iosapic.c

xen虚拟机源代码安装包

/*
 * I/O SAPIC support.
 *
 * Copyright (C) 1999 Intel Corp.
 * Copyright (C) 1999 Asit Mallick <[e-mail]>
 * Copyright (C) 2000-2002 J.I. Lee <[e-mail]>
 * Copyright (C) 1999-2000, 2002-2003 Hewlett-Packard Co.
 *	David Mosberger-Tang <[e-mail]>
 * Copyright (C) 1999 VA Linux Systems
 * Copyright (C) 1999,2000 Walt Drummond <[e-mail]>
 *
 * 00/04/19	D. Mosberger	Rewritten to mirror more closely the x86 I/O APIC code.
 *				In particular, we now have separate handlers for edge
 *				and level triggered interrupts.
 * 00/10/27	Asit Mallick, Goutham Rao <[e-mail]> IRQ vector allocation
 *				PCI to vector mapping, shared PCI interrupts.
 * 00/10/27	D. Mosberger	Document things a bit more to make them more understandable.
 *				Clean up much of the old IOSAPIC cruft.
 * 01/07/27	J.I. Lee	PCI irq routing, Platform/Legacy interrupts and fixes for
 *				ACPI S5(SoftOff) support.
 * 02/01/23	J.I. Lee	iosapic pgm fixes for PCI irq routing from _PRT
 * 02/01/07     E. Focht        <[e-mail]> Redirectable interrupt vectors in
 *                              iosapic_set_affinity(), initializations for
 *                              /proc/irq/#/smp_affinity
 * 02/04/02	P. Diefenbaugh	Cleaned up ACPI PCI IRQ routing.
 * 02/04/18	J.I. Lee	bug fix in iosapic_init_pci_irq
 * 02/04/30	J.I. Lee	bug fix in find_iosapic to fix ACPI PCI IRQ to IOSAPIC mapping
 *				error
 * 02/07/29	T. Kochi	Allocate interrupt vectors dynamically
 * 02/08/04	T. Kochi	Cleaned up terminology (irq, global system interrupt, vector, etc.)
 * 02/09/20	D. Mosberger	Simplified by taking advantage of ACPI's pci_irq code.
 * 03/02/19	B. Helgaas	Make pcat_compat system-wide, not per-IOSAPIC.
 *				Remove iosapic_address & gsi_base from external interfaces.
 *				Rationalize __init/__devinit attributes.
 * 04/12/04 Ashok Raj	<[e-mail]> Intel Corporation 2004
 *				Updated to work with irq migration necessary for CPU Hotplug
 */
/*
 * Here is what the interrupt logic between a PCI device and the kernel looks like:
 *
 * (1) A PCI device raises one of the four interrupt pins (INTA, INTB, INTC, INTD).  The
 *     device is uniquely identified by its bus--, and slot-number (the function
 *     number does not matter here because all functions share the same interrupt
 *     lines).
 *
 * (2) The motherboard routes the interrupt line to a pin on a IOSAPIC controller.
 *     Multiple interrupt lines may have to share the same IOSAPIC pin (if they're level
 *     triggered and use the same polarity).  Each interrupt line has a unique Global
 *     System Interrupt (GSI) number which can be calculated as the sum of the controller's
 *     base GSI number and the IOSAPIC pin number to which the line connects.
 *
 * (3) The IOSAPIC uses an internal routing table entries (RTEs) to map the IOSAPIC pin
 *     into the IA-64 interrupt vector.  This interrupt vector is then sent to the CPU.
 *
 * (4) The kernel recognizes an interrupt as an IRQ.  The IRQ interface is used as
 *     architecture-independent interrupt handling mechanism in Linux.  As an
 *     IRQ is a number, we have to have IA-64 interrupt vector number <-> IRQ number
 *     mapping.  On smaller systems, we use one-to-one mapping between IA-64 vector and
 *     IRQ.  A platform can implement platform_irq_to_vector(irq) and
 *     platform_local_vector_to_irq(vector) APIs to differentiate the mapping.
 *     Please see also include/asm-ia64/hw_irq.h for those APIs.
 *
 * To sum up, there are three levels of mappings involved:
 *
 *	PCI pin -> global system interrupt (GSI) -> IA-64 vector <-> IRQ
 *
 * Note: The term "IRQ" is loosely used everywhere in Linux kernel to describe interrupts.
 * Now we use "IRQ" only for Linux IRQ's.  ISA IRQ (isa_irq) is the only exception in this
 * source code.
 */
#include <linux/config.h>

