries / hotplug-cpu.c

/*
 * pseries CPU Hotplug infrastructure.
 *
 * Split out from arch/powerpc/platforms/pseries/setup.c
 *  arch/powerpc/kernel/rtas.c, and arch/powerpc/platforms/pseries/smp.c
 *
 * Peter Bergner, IBM	March 2001.
 * Copyright (C) 2001 IBM.
 * Dave Engebretsen, Peter Bergner, and
 * Mike Corrigan {engebret|bergner|mikec}@us.ibm.com
 * Plus various changes from other IBM teams...
 *
 * Copyright (C) 2006 Michael Ellerman, IBM Corporation
 *
 *      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.
 */

#include <linux/kernel.h>
#include <linux/interrupt.h>
#include <linux/delay.h>
#include <linux/sched.h>	/* for idle_task_exit */
#include <linux/cpu.h>
#include <linux/of.h>
#include <asm/prom.h>
#include <asm/rtas.h>
#include <asm/firmware.h>
#include <asm/machdep.h>
#include <asm/vdso_datapage.h>
#include <asm/xics.h>
#include <asm/plpar_wrappers.h>

#include "offline_states.h"

/* This version can't take the spinlock, because it never returns */
static int rtas_stop_self_token = RTAS_UNKNOWN_SERVICE;

static DEFINE_PER_CPU(enum cpu_state_vals, preferred_offline_state) =
							CPU_STATE_OFFLINE;
static DEFINE_PER_CPU(enum cpu_state_vals, current_state) = CPU_STATE_OFFLINE;

static enum cpu_state_vals default_offline_state = CPU_STATE_OFFLINE;

static int cede_offline_enabled __read_mostly = 1;

/*
 * Enable/disable cede_offline when available.
 */
static int __init setup_cede_offline(char *str)
{
	if (!strcmp(str, "off"))
		cede_offline_enabled = 0;
	else if (!strcmp(str, "on"))
		cede_offline_enabled = 1;
	else
		return 0;
	return 1;
}

__setup("cede_offline=", setup_cede_offline);

enum cpu_state_vals get_cpu_current_state(int cpu)
{
	return per_cpu(current_state, cpu);
}

void set_cpu_current_state(int cpu, enum cpu_state_vals state)
{
	per_cpu(current_state, cpu) = state;
}

enum cpu_state_vals get_preferred_offline_state(int cpu)
{
	return per_cpu(preferred_offline_state, cpu);
}

void set_preferred_offline_state(int cpu, enum cpu_state_vals state)
{
	per_cpu(preferred_offline_state, cpu) = state;
}

void set_default_offline_state(int cpu)
{
	per_cpu(preferred_offline_state, cpu) = default_offline_state;
}

static void rtas_stop_self(void)
{
	static struct rtas_args args = {
		.nargs = 0,
		.nret = cpu_to_be32(1),
		.rets = &args.args[0],
	};

	args.token = cpu_to_be32(rtas_stop_self_token);

	local_irq_disable();

	BUG_ON(rtas_stop_self_token == RTAS_UNKNOWN_SERVICE);

	printk("cpu %u (hwid %u) Ready to die...\n",
	       smp_processor_id(), hard_smp_processor_id());
	enter_rtas(__pa(&args));

	panic("Alas, I survived.\n");
}

static void pseries_mach_cpu_die(void)
{
	unsigned int cpu = smp_processor_id();
	unsigned int hwcpu = hard_smp_processor_id();
	u8 cede_latency_hint = 0;

	local_irq_disable();
	idle_task_exit();
	xics_teardown_cpu();

	if (get_preferred_offline_state(cpu) == CPU_STATE_INACTIVE) {
		set_cpu_current_state(cpu, CPU_STATE_INACTIVE);
		if (ppc_md.suspend_disable_cpu)
			ppc_md.suspend_disable_cpu();

		cede_latency_hint = 2;

		get_lppaca()->idle = 1;
		if (!lppaca_shared_proc(get_lppaca()))
			get_lppaca()->donate_dedicated_cpu = 1;

		while (get_preferred_offline_state(cpu) == CPU_STATE_INACTIVE) {
			while (!prep_irq_for_idle()) {
				local_irq_enable();
				local_irq_disable();
			}

			extended_cede_processor(cede_latency_hint);
		}

		local_irq_disable();

		if (!lppaca_shared_proc(get_lppaca()))
			get_lppaca()->donate_dedicated_cpu = 0;
		get_lppaca()->idle = 0;

		if (get_preferred_offline_state(cpu) == CPU_STATE_ONLINE) {
			unregister_slb_shadow(hwcpu);

			hard_irq_disable();
			/*
			 * Call to start_secondary_resume() will not return.
			 * Kernel stack will be reset and start_secondary()
			 * will be called to continue the online operation.
			 */
			start_secondary_resume();
		}
	}

