linux/arch/x86_64/pci/mmconfig.c

/*
 * mmconfig.c - Low-level direct PCI config space access via MMCONFIG
 *
 * This is an 64bit optimized version that always keeps the full mmconfig
 * space mapped. This allows lockless config space operation.
 */

#include <linux/pci.h>
#include <linux/init.h>
#include <linux/acpi.h>
#include <linux/bitmap.h>
#include <asm/e820.h>

#include "pci.h"

/* aperture is up to 256MB but BIOS may reserve less */
#define MMCONFIG_APER_MIN	(2 * 1024*1024)
#define MMCONFIG_APER_MAX	(256 * 1024*1024)

/* Verify the first 16 busses. We assume that systems with more busses
   get MCFG right. */
#define MAX_CHECK_BUS 16

static DECLARE_BITMAP(fallback_slots, 32*MAX_CHECK_BUS);

/* Static virtual mapping of the MMCONFIG aperture */
struct mmcfg_virt {
	struct acpi_mcfg_allocation *cfg;
	char __iomem *virt;
};
static struct mmcfg_virt *pci_mmcfg_virt;

static char __iomem *get_virt(unsigned int seg, unsigned bus)
{
	int cfg_num = -1;
	struct acpi_mcfg_allocation *cfg;

	while (1) {
		++cfg_num;
		if (cfg_num >= pci_mmcfg_config_num)
			break;
		cfg = pci_mmcfg_virt[cfg_num].cfg;
		if (cfg->pci_segment != seg)
			continue;
		if ((cfg->start_bus_number <= bus) &&
		    (cfg->end_bus_number >= bus))
			return pci_mmcfg_virt[cfg_num].virt;
	}

	/* Handle more broken MCFG tables on Asus etc.
	   They only contain a single entry for bus 0-0. Assume
 	   this applies to all busses. */
	cfg = &pci_mmcfg_config[0];
	if (pci_mmcfg_config_num == 1 &&
		cfg->pci_segment == 0 &&
		(cfg->start_bus_number | cfg->end_bus_number) == 0)
		return pci_mmcfg_virt[0].virt;

	/* Fall back to type 0 */
	return NULL;
}

static char __iomem *pci_dev_base(unsigned int seg, unsigned int bus, unsigned int devfn)
{
	char __iomem *addr;
	if (seg == 0 && bus < MAX_CHECK_BUS &&
		test_bit(32*bus + PCI_SLOT(devfn), fallback_slots))
		return NULL;
	addr = get_virt(seg, bus);
	if (!addr)
		return NULL;
 	return addr + ((bus << 20) | (devfn << 12));
}

static int pci_mmcfg_read(unsigned int seg, unsigned int bus,
			  unsigned int devfn, int reg, int len, u32 *value)
{
	char __iomem *addr;

	/* Why do we have this when nobody checks it. How about a BUG()!? -AK */
	if (unlikely((bus > 255) || (devfn > 255) || (reg > 4095))) {
		*value = -1;
		return -EINVAL;
	}

	addr = pci_dev_base(seg, bus, devfn);
	if (!addr)
		return pci_conf1_read(seg,bus,devfn,reg,len,value);

	switch (len) {
	case 1:
		*value = readb(addr + reg);
		break;
	case 2:
		*value = readw(addr + reg);
		break;
	case 4:
		*value = readl(addr + reg);
		break;
	}

	return 0;
}

static int pci_mmcfg_write(unsigned int seg, unsigned int bus,
			   unsigned int devfn, int reg, int len, u32 value)
{
	char __iomem *addr;

	/* Why do we have this when nobody checks it. How about a BUG()!? -AK */
	if (unlikely((bus > 255) || (devfn > 255) || (reg > 4095)))
		return -EINVAL;

	addr = pci_dev_base(seg, bus, devfn);
	if (!addr)
		return pci_conf1_write(seg,bus,devfn,reg,len,value);

	switch (len) {
	case 1:
		writeb(value, addr + reg);
		break;
	case 2:
		writew(value, addr + reg);
		break;
	case 4:
		writel(value, addr + reg);
		break;
	}

	return 0;
}

static struct pci_raw_ops pci_mmcfg = {
	.read =		pci_mmcfg_read,
	.write =	pci_mmcfg_write,
};

/* K8 systems have some devices (typically in the builtin northbridge)
   that are only accessible using type1
   Normally this can be expressed in the MCFG by not listing them
   and assigning suitable _SEGs, but this isn't implemented in some BIOS.
   Instead try to discover all devices on bus 0 that are unreachable using MM
   and fallback for them. */
static __init void unreachable_devices(void)
{
	int i, k;
	/* Use the max bus number from ACPI here? */
	for (k = 0; k < MAX_CHECK_BUS; k++) {
		for (i = 0; i < 32; i++) {
			u32 val1;
			char __iomem *addr;

			pci_conf1_read(0, k, PCI_DEVFN(i,0), 0, 4, &val1);
			if (val1 == 0xffffffff)
				continue;
			addr = pci_dev_base(0, k, PCI_DEVFN(i, 0));
			if (addr == NULL|| readl(addr) != val1) {
				set_bit(i + 32*k, fallback_slots);
				printk(KERN_NOTICE "PCI: No mmconfig possible"
				       " on device %02x:%02x\n", k, i);
			}
		}
	}
}

void __init pci_mmcfg_init(int type)
{
	int i;

	if ((pci_probe & PCI_PROBE_MMCONF) == 0)
		return;

	acpi_table_parse(ACPI_SIG_MCFG, acpi_parse_mcfg);
	if ((pci_mmcfg_config_num == 0) ||
	    (pci_mmcfg_config == NULL) ||
	    (pci_mmcfg_config[0].address == 0))
		return;

	/* Only do this check when type 1 works. If it doesn't work
           assume we run on a Mac and always use MCFG */
	if (type == 1 && !e820_all_mapped(pci_mmcfg_config[0].address,
			pci_mmcfg_config[0].address + MMCONFIG_APER_MIN,
			E820_RESERVED)) {
		printk(KERN_ERR "PCI: BIOS Bug: MCFG area at %lx is not E820-reserved\n",
				(unsigned long)pci_mmcfg_config[0].address);
		printk(KERN_ERR "PCI: Not using MMCONFIG.\n");
		return;
	}

	pci_mmcfg_virt = kmalloc(sizeof(*pci_mmcfg_virt) * pci_mmcfg_config_num, GFP_KERNEL);
	if (pci_mmcfg_virt == NULL) {
		printk(KERN_ERR "PCI: Can not allocate memory for mmconfig structures\n");
		return;
	}
	for (i = 0; i < pci_mmcfg_config_num; ++i) {
		pci_mmcfg_virt[i].cfg = &pci_mmcfg_config[i];
		pci_mmcfg_virt[i].virt = ioremap_nocache(pci_mmcfg_config[i].address,
							 MMCONFIG_APER_MAX);
		if (!pci_mmcfg_virt[i].virt) {
			printk(KERN_ERR "PCI: Cannot map mmconfig aperture for "
					"segment %d\n",
				pci_mmcfg_config[i].pci_segment);
			return;
		}
		printk(KERN_INFO "PCI: Using MMCONFIG at %lx\n",
			(unsigned long)pci_mmcfg_config[i].address);
	}

	unreachable_devices();

	raw_pci_ops = &pci_mmcfg;
	pci_probe = (pci_probe & ~PCI_PROBE_MASK) | PCI_PROBE_MMCONF;
}