xref: /openbmc/linux/drivers/edac/amd64_edac.c (revision d16149e8)

#include "amd64_edac.h"
#include <asm/k8.h>

static struct edac_pci_ctl_info *amd64_ctl_pci;

static int report_gart_errors;
module_param(report_gart_errors, int, 0644);

/*
 * Set by command line parameter. If BIOS has enabled the ECC, this override is
 * cleared to prevent re-enabling the hardware by this driver.
 */
static int ecc_enable_override;
module_param(ecc_enable_override, int, 0644);

/* Lookup table for all possible MC control instances */
struct amd64_pvt;
static struct mem_ctl_info *mci_lookup[EDAC_MAX_NUMNODES];
static struct amd64_pvt *pvt_lookup[EDAC_MAX_NUMNODES];

/*
 * See F2x80 for K8 and F2x[1,0]80 for Fam10 and later. The table below is only
 * for DDR2 DRAM mapping.
 */
u32 revf_quad_ddr2_shift[] = {
	0,	/* 0000b NULL DIMM (128mb) */
	28,	/* 0001b 256mb */
	29,	/* 0010b 512mb */
	29,	/* 0011b 512mb */
	29,	/* 0100b 512mb */
	30,	/* 0101b 1gb */
	30,	/* 0110b 1gb */
	31,	/* 0111b 2gb */
	31,	/* 1000b 2gb */
	32,	/* 1001b 4gb */
	32,	/* 1010b 4gb */
	33,	/* 1011b 8gb */
	0,	/* 1100b future */
	0,	/* 1101b future */
	0,	/* 1110b future */
	0	/* 1111b future */
};

/*
 * Valid scrub rates for the K8 hardware memory scrubber. We map the scrubbing
 * bandwidth to a valid bit pattern. The 'set' operation finds the 'matching-
 * or higher value'.
 *
 *FIXME: Produce a better mapping/linearisation.
 */

struct scrubrate scrubrates[] = {
	{ 0x01, 1600000000UL},
	{ 0x02, 800000000UL},
	{ 0x03, 400000000UL},
	{ 0x04, 200000000UL},
	{ 0x05, 100000000UL},
	{ 0x06, 50000000UL},
	{ 0x07, 25000000UL},
	{ 0x08, 12284069UL},
	{ 0x09, 6274509UL},
	{ 0x0A, 3121951UL},
	{ 0x0B, 1560975UL},
	{ 0x0C, 781440UL},
	{ 0x0D, 390720UL},
	{ 0x0E, 195300UL},
	{ 0x0F, 97650UL},
	{ 0x10, 48854UL},
	{ 0x11, 24427UL},
	{ 0x12, 12213UL},
	{ 0x13, 6101UL},
	{ 0x14, 3051UL},
	{ 0x15, 1523UL},
	{ 0x16, 761UL},
	{ 0x00, 0UL},        /* scrubbing off */
};

/*
 * Memory scrubber control interface. For K8, memory scrubbing is handled by
 * hardware and can involve L2 cache, dcache as well as the main memory. With
 * F10, this is extended to L3 cache scrubbing on CPU models sporting that
 * functionality.
 *
 * This causes the "units" for the scrubbing speed to vary from 64 byte blocks
 * (dram) over to cache lines. This is nasty, so we will use bandwidth in
 * bytes/sec for the setting.
 *
 * Currently, we only do dram scrubbing. If the scrubbing is done in software on
 * other archs, we might not have access to the caches directly.
 */

/*
 * scan the scrub rate mapping table for a close or matching bandwidth value to
 * issue. If requested is too big, then use last maximum value found.
 */
static int amd64_search_set_scrub_rate(struct pci_dev *ctl, u32 new_bw,
				       u32 min_scrubrate)
{
	u32 scrubval;
	int i;

	/*
	 * map the configured rate (new_bw) to a value specific to the AMD64
	 * memory controller and apply to register. Search for the first
	 * bandwidth entry that is greater or equal than the setting requested
	 * and program that. If at last entry, turn off DRAM scrubbing.
	 */
	for (i = 0; i < ARRAY_SIZE(scrubrates); i++) {
		/*
		 * skip scrub rates which aren't recommended
		 * (see F10 BKDG, F3x58)
		 */
		if (scrubrates[i].scrubval < min_scrubrate)
			continue;

		if (scrubrates[i].bandwidth <= new_bw)
			break;

		/*
		 * if no suitable bandwidth found, turn off DRAM scrubbing
		 * entirely by falling back to the last element in the
		 * scrubrates array.
		 */
	}

	scrubval = scrubrates[i].scrubval;
	if (scrubval)
		edac_printk(KERN_DEBUG, EDAC_MC,
			    "Setting scrub rate bandwidth: %u\n",
			    scrubrates[i].bandwidth);
	else
		edac_printk(KERN_DEBUG, EDAC_MC, "Turning scrubbing off.\n");

	pci_write_bits32(ctl, K8_SCRCTRL, scrubval, 0x001F);

	return 0;
}

static int amd64_set_scrub_rate(struct mem_ctl_info *mci, u32 *bandwidth)
{
	struct amd64_pvt *pvt = mci->pvt_info;
	u32 min_scrubrate = 0x0;

	switch (boot_cpu_data.x86) {
	case 0xf:
		min_scrubrate = K8_MIN_SCRUB_RATE_BITS;
		break;
	case 0x10:
		min_scrubrate = F10_MIN_SCRUB_RATE_BITS;
		break;
	case 0x11:
		min_scrubrate = F11_MIN_SCRUB_RATE_BITS;
		break;

	default:
		amd64_printk(KERN_ERR, "Unsupported family!\n");
		break;
	}
	return amd64_search_set_scrub_rate(pvt->misc_f3_ctl, *bandwidth,
			min_scrubrate);
}

static int amd64_get_scrub_rate(struct mem_ctl_info *mci, u32 *bw)
{
	struct amd64_pvt *pvt = mci->pvt_info;
	u32 scrubval = 0;
	int status = -1, i;

	amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_SCRCTRL, &scrubval);

	scrubval = scrubval & 0x001F;

	edac_printk(KERN_DEBUG, EDAC_MC,
		    "pci-read, sdram scrub control value: %d \n", scrubval);

	for (i = 0; ARRAY_SIZE(scrubrates); i++) {
		if (scrubrates[i].scrubval == scrubval) {
			*bw = scrubrates[i].bandwidth;
			status = 0;
			break;
		}
	}

	return status;
}

/* Map from a CSROW entry to the mask entry that operates on it */
static inline u32 amd64_map_to_dcs_mask(struct amd64_pvt *pvt, int csrow)
{
	if (boot_cpu_data.x86 == 0xf && pvt->ext_model < OPTERON_CPU_REV_F)
		return csrow;
	else
		return csrow >> 1;
}

/* return the 'base' address the i'th CS entry of the 'dct' DRAM controller */
static u32 amd64_get_dct_base(struct amd64_pvt *pvt, int dct, int csrow)
{
	if (dct == 0)
		return pvt->dcsb0[csrow];
	else
		return pvt->dcsb1[csrow];
}

/*
 * Return the 'mask' address the i'th CS entry. This function is needed because
 * there number of DCSM registers on Rev E and prior vs Rev F and later is
 * different.
 */
static u32 amd64_get_dct_mask(struct amd64_pvt *pvt, int dct, int csrow)
{
	if (dct == 0)
		return pvt->dcsm0[amd64_map_to_dcs_mask(pvt, csrow)];
	else
		return pvt->dcsm1[amd64_map_to_dcs_mask(pvt, csrow)];
}


/*
 * In *base and *limit, pass back the full 40-bit base and limit physical
 * addresses for the node given by node_id.  This information is obtained from
 * DRAM Base (section 3.4.4.1) and DRAM Limit (section 3.4.4.2) registers. The
 * base and limit addresses are of type SysAddr, as defined at the start of
 * section 3.4.4 (p. 70).  They are the lowest and highest physical addresses
 * in the address range they represent.
 */
static void amd64_get_base_and_limit(struct amd64_pvt *pvt, int node_id,
			       u64 *base, u64 *limit)
{
	*base = pvt->dram_base[node_id];
	*limit = pvt->dram_limit[node_id];
}

/*
 * Return 1 if the SysAddr given by sys_addr matches the base/limit associated
 * with node_id
 */
static int amd64_base_limit_match(struct amd64_pvt *pvt,
					u64 sys_addr, int node_id)
{
	u64 base, limit, addr;

	amd64_get_base_and_limit(pvt, node_id, &base, &limit);

	/* The K8 treats this as a 40-bit value.  However, bits 63-40 will be
	 * all ones if the most significant implemented address bit is 1.
	 * Here we discard bits 63-40.  See section 3.4.2 of AMD publication
	 * 24592: AMD x86-64 Architecture Programmer's Manual Volume 1
	 * Application Programming.
	 */
	addr = sys_addr & 0x000000ffffffffffull;

	return (addr >= base) && (addr <= limit);
}

/*
 * Attempt to map a SysAddr to a node. On success, return a pointer to the
 * mem_ctl_info structure for the node that the SysAddr maps to.
 *
 * On failure, return NULL.
 */
static struct mem_ctl_info *find_mc_by_sys_addr(struct mem_ctl_info *mci,
						u64 sys_addr)
{
	struct amd64_pvt *pvt;
	int node_id;
	u32 intlv_en, bits;

	/*
	 * Here we use the DRAM Base (section 3.4.4.1) and DRAM Limit (section
	 * 3.4.4.2) registers to map the SysAddr to a node ID.
	 */
	pvt = mci->pvt_info;

	/*
	 * The value of this field should be the same for all DRAM Base
	 * registers.  Therefore we arbitrarily choose to read it from the
	 * register for node 0.
	 */
	intlv_en = pvt->dram_IntlvEn[0];

	if (intlv_en == 0) {
		for (node_id = 0; node_id < DRAM_REG_COUNT; node_id++) {
			if (amd64_base_limit_match(pvt, sys_addr, node_id))
				goto found;
		}
		goto err_no_match;
	}

	if (unlikely((intlv_en != 0x01) &&
		     (intlv_en != 0x03) &&
		     (intlv_en != 0x07))) {
		amd64_printk(KERN_WARNING, "junk value of 0x%x extracted from "
			     "IntlvEn field of DRAM Base Register for node 0: "
			     "this probably indicates a BIOS bug.\n", intlv_en);
		return NULL;
	}

	bits = (((u32) sys_addr) >> 12) & intlv_en;

	for (node_id = 0; ; ) {
		if ((pvt->dram_IntlvSel[node_id] & intlv_en) == bits)
			break;	/* intlv_sel field matches */

		if (++node_id >= DRAM_REG_COUNT)
			goto err_no_match;
	}

	/* sanity test for sys_addr */
	if (unlikely(!amd64_base_limit_match(pvt, sys_addr, node_id))) {
		amd64_printk(KERN_WARNING,
			     "%s(): sys_addr 0x%llx falls outside base/limit "
			     "address range for node %d with node interleaving "
			     "enabled.\n",
			     __func__, sys_addr, node_id);
		return NULL;
	}

found:
	return edac_mc_find(node_id);

err_no_match:
	debugf2("sys_addr 0x%lx doesn't match any node\n",
		(unsigned long)sys_addr);

	return NULL;
}

/*
 * Extract the DRAM CS base address from selected csrow register.
 */
static u64 base_from_dct_base(struct amd64_pvt *pvt, int csrow)
{
	return ((u64) (amd64_get_dct_base(pvt, 0, csrow) & pvt->dcsb_base)) <<
				pvt->dcs_shift;
}

/*
 * Extract the mask from the dcsb0[csrow] entry in a CPU revision-specific way.
 */
static u64 mask_from_dct_mask(struct amd64_pvt *pvt, int csrow)
{
	u64 dcsm_bits, other_bits;
	u64 mask;

	/* Extract bits from DRAM CS Mask. */
	dcsm_bits = amd64_get_dct_mask(pvt, 0, csrow) & pvt->dcsm_mask;

	other_bits = pvt->dcsm_mask;
	other_bits = ~(other_bits << pvt->dcs_shift);

	/*
	 * The extracted bits from DCSM belong in the spaces represented by
	 * the cleared bits in other_bits.
	 */
	mask = (dcsm_bits << pvt->dcs_shift) | other_bits;

	return mask;
}

/*
 * @input_addr is an InputAddr associated with the node given by mci. Return the
 * csrow that input_addr maps to, or -1 on failure (no csrow claims input_addr).
 */
static int input_addr_to_csrow(struct mem_ctl_info *mci, u64 input_addr)
{
	struct amd64_pvt *pvt;
	int csrow;
	u64 base, mask;

	pvt = mci->pvt_info;

	/*
	 * Here we use the DRAM CS Base and DRAM CS Mask registers. For each CS
	 * base/mask register pair, test the condition shown near the start of
	 * section 3.5.4 (p. 84, BKDG #26094, K8, revA-E).
	 */
	for (csrow = 0; csrow < pvt->cs_count; csrow++) {

		/* This DRAM chip select is disabled on this node */
		if ((pvt->dcsb0[csrow] & K8_DCSB_CS_ENABLE) == 0)
			continue;

		base = base_from_dct_base(pvt, csrow);
		mask = ~mask_from_dct_mask(pvt, csrow);

		if ((input_addr & mask) == (base & mask)) {
			debugf2("InputAddr 0x%lx matches csrow %d (node %d)\n",
				(unsigned long)input_addr, csrow,
				pvt->mc_node_id);

			return csrow;
		}
	}

	debugf2("no matching csrow for InputAddr 0x%lx (MC node %d)\n",
		(unsigned long)input_addr, pvt->mc_node_id);

	return -1;
}

/*
 * Return the base value defined by the DRAM Base register for the node
 * represented by mci.  This function returns the full 40-bit value despite the
 * fact that the register only stores bits 39-24 of the value. See section
 * 3.4.4.1 (BKDG #26094, K8, revA-E)
 */
static inline u64 get_dram_base(struct mem_ctl_info *mci)
{
	struct amd64_pvt *pvt = mci->pvt_info;

	return pvt->dram_base[pvt->mc_node_id];
}

/*
 * Obtain info from the DRAM Hole Address Register (section 3.4.8, pub #26094)
 * for the node represented by mci. Info is passed back in *hole_base,
 * *hole_offset, and *hole_size.  Function returns 0 if info is valid or 1 if
 * info is invalid. Info may be invalid for either of the following reasons:
 *
 * - The revision of the node is not E or greater.  In this case, the DRAM Hole
 *   Address Register does not exist.
 *
 * - The DramHoleValid bit is cleared in the DRAM Hole Address Register,
 *   indicating that its contents are not valid.
 *
 * The values passed back in *hole_base, *hole_offset, and *hole_size are
 * complete 32-bit values despite the fact that the bitfields in the DHAR
 * only represent bits 31-24 of the base and offset values.
 */
int amd64_get_dram_hole_info(struct mem_ctl_info *mci, u64 *hole_base,
			     u64 *hole_offset, u64 *hole_size)
{
	struct amd64_pvt *pvt = mci->pvt_info;
	u64 base;

	/* only revE and later have the DRAM Hole Address Register */
	if (boot_cpu_data.x86 == 0xf && pvt->ext_model < OPTERON_CPU_REV_E) {
		debugf1("  revision %d for node %d does not support DHAR\n",
			pvt->ext_model, pvt->mc_node_id);
		return 1;
	}

