CPU Hotplug support for X86_64

Index: linux-2.6.12-rc4-mm2/include/linux/init.h
===================================================================
@@ -222,6 +222,19 @@ void __init parse_early_param(void);
 #define __devinitdata
 #define __devexit
 #define __devexitdata
+
+#ifdef CONFIG_HOTPLUG_CPU
+#define __cpuinit
+#define __cpuinitdata
+#define __cpuexit
+#define __cpuexitdata
+#else
+#define __cpuinit	__init
+#define __cpuinitdata __initdata
+#define __cpuexit __exit
+#define __cpuexitdata	__exitdata
+#endif
+
 #else
 #define __devinit __init
 #define __devinitdata __initdata

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/smpboot.c
===================================================================
@@ -58,11 +58,6 @@
 #include <asm/proto.h>
 #include <asm/nmi.h>
 
-/* Change for real CPU hotplug. Note other files need to be fixed
-   first too. */
-#define __cpuinit __init
-#define __cpuinitdata __initdata
-
 /* Number of siblings per CPU package */
 int smp_num_siblings = 1;
 /* Package ID of each logical CPU */

@@ -822,7 +817,7 @@ static __cpuinit void smp_cleanup_boot(v
  *
  * RED-PEN audit/test this more. I bet there is more state messed up here.
  */
-static __cpuinit void disable_smp(void)
+static __init void disable_smp(void)
 {
 	cpu_present_map = cpumask_of_cpu(0);
 	cpu_possible_map = cpumask_of_cpu(0);

@@ -837,7 +832,7 @@ static __cpuinit void disable_smp(void)
 /*
  * Handle user cpus=... parameter.
  */
-static __cpuinit void enforce_max_cpus(unsigned max_cpus)
+static __init void enforce_max_cpus(unsigned max_cpus)
 {
 	int i, k;
 	k = 0;

@@ -854,7 +849,7 @@ static __cpuinit void enforce_max_cpus(u
 /*
  * Various sanity checks.
  */
-static int __cpuinit smp_sanity_check(unsigned max_cpus)
+static int __init smp_sanity_check(unsigned max_cpus)
 {
 	if (!physid_isset(hard_smp_processor_id(), phys_cpu_present_map)) {
 		printk("weird, boot CPU (#%d) not listed by the BIOS.\n",

@@ -912,7 +907,7 @@ static int __cpuinit smp_sanity_check(un
  * Prepare for SMP bootup.  The MP table or ACPI has been read
  * earlier.  Just do some sanity checking here and enable APIC mode.
  */
-void __cpuinit smp_prepare_cpus(unsigned int max_cpus)
+void __init smp_prepare_cpus(unsigned int max_cpus)
 {
 	int i;
 

@@ -1018,7 +1013,7 @@ int __cpuinit __cpu_up(unsigned int cpu)
 /*
  * Finish the SMP boot.
  */
-void __cpuinit smp_cpus_done(unsigned int max_cpus)
+void __init smp_cpus_done(unsigned int max_cpus)
 {
 	zap_low_mappings();
 	smp_cleanup_boot();

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/process.c
===================================================================
@@ -204,7 +204,7 @@ static void mwait_idle(void)
 	}
 }
 
-void __init select_idle_routine(const struct cpuinfo_x86 *c)
+void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
 {
 	static int printed;
 	if (cpu_has(c, X86_FEATURE_MWAIT)) {

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/setup.c
===================================================================
@@ -696,7 +696,7 @@ void __init setup_arch(char **cmdline_p)
 #endif
 }
 
-static int __init get_model_name(struct cpuinfo_x86 *c)
+static int __cpuinit get_model_name(struct cpuinfo_x86 *c)
 {
 	unsigned int *v;
 

@@ -712,7 +712,7 @@ static int __init get_model_name(struct 
 }
 
 
-static void __init display_cacheinfo(struct cpuinfo_x86 *c)
+static void __cpuinit display_cacheinfo(struct cpuinfo_x86 *c)
 {
 	unsigned int n, dummy, eax, ebx, ecx, edx;
 