#include <linux/acpi.h>
#include <linux/init.h>
#include <linux/irq.h>
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/pci.h>
#ifdef XEN
#include <xen/errno.h>
#endif
#include <linux/smp.h>
#include <linux/smp_lock.h>
#include <linux/string.h>
#include <linux/bootmem.h>

#include <asm/delay.h>
#include <asm/hw_irq.h>
#include <asm/io.h>
#include <asm/iosapic.h>
#include <asm/machvec.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/system.h>


#undef DEBUG_INTERRUPT_ROUTING

#ifdef DEBUG_INTERRUPT_ROUTING
#define DBG(fmt...)	printk(fmt)
#else
#define DBG(fmt...)
#endif

#define NR_PREALLOCATE_RTE_ENTRIES	(PAGE_SIZE / sizeof(struct iosapic_rte_info))
#define RTE_PREALLOCATED	(1)

static DEFINE_SPINLOCK(iosapic_lock);

/* These tables map IA-64 vectors to the IOSAPIC pin that generates this vector. */

struct iosapic_rte_info {
	struct list_head rte_list;	/* node in list of RTEs sharing the same vector */
	char __iomem	*addr;		/* base address of IOSAPIC */
	unsigned int	gsi_base;	/* first GSI assigned to this IOSAPIC */
	char		rte_index;	/* IOSAPIC RTE index */
	int		refcnt;		/* reference counter */
	unsigned int	flags;		/* flags */
} ____cacheline_aligned;

static struct iosapic_intr_info {
	struct list_head rtes;		/* RTEs using this vector (empty => not an IOSAPIC interrupt) */
	int		count;		/* # of RTEs that shares this vector */
	u32		low32;		/* current value of low word of Redirection table entry */
	unsigned int	dest;		/* destination CPU physical ID */
	unsigned char	dmode	: 3;	/* delivery mode (see iosapic.h) */
	unsigned char 	polarity: 1;	/* interrupt polarity (see iosapic.h) */
	unsigned char	trigger	: 1;	/* trigger mode (see iosapic.h) */
} iosapic_intr_info[IA64_NUM_VECTORS];

static struct iosapic {
	char __iomem	*addr;		/* base address of IOSAPIC */
	unsigned int 	gsi_base;	/* first GSI assigned to this IOSAPIC */
	unsigned short 	num_rte;	/* number of RTE in this IOSAPIC */
	int		rtes_inuse;	/* # of RTEs in use on this IOSAPIC */
#ifdef CONFIG_NUMA
	unsigned short	node;		/* numa node association via pxm */
#endif
} iosapic_lists[NR_IOSAPICS];

static unsigned char pcat_compat __devinitdata;	/* 8259 compatibility flag */

static int iosapic_kmalloc_ok;
static LIST_HEAD(free_rte_list);

/*
 * Find an IOSAPIC associated with a GSI
 */
static inline int
find_iosapic (unsigned int gsi)
{
	int i;

	for (i = 0; i < NR_IOSAPICS; i++) {
		if ((unsigned) (gsi - iosapic_lists[i].gsi_base) < iosapic_lists[i].num_rte)
			return i;
	}

	return -1;
}

static inline int
_gsi_to_vector (unsigned int gsi)
{
	struct iosapic_intr_info *info;
	struct iosapic_rte_info *rte;

	for (info = iosapic_intr_info; info < iosapic_intr_info + IA64_NUM_VECTORS; ++info)
		list_for_each_entry(rte, &info->rtes, rte_list)
			if (rte->gsi_base + rte->rte_index == gsi)
				return info - iosapic_intr_info;
	return -1;
}

