12 files changed, 184 insertions, 19 deletions

diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index a85ea9d37f05..ff7ae6b664e8 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -271,6 +271,25 @@ config SCHED_SMT
           Intel IA64 chips with MultiThreading at a cost of slightly increased
           overhead in some places. If unsure say N here.
 
+config PERMIT_BSP_REMOVE
+        bool "Support removal of Bootstrap Processor"
+        depends on HOTPLUG_CPU
+        default n
+        ---help---
+        Say Y here if your platform SAL will support removal of BSP with HOTPLUG_CPU
+        support. 
+
+config FORCE_CPEI_RETARGET
+        bool "Force assumption that CPEI can be re-targetted"
+        depends on PERMIT_BSP_REMOVE
+        default n
+        ---help---
+        Say Y if you need to force the assumption that CPEI can be re-targetted to
+        any cpu in the system. This hint is available via ACPI 3.0 specifications.
+        Tiger4 systems are capable of re-directing CPEI to any CPU other than BSP.
+        This option it useful to enable this feature on older BIOS's as well.
+        You can also enable this by using boot command line option force_cpei=1.
+
 config PREEMPT
         bool "Preemptible Kernel"
         help
diff --git a/arch/ia64/configs/tiger_defconfig b/arch/ia64/configs/tiger_defconfig
index 125568118b84..766bf4955432 100644
--- a/arch/ia64/configs/tiger_defconfig
+++ b/arch/ia64/configs/tiger_defconfig
@@ -116,6 +116,8 @@ CONFIG_FORCE_MAX_ZONEORDER=17
 CONFIG_SMP=y
 CONFIG_NR_CPUS=4
 CONFIG_HOTPLUG_CPU=y
+CONFIG_PERMIT_BSP_REMOVE=y
+CONFIG_FORCE_CPEI_RETARGET=y
 # CONFIG_SCHED_SMT is not set
 # CONFIG_PREEMPT is not set
 CONFIG_SELECT_MEMORY_MODEL=y
diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c
index ecd44bdc8394..4722ec51c70c 100644
--- a/arch/ia64/kernel/acpi.c
+++ b/arch/ia64/kernel/acpi.c
@@ -284,19 +284,24 @@ acpi_parse_plat_int_src(acpi_table_entry_header * header,
         return 0;
 }
 
+#ifdef CONFIG_HOTPLUG_CPU
 unsigned int can_cpei_retarget(void)
 {
         extern int cpe_vector;
+        extern unsigned int force_cpei_retarget;
 
         /*
          * Only if CPEI is supported and the override flag
          * is present, otherwise return that its re-targettable
          * if we are in polling mode.
          */
-        if (cpe_vector > 0 && !acpi_cpei_override)
-                return 0;
-        else
-                return 1;
+        if (cpe_vector > 0) {
+                if (acpi_cpei_override || force_cpei_retarget)
+                        return 1;
+                else
+                        return 0;
+        }
+        return 1;
 }
 
 unsigned int is_cpu_cpei_target(unsigned int cpu)
@@ -315,6 +320,7 @@ void set_cpei_target_cpu(unsigned int cpu)
 {
         acpi_cpei_phys_cpuid = cpu_physical_id(cpu);
 }
+#endif
 
 unsigned int get_cpei_target_cpu(void)
 {
diff --git a/arch/ia64/kernel/iosapic.c b/arch/ia64/kernel/iosapic.c
index 574084f343fa..8832c553230a 100644
--- a/arch/ia64/kernel/iosapic.c
+++ b/arch/ia64/kernel/iosapic.c
@@ -631,6 +631,7 @@ get_target_cpu (unsigned int gsi, int vector)
 {
 #ifdef CONFIG_SMP
         static int cpu = -1;
+        extern int cpe_vector;
 
         /*
          * In case of vector shared by multiple RTEs, all RTEs that
@@ -653,6 +654,11 @@ get_target_cpu (unsigned int gsi, int vector)
         if (!cpu_online(smp_processor_id()))
                 return cpu_physical_id(smp_processor_id());
 
