skx_common.c 18.2 KB
Newer Older
1 2
// SPDX-License-Identifier: GPL-2.0
/*
3 4 5 6 7 8 9 10 11 12
 *
 * Shared code by both skx_edac and i10nm_edac. Originally split out
 * from the skx_edac driver.
 *
 * This file is linked into both skx_edac and i10nm_edac drivers. In
 * order to avoid link errors, this file must be like a pure library
 * without including symbols and defines which would otherwise conflict,
 * when linked once into a module and into a built-in object, at the
 * same time. For example, __this_module symbol references when that
 * file is being linked into a built-in object.
13 14 15 16 17 18 19 20 21 22 23 24 25
 *
 * Copyright (c) 2018, Intel Corporation.
 */

#include <linux/acpi.h>
#include <linux/dmi.h>
#include <linux/adxl.h>
#include <acpi/nfit.h>
#include <asm/mce.h>
#include "edac_module.h"
#include "skx_common.h"

static const char * const component_names[] = {
26 27 28 29
	[INDEX_SOCKET]		= "ProcessorSocketId",
	[INDEX_MEMCTRL]		= "MemoryControllerId",
	[INDEX_CHANNEL]		= "ChannelId",
	[INDEX_DIMM]		= "DimmSlotId",
30
	[INDEX_CS]		= "ChipSelect",
31 32 33
	[INDEX_NM_MEMCTRL]	= "NmMemoryControllerId",
	[INDEX_NM_CHANNEL]	= "NmChannelId",
	[INDEX_NM_DIMM]		= "NmDimmSlotId",
34
	[INDEX_NM_CS]		= "NmChipSelect",
35 36 37 38 39 40 41
};

static int component_indices[ARRAY_SIZE(component_names)];
static int adxl_component_count;
static const char * const *adxl_component_names;
static u64 *adxl_values;
static char *adxl_msg;
42
static unsigned long adxl_nm_bitmap;
43 44

static char skx_msg[MSG_SIZE];
45
static skx_decode_f driver_decode;
46
static skx_show_retry_log_f skx_show_retry_rd_err_log;
47 48
static u64 skx_tolm, skx_tohm;
static LIST_HEAD(dev_edac_list);
49
static bool skx_mem_cfg_2lm;
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

int __init skx_adxl_get(void)
{
	const char * const *names;
	int i, j;

	names = adxl_get_component_names();
	if (!names) {
		skx_printk(KERN_NOTICE, "No firmware support for address translation.\n");
		return -ENODEV;
	}

	for (i = 0; i < INDEX_MAX; i++) {
		for (j = 0; names[j]; j++) {
			if (!strcmp(component_names[i], names[j])) {
				component_indices[i] = j;
66 67 68 69

				if (i >= INDEX_NM_FIRST)
					adxl_nm_bitmap |= 1 << i;

70 71 72 73
				break;
			}
		}

74
		if (!names[j] && i < INDEX_NM_FIRST)
75 76 77
			goto err;
	}

78 79 80 81 82 83 84
	if (skx_mem_cfg_2lm) {
		if (!adxl_nm_bitmap)
			skx_printk(KERN_NOTICE, "Not enough ADXL components for 2-level memory.\n");
		else
			edac_dbg(2, "adxl_nm_bitmap: 0x%lx\n", adxl_nm_bitmap);
	}

85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
	adxl_component_names = names;
	while (*names++)
		adxl_component_count++;

	adxl_values = kcalloc(adxl_component_count, sizeof(*adxl_values),
			      GFP_KERNEL);
	if (!adxl_values) {
		adxl_component_count = 0;
		return -ENOMEM;
	}

	adxl_msg = kzalloc(MSG_SIZE, GFP_KERNEL);
	if (!adxl_msg) {
		adxl_component_count = 0;
		kfree(adxl_values);
		return -ENOMEM;
	}