	/* only valid for Fam10h */
	if (boot_cpu_data.x86 == 0x10 &&
	    (pvt->dhar & F10_DRAM_MEM_HOIST_VALID) == 0) {
		debugf1("  Dram Memory Hoisting is DISABLED on this system\n");
		return 1;
	}

	if ((pvt->dhar & DHAR_VALID) == 0) {
		debugf1("  Dram Memory Hoisting is DISABLED on this node %d\n",
			pvt->mc_node_id);
		return 1;
	}

	/* This node has Memory Hoisting */

	/* +------------------+--------------------+--------------------+-----
	 * | memory           | DRAM hole          | relocated          |
	 * | [0, (x - 1)]     | [x, 0xffffffff]    | addresses from     |
	 * |                  |                    | DRAM hole          |
	 * |                  |                    | [0x100000000,      |
	 * |                  |                    |  (0x100000000+     |
	 * |                  |                    |   (0xffffffff-x))] |
	 * +------------------+--------------------+--------------------+-----
	 *
	 * Above is a diagram of physical memory showing the DRAM hole and the
	 * relocated addresses from the DRAM hole.  As shown, the DRAM hole
	 * starts at address x (the base address) and extends through address
	 * 0xffffffff.  The DRAM Hole Address Register (DHAR) relocates the
	 * addresses in the hole so that they start at 0x100000000.
	 */

	base = dhar_base(pvt->dhar);

	*hole_base = base;
	*hole_size = (0x1ull << 32) - base;

	if (boot_cpu_data.x86 > 0xf)
		*hole_offset = f10_dhar_offset(pvt->dhar);
	else
		*hole_offset = k8_dhar_offset(pvt->dhar);

	debugf1("  DHAR info for node %d base 0x%lx offset 0x%lx size 0x%lx\n",
		pvt->mc_node_id, (unsigned long)*hole_base,
		(unsigned long)*hole_offset, (unsigned long)*hole_size);

	return 0;
}
EXPORT_SYMBOL_GPL(amd64_get_dram_hole_info);

/*
 * Return the DramAddr that the SysAddr given by @sys_addr maps to.  It is
 * assumed that sys_addr maps to the node given by mci.
 *
 * The first part of section 3.4.4 (p. 70) shows how the DRAM Base (section
 * 3.4.4.1) and DRAM Limit (section 3.4.4.2) registers are used to translate a
 * SysAddr to a DramAddr. If the DRAM Hole Address Register (DHAR) is enabled,
 * then it is also involved in translating a SysAddr to a DramAddr. Sections
 * 3.4.8 and 3.5.8.2 describe the DHAR and how it is used for memory hoisting.
 * These parts of the documentation are unclear. I interpret them as follows:
 *
 * When node n receives a SysAddr, it processes the SysAddr as follows:
 *
 * 1. It extracts the DRAMBase and DRAMLimit values from the DRAM Base and DRAM
 *    Limit registers for node n. If the SysAddr is not within the range
 *    specified by the base and limit values, then node n ignores the Sysaddr
 *    (since it does not map to node n). Otherwise continue to step 2 below.
 *
 * 2. If the DramHoleValid bit of the DHAR for node n is clear, the DHAR is
 *    disabled so skip to step 3 below. Otherwise see if the SysAddr is within
 *    the range of relocated addresses (starting at 0x100000000) from the DRAM
 *    hole. If not, skip to step 3 below. Else get the value of the
 *    DramHoleOffset field from the DHAR. To obtain the DramAddr, subtract the
 *    offset defined by this value from the SysAddr.
 *
 * 3. Obtain the base address for node n from the DRAMBase field of the DRAM
 *    Base register for node n. To obtain the DramAddr, subtract the base
 *    address from the SysAddr, as shown near the start of section 3.4.4 (p.70).
 */
static u64 sys_addr_to_dram_addr(struct mem_ctl_info *mci, u64 sys_addr)
{
	u64 dram_base, hole_base, hole_offset, hole_size, dram_addr;
	int ret = 0;

	dram_base = get_dram_base(mci);

	ret = amd64_get_dram_hole_info(mci, &hole_base, &hole_offset,
				      &hole_size);
	if (!ret) {
		if ((sys_addr >= (1ull << 32)) &&
		    (sys_addr < ((1ull << 32) + hole_size))) {
			/* use DHAR to translate SysAddr to DramAddr */
			dram_addr = sys_addr - hole_offset;

			debugf2("using DHAR to translate SysAddr 0x%lx to "
				"DramAddr 0x%lx\n",
				(unsigned long)sys_addr,
				(unsigned long)dram_addr);

			return dram_addr;
		}
	}

	/*
	 * Translate the SysAddr to a DramAddr as shown near the start of
	 * section 3.4.4 (p. 70).  Although sys_addr is a 64-bit value, the k8
	 * only deals with 40-bit values.  Therefore we discard bits 63-40 of
	 * sys_addr below.  If bit 39 of sys_addr is 1 then the bits we
	 * discard are all 1s.  Otherwise the bits we discard are all 0s.  See
	 * section 3.4.2 of AMD publication 24592: AMD x86-64 Architecture
	 * Programmer's Manual Volume 1 Application Programming.
	 */
	dram_addr = (sys_addr & 0xffffffffffull) - dram_base;

	debugf2("using DRAM Base register to translate SysAddr 0x%lx to "
		"DramAddr 0x%lx\n", (unsigned long)sys_addr,
		(unsigned long)dram_addr);
	return dram_addr;
}

/*
 * @intlv_en is the value of the IntlvEn field from a DRAM Base register
 * (section 3.4.4.1).  Return the number of bits from a SysAddr that are used
 * for node interleaving.
 */
static int num_node_interleave_bits(unsigned intlv_en)
{
	static const int intlv_shift_table[] = { 0, 1, 0, 2, 0, 0, 0, 3 };
	int n;

	BUG_ON(intlv_en > 7);
	n = intlv_shift_table[intlv_en];
	return n;
}

/* Translate the DramAddr given by @dram_addr to an InputAddr. */
static u64 dram_addr_to_input_addr(struct mem_ctl_info *mci, u64 dram_addr)
{
	struct amd64_pvt *pvt;
	int intlv_shift;
	u64 input_addr;

	pvt = mci->pvt_info;

	/*
	 * See the start of section 3.4.4 (p. 70, BKDG #26094, K8, revA-E)
	 * concerning translating a DramAddr to an InputAddr.
	 */
	intlv_shift = num_node_interleave_bits(pvt->dram_IntlvEn[0]);
	input_addr = ((dram_addr >> intlv_shift) & 0xffffff000ull) +
	    (dram_addr & 0xfff);

	debugf2("  Intlv Shift=%d DramAddr=0x%lx maps to InputAddr=0x%lx\n",
		intlv_shift, (unsigned long)dram_addr,
		(unsigned long)input_addr);

	return input_addr;
}

/*
 * Translate the SysAddr represented by @sys_addr to an InputAddr.  It is
 * assumed that @sys_addr maps to the node given by mci.
 */
static u64 sys_addr_to_input_addr(struct mem_ctl_info *mci, u64 sys_addr)
{
	u64 input_addr;

	input_addr =
	    dram_addr_to_input_addr(mci, sys_addr_to_dram_addr(mci, sys_addr));

	debugf2("SysAdddr 0x%lx translates to InputAddr 0x%lx\n",
		(unsigned long)sys_addr, (unsigned long)input_addr);

	return input_addr;
}


/*
 * @input_addr is an InputAddr associated with the node represented by mci.
 * Translate @input_addr to a DramAddr and return the result.
 */
static u64 input_addr_to_dram_addr(struct mem_ctl_info *mci, u64 input_addr)
{
	struct amd64_pvt *pvt;
	int node_id, intlv_shift;
	u64 bits, dram_addr;
	u32 intlv_sel;

	/*
	 * Near the start of section 3.4.4 (p. 70, BKDG #26094, K8, revA-E)
	 * shows how to translate a DramAddr to an InputAddr. Here we reverse
	 * this procedure. When translating from a DramAddr to an InputAddr, the
	 * bits used for node interleaving are discarded.  Here we recover these
	 * bits from the IntlvSel field of the DRAM Limit register (section
	 * 3.4.4.2) for the node that input_addr is associated with.
	 */
	pvt = mci->pvt_info;
	node_id = pvt->mc_node_id;
	BUG_ON((node_id < 0) || (node_id > 7));

	intlv_shift = num_node_interleave_bits(pvt->dram_IntlvEn[0]);

	if (intlv_shift == 0) {
		debugf1("    InputAddr 0x%lx translates to DramAddr of "
			"same value\n",	(unsigned long)input_addr);

		return input_addr;
	}

	bits = ((input_addr & 0xffffff000ull) << intlv_shift) +
	    (input_addr & 0xfff);

	intlv_sel = pvt->dram_IntlvSel[node_id] & ((1 << intlv_shift) - 1);
	dram_addr = bits + (intlv_sel << 12);

	debugf1("InputAddr 0x%lx translates to DramAddr 0x%lx "
		"(%d node interleave bits)\n", (unsigned long)input_addr,
		(unsigned long)dram_addr, intlv_shift);

	return dram_addr;
}

/*
 * @dram_addr is a DramAddr that maps to the node represented by mci. Convert
 * @dram_addr to a SysAddr.
 */
static u64 dram_addr_to_sys_addr(struct mem_ctl_info *mci, u64 dram_addr)
{
	struct amd64_pvt *pvt = mci->pvt_info;
	u64 hole_base, hole_offset, hole_size, base, limit, sys_addr;
	int ret = 0;

	ret = amd64_get_dram_hole_info(mci, &hole_base, &hole_offset,
				      &hole_size);
	if (!ret) {
		if ((dram_addr >= hole_base) &&
		    (dram_addr < (hole_base + hole_size))) {
			sys_addr = dram_addr + hole_offset;

			debugf1("using DHAR to translate DramAddr 0x%lx to "
				"SysAddr 0x%lx\n", (unsigned long)dram_addr,
				(unsigned long)sys_addr);

			return sys_addr;
		}
	}

	amd64_get_base_and_limit(pvt, pvt->mc_node_id, &base, &limit);
	sys_addr = dram_addr + base;

	/*
	 * The sys_addr we have computed up to this point is a 40-bit value
	 * because the k8 deals with 40-bit values.  However, the value we are
	 * supposed to return is a full 64-bit physical address.  The AMD
	 * x86-64 architecture specifies that the most significant implemented
	 * address bit through bit 63 of a physical address must be either all
	 * 0s or all 1s.  Therefore we sign-extend the 40-bit sys_addr to a
	 * 64-bit value below.  See section 3.4.2 of AMD publication 24592:
	 * AMD x86-64 Architecture Programmer's Manual Volume 1 Application
	 * Programming.
	 */
	sys_addr |= ~((sys_addr & (1ull << 39)) - 1);

	debugf1("    Node %d, DramAddr 0x%lx to SysAddr 0x%lx\n",
		pvt->mc_node_id, (unsigned long)dram_addr,
		(unsigned long)sys_addr);

	return sys_addr;
}

/*
 * @input_addr is an InputAddr associated with the node given by mci. Translate
 * @input_addr to a SysAddr.
 */
static inline u64 input_addr_to_sys_addr(struct mem_ctl_info *mci,
					 u64 input_addr)
{
	return dram_addr_to_sys_addr(mci,
				     input_addr_to_dram_addr(mci, input_addr));
}

/*
 * Find the minimum and maximum InputAddr values that map to the given @csrow.
 * Pass back these values in *input_addr_min and *input_addr_max.
 */
static void find_csrow_limits(struct mem_ctl_info *mci, int csrow,
			      u64 *input_addr_min, u64 *input_addr_max)
{
	struct amd64_pvt *pvt;
	u64 base, mask;

	pvt = mci->pvt_info;
	BUG_ON((csrow < 0) || (csrow >= pvt->cs_count));

	base = base_from_dct_base(pvt, csrow);
	mask = mask_from_dct_mask(pvt, csrow);

	*input_addr_min = base & ~mask;
	*input_addr_max = base | mask | pvt->dcs_mask_notused;
}

/*
 * Extract error address from MCA NB Address Low (section 3.6.4.5) and MCA NB
 * Address High (section 3.6.4.6) register values and return the result. Address
 * is located in the info structure (nbeah and nbeal), the encoding is device
 * specific.
 */
static u64 extract_error_address(struct mem_ctl_info *mci,
				 struct err_regs *info)
{
	struct amd64_pvt *pvt = mci->pvt_info;

	return pvt->ops->get_error_address(mci, info);
}


/* Map the Error address to a PAGE and PAGE OFFSET. */
static inline void error_address_to_page_and_offset(u64 error_address,
						    u32 *page, u32 *offset)
{
	*page = (u32) (error_address >> PAGE_SHIFT);
	*offset = ((u32) error_address) & ~PAGE_MASK;
}

/*
 * @sys_addr is an error address (a SysAddr) extracted from the MCA NB Address
 * Low (section 3.6.4.5) and MCA NB Address High (section 3.6.4.6) registers
 * of a node that detected an ECC memory error.  mci represents the node that
 * the error address maps to (possibly different from the node that detected
 * the error).  Return the number of the csrow that sys_addr maps to, or -1 on
 * error.
 */
static int sys_addr_to_csrow(struct mem_ctl_info *mci, u64 sys_addr)
{
	int csrow;

	csrow = input_addr_to_csrow(mci, sys_addr_to_input_addr(mci, sys_addr));

	if (csrow == -1)
		amd64_mc_printk(mci, KERN_ERR,
			     "Failed to translate InputAddr to csrow for "
			     "address 0x%lx\n", (unsigned long)sys_addr);
	return csrow;
}

static int get_channel_from_ecc_syndrome(unsigned short syndrome);

static void amd64_cpu_display_info(struct amd64_pvt *pvt)
{
	if (boot_cpu_data.x86 == 0x11)
		edac_printk(KERN_DEBUG, EDAC_MC, "F11h CPU detected\n");
	else if (boot_cpu_data.x86 == 0x10)
		edac_printk(KERN_DEBUG, EDAC_MC, "F10h CPU detected\n");
	else if (boot_cpu_data.x86 == 0xf)
		edac_printk(KERN_DEBUG, EDAC_MC, "%s detected\n",
			(pvt->ext_model >= OPTERON_CPU_REV_F) ?
			"Rev F or later" : "Rev E or earlier");
	else
		/* we'll hardly ever ever get here */
		edac_printk(KERN_ERR, EDAC_MC, "Unknown cpu!\n");
}

/*
 * Determine if the DIMMs have ECC enabled. ECC is enabled ONLY if all the DIMMs
 * are ECC capable.
 */
static enum edac_type amd64_determine_edac_cap(struct amd64_pvt *pvt)
{
	int bit;
	enum dev_type edac_cap = EDAC_FLAG_NONE;

	bit = (boot_cpu_data.x86 > 0xf || pvt->ext_model >= OPTERON_CPU_REV_F)
		? 19
		: 17;

	if (pvt->dclr0 & BIT(bit))
		edac_cap = EDAC_FLAG_SECDED;

	return edac_cap;
}


static void amd64_debug_display_dimm_sizes(int ctrl, struct amd64_pvt *pvt);

static void amd64_dump_dramcfg_low(u32 dclr, int chan)
{
	debugf1("F2x%d90 (DRAM Cfg Low): 0x%08x\n", chan, dclr);

	debugf1("  DIMM type: %sbuffered; all DIMMs support ECC: %s\n",
		(dclr & BIT(16)) ?  "un" : "",
		(dclr & BIT(19)) ? "yes" : "no");

	debugf1("  PAR/ERR parity: %s\n",
		(dclr & BIT(8)) ?  "enabled" : "disabled");

	debugf1("  DCT 128bit mode width: %s\n",
		(dclr & BIT(11)) ?  "128b" : "64b");

	debugf1("  x4 logical DIMMs present: L0: %s L1: %s L2: %s L3: %s\n",
		(dclr & BIT(12)) ?  "yes" : "no",
		(dclr & BIT(13)) ?  "yes" : "no",
		(dclr & BIT(14)) ?  "yes" : "no",
		(dclr & BIT(15)) ?  "yes" : "no");
}