	/* Requested state is CPU_STATE_OFFLINE at this point */
	WARN_ON(get_preferred_offline_state(cpu) != CPU_STATE_OFFLINE);

	set_cpu_current_state(cpu, CPU_STATE_OFFLINE);
	unregister_slb_shadow(hwcpu);
	rtas_stop_self();

	/* Should never get here... */
	BUG();
	for(;;);
}

static int pseries_cpu_disable(void)
{
	int cpu = smp_processor_id();

	set_cpu_online(cpu, false);
	vdso_data->processorCount--;

	/*fix boot_cpuid here*/
	if (cpu == boot_cpuid)
		boot_cpuid = cpumask_any(cpu_online_mask);

	/* FIXME: abstract this to not be platform specific later on */
	xics_migrate_irqs_away();
	return 0;
}

/*
 * pseries_cpu_die: Wait for the cpu to die.
 * @cpu: logical processor id of the CPU whose death we're awaiting.
 *
 * This function is called from the context of the thread which is performing
 * the cpu-offline. Here we wait for long enough to allow the cpu in question
 * to self-destroy so that the cpu-offline thread can send the CPU_DEAD
 * notifications.
 *
 * OTOH, pseries_mach_cpu_die() is called by the @cpu when it wants to
 * self-destruct.
 */
static void pseries_cpu_die(unsigned int cpu)
{
	int tries;
	int cpu_status = 1;
	unsigned int pcpu = get_hard_smp_processor_id(cpu);

	if (get_preferred_offline_state(cpu) == CPU_STATE_INACTIVE) {
		cpu_status = 1;
		for (tries = 0; tries < 5000; tries++) {
			if (get_cpu_current_state(cpu) == CPU_STATE_INACTIVE) {
				cpu_status = 0;
				break;
			}
			msleep(1);
		}
	} else if (get_preferred_offline_state(cpu) == CPU_STATE_OFFLINE) {

		for (tries = 0; tries < 25; tries++) {
			cpu_status = smp_query_cpu_stopped(pcpu);
			if (cpu_status == QCSS_STOPPED ||
			    cpu_status == QCSS_HARDWARE_ERROR)
				break;
			cpu_relax();
		}
	}

	if (cpu_status != 0) {
		printk("Querying DEAD? cpu %i (%i) shows %i\n",
		       cpu, pcpu, cpu_status);
	}

	/* Isolation and deallocation are definitely done by
	 * drslot_chrp_cpu.  If they were not they would be
	 * done here.  Change isolate state to Isolate and
	 * change allocation-state to Unusable.
	 */
	paca[cpu].cpu_start = 0;
}

/*
 * Update cpu_present_mask and paca(s) for a new cpu node.  The wrinkle
 * here is that a cpu device node may represent up to two logical cpus
 * in the SMT case.  We must honor the assumption in other code that
 * the logical ids for sibling SMT threads x and y are adjacent, such
 * that x^1 == y and y^1 == x.
 */
static int pseries_add_processor(struct device_node *np)
{
	unsigned int cpu;
	cpumask_var_t candidate_mask, tmp;
	int err = -ENOSPC, len, nthreads, i;
	const __be32 *intserv;

	intserv = of_get_property(np, "ibm,ppc-interrupt-server#s", &len);
	if (!intserv)
		return 0;

	zalloc_cpumask_var(&candidate_mask, GFP_KERNEL);
	zalloc_cpumask_var(&tmp, GFP_KERNEL);

	nthreads = len / sizeof(u32);
	for (i = 0; i < nthreads; i++)
		cpumask_set_cpu(i, tmp);

	cpu_maps_update_begin();

	BUG_ON(!cpumask_subset(cpu_present_mask, cpu_possible_mask));

