sb_edac.c 54.0 KB
Newer Older
1 2 3 4 5 6 7 8 9
/* Intel Sandy Bridge -EN/-EP/-EX Memory Controller kernel module
 *
 * This driver supports the memory controllers found on the Intel
 * processor family Sandy Bridge.
 *
 * This file may be distributed under the terms of the
 * GNU General Public License version 2 only.
 *
 * Copyright (c) 2011 by:
10
 *	 Mauro Carvalho Chehab
11 12 13 14 15 16 17 18 19 20 21 22
 */

#include <linux/module.h>
#include <linux/init.h>
#include <linux/pci.h>
#include <linux/pci_ids.h>
#include <linux/slab.h>
#include <linux/delay.h>
#include <linux/edac.h>
#include <linux/mmzone.h>
#include <linux/smp.h>
#include <linux/bitmap.h>
23
#include <linux/math64.h>
24
#include <asm/processor.h>
25
#include <asm/mce.h>
26 27 28 29 30 31 32 33 34 35 36

#include "edac_core.h"

/* Static vars */
static LIST_HEAD(sbridge_edac_list);
static DEFINE_MUTEX(sbridge_edac_lock);
static int probed;

/*
 * Alter this version for the module when modifications are made
 */
37
#define SBRIDGE_REVISION    " Ver: 1.1.0 "
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
#define EDAC_MOD_STR      "sbridge_edac"

/*
 * Debug macros
 */
#define sbridge_printk(level, fmt, arg...)			\
	edac_printk(level, "sbridge", fmt, ##arg)

#define sbridge_mc_printk(mci, level, fmt, arg...)		\
	edac_mc_chipset_printk(mci, level, "sbridge", fmt, ##arg)

/*
 * Get a bit field at register value <v>, from bit <lo> to bit <hi>
 */
#define GET_BITFIELD(v, lo, hi)	\
53
	(((v) & GENMASK_ULL(hi, lo)) >> (lo))
54 55 56 57 58 59 60

/*
 * sbridge Memory Controller Registers
 */

/*
 * FIXME: For now, let's order by device function, as it makes
D
David Mackey 已提交
61
 * easier for driver's development process. This table should be
62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85
 * moved to pci_id.h when submitted upstream
 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_SAD0	0x3cf4	/* 12.6 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_SAD1	0x3cf6	/* 12.7 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_BR		0x3cf5	/* 13.6 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_HA0	0x3ca0	/* 14.0 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TA	0x3ca8	/* 15.0 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_RAS	0x3c71	/* 15.1 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD0	0x3caa	/* 15.2 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD1	0x3cab	/* 15.3 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD2	0x3cac	/* 15.4 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD3	0x3cad	/* 15.5 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_DDRIO	0x3cb8	/* 17.0 */

	/*
	 * Currently, unused, but will be needed in the future
	 * implementations, as they hold the error counters
	 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_ERR0	0x3c72	/* 16.2 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_ERR1	0x3c73	/* 16.3 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_ERR2	0x3c76	/* 16.6 */
#define PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_ERR3	0x3c77	/* 16.7 */

/* Devices 12 Function 6, Offsets 0x80 to 0xcc */
86
static const u32 sbridge_dram_rule[] = {
87 88 89 90
	0x80, 0x88, 0x90, 0x98, 0xa0,
	0xa8, 0xb0, 0xb8, 0xc0, 0xc8,
};

91 92 93 94 95 96
static const u32 ibridge_dram_rule[] = {
	0x60, 0x68, 0x70, 0x78, 0x80,
	0x88, 0x90, 0x98, 0xa0,	0xa8,
	0xb0, 0xb8, 0xc0, 0xc8, 0xd0,
	0xd8, 0xe0, 0xe8, 0xf0, 0xf8,
};
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116

#define SAD_LIMIT(reg)		((GET_BITFIELD(reg, 6, 25) << 26) | 0x3ffffff)
#define DRAM_ATTR(reg)		GET_BITFIELD(reg, 2,  3)
#define INTERLEAVE_MODE(reg)	GET_BITFIELD(reg, 1,  1)
#define DRAM_RULE_ENABLE(reg)	GET_BITFIELD(reg, 0,  0)

static char *get_dram_attr(u32 reg)
{
	switch(DRAM_ATTR(reg)) {
		case 0:
			return "DRAM";
		case 1:
			return "MMCFG";
		case 2:
			return "NXM";
		default:
			return "unknown";
	}
}

117
static const u32 sbridge_interleave_list[] = {
118 119 120 121
	0x84, 0x8c, 0x94, 0x9c, 0xa4,
	0xac, 0xb4, 0xbc, 0xc4, 0xcc,
};

122 123 124 125 126 127 128
static const u32 ibridge_interleave_list[] = {
	0x64, 0x6c, 0x74, 0x7c, 0x84,
	0x8c, 0x94, 0x9c, 0xa4, 0xac,
	0xb4, 0xbc, 0xc4, 0xcc, 0xd4,
	0xdc, 0xe4, 0xec, 0xf4, 0xfc,
};

A
Aristeu Rozanski 已提交
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
struct interleave_pkg {
	unsigned char start;
	unsigned char end;
};

static const struct interleave_pkg sbridge_interleave_pkg[] = {
	{ 0, 2 },
	{ 3, 5 },
	{ 8, 10 },
	{ 11, 13 },
	{ 16, 18 },
	{ 19, 21 },
	{ 24, 26 },
	{ 27, 29 },
};

145 146 147 148 149 150 151 152 153 154 155
static const struct interleave_pkg ibridge_interleave_pkg[] = {
	{ 0, 3 },
	{ 4, 7 },
	{ 8, 11 },
	{ 12, 15 },
	{ 16, 19 },
	{ 20, 23 },
	{ 24, 27 },
	{ 28, 31 },
};

A
Aristeu Rozanski 已提交
156 157
static inline int sad_pkg(const struct interleave_pkg *table, u32 reg,
			  int interleave)
158
{
A
Aristeu Rozanski 已提交
159 160
	return GET_BITFIELD(reg, table[interleave].start,
			    table[interleave].end);
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
}

/* Devices 12 Function 7 */

#define TOLM		0x80
#define	TOHM		0x84

#define GET_TOLM(reg)		((GET_BITFIELD(reg, 0,  3) << 28) | 0x3ffffff)
#define GET_TOHM(reg)		((GET_BITFIELD(reg, 0, 20) << 25) | 0x3ffffff)

/* Device 13 Function 6 */

#define SAD_TARGET	0xf0

#define SOURCE_ID(reg)		GET_BITFIELD(reg, 9, 11)

#define SAD_CONTROL	0xf4

/* Device 14 function 0 */

static const u32 tad_dram_rule[] = {
	0x40, 0x44, 0x48, 0x4c,
	0x50, 0x54, 0x58, 0x5c,
	0x60, 0x64, 0x68, 0x6c,
};
#define MAX_TAD	ARRAY_SIZE(tad_dram_rule)

#define TAD_LIMIT(reg)		((GET_BITFIELD(reg, 12, 31) << 26) | 0x3ffffff)
#define TAD_SOCK(reg)		GET_BITFIELD(reg, 10, 11)
#define TAD_CH(reg)		GET_BITFIELD(reg,  8,  9)
#define TAD_TGT3(reg)		GET_BITFIELD(reg,  6,  7)
#define TAD_TGT2(reg)		GET_BITFIELD(reg,  4,  5)
#define TAD_TGT1(reg)		GET_BITFIELD(reg,  2,  3)
#define TAD_TGT0(reg)		GET_BITFIELD(reg,  0,  1)

/* Device 15, function 0 */

#define MCMTR			0x7c

#define IS_ECC_ENABLED(mcmtr)		GET_BITFIELD(mcmtr, 2, 2)
#define IS_LOCKSTEP_ENABLED(mcmtr)	GET_BITFIELD(mcmtr, 1, 1)
#define IS_CLOSE_PG(mcmtr)		GET_BITFIELD(mcmtr, 0, 0)

/* Device 15, function 1 */

#define RASENABLES		0xac
#define IS_MIRROR_ENABLED(reg)		GET_BITFIELD(reg, 0, 0)

/* Device 15, functions 2-5 */

static const int mtr_regs[] = {
	0x80, 0x84, 0x88,
};

#define RANK_DISABLE(mtr)		GET_BITFIELD(mtr, 16, 19)
#define IS_DIMM_PRESENT(mtr)		GET_BITFIELD(mtr, 14, 14)
#define RANK_CNT_BITS(mtr)		GET_BITFIELD(mtr, 12, 13)
#define RANK_WIDTH_BITS(mtr)		GET_BITFIELD(mtr, 2, 4)
#define COL_WIDTH_BITS(mtr)		GET_BITFIELD(mtr, 0, 1)

static const u32 tad_ch_nilv_offset[] = {
	0x90, 0x94, 0x98, 0x9c,
	0xa0, 0xa4, 0xa8, 0xac,
	0xb0, 0xb4, 0xb8, 0xbc,
};
#define CHN_IDX_OFFSET(reg)		GET_BITFIELD(reg, 28, 29)
#define TAD_OFFSET(reg)			(GET_BITFIELD(reg,  6, 25) << 26)

static const u32 rir_way_limit[] = {
	0x108, 0x10c, 0x110, 0x114, 0x118,
};
#define MAX_RIR_RANGES ARRAY_SIZE(rir_way_limit)

#define IS_RIR_VALID(reg)	GET_BITFIELD(reg, 31, 31)
#define RIR_WAY(reg)		GET_BITFIELD(reg, 28, 29)

#define MAX_RIR_WAY	8

static const u32 rir_offset[MAX_RIR_RANGES][MAX_RIR_WAY] = {
	{ 0x120, 0x124, 0x128, 0x12c, 0x130, 0x134, 0x138, 0x13c },
	{ 0x140, 0x144, 0x148, 0x14c, 0x150, 0x154, 0x158, 0x15c },
	{ 0x160, 0x164, 0x168, 0x16c, 0x170, 0x174, 0x178, 0x17c },
	{ 0x180, 0x184, 0x188, 0x18c, 0x190, 0x194, 0x198, 0x19c },
	{ 0x1a0, 0x1a4, 0x1a8, 0x1ac, 0x1b0, 0x1b4, 0x1b8, 0x1bc },
};

