C++ put_online_cpus函数代码示例

void arch_jump_label_transform(struct jump_entry *entry,
			       enum jump_label_type type)
{
	union jump_code_union code;

	if (type == JUMP_LABEL_ENABLE) {
		code.jump = 0xe9;
		code.offset = entry->target -
				(entry->code + JUMP_LABEL_NOP_SIZE);
	} else
		memcpy(&code, ideal_nops[NOP_ATOMIC5], JUMP_LABEL_NOP_SIZE);
	get_online_cpus();
	mutex_lock(&text_mutex);
	text_poke_smp((void *)entry->code, &code, JUMP_LABEL_NOP_SIZE);
	mutex_unlock(&text_mutex);
	put_online_cpus();
}

int config_L2(int size)
{ 
	int cur_size = get_l2c_size();
	if (size != SZ_256K && size != SZ_512K) {
		printk("inlvalid input size %x\n", size);
		return -1;
	}
	if (in_interrupt()) {
		printk(KERN_ERR "Cannot use %s in interrupt/softirq context\n",
		       __func__);
		return -1;
	}
	if (size == cur_size) {
		printk("Config L2 size %x is equal to current L2 size %x\n",
		       size, cur_size);
		return 0;
	}

    atomic_set(&L1_flush_done, 0);
	get_online_cpus();
	//printk("[Config L2] Config L2 start, on line cpu = %d\n",num_online_cpus());	
	
	/* disable cache and flush L1 */
	on_each_cpu((smp_call_func_t)atomic_flush, NULL, true);
	//while(atomic_read(&L1_flush_done) != num_online_cpus());	
    //printk("[Config L2] L1 flush done\n");
    
    /* flush L2 */	
	inner_dcache_flush_L2();
	//printk("[Config L2] L2 flush done\n");
	
	/* change L2 size */	
	config_L2_size(size);
	//printk("[Config L2] Change L2 flush size done(size = %d)\n",size);
		
	/* enable cache */
	atomic_set(&L1_flush_done, 0);
	on_each_cpu((smp_call_func_t)__enable_cache, NULL, true);
	
	//update cr_alignment for other kernel function usage 
	cr_alignment = cr_alignment | (0x4); //C1_CBIT
	put_online_cpus();
	printk("Config L2 size %x done\n", size);
	return 0;
}

static void nmi_shutdown(void)
{
	struct op_msrs *msrs;

	get_online_cpus();
	unregister_cpu_notifier(&oprofile_cpu_nb);
	on_each_cpu(nmi_cpu_shutdown, NULL, 1);
	nmi_enabled = 0;
	ctr_running = 0;
	put_online_cpus();
	/* make variables visible to the nmi handler: */
	smp_mb();
	unregister_nmi_handler(NMI_LOCAL, "oprofile");
	msrs = &get_cpu_var(cpu_msrs);
	model->shutdown(msrs);
	free_msrs();
	put_cpu_var(cpu_msrs);
}

/* Input boost main boost function */
static void __cpuinit ib_boost_main(struct work_struct *work)
{
	unsigned int cpu, nr_cpus_to_boost, nr_boosted = 0;

	get_online_cpus();
	/* Max. of 3 CPUs can be boosted at any given time */
	nr_cpus_to_boost = num_online_cpus() > 2 ? num_online_cpus() - 1 : 1;

	for_each_online_cpu(cpu) {
		/* Calculate boost duration for each CPU (CPU0 is boosted the longest) */
		/* TODO: Make this more standard and configurable from sysfs */
		boost_ms[cpu] = 1500 - (cpu * 200) - (nr_cpus_to_boost * 250);
		cpu_boost_cpu(cpu);
		nr_boosted++;
		if (nr_boosted == nr_cpus_to_boost)
			break;
	}
	put_online_cpus();
}

/*
 * STP work. Check for the STP state and take over the clock
 * synchronization if the STP clock source is usable.
 */
static void stp_work_fn(struct work_struct *work)
{
	struct clock_sync_data stp_sync;
	int rc;

	/* prevent multiple execution. */
	mutex_lock(&stp_work_mutex);

	if (!stp_online) {
		chsc_sstpc(stp_page, STP_OP_CTRL, 0x0000, NULL);
		del_timer_sync(&stp_timer);
		goto out_unlock;
	}

	rc = chsc_sstpc(stp_page, STP_OP_CTRL, 0xb0e0, NULL);
	if (rc)
		goto out_unlock;

	rc = chsc_sstpi(stp_page, &stp_info, sizeof(struct stp_sstpi));
	if (rc || stp_info.c == 0)
		goto out_unlock;