	/* Get a bitmap of unoccupied slots. */
	cpumask_xor(candidate_mask, cpu_possible_mask, cpu_present_mask);
	if (cpumask_empty(candidate_mask)) {
		/* If we get here, it most likely means that NR_CPUS is
		 * less than the partition's max processors setting.
		 */
		printk(KERN_ERR "Cannot add cpu %s; this system configuration"
		       " supports %d logical cpus.\n", np->full_name,
		       cpumask_weight(cpu_possible_mask));
		goto out_unlock;
	}

	while (!cpumask_empty(tmp))
		if (cpumask_subset(tmp, candidate_mask))
			/* Found a range where we can insert the new cpu(s) */
			break;
		else
			cpumask_shift_left(tmp, tmp, nthreads);

	if (cpumask_empty(tmp)) {
		printk(KERN_ERR "Unable to find space in cpu_present_mask for"
		       " processor %s with %d thread(s)\n", np->name,
		       nthreads);
		goto out_unlock;
	}

	for_each_cpu(cpu, tmp) {
		BUG_ON(cpu_present(cpu));
		set_cpu_present(cpu, true);
		set_hard_smp_processor_id(cpu, be32_to_cpu(*intserv++));
	}
	err = 0;
out_unlock:
	cpu_maps_update_done();
	free_cpumask_var(candidate_mask);
	free_cpumask_var(tmp);
	return err;
}

/*
 * Update the present map for a cpu node which is going away, and set
 * the hard id in the paca(s) to -1 to be consistent with boot time
 * convention for non-present cpus.
 */
static void pseries_remove_processor(struct device_node *np)
{
	unsigned int cpu;
	int len, nthreads, i;
	const __be32 *intserv;
	u32 thread;

	intserv = of_get_property(np, "ibm,ppc-interrupt-server#s", &len);
	if (!intserv)
		return;

	nthreads = len / sizeof(u32);

	cpu_maps_update_begin();
	for (i = 0; i < nthreads; i++) {
		thread = be32_to_cpu(intserv[i]);
		for_each_present_cpu(cpu) {
			if (get_hard_smp_processor_id(cpu) != thread)
				continue;
			BUG_ON(cpu_online(cpu));
			set_cpu_present(cpu, false);
			set_hard_smp_processor_id(cpu, -1);
			break;
		}
		if (cpu >= nr_cpu_ids)
			printk(KERN_WARNING "Could not find cpu to remove "
			       "with physical id 0x%x\n", thread);
	}
	cpu_maps_update_done();
}

static int pseries_smp_notifier(struct notifier_block *nb,
				unsigned long action, void *node)
{
	int err = 0;

	switch (action) {
	case OF_RECONFIG_ATTACH_NODE:
		err = pseries_add_processor(node);
		break;
	case OF_RECONFIG_DETACH_NODE:
		pseries_remove_processor(node);
		break;
	}
	return notifier_from_errno(err);
}

static struct notifier_block pseries_smp_nb = {
	.notifier_call = pseries_smp_notifier,
};

#define MAX_CEDE_LATENCY_LEVELS		4
#define	CEDE_LATENCY_PARAM_LENGTH	10
#define CEDE_LATENCY_PARAM_MAX_LENGTH	\
	(MAX_CEDE_LATENCY_LEVELS * CEDE_LATENCY_PARAM_LENGTH * sizeof(char))
#define CEDE_LATENCY_TOKEN		45

static char cede_parameters[CEDE_LATENCY_PARAM_MAX_LENGTH];

static int parse_cede_parameters(void)
{
	memset(cede_parameters, 0, CEDE_LATENCY_PARAM_MAX_LENGTH);
	return rtas_call(rtas_token("ibm,get-system-parameter"), 3, 1,
			 NULL,
			 CEDE_LATENCY_TOKEN,
			 __pa(cede_parameters),
			 CEDE_LATENCY_PARAM_MAX_LENGTH);
}

static int __init pseries_cpu_hotplug_init(void)
{
	struct device_node *np;
	const char *typep;
	int cpu;
	int qcss_tok;

	for_each_node_by_name(np, "interrupt-controller") {
		typep = of_get_property(np, "compatible", NULL);
		if (strstr(typep, "open-pic")) {
			of_node_put(np);

			printk(KERN_INFO "CPU Hotplug not supported on "
				"systems using MPIC\n");
			return 0;
		}
	}

	rtas_stop_self_token = rtas_token("stop-self");
	qcss_tok = rtas_token("query-cpu-stopped-state");

	if (rtas_stop_self_token == RTAS_UNKNOWN_SERVICE ||
			qcss_tok == RTAS_UNKNOWN_SERVICE) {
		printk(KERN_INFO "CPU Hotplug not supported by firmware "
				"- disabling.\n");
		return 0;
	}

	ppc_md.cpu_die = pseries_mach_cpu_die;
	smp_ops->cpu_disable = pseries_cpu_disable;
	smp_ops->cpu_die = pseries_cpu_die;