+#ifdef CONFIG_ACPI
+        if (cpe_vector > 0 && vector == IA64_CPEP_VECTOR)
+                return get_cpei_target_cpu();
+#endif
+
 #ifdef CONFIG_NUMA
         {
                 int num_cpus, cpu_index, iosapic_index, numa_cpu, i = 0;
diff --git a/arch/ia64/kernel/irq.c b/arch/ia64/kernel/irq.c
index d33244c32759..5ce908ef9c95 100644
--- a/arch/ia64/kernel/irq.c
+++ b/arch/ia64/kernel/irq.c
@@ -163,8 +163,19 @@ void fixup_irqs(void)
 {
         unsigned int irq;
         extern void ia64_process_pending_intr(void);
+        extern void ia64_disable_timer(void);
+        extern volatile int time_keeper_id;
+
+        ia64_disable_timer();
+
+        /*
+         * Find a new timesync master
+         */
+        if (smp_processor_id() == time_keeper_id) {
+                time_keeper_id = first_cpu(cpu_online_map);
+                printk ("CPU %d is now promoted to time-keeper master\n", time_keeper_id);
+        }
 
-        ia64_set_itv(1<<16);
         /*
          * Phase 1: Locate irq's bound to this cpu and
          * relocate them for cpu removal.
diff --git a/arch/ia64/kernel/mca.c b/arch/ia64/kernel/mca.c
index ee7eec9ee576..87fb7cecead0 100644
--- a/arch/ia64/kernel/mca.c
+++ b/arch/ia64/kernel/mca.c
@@ -289,6 +289,7 @@ ia64_mca_log_sal_error_record(int sal_info_type)
 #ifdef CONFIG_ACPI
 
 int cpe_vector = -1;
+int ia64_cpe_irq = -1;
 
 static irqreturn_t
 ia64_mca_cpe_int_handler (int cpe_irq, void *arg, struct pt_regs *ptregs)
@@ -1444,11 +1445,13 @@ void __devinit
 ia64_mca_cpu_init(void *cpu_data)
 {
         void *pal_vaddr;
+        static int first_time = 1;
 
-        if (smp_processor_id() == 0) {
+        if (first_time) {
                 void *mca_data;
                 int cpu;
 
+                first_time = 0;
                 mca_data = alloc_bootmem(sizeof(struct ia64_mca_cpu)
                                          * NR_CPUS + KERNEL_STACK_SIZE);
                 mca_data = (void *)(((unsigned long)mca_data +
@@ -1704,6 +1707,7 @@ ia64_mca_late_init(void)
                                         desc = irq_descp(irq);
                                         desc->status |= IRQ_PER_CPU;
                                         setup_irq(irq, &mca_cpe_irqaction);
+                                        ia64_cpe_irq = irq;
                                 }
                         ia64_mca_register_cpev(cpe_vector);
                         IA64_MCA_DEBUG("%s: CPEI/P setup and enabled.\n", __FUNCTION__);
diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c
index 9c5194b385da..077f21216b65 100644
--- a/arch/ia64/kernel/perfmon.c
+++ b/arch/ia64/kernel/perfmon.c
@@ -6722,6 +6722,7 @@ __initcall(pfm_init);
 void
 pfm_init_percpu (void)
 {
+        static int first_time=1;
         /*
          * make sure no measurement is active
          * (may inherit programmed PMCs from EFI).
@@ -6734,8 +6735,10 @@ pfm_init_percpu (void)
          */
         pfm_unfreeze_pmu();
 
-        if (smp_processor_id() == 0)
+        if (first_time) {
                 register_percpu_irq(IA64_PERFMON_VECTOR, &perfmon_irqaction);
+                first_time=0;
+        }
 
         ia64_setreg(_IA64_REG_CR_PMV, IA64_PERFMON_VECTOR);
         ia64_srlz_d();
diff --git a/arch/ia64/kernel/smpboot.c b/arch/ia64/kernel/smpboot.c
index b681ef34a86e..c4b633b36dab 100644
--- a/arch/ia64/kernel/smpboot.c
+++ b/arch/ia64/kernel/smpboot.c
@@ -70,6 +70,12 @@
 #endif
 
 #ifdef CONFIG_HOTPLUG_CPU
+#ifdef CONFIG_PERMIT_BSP_REMOVE
+#define bsp_remove_ok   1
+#else
+#define bsp_remove_ok   0
+#endif
+
 /*
  * Store all idle threads, this can be reused instead of creating
  * a new thread. Also avoids complicated thread destroy functionality
@@ -104,7 +110,7 @@ struct sal_to_os_boot *sal_state_for_booting_cpu = &sal_boot_rendez_state[0];
 /*
  * ITC synchronization related stuff:
  */
-#define MASTER  0
+#define MASTER  (0)
 #define SLAVE   (SMP_CACHE_BYTES/8)
 
 #define NUM_ROUNDS      64      /* magic value */
@@ -151,6 +157,27 @@ char __initdata no_int_routing;
 
 unsigned char smp_int_redirect; /* are INT and IPI redirectable by the chipset? */
 