	/* Skip synchronization if the clock is already in sync. */
	if (check_sync_clock())
		goto out_unlock;

	memset(&stp_sync, 0, sizeof(stp_sync));
	get_online_cpus();
	atomic_set(&stp_sync.cpus, num_online_cpus() - 1);
	stop_machine(stp_sync_clock, &stp_sync, cpu_online_mask);
	put_online_cpus();

	if (!check_sync_clock())
		/*
		 * There is a usable clock but the synchonization failed.
		 * Retry after a second.
		 */
		mod_timer(&stp_timer, jiffies + HZ);

out_unlock:
	mutex_unlock(&stp_work_mutex);
}

static void fb_bpf_cleanup_filter_cpus(struct fblock *fb)
{
	unsigned int cpu;
	struct fb_bpf_priv __percpu *fb_priv;

	if (!fb)
		return;

	rcu_read_lock();
	fb_priv = (struct fb_bpf_priv __percpu *) rcu_dereference_raw(fb->private_data);
	rcu_read_unlock();

	get_online_cpus();
	for_each_online_cpu(cpu) {
		struct fb_bpf_priv *fb_priv_cpu;
		fb_priv_cpu = per_cpu_ptr(fb_priv, cpu);
		fb_bpf_cleanup_filter(fb_priv_cpu);
	}
	put_online_cpus();
}

static void __init acpi_cpufreq_boost_init(void)
{
	if (boot_cpu_has(X86_FEATURE_CPB) || boot_cpu_has(X86_FEATURE_IDA)) {
		msrs = msrs_alloc();

		if (!msrs)
			return;

		acpi_cpufreq_driver.boost_supported = true;
		acpi_cpufreq_driver.boost_enabled = boost_state(0);
		get_online_cpus();

		/* Force all MSRs to the same value */
		boost_set_msrs(acpi_cpufreq_driver.boost_enabled,
			       cpu_online_mask);

		register_cpu_notifier(&boost_nb);

		put_online_cpus();
	}
}

static int fb_bpf_proc_show_filter(struct seq_file *m, void *v)
{
	unsigned long flags;
	struct fblock *fb = (struct fblock *) m->private;
	struct fb_bpf_priv *fb_priv_cpu;
	struct sk_filter *sf;

	get_online_cpus();
	rcu_read_lock();
	fb_priv_cpu = this_cpu_ptr(rcu_dereference_raw(fb->private_data));
	rcu_read_unlock();

	spin_lock_irqsave(&fb_priv_cpu->flock, flags);
	sf = fb_priv_cpu->filter;
	if (sf) {
		unsigned int i;
		if (sf->bpf_func == sk_run_filter)
			seq_puts(m, "bpf jit: 0\n");
		else
			seq_puts(m, "bpf jit: 1\n");
		seq_puts(m, "code:\n");
		for (i = 0; i < sf->len; ++i) {
			char sline[32];
			memset(sline, 0, sizeof(sline));
			snprintf(sline, sizeof(sline),
				 "{ 0x%x, %u, %u, 0x%x }\n",
				 sf->insns[i].code,
				 sf->insns[i].jt,
				 sf->insns[i].jf,
				 sf->insns[i].k);
			sline[sizeof(sline) - 1] = 0;
			seq_puts(m, sline);
		}
	}
	spin_unlock_irqrestore(&fb_priv_cpu->flock, flags);
	put_online_cpus();

	return 0;
}

static void acpi_processor_remove(struct acpi_device *device)
{
	struct acpi_processor *pr;

	if (!device || !acpi_driver_data(device))
		return;

	pr = acpi_driver_data(device);
	if (pr->id >= nr_cpu_ids)
		goto out;

	/*
	 * The only reason why we ever get here is CPU hot-removal.  The CPU is
	 * already offline and the ACPI device removal locking prevents it from
	 * being put back online at this point.
	 *
	 * Unbind the driver from the processor device and detach it from the
	 * ACPI companion object.
	 */
	device_release_driver(pr->dev);
	acpi_unbind_one(pr->dev);