	/* Processors can be added/removed only on LPAR */
	if (firmware_has_feature(FW_FEATURE_LPAR)) {
		of_reconfig_notifier_register(&pseries_smp_nb);
		cpu_maps_update_begin();
		if (cede_offline_enabled && parse_cede_parameters() == 0) {
			default_offline_state = CPU_STATE_INACTIVE;
			for_each_online_cpu(cpu)
				set_default_offline_state(cpu);
		}
		cpu_maps_update_done();
	}

	return 0;
}
machine_arch_initcall(pseries, pseries_cpu_hotplug_init);
Commit	Line	Data
0332c2d4 ME	1	/*
	2	* pseries CPU Hotplug infrastructure.
	3	*
413f7c40 ME	4	* Split out from arch/powerpc/platforms/pseries/setup.c
413f7c40 ME	5	* arch/powerpc/kernel/rtas.c, and arch/powerpc/platforms/pseries/smp.c
0332c2d4 ME	6	*
	7	* Peter Bergner, IBM March 2001.
	8	* Copyright (C) 2001 IBM.
413f7c40 ME	9	* Dave Engebretsen, Peter Bergner, and
	10	* Mike Corrigan {engebret\|bergner\|mikec}@us.ibm.com
	11	* Plus various changes from other IBM teams...
0332c2d4 ME	12	*
	13	* Copyright (C) 2006 Michael Ellerman, IBM Corporation
	14	*
	15	* This program is free software; you can redistribute it and/or
	16	* modify it under the terms of the GNU General Public License
	17	* as published by the Free Software Foundation; either version
	18	* 2 of the License, or (at your option) any later version.
	19	*/
	20
	21	#include <linux/kernel.h>
0b05ac6e	22	#include <linux/interrupt.h>
0332c2d4	23	#include <linux/delay.h>
62fe91bb	24	#include <linux/sched.h> /* for idle_task_exit */
0332c2d4	25	#include <linux/cpu.h>
1cf3d8b3	26	#include <linux/of.h>
0332c2d4 ME	27	#include <asm/prom.h>
	28	#include <asm/rtas.h>
	29	#include <asm/firmware.h>
	30	#include <asm/machdep.h>
	31	#include <asm/vdso_datapage.h>
0b05ac6e	32	#include <asm/xics.h>
212bebb4 DD	33	#include <asm/plpar_wrappers.h>
212bebb4 DD	34
3aa565f5	35	#include "offline_states.h"
0332c2d4 ME	36
0332c2d4 ME	37	/* This version can't take the spinlock, because it never returns */
41dd03a9	38	static int rtas_stop_self_token = RTAS_UNKNOWN_SERVICE;
0332c2d4	39
3aa565f5 GS	40	static DEFINE_PER_CPU(enum cpu_state_vals, preferred_offline_state) =
	41	CPU_STATE_OFFLINE;
	42	static DEFINE_PER_CPU(enum cpu_state_vals, current_state) = CPU_STATE_OFFLINE;
	43
	44	static enum cpu_state_vals default_offline_state = CPU_STATE_OFFLINE;
	45
	46	static int cede_offline_enabled __read_mostly = 1;
	47
	48	/*
	49	* Enable/disable cede_offline when available.
	50	*/
	51	static int __init setup_cede_offline(char *str)
	52	{
	53	if (!strcmp(str, "off"))
	54	cede_offline_enabled = 0;
	55	else if (!strcmp(str, "on"))
	56	cede_offline_enabled = 1;
	57	else
	58	return 0;
	59	return 1;
	60	}
	61
	62	__setup("cede_offline=", setup_cede_offline);
	63
	64	enum cpu_state_vals get_cpu_current_state(int cpu)
	65	{
	66	return per_cpu(current_state, cpu);
	67	}
	68
	69	void set_cpu_current_state(int cpu, enum cpu_state_vals state)
	70	{
	71	per_cpu(current_state, cpu) = state;
	72	}
	73
	74	enum cpu_state_vals get_preferred_offline_state(int cpu)
	75	{
	76	return per_cpu(preferred_offline_state, cpu);