@@ -823,7 +823,7 @@ static int __init init_amd(struct cpuinf
 	return r;
 }
 
-static void __init detect_ht(struct cpuinfo_x86 *c)
+static void __cpuinit detect_ht(struct cpuinfo_x86 *c)
 {
 #ifdef CONFIG_SMP
 	u32 	eax, ebx, ecx, edx;

@@ -884,7 +884,7 @@ static void __init detect_ht(struct cpui
 /*
  * find out the number of processor cores on the die
  */
-static int __init intel_num_cpu_cores(struct cpuinfo_x86 *c)
+static int __cpuinit intel_num_cpu_cores(struct cpuinfo_x86 *c)
 {
 	unsigned int eax;
 

@@ -902,7 +902,7 @@ static int __init intel_num_cpu_cores(st
 		return 1;
 }
 
-static void __init init_intel(struct cpuinfo_x86 *c)
+static void __cpuinit init_intel(struct cpuinfo_x86 *c)
 {
 	/* Cache sizes */
 	unsigned n;

@@ -922,7 +922,7 @@ static void __init init_intel(struct cpu
  	c->x86_num_cores = intel_num_cpu_cores(c);
 }
 
-void __init get_cpu_vendor(struct cpuinfo_x86 *c)
+void __cpuinit get_cpu_vendor(struct cpuinfo_x86 *c)
 {
 	char *v = c->x86_vendor_id;
 

@@ -943,7 +943,7 @@ struct cpu_model_info {
 /* Do some early cpuid on the boot CPU to get some parameter that are
    needed before check_bugs. Everything advanced is in identify_cpu
    below. */
-void __init early_identify_cpu(struct cpuinfo_x86 *c)
+void __cpuinit early_identify_cpu(struct cpuinfo_x86 *c)
 {
 	u32 tfms;
 

@@ -998,7 +998,7 @@ void __init early_identify_cpu(struct cp
 /*
  * This does the hard work of actually picking apart the CPU stuff...
  */
-void __init identify_cpu(struct cpuinfo_x86 *c)
+void __cpuinit identify_cpu(struct cpuinfo_x86 *c)
 {
 	int i;
 	u32 xlvl;

@@ -1075,7 +1075,7 @@ void __init identify_cpu(struct cpuinfo_
 }
  
 
-void __init print_cpu_info(struct cpuinfo_x86 *c)
+void __cpuinit print_cpu_info(struct cpuinfo_x86 *c)
 {
 	if (c->x86_model_id[0])
 		printk("%s", c->x86_model_id);

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/apic.c
===================================================================
@@ -321,7 +321,7 @@ void __init init_bsp_APIC(void)
 	apic_write_around(APIC_LVT1, value);
 }
 
-void __init setup_local_APIC (void)
+void __cpuinit setup_local_APIC (void)
 {
 	unsigned int value, ver, maxlvt;
 

@@ -570,7 +570,7 @@ static struct sys_device device_lapic = 
 	.cls		= &lapic_sysclass,
 };
 
-static void __init apic_pm_activate(void)
+static void __cpuinit apic_pm_activate(void)
 {
 	apic_pm_state.active = 1;
 }

@@ -810,14 +810,14 @@ void __init setup_boot_APIC_clock (void)
 	local_irq_enable();
 }
 
-void __init setup_secondary_APIC_clock(void)
+void __cpuinit setup_secondary_APIC_clock(void)
 {
 	local_irq_disable(); /* FIXME: Do we need this? --RR */
 	setup_APIC_timer(calibration_result);
 	local_irq_enable();
 }
 
-void __init disable_APIC_timer(void)
+void __cpuinit disable_APIC_timer(void)
 {
 	if (using_apic_timer) {
 		unsigned long v;

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/setup64.c
===================================================================
@@ -29,7 +29,7 @@
 
 char x86_boot_params[BOOT_PARAM_SIZE] __initdata = {0,};
 