	/* Clean up. */
	per_cpu(processor_device_array, pr->id) = NULL;
	per_cpu(processors, pr->id) = NULL;
	try_offline_node(cpu_to_node(pr->id));

	/* Remove the CPU. */
	get_online_cpus();
	arch_unregister_cpu(pr->id);
	acpi_unmap_lsapic(pr->id);
	put_online_cpus();

 out:
	free_cpumask_var(pr->throttling.shared_cpu_map);
	kfree(pr);
}

/*
 * Take a map of online CPUs and the number of available interrupt vectors
 * and generate an output cpumask suitable for spreading MSI/MSI-X vectors
 * so that they are distributed as good as possible around the CPUs.  If
 * more vectors than CPUs are available we'll map one to each CPU,
 * otherwise we map one to the first sibling of each socket.
 *
 * If there are more vectors than CPUs we will still only have one bit
 * set per CPU, but interrupt code will keep on assigning the vectors from
 * the start of the bitmap until we run out of vectors.
 */
struct cpumask *irq_create_affinity_mask(unsigned int *nr_vecs)
{
	struct cpumask *affinity_mask;
	unsigned int max_vecs = *nr_vecs;

	if (max_vecs == 1)
		return NULL;

	affinity_mask = kzalloc(cpumask_size(), GFP_KERNEL);
	if (!affinity_mask) {
		*nr_vecs = 1;
		return NULL;
	}

	get_online_cpus();
	if (max_vecs >= num_online_cpus()) {
		cpumask_copy(affinity_mask, cpu_online_mask);
		*nr_vecs = num_online_cpus();
	} else {
		unsigned int vecs = 0, cpu;

		for_each_online_cpu(cpu) {
			if (cpu == get_first_sibling(cpu)) {
				cpumask_set_cpu(cpu, affinity_mask);
				vecs++;
			}

			if (--max_vecs == 0)
				break;
		}
		*nr_vecs = vecs;
	}
	put_online_cpus();

	return affinity_mask;
}

static void etm_disable_sysfs(struct coresight_device *csdev)
{
	struct etm_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);

	/*
	 * Taking hotplug lock here protects from clocks getting disabled
	 * with tracing being left on (crash scenario) if user disable occurs
	 * after cpu online mask indicates the cpu is offline but before the
	 * DYING hotplug callback is serviced by the ETM driver.
	 */
	get_online_cpus();
	spin_lock(&drvdata->spinlock);

	/*
	 * Executing etm_disable_hw on the cpu whose ETM is being disabled
	 * ensures that register writes occur when cpu is powered.
	 */
	smp_call_function_single(drvdata->cpu, etm_disable_hw, drvdata, 1);

	spin_unlock(&drvdata->spinlock);
	put_online_cpus();

	dev_info(drvdata->dev, "ETM tracing disabled\n");
}

//.........这里部分代码省略.........
		 * for each CPU.  We keep a descriptor for the GDT which says
		 * where it is and how big it is (the size is actually the last
		 * byte, not the size, hence the "-1").
		 */
		state->host_gdt_desc.size = GDT_SIZE-1;
		state->host_gdt_desc.address = (long)get_cpu_gdt_table(i);

		/*
		 * All CPUs on the Host use the same Interrupt Descriptor
		 * Table, so we just use store_idt(), which gets this CPU's IDT
		 * descriptor.
		 */
		store_idt(&state->host_idt_desc);

		/*
		 * The descriptors for the Guest's GDT and IDT can be filled
		 * out now, too.  We copy the GDT & IDT into ->guest_gdt and
		 * ->guest_idt before actually running the Guest.
		 */
		state->guest_idt_desc.size = sizeof(state->guest_idt)-1;
		state->guest_idt_desc.address = (long)&state->guest_idt;
		state->guest_gdt_desc.size = sizeof(state->guest_gdt)-1;
		state->guest_gdt_desc.address = (long)&state->guest_gdt;

		/*
		 * We know where we want the stack to be when the Guest enters
		 * the Switcher: in pages->regs.  The stack grows upwards, so
		 * we start it at the end of that structure.
		 */
		state->guest_tss.sp0 = (long)(&pages->regs + 1);
		/*
		 * And this is the GDT entry to use for the stack: we keep a
		 * couple of special LGUEST entries.
		 */
		state->guest_tss.ss0 = LGUEST_DS;