	77	}
	78
	79	void set_preferred_offline_state(int cpu, enum cpu_state_vals state)
	80	{
	81	per_cpu(preferred_offline_state, cpu) = state;
	82	}
	83
	84	void set_default_offline_state(int cpu)
	85	{
	86	per_cpu(preferred_offline_state, cpu) = default_offline_state;
	87	}
	88
04da6af9	89	static void rtas_stop_self(void)
0332c2d4	90	{
4fb8d027	91	static struct rtas_args args = {
41dd03a9	92	.nargs = 0,
e36d1227	93	.nret = cpu_to_be32(1),
41dd03a9 TB	94	.rets = &args.args[0],
41dd03a9 TB	95	};
0332c2d4	96
4fb8d027 LZ	97	args.token = cpu_to_be32(rtas_stop_self_token);
4fb8d027 LZ	98
0332c2d4 ME	99	local_irq_disable();
0332c2d4 ME	100
41dd03a9	101	BUG_ON(rtas_stop_self_token == RTAS_UNKNOWN_SERVICE);
0332c2d4 ME	102
	103	printk("cpu %u (hwid %u) Ready to die...\n",
	104	smp_processor_id(), hard_smp_processor_id());
41dd03a9	105	enter_rtas(__pa(&args));
0332c2d4 ME	106
	107	panic("Alas, I survived.\n");
	108	}
	109
06ba30b6	110	static void pseries_mach_cpu_die(void)
04da6af9	111	{
3aa565f5 GS	112	unsigned int cpu = smp_processor_id();
	113	unsigned int hwcpu = hard_smp_processor_id();
	114	u8 cede_latency_hint = 0;
	115
04da6af9 ME	116	local_irq_disable();
04da6af9 ME	117	idle_task_exit();
c3e8506c	118	xics_teardown_cpu();
3aa565f5 GS	119
	120	if (get_preferred_offline_state(cpu) == CPU_STATE_INACTIVE) {
	121	set_cpu_current_state(cpu, CPU_STATE_INACTIVE);
32d8ad4e BK	122	if (ppc_md.suspend_disable_cpu)
	123	ppc_md.suspend_disable_cpu();
	124
3aa565f5 GS	125	cede_latency_hint = 2;
	126
	127	get_lppaca()->idle = 1;
f13c13a0	128	if (!lppaca_shared_proc(get_lppaca()))
3aa565f5 GS	129	get_lppaca()->donate_dedicated_cpu = 1;
3aa565f5 GS	130
3aa565f5	131	while (get_preferred_offline_state(cpu) == CPU_STATE_INACTIVE) {
fb912568 LZ	132	while (!prep_irq_for_idle()) {
	133	local_irq_enable();
	134	local_irq_disable();
	135	}
	136
3aa565f5	137	extended_cede_processor(cede_latency_hint);
3aa565f5 GS	138	}
3aa565f5 GS	139
fb912568 LZ	140	local_irq_disable();
fb912568 LZ	141
f13c13a0	142	if (!lppaca_shared_proc(get_lppaca()))
3aa565f5 GS	143	get_lppaca()->donate_dedicated_cpu = 0;
3aa565f5 GS	144	get_lppaca()->idle = 0;
3aa565f5	145
0212f260	146	if (get_preferred_offline_state(cpu) == CPU_STATE_ONLINE) {
598c8231	147	unregister_slb_shadow(hwcpu);
3aa565f5	148
fb912568	149	hard_irq_disable();
0212f260 VS	150	/*
	151	* Call to start_secondary_resume() will not return.
	152	* Kernel stack will be reset and start_secondary()
	153	* will be called to continue the online operation.
	154	*/
	155	start_secondary_resume();
	156	}
	157	}
3aa565f5	158
0212f260 VS	159	/* Requested state is CPU_STATE_OFFLINE at this point */
0212f260 VS	160	WARN_ON(get_preferred_offline_state(cpu) != CPU_STATE_OFFLINE);
3aa565f5	161
0212f260	162	set_cpu_current_state(cpu, CPU_STATE_OFFLINE);
598c8231	163	unregister_slb_shadow(hwcpu);
0212f260	164	rtas_stop_self();
3aa565f5	165
04da6af9 ME	166	/* Should never get here... */
	167	BUG();
	168	for(;;);
	169	}
	170
06ba30b6	171	static int pseries_cpu_disable(void)