-cpumask_t cpu_initialized __initdata = CPU_MASK_NONE;
+cpumask_t cpu_initialized __cpuinitdata = CPU_MASK_NONE;
 
 struct x8664_pda cpu_pda[NR_CPUS] __cacheline_aligned; 
 

@@ -171,7 +171,7 @@ void syscall_init(void)
 	wrmsrl(MSR_SYSCALL_MASK, EF_TF|EF_DF|EF_IE|0x3000); 
 }
 
-void __init check_efer(void)
+void __cpuinit check_efer(void)
 {
 	unsigned long efer;
 

@@ -188,7 +188,7 @@ void __init check_efer(void)
  * 'CPU state barrier', nothing should get across.
  * A lot of state is already set up in PDA init.
  */
-void __init cpu_init (void)
+void __cpuinit cpu_init (void)
 {
 #ifdef CONFIG_SMP
 	int cpu = stack_smp_processor_id();

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/i387.c
===================================================================
@@ -42,7 +42,7 @@ void mxcsr_feature_mask_init(void)
  * Called at bootup to set up the initial FPU state that is later cloned
  * into all processes.
  */
-void __init fpu_init(void)
+void __cpuinit fpu_init(void)
 {
 	unsigned long oldcr0 = read_cr0();
 	extern void __bad_fxsave_alignment(void);

Index: linux-2.6.12-rc4-mm2/mm/page_alloc.c
===================================================================
@@ -72,7 +72,7 @@ EXPORT_SYMBOL(zone_table);
 
 #ifdef CONFIG_NUMA
 static struct per_cpu_pageset
-	pageset_table[MAX_NR_ZONES*MAX_NUMNODES*NR_CPUS] __initdata;
+	pageset_table[MAX_NR_ZONES*MAX_NUMNODES*NR_CPUS] __cpuinitdata;
 #endif
 
 static char *zone_names[MAX_NR_ZONES] = { "DMA", "Normal", "HighMem" };

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/nmi.c
===================================================================
@@ -98,7 +98,7 @@ static unsigned int nmi_p4_cccr_val;
 	(P4_CCCR_OVF_PMI0|P4_CCCR_THRESHOLD(15)|P4_CCCR_COMPLEMENT|	\
 	 P4_CCCR_COMPARE|P4_CCCR_REQUIRED|P4_CCCR_ESCR_SELECT(4)|P4_CCCR_ENABLE)
 
-static __init inline int nmi_known_cpu(void)
+static __cpuinit inline int nmi_known_cpu(void)
 {
 	switch (boot_cpu_data.x86_vendor) {
 	case X86_VENDOR_AMD:

@@ -110,7 +110,7 @@ static __init inline int nmi_known_cpu(v
 }
 
 /* Run after command line and cpu_init init, but before all other checks */
-void __init nmi_watchdog_default(void)
+void __cpuinit nmi_watchdog_default(void)
 {
 	if (nmi_watchdog != NMI_DEFAULT)
 		return;

Index: linux-2.6.12-rc4-mm2/arch/x86_64/mm/numa.c
===================================================================
@@ -243,7 +243,7 @@ void __init numa_initmem_init(unsigned l
 	setup_node_bootmem(0, start_pfn << PAGE_SHIFT, end_pfn << PAGE_SHIFT);
 }
 
-__init void numa_add_cpu(int cpu)
+__cpuinit void numa_add_cpu(int cpu)
 {
 	/* BP is initialized elsewhere */
 	if (cpu) 

332 lines not on this page · raw message

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/irq.c
===================================================================
@@ -14,6 +14,7 @@
 #include <linux/interrupt.h>
 #include <linux/seq_file.h>
 #include <linux/module.h>
+#include <linux/delay.h>
 #include <asm/uaccess.h>
 #include <asm/io_apic.h>
 