		/*
		 * x86 can have a finegrained bitmap which indicates what I/O
		 * ports the process can use.  We set it to the end of our
		 * structure, meaning "none".
		 */
		state->guest_tss.io_bitmap_base = sizeof(state->guest_tss);

		/*
		 * Some GDT entries are the same across all Guests, so we can
		 * set them up now.
		 */
		setup_default_gdt_entries(state);
		/* Most IDT entries are the same for all Guests, too.*/
		setup_default_idt_entries(state, default_idt_entries);

		/*
		 * The Host needs to be able to use the LGUEST segments on this
		 * CPU, too, so put them in the Host GDT.
		 */
		get_cpu_gdt_table(i)[GDT_ENTRY_LGUEST_CS] = FULL_EXEC_SEGMENT;
		get_cpu_gdt_table(i)[GDT_ENTRY_LGUEST_DS] = FULL_SEGMENT;
	}

	/*
	 * In the Switcher, we want the %cs segment register to use the
	 * LGUEST_CS GDT entry: we've put that in the Host and Guest GDTs, so
	 * it will be undisturbed when we switch.  To change %cs and jump we
	 * need this structure to feed to Intel's "lcall" instruction.
	 */
	lguest_entry.offset = (long)switch_to_guest + switcher_offset();
	lguest_entry.segment = LGUEST_CS;

	/*
	 * Finally, we need to turn off "Page Global Enable".  PGE is an
	 * optimization where page table entries are specially marked to show
	 * they never change.  The Host kernel marks all the kernel pages this
	 * way because it's always present, even when userspace is running.
	 *
	 * Lguest breaks this: unbeknownst to the rest of the Host kernel, we
	 * switch to the Guest kernel.  If you don't disable this on all CPUs,
	 * you'll get really weird bugs that you'll chase for two days.
	 *
	 * I used to turn PGE off every time we switched to the Guest and back
	 * on when we return, but that slowed the Switcher down noticibly.
	 */

	/*
	 * We don't need the complexity of CPUs coming and going while we're
	 * doing this.
	 */
	get_online_cpus();
	if (boot_cpu_has(X86_FEATURE_PGE)) { /* We have a broader idea of "global". */
		/* Remember that this was originally set (for cleanup). */
		cpu_had_pge = 1;
		/*
		 * adjust_pge is a helper function which sets or unsets the PGE
		 * bit on its CPU, depending on the argument (0 == unset).
		 */
		on_each_cpu(adjust_pge, (void *)0, 1);
		/* Turn off the feature in the global feature set. */
		clear_cpu_cap(&boot_cpu_data, X86_FEATURE_PGE);
	}
	put_online_cpus();
}

/*
 * Allow secondary CPU threads to come online again
 */
void uninhibit_secondary_onlining(void)
{
	get_online_cpus();
	atomic_dec(&secondary_inhibit_count);
	put_online_cpus();
}

/*
 * mipsmt_sys_sched_setaffinity - set the cpu affinity of a process
 */
asmlinkage long mipsmt_sys_sched_setaffinity(pid_t pid, unsigned int len,
				      unsigned long __user *user_mask_ptr)
{
	cpumask_var_t cpus_allowed, new_mask, effective_mask;
	struct thread_info *ti;
	struct task_struct *p;
	int retval;

	if (len < sizeof(new_mask))
		return -EINVAL;

	if (copy_from_user(&new_mask, user_mask_ptr, sizeof(new_mask)))
		return -EFAULT;

	get_online_cpus();
	rcu_read_lock();

	p = find_process_by_pid(pid);
	if (!p) {
		rcu_read_unlock();
		put_online_cpus();
		return -ESRCH;
	}

	/* Prevent p going away */
	get_task_struct(p);
	rcu_read_unlock();

	if (!alloc_cpumask_var(&cpus_allowed, GFP_KERNEL)) {
		retval = -ENOMEM;
		goto out_put_task;
	}
	if (!alloc_cpumask_var(&new_mask, GFP_KERNEL)) {
		retval = -ENOMEM;
		goto out_free_cpus_allowed;
	}
	if (!alloc_cpumask_var(&effective_mask, GFP_KERNEL)) {
		retval = -ENOMEM;
		goto out_free_new_mask;
	}
	retval = -EPERM;
	if (!check_same_owner(p) && !capable(CAP_SYS_NICE))
		goto out_unlock;

	retval = security_task_setscheduler(p);
	if (retval)
		goto out_unlock;

