sn2_smp.c

xen虚拟机源代码安装包

#else
	rr_value = REGION_NUMBER(start);
#endif

	shub1 = is_shub1();
	if (shub1) {
		data0 = (1UL << SH1_PTC_0_A_SHFT) |
		    	(nbits << SH1_PTC_0_PS_SHFT) |
			(rr_value << SH1_PTC_0_RID_SHFT) |
		    	(1UL << SH1_PTC_0_START_SHFT);
#ifndef XEN
		ptc0 = (long *)GLOBAL_MMR_PHYS_ADDR(0, SH1_PTC_0);
		ptc1 = (long *)GLOBAL_MMR_PHYS_ADDR(0, SH1_PTC_1);
#else
		ptc0 = (unsigned long *)GLOBAL_MMR_PHYS_ADDR(0, SH1_PTC_0);
		ptc1 = (unsigned long *)GLOBAL_MMR_PHYS_ADDR(0, SH1_PTC_1);
#endif
	} else {
		data0 = (1UL << SH2_PTC_A_SHFT) |
			(nbits << SH2_PTC_PS_SHFT) |
		    	(1UL << SH2_PTC_START_SHFT);
#ifndef XEN
		ptc0 = (long *)GLOBAL_MMR_PHYS_ADDR(0, SH2_PTC + 
#else
		ptc0 = (unsigned long *)GLOBAL_MMR_PHYS_ADDR(0, SH2_PTC + 
#endif
			(rr_value << SH2_PTC_RID_SHFT));
		ptc1 = NULL;
	}
	

	mynasid = get_nasid();
	use_cpu_ptcga = local_node_uses_ptc_ga(shub1);
	max_active = max_active_pio(shub1);

	itc = ia64_get_itc();
	spin_lock_irqsave(PTC_LOCK(shub1), flags);
	itc2 = ia64_get_itc();

	__get_cpu_var(ptcstats).lock_itc_clocks += itc2 - itc;
	__get_cpu_var(ptcstats).shub_ptc_flushes++;
	__get_cpu_var(ptcstats).nodes_flushed += nix;
	if (!mymm)
		 __get_cpu_var(ptcstats).shub_ptc_flushes_not_my_mm++;

	if (use_cpu_ptcga && !mymm) {
		old_rr = ia64_get_rr(start);
		ia64_set_rr(start, (old_rr & 0xff) | (rr_value << 8));
		ia64_srlz_d();
	}

	wait_piowc();
	do {
		if (shub1)
			data1 = start | (1UL << SH1_PTC_1_START_SHFT);
		else
			data0 = (data0 & ~SH2_PTC_ADDR_MASK) | (start & SH2_PTC_ADDR_MASK);
		deadlock = 0;
		active = 0;
		for (ibegin = 0, i = 0; i < nix; i++) {
			nasid = nasids[i];
			if (use_cpu_ptcga && unlikely(nasid == mynasid)) {
				ia64_ptcga(start, nbits << 2);
				ia64_srlz_i();
			} else {
				ptc0 = CHANGE_NASID(nasid, ptc0);
				if (ptc1)
					ptc1 = CHANGE_NASID(nasid, ptc1);
				pio_atomic_phys_write_mmrs(ptc0, data0, ptc1, data1);
				active++;
			}
			if (active >= max_active || i == (nix - 1)) {
				if ((deadlock = wait_piowc())) {
					sn2_ptc_deadlock_recovery(nasids, ibegin, i, mynasid, ptc0, data0, ptc1, data1);
					if (reset_max_active_on_deadlock())
						max_active = 1;
				}
				active = 0;
				ibegin = i + 1;
			}
		}
		start += (1UL << nbits);
	} while (start < end);

	itc2 = ia64_get_itc() - itc2;
	__get_cpu_var(ptcstats).shub_itc_clocks += itc2;
	if (itc2 > __get_cpu_var(ptcstats).shub_itc_clocks_max)
		__get_cpu_var(ptcstats).shub_itc_clocks_max = itc2;