+#ifdef CONFIG_FORCE_CPEI_RETARGET
+#define CPEI_OVERRIDE_DEFAULT   (1)
+#else
+#define CPEI_OVERRIDE_DEFAULT   (0)
+#endif
+
+unsigned int force_cpei_retarget = CPEI_OVERRIDE_DEFAULT;
+
+static int __init
+cmdl_force_cpei(char *str)
+{
+        int value=0;
+
+        get_option (&str, &value);
+        force_cpei_retarget = value;
+
+        return 1;
+}
+
+__setup("force_cpei=", cmdl_force_cpei);
+
 static int __init
 nointroute (char *str)
 {
@@ -161,6 +188,27 @@ nointroute (char *str)
 
 __setup("nointroute", nointroute);
 
+static void fix_b0_for_bsp(void)
+{
+#ifdef CONFIG_HOTPLUG_CPU
+        int cpuid;
+        static int fix_bsp_b0 = 1;
+
+        cpuid = smp_processor_id();
+
+        /*
+         * Cache the b0 value on the first AP that comes up
+         */
+        if (!(fix_bsp_b0 && cpuid))
+                return;
+
+        sal_boot_rendez_state[0].br[0] = sal_boot_rendez_state[cpuid].br[0];
+        printk ("Fixed BSP b0 value from CPU %d\n", cpuid);
+
+        fix_bsp_b0 = 0;
+#endif
+}
+
 void
 sync_master (void *arg)
 {
@@ -327,8 +375,9 @@ smp_setup_percpu_timer (void)
 static void __devinit
 smp_callin (void)
 {
-        int cpuid, phys_id;
+        int cpuid, phys_id, itc_master;
         extern void ia64_init_itm(void);
+        extern volatile int time_keeper_id;
 
 #ifdef CONFIG_PERFMON
         extern void pfm_init_percpu(void);
@@ -336,6 +385,7 @@ smp_callin (void)
 
         cpuid = smp_processor_id();
         phys_id = hard_smp_processor_id();
+        itc_master = time_keeper_id;
 
         if (cpu_online(cpuid)) {
                 printk(KERN_ERR "huh, phys CPU#0x%x, CPU#0x%x already present??\n",
@@ -343,6 +393,8 @@ smp_callin (void)
                 BUG();
         }
 
+        fix_b0_for_bsp();
+
         lock_ipi_calllock();
         cpu_set(cpuid, cpu_online_map);
         unlock_ipi_calllock();
@@ -365,8 +417,8 @@ smp_callin (void)
                  * calls spin_unlock_bh(), which calls spin_unlock_bh(), which calls
                  * local_bh_enable(), which bugs out if irqs are not enabled...
                  */
-                Dprintk("Going to syncup ITC with BP.\n");
-                ia64_sync_itc(0);
+                Dprintk("Going to syncup ITC with ITC Master.\n");
+                ia64_sync_itc(itc_master);
         }
 
         /*
@@ -635,6 +687,47 @@ remove_siblinginfo(int cpu)
 }
 
 extern void fixup_irqs(void);
+
+int migrate_platform_irqs(unsigned int cpu)
+{
+        int new_cpei_cpu;
+        irq_desc_t *desc = NULL;
+        cpumask_t       mask;
+        int             retval = 0;
+
+        /*
+         * dont permit CPEI target to removed.
+         */
+        if (cpe_vector > 0 && is_cpu_cpei_target(cpu)) {
+                printk ("CPU (%d) is CPEI Target\n", cpu);
+                if (can_cpei_retarget()) {
+                        /*
+                         * Now re-target the CPEI to a different processor
+                         */
+                        new_cpei_cpu = any_online_cpu(cpu_online_map);
+                        mask = cpumask_of_cpu(new_cpei_cpu);
+                        set_cpei_target_cpu(new_cpei_cpu);
+                        desc = irq_descp(ia64_cpe_irq);
+                        /*
+                         * Switch for now, immediatly, we need to do fake intr
+                         * as other interrupts, but need to study CPEI behaviour with
+                         * polling before making changes.
+                         */
+                        if (desc) {
+                                desc->handler->disable(ia64_cpe_irq);
+                                desc->handler->set_affinity(ia64_cpe_irq, mask);
+                                desc->handler->enable(ia64_cpe_irq);
+                                printk ("Re-targetting CPEI to cpu %d\n", new_cpei_cpu);
+                        }
+                }
+                if (!desc) {
+                        printk ("Unable to retarget CPEI, offline cpu [%d] failed\n", cpu);
+                        retval = -EBUSY;
+                }
+        }
+        return retval;
+}
+
 /* must be called with cpucontrol mutex held */
 int __cpu_disable(void)
 {
@@ -643,8 +736,17 @@ int __cpu_disable(void)
         /*
          * dont permit boot processor for now
          */
-        if (cpu == 0)
-                return -EBUSY;
+        if (cpu == 0 && !bsp_remove_ok) {
+                printk ("Your platform does not support removal of BSP\n");
+                return (-EBUSY);
+        }
+
+        cpu_clear(cpu, cpu_online_map);
+
+        if (migrate_platform_irqs(cpu)) {
+                cpu_set(cpu, cpu_online_map);
+                return (-EBUSY);
+        }
 