/* Display and decode various NB registers for debug purposes. */
static void amd64_dump_misc_regs(struct amd64_pvt *pvt)
{
	int ganged;

	debugf1("F3xE8 (NB Cap): 0x%08x\n", pvt->nbcap);

	debugf1("  NB two channel DRAM capable: %s\n",
		(pvt->nbcap & K8_NBCAP_DCT_DUAL) ? "yes" : "no");

	debugf1("  ECC capable: %s, ChipKill ECC capable: %s\n",
		(pvt->nbcap & K8_NBCAP_SECDED) ? "yes" : "no",
		(pvt->nbcap & K8_NBCAP_CHIPKILL) ? "yes" : "no");

	amd64_dump_dramcfg_low(pvt->dclr0, 0);

	debugf1("F3xB0 (Online Spare): 0x%08x\n", pvt->online_spare);

	debugf1("F1xF0 (DRAM Hole Address): 0x%08x, base: 0x%08x, "
			"offset: 0x%08x\n",
			pvt->dhar,
			dhar_base(pvt->dhar),
			(boot_cpu_data.x86 == 0xf) ? k8_dhar_offset(pvt->dhar)
						   : f10_dhar_offset(pvt->dhar));

	debugf1("  DramHoleValid: %s\n",
		(pvt->dhar & DHAR_VALID) ? "yes" : "no");

	/* everything below this point is Fam10h and above */
	if (boot_cpu_data.x86 == 0xf) {
		amd64_debug_display_dimm_sizes(0, pvt);
		return;
	}

	/* Only if NOT ganged does dclr1 have valid info */
	if (!dct_ganging_enabled(pvt))
		amd64_dump_dramcfg_low(pvt->dclr1, 1);

	/*
	 * Determine if ganged and then dump memory sizes for first controller,
	 * and if NOT ganged dump info for 2nd controller.
	 */
	ganged = dct_ganging_enabled(pvt);

	amd64_debug_display_dimm_sizes(0, pvt);

	if (!ganged)
		amd64_debug_display_dimm_sizes(1, pvt);
}

/* Read in both of DBAM registers */
static void amd64_read_dbam_reg(struct amd64_pvt *pvt)
{
	amd64_read_pci_cfg(pvt->dram_f2_ctl, DBAM0, &pvt->dbam0);

	if (boot_cpu_data.x86 >= 0x10)
		amd64_read_pci_cfg(pvt->dram_f2_ctl, DBAM1, &pvt->dbam1);
}

/*
 * NOTE: CPU Revision Dependent code: Rev E and Rev F
 *
 * Set the DCSB and DCSM mask values depending on the CPU revision value. Also
 * set the shift factor for the DCSB and DCSM values.
 *
 * ->dcs_mask_notused, RevE:
 *
 * To find the max InputAddr for the csrow, start with the base address and set
 * all bits that are "don't care" bits in the test at the start of section
 * 3.5.4 (p. 84).
 *
 * The "don't care" bits are all set bits in the mask and all bits in the gaps
 * between bit ranges [35:25] and [19:13]. The value REV_E_DCS_NOTUSED_BITS
 * represents bits [24:20] and [12:0], which are all bits in the above-mentioned
 * gaps.
 *
 * ->dcs_mask_notused, RevF and later:
 *
 * To find the max InputAddr for the csrow, start with the base address and set
 * all bits that are "don't care" bits in the test at the start of NPT section
 * 4.5.4 (p. 87).
 *
 * The "don't care" bits are all set bits in the mask and all bits in the gaps
 * between bit ranges [36:27] and [21:13].
 *
 * The value REV_F_F1Xh_DCS_NOTUSED_BITS represents bits [26:22] and [12:0],
 * which are all bits in the above-mentioned gaps.
 */
static void amd64_set_dct_base_and_mask(struct amd64_pvt *pvt)
{

	if (boot_cpu_data.x86 == 0xf && pvt->ext_model < OPTERON_CPU_REV_F) {
		pvt->dcsb_base		= REV_E_DCSB_BASE_BITS;
		pvt->dcsm_mask		= REV_E_DCSM_MASK_BITS;
		pvt->dcs_mask_notused	= REV_E_DCS_NOTUSED_BITS;
		pvt->dcs_shift		= REV_E_DCS_SHIFT;
		pvt->cs_count		= 8;
		pvt->num_dcsm		= 8;
	} else {
		pvt->dcsb_base		= REV_F_F1Xh_DCSB_BASE_BITS;
		pvt->dcsm_mask		= REV_F_F1Xh_DCSM_MASK_BITS;
		pvt->dcs_mask_notused	= REV_F_F1Xh_DCS_NOTUSED_BITS;
		pvt->dcs_shift		= REV_F_F1Xh_DCS_SHIFT;

		if (boot_cpu_data.x86 == 0x11) {
			pvt->cs_count = 4;
			pvt->num_dcsm = 2;
		} else {
			pvt->cs_count = 8;
			pvt->num_dcsm = 4;
		}
	}
}

/*
 * Function 2 Offset F10_DCSB0; read in the DCS Base and DCS Mask hw registers
 */
static void amd64_read_dct_base_mask(struct amd64_pvt *pvt)
{
	int cs, reg;

	amd64_set_dct_base_and_mask(pvt);

	for (cs = 0; cs < pvt->cs_count; cs++) {
		reg = K8_DCSB0 + (cs * 4);
		if (!amd64_read_pci_cfg(pvt->dram_f2_ctl, reg, &pvt->dcsb0[cs]))
			debugf0("  DCSB0[%d]=0x%08x reg: F2x%x\n",
				cs, pvt->dcsb0[cs], reg);

		/* If DCT are NOT ganged, then read in DCT1's base */
		if (boot_cpu_data.x86 >= 0x10 && !dct_ganging_enabled(pvt)) {
			reg = F10_DCSB1 + (cs * 4);
			if (!amd64_read_pci_cfg(pvt->dram_f2_ctl, reg,
						&pvt->dcsb1[cs]))
				debugf0("  DCSB1[%d]=0x%08x reg: F2x%x\n",
					cs, pvt->dcsb1[cs], reg);
		} else {
			pvt->dcsb1[cs] = 0;
		}
	}

	for (cs = 0; cs < pvt->num_dcsm; cs++) {
		reg = K8_DCSM0 + (cs * 4);
		if (!amd64_read_pci_cfg(pvt->dram_f2_ctl, reg, &pvt->dcsm0[cs]))
			debugf0("    DCSM0[%d]=0x%08x reg: F2x%x\n",
				cs, pvt->dcsm0[cs], reg);

		/* If DCT are NOT ganged, then read in DCT1's mask */
		if (boot_cpu_data.x86 >= 0x10 && !dct_ganging_enabled(pvt)) {
			reg = F10_DCSM1 + (cs * 4);
			if (!amd64_read_pci_cfg(pvt->dram_f2_ctl, reg,
						&pvt->dcsm1[cs]))
				debugf0("    DCSM1[%d]=0x%08x reg: F2x%x\n",
					cs, pvt->dcsm1[cs], reg);
		} else {
			pvt->dcsm1[cs] = 0;
		}
	}
}

static enum mem_type amd64_determine_memory_type(struct amd64_pvt *pvt)
{
	enum mem_type type;

	if (boot_cpu_data.x86 >= 0x10 || pvt->ext_model >= OPTERON_CPU_REV_F) {
		/* Rev F and later */
		type = (pvt->dclr0 & BIT(16)) ? MEM_DDR2 : MEM_RDDR2;
	} else {
		/* Rev E and earlier */
		type = (pvt->dclr0 & BIT(18)) ? MEM_DDR : MEM_RDDR;
	}

	debugf1("  Memory type is: %s\n",
		(type == MEM_DDR2) ? "MEM_DDR2" :
		(type == MEM_RDDR2) ? "MEM_RDDR2" :
		(type == MEM_DDR) ? "MEM_DDR" : "MEM_RDDR");

	return type;
}

/*
 * Read the DRAM Configuration Low register. It differs between CG, D & E revs
 * and the later RevF memory controllers (DDR vs DDR2)
 *
 * Return:
 *      number of memory channels in operation
 * Pass back:
 *      contents of the DCL0_LOW register
 */
static int k8_early_channel_count(struct amd64_pvt *pvt)
{
	int flag, err = 0;

	err = amd64_read_pci_cfg(pvt->dram_f2_ctl, F10_DCLR_0, &pvt->dclr0);
	if (err)
		return err;

	if ((boot_cpu_data.x86_model >> 4) >= OPTERON_CPU_REV_F) {
		/* RevF (NPT) and later */
		flag = pvt->dclr0 & F10_WIDTH_128;
	} else {
		/* RevE and earlier */
		flag = pvt->dclr0 & REVE_WIDTH_128;
	}

	/* not used */
	pvt->dclr1 = 0;

	return (flag) ? 2 : 1;
}

/* extract the ERROR ADDRESS for the K8 CPUs */
static u64 k8_get_error_address(struct mem_ctl_info *mci,
				struct err_regs *info)
{
	return (((u64) (info->nbeah & 0xff)) << 32) +
			(info->nbeal & ~0x03);
}

/*
 * Read the Base and Limit registers for K8 based Memory controllers; extract
 * fields from the 'raw' reg into separate data fields
 *
 * Isolates: BASE, LIMIT, IntlvEn, IntlvSel, RW_EN
 */
static void k8_read_dram_base_limit(struct amd64_pvt *pvt, int dram)
{
	u32 low;
	u32 off = dram << 3;	/* 8 bytes between DRAM entries */

	amd64_read_pci_cfg(pvt->addr_f1_ctl, K8_DRAM_BASE_LOW + off, &low);

	/* Extract parts into separate data entries */
	pvt->dram_base[dram] = ((u64) low & 0xFFFF0000) << 8;
	pvt->dram_IntlvEn[dram] = (low >> 8) & 0x7;
	pvt->dram_rw_en[dram] = (low & 0x3);

	amd64_read_pci_cfg(pvt->addr_f1_ctl, K8_DRAM_LIMIT_LOW + off, &low);

	/*
	 * Extract parts into separate data entries. Limit is the HIGHEST memory
	 * location of the region, so lower 24 bits need to be all ones
	 */
	pvt->dram_limit[dram] = (((u64) low & 0xFFFF0000) << 8) | 0x00FFFFFF;
	pvt->dram_IntlvSel[dram] = (low >> 8) & 0x7;
	pvt->dram_DstNode[dram] = (low & 0x7);
}

static void k8_map_sysaddr_to_csrow(struct mem_ctl_info *mci,
					struct err_regs *info,
					u64 SystemAddress)
{
	struct mem_ctl_info *src_mci;
	unsigned short syndrome;
	int channel, csrow;
	u32 page, offset;

	/* Extract the syndrome parts and form a 16-bit syndrome */
	syndrome  = HIGH_SYNDROME(info->nbsl) << 8;
	syndrome |= LOW_SYNDROME(info->nbsh);

	/* CHIPKILL enabled */
	if (info->nbcfg & K8_NBCFG_CHIPKILL) {
		channel = get_channel_from_ecc_syndrome(syndrome);
		if (channel < 0) {
			/*
			 * Syndrome didn't map, so we don't know which of the
			 * 2 DIMMs is in error. So we need to ID 'both' of them
			 * as suspect.
			 */
			amd64_mc_printk(mci, KERN_WARNING,
				       "unknown syndrome 0x%x - possible error "
				       "reporting race\n", syndrome);
			edac_mc_handle_ce_no_info(mci, EDAC_MOD_STR);
			return;
		}
	} else {
		/*
		 * non-chipkill ecc mode
		 *
		 * The k8 documentation is unclear about how to determine the
		 * channel number when using non-chipkill memory.  This method
		 * was obtained from email communication with someone at AMD.
		 * (Wish the email was placed in this comment - norsk)
		 */
		channel = ((SystemAddress & BIT(3)) != 0);
	}

	/*
	 * Find out which node the error address belongs to. This may be
	 * different from the node that detected the error.
	 */
	src_mci = find_mc_by_sys_addr(mci, SystemAddress);
	if (!src_mci) {
		amd64_mc_printk(mci, KERN_ERR,
			     "failed to map error address 0x%lx to a node\n",
			     (unsigned long)SystemAddress);
		edac_mc_handle_ce_no_info(mci, EDAC_MOD_STR);
		return;
	}

	/* Now map the SystemAddress to a CSROW */
	csrow = sys_addr_to_csrow(src_mci, SystemAddress);
	if (csrow < 0) {
		edac_mc_handle_ce_no_info(src_mci, EDAC_MOD_STR);
	} else {
		error_address_to_page_and_offset(SystemAddress, &page, &offset);

		edac_mc_handle_ce(src_mci, page, offset, syndrome, csrow,
				  channel, EDAC_MOD_STR);
	}
}

/*
 * determrine the number of PAGES in for this DIMM's size based on its DRAM
 * Address Mapping.
 *
 * First step is to calc the number of bits to shift a value of 1 left to
 * indicate show many pages. Start with the DBAM value as the starting bits,
 * then proceed to adjust those shift bits, based on CPU rev and the table.
 * See BKDG on the DBAM
 */
static int k8_dbam_map_to_pages(struct amd64_pvt *pvt, int dram_map)
{
	int nr_pages;

	if (pvt->ext_model >= OPTERON_CPU_REV_F) {
		nr_pages = 1 << (revf_quad_ddr2_shift[dram_map] - PAGE_SHIFT);
	} else {
		/*
		 * RevE and less section; this line is tricky. It collapses the
		 * table used by RevD and later to one that matches revisions CG
		 * and earlier.
		 */
		dram_map -= (pvt->ext_model >= OPTERON_CPU_REV_D) ?
				(dram_map > 8 ? 4 : (dram_map > 5 ?
				3 : (dram_map > 2 ? 1 : 0))) : 0;

		/* 25 shift is 32MiB minimum DIMM size in RevE and prior */
		nr_pages = 1 << (dram_map + 25 - PAGE_SHIFT);
	}

	return nr_pages;
}

/*
 * Get the number of DCT channels in use.
 *
 * Return:
 *	number of Memory Channels in operation
 * Pass back:
 *	contents of the DCL0_LOW register
 */
static int f10_early_channel_count(struct amd64_pvt *pvt)
{
	int dbams[] = { DBAM0, DBAM1 };
	int i, j, channels = 0;
	u32 dbam;

	/* If we are in 128 bit mode, then we are using 2 channels */
	if (pvt->dclr0 & F10_WIDTH_128) {
		channels = 2;
		return channels;
	}