/*
 * Translate GSI number to the corresponding IA-64 interrupt vector.  If no
 * entry exists, return -1.
 */
inline int
gsi_to_vector (unsigned int gsi)
{
	return _gsi_to_vector(gsi);
}

int
gsi_to_irq (unsigned int gsi)
{
	unsigned long flags;
	int irq;
	/*
	 * XXX fix me: this assumes an identity mapping vetween IA-64 vector and Linux irq
	 * numbers...
	 */
	spin_lock_irqsave(&iosapic_lock, flags);
	{
		irq = _gsi_to_vector(gsi);
	}
	spin_unlock_irqrestore(&iosapic_lock, flags);

	return irq;
}

static struct iosapic_rte_info *gsi_vector_to_rte(unsigned int gsi, unsigned int vec)
{
	struct iosapic_rte_info *rte;

	list_for_each_entry(rte, &iosapic_intr_info[vec].rtes, rte_list)
		if (rte->gsi_base + rte->rte_index == gsi)
			return rte;
	return NULL;
}

static void
set_rte (unsigned int gsi, unsigned int vector, unsigned int dest, int mask)
{
	unsigned long pol, trigger, dmode;
	u32 low32, high32;
	char __iomem *addr;
	int rte_index;
	char redir;
	struct iosapic_rte_info *rte;

	DBG(KERN_DEBUG"IOSAPIC: routing vector %d to 0x%x\n", vector, dest);

	rte = gsi_vector_to_rte(gsi, vector);
	if (!rte)
		return;		/* not an IOSAPIC interrupt */

	rte_index = rte->rte_index;
	addr	= rte->addr;
	pol     = iosapic_intr_info[vector].polarity;
	trigger = iosapic_intr_info[vector].trigger;
	dmode   = iosapic_intr_info[vector].dmode;

	redir = (dmode == IOSAPIC_LOWEST_PRIORITY) ? 1 : 0;

#ifdef CONFIG_SMP
	{
		unsigned int irq;

		for (irq = 0; irq < NR_IRQS; ++irq)
			if (irq_to_vector(irq) == vector) {
				set_irq_affinity_info(irq, (int)(dest & 0xffff), redir);
				break;
			}
	}
#endif

	low32 = ((pol << IOSAPIC_POLARITY_SHIFT) |
		 (trigger << IOSAPIC_TRIGGER_SHIFT) |
		 (dmode << IOSAPIC_DELIVERY_SHIFT) |
		 ((mask ? 1 : 0) << IOSAPIC_MASK_SHIFT) |
		 vector);

	/* dest contains both id and eid */
	high32 = (dest << IOSAPIC_DEST_SHIFT);

	iosapic_write(addr, IOSAPIC_RTE_HIGH(rte_index), high32);
	iosapic_write(addr, IOSAPIC_RTE_LOW(rte_index), low32);
	iosapic_intr_info[vector].low32 = low32;
	iosapic_intr_info[vector].dest = dest;
}

static void
nop (unsigned int vector)
{
	/* do nothing... */
}

void
kexec_disable_iosapic(void)
{
        struct iosapic_intr_info *info;
        struct iosapic_rte_info *rte;
        u8 vec = 0;
        for (info = iosapic_intr_info; info <
                        iosapic_intr_info + IA64_NUM_VECTORS; ++info, ++vec) {
                list_for_each_entry(rte, &info->rtes,
                                rte_list) {
                        iosapic_write(rte->addr,
                                        IOSAPIC_RTE_LOW(rte->rte_index),
                                        IOSAPIC_MASK|vec);
                        iosapic_eoi(rte->addr, vec);
                }
        }
}

static void
mask_irq (unsigned int irq)
{
	unsigned long flags;
	char __iomem *addr;
	u32 low32;
	int rte_index;
	ia64_vector vec = irq_to_vector(irq);
	struct iosapic_rte_info *rte;

	if (list_empty(&iosapic_intr_info[vec].rtes))
		return;			/* not an IOSAPIC interrupt! */