#define RIR_RNK_TGT(reg)		GET_BITFIELD(reg, 16, 19)
#define RIR_OFFSET(reg)		GET_BITFIELD(reg,  2, 14)

/* Device 16, functions 2-7 */

/*
 * FIXME: Implement the error count reads directly
 */

static const u32 correrrcnt[] = {
	0x104, 0x108, 0x10c, 0x110,
};

#define RANK_ODD_OV(reg)		GET_BITFIELD(reg, 31, 31)
#define RANK_ODD_ERR_CNT(reg)		GET_BITFIELD(reg, 16, 30)
#define RANK_EVEN_OV(reg)		GET_BITFIELD(reg, 15, 15)
#define RANK_EVEN_ERR_CNT(reg)		GET_BITFIELD(reg,  0, 14)

static const u32 correrrthrsld[] = {
	0x11c, 0x120, 0x124, 0x128,
};

#define RANK_ODD_ERR_THRSLD(reg)	GET_BITFIELD(reg, 16, 30)
#define RANK_EVEN_ERR_THRSLD(reg)	GET_BITFIELD(reg,  0, 14)


/* Device 17, function 0 */

275
#define SB_RANK_CFG_A		0x0328
276

277
#define IB_RANK_CFG_A		0x0320
278 279 280 281 282 283 284 285

/*
 * sbridge structs
 */

#define NUM_CHANNELS	4
#define MAX_DIMMS	3		/* Max DIMMS per channel */

286 287 288 289 290
enum type {
	SANDY_BRIDGE,
	IVY_BRIDGE,
};

A
Aristeu Rozanski 已提交
291
struct sbridge_pvt;
292
struct sbridge_info {
293
	enum type	type;
294 295 296 297
	u32		mcmtr;
	u32		rankcfgr;
	u64		(*get_tolm)(struct sbridge_pvt *pvt);
	u64		(*get_tohm)(struct sbridge_pvt *pvt);
298
	u64		(*rir_limit)(u32 reg);
299
	const u32	*dram_rule;
300
	const u32	*interleave_list;
A
Aristeu Rozanski 已提交
301
	const struct interleave_pkg *interleave_pkg;
302
	u8		max_sad;
303
	u8		max_interleave;
304
	u8		(*get_node_id)(struct sbridge_pvt *pvt);
305
	enum mem_type	(*get_memory_type)(struct sbridge_pvt *pvt);
306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335
};

struct sbridge_channel {
	u32		ranks;
	u32		dimms;
};

struct pci_id_descr {
	int			dev;
	int			func;
	int 			dev_id;
	int			optional;
};

struct pci_id_table {
	const struct pci_id_descr	*descr;
	int				n_devs;
};

struct sbridge_dev {
	struct list_head	list;
	u8			bus, mc;
	u8			node_id, source_id;
	struct pci_dev		**pdev;
	int			n_devs;
	struct mem_ctl_info	*mci;
};

struct sbridge_pvt {
	struct pci_dev		*pci_ta, *pci_ddrio, *pci_ras;
336 337 338
	struct pci_dev		*pci_sad0, *pci_sad1;
	struct pci_dev		*pci_ha0, *pci_ha1;
	struct pci_dev		*pci_br0, *pci_br1;
339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362
	struct pci_dev		*pci_tad[NUM_CHANNELS];

	struct sbridge_dev	*sbridge_dev;

	struct sbridge_info	info;
	struct sbridge_channel	channel[NUM_CHANNELS];

	/* Memory type detection */
	bool			is_mirrored, is_lockstep, is_close_pg;

	/* Fifo double buffers */
	struct mce		mce_entry[MCE_LOG_LEN];
	struct mce		mce_outentry[MCE_LOG_LEN];

	/* Fifo in/out counters */
	unsigned		mce_in, mce_out;

	/* Count indicator to show errors not got */
	unsigned		mce_overrun;

	/* Memory description */
	u64			tolm, tohm;
};

363 364 365 366 367
#define PCI_DESCR(device, function, device_id, opt)	\
	.dev = (device),				\
	.func = (function),				\
	.dev_id = (device_id),				\
	.optional = opt
368 369 370

static const struct pci_id_descr pci_dev_descr_sbridge[] = {
		/* Processor Home Agent */
371
	{ PCI_DESCR(14, 0, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_HA0, 0)	},
372 373

		/* Memory controller */
374 375 376 377 378 379 380
	{ PCI_DESCR(15, 0, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TA, 0)	},
	{ PCI_DESCR(15, 1, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_RAS, 0)	},
	{ PCI_DESCR(15, 2, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD0, 0)	},
	{ PCI_DESCR(15, 3, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD1, 0)	},
	{ PCI_DESCR(15, 4, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD2, 0)	},
	{ PCI_DESCR(15, 5, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TAD3, 0)	},
	{ PCI_DESCR(17, 0, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_DDRIO, 1)	},
381 382

		/* System Address Decoder */
383 384
	{ PCI_DESCR(12, 6, PCI_DEVICE_ID_INTEL_SBRIDGE_SAD0, 0)		},
	{ PCI_DESCR(12, 7, PCI_DEVICE_ID_INTEL_SBRIDGE_SAD1, 0)		},
385 386

		/* Broadcast Registers */
387
	{ PCI_DESCR(13, 6, PCI_DEVICE_ID_INTEL_SBRIDGE_BR, 0)		},
388 389 390 391 392 393 394 395
};

#define PCI_ID_TABLE_ENTRY(A) { .descr=A, .n_devs = ARRAY_SIZE(A) }
static const struct pci_id_table pci_dev_descr_sbridge_table[] = {
	PCI_ID_TABLE_ENTRY(pci_dev_descr_sbridge),
	{0,}			/* 0 terminated list. */
};

396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458
/* This changes depending if 1HA or 2HA:
 * 1HA:
 *	0x0eb8 (17.0) is DDRIO0
 * 2HA:
 *	0x0ebc (17.4) is DDRIO0
 */
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_1HA_DDRIO0	0x0eb8
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_2HA_DDRIO0	0x0ebc

/* pci ids */
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0		0x0ea0
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TA		0x0ea8
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_RAS		0x0e71
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD0	0x0eaa
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD1	0x0eab
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD2	0x0eac
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD3	0x0ead
#define PCI_DEVICE_ID_INTEL_IBRIDGE_SAD			0x0ec8
#define PCI_DEVICE_ID_INTEL_IBRIDGE_BR0			0x0ec9
#define PCI_DEVICE_ID_INTEL_IBRIDGE_BR1			0x0eca
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1		0x0e60
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_TA		0x0e68
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_RAS		0x0e79
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_TAD0	0x0e6a
#define PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_TAD1	0x0e6b

static const struct pci_id_descr pci_dev_descr_ibridge[] = {
		/* Processor Home Agent */
	{ PCI_DESCR(14, 0, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0, 0)	},

		/* Memory controller */
	{ PCI_DESCR(15, 0, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TA, 0)	},
	{ PCI_DESCR(15, 1, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_RAS, 0)	},
	{ PCI_DESCR(15, 2, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD0, 0)	},
	{ PCI_DESCR(15, 3, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD1, 0)	},
	{ PCI_DESCR(15, 4, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD2, 0)	},
	{ PCI_DESCR(15, 5, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TAD3, 0)	},

		/* System Address Decoder */
	{ PCI_DESCR(22, 0, PCI_DEVICE_ID_INTEL_IBRIDGE_SAD, 0)		},

		/* Broadcast Registers */
	{ PCI_DESCR(22, 1, PCI_DEVICE_ID_INTEL_IBRIDGE_BR0, 1)		},
	{ PCI_DESCR(22, 2, PCI_DEVICE_ID_INTEL_IBRIDGE_BR1, 0)		},

		/* Optional, mode 2HA */
	{ PCI_DESCR(28, 0, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1, 1)	},
#if 0
	{ PCI_DESCR(29, 0, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_TA, 1)	},
	{ PCI_DESCR(29, 1, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_RAS, 1)	},
#endif
	{ PCI_DESCR(29, 2, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_TAD0, 1)	},
	{ PCI_DESCR(29, 3, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1_TAD1, 1)	},

	{ PCI_DESCR(17, 0, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_1HA_DDRIO0, 1) },
	{ PCI_DESCR(17, 4, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_2HA_DDRIO0, 1) },
};

static const struct pci_id_table pci_dev_descr_ibridge_table[] = {
	PCI_ID_TABLE_ENTRY(pci_dev_descr_ibridge),
	{0,}			/* 0 terminated list. */
};

459 460 461
/*
 *	pci_device_id	table for which devices we are looking for
 */
462
static const struct pci_device_id sbridge_pci_tbl[] = {
463
	{PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_SBRIDGE_IMC_TA)},
464
	{PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TA)},
465 466 467 468 469
	{0,}			/* 0 terminated list. */
};


/****************************************************************************
D
David Mackey 已提交
470
			Ancillary status routines
471 472 473 474 475 476 477
 ****************************************************************************/

static inline int numrank(u32 mtr)
{
	int ranks = (1 << RANK_CNT_BITS(mtr));

	if (ranks > 4) {
478 479
		edac_dbg(0, "Invalid number of ranks: %d (max = 4) raw value = %x (%04x)\n",
			 ranks, (unsigned int)RANK_CNT_BITS(mtr), mtr);
480 481 482 483 484 485 486 487 488 489 490
		return -EINVAL;
	}

	return ranks;
}

static inline int numrow(u32 mtr)
{
	int rows = (RANK_WIDTH_BITS(mtr) + 12);

	if (rows < 13 || rows > 18) {
491 492
		edac_dbg(0, "Invalid number of rows: %d (should be between 14 and 17) raw value = %x (%04x)\n",
			 rows, (unsigned int)RANK_WIDTH_BITS(mtr), mtr);
493 494 495 496 497 498 499 500 501 502 503
		return -EINVAL;
	}

	return 1 << rows;
}

static inline int numcol(u32 mtr)
{
	int cols = (COL_WIDTH_BITS(mtr) + 10);

	if (cols > 12) {
504 505
		edac_dbg(0, "Invalid number of cols: %d (max = 4) raw value = %x (%04x)\n",
			 cols, (unsigned int)COL_WIDTH_BITS(mtr), mtr);
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553
		return -EINVAL;
	}

	return 1 << cols;
}

static struct sbridge_dev *get_sbridge_dev(u8 bus)
{
	struct sbridge_dev *sbridge_dev;

	list_for_each_entry(sbridge_dev, &sbridge_edac_list, list) {
		if (sbridge_dev->bus == bus)
			return sbridge_dev;
	}

	return NULL;
}

static struct sbridge_dev *alloc_sbridge_dev(u8 bus,
					   const struct pci_id_table *table)
{
	struct sbridge_dev *sbridge_dev;

	sbridge_dev = kzalloc(sizeof(*sbridge_dev), GFP_KERNEL);
	if (!sbridge_dev)
		return NULL;

	sbridge_dev->pdev = kzalloc(sizeof(*sbridge_dev->pdev) * table->n_devs,
				   GFP_KERNEL);
	if (!sbridge_dev->pdev) {
		kfree(sbridge_dev);
		return NULL;
	}

	sbridge_dev->bus = bus;
	sbridge_dev->n_devs = table->n_devs;
	list_add_tail(&sbridge_dev->list, &sbridge_edac_list);

	return sbridge_dev;
}

static void free_sbridge_dev(struct sbridge_dev *sbridge_dev)
{
	list_del(&sbridge_dev->list);
	kfree(sbridge_dev->pdev);
	kfree(sbridge_dev);
}

A
Aristeu Rozanski 已提交
554 555 556 557 558 559 560 561 562
static u64 sbridge_get_tolm(struct sbridge_pvt *pvt)
{
	u32 reg;

	/* Address range is 32:28 */
	pci_read_config_dword(pvt->pci_sad1, TOLM, &reg);
	return GET_TOLM(reg);
}

A
Aristeu Rozanski 已提交
563 564 565 566 567 568 569 570
static u64 sbridge_get_tohm(struct sbridge_pvt *pvt)
{
	u32 reg;

	pci_read_config_dword(pvt->pci_sad1, TOHM, &reg);
	return GET_TOHM(reg);
}

571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588
static u64 ibridge_get_tolm(struct sbridge_pvt *pvt)
{
	u32 reg;

	pci_read_config_dword(pvt->pci_br1, TOLM, &reg);

	return GET_TOLM(reg);
}

static u64 ibridge_get_tohm(struct sbridge_pvt *pvt)
{
	u32 reg;

	pci_read_config_dword(pvt->pci_br1, TOHM, &reg);

	return GET_TOHM(reg);
}

589 590 591 592 593
static u64 rir_limit(u32 reg)
{
	return ((u64)GET_BITFIELD(reg,  1, 10) << 29) | 0x1fffffff;
}

594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612
static enum mem_type get_memory_type(struct sbridge_pvt *pvt)
{
	u32 reg;
	enum mem_type mtype;

	if (pvt->pci_ddrio) {
		pci_read_config_dword(pvt->pci_ddrio, pvt->info.rankcfgr,
				      &reg);
		if (GET_BITFIELD(reg, 11, 11))
			/* FIXME: Can also be LRDIMM */
			mtype = MEM_RDDR3;
		else
			mtype = MEM_DDR3;
	} else
		mtype = MEM_UNKNOWN;

	return mtype;
}

613 614 615 616 617 618 619
static u8 get_node_id(struct sbridge_pvt *pvt)
{
	u32 reg;
	pci_read_config_dword(pvt->pci_br0, SAD_CONTROL, &reg);
	return GET_BITFIELD(reg, 0, 2);
}

620 621 622 623 624 625 626 627 628 629 630
static inline u8 sad_pkg_socket(u8 pkg)
{
	/* on Ivy Bridge, nodeID is SASS, where A is HA and S is node id */
	return (pkg >> 3) | (pkg & 0x3);
}

static inline u8 sad_pkg_ha(u8 pkg)
{
	return (pkg >> 2) & 0x1;
}

631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648
/****************************************************************************
			Memory check routines
 ****************************************************************************/
static struct pci_dev *get_pdev_slot_func(u8 bus, unsigned slot,
					  unsigned func)
{
	struct sbridge_dev *sbridge_dev = get_sbridge_dev(bus);
	int i;

	if (!sbridge_dev)
		return NULL;

	for (i = 0; i < sbridge_dev->n_devs; i++) {
		if (!sbridge_dev->pdev[i])
			continue;

		if (PCI_SLOT(sbridge_dev->pdev[i]->devfn) == slot &&
		    PCI_FUNC(sbridge_dev->pdev[i]->devfn) == func) {
649 650
			edac_dbg(1, "Associated %02x.%02x.%d with %p\n",
				 bus, slot, func, sbridge_dev->pdev[i]);
651 652 653 654 655 656 657 658
			return sbridge_dev->pdev[i];
		}
	}

	return NULL;
}

/**
659
 * check_if_ecc_is_active() - Checks if ECC is active
660 661
 * bus:		Device bus
 */
662
static int check_if_ecc_is_active(const u8 bus)
663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682
{
	struct pci_dev *pdev = NULL;
	u32 mcmtr;

	pdev = get_pdev_slot_func(bus, 15, 0);
	if (!pdev) {
		sbridge_printk(KERN_ERR, "Couldn't find PCI device "
					"%2x.%02d.%d!!!\n",
					bus, 15, 0);
		return -ENODEV;
	}

	pci_read_config_dword(pdev, MCMTR, &mcmtr);
	if (!IS_ECC_ENABLED(mcmtr)) {
		sbridge_printk(KERN_ERR, "ECC is disabled. Aborting\n");
		return -ENODEV;
	}
	return 0;
}

683
static int get_dimm_config(struct mem_ctl_info *mci)
684 685
{
	struct sbridge_pvt *pvt = mci->pvt_info;
686
	struct dimm_info *dimm;
687 688
	unsigned i, j, banks, ranks, rows, cols, npages;
	u64 size;
689 690
	u32 reg;
	enum edac_type mode;
691
	enum mem_type mtype;
692

A
Aristeu Rozanski 已提交
693
	pci_read_config_dword(pvt->pci_br0, SAD_TARGET, &reg);
694 695
	pvt->sbridge_dev->source_id = SOURCE_ID(reg);

696
	pvt->sbridge_dev->node_id = pvt->info.get_node_id(pvt);
697 698 699 700
	edac_dbg(0, "mc#%d: Node ID: %d, source ID: %d\n",
		 pvt->sbridge_dev->mc,
		 pvt->sbridge_dev->node_id,
		 pvt->sbridge_dev->source_id);
701 702 703

	pci_read_config_dword(pvt->pci_ras, RASENABLES, &reg);
	if (IS_MIRROR_ENABLED(reg)) {
704
		edac_dbg(0, "Memory mirror is enabled\n");
705 706
		pvt->is_mirrored = true;
	} else {
707
		edac_dbg(0, "Memory mirror is disabled\n");
708 709 710 711 712
		pvt->is_mirrored = false;
	}

	pci_read_config_dword(pvt->pci_ta, MCMTR, &pvt->info.mcmtr);
	if (IS_LOCKSTEP_ENABLED(pvt->info.mcmtr)) {
713
		edac_dbg(0, "Lockstep is enabled\n");
714 715 716
		mode = EDAC_S8ECD8ED;
		pvt->is_lockstep = true;
	} else {
717
		edac_dbg(0, "Lockstep is disabled\n");
718 719 720 721
		mode = EDAC_S4ECD4ED;
		pvt->is_lockstep = false;
	}
	if (IS_CLOSE_PG(pvt->info.mcmtr)) {
722
		edac_dbg(0, "address map is on closed page mode\n");
723 724
		pvt->is_close_pg = true;
	} else {
725
		edac_dbg(0, "address map is on open page mode\n");
726 727 728
		pvt->is_close_pg = false;
	}

729 730 731 732
	mtype = pvt->info.get_memory_type(pvt);
	if (mtype == MEM_RDDR3)
		edac_dbg(0, "Memory is registered\n");
	else if (mtype == MEM_UNKNOWN)
733
		edac_dbg(0, "Cannot determine memory type\n");
734 735
	else
		edac_dbg(0, "Memory is unregistered\n");
736 737 738 739 740 741 742 743

	/* On all supported DDR3 DIMM types, there are 8 banks available */
	banks = 8;

	for (i = 0; i < NUM_CHANNELS; i++) {
		u32 mtr;

		for (j = 0; j < ARRAY_SIZE(mtr_regs); j++) {
744 745
			dimm = EDAC_DIMM_PTR(mci->layers, mci->dimms, mci->n_layers,
				       i, j, 0);
746 747
			pci_read_config_dword(pvt->pci_tad[i],
					      mtr_regs[j], &mtr);
748
			edac_dbg(4, "Channel #%d  MTR%d = %x\n", i, j, mtr);
749 750 751 752 753 754 755 756
			if (IS_DIMM_PRESENT(mtr)) {
				pvt->channel[i].dimms++;

				ranks = numrank(mtr);
				rows = numrow(mtr);
				cols = numcol(mtr);

				/* DDR3 has 8 I/O banks */
757
				size = ((u64)rows * cols * banks * ranks) >> (20 - 3);
758 759
				npages = MiB_TO_PAGES(size);

760
				edac_dbg(0, "mc#%d: channel %d, dimm %d, %Ld Mb (%d pages) bank: %d, rank: %d, row: %#x, col: %#x\n",
761 762 763
					 pvt->sbridge_dev->mc, i, j,
					 size, npages,
					 banks, ranks, rows, cols);
764

765
				dimm->nr_pages = npages;
766 767 768 769 770
				dimm->grain = 32;
				dimm->dtype = (banks == 8) ? DEV_X8 : DEV_X4;
				dimm->mtype = mtype;
				dimm->edac_mode = mode;
				snprintf(dimm->label, sizeof(dimm->label),
771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786
					 "CPU_SrcID#%u_Channel#%u_DIMM#%u",
					 pvt->sbridge_dev->source_id, i, j);
			}
		}
	}

	return 0;
}

static void get_memory_layout(const struct mem_ctl_info *mci)
{
	struct sbridge_pvt *pvt = mci->pvt_info;
	int i, j, k, n_sads, n_tads, sad_interl;
	u32 reg;
	u64 limit, prv = 0;
	u64 tmp_mb;
787
	u32 mb, kb;
788 789 790 791 792 793
	u32 rir_way;

	/*
	 * Step 1) Get TOLM/TOHM ranges
	 */

A
Aristeu Rozanski 已提交
794
	pvt->tolm = pvt->info.get_tolm(pvt);
795 796
	tmp_mb = (1 + pvt->tolm) >> 20;

797
	mb = div_u64_rem(tmp_mb, 1000, &kb);
798
	edac_dbg(0, "TOLM: %u.%03u GB (0x%016Lx)\n", mb, kb, (u64)pvt->tolm);
799 800

	/* Address range is already 45:25 */
A
Aristeu Rozanski 已提交
801
	pvt->tohm = pvt->info.get_tohm(pvt);
802 803
	tmp_mb = (1 + pvt->tohm) >> 20;

804
	mb = div_u64_rem(tmp_mb, 1000, &kb);
805
	edac_dbg(0, "TOHM: %u.%03u GB (0x%016Lx)\n", mb, kb, (u64)pvt->tohm);
806 807 808 809 810 811 812 813

	/*
	 * Step 2) Get SAD range and SAD Interleave list
	 * TAD registers contain the interleave wayness. However, it
	 * seems simpler to just discover it indirectly, with the
	 * algorithm bellow.
	 */
	prv = 0;
814
	for (n_sads = 0; n_sads < pvt->info.max_sad; n_sads++) {
815
		/* SAD_LIMIT Address range is 45:26 */
816
		pci_read_config_dword(pvt->pci_sad0, pvt->info.dram_rule[n_sads],
817 818 819 820 821 822 823 824 825 826
				      &reg);
		limit = SAD_LIMIT(reg);

		if (!DRAM_RULE_ENABLE(reg))
			continue;

		if (limit <= prv)
			break;

		tmp_mb = (limit + 1) >> 20;
827
		mb = div_u64_rem(tmp_mb, 1000, &kb);
828 829 830 831 832 833 834
		edac_dbg(0, "SAD#%d %s up to %u.%03u GB (0x%016Lx) Interleave: %s reg=0x%08x\n",
			 n_sads,
			 get_dram_attr(reg),
			 mb, kb,
			 ((u64)tmp_mb) << 20L,
			 INTERLEAVE_MODE(reg) ? "8:6" : "[8:6]XOR[18:16]",
			 reg);
835 836
		prv = limit;

837
		pci_read_config_dword(pvt->pci_sad0, pvt->info.interleave_list[n_sads],
838
				      &reg);
A
Aristeu Rozanski 已提交
839
		sad_interl = sad_pkg(pvt->info.interleave_pkg, reg, 0);
840
		for (j = 0; j < 8; j++) {
A
Aristeu Rozanski 已提交
841 842
			u32 pkg = sad_pkg(pvt->info.interleave_pkg, reg, j);
			if (j > 0 && sad_interl == pkg)
843 844
				break;

845
			edac_dbg(0, "SAD#%d, interleave #%d: %d\n",
A
Aristeu Rozanski 已提交
846
				 n_sads, j, pkg);
847 848 849 850 851 852 853 854 855 856 857 858 859 860 861
		}
	}

	/*
	 * Step 3) Get TAD range
	 */
	prv = 0;
	for (n_tads = 0; n_tads < MAX_TAD; n_tads++) {
		pci_read_config_dword(pvt->pci_ha0, tad_dram_rule[n_tads],
				      &reg);
		limit = TAD_LIMIT(reg);
		if (limit <= prv)
			break;
		tmp_mb = (limit + 1) >> 20;

862
		mb = div_u64_rem(tmp_mb, 1000, &kb);
863 864 865 866 867 868 869 870 871 872
		edac_dbg(0, "TAD#%d: up to %u.%03u GB (0x%016Lx), socket interleave %d, memory interleave %d, TGT: %d, %d, %d, %d, reg=0x%08x\n",
			 n_tads, mb, kb,
			 ((u64)tmp_mb) << 20L,
			 (u32)TAD_SOCK(reg),
			 (u32)TAD_CH(reg),
			 (u32)TAD_TGT0(reg),
			 (u32)TAD_TGT1(reg),
			 (u32)TAD_TGT2(reg),
			 (u32)TAD_TGT3(reg),
			 reg);
873
		prv = limit;
874 875 876 877 878 879 880 881 882 883 884 885 886
	}

	/*
	 * Step 4) Get TAD offsets, per each channel
	 */
	for (i = 0; i < NUM_CHANNELS; i++) {
		if (!pvt->channel[i].dimms)
			continue;
		for (j = 0; j < n_tads; j++) {
			pci_read_config_dword(pvt->pci_tad[i],
					      tad_ch_nilv_offset[j],
					      &reg);
			tmp_mb = TAD_OFFSET(reg) >> 20;
887
			mb = div_u64_rem(tmp_mb, 1000, &kb);
888 889 890 891 892
			edac_dbg(0, "TAD CH#%d, offset #%d: %u.%03u GB (0x%016Lx), reg=0x%08x\n",
				 i, j,
				 mb, kb,
				 ((u64)tmp_mb) << 20L,
				 reg);
893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909
		}
	}

	/*
	 * Step 6) Get RIR Wayness/Limit, per each channel
	 */
	for (i = 0; i < NUM_CHANNELS; i++) {
		if (!pvt->channel[i].dimms)
			continue;
		for (j = 0; j < MAX_RIR_RANGES; j++) {
			pci_read_config_dword(pvt->pci_tad[i],
					      rir_way_limit[j],
					      &reg);

			if (!IS_RIR_VALID(reg))
				continue;

910
			tmp_mb = pvt->info.rir_limit(reg) >> 20;
911
			rir_way = 1 << RIR_WAY(reg);
912
			mb = div_u64_rem(tmp_mb, 1000, &kb);
913 914 915 916 917 918
			edac_dbg(0, "CH#%d RIR#%d, limit: %u.%03u GB (0x%016Lx), way: %d, reg=0x%08x\n",
				 i, j,
				 mb, kb,
				 ((u64)tmp_mb) << 20L,
				 rir_way,
				 reg);
919 920 921 922 923 924 925

			for (k = 0; k < rir_way; k++) {
				pci_read_config_dword(pvt->pci_tad[i],
						      rir_offset[j][k],
						      &reg);
				tmp_mb = RIR_OFFSET(reg) << 6;

926
				mb = div_u64_rem(tmp_mb, 1000, &kb);
927 928 929 930 931 932
				edac_dbg(0, "CH#%d RIR#%d INTL#%d, offset %u.%03u GB (0x%016Lx), tgt: %d, reg=0x%08x\n",
					 i, j, k,
					 mb, kb,
					 ((u64)tmp_mb) << 20L,
					 (u32)RIR_RNK_TGT(reg),
					 reg);
933 934 935 936 937
			}
		}
	}
}

938
static struct mem_ctl_info *get_mci_for_node_id(u8 node_id)
939 940 941 942 943 944 945 946 947 948 949 950 951 952 953
{
	struct sbridge_dev *sbridge_dev;

	list_for_each_entry(sbridge_dev, &sbridge_edac_list, list) {
		if (sbridge_dev->node_id == node_id)
			return sbridge_dev->mci;
	}
	return NULL;
}

static int get_memory_error_data(struct mem_ctl_info *mci,
				 u64 addr,
				 u8 *socket,
				 long *channel_mask,
				 u8 *rank,
954
				 char **area_type, char *msg)
955 956 957
{
	struct mem_ctl_info	*new_mci;
	struct sbridge_pvt *pvt = mci->pvt_info;
958
	struct pci_dev		*pci_ha;
959 960 961
	int 			n_rir, n_sads, n_tads, sad_way, sck_xch;
	int			sad_interl, idx, base_ch;
	int			interleave_mode;
962
	unsigned		sad_interleave[pvt->info.max_interleave];
963
	u32			reg;
964
	u8			ch_way, sck_way, pkg, sad_ha = 0;
965 966
	u32			tad_offset;
	u32			rir_way;
967
	u32			mb, kb;
968
	u64			ch_addr, offset, limit = 0, prv = 0;
969 970 971 972 973 974 975 976 977


	/*
	 * Step 0) Check if the address is at special memory ranges
	 * The check bellow is probably enough to fill all cases where
	 * the error is not inside a memory, except for the legacy
	 * range (e. g. VGA addresses). It is unlikely, however, that the
	 * memory controller would generate an error on that range.
	 */
978
	if ((addr > (u64) pvt->tolm) && (addr < (1LL << 32))) {
979 980 981 982 983 984 985 986 987 988 989
		sprintf(msg, "Error at TOLM area, on addr 0x%08Lx", addr);
		return -EINVAL;
	}
	if (addr >= (u64)pvt->tohm) {
		sprintf(msg, "Error at MMIOH area, on addr 0x%016Lx", addr);
		return -EINVAL;
	}

	/*
	 * Step 1) Get socket
	 */
990 991
	for (n_sads = 0; n_sads < pvt->info.max_sad; n_sads++) {
		pci_read_config_dword(pvt->pci_sad0, pvt->info.dram_rule[n_sads],
992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
				      &reg);

		if (!DRAM_RULE_ENABLE(reg))
			continue;

		limit = SAD_LIMIT(reg);
		if (limit <= prv) {
			sprintf(msg, "Can't discover the memory socket");
			return -EINVAL;
		}
		if  (addr <= limit)
			break;
		prv = limit;
	}
1006
	if (n_sads == pvt->info.max_sad) {
1007 1008 1009
		sprintf(msg, "Can't discover the memory socket");
		return -EINVAL;
	}
1010
	*area_type = get_dram_attr(reg);
1011 1012
	interleave_mode = INTERLEAVE_MODE(reg);

1013
	pci_read_config_dword(pvt->pci_sad0, pvt->info.interleave_list[n_sads],
1014
			      &reg);
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039

	if (pvt->info.type == SANDY_BRIDGE) {
		sad_interl = sad_pkg(pvt->info.interleave_pkg, reg, 0);
		for (sad_way = 0; sad_way < 8; sad_way++) {
			u32 pkg = sad_pkg(pvt->info.interleave_pkg, reg, sad_way);
			if (sad_way > 0 && sad_interl == pkg)
				break;
			sad_interleave[sad_way] = pkg;
			edac_dbg(0, "SAD interleave #%d: %d\n",
				 sad_way, sad_interleave[sad_way]);
		}
		edac_dbg(0, "mc#%d: Error detected on SAD#%d: address 0x%016Lx < 0x%016Lx, Interleave [%d:6]%s\n",
			 pvt->sbridge_dev->mc,
			 n_sads,
			 addr,
			 limit,
			 sad_way + 7,
			 !interleave_mode ? "" : "XOR[18:16]");
		if (interleave_mode)
			idx = ((addr >> 6) ^ (addr >> 16)) & 7;
		else
			idx = (addr >> 6) & 7;
		switch (sad_way) {
		case 1:
			idx = 0;
1040
			break;
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
		case 2:
			idx = idx & 1;
			break;
		case 4:
			idx = idx & 3;
			break;
		case 8:
			break;
		default:
			sprintf(msg, "Can't discover socket interleave");
			return -EINVAL;
		}
		*socket = sad_interleave[idx];
		edac_dbg(0, "SAD interleave index: %d (wayness %d) = CPU socket %d\n",
			 idx, sad_way, *socket);
	} else {
		/* Ivy Bridge's SAD mode doesn't support XOR interleave mode */
1058
		idx = (addr >> 6) & 7;
1059 1060 1061 1062 1063
		pkg = sad_pkg(pvt->info.interleave_pkg, reg, idx);
		*socket = sad_pkg_socket(pkg);
		sad_ha = sad_pkg_ha(pkg);
		edac_dbg(0, "SAD interleave package: %d = CPU socket %d, HA %d\n",
			 idx, *socket, sad_ha);
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
	}

	/*
	 * Move to the proper node structure, in order to access the
	 * right PCI registers
	 */
	new_mci = get_mci_for_node_id(*socket);
	if (!new_mci) {
		sprintf(msg, "Struct for socket #%u wasn't initialized",
			*socket);
		return -EINVAL;
	}
	mci = new_mci;
	pvt = mci->pvt_info;

	/*
	 * Step 2) Get memory channel
	 */
	prv = 0;
1083 1084 1085 1086 1087 1088 1089 1090
	if (pvt->info.type == SANDY_BRIDGE)
		pci_ha = pvt->pci_ha0;
	else {
		if (sad_ha)
			pci_ha = pvt->pci_ha1;
		else
			pci_ha = pvt->pci_ha0;
	}
1091
	for (n_tads = 0; n_tads < MAX_TAD; n_tads++) {
1092
		pci_read_config_dword(pci_ha, tad_dram_rule[n_tads], &reg);
1093 1094 1095 1096 1097 1098 1099 1100 1101
		limit = TAD_LIMIT(reg);
		if (limit <= prv) {
			sprintf(msg, "Can't discover the memory channel");
			return -EINVAL;
		}
		if  (addr <= limit)
			break;
		prv = limit;
	}
1102 1103 1104 1105 1106
	if (n_tads == MAX_TAD) {
		sprintf(msg, "Can't discover the memory channel");
		return -EINVAL;
	}

1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137
	ch_way = TAD_CH(reg) + 1;
	sck_way = TAD_SOCK(reg) + 1;

	if (ch_way == 3)
		idx = addr >> 6;
	else
		idx = addr >> (6 + sck_way);
	idx = idx % ch_way;

	/*
	 * FIXME: Shouldn't we use CHN_IDX_OFFSET() here, when ch_way == 3 ???
	 */
	switch (idx) {
	case 0:
		base_ch = TAD_TGT0(reg);
		break;
	case 1:
		base_ch = TAD_TGT1(reg);
		break;
	case 2:
		base_ch = TAD_TGT2(reg);
		break;
	case 3:
		base_ch = TAD_TGT3(reg);
		break;
	default:
		sprintf(msg, "Can't discover the TAD target");
		return -EINVAL;
	}
	*channel_mask = 1 << base_ch;

1138 1139 1140 1141
	pci_read_config_dword(pvt->pci_tad[base_ch],
				tad_ch_nilv_offset[n_tads],
				&tad_offset);

1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160
	if (pvt->is_mirrored) {
		*channel_mask |= 1 << ((base_ch + 2) % 4);
		switch(ch_way) {
		case 2:
		case 4:
			sck_xch = 1 << sck_way * (ch_way >> 1);
			break;
		default:
			sprintf(msg, "Invalid mirror set. Can't decode addr");
			return -EINVAL;
		}
	} else
		sck_xch = (1 << sck_way) * ch_way;

	if (pvt->is_lockstep)
		*channel_mask |= 1 << ((base_ch + 1) % 4);

	offset = TAD_OFFSET(tad_offset);

1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
	edac_dbg(0, "TAD#%d: address 0x%016Lx < 0x%016Lx, socket interleave %d, channel interleave %d (offset 0x%08Lx), index %d, base ch: %d, ch mask: 0x%02lx\n",
		 n_tads,
		 addr,
		 limit,
		 (u32)TAD_SOCK(reg),
		 ch_way,
		 offset,
		 idx,
		 base_ch,
		 *channel_mask);
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184

	/* Calculate channel address */
	/* Remove the TAD offset */

	if (offset > addr) {
		sprintf(msg, "Can't calculate ch addr: TAD offset 0x%08Lx is too high for addr 0x%08Lx!",
			offset, addr);
		return -EINVAL;
	}
	addr -= offset;
	/* Store the low bits [0:6] of the addr */
	ch_addr = addr & 0x7f;
	/* Remove socket wayness and remove 6 bits */
	addr >>= 6;
1185
	addr = div_u64(addr, sck_xch);
1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203
#if 0
	/* Divide by channel way */
	addr = addr / ch_way;
#endif
	/* Recover the last 6 bits */
	ch_addr |= addr << 6;

	/*
	 * Step 3) Decode rank
	 */
	for (n_rir = 0; n_rir < MAX_RIR_RANGES; n_rir++) {
		pci_read_config_dword(pvt->pci_tad[base_ch],
				      rir_way_limit[n_rir],
				      &reg);

		if (!IS_RIR_VALID(reg))
			continue;

1204
		limit = pvt->info.rir_limit(reg);
1205
		mb = div_u64_rem(limit >> 20, 1000, &kb);
1206 1207 1208 1209 1210
		edac_dbg(0, "RIR#%d, limit: %u.%03u GB (0x%016Lx), way: %d\n",
			 n_rir,
			 mb, kb,
			 limit,
			 1 << RIR_WAY(reg));
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230
		if  (ch_addr <= limit)
			break;
	}
	if (n_rir == MAX_RIR_RANGES) {
		sprintf(msg, "Can't discover the memory rank for ch addr 0x%08Lx",
			ch_addr);
		return -EINVAL;
	}
	rir_way = RIR_WAY(reg);
	if (pvt->is_close_pg)
		idx = (ch_addr >> 6);
	else
		idx = (ch_addr >> 13);	/* FIXME: Datasheet says to shift by 15 */
	idx %= 1 << rir_way;

	pci_read_config_dword(pvt->pci_tad[base_ch],
			      rir_offset[n_rir][idx],
			      &reg);
	*rank = RIR_RNK_TGT(reg);

1231 1232 1233 1234 1235 1236
	edac_dbg(0, "RIR#%d: channel address 0x%08Lx < 0x%08Lx, RIR interleave %d, index %d\n",
		 n_rir,
		 ch_addr,
		 limit,
		 rir_way,
		 idx);
1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252

	return 0;
}

/****************************************************************************
	Device initialization routines: put/get, init/exit
 ****************************************************************************/

/*
 *	sbridge_put_all_devices	'put' all the devices that we have
 *				reserved via 'get'
 */
static void sbridge_put_devices(struct sbridge_dev *sbridge_dev)
{
	int i;

1253
	edac_dbg(0, "\n");
1254 1255 1256 1257
	for (i = 0; i < sbridge_dev->n_devs; i++) {
		struct pci_dev *pdev = sbridge_dev->pdev[i];
		if (!pdev)
			continue;
1258 1259 1260
		edac_dbg(0, "Removing dev %02x:%02x.%d\n",
			 pdev->bus->number,
			 PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
		pci_dev_put(pdev);
	}
}

static void sbridge_put_all_devices(void)
{
	struct sbridge_dev *sbridge_dev, *tmp;

	list_for_each_entry_safe(sbridge_dev, tmp, &sbridge_edac_list, list) {
		sbridge_put_devices(sbridge_dev);
		free_sbridge_dev(sbridge_dev);
	}
}

static int sbridge_get_onedevice(struct pci_dev **prev,
				 u8 *num_mc,
				 const struct pci_id_table *table,
				 const unsigned devno)
{
	struct sbridge_dev *sbridge_dev;
	const struct pci_id_descr *dev_descr = &table->descr[devno];

	struct pci_dev *pdev = NULL;
	u8 bus = 0;

1286
	sbridge_printk(KERN_DEBUG,
1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
		"Seeking for: dev %02x.%d PCI ID %04x:%04x\n",
		dev_descr->dev, dev_descr->func,
		PCI_VENDOR_ID_INTEL, dev_descr->dev_id);

	pdev = pci_get_device(PCI_VENDOR_ID_INTEL,
			      dev_descr->dev_id, *prev);

	if (!pdev) {
		if (*prev) {
			*prev = pdev;
			return 0;
		}

		if (dev_descr->optional)
			return 0;

		if (devno == 0)
			return -ENODEV;

		sbridge_printk(KERN_INFO,
			"Device not found: dev %02x.%d PCI ID %04x:%04x\n",
			dev_descr->dev, dev_descr->func,
			PCI_VENDOR_ID_INTEL, dev_descr->dev_id);

		/* End of list, leave */
		return -ENODEV;
	}
	bus = pdev->bus->number;

	sbridge_dev = get_sbridge_dev(bus);
	if (!sbridge_dev) {
		sbridge_dev = alloc_sbridge_dev(bus, table);
		if (!sbridge_dev) {
			pci_dev_put(pdev);
			return -ENOMEM;
		}
		(*num_mc)++;
	}

	if (sbridge_dev->pdev[devno]) {
		sbridge_printk(KERN_ERR,
			"Duplicated device for "
			"dev %02x:%d.%d PCI ID %04x:%04x\n",
			bus, dev_descr->dev, dev_descr->func,
			PCI_VENDOR_ID_INTEL, dev_descr->dev_id);
		pci_dev_put(pdev);
		return -ENODEV;
	}

	sbridge_dev->pdev[devno] = pdev;

	/* Sanity check */
	if (unlikely(PCI_SLOT(pdev->devfn) != dev_descr->dev ||
			PCI_FUNC(pdev->devfn) != dev_descr->func)) {
		sbridge_printk(KERN_ERR,
			"Device PCI ID %04x:%04x "
			"has dev %02x:%d.%d instead of dev %02x:%02x.%d\n",
			PCI_VENDOR_ID_INTEL, dev_descr->dev_id,
			bus, PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn),
			bus, dev_descr->dev, dev_descr->func);
		return -ENODEV;
	}

	/* Be sure that the device is enabled */
	if (unlikely(pci_enable_device(pdev) < 0)) {
		sbridge_printk(KERN_ERR,
			"Couldn't enable "
			"dev %02x:%d.%d PCI ID %04x:%04x\n",
			bus, dev_descr->dev, dev_descr->func,
			PCI_VENDOR_ID_INTEL, dev_descr->dev_id);
		return -ENODEV;
	}

1360 1361 1362
	edac_dbg(0, "Detected dev %02x:%d.%d PCI ID %04x:%04x\n",
		 bus, dev_descr->dev, dev_descr->func,
		 PCI_VENDOR_ID_INTEL, dev_descr->dev_id);
1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375

	/*
	 * As stated on drivers/pci/search.c, the reference count for
	 * @from is always decremented if it is not %NULL. So, as we need
	 * to get all devices up to null, we need to do a get for the device
	 */
	pci_dev_get(pdev);

	*prev = pdev;

	return 0;
}

1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387
/*
 * sbridge_get_all_devices - Find and perform 'get' operation on the MCH's
 *			     device/functions we want to reference for this driver.
 *			     Need to 'get' device 16 func 1 and func 2.
 * @num_mc: pointer to the memory controllers count, to be incremented in case
 * 	    of success.
 * @table: model specific table
 *
 * returns 0 in case of success or error code
 */
static int sbridge_get_all_devices(u8 *num_mc,
				   const struct pci_id_table *table)
1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
{
	int i, rc;
	struct pci_dev *pdev = NULL;

	while (table && table->descr) {
		for (i = 0; i < table->n_devs; i++) {
			pdev = NULL;
			do {
				rc = sbridge_get_onedevice(&pdev, num_mc,
							   table, i);
				if (rc < 0) {
					if (i == 0) {
						i = table->n_devs;
						break;
					}
					sbridge_put_all_devices();
					return -ENODEV;
				}
			} while (pdev);
		}
		table++;
	}

	return 0;
}

A
Aristeu Rozanski 已提交
1414 1415
static int sbridge_mci_bind_devs(struct mem_ctl_info *mci,
				 struct sbridge_dev *sbridge_dev)
1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442
{
	struct sbridge_pvt *pvt = mci->pvt_info;
	struct pci_dev *pdev;
	int i, func, slot;

	for (i = 0; i < sbridge_dev->n_devs; i++) {
		pdev = sbridge_dev->pdev[i];
		if (!pdev)
			continue;
		slot = PCI_SLOT(pdev->devfn);
		func = PCI_FUNC(pdev->devfn);
		switch (slot) {
		case 12:
			switch (func) {
			case 6:
				pvt->pci_sad0 = pdev;
				break;
			case 7:
				pvt->pci_sad1 = pdev;
				break;
			default:
				goto error;
			}
			break;
		case 13:
			switch (func) {
			case 6:
A
Aristeu Rozanski 已提交
1443
				pvt->pci_br0 = pdev;
1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488
				break;
			default:
				goto error;
			}
			break;
		case 14:
			switch (func) {
			case 0:
				pvt->pci_ha0 = pdev;
				break;
			default:
				goto error;
			}
			break;
		case 15:
			switch (func) {
			case 0:
				pvt->pci_ta = pdev;
				break;
			case 1:
				pvt->pci_ras = pdev;
				break;
			case 2:
			case 3:
			case 4:
			case 5:
				pvt->pci_tad[func - 2] = pdev;
				break;
			default:
				goto error;
			}
			break;
		case 17:
			switch (func) {
			case 0:
				pvt->pci_ddrio = pdev;
				break;
			default:
				goto error;
			}
			break;
		default:
			goto error;
		}

1489 1490 1491 1492
		edac_dbg(0, "Associated PCI %02x.%02d.%d with dev = %p\n",
			 sbridge_dev->bus,
			 PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn),
			 pdev);
1493 1494 1495 1496
	}

	/* Check if everything were registered */
	if (!pvt->pci_sad0 || !pvt->pci_sad1 || !pvt->pci_ha0 ||
1497
	    !pvt-> pci_tad || !pvt->pci_ras  || !pvt->pci_ta)
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
		goto enodev;

	for (i = 0; i < NUM_CHANNELS; i++) {
		if (!pvt->pci_tad[i])
			goto enodev;
	}
	return 0;

enodev:
	sbridge_printk(KERN_ERR, "Some needed devices are missing\n");
	return -ENODEV;

error:
	sbridge_printk(KERN_ERR, "Device %d, function %d "
		      "is out of the expected range\n",
		      slot, func);
	return -EINVAL;
}

1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
static int ibridge_mci_bind_devs(struct mem_ctl_info *mci,
				 struct sbridge_dev *sbridge_dev)
{
	struct sbridge_pvt *pvt = mci->pvt_info;
	struct pci_dev *pdev, *tmp;
	int i, func, slot;
	bool mode_2ha = false;

	tmp = pci_get_device(PCI_VENDOR_ID_INTEL,
			     PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA1, NULL);
	if (tmp) {
		mode_2ha = true;
		pci_dev_put(tmp);
	}

	for (i = 0; i < sbridge_dev->n_devs; i++) {
		pdev = sbridge_dev->pdev[i];
		if (!pdev)
			continue;
		slot = PCI_SLOT(pdev->devfn);
		func = PCI_FUNC(pdev->devfn);

		switch (slot) {
		case 14:
			if (func == 0) {
				pvt->pci_ha0 = pdev;
				break;
			}
			goto error;
		case 15:
			switch (func) {
			case 0:
				pvt->pci_ta = pdev;
				break;
			case 1:
				pvt->pci_ras = pdev;
				break;
			case 4:
			case 5:
				/* if we have 2 HAs active, channels 2 and 3
				 * are in other device */
				if (mode_2ha)
					break;
				/* fall through */
			case 2:
			case 3:
				pvt->pci_tad[func - 2] = pdev;
				break;
			default:
				goto error;
			}
			break;
		case 17:
			if (func == 4) {
				pvt->pci_ddrio = pdev;
				break;
			} else if (func == 0) {
				if (!mode_2ha)
					pvt->pci_ddrio = pdev;
				break;
			}
			goto error;
		case 22:
			switch (func) {
			case 0:
				pvt->pci_sad0 = pdev;
				break;
			case 1:
				pvt->pci_br0 = pdev;
				break;
			case 2:
				pvt->pci_br1 = pdev;
				break;
			default:
				goto error;
			}
			break;
		case 28:
			if (func == 0) {
				pvt->pci_ha1 = pdev;
				break;
			}
			goto error;
		case 29:
			/* we shouldn't have this device if we have just one
			 * HA present */
			WARN_ON(!mode_2ha);
			if (func == 2 || func == 3) {
				pvt->pci_tad[func] = pdev;
				break;
			}
			goto error;
		default:
			goto error;
		}

		edac_dbg(0, "Associated PCI %02x.%02d.%d with dev = %p\n",
			 sbridge_dev->bus,
			 PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn),
			 pdev);
	}

	/* Check if everything were registered */
	if (!pvt->pci_sad0 || !pvt->pci_ha0 || !pvt->pci_br0 ||
	    !pvt->pci_br1 || !pvt->pci_tad || !pvt->pci_ras  ||
	    !pvt->pci_ta)
		goto enodev;

	for (i = 0; i < NUM_CHANNELS; i++) {
		if (!pvt->pci_tad[i])
			goto enodev;
	}
	return 0;

enodev:
	sbridge_printk(KERN_ERR, "Some needed devices are missing\n");
	return -ENODEV;

error:
	sbridge_printk(KERN_ERR,
		       "Device %d, function %d is out of the expected range\n",
		       slot, func);
	return -EINVAL;
}

1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656
/****************************************************************************
			Error check routines
 ****************************************************************************/

/*
 * While Sandy Bridge has error count registers, SMI BIOS read values from
 * and resets the counters. So, they are not reliable for the OS to read
 * from them. So, we have no option but to just trust on whatever MCE is
 * telling us about the errors.
 */
static void sbridge_mce_output_error(struct mem_ctl_info *mci,
				    const struct mce *m)
{
	struct mem_ctl_info *new_mci;
	struct sbridge_pvt *pvt = mci->pvt_info;
1657
	enum hw_event_mc_err_type tp_event;
1658
	char *type, *optype, msg[256];
1659 1660 1661
	bool ripv = GET_BITFIELD(m->mcgstatus, 0, 0);
	bool overflow = GET_BITFIELD(m->status, 62, 62);
	bool uncorrected_error = GET_BITFIELD(m->status, 61, 61);
1662
	bool recoverable;
1663 1664 1665 1666 1667 1668 1669
	u32 core_err_cnt = GET_BITFIELD(m->status, 38, 52);
	u32 mscod = GET_BITFIELD(m->status, 16, 31);
	u32 errcode = GET_BITFIELD(m->status, 0, 15);
	u32 channel = GET_BITFIELD(m->status, 0, 3);
	u32 optypenum = GET_BITFIELD(m->status, 4, 6);
	long channel_mask, first_channel;
	u8  rank, socket;
1670
	int rc, dimm;
1671
	char *area_type = NULL;
1672

1673 1674 1675 1676 1677
	if (pvt->info.type == IVY_BRIDGE)
		recoverable = true;
	else
		recoverable = GET_BITFIELD(m->status, 56, 56);

1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689
	if (uncorrected_error) {
		if (ripv) {
			type = "FATAL";
			tp_event = HW_EVENT_ERR_FATAL;
		} else {
			type = "NON_FATAL";
			tp_event = HW_EVENT_ERR_UNCORRECTED;
		}
	} else {
		type = "CORRECTED";
		tp_event = HW_EVENT_ERR_CORRECTED;
	}
1690 1691

	/*
D
David Mackey 已提交
1692
	 * According with Table 15-9 of the Intel Architecture spec vol 3A,
1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706
	 * memory errors should fit in this mask:
	 *	000f 0000 1mmm cccc (binary)
	 * where:
	 *	f = Correction Report Filtering Bit. If 1, subsequent errors
	 *	    won't be shown
	 *	mmm = error type
	 *	cccc = channel
	 * If the mask doesn't match, report an error to the parsing logic
	 */
	if (! ((errcode & 0xef80) == 0x80)) {
		optype = "Can't parse: it is not a mem";
	} else {
		switch (optypenum) {
		case 0:
1707
			optype = "generic undef request error";
1708 1709
			break;
		case 1:
1710
			optype = "memory read error";
1711 1712
			break;
		case 2:
1713
			optype = "memory write error";
1714 1715
			break;
		case 3:
1716
			optype = "addr/cmd error";
1717 1718
			break;
		case 4:
1719
			optype = "memory scrubbing error";
1720 1721 1722 1723 1724 1725 1726
			break;
		default:
			optype = "reserved";
			break;
		}
	}

1727 1728 1729 1730
	/* Only decode errors with an valid address (ADDRV) */
	if (!GET_BITFIELD(m->status, 58, 58))
		return;

1731
	rc = get_memory_error_data(mci, m->addr, &socket,
1732
				   &channel_mask, &rank, &area_type, msg);
1733
	if (rc < 0)
1734
		goto err_parsing;
1735 1736
	new_mci = get_mci_for_node_id(socket);
	if (!new_mci) {
1737 1738
		strcpy(msg, "Error: socket got corrupted!");
		goto err_parsing;
1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
	}
	mci = new_mci;
	pvt = mci->pvt_info;

	first_channel = find_first_bit(&channel_mask, NUM_CHANNELS);

	if (rank < 4)
		dimm = 0;
	else if (rank < 8)
		dimm = 1;
	else
		dimm = 2;


	/*
1754 1755 1756 1757
	 * FIXME: On some memory configurations (mirror, lockstep), the
	 * Memory Controller can't point the error to a single DIMM. The
	 * EDAC core should be handling the channel mask, in order to point
	 * to the group of dimm's where the error may be happening.
1758
	 */
1759
	snprintf(msg, sizeof(msg),
1760
		 "%s%s area:%s err_code:%04x:%04x socket:%d channel_mask:%ld rank:%d",
1761 1762 1763 1764 1765 1766 1767
		 overflow ? " OVERFLOW" : "",
		 (uncorrected_error && recoverable) ? " recoverable" : "",
		 area_type,
		 mscod, errcode,
		 socket,
		 channel_mask,
		 rank);
1768

1769
	edac_dbg(0, "%s\n", msg);
1770

1771 1772
	/* FIXME: need support for channel mask */

1773
	/* Call the helper to output message */
1774
	edac_mc_handle_error(tp_event, mci, core_err_cnt,
1775 1776
			     m->addr >> PAGE_SHIFT, m->addr & ~PAGE_MASK, 0,
			     channel, dimm, -1,
1777
			     optype, msg);
1778 1779
	return;
err_parsing:
1780
	edac_mc_handle_error(tp_event, mci, core_err_cnt, 0, 0, 0,
1781
			     -1, -1, -1,
1782
			     msg, "");
1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844

}

/*
 *	sbridge_check_error	Retrieve and process errors reported by the
 *				hardware. Called by the Core module.
 */
static void sbridge_check_error(struct mem_ctl_info *mci)
{
	struct sbridge_pvt *pvt = mci->pvt_info;
	int i;
	unsigned count = 0;
	struct mce *m;

	/*
	 * MCE first step: Copy all mce errors into a temporary buffer
	 * We use a double buffering here, to reduce the risk of
	 * loosing an error.
	 */
	smp_rmb();
	count = (pvt->mce_out + MCE_LOG_LEN - pvt->mce_in)
		% MCE_LOG_LEN;
	if (!count)
		return;

	m = pvt->mce_outentry;
	if (pvt->mce_in + count > MCE_LOG_LEN) {
		unsigned l = MCE_LOG_LEN - pvt->mce_in;

		memcpy(m, &pvt->mce_entry[pvt->mce_in], sizeof(*m) * l);
		smp_wmb();
		pvt->mce_in = 0;
		count -= l;
		m += l;
	}
	memcpy(m, &pvt->mce_entry[pvt->mce_in], sizeof(*m) * count);
	smp_wmb();
	pvt->mce_in += count;

	smp_rmb();
	if (pvt->mce_overrun) {
		sbridge_printk(KERN_ERR, "Lost %d memory errors\n",
			      pvt->mce_overrun);
		smp_wmb();
		pvt->mce_overrun = 0;
	}

	/*
	 * MCE second step: parse errors and display
	 */
	for (i = 0; i < count; i++)
		sbridge_mce_output_error(mci, &pvt->mce_outentry[i]);
}

/*
 * sbridge_mce_check_error	Replicates mcelog routine to get errors
 *				This routine simply queues mcelog errors, and
 *				return. The error itself should be handled later
 *				by sbridge_check_error.
 * WARNING: As this routine should be called at NMI time, extra care should
 * be taken to avoid deadlocks, and to be as fast as possible.
 */
1845 1846
static int sbridge_mce_check_error(struct notifier_block *nb, unsigned long val,
				   void *data)
1847
{
1848 1849 1850
	struct mce *mce = (struct mce *)data;
	struct mem_ctl_info *mci;
	struct sbridge_pvt *pvt;
1851
	char *type;
1852

1853 1854 1855
	if (get_edac_report_status() == EDAC_REPORTING_DISABLED)
		return NOTIFY_DONE;

1856 1857 1858 1859
	mci = get_mci_for_node_id(mce->socketid);
	if (!mci)
		return NOTIFY_BAD;
	pvt = mci->pvt_info;
1860 1861 1862 1863 1864 1865 1866 1867

	/*
	 * Just let mcelog handle it if the error is
	 * outside the memory controller. A memory error
	 * is indicated by bit 7 = 1 and bits = 8-11,13-15 = 0.
	 * bit 12 has an special meaning.
	 */
	if ((mce->status & 0xefff) >> 7 != 1)
1868
		return NOTIFY_DONE;
1869

1870 1871 1872 1873 1874
	if (mce->mcgstatus & MCG_STATUS_MCIP)
		type = "Exception";
	else
		type = "Event";

1875
	sbridge_mc_printk(mci, KERN_DEBUG, "HANDLING MCE MEMORY ERROR\n");
1876

1877 1878 1879 1880 1881 1882
	sbridge_mc_printk(mci, KERN_DEBUG, "CPU %d: Machine Check %s: %Lx "
			  "Bank %d: %016Lx\n", mce->extcpu, type,
			  mce->mcgstatus, mce->bank, mce->status);
	sbridge_mc_printk(mci, KERN_DEBUG, "TSC %llx ", mce->tsc);
	sbridge_mc_printk(mci, KERN_DEBUG, "ADDR %llx ", mce->addr);
	sbridge_mc_printk(mci, KERN_DEBUG, "MISC %llx ", mce->misc);
1883

1884 1885 1886
	sbridge_mc_printk(mci, KERN_DEBUG, "PROCESSOR %u:%x TIME %llu SOCKET "
			  "%u APIC %x\n", mce->cpuvendor, mce->cpuid,
			  mce->time, mce->socketid, mce->apicid);
1887 1888 1889

	/* Only handle if it is the right mc controller */
	if (cpu_data(mce->cpu).phys_proc_id != pvt->sbridge_dev->mc)
1890
		return NOTIFY_DONE;
1891 1892 1893 1894 1895

	smp_rmb();
	if ((pvt->mce_out + 1) % MCE_LOG_LEN == pvt->mce_in) {
		smp_wmb();
		pvt->mce_overrun++;
1896
		return NOTIFY_DONE;
1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
	}

	/* Copy memory error at the ringbuffer */
	memcpy(&pvt->mce_entry[pvt->mce_out], mce, sizeof(*mce));
	smp_wmb();
	pvt->mce_out = (pvt->mce_out + 1) % MCE_LOG_LEN;

	/* Handle fatal errors immediately */
	if (mce->mcgstatus & 1)
		sbridge_check_error(mci);

	/* Advice mcelog that the error were handled */
1909
	return NOTIFY_STOP;
1910 1911
}

1912 1913 1914 1915
static struct notifier_block sbridge_mce_dec = {
	.notifier_call      = sbridge_mce_check_error,
};

1916 1917 1918 1919 1920 1921 1922 1923 1924 1925
/****************************************************************************
			EDAC register/unregister logic
 ****************************************************************************/

static void sbridge_unregister_mci(struct sbridge_dev *sbridge_dev)
{
	struct mem_ctl_info *mci = sbridge_dev->mci;
	struct sbridge_pvt *pvt;

	if (unlikely(!mci || !mci->pvt_info)) {
1926
		edac_dbg(0, "MC: dev = %p\n", &sbridge_dev->pdev[0]->dev);
1927 1928 1929 1930 1931 1932 1933

		sbridge_printk(KERN_ERR, "Couldn't find mci handler\n");
		return;
	}

	pvt = mci->pvt_info;

1934 1935
	edac_dbg(0, "MC: mci = %p, dev = %p\n",
		 mci, &sbridge_dev->pdev[0]->dev);
1936 1937

	/* Remove MC sysfs nodes */
1938
	edac_mc_del_mc(mci->pdev);
1939

1940
	edac_dbg(1, "%s: free mci struct\n", mci->ctl_name);
1941 1942 1943 1944 1945
	kfree(mci->ctl_name);
	edac_mc_free(mci);
	sbridge_dev->mci = NULL;
}

1946
static int sbridge_register_mci(struct sbridge_dev *sbridge_dev, enum type type)
1947 1948
{
	struct mem_ctl_info *mci;
1949
	struct edac_mc_layer layers[2];
1950
	struct sbridge_pvt *pvt;
1951
	struct pci_dev *pdev = sbridge_dev->pdev[0];
1952
	int rc;
1953 1954

	/* Check the number of active and not disabled channels */
1955
	rc = check_if_ecc_is_active(sbridge_dev->bus);
1956 1957 1958 1959
	if (unlikely(rc < 0))
		return rc;

	/* allocate a new MC control structure */
1960 1961 1962 1963 1964 1965
	layers[0].type = EDAC_MC_LAYER_CHANNEL;
	layers[0].size = NUM_CHANNELS;
	layers[0].is_virt_csrow = false;
	layers[1].type = EDAC_MC_LAYER_SLOT;
	layers[1].size = MAX_DIMMS;
	layers[1].is_virt_csrow = true;
1966
	mci = edac_mc_alloc(sbridge_dev->mc, ARRAY_SIZE(layers), layers,
1967 1968
			    sizeof(*pvt));

1969 1970 1971
	if (unlikely(!mci))
		return -ENOMEM;

1972
	edac_dbg(0, "MC: mci = %p, dev = %p\n",
1973
		 mci, &pdev->dev);
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986

	pvt = mci->pvt_info;
	memset(pvt, 0, sizeof(*pvt));

	/* Associate sbridge_dev and mci for future usage */
	pvt->sbridge_dev = sbridge_dev;
	sbridge_dev->mci = mci;

	mci->mtype_cap = MEM_FLAG_DDR3;
	mci->edac_ctl_cap = EDAC_FLAG_NONE;
	mci->edac_cap = EDAC_FLAG_NONE;
	mci->mod_name = "sbridge_edac.c";
	mci->mod_ver = SBRIDGE_REVISION;
1987
	mci->dev_name = pci_name(pdev);
1988 1989 1990 1991 1992
	mci->ctl_page_to_phys = NULL;

	/* Set the function pointer to an actual operation function */
	mci->edac_check = sbridge_check_error;

1993 1994 1995 1996 1997 1998
	pvt->info.type = type;
	if (type == IVY_BRIDGE) {
		pvt->info.rankcfgr = IB_RANK_CFG_A;
		pvt->info.get_tolm = ibridge_get_tolm;
		pvt->info.get_tohm = ibridge_get_tohm;
		pvt->info.dram_rule = ibridge_dram_rule;
1999
		pvt->info.get_memory_type = get_memory_type;
2000
		pvt->info.get_node_id = get_node_id;
2001
		pvt->info.rir_limit = rir_limit;
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
		pvt->info.max_sad = ARRAY_SIZE(ibridge_dram_rule);
		pvt->info.interleave_list = ibridge_interleave_list;
		pvt->info.max_interleave = ARRAY_SIZE(ibridge_interleave_list);
		pvt->info.interleave_pkg = ibridge_interleave_pkg;
		mci->ctl_name = kasprintf(GFP_KERNEL, "Ivy Bridge Socket#%d", mci->mc_idx);

		/* Store pci devices at mci for faster access */
		rc = ibridge_mci_bind_devs(mci, sbridge_dev);
		if (unlikely(rc < 0))
			goto fail0;
	} else {
		pvt->info.rankcfgr = SB_RANK_CFG_A;
		pvt->info.get_tolm = sbridge_get_tolm;
		pvt->info.get_tohm = sbridge_get_tohm;
		pvt->info.dram_rule = sbridge_dram_rule;
2017
		pvt->info.get_memory_type = get_memory_type;
2018
		pvt->info.get_node_id = get_node_id;
2019
		pvt->info.rir_limit = rir_limit;
2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031
		pvt->info.max_sad = ARRAY_SIZE(sbridge_dram_rule);
		pvt->info.interleave_list = sbridge_interleave_list;
		pvt->info.max_interleave = ARRAY_SIZE(sbridge_interleave_list);
		pvt->info.interleave_pkg = sbridge_interleave_pkg;
		mci->ctl_name = kasprintf(GFP_KERNEL, "Sandy Bridge Socket#%d", mci->mc_idx);

		/* Store pci devices at mci for faster access */
		rc = sbridge_mci_bind_devs(mci, sbridge_dev);
		if (unlikely(rc < 0))
			goto fail0;
	}

2032 2033 2034 2035 2036 2037

	/* Get dimm basic config and the memory layout */
	get_dimm_config(mci);
	get_memory_layout(mci);

	/* record ptr to the generic device */
2038
	mci->pdev = &pdev->dev;
2039 2040 2041

	/* add this new MC control structure to EDAC's list of MCs */
	if (unlikely(edac_mc_add_mc(mci))) {
2042
		edac_dbg(0, "MC: failed edac_mc_add_mc()\n");
2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063
		rc = -EINVAL;
		goto fail0;
	}

	return 0;

fail0:
	kfree(mci->ctl_name);
	edac_mc_free(mci);
	sbridge_dev->mci = NULL;
	return rc;
}

/*
 *	sbridge_probe	Probe for ONE instance of device to see if it is
 *			present.
 *	return:
 *		0 for FOUND a device
 *		< 0 for error code
 */

2064
static int sbridge_probe(struct pci_dev *pdev, const struct pci_device_id *id)
2065 2066 2067 2068
{
	int rc;
	u8 mc, num_mc = 0;
	struct sbridge_dev *sbridge_dev;
2069
	enum type type;
2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082

	/* get the pci devices we want to reserve for our use */
	mutex_lock(&sbridge_edac_lock);

	/*
	 * All memory controllers are allocated at the first pass.
	 */
	if (unlikely(probed >= 1)) {
		mutex_unlock(&sbridge_edac_lock);
		return -ENODEV;
	}
	probed++;

2083 2084 2085 2086 2087 2088 2089
	if (pdev->device == PCI_DEVICE_ID_INTEL_IBRIDGE_IMC_HA0_TA) {
		rc = sbridge_get_all_devices(&num_mc, pci_dev_descr_ibridge_table);
		type = IVY_BRIDGE;
	} else {
		rc = sbridge_get_all_devices(&num_mc, pci_dev_descr_sbridge_table);
		type = SANDY_BRIDGE;
	}
2090 2091 2092 2093 2094
	if (unlikely(rc < 0))
		goto fail0;
	mc = 0;

	list_for_each_entry(sbridge_dev, &sbridge_edac_list, list) {
2095 2096
		edac_dbg(0, "Registering MC#%d (%d of %d)\n",
			 mc, mc + 1, num_mc);
2097
		sbridge_dev->mc = mc++;
2098
		rc = sbridge_register_mci(sbridge_dev, type);
2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
		if (unlikely(rc < 0))
			goto fail1;
	}

	sbridge_printk(KERN_INFO, "Driver loaded.\n");

	mutex_unlock(&sbridge_edac_lock);
	return 0;

fail1:
	list_for_each_entry(sbridge_dev, &sbridge_edac_list, list)
		sbridge_unregister_mci(sbridge_dev);

	sbridge_put_all_devices();
fail0:
	mutex_unlock(&sbridge_edac_lock);
	return rc;
}

/*
 *	sbridge_remove	destructor for one instance of device
 *
 */
2122
static void sbridge_remove(struct pci_dev *pdev)
2123 2124 2125
{
	struct sbridge_dev *sbridge_dev;

2126
	edac_dbg(0, "\n");
2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162

	/*
	 * we have a trouble here: pdev value for removal will be wrong, since
	 * it will point to the X58 register used to detect that the machine
	 * is a Nehalem or upper design. However, due to the way several PCI
	 * devices are grouped together to provide MC functionality, we need
	 * to use a different method for releasing the devices
	 */

	mutex_lock(&sbridge_edac_lock);

	if (unlikely(!probed)) {
		mutex_unlock(&sbridge_edac_lock);
		return;
	}

	list_for_each_entry(sbridge_dev, &sbridge_edac_list, list)
		sbridge_unregister_mci(sbridge_dev);

	/* Release PCI resources */
	sbridge_put_all_devices();

	probed--;

	mutex_unlock(&sbridge_edac_lock);
}

MODULE_DEVICE_TABLE(pci, sbridge_pci_tbl);

/*
 *	sbridge_driver	pci_driver structure for this module
 *
 */
static struct pci_driver sbridge_driver = {
	.name     = "sbridge_edac",
	.probe    = sbridge_probe,
2163
	.remove   = sbridge_remove,
2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174
	.id_table = sbridge_pci_tbl,
};

/*
 *	sbridge_init		Module entry function
 *			Try to initialize this module for its devices
 */
static int __init sbridge_init(void)
{
	int pci_rc;

2175
	edac_dbg(2, "\n");
2176 2177 2178 2179 2180

	/* Ensure that the OPSTATE is set correctly for POLL or NMI */
	opstate_init();

	pci_rc = pci_register_driver(&sbridge_driver);
2181 2182
	if (pci_rc >= 0) {
		mce_register_decode_chain(&sbridge_mce_dec);
2183 2184
		if (get_edac_report_status() == EDAC_REPORTING_DISABLED)
			sbridge_printk(KERN_WARNING, "Loading driver, error reporting disabled.\n");
2185
		return 0;
2186
	}
2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199

	sbridge_printk(KERN_ERR, "Failed to register device with error %d.\n",
		      pci_rc);

	return pci_rc;
}

/*
 *	sbridge_exit()	Module exit function
 *			Unregister the driver
 */
static void __exit sbridge_exit(void)
{
2200
	edac_dbg(2, "\n");
2201
	pci_unregister_driver(&sbridge_driver);
2202
	mce_unregister_decode_chain(&sbridge_mce_dec);
2203 2204 2205 2206 2207 2208 2209 2210 2211
}

module_init(sbridge_init);
module_exit(sbridge_exit);

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

MODULE_LICENSE("GPL");
2212
MODULE_AUTHOR("Mauro Carvalho Chehab");
2213
MODULE_AUTHOR("Red Hat Inc. (http://www.redhat.com)");
2214
MODULE_DESCRIPTION("MC Driver for Intel Sandy Bridge and Ivy Bridge memory controllers - "
2215
		   SBRIDGE_REVISION);