	return 0;
err:
	skx_printk(KERN_ERR, "'%s' is not matched from DSM parameters: ",
		   component_names[i]);
	for (j = 0; names[j]; j++)
		skx_printk(KERN_CONT, "%s ", names[j]);
	skx_printk(KERN_CONT, "\n");

	return -ENODEV;
}

void __exit skx_adxl_put(void)
{
	kfree(adxl_values);
	kfree(adxl_msg);
}

120
static bool skx_adxl_decode(struct decoded_addr *res, bool error_in_1st_level_mem)
121
{
122
	struct skx_dev *d;
123 124 125 126 127 128 129 130 131 132 133 134 135 136
	int i, len = 0;

	if (res->addr >= skx_tohm || (res->addr >= skx_tolm &&
				      res->addr < BIT_ULL(32))) {
		edac_dbg(0, "Address 0x%llx out of range\n", res->addr);
		return false;
	}

	if (adxl_decode(res->addr, adxl_values)) {
		edac_dbg(0, "Failed to decode 0x%llx\n", res->addr);
		return false;
	}

	res->socket  = (int)adxl_values[component_indices[INDEX_SOCKET]];
137 138 139 140 141 142 143
	if (error_in_1st_level_mem) {
		res->imc     = (adxl_nm_bitmap & BIT_NM_MEMCTRL) ?
			       (int)adxl_values[component_indices[INDEX_NM_MEMCTRL]] : -1;
		res->channel = (adxl_nm_bitmap & BIT_NM_CHANNEL) ?
			       (int)adxl_values[component_indices[INDEX_NM_CHANNEL]] : -1;
		res->dimm    = (adxl_nm_bitmap & BIT_NM_DIMM) ?
			       (int)adxl_values[component_indices[INDEX_NM_DIMM]] : -1;
144 145
		res->cs      = (adxl_nm_bitmap & BIT_NM_CS) ?
			       (int)adxl_values[component_indices[INDEX_NM_CS]] : -1;
146 147 148 149
	} else {
		res->imc     = (int)adxl_values[component_indices[INDEX_MEMCTRL]];
		res->channel = (int)adxl_values[component_indices[INDEX_CHANNEL]];
		res->dimm    = (int)adxl_values[component_indices[INDEX_DIMM]];
150
		res->cs      = (int)adxl_values[component_indices[INDEX_CS]];
151
	}
152

153
	if (res->imc > NUM_IMC - 1 || res->imc < 0) {
154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
		skx_printk(KERN_ERR, "Bad imc %d\n", res->imc);
		return false;
	}

	list_for_each_entry(d, &dev_edac_list, list) {
		if (d->imc[0].src_id == res->socket) {
			res->dev = d;
			break;
		}
	}

	if (!res->dev) {
		skx_printk(KERN_ERR, "No device for src_id %d imc %d\n",
			   res->socket, res->imc);
		return false;
	}

171 172 173 174 175 176 177 178 179 180
	for (i = 0; i < adxl_component_count; i++) {
		if (adxl_values[i] == ~0x0ull)
			continue;

		len += snprintf(adxl_msg + len, MSG_SIZE - len, " %s:0x%llx",
				adxl_component_names[i], adxl_values[i]);
		if (MSG_SIZE - len <= 0)
			break;
	}

181 182
	res->decoded_by_adxl = true;

183 184 185
	return true;
}

186 187 188 189 190
void skx_set_mem_cfg(bool mem_cfg_2lm)
{
	skx_mem_cfg_2lm = mem_cfg_2lm;
}

191
void skx_set_decode(skx_decode_f decode, skx_show_retry_log_f show_retry_log)
192
{
193
	driver_decode = decode;
194
	skx_show_retry_rd_err_log = show_retry_log;
195 196
}

197
int skx_get_src_id(struct skx_dev *d, int off, u8 *id)
198 199 200
{
	u32 reg;

201
	if (pci_read_config_dword(d->util_all, off, &reg)) {
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
		skx_printk(KERN_ERR, "Failed to read src id\n");
		return -ENODEV;
	}