	/*
	 * Need to check if in unganged mode: In such, there are 2 channels,
	 * but they are not in 128 bit mode and thus the above 'dclr0' status
	 * bit will be OFF.
	 *
	 * Need to check DCT0[0] and DCT1[0] to see if only one of them has
	 * their CSEnable bit on. If so, then SINGLE DIMM case.
	 */
	debugf0("Data width is not 128 bits - need more decoding\n");

	/*
	 * Check DRAM Bank Address Mapping values for each DIMM to see if there
	 * is more than just one DIMM present in unganged mode. Need to check
	 * both controllers since DIMMs can be placed in either one.
	 */
	for (i = 0; i < ARRAY_SIZE(dbams); i++) {
		if (amd64_read_pci_cfg(pvt->dram_f2_ctl, dbams[i], &dbam))
			goto err_reg;

		for (j = 0; j < 4; j++) {
			if (DBAM_DIMM(j, dbam) > 0) {
				channels++;
				break;
			}
		}
	}

	if (channels > 2)
		channels = 2;

	debugf0("MCT channel count: %d\n", channels);

	return channels;

err_reg:
	return -1;

}

static int f10_dbam_map_to_pages(struct amd64_pvt *pvt, int dram_map)
{
	return 1 << (revf_quad_ddr2_shift[dram_map] - PAGE_SHIFT);
}

/* Enable extended configuration access via 0xCF8 feature */
static void amd64_setup(struct amd64_pvt *pvt)
{
	u32 reg;

	amd64_read_pci_cfg(pvt->misc_f3_ctl, F10_NB_CFG_HIGH, &reg);

	pvt->flags.cf8_extcfg = !!(reg & F10_NB_CFG_LOW_ENABLE_EXT_CFG);
	reg |= F10_NB_CFG_LOW_ENABLE_EXT_CFG;
	pci_write_config_dword(pvt->misc_f3_ctl, F10_NB_CFG_HIGH, reg);
}

/* Restore the extended configuration access via 0xCF8 feature */
static void amd64_teardown(struct amd64_pvt *pvt)
{
	u32 reg;

	amd64_read_pci_cfg(pvt->misc_f3_ctl, F10_NB_CFG_HIGH, &reg);

	reg &= ~F10_NB_CFG_LOW_ENABLE_EXT_CFG;
	if (pvt->flags.cf8_extcfg)
		reg |= F10_NB_CFG_LOW_ENABLE_EXT_CFG;
	pci_write_config_dword(pvt->misc_f3_ctl, F10_NB_CFG_HIGH, reg);
}

static u64 f10_get_error_address(struct mem_ctl_info *mci,
			struct err_regs *info)
{
	return (((u64) (info->nbeah & 0xffff)) << 32) +
			(info->nbeal & ~0x01);
}

/*
 * Read the Base and Limit registers for F10 based Memory controllers. Extract
 * fields from the 'raw' reg into separate data fields.
 *
 * Isolates: BASE, LIMIT, IntlvEn, IntlvSel, RW_EN.
 */
static void f10_read_dram_base_limit(struct amd64_pvt *pvt, int dram)
{
	u32 high_offset, low_offset, high_base, low_base, high_limit, low_limit;

	low_offset = K8_DRAM_BASE_LOW + (dram << 3);
	high_offset = F10_DRAM_BASE_HIGH + (dram << 3);

	/* read the 'raw' DRAM BASE Address register */
	amd64_read_pci_cfg(pvt->addr_f1_ctl, low_offset, &low_base);

	/* Read from the ECS data register */
	amd64_read_pci_cfg(pvt->addr_f1_ctl, high_offset, &high_base);

	/* Extract parts into separate data entries */
	pvt->dram_rw_en[dram] = (low_base & 0x3);

	if (pvt->dram_rw_en[dram] == 0)
		return;

	pvt->dram_IntlvEn[dram] = (low_base >> 8) & 0x7;

	pvt->dram_base[dram] = (((u64)high_base & 0x000000FF) << 40) |
			       (((u64)low_base  & 0xFFFF0000) << 8);

	low_offset = K8_DRAM_LIMIT_LOW + (dram << 3);
	high_offset = F10_DRAM_LIMIT_HIGH + (dram << 3);

	/* read the 'raw' LIMIT registers */
	amd64_read_pci_cfg(pvt->addr_f1_ctl, low_offset, &low_limit);

	/* Read from the ECS data register for the HIGH portion */
	amd64_read_pci_cfg(pvt->addr_f1_ctl, high_offset, &high_limit);

	debugf0("  HW Regs: BASE=0x%08x-%08x      LIMIT=  0x%08x-%08x\n",
		high_base, low_base, high_limit, low_limit);

	pvt->dram_DstNode[dram] = (low_limit & 0x7);
	pvt->dram_IntlvSel[dram] = (low_limit >> 8) & 0x7;

	/*
	 * Extract address values and form a LIMIT address. Limit is the HIGHEST
	 * memory location of the region, so low 24 bits need to be all ones.
	 */
	pvt->dram_limit[dram] = (((u64)high_limit & 0x000000FF) << 40) |
				(((u64) low_limit & 0xFFFF0000) << 8) |
				0x00FFFFFF;
}

static void f10_read_dram_ctl_register(struct amd64_pvt *pvt)
{

	if (!amd64_read_pci_cfg(pvt->dram_f2_ctl, F10_DCTL_SEL_LOW,
				&pvt->dram_ctl_select_low)) {
		debugf0("F2x110 (DCTL Sel. Low): 0x%08x, "
			"High range addresses at: 0x%x\n",
			pvt->dram_ctl_select_low,
			dct_sel_baseaddr(pvt));

		debugf0("  DCT mode: %s, All DCTs on: %s\n",
			(dct_ganging_enabled(pvt) ? "ganged" : "unganged"),
			(dct_dram_enabled(pvt) ? "yes"   : "no"));

		if (!dct_ganging_enabled(pvt))
			debugf0("  Address range split per DCT: %s\n",
				(dct_high_range_enabled(pvt) ? "yes" : "no"));

		debugf0("  DCT data interleave for ECC: %s, "
			"DRAM cleared since last warm reset: %s\n",
			(dct_data_intlv_enabled(pvt) ? "enabled" : "disabled"),
			(dct_memory_cleared(pvt) ? "yes" : "no"));

		debugf0("  DCT channel interleave: %s, "
			"DCT interleave bits selector: 0x%x\n",
			(dct_interleave_enabled(pvt) ? "enabled" : "disabled"),
			dct_sel_interleave_addr(pvt));
	}

	amd64_read_pci_cfg(pvt->dram_f2_ctl, F10_DCTL_SEL_HIGH,
			   &pvt->dram_ctl_select_high);
}

/*
 * determine channel based on the interleaving mode: F10h BKDG, 2.8.9 Memory
 * Interleaving Modes.
 */
static u32 f10_determine_channel(struct amd64_pvt *pvt, u64 sys_addr,
				int hi_range_sel, u32 intlv_en)
{
	u32 cs, temp, dct_sel_high = (pvt->dram_ctl_select_low >> 1) & 1;

	if (dct_ganging_enabled(pvt))
		cs = 0;
	else if (hi_range_sel)
		cs = dct_sel_high;
	else if (dct_interleave_enabled(pvt)) {
		/*
		 * see F2x110[DctSelIntLvAddr] - channel interleave mode
		 */
		if (dct_sel_interleave_addr(pvt) == 0)
			cs = sys_addr >> 6 & 1;
		else if ((dct_sel_interleave_addr(pvt) >> 1) & 1) {
			temp = hweight_long((u32) ((sys_addr >> 16) & 0x1F)) % 2;

			if (dct_sel_interleave_addr(pvt) & 1)
				cs = (sys_addr >> 9 & 1) ^ temp;
			else
				cs = (sys_addr >> 6 & 1) ^ temp;
		} else if (intlv_en & 4)
			cs = sys_addr >> 15 & 1;
		else if (intlv_en & 2)
			cs = sys_addr >> 14 & 1;
		else if (intlv_en & 1)
			cs = sys_addr >> 13 & 1;
		else
			cs = sys_addr >> 12 & 1;
	} else if (dct_high_range_enabled(pvt) && !dct_ganging_enabled(pvt))
		cs = ~dct_sel_high & 1;
	else
		cs = 0;

	return cs;
}

static inline u32 f10_map_intlv_en_to_shift(u32 intlv_en)
{
	if (intlv_en == 1)
		return 1;
	else if (intlv_en == 3)
		return 2;
	else if (intlv_en == 7)
		return 3;

	return 0;
}

/* See F10h BKDG, 2.8.10.2 DctSelBaseOffset Programming */
static inline u64 f10_get_base_addr_offset(u64 sys_addr, int hi_range_sel,
						 u32 dct_sel_base_addr,
						 u64 dct_sel_base_off,
						 u32 hole_valid, u32 hole_off,
						 u64 dram_base)
{
	u64 chan_off;

	if (hi_range_sel) {
		if (!(dct_sel_base_addr & 0xFFFFF800) &&
		   hole_valid && (sys_addr >= 0x100000000ULL))
			chan_off = hole_off << 16;
		else
			chan_off = dct_sel_base_off;
	} else {
		if (hole_valid && (sys_addr >= 0x100000000ULL))
			chan_off = hole_off << 16;
		else
			chan_off = dram_base & 0xFFFFF8000000ULL;
	}

	return (sys_addr & 0x0000FFFFFFFFFFC0ULL) -
			(chan_off & 0x0000FFFFFF800000ULL);
}

/* Hack for the time being - Can we get this from BIOS?? */
#define	CH0SPARE_RANK	0
#define	CH1SPARE_RANK	1

/*
 * checks if the csrow passed in is marked as SPARED, if so returns the new
 * spare row
 */
static inline int f10_process_possible_spare(int csrow,
				u32 cs, struct amd64_pvt *pvt)
{
	u32 swap_done;
	u32 bad_dram_cs;

	/* Depending on channel, isolate respective SPARING info */
	if (cs) {
		swap_done = F10_ONLINE_SPARE_SWAPDONE1(pvt->online_spare);
		bad_dram_cs = F10_ONLINE_SPARE_BADDRAM_CS1(pvt->online_spare);
		if (swap_done && (csrow == bad_dram_cs))
			csrow = CH1SPARE_RANK;
	} else {
		swap_done = F10_ONLINE_SPARE_SWAPDONE0(pvt->online_spare);
		bad_dram_cs = F10_ONLINE_SPARE_BADDRAM_CS0(pvt->online_spare);
		if (swap_done && (csrow == bad_dram_cs))
			csrow = CH0SPARE_RANK;
	}
	return csrow;
}

/*
 * Iterate over the DRAM DCT "base" and "mask" registers looking for a
 * SystemAddr match on the specified 'ChannelSelect' and 'NodeID'
 *
 * Return:
 *	-EINVAL:  NOT FOUND
 *	0..csrow = Chip-Select Row
 */
static int f10_lookup_addr_in_dct(u32 in_addr, u32 nid, u32 cs)
{
	struct mem_ctl_info *mci;
	struct amd64_pvt *pvt;
	u32 cs_base, cs_mask;
	int cs_found = -EINVAL;
	int csrow;

	mci = mci_lookup[nid];
	if (!mci)
		return cs_found;

	pvt = mci->pvt_info;

	debugf1("InputAddr=0x%x  channelselect=%d\n", in_addr, cs);

	for (csrow = 0; csrow < pvt->cs_count; csrow++) {

		cs_base = amd64_get_dct_base(pvt, cs, csrow);
		if (!(cs_base & K8_DCSB_CS_ENABLE))
			continue;

		/*
		 * We have an ENABLED CSROW, Isolate just the MASK bits of the
		 * target: [28:19] and [13:5], which map to [36:27] and [21:13]
		 * of the actual address.
		 */
		cs_base &= REV_F_F1Xh_DCSB_BASE_BITS;

		/*
		 * Get the DCT Mask, and ENABLE the reserved bits: [18:16] and
		 * [4:0] to become ON. Then mask off bits [28:0] ([36:8])
		 */
		cs_mask = amd64_get_dct_mask(pvt, cs, csrow);

		debugf1("    CSROW=%d CSBase=0x%x RAW CSMask=0x%x\n",
				csrow, cs_base, cs_mask);

		cs_mask = (cs_mask | 0x0007C01F) & 0x1FFFFFFF;

		debugf1("              Final CSMask=0x%x\n", cs_mask);
		debugf1("    (InputAddr & ~CSMask)=0x%x "
				"(CSBase & ~CSMask)=0x%x\n",
				(in_addr & ~cs_mask), (cs_base & ~cs_mask));

		if ((in_addr & ~cs_mask) == (cs_base & ~cs_mask)) {
			cs_found = f10_process_possible_spare(csrow, cs, pvt);

			debugf1(" MATCH csrow=%d\n", cs_found);
			break;
		}
	}
	return cs_found;
}

/* For a given @dram_range, check if @sys_addr falls within it. */
static int f10_match_to_this_node(struct amd64_pvt *pvt, int dram_range,
				  u64 sys_addr, int *nid, int *chan_sel)
{
	int node_id, cs_found = -EINVAL, high_range = 0;
	u32 intlv_en, intlv_sel, intlv_shift, hole_off;
	u32 hole_valid, tmp, dct_sel_base, channel;
	u64 dram_base, chan_addr, dct_sel_base_off;

	dram_base = pvt->dram_base[dram_range];
	intlv_en = pvt->dram_IntlvEn[dram_range];

	node_id = pvt->dram_DstNode[dram_range];
	intlv_sel = pvt->dram_IntlvSel[dram_range];

	debugf1("(dram=%d) Base=0x%llx SystemAddr= 0x%llx Limit=0x%llx\n",
		dram_range, dram_base, sys_addr, pvt->dram_limit[dram_range]);

	/*
	 * This assumes that one node's DHAR is the same as all the other
	 * nodes' DHAR.
	 */
	hole_off = (pvt->dhar & 0x0000FF80);
	hole_valid = (pvt->dhar & 0x1);
	dct_sel_base_off = (pvt->dram_ctl_select_high & 0xFFFFFC00) << 16;

	debugf1("   HoleOffset=0x%x  HoleValid=0x%x IntlvSel=0x%x\n",
			hole_off, hole_valid, intlv_sel);

	if (intlv_en ||
	    (intlv_sel != ((sys_addr >> 12) & intlv_en)))
		return -EINVAL;

	dct_sel_base = dct_sel_baseaddr(pvt);

	/*
	 * check whether addresses >= DctSelBaseAddr[47:27] are to be used to
	 * select between DCT0 and DCT1.
	 */
	if (dct_high_range_enabled(pvt) &&
	   !dct_ganging_enabled(pvt) &&
	   ((sys_addr >> 27) >= (dct_sel_base >> 11)))
		high_range = 1;

	channel = f10_determine_channel(pvt, sys_addr, high_range, intlv_en);

	chan_addr = f10_get_base_addr_offset(sys_addr, high_range, dct_sel_base,
					     dct_sel_base_off, hole_valid,
					     hole_off, dram_base);

	intlv_shift = f10_map_intlv_en_to_shift(intlv_en);

	/* remove Node ID (in case of memory interleaving) */
	tmp = chan_addr & 0xFC0;

	chan_addr = ((chan_addr >> intlv_shift) & 0xFFFFFFFFF000ULL) | tmp;