	if (old_rr) {
		ia64_set_rr(start, old_rr);
		ia64_srlz_d();
	}

	spin_unlock_irqrestore(PTC_LOCK(shub1), flags);

	preempt_enable();
}
#endif

/*
 * sn2_ptc_deadlock_recovery
 *
 * Recover from PTC deadlocks conditions. Recovery requires stepping thru each 
 * TLB flush transaction.  The recovery sequence is somewhat tricky & is
 * coded in assembly language.
 */

void
sn2_ptc_deadlock_recovery(short *nasids, short ib, short ie, int mynasid,
			  volatile unsigned long *ptc0, unsigned long data0,
			  volatile unsigned long *ptc1, unsigned long data1)
{
	short nasid, i;
	unsigned long *piows, zeroval, n;

	__get_cpu_var(ptcstats).deadlocks++;

	piows = (unsigned long *) pda->pio_write_status_addr;
	zeroval = pda->pio_write_status_val;


	for (i=ib; i <= ie; i++) {
		nasid = nasids[i];
		if (local_node_uses_ptc_ga(is_shub1()) && nasid == mynasid)
			continue;
		ptc0 = CHANGE_NASID(nasid, ptc0);
		if (ptc1)
			ptc1 = CHANGE_NASID(nasid, ptc1);

		n = sn2_ptc_deadlock_recovery_core(ptc0, data0, ptc1, data1, piows, zeroval);
		__get_cpu_var(ptcstats).deadlocks2 += n;
	}

}

/**
 * sn_send_IPI_phys - send an IPI to a Nasid and slice
 * @nasid: nasid to receive the interrupt (may be outside partition)
 * @physid: physical cpuid to receive the interrupt.
 * @vector: command to send
 * @delivery_mode: delivery mechanism
 *
 * Sends an IPI (interprocessor interrupt) to the processor specified by
 * @physid
 *
 * @delivery_mode can be one of the following
 *
 * %IA64_IPI_DM_INT - pend an interrupt
 * %IA64_IPI_DM_PMI - pend a PMI
 * %IA64_IPI_DM_NMI - pend an NMI
 * %IA64_IPI_DM_INIT - pend an INIT interrupt
 */
void sn_send_IPI_phys(int nasid, long physid, int vector, int delivery_mode)
{
	long val;
	unsigned long flags = 0;
	volatile long *p;

	p = (long *)GLOBAL_MMR_PHYS_ADDR(nasid, SH_IPI_INT);
	val = (1UL << SH_IPI_INT_SEND_SHFT) |
	    (physid << SH_IPI_INT_PID_SHFT) |
	    ((long)delivery_mode << SH_IPI_INT_TYPE_SHFT) |
	    ((long)vector << SH_IPI_INT_IDX_SHFT) |
	    (0x000feeUL << SH_IPI_INT_BASE_SHFT);

	mb();
	if (enable_shub_wars_1_1()) {
		spin_lock_irqsave(&sn2_global_ptc_lock, flags);
	}
	pio_phys_write_mmr(p, val);
	if (enable_shub_wars_1_1()) {
		wait_piowc();
		spin_unlock_irqrestore(&sn2_global_ptc_lock, flags);
	}

}

EXPORT_SYMBOL(sn_send_IPI_phys);

/**
 * sn2_send_IPI - send an IPI to a processor
 * @cpuid: target of the IPI
 * @vector: command to send
 * @delivery_mode: delivery mechanism
 * @redirect: redirect the IPI?
 *
 * Sends an IPI (InterProcessor Interrupt) to the processor specified by
 * @cpuid.  @vector specifies the command to send, while @delivery_mode can 
 * be one of the following
 *
 * %IA64_IPI_DM_INT - pend an interrupt
 * %IA64_IPI_DM_PMI - pend a PMI
 * %IA64_IPI_DM_NMI - pend an NMI
 * %IA64_IPI_DM_INIT - pend an INIT interrupt
 */
void sn2_send_IPI(int cpuid, int vector, int delivery_mode, int redirect)
{
	long physid;
	int nasid;

	physid = cpu_physical_id(cpuid);
#ifdef XEN
	if (!sn_nodepda) {
		ia64_sn_get_sapic_info(physid, &nasid, NULL, NULL);
	} else
#endif
	nasid = cpuid_to_nasid(cpuid);