	spin_lock_irqsave(&iosapic_lock, flags);
	{
		/* set only the mask bit */
		low32 = iosapic_intr_info[vec].low32 |= IOSAPIC_MASK;
		list_for_each_entry(rte, &iosapic_intr_info[vec].rtes, rte_list) {
			addr = rte->addr;
			rte_index = rte->rte_index;
			iosapic_write(addr, IOSAPIC_RTE_LOW(rte_index), low32);
		}
	}
	spin_unlock_irqrestore(&iosapic_lock, flags);
}

static void
unmask_irq (unsigned int irq)
{
	unsigned long flags;
	char __iomem *addr;
	u32 low32;
	int rte_index;
	ia64_vector vec = irq_to_vector(irq);
	struct iosapic_rte_info *rte;

	if (list_empty(&iosapic_intr_info[vec].rtes))
		return;			/* not an IOSAPIC interrupt! */

	spin_lock_irqsave(&iosapic_lock, flags);
	{
		low32 = iosapic_intr_info[vec].low32 &= ~IOSAPIC_MASK;
		list_for_each_entry(rte, &iosapic_intr_info[vec].rtes, rte_list) {
			addr = rte->addr;
			rte_index = rte->rte_index;
			iosapic_write(addr, IOSAPIC_RTE_LOW(rte_index), low32);
		}
	}
	spin_unlock_irqrestore(&iosapic_lock, flags);
}


static void
iosapic_set_affinity (unsigned int irq, cpumask_t mask)
{
#ifdef CONFIG_SMP
	unsigned long flags;
	u32 high32, low32;
	int dest, rte_index;
	char __iomem *addr;
	int redir = (irq & IA64_IRQ_REDIRECTED) ? 1 : 0;
	ia64_vector vec;
	struct iosapic_rte_info *rte;

	irq &= (~IA64_IRQ_REDIRECTED);
	vec = irq_to_vector(irq);

	if (cpus_empty(mask))
		return;

	dest = cpu_physical_id(first_cpu(mask));

	if (list_empty(&iosapic_intr_info[vec].rtes))
		return;			/* not an IOSAPIC interrupt */

	set_irq_affinity_info(irq, dest, redir);

	/* dest contains both id and eid */
	high32 = dest << IOSAPIC_DEST_SHIFT;

	spin_lock_irqsave(&iosapic_lock, flags);
	{
		low32 = iosapic_intr_info[vec].low32 & ~(7 << IOSAPIC_DELIVERY_SHIFT);

		if (redir)
		        /* change delivery mode to lowest priority */
			low32 |= (IOSAPIC_LOWEST_PRIORITY << IOSAPIC_DELIVERY_SHIFT);
		else
		        /* change delivery mode to fixed */
			low32 |= (IOSAPIC_FIXED << IOSAPIC_DELIVERY_SHIFT);

		iosapic_intr_info[vec].low32 = low32;
		iosapic_intr_info[vec].dest = dest;
		list_for_each_entry(rte, &iosapic_intr_info[vec].rtes, rte_list) {
			addr = rte->addr;
			rte_index = rte->rte_index;
			iosapic_write(addr, IOSAPIC_RTE_HIGH(rte_index), high32);
			iosapic_write(addr, IOSAPIC_RTE_LOW(rte_index), low32);
		}
	}
	spin_unlock_irqrestore(&iosapic_lock, flags);
#endif
}

/*
 * Handlers for level-triggered interrupts.
 */

static unsigned int
iosapic_startup_level_irq (unsigned int irq)
{
	unmask_irq(irq);
	return 0;
}

static void
iosapic_end_level_irq (unsigned int irq)
{
	ia64_vector vec = irq_to_vector(irq);
	struct iosapic_rte_info *rte;

	move_irq(irq);
	list_for_each_entry(rte, &iosapic_intr_info[vec].rtes, rte_list)
		iosapic_eoi(rte->addr, vec);
}

#define iosapic_shutdown_level_irq	mask_irq
#define iosapic_enable_level_irq	unmask_irq
#define iosapic_disable_level_irq	mask_irq
#define iosapic_ack_level_irq		nop

struct hw_interrupt_type irq_type_iosapic_level = {
	.typename =	"IO-SAPIC-level",