         remove_siblinginfo(cpu);
         cpu_clear(cpu, cpu_online_map);
diff --git a/arch/ia64/kernel/time.c b/arch/ia64/kernel/time.c
index 307d01e15b2e..ac167436e936 100644
--- a/arch/ia64/kernel/time.c
+++ b/arch/ia64/kernel/time.c
@@ -32,7 +32,7 @@
 
 extern unsigned long wall_jiffies;
 
-#define TIME_KEEPER_ID  0       /* smp_processor_id() of time-keeper */
+volatile int time_keeper_id = 0; /* smp_processor_id() of time-keeper */
 
 #ifdef CONFIG_IA64_DEBUG_IRQ
 
@@ -71,7 +71,7 @@ timer_interrupt (int irq, void *dev_id, struct pt_regs *regs)
 
                 new_itm += local_cpu_data->itm_delta;
 
-                if (smp_processor_id() == TIME_KEEPER_ID) {
+                if (smp_processor_id() == time_keeper_id) {
                         /*
                          * Here we are in the timer irq handler. We have irqs locally
                          * disabled, but we don't know if the timer_bh is running on
@@ -236,6 +236,11 @@ static struct irqaction timer_irqaction = {
         .name =         "timer"
 };
 
+void __devinit ia64_disable_timer(void)
+{
+        ia64_set_itv(1 << 16);
+}
+
 void __init
 time_init (void)
 {
diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index 6e5eea19fa67..3b6fd798c4d6 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -36,7 +36,7 @@ int arch_register_cpu(int num)
         parent = &sysfs_nodes[cpu_to_node(num)];
 #endif /* CONFIG_NUMA */
 
-#ifdef CONFIG_ACPI
+#if defined (CONFIG_ACPI) && defined (CONFIG_HOTPLUG_CPU)
         /*
          * If CPEI cannot be re-targetted, and this is
          * CPEI target, then dont create the control file
diff --git a/arch/ia64/mm/contig.c b/arch/ia64/mm/contig.c
index acaaec4e4681..9855ba318094 100644
--- a/arch/ia64/mm/contig.c
+++ b/arch/ia64/mm/contig.c
@@ -181,13 +181,15 @@ per_cpu_init (void)
 {
         void *cpu_data;
         int cpu;
+        static int first_time=1;
 
         /*
          * get_free_pages() cannot be used before cpu_init() done.  BSP
          * allocates "NR_CPUS" pages for all CPUs to avoid that AP calls
          * get_zeroed_page().
          */
-        if (smp_processor_id() == 0) {
+        if (first_time) {
+                first_time=0;
                 cpu_data = __alloc_bootmem(PERCPU_PAGE_SIZE * NR_CPUS,
                                            PERCPU_PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
                 for (cpu = 0; cpu < NR_CPUS; cpu++) {
diff --git a/arch/ia64/mm/discontig.c b/arch/ia64/mm/discontig.c
index c87d6d1d5813..573d5cc63e2b 100644
--- a/arch/ia64/mm/discontig.c
+++ b/arch/ia64/mm/discontig.c
@@ -528,12 +528,17 @@ void __init find_memory(void)
 void *per_cpu_init(void)
 {
         int cpu;
+        static int first_time = 1;
+
 
         if (smp_processor_id() != 0)
                 return __per_cpu_start + __per_cpu_offset[smp_processor_id()];
 
-        for (cpu = 0; cpu < NR_CPUS; cpu++)
-                per_cpu(local_per_cpu_offset, cpu) = __per_cpu_offset[cpu];
+        if (first_time) {
+                first_time = 0;
+                for (cpu = 0; cpu < NR_CPUS; cpu++)
+                        per_cpu(local_per_cpu_offset, cpu) = __per_cpu_offset[cpu];
+        }
 
         return __per_cpu_start + __per_cpu_offset[smp_processor_id()];
 }