#include <asm/cpudata.h>
#include <asm/irq.h>
+#include <asm/irq_regs.h>
#include <asm/page.h>
#include <asm/pgtable.h>
#include <asm/oplib.h>
#include <asm/starfire.h>
#include <asm/tlb.h>
#include <asm/sections.h>
+#include <asm/prom.h>
extern void calibrate_delay(void);
void __init smp_store_cpu_info(int id)
{
- int cpu_node, def;
+ struct device_node *dp;
+ int def;
/* multiplier and counter set by
smp_setup_percpu_timer() */
cpu_data(id).udelay_val = loops_per_jiffy;
- cpu_find_by_mid(id, &cpu_node);
- cpu_data(id).clock_tick = prom_getintdefault(cpu_node,
- "clock-frequency", 0);
+ cpu_find_by_mid(id, &dp);
+ cpu_data(id).clock_tick =
+ of_getintprop_default(dp, "clock-frequency", 0);
def = ((tlb_type == hypervisor) ? (8 * 1024) : (16 * 1024));
- cpu_data(id).dcache_size = prom_getintdefault(cpu_node, "dcache-size",
- def);
+ cpu_data(id).dcache_size =
+ of_getintprop_default(dp, "dcache-size", def);
def = 32;
cpu_data(id).dcache_line_size =
- prom_getintdefault(cpu_node, "dcache-line-size", def);
+ of_getintprop_default(dp, "dcache-line-size", def);
def = 16 * 1024;
- cpu_data(id).icache_size = prom_getintdefault(cpu_node, "icache-size",
- def);
+ cpu_data(id).icache_size =
+ of_getintprop_default(dp, "icache-size", def);
def = 32;
cpu_data(id).icache_line_size =
- prom_getintdefault(cpu_node, "icache-line-size", def);
+ of_getintprop_default(dp, "icache-line-size", def);
def = ((tlb_type == hypervisor) ?
(3 * 1024 * 1024) :
(4 * 1024 * 1024));
- cpu_data(id).ecache_size = prom_getintdefault(cpu_node, "ecache-size",
- def);
+ cpu_data(id).ecache_size =
+ of_getintprop_default(dp, "ecache-size", def);
def = 64;
cpu_data(id).ecache_line_size =
- prom_getintdefault(cpu_node, "ecache-line-size", def);
+ of_getintprop_default(dp, "ecache-line-size", def);
printk("CPU[%d]: Caches "
"D[sz(%d):line_sz(%d)] "
prom_startcpu_cpuid(cpu, entry, cookie);
} else {
- int cpu_node;
+ struct device_node *dp;
- cpu_find_by_mid(cpu, &cpu_node);
- prom_startcpu(cpu_node, entry, cookie);
+ cpu_find_by_mid(cpu, &dp);
+ prom_startcpu(dp->node, entry, cookie);
}
for (timeout = 0; timeout < 5000000; timeout++) {
int wait;
};
-static DEFINE_SPINLOCK(call_lock);
+static __cacheline_aligned_in_smp DEFINE_SPINLOCK(call_lock);
static struct call_data_struct *call_data;
extern unsigned long xcall_call_function;
-/*
+/**
+ * smp_call_function(): Run a function on all other CPUs.
+ * @func: The function to run. This must be fast and non-blocking.
+ * @info: An arbitrary pointer to pass to the function.
+ * @nonatomic: currently unused.
+ * @wait: If true, wait (atomically) until function has completed on other CPUs.
+ *
+ * Returns 0 on success, else a negative status code. Does not return until
+ * remote CPUs are nearly ready to execute <<func>> or are or have executed.
+ *
* You must not call this function with disabled interrupts or from a
* hardware interrupt handler or from a bottom half handler.
*/
{
struct call_data_struct data;
int cpus;
- long timeout;
/* Can deadlock when called with interrupts disabled */
WARN_ON(irqs_disabled());
goto out_unlock;
call_data = &data;
+ mb();
smp_cross_call_masked(&xcall_call_function, 0, 0, 0, mask);
- /*
- * Wait for other cpus to complete function or at
- * least snap the call data.