	/* remove channel interleave and hash */
	if (dct_interleave_enabled(pvt) &&
	   !dct_high_range_enabled(pvt) &&
	   !dct_ganging_enabled(pvt)) {
		if (dct_sel_interleave_addr(pvt) != 1)
			chan_addr = (chan_addr >> 1) & 0xFFFFFFFFFFFFFFC0ULL;
		else {
			tmp = chan_addr & 0xFC0;
			chan_addr = ((chan_addr & 0xFFFFFFFFFFFFC000ULL) >> 1)
					| tmp;
		}
	}

	debugf1("   (ChannelAddrLong=0x%llx) >> 8 becomes InputAddr=0x%x\n",
		chan_addr, (u32)(chan_addr >> 8));

	cs_found = f10_lookup_addr_in_dct(chan_addr >> 8, node_id, channel);

	if (cs_found >= 0) {
		*nid = node_id;
		*chan_sel = channel;
	}
	return cs_found;
}

static int f10_translate_sysaddr_to_cs(struct amd64_pvt *pvt, u64 sys_addr,
				       int *node, int *chan_sel)
{
	int dram_range, cs_found = -EINVAL;
	u64 dram_base, dram_limit;

	for (dram_range = 0; dram_range < DRAM_REG_COUNT; dram_range++) {

		if (!pvt->dram_rw_en[dram_range])
			continue;

		dram_base = pvt->dram_base[dram_range];
		dram_limit = pvt->dram_limit[dram_range];

		if ((dram_base <= sys_addr) && (sys_addr <= dram_limit)) {

			cs_found = f10_match_to_this_node(pvt, dram_range,
							  sys_addr, node,
							  chan_sel);
			if (cs_found >= 0)
				break;
		}
	}
	return cs_found;
}

/*
 * This the F10h reference code from AMD to map a @sys_addr to NodeID,
 * CSROW, Channel.
 *
 * The @sys_addr is usually an error address received from the hardware.
 */
static void f10_map_sysaddr_to_csrow(struct mem_ctl_info *mci,
				     struct err_regs *info,
				     u64 sys_addr)
{
	struct amd64_pvt *pvt = mci->pvt_info;
	u32 page, offset;
	unsigned short syndrome;
	int nid, csrow, chan = 0;

	csrow = f10_translate_sysaddr_to_cs(pvt, sys_addr, &nid, &chan);

	if (csrow >= 0) {
		error_address_to_page_and_offset(sys_addr, &page, &offset);

		syndrome  = HIGH_SYNDROME(info->nbsl) << 8;
		syndrome |= LOW_SYNDROME(info->nbsh);

		/*
		 * Is CHIPKILL on? If so, then we can attempt to use the
		 * syndrome to isolate which channel the error was on.
		 */
		if (pvt->nbcfg & K8_NBCFG_CHIPKILL)
			chan = get_channel_from_ecc_syndrome(syndrome);

		if (chan >= 0) {
			edac_mc_handle_ce(mci, page, offset, syndrome,
					csrow, chan, EDAC_MOD_STR);
		} else {
			/*
			 * Channel unknown, report all channels on this
			 * CSROW as failed.
			 */
			for (chan = 0; chan < mci->csrows[csrow].nr_channels;
								chan++) {
					edac_mc_handle_ce(mci, page, offset,
							syndrome,
							csrow, chan,
							EDAC_MOD_STR);
			}
		}

	} else {
		edac_mc_handle_ce_no_info(mci, EDAC_MOD_STR);
	}
}

/*
 * Input (@index) is the DBAM DIMM value (1 of 4) used as an index into a shift
 * table (revf_quad_ddr2_shift) which starts at 128MB DIMM size. Index of 0
 * indicates an empty DIMM slot, as reported by Hardware on empty slots.
 *
 * Normalize to 128MB by subracting 27 bit shift.
 */
static int map_dbam_to_csrow_size(int index)
{
	int mega_bytes = 0;

	if (index > 0 && index <= DBAM_MAX_VALUE)
		mega_bytes = ((128 << (revf_quad_ddr2_shift[index]-27)));

	return mega_bytes;
}

/*
 * debug routine to display the memory sizes of all logical DIMMs and its
 * CSROWs as well
 */
static void amd64_debug_display_dimm_sizes(int ctrl, struct amd64_pvt *pvt)
{
	int dimm, size0, size1;
	u32 dbam;
	u32 *dcsb;

	if (boot_cpu_data.x86 == 0xf) {
		/* K8 families < revF not supported yet */
	       if (pvt->ext_model < OPTERON_CPU_REV_F)
			return;
	       else
		       WARN_ON(ctrl != 0);
	}

	debugf1("F2x%d80 (DRAM Bank Address Mapping): 0x%08x\n",
		ctrl, ctrl ? pvt->dbam1 : pvt->dbam0);

	dbam = ctrl ? pvt->dbam1 : pvt->dbam0;
	dcsb = ctrl ? pvt->dcsb1 : pvt->dcsb0;

	edac_printk(KERN_DEBUG, EDAC_MC, "DCT%d chip selects:\n", ctrl);

	/* Dump memory sizes for DIMM and its CSROWs */
	for (dimm = 0; dimm < 4; dimm++) {

		size0 = 0;
		if (dcsb[dimm*2] & K8_DCSB_CS_ENABLE)
			size0 = map_dbam_to_csrow_size(DBAM_DIMM(dimm, dbam));

		size1 = 0;
		if (dcsb[dimm*2 + 1] & K8_DCSB_CS_ENABLE)
			size1 = map_dbam_to_csrow_size(DBAM_DIMM(dimm, dbam));

		edac_printk(KERN_DEBUG, EDAC_MC, " %d: %5dMB %d: %5dMB\n",
			    dimm * 2, size0, dimm * 2 + 1, size1);
	}
}

/*
 * Very early hardware probe on pci_probe thread to determine if this module
 * supports the hardware.
 *
 * Return:
 *      0 for OK
 *      1 for error
 */
static int f10_probe_valid_hardware(struct amd64_pvt *pvt)
{
	int ret = 0;

	/*
	 * If we are on a DDR3 machine, we don't know yet if
	 * we support that properly at this time
	 */
	if ((pvt->dchr0 & F10_DCHR_Ddr3Mode) ||
	    (pvt->dchr1 & F10_DCHR_Ddr3Mode)) {

		amd64_printk(KERN_WARNING,
			"%s() This machine is running with DDR3 memory. "
			"This is not currently supported. "
			"DCHR0=0x%x DCHR1=0x%x\n",
			__func__, pvt->dchr0, pvt->dchr1);

		amd64_printk(KERN_WARNING,
			"   Contact '%s' module MAINTAINER to help add"
			" support.\n",
			EDAC_MOD_STR);

		ret = 1;

	}
	return ret;
}

/*
 * There currently are 3 types type of MC devices for AMD Athlon/Opterons
 * (as per PCI DEVICE_IDs):
 *
 * Family K8: That is the Athlon64 and Opteron CPUs. They all have the same PCI
 * DEVICE ID, even though there is differences between the different Revisions
 * (CG,D,E,F).
 *
 * Family F10h and F11h.
 *
 */
static struct amd64_family_type amd64_family_types[] = {
	[K8_CPUS] = {
		.ctl_name = "RevF",
		.addr_f1_ctl = PCI_DEVICE_ID_AMD_K8_NB_ADDRMAP,
		.misc_f3_ctl = PCI_DEVICE_ID_AMD_K8_NB_MISC,
		.ops = {
			.early_channel_count = k8_early_channel_count,
			.get_error_address = k8_get_error_address,
			.read_dram_base_limit = k8_read_dram_base_limit,
			.map_sysaddr_to_csrow = k8_map_sysaddr_to_csrow,
			.dbam_map_to_pages = k8_dbam_map_to_pages,
		}
	},
	[F10_CPUS] = {
		.ctl_name = "Family 10h",
		.addr_f1_ctl = PCI_DEVICE_ID_AMD_10H_NB_MAP,
		.misc_f3_ctl = PCI_DEVICE_ID_AMD_10H_NB_MISC,
		.ops = {
			.probe_valid_hardware = f10_probe_valid_hardware,
			.early_channel_count = f10_early_channel_count,
			.get_error_address = f10_get_error_address,
			.read_dram_base_limit = f10_read_dram_base_limit,
			.read_dram_ctl_register = f10_read_dram_ctl_register,
			.map_sysaddr_to_csrow = f10_map_sysaddr_to_csrow,
			.dbam_map_to_pages = f10_dbam_map_to_pages,
		}
	},
	[F11_CPUS] = {
		.ctl_name = "Family 11h",
		.addr_f1_ctl = PCI_DEVICE_ID_AMD_11H_NB_MAP,
		.misc_f3_ctl = PCI_DEVICE_ID_AMD_11H_NB_MISC,
		.ops = {
			.probe_valid_hardware = f10_probe_valid_hardware,
			.early_channel_count = f10_early_channel_count,
			.get_error_address = f10_get_error_address,
			.read_dram_base_limit = f10_read_dram_base_limit,
			.read_dram_ctl_register = f10_read_dram_ctl_register,
			.map_sysaddr_to_csrow = f10_map_sysaddr_to_csrow,
			.dbam_map_to_pages = f10_dbam_map_to_pages,
		}
	},
};

static struct pci_dev *pci_get_related_function(unsigned int vendor,
						unsigned int device,
						struct pci_dev *related)
{
	struct pci_dev *dev = NULL;

	dev = pci_get_device(vendor, device, dev);
	while (dev) {
		if ((dev->bus->number == related->bus->number) &&
		    (PCI_SLOT(dev->devfn) == PCI_SLOT(related->devfn)))
			break;
		dev = pci_get_device(vendor, device, dev);
	}

	return dev;
}

/*
 * syndrome mapping table for ECC ChipKill devices
 *
 * The comment in each row is the token (nibble) number that is in error.
 * The least significant nibble of the syndrome is the mask for the bits
 * that are in error (need to be toggled) for the particular nibble.
 *
 * Each row contains 16 entries.
 * The first entry (0th) is the channel number for that row of syndromes.
 * The remaining 15 entries are the syndromes for the respective Error
 * bit mask index.
 *
 * 1st index entry is 0x0001 mask, indicating that the rightmost bit is the
 * bit in error.
 * The 2nd index entry is 0x0010 that the second bit is damaged.
 * The 3rd index entry is 0x0011 indicating that the rightmost 2 bits
 * are damaged.
 * Thus so on until index 15, 0x1111, whose entry has the syndrome
 * indicating that all 4 bits are damaged.
 *
 * A search is performed on this table looking for a given syndrome.
 *
 * See the AMD documentation for ECC syndromes. This ECC table is valid
 * across all the versions of the AMD64 processors.
 *
 * A fast lookup is to use the LAST four bits of the 16-bit syndrome as a
 * COLUMN index, then search all ROWS of that column, looking for a match
 * with the input syndrome. The ROW value will be the token number.
 *
 * The 0'th entry on that row, can be returned as the CHANNEL (0 or 1) of this
 * error.
 */
#define NUMBER_ECC_ROWS  36
static const unsigned short ecc_chipkill_syndromes[NUMBER_ECC_ROWS][16] = {
	/* Channel 0 syndromes */
	{/*0*/  0, 0xe821, 0x7c32, 0x9413, 0xbb44, 0x5365, 0xc776, 0x2f57,
	   0xdd88, 0x35a9, 0xa1ba, 0x499b, 0x66cc, 0x8eed, 0x1afe, 0xf2df },
	{/*1*/  0, 0x5d31, 0xa612, 0xfb23, 0x9584, 0xc8b5, 0x3396, 0x6ea7,
	   0xeac8, 0xb7f9, 0x4cda, 0x11eb, 0x7f4c, 0x227d, 0xd95e, 0x846f },
	{/*2*/  0, 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006, 0x0007,
	   0x0008, 0x0009, 0x000a, 0x000b, 0x000c, 0x000d, 0x000e, 0x000f },
	{/*3*/  0, 0x2021, 0x3032, 0x1013, 0x4044, 0x6065, 0x7076, 0x5057,
	   0x8088, 0xa0a9, 0xb0ba, 0x909b, 0xc0cc, 0xe0ed, 0xf0fe, 0xd0df },
	{/*4*/  0, 0x5041, 0xa082, 0xf0c3, 0x9054, 0xc015, 0x30d6, 0x6097,
	   0xe0a8, 0xb0e9, 0x402a, 0x106b, 0x70fc, 0x20bd, 0xd07e, 0x803f },
	{/*5*/  0, 0xbe21, 0xd732, 0x6913, 0x2144, 0x9f65, 0xf676, 0x4857,
	   0x3288, 0x8ca9, 0xe5ba, 0x5b9b, 0x13cc, 0xaded, 0xc4fe, 0x7adf },
	{/*6*/  0, 0x4951, 0x8ea2, 0xc7f3, 0x5394, 0x1ac5, 0xdd36, 0x9467,
	   0xa1e8, 0xe8b9, 0x2f4a, 0x661b, 0xf27c, 0xbb2d, 0x7cde, 0x358f },
	{/*7*/  0, 0x74e1, 0x9872, 0xec93, 0xd6b4, 0xa255, 0x4ec6, 0x3a27,
	   0x6bd8, 0x1f39, 0xf3aa, 0x874b, 0xbd6c, 0xc98d, 0x251e, 0x51ff },
	{/*8*/  0, 0x15c1, 0x2a42, 0x3f83, 0xcef4, 0xdb35, 0xe4b6, 0xf177,
	   0x4758, 0x5299, 0x6d1a, 0x78db, 0x89ac, 0x9c6d, 0xa3ee, 0xb62f },
	{/*9*/  0, 0x3d01, 0x1602, 0x2b03, 0x8504, 0xb805, 0x9306, 0xae07,
	   0xca08, 0xf709, 0xdc0a, 0xe10b, 0x4f0c, 0x720d, 0x590e, 0x640f },
	{/*a*/  0, 0x9801, 0xec02, 0x7403, 0x6b04, 0xf305, 0x8706, 0x1f07,
	   0xbd08, 0x2509, 0x510a, 0xc90b, 0xd60c, 0x4e0d, 0x3a0e, 0xa20f },
	{/*b*/  0, 0xd131, 0x6212, 0xb323, 0x3884, 0xe9b5, 0x5a96, 0x8ba7,
	   0x1cc8, 0xcdf9, 0x7eda, 0xafeb, 0x244c, 0xf57d, 0x465e, 0x976f },
	{/*c*/  0, 0xe1d1, 0x7262, 0x93b3, 0xb834, 0x59e5, 0xca56, 0x2b87,
	   0xdc18, 0x3dc9, 0xae7a, 0x4fab, 0x542c, 0x85fd, 0x164e, 0xf79f },
	{/*d*/  0, 0x6051, 0xb0a2, 0xd0f3, 0x1094, 0x70c5, 0xa036, 0xc067,
	   0x20e8, 0x40b9, 0x904a, 0x601b, 0x307c, 0x502d, 0x80de, 0xe08f },
	{/*e*/  0, 0xa4c1, 0xf842, 0x5c83, 0xe6f4, 0x4235, 0x1eb6, 0xba77,
	   0x7b58, 0xdf99, 0x831a, 0x27db, 0x9dac, 0x396d, 0x65ee, 0xc12f },
	{/*f*/  0, 0x11c1, 0x2242, 0x3383, 0xc8f4, 0xd935, 0xeab6, 0xfb77,
	   0x4c58, 0x5d99, 0x6e1a, 0x7fdb, 0x84ac, 0x956d, 0xa6ee, 0xb72f },