413f7c40 ME	172	{
	173	int cpu = smp_processor_id();
	174
ea0f1cab	175	set_cpu_online(cpu, false);
413f7c40 ME	176	vdso_data->processorCount--;
	177
	178	/fix boot_cpuid here/
	179	if (cpu == boot_cpuid)
8729faaa	180	boot_cpuid = cpumask_any(cpu_online_mask);
413f7c40 ME	181
	182	/* FIXME: abstract this to not be platform specific later on */
	183	xics_migrate_irqs_away();
	184	return 0;
	185	}
	186
3aa565f5 GS	187	/*
	188	* pseries_cpu_die: Wait for the cpu to die.
	189	* @cpu: logical processor id of the CPU whose death we're awaiting.
	190	*
	191	* This function is called from the context of the thread which is performing
	192	* the cpu-offline. Here we wait for long enough to allow the cpu in question
	193	* to self-destroy so that the cpu-offline thread can send the CPU_DEAD
	194	* notifications.
	195	*
	196	* OTOH, pseries_mach_cpu_die() is called by the @cpu when it wants to
	197	* self-destruct.
	198	*/
06ba30b6	199	static void pseries_cpu_die(unsigned int cpu)
413f7c40 ME	200	{
413f7c40 ME	201	int tries;
3aa565f5	202	int cpu_status = 1;
413f7c40 ME	203	unsigned int pcpu = get_hard_smp_processor_id(cpu);
413f7c40 ME	204
3aa565f5 GS	205	if (get_preferred_offline_state(cpu) == CPU_STATE_INACTIVE) {
3aa565f5 GS	206	cpu_status = 1;
940ce422	207	for (tries = 0; tries < 5000; tries++) {
3aa565f5 GS	208	if (get_cpu_current_state(cpu) == CPU_STATE_INACTIVE) {
	209	cpu_status = 0;
	210	break;
	211	}
940ce422	212	msleep(1);
3aa565f5 GS	213	}
	214	} else if (get_preferred_offline_state(cpu) == CPU_STATE_OFFLINE) {
	215
	216	for (tries = 0; tries < 25; tries++) {
f8b67691 MN	217	cpu_status = smp_query_cpu_stopped(pcpu);
	218	if (cpu_status == QCSS_STOPPED \|\|
	219	cpu_status == QCSS_HARDWARE_ERROR)
3aa565f5 GS	220	break;
	221	cpu_relax();
	222	}
413f7c40	223	}
3aa565f5	224
413f7c40 ME	225	if (cpu_status != 0) {
	226	printk("Querying DEAD? cpu %i (%i) shows %i\n",
	227	cpu, pcpu, cpu_status);
	228	}
	229
25985edc	230	/* Isolation and deallocation are definitely done by
413f7c40 ME	231	* drslot_chrp_cpu. If they were not they would be
	232	* done here. Change isolate state to Isolate and
	233	* change allocation-state to Unusable.
	234	*/
	235	paca[cpu].cpu_start = 0;
	236	}
	237
	238	/*
828a6986	239	* Update cpu_present_mask and paca(s) for a new cpu node. The wrinkle
413f7c40 ME	240	* here is that a cpu device node may represent up to two logical cpus
	241	* in the SMT case. We must honor the assumption in other code that
	242	* the logical ids for sibling SMT threads x and y are adjacent, such
	243	* that x^1 == y and y^1 == x.
	244	*/
06ba30b6	245	static int pseries_add_processor(struct device_node *np)
413f7c40 ME	246	{
413f7c40 ME	247	unsigned int cpu;
8729faaa	248	cpumask_var_t candidate_mask, tmp;
413f7c40	249	int err = -ENOSPC, len, nthreads, i;
d6f1e7ab	250	const __be32 *intserv;
413f7c40	251
e2eb6392	252	intserv = of_get_property(np, "ibm,ppc-interrupt-server#s", &len);
413f7c40 ME	253	if (!intserv)
	254	return 0;
	255