@@ -106,3 +107,32 @@ asmlinkage unsigned int do_IRQ(struct pt
 
 	return 1;
 }
+
+#ifdef CONFIG_HOTPLUG_CPU
+void fixup_irqs(cpumask_t map)
+{
+	unsigned int irq;
+	static int warned;
+
+	for (irq = 0; irq < NR_IRQS; irq++) {
+		cpumask_t mask;
+		if (irq == 2)
+			continue;
+
+		cpus_and(mask, irq_affinity[irq], map);
+		if (any_online_cpu(mask) == NR_CPUS) {
+			printk("Breaking affinity for irq %i\n", irq);
+			mask = map;
+		}
+		if (irq_desc[irq].handler->set_affinity)
+			irq_desc[irq].handler->set_affinity(irq, mask);
+		else if (irq_desc[irq].action && !(warned++))
+			printk("Cannot set affinity for irq %i\n", irq);
+	}
+
+	/* That doesn't seem sufficient.  Give it 1ms. */
+	local_irq_enable();
+	mdelay(1);
+	local_irq_disable();
+}
+#endif

Index: linux-2.6.12-rc4-mm2/include/asm-x86_64/irq.h
===================================================================
@@ -52,4 +52,9 @@ struct irqaction;
 struct pt_regs;
 int handle_IRQ_event(unsigned int, struct pt_regs *, struct irqaction *);
 
+#ifdef CONFIG_HOTPLUG_CPU
+#include <linux/cpumask.h>
+extern void fixup_irqs(cpumask_t map);
+#endif
+
 #endif /* _ASM_IRQ_H */

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/process.c
===================================================================
@@ -8,7 +8,8 @@
  * 
  *  X86-64 port
  *	Andi Kleen.
- * 
+ *
+ *	CPU hotplug support - [email protected]
  *  $Id: process.c,v 1.38 2002/01/15 10:08:03 ak Exp $
  */
 

@@ -18,6 +19,7 @@
 
 #include <stdarg.h>
 
+#include <linux/cpu.h>
 #include <linux/errno.h>
 #include <linux/sched.h>
 #include <linux/kernel.h>

@@ -154,6 +156,33 @@ void cpu_idle_wait(void)
 }
 EXPORT_SYMBOL_GPL(cpu_idle_wait);
 
+#ifdef CONFIG_HOTPLUG_CPU
+DECLARE_PER_CPU(int, cpu_state);
+
+#include <asm/nmi.h>
+/* We don't actually take CPU down, just spin without interrupts. */
+static inline void play_dead(void)
+{
+	idle_task_exit();
+	mb();
+	/* Ack it */
+	__get_cpu_var(cpu_state) = CPU_DEAD;
+
+	local_irq_disable();
+	while (1)
+		safe_halt();
+#if 0
+	while (1)
+		__asm__ __volatile__("hlt":::"memory");
+#endif
+}
+#else
+static inline void play_dead(void)
+{
+	BUG();
+}
+#endif /* CONFIG_HOTPLUG_CPU */
+
 /*
  * The idle thread. There's no useful work to be
  * done, so just try to conserve power and have a

@@ -174,6 +203,8 @@ void cpu_idle (void)
 			idle = pm_idle;
 			if (!idle)
 				idle = default_idle;
+			if (cpu_is_offline(smp_processor_id()))
+				play_dead();
 			idle();
 		}
 

Index: linux-2.6.12-rc4-mm2/include/asm-x86_64/smp.h
===================================================================
@@ -44,6 +44,8 @@ extern void smp_alloc_memory(void);
 extern volatile unsigned long smp_invalidate_needed;
 extern int pic_mode;
 extern int smp_num_siblings;
+extern void lock_ipi_calllock(void);
+extern void unlock_ipi_calllock(void);
 extern void smp_flush_tlb(void);
 extern void smp_message_irq(int cpl, void *dev_id, struct pt_regs *regs);
 extern void smp_send_reschedule(int cpu);

@@ -77,6 +79,8 @@ extern __inline int hard_smp_processor_i
 }
 
 extern int safe_smp_processor_id(void);
+extern int __cpu_disable(void);
+extern void __cpu_die(unsigned int cpu);
 
 #endif /* !ASSEMBLY */
 

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/traps.c
===================================================================
@@ -587,11 +587,18 @@ static void unknown_nmi_error(unsigned c
 asmlinkage void default_do_nmi(struct pt_regs *regs)
 {
 	unsigned char reason = 0;
+	int cpu;
+
+	cpu = smp_processor_id();
 