	/* Channel 1 syndromes */
	{/*10*/ 1, 0x45d1, 0x8a62, 0xcfb3, 0x5e34, 0x1be5, 0xd456, 0x9187,
	   0xa718, 0xe2c9, 0x2d7a, 0x68ab, 0xf92c, 0xbcfd, 0x734e, 0x369f },
	{/*11*/ 1, 0x63e1, 0xb172, 0xd293, 0x14b4, 0x7755, 0xa5c6, 0xc627,
	   0x28d8, 0x4b39, 0x99aa, 0xfa4b, 0x3c6c, 0x5f8d, 0x8d1e, 0xeeff },
	{/*12*/ 1, 0xb741, 0xd982, 0x6ec3, 0x2254, 0x9515, 0xfbd6, 0x4c97,
	   0x33a8, 0x84e9, 0xea2a, 0x5d6b, 0x11fc, 0xa6bd, 0xc87e, 0x7f3f },
	{/*13*/ 1, 0xdd41, 0x6682, 0xbbc3, 0x3554, 0xe815, 0x53d6, 0xce97,
	   0x1aa8, 0xc7e9, 0x7c2a, 0xa1fb, 0x2ffc, 0xf2bd, 0x497e, 0x943f },
	{/*14*/ 1, 0x2bd1, 0x3d62, 0x16b3, 0x4f34, 0x64e5, 0x7256, 0x5987,
	   0x8518, 0xaec9, 0xb87a, 0x93ab, 0xca2c, 0xe1fd, 0xf74e, 0xdc9f },
	{/*15*/ 1, 0x83c1, 0xc142, 0x4283, 0xa4f4, 0x2735, 0x65b6, 0xe677,
	   0xf858, 0x7b99, 0x391a, 0xbadb, 0x5cac, 0xdf6d, 0x9dee, 0x1e2f },
	{/*16*/ 1, 0x8fd1, 0xc562, 0x4ab3, 0xa934, 0x26e5, 0x6c56, 0xe387,
	   0xfe18, 0x71c9, 0x3b7a, 0xb4ab, 0x572c, 0xd8fd, 0x924e, 0x1d9f },
	{/*17*/ 1, 0x4791, 0x89e2, 0xce73, 0x5264, 0x15f5, 0xdb86, 0x9c17,
	   0xa3b8, 0xe429, 0x2a5a, 0x6dcb, 0xf1dc, 0xb64d, 0x783e, 0x3faf },
	{/*18*/ 1, 0x5781, 0xa9c2, 0xfe43, 0x92a4, 0xc525, 0x3b66, 0x6ce7,
	   0xe3f8, 0xb479, 0x4a3a, 0x1dbb, 0x715c, 0x26dd, 0xd89e, 0x8f1f },
	{/*19*/ 1, 0xbf41, 0xd582, 0x6ac3, 0x2954, 0x9615, 0xfcd6, 0x4397,
	   0x3ea8, 0x81e9, 0xeb2a, 0x546b, 0x17fc, 0xa8bd, 0xc27e, 0x7d3f },
	{/*1a*/ 1, 0x9891, 0xe1e2, 0x7273, 0x6464, 0xf7f5, 0x8586, 0x1617,
	   0xb8b8, 0x2b29, 0x595a, 0xcacb, 0xdcdc, 0x4f4d, 0x3d3e, 0xaeaf },
	{/*1b*/ 1, 0xcce1, 0x4472, 0x8893, 0xfdb4, 0x3f55, 0xb9c6, 0x7527,
	   0x56d8, 0x9a39, 0x12aa, 0xde4b, 0xab6c, 0x678d, 0xef1e, 0x23ff },
	{/*1c*/ 1, 0xa761, 0xf9b2, 0x5ed3, 0xe214, 0x4575, 0x1ba6, 0xbcc7,
	   0x7328, 0xd449, 0x8a9a, 0x2dfb, 0x913c, 0x365d, 0x688e, 0xcfef },
	{/*1d*/ 1, 0xff61, 0x55b2, 0xaad3, 0x7914, 0x8675, 0x2ca6, 0xd3c7,
	   0x9e28, 0x6149, 0xcb9a, 0x34fb, 0xe73c, 0x185d, 0xb28e, 0x4def },
	{/*1e*/ 1, 0x5451, 0xa8a2, 0xfcf3, 0x9694, 0xc2c5, 0x3e36, 0x6a67,
	   0xebe8, 0xbfb9, 0x434a, 0x171b, 0x7d7c, 0x292d, 0xd5de, 0x818f },
	{/*1f*/ 1, 0x6fc1, 0xb542, 0xda83, 0x19f4, 0x7635, 0xacb6, 0xc377,
	   0x2e58, 0x4199, 0x9b1a, 0xf4db, 0x37ac, 0x586d, 0x82ee, 0xed2f },

	/* ECC bits are also in the set of tokens and they too can go bad
	 * first 2 cover channel 0, while the second 2 cover channel 1
	 */
	{/*20*/ 0, 0xbe01, 0xd702, 0x6903, 0x2104, 0x9f05, 0xf606, 0x4807,
	   0x3208, 0x8c09, 0xe50a, 0x5b0b, 0x130c, 0xad0d, 0xc40e, 0x7a0f },
	{/*21*/ 0, 0x4101, 0x8202, 0xc303, 0x5804, 0x1905, 0xda06, 0x9b07,
	   0xac08, 0xed09, 0x2e0a, 0x6f0b, 0x640c, 0xb50d, 0x760e, 0x370f },
	{/*22*/ 1, 0xc441, 0x4882, 0x8cc3, 0xf654, 0x3215, 0xbed6, 0x7a97,
	   0x5ba8, 0x9fe9, 0x132a, 0xd76b, 0xadfc, 0x69bd, 0xe57e, 0x213f },
	{/*23*/ 1, 0x7621, 0x9b32, 0xed13, 0xda44, 0xac65, 0x4176, 0x3757,
	   0x6f88, 0x19a9, 0xf4ba, 0x829b, 0xb5cc, 0xc3ed, 0x2efe, 0x58df }
};

/*
 * Given the syndrome argument, scan each of the channel tables for a syndrome
 * match. Depending on which table it is found, return the channel number.
 */
static int get_channel_from_ecc_syndrome(unsigned short syndrome)
{
	int row;
	int column;

	/* Determine column to scan */
	column = syndrome & 0xF;

	/* Scan all rows, looking for syndrome, or end of table */
	for (row = 0; row < NUMBER_ECC_ROWS; row++) {
		if (ecc_chipkill_syndromes[row][column] == syndrome)
			return ecc_chipkill_syndromes[row][0];
	}

	debugf0("syndrome(%x) not found\n", syndrome);
	return -1;
}

/*
 * Check for valid error in the NB Status High register. If so, proceed to read
 * NB Status Low, NB Address Low and NB Address High registers and store data
 * into error structure.
 *
 * Returns:
 *	- 1: if hardware regs contains valid error info
 *	- 0: if no valid error is indicated
 */
static int amd64_get_error_info_regs(struct mem_ctl_info *mci,
				     struct err_regs *regs)
{
	struct amd64_pvt *pvt;
	struct pci_dev *misc_f3_ctl;

	pvt = mci->pvt_info;
	misc_f3_ctl = pvt->misc_f3_ctl;

	if (amd64_read_pci_cfg(misc_f3_ctl, K8_NBSH, &regs->nbsh))
		return 0;

	if (!(regs->nbsh & K8_NBSH_VALID_BIT))
		return 0;

	/* valid error, read remaining error information registers */
	if (amd64_read_pci_cfg(misc_f3_ctl, K8_NBSL, &regs->nbsl) ||
	    amd64_read_pci_cfg(misc_f3_ctl, K8_NBEAL, &regs->nbeal) ||
	    amd64_read_pci_cfg(misc_f3_ctl, K8_NBEAH, &regs->nbeah) ||
	    amd64_read_pci_cfg(misc_f3_ctl, K8_NBCFG, &regs->nbcfg))
		return 0;

	return 1;
}

/*
 * This function is called to retrieve the error data from hardware and store it
 * in the info structure.
 *
 * Returns:
 *	- 1: if a valid error is found
 *	- 0: if no error is found
 */
static int amd64_get_error_info(struct mem_ctl_info *mci,
				struct err_regs *info)
{
	struct amd64_pvt *pvt;
	struct err_regs regs;

	pvt = mci->pvt_info;

	if (!amd64_get_error_info_regs(mci, info))
		return 0;

	/*
	 * Here's the problem with the K8's EDAC reporting: There are four
	 * registers which report pieces of error information. They are shared
	 * between CEs and UEs. Furthermore, contrary to what is stated in the
	 * BKDG, the overflow bit is never used! Every error always updates the
	 * reporting registers.
	 *
	 * Can you see the race condition? All four error reporting registers
	 * must be read before a new error updates them! There is no way to read
	 * all four registers atomically. The best than can be done is to detect
	 * that a race has occured and then report the error without any kind of
	 * precision.
	 *
	 * What is still positive is that errors are still reported and thus
	 * problems can still be detected - just not localized because the
	 * syndrome and address are spread out across registers.
	 *
	 * Grrrrr!!!!!  Here's hoping that AMD fixes this in some future K8 rev.
	 * UEs and CEs should have separate register sets with proper overflow
	 * bits that are used! At very least the problem can be fixed by
	 * honoring the ErrValid bit in 'nbsh' and not updating registers - just
	 * set the overflow bit - unless the current error is CE and the new
	 * error is UE which would be the only situation for overwriting the
	 * current values.
	 */

	regs = *info;

	/* Use info from the second read - most current */
	if (unlikely(!amd64_get_error_info_regs(mci, info)))
		return 0;

	/* clear the error bits in hardware */
	pci_write_bits32(pvt->misc_f3_ctl, K8_NBSH, 0, K8_NBSH_VALID_BIT);

	/* Check for the possible race condition */
	if ((regs.nbsh != info->nbsh) ||
	     (regs.nbsl != info->nbsl) ||
	     (regs.nbeah != info->nbeah) ||
	     (regs.nbeal != info->nbeal)) {
		amd64_mc_printk(mci, KERN_WARNING,
				"hardware STATUS read access race condition "
				"detected!\n");
		return 0;
	}
	return 1;
}

/*
 * Handle any Correctable Errors (CEs) that have occurred. Check for valid ERROR
 * ADDRESS and process.
 */
static void amd64_handle_ce(struct mem_ctl_info *mci,
			    struct err_regs *info)
{
	struct amd64_pvt *pvt = mci->pvt_info;
	u64 SystemAddress;

	/* Ensure that the Error Address is VALID */
	if ((info->nbsh & K8_NBSH_VALID_ERROR_ADDR) == 0) {
		amd64_mc_printk(mci, KERN_ERR,
			"HW has no ERROR_ADDRESS available\n");
		edac_mc_handle_ce_no_info(mci, EDAC_MOD_STR);
		return;
	}

	SystemAddress = extract_error_address(mci, info);

	amd64_mc_printk(mci, KERN_ERR,
		"CE ERROR_ADDRESS= 0x%llx\n", SystemAddress);

	pvt->ops->map_sysaddr_to_csrow(mci, info, SystemAddress);
}

/* Handle any Un-correctable Errors (UEs) */
static void amd64_handle_ue(struct mem_ctl_info *mci,
			    struct err_regs *info)
{
	int csrow;
	u64 SystemAddress;
	u32 page, offset;
	struct mem_ctl_info *log_mci, *src_mci = NULL;

	log_mci = mci;

	if ((info->nbsh & K8_NBSH_VALID_ERROR_ADDR) == 0) {
		amd64_mc_printk(mci, KERN_CRIT,
			"HW has no ERROR_ADDRESS available\n");
		edac_mc_handle_ue_no_info(log_mci, EDAC_MOD_STR);
		return;
	}

	SystemAddress = extract_error_address(mci, info);

	/*
	 * Find out which node the error address belongs to. This may be
	 * different from the node that detected the error.
	 */
	src_mci = find_mc_by_sys_addr(mci, SystemAddress);
	if (!src_mci) {
		amd64_mc_printk(mci, KERN_CRIT,
			"ERROR ADDRESS (0x%lx) value NOT mapped to a MC\n",
			(unsigned long)SystemAddress);
		edac_mc_handle_ue_no_info(log_mci, EDAC_MOD_STR);
		return;
	}

	log_mci = src_mci;

	csrow = sys_addr_to_csrow(log_mci, SystemAddress);
	if (csrow < 0) {
		amd64_mc_printk(mci, KERN_CRIT,
			"ERROR_ADDRESS (0x%lx) value NOT mapped to 'csrow'\n",
			(unsigned long)SystemAddress);
		edac_mc_handle_ue_no_info(log_mci, EDAC_MOD_STR);
	} else {
		error_address_to_page_and_offset(SystemAddress, &page, &offset);
		edac_mc_handle_ue(log_mci, page, offset, csrow, EDAC_MOD_STR);
	}
}

static inline void __amd64_decode_bus_error(struct mem_ctl_info *mci,
					    struct err_regs *info)
{
	u32 ec  = ERROR_CODE(info->nbsl);
	u32 xec = EXT_ERROR_CODE(info->nbsl);
	int ecc_type = (info->nbsh >> 13) & 0x3;

	/* Bail early out if this was an 'observed' error */
	if (PP(ec) == K8_NBSL_PP_OBS)
		return;

	/* Do only ECC errors */
	if (xec && xec != F10_NBSL_EXT_ERR_ECC)
		return;

	if (ecc_type == 2)
		amd64_handle_ce(mci, info);
	else if (ecc_type == 1)
		amd64_handle_ue(mci, info);

	/*
	 * If main error is CE then overflow must be CE.  If main error is UE
	 * then overflow is unknown.  We'll call the overflow a CE - if
	 * panic_on_ue is set then we're already panic'ed and won't arrive
	 * here. Else, then apparently someone doesn't think that UE's are
	 * catastrophic.
	 */
	if (info->nbsh & K8_NBSH_OVERFLOW)
		edac_mc_handle_ce_no_info(mci, EDAC_MOD_STR "Error Overflow");
}

void amd64_decode_bus_error(int node_id, struct err_regs *regs)
{
	struct mem_ctl_info *mci = mci_lookup[node_id];

	__amd64_decode_bus_error(mci, regs);

	/*
	 * Check the UE bit of the NB status high register, if set generate some
	 * logs. If NOT a GART error, then process the event as a NO-INFO event.
	 * If it was a GART error, skip that process.
	 *
	 * FIXME: this should go somewhere else, if at all.
	 */
	if (regs->nbsh & K8_NBSH_UC_ERR && !report_gart_errors)
		edac_mc_handle_ue_no_info(mci, "UE bit is set");

}

/*
 * The main polling 'check' function, called FROM the edac core to perform the
 * error checking and if an error is encountered, error processing.
 */
static void amd64_check(struct mem_ctl_info *mci)
{
	struct err_regs regs;

	if (amd64_get_error_info(mci, &regs)) {
		struct amd64_pvt *pvt = mci->pvt_info;
		amd_decode_nb_mce(pvt->mc_node_id, &regs, 1);
	}
}

/*
 * Input:
 *	1) struct amd64_pvt which contains pvt->dram_f2_ctl pointer
 *	2) AMD Family index value
 *
 * Ouput:
 *	Upon return of 0, the following filled in:
 *
 *		struct pvt->addr_f1_ctl
 *		struct pvt->misc_f3_ctl
 *
 *	Filled in with related device funcitions of 'dram_f2_ctl'
 *	These devices are "reserved" via the pci_get_device()
 *
 *	Upon return of 1 (error status):
 *
 *		Nothing reserved
 */
static int amd64_reserve_mc_sibling_devices(struct amd64_pvt *pvt, int mc_idx)
{
	const struct amd64_family_type *amd64_dev = &amd64_family_types[mc_idx];