8729faaa AB	256	zalloc_cpumask_var(&candidate_mask, GFP_KERNEL);
	257	zalloc_cpumask_var(&tmp, GFP_KERNEL);
	258
413f7c40 ME	259	nthreads = len / sizeof(u32);
413f7c40 ME	260	for (i = 0; i < nthreads; i++)
8729faaa	261	cpumask_set_cpu(i, tmp);
413f7c40	262
86ef5c9a	263	cpu_maps_update_begin();
413f7c40	264
8729faaa	265	BUG_ON(!cpumask_subset(cpu_present_mask, cpu_possible_mask));
413f7c40 ME	266
413f7c40 ME	267	/* Get a bitmap of unoccupied slots. */
8729faaa AB	268	cpumask_xor(candidate_mask, cpu_possible_mask, cpu_present_mask);
8729faaa AB	269	if (cpumask_empty(candidate_mask)) {
413f7c40 ME	270	/* If we get here, it most likely means that NR_CPUS is
	271	* less than the partition's max processors setting.
	272	*/
	273	printk(KERN_ERR "Cannot add cpu %s; this system configuration"
	274	" supports %d logical cpus.\n", np->full_name,
8729faaa	275	cpumask_weight(cpu_possible_mask));
413f7c40 ME	276	goto out_unlock;
	277	}
	278
8729faaa AB	279	while (!cpumask_empty(tmp))
8729faaa AB	280	if (cpumask_subset(tmp, candidate_mask))
413f7c40 ME	281	/* Found a range where we can insert the new cpu(s) */
	282	break;
	283	else
8729faaa	284	cpumask_shift_left(tmp, tmp, nthreads);
413f7c40	285
8729faaa	286	if (cpumask_empty(tmp)) {
828a6986	287	printk(KERN_ERR "Unable to find space in cpu_present_mask for"
413f7c40 ME	288	" processor %s with %d thread(s)\n", np->name,
	289	nthreads);
	290	goto out_unlock;
	291	}
	292
8729faaa	293	for_each_cpu(cpu, tmp) {
104699c0	294	BUG_ON(cpu_present(cpu));
ea0f1cab	295	set_cpu_present(cpu, true);
d6f1e7ab	296	set_hard_smp_processor_id(cpu, be32_to_cpu(*intserv++));
413f7c40 ME	297	}
	298	err = 0;
	299	out_unlock:
86ef5c9a	300	cpu_maps_update_done();
8729faaa AB	301	free_cpumask_var(candidate_mask);
8729faaa AB	302	free_cpumask_var(tmp);
413f7c40 ME	303	return err;
	304	}
	305
	306	/*
	307	* Update the present map for a cpu node which is going away, and set
	308	* the hard id in the paca(s) to -1 to be consistent with boot time
	309	* convention for non-present cpus.
	310	*/
06ba30b6	311	static void pseries_remove_processor(struct device_node *np)
413f7c40 ME	312	{
	313	unsigned int cpu;
	314	int len, nthreads, i;
e36d1227 TF	315	const __be32 *intserv;
e36d1227 TF	316	u32 thread;
413f7c40	317
e2eb6392	318	intserv = of_get_property(np, "ibm,ppc-interrupt-server#s", &len);
413f7c40 ME	319	if (!intserv)
	320	return;
	321
	322	nthreads = len / sizeof(u32);
	323
86ef5c9a	324	cpu_maps_update_begin();
413f7c40	325	for (i = 0; i < nthreads; i++) {
e36d1227	326	thread = be32_to_cpu(intserv[i]);
413f7c40	327	for_each_present_cpu(cpu) {
e36d1227	328	if (get_hard_smp_processor_id(cpu) != thread)
413f7c40 ME	329	continue;
413f7c40 ME	330	BUG_ON(cpu_online(cpu));
ea0f1cab	331	set_cpu_present(cpu, false);
413f7c40 ME	332	set_hard_smp_processor_id(cpu, -1);
	333	break;
	334	}
8729faaa	335	if (cpu >= nr_cpu_ids)
413f7c40	336	printk(KERN_WARNING "Could not find cpu to remove "
e36d1227	337	"with physical id 0x%x\n", thread);
413f7c40	338	}
86ef5c9a	339	cpu_maps_update_done();