 	/* Only the BSP gets external NMIs from the system.  */
-	if (!smp_processor_id())
+	if (!cpu)
 		reason = get_nmi_reason();
 
+#ifdef CONFIG_HOTPLUG_CPU
+	if (!cpu_online(cpu))
+		return;
+#endif
 	if (!(reason & 0xc0)) {
 		if (notify_die(DIE_NMI_IPI, "nmi_ipi", regs, reason, 0, SIGINT)
 								== NOTIFY_STOP)

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/smp.c
===================================================================
@@ -295,6 +295,18 @@ struct call_data_struct {
 
 static struct call_data_struct * call_data;
 
+void
+lock_ipi_calllock(void)
+{
+	spin_lock_irq(&call_lock);
+}
+
+void
+unlock_ipi_calllock(void)
+{
+	spin_unlock_irq(&call_lock);
+}
+
 /*
  * this function sends a 'generic call function' IPI to all other CPUs
  * in the system.

Index: linux-2.6.12-rc4-mm2/arch/i386/mach-default/topology.c
===================================================================
@@ -88,8 +88,11 @@ static int __init topology_init(void)
 {
 	int i;
 
-	for (i = 0; i < NR_CPUS; i++)
-		if (cpu_possible(i)) arch_register_cpu(i);
+	for (i = 0; i < NR_CPUS; i++) {
+		if (cpu_possible(i))  {
+			arch_register_cpu(i);
+		}
+	}
 	return 0;
 }
 

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/smpboot.c
===================================================================
@@ -62,9 +62,12 @@
 /* Number of siblings per CPU package */
 int smp_num_siblings = 1;
 /* Package ID of each logical CPU */
-u8 phys_proc_id[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
-u8 cpu_core_id[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
+u8 phys_proc_id[NR_CPUS] __cacheline_aligned = 
+	{ [0 ... NR_CPUS-1] = BAD_APICID };
 EXPORT_SYMBOL(phys_proc_id);
+
+u8 cpu_core_id[NR_CPUS] __cacheline_aligned = 
+	{ [0 ... NR_CPUS-1] = BAD_APICID };
 EXPORT_SYMBOL(cpu_core_id);
 
 /* Bitmask of currently online CPUs */

@@ -434,6 +437,34 @@ void __cpuinit smp_callin(void)
 	local_flush_tlb();
 }
 
+static inline void
+set_cpu_sibling_map(int cpu)
+{
+	int i;
+
+	if (smp_num_siblings > 1) {
+		for_each_online_cpu(i) {
+			if (cpu_core_id[cpu] == cpu_core_id[i]) {
+				cpu_set(i, cpu_sibling_map[cpu]);
+				cpu_set(cpu, cpu_sibling_map[i]);
+			}
+		}
+	} else {
+		cpu_set(cpu, cpu_sibling_map[cpu]);
+	}
+
+	if (current_cpu_data.x86_num_cores > 1) {
+		for_each_online_cpu(i) {
+			if (phys_proc_id[cpu] == phys_proc_id[i]) {
+				cpu_set(i, cpu_core_map[cpu]);
+				cpu_set(cpu, cpu_core_map[i]);
+			}
+		}
+	} else {
+		cpu_core_map[cpu] = cpu_sibling_map[cpu];
+	}
+}
+
 /*
  * Setup code on secondary processor (after comming out of the trampoline)
  */

@@ -463,6 +494,12 @@ void __cpuinit start_secondary(void)
 
 	enable_APIC_timer();
 
+	/* 
+	 * The sibling maps must be set before turing the online map on for 
+	 * this cpu 
+	 */
+	set_cpu_sibling_map(smp_processor_id());
+
 	/*
 	 * Allow the master to continue.
 	 */

@@ -798,51 +835,6 @@ cycles_t cacheflush_time;
 unsigned long cache_decay_ticks;
 