	/* Reserve the ADDRESS MAP Device */
	pvt->addr_f1_ctl = pci_get_related_function(pvt->dram_f2_ctl->vendor,
						    amd64_dev->addr_f1_ctl,
						    pvt->dram_f2_ctl);

	if (!pvt->addr_f1_ctl) {
		amd64_printk(KERN_ERR, "error address map device not found: "
			     "vendor %x device 0x%x (broken BIOS?)\n",
			     PCI_VENDOR_ID_AMD, amd64_dev->addr_f1_ctl);
		return 1;
	}

	/* Reserve the MISC Device */
	pvt->misc_f3_ctl = pci_get_related_function(pvt->dram_f2_ctl->vendor,
						    amd64_dev->misc_f3_ctl,
						    pvt->dram_f2_ctl);

	if (!pvt->misc_f3_ctl) {
		pci_dev_put(pvt->addr_f1_ctl);
		pvt->addr_f1_ctl = NULL;

		amd64_printk(KERN_ERR, "error miscellaneous device not found: "
			     "vendor %x device 0x%x (broken BIOS?)\n",
			     PCI_VENDOR_ID_AMD, amd64_dev->misc_f3_ctl);
		return 1;
	}

	debugf1("    Addr Map device PCI Bus ID:\t%s\n",
		pci_name(pvt->addr_f1_ctl));
	debugf1("    DRAM MEM-CTL PCI Bus ID:\t%s\n",
		pci_name(pvt->dram_f2_ctl));
	debugf1("    Misc device PCI Bus ID:\t%s\n",
		pci_name(pvt->misc_f3_ctl));

	return 0;
}

static void amd64_free_mc_sibling_devices(struct amd64_pvt *pvt)
{
	pci_dev_put(pvt->addr_f1_ctl);
	pci_dev_put(pvt->misc_f3_ctl);
}

/*
 * Retrieve the hardware registers of the memory controller (this includes the
 * 'Address Map' and 'Misc' device regs)
 */
static void amd64_read_mc_registers(struct amd64_pvt *pvt)
{
	u64 msr_val;
	int dram;

	/*
	 * Retrieve TOP_MEM and TOP_MEM2; no masking off of reserved bits since
	 * those are Read-As-Zero
	 */
	rdmsrl(MSR_K8_TOP_MEM1, pvt->top_mem);
	debugf0("  TOP_MEM:  0x%016llx\n", pvt->top_mem);

	/* check first whether TOP_MEM2 is enabled */
	rdmsrl(MSR_K8_SYSCFG, msr_val);
	if (msr_val & (1U << 21)) {
		rdmsrl(MSR_K8_TOP_MEM2, pvt->top_mem2);
		debugf0("  TOP_MEM2: 0x%016llx\n", pvt->top_mem2);
	} else
		debugf0("  TOP_MEM2 disabled.\n");

	amd64_cpu_display_info(pvt);

	amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_NBCAP, &pvt->nbcap);

	if (pvt->ops->read_dram_ctl_register)
		pvt->ops->read_dram_ctl_register(pvt);

	for (dram = 0; dram < DRAM_REG_COUNT; dram++) {
		/*
		 * Call CPU specific READ function to get the DRAM Base and
		 * Limit values from the DCT.
		 */
		pvt->ops->read_dram_base_limit(pvt, dram);

		/*
		 * Only print out debug info on rows with both R and W Enabled.
		 * Normal processing, compiler should optimize this whole 'if'
		 * debug output block away.
		 */
		if (pvt->dram_rw_en[dram] != 0) {
			debugf1("  DRAM-BASE[%d]: 0x%016llx "
				"DRAM-LIMIT:  0x%016llx\n",
				dram,
				pvt->dram_base[dram],
				pvt->dram_limit[dram]);

			debugf1("        IntlvEn=%s %s %s "
				"IntlvSel=%d DstNode=%d\n",
				pvt->dram_IntlvEn[dram] ?
					"Enabled" : "Disabled",
				(pvt->dram_rw_en[dram] & 0x2) ? "W" : "!W",
				(pvt->dram_rw_en[dram] & 0x1) ? "R" : "!R",
				pvt->dram_IntlvSel[dram],
				pvt->dram_DstNode[dram]);
		}
	}

	amd64_read_dct_base_mask(pvt);

	amd64_read_pci_cfg(pvt->addr_f1_ctl, K8_DHAR, &pvt->dhar);
	amd64_read_dbam_reg(pvt);

	amd64_read_pci_cfg(pvt->misc_f3_ctl,
			   F10_ONLINE_SPARE, &pvt->online_spare);

	amd64_read_pci_cfg(pvt->dram_f2_ctl, F10_DCLR_0, &pvt->dclr0);
	amd64_read_pci_cfg(pvt->dram_f2_ctl, F10_DCHR_0, &pvt->dchr0);

	if (!dct_ganging_enabled(pvt)) {
		amd64_read_pci_cfg(pvt->dram_f2_ctl, F10_DCLR_1, &pvt->dclr1);
		amd64_read_pci_cfg(pvt->dram_f2_ctl, F10_DCHR_1, &pvt->dchr1);
	}
	amd64_dump_misc_regs(pvt);
}

/*
 * NOTE: CPU Revision Dependent code
 *
 * Input:
 *	@csrow_nr ChipSelect Row Number (0..pvt->cs_count-1)
 *	k8 private pointer to -->
 *			DRAM Bank Address mapping register
 *			node_id
 *			DCL register where dual_channel_active is
 *
 * The DBAM register consists of 4 sets of 4 bits each definitions:
 *
 * Bits:	CSROWs
 * 0-3		CSROWs 0 and 1
 * 4-7		CSROWs 2 and 3
 * 8-11		CSROWs 4 and 5
 * 12-15	CSROWs 6 and 7
 *
 * Values range from: 0 to 15
 * The meaning of the values depends on CPU revision and dual-channel state,
 * see relevant BKDG more info.
 *
 * The memory controller provides for total of only 8 CSROWs in its current
 * architecture. Each "pair" of CSROWs normally represents just one DIMM in
 * single channel or two (2) DIMMs in dual channel mode.
 *
 * The following code logic collapses the various tables for CSROW based on CPU
 * revision.
 *
 * Returns:
 *	The number of PAGE_SIZE pages on the specified CSROW number it
 *	encompasses
 *
 */
static u32 amd64_csrow_nr_pages(int csrow_nr, struct amd64_pvt *pvt)
{
	u32 dram_map, nr_pages;

	/*
	 * The math on this doesn't look right on the surface because x/2*4 can
	 * be simplified to x*2 but this expression makes use of the fact that
	 * it is integral math where 1/2=0. This intermediate value becomes the
	 * number of bits to shift the DBAM register to extract the proper CSROW
	 * field.
	 */
	dram_map = (pvt->dbam0 >> ((csrow_nr / 2) * 4)) & 0xF;

	nr_pages = pvt->ops->dbam_map_to_pages(pvt, dram_map);

	/*
	 * If dual channel then double the memory size of single channel.
	 * Channel count is 1 or 2
	 */
	nr_pages <<= (pvt->channel_count - 1);

	debugf0("  (csrow=%d) DBAM map index= %d\n", csrow_nr, dram_map);
	debugf0("    nr_pages= %u  channel-count = %d\n",
		nr_pages, pvt->channel_count);

	return nr_pages;
}

/*
 * Initialize the array of csrow attribute instances, based on the values
 * from pci config hardware registers.
 */
static int amd64_init_csrows(struct mem_ctl_info *mci)
{
	struct csrow_info *csrow;
	struct amd64_pvt *pvt;
	u64 input_addr_min, input_addr_max, sys_addr;
	int i, empty = 1;

	pvt = mci->pvt_info;

	amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_NBCFG, &pvt->nbcfg);

	debugf0("NBCFG= 0x%x  CHIPKILL= %s DRAM ECC= %s\n", pvt->nbcfg,
		(pvt->nbcfg & K8_NBCFG_CHIPKILL) ? "Enabled" : "Disabled",
		(pvt->nbcfg & K8_NBCFG_ECC_ENABLE) ? "Enabled" : "Disabled"
		);

	for (i = 0; i < pvt->cs_count; i++) {
		csrow = &mci->csrows[i];

		if ((pvt->dcsb0[i] & K8_DCSB_CS_ENABLE) == 0) {
			debugf1("----CSROW %d EMPTY for node %d\n", i,
				pvt->mc_node_id);
			continue;
		}

		debugf1("----CSROW %d VALID for MC node %d\n",
			i, pvt->mc_node_id);

		empty = 0;
		csrow->nr_pages = amd64_csrow_nr_pages(i, pvt);
		find_csrow_limits(mci, i, &input_addr_min, &input_addr_max);
		sys_addr = input_addr_to_sys_addr(mci, input_addr_min);
		csrow->first_page = (u32) (sys_addr >> PAGE_SHIFT);
		sys_addr = input_addr_to_sys_addr(mci, input_addr_max);
		csrow->last_page = (u32) (sys_addr >> PAGE_SHIFT);
		csrow->page_mask = ~mask_from_dct_mask(pvt, i);
		/* 8 bytes of resolution */

		csrow->mtype = amd64_determine_memory_type(pvt);

		debugf1("  for MC node %d csrow %d:\n", pvt->mc_node_id, i);
		debugf1("    input_addr_min: 0x%lx input_addr_max: 0x%lx\n",
			(unsigned long)input_addr_min,
			(unsigned long)input_addr_max);
		debugf1("    sys_addr: 0x%lx  page_mask: 0x%lx\n",
			(unsigned long)sys_addr, csrow->page_mask);
		debugf1("    nr_pages: %u  first_page: 0x%lx "
			"last_page: 0x%lx\n",
			(unsigned)csrow->nr_pages,
			csrow->first_page, csrow->last_page);

		/*
		 * determine whether CHIPKILL or JUST ECC or NO ECC is operating
		 */
		if (pvt->nbcfg & K8_NBCFG_ECC_ENABLE)
			csrow->edac_mode =
			    (pvt->nbcfg & K8_NBCFG_CHIPKILL) ?
			    EDAC_S4ECD4ED : EDAC_SECDED;
		else
			csrow->edac_mode = EDAC_NONE;
	}

	return empty;
}

/* get all cores on this DCT */
static void get_cpus_on_this_dct_cpumask(struct cpumask *mask, int nid)
{
	int cpu;

	for_each_online_cpu(cpu)
		if (amd_get_nb_id(cpu) == nid)
			cpumask_set_cpu(cpu, mask);
}

/* check MCG_CTL on all the cpus on this node */
static bool amd64_nb_mce_bank_enabled_on_node(int nid)
{
	cpumask_var_t mask;
	struct msr *msrs;
	int cpu, nbe, idx = 0;
	bool ret = false;

	if (!zalloc_cpumask_var(&mask, GFP_KERNEL)) {
		amd64_printk(KERN_WARNING, "%s: error allocating mask\n",
			     __func__);
		return false;
	}

	get_cpus_on_this_dct_cpumask(mask, nid);

	msrs = kzalloc(sizeof(struct msr) * cpumask_weight(mask), GFP_KERNEL);
	if (!msrs) {
		amd64_printk(KERN_WARNING, "%s: error allocating msrs\n",
			      __func__);
		free_cpumask_var(mask);
		 return false;
	}

	rdmsr_on_cpus(mask, MSR_IA32_MCG_CTL, msrs);

	for_each_cpu(cpu, mask) {
		nbe = msrs[idx].l & K8_MSR_MCGCTL_NBE;

		debugf0("core: %u, MCG_CTL: 0x%llx, NB MSR is %s\n",
			cpu, msrs[idx].q,
			(nbe ? "enabled" : "disabled"));

		if (!nbe)
			goto out;

		idx++;
	}
	ret = true;

out:
	kfree(msrs);
	free_cpumask_var(mask);
	return ret;
}

static int amd64_toggle_ecc_err_reporting(struct amd64_pvt *pvt, bool on)
{
	cpumask_var_t cmask;
	struct msr *msrs = NULL;
	int cpu, idx = 0;

	if (!zalloc_cpumask_var(&cmask, GFP_KERNEL)) {
		amd64_printk(KERN_WARNING, "%s: error allocating mask\n",
			     __func__);
		return false;
	}

	get_cpus_on_this_dct_cpumask(cmask, pvt->mc_node_id);

	msrs = kzalloc(sizeof(struct msr) * cpumask_weight(cmask), GFP_KERNEL);
	if (!msrs) {
		amd64_printk(KERN_WARNING, "%s: error allocating msrs\n",
			     __func__);
		return -ENOMEM;
	}

	rdmsr_on_cpus(cmask, MSR_IA32_MCG_CTL, msrs);

	for_each_cpu(cpu, cmask) {

		if (on) {
			if (msrs[idx].l & K8_MSR_MCGCTL_NBE)
				pvt->flags.ecc_report = 1;

			msrs[idx].l |= K8_MSR_MCGCTL_NBE;
		} else {
			/*
			 * Turn off ECC reporting only when it was off before
			 */
			if (!pvt->flags.ecc_report)
				msrs[idx].l &= ~K8_MSR_MCGCTL_NBE;
		}
		idx++;
	}
	wrmsr_on_cpus(cmask, MSR_IA32_MCG_CTL, msrs);

	kfree(msrs);
	free_cpumask_var(cmask);

	return 0;
}

/*
 * Only if 'ecc_enable_override' is set AND BIOS had ECC disabled, do "we"
 * enable it.
 */
static void amd64_enable_ecc_error_reporting(struct mem_ctl_info *mci)
{
	struct amd64_pvt *pvt = mci->pvt_info;
	u32 value, mask = K8_NBCTL_CECCEn | K8_NBCTL_UECCEn;

	if (!ecc_enable_override)
		return;

	amd64_printk(KERN_WARNING,
		"'ecc_enable_override' parameter is active, "
		"Enabling AMD ECC hardware now: CAUTION\n");

	amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_NBCTL, &value);

	/* turn on UECCn and CECCEn bits */
	pvt->old_nbctl = value & mask;
	pvt->nbctl_mcgctl_saved = 1;

	value |= mask;
	pci_write_config_dword(pvt->misc_f3_ctl, K8_NBCTL, value);

	if (amd64_toggle_ecc_err_reporting(pvt, ON))
		amd64_printk(KERN_WARNING, "Error enabling ECC reporting over "
					   "MCGCTL!\n");

	amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_NBCFG, &value);

	debugf0("NBCFG(1)= 0x%x  CHIPKILL= %s ECC_ENABLE= %s\n", value,
		(value & K8_NBCFG_CHIPKILL) ? "Enabled" : "Disabled",
		(value & K8_NBCFG_ECC_ENABLE) ? "Enabled" : "Disabled");

	if (!(value & K8_NBCFG_ECC_ENABLE)) {
		amd64_printk(KERN_WARNING,
			"This node reports that DRAM ECC is "
			"currently Disabled; ENABLING now\n");