- */
- timeout = 1000000;
- while (atomic_read(&data.finished) != cpus) {
- if (--timeout <= 0)
- goto out_timeout;
- barrier();
- udelay(1);
- }
+ /* Wait for response */
+ while (atomic_read(&data.finished) != cpus)
+ cpu_relax();
out_unlock:
spin_unlock(&call_lock);
return 0;
-
-out_timeout:
- spin_unlock(&call_lock);
- printk("XCALL: Remote cpus not responding, ncpus=%d finished=%d\n",
- cpus, atomic_read(&data.finished));
- return 0;
}
int smp_call_function(void (*func)(void *info), void *info,
static void tsb_sync(void *info)
{
+ struct trap_per_cpu *tp = &trap_block[raw_smp_processor_id()];
struct mm_struct *mm = info;
- if (current->active_mm == mm)
+ /* It is not valid to test "currrent->active_mm == mm" here.
+ *
+ * The value of "current" is not changed atomically with
+ * switch_mm(). But that's OK, we just need to check the
+ * current cpu's trap block PGD physical address.
+ */
+ if (tp->pgd_paddr == __pa(mm->pgd))
tsb_context_switch(mm);
}
unsigned long compare, tick, pstate;
int cpu = smp_processor_id();
int user = user_mode(regs);
+ struct pt_regs *old_regs;
/*
* Check for level 14 softint.
clear_softint(tick_mask);
}
+ old_regs = set_irq_regs(regs);
do {
- profile_tick(CPU_PROFILING, regs);
+ profile_tick(CPU_PROFILING);
if (!--prof_counter(cpu)) {
irq_enter();
: /* no outputs */
: "r" (pstate));
} while (time_after_eq(tick, compare));
+ set_irq_regs(old_regs);
}
static void __init smp_setup_percpu_timer(void)
boot_cpu_id = hard_smp_processor_id();
current_tick_offset = timer_tick_offset;
- cpu_set(boot_cpu_id, cpu_online_map);
prof_counter(boot_cpu_id) = prof_multiplier(boot_cpu_id) = 1;
}
return -EINVAL;
spin_lock_irqsave(&prof_setup_lock, flags);
- for_each_cpu(i)
+ for_each_possible_cpu(i)
prof_multiplier(i) = multiplier;
current_tick_offset = (timer_tick_offset / multiplier);
spin_unlock_irqrestore(&prof_setup_lock, flags);
return 0;
}
+static void __init smp_tune_scheduling(void)
+{
+ struct device_node *dp;
+ int instance;
+ unsigned int def, smallest = ~0U;
+
+ def = ((tlb_type == hypervisor) ?
+ (3 * 1024 * 1024) :
+ (4 * 1024 * 1024));
+
+ instance = 0;
+ while (!cpu_find_by_instance(instance, &dp, NULL)) {
+ unsigned int val;
+
+ val = of_getintprop_default(dp, "ecache-size", def);
+ if (val < smallest)
+ smallest = val;
+
+ instance++;
+ }
+
+ /* Any value less than 256K is nonsense. */
+ if (smallest < (256U * 1024U))
+ smallest = 256 * 1024;
+
+ max_cache_size = smallest;
+
+ if (smallest < 1U * 1024U * 1024U)
+ printk(KERN_INFO "Using max_cache_size of %uKB\n",
+ smallest / 1024U);
+ else
+ printk(KERN_INFO "Using max_cache_size of %uMB\n",
+ smallest / 1024U / 1024U);
+}
+
/* Constrain the number of cpus to max_cpus. */
void __init smp_prepare_cpus(unsigned int max_cpus)
{
}
}
- for_each_cpu(i) {
+ for_each_possible_cpu(i) {
if (tlb_type == hypervisor) {
int j;
/* XXX get this mapping from machine description */
- for_each_cpu(j) {
+ for_each_possible_cpu(j) {
if ((j >> 2) == (i >> 2))
cpu_set(j, cpu_sibling_map[i]);
}
}
smp_store_cpu_info(boot_cpu_id);
+ smp_tune_scheduling();
}
/* Set this up early so that things like the scheduler can init
void __devinit smp_prepare_boot_cpu(void)
{
- int cpu = hard_smp_processor_id();
-
- if (cpu >= NR_CPUS) {
- prom_printf("Serious problem, boot cpu id >= NR_CPUS\n");
- prom_halt();
- }
-
- current_thread_info()->cpu = cpu;
- __local_per_cpu_offset = __per_cpu_offset(cpu);
-
- cpu_set(smp_processor_id(), cpu_online_map);
- cpu_set(smp_processor_id(), phys_cpu_present_map);
}
int __devinit __cpu_up(unsigned int cpu)
for (i = 0; i < NR_CPUS; i++, ptr += size)
memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
+
+ /* Setup %g5 for the boot cpu. */
+ __local_per_cpu_offset = __per_cpu_offset(smp_processor_id());
}