 /*
- * Construct cpu_sibling_map[], so that we can tell the sibling CPU
- * on SMT systems efficiently.
- */
-static __cpuinit void detect_siblings(void)
-{
-	int cpu;
-
-	for (cpu = 0; cpu < NR_CPUS; cpu++) {
-		cpus_clear(cpu_sibling_map[cpu]);
-		cpus_clear(cpu_core_map[cpu]);
-	}
-
-	for_each_online_cpu (cpu) {
-		struct cpuinfo_x86 *c = cpu_data + cpu;
-		int siblings = 0;
-		int i;
-		if (smp_num_siblings > 1) {
-			for_each_online_cpu (i) {
-				if (cpu_core_id[cpu] == cpu_core_id[i]) {
-					siblings++;
-					cpu_set(i, cpu_sibling_map[cpu]);
-				}
-			}
-		} else {
-			siblings++;
-			cpu_set(cpu, cpu_sibling_map[cpu]);
-		}
-
-		if (siblings != smp_num_siblings) {
-			printk(KERN_WARNING
-	       "WARNING: %d siblings found for CPU%d, should be %d\n",
-			       siblings, cpu, smp_num_siblings);
-			smp_num_siblings = siblings;
-		}
-		if (c->x86_num_cores > 1) {
-			for_each_online_cpu(i) {
-				if (phys_proc_id[cpu] == phys_proc_id[i])
-					cpu_set(i, cpu_core_map[cpu]);
-			}
-		} else
-			cpu_core_map[cpu] = cpu_sibling_map[cpu];
-	}
-}
-
-/*
  * Cleanup possible dangling ends...
  */
 static __cpuinit void smp_cleanup_boot(void)

@@ -1036,6 +1028,8 @@ void __init smp_prepare_boot_cpu(void)
 	int me = smp_processor_id();
 	cpu_set(me, cpu_online_map);
 	cpu_set(me, cpu_callout_map);
+	cpu_set(0, cpu_sibling_map[0]);
+	cpu_set(0, cpu_core_map[0]);
 }
 
 /*

@@ -1100,7 +1094,6 @@ void __init smp_cpus_done(unsigned int m
 	setup_ioapic_dest();
 #endif
 
-	detect_siblings();
 	time_init_gtod();
 
 	check_nmi_watchdog();

Index: linux-2.6.12-rc4-mm2/arch/x86_64/kernel/smp.c
===================================================================
@@ -43,6 +43,13 @@ static cpumask_t flush_cpumask;
 static struct mm_struct * flush_mm;
 static unsigned long flush_va;
 static DEFINE_SPINLOCK(tlbstate_lock);
+
+#ifdef CONFIG_SAFE_IPI_CALLFUNC
+static int safe_ipi_call = 1;
+#else
+static int safe_ipi_call = 0;
+#endif
+
 #define FLUSH_ALL	-1ULL
 
 /*

@@ -307,6 +314,15 @@ unlock_ipi_calllock(void)
 	spin_unlock_irq(&call_lock);
 }
 
+static int __init safe_ipi_setup(char *s)
+{ 
+     safe_ipi_call = 1; 
+	 printk ("Turning off broadcast IPI in smp_call_function\n");
+     return 0; 
+} 
+
+__setup("safe_ipi", safe_ipi_setup); 
+
 /*
  * this function sends a 'generic call function' IPI to all other CPUs
  * in the system.

@@ -330,7 +346,13 @@ static void __smp_call_function (void (*
 	call_data = &data;
 	wmb();
 	/* Send a message to all other CPUs and wait for them to respond */
-	send_IPI_allbutself(CALL_FUNCTION_VECTOR);
+
+	if (safe_ipi_call) {
+		cpumask_t cpu_mask = cpu_online_map;
+		cpu_clear(smp_processor_id(), cpu_mask);
+		send_IPI_mask(cpu_mask, CALL_FUNCTION_VECTOR);
+	} else
+		send_IPI_allbutself(CALL_FUNCTION_VECTOR);
 
 	/* Wait for response */
 	while (atomic_read(&data.started) != cpus)