		/* Attempt to turn on DRAM ECC Enable */
		value |= K8_NBCFG_ECC_ENABLE;
		pci_write_config_dword(pvt->misc_f3_ctl, K8_NBCFG, value);

		amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_NBCFG, &value);

		if (!(value & K8_NBCFG_ECC_ENABLE)) {
			amd64_printk(KERN_WARNING,
				"Hardware rejects Enabling DRAM ECC checking\n"
				"Check memory DIMM configuration\n");
		} else {
			amd64_printk(KERN_DEBUG,
				"Hardware accepted DRAM ECC Enable\n");
		}
	}
	debugf0("NBCFG(2)= 0x%x  CHIPKILL= %s ECC_ENABLE= %s\n", value,
		(value & K8_NBCFG_CHIPKILL) ? "Enabled" : "Disabled",
		(value & K8_NBCFG_ECC_ENABLE) ? "Enabled" : "Disabled");

	pvt->ctl_error_info.nbcfg = value;
}

static void amd64_restore_ecc_error_reporting(struct amd64_pvt *pvt)
{
	u32 value, mask = K8_NBCTL_CECCEn | K8_NBCTL_UECCEn;

	if (!pvt->nbctl_mcgctl_saved)
		return;

	amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_NBCTL, &value);
	value &= ~mask;
	value |= pvt->old_nbctl;

	/* restore the NB Enable MCGCTL bit */
	pci_write_config_dword(pvt->misc_f3_ctl, K8_NBCTL, value);

	if (amd64_toggle_ecc_err_reporting(pvt, OFF))
		amd64_printk(KERN_WARNING, "Error restoring ECC reporting over "
					   "MCGCTL!\n");
}

/*
 * EDAC requires that the BIOS have ECC enabled before taking over the
 * processing of ECC errors. This is because the BIOS can properly initialize
 * the memory system completely. A command line option allows to force-enable
 * hardware ECC later in amd64_enable_ecc_error_reporting().
 */
static const char *ecc_warning =
	"WARNING: ECC is disabled by BIOS. Module will NOT be loaded.\n"
	" Either Enable ECC in the BIOS, or set 'ecc_enable_override'.\n"
	" Also, use of the override can cause unknown side effects.\n";

static int amd64_check_ecc_enabled(struct amd64_pvt *pvt)
{
	u32 value;
	u8 ecc_enabled = 0;
	bool nb_mce_en = false;

	amd64_read_pci_cfg(pvt->misc_f3_ctl, K8_NBCFG, &value);

	ecc_enabled = !!(value & K8_NBCFG_ECC_ENABLE);
	if (!ecc_enabled)
		amd64_printk(KERN_WARNING, "This node reports that Memory ECC "
			     "is currently disabled, set F3x%x[22] (%s).\n",
			     K8_NBCFG, pci_name(pvt->misc_f3_ctl));
	else
		amd64_printk(KERN_INFO, "ECC is enabled by BIOS.\n");

	nb_mce_en = amd64_nb_mce_bank_enabled_on_node(pvt->mc_node_id);
	if (!nb_mce_en)
		amd64_printk(KERN_WARNING, "NB MCE bank disabled, set MSR "
			     "0x%08x[4] on node %d to enable.\n",
			     MSR_IA32_MCG_CTL, pvt->mc_node_id);

	if (!ecc_enabled || !nb_mce_en) {
		if (!ecc_enable_override) {
			amd64_printk(KERN_WARNING, "%s", ecc_warning);
			return -ENODEV;
		}
	} else
		/* CLEAR the override, since BIOS controlled it */
		ecc_enable_override = 0;

	return 0;
}

struct mcidev_sysfs_attribute sysfs_attrs[ARRAY_SIZE(amd64_dbg_attrs) +
					  ARRAY_SIZE(amd64_inj_attrs) +
					  1];

struct mcidev_sysfs_attribute terminator = { .attr = { .name = NULL } };

static void amd64_set_mc_sysfs_attributes(struct mem_ctl_info *mci)
{
	unsigned int i = 0, j = 0;

	for (; i < ARRAY_SIZE(amd64_dbg_attrs); i++)
		sysfs_attrs[i] = amd64_dbg_attrs[i];

	for (j = 0; j < ARRAY_SIZE(amd64_inj_attrs); j++, i++)
		sysfs_attrs[i] = amd64_inj_attrs[j];

	sysfs_attrs[i] = terminator;

	mci->mc_driver_sysfs_attributes = sysfs_attrs;
}

static void amd64_setup_mci_misc_attributes(struct mem_ctl_info *mci)
{
	struct amd64_pvt *pvt = mci->pvt_info;

	mci->mtype_cap		= MEM_FLAG_DDR2 | MEM_FLAG_RDDR2;
	mci->edac_ctl_cap	= EDAC_FLAG_NONE;

	if (pvt->nbcap & K8_NBCAP_SECDED)
		mci->edac_ctl_cap |= EDAC_FLAG_SECDED;

	if (pvt->nbcap & K8_NBCAP_CHIPKILL)
		mci->edac_ctl_cap |= EDAC_FLAG_S4ECD4ED;

	mci->edac_cap		= amd64_determine_edac_cap(pvt);
	mci->mod_name		= EDAC_MOD_STR;
	mci->mod_ver		= EDAC_AMD64_VERSION;
	mci->ctl_name		= get_amd_family_name(pvt->mc_type_index);
	mci->dev_name		= pci_name(pvt->dram_f2_ctl);
	mci->ctl_page_to_phys	= NULL;

	/* IMPORTANT: Set the polling 'check' function in this module */
	mci->edac_check		= amd64_check;

	/* memory scrubber interface */
	mci->set_sdram_scrub_rate = amd64_set_scrub_rate;
	mci->get_sdram_scrub_rate = amd64_get_scrub_rate;
}

/*
 * Init stuff for this DRAM Controller device.
 *
 * Due to a hardware feature on Fam10h CPUs, the Enable Extended Configuration
 * Space feature MUST be enabled on ALL Processors prior to actually reading
 * from the ECS registers. Since the loading of the module can occur on any
 * 'core', and cores don't 'see' all the other processors ECS data when the
 * others are NOT enabled. Our solution is to first enable ECS access in this
 * routine on all processors, gather some data in a amd64_pvt structure and
 * later come back in a finish-setup function to perform that final
 * initialization. See also amd64_init_2nd_stage() for that.
 */
static int amd64_probe_one_instance(struct pci_dev *dram_f2_ctl,
				    int mc_type_index)
{
	struct amd64_pvt *pvt = NULL;
	int err = 0, ret;

	ret = -ENOMEM;
	pvt = kzalloc(sizeof(struct amd64_pvt), GFP_KERNEL);
	if (!pvt)
		goto err_exit;

	pvt->mc_node_id = get_node_id(dram_f2_ctl);

	pvt->dram_f2_ctl	= dram_f2_ctl;
	pvt->ext_model		= boot_cpu_data.x86_model >> 4;
	pvt->mc_type_index	= mc_type_index;
	pvt->ops		= family_ops(mc_type_index);

	/*
	 * We have the dram_f2_ctl device as an argument, now go reserve its
	 * sibling devices from the PCI system.
	 */
	ret = -ENODEV;
	err = amd64_reserve_mc_sibling_devices(pvt, mc_type_index);
	if (err)
		goto err_free;

	ret = -EINVAL;
	err = amd64_check_ecc_enabled(pvt);
	if (err)
		goto err_put;

	/*
	 * Key operation here: setup of HW prior to performing ops on it. Some
	 * setup is required to access ECS data. After this is performed, the
	 * 'teardown' function must be called upon error and normal exit paths.
	 */
	if (boot_cpu_data.x86 >= 0x10)
		amd64_setup(pvt);

	/*
	 * Save the pointer to the private data for use in 2nd initialization
	 * stage
	 */
	pvt_lookup[pvt->mc_node_id] = pvt;

	return 0;

err_put:
	amd64_free_mc_sibling_devices(pvt);

err_free:
	kfree(pvt);

err_exit:
	return ret;
}

/*
 * This is the finishing stage of the init code. Needs to be performed after all
 * MCs' hardware have been prepped for accessing extended config space.
 */
static int amd64_init_2nd_stage(struct amd64_pvt *pvt)
{
	int node_id = pvt->mc_node_id;
	struct mem_ctl_info *mci;
	int ret, err = 0;

	amd64_read_mc_registers(pvt);

	ret = -ENODEV;
	if (pvt->ops->probe_valid_hardware) {
		err = pvt->ops->probe_valid_hardware(pvt);
		if (err)
			goto err_exit;
	}

	/*
	 * We need to determine how many memory channels there are. Then use
	 * that information for calculating the size of the dynamic instance
	 * tables in the 'mci' structure
	 */
	pvt->channel_count = pvt->ops->early_channel_count(pvt);
	if (pvt->channel_count < 0)
		goto err_exit;

	ret = -ENOMEM;
	mci = edac_mc_alloc(0, pvt->cs_count, pvt->channel_count, node_id);
	if (!mci)
		goto err_exit;

	mci->pvt_info = pvt;

	mci->dev = &pvt->dram_f2_ctl->dev;
	amd64_setup_mci_misc_attributes(mci);

	if (amd64_init_csrows(mci))
		mci->edac_cap = EDAC_FLAG_NONE;

	amd64_enable_ecc_error_reporting(mci);
	amd64_set_mc_sysfs_attributes(mci);

	ret = -ENODEV;
	if (edac_mc_add_mc(mci)) {
		debugf1("failed edac_mc_add_mc()\n");
		goto err_add_mc;
	}

	mci_lookup[node_id] = mci;
	pvt_lookup[node_id] = NULL;

	/* register stuff with EDAC MCE */
	if (report_gart_errors)
		amd_report_gart_errors(true);

	amd_register_ecc_decoder(amd64_decode_bus_error);

	return 0;

err_add_mc:
	edac_mc_free(mci);

err_exit:
	debugf0("failure to init 2nd stage: ret=%d\n", ret);

	amd64_restore_ecc_error_reporting(pvt);

	if (boot_cpu_data.x86 > 0xf)
		amd64_teardown(pvt);

	amd64_free_mc_sibling_devices(pvt);

	kfree(pvt_lookup[pvt->mc_node_id]);
	pvt_lookup[node_id] = NULL;

	return ret;
}


static int __devinit amd64_init_one_instance(struct pci_dev *pdev,
				 const struct pci_device_id *mc_type)
{
	int ret = 0;

	debugf0("(MC node=%d,mc_type='%s')\n", get_node_id(pdev),
		get_amd_family_name(mc_type->driver_data));

	ret = pci_enable_device(pdev);
	if (ret < 0)
		ret = -EIO;
	else
		ret = amd64_probe_one_instance(pdev, mc_type->driver_data);

	if (ret < 0)
		debugf0("ret=%d\n", ret);

	return ret;
}

static void __devexit amd64_remove_one_instance(struct pci_dev *pdev)
{
	struct mem_ctl_info *mci;
	struct amd64_pvt *pvt;

	/* Remove from EDAC CORE tracking list */
	mci = edac_mc_del_mc(&pdev->dev);
	if (!mci)
		return;

	pvt = mci->pvt_info;

	amd64_restore_ecc_error_reporting(pvt);

	if (boot_cpu_data.x86 > 0xf)
		amd64_teardown(pvt);

	amd64_free_mc_sibling_devices(pvt);

	kfree(pvt);
	mci->pvt_info = NULL;

	mci_lookup[pvt->mc_node_id] = NULL;

	/* unregister from EDAC MCE */
	amd_report_gart_errors(false);
	amd_unregister_ecc_decoder(amd64_decode_bus_error);

	/* Free the EDAC CORE resources */
	edac_mc_free(mci);
}

/*
 * This table is part of the interface for loading drivers for PCI devices. The
 * PCI core identifies what devices are on a system during boot, and then
 * inquiry this table to see if this driver is for a given device found.
 */
static const struct pci_device_id amd64_pci_table[] __devinitdata = {
	{
		.vendor		= PCI_VENDOR_ID_AMD,
		.device		= PCI_DEVICE_ID_AMD_K8_NB_MEMCTL,
		.subvendor	= PCI_ANY_ID,
		.subdevice	= PCI_ANY_ID,
		.class		= 0,
		.class_mask	= 0,
		.driver_data	= K8_CPUS
	},
	{
		.vendor		= PCI_VENDOR_ID_AMD,
		.device		= PCI_DEVICE_ID_AMD_10H_NB_DRAM,
		.subvendor	= PCI_ANY_ID,
		.subdevice	= PCI_ANY_ID,
		.class		= 0,
		.class_mask	= 0,
		.driver_data	= F10_CPUS
	},
	{
		.vendor		= PCI_VENDOR_ID_AMD,
		.device		= PCI_DEVICE_ID_AMD_11H_NB_DRAM,
		.subvendor	= PCI_ANY_ID,
		.subdevice	= PCI_ANY_ID,
		.class		= 0,
		.class_mask	= 0,
		.driver_data	= F11_CPUS
	},
	{0, }
};
MODULE_DEVICE_TABLE(pci, amd64_pci_table);

static struct pci_driver amd64_pci_driver = {
	.name		= EDAC_MOD_STR,
	.probe		= amd64_init_one_instance,
	.remove		= __devexit_p(amd64_remove_one_instance),
	.id_table	= amd64_pci_table,
};

static void amd64_setup_pci_device(void)
{
	struct mem_ctl_info *mci;
	struct amd64_pvt *pvt;

	if (amd64_ctl_pci)
		return;

	mci = mci_lookup[0];
	if (mci) {

		pvt = mci->pvt_info;
		amd64_ctl_pci =
			edac_pci_create_generic_ctl(&pvt->dram_f2_ctl->dev,
						    EDAC_MOD_STR);

		if (!amd64_ctl_pci) {
			pr_warning("%s(): Unable to create PCI control\n",
				   __func__);

			pr_warning("%s(): PCI error report via EDAC not set\n",
				   __func__);
			}
	}
}

static int __init amd64_edac_init(void)
{
	int nb, err = -ENODEV;

	edac_printk(KERN_INFO, EDAC_MOD_STR, EDAC_AMD64_VERSION "\n");

	opstate_init();

	if (cache_k8_northbridges() < 0)
		return err;

	err = pci_register_driver(&amd64_pci_driver);
	if (err)
		return err;

	/*
	 * At this point, the array 'pvt_lookup[]' contains pointers to alloc'd
	 * amd64_pvt structs. These will be used in the 2nd stage init function
	 * to finish initialization of the MC instances.
	 */
	for (nb = 0; nb < num_k8_northbridges; nb++) {
		if (!pvt_lookup[nb])
			continue;

		err = amd64_init_2nd_stage(pvt_lookup[nb]);
		if (err)
			goto err_2nd_stage;
	}

	amd64_setup_pci_device();

	return 0;

err_2nd_stage:
	debugf0("2nd stage failed\n");
	pci_unregister_driver(&amd64_pci_driver);

	return err;
}

static void __exit amd64_edac_exit(void)
{
	if (amd64_ctl_pci)
		edac_pci_release_generic_ctl(amd64_ctl_pci);

	pci_unregister_driver(&amd64_pci_driver);
}

module_init(amd64_edac_init);
module_exit(amd64_edac_exit);

MODULE_LICENSE("GPL");
MODULE_AUTHOR("SoftwareBitMaker: Doug Thompson, "
		"Dave Peterson, Thayne Harbaugh");
MODULE_DESCRIPTION("MC support for AMD64 memory controllers - "
		EDAC_AMD64_VERSION);

module_param(edac_op_state, int, 0444);
MODULE_PARM_DESC(edac_op_state, "EDAC Error Reporting state: 0=Poll,1=NMI");