	/* Record new user-specified CPU set for future reference */
	cpumask_copy(&p->thread.user_cpus_allowed, new_mask);

 again:
	/* Compute new global allowed CPU set if necessary */
	ti = task_thread_info(p);
	if (test_ti_thread_flag(ti, TIF_FPUBOUND) &&
	    cpus_intersects(*new_mask, mt_fpu_cpumask)) {
		cpus_and(*effective_mask, *new_mask, mt_fpu_cpumask);
		retval = set_cpus_allowed_ptr(p, effective_mask);
	} else {
		cpumask_copy(effective_mask, new_mask);
		clear_ti_thread_flag(ti, TIF_FPUBOUND);
		retval = set_cpus_allowed_ptr(p, new_mask);
	}

	if (!retval) {
		cpuset_cpus_allowed(p, cpus_allowed);
		if (!cpumask_subset(effective_mask, cpus_allowed)) {
			/*
			 * We must have raced with a concurrent cpuset
			 * update. Just reset the cpus_allowed to the
			 * cpuset's cpus_allowed
			 */
			cpumask_copy(new_mask, cpus_allowed);
			goto again;
		}
	}
out_unlock:
	free_cpumask_var(effective_mask);
out_free_new_mask:
	free_cpumask_var(new_mask);
out_free_cpus_allowed:
	free_cpumask_var(cpus_allowed);
out_put_task:
	put_task_struct(p);
	put_online_cpus();
	return retval;
}

//.........这里部分代码省略.........
			unsigned long ecx = 1;
			unsigned long eax = target_mwait;

			/*
			 * REVISIT: may call enter_idle() to notify drivers who
			 * can save power during cpu idle. same for exit_idle()
			 */
			local_touch_nmi();
			stop_critical_timings();
			mwait_idle_with_hints(eax, ecx);
			start_critical_timings();
			atomic_inc(&idle_wakeup_counter);
		}
		preempt_enable();
	}
	del_timer_sync(&wakeup_timer);
	clear_bit(cpunr, cpu_clamping_mask);

	return 0;
}

/*
 * 1 HZ polling while clamping is active, useful for userspace
 * to monitor actual idle ratio.
 */
static void poll_pkg_cstate(struct work_struct *dummy);
static DECLARE_DELAYED_WORK(poll_pkg_cstate_work, poll_pkg_cstate);
static void poll_pkg_cstate(struct work_struct *dummy)
{
	static u64 msr_last;
	static u64 tsc_last;
	static unsigned long jiffies_last;

	u64 msr_now;
	unsigned long jiffies_now;
	u64 tsc_now;
	u64 val64;

	msr_now = pkg_state_counter();
	tsc_now = rdtsc();
	jiffies_now = jiffies;

	/* calculate pkg cstate vs tsc ratio */
	if (!msr_last || !tsc_last)
		pkg_cstate_ratio_cur = 1;
	else {
		if (tsc_now - tsc_last) {
			val64 = 100 * (msr_now - msr_last);
			do_div(val64, (tsc_now - tsc_last));
			pkg_cstate_ratio_cur = val64;
		}
	}

	/* update record */
	msr_last = msr_now;
	jiffies_last = jiffies_now;
	tsc_last = tsc_now;

	if (true == clamping)
		schedule_delayed_work(&poll_pkg_cstate_work, HZ);
}

static int start_power_clamp(void)
{
	unsigned long cpu;
	struct task_struct *thread;

	set_target_ratio = clamp(set_target_ratio, 0U, MAX_TARGET_RATIO - 1);
	/* prevent cpu hotplug */
	get_online_cpus();

	/* prefer BSP */
	control_cpu = 0;
	if (!cpu_online(control_cpu))
		control_cpu = smp_processor_id();

	clamping = true;
	schedule_delayed_work(&poll_pkg_cstate_work, 0);

	/* start one thread per online cpu */
	for_each_online_cpu(cpu) {
		struct task_struct **p =
			per_cpu_ptr(powerclamp_thread, cpu);

		thread = kthread_create_on_node(clamp_thread,
						(void *) cpu,
						cpu_to_node(cpu),
						"kidle_inject/%ld", cpu);
		/* bind to cpu here */
		if (likely(!IS_ERR(thread))) {
			kthread_bind(thread, cpu);
			wake_up_process(thread);
			*p = thread;
		}

	}
	put_online_cpus();

	return 0;
}