	/* the following is used only when starting cpus at boot time */
	if (unlikely(nasid == -1))
		ia64_sn_get_sapic_info(physid, &nasid, NULL, NULL);

 	sn_send_IPI_phys(nasid, physid, vector, delivery_mode);
}

#ifdef CONFIG_PROC_FS

#define PTC_BASENAME	"sgi_sn/ptc_statistics"

static void *sn2_ptc_seq_start(struct seq_file *file, loff_t * offset)
{
	if (*offset < NR_CPUS)
		return offset;
	return NULL;
}

static void *sn2_ptc_seq_next(struct seq_file *file, void *data, loff_t * offset)
{
	(*offset)++;
	if (*offset < NR_CPUS)
		return offset;
	return NULL;
}

static void sn2_ptc_seq_stop(struct seq_file *file, void *data)
{
}

static int sn2_ptc_seq_show(struct seq_file *file, void *data)
{
	struct ptc_stats *stat;
	int cpu;

	cpu = *(loff_t *) data;

	if (!cpu) {
		seq_printf(file,
			   "# cpu ptc_l newrid ptc_flushes nodes_flushed deadlocks lock_nsec shub_nsec shub_nsec_max not_my_mm deadlock2\n");
		seq_printf(file, "# ptctest %d\n", sn2_ptctest);
	}

	if (cpu < NR_CPUS && cpu_online(cpu)) {
		stat = &per_cpu(ptcstats, cpu);
		seq_printf(file, "cpu %d %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld\n", cpu, stat->ptc_l,
				stat->change_rid, stat->shub_ptc_flushes, stat->nodes_flushed,
				stat->deadlocks,
				1000 * stat->lock_itc_clocks / per_cpu(cpu_info, cpu).cyc_per_usec,
				1000 * stat->shub_itc_clocks / per_cpu(cpu_info, cpu).cyc_per_usec,
				1000 * stat->shub_itc_clocks_max / per_cpu(cpu_info, cpu).cyc_per_usec,
				stat->shub_ptc_flushes_not_my_mm,
				stat->deadlocks2);
	}
	return 0;
}

static struct seq_operations sn2_ptc_seq_ops = {
	.start = sn2_ptc_seq_start,
	.next = sn2_ptc_seq_next,
	.stop = sn2_ptc_seq_stop,
	.show = sn2_ptc_seq_show
};

static int sn2_ptc_proc_open(struct inode *inode, struct file *file)
{
	return seq_open(file, &sn2_ptc_seq_ops);
}

static struct file_operations proc_sn2_ptc_operations = {
	.open = sn2_ptc_proc_open,
	.read = seq_read,
	.llseek = seq_lseek,
	.release = seq_release,
};

static struct proc_dir_entry *proc_sn2_ptc;

static int __init sn2_ptc_init(void)
{
	if (!ia64_platform_is("sn2"))
		return 0;

	if (!(proc_sn2_ptc = create_proc_entry(PTC_BASENAME, 0444, NULL))) {
		printk(KERN_ERR "unable to create %s proc entry", PTC_BASENAME);
		return -EINVAL;
	}
	proc_sn2_ptc->proc_fops = &proc_sn2_ptc_operations;
	spin_lock_init(&sn2_global_ptc_lock);
	return 0;
}

static void __exit sn2_ptc_exit(void)
{
	remove_proc_entry(PTC_BASENAME, NULL);
}

module_init(sn2_ptc_init);
module_exit(sn2_ptc_exit);
#endif /* CONFIG_PROC_FS */