	*id = GET_BITFIELD(reg, 12, 14);
	return 0;
}

int skx_get_node_id(struct skx_dev *d, u8 *id)
{
	u32 reg;

	if (pci_read_config_dword(d->util_all, 0xf4, &reg)) {
		skx_printk(KERN_ERR, "Failed to read node id\n");
		return -ENODEV;
	}

	*id = GET_BITFIELD(reg, 0, 2);
	return 0;
}

static int get_width(u32 mtr)
{
	switch (GET_BITFIELD(mtr, 8, 9)) {
	case 0:
		return DEV_X4;
	case 1:
		return DEV_X8;
	case 2:
		return DEV_X16;
	}
	return DEV_UNKNOWN;
}

/*
237
 * We use the per-socket device @cfg->did to count how many sockets are present,
238 239 240
 * and to detemine which PCI buses are associated with each socket. Allocate
 * and build the full list of all the skx_dev structures that we need here.
 */
241
int skx_get_all_bus_mappings(struct res_config *cfg, struct list_head **list)
242 243 244 245 246 247 248 249
{
	struct pci_dev *pdev, *prev;
	struct skx_dev *d;
	u32 reg;
	int ndev = 0;

	prev = NULL;
	for (;;) {
250
		pdev = pci_get_device(PCI_VENDOR_ID_INTEL, cfg->decs_did, prev);
251 252 253 254 255 256 257 258 259
		if (!pdev)
			break;
		ndev++;
		d = kzalloc(sizeof(*d), GFP_KERNEL);
		if (!d) {
			pci_dev_put(pdev);
			return -ENOMEM;
		}

260
		if (pci_read_config_dword(pdev, cfg->busno_cfg_offset, &reg)) {
261 262 263 264 265 266 267 268
			kfree(d);
			pci_dev_put(pdev);
			skx_printk(KERN_ERR, "Failed to read bus idx\n");
			return -ENODEV;
		}

		d->bus[0] = GET_BITFIELD(reg, 0, 7);
		d->bus[1] = GET_BITFIELD(reg, 8, 15);
269
		if (cfg->type == SKX) {
270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294
			d->seg = pci_domain_nr(pdev->bus);
			d->bus[2] = GET_BITFIELD(reg, 16, 23);
			d->bus[3] = GET_BITFIELD(reg, 24, 31);
		} else {
			d->seg = GET_BITFIELD(reg, 16, 23);
		}

		edac_dbg(2, "busses: 0x%x, 0x%x, 0x%x, 0x%x\n",
			 d->bus[0], d->bus[1], d->bus[2], d->bus[3]);
		list_add_tail(&d->list, &dev_edac_list);
		prev = pdev;
	}

	if (list)
		*list = &dev_edac_list;
	return ndev;
}

int skx_get_hi_lo(unsigned int did, int off[], u64 *tolm, u64 *tohm)
{
	struct pci_dev *pdev;
	u32 reg;

	pdev = pci_get_device(PCI_VENDOR_ID_INTEL, did, NULL);
	if (!pdev) {
295
		edac_dbg(2, "Can't get tolm/tohm\n");
296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342
		return -ENODEV;
	}

	if (pci_read_config_dword(pdev, off[0], &reg)) {
		skx_printk(KERN_ERR, "Failed to read tolm\n");
		goto fail;
	}
	skx_tolm = reg;

	if (pci_read_config_dword(pdev, off[1], &reg)) {
		skx_printk(KERN_ERR, "Failed to read lower tohm\n");
		goto fail;
	}
	skx_tohm = reg;

	if (pci_read_config_dword(pdev, off[2], &reg)) {
		skx_printk(KERN_ERR, "Failed to read upper tohm\n");
		goto fail;
	}
	skx_tohm |= (u64)reg << 32;

	pci_dev_put(pdev);
	*tolm = skx_tolm;
	*tohm = skx_tohm;
	edac_dbg(2, "tolm = 0x%llx tohm = 0x%llx\n", skx_tolm, skx_tohm);
	return 0;
fail:
	pci_dev_put(pdev);
	return -ENODEV;
}

static int skx_get_dimm_attr(u32 reg, int lobit, int hibit, int add,
			     int minval, int maxval, const char *name)
{
	u32 val = GET_BITFIELD(reg, lobit, hibit);

	if (val < minval || val > maxval) {
		edac_dbg(2, "bad %s = %d (raw=0x%x)\n", name, val, reg);
		return -EINVAL;
	}
	return val + add;
}