413f7c40 ME	340	}
413f7c40 ME	341
06ba30b6 ME	342	static int pseries_smp_notifier(struct notifier_block *nb,
06ba30b6 ME	343	unsigned long action, void *node)
413f7c40	344	{
de2780a3	345	int err = 0;
413f7c40 ME	346
413f7c40 ME	347	switch (action) {
1cf3d8b3	348	case OF_RECONFIG_ATTACH_NODE:
de2780a3	349	err = pseries_add_processor(node);
413f7c40	350	break;
1cf3d8b3	351	case OF_RECONFIG_DETACH_NODE:
06ba30b6	352	pseries_remove_processor(node);
413f7c40	353	break;
413f7c40	354	}
de2780a3	355	return notifier_from_errno(err);
413f7c40 ME	356	}
413f7c40 ME	357
06ba30b6 ME	358	static struct notifier_block pseries_smp_nb = {
06ba30b6 ME	359	.notifier_call = pseries_smp_notifier,
413f7c40 ME	360	};
413f7c40 ME	361
3aa565f5 GS	362	#define MAX_CEDE_LATENCY_LEVELS 4
	363	#define CEDE_LATENCY_PARAM_LENGTH 10
	364	#define CEDE_LATENCY_PARAM_MAX_LENGTH \
	365	(MAX_CEDE_LATENCY_LEVELS * CEDE_LATENCY_PARAM_LENGTH * sizeof(char))
	366	#define CEDE_LATENCY_TOKEN 45
	367
	368	static char cede_parameters[CEDE_LATENCY_PARAM_MAX_LENGTH];
	369
	370	static int parse_cede_parameters(void)
	371	{
3aa565f5	372	memset(cede_parameters, 0, CEDE_LATENCY_PARAM_MAX_LENGTH);
20a8ab97 AB	373	return rtas_call(rtas_token("ibm,get-system-parameter"), 3, 1,
	374	NULL,
	375	CEDE_LATENCY_TOKEN,
	376	__pa(cede_parameters),
	377	CEDE_LATENCY_PARAM_MAX_LENGTH);
3aa565f5 GS	378	}
3aa565f5 GS	379
0332c2d4 ME	380	static int __init pseries_cpu_hotplug_init(void)
0332c2d4 ME	381	{
64f27585 OJ	382	struct device_node *np;
64f27585 OJ	383	const char *typep;
3aa565f5	384	int cpu;
f8b67691	385	int qcss_tok;
64f27585 OJ	386
	387	for_each_node_by_name(np, "interrupt-controller") {
	388	typep = of_get_property(np, "compatible", NULL);
	389	if (strstr(typep, "open-pic")) {
	390	of_node_put(np);
	391
	392	printk(KERN_INFO "CPU Hotplug not supported on "
	393	"systems using MPIC\n");
	394	return 0;
	395	}
	396	}
	397
41dd03a9	398	rtas_stop_self_token = rtas_token("stop-self");
674fa677	399	qcss_tok = rtas_token("query-cpu-stopped-state");
0332c2d4	400
41dd03a9	401	if (rtas_stop_self_token == RTAS_UNKNOWN_SERVICE \|\|
674fa677 ME	402	qcss_tok == RTAS_UNKNOWN_SERVICE) {
	403	printk(KERN_INFO "CPU Hotplug not supported by firmware "
	404	"- disabling.\n");
	405	return 0;
	406	}
04da6af9	407
06ba30b6 ME	408	ppc_md.cpu_die = pseries_mach_cpu_die;
	409	smp_ops->cpu_disable = pseries_cpu_disable;
	410	smp_ops->cpu_die = pseries_cpu_die;
413f7c40 ME	411
413f7c40 ME	412	/* Processors can be added/removed only on LPAR */
3aa565f5	413	if (firmware_has_feature(FW_FEATURE_LPAR)) {
1cf3d8b3	414	of_reconfig_notifier_register(&pseries_smp_nb);
3aa565f5 GS	415	cpu_maps_update_begin();
	416	if (cede_offline_enabled && parse_cede_parameters() == 0) {
	417	default_offline_state = CPU_STATE_INACTIVE;
	418	for_each_online_cpu(cpu)
	419	set_default_offline_state(cpu);
	420	}
	421	cpu_maps_update_done();
	422	}
413f7c40	423
0332c2d4 ME	424	return 0;
0332c2d4 ME	425	}
d2a36071	426	machine_arch_initcall(pseries, pseries_cpu_hotplug_init);