#define numrank(reg)	skx_get_dimm_attr(reg, 12, 13, 0, 0, 2, "ranks")
#define numrow(reg)	skx_get_dimm_attr(reg, 2, 4, 12, 1, 6, "rows")
#define numcol(reg)	skx_get_dimm_attr(reg, 0, 1, 10, 0, 2, "cols")

343
int skx_get_dimm_info(u32 mtr, u32 mcmtr, u32 amap, struct dimm_info *dimm,
344 345
		      struct skx_imc *imc, int chan, int dimmno,
		      struct res_config *cfg)
346
{
347 348
	int  banks, ranks, rows, cols, npages;
	enum mem_type mtype;
349 350 351 352
	u64 size;

	ranks = numrank(mtr);
	rows = numrow(mtr);
353
	cols = imc->hbm_mc ? 6 : numcol(mtr);
354

355 356 357 358
	if (imc->hbm_mc) {
		banks = 32;
		mtype = MEM_HBM2;
	} else if (cfg->support_ddr5 && (amap & 0x8)) {
359 360 361 362 363 364 365
		banks = 32;
		mtype = MEM_DDR5;
	} else {
		banks = 16;
		mtype = MEM_DDR4;
	}

366 367 368 369 370 371 372 373 374 375
	/*
	 * Compute size in 8-byte (2^3) words, then shift to MiB (2^20)
	 */
	size = ((1ull << (rows + cols + ranks)) * banks) >> (20 - 3);
	npages = MiB_TO_PAGES(size);

	edac_dbg(0, "mc#%d: channel %d, dimm %d, %lld MiB (%d pages) bank: %d, rank: %d, row: 0x%x, col: 0x%x\n",
		 imc->mc, chan, dimmno, size, npages,
		 banks, 1 << ranks, rows, cols);

376 377
	imc->chan[chan].dimms[dimmno].close_pg = GET_BITFIELD(mcmtr, 0, 0);
	imc->chan[chan].dimms[dimmno].bank_xor_enable = GET_BITFIELD(mcmtr, 9, 9);
378 379 380 381 382 383 384
	imc->chan[chan].dimms[dimmno].fine_grain_bank = GET_BITFIELD(amap, 0, 0);
	imc->chan[chan].dimms[dimmno].rowbits = rows;
	imc->chan[chan].dimms[dimmno].colbits = cols;

	dimm->nr_pages = npages;
	dimm->grain = 32;
	dimm->dtype = get_width(mtr);
385
	dimm->mtype = mtype;
386
	dimm->edac_mode = EDAC_SECDED; /* likely better than this */
387 388 389 390 391 392 393

	if (imc->hbm_mc)
		snprintf(dimm->label, sizeof(dimm->label), "CPU_SrcID#%u_HBMC#%u_Chan#%u",
			 imc->src_id, imc->lmc, chan);
	else
		snprintf(dimm->label, sizeof(dimm->label), "CPU_SrcID#%u_MC#%u_Chan#%u_DIMM#%u",
			 imc->src_id, imc->lmc, chan, dimmno);
394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447

	return 1;
}

int skx_get_nvdimm_info(struct dimm_info *dimm, struct skx_imc *imc,
			int chan, int dimmno, const char *mod_str)
{
	int smbios_handle;
	u32 dev_handle;
	u16 flags;
	u64 size = 0;

	dev_handle = ACPI_NFIT_BUILD_DEVICE_HANDLE(dimmno, chan, imc->lmc,
						   imc->src_id, 0);

	smbios_handle = nfit_get_smbios_id(dev_handle, &flags);
	if (smbios_handle == -EOPNOTSUPP) {
		pr_warn_once("%s: Can't find size of NVDIMM. Try enabling CONFIG_ACPI_NFIT\n", mod_str);
		goto unknown_size;
	}

	if (smbios_handle < 0) {
		skx_printk(KERN_ERR, "Can't find handle for NVDIMM ADR=0x%x\n", dev_handle);
		goto unknown_size;
	}

	if (flags & ACPI_NFIT_MEM_MAP_FAILED) {
		skx_printk(KERN_ERR, "NVDIMM ADR=0x%x is not mapped\n", dev_handle);
		goto unknown_size;
	}

	size = dmi_memdev_size(smbios_handle);
	if (size == ~0ull)
		skx_printk(KERN_ERR, "Can't find size for NVDIMM ADR=0x%x/SMBIOS=0x%x\n",
			   dev_handle, smbios_handle);

unknown_size:
	dimm->nr_pages = size >> PAGE_SHIFT;
	dimm->grain = 32;
	dimm->dtype = DEV_UNKNOWN;
	dimm->mtype = MEM_NVDIMM;
	dimm->edac_mode = EDAC_SECDED; /* likely better than this */

	edac_dbg(0, "mc#%d: channel %d, dimm %d, %llu MiB (%u pages)\n",
		 imc->mc, chan, dimmno, size >> 20, dimm->nr_pages);

	snprintf(dimm->label, sizeof(dimm->label), "CPU_SrcID#%u_MC#%u_Chan#%u_DIMM#%u",
		 imc->src_id, imc->lmc, chan, dimmno);

	return (size == 0 || size == ~0ull) ? 0 : 1;
}

int skx_register_mci(struct skx_imc *imc, struct pci_dev *pdev,
		     const char *ctl_name, const char *mod_str,
448 449
		     get_dimm_config_f get_dimm_config,
		     struct res_config *cfg)
450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483
{
	struct mem_ctl_info *mci;
	struct edac_mc_layer layers[2];
	struct skx_pvt *pvt;
	int rc;

	/* Allocate a new MC control structure */
	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 = NUM_DIMMS;
	layers[1].is_virt_csrow = true;
	mci = edac_mc_alloc(imc->mc, ARRAY_SIZE(layers), layers,
			    sizeof(struct skx_pvt));

	if (unlikely(!mci))
		return -ENOMEM;

	edac_dbg(0, "MC#%d: mci = %p\n", imc->mc, mci);

	/* Associate skx_dev and mci for future usage */
	imc->mci = mci;
	pvt = mci->pvt_info;
	pvt->imc = imc;

	mci->ctl_name = kasprintf(GFP_KERNEL, "%s#%d IMC#%d", ctl_name,
				  imc->node_id, imc->lmc);
	if (!mci->ctl_name) {
		rc = -ENOMEM;
		goto fail0;
	}

	mci->mtype_cap = MEM_FLAG_DDR4 | MEM_FLAG_NVDIMM;
484 485
	if (cfg->support_ddr5)
		mci->mtype_cap |= MEM_FLAG_DDR5;
486 487 488 489 490 491
	mci->edac_ctl_cap = EDAC_FLAG_NONE;
	mci->edac_cap = EDAC_FLAG_NONE;
	mci->mod_name = mod_str;
	mci->dev_name = pci_name(pdev);
	mci->ctl_page_to_phys = NULL;

492
	rc = get_dimm_config(mci, cfg);
493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537
	if (rc < 0)
		goto fail;

	/* Record ptr to the generic device */
	mci->pdev = &pdev->dev;

	/* Add this new MC control structure to EDAC's list of MCs */
	if (unlikely(edac_mc_add_mc(mci))) {
		edac_dbg(0, "MC: failed edac_mc_add_mc()\n");
		rc = -EINVAL;
		goto fail;
	}

	return 0;

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

static void skx_unregister_mci(struct skx_imc *imc)
{
	struct mem_ctl_info *mci = imc->mci;

	if (!mci)
		return;

	edac_dbg(0, "MC%d: mci = %p\n", imc->mc, mci);

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

	edac_dbg(1, "%s: free mci struct\n", mci->ctl_name);
	kfree(mci->ctl_name);
	edac_mc_free(mci);
}

static void skx_mce_output_error(struct mem_ctl_info *mci,
				 const struct mce *m,
				 struct decoded_addr *res)
{
	enum hw_event_mc_err_type tp_event;
538
	char *optype;
539 540 541
	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);
542
	bool scrub_err = false;
543
	bool recoverable;
544
	int len;
545 546 547 548 549 550 551 552 553 554 555
	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 optypenum = GET_BITFIELD(m->status, 4, 6);

	recoverable = GET_BITFIELD(m->status, 56, 56);

	if (uncorrected_error) {
		core_err_cnt = 1;
		if (ripv) {
			tp_event = HW_EVENT_ERR_UNCORRECTED;
556 557
		} else {
			tp_event = HW_EVENT_ERR_FATAL;
558 559 560 561 562 563
		}
	} else {
		tp_event = HW_EVENT_ERR_CORRECTED;
	}

	/*
564 565 566
	 * According to Intel Architecture spec vol 3B,
	 * Table 15-10 "IA32_MCi_Status [15:0] Compound Error Code Encoding"
	 * memory errors should fit one of these masks:
567
	 *	000f 0000 1mmm cccc (binary)
568
	 *	000f 0010 1mmm cccc (binary)	[RAM used as cache]
569 570 571 572 573 574 575
	 * 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
	 */
576
	if (!((errcode & 0xef80) == 0x80 || (errcode & 0xef80) == 0x280)) {
577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593
		optype = "Can't parse: it is not a mem";
	} else {
		switch (optypenum) {
		case 0:
			optype = "generic undef request error";
			break;
		case 1:
			optype = "memory read error";
			break;
		case 2:
			optype = "memory write error";
			break;
		case 3:
			optype = "addr/cmd error";
			break;
		case 4:
			optype = "memory scrubbing error";
594
			scrub_err = true;
595 596 597 598 599 600
			break;
		default:
			optype = "reserved";
			break;
		}
	}
601
	if (res->decoded_by_adxl) {
602
		len = snprintf(skx_msg, MSG_SIZE, "%s%s err_code:0x%04x:0x%04x %s",
603 604 605 606
			 overflow ? " OVERFLOW" : "",
			 (uncorrected_error && recoverable) ? " recoverable" : "",
			 mscod, errcode, adxl_msg);
	} else {
607
		len = snprintf(skx_msg, MSG_SIZE,
608
			 "%s%s err_code:0x%04x:0x%04x ProcessorSocketId:0x%x MemoryControllerId:0x%x PhysicalRankId:0x%x Row:0x%x Column:0x%x Bank:0x%x BankGroup:0x%x",
609 610 611 612
			 overflow ? " OVERFLOW" : "",
			 (uncorrected_error && recoverable) ? " recoverable" : "",
			 mscod, errcode,
			 res->socket, res->imc, res->rank,
613
			 res->row, res->column, res->bank_address, res->bank_group);
614 615
	}

616
	if (skx_show_retry_rd_err_log)
617
		skx_show_retry_rd_err_log(res, skx_msg + len, MSG_SIZE - len, scrub_err);
618

619 620 621 622 623 624 625 626 627
	edac_dbg(0, "%s\n", skx_msg);

	/* Call the helper to output message */
	edac_mc_handle_error(tp_event, mci, core_err_cnt,
			     m->addr >> PAGE_SHIFT, m->addr & ~PAGE_MASK, 0,
			     res->channel, res->dimm, -1,
			     optype, skx_msg);
}

628 629 630 631 632 633 634 635 636 637 638 639 640 641 642
static bool skx_error_in_1st_level_mem(const struct mce *m)
{
	u32 errcode;

	if (!skx_mem_cfg_2lm)
		return false;

	errcode = GET_BITFIELD(m->status, 0, 15);

	if ((errcode & 0xef80) != 0x280)
		return false;

	return true;
}

643 644 645 646 647 648 649 650
int skx_mce_check_error(struct notifier_block *nb, unsigned long val,
			void *data)
{
	struct mce *mce = (struct mce *)data;
	struct decoded_addr res;
	struct mem_ctl_info *mci;
	char *type;

651 652 653
	if (mce->kflags & MCE_HANDLED_CEC)
		return NOTIFY_DONE;

654 655 656 657 658
	/* ignore unless this is memory related with an address */
	if ((mce->status & 0xefff) >> 7 != 1 || !(mce->status & MCI_STATUS_ADDRV))
		return NOTIFY_DONE;

	memset(&res, 0, sizeof(res));
659
	res.mce  = mce;
660 661
	res.addr = mce->addr;

662 663 664 665
	/* Try driver decoder first */
	if (!(driver_decode && driver_decode(&res))) {
		/* Then try firmware decoder (ACPI DSM methods) */
		if (!(adxl_component_count && skx_adxl_decode(&res, skx_error_in_1st_level_mem(mce))))
666 667 668
			return NOTIFY_DONE;
	}

669 670
	mci = res.dev->imc[res.imc].mci;

671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693
	if (!mci)
		return NOTIFY_DONE;

	if (mce->mcgstatus & MCG_STATUS_MCIP)
		type = "Exception";
	else
		type = "Event";

	skx_mc_printk(mci, KERN_DEBUG, "HANDLING MCE MEMORY ERROR\n");

	skx_mc_printk(mci, KERN_DEBUG, "CPU %d: Machine Check %s: 0x%llx "
			   "Bank %d: 0x%llx\n", mce->extcpu, type,
			   mce->mcgstatus, mce->bank, mce->status);
	skx_mc_printk(mci, KERN_DEBUG, "TSC 0x%llx ", mce->tsc);
	skx_mc_printk(mci, KERN_DEBUG, "ADDR 0x%llx ", mce->addr);
	skx_mc_printk(mci, KERN_DEBUG, "MISC 0x%llx ", mce->misc);

	skx_mc_printk(mci, KERN_DEBUG, "PROCESSOR %u:0x%x TIME %llu SOCKET "
			   "%u APIC 0x%x\n", mce->cpuvendor, mce->cpuid,
			   mce->time, mce->socketid, mce->apicid);

	skx_mce_output_error(mci, mce, &res);

694
	mce->kflags |= MCE_HANDLED_EDAC;
695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723
	return NOTIFY_DONE;
}

void skx_remove(void)
{
	int i, j;
	struct skx_dev *d, *tmp;

	edac_dbg(0, "\n");

	list_for_each_entry_safe(d, tmp, &dev_edac_list, list) {
		list_del(&d->list);
		for (i = 0; i < NUM_IMC; i++) {
			if (d->imc[i].mci)
				skx_unregister_mci(&d->imc[i]);

			if (d->imc[i].mdev)
				pci_dev_put(d->imc[i].mdev);

			if (d->imc[i].mbase)
				iounmap(d->imc[i].mbase);

			for (j = 0; j < NUM_CHANNELS; j++) {
				if (d->imc[i].chan[j].cdev)
					pci_dev_put(d->imc[i].chan[j].cdev);
			}
		}
		if (d->util_all)
			pci_dev_put(d->util_all);
724 725
		if (d->pcu_cr3)
			pci_dev_put(d->pcu_cr3);
726 727 728 729 730 731 732 733
		if (d->sad_all)
			pci_dev_put(d->sad_all);
		if (d->uracu)
			pci_dev_put(d->uracu);

		kfree